Journal of Experimental Botany

Messina, Carlos D.;Podlich, Dean;Dong, Zhanshan;Samples, Mitch;Cooper, Mark

doi: 10.1093/jxb/erq329pmid: 21041371

Abstract The effectiveness of breeding strategies to increase drought resistance in crops could be increased further if some of the complexities in gene-to-phenotype (G→P) relations associated with epistasis, pleiotropy, and genotype-by-environment interactions could be captured in realistic G→P models, and represented in a quantitative manner useful for selection. This paper outlines a promising methodology. First, the concept of landscapes was extended from the study of fitness landscapes used in evolutionary genetics to the characterization of yield–trait-performance landscapes for agricultural environments and applications in plant breeding. Second, the E(NK) model of trait genetic architecture was extended to incorporate biophysical, physiological, and statistical components. Third, a graphical representation is proposed to visualize the yield–trait performance landscape concept for use in selection decisions. The methodology was demonstrated at a particular stage of a maize breeding programme with the objective of improving the drought tolerance of maize hybrids for the US Western Corn-Belt. The application of the framework to the genetic improvement of drought tolerance in maize supported selection of Doubled Haploid (DH) lines with improved levels of drought tolerance based on physiological genetic knowledge, prediction of test-cross yield within the target population of environments, and their predicted potential to sustain further genetic progress with additional cycles of selection. The existence of rugged yield-performance landscapes with multiple peaks and intervening valleys of lower performance, as shown in this study, supports the proposition that phenotyping strategies, and the directions emphasized in genomic selection can be improved by creating knowledge of the topology of yield–trait performance landscapes. Complex traits, fitness landscape, gene-to-phenotype, maize, maize physiology, plant breeding, yield trait performance landscape Introduction The formulation of the theory of selection, genetics, and quantitative genetics and further systematic application by plant breeders provided a framework for the sustained genetic improvement of yield and agronomic traits of crop plants for most of the 20th century (Duvick et al., 2004; Janick, 2004). Yield improvement for drought tolerance in modern maize has resulted from cycles of selection among genotypes sampled from elite germplasm and evaluated for performance in a sample of environments taken to represent the target population of environments (TPE) (Bolaños and Edmeades, 1993a; Bruce et al., 2002; Duvick et al., 2004). Although applied breeding has proved to be effective for yield improvement, scientists continually seek opportunities to improve the efficiency and effectiveness of breeding methodologies (Bolaños and Edmeades, 1996; Cooper and Hammer, 1996; Podlich et al., 2004; Jannink et al., 2010; van Eeuwijk et al., 2010b). Recent investigations have provided an extension of the theoretical framework for quantitative genetics that is built on allelic variation for genes functioning within networks (Cooper et al., 2005) and gene-to-phenotype (G→P) models structured around crop growth and development principles (Hammer et al., 2006). In this paper, we report on an implementation of this framework within a maize breeding programme with the objective of improving the drought tolerance of maize hybrids for the US Western Corn-Belt. Gene-to-phenotype modelling Advances in information technologies and prediction methods based on a mixed-model framework (Podlich et al., 2004; van Eeuwijk et al., 2005, 2010a, b; Boer et al., 2007) contributed to the realization of the potential use of DNA markers to predict the performance of new genotypes from founder genotype information. A toolkit of methodologies that quantifies trait genetic architecture based on mapping studies is now available to help breeders confront the problem of predicting the performance of new genotypes (Podlich et al., 2004; Cooper et al., 2009; van Eeuwijk et al., 2010a, b). Theoretical studies indicate that the effectiveness of breeding strategies could be increased further if complexities in G→P relations associated with epistasis, pleiotropy, and genotype×environment interactions (GEI) could be captured in realistic G→P models (Cooper et al., 2005, 2009). A generalized quantitative approach to consider G→P relations was proposed by Cooper et al. (2005) on the basis of the NK gene network model (Kauffman, 1993), extensions to consider the environmental context E(NK) (Cooper and Podlich, 2002), and limitations in our capacity to detect the genetic signal of complex quantitative traits fully (Podlich et al., 2004),  (1)where Pijk denotes phenotype for observation k on genotype i in environment type j, Ej identifies different environment type j within a reference TPE, N identifies the different genes influencing a trait within a given reference population of genotypes, K identifies the level of epistatic interaction between subsets of the total N genes that makeup genotype i in the form of a network graph where the edges of the graph identify the genes that interact, and ϵijk is the residual source of variation. The ( ) parenthesis notation indicates that the number of genes and their interactions are conditioned or can vary with the environment. The [ ] bracket notation is used here to identify subsets of the total N genes and interactions that are detected (D), undetected (U), and interactions arising from the intersection (I) between subsets of detected and undetected genes. Components including undetected genes can be considered to represent the genetic background effect. This framework incorporates explicitly epistasis and GEI within the genetic model and it can be applied to the modelling of trait genetic architecture for the continuum from simple to complex trait genetics. A unique feature of this framework is the explicit consideration of unknown components of the genetic architecture of the trait, which has important implications for prediction of response to selection within the context of a breeding programme (Cooper et al., 2005, 2009). The assumption about there being ‘no unknown QTL’ makes for overly-optimistic predictions of the value of known QTL and of genetic progress of QTL-based selection strategies. The G→P prediction problem remains a major challenge for most important traits in plant breeding (Cooper et al., 2002). Different concepts have been proposed to tackle this problem (White and Hoogenboom, 1996; Chapman et al., 2003; Reymond et al., 2003; Tardieu, 2003; Hoogenboom et al., 2004; Peccoud et al., 2004; Podlich et al., 2004; Yin et al., 2005; Hammer et al., 2005; Welch et al., 2005; Messina et al., 2006; Welcker et al., 2007; Chenu et al., 2008; Bertin et al., 2010). A synthesis of these methods suggests that a generalized framework can be proposed in which the genetic architecture of the trait is represented by specific parameterizations of the NK model, and phenotypes are predicted by applying a function (Γ) with parameters determined by NK  (2) The nature of the function Γ is determined by the biophysical properties of the trait under consideration, the degree of model simplification chosen to represent the trait, and by the number and connectivity between traits and the environment in multi-trait modelling (Γ). For example, Γ can represent a leaf elongation rate (dL/dt) model for maize (Salah and Tardieu, 1997; Reymond et al., 2003)  (3)where T is meristem temperature, VPD is vapour pressure deficit, ψ is soil water potential; b and c are constants coding for the response of leaf elongation rate to VPD and soil water potential after correction for T effects; a and T0 are the slope and x-intercept of the leaf elongation rate response to meristem temperature. This model can be rewritten using the NK framework (equation 2) as  (4)where now the phenotype (dL/dt or its integral form) is parameterized using the NK model, which is allowed to vary among parameters p. That is, the genetic networks associated with each of the model parameters could have common components. Therefore,pleiotropic effects are formally incorporated via shared nodes among gene networks, epistasis is determined by K as describe before and GEI is an emergent property of the model within different environmental conditions. Crop growth models that are structured to capture dynamic interactions of the physiological determinants of crop growth and development can be used as a framework for muti-trait integration (Γ) and to predict consequences of genotype-by-environment-by-management interactions (Chapman et al., 2003; Hammer et al., 2006; Messina et al., 2009). Crop models estimate growth and development using environmental resource capture and conversion efficiency concepts for radiation and water, while allowing for influences of major nutrients such as nitrogen (Muchow et al., 1990; Boote et al., 1999; Keating et al., 2003). The value of the framework to assist plant breeding depends on the extent to which the algorithms included in the model adequately capture the physiological determinants of genetic variation for adaptive traits of interest to the breeder (Tardieu, 2003; Cooper et al., 2009; Messina et al., 2009; Tardieu and Tuberosa, 2010). Fundamental physiological and genetic studies are often necessary to improve the model architecture (Tardieu, 2003; Messina et al., 2006; Chenu et al., 2008; Hammer et al., 2009; Messina et al., 2009; Bertin et al., 2010; Sinclair et al., 2010; Yin and Struik, 2010). The level of detail needed for crop growth models to be able to integrate processes across levels of organization while predicting emergent functional consequences for the organism is under debate (Hammer et al., 2006; Bertin et al., 2010). The combination of equations 1 and 2 embedded within the crop growth model provide a complete and continuous framework to model G→P relations for multiple traits that can be used to evaluate breeding strategies  (5) A fundamental attribute of this framework is that it explicitly incorporates sources of uncertainty and error in the prediction. This feature has largely been ignored in process-based G→P frameworks (Bertin et al., 2010). A simplified framework based on the concepts represented in equation 4 was applied in investigations seeking to assess the value of molecular breeding and physiological knowledge to improve the rate of genetic gain relative to conventional breeding strategies (Hammer et al., 2005; Cooper et al., 2009). Fitness landscapes The E(NK) model, as described above, has an associated trait performance landscape with a topography and ruggedness that varies in response to the complexity at the trait genetic model via combinations of E, N, and K (Kauffman, 1993; Cooper and Podlich, 2002; Cooper et al., 2005). The concept of a fitness landscape, which originated in evolutionary biology, introduces the notion of a ‘potential surface’ or function underlying the dynamics of evolution. This landscape metaphor was first proposed by Sewall Wright (1932) as a framework to advance the theoretical grounds for evolution when assumptions about non-additive variance, pleiotropy, and epistasis are included in the genetic model. The assumptions of additivity were fundamental to the paradigm at the time, exemplified by the infinitesimal and geometric models proposed by Fisher (Orr, 2005; Gavrilets, 2004), which are determinants of a smooth and single peak landscape. Wright, contrary to Fisher, envisioned that pleiotropy and epistasis would generate design constraints leading to rugged fitness landscapes, which, in turn, would determine the phases of adaptation and speciation as populations moved across the landscapes and reached different peaks in a complex surface (Gavrilets, 2004). The implications of rugged landscapes for plant breeding are consequential as they suggest trajectories of genetic improvement conditioned by the presence of hills (high performance) and valleys (low performance). In rugged trait performance landscapes, the response to selection is variable and conditioned upon population size, position of breeding populations in the landscape and genetic variation for adaptive traits that determine the number and height of peaks accessible to individuals in the population subjected to cycles of selection by the plant breeder (Cooper et al., 2002, 2005). Despite the important implications of rugged trait performance landscapes in plant breeding and the associated potential benefits, the creation of biologically sound landscapes for use by the breeder has been elusive. Recent work with RNA folding models has provided some views of the structure of fitness landscapes for a well-defined system (Fontana, 2002). Advances in computing capacity, data management, graph theory, and G→P models open the opportunity to generate the first partial views of trait performance landscapes for crops, as discussed in this paper. This information offers the breeder the opportunity to examine the positions of the genotypes that comprise the reference germplasm pool under consideration in the breeding programme relative to the accessible peaks in performance, and to gain insights into the topography of the adjacent but unexplored G→P space. The complexity of trait performance landscapes and the inability of breeders to explore fully a vast genotype space, that remains mostly unobservable to the breeder, has been identified as a major constraint to the design of breeding programme strategies for complex traits to achieve genetic progress (Cooper and Hammer, 1996). The problem of a limited capacity empirically to explore a vast and unobserved G→P space will not be resolved in the foreseeable future. However, recent theoretical work (Chapman et al., 2003; Podlich et al., 2004; Hammer et al., 2005; Chenu et al., 2009; Messina et al., 2009) suggests that breeding simulation (Podlich and Cooper, 1998) applied to trait performance landscapes can enable breeders to explore the unknown G→P space in silico. This G→P space can only be created by means of G→P models grounded on physiological and genetic principles. The predicted breeding programme trajectories become testable genetic hypotheses. Breeders can create genotypes to test these hypotheses by means of creating genotypic novelty through strategic sampling of germplasm, recombination and segregation, evaluation in environments representative of the TPE, and by genotypic and phenotypic selection. This paper documents the first steps towards generating the evidence to answer the question, ‘Does a knowledge-based approach that creates greater G→P knowledge than we currently have help manage resources and accelerate genetic progress relative to what we can achieve using current technologies?’ From fitness landscapes to yield–trait performance landscapes for plant breeding applications The objective of a modelling framework developed and structured around G→P models, landscapes, and breeding simulation is to enable plant breeders to achieve greater genetic progress relative to what could be achieved using molecular breeding and phenotypic selection approaches. Considering this context, two necessary conditions are immediate and common to other breeding technologies: (i) the framework should be relevant to the germplasm developed by the breeder; and (ii) be operational in the context an ongoing breeding programme. In addition, the proposed framework must have prediction capabilities that apply beyond the set of environments and germplasm sampled in the breeding programme, a distinct characteristic from current statistically-based approaches, and should enable continuous improvements concomitant with the generation of knowledge in each cycle of selection (Podlich et al., 2004). Figure 1 is a graphical representation of the concept map after Casti (1997). The real world component represents the ongoing breeding programme with the key processes represented: (i) testing of a sample of genotypes in a sample of environments; (ii) hypothesis driven ad hoc experimentation; and (iii) selection decisions (Cooper and Hammer, 1996). The structure and operation of the breeding programme and the data produced within the breeding programme inform modelling activities. The mathematical world represents the ongoing mathematical modelling and analyses of trait genetic architecture and G→P prediction. Trait modelling informs phenotyping in the real world and ad hoc experimentation. Exercising the models, genetic, statistical, biophysical, or a combination of these, and the analyses of results inform selection decisions and put forward genetic hypotheses for testing within the breeding programme. Arrows emphasize the iterative nature of this modelling-testing process, which is fundamental to the continuous generation of knowledge concomitant with the changes in the breeding programme, the environment, and the germplasm over cycles of improvement. Map-As-You-Go implements this iterative model building for application in molecular breeding (Podlich et al., 2004). Fig. 1. View largeDownload slide Concept map of the iterative gene-to-phenotype modelling framework. Fig. 1. View largeDownload slide Concept map of the iterative gene-to-phenotype modelling framework. `The framework considered in this paper (herein the G→P framework) is an implementation of the iterative cycle depicted in Fig. 1 and integrates the knowledge creation capabilities of the crop model within the Map-As-You-Go methodology. The framework has five key components, which can be considered in the form of an algorithm, summarized below. (i) Define components for model Γ Understand the physiological basis of yield and adaptation in the TPE relevant to the germplasm in the breeding programme (Cooper and Hammer, 1996). Represent this knowledge as meta-processes (equation 3; see Tardieu, 2003, for concept details and an example). Design a high-throughput precision phenotyping strategy that could be executed within the breeding programme to characterize genetic variation for key components of the meta-processes (see Tardieu and Tuberosa, 2010, for a recent review on the utilization of high throughput phenotyping platforms). Integrate a suitable trait model within a crop growth and development framework (Chenu et al., 2008; Hammer et al., 2009; Messina et al., 2009; Sinclair et al., 2010; Tardieu and Tuberosa, 2010). (ii) Develop the model Γ Undertake genetic modelling to define the genetic architecture of adaptive traits. We refer here to parameters in physical or control equations (Tardieu, 2003) in the context of the NK model (e.g. equation 4; Cooper et al., 2005). The recent advances in genotyping and phenotyping, together with suitable mixed model methodology to map traits, enables the estimation of allele effects for components of trait genetic architecture accessible through mapping in suitable populations (Boer et al., 2007; Cooper et al., 2009). (iii) Create performance landscapes for the defined model Γ Conduct environmental characterization and define the TPE for the geography and management of interest (Chapman et al., 2000; Löffler et al., 2004). Using models developed in (ii) define the range of values for adaptive traits. Define the factorial experiment for all combinations of trait levels (G), environments (E), and management (M). Conduct in silico experiments using a model Γ as represented by equation 2. Represent the results in a graphical form to enable decision-making for the selection and creation of new populations of genotypes. Graphical representation of yield–trait performance landscapes to support decisions in the plant breeding programme cycle is a focus of this paper. (iv) Breeding modelling and simulation for Γ Represent the target breeding programme in an abstract form suitable for simulation (Podlich and Cooper, 1998). Define the TPE as in (iii), germplasm as gene–genotype frequencies and organization of alleles within a reference population structure, genetic architecture of each trait as in (ii), and the breeding strategy (i.e. number and size of populations, testers, etc.). Undertake breeding simulation utilizing a specific parameterization of equation 5 for modelled and detected components, and an ensemble approach to accommodate stochastic components associated with undetected components and sources of error. The latter two stochastic terms affect G→P prediction, and the trajectories that are realized exploring the adjacent possible genetic space in silico, in a manner that can be anticipated from the behaviour of dynamic complex systems (Kauffman, 1993; Hammer et al., 2005; Cooper et al., 2005). (v) Predict and evaluate Predict the performance of genotypes that can be created within the context of the reference germplasm, create new genotypes to test genetic hypotheses, evaluate predictions, advance the cycle of the breeding programme, and refine biophysical and genetic models [i.e. return to step (i) above] when necessary. Application to yield improvement of maize under drought stress The G→P framework described in section (ii) was applied within an operational breeding programme, which has as an objective to improve the levels of drought tolerance in maize beyond the levels currently existing in Pioneer elite germplasm deployed in the Western Corn-Belt. This breeding programme seeks genetic improvement in two complementary heterotic groups, hereafter referred to as heterotic groups 1 and 2 (Duvick et al., 2004). Doubled haploid (DH) lines in test-cross combination are evaluated in managed drought environments under stress and in a sample of environments in the Western Corn-Belt taken to represent the TPE (Cooper, 2009). Irrigation management is designed at key locations to expose genetic variation for target adaptive traits; irrigation management is informed by computer modelling and simulation as part of the iterative modelling and testing cycle (Fig. 1). Newly developed lines are evaluated and selected on performance across multiple locations in one or more years, after which selected lines are used to create new hybrid combinations and cycled back to the germplasm pool to be used as parents for future breeding cycles. The breeding programme implements reciprocal recurrent selection with molecular enhanced pedigree selection operating within the two heterotic groups. Selection decisions are informed by predictions based on G→P statistical (equation 1) and biophysical models (equation 4) combined with yield–trait performance landscapes and breeding simulation (Podlich et al., 2004; Cooper et al., 2005; Messina et al., 2009). Further evaluation and selection takes place in an increasing number of locations and hybrid combinations. The breeding programme was formalized within the QU-GENE software (Podlich and Cooper, 1998) to enable breeding simulation. Environmental characterization for drought stress is conducted using the EnClass® system (Löffler et al., 2004) based on patterns of water deficit characterized by the ratio between water supply and demand (Messina et al., 2009). For demonstration purposes, illustrative results will be presented for combinations of well-watered irrigated (ETWW) and drought-stress (ETDR) conditions. Trait understanding and modelling The APSIM crop growth modelling framework (Keating et al., 2003) provided the organizing principle to assess the needs for trait modelling, and the simulation platform for trait integration that enables the mapping of trait phenotypes to yield performance (used here in place of fitness as applied in the evolutionary context) in the TPE. Trait modules were developed to enable APSIM-maize to simulate the effects of leaf angle, root angle, ear growth, and competition among sinks on light use efficiency, light interception, water uptake, biomass allocation, secondary traits, and yield (Hammer et al., 2009; Messina et al., 2009). Root angle (RA) and leaf angle (LA) were considered as quantitative overall descriptors of canopy and root system architecture. The effects of variation in canopy architecture were modelled by means of a layered canopy photosynthesis model (Hammer and Wright, 1994). Leaf angle effects on light interception and distribution among canopy layers were modelled after Duncan et al. (1967). Potential water uptake rate in each soil layer was modelled using an exponential decay function (Passioura, 1983). Root angle was used to represent root exploration of each soil layer that modulates the extent of potential water uptake that is realized (Hammer et al., 2009). The root angle parameter determines the shape of the cross-section plane perpendicular to the planting row (ranging from semi-circular to semi-elliptical) that it is then compared with the area of the soil layer. Algorithms were incorporated to simulate kernel numbers in relation to ear growth and sink dynamics (Messina et al., 2009). A threshold in ear mass (EBT), which determines biomass allocation to the ear, ear growth, and the number of fertilized spikelets at flowering time (Uhart and Andrade, 1995; Borras et al., 2007, 2009), connects development to growth in such a manner that changes in ear growth can affect the time to silking, the anthesis–silking interval (discussed in Hammer et al., 2009), and rooting depth (Keating et al., 2003). Hammer et al. (2009) describe the model parameterization process and verified the model to simulate faithfully diurnal patterns of photosynthesis and response to leaf angle and leaf area index. Further testing consisted in contrasting simulated and observed yield responses to irrigation applied at different times around flowering time and during grain fill (Messina et al., 2009). Genetic architecture of adaptive traits Phenotyping strategies were designed and executed using Pioneer proprietary technologies to quantify the parameters in the process equations. However, protocols for model parameterization for the processes considered in this work have been developed and described elsewhere (Muchow and Carberry, 1989; Edmeades et al., 1993; Uhart and Andrade, 1995; Birch et al., 1998a, b, c; Vega et al., 2001; Echarte et al., 2004; Monneveux et al., 2006; Padilla and Otegui, 2005; Borras et al., 2007, 2009). Three-hundred-and-fifty DH lines from a single cross representing one heterotic group were genotyped and phenotyped. Phenotyping was conducted to characterize the DH lines in a test-cross combination for thermal time from emergence to tassel initiation (ETI), leaf appearance rates (LAR), canopy leaf area distribution (LAD), maximum number of ovules in the apical ear (ME), and ear biomass threshold (EBT). All DH lines were evaluated for the traits as test-cross combinations with a single tester selected from the complementary heterotic group. Genome scans were applied to trait data using QTL mapping methodology within a mixed model framework following the methodology described by Boer et al. (2007). The results were used to describe the genetic architecture of the traits and to parameterize the NK model. The simplest genetic architecture suggested by the data, used here for the purpose of demonstrating the framework, was an additive genetic model with three QTL with equal effect sizes for each trait and two alleles at each locus (NK=3:0). This genetic architecture is assumed for all traits and does not consider epistasis or pleiotropic effects at the individual trait QTL level. It is noted, however, that GEI, trait interactions and pleiotropic effects on yield for the trait QTL are possible and emergent properties that result from the interaction of the adaptive traits as they influence crop growth, development, and yield within the environmental context. The additive three QTL models for each trait define 27 genotypes but only seven expression states for each trait. Given the additive genetic model within a trait, different combinations of alleles for the three QTL correspond to the same trait expression state. If one considers for a given QTL A that allele A results in increased expression (+) relative to the allele a (–), then the expression state for the genotype is defined by the sum across the QTL of the + alleles. For example, the allelic combinations AAbbcc, aaBBcc, aabbCC, AaBbcc, aaBbCc, and AabbCc have the same expression state. As a convention for the additive QTL model, the highest expression state value 1 was defined to correspond to the genotype AABBCC and the lowest value 0 to aabbcc. Extensions to the additive genetic model considered here are possible and their main impact is on the computational load and the outcomes of the simulation strategies as discussed below. Simulating performance for quantitative traits Two in silico factorial experiments were conducted in a high performance computing environment. For both experiments there was one reference genotype population for each of two heterotic groups, referred to hereafter as heterotic groups 1 and 2. The simulation experiment for heterotic group 1 was described by Messina et al. (2009). The simulation experiment for heterotic group 2 was based on five adaptive traits: ETI, LAR, LAD, ME, and EBT. The genetic models for the five traits define 1.4×107 genotypes. Because of the use of the approximation of equal additive QTL effects within each trait with three QTL per trait, only 1.6×104 expression states were defined. Maize phenotypes were simulated using APSIM-Maize for each expression state, 59 years of weather data for Belgrade and Holdrege, Nebraska, US. Four soil water initial conditions for a loam soil with a soil water-holding capacity of 508 mm were estimated using a long-term simulation following a procedure outlined by Hammer et al., 2009. The planting date was set to 5 May for all years and plant population to 8.6 plants m−2. Model parameters not phenotyped for DH lines in a test-cross combination in the mapping population were set to the values estimated for two reference hybrids for both heterotic groups. Representing landscapes for plant breeding applications Different graphical and numerical representations have been proposed to represent trait landscapes (Wright, 1932; Kauffman, 1993; Cooper and Podlich, 2002; Fontana, 2002; Gavrilets, 2004; Cooper et al., 2009). Gavrilets (2004) provides a clear synthesis in which he classifies these into three landscape types. A first class is the fitness of gene combinations, the original concept proposed by Wright (1932), in which each genotype is assigned a value of fitness. For the application of the landscape concept in breeding, this class of landscape is referred to as the trait performance of the genotypes defined by the QTL combinations. A second type of landscape that relates the gene (QTL) allele frequencies to the mean fitness (i.e. yield performance) of the reference population can be constructed if the individual fitness (i.e. yield) values for each genotype and the population structure are known. A third type of landscape is for quantitative traits in which trait values are related to fitness (here yield performance). This third type forms the basis for representing the performance landscapes for plant breeding applications discussed in this paper. Herein this yield–trait performance landscape view is referred to as the gene-to-phenotype plot, or GP plot for shorthand, since there is a G→P mapping of genetic variation for genes (QTL) to yield. The connection between genes (QTL) and genes is via their modelled influence on the physiological traits defined by the function Γ. The GP plot represents a landscape as a set of conditional cross-sections for grain yield in the G and P dimensions (Fig. 2A–D). Each cross-section is a projection Rn→R2 of n–1 traits onto a two-dimensional space determined by the trait t and yield (Fig. 2A–D). Grain yield distributions are thus conditional on the expression states for a defined trait. Taking the landscape created for root angle (Fig. 2A) in environment ETDR as an example, seven expression states were observed for root angle, arranged on the x-axis of the GP plots, and therefore seven yield distributions, depicted on the y-axis of the GP plot. The yield distribution is determined by the simulated yield variation created by all other traits conditional on each expression level for root angle. Genotype relative frequencies for yield are shown as a heat map to visualize the grain yield distributions associated with variation for the other traits (Fig. 2). Fig. 2. View largeDownload slide Yield–trait performance landscapes for a population from heteroric group 1, simulated for drought stress (ETDR) environments. Cross-sections correspond to traits: root angle (RA) (A, E, I), leaf angle (LA) (B, F, J), potential number of reproductive sinks (ME) (C, G, K), and ear mass threshold (EBT) (D, H, L). Frequencies of genotypes with a yield level increases from green to red. The central tendency in the landscape is shown as a black line. Individual or clusters of genotypes are shown as black dots with sizes proportional to the frequency of the genotype in the population (E–L). The population mean breeding trajectory over cycles of selection is shown in blue (I–L). Fig. 2. View largeDownload slide Yield–trait performance landscapes for a population from heteroric group 1, simulated for drought stress (ETDR) environments. Cross-sections correspond to traits: root angle (RA) (A, E, I), leaf angle (LA) (B, F, J), potential number of reproductive sinks (ME) (C, G, K), and ear mass threshold (EBT) (D, H, L). Frequencies of genotypes with a yield level increases from green to red. The central tendency in the landscape is shown as a black line. Individual or clusters of genotypes are shown as black dots with sizes proportional to the frequency of the genotype in the population (E–L). The population mean breeding trajectory over cycles of selection is shown in blue (I–L). Because the model presented in equation 2 is deterministic, a given genotype map to yield (y-axis) through the value determined by the genetic model considered for the trait phenotype (x-axis). The landscape represented by the GP plot is thus a projection of the yield–trait performance landscape for gene combinations, as conceived for fitness by Wright (1932), in two dimensions. Individuals or clusters of individuals in a breeding population can be represented as points (here shown as black dots) in the landscape with the size of the dot proportional to the genotype frequency (Fig. 2E–L). The dots in Fig. 2E–H refer to individuals in the initial population, while those in Fig. 2I–L represent groups of individuals at the end of selection. By representing in the GP plot both the frequency of the set of genotypes in a breeding population at any given stage in the breeding programme cycle and the breeding trajectory (blue line traces the average for the population at each cycle of selection for a single breeding simulation run), the GP plot is a variant of the performance (fitness) landscape as mean performance (fitness) of populations (Fig. 2I–L). The GP plot views are constructed for each environment type in the TPE to describe how the landscape shape changes in response to environment change (e.g. Fig. 3A compared with E). Fig. 3. View largeDownload slide Yield–trait performance landscapes for a population from heteroric group 1, simulated for drought-stress (ETDR) and well-watered (ETWW) environments. Cross-sections correspond to traits: root angle (RA) (A–E), leaf angle (LA) (B–F), potential number of reproductive sinks (ME) (C–G), and ear mass threshold (EBT) (D–H). The central tendency in the landscape is shown as a black line. The population mean breeding trajectory over cycles of selection is shown in blue and individual or clusters of genotypes are shown as black dots with sizes proportional to the frequency of the genotype in the population. Frequencies of genotypes within a yield level increases from green to red. Fig. 3. View largeDownload slide Yield–trait performance landscapes for a population from heteroric group 1, simulated for drought-stress (ETDR) and well-watered (ETWW) environments. Cross-sections correspond to traits: root angle (RA) (A–E), leaf angle (LA) (B–F), potential number of reproductive sinks (ME) (C–G), and ear mass threshold (EBT) (D–H). The central tendency in the landscape is shown as a black line. The population mean breeding trajectory over cycles of selection is shown in blue and individual or clusters of genotypes are shown as black dots with sizes proportional to the frequency of the genotype in the population. Frequencies of genotypes within a yield level increases from green to red. Analyses of performance landscapes Two approaches were used to study the properties of the yield–trait performance landscapes using the GP plots. The first approach focuses on measuring static properties of the landscape for a given environment and population and seeks to describe the geometric properties of the landscape. In the second approach, dynamic properties were studied by simulating the trajectories of breeding populations over cycles of selection. The geometric properties considered here include local and global maxima and minima, trends in central tendency, trends in maxima and minima, ridges, valleys, and plateaus. The global maximum is identified as the maximum value in the y-axis across all frequency distributions for a given trait conditioned to G. The global maximum has coordinates in n dimensions, one for each trait. For a given genetic and environment context the global maximum could be conceived as an ideotype conditional to the germplasm and the TPE. In the example presented in Fig. 3A–D for drought-stress environments, this maximum corresponds to genotype QTL AABBCC ddeeff GGHHII JJKKLL. Local maxima (minima) correspond to each maximum (minimum) value for yield conditional on each expression state and G. Local maxima provide information about the potential for yield improvement for a population under selection with access to a subset of the peaks on the landscape. Local minima emphasize the role of the traits and the expression state on the probability of yield failure. Thus, knowledge of these features of the landscape helps assess the role of the trait in determining yield stability. For example, increasing rooting depth associated with increasing frequency of (+) alleles at loci A, B, and C increases the minimum yield under drought stress. This breeding strategy has been advocated and genetic variation for root angle has been documented (Sinclair and Muchow, 2001; Tuberosa et al., 2002; Giuliani et al., 2005; Manschadi et al., 2006; Messina et al., 2009). Variations in the location of minima and maxima for any given trait provide insights to the interpretation of the yield distributions for other traits. Under well-watered conditions, the predominantly flat landscape (green area) across expression states for RA is largely determined by the variation in minimum yields for ME (Fig. 3G). Increasing the potential number of sinks underpinned yield improvement in Argentina (Echarte et al., 2004) and it has been proposed as a possible path towards yield improvement under well-watered conditions in the US Corn-Belt (Messina et al., 2009). When local minima are zero the GP plot provides a view of what Gavrilets (2004) refers to as holey landscapes, i.e. landscapes with valleys or holes defined by unviable genotypes, zero yields in this case. The yield distribution for a given expression state as shown in Fig. 3, can be viewed as analogous to a Poincare section of a multi-dimensional space within which trajectories cross the landscape at that point in G→P space. Local maxima and minima define the size of this section. The landscape for ME (Fig. 3G) provides an example of a tight cross-section for genotype GGHHII that contrasts with a wide section for genotype gghhii. While population trajectories are strongly constrained by the genotype GGHHII, they are not by the genotype gghhii. At the trait expression state associated with genotype gghhii, other traits have more influence on yield performance than ME; this is reflected by the yield range difference between these two genotypes, mainly determined by differences among genotypes in local minima (Δ=200 g m−2; Fig. 3G) rather than differences in local maxima (Δ=100 g m−2). The size of the Poincare section is an important descriptor of the landscape. This statistic could be used to help the breeder anticipate the expected genetic variation in a breeding population and potential changes that can be expected over cycles of selection. Trends in local maxima and central tendency statistics (Fig. 3, black line) help identify ridges and valleys, and provide a first view of gradients that can shape the path of populations under selection pressure towards local or global maxima. The identification of valleys is of particular interest to the breeder because they provide a first assessment about the possibility to attain maximum yields given the position of the population in the landscape and the need for populations to move downhill before reaching a new position that provides access to the global maximum or higher local maxima. Figure 3 provides two examples (Fig. 3B, D) that illustrate the presence of saddles that can constrain the population access to the global maxima. Figure 3B suggests that increased levels of drought tolerance are feasible by either increasing or decreasing leaf inclination angle. Both strategies lead to improved water status during reproductive stages associated with reduced growth and water conservation. But only reduced growth due to lower radiation use efficiency and lower average leaf inclination angle leads to the global maxima under drought stress. However, this strategy has the associated cost of reduced productivity under well-watered conditions (Fig. 3F). Water conservation strategies that improve the partitioning of water use between vegetative and reproductive stages (Sinclair et al., 1984; Condon et al., 2004) and opportunities for breeding for specific adaptation in the western Corn Belt (Messina et al., 2009) have been discussed previously. A second approach to study performance landscapes is through the study of the dynamics of populations over cycles of selection. Breeding simulation enables an estimation of the length of the adaptive walks via higher yielding neighbours from a starting reference population towards a local maxima, the overall complexity of the landscape through the quantification of the correlation structure of the landscape (Kaufmann, 1993; Cooper et al., 2005), how this complexity affects rates of genetic gain over cycles of selection (Cooper et al., 2009), and the dependency of such breeding trajectories on the initial structure of the population. Sensitive dependence on initial conditions is a fundamental characteristic of complex systems (Lorenz, 1995; Kauffman, 1993). It can be anticipated that this feature of the breeding system will have some importance for the plant breeder in the use of prediction methods to enable genetic improvement of complex traits for different crosses within different reference populations of genotypes and different target environments. Figure 4 presents the results of a simulation experiment designed to study trajectories of maize populations and how these are affected by changes in the structure of the initial breeding populations. Results for the last run from an ensemble of 50 simulations are shown for cycle 8 of selection under drought stress conditions (ETDR). Drought environments were sampled over time from the classification of environments described in Messina et al. (2009). When the population structure was defined by the frequency of (+) alleles and set at 0.5 within the reference population for all traits (control simulation experiment), yield increased with increasing rooting depth (Fig. 4A), decreasing leaf angle and radiation use efficiency (Fig. 4B), and decreasing EBT (Fig 4D). The reduction in EBT was associated with simulated lower anthesis silking interval (ASI) (data not shown), as discussed by Campos et al. (2004) and Borras et al. (2009). This result from simulation conforms to empirical data that show a correlated reduction in ASI with increased ear growth and yield (Edmeades et al., 1993; Campos et al., 2004; Borras et al., 2007, 2009). Reduction in EBT and increased partitioning to the ear has been implicated in the genetic improvement for yield (Bolaños and Edmeades, 1993,b; Edmeades et al., 1993; Echarte et al., 2004; Hammer et al., 2009; Messina, 2009). Increasing the frequency of the (+) alleles in the reference population for EBT from 0.5 to 0.55 led to a bifurcation of trajectories with populations selected towards any of the two highest local maxima (data not shown). Further increases in the frequency of the (+) alleles from 0.55 to 0.8 resulted in the population following a trajectory in the opposite direction from the one shown for the control run (compare Fig. 4D with H). Higher values of EBT increased ASI but also increased rooting depth associated with delayed flowering time, resulting in a trade-off between resource use efficiency and resource capture. Fig. 4. View largeDownload slide Breeding simulation applied to the study of performance landscapes. Simulations are for a breeding population from heteroric group 1 and selection in drought-stress environments (ETDR). Results are shown for cycle 8 of selection. Cross-sections correspond to traits: root angle (RA) (A, E, I, M), leaf angle (LA) (B, F, J, N), potential number of reproductive sinks (ME) (C, G, K, O), and ear mass threshold (EBT) (D, H, L, P). The central tendency in the landscape is shown as a black line. The population mean breeding trajectory over cycles of selection is shown in blue and individual or clusters of genotypes are shown as black dots with sizes proportional to the frequency of the genotype in the population. Frequencies of genotypes with a yield level increases from green to red. Panels (A–D) correspond to a control simulation experiment where (+) allele frequencies were set at 0.5 in the cycle 0 reference population of genotypes. Three simulation experiments where the frequency of (+) alleles in the cycle 0 reference population of genotypes increased from 0.5 to 0.8 for the trait EBT (E–H), from 0.5 to 0.8 for the trait LA (I–L) and from 0.5 to 0.9 for the trait LA (M–P). Fig. 4. View largeDownload slide Breeding simulation applied to the study of performance landscapes. Simulations are for a breeding population from heteroric group 1 and selection in drought-stress environments (ETDR). Results are shown for cycle 8 of selection. Cross-sections correspond to traits: root angle (RA) (A, E, I, M), leaf angle (LA) (B, F, J, N), potential number of reproductive sinks (ME) (C, G, K, O), and ear mass threshold (EBT) (D, H, L, P). The central tendency in the landscape is shown as a black line. The population mean breeding trajectory over cycles of selection is shown in blue and individual or clusters of genotypes are shown as black dots with sizes proportional to the frequency of the genotype in the population. Frequencies of genotypes with a yield level increases from green to red. Panels (A–D) correspond to a control simulation experiment where (+) allele frequencies were set at 0.5 in the cycle 0 reference population of genotypes. Three simulation experiments where the frequency of (+) alleles in the cycle 0 reference population of genotypes increased from 0.5 to 0.8 for the trait EBT (E–H), from 0.5 to 0.8 for the trait LA (I–L) and from 0.5 to 0.9 for the trait LA (M–P). A third simulation experiment was defined such that the breeding population was positioned on the landscape at the centre of the valley identified in the cross-section created for LA (Fig. 4I–L). Two key results of these two simulation experiments are the reduction in the rate of genetic gain (compare yields after 8 cycles of selection as indicated by the end of the trajectory show in blue in Fig. 4A with I) and the change in the trajectory in EBT (compare Fig. 4D with L) in response to the increase in initial frequency of (+) alleles for LA. Further increases in the frequency of (+) alleles for LA led to yield stagnation due to convergence of the population on a local maxima determined by the highest average leaf inclination angle (Fig. 4N) and a population that largely remains within the valley observed for EBT (Fig. 4P). Application to selection decisions The G→P framework was applied to enable selection decisions. For all breeding lines with acceptable agronomic phenotypes (i.e. reduced lodging), the merit of the inbred lines was determined from the lines relative position in the performance landscape, the predicted performance, and the potential to contribute to further yield improvement. Inbred lines in test-cross combinations and reference hybrids were characterized for five adaptive traits. A performance landscape was created for a range of quantitative traits that encompass both the breeding population and reference hybrids and was extended to include an adjacent possible genotype space (Fig. 5A–E). The predicted test-cross yield performance values of the inbred lines for their trait combinations were then positioned in this landscape (Fig. 5F–J) and the plausible trajectories of this population in G→P space, based on selection under drought stress, were investigated via simulation (Fig. 5K–O). Fig. 5. View largeDownload slide Yield–trait performance landscapes applied to plant breeding. Simulated performance for a breeding population from the heteroric group 2 and representation of landscape (A–E). Phenotyping of inbred lines and projection of performance in the performance landscape (F–J). Simulation of breeding strategy for a specific breeding population and projected trajectory over cycles of selection (K–O). Selection experiments were conducted by sampling drought-stress environments (ETDR). Results are shown for cycle 4 of selection (K–O). Cross-sections correspond to traits: potential number of reproductive sinks (ME) (A, F, K), ear mass threshold (EBT) (B, G, L), thermal time from emergence to flower initiation (ETI) (C, H, M), leaf appearance rate (LAR) (D, I, N), and leaf area distribution (LAD) (E, J, O). The central tendency in the landscape is shown as a black line. The population mean breeding trajectory over simulated cycles of selection is shown in blue and individual or clusters of genotypes are shown as black dots with sizes proportional to the frequency of the genotype in the population. Frequencies of genotypes with a yield level increases from green to red. Observed phenotypes for 350 inbred lines in test-cross combination are represented as red dots. Fig. 5. View largeDownload slide Yield–trait performance landscapes applied to plant breeding. Simulated performance for a breeding population from the heteroric group 2 and representation of landscape (A–E). Phenotyping of inbred lines and projection of performance in the performance landscape (F–J). Simulation of breeding strategy for a specific breeding population and projected trajectory over cycles of selection (K–O). Selection experiments were conducted by sampling drought-stress environments (ETDR). Results are shown for cycle 4 of selection (K–O). Cross-sections correspond to traits: potential number of reproductive sinks (ME) (A, F, K), ear mass threshold (EBT) (B, G, L), thermal time from emergence to flower initiation (ETI) (C, H, M), leaf appearance rate (LAR) (D, I, N), and leaf area distribution (LAD) (E, J, O). The central tendency in the landscape is shown as a black line. The population mean breeding trajectory over simulated cycles of selection is shown in blue and individual or clusters of genotypes are shown as black dots with sizes proportional to the frequency of the genotype in the population. Frequencies of genotypes with a yield level increases from green to red. Observed phenotypes for 350 inbred lines in test-cross combination are represented as red dots. The G→P space of the target germplasm is characterized by a complex and holey landscape. In contrast with previous examples (Figs 3, 4), the presence of zero yields are common when these genotypes are exposed to severe drought stress and largely determined by genotypes with a large expression of EBT. Presence of two saddles (see cross-section corresponding to the thermal time from emergence to tassel initiation; Fig. 5C) and the absence of a clear single trait determinant of yield bring complexity to the landscape. However, trends could be indentified and indicate yield under drought stress increasing with increasing ME, and decreasing EBT, ETI, LAR, and LAD. Together, the trends in these traits indicates that yield improvement for this germplasm could be attained by reducing canopy size (both reduction in LAD, ETI, and LAR) and time to flowering, and by increasing the potential number of reproductive sinks and the allocation of resources to the ear. The strategy has components of stress avoidance (early flowering), water conservation and optimal partitioning of water use between pre-anthesis and post-anthesis (early flowering and small canopy size), and stress resistance by means of improved kernel set. The empirical evidence indicates that there is little room for further improvement of drought resistance for the characterized breeding material by further change in the expression of EBT; individuals in this population have already explored the edge of the G→P space (Fig. 5G). Breeding simulation suggests that opportunities for further yield improvements exist. Earliness and smaller canopy size can improve further the water use pattern provided seed fill duration remains constant (Fig. 5M, O). Bolaños et al. (1993) reported similar changes in development and canopy architecture with cycles of selection in CIMMYT drought-tolerant germplasm. Increasing the potential number of reproductive sinks can contribute further to yield improvement, but only after improvement in the water use pattern and stress avoidance were in place (compare yields after 8 cycles of simulated selection indicated by the ends of the blue lines; Fig. 5K and M). Discussion and remarks Plant breeders and physiologists recognize drought tolerance and yield as complex traits (Cooper and Hammer, 1996; Bruce et al., 2002; Campos et al., 2004; Duvick et al., 2004, Ribaut et al., 2004). Many studies have sought to unravel the genetic and physiological basis of drought tolerance and yield improvement in field crops (Sinclair et al., 1984; Ludlow and Muchow, 1990; Condon et al., 2004; Ribaut, 2006). The difficulty of studying complex systems in multiple trait dimensions with the available methodologies has forced discussions about the value of physiological traits to improved drought resistance to orthogonal contrasts. This study has implemented a theoretical framework grounded on realistic G→P models and fitness landscapes within a breeding programme to help breeders deal with the complexity of how these traits interact with each other during crop growth and development to determine yield within an environmental context. Previous studies advocated for the implementation of similar frameworks but these have remained on theoretical grounds (Chapman et al., 2003; Cooper et al., 2005; Hammer et al., 2005; Chenu et al., 2009; Messina et al., 2009). This paper documents the first application of a G→P framework within an operational breeding programme. The application of the framework to the genetic improvement of drought tolerance in maize supported selection of DH lines with improved levels of drought tolerance based on physiological and genetic knowledge, prediction of yield within the TPE, and their predicted potential to sustain further genetic progress with additional cycles of selection. The G→P modelling framework considered in this work opens up new opportunities for selection that are difficult to consider with the current statistical approaches to molecular breeding. A limitation of current methods in plant breeding is that they provide a static view of marker–trait associations, leaving to the breeder the subjective interpretation of the interplay and emergent trade-offs among traits and the subjective prediction of the selection trajectories in genotype and phenotype space. Because the G→P model considered here is constructed on genetic and biophysical principles, emergent behaviour resulting from the interplay between the traits is captured and enables prediction for genotypes that were not created or environments in the TPE that were inadequately sampled. The framework proved useful to enable selection decisions due to previous work conducted on model adaptive traits relevant to the genetic variation within a set of elite breeding populations. When considering applying this technology to other populations it is easy to foresee the need to advance the crop growth model framework to incorporate or improve trait models incorporated in APSIM. Some targets for improvement are routines to enable the simulation of root growth (Tuberosa et al., 2002; Bengough et al., 2006; Whitmore and Whalley, 2009; Dupuy et al., 2010; White and Kirkegaard, 2010), the regulation of biomass and nitrogen allocation at the organ level (Horton, 2000; Echarte et al., 2004; Condon et al., 2004; Messina et al., 2009; van Oosterom et al., 2009a, b ), and the mechanisms underpinning growth maintenance under drought (Tardieu, 2003; Horton, 2000) and high temperature stress (Suwa et al., 2010). The proposed iterative nature of the framework emphasizes the need to conduct this work in the context of the target germplasm, management and TPE such that breeding objectives, physiological questions, phenotyping strategies, and genetic hypotheses are co-ordinated. This paper provides a set of empirical views of relevant yield–trait performance landscapes with projections of maize elite germplasm onto the modelled G→P space (Fig. 5). This moves us beyond those views created for the folding properties of RNA molecules (Fontana, 2002). The analyses of the landscapes provided a wealth of knowledge to inform selection decisions, phenotyping, and target traits for genetic analyses. Some of the key outcomes from the simulation studies undertaken in this work to support selection decisions are: (i) the existence of multiple paths towards yield improvement for drought tolerance within a breeding population; and (ii) the outcomes of the cycles of selection for a given breeding population are conditional on the environmental challenges and the physiological genetic background of the germplasm. Studies seeking to understand the causes of improved drought tolerance in tropical maize support this proposition and provide empirical evidence for the existence of multiple selection trajectories. Reduced canopy temperature, reduced rate of leaf senescence, and increased leaf elongation rates were shown to be correlated with improved yield under drought stress (Fischer et al., 1989). These traits could be associated with improved water status, water capture, and growth maintenance under drought stress. Chenu et al. (2009) suggest that growth maintenance under drought can improve maize adaptation to drought stress. A body of evidence supports an alternative path towards yield improvement under drought stress associated with a reduction in the anthesis–silking interval caused by increased biomass allocation to the ear, and ear growth rate per ovule (Bolaños et al., 1993; Bolaños and Edmeades, 1993b, 1996). The results obtained from the simulation experiments show the importance of both improvements in resource capture and use relative to improvements in reproductive efficiency, and how these change with cycles of selection and breeding populations (Figs 4, 5). Simulation results can thus help breeders design phenotyping strategies and prioritize trait phenotyping based on their immediate and predicted long-term effects of the trait on yield improvement. This paper showed that small changes in initial structure of the germplasm in the reference population, and types of environmental conditions used to phenotype populations and rank genotypes for selection, can have a profound impact on the selection trajectories achieved in a breeding population and thus rate of genetic gain (Fig. 4). This result is of significance as it affects how we collectively think about the utilization of QTL information and other sources of genetic diversity in plant breeding. This study demonstrated that yield performance landscapes can be rugged and that genetic gain can be constrained by such landscape structure (Fig. 4). However, the landscapes were continuous and conform to the notion that the effects of single alterations in the physiological process were dampened so that small changes were observed in plant growth and yield (Sinclair and Purcell, 2005). The absence of discontinuities in the performance landscapes suggests that the impact of a single QTL in a complex trait would be small (Figs 3, 4, 5) supporting the arguments for the utilization of genomic selection for complex traits (Heffner et al., 2009; Jannink et al., 2010). The existence of bifurcation, saddles, and rugged landscapes supports the proposition that the directions emphasized in such genomic selection can be improved by the knowledge of the yield–trait performance landscapes structure as demonstrated here (Podlich et al., 2004). The utilization of genome-wide prediction methods (van Eeuwijk et al., 2010b) applied to the prediction of yield is one path forward. The G→P framework as represented in equation 5 provides a prediction methodology that can help the breeder make selection decisions utilizing both the biological knowledge created for single QTL or other sources of genetic variation (set the genetic context that determines the trajectory of a breeding population), and genome-wide information (genome-wide prediction applied to non-modelled traits) to anticipate and assess outcomes of alternative selection strategies in the short and long term by means of landscape methodology. To enable such a comprehensive knowledge-based selection methodology will require implementation of high throughput phenotyping strategies, advanced information management systems, and methodological developments that further integrate crop growth modelling platforms with quantitative genetics models. We acknowledge the field support provided by Darren Schneider, Carla Gho, Andrea Salinas, Andres Reyes, Karen Thompson, and Neil Hausmann, and the information management support provided Zac Oler, Andy Beatty, Jason Thompson, Duhong Chen, and Tim Fast. References Bengough GA,  Bransby MF,  Hans J,  MacKenna SJ,  Roberts TJ,  Valentine TA.  Root responses to soil physical conditions: growth dynamics from field to cell,  Journal of Experimental Botany ,  2006, vol.  57 (pg.  437- 447) Google Scholar CrossRef Search ADS PubMed  Bertin N,  Martre P,  Genard M,  Quilot B,  Salon C.  Under what circumstances can process-based simulation models link genotype to phenotype for complex traits? Case study of fruit and grain quality traits,  Journal of Experimental Botany ,  2010, vol.  61 (pg.  955- 697) Google Scholar CrossRef Search ADS PubMed  Birch CJ,  Hammer GL,  Rickert KG.  Temperature and photoperiod sensitivity of development in five cultivars of maize (Zea mays L.) from emergence to tassel initiation,  Field Crops Research ,  1998, vol.  55 (pg.  93- 107) Google Scholar CrossRef Search ADS   Birch CJ,  Hammer GL,  Rickert KG.  Improved methods for predicting individual leaf area and leaf senescence in maize (Zea mays L.),  Australian Journal of Agricultural Research ,  1998, vol.  49 (pg.  249- 262) Google Scholar CrossRef Search ADS   Birch CJ,  Rickert KG,  Hammer GL.  Modelling leaf production and crop development in maize (Zea mays L.) after tassel initiation under diverse conditions of temperature and photoperiod,  Field Crops Research ,  1998, vol.  58 (pg.  81- 95) Google Scholar CrossRef Search ADS   Boer MP,  Wright D,  Feng L,  Podlich DW,  Luo Lang,  Cooper M,  van Eeuwijk FA.  A mixed-model quantitative trait loci (QTL) analysis for multipleenvironment trial data using environmental covariables for QTL-by-environment interactions, with an example in maize,  Genetics ,  2007, vol.  177 (pg.  1801- 1813) Google Scholar CrossRef Search ADS PubMed  Bolaños J,  Edmeades GO.  Eight cycles of selection for drought tolerance in lowland tropical maize. I. Responses in grain yield, biomass, and radiation utilization,  Field Crops Research ,  1993, vol.  31 (pg.  233- 252) Google Scholar CrossRef Search ADS   Bolaños J,  Edmeades GO.  Eight cycles of selection for drought tolerance in lowland tropical maize. II. Responses in reproductive behavior,  Field Crops Research ,  1993, vol.  31 (pg.  253- 268) Google Scholar CrossRef Search ADS   Bolaños J,  Edmeades GO.  The importance of the anthesis–silking interval in breeding for drought tolerance in tropical maize,  Field Crops Research ,  1996, vol.  48 (pg.  65- 80) Google Scholar CrossRef Search ADS   Bolaños J,  Edmeades GO,  Martinez L.  Eight cycles of selection for drought tolerance in lowland tropical maize. III. Responses in drought-adaptive physiological and morphological traits,  Field Crops Research ,  1993, vol.  31 (pg.  269- 286) Google Scholar CrossRef Search ADS   Boote KJ,  Jones JW,  Hoogenboom GH.  Peart RM,  Curry RB.  Simulation of crop growth: CROPGRO Model,  Agricultural systems modelling and simulation ,  1998 New York Marcel Dekker(pg.  651- 693) Borras L,  Astini JP,  Westgate ME,  Severini AD.  Modeling anthesis to silking in maize using a plant biomass framework,  Crop Science ,  2009, vol.  49 (pg.  937- 948) Google Scholar CrossRef Search ADS   Borras L,  Westgate ME,  Astini JP,  Echarte L.  Coupling time to silking with plant growth rate in maize,  Field Crops Research ,  2007, vol.  102 (pg.  73- 85) Google Scholar CrossRef Search ADS   Bruce WB,  Edmeades GO,  Barker TC.  Molecular and physiological approaches to maize improvement for drought tolerance,  Journal of Experimental Botany ,  2002, vol.  53 (pg.  13- 25) Google Scholar CrossRef Search ADS PubMed  Campos H,  Cooper M,  Habben JE,  Edmeades GO,  Schussler JR.  Improving drought tolerance in maize: a view from industry,  Field Crops Research ,  2004, vol.  90 (pg.  19- 34) Google Scholar CrossRef Search ADS   Casti JL. ,  Would-be worlds ,  1997 United States John Wiley and Sons Chapman S,  Cooper M,  Podlich D,  Hammer G.  Evaluating plant breeding strategies by simulating gene action and dryland environment effects,  Agronomy Journal ,  2003, vol.  95 (pg.  99- 113) Google Scholar CrossRef Search ADS   Chapman SC,  Cooper M,  Hammer GL,  Butler D.  Genotype by environment interactions affecting grain sorghum. II. Frequencies of different seasonal patterns of drought stress are related to location effects on hybrid yields,  Australian Journal of Agricultural Research ,  2000, vol.  50 (pg.  209- 222) Google Scholar CrossRef Search ADS   Chenu K,  Chapman SC,  Hammer GL,  Mclean G,  Ben Haj Salah,  Tardieu F.  Short-term responses of leaf growth rate to water deficit scale up to whole-plant and crop levels: an integrated modelling approach in maize,  Plant, Cell and Environment ,  2008, vol.  31 (pg.  378- 391) Google Scholar CrossRef Search ADS   Chenu K,  Chapman SC,  Tardieu F,  McLean G,  Welcker C,  Hammer GL.  Simulating the yield impacts of organ-level quantitative trait loci associated with drought response in maize: a ‘Gene-to-Phenotype’ modeling approach,  Genetics ,  2009, vol.  183 (pg.  1507- 1523) Google Scholar CrossRef Search ADS PubMed  Condon AG,  Richards RA,  Rebetzke GJ,  Farquhar GD.  Breeding for high water-use efficiency,  Journal of Experimental Botany ,  2004, vol.  55 (pg.  2447- 2460) Google Scholar CrossRef Search ADS PubMed  Cooper M. ,  Applications of molecular breeding: drought tolerance in corn ,  2009  64th Corn and Sorghum Seed Research Conference. 8–11 December. Chicago, Illinois Cooper M,  Chapman SC,  Podlich DW,  Hammer GL.  The GP problem: quantifying gene-to-phenotype relationships,  In Silico Biology ,  2002  (Available online at http://www.bioinfo.de/isb/2002/02/0013/; verified 14 August 2008) Cooper M,  Hammer GL. ,  Plant adaptation and crop improvement ,  1996 Wallingford, UK CAB International Cooper M,  Podlich D.  The E(NK) model: extending the NK model to incorporate gene-by-environment interactions and epistasis for diploid genomes,  Complexity ,  2002, vol.  7 (pg.  31- 47) Google Scholar CrossRef Search ADS   Cooper M,  Podlich DW,  Smith OS.  Gene-to-phenotype models and complex trait genetics,  Australian Journal of Agricultural Research ,  2005, vol.  56 (pg.  895- 918) Google Scholar CrossRef Search ADS   Cooper M,  van Eeuwijk FA,  Hammer GL,  Podlich DW,  Messina C.  Modeling QTL for complex traits: detection and context for plant breeding,  Current Opinion in Plant Biology ,  2009, vol.  12 (pg.  231- 240) Google Scholar CrossRef Search ADS PubMed  Duncan WG,  Loomis RS,  Williams WA,  Hanau R.  A model for simulating photosynthesis in plant communities,  Hilgardia ,  1967, vol.  38 (pg.  181- 205) Google Scholar CrossRef Search ADS   Dupuy L,  Gregory PJ,  Bengough G.  Root growth models: towards a new generation of continuous approaches,  Journal of Experimental Botany ,  2010, vol.  61 (pg.  2131- 2148) Google Scholar CrossRef Search ADS PubMed  Duvick DN,  Smith JSC,  Cooper M.  Long-term selection in a commercial hybrid maize breeding programme,  Plant Breeding Review ,  2004, vol.  24 (pg.  109- 151) Echarte L,  Andrade FH,  Vega CRC,  Tollenaar M.  Kernel number determination in Argentinean maize hybrids released between 1965 and 1993,  Crop Science ,  2004, vol.  44 (pg.  1654- 1661) Google Scholar CrossRef Search ADS   Edmeades GO,  Bolaños J,  Hernandez M,  Bello S.  Causes for silk delay in lowland tropical maize population,  Crop Science ,  1993, vol.  33 (pg.  1029- 1035) Google Scholar CrossRef Search ADS   Fischer KS,  Edmeades GO,  Johnson EC.  Selection for the improvement of maize under moisture deficits,  Field Crops Research ,  1989, vol.  22 (pg.  227- 243) Google Scholar CrossRef Search ADS   Fontana W.  Modelling ‘Evo-Devo’ with RNA,  BioEssays ,  2002, vol.  24 (pg.  1164- 1177) Google Scholar CrossRef Search ADS PubMed  Gavrilets S. ,  Fitness landscapes and the origin of species ,  2004 United States Princeton University Press Giuliani S,  Sanguineti MC,  Tuberosa R,  Belloti M,  Salvi S,  Landi P.  Root-ABA1, a major constitutive QTL, affects maize root architecture and leaf ABA concentration at different water regimes,  Journal of Experimental Botany ,  2005, vol.  56 (pg.  3061- 3070) Google Scholar CrossRef Search ADS PubMed  Hammer GL,  Chapman S,  van Oosterom E,  Podlich DW.  Trait physiology and crop modelling as a framework to link phenotypic complexity to underlying genetic systems,  Australian Journal of Agricultural Research ,  2005, vol.  56 (pg.  947- 960) Google Scholar CrossRef Search ADS   Hammer G,  Cooper M,  Tardieu F,  Welch S,  Walsh B,  van Eeuwijk F,  Chapman S,  Podlich D.  Models for navigating biological complexity in breeding improved crop plants,  Trends in Plant Science ,  2006, vol.  11 (pg.  1360- 1385) Google Scholar CrossRef Search ADS   Hammer GL,  Dong Z,  McLean G,  Doherty A,  Messina C,  Schussler J,  Zinselmeier C,  Paszkiewicz S,  Cooper M.  Can changes in canopy and/or root systems architecture explain historical maize yield trends in the U.S. Corn Belt?,  Crop Science ,  2009, vol.  49 (pg.  299- 312) Google Scholar CrossRef Search ADS   Hammer GL,  Wright GC.  A theoretical analysis of nitrogen and radiation use efficiency in peanut,  Australian Journal of Agricultural Research ,  1994, vol.  45 (pg.  575- 579) Google Scholar CrossRef Search ADS   Heffner EL,  Sorrells ME,  Jannink J- L.  Genomic selection for crop improvement,  Crop Science ,  2009, vol.  49 (pg.  1- 12) Google Scholar CrossRef Search ADS   Hoogenboom G,  White JW,  Messina CD.  From genome to crop: integration through simulation modeling,  Field Crops Research ,  2004, vol.  90 (pg.  145- 163) Google Scholar CrossRef Search ADS   Horton P.  Prospects for crop improvement through the genetic manipulation of photosynthesis: morphological and biochemical aspects of light capture,  Journal of Experimental Botany ,  2000, vol.  51 (pg.  475- 485) Google Scholar CrossRef Search ADS PubMed  Janick J. ,  Plant Breeding Reviews 24, Part 1: Long-term selection: Maize ,  2004 New Jersey John Wiley & Sons Inc Jannink J-L,  Lorenz AJ,  Iwata H.  Genomic selection in plant breeding: from theory to practice,  Briefings in Functional Genomics and Proteomics ,  2010, vol.  9 (pg.  166- 177) Google Scholar CrossRef Search ADS   Kauffman SA. ,  The origins of order: self-organization and selection in evolution ,  1993 Oxford University Press Keating BA,  Carberry PS,  Hammer GL, et al.  An overview of APSIM, a model designed for farming systems simulation,  European Journal of Agronomy ,  2003, vol.  18 (pg.  267- 288) Google Scholar CrossRef Search ADS   Löffler CM,  Wei J,  Fast T,  Gogerty J,  Langton S,  Bergman M,  Merrill RE,  Cooper M.  Classification of maize environments using crop simulation and geographic information systems,  Crop Science ,  2005, vol.  45 (pg.  1708- 1716) Google Scholar CrossRef Search ADS   Lorenz EN. ,  The essence of chaos ,  1995 Seattle University of Washington Press Ludlow MM,  Muchow RC.  A critical evaluation of traits for improving crop yields in water-limited environments,  Advances in Agronomy ,  1990, vol.  43 (pg.  107- 153) Manschadi AM,  Christopher J,  deVoil P,  Hammer GL.  The role of root architectural traits in adaptation of wheat to water-limited environments,  Functional Plant Biology ,  2006, vol.  33 (pg.  823- 837) Google Scholar CrossRef Search ADS   Messina CD. ,  Understanding maize yield trends in the U.S. Corn Belt ,  2009  64th Corn and Sorghum Seed Research Conference, 8–11 December, Chicago, Illinois Messina CD,  Hammer GL,  Dong Z,  Podlich D,  Cooper M.  Sadras V,  Calderini D.  Modelling crop improvement in a G*E*M framework via gene–trait–phenotype relationships,  Crop physiology: interfacing with genetic improvement and agronomy ,  2009 The Netherlands Elsevier(pg.  235- 265) Messina CD,  Jones JW,  Boote KJ,  Vallejos CE.  A gene-based model to simulate soybean development and yield responses to environment,  Crop Science ,  2006, vol.  46 (pg.  456- 466) Google Scholar CrossRef Search ADS   Monneveux P,  Sanchez C,  Beck D,  Edmeades GO.  Drought tolerance improvement in tropical maize source populations: evidence of progress,  Crop Science ,  2006, vol.  46 (pg.  180- 1991) Google Scholar CrossRef Search ADS   Muchow RC,  Carberry PS.  Environmental control of phenology and leaf growth in tropically adapted maize,  Field Crops Research ,  1989, vol.  20 (pg.  221- 236) Google Scholar CrossRef Search ADS   Muchow RC,  Sinclair TR,  Bennett JM.  Temperature and solar radiation effects on potential maize yield across locations,  Agronomy Journal ,  1990, vol.  82 (pg.  338- 343) Google Scholar CrossRef Search ADS   Orr HA.  The genetic theory of adaptation: a brief history,  Nature Reviews Genetics ,  2005, vol.  6 (pg.  119- 127) Google Scholar CrossRef Search ADS PubMed  Padilla JM,  Otegui ME.  Co-ordination between leaf initiation and leaf appearance in field-grown maize (Zea mays): gGenotypic differences in response of rates to temperature,  Annals of Botany ,  2005, vol.  96 (pg.  997- 1007) Google Scholar CrossRef Search ADS PubMed  Passioura JB.  Roots and drought resistance,  Agricultural Water Management ,  1983, vol.  7 (pg.  265- 280) Google Scholar CrossRef Search ADS   Peccoud J,  Vander Velden K,  Podlich DW,  Winkler C,  Arthur L,  Cooper M.  The selective values of alleles in a molecular network model are context-dependent,  Genetics ,  2004, vol.  166 (pg.  1715- 1725) Google Scholar CrossRef Search ADS PubMed  Podlich DW,  Cooper M.  QU-GENE:a platform for quantitative analysis of genetic models,  Bioinformatics ,  1998, vol.  14 (pg.  632- 653) Google Scholar CrossRef Search ADS PubMed  Podlich DW,  Winkler CR,  Cooper M.  Mapping as you go: an effective approach for marker-assisted selection of complex traits,  Crop Science ,  2004, vol.  44 (pg.  1560- 1571) Google Scholar CrossRef Search ADS   Reymond M,  Muller B,  Leonardi A,  Charcosset A,  Tardieu F.  Combining quantitative trait loci analysis and an ecophysiological model to analyze the genetic variability of the responses of maize leaf growth to temperature and water deficit,  Plant Physiology ,  2003, vol.  131 (pg.  664- 675) Google Scholar CrossRef Search ADS PubMed  Ribaut J-M. ,  Drought adaptation in cereals ,  2006 United States The Hawthorn Press Inc Ribaut J-M,  Hoisington D,  Bänziger M,  Setter TL,  Edmeades GO.  Nguyen HT,  Blum A.  Genetic dissection of drought tolerance in maize: a case study,  Physiology and biotechnology integration for plant breeding ,  2004 New York Marcel Dekker Inc(pg.  571- 609) Salah HBH,  Tardieu F.  Control of leaf expansion rate of droughted maize plants under fluctuating evaporative demand. A superposition of hydraulic and chemical messages?,  Plant Physiology ,  1997, vol.  114 (pg.  893- 900) Google Scholar CrossRef Search ADS PubMed  Sinclair T,  Messina CD,  Beatty A,  Samples M.  Assessment across the United States of the benefits of altered soybean drought traits,  Agronony Journal ,  2010, vol.  102 (pg.  475- 482) Google Scholar CrossRef Search ADS   Sinclair TR,  Muchow RC.  System analysis of plant traits to increase grain yield on limited water supplies,  Agronomy Journal ,  2001, vol.  93 (pg.  263- 270) Google Scholar CrossRef Search ADS   Sinclair TR,  Purcell LC.  Is a physiological perspective relevant in a ‘genocentric’ age?,  Journal of Experimental Botany ,  2005, vol.  421 (pg.  2777- 2782) Google Scholar CrossRef Search ADS   Sinclair TR,  Tanner CB,  Bennett JM.  Water-use efficiency in crop production,  Bioscience ,  1984, vol.  34 (pg.  36- 40) Google Scholar CrossRef Search ADS   Suwa R,  Hakata H,  Hara H,  El-Shemy HA,  Adu-Gyamfi JJ,  Nguyen NT,  Kanai S,  Lightfoot DA,  Mohapatra PK,  Fujita K.  High temperature effects on photosynthate partitioning and sugar metabolism during ear expansion in maize (Zea mays L.) genotypes,  Plant Physiology and Biochemistry ,  2010, vol.  48 (pg.  124- 130) Google Scholar CrossRef Search ADS PubMed  Tardieu F.  Virtual plants: modelling as a tool for the genomics of tolerance to water deficit,  Trends in Plant Science ,  2003, vol.  8 (pg.  9- 14) Google Scholar CrossRef Search ADS PubMed  Tardieu F,  Tuberosa R.  Dissection and modelling of abiotic stress tolerance in plants,  Current Opinion in Plant Biology ,  2010, vol.  13 (pg.  206- 212) Google Scholar CrossRef Search ADS PubMed  Tuberosa R,  Salvi S,  Sanguineti MC,  Landi P,  Maccaferri M,  Conti S.  Mapping QTL regulating morpho-physiological traits and yield: case studies, shortcomings and perspectives in drought-stressed maize,  Annals of Botany ,  2002, vol.  89 (pg.  941- 963) Google Scholar CrossRef Search ADS PubMed  Uhart SA,  Andrade FH.  Nitrogen deficiency in maize. II. Carbon–nitrogen interaction effects on kernel number and yield,  Crop Science ,  1995, vol.  35 (pg.  1384- 1389) Google Scholar CrossRef Search ADS   van Eeuwijk FA,  Bink MCAM,  Chenu K,  Chapman SC.  Detection and use of QTL for complex traits in multiple environments,  Current Opinion in Plant Biology ,  2010, vol.  13 (pg.  1- 13) Google Scholar CrossRef Search ADS PubMed  van Eeuwijk FA,  Boer M,  Totir LR, et al.  Mixed model approaches for the identification of QTLs within a maize hybrid breeding programme,  Theoretical and Applied Genetics ,  2010, vol.  120 (pg.  429- 440) Google Scholar CrossRef Search ADS PubMed  van Eeuwijk FA,  Malosetti M,  Yin X,  Struik PC,  Stam P.  Statistical models for genotype by environment data: from conventional ANOVA models to ecophysiological QTL models,  Australian Journal of Agricultural Research ,  2005, vol.  56 (pg.  883- 894) Google Scholar CrossRef Search ADS   van Oosterom EJ,  Borrell AK,  Chapman SC,  Broad IJ,  Hammer GL.  Functional dynamics of the nitrogen balance of sorghum. I. N demand of vegetative plant parts,  Field Crops Research ,  2009, vol.  115 (pg.  19- 28) Google Scholar CrossRef Search ADS   van Oosterom EJ,  Chapman SC,  Borrell AK,  Broad IJ,  Hammer GL.  Dynamics of the nitrogen balance of sorghum. II. Grain filling period,  Field Crops Research ,  2009, vol.  115 (pg.  29- 38) Google Scholar CrossRef Search ADS   Vega CRC,  Andrade FH,  Sadras VO,  Uhart SA,  Valentinuz OR.  Seed number as a function of growth. A comparative study in soybean, sunflower, and maize,  Crop Science ,  2001, vol.  41 (pg.  748- 754) Google Scholar CrossRef Search ADS   Welch SM,  Dong Z,  Roe JL,  Das S.  Flowering time control: gene network modelling and the link to quantitative genetics,  Australian Journal of Agricultural Research ,  2005, vol.  56 (pg.  919- 936) Google Scholar CrossRef Search ADS   Welcker C,  Boussuge1 B,  Bencivenni C,  Ribaut J-M,  Tardieu F.  Are source and sink strengths genetically linked in maize plants subjected to water deficit? A QTL study of the responses of leaf growth and of anthesis–silking interval to water deficit,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  339- 349) Google Scholar CrossRef Search ADS PubMed  White RG,  Kirkegaard JA.  The distribution and abundance of wheat roots in a dense, structured subsoil: implications for water uptake,  Plant, Cell and Environment ,  2010, vol.  33 (pg.  133- 148) Google Scholar CrossRef Search ADS   White JW,  Hoogenboom G.  Simulating effects of genes for physiological traits in a process-oriented crop model,  Agronomy Journal ,  1996, vol.  88 (pg.  416- 422) Google Scholar CrossRef Search ADS   Whitmore AP,  Whalley WR.  Physical effects of soil drying on roots and crop growth,  Journal of Experimental Botany ,  2009, vol.  60 (pg.  2845- 2857) Google Scholar CrossRef Search ADS PubMed  Wright S.  The roles of mutation, inbreeding, crossbreeding and selection in evolution,  Proceedings of the 6th International Congress of Genetics ,  1932  Ithaca, NY, 356–366 Yin X,  Struik PC,  van Eeuwijk FA,  Stam P,  Tang J.  QTL analysis and QTL-based prediction of flowering phenology in recombinant inbred lines of barley,  Journal of Experimental Botany ,  2005, vol.  56 (pg.  967- 976) Google Scholar CrossRef Search ADS PubMed  Yin X,  Struik PC.  Modelling the crop: from system dynamics to systems biology,  Journal of Experimental Botany ,  2010, vol.  61 (pg.  2171- 2183) Google Scholar CrossRef Search ADS PubMed  © The Author [2010]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: [email protected]

Pinheiro, C.;Chaves, M. M.

doi: 10.1093/jxb/erq340pmid: 21172816

Abstract Photosynthesis is one of the key processes to be affected by water deficits, via decreased CO2 diffusion to the chloroplast and metabolic constraints. The relative impact of those limitations varies with the intensity of the stress, the occurrence (or not) of superimposed stresses, and the species we are dealing with. Total plant carbon uptake is further reduced due to the concomitant or even earlier inhibition of growth. Leaf carbohydrate status, altered directly by water deficits or indirectly (via decreased growth), acts as a metabolic signal although its role is not totally clear. Other relevant signals acting under water deficits comprise: abscisic acid (ABA), with an impact on stomatal aperture and the regulation at the transcription level of a large number of genes related to plant stress response; other hormones that act either concurrently (brassinosteroids, jasmonates, and salycilic acid) or antagonistically (auxin, cytokinin, or ethylene) with ABA; and redox control of the energy balance of photosynthetic cells deprived of CO2 by stomatal closure. In an attempt to systematize current knowledge on the complex network of interactions and regulation of photosynthesis in plants subjected to water deficits, a meta-analysis has been performed covering >450 papers published in the last 15 years. This analysis shows the interplay of sugars, reactive oxygen species (ROS), and hormones with photosynthetic responses to drought, involving many metabolic events. However, more significantly it highlights (i) how fragmented and often non-comparable the results are and (ii) how hard it is to relate molecular events to plant physiological status, namely photosynthetic activity, and to stress intensity. Indeed, the same data set usually does not integrate these different levels of analysis. Considering these limitations, it was hard to find a general trend, particularly concerning molecular responses to drought, with the exception of the genes ABI1 and ABI3. These genes, irrespective of the stress type (acute versus chronic) and intensity, show a similar response to water shortage in the two plant systems analysed (Arabidopsis and barley). Both are associated with ABA-mediated metabolic responses to stress and the regulation of stomatal aperture. Under drought, ABI1 transcription is up-regulated while ABI3 is usually down-regulated. Recently ABI3 has been hypothesized to be essential for successful drought recovery. ABA, carbon metabolism, drought, photosynthesis, sugars, stress imposition rate and intensity Introduction Dictated by evolution, plant success in unfriendly environments (including drought) involves a plethora of responses, from early responses to longer term metabolic and physionomic alterations that can sustain acclimation and survival (Lawlor, 2009). This requires a tight coordination at the whole plant level. From the work being produced in the last decade, it became apparent that plants perceive and respond rapidly to alterations (even small) in water status via a series of physiological, cellular, and molecular events developing in parallel (Chaves et al., 2009). The responses at various levels are modulated by the intensity, duration, and rate of progression of imposed drought. From the methodological point of view, this complexity poses additional challenges to the compilation and integration of the available data. This brings about the need to define and monitor the drought that plants are facing in each experiment, in terms of water availability in the substrate, photosynthetic activity, plant water status, as well as the radiation to which plants are subjected, since this combination will determine the elicited responses. Only such fine monitoring of conditions will allow the results to be reproduced and correctly interpreted, and for distinct data sets to be compared (Jones, 2007; Dehyolos, 2010). Often the studies of molecular responses of plants to drought use very artificial systems of stress imposition, such as an instantaneous decline in water availability produced by detaching organs or removing the plants from substrates. Such experimental conditions cannot provide information on relevant acclimation processes that might occur under field conditions. Furthermore, rapid alterations (within hours) would not necessarily reflect a response to a long-term water shortage but instead a short-term adjustment to a new environmental condition. Although some of the effects of a rapidly imposed water deficit might be common to those when the deficit is imposed slowly, reproduction of slowly imposed water deficits under field conditions is required when considering a crop's response to drought. This type of study will allow the evaluation of acclimation processes in mature plants as well as plant resistance to a multistress situation that often is the cause of dramatic losses in agricultural production. Recent studies revealed that molecular and metabolic responses of plants to a combination of stresses are unique and cannot be extrapolated from the separate study of individual stresses (Mittler, 2006). Moreover, from an agricultural perspective, drought is ultimately defined in terms of its effects on yield, since this is the relevant issue when addressing the improvement of crop production under water-limited environments (Passioura, 2007). Consequently, the timing of water deficits during the season (e.g. sowing, crop establishment, flowering, or grain filling) may have a much larger impact on yield than the intensity of drought per se. As the key process of primary metabolism, photosynthesis plays a central role in plant performance under drought (see reviews by Chaves et al., 2003, 2009; Flexas et al., 2004; Lawlor and Tezara, 2009). The decline observed in leaf net carbon uptake as a result of plant water deficits is followed by an alteration in partitioning of the photoassimilates at the whole plant level, corresponding in general to an increase in the root to shoot ratio. This is the result of the decline in shoot growth and the maintenance of root growth under decreasing water in the soil (Sharp, 2002). Such a response is mediated by hormonal control, namely by abscisic acid (ABA), ethylene, and their interactions (Wilkinson and Davies, 2010), as will be discussed further on. The changes in the root–shoot ratio as well as the temporary accumulation of reserves in the stem that occur in several species under water deficits (Blum et al., 1994; Chaves et al., 2002) are accompanied by alterations in carbon and nitrogen metabolism in the different organs (Pinheiro et al., 2001; António et al., 2008), whose fine regulation is still largely unknown. In this context, sugars are likely to be key players in the integration, at the whole plant level, of the cellular responses to internal and environmental alterations. They act as substrates and modulators of enzyme activity in carbon-related pathways and via the control of expression of different genes related to carbon, lipid, and nitrogen metabolism (Koch, 1996; Gibson, 2000; Rolland et al., 2006). The interplay of sugars with other stress elicitors, such as redox and hormone signals, is at the forefront of present research efforts (Couée et al., 2006; Usadel et al., 2008; Bolouri-Moghaddam et al., 2010). Sulpice et al. (2009) recently suggested that starch is a major integrator of plant metabolism and growth, in response to changes in development or the environment, reflecting a regulatory network that balances growth with carbon supply (see Discussion and Fig. 1 below). Fig. 1. View large Download slide Biological networks generated for drought and photosynthesis interactions considering the literature available (1995 to February 2010) by making use of Pathway Studio software. The complete list of interactions as well as details of the pathway is available in Supplementary Tables S1 and Supplementary Data at JXB online. A, small molecules (loc al connectivity ≥6); B, proteins and/or genes (local connectivity ≥ 6). (This figure is available in colour at JXB online.) Fig. 1. View large Download slide Biological networks generated for drought and photosynthesis interactions considering the literature available (1995 to February 2010) by making use of Pathway Studio software. The complete list of interactions as well as details of the pathway is available in Supplementary Tables S1 and Supplementary Data at JXB online. A, small molecules (loc al connectivity ≥6); B, proteins and/or genes (local connectivity ≥ 6). (This figure is available in colour at JXB online.) In the present paper, the current status of the physiological limitations to photosynthesis under drought was revised and a meta-analysis (covering >450 papers published in the last 15 years) was performed with the goal of strengthening our understanding of the complex network of interactions and regulations of photosynthesis in plants subjected to water deficits. The working hypothesis is that if a general/unifying response exists it will emerge from these data and would be a useful starting point for future experiments. Revisiting drought constraints to photosynthesis Several recent review papers have dealt with this issue in a comprehensive way (Chaves et al., 2009; Lawlor and Tezara, 2009). However, there is still some controversy regarding the relative importance and timing of the main physiological targets responsible for limiting photosynthesis under drought. Decreased CO2 diffusion from the atmosphere to the site of carboxylation is generally considered the main cause for decreased photosynthesis under mild to moderate water limitation (Chaves et al., 2003, 2009; Flexas et al., 2004; Grassi and Magnani, 2005). This limitation includes a stomatal and a mesophyll component (Flexas et al., 2008). The magnitude of the latter is still under debate, with criticisms arising over methodological issues related to the estimation of the intercellular or the chloroplastic CO2 concentration (Bunce, 2009; Lawlor and Tezara, 2009). Mesophyll conductance (gm) comprises physical (solubility of CO2, surface area of the apoplastic, and symplastic routes of CO2) and metabolic components (aquaporins and carbonic anydrase). Both are dependent on the species concerned, presumably as a result of differences in the relative contribution of anatomical versus biochemical components and on the experimental conditions, namely water deficits and temperature. It has been shown that soil water deficits can substantially reduce gm (see Flexas et al., 2008), although in general gm is less sensitive to water stress than gs (Bunce, 2009). In species well adapted to dry environments the feed-forward responses of stomata to soil and atmospheric dryness are important components of plant water saving (Maroco et al., 1997; Chaves et al., 2003, Chaves and Oliveira, 2004; David et al., 2007). Stomata act as pressure regulators that prevent xylem pressure from runaway cavitation thresholds (David et al., 2007). This is visible in the midday closure of stomata on hot days or in the decreased stomatal conductance in response to mild soil dehydration, in plants whose tissue water status is high. Both responses seem to be mediated by ABA synthesized in or transported to the leaves (from dehydrating roots) and are modulated by numerous internal and external factors, as will be discussed later on (see also a recent review by Wilkinson and Davies, 2010). Under field conditions, together with the intensity and duration of midday depression of stomatal conductance, how early in the day it starts to appear is an important indicator of the degree of stress being endured by the plant. Furthermore, when decreased stomatal conductance is combined with sustained high irradiance, leaves are subjected to excess incident energy relative to the available intercellular CO2, and the rate of reducing power production can overcome the rate of its use by the Calvin cycle. Under such circumstances, down-regulation of photosynthesis, or even photoinhibition, can be a powerful defence mechanism in C3 plants. Such protection may be achieved by the regulated thermal dissipation occurring in the light harvesting complexes, involving the xanthophyll cycle (Demmig-Adams and Adams, 1996; Demmig-Adams et al., 2006) and the lutein cycle (Garcia-Plazaola et al., 2003). These photoprotective mechanisms compete with photochemistry for the absorbed energy, leading to a down-regulation of photosynthesis evidenced by the decrease in quantum yield of photosystem II (PSII) (Genty et al., 1989). If the limitation of the rate of CO2 assimilation is accompanied by an increase in the activity of another sink for the absorbed energy, for example photorespiration (Harbinson et al., 1990; Wingler et al., 1999) or the Mehler-peroxidase reaction (Biehler and Fock, 1996), the decline in non-cyclic electron transport will be proportionally less than the decrease observed in the rate of CO2 assimilation. This type of response has been well documented in C3 plants native of semi-arid regions and less so in C4 plants (Ghannoum, 2009). Recent evidence suggests that the equal or even stronger susceptibility to water deficits observed in C4 plants as compared with C3 plants, in spite of the CO2-concentrating mechanism in the former, may be ascribed to the limited capacity of photorespiration or the Mehler reaction to act as alternative electron sinks for excess reducing power (Ghannoum, 2009). The biochemical component of the limitation of photosynthesis under water deficits is generally estimated as much smaller than the diffusion limitation (Galmes et al., 2007a). However, its importance should not be underestimated. Indeed, alterations in gene expression may develop early on in response to the decline of plant water status, preceding acclimation mechanisms, although the impact on metabolites may not occur immediately (Chaves et al., 2009). Energy balance was also recognized as a key component of cell functioning under limited supply of CO2 and high light (Lawlor and Tezara, 2009; Pfannschmidt et al., 2009). Under such conditions, Tezara et al. (1999) had found an impaired ATP production and thus ribulose bisphosphate (RuBP) regeneration, and recently reactive oxygen species (ROS) generated under highly reduced conditions in the chloroplast were shown to damage ATP synthase (Lawlor and Tezara, 2009). As for the impact of water deficits on Rubisco, although the results are very variable, it has generally been found that its activity and quantity are affected under severe stress (Maroco et al., 2002; Parry et al., 2002; Flexas et al., 2006b), although there is also evidence of alterations at the transcript level under milder stress (Supplementary Table S5 at JXB online). The drop in Rubisco activase is presumably a key factor in slowing down Rubisco activity (Lawlor and Tezara, 2009). A recent study by Galmés et al. (2010) in 11 Mediterranean species suggests that low chloroplastic CO2 concentration (Cc) occurring under water stress could induce de-activation of Rubisco sites, the threshold of Cc triggering de-activation of Rubisco being dependent on leaf characteristics. It is also suggested that species adapted to functioning at low Cc can maintain active Rubisco under more intense drought. The transcriptional control of photosynthetic genes by transcription factors (TFs) in response to abiotic stresses was recently reviewed (Saibo et al., 2009), pointing out the role of several TFs belonging to the MYB family in both stomatal and non-stomatal limitations of photosynthesis. They are involved in the regulation of stomatal number and size, and of metabolic components of the photosynthetic system. As highlighted by Pourkeirandish and Komatsuda (2007), breeding for increased crop yield seems to have altered the functionality of some TFs, with potential impact on their responses to the environment. Adding the whole plant photosynthesis dimension To study the response of crop photosynthesis to drought, it is relevant to approach it at the canopy scale as well, since crop productivity is dependent on photoassimilates produced at the whole plant level. It is known that the decline in stomatal aperture is accompanied by the adjustment of leaf area at the whole plant level. It occurs either via the inhibition of new leaf growth or via the earlier senescence of older leaves, in the case of prolonged stress. This reduction in foliage dimension leads to decreased transpirational area but also to lower intercepted radiation throughout the growing season and ultimately to decreased biomass production (Pereira and Chaves, 1993). In many crops, alteration of the leaf angle with dehydration, towards smaller angles, will also diminish total intercepted radiation and therefore carbon assimilation by the plant, but will have an important protective role against excess solar energy. Photosynthetic resilience to drought is known to vary with leaf age (Chaves, 1991). Younger leaves tend to be more resistant to drought than older leaves, and this increased tolerance may be particularly relevant in plants where a severe reduction in the size of the leaf canopy occurs as a result of shedding of older leaves, because it allows a fast recovery following rehydration (Pereira and Chaves, 1993). In addition to a plant's ability to avoid and/or endure water stress, photosynthetic recovery following rehydration is pivotal to dictate a plant's resistance to drought and to prevent dramatic declines in crop yield (Chaves et al., 2009). It was shown that recovery from a severe stress was a two-stage process: the first stage occurs during the first hours or days upon re-watering, corresponding to the improvement of leaf water status and stomatal re-opening (Pinheiro et al., 2005; António et al., 2008; Hayano-Kanashiro et al., 2009); and the second stage lasts several days and requires de novo synthesis of photosynthetic proteins (Kirschbaum, 1988). Previous stress intensity and/or duration are crucial factors affecting both the velocity and the extent of recovery of photosynthesis (Miyashita et al., 2005; Flexas et al., 2006a). Long-term down-regulation of gs after re-watering may be derived from limited recovery of leaf-specific hydraulic conductivity (Galmés et al., 2007c). From the molecular point of view, the comparison between susceptible and tolerant genotypes suggests that drought tolerance is associated with a rapid modulation of genes from different TF gene families during recovery. For example, the greatest difference between drought-tolerant and drought-sensitive maize genotypes was observed in the speed of transcriptional down-regulation during recovery from drought (Hayano-Kanashiro et al., 2009). The respiration connection: support for photosynthesis recovery? Net carbon gain that ultimately dictates plant growth and development reflects the balance between photosynthesis and respiration (in auto- and heterotrophic tissues). Indeed, 30–70% of the CO2 fixed per day by net photosynthesis in well-watered plants is released back into the atmosphere by plant respiration, the larger part through the leaves (Aktin and Macherel, 2009). The impact of water deficits on dark respiration is still far from clear, with reports in the literature comprising decreases, maintenance, or increases in the rates of this process (Gimeno et al., 2010). Inhibition of respiration under drought has been observed in actively growing roots and mature leaves of crops and herbaceous species (e.g. Haupt-Herting et al., 2001; Ribas-Carbo et al., 2005; Galmes et al., 2007b). Decreased availability of the substrate to the mitochondria under conditions of low photosynthesis as well as inhibition of leaf growth may explain reduced respiration, mostly in its growth component (Flexas et al., 2006a; Gimeno et al., 2010). However, a higher demand for respiratory ATP under severe water stress (to compensate for the lowered ATP production in the chloroplasts) may be required to support photosynthesis repair mechanisms, as suggested by Flexas et al. (2005, 2006a) and Atkin and Macherel (2009). Higher respiration rates, mainly as the maintenance component, are then observed in droughted plants, underlying acclimation mechanisms of drought (Gratani et al., 2007; Slot et al., 2008). Finally, a third response pattern, with no alterations in the rates of dark respiration under drought, was reported in several species, mostly in evergreen perennials (Galmes et al., 2007b; Gimeno et al., 2010). Elaborating on such contrasting results, Atkin and Macherel (2009) proposed a model where mitochondrial respiration dictates plant survival and rapid recovery of productivity under water stress conditions, by ensuring survival under extended periods of drought. According to some authors (Gimeno et al., 2010), shrubs and trees that possess long-lived leaves are likely to show slower responses to drought than short-lived species that need to optimize their carbon gain over shorter periods and therefore may respond quickly to water scarcity, lowering their respiration rates. From the biochemical point of view it has been reported that the electron partitioning towards the alternative respiration pathway sharply increases under severe drought, even when total respiration rates are not greatly affected (Ribas-Carbo et al., 2005). Unlike many other stresses, water stress does not affect the quantity of mitochondrial alternative oxidase protein, suggesting that a biochemical regulation causes this mitochondrial electron shift. This shift may have a physiological significance, since evidence is accumulating to support a role for the alternative oxidase in the prevention of the formation of ROS (Lambers et al., 2005). Overall, the changes observed in respiration in response to drought are smaller as compared with the large decreases in photosynthesis; therefore, as carbon uptake becomes more limiting under water scarcity, respiration increases proportionally, leading to increased leaf intercellular CO2 and altered plant carbon balance (Lawlor and Tezara, 2009). As already mentioned, the ratio between the respiratory needs for growth and maintenance will also change in plants under water stress, the component devoted to shoot growth being drastically decreased (Flexas et al., 2006a). Drought and photosynthesis: the metabolic connections Although impressive advances have been made in the last decade with respect to the nature of events occurring in plants subjected to drought, an integrative picture of the metabolic regulation taking place is still missing (Rolland et al., 2006; Shinozaki et al., 2007). This is partly related to the disparate experimental conditions of the studies being done and the very artificial conditions of the applied stress, frequently acute and/or too severe. Moreover, in many studies, particularly those dealing with the molecular responses to drought, plant water status, leaf conductance, and photosynthetic rate are usually not measured, which makes comparative analysis of these data very difficult to perform. Details on regulatory mechanisms and interactions are available for specific situations, although systematic information on common/general effects is still scarce. It is compelling that the vast amount of information on plant transcriptomes under drought has not yet been translated into genotype selection. This is for the most part due to low correlations between transcript abundance and corresponding protein and enzyme activities, as well as plant physiological performance, the question ‘what do these genes contribute to stress tolerance?’ still being largely unanswered (Chaves et al., 2009; Deyholos, 2010). New experimental and computational methods are starting to allow multilevel analysis that can integrate physiological, transcriptome, proteome, and metabolome data, thus providing a more detailed view of the cellular events (Eberhard et al., 2008), and contributing to disclosure of the existence of common metabolic features in photosynthetic responses to drought. Drought effects on photosynthesis: description of the meta-analysis For a comprehensive analysis of the effects of drought on photosynthesis and related molecular and metabolic events, a literature survey (from 1995 to February 2010) was performed using the tools MedScan Reader v3 and Pathway Studio v7 from Ariadne Genomics (www.ariadnegenomics.com). This meta-analysis integrates information from most studies on drought and photosynthesis published during the 15 year period mentioned above and allows a picture to be obtained of the main processes involved and how often they have been reported. A total of 469 publications for proteins/genes and 515 for metabolites associated with drought and photosynthesis were screened. Taking into consideration the number of publications that relate the involvement of a particular protein, gene, or metabolite in photosynthesis responses to drought, those associations can be considered as strong (high number of records), weak (few records), or inexistent, the latter indicating either a weak association at the biological level or few references available. Using this tool it was possible to identify proteins, genes, or metabolites (small molecules) that are most strongly associated with drought and photosynthesis (the complete list of relationships extracted from the literature is available as Supplementary Table S1 at JXB online). It includes sugars, hormones, and ROS pathways. By focusing on the association of such pathways with drought and photosynthesis, it was possible to identify 389 relationships at the metabolite level and 256 relationships at the protein/gene level (Supplementary Table S2). In order to disclose strong associations, the relationships with high connectivity (which imply that a given identity is associated with several of the processes) are considered (Fig. 1). Drought, photosynthesis, and strongly associated pathways At the metabolite level (Fig. 1A), a high level of connectivity is found between drought, photosynthesis, ROS, ABA, sucrose, and starch. Cytokinin-related processes are not well documented, presumably because they have not been studied to any great extent. Interactions between the different hormone pathways are also noticeable, namely for ABA, auxins, and ethylene. At the protein/gene level (Fig. 1B), sugars, starch, ROS, and ABA pathways are well represented in this network relating drought and photosynthesis. In contrast, auxins and ethylene appear with weaker associations (Supplementary Table S2 at JXB online). The literature survey reveals proteins with regulatory functions that connect photosynthesis responses and drought, as is the case for the TFs T6L1.5, HY5, AHBP-1B, and GBF3, members of the bZIP (basic leucine zipper domain) family, and one TF belonging to the ABI3 (abscisic acid-insensitive 3) family. The involvement of the bZIP and ABI3 transcription factors, both ABA dependent, in the connection between drought and photosynthesis has recently been reviewed (Saibo et al., 2009). Members of the bZIP family are associated with Rubisco regulation (Saibo et al., 2009); more specifically, the bZIP TFs found in the present analysis were found to be associated with: photomorphogenesis and cytokinin pathways (HY5); salicylic acid-mediated responses and defence genes (AHBP-1B); the sucrose-sensing pathway; the ABA pathway; and interactions with HY5 and AHBP-1B (GBF3) (wikigenes platform, www.wikigenes.org; Hoffmann, 2008). The transcriptional regulator ABI3 is associated with the regulation of stomatal aperture, involved in the auxin pathways (Brady et al., 2003), and interacts with ABI1 (Parcy and Giraudat, 1997). ABI1 codes for a serine/threonine phosphatase and is related to stomatal regulation and ABA-mediated responses. ETR1 (ethylene response 1, a protein histidine kinase) is responsive to ABA, auxins, ethylene, cytokininis, and gibberellins, and was shown to be involved in the regulation of stomatal movement, the glucose-sensing pathway, and H2O2 biosynthesis. This gene and the protein it encodes can thus link ROS, sugar, and hormone pathways. All TFs identified in the meta-analysis as strong connections, linked drought and photosynthesis to the regulation of stomatal aperture. The literature survey also highlights strong associations concerning Rubisco (CO2 assimilation), catalase (ROS detoxification), nitrate reductase (nitrate assimilation; NO synthesis), invertase (sucrose breakdown through an irreversible reaction), sucrose synthase (which catalyses a reversible reaction depending on the cellular homeostasis), and amylase (starch metabolism) with photosynthesis under drought. Several questions arise from this analysis. How do the readjustments of the mentioned metabolic pathways affect and/or are affected by photosynthesis? Do they contribute to the plant's ability to cope with the stress? How are such adjustments coordinated? Are such effects strongly dependent on transcription or could they be achieved through metabolic reorganization? Effects at the transcriptional level: (i) methodological information The results can be further explored by making use of the publicly available databases on drought effects at the transcription level. This approach can allow the detection of potentially relevant trends to be considered in future studies. One of the points of regulation in the photosynthetic response to drought is at the gene transcription level, and there is evidence indicating that although most photosynthetic genes are down-regulated in response to a multitude of stress conditions, there are certain subsets of photosynthetic genes linked with protective functions that are up-regulated. How drought affects transcription of the genes coding for the proteins represented in Fig. 1B was analysed, aiming at (i) identifying a general trend that would unify these responses; and (ii) finding out whether responses at the transcription level were different according to the stress imposition rate (fast versus slow) and the intensity (mild versus intense drought). The publicly available data deposited in the EMBL-EBI ArrayExpress (www.ebi.ac.uk/microarray-as/ae/) were exploited, and by using the keywords drought (62 experiments), water stress (34 experiments), or dehydration (9 experiments), 56 distinct experiments for several species were detected, 26 of them in Arabidopsis (Supplementary Table S4 at JXB online). Several experiments were eliminated due to several drawbacks: use of osmotic stress rather than drought (five data sets); no information on stress type and duration (two data sets); only recovery data reported (one data set); and only mutant information available (six data sets). In the 12 remaining Arabidopsis arrays, five experiments were done using soil and seven using paper/detached leaves. Relative water content is only available for three assays and none has information on leaf conductance or photosynthesis. From the six rice data sets only two are suitable for this analysis (one in paper; one in soil). For barley, three of the four experiments were considered and represent soil experiments. In order to be able to use the array information, Arabidopsis genes/proteins (detailed information on their identity is available as Supplementary Table 3 at JXB online) were translated to the corresponding probe set IDs via the Plant Expression database (www.plexdb.org/modules/glSuite/gl_main.php; Wise et al., 2007). Since in Pathway Studio there was no information for barley proteins, the Arabidopsis and rice probe sets were respectively translated from the Arabidopsis and rice microarray platform to the barley platform (http://www.plexdb.org/modules/MPT/mpt_Input.php); the identity of the resulting barley probe sets was confirmed (www.plexdb.org/modules/PD_probeset/GO_annotation.php). Effects at the transcriptional level: (ii) drought, photosynthesis, and strongly associated pathways Supplementary Table S1 shows the effect of drought at the gene level in the model plant Arabidopsis and the crop plants rice and barley for the genes/proteins represented in Fig. 1B. The main findings may be summarized as follows. (i) The stress imposition method (soil versus paper) significantly affects the responses. The principal component analysis bi-plot generated with the Arabidopsis arrays shows that the soil data sets group together, while paper data sets are more dispersed and not mixed with the soil experiments (data not shown). (ii) In the same data set, distinct effects are observed within the same gene family, which allows the hypothesis that at the protein level the effects can be cancelled out. Furthermore, protein activity is modulated at many levels including post-translational modifications (see Eberhard et al., 2008), with activity being modulated via substrate flux and availability and cellular compartmentation (Deyholos, 2010). (iii) The drought effect in a given gene is highly variable, with few exceptions (notably ABI1 and ABI3). (iv) Usually the observed responses (up- or down-regulation) are not reversed with stress progression. The present meta-analysis (Fig. 1), taken together with the transcriptomic data (Supplementary Table S5 at JXB online), highlights the difficulties faced when searching for metabolic events associated with stress and in gaining insight into the relevant pathways, because not all biological and methodological variables are considered in the different experiments. Although the association of a given response with stress perception, intensity, tolerance, or sensitivity is still rare, the analysis allowed recognition of some potentially relevant features. For example, ABI1 is up-regulated under water deficits in both Arabidopsis and barley plant systems and stress types, while for ABI3 the opposite trend is observed. ETR1 showed a similar response to ABI1, although not so marked, and seems not to respond to acute stress. These genes are related to stomatal closure regulation and provide a link between several hormone pathways. Recently, Khandelwal et al. (2010) highlighted a new target for ABI3 action in Physcomitrella patens (in an acute stress experiment). They inferred that several transcripts produced during ABA pre-treatment (necessary for P. patens desiccation survival) are necessary for recovery (ABI3 mutants do not survive). This may be linked with previous findings of gene expression required for stress recovery being already operative during desiccation (Bray, 1993). Accordingly, very few rehydration-specific proteins are known (Bartels and Salamini, 2001), and in the leaves of two resurrection plants (Xerophyta humilis and Craterostigma wilmsii) recovery is largely independent on de novo gene transcription and protein translation (Dace et al., 1998; Cooper and Farrant, 2002). In lupins (Pinheiro et al., 2005) and wheatgrass (Gazanchian et al., 2007), it became apparent that the proteins needed for early plant recovery could already be present during the severe stress phase. Regarding the invertase multigenic family, for three genes (one coding for cell wall invertase and two for neutral invertases) it was possible to distinguish the effects from acute (paper) stress and soil stress experiments. The genes AT3G13970 (cell wall invertases) and AT4G09510 (neutral invertase) are down-regulated under soil water stress but up-regulated with acute stress; the gene AT3G06500 (neutral invertase) is up-regulated in both systems but it seems to be affected more at the very early stages of the acute stress (1–2 h). This is an example of differences in plant response to the velocity of stress imposition—sucrose metabolism will be affected in distinct ways when plants acclimate to slowly imposed water or with a fast response to a dramatic change in tissue water status. Moreover, the light regime under which plants are grown may also drastically influence the results. For example, when water deficits were imposed on plants adapted to low light, as in a recent study with Arabidopsis (180 μmol m−2 s−1), the expression of a set of sugar-responsive genes indicates increased, rather than decreased, carbon availability (Hummel et al., 2010). Indeed, under such conditions photosynthesis was not affected under severe stress (because it was light limited) and the concomitant inhibition of shoot growth gave rise to a surplus of carbon, which was re-directed to root growth. It must be emphasized that these results cannot be extrapolated to field conditions where net carbon uptake will be decreased and carbon limitation will be apparent. Co-expression transcription analyses The tool ‘Correlated Gene Search’ available at PRIMe (http://prime.psc.riken.jp; Akiyama et al., 2008) was also used in order to detect correlated transcription of the Arabidopsis genes described in Supplementary Table S3 at JXB online. Co-expression analysis asks the question ‘what are the genes that show similar expression profiles across many experiments to my gene of interest?’ Genes that are highly co-expressed may be involved in the biological process or processes of the query gene. Considering the available experiments (AtGenExpress arrays that deal with acute stress treatments) the following correlation groups were found. (i) ABI1 transcription is correlated with sucrose synthase genes (AT5G20830, AT4G02280), neutral invertase (At3g06500), β-amylase (AT3G23920), and the TF GBF3 (AT2G46270); this group has a branch where GBF3 and SUS3 (AT4G02280) also correlate with another neutral invertase (AT4G34860), the neutral invertase and SUS3 also correlating with AMY1. (ii) This group relates several α-amylase genes to each other (AT5G04360, AT1G69830, AT2G39930, and AT4G09020). It also associates ISA3 (AT4G09020) with catalase transcription (AT1G20630, AT1G20620). (iii) In this group, another catalase gene (AT4G35090) is associated with a β-amylase (AT5G18670) in an independent node. (iv) In this group, ABI3 is associated with SUS2 (AT5G49190). This exercise shows the interplay between ABA, sucrose, starch, and ROS metabolism and points to the role of neutral invertase (not as well studied as the acid form) in stress responses. The meta-analysis presented here documents the strong association between drought, photosynthesis, sugars, hormones, and ROS (Fig. 1), but more importantly demonstrates that it is necessary to make additional efforts in stress characterization and quantification to be able to associate the alterations firmly with stress type and intensity. Moreover, comparisons between tolerant and susceptible genotypes are still scarce and need to be strengthened because this is a faster way to produce information to design more efficient breeding programmes to produce genotypes better adapted to water-limiting conditions. Signals and metabolic cross-talk In the chain of events triggered by drought, one relevant issue relates to signals and signalling cascades and their interactions. Although these terms (signals and signalling cascades) are often used, the biological rationale and supportive data are not always obvious. Some questions remain elusive. Are the observed effects a signal or a consequence of the stimulus (direct or indirect)? How do such changes lead to metabolic rearrangements in photosynthesis? Hormones Water deficit affects biosynthesis, accumulation, and redistribution of major plant hormones, with ABA (synthesized either in leaves or in roots) playing the major role in controlling stomatal aperture and therefore photosynthetic carbon uptake under conditions of water scarcity (Dodd, 2003; Hirayama and Shinozaki, 2007). Stomatal sensitivity to ABA is modulated by a number of external drivers, such as temperature, ozone, nitrogen nutrition (often altered in drying soil), and endogenous components, including cytosolic pH or hydraulic signals that can either reinforce or moderate ABA-based signals (Wilkinson and Davies, 2002, 2010; Parent et al., 2009). Recent reports show that stomatal function is also dependent on other hormones (auxin, cytokinin, ethylene, brassinosteroids, jasmonates, and salicylic acid) and on the degree of their interactions (see the reviews by Acharya and Assmann, 2009; Wilkinson and Davies, 2010). In general, auxin, cytokinin, or ethylene tend to inhibit ABA-mediated stomatal closure, whereas brassinosteroids, jasmonates, and salycilic acid display a concurrent action with ABA. Moreover, all these hormones modulate the expression of different drought-related genes (Shinozaki and Yamaguchi-Shinozaki, 2007; Huang et al., 2008). Multiple cascades of cellular biochemical events have also been associated with the regulation of stomatal guard cells, such as the activation of G-proteins, the production of ROS (ABA stimulated) and NO, cytosolic Ca2, protein phosphorylation/dephosphorylation, and reorganization of the cytoskeleton (Acharya and Assmann, 2009). The phosphatases ABI1 and ABI2 were shown to be crucial for ABA-mediated stomatal regulation (Merlot et al., 2001) and are one of the strong connection points in photosynthesis responses to drought ABI1 in Fig. 1B. Still, many questions remain to be clarified, namely the molecular basis of cross-talk among different hormones or the underlying causes for dual roles played by some of them, such as ethylene or even ABA, as recognized by Parent et al. (2009). Recent attention was given to the interactions observed between hormonal and circadian networks, since it has been demonstrated that a large proportion of transcripts involved in hormonal metabolism, catabolism, and signalling are also regulated by the circadian clock (Dodd et al., 2007; Robertson et al., 2009). Daily rhythms have been recognized for a long time in different plant processes, namely photosynthesis and stomatal aperture, and this may have resulted from an evolutionary pressure in order to prevent physiological responses that might be counterproductive during some parts of the day, when temperature and radiation are excessive. Circadian clocks may therefore moderate or produce antagonistic effects relative to hormones, such as those they produce with sugars, as is highlighted further on. Redox signals Maintaining homeostasis of redox and adenylate systems is essential for cell functioning. Whenever an imbalance develops between capture of light and its utilization via CO2 and NO3− reduction, as may happen under drought, redox signals from photosynthetic electron transport and production of ROS may occur (Lawlor, 2009) (see Fig. 1). It has now been extensively demonstrated in several biological systems that these redox signals and ROS have an important function in the plant's acclimation to stress (Buchanan and Balmer, 2005; Hayano-Kanashiro et al., 2009). ROS are produced in plant tissues due to the partial reduction of oxygen as, for example, in the photosynthetic and the respiratory electron chains or the photorespiration pathway, or they accumulate as a result of the activity of peroxidases, membrane-located NADPH oxidases, etc., and this production increases dramatically under environmental stress (Mittler et al., 2004). On the other hand, the intensity, duration, and localization of the different ROS signals are determined by the interplay between the ROS-producing and ROS-scavenging pathways of the cell, as highlighted in Fig. 1B (nitrate reductase and catalase). Further, antioxidants such as ascorbate, tocopherol, or glutathione (Fig. 1A; Supplementary Tables S1, Supplementary Data at JXB online) are able to control the lifetime of ROS signals and therefore participate in the overall redox regulation that ultimately controls the energy balance in plants (Foyer and Noctor, 2009). Although ROS can modulate many pathways (e.g. mitogen-activated protein kinase cascades) and influence the activity of TFs, redox control over photosynthesis is still largely unknown. It may occur, at least partly, through the monitoring of the cell redox status by several molecules in different cellular compartments, reporting the functional state of the chloroplast to the nucleus (Pfannschmidt et al., 2009), as suggested by Jaspers and Kangasjarvi (2010), since ROS are mostly ephemeral molecules. The sugar connection Soluble sugars play a central role in plant metabolism as sources of carbon and energy in cells, and their pools are continually adjusting as a result of the balance between supply and utilization of carbon at the whole plant level and of the cell sucrose–starch partition, which is under the control of several factors, including drought (Chaves et al., 1991). Nowadays, it is also recognized that sugars are important signalling molecules (see the recent review by Hanson and Smeekens, 2009) and may play important roles in the adaptive mechanisms to stress, including, for example, sucrose induction of stress defences (Ramel et al., 2009). Different neutral sugars and sugar intermediates are known to be sensed by specific sensors, although most still remain unidentified. It was proposed by Usadel et al. (2008) that plants respond in an acclimatory manner to the balance in the supply–use of sugars and that signalling events may be initiated by small changes in carbon status. The intracellular concentrations of sugars exert a feedback control on the rate of photosynthesis, these feedback mechanisms leading to significant changes in enzyme activities and gene expression (Koch, 1996). In general, source activities such as photosynthesis, nutrient mobilization, and export are up-regulated under low sugar conditions, as a result of gene de-repression, whereas an accumulation of sugars has the opposite effect (Pego et al., 2000). Contrastingly, sink activities such as growth and storage are up-regulated under carbon abundance. Eberhard et al. (2008) suggest that re-modelling of photosynthesis induced by a high sugar content may play an important role in minimizing the deleterious effects of excess light under conditions in which no net photosynthesis is required (e.g. when growth is arrested, as may happen in the early stages of water deficits). Moreover, quantitative studies of biochemical and physiological traits in plants under stress revealed that pre-stress sugar concentrations were correlated with subsequent stress tolerance (Ramel et al., 2009). Among the different sugar traits, the sucrose concentration at the end of the day was found to be particularly important for stress tolerance. A recent study by Meyer et al. (2007) also suggests that metabolic signals and not the availability of metabolic substrates determine the growth rate, with fast-growing species using available carbohydrates in growth and thus having low tissue sugar levels. In contrast, slow-growing plants save carbon resources that might be used under unfavourable conditions. Strong evidence is also available for the role of sugar signalling in the regulation of diurnal gene expression, with sugar content affecting 25–50% of transcripts that are subjected to circadian regulation (Blasing et al., 2005). This also explains major sugar control of photosynthetic activity during the day/night cycles, with photosynthesis-related transcripts displaying the lowest abundance during the night. Sugar metabolism in plants is highly dynamic as it varies with the stage of development and in response to biotic and abiotic stress (Rolland et al., 2006). For example, phloem carbohydrates are known to control reproductive development. Indeed, starch mobilization and an increase in leaf carbohydrate export to the shoot apical meristem underlie flowering induction (Corbesier et al., 2002). Figure 1 show that starch is a connection point for drought, photosynthesis, ABA, and ROS, and in Arabidopsis transcription of several amylase, sucrose synthase, and neutral invertase genes correlates with ABI1 and with GBF3 genes, while in another group it is possible to associate amylase and catalase gene transcription. Under water limitation, the balance between photosynthetic carbon uptake and the use of photoassimilates by the sinks is affected, leading to alterations in the pools of carbon (sugars) in the various compartments of the plant. The concentration of soluble sugars in leaves may increase (under the initial stages of moderate stress; Pinheiro et al., 2001), stay constant, or decrease (under intense stress). Inhibition of growth and export explains why under low carbon assimilation sugars may increase in leaf blades (Chaves and Oliveira, 2004). Starch synthesis is generally repressed under water deficits (Chaves, 1991), but there are indications that in the early stages of water stress a transitory increase in its concentration may occur (authors’ unpublished results). Alterations in the magnitude of the sugar pools are supported by changes in the enzyme activities involved in sugar- and starch-elated pathways such as α-amylase, sucrose synthase, and invertase, their association with drought and photosynthesis being highlighted in the literature survey (Fig. 1B). An increase in total acid invertase activity, coinciding with the rapid accumulation of glucose and fructose, was reported in leaves of maize plants (Trouverie et al., 2003) and of lupins (Pinheiro et al., 2001) subjected to drought. Transcription analysis (Supplementary Table S5 at JXB online) shows that the same stress has distinct effects within the gene family, which can be easily justified due to the fact that the invertase gene family comprises three types of invertase enzymes distinct in terms of cellular location and kinetic properties. However, Supplementary Table S5 also show that drought effects are variable for the same gene in several data sets, not allowing a general response to be extracted. Possible exceptions are one cell wall invertase and two neutral invertase genes. While acute stress lead to up-regulation of two invertases (cell wall AT3G13790 and neutral AT4G09510), slowly imposed stress leads to down-regulation. Although stress leads to the up-regulation of the neutral invertase gene At3G06500, acute stress induces higher alterations very early (1–2 h). It is well known that ADP-glucose pyrophosphorylase (AGPase; Supplementary Table S1, Supplementary Data at JXB online), a key enzyme in starch synthesis, is highly regulated by sugars (Geigenberger et al., 2005), with starch breakdown (namely that which occurs each night) being a major source of glucose signals (Rolland et al., 2006). On the other hand, KIN10 and KIN11 protein kinases are regarded as central in the coordination of several plant responses to sugars and stress, whereas bZIP TFs (such as GBF3, identified in Fig. 1, that can interact with the sucrose-sensing pathway) were shown to mediate effects of sugar signalling on gene expression and metabolite content (Hanson and Smeekens, 2009). The interconnection between starch and sugar metabolism is highlighted when looking at the correlated transcription of GBF3, ABI1, several amylases, sucrose synthase, and neutral invertase genes under stress. It can be concluded that the imbalance between sugar production (photosynthesis) and utilization (growth) observed under moderate drought constitutes an important signal to modulate photoassimilate investment in different organs of the plant (increasing, for example, the root versus the shoot) or, by accumulating in leaves, to play a protective role against oxidative stress, as is discussed below. The interplay between sugars, ROS, and hormones Figure 1A shows that ROS generation, sugar and starch metabolism, and ABA are clearly associated. Considering ETR1, a cross-talk between sugars and ROS is becoming apparent, with either converging or antagonistic effects (see the review by Couée et al., 2006). Using the tool ‘Correlated Gene Search’, it was possible to observe that stress induces correlated effects on several α-amylase and catalase genes, as well as catalase and β-amylases genes. Sugars are known to feed the oxidative pentose phosphate pathway that can contribute to ROS scavenging, as shown in mammalian cells, but can also increase ROS production through, for example, glucose auto-oxidation. Sucrose protection against oxidative stress seems to be partly due to activation of specific ROS scavenging systems (e.g. superoxide dismutase, Supplementary Tables S1, Supplementary Data at JXB online), with consequent reduction of oxidative damage, as confirmed by transcriptome analysis (Koch, 1996; Ramel et al., 2009). Fluctuations in sugar content that accompany alterations in the environment will therefore influence ROS production, placing sugars as key players in the redox balance in plants. Interestingly, the relationship between ROS production under excessive light and sugar accumulation may have been the basis for the selection of the parallel induction of gene expression by light and sugar in plant cells, as suggested by Couée et al. (2006). However, pathways of sugar-induced responses to stress remain to be characterized, and further investigation of the interactions between metabolic (sugar) and other stress signals needs to be pursued (Gibson 2000, 2005). The interplay between sugar and plant hormone pathways is also well established (Hanson and Smeekens, 2009). Sugars activate specific, or hormone cross-talk, transduction pathways in response to stress (Ramel et al., 2009). In particular, sugar is closely related to the ABA signalling cascade and to a lesser extent with auxins and ethylene signalling, which is exemplified by ABI1, ABI3, and ETR1 (Fig. 1; Supplementary Table S2 at JXB online). Furthermore, under stress, co-expression was found for one sucrose synthase gene and ABI3, while ABI1 gene was co-expressed with the genes of one neutral invertase, two sucrose synthases, one β-amylase, and the GBF3 TF. Such an association can lead to amplification of the signals as, for example, sugars travelling in the xylem of droughted plants are likely to exert an influence on stomatal sensitivity to ABA and, on the other hand, ABA can regulate the activity and expression of invertases, enzymes that hydrolyse sucrose to its hexoses. Conclusions and the way foreward Great progress has occurred in recent years in elucidating the nature of the various factors affecting photosynthesis in plants subjected to water deficits. The alterations that do occur in response to stress comprise the restriction of CO2 diffusion to the chloroplast, as well as metabolic changes, including the modulation of the expression of photosynthesis-related genes. However, when trying to make use of publicly available data to establish which events are regulated by and/or regulate photosynthesis, the lack of stress characterization is immediately revealed, impairing the possibility to compare and integrate data. From the meta-analysis of the literature, it was not possible to establish a firm answer regarding how common are the sets of metabolic responses (genes and proteins) that are active in regulating photosynthetic activity under water stress. Exceptions are ABI3 (down-regulated under stress) that responds to both auxin and ABA, and the serine/threonine phosphatase ABI1 (up-regulated under stress), which acts as a negative regulator of ABA promotion of stomatal closure. This highlights the role of post-translational regulation of protein activity in drought responses. The meta-analysis also reveals how interconnected sugar and starch metabolism (directly affected by the photosynthetic performance) are with hormone and ROS pathways. The interconnections between the different pathways that act on photosynthesis in response to dehydration are being unravelled and indicate a multitude of responses acting in parallel, which may explain the flexibility and resilience of photosynthesis under drought as well as the diversity of responses across species. On the other hand, as rightly pointed out by Hanson and Smeekens (2009), sugar signalling research is likely to present new opportunities for crop improvement, by acting on pathways determining source–sink relationships or spending–saving strategies in plants under stress. Ultimately, many questions remain unresolved regarding carbon assimilation (and plant) response to drought, partially associated with discrepancies observed with different species and experimental conditions. The physiological significance (stress response and/or tolerance) of alterations in expression observed in many drought-responsive genes, including those related to photosynthesis, are still not fully understood. Moreover, post-transcriptional and post-translational studies have been largely neglected. Proteomic and metabolomic approaches will have to be reinforced in order to obtain information on the relevance of the different metabolic responses for photosynthetic acclimation to drought. Furthermore, such approaches need to consider and integrate the physiological data in order to allow the full use of the computational methods being developed. Through the integration of multilevel data that can be compared, it will be possible to provide insights and directions on future research. We are grateful to Ariadne Genomics Inc. and Dr Oliver Braun for the opportunity to use the Pathway Studio trial version and for the most helpful WebEx training sessions. References Acharya BR,  Assmann SM.  Hormone interactions in stomatal function,  Plant Molecular Biology ,  2009, vol.  69 (pg.  451- 462) Google Scholar CrossRef Search ADS PubMed  Akiyama K,  Chikayama E,  Yuasa H,  Shimada Y,  Tohge T,  Shinozaki K,  Hirai MY,  Sakurai T,  Kikuchi J,  Saito K.  PRIMe: a Web site that assembles tools for metabolomics and transcriptomics,  In Silico Biology ,  2008, vol.  8 pg.  0027  Antonio C,  Pinheiro C,  Chaves MM,  Ricardo CP,  Ortuno MF,  Thomas-Oates J.  Analysis of carbohydrates in Lupinus albus stems on imposition of water deficit, using porous graphitic carbon liquid chromatography-electrospray ionization mass spectrometry,  Journal of Chromatography A ,  2008, vol.  1187 (pg.  111- 118) Google Scholar CrossRef Search ADS PubMed  Atkin OK,  Macherel D.  The crucial role of plant mitochondria in orchestrating drought tolerance,  Annals of Botany ,  2009, vol.  103 (pg.  581- 597) Google Scholar CrossRef Search ADS PubMed  Bartels D,  Salamini F.  Desiccation tolerance in the resurrection plant Craterostigma plantagineum. A contribution to the study of drought tolerance at the molecular level,  Plant Physiology ,  2001, vol.  127 (pg.  1346- 1353) Google Scholar CrossRef Search ADS PubMed  Biehler K,  Fock H.  Evidence for the contribution of the Mehler-peroxidase reaction in dissipating excess electrons in drought-stressed wheat,  Plant Physiology ,  1996, vol.  112 (pg.  265- 272) Google Scholar CrossRef Search ADS PubMed  Blasing OE,  Gibon Y,  Gunther M,  Hohne M,  Morcuende R,  Osuna D,  Thimm O,  Usadel B,  Scheible WR,  Stitt M.  Sugars and circadian regulation make major contributions to the global regulation of diurnal gene expression in Arabidopsis,  The Plant Cell ,  2005, vol.  17 (pg.  3257- 3281) Google Scholar CrossRef Search ADS PubMed  Blum A,  Sinmena B,  Mayer J,  Golan G,  Shpiler L.  Stem reserve mobilization supports wheat-grain filling under heat stress,  Australian Journal of Plant Physiology ,  1994, vol.  21 (pg.  771- 781) Google Scholar CrossRef Search ADS   Bolouri-Moghaddam MR,  Le Roy K,  Xiang L,  Rolland F,  Van den Ende W.  Sugar signalling and antioxidant network connections in plant cells,  FEBS Journal ,  2010, vol.  277 (pg.  2022- 2037) Google Scholar CrossRef Search ADS PubMed  Brady SM,  Sarkar SF,  Bonetta D,  McCourt P.  The ABSCISIC ACID INSENSITIVE 3 (ABI3) gene is modulated by farnesylation and is involved in auxin signaling and lateral root development in Arabidopsis,  The Plant Journal ,  2003, vol.  34 (pg.  67- 75) Google Scholar CrossRef Search ADS PubMed  Bray EA.  Molecular responses to water deficit,  Plant Physiology ,  1993, vol.  103 (pg.  1035- 1040) Google Scholar CrossRef Search ADS PubMed  Buchanan BB,  Balmer Y.  Redox regulation: a broadening horizon,  Annual Review of Plant Biology ,  2005, vol.  56 (pg.  187- 220) Google Scholar CrossRef Search ADS PubMed  Bunce JA.  Use of the response of photosynthesis to oxygen to estimate mesophyll conductance to carbon dioxide in water-stressed soybean leaves,  Plant, Cell and Environment ,  2009, vol.  32 (pg.  875- 881) Google Scholar CrossRef Search ADS   Chaves MM.  Effects of water deficits on carbon assimilation,  Journal of Experimental Botany ,  1991, vol.  42 (pg.  1- 16) Google Scholar CrossRef Search ADS   Chaves MM,  Flexas J,  Pinheiro C.  Photosynthesis under drought and salt stress: regulation mechanisms from whole plant to cell,  Annals of Botany ,  2009, vol.  103 (pg.  551- 560) Google Scholar CrossRef Search ADS PubMed  Chaves MM,  Oliveira MM.  Mechanisms underlying plant resilience to water deficits: prospects for water-saving agriculture,  Journal of Experimental Botany ,  2004, vol.  55 (pg.  2365- 2384) Google Scholar CrossRef Search ADS PubMed  Chaves MM,  Pereira JS,  Maroco J.  Understanding plant response to drought—from genes to the whole plant,  Functional Plant Biology ,  2003, vol.  30 (pg.  239- 264) Google Scholar CrossRef Search ADS   Chaves MM,  Pereira JS,  Maroco JP,  Rodrigues ML,  Ricardo CPP,  Osório ML,  Carvalho I,  Faria T,  Pinheiro C.  How plants cope with water stress in the field: photosynthesis and growth,  Annals of Botany ,  2002, vol.  89 (pg.  907- 916) Google Scholar CrossRef Search ADS PubMed  Cooper K,  Farrant JM.  Recovery of the resurrection plant Craterostigma wilmsii from desiccation: protection versus repair,  Journal of Experimental Botany ,  2002, vol.  53 (pg.  1805- 1813) Google Scholar CrossRef Search ADS PubMed  Corbesier L,  Bernier G,  Perilleux C.  C: N ratio increases in the phloem sap during floral transition of the long-day plants Sinapis alba and Arabidopsis thaliana,  Plant and Cell Physiology ,  2002, vol.  43 (pg.  684- 688) Google Scholar CrossRef Search ADS PubMed  Couée I,  Sulmon C,  Gouesbet G,  El Amrani A.  Involvement of soluble sugars in reactive oxygen species balance and responses to oxidative stress in plants,  Journal of Experimental Botany ,  2006, vol.  57 (pg.  449- 459) Google Scholar CrossRef Search ADS PubMed  Dace H,  Sherwin HW,  Illing N,  Farrant JM.  Use of metabolic inhibitors to elucidate mechanisms of recovery from desiccation stress in the resurrection plant Xerophyta humilis,  Plant Growth Regulation ,  1998, vol.  24 (pg.  171- 177) Google Scholar CrossRef Search ADS   David TS,  Henriques MO,  Kurz-Besson C, et al.  Water use strategies in two co-occurring Mediterranean evergreen oaks: surviving the summer drought,  Tree Physiology ,  2007, vol.  27 (pg.  793- 803) Google Scholar CrossRef Search ADS PubMed  Demmig-Adams B,  Adams WWIII.  The role of xanthophyll cycle carotenoids in the protection of photosynthesis,  Trends in Plant Science ,  1996, vol.  1 (pg.  21- 26) Google Scholar CrossRef Search ADS   Demmig-Adams B,  Adams WWIII,  Mattoo A. ,  Photoprotection, photoinhibition, gene regulation and environment. Advances in photosynthesis and respiration ,  2006, vol.  Vol. 21  Dordrecht Springer Deyholos MK.  Making the most of drought and salinity transcriptomics,  Plant, Cell and Environment ,  2010, vol.  33 (pg.  648- 654) Google Scholar CrossRef Search ADS   Dodd AN,  Gardner MJ,  Hotta CT, et al.  The Arabidopsis circadian clock incorporates a cADPR-based feedback loop,  Science ,  2007, vol.  318 (pg.  1789- 1792) Google Scholar CrossRef Search ADS PubMed  Dodd IC.  Hormonal interactions and stomatal responses,  Journal of Plant Growth Regulation ,  2003, vol.  22 (pg.  32- 46) Google Scholar CrossRef Search ADS   Eberhard S,  Finazzi G,  Wollman F- A.  The dynamics of photosynthesis,  Annual Review of Genetics ,  2008, vol.  42 (pg.  463- 515) Google Scholar CrossRef Search ADS PubMed  Flexas J,  Bota J,  Galmés J,  Medrano H,  Ribas-Carbó M.  Keeping a positive carbon balance under adverse conditions: responses of photosynthesis and respiration to water stress,  Physiologia Plantarum ,  2006, vol.  127 (pg.  343- 352) Google Scholar CrossRef Search ADS   Flexas J,  Bota J,  Loreto F,  Cornic G,  Sharkey TD.  Diffusive and metabolic limitations to photosynthesis under drought and salinity in C3 plants,  Plant Biology ,  2004, vol.  6 (pg.  269- 279) Google Scholar CrossRef Search ADS PubMed  Flexas J,  Galmes J,  Ribas-Carbo M,  Medrano H.  Lambers H,  Ribas-Carbo M.  The effects of water stress on plant respiration,  Plant respiration: from cell to ecosystem ,  2005 Dordrecht Springer-Verlag(pg.  85- 94) Flexas J,  Ribas-Carbó M,  Bota J,  Galmés J,  Henkle M,  Martínez-Cañellas S,  Medrano H.  Decreased Rubisco activity during water stress is not induced by decreased relative water content but related to conditions of low stomatal conductance and chloroplast CO2concentration,  New Phytologist ,  2006, vol.  172 (pg.  73- 82) Google Scholar CrossRef Search ADS PubMed  Flexas J,  Ribas-Carbo M,  Diaz-Espejo A,  Galmés J,  Medrano H.  Mesophyll conductance to CO2: current knowledge and future prospects,  Plant, Cell and Environment ,  2008, vol.  31 (pg.  602- 612) Google Scholar CrossRef Search ADS   Foyer CH,  Noctor G.  Redox regulation in photosynthetic organisms: signaling, acclimation, and practical implications,  Antioxidants and Redox Signaling ,  2009, vol.  11 (pg.  861- 905) Google Scholar CrossRef Search ADS PubMed  Galmés J,  Flexas J,  Savé R,  Medrano H.  Water relations and stomatal characteristics of Mediterranean plants with different growth forms and leaf habits: responses to water stress and recovery,  Plant and Soil ,  2007, vol.  290 (pg.  139- 155) Google Scholar CrossRef Search ADS   Galmés J,  Medrano H,  Flexas J.  Photosynthetic limitations in response to water stress and recovery in Mediterranean plants with different growth forms,  New Phytologist ,  2007, vol.  175 (pg.  81- 93) Google Scholar CrossRef Search ADS PubMed  Galmés J,  Ribas-Carbo M,  Medrano H,  Flexas J.  Response of leaf respiration to water stress in Mediterranean species with different growth forms,  Journal of Arid Environments ,  2007, vol.  68 (pg.  206- 222) Google Scholar CrossRef Search ADS   Galmés J,  Ribas-Carbó M,  Medrano H,  Flexas J.  Rubisco activity in Mediterranean species is regulated by the chloroplastic CO2 concentration under water stress,  Journal of Experimental Botany ,  2010, vol.  62 (pg.  653- 665) Google Scholar CrossRef Search ADS PubMed  Garcia-Plazaola JI,  Hernández A,  Olano JM,  Becerril JM.  The operation of the lutein epoxide cycle correlates with energy dissipation,  Functional Plant Biology ,  2003, vol.  30 (pg.  319- 324) Google Scholar CrossRef Search ADS   Gazanchian A,  Hajheidari M,  Sima NK,  Ghasem Hosseini Salekdeh GH.  Proteome response of Elymus elongatum to severe water stress and recovery,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  291- 300) Google Scholar CrossRef Search ADS PubMed  Geigenberger P,  Kolbe A,  Tiessen A.  Redox regulation of carbon storage and partitioning in response to light and sugars,  Journal of Experimental Botany ,  2005, vol.  56 (pg.  1469- 1479) Google Scholar CrossRef Search ADS PubMed  Genty B,  Briantais JM,  Baker JM.  The relationship between the quantum yield of photosynthetic electron transport and quenching of chlorophyll fluorescence,  Biochimica et Biophysica Acta ,  1989, vol.  990 (pg.  87- 92) Google Scholar CrossRef Search ADS   Ghannoum O.  C4 photosynthesis and water stress,  Annals of Botany ,  2009, vol.  103 (pg.  635- 644) Google Scholar CrossRef Search ADS PubMed  Gibson SI.  Plant sugar-response pathways. Part of a complex regulatory web,  Plant Physiology ,  2000, vol.  124 (pg.  1532- 1539) Google Scholar CrossRef Search ADS PubMed  Gibson SI.  Control of plant development and gene expression by sugar signaling,  Current Opinion in Plant Biology ,  2005, vol.  8 (pg.  93- 102) Google Scholar CrossRef Search ADS PubMed  Gimeno TE,  Sommerville KE,  Valladares F,  Atkin OK.  Homeostasis of respiration under drought and its important consequences for foliar carbon balance in a drier climate: insights from two contrasting Acacia species,  Functional Plant Biology ,  2010, vol.  37 (pg.  323- 333) Google Scholar CrossRef Search ADS   Grassi G,  Magnani F.  Stomatal, mesophyll conductance and biochemical limitations to photosynthesis as affected by drought and leaf ontogeny in ash and oak trees,  Plant, Cell and Environment ,  2005, vol.  28 (pg.  834- 849) Google Scholar CrossRef Search ADS   Gratani L,  Varone L,  Bonito A.  Environmental induced variations in leaf dark respiration and net photosynthesis of Quercus ilex L,  Photosynthetica ,  2007, vol.  45 (pg.  633- 636) Google Scholar CrossRef Search ADS   Hanson J,  Smeekens S.  Sugar perception and signaling—an update,  Current Opinion in Plant Biology ,  2009, vol.  12 (pg.  562- 567) Google Scholar CrossRef Search ADS PubMed  Harbinson J,  Genty B,  Baker NR.  The relationship between CO2 assimilation and electron transport in leaves,  Photosynthesis Research ,  1990, vol.  25 (pg.  199- 212) Google Scholar CrossRef Search ADS PubMed  Hayano-Kanashiro C,  Calderón-Vázquez C,  Ibarra-Laclette E,  Herrera-Estrella L,  Simpson J.  Analysis of gene expression and physiological responses in three Mexican maize landraces under drought stress and recovery irrigation,  PLoS ONE ,  2009, vol.  4 pg.  e7531  Google Scholar CrossRef Search ADS PubMed  Hirayama T,  Shinozaki K.  Perception and transduction of abscisic acid signals: keys to the function of the versatile plant hormone ABA,  Trends in Plant Science ,  2007, vol.  12 (pg.  343- 351) Google Scholar CrossRef Search ADS PubMed  Hoffmann R.  A wiki for the life sciences where authorship matters,  Nature Genetics ,  2008, vol.  40 (pg.  1047- 1051) Google Scholar CrossRef Search ADS PubMed  Huang D,  Wu W,  Abrams SR,  Cutler AJ.  The relationship of drought-related gene expression in Arabidopsis thaliana to hormonal and environmental factors,  Journal of Experimental Botany ,  2008, vol.  59 (pg.  2991- 3007) Google Scholar CrossRef Search ADS PubMed  Hummel I,  Pantin F,  Sulpice R, et al.  Arabidopsis plants acclimate to water deficit at low cost through changes of carbon usage: an integrated perspective using growth, metabolite, enzyme, and gene expression analysis,  Plant Physiology ,  2010, vol.  154 (pg.  357- 372) Google Scholar CrossRef Search ADS PubMed  Jaspers P,  Kangasjarvi J.  Reactive oxygen species in abiotic stress signaling,  Physiologia Plantarum ,  2010, vol.  138 (pg.  405- 413) Google Scholar CrossRef Search ADS PubMed  Jones HG.  Monitoring plant and soil water status: established and novel methods revisited and their relevance to studies of drought tolerance,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  119- 130) Google Scholar CrossRef Search ADS PubMed  Khandelwal A,  Cho SH,  Marella H,  Sakata Y,  Perroud P-F,  Pan A,  Quatrano RS.  The hormone pathway that stabilizes seeds may have served more primitive seedless plants in supporting desiccation tolerance,  Science ,  2010, vol.  327 pg.  546  Google Scholar CrossRef Search ADS PubMed  Kirschbaum MUF.  Recovery of photosynthesis from water stress in Eucalyptus pauciflora—a process in two stages,  Plant, Cell and Environment ,  1988, vol.  11 (pg.  685- 694) Google Scholar CrossRef Search ADS   Koch KE.  Carbohydrate-modulated gene expression in plants,  Annual Review of Plant Physiology and Plant Molecular Biology ,  1996, vol.  47 (pg.  509- 540) Google Scholar CrossRef Search ADS PubMed  Lambers H,  Robinson SA,  Ribas-Carbo M.  Lambers H,  Ribas-Carbo M.  Regulation of respiration in vivo,  Plant respiration: from cell to ecosystem. Advances in photosynthesis and respiration series ,  2005, vol.  Vol. 18  Dordrecht Springer(pg.  1- 15) Lawlor DW.  Musings about the effects of environment on photosynthesis,  Annals of Botany ,  2009, vol.  103 (pg.  543- 549) Google Scholar CrossRef Search ADS PubMed  Lawlor DW,  Tezara W.  Causes of decreased photosynthetic rate and metabolic capacity in water-deficient leaf cells: a critical evaluation of mechanisms and integration of processes,  Annals of Botany ,  2009, vol.  103 (pg.  561- 579) Google Scholar CrossRef Search ADS PubMed  Maroco JP,  Pereira JS,  Chaves MM.  Stomatal responses to leaf-to-air vapour pressure deficit in Sahelian species,  Australian Journal of Plant Physiology ,  1997, vol.  24 (pg.  381- 387) Google Scholar CrossRef Search ADS   Maroco JP,  Rodrigues ML,  Lopes C,  Chaves MM.  Limitations to leaf photosynthesis in grapevine under drought—metabolic and modeling approaches,  Functional Plant Biology ,  2002, vol.  29 (pg.  1- 9) Google Scholar CrossRef Search ADS   Merlot S,  Gosti F,  Guerrier D,  Vavasseur A,  Giraudat J.  The ABI1 and ABI2 protein phosphatases 2C act in a negative feedback regulatory loop of the abscisic acid signalling pathway,  The Plant Journal ,  2001, vol.  25 (pg.  295- 303) Google Scholar CrossRef Search ADS PubMed  Mittler R.  Abiotic stress, the field environment and stress combination,  Trends in Plant Science ,  2006, vol.  11 (pg.  15- 19) Google Scholar CrossRef Search ADS PubMed  Mittler R,  Vanderauwera S,  Gollery M,  Van Breusegem F.  Reactive oxygen gene network of plants,  Trends in Plant Science ,  2004, vol.  9 (pg.  490- 498) Google Scholar CrossRef Search ADS PubMed  Miyashita K,  Tanakamaru S,  Maitani T,  Kimura K.  Recovery responses of photosynthesis, transpiration, and stomatal conductance in kidney bean following drought stress,  Environmental and Experimental Botany ,  2005, vol.  53 (pg.  205- 214) Google Scholar CrossRef Search ADS   Parcy F,  Giraudat J.  Interactions between the ABI1 and the ectopically expressed ABI3 genes in controlling abscisic acid responses in Arabidopsis vegetative tissues,  The Plant Journal ,  1997, vol.  11 (pg.  693- 702) Google Scholar CrossRef Search ADS PubMed  Parent B,  Hachez C,  Redondo E,  Simonneau T,  Chaumont F,  Tardieu F.  Drought and abscisic acid effects on aquaporin content translate into changes in hydraulic conductivity and leaf growth rate: a trans-scale approach,  Plant Physiology ,  2009, vol.  149 (pg.  2000- 2012) Google Scholar CrossRef Search ADS PubMed  Parry MAJ,  Andralojc PJ,  Khan S,  Lea P,  Keys AJ.  Rubisco activity: effects of drought stress,  Annals of Botany ,  2002, vol.  89 (pg.  833- 839) Google Scholar CrossRef Search ADS PubMed  Passioura J.  The drought environment: physical, biological and agricultural perspectives,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  113- 117) Google Scholar CrossRef Search ADS PubMed  Pego JV,  Kortstee AJ,  Huijser C,  Smeekens SCM.  Photosynthesis, sugars and the regulation of gene expression,  Journal of Experimental Botany ,  2000, vol.  51 (pg.  407- 416) Google Scholar CrossRef Search ADS PubMed  Pereira JS,  Chaves MM.  Smith JAC,  Griffiths H.  Plant water deficits in Mediterranean ecosystems,  Plant responses to water deficits—from cell to community ,  1993 Oxford BIOS Scientific(pg.  237- 251) Pfannschmidt T,  Brautigam K,  Wagner R,  Dietzel L,  Schröter Y,  Steiner S,  Nykytenko A.  Potential regulation of gene expression in photosynthetic cells by redox and energy state: approaches towards better understanding,  Annals of Botany ,  2009, vol.  103 (pg.  599- 607) Google Scholar CrossRef Search ADS PubMed  Pinheiro C,  Chaves MM,  Ricardo CP.  Alterations in carbon and nitrogen metabolism induced by water deficit in stem and leaves of Lupinus albus (L.),  Journal of Experimental Botany ,  2001, vol.  52 (pg.  1063- 1070) Google Scholar CrossRef Search ADS PubMed  Pinheiro C,  Kehr J,  Ricardo CP.  Effect of water stress on lupin stem protein analysed by two-dimensional gel electrophoresis,  Planta ,  2005, vol.  221 (pg.  716- 728) Google Scholar CrossRef Search ADS PubMed  Pourkeirandish M,  Komatsuda T.  The importance of barley genetics and domestication in a global perspective,  Annals of Botany ,  2007, vol.  100 (pg.  999- 1008) Google Scholar CrossRef Search ADS PubMed  Ramel F,  Sulmon C,  Gouesbet G,  Couee I.  Natural variation reveals relationships between pre-stress carbohydrate nutritional status and subsequent responses to xenobiotic and oxidative stress in Arabidopsis thaliana,  Annals of Botany ,  2009, vol.  104 (pg.  1323- 1337) Google Scholar CrossRef Search ADS PubMed  Ribas-Carbo M,  Taylor NL,  Giles L,  Busquets S,  Finnegan PM,  Day DA,  Lambers H,  Medrano H,  Berry JA,  Flexas J.  Effects of water stress on respiration in soybean (Glycine max. L.) leaves,  Plant Physiology ,  2005, vol.  139 (pg.  466- 473) Google Scholar CrossRef Search ADS PubMed  Robertson FC,  Skeffington AW,  Gardner MJ,  Webb AAR.  Interactions between circadian and hormonal signaling in plants,  Plant Molecular Biology ,  2009, vol.  69 (pg.  419- 427) Google Scholar CrossRef Search ADS PubMed  Rolland F,  Baena-Gonzalez E,  Sheen J.  Sugar sensing and signaling in plants: conserved and novel mechanisms,  Annual Review of Plant Biology ,  2006, vol.  57 (pg.  675- 709) Google Scholar CrossRef Search ADS PubMed  Saibo NJM,  Lourenço T,  Oliveira MM.  Transcription factors and regulation of photosynthetic and related metabolism under environmental stresses,  Annals of Botany ,  2009, vol.  103 (pg.  609- 623) Google Scholar CrossRef Search ADS PubMed  Sharp RE.  Interaction with ethylene: changing views on the role of abscisic acid in root and shoot growth responses to water stress,  Plant, Cell and Environment ,  2002, vol.  25 (pg.  211- 222) Google Scholar CrossRef Search ADS   Shinozaki K,  Yamaguchi-Shinozaki K.  Gene networks involved in drought stress response and tolerance,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  221- 227) Google Scholar CrossRef Search ADS PubMed  Slot M,  Zaragoza-Castells J,  Atkin OK.  Transient shade and drought have divergent impacts on the temperature sensitivity of dark respiration in leaves of Geum urbanum,  Functional Plant Biology ,  2008, vol.  35 (pg.  1135- 1146) Google Scholar CrossRef Search ADS   Sulpice R,  Pyl E-T,  Ishihara H, et al.  Starch as a major integrator in the regulation of plant growth,  Proceedings of the National Academy of Sciences, USA ,  2009, vol.  106 (pg.  10348- 10353) Google Scholar CrossRef Search ADS   Tezara W,  Mitchell VJ,  Driscoll SD,  Lawlor DW.  Water stress inhibits plant photosynthesis by decreasing coupling factor and ATP,  Nature ,  1999, vol.  401 (pg.  914- 917) Google Scholar CrossRef Search ADS   Trouverie J,  Thâevenot C,  Rocher JP,  Sotta B,  Prioul JL.  The role of abscisic acid in the response of a specific vacuolar invertase to water stress in the adult maize leaf,  Journal of Experimental Botany ,  2003, vol.  54 (pg.  2177- 2186) Google Scholar CrossRef Search ADS PubMed  Usadel B,  Blasing OE,  Gibon Y,  Retzlaff K,  Hoehne M,  Gunther M,  Stitt M.  Global transcript levels respond to small changes of the carbon status during progressive exhaustion of carbohydrates in Arabidopsis rosettes,  Plant Physiology ,  2008, vol.  146 (pg.  1834- 1861) Google Scholar CrossRef Search ADS PubMed  Wilkinson S,  Davies WJ.  ABA-based chemical signaling: the co-ordination of responses to stress in plants,  Plant, Cell and Environment ,  2002, vol.  25 (pg.  195- 210) Google Scholar CrossRef Search ADS   Wilkinson S,  Davies WJ.  Drought, ozone, ABA and ethylene: new insights from cell to plant to community,  Plant, Cell and Environment ,  2010, vol.  33 (pg.  510- 525) Google Scholar CrossRef Search ADS   Wingler A,  Quick WP,  Bungard RA,  Bailey KJ,  Lea PJ,  Leegood RC.  The role of photorespiration during drought stress: an analysis utilizing barley mutants with reduced activities of photorespiratory enzymes,  Plant, Cell and Environment ,  1999, vol.  22 (pg.  361- 373) Google Scholar CrossRef Search ADS   Wise RP,  Caldo RA,  Hong L,  Shen L,  Cannon EK,  Dickerson JA.  BarleyBase/PLEXdb: a unified expression profiling database for plants and plant pathogens,  Methods in Molecular Biology ,  2007, vol.  406 (pg.  347- 363) Google Scholar PubMed  © The Author [2010]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: [email protected]

Coello, Patricia;Hey, Sandra J.;Halford, Nigel G.

doi: 10.1093/jxb/erq331pmid: 20974737

Abstract Sucrose non-fermenting-1 (SNF1)-related protein kinases (SnRKs) take their name from their fungal homologue, SNF1, a global regulator of carbon metabolism. The plant family has burgeoned to comprise 38 members which can be subdivided into three sub-families: SnRK1, SnRK2, and SnRK3. There is now good evidence that this has occurred to allow plants to link metabolic and stress signalling in a way that does not occur in other organisms. The role of SnRKs, focusing in particular on abscisic acid-induced signalling pathways, salinity tolerance, responses to nutritional stress and disease, and the regulation of carbon metabolism and, therefore, yield, is reviewed here. The key role that SnRKs play at the interface between metabolic and stress signalling make them potential candidates for manipulation to improve crop performance in extreme environments. ABA, biotic stress, carbon metabolism, crop yield, plant nutrition, salt tolerance, signalling, stress Introduction Due to the close and direct dependence upon the land and agriculture of much of the African population, it is likely that changes in weather patterns and climate will be keenly felt by many on the continent. The Intergovernmental Panel on Climate Change (IPCC), in their report of 2007 on Regional Climatic Projections (Christensen et al., 2007), predicted that climate change will impact heavily on Africa over the coming century. It concludes that all of Africa is very likely to get warmer during this century, with the drier subtropical regions warming more than the moister tropics. In the west, east, south, and Saharan sub-regions, the projected median temperature increase in the 100 years between the 1980–1999 and 2080–2099 20-year averages, lies between 3 °C and 4 °C, roughly 1.5 times the global mean increase. Smaller increases, near 3 °C, are predicted in the equatorial and coastal areas, but larger increases, above 4 °C, are predicted for the Western Sahara. Annual rainfall is predicted to decrease in much of Mediterranean Africa, the Northern Sahara, and the winter rainfall region and western margins of Southern Africa, although increases in rainfall are predicted for East Africa. In addition, a general increase is predicted in the intensity of high rainfall events, in conjunction with a decrease in the number of rain days in regions of mean drying (Christensen et al., 2007). Decreasing rainfall and rising temperatures will increase both the requirement for irrigation and the rates of evapotranspiration, in turn exacerbating the risk of soil salination. Salts from precipitation and irrigation water remain in the soil and accumulate in the root zone when water evaporates from the soil or is taken up by the crop. If the salt is not leached from the soil by a sufficient flow of water, either from precipitation or by the application of excess irrigation water, the concentration increases until it reduces crop yields by impeding water uptake by the plant. Conversely, the more frequent and severe flooding events predicted for parts of East Africa could potentially lead to increased hypoxia of crop roots. These additional challenges will further compound those already faced by African agriculture, such as the often poor nutrient status of its older soils. In the face of these challenges, there is clearly scope for research into developing crops with improved resistance to nutrient limitation, drought, heat, salt, and flooding, to be exploited by African agriculture. As a further consequence of predicted climatic changes, it is also to be expected that crop diseases and insect pests may be able to spread to regions where they have not posed a serious problem in the past and where traditional agricultural practices may not be best suited to protect against them. Development of crops with improved disease resistance could offer a degree of protection from yield losses under these circumstances. Development of crops better able to tolerate adverse conditions would be of particular value in places like Africa where resources to ameliorate the environment, such as fertilizers, irrigation water or pesticides, are often difficult or uneconomic to access given the farming practices frequently employed. Many of the reductions in yield that occur in response to stress are evolutionary adaptations that enable the plant to conserve energy and resources in the expectation of better future conditions either for itself or its offspring, thereby increasing its long-term chances of survival. In the artificial conditions imposed by agriculture, depending on the particular crop product being produced, long-term survival strategies may be less important than the crop's value to the farmer. It is therefore likely that potential exists for manipulating plant responses to stress to improve yields significantly under adverse conditions. Interactions between metabolic and stress signalling networks Recent developments in plant cell signalling have highlighted intricate interconnections between metabolic regulation and stress signalling systems which could enable the development of crops better able to resist and adjust to environmental stresses, whilst maintaining yields. High salt concentrations, drought, and heat stress in plants have much in common. All lead to cellular osmotic stress and are associated with increases in the concentration of the plant hormone, abscisic acid (ABA), which acts as a trigger for a number of physiological responses that can help to ameliorate the effects of stress on the plant. It is becoming clear that metabolic regulation is yet another common factor linking these stresses. This is, perhaps, not surprizing, because plants modify the balance between soluble and insoluble compounds in the cell in order to cope with osmotic stress. Recently, considerable developments have been made in elucidating the signalling networks that have formed to enable metabolic and stress signalling pathways to interact and cross-talk (Hey et al., 2010). Early evidence of links between ABA, stress, and metabolic signalling pathways materialized when several mutant plants that were identified in screens for impairment in response to sugar, turned out also to be ABA signalling mutants (reviewed by Halford, 2006). This has since been further supported by evidence that the key metabolic regulator SnRK1 (sucrose non-fermenting-1-related protein kinase-1) is involved in stress signalling. SnRK1 regulates carbon metabolism, both through the modulation of enzyme activity by direct phosphorylation or redox activation of metabolic enzymes, and through the regulation of gene expression (reviewed by Halford and Hey, 2009). The enzymes that are known to be directly phosphorylated and inactivated by SnRK1 are 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMG-CoA reductase: sterol biosynthesis), sucrose phosphate synthase (SPS: sucrose synthesis), nitrate reductase (NR: nitrogen assimilation), trehalose-phosphate synthase (TPS: desiccation tolerance; signalling), and 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (F2KP: signalling; photosynthate partitioning). Note that in the case of NR, TPS, and F2KP, inactivation also requires the binding of a 14-3-3 protein. Another key enzyme in carbon metabolism, ADP-glucose pyrophosphorylase, is regulated by SnRK1 through both modulation of its redox state and gene expression, while sucrose synthase, α-amylase, and sugar-repressed/dark-induced asparagine synthetase are all regulated by SnRK1 at the level of gene expression. Furthermore, a recent transcriptomics study has shown that SnRK1 is involved in responses to sugar, darkness, and a range of stresses that limit photosynthesis and respiration, including herbicide [(3,4-dichlorophenyl)-1,1-dimethylurea, (DCMU)] treatment and flooding/hypoxia (Baena-Gonzáles et al., 2007). SnRK1 is closely related to the metabolic regulators of mammals (5'-AMP-activated protein kinase, AMPK) and yeast (sucrose non-fermenting-1, SNF1), with which it shares about 47% amino acid sequence identity and similar substrate specificity. However, in plants, the SnRK family of protein kinases has proliferated, with the sub-families SnRK2 and SnRK3 having emerged during plant evolution since the divergence of plants, animals, and fungi. Unlike SnRK1, which in Arabidopsis has only two active representatives, these sub-families are large and diverse with 10 and 25 members, respectively, in Arabidopsis (Halford and Hey, 2009). SnRK2s and 3s have diverged further from AMPK and SNF1 than have SnRK1s; it is suggested that they emerged as a result of duplication and then evolved rapidly, taking on new roles to enable plants to link metabolic and stress signalling (Halford and Hey, 2009). Compelling evidence suggests that members of the SnRK2 and SnRK3 sub-families are involved in signalling pathways that regulate plant responses to nutrient limitation, drought, cold, salt, and osmotic stresses. Expression of the entire SnRK2 sub-family of rice (10 members), for example, is induced by osmotic stress, and, of these, three are also induced by ABA (Kobayashi et al., 2004). A SnRK2 in Arabidopsis has also been shown to control stress responsive gene expression and improve drought tolerance when over-expressed (Umezawa et al., 2004), while PKABA1, a SnRK2 from wheat, mediates ABA-induced changes in gene expression in response to cold and other stresses (Gómez-Cadenas et al., 1999). Very recently, a central role for SnRK2 in ABA signalling has been discovered with the identification of PYR/PYL/RCAR proteins as ABA receptors and elucidation of the ABA signalling pathway. PYR/PYL/RCAR proteins bind to ABA and inhibit the activity of phosphatase PP2Cs encoded by ABSCISIC ACID-INSENSITIVE 1 and 2 (ABI1 and ABI2) (Leung et al., 1997). In the absence of ABA, these PP2Cs dephosphorylate and inactivate SnRK2, the phosphorylation of which is required to activate downstream transcription factors (ABF) (Ma et al., 2009; Park et al., 2009; Sheard and Zheng, 2009; Umezawa et al., 2009). Arabidopsis mutants defective in three SnRK2s, SnRK2.2, 2.3, and 2.6, have been shown to be almost completely insensitive to ABA (Fujii and Zhu, 2009). Several of the SnRK3-type protein kinases have also been implicated in stress responses. For example, one of the SnRK3s of Arabidopsis [SnRK3.11 or Salt Overly Sensitive 2 (SOS2)], is involved in conferring salt tolerance (Liu et al., 2000), whilst another (SnRK3.1) functions as a global regulator of ABA responses in a calcium-responsive negative regulatory loop controlling ABA sensitivity (Guo et al., 2002). SnRK3s are thought to be calcium-dependent because they interact with calcineurin B-like (CBL) calcium-binding proteins (Guo et al., 2002) and for this reason are also known as CBL-interacting kinases (CIPKs) (reviewed by Luan, 2009). The involvement of Ca2+ signalling in mediating responses to abiotic stresses, and particularly those mediated by ABA, is well known. Involvement of SnRKs in salinity tolerance Salinity has already become a serious risk to agricultural production, limiting plant growth and productivity worldwide. Salinity in the crop root zone is likely in drying areas of Africa because of the predicted reductions in rainfall and increased evapotranspiration caused by lower humidity (reviewed by Semenov and Halford, 2009). This will lead to reduced water uptake, with concomitant effects on growth and crop yield. It is also likely to activate SnRK stress signalling, potentially affecting carbon allocation. Sodium ions (Na+) are cytotoxic to plants when they accumulate to high concentrations, causing an ionic stress resulting from a solute imbalance and an osmotic stress resulting from reduced water availability (Silva and Gerós, 2009). Plant responses involve changes in the expression of multiple genes, some of which are considered to be part of the general stress response, others of which are regulated specifically in response to high salinity (Chinnusamy et al., 2004). High salinity produces dehydration, but dehydration can also be caused by drought and low temperatures, so some of the responses are common to these different abiotic stresses (Boudsocq and Lauriere, 2005; Shinozaki and Yamaguchi-Shinozaki, 2007). A critical component of the salt stress response is the maintenance of ion homeostasis. The role of SnRK3-type protein kinases in salt-stress responses To prevent Na+ accumulation in the cytoplasm, plants can reduce its entry into cells, activate its efflux from cells or compartmentalize it in the vacuole (Bertorello and Zhu, 2009). In Arabidopsis, ion homeostasis is mediated by the Salt Overly Sensitive (SOS) signalling pathway, which consists of three main components: SOS1, which is a Na+/H+ antiporter that effluxes excess ions out of the cytosol; SOS2, which is a SnRK3-type protein kinase, and SOS3, which is a Ca2+ sensor protein with an N-terminal myristoylation site and four Ca2+-binding EF hands (helix–loop–helix structures) (Sanchez-Barrena et al., 2005; Mahajan et al., 2008). SOS1 gene expression is up-regulated in response to salt stress and its over-expression improves salt tolerance in transgenic plants (Shi et al., 2003; Yang et al., 2009). By contrast, sos1 mutants are extremely sensitive to high salt concentrations and accumulate more salt than wild-type plants (Qiu et al., 2004). sos2 and sos3 mutants show a similar phenotype, suggesting that the three genes function in the same pathway (Mahajan et al., 2008). SOS2 has an N-terminal catalytic domain and a C-terminal regulatory domain, both of which are important for its function in salt tolerance. In vitro and in vivo experiments have shown that SOS2 directly interacts with SOS3, and that a 21-amino acid motif in the C-terminal regulatory domain (FISL motif) is sufficient for the interaction to occur (Guo et al., 2001). Moreover, the FISL motif serves as an autoinhibitory sequence, keeping SOS2 in an inactive state (Guo et al., 2001). Increases in Ca2+ concentration in response to salt stress can be detected by SOS3, which, in turn, binds to and activates the kinase domain of SOS2 by releasing the autoinhibitory effect of the FISL domain. Myristoylation of SOS3 enables the complex to associate with the cell membrane and lack of such a modification leads to failure of SOS3 to confer salt tolerance (Ishitani et al., 2000). The active SOS2/SOS3 complex phosphorylates and activates the SOS1 Na+/H+ antiporter. Other substrates of SOS2 may include vacuolar Na+/H+ antiporters and H+-ATPases (Silva and Gerós, 2009). All 25 members of the Arabidopsis SnRK3 family have a C-terminal FISL motif and expression analysis of seven of them showed four to be up-regulated by Na+, two to be repressed, and one to be unaffected, This differential regulation of expression was organ-specific. Furthermore, all seven protein kinases that were studied interacted with SOS3, although the interactions were weak compared with the SOS2–SOS3 interaction (Guo et al., 2001; Batistic and Kudla, 2009). SOS3 is a member of a small family of calcium sensors (CBL, calcineurin-B like proteins). Besides the SOS2 (CIPK24)/SOS3 (CBL4) partners, other CBL/SnRK3 partners have been shown to be involved in salt responses. SOS2 can also interact with CBL10, and the complex seems to regulate the transport of Na+ to the vacuole (Kim et al., 2007). It has also been shown that SOS2 phosphorylates CBL10 and that, in so doing, it stabilizes the complex at the membrane, so it has been proposed that SOS1 could also be a target of the SOS2/CBL10 complex (Lin et al., 2009). Interestingly, SOS3 is preferentially expressed in roots, whereas CBL10 is mainly expressed in shoots, so SOS2/SOS3 partners might function in conferring root salt tolerance whilst SOS2/CBL10 could confer shoot salt tolerance (Quan et al., 2007; Bertorello and Zhu, 2009; Lin et al., 2009). CBL2 interacts with and activates a SOS2-like protein kinase (PKS5), and this complex negatively regulates the activity of a plasma membrane H+-ATPase (Fugslang et al., 2007). CBL1 and CBL9 interact with CIPK23 and this complex increases the activity of AKT1, a plasma membrane K+ channel (Lin et al., 2009). The SOS pathway for salt tolerance appears to be conserved in other plant species. In rice the Sos1, Sos2, and Sos3 homologues have been identified and OsSos2 and OsSos3 also belong to small gene families, suggesting that some other SnRK3/CBL partners could also be involved in salt tolerance (Martinez-Atienza et al., 2007). In maize, a SOS3 homologue, ZmCBL4, can complement a sos3 mutant and reconstitute salt-tolerance responses (Wang et al., 2007) whilst a Sos2 homologue, ZmCIPK16 enhances salt tolerance when expressed in Arabidopsis (Zhao et al., 2009). Interconnection with other signalling networks can be inferred from the analysis of other interacting proteins. For example, interactions between SOS2 and members of the protein phosphatase 2C (PP2C) have been demonstrated using the yeast two-hybrid assay. Mutational analysis of the protein phosphatase interaction (PPI) domain, demonstrated that a region of 37 amino acids was sufficient and necessary for interaction with ABI2. The PPI domain was found to be conserved in all the SnRK3s that were analysed and differences in the interactions between the kinases and the ABI1- and ABI2-encoded PP2Cs were discovered (Ohta et al., 2003). Presently, it is not known if the phosphatase is able to dephosphorylate the kinase or if the kinase can phosphorylate the phosphatase. However, what is known is that binding of the phosphatase at the PPI domain might prevent interaction with the CBL protein and vice versa (Sanchez-Barrena et al., 2007). The functional relevance of this interaction clearly requires further study. The role of SnRK2-type protein kinases in salt-stress responses The SnRK2 family of Arabidopsis contains 10 members and, with the exception of SnRK2.9, all of them are activated by saline stress (Boudsocq et al., 2004). Furthermore, over-expression of one member of the family, SnRK2.8, has been shown to increase drought tolerance and up-regulate stress-induced genes (Umezawa et al., 2004). Activation of SnRK2 in response to hyperosmotic treatments depends on phosphorylation of a specific serine residue located in the activation loop, although some other sites may also be phosphorylated (Burza et al., 2006; Boudsocq et al., 2007). Other plant species also have members of the SnRK2 family that have been shown to be involved in salt signalling. In rice, for example, the entire SnRK2 family has been shown to be activated by hyperosmotic stress and this activation involves phosphorylation (Kobayashi et al., 2004). Interestingly, not all members of the family are activated similarly in response to various salt concentrations. Activation of one member, SAPK1, for example, has been observed at NaCl concentrations higher than 300 mM, whereas another, SAPK2, becomes active at lower concentrations. Domain exchange experiments have revealed that the C-terminal domain is responsible for these responses (Kobayashi et al., 2004). The C-terminal domain of SnRK2s is short and contains a characteristic acidic patch. Paradoxically, for understanding the evolution of the family, the acidic patch is highly aspartic acid-rich in some SnRK2s but glutamic acid-rich in others (Halford and Hardie, 1998). Members of the soybean SnRK2 family, SPK1 and SPK2, have also been shown to be activated by NaCl when expressed in yeast, although concentrations higher than 0.5 M are required, and to phosphorylate a soybean phosphatidylinositol transfer protein (Ssh1p) in response to saline stress. Ssh1p might be involved in phosphoinositide metabolism, playing an essential role in hyperosmotic signalling (Monks et al., 2001). In wheat, three SnRK2 family members have been shown to be induced by saline treatments. Expression of PKABA1, W55a, and TaSnRK2.4 is stimulated by high salt treatment and expression of W55a and TaSnRK2.4 in Arabidopsis has been shown to enhance salt tolerance (Holappa and Walker-Simmons, 1995; Xu et al., 2009; Mao et al., 2010). Interactions between ABA and stress signalling through SnRKs It has been known for some time that an important set of genes induced by drought, salt, and cold stress, are also activated by ABA. As already described, several studies have shown that Arabidopsis and rice SnRK2s are activated in response to salinity stress. Some are also regulated by ABA, although not all are, demonstrating that osmotic and ABA signalling networks are distinct. Analysis of the C-terminal domain of SnRK2 showed that this region was responsible for ABA activation (Boudsocq et al., 2004; Kobayashi et al., 2004). Furthermore, Arabidopsis SnRK2.6, which is activated by osmotic stress and is involved in stomatal closure in response to ABA (Merlot et al., 2002; Mustilli et al., 2002; Yoshida et al., 2002), has two regulatory domains at its C-terminus. When part of the Domain II was deleted, ABA-dependent activation was inhibited, whereas osmotic-dependent activation remained as normal. The PP2C encoded by ABI1 might play an important role in the ABA-dependent activation of SnRK2.6, since the abi1-1 mutation inhibits its ABA-dependent activation and the PP2C interacts directly with SnRK2.6 (Yoshida et al., 2006). Just recently, HAB1, another type of PP2C in Arabidopsis, was shown to regulate the abscisic acid-dependent activation of SnRK2.6 (Vlad et al., 2009). A simple model depicting the interaction of Ca2+, ABA, and SnRKs in salinity responses is shown in Fig. 1. Fig. 1. View largeDownload slide Schematic diagram showing calcium- and ABA-dependent and -independent signalling pathways involving SnRK1, SnRK2, and SnRK3. High accumulation of Na+ in the cytoplasm triggers a cytosolic Ca2+ signal, which is sensed by an SnRK3/calcineurin B-like (CBL) calcium-binding protein complex (SOS2/SOS3). ABA regulates the activity of SnRK2, which, in turn, activates AREBPs inducing gene expression. ABA might also regulate the activity of SnRK3 through the binding of PP2C. Other abiotic stresses such as nutrient deprivation induce or repress the expression/activity of SnRKs. However, additional components of the signal transduction pathways involved need to be identified. Fig. 1. View largeDownload slide Schematic diagram showing calcium- and ABA-dependent and -independent signalling pathways involving SnRK1, SnRK2, and SnRK3. High accumulation of Na+ in the cytoplasm triggers a cytosolic Ca2+ signal, which is sensed by an SnRK3/calcineurin B-like (CBL) calcium-binding protein complex (SOS2/SOS3). ABA regulates the activity of SnRK2, which, in turn, activates AREBPs inducing gene expression. ABA might also regulate the activity of SnRK3 through the binding of PP2C. Other abiotic stresses such as nutrient deprivation induce or repress the expression/activity of SnRKs. However, additional components of the signal transduction pathways involved need to be identified. Another route whereby the SnRKs could affect stress signalling is through ABA response element binding proteins (AREBPs), a family of bZIP transcription factors that are unique to plants and which regulate the expression of ABA responsive genes. AREBPs contain highly conserved SnRK1 target sites, and peptides with amino acid sequences based on these sites are very good substrates for phosphorylation by SnRK1 (Zhang et al., 2008). Like SnRK1, SnRK2-type protein kinases have also been shown to phosphorylate AREBPs (Kobayashi et al., 2005; Furihata et al., 2006). It is also possible that SnRK3s phosphorylate AREBPs because a calcium-dependent activity from stressed Arabidopsis seedlings has been shown to phosphorylate the same AREBP-based peptides, which would make AREBPs potential hubs in a signalling network where multiple pathways converge (Halford and Hey, 2009). Nutritional stress Drying and increasing salinity are not the only challenges faced by African farmers. African soils are very ancient and degraded and nutrient poor because of weathering and failure to replace nutrients that are removed in cropping (Keith Goulding, Rothamsted, personal communication). Artificial fertilizers are expensive and often difficult to obtain in remote areas, leaving many farmers to rely on traditional inputs from livestock, which can be limited and patchy in their availability. Nitrogen (N), phosphorus (P), and sulphur (S) are essential macronutrients required by plants for growth and productivity (Gojon et al., 2009). Although their availability in soils is generally low, their amount can fluctuate greatly due to factors such as soil type, soil pH, temperature, and precipitation. As sessile organisms, plants have developed adaptations on morphological, biochemical, and molecular levels that allow them to cope with nutrient limitation. These adaptive responses include improvements in nutrient uptake and the modulation of metabolic processes to optimize the use of assimilated nutrients (Poirier and Bucher, 2002). Increasing evidence has shown that members of the different SnRK sub-families play central roles in deciphering stress signals and their participation in NPS signalling has been documented. Much of the work on N sensing has focused on the induction of N metabolism after the addition of nitrate (Schachtman and Shin, 2007). Nitrate reductase (NR) is the first enzyme involved in nitrate assimilation; it catalyses the transfer of two electrons from NAD(P)H to nitrate, which is further reduced to nitrite and ammonium (Kaiser and Huber, 2001). Regulation of NR occurs on at least two levels: NR genes can be induced by addition of N to starved plants (Wang et al., 2000, 2003), and the activity of the NR enzyme can be rapidly and reversibly modulated by phosphorylation (Lillo, 2008). Phosphorylation occurs on a light/dark cycle, with the active, dephosphorylated form of NR being mainly present in the day, whereas the inactive, phosphorylated enzyme is present at night (Kaiser and Huber, 1994). NR is phosphorylated and inactivated by Ca2+-dependent and -independent protein kinases, including SnRK1 (Douglas et al., 1997; Sugden et al., 1999b). Inactivation by SnRK1 occurs in a two-step mechanism, in which phosphorylation takes place first, followed by the binding of 14-3-3 proteins to the regulatory phosphorylation site (Ser 543) (Moorhead et al., 1999). SnRK1 is a heterotrimeric complex comprising a catalytic α subunit with β and γ regulatory subunits. Uniquely, plants also have a protein that appears to be a fusion of the two regulatory subunits, called the βγ subunit (this is encoded by a single gene; it does not arise from a post-translational fusion) and there is further complexity in that there are three different forms of the β subunit. At the transcriptional level, all of these subunits are modulated by nutrient availability, as well as other stresses (Buitink et al., 2004; Polge et al., 2008). Evaluation of publicly-available microarray expression data of the genes that encode the SnRK1 complex subunits using Genevestigator shows that low N conditions have a strong influence on the expression of the α, β1, and γ subunits (Fig. 2). Variations in the amount of some of the β subunits have also been detected during the light/dark transition, the β1 and β3 genes being up-regulated in the dark (Polge et al., 2008). Experiments using the two hybrid system to characterize the interaction between different β subunits and NR revealed that both β1 and β2 subunits were associated with the enzyme (Polge et al., 2008; Li et al., 2009). These results support the idea that the β subunits could interact specifically with different targets, as has been proposed in yeast (Vincent et al., 2001). Fig. 2. View largeDownload slide Meta-analysis of microarray data for expression of genes encoding SnRK1, SnRK2, and SnRK3 in Arabidopsis in response to nitrate starvation, different combinations of nitrate and sucrose, phosphorus (P) deficiency, and sulphate deprivation. Results are shown for genes encoding all three subunits (α, β, and γ/βγ) of the SnRK1 complex (the α subunit is encoded by AKIN10 and AKIN11), all ten members of the SnRK2 family, and all 25 members of the SnRK3 family. The SnRK2s have an N-terminal catalytic domain and a C-terminal domain containing an acidic ‘patch’ that is important for activation, while the SnRK3s have an N-terminal catalytic domain and a C-terminal domain containing a FISL motif that is important for binding the calcineurin B-like (CBL) calcium-binding protein partner. These structures are represented above the expression data. The data were obtained using the microarray database Meta-Analyzer provided by Genevestigator (Zimmermann et al., 2004). Fig. 2. View largeDownload slide Meta-analysis of microarray data for expression of genes encoding SnRK1, SnRK2, and SnRK3 in Arabidopsis in response to nitrate starvation, different combinations of nitrate and sucrose, phosphorus (P) deficiency, and sulphate deprivation. Results are shown for genes encoding all three subunits (α, β, and γ/βγ) of the SnRK1 complex (the α subunit is encoded by AKIN10 and AKIN11), all ten members of the SnRK2 family, and all 25 members of the SnRK3 family. The SnRK2s have an N-terminal catalytic domain and a C-terminal domain containing an acidic ‘patch’ that is important for activation, while the SnRK3s have an N-terminal catalytic domain and a C-terminal domain containing a FISL motif that is important for binding the calcineurin B-like (CBL) calcium-binding protein partner. These structures are represented above the expression data. The data were obtained using the microarray database Meta-Analyzer provided by Genevestigator (Zimmermann et al., 2004). Analysis of the expression of the SnRK3 sub-family in Arabidopsis shows that most SnRK3s are affected by N starvation and different combinations of N and sucrose, but that the expression of some increases while the expression of others decrease under the same treatment (Fig. 2). These results are consistent with other studies that have demonstrated the participation of SnRK3s in N and other nutrient signalling processes. WPK4, for example, a SnRK3 from wheat, is induced by low nutrient availability, in particular N, P, and S (Sano and Youssefian, 1994). WPK4 interacts with TaWIN1, which is a 14-3-3 protein that is able to bind WPK4-phosphorylated NR (Ikeda et al., 2000). A model has been proposed in which WPK4 autophosphorylates on its catalytic domain in response to nutrient limitation, allowing the binding of the 14-3-3 protein. WPK4 then phosphorylates NR and transfers the 14-3-3 protein to inactivate the enzyme (Ikeda et al., 2000). Just recently, another member of the SnRK3 sub-family, CIPK6, has been identified as being involved in N signalling: CIPK6 is partly responsible for the regulation of expression of some of the genes involved in N assimilation and transport in Arabidopsis (Hu et al., 2009). SnRK2-type protein kinases are also regulated at the transcript level by low N (Fig. 2) and over-expression of one, SnRK2.8, has been shown to lead to increased biomass accumulation in plants under nutrient-deprived conditions, particularly N, P, and potassium (K). Phosphoproteomic analysis of the Arabidopsis snrk2.8 mutant showed one of its in vivo targets to be a 14-3-3 protein. The expression of SnRK2.8 is regulated diurnally, with its activity enhanced during the day, leading to phosphorylation of the 14-3-3 protein and preventing the 14-3-3 protein's interaction with NR (Shin et al., 2007). S is another of the macronutrients that plants require for growth and development. The preferred form of S that is assimilated by plants is the sulphate ion, SO42−. Most organisms have a limited capacity to store S and thus require a continuous supply of S-containing nutrients for survival. Once sulphate is taken up, it is assimilated to cysteine and then converted into methionine, glutathione (GSH) or other S-containing organic compounds (Schachtman and Shin, 2007). During the early stages of S deprivation, plants activate mechanisms which increase S acquisition from the soil, such as arylsulphatase and sulphate transporters (Gonzalez-Ballester et al., 2008). However, if they still cannot get enough sulphate, leading to a decrease in the plant S content, this will lead to changes that reduce metabolism and growth rate (Nikiforova et al., 2003) and affect composition (Muttucumaru et al., 2006; Elmore et al., 2007; Curtis et al., 2009). The first report on the involvement of SnRKs in S deprivation responses came from Chlamydomonas studies. Chlamydomonas cells have similar S limitation responses to those exhibited by vascular plants (Davies et al., 1999) and SnRK2.2 (formerly named Sac3) regulates some of the responses to S starvation. snrk2.2 mutants exhibit both positive and negative effects on regulation because they are unable to repress arylsulphatase activity fully when grown in nutrient-replete medium and are also unable to activate sulphate uptake fully upon S starvation (Davies et al., 1999). In addition, the concentration of chloroplast RNA (cpRNA) in S-starved cells is associated with SnRK2.2 activity: mutant cells show higher levels of cpRNA than wild-type after several hours of S starvation. It has been proposed that a chloroplast sigma factor that controls chloroplast transcription might be deactivated by SnRK2.2 (Irihimovitch and Stern, 2006). Another SnRK2 family member in Chlamydomonas, SnRK2.1, is also involved in regulating S-responsive gene expression, playing a crucial role in the control of S deprivation responses. snrk2.1 mutants show little or no increase in the levels of known S deprivation-responsive transcripts when starved for S, and after transfer to medium lacking S they bleach rapidly. Furthermore, snrk2.1 is epistatic to snrk2.2, reflecting its key position in the control of S-deprivation responses (Gonzalez-Ballester et al., 2008). In Arabidopsis plants, there is little evidence that members of the SnRK1, SnRK2 or SnRK3 sub-families are regulated at the transcriptional level by S deprivation (Fig. 2). In an attempt to determine whether SnRK2s play an important role in S deprivation responses in the way that they do in Chlamydomonas, mutations were induced in snrk2.3, the gene with most similarity with the Chlamydomonas SnRK2.2 gene. The Arabidopsis mutant did not show the usual induction of sultr2.2 sulphate transporter genes under S deprivation, but no other S-starvation responses were affected (Kimura et al., 2006). These results may indicate that SnRK2s have a less important role in S signalling in higher plants than in Chlamydomonas, or that other members of the Arabidopsis SnRK2 family were able to fulfil the function of SnRK2.3 in its absence. P, like N and S, is a vital macronutrient for living organisms. It forms part of important macromolecules such as nucleic acids, phospholipids, and adenosine triphosphate (ATP), and is directly involved in the regulation of diverse metabolic pathways through the participation of phosphorylated intermediates (Poirier and Bucher, 2002). Adaptations to P starvation include changes to optimize uptake from the soil (morphological modification of roots, induction of phosphatases, RNAses, phosphate transporters) and changes to promote the efficient use of internal phosphate (Yuan and Liu, 2008). Limited supplies of phosphate reduce the internal adenylate pools (ATP, ADP, and AMP), which could affect the energy status (Poirier and Bucher, 2002). The animal homologue of the SnRKs, AMP-activated protein kinase, as its name suggests, is activated allosterically by AMP. While this is not the case for SnRKs or the fungal homologue, SNF1, 5'AMP does modulate the phosphorylation state of SnRK1 (Sugden et al., 1999a). It is therefore not surprizing that SnRKs could be involved in P-starvation responses. Analyses of the expression of SnRK genes under phosphate starvation show that most of the genes encoding SnRK1, SnRK2, and SnRK3 are up- or down-regulated and variations in gene expression are dependent upon starvation time and the affected organ (Fig. 2). In the case of genes AKIN10 and AKIN11 (SnRK1.1 and SnRK1.2), which encode the catalytic subunits of the SnRK1 complex in Arabidopsis, there is no regulation at the transcript level. However, the AKIN11/SnRK1.2 protein is specifically degraded under phosphate starvation, indicating that SnRK1 complexes under this nutritional condition contain AKIN10/SnRK1.1. Akin10 mutants growing under P-starvation show reduced starch mobilization at night and important differences in the expression of genes involved in carbohydrate metabolism and general stress responses (Fragoso et al., 2009). Role of SnRKs in biotic stress responses There is increasing evidence that metabolic signalling pathways in plants may also be activated by biotic stresses such as viral infection. SnRK1, like its animal and fungal counterparts, AMPK and SNF1, is regulated in part by phosphorylation of a threonine residue in its so-called T-loop (Sugden et al. 1999a). Two of the kinases responsible for this activation have recently been identified (Shen and Hanley-Bowdoin, 2006; Hey et al., 2007; Shen et al., 2009). These are known as SnRK1-activating kinase-1 and -2 (SnAK1 and SnAK2) (Hey et al., 2007) or geminivirus rep-interacting kinases (GRIK1 and GRIK2) (Shen and Hanley-Bowdoin, 2006). These kinases are expressed in response to geminivirus infection and interact with geminivirus replication protein AL1 (Shen and Hanley-Bowdoin, 2006), suggesting a metabolic signalling response to pathogen attack. As our understanding of the details of the pathways involved increases, manipulation of the interactions between metabolic and pathogen signalling might permit advanced priming of crops either to protect them from pathogen infection or from yield losses in the event of pathogen infection. Potential for manipulation of carbon metabolism to increase yield In addition to its potential for developing crops with greater tolerance to stress, SnRK1 regulation of starch accumulation in storage organs also gives us a potential handle on manipulating carbohydrate metabolism to increase the energy/nutritional value of crops and crop products. In this way, traditional crops, for which agricultural practices are already well established, could be adapted to provide better for the needs of the communities that rely on them. Sucrose synthesis and accumulation is co-ordinated with photosynthesis, the ultimate determinant of crop yield, and photosynthesis is inhibited if too much sucrose accumulates in the leaves (Stitt et al., 1988; Smith and Stitt, 2007). Hypothetically, if the flow of carbon into starch and other storage products could be enhanced, sink strength would increase, reducing the build-up of sucrose. Feedback inhibition of photosynthesis would decrease, increasing carbon fixation and maximizing yield. Biotechnologies could alternatively be used to improve productivity of particular products. For example, sucrose phosphate synthase (SPS) is an obvious target for manipulation to increase sucrose production in sugar cane. SPS is subject to inactivation by SnRK1, so simple over-expression of a wild-type SPS may not result in an increase in SPS activity in line with increased transcript and protein levels. This could be circumvented by uncoupling SPS from SnRK1 regulation by mutating the gene to remove the SnRK1 phosphorylation site. This technique has been used successfully to increase sterol production by uncoupling HMG-CoA reductase from regulation by SnRK1 (Hey et al., 2006). Maximizing yield is vital to provide food and fuel security to the rapidly increasing world population while reducing the contribution of agriculture to greenhouse gas emissions and climate change. Clearly, manipulation of the metabolic and stress signalling pathways mediated by the SnRK protein kinase families in plants has the potential to contribute to improving crop production and development, both in Africa and around the world. Patricia Coello was supported at Rothamsted Research through a sabbatical fellowship from Dirección General de Asuntos del Personal Académico, Universidad Nacional Autónoma de México (DGAPA-UNAM). Rothamsted Research receives grant-aided support from the Biotechnology and Biological Sciences Research Council (BBSRC) of the United Kingdom. References Baena-Gonzáles E,  Rolland F,  Thevelein JM,  Sheen J.  A central integrator of transcription networks in plant stress and energy signalling,  Nature ,  2007, vol.  448 (pg.  938- 942) Google Scholar CrossRef Search ADS PubMed  Batistic O,  Kudla J.  Plant calcineurin B-like proteins and their interacting protein kinases,  Biochimica et Biophysica Acta ,  2009, vol.  1793 (pg.  985- 992) Google Scholar CrossRef Search ADS PubMed  Bertorello AM,  Zhu JK.  SIK1/SOS2 networks: decoding sodium signals via calcium-responsive protein kinase pathways,  European Journal of Physiology ,  2009, vol.  458 (pg.  613- 619) Google Scholar CrossRef Search ADS PubMed  Boudsocq M,  Barbier-Brygoo H,  Laurière C.  Identification of nine sucrose non-fermenting 1-related protein kinases 2 activated by hyperosmotic and saline stresses in Arabidopsis thaliana,  Journal of Biological Chemistry ,  2004, vol.  279 (pg.  41758- 41766) Google Scholar CrossRef Search ADS PubMed  Boudsocq M,  Droillard MJ,  Barbier-Brygoo H,  Laurière C.  Different phosphorylation mechanisms are involved in activation of sucrose non-fermenting 1-related protein kinases 2 by osmotic stresses and abscisic acid,  Plant Molecular Biology ,  2007, vol.  63 (pg.  491- 503) Google Scholar CrossRef Search ADS PubMed  Boudsocq M,  Laurière C.  Osmotic signalling in plants: multiple pathways mediated by emerging kinase families,  Plant Physiology ,  2005, vol.  138 (pg.  1185- 1194) Google Scholar CrossRef Search ADS PubMed  Buitink J,  Thomas M,  Gissot L,  Leprince O.  Starvation, osmotic stress and dessication tolerance lead to expression of different genes of the regulatory β and γ subunits of the SnRK1 complex in germinating seeds of Medicago truncatula,  Plant, Cell and Environment ,  2004, vol.  27 (pg.  55- 67) Google Scholar CrossRef Search ADS   Burza AM,  Pekala I,  Sikora J,  Sledlecki P,  Malagocki P,  Bucholc M,  Koper L,  Zielenkiewicz P,  Dadlez M,  Dobrowolska G.  Nicotiana tabacum osmotic stress-activated kinase is regulated by phosphorylation on Ser-154 and Ser-158 in the kinase activation loop,  Journal of Biological Chemistry ,  2006, vol.  281 (pg.  34299- 34311) Google Scholar CrossRef Search ADS PubMed  Chinnusamy V,  Schumaker K,  Zhu JK.  Molecular genetic perspectives on cross-talk and specificity in abiotic stress signalling in plants,  Journal of Experimental Botany ,  2004, vol.  55 (pg.  225- 236) Google Scholar CrossRef Search ADS PubMed  Christensen JH,  Hewitson B,  Busuioc A, et al.  Solomon S,  Qin D,  Manning M,  Chen Z,  Marquis M,  Averyt KB,  Tignor M,  Miller HL.  Regional climate projections,  Climate change 2007: the physical science basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change ,  2007 Cambridge Cambridge University Press(pg.  847- 940) Curtis TY,  Muttucumaru N,  Shewry PR,  Parry MA,  Powers SJ,  Elmore JS,  Mottram DS,  Hook S,  Halford NG.  Evidence for genetic and environmental effects on free amino acid levels in wheat grain: implications for acrylamide formation during processing,  Journal of Agricultural and Food Chemistry ,  2009, vol.  57 (pg.  1013- 1021) Google Scholar CrossRef Search ADS PubMed  Davies JP,  Yildiz FH,  Grossman AR.  Sac3, an Snf1-like serine/threonine kinase that positively and negatively regulates the responses of Chlamydomonas to sulphur limitation,  The Plant Cell ,  1999, vol.  11 (pg.  1179- 1190) Google Scholar CrossRef Search ADS PubMed  Douglas P,  Pigaglio E,  Ferrer A,  Halford NG,  MacKintosh C.  Three spinach leaf nitrate reductase-3-hydroxy-3-methylglutaryl-CoA reductase kinases that are regulated by reversible phosphorylation and/or Ca2+ions,  Biochemical Journal ,  1997, vol.  325 (pg.  101- 109) Google Scholar CrossRef Search ADS PubMed  Elmore JS,  Mottram DS,  Muttucumaru N,  Dodson AT,  Parry MAJ,  Halford NG.  Changes in free amino acids and sugars in potatoes due to sulfate fertilization and the effect on acrylamide formation,  Journal of Agricultural and Food Chemistry ,  2007, vol.  55 (pg.  5363- 5366) Google Scholar CrossRef Search ADS PubMed  Fragoso S,  Espíndola L,  Páez-Valencia J,  Gamboa A,  Camacho Y,  Martinez-Barajas E,  Coello P.  SnRK1 isoforms AKIN10 and AKIN11 are differentially regulated in Arabidopsis plants under phosphate starvation,  Plant Physiology ,  2009, vol.  149 (pg.  1906- 1916) Google Scholar CrossRef Search ADS PubMed  Fugslang AT,  Guo Y,  Cuin TA, et al.  Arabidopsis protein kinase PKS5 inhibits the plasma membrane H+-ATPase by preventing interaction with 14-3-3 protein,  The Plant Cell ,  2007, vol.  19 (pg.  1617- 1634) Google Scholar CrossRef Search ADS PubMed  Fujii H,  Zhu J-K.  Arabidopsis mutant deficient in 3 abscisic acid-activated protein kinases reveals critical roles in growth, reproduction, and stress,  Proceedings of the National Academy of Sciences, USA ,  2009, vol.  1106 (pg.  8380- 8385) Google Scholar CrossRef Search ADS   Furihata T,  Maruyama K,  Fujita Y,  Umezawa T,  Yoshida R,  Shinozaki K,  Yamaguchi-Shinozaki K.  Abscisic acid-dependent multisite phosphorylation regulates the activity of a transcription activator AREB1,  Proceedings of the National Academy of Sciences, USA ,  2006, vol.  103 (pg.  1988- 1993) Google Scholar CrossRef Search ADS   Gojon A,  Nacry P,  Davidian JC.  Root uptake regulation: a central process for NPS homeostasis in plants,  Current Opinion in Plant Biology ,  2009, vol.  12 (pg.  328- 338) Google Scholar CrossRef Search ADS PubMed  Gomez-Cadenas A,  Verhey SD,  Holappa LD,  Shen Q,  Ho T-HD,  Walker-Simmons MK.  An abscisic acid-induced protein kinase, PKABA1, mediates abscisic acid-suppressed gene expression in barley aleurone layers,  Proceedings of the National Academy of Sciences, USA ,  1999, vol.  96 (pg.  1767- 1772) Google Scholar CrossRef Search ADS   Gonzalez-Ballester D,  Pollock SV,  Pootakham W,  Grossman AR.  The central role of a SnRK2 kinase in sulfur deprivation responses,  Plant Physiology ,  2008, vol.  147 (pg.  216- 227) Google Scholar CrossRef Search ADS PubMed  Guo Y,  Halfter U,  Ishitani M,  Zhu JK.  Molecular characterization of functional domains in the protein kinase SOS2 that is required for plant salt tolerance,  The Plant Cell ,  2001, vol.  13 (pg.  1383- 1399) Google Scholar CrossRef Search ADS PubMed  Guo Y,  Xiong L,  Song C-P,  Gong D,  Halfter U,  Zhu J-K.  A calcium sensor and its interacting protein kinase are global regulators of abscisic acid signalling in Arabidopsis,  Developmental Cell ,  2002, vol.  3 (pg.  233- 244) Google Scholar CrossRef Search ADS PubMed  Halford NG.  Regulation of carbon and amino acid metabolism, roles of sucrose non-fermenting-1-related protein kinase-1 and general control nonderepressible-2-related protein kinase,  Advances in Botanical Research including Advances in Plant Pathology ,  2006, vol.  43 (pg.  93- 142) Halford NG,  Hardie DG.  SNF1-related protein kinases: global regulators of carbon metabolism in plants?,  Plant Molecular Biology ,  1998, vol.  37 (pg.  735- 748) Google Scholar CrossRef Search ADS PubMed  Halford NG,  Hey SJ.  SNF1-related protein kinases (SnRKs) act within an intricate network that links metabolic and stress signalling in plants,  Biochemical Journal ,  2009, vol.  419 (pg.  247- 259) Google Scholar CrossRef Search ADS PubMed  Hey SJ,  Powers SJ,  Beale M,  Hawkins ND,  Ward J,  Halford NG.  Enhanced seed phytosterol accumulation through expression of a modified HMG-CoA reductase,  Plant Biotechnology Journal ,  2006, vol.  4 (pg.  219- 229) Google Scholar CrossRef Search ADS PubMed  Hey S,  Mayerhofer H,  Halford NG,  Dickinson JR.  DNA sequences from Arabidopsis which encode protein kinases and function as upstream regulators of Snf1 in yeast,  Journal of Biological Chemistry ,  2007, vol.  282 (pg.  10472- 10479) Google Scholar CrossRef Search ADS PubMed  Hey SJ,  Byrne E,  Halford NG.  The interface between metabolic and stress signalling,  Annals of Botany ,  2010, vol.  105 (pg.  197- 203) Google Scholar CrossRef Search ADS PubMed  Holappa LD,  Walker-Simmons MK.  The wheat abscisic acid-responsive protein kinase messenger RNA, PKABA1, is up-regulated by dehydration, cold temperature and osmotic stress,  Plant Physiology ,  1995, vol.  108 (pg.  1203- 1210) Google Scholar PubMed  Hu H-C,  Wang Y-Y,  Tsay Y-F.  AtCIPK8, a CBL-interacting protein kinase, regulates the low-affinity phase of the primary nitrate response,  The Plant Journal ,  2009, vol.  57 (pg.  264- 278) Google Scholar CrossRef Search ADS PubMed  Ikeda Y,  Koizumi N,  Kusano T,  Sano H.  Specific binding of a 14-3-3- protein to autophosphorylated WPK4, an SNF1-related wheat protein kinase, and to WPK4-phosphorylated nitrate reductase,  Journal of Biological Chemistry ,  2000, vol.  275 (pg.  31695- 31700) Google Scholar CrossRef Search ADS PubMed  Irihimovitch V,  Stern DB.  The sulphur acclimation SAC3 kinase is required for chloroplast transcriptional repression under sulphur limitation in Chlamydomonas reinhardtii,  Proceedings of the National Academy of Sciences, USA ,  2006, vol.  103 (pg.  7911- 7916) Google Scholar CrossRef Search ADS   Ishitani M,  Liu J,  Halfter U,  Kim CS,  Shi W,  Zhu JK.  SOS3 function in plant salt tolerance requires N-myristoylation and calcium binding,  The Plant Cell ,  2000, vol.  12 (pg.  1667- 1678) Google Scholar CrossRef Search ADS PubMed  Kaiser WM,  Huber SC.  Post-translational regulation of nitrate reductase in higher plants,  Plant Physiology ,  1994, vol.  106 (pg.  817- 821) Google Scholar PubMed  Kaiser WM,  Huber SC.  Post-translational regulation of nitrate reductase: mechanism, physiological relevance and environmental triggers,  Journal of Experimental Botany ,  2001, vol.  52 (pg.  1981- 1989) Google Scholar CrossRef Search ADS PubMed  Kim BG,  Waadt R,  Cheong Y-H,  Pandey GK,  Dominquez-Solis JR,  Schültke S,  Lee SC,  Kudla J,  Luan S.  The calcium sensor CBL10 mediates salt tolerance by regulating ion homeostasis in Arabidopsis,  The Plant Journal ,  2007, vol.  52 (pg.  473- 484) Google Scholar CrossRef Search ADS PubMed  Kimura T,  Shibagaki N,  Ohkama-Ohtsu N,  Hayashi H,  Yoneyama T,  Davies JP,  Fujiwara T.  Arabidopsis SnRK2.3 protein kinase is involved in the regulation of sulphur-responsive gene expression and O-acetyl-L-serine accumulation under limited sulphur supply,  Soil Science and Plant Nutrition ,  2006, vol.  52 (pg.  211- 220) Google Scholar CrossRef Search ADS   Kobayashi Y,  Murata M,  Minami H,  Yamamoto S,  Kagaya Y,  Hobo T,  Yamamoto A,  Hattori T.  Abscisic acid-activated SNRK2 protein kinases function in the gene-regulation pathway of ABA signal transduction by phosphorylating ABA response element-binding factors,  The Plant Journal ,  2005, vol.  44 (pg.  939- 949) Google Scholar CrossRef Search ADS PubMed  Kobayashi Y,  Yamamoto S,  Minami H,  Kagaya Y,  Hattori H.  Differential activation of the rice sucrose nonfermenting1-related protein kinase2 family by hyperosmotic stress and abscisic acid,  The Plant Cell ,  2004, vol.  16 (pg.  1163- 1177) Google Scholar CrossRef Search ADS PubMed  Leung J,  Merlot S,  Giraudat J.  The Arabidopsis ABSCISIC ACID-INSENSITIVE2 (ABI2) and ABI1 genes encode homologous protein phophatases 2C involved in abscisic acid signal transduction,  The Plant Cell ,  1997, vol.  9 (pg.  759- 771) Google Scholar CrossRef Search ADS PubMed  Li X-F,  Li Y-J,  An Y-H,  Xiong L-J,  Shao X-H,  Wang Y,  Sun Y.  AKINβ1 is involved in the regulation of nitrogen metabolism and sugar signalling in Arabidopsis,  Journal of Integrative Plant Biology ,  2009, vol.  51 (pg.  513- 520) Google Scholar CrossRef Search ADS PubMed  Lillo C.  Signalling cascades integrating light-enhanced nitrate metabolism,  Biochemical Journal ,  2008, vol.  415 (pg.  11- 19) Google Scholar CrossRef Search ADS PubMed  Lin H-X,  Yang Y-Q,  Quan RD,  Mendoza I,  Wu Y-S,  Du WN,  Zhao SS,  Shumaker KS,  Pardo JM,  Guo Y.  Phosphorylation of SOS3-LIKE CALCIUM BINDING PROTEIN8 by SOS2 protein kinase stabilizes their protein complex and regulates salt tolerance in Arabidopsis,  The Plant Cell ,  2009, vol.  21 (pg.  1607- 1619) Google Scholar CrossRef Search ADS PubMed  Liu J,  Ishitani M,  Halfter U,  Kim C-S,  Shu J-K.  The Arabidopsis thaliana SOS2 gene encodes a protein kinase that is required for salt tolerance,  Proceedings of the National Academy of Sciences, USA ,  2000, vol.  97 (pg.  3730- 3734) Google Scholar CrossRef Search ADS   Luan S.  The CBL–CIPK network in plant calcium signaling,  Trends in Plant Science ,  2009, vol.  14 (pg.  37- 42) Google Scholar CrossRef Search ADS PubMed  Ma Y,  Szostkiewicz I,  Korte A,  Moes D,  Yang Y,  Christmann A,  Grill E.  Regulators of PP2C phosphatase activity function as abscisic acid sensors,  Science ,  2009, vol.  324 (pg.  1064- 1068) Google Scholar PubMed  Mahajan S,  Pandey GK,  Tuteja N.  Calcium- and salt-stress signalling in plants: shedding light on SOS pathway,  Archives of Biochemistry and Biophysics ,  2008, vol.  471 (pg.  146- 158) Google Scholar CrossRef Search ADS PubMed  Mao X-G,  Zhang H-Y,  Tian S-J,  Chang X-P,  Jing R-L.  TaSnRK2.4, an SNF1-type serine/threonine protein kinase of wheat (Triticum aestivum L.), confers enhanced multistress tolerance in Arabidopsis,  Journal of Experimental Botany ,  2010, vol.  61 (pg.  683- 696) Google Scholar CrossRef Search ADS PubMed  Martinez-Atienza J,  Jiang X,  Garciadeblas B,  Mendoza I,  Zhu JK,  Pardo JM,  Quintero FJ.  Conservation of the salt overly sensitive pathway in rice,  Plant Physiology ,  2007, vol.  143 (pg.  1001- 1012) Google Scholar CrossRef Search ADS PubMed  Merlot S,  Mustilli A-C,  Genty B,  North H,  Lefebvre V,  Sotta B,  Vavasseur A,  Giraudat J.  Use of infrared thermal imaging to isolate Arabidopsis mutants defective in stomatal regulation,  The Plant Journal ,  2002, vol.  30 (pg.  601- 609) Google Scholar CrossRef Search ADS PubMed  Monks DE,  Aghoram K,  Courtney PD,  De Wald DB,  Dewey RE.  Hyperosmotic stress induces the rapid phosphorylation of a soybean phosphatidylinositol transfer protein homolog through activation of the protein kinases SPK1 and SPK2,  The Plant Cell ,  2001, vol.  13 (pg.  1205- 1219) Google Scholar CrossRef Search ADS PubMed  Moorhead G,  Douglas P,  Cotelle V, et al.  Phosphorylation-dependent interactions between enzymes of plant metabolism and 14-3-3 proteins,  The Plant Journal ,  1999, vol.  18 (pg.  1- 12) Google Scholar CrossRef Search ADS PubMed  Mustilli A-C,  Merlot S,  Vavasseur A,  Fenzi F,  Giraudat J.  Arabidopsis OST1 protein kinase mediates the regulation of stomatal aperture by abscisic acid and acts upstream of reactive oxygen species production,  The Plant Cell ,  2002, vol.  14 (pg.  3089- 3099) Google Scholar CrossRef Search ADS PubMed  Muttucumaru N,  Halford NG,  Elmore JS,  Dodson AT,  Parry M,  Shewry PR,  Mottram DS.  The formation of high levels of acrylamide during the processing of flour derived from sulfate-deprived wheat,  Journal of Agricultural and Food Chemistry ,  2006, vol.  54 (pg.  8951- 8955) Google Scholar CrossRef Search ADS PubMed  Nikiforova V,  Freitag J,  Kempa S,  Adamik M,  Hesse H,  Hoefgen R.  Transcriptome analysis of sulphur depletion in Arabidopsis thaliana: interlacing of biosynthetic pathways provides response specificity,  The Plant Journal ,  2003, vol.  33 (pg.  633- 650) Google Scholar CrossRef Search ADS PubMed  Ohta M,  Guo Y,  Halfter U,  Zhu JK.  A novel domain in the protein kinase SOS2 mediates interaction with the protein phosphatase 2C ABI2,  Proceedings of the National Academy of Sciences, USA ,  2003, vol.  100 (pg.  11771- 11776) Google Scholar CrossRef Search ADS   Park SY,  Fung P,  Nishimura N, et al.  Abscisic acid inhibits type 2C protein phosphatases via the PYR/PYL family of START proteins,  Science ,  2009, vol.  324 (pg.  1068- 1071) Google Scholar PubMed  Poirier Y,  Bucher M.  Somerville CR,  Meyerowitz EM.  Phosphate transport and homeostasis in Arabidopsis,  The Arabidopsis book ,  2002 Rockville, MD American Society of Plant Biologists(pg.  2- 35) Polge C,  Jossier M,  Crozet P,  Gissot L,  Thomas M.  β-subunits of the SnRK1 complexes share a common ancestral function together with expression and function specificities; physical interaction with nitrate reductase specifically occurs via AKINβ1-subunit,  Plant Physiology ,  2008, vol.  148 (pg.  1570- 1582) Google Scholar CrossRef Search ADS PubMed  Qiu QS,  Guo Y,  Quintero FJ,  Pardo JM,  Schumaker KS.  Regulation of vacuolar Na+/H+ exchange in Arabidopsis thaliana by the salt-overly-sensitive (SOS) pathway,  Journal of Biological Chemistry ,  2004, vol.  279 (pg.  207- 215) Google Scholar CrossRef Search ADS PubMed  Quan R,  Lin H,  Mendoza I,  Zhang Y,  Cao W,  Yang Y,  Shang M,  Chen S,  Pardo JM,  Guo Y.  SCABP8/CBL10, a putative calcium sensor, interacts with the protein kinase SOS2 to protect Arabidopsis shoots from salt stress,  The Plant Cell ,  2007, vol.  19 (pg.  1415- 1431) Google Scholar CrossRef Search ADS PubMed  Sánchez-Barrena MJ,  Fujii H,  Angulo I,  Martínez-Ripoll M,  Zhu JK,  Albert A.  The structure of the C-terminal domain of the protein kinase AtSOS2 bound to the calcium sensor AtSOS3,  Molecular Cell ,  2007, vol.  26 (pg.  427- 435) Google Scholar CrossRef Search ADS PubMed  Sanchez-Barrena MJ,  Martinez-Ripoll M,  Zhu JK,  Albert A.  The structure of the Arabidopsis thaliana SOS3: molecular mechanism of sensing calcium for salt stress responses,  Journal of Molecular Biology ,  2005, vol.  345 (pg.  1253- 1264) Google Scholar CrossRef Search ADS PubMed  Sano H,  Youssefian S.  Light and nutritional regulation of transcripts encoding a wheat protein kinase homolog is mediated by cytokinins,  Proceedings of the National Academy of Sciences, USA ,  1994, vol.  91 (pg.  2582- 2586) Google Scholar CrossRef Search ADS   Schachtman DP,  Shin R.  Nutrient sensing and signalling: NPKS,  Annual Review of Plant Biology ,  2007, vol.  58 (pg.  47- 69) Google Scholar CrossRef Search ADS PubMed  Semenov MA,  Halford NG.  Identifying target traits and molecular mechanisms for wheat breeding under a changing climate,  Journal of Experimental Botany ,  2009, vol.  60 (pg.  2791- 2804) Google Scholar CrossRef Search ADS PubMed  Sheard LB,  Zheng N.  Signal advance for abscisic acid,  Nature ,  2009, vol.  462 (pg.  575- 576) Google Scholar CrossRef Search ADS PubMed  Shen W,  Hanley-Bowdoin L.  Geminivirus infection up-regulates the expression of two Arabidopsis protein kinases related to yeast SNF1- and mammalian AMPK-activating kinases,  Plant Physiology ,  2006, vol.  142 (pg.  1642- 1655) Google Scholar CrossRef Search ADS PubMed  Shen W,  Reyes MI,  Hanley-Bowdoin L.  Arabidopsis protein kinases GRIK1 and GRIK2 specifically activate SnRK1 by phosphorylating its activation loop,  Plant Physiology ,  2009, vol.  150 (pg.  996- 1005) Google Scholar CrossRef Search ADS PubMed  Shi H,  Lee BH,  Wu SJ,  Zhu JK.  Overexpression of a plasma membrane Na+/H+ antiporter gene improves salt tolerance in Arabidopsis thaliana,  Nature Biotechnology ,  2003, vol.  21 (pg.  81- 85) Google Scholar CrossRef Search ADS PubMed  Shin R,  Alvarez S,  Burch AY,  Jez JM,  Schachtman DP.  Phosphoproteomic identification of targets of the Arabidopsis sucrose nonfermenting-like kinase SnRK2.8 reveals a connection to metabolic processes,  Proceedings of the National Academy of Sciences, USA ,  2007, vol.  104 (pg.  6460- 6465) Google Scholar CrossRef Search ADS   Shinozaki K,  Yamaguchi-Shinozaki K.  Gene networks involved in drought stress response and tolerance,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  221- 227) Google Scholar CrossRef Search ADS PubMed  Silva P,  Gerós H.  Regulation by salt of vacuolar H+-ATPase and H+-pyrophosphatase activities and Na+/H+ exchange,  Plant Signaling and Behavior ,  2009, vol.  8 (pg.  718- 726) Google Scholar CrossRef Search ADS   Smith AM,  Stitt M.  Coordination of carbon supply and plant growth,  Plant, Cell and Environment ,  2007, vol.  30 (pg.  1128- 1149) Google Scholar CrossRef Search ADS   Stitt M,  Wilke I,  Feil R,  Heldt HW.  Coarse control of sucrose-phosphate synthase in leaves: alterations of the kinetic properties in response to the rate of photosynthesis and the accumulation of sucrose,  Planta ,  1988, vol.  174 (pg.  217- 230) Google Scholar CrossRef Search ADS PubMed  Sugden C,  Crawford RM,  Halford NG,  Hardie DG.  Regulation of spinach SNF1-related (SnRK1) kinases by protein kinases and phosphatises is associated with phosphorylation of the T-loop and is regulated by 5'AMP,  The Plant Journal ,  1999, vol.  19 (pg.  433- 439) Google Scholar CrossRef Search ADS PubMed  Sugden C,  Donaghy PG,  Halford NG,  Hardie DG.  Two SNF1-related protein kinases from spinach leaf phosphorylate and inactivate 3-hydroxy-3-methylglutaryl-coenzyme A reductase, nitrate reductase, and sucrose phosphate synthase in vitro,  Plant Physiology ,  1999, vol.  20 (pg.  257- 274) Google Scholar CrossRef Search ADS   Umezawa T,  Sugiyama N,  Mizoguchi M,  Hayashi S,  Myouga F,  Yamaguchi-Shinozaki K,  Ishihama Y,  Hirayama T,  Shinozaki K.  Type 2C protein phosphatases directly regulate abscisic acid-activated protein kinases in Arabidopsis,  Proceedings of the National Academy of Sciences, USA ,  2009, vol.  106 (pg.  17588- 17593) Google Scholar CrossRef Search ADS   Umezawa T,  Yoshida R,  Maruyama K,  Yamaguchi-Shinozaki K,  Shinozaki K.  SRK2C, a SNF1-related protein kinase 2, improves drought tolerance by controlling stress-response gene expression in Arabidopsis thaliana,  Proceedings of the National Academy of Sciences, USA ,  2004, vol.  101 (pg.  17306- 17311) Google Scholar CrossRef Search ADS   Vincent O,  Townley R,  Kuchin S,  Carlson M.  Subcellular localization of the Snf1 kinase is regulated by specific beta subunits and a novel glucose signalling mechanism,  Genes and Development ,  2001, vol.  15 (pg.  1104- 1114) Google Scholar CrossRef Search ADS PubMed  Vlad F,  Rubio S,  Rodrigues A,  Sirichandra C,  Belin C,  Robert N,  Leung J,  Rodriguez PL,  Laurière C,  Merlot S.  Protein phosphatases 2C regulate the activation of the Snf1-related kinase OST1 by abscisic acid in Arabidopsis,  The Plant Cell ,  2009, vol.  21 (pg.  3170- 3184) Google Scholar CrossRef Search ADS PubMed  Wang M,  Gu D,  Liu T,  Wang Z,  Guo X,  Hou W,  Bai Y,  Chen X,  Wang G.  Overexpression of a putative maize calcineurin B-like protein in Arabidopsis confers salt tolerance,  Plant Molecular Biology ,  2007, vol.  65 (pg.  733- 746) Google Scholar CrossRef Search ADS PubMed  Wang R,  Guegler K,  LaBrie ST,  Crawford NM.  Genomic analysis of a nutrient response in Arabidopsis reveals diverse expression patterns and novel metabolic and potential regulatory genes induced by nitrate,  The Plant Cell ,  2000, vol.  12 (pg.  1491- 1509) Google Scholar CrossRef Search ADS PubMed  Wang R,  Okamoto M,  Xing X,  Crawford NM.  Microarray analysis of the nitrate response in Arabidopsis roots and shoots reveals over 1000 rapidly responding genes and new linkages to glucose, trehalose 6-phosphate, iron and sulphate metabolism,  Plant Physiology ,  2003, vol.  132 (pg.  556- 567) Google Scholar CrossRef Search ADS PubMed  Xu Z-S,  Liu L,  Ni Z-Y,  Liu P,  Chen M,  Li L-C,  Chen Y-F,  Ma Y- Z.  W55a encodes a novel protein kinase that is involved in multiple stress responses,  Journal of Integrative Plant Biology ,  2009, vol.  51 (pg.  58- 66) Google Scholar CrossRef Search ADS PubMed  Yang Q,  Chen ZZ,  Zhou XF,  Yin HB,  Li X,  Xin XF,  Hong XH,  Zhu JK,  Gong Z.  Overexpression of SOS (Salt Overly Sensitive) genes increases salt tolerance in transgenic Arabidopsis,  Molecular Plant ,  2009, vol.  2 (pg.  22- 31) Google Scholar CrossRef Search ADS PubMed  Yoshida R,  Hobo T,  Ichimura K,  Mizoguchi T,  Takahashi F,  Alonso J,  Ecker JR,  Shinozaki K.  ABA-activated SnRK2 protein kinase is required for dehydration stress signalling in Arabidopsis,  Plant and Cell Physiology ,  2002, vol.  43 (pg.  1473- 1483) Google Scholar CrossRef Search ADS PubMed  Yoshida R,  Umezawa T,  Mizoguchi T,  Takahashi S,  Takahashi F,  Shinozaki K.  The regulatory domain of SRK2E/OST1/SnRK2.6 interacts with ABI1 and integrates abscisic acid (ABA) and osmotic stress signals,  Journal of Biological Chemistry ,  2006, vol.  281 (pg.  5310- 5318) Google Scholar CrossRef Search ADS PubMed  Yuan H,  Liu D.  Signaling components involved in plant responses to phosphate starvation,  Journal of Integrative Plant Biology ,  2008, vol.  50 (pg.  849- 859) Google Scholar CrossRef Search ADS PubMed  Zhang Y,  Andralojc PJ,  Hey SJ,  Primavesi LF,  Specht M,  Koehler J,  Parry MAJ,  Halford NG.  Arabidopsis sucrose non-fermenting-1-related protein kinase-1 and calcium-dependent protein kinase phosphorylate conserved target sites in ABA response element binding proteins,  Annals of Applied Biology ,  2008, vol.  153 (pg.  401- 409) Google Scholar CrossRef Search ADS   Zhao J,  Sun Z,  Zhang J, et al.  Cloning and characterization of a novel CBL-interacting protein kinase from maize,  Plant Molecular Biology ,  2009, vol.  69 (pg.  661- 674) Google Scholar CrossRef Search ADS PubMed  Zimmermann P,  Hirsch-Hoffmann M,  Hennig L,  Gruissem W.  GENEVESTIGATOR. Arabidopsis microarray database and analysis toolbox,  Plant Physiology ,  2004, vol.  136 (pg.  2621- 2632) Google Scholar CrossRef Search ADS PubMed  © The Author [2010]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: [email protected]

Černý, Martin;Dyčka, Filip;Bobál'ová, Janette;Brzobohatý, Břetislav

doi: 10.1093/jxb/erq322pmid: 20974740

Abstract Cytokinins are plant hormones involved in regulation of diverse developmental and physiological processes in plants whose molecular mechanisms of action are being intensely researched. However, most rapid responses to cytokinin signals at the proteomic and phosphoproteomic levels are unknown. Early cytokinin responses were investigated through proteome-wide expression profiling based on image and mass spectrometric analysis of two-dimensionally separated proteins and phosphoproteins. The effects of 15 min treatments of 7-day-old Arabidopsis thaliana seedlings with four main cytokinins representing hydroxyisopentenyl, isopentenyl, aromatic, and urea-derived type cytokinins were compared to help elucidate their common and specific function(s) in regulating plant development. In proteome and phosphoproteome maps, significant differences were reproducibly observed for 53 and 31 protein spots, respectively. In these spots, 96 proteins were identified by matrix-assisted laser desorption/ionization time-of-flight/time-of-flight mass spectrometry (MALDI-TOF/TOF MS), providing a snapshot of early links in cytokinin-regulated signalling circuits and cellular processes, including light signalling and photosynthesis, nitrogen metabolism, the CLAVATA pathway, and protein and gene expression regulation, in accordance with previously described cytokinin functions. Furthermore, they indicate novel links between temperature and cytokinin signalling, and an involvement of calcium ions in cytokinin signalling. Most of the differentially regulated proteins and phosphoproteins are located in chloroplasts, suggesting an as yet uncharacterized direct signalling chain responsible for cytokinin action in chloroplasts. Finally, first insights into the degree of specificity of cytokinin receptors on phosphoproteomic effects were obtained from analyses of cytokinin action in a set of cytokinin receptor double mutants. Arabidopsis thaliana, cytokinin, phosphoproteome, proteome Introduction Cytokinins were first identified by their ability to promote division in cultured plant cells (Miller et al., 1955). They have since been shown to play roles in diverse aspects of plant growth and development including cell division, shoot initiation, apical meristem function, and vascular formation (Mok and Mok, 2001). Naturally occurring cytokinins are adenine derivatives substituted at the N6 position with an isoprenoid or aromatic side chain. Isoprenoid cytokinins are the most abundant cytokinins, while aromatic cytokinins, including N6-benzyladenine (BA), are minor components of the cytokinin pool (Strnad, 1997). Isoprenoid cytokinins are either of the isopentenyl (iP) type, with an isopentenyl N6 side chain, or of the zeatin (Z) type, with a hydroxylated isopentenyl N6 side chain in either trans (t-Z) or cis (c-Z) configuration. Reduction of the double bond in the side chain results in dihydrozeatin (DZ) (Brzobohatý et al., 1994; Mok and Mok, 2001). In addition, structurally unrelated (synthetic) phenylurea-type cytokinins, for example thidiazuron (TDZ), show high activity in most cytokinin bioassays (Mok and Mok, 2001). The diversity of compounds exerting cytokinin activity might underlie the molecular fine tuning of their numerous functions. Indeed, differences in biological activities of specific cytokinins have long been recognized, for instance in growth and morphogenic responses (e.g. Mok et al., 1978; Sujatha and Reddy, 1998; Lexa et al., 2003), but molecular mechanisms underlying between-cytokinin differences in activity are just emerging. Between-cytokinin differences in activity can be at least partially explained by differences in the receptors that perceive them and trigger biological responses. Cytokinin perception and signalling apparently evolved from bacterial two-component phosphorelays (Ferreira and Kieber, 2005). Binding of cytokinins to the Arabidopsis sensor hybrid histidine kinases AHK2, AHK3, and AHK4/CRE1/WOL1 initiates a phosphorelay in which Arabidopsis histidine-containing phosphotransfer proteins (AHPs) are phosphorylated then translocated into the nucleus, where they transfer the phosphate to Arabidopsis type-B response regulators (ARRs) (Kakimoto, 2003; Rashotte et al., 2003; Kiba et al., 2005; Choi and Hwang, 2007). The latter play roles in mediating transcriptional responses to cytokinin, including rapid induction of another class of response regulators, type-A ARRs (Rashotte et al., 2003), which act as negative regulators of the primary signal transduction pathway (Argueso et al., 2009). The first evidence for differential ligand specificity of cytokinin receptors has been obtained from their characterization in bacterial expression systems (Spíchal et al., 2004; Yonekura-Sakakibara et al., 2004; Romanov et al., 2006). Global genome expression profiling of cytokinin action in Arabidopsis has yielded a genome-wide view of changes in abundance of cytokinin-responsive transcripts that might be relevant for the many biological processes governed by cytokinins (Hoth et al., 2003; Rashotte et al., 2003; Brenner et al., 2005). However, since changes in transcript abundance are not necessarily linearly related to changes in levels and/or activities of corresponding proteins, proteome profiling can provide valuable complementary information regarding molecular mechanisms linking cytokinin signals and their diverse effects in plants. In addition to protein abundance, post-translational modifications (PTMs) of proteins are crucial determinants of protein activity and subcellular location. Phosphorylation is a key PTM; at least 5% of the Arabidopsis thaliana genome is involved in regulating protein phosphorylation (Laugesen et al., 2004), indicating that nearly all aspects of cell function may involve reversible phosphorylation. A set of proteins involved in cytokinin-induced photomorphogenesis has been identified by proteomic analysis (Lochmanová et al., 2008). In addition, rapid alterations of the phosphoproteome following cytokinin treatment have been examined in the moss Physcomitrella patens (Heintz et al., 2006), although comprehensive interpretation of the data was hindered by gaps in knowledge of its genome sequence. Nevertheless, our understanding of early cytokinin-responsive proteins and protein PTMs is still rudimentary. Hence, further analysis of proteome and phosphoproteome alterations caused by cytokinins before proteins encoded by the immediate cytokinin response genes (Brenner et al., 2006) accumulate significantly is needed to elucidate aspects of cytokinin signalling and action networks that cannot be deduced solely from transcriptome profiling. Therefore, proteomic analysis was applied to identify early cytokinin response proteins and phosphoproteins in Arabidopsis seedlings treated with four main cytokinins—t-zeatin (t-Z), isopentenyladenine (i-P), 6-benzylaminopurine (BA), and thidiazuron (TDZ). Detection of proteins involved in processes known to be regulated by cytokinins validated the experimental approach, and unexpected cytokinin targets were identified. Contributions of specific cytokinin receptors to the phosphoproteome alterations were assessed by examining effects of the cytokinins in ahk2ahk3, ahk2cre1, and ahk3cre1 mutants. Materials and methods Plant material, growth conditions, and cytokinin treatment Seeds of A. thaliana ecotype Columbia (Col-0), and ahk2ahk3, ahk2cre1, and ahk3cre1 double mutants (provided by Professor Thomas Schmülling, Free University of Berlin) were surface-sterilized and sown on Uhelon 120T (Silk & Progress, Czech Republic) mesh placed on 1% (w/v) agar containing Murashige and Skoog (MS) medium (pH 5.7) supplemented with 5×10−4% (v/v) dimethylsulphoxide (DMSO), stratified at 4 °C for 3 d, and cultivated at 21 °C/19 °C day/night temperatures, with a 16 h photoperiod (90 μmol m−2 s−1 light intensity) for 7 d. On the seventh day (after the first 2 h of the day period), the Uhelon mesh with the seedlings was transferred onto liquid MS medium supplemented with (i) 5×10−4% (v/v) DMSO (mock buffer); (ii) 5 μM individual cytokinins (BA, TDZ, iP, and t-Z; Duchefa) in DMSO (final concentration, as for the mock); (iii) 30 μM D600 and 60 μM LaCl3 (Sigma); or (iv) 30 μM D600, 60 μM LaCl3 (Sigma), and 5 μM t-Z, and incubated for 15 min. The concentrations of the calcium signalling inhibitors (D600 and LaCl3) followed Saunders and Hepler (1983) who observed disruption of cytokinin-induced bud formation in the moss Funaria in response to them. Seedlings were rapidly harvested, dried, then frozen and ground in liquid nitrogen. Protein extraction Total protein was extracted from frozen seedlings (250–300 mg) by acetone/trichloroacetic acid (TCA) extraction (Damerval et al., 1986). Dried protein was solubilized for 2 h at 30 °C in SOL buffer: 7 M urea, 2 M thiourea, 2% (w/v) CHAPS, 90 mM dithiothreitol (DTT). Insoluble matter was removed by centrifugation (15 000 g for 10 min) and the protein concentration was determined (Bradford, 1976) (Sigma-Aldrich, http://www.sigmaaldrich.com/) after diluting 1 μl of the total protein extract in 1 ml of reaction mix to prevent the SOL buffer interfering with the Bradford assay. Solubilized protein was then diluted 1:1 with rehydration solution [SOL supplemented with 1% (v/v) ampholytes pH 3–10, 0.2% (w/v) bromophenol blue] and loaded onto IPG strips (Bio-Rad, http://www.bio-rad.com/). For phosphoproteome analysis, an isolation procedure was established using a PhosphoProtein Purification Kit (Qiagen, http://www.qiagen.com/). Briefly, 350–400 mg of seedlings ground in liquid nitrogen were extracted with 4 ml of lysis buffer supplemented with protease inhibitors and benzonase (Qiagen kit). Each sample was then diluted to 25 ml with lysis buffer, applied to an affinity column and processed according to the supplier's manual (Qiagen). Protein concentration was determined by the Bradford assay. Desalted phosphoproteins in TRIS-HCl buffer (pH 7.0) were diluted with rehydration solution:SOL (1:1) and loaded onto IPG strips. 2D gel electrophoresis Proteins were separated essentially as previously described (Lochmanová et al., 2008). Briefly, portions containing 500 μg of protein or 150 μg of phosphoprotein were applied to 18 cm and 7 cm IPG strips, respectively, with a linear pH gradient (4–7), the strips were rehydrated for 16 h at room temperature in buffer containing the extracts, then the proteins were isoelectrically focused at 22 °C in six steps in a PROTEAN IEF Cell unit (Bio-Rad): 150 V (30 min), 300 V (60 min), 600 V (60 min), 1500 V (180 min), 3500 V (300 min), and 10 000 V to 80 000 Vh for long strips; 150 V (20 min), 300 V (20 min), 600 V (20 min), 1500 V (20 min), 3000 V (20 min), and 4000 V up to 12 000 Vh for short strips. The strips were then treated with buffers containing DTT and iodoacetamide (Sigma-Aldrich) to reduce and alkylate the proteins, which were then separated by 11% polyacrylamide SDS–PAGE with the following settings: 50 V (120 min) followed by 100 V (16 h) for large gels (proteome analysis), and 100 V (10 min) followed by 150 V (50 min) for small gels (phosphoproteome analysis), using a PROTEAN Plus Dodeca Cell, and a Mini-PROTEAN 3 Dodeca Cell (Bio-Rad), respectively. Protein staining and image analysis Gels were stained with colloidal Bio-Safe Coomassie G-250 (Bio-Rad) and scanned with a Bio-Rad GS-800 Calibrated Densitometer (400 dpi and 700 dpi for large and small gels, respectively). Acquired images were analysed using Decodon Delta 2D software (http://www.decodon.com). Three, six, three, and four biological replicates were used in the 2-DE total proteome comparisons of the wild type, phosphoproteome comparisons of the wild type, phosphoproteome comparisons of the ahk double mutant, and phosphoproteome comparisons of wild-type samples in the presence of calcium signalling inhibitors, respectively. Cytokinin responses of proteins corresponding to detected spots were deemed significant if there was a cytokinin/mock, BA/TDZ, BA/iP, or BA/t-Z spot volume ratio of ±1.4 or more (for at least one variant), with t-test values ≥95% and similar profiles in (i) ≥2 biological replicates for total protein comparisons (with three parallel SDS–PAGE analyses for each treatment, i.e. 15 parallel SDS–PAGE analyses for each biological replicate); (ii) ≥3 biological replicates for phosphoproteome comparisons (with two parallel SDS–PAGE analyses per treatment, i.e. 10 parallel SDS–PAGE analyses for each biological replicate); (iii) three biological replicates for phosphoproteome comparisons in the ahk double mutants (with two parallel SDS–PAGE analyses per treatment, i.e. 8 parallel SDS–PAGE analyses for each biological replicate); or (iv) ≥2 biological replicates for phosphoproteome comparisons in the wild type in the presence of calcium signalling inhibitors. Only spots with significant and reproducible changes were considered for mass spectroscopic identification. The experimental design is outlined schematically in Supplementary Fig. S1 available at JXB onluine. Protein identification Proteins were identified as previously described (Hradilová et al., 2010) with minor modifications. Briefly, selected protein spots were digested with trypsin. The dried tryptic peptides were each dissolved in 10 μl of 0.1% trifluoroacetic acid and purified using ZipTip C18 tips. The eluate was mixed with 1 vol. of 10 mg ml−1 α-cyano-4-hydroxycinnamic acid (CHCA) in 50% (v/v) acetonitrile and 0.1% trifluoroacetic acid for spotting onto sample plates, and dried for matrix-assisted laser desorption/ionization time-of-flight/time-of-flight mass spectrometry (MALDI-TOF/TOF MS) analysis. To demonstrate phosphorylation of selected peptides, phosphopeptides were first enriched from tryptic peptides dissolved in 10% acetonitrile and 0.1% acetic acid using IMAC tips (Millipore) containing iron ions. After loading, the tips were washed with 10% acetonitrile and 0.1% acetic acid, then rinsed with water. Phosphopeptides were eluted by 0.3 N ammonium hydroxide and measured using 15 mg ml−1 2,5-dihydroxybenzoic acid in 50% (v/v) acetonitrile and 6% phosphoric acid solution as a matrix. Alkaline phosphatase treatment was used to confirm the phosphorylation of the phosphopeptides according to Larsen et al. (2001). Briefly, IMAC-purified phosphopeptides were incubated with 0.05 U μl−1 alkaline phosphatase in 50 mM NH4HCO3, pH 7.8 at 37 °C for 30 min then acidified with 2.5 μl of 5% trifluoroacetic acid. Phosphopeptides were identified by single or multiple 80 Da (HPO3) losses in MALDI-TOF-MS following alkaline phosphatase treatment, for mono- and multiphosphorylated peptides, respectively. MALDI-TOF/TOF measurements were performed with an Applied Biosystems 4700 Proteomic Analyzer (Applied Biosystems, http://www.appliedbiosystems.com/) equipped with an Nd:YAG laser (355 nm) operated with 3–7 ns pulses and 200 Hz firing rate in positive reflectron mode for both MS and MS/MS analyses. The accelerating voltage in the ion source for MS and MS/MS analyses was set at 20 kV and 8 kV, respectively. Acquired sequences were searched against the NCBInr sequence database (version 09/2009) using Mascot (http://www.matrixscience.com/), and peaks generated from the acquired mass spectra by the Peak-to-Mascot function incorporated in the software. In the MS analyses, peaks in the 900–4000 m/z range with signal to noise (S/N) ratios >4 were sought. In the MS/MS analyses, peaks with S/N ratios >4 in the m/z range from 68 m/z up to 20 m/z units lower than their precursors’ m/z values were used. The resulting peak lists contained information from both MS and MS/MS runs concerning fragmentation patterns of selected precursors. Parameters for both MS and MS/MS data searches in Mascot were: taxonomy, Arabidopsis thaliana; enzyme, trypsin; allowed missed cleavages, 1 [except for the peptide VGKDSKDKELKEAFK of endoplasmin homologue (SHD), where allowed missed cleavages were set to 4]; fixed modification, carbamidomethyl (C); variable modifications, none or Phospho (ST) and Phospho (Y) (for searching phosphopeptides); peptide tolerance, 0.5 Da; MS/MS tolerance, 0.5 Da; peptide charge, (+1); instrument, MALDI-TOF/TOF. Protein matches in MS/MS identification were considered valid if there was at least one peptide with a Mascot score corresponding to identity or extensive homology (P <0.05). Protein scores were derived from ion scores as a non-probabilistic basis for ranking protein hits by Mascot. Similar parameters were set for peptide mass fingerprint analysis—only protein matches with Mascot scores indicating extensive homology were accepted. Gene ontology Gene ontology was evaluated by BiNGO 2.3 in Cytoscape 2.6.2, with data from the NCBI (http://www.ncbi.nlm.nih.gov) and TAIR (http://www.arabidopsis.org) databases. Results Identification of early cytokinin response proteins To identify early cytokinin response proteins, 7-day-old Arabidopsis seedlings were treated (separately) with four main cytokinins (BA, iP, TDZ, and t-Z) at a concentration of 5 μM for 15 min. Total proteins were then extracted and subjected to 2-DE (Fig. 1A, B). Image analysis of the resulting proteome maps revealed >850 reproducibly resolved spots in gels over pI and molecular mass ranges of 4–7 and 10–120 kDa, respectively, then proteome patterns of seedlings treated with the individual cytokinins were compared separately with the proteome patterns of seedlings treated with mock buffer. Significant differences (P <0.05) in all biological replicates were found for 160 resolved spots, but only 53 spots were reproducibly significant in two or more independent experiments and were then subjected to protein identification. Altogether, 67 proteins were identified in the 53 spots, including 10 protein mixtures and a non-dissociated heterodimer consisting of small and large Rubisco subunits (T13), by MALDI-TOF/TOF MS analysis followed by Mascot database searches of the full NCBI Arabidopsis protein database (Table 1; Supplementary Table S1 at JXB online). The ratio of numbers of up-regulated to down-regulated proteins was ∼1:2. Identified protein spots are marked in protein maps shown in Fig. 1A, and corresponding partial amino acid sequences are listed in Supplementary Table S3. Protein identifications and relative fold changes based on mean percentage volumes of each of these spots are presented in Table 1. The apparent strength of effects of the cytokinins on expression of the early cytokinin response proteins decreased in the order BA=TDZ>t-Z=iP. Table 1. Early cytokinin response proteins of Arabidopsis Spot/protein no.  AGI code  Protein name  MALDI-MS (protein score/%cov/pep#)  Relative fold change           BA  iP  TDZ  t-Z  T1  At2g28000  Rubisco large subunit-binding protein subunit α, chloroplastic  666/32/12  –1.5±0.23  –1.5±0.23  –1.4±0.21  –1.4±0.21  T2  At4g24190  Endoplasmin homologue (SHD)  27/1/1  1.5±0.10  1.8±0.20  1.6±0.32  1.4±0.28  T3  At5g56000  Heat shock protein 81-4  153/8/5  –2.0±0.20  –1.5±0.23  –2.0±0.30  –1.3±0.20  T4  At1g19570  Glutathione S-transferase DHAR1, mitochondrial  35/17/2  –1.4±0.21  –1.3±0.20  –1.3±0.20  –1.5±0.23  T5  At5g17920  Cobalamin-independent methionine synthase  49/6/3 PMF: 127/23/11  1.4±0.14  1.4±0.26  1.6±0.33  1.7±0.30  T6  At5g17920  Cobalamin-independent methionine synthase  55/7/3 PMF: 104/21/9  1.3±0.26  1.4±0.28  1.5±0.30  1.4±0.21  T7  At3g60750  Putative transketolase  236/17/4  –1.5±0.10  –1.5±0.30  –1.6±0.30  –1.4±0.20  T8  At5g02500  Heat shock cognate 70 kDa protein 1  104/10/3  –1.3±0.20  –1.4±0.21  –1.5±0.23  –1.3±0.20  T9  At2g30950  Cell division protease FtsH homologue 2, chloroplastic  98/8/3  1.5±0.30  1.4±0.28  1.3±0.26  1.4±0.21  T10  At2g30950  Cell division protease FtsH homologue 2, chloroplastic  101/9/3  1.4±0.10  1.7±0.12  1.5±0.30  1.8±0.10  T11  At5g60640  Protein disulphide isomerase-like protein  96/25/8  –2.5±0.38  –2.0±0.30  –1.6±0.24  –2.0±0.30  T12  AtCg00120  ATP synthase subunit α, chloroplastic  114/26/9  1.3±0.26  1.4±0.28  1.4±0.21  1.4±0.21  T13  At5g38420  Rubisco small chain 2β, chloroplastic  45/8/1  –1.4±0.21  –1.3±0.20  –1.3±0.20  –1.4±0.21    AtCg00490  Rubisco large chain  PMF: 159/34/14          T14  At1g21750  Probable protein disulphide-isomerase 1  41/6/2  –1.7±0.23  –1.4±0.21  –1.6±0.30  –1.4±0.15  T15  At1g20020  Ferredoxin-NADP reductase, leaf 2, chloroplastic  383/29/7  1.6±0.32  1.3±0.30  1.0±0.30  1.0±0.32  T16  AtCg00120  ATP synthase subunit α, chloroplastic  238/20/6  1.5±0.30  1.2±0.24  1.2±0.22  1.4±0.28  T17  At5g08690  ATP synthase subunit β-2, mitochondrial  84/12/4  –1.3±0.20  –1.2±0.20  –1.2±0.18  –1.5±0.23  T18  AtCg00490  Rubisco large chain  269/22/7  2.0±0.40  1.5±0.30  2.0±0.40  1.5±0.30  T19  AtCg00480  ATP synthase subunit β, chloroplastic  128/10/4  –1.3±0.24  –1.4±0.30  –1.5±0.23  –1.4±0.21  T20  At2g39730  Rubisco activase, chloroplastic  714/39/10  1.4±0.28  1.5±0.34  1.5±0.30  1.4±0.25  T21  At2g39730  Rubisco activase, chloroplastic  465/17/5  1.4±0.24  1.5±0.30  1.3±0.26  1.6±0.25  T22  At3g54050  Fructose-1,6-bisphosphatase, chloroplastic  100/12/4  –1.3±0.15  –1.2±0.18  –1.5±0.08  –1.5±0.10  T23  At5g35630  Glutamine synthetase, chloroplastic/mitochondrial  PMF: 72/29/5  –1.3±0.20  –1.2±0.18  –1.4±0.21  –1.2±0.18  T24  At4g02520  Glutathione S-transferase PM24  230/45/8  1.7±0.20  1.9±0.10  1.5±0.30  1.3±0.26    At5g61410  Ribulose-5-phosphate-3-epimerase  302/19/3          T25  At1g09780  2,3-Bisphosphoglycerate-independent phosphoglycerate mutase 1  435/25/9  1.4±0.28  –1.2±0.13  –1.4±0.20  –1.3±0.20  T26  At5g26000  Myrosinase  PMF: 61/13/5  –1.5±0.24  –1.6±0.24  –1.5±0.23  –1.3±0.12  T27  At3g18780  Actin-2  147/23/5  –1.6±0.24  –1.5±0.14  –1.5±0.23  –1.4±0.22    At1g49240  Actin-8  147/23/5            At5g35630  Glutamine synthetase, chloroplastic/mitochondrial  133/15/3          T28  At4g20360  Elongation factor Tu, chloroplastic  66/7/2  –1.5±0.23  –1.3±0.20  –1.8±0.26  –1.5±0.10  T29  At2g39730  Rubisco activase, chloroplastic  396/22/5  1.4±0.20  1.4±0.15  1.4±0.30  1.4±0.21  T30  At1g32060  Phosphoribulokinase, chloroplastic  287/26/7  –1.6±0.15  –1.4±0.22  –1.5±0.10  –1.5±0.20  T31  At3g12780  Phosphoglycerate kinase  254/17/5  –1.5±0.20  –1.5±0.10  –1.5±0.10  –1.6±0.20  T32  At3g52930  Fructose-bisphosphate aldolase  318/23/6  1.6±0.32  1.7±0.21  2.1±0.42  2.2±0.30  T33  At2g43910  Thiol methyltransferase, putative  64/13/2  –1.8±0.27  –1.6±0.24  –1.1±0.22  –1.2±0.30  T34  At3g09200  60S Acidic ribosomal protein P0-2  59/14/2  –1.4±0.21  –1.7±0.22  –2.1±0.31  –2.0±0.30  T35  At1g30230  Elongation factor 1- δ 1  897/46/8  –1.5±0.20  –1.4±0.23  –1.5±0.15  –1.4±0.21  T36  At2g05990  Enoyl-[acyl-carrier-protein] reductase  206/23/6  1.7±0.15  1.7±0.34  1.6±0.32  1.7±0.02  T37  At3g10920  Superoxide dismutase [Mn], mitochondrial  462/38/7  –1.2±0.19  –1.4±0.08  –1.4±0.21  –1.3±0.20    At2g47730  Glutathione S-transferase 6, chloroplastic  147/35/4          T38  At3g16420  PBP1  35/6/1  –1.4±0.21  –1.2±0.18  –1.4±0.21  –1.3±0.20  T39  At3g53460  29 kDa ribonucleoprotein, chloroplastic  71/4/1  –1.4±0.22  –1.4±0.21  –1.6±0.24  –1.4±0.21  T40  At4g28520  Cruciferin 3  475/41/10  –2.4±0.36  2.7±0.54  1.1±0.30  –1.3±0.20  T41  At5g38480  14-3-3-like protein GF14 ψ  403/39/6  –1.5±0.23  –1.3±0.20  –1.6±0.25  –1.2±0.18    At1g22300  14-3-3-like protein GF14 ϵ  326/26/5          T42  At5g14740  β-Carbonic anhydrase 2  333/35/5  1.6±0.20  1.6±0.10  1.7±0.34  1.9±0.38  T43  At2g37220  Putative ribonucleoprotein, chloroplastic  440/24/6  –1.4±0.21  –1.2±0.11  –1.1±0.17  –1.2±0.18    At5g50250  Putative 31 kDa ribonucleoprotein, chloroplastic  85/7/2          T44  At5g10450  14-3-3-like protein GF14 λ  39/9/1  1.9±0.35  1.5±0.30  1.4±0.24  1.4±0.28  T45  At2g34430  Photosystem II type I chlorophyll a/b-binding protein (LHB1B1)  229/31/3  –1.6±0.24  –1.7±0.26  –1.1±0.20  –1.5±0.23    At2g34420  Photosystem II type I chlorophyll a/b-binding protein (LHB1B2)  229/31/3          T46  At2g21330  Fructose-bisphosphate aldolase (FBA1)  105/12/3  –1.2±0.23  –1.3±0.20  –1.5±0.23  –1.6±0.24    At4g38970  Fructose-bisphosphate aldolase (FBA2)  76/19/4          T47  At1g54870  Glucose and ribitol dehydrogenase homologue 1  PMF: 61/31/5  –1.5±0.23  –1.5±0.23  –1.5±0.23  –1.4±0.21  T48  At1g29910, At1g29920  Chlorophyll a-b-binding protein 165/180, chloroplastic (CAB2/3)  419/35/4  –1.5±0.23  –1.5±0.30  –1.4±0.21  –1.3±0.20    At1g29910, At1g29930  Chlorophyll a-b-binding protein 2, chloroplastic (CAB1)            T49  At3g55440  Triosephosphate isomerase, cytosolic  122/23/3  –1.4±0.22  –1.5±0.23  –2±0.30  –1.4±0.15  T50  At3g14290  Proteasome subunit α type-5-B  121/27/4  –1.5±0.23  –1.7±0.30  –1.2±0.20  –1.5±0.23    At1g53850  Proteasome subunit α type-5-A  88/21/3          T51  At3g27830  50S Ribosomal protein L12-1, chloroplastic  313/36/3  –1.6±0.18  –1.1±0.17  –1.3±0.20  –1.3±0.20    At3g27850  50S Ribosomal protein L12-3, chloroplastic            T52  At4g38680  Glycine-rich protein 2/cold shock domain protein 2  74/21/2  –1.6±0.10  –1.5±0.23  –1.6±0.24  –1.3±0.20  T53  At1g61520  LHCA3 (PSI type III chlorophyll a/b-binding protein); chlorophyll binding  207/13/3  –1.3±0.20  –1.4±0.20  –1.3±0.32  –1.7±0.26  Spot/protein no.  AGI code  Protein name  MALDI-MS (protein score/%cov/pep#)  Relative fold change           BA  iP  TDZ  t-Z  T1  At2g28000  Rubisco large subunit-binding protein subunit α, chloroplastic  666/32/12  –1.5±0.23  –1.5±0.23  –1.4±0.21  –1.4±0.21  T2  At4g24190  Endoplasmin homologue (SHD)  27/1/1  1.5±0.10  1.8±0.20  1.6±0.32  1.4±0.28  T3  At5g56000  Heat shock protein 81-4  153/8/5  –2.0±0.20  –1.5±0.23  –2.0±0.30  –1.3±0.20  T4  At1g19570  Glutathione S-transferase DHAR1, mitochondrial  35/17/2  –1.4±0.21  –1.3±0.20  –1.3±0.20  –1.5±0.23  T5  At5g17920  Cobalamin-independent methionine synthase  49/6/3 PMF: 127/23/11  1.4±0.14  1.4±0.26  1.6±0.33  1.7±0.30  T6  At5g17920  Cobalamin-independent methionine synthase  55/7/3 PMF: 104/21/9  1.3±0.26  1.4±0.28  1.5±0.30  1.4±0.21  T7  At3g60750  Putative transketolase  236/17/4  –1.5±0.10  –1.5±0.30  –1.6±0.30  –1.4±0.20  T8  At5g02500  Heat shock cognate 70 kDa protein 1  104/10/3  –1.3±0.20  –1.4±0.21  –1.5±0.23  –1.3±0.20  T9  At2g30950  Cell division protease FtsH homologue 2, chloroplastic  98/8/3  1.5±0.30  1.4±0.28  1.3±0.26  1.4±0.21  T10  At2g30950  Cell division protease FtsH homologue 2, chloroplastic  101/9/3  1.4±0.10  1.7±0.12  1.5±0.30  1.8±0.10  T11  At5g60640  Protein disulphide isomerase-like protein  96/25/8  –2.5±0.38  –2.0±0.30  –1.6±0.24  –2.0±0.30  T12  AtCg00120  ATP synthase subunit α, chloroplastic  114/26/9  1.3±0.26  1.4±0.28  1.4±0.21  1.4±0.21  T13  At5g38420  Rubisco small chain 2β, chloroplastic  45/8/1  –1.4±0.21  –1.3±0.20  –1.3±0.20  –1.4±0.21    AtCg00490  Rubisco large chain  PMF: 159/34/14          T14  At1g21750  Probable protein disulphide-isomerase 1  41/6/2  –1.7±0.23  –1.4±0.21  –1.6±0.30  –1.4±0.15  T15  At1g20020  Ferredoxin-NADP reductase, leaf 2, chloroplastic  383/29/7  1.6±0.32  1.3±0.30  1.0±0.30  1.0±0.32  T16  AtCg00120  ATP synthase subunit α, chloroplastic  238/20/6  1.5±0.30  1.2±0.24  1.2±0.22  1.4±0.28  T17  At5g08690  ATP synthase subunit β-2, mitochondrial  84/12/4  –1.3±0.20  –1.2±0.20  –1.2±0.18  –1.5±0.23  T18  AtCg00490  Rubisco large chain  269/22/7  2.0±0.40  1.5±0.30  2.0±0.40  1.5±0.30  T19  AtCg00480  ATP synthase subunit β, chloroplastic  128/10/4  –1.3±0.24  –1.4±0.30  –1.5±0.23  –1.4±0.21  T20  At2g39730  Rubisco activase, chloroplastic  714/39/10  1.4±0.28  1.5±0.34  1.5±0.30  1.4±0.25  T21  At2g39730  Rubisco activase, chloroplastic  465/17/5  1.4±0.24  1.5±0.30  1.3±0.26  1.6±0.25  T22  At3g54050  Fructose-1,6-bisphosphatase, chloroplastic  100/12/4  –1.3±0.15  –1.2±0.18  –1.5±0.08  –1.5±0.10  T23  At5g35630  Glutamine synthetase, chloroplastic/mitochondrial  PMF: 72/29/5  –1.3±0.20  –1.2±0.18  –1.4±0.21  –1.2±0.18  T24  At4g02520  Glutathione S-transferase PM24  230/45/8  1.7±0.20  1.9±0.10  1.5±0.30  1.3±0.26    At5g61410  Ribulose-5-phosphate-3-epimerase  302/19/3          T25  At1g09780  2,3-Bisphosphoglycerate-independent phosphoglycerate mutase 1  435/25/9  1.4±0.28  –1.2±0.13  –1.4±0.20  –1.3±0.20  T26  At5g26000  Myrosinase  PMF: 61/13/5  –1.5±0.24  –1.6±0.24  –1.5±0.23  –1.3±0.12  T27  At3g18780  Actin-2  147/23/5  –1.6±0.24  –1.5±0.14  –1.5±0.23  –1.4±0.22    At1g49240  Actin-8  147/23/5            At5g35630  Glutamine synthetase, chloroplastic/mitochondrial  133/15/3          T28  At4g20360  Elongation factor Tu, chloroplastic  66/7/2  –1.5±0.23  –1.3±0.20  –1.8±0.26  –1.5±0.10  T29  At2g39730  Rubisco activase, chloroplastic  396/22/5  1.4±0.20  1.4±0.15  1.4±0.30  1.4±0.21  T30  At1g32060  Phosphoribulokinase, chloroplastic  287/26/7  –1.6±0.15  –1.4±0.22  –1.5±0.10  –1.5±0.20  T31  At3g12780  Phosphoglycerate kinase  254/17/5  –1.5±0.20  –1.5±0.10  –1.5±0.10  –1.6±0.20  T32  At3g52930  Fructose-bisphosphate aldolase  318/23/6  1.6±0.32  1.7±0.21  2.1±0.42  2.2±0.30  T33  At2g43910  Thiol methyltransferase, putative  64/13/2  –1.8±0.27  –1.6±0.24  –1.1±0.22  –1.2±0.30  T34  At3g09200  60S Acidic ribosomal protein P0-2  59/14/2  –1.4±0.21  –1.7±0.22  –2.1±0.31  –2.0±0.30  T35  At1g30230  Elongation factor 1- δ 1  897/46/8  –1.5±0.20  –1.4±0.23  –1.5±0.15  –1.4±0.21  T36  At2g05990  Enoyl-[acyl-carrier-protein] reductase  206/23/6  1.7±0.15  1.7±0.34  1.6±0.32  1.7±0.02  T37  At3g10920  Superoxide dismutase [Mn], mitochondrial  462/38/7  –1.2±0.19  –1.4±0.08  –1.4±0.21  –1.3±0.20    At2g47730  Glutathione S-transferase 6, chloroplastic  147/35/4          T38  At3g16420  PBP1  35/6/1  –1.4±0.21  –1.2±0.18  –1.4±0.21  –1.3±0.20  T39  At3g53460  29 kDa ribonucleoprotein, chloroplastic  71/4/1  –1.4±0.22  –1.4±0.21  –1.6±0.24  –1.4±0.21  T40  At4g28520  Cruciferin 3  475/41/10  –2.4±0.36  2.7±0.54  1.1±0.30  –1.3±0.20  T41  At5g38480  14-3-3-like protein GF14 ψ  403/39/6  –1.5±0.23  –1.3±0.20  –1.6±0.25  –1.2±0.18    At1g22300  14-3-3-like protein GF14 ϵ  326/26/5          T42  At5g14740  β-Carbonic anhydrase 2  333/35/5  1.6±0.20  1.6±0.10  1.7±0.34  1.9±0.38  T43  At2g37220  Putative ribonucleoprotein, chloroplastic  440/24/6  –1.4±0.21  –1.2±0.11  –1.1±0.17  –1.2±0.18    At5g50250  Putative 31 kDa ribonucleoprotein, chloroplastic  85/7/2          T44  At5g10450  14-3-3-like protein GF14 λ  39/9/1  1.9±0.35  1.5±0.30  1.4±0.24  1.4±0.28  T45  At2g34430  Photosystem II type I chlorophyll a/b-binding protein (LHB1B1)  229/31/3  –1.6±0.24  –1.7±0.26  –1.1±0.20  –1.5±0.23    At2g34420  Photosystem II type I chlorophyll a/b-binding protein (LHB1B2)  229/31/3          T46  At2g21330  Fructose-bisphosphate aldolase (FBA1)  105/12/3  –1.2±0.23  –1.3±0.20  –1.5±0.23  –1.6±0.24    At4g38970  Fructose-bisphosphate aldolase (FBA2)  76/19/4          T47  At1g54870  Glucose and ribitol dehydrogenase homologue 1  PMF: 61/31/5  –1.5±0.23  –1.5±0.23  –1.5±0.23  –1.4±0.21  T48  At1g29910, At1g29920  Chlorophyll a-b-binding protein 165/180, chloroplastic (CAB2/3)  419/35/4  –1.5±0.23  –1.5±0.30  –1.4±0.21  –1.3±0.20    At1g29910, At1g29930  Chlorophyll a-b-binding protein 2, chloroplastic (CAB1)            T49  At3g55440  Triosephosphate isomerase, cytosolic  122/23/3  –1.4±0.22  –1.5±0.23  –2±0.30  –1.4±0.15  T50  At3g14290  Proteasome subunit α type-5-B  121/27/4  –1.5±0.23  –1.7±0.30  –1.2±0.20  –1.5±0.23    At1g53850  Proteasome subunit α type-5-A  88/21/3          T51  At3g27830  50S Ribosomal protein L12-1, chloroplastic  313/36/3  –1.6±0.18  –1.1±0.17  –1.3±0.20  –1.3±0.20    At3g27850  50S Ribosomal protein L12-3, chloroplastic            T52  At4g38680  Glycine-rich protein 2/cold shock domain protein 2  74/21/2  –1.6±0.10  –1.5±0.23  –1.6±0.24  –1.3±0.20  T53  At1g61520  LHCA3 (PSI type III chlorophyll a/b-binding protein); chlorophyll binding  207/13/3  –1.3±0.20  –1.4±0.20  –1.3±0.32  –1.7±0.26  Spot no., spot number (as given in Fig. 1A); AGI code, accession number in the TAIR database; Protein name, entry name in the NCBI database; %cov, percentage of protein coverage; pep#, number of peptides; PMF, peptide mass fingerprint; Relative fold change, fold change relative to the mock control (calculated by DECODON DELTA 2D software) ±SE. Full information on the proteins including their classification, peptide sequences and peak list is given in Supplementary Tables S1 and Supplementary Data at JXB online. View Large Fig. 1. View large Download slide Effects of cytokinin treatment on the proteome and phosphoproteome of Arabidopsis seedlings. (A) Average two-dimensional gel electrophoresis proteome map of 7-day-old Arabidopsis seedlings treated with cytokinin/mock buffer for 15 min. Differentially regulated protein spots are indicated. See Table 1, and Supplementary Table S1 at JXB online, for detailed information on the corresponding identified proteins. Proteins (500 μg) were separated in the first and second dimensions by IPG (18 cm strips, pH 4–7) followed by 11% SDS–PAGE then visualized by Bio-Safe Coomassie G250 staining. Isoelectric points (pI) and migrating positions of molecular mass (kDa) markers are marked. (B) Examples of spots corresponding to the differentially regulated proteins in Arabidopsis seedlings treated with 5 μM cytokinin (BA, iP, TDZ, or t-Z) or mock buffer for 15 min. For details see Materials and methods. (C) Average 2-DE phosphoproteome map of 7-day-old Arabidopsis seedlings treated with cytokinin/mock buffer for 15 min. Differentially regulated protein spots are indicated. See Table 2, and Supplementary Table S2, for detailed information on the corresponding identified proteins. Phosphoprotein fractions were obtained using a PhosphoProtein Purification Kit. Phosphoproteins (150 μg) were separated in the first and second dimensions by IPG (7 cm strips, pH 4–7) followed by 11% SDS–PAGE then visualized by Bio-Safe Coomassie G250 staining. Isoelectric points (pI) and relative migrating positions of molecular mass (kDa) markers are marked. (D) Examples of spots corresponding to the differentially regulated phosphoproteins in Arabidopsis seedlings treated with 5 μM cytokinin (BA, iP, TDZ, or t-Z) or mock buffer for 15 min. For details see Materials and methods. (This figure is available in colour at JXB online.) Fig. 1. View large Download slide Effects of cytokinin treatment on the proteome and phosphoproteome of Arabidopsis seedlings. (A) Average two-dimensional gel electrophoresis proteome map of 7-day-old Arabidopsis seedlings treated with cytokinin/mock buffer for 15 min. Differentially regulated protein spots are indicated. See Table 1, and Supplementary Table S1 at JXB online, for detailed information on the corresponding identified proteins. Proteins (500 μg) were separated in the first and second dimensions by IPG (18 cm strips, pH 4–7) followed by 11% SDS–PAGE then visualized by Bio-Safe Coomassie G250 staining. Isoelectric points (pI) and migrating positions of molecular mass (kDa) markers are marked. (B) Examples of spots corresponding to the differentially regulated proteins in Arabidopsis seedlings treated with 5 μM cytokinin (BA, iP, TDZ, or t-Z) or mock buffer for 15 min. For details see Materials and methods. (C) Average 2-DE phosphoproteome map of 7-day-old Arabidopsis seedlings treated with cytokinin/mock buffer for 15 min. Differentially regulated protein spots are indicated. See Table 2, and Supplementary Table S2, for detailed information on the corresponding identified proteins. Phosphoprotein fractions were obtained using a PhosphoProtein Purification Kit. Phosphoproteins (150 μg) were separated in the first and second dimensions by IPG (7 cm strips, pH 4–7) followed by 11% SDS–PAGE then visualized by Bio-Safe Coomassie G250 staining. Isoelectric points (pI) and relative migrating positions of molecular mass (kDa) markers are marked. (D) Examples of spots corresponding to the differentially regulated phosphoproteins in Arabidopsis seedlings treated with 5 μM cytokinin (BA, iP, TDZ, or t-Z) or mock buffer for 15 min. For details see Materials and methods. (This figure is available in colour at JXB online.) Table 2. Early cytokinin response phosphoproteins of Arabidopsis Spot/protein no.  AGI code  Protein name  PhosPhAt database  MALDI-MS (protein score/%cov/pep#)  Relative fold change             BA  iP  TDZ  t–Z  P1  At1g22530  Patellin-2 (PATL-2)  +  42/1/3  2.0±0.05  2.1±0.10  1.8±0.08  1.9±0.06  P2  At4g24190  Endoplasmin homologue (SHD)    76/10/7  1.9±0.07  1.4±0.11  1.7±0.09  1.7±0.20  P3  At5g56030  Heat shock protein 81-2  +  227/10/5  2.0±0.15  1.7±0.06  1.6±0.05  2.0±0.15  P4  At5g11170  DEAD-box ATP-dependent RNA helicase 15    33/4/2  1.6±0.18  1.7±0.05  1.8±0.06  1.5±0.15  P5  At5g22650  Histone deacetylase HDT2  +  86/11/3  1.6±0.13  2.0±0.31  2.3±0.28  1.6±0.18  P6  X  X  ?  X  2.3±0.15  2.5±0.34  2.5±0.26  2.0±0.21  P7  X  X  ?  X  1.7±0.15  2.0±0.32  1.7±0.21  2.5±0.13  P8  X  X  ?  X  2.2±0.10  1.7±0.09  1.6±0.05  1.4±0.08  P9  At1g09640  Probable elongation factor 1-γ 1    84/10/3  2.0±0.08  1.6±0.06  1.5±0.07  1.5±0.05  P10  At1g76180  Dehydrin ERD14  +  50/20/2  1.7±0.04  2.1±0.13  2.3±0.15  1.3±0.07  P11  At5g60640  Protein disulphide isomerase-like protein    333/22/9  –1.5±0.05  –1.4±0.07  –1.6±0.09  –1.8±0.09  P12  AtCg00120  ATP synthase subunit α, chloroplastic  +  701/25/9  –1.3±0.06  –1.8±0.11  –1.5±0.10  –1.5±0.06  P13  AtCg00490  Rubisco large chain  +  80/12/5  –1.5±0.09  –1.7±0.15  –1.9±0.12  –2.0±0.23    At1g67090  Rubisco small chain 1A, chloroplastic  +  78/19/3          P14  At2g39990  eIF2 (eukaryotic translation initiation factor)    139/17/3  –1.6±0.05  –1.5±0.17  –1.2±0.25  –2.0±0.12  P15  At5g14740  β-Carbonic anhydrase 2  +  383/25/4  –1.5±0.08  –1.5±0.09  –1.4±0.08  –1.4±0.10  P16  At5g43830  GATase-like protein  +  119/9/2  –1.4±0.04  –2.0±0.06  –1.5±0.05  –1.5±0.10  P17  At5g56030  Heat shock protein 81-2  +  574/19/11  –2.4±0.20  –2.5±0.33  –2.6±0.24  1.0±0.30  P18  At5g56030  Heat shock protein 81-2/3/4  +  63/8/4  –1.7±0.08  –1.6±0.06  –2.0±0.30  –1.5±0.05  P19  At3g16420  PBP1  +  70/12/2  –2.5±0.35  –1.8±0.18  –1.8±0.25  –1.3±0.20  P20  At5g42790  Proteasome subunit a type-1-A    94/27/4  –1.9±0.23  –2.0±0.15  –1.5±0.08  –1.5±0.08  P21  At3g51880  HMGB1  +  48/7/2  –1.6±0.10  –1.4±0.12  –1.3±0.07  –1.6±0.14  P22  At3g09200  60S Acidic ribosomal protein P0-2  +  124/18/3  –2.5±0.22  –2.0±0.32  –2.0±0.21  –1.2±0.20  P23  At1g20440  Dehydrin COR47  +  PMF: 106/43/9  1.7±0.08  1.6±0.15  1.3±0.15  1.1±0.18    At4g26110  NAP1  +  PMF: 61/23/7          P24  AtCg00490  Rubisco large subunit  +  291/22/7  1.6±0.10  1.6±0.08  1.4±0.18  1.3±0.15  P25  AtCg00490  Rubisco large subunit  +  317/16/7  –1.6±0.07  –1.6±0.05  –1.2±0.40  1.3±0.15  P26  At1g76180  Dehydrin ERD14  +  87/25/3  –1.8±0.05  –1.6±0.08  –1.6±0.05  –1.3±0.16  P27  At1g26630  eIF5A-2 (eukaryotic translation initiation factor)  +  95/28/3  1.7±0.08  1.4±0.14  1.3±0.10  1.5±0.05  P28  At1g20010  Tubulin β-5 chain    223/14/5  1.6±0.11  1.5±0.29  1.8±0.07  1.6±0.07  P29  X  X  ?  X  1.3±0.23  1.5±0.07  1.7±0.05  1.7±0.05  P30  At5g44340  Tubulin β-4 chain  +  488/23/11  1.4±0.15  1.4±0.05  1.5±0.04  1.4±0.15  P31  At3g09200  60S Acidic ribosomal protein P0-2  +  309/16/3  –1.3±0.20  –1.4±0.08  –1.5±0.10  –1.6±0.10  Spot/protein no.  AGI code  Protein name  PhosPhAt database  MALDI-MS (protein score/%cov/pep#)  Relative fold change             BA  iP  TDZ  t–Z  P1  At1g22530  Patellin-2 (PATL-2)  +  42/1/3  2.0±0.05  2.1±0.10  1.8±0.08  1.9±0.06  P2  At4g24190  Endoplasmin homologue (SHD)    76/10/7  1.9±0.07  1.4±0.11  1.7±0.09  1.7±0.20  P3  At5g56030  Heat shock protein 81-2  +  227/10/5  2.0±0.15  1.7±0.06  1.6±0.05  2.0±0.15  P4  At5g11170  DEAD-box ATP-dependent RNA helicase 15    33/4/2  1.6±0.18  1.7±0.05  1.8±0.06  1.5±0.15  P5  At5g22650  Histone deacetylase HDT2  +  86/11/3  1.6±0.13  2.0±0.31  2.3±0.28  1.6±0.18  P6  X  X  ?  X  2.3±0.15  2.5±0.34  2.5±0.26  2.0±0.21  P7  X  X  ?  X  1.7±0.15  2.0±0.32  1.7±0.21  2.5±0.13  P8  X  X  ?  X  2.2±0.10  1.7±0.09  1.6±0.05  1.4±0.08  P9  At1g09640  Probable elongation factor 1-γ 1    84/10/3  2.0±0.08  1.6±0.06  1.5±0.07  1.5±0.05  P10  At1g76180  Dehydrin ERD14  +  50/20/2  1.7±0.04  2.1±0.13  2.3±0.15  1.3±0.07  P11  At5g60640  Protein disulphide isomerase-like protein    333/22/9  –1.5±0.05  –1.4±0.07  –1.6±0.09  –1.8±0.09  P12  AtCg00120  ATP synthase subunit α, chloroplastic  +  701/25/9  –1.3±0.06  –1.8±0.11  –1.5±0.10  –1.5±0.06  P13  AtCg00490  Rubisco large chain  +  80/12/5  –1.5±0.09  –1.7±0.15  –1.9±0.12  –2.0±0.23    At1g67090  Rubisco small chain 1A, chloroplastic  +  78/19/3          P14  At2g39990  eIF2 (eukaryotic translation initiation factor)    139/17/3  –1.6±0.05  –1.5±0.17  –1.2±0.25  –2.0±0.12  P15  At5g14740  β-Carbonic anhydrase 2  +  383/25/4  –1.5±0.08  –1.5±0.09  –1.4±0.08  –1.4±0.10  P16  At5g43830  GATase-like protein  +  119/9/2  –1.4±0.04  –2.0±0.06  –1.5±0.05  –1.5±0.10  P17  At5g56030  Heat shock protein 81-2  +  574/19/11  –2.4±0.20  –2.5±0.33  –2.6±0.24  1.0±0.30  P18  At5g56030  Heat shock protein 81-2/3/4  +  63/8/4  –1.7±0.08  –1.6±0.06  –2.0±0.30  –1.5±0.05  P19  At3g16420  PBP1  +  70/12/2  –2.5±0.35  –1.8±0.18  –1.8±0.25  –1.3±0.20  P20  At5g42790  Proteasome subunit a type-1-A    94/27/4  –1.9±0.23  –2.0±0.15  –1.5±0.08  –1.5±0.08  P21  At3g51880  HMGB1  +  48/7/2  –1.6±0.10  –1.4±0.12  –1.3±0.07  –1.6±0.14  P22  At3g09200  60S Acidic ribosomal protein P0-2  +  124/18/3  –2.5±0.22  –2.0±0.32  –2.0±0.21  –1.2±0.20  P23  At1g20440  Dehydrin COR47  +  PMF: 106/43/9  1.7±0.08  1.6±0.15  1.3±0.15  1.1±0.18    At4g26110  NAP1  +  PMF: 61/23/7          P24  AtCg00490  Rubisco large subunit  +  291/22/7  1.6±0.10  1.6±0.08  1.4±0.18  1.3±0.15  P25  AtCg00490  Rubisco large subunit  +  317/16/7  –1.6±0.07  –1.6±0.05  –1.2±0.40  1.3±0.15  P26  At1g76180  Dehydrin ERD14  +  87/25/3  –1.8±0.05  –1.6±0.08  –1.6±0.05  –1.3±0.16  P27  At1g26630  eIF5A-2 (eukaryotic translation initiation factor)  +  95/28/3  1.7±0.08  1.4±0.14  1.3±0.10  1.5±0.05  P28  At1g20010  Tubulin β-5 chain    223/14/5  1.6±0.11  1.5±0.29  1.8±0.07  1.6±0.07  P29  X  X  ?  X  1.3±0.23  1.5±0.07  1.7±0.05  1.7±0.05  P30  At5g44340  Tubulin β-4 chain  +  488/23/11  1.4±0.15  1.4±0.05  1.5±0.04  1.4±0.15  P31  At3g09200  60S Acidic ribosomal protein P0-2  +  309/16/3  –1.3±0.20  –1.4±0.08  –1.5±0.10  –1.6±0.10  Spot no., spot number (as given in Fig. 1C); AGI code, accession number in the TAIR atabase; PhosPhAt, the Arabidopsis Protein Phosphorylation Site Database (Heazlewood et al., 2008); Protein name, entry name in the NCBI database; %cov, percentage of protein coverage; pep#, number of peptides; PMF, peptide mass fingerprint; Relative fold change, fold change relative to the mock control (calculated by DECODON DELTA 2D software) ±SE. Full information on the phosphoproteins including their classification, peptide sequences, and peak list is given in Supplementary Tables S2 and S4 at JXB online. View Large Previously, cytokinin early response transcripts were identified following 15 min treatment of 7-day-old Arabidopsis seedlings with 5 μM BA. Here it was confirmed that levels of type-A ARR genes (ARR3 and ARR5) increased following BA treatment in the experimental set-up employed using quantitative RT-PCR (P. Souček, unpublished data) as outlined in Souček et al. (2007). Identification of early cytokinin response phosphoproteins A fraction of phosphoproteins phosphorylated at serine and threonine residues was isolated from seedlings treated with cytokinins using a Quiagen Phosphoprotein enrichment kit with an optimized procedure, as outlined above. In addition, cytokinin receptor double mutants (ahk2ahk3, ahk2cre1, and ahk3cre1) treated with 5 μM t-Z for 15 min were analysed to assess how much the individual cytokinin receptors contribute to phosphoproteome regulation. Phosphoprotein fractions were subjected to 2-DE, and image analysis was used to reveal phosphoproteins differentially regulated by cytokinins (Fig. 1C, D), essentially as described above for early cytokinin response proteins. Of 450 reproducibly resolved spots in phosphoproteome maps from wild-type samples, significant differences (P <0.05) in all 10 independent experiments (including the four pilot experiments using only one cytokinin each) were found for 90 resolved spots (for a schematic representation of the experimental design see Supplementary Fig. S1 at JXB online). Reproducible significant changes in at least three biological replicates were found for 31 spots. Subsequently, >90% of them were reproducibly resolved in phosphoproteome maps displaying phosphoproteins from each of the three cytokinin receptor double mutants (Table 3). Table 3. Regulation of the early cytokinin response phosphoproteins by t-Z in the cytokinin receptor double mutants ahk2cre1, ahk3cre1, and ahk2ahk3 Spot/protein no.  AGI code  Protein name  Relative fold change         ahk2ahk3  ahk2cre1  ahk3cre1  Wild type  Significant response apparently mediated by a single cytokinin receptor  P1  At1g22530  Patellin-2 (PATL-2)  –1.1±0.13  1.5±0.25  1.1±0.04  1.9±0.06  P2  At4g24190  Endoplasmin homologue (SHD)  1.2±0.30  1.4±0.10  1.0±0.30  1.7±0.20  P3  At5g56030  Heat shock protein 81-2  1.5±0.08  1.0±0.05  1.3±0.32  2.0±0.15  P4  At5g11170  DEAD-box ATP-dependent RNA helicase 15  1.4±0.06  1.0±0.03  1.0±0.18  1.5±0.15  P7  X  X  1.0±0.26  1.5±0.07  1.0±0.26  2.5±0.13  P9  At1g09640  Probable elongation factor 1-γ 1  1.5±0.05  1.0±0.12  1.0±0.15  1.5±0.05  P11  At5g60640  Protein disulphide isomerase-like protein  1.0±0.28  –1.4±0.10  –1.1±0.15  –1.8±0.09  P12  AtCg00120  ATP synthase subunit α, chloroplastic  –1.5±0.19  –1.3±0.09  –1.2±0.11  –1.5±0.06  P14  At2g39990  eIF2 (eukaryotic translation initiation factor)  1.0±0.25  –1.1±0.17  –1.5±0.30  –2.0±0.12  P18  At5g56030  Heat shock protein 81-2/3/4  1.2±0.35  –1.4±0.01  –1.0±0.15  –1.5±0.05  P21  At3g51880  HMGB1  –1.5±0.15  –1.3±0.01  1.0±0.20  –1.6±0.14  P27  At1g26630  eIF5A-2 (eukaryotic translation initiation factor)  1.5±0.08  –1.1±0.14  –1.2±0.29  1.5±0.05  P28  At1g20010  Tubulin β-5 chain  1.7±0.23  1.0±0.26  1.0±0.05  1.6±0.07  P29  X  X  1.0±0.08  1.6±0.24  1.2±0.11  1.7±0.05  P31  At3g09200  60S Acidic ribosomal protein P0-2  –1.1±0.11  –1.4±0.02  –1.2±0.16  –1.6±0.10  Significant response apparently mediated by two cytokinin receptors  P13  AtCg00490  Rubisco large chain  1.6±0.21  –1.6±0.57  +/–  –2.0±0.23    At1g67090  Rubisco small chain 1A, chloroplastic          P15  At5g14740  β-Carbonic anhydrase 2  1.0±0.18  –1.4±0.14  –1.5±0.34  –1.4±0.10  P19  At3g16420  PBP1  –1.2±0.16  –1.5±0.21  –1.4±0.13  –1.3±0.20  P24  AtCg00490  Rubisco large subunit  +/–  –1.5±0.26  2.0±0.14  1.3±0.15  P25  AtCg00490  Rubisco large subunit  1.4±0.12  1.4±0.21  1.0±0.17  1.3±0.15  P26  At1g76180  Dehydrin ERD14  1.5±0.15  –1.4±0.32  –1.3±0.27  –1.3±0.16  P30  At5g44340  Tubulin β-4 chain  1.4±0.20  1.4±0.07  1.3±0.11  1.4±0.15  Non-significant response to t-Z in wild type and/or mutants  P10  At1g76180  Dehydrin ERD14  +/–  –1.3±0.33  +/–  1.3±0.07  P16  At5g43830  GATase like protein  1.1±0.20  1.0±0.20  –1.1±0.22  –1.5±0.10  P17  At5g56030  Heat shock protein 81-2  1.0±0.16  1.0±0.18  1.0±0.05  1.0±0.30  P20  At5g42790  Proteasome subunit a type-1-A  1.2±0.08  1.0±0.01  1.2±0.14  –1.5±0.08  P22  At3g09200  60S Acidic ribosomal protein P0-2  1.0±0.16  1.0±0.30  1.0±0.19  –1.2±0.20  P23  At1g20440  Dehydrin COR47  1.0±0.02  –1.2±0.34  1.0±0.22  1.1±0.18    At4g26110  NAP1          Spot/protein no.  AGI code  Protein name  Relative fold change         ahk2ahk3  ahk2cre1  ahk3cre1  Wild type  Significant response apparently mediated by a single cytokinin receptor  P1  At1g22530  Patellin-2 (PATL-2)  –1.1±0.13  1.5±0.25  1.1±0.04  1.9±0.06  P2  At4g24190  Endoplasmin homologue (SHD)  1.2±0.30  1.4±0.10  1.0±0.30  1.7±0.20  P3  At5g56030  Heat shock protein 81-2  1.5±0.08  1.0±0.05  1.3±0.32  2.0±0.15  P4  At5g11170  DEAD-box ATP-dependent RNA helicase 15  1.4±0.06  1.0±0.03  1.0±0.18  1.5±0.15  P7  X  X  1.0±0.26  1.5±0.07  1.0±0.26  2.5±0.13  P9  At1g09640  Probable elongation factor 1-γ 1  1.5±0.05  1.0±0.12  1.0±0.15  1.5±0.05  P11  At5g60640  Protein disulphide isomerase-like protein  1.0±0.28  –1.4±0.10  –1.1±0.15  –1.8±0.09  P12  AtCg00120  ATP synthase subunit α, chloroplastic  –1.5±0.19  –1.3±0.09  –1.2±0.11  –1.5±0.06  P14  At2g39990  eIF2 (eukaryotic translation initiation factor)  1.0±0.25  –1.1±0.17  –1.5±0.30  –2.0±0.12  P18  At5g56030  Heat shock protein 81-2/3/4  1.2±0.35  –1.4±0.01  –1.0±0.15  –1.5±0.05  P21  At3g51880  HMGB1  –1.5±0.15  –1.3±0.01  1.0±0.20  –1.6±0.14  P27  At1g26630  eIF5A-2 (eukaryotic translation initiation factor)  1.5±0.08  –1.1±0.14  –1.2±0.29  1.5±0.05  P28  At1g20010  Tubulin β-5 chain  1.7±0.23  1.0±0.26  1.0±0.05  1.6±0.07  P29  X  X  1.0±0.08  1.6±0.24  1.2±0.11  1.7±0.05  P31  At3g09200  60S Acidic ribosomal protein P0-2  –1.1±0.11  –1.4±0.02  –1.2±0.16  –1.6±0.10  Significant response apparently mediated by two cytokinin receptors  P13  AtCg00490  Rubisco large chain  1.6±0.21  –1.6±0.57  +/–  –2.0±0.23    At1g67090  Rubisco small chain 1A, chloroplastic          P15  At5g14740  β-Carbonic anhydrase 2  1.0±0.18  –1.4±0.14  –1.5±0.34  –1.4±0.10  P19  At3g16420  PBP1  –1.2±0.16  –1.5±0.21  –1.4±0.13  –1.3±0.20  P24  AtCg00490  Rubisco large subunit  +/–  –1.5±0.26  2.0±0.14  1.3±0.15  P25  AtCg00490  Rubisco large subunit  1.4±0.12  1.4±0.21  1.0±0.17  1.3±0.15  P26  At1g76180  Dehydrin ERD14  1.5±0.15  –1.4±0.32  –1.3±0.27  –1.3±0.16  P30  At5g44340  Tubulin β-4 chain  1.4±0.20  1.4±0.07  1.3±0.11  1.4±0.15  Non-significant response to t-Z in wild type and/or mutants  P10  At1g76180  Dehydrin ERD14  +/–  –1.3±0.33  +/–  1.3±0.07  P16  At5g43830  GATase like protein  1.1±0.20  1.0±0.20  –1.1±0.22  –1.5±0.10  P17  At5g56030  Heat shock protein 81-2  1.0±0.16  1.0±0.18  1.0±0.05  1.0±0.30  P20  At5g42790  Proteasome subunit a type-1-A  1.2±0.08  1.0±0.01  1.2±0.14  –1.5±0.08  P22  At3g09200  60S Acidic ribosomal protein P0-2  1.0±0.16  1.0±0.30  1.0±0.19  –1.2±0.20  P23  At1g20440  Dehydrin COR47  1.0±0.02  –1.2±0.34  1.0±0.22  1.1±0.18    At4g26110  NAP1          Spot no., spot number (as given in Fig. 1C); AGI code, accession number in the TAIR database; Protein name, entry name according to the NCBI database; Relative fold change, fold change relative to the mock control (calculated by DECODON DELTA 2D software) ±SE; +/–, inconsistent regulation in three biological replicas (down-, up-, and non-regulated in the individual biological replicas). Full information on the phosphoproteins including their classification, peptide sequences, and peak list is given in Supplementary Tables S2 and S4 at JXB online. View Large In total, 29 proteins were identified in these spots, including two protein mixtures, by MALDI-TOF/TOF MS followed by Mascot database searches of the full NCBI protein database (Table 2; Supplementary Table S2 at JXB online). Phosphorylation has been previously reported for 22 of these proteins (Table 2; PhosPhAt 3.0, http://phosphat.mpimp-golm.mpg.de). Here, phosphorylation was confirmed for 60S acidic ribosomal protein P0-2 (P22) and endoplasmin homologue (SHD; P2) by comparing MS spectra of their IMAC-purified peptides VEEKEESDEEDYGGDFGLFDEE and VGKDSKDKELKEAFK, respectively, before and after alkaline phosphatase treatment (Supplementary Fig. S2). Serine was previously shown to be a phosphorylation site on the peptide of 60S acidic ribosomal protein P0-2 (Laugesen et al., 2006), but phosphorylation of endoplasmin homologue (SHD) has not been previously reported. In addition, 29 of the 31 spots were stained by the phosphoprotein-specific stain Phos-tag™ (Supplementary Fig. S3). The ratio of numbers of up-regulated to down-regulated phosphoproteins was ∼1:1. Alterations in levels of individual phosphoproteins may result from phosphorylation/dephosphorylation events and/or modulation of turnover rates of the phosphoproteins. Identified protein spots are marked in protein maps shown in Fig. 1C, and examples of spots containing phosphoproteins differentially regulated by the individual cytokinins in Fig. 1D. The corresponding partial amino acid sequences are listed in Supplementary Table S4. Protein identifications and relative fold changes based on mean percentage volumes of these spots are presented in Tables 2 and 3 for phosphoproteins of the wild type and cytokinin receptor double mutants, respectively. Apparent strength of effects of the cytokinins on expression of the early cytokinin response phosphoproteins decreased in the order BA>TDZ>t-Z=iP. Responses specific for a single receptor were found in 15 spots, while seven spots were regulated by two individual receptors (Table 3). The remaining spots were either non-significantly or inconsistently (down-, up-, and non-regulated in the individual biological replicas) regulated. Interestingly, regulation was apparent in the mutants for four spots (P19, P24, P25, and P26) that remained below cut-off limits in wild-type seedlings. The opposite regulation was found for spots P13, P26 (ahk2ahk3 and ahk2cre1), and P24 (ahk2cre1 and ahk3cre1), suggesting receptor interactions in response regulation. Further, loss of consistent regulation in the double mutants for two spots regulated by t-Z in the wild type (P16, P20) was observed, implying that simultaneous activity of at least two receptors may be needed for correct regulation of the corresponding proteins. In addition, responses of a fraction of the 15 spots primarily regulated by a single receptor to cytokinin treatment were lower in all three double mutants than in wild-type plants, suggesting they may require simultaneous activity of one or more other receptor(s) for a full response. The highest numbers of regulated spots were found in the ahk2cre1 (14) mutant, followed by ahk2ahk3 (11) and ahk3cre1 (four). Calcium signalling in regulation of early cytokinin response phosphoproteins Recognition of ERD14 (P10, P26) and COR47 (P23), in which phosphorylation status and Ca2+ binding are reportedly interlinked, as early cytokinin response phosphoproteins suggested a molecular link between cytokinin action and calcium signalling. To test the involvement of calcium signalling in early phosphoproteome regulation by cytokinin, 7-day-old Arabidopsis seedlings were treated with 5 μM t-Z in the presence and absence of a calcium channel blocker (30 μM D600) and a competitive inhibitor of calcium uptake (60 μM LaCl3) for 15 min, and phosphoproteome alterations were analysed as outlined above. This resulted in identification of five phosphoproteins in which regulation by t-Z was lost in the presence of the calcium signalling inhibitors (Fig. 2), while the remaining pattern of phosphoprotein regulation remained unaltered compared with data given in Table 2. Fig. 2. View largeDownload slide Effect of calcium signalling inhibitors on regulation by cytokinin of early cytokinin response phosphoproteins. (A) Selected regions of 2D gels showing early cytokinin response phosphoproteins (indicated by arrows) whose regulation by 15 min treatment with 5 μM t-Z in Arabidopsis seedlings (t-Z) was abolished when calcium signalling inhibitors 30 μM D600 and 60 μM LaCl3 were administered simultaneously with 5 μM t-Z (t-Z + INH). Control samples were treated with the inhibitors only (INH). For details see Materials and methods. Spot numbers as in Fig. 1C and Table 2. (B) Relative volumes of the individual spots. Fig. 2. View largeDownload slide Effect of calcium signalling inhibitors on regulation by cytokinin of early cytokinin response phosphoproteins. (A) Selected regions of 2D gels showing early cytokinin response phosphoproteins (indicated by arrows) whose regulation by 15 min treatment with 5 μM t-Z in Arabidopsis seedlings (t-Z) was abolished when calcium signalling inhibitors 30 μM D600 and 60 μM LaCl3 were administered simultaneously with 5 μM t-Z (t-Z + INH). Control samples were treated with the inhibitors only (INH). For details see Materials and methods. Spot numbers as in Fig. 1C and Table 2. (B) Relative volumes of the individual spots. Comparison of proteome and phosphoproteome data sets The sets of differentially expressed proteins and phosphoproteins (Fig. 1) included seven overlapping proteins: At3g09200 (T34, P22, P31), At3g16420 (T38, P19), At4g24190 (T2, P2), At5g14740 (T42, P15), At5g60640 (T11, P11), AtCg00120 (T12, T16, P12), and AtCg00490 (T13, T18, P13). However, the apparent pI and molecular mass values were indistinguishable in proteome and phosphoproteome maps for only one of them, endoplasmin homologue (At4g24190), indicating that the proteins displayed in the protein and phosphoprotein maps may be subjected to distinct PTMs affecting their apparent pI, molecular mass, or both. Classification of identified proteins and phosphoproteins Identified proteins were categorized using criteria described by Bevan et al. (1998). As shown in Fig. 3A, significant fractions of the phosphoproteins are involved in responses to environmental (stress, light, and temperature) stimuli (36%), and signalling and development pathways (27%). Other highly represented functional categories are protein biosynthesis (17%) and transcription regulators (10%). Given previously reported data on the functional significance of phosphorylation of the phosphoproteins identified here, the present data set also indicates that phosphorylation regulation by cytokinins might mainly regulate protein–protein and protein–ligand/substrate interactions (Fig. 3B). Fig. 3. View largeDownload slide (A) Classification of the early cytokinin response phosphoproteins according to their cellular functions (Bevan et al., 1998) and (B) molecular processes reportedly controlled by their phosphorylation as deduced from database entries and literature review. Fig. 3. View largeDownload slide (A) Classification of the early cytokinin response phosphoproteins according to their cellular functions (Bevan et al., 1998) and (B) molecular processes reportedly controlled by their phosphorylation as deduced from database entries and literature review. To gain deeper insights into biological processes in which the differentially regulated proteins are involved, the gene ontology (GO) of proteins showing significant (P <0.05), ≥1.4-fold changes in expression between cytokinin-challenged and control samples was analysed. The results were visualized using BiNGO, a graphical tool enabling GO classes in clustered data to be highlighted (http://www.psb.ugent.be/cbd/papers/BiNGO/) (Fig. 4). The GO categories that were identified as being significantly over-represented were ‘Growth and development’ (including subclasses such as ‘Thylakoid development’ and ’Protein transport’), ‘Nitrogen metabolism’, ‘Hormone signalling’, ‘Photosynthesis and light’ (e.g. ‘Response to light stimulus’, ‘Response to light intensity’, and ‘Photosynthesis–light reaction and light harvesting’), and ‘Toxicity’. Further, ‘Temperature response’ was the only down-regulated GO category. Fig. 4. View largeDownload slide Gene ontology (GO) analysis of the early cytokinin response proteins in Arabidopsis (performed in Cytoscape using BiNGO plugin version 2.3). GO categories that were significantly over-represented among the differentially expressed proteins were identified. The yellow to orange colour of the circles indicates the level of significance of over-represented categories (P=0.05, yellow; P=10−7, orange). The size of the circles is proportional to the number of proteins in each category. Links with low significance were removed manually to reduce complexity of the image. Fig. 4. View largeDownload slide Gene ontology (GO) analysis of the early cytokinin response proteins in Arabidopsis (performed in Cytoscape using BiNGO plugin version 2.3). GO categories that were significantly over-represented among the differentially expressed proteins were identified. The yellow to orange colour of the circles indicates the level of significance of over-represented categories (P=0.05, yellow; P=10−7, orange). The size of the circles is proportional to the number of proteins in each category. Links with low significance were removed manually to reduce complexity of the image. The subcellular location of each identified protein was determined according to the TAIR database (http://www.arabidopsis.org), and the results are summarized in Fig. 5. The largest group of proteins was localized to chloroplasts (52%), followed by the cytoplasm (17%) and mitochondria (14%). Among phosphoproteins, 31, 25, and 19% were localized to chloroplasts, the nucleus, and cytoplasm, respectively. Fig. 5. View largeDownload slide Subcellular distribution of the early cytokinin response proteins (white) and phosphoproteins (grey) according to the UniProt database (http://www.uniprot.org). The numbers above the columns represent sums of the identified proteins and phosphoproteins located in the respective cellular compartment. Fig. 5. View largeDownload slide Subcellular distribution of the early cytokinin response proteins (white) and phosphoproteins (grey) according to the UniProt database (http://www.uniprot.org). The numbers above the columns represent sums of the identified proteins and phosphoproteins located in the respective cellular compartment. Discussion Proteomic and phosphoproteomic effects of cytokinin treatment on Arabidopsis seedlings were analysed to identify early cytokinin response proteins and phosphoproteins in order to elucidate molecular mechanisms involved in cytokinin action. The identified proteins and phosphoproteins represent a snapshot of early links in various well known cytokinin-regulated signalling circuits and cellular processes. The results also indicate as yet unrecognized links between temperature, calcium, and cytokinin signalling. Comparative analysis revealed differences in both the potency of the four representative cytokinins to trigger the responses, and the contributions of specific cytokinin receptors to phosphoproteomic responses to t-Z treatment. Phosphoproteome isolation and analysis A previously established procedure for isolating mammalian and yeast phosphoproteins by affinity chromatography prior to further analysis by 2-DE and MS (Makrantoni et al., 2005) was optimized and employed in this study. The procedure is specific for proteins phosphorylated at serine or threonine residues. Thus, it is capable of detecting most phosphorylated proteins in eukaryotic cells since their pSer:pThr:pTyr and pHis/pTyr ratios are typically 1800:200:1 and 10–100:1, respectively (Klumpp and Krieglstein, 2002; Laugesen et al., 2004). The procedure is reportedly reliable for plant phosphoproteome analysis (Laugesen et al., 2006; Meimoun et al., 2007), and sample preparation for the 2-DE stage of the protocol has been further improved. All proteins identified on the phosphoproteome map show the predicted phosphorylation sites in NetPhos (Blom et al., 1999). Most are already included in the PhosPhAt database (Heazlewood et al., 2008), and the possibility of PTM by phosphorylation of the others has been previously documented (Table 2). Here, phosphorylation for 60S acidic ribosomal protein P0-2 (P22) and endoplasmin homologue (SHD; P2) was confirmed directly, and it was shown that all but two differentially regulated spots are stained by the phosphoprotein-specific stain Phos-Tag (Supplementary Fig. S3 at JXB online). Further, phosphorylation-dependent shifts in apparent (SDS–PAGE) molecular masses of COR47 and ERD14 have been reported (Alsheikh et al., 2005), and increases in apparent molecular masses of these proteins after cytokinin stimulation (P10, P23) were observed. Based on previous reports, the changes in phosphoprotein levels detected here may reflect phosphorylation events mainly involved in regulation of protein–protein or protein–ligand/substrate interactions. Modulation of enzyme activity and specificity by phosphorylation has been shown for histone deacetylase (Pflum et al., 2001), protein disulphide isomerase (Guthapfel et al., 1996), β-carbonic anhydrase (Church et al., 1980), Rubisco (Aggarwal et al., 1993), RNA helicase (Yang et al., 2005), and ATP synthase (Murtazina et al., 2001). Phosphorylation also reportedly mediates association and assembly of protein complexes of Rubisco (Aggarwal et al., 1993), 60S acidic ribosomal protein P0-2 (Naranda and Ballesta, 1991), endoplasmin (Brunati et al., 2000), tubulins (Blume et al., 2009), proteasome subunits (Umeda et al., 1997), and heat shock proteins (Picard, 2002). Down-regulation of eIF2 (P14), and up-regulation of eIF5A-2 and probable elongation factor 1-γ1 (P28, P9) are consistent with previously reported stimulation of proteosynthesis by cytokinins as eIF2 reportedly inhibits initiation of proteosynthesis in its phosphorylated form (Zhang et al., 2008) while phosphorylation of eIF5 and EF-1 reportedly stimulates formation of initiation complexes (Majumdar et al., 2002) and enhances proteosynthesis (Belle et al., 1995), respectively. Further, phosphorylation reportedly promotes NAP1 import into the nucleus (Calvert et al., 2008), while HMGB1 requires phosphorylation for export from the nucleus (Youn and Shin, 2006). Finally, several phosphoproteins reportedly related to other signalling pathways were identified that may be involved in as yet unrecognized branches of cytokinin signalling and/or as molecular players in cross-talk between cytokinins and other stimuli. For example, PATL-2 (P1) is reportedly involved in membrane trafficking events (Peterman et al., 2004) and its phosphorylation has been localized to its phosphoinositide-binding pocket (Jones et al., 2009). This is consistent with recent indications of a role for intracellular trafficking in cytokinin signalling (Dortay et al., 2008). Temperature perception A novel theme highlighted by the present analysis is differential regulation by cytokinins of proteins and phosphoproteins reportedly involved in responses to high and low temperatures (Nylander et al., 2001; Sung et al., 2001; Bae et al., 2003; Goulas et al., 2006; Lim et al., 2006; Sasaki et al., 2007). The proteins include heat shock cognate 70 kDa protein 1, fructose-1,6-bisphosphatase, phosphoribulokinase, phosphoglycerate kinase, 60S acidic ribosomal protein P0-2, putative ribonucleoprotein At2g37220, and cold shock domain protein 2, and the phosphoproteins comprise heat shock protein 81-2, 60S acidic ribosomal protein P0-2, and dehydrins COR47 and ERD14. Interestingly, all the proteins were down-regulated by cytokinins, suggesting there may be shared components of cytokinin and temperature signalling pathways. In plants, mechanisms underlying temperature perception are poorly understood (Penfield, 2008). However, a two-component signalling pathway is known to act in temperature perception in cyanobacteria (e.g. Synechocystis), with a histidine kinase perceiving and relaying temperature signals (Suzuki et al., 2000), and a two-component signalling pathway is the main known component of cytokinin signalling chains in plants. Accordingly, a role for cytokinins in responses to cold stress was recently deduced from the apparent attenuation of cytokinin signalling under cold stress (Argueso et al., 2009). Decreases in levels of endogenous cytokinins in heat shock-treated plants have also been documented (Hare et al., 1997), and Burkhanova et al. (2001) found that responses to BA were enhanced in heat-shocked Arabidopsis thaliana, prompting speculation that heat shock proteins may be involved in cytokinin signalling. Chloroplast biogenesis and function The differentially regulated proteins and phosphoproteins detected in the presented experiments include a remarkably high percentage of proteins located to chloroplasts—45%—compared with 7.9% chloroplast proteins predicted for the whole genome (Bevan et al., 1998). They reflect most processes involved in chloroplast biogenesis and function, including: mRNA processing; protein biosynthesis, folding, and degradation; light and dark reactions of photosynthesis; carbon utilization; carbohydrate metabolism; glycolysis; fatty acid biosynthesis; and stress responses. Both the number and variety of functions in which they are implicated substantially exceed estimates obtained in two previous proteomic analyses of cytokinin action. In Physcomitrella patens, four early response phosphoproteins located to chloroplasts were identified following short cytokinin treatment (Heintz et al., 2006), and 10 up-regulated chloroplast proteins were found when effects of continuously increasing endogenous cytokinin levels in dark-grown Arabidopsis seedlings were investigated (Lochmanová et al., 2008). For two of the Physcomitrella phosphoproteins (Rubisco large subunit and carbonic anhydrase) functional matches are present in our data set. The high number of early cytokinin response proteins and phosphoproteins located in chloroplasts might indicate an as yet uncharacterized direct branch in cytokinin signalling responsible for cytokinin action in chloroplasts. Consistently, 10 interactors of cytokinin receptors have been located to chloroplasts (Dortay et al., 2008), and one of them, Rubisco small chain 1A (At1g67090), was identified here as a constituent of the differentially regulated spot P13. Brenner et al. (2005) have proposed that rapid transfer of cytokinin signals to plastids, or direct perception and interpretation of the signals by the plastids, may explain the fast regulation of five chloroplast transcripts by cytokinin treatment. Interestingly, the chloroplast cytokinin pool has been found to be dynamic (Benková et al., 1999), and compartmentation into chloroplasts of some cytokinin biosynthesis and metabolism pathways has been reported (Brzobohatý et al., 1993; Kristoffersen et al., 2000; Takei et al., 2004; Kiran et al., 2006). Chromatin remodelling and nuclear proteins Novel insights into possible involvement of cytokinins in chromatin remodelling were obtained by identifying HMGB1, histone deacetylase HDT2, and possibly NAP1;1 as early cytokinin response phosphoproteins. Interestingly, cytokinin response genes have been found to be up-regulated in hmgb1 knockout mutants (Lildballe et al., 2008), while NAP1;2 protein reportedly interacts with ARR7 (Dortay et al., 2008), and NAP1 proteins are positive regulators in the ABA signalling pathway (Liu et al., 2009). The expression level of genes involved in chromatin remodelling has been found to change after just 120 min cytokinin treatment (Brenner et al., 2005), clearly demonstrating the power of proteomic profiling for identifying primary events in cytokinin action. Cytokinin and cytokinin receptor specificity in eliciting proteomic/phosphoproteomic alterations The present experiments showed that the representatives of the four cytokinin classes had largely similar qualitative proteomic and phosphoproteomic effects, although the extent of up- or down-regulation varied for most proteins and phosphoproteins. The varying degree of responsiveness might reflect the molecular basis of differential activities of different cytokinin classes previously observed in various physiological experiments (e.g. Mok et al., 1978; Sujatha and Reddy, 1998; Lexa et al., 2003; Hradilová et al., 2007), and is consistent with distinct specificities of the individual cytokinin receptors for individual cytokinin moieties (Spíchal et al., 2004; Yonekura-Sakakibara et al., 2004; Romanov et al., 2006). Generally, the highest degree of differential regulation of proteins and phosphoproteins was elicited by BA and TDZ (reportedly the most potent cytokinin in activation of the cytokinin primary response gene ARR5 in Arabidopsis; Spíchal et al., 2004). The present study is the first global comparison of the effects of all four classes of cytokinins. However, Rashotte et al. (2003) compared the effects of BA and t-Z on expression profiles in Arabidopsis and found that while both cytokinins up-regulated largely overlapping sets of genes, far fewer genes were found to be down-regulated by t-Z than by BA, concluding that some of these genes may be specifically down-regulated by BA or not actually regulated by cytokinin. The profiling of t-Z action in the cytokinin receptor double mutants ahk2ahk3, ahk2cre1, and ahk3cre1 provides first insights into the specificity of outputs of specific cytokinin receptors at the proteome level. Most of the identified phosphoproteins were found to be differentially regulated primarily by a single cytokinin receptor. However, indications of inter-receptor cooperation were seen for some of the differentially regulated phosphoproteins. Detection of a significant number of phosphoproteins regulated by two individual cytokinin receptors was consistent with reportedly overlapping functions of the cytokinin receptors (Inoue et al., 2001; Higuchi et al., 2004; Nishimura et al., 2004; Riefler et al., 2006). Defining the output of cytokinin receptors as the number of phosphoproteins they apparently regulate, their output decreases in the order AHK3>AHK4/CRE1/WOL1>>AHK2. This observation is consistent with degrees of involvement of the cytokinin receptors in various biological processes previously found in morphological, physiological, and molecular analysis of cytokinin receptor mutants (Riefler et al., 2006). Calcium signalling The present work provides the first molecular link between cytokinin action and signalling pathways involving modulation of free Ca2+ levels, showing that early cytokinin response phosphoproteins include ERD14 and COR47, for which correlations between phosphorylation status and Ca2+ binding have been demonstrated (Alsheikh et al., 2005). In addition, it is shown that inhibition of calcium signalling abolishes cytokinin regulation of several phosphoproteins, further supporting the interlinking of cytokinin and calcium signalling. In planta, inhibition of calcium signalling disrupts cytokinin-induced bud formation in the moss Funaria (Saunders and Helper, 1983). Ca2+ signalling is reportedly involved in the transduction of diverse abiotic, biotic, and developmental stimuli including temperature and plant hormones (Sanders et al., 2002). In this context, increases in the phosphorylation of SHD and HSP 81-2 (P2, P3), proteins related to the hsp-90 family, whose members are known to be autophoshorylated in the presence of Ca2+ (Csermely and Kahn, 1991), were also observed. Further, Ca2+-mediated signalling may represent a rapid mechanism of transmitting cytokinin signals into chloroplasts. It has long been established that chloroplast-localized physiological processes are subject to regulation by Ca2+ and, accordingly, a Ca2+-sensing receptor has been localized to the chloroplast and found to modulate cytoplasmic Ca2+ concentrations (Nomura et al., 2008; Weinl et al., 2008). Conclusion In conclusion, a novel proteome- and phosphoproteome-wide view of changes in abundance of proteins and phosphoproteins is presented that might be functionally relevant for the many biological processes regulated by cytokinins. Importantly, the results indicate as yet unrecognized links between temperature, calcium, and cytokinin signalling. Interlinking of cytokinin and calcium signalling is further supported by loss of cytokinin regulation of several phosphoproteins following inhibition of calcium signalling. Rapid regulation of a number of chloroplast phosphoproteins suggests a currently uncharacterized direct signalling chain responsible for cytokinin action in chloroplast. Comparative analysis of the representatives of the four cytokinin classes revealed largely similar regulation patterns in the 7-day-old Arabidopsis seedlings. First insights into the specificity of cytokinin receptors on phosphoproteomic effects were obtained from analysis of cytokinin action in the set of cytokinin receptor double mutants. The presented data provide a new framework for further detailed investigations, using, for example, mutants and transgenic plants, of molecular mechanisms involved in cytokinin action. Further identification of kinase(s) and phosphatase(s) involved in phosphorylation and dephosphorylation events triggered by cytokinins and elucidation of their relationships to cytokinin receptors are important challenges for future work. Interestingly, two protein kinases have been identified as AHK2 and AHK4 interactors, and a phosphatase as an ARR4 interactor (Dortay et al., 2008). We thank Professor Thomas Schmülling for ahk2ahk3, ahk2cre1, and ahk3cre1 seeds, and Dr Přemysl Souček for providing us with unpublished ARR3 and ARR5 expression data. This work was supported by grants LC06034 and 1M06030 (Ministry of Education of the Czech Republic), IAA600040701 (Grant Agency of the Academy of Sciences of the Czech Republic), GACR 206/09/2062 (Grant Agency of the Czech Republic), and AV0Z50040507, AV0Z50040702 and AV0Z40310501 (Academy of Sciences of the Czech Republic). References Aggarwal KK,  Saluja D,  Sachar RC.  Phosphorylation of rubisco in Cicer arietinum: non-phosphoprotein nature of rubisco in Nicotiana tabacum,  Phytochemistry ,  1993, vol.  34 (pg.  329- 335) Google Scholar CrossRef Search ADS   Alsheikh MK,  Svensson JT,  Randall SK.  Phosphorylation regulated ion-binding is a property shared by the acidic subclass dehydrins,  Plant, Cell and Environment ,  2005, vol.  28 (pg.  1114- 1122) Google Scholar CrossRef Search ADS   Argueso CT,  Ferreira FJ,  Kieber JJ.  Environmental perception avenues: the interaction of cytokinin and environmental response pathways,  Plant, Cell and Environment ,  2009, vol.  32 (pg.  1147- 1160) Google Scholar CrossRef Search ADS   Bae MS,  Cho EJ,  Choi E,  Park OK.  Analysis of the Arabidopsis nuclear proteome and its response to cold stress,  The Plant Journal ,  2003, vol.  36 (pg.  652- 663) Google Scholar CrossRef Search ADS PubMed  Benková E,  Witters E,  Van Dongen W,  Kolář J,  Motyka V,  Brzobohatý B,  Van Onckelen HA,  Macháčková I.  Cytokinins in tobacco and wheat chloroplasts: occurrence and changes due to light/dark treatment,  Plant Physiology ,  1999, vol.  121 (pg.  245- 251) Google Scholar CrossRef Search ADS PubMed  Belle R,  Minella O,  Cormier P,  Morales J,  Poulhe R,  Mulner-Lorillon O.  Phosphorylation of elongation factor-1 (EF-1) by cdc2 kinase,  Progress in Cell Cycle Research ,  1995, vol.  1 (pg.  265- 270) Google Scholar PubMed  Bevan M,  Bancroft I,  Bent E, et al.  Analysis of 1.9 Mb of contiguous sequence from chromosome 4 of Arabidopsis thaliana,  Nature ,  1998, vol.  391 (pg.  485- 488) Google Scholar CrossRef Search ADS PubMed  Blom N,  Gammeltoft S,  Brunak S.  Sequence and structure-based prediction of eukaryotic protein phosphorylation sites,  Journal of Molecular Biology ,  1999, vol.  294 (pg.  1351- 1362) Google Scholar CrossRef Search ADS PubMed  Blume YB,  Lloyd CW,  Yemets AI.  Blume Y,  Baird W,  Yemets A,  Breviario D.  Plant tubulin phosphorylation and its role in cell cycle progression,  The plant cytoskeleton: a key tool for agro-biotechnology ,  2009, vol.  Vol. III.  Dordrecht, The Netherlands Springer(pg.  145- 159) Bradford MM.  A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein–dye binding,  Analytical Biochemistry ,  1976, vol.  72 (pg.  248- 254) Google Scholar CrossRef Search ADS PubMed  Brenner WG,  Romanov GA,  Köllmer I,  Bürkle L,  Schmülling T.  Immediate-early and delayed cytokinin response genes of Arabidopsis thaliana identified by genome-wide expression profiling reveal novel cytokinin-sensitive processes and suggest cytokinin action through transcriptional cascades,  The Plant Journal ,  2005, vol.  44 (pg.  314- 333) Google Scholar CrossRef Search ADS PubMed  Brunati AM,  Contri A,  Muenchbach M,  James P,  Marin O,  Pinna LA.  Grp94 (endoplasmin) co-purifies with and is phosphorylated by golgi apparatus casein kinase,  FEBS Letters ,  2000, vol.  471 (pg.  151- 155) Google Scholar CrossRef Search ADS PubMed  Brzobohatý B,  Moore I,  Kristoffersen P,  Bako L,  Campos N,  Schell J,  Palme K.  Release of active cytokinin by a β-glucosidase localized to the maize root meristem,  Science ,  1993, vol.  262 (pg.  1051- 1054) Google Scholar CrossRef Search ADS PubMed  Brzobohatý B,  Moore I,  Palme K.  Cytokinin metabolism: implications for regulation of plant growth and development,  Plant Molecular Biology ,  1994, vol.  26 (pg.  1483- 1497) Google Scholar CrossRef Search ADS PubMed  Burkhanova EA,  Mikulovich TP,  Kudryakova NV,  Kukina IM,  Smith AR,  Hall MA,  Kulaeva ON.  Heat shock pre-treatment enhances the response of Arabidopsis thaliana leaves and Cucurbita pepo cotyledons to benzyladenine,  Plant Growth Regulation ,  2001, vol.  33 (pg.  195- 198) Google Scholar CrossRef Search ADS   Calvert ME,  Keck KM,  Ptak C,  Shabanowitz J,  Hunt DF,  Pemberton LF.  Phosphorylation by casein kinase 2 regulates Nap1 localization and function,  Molecular and Cellular Biology ,  2008, vol.  28 (pg.  1313- 1325) Google Scholar CrossRef Search ADS PubMed  Choi J,  Hwang I.  Cytokinin: perception, signal transduction, and role in plant growth and development,  Journal of Plant Biology ,  2007, vol.  50 (pg.  98- 108) Google Scholar CrossRef Search ADS   Church GA,  Kimelberg HK,  Sapirstein VS.  Stimulation of carbonic anhydrase activity and phosphorylation in primary astroglial cultures by norepinephrine,  Journal of Neurochemistry ,  1980, vol.  34 (pg.  873- 879) Google Scholar CrossRef Search ADS PubMed  Csermely P,  Kahn CR.  The 90-kDa heat shock protein (hsp-90) possesses an ATP binding site and autophosphorylating activity,  Journal of Biological Chemistry ,  1991, vol.  266 (pg.  4943- 4950) Google Scholar PubMed  Damerval C,  De Vienne D,  Zivy M,  Thiellement H.  Technical improvements in two-dimensional electrophoresis increase the level of genetic variation detected in wheat-seedling proteins,  Electrophoresis ,  1986, vol.  7 (pg.  52- 54) Google Scholar CrossRef Search ADS   Dortay H,  Gruhn N,  Pfeifer A,  Schwerdtner M,  Schmülling T,  Heyl A.  Toward an interaction map of the two-component signaling pathway of Arabidopsis thaliana,  Journal of Proteome Research ,  2008, vol.  7 (pg.  3649- 3660) Google Scholar CrossRef Search ADS PubMed  Ferreira FJ,  Kieber JJ.  Cytokinin signaling,  Current Opinion in Plant Biology ,  2005, vol.  8 (pg.  518- 525) Google Scholar CrossRef Search ADS PubMed  Goulas E,  Schubert M,  Kieselbach T,  Kleczkowski LA,  Gardeström P,  Schröder W,  Hurry V.  The chloroplast lumen and stromal proteomes of Arabidopsis thaliana show differential sensitivity to short- and long-term exposure to low temperature,  The Plant Journal ,  2006, vol.  47 (pg.  720- 734) Google Scholar CrossRef Search ADS PubMed  Guthapfel R,  Gueguen P,  Quemeneur E.  ATP binding and hydrolysis by the multifunctional protein disulfide isomerase,  Journal of Biological Chemistry ,  1996, vol.  271 (pg.  2663- 2666) Google Scholar CrossRef Search ADS PubMed  Hare PD,  Cress WA,  van Staden J.  The involvement of cytokinins in plant responses to environmental stress,  Plant Growth Regulation ,  1997, vol.  23 (pg.  79- 103) Google Scholar CrossRef Search ADS   Heazlewood JL,  Durek P,  Hummel J,  Selbig J,  Weckwerth W,  Walther D,  Schulze WX.  Phosphat: a database of phosphorylation sites in Arabidopsis thaliana and a plant-specific phosphorylation site predictor,  Nucleic Acids Research ,  2008, vol.  36 (pg.  D1015- D1021) Google Scholar CrossRef Search ADS PubMed  Heintz D,  Erxleben A,  High AA,  Wurtz V,  Reski R,  Van Dorsselaer A,  Sarnighausen E.  Rapid alteration of the phosphoproteome in the Moss physcomitrella patens after cytokinin treatment,  Journal of Proteome Research ,  2006, vol.  5 (pg.  2283- 2293) Google Scholar CrossRef Search ADS PubMed  Higuchi M,  Pischke MS,  Mähönen AP, et al.  In planta functions of the Arabidopsis cytokinin receptor family,  Proceedings of the National Academy of Sciences, USA ,  2004, vol.  101 (pg.  8821- 8826) Google Scholar CrossRef Search ADS   Hoth S,  Ikeda Y,  Morgante M,  Wang X,  Zuo J,  Hanafey MK,  Gaasterland T,  Tingey SV,  Chua N.  Monitoring genome-wide changes in gene expression in response to endogenous cytokinin reveals targets in Arabidopsis thaliana,  FEBS Letters ,  2003, vol.  554 (pg.  373- 380) Google Scholar CrossRef Search ADS PubMed  Hradilová J,  Malbeck J,  Brzobohatý B.  Cytokoinin regulation of gene expression in the AHP gene family in Arabidopsis thaliana,  Journal of Plant Growth Regulation ,  2007, vol.  26 (pg.  229- 244) Google Scholar CrossRef Search ADS   Hradilová J,  Řehulka P,  Řehulková H,  Vrbová M,  Griga M,  Brzobohatý B.  Comparative analysis of proteomic changes in contrasting flax cultivars upon cadmium exposure,  Electrophoresis ,  2010, vol.  31 (pg.  421- 431) Google Scholar CrossRef Search ADS PubMed  Inoue T,  Higuchi M,  Hashimoto Y,  Seki M,  Kobayashi M,  Kato T,  Tabata S,  Shinozaki K,  Kakimoto T.  Identification of CRE1 as a cytokinin receptor from Arabidopsis,  Nature ,  2001, vol.  409 (pg.  1060- 1063) Google Scholar CrossRef Search ADS PubMed  Jones AME,  MacLean D,  Studholme DJ,  Serna-Sanz A,  Andreasson E,  Rathjen JP,  Peck SC.  Phosphoproteomic analysis of nuclei-enriched fractions from Arabidopsis thaliana,  Journal of Proteomics ,  2009, vol.  72 (pg.  439- 451) Google Scholar CrossRef Search ADS PubMed  Kakimoto T.  Biosynthesis of cytokinins,  Journal of Plant Research ,  2003, vol.  116 (pg.  233- 239) Google Scholar CrossRef Search ADS PubMed  Kiba T,  Naitou T,  Koizumi N,  Yamashino T,  Sakakibara H,  Mizuno T.  Combinatorial microarray analysis revealing Arabidopsis genes implicated in cytokinin responses through the His→Asp phosphorelay circuitry,  Plant and Cell Physiology ,  2005, vol.  46 (pg.  339- 355) Google Scholar CrossRef Search ADS PubMed  Kiran NS,  Polanská L,  Fohlerová R, et al.  Ectopic over-expression of the maize β-glucosidase Zm-p60.1 perturbs cytokinin homeostasis in transgenic tobacco,  Journal of Experimental Botany ,  2006, vol.  57 (pg.  985- 996) Google Scholar CrossRef Search ADS PubMed  Klumpp S,  Krieglstein J.  Phosphorylation and dephosphorylation of histidine residues in proteins,  European Journal of Biochemistry ,  2002, vol.  269 (pg.  1067- 1071) Google Scholar CrossRef Search ADS PubMed  Kristoffersen P,  Brzobohatý B,  Höhfeld I,  Bako L,  Melkonian M,  Palme K.  Developmental regulation of the maize Zm-p60.1 gene encoding a β-glucosidase located to plastids,  Planta ,  2000, vol.  210 (pg.  407- 415) Google Scholar CrossRef Search ADS PubMed  Larsen MR,  Sørensen GL,  Fey SJ,  Larsen PM,  Roepstorff P.  Phospho-proteomics: evaluation of the use of enzymatic de-phosphorylation and differential mass spectrometric peptide mass mapping for site specific phosphorylation assignment in proteins separated by gel electrophoresis,  Proteomics ,  2001, vol.  1 (pg.  223- 228) Google Scholar CrossRef Search ADS PubMed  Laugesen S,  Bergoin A,  Rossignol M.  Deciphering the plant phosphoproteome: tools and strategies for a challenging task,  Plant Physiology and Biochemistry ,  2004, vol.  42 (pg.  929- 936) Google Scholar CrossRef Search ADS PubMed  Laugesen S,  Messinese E,  Hem S,  Pichereaux C,  Grat S,  Ranjeva R,  Rossignol M,  Bono J.  Phosphoproteins analysis in plants: a proteomic approach,  Phytochemistry ,  2006, vol.  67 (pg.  2208- 2214) Google Scholar CrossRef Search ADS PubMed  Lexa M,  Genkov T,  Malbeck J,  Macháčková I,  Brzobohatý B.  Dynamics of endogenous cytokinin pools in tobacco seedlings: a modelling approach,  Annals of Botany ,  2003, vol.  91 (pg.  585- 597) Google Scholar CrossRef Search ADS PubMed  Lildballe DL,  Pedersen DS,  Kalamajka R,  Emmersen J,  Houben A,  Grasser KD.  The expression level of the chromatin-associated HMGB1 protein influences growth, stress tolerance, and transcriptome in Arabidopsis,  Journal of Molecular Biology ,  2008, vol.  384 (pg.  9- 21) Google Scholar CrossRef Search ADS PubMed  Lim CJ,  Yang KA,  Hong JK,  Choi JS,  Yun D,  Hong JC,  Chung WS,  Lee SY,  Cho MJ,  Lim CO.  Gene expression profiles during heat acclimation in Arabidopsis thaliana suspension-culture cells,  Journal of Plant Research ,  2006, vol.  119 (pg.  373- 383) Google Scholar CrossRef Search ADS PubMed  Liu Z,  Gao J,  Dong A,  Shen W.  A truncated Arabidopsis nucleosome assembly protein 1, atNAP1;3T, alters plant growth responses to abscisic acid and salt in the Atnap1;3-2 mutant,  Molecular Plant ,  2009, vol.  2 (pg.  688- 699) Google Scholar CrossRef Search ADS PubMed  Lochmanová G,  Zdráhal Z,  Konečná H,  Koukalová Š,  Malbeck J,  Souček P,  Válková M,  Kiran NS,  Brzobohatý B.  Cytokinin-induced photomorphogenesis in dark-grown Arabidopsis: a proteomic analysis,  Journal of Experimental Botany ,  2008, vol.  59 (pg.  3705- 3719) Google Scholar CrossRef Search ADS PubMed  Majumdar R,  Bandyopadhyay A,  Deng H,  Maitra U.  Phosphorylation of mammalian translation initiation factor 5 eIF5) in vitroand in vivo,  Nucleic Acids Research ,  2002, vol.  30 (pg.  1154- 1162) Google Scholar CrossRef Search ADS PubMed  Makrantoni V,  Antrobus R,  Botting CH,  Coote PJ.  Rapid enrichment and analysis of yeast phosphoproteins using affinity chromatography, 2D-PAGE and peptide mass fingerprinting,  Yeast ,  2005, vol.  22 (pg.  401- 414) Google Scholar CrossRef Search ADS PubMed  Meimoun P,  Ambard-Bretteville F,  Colas-des Francs-Small C,  Valot B,  Vidal J.  Analysis of plant phosphoproteins,  Analytical Biochemistry ,  2007, vol.  371 (pg.  238- 246) Google Scholar CrossRef Search ADS PubMed  Miller CO,  Skoog F,  von Saltza MH,  Strong F.  Kinetin, a cell division factor from deoxyribonucleic acid,  Journal of the American Chemical Society ,  1955, vol.  77 pg.  1392  Google Scholar CrossRef Search ADS   Mok D,  Mok M.  Cytokinin metabolism and action,  Annual Review of Plant Physiology and Plant Molecular Biology ,  2001, vol.  52 (pg.  89- 118) Google Scholar CrossRef Search ADS PubMed  Mok M,  Mok D,  Armstrong D.  Differential cytokinin structure–activity relationships in Phaseolus,  Plant Physiology ,  1978, vol.  61 (pg.  72- 75) Google Scholar CrossRef Search ADS PubMed  Murtazina DA,  Petukhov SP,  Rubtsov AM,  Storey KB,  Lopina OD.  Phosphorylation of the alpha-subunit of Na, K-ATPase from duck salt glands by cAMP-dependent protein kinase inhibits the enzyme activity,  Biochemistry ,  2001, vol.  66 (pg.  865- 874) Google Scholar PubMed  Naranda T,  Ballesta JP.  Phosphorylation controls binding of acidic proteins to the ribosome,  Proceedings of the National Academy of Sciences, USA ,  1991, vol.  88 (pg.  10563- 10567) Google Scholar CrossRef Search ADS   Nishimura C,  Ohashi Y,  Sato S,  Kato T,  Tabata S,  Ueguchi C.  Histidine kinase homologs that act as cytokinin receptors possess overlapping functions in the regulation of shoot and root growth in Arabidopsis,  The Plant Cell ,  2004, vol.  16 (pg.  1365- 1377) Google Scholar CrossRef Search ADS PubMed  Nomura H,  Komori T,  Kobori M,  Nakahira Y,  Shiina T.  Evidence for chloroplast control of external Ca2+-induced cytosolic Ca2+transients and stomatal closure,  The Plant Journal ,  2008, vol.  53 (pg.  988- 998) Google Scholar CrossRef Search ADS PubMed  Nylander M,  Svensson J,  Palva ET,  Welin BV.  Stress-induced accumulation and tissue-specific localization of dehydrins in Arabidopsis thaliana,  Plant Molecular Biology ,  2001, vol.  45 (pg.  263- 279) Google Scholar CrossRef Search ADS PubMed  Penfield S.  Temperature perception and signal transduction in plants,  New Phytologist ,  2008, vol.  179 (pg.  615- 628) Google Scholar CrossRef Search ADS PubMed  Peterman TK,  Ohol YM,  McReynolds LJ,  Luna EJ.  Patellin1, a novel sec14-like protein, localizes to the cell plate and binds phosphoinositides,  Plant Physiology ,  2004, vol.  136 (pg.  3080- 3094)  discussion 3001–3002 Google Scholar CrossRef Search ADS PubMed  Pflum MK,  Tong JK,  Lane WS,  Schreiber SL.  Histone deacetylase 1 phosphorylation promotes enzymatic activity and complex formation,  Journal of Biological Chemistry ,  2001, vol.  276 (pg.  47733- 47741) Google Scholar CrossRef Search ADS PubMed  Picard D.  Heat-shock protein 90, a chaperone for folding and regulation,  Cellular and Molecular Life Sciences ,  2002, vol.  59 (pg.  1640- 1648) Google Scholar CrossRef Search ADS PubMed  Rashotte AM,  Carson SDB,  To JPC,  Kieber JJ.  Expression profiling of cytokinin action in Arabidopsis,  Plant Physiology ,  2003, vol.  132 (pg.  1998- 2011) Google Scholar CrossRef Search ADS PubMed  Riefler M,  Novak O,  Strnad M,  Schmülling T.  Arabidopsis cytokinin receptor mutants reveal functions in shoot growth, leaf senescence, seed size, germination, root development, and cytokinin metabolism,  The Plant Cell ,  2006, vol.  18 (pg.  40- 54) Google Scholar CrossRef Search ADS PubMed  Romanov GA,  Lomin SN,  Schmülling T.  Biochemical characteristics and ligand-binding properties of Arabidopsis cytokinin receptor AHK3 compared to CRE1/AHK4 as revealed by a direct binding assay,  Journal of Experimental Botany ,  2006, vol.  57 (pg.  4051- 4058) Google Scholar CrossRef Search ADS PubMed  Sanders D,  Pelloux J,  Brownlee C,  Harper JF.  Calcium at the crossroads of signaling,  The Plant Cell ,  2002, vol.  14  Supplement(pg.  S401- S417) Google Scholar PubMed  Saunders MJ,  Helper PK.  Calcium antagonists and calmodulin inhibitors block cytokinin-induced bud formation in Funaria,  Developmental Biology ,  1983, vol.  99 (pg.  41- 49) Google Scholar CrossRef Search ADS PubMed  Sasaki K,  Kim M,  Imai R.  Arabidopsis cold shock domain protein2 is a RNA chaperone that is regulated by cold and developmental signals,  Biochemical and Biophysical Research Communications ,  2007, vol.  364 (pg.  633- 638) Google Scholar CrossRef Search ADS PubMed  Souček P,  Klíma P,  Reková A,  Brzobohatý B.  Involvement of hormones and KNOXI genes in early Arabidopsis seedling development,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  3797- 3810) Google Scholar CrossRef Search ADS PubMed  Spíchal L,  Rakova NY,  Riefler M,  Mizuno T,  Romanov GA,  Strnad M,  Schmülling T.  Two cytokinin receptors of Arabidopsis thaliana, CRE1/AHK4 and AHK3, differ in their ligand specificity in a bacterial assay,  Plant and Cell Physiology ,  2004, vol.  45 (pg.  1299- 1305) Google Scholar CrossRef Search ADS PubMed  Strnad M.  The aromatic cytokinins,  Physiologia Plantarum ,  1997, vol.  101 (pg.  674- 688) Google Scholar CrossRef Search ADS   Sujatha M,  Reddy TP.  Differential cytokinin effects on the stimulation of in vitro shoot proliferation from meristematic explants of castor (Ricinus communis L.),  Plant Cell Reports ,  1998, vol.  17 (pg.  561- 566) Google Scholar CrossRef Search ADS   Sung DY,  Vierling E,  Guy CL.  Comprehensive expression profile analysis of the Arabidopsis Hsp70 gene family,  Plant Physiology ,  2001, vol.  126 (pg.  789- 800) Google Scholar CrossRef Search ADS PubMed  Suzuki I,  Los DA,  Kanesaki Y,  Mikami K,  Murata N.  The pathway for perception and transduction of low-temperature signals in Synechocystis,  The EMBO Journal ,  2000, vol.  19 (pg.  1327- 1334) Google Scholar CrossRef Search ADS PubMed  Takei K,  Ueda N,  Aoki K,  Kuromori T,  Hirayama T,  Shinozaki K,  Yamaya T,  Sakakibara H.  AtIPT3 is a key determinant of nitrate-dependent cytokinin biosynthesis in Arabidopsis,  Plant and Cell Physiology ,  2004, vol.  45 (pg.  1053- 1062) Google Scholar CrossRef Search ADS PubMed  Umeda M,  Manabe Y,  Uchimiya H.  Phosphorylation of the C2 subunit of the proteasome in rice (Oryza sativa L.),  FEBS Letters ,  1997, vol.  403 (pg.  313- 317) Google Scholar CrossRef Search ADS PubMed  Weinl S,  Held K,  Schlücking K,  Steinhorst L,  Kuhlgert S,  Hippler M,  Kudla J.  A plastid protein crucial for Ca2+-regulated stomatal responses,  New Phytologist ,  2008, vol.  179 (pg.  675- 686) Google Scholar CrossRef Search ADS PubMed  Yang L,  Lin C,  Liu ZR.  Phosphorylations of DEAD box p68 RNA helicase are associated with cancer development and cell proliferation,  Molecular Cancer Research ,  2005, vol.  3 (pg.  355- 363) Google Scholar CrossRef Search ADS PubMed  Yonekura-Sakakibara K,  Kojima M,  Yamaya T,  Sakakibara H.  Molecular characterization of cytokinin-responsive histidine kinases in maize. Differential ligand preferences and response to cis-zeatin,  Plant Physiology ,  2004, vol.  134 (pg.  1654- 1661) Google Scholar CrossRef Search ADS PubMed  Youn JH,  Shin J- S.  Nucleocytoplasmic shuttling of HMGB1 is regulated by phosphorylation that redirects it toward secretion,  Journal of Immunology ,  2006, vol.  177 (pg.  7889- 7897) Google Scholar CrossRef Search ADS   Zhang Y,  Wang Y,  Kanyuka K,  Parry MAJ,  Powers SJ,  Halford NG.  GCN2-dependent phosphorylation of eukaryotic translation initiation factor-2alpha in Arabidopsis,  Journal of Experimental Botany ,  2008, vol.  59 (pg.  3131- 3141) Google Scholar CrossRef Search ADS PubMed  © 2010 The Author(s). This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.5), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)

Balzergue, Coline;Puech-Pagès, Virginie;Bécard, Guillaume;Rochange, Soizic F.

doi: 10.1093/jxb/erq335pmid: 21045005

Abstract Most plants form root symbioses with arbuscular mycorrhizal (AM) fungi, which provide them with phosphate and other nutrients. High soil phosphate levels are known to affect AM symbiosis negatively, but the underlying mechanisms are not understood. This report describes experimental conditions which triggered a novel mycorrhizal phenotype under high phosphate supply: the interaction between pea and two different AM fungi was almost completely abolished at a very early stage, prior to the formation of hyphopodia. As demonstrated by split-root experiments, down-regulation of AM symbiosis occurred at least partly in response to plant-derived signals. Early signalling events were examined with a focus on strigolactones, compounds which stimulate pre-symbiotic fungal growth and metabolism. Strigolactones were also recently identified as novel plant hormones contributing to the control of shoot branching. Root exudates of plants grown under high phosphate lost their ability to stimulate AM fungi and lacked strigolactones. In addition, a systemic down-regulation of strigolactone release by high phosphate supply was demonstrated using split-root systems. Nevertheless, supplementation with exogenous strigolactones failed to restore root colonization under high phosphate. This observation does not exclude a contribution of strigolactones to the regulation of AM symbiosis by phosphate, but indicates that they are not the only factor involved. Together, the results suggest the existence of additional early signals that may control the differentiation of hyphopodia. Phosphorus, arbuscular mycorrhiza, strigolactone, symbiosis, hyphopodium Introduction Roots of the vast majority of plant species develop symbiotic associations with arbuscular mycorrhizal (AM) soil fungi. Fungal hyphae develop in the root cortex where they form intracellular highly branched structures called arbuscules, and simultaneously in the soil where they form a dense mycelial network. Within the root the plant supplies the fungus with hexoses, at a cost of up to 20% of the carbon fixed by photosynthesis (Smith and Read, 2008). In return, it obtains water and minerals taken up from soil by the mycelial network. The main benefit of the symbiosis for the plant is an enhanced acquisition of phosphorus (P), a frequent limiting factor in plant growth due to its poor solubility and mobility in soils. Despite the importance of AM symbiosis, cellular and molecular events underlying this interaction are only beginning to be unravelled (Parniske, 2008). Direct genetic screens to identify mycorrhizal (myc−) mutants are extremely cumbersome. As a result, most myc− mutants in fact belong to a subset of mutants initially isolated as deficient in nitrogen-fixing symbiosis, this latter interaction being easier to examine. A consequence of this bias is the relative scarcity of mutants affected in events unique to the AM symbiosis, including pre-colonization signalling and arbuscule development and function (Marsh and Schultze, 2001). Nonetheless, several specific myc− mutants have been identified in the past few years. They can be affected in different stages of the interaction as summarized in Pumplin et al. (2009): pre-symbiotic fungal growth, formation of hyphopodia (root attachment and penetration structures, formerly referred to as appressoria), epidermal penetration, and arbuscule development (see also Zhang et al., 2010). Various physiological situations are known to affect the development of AM symbiosis. For instance, plants control the extent to which AM fungi can colonize their roots according to their own nutritional requirements. The best known example of such regulations is the control of AM symbiosis according to P availability. Roots can acquire P as inorganic orthophosphate (Pi) through different pathways (Bucher, 2007). In certain conditions the mycorrhizal uptake pathway, which involves specific Pi transporters (Rausch et al., 2001; Harrison et al., 2002; Paszkowski et al., 2002), can be the major route for P uptake (Smith et al., 2003). When P is abundant, a direct, probably less costly uptake pathway is preferred (Nagy et al., 2008), and a reduced root colonization by AM fungi is observed. This down-regulation of the symbiosis by P has been known for a long time (Graham et al., 1981; Thomson et al., 1986; Elias and Safir, 1987; Rausch et al., 2001; and many others). It seems to be a general phenomenon, although its magnitude can vary (Javot et al. 2007; Smith and Read, 2008). It has far-reaching consequences in natural ecosystems where it modulates the effect of AM fungi on plant species diversity (Collins and Foster, 2009), as well as in agriculture where strong P fertilization may in the long term decrease the presence and richness of soil AM communities (Johnson, 1993). Little is known about mechanisms underlying the regulation of AM symbiosis by P. A recent study (Branscheid et al., 2010) has documented this down-regulation in Medicago truncatula, and investigated the identity of the internal signal that triggers suppression of the interaction under high P. Nonetheless, the downstream mechanisms that prevent or limit root colonization by AM fungi remain largely unknown. Early studies led to conflicting results and interpretations, partly due to the variety of species combinations and experimental systems. Some of these early studies interpreted the impact of high P on the fungus in terms of trophic effects: high P would decrease the root secretion of metabolites used by the fungus, such as amino acids or carbohydrates (e.g. Graham et al., 1981; Thomson et al., 1986). An alternative proposition was that qualitative rather than quantitative differences between root exudates of P-replete and P-deficient plants could account for their differential effects on the fungus (Elias and Safir, 1987). This led to the suggestion that P-deprived roots exuded important flavonoid signals that triggered pre-symbiotic fungal growth and activity (Nair et al., 1991). Advances made in the last 10 years have indeed emphasized the importance of signalling events in mycorrhizal interactions, and the recent identification of some signals may shed new light on the regulation of AM symbiosis by P. Plants and AM fungi are known to exchange molecular signals prior to physical contact, at the so-called pre-symbiotic stage. Various lines of evidence indicate that AM fungi produce diffusible compounds able to modulate root gene expression (Kosuta et al., 2003; Weidmann et al., 2004), intracellular signalling (Navazio et al., 2007; Kosuta et al., 2008), development (Olah et al., 2005), and metabolism (Gutjahr et al., 2009). Reciprocally, plant roots secrete compounds that stimulate the fungus (Gianinazzi-Pearson et al., 1989; Siqueira et al., 1991; Tsai and Phillips, 1991; Giovannetti et al., 1996; Buée et al., 2000). A group of secondary metabolites called strigolactones were identified as major contributors to this effect (Akiyama et al., 2005; Besserer et al., 2006). Strigolactones trigger morphological and developmental responses in the fungus such as hyphal branching and spore germination, and enhance fungal mitochondrial activity and respiration (Besserer et al., 2006, 2008). Strigolactone-mediated signalling is necessary for a normal level of root colonization, as demonstrated using strigolactone-deficient mutants (Gomez-Roldan et al., 2008). Most interestingly, these root-exuded compounds also play an important role in planta, acting as hormones that contribute to the regulation of shoot branching (Gomez-Roldan et al., 2008; Umehara et al., 2008). Prior to the discovery of their roles in AM symbiosis and plant development, strigolactones were known as germination stimulants for the seeds of the parasitic plants Striga and Orobanche (Bouwmeester et al., 2007). Damage caused to crops by these weeds is lower under strong nutrient fertilization, which led to the investigation of whether P availability influenced strigolactone release into the soil. Indeed, several studies demonstrated a strong negative effect of high P supply on strigolactone production and exudation in various species (Yoneyama et al., 2007a, b; Lopez-Raez et al., 2008). A reasonable hypothesis is that high P availability would decrease the extent of AM symbiosis by reducing strigolactone production in roots (Bouwmeester et al., 2007; Yoneyama et al., 2007b). In this report, P fertilization conditions which lead to an almost complete arrest of the first stages of the interaction between pea (Pisum sativum L.) and two species of AM fungi are described. This strong effect is at least partly linked to regulatory events occurring in the plant partner, as shown by split-root experiments. Furthermore, it is demonstrated that like root colonization, strigolactone production is controlled in a systemic manner by P supply. Hence, strigolactones may contribute to the regulation of AM symbiosis by P, but supplementation experiments indicate that they are not the only factor involved. Materials and methods Plant and fungal materials Seeds of garden pea (Pisum sativum L., cv Terese) were surface sterilized with 3.2% sodium hypochlorite for 10 min and 95% ethanol for 2 min, and washed four times with sterile distilled water. Seeds were germinated on agar–water [0.8% (w/v)] solid medium for 4 d at 25 °C. Gigaspora rosea spores (DAOM 194757) were produced and surface sterilized as described in Besserer et al. (2006). Sterile Glomus intraradices spores (DAOM 197198) were purchased from Premier Tech Ltée (Rivière du loup, Québec, Canada) or produced according to St-Arnaud et al. (1996). Growth and inoculation of plants Plants were grown in a growth chamber under a 16 h photoperiod (22 °C day, 20 °C night), in pots containing sterilized charred clay (Oil-Dri, Klasmann, France) as substrate. They were fertilized daily with half-strength Long Ashton Nutrient Solution (LANS; Hewitt, 1966) containing a final concentration of 7.5 μM (low P; LP), 75 μM (medium P; MP), or 750 μM (high P; HP) sodium dihydrogen phosphate (NaH2PO4). Phosphate was supplied as KH2PO4 instead of NaH2PO4 in two experiments. For the determination of mycorrhizal ability, germinated seedlings were transferred to 250 ml pots. They were inoculated with 150 or 600 spores of Gl. intraradices, or 100 spores of Gi. rosea. Two-thirds of the spores were mixed with the substrate and one-third was added close to the seedling. The percentage of root length colonized by the fungus (i.e. showing arbuscules, vesicles, or both) was determined by the gridline intersection method (Giovannetti and Mosse, 1980) using a dissecting microscope after staining with Schaeffer black ink (Vierheilig et al., 1998). For the quantification of early symbiotic structures, roots were handled very carefully during rinses and staining to avoid tearing off hyphae from the roots. For each plant, 60 randomly picked 1 cm long root fragments mounted on glass slides were examined under a microscope at ×40 magnification, and hyphopodia were counted. For split-root experiments, sterile seedlings were grown on solid medium (nutrient solution solidified with 0.4% phytagel) until radicles were ∼2 cm long. The root apex was cut off and the primary root was divided lengthwise into two equal parts. Seedlings were kept on solid medium for another week during which lateral roots developed on both sides of the split root. They were then transferred to two-compartment pot systems with 150 ml of substrate per compartment. Each compartment was inoculated with 90 spores of Gl. intraradices mixed with the substrate. Following plant harvest, root fresh weight was determined and plants for which one side of the root system was >2-fold heavier than the other side were excluded from the analysis. Preparation of root exudate extracts Three-week-old non-inoculated plants fertilized with LP or HP nutrient solution were removed from the substrate. Their roots were rinsed and immersed in the same nutrient solution for 24 h. Exudates were extracted with 1 vol. of ethyl acetate, then the organic phase was treated with 1 vol. of 0.2 M K2HPO4. Residual water was removed with anhydrous MgSO4, and ethyl acetate extracts were filtered and dried under vacuum. Exudate extracts were resuspended in the appropriate solvents for branching bioassays or mass spectrometry analysis, and their concentration was adjusted on a root dry weight basis. Determination of P contents Inorganic phosphate (Pi) content was measured using the colorimetric method based on molybdenum blue described in Nanamori et al. (2004). The only modification was that plant tissues were ground in perchloric acid using a FASTPREP® system (MP Biomedicals) with lysing matrix A. Gigaspora rosea hyphal branching bioassay Spores of Gi. rosea were germinated on solid M medium (Bécard and Fortin, 1988) as described in Besserer et al. (2006). Root exudate extracts produced by the equivalent of 120 mg of root dry weight were resuspended in 1 ml of 10% (v:v) acetonitrile. Samples of 5 μl were applied on both sides of the main hypha of a 6-day-old germinated spore. Newly formed apices were counted 48 h after treatment. The experiment included spores treated with 10% acetonitrile (negative control) and with 100 nM GR24 in 10% acetonitrile (positive control). For the experiment described in Supplementary Fig. S1 available at JXB online, spores were germinated on LP or HP half-strength LANS solidified with 0.7% high gel strength agar, and treated either with 100 nM GR24 or with the solvent only. Glomus intraradices germination assays The assays were carried out in 25-compartment plates. Four compartments were used for each treatment. In each compartment, 1 ml of a sterile suspension of Gl. intraradices spores at 30 spores ml−1 was added to 1 ml of sterile test solution. The test solutions corresponded to full-strength LP or HP LANS, so the final concentrations of nutrients were equivalent to those of the watering solutions. The nutrient solutions alone were tested, as well as the same solutions containing root exudate extracts (1 ml of test solution then contained root exudates produced by the equivalent of 1 mg of root dry weight). Plates were incubated at 30 °C under 2% CO2 in the dark. Spore germination rates were determined 5 d after treatment. Chromatography and mass spectrometry analyses Root exudate extracts were dissolved in acetonitrile:water [1:1 (v/v)]. Strigolactone detection was performed using a 4000 Q Trap mass spectrometer with a Turbo V ESI source in the positive mode, coupled to an Agilent 1100 series HPLC system as described in Gomez-Roldan et al. (2008), except for the following modifications. HPLC separation was performed using a C18 column (5 μm, 2.1×250 mm, ACCLAIM 120C18, Dionex). Solutions of formic acid:water [1:103 (v/v); A] and formic acid:acetonitrile [1:103 (v/v); B] were pumped at 0.2 ml min−1. The gradient was: 50% B for 5 min, 50–70% B in 5 min, 70% B for 10 min, 70–100% B in 10 min, and 100% B for 5 min. The reported peak intensities correspond to extracts obtained with the equivalent of 75 mg of root dry weight. GR24 was added as an external standard to all samples at a final concentration of 100 nM. The two major pea strigolactones and GR24 were detected in the MRM mode by monitoring the transitions 405>97 m/z and 405>345 m/z for fabacyl acetate, 389>233 m/z and 411>254 m/z for orobanchyl acetate, and 299>202 m/z for GR24. Statistical analyses Results were analysed by analysis of variance (ANOVA) followed by Tukey's HSD test or Student's t-test using Statgraphics Centurion software (Sigma Plus). Data in Figs 2, 3D, and 5 were subjected to logarithmic, cosine, and arc sine transformation, respectively, prior to analysis. Data in Supplementary Fig. S1 at JXB online were analysed by a Kruskal–Wallis test. Fig. 1. View largeDownload slide Effect of phosphate fertilization on mycorrhizal root colonization. Plants inoculated with 600 spores of Gl. intraradices (grey bars) or 100 spores of Gi. rosea (white bars) were grown under low (LP) or high (HP) phosphate fertilization. The extent of root colonization was determined after observation of stained root samples as the fraction of root length showing arbuscules, vesicles, or both in the case of Gl. intraradices, and arbuscules in the case of Gi. rosea. Error bars show the SEM; n=5–6 plants when inoculated with Gl. intraradices and n=7–8 plants when inoculated with Gi. rosea. Different letters indicate statistically significant differences according to Student's t-test (P <0.05). Fig. 1. View largeDownload slide Effect of phosphate fertilization on mycorrhizal root colonization. Plants inoculated with 600 spores of Gl. intraradices (grey bars) or 100 spores of Gi. rosea (white bars) were grown under low (LP) or high (HP) phosphate fertilization. The extent of root colonization was determined after observation of stained root samples as the fraction of root length showing arbuscules, vesicles, or both in the case of Gl. intraradices, and arbuscules in the case of Gi. rosea. Error bars show the SEM; n=5–6 plants when inoculated with Gl. intraradices and n=7–8 plants when inoculated with Gi. rosea. Different letters indicate statistically significant differences according to Student's t-test (P <0.05). Fig. 2. View largeDownload slide Gigaspora rosea hyphal branching in response to GR24, LP or HP root exudates. Germinated spores of Gi. rosea were treated with GR24 and/or root exudates of low (LP) or high (HP) phosphate-grown plants, or with the solvent alone as negative control (10% acetonitrile; AcN). Newly formed hyphal apices were counted 48 h after treatment. White bars, controls; grey bars, root exudates alone; black bars, root exudates+GR24. Error bars show the SEM; n=24–26 treated spores for each condition. Different letters indicate statistically significant differences according to one-way ANOVA followed by Tukey's test (P <0.05). Fig. 2. View largeDownload slide Gigaspora rosea hyphal branching in response to GR24, LP or HP root exudates. Germinated spores of Gi. rosea were treated with GR24 and/or root exudates of low (LP) or high (HP) phosphate-grown plants, or with the solvent alone as negative control (10% acetonitrile; AcN). Newly formed hyphal apices were counted 48 h after treatment. White bars, controls; grey bars, root exudates alone; black bars, root exudates+GR24. Error bars show the SEM; n=24–26 treated spores for each condition. Different letters indicate statistically significant differences according to one-way ANOVA followed by Tukey's test (P <0.05). Fig. 3. View largeDownload slide Mycorrhizal root colonization in split-root systems. (A) Experimental design. Each root system was divided into two parts placed in different pots to allow differential phosphate fertilization. Both sides were inoculated with 90 spores of Gl. intraradices and plants were grown for 6 weeks. Control plants were fertilized with the same solution on both sides. Results in B, C, and D correspond to the same plants. (B) Root colonization levels determined by observation of stained root samples. For control plants, colonization levels measured on both sides were averaged. (C, D) Inorganic orthophosphate (Pi) content in leaves (C) and roots (D). Error bars show the SEM; n=5–7 plants for each condition. Different letters indicate statistically significant differences according to one-way ANOVA followed by Tukey's test (P <0.05). Fig. 3. View largeDownload slide Mycorrhizal root colonization in split-root systems. (A) Experimental design. Each root system was divided into two parts placed in different pots to allow differential phosphate fertilization. Both sides were inoculated with 90 spores of Gl. intraradices and plants were grown for 6 weeks. Control plants were fertilized with the same solution on both sides. Results in B, C, and D correspond to the same plants. (B) Root colonization levels determined by observation of stained root samples. For control plants, colonization levels measured on both sides were averaged. (C, D) Inorganic orthophosphate (Pi) content in leaves (C) and roots (D). Error bars show the SEM; n=5–7 plants for each condition. Different letters indicate statistically significant differences according to one-way ANOVA followed by Tukey's test (P <0.05). Fig. 4. View largeDownload slide Systemic control of strigolactone production. Split-root plants were fertilized with low phosphorus (LP) on one side and high phosphorus (HP) on the other (LP/HP plants), or with the same solution on both sides (LP/LP and HP/HP plants). Root exudate extracts were analysed by LC-MS/MS in the MRM mode. The synthetic strigolactone analogue GR24 was added in equal quantity to all samples as an external standard. Chromatograms show the most abundant mass transition for each of the two major pea strigolactones, fabacyl acetate and orobanchyl acetate. Insets show the mass transition corresponding to the external standard GR24 (299>202 m/z). Fig. 4. View largeDownload slide Systemic control of strigolactone production. Split-root plants were fertilized with low phosphorus (LP) on one side and high phosphorus (HP) on the other (LP/HP plants), or with the same solution on both sides (LP/LP and HP/HP plants). Root exudate extracts were analysed by LC-MS/MS in the MRM mode. The synthetic strigolactone analogue GR24 was added in equal quantity to all samples as an external standard. Chromatograms show the most abundant mass transition for each of the two major pea strigolactones, fabacyl acetate and orobanchyl acetate. Insets show the mass transition corresponding to the external standard GR24 (299>202 m/z). Fig. 5. View largeDownload slide Effect of strigolactone supplementation on root colonization. Plants inoculated with 150 spores of Gl. intraradices were fertilized daily with LP or HP nutrient solution, supplemented or not with 10 nM GR24. The extent of root colonization was determined by observation of stained root samples. Error bars show the SEM; n=7–8 plants for each condition. Different letters indicate statistically significant differences according to one-way ANOVA followed by Tukey's test (P <0.05). Fig. 5. View largeDownload slide Effect of strigolactone supplementation on root colonization. Plants inoculated with 150 spores of Gl. intraradices were fertilized daily with LP or HP nutrient solution, supplemented or not with 10 nM GR24. The extent of root colonization was determined by observation of stained root samples. Error bars show the SEM; n=7–8 plants for each condition. Different letters indicate statistically significant differences according to one-way ANOVA followed by Tukey's test (P <0.05). Results Experimental system Pea plants in interaction with Gl. intraradices have been used previously to determine the importance of strigolactones in AM symbiosis (Gomez-Roldan et al., 2008). For the present study the main advantage of pea was that the strigolactones produced by this species are characterized (Yoneyama et al., 2008; Xie et al., 2009) and readily detectable. To evaluate the effect of P supply on AM interactions, plants were fertilized with half-strength LANS (Hewitt, 1966) containing different concentrations of P. The HP solution contained the normal concentration of phosphate of half-strength LANS—that is, 750 μM P. MP and LP corresponded to 10- and 100-fold lower phosphate concentrations, respectively. Plant growth evaluated by fresh weight was the lowest under LP, intermediate under MP, and maximal under HP (Supplementary Table S1 at JXB online). Therefore, HP fertilization provides sufficient but not excess P, while LP corresponds to P starvation conditions. Pea plants exhibited a remarkably strong mycorrhizal response to P fertilization (Fig. 1): when inoculated with Gl. intraradices, plants grown under LP exhibited colonization levels of ∼60%, while hardly any fungal structures could be observed in roots under HP (<1% root length colonized). The effect of HP was also tested in the interaction between P. sativum and Gi. rosea, a fungus phylogenetically distant from Gl. intraradices. Root colonization levels were lower with Gi. rosea than with Gl. intraradices (Fig. 1). Similar observations have been reported in another legume, Medicago sativa (Douds et al. 1998) and may reflect host preferences in AM interactions. Nonetheless, high P exerted a similarly strong negative effect on root colonization with both fungal species, suggesting that the regulation mechanisms involved are not fungus specific. Mycorrhizal phenotype of plants grown under HP Plants seem to possess multiple checkpoints for mycorrhizal invasion, and the interaction can be stopped at distinct stages in various mutant backgrounds (Pumplin et al., 2009). To determine at what stage the interaction was arrested under HP, inoculated roots were subjected to closer microscopic examination. This allows the observation of all fungal structures including hyphopodia, which are visible as flattened, lenticular hyphal tips attached to the root epidermal surface (Garriock et al., 1989). Like the frequency of arbuscules and vesicles reported in Fig. 1, the number of hyphopodia per unit of root length was markedly reduced under HP as compared with LP, in plants inoculated either with Gl. intraradices or with Gi. rosea (Table 1). Similar observations were made when P was supplied as NaH2PO4 or KH2PO4, indicating that the observed effect was not due to the phosphate counterion. It has to be noted that when present, hyphopodia, arbuscules, and vesicles in HP-grown roots could not be distinguished morphologically from those observed in LP-grown roots (results not shown). In an experiment with Gl. intraradices, roots were also observed at an earlier time point, 4 weeks post-inoculation (wpi). Again, the frequency of hyphopodia was much lower under HP than under LP. Under the present conditions, this time point corresponds to the very first stages of root colonization. (At 3 wpi no fungal structures can be observed on roots, and at 4 wpi the root colonization level is <5%; data not shown. This slow progression of AM symbiosis establishment is most probably related to the inoculation with spores rather than with more infectious sources of inoculum.) Together, these results indicate that under HP the interaction was arrested prior to the formation of hyphopodia on the root epidermis. Table 1. Frequency of hyphopodia on inoculated roots   Weeks post-inoculation    No. of hyphopodia m−1 root  Gl. intraradices KH2PO4  6  LP  112±42.4 a      HP  6.67±1.67 b  Gl. intraradices NaH2PO4  4  LP  38.8±5.42 a      HP  1.33±0.83 b    6  LP  42.9±2.67 a      HP  0.42±0.42 b  Gi. rosea KH2PO4  7  LP  61.7±10.63 a      HP  10.2±3.94 b    Weeks post-inoculation    No. of hyphopodia m−1 root  Gl. intraradices KH2PO4  6  LP  112±42.4 a      HP  6.67±1.67 b  Gl. intraradices NaH2PO4  4  LP  38.8±5.42 a      HP  1.33±0.83 b    6  LP  42.9±2.67 a      HP  0.42±0.42 b  Gi. rosea KH2PO4  7  LP  61.7±10.63 a      HP  10.2±3.94 b  Plants inoculated with 150 spores of Gl. intraradices or 100 spores of Gi. rosea were grown under LP or HP with KH2PO4 or NaH2PO4 as phosphorus source. Data correspond to three independent experiments: one with Gl. intraradices and KH2PO4, one with Gl. intraradices and NaH2PO4, and one with Gi. rosea and KH2PO4. Root samples were examined microscopically 4, 6, or 7 weeks post-inoculation for the presence of hyphopodia. Values indicate the average number of hyphopodia per metre of root length, ±SEM. n=3–4 plants when inoculated with Gl. intraradices and n=7–8 plants when inoculated with Gi. rosea (60 cm of roots analysed per plant). Data were analysed separately for each experiment and time point. Different letters indicate statistically significant differences according to Student's t-test (P <0.05). View Large Effect of P supply on pre-symbiotic signalling The possible involvement of diffusible signals acting at the pre-symbiotic stage was considered. Such signals could include activators or inhibitors of early fungal development. Germination rates of Gl. intraradices spores exceeded 96% within 5 d in LP and HP nutrient solutions alone, and in these solutions supplemented with extracts of root exudates prepared from LP- or HP-grown plants (data not shown). This suggests the absence of a negative impact of HP conditions on this process. Hyphal branching activities of root exudates from HP- and LP-grown plants were then evaluated. This experiment was carried out on germinated spores of Gi. rosea, for which hyphal branching can more easily be observed. Root exudate extracts of LP-grown plants, as well as GR24, stimulated hyphal branching (Fig. 2). The combination of both treatments showed an additive effect. Root exudate extracts of HP-grown plants did not enhance hyphal branching relative to the control, which could be due either to the lack of stimulants or to the presence of inhibitors. The addition of GR24 to exudate extracts of HP-grown plants resulted in an activity similar to that of GR24 alone, suggesting that these extracts do not contain inhibitors of the strigolactone effect. Rather, they may lack important stimulants of hyphal branching. Systemic control of AM symbiosis by P The effect of HP fertilization on AM symbiosis reported above is unusually strong. This raises the possibility that, although HP conditions do not correspond to a very high phosphate concentration, they somehow disturb early fungal development (the reduced root colonization would then be a secondary effect of these perturbations). Indeed, AM fungi can sense and react to P availability (Requena et al., 2003). As shown above, HP does not prevent germination of Gl. intraradices spores. Moreover, whether HP could prevent fungal responsiveness to GR24 was tested with the branching bioassay. Gigaspora rosea spores grown in LP or HP conditions responded equally well to GR24 (Supplementary Fig. S1 at JXB online), suggesting that P fertilization should not affect fungal ability to respond to strigolactones. To investigate further whether the early arrest of AM symbiosis could be due to a post-germination direct effect of HP on the fungus, split-root experiments were carried out in which two halves of a root system were fertilized independently. In the experimental set-up described in Fig. 3A, all compartments were inoculated with spores of Gl. intraradices. Test plants (denoted LP/HP) were watered with LP on one side and HP on the other. Both sides of control plants (denoted LP/LP and HP/HP) were watered with the same solution. In these control plants, colonization rates were comparable with those observed in intact plants—that is, high in LP and very low in HP (<0.1% root length colonized; Fig 3B). The HP-watered roots of test LP/HP plants behaved like those of HP/HP plants, showing hardly any colonization. The most striking observation was that LP-watered roots of LP/HP plants were markedly less colonized than LP/LP plants (2% of root length versus 60%). In other words, root colonization on the LP side of LP/HP plants did not respond to local fertilization conditions but to the fertilization of a distant part of the plant. This systemic regulation indicates that the effects of HP are mediated by the plant in split LP/HP plants. This does not exclude the possibility that additional direct effects on the fungus contribute to the reduced root colonization in HP/HP or intact HP plants. Pi contents were determined in roots and leaves of the same inoculated split-root plants (Fig. 3C, D). The results are expressed in micromol Pi g−1 fresh weight to reflect the actual availability of Pi in the different tissues [NB: a high P supply resulted in increased root and leaf biomass in both LP/HP and HP/HP plants (results not shown), so the total amount of P taken up by these plants was higher than in LP/LP plants]. The results in Fig. 3C and D show that LP/HP plants accumulated Pi in leaves rather than in roots, indicating that in these conditions leaves acted as a stronger sink than roots. Roots on the LP side displayed Pi contents similar to those of LP/LP plants, yet their colonization rates were much lower (Fig. 3B). Therefore, down-regulation of AM symbiosis is not triggered by root Pi content. In contrast, leaves of LP/HP plants accumulated Pi at levels comparable with HP/HP plants. It can thus be hypothesized that the low root colonization levels observed in both types of plants may be related to high leaf Pi contents. Effect of P supply on strigolactone production in split-root systems Given that previous results point towards an effect of HP on early events in the AM interaction, that strigolactones are important pre-symbiotic signals (Gomez-Roldan et al., 2008), and that their production is regulated by P supply (Yoneyama et al., 2007a, b; Lopez-Raez et al., 2008), it is reasonable to envisage that these compounds mediate the effect of HP. If such is the case, one would expect strigolactone synthesis to be systemically regulated by P, like AM symbiosis. Root exudates obtained with split-root plants were analysed to address this question. The experimental set-up was similar to that described in Fig. 3, except that the plants were not inoculated. Figure 4 shows LC-MS/MS chromatograms obtained with extracts of root exudates. In the MRM detection mode used, each line corresponds to an MS/MS mass transition characteristic of one of the two major strigolactones produced by pea: fabacyl acetate (Xie et al., 2009) and orobanchyl acetate (Yoneyama et al., 2008). Two mass transitions were monitored for each strigolactone and gave a signal at the same retention time; for clarity only one transition is shown. Synthetic standards of these two strigolactones also eluted at the same retention times (data not shown), demonstrating that the monitored signals truly corresponded to strigolactones. GR24 was added in equal quantities to all samples as an external standard to visualize any possible artefacts due to sample loading or matrix effects. The signal obtained with GR24 was similar between samples, indicating that the amounts of other strigolactones in the samples could be appropriately compared. Chromatograms obtained with control LP/LP and HP/HP plants (Fig. 4) were similar to those obtained with intact plants grown under LP or HP (data not shown), and confirm the previously reported inhibitory effect of P supply on strigolactone production (Yoneyama et al., 2007a, b; Lopez-Raez et al., 2008). The analysis of root extracts rather than exudates led to similar observations (data not shown), indicating that the regulation occurs at the level of strigolactone biosynthesis rather than exudation. Furthermore, the HP side of LP/HP split-root plants barely produced detectable strigolactones, and therefore behaved as an intact HP root system. In contrast, the LP side of LP/HP plants produced much less strigolactone than control LP/LP plants (Fig. 4). This indicates that the HP side of these split-root plants negatively regulated strigolactone production on the LP side through systemic signalling. Supplementation of HP-grown plants with exogenous strigolactones The strong down-regulation of strigolactone synthesis by HP (Fig. 4) could account for the reduced root colonization (Fig. 1) and the absence of hyphal branching activity of root exudates (Fig. 2) observed under these conditions. To address this hypothesis, HP-grown plants were supplemented with exogenous strigolactones. The experimental conditions used (treatment with the synthetic strigolactone GR24, concentration and frequency of application) were previously demonstrated to be effective since they could rescue the mycorrhizal phenotype of strigolactone-deficient mutants (Gomez-Roldan et al., 2008). This strigolactone treatment was also sufficient to enhance mycorrhizal symbiosis establishment in LP-grown plants inoculated with Gl. intraradices (Fig. 5). Surprisingly, root colonization of HP-grown plants was not improved by strigolactone supplementation. Similar observations were made for HP-grown plants inoculated with Gi. rosea (data not shown). Strigolactone treatment did not stimulate the formation of hyphopodia on HP-grown roots either: 5.0±1.7 hyphopodia m−1 of root were observed in treated roots versus 6.7±1.7 in untreated roots in plants inoculated with Gl. intraradices, and 6.04±1.31 in treated roots versus 10.2±3.94 in untreated roots in plants inoculated with Gi. rosea. Discussion HP supply can strongly inhibit AM symbiosis The regulation of AM symbiosis by P supply has been observed repeatedly and is considered a general phenomenon. In contrast to most previous studies, however, the experimental conditions described in the present report lead to a clear-cut mycorrhizal phenotype under HP supply, since hardly any symbiotic structures are observed (Fig. 1). Similar effects of HP were observed with 150 and 600 Glomus spores per plant (Fig. 5 and Fig. 1, respectively), indicating that a higher inoculum density was not able to circumvent the regulatory mechanisms. The discrepancy between the strong mycorrhizal phenotype reported here and the more moderate effects of P reported previously may relate to the plant species used, and/or to the experimental conditions: in this study, P was supplied daily in the nutrient solution, and plants were inoculated with spores rather than fragments of infected roots containing different kinds of propagules. Inoculation with spores, often regarded as less virulent, probably helps to reveal moderate phenotypes that could be masked with stronger sources of inoculum. For example, the pmi1 mutant of tomato exhibits a severe phenotype when inoculated with spores, but is colonized normally when inoculated with mycorrhizal nurse plants (David-Schwartz et al., 2001). The effect of HP supply on AM symbiosis is partly mediated by the plant Among the conditions tested, plant growth was maximal under HP conditions (Supplementary Table S1 at JXB online), which correspond to a moderate P supply (750 μM): in studies on P starvation responses, the P-replete condition usually falls in the 1–3 mM range [e.g. Bonser et al. (1996) on pea; Valdes-Lopez et al. (2008) on bean; Pant et al. (2008) on Arabidopsis]. The HP nutrient solution does not exhibit toxicity towards the fungal partner, as evaluated by spore germination tests. It also does not seem to modify the ability of the fungus to respond to strigolactones (Supplementary Fig. S1). In addition, in split-root experiments the inhibition of root colonization can be observed in a compartment where the fungus is only exposed to LP (Fig. 3). This regulation of AM symbiosis through systemic signalling is consistent with previous reports (e.g. Thomson et al., 1991; Rausch et al., 2001). It shows that the very strong inhibition of root colonization triggered by HP in the present report involves plant-driven processes and is not only due to a direct effect of local P concentration on the fungus. Yet, the existence of such direct effects cannot be excluded. HP supply arrests AM symbiosis in its first stages Microscopic examination of root samples revealed that HP fertilization reduced the number of hyphopodia formed on the root epidermis. This represents a novel HP-related mycorrhizal phenotype. A straightforward interpretation is that HP prevents hyphopodium formation per se. Alternatively, one could imagine that defects in later symbiotic stages could also lead to a reduced number of hyphopodia; for example, an impaired progression of the fungus within roots could delay or reduce the number of secondary infection events, which would in turn result in a smaller number of attached external hyphae. Several arguments lead us to conclude that the block in AM symbiosis triggered by HP occurs prior to primary hyphopodium formation, rather than later in the symbiotic process. First, the steps of clearing and staining the roots prior to microscopic observation were performed with particular care to prevent possible stripping and loss of hyphopodia (particularly those that did not lead to root colonization). Second, similar observations were made at 4 and 6 wpi (Table 1). The first time point (4 wpi) corresponds to the very beginning of the infection process, when the first arbuscules become visible (<5% root length colonized). Hyphopodia observed at this time point therefore most probably derived from primary hyphae of germinated spores, rather than from secondary infections. At this time point, a very strong effect of HP fertilization was already noted. Third, mutants affected in later stages of the interaction typically exhibit a normal (sometimes even higher) number of hyphopodia (Bradbury et al., 1991; Bonfante et al., 2000). Therefore, the present observations strongly suggest that HP conditions prevent either pre-symbiotic fungal development or attachment to roots. In contrast to the formation of appressoria by pathogenic fungi, the differentiation of these attachment and penetration structures by AM fungi is still poorly understood. Plants grown under HP are reminiscent of tomato pmi1 and pmi2 (David-Schwartz et al., 2001, 2003), and maize nope1 and taci1 mutants (Paszkowski et al., 2006), in which a reduced frequency of hyphopodia was observed. Unfortunately the genes affected by these mutations have not been identified yet. Nonetheless these mutants, together with the HP conditions described in the present report, should be useful to decipher the mechanisms involved in hyphopodium differentiation. Different kinds of mechanisms could regulate the formation of hyphopodia under HP. One of them is the production by plant roots of stimulatory or inhibitory diffusible compounds. Candidate compounds include flavonoids, some of which have been reported to stimulate AM root colonization by enhancing the number of fungal entry points (Scervino et al., 2007). Polyamines have also been proposed to favour the formation of hyphopodia (El Ghachtouli et al., 1995). P availability also affects the production of compounds known to affect fungal development more generally (reviewed in Vierheilig, 2004), but in most instances their contribution to the regulation of AM symbiosis by P has not been tested functionally. An exception is the report by Akiyama et al. (2002) that a C-glycosylflavonoid accumulated in melon roots upon P starvation, and that supplementation with this compound restored normal mycorrhizal rates under HP. The reduced accumulation of this compound may therefore account for the decreased root colonization under HP. In contrast to the present report, however, the effects of HP were not observed in the first visible stages of the interaction. Two time points were examined by Akiyama et al. (2002): 25 d and 45 d post-inoculation (dpi). At 25 dpi, the root colonization levels were similar under LP and HP. HP triggered down-regulation of AM symbiosis only at 45 dpi. In agreement with this, an AM-stimulating effect of the C-glycosylflavonoid on HP-grown plants was only observed at 45 dpi. In contrast, in the present conditions the negative impact of HP on root colonization could be observed as soon as the control roots became colonized (28 dpi, Table 1). Therefore, the mechanisms underlying suppression of AM symbiosis by HP may be different in the two systems. In addition, different plant species produce distinct arrays of flavonoids, making it difficult to extrapolate results from one species to another. Still, flavonoids remain interesting candidates as mediators of the P effect. In the present experimental system, branching bioassays supported the hypothesis of an effect of P on pre-symbiotic fungal development, since root exudate extracts of HP-grown plants failed to stimulate hyphal branching (Fig. 2). These extracts did not inhibit the effect of GR24 on the fungus, and therefore appeared to lack branching stimulants that are present in exudates of LP-grown plants. It must be noted, however, that in these experiments ethyl acetate extracts of root exudates were used in order to allow an adequate concentration of the samples. Therefore, it cannot be excluded that in addition to the lack of stimulants in the organic fraction, fungal inhibitors could be found in the aqueous fraction of HP root exudates. Another possible type of regulatory process is the display of signals on the root epidermal surface. For example, AM fungal hyphae can recognize specific patterns displayed by epidermal cells and differentiate hyphopodia on cell wall fragments of the epidermis, but not of other root tissues (Nagahashi and Douds, 1997). In addition, hyphopodia are formed on grooves between adjacent epidermal cells rather than on the outer cell wall. Interestingly, cell walls in these grooves appear thinner, looser, and richer in non-esterified pectin as compared with the tangential walls of epidermal cells (Bonfante et al., 2000). Whether such changes in cell wall composition contribute to the effect of HP on the formation of hyphopodia deserves further investigation. P supply affects strigolactone production in a systemic manner Strigolactones, identified as important contributors to the effect of host roots on pre-symbiotic fungal growth and metabolism (Akiyama et al., 2005; Besserer et al., 2006, 2008), were obvious candidates to mediate the effect of P supply because their synthesis is known to correlate inversely with P supply (Yoneyama et al., 2007a, b; Lopez-Raez et al., 2008). In agreement with this, strigolactones were undetectable in root exudates of plants grown under HP. Furthermore, HP supply was able to down-regulate strigolactone production in a systemic manner, as evidenced by the analysis of split-root plants (Fig. 4). This novel finding is particularly interesting in the context of the hormonal function of strigolactones. Indeed, a recent study has proposed that strigolactones mediate the tillering response to P starvation in rice (Umehara et al., 2010). In addition to the effect of strigolactones on lateral bud outgrowth, a role in the control of root architecture has recently been suggested (Koltai et al., 2009). This raises the possibility that P supply on one side of a plant affects development of a distant part of the root system through a modulation of strigolactone synthesis. It is already known that modifications of root architecture in response to P availability are integrated at the whole-plant level (Williamson et al., 2001), and it would be worth investigating the contribution of strigolactones to this phenomenon. The analysis of Pi contents in root and shoot tissues of split-root plants (Fig. 3C, D) revealed that root colonization levels and strigolactone production were linked to shoot Pi rather than to external P availability or local Pi concentrations in roots. This is consistent with the fact that HP exerts a dominant effect over LP in LP/HP split-root plant with regards to mycorrhizal and strigolactone exudation responses, and also with regards to Pi content (in LP/HP plants shoot Pi contents are similar to those of HP/HP plants). However, the signal underlying this systemic signalling remains unknown. Branscheid et al. (2010) proposed that the microRNA miR399 could act as a P starvation-induced signal to stimulate AM symbiosis under low P. MiR399 is known to accumulate in shoots under P deprivation, and to be transported to roots where it targets PHO2, a negative regulator of several P starvation responses (Lin et al., 2008; Pant et al., 2008). Interestingly, miR399 expression responded to AM root colonization, but overexpression of miR399 was not sufficient to improve AM root colonization under HP (Branscheid et al., 2010), suggesting that additional internal signals are required. Other microRNAs expressed in response to AM colonization and/or P supply (Gu et al. 2010) may be alternative candidates as systemic signals. Strigolactones are not solely responsible for P-triggered down-regulation of AM symbiosis The putative role of strigolactones as mediators of the P effect on AM symbiosis was supported by the good correlation between mycorrhizal colonization and strigolactone exudation in split-root plants (Figs 3, 4). Supplementation with exogenous GR24, however, failed to restore AM symbiosis in HP-grown plants (Fig. 5). These novel results rule out the proposed hypothesis that HP-grown plants are poorly colonized by AM fungi simply because they do not produce strigolactones (Yoneyama et al., 2007,b; Lopez-Raez et al., 2008). Although a role for strigolactones in the process is still possible and indeed likely, additional mechanisms remain to be discovered. This is consistent with the proposition that hyphal branching is a complex response involving several classes of compounds (Nagahashi and Douds, 2007). The present observations do not imply, however, that the absence of additional stimulatory compounds in HP root exudates is the only explanation for the lack of root colonization under HP. An additional possibility is that the hormonal function of strigolactones (rather than their role as rhizospheric signals) is involved in the regulation of AM symbiosis, for example by influencing root development or the ability of root cells to accommodate AM fungi. This question was not addressed in the present study, and the concentration of GR24 necessary to restore the putative hormonal function(s) of strigolactones in roots is not known. Nevertheless, the hypothesis of a strigolactone requirement at the plant hormonal level is not supported by previous observations that strigolactone-deficient mutants could still be slightly colonized by AM fungi (Gomez-Roldan et al., 2008). As determined by hyphal branching bioassays, HP conditions do not seem to prevent the stimulation of the fungus by strigolactones. The combination of HP root exudates with GR24 results in an activity similar to that of LP root exudates (Fig. 2). This suggests that hyphal branching and associated metabolic processes are restored in the supplementation experiment (HP+GR24) described in Fig. 5. The observation that this is not sufficient to allow root colonization by the fungus or the formation of hyphopodia points towards an effect of HP on steps other than hyphal branching, possibly including the differentiation of hyphopodia. Conclusion Collectively, the various reports on the down-regulation of AM symbiosis by P suggest that several successive layers of control operate in roots grown under HP. The experimental conditions used by different authors shed light on one or the other of these control mechanisms. Those described in this report allow the manipulatation of mycorrhizal symbiosis by targeting some of the first events in the interaction, and the testing of a number of hypotheses related to these events. It is demonstrated for the first time that the regulation of AM symbiosis by P is accompanied by a systemic regulation of strigolactone production, an important observation with regards to the hormonal function of these compounds. The decreased strigolactone content under HP, however, does not solely account for the strong mycorrhizal phenotype. The results therefore suggest the existence of additional early signalling events, some of which probably affect the differentiation of hyphopodia. A better understanding of this regulation should reveal important mechanisms required for the symbiosis under favourable conditions, and help circumvent the limitations for this symbiosis associated with the extensive use of P fertilizers in agriculture. The authors would like to thank Dr Koichi Yoneyama (Utsunomiya University, Japan) for the gift of synthetic standards of orobanchyl acetate and fabacyl acetate, and Dr Christian Brière (CNRS, Toulouse, France) for advice on statistical analyses. This study was partly funded by ASEDIS-SO, Toulouse, France. The Q Trap mass spectrometer was made available to us by the Metabolomics and Fluxomics platform of Toulouse (MetaToul). References Akiyama K,  Matsuoka H,  Hayashi H.  Isolation and identification of a phosphate deficiency-induced C-glycosylflavonoid that stimulates arbuscular mycorrhiza formation in melon roots,  Molecular Plant-Microbe Interactions ,  2002, vol.  15 (pg.  334- 340) Google Scholar CrossRef Search ADS PubMed  Akiyama K,  Matsuzaki K,  Hayashi H.  Plant sesquiterpenes induce hyphal branching in arbuscular mycorrhizal fungi,  Nature ,  2005, vol.  435 (pg.  824- 827) Google Scholar CrossRef Search ADS PubMed  Bécard G,  Fortin JA.  Early events of vesicular-arbuscular mycorrhiza formation on Ri T-DNA transformed roots,  New Phytologist ,  1988, vol.  108 (pg.  211- 218) Google Scholar CrossRef Search ADS   Besserer A,  Bécard G,  Jauneau A,  Roux C,  Séjalon-Delmas N.  GR24, a synthetic analog of strigolactones, stimulates the mitosis and growth of the arbuscular mycorrhizal fungus Gigaspora rosea by boosting its energy metabolism,  Plant Physiology ,  2008, vol.  148 (pg.  402- 413) Google Scholar CrossRef Search ADS PubMed  Besserer A,  Puech-Pages V,  Kiefer P,  Gomez-Roldan V,  Jauneau A,  Roy S,  Portais JC,  Roux C,  Bécard G,  Séjalon-Delmas N.  Strigolactones stimulate arbuscular mycorrhizal fungi by activating mitochondria,  PLoS Biology ,  2006, vol.  4 pg.  e226  Google Scholar CrossRef Search ADS PubMed  Bonfante P,  Genre A,  Faccio A,  Martini I,  Schauser L,  Stougaard J,  Webb J,  Parniske M.  The Lotus japonicus LjSym4 gene is required for the successful symbiotic infection of root epidermal cells,  Molecular Plant-Microbe Interactions ,  2000, vol.  13 (pg.  1109- 1120) Google Scholar CrossRef Search ADS PubMed  Bonser AM,  Lynch J,  Snapp S.  Effect of phosphorus deficiency on growth angle of basal roots in Phaseolus vulgaris,  New Phytologist ,  1996, vol.  132 (pg.  281- 288) Google Scholar CrossRef Search ADS PubMed  Bouwmeester HJ,  Roux C,  Lopez-Raez JA,  Bécard G.  Rhizosphere communication of plants, parasitic plants and AM fungi,  Trends in Plant Science ,  2007, vol.  12 (pg.  224- 230) Google Scholar CrossRef Search ADS PubMed  Bradbury SM,  Peterson RL,  Bowley SR.  Interactions between three alfalfa nodulation genotypes and two Glomus species,  New Phytologist ,  1991, vol.  119 (pg.  115- 120) Google Scholar CrossRef Search ADS   Branscheid A,  Sieh D,  Pant BD,  May P,  Devers EA,  Elkrog A,  Schauser L,  Scheible WR,  Krajinski F.  Expression pattern suggests a role of MiR399 in the regulation of the cellular response to local Pi increase during arbuscular mycorrhizal symbiosis,  Molecular Plant-Microbe Interactions ,  2010, vol.  23 (pg.  915- 926) Google Scholar CrossRef Search ADS PubMed  Bucher M.  Functional biology of plant phosphate uptake at root and mycorrhiza interfaces,  New Phytologist ,  2007, vol.  173 (pg.  11- 26) Google Scholar CrossRef Search ADS PubMed  Buée M,  Rossignol M,  Jauneau A,  Ranjeva R,  Bécard G.  The pre-symbiotic growth of arbuscular mycorrhizal fungi is induced by a branching factor partially purified from root exudates,  Molecular Plant-Microbe Interactions ,  2000, vol.  13 (pg.  693- 698) Google Scholar CrossRef Search ADS PubMed  Collins CD,  Foster BL.  Community-level consequences of mycorrhizae depend on phosphorus availability,  Ecology ,  2009, vol.  90 (pg.  2567- 2576) Google Scholar CrossRef Search ADS PubMed  David-Schwartz R,  Badani H,  Smadar W,  Levy AA,  Galili G,  Kapulnik Y.  Identification of a novel genetically controlled step in mycorrhizal colonization: plant resistance to infection by fungal spores but not extra-radical hyphae,  The Plant Journal ,  2001, vol.  27 (pg.  561- 569) Google Scholar CrossRef Search ADS PubMed  David-Schwartz R,  Gadkar V,  Wininger S,  Bendov R,  Galili G,  Levy AA,  Kapulnik Y.  Isolation of a premycorrhizal infection (pmi2) mutant of tomato, resistant to arbuscular mycorrhizal fungal colonization,  Molecular Plant-Microbe Interactions ,  2003, vol.  16 (pg.  382- 388) Google Scholar CrossRef Search ADS PubMed  Douds DD,  Galvez L,  Bécard G,  Kapulnik Y.  Regulation of arbuscular mycorrhizal development by plant host and fungus species in alfalfa,  New Phytologist ,  1998, vol.  138 (pg.  27- 35) Google Scholar CrossRef Search ADS   El Ghachtouli N,  Paynot M,  Morandi D,  Martin-Tanguy J,  Gianinazzi S.  The effect of polyamines on endomycorrhizal infection of wild-type Pisum sativum, cv. Frisson (nod+myc+) and two mutants (nod−myc+and nod−myc−),  Mycorrhiza ,  1995, vol.  5 (pg.  189- 192) Elias KS,  Safir GR.  Hyphal elongation of Glomus fasciculatus in response to root exudates,  Applied and Environmental Microbiology ,  1987, vol.  53 (pg.  1928- 1933) Google Scholar PubMed  Garriock ML,  Peterson RL,  Ackerley CA.  Early stages in colonization of Allium porum (leek) roots by the vesicular-arbuscular mycorrhizal fungus, Glomus versiforme,  New Phytologist ,  1989, vol.  112 (pg.  85- 92) Google Scholar CrossRef Search ADS   Gianinazzi-Pearson V,  Branzanti B,  Gianinazzi S.  In vitro enhancement of spore germination and early hyphal growth of a vesicular-arbuscular mycorrhizal fungus by root exudates and plant flavonoids,  Symbiosis ,  1989, vol.  7 (pg.  243- 255) Giovannetti M,  Mosse B.  An evaluation of techniques for measuring vesicular-arbuscular infection in roots,  New Phytologist ,  1980, vol.  84 (pg.  489- 500) Google Scholar CrossRef Search ADS   Giovannetti M,  Sbrana C,  Citernesi AS,  Avio L.  Analysis of factors involved in fungal recognition responses to host-derived signals by arbuscular mycorrhizal fungi,  New Phytologist ,  1996, vol.  133 (pg.  65- 71) Google Scholar CrossRef Search ADS   Gomez-Roldan V,  Fermas S,  Brewer PB, et al.  Strigolactone inhibition of shoot branching,  Nature ,  2008, vol.  455 (pg.  189- 194) Google Scholar CrossRef Search ADS PubMed  Graham JH,  Leonard RT,  Menge JA.  Membrane-mediated decrease in root exudation responsible for phosphorus inhibition of vesicular-arbuscular mycorrhiza formation,  Plant Physiology ,  1981, vol.  68 (pg.  548- 552) Google Scholar CrossRef Search ADS PubMed  Gu M,  Xu K,  Chen A,  Zhu Y,  Tang G,  Xu G.  Expression analysis suggests potential roles of microRNAs for phosphate and arbuscular mycorrhizal signaling in Solanum lycopersicum,  Physiologia Plantarum ,  2010, vol.  138 (pg.  226- 237) Google Scholar CrossRef Search ADS PubMed  Gutjahr C,  Novero M,  Guether M,  Montanari O,  Udvardi M,  Bonfante P.  Presymbiotic factors released by the arbuscular mycorrhizal fungus Gigaspora margarita induce starch accumulation in Lotus japonicus roots,  New Phytologist ,  2009, vol.  183 (pg.  53- 61) Google Scholar CrossRef Search ADS PubMed  Harrison MJ,  Dewbre GR,  Liu J.  A phosphate transporter from Medicago truncatula involved in the acquisition of phosphate released by arbuscular mycorrhizal fungi,  The Plant Cell ,  2002, vol.  14 (pg.  2413- 2429) Google Scholar CrossRef Search ADS PubMed  Hewitt EJ. ,  Sand and water culture methods used in the study of plant nutrition ,  1966 2nd edn London Commonwealth Agricultural Bureau Javot H,  Pumplin N,  Harrison MJ.  Phosphate in the arbuscular mycorrhizal symbiosis: transport properties and regulatory roles,  Plant, Cell and Environment ,  2007, vol.  30 (pg.  310- 322) Google Scholar CrossRef Search ADS   Johnson NC.  Can fertilization of soil select less mutualistic mycorrhizae?,  Ecological Applications ,  1993, vol.  3 (pg.  749- 757) Google Scholar CrossRef Search ADS PubMed  Koltai H,  Dor E,  Hershenhorn J, et al.  Strigolactones’ effect on root growth and root-hair elongation may be mediated by auxin-efflux carriers,  Journal of Plant Growth Regulation ,  2009, vol.  26 (pg.  129- 136) Kosuta S,  Chabaud M,  Lougnon G,  Gough C,  Dénarié J,  Barker DG,  Bécard G.  A diffusible factor from arbuscular mycorrhizal fungi induces symbiosis-specific MtENOD11 expression in roots of Medicago truncatula,  Plant Physiology ,  2003, vol.  131 (pg.  952- 962) Google Scholar CrossRef Search ADS PubMed  Kosuta S,  Hazledine S,  Sun J,  Miwa H,  Morris RJ,  Downie JA,  Oldroyd GE.  Differential and chaotic calcium signatures in the symbiosis signaling pathway of legumes,  Proceedings of the National Academy of Sciences, USA ,  2008, vol.  105 (pg.  9823- 9828) Google Scholar CrossRef Search ADS   Lin S-I,  Chiang SF,  Lin W-Y,  Chen J-W,  Tseng CY,  Wu PC,  Chiou T-J.  Regulatory network of microRNA399 and PHO2 by systemic signaling,  Plant Physiology ,  2008, vol.  147 (pg.  732- 746) Google Scholar CrossRef Search ADS PubMed  López-Ráez JA,  Charnikhova T,  Gómez-Roldán V, et al.  Tomato strigolactones are derived from carotenoids and their biosynthesis is promoted by phosphate starvation,  New Phytologist ,  2008, vol.  178 (pg.  863- 874) Google Scholar CrossRef Search ADS PubMed  Marsh JF,  Schultze M.  Analysis of arbuscular mycorrhizas using symbiosis-defective plant mutants,  New Phytologist ,  2001, vol.  150 (pg.  525- 532) Google Scholar CrossRef Search ADS   Nagahashi G,  Douds DD.  Appressorium formation by AM fungi on isolated cell walls of carrot roots,  New Phytologist ,  1997, vol.  136 (pg.  299- 304) Google Scholar CrossRef Search ADS   Nagahashi G,  Douds DJr.  Separated components of root exudate and cytosol stimulate different morphologically identifiable types of branching responses by arbuscular mycorrhizal fungi,  Mycological Research ,  2007, vol.  111 (pg.  487- 492) Google Scholar CrossRef Search ADS PubMed  Nagy R,  Drissner D,  Amrhein N,  Jakobsen I,  Bucher M.  Mycorrhizal phosphate uptake pathway in tomato is phosphorus-repressible and transcriptionally regulated,  New Phytologist ,  2008, vol.  181 (pg.  950- 959) Google Scholar CrossRef Search ADS   Nair MG,  Safir GR,  Siqueira JO.  Isolation and identification of vesicular-arbuscular mycorrhiza-stimulatory compounds from clover (Trifolium repens) roots,  Applied and Environmental Microbiology ,  1991, vol.  57 (pg.  434- 439) Google Scholar PubMed  Nanamori M,  Shinano T,  Wasaki J,  Yamamura T,  Rao IM,  Osaki M.  Low phosphorus tolerance mechanisms: phosphorus recycling and photosynthate partitioning in the tropical forage grass, Brachiaria hybrid cultivar Mulato compared with rice,  Plant and Cell Physiology ,  2004, vol.  45 (pg.  460- 469) Google Scholar CrossRef Search ADS PubMed  Navazio L,  Moscatiello R,  Genre A,  Novero M,  Baldan B,  Bonfante P,  Mariani P.  A diffusible signal from arbuscular mycorrhizal fungi elicits a transient cytosolic calcium elevation in host plant cells,  Plant Physiology ,  2007, vol.  144 (pg.  673- 681) Google Scholar CrossRef Search ADS PubMed  Oláh B,  Brière C,  Bécard G,  Dénarié J,  Gough C.  Nod factors and a diffusible factor from arbuscular mycorrhizal fungi stimulate lateral root formation in Medicago truncatula via the DMI1/DMI2 signalling pathway,  The Plant Journal ,  2005, vol.  44 (pg.  195- 207) Google Scholar CrossRef Search ADS PubMed  Pant BD,  Buhtz A,  Kehr J,  Scheible WR.  MicroRNA399 is a long-distance signal for the regulation of plant phosphate homeostasis,  The Plant Journal ,  2008, vol.  53 (pg.  731- 738) Google Scholar CrossRef Search ADS PubMed  Parniske M.  Arbuscular mycorrhiza: the mother of plant root endosymbioses,  Nature Reviews Microbiology ,  2008, vol.  6 (pg.  763- 775) Google Scholar CrossRef Search ADS PubMed  Paszkowski U,  Jakovleva L,  Boller T.  Maize mutants affected at distinct stages of the arbuscular mycorrhizal symbiosis,  The Plant Journal ,  2006, vol.  47 (pg.  165- 173) Google Scholar CrossRef Search ADS PubMed  Paszkowski U,  Kroken S,  Roux C,  Briggs SP.  Rice phosphate transporters include an evolutionarily divergent gene specifically activated in arbuscular mycorrhizal symbiosis,  Proceedings of the National Academy of Sciences, USA ,  2002, vol.  99 (pg.  13324- 13329) Google Scholar CrossRef Search ADS   Pumplin N,  Mondo SJ,  Topp S,  Starker CG,  Gantt JS,  Harrison MJ.  Medicago truncatula Vapyrin is a novel protein required for arbuscular mycorrhizal symbiosis,  The Plant Journal ,  2009, vol.  61 (pg.  482- 494) Google Scholar CrossRef Search ADS PubMed  Rausch C,  Daram P,  Brunner S,  Jansa J,  Laloi M,  Leggewie G,  Amrhein N,  Bucher M.  A phosphate transporter expressed in arbuscule-containing cells in potato,  Nature ,  2001, vol.  414 (pg.  462- 470) Google Scholar CrossRef Search ADS PubMed  Requena N,  Breuninger M,  Franken P,  Ocón A.  Symbiotic status, phosphate, and sucrose regulate the expression of two plasma membrane H+-ATPase genes from the mycorrhizal fungus Glomus mosseae,  Plant Physiology ,  2003, vol.  132 (pg.  1540- 1549) Google Scholar CrossRef Search ADS PubMed  Scervino JM,  Ponce MA,  Erra-Bassells R,  Bompadre J,  Vierheilig H,  Ocampo JA,  Godeas A.  The effect of flavones and flavonols on colonization of tomato plants by arbuscular mycorrhizal fungi of the genera Gigaspora and Glomus,  Canadian Journal of Microbiology ,  2007, vol.  53 (pg.  702- 709) Google Scholar CrossRef Search ADS PubMed  Siqueira JO,  Safir GR,  Nair MG.  Stimulation of vesicular-arbuscular mycorrhiza formation and growth of white clover by flavonoid compounds,  New Phytologist ,  1991, vol.  118 (pg.  87- 93) Google Scholar CrossRef Search ADS   Smith SE,  Read D. ,  Mycorrhizal symbiosis ,  2008 3rd edn London Academic Press Smith SE,  Smith FA,  Jakobsen I.  Mycorrhizal fungi can dominate phosphate supply to plants irrespective of growth responses,  Plant Physiology ,  2003, vol.  133 (pg.  16- 20) Google Scholar CrossRef Search ADS PubMed  St-Arnaud M,  Hamel C,  Vimard B,  Caron M,  Fortin JA.  Enhanced hyphal growth and spore production of the arbuscular mycorrhizal fungus Glomus intraradices in an in vitro system in the absence of host roots,  Mycological Research ,  1996, vol.  100 (pg.  328- 332) Google Scholar CrossRef Search ADS   Thomson BD,  Robson AD,  Abbott LK.  Effects of phsophorus on the formation of mycorrhizas by Gigaspora calospora and Glomus fasciculatum in relation to root carbohydrates,  New Phytologist ,  1986, vol.  103 (pg.  751- 765) Google Scholar CrossRef Search ADS   Thomson BD,  Robson AD,  Abbott LK.  Soil mediated effects of phosphorus supply on the formation of mycorrhizas by Scutellispora calospora (Nicol. & Gerd.) Walker & Sanders on subterranean clover,  New Phytologist ,  1991, vol.  118 (pg.  463- 469) Google Scholar CrossRef Search ADS   Tsai SM,  Phillips DA.  Flavonoids released naturally from alfalfa promote development of symbiotic Glomus spores in vitro,  Applied and Environmental Microbiology ,  1991, vol.  57 (pg.  1485- 1488) Google Scholar PubMed  Umehara M,  Hanada A,  Magome H,  Takeda-Kamiya N,  Yamaguchi S.  Contribution of strigolactones to the inhibition of tiller bud outgrowth under phosphate deficiency in rice,  Plant and Cell Physiology ,  2010, vol.  51 (pg.  1118- 1126) Google Scholar CrossRef Search ADS PubMed  Umehara M,  Hanada A,  Yoshida S, et al.  Inhibition of shoot branching by new terpenoid plant hormones,  Nature ,  2008, vol.  455 (pg.  195- 200) Google Scholar CrossRef Search ADS PubMed  Valdés-López O,  Arenas-Huertero C,  Ramírez M,  Girard L,  Sánchez F,  Vance CP,  Luis Reyes J,  Hernández G.  Essential role of MYB transcription factor: PvPHR1 and microRNA: PvmiR399 in phosphorus-deficiency signalling in common bean roots,  Plant, Cell and Environment ,  2008, vol.  31 (pg.  1834- 1843) Google Scholar CrossRef Search ADS   Vierheilig H.  Regulatory mechanisms during the plant–arbuscular mycorrhizal fungus interaction,  Canadian Journal of Botany ,  2004, vol.  82 (pg.  1166- 1176) Google Scholar CrossRef Search ADS   Vierheilig H,  Coughlan AP,  Wyss U,  Piche Y.  Ink and vinegar, a simple staining technique for arbuscular-mycorrhizal fungi,  Applied and Environmental Microbiology ,  1998, vol.  64 (pg.  5004- 5007) Google Scholar PubMed  Weidmann S,  Sanchez L,  Descombin J,  Chatagnier O,  Gianinazzi S,  Gianinazzi-Pearson V.  Fungal elicitation of signal transduction-related plant genes precedes mycorrhiza establishment and requires the dmi3 gene in Medicago truncatula,  Molecular Plant-Microbe Interactions ,  2004, vol.  17 (pg.  1385- 1393) Google Scholar CrossRef Search ADS PubMed  Williamson LC,  Ribrioux SP,  Fitter AH,  Leyser HM.  Phosphate availability regulates root system architecture in Arabidopsis,  Plant Physiology ,  2001, vol.  126 (pg.  875- 882) Google Scholar CrossRef Search ADS PubMed  Xie X,  Yoneyama K,  Harada Y,  Fusegi N,  Yamada Y,  Ito S,  Yokota T,  Takeuchi Y,  Yoneyama K.  Fabacyl acetate, a germination stimulant for root parasitic plants from Pisum sativum,  Phytochemistry ,  2009, vol.  70 (pg.  211- 215) Google Scholar CrossRef Search ADS PubMed  Yoneyama K,  Xie X,  Kusumoto D,  Sekimoto H,  Sugimoto Y,  Takeuchi Y,  Yoneyama K.  Nitrogen deficiency as well as phosphorus deficiency in sorghum promotes the production and exudation of 5-deoxystrigol, the host recognition signal for arbuscular mycorrhizal fungi and root parasites,  Planta ,  2007, vol.  227 (pg.  125- 132) Google Scholar CrossRef Search ADS PubMed  Yoneyama K,  Xie X,  Sekimoto H,  Takeuchi Y,  Ogasawara S,  Akiyama K,  Hayashi H,  Yoneyama K.  Strigolactones, host recognition signals for root parasitic plants and arbuscular mycorrhizal fungi, from Fabaceae plants,  New Phytologist ,  2008, vol.  179 (pg.  484- 494) Google Scholar CrossRef Search ADS PubMed  Yoneyama K,  Yoneyama K,  Takeuchi Y,  Sekimoto H.  Phosphorus deficiency in red clover promotes exudation of orobanchol, the signal for mycorrhizal symbionts and germination stimulant for root parasites,  Planta ,  2007, vol.  225 (pg.  1031- 1038) Google Scholar CrossRef Search ADS PubMed  Zhang Q,  Blaylock LA,  Harrison MJ.  Two Medicago truncatula half-ABC transporters are essential for arbuscule development in arbuscular mycorrhizal symbiosis,  The Plant Cell ,  2010, vol.  22 (pg.  1483- 1497) Google Scholar CrossRef Search ADS PubMed  © 2010 The Author(s). This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.5), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)

Dezar, Carlos A.;Giacomelli, Jorge I.;Manavella, Pablo A.;Ré, Delfina A.;Alves-Ferreira, Marcio;Baldwin, Ian T.;Bonaventure, Gustavo;Chan, Raquel L.

doi: 10.1093/jxb/erq339pmid: 21030388

Abstract The transcription factor HAHB10 belongs to the sunflower (Helianthus annuus) HD-Zip II subfamily and it has been previously associated with the induction of flowering. In this study it is shown that HAHB10 is expressed in sunflower leaves throughout the vegetative stage and in stamens during the reproductive stage. In short-day inductive conditions the expression of this gene is induced in shoot apexes together with the expression of the flowering genes HAFT and HAAP1. Transgenic Arabidopsis plants expressing HAHB10 cDNA under regulation either by its own promoter or by cauliflower mosaic virus (CaMV) 35S exhibited an early flowering phenotype. This phenotype was completely reverted in a non-inductive light regime, indicating a photoperiod-dependent action for this transcription factor. Gene expression profiling of Arabidopsis plants constitutively expressing HAHB10 indicated that specific flowering transition genes such as FT, FUL, and SEP3 were induced several fold, whereas genes related to biotic stress responses, such as PR1, PR2, ICS1, AOC1, EDS5, and PDF1-2a, were repressed. The expression of HAHB10 and of the flowering genes HASEP3 and HAFT was up-regulated by both salicylic acid (SA) treatment and infection with a virulent strain of Pseudomonas syringae. Basal SA and jasmonic acid (JA) levels in Arabidopsis plants ectopically expressing HAHB10 were similar to those of control plants; however, SA levels differentially increased in the transgenic plants after wounding and infection with P. syringae while JA levels differentially decreased. Taken together, the results indicated that HAHB10 participates in two different processes in plants: the transition from the vegetative to the flowering stage via the induction of specific flowering transition genes and the accumulation of phytohormones upon biotic stresses. FT, HAHB10, HD-Zip, plant defence mechanisms, salicylic acid, SEP3, sunflower transcription factor Introduction The transition from the vegetative to the reproductive stage occurs via complex mechanisms involving genes participating in different signal transduction pathways (Levy and Dean, 1998). The timing of this process is governed by several external factors such as light quality, photoperiod, and vernalization (Levy and Dean, 1998). Additionally, biotic and abiotic stresses can accelerate the entrance into the reproductive stage. For example, salicylic acid (SA), a critical hormone involved in the response to pathogens (Wildermuth, 2001; Loake and Grant, 2007; Park et al., 2007), has been described as an inducer of flowering in several plant species (Cleland and Ajami 1974; Khurana and Cleland, 1992). Martinez et al. (2004) analysed the transcript levels of several genes participating in flowering in SA-deficient plants and found that the FLOWERING TIME (FT) gene displays lower expression levels in these plants than in wild-type plants. Concomitantly, stress conditions increasing the endogenous SA content activate the expression of FT (Martínez et al., 2004). The ectopic expression of the Arabidopsis and tomato (Solanum lycopersicum) FT genes causes both early flowering (Kardailsky et al., 1999; Carmel-Goren et al., 2003) and enhanced levels of FRUITFULL (FUL) and SEPALLATA3 (SEP3) gene expression (Teper-Bamnolker and Samach, 2005). SA also participates as a signal molecule during infection with bacterial pathogens to control the expression of specific defence genes (Després et al., 2000; Spoel et al., 2003; Weigel et al., 2005). In contrast to SA, jasmonic acid (JA) levels usually decrease during bacterial infection, but they increase after wounding or attack by herbivores or necrotrophic pathogens (Balbi and Devoto, 2008). Although both hormones trigger specific healing and defence responses in plants, a very delicate balance between the levels of these two hormones is required to tailor these responses (Anderson et al., 2004). Moreover, SA and JA also interact with ethylene and abscisic acid (ABA) during defence responses to act either synergistically, additively, or antagonistically depending on the stimuli (Xu et al., 1994; O'Donnell et al., 1996; Penninckx et al., 1998; Lorenzo et al., 2003; Anderson et al., 2004). In a previous study, it was reported that HAHB10, a sunflower HD-Zip transcription factor belonging to subfamily II, was able to accelerate flowering and thereby shortens the plant's life cycle when ectopically expressed in Arabidopsis plants (Rueda et al., 2005). Transcription factors from the HD-Zip family are characterized by the association of an HD with a LZ domain, an association unique to plants (Chan et al., 1998). These proteins are classified in four subfamilies (I–IV) according to sequence conservation, gene structures, and functions, among other features (Chan et al., 1998; Ariel et al., 2007). Members of the subfamily II (like HAHB10) bind in vitro to the pseudopalindromic sequence CAAT(C/G)ATTG and those that have been characterized thus far are involved in light signalling pathways. For example, Arabidopsis ATHB2/HAT4 induces early flowering when ectopically expressed and it has been described as a master switch in the shade avoidance response (Schena et al., 1993; Steindler et al., 1999; Ciarbelli et al., 2008; Sorin et al., 2009). Several HD-Zip transcription factors from subfamily I, first assigned as regulators of abiotic stress responses, have been shown also to participate in biotic stress responses. For example, the sunflower HAHB4 confers drought tolerance and enhanced defence responses against insect attack when ectopically expressed in Arabidopsis (Manavella et al., 2008), the tomato H52 participates in the regulation of pathogen resistance and cell death (Mayda et al., 1999), and the Nicotiana benthaniana NbHB1 is a JA-dependent positive regulator of pathogen-induced plant cell death (Yoon et al., 2009). In this study, it was demonstrated that HAHB10 induces flowering by affecting the expression of specific genes operating during the vegetative to flowering transition. In addition to the up-regulation of flowering genes in these plants, several genes involved in biotic stress responses were repressed. Levels of SA, JA and ethylene were not changed in Arabidopsis plants expressing HAHB10 in control conditions but were significantly affected after bacterial infection and wounding, and these plants were affected in their response to a compatible interaction with Pseudomonas syringae. Materials and methods Plant material and growth conditions Arabidopsis thaliana Heyhn. ecotype Columbia (Col-0) and Helianthus annuus L. (sunflower CF33 cv. from Advanta) were grown in a growth chamber at 22–24 °C or 26–28 °C, respectively, under long-day or short-day photoperiods (LD or SD) as indicated in the figures (16 h or 8 h of illumination by a mixture of cool-white and GroLux fluorescent lamps). The light intensity in the culture chamber was ∼150 μE m−2 s−1. Arabidopsis plants were grown in Petri dishes containing 0.8% agar–Murashige and Skoog (MS) medium or in soil pots (8×7 cm) depending on the experiment and during the periods indicated in the figures. Sunflowers were grown in 27×30 cm soil pots until different developmental stages as defined by Shneiter (1981). In order to test inductive conditions, V10 plants were placed under a 8 h illumination regime during 96 h while their controls remained in LD conditions. Constructs For isolation of the HAHB10 promoter, a sunflower bacterial artificial chromosome (BAC) library (BAC Library HA_HBa, CUGI-Clemson University Genome Institute) was screened with a 32P-labelled probe corresponding to the 5′-non-coding region of the HAHB10 cDNA plus the first 241 nucleotides of the coding region, which does not include the HD-Zip domain (EcoRI/SpeI fragment). One of the isolated positive clones was digested with EcoRI and HindIII, and analysed in a Southern blot using the same probe. A 3025 bp hybridized fragment was subcloned in the pBluescript SK vector and sequenced (Macrogen-Korea). A 1399 bp DNA fragment (accession no. GQ470994) located upstream of the transcription initiation site was amplified by PCR using the oligonucleotides H10-1R (5′-CCGGGATCCCCATCTGAATAAAAAATGTGT-3′) and H10-8F (5′- CGCAAGCTTCTTGGTACCGATACCCAGAAC-3′) bearing the BamHI and HindIII sites, restricted with these enzymes, and cloned in the pBI 101.3 binary vector directing expression of the GUS (β-glucuronidase) reporter gene. The construct 35S:HAHB10 in the pBI121 vector was previously described (Rueda et al., 2005). PromHAHB10:HAHB10 was obtained by replacing the 35S cauliflower mosaic virus (CaMV) promoter (by BamHI/HindIII restriction) by the HAHB10 promoter fragment (1399 bp) in the above-mentioned 35S:HAHB10 construct. Escherichia coli DH5α cells were transformed with each construct and, once a positive clone was identified, it was used to transform Agrobacterium tumefaciens cells (LBA4404) (Höfgen and Willmitzer, 1988). Transformation and identification of transformed plants Stable transformation of Arabidopsis plants was carried out by the floral dip procedure (Clough and Bent, 1998). At least three independent homozygous lines of each transgenic genotype were used in each assay. Transiently transformed sunflower leaf discs (11 mm diameter) were obtained as described (Manavella and Chan, 2009). For each construct used, at least six discs cut from at least three different plants were analysed. To test the infiltration and transformation efficiencies, the expression of the simultaneously introduced kanamycin resistance gene was measured by real-time reverse transcription-PCR (RT-PCR) as described below. Microarray set-up Microarrays were based on the Arabidopsis Genome Oligo Set version 1.0 (Operon). This set consists of a total of 26 090 oligonucleotides that correspond to 22 361 annotated genes according to The Arabidopsis Information Resource (TAIR) genome annotation version 6. Microarrays were manufactured as previously described (Alves-Ferreira et al., 2007). Tissue collection and microarray experiments Tissue collection for the different biologically independent sets of samples was done on different days but at the same time of day to minimize any diurnal effects on gene expression. Total RNA was isolated from all tissue samples using the Trizol reagent and the RNA was cleaned up with an RNeasy RNA isolation kit (Qiagen) according to the manufacturer's instructions. Dye-labelled antisense RNA was generated from these total RNA preparations and hybridized to microarrays using a MAUI hybridization system (BioMicro Systems) as previously described (Alves-Ferreira et al., 2007). The dyes used for labelling RNA from the individual samples were switched in the replicate experiments to reduce dye-related artefacts. Data analysis Microarrays were scanned with an Axon GenePix 4200A scanner, using the Gene Pix 5.0 analysis software (Axon Instruments). Raw data were imported into the Resolver gene expression data analysis system (Rosetta Biosoftware) and processed as previously described (Alves-Ferreira et al., 2007). The P-values calculated by this software were adjusted for each experiment using the Benjamini and Hochberg procedure as implemented in the Bioconductor multitest package (http://www.bioconductor.org/packages/bioc/stable/src/contrib/html/multtest.html). Genes were considered as differentially expressed if they showed an absolute FC value of ≥2 between the wild type and a mutant and had been assigned an adjusted P-value of 0.05. All analyses in Resolver were done at the so-called sequence level; that is, data from reporters (probes) representing the same gene were combined. The percentage of promoters bearing the pseudopalindrome CAAT(C/G)ATTG was calculated by searching this sequence six times in 781 random chosen promoters in the TAIR9 database. Standard error was 0.1% taking these six samples. Histochemical GUS staining In situ assays of GUS activity were performed as described by Jefferson et al. (1987). Whole plants were immersed in a 1 mM 5-bromo-4-chloro-3-indolyl-glucuronic acid solution in 100 mM sodium phosphate pH 7.0 and 0.1% Triton X-100, and, after applying a vacuum for 5 min, they were incubated at 37 °C overnight. Chlorophyll was cleared from the plant tissues by immersion in 70% ethanol. RNA isolation and analysis by real-time RT-PCR measurements Total RNA from Arabidopsis or sunflower plants (at the developmental stages indicated in the figure legends) for quantitative real-time PCR (qPCR) was prepared with Trizol reagent (Invitrogen, http://www.invitrogen.com/) following the manufacturer's instructions (Invitrogen). qPCRs were carried out using an MJ-Cromos 4 (Bio-Rad, Hercules, CA, USA) apparatus as previously described (Manavella et al., 2006). The oligonucleotides used for these determinations are listed in Supplementary Table S2 available at JXB online. Hormone treatments Sunflower plants in the V2 and V4 developmental stage were sprayed with 100 μM SA, 200 μM JA, or left untreated (control). The hormone solutions were prepared in 0.2% Tween-20. Pathogen infections Pseudomonas syringae infections were carried out by spraying a suspension of virulent or avirulent strains (Pst DC3000 and Pst DC3000/avrRpt2, respectively) as described by Katagiri et al. (2002). Wounding and insect bioassays For the wounding treatments of sunflower leaves, one-half of the lamina was damaged by crushing with fine tweezers (∼50% of the surface was damaged). For the wounding treatment of Arabidopsis plants, the leaf (one-third of the surface) was crushed with fine tweezers and the remaining two-thirds were used for analysis. Larval mass gain was determined by placing one Spodoptera exigua larva per Arabidopsis plant (transgenic or controls). The plants were replaced daily and larval mass was determined daily during a period of 5 d with a microbalance. These tests were repeated at least 10 times using three independent transgenic lines for each genotype. For each experiment, at least 30 larvae were used per genotype. Phytohormone extraction and quantification Phytohormone extraction and quantification was carried out as previously described (von Dahl et al., 2007; Manavella et al., 2008) on flowering plants. Results HAHB10 transcripts accumulate in both vegetative and reproductive stages The sunflower HAHB10 gene accelerates the transition from the vegetative to the reproductive stage, leading to a shortening of the plant's life cycle when ectopically expressed in Arabidopsis (Rueda et al., 2005). In order to investigate the mechanisms underlying this process, the expression pattern of HAHB10 was first analysed in different sunflower tissues and developmental stages. Figure 1A shows that this gene was almost constitutively expressed in leaves, stems, and shoot apexes during the vegetative stage in LD while an inductive condition (SD) strongly induced the expression in shoot apexes. HAFT (HELIANTHUS ANNUS FLOWERING TIME) and HAAP1 (HELIANTHUS ANNUUS APETALLA 1) transcripts were identified by phylogenetic analysis (data not shown) and quantified as controls in the same samples (leaves, shoot apexes, and stems), showing a very similar pattern with an additional and expected induction of HAFT in leaves (Fig. 1B, C). HAHB10 and HAFT also showed a similar pattern in floral organs, while HAAP1, as expected, did not show expression at all in these organs (Fig. 1D). In addition, HASEP3 (HELIANTHUS ANNUUS SEPALLATA 3) and HAFUL (HELIANTHUS ANNUUS FRUITFULL) transcripts (identified by phylogenetic analysis) were quantified in these organs as controls, showing a similar pattern to their homologues from Arabidopsis. Fig. 1. View largeDownload slide Expression pattern of sunflower HAHB10 during vegetative and reproductive stages. (A) HAHB10 transcript levels in leaves (left panel), shoot apexes (central panel), and stems (right panel) at different developmental stages were measured by quantitative RT-PCR and related to the level quantified in young leaves (V2) arbitrarily taken as 1. Vx (vegetative stages from V2 to V10) in which x is a number representing the sunflower developmental stage according to Schneiter and Miller (1981). (B) Transcript levels of HAFT1 quantified in the same samples. (C) Transcript levels of HAAP1 quantified in the same samples. LD, plants grown in long-day conditions (16 h light/8 h darkness); SD, V10 plants grown in LD were placed under a 8 h illumination regime during 96 h prior to RNA isolation. (D) HAHB10, HAFT, HAAP1, HAFUL, and HASEP3 transcript levels quantified in reproductive organs (stamens and gynoecium) at the R5 developmental stage. Quantitiative RT-PCR values are related to that obtained in the organ in which the expression was the lowest and this last value was arbitrarily taken as 1. Standard errors were calculated taking three independent biological samples in which actin transcripts (ACTIN2 plus ACTIN8) were used as internal controls. Differences were considered significant when the P-values were <0.05 (Students t-test). (E) 35S:HAHB10 transgenic plants grown in LD (a, b, c, d) or SD (e, f, g, h) conditions compared with wild-type plants (35S:GUS) grown in the same conditions. In each panel, transgenic plants are on the right side while controls are on the left. a and b, 25-day-old plants; e and f, 50-day-old plants; c and d, 45-day-old plants; g and h, 60-day-old plants. Fig. 1. View largeDownload slide Expression pattern of sunflower HAHB10 during vegetative and reproductive stages. (A) HAHB10 transcript levels in leaves (left panel), shoot apexes (central panel), and stems (right panel) at different developmental stages were measured by quantitative RT-PCR and related to the level quantified in young leaves (V2) arbitrarily taken as 1. Vx (vegetative stages from V2 to V10) in which x is a number representing the sunflower developmental stage according to Schneiter and Miller (1981). (B) Transcript levels of HAFT1 quantified in the same samples. (C) Transcript levels of HAAP1 quantified in the same samples. LD, plants grown in long-day conditions (16 h light/8 h darkness); SD, V10 plants grown in LD were placed under a 8 h illumination regime during 96 h prior to RNA isolation. (D) HAHB10, HAFT, HAAP1, HAFUL, and HASEP3 transcript levels quantified in reproductive organs (stamens and gynoecium) at the R5 developmental stage. Quantitiative RT-PCR values are related to that obtained in the organ in which the expression was the lowest and this last value was arbitrarily taken as 1. Standard errors were calculated taking three independent biological samples in which actin transcripts (ACTIN2 plus ACTIN8) were used as internal controls. Differences were considered significant when the P-values were <0.05 (Students t-test). (E) 35S:HAHB10 transgenic plants grown in LD (a, b, c, d) or SD (e, f, g, h) conditions compared with wild-type plants (35S:GUS) grown in the same conditions. In each panel, transgenic plants are on the right side while controls are on the left. a and b, 25-day-old plants; e and f, 50-day-old plants; c and d, 45-day-old plants; g and h, 60-day-old plants. In order to correlate the expression pattern of HAHB10 to flowering, sunflower plants, responsive to SD, were placed in this condition when they reached the V10 stage. Notably, HAHB10 expression was clearly induced together with HAFT and HAAP1, indicating that this gene may function in the transition from the vegetative to the reproductive stage. The phenotype of 35S:HAHB10 Arabidopsis transgenic plants in SD was analysed and, as is shown in Fig. 1E, a clear reversion of the flowering acceleration takes place in LD, supporting a specific role for HAHB10 in flowering induction depending on the photoperiod. As a second approach to characterize the function of this gene, a 1399 bp fragment corresponding to the promoter region of HAHB10 was isolated from a sunflower genomic library and inserted upstream of the GUS reporter gene to generate transgenic Arabidopsis plants. In seedlings, GUS expression was evident in cotyledons, mainly in the vasculature (Fig. 2A), while in the vegetative stage expression was detected in leaves (primarily in the secondary veins) and roots (Fig. 2A). GUS expression was not detected in the central vascular system but it was detected in primary buds and stems (Fig. 2A). During the reproductive stage, GUS expression was evident in stamens and in stigmatic papillae and style of carpels at late stages of flower development (Fig. 2A). Thus, a strong correlation was observed between expression of HAHB10 in sunflower tissues and in Arabidopsis transgenic promHAHB10:GUS plants, suggesting the presence of conserved regulatory mechanisms for the expression of this gene in Arabidopsis. Fig. 2. View largeDownload slide promHAHB10 directs the expression of GUS in different stages of development and promotes an early flowering in Arabidopsis plants transformed with promHAHB10:HAHB10. (A) Histochemistry of transgenic Arabidopsis plants expressing the GUS reporter gene under the control of the HAHB10 promoter. (a, b, c) Three-day-old seedlings in control conditions, grown in darkness, and under far red illumination respectively. Ten-day-old aerial organs. (d, e, f) Seven-day-old plants; (g, h, i) 21-day old plants; (j) different stages of the inflorescences; (k) stems of a 35-day-old plant; (l, m) immature and mature inflorescences; (n) 35-day-old leaf; (o) 50-day-old plant silique. (B) Stem length (mm) measured with a ruler in 25-day-old plants from the three genotypes. (C) Front and upper view of three Arabidopsis transgenic genotypes. From left to right: Arabidopsis plants transformed with 35S:HAHB10, control plants transformed with 35S:GUS, and plants transformed with promHAHB10:HAHB10. This is a representative experiment performed with 32 individuals of each genotype. Fig. 2. View largeDownload slide promHAHB10 directs the expression of GUS in different stages of development and promotes an early flowering in Arabidopsis plants transformed with promHAHB10:HAHB10. (A) Histochemistry of transgenic Arabidopsis plants expressing the GUS reporter gene under the control of the HAHB10 promoter. (a, b, c) Three-day-old seedlings in control conditions, grown in darkness, and under far red illumination respectively. Ten-day-old aerial organs. (d, e, f) Seven-day-old plants; (g, h, i) 21-day old plants; (j) different stages of the inflorescences; (k) stems of a 35-day-old plant; (l, m) immature and mature inflorescences; (n) 35-day-old leaf; (o) 50-day-old plant silique. (B) Stem length (mm) measured with a ruler in 25-day-old plants from the three genotypes. (C) Front and upper view of three Arabidopsis transgenic genotypes. From left to right: Arabidopsis plants transformed with 35S:HAHB10, control plants transformed with 35S:GUS, and plants transformed with promHAHB10:HAHB10. This is a representative experiment performed with 32 individuals of each genotype. Thirdly, to investigate whether the early flowering phenotype observed in Arabidopsis plants was the result of an indirect effect induced by the constitutive expression of HAHB10 (35S:HAHB10), the HAHB10 cDNA was cloned downstream of its native promoter and used to generate transgenic Arabidopsis plants (promHAHB10:HAHB10). There were no morphological differences between promHAHB10:HAHB10 and control plants, whereas 35S-HAHB10 did show changes. When grown in LD, promHAHB10:HAHB10 leaves were similar to wild-type leaves in shape, size, and number (Fig, 2C). Moreover, the darker green colour, observed in 35S:HAHB10 Arabidopsis plants, was not evident in promHAHB10:HAHB10 plants (Fig. 2C, lower panel). Both, 35S:HAHB10 and promHAHB10:HAHB10 plants exhibited shorter life cycles compared with control plants (Fig. 2B and Table 1). In this regard, several reproduction-associated developmental processes including bolting time, stem length, number of siliques, and silique maturation time were markedly different between the genotypes. The main difference was observed between control plants and 35S:HAHB10 plants, while promHAHB10:HAHB10 plants showed intermediate values (Table 1). These results strongly suggested that the early flowering induced by the constitutive expression of HAHB10 in Arabidopsis (Rueda et al., 2005) was not an artefact of its ectopic expression. Table 1. PromHAHB10:HAHB10 plants have shorter life cycles compared with control plants Genotype and line name  No. of plants per experiment  No. of rosette leaves  Bolting time (d)  Stem length of 25-day-old plants (mm)  No. of cauline leaves of 25-day-old plants  No. of silique leaves of 30-day-old plants  Plant age for harvest (d)  35S:GUS  32  10±1  21±1  38.12±2.4  2  2±1  65±2  35S: HAHB10-A  32  7±1  19±1  90.65±2.8  3  5±1  50±1  promHAHB10:HAHB10-A  32  9±1  20±1  78.59±2.4  3  4±2  57±2  promHAHB10:HAHB10-B  32  10±1  20±1  52.62±2.5  3  3±1  60±1  promHAHB10:HAHB10-C  32  10±1  20±1  54.75±1.7  3  3±1  59±2  Genotype and line name  No. of plants per experiment  No. of rosette leaves  Bolting time (d)  Stem length of 25-day-old plants (mm)  No. of cauline leaves of 25-day-old plants  No. of silique leaves of 30-day-old plants  Plant age for harvest (d)  35S:GUS  32  10±1  21±1  38.12±2.4  2  2±1  65±2  35S: HAHB10-A  32  7±1  19±1  90.65±2.8  3  5±1  50±1  promHAHB10:HAHB10-A  32  9±1  20±1  78.59±2.4  3  4±2  57±2  promHAHB10:HAHB10-B  32  10±1  20±1  52.62±2.5  3  3±1  60±1  promHAHB10:HAHB10-C  32  10±1  20±1  54.75±1.7  3  3±1  59±2  Phenotypic characteristics of transgenic plants expressing HAHB10 under the control of its own promoter. Thirty-two individuals from each genotype as named in the first column were grown sharing the tray with an equal number of 35S:GUS individuals, used as controls, under standard conditions as described in the Materials and methods. Phenotypic parameters were taken at the periods indicated in the respective columns. The number of rosette leaves was determined in the transition from vegetative to reproductive stage. The experiment was repeated at least three times with these lines and the data shown are the average of the replicate. Plants with a high level expression of 35S:HAHB10 were advanced in their development and attained their maximal heights earlier. These are representative examples of experiments using other transgenic lines for each construction showing similar results (not included). Standard errors are expressed as the ratio between standard deviations and the square of the number of independent measurements. View Large The expression of HAHB10 induces significant changes in the Arabidopsis transcriptome In order to investigate the mechanisms involved in the developmental phenotype conferred by HAHB10, a comparative transcriptome analysis of Arabidopsis transgenic (35S:HAHB10) and control plants was performed. From a total of 30 081 genes analysed, 781 showed altered expression levels after a Benjamin and Hochberg false discovery rate (BDH-FDR) correction and selection for P-values <0.05 (Supplementary Table S1 at JXB online). The differentially expressed transcripts belonged to several different metabolic and signalling pathways according to their gene ontogeny (GO) annotation (Al-Shahrour et al., 2006). The signalling pathways exhibiting a significant over-representation were those of defence responses and flowering. A list of genes related to development and especially to flowering initiation, as well as those related to the photoperiod pathway, is presented in Table 2. Table 2. Genes involved in flowering which change their transcript levels in transgenic plants expressing HAHB10 ID  Description  Log2 ratio  P-value  Data obtained in the microarray analysis  At1g02230  NAM (no apical meristem)  −1.2  4.00E-02  At1g18810  Phytochrome kinase substrate 1  −0.99  1.00E-10  At4g32980  ATH1 (HD-BELL homeobox protein)  −0.86  7.00E-04  At5g14920  Gibberellins-regulated protein 1 precursor  −0.61  6.00E-06  At2g43010  PIF4 (phytochrome-interacting factor 4)  −0.52  3.00E-05  At5g62430  CDF1 (cycling 2 factor 1)  −0.44  3.00E-02  At2g02950  PKS1 (phytochrome kinase substrate 1)  −0.36  1.00E-02  AT1g14280  PKS2 (phytochrome kinase substrate 1)  −0.35  6.00E-05  At1g75820  CLV1 (CLAVATA 1 receptor kinase)  0.21  5.00E-02  At1g68050  FKF1 (E3 ubiquitin ligase SCF complex F-box subunit)  0.24  2.00E-02  At1g14920  GAI/RGA2 (GA insensitive-gibberellin response modulator)  0.3  2.00E-02  At1g22770  GI (gigantea protein)  0.31  1.00E-02  At1g56170  HAP5 (CCAAT-box binding transcription factor Hap5 putative)  0.32  4.00E-02  At1g04400  CRY2 (cryptochrome 2 apoprotein)  0.35  2.00E-02  At3g58070  GIS (GLABROUS INFLORESCENCE STEMS)  0.35  4.00E-02  At5g47640  HAP3b (CCAAT-box binding transcription factor Hap3b)  0.39  6.00E-05  At2g45660  SOC1 (suppressor of overexpression of CO 1-AGL20)  0.44  9.00E-04  At5g15840  CO (zinc finger protein CONSTANS)  0.45  6.00E-02  At1g54830  HAP5a (CCAAT-box binding transcription factor Hap5a)  0.47  2.00E-03  At1g53160  SPL4 (squamosa promoter-binding protein-like 4)  0.49  1.00E-02  At1g62360  STM (SHOOT MERISTEMLESS)  0.71  5.00E-07  At1g66350  RGL1 (gibberellin regulatory protein)  0.79  7.00E-06  At3g54340  AP3 (floral homeotic protein APETALA3)  0.91  2.00E-03  At5g03840  TFL1 (terminal flower 1 protein)  1.05  6.00E-08  At4g08150  KNAT1 (homeobox protein knotted-1 like 1)  1.14  6.00E-03  At1g65480  FT (flowering locus T protein)  1.25  1.00E-02  At5g24780  VSP1 (vegetative storage protein 1)  1.28  2.00E-02  At1g74670  GASA4 (gibberellins-regulated protein 4 precursor)  1.41  6.00E-09  At1g69600  ATHB29 (ZF-HD homeobox family protein)  1.58  1.00E-02  At5g60910  FUL (MADS-box protein FRUITFULL)  1.83  —  At5g65080  MAF5 (MADS AFFECTING FLOWERING 5)  1.85  1.00E-04  At3g02310  AGL4 (floral homeotic protein)  2.2  —  At5g20240  PI (floral homeotic protein PISTILLATA)  2.46  8.00E-05  At5g15800  AGL2 (floral homeotic protein)  2.61  —  At1g24260  SEP3 (MADS-box protein)  3.37  —  Data obtained in quantitative RT-PCR  At5g03840  TFL1 (terminal flower 1 protein)  1.26  6.E-02  At1g65480  FT (flowering locus T protein)  1.68  1.E-02  At5g60910  FUL (MADS-box protein FRUITFULL)  1.64  1.E-02  At5g65080  MAF5 (MADS AFFECTING FLOWERING 5)  1.41  5.E-02  At3g02310  AGL4 (floral homeotic protein)  1.30  5.E-02  At5g20240  PI (floral homeotic protein PISTILLATA)  0.92  4.E-02  At1g24260  SEP3 (MADS-box protein)  2.17  1.E-02  At5g15840  CO (zinc finger protein CONSTANS)  0.93  3.E-02  ID  Description  Log2 ratio  P-value  Data obtained in the microarray analysis  At1g02230  NAM (no apical meristem)  −1.2  4.00E-02  At1g18810  Phytochrome kinase substrate 1  −0.99  1.00E-10  At4g32980  ATH1 (HD-BELL homeobox protein)  −0.86  7.00E-04  At5g14920  Gibberellins-regulated protein 1 precursor  −0.61  6.00E-06  At2g43010  PIF4 (phytochrome-interacting factor 4)  −0.52  3.00E-05  At5g62430  CDF1 (cycling 2 factor 1)  −0.44  3.00E-02  At2g02950  PKS1 (phytochrome kinase substrate 1)  −0.36  1.00E-02  AT1g14280  PKS2 (phytochrome kinase substrate 1)  −0.35  6.00E-05  At1g75820  CLV1 (CLAVATA 1 receptor kinase)  0.21  5.00E-02  At1g68050  FKF1 (E3 ubiquitin ligase SCF complex F-box subunit)  0.24  2.00E-02  At1g14920  GAI/RGA2 (GA insensitive-gibberellin response modulator)  0.3  2.00E-02  At1g22770  GI (gigantea protein)  0.31  1.00E-02  At1g56170  HAP5 (CCAAT-box binding transcription factor Hap5 putative)  0.32  4.00E-02  At1g04400  CRY2 (cryptochrome 2 apoprotein)  0.35  2.00E-02  At3g58070  GIS (GLABROUS INFLORESCENCE STEMS)  0.35  4.00E-02  At5g47640  HAP3b (CCAAT-box binding transcription factor Hap3b)  0.39  6.00E-05  At2g45660  SOC1 (suppressor of overexpression of CO 1-AGL20)  0.44  9.00E-04  At5g15840  CO (zinc finger protein CONSTANS)  0.45  6.00E-02  At1g54830  HAP5a (CCAAT-box binding transcription factor Hap5a)  0.47  2.00E-03  At1g53160  SPL4 (squamosa promoter-binding protein-like 4)  0.49  1.00E-02  At1g62360  STM (SHOOT MERISTEMLESS)  0.71  5.00E-07  At1g66350  RGL1 (gibberellin regulatory protein)  0.79  7.00E-06  At3g54340  AP3 (floral homeotic protein APETALA3)  0.91  2.00E-03  At5g03840  TFL1 (terminal flower 1 protein)  1.05  6.00E-08  At4g08150  KNAT1 (homeobox protein knotted-1 like 1)  1.14  6.00E-03  At1g65480  FT (flowering locus T protein)  1.25  1.00E-02  At5g24780  VSP1 (vegetative storage protein 1)  1.28  2.00E-02  At1g74670  GASA4 (gibberellins-regulated protein 4 precursor)  1.41  6.00E-09  At1g69600  ATHB29 (ZF-HD homeobox family protein)  1.58  1.00E-02  At5g60910  FUL (MADS-box protein FRUITFULL)  1.83  —  At5g65080  MAF5 (MADS AFFECTING FLOWERING 5)  1.85  1.00E-04  At3g02310  AGL4 (floral homeotic protein)  2.2  —  At5g20240  PI (floral homeotic protein PISTILLATA)  2.46  8.00E-05  At5g15800  AGL2 (floral homeotic protein)  2.61  —  At1g24260  SEP3 (MADS-box protein)  3.37  —  Data obtained in quantitative RT-PCR  At5g03840  TFL1 (terminal flower 1 protein)  1.26  6.E-02  At1g65480  FT (flowering locus T protein)  1.68  1.E-02  At5g60910  FUL (MADS-box protein FRUITFULL)  1.64  1.E-02  At5g65080  MAF5 (MADS AFFECTING FLOWERING 5)  1.41  5.E-02  At3g02310  AGL4 (floral homeotic protein)  1.30  5.E-02  At5g20240  PI (floral homeotic protein PISTILLATA)  0.92  4.E-02  At1g24260  SEP3 (MADS-box protein)  2.17  1.E-02  At5g15840  CO (zinc finger protein CONSTANS)  0.93  3.E-02  The first column shows gene identity; the second column shows the gene name and description; the third column shows the log2 of the ratio between transcript levels in transgenic (35S:HAHB10) plants related to those in control plants. The P-value was determined according to the Bonferroni test; – indicates that the P-value is <E−10. View Large Validation of the microarray results for some of these genes was performed by qPCR on three biological replicates coming from independent transgenic lines. A good correlation in the changes of gene expression was found between the two methods (Tables 2, 3 and 4). Table 3. Genes involved in the defence response which change their expression level in plants transformed with HAHB10 ID  Description  Log2 ratio  P-value  Data obtained in the microarray analysis  At1g02360  Chitinase putative  −2.33  —  At3g23220  AP2/ATERF  −1.83  3.E-02  At1g73330  DR4 (protease inhibitor)  −1.69  —  At1g18870  ICS2 (isochorismate synthase 2)  −1.66  2.E-03  At3g22231  PCC1 (PATHOGEN AND CIRCADIAN CONTROLLED 1)  −1.52  5.E-02  At2g32680  ATRLP23 (receptor like protein 23)  −1.51  6.E-09  At4g36470  Similar to SAM:JMT and to SAM:SAMT  −1.44  2.E-05  At3g05730  DEFL (defensin-like family protein)  −1.32  7.E-07  At1g72930  TIR (Toll-interleukin-resistance)  −1.31  1.E-03  At3g50470  HR3 (hypersensitive response protein 3)  −1.25  2.E-02  At3g56400  WRKY70 (WRKY family transcription factor 70)  −1.23  2.E-09  At1g02450  NIMIN-1 (NPR1/NIM1-interacting protein 1)  −1.21  4.E-04  At3g25760  AOC1 (allene oxide cyclase)  −1.09  2.E-02  At1g75040  PR5 (pathogenesis-related protein 5)  −1.03  3.E-06  At3g48080  Disease resistance protein-related  −1.00  9.E-03  At3g25882  NIMIN-2 (NPR1/NIM1-interacting protein 2)  −0.99  2.E-04  At1g33590  LRR protein-related similar to Hcr2-5D  −0.96  2.E-08  At3g23110  AtRLP37 (disease resistance family protein similar to Cf-2.2)  −0.92  5.E-07  At1g17600  TIR-NBS-LRR class  −0.89  2.E-03  At5G48657  Defence protein-related  −0.88  4.E-02  At3g20600  NDR1 (non-race-specific disease resistance protein)  −0.87  —  At1g72940  TIR-NBS class  −0.85  1.E-07  At2g40750  WRKY54 (WRKY family transcription factor 54)  −0.85  6.E-07  At1g73325  Trypsin and protease inhibitor family protein  0.85  9.E-04  At3g13662  Disease resistance-responsive protein-related  0.88  3.E-02  At4g23600  CORI3 (coronatine-responsive tyrosine aminotransferase)  0.93  7.E-04  At3g45140  ATLOX2 (lipoxygenase 2)  1.02  1.E-03  At5g42500  Similar to disease resistance response protein 206-d  1.02  2.E-02  At2g42885  DEFL (encodes a defensin-like family protein)  1.23  2.E-03  At5g24780  VSP1 (vegetative storage protein 1)  1.28  2.E-02  At1g19640  JMT (S-Ade-L-Met:JA carboxyl methyltransferase)  1.31  2.E-03  At4g10265  Similar to wound-induced protein of L. esculentum  1.41  1.E-02  Data obtained in quantitative RT-PCR  At1g73330  DR4 (protease inhibitor)  −1.41  5.E-02  At1g18870  ICS2 (isochorismate synthase 2)  −0.93  1.E-02  At1g02450  NIMIN-1 (NPR1/NIM1-interacting protein 1)  −1.63  5.E-02  At3g25760  ERD12 (allene oxide cyclase)  −0.52  4.E-02  At3g25882  NIMIN-2 (NPR1/NIM1-interacting protein 2)  −2.24  5.E-02  ID  Description  Log2 ratio  P-value  Data obtained in the microarray analysis  At1g02360  Chitinase putative  −2.33  —  At3g23220  AP2/ATERF  −1.83  3.E-02  At1g73330  DR4 (protease inhibitor)  −1.69  —  At1g18870  ICS2 (isochorismate synthase 2)  −1.66  2.E-03  At3g22231  PCC1 (PATHOGEN AND CIRCADIAN CONTROLLED 1)  −1.52  5.E-02  At2g32680  ATRLP23 (receptor like protein 23)  −1.51  6.E-09  At4g36470  Similar to SAM:JMT and to SAM:SAMT  −1.44  2.E-05  At3g05730  DEFL (defensin-like family protein)  −1.32  7.E-07  At1g72930  TIR (Toll-interleukin-resistance)  −1.31  1.E-03  At3g50470  HR3 (hypersensitive response protein 3)  −1.25  2.E-02  At3g56400  WRKY70 (WRKY family transcription factor 70)  −1.23  2.E-09  At1g02450  NIMIN-1 (NPR1/NIM1-interacting protein 1)  −1.21  4.E-04  At3g25760  AOC1 (allene oxide cyclase)  −1.09  2.E-02  At1g75040  PR5 (pathogenesis-related protein 5)  −1.03  3.E-06  At3g48080  Disease resistance protein-related  −1.00  9.E-03  At3g25882  NIMIN-2 (NPR1/NIM1-interacting protein 2)  −0.99  2.E-04  At1g33590  LRR protein-related similar to Hcr2-5D  −0.96  2.E-08  At3g23110  AtRLP37 (disease resistance family protein similar to Cf-2.2)  −0.92  5.E-07  At1g17600  TIR-NBS-LRR class  −0.89  2.E-03  At5G48657  Defence protein-related  −0.88  4.E-02  At3g20600  NDR1 (non-race-specific disease resistance protein)  −0.87  —  At1g72940  TIR-NBS class  −0.85  1.E-07  At2g40750  WRKY54 (WRKY family transcription factor 54)  −0.85  6.E-07  At1g73325  Trypsin and protease inhibitor family protein  0.85  9.E-04  At3g13662  Disease resistance-responsive protein-related  0.88  3.E-02  At4g23600  CORI3 (coronatine-responsive tyrosine aminotransferase)  0.93  7.E-04  At3g45140  ATLOX2 (lipoxygenase 2)  1.02  1.E-03  At5g42500  Similar to disease resistance response protein 206-d  1.02  2.E-02  At2g42885  DEFL (encodes a defensin-like family protein)  1.23  2.E-03  At5g24780  VSP1 (vegetative storage protein 1)  1.28  2.E-02  At1g19640  JMT (S-Ade-L-Met:JA carboxyl methyltransferase)  1.31  2.E-03  At4g10265  Similar to wound-induced protein of L. esculentum  1.41  1.E-02  Data obtained in quantitative RT-PCR  At1g73330  DR4 (protease inhibitor)  −1.41  5.E-02  At1g18870  ICS2 (isochorismate synthase 2)  −0.93  1.E-02  At1g02450  NIMIN-1 (NPR1/NIM1-interacting protein 1)  −1.63  5.E-02  At3g25760  ERD12 (allene oxide cyclase)  −0.52  4.E-02  At3g25882  NIMIN-2 (NPR1/NIM1-interacting protein 2)  −2.24  5.E-02  The first column shows gene identity; the second column shows the gene name and description; the third column shows the log2 of the ratio between transcript levels in transgenic (35S:HAHB10) plants related to those in control plants. The P-value was determined according to the Bonferroni test; – indicates that the P-value is <E−10. View Large Table 4. Arabidopsis HD-Zip II members which change their transcript levels in transgenic plants expressing HAHB10 ID  Description  Log2 ratio  P-value  Data obtained in the microarray analysis        At4g17460  HAT1 (homeobox-leucine zipper protein 1)  −2.35  —  At4g37790  HAT22 (homeobox-leucine zipper protein 22)  −1.60  5.E-09  At4g16780  HAT4 (homeobox-leucine zipper protein 4)  −1.58  3.E-06  At5g06710  HAT14 (homeobox-leucine zipper protein 14)  −1.29  3.E-03  At5g47370  HAT2 (homeobox-leucine zipper protein 2)  −1.18  —  At3g60390  HAT3 (homeobox-leucine zipper protein)  −1.15  3.E-03  At2g22430  ATHB-6 (homeobox-leucine zipper protein 6)  0.67  1.E-05  At2g46680  ATHB-7 (homeobox-leucine zipper protein 7)  0.69  5.E-02  At4g40060  ATHB-16 (homeobox-leucine zipper protein 16)  0.79  2.E-03  At5g65310  ATHB-5 (homeobox-leucine zipper protein 5)  0.93  1.E-02  Data obtained in quantitative RT-PCR        At4g17460  HAT1 (homeobox-leucine zipper protein 1)  −1.31  1.E-02  At3g60390  HAT3 (homeobox-leucine zipper protein 3)  −1.48  1.E-02  At4g16780  HAT4 (homeobox-leucine zipper protein 4)  −2.20  1.E-02  ID  Description  Log2 ratio  P-value  Data obtained in the microarray analysis        At4g17460  HAT1 (homeobox-leucine zipper protein 1)  −2.35  —  At4g37790  HAT22 (homeobox-leucine zipper protein 22)  −1.60  5.E-09  At4g16780  HAT4 (homeobox-leucine zipper protein 4)  −1.58  3.E-06  At5g06710  HAT14 (homeobox-leucine zipper protein 14)  −1.29  3.E-03  At5g47370  HAT2 (homeobox-leucine zipper protein 2)  −1.18  —  At3g60390  HAT3 (homeobox-leucine zipper protein)  −1.15  3.E-03  At2g22430  ATHB-6 (homeobox-leucine zipper protein 6)  0.67  1.E-05  At2g46680  ATHB-7 (homeobox-leucine zipper protein 7)  0.69  5.E-02  At4g40060  ATHB-16 (homeobox-leucine zipper protein 16)  0.79  2.E-03  At5g65310  ATHB-5 (homeobox-leucine zipper protein 5)  0.93  1.E-02  Data obtained in quantitative RT-PCR        At4g17460  HAT1 (homeobox-leucine zipper protein 1)  −1.31  1.E-02  At3g60390  HAT3 (homeobox-leucine zipper protein 3)  −1.48  1.E-02  At4g16780  HAT4 (homeobox-leucine zipper protein 4)  −2.20  1.E-02  The first column shows gene identity; the second column shows the gene name and description; the third column shows the log2 of the ratio between transcript levels in transgenic (35S:HAHB10) plants related to those in control plants. The P-value was determined according to the Bonferroni test; – indicates that the P-value is <E−10. View Large Like other members of the HD-Zip II subfamily, HAHB10 binds in vitro to the pseudopalindromic sequence CAAT(C/G)ATTG (Tron et al., 2002). An investigation was therefore carries out to determine which of the differentially expressed genes contain this pseudopalindromic sequence in their promoter regions. For this analysis, a region of 1000 bp upstream of the transcription initiation site of the corresponding genes was extracted and evaluated [Arabidopsis promoters (TAIR9 genome release)]. Of the 781 differentially expressed genes, 2.8% contained the CAAT(C/G)ATTG element, indicating that this percentage of genes could be direct targets of HAHB10. This value was 1.1% higher than the value corresponding to the percentage of genes containing the promoter sequence [CAAT(C/G)ATTG] in the complete Arabidopsis genome (1.7%). Early flowering is associated with up-regulation of a key set of genes in transgenic plants Based on the early flowering phenotype conferred by HAHB10 and the gene expression data, the analysis was focused on those genes involved in the transition from the reproductive to the flowering stage (which presented the largest changes in the microarray analysis) and their transcript levels were measured in 5-week-old leaves of 35S:HAHB10 transgenic plants grown in LD. SEP3 transcript levels were induced ∼25-fold in all the transgenic lines tested, whereas FT and FUL transcript levels were increased ∼5-fold compared with control plants (Fig. 3). To understand better the numerical differences between the microarray and the qPCR results, it must be considered that these strong inductions were detected in leaves from mature plants (already in the reproductive–fructification stages) while the microarray analysis was performed with leaves from 21-day-old plants (entering the vegetative–reproductive stage). Fig. 3. View largeDownload slide Genes involved in flowering are induced in transgenic plants expressing HAHB10. (A) Transcript levels of SEP3, FUL, and FT genes, associated with flowering initiation, were analysed by qRT-PCR in three independent transgenic lines (A, B, and C) transformed with the construct 35S:HAHB10. As controls, plants transformed with 35S:GUS (named WT) were used. Transcript levels of each gene determined in three biological replicates were related to the level detected in WT leaves. ACTIN (ACTIN2 plus ACTIN8) and UBIQUITIN (UBI9) were used as internal controls. Standard errors were calculated taking three independent experiments, and differences were considered significant when P-values were <0.05 (Students t-test). (B) Sunflower leaf discs were transformed with 35S:GUS used as control (C) or with 35S:HAHB10 (T). Transcript levels of HAHB10, HAFT (DY917234.1), and HASEP3 (EL489638.1) were measured by qRT-PCR 72 h after transformation. ACTIN (ACTIN2 plus ACTIN8) and UBIQUITIN (UBI9) were used as internal controls. Standard errors were calculated from at least three independent experiments with biological sextuplicates, and differences were considered significant when the P-values were <0.05 (Students t-test). Fig. 3. View largeDownload slide Genes involved in flowering are induced in transgenic plants expressing HAHB10. (A) Transcript levels of SEP3, FUL, and FT genes, associated with flowering initiation, were analysed by qRT-PCR in three independent transgenic lines (A, B, and C) transformed with the construct 35S:HAHB10. As controls, plants transformed with 35S:GUS (named WT) were used. Transcript levels of each gene determined in three biological replicates were related to the level detected in WT leaves. ACTIN (ACTIN2 plus ACTIN8) and UBIQUITIN (UBI9) were used as internal controls. Standard errors were calculated taking three independent experiments, and differences were considered significant when P-values were <0.05 (Students t-test). (B) Sunflower leaf discs were transformed with 35S:GUS used as control (C) or with 35S:HAHB10 (T). Transcript levels of HAHB10, HAFT (DY917234.1), and HASEP3 (EL489638.1) were measured by qRT-PCR 72 h after transformation. ACTIN (ACTIN2 plus ACTIN8) and UBIQUITIN (UBI9) were used as internal controls. Standard errors were calculated from at least three independent experiments with biological sextuplicates, and differences were considered significant when the P-values were <0.05 (Students t-test). To correlate the data obtained in Arabidopsis with the regulatory networks involving HAHB10 in sunflower, the expression pattern of putative target genes, homologous to those identified as differentially expressed in Arabidopsis, was analyzed in transformed leaf discs. HASEP3 and HAFT transcripts were evaluated in this tissue transformed with the construct 35S:HAHB10 or with the control construct 35S:GUS. Consistent with the results obtained in Arabidopsis, overexpression of HAHB10 induced HASEP3 and HAFT genes ∼15-fold. HAHB10 down-regulates genes involved in the defence response A significant number of genes related to defence responses were identified as differentially expressed (most of them down-regulated) in transgenic plants constitutively expressing HAHB10 (Table 3). Among them, there were some genes participating in the initial steps of the defence response such as ICS2 (ISOCHORISMATE SYNTHASE 2) and EDS1 (ENHANCED DISEASE SUSCEPTIBILITY 1), both required for SA synthesis/signalling in pathogen-challenged plants, and genes encoding pathogenesis-related proteins such as PR2 and PR5 (Table 3). Some of these genes together with other defence marker genes were selected to analyse their expression by qPCR. PR1 and PDF1.2a mRNA levels were repressed in HAHB10-expressing plants compared with control plants (Fig. 4A). The mRNA levels of ICS1 (involved in SA synthesis), EDS5 (encoding a transporter in SA signalling), AOC1 (involved in JA synthesis), and PR2 (involved in the defence response) were also lower in these plants (Fig. 4). These results showed that the basal levels of expression of some genes involved in both SA and JA biosynthesis and some of their induced defence responses were affected in 35S:HAHB10 Arabidopsis plants. Fig. 4. View largeDownload slide Genes related to the biotic stress response are repressed in transgenic plants expressing HAHB10. (A) Transcript levels of Arabidopsis ICS1, AOC1, EDS5, PR1, PR2, and PDF1.2 genes associated with the defence response were analysed by qRT-PCR in three independent transgenic lines (A, B, and C) transformed with the 35S:HAHB10 construct (35S:HAHB10). Control plants (WT) were transformed with pBI121. Quantifications were related to the level of each gene in WT leaves and repeated at least three times with biological triplicates. (B) Sunflower leaf discs were transformed with 35S:GUS (C) or with 35S:HAHB10 (T). Transcript levels of HAHB10, HAPR1, HAPR3, and HALOX2 were quantified by qRT-PCR 72 h after transformation. In both Arabidopsis and sunflower, A and B, ACTIN (ACTIN2 plus ACTIN8) and UBIQUITIN (UBI9) were used as internal controls. Standard errors were calculated from at least three independent experiments with six biological replicates, and differences were considered significant when the P-values were <0.05 (Students t-test). Fig. 4. View largeDownload slide Genes related to the biotic stress response are repressed in transgenic plants expressing HAHB10. (A) Transcript levels of Arabidopsis ICS1, AOC1, EDS5, PR1, PR2, and PDF1.2 genes associated with the defence response were analysed by qRT-PCR in three independent transgenic lines (A, B, and C) transformed with the 35S:HAHB10 construct (35S:HAHB10). Control plants (WT) were transformed with pBI121. Quantifications were related to the level of each gene in WT leaves and repeated at least three times with biological triplicates. (B) Sunflower leaf discs were transformed with 35S:GUS (C) or with 35S:HAHB10 (T). Transcript levels of HAHB10, HAPR1, HAPR3, and HALOX2 were quantified by qRT-PCR 72 h after transformation. In both Arabidopsis and sunflower, A and B, ACTIN (ACTIN2 plus ACTIN8) and UBIQUITIN (UBI9) were used as internal controls. Standard errors were calculated from at least three independent experiments with six biological replicates, and differences were considered significant when the P-values were <0.05 (Students t-test). To assess whether the expression of these genes was also affected in sunflower leaves overexpressing HAHB10, their putative homologues in sunflower were identified by phylogenetic analysis (data not shown). The expression of HAPR1, HAPR3, and HALOX2 transcripts was quantified in transiently transformed sunflower leaf discs. Consistent with the results obtained in Arabidopsis, overexpression of HAHB10 repressed the levels of these mRNAs (Fig. 4B). HAHB10 is regulated by phytohormones and conditions related to biotic stresses In sunflower leaves, the expression of HAHB10 and HAFT was induced after 12 h of SA treatment (Fig. 5A, B), whereas the expression of HASEP3 did not change (Fig. 5C). HAHB10 and HAFT presented similar kinetics of induction, with a peak at 12 h, slowly decreasing thereafter. The defence marker HAPR1 presented different kinetics of induction, with a continuous increase up to 72 h (Fig. 5D). HAHB10 was also induced after infection of sunflower leaves with a virulent strain of P. syringae, showing a peak at 48 h (Fig. 5E). In contrast, when the sunflower leaves were wounded, a significant repression of HAHB10 mRNA levels was observed (Fig. 5F). Fig. 5. View largeDownload slide HAHB10 expression in sunflower is regulated by phytohormones and biotic stress. Kinetics of induction of HAHB10 (A), HAFT (B), HASEP3 (C), and HAPR1 (D) with 100 μM SA treatment. Expression kinetics of HAHB10 after infection with virulent (avir) and avirulent (vir) strains of Pseudomonas syringae DC3000. (F) Kinetics of HAHB10 repression after wounding. The assays in A, B, C, and D were performed on V4 stage leaves. Time periods in which samples were collected are expressed in hours (A–E) or minutes (F). (G–I) Expression levels of FT, PR1, and PDF1.2a in 3-week-old transgenic (35S:HAHB10 or promHAHB10:HAHB10) and WT (transformed with pBI 121) plants after a treatment with 1 mM SA during 6 h. ACTIN genes (ACTIN2 plus ACTIN8) were used as internal controls. Standard deviations were calculated from at least three independent experiments with three biological replicates and differences were considered significant when the P-values were <0.05 (Students t-test). Fig. 5. View largeDownload slide HAHB10 expression in sunflower is regulated by phytohormones and biotic stress. Kinetics of induction of HAHB10 (A), HAFT (B), HASEP3 (C), and HAPR1 (D) with 100 μM SA treatment. Expression kinetics of HAHB10 after infection with virulent (avir) and avirulent (vir) strains of Pseudomonas syringae DC3000. (F) Kinetics of HAHB10 repression after wounding. The assays in A, B, C, and D were performed on V4 stage leaves. Time periods in which samples were collected are expressed in hours (A–E) or minutes (F). (G–I) Expression levels of FT, PR1, and PDF1.2a in 3-week-old transgenic (35S:HAHB10 or promHAHB10:HAHB10) and WT (transformed with pBI 121) plants after a treatment with 1 mM SA during 6 h. ACTIN genes (ACTIN2 plus ACTIN8) were used as internal controls. Standard deviations were calculated from at least three independent experiments with three biological replicates and differences were considered significant when the P-values were <0.05 (Students t-test). Transgenic Arabidopsis plants (3 weeks old) expressing HAHB10 under the control of its own promoter were treated with SA and the levels of FT, PR1, and PDF1.2a transcripts were quantified in leaves 6 h after the treatment. The results showed (Fig. 5G–I) that under control conditions the transcript levels of FT were higher in transgenic plants than in the wild type, both in the constitutive genotype and in promHAHB10:HAHB10 plants, and did not change significantly after the treatment. Under control conditions, PR1 transcript levels were repressed in both transgenic genotypes; however, they were induced to similar levels in control plants after SA treatment. In contrast, PDF1.2a transcript levels remained repressed in all genotypes. Levels of SA and JA are affected in Arabidopsis HAHB10-expressing plants after wounding and bacterial infection As mentioned above, the constitutive expression of HAHB10 in Arabidopsis reduced the basal levels of expression of genes regulated by SA and JA. Moreover, HAHB10 mRNA levels were up-regulated by SA treatment but were, however, slightly repressed by wounding (Fig. 5). In order to elucidate some aspects of the complex relationship between HAHB10 and SA- and JA-mediated responses, the amounts of these two phytohormones were quantified in 35S:HAHB10 Arabidopsis plants and in control plants after mechanical damage and P. syringae infection. The results indicated that the basal levels of JA and SA were similar between control and transgenic plants (Fig. 6A, 6B). However, 24 h after infection with a virulent strain of P. syringae, a significant increase in SA levels was observed in transgenic plants compared with control plants (Fig. 6A; all the differences were statistically significant; P <0.05, t-test). At 48 h, SA levels were reduced to control levels (Fig. 6A) while JA levels did not change significantly during infection (Fig. 6B). As expected, mechanical damage induced JA levels with a peak at 60 min in control plants; however, the levels were ∼2.5-fold lower in transgenic plants than in controls (Fig. 6C; all the differences were statistically significant; P <0.05, t-test). In contrast, SA levels did not change significantly in wounded control plants whereas, similar to bacterial infection, they were elevated between 3- and 4.5-fold in wounded transgenic plants (Fig. 6D; all the differences were statistically significant; P <0.05, t-test). Fig. 6. View largeDownload slide The ectopic expression of HAHB10 modulates the synthesis of SA and JA in Arabidopsis plants. (A and B) SA and JA quantification was performed in Arabidopsis plants transformed with 35S:HAHB10 (TG-A, TG-B) or with pBI121 (121, control). Levels of SA and JA were quantified after infection with an avirulent (Vir, left panel) or virulent (Avir, right panel) strain of P. syringae. Phytohormone levels were determined in leaves at 0, 24, and 48 h post-infection in four independent samples (n=4, bars: ±SD). (C and D) JA and SA quantification was performed in Arabidopsis plants transformed with 35S:HAHB10 (TG-A, TG-B) or with pBI121 (121, control) after 0, 30, 60, 90, and 120 min of mechanical wounding. Phytohormone levels were determined in leaves in four independent samples (n=4, bars: ±SD). (E) Bacterial colony-forming units (CFU) in Arabidopsis plants transformed with 35S:HAHB10 (TG-A, TG-B) or pBI121 (121, control). Bacterial density was quantified 24 h (white bars) and 48 h (black bars) post-infection with a virulent (Vir) and an avirulent (Avir) strain of P. syringae. Each determination was performed in triplicate. (F) Mass gain of Spodoptera exigua larvae on HAHB10-expressing Arabidopsis transgenic plants (TG) or plants transformed with pBI121 (121, control). Fig. 6. View largeDownload slide The ectopic expression of HAHB10 modulates the synthesis of SA and JA in Arabidopsis plants. (A and B) SA and JA quantification was performed in Arabidopsis plants transformed with 35S:HAHB10 (TG-A, TG-B) or with pBI121 (121, control). Levels of SA and JA were quantified after infection with an avirulent (Vir, left panel) or virulent (Avir, right panel) strain of P. syringae. Phytohormone levels were determined in leaves at 0, 24, and 48 h post-infection in four independent samples (n=4, bars: ±SD). (C and D) JA and SA quantification was performed in Arabidopsis plants transformed with 35S:HAHB10 (TG-A, TG-B) or with pBI121 (121, control) after 0, 30, 60, 90, and 120 min of mechanical wounding. Phytohormone levels were determined in leaves in four independent samples (n=4, bars: ±SD). (E) Bacterial colony-forming units (CFU) in Arabidopsis plants transformed with 35S:HAHB10 (TG-A, TG-B) or pBI121 (121, control). Bacterial density was quantified 24 h (white bars) and 48 h (black bars) post-infection with a virulent (Vir) and an avirulent (Avir) strain of P. syringae. Each determination was performed in triplicate. (F) Mass gain of Spodoptera exigua larvae on HAHB10-expressing Arabidopsis transgenic plants (TG) or plants transformed with pBI121 (121, control). Quantification of bacterial density [colony-forming units (CFU)] in leaves of plants infected with an avirulent and a virulent strain of P. syringae showed that the avirulent strain grew at similar rates in control and transgenic plants whereas the virulent strain grew slightly faster (significant differences in CFU at 24 h); however, it reached similar CFU levels at 48 h of infection (Fig. 6E). Because Arabidopsis plants ectopically expressing HAHB10 also accumulated lower levels of JA after wounding compared with control plants, whether these plants were more susceptible to insect herbivores was also evaluated. First instar larvae of S. exigua were placed on transgenic and control plants and their gain in mass quantified every day for 6 d. The gain in mass of the larvae was similar between transgenic and control plants, indicating that the reduced levels of JA induced by the ectopic expression of HAHB10 did not affect the susceptibility of the plants to S. exigua larvae (Fig. 6F). Based on the effect of HAHB4 on ethylene levels (Manavella et al., 2008), the amount of ethylene released by Arabidopsis HAHB10-expressing plants and control plants was also quantified. The levels of this hormone were similar between genotypes (data not shown), indicating that, in contrast to HAHB4, ectopic expression of HAHB10 in Arabidopsis does not affect ethylene levels. Discussion The expression pattern of HAHB10 is consistent with the induction of early flowering HD-Zip transcription factors are usually expressed at very low levels; however, their expression increases as a result of specific external stimuli or internal signals (Ariel et al., 2007). It has been shown that the expression of HAHB10 is high in sunflower mature leaves, whereas it was almost undetectable in other tissues (Rueda et al., 2005). A more extensive analysis performed in this study indicated that the expression of HAHB10 in sunflower leaves was almost constitutive during the vegetative stage together with a lower expression of the newly identified HAAP1 and HAFT. During the reproductive stage, HAHB10 is expressed in stamen together with HASEP3 and HAFUL but not in the gynoecium, while HAAP1 transcripts disappear in both floral organs, as expected, and HASEP3 is still present in the gynoecium. Sunflower CF33 is an SD-responsive genotype while Arabidopsis Col 0 is LD responsive (de la Vega and Chapman, 2010). Interestingly, HAHB10 as well as HAFT and HAAP1 expression was strongly induced in apexes when sunflower plants were placed in an SD inductive condition while they remained almost constant in the non-inductive LD. In addition, when transgenic 35S:HAHB10 plants were grown in SD (non-inductive for Arabidopsis), HAHB10 did not induce flowering. Moreover, the phenotype was reverted in this condition. These observations indicate that HAHB10 action is photoperiod dependent, probably needing other photoperiod-dependent partners such as CO to exert its function. Similar observations were made for transcription factors belonging to the HAP family (Kumimoto et al., 2008). In general, the GUS expression pattern in leaves of promHAHB10:GUS plants was very similar to the reported expression pattern of CO, FLC, FT, and FUL in Arabidopsis (Mandel and Yanofski, 1995; Kardailsky et al., 1999; Teper-Bamnolker and Samach, 2005; Kim et al., 2008). Thus, the GUS expression pattern in promHAHB10:GUS plants was consistent with a potential function of HAHB10 in regulating flowering by affecting the expression of key flowering genes (Fig. 2A). Consistently, Arabidopsis transgenic plants transformed with a construct bearing this gene fused to its own promoter (promHAHB10:HAHB10), directing a tissue-specific expression, exhibited an early flowering and a shorter life cycle compared with control plants as was observed for constitutively HAHB10-expressing plants (35S:HAHB10) (Rueda et al., 2005). This early flowering could be explained based on the high activity of the HAHB10 promoter during the transition from the vegetative to the reproductive stage in the flower primordia, as it was observed in the promHAHB10:GUS transgenic plants. It is likely that the promoter responds to some environmental or internal signal and activates the transcription of HAHB10, thereby inducing the early flowering phenotype. Unfortunately, sunflower mutants (in this or other genes) are not available, precluding the analysis of the role of HAHB10 in flowering regulation in this plant species. Accelerated flowering was also observed in transgenic plants ectopically overexpressing the HAHB10 homologue, ATHB2/HAT4 (Schena et al., 1993), even though the phenotypic characteristics of ATHB2/HAT4 ectopic overexpression and knock-out plants indicate that these two genes should not be considered as orthologues (Rueda et al., 2005). However, another member of the Arabidopsis HD-Zip II family could be the actual orthologue of HAHB10. Based on the expression pattern, HAT22 seems to be the best candidate (A. L. Arce et al., unpublished results). Analysis of gene expression supports a role for HAHB10 in the induction of early flowering To understand the molecular mechanisms involved in the induction of flowering by HAHB10, the transcriptome of 35S:HAHB10 and control plants entering the flowering transition stage was compared. Several key genes involved in flowering initiation were found to be significantly up-regulated in 35S:HAHB10 plants, consistent with the early flowering phenotype induced by this transcription factor. For example, SEP3 and FUL were described as key genes in floral organogenesis (Mandel and Yanofski, 1995; Pelaz et al., 2001; Teper-Bamnolker and Samach, 2005). FT is considered the trigger of universal florigenic signals and regulates the flowering cycles in many plant species (Kardailsky et al., 1999; Huang et al., 2005; Lifschitz and Eshed, 2006; Lifschitz et al., 2006). SEP3 showed the highest induction in HAHB10-expressing plants, ∼30-fold, and harboured in its promoter the pseudopalindrome CAAT(A/T)ATTG, which binds HAHB10 in vitro (Tron et al., 2002). Hence, SEP3 may be a direct target of this HD-Zip. FT and FUL were also up-regulated in 35S:HAHB10 plants; however, these genes did not contain the CAAT(A/T)ATTG sequence in their promoter regions, suggesting that they are indirectly affected by HAHB10 expression. CO was slightly induced in the transgenic plants but to a lower level. In addition to flowering initiation genes, others, known to be involved in the photoperiod pathway, were also up-regulated. In agreement with the results obtained with Arabidopsis, putative sunflower homologues of Arabidopsis FT and SEP3 (HAFT and HASEP3) were strongly induced when HAHB10 was overexpressed in leaf discs. The high expression levels of FT, SEP3, and FUL were most probably sufficient to induce early flowering in Arabidopsis ectopically expressing HAHB10, since their overexpression generates a similar phenotype in this plant species (Kardailsky et al., 1999; Pelaz et al., 2001; Jaeger and Wigge, 2007). The microarray analysis also revealed that transcript levels of Arabidopsis HD-Zip II members homologous to HAHB10 were reduced in the HAHB10 transgenic plants (Table 4). These results were in accordance with the already described negative autoregulation of HD-Zip II members (Ohgishi et al., 2001; Ciarbelli et al., 2008) and further suggested that HAHB10 was recognized in Arabidopsis as a functional HD-Zip II member. Regarding these results indicating that HAHB10 is involved both in flowering and in the defence response, HAHB10 expression was analysed when sunflower plants were treated with exogenous SA, infected with P. syringae, or subjected to wounding. The expression of this gene was induced by SA, presenting similar kinetics to those of SA induction of HAFT, while HASEP3 transcripts remained almost constant 72 h after the treatment. This last result is not in accordance with the up-regulation of HASEP3 observed in transiently HAHB10-transformed leaf discs, suggesting that the concentration of HAHB10 reached 72 h after SA treatment was not enough to induce HASEP3 expression or, alternatively, that HASEP3 is induced by HAHB10 via an SA-independent pathway. The accumulation of SA and JA is affected in Arabidopsis plants constitutively expressing HAHB10 after infection with P. syringae and wounding Several defence-related genes were down-regulated in 35S:HAHB10 plants (Table 3). Among these genes were ICS1 and EDS5, involved in SA synthesis and signalling, respectively, and PR1 and PDF1.2a (Fig. 4). Moreover, the sunflower homologues of PR1, PR3, and LOX2 (HAPR1, HAPR3, and HALOX2, respectively) were repressed at the mRNA level when HAHB10 was transiently over-expressed in sunflower leaf discs. This general down-regulation of SA- and JA-dependent defence-related genes could be associated with a negative role for HAHB10 in the regulation of SA and JA biosynthesis or their induced defence responses. Plants ectopically expressing HAHB10 exhibited the same basal SA levels as their controls but accumulated more SA after a compatible interaction with P. syringae or wounding. Virulent bacteria showed an initial accelerated growth on HAHB10-expressing plants, in agreement with the basal reduced expression of some SA-responsive genes in these plants. However, after 48 h the bacterial density was similar to that of control plants, suggesting that the activation of SA-mediated responses was not impaired in HAHB10-expressing plants (consistent with the wild-type levels of activation of PR1 gene expression after application of exogenous SA in this genotype). The increased accumulation of SA after infection could be the result of a compensatory effect for the reduced basal defence gene expression or a more direct de-regulation of SA biosynthesis or metabolism in HAHB10-expressing plants. The actual mechanism remains at present unknown. Prithiviraj et al. (2005) showed that SA is able to attenuate P. aeuroginosa virulence via the transcriptional repression of exoproteins and other virulence factors; however, this hormone did not inhibit bacterial growth. The lower than control levels of JA (and JA–Ile) quantified in wounded and infected leaves of 35S:HAHB10 plants could be the result of a negative effect of the increased SA levels (Fig. 6). These reduced JA levels did not, however, affect the growth of larvae of the folivorous insect S. exigua. A related, however opposite, mechanism was described previously for the sunflower HAHB4 (Manavella et al., 2008); Arabidopsis plants ectopically expressing this transcription factor accumulated higher levels of JA and ethylene after wounding but reduced levels of SA after bacterial infection. Plants expressing HAHB4 were more and less resistant to bacterial pathogens and insect herbivores, respectively (Manavella et al., 2008). These results indicated that both HAHB10 and HAHB4 participate in the control of phytohormone synthesis and in signalling pathways affecting biotic stress responses (this study and Manavella et al., 2006, 2008). Conclusion The results presented in this study indicated that HAHB10 plays two different roles in plants. It induces the transition from vegetative to flowering stages via the activation of specific flowering transition genes in a photoperiod-dependent way and it affects the accumulation of SA and JA during a compatible interaction with P. syringae and wounding. A proposed model schematizing the participation of this transcription factor in these pathways is represented in Fig. 7. Whether the role of HAHB10 in early flowering is associated with the regulation of SA biosynthesis/signalling remains a possible scenario and it is the topic of future work. Moreover, future experiments will also investigate the mechanisms induced by HAHB10 that affect phytohormone accumulation during the wound response and defence against virulent P. syringae. Fig. 7. View largeDownload slide Schematic model illustrating the role of HAHB10 in flowering and biotic stress response. (→) indicates induction while (—|) indicates repression. Fig. 7. View largeDownload slide Schematic model illustrating the role of HAHB10 in flowering and biotic stress response. (→) indicates induction while (—|) indicates repression. Abbreviations Abbreviations ABA abscisic acid AP1 APETALLA1 CO CONSTANS ET ethylene FT flowering time FUL FRUITFULL HAHB10 Helianthus annuus homeobox 10 JA jasmonic acid LD long day regime SA salicylic acid SD short day regime SEP3 SEPALLATA3 This work was supported by ANPCyT (PICT 2005 38103 and PICT-PAE 37100) and UNL. CAD and RLC are members of CONICET; PAM was a Fellow of the same Institution; DR is a Fellow of CONICET and DAAD; JIG was a Fellow of Foncyt (PICT 38103) and is currently a CONICET Fellow; GB and ITB are funded by the Max Planck Society. MA-F is supported by grants from CNPq (310254/2007-8) and FAPERJ (E-26/102.861/2008). References Al-Shahrour F,  Mínguez P,  Tárraga J,  Montaner D,  Alloza E,  Vaquerizas JMM,  Conde L,  Blaschke C,  Vera J,  Dopazo J.  BABELOMICS: a systems biology perspective in the functional annotation of genome-scale experiments,  Nucleic Acids Research ,  2006, vol.  34 (pg.  W472- W476) Google Scholar CrossRef Search ADS PubMed  Alves-Ferreira M,  Wellmer F,  Banhara A,  Kumar V,  Riechmann JL,  Meyerowitz EM.  Global expression profiling applied to the analysis of Arabidopsis stamen development,  Plant Physiology ,  2007, vol.  145 (pg.  747- 762) Google Scholar CrossRef Search ADS PubMed  Anderson JP,  Badruzsaufari E,  Schenk PM,  Manners JM,  Desmond OJ,  Ehlert C,  Maclean DJ,  Ebert PR,  Kazan K.  Antagonistic interaction between abscisic acid and jasmonate–ethylene signalling pathways modulates defense gene expression and disease resistance in Arabidopsis,  The Plant Cell ,  2004, vol.  16 (pg.  3460- 3479) Google Scholar CrossRef Search ADS PubMed  Ariel FD,  Manavella PA,  Dezar CA,  Chan RL.  The true story of the HD-Zip family,  Trends in Plant Science ,  2007, vol.  12 (pg.  419- 426) Google Scholar CrossRef Search ADS PubMed  Balbi V,  Devoto A.  Jasmonate signalling network in Arabidopsis thaliana: crucial regulatory nodes and new physiological scenarios,  New Phytologist ,  2008, vol.  177 (pg.  301- 318) Google Scholar CrossRef Search ADS PubMed  Carmel-Goren L,  Liu YS,  Lifschitz E,  Zamir D.  The SELF-PRUNING gene family in tomato,  Plant Molecular Biology ,  2003, vol.  52 (pg.  1215- 1222) Google Scholar CrossRef Search ADS PubMed  Chan RL,  Gago GM,  Palena CM,  Gonzalez DH.  Homeoboxes in plant development,  Biochimica et Biophysica Acta ,  1998, vol.  1442 (pg.  1- 19) Google Scholar CrossRef Search ADS PubMed  Ciarbelli AR,  Ciolfi A,  Salvucci S,  Ruzza V,  Possenti M,  Carabelli M,  Fruscalzo A,  Sessa G,  Morelli G,  Ruberti I.  The Arabidopsis homeodomain-leucine zipper II gene family: diversity and redundancy,  Plant Molecular Biology ,  2008, vol.  68 (pg.  465- 478) Google Scholar CrossRef Search ADS PubMed  Cleland CF,  Ajami A.  Identification of the flower-inducing factor isolated from Aphid Honeydew as being salicylic scid,  Plant Physiology ,  1974, vol.  54 (pg.  904- 906) Google Scholar CrossRef Search ADS PubMed  Clough SJ,  Bent AF.  Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana,  The Plant Journal ,  1998, vol.  16 (pg.  735- 743) Google Scholar CrossRef Search ADS PubMed  de la Vega AJ,  Chapman SC.  Mega-environment differences affecting genetic progress for yield and relative value of component traits,  Crop Science ,  2010, vol.  50 (pg.  574- 583) Google Scholar CrossRef Search ADS   Despres C,  DeLong C,  Glaze S,  Liu E,  Fobert PR.  The Arabidopsis NPR1/NIM1 protein enhances the DNA binding activity of a subgroup of the TGA family of bZIP transcription factors,  The Plant Cell ,  2000, vol.  12 (pg.  279- 290) Google Scholar CrossRef Search ADS PubMed  Höfgen R,  Willmitzer L.  Storage of competent cells for Agrobacterium transformation,  Nucleic Acids Research ,  1988, vol.  16 pg.  9977  Google Scholar CrossRef Search ADS   Huang T,  Böhlenius H,  Eriksson S,  Parcy F,  Nilsson O.  The mRNA of the Arabidopsis gene FT moves from leaf to shoot apex and induces flowering,  Science ,  2005, vol.  309 (pg.  1694- 1696) Google Scholar CrossRef Search ADS PubMed  Jaeger KE,  Wigge PA.  FT protein acts as a long-range signal in Arabidopsis,  Current Biology ,  2007, vol.  17 (pg.  1050- 1054) Google Scholar CrossRef Search ADS PubMed  Jefferson RA,  Kavanagh TA,  Bevan MW.  GUS fusions: beta-glucuronidase as a sensitive and versatile gene fusion marker in higher plants,  EMBO Journal ,  1987, vol.  6 (pg.  3901- 3907) Google Scholar PubMed  Kardailsky I,  Shukla VK,  Ahn JH,  Dagenais N,  Christensen SK,  Nguyen JT,  Chory J,  Harrison MJ,  Weigel D.  Activation tagging of the floral inducer FT,  Science ,  1999, vol.  286 (pg.  1962- 1965) Google Scholar CrossRef Search ADS PubMed  Katagiri F,  Thilmony R,  He SY.  Somerville CR,  Meyerowitz EM.  The Arabidopsis thaliana–Pseudomonas syringae interaction,  The Arabidopsis mok. American Society of Plant Biologists ,  2002  Rockville, MD, doi: 10.1199/tab.0039: www.aspb.org/publications/arabidopsis/ Khurana JP,  Cleland CF.  Role of salicylic acid and benzoic acid in flowering of a photoperiod-insensitive strain, Lemna paucicostata LP6,  Plant Physiology ,  1992, vol.  100 (pg.  1541- 1546) Google Scholar CrossRef Search ADS PubMed  Kim SY,  Yu X,  Michaels SD.  Regulation of CONSTANS and FLOWERING LOCUS T expression in response to changing light quality,  Plant Physiology ,  2008, vol.  148 (pg.  269- 279) Google Scholar CrossRef Search ADS PubMed  Kumimoto R,  Adam L,  Hymus G,  Repetti P,  Lynne Reuber T,  Marion C,  Hempel F,  Ratcliffe O.  The nuclear factor Y subunits NF-YB2 and NF-YB3 play additive roles in the promotion of flowering by inductive long-day photoperiods in Arabidopsis,  Planta ,  2008, vol.  228 (pg.  709- 723) Google Scholar CrossRef Search ADS PubMed  Levy YY,  Dean C.  The transition to flowering,  The Plant Cell ,  1998, vol.  10 (pg.  1973- 1990) Google Scholar CrossRef Search ADS PubMed  Lifschitz E,  Eshed Y.  Universal florigenic signal triggered by FT homologues regulate growth and flowering cycles in perennial day neutral tomato,  Journal of Experimental Botany ,  2006, vol.  57 (pg.  3405- 3414) Google Scholar CrossRef Search ADS PubMed  Lifschitz E,  Eviatar T,  Rozman A,  Shalit A,  Goldshmidt A,  Amsellem Z,  Alvarez JP,  Eshed Y.  The tomato FT ortholog triggers systemic signals that regulate growth and flowering and substitute for diverse environmental stimuli,  Proceedings of the National Academy of Sciences, USA ,  2006, vol.  103 (pg.  6398- 6403) Google Scholar CrossRef Search ADS   Loake G,  Grant M.  Salicylic acid in plant defence—the players and protagonists. Current Opinion in Plant Biology,  2007, vol.  10 (pg.  466- 472) Lorenzo O,  Piqueras R,  Sanchez-Serrano JJ,  Solano R.  ETHYLENE RESPONSE FACTOR1 integrates signals from ethylene and jasmonate pathways in plant defense,  The Plant Cell ,  2003, vol.  14 (pg.  165- 178) Google Scholar CrossRef Search ADS   Manavella PA,  Arce AL,  Dezar CA,  Bitton F,  Renou JP,  Crespi M,  Chan RL.  Cross-talk between ethylene and drought signalling pathways is mediated by the sunflower Hahb-4 transcription factor,  The Plant Journal ,  2006, vol.  48 (pg.  125- 137) Google Scholar CrossRef Search ADS PubMed  Manavella PA,  Dezar CA,  Bonaventure G,  Baldwin IT,  Chan RL.  HAHB4, a sunflower HD-Zip protein, integrates signals from the jasmonic acid and ethylene pathways during wounding and biotic stress responses,  The Plant Journal ,  2008, vol.  56 (pg.  376- 388) Google Scholar CrossRef Search ADS PubMed  Manavella PA,  Chan RL.  Transient transformation of sunflower leaf discs via an Agrobacterium-mediated method: applications for gene expression and silencing studies,  Nature Protocols ,  2009, vol.  4 (pg.  1699- 1707) Google Scholar CrossRef Search ADS PubMed  Mandel MA,  Yanofsky MF.  The Arabidopsis AGL8 MADS box gene is expressed in inflorescence meristems and is negatively regulated by APETALA1,  The Plant Cell ,  1995, vol.  7 (pg.  1763- 1771) Google Scholar CrossRef Search ADS PubMed  Martinez C,  Pons E,  Prats G,  Leon J.  Salicylic acid regulates flowering time and links defence responses and reproductive development,  The Plant Journal ,  2004, vol.  37 (pg.  209- 217) Google Scholar CrossRef Search ADS PubMed  Mayda E,  Tornero P,  Conejero V,  Vera P.  A tomato homeobox gene (HD-Zip) is involved in limiting the spread of programmed cell death,  The Plant Journal ,  1999, vol.  20 (pg.  591- 600) Google Scholar CrossRef Search ADS PubMed  O'Donnell PJ,  Calvert C,  Atzorn R,  Wasternack C,  Leyser HMO,  Bowles DJ.  Ethylene as a signal mediating the wound response of tomato plants,  Science ,  1996, vol.  274 (pg.  1914- 1917) Google Scholar CrossRef Search ADS PubMed  Ohgishi M,  Oka A,  Morelli G,  Ruberti I,  Aoyama T.  Negative autoregulation of the Arabidopsis homeobox gene ATHB-2,  The Plant Journal ,  2001, vol.  24 (pg.  389- 398) Google Scholar CrossRef Search ADS   Park S-W,  Kaimoyo E,  Kumar D,  Mosher S,  Klessig DF.  Methyl salicylate is a critical mobile signal for plant systemic acquired resistance,  Science ,  2007, vol.  318 (pg.  113- 116) Google Scholar CrossRef Search ADS PubMed  Pelaz S,  Gustafson-Brown C,  Kohalmi SE,  Crosby WL,  Yanofsky MF.  APETALA1 and SEPALLATA3 interact to promote flower development,  The Plant Journal ,  2001, vol.  26 (pg.  385- 394) Google Scholar CrossRef Search ADS PubMed  Penninckx IAMA,  Thomma BPHJ,  Buchala A,  Metraux JP,  Broekaert WF.  Concomitant activation of jasmonate and ethylene response pathways is required for induction of a plant defensin gene in Arabidopsis,  The Plant Cell ,  1998, vol.  10 (pg.  2103- 2114) Google Scholar CrossRef Search ADS PubMed  Prithiviraj B,  Bais HP,  Weir T,  Suresh B,  Najarro EH,  Dayakar BV,  Schweizer HP,  Vivanco JM.  Down regulation of virulence factors of Pseudomonas aeruginosa by salicylic acid attenuates its virulence on Arabidopsis thaliana and,  Caenorhabditis elegans. Infection and Immunity ,  2005, vol.  73 (pg.  5319- 5328) Google Scholar CrossRef Search ADS   Rueda EC,  Dezar CA,  Gonzalez DH,  Chan RL.  Hahb-10, a sunflower homeobox-leucine zipper gene, is regulated by light quality and quantity, and promotes early flowering when expressed in Arabidopsis,  Plant and Cell Physiology ,  2005, vol.  46 (pg.  1954- 1963) Google Scholar CrossRef Search ADS PubMed  Schena M,  Lloyd AM,  Davis RW.  The HAT4 gene of Arabidopsis encodes a developmental regulator. Genes and,  Development ,  1993, vol.  7 (pg.  367- 379) Schneiter AA,  Miller JF.  Description of sunflower growth stages,  Crop Science ,  1981, vol.  21 (pg.  901- 903) Google Scholar CrossRef Search ADS   Sorin C,  Salla-Martret M,  Bou-Torrent J,  Roig-Villanova I,  Martínez-García JF.  ATHB4, a regulator of shade avoidance, modulates hormone response in Arabidopsis seedlings,  The Plant Journal ,  2009, vol.  59 (pg.  266- 277) Google Scholar CrossRef Search ADS PubMed  Spoel SH,  Koornneef A,  Claessens SMC, et al.  NPR1 modulates cross-talk between salicylate- and jasmonate-dependent defense pathways through a novel function in the cytosol,  The Plant Cell ,  2003, vol.  14 (pg.  760- 770) Google Scholar CrossRef Search ADS   Steindler C,  Matteucci A,  Sessa G,  Weimar T,  Ohgishi M,  Aoyama T,  Morelli G,  Ruberti I.  Shade avoidance responses are mediated by the ATHB-2 HD-zip protein, a negative regulator of gene expression,  Development ,  1999, vol.  126 (pg.  4235- 4245) Google Scholar PubMed  Teper-Bamnolker P,  Samach A.  The flowering integrator FT regulates SEPALLATA3 and FRUITFULL accumulation in Arabidopsis leaves,  The Plant Cell ,  2005, vol.  17 (pg.  2661- 2675) Google Scholar CrossRef Search ADS PubMed  Tron AE,  Bertoncini CW,  Chan RL,  González DH.  Redox regulation of plant homeodomain transcription factors,  Journal of Biological Chemistry ,  2002, vol.  277 (pg.  34800- 34807) Google Scholar CrossRef Search ADS PubMed  von Dahl CC,  Winz RA,  Halitschke R,  Kuhnemann F,  Gase K,  Baldwin IT.  Tuning the herbivore-induced ET burst: the role of transcript accumulation and ET perception in Nicotiana attenuata,  The Plant Journal ,  2007, vol.  51 (pg.  293- 307) Google Scholar CrossRef Search ADS PubMed  Weigel RR,  Pfitzner UM,  Gatz C.  Interaction of NIMIN1 with NPR1 modulates PR gene expression in Arabidopsis,  The Plant Cell ,  2005, vol.  17 (pg.  1279- 1291) Google Scholar CrossRef Search ADS PubMed  Wildermuth MC,  Dewdney J,  Wu G,  Ausubel FM.  Isochorismate synthase is required to synthesize salicylic acid for plant defence,  Nature ,  2001, vol.  414 (pg.  562- 565) Google Scholar CrossRef Search ADS PubMed  Yoon J,  Chung W-I,  Choi D.  NbHB1, Nicotiana benthamiana homeobox 1, is a jasmonic acid-dependent positive regulator of pathogen-induced plant cell death,  New Phytologist ,  2009, vol.  184 (pg.  71- 84) Google Scholar CrossRef Search ADS PubMed  Xu Y,  Chang PFL,  Liu D,  Narasimhan ML,  Raghothama KG,  Hasegawa PM,  Bressan RA.  Plant defense genes are synergistically induced by ethylene and methyl jasmonate,  The Plant Cell ,  1994, vol.  6 (pg.  1077- 1085) Google Scholar CrossRef Search ADS PubMed  © The Author [2010]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: [email protected]

Gambino, Giorgio;Minuto, Martina;Boccacci, Paolo;Perrone, Irene;Vallania, Rosalina;Gribaudo, Ivana

doi: 10.1093/jxb/erq349pmid: 21127025

Abstract Different cultivars of Vitis vinifera vary in their potential to form embryogenic tissues. The WUSCHEL (WUS)-related homeobox (WOX) genes have been shown to play an important role in coordinating the gene transcription involved in the early phases of embryogenesis. The expression dynamics of 12 VvWOX genes present in the V. vinifera genome in embryogenic and other tissues of ‘Chardonnay’ were analysed. In order to understand the influence of WOX genes on the somatic embryogenic process, their expression profiles were compared in two cultivars of V. vinifera (‘Chardonnay’ and ‘Cabernet Sauvignon’) that show different aptitudes for embryogenesis. The expression of all VvWOX genes was influenced by culture conditions. VvWOX2 and VvWOX9 were the principal WOX genes expressed during the somatic embryogenesis process, and the low aptitude for embryogenesis of ‘Cabernet Sauvignon’ was generally correlated with the low expression levels of these VvWOX genes. VvWOX3 and VvWOX11 were strongly activated in correspondence to torpedo and cotyledonary stages of somatic embryos, with low expression in the earlier developmental stages (pre-embryogenic masses and globular embryos) and during embryo germination. VvWOX genes appeared to be key regulators of somatic embryogenesis in grapevine, and the regulation of these genes during early phases of somatic embryogenesis differed between the two cultivars of the same species. Grapevine, phytohormones, qRT-PCR, somatic embryogenesis, WUSCHEL (WUS)-related homeobox genes Introduction Somatic embryogenesis is the initiation of embryos from plant somatic tissues. This morphological event occurs spontaneously in some plant species and is usually induced in tissue cultures, potentially from almost any part of the plant body, either directly from the explants or after the callus stage (Martinelli and Gribaudo, 2009). Somatic embryogenesis offers an attractive model system for studies on embryo development: somatic embryos (SEs) follow a developmental pathway very similar to that of their zygotic counterparts (Dodeman et al., 1997) and are more accessible than zygotic embryos. Somatic embryogenesis was first developed in grape in the 1970s, and today is one of the most suitable tools for the application of in vitro manipulation in the Vitis genus (Martinelli and Gribaudo, 2009). Somatic embryogenesis in grapevine is affected by many factors such as genotype, explant type, and culture conditions. Anthers are the most widely used explants, while ovaries and whole flowers have also proven to be suitable material for somatic embryogenesis in several cultivars of Vitis vinifera (Gambino et al., 2007). Somatic embryogenesis is a hormone-dependent process, and in grapevine, as well as in other species (Mordhorst et al., 1997), the combination of auxins—particularly 2,4-dichloro-phenoxyacetic acid (2,4-D)—and cytokinins—most frequently 6-benzyl-adenine (BA)—is essential for induction of SE differentiation. Different cultivars of V. vinifera vary relatively widely in their potential to form embryogenic tissues, and although several protocols have been published over the years, for some cultivars the method still needs improvement. Moreover, the molecular mechanisms that cause a plant cell to change its fate and become embryogenic are not well understood. In grapevine, Schellenbaum et al. (2008) and Maillot et al. (2009) recently demonstrated that the VvSERK and VvL1L genes are involved in somatic embryogenesis. The WUSCHEL (WUS)-related homeobox (WOX) gene family is a class of homeodomain (HD) transcription factors that are involved in the early phases of embryogenesis and lateral organ development in plants (Haecker et al, 2004). WUS, the founding member of the WOX gene family, which consists of at least 15 members in the Arabidopsis thaliana genome (Haecker et al., 2004), is required for stem maintenance of the Arabidopsis shoot (Mayer et al., 1998). WOX5 performs a similar function in the root apical meristem (Sarkar et al., 2007). WOX3 is involved in the development of lateral sepals and stamens in the flower (Nardmann et al., 2004), while WOX6 is required for ovule development (Park et al., 2005). WOX2, WOX8, and WOX9 are important cell fate regulators of early pre-embryos (Haecker et al., 2004; Breuninger et al., 2008); WOX9 is also involved in other aspects of development, such as the regulation of WUS expression in the shoot apical meristem (Wu et al., 2005). In animals, the HD contains 60 amino acids, while in plants all the WOX proteins have an HD of 65/66 amino acids. The WOX genes were first characterized in Arabidopsis (Haecker et al., 2004), but in recent years their function has been analysed in several species (Nardmann et al., 2007; Imin et al., 2007; Jain et al., 2008; Palovaara and Hakman, 2008). Recently Vandenbussche et al. (2009) carried out a phylogenetic analysis of the WOX proteins that are present in several fully sequenced plant genomes, and through a bioinformatics approach identified 12 WOX genes in the sequences of two grapevine genotypes (Jaillon et al., 2007; Velasco et al., 2007). In this work, the 12 VvWOX genes identified in V. vinifera were cloned and sequenced, and their expression dynamics were analysed in embryogenic and other tissues of the cultivar Chardonnay. In order to understand the influence of WOX genes on the somatic embryogenic process, their expression profiles were compared in embryogenic tissues of two cultivars of V. vinifera (‘Chardonnay’ and ‘Cabernet Sauvignon’) that have different aptitudes for embryogenesis. The results show that the VvWOX genes are important key regulators of somatic embryogenesis in grapevine, and demonstrate that the regulation of WOX genes during the early phase of somatic embryogenesis differs between two cultivars of the same species. Materials and methods Plant materials Embryogenic tissues were induced to form from immature whole flowers, anthers, and ovaries of two V. vinifera cultivars ‘Chardonnay’ and ‘Cabernet Sauvignon’. Over a 2 week period during spring (May 2009), flower clusters (Fig. 1A) were field collected from vineyards in North-Western Italy. The developmental stage of explants was preliminarily determined by observing the flowers and anthers under a stereomicroscope and examining the stage of microsporogenesis under an optical microscope after anthers were squashed in a drop of acetocarmine (Gribaudo et al., 2004). The inflorescences were surface sterilized for 15 min with sodium hypochlorite (1.5% available chlorine), and rinsed several times with sterile distilled water. Whole flowers (Fig. 1C) were removed under a laminar flow hood from the inflorescence by cutting the pedicels; stamens (anthers plus filaments) (Fig. 1B) and pistils (ovaries plus styles, stigmas, and receptacles) (Fig. 1D) were excised from flowers under a stereomicroscope. Below these are referred to simply as anthers and ovaries. Explants were cultured on a previously described callus induction medium (Gribaudo et al., 2004; Gambino et al., 2007) supplemented with 4.5 μM 2,4-D and 8.9 μM BA. The cultures were maintained at 26 °C in the dark. Three months after the initiation of the culture, the calli were classified as embryogenic (EC; Fig. 1E–J) or non-embryogenic (NEC; Fig. 1K–N) under a stereomicroscope. EC and NEC were transferred to an embryo proliferation medium supplemented with 10 μM 2-naphthoxyacetic acid (NOA), 1 μM BA, and 20 μM indole-3-acetic acid (IAA) (Gambino et al., 2007). The number of explants differentiating somatic embryos was recorded 3 and 5 months after culture initiation. The final percentage data were arcsin transformed and then subjected to analysis of variance (GLM procedure; SAS statistical software, version 8.2, SAS Institute, Cary, NC, USA). Fig. 1. View largeDownload slide Explants analysed for the VvWOX gene expression profile. Immature inflorescence (A), whole flower (C), stamens (in the text we refer to them as anthers: B) and pistil (in the text we refer to this as ovary: D) of ‘Chardonnay’ collected in May. Embryogenic calli (EC) after 3 months of culture from anthers (E, H, J), flowers (G), and ovaries (I) of ‘Chardonnay’. EC after 3 months of culture from anthers (F) of ‘Cabernet Sauvignon’. Non-embryogenic calli (NEC) from anthers of ‘Chardonnay’ (K, N) and ‘Cabernet Sauvignon’ (L, M). se, somatic embryos; pe, pre-embryogenic tissues. Size bar=1 mm. Fig. 1. View largeDownload slide Explants analysed for the VvWOX gene expression profile. Immature inflorescence (A), whole flower (C), stamens (in the text we refer to them as anthers: B) and pistil (in the text we refer to this as ovary: D) of ‘Chardonnay’ collected in May. Embryogenic calli (EC) after 3 months of culture from anthers (E, H, J), flowers (G), and ovaries (I) of ‘Chardonnay’. EC after 3 months of culture from anthers (F) of ‘Cabernet Sauvignon’. Non-embryogenic calli (NEC) from anthers of ‘Chardonnay’ (K, N) and ‘Cabernet Sauvignon’ (L, M). se, somatic embryos; pe, pre-embryogenic tissues. Size bar=1 mm. SEs at the torpedo stage (SE1) of ‘Chardonnay’ were isolated from the EC and transferred to a germination medium without plant growth regulators (PGRs). To obtain somatic embryos in germination (SEG), embryos at the cotyledonary stage (SE2) were maintained on the germination medium but transferred to a growth chamber at 24 °C with a 16 h photoperiod. Shoot apexes (SAs), mature leaves (MLs), and roots (Rs) were collected from 3-year-old greenhouse-grown potted plants of V. vinifera ‘Chardonnay’. Phloem scraped from dormant woody canes (PWs) and immature inflorescences (IFs) were collected from plants of field-grown ‘Chardonnay’. Root tips (TRs) were isolated from in vitro plantlets of ‘Chardonnay’ micropropagated by culturing apical cuttings on a PGR-free medium in a growth chamber at 24 °C with a 16 h photoperiod. Bioinformatics analyses and cDNA cloning The putative VvWOX sequences were retrieved from the two available draft genome sequences of grapevine [PN40024: http://www.cns.fr/spip/Vitis-vinifera-whole-genome.html (Jaillon et al., 2007); ‘Pinot Noir’ ENTAV115: http://genomics.research.iasma.it/cgi-bin/gbrowse/dasrelease3/ (Velasco et al., 2007)] and the National Centre for Biotechnology Information database (NCBI, http://www.ncbi.nlm.nih.gov/) using the AtWOX proteins from Arabidopsis (Haecker et al., 2004) as the search term. The nomenclature of WOX sequences proposed by Vandenbussche et al. (2009) on the basis of sequence homology with members of the Arabidopsis WOX family was maintained. Primers used for preliminary end-point RT-PCR analyses and cDNA cloning (Supplementary Table S1 at JXB online) were designed by means of the software Primer 3 (Rozen and Skaletsky, 2000) (http://frodo.wi.mit.edu/primer3/) in correspondence to DNA sequences outside the HD, these being the most conserved region within the WOX genes. Since the coding sequences of VvWOX13 genes were very similar, for these genes the primers were designed to correspond with untranslated regions. To ensure that the primer sequence was unique for a given gene, similarity to other known sequences was checked using BLAST (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi). Total RNA was extracted from IFs, SE2, MLs, Rs, and PWs of ‘Chardonnay’ following the protocol described in Gambino et al. (2008). RNA purity and concentration were assessed by determining the spectrophotometric absorbance of the samples at 230, 260, and 280 nm and ratios of A260/A280 and A260/A230. RNA quality was evaluated by 1% formaldehyde–agarose gel electrophoresis and visualization by UV light after staining with ethidium bromide. First-strand cDNA synthesis was performed following the protocol previously described by Gambino and Gribaudo (2006). Cycling conditions for all primer pairs (Supplementary Table S1 at JXB online) consisted of initial denaturation at 94 °C for 2 min, followed by 35 cycles at 94 °C for 30 s, 63 °C for 45 s, and 72 °C for 1 min. Reaction products were analysed by electrophoresis on 1.5% agarose gels buffered in TBE (45 mM Tris-borate, 1 mM EDTA), and visualized by UV light after staining with ethidium bromide. Amplified RT-PCR products were gel-purified by the Wizard® SV Gel and PCR Clean-Up System (Promega) and cloned into the pGEM-T Easy vector (Promega). The plasmid DNA was isolated by the Wizard Plus SV Minipreps DNA Purification System (Promega) following the Promega protocol and sequenced using M13 forward and reverse primers. Two independent clones from each RT-PCR product were sequenced using a Big-Dye Terminator v1.1 Cycle Sequencing kit (Applied Biosystems), following the manufacturer's instructions. PCR products were purified using an AutoSeq G-50 Dye Terminator Removal kit (GE Healthcare) and analysed using a 3130 Genetic Analyzer capillary sequencer (Applied Biosystems). Phylogenetic analysis After the sequencing of V. vinifera WOX genes, the protein sequences deduced through the ExPasy translate program (http://expasy.org/tools/dna.html) (Supplementary Table S2 at JXB online) were used for a phylogenetic analysis of the WOX family. The WOX proteins in the fully sequenced genomes of Arabidopsis, Populus trichocarpa, and Oryza sativa, and the WOX sequences of Zea mays, Solanum lycopersicum, Picea abies, Pinus sitchensis, Pinus sylvestris, and Citrus sinensis available in the NCBI database were analysed. All accession numbers are listed in Supplementary Table S3. When available, the original nomenclature of WOX proteins was maintained [Nardmann et al. (2007) and Vandenbussche et al. (2009) for P. trichocarpa; Haecker et al. (2004) for Arabidopsis; Nardmann et al. (2007) for 12 of 18 WOX proteins of Z. mays and for 9 of 13 proteins of O. sativa; Palovaara and Hakmann (2008) for P. abies]; for the remaining proteins the sequences was named on the basis of the phylogenetic position. The protein sequences were aligned with the ClustalW 1.82 software (http://www.ebi.ac.uk/Tools/clustalw2/index.html) and the CLC Sequence Viewer 6.3 software (http://www.clcbio.com/index.php?id=1052) using default settings. The phylogenetic analysis based on the Neighbor–Joining (NJ) method was carried out on the HD region and on complete protein sequences using the CLC Sequence Viewer 6.3 software. To assess support for the relationship, 1000 bootstrap samples were generated. Relative real-time qRT-PCR Total RNA was extracted following the protocol previously described (Gambino et al., 2008) from anthers, ovaries, and flowers of ‘Chardonnay’ and ‘Cabernet Sauvignon’ collected from different inflorescences at culture initiation (May 2009). After 40 d of culture (June 2009) ovaries and flowers were sampled, but not anthers, as, due to their light weight, such sampling would have reduced the remaining cultures too much. At least five different EC (whole calli, including pre-embryogenic masses and SEs from globular to heart stages) and NEC, each generated from single anthers, ovaries, or flowers, were sampled 3 months after culture initiation. SE1 of ‘Chardonnay’ were sampled at the time of their transfer to the PGR-free medium, SE2 of ‘Chardonnay’ at the time of their transfer to the 16 h photoperiod, and SEG of ‘Chardonnay’ 1 month after transfer to the 16 h photoperiod when the shoot apex was visible between the cotyledons. Primers for qRT-PCR (Supplementary Table S1 at JXB online) were designed outside the HD using the software Primer Express 3.0 (Applied Biosystems) in correspondence to sequenced regions. For the qRT-PCR, first-strand cDNA synthesis was performed in duplicate using 5 μg of total RNA treated with DNase and the High Capacity cDNA Reverse Transcription kit (Applied Biosystems). Three endogenous housekeeping genes, actin (ACT), ubiquitin (UBI), and glyceraldehyde-3-phosphate-dehydrogenase (GAPDH; primers from Reid et al., 2006), were considered for use in qRT-PCR after analysis of gene expression stability in the geNorm software (http://medgen.ugent.be/∼jvdesomp/genorm/). The relative expression was computed based on the comparative Ct (2–ΔΔCt) method as described by Livak and Schmittgen (2001). The geometric mean of the expression ratios of the two most stable housekeeping genes was used as the normalization factor in all samples. The PCR mix (20 μl) contained 10 μl of PowerSYBR Green master mix (Applied Biosystems), 0.25 μM of each primer (Supplementary Table S1), and 2 μl of cDNA diluted 1:10. Cycling conditions for all primer pairs consisted of initial denaturation at 95 °C for 10 min, followed by 40 cycles at 95 °C for 15 s, and 60 °C for 1 min. PCR was performed in triplicate, and specific annealing of the primers was controlled on dissociation kinetics performed at the end of each PCR run. Gene expression was expressed as the mean and standard error calculated over all biological and technical replicates. In situ hybridization EC originating from anthers of ‘Chardonnay’ were collected 3 months after culture initiation. Before the paraffin embedding of samples for histological sections, the friable calli were embedded in 3% agar to preserve the original morphology of the calli and the cell arrangement within the samples. Immature flowers of ‘Chardonnay’ were collected during spring (May 2009). Samples were then fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS; 130 mM NaCl; 7 mM Na2HPO4; 3 mM NaH2PO4, pH 7.4) overnight at 4 °C. For the first 30–40 min, samples were fixed under vacuum to facilitate infiltration with the fixative. Thereafter, the fixative was removed by washing in saline solution (150 mM NaCl) for 15 min and the tissues were dehydrated successively in solutions of 30, 50, 70, 80, 95 (in 150 mM NaCl), and 100% ethanol, and 100% xylene for 2 h each step. Finally, the samples were embedded in Paraplast (Paraplast plus, Sigma-Aldrich, St Louis, MO, USA) at 60 °C. Sections (8 μm) were obtained with a rotative Leitz microtome, transferred onto slides treated with 100 μg ml−1 poly-L-lysine (Sigma), and dried overnight on a warming plate at 40 °C. RT-PCR fragments of VvWUS, VvWOX1, VvWOX2, and VvWOX9 cloned as indicated above into the pGEM-T Easy vector (Promega) were transcribed in vitro from either the T7 or the SP6 promoter for sense (used as controls) and antisense RNA strand synthesis using the digoxigenin RNA labelling kit (Roche Applied Science). To exclude cross-hybridization between WOX genes, probes excluding the HD were generated (Supplementary Table S1 at JXB online) and a Southern blot hybridization of grapevine genomic DNA was performed. For all four genes, a single band only was detected, indicating that each probe hybridized to a single locus (data not shown). Hybridization and immunological detection were performed as described by Mayer et al. (1998) with some modifications: the rehydrated sections were incubated in proteinase K (1 μg ml−1 in 100 mM TRIS-HCl, and 5 mM EDTA pH 8.0) at 37 °C for 30 min and in 4% paraformaldehyde at 4 °C for 5 min; the dehydrated sections were then pre-hybridized for 10 min at 45 °C in 4× SSC (0.6 M NaCl, 60 mM sodium citrate) and 50% (v/v) deionized formamide. Hybridization was carried out overnight at 45 °C in the hybridization buffer [1× Denhardt's solution, 10% dextran sulphate, 10 mM dithiothreitol (DTT), 4× SSC, 1 mg ml−1 yeast tRNA, 1 mg ml−1 denatured and sheared salmon sperm DNA, and 50% deionized formamide] containing 500 ng of probes per ml of solution. The probes were detected using the chromogenic substrates nitroblue tetrazolium and 5-bromo-4-cloro-3-indolyl phosphate (NPT/BCIP), resulting in a blue precipitate that was viewed under a light microscope. Results and Discussion The VvWOX gene family The WOX gene family has been shown to play an important role in coordinating the gene transcription that is involved in the early phases of embryogenesis, in both shoot and root meristem functions, and in organogenesis in plants (Haecker et al., 2004). The two available draft genome sequences of grapevine (Jaillon et al., 2007; Velasco et al., 2007) were screened using the AtWOX proteins from Arabidopsis (Haecker et al., 2004) as the search term. Twelve non-redundant potential WOX proteins were identified and named VvWUS and VvWOXn, according to their sequence homology with the Arabidopsis WOX family. Recently, in a phylogenetic analysis of the WOX proteins present in several fully sequenced plant genomes, Vandenbussche et al. (2009) identified the same number of WOX genes in grapevine. AtWOX7, AtWOX8, AtWOX10, AtWOX12, and AtWOX14 lack orthologues in grapevine, while three different proteins orthologous to AtWOX13 were identified. The VvWOX genes identified through the bioinformatics approach were analysed by sequencing the amplification products obtained from embryogenic and other grapevine tissues (Supplementary Table S1 at JXB online). In the preliminary analyses by end-point RT-PCR, it was observed that almost all VvWOX genes are expressed in young tissues such as those of SE2 (11 genes out of 12; no amplification was observed for VvWUS) and IFs (8 genes out of 12; no amplification was observed for VvWOX2, VvWOX5, VvWOX11, and VvWOX13B), confirming the involvement of these transcription factors in grapevine somatic embryogenesis (Supplementary Fig. S1). For some genes (VvWUS, VvWOX5, and VvWOX13B), amplification in SE2 or IFs was not observed, probably because of the low sensitivity of the technique. In tissues from 3-year-old plants, only some VvWOX genes were amplified; in Rs, high expression was observed for VvWOX5, and VvWOX13A was the most expressed gene in all organs analysed including PWs sampled in winter (Supplementary Fig. S1). The sequencing of the RT-PCR products from ‘Chardonnay’ showed that in all genes at least one nucleotide was different from the sequences reported in the grapevine databases (Supplementary data set 1 at JXB online). These differences are generally localized outside the HD region and only in some cases do they cause an amino acid change (Supplementary Table S2). The most variable sequences were observed in VvWOX2, with a deletion of an amino acid near the C-terminus of the protein, and in VvWOX5. In the latter gene an uncorrected annotation of the N-terminal region in the PN40024 database (Supplementay Table S2 and data set 1) and an amino acid change in the HD region near helix I were observed. Uncorrected annotations were also observed in the N-terminal regions of VvWOX6 and VvWOX11, and in the C-terminus of VvWOX13A of the ‘Pinot Noir’ ENTAV115 database (Velasco et al., 2007). To investigate the relationship between the VvWOX protein deduced from the ‘Chardonnay’ sequences and other WOX proteins, the HD regions (Supplementary Table S2 at JXB online) and the complete protein sequences were examined to create phylogenetic trees. In the present analysis, in addition to 12 grapevine sequences, 76 sequences from representative dicot species (A. thaliana, P. trichocarpa, S. lycopersicum, P. abies, P. sitchensis, P. sylvestris, and C. sinensis) and two monocot species (O. sativa and Z. mays) were included. The phylogenetic analyses of the WOX-HD sequences (Fig. 2) and of the complete WOX sequences (Supplementary Fig. S3) showed the typical subdivision into three major evolutionary lineages, with a major clade containing the WUS and most WOX (WOX1–WOX7) proteins (Haecker et al., 2004; Deveaux et al., 2008; Vandenbussche et al., 2009). The VvWOX proteins are generally strictly related to the PtWOX and, interestingly, the SlWOX sequences (Fig. 2, and Supplementary S3). In all WOX clades the proteins from monocot species form subclusters distinct from those of dicot species. In addition, in the WUS/WOX1–WOX7 clade, the WUS-box was present (Supplementary Fig. S4), a domain conserved in different plants located downstream from the HD domain (Haecker et al., 2004). Recently Ikeda et al. (2009) demonstrated that the WUS-box is the domain essential for all the activities of WUS, namely for the regulation of stem cell identity and the size of the floral meristem. The sequence SLELSLN is also present in the C-terminal end of VvWUS (Supplementary Fig. S4), similar to the EAR-like motif that is involved in the repression of transcription (Kieffer et al., 2006). In grapevine, two proteins appear to be orthologous to AtWOX1. One protein, which was named VvWOX1, has a structure and peptide motif similar to those of AtWOX1, as also supported by phylogenetic analysis (Fig. 2). A second protein is correlated with AtWOX1 with regard to exon structure and phylogeny, but is also correlated with AtWOX6. Vandenbussche and colleagues (2009) named this protein VvWOX1B, whereas here it was preferred to rename it VvWOX6 because, according to the present experimental observations, the gene appears to be related to AtWOX6 and PtWOX6. In the WOX 13 clade, considered evolutionarily the ancient clade of WOX protein (van der Graaff et al., 2009), three grapevine genes are included. In particular, VvWOX13B and VvWOX13C are located next to each other in tandem on chromosome 4 with very similar sequences (94% nucleotide identity), suggestive of a duplication event; out of the 280 amino acids constituting the two proteins, differences were observed in 23 amino acids and only two substitutions in the HD sequence (Supplementary Figs S2, S4 at JXB online). Fig. 2. View largeDownload slide The Neighbor–Joining tree of the homeodomain (HD) region of WOX proteins from Vitis vinifera (Vv), Populus trichocarpa (Pt), Zea mays (Zm), Oryza sativa (Os), Arabidopsis thaliana (At), Solanum lycopersicum (Sl), Picea abies (Pa), Citrus sinensis (Cs), Pinus sylvestris (Ps), and Pinus sitchensis (Psit). The significance of each node was tested using 1000 bootstrap replicates. Fig. 2. View largeDownload slide The Neighbor–Joining tree of the homeodomain (HD) region of WOX proteins from Vitis vinifera (Vv), Populus trichocarpa (Pt), Zea mays (Zm), Oryza sativa (Os), Arabidopsis thaliana (At), Solanum lycopersicum (Sl), Picea abies (Pa), Citrus sinensis (Cs), Pinus sylvestris (Ps), and Pinus sitchensis (Psit). The significance of each node was tested using 1000 bootstrap replicates. The WOX family of transcription factors exhibits a plant-specific taxonomic profile and, according to Richardt et al. (2007), WUS probably evolved from an ancestral homeobox gene that was already present after the water-to-land transition of plants. Embryogenic competence It is known that in grapevine both the genotype and the explant type have significant effects on the differentiation of somatic embryos (Martinelli and Gribaudo, 2009). The WOX expression in flower explants (Fig. 1B, C, D) and in embryogenic tissues was therefore analysed in two cultivars of V. vinifera with different aptitudes for embryogenesis. Based on previous data (Gambino et al., 2007; and unpublished results), the cv Chardonnay, which has a high aptitude for embryogenesis, and the cv Cabernet Sauvignon, which has a low aptitude for embryogenesis in the present culture conditions, were chosen. After 3 months of culture, calli were obtained from all explant types of the two cultivars (Table 1). Most granular white or yellow calli, which were eventually associated with dark callus, differentiated SEs (Fig. 1E– J), whereas dark and compact or watery and soft callus showed little or no embryogenic competence (Fig. 1K–N). After 5 months of culture the percentages of calli differentiating SEs increased for all cultures. Calli from ovaries and whole flowers often expressed their embryogenic competence later than those from anthers. All explant types of ‘Chardonnay’ originated pre-embryogenic masses and EC, whereas only anthers of ‘Cabernet Sauvignon’ differentiated EC. Table 1. Frequency (%) of callogenesis and somatic embryogenesis starting from whole flowers, ovaries, and anthers of V. vinifera cv Chardonnay and Cabernet Sauvignon, after 3 and 5 months of culture Cultivar  Explant type (no. of explants in culture)  Callogenesis (%) after 3 months of culturea  Embryogenesis (%) after 3 months of culturea  Embryogenesis (%) after 5 months of culturea  ‘Chardonnay’  Flower (317)  87.7 a  2.8 a  13.2 a    Ovary (470)  38.7 b  13 a  17.6 a    Anther (1585)  8.4 c  6.2 a  7.1 a  ‘Cabernet Sauvignon’  Flower (440)  90.4 a  0 a  0 b  Ovary (525)  86.3 a  0 a  0 b  Anther (1600)  21.3 b  1.1 a  1.9 a  Cultivar  Explant type (no. of explants in culture)  Callogenesis (%) after 3 months of culturea  Embryogenesis (%) after 3 months of culturea  Embryogenesis (%) after 5 months of culturea  ‘Chardonnay’  Flower (317)  87.7 a  2.8 a  13.2 a    Ovary (470)  38.7 b  13 a  17.6 a    Anther (1585)  8.4 c  6.2 a  7.1 a  ‘Cabernet Sauvignon’  Flower (440)  90.4 a  0 a  0 b  Ovary (525)  86.3 a  0 a  0 b  Anther (1600)  21.3 b  1.1 a  1.9 a  a For each cultivar, means followed by the same letter do not differ significantly at P ≤0.05 as determined by the Duncan's multiple range test. View Large Expression dynamics of VvWOX genes in ‘Chardonnay’ The expression patterns of WOX genes were initially analysed in the starting explants and calli described above, in embryos at the torpedo stage (SE1) and at the cotyledonary stage (SE2), in SEG, as well as in tissues from 3-year-old plant: SAs, TRs, and MLs of ‘Chardonnay’. The most stable housekeeping genes in these grapevine tissues were ACT and UBI (results not shown), and the geometric mean of their expression ratios was used as a normalization factor in all analyses. The relative expression levels of the VvWOX genes in the various tissues and culture phases, as determined by qRT-PCR, are shown in Figs 3 and 5. Fig. 3. View largeDownload slide Relative expression levels of VvWUS, VvWOX1, VvWOX2, VvWOX3, VvWOX4, VvWOX5, VvWOX6, and VvWOX9 in ‘Chardonnay’ and ‘Cabernet Sauvignon’ determined by qRT-PCR. The PCR signals were normalized with those of ACT and UBI transcripts. Data are means ±SEs of three replicates. The asterisks indicate samples not analysed in ‘Cabernet Sauvignon’. Fig. 3. View largeDownload slide Relative expression levels of VvWUS, VvWOX1, VvWOX2, VvWOX3, VvWOX4, VvWOX5, VvWOX6, and VvWOX9 in ‘Chardonnay’ and ‘Cabernet Sauvignon’ determined by qRT-PCR. The PCR signals were normalized with those of ACT and UBI transcripts. Data are means ±SEs of three replicates. The asterisks indicate samples not analysed in ‘Cabernet Sauvignon’. High levels of VvWUS mRNA were detected in immature anthers of ‘Chardonnay’, and at lower levels in ovaries and flowers. After 40 d of culture, the expression levels increased in cultured flowers, and in 3-month-old cultures seemed to be associated with embryogenesis (Fig. 3). In SE1, SE2, and SEG, the gene was expressed, although the levels were lower than those observed in EC. WUS is known to play a critical role during embryogenesis by promoting the vegetative-to-embryogenic transition and maintaining the identity of the embryonic stem cells (Zuo et al., 2002). WUS is also involved in the development of the floral organs: the gene is expressed in the apical nucellus of the ovule primordium (Sieber et al., 2004). In anthers of Arabidopsis, WUS expression is localized in stomium cells, maintaining the anther primordium cells in an undifferentiated state (Deyhle et al., 2007). In grapevine, WUS was found to be localized in several epidermal cells in the anthers through in situ hybridization experiments carried out on immature flowers of ‘Chardonnay’ (Fig. 4A, D). The WUS protein was previously characterized in Arabidopsis as a key regulator for the specification of meristem cell fate (Laux et al., 1996), and its expression in meristems is restricted to the organizing centre of the shoot meristem during the post-embryonic stages (Mayer et al., 1998). Recently Su and colleagues (2009) in Arabidopsis and Chen et al. (2009) in Medicago truncatula showed that auxin regulates WUS expression during somatic embryogenesis. Su et al. (2009) also reported that WUS was induced in EC prior to the point at which SEs could be identified morphologically, and that WUS induction is correlated with auxin gradients that are established in specific regions of the EC. Similarly, activation of the gene in the explants was observed after 40 d of culture and in the EC: its expression in the explants before the beginning of embryo differentiation may be associated with the gene activation caused by the exogenous PGRs present in the culture medium. The overexpression of WUS increases somatic embryogenesis from mature tissues of different species without the addition of exogenous PGRs (Zuo et al., 2002; Arroyo-Herrera et al., 2008; Solis-Ramos et al., 2009). Fig. 4. View largeDownload slide In situ hybridization detection of VvWOX genes in several tissues of grapevine. Sections of immature flowers of ‘Chardonnay’ were hybridized with digoxigenin-labelled antisense (A, D) and sense (B, E) VvWUS RNA probes, and antisense (C) and sense (F) VvWOX1 RNA probes. Pre-embryogenic masses in EC of ‘Chardonnay’ were hybridized with antisense (G, H) and sense (I) VvWOX9 RNA probes, and antisense (J, K) and sense (L) VvWOX2 RNA probes. The black arrows indicate the blue precipitate corresponding to VvWOX gene expression. c, caliptra; e, epidermis cells of anther; f, anther filament; p, pollen sacs; r, receptacle; s, sepal; st, stylus. Size bar=200 μm. Fig. 4. View largeDownload slide In situ hybridization detection of VvWOX genes in several tissues of grapevine. Sections of immature flowers of ‘Chardonnay’ were hybridized with digoxigenin-labelled antisense (A, D) and sense (B, E) VvWUS RNA probes, and antisense (C) and sense (F) VvWOX1 RNA probes. Pre-embryogenic masses in EC of ‘Chardonnay’ were hybridized with antisense (G, H) and sense (I) VvWOX9 RNA probes, and antisense (J, K) and sense (L) VvWOX2 RNA probes. The black arrows indicate the blue precipitate corresponding to VvWOX gene expression. c, caliptra; e, epidermis cells of anther; f, anther filament; p, pollen sacs; r, receptacle; s, sepal; st, stylus. Size bar=200 μm. Like WUS, WOX5 has important functions in meristem regulation: Kamiya et al. (2003) in rice and Sarkar et al. (2007) in Arabidopsis showed that WOX5 is expressed in the quiescent centre cells of the root meristem and maintains stem cells in the undifferentiated state in the root in the same way as WUS maintains the stem cells in the shoot. As predicted, VvWOX5 was detected in grapevine TR, and its expression was activated during culture on the auxin-containing substrate (Fig. 3), similarly to that observed by Gonzali et al. (2005) and Chen et al. (2009). In the present experiments a close association between VvWOX5 expression and NEC was observed. In addition, the gene seems to be involved in the initial phases of SE development: in EC and in young embryos (SE1) its expression levels are significantly higher than those found in the following stages SE2 and SEG (Fig. 3). In Arabidopsis some WOX genes are expressed very early during embryo development (Haecker et al., 2004), and the interaction between WOX2, WOX8, and WOX9 (a close homologue of WOX8) determines the correct developmental process of embryos (Breuninger et al., 2008). In Arabidopsis, WOX8 and WOX9 regulate gene expression programmes and division patterns in both the basal and apical lineage of the embryo, and WOX2 expression in the apical lineage is an early downstream function of WOX8/WOX9 activity (Breuninger et al., 2008). In grapevine, the very early stages (first divisions) of embryo development were not analysed as was done for the zygotes in Arabidopsis (Haecker et al., 2004; Breuninger et al., 2008). However, the results from EC (containing pre-embryogenic masses and SEs in early stages of development) and from developing somatic embryos (SE1, SE2, and SEG) can provide important information about the initial phases of embryogenic processes in this plant. In grapevine there is no gene orthologue to AtWOX8, but its role in embryo development seems to be replaced by VvWOX9, which, with VvWOX2, is the principal WOX gene expressed in EC (Fig. 3). VvWOX2 was not expressed in immature anthers, ovaries, or flowers at culture initiation (Fig. 3). After 40 d, an initial expression was found in ovaries and flowers, while high levels of expression of VvWOX2 were found in embryogenic tissues originating from all explant types. Its expression levels in SE1 were ∼15-fold lower than in EC, were further reduced in SE2, and were undetectable in SEG (Fig. 3). In situ hybridization experiments carried out on pre-embryogenic masses showed VvWOX2 expression in small young cells in the peripheral zone of the EC (Fig. 4J, K). The expression profile of VvWOX9 was similar to the profile observed for VvWOX2: high levels of expression of VvWOX9 were found in the embryogenic tissues originating from all explant types (Fig. 3), and in situ hybridizations revealed its expression in peripheral cells of the EC (Fig. 4G, H). However, unlike VvWOX2, the reduction of VvWOX9 expression in tissues from EC to SEG occurred more slowly: in SE2 the expression level was ∼5-fold lower than in EC from anthers, and it was further reduced in SEG, although it did not disappear completely (Fig. 3). Both VvWOX2 and VvWOX9 seem to be associated with early embryogenic phases in grapevine. However, while VvWOX2 was deactivated in late stages of embryo development (SEG) and in mature tissues (SAs, TRs, and MLs), the expression of VvWOX9 was also detected in SEG and SAs. Probably, besides embryogenesis, the gene is involved in other phases of plant development, as has been observed in Arabidopsis where it regulates WUS expression in shoot apical meristems (Wu et al., 2005). The association between VvWOX2 and EC was also observed in Picea abies (Palovaara and Hakmann, 2008) and in microspore-derived embryos and developing seeds of Brassica napus (Malik et al., 2007), in which the authors identified the BnWOX9 gene as a marker for SEs. In grapevine VvWOX2 and VvWOX9 could be used as markers for somatic embryogenesis. Besides the high level of expression of these genes in EC, low levels of expression were also detected in 3-month-old NEC. Some of the NEC can subsequently turn into EC, as reported previously (Gambino et al., 2007) and as shown in Table 1. Somatic embryogenesis in grapevine is a long process that lasts several months. Although no pre-embryogenic mass and/or SE was visually detectable in NEC after 3 months of culture, the embryogenic process could have started in some cells of the callus, thus explaining the expression of those genes at a low level in NEC. VvWOX1 mRNA was detected in immature ovaries and flowers at culture initiation; after 40 d of culture gene expression was greatly reduced (Fig. 3). The culture conditions led to its deactivation and the gene does not seem to be associated with early embryogenic processes. However, its expression increases during embryo development (SE1 and SE2), is very high in SEG, and continues to be high in organs such as SAs and MLs from fully developed plants (Fig. 3). In Arabidopsis the WOX1 expression patterns are not well characterized; its expression is confined to the initiating vascular primordium of the cotyledons during the heart and torpedo stages of the embryos (Haecker et al., 2004). In situ hybridization experiments indicated VvWOX1 expression in the ovary and in the vascular tissue localized in the receptacle of the flower (Fig. 4C). In grapevine, VvWOX6 was expressed in the floral organs and in young organs, as also reported in Arabidopsis (Park et al., 2005), with high expression in the SAs and absence in MLs (Fig. 3). The gene was also expressed in all explants at the time of culture initiation, and in all calli its expression was higher in EC than in NEC, similar to that observed for VvWUS, VvWOX2, and VvWOX9. VvWOX1 and VvWOX6 are correlated in their exon structure and in their peptide motif, and on the basis of a bioinformatics approach Vandenbussche et al. (2009) named these genes VvWOX1A and VvWOX1B. However the expression profiles of these genes in the EC and SEs have significant differences: the expression of VvWOX6 remained nearly constant in EC, SE1, SE2, and SEG, whereas the expression of VvWOX1 was very low in EC and increased during the embryo developmental stages. Therefore, the genes were renamed VvWOX1 and VvWOX6 because in the present experimental conditions they showed expression profiles correlated with AtWOX1 and AtWOX6, respectively. VvWOX4 was expressed in EC at lower levels, while it increased during embryo development (from SE1 to SEG) (Fig. 3), with a profile very similar to that observed in VvWOX1. WOX4 expression has not been well characterized in other species; however, also in Arabidopsis and Z. mays it was expressed in young embryos with a profile similar to WOX1 (Nardmann et al., 2007). Unlike VvWOX1, when organs from fully developed plants were examined, high VvWOX4 mRNA was detected in TRs and low levels in MLs. VvWOX3 mRNA was detected in all floral tissues at culture initiation (Fig. 3), and in ovaries and flowers after 40 d of culture, while its expression was very low in callus tissues (EC and NEC). The mRNA was found at very high levels in SE1 and SE2 (∼200-fold higher than in EC) and decreased in SEG (Fig. 3). The VvWOX3 expression profile in somatic embryos of ‘Chardonnay’ was similar to that published for Arabidopsis, where it has been demonstrated that AtWOX3 was expressed at the margins of the cotyledonary primordia of heart stage embryos (Haecker et al., 2004). A close association between VvWOX3 expression and the pre-embryogenic tissues or globular SEs localized in EC was not observeed. VvWOX11 was not expressed in the explants at culture initiation, but after 40 d of culture its expression was activated in ovaries and flowers (Fig. 5). A close association between VvWOX11 expression and NEC was observed, particularly in the anthers. The gene showed an expression profile in EC and SEs very similar to that observed for VvWOX3: low expression in EC, increasing in SE1 and SE2, and decreasing in SEG. To our knowledge, the involvement of this gene in the embryo developmental processes between torpedo and cotyledonary stages was never observed before in plants. Recently Zhao et al. (2009) showed that in rice WOX11 is involved in the activation of crown root emergence and growth, and is also expressed in root and shoot meristems. The gene was induced by auxins and BAP. Also in grapevine VvWOX11 seems to be activated by the PGRs of the culture medium. Fig. 5. View largeDownload slide Relative expression levels of VvWOX11, VvWOX13A, VvWOX13B, and VvWOX13C in ‘Chardonnay’ and ‘Cabernet Sauvignon’ determined by qRT-PCR. The PCR signals were normalized with those of ACT and UBI transcripts. Data are means ±SEs of three replicates. The asterisks indicate samples not analysed in ‘Cabernet Sauvignon’. Fig. 5. View largeDownload slide Relative expression levels of VvWOX11, VvWOX13A, VvWOX13B, and VvWOX13C in ‘Chardonnay’ and ‘Cabernet Sauvignon’ determined by qRT-PCR. The PCR signals were normalized with those of ACT and UBI transcripts. Data are means ±SEs of three replicates. The asterisks indicate samples not analysed in ‘Cabernet Sauvignon’. In grapevine three VvWOX genes were identified that are orthologous to AtWOX13 with a close sequence homology, but with different expression profiles in the tissues analysed. VvWOX13B was detected in all tissues and calli as well as in SEs, SEG, SAs, TRs, and MLs (Fig. 5). However, this gene had the lowest level of expression in all tissues analysed, and culture conditions apparently did not influence its expression profile. VvWOX13C was expressed in all tissues at the time of culture initiation, and after 40 d an increase in its mRNA was detected in ovaries and flowers (Fig. 5). In calli and in other tissues such as SE2, SEG, TRs, and SAs, limited variations in its expression profile were observed. VvWOX13B and VvWOX13C show a close sequence homology and are located in tandem configuration on chromosome 4. The characteristics of the two genes are suggestive of a duplication event; however, VvWOX13B expression was lower than VvWOX13C expression. The expression profile of VvWOX13A showed a trend similar to that of VvWOX13C: the gene was activated after 40 d of culture and was detected in EC as well as in NEC. The expression profiles of VvWOX13A and VvWOX13C are in partial agreement with the expression of AtWOX13 in Arabidopsis (Deveaux et al., 2008). AtWOX13 was expressed in several young tissues and appeared to affect the floral transition. Expression dynamics of VvWOX genes in two grapevine cultivars After the complete characterization of VvWOX expression in ‘Chardonnay’, the expression patterns of the genes were analysed in the starting explants for embryogenesis, in EC and in NEC of ‘Cabernet Sauvignon’. As reported above ‘Cabernet Sauvignon’ showed a low aptitude for embryogenesis in the present culture conditions in comparison with ‘Chardonnay’. In the floral explants (anthers, ovaries, and flowers) used for the initiation of cultures in May, the expression dynamics of VvWOX genes were similar in ‘Chardonnay’ and ‘Cabernet Sauvignon’ as regards expression profiles and relative amount of transcripts (Figs 3, 5). Later on, the culture conditions that allowed the formation of EC induced different expression patterns of WOX genes. It can be hypothesized that the exogenous PGRs added to the culture medium are largely responsible for the changes in gene expression. Several authors have reported the influence of exogenous PGRs, in particular auxins, on the expression of various WOX genes (Gonzali et al., 2005; Chen et al., 2009; Su et al., 2009; Zhao et al., 2009). However, in the two cultivars of V. vini fera the VvWOX genes were activated differently in response to the culture. VvWUS, VvWOX2, and VvWOX9 were activated in ovaries of ‘Chardonnay’ after 40 d of culture on medium containing 2,4-D and BA, while in ‘Cabernet Sauvignon’ the expression of these genes was reduced (VvWUS) or not activated (VvWOX2 and VvWOX9; Fig. 3). Otherwise in the same conditions in ‘Cabernet Sauvignon’, induction of VvWOX4, VvWOX6, VvWOX13A, and VvWOX13C was observed (Figs 3, 5), these being genes that do not seem to be involved or are only partially (VvWOX6) involved in somatic embryogenesis. Conversely, ‘Cabernet Sauvignon’ produced abundant NEC and these genes seem to be implicated in the formation of callus that is unable to differentiate SEs. After 3 months of culture, in EC from anthers of ‘Chardonnay’ the expression of VvWOX2 and VvWOX9 was at least 10 times higher than in EC from the same explants of ‘Cabernet Sauvignon’, while VvWOX1, VvWOX3, VvWOX4, and VvWOX13A were expressed more highly in ‘Cabernet Sauvignon’ (Figs 3, 5). VvWOX2 and VvWOX9 seem to be clearly correlated with the precocious phases of SE formation. In non-embryogenic calli of ‘Chardonnay’ they were also expressed, but at lower levels. As discussed above, the expression of these genes in NEC of ‘Chardonnay’ could explain the ability of this cultivar to differentiate EC and SE from NEC after a long period of culture (between 3 and 5 months after culture initiation). This aptitude is absent in ovaries and flowers and is very low in anthers of ‘Cabernet Sauvignon’. In addition, in EC of ‘Cabernet Sauvignon’, the most activated WOX genes were VvWOX1, VvWOX3, and VvWOX4, genes that in ‘Chardonnay’ were associated with torpedo and cotyledonary embryos (SE1 and SE2). It could be assumed that the activation of these genes in ‘Cabernet Sauvignon’ occurred earlier in comparison with ‘Chardonnay’; however, there is not currently sufficient evidence to corroborate this hypothesis with certainty. In ‘Chardonnay’ the VvWOX13B was the gene least expressed in all tissues, and in ‘Cabernet Sauvignon’ its expression was even lower, as very low levels of mRNA were detected in anthers and NEC from flowers only (Fig. 5). Conclusions The analyses carried out on different tissues of grapevine using qRT-PCR and in situ hybridization confirmed the previous observations in Arabidopsis and provided new information about the expression dynamics of WOX genes in plants. The results indicated that in grapevine, as well as in other species, some VvWOX genes are involved in somatic embryogenesis. It was shown that during early phases of embryogenesis VvWOX regulation differs in two cultivars of the same species. As different cultivars of V. vinifera vary quite widely in their potential to form embryogenic tissues, VvWOX genes may be key regulators (or some of the key regulators) of somatic embryogenesis in grapevine. In addition to knowledge of the regulatory pathway leading to the formation of SEs, the results could have applications in some important aspects of somatic embryogenesis in grapevine. The publication of the genome sequence of grapevine offers new perspectives in genomic research, but the new ‘omics’ technologies need to be complemented by efficient transformation systems for the study of genes. VvWOX2 and VvWOX9 can be used as early expression markers associated with the developmental programme of SEs, allowing the identification of potential EC before the morphological appearance of SEs or pre-embryogenic masses. VvWOX genes could also allow improvement of the transformation methodology in grapevine. VvWOX2 and/or VvWOX9 and/or VvWUS could be transferred through an Agrobacterium-mediated system and, under inducible promoters, expressed in undifferentiated calli or somatic tissues, thus allowing the differentiation of SE from tissues that are normally unable to differentiate EC. In this way it could be possible to transform recalcitrant grapevine cultivars. These VvWOX genes could also be used as positive selectable markers, eliminating the need for genes for antibiotic resistance: in principle, only embryos containing the expression cassette will differentiate from transformed tissues. Testing of these hypotheses is currently under way. References Arroyo-Herrera A,  Gonzalez AK,  Canche Moo R,  Quirez-Figueroa FR,  Loyola-Vargas VM,  Rodriguez-Zapata LC,  Burgeff D'Hondt C,  Suárez-Solís VM,  Castaño E.  Expression of WUSCHEL in Coffea canephora causes ectopic morphogenesis and increases somatic embryogenesis,  Plant Cell, Tissue and Organ Culture ,  2008, vol.  94 (pg.  171- 180) Google Scholar CrossRef Search ADS   Breuninger H,  Rikirsch E,  Hermann M,  Ueda M,  Laux T.  Differential expression of WOX genes mediates apical–basal axis formation in the Arabidopsis embryo,  Developmental Cell ,  2008, vol.  14 (pg.  867- 876) Google Scholar CrossRef Search ADS PubMed  Chen SK,  Kurdyukov S,  Kerezst A,  Wang XD,  Gresshoff PM,  Rose RJ.  The association of homeobox gene expression with stem cell formation and morphogenesis in cultured Medicago truncatula,  Planta ,  2009, vol.  230 (pg.  827- 840) Google Scholar CrossRef Search ADS PubMed  Deveaux Y,  Toffano-Nioche C,  Claisse G,  Thareau V,  Morin H,  Laufs P,  Moreau H,  Kreis M,  Lecharny A.  Genes of the most conserved WOX clade in plants affect root and flower development in Arabidopsis,  BMC Evolutionary Biology ,  2008, vol.  8 pg.  291  Google Scholar CrossRef Search ADS PubMed  Deyhle F,  Sarkar AK,  Tucker EJ,  Laux T.  WUSCHEL regulates cell differentiation during anther development,  Developmental Biology ,  2007, vol.  302 (pg.  154- 159) Google Scholar CrossRef Search ADS PubMed  Dodeman VL,  Ducreux G,  Kreis M.  Zygotic embryogenesis versus somatic embryogenesis,  Journal of Experimental Botany ,  1997, vol.  48 (pg.  1493- 1509) Gambino G,  Gribaudo I.  Simultaneous detection of nine grapevine viruses by multiplex RT-PCR with coamplification of a plant RNA internal control,  Phytopathology ,  2006, vol.  96 (pg.  1223- 1229) Google Scholar CrossRef Search ADS PubMed  Gambino G,  Perrone I,  Gribaudo I.  A rapid and effective method for RNA extraction from different tissues of grapevine and other woody plants,  Phytochemical Analysis ,  2008, vol.  19 (pg.  520- 525) Google Scholar CrossRef Search ADS PubMed  Gambino G,  Ruffa P,  Vallania R,  Gribaudo I.  Somatic embryogenesis from whole flowers, anthers and ovaries of grapevine (Vitis spp.),  Plant Cell, Tissue and Organ Culture ,  2007, vol.  90 (pg.  79- 83) Google Scholar CrossRef Search ADS   Gonzali S,  Novi G,  Loreti E,  Paolicchi F,  Poggi A,  Alpi A,  Perata P.  A turanose insensitive mutant suggests a role for WOX5 in auxin homeostasis in Arabidopsis thaliana,  The Plant Journal ,  2005, vol.  44 (pg.  633- 645) Google Scholar CrossRef Search ADS PubMed  Gribaudo I,  Gambino G,  Vallania R.  Somatic embryogenesis from grapevine anthers: identification of the optimal developmental stage for collecting explants,  American Journal of Enology and Viticulture ,  2004, vol.  55 (pg.  427- 430) Haecker A,  Gross-Hardt R,  Geiges B,  Sarkar A,  Breuninger H,  Herrmann M,  Laux T.  Expression dynamics of WOX genes mark cell fate decisions during early embryonic patterning in Arabidopsis thaliana,  Development ,  2004, vol.  131 (pg.  657- 668) Google Scholar CrossRef Search ADS PubMed  Ikeda M,  Mitsuda N,  Ohme-Takagi M.  Arabidopsis WUSCHEL is a bifunctional transcription factor that acts as a repressor in stem cell regulation and as an activator in floral patterning,  The Plant Cell ,  2009, vol.  21 (pg.  3493- 3505) Google Scholar CrossRef Search ADS PubMed  Imin N,  Nizamidin M,  Wu T,  Rolfe BG.  Factors involved in root formation in Medicago truncatula,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  439- 451) Google Scholar CrossRef Search ADS PubMed  Jaillon O,  Aury JM,  Noel B, et al.  The grapevine genome sequence suggests ancestral hexaploidization in major angiosperm phyla,  Nature ,  2007, vol.  449 (pg.  463- 468) Google Scholar CrossRef Search ADS PubMed  Jain M,  Tyagi AK,  Khurana JP.  Genome-wide identification, classification, evolutionary expansion and expression analyses of homeobox genes in rice,  FEBS Journal ,  2008, vol.  275 (pg.  2845- 2861) Google Scholar CrossRef Search ADS PubMed  Kamiya N,  Nagasaki H,  Morikami A,  Sato Y,  Matsuoka M.  Isolation and characterization of a rice WUSCHEL-type homeobox gene that is specifically expressed in the central cells of a quiescent center in the root apical meristem,  The Plant Journal ,  2003, vol.  35 (pg.  429- 441) Google Scholar CrossRef Search ADS PubMed  Kieffer M,  Stern Y,  Cook H,  Clerici E,  Maulbetsch C,  Laux T,  Davies B.  Analysis of the transcription factor WUSCHEL and its functional homologue in Antirrhinum reveals a potential mechanism for their roles in meristem maintenance,  The Plant Cell ,  2006, vol.  18 (pg.  560- 573) Google Scholar CrossRef Search ADS PubMed  Laux T,  Mayer KF,  Berger J,  Jurgens G.  The WUSCHEL gene is required for shoot and floral meristem integrity in Arabidopsis,  Development ,  1996, vol.  122 (pg.  87- 96) Google Scholar PubMed  Livak KJ,  Schmittgen TD.  Analysis of relative gene expression data using real-time quantitative PCR and the 2–ΔΔCTmethod,  Methods ,  2001, vol.  25 (pg.  402- 408) Google Scholar CrossRef Search ADS PubMed  Maillot P,  Lebel S,  Schellenbaum P,  Jacques A,  Walter B.  Differential regulation of SERK, LEC-1Like and Pathogenesis-Related genes during indirect secondary somatic embryogenesis in grapevine,  Plant Physiology and Biochemistry ,  2009, vol.  47 (pg.  743- 752) Google Scholar CrossRef Search ADS PubMed  Malik MR,  Wang F,  Dirpaul JM,  Zhou N,  Polowick PL,  Ferrie AMR,  Krochko JE.  Transcript profiling and identification of molecular markers for early microspore embryogenesis in Brassica napus,  Plant Physiology ,  2007, vol.  144 (pg.  134- 154) Google Scholar CrossRef Search ADS PubMed  Martinelli L,  Gribaudo I.  Roubelakis-Angelakis K.  Strategies for effective somatic embryogenesis in grapevine (Vitis spp.): an appraisal,  Grapevine molecular physiology & biotechnology ,  2009 2nd edn Dordrecht Springer(pg.  461- 494) Mayer KF,  Schoof H,  Haecker A,  Lenhard M,  Jurgens G,  Laux T.  Role of WUSCHEL in regulating stem cell fate in the Arabidopsis shoot meristem,  Cell ,  1998, vol.  95 (pg.  805- 815) Google Scholar CrossRef Search ADS PubMed  Mordhorst AP,  Toonen MAJ,  de Vries S.  Plant embryogenesis,  Plant Science ,  1997, vol.  16 (pg.  535- 576) Google Scholar CrossRef Search ADS   Nardmann J,  Ji J,  Werr W,  Scanlon MJ.  The maize duplicate genes narrow sheath1 and narrow sheath2 encode a conserved homeobox gene function in a lateral domain of shoot apical meristems,  Development ,  2004, vol.  131 (pg.  2827- 2839) Google Scholar CrossRef Search ADS PubMed  Nardmann J,  Zimmermann R,  Durantini D,  Kranz E,  Werr W.  WOX gene phylogeny in Poaceae: a comparative approach addressing leaf and embryo development,  Molecular Biology and Evolution ,  2007, vol.  24 (pg.  2474- 2484) Google Scholar CrossRef Search ADS PubMed  Palovaara J,  Hakman I.  Conifer WOX-related homeodomain transcription factors, developmental consideration and expression dynamic of WOX2 during Picea abies somatic embryogenesis,  Plant Molecular Biology ,  2008, vol.  66 (pg.  533- 549) Google Scholar CrossRef Search ADS PubMed  Park SO,  Zheng Z,  Oppenheimer DG,  Hauser BA.  The PRETTY FEW SEEDS2 gene encodes an Arabidopsis homeodomain protein that regulates ovule development,  Development ,  2005, vol.  132 (pg.  841- 849) Google Scholar CrossRef Search ADS PubMed  Reid KE,  Olsson N,  Schlosser J,  Peng F,  Lund ST.  An optimized grapevine RNA isolation procedure and statistical determination of reference genes for real-time RT-PCR during berry development,  BMC Plant Biology ,  2006, vol.  6 pg.  27  Google Scholar CrossRef Search ADS PubMed  Richardt S,  Lang D,  Reski R,  Frank W,  Rensing SA.  PlantTAPDB, a phylogeny-based resource of plant transcription-associated proteins,  Plant Physiology ,  2007, vol.  143 (pg.  1452- 1466) Google Scholar CrossRef Search ADS PubMed  Rozen S,  Skaletsky HJ.  Primer3 on the WWW for general users and for biologist programmers,  Methods in Molecular Biology ,  2000, vol.  132 (pg.  365- 386) Google Scholar PubMed  Sarkar AK,  Luijten M,  Miyashima S,  Lenhard M,  Hashimoto T,  Nakajima K,  Scheres B,  Heidstra R,  Laux T.  Conserved factors regulate signaling in Arabidopsis thaliana shoot and root stem cell organizers,  Nature ,  2007, vol.  446 (pg.  811- 814) Google Scholar CrossRef Search ADS PubMed  Schellenbaum P,  Jacques A,  Maillot P,  Bertsh C,  Mazet F,  Farine S,  Walter B.  Characterization of VvSERK1, VvSERK2, VvSERK3 and VvL1L genes and their expression during somatic embryogenesis of grapevine (Vitis vinifera L.),  Plant Cell Reports ,  2008, vol.  27 (pg.  1799- 1809) Google Scholar CrossRef Search ADS PubMed  Sieber P,  Gheyselinck J,  Gross-Hardt R,  Laux T,  Grossniklaus U,  Schneitz K.  Pattern formation during early ovule development in Arabidospis thaliana,  Developmental Biology ,  2004, vol.  273 (pg.  321- 334) Google Scholar CrossRef Search ADS PubMed  Solis-Ramos LY,  Gonzalez-Estrada T,  Nahuath-Dzib S,  Zapata-Rodriguez LC,  Castano E.  Overexpression of WUSCHEL in C. chinense causes ectopic morphogenesis,  Plant Cell, Tissue and Organ Culture ,  2009, vol.  96 (pg.  279- 287) Google Scholar CrossRef Search ADS   Su YH,  Zhao HY,  Liu YB,  Zhang CL,  O'Neil SD,  Zhang XS.  Auxin-induced WUS expression is essential for embryonic stem cell renewal during somatic embryogenesis in Arabidopsis,  The Plant Journal ,  2009, vol.  59 (pg.  448- 460) Google Scholar CrossRef Search ADS PubMed  Vandenbussche M,  Horstman A,  Zethof J,  Koes R,  Rijpkema AS,  Gerats T.  Differential recruitment of WOX transcription factors for lateral development and organ fusion in Petunia and Arabidopsis,  The Plant Cell ,  2009, vol.  21 (pg.  2269- 2283) Google Scholar CrossRef Search ADS PubMed  van der Graaff E,  Laux T,  Rensing SA.  The WUS homeobox-containing (WOX) protein family,  Genome Biology ,  2009, vol.  10 pg.  248  Google Scholar CrossRef Search ADS PubMed  Velasco R,  Zharkikh A,  Troggio M, et al.  A high quality draft consensus sequence of the genome of a heterozygous grapevine variety,  PloS ONE ,  2007, vol.  2 pg.  e1326  Google Scholar CrossRef Search ADS PubMed  Wu X,  Dabi T,  Weigel D.  Requirement of homeobox gene STIMPY/WOX9 for Arabidopsis meristem growth and maintenance,  Current Biology ,  2005, vol.  15 (pg.  436- 440) Google Scholar CrossRef Search ADS PubMed  Zhao Y,  Hu Y,  Dai M,  Huang L,  Zhou DX.  The WUSCHEL-related homeobox gene WOX11 is required to activate shoot-borne crown root development in rice,  The Plant Cell ,  2009, vol.  21 (pg.  736- 748) Google Scholar CrossRef Search ADS PubMed  Zuo J,  Niu QW,  Frugis G,  Chua NH.  The WUSCHEL gene promotes vegetative-to-embryonic transition in Arabidopsis,  The Plant Journal ,  2002, vol.  30 (pg.  349- 359) Google Scholar CrossRef Search ADS PubMed  © The Author [2010]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: [email protected]

Zhang, Juanjuan;Wang, Xin;Yu, Oliver;Tang, Juanjuan;Gu, Xungang;Wan, Xiaochun;Fang, Congbing

doi: 10.1093/jxb/erq343pmid: 21041374

Abstract Strawberry (Fragaria×ananassa Duch), a fruit of economic and nutritional importance, is also a model species for fleshy fruits and genomics in Rosaceae. Strawberry fruit quality at different harvest stages is a function of the fruit's metabolite content, which results from physiological changes during fruit growth and ripening. In order to investigate strawberry fruit development, untargeted (GC-MS) and targeted (HPLC) metabolic profiling analyses were conducted. Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were employed to explore the non-polar and polar metabolite profiles from fruit samples at seven developmental stages. Different cluster patterns and a broad range of metabolites that exerted influence on cluster formation of metabolite profiles were observed. Significant changes in metabolite levels were found in both fruits turning red and fruits over-ripening in comparison with red-ripening fruits. The levels of free amino acids decreased gradually before the red-ripening stage, but increased significantly in the over-ripening stage. Metabolite correlation and network analysis revealed the interdependencies of individual metabolites and metabolic pathways. Activities of several metabolic pathways, including ester biosynthesis, the tricarboxylic acid cycle, the shikimate pathway, and amino acid metabolism, shifted during fruit growth and ripening. These results not only confirmed published metabolic data but also revealed new insights into strawberry fruit composition and metabolite changes, thus demonstrating the value of metabolomics as a functional genomics tool in characterizing the mechanism of fruit quality formation, a key developmental stage in most economically important fruit crops. Fragaria×ananassa, fruit maturation, fruit quality, GC-MS profiling, metabolic profiling, metabolite composition, metabolomics, strawberry Introduction Strawberry (Fragaria×ananassa Duch.) is one of the most economically important fresh and processed fruits, consumed for both its pleasant flavour and its nutrient content (Hancock, 1999). Strawberry is cultivated throughout the world; ∼4.07 million tons were produced in 2008 (http://faostat.fao.org/site/567/default.aspx). Strawberry flavour is a result of a complex mixture of numerous volatile and organoleptic compounds combined with characteristics such as texture and taste. More than 300 volatile compounds have been identified in ripening strawberry (Honkanen and Hivi, 1990; Latrasse, 1991), which can be grouped into several chemical classes, including the major components acids, aldehydes, ketones, alcohols, esters, and lactones, and the contributing groups of sulphur compounds, acetals, furans, phenols, epoxides, and alkanes (Zabetakis and Holden, 1997). Although the individual volatiles of these groups are often present in minute quantities, typically in concentrations of 10–100 ppm of the fruit fresh weight (Maarse, 1991), they have a significant impact on the overall flavour of strawberry fruits. The nutritional quality of strawberry fruits is closely correlated with the presence of soluble sugars, organic acids, amino acids, and some major secondary metabolites. These compounds play an important role in maintaining fruit quality and nutritive value; for this reason, fruit compositional analysis is of interest to food chemists and processors. Phenolic acids and their derivatives are often conjugated with sugars and have been frequently reported in strawberry fruits and leaves (Määttä-Riihinen et al., 2004; Aaby et al., 2007; Hukkanen et al., 2007). Anthocyanins are a major class of polyphenols producing fruit pigmentation and are strong antioxidants that contribute to the health-beneficial effect of strawberry. (Santos-Buelga and Scalbert, 2000; Gu et al., 2003). Strawberry fruits possess high in vitro antioxidative activity, which has been positively correlated with the content of polyphenolic compounds and, specifically, anthocyanins (Heinonen et al., 1998; Wang and Jiao, 2000; Wang and Lin, 2000). The physiological changes that contribute to strawberry fruit quality during development and maturation result from changes in gene expression and enzyme activities (Medina et al., 1997; Manning, 1998; Nam et al., 1999). The most noticeable changes involve alterations to fruit shape, size, texture, and pigmentation that coincide with an increase in the content of soluble solids and the production of natural aroma and flavour compounds (Perkins-Veazie, 1995). So far, great efforts have been made to investigate fruit composition and to explain the relationship between fruit quality and numerous factors, including, but not limited to, strawberry varieties (Wang and Lin, 2000; Määttä-Riihinen et al., 2004; Atkinson et al., 2006), developmental stage (Wang and Lin, 2000;), degree of maturity (Perez et al., 1992; Perkins-Veazie, 1995), climatic conditions (Anttonen et al., 2006), cultivation practices (Zabetakis and Holden, 1997; Anttonen et al., 2006; Atkinson et al., 2006), and biotic or abiotic stresses (Terry et al., 2007). These studies focused on either individual or just a few physiochemical properties, but failed to characterize the global changes to fruit biochemical composition. Metabolomic approaches have increasingly been used to gain insight into the metabolic composition of plant organs and to characterize the natural variance in metabolite content (Schauer and Fernie, 2006). Metabolomics can provide a diagnostic tool for better understanding of a biological system and has now been successfully performed on a diverse array of plant species, including models such as Arabidopsis (Kim et al., 2007), Medicago (Chen et al., 2003; Broeckling et al., 2005), and tobacco (Blount et al., 2002), and the important crops tomato (Roessner-Tunali et al., 2003; Schauer et al., 2005), potato (Roessner et al., 2001), rice (Sato et al., 2004), wheat (Hamzehzarghani et al., 2005), cucumber (Tagashira et al., 2005), and strawberry (Aharoni et al., 2002; Fait et al., 2008). In contrast to transcriptomics and proteomics, which rely to a great extent on genome information, metabolomics is mainly metabolite dependent. The cultivated varieties of commercial strawberries, usually designated as F.×ananassa, are almost all octoploids (Hancock, 1999); therefore, the whole genome sequence of strawberry is still unavailable, which hinders genetic research based on transcriptomics and proteomics (Aharoni et al., 2000). Consequently, it is essential to apply metabolomic approaches in strawberry research, which seems to have at least three advantages. First, strawberry fruit quality at harvest is a direct function of metabolite content, and the compositional analysis of plant metabolites is an established application in metabolomic research (Moco et al., 2007). Secondly, metabolomics relies on the analysis of the multitude of small molecules (metabolites) present in a biological system (Schauer and Fernie, 2006; Moco et al., 2007), which is helpful for the comprehensive understanding of the correlation between all the metabolites investigated and their metabolic network, instead of just a few. Thirdly, without the presence of genome information, metabolomics is a key tool in comprehensive functional genomics that aims to decipher gene function, investigate metabolic regulation, and analyse the systemic response to environmental or genetic perturbations (Schauer and Fernie, 2006). Aharoni et al. (2002) analysed four consecutive stages (green, white, turning, and red) of strawberry fruit development and identified the changes of fruit metabolites by use of Fourier transform ion cyclotron mass spectrometry. Metabolic networks were reconfigurated between the achene and the receptacle to investigate the cross-talk between primary and secondary metabolism during strawberry fruit development (Fait et al., 2008). The data revealed novel information on the metabolic transition from immature to ripe fruit, but failed to consider the investigation of volatile compounds that contribute to fruit flavour formation. The purpose of the present study was, therefore, to investigate the changes in metabolic composition from immature strawberry to ripe strawberry fruits that result in the formation of taste and nutritional quality. The biochemical changes were determined by gas chromatography–mass spectrometry (GC-MS) using either non-targeted or targeted quantitative profiling of both non-polar and polar extracts. Fruit volatile components were detected by non-targeted metabolome analysis in order to investigate fruit flavour formation. These data were complemented with high-performance liquid chromatography (HPLC) analyses of amino acids and anthocyanins. All the data underwent a variety of chemometric analyses, including principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA), to identify the true differences between samples. Both metabolic correlation analysis and metabolic network analysis based on pairwise coefficients were carried out to characterize the physiological response to developmental changes, to explore metabolic pathways, and to find bottlenecks or metabolic shifts in pathways during strawberry fruit development and ripening. Materials and methods Plant material and sample collection Strawberry (F.×ananassa Duch. cv. Troyonoka) plants were grown in a greenhouse in the Strawberry Germplasm Resource Garden of Anhui Agricultural University with a diurnal rhythm of 16 h light and 8 h darkness following normal cultivation practices. Fruit samples were collected from 4–5 individual plants beginning with the small green fruit stage [∼10 days post-anthesis (DPA)]. Between 10 and 18 fruits of the same maturation degree were harvested every 5 d, with sample collection ending at the over-ripe stage (5 d after the red-ripening stage). These fruit samples coincided with the fruit ripening stages of small green fruit (stage 1), large green fruit (stage 2), green–white fruit (stage 3), white fruit (stage 4), red-turning fruit (stage 5), red-ripening fruit (stage 6), and over-ripening fruit (stage 7) as shown in Fig. 1. After harvest, all the fruits from different ripening stages were washed in water, cut into quarters, pooled, and immediately frozen in liquid nitrogen and kept at –80 °C until use for sample preparation. Fig. 1. View largeDownload slide Growth and development of ‘Troyonoka’ strawberry fruits cultivated in a glass greenhouse. (a) Changes in fruit transverse diameter, vertical diameter, and fruit fresh weight. Mean of 10–18 fruits. Vertical bars represent the SD. (b) Stages of fruit development from flower (FL) to over-ripening fruit and the corresponding days after anthesis (DPA) separated into three phases of development characterized by cell division, cell expansion, and fruit ripening. (This figure is available in colour at JXB online.) Fig. 1. View largeDownload slide Growth and development of ‘Troyonoka’ strawberry fruits cultivated in a glass greenhouse. (a) Changes in fruit transverse diameter, vertical diameter, and fruit fresh weight. Mean of 10–18 fruits. Vertical bars represent the SD. (b) Stages of fruit development from flower (FL) to over-ripening fruit and the corresponding days after anthesis (DPA) separated into three phases of development characterized by cell division, cell expansion, and fruit ripening. (This figure is available in colour at JXB online.) Chemicals and reagents The standard samples of sugars and organic acids were of analytical reagent grade. Pyridine and cyclohexane were doubly distilled before use. Methoxamine hydrochloride was from the Institute of Biological Drugs Examination (Beijing, China). The reagent for trimethylsilylation, N,O-bis (trimethylsilyl)-trifluoroacetamide (BSTFA) containing 1% trimethylchlorosilane (TMCS), was purchased in 1 ml ampoules from Anpel Scientific Instrument Co., Ltd (Shanghai, China). A Waters ACCQ Fluor kit for HPLC analyses of amino acids was purchased from Waters Chemical Industry Co. (Milford, MA, USA). Authentic reference compounds of six anthocyanins, 3-O-β-glucopyranosides of delphindin (Dp3glc), 3-O-β-glucopyranosides of cyanidin (Cy3glc), 3-O-β-glucopyranosides of petunidin (Pt3glc), 3-O-β-glucopyranosides of pelargonidi (Pg3glc), 3-O-β-glucopyranosides of peonidin (Pn3glc), and 3-O-β-glucopyranosides of malvidin (Mv3glc), were kindly supplied as a gift from Dr Jianhui Cheng (Institute of Horticulture, Zhejiang Agricultural Academy, Hangzhou, China) and obtained from Polyphenols Lab (Sandnes, Norway). Methanol, acetonitrile, and trichloroacetic acid were all from TEDIA (Fairfield, OH, USA) and were of HPLC grade. Water was doubly distilled and further prepared with a Purelab Classic UF purification system from Pall Co. (New York, USA). Sample preparation Each of the frozen fruit samples (∼0.5 g) from the different ripening stages was ground into fine powder in liquid nitrogen. Each sample was mixed with 2 ml of trichloroacetic acid (0.6 mg ml−1) for sample quenching and then extracted three times with 100 ml of cold (–20 °C) petroleum ether for 2 h at 4 °C. The organic phase was evaporated under nitrogen and concentrated to 1 ml. After extraction with petroleum ether, the residue was successively extracted with cold cyclohexane following the same procedure as above. Both the petroleum ether extract and the cyclohexane extract were kept with anhydrous sodium sulphate to eliminate water before GC-MS analysis. All the samples were prepared in triplicate. The polar phase of the fruit sample was centrifuged at 12 000 g at 4 °C for 10 min. The fruit debris was extracted twice with 3 ml of pre-cooled methanol/water solution [1:1 (v/v) –20 °C]. The supernatant was pooled and adjusted to 5 ml with a suitable volume of the methanol/water solution. To correct for minor variations occurring during sample preparation and analysis, ribitol (0.2 mg ml−1 in water) was used as an internal standard. A 100 μl aliquot of the polar extract containing all hydrophilic metabolites was lyophilized under low temperature (–60 °C) in an LGJ-12 lyophilizer (Songyuan Huaxing Scientific Co., Beijing, China). Three analytical replicates were prepared for each sample. Two-step chemical derivatization was performed on the extracted hydrophilic metabolites according to the protocol of Roessner et al. (2000) with a few modifications. Briefly, oximation was carried out by dissolving the samples in 100 μl of methoxamine hydrochloride (25 mg ml−1 in pyridine) and incubating at 50 °C for 30 min. Samples were further derivatized with the addition of BSTFA containing 1% TMCS (100 μl) at 60 °C for 30 min to trimethylsilylate the polar functional groups. The derived samples were equilibrated to room temperature before injection. GC-MS analysis Capillary GC-MS analysis of the petroleum ether extract and the cyclohexane extract was performed with a SHIMADZU QP2010 series (Shimadzu Instruments, Japan) coupled to a Fisons MD800 quadrupole mass detector (Fisons Instruments, CA, USA) fitted with a split injector (1:10). Each sample (1 μl) was injected into the gas chromatograph through a fused-silica capillary column (30 m×0.25 mm i.d., 0.25 μm DB-5 MS stationary phase, J&W Scientific, Folsom, CA, USA). It was operated in constant pressure mode at 91 kPa and the injector temperature was 250 °C. The column temperature was held at 100 °C for 1 min; increased to 150 °C with a temperature gradient of 20 °C min−1; increased to 200 °C at 4 °C min−1; increased to 280 °C at 10 °C min−1; then held for 10 min. Helium (99.999%) was used as the carrier gas with the flow rate at 1 ml min−1. The significant MS operating parameters were as follows: ionization voltage was 70 eV (electron impact ionization), ion source temperature was 200 °C, and interface temperature was 250 °C. Ions were generated by a 70 eV electron beam at an ionization current of 40 μA. TIC (total ion current) spectra were recorded in the mass range of 40–600 atomic mass units (amu) in scanning mode. Shimadzu GCMS solution software (version 2.30) was used for data acquisition. Polar extract analysis by GC-MS was carried out using the same equipment as above, but with modifications in the GC program. The GC was programmed at an initial temperature of 100 °C for 2 min; increased to 184 °C at 3 °C min−1; increased to 200 °C at 0.5 °C min−1; held for 3 min; increased to 280 °C at 15 °C min−1; and finally held for 10 min. The temperatures of the injection port (splitless mode), interface, and MS source were 250, 250, and 200 °C, respectively. HPLC analysis on amino acids and anthocyanins For amino acid analysis, a pre-column derivatization method was used for sample preparation. Frozen fruit slices were homogenized in 20 μl of boiling water per mg of fresh weight of tissues and extracted at 100 °C for 30 min. After centrifugation at 11 000 g for 5 min, the fruit debris was discarded, and the extracts were immediately used for derivatization. A 10 μl aliquot of supernatant was transferred and mixed successively with 70 μl of borax buffer (pH 8.0) and 20 μl of ACCQ Fluor derivatization reagent (6-aminoquinolyl-N-hydroxyl-succinimidyl carbamate). The mixture was kept at 55 °C for 10 min and then subjected to HPLC analysis. The chromatographic analysis was performed on a C18 reverse phase column (Nova-Pak™, 250 mm×4.6 mm, particle size 5 μm) at 37 °C. The HPLC system was a Waters 600E (Milford, MA, USA) equipped with a binary 2489 UV detector and a 2475 fluorescence detector. The excitation and emission wavelengths of the fluorophotomeric detector were 250 nm and 395 nm, respectively, and the UV detector wavelength was set at 248 nm. The elution gradient consisted of acetate phosphate buffer (A), acetonitrile (B), and water (C), and the elution profile was: 0 min, 100% A; 18 min, 5% B; 19 min, 9% B; 29.5 min, 17% B; 33 min, 60% B, 40% C; 36 min, 100% A; 45 min, 60% B using linear gradients in between the time points. A 5 μl aliquot of each fruit sample was manually injected and the flow rate was maintained at 1 ml min−1. All data were processed using Waters Empower Station. For anthocyanin analysis, fine powders of each fruit sample (0.5 g) were extracted for 20 min with 5 ml of extraction buffer (acetone:water:formic acid 80:19.8:0.2) in the dark. The mixture was centrifuged at 11 000 g for 10 min and adjusted to 5 ml with extraction buffer. The supernatant was further filtered through a syringe-tip cellulose acetate filter (0.45 μm) prior to use. HPLC was performed on a Waters 600E system equipped with a 2487 UV detector. Chromatographic separation was carried out on a (250 mm×4.6 mm) Hyersil ODS 5 μm C18 HPLC column (Supelco Inc., Bellefonte, PA, USA). The stationary phase consisted of 5% formic acid in water (A), while the mobile phase was acetonitrile (B). The gradient program was linear from 0% to 20% B (0–13 min); 20–30% B (13–20 min); and 30–0% B (20–25 min). Prior to the next injection, the column was equilibrated for 30 min with 5% formic acid. The column temperature was maintained at 25 °C and the flow rate was maintained at 1 ml min−1. The chromatographic profile was recorded at 520 nm. All data were processed using Millinum32 software (Waters Co., MA, USA). Identification of metabolites detected by GC-MS and HPLC Recorded data profiles from GC-MS analysis were converted into NetCDF format and then exported to the MetAlign software package (RIKILT-WUR Institute of Food Safety, Wageningen, The Netherlands), where all data pre-treatment procedures including noise reduction, baseline correction, and peak alignment were carried out. Artefact peaks, such as column bleeding, and solvent peaks were removed. For metabolite identification and annotation, peaks were matched against customized reference spectrum databases including the National Institute of Standards and Technology (NIST) and the Wiley Registry, based on retention indices and mass spectral similarities. Some constituents, including the major sugars and organic acids, were further confirmed by comparing their mass spectra and retention indices with those of authentic reference compounds (Supplementary Figs S1, Supplementary Data available at JXB online). Metabolite contents were calculated by the internal standard method and independently normalized to plant mg of fresh weight and to internal references (n-heptadecane in the non-polar phase and ribitol in the polar phases) with each sample collection point. The chromatographic peaks in the polar extracts were identified and quantified by comparing their retention times and UV visible absorption spectra with those of authentic reference compounds. The amounts of each amino acid and anthocyanin were calculated using regression equations. The data presented in Supplementary Table S1 at JXB online are the average of the three replicate samples. The ratio of metabolite content at each sample collection point to the average content of each metabolite across seven developmental stages was shown by the colour scale. The heat map analysis was done on a Microsoft Excel-implemented macro program available at http://bitesizebio.com/2009/02/03/how-to-create-a-heatmap-in-excel. Multivariate analysis For individual metabolites, means ±SD were calculated from three biological replicates. Mean comparisons between the harvest stages were conducted by a Duncan's t-test using SPSS software version 13.0 (SPSS Institute, Chicago, IL, USA). A difference with P <0.05 was considered significant. The ratios calculated by the means of the metabolite contents were used to explain the up-regulation or down-regulation of identified metabolites. The GC-MS data matrix after peak annotation and normalization was used for multivariate analysis with SIMCA-P 11.5 software (Umetrics AB, Umeå, Sweden). PCA was used for an unsupervised analysis and PLS-DA for a supervised analysis. A PCA scatter plot of variables showing different metabolites was obtained to explain the separation of polar phase samples from different strawberry fruit stages. A range of metabolites was selected as the variable importance in the projection (VIP) based on PCA. Metabolite correlation and network analysis To visualize the correlation matrix, pairwise metabolite correlations were calculated by Pearson's correlation coefficient (rij) within and between 224 annotated peaks in the non-polar phase and 107 annotated peaks and 17 amino acids in the polar phase. The level of significance was set as |rMet| ≥0.85 for both the non-polar and polar phase. An |rMet| ≥0.80 was used for the correlation analysis between the non-polar phase and the polar phase. Correlation networks were reconstructed from the measured metabolite concentrations by pairwise combination of the metabolite concentrations, and 1111 pairwise analyses with strong dependencies (correlation coefficients: |rMet| ≥0.85) were selected and applied for metabolic network analysis. The network of pairwise metabolite correlation is drawn using the Kamada kawei-algorithm in Pajek (http://vlado.fmf.uni-lj.si/pub/networks/pajek/). In the correlation network, there is an edge between two metabolites if the absolute value of the correlation coefficient resulting from two concentration vectors exceeds a certain threshold |rMet|, and all the scattered vertexes are deleted. Publicly available data from the KEGG pathway database were obtained to understand and confirm the relationship between metabolite–metabolite correlations. Results Fruit metabolite screening by GC-MS Strawberry was grown in a greenhouse and different stages of fruits were harvested for metabolite analysis. These fruit samples were collected every 5 d as shown in Fig. 1. Initial GC-MS analyses of the non-polar and polar metabolites from all seven stages identified >280 resolved peaks, including the small green, large green, green–white, white, red-turning, red-ripening, and over-ripening fruit stages. Approximately 40% of these peaks could be identified as discrete metabolites with known chemical structure (Supplementary Table S1 at JXB online). These identified metabolites represent numerous metabolic pathways, including biosynthesis of isobutyl phthalate, isooctyl phthalate, and bis(2-ethylhexyl) phthalate, and the metabolism of related compounds [e.g. the tricarboxylic acid cycle (TCA) and the shikimate pathway]. The presence of 2,6-dimethyl-α-D-galactopyranoside and 1-methyl-α-D-galactopyranoside, which are considered possible precursors for compounds responsible for strawberry fruit aroma, was also noted. These compounds have been reported previously in flavour components of grape fruit (Mateo et al., 1997) and oolong tea (Guo et al., 1993). In the petroleum ether extracts of red-ripening fruits, 21 components from 35 chromatographic peaks were identified. Through peak area integration, the relative amounts of isobutyl phthalate, isooctyl phthalate, 1,2-benzenedicarboxylic-bis (2-methylpropyl) ester, and dibutyl phthalate were 856.44, 1336.54, 269.71, and 168.89 μg g−1, respectively. Around 615.76 μg g−1 of fatty acids and alcohols were represented by n-hexadecanoic acid; 9,12-(Z, Z)-octadecadienoic acid; 3,7,11-trimethyl-1,6,10-dodecatrien-3-ol; and 9,12,15-(Z, Z, Z)-octadecatrien-1-ol, which have an important influence on the flavour of strawberry fruit (Fig. 2; Supplementary Table S1 at JXB online). Fig. 2. View largeDownload slide GC-MS analysis on petroleum ether extract from strawberry fruits (TIC). Labelled peaks are: (1) 3,7,11-trimethyl-1,6,10-dodecatrien-3-ol; (2) isobutyl phthalate; (3) 1,2-benzenedicarboxylic-bis(2-methylpropyl) ester; (4) n-hexadecanoic acid; (5) dibutyl phthalate; (6) 9,12-(Z, Z)-octadecadienoic acid; (7) 9,12,15-(Z, Z, Z)-octadecatrien-1-ol; (8) isooctyl phthalate; IS, internal standard. Fig. 2. View largeDownload slide GC-MS analysis on petroleum ether extract from strawberry fruits (TIC). Labelled peaks are: (1) 3,7,11-trimethyl-1,6,10-dodecatrien-3-ol; (2) isobutyl phthalate; (3) 1,2-benzenedicarboxylic-bis(2-methylpropyl) ester; (4) n-hexadecanoic acid; (5) dibutyl phthalate; (6) 9,12-(Z, Z)-octadecadienoic acid; (7) 9,12,15-(Z, Z, Z)-octadecatrien-1-ol; (8) isooctyl phthalate; IS, internal standard. In the cyclohexane extract of red-ripening strawberry fruits, 107 volatile metabolites from 160 chromatographic peaks were identified using the NIST and Wiley databases. The most relatively abundant metabolites were alkanes, esters, and alcohols, including hexadecane (643.36 μg g−1), n-tetradecane (52.23 μg g−1), nonadecane (49.20 μg g−1), n-docosane (69.77 μg g−1), n-tricosane (29.71 μg g−1), n-octadecane (48.01 μg g−1), undecane (39.63 μg g−1), eicosane (25.59 μg g−1), n-dodecane (11.20 μg g−1), isobutyl phthalate (453.08 μg g−1), bis(2-ethylhexyl) phthalate (716.13 μg g−1), 1,2-benzenedicarboxylic acid, butyl 8-methylnonyl ester (69.48 μg g−1), 5-methyl-1-heptanol (118.98 μg g−1), and 8-dodecen-1-ol (5.83 μg g−1) (Fig. 3; Supplementary Table S1 at JXB online). Comparing compounds identified in this study with the perceived aroma characteristics of different volatile compounds reveals that strawberry aroma is the result of the combined perception of fruity, green, sweaty, peach-like, and caramel-like flavour (Larsen and Poll, 1992). Fig. 3. View largeDownload slide GC-MS analysis on cyclohexane extract from strawberry fruits (TIC). (1) n-Dodecane; (2) 8-dodecen-1-ol; (3) 5-methyl-1-heptanol; (4) 2,6,10,14-tetramethyl-heptadecane; (5) n-tetradecane; (6) hexadecane; (7) n-tricosane; (8) nonadecane; (9) isobutyl phthalate; (10) n-docosane; (11) 1,2-benzenedicarboxylic acid, butyl 8-methylnonyl ester; (12) n-octadecane; (13) eicosane; (14) undecane; (15) bis(2-ethylhexyl) phthalate; IS, internal standard. Fig. 3. View largeDownload slide GC-MS analysis on cyclohexane extract from strawberry fruits (TIC). (1) n-Dodecane; (2) 8-dodecen-1-ol; (3) 5-methyl-1-heptanol; (4) 2,6,10,14-tetramethyl-heptadecane; (5) n-tetradecane; (6) hexadecane; (7) n-tricosane; (8) nonadecane; (9) isobutyl phthalate; (10) n-docosane; (11) 1,2-benzenedicarboxylic acid, butyl 8-methylnonyl ester; (12) n-octadecane; (13) eicosane; (14) undecane; (15) bis(2-ethylhexyl) phthalate; IS, internal standard. Major metabolites in the polar metabolic extracts of strawberry fruits are sugars, glucosides, sugar alcohols, and organic acids (Fig. 4), which showed significant differences in levels during fruit growth and ripening. For quantifications, oximation and derivatization treatments were applied. GC-MS profiling analysis on red-ripening strawberry fruits showed that the average contents of fructose, sucrose, galactose, and turanose were 14.76, 9.80, 7.75, and 1.87 mg g−1, respectively, while those of glucose and palatinose were 210.42 μg g−1 and 190.17  μg g−1, respectively. Organic acids, primarily comprised of citric acid and malic acid, were 7.74 mg g−1 and 4.22 mg g−1, respectively, while other compounds (e.g. 2-keto-D-gluconic acid, palmitic acid, n-hexadecanoic acid, arabinonic acid, butanedioic acid, gulonic acid, gluconic acid, mannonic acid, and 2-hydroxy- ethyl-sulphonic acid) made up ∼2% of the identified polar metabolites. Myo-inositol-1-phosphate was at 1.37 mg g−1. Fig. 4. View largeDownload slide GC-MS analysis on polar metabolites from strawberry fruits (TIC). (1) Butanedioic acid; (2) 2,5-dimethyl-4-hydroxy-3(2H)-furanone; (3) malic acid; (4) arabinonic acid; (5) gluconic acid; (6) mannonic acid; (7) 2-keto-D-gluconic acid; (8) citric acid; (9) 2-hydroxyethylsulphonic acid; (10) gulonic acid; (11) glucose; (12 and 13) fructose; (14 and 16) galactose; (15) palmitic acid; (17) myo-inositol-1-phosphate; (18) hexadecanoic acid; (19) sucrose; (20) turanose; (21) palatinose; IS, internal standard. Fig. 4. View largeDownload slide GC-MS analysis on polar metabolites from strawberry fruits (TIC). (1) Butanedioic acid; (2) 2,5-dimethyl-4-hydroxy-3(2H)-furanone; (3) malic acid; (4) arabinonic acid; (5) gluconic acid; (6) mannonic acid; (7) 2-keto-D-gluconic acid; (8) citric acid; (9) 2-hydroxyethylsulphonic acid; (10) gulonic acid; (11) glucose; (12 and 13) fructose; (14 and 16) galactose; (15) palmitic acid; (17) myo-inositol-1-phosphate; (18) hexadecanoic acid; (19) sucrose; (20) turanose; (21) palatinose; IS, internal standard. Untargeted profiling of metabolites by GC-MS The parallel analysis of non-polar and polar extracts using GC-MS enabled the detection of a large number of compounds of different classes, mainly alkanes, alcohols, esters, sugars and sugar alcohols, organic acids, and fatty acids (Supplementary Table S1 at JXB online). The heat map showed that the metabolites have a differential distribution during development and maturation. Almost all alkanes exhibited significance increases during fruit development, while the accumulation pattern of each metabolite was very different. Some alkanes (e.g. tridecane, tetradecane, pentadecane, 2,6,10-trimethyltetradecane, octadecane, 2,6,10,14-tetramethyl-hexadecane, and nonadecane) began to accumulate at the red-turning stage, and were maintained at high levels; others increased through stages 4–6 initially but decreased thereafter (Fig. 5). Most of the alcohols and esters reached the highest levels in red-turning and red-ripening fruits, and these contributed to the formation of fruity, sweaty, and peach-like flavour in strawberry fruits (Larsen and Poll, 1992). Sugars, organic acids, and fatty acids were the metabolites that exhibited a high degree of variance during fruit development (Fig. 5; Supplementary Table S1 at JXB online), while fruit ripening was characterized by the increases in levels of fructose, sucrose, galactose, citric acid, and malic acid, which comprised the major soluble sugars and organic acids in strawberry fruits. Fig. 5. View large Download slide Metabolite levels of representatives of major metabolite groups which changed significantly during fruit development. The amounts were normalized to the fresh weight of each developmental stage (μg g−1) and are provided in Supplementary Table S1 at JXB online. The ratio of metabolite content at each sample collection point to the average metabolite content across seven developmental stages is shown by the colour scale. The lowest ratios are in dark blue, and the highest ratio in dark red. S1–S7 refer to the seven developmental stages of strawberry fruits shown in Fig. 1. (This figure is available in colour at JXB online.) Fig. 5. View large Download slide Metabolite levels of representatives of major metabolite groups which changed significantly during fruit development. The amounts were normalized to the fresh weight of each developmental stage (μg g−1) and are provided in Supplementary Table S1 at JXB online. The ratio of metabolite content at each sample collection point to the average metabolite content across seven developmental stages is shown by the colour scale. The lowest ratios are in dark blue, and the highest ratio in dark red. S1–S7 refer to the seven developmental stages of strawberry fruits shown in Fig. 1. (This figure is available in colour at JXB online.) Cluster analysis allowed the distinction between samples that were collected at the different stages. Important metabolites in these samples that affected cluster formation could be determined. Multivariate statistical analysis PCA and PLS-DA were applied to non-polar and polar extracts separately. Table 1 lists the top 20 metabolites that influenced cluster formation within the non-polar metabolite profiles, which were generated from PLS-DA analysis of the cyclohexane extract. These compounds include alkanes, aldehydes, and esters (Table 1 and Fig. 7). A different cluster pattern from the polar metabolite extracts was observed through PLS-DA analysis. Specifically, five metabolite profiles from stage 3 to stage 7 fruit samples clearly separated from each other, forming distinct clusters (Fig. 6). Stage 1 and stage 2 fruit profiles (both green fruit) clustered together and were clearly separated from the five ripening profiles (Fig. 6). Table 1. Compounds determined to be of variable importance in the projection through PLS-DA analysis on the identified compounds from the cyclohexane extract No.  RT (min)  Name  VIP  1  65.008  n-Tetradecane  1.723  2  72.558  Tridecanal  1.671  3  42.858  n-Dodecane  1.663  4  79.217  Nonacosane  1.660  5  71.687  2-Methyl-octadecane  1.650  6  69.008  Hexadecane  1.618  7  79.425  1,2-Benzenedicarboxylic acid, butyl 8-methylnonyl ester  1.617  8  66.991  Isozonarol  1.612  9  72.814  Nonadecane  1.598  10  77.784  n-Docosane  1.576  11  71.000  n-Tricosane  1.541  12  84.892  Eicosane  1.537  13  94.408  2-Octadecyloxy ethyl ester  1.531  14  74.159  Methyl α-ketopalmitate  1.516  15  91.159  n-Heptacosane  1.513  16  80.300  n-Octadecane  1.512  17  92.217  Bis(2-ethylhexyl) phthalate  1.462  18  85.367  Undecane  1.451  19  69.348  Heneicosane  1.435  20  76.035  Isobutyl phthalate  1.409  No.  RT (min)  Name  VIP  1  65.008  n-Tetradecane  1.723  2  72.558  Tridecanal  1.671  3  42.858  n-Dodecane  1.663  4  79.217  Nonacosane  1.660  5  71.687  2-Methyl-octadecane  1.650  6  69.008  Hexadecane  1.618  7  79.425  1,2-Benzenedicarboxylic acid, butyl 8-methylnonyl ester  1.617  8  66.991  Isozonarol  1.612  9  72.814  Nonadecane  1.598  10  77.784  n-Docosane  1.576  11  71.000  n-Tricosane  1.541  12  84.892  Eicosane  1.537  13  94.408  2-Octadecyloxy ethyl ester  1.531  14  74.159  Methyl α-ketopalmitate  1.516  15  91.159  n-Heptacosane  1.513  16  80.300  n-Octadecane  1.512  17  92.217  Bis(2-ethylhexyl) phthalate  1.462  18  85.367  Undecane  1.451  19  69.348  Heneicosane  1.435  20  76.035  Isobutyl phthalate  1.409  RT, retention time; VIP, variable importance in the projection. View Large Fig. 6. View largeDownload slide PLS-DA score plot of polar phase samples from different strawberry fruit stages. The coloured dots represent polar phase samples from different strawberry fruit stages. Seven independent time course studies, each with triplicate samples taken for extraction at each time point, were used in the analysis, for a total of 21 data points. (This figure is available in colour at JXB online.) Fig. 6. View largeDownload slide PLS-DA score plot of polar phase samples from different strawberry fruit stages. The coloured dots represent polar phase samples from different strawberry fruit stages. Seven independent time course studies, each with triplicate samples taken for extraction at each time point, were used in the analysis, for a total of 21 data points. (This figure is available in colour at JXB online.) Fig. 7. View largeDownload slide Key compounds identified from the PLS-DA (VIPs) showed major differences over fruit maturation. Metabolites from the non-polar phase and polar phase were detected by GC-MS. The bars represent metabolite contents across the seven different developmental stages (S1–S7) of strawberry fruits. The amounts are normalized to the fresh weight of each developmental stage (μg g−1). The error bars are the standard deviations from three biological repeats. Fig. 7. View largeDownload slide Key compounds identified from the PLS-DA (VIPs) showed major differences over fruit maturation. Metabolites from the non-polar phase and polar phase were detected by GC-MS. The bars represent metabolite contents across the seven different developmental stages (S1–S7) of strawberry fruits. The amounts are normalized to the fresh weight of each developmental stage (μg g−1). The error bars are the standard deviations from three biological repeats. PCA uses an n-dimensional vector approach to separate samples based on the cumulative correlation of all data. The resulting vectors that yield the greatest separation between samples are identified and then used to calculate their factor scores (Roessner et al., 2001). The first two highest-ranked vectors, which represent the major variance among samples, were then displayed in two-dimensional plots. A broad range of metabolites were found to influence cluster formation of polar metabolite profiles, including sugars, glycosides, sugar alcohols, organic acids, and 2-deoxyerythropentono-1,4-lactone (Supplementary Fig. S3 at JXB online; Table 2). The accumulation profiles of VIPs in the polar phase showed that strawberries underwent major changes in carbohydrate concentrations during fruit development (Fig. 7). Table 2. Compounds determined to be of variable importance in the projection through PLS-DA analysis on the identified compounds from the polar phase extract No.  RT (min)  Name  VIP  1  18.325  2-Deoxyerythropentono-1,4-lactone  1.799  2  21.442  Ethanedioic acid  1.668  3  24.283  Gluconic acid  1.609  4  71.523  Turanose  1.578  5  8.675  1-Methyl-α-D-galactopyranoside  1.569  6  72.219  Sucrose  1.557  7  38.242  β-D-Glucopyranose  1.539  8  70.685  Turanose  1.528  9  27.550  Mannonic acid  1.468  10  24.008  Monoamidoethylmalonic acid  1.444  11  5.475  1-Methyl-α-D-galactopyranoside  1.433  12  18.117  2,6-di-Methyl-α-D-galactopyranoside  1.431  13  27.192  Pentonic acid  1.294  14  40.752  Glucohexodialdose  1.293  15  41.183  Threitol  1.281  16  73.167  Scopolin  1.269  17  36.425  Fructose  1.252  18  29.817  Butanoic acid  1.223  19  40.147  Sorbit  1.196  20  69.575  Turanose  1.185  No.  RT (min)  Name  VIP  1  18.325  2-Deoxyerythropentono-1,4-lactone  1.799  2  21.442  Ethanedioic acid  1.668  3  24.283  Gluconic acid  1.609  4  71.523  Turanose  1.578  5  8.675  1-Methyl-α-D-galactopyranoside  1.569  6  72.219  Sucrose  1.557  7  38.242  β-D-Glucopyranose  1.539  8  70.685  Turanose  1.528  9  27.550  Mannonic acid  1.468  10  24.008  Monoamidoethylmalonic acid  1.444  11  5.475  1-Methyl-α-D-galactopyranoside  1.433  12  18.117  2,6-di-Methyl-α-D-galactopyranoside  1.431  13  27.192  Pentonic acid  1.294  14  40.752  Glucohexodialdose  1.293  15  41.183  Threitol  1.281  16  73.167  Scopolin  1.269  17  36.425  Fructose  1.252  18  29.817  Butanoic acid  1.223  19  40.147  Sorbit  1.196  20  69.575  Turanose  1.185  RT, retention time; VIP, variable importance in the projection. View Large Targeted profiling of amino acids and anthocyanins by HPLC Free amino acids not only participate in a wide range of physiological reactions, especially in carbon–nitrogen metabolism in fruits, but also contribute to fruit quality formation. HPLC coupled with fluorescence detection was used to investigate the changes of the main amino acid components during the development of strawberry fruits. The amino acid components in strawberry fruits can be well separated (Supplementary Fig. S4 at JXB online). Serine, arginine, glutamic acid, histidine, aspartic acid, and proline were all present at significantly high levels in strawberry fruits (Supplementary Table S1). PLS-DA analysis on the profiles of amino acids gave a clear cluster pattern (Fig. 8). During the development of strawberry fruits, the total content of amino acids decreased gradually, from 4911.91 μg g−1 in stage 1 to 894.04 μg g−1 fresh weight in stage 6 (Supplementary Table S1). During stage 1 to stage 6, the level of each amino acid component decreased (Supplementary Table S1; Fig. 5). However, the levels of most amino acids showed a significant increase in stage 7 when compared with stage 6, which might be related to cell degradation in over-ripening fruits (Fig. 9B). Fig. 8. View largeDownload slide PLS-DA score plot of amino acid extracts from different strawberry fruit stages. The coloured dots represent polar phase samples from different strawberry fruit stages. Seven independent time course studies, each with triplicate samples taken for extraction at each time point, were used in the analysis, for a total of 21 data points. (This figure is available in colour at JXB online.) Fig. 8. View largeDownload slide PLS-DA score plot of amino acid extracts from different strawberry fruit stages. The coloured dots represent polar phase samples from different strawberry fruit stages. Seven independent time course studies, each with triplicate samples taken for extraction at each time point, were used in the analysis, for a total of 21 data points. (This figure is available in colour at JXB online.) Fig. 9. View largeDownload slide Changes in the levels of metabolites during fruit development shown in a metabolic diagram. (A) The ratios between stage 6 and stage 5. (B) The ratios between stage 7 and stage 6. The changes in metabolite contents were calculated by dividing the metabolite level in fruit ripening stages 5, 6, and 7. The level of significance was set at P <0.05. The metabolites with grey characters were undetectable. 3-PGA, 3-phosphoglycerate; AA, ascorbate; AbuA, 4-aminobutyrate; Ala, alanine; Arg, arginine; Asn, asparagine; Asp, aspartate; CitA, citric acid; Cy3glc, 3-O-β-glucopyranosides of cyanidin; Cys, cysteine; DhAA, dehydroascorbate; Ery, erythrose; Fru, fructose; G6P, glucose-6-phosphate; Gal, galactose; Gcn, gluconic acid; Glc, glucose; Gln, glutamine; Glu, glutamate; Gly, glycine; Gly3P, glycerol-3-phosphate; GtA, galacturonate; HoP, hydroxyproline; hSer, homoserine; Ile, isoleucine; Ino, myo-inositol; Ino1P, myo-inositol-1-phosphate; Leu, leucine; Lys, lysine; Mal, maltose; MalA, malic acid; Man, mannose; Met, methionine; Mnt, mannitol; OAS, O-acetyl serine; OgA, α-oxoglutarate; OxP, 5-oxoproline; PEP, phosphoenolpyruvic acid; Pg3glc, 3-O-β-glucopyranosides of Pelargonidi; Phe, phenylalanine; Pro, proline; Put, putrescine; Raf, raffinose; Rha, rhamnose; Sbt, D-sorbitol; Ser, serine; ShikA, shikimate; SMC, S-Me-cysteine; Spd, spermidine; Suc, sucrose; SucA, succinate; Thr, threonine; ThrA, threonate; Top, α-tocopherol; Tre, trehalose; Trp, tryptophan; Tyr, tyrosine; Val, valine. (This figure is available in colour at JXB online.) Fig. 9. View largeDownload slide Changes in the levels of metabolites during fruit development shown in a metabolic diagram. (A) The ratios between stage 6 and stage 5. (B) The ratios between stage 7 and stage 6. The changes in metabolite contents were calculated by dividing the metabolite level in fruit ripening stages 5, 6, and 7. The level of significance was set at P <0.05. The metabolites with grey characters were undetectable. 3-PGA, 3-phosphoglycerate; AA, ascorbate; AbuA, 4-aminobutyrate; Ala, alanine; Arg, arginine; Asn, asparagine; Asp, aspartate; CitA, citric acid; Cy3glc, 3-O-β-glucopyranosides of cyanidin; Cys, cysteine; DhAA, dehydroascorbate; Ery, erythrose; Fru, fructose; G6P, glucose-6-phosphate; Gal, galactose; Gcn, gluconic acid; Glc, glucose; Gln, glutamine; Glu, glutamate; Gly, glycine; Gly3P, glycerol-3-phosphate; GtA, galacturonate; HoP, hydroxyproline; hSer, homoserine; Ile, isoleucine; Ino, myo-inositol; Ino1P, myo-inositol-1-phosphate; Leu, leucine; Lys, lysine; Mal, maltose; MalA, malic acid; Man, mannose; Met, methionine; Mnt, mannitol; OAS, O-acetyl serine; OgA, α-oxoglutarate; OxP, 5-oxoproline; PEP, phosphoenolpyruvic acid; Pg3glc, 3-O-β-glucopyranosides of Pelargonidi; Phe, phenylalanine; Pro, proline; Put, putrescine; Raf, raffinose; Rha, rhamnose; Sbt, D-sorbitol; Ser, serine; ShikA, shikimate; SMC, S-Me-cysteine; Spd, spermidine; Suc, sucrose; SucA, succinate; Thr, threonine; ThrA, threonate; Top, α-tocopherol; Tre, trehalose; Trp, tryptophan; Tyr, tyrosine; Val, valine. (This figure is available in colour at JXB online.) In previous studies, 25 anthocyanin pigments were detected from five different strawberry cultivars, with most cultivars containing pelargonidin (Pg) and cyanidin (Cy) (Lopes et al., 2007). It was found that the major anthocyanin in the ripening fruits of cultivar ‘Troyonoka’ is Pg3glc, which began to accumulate in red-turning fruits (Supplementary Table S1 at JXB online). A significant increase (30.4-fold) was observed in red-ripening fruits (stage 6). The level of Pg3glc continued to increase (1.2-fold) in over-ripening fruits (stage 7) as compared with stage 6 (Fig. 9). The smaller proportions of Cy3glc varied drastically from stage 5 to stage 7. Cy3glc showed a large increase (99.4-fold) between stage 5 and stage 6, then decreased significantly (0.5-fold) by stage 7 (as compared with stage 6). Pn3glc and Mv3glc were detected only in red-ripening fruits in a minute quantity (Supplementary Table S1). Notable variability was found among the anthocyanin concentrations in fruit samples at different harvest times, indicating a strong influence from the degree of maturity of the strawberry fruits on organoleptic quality. Metabolite changes in red-turning and over-ripening stages of strawberry fruits In order to identify the metabolic pathways that modulate metabolite levels linked to peak market quality, significant metabolite changes (P <0.05) observed between either stage 5 or stage 7 and stage 6 were highlighted on the metabolic map (Fig. 9). In stage 6, significant increases in fructose (1.2-fold), sucrose (1.3-fold), galactose (1.5-fold), citric acid (1.5-fold), and malic acid (1.5-fold), which are the main polar metabolites contributing to quality, were found. Up-regulation was also observed in maltose (5.5-fold), glucose (1.6-fold), myo-inositol-1-phosphate (1.9-fold), and glucose-6-phosphate (1.5-fold) (Fig. 9A). In contrast, there was a significant decrease in the levels of glycerol (0.7-fold), rhamnose (0.8-fold), erythrose (0.7-fold), gluconic acid (0.6-fold), and sorbitol (0.5-fold). The levels of most of the amino acids also decreased from stage 5 to stage 6. The development through stage 6 into stage 7 results in an almost complete metabolic reversal. The levels of major sugars and organic acids, such as fructose (0.9-fold), galactose (0.8-fold), citric acid (0.7-fold), and malic acid (0.7-fold), showed significant decreases (Fig. 9B). In addition, myo-inositol-1-phosphate (0.7-fold), myo-inositol (0.7-fold), and glycerol-3-phosphate (0.6-fold) were sharply decreased in stage 7. The levels of sucrose and some amino acids, such as serine, cysteine, leucine, proline, aspartate, arginine, and lysine, remain the same. As shown above, most other amino acids showed a significant increase in stage 7, which suggests that cell degradation occurs in over-ripening fruits (Knee et al., 1977). Correlation of metabolite levels and analysis of dependencies of non-polar metabolites Correlation analysis is a useful tool to explore metabolic pathways and networks (Raamsdonk et al., 2001; Roessner et al., 2001; Fiehn, 2003; Steuer et al., 2003a, b; Weckwerth et al., 2004). In order to delineate metabolic changes relevant to fruit ripening, the pairwise correlation for each metabolite at different stages was analysed against every other metabolite within the samples. For non-polar compounds, strong correlations (|rMet| ≥0.8) were found in three main metabolites [hexadecane, isobutyl phthalate, and bis(2-ethylhexyl) phthalate] from the cyclohexane extract (Supplementary Table S2 at JXB online). These relationships could be classified into three types: (i) those between compounds having similar chemical structures or belonging to the same chemical class, such as alkanes (e.g. n-dodecane, n-etradecane, hexadecane, nonadecane, and octadecane), and esters (e.g. isobutyl phthalate, 1,2-benzenedicarboxylic acid, and butyl 8-methylnonyl ester); (ii) those between compounds having different chemical structures, but that are known to be connected by close biosynthetic relationships (e.g. 2-ethylhexyl phthalate and 2-ethyl-1-hexanol, because 2-ethyl-1-hexanol is used as the precursor in the synthesis of 2-ethylhexyl phthalate); and (iii) those that seem unrelated based on available knowledge (e.g. isobutyl phthalate and alkanes) (Supplementary Table S2). Correlation analysis of polar metabolites In the polar extract of strawberry fruits, 107 polar metabolites identified from untargeted metabolic profiling analysis by GC-MS and 17 amino acids from targeted metabolic profiling analysis by HPLC were subjected to correlation analysis. In total 7626 pairwise analyses and their Pearson correlation coefficients (rij) were computed. Metabolites with correlation coefficients (|rMet| ≥0.85) were set as strong dependencies and 1111 pairwise analyses were selected (14.57% of total correlations). Overall, both positive and negative dependencies were observed; the most significant correlations were positive among these polar metabolites, albeit few in number. Most of the coefficients were negative. The correlation pairs of polar metabolites were categorized into similar groups as described above with respect to biosynthetic relationships: (i) metabolite pairs placed close to each other in a metabolic pathway for the concerned reactions, for example the pair of sucrose:fructose (r=0.944, P=0.14%), sucrose:glucose (r=0.956, P=0.08%), citric acid:malic acid (in the TCA cycle, r=0.989, P=0.02%), and phenylalanine:tyrosine (in the shikimate pathway, r=0.951, P=0.10%); (ii) metabolite pairs with a similar chemical skeleton derived from the same biosynthetic pathway, for example the pairs galactose:1-methyl-α-D-galactopyranoside (r=0.852, P=1.48%) and glucose:glucopyranoside (r=0.866, P=1.17%); and (iii) metabolite pairs located at a distance in a metabolic network, for example the pairs serine:glutamate (r=0.956, P=0.08%) and glutamic acid:arginine (r=0.980, P=0.01%) in arginine metabolism (Supplementary Fig. S5 at JXB online). Correlation analysis between non-polar metabolites and polar metabolites Although the non-polar metabolites and polar metabolites were measured in two separate chromatographic analyses for any given time point, they came from the same tissue sample. It is therefore possible to make connections between these profiles by correlation analysis. Pairwise metabolite correlations between 224 non-polar metabolites and 124 polar metabolites (107 annotated peaks and 17 amino acids in the polar phase) were calculated by Pearson's correlation coefficient, with the threshold set as |rMet| ≥0.80. It was found that isobutyl phthalate showed strong positive correlations with mannonic acid (r=0.885, P=0.08%) and glucohexodialdose (r=0.842, P=1.74%). Bis(2-ethylhexyl) phthalate exhibited strong positive correlations with 2,3,4-trihydroxybutyric acid (r=0.897, P=0.62%), turanose (r=0.886, P=0.79%), sucrose (r=0.927, P=0.27%), and palatinose (r=0.849, P=1.56%). 5-Methyl-1-heptanol strongly correlated with ethanedioic acid (r=0.937, P=0.18%), fructose (r=0.845, P=1.66%), turanose (r=0.900, P=0.57%), sucrose (r=0.991, P=0.01%), and palatinose (r=0.972, P=0.01%). Fructose, galactose, and malic acid showed strong correlations with 2-butyl-1-octanol, with the correlates 0.909 (P=0.05%), 0.821 (P=2.35%), and 0.865 (P=1.20%), respectively. Fructose, malic acid, and citric acid all have positive correlations with n-octadecene, with the correlates 0.810 (P=2.73%), 0.879 (P=0.09%), and 0.883 (P=0.08%), respectively. The data presented here provide the correlative information that may, with further experimentation, allow the elucidation of the linkage between the biosynthesis pathways of sugars, organic acids, alcohols, esters, and alkenes. Metabolic network analysis In addition to correlation analysis, network analysis based on a correlation matrix was carried out to visualize the network and focus on a few metabolites of interest. This network analysis allowed the definition of the key points at which metabolism changed in metabolic networks using the connectivity matrix (Weckwerth et al., 2004). After correlation analysis of polar metabolites and amino acids, 1111 pairwise correlations with strong dependencies (correlation coefficients: |rMet| ≥0.85) were picked and applied for metabolic network analysis. Supplementary Figure S5 at JXB online presents the metabolite correlation network in the polar phase extract of strawberry fruits, and each vertex corresponds to a metabolite. All the scattered vertexes are deleted and the metabolites with significant correlations are linked together. Fructose, galactose, citric acid, and malic acid, the four main polar metabolites closely related to ripe fruit, exhibited strong positive correlations with the following metabolites: sucrose, glucose, ribose, turanose, glucohexodialdose, palatinose, glucopyranoside, 1-methyl-α-D-galactopyranoside, mannonic acid, and arabinonic acid; but had negative correlations with serine, glutamic acid, threonine, and isoleucine. Moreover, fructose showed strong positive correlations with other metabolites, including erythrose, hexadecanoic acid, and scopolin, and negative correlations with glycine and histidine. Fructose was placed close to erythrose in the metabolic network because fructose 6-phosphate can be converted into erythrose 4-phosphate by transketolase. Galactose had strong positive correlations with 2-keto-D-gluconic acid, threitol, D-glycero-L-manno-heptonic acid, idonic acid, and scopolin, but had negative correlations with glycine, histidine, arginine, and methionine. Citric acid exhibited strong positive correlations with 2-hydroxybutanoic acid, 5-hydroxymethyl-2-furoic acid, 2-keto-D-gluconic acid, threitol, 3,4,5-trihydroxy-pentanoic acid, allonic acid, D-glycero-L-manno-heptonic acid, idonic acid, and scopolin, but had negative correlations with glycine, histidine, arginine, alanine, methionine, and lysine. Malic acid showed strong positive correlations with citric acid, 2-hydroxybutanoic acid, 5-hydroxymethyl-2-furoic acid, 2-keto-D-gluconic acid, threitol, 3,4,5-trihydroxypentanoic acid, idonic acid, and scopolin. Sucrose showed strong positive correlations with ribose, 2-keto-D-gluconic acid, D-glycero-L-manno-heptonic acid, fructose, glucose, galactose, arabinofuranose, glucopyranoside, D-glycero-L-manno-heptonic acid, mannonic acid, scopolin, and palatinose. Glycerol 3-phosphate and glycerol were placed close to each other in the metabolic network, because glycerol 3-phosphate is produced from glycerol by the enzyme glycerol kinase and the conversion of glycerol 3-phosphate to glycerol is catalysed by glycerol 3-phosphatase. Obviously, these data suggested that during fruit maturation, carbohydrate biosynthesis increased only when free amino acid accumulation decreased. Discussion Several promising functional genomic approaches, including transcriptomics, proteomics, and metabolomics, are employed in the systematic and comprehensive understanding of the complex events of life. Currently, whole genome information of strawberry, an octoploid species, is unavailable, restricting transcriptomic and proteomic studies. Metabolomics can be independent of genomic data, identifying sample constituents and the actual biochemical status of the tissues. Despite the wide application of metabolomics in plant research, there are few published reports on the metabolite profiles of strawberry fruits. These reports either described non-targeted metabolomic analysis of dissected strawberry floral organs by the use of UPLC-qTOF-MS (Hanhineva et al., 2008) or were focused on the metabolite composition of a fruit polar extract by use of Fourier transform ion cyclotron mass spectrometry (Aharoni et al., 2002) and GC-MS (Aprea et al., 2009). Thus, an extensive and comprehensive metabolite profile in strawberry fruits (including non-polar and polar phases) has not been described previously. This study provides a comprehensive, comparative analysis of the metabolite composition of strawberry fruits from seven different developmental stages. The resulting information is important for investigating the changes of key nutritional metabolites and for illustrating the mechanism of fruit quality formation. A derivatization step is often necessary before GC-MS analysis in order to improve the volatility of polar metabolites. At the same time, due to the cyclic and open chain structures of sugars, silylation of monosaccharides without an oximation step leads to multiple peaks belonging to each individual sugar compound. By introducing an oximation step prior to silylation, cyclization is inhibited, resulting in fewer peaks per sugar. However, trimethylsilylation of some polar metabolites in this study still gave 2–3 peaks, such as in the case of fructose, galactose, and turanose (Figs 5, 7; Supplementary Table S1 at JXB online). When data from polar extracts and non-polar extracts were combined, stage 1 (small green fruit) and stage 2 (large green fruit) have the most similar metabolite content of the seven developmental stages (Fig. 6). It is probable that both stages actually belong to the same fruit developmental phase (the cell division phase). After that, during 15–35 DPA, fruits were characterized by the accumulation of soluble sugars and organic acids, which contribute to the acquisition of the fleshy trait associated with cell expansion. Significant distinctions of metabolite composition between stage 4 (white fruit) and stage 5 (red-turning fruit) were detected based on metabolic clusters using PLS-DA (Fig. 6). This is due to the developmental transition from cell expansion to cell maturation (Knee et al., 1977), in agreement with increases in pigment accumulation (Lopes et al., 2007) and shifts in gene expression (Medina et al., 1997; Manning, 1998; Nam et al., 1999). Because many proteins serve as storage units of fruit tissues, in contrast to the accumulation of most other metabolites, the levels of free amino acids decreased sharply. In Fig. 8, the fruit samples from 25 DPA mapped with the samples from 30 DPA, reflecting the compositional similarities of amino acid levels between those two phases. During the ripening phase (30–40 DPA), major pigments Pg3glc and Cy3glc, which change the colour of the flesh, increased drastically. In parallel, fructose, galactose, sucrose, citrate, and malate concentrations continue to increase, leading to a higher sugar/acid ratio, which is a major determinant of fruit taste. Additionally, there is a relationship between the softening of the strawberry fruit and the degradation of the middle lamella and cell wall, which mostly occurs during the last stages of the ripening process (Knee et al., 1977). As a result of hydration of the cell walls (Knee et al., 1977) and alterations in the cross-linking of carbohydrates (Huber, 1984; Manning, 1993), >70% of the polyuronide in the cell wall became freely soluble. The observed increases of galactose (1.5-fold) and arabinose (4.2-fold) in stage 6 are indicative of possible changes in cell wall structure linked with fruit softening. The increase in galactose during stage 6 followed by a decrease in stage 7 was also observed, which suggests that the process of fruit senescence was slower in the red-ripening stage than in the over-ripening stage. The metabolite profiling analysis indicated the perturbation of several metabolic pathways, including ester biosynthesis, the shikimate pathway, the TCA cycle, and amino acid synthesis, during fruit growth and ripening. The amino acid alanine has been implicated in the formation of ethyl esters (Perez et al., 1992), and two aromatic amino acids (phenylalanine and tyrosine) are known to be the precursors for the biosynthesis of anthocyanins and flavonoids through the shikimate pathway. Four central amino acids, glutamine, glutamate, aspartate, and asparagine (Gln, Glu, Asp, and Asn), are first derived from α-oxoglutarate and oxaloacetate in the TCA cycle (Fig. 9) and then converted into all other amino acids by various biochemical processes (Galili et al., 2008). Taken together, the present data suggest that amino acid metabolism is central to fruit development and that the synthesis of storage proteins and other amino acid-derived compounds may have major impacts on the levels of free amino acids. Volatile compounds play a dual role in fruit development and ripening, serving both as biological perfumes to entice living creatures, including humans, and as protectants against pathogens. Thus, the blends of flavour compounds produced by fruits constitute fruit organoleptic properties. The volatile components of strawberry aroma formed during ripening are the result of the combined perception of esters, alcohols, and alkanes. The formation of these compounds is closely correlated with the metabolic changes occurring during fruit maturation. The most likely precursors for the esters are lipids and amino acids. The degradation of fatty acids results in the production of volatile aldehydes, which are subsequently converted to alcohols and hexyl esters (Perez et al., 1996). The last step in ester biosynthesis is catalysed by alcohol acyltransferases, which link alcohols to acyl moieties (Shalit et al., 2001). Alkanes also serve as volatile metabolites and contribute to fruit flavour. In ripening fruits of ‘Kensington Pride’ mango, the most abundant group of volatile compounds was alkanes, accounting for ∼59% of the total identified compounds (Lalel et al., 2003). The amount of most of the alkanes in strawberry increased from the small green fruit stage, reached the highest point at the white fruit stage, and gradually decreased through the red-ripening and over-ripening stages. Among these volatiles, it was found that 2,5-dimethyl-4-hydroxy-3(2H)-furanone is a major flavour compound because of its high concentration (up to 48.53 mg kg−1 strawberry fruit fresh weight) and its low odour threshold (10 ppb), in agreement with previous data (Larsen and Poll, 1992; Schwab and Roscher, 1997). Fructose-1,6-bisphosphate represents the major precursor of 2,5-dimethyl-4-hydroxy-3(2H)-furanone in strawberry fruits (Roscher et al., 1998), while the enzyme quinone oxidoreductase was reported to be involved in the biosynthesis of 2,5-dimethyl-4-hydroxy-3(2H)-furanone (Raab et al., 2006), which can be confirmed with the relatively high correlation between fructose and 2,5-dimethyl-4-hydroxy-3(2H)-furanone (r=0.776, P=4.04%) in this study. Compared with published strawberry metabolomics data, the present report covered more developmental stages and analysed more classes of metabolites encompassing all major types of small molecules. More importantly, the correlation analysis found many interesting connections between different metabolic pathways. For example, sucrose was found to be closely linked to the biosynthesis of other monosaccharides and disaccharides, which share the same Calvin cycle origin. Surprisingly, increased glycoside conjugation, such as 1-methyl-D-galactopyranoside, is negatively correlated with sucrose content, suggesting the carbon source at the mature fruit may be limited. Several correlations in the metabolic network that are difficult to explain were also discovered. For example, fructose, galactose, and malic acid strongly correlated with 2-butyl-l-octanol accumulation. Gutierrez-Gonzalez et al. (2009) previously suggested that in seed, biosynthesis of sugar and starch might be in direct competition with the biosynthesis of fatty acid and lipid, which is the precursor of 2-butyl-l-octanol. Therefore, a reverse correlation between monosaccharides and long-chain fatty acid or alkanes is expected. This and other unexpected connections deserve further investigations. In conclusion, by metabolic profiling of both polar and non-polar compounds, it was possible to monitor the alterations in several major groups of compounds during strawberry fruit growth and maturation. The discoveries in these pathways are consistent with strawberry pigment and flavour formation, using mostly materials from primary metabolism. Each stage of fruit development has its own unique metabolic profiles, with the most drastic changes occurring at the transition toward the red-ripened stage. Amino acid biosynthesis plays an important role in generating several classes of compounds related to the quality of the fruits. This information will help strawberry breeders detect and monitor key components that are important for these output traits. We are grateful to Dr Jianhui Cheng (Institute of Horticulture, Zhejiang Agricultural Academy, Hangzhou, China) for kindly supplying the six anthocyanin standards. We greatly appreciate the help of Dr Kazuki Saito and Dr Atsushi Fukushima (RIKEN Yokohama Institute, Plant Science Center, Japan) in metabolic network and heat map analysis. The project was funded by the National Natural Science Foundation of China (grant no. 31071756) and the Science and Technological Fund of Anhui Province for Outstanding Youth (grant no. 08040106801). References Aaby K,  Ekeberg D,  Skrede G.  Characterization of phenolic compounds in strawberry (Fragaria × ananassa) fruits by different HPLC detectors and contribution of individual compounds to total antioxidant capacity,  Journal of Agricultural and Food Chemistry ,  2007, vol.  55 (pg.  4395- 4406) Google Scholar CrossRef Search ADS PubMed  Aharoni A,  Keizer LCP,  Bouwmeester HJ, et al.  Identification of the SAAT gene involved in strawberry flavor biogenesis by use of DNA microarrays,  The Plant Cell ,  2000, vol.  12 (pg.  647- 661) Google Scholar CrossRef Search ADS PubMed  Aharoni A,  de Vos CH,  Maliepaard CA,  Kruppa G,  Bino RJ,  Goodenough D.  Nontargeted metabolome analysis by use of fourier transform ion cyclotron mass spectrometry,  OMICS ,  2002, vol.  6 (pg.  217- 234) Google Scholar CrossRef Search ADS PubMed  Anttonen MJ,  Hoppula KI,  Nestby R,  Verheul MJ,  Karjalainen RO.  Influence of fertilization, mulch color, early forcing, fruit order, planting date, shading, growing environment, and genotype on the contents of selected phenolics in strawberry (Fragaria × ananassa Duch.) fruits,  Journal of Agricultural and Food Chemistry ,  2006, vol.  54 (pg.  2614- 2620) Google Scholar CrossRef Search ADS PubMed  Aprea E,  Biasioli F,  Carlin S,  Endrizzi I,  Gasperi F.  Investigation of volatile compounds in two raspberry cultivars by two headspace techniques: solid-phase microextraction/gas chromatography-mass spectrometry (SPME/GC-MS) and proton-transfer reaction-mass spectrometry (PTR-MS),  Journal of Agricultural and Food Chemistry ,  2009, vol.  57 (pg.  4011- 4018) Google Scholar CrossRef Search ADS PubMed  Atkinson CJ,  Dodds PA,  Ford YY,  Le Mière J,  Taylor JM,  Blake PS,  Paul N.  Effects of cultivar, fruit number and reflected photosynthetically active radiation on Fragaria× ananassa productivity and fruit ellagic acid and ascorbic acid concentrations,  Annals of Botany ,  2006, vol.  97 (pg.  429- 444) Google Scholar CrossRef Search ADS PubMed  Blount JW,  Masoud S,  Sumner LW,  Huhman D,  Dixon RA.  Over-expression of cinnamate 4-hydroxylase leads to increased accumulation of acetosyringone in elicited tobacco cell-suspension cultures,  Planta ,  2002, vol.  214 (pg.  902- 910) Google Scholar CrossRef Search ADS PubMed  Broeckling CD,  Huhman DV,  Farag MA,  Smith JT,  May GD,  Mendes P,  Dixon RA,  Sumner LW.  Metabolic profiling of Medicago truncatula cell cultures reveals the effects of biotic and abiotic elicitors on metabolism,  Journal of Experimental Botany ,  2005, vol.  56 (pg.  323- 326) Google Scholar CrossRef Search ADS PubMed  Chen F,  Kota P,  Blount JW,  Sumner LW,  Dixon RA.  Profiling phenolic metabolites in transgenic alfalfa modified in lignin biosynthesis,  Phytochemistry ,  2003, vol.  64 (pg.  1013- 1021) Google Scholar CrossRef Search ADS PubMed  Fait A,  Hanhineva K,  Beleggia R,  Dai N,  Rogachev I,  Nikiforova VJ,  Fernie AR,  Aharoni A.  Reconfiguration of the achene and receptacle metabolic networks during strawberry fruit development,  Plant Physiology ,  2008, vol.  148 (pg.  730- 750) Google Scholar CrossRef Search ADS PubMed  Fiehn O.  Metabolic networks of Cucurbita maxima phloem,  Phytochemistry ,  2003, vol.  62 (pg.  875- 886) Google Scholar CrossRef Search ADS PubMed  Galili S,  Amir R,  Galili G.  Genetic engineering of amino acid metabolism in plants,  Advances in Plant Biochemistry and Molecular Biology ,  2008, vol.  1 (pg.  49- 80) Gu L,  Kelm MA,  Hammerstone JF,  Beecher G,  Holden J,  Haytowitz D,  Prior RL.  Screening of foods containing proanthocyanidins and their structural characterization using LC–MS/MS and thiolytic degradation,  Journal of Agricultural and Food Chemistry ,  2003, vol.  51 (pg.  7513- 7521) Google Scholar CrossRef Search ADS PubMed  Guo W,  Sakata K,  Watanabe N,  Nakajima R,  Yagi A,  Ina K,  Luo S.  Geranyl 6-O-β-D-xylopyranosyl-β-D-glucopyranoside isolated as an aroma precursor from tea leaves for oolong tea,  Phytochemistry ,  1993, vol.  33 (pg.  1373- 1375) Google Scholar CrossRef Search ADS PubMed  Gutierrez-Gonzalez JJ,  Wu X,  Zhang J,  Lee JD,  Ellersieck M,  Shannon GJ,  Yu O,  Nguyen HT,  Sleper DA.  Genetic control of soybean seed isoflavone content: importance of statistical model and epistasis in complex traits,  Theoretical and Applied Genetics ,  2009, vol.  119 (pg.  1069- 1083) Google Scholar CrossRef Search ADS PubMed  Hamzehzarghani H,  Kushalappa AC,  Dion Y,  Rioux S,  Comeau A,  Yaylayan V,  Marshall WD,  Mather DE.  Metabolic profiling and factor analysis to discriminate quantitative resistance in wheat cultivars against fusarium head blight,  Physiological and Molecular Plant Pathology ,  2005, vol.  66 (pg.  119- 133) Google Scholar CrossRef Search ADS   Hancock JF. ,  Strawberries ,  1999 Wallingford, UK CABI Publishing Hanhineva K,  Rogachev I,  Kokko H,  Mintz-Oron S,  Venger I,  Kärenlampi S,  Aharoni A.  Non-targeted analysis of spatial metabolite composition in strawberry (Fragaria × ananassa) flowers,  Phytochemistry ,  2008, vol.  69 (pg.  2463- 2481) Google Scholar CrossRef Search ADS PubMed  Heinonen IM,  Meyer AS,  Frankel EN.  Antioxidant activity of berry phenolics on human low-density lipoprotein and liposome oxidation,  Journal of Agricultural and Food Chemistry ,  1998, vol.  46 (pg.  4107- 4112) Google Scholar CrossRef Search ADS   Honkanen E,  Hirvi T.  Morton ID,  MacLeod AJ.  The flavour of berries,  Food flavours ,  1990 Amsterdam Elsevier Scientific Publications(pg.  125- 193) Huber DJ.  Strawberry fruit softening: the potential roles of polyuronides and hemicelluloses,  Journal of Food Science ,  1984, vol.  49 (pg.  1310- 1315) Google Scholar CrossRef Search ADS   Hukkanen AT,  Kokko HI,  Buchala AJ,  McDougall GJ,  Stewart D,  Kärenlampi SO,  Karjalainen RO.  Benzothiadiazole induces the accumulation of phenolics and improves resistance to powdery mildew in strawberries,  Journal of Agricultural and Food Chemistry ,  2007, vol.  55 (pg.  1862- 1870) Google Scholar CrossRef Search ADS PubMed  Kim JK,  Bamba T,  Harada K,  Fukusaki E,  Kobayashi A.  Time-course metabolic processing in Arabidopsis thaliana cell cultures after salt stress treatment,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  415- 424) Google Scholar CrossRef Search ADS PubMed  Knee M,  Sargent JA,  Osborne DJ.  Cell-wall metabolism in developing strawberry fruits,  Journal of Experimental Botany ,  1977, vol.  28 (pg.  377- 396) Google Scholar CrossRef Search ADS   Lalel HJD,  Singh Z,  Tan SC.  Aroma volatiles production during fruit ripening of ‘Kensington Pride’ mango,  Postharvest Biology and Technology ,  2003, vol.  27 (pg.  323- 336) Google Scholar CrossRef Search ADS   Larsen M,  Poll L.  Odor thresholds of some important aroma compounds in strawberries,  Zeitschrift Lebensmittel-Untersuchung und -Forschung ,  1992, vol.  195 (pg.  120- 123) Google Scholar CrossRef Search ADS   Latrasse A.  Maarse H.  Fruits III,  Volatile compounds in foods and beverages ,  1991 New York Dekker(pg.  329- 387) Lopes da Silva F,  Escribano-Bailón MT,  Pérez Alonso JJ,  Rivas-Gonzalo JC,  Santos-Buelga C.  Anthocyanin pigments in strawberry,  LWT–Food Science and Technology ,  2007, vol.  40 (pg.  374- 382) Google Scholar CrossRef Search ADS   Maarse H. ,  Volatile compounds in foods and beverages ,  1991 New York Marcel Dekker(pg.  483- 546) Määttä-Riihinen KR,  Kamal-Eldin A,  Torronen AR.  Identification and quantification of phenolic compounds in berries of Fragaria and Rubus species (family Rosaceae),  Journal of Agricultural and Food Chemistry ,  2004, vol.  52 (pg.  6178- 6187) Google Scholar CrossRef Search ADS PubMed  Manning K.  Isolation of a set of ripening-related genes from strawberry: their identification and possible relationship to fruit quality traits,  Planta ,  1998, vol.  205 (pg.  622- 631) Google Scholar CrossRef Search ADS PubMed  Manning K.  Taylor JE,  Tucker GA.  Soft fruits,  Biochemistry of fruit ripening ,  1993 Cambridge, UK Chapman and Hall(pg.  347- 373) Mateo JJ,  Gentilini N,  Huerta T,  Jimenez M,  Distefano R.  Fractionation of glycoside precursors of aroma in grapes and wine,  Journal of Chromatography A ,  1997, vol.  778 (pg.  219- 224) Google Scholar CrossRef Search ADS PubMed  Medina Escobar N,  Cardenas J,  Valpuesta V,  Munoz Blanco J,  Caballero JL.  Cloning and characterization of cDNAs from genes differentially expressed during the strawberry fruit ripening process by a MAST-PCR-SBDS method,  Analytical Biochemistry ,  1997, vol.  248 (pg.  288- 296) Google Scholar CrossRef Search ADS PubMed  Moco SIA,  Bino RJ,  Vos de CH,  Vervoort JJM.  Metabolomics technologies and metabolite identification,  Trends in Analytical Chemistry ,  2007, vol.  26 (pg.  855- 866) Google Scholar CrossRef Search ADS   Nam YW,  Tichit L,  Leperlier M,  Cuerq B,  Marty I,  Lelievre JM.  Isolation and characterization of mRNAs differentially expressed during ripening of wild strawberry (Fragaria vesca L.) fruits,  Plant Molecular Biology ,  1999, vol.  39 (pg.  629- 636) Google Scholar CrossRef Search ADS PubMed  Perez AG,  Rios JJ,  Sanz C,  Olias JM.  Aroma components and free amino acids in strawberry variety Chandler during ripening,  Journal of Agricultural and Food Chemistry ,  1992, vol.  40 (pg.  2232- 2235) Google Scholar CrossRef Search ADS   Perez AG,  Sanz C,  Olias R,  Rios JJ,  Olías JM.  Evolution of strawberry alcohol acyltransferase activity during fruit development and storage,  Journal of Agricultural and Food Chemistry ,  1996, vol.  44 (pg.  3286- 3290) Google Scholar CrossRef Search ADS   Perkins-Veazie P.  Growth and ripening of strawberry fruit,  Horticultural Reviews ,  1995, vol.  17 (pg.  267- 297) Raab T,  Lopez-raez JA,  Klein D,  Caballero JL,  Moyano E,  Schwab W,  Munoz-Blanco J.  FaQR, required for the biosynthesis of the strawberry flavor compound 4-hydroxy-2,5-dimethyl-3(2H)-furanone, encodes an enone oxidoreductase,  The Plant Cell ,  2006, vol.  18 (pg.  1023- 1037) Google Scholar CrossRef Search ADS PubMed  Raamsdonk LH,  Teusink B,  Broadhurst D, et al.  A functional genomics strategy that uses metabolome data to reveal the phenotype of silent mutations,  Nature Biotechnology ,  2001, vol.  19 (pg.  45- 50) Google Scholar CrossRef Search ADS PubMed  Roessner-Tunali U,  Hegemann B,  Lytovchenko A,  Carrari F,  Bruedigam C.  Metabolic profiling of transgenic tomato plants overexpressing hexokinase reveals that the influence of hexose phosphorylation diminishes during fruit development,  Plant Physiology ,  2003, vol.  133 (pg.  84- 99) Google Scholar CrossRef Search ADS PubMed  Roessner U,  Luedemann A,  Brust D,  Fiehn O,  Linke T,  Willmitzer L,  Fernie AR.  Metabolic profiling allows comprehensive phenotyping of genetically or environmentally modified plant systems,  The Plant Cell ,  2001, vol.  13 (pg.  11- 29) Google Scholar CrossRef Search ADS PubMed  Roessner U,  Wagner C,  Kopka J,  Trethewey RN,  Willmitzer L.  Simultaneous analysis of metabolites in potato tuber by gas chromatography–mass spectrometry,  The Plant Journal ,  2000, vol.  23 (pg.  131- 142) Google Scholar CrossRef Search ADS PubMed  Roscher R,  Bringmann G,  Schreier P,  Schwab W.  Radiotracer studies on the formation of 2, 5-dimethyl-4-hydroxy-3[2H]-furanone in detached ripening strawberry fruits,  Journal of Agricultural and Food Chemistry ,  1998, vol.  46 (pg.  1488- 1493) Google Scholar CrossRef Search ADS   Santos-Buelga C,  Scalbert A.  Proanthocyanidins and tannin-like compounds: nature, occurrence, dietary intake, and effects on nutrition and health,  Journal of the Science of Food and Agriculture ,  2000, vol.  80 (pg.  1094- 1117) Google Scholar CrossRef Search ADS   Sato S,  Soga T,  Tomita M.  Simultaneous determination of the main metabolites in rice leaves using capillary electrophoresis mass spectrometry and capillary electrophoresis diode array detection,  The Plant Journal ,  2004, vol.  40 (pg.  151- 163) Google Scholar CrossRef Search ADS PubMed  Schauer N,  Fernie AR.  Plant metabolomics: towards biological function and mechanism,  Trends in Plant Science ,  2006, vol.  11 (pg.  508- 516) Google Scholar CrossRef Search ADS PubMed  Schauer N,  Zamir D,  Fernie AR.  Metabolic profiling of leaves and fruit of wild species tomato: a survey of the Solanum lycopersicum complex,  Journal of Experimental Botany ,  2005, vol.  56 (pg.  297- 307) Google Scholar CrossRef Search ADS PubMed  Schwab W,  Roscher R.  4-Hydroxy-3(2H)-furanones: natural and maillard products,  Recent Research Developments in Phytochemistry ,  1997, vol.  1 (pg.  643- 673) Shalit M,  Katzir N,  Tadmor Y,  Larkov O,  Burger Y,  Schalechet F,  Lastochkin E,  Ravid U,  Amar O,  Edelstein M.  Acetyl-CoA:alcohol acetyl transferase activity and aroma formation in ripening melon fruits,  Journal of Agricultural and Food Chemistry ,  2001, vol.  49 (pg.  794- 799) Google Scholar CrossRef Search ADS PubMed  Steuer R,  Kurths J,  Fiehn O,  Weckwerth W.  Interpreting correlations in metabolomic networks,  Biochemical Society Transactions ,  2003, vol.  31 (pg.  1476- 1478) Google Scholar CrossRef Search ADS PubMed  Steuer R,  Kurths J,  Fiehn O,  Weckwerth W.  Observing and interpreting correlations in metabolomic networks,  Bioinformantics ,  2003, vol.  19 (pg.  1019- 1026) Google Scholar CrossRef Search ADS   Tagashira N,  Plader W,  Filipecki M, et al.  The metabolic profiles of transgenic cucumber lines vary with different chromosomal locations of the transgene,  Cellular and Molecular Biology Letters ,  2005, vol.  10 (pg.  697- 710) Google Scholar PubMed  Terry LA,  Chope GA,  Bordonaba JG.  Effect of water deficit irrigation and inoculation with Botrytis cinerea on strawberry (Fragaria × ananassa) fruit quality,  Journal of Agricultural and Food Chemistry ,  2007, vol.  55 (pg.  10812- 10819) Google Scholar CrossRef Search ADS PubMed  Wang SY,  Jiao H.  Scavenging capacity of berry crops on superoxide radicals, hydrogen peroxide, hydroxyl radicals and singlet oxygen,  Journal of Agricultural and Food Chemistry ,  2000, vol.  48 (pg.  5677- 5684) Google Scholar CrossRef Search ADS PubMed  Wang SY,  Lin HS.  Antioxidant activity in fruits and leaves of blackberry, raspberry and strawberry varies with cultivar and developmental stage,  Journal of Agricultural and Food Chemistry ,  2000, vol.  48 (pg.  140- 146) Google Scholar CrossRef Search ADS PubMed  Weckwerth W,  Loureiro M,  Wenzel K,  Fiehn O.  Differential metabolic networks unravel the effects of silent plant phenotypes,  Proceedings of the National Academy of Sciences, USA ,  2004, vol.  101 (pg.  7809- 7814) Google Scholar CrossRef Search ADS   Zabetakis I,  Holden MA.  Strawberry flavor: analysis and biosynthesis,  Journal of the Science of Food and Agriculture ,  1997, vol.  74 (pg.  421- 434) Google Scholar CrossRef Search ADS   © The Author [2010]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: [email protected]

Espino, Susana;Schenk, H. Jochen

doi: 10.1093/jxb/erq338pmid: 21147811

Abstract The maximum specific hydraulic conductivity (kmax) of a plant sample is a measure of the ability of a plants’ vascular system to transport water and dissolved nutrients under optimum conditions. Precise measurements of kmax are needed in comparative studies of hydraulic conductivity, as well as for measuring the formation and repair of xylem embolisms. Unstable measurements of kmax are a common problem when measuring woody plant samples and it is commonly observed that kmax declines from initially high values, especially when positive water pressure is used to flush out embolisms. This study was designed to test five hypotheses that could potentially explain declines in kmax under positive pressure: (i) non-steady-state flow; (ii) swelling of pectin hydrogels in inter-vessel pit membranes; (iii) nucleation and coalescence of bubbles at constrictions in the xylem; (iv) physiological wounding responses; and (v) passive wounding responses, such as clogging of the xylem by debris. Prehydrated woody stems from Laurus nobilis (Lauraceae) and Encelia farinosa (Asteraceae) collected from plants grown in the Fullerton Arboretum in Southern California, were used to test these hypotheses using a xylem embolism meter (XYL'EM). Treatments included simultaneous measurements of stem inflow and outflow, enzyme inhibitors, stem-debarking, low water temperatures, different water degassing techniques, and varied concentrations of calcium, potassium, magnesium, and copper salts in aqueous measurement solutions. Stable measurements of kmax were observed at concentrations of calcium, potassium, and magnesium salts high enough to suppress bubble coalescence, as well as with deionized water that was degassed using a membrane contactor under strong vacuum. Bubble formation and coalescence under positive pressure in the xylem therefore appear to be the main cause for declining kmax values. Our findings suggest that degassing of water is essential for achieving stable and precise measurements of kmax through woody plant samples. For complete rehydration of woody samples, incubation in water under vacuum for 24 h is suggested as a reliable technique that avoids bubble problems associated with flushing under high positive pressure. Bubble coalescence, Encelia farinosa, Laurus nobilis, hydraulic conductivity, pectin hydrogel hypothesis, plant hydraulics, woody stems, xylem vulnerability curves Introduction The maximum hydraulic conductivity (kmax) is a measure of the plant vascular system's ability to transport water and solutes under optimum conditions. It is defined as the hydraulic conductivity in the absence of any reversible air embolisms in xylem conduits. Precise measurements of kmax are needed for characterizing the resistance of plants to embolism formation (Sperry and Sullivan, 1992; Alder et al., 1997), in studies of embolism repair (Salleo et al., 2004), and in comparative studies of hydraulic conductivity in different plant species (Maherali et al., 2004). Xylem traits that affect kmax include the abundance, diameters, lengths, and degree of connectedness of xylem conduits (Loepfe et al., 2007). Plants can lower kmax by filling xylem conduits with tyloses (Cochard and Tyree, 1990), gels (Bonsen and Kučera, 1990; Crews et al., 2003), or both (Sun et al., 2008), usually in response to wounding or to pathogen attack or during heartwood formation. The ionic composition of xylem sap also affects kmax (van Ieperen et al., 2000; Zwieniecki et al., 2001; Lopez-Portillo et al., 2005; Gascó et al., 2007). This process is hypothesized to involve the swelling of pectin hydrogels embedded into the pit membranes that connect xylem conduits (Zwieniecki et al., 2001), which potentially could decrease kmax. When water or a solution with a constant ionic composition is fed through stems, roots, or petioles whose conduits are completely filled with water, one would not expect kmax to change substantially in the short term (minutes to hours), because structural features do not change over such periods and because pit membranes are thought to retain a constant hydraulic conductivity for a given ionic composition in water (Zwieniecki et al., 2001). Over longer periods (many hours to days), microbial growth and decomposition could clog or alter the structure of xylem conduits (Sperry et al., 1988) and tyloses may plug conduits as well (Sun et al., 2007). Despite the fact that short-term changes of kmax are not expected in theory, they are frequently observed. When deionized water is used as the fluid medium, kmax typically declines from initially high values, even in samples without embolized xylem conduits (Kelso et al., 1963; Sperry et al., 1988; Sperry and Tyree, 1990; van Ieperen and van Gelder, 2006; Canny et al., 2007; Cochard et al., 2010). Such short-term declines lead to unreliable measurements of kmax, which can be a serious problem for studies in plant hydraulics. This study was designed to test the hypothesis that a short-term decline in kmax was caused by one or more of the following mechanisms. (i) Non-steady-state flow. Water forced into a cut plant sample under positive pressure may enter air-filled fibre cells, extracellular spaces, or parenchyma cells (van Ieperen et al., 2000). Lateral water flow from conduits may cause initially high rates of inflow, declining over time as cells and extracellular spaces are filled (Tyree and Yang, 1992). Hydraulic conductance measurements based on inflow, for example, using the XYL'EM apparatus (Cochard et al., 2000), could be affected by artefacts due to lateral flow. (ii) Swelling of pectin hydrogels in pit membranes. Use of deionized water for flushing xylem could potentially deplete pit-membrane-residing pectins of cations, especially Ca2+, thereby cause hydrogel swelling, and thus decrease hydraulic conductance (Zwieniecki et al., 2001). Ca2+ ions are a structural component of some pectins, provide rigidity, and reduce pectin swelling potential (Ryden et al., 2000). (iii) Formation and coalescence of bubbles. Gassing-out can occur when air-saturated water flows through constrictions (Kelso et al., 1963; Scardina and Edwards, 2004; Blatteau et al., 2006; Canny et al., 2007), such as xylem pit membranes or perforation plates. In a process termed ‘air binding’, small bubbles form through heterogeneous nucleation and then coalesce into larger ones that can block pit membranes or whole vessels. (iv) Active wounding responses. Reduced water uptake by stems can be due to a physiological wounding response that may include mechanisms of damage control and repair or defences against pathogens that may take advantage of the wounded plant tissue (Cheong et al., 2002; Ramonell and Somerville, 2002). Wounding responses that have been associated with reduced water uptake in cut stems appear to involve cell wall enzyme activity, possibly related to suberin, tylose, and gel formation (van Doorn and Cruz, 2000; van Doorn and Vaslier, 2002; Loubaud and van Doorn, 2004) or in defence against pathogenic cell-wall degrading enzymes (Cheong et al., 2002; Li et al., 2003). (v) Passive wounding effects. A decline in kmax could result if cutting a sample caused cell-wall debris to dislodge or cause gums and resins to distribute across the cut surface of the xylem (Schulte et al., 1987; Sperry et al., 1988). Flushing the samples under pressure could force such materials into vessels and clog them. This study was designed to test for the five mechanisms described above for stem samples from two woody plant species, Laurus nobilis L. (Lauraceae), sweet bay, and Encelia farinosa Torrey & A. Gray (Asteraceae), brittlebush. Laurus, a small evergreen tree of Mediterranean origin, was chosen because it has been the subject of numerous studies in plant hydraulics (Tyree et al., 1999; Zwieniecki et al., 2001; Hacke and Sperry, 2003; Salleo et al., 2004; Gascó et al., 2006), while Encelia, a North American, drought-deciduous, desert shrub was chosen because we had previously experienced problems with unstable measurements of kmax in this species. Materials and methods Collection of plant samples Branch samples were collected between July 2008 and September 2010 from a single Laurus nobilis tree and a group of Encelia farinosa shrubs growing in the Fullerton Arboretum in Fullerton, California, USA. At least 50-cm-long branches with leaves were cut off from the plants under water and, with their cut ends submerged under water, were transported to the laboratory, where they were placed in a vase filled with tap water or with an aqueous-solution, depending on treatment (see below), for 24 h to allow the stems to rehydrate. In one experiment, stems were cut under water from very well-watered plants at predawn and used as a control treatment to test for the effectiveness of the vase rehydration treatment. The stems were then submerged in a water-filled tray, where a 15-cm-long central portion was cut out with sharp anvil clippers. Stem segments were 2–6 mm in diameter (mean 3.5 mm for both species), 1–2-years-old, and did not contain any heartwood. Maximum vessel lengths in both species exceed 15 cm (Laurus 40 cm; Gascó et al., 2006; Encelia 50 cm; S Espino, unpublished data), thus, due to a few open vessels, measured kmax values may slightly overestimate true values of kmax in these species. For each treatment, 10 replicates were used for each species, but in some cases replication was reduced to 6–9 replicates, usually because leaky or damaged stems had to be discarded. Measurements of hydraulic conductivity A XYL'EM embolism meter (Bronkhorst, Montigny les Cormeilles, France; Cochard et al., 2000) was used for all measurements of hydraulic conductivity. Keeping stem segments submerged under water at all times, they were connected at their basal inflow end to the tubing of the apparatus. Except for treatments measured in ice water (5–11 °C), all measurements were conducted at room temperature (18.1–22.3 °C). Aqueous solutions used for conductance measurements and for flushing stems under high pressure varied between treatments. Treatments and their predicted effects are summarized in Table 1. Except where noted in Table 1, all solutions were degassed under laboratory vacuum for at least 3 h in a side-arm flask on a magnetic stirrer at approximately 30 kPa absolute pressure (=71 kPa vacuum) and passed through a 0.2 μm filter (model Polycap AS 75, Whatman Inc., Piscataway, NJ). All flow rate measurements were preceded by a measurement without a pressure differential (Δp=0) between inflow and outflow to determine baseline flow-rates (F0) into the samples. Water pressure for flushing of stems was generated with the high pressure vessel in the XYL'EM apparatus, which prevents contact between air and water during pressurization. All measurements were recorded after flow rates stabilized, typically within 1–5 min. The first measurement was conducted under a Δp of 3 kPa (FΔp). Hydraulic conductance, k, was determined as k=(FΔp–F0)/Δp and specific hydraulic conductivity, ks, for a given length, L, and wood-cross-sectional area, A, as:  (1) Table 1. Predicted outcomes of Experimental treatments on the decline of kmax measurements in woody stems, assuming that the mechanism listed in the corresponding column header is operative (see Introduction for further explanations) Experimental treatment  Non-steady-state flow  Pectin swelling  Bubbles  Active wounding response  Passive wounding effects  Not degassed  o  o  o  o  o  Flask-degassed at 30 kPa  o  o  +  o  o  Degasser at 30 kPa  o  o  +  o  o  Degasser at 3 kPa  o  o  ++  o  o  Hydrated at 3 kPa, degasser at 3 kPa  o  o  ++  o  o  Ice water  o  +  +  +  o  Ice waterb  o  o  +  +  o  Bark removed before cutting  o  o  o  +  +  KCl, 10 mM  o  ++  o  o  o  KCl, 100 mM  o  ++  +  o  o  KCl, 300 mM  o  ++  ++  o  o  CaCl2, 10 mM  o  ++  o  o  o  CaCl2, 40 mM  o  ++  +  o  o  CaCl2, 100 mM  o  ++  ++  o  o  MgSO4, 0.5 mM  o  +  o  o  o  MgSO4, 20 mM  o  ++  +  o  o  MgSO4, 50 mM  o  ++  ++  o  o  (NH4)2SO4, 100 mM  o  o  ++  o  o  Cu(II)SO4, 0.1 mMb  o  o  o  +  o  Cu(II)SO4, 0.25 mM  o  o  o  +  o  4-HR, 10 mMb  o  o  o  +  o  Hydroquinone, 10 mMb  o  o  o  +  o  Degasser, in- and outflow measured  +a  o  +  o  o  Experimental treatment  Non-steady-state flow  Pectin swelling  Bubbles  Active wounding response  Passive wounding effects  Not degassed  o  o  o  o  o  Flask-degassed at 30 kPa  o  o  +  o  o  Degasser at 30 kPa  o  o  +  o  o  Degasser at 3 kPa  o  o  ++  o  o  Hydrated at 3 kPa, degasser at 3 kPa  o  o  ++  o  o  Ice water  o  +  +  +  o  Ice waterb  o  o  +  +  o  Bark removed before cutting  o  o  o  +  +  KCl, 10 mM  o  ++  o  o  o  KCl, 100 mM  o  ++  +  o  o  KCl, 300 mM  o  ++  ++  o  o  CaCl2, 10 mM  o  ++  o  o  o  CaCl2, 40 mM  o  ++  +  o  o  CaCl2, 100 mM  o  ++  ++  o  o  MgSO4, 0.5 mM  o  +  o  o  o  MgSO4, 20 mM  o  ++  +  o  o  MgSO4, 50 mM  o  ++  ++  o  o  (NH4)2SO4, 100 mM  o  o  ++  o  o  Cu(II)SO4, 0.1 mMb  o  o  o  +  o  Cu(II)SO4, 0.25 mM  o  o  o  +  o  4-HR, 10 mMb  o  o  o  +  o  Hydroquinone, 10 mMb  o  o  o  +  o  Degasser, in- and outflow measured  +a  o  +  o  o  Symbols: o=no alleviation predicted; + =some alleviation predicted; ++ =strong alleviation predicted. a Experimental treatment accounts for, but does not alleviate, declines in kmax. b Treatment applied during vase rehydration and during measurement. View Large The reported conductivities include a temperature correction to 20 °C to allow for changes of water viscosity with temperature, except for ice water, because it was impossible to control the ice-water temperature carefully during measurements and the exact water temperature inside the stems was therefore unknown. Stems were then flushed under a pressure differential of 150 kPa for 3 min, after which FΔp was determined again. In one treatment, 3 kPa ‘flushes’ were used to test for effects of high-pressure versus low-pressure flushing. This was followed by another high-pressure flush for 3 min, followed by measurement of FΔp. In a few cases indicated in the results, this sequence was repeated five times to determine kmax stability. High-pressure flushing was used in this study because it is the most commonly used method for measuring kmax in stems (Table 2). Table 2. Methods used in representative studies of maximum hydraulic conductivity through stem and root samples Reference  Sample length (cm)  Measured inflow or outflow?a  Filtered and degassed?b  Measuring pressure (kPa)b  Flushing pressure and flushing timeb  Ionic composition  Sperry et al. (1987)  Various  Out  0.22 μm filter, degassed5  ≤105  170 kPa, repeated flushes to kmax  10 mM NaCl, 0.05% formaldehyde  Sperry et al. (1988)  10–15  Out*  0.22 μm filter, degassed5  ≤10  175 kPa, 20–200 min  10 mM NaCl, (1 mM NaCl+0.5 mM CaCl2+0.2 mM KCl), ‘maple sap’, 10 mM citric acid (pH 4), 10 mM citric acid (pH <3), formaldehyde (0.05%, 0.5%), gluteraldehyde (0.05%), 10 mM oxalic acid (pH 1.3–2.4)  Tyree and Yang (1992)  25–75  In and out  0.2 and 0.1 μm filter  2–14  150 kPa, both ends, up to 200 h  10 mM oxalic acid  Sperry and Sullivan (1992)  4–15  Out*  0.22 μm filter, degassed5  40 or 70  175 kPa, repeated flushes to kmax  10 mM oxalic acid  Sperry and Saliendra (1994)  15–25  Out*  0.2 μm filter, not degassed5  10  175 kPa, no time info  HCl, pH 2  Jarbeau et al. (1995)  10  Out*  0.1 μm filter, degassed  3  175 kPa, 1 h  10 mM citric acid  Pockman and Sperry (2000)  10  Out*  0.22 μm filter, not degassed5  ≤10  100 kPa, repeated flushes to kmax  HCl, pH 2  Hacke et al. (2000)  14  Out*  0.2 μm filter, not degassed4  –7 to –10, –3 for roots  50–70 kPa, 30 min, kmax at –0.5 MPa  Deionized water  Cochard et al. (2000)  3–4  In**  0.2 μm filter, degassed  3  100 kPa, no time info  Water  van Ieperen et al. (2000)  20  In†  ?  –40  N/A  Deionized water, 0.01–200 mM KCl, 6.7 mM K2SO4, 10 mM NaCl, 67 mM CaCl2, 20 mM mannitol, 20 mM melizitose  Zwieniecki et al. (2001)  3–6  Out  0.2 μm filter, not degassed6  40  200 kPa, 10 min  Deionized water, 0.1–100 mM KCl, 10 mM sucrose, 10 mM ethanol, 10 mM NaCl, 10 mM KNO2, 10 mM CaCl2, 10 to 95% ethanol  Bucci et al. (2003)  10–15  Out8  0.22 μm filter, degassed  4.9  200 kPa, repeated 15 min flushes to kmax  Distilled water  Hukin et al. (2005)  2  In**  0.2 μm filter, degassed2  1.5  150 kPa, no time info  10 mM KCl  Jacobsen et al. (2005)  10–27  Out*  0.1 μm filter, degassed1  1.5–3.51  100 kPa, 1 h, kmax at –0.5 MPa  HCl, pH 2  Gascò et al. (2006)  1–36  In**  0.1 μmm filter  9  190 kPa, 10 min  Deionized water, 5–150 mM KCl, 200 mM sucrose, 100 mM NaCl  Maherali et al. (2006)  Stems 14, roots: 27–59  Out*  0.2 μm filter, not degassed3  1.5–2  100 kPa, 15–20 min  Distilled water  van Meeteren et al. (2006)  22  In†  Degassed and not degassed  –40  –40 kPa 1.5 h, then 3.4–3 kPa 30 min  0.7 mM CaCl2, 1.5 mM NaHCO3, 50 mM CuSO4  van Ieperen and van Gelder (2006)  7–13  In†  ?  –20  N/A  Ultrapure deionized water, 0.1 mM CaCl2, 1 mM CaCl2, 0.1 mM CaCl2+10 mM KCl, 1 mM CaCl2+10 mM KCl, 0.1 mM CaCl2+100 mM KCl, 1 mM CaCl2+100 mM KCl, 0.1–100 mM KCl, 0.1–10 mM CaCl2  Nardini et al. (2007)  6–16  In**  0.1 μm filter  9  190 kPa, 10 min  Deionized water, 25 mM KCl, 0.5 mM CaCl2, 1 mM CaCl2, 25 mM KCl+0.5 mM CaCl2, 25 mM KCl+1 mM CaCl2, mineral water  Lovisolo et al. (2008)  Petioles 1, shoots 40, roots ?  In††  0.1 μm filter, degassed  Petioles 40; shoots 20; roots 10  Petioles 600 kPa7; shoots 300 kPa7; roots 300 kPa7  15 mM KCl  Reference  Sample length (cm)  Measured inflow or outflow?a  Filtered and degassed?b  Measuring pressure (kPa)b  Flushing pressure and flushing timeb  Ionic composition  Sperry et al. (1987)  Various  Out  0.22 μm filter, degassed5  ≤105  170 kPa, repeated flushes to kmax  10 mM NaCl, 0.05% formaldehyde  Sperry et al. (1988)  10–15  Out*  0.22 μm filter, degassed5  ≤10  175 kPa, 20–200 min  10 mM NaCl, (1 mM NaCl+0.5 mM CaCl2+0.2 mM KCl), ‘maple sap’, 10 mM citric acid (pH 4), 10 mM citric acid (pH <3), formaldehyde (0.05%, 0.5%), gluteraldehyde (0.05%), 10 mM oxalic acid (pH 1.3–2.4)  Tyree and Yang (1992)  25–75  In and out  0.2 and 0.1 μm filter  2–14  150 kPa, both ends, up to 200 h  10 mM oxalic acid  Sperry and Sullivan (1992)  4–15  Out*  0.22 μm filter, degassed5  40 or 70  175 kPa, repeated flushes to kmax  10 mM oxalic acid  Sperry and Saliendra (1994)  15–25  Out*  0.2 μm filter, not degassed5  10  175 kPa, no time info  HCl, pH 2  Jarbeau et al. (1995)  10  Out*  0.1 μm filter, degassed  3  175 kPa, 1 h  10 mM citric acid  Pockman and Sperry (2000)  10  Out*  0.22 μm filter, not degassed5  ≤10  100 kPa, repeated flushes to kmax  HCl, pH 2  Hacke et al. (2000)  14  Out*  0.2 μm filter, not degassed4  –7 to –10, –3 for roots  50–70 kPa, 30 min, kmax at –0.5 MPa  Deionized water  Cochard et al. (2000)  3–4  In**  0.2 μm filter, degassed  3  100 kPa, no time info  Water  van Ieperen et al. (2000)  20  In†  ?  –40  N/A  Deionized water, 0.01–200 mM KCl, 6.7 mM K2SO4, 10 mM NaCl, 67 mM CaCl2, 20 mM mannitol, 20 mM melizitose  Zwieniecki et al. (2001)  3–6  Out  0.2 μm filter, not degassed6  40  200 kPa, 10 min  Deionized water, 0.1–100 mM KCl, 10 mM sucrose, 10 mM ethanol, 10 mM NaCl, 10 mM KNO2, 10 mM CaCl2, 10 to 95% ethanol  Bucci et al. (2003)  10–15  Out8  0.22 μm filter, degassed  4.9  200 kPa, repeated 15 min flushes to kmax  Distilled water  Hukin et al. (2005)  2  In**  0.2 μm filter, degassed2  1.5  150 kPa, no time info  10 mM KCl  Jacobsen et al. (2005)  10–27  Out*  0.1 μm filter, degassed1  1.5–3.51  100 kPa, 1 h, kmax at –0.5 MPa  HCl, pH 2  Gascò et al. (2006)  1–36  In**  0.1 μmm filter  9  190 kPa, 10 min  Deionized water, 5–150 mM KCl, 200 mM sucrose, 100 mM NaCl  Maherali et al. (2006)  Stems 14, roots: 27–59  Out*  0.2 μm filter, not degassed3  1.5–2  100 kPa, 15–20 min  Distilled water  van Meeteren et al. (2006)  22  In†  Degassed and not degassed  –40  –40 kPa 1.5 h, then 3.4–3 kPa 30 min  0.7 mM CaCl2, 1.5 mM NaHCO3, 50 mM CuSO4  van Ieperen and van Gelder (2006)  7–13  In†  ?  –20  N/A  Ultrapure deionized water, 0.1 mM CaCl2, 1 mM CaCl2, 0.1 mM CaCl2+10 mM KCl, 1 mM CaCl2+10 mM KCl, 0.1 mM CaCl2+100 mM KCl, 1 mM CaCl2+100 mM KCl, 0.1–100 mM KCl, 0.1–10 mM CaCl2  Nardini et al. (2007)  6–16  In**  0.1 μm filter  9  190 kPa, 10 min  Deionized water, 25 mM KCl, 0.5 mM CaCl2, 1 mM CaCl2, 25 mM KCl+0.5 mM CaCl2, 25 mM KCl+1 mM CaCl2, mineral water  Lovisolo et al. (2008)  Petioles 1, shoots 40, roots ?  In††  0.1 μm filter, degassed  Petioles 40; shoots 20; roots 10  Petioles 600 kPa7; shoots 300 kPa7; roots 300 kPa7  15 mM KCl  All studies were conducted at room temperature, except for Tyree and Yang (1992), which was conducted at 1–3 °C. a Apparatus used for hydraulic conductivity measurements: Sperry apparatus (Sperry et al., 1988), **XYL'EM apparatus (Instrutec, Montigny les Cormeilles, France; Cochard et al., 2000), †van Ieperen et al. (2000). ††Hydraulic conductance flow meter (model HCFM-XP, Dynamax, Houston, TX; Tyree et al., 1995). b 1Anna Jacobsen, personal communication; 2Hervé Cochard, personal communication; 3Hafiz Maherali, personal communication; 4Uwe Hacke personal communication; 5John Sperry, personal communication; 6Maciej Zwieniecki, personal communication; 7Pressure gradually increased to the maximum listed, Claudio Lovisolo personal communication; 8Sandra Bucci, personal communication. View Large The following experimental protocols were used to test the five alternative hypotheses. Some treatments potentially affect more then one of the five mechanisms. See Table 1 for a comprehensive listing of treatments and their predicted effects on the five mechanisms. Control treatments The two control treatments included non-degassed, deionized water, and deionized water degassed as described above. Non-steady-state flow To test for lateral flow from vessels into adjacent cells and extracellular spaces, the outflow of the measurement solution was measured simultaneously with the inflow by combining the XYL'EM apparatus with a Sperry apparatus (Sperry et al., 1988). In this set-up, the XYL'EM apparatus provided the water supply under gravitational pressure for measurements and high pressure for flushing. The distal outflow end of the stem segments were connected with tubing to a reservoir placed on an electronic analytical balance (model Explorer Pro EP214DC, Ohaus, Pine Brook, NJ, USA), which was connected to a computer and recorded the outflow rate. Inflow and outflow rates were measured simultaneously at 5 s intervals. To calibrate measurements using the balance against measurements with the high-precision flowmeter (model LIQUI-FLOW L1, Bronkhorst, Montigny les Cormeilles, France) that is at the core of the XYL'EM apparatus, stem samples were replaced with a 0.45 μm syringe filter and 15 cm of luer PVC tubing, which was compressed with a clamp to generate flow rates similar to those observed in Encelia stems. Flow rates were measured the same way as in all other experiments, first at Δp=0 kPa, then at Δp=3 kPa, then at Δp=3 kPa after a first 3-min flush under Δp=150 kPa, and again at Δp=3 kPa after a second 3-min flush under Δp=150 kPa. Comparisons of inflow and outflow rates for ten replicates of Encelia stems, Laurus stems, and tubing were conducted using separate paired, two-sided t tests for the four flow rates measured. Observed differences between the four measured inflow and outflow rates (ΔF=Fin–Fout) were also compared between Encelia stems, Laurus stems, and tubing using two-sided t tests. Test results were corrected for false discovery rate in multiple comparisons (Benjamini and Hochberg, 1995). Swelling of pectin hydrogels in pit membranes To reduce the chance that swelling pectin hydrogels cause a decline in ks in response to flushing, aqueous solutions of 10 mM KCl and 10 mM CaCl2, were used, which are hypothesized to prevent hydrogel swelling at these low concentrations and have been shown to increase flow rates through stems compared to deionized water (Zwieniecki et al., 2001; Gascó et al., 2006; van Ieperen and van Gelder, 2006). A solution of 0.5 mM MgSO4 was also tested, because Mg2+ ions are also known to prevent pectin swelling (Zsivánovits et al., 2005). Ice water was also used as a treatment, because it can cause pectin hydrogels to shrink and become more rigid by forming hydrogen bonds (Lootens et al., 2003; Kjøniksen et al., 2004). Formation and coalescence of bubbles To prevent bubble formation in the stem completely, deionized water was degassed using a membrane contactor (Liqui-Cel mini-module 1.7×5.5, Membrana, Charlotte, NC, USA) in line between the external water supply and the XYL'EM apparatus. The mini-module was used in vacuum mode under 3 kPa absolute pressure, generated using a vacuum diaphragm pump (model DAA-V715A-EB, Gast, Benton Harbor, MI, USA). In addition, solutions were created that contained ionic solutes at concentrations that have been found to prevent the coalescence of small bubbles into larger ones (Craig et al., 1993). Solutions of 50 mM MgSO4, 100 mM CaCl2, 300 mM KCl, and 100 mM (NH4)2SO4 were used because they completely inhibit bubble coalescence, while solutions of 20 mM MgSO4, 40 mM CaCl2, and 100 mM KCl were used because they prevent about 50% of bubble coalescence (Craig et al., 1993). Solutions of 0.5 mM MgSO4, 10 mM CaCl2, and 10 mM KCl used to test for pectin swelling are not expected to inhibit bubble coalescence (Craig et al., 1993). Di-ammonium sulphate (NH4)2SO4 at 100 mM is expected to inhibit bubble coalescence completely (Craig et al., 1993), while only having minor osmotic effects on pectin swelling. Ice water was also used as a treatment, because it can reduce bubble coalescence in deionized water (Ribeiro and Mewes, 2006). Active wounding responses To slow enzyme activities at low water temperatures, branches were both rehydrated and measured in ice water for one treatment and rehydrated at room temperature, but measured in ice water for another treatment. The latter treatment was expected only to affect immediate wounding responses to cutting stems to 15 cm in length just before taking the measurements. In addition, three chemical treatments were applied that previously had been found to reduce xylem blockage and increase the water uptake into cut plant stems, including treatment with 0.25 mM Cu(II)SO4 (Loubaud and van Doorn, 2004) during measurement or 0.10 mM Cu(II)SO4 (Vaslier and van Doorn, 2003) during prehydration and measurement, 10 mM hydroquinone (Loubaud and van Doorn, 2004) during prehydration and measurement, and 10 mM 4-hexylresorcinol (4-HR) (Vaslier and van Doorn, 2003; He et al., 2006) during prehydration and measurement. Ice water as a measuring solution also was expected to slow enzyme activity and thereby inhibit active wound responses. Passive wounding effects The only treatment directly to address passive wounding responses was to remove the bark of stems before cutting them in order to avoid the transfer of resins from the bark to the cut surface, especially in the resin-rich Encelia. Rehydration of partially embolized stem sections Because most woody plant stems under natural conditions will at least be partially embolized and because flushing of stems under high pressure is time-consuming (Tyree and Yang, 1992; Yang and Tyree, 1992) and can cause a decline in ks, a prehydration treatment under vacuum (Sellin, 1991; Hietz et al., 2008) was tested to determine whether it could result in stable measurements of kmax. Ten stem segments (15 cm long) each of Encelia and Laurus were placed on a laboratory bench for 24 h, as previous studies (data not shown) had found that this treatment would induce a high degree of embolism. They were then infiltrated with deionized water under vacuum for 24 h and hydraulic conductance measured as described above. Data analysis The effect of flushing time and flushing pressure on specific hydraulic conductance, ks, through stems of Encelia farinosa and Laurus nobilis were tested using repeated measures ANOVA using SYSTAT (version 12.02, SYSTAT Inc., San Jose, CA, USA). Because the experiments included measurements of kmax spread out over 24 dates during a period of 8 months, for temporal effects on kmax were tested by linearly regressing the treatment means of kmax against date separately for each species using SYSTAT. To test for serial correlation between measurement dates, Durbin–Watson D statistics, calculated by SYSTAT for these regressions, were compared to critical values given in Savin and White (1977). ANOVA was used to analyse kmax as a function of species, treatments, and species×treatment interactions using SYSTAT. To identify kmax values that were significantly higher or lower than the respective overall mean kmax value for the species calculated across 22 separate experiments, post hoc tests were conducted for each species×treatment combination to test the null hypothesis that the treatment mean was equal to the overall species mean. Resulting p-values were corrected for false discovery rate in multiple comparisons (Benjamini and Hochberg, 1995). Responses of ks to repeated high-pressure flushing were characterized by calculating the mean percent change in ks after each 3 min flush for each treatment. These rates of change were analysed by ANOVA in SYSTAT with species, treatments, and species×treatment interactions as effects. To test the null hypothesis of no change in ks, post hoc tests were conducted for each species×treatment combination. Resulting p-values were corrected for false discovery rate in multiple comparisons (Benjamini and Hochberg, 1995). Results The overnight vase rehydration treatment used to refill gas-filled vessels worked very well in Encelia, resulting in a specific hydraulic conductivity, ks [3.29±0.55 (standard error) kg m−1 s−1 MPa−1] that was not statistically different (P=0.18) from that measured in stems harvested at predawn from the same plants after they had been well-watered (4.13±0.25 kg m−1 s−1 MPa−1). In Laurus, vase rehydration was less successful, as rehydrated stems had a lower ks (1.28±0.15 kg m−1 s−1 MPa−1) than stems from the same tree after thorough watering (1.83±0.26 kg m−1 s−1 MPa−1). The difference, which was not statistically significant (p=0.08), was alleviated by rehydrating the stems under vacuum for 20 h, which resulted in a ks of 2.40±0.44 and 1.83±0.26 kg m−1 s−1 MPa−1, which was similar (p=0.27) to that observed after thorough watering. Consecutive 150 kPa flushes with deionized, non-degassed water resulted in significant declines in ks over the 15 min flushing period in both species, while 15 min of low pressure flushing at 3 kPa did not cause such a decline (Fig. 1). No significant differences were observed between inflow rates measured with the XYL'EM apparatus and outflow rates measured with a Sperry apparatus for the control (PVC tubing) and Laurus nobilis (Fig. 2a, c). Inflow rates for Encelia farinosa before flushing under high pressure were slightly higher than outflow rates (p <0.05; Fig. 2b), but the difference disappeared after flushing, and none of the differences between inflow and outflow for Laurus or Encelia were significantly different from those observed for the tubing control. Fig. 1. View largeDownload slide Effect of flushing time and flushing pressure on specific hydraulic conductance, ks, through stems of Encelia farinosa and Laurus nobilis when flushing with deionized, non-degassed water. Effects of species, pressure, time, and all their interactions on ks were tested with repeated measures ANOVA (Species: Sum of Squares (SS) 639.19, degrees of freedom (df) 1, F-ratio (F) 58.982, p-value (p) <0.001; pressure: SS 16.17, df 1, F 1.492, p 0.230; species×pressure: SS 35.68, df 1, F 3.292, p 0.078; time: SS 6.43, df 5, F 7.832, p <0.0001; time×species: SS 0.74, df 5, F 0.899, p 0.48313; time×pressure SS 3.486, df 5, F 4.247, p 0.001; time×species×pressure: SS 1.542, df 5, F 1.878, p 0.100). Slopes of linear trend-lines shown in the graph were significantly different from zero only in the 150 kPa treatments for both species. Fig. 1. View largeDownload slide Effect of flushing time and flushing pressure on specific hydraulic conductance, ks, through stems of Encelia farinosa and Laurus nobilis when flushing with deionized, non-degassed water. Effects of species, pressure, time, and all their interactions on ks were tested with repeated measures ANOVA (Species: Sum of Squares (SS) 639.19, degrees of freedom (df) 1, F-ratio (F) 58.982, p-value (p) <0.001; pressure: SS 16.17, df 1, F 1.492, p 0.230; species×pressure: SS 35.68, df 1, F 3.292, p 0.078; time: SS 6.43, df 5, F 7.832, p <0.0001; time×species: SS 0.74, df 5, F 0.899, p 0.48313; time×pressure SS 3.486, df 5, F 4.247, p 0.001; time×species×pressure: SS 1.542, df 5, F 1.878, p 0.100). Slopes of linear trend-lines shown in the graph were significantly different from zero only in the 150 kPa treatments for both species. Fig. 2. View largeDownload slide Comparison of inflow rates into plant stems of Encelia farinosa, Laurus nobilis, and a tubing control, measured with a XYL'EM apparatus (Cochard et al., 2000) and simultaneous outflow rates from these samples measured with a Sperry apparatus (Sperry et al. 1988). Significant differences (p <0.05) between inflow and outflow rates are designated by an asterisk. Fig. 2. View largeDownload slide Comparison of inflow rates into plant stems of Encelia farinosa, Laurus nobilis, and a tubing control, measured with a XYL'EM apparatus (Cochard et al., 2000) and simultaneous outflow rates from these samples measured with a Sperry apparatus (Sperry et al. 1988). Significant differences (p <0.05) between inflow and outflow rates are designated by an asterisk. The overall mean kmax across 20 experiments (excluding the two ice-water treatments) for Encelia was 2.75±0.20 kg m−1 s−1 MPa−1 and for Laurus was 1.24±0.10 kg m−1 s−1 MPa−1. Means for individual experiments varied from 1.40 to 4.70 kg m−1 s−1 MPa−1 for Encelia and from 0.52 to 2.61 kg m−1 s−1 MPa−1 for Laurus. Small rates of inflow into well-hydrated stems without any driving pressure differential were observed for all treatments. No positive serial correlations among kmax measurements conducted over an 8-month period were found for either species (data not shown), suggesting that treatment effects on kmax masked any seasonal variation that may have existed. This was also evident by significant differences between treatments conducted within just a few days of each other (data not shown). For Encelia, four treatment means were significantly (p <0.05) higher than the overall species mean, including 10 mM KCl, 10 mM CaCl2, 50 mM MgSO4, and prehydration under vacuum and using a degasser at 3 kPa. Hydration and measurement in ice water, 100 mM KCl, 100 mM (NH4)2SO4, and 0.25 mM Cu(II)SO4 in the measurement solution resulted in kmax values that were significantly lower (p <0.05) than the overall mean. For Laurus, the only treatment that differed significantly from the overall mean kmax was an exceptionally high kmax measured with 0.25 mM Cu(II)SO4 solution. Most treatments, including the two control treatments, resulted in declines of ks after flushing at high pressure (Figs 3, 4), with slopes of decline varying from 0% to –41.4% min−1 (Fig. 5). The only significant increase in ks was observed for treatment with 300 mM KCl for Laurus (Fig. 5). Overall treatment effects on rates of decline in ks were significant, but differences between the two species were not, and the species responded similarly to the 22 experimental treatments (Fig. 5), as there was no significant species×treatment interaction. All treatments that were expected to alleviate declines of ks due to prevention of bubble formation or coalescence (Table 1) resulted in rates of change in ks that were not significantly different from zero, with the exception of 100 mM CaCl2 in Laurus. Most treatments predicted to cause at least some alleviation of decline in ks (Table 1), through the prevention of bubble formation or coalescence, also had the predicted effect, except, notably, for measurement with flask-degassed, deionized water and for treatment with ice water during prehydration and measurement for both species. Those treatments predicted to prevent a decline in ks by reducing the swelling of pectins in pit membranes (Table 1), but not by reducing bubble coalescence, such as 10 mM KCl, 10 mM CaCl2, or 0.5 mM MgSO4, did not prevent a decline in ks, but a treatment that strongly affected bubble coalescence but not pectin swelling, 100 mM (NH4)2SO4, did (Fig. 5). Treatments applied to reduce active wounding responses did not prevent a decline of ks, except for 0.25 mM Cu(II)SO4 in the measurement solution. Removing the bark before cutting the stems under water did not prevent a decline in ks, in Encelia, but produced what appeared to be an experimental artefact in Laurus, in which bark removal was difficult and required substantial handling of the stems. This resulted in low initial flow rates that increased after the first flush and then decreased after the second flush, causing an overall change in ks that was not significantly different from zero. Prehydration for 24 h under 3 kPa vacuum resulted in stable measurements of ks, whether the stems were well hydrated or embolized (Fig. 6). Fig. 3. View largeDownload slide Effects of physical treatments on the specific hydraulic conductivity, ks, of fully hydrated Encelia and Laurus stems before high-pressure flushing (initial), after the first 3-min high-pressure flush, and after a second 3-min high-pressure flush. *Treatment applied during vase rehydration and during measurement. Fig. 3. View largeDownload slide Effects of physical treatments on the specific hydraulic conductivity, ks, of fully hydrated Encelia and Laurus stems before high-pressure flushing (initial), after the first 3-min high-pressure flush, and after a second 3-min high-pressure flush. *Treatment applied during vase rehydration and during measurement. Fig. 4. View largeDownload slide Effects of chemical treatments on the specific hydraulic conductivity, ks, of fully hydrated Encelia and Laurus stems before high-pressure flushing (initial), after the first 3-min high-pressure flush, and after a second 3-min high-pressure flush. *Treatment applied during vase rehydration and during measurement. Fig. 4. View largeDownload slide Effects of chemical treatments on the specific hydraulic conductivity, ks, of fully hydrated Encelia and Laurus stems before high-pressure flushing (initial), after the first 3-min high-pressure flush, and after a second 3-min high-pressure flush. *Treatment applied during vase rehydration and during measurement. Fig. 5. View largeDownload slide Effects of treatments on the change in specific hydraulic conductivity ks (±95% confidence intervals) of fully hydrated Encelia and Laurus stems in response to high-pressure flushing. Values with 95% confidence intervals overlapping the zero-line are not significantly different from zero. Overall effects of treatments on change in ks were highly significant (sum of squares, SS=5,638.4, degrees of freedom, df=21, F-ratio=6.765, p <0.001), but differences in rates of change in ks between species were not (SS=59.457, df=1, F-ratio=1.498, p=0.222), nor were species×treatment interactions (SS=1,215.037, df=21, F-ratio=1.458, p=0.089). Fig. 5. View largeDownload slide Effects of treatments on the change in specific hydraulic conductivity ks (±95% confidence intervals) of fully hydrated Encelia and Laurus stems in response to high-pressure flushing. Values with 95% confidence intervals overlapping the zero-line are not significantly different from zero. Overall effects of treatments on change in ks were highly significant (sum of squares, SS=5,638.4, degrees of freedom, df=21, F-ratio=6.765, p <0.001), but differences in rates of change in ks between species were not (SS=59.457, df=1, F-ratio=1.498, p=0.222), nor were species×treatment interactions (SS=1,215.037, df=21, F-ratio=1.458, p=0.089). Fig. 6. View largeDownload slide Effects of 24 h prehydration of submerged woody plant stems under 3 kPa (absolute pressure) vacuum on the change in specific hydraulic conductivity ks (±95% confidence intervals) in response to repeated high pressure flushing. Values with 95% confidence intervals overlapping the zero-line are not significantly different from zero. Multiple points for Laurus and Encelia stand for separate experiments. Fig. 6. View largeDownload slide Effects of 24 h prehydration of submerged woody plant stems under 3 kPa (absolute pressure) vacuum on the change in specific hydraulic conductivity ks (±95% confidence intervals) in response to repeated high pressure flushing. Values with 95% confidence intervals overlapping the zero-line are not significantly different from zero. Multiple points for Laurus and Encelia stand for separate experiments. Discussion Stems of Laurus nobilis proved to be very difficult to rehydrate. Most flushing treatments did not result in specific hydraulic conductivities, ks, as high as those measured in stems taken at predawn after thorough watering over two nights. The failure of vase rehydration was most probably due to stomatal closure, observed for leaves of cut branches (data not shown), and rehydrating the stems under vacuum for 20 h restored ks to the levels seen in well-watered plants. This suggests that vase storage overnight did not induce rapid gel or tylose formation in vessels. Encelia farinosa branches rehydrated well in the vase, and their stomata did not close. Thus, for Laurus, measurements of ks probably remained below kmax in this study, as Laurus stems remained partially embolized. This does not affect the conclusions to be drawn from this study, which addressed the problem of declines in ks rather than the question of how to achieve true measurements of kmax, but it raises the question of how reliable measurements of kmax can be achieved in different species. Our results show that flushing stems under high pressure can be problematic. Rehydration under vacuum (Hietz et al., 2008) may affect xylem structure and function, so it may not be suitable for all studies. Our comparison of methods between a XYL'EM apparatus (Cochard et al., 2000) that measures inflow into a stem and a Sperry apparatus (Sperry et al., 1988) that measures outflow from a stem showed no significant differences between flow rates, suggesting that both methods lead to similar measurements of maximum hydraulic conductivity (kmax). Initial inflow into Encelia stems slightly exceeded outflow, but did not significantly exceed inflow and outflow comparisons for a tubing control. Observations of higher inflow than outflow had been previously reported by Tyree and Yang (1992). As the differences observed in this study were minor, it appears that there is little concern about potential unpredictable secondary responses to lateral water flow, such as the movement of ions into vessels, that may affect hydraulic conductance (van Ieperen, 2007). Interestingly, inflow into well-hydrated stems was consistently observed, even without any driving pressure differential. Interestingly, this no-pressure flow declined to zero in Laurus after repeated flushes, but not in Encelia (data not shown). Initial flow through stems without a driving pressure differential is commonly observed in plant hydraulics studies (John Sperry, personal communication) and is routinely deducted from measured flow rates under pressure. The mechanisms responsible for such no-pressure flow would certainly seem to warrant some attention. Declines in hydraulic conductance have been attributed to the swelling of pectin hydrogels in pit membranes as Ca2+ ions are flushed from the pectins by deionized water (Zwieniecki et al., 2001). Cations, such as K+, Na+, and Ca2+, can reduce hydration of pectin gels by associating with negatively-charged galacturonic acid groups (Zwieniecki et al., 2001; van Ieperen, 2007). In addition, Ca2+ ions crosslink demthylesterized pectin molecules (Jarvis, 1984; Willats et al., 2001) and thereby increase gel rigidity and decrease hydration. Mg2+ ions appear to have a similar effect (Zsivánovits et al., 2005). In our experiments, low concentrations of K+, Mg2+, and Ca2+ salts caused high values of kmax in Encelia, but did not prevent decline in ks in either species. Observed alleviations of declines at high salt concentrations were probably due to the property of these salts to inhibit bubble coalescence almost completely at these higher concentrations (Marrucci and Nicodemo, 1967; Lessard and Zieminski, 1971; Craig et al., 1993; Zahradník et al., 1995). This interpretation is strongly supported by the finding that (NH4)2SO4, a very effective bubble coalescence inhibitor (Craig et al., 1993), alleviated the decline in ks, even though ammonium ions are unlikely to affect pectin hydration. The stem segments used in this study were shorter than the maximum vessel length, which means that at least some water could pass through open vessels without crossing pit membranes. Other studies conducted to date to test the pectin hydrogel hypotheses have used stem lengths ranging from 3 cm to 16 cm (Zwieniecki et al., 2001; Boyce et al., 2004; Nardini et al., 2007), with only one study examining stem lengths of up to 36 cm (Gascó et al., 2006). Flow through open vessels did not obscure membrane effects in any of these studies. Four chemical and two physical treatments that were predicted to alleviate declines in ks by preventing either the formation or coalescence of gas bubbles in the xylem all had the predicted effect, providing very strong support for the conclusion that declines in ks are caused by the blockage of vessels by bubbles. Similarly, cut flowers have been found to take up degassed water more rapidly than non-degassed water (van Ieperen et al., 2002; van Meeteren et al., 2006). Our findings confirm the conclusions of Kelso et al. (1963) and Canny et al. (2007), who proposed that a thorough degassing of the measurement solution is essential for stable and precise measurements of hydraulic conductivity. Our findings show that thorough degassing, as provided by the membrane contactors used in this study, is needed to obtain stable kmax. It is important to note that the common degassing method of stirring water under laboratory vacuum was not sufficient to prevent a decline of kmax (Fig. 5). Our findings of bubbles as the main cause for declining hydraulic conductivity through woody stems while under positive pressure raise the question whether similar effects occur in the transpiration stream under natural conditions, as proposed by Canny et al. (2007). This could indeed be the case if xylem sap is air-saturated, as normally assumed (Lybeck, 1959; Hammel, 1967; Sperry and Sullivan, 1992; Yang and Tyree, 1992; McCully et al., 2000; Cobb et al., 2007). Passage of air-saturated water through pit membranes could cause the formation of bubbles through heterogeneous nucleation, followed by bubble coalescence, and the air blockage of vessels. It should be noted, however, that contrary to the suggestion by Canny et al. (2007), this would not occur due to the water pressure drop that occurs as water flows through constrictions, because decreasing water pressure actually increases air solubility (Mercury and Tardy, 2001; Mercury, 2006). The familiar gassing-out of water under vacuum only occurs when water is in contact with a gas phase under partial vacuum (Mercury et al., 2003), which is not the case in xylem because it is in contact with a gas phase under atmospheric pressure. Gassing-out of xylem sap could still occur due to heterogeneous nucleation or in response to lower air solubility with rising temperatures as the xylem sap emerges from cool roots into warmer stems and branches. Physiological wounding responses are a concern in hydraulic studies of cut plant samples. Responses to wounding could be induced by a rapid increase in water potential, which sets off hydraulic and/or electrical signals (Stahlberg and Cosgrove, 1995; Stahlberg et al., 2005; Fromm and Lautner, 2007). Cell-wall proteins, some of which reside in pit membranes (Harrak et al., 1999), can react within minutes to wounding (Bradley et al., 1992). Experiments with cut flowers have shown that physiological wounding responses can be reduced by keeping stems at low temperatures (van Meeteren, 1992; van Doorn and Cruz, 2000). In this study, measurement in deionized water at 4–6 °C reduced the decline in ks, but this effect may have been due to several causes, including increasing air solubility that reduced bubble formation, reduced bubble coalescence (Ribeiro and Mewes, 2006), shrinking of pectin hydrogels (Kjøniksen et al., 2004; Lootens et al., 2003), or reduced physiological activity. A more specific inhibition of active wounding responses was expected from chemical treatments that had been found previously to reduce active wounding responses, reduce xylem blockage, and increase water uptake into cut plant stems (van Meeteren et al., 2000; van Doorn and Vaslier, 2002; Vaslier and van Doorn, 2003; Loubaud and van Doorn, 2004; He et al., 2006). In this study, Cu2+ ions, applied at a concentration of 0.25 mM for measurement and flushing, reduced the decline in hydraulic conductance, as found in previous studies (van Doorn and Vaslier, 2002; Vaslier and van Doorn, 2003; Loubaud and van Doorn, 2004). Copper is an essential micronutrient, is toxic at super-optimal levels, and plays a wide variety of roles in plant biochemistry (Maksymiec, 1997). Its functions in xylem may include effects on peroxidases, which are among the most common enzymes in xylem sap (Buhtz et al., 2004; Kehr et al., 2005), and whose activities Cu2+ ions have been found to increase (Chen et al., 2002) or inhibit (Zancani et al., 1995), probably depending on the nature of specific peroxidases. Cu2+ ions also bind strongly to cell wall pectins (Dronnet et al., 1996; Wehr et al., 2004) and may thereby inhibit enzymatic attack (Wehr et al., 2004) and/or contribute to cell wall loosening by causing non-enzymatic scission of other wall polysaccharides (Fry et al., 2002). In our study, 0.25 mM Cu(II)SO4 caused exceptionally high values of kmax in Laurus and exceptionally low values of kmax in Encelia, which means that this is probably not a treatment to be recommended for a standardized protocol of kmax measurements. The other two chemicals applied to reduce active wounding responses, hydroquinone and 4-hexylresorcinol (4-HR), had no effect on preventing the decline in kmax. These chemicals were chosen because they have been found to reduce xylem blockage and delay wilting in cut flowers (van Doorn and Vaslier, 2002; Vaslier and van Doorn, 2003; Loubaud and van Doorn, 2004). Hydroquinone, a peroxidase substrate (Bagirova et al., 2001) and inhibitor (Martinez et al., 2001) is a powerful phytotoxin at the concentration used in this study (10 mM) that strongly affects cell membrane integrity (Pandey et al., 2005). The polyphenol oxidase inhibitor 4-hexylresorcinol (Dawley and Flurkey, 1993) had no effect in our study. Polyphenol oxidases play roles in lignifying xylem and in wounding responses (Richardson et al., 2000; Mayer, 2006). They do not appear to occur normally in mature xylem (Buhtz et al., 2004; Kehr et al., 2005), so it is not clear how they could affect water uptake through cut stems. Cutting of plant stems could decrease hydraulic conductivity by clogging of vessels with gums, resins, or debris from the xylem and the bark. To reduce the transfer of substances from the bark to the xylem, bark was removed from submerged stems before cutting them. This treatment had no effect on alleviating the decline of ks in Encelia, and produced experimental artefacts in Laurus that lead to inconclusive results. Examination of stems for evidence of debris in the xylem was initially planned for this study, but the plans were dropped as too labour-intensive and unnecessary after it was found that the declines in ks could be alleviated with methods that affect bubble formation and coalescence. Arguing that a lack of method standardization in plant hydraulics could negatively affect progress in the field, Pratt et al. (2008) recently called for a common toolbox of methods. A comparison of methods used by different plant hydraulics research laboratories to measure kmax through stems (Table 2) shows that methods vary widely in the length of samples, pressure differentials and flushing times, chemical composition of measurements solutions, and in whether or not solutions are degassed. As a result, measurements of kmax and the percentage loss of conductance (PLC) in xylem vulnerability curves from different published studies are strictly comparable only within laboratories and among collaborating researchers. Two conclusions from this study would seem to merit incorporation into future protocols to measure kmax: (i) Thorough degassing of measurement solutions using a degassing unit, such as the Liqui-Cel mini-module used in this study, and complete air removal from the measuring apparatus. (ii) Hydration of submerged samples under vacuum instead of flushing at high pressure. The latter is recommended based on the findings of this study for angiosperm stems, based on a previous study of conifer stems (Hietz et al., 2008), and, most importantly, based on theoretical considerations and measurements which showed that complete air removal from a severely embolized stems by high-pressure flushing takes many hours (Yang and Tyree, 1992). Eliminating high-pressure flushing from protocols to measure native embolisms and kmax has the added advantage of allowing increased sample sizes by shifting measurements of kmax to the next day after the samples have been incubated for 20–24 h under vacuum. Subsequent to the experiments described in this paper, it was found that vacuum hydration lead to stable measurements of kmax for eight out of ten species tested, with relatively small declines for the remaining two species (1–3% min−1; data not shown). A detailed protocol for measurements of native embolisms and kmax,, especially for XYL'EM users, is available from the authors upon request. Vacuum hydration was found to lead to stable measurements of kmax in this study, but further comparative research on the effects of vacuum versus high pressure infiltration of angiosperm wood seems warranted. Both techniques could cause artefacts involving non-conducting xylem cells and could potentially damage xylem structures. Several hours of flushing with thoroughly degassed water under very low pressure may be the optimal solution. Another technique that reduces the chance of bubble formation during flushing is to measure and flush very short stem segments (2–4 cm). This technique used by Hervé Cochard and colleagues (e.g. Hukin et al., 2005) greatly reduces the number of vessel endings at which bubbles could form or get entrapped within a stem segment. By allowing water to flow through a stem without crossing vessel ends, this technique would tend to overestimate kmax, but this may be of little concern for measurements of PLC within a species. Obviously, the choice of methods used for measuring kmax will depend on the study system and the questions asked. Adding chemicals to the measurement solution provided no clear advantage in this study over the use of thoroughly degassed, deionized water, which is why the latter is suggested as the easiest choice for a standard solution. Responses of kmax to salts vary widely between species and seasonally (Gascó et al., 2007; Trifilò et al., 2008), which means that there is no recipe for a chemical composition of the measuring solution that could be a standard choice for all studies. Deionized and thoroughly degassed water therefore appears to be the most parsimonious choice for a standard measurement solution. We thank Judy Quang, Tracy Vo, and Daisha Ortega for help with measurements, Ernesto Cassillas, Vickie Nguyen, and Matthew Sutton for help with sampling. Anna Jacobsen, Sandra Bucci, Hervé Cochard, Hafiz Maherali, Uwe Hacke, John Sperry, Maciej Zwieniecki, Claudio Lovisolo, and Mel Tyree generously shared details about their experimental protocols summarized in Table 2. Comments from three anonymous reviewers greatly helped to improve the manuscript. This work was supported by the Andrew W Mellon Foundation and the National Science Foundation (IOS-0641765). References Alder NN,  Pockman WT,  Sperry JS,  Nuismer SM.  Use of centrifugal force in the study of xylem cavitation,  Journal of Experimental Botany ,  1997, vol.  48 (pg.  665- 674) Google Scholar CrossRef Search ADS   Bagirova NA,  Shekhovtsova TN,  van Huystee RB.  Enzymatic determination of phenols using peanut peroxidase,  Talanta ,  2001, vol.  55 (pg.  1151- 1164) Google Scholar CrossRef Search ADS PubMed  Benjamini Y,  Hochberg Y.  Controlling the false discovery rate: a practical and powerful approach to multiple testing,  Journal of the Royal Statistical Society B ,  1995, vol.  57 (pg.  289- 300) Blatteau J-E,  Souraud J-B,  Gempp E,  Boussuges A.  Gas nuclei, their origin, and their role in bubble formation,  Aviation, Space, and Environmental Medicine ,  2006, vol.  77 (pg.  1068- 1076) Google Scholar PubMed  Bonsen KJM,  Kučera LJ.  Vessel occlusions in plants - morphological, functional and evolutionary aspects,  Iawa Bulletin ,  1990, vol.  11 (pg.  393- 399) Google Scholar CrossRef Search ADS   Boyce CK,  Zwieniecki MA,  Cody GD,  Jacobsen C,  Wirick S,  Knoll AH,  Holbrook NM.  Evolution of xylem lignification and hydrogel transport regulation,  Proceedings of the National Academy of Sciences (USA) ,  2004, vol.  101 (pg.  17555- 17558) Google Scholar CrossRef Search ADS   Bradley DJ,  Kjellbom P,  Lamb CJ.  Elicitor- and wound-induced oxidative cross-linking of a proline-rich plant cell wall protein: A novel, rapid defense response,  Cell ,  1992, vol.  70 (pg.  21- 30) Google Scholar CrossRef Search ADS PubMed  Bucci SJ,  Scholz FG,  Goldstein G,  Meinzer FC,  Sternberg LDSL.  Dynamic changes in hydraulic conductivity in petioles of two savanna tree species: factors and mechanisms contributing to the refilling of embolized vessels,  Plant, Cell and Environment ,  2003, vol.  26 (pg.  1633- 1645) Google Scholar CrossRef Search ADS   Buhtz A,  Kolasa A,  Arlt K,  Walz C,  Kehr J.  Xylem sap protein composition is conserved among different plant species,  Planta ,  2004, vol.  219 (pg.  610- 618) Google Scholar CrossRef Search ADS PubMed  Canny MJ,  Sparks JP,  Huang CX,  Roderick ML.  Air embolisms exsolving in the transpiration water: the effect of constrictions in the xylem pipes,  Functional Plant Biology ,  2007, vol.  34 (pg.  95- 111) Google Scholar CrossRef Search ADS   Chen EL,  Chen YA,  Chen LM,  Liu ZH.  Effect of copper on peroxidase activity and lignin content in Raphanus sativus,  Plant Physiology and Biochemistry ,  2002, vol.  40 (pg.  439- 444) Google Scholar CrossRef Search ADS   Cheong YH,  Chang HS,  Gupta R,  Wang X,  Zhu T,  Luan S.  Transcriptional profiling reveals novel interactions between wounding, pathogen, abiotic stress, and hormonal responses in Arabidopsis,  Plant Physiology ,  2002, vol.  129 (pg.  661- 677) Google Scholar CrossRef Search ADS PubMed  Cobb AR,  Choat B,  Holbrook NM.  Dynamics of freeze–thaw embolism in Smilax rotundifolia (Smilacaceae),  American Journal of Botany ,  2007, vol.  94 (pg.  640- 649) Google Scholar CrossRef Search ADS PubMed  Cochard H,  Bodet C,  Améglio T,  Cruiziat P.  Cryo-scanning electron microscopy observations of vessel content during transpiration in walnut petioles. Facts or artifacts?,  Plant Physiology ,  2000, vol.  124 (pg.  1191- 1202) Google Scholar CrossRef Search ADS PubMed  Cochard H,  Herbette S,  Hernández E,  Hölttä T,  Mencuccini M.  The effects of sap ionic composition on xylem vulnerability to cavitation,  Journal of Experimental Botany ,  2010, vol.  61 (pg.  275- 285) Google Scholar CrossRef Search ADS PubMed  Cochard H,  Tyree MT.  Xylem dysfunction in Quercus: vessel sizes, tyloses, cavitation and seasonal changes in embolism,  Tree Physiology ,  1990, vol.  6 (pg.  393- 407) Google Scholar CrossRef Search ADS PubMed  Craig VSJ,  Ninham BW,  Pashley RM.  The effect of electrolytes on bubble coalescence in water,  Journal of Physical Chemistry ,  1993, vol.  97 (pg.  10192- 10197) Google Scholar CrossRef Search ADS   Crews LJ,  McCully ME,  Canny MJ.  Mucilage production by wounded xylem tissue of maize roots – time course and stimulus,  Functional Plant Biology ,  2003, vol.  30 (pg.  755- 766) Google Scholar CrossRef Search ADS   Dawley RM,  Flurkey WH.  Differentiation of tyrosinase and laccase using 4-hexyl-resorcinol, a tyrosinase inhibitor,  Phytochemistry ,  1993, vol.  33 (pg.  281- 284) Google Scholar CrossRef Search ADS   Dronnet VM,  Renard CMGC,  Axelos MAV,  Thibault JF.  Characterisation and selectivity of divalent metal ions binding by citrus and sugar-beet pectins,  Carbohydrate Polymers ,  1996, vol.  30 (pg.  253- 263) Google Scholar CrossRef Search ADS   Fromm J,  Lautner S.  Electrical signals and their physiological significance in plants,  Plant, Cell and Environment ,  2007, vol.  30 (pg.  249- 257) Google Scholar CrossRef Search ADS   Fry SC,  Miller JG,  Dumville JC.  A proposed role for copper ions in cell wall loosening,  Plant and Soil ,  2002, vol.  247 (pg.  57- 67) Google Scholar CrossRef Search ADS   Gascó A,  Nardini A,  Gortan E,  Salleo S.  Ion-mediated increase in the hydraulic conductivity of Laurel stems: role of pits and consequences for the impact of cavitation on water transport,  Plant, Cell and Environment ,  2006, vol.  29 (pg.  1946- 1955) Google Scholar CrossRef Search ADS   Gascó A,  Salleo S,  Gortan E,  Nardini A.  Seasonal changes in the ion-mediated increase of xylem hydraulic conductivity in stems of three evergreens: any functional role?,  Physiologia Plantarum ,  2007, vol.  129 (pg.  597- 606) Google Scholar CrossRef Search ADS   Hacke UG,  Sperry JS.  Limits to xylem refilling under negative pressure in Laurus nobilis and Acer negundo,  Plant, Cell and Environment ,  2003, vol.  26 (pg.  303- 311) Google Scholar CrossRef Search ADS   Hacke UG,  Sperry JS,  Pittermann J.  Drought experience and cavitation resistance in six shrubs from the Great Basin, Utah,  Basic and Applied Ecology ,  2000, vol.  1 (pg.  31- 41) Google Scholar CrossRef Search ADS   Hammel HT.  Freezing of xylem sap without cavitation,  Plant Physiology ,  1967, vol.  42 (pg.  55- 66) Google Scholar CrossRef Search ADS PubMed  Harrak H,  Chamberland H,  Plante M,  Bellemare C,  Lafontaine JG,  Tabaeizadeh Z.  A proline-, threonine-, and glycine-rich protein down-regulated by drought is localized in the cell wall of xylem elements,  Plant Physiology ,  1999, vol.  121 (pg.  557- 564) Google Scholar CrossRef Search ADS PubMed  He S,  Joyce DC,  Irving DE,  Faragher JD.  Stem end blockage in cut Grevillea 'Crimson Yul-lo' inflorescences,  Postharvest Biology and Technology ,  2006, vol.  41 (pg.  78- 84) Google Scholar CrossRef Search ADS   Hietz P,  Rosner S,  Sorz J,  Mayr S.  Comparison of methods to quantify loss of hydraulic conductivity in Norway spruce,  Annals of Forest Science ,  2008, vol.  65 pg.  502  Google Scholar CrossRef Search ADS   Hukin D,  Cochard H,  Dreyer E,  Thiec DL,  Bogeat-Triboulot MB.  Cavitation vulnerability in roots and shoots: does Populus euphratica Oliv., a poplar from arid areas of Central Asia, differ from other poplar species?,  Journal of Experimental Botany ,  2005, vol.  56 (pg.  2003- 2010) Google Scholar CrossRef Search ADS PubMed  Jacobsen AL,  Ewers FW,  Pratt RB,  Paddock WAIII,  Davis SD.  Do xylem fibers affect vessel cavitation resistance?,  Plant Physiology ,  2005, vol.  139 (pg.  546- 556) Google Scholar CrossRef Search ADS PubMed  Jarbeau JA,  Ewers FW,  Davis SD.  The mechanism of water stress-induced embolism in two species of chaparral shrubs,  Plant, Cell and Environment ,  1995, vol.  18 (pg.  189- 196) Google Scholar CrossRef Search ADS   Jarvis MC.  Structure and properties of pectin gels in plant cell walls,  Plant, Cell and Environment ,  1984, vol.  7 (pg.  153- 164) Kehr J,  Buhtz A,  Giavalisco P.  Analysis of xylem sap proteins from Brassica napus,  BMC Plant Biology ,  2005, vol.  5 pg.  11  Google Scholar CrossRef Search ADS PubMed  Kelso WCJr,  Gertjejansen RO,  Hossfeld RL.  The effect of air blockage upon the permeability of wood to liquids,  University of Minnesota Agricultural Experiment Station Technical Bulletin ,  1963, vol.  242 (pg.  1- 40) Kjøniksen A-L,  Hiorth M,  Nyström B.  Temperature-induced association and gelation of aqueous solutions of pectin. A dynamic light scattering study,  European Polymer Journal ,  2004, vol.  40 (pg.  2427- 2435) Google Scholar CrossRef Search ADS   Lessard RR,  Zieminski SA.  Bubble coalescence and gas transfer in aqueous electrolytic solutions,  Industrial and Engineering Chemistry Fundamentals ,  1971, vol.  10 (pg.  260- 269) Google Scholar CrossRef Search ADS   Li R,  Rimmer R,  Yu M,  Sharpe AG,  Séguin-Swartz G,  Lydiate D,  Hegedus DD.  Two Brassica napus polygalacturonase inhibitory protein genes are expressed at different levels in response to biotic and abiotic stresses,  Planta ,  2003, vol.  217 (pg.  299- 308) Google Scholar PubMed  Loepfe L,  Martienez-Vilalta J,  Piñol J,  Mencuccini M.  The relevance of xylem network structure for plant hydraulic efficiency and safety,  Journal of Theoretical Biology ,  2007, vol.  247 (pg.  788- 803) Google Scholar CrossRef Search ADS PubMed  Lootens D,  Capel F,  Durand D,  Nicolai T,  Boulenguer P,  Langendorff V.  Influence of pH, Ca concentration, temperature and amidation on the gelation of low methoxyl pectin,  Food Hydrocolloids ,  2003, vol.  17 (pg.  237- 244) Google Scholar CrossRef Search ADS   Lopez-Portillo J,  Ewers FW,  Angeles G.  Sap salinity effects on xylem conductivity in two mangrove species,  Plant, Cell and Environment ,  2005, vol.  28 (pg.  1285- 1292) Google Scholar CrossRef Search ADS   Loubaud M,  van Doorn WG.  Wound-induced and bacteria-induced xylem blockage in roses, Astilbe, and Viburnum,  Postharvest Biology and Technology ,  2004, vol.  32 (pg.  281- 288) Google Scholar CrossRef Search ADS   Lovisolo C,  Perrone I,  Hartung W,  Schubert A.  An abscisic acid-related reduced transpiration promotes gradual embolism repair when grapevines are rehydrated after drought,  New Phytologist ,  2008, vol.  180 (pg.  642- 651) Google Scholar CrossRef Search ADS PubMed  Lybeck BR.  Winter freezing in relation to the rise of sap in tall trees,  Plant Physiology ,  1959, vol.  34 (pg.  482- 486) Google Scholar CrossRef Search ADS PubMed  Maherali H,  Moura CF,  Caldeira MC,  Willson CJ,  Jackson RB.  Functional coordination between leaf gas exchange and vulnerability to xylem cavitation in temperate forest trees,  Plant, Cell and Environment ,  2006, vol.  29 (pg.  571- 583) Google Scholar CrossRef Search ADS   Maherali H,  Pockman WT,  Jackson RB.  Adaptive variation in the vulnerability of woody plants to xylem cavitation,  Ecology ,  2004, vol.  85 (pg.  2184- 2199) Google Scholar CrossRef Search ADS   Maksymiec W.  Effect of copper on cellular processes in higher plants,  Photosynthetica ,  1997, vol.  34 (pg.  321- 342) Google Scholar CrossRef Search ADS   Marrucci G,  Nicodemo L.  Coalescence of gas bubbles in aqueous solutions of inorganic electrolytes,  Chemical Engineering Science ,  1967, vol.  22 (pg.  1257- 1265) Google Scholar CrossRef Search ADS   Martinez GA,  Civello PM,  Chaves AR,  Añón MC.  Characterization of peroxidase-mediated chlorophyll bleaching in strawberry fruit,  Phytochemistry ,  2001, vol.  58 (pg.  379- 387) Google Scholar CrossRef Search ADS PubMed  Mayer AM.  Polyphenol oxidases in plants and fungi: going places? A review,  Phytochemistry ,  2006, vol.  67 (pg.  2318- 2331) Google Scholar CrossRef Search ADS PubMed  McCully ME,  Shane MW,  Baker AN,  Huang CX,  Ling LEC,  Canny MJ.  The reliability of cryoSEM for the observation and quantification of xylem embolisms and quantitative analysis of xylem sap in situ,  Journal of Microscopy ,  2000, vol.  198 (pg.  24- 33) Google Scholar CrossRef Search ADS PubMed  Mercury L.  Somasundaran P.  Gas solubilities in capillary water confined in finely dispersed systems,  Encyclopedia of surface and colloid science ,  2006, vol.  Vol. 4  New York Taylor & Francis(pg.  2665- 2677) Mercury L,  Azaroual M,  Zeyen H,  Tardy Y.  Thermodynamic properties of solutions in metastable systems under negative or positive pressures,  Geochimica et Cosmochimica Acta ,  2003, vol.  67 (pg.  1769- 1785) Google Scholar CrossRef Search ADS   Mercury L,  Tardy Y.  Negative pressure of stretched liquid water. geochemistry of soil capillaries,  Geochimica et Cosmochimica Acta ,  2001, vol.  65 (pg.  3391- 3408) Google Scholar CrossRef Search ADS   Nardini A,  Gascò A,  Trifilò P,  Lo Gullo MA,  Salleo S.  Ion-mediated enhancement of xylem hydraulic conductivity is not always suppressed by the presence of Ca2+in the sap,  Journal of Experimental Botany ,  2007, vol.  58 (pg.  2609- 2615) Google Scholar CrossRef Search ADS PubMed  Pandey DK,  Mishra N,  Singh P.  Relative phytotoxicity of hydroquinone on rice (Oryza sativa L.) and associated aquatic weed green musk chary (Chary zeylanica Willd.),  Pesticide Biochemistry and Physiology ,  2005, vol.  83 (pg.  82- 96) Google Scholar CrossRef Search ADS   Pockman WT,  Sperry JS.  Vulnerability to cavitation and the distribution of Sonoran Desert vegetation,  American Journal of Botany ,  2000, vol.  87 (pg.  1287- 1299) Google Scholar CrossRef Search ADS PubMed  Pratt RB,  Jacobsen AL,  North GB,  Sack L,  Schenk HJ.  Plant hydraulics: new discoveries in the pipeline,  New Phytologist ,  2008, vol.  179 (pg.  590- 593) Google Scholar CrossRef Search ADS PubMed  Ramonell KM,  Somerville S.  The genomics parade of defense responses: to infinity and beyond,  Current Opinion in Plant Biology ,  2002, vol.  5 (pg.  1- 4) Google Scholar CrossRef Search ADS   Ribeiro CP,  Mewes D.  On the effect of liquid temperature upon bubble coalescence,  Chemical Engineering Science ,  2006, vol.  61 (pg.  5704- 5716) Google Scholar CrossRef Search ADS   Richardson A,  Duncan J,  McDougall GJ.  Oxidase activity in lignifying xylem of a taxonomically diverse range of trees: identification of a conifer laccase,  Tree Physiology ,  2000, vol.  20 (pg.  1039- 1047) Google Scholar CrossRef Search ADS PubMed  Ryden P,  MacDougall AJ,  Tibbits CW,  Ring SG.  Hydration of pectic polysaccharides,  Biopolymers ,  2000, vol.  54 (pg.  398- 405) Google Scholar CrossRef Search ADS PubMed  Salleo S,  Lo Gullo MA,  Trifilò P,  Nardini A.  New evidence for a role of vessel-associated cells and phloem in the rapid xylem refilling of cavitated stems of Laurus nobilis L,  Plant, Cell and Environment ,  2004, vol.  27 (pg.  1065- 1076) Google Scholar CrossRef Search ADS   Savin NE,  White KJ.  The Durbin–Watson test for serial correlation with extreme sample sizes or many regressors,  Econometrica ,  1977, vol.  45 (pg.  1989- 1996) Google Scholar CrossRef Search ADS   Scardina P,  Edwards M.  Air binding of granular media filters,  Journal of Environmental Engineering ,  2004, vol.  130 (pg.  1126- 1138) Google Scholar CrossRef Search ADS   Schulte PJ,  Gibson AC,  Nobel PS.  Xylem anatomy and hydraulic conductance of Psilotum nudum,  American Journal of Botany ,  1987, vol.  74 (pg.  1438- 1445) Google Scholar CrossRef Search ADS   Sellin A.  Hydraulic conductivity of xylem depending on water saturation level in Norway spruce (Picea abies (L.) Karst),  Journal of Plant Physiology ,  1991, vol.  138 (pg.  466- 469) Google Scholar CrossRef Search ADS   Sperry JS,  Donnelly JR,  Tyree MT.  A method for measuring hydraulic conductivity and embolism in xylem,  Plant, Cell and Environment ,  1988, vol.  11 (pg.  35- 40) Google Scholar CrossRef Search ADS   Sperry JS,  Holbrook NM,  Zimmermann MH,  Tyree MT.  Spring filling of xylem vessels in wild grapevine,  Plant Physiology ,  1987, vol.  83 (pg.  414- 417) Google Scholar CrossRef Search ADS PubMed  Sperry JS,  Saliendra NZ.  Intra- and inter-plant variation in xylem cavitation in Betula occidentalis,  Plant, Cell and Environment ,  1994, vol.  17 (pg.  1233- 1241) Google Scholar CrossRef Search ADS   Sperry JS,  Sullivan JEM.  Xylem embolism in response to freeze–thaw cycles and water stress in ring-porous, diffuse-porous, and conifer species,  Plant Physiology ,  1992, vol.  100 (pg.  605- 613) Google Scholar CrossRef Search ADS PubMed  Sperry JS,  Tyree MT.  Water-stress-induced xylem embolism in three species of conifers,  Plant, Cell and Environment ,  1990, vol.  13 (pg.  427- 436) Google Scholar CrossRef Search ADS   Stahlberg R,  Cosgrove DJ.  Comparison of electric and growth responses to excision in cucumber and pea-seedlings. 2. Long-distance effects are caused by the release of xylem pressure,  Plant, Cell and Environment ,  1995, vol.  18 (pg.  33- 41) Google Scholar CrossRef Search ADS   Stahlberg R,  Cleland RE,  Van Volkenburgh E.  Decrement and amplification of slow wave potentials during their propagation in Helianthus annuus L. shoots,  Planta ,  2005, vol.  220 (pg.  550- 558) Google Scholar CrossRef Search ADS PubMed  Sun Q,  Rost TL,  Matthews MA.  Wound-induced vascular occlusions in Vitis vinifera (Vitaceae): tyloses in summer and gels in winter,  American Journal of Botany ,  2008, vol.  95 (pg.  1498- 1505) Google Scholar CrossRef Search ADS PubMed  Sun Q,  Rost TL,  Reid MS,  Matthews MA.  Ethylene and not embolism is required for wound-induced tylose development in stems of grapevines,  Plant Physiology ,  2007, vol.  145 (pg.  1629- 1636) Google Scholar CrossRef Search ADS PubMed  Trifilò P,  Lo Gullo MA,  Salleo S,  Callea K,  Nardini A.  Xylem embolism alleviated by ion-mediated increase in hydraulic conductivity of functional xylem: insights from field measurements,  Tree Physiology ,  2008, vol.  28 (pg.  1505- 1512) Google Scholar CrossRef Search ADS PubMed  Tyree MT,  Patiño S,  Bennink J,  Alexander J.  Dynamic measurements of root hydraulic conductance using a high-pressure flowmeter in the laboratory and field,  Journal of Experimental Botany ,  1995, vol.  46 (pg.  83- 94) Google Scholar CrossRef Search ADS   Tyree MT,  Salleo S,  Nardini A,  Lo Gullo MA,  Mosca R.  Refilling of embolized vessels in young stems of laurel. Do we need a new paradigm?,  Plant Physiology ,  1999, vol.  120 (pg.  11- 21) Google Scholar CrossRef Search ADS PubMed  Tyree MT,  Yang S.  Hydraulic conductivity recovery versus water pressure in xylem of Acer saccharum,  Plant Physiology ,  1992, vol.  100 (pg.  669- 676) Google Scholar CrossRef Search ADS PubMed  van Doorn WG,  Cruz P.  Evidence for a wounding-induced xylem occlusion in stems of cut chrysanthemum flowers,  Postharvest Biology and Technology ,  2000, vol.  19 (pg.  73- 83) Google Scholar CrossRef Search ADS   van Doorn WG,  Vaslier N.  Wounding-induced xylem occlusion in stems of cut chrysanthemum flowers: roles of peroxidase and cathechol oxidase,  Postharvest Biology and Technology ,  2002, vol.  26 (pg.  275- 284) Google Scholar CrossRef Search ADS   van Ieperen W.  Ion-mediated changes of xylem hydraulic resistance in planta: fact or fiction?,  Trends in Plant Science ,  2007, vol.  12 (pg.  137- 142) Google Scholar CrossRef Search ADS PubMed  van Ieperen W,  van Gelder A.  Ion-mediated flow changes suppressed by minimal calcium presence in xylem sap in Chrysanthemum and Prunus laurocerasus,  Journal of Experimental Botany ,  2006, vol.  57 (pg.  2743- 2750) Google Scholar CrossRef Search ADS PubMed  van Ieperen W,  van Meeteren U,  Nijsse J.  Embolism repair in cut flower stems: a physical approach,  Postharvest Biology and Technology ,  2002, vol.  25 (pg.  1- 14) Google Scholar CrossRef Search ADS   van Ieperen W,  van Meeteren U,  van Gelder H.  Fluid ionic composition influences hydraulic conductance of xylem conduits,  Journal of Experimental Botany ,  2000, vol.  51 (pg.  769- 776) Google Scholar CrossRef Search ADS PubMed  van Meeteren U.  Role of air embolism and low water temperature in water balance of cut chrysanthemum flowers,  Scientia Horticulturae ,  1992, vol.  51 (pg.  275- 284) Google Scholar CrossRef Search ADS   van Meeteren U,  Arevalo-Galarza L,  van Doorn WG.  Inhibition of water uptake after dry storage of cut flowers: role of aspired air and wound-induced processes in Chrysanthemum,  Postharvest Biology and Technology ,  2006, vol.  41 (pg.  70- 77) Google Scholar CrossRef Search ADS   van Meeteren U,  van Gelder H,  van Ieperen W.  Reconsideration of the use of deionized water as vase water in postharvest experiments on cut flowers,  Postharvest Biology and Technology ,  2000, vol.  18 (pg.  169- 181) Google Scholar CrossRef Search ADS   Vaslier N,  van Doorn WG.  Xylem occlusion in bouvardia flowers: evidence for a role of peroxidase and cathechol oxidase,  Postharvest Biology and Technology ,  2003, vol.  28 (pg.  231- 237) Google Scholar CrossRef Search ADS   Wehr JB,  Menzies NW,  Blamey FPC.  Inhibition of cell-wall autolysis and pectin degradation by cations,  Plant Physiology and Biochemistry ,  2004, vol.  42 (pg.  485- 492) Google Scholar CrossRef Search ADS PubMed  Willats WGT,  McCartney L,  Mackie W,  Knox JP.  Pectin: cell biology and prospects for functional analysis,  Plant Molecular Biology ,  2001, vol.  47 (pg.  9- 27) Google Scholar CrossRef Search ADS PubMed  Yang S,  Tyree MT.  A theoretical-model of hydraulic conductivity recovery from embolism with comparison to experimental data on Acer saccharum,  Plant, Cell and Environment ,  1992, vol.  15 (pg.  633- 643) Google Scholar CrossRef Search ADS   Zahradník J,  Fialová M,  Kastanek F,  Green KD,  Thomas NH.  The effect of electrolytes on bubble coalescence and gas holdup in bubble-column reactors,  Chemical Engineering Research and Design ,  1995, vol.  73 (pg.  341- 346) Zancani M,  Nagy G,  Vianello A,  Macri F.  Copper-inhibited NADH-dependent peroxidase-activity of purified soya bean plasma-membranes,  Phytochemistry ,  1995, vol.  40 (pg.  367- 371) Google Scholar CrossRef Search ADS   Zsivánovits G,  Marudova M,  Ring S.  Influence of mechanical properties of pectin films on charge density and charge density distribution in pectin macromolecule,  Colloid and Polymer Science ,  2005, vol.  284 (pg.  301- 308) Google Scholar CrossRef Search ADS   Zwieniecki MA,  Melcher PJ,  Holbrook NM.  Hydrogel control of xylem hydraulic resistance in plants,  Science ,  2001, vol.  291 (pg.  1059- 1062) Google Scholar CrossRef Search ADS PubMed  © 2010 The Author(s). This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.5), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)

Fincato, Paola;Moschou, Panagiotis N.;Spedaletti, Valentina;Tavazza, Raffaela;Angelini, Riccardo;Federico, Rodolfo;Roubelakis-Angelakis, Kalliopi A.;Tavladoraki, Paraskevi

doi: 10.1093/jxb/erq341pmid: 21081665

Abstract Polyamine oxidases (PAOs) are FAD-dependent enzymes involved in polyamine catabolism. All so far characterized PAOs from monocotyledonous plants, such as the apoplastic maize PAO, oxidize spermine (Spm) and spermidine (Spd) to produce 1,3-diaminopropane, H2O2, and an aminoaldehyde, and are thus considered to be involved in a terminal catabolic pathway. Mammalian PAOs oxidize Spm or Spd (and/or their acetyl derivatives) differently from monocotyledonous PAOs, producing Spd or putrescine, respectively, in addition to H2O2 and an aminoaldehyde, and are therefore involved in a polyamine back-conversion pathway. In Arabidopsis thaliana, five PAOs (AtPAO1–AtPAO5) are present with cytosolic or peroxisomal localization and three of them (the peroxisomal AtPAO2, AtPAO3, and AtPAO4) form a distinct PAO subfamily. Here, a comparative study of the catalytic properties of recombinant AtPAO1, AtPAO2, AtPAO3, and AtPAO4 is presented, which shows that all four enzymes strongly resemble their mammalian counterparts, being able to oxidize the common polyamines Spd and/or Spm through a polyamine back-conversion pathway. The existence of this pathway in Arabidopsis plants is also evidenced in vivo. These enzymes are also able to oxidize the naturally occurring uncommon polyamines norspermine and thermospermine, the latter being involved in important plant developmental processes. Furthermore, data herein reveal some important differences in substrate specificity among the various AtPAOs, which suggest functional diversity inside the AtPAO gene family. These results represent a new starting point for further understanding of the physiological role(s) of the polyamine catabolic pathways in plants. polyamines, polyamine catabolism, polyamine oxidase, spermine, spermidine, thermospermine Introduction The polyamines putrescine (Put), spermidine (Spd), and spermine (Spm) are low molecular weight organic cations that are found in a wide range of organisms from bacteria to plants and animals. In plants, polyamines are involved in various physiological processes, such as growth, development, and stress responses (Alcázar et al., 2010; Mattoo et al., 2010). Polyamine oxidases (PAOs, EC 1.5.3.11) are FAD-dependent enzymes involved in polyamine catabolism (Cona et al., 2006). They catalyse the oxidation of Spm, Spd, and/or their acetylated derivatives at the secondary amino groups (Wang et al., 2001; Wu et al., 2003; Cona et al., 2006). The products of the PAO-catalysed reactions depend on the enzyme source and reflect the mode of substrate oxidation. Animal PAOs and yeast Spm-oxidase (Fms1) oxidize N1-acetyl-Spm, N1-acetyl-Spd, and N1,N12-bis-acetyl-Spm at the carbon on the exo-side of N4-nitrogen to produce Spd, Put, and N1-acetyl-Spd, respectively, in addition to 3-acetamidopropanal and H2O2 (Landry and Sternglanz, 2003; Vujcic et al., 2003; Wu et al., 2003; Cona et al., 2006). In this catabolic pathway, polyamine acetylation is catalysed by the tightly regulated Spd/Spm N1-acetyltransferase (SSAT; EC 2.3.1.57), which is the rate-limiting enzyme of this pathway (Wallace et al., 2003). Similarly, animal Spm-oxidases (SMOs) and Fms1 oxidize Spm at the carbon on the exo-side of N4-nitrogen to produce Spd, 3-aminopropanal, and H2O2 (Wang et al., 2001; Vujcic et al., 2002; Cervelli et al., 2003; Landry and Sternglanz, 2003). Thus, animal PAOs and SMOs, as well as yeast Fms1, all are involved in a polyamine back-conversion pathway (Seiler, 2004). All PAOs from monocotyledonous plants characterized so far, such as Zea mays and Hordeum vulgare PAOs (ZmPAO and HvPAO, respectively), oxidize the carbon at the endo-side of the N4-nitrogen of Spd and Spm, producing 4-aminobutanal and N-(3-aminopropyl)-4-aminobutanal, respectively, in addition to 1,3-diaminopropane (Dap) and H2O2 (Cona et al., 2006), and are considered to be involved in terminal catabolism of polyamines. In contrast, in the dicotyledonous model plant Arabidopsis thaliana three PAOs (AtPAO1, AtPAO3, and AtPAO4) were recently shown to oxidize Spm and Spd in a similar mode to that of animal PAOs/SMOs (Tavladoraki et al., 2006; Moschou et al., 2008,a; Kamada-Nobusada et al., 2008), thus changing the current prevailing idea that plants and animals have distinct polyamine catabolic pathways. Here, a comparative study of the catalytic properties of AtPAOs towards common and uncommon polyamines is presented, revealing some important differences in substrate specificity and catalytic constants among the various AtPAOs. This study also evidences that all AtPAOs characterized thus far are involved in a polyamine back-conversion pathway. The existence in vivo of this pathway is shown not only in A. thaliana but also in Nicotiana tabacum plants, in which a PAO (NtPAO) with high sequence homology to AtPAO1 is present (Tavladoraki et al., 2006; Yoda et al., 2006). This study suggests functional diversification among the members of the AtPAO gene family and presents a new starting point for further understanding of the physiological role(s) of the polyamine catabolic pathways in plants. Materials and methods Materials Put, Spd, Spm, N1-acetyl-Spm, 4-aminoantipyrine, 3,5-dichloro-2-hydroxybenzenesulphonic acid, and horseradish peroxidase were purchased from Sigma-Aldrich-Fluka. Restriction and DNA-modifying enzymes were purchased from New England Biolabs, Invitrogen, Stratagene, and Promega. Other chemicals were obtained from Bio-Rad and J. T. Baker. All oligonucleotides were synthesized by Invitrogen and PRIMM. [14C]Spd {N-(3-aminopropyl)-[1,4-14C]tetramethylene-1,4-diamine} trihydrochloride (112 mCi mmol−1) and [14C]Spm {N',N'-bis-(3-aminopropyl)-[1,4-14C]tetramethylene-1,4-diamine} trihydrochloride (113 mCi mmol−1) were purchased from GE Healthcare. Synthetic thermospermine (Therm-Spm) was a kind gift from Professor Armin Geyer, Marburg. Sequence analysis and cDNA acquisition Expressed sequence tag (EST) and genomic database searches were performed using the Basic Local Alignment Search Tool (BLAST) (Altschul et al., 1990) and The Arabidopsis Information Resource (TAIR) database. Multiple sequence alignment of the amino acid sequences was done using the program CLUSTALW (Thompson et al., 1994). The cDNA encoding AtPAO2 (At2g43020; GenBank accession no. NM_129863) was obtained from the EST bank of RIKEN BioResource Center (Seki et al., 1998, 2002) (pda07387 clone) and the cDNA encoding AtPAO3 (At3g5950; GenBank accession no. NM_115767) was obtained from the ABRC stock center (http://www.arabidopsis.org) (clone no. U21196). The T-DNA insertional mutant of A. thaliana SALK_121288 for the AtPAO3 gene was obtained from the SALK Institute. DNA sequencing DNA sequencing was performed on double-stranded plasmid DNA using the automated fluorescent dye terminator technique (Applied Biosystems, model 3730). Plant material and growth conditions Plants of A. thaliana (ecotype Columbia-0) and N. tabacum (cv Petit Havana) were grown in a growth chamber at a temperature of 23 °C, a 16/8 h photoperiod (50 μmol m−2 s−1), and 55% relative humidity. For treatment of seedlings with abscisic acid (ABA), Spd, or guazatine, A. thaliana seeds were surface sterilized and sown in 24-well (10 seeds per well) tissue culture clusters, containing 1 ml per well of sterile Murashige and Skoog salts (MS salts; Duchefa; Murashige and Skoog, 1962) supplemented with 0.5% (w/v) sucrose, and grown with shaking in the growth chamber. Fresh medium, 500 μl per well, was added 9 d after sowing, and treatments were started 1 d later (10 d after sowing; Moschou et al., 2008,a). Prior to treatment initiation, the medium was removed and 1 ml of fresh medium containing the different compounds to be tested (10 μM ABA, 1 mM Spd, or 10 μM guazatine) was added. After treatment for 24 h, seedlings were washed with MS salts and homogenized with 5% (v/v) perchloric acid (PCA) to determine the content of total free polyamines as described by Moschou et al. (2008a). Preparation of constructs for AtPAO2, AtPAO4, and AtPAO5 expression in Escherichia coli and Pichia pastoris The AtPAO2 cDNA was initially obtained from RIKEN BioResource Center. However, sequencing of this cDNA revealed the presence of a non-conservative substitution (GAA to GGA) at position 863 of the AtPAO2 open reading frame or at position 288 of the protein (Fig. 1). Consequently, the correct whole coding region of AtPAO2 was amplified by reverse transcription-PCR (RT-PCR) from A. thaliana leaves using sequence-specific primers designed in such a way as to allow AtPAO2 cDNA cloning in the pET17b vector which guides cytoplasmic expression of recombinant proteins in Escherichia coli and to add a sequence coding for a 6-His tag at the 3' terminus of the AtPAO2 cDNA. AtPAO4 (At1g65840; NM_105256) and AtPAO5 (At4g29720; NM_119117) cDNAs were also isolated by RT-PCR from Arabidopsis leaves. Total RNA was isolated from leaves of A. thaliana plants using TRIZOL reagent (Invitrogen) according to the manufacturer's instructions. RNA samples were treated with RNase-free DNase I (Invitrogen) to avoid amplification from genomic DNA. The first cDNA strand was synthesized from total RNA using the SuperScript III first-strand synthesis system for RT-PCR (Invitrogen) and random primers. The cDNAs encoding AtPAO2, AtPAO4, and AtPAO5 were obtained from first-strand cDNA using gene-specific oligonucleotides. The AtPAO2-specific primers were AtPAO2-for1 (5′-GCATATGGAGTCCAGGAAAA ACTCTGATCG-3′) and AtPAO2-rev1 (5′-GGCGGCCGCCTAGTGGTGG TGGTGGTGGTGTCCTCCGAGACGAGATATAAGAAGAGGTACAGAGGC-3′); the AtPAO4-specific primers were AtPAO4-for1 (5′-GCCTCGCATATGGATAAGAAGAAGAATTCGTTTCCAG-3′) and AtPAO4-rev1 (5′-GGTACGCTCGAGCTAGTGGTGGTGGTGGTGGTGTCCTCCCATCCTGGAGATTTGGAGAGGCACAG-3′); and the AtPAO5-specific primers were AtPAO5-for1 (5′-GGCTCCCATATGGCGAAGAAAGCAAGAATTGTTATAATCG-3′) and AtPAO5-rev1 (5′-GTGTACTGCGGCCGCCTAGTGGTGGTGGTGGTGGTGTCCTCCAAAATTACATTTGTAATGCTTGAGAAG-3′). The underlined regions in AtPAO2-for1, AtPAO4-for1, and AtPAO5-for1 oligonucleotides indicate NdeI restriction sites, in AtPAO2-rev1 and AtPAO5-rev1 oligonucleotides they indicate NotI restriction sites, and in the AtPAO4-rev1 oligonucleotide they indicate an XhoI restriction site. The AtPAO2-rev1, AtPAO4-rev1, and AtPAO5-rev1 primers were designed to insert the coding sequence for two glycine residues followed by a 6-His tag prior to the stop codon of the corresponding cDNA. PCR amplification was carried out with the Pfu Turbo® DNA polymerase (Stratagene) in an iCycler thermal cycler (Bio-Rad) with the following parameters: 2 min of denaturation at 95 °C, 30 cycles of 95 °C for 1 min, 58 °C for 1 min and 72 °C for 2 min; and 10 min at 72 °C for final extension. PCR products were purified using the QIAquick gel extraction kit (Qiagen), ligated into the pDrive vector (Qiagen) or pGEMT-easy vector (Promega), completely sequenced, and then cloned in the pET17b plasmid (Novagen) for bacterial expression, yielding the AtPAO2-pET17b, AtPAO4-pET17b, and AtPAO5-pET17b constructs. These plasmids were then used to transform E. coli BL21 (DE3) cells. The AtPAO5 cDNA was also amplified from the AtPAO5-pET17b construct with AtPAO5-for2 (5′-GCCTCGGAATTCGCGAAGAAAGCAAGAATTGTTATAATC-3′) and AtPAO5-rev1 oligonucleotides, the underlined region in the AtPAO5-for2 oligonucleotide indicating an EcoRI restriction site. This amplification product was cloned between EcoRI and NotI restriction sites in the pGAPZαA vector (Invitrogen) for heterologous expression in the culture medium of Pichia pastoris, thus giving the AtPAO5-pGAP construct which was used to transform P. pastoris strain X-33 according to the Pichia EasyComp™ kit (Invitrogen). Fig. 1. View largeDownload slide Alignment of the amino acid sequence of AtPAOs and ZmPAO. Multialignment was accomplished by using the program CLUSTALW sequence alignment. Numbering of amino acid residues is shown on the right. In ZmPAO, numbering starts from the first amino acid of the mature protein. The signal peptide of ZmPAO is underlined. Identical residues to those of AtPAO2 are indicated by grey boxes. Red boxes indicate residues constituting the ZmPAO catalytic site and the corresponding residues conserved in the various AtPAOs. AtPAO residues which are different from the corresponding residues in the ZmPAO catalytic site and which are conserved (yellow boxes) or non-conserved (green boxes) among a number of the AtPAOs are indicated. The position which in human MAO-A and MAO-B is involved in covalent binding to the isoalloxazine ring of the FAD through a cysteine residue is indicated by the symbol #. Green lines above the sequences indicate FAD-interacting regions in MAO-B and ZmPAO (Binda et al., 1999, 2002; Wu et al., 2003). Peroxisomal targeting signals of AtPAO2, AtPAO3, and AtPAO4 are indicated in red. Numbers in parentheses indicate the percentage amino acid sequence identity of the various proteins with respect to AtPAO2. Fig. 1. View largeDownload slide Alignment of the amino acid sequence of AtPAOs and ZmPAO. Multialignment was accomplished by using the program CLUSTALW sequence alignment. Numbering of amino acid residues is shown on the right. In ZmPAO, numbering starts from the first amino acid of the mature protein. The signal peptide of ZmPAO is underlined. Identical residues to those of AtPAO2 are indicated by grey boxes. Red boxes indicate residues constituting the ZmPAO catalytic site and the corresponding residues conserved in the various AtPAOs. AtPAO residues which are different from the corresponding residues in the ZmPAO catalytic site and which are conserved (yellow boxes) or non-conserved (green boxes) among a number of the AtPAOs are indicated. The position which in human MAO-A and MAO-B is involved in covalent binding to the isoalloxazine ring of the FAD through a cysteine residue is indicated by the symbol #. Green lines above the sequences indicate FAD-interacting regions in MAO-B and ZmPAO (Binda et al., 1999, 2002; Wu et al., 2003). Peroxisomal targeting signals of AtPAO2, AtPAO3, and AtPAO4 are indicated in red. Numbers in parentheses indicate the percentage amino acid sequence identity of the various proteins with respect to AtPAO2. Purification of recombinant AtPAOs from bacterial extracts Expression of recombinant AtPAO2 and AtPAO4 was induced by 0.4 mM isopropyl-β-d-thiogalactopyranoside at 25 °C for 5 h. Cells expressing recombinant AtPAO2 or AtPAO4 were resuspended in 100 mM TRIS-HCl, pH 8.0, 0.5 M NaCl, 30% (v/v) glycerol, 1 mM phenylmethylsulphonyl fluoride (PMSF) (extraction buffer I) or 100 mM TRIS-HCl, pH 8.0, 30% (v/v) glycerol, 1 mM PMSF (extraction buffer II), respectively, and disrupted by sonication. After centrifugation at 13 000 g for 30 min at 4 °C, the clear supernatant (bacterial extract) containing the soluble proteins was either analysed for recombinant protein accumulation by immunoblotting or further processed for recombinant protein purification. Recombinant AtPAO2 and AtPAO4 were purified from bacterial extracts by affinity chromatography. In detail, bacterial extracts were applied to an Ni2+-charged resin (Amersham Biosciences) equilibrated in extraction buffer. The resin was washed first with extraction buffer and then with 100 mM TRIS-HCl, pH 8.0, 30% (v/v) glycerol. The recombinant proteins were eluted with 100 mM TRIS-HCl, pH 8.0, 30% (v/v) glycerol, 300 mM imidazole and immediately dialysed against 100 mM TRIS-HCl, pH 8.0, 30% (v/v) glycerol. This expression/purification protocol resulted in the production of high amounts of recombinant AtPAO2 and AtPAO4 (∼2 mg l−1 of culture) and permitted purification of the enzymes to almost electrophoretic homogeneity. In this protocol, the constant presence of 30% glycerol in all buffers was of particular importance to avoid aggregation of recombinant AtPAO2 and to confer stability to recombinant AtPAO4. Indeed, in the absence of glycerol, AtPAO2 formed enzymatically active aggregates and AtPAO4 was rapidly inactivated under high ionic strength conditions. Expression and purification of AtPAO1 and AtPAO3 recombinant proteins was performed as described by Tavladoraki et al. (2006) and Moschou et al. (2008a), respectively. The purity of the recombinant enzymes was evaluated by SDS–PAGE analysis and by the A280/A458 ratio which for preparations of homogeneity >97% was ∼10. A typical SDS–PAGE analysis of the purified AtPAO proteins is shown in Supplementary Fig. S1 available at JXB online. Production of transgenic plants To obtain 35S::AtPAO3 and 35S::GFP-AtPAO3 Arabidopsis transgenic plants, the AtPAO3 cDNA was amplified using the gene-specific oligonucleotides AtPAO3-for (5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTGATGGAGTCCGGAGGCAAC-3′) and AtPAO3-rev (5′-GGGGACCACTTTGTACAAGAAAGCTGGGTATTACATACGGGAGATCAGAAG-3′) and cloned initially into the pDONR207 vector (Invitrogen) and then into the pGWB2 and pGWB6 vectors through the GATEWAY recombination system (Invitrogen). Constructs were used to transform A. thaliana Col-0 wild-type plants by the Agrobacterium tumefaciens (strain C58C1)-mediated floral dip transformation method as described by Clough and Bent (1998). Recombinant AtPAO3 expression in 35S::AtPAO3 and 35S::GFP-AtPAO3 transgenic plants was determined by northern blot and western blot analysis using in the latter case a rabbit anti-AtPAO3 antibody raised (Davids Biotechnologie, Germany) against AtPAO3 protein in fusion with the maltose-binding protein (MBP). The antibody was purified by immune-affinity absorbance using the MBP–AtPAO3 protein as absorbant. For polyamine back-conversion assay, 2-week-old homozygous T3 transgenic plants grown in a 24-well plate were used. Polyamine levels were determined by HPLC analysis as described previously (Moschou et al., 2008a). Characterization of a T-DNA insertional mutant for AtPAO3 Homozygous lines for the atpao3 mutant (SALK_121288) were selected by PCR using the gene-specific primers AtPAO3-LP1 (5′-TTGAGCACATTCTGGAAGAGGTG) and AtPAO3-RP1 (5′-ACACTTTGCCGAGATGGTTTCAG). The lack of AtPAO3-specific mRNA in the homozygous mutant line was confirmed by RNA gel blot analysis performed as described by Moschou et al. (2008a). Polyamine levels were determined in whole seedlings. Determination of AtPAO catalytic parameters The catalytic parameters (Km and kcat) for the oxidation of Spm, Spd, N1-acetyl-Spm, Therm-Spm, and norspermine (Nor-Spm) by E. coli-expressed recombinant AtPAO2, AtPAO3, and AtPAO4 were determined using purified proteins and following spectrophotometrically the formation of a pink adduct (ε515=2.6×104 M−1 cm−1), as a result of oxidation of 4-aminoantipyrine and 3,5-dichloro-2-hydroxybenzenesulphonic acid catalysed by horseradish peroxidase in 100 mM TRIS-HCl buffer pH 6.5–8.5, at 25 °C (Tavladoraki et al., 2006). Km values were determined from Michaelis–Menten plots, and non-linear least-squares fitting of data was performed using Graphpad Prism software. Analysis of AtPAO2 and AtPAO4 reaction products Analysis of the reaction products of Spm and Spd oxidation by recombinant AtPAO2 and AtPAO4 was performed as described by Tavladoraki et al. (2006). More specifically, a reaction mixture containing purified recombinant AtPAO2 (at a final concentration of 1 nM) or AtPAO4 (at a final concentration of 5 nM) and 0.2 mM Spm or Spd in 100 mM TRIS-HCl, pH 7.5 was prepared and aliquots were analysed for polyamine content at various time intervals after addition of an equal volume of 5% (v/v) PCA containing 0.12 mM 1,6-diaminohexane as internal standard. Polyamines were quantified after derivatization with dansyl chloride according to Smith and Davies (1985) with minor modifications, and dansylated polyamines were separated by HPLC (Spectra SystemP 2000; ThermoFinnigan) on a reverse-phase C18 column (Spherisorb S5 ODS2, 5 μm particle diameter, 4.6×250 mm) using a discontinued solution A [acetonitrile, methanol, and water in a ratio 3:2:5 (v/v/v)] to solution B [acetonitrile and methanol in a ratio 3:2 (v/v)] gradient (78% solution A for 5 min, 78–36% solution A in 42 min, 36–20% in 3 min, 20–10% in 10 min, 10–78% in 10 min at a flow rate of 1.0 ml min−1). Eluted peaks were detected by a spectrofluorometer (Spectra System FL 3000; excitation 365 nm, emission 510 nm), recorded, and integrated using Thermo Finnigan Chrom-Card 32-bit software. Determination of a polyamine back-conversion pathway in A. thaliana and N. tabacum protoplasts Protoplasts were prepared from A. thaliana and N. tabacum leaves as described by Molinari et al. (1998), resuspended in K3 medium (B5 medium including vitamins, 136.92 g l−1 sucrose, 25 mg l−1 xylose, 250 mg l−1 NH4NO3, 750 mg l−1 CaCl2·2H2O, 63.3 mg l−1 CaHPO4·2H2O, 25.35 mg l−1 NaH2PO4·2H2O, 1 mg l−1 naphthaleneacetic acid (NAA), 0.2 mg l−1 6-benzoamino purine (6-BAP), 0.1 mg l−1 2,4-D, pH 5.6) at a final concentration of 1.5×106 protoplasts ml−1 and incubated with 7 μM [14C]Spd (112 mCi mmol−1) or [14C]Spm (113 mCi mmol−1). For inhibition experiments, protoplasts were simultaneously incubated with 200 μM 2-bromoethylamine [a copper amine oxidase (CuAO)-specific inhibitor] or 50 μM guazatine (a PAO-specific inhibitor). Aliquots of protoplasts (1.5×106) were removed at various time intervals, washed with washing buffer (9 g l−1 NaCl, 18.4 g l−1 CaCl2, 0.4 g l−1 KCl, 1.0 g l−1 glucose) and homogenized with 5% (v/v) PCA containing 0.12 mM 1,6-diaminohexane. As a control, recombinant AtPAO2, AtPAO4, and ZmPAO purified from E. coli (AtPAO2, AtPAO4) or P. pastoris (ZmPAO) (Polticelli et al., 2005) were incubated with [14C]Spd or [14C]Spm. Polyamines were derivatized by dansyl chloride, analysed by thin-layer chromatography (TLC) on silica gel TLC plates (AL SIL G, Whatman) in a solvent system of chloroform/triethylamine (25:2, v/v) and visualized under UV light. To detect radioactivity, TLC plates were exposed to radiographic film for 2–4 d. Results Characterization of the PAO gene family in A. thaliana In A. thaliana, five putative PAO genes [AtPAO1 (At5g13700), AtPAO2 (At2g43020), AtPAO3 (At3g59050), AtPAO4 (At1g65840), and AtPAO5 (At4g29720)] are present, showing varying sequence homology with ZmPAO (Tavladoraki et al., 2006). AtPAO2, AtPAO3, and AtPAO4 display low amino acid sequence identity (23%) with ZmPAO, AtPAO1, and AtPAO5, but a high sequence identity with respect to each other (Fig. 1). In particular, the amino acid sequence identity between AtPAO2 and AtPAO3 is 85%, while it is 58% between AtPAO2 and AtPAO4 and 50% between AtPAO3 and AtPAO4. Furthermore, AtPAO2, AtPAO3, and AtPAO4 genes have the same intron/exon organization, all containing eight introns with highly conserved positions (Fig. 2). This, together with the elevated sequence homology to each other, suggests that these three Arabidopsis genes are recent derivatives of a common ancestor, thus forming a distinct PAO subfamily (AtPAO2–AtPAO4 subfamily). In contrast, the intron/exon organization of AtPAO2–AtPAO4 genes is different from that of AtPAO1 (Fig. 2) and ZmPAO, the latter two sharing a very similar gene organization (Cervelli et al., 2001). AtPAO5 has low sequence identity with ZmPAO and the other AtPAOs (21–23% amino acid identity), but it exhibits good amino acid sequence identity (31%) with mouse PAO and SMO. Furthermore, the AtPAO5 gene presents a very different gene organization from that of the other AtPAO genes and ZmPAO. Indeed, AtPAO5 lacks introns (Fig. 2) and is the first PAO gene characterized so far with such organization. Fig. 2. View largeDownload slide Schematic representation of the exon/intron organization of AtPAO genes. Introns are represented by lines and exons by boxes. Exons are numbered in Roman numerals. Open and filled boxes indicate shared and unshared exons among the various AtPAO genes, respectively. Stripes and stipples show shared exon domains which are found either joined to each other or separated by the presence of an intron, according to the specific gene considered. Exons and introns are drawn in scale. Fig. 2. View largeDownload slide Schematic representation of the exon/intron organization of AtPAO genes. Introns are represented by lines and exons by boxes. Exons are numbered in Roman numerals. Open and filled boxes indicate shared and unshared exons among the various AtPAO genes, respectively. Stripes and stipples show shared exon domains which are found either joined to each other or separated by the presence of an intron, according to the specific gene considered. Exons and introns are drawn in scale. To determine the subcellular localization of the various AtPAOs, their amino acid sequences were analysed by PSORT (www.psort.org). This analysis predicted the presence of a sequence for peroxisomal targeting in AtPAO2, AtPAO3, and AtPAO4 (Fig. 1), and indeed peroxisomal targeting of all three enzymes has been recently demonstrated (Kamada-Nobusada et al., 2008; Moschou et al., 2008,a). For AtPAO5, PSORT analysis and sequence alignment with all the plant PAOs so far sequenced revealed the absence of any signal peptide for protein targeting to a specific extracellular or intracellular compartment, suggesting cytosolic localization, as has been recently proposed for AtPAO1 (Tavladoraki et al., 2006). Heterologous expression of AtPAO2 and AtPAO4 in E. coli and purification of recombinant proteins To perform a comparative study of the catalytic properties of all Arabidopsis PAOs, heterologous expression was performed in E. coli. Similarly to AtPAO1 and AtPAO3 (Tavladoraki et al., 2006; Moschou et al., 2008,a), recombinant AtPAO2 and AtPAO4 proteins had an apparent molecular weight of ∼54 500 Da (Supplementary Fig. S1 at JXB online), which is similar to that predicted from the amino acid sequence, and displayed the characteristic UV-visible spectrum of the oxidized flavoproteins with three absorbance peaks at 280, 380, and 460 nm (data not shown). Precipitation of purified AtPAO2 and AtPAO4 with trichloroacetic acid led to cofactor release into the supernatant, suggesting a non-covalent linkage to both proteins. This is in agreement with the presence of a serine residue in position 405 in AtPAO2 (Fig. 1), a position which in human monoamine oxidase (MAO)-A and MAO-B is involved in covalent binding to the isoalloxazine ring of the FAD through a cysteine residue (Edmondson et al., 2004). Indeed, while cysteine, histidine, and tyrosine residues were shown to be involved in covalent binding to the flavin ring in some flavoenzymes (Edmondson and Newton-Vinson, 2001; Heuts et al., 2009), such a covalent linkage has not been observed until now for serine residues. The non-convalent binding of the FAD molecule also in recombinant AtPAO4 was somehow unexpected since a cysteine residue is present at position 407 (position 405 in AtPAO2; Fig. 1). This probably reflects that for covalent flavinylation some other amino acid residues play a crucial role in activating the process (Fraaije et al., 2000). As far as AtPAO5 is concerned, all efforts to express this gene in the BL21(DE3) strain of E. coli were unsuccessful. In particular, induction of recombinant protein was performed at various temperatures (25, 30, and 37 °C) and for various time periods (1, 3, 5, and 24 h), and catalytic activity was tested at various pH values (pH 6.0–8.0). Recombinant AtPAO5 expression was also attempted in two other E. coli strains, one [BL21(DE3)pLysS] strain permitting more stringent expression conditions and thus being useful for the expression of insoluble or toxic proteins, and the other [BL21-CodonPlus(DE3)-RIPL] strain being characterized by an enriched tRNA pool. However, in all cases AtPAO5 heterologous expression was unsuccessful. Similarly, expression of AtPAO5 in P. pastoris through the pGAPZαA vector, which guides constitutive expression of recombinant proteins in the culture medium, and in plants using an A. tumefaciens-based transient expression system was unsuccessful. Furthermore, analysis of Arabidopsis plants stably transformed with a construct for AtPAO5 overexpression (35SCaMV::AtPAO5-6His/pK2GW7) by Western blot using an anti-6-His antibody and enzyme activity assay using polyamines as substrates demonstrated that, despite the presence of the corresponding RNA, recombinant AtPAO5 protein did not accumulate to detectable amounts. Biochemical characterization of recombinant AtPAOs Analysis of the catalytic constants of the purified recombinant AtPAO1 and AtPAO3 (Table 1) confirmed that AtPAO1 oxidizes only Spm and not Spd (Tavladoraki et al., 2006) and that Spd is a better substrate than Spm for AtPAO3 (Moschou et al., 2008,a). Furthermore, analysis of the catalytic constants of the purified recombinant AtPAO2 and AtPAO4 indicated that these enzymes oxidize both Spm and Spd (Table 1) but not Put. In particular, AtPAO2 exhibits similar kcat values for the two substrates, whereas AtPAO4 has a kcat value for Spm which is ∼10-fold higher than that for Spd. These data differ in part from recently published data showing activity of recombinant AtPAO4 only towards Spm and not Spd (Kamada-Nobusada et al., 2008; Takahashi et al., 2010). This discrepancy could be due to differences in experimental conditions, such as protein concentration, ionic strength, and pH values during the purification procedure and the enzyme activity assays, conditions that greatly affect AtPAO4 catalytic activity. Indeed, a rapid reduction in AtPAO4 catalytic activity was observed in the presence of a high salt concentration. Furthermore, the AtPAO4 catalytic activity towards Spm displays a pH dependence which is different from that towards Spd. More specifically, the optimum pH for AtPAO4 catalytic activity towards Spd is 8.0, whereas that towards Spm is 7.0 (see below). On the other hand, this apparent discrepancy is probably of low physiological significance, since for AtPAO4 much lower catalytic activity with Spd than with Spm was found in the present study. Table 1. Kinetic constants of substrate oxidation by recombinant AtPAOs   AtPAO1   AtPAO2   AtPAO3   AtPAO4     kcat (s−1)  Km (μM)  kcat (s−1)  Km (μM)  kcat (s−1)  Km (μM)  kcat (s−1)  Km (μM)  Spm  2.5±0.4  120±20  4.2±1.2  270±30  1.7±0.5  580±40  4.6±1.0  47±5  Spd  0  0  4.6±1.5  409±40  3.4±1.4  274±50  0.1±0.03  139±18  N1-acetyl-Spm  0.2±0.4  470±20  0.8±0.2  233±20  0.02±0.01  2000±200  0.014±0.004  ND  Nor-Spm  6.9±1.3  90±10  2.9±0.8  ND  1.1±0.2  45±12  0.45±0.1  ND  Therm-Spm  5.7±1.1  20±3.0  0.4±0.1  ND  0.5±0.1  50±10  0.1±0.04  ND    AtPAO1   AtPAO2   AtPAO3   AtPAO4     kcat (s−1)  Km (μM)  kcat (s−1)  Km (μM)  kcat (s−1)  Km (μM)  kcat (s−1)  Km (μM)  Spm  2.5±0.4  120±20  4.2±1.2  270±30  1.7±0.5  580±40  4.6±1.0  47±5  Spd  0  0  4.6±1.5  409±40  3.4±1.4  274±50  0.1±0.03  139±18  N1-acetyl-Spm  0.2±0.4  470±20  0.8±0.2  233±20  0.02±0.01  2000±200  0.014±0.004  ND  Nor-Spm  6.9±1.3  90±10  2.9±0.8  ND  1.1±0.2  45±12  0.45±0.1  ND  Therm-Spm  5.7±1.1  20±3.0  0.4±0.1  ND  0.5±0.1  50±10  0.1±0.04  ND  The enzymatic activity of recombinant enzymes was determined in 100 mM TRIS-HCl at the optimum pH (pH 8.0 for AtPAO1 and pH 7.5 for AtPAO2, AtPAO3, and AtPAO4). Data are the mean ±SE of at least three independent experiments. ND, not determined. View Large The substrate specificity of AtPAO2 and AtPAO4 is different from that of AtPAO1, which oxidizes only Spm, but similar to that of ZmPAO and AtPAO3, which are active towards both Spd and Spm with a kcat (Spd)/kcat (Spm) ratio of 1.5 and 2.0, respectively (Polticelli et al., 2005; Table 1). Furthermore, the kcat values for Spm of recombinant AtPAO1, AtPAO2, AtPAO3, and AtPAO4 (ranging from 1.7 s−1 to 4.6 s−1) and for Spd of recombinant AtPAO3 (3.4 s−1) are similar to each other and ∼10-fold lower than that of recombinant ZmPAO for Spm and Spd (32.9 s−1 and 50.2 s−1, respectively) (Table 1; Polticelli et al., 2005). Similarly, the Km values for Spm of recombinant AtPAO1, AtPAO2, AtPAO3, and AtPAO4 (47–580 μM) and for Spd of recombinant AtPAO3 (274 μM) are ∼30- to 360-fold higher than that of recombinant ZmPAO (1.6 μM) for the same substrates (Polticelli et al., 2005). The data suggest that all four recombinant AtPAOs, having kcat/Km values for Spm and Spd in the range of 15.4–97.9 mM−1 s−1, are as efficient as murine SMO (kcat/Km=50 mM−1 s−1; Cervelli et al., 2003) and less efficient than recombinant ZmPAO (kcat/Km values for Spm and Spd of 20.6×103 mM−1 s−1 and 23.9×103 mM−1 s−1, respectively) and murine PAO (with a kcat/Km value of 2.5×103 mM−1 s−1, though with a kcat value of 4.5 s−1, for N1-acetyl-Spm). Recombinant AtPAO2, AtPAO3, and AtPAO4 are able to oxidize N1-acetyl-Spm, but less efficiently than Spm (Table 1). This suggests that acetylated polyamines are probably not the physiological substrates of these enzymes despite the fact that they have a peroxisomal localization (Kamada-Nobusada et al., 2008; Moschou et al., 2008,a) as do the animal PAOs, which prevalently oxidize the acetylated polyamines (Wallace et al., 2003). Recently, it was shown that recombinant AtPAO1 oxidizes the uncommon polyamine Nor-Spm more efficiently than Spm (Tavladoraki et al., 2006). In the present study, it is shown that recombinant AtPAO2, AtPAO3, and AtPAO4 are also able to oxidize Nor-Spm (Table 1), even though with lower kcat values (1.5-, 3-, and 10-fold, respectively) when compared with those for Spm (in the case of AtPAO2 and AtPAO4) and for Spd (in the case of AtPAO3). Recombinant AtPAOs were also analysed for their ability to oxidize the polyamine Therm-Spm, an isomer of Spm, which was recently detected in A. thaliana plants (Kakehi et al., 2008). Furthermore, ACL5 from A. thaliana, which was initially described as a Spm synthase (Hanzawa et al., 1997), was recently shown to catalyse the synthesis of Therm-Spm from Spd (Knott et al., 2007). Interestingly, exogenous application of Therm-Spm partially rescued the severe dwarf phenotype of the acl5 loss-of-function A. thaliana mutant (Kakehi et al., 2008), suggesting a role for Therm-Spm in important developmental processes, such as vascular development (Clay and Nelson, 2005; Muñiz et al., 2008; Vera-Sirera et al., 2010). The data herein show that recombinant AtPAO1 is able to oxidize Therm-Spm with a kcat value and a kcat/Km ratio which are 2- and 10-fold higher, respectively, than those for Spm (Table 1), suggesting that Therm-Spm is a better substrate than Spm for AtPAO1 and it may be the in vivo physiological substrate of this enzyme. Recombinant AtPAO2, AtPAO3, and AtPAO4 are also able to oxidize Therm-Spm, but with kcat values 7- to 50-fold lower than those for their best substrate (Table 1). On the other hand, recombinant ZmPAO expressed in P. pastoris is able to oxidize Therm-Spm with a kcat value of 1.1 s−1, which is ∼30-fold lower than that for Spm, and recombinant murine SMO expressed in E. coli is not active with Therm-Spm. The pH dependence of the purified recombinant AtPAO2 and AtPAO4 catalytic activity was also examined using Spm and Spd as substrates (data not shown). The results show an increase in the catalytic activity of both enzymes and for both substrates with increasing pH, reaching a maximum between pH 7.0 and pH 8.0. More specifically, the optimum pH for AtPAO2 catalytic activity towards both Spd and Spm is 7.5, while that for AtPAO4 activity towards Spd is 8.0 and towards Spm 7.0. At higher pH values, the catalytic activity of the recombinant enzymes decreases. Characterization of AtPAO2 and AtPAO4 reaction products AtPAO1, AtPAO3, and AtPAO4 have recently been shown to oxidize Spm mainly through a polyamine back-conversion pathway producing Spd from Spm and Put from Spd (Tavladoraki et al., 2006; Kamada-Nobusada et al., 2008; Moschou et al., 2008a). Only in the case of recombinant AtPAO1 was a very small amount of Dap produced during Spm oxidation in parallel to the production of Spd, suggesting that this enzyme is also able to oxidize Spm through a terminal catabolic pathway, though to a much less extent than it does through the polyamine back-conversion catabolic pathway. In the present study, analysis of the AtPAO2 and AtPAO4 reaction products by HPLC demonstrated that AtPAO2 and AtPAO4 oxidize Spm and Spd exclusively through a polyamine back-conversion pathway. Indeed, Spd oxidation by either recombinant AtPAO2 or AtPAO4 resulted in the production of increasing amounts of Put in parallel with a decrease in the amount of Spd (data not shown). Formation of product with the retention time of Dap was not observed. Similarly, analysis of AtPAO2 and AtPAO4 reaction products using Spm as substrate showed the production of Put and Spd in parallel with a decrease in Spm levels (Fig. 3). Also in this case, formation of Dap was not observed. It should be noticed that in the case of AtPAO4 the amount of Put produced was very low (Fig. 3), which correlates well with the very low kcat value of this protein for Spd. Fig. 3. View largeDownload slide Time-course of Spm oxidation and reaction product accumulation by recombinant AtPAO2 and AtPAO4 proteins. Purified recombinant proteins (1 nM AtPAO2 and 5 nM AtPAO4) were incubated with 0.2 mM Spm. Aliquots were analysed for polyamine levels at various time intervals. Data are from a single representative experiment, which was repeated twice. Fig. 3. View largeDownload slide Time-course of Spm oxidation and reaction product accumulation by recombinant AtPAO2 and AtPAO4 proteins. Purified recombinant proteins (1 nM AtPAO2 and 5 nM AtPAO4) were incubated with 0.2 mM Spm. Aliquots were analysed for polyamine levels at various time intervals. Data are from a single representative experiment, which was repeated twice. Polyamine back-conversion pathway in vivo The existence of a polyamine back-conversion pathway in A. thaliana was also demonstrated in vivo following incubation of A. thaliana protoplasts with Spm or Spd radiolabelled at internal carbon atoms. In the case of a polyamine back-conversion pathway, radiolabelled Spm and Spd should be converted to radiolabelled Spd and Put, respectively (Fig. 4), as was found in vitro using purified recombinant AtPAO2 and AtPAO4 (Fig. 5). In contrast, oxidation of radiolabelled Spd or Spm by recombinant ZmPAO resulted in the production of radiolabelled 4-aminobutanal and N-(3-aminopropyl)-4-aminobutanal, respectively (Fig. 5), as expected by a terminal catabolic pathway (Fig. 4). The Dap produced is not radiolabelled and thus not observable by radiography (Figs 4, 5). Fig. 4. View largeDownload slide Schematic representation of [14C]Spd and [14C]Spm oxidation through a polyamine back-conversion pathway and a terminal catabolic pathway of polyamines. Asterisks indicate 14C atoms. Arrowheads B or T indicate the Spm or Spd carbon atoms oxidized in a polyamine back-conversion pathway or the terminal catabolic pathway of the polyamines, respectively. Fig. 4. View largeDownload slide Schematic representation of [14C]Spd and [14C]Spm oxidation through a polyamine back-conversion pathway and a terminal catabolic pathway of polyamines. Asterisks indicate 14C atoms. Arrowheads B or T indicate the Spm or Spd carbon atoms oxidized in a polyamine back-conversion pathway or the terminal catabolic pathway of the polyamines, respectively. Fig. 5. View largeDownload slide Polyamine back-conversion pathway in A. thaliana protoplasts. Aliquots of protoplasts were incubated with [14C]Spd (A) or [14C]Spm (B) in the presence or absence of the CuAO-specific inhibitor 2-bromoethylamine or the PAO-specific inhibitor guazatine, and were analysed at various time intervals for total free polyamine content by TLC after dansyl chloride derivatization. TLC plates were exposed to radiographic film to detect radioactivity. As a control, the AtPAO2, AtPAO4, and ZmPAO reaction products with radiolabelled [14C]Spm or [14C]Spd are also shown using recombinant proteins purified from E. coli (AtPAO2 and AtPAO4) or P. pastoris (ZmPAO). Arrowheads in ZmPAO lanes indicate the aminoaldehydes 4-aminobutanal and N-(3-aminopropyl)-4-aminobutanal produced during ZmPAO reaction with Spd and Spm, respectively. Dap, 1,3-diaminopropane. Graphs represent densitometric analysis. Fig. 5. View largeDownload slide Polyamine back-conversion pathway in A. thaliana protoplasts. Aliquots of protoplasts were incubated with [14C]Spd (A) or [14C]Spm (B) in the presence or absence of the CuAO-specific inhibitor 2-bromoethylamine or the PAO-specific inhibitor guazatine, and were analysed at various time intervals for total free polyamine content by TLC after dansyl chloride derivatization. TLC plates were exposed to radiographic film to detect radioactivity. As a control, the AtPAO2, AtPAO4, and ZmPAO reaction products with radiolabelled [14C]Spm or [14C]Spd are also shown using recombinant proteins purified from E. coli (AtPAO2 and AtPAO4) or P. pastoris (ZmPAO). Arrowheads in ZmPAO lanes indicate the aminoaldehydes 4-aminobutanal and N-(3-aminopropyl)-4-aminobutanal produced during ZmPAO reaction with Spd and Spm, respectively. Dap, 1,3-diaminopropane. Graphs represent densitometric analysis. During incubation of A. thaliana protoplasts with radiolabelled Spd, the production of increasing amount of radiolabelled Put was observed (Fig. 5A). This production was inhibited by guazatine, a PAO-specific inhibitor, but not by 2-bromoethylamine, a CuAO-specific inhibitor (Cona et al., 2006). Similarly, during incubation of Arabidopsis protoplasts with radiolabelled Spm, an increasing amount of radiolabelled Spd and Put was observed (Fig. 5B). Very similar results were also obtained using protoplasts (see Supplementary Fig. S2 at JXB online) and protein extracts from N. tabacum plants (data not shown), in which a PAO gene (NtPAO) with very high sequence similarity (74% amino acid identity) to AtPAO1 is present (Tavladoraki et al., 2006; Yoda et al., 2006). The presence of a polyamine back-conversion pathway was also shown in whole A. thaliana plants. Indeed, exogenously supply of Spd to wild-type A. thaliana plants resulted in a small, but statistically significant, increase in Put levels (Fig. 6). Interestingly, this increase was much more pronounced in the presence of the plant hormone ABA, which was previously shown to induce the peroxisomal polyamine catabolic pathway in A. thaliana (Moschou et al., 2008a). In contrast, the ABA-induced increase in Put levels was not observed when Spd and ABA were added in abi1-1 Arabidopsis mutants defective in the ABA signal transduction pathway (Fig. 6). Fig. 6. View largeDownload slide Polyamine back-conversion pathway in A. thaliana whole plants. Put levels in Arabidopsis wild-type (WT) plants and mutants (abi1-1) defective in the ABA signal transduction pathway are shown. Polyamine content was determined in untreated control plants and in plants treated with 10 μM ABA and/or with 1 mM Spd for 24 h. Data are from a single representative experiment, which was repeated twice. Each point represents the mean value from three independent analyses of polyamine levels and the bars indicate the SE. Asterisk indicates values significantly different from those of the WT untreated plants by one-way ANOVA test (P <0.05). Fig. 6. View largeDownload slide Polyamine back-conversion pathway in A. thaliana whole plants. Put levels in Arabidopsis wild-type (WT) plants and mutants (abi1-1) defective in the ABA signal transduction pathway are shown. Polyamine content was determined in untreated control plants and in plants treated with 10 μM ABA and/or with 1 mM Spd for 24 h. Data are from a single representative experiment, which was repeated twice. Each point represents the mean value from three independent analyses of polyamine levels and the bars indicate the SE. Asterisk indicates values significantly different from those of the WT untreated plants by one-way ANOVA test (P <0.05). The polyamine back-conversion pathway was also analysed in Arabidopsis transgenic plants constitutively overexpressing AtPAO3 alone (35S::AtPAO3 transgenic plants; Supplementary Fig. S3 at JXB online) or as a fusion to the C-terminus of the green fluorescent protein (GFP; 35S::GFP-AtPAO3 transgenic plants; Supplementary Fig. S3), as well as in an Arabidopsis knock-out mutant for AtPAO3 (atpao3 mutant; Supplementary Fig. S4). All these Arabidopsis transgenic lines exhibited no morphological alterations as compared with the wild-type plants. The transgenic lines with increased AtPAO3 expression levels (line 2 in the case of 35S::AtPAO3 transgenic plants and line 6 in the case of 35S::GFP-AtPAO3 transgenic plants; Supplementary Fig. S3) had increased Put levels as compared with the wild-type plants (Fig. 7A), in accordance with the recent demonstration that recombinant AtPAO3 produces Put as the end-product (Moschou et al., 2008a). In contrast, Spm and Spd levels remained unchanged, which may be due to the fact that Spd is both a product and a substrate of AtPAO3 and/or to the activation of homeostatic mechanisms also involving biosynthetic transport and conjugation pathways. Supply of exogenous Spd (1 mM) resulted in a much higher increase in Put levels in both the 35S::AtPAO3 and the 35S::GFP-AtPAO3 transgenic plants as compared with the Spd-treated wild-type plants (Fig. 7B). In all cases, the increase in Put levels was partially inhibited when plants were treated with guazatine (Fig. 7B). Fig. 7. View largeDownload slide Recombinant AtPAO3 constitutively expressed in Arabidopsis transgenic plants is involved in a polyamine back-conversion pathway. (A) Polyamine levels in 35S::AtPAO3 and 35S::GFP-AtPAO3 transgenic plants (transgenic lines 2 and 6, respectively) at the T3 generation. (B) Put levels in 35S::AtPAO3 and 35S::GFP-AtPAO3 transgenic plants. Polyamine levels were determined in untreated plants (control) and in plants treated with 1 mM Spd alone or together with 10 μM of the PAO-specific inhibitor guazatine (GU) for 24 h. Data are from a single representative experiment, which was repeated twice. Each point represents the mean value from three independent analyses of polyamine levels and the bars indicate the SE. Asterisks indicate values significantly different from the corresponding values of the wild-type plants (WT) by one-way ANOVA test (P <0.05). Fig. 7. View largeDownload slide Recombinant AtPAO3 constitutively expressed in Arabidopsis transgenic plants is involved in a polyamine back-conversion pathway. (A) Polyamine levels in 35S::AtPAO3 and 35S::GFP-AtPAO3 transgenic plants (transgenic lines 2 and 6, respectively) at the T3 generation. (B) Put levels in 35S::AtPAO3 and 35S::GFP-AtPAO3 transgenic plants. Polyamine levels were determined in untreated plants (control) and in plants treated with 1 mM Spd alone or together with 10 μM of the PAO-specific inhibitor guazatine (GU) for 24 h. Data are from a single representative experiment, which was repeated twice. Each point represents the mean value from three independent analyses of polyamine levels and the bars indicate the SE. Asterisks indicate values significantly different from the corresponding values of the wild-type plants (WT) by one-way ANOVA test (P <0.05). The atpao3 mutant, in which AtPAO3 expression is interrupted (Supplementary Fig. S4 at JXB online), did not show statistically significant changes in the levels of the specific polyamines as compared with the wild-type plants (Fig. 8A), probably due to gene redundancy and/or activation of homeostatic mechanisms. However, supply of exogenous Spd (1 mM) resulted in a less pronounced increase in Put levels in the atpao3 plants than in the wild-type plants, and in both cases this Spd-induced increase in Put levels was inhibited by guazatine (Fig. 8B). These data are in line with the idea of a polyamine back-conversion pathway in vivo. The recently published results, showing increased levels of Spd and Spm and decreased levels of Put in the roots and the shoots of atpao4 knock-out mutants as compared with the wild-type plants (Kamada-Nobusada et al., 2008), are also in line with the existence of a polyamine back-conversion pathway in vivo. The differences between the atpao4 (Kamada-Nobusada et al., 2008) and atpao3 (this study) mutants in polyamine levels may be due either to a different contribution of the two enzymes in the polyamine back-conversion pathway or to different tissues tested. Fig. 8. View largeDownload slide Analysis of the polyamine back-conversion pathway in the atpao3 mutant. (A) Polyamine levels in atpao3 and wild-type (WT) Arabidopsis plants. (B) Put levels in atpao3 and WT plants. Polyamine levels were determined in untreated whole seedlings (control) and in seedlings treated with 1 mM Spd alone or together with 10 μM of the PAO-specific inhibitor guazatine (GU) for 24 h. Each point represents the mean value from three independent analyses of polyamine levels and the bars indicate the SD. Asterisks indicate values significantly different from those of the corresponding control plants (a) or from the corresponding values of the WT plants (b) by one-way ANOVA test (P <0.05). Fig. 8. View largeDownload slide Analysis of the polyamine back-conversion pathway in the atpao3 mutant. (A) Polyamine levels in atpao3 and wild-type (WT) Arabidopsis plants. (B) Put levels in atpao3 and WT plants. Polyamine levels were determined in untreated whole seedlings (control) and in seedlings treated with 1 mM Spd alone or together with 10 μM of the PAO-specific inhibitor guazatine (GU) for 24 h. Each point represents the mean value from three independent analyses of polyamine levels and the bars indicate the SD. Asterisks indicate values significantly different from those of the corresponding control plants (a) or from the corresponding values of the WT plants (b) by one-way ANOVA test (P <0.05). Discussion Here, a comparative study on the catalytic properties of the various AtPAOs is presented, which evidences functional diversity inside the A. thaliana PAO gene family. In particular, recombinant AtPAO1 seems to be active only with Spm and not with Spd, in contrast to recombinant AtPAO2 and AtPAO3 which oxidize both Spm and Spd. Also recombinant AtPAO4 oxidizes both Spm and Spm; however, since its catalytic activity towards Spm is 40-fold higher than towards Spd, Spm can be considered as its physiological substrate. Furthermore, although all four recombinant proteins oxidize the two ‘uncommon’ polyamines Therm-Spm and Nor-Spm, for AtPAO1 these are better substrates than Spm, implying that they may be its physiological substrates. The four recombinant proteins also present differences in their catalytic activity towards N1-acetyl-Spm. Indeed, while for AtPAO1, AtPAO3, and AtPAO4 the catalytic activity towards N1-acetyl-Spm is 10- to 300-fold lower than towards Spm, for AtPAO2 this is only 5-fold lower. These results fully agree with a work published while the present article was under review (Takahashi et al., 2010). Some existing discrepancies between the two studies, mainly regarding AtPAO4 substrate specificity found restricted to Spm by Takahashi et al., can be attributed to differences in the experimental conditions and are probably of low physiological significance considering the much lower catalytic activity of AtPAO4 towards Spd than Spm found in the present study. The functional diversification among the members of the AtPAO gene family probably reflects distinct physiological roles, a hypothesis also supported by the distinct subcellular localization (Kamada-Nobusada et al., 2008; Moschou et al., 2008,a) and tissue-specific expression pattern (Takahashi et al., 2010; PT, unpublished data) of the various AtPAOs. In the present study it was also shown that all recombinant AtPAOs characterized to date (i.e. AtPAO1, AtPAO2, AtPAO3, and AtPAO4) have lower catalytic efficiency and different pH optima (7.0–8.0) than ZmPAO (pH optimum of 6.0; Polticelli et al., 2005). These differences could be related to different subcellular localization and/or physiological role(s). Indeed, in contrast to the extracellular localization of ZmPAO, a peroxisomal localization was shown for the three members of the Arabidopsis AtPAO2–AtPAO4 subfamily (Kamada-Nobusada et al., 2008; Moschou et al., 2008,a), and a cytosolic localization is predicted for AtPAO1 (Tavladoraki et al., 2006) and AtPAO5 (this study). It is possible that the catalytic properties of the various plant PAOs match up with the endogenous levels of substrates and/or products in the site of enzyme expression in harmony with specific physiological functions. However, to make such a correlation, more information on the subcellular localization, tissue-specific distribution, and intra- or intercellular transport of polyamines in plants is still necessary. Recent data and data herein support that all characterized A. thaliana PAOs are involved in a polyamine back-conversion pathway, similar to the animal PAOs/SMOs and in contrast to all PAOs from monocotyledonous plants characterized thus far (such as, for example, the apoplastic ZmPAO) which are involved in a terminal polyamine catabolic pathway. Indeed, the existence of a polyamine back-conversion pathway was also shown in vivo in Arabidopsis plants as well as in N. tabacum plants in which a PAO with a high sequence homology to AtPAO1 is present. Interestingly, a polyamine back-conversion pathway may also exist in maize, since putative PAOs with high sequence homology to AtPAO2–AtPAO4 (B6SYR8, B4F9F6, B6SW44) and to AtPAO5 (C0PE4) and with predicted intracellular localization were found by similarity searches in the maize genome, suggesting that the two polyamine catabolic pathways (the terminal polyamine catabolic pathway and the polyamine back-conversion pathway) co-exist in this plant. In this regard, the information so far available allows the proposal that the terminal catabolic pathway of polyamines is specifically active in the extracellular compartments while the polyamine back-conversion pathway is mostly intracellular. In mammals, polyamines are acetylated by the enzyme SSAT and peroxisomal PAOs are implicated in the catabolism of these polyamine derivatives. In Arabidopsis the much lower catalytic efficiency of the three peroxisomal PAOs (AtPAO2, AtPAO3, and AtPAO4) towards N1-acetyl-Spm than towards Spm and Spd suggests that acetylated polyamines may not be their physiological substrates. Indeed, AtPAO4-deficient mutants do not show differences in the levels of N1-acetyl-Spm compared with the wild-type plants (Kamada-Nobusada et al., 2008). However, an in vivo role for AtPAO2, which exhibits the highest activity towards N1-acetyl Spm as compared with AtPAO3 and AtPAO4 (Table 1), in the catabolism of acetylated polyamines cannot be excluded. In mammals, acetylated polyamines and their oxidation are considered to have a regulatory role. Indeed, it is believed that polyamines are acetylated in order to be excreted from the cells (Seiler, 1995; Wu et al., 2003) and PAOs may prevent this transport, resulting in increased intracellular Spd and Put levels (Wu et al., 2003). Furthermore, acetylation reduces the charge of polyamines and this probably alters their ability to interact with other molecules and thereby their biological function (Pegg, 2008). In this case, peroxisomal PAOs may play an important role in the homeostasis of acetylated polyamines. In contrast, the specific biological role of the different SSAT-independent polyamine back-conversion pathways with different localization is still unknown in both mammals and plants. They may have a role not only in polyamine homeostasis but also in signalling through H2O2 (Moschou et al., 2008,b) and/or aminoaldehyde production. More studies are necessary for a detailed understanding of the specific physiological roles of the distinct polyamine back-conversion pathways in plants as opposed to the terminal catabolic pathways of polyamines which have been shown to be involved in plant development and defence responses to biotic and abiotic stresses (Cona et al., 2006; Yoda et al., 2006, 2009; Kusano et al., 2008; Moschou et al., 2008b). The increased catalytic activity of AtPAO1 towards the polyamines Therm-Spm and Nor-Spm is of great significance considering that in A. thaliana a gene (ACL5) encoding a protein able to synthesize Therm-Spm from Spd has been characterized (Knott et al., 2007) and Therm-Spm has been detected (Kakehi et al., 2008; Naka et al., 2010; Rambla et al., 2010). Furthermore, the acl5 Arabidopsis mutant shows defects in stem elongation (Hanzawa et al., 1997) and vascular development (Clay and Nelson, 2005; Muñiz et al., 2008; Vera-Sirera et al., 2010). Characterization of loss-of-function Arabidopsis mutants for AtPAO1 as well as of Arabidopsis transgenic plants overexpressing AtPAO1 is necessary to determine the AtPAO1 function in vivo. Despite the elevated sequence homology to each other, the peroxisomal AtPAO2, AtPAO3, and AtPAO4 show some important differences in substrate specificity, which probably suggest distinct physiological role(s). In particular, while AtPAO2 is equally active with Spm and Spd, AtPAO3 is 2-fold more active with Spd than with Spm, whereas AtPAO4 is 10-fold more active with Spm than with Spd. Differences also exist in their catalytic activity towards N1-acetyl-Spm and the uncommon polyamines. On the other hand, AtPAO2 and ZmPAO have similar substrate specificity despite the low sequence homology to each other, while, in contrst, AtPAO1 and ZmPAO have different substrate specificity despite the elevated sequence homology to each other (Tavladoraki et al., 2006; this study). These data, together with a detailed comparative study of the amino acid sequence and structure of the various PAOs/SMOs characterized to date, will probably allow the discovery of the determinants for substrate specificity and catalytic mechanism(s). This study is a new starting point for further analysis of the physiological roles of the polyamine back-conversion pathway in plants, and sensu lato of polyamine catabolism. Abbreviations Abbreviations ABA abscisic acid AtPAO Arabidopsis thaliana polyamine oxidase Dap 1,3-diaminopropane Fms1 yeast Spm-oxidase HvPAO Hordeum vulgare polyamine oxidase MBP maltose-binding protein Nor-Spm norspermine NtPAO Nicotiana tabacum polyamine oxidase PAO polyamine oxidase PCA perchloric acid PMSF phenylmethysulphonyl fluoride Put putrescine SMO Spm-oxidase Spd spermidine Spm spermine Therm-Spm thermospermine ZmPAO Zea mays polyamine oxidase We wish to thank Dr Lucia Pomettini and Dr Sylvia Ciferri for technical assistance. We are grateful to Plant Systems Biology (University of Gent) for the kind gift of the pK2GW7 binary vector, RIKEN BioResource Center for the cDNA clone pda07387, the ABRC stock center for the cDNA U21196, and the SALK Institute for the SALK_121288 mutant. We are also grateful to Professor Armin Geyer for providing thermospermine. This work was supported by the University ‘Roma Tre’ and the Italian Ministry of University and Research (PRIN 2005 and PRIN 2007), and implemented in the frame of COST FA605 Action. References Alcázar R,  Altabella T,  Marco F,  Bortolotti C,  Reymond M,  Koncz C,  Carrasco P,  Tiburcio AF.  Polyamines: molecules with regulatory functions in plant abiotic stress tolerance,  Planta ,  2010, vol.  231 (pg.  1237- 1249) Google Scholar CrossRef Search ADS PubMed  Altschul SF,  Gish W,  Miller W,  Myers EW,  Lipman DJ.  Basic local alignment search tool,  Journal of Molecular Biology ,  1990, vol.  215 (pg.  403- 410) Google Scholar CrossRef Search ADS PubMed  Binda C,  Coda A,  Angelini R,  Federico R,  Ascenzi P,  Mattevi A.  A 30-Å-long U-shaped catalytic tunnel in the crystal structure of polyamine oxidase,  Structure ,  1999, vol.  7 (pg.  265- 276) Google Scholar CrossRef Search ADS PubMed  Binda C,  Newton-Vinson P,  Hubalek F,  Edmondson DE,  Mattevi A.  Structure of human monoamine oxidase B, a drug target for the treatment of neurological disorders,  Nature Structural Molecular Biology ,  2002, vol.  9 (pg.  22- 26) Google Scholar CrossRef Search ADS   Cervelli M,  Cona A,  Angelini R,  Polticelli F,  Federico R,  Mariottini P.  A barley polyamine oxidase isoform with distinct structural features and subcellular localization,  European Journal of Biochemistry ,  2001, vol.  268 (pg.  3816- 3830) Google Scholar CrossRef Search ADS PubMed  Cervelli M,  Polticelli F,  Federico R,  Mariottini P.  Heterologous expression and characterization of mouse spermine oxidase,  Journal of Biological Chemistry ,  2003, vol.  278 (pg.  5271- 5276) Google Scholar CrossRef Search ADS PubMed  Clay NK,  Nelson T.  Arabidopsis thickvein mutation affects vein thickness and organ vascularization, and resides in a provascular cell specific spermine synthase involved in vein definition and in polar auxin transport,  Plant Physiology ,  2005, vol.  138 (pg.  767- 777) Google Scholar CrossRef Search ADS PubMed  Clough SJ,  Bent AF.  Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana,  The Plant Journal ,  1998, vol.  16 (pg.  735- 743) Google Scholar CrossRef Search ADS PubMed  Cona A,  Rea G,  Angelini R,  Federico R,  Tavladoraki P.  Functions of amine oxidases in plant development and defence,  Trends in Plant Science ,  2006, vol.  11 (pg.  80- 88) Google Scholar CrossRef Search ADS PubMed  Edmondson DE,  Newton-Vinson P.  The covalent FAD of monoamine oxidase: structural and functional role and mechanism of the flavinylation reaction,  Antioxidants and Redox Signaling ,  2001, vol.  3 (pg.  789- 806) Google Scholar CrossRef Search ADS PubMed  Edmondson DE,  Binda C,  Mattevi A.  The FAD binding sites of human monoamine oxidases A and B,  Neurotoxicology ,  2004, vol.  25 (pg.  63- 72) Google Scholar CrossRef Search ADS PubMed  Fraaije MW,  van den Heuvel RHH,  van Berkel WJH,  Mattevi A.  Structural analysis of flavinylation in vanillyl-alcohol oxidase,  Journal of Biological Chemistry ,  2000, vol.  275 (pg.  38654- 38658) Google Scholar CrossRef Search ADS PubMed  Hanzawa Y,  Takahashi T,  Komeda Y.  ACL5: an Arabidopsis gene required for internodal elongation after flowering,  The Plant Journal ,  1997, vol.  12 (pg.  863- 874) Google Scholar CrossRef Search ADS PubMed  Heuts DPHM,  Scrutton NS,  McIntire WS,  Fraaije MW.  What's in a covalent bond? On the role and formation of covalently bound flavin cofactors,  FEBS Journal ,  2009, vol.  276 (pg.  3405- 3427) Google Scholar CrossRef Search ADS PubMed  Kakehi J,  Kuwashiro Y,  Niitsu M,  Takahashi T.  Thermospermine is required for stem elongation in Arabidopsis thaliana,  Plant and Cell Physiology ,  2008, vol.  49 (pg.  1342- 1349) Google Scholar CrossRef Search ADS PubMed  Kamada-Nobusada T,  Hayashi M,  Fukazawa M,  Sakakibara H,  Nishimura M.  A putative peroxisomal polyamine oxidase, AtPAO4, is involved in polyamine catabolism in Arabidopsis thaliana,  Plant and Cell Physiology ,  2008, vol.  49 (pg.  1272- 1282) Google Scholar CrossRef Search ADS PubMed  Knott JM,  Römer P,  Sumper M.  Putative spermine synthases from Thalassiosira pseudonana and Arabidopsis thaliana synthesize thermospermine rather than spermine,  FEBS Letters ,  2007, vol.  581 (pg.  3081- 3086) Google Scholar CrossRef Search ADS PubMed  Kusano T,  Berberich T,  Tateda C,  Takahashi Y.  Polyamines: essential factors for growth and survival,  Planta ,  2008, vol.  228 (pg.  367- 381) Google Scholar CrossRef Search ADS PubMed  Landry J,  Sternglanz R.  Yeast Fms1 is a FAD-utilizing polyamine oxidase,  Biochemical and Biophysical Research Communications ,  2003, vol.  303 (pg.  771- 776) Google Scholar CrossRef Search ADS PubMed  Mattoo AK,  Minocha SC,  Minocha R,  Handa AK.  Polyamines and cellular metabolism in plants: transgenic approaches reveal different responses to diamine putrescine versus higher polyamines spermidine and spermine,  Amino Acids ,  2010, vol.  38 (pg.  405- 413) Google Scholar CrossRef Search ADS PubMed  Molinari P,  Marusic C,  Lucioli A,  Tavazza R,  Tavazza M.  Identification of artichoke mottled crinkle virus (AMCV) proteins required for viral replication: complementation of AMCV p33 and p92 replication-defective mutants,  Journal of General Virology ,  1998, vol.  79 (pg.  639- 647) Google Scholar CrossRef Search ADS PubMed  Moschou PN,  Paschalidis KA,  Delis ID,  Andriopoulou AH,  Lagiotis GD,  Yakoumakis DI,  Roubelakis-Angelakis KA.  b. Spermidine exodus and oxidation in the apoplast induced by abiotic stress is responsible for H2O2signatures that direct tolerance responses in tobacco,  The Plant Cell ,  2008, vol.  20 (pg.  1708- 1724) Google Scholar CrossRef Search ADS PubMed  Moschou PN,  Sanmartin M,  Andriopoulou AH,  Rojo E,  Sanchez-Serrano JJ,  Roubelakis-Angelakis KA.  Bridging the gap between plant and mammalian polyamine catabolism: a novel peroxisomal polyamine oxidase responsible for a full back-conversion pathway in Arabidopsis,  Plant Physiology ,  2008, vol.  47 (pg.  1845- 1857) Google Scholar CrossRef Search ADS   Muñiz L,  Minguet EG,  Singh SK,  Pesquet E,  Vera-Sirera F,  Moreau-Courtois CL,  Carbonell J,  Blázquez MA,  Tuominen H.  ACAULIS5 controls Arabidopsis xylem specification through the prevention of premature cell death,  Development ,  2008, vol.  135 (pg.  2573- 2582) Google Scholar CrossRef Search ADS PubMed  Murashige T,  Skoog F.  A revised medium for rapid growth and bioassays with tobacco tissue cultures,  Physiologia Plantarum ,  1962, vol.  15 (pg.  473- 497) Google Scholar CrossRef Search ADS   Naka Y,  Watanabe K,  Sagor GH,  Niitsu M,  Pillai MA,  Kusano T,  Takahashi Y.  Quantitative analysis of plant polyamines including thermospermine during growth and salinity stress,  Plant Physiology and Biochemistry ,  2010, vol.  48 (pg.  527- 533) Google Scholar CrossRef Search ADS PubMed  Pegg AE.  Spermidine/spermine-N1-acetyltransferase: a key metabolic regulator,  American Journal of Physiology–Endocrinology and Metabolism ,  2008, vol.  294 (pg.  E995- E1010) Google Scholar CrossRef Search ADS PubMed  Polticelli F,  Basran J,  Faso C,  Cona A,  Minervini G,  Angelini R,  Federico R,  Scrutton NS,  Tavladoraki P.  Lys300 plays a major role in the catalytic mechanism of maize polyamine oxidase,  Biochemistry ,  2005, vol.  44 (pg.  16108- 16120) Google Scholar CrossRef Search ADS PubMed  Rambla JL,  Vera-Sirera F,  Blázquez MA,  Carbonell J,  Granell A.  Quantitation of biogenic tetraamines in Arabidopsis thaliana,  Analytical Biochemistry ,  2010, vol.  397 (pg.  208- 211) Google Scholar CrossRef Search ADS PubMed  Seiler N.  Polyamine oxidase, properties and functions,  Progress in Brain Research ,  1995, vol.  106 (pg.  333- 344) Google Scholar PubMed  Seiler N.  Catabolism of polyamines,  Amino Acids ,  2004, vol.  26 (pg.  217- 233) Google Scholar PubMed  Seki M,  Carninci P,  Nishiyama Y,  Hayashizaki Y,  Shinozaki K.  High-efficiency cloning of Arabidopsis full-length cDNA by biotinylated CAP trapper,  The Plant Journal ,  1998, vol.  15 (pg.  707- 720) Google Scholar CrossRef Search ADS PubMed  Seki M,  Narusaka M,  Kamiya A, et al.  Functional annotation of a full-length Arabidopsis cDNA collection,  Science ,  2002, vol.  296 (pg.  141- 144) Google Scholar CrossRef Search ADS PubMed  Smith TA,  Davies PJ.  Separation and quantitation of polyamines in plant tissue by high performance liquid chromatography of their dansyl derivatives,  Plant Physiology ,  1985, vol.  78 (pg.  89- 91) Google Scholar CrossRef Search ADS PubMed  Takahashi Y,  Cong R,  Sagor GH,  Niitsu M,  Berberich T,  Kusano T.  Characterization of five polyamine oxidase isoforms in Arabidopsis thaliana,  Plant Cell Reports ,  2010, vol.  29 (pg.  955- 965) Google Scholar CrossRef Search ADS PubMed  Tavladoraki P,  Rossi MN,  Saccuti G,  Perez-Amador MA,  Polticelli F,  Angelini R,  Federico R.  Heterologous expression and biochemical characterization of a polyamine oxidase from Arabidopsis involved in polyamine back-conversion,  Plant Physiology ,  2006, vol.  141 (pg.  1519- 1532) Google Scholar CrossRef Search ADS PubMed  Thompson JD,  Higgins DG,  Gibson TJ.  CLUSTALW: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice,  Nucleic Acids Research ,  1994, vol.  22 (pg.  4673- 4680) Google Scholar CrossRef Search ADS PubMed  Vera-Sirera F,  Minguet EG,  Singh SK,  Ljung K,  Tuominen H,  Blázquez MA,  Carbonell J.  Role of polyamines in plant vascular development,  Plant Physiology and Biochemistry ,  2010, vol.  48 (pg.  534- 539) Google Scholar CrossRef Search ADS PubMed  Vujcic S,  Diegelman P,  Bacchi CJ,  Kramer DL,  Porter CW.  Identification and characterization of a novel flavin-containing spermine oxidase of mammalian cell origin,  Biochemical Journal ,  2002, vol.  367 (pg.  665- 675) Google Scholar CrossRef Search ADS PubMed  Vujcic S,  Liang P,  Diegelman P,  Kramer DL,  Porter CW.  Genomic identification and biochemical characterization of the mammalian polyamine oxidase involved in polyamine back-conversion,  Biochemical Journal ,  2003, vol.  370 (pg.  19- 28) Google Scholar CrossRef Search ADS PubMed  Wallace HM,  Fraser AV,  Hughes A.  A perspective of polyamine metabolism,  Biochemical Journal ,  2003, vol.  76 (pg.  1- 14) Google Scholar CrossRef Search ADS   Wang Y,  Devereux W,  Woster PM,  Stewart TM,  Hacker A,  Casero RA Jr.  Cloning and characterization of a human polyamine oxidase that is inducible by polyamine analogue exposure,  Cancer Research ,  2001, vol.  61 (pg.  5370- 5373) Google Scholar PubMed  Wu T,  Yankovskaya V,  McIntire WS.  Cloning, sequencing, and heterologous expression of the murine peroxisomal flavoprotein, N1-acetylated polyamine oxidase,  Journal of Biological Chemistry ,  2003, vol.  278 (pg.  20514- 20525) Google Scholar CrossRef Search ADS PubMed  Yoda H,  Fujimura K,  Takahashi H,  Munemura I,  Uchimiya H,  Sano H.  Polyamines as a common source of hydrogen peroxide in host- and nonhost hypersensitive response during pathogen infection,  Plant Molecular Biology ,  2009, vol.  70 (pg.  103- 112) Google Scholar CrossRef Search ADS PubMed  Yoda H,  Hiroi Y,  Sano H.  Polyamine oxidase is one of the key elements for oxidative burst to induce programmed cell death in tobacco cultured cells,  Plant Physiology ,  2006, vol.  142 (pg.  193- 206) Google Scholar CrossRef Search ADS PubMed  © The Author [2010]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: [email protected]