Plant Physiology

Minorsky, Peter V.

doi: 10.1104/pp.15.00536pmid: N/A

Parental Age Affects Filial Somatic Mutation Rates In humans, it is well known that the reproductive age of the parents has a strong influence on the health of their progeny. For example, a mother of older reproductive age is more apt to produce aneuploid gametes. The risk for a trisomic offspring increases from 2% to 3% for mothers in their 20s to more than 30% for mothers in their 40s. The age of the father also has an effect on the frequency of spontaneous congenital disorders and common complex diseases, such as autism and some cancers. Sperm from 36- to 57-year-old men have more double-strand breaks than those of 20- to 35-year-old individuals. Hence, it is clear that the germinal mutation rates are a function of both maternal and paternal age in humans. In contrast, it is unknown whether the parental reproductive age has an effect on somatic mutation rates in the progeny because these are rare and difficult to detect. To address this question, Singh et al. (pp. 247–257) took advantage of Arabidopsis (Arabidopsis thaliana), where mutation detector lines allow for an easy quantitation of somatic mutations, to test the effect of parental age on somatic mutation rates in the progeny. They found no significant effect of parental age on base substitutions, but frameshift mutations and transposition events were enhanced in the progeny of older parents, an effect that is stronger through the maternal line. In contrast, intrachromosomal recombination events in the progeny decrease with the age of the parents in a parent-of-origin-dependent manner. These results show that parental reproductive age affects somatic mutation rates in the progeny and, thus, that some form of age-dependent information is transmitted through the gametes, which affects the frequency of double-strand breaks. S-Sulfhydration: A Posttranslational Modification in Plants Hydrogen sulfide (H2S) is a highly reactive molecule that has recently emerged as an important signaling compound with many physiological functions. H2S has been shown to participate in diverse physiological processes in animals, including cardioprotection, neuromodulation, inflammation, apoptosis, and gastrointestinal functions, among others. The possible role of H2S as an endogenous neuromodulator was first described in 1996, and the molecule is now accepted as the third most prevalent gasotransmitter after nitric oxide and carbon monoxide. Although many studies have been conducted on the physiological effects of H2S in mammals, and more recently in plants, the underlying mechanisms are poorly understood. Nonetheless, two mechanisms have been proposed based on the chemical properties of H2S. The nucleophilic properties of this molecule and its capacity to react with oxygen, hydrogen peroxide, or peroxynitrite suggest that it acts by reducing cellular oxidative stress. The second mechanism involves the posttranslational modification of protein Cys residues to form a persulfide group. This process is called S-sulfhydration, as opposed to S-nitrosylation (i.e. the posttranslational modification of protein Cys residues by nitric oxide to form S-nitroso-Cys residues). Although S-nitrosylation typically inhibits protein function, the effect of S-sulfhydration can either activate or inactivate enzymatic activities. Aroca et al. (334–342) have examined Arabidopsis proteins modified by S-sulfhydration under physiological conditions. A total of 106 S-sulfhydrated proteins were identified. Immunoblot and enzyme activity analyses of some of these proteins showed that the sulfide added through S-sulfhydration reversibly regulates the functions of plant proteins in a manner similar to that described in mammalian systems. These results constitute the first report of S-sulfhydration as a posttranslational modification in plants. Potassium Channel Regulation by H2S H2S affects many physiological processes in animals by modulating channels. H2S has also been reported to close stomata, but the underlying mechanism remains elusive. In plants, H2S action was originally related to pathogenesis resistance, but in the last decade, it has been shown to have an active role in development, responses to various abiotic stresses, and stomatal movement. Moreover, H2S has been reported to participate in the signaling of several plant hormones, including abscisic acid (ABA). In guard cells, ABA induces an increase of cytosolic-free Ca2+ concentration, elevates cytosolic pH, and activates the efflux of anions, mainly chloride, through S- and R-type anion channels. The increase in cytosolic-free Ca2+ concentration inactivates the inward-rectifying K+ channel (IKIN); anion efflux depolarizes the plasma membrane, and together with the rise in cytosolic pH, it activates K+ efflux through the outward-rectifying K+ channel. These changes in ion flux, in turn, generate an osmotically driven reduction in turgor and volume and closure of the stomatal pore. Papanatsiou et al. (pp. 29–35) have employed two-electrode voltage clamp measurements to study the role of H2S in the regulation of the guard cell K+ channels of tobacco (Nicotiana tabacum). Their results show that H2S selectively inactivates IKIN and that this action parallels that of stomatal closure. Treatments that scavenge H2S suggest that the effect of H2S is separable from that of ABA. Thus, H2S seems to define a unique and unresolved signaling pathway that selectively targets IKINs. Plant Response to Nitrogen and Nocturnal CO2 The CO2 concentration in Earth’s atmosphere has increased from about 270 to 400 mmol mol–1 since 1800 and may double before the end of the century. Plant respiration is a major contributor to the global carbon cycle. Elevated atmospheric CO2 concentrations may either accelerate or decelerate plant respiration for reasons that are unclear. Plant responses to elevated CO2 are highly variable, but plant nitrogen (N) concentrations generally decline. It has recently been established that elevated CO2 during the daytime decreases plant mitochondrial respiration in the light as well as protein titer because CO2 slows the daytime conversion of nitrate (NO3−) into protein. This effect is traceable to the inhibitory influence of CO2 on photorespiration and the dependence of shoot NO3− assimilation on photorespiration. Elevated CO2 also inhibits the translocation of nitrite into the chloroplast, a response that influences shoot NO3− assimilation during both day and night. Asensio et al. (pp. 156–163) exposed Arabidopsis and wheat (Triticum aestivum) plants to daytime or nighttime elevated CO2 and supplied them with NO3− or ammonium as a sole N source. Six independent measures, including plant biomass, shoot NO3−, shoot organic N, 15N isotope fractionation, 15NO3− assimilation, and the ratio of shoot CO2 evolution to O2 consumption, indicated that elevated CO2 at night slowed NO3− assimilation and thus decreased dark respiration in the plants reliant on NO3−. These results provide a straightforward explanation for the diverse responses of plants to elevated CO2 at night and suggest that soil N source will have an increasing influence on the capacity of plants to mitigate human greenhouse gas emissions. Autophagy Contributes to the Growth of Rice N is an essential nutrient that strongly influences plant growth and productivity. A large input of N fertilizer is required to maintain high crop yields but comes with high costs to both the farmer and the environment. N use efficiency is recognized as an important target for improvement to promote sustainable and productive agriculture. The remobilization of assimilated N from senescent leaves greatly affects N use efficiency in cereal crops. Autophagy facilitates the degradation of intracellular components for nutrient recycling in all eukaryotes, including plants. Much of the N in leaves is distributed to chloroplasts, mainly in photosynthetic proteins. During leaf senescence, chloroplastic proteins, including Rubisco, are rapidly degraded, and the released N is remobilized and reused in newly developing tissues. Wada et al. (pp. 60–73) have examined in particular the function of autophagy in the vegetative growth and N usage of rice (Oryza sativa). An autophagy-disrupted rice mutant (Osatg7-1) showed reduced biomass production and N use efficiency compared with the wild type. Although Osatg7-1 showed early visible leaf senescence, its N concentration remained high in the senescent leaves. 15N pulse chase analysis revealed the suppression of N remobilization during leaf senescence in Osatg7-1. Accordingly, the reduction of N available for newly developing tissues in Osatg7-1 probably led to its reduced leaf area and tillering. The limited leaf growth in Osatg7-1 decreased the photosynthetic capacity of the plant. Much of the N remaining in senescent leaves of Osatg7-1 was in soluble proteins, and the Rubisco concentration in senescing leaves of Osatg7-1 was about 2.5 times higher than in the wild type. These results suggest that autophagy contributes to efficient N remobilization at the whole-plant level by facilitating protein degradation for N recycling in senescent leaves. Salt Reduces Meristem Size via Nitric Oxide and Auxin Soil salinity is a serious factor limiting the productivity and quality of agricultural crops. Worldwide, high salinity in the soil damages approximately 20% of total irrigated lands and takes 1.5 million ha out of production each year. In general, high salinity affects plant growth and development by reducing plant water potential, altering nutrient uptake, and increasing the accumulation of toxic ions. Together, these effects severely reduce plant growth and survival. Because the root is the first organ to sense high salinity, salt stress plays a direct, important role in modulating root system architecture. For instance, salt stress negatively regulates root hair formation and gravitropism. High NaCl levels also inhibit lateral root formation, whereas lower NaCl levels stimulate lateral root formation in an auxin-dependent manner. Salt stress inhibits primary root elongation by decreasing the size and activity of root meristems. However, the mechanisms underlying this inhibition remain unclear. Liu et al. (pp. 343–356) have examined whether and how auxin and nitric oxide (NO) are involved in salt-mediated inhibition of root meristem growth in Arabidopsis using physiological, pharmacological, and genetic approaches. They report that salt stress significantly reduced root meristem size by down-regulating the expression of PINFORMED (PIN) auxin transporters, thereby reducing auxin levels. In addition, salt stress promoted AUXIN RESISTANT3 (AXR3)/INDOLE-3-ACETIC ACID17 (IAA17) stabilization, which repressed auxin signaling during this process. Furthermore, salt stress stimulated NO accumulation, whereas pharmacologically blocking NO production compromised the salt-mediated reduction of root meristem size, PIN down-regulation, and stabilization of AXR3/IAA17, indicating that NO is involved in salt-mediated inhibition of root meristem growth. Taken together, these findings suggest that salt stress inhibits root meristem growth by repressing PIN expression (thereby reducing auxin levels) and stabilizing IAA17 (thereby repressing auxin signaling) via increasing NO levels. Glossary H2S hydrogen sulfide ABA abscisic acid IKIN inward-rectifying K+ channel N nitrogen NO3− nitrate NO nitric oxide © 2015 American Society of Plant Biologists. All Rights Reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Oehme, Daniel P.; Downton, Matthew T.; Doblin, Monika S.; Wagner, John; Gidley, Michael J.; Bacic, Antony

doi: 10.1104/pp.114.254664pmid: 25786828

Abstract The question of how many chains an elementary cellulose microfibril contains is critical to understanding the molecular mechanism(s) of cellulose biosynthesis and regulation. Given the hexagonal nature of the cellulose synthase rosette, it is assumed that the number of chains must be a multiple of six. We present molecular dynamics simulations on three different models of Iβ cellulose microfibrils, 18, 24, and 36 chains, to investigate their structure and dynamics in a hydrated environment. The 36-chain model stays in a conformational space that is very similar to the initial crystalline phase, while the 18- and 24-chain models sample a conformational space different from the crystalline structure yet similar to conformations observed in recent high-temperature molecular dynamics simulations. Major differences in the conformations sampled between the different models result from changes to the tilt of chains in different layers, specifically a second stage of tilt, increased rotation about the O2-C2 dihedral, and a greater sampling of non-TG exocyclic conformations, particularly the GG conformation in center layers and GT conformation in solvent-exposed exocyclic groups. With a reinterpretation of nuclear magnetic resonance data, specifically for contributions made to the C6 peak, data from the simulations suggest that the 18- and 24-chain structures are more viable models for an elementary cellulose microfibril, which also correlates with recent scattering and diffraction experimental data. These data inform biochemical and molecular studies that must explain how a six-particle cellulose synthase complex rosette synthesizes microfibrils likely comprised of either 18 or 24 chains. Cellulose is the main structural polysaccharide in higher plant cell walls and has been studied intensively since it was first isolated by Anselme Payen in 1838 (Payen, 1838; Zugenmaier, 2001). As a result, there is significant accumulated knowledge of the most abundant biological macromolecule on earth. Cellulose plays an important role in defining the physical and mechanical properties of plant cells, such as their growth, flexibility, structural support, and responses to stresses (biotic and abiotic). These properties vary depending on whether the cell/tissue is undergoing rapid growth and expansion or is undergoing differentiation following the cessation of growth. Thus, the primary wall laid down at division must yield to stress produced by turgor pressure, leading to cell expansion. The shape of the cell is determined by differential yielding of walls in specific locations. At the completion of expansion, the mechanical properties of the wall can change so that it no longer yields to turgor pressure, but is rigidified so that it can withstand large compressive forces. Despite this, the mechanism(s) involved in how cellulose confers these properties are not well understood. One reason stems from the lack of consensus on the fine structure and dynamics of the elementary microfibril, the minimal functional unit of cellulose. Produced at the plasma membrane via a rosette of cellulose synthase (CesA) catalytic subunits (in plants) utilizing cytoplasmic UDP-Glc, cellulose is made up of layers of planar (1,4)-β-glucan chains, stacked upon each other to form microfibrils, whose directionality is controlled initially by the underlying cytoskeleton of microtubules on which the cellulose synthase tracks and which are usually transverse to the direction of turgor pressure-driven growth. These microfibrils can also form larger order aggregates that differ in their size depending upon the species/tissue/developmental state. X-ray and neutron diffraction experiments performed on cellulose from algae, such as Valonia spp., and in tunicates have identified single crystalline aggregates being up to 40 nm in diameter (Thomas et al., 2013), with those from higher plants generally ranging between 2 and 5 nm (Newman, 1999; Fernandes et al., 2011; Newman et al., 2013) and thought to be composed of single microfibrils. A number of different polymorphs have been identified for crystalline phases of cellulose. The native state is defined as cellulose I, with two allomorphs, β and α, that differ slightly in their unit cells. The β form dominates in plants and the α form in bacteria and algae, although both the β and α forms can be found from the same source and possibly in the same microfibril (Atalla and Vanderhart, 1984; VanderHart and Atalla, 1984; Horii et al., 1987; Sugiyama et al., 1991a, 1991b). The crystal and molecular structure of cellulose has been determined at atomic resolution for both the Iβ (Nishiyama et al., 2002) and Iα allomorphs (Nishiyama et al., 2003) using x-ray and neutron diffraction. Iβ cellulose crystals were found to have the monoclinic P21 space group, with two nonidentical chains arranged in a parallel manner. Iα cellulose crystals were found to have the triclinic P1 space group, with parallel chains of identical conformation. For both allomorphs, hydrogen bonding was reported within chains and between chains of the same layer, but not between layers. Additionally, the Iβ form is proposed to have two differing H-bond patterns, A and B, which are dependent on the position of hydrogens HO2 and HO6 (Fig. 1). The conformation of the exocyclic group at carbon C6 was also reported to reside in the TG conformation (Fig. 2). Figure 1. Open in new tabDownload slide Cellobiose dimers in the A (left) and B (right) hydrogen bonding patterns. Oxygen (red) atoms are labeled in the left dimer, while key dihedral angles are labeled in the right dimer. Hydrogens (gray) are displayed if involved in hydrogen bonds. Otherwise, they are omitted. Figure 1. Open in new tabDownload slide Cellobiose dimers in the A (left) and B (right) hydrogen bonding patterns. Oxygen (red) atoms are labeled in the left dimer, while key dihedral angles are labeled in the right dimer. Hydrogens (gray) are displayed if involved in hydrogen bonds. Otherwise, they are omitted. Figure 2. Open in new tabDownload slide Rotation about the O5-C5-C6-O6 dihedral allows for three different conformations of the C6 exocyclic group of each Glc monomer. The C6-O6 bond is colored yellow for the TG conformation (dihedral angle of 180°), green for GT (60°), and blue for GG (–60°). Hydrogens are not shown. Figure 2. Open in new tabDownload slide Rotation about the O5-C5-C6-O6 dihedral allows for three different conformations of the C6 exocyclic group of each Glc monomer. The C6-O6 bond is colored yellow for the TG conformation (dihedral angle of 180°), green for GT (60°), and blue for GG (–60°). Hydrogens are not shown. Numerous 13C-NMR studies have been performed to investigate the structure of cellulose microfibrils and aggregates (Earl and VanderHart, 1981; Horii et al., 1983, 1987; Isogai and Usuda, 1989; Newman and Hemmingson, 1994; Larsson et al., 1997; Ha et al., 1998; Wickholm et al., 1998; Viëtor et al., 2002; Larsson and Westlund, 2005; Park et al., 2009; Malm et al., 2010). Peaks attributed to the C4 and C6 carbons suggest that surface (amorphous) and interior (crystalline) chains have differing chemical shifts due to differences in either their environment, structure, or a combination of both (Viëtor et al., 2002). Data from the C6 peaks have also suggested that the exocyclic group for interior chains reside in the TG conformation, with this preference changing at the surface of the microfibril, where the GT and GG conformations are preferred (Viëtor et al., 2002). These conclusions were reached as a result of the earlier work of Horii et al. (1983) in which NMR studies on a range of oligosaccharides whose crystal structures had previously been solved identified three separate C6 peaks at 60 to 62, 62.5 to 64.5, and 65.6 to 66.6 ppm that corresponded to the GG, GT, and TG exocyclic conformations, respectively. Spectral fitting of NMR spectra has also been performed to deconvolute peaks and extract ultrastructural information about microfibril aggregates, such as crystalline, paracrystalline, surface, and allomorph proportions (Larsson et al., 1997; Wickholm et al., 1998; Larsson and Westlund, 2005). Based upon these physical and ultrastructural measurements, a number of different models have been proposed for an elementary cellulose microfibril. A 36-chain, hexagonal-shaped model was suggested by Ding and Himmel (2006) and thought to be an optimal model from transmission electron microscopy and atomic force microscopy studies (Ding et al., 2014), because the cellulose synthase rosette complex from which cellulose is produced in plants is thought to be composed of 36 cellulose synthase catalytic proteins (Herth, 1983). Other 36-chain complexes have been suggested that have a square shape, in which the hydrophobic (100) and hydrophilic (010) faces are exposed to solvent (Mazeau, 2005; Matthews et al., 2006; Bergenstråhle et al., 2007), and a diamond-like shape where two hydrophilic (110 and 1-10) faces are exposed (Matthews et al., 2006; Zhang et al., 2011; Chen et al., 2014). However, there is now evidence from wide-angle x-ray scattering (WAXS) and small-angle neutron scattering studies to suggest that 36 chains in a microfibril may be an overestimation (Fernandes et al., 2011; Thomas et al., 2013). Computer-simulated diffractograms for 18- and 24-chain models, with and without constant or regular shape, show a much better fit to the experimental data than the 36-chain model. Newman et al. (2013) also suggest that there are only six layers in the 100 plane of microfibrils, which differs from eight layers in the Ding and Himmel (2006) 36-chain model. Thus, the structure of an elementary cellulose microfibril is yet to be determined unequivocally. This is related to the significant experimental challenge of characterizing small-diameter cellulose (Matthews et al., 2012). X-ray and neutron diffraction studies have given us models of crystalline cellulose, but these are generally acquired at nonnative temperatures and environments. NMR gives information on microfibril aggregates in a more natural environment through solid-state techniques. These techniques allow spectra to be produced with water contents similar to those found in the primary cell wall of plants (Brett and Hillman, 1985) and provide a bridge between structures determined by diffraction techniques and those which exist in solution (Zugenmaier, 2001). However, to understand the natural structure and dynamics of microfibrils, how they interact with other microfibrils and components of the cell wall, and their chemical reactivity, we must be able to investigate single microfibrils at the atomic level in an environment as close to solution as possible, considering that in primary walls the aqueous phase can comprise 60% to 70% by weight (Doblin et al., 2010). One alternative is to use computational approaches, with numerous studies published in which molecular dynamics (MD) simulations have been performed on cellulose microfibrils, generally with a 36-chain model (Mazeau, 2005; Matthews et al., 2006, 2011a, 2011b, 2012; Bergenstråhle et al., 2007; Gross and Chu, 2010; Zhang et al., 2011; Chen et al., 2014). These simulations have investigated the structure and dynamics of the microfibrils, temperature dependence, H-bonding patterns, and effect of solvent water, and compared different carbohydrate force fields. Simulations have also been performed in different environments, whether that be purely crystalline or fully solvated. Additionally, different-shaped microfibrils have been simulated, as have microfibrils of finite or infinite length. We are only aware of one study where MD simulations have been performed on 18- or 24-chain models of cellulose (Busse-Wicher et al., 2014). In this study, we investigate the effect that different numbers of chains have on the structure and dynamics of a cellulose microfibril with the aim of obtaining a better understanding of how many chains make up an elementary cellulose microfibril. Considering the Iβ allomorph, we perform MD simulations on infinite-length microfibrils for long time scales using the Charmm carbohydrate force field. All microfibrils are built with the hexagonal shape suggested by Ding and Himmel (2006) such that the 1-10, 110, and 100/200 faces are exposed (Fig. 3A). This shape was chosen, as NMR data suggests that two hydrophilic faces (1-10 and 110) are exposed (Wickholm et al., 1998), while hydrophobic faces (100/200) have been included to correlate with the hexagonal shape of cellulose synthase rosettes. This also satisfies the requirement of the microfibril having a hydrophobic face with multiple chains for cellulose binding domains to interact with (Tormo et al., 1996; Lehtiö et al., 2003). From this work, we can define the dynamic characteristics of a microfibril containing 18, 24, or 36 chains in a natural environment with regards to exocyclic group conformation, O2 dihedral, H-bond patterns, tilt of chain with respect to the polymerization axis, and unit cell dimensions. We also obtain an understanding of how the different number of chains and, as a result, different surfaces affect the structuring of water around the microfibril. These insights will be useful for further investigating the role that cellulose plays in the structure and dynamics of the plant cell wall, particularly how it interacts with other microfibrils and the matrix phase polysaccharides during growth and development. This knowledge may also lead to insights into the underlying causes of the recalcitrance of plant walls to biotechnological utilization and also the mechanism(s) of cellulose synthesis. Figure 3. Open in new tabDownload slide Initial conformations for the 36-chain (A), 24-chain (B), and 18-chain (C) microfibrils, viewed from the nonreducing end. In A, the different hydrophobic (100/200) and hydrophilic (110/1-10) surfaces are labeled, as are the spacings between the different layers and the unit cell vectors. Note that the c vector runs into the page, along the polymerization axis. Atoms are colored red for oxygen, cyan for carbon, and gray for hydrogen. C6 and C4 carbons are highlighted in green and purple, respectively, for one surface chain. The inset to A shows the environment directly surrounding this chain, with an interior chain to the left and water solvent to the right. Viewed from above the microfibril, this highlights that the two C4 carbons of a cellobiose repeat are in the same environment, while the two C6 carbons are not. Figure 3. Open in new tabDownload slide Initial conformations for the 36-chain (A), 24-chain (B), and 18-chain (C) microfibrils, viewed from the nonreducing end. In A, the different hydrophobic (100/200) and hydrophilic (110/1-10) surfaces are labeled, as are the spacings between the different layers and the unit cell vectors. Note that the c vector runs into the page, along the polymerization axis. Atoms are colored red for oxygen, cyan for carbon, and gray for hydrogen. C6 and C4 carbons are highlighted in green and purple, respectively, for one surface chain. The inset to A shows the environment directly surrounding this chain, with an interior chain to the left and water solvent to the right. Viewed from above the microfibril, this highlights that the two C4 carbons of a cellobiose repeat are in the same environment, while the two C6 carbons are not. RESULTS MD simulations were performed to investigate the effect that differing numbers of chains had on the dynamics and structure of cellulose microfibrils in a hydrated environment. Simulations were performed on models of Iβ cellulose containing 18, 24, and 36 chains. The initial structures for each microfibril looking along the polymerization axis from the nonreducing end are shown in Figure 3, and the final simulation structures are shown in Figure 4. The work of Matthews et al. (2012) has shown that significant changes to the microfibril can occur over long time scales, and thus long MD simulations up to 640 ns were performed. Limited changes to microfibril structure and dynamics were observed after 200 ns of simulation (Supplemental Fig. S1), so analysis was limited to this time scale given the computational requirements of such simulations. Figure 4. Open in new tabDownload slide Final conformations for the 36-chain (A), 36-chain (B), 24-chain (C), and 18-chain (D) microfibrils, viewed from the nonreducing end. Layers are colored red for center and blue for origin. Figure 4. Open in new tabDownload slide Final conformations for the 36-chain (A), 36-chain (B), 24-chain (C), and 18-chain (D) microfibrils, viewed from the nonreducing end. Layers are colored red for center and blue for origin. The largest system considered in this work contained 36 chains in a 3, 4, 5, 6/6, 5, 4, 3 configuration (Fig. 4, A and B). This microfibril had 50% of its chains in the interior and 50% at the surface. Despite this, only 33% of exocyclic groups could be considered solvent exposed. This value was significantly less than 50% because, except for solvent-exposed chains on hydrophobic surfaces, every second exocyclic group in solvent-exposed chains points toward the center of the microfibril instead of out into the solvent. With an initial cross-sectional area of just under 1,200 Å2, the initial water content of the system was 77%. After equilibration, the water content came down to 73% due to a decrease in the total volume of the system and expansion of the microfibril (Table I). The next system considered consisted of 24 chains in a 3, 4, 5/5, 4, 3 arrangement (Fig. 4C). With 58% of its chains at the surface and 42% of exocyclic groups considered solvent exposed, the equilibrated water content was 71%. The smallest system simulated consisted of 18 chains in a 2, 3, 4/4, 3, 2 configuration, with an initial cross-sectional area of 590 Å2 (Fig. 4D). This system had 44% of its exocyclic groups solvent exposed (Table I), and its equilibrated water content was 74%. Microfibril properties Table I. Microfibril properties Simulation . Interior . Surface . Solventa Exposed . Initial . Equilibrated . Crossb Section . Water Content . Crossb Section . Water Content . % A2 v/v% A2 v/v% 18 33 67 44 588.2 80.2 680.2 74.0 24 42 58 42 769.8 76.6 858.1 71.1 36 (A) 50 50 33 1,188.6 76.7 1,261.3 73.3 36 (B) 1,279.4 72.9 Simulation . Interior . Surface . Solventa Exposed . Initial . Equilibrated . Crossb Section . Water Content . Crossb Section . Water Content . % A2 v/v% A2 v/v% 18 33 67 44 588.2 80.2 680.2 74.0 24 42 58 42 769.8 76.6 858.1 71.1 36 (A) 50 50 33 1,188.6 76.7 1,261.3 73.3 36 (B) 1,279.4 72.9 a Solvent-exposed exocyclic groups, assuming that all exocyclic groups on hydrophobic surfaces are solvent exposed.  bCross-sectional area in the xy plane. Open in new tab Table I. Microfibril properties Simulation . Interior . Surface . Solventa Exposed . Initial . Equilibrated . Crossb Section . Water Content . Crossb Section . Water Content . % A2 v/v% A2 v/v% 18 33 67 44 588.2 80.2 680.2 74.0 24 42 58 42 769.8 76.6 858.1 71.1 36 (A) 50 50 33 1,188.6 76.7 1,261.3 73.3 36 (B) 1,279.4 72.9 Simulation . Interior . Surface . Solventa Exposed . Initial . Equilibrated . Crossb Section . Water Content . Crossb Section . Water Content . % A2 v/v% A2 v/v% 18 33 67 44 588.2 80.2 680.2 74.0 24 42 58 42 769.8 76.6 858.1 71.1 36 (A) 50 50 33 1,188.6 76.7 1,261.3 73.3 36 (B) 1,279.4 72.9 a Solvent-exposed exocyclic groups, assuming that all exocyclic groups on hydrophobic surfaces are solvent exposed.  bCross-sectional area in the xy plane. Open in new tab A characteristic feature of the simulations was the tilting of the glucan chains in the hexagonal structures. Chain tilt occurred from the beginning of our simulations, consistent with observations in other MD simulations of cellulose using different force fields, crystalline models, and finite models (Bergenstråhle et al., 2007; Matthews et al., 2011a, 2012; Zhang et al., 2011; Chen et al., 2014). When microfibrils were viewed from the nonreducing end, origin layers had a clockwise tilt (positive) compared with the crystal structure, while the center layers had an anticlockwise tilt (negative; Fig. 4, blue and red colored, respectively). The 36-chain model experienced tilts of –4° and 7.5° for the center and origin layers, respectively, while the 18- and 24-chain models experience larger changes in tilt of approximately ± 10° (Table II). Degree of change of tilt for center and origin chains averaged over the last 50 ns of duplicate simulations Table II. Degree of change of tilt for center and origin chains averaged over the last 50 ns of duplicate simulations Positive number indicates clockwise tilt, and negative indicates anticlockwise tilt. Simulation . Center . Origin . ° 18 –9.29 10.01 24 –10.43 9.06 36 (A) –4.26 7.67 36 (B) –7.92 10.73 18 (O2-C2) –2.27 6.26 Simulation . Center . Origin . ° 18 –9.29 10.01 24 –10.43 9.06 36 (A) –4.26 7.67 36 (B) –7.92 10.73 18 (O2-C2) –2.27 6.26 Open in new tab Table II. Degree of change of tilt for center and origin chains averaged over the last 50 ns of duplicate simulations Positive number indicates clockwise tilt, and negative indicates anticlockwise tilt. Simulation . Center . Origin . ° 18 –9.29 10.01 24 –10.43 9.06 36 (A) –4.26 7.67 36 (B) –7.92 10.73 18 (O2-C2) –2.27 6.26 Simulation . Center . Origin . ° 18 –9.29 10.01 24 –10.43 9.06 36 (A) –4.26 7.67 36 (B) –7.92 10.73 18 (O2-C2) –2.27 6.26 Open in new tab In addition to the tilting of the chains within the microfibrils, significant changes also occurred to the conformation of exocyclic groups. This is highlighted in Figure 5, which separately tracks the changes to the exocyclic conformation occupancy for groups pointing to the left and right, when looking down the polymerization axis from the nonreducing end, over the length of the simulations for each chain. Additionally, Supplemental Figure S2 shows the exocyclic conformation for all Glc residues of the microfibril at different time points throughout the simulations. It was apparent that the first changes occur to exocyclic groups that were exposed to solvent, with the changes occurring during equilibration. For all microfibrils, there was a preference for the GT conformation to dominate for solvent-exposed groups, although there was a high degree of variation (Table III). For the 36-chain model, the GT:TG:GG ratio for solvent exposed groups was 50:35:15. This was similar for the 24- and 18-chain models, except that the GG proportion increases and TG decreases. Interestingly, it was found that there were different exocyclic conformation ratios for the two different hydrophilic surfaces, with the 110 surface having a higher occupancy for GT and lower occupancy for GG (Supplemental Table S1). The TG conformations were of similar occupancy for the two surfaces. Figure 5. Open in new tabDownload slide Plots of the exocyclic group conformation occupancies (TG in yellow, GG in blue, and GT in green) of each chain in the 36-chain (A), 36-chain (B), 24-chain (C), and 18-chain (D) models over the length of 200 ns simulations. Chains from center layers have plots with a red border, and origin layers have blue borders. Due to the different environments for the two exocyclic groups of each cellobiose unit, each chain has two plots; the left plot details those exocyclic groups that point to the left when looking down the microfibril polymerization axis from the nonreducing end, and the right plot details those that point to the right. Plots on the outside for each model represent exocyclic groups exposed to solvent, and these plots have high variance (as noted by the spiky nature of the plots) and sample all three conformations (all three colors are distinctly visible). For all other exocyclic groups, including those from surface chains but not exposed to the solvent, little variation is observed and generally one conformation (one color) dominates. Figure 5. Open in new tabDownload slide Plots of the exocyclic group conformation occupancies (TG in yellow, GG in blue, and GT in green) of each chain in the 36-chain (A), 36-chain (B), 24-chain (C), and 18-chain (D) models over the length of 200 ns simulations. Chains from center layers have plots with a red border, and origin layers have blue borders. Due to the different environments for the two exocyclic groups of each cellobiose unit, each chain has two plots; the left plot details those exocyclic groups that point to the left when looking down the microfibril polymerization axis from the nonreducing end, and the right plot details those that point to the right. Plots on the outside for each model represent exocyclic groups exposed to solvent, and these plots have high variance (as noted by the spiky nature of the plots) and sample all three conformations (all three colors are distinctly visible). For all other exocyclic groups, including those from surface chains but not exposed to the solvent, little variation is observed and generally one conformation (one color) dominates. Exocyclic conformation occupancies, expressed as a percentage, averaged over the last 50 ns of duplicate simulations Table III. Exocyclic conformation occupancies, expressed as a percentage, averaged over the last 50 ns of duplicate simulations Chain . 18 . 24 . 36 (A) . 36 (B) . 18 (O2-C2) . GG . GT . TG . GG . GT . TG . GG . GT . TG . GG . GT . TG . GG . GT . TG . Surface 23 54 23 24 52 24 17 49 34 21 52 27 8 19 72 Interior 46 12 42 47 7 47 7 2 91 39 28 33 0 2 98 Center 92 7 2 93 6 1 14 3 83 77 18 4 1 3 96 Origin 0 17 83 0 8 92 0 1 99 0 38 62 0 0 100 Total 36 30 34 37 26 37 10 18 72 33 36 31 4 9 87 Chain . 18 . 24 . 36 (A) . 36 (B) . 18 (O2-C2) . GG . GT . TG . GG . GT . TG . GG . GT . TG . GG . GT . TG . GG . GT . TG . Surface 23 54 23 24 52 24 17 49 34 21 52 27 8 19 72 Interior 46 12 42 47 7 47 7 2 91 39 28 33 0 2 98 Center 92 7 2 93 6 1 14 3 83 77 18 4 1 3 96 Origin 0 17 83 0 8 92 0 1 99 0 38 62 0 0 100 Total 36 30 34 37 26 37 10 18 72 33 36 31 4 9 87 Open in new tab Table III. Exocyclic conformation occupancies, expressed as a percentage, averaged over the last 50 ns of duplicate simulations Chain . 18 . 24 . 36 (A) . 36 (B) . 18 (O2-C2) . GG . GT . TG . GG . GT . TG . GG . GT . TG . GG . GT . TG . GG . GT . TG . Surface 23 54 23 24 52 24 17 49 34 21 52 27 8 19 72 Interior 46 12 42 47 7 47 7 2 91 39 28 33 0 2 98 Center 92 7 2 93 6 1 14 3 83 77 18 4 1 3 96 Origin 0 17 83 0 8 92 0 1 99 0 38 62 0 0 100 Total 36 30 34 37 26 37 10 18 72 33 36 31 4 9 87 Chain . 18 . 24 . 36 (A) . 36 (B) . 18 (O2-C2) . GG . GT . TG . GG . GT . TG . GG . GT . TG . GG . GT . TG . GG . GT . TG . Surface 23 54 23 24 52 24 17 49 34 21 52 27 8 19 72 Interior 46 12 42 47 7 47 7 2 91 39 28 33 0 2 98 Center 92 7 2 93 6 1 14 3 83 77 18 4 1 3 96 Origin 0 17 83 0 8 92 0 1 99 0 38 62 0 0 100 Total 36 30 34 37 26 37 10 18 72 33 36 31 4 9 87 Open in new tab Analysis of the exocyclic groups that were not solvent exposed indicates that the differing layers can sample different conformations. In origin chains, the TG conformation dominates, with occupancies greater than 80%. However, it was in the center chains where the relationship between an increase in GG sampling (and a corresponding decrease in TG) with decreasing numbers of chains was most evident. Simulations with the 36-chain model had a TG occupancy of 83% for center chains, while in the 18- and 24-chain simulations, significant changes were observed and the GG conformation now dominated with occupancies greater than 90% (Table III). The variability in interior chains was significantly reduced compared with the surface-exposed groups; however, it does appear to be smaller for chains closer to the center of the microfibril and when there were more chains in the microfibril. Apart from the dihedral angles that dictate the conformation of exocyclic groups, there were four other key dihedral angles. The ϕ and ψ angles determine the planarity of the β-glucan chains, while the O2-C2 (τ2) and O3-C3 (τ3) dihedrals affect possible interactions between the chains in the same layer of microfibrils (Fig. 1). In the crystal structure, the ϕ and ψ angles differed for origin and center chains, with ϕ and ψ being –88.7° and –147.1° for center chains and –98.5° and –142.3° for origin chains, respectively (Supplemental Tables S2 and S3). From the simulations, ϕ and ψ converge for the different layers to an average of –93° and –150°, respectively. This was consistent irrespective of whether a chain was in the origin or center layer or was solvent exposed or located in the interior. Given that we simulated an infinite microfibril, it was expected that the ϕ and ψ angles would not change markedly to keep the planar arrangement of chains. However, it was unexpected that they would equilibrate to the same value considering the differences that had been observed in the crystal structures. From neutron diffraction studies (Nishiyama et al., 2002), it was suggested that hydrogens bonded to O2 and O3 were oriented in opposite directions (trans compared with the C2-C3 bond), and thus the initial dihedral angles about these O-C bonds were close to 180°. This was in contrast to a more cis-like conformation where the angle would be approximately 60° (Fig. 1). Similarly to the ϕ and ψ dihedrals, there was little variation about the τ3 dihedral when compared with the crystal structure, with an average angle of approximately 161° for all simulations (Supplemental Table S4). This lack of variation resulted from the stabilizing O3-to-O5 intrachain H bond (see below). There were far greater changes observed for the τ2 dihedral angle (Table IV). For the 36-chain simulations, the interior chains maintained their trans-like conformation (approximately 158°), while there was a drop in the angle for surface exposed chains (approximately 100°), suggesting that cis conformations were being sampled more regularly. For the 18- and 24-chain simulations, there were further reductions in the dihedral angle, most significantly for center chains. The dihedrals here were less than 60°, which suggests that only cis conformations were being sampled. Reductions in the dihedral angle of origin chains, though to a lesser extent for the 24-chain model, suggest more cis-like conformations were being sampled. O2 dihedral angles from diffraction data and averaged over the last 50 ns of duplicate simulations Table IV. O2 dihedral angles from diffraction data and averaged over the last 50 ns of duplicate simulations Simulation . Center . Origin . Surface . Crystal . Total . ° Iβ (Crystal) 166.1 165.3 165.7 18 58.9 111.8 85.3 90.9 85.4 24 56.3 124.7 81.6 100.6 90.5 36 (A) 117.3 140.5 100.2 157.6 128.9 36 (B) 52.8 83.9 76.4 60.2 68.3 18 (O2-C2) 172.3 173.2 172.5 173.2 172.7 Simulation . Center . Origin . Surface . Crystal . Total . ° Iβ (Crystal) 166.1 165.3 165.7 18 58.9 111.8 85.3 90.9 85.4 24 56.3 124.7 81.6 100.6 90.5 36 (A) 117.3 140.5 100.2 157.6 128.9 36 (B) 52.8 83.9 76.4 60.2 68.3 18 (O2-C2) 172.3 173.2 172.5 173.2 172.7 Open in new tab Table IV. O2 dihedral angles from diffraction data and averaged over the last 50 ns of duplicate simulations Simulation . Center . Origin . Surface . Crystal . Total . ° Iβ (Crystal) 166.1 165.3 165.7 18 58.9 111.8 85.3 90.9 85.4 24 56.3 124.7 81.6 100.6 90.5 36 (A) 117.3 140.5 100.2 157.6 128.9 36 (B) 52.8 83.9 76.4 60.2 68.3 18 (O2-C2) 172.3 173.2 172.5 173.2 172.7 Simulation . Center . Origin . Surface . Crystal . Total . ° Iβ (Crystal) 166.1 165.3 165.7 18 58.9 111.8 85.3 90.9 85.4 24 56.3 124.7 81.6 100.6 90.5 36 (A) 117.3 140.5 100.2 157.6 128.9 36 (B) 52.8 83.9 76.4 60.2 68.3 18 (O2-C2) 172.3 173.2 172.5 173.2 172.7 Open in new tab To understand how the changes in the τ2 dihedral allow for the sampling of different exocyclic conformations, we also analyzed changes to the H-bonding patterns. For all these systems, the simulations were started with the microfibrils in the A H-bonding pattern. In this pattern, intrachain H bonds are formed where O2 donates its hydrogen to O6 and O3 to O5 (Fig. 1). Intralayer H bonds exist from O6 to O3, with those from O6 to O2 only found for center layers, while there were no interlayer H bonds. The 36-chain microfibril showed limited changes to its H bonding throughout the simulations and can thus be thought to only sample the A H-bonding pattern (Table V). The O3-to-O5 intrachain H bonds continue to have strong occupancy, as does the O2-to-O6 intrachain H bonds. The O6-to-O3 intralayer H bond was observed to have high occupancy, coexisting with the O6-to-O2 intralayer H bond that formed in origin chains to a similar moderate level as those in center chains. The O6-to-O2 intrachain and O2-to-O6 intralayer H bonds are representative of the B H-bond pattern and were sampled to a low extent, generally due to chains at the surface of the microfibril (Supplemental Fig. S3). There were only sporadic interlayer H bonds, mainly from surface chains to interior chains (Supplemental Table S5). Occupancy of hydrogen bonds averaged over the last 50 ns of duplicate simulations Table V. Occupancy of hydrogen bonds averaged over the last 50 ns of duplicate simulations Simulation . Chain . Intralayer . Intrachain . O6→O3a . O6→O2a . O2→O6b . O3→O5 . O2→O6a . O6→O2b . % 18 Center 3 2 89 86 1 2 Origin 58 24 39 89 35 25 24 Center 4 2 87 89 1 19 Origin 68 3 29 90 49 18 36 (A) Center 58 28 2 87 6 1 Origin 87 41 1 91 7 13 36 (B) Center 1 0 91 88 1 23 Origin 25 6 8 91 12 4 18 (O2-C2) Center 71 37 0 8 84 0 Origin 98 44 0 88 9 0 Simulation . Chain . Intralayer . Intrachain . O6→O3a . O6→O2a . O2→O6b . O3→O5 . O2→O6a . O6→O2b . % 18 Center 3 2 89 86 1 2 Origin 58 24 39 89 35 25 24 Center 4 2 87 89 1 19 Origin 68 3 29 90 49 18 36 (A) Center 58 28 2 87 6 1 Origin 87 41 1 91 7 13 36 (B) Center 1 0 91 88 1 23 Origin 25 6 8 91 12 4 18 (O2-C2) Center 71 37 0 8 84 0 Origin 98 44 0 88 9 0 a A-type H bond.   b B-type H bond. Open in new tab Table V. Occupancy of hydrogen bonds averaged over the last 50 ns of duplicate simulations Simulation . Chain . Intralayer . Intrachain . O6→O3a . O6→O2a . O2→O6b . O3→O5 . O2→O6a . O6→O2b . % 18 Center 3 2 89 86 1 2 Origin 58 24 39 89 35 25 24 Center 4 2 87 89 1 19 Origin 68 3 29 90 49 18 36 (A) Center 58 28 2 87 6 1 Origin 87 41 1 91 7 13 36 (B) Center 1 0 91 88 1 23 Origin 25 6 8 91 12 4 18 (O2-C2) Center 71 37 0 8 84 0 Origin 98 44 0 88 9 0 Simulation . Chain . Intralayer . Intrachain . O6→O3a . O6→O2a . O2→O6b . O3→O5 . O2→O6a . O6→O2b . % 18 Center 3 2 89 86 1 2 Origin 58 24 39 89 35 25 24 Center 4 2 87 89 1 19 Origin 68 3 29 90 49 18 36 (A) Center 58 28 2 87 6 1 Origin 87 41 1 91 7 13 36 (B) Center 1 0 91 88 1 23 Origin 25 6 8 91 12 4 18 (O2-C2) Center 71 37 0 8 84 0 Origin 98 44 0 88 9 0 a A-type H bond.   b B-type H bond. Open in new tab The 18- and 24-chain models sample very similar H-bonding patterns to each other that can be classified as being in both the A and B H-bonding patterns (Table V). As was expected, the O3-to-O5 intrachain H bond has very high occupancy. With regard to H bonds that were representative of the A pattern, the O2-to-O6 H bond had similar occupancy to the 36-chain model for origin chains but had almost no occupancy in center chains. This relationship was also observed for other A intralayer H bonds where the O6-to-O3-and-O2 occupancies for the origin chains were similar to those of the 36-chain microfibril, while those in center chains were nonexistent. Instead, in the center chains, occupancy was greater for those H bonds that represent the B pattern, such as the intrachain O2-to-O6 and intralayer O2-to-O6 H bonds. Interestingly, there was some occupancy in the origin chains for these B-pattern H bonds. In what was a significant difference to the 36-chain simulations, there were significant interlayer H bonds for the 18- and 24-chain models, specifically for the O6-to-O2 1-10 layer, with O6 in a center chain and O2 in an origin chain (Supplemental Table S5). This type of H bond was only found to occur when the exocyclic group was in the GG conformation. Initial expansion of the microfibrils can be evidenced by limited increases to the unit cell parameters (Table VI). The majority of changes were below 2%, with the exception being the a-unit cell length (Fig. 3), which could differ by almost 5%. This discrepancy results from a change in the distance between 200/100 layers. In the crystal form, this distance was 7.78 Å, while from the simulations, the distance increases to 7.94 Å for the 36-chain model and 8.15 Å for the 18- and 24-chain models. This increase correlates with WAXS data (Newman et al., 2013), where the lattice spacing between (200) planes increased to 0.41 nm (8.2 Å between layers). The largest change in unit cell angle parameters occurs for the γ angle and represents a slight slippage between 200 layers; the top layer moves to the left and the bottom layer to the right when viewed from the nonreducing end. Unit cell distance and angles from diffraction studies and averaged over the last 50 ns of duplicate simulations Table VI. Unit cell distance and angles from diffraction studies and averaged over the last 50 ns of duplicate simulations Chains . a . b . c . α . β . γ . Å o Iβ (Crystal) 7.78 8.20 10.38 90.00 90.00 96.50 Iβ (203°C)a 8.19 8.18 10.37 90.00 90.00 96.40 18 8.15 8.25 10.44 91.52 89.21 98.47 24 8.14 8.25 10.44 90.70 90.20 97.83 36 (A) 7.94 8.35 10.44 90.41 90.35 98.41 36 (B) 8.06 8.40 10.43 93.77 90.00 97.79 18 (O2-C2) 7.95 8.33 10.45 90.24 90.43 98.46 Chains . a . b . c . α . β . γ . Å o Iβ (Crystal) 7.78 8.20 10.38 90.00 90.00 96.50 Iβ (203°C)a 8.19 8.18 10.37 90.00 90.00 96.40 18 8.15 8.25 10.44 91.52 89.21 98.47 24 8.14 8.25 10.44 90.70 90.20 97.83 36 (A) 7.94 8.35 10.44 90.41 90.35 98.41 36 (B) 8.06 8.40 10.43 93.77 90.00 97.79 18 (O2-C2) 7.95 8.33 10.45 90.24 90.43 98.46 a Wada et al. (2010). Open in new tab Table VI. Unit cell distance and angles from diffraction studies and averaged over the last 50 ns of duplicate simulations Chains . a . b . c . α . β . γ . Å o Iβ (Crystal) 7.78 8.20 10.38 90.00 90.00 96.50 Iβ (203°C)a 8.19 8.18 10.37 90.00 90.00 96.40 18 8.15 8.25 10.44 91.52 89.21 98.47 24 8.14 8.25 10.44 90.70 90.20 97.83 36 (A) 7.94 8.35 10.44 90.41 90.35 98.41 36 (B) 8.06 8.40 10.43 93.77 90.00 97.79 18 (O2-C2) 7.95 8.33 10.45 90.24 90.43 98.46 Chains . a . b . c . α . β . γ . Å o Iβ (Crystal) 7.78 8.20 10.38 90.00 90.00 96.50 Iβ (203°C)a 8.19 8.18 10.37 90.00 90.00 96.40 18 8.15 8.25 10.44 91.52 89.21 98.47 24 8.14 8.25 10.44 90.70 90.20 97.83 36 (A) 7.94 8.35 10.44 90.41 90.35 98.41 36 (B) 8.06 8.40 10.43 93.77 90.00 97.79 18 (O2-C2) 7.95 8.33 10.45 90.24 90.43 98.46 a Wada et al. (2010). Open in new tab More information about the changes that occur to the structure of the microfibril was extracted from the simulations by calculating changes in the distance between hydrophobic (100/200) and hydrophilic (1-10/110) layers (Table VII). The average 100/200 distance increased by 0.1 Å (2.6%) for the 36-chain model and by 0.2 Å (4.7%) for the 18- and 24-chain models, whereas distances between 110 layers increased by under 0.1 Å, with the greatest change (>0.2 Å) occurring for the 1-10 layers. The extra increase for the 1-10-layer distances was not unexpected given the increase in the γ unit cell angle. These average layer spacings correlate as well, if not better, than the crystal structure with WAXS (Newman et al., 2013) and x-ray diffraction (Nishiyama, 2009) data, where spacings were calculated to be 4.1, 5.6, and 5.6/6.0 Å for the 100/200,110, and 1-10 planes, respectively. Overall, the changes in these distances, as well as the changes to the unit cell parameters, present themselves as an anticlockwise rotation of the microfibril about the polymerization axis (looking from the nonreducing end) compared with the crystal structure. Additionally, we have measured the change in the distance along the polymerization axis of an origin chain compared with a center chain. This gives a measure of slippage of chains along the polymerization axis, and from these simulations, an average change of 0.17 Å for the 36-chain simulation was observed, while this increased to 0.26 Å for the 18- and 24-chain simulations. Given that there was little change in the α- and β-unit cell angles, the slippage described here results solely from slippage between origin and center chains and not from slippage across the entire microfibril. Distance between layers and chain shift along the polymerization axis from diffraction data and averaged over the last 50 ns of duplicate simulations Table VII. Distance between layers and chain shift along the polymerization axis from diffraction data and averaged over the last 50 ns of duplicate simulations Chains . 100 . 110 . 1-10 . Z . Å Iβ (Crystal) 3.84 5.31 5.97 2.71 18 4.01 5.39 6.23 2.96 24 4.04 5.40 6.18 2.98 36 (A) 3.94 5.36 6.22 2.88 36 (B) 4.00 5.42 6.21 2.93 18 (O2-C2) 3.83 5.34 6.18 2.89 Chains . 100 . 110 . 1-10 . Z . Å Iβ (Crystal) 3.84 5.31 5.97 2.71 18 4.01 5.39 6.23 2.96 24 4.04 5.40 6.18 2.98 36 (A) 3.94 5.36 6.22 2.88 36 (B) 4.00 5.42 6.21 2.93 18 (O2-C2) 3.83 5.34 6.18 2.89 Open in new tab Table VII. Distance between layers and chain shift along the polymerization axis from diffraction data and averaged over the last 50 ns of duplicate simulations Chains . 100 . 110 . 1-10 . Z . Å Iβ (Crystal) 3.84 5.31 5.97 2.71 18 4.01 5.39 6.23 2.96 24 4.04 5.40 6.18 2.98 36 (A) 3.94 5.36 6.22 2.88 36 (B) 4.00 5.42 6.21 2.93 18 (O2-C2) 3.83 5.34 6.18 2.89 Chains . 100 . 110 . 1-10 . Z . Å Iβ (Crystal) 3.84 5.31 5.97 2.71 18 4.01 5.39 6.23 2.96 24 4.04 5.40 6.18 2.98 36 (A) 3.94 5.36 6.22 2.88 36 (B) 4.00 5.42 6.21 2.93 18 (O2-C2) 3.83 5.34 6.18 2.89 Open in new tab The final characteristic measured was the degree of water structuring surrounding the microfibrils. Densities were measured above all faces as a function of distance away from the layer. The water structuring was very similar for each model, and for the most part, there appeared to be two peaks in each layer’s density plots. However, it was evident that water was structured differently above the different layers (Fig. 6A; Supplemental Figs. S4A–S6A). Water was least structured above the 1-10 surface, as evidenced by the much smaller initial peak in the density curves. Increased structuring was observed over the 110 surface, with the most significant structuring occurring over the hydrophobic (100/200) surfaces. With the densities measured starting from the heavy atoms that were most exposed to the surface (Fig. 6B), it was apparent that there was water density preceding these heavy atoms for the hydrophilic surfaces. This was due to the apparent tilting of the chains at these surfaces, which allowed water molecules to position themselves in the gaps between the center and origin layers (Supplemental Fig. S7). By analyzing the densities starting at the middle of the surface layers, we can deduce that the initial primary peaks in density start between 3.5 and 4.5 Å away from the middle of the layers. Additionally, minima in density between the primary and secondary peaks occur between 5 to 6 Å. Figure 6. Open in new tabDownload slide Relative water densities about the 100 (purple), 110 (orange), and 1-10 (brown) layers for the 36 A simulations as a function of distance above the top heavy atom of the layer of interest in A, above the middle of the layer in B, and along the polymerization axis in C. Figure 6. Open in new tabDownload slide Relative water densities about the 100 (purple), 110 (orange), and 1-10 (brown) layers for the 36 A simulations as a function of distance above the top heavy atom of the layer of interest in A, above the middle of the layer in B, and along the polymerization axis in C. Another approach to analyze the water structuring is to measure the density as a function of distance along the polymerization axis (Fig. 6C; Supplemental Figs. S4C–S6C). From these plots, it was notable that, for the hydrophobic surfaces, there were two peaks per cellobiose repeat, while there were four for the hydrophilic surfaces. These peaks coincide with glycosidic linkages exposed on the surface. The reason that four peaks were found for the hydrophilic surfaces was that these surfaces were composed of both center and origin layers, and because the center layers were displaced by approximately 2.6 Å compared with the origin layers, we see peaks with this separation. From the analysis of these simulations it was clear that the 18- and 24-chain microfibrils were sampling a different conformational space to the 36-chain microfibril. This conformational space appeared to be similar to high-temperature structures (I-HT) simulated by Matthews et al. (2011a). Subsequently, a question that needed to be answered was what is causing these changes in the conformation of the microfibrils. While it appeared that the number of chains in the microfibril was having an effect on the simulations, it was thought it would be informative to pinpoint the changes that occurred in the 18- and 24-chain systems yet did not occur in the 36-chain model. To investigate this, two further sets of simulations were performed; one with a different initial H-bonding scheme and the other with the τ2 dihedral restrained to its initial value. 36-Chain B As noted above, a major difference between surface-exposed and interior chains for the 36-chain simulations and throughout the microfibril for simulations with smaller numbers of chains was the τ2 dihedral angle. There was a dramatic change such that the O2 hydroxyl no longer sat in the crystal-like conformation pointing away from O3 of the same residue but now pointed toward it. This conformation was similar to one sampled by Matthews et al. (2011a), who found that at high temperature, and when simulating for long time scales, this τ2 dihedral would change. Moreover, this also aligns with some uncertainty in the diffraction data of Nishiyama et al. (2002) for which two potential H-bond schemes were suggested. We therefore performed simulations on the 36-chain model with the B H-bond scheme (Fig. 1), based on the proposed structures from the x-ray diffraction work (Nishiyama et al., 2002). In these simulations, we changed the τ2 dihedral such that the simulations started with the O2 hydroxyl group pointed toward the O3 atom of the same Glc residue (i.e. the cis instead of the trans orientation). We also rotated the position of the HO6 (hydrogen of exocyclic hydroxyl group) so that there were no steric clashes, while the exocyclic group was kept in the TG conformation. These changes resulted in the O2-to-O6 H bond being the only interchain H bonds existing at the beginning of the simulation. Many of the dynamic characteristics of this model were similar to those of the 18- and 24-chain models. The tilt of chains was very similar (Table II), as were the unit cell parameters (Table VI) and the distances between layers (Table VII). The τ2 dihedral from center and solvent-exposed chains was similar to those seen for previous simulations; however, the origin and interior chains had greatly reduced dihedrals and were therefore much more cis like (Table IV). These chains did not revert to the trans-like conformation of the A H-bonding pattern model as could be expected given the stability shown in the 36 A H-bond simulations. As expected, by starting these simulations in a different H-bond pattern, an entirely different sampling of H bonds and exocyclic group conformations was observed when compared with the 36 A simulations (Table III; Fig. 5). However, the conformations sampled did resemble those sampled with the 18- and 24-chain models, though with some small differences. There was still some A character to origin chains for the O2-to-O6 intrachain and O6-to-O3 intralayer H bonds, but this was significantly reduced. It was also noted that the occupancies for H bonds that are typical of the B pattern were far higher. Importantly, interlayer H bonds were observed between O6 and O2 for 1-10 layers at a similar occupancy to the 18- and 24-chain models. With regards to exocyclic group conformation, center chains had a greater likelihood to sample GG, while solvent-exposed chains more often sampled GT. Interestingly, origin chains were able to sample the GT conformation to a greater extent, although TG was still the preferred exocyclic conformation for these chains. 18-Chain τ2 Restrained From the simulations performed thus far, three major events took place to move the microfibril from the crystal phase to the so-called I-HT phase in the 18, 24, and 36 B simulations. These were the opposing tilts of center and origin layers, reduced sampling of the TG exocyclic conformation, and the rotation of the τ2 dihedral away from the trans conformation. The effect of this final event can be investigated by restraining the dihedral angle to the trans conformation as observed in the crystal structure, and this was performed for the 18-chain model. Unsurprisingly, there were limited changes that took place to the structural characteristics of the microfibril compared with the initial structure. There were limited changes to dihedral angles, which was to be expected given that the only dihedral angle to show significant variation in the other simulations was the τ2 dihedral, and this was now restrained. Exocyclic groups not exposed to solvent were found to reside in the TG conformation for 98% of the simulation. There was also limited change in the exocyclic conformations of surface-exposed groups, with 72% still in the TG conformation. This suggested a link between τ2 dihedral rotation and variation in exocyclic conformation. Characteristics that showed slight changes, but only such that they sample values as observed in the 36 A simulations, were the unit cell parameters, chain shifts, and H-bond occupancies. One specific characteristic that showed significant changes compared with the initial structure, but still similar to values observed in the 36 A simulations, was the tilt of center and origin layers (Table II). As this tilt occurs, we can suggest that the tilt is not dependent on τ2 rotation. However, as the tilt was reduced compared with the other simulations, it can be used as further evidence to suggest that tilt is a two-stage process, with the first stage innate to all structures and the second stage being dependent on rotation of the τ2 dihedral. DISCUSSION We have performed MD simulations on three different models of Iβ cellulose microfibrils to investigate the structural and dynamical dependence on the number of chains in the microfibril. Models with 18, 24, and 36 chains were produced, and the smaller 18- and 24-chain models were shown to sample a conformational space different to that of the initial crystalline phase and any phase identified experimentally, but similar to that observed in high-temperature MD simulations. Here, we discuss why particular exocyclic group conformations are sampled, why this depends on a second stage of tilt and dihedral τ2 rotation, and why it only occurs for smaller microfibrils. We also compare the results of our simulations with other computational studies and the available experimental data. Differences in Microfibril Structure It was clearly evident that the simulations performed with 18- and 24-chain microfibrils, and those with 36 chains in the B H-bonding pattern, quickly moved from the crystalline conformation to one more reminiscent of the I-HT structures produced in the simulations of Matthews et al. (2011a). This was, however, different from the 36-chain simulations with the A H-bonding pattern in which the microfibrils stayed close to the crystalline phase. One of the major differences between the simulations was the sampling of the exocyclic group conformations. The 36-chain A simulations predominantly stayed in the TG conformation for interior chains, while the other simulations sampled the GG conformation if the chain belonged to a center layer and the traditional TG if in the origin layer. Exocyclic groups that were exposed to solvent behave in a similar manner no matter the number of chains in the microfibril. The predominant conformation for these groups was GT, with the next favored conformation dependent on the chains layer and following the same pattern as observed for interior chains. The ability of the 18- and 24-chain simulations to sample different exocyclic conformations compared with the 36-chain A model was dependent on extra tilt and the rotation of the τ2 dihedral. In all simulations, an initial stage of tilt was observed, with chains in the origin layer tilting in a clockwise manner and those in center layers exhibiting an anticlockwise tilt when looking down the microfibril from the nonreducing end. Tilt was able to occur in two different directions as a result of the shift of cellobiose repeat units along the polymerization axis between origin and center chains that positions the chains in different environments. This becomes apparent by analyzing the environment surrounding a cellobiose repeat from the different chains (Fig. 7). Firstly, it was noted that the chains above and below a center chain are oriented differently. For the first (Fig. 7A, bottom) residue of a center chain, the glycosidic linkage in the origin chains above and below the exocyclic group point down. This means that the hydrogens attached to the C1 and C4 carbons of the glycosidic linkage point upwards. This produces a shallow cavity above the exocyclic group and extra volume below it. For the second (Fig. 7A, top) residue of the cellobiose repeat, the shallow cavity is below the exocyclic group and extra volume above it. The extra volume of the hydrogens places steric strain on the alternating exocyclic groups, which is alleviated by a slight anticlockwise tilt to the chain. For the origin chain cellobiose repeat, the opposite orientation of glycosidic linkages occurs in the center chains above and below, and as a result, we see a clockwise tilt to alleviate the strain due to the hydrogens. Figure 7. Open in new tabDownload slide Initial structures of center chain (in licorice) above an origin layer (surface) in A and origin chain (licorice) above a center layer in B, colored by atom type (hydrogen, gray; oxygen, red; and carbon, cyan). This figure highlights the slight cavity below the exocyclic group in the top Glc of the center chain in A and the bottom Glc of the origin chain in B, while also highlighting the raised volume below the exocyclic group in the bottom Glc of the center chain in A and the top Glc of the origin chain in B. It is this steric environment that is responsible for the clockwise tilt in the origin chains and the anticlockwise tilt in the center chains that occurs early. Figure 7. Open in new tabDownload slide Initial structures of center chain (in licorice) above an origin layer (surface) in A and origin chain (licorice) above a center layer in B, colored by atom type (hydrogen, gray; oxygen, red; and carbon, cyan). This figure highlights the slight cavity below the exocyclic group in the top Glc of the center chain in A and the bottom Glc of the origin chain in B, while also highlighting the raised volume below the exocyclic group in the bottom Glc of the center chain in A and the top Glc of the origin chain in B. It is this steric environment that is responsible for the clockwise tilt in the origin chains and the anticlockwise tilt in the center chains that occurs early. This explains why, in all simulations, an initial tilt of approximately 5° in either direction was observed and sets up each system for a second stage of tilt. For this second stage to occur, the intrachain H bond from O2 to O6 must break. In the 36-chain A and 18-chain O2-C2 restrained simulations, this H bond did not break, and as a result, the second stage of tilt was not observed. As this H bond was the major interaction that kept the exocyclic group in the TG conformation, these two simulations seldom sample conformations other than TG in interior chains. Additionally, the a-unit cell distance was smaller for these simulations, suggesting that the layers must increase their distance between each other to allow for the second stage of tilt. The link between τ2 dihedral and exocyclic conformation was quite apparent. From the τ2 restrained simulations, there was no rotation of the τ2 dihedral, and as a result, almost no change to the exocyclic conformation. However, in all other simulations, coinciding with the rotation of τ2 dihedral into a cis-like conformation, the O2-to-O6 intrachain H bond broke. As a result, the exocyclic group was free to rotate, and a second stage of tilt was observed. Despite this, it was predominantly only in the center chains where changes in the exocyclic group conformation were observed for interior chains. We suggest this was due to the steric restraints/freedoms placed on the exocyclic groups by the initial tilt of chains, causing steric clashes with exocyclic rotations of certain directions. For center chains, the TG-GG rotation was made easier by the anticlockwise tilt. With the τ2 dihedral in a cis-like conformation and, as a result, the O2-to-O6 intrachain H bond being broken, the exocyclic groups of center chains could easily rotate from TG to GG. This rotation was further stabilized by the second stage of tilt and the forming of interlayer H bonds, specifically the O6-to-O2 1-10 layer H bonds. Stabilization was so strong that conversion back to TG was not observed. With the clockwise tilt of origin chains, the TG-to-GG rotation was hindered, while the TG-to-GT rotation was more favored. However, unlike the TG and GG conformation, there are no other interactions to stabilize the GT conformation in the interior chains and that is why the TG conformation still dominates. It should be noted that solvent-exposed exocyclic groups were not hindered in any way and could sample all conformations, with GT being of lowest energy. Rotation of the τ2 dihedral was observed to a high degree for the 18- and 24-chain simulations but was limited for 36 A chain simulations. The only difference between the simulations was the size of the microfibrils and, as a result, the greater the influence that the surrounding solvent had on the dynamics of each chain. It was clear from the plots of the exocyclic conformation (Fig. 5) that there was far more variation the closer a chain was to the solvent. The solvent must affect the dynamics of the exocyclic group and the τ2 dihedral. Increased fluctuations led to increased chances of the O2-to-O6 intrachain H bond breaking and eventual freeing of the exocyclic group to sample different conformations. It was evident from Supplemental Figure S2 that these changes filter from the outside chains inwards, and as there are less interior chains in the 18- and 24-chain models, the changes filter throughout the entire microfibril. The 36-chain microfibril has enough interior chains that these changes were not able to infiltrate the microfibril. This is shown in Figure 5A for the center chains of the 36-chain A microfibril, where there was greater variation in the exocyclic conformation with a consistent GG conformation but at low levels, preventing adoption of a 100% TG exocyclic conformation. Comparison with Previous Experimental and in Silico Data The simulations presented here resemble those performed on 36-chain microfibrils with three different force fields (Matthews et al., 2012) and those that heated the microfibrils to 500 K, which resulted in a so-called I-HT conformation (Matthews et al., 2011a). Though their simulations were performed on diamond-shaped finite microfibrils, our 18- and 24-chain simulations with infinite hexagonally shaped microfibrils were able to sample a similar I-HT conformation at simulated ambient temperature with regards to the tilt, exocyclic group conformation, and H-bond patterns. By contrast, our 36-chain model simulations sampled the more crystalline-like Iβ conformation and not the I-HT conformation at ambient temperature. The microfibrils in the Matthews et al. (2012) work were modeled as finite microfibrils and, as a result, were solvated in all directions, including at the reducing and nonreducing end, and thus had significantly more solvent exposure than our 36-chain model. Given the reduced degree of polymerization of these microfibrils, we suggest these models will be oversolvated compared with natural microfibrils, and this is a potential limitation of their simulations. As mentioned above, we believe the reason our 18/24- and 36-chain simulations sampled a different conformational space was the difference in degree of solvent exposure, and we propose that this was the reason that the Matthews et al. (2012) 36-chain simulations also sampled the I-HT conformation. From Matthews et al. (2012), specifically the simulations performed with the Charmm and Glycam force fields, we note that changes away from the crystalline TG conformation occur from the outside of the microfibril inwards. This correlates with what was observed in our 18- and 24-chain simulations, suggesting that changes to structure occur initially at solvent-exposed residues and then filter through the microfibril. With our 36-chain simulations, the decreased solvent exposure and higher number of interior chains means that the changes were not able infiltrate the entire microfibril. It was evident from the exocyclic conformation occupancy plots that the favored conformation for solvent-exposed groups was predicted to be GT. This goes against both experimental and computational studies of Glc and cello-oligosaccharides in solution where GG is reported to be the favored form (Nishida et al., 1984; Cramer and Truhlar, 1993; Bock and Duus, 1994; Kirschner and Woods, 2001; Kuttel et al., 2002; Pereira et al., 2006; Shen et al., 2009). In these studies, the general consensus is that exocyclic groups have a distribution of about 55:40:5 for the GG:GT:TG conformations, respectively. With the NMR and x-ray data suggesting that TG was the only conformation in a cellulose microfibril, we further questioned this preference for GT. However, there is some evidence to suggest that the simulations presented here are sampling the correct exocyclic conformation for these solvent-exposed groups. Recent experimental data suggests that GT may be more populated than GG for Glc in solution (Hansen and Hünenberger, 2011). X-ray data for cellobiose has shown that the two exocyclic groups in the crystal structure adopt the GT conformation (Chu and Jeffrey, 1968). Additionally, atomic force microscopy images and NMR data suggest that exocyclic groups at surfaces have the GT conformation (Baker et al., 2000). Finally, other MD simulations have also seen the GT conformation favored at the surface (Matthews et al., 2012). Thus, it appears that when Glc is formed into a polymer, and does not reside in an environment where it can interact with other chains, GT is its favored conformation, though with low-energy barriers such that significant variation between conformers is likely. Experimental work suggests that the majority of the exocyclic groups of cellulose reside in the TG conformation. X-ray and neutron diffraction data suggest that this is the case for all crystalline chains. NMR data has generally been interpreted to suggest that chains in the interior of microfibril aggregates have a TG exocyclic conformation, while those that were solvent exposed were either GG or GT. The simulations performed here with the 36-chain model correlate well with this interpretation of the experimental data. However, as mentioned previously, simulations with the 18- and 24-chain models move away from the Iβ phase to a I-HT conformational space, which does not fit with the experimental data. However, we suggest that it is possible that the NMR data are being interpreted incorrectly. It is well documented that the dual C4 peaks are separately due to the C4 carbons from solvent-exposed and interior chains (Ha et al., 1998; Newman, 1998). Given the almost identical sizes of the double peaks for C4 and C6 in cellulose, it is commonly thought that the upfield peaks result from the contribution of carbons in surface chains and the downfield peaks from interior chains (Viëtor et al., 2002). To make these conclusions, two assumptions must be made: all C4 and C6 carbons on surface chains will have the same chemical shifts, and exocyclic groups are restricted to GG and GT for surface and TG for interior. We have doubts about the validity of these assumptions. Given the 2-fold axis of cellulose, there is potential for the equivalent carbons on each cellobiose repeat to have two different chemical shifts. If we focus on surface chains, the environment around the two C4s is very similar, with comparable exposure to solvent (Fig. 3A, inset). However, the two C6 carbons and their attached hydroxyl groups exist in different environments. One points directly out into the solvent, while the other is positioned away from the solvent, directly interacting with another cellulose chain, and can therefore be thought of as actually being an interior exocyclic group. Thus, the two different C6 carbons should have different chemical shifts, with the nonsolvent-exposed group having a similar shift to the interior C6 and thus contributing to the downfield peak. As a result, we should characterize these chains, and the atoms of interest, with regards to their solvent exposure instead of whether they are on surface or interior chains. As we have identified that it is solvent exposure that we need to focus on to analyze NMR spectra, the second assumption should now be that exocyclic groups that are solvent exposed are restricted to the GG and GT conformations, while all other exocyclic groups are TG. If this is true, then for the 36-chain microfibril, the upfield C6 peak that represents all exocyclic groups that are solvent exposed should be 33% of the total C6 peak (Table I). However, the C4 peak will have a 50% contribution to the total C4 peak, and this does not correlate with the fact that the two peaks should be of similar size. Using the restriction assumption, the C6 upfield peak sizes for the 18- and 24-chain models are not similar to the C4 upfield peaks either (44% versus 67% and 42% versus 58% for 18 and 24 chains, respectively). Under the second assumption, none of the models fit the NMR data, suggesting that this assumption is also invalid. Our simulations provide more evidence that this assumption is invalid and also suggest that the solvent-exposed exocyclic groups are not restricted to GT or GG, while interior groups are not restricted to TG. Therefore, we suggest that the percentage of exocyclic groups in the GT and GG conformation be a measure of the size of the upfield C6 peak, as is suggested by Horii et al. (1983). Using this criterion, we find that both the 18- and 24-chain models give C6 upfield peaks that would be of a very similar size to the C4 upfield peak. For the 24-chain microfibril, the C6 peak would be 63%, which compares favorably with the C4 surface peak of 58%, with the 18-chain microfibrils giving even better correlation with the C6 peak predicted to be 66% and the C4 peak predicted to be 67%. Additionally, the data for the 18- and 24-chain microfibrils correlate well with an NMR spectrum of sugar beet (Beta vulgaris) cellulose from parenchymatous primary cell walls, which was interpreted as having 54% of its chains located at the surface (Viëtor et al., 2002). Thus, using this new interpretation of the NMR data, it appears that the size of the C6 upfield peak is dictated by the percentage of exocyclic groups in the GT and GG conformations, and these conformations are not restricted to just surface chains or solvent-exposed groups. Further evidence to support the theory that the elementary microfibril consists of either 18 or 24 chains comes from scattering and diffraction work (Fernandes et al., 2011; Newman et al., 2013). Data from these recent WAXS, wide-angle neutron scattering, and small-angle neutron scattering studies on mung bean (Vigna radiata) primary cell walls and spruce (Picea sitchensis) wood secondary cell walls suggest that microfibril diameters should be about 3 nm and that there should be six layers of 100 planes per microfibril. As a result, the microfibril should have a cross-sectional area of around 7 nm2. After the expansion that was observed by performing these simulations at 300 K in a hydrated environment, the microfibrils had cross-sectional areas of 6.8, 8.6, and 12.6 nm2 for the 18-, 24-, and 36-chain microfibrils, respectively. This suggests that the 36-chain microfibril is too big and that 18 or 24 chains are more credible fits to the recent experimental data. Given the results with the 36-chain model and all other experimental x-ray diffraction work, we still believe that cellulose fibers of larger diameter have a crystalline core. However, the results with the 18- and 24-chain models suggest that we should question the model of a purely crystalline cellulose core for small-diameter microfibrils. Smaller fibers are likely to have a very different structure, which we believe is a result of the greater solvent exposure that these chains experience. Comparisons with experimental structural data are dependent on the systems that we are simulating being closely related to the systems used in the experiments. With the NMR, diffraction, and scattering work, this is an assumption that may not hold true. Due to issues with experimental isolation protocols, it is likely that the experimental data are produced with microfibril aggregates instead of small-diameter microfibrils, in an environment with significantly less solvent than what was modeled in this work. This may not be as great an issue for NMR studies, as it has been shown that water content of systems studied are close to the in vivo water content and that changes to spectra are observed when working on cellulose samples from different sources and different aggregate sizes (Malm et al., 2010). There is also a potential issue with the mechanical and chemical pretreatments applied to cellulose that allow them to be investigated with the differing experimental methods, which could lead to an altered cellulose microfibril structure. It is not only the experiments that could affect our comparisons but also the computational methods, specifically the force field used. It could be possible that the C36 Charmm force field does not give a good account of Glc when in a cellulose microfibril. This force field was parameterized based on the structure and dynamics of Glc in solution and, as a result, may not be able to sample the correct energy profiles when a residue is in an interior chain, without any solvent exposure and interacting with many other Glc residues. It would be of interest to perform some ab initio calculations on small cellulose systems, such as those performed by Watts et al. (2013), to investigate the structure and energetics of Glc in a cellulose microfibril and how this compares with the structures and dynamics observed in these simulations. CONCLUSION We have performed MD simulations on 18-, 24-, and 36-chain models of Iβ cellulose and found that the smaller 18- and 24-chain models sample a conformational space different from that of the initial crystalline phase and any phase identified experimentally but similar to that observed in previous high-temperature MD simulations of a 36-chain model. The chains of all microfibrils became tilted, with the origin chains experiencing a clockwise tilt and center chains an anticlockwise tilt due to the different steric environments in which the two chains were situated. As a result of increased variation in the τ2 dihedral and the steric limitations of tilt, the 18- and 24-chain microfibrils were able to change exocyclic conformations in center layers, virtually exclusively to GG, while there were sporadic changes to GT in the origin layers. To accommodate these changes, a further stage of tilt was required. For all simulations, solvent-exposed exocyclic groups preferred to reside in a GT conformation; however, there was significant variability in the conformations sampled. As a result of these simulations and a reinterpretation of NMR data, it was concluded that the 18- and 24-chain models were more parsimonious than the 36-chain model, in agreement with recent scattering and diffraction data. Determination of the detailed structure of cellulose microfibrils in a hydrated environment will allow for more rigorous investigations of cellulose in larger systems. Aggregation of microfibrils can be studied, which can, in turn, be used to understand more about the recalcitrance to enzymic and chemical hydrolysis of plant-derived cellulose. Simulation studies with atomic detail of the interaction of cellulose with noncellulosic polymers are now possible, leading to a better understanding of atomic interactions underlying the structure, dynamics, and physical properties of the plant cell wall matrix. Identifying the number of chains in a microfibril also gives insight into the structure of the CesA rosette complex. The interpretation of simulation and NMR data to suggest 18 to 24 chains in a microfibril means that either the 36 proposed CesA proteins in a rosette are regulated such that not all subunits are simultaneously active or that there are less than 36 CesA proteins in a rosette. There have been suggestions that each six elementary particles of the rosette could contain less than six CesA proteins (Newman et al., 2013) or that the particle could contain both initiating and extending CesA proteins (Read and Bacic, 2002). It would also be of interest to run simulations similar to the ones reported in this study on cellulose aggregates made up of many small-diameter microfibrils or microfibrils of much larger diameter to further investigate the effect of chain numbers on crystallinity, chain conformation, surface features, and dynamic characteristics of cellulose. MATERIALS AND METHODS Models of cellulose microfibrils were constructed with Cellulose Builder (Gomes and Skaf, 2012). Initial models containing 36 chains (eight layers) were created for the Iβ form with a hexagonal shape (Fig. 3). These models were then used as templates to produce the smaller 18- and 24-chain models (six layers in each; Fig. 3, B and C). Models in which the 36-chain model had the B H-bond pattern of Nishiyama et al. (2002) were produced by rotating the position of HO2 of every residue such that the C3-C2-O2-HO2 dihedral angle was –14° and rotation of HO6 such that the C5-C6-O6-HO6 dihedral angle was 26°. Each chain contained 20 Glc residues (i.e. 10 cellobiose repeat units) so that the microfibril had a length of approximately 102 Å. Periodic boundary conditions were applied along all three directions. The bonding between Glc residues was arranged so that residues at one end of the simulation cell were covalently bonded to Glc residues on the other side of the periodic boundary. This creates a microfibril that is unaffected by boundary effects and is effectively infinite in length. Each microfibril was aligned so that its polymerization axis ran along the z axis of the simulation cell and was solvated in a box of transferable intermolecular potential 3P water so that the microfibril was padded in the x and y directions by at least 12 Å, with no padding in the z direction. All simulations were performed in duplicate using NAMD 2.9 (Kale et al., 1999) with the Charmm 36 carbohydrate force field (Guvench et al., 2008) at 300 K. An initial simulation step was performed to optimize the position of the added waters. With a time step of 1 fs, a nonbonded cutoff of 10 Å, and the RATTLE/SETTLE algorithms (Andersen, 1983; Miyamoto and Kollman, 1992) used to constrain the bond lengths involving hydrogens in water, 5,000 steps of minimization were followed by 500,000 steps of constant pressure-temperature Langevin dynamics. Pressure was kept constant using a Langevin piston barostat, and the particle mesh Ewald method was used to calculate long-range electrostatic interactions (Darden et al., 1993). Production phase simulations were performed in the constant volume-temperature ensemble with a time step of 2 fs and all bonds from hydrogen to heavy atoms constrained. To analyze the structure and dynamics of the microfibrils, multiple characteristics were calculated for the initial structure of each simulation and the last 50 ns. Unit cell parameters were calculated by splitting each microfibril into discrete unit cells and then averaging over all unit cells. The exocyclic group around C6 can take on three different orientations that can be described by the dihedral angle about the atoms O5-C5-C6-O6. With values of 0° to 120°, 120° to 240°, and 240° to 360°, these angles correspond to the GT, TG, and GG conformations, respectively (Fig. 2). The color scheme used in Matthews et al. (2012) has been used in this work to color cellulose chains and individual Glc residues. Intrachain, intralayer, and interlayer H-bond occupancies were calculated using the Visual Molecular Dynamics h_bonds command with a heavy atom cutoff of 3.4 Å and angle cutoff of 60°. The internal motion of the microfibrils was tracked by calculating the change in distance between 100/200 layers, 110 layers, 1-10 layers, and the shift of chains along the z axis (Fig. 1). The tilt in each chain was calculated by determining how much the plane of each chain rotates around its initial position along the xz plane. A positive value means there has been a clockwise rotation and a negative value anticlockwise rotation, when viewed from the nonreducing end. Average ϕ, ψ, τ2, and τ3 dihedral angles for each chain have also been calculated. The density of water above each layer was calculated as a function of distance from the microfibril and as a function of distance along the polymerization axis. Supplemental Data The following supplemental materials are available. Supplemental Figure S1 . Plot of exocyclic conformation occupancies for 640-ns simulation of 36 A model. Supplemental Figure S2. Exocyclic conformations at different time points up to 200 ns. Supplemental Figure S3. O6-to-O2 intrachain and O2-to-O6 intralayer H-bond occupancies. Supplemental Figure S4. Relative water densities for 36 B chain model. Supplemental Figure S5. Relative water densities for 24-chain model. Supplemental Figure S6. Relative water densities for 18-chain model. Supplemental Figure S7. Water densities above cellulose surfaces. Supplemental Table S1. Exocyclic occupancies for 110 and 1-10 hydrophilic surfaces. Supplemental Table S2. Average ϕ dihedral angles. Supplemental Table S3. Average ψ dihedral angles. Supplemental Table S4. Average τ3 dihedral angles. Supplemental Table S5. Occupancy of interlayer hydrogen bonds. LITERATURE CITED Andersen C ( 1983 ) Rattle: A “velocity” molecular version of the shake dynamics calculations for molecular dynamics calculations . J Comput Phys 52 : 24 – 34 Google Scholar Crossref Search ADS WorldCat Atalla RH , Vanderhart DL ( 1984 ) Native cellulose: a composite of two distinct crystalline forms . Science 223 : 283 – 285 Google Scholar Crossref Search ADS PubMed WorldCat Baker AA , Helbert W, Sugiyama J, Miles MJ ( 2000 ) New insight into cellulose structure by atomic force microscopy shows the i(α) crystal phase at near-atomic resolution . Biophys J 79 : 1139 – 1145 Google Scholar Crossref Search ADS PubMed WorldCat Bergenstråhle M , Berglund LA, Mazeau K ( 2007 ) Thermal response in crystalline Iβ cellulose: a molecular dynamics study . J Phys Chem B 111 : 9138 – 9145 Google Scholar Crossref Search ADS PubMed WorldCat Bock K , Duus JØ ( 1994 ) A conformational study of hydroxymethyl groups in carbohydrates investigated by 1H NMR spectroscopy . J Carbohydr Chem 13 : 513 – 543 Google Scholar Crossref Search ADS WorldCat Brett CT, Hillman JR (1985) Biochemistry of Plant Cell Walls, Ed 70. Cambridge University Press, Cambridge, UK Busse-Wicher M , Gomes TCF, Tryfona T, Nikolovski N, Stott K, Grantham NJ, Bolam DN, Skaf MS, Dupree P ( 2014 ) The pattern of xylan acetylation suggests xylan may interact with cellulose microfibrils as a twofold helical screw in the secondary plant cell wall of Arabidopsis thaliana . Plant J 79 : 492 – 506 Google Scholar Crossref Search ADS PubMed WorldCat Chen P , Nishiyama Y, Putaux JL, Mazeau K ( 2014 ) Diversity of potential hydrogen bonds in cellulose I revealed by molecular dynamics simulation . Cellulose 21 : 897 – 908 Google Scholar Crossref Search ADS WorldCat Chu SSC , Jeffrey GA ( 1968 ) The refinement of the crystal structures of β-D-glucose and cellobiose . Acta Crystallogr B 24 : 830 – 838 Google Scholar Crossref Search ADS WorldCat Cramer CJ , Truhlar DG ( 1993 ) Quantum chemical conformational analysis of glucose in aqueous solution . J Am Chem Soc 115 : 5745 – 5753 Google Scholar Crossref Search ADS WorldCat Darden T , York D, Pedersen L ( 1993 ) Particle mesh Ewald: An N.log(N) method for Ewald sums in large systems . J Chem Phys 98 : 10089 – 10092 Google Scholar Crossref Search ADS WorldCat Ding SY , Himmel ME ( 2006 ) The maize primary cell wall microfibril: a new model derived from Direct visualization . J Agric Food Chem 54 : 597 – 606 Google Scholar Crossref Search ADS PubMed WorldCat Ding SY , Zhao S, Zeng Y ( 2014 ) Size, shape, and arrangement of native cellulose fibrils in maize cell walls . Cellulose 21 : 863 – 871 Google Scholar Crossref Search ADS WorldCat Doblin MS , Pettolino F, Bacic A ( 2010 ) Evans Review: plant cell walls: the skeleton of the plant world . Funct Plant Biol 37 : 357 – 381 Google Scholar Crossref Search ADS WorldCat Earl WL , VanderHart DL ( 1981 ) Observations by high-resolution carbon-13 nuclear magnetic resonance of cellulose I related to morphology and crystal structure . Macromolecules 14 : 570 – 574 Google Scholar Crossref Search ADS WorldCat Fernandes AN , Thomas LH, Altaner CM, Callow P, Forsyth VT, Apperley DC, Kennedy CJ, Jarvis MC ( 2011 ) Nanostructure of cellulose microfibrils in spruce wood . Proc Natl Acad Sci USA 108 : E1195 – E1203 Google Scholar Crossref Search ADS PubMed WorldCat Gomes TCF , Skaf MS ( 2012 ) Cellulose-builder: a toolkit for building crystalline structures of cellulose . J Comput Chem 33 : 1338 – 1346 Google Scholar Crossref Search ADS PubMed WorldCat Gross AS , Chu JW ( 2010 ) On the molecular origins of biomass recalcitrance: the interaction network and solvation structures of cellulose microfibrils . J Phys Chem B 114 : 13333 – 13341 Google Scholar Crossref Search ADS PubMed WorldCat Guvench O , Greene SN, Kamath G, Brady JW, Venable RM, Pastor RW, Mackerell AD Jr ( 2008 ) Additive empirical force field for hexopyranose monosaccharides . J Comput Chem 29 : 2543 – 2564 Google Scholar Crossref Search ADS PubMed WorldCat Ha MA , Apperley DC, Evans BW, Huxham IM, Jardine WG, Viëtor RJ, Reis D, Vian B, Jarvis MC ( 1998 ) Fine structure in cellulose microfibrils: NMR evidence from onion and quince . Plant J 16 : 183 – 190 Google Scholar Crossref Search ADS PubMed WorldCat Hansen HS , Hünenberger PH ( 2011 ) A reoptimized GROMOS force field for hexopyranose-based carbohydrates accounting for the relative free energies of ring conformers, anomers, epimers, hydroxymethyl rotamers, and glycosidic linkage conformers . J Comput Chem 32 : 998 – 1032 Google Scholar Crossref Search ADS PubMed WorldCat Herth W ( 1983 ) Arrays of plasma-membrane “rosettes” involved in cellulose microfibril formation of Spirogyra . Planta 159 : 347 – 356 Google Scholar Crossref Search ADS PubMed WorldCat Horii F , Hirai A, Kitamaru R ( 1983 ) Solid-state 13C-NMR study of conformations of oligosaccharides and cellulose conformation of CH2OH group about the exo-cyclic C-C Bond . Polym Bull 10 : 357 – 361 Google Scholar Crossref Search ADS WorldCat Horii F , Hirai A, Kitamaru R ( 1987 ) CP/MAS 13C NMR spectra of the crystalline components of native celluloses . Macromolecules 20 : 2117 – 2120 Google Scholar Crossref Search ADS WorldCat Isogai A , Usuda M ( 1989 ) Solid-state CP/MAS 13C NMR study of cellulose polymorphs . Macromolecules 22 : 3168 – 3172 Kale L , Skeel R, Bhandarkar M, Brunner R, Gursoy A, Krawetz N, Phillips J, Shinozaki A, Varadarajan K, Schulten K ( 1999 ) NAMD2: greater scalability for parallel molecular dynamics . J Comput Phys 312 : 283 – 312 Google Scholar OpenURL Placeholder Text WorldCat Kirschner KN , Woods RJ ( 2001 ) Solvent interactions determine carbohydrate conformation . Proc Natl Acad Sci USA 98 : 10541 – 10545 Google Scholar Crossref Search ADS PubMed WorldCat Kuttel M , Brady JW, Naidoo KJ ( 2002 ) Carbohydrate solution simulations: producing a force field with experimentally consistent primary alcohol rotational frequencies and populations . J Comput Chem 23 : 1236 – 1243 Google Scholar Crossref Search ADS PubMed WorldCat Larsson PT , Westlund PO ( 2005 ) Line shapes in CP/MAS 13C NMR spectra of cellulose I . Spectrochim Acta A Mol Biomol Spectrosc 62 : 539 – 546 Google Scholar Crossref Search ADS PubMed WorldCat Larsson PT , Wickholm K, Iversen T ( 1997 ) A CP/MAS 13C NMR investigation of molecular ordering in celluloses . Carbohydr Res 302 : 19 – 25 Google Scholar Crossref Search ADS WorldCat Lehtiö J , Sugiyama J, Gustavsson M, Fransson L, Linder M, Teeri TT ( 2003 ) The binding specificity and affinity determinants of family 1 and family 3 cellulose binding modules . Proc Natl Acad Sci USA 100 : 484 – 489 Google Scholar Crossref Search ADS PubMed WorldCat Malm E , Bulone V, Wickholm K, Larsson PT, Iversen T ( 2010 ) The surface structure of well-ordered native cellulose fibrils in contact with water . Carbohydr Res 345 : 97 – 100 Google Scholar Crossref Search ADS PubMed WorldCat Matthews JF , Beckham GT, Bergenstrahle-Wohlert M, Brady JW, Himmel ME, Crowley MF ( 2012 ) Comparison of cellulose Iβ simulations with three carbohydrate force fields . J Chem Theory Comput 8 : 735 – 748 Google Scholar Crossref Search ADS PubMed WorldCat Matthews JF , Bergenstråhle M, Beckham GT, Himmel ME, Nimlos MR, Brady JW, Crowley MF ( 2011 a ) High-temperature behavior of cellulose I . J Phys Chem B 115 : 2155 – 2166 Google Scholar Crossref Search ADS PubMed WorldCat Matthews JF , Himmel ME, Crowley MF ( 2011 b ) Conversion of cellulose Iα to Iβ via a high temperature intermediate (I-HT) and other cellulose phase transformations . Cellulose 19 : 297 – 306 Google Scholar Crossref Search ADS WorldCat Matthews JF , Skopec CE, Mason PE, Zuccato P, Torget RW, Sugiyama J, Himmel ME, Brady JW ( 2006 ) Computer simulation studies of microcrystalline cellulose Iβ . Carbohydr Res 341 : 138 – 152 Google Scholar Crossref Search ADS PubMed WorldCat Mazeau K ( 2005 ) Structural micro-heterogeneities of crystalline Iβ-cellulose . Cellulose 12 : 339 – 349 Google Scholar Crossref Search ADS WorldCat Miyamoto S , Kollman PA ( 1992 ) SETTLE: An analytical version of the SHAKE and RATTLE algorithm for rigid water models . J Comput Chem 13 : 952 – 962 Google Scholar Crossref Search ADS WorldCat Newman RH ( 1998 ) Evidence for assignment of 13C NMR signals to cellulose crystallite surfaces in wood, pulp and isolated celluloses . Holzforschung 52 : 157 – 159 Google Scholar OpenURL Placeholder Text WorldCat Newman RH ( 1999 ) Estimation of the lateral dimensions of cellulose crystallites using 13C NMR signal strengths . Solid State Nucl Magn Reson 15 : 21 – 29 Google Scholar Crossref Search ADS PubMed WorldCat Newman RH , Hemmingson JA ( 1994 ) Carbon-13 NMR distinction between categories of molecular order and disorder in cellulose . Cellulose 2 : 95 – 110 Google Scholar Crossref Search ADS WorldCat Newman RH , Hill SJ, Harris PJ ( 2013 ) Wide-angle x-ray scattering and solid-state nuclear magnetic resonance data combined to test models for cellulose microfibrils in mung bean cell walls . Plant Physiol 163 : 1558 – 1567 Google Scholar Crossref Search ADS PubMed WorldCat Nishida Y , Ohrui H, Meguro H ( 1984 ) 1H-NMR studies of (6R)- and (6S)-deuterated D-hexoses: assignment of the preferred rotamers about C5-C6 bond of D-glucose and D-galactose derivatives in solutions . Tetrahedron Lett 25 : 1575 – 1578 Google Scholar Crossref Search ADS WorldCat Nishiyama Y ( 2009 ) Structure and properties of the cellulose microfibril . J Wood Sci 55 : 241 – 249 Google Scholar Crossref Search ADS WorldCat Nishiyama Y , Langan P, Chanzy H ( 2002 ) Crystal structure and hydrogen-bonding system in cellulose Iβ from synchrotron x-ray and neutron fiber diffraction . J Am Chem Soc 124 : 9074 – 9082 Google Scholar Crossref Search ADS PubMed WorldCat Nishiyama Y , Sugiyama J, Chanzy H, Langan P ( 2003 ) Crystal structure and hydrogen bonding system in cellulose I(α) from synchrotron x-ray and neutron fiber diffraction . J Am Chem Soc 125 : 14300 – 14306 Google Scholar Crossref Search ADS PubMed WorldCat Park S , Johnson DK, Ishizawa CI, Parilla PA, Davis MF ( 2009 ) Measuring the crystallinity index of cellulose by solid state 13C nuclear magnetic resonance . Cellulose 16 : 641 – 647 Google Scholar Crossref Search ADS WorldCat Payen A ( 1838 ) Memoir on the composition of the tissue of plants and of woody material . C R Biol 7 : 1052 – 1056 Google Scholar OpenURL Placeholder Text WorldCat Pereira CS , Kony D, Baron R, Müller M, van Gunsteren WF, Hünenberger PH ( 2006 ) Conformational and dynamical properties of disaccharides in water: a molecular dynamics study . Biophys J 90 : 4337 – 4344 Google Scholar Crossref Search ADS PubMed WorldCat Read SM , Bacic T ( 2002 ) Plant biology. Prime time for cellulose . Science 295 : 59 – 60 Google Scholar Crossref Search ADS PubMed WorldCat Shen T , Langan P, French AD, Johnson GP, Gnanakaran S ( 2009 ) Conformational flexibility of soluble cellulose oligomers: chain length and temperature dependence . J Am Chem Soc 131 : 14786 – 14794 Google Scholar Crossref Search ADS PubMed WorldCat Sugiyama J , Persson J, Chanzy H ( 1991 a ) Combined infrared and electron diffraction study of the polymorphism of native celluloses . Macromolecules 24 : 2461 – 2466 Google Scholar Crossref Search ADS WorldCat Sugiyama J , Vuong R, Chanzy H ( 1991 b ) Electron diffraction study on the two crystalline phases occurring in native cellulose from an algal cell wall . Macromolecules 24 : 4168 – 4175 Google Scholar Crossref Search ADS WorldCat Thomas LH , Forsyth VT, Sturcová A, Kennedy CJ, May RP, Altaner CM, Apperley DC, Wess TJ, Jarvis MC ( 2013 ) Structure of cellulose microfibrils in primary cell walls from collenchyma . Plant Physiol 161 : 465 – 476 Google Scholar Crossref Search ADS PubMed WorldCat Tormo J , Lamed R, Chirino AJ, Morag E, Bayer EA, Shoham Y, Steitz TA ( 1996 ) Crystal structure of a bacterial family-III cellulose-binding domain: a general mechanism for attachment to cellulose . EMBO J 15 : 5739 – 5751 Google Scholar Crossref Search ADS PubMed WorldCat VanderHart DL , Atalla RH ( 1984 ) Studies of microstructure in native celluloses using solid-state 13C NMR . Macromolecules 17 : 1465 – 1472 Google Scholar Crossref Search ADS WorldCat Viëtor RJ , Newman RH, Ha MA, Apperley DC, Jarvis MC ( 2002 ) Conformational features of crystal-surface cellulose from higher plants . Plant J 30 : 721 – 731 Google Scholar Crossref Search ADS PubMed WorldCat Wada M , Hori R, Kim UJ, Sasaki S ( 2010 ) X-ray diffraction study on the thermal expansion behavior of cellulose Iβ and its high-temperature phase . Polym Degrad Stabil 95 : 1330 – 1334 Google Scholar Crossref Search ADS WorldCat Watts HD , Mohamed MNA, Kubicki JD ( 2013 ) A DFT study of vibrational frequencies and 13C NMR chemical shifts of model cellulosic fragments as a function of size . Cellulose 21 : 53 – 70 Google Scholar Crossref Search ADS WorldCat Wickholm K , Larsson PT, Iversen T ( 1998 ) Assignment of non-crystalline forms in cellulose I by CP/MAS 13C NMR spectroscopy . Carbohydr Res 312 : 123 – 129 Google Scholar Crossref Search ADS WorldCat Zhang Q , Bulone V, Ågren H, Tu Y ( 2011 ) A molecular dynamics study of the thermal response of crystalline cellulose Iβ . Cellulose 18 : 207 – 221 Google Scholar Crossref Search ADS WorldCat Zugenmaier P ( 2001 ) Conformation and packing of various crystalline cellulose fibers . Prog Polym Sci 26 : 1341 – 1417 Google Scholar Crossref Search ADS WorldCat Author notes 1 This work was supported by the Australia Research Council to the Australian Research Council Centre of Excellence in Plant Cell Walls (grant no. CE110001007 to M.J.G., M.S.D., and A.B.). * Address correspondence to [email protected] and [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Antony Bacic ([email protected]). D.P.O. performed most of the experiments assisted by M.T.D.; all authors contributed to the experimental design and data interpretation; M.J.G., J.W., M.S.D., and A.B. conceived and supervised the project; D.P.O. wrote the first article draft; all authors contributed to the writing. [OPEN] Articles can be viewed without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.114.254664 © 2015 American Society of Plant Biologists. All Rights Reserved. © The Author(s) 2015. Published by Oxford University Press on behalf of American Society of Plant Biologists. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Littlejohn, George R.; Mansfield, Jessica C.; Parker, David; Lind, Rob; Perfect, Sarah; Seymour, Mark; Smirnoff, Nicholas; Love, John; Moger, Julian

doi: 10.1104/pp.15.00119pmid: 25783412

Abstract The cuticle is a ubiquitous, predominantly waxy layer on the aerial parts of higher plants that fulfils a number of essential physiological roles, including regulating evapotranspiration, light reflection, and heat tolerance, control of development, and providing an essential barrier between the organism and environmental agents such as chemicals or some pathogens. The structure and composition of the cuticle are closely associated but are typically investigated separately using a combination of structural imaging and biochemical analysis of extracted waxes. Recently, techniques that combine stain-free imaging and biochemical analysis, including Fourier transform infrared spectroscopy microscopy and coherent anti-Stokes Raman spectroscopy microscopy, have been used to investigate the cuticle, but the detection sensitivity is severely limited by the background signals from plant pigments. We present a new method for label-free, in vivo structural and biochemical analysis of plant cuticles based on stimulated Raman scattering (SRS) microscopy. As a proof of principle, we used SRS microscopy to analyze the cuticles from a variety of plants at different times in development. We demonstrate that the SRS virtually eliminates the background interference compared with coherent anti-Stokes Raman spectroscopy imaging and results in label-free, chemically specific confocal images of cuticle architecture with simultaneous characterization of cuticle composition. This innovative use of the SRS spectroscopy may find applications in agrochemical research and development or in studies of wax deposition during leaf development and, as such, represents an important step in the study of higher plant cuticles. The majority of land plants possess an extracellular, waxy cuticle that covers the surface of their aerial parts and protects them against desiccation, external physical and chemical stresses, and a variety of biological agents (Grncarevic and Radler, 1967; Barthlott and Neinhuis, 1997; Krauss et al., 1997; Ristic and Jenks, 2002; Yeats and Rose, 2013). The cuticle is a composite layer composed mainly of cutin and overlaid by cuticular waxes. Cutin is a macromolecular structure consisting primarily of hexadecanoic (palmitic) and octadecenoic (vaccenic) acids that are covalently linked by ester bonds to generate a rigid, three-dimensional network that is embedded with polysaccharides. Cuticular waxes are composed of long-chain (C20–C40) aliphatic molecules derived from fatty acids (Samuels et al., 2008), and studies over the last several decades have identified structural and regulatory constituents of the biosynthetic pathways of cuticular components (Kolattukudy, 1981; Beisson et al., 2012). In addition to the physiochemical properties conferred by its lipid components, the architecture of the cuticle plays an essential role in physiological function. For example, through understanding the properties of the cuticular structure, the extraordinary superhydrophobicity of the Lotus spp. leaf has been mimicked in micro- and nanotechnology to generate self-cleaning surfaces (Bhushan and Jung, 2006; Bhushan et al., 2009; Koch et al., 2009). As may be expected, given the diversity of plants, the habitats they inhabit, and individual life histories, the morphology and composition of plant cuticle varies extensively between and within species and includes plate-, needle-, and pillar-shaped wax crystals (Barthlott et al., 2008). In some species, cuticular wax composition is known to vary with depth, giving rise to chemically distinguishable layers (Yeats and Rose, 2013). Finally, the cuticle is increasingly shown to be important in development (Koornneef et al., 1989; Yeats and Rose, 2013) and pathogenesis (Lee and Dean, 1994; Gilbert et al., 1996; Bessire et al., 2007; Delventhal et al., 2014). It is therefore unsurprising that interest in cuticle composition, structure, and physiology is increasing (Buschhaus et al., 2014; Hen-Avivi et al., 2014; Heredia-Guerrero et al., 2014; Xu et al., 2014). Moreover, a greater understanding of the relationship between structure and chemical composition of cuticle waxes is vital for enhancing agriculture yields, as it will further our knowledge of how plants regulate water balance and inform the application of nutrition (foliar feeds) and pesticides, leading to improved formulation strategies for agrochemicals. The chemical composition and topological architecture of cuticular waxes are both critical for optimal physiological function. Analyses of these essential properties have typically been performed separately. Cuticle wax composition is normally determined using gas chromatography; cuticle ultrastructure may be analyzed using destructive imaging techniques such as scanning electron microscopy (SEM; Baker and Holloway, 1971; Jetter et al., 2000; Barthlott et al., 2008) and laser desorption ionizing mass spectroscopy (Jun et al., 2010) or, in vivo, using nondestructive real-time techniques, including white-light scanning interferometry (Kim et al., 2011), atomic force microscopy (Koch et al., 2004), confocal microscopy in reflectance mode (Veraverbeke et al., 2001), fluorescence microscopy of chemical stains (Pighin et al., 2004), coherent anti-Stokes Raman scattering (CARS) microscopy (Yu et al., 2008; Weissflog et al., 2010), and total internal reflection Raman spectroscopy (Greene and Bain, 2005). Despite the advances in our understanding of the cuticle that have been made with these techniques, there is a great need for techniques that combine chemical and structural information to provide in situ high-resolution chemical analysis of epicuticle waxes. Techniques based on vibrational spectroscopy offer in situ chemical analysis derived from the vibrational frequencies of molecular bonds within a sample. However, due to water absorption and the intrinsically low spatial resolution associated with the long infrared (IR) wavelengths required to directly excite molecular vibrations, IR absorption techniques have limited value for bioimaging. Raman scattering, however, provides analysis of vibrational frequencies by examining the inelastic scattering of visible light. Raman scattered light is frequency shifted with respect to the incident light by discrete amounts that correspond to the vibrational frequencies of molecular bonds within the sample. The spectrum of Raman scattered light consists of a series of discrete peaks that each correspond to a molecular bond and can be regarded as a chemical fingerprint holding a wealth of information regarding chemical composition. Unfortunately, Raman scattering is an extremely weak effect, and typical signals from biological samples are at least six orders of magnitude weaker than those from fluorescent labels. This severely limits the application of Raman for studying living systems because long acquisition times (100 ms–1 s per pixel) and relatively high excitation powers (several hundred milliwatts) are required to image most biomolecules with sufficient sensitivity. Furthermore, the lack of sensitivity is compounded by autofluorescence, which in plant tissues completely overwhelms the Raman signal, prohibiting its application in planta. Far stronger Raman signals can be obtained using coherent Raman scattering (CRS; Min et al., 2011). CRS achieves a Raman signal enhancement by focusing the excitation energy onto a specific molecular vibrational frequency (Fig. 1A). A pump and Stokes beam, with frequencies ωp and ωS, respectively, are incident upon the sample, with their frequency difference (ωp–ωS) tuned to match the molecular vibrational frequency of interest. Under this resonant condition, the excitation fields efficiently drive bonds to produce a strong nonlinear coherent Raman signal. When applied in microscopy format, the nonlinear nature of the CRS process confines the signal to a submicron focus that can be scanned in space, allowing three-dimensional confocal-like mapping of biomolecules. CRS microscopy has particular advantages for bioimaging: (1) Chemically specific contrast is derived from the vibrational signature of endogenous biomolecules within the sample, negating the need for extraneous labels/stains; (2) Low-energy, near-IR excitation wavelengths can be employed, which reduces photodamage and increases depth penetration into scattering tissues; and (3) The CRS process does not leave sample molecules in an excited state, does not suffer from photobleaching, and can be used for time course studies. Figure 1. Open in new tabDownload slide Schematic representation of the two CRS processes: CARS and SRS. A, Energy level diagrams for the CARS and SRS processes, showing the pump (green), Stokes (red), and anti-Stokes (blue) photon energies. B, Diagrammatic representation of the input and output spectra for CARS and SRS, showing the gain and loss in the pump (red) and Stokes (green) beams, respectively. ƊIS, Change in Stokes beam intensity; ƊIp, change in pump beam intensity. C, Diagrammatic representation of the modulation transfer detection scheme used to detect stimulated Raman gain and loss with high sensitivity. Figure 1. Open in new tabDownload slide Schematic representation of the two CRS processes: CARS and SRS. A, Energy level diagrams for the CARS and SRS processes, showing the pump (green), Stokes (red), and anti-Stokes (blue) photon energies. B, Diagrammatic representation of the input and output spectra for CARS and SRS, showing the gain and loss in the pump (red) and Stokes (green) beams, respectively. ƊIS, Change in Stokes beam intensity; ƊIp, change in pump beam intensity. C, Diagrammatic representation of the modulation transfer detection scheme used to detect stimulated Raman gain and loss with high sensitivity. CRS microscopy may be achieved by detecting either CARS or stimulated Raman scattering (SRS). CARS MICROSCOPY CARS microscopy relies on detection of the anti-Stokes signal generated at frequency ωas = 2ωp–ωS, which, by using filters, is spectrally isolated from the pump and Stokes beams, and its intensity is used to map the location of biomolecules of interest (Zumbusch et al., 1999). The CARS signal is blue shifted with respect to the pump and Stokes wavelengths (Fig. 1B), making CARS more resilient to sample autofluorescence than spontaneous Raman. However, in highly autofluorescent samples such as plant tissues, the usually weak two-photon fluorescence (also blue shifted with respect to the excitation wavelengths) overwhelms the CARS signal. Consequently, CARS imaging in planta has only been applied to samples with reduced chlorophyll autofluorescence including dried tissues (Zeng et al., 2010; Ding et al., 2012), roots (Ly et al., 2007), and cuticle waxes after they have been stripped away from the leaf (Weissflog et al., 2010). SRS MICROSCOPY SRS relies on detecting subtle changes in the intensities of the excitation fields that occur by virtue of stimulated excitation (Freudiger et al., 2008). When the difference frequency, ωp – ωS, matches the frequency of a molecular vibration, the Stokes beam intensity experiences a gain, while the intensity of the pump beam experiences a loss, shown in Figure 1C. This transfer of intensity between the excitation beams only occurs when both beams are incident simultaneously on the sample and is measured by detecting the modulation that is transferred to the pump beam after it has passed through the sample when the intensity of the Stokes beam is modulated. The amplitude of the transferred intensity modulation is directly proportional to the concentration of target molecules and, by modulating at frequencies above laser noise (>1 MHz), can be detected with a lock-in amplifier with great sensitivity (Ye et al., 2009). Because SRS is detected at the same wavelength as the excitation fields, it is not affected by fluorescent emission and, as recently demonstrated by Mansfield et al. (2013), may also be used in the presence of highly pigmented samples such as plant tissues. We have recently shown that it is possible use phase-sensitive detection to separate the vibrational SRS signal from the electronic absorption processes (Mansfield et al., 2013). In this investigation, we used SRS imaging to investigate both the structure and chemical composition of plant cuticular waxes, simultaneously and in vivo. We compared SRS images of cuticle wax structures from a variety of plant species that display characteristic wax morphologies with SEM and CARS images of the leaf surface. We demonstrated that spectroscopic information can be obtained to provide chemically specific, in situ analysis of waxes on living leaves with submicron spatial resolution. Using SRS microscopy, we compared the cuticle structure and composition between wild-type Arabidopsis (Arabidopsis thaliana) and an eciferum1 (cer1) mutant with impaired cuticle deposition and have also shown that SRS imaging is sufficiently sensitive to track temperature-induced changes in cuticle formation in Thellungiella salsuginea. We conclude that, with the increasing availability of commercial Raman-based microscopes, SRS imaging has the potential to radically advance our understanding of plant cuticular structure and composition and their effects on physiology and development. RESULTS Characterization of the Raman Spectra of Dissolved Cuticle Waxes Spontaneous Raman spectra of hexane-extracted wax samples over the CH vibrational region from 2,700 to 3,200 cm–1 were acquired as a baseline reference for each plant species (Fig. 2). Gaussian peaks were fitted underneath each of the spectra to visualize the relative contribution of different molecular vibrations, based on peaks previously identified in the literature (Snyder et al., 1978; Ho and Pemberton, 1998; Greene and Bain, 2005). The wax extracts from Arabidopsis, Thellungiella parvula, banana (Musa acuminata), and cheese plant (Monstera delicosa) had similar spectral profiles that matched those reported for other waxes and alkanes reported in the literature (Snyder et al., 1978; Greene and Bain, 2005) and deconvolved into seven identifiable peaks (Table I). The peaks, corresponding to the symmetrical and antisymmetrical CH2 stretch, are prominent on all spectra. The ratio antisymmetric-to-symmetric peak height is known to give a strong indication of the alkyl chain conformational order with a ratio of 1.6 to 2 indicating a highly crystalline structure, and a ratio of 0.6 to 0.9 indicating a liquid structure (Snyder et al., 1978; Ho and Pemberton, 1998, Greene and Bain, 2005). For the wax samples analyzed here, the ratios were as follows: Arabidopsis, 1.13; T. parvula 1.58; banana, 1.34; and cheese plant, 1.54; indicating structures intermediate between a highly crystalline structure and a liquid. The curve fits to the spectra also included two Fermi resonances of the CH2 bond one at 2,930 cm–1 and another very broad Fermi resonance at approximately 2,870 cm–1 (Snyder et al., 1978). The contributions due to the CH3 symmetric and antisymmetric stretches were small, as proportionally there are much less of these groups due to the long chain lengths. The fitting parameters for each wax along with the goodness of fit are included in Supplemental Text S1. Figure 2. Open in new tabDownload slide Raman spectra and curve fits of plant leaf waxes. A, Raman spectra of cuticle waxes purified from silver dollar plant, D. anthonyi, Arabidopsis, banana, and T. parvula. Peak fitting to the Raman spectra of T. parvula (B), D. anthonyi (C), and silver dollar plant (D). Figure 2. Open in new tabDownload slide Raman spectra and curve fits of plant leaf waxes. A, Raman spectra of cuticle waxes purified from silver dollar plant, D. anthonyi, Arabidopsis, banana, and T. parvula. Peak fitting to the Raman spectra of T. parvula (B), D. anthonyi (C), and silver dollar plant (D). Raman peaks used in curve fitting for cuticular waxes Table I. Raman peaks used in curve fitting for cuticular waxes Peak Wavelength (Approximate) . Assignment . 2,840 (crystalline) 2,850 (liquid) CH2 symmetric stretch 2,870 (crystalline) 2,880 (liquid) CH2 antisymmetric stretch 2,930 CH2 Fermi resonance 2,952 CH2 antisymmetric (out of plane) 2,964 CH2 antisymmetric (in plane) 2,880 to 2,901 CH2 symmetric 2,870 (very broad) CH2 Fermi resonance Peak Wavelength (Approximate) . Assignment . 2,840 (crystalline) 2,850 (liquid) CH2 symmetric stretch 2,870 (crystalline) 2,880 (liquid) CH2 antisymmetric stretch 2,930 CH2 Fermi resonance 2,952 CH2 antisymmetric (out of plane) 2,964 CH2 antisymmetric (in plane) 2,880 to 2,901 CH2 symmetric 2,870 (very broad) CH2 Fermi resonance Open in new tab Table I. Raman peaks used in curve fitting for cuticular waxes Peak Wavelength (Approximate) . Assignment . 2,840 (crystalline) 2,850 (liquid) CH2 symmetric stretch 2,870 (crystalline) 2,880 (liquid) CH2 antisymmetric stretch 2,930 CH2 Fermi resonance 2,952 CH2 antisymmetric (out of plane) 2,964 CH2 antisymmetric (in plane) 2,880 to 2,901 CH2 symmetric 2,870 (very broad) CH2 Fermi resonance Peak Wavelength (Approximate) . Assignment . 2,840 (crystalline) 2,850 (liquid) CH2 symmetric stretch 2,870 (crystalline) 2,880 (liquid) CH2 antisymmetric stretch 2,930 CH2 Fermi resonance 2,952 CH2 antisymmetric (out of plane) 2,964 CH2 antisymmetric (in plane) 2,880 to 2,901 CH2 symmetric 2,870 (very broad) CH2 Fermi resonance Open in new tab Two of the species investigated, Dudleya anthonyi and silver dollar plant (Xerosicyos danguyi), showed dramatically different Raman spectra from Arabidopsis, T. parvula, banana, and cheese plant, characterized by additional peaks, which are indicative of a more complex chemical structure, with additional chemical bonds to those described in Table I. In D. anthonyi, the Raman spectrum changed depending on the number of the hexane wash and hence, we surmise, on the depth of penetration into the cuticle; the initial, most superficial hexane wash showed an unusual Raman spectrum, but subsequent washes that solubilize waxes deeper in the cuticle showed spectra that matched more closely with those of a typical alkane. These differences are most readily explained by both these species possessing a thick, glaucous cuticle compared with the other plant species studied, an adaptation enabling survival in xeric environments. In Vivo Comparison of CARS and SRS Images of Plant Cuticle Having characterized the Raman spectra of waxes extracted from the cuticle of a variety of plants, we imaged the cuticles of the same plants in vivo, using both CARS microscopy (Weissflog et al., 2010) and SRS microscopy (Fig. 3). SRS and CARS images of the epicuticular wax layer of T. parvula leaves were acquired simultaneously (Fig. 3) and are presented as the exemplar to compare the two techniques. Figure 3. Open in new tabDownload slide CARS and SRS images of the T. parvulaleaf surface. The surface of a leaf from T. parvula was simultaneously imaged using CARS microscopy (A) and SRS microscopy (B). Both images were acquired at a 2,845 cm–1 CH2 symmetric stretch. The CARS image (A) is dominated by autofluorescence, and only the largest wax crystals are visible against the background. Conversely, the SRS image (B) is almost background free, enabling the cell walls and wax crystals to be clearly visualized. C shows the spectral scan of the SRS (red crosses) from the wax crystals in situ overlaid onto the spontaneous Raman spectra of purified T. parvula wax (blue line), purified cellulose (green line), and purified pectin (purple line). Figure 3. Open in new tabDownload slide CARS and SRS images of the T. parvulaleaf surface. The surface of a leaf from T. parvula was simultaneously imaged using CARS microscopy (A) and SRS microscopy (B). Both images were acquired at a 2,845 cm–1 CH2 symmetric stretch. The CARS image (A) is dominated by autofluorescence, and only the largest wax crystals are visible against the background. Conversely, the SRS image (B) is almost background free, enabling the cell walls and wax crystals to be clearly visualized. C shows the spectral scan of the SRS (red crosses) from the wax crystals in situ overlaid onto the spontaneous Raman spectra of purified T. parvula wax (blue line), purified cellulose (green line), and purified pectin (purple line). The images acquired using CARS are dominated by high levels of background autofluorescence from cell walls and chloroplasts (Fig. 3A). This autofluorescence is due to inevitable two-photon absorption by the compounds in these organelles and the resulting fluorescence emission at similar wavelengths to the CARS signal. Consequently, in the CARS images, only the largest wax crystals can be visualized. Conversely, in the SRS image (Fig. 3B), there is no fluorescent background because the signal measured is not due to a change in wavelength (i.e. fluorescence relative to excitation) but a change in intensity between the excitation lasers (the pump and Stokes beams) following stimulated Raman excitation of the samples. The cuticular wax crystals are clearly visible, and the SRS signal spectrally matches that of the spontaneous Raman spectrum of purified cuticular waxes (Fig. 3C). Cell walls are also visible in the SRS images, although with a lesser intensity than the wax crystals, which can be explained by the overlap between the Raman spectra of the cuticular waxes and cellulose. SRS microscopy therefore, not only provides clearer images of plant cuticle structure than CARS microscopy, but, because of the low levels of background noise in the images, also has the capacity to yield information on specific compounds that compose the imaged structures. Comparison of SRS and Scanning Electron Microscope Images of Plant Cuticle To confirm the capacity of SRS microscopy for accurately resolving cuticular structures, the surface of Arabidopsis stems were imaged using SRS microscopy and the more conventional SEM (Fig. 4). SRS and SEM images of untreated, wild-type stems (Fig. 4, A and B) show an uneven surface composed of similarly proportioned globules, although the level of detail is higher in the SEM image. We ascertained that these globular structures were waxes by washing the stems in hexane prior to imaging (Fig. 4, B and C). In this case, both images show that the surface of the stem is smooth, with clearly visible cells and stomata. Finally, stems from Arabidopsis defective for CER1 gene expression were imaged using either technique (Fig. 4, E and F). cer1 mutants are impaired in cuticle wax biosynthesis (Aarts et al., 1995; Bernard et al., 2012), and the images show stems with markedly reduced surface structures compared with the wild type (Fig. 4, A and B). Figure 4. Open in new tabDownload slide SRS images and scanning electron micrographs of Arabidopsis cuticle. The surface of Arabidopsis stems were imaged using SRS microscopy (A, C, and D) and SEM (B, D, and E). A and B, The surface of the untreated stem of wild-type Arabidopsis, ecotype Landsberg erecta, with the structure of cuticle wax crystals clearly visible. C and D, The surface of the wild-type Arabidopsis stem following a wash in hexane that dissolved the cuticle waxes. E and F, The surface of the stem from a cer1 Arabidopsis mutant. A, C, and E were generated from three-dimensional stacks taken of a 64- × 64-µm field of view and are displayed in false color. Figure 4. Open in new tabDownload slide SRS images and scanning electron micrographs of Arabidopsis cuticle. The surface of Arabidopsis stems were imaged using SRS microscopy (A, C, and D) and SEM (B, D, and E). A and B, The surface of the untreated stem of wild-type Arabidopsis, ecotype Landsberg erecta, with the structure of cuticle wax crystals clearly visible. C and D, The surface of the wild-type Arabidopsis stem following a wash in hexane that dissolved the cuticle waxes. E and F, The surface of the stem from a cer1 Arabidopsis mutant. A, C, and E were generated from three-dimensional stacks taken of a 64- × 64-µm field of view and are displayed in false color. Unlike SEM, which provides information only on the surface topology of imaged samples, SRS microscopy is a technique of multiphoton confocal microscopy that, like conventional confocal microscopy, can penetrate materials and provide information on their subsurface structure. To illustrate this technical capability, we imaged the cuticles of banana and silver dollar plant (Fig. 5). Banana was chosen, as it is a monocot, with a visibly waxy cuticle and regular files of elongated, epidermal cells. Silver dollar plant was chosen because of its thick, glaucous cuticle composed of different waxes with different Raman spectra. Figure 5. Open in new tabDownload slide SRS images and scanning electron micrographs of cuticle of banana and of silver dollar plant. The surface of banana and of silver dollar plant leaves were imaged using SRS microscopy (A–D) and SEM (E and F). A, An SRS image of the adaxial leaf surface of surface banana constructed from an image stack from a 64- × 64-µm field of view. B, An SRS image of silver dollar plant leaf reconstructed from an image stack from a 126- × 126-µm field of view. C and D, Orthogonal views projected from the image stacks presented in A and B, respectively, showing the depth profile of cuticles. E and F, SEM images of the cuticles of banana and silver dollar plant leaves, respectively. Figure 5. Open in new tabDownload slide SRS images and scanning electron micrographs of cuticle of banana and of silver dollar plant. The surface of banana and of silver dollar plant leaves were imaged using SRS microscopy (A–D) and SEM (E and F). A, An SRS image of the adaxial leaf surface of surface banana constructed from an image stack from a 64- × 64-µm field of view. B, An SRS image of silver dollar plant leaf reconstructed from an image stack from a 126- × 126-µm field of view. C and D, Orthogonal views projected from the image stacks presented in A and B, respectively, showing the depth profile of cuticles. E and F, SEM images of the cuticles of banana and silver dollar plant leaves, respectively. Three-dimensional reconstructions of confocal SRS images of banana (Fig. 5, A and C) and silver dollar plant (Fig. 5, B and C) cuticles show very different structures. The banana cuticle is formed of a number of hair-like projections from the leaf surface. In the comparator SEM image (Fig. 5E), these projections clump together, most likely due to the process of fixation, and show a less even distribution. By contrast, the silver dollar plant cuticle has an amorphous surface, supported by a reticulate pattern (6–10 µm below the surface) of columnar structures emerging from the leaf epidermis. The amorphous surface of the silver dollar plant cuticle is clear on the comparator SEM image (Fig. 5F), but the intriguing cuticle substructure is not. The comparison between SRS and SEM images therefore demonstrates that sufficient contrast can be produced using SRS microscopy to provide credible, structural images of waxes with submicron resolution that correlate well with images from SEM and can provide critical, structural information not previously available. Simultaneous Ultrastructural Imaging and Chemical Analysis of Cuticle by SRS Microscopy As previously noted, SRS imaging relies on the chemically specific Raman spectrum of the cuticle constituents. To verify the potential of the technique to provide structural and chemical information, simultaneously and in vivo, we first analyzed the cuticle of D. anthonyi (Fig. 6). D. anthonyiwas selected for this aspect of the investigation because previous work had shown stratification of the composition of cuticular wax in the D. anthonyi cuticle (Jetter et al., 2000). Figure 6. Open in new tabDownload slide SRS spectral images of D. anthonyi cuticle. SRS spectral imaging of D. anthonyi cuticle waxes shows changes in chemical composition with depth. A and B, The SRS and SEM images, respectively, of D. anthonyi leaf showing the crystalline structure of cuticle wax deposits on the adaxial surface. The image in A is a three-dimensional reconstruction of an image stack taken from a 126- × 126-µm field of view at the 2,840 cm–1 Raman shift. C, SRS spectral scans taken from the more superficial and deeper wax areas indicated on D. Blue rectangle, An area of superficial wax; red line, an area deeper in the cuticle. The imagess on the right and bottom of the main image in D are from the orthogonal views through the stack showing the depth profile image of the cuticle wax. Figure 6. Open in new tabDownload slide SRS spectral images of D. anthonyi cuticle. SRS spectral imaging of D. anthonyi cuticle waxes shows changes in chemical composition with depth. A and B, The SRS and SEM images, respectively, of D. anthonyi leaf showing the crystalline structure of cuticle wax deposits on the adaxial surface. The image in A is a three-dimensional reconstruction of an image stack taken from a 126- × 126-µm field of view at the 2,840 cm–1 Raman shift. C, SRS spectral scans taken from the more superficial and deeper wax areas indicated on D. Blue rectangle, An area of superficial wax; red line, an area deeper in the cuticle. The imagess on the right and bottom of the main image in D are from the orthogonal views through the stack showing the depth profile image of the cuticle wax. Both the SRS (Fig. 6A) and SEM (Fig. 6B) images of the D. anthonyi cuticle show a complex structure; the outer cuticle layer has a structure in which many small wax crystals are amassed together, with the deeper layers of the cuticle having a smoother, somewhat reticulated or wavelet-like appearance (Fig. 6D). Most importantly for this investigation, it is clear that the wax composition, as reported by the SRS spectra (Fig. 6C), was different depending on cuticle depth. The SRS spectrum acquired from deeper in the cuticle was most similar to an alkane, whereas the SRS spectrum from the surface components was markedly different, indicating a different chemical composition. It is well documented that changes in wax composition and deposition during development and in response to abiotic stress or pathogen attack are important in the life history of many plant species (Raffaele et al., 2009; Bourdenx et al., 2011). T. salsuginea is particularly interesting because it is a close relative of Arabidopsis and T. parvula, and its cuticle composition and structure alters following exposure to prolonged cold treatment or vernalization (Teusink et al., 2002; Amtmann, 2009; Xu et al., 2014). To determine whether SRS microscopy could, in principle, document the ultrastructural and chemical changes in cuticle that may occur during plant development, we imaged leaves from T. salsuginea plants that had been grown either in constant temperature or exposed to a 14-d-long cold shock using SRS microscopy and SEM (Fig. 7). Plants grown at a constant temperature of 20°C show a cuticle containing aggregates of waxes on the adaxial surface of the leaf (Fig. 7, A and B) and relatively little cuticular waxes on the abaxial surface (Fig. 7, C and D). However, following incubation at 4°C for 14 d, a marked alteration in cuticle structure can be observed on both the adaxial and abaxial surfaces (Fig. 7, C–H): a marked increase in waxy deposits in the cuticle was observed covering the entire leaf. Moreover, the ultrastructural aggregation of the cuticle waxes was smaller and more regular than that observed on the adaxial surface of leaves grown in a constant temperature of 20°C (Fig. 7, A and B). SRS microscopy therefore precisely reported the anticipated changes in cuticle formation that occur following exposure to cold in T. salsuginea and, by extension, may prove a useful, in vivo tool for investigating other dynamic changes in cuticle architecture and composition during development and in response to environmental or biotic perturbation. Figure 7. Open in new tabDownload slide SRS images and scanning electron micrographs of the cuticle of T. salsuginea following cold-induced wax biogenesis. T. salsuginea leaves were imaged using SRS spectral imaging and SEM before and after cold treatment. A, C, E, and G, Three-dimensional reconstructions of image stacks from a 250- × 250-µm field of view at the 2,840 cm–1 Raman shift. B, D, F, and H, SEM images. A to D are, respectively, from the adaxial and abaxial leaf surfaces of plants grown at 20°C for 8 weeks. E to H are, respectively, from adaxial and abaxial leaf surfaces of plants grown at 20°C for 4 weeks and then at 4°C for 2 weeks and at 20°C for an additional 2 weeks. Bars = 50 µm. Figure 7. Open in new tabDownload slide SRS images and scanning electron micrographs of the cuticle of T. salsuginea following cold-induced wax biogenesis. T. salsuginea leaves were imaged using SRS spectral imaging and SEM before and after cold treatment. A, C, E, and G, Three-dimensional reconstructions of image stacks from a 250- × 250-µm field of view at the 2,840 cm–1 Raman shift. B, D, F, and H, SEM images. A to D are, respectively, from the adaxial and abaxial leaf surfaces of plants grown at 20°C for 8 weeks. E to H are, respectively, from adaxial and abaxial leaf surfaces of plants grown at 20°C for 4 weeks and then at 4°C for 2 weeks and at 20°C for an additional 2 weeks. Bars = 50 µm. DISCUSSION The cuticle is a composite, hydrophobic layer composed mainly of waxes and cutin secreted from the aerial epidermis of land plants. The cuticle performs a number of physiological roles, notably acting as a barrier to water loss, which, in higher plants, enables the exquisite control of evapotranspiration by stomatal guard cells, increasing light reflection and heat tolerance in xerophytes and providing an essential barrier to the ingress of toxins and pathogens into the plant. Cuticle topography and composition has mainly been investigated using destructive techniques that provide only a fragmented understanding of the complexity of this essential structure. In this investigation, we have generated in vivo images of cuticle waxes at submicron resolution, using SRS microscopy. Like other forms of nonlinear Raman-based imaging techniques, SRS is label free and enables the simultaneous acquisition of chemically specific data and confocal quality images. Most importantly, SRS microscopy relies on a change in signal intensity rather than wavelength, so images can be acquired without the strong autofluorescent background from chloroplasts and cell walls that hampers the resolution and interpretation of CARS and spontaneous Raman images (Gierlinger et al., 2010, 2012). The structures observed using SRS microscopy compared well to those seen under the SEM, albeit with a lesser degree of resolution due to the inherent limitations of optical microscopy relative to electron microscopy. However, unlike SEM, which generates images from the surface of fixed or cryogenically preserved samples, SRS imaging, like other forms of single- or multiphoton laser microscopy, enables information to be acquired in vivo from within the sample and three-dimensional reconstructions of internal structures to be reconstructed. This capacity is particularly advantageous for documenting dynamic changes of topography or composition that may occur and that are otherwise undetectable using fixed samples. As demonstrated here, the chemical specificity of the SRS signal allows differentiation of different wax components from the main constituents of cell walls, cellulose, and pectins. In this study, we showed that the xerophytes D. anthonyi and silver dollar plant had different cuticular wax compositions from the other mesophytic plants investigated. Although this result is unsurprising, as these plants were selected for precisely their singularly thick and glaucous cuticles, it does demonstrate the potential of SRS for in vivo characterization of cuticle constituents. Moreover, although we did not attempt in this study to quantify, in situ, the different compounds for which we acquired SRS spectra, it may be possible to use this technique in a more quantifiable manner to characterize, in vivo, the relative abundance of cuticle or cell wall components and the changes that may occur in response to environmental or developmental stimuli. In this investigation, we have shown that SRS imaging is sufficiently sensitive enough to show developmental changes in wax production in T. salsuginea. Moreover, the technique may be useful for characterizing wax biosynthesis mutants, such as the exemplar used here, cer1, which has been shown to have a role in both water use and Arabidopsis interaction with fungal and bacterial pathogens (Raffaele et al., 2009; Bourdenx et al., 2011). SRS microscopy may therefore provide new insights into the physiological responses of plants to drought, temperature, chemicals, or pathogens and underpin the use of alternative model systems, such as T. salsuginea (Amtmann 2009), to investigate the control of cuticle deposition. However promising, any new technology or application must be simplified to enable widespread use. Currently, SRS microscopy is somewhat niche and requires specialist knowledge. However, commercial, user-friendly CARS microscopes that can be converted to SRS imaging exist and, like laser confocal microscopy, will become increasingly common and add real-time, label-free Raman microscopy to the range of imaging techniques widely available to researchers in plant science. CONCLUSION The chemical composition and structure of the cuticle in mesophyllous plants are crucial elements for survival, physiology, and development. We have shown that SRS microscopy can be used to acquire in vivo, three-dimensional images of the cuticle in a variety of plant species, enabling simultaneous analysis of cuticle structure and chemical composition. In contrast to CARS microscopy, which is more common, SRS images of plant tissues contain very low autofluorescence from chloroplasts or cell walls and are therefore easier to interpret. Moreover, SRS images compare well with those acquired using SEM, albeit with the lower resolution that is due to the use of laser light rather than an electron beam. Moreover, we have shown that SRS can be used to track and quantify dynamic changes in cuticle structure and composition in response to environmental stimuli and can therefore increase our understanding of this essential and often-overlooked structure. MATERIALS AND METHODS Plants Silver dollar plant (Xerosicyos danguyi), Dudleya anthonyi, banana (Musa acuminate), and cheese plant (Monstera delicosa) were grown in soil in the University of Exeter greenhouses and harvested between February and April 2012. The photoperiod was set to 16 h, from 5 am to 9 pm. Supplemental lighting and shading were provided to ensure irradiance between 540 and 720 µmol s–1 m–2. The greenhouse temperature was set to 19.5°C with an sd of 3.7°C. Arabidopsis (Arabidopsis thaliana), Thellungiella salsuginea, and Thellungiella parvula were grown in a 16-h-light/8-h-dark photoperiod at 20°C in growth rooms controlled for temperature and humidity. Four weeks after germination, the Thellungiella spp. were transferred to an ambient temperature of 4°C for 2 weeks and then returned to 20°C for an additional 2 weeks. The light regime remained at 16 h of light/8 h of dark during all experiments. Leaves were harvested for imaging and biochemical analysis. Purification of Wax from Plant Cuticle Following excision from plants, leaf surface areas were measured and logged. Leaf surfaces were washed for 30 s in 15 mL of high-performance liquid chromatography-grade hexane (Sigma). The hexane solvent and dissolved cuticular waxes were decanted into a glass vial that had been washed with acetone and dried. Leaves were washed a second time with 2 mL of hexane, which was added to the appropriate sample vial. Hexane was evaporated under a continuous stream of N2 to dryness, and the cuticle wax was redissolved in 250 µL of hexane. Spontaneous Raman Spectroscopy Spontaneous Raman spectra of purified wax samples were acquired using a Renishaw RM100 Raman microscope equipped with a 785-nm diode laser and a 1,200 line mm–1 spectral grating, which gave a spectral resolution of 1 cm–1. Samples were mounted on aluminum-coated microscope slides. SRS Microscopy A detailed technical explanation of the materials and methods for SRS microscopy is provided to promote the implementation of the technique as widely as possible. For stimulated Raman gain microscopy, two laser systems were used: The Stokes beam was provided by the signal output from the picosecond Optical Parametric Oscillator (Levante, Emerald APE) pumped by the frequency doubled output from a neodymium vanadate ps oscillator (picoTrain, HighQ laser). The pump beam was provided by a titanium sapphire laser (MIRA 900 D, Coherent) tuned to 770 nm and operating in ps mode, which was electronically synchronized to the neodymium vanadate laser (Coherent, Synchro-lock AP), thereby ensuring the laser pulses were temporally overlapped. The 770-nm pump beam was amplitude modulated at 1.7 MHz using an acousto-optical modulator (Crystal Technology, model 3080-122) and combined with the signal output from the optical parametric oscillator using an 850-nm short-pass dichroic mirror. Imaging was performed using a modified confocal laser scanning unit (Flouview 300, Olympus) and Olympus IX71 inverted microscope. The laser light was focused onto the sample using a 60× 1.2 numerical aperture water immersion microscope objective (UPlanS Apo, Olympus). The transmitted light from the sample was collected with a 60× 1.0 numerical aperture water-dipping condenser (LUMFI, Olympus). The transmitted Stokes beam was detected using an silicon photodiode (Thorlabs, FDS1010) with a 70-V reverse bias. The pump beam was blocked from the photodiode using an 850-nm-long pass filter (hq850lp, Chroma Technologies). The output from the photodiode was passed through a long-pass filter (minicircuits, BLP-1.9+) to remove modulations at the 76-laser repetition rate and terminated by a 50-Ω resistor. The filtered beam was then fed into a lock-in amplifier (Zurich Instruments, HF2L1 Lock-in Amplifier) that separated out the modulated SRS signal at 1.7 MHz. SRS spectra were obtained by tuning wavelength of signal beam in 0.2-nm intervals and acquiring a series of images. Stimulated Raman gain was measured in preference to stimulated Raman loss, as it allowed the chemically specific Raman signal to be separated from any additional signal due to two photon absorption (Ye et al., 2009) or photothermal lensing (Lu et al., 2010; Moger et al., 2012) by phase-sensitive lock-in detection (Mansfield et al., 2013). Fresh, excised plant leaves were mounted in perfluorodecalin (Littlejohn et al., 2010) prior to imaging, which has been shown not to interfere with Raman-based imaging in plants (Mansfield et al., 2013; Littlejohn et al., 2014). SEM Leaf samples were imaged using cryogenic SEM. Samples were flash frozen in liquid N2 slush, transferred to a vacuum, and coated in gold using the Gatan Alto 2100 system. Images were acquired using a JEOL JSM-6390 LV scanning electron microscope operating at 5 kV with a working distance of 10 to 12 nm. Supplemental Data Supplemental Text S1. The Raman fitting parameters for each purified cuticular wax. ACKNOWLEDGMENTS We thank Peter Splatt (Exeter Bioimaging Centre) for technical assistance with the SEM, James Chidlow for plant samples from the University of Exeter greenhouse collection, and Dr. Anna Amtmann (University of Glasgow) for generously providing seeds of Thellungiella species used. Glossary SRS stimulated Raman scattering CARS coherent anti-Stokes Raman scattering IR long infrared CRS coherent Raman scattering SEM scanning electron microscopy ps picosecond LITERATURE CITED Aarts MG , Keijzer CJ, Stiekema WJ, Pereira A ( 1995 ) Molecular characterization of the CER1 gene of Arabidopsis involved in epicuticular wax biosynthesis and pollen fertility . Plant Cell 7 : 2115 – 2127 Google Scholar PubMed OpenURL Placeholder Text WorldCat Amtmann A ( 2009 ) Learning from evolution: Thellungiella generates new knowledge on essential and critical components of abiotic stress tolerance in plants . Mol Plant 2 : 3 – 12 Google Scholar Crossref Search ADS PubMed WorldCat Baker EA , Holloway PJ ( 1971 ) Scanning electron microscopy of waxes on plant surfaces . Micron 2 : 364 – 380 Google Scholar OpenURL Placeholder Text WorldCat Barthlott W , Neinhuis C ( 1997 ) Purity of the sacred lotus, or escape from contamination in biological surfaces . Planta 202 : 1 – 8 Google Scholar Crossref Search ADS WorldCat Barthlott W , Neinhuis C, Cutler D, Ditsch F, Meusel I, Theisen I, Wilhelmi H ( 2008 ) Classification and terminology of plant epicuticular waxes . Bot J Linn Soc 126 : 237 – 260 Google Scholar Crossref Search ADS WorldCat Beisson F , Li-Beisson Y, Pollard M ( 2012 ) Solving the puzzles of cutin and suberin polymer biosynthesis . Curr Opin Plant Biol 15 : 329 – 337 Google Scholar Crossref Search ADS PubMed WorldCat Bernard A , Domergue F, Pascal S, Jetter R, Renne C, Faure JD, Haslam RP, Napier JA, Lessire R, Joubès J ( 2012 ) Reconstitution of plant alkane biosynthesis in yeast demonstrates that Arabidopsis ECERIFERUM1 and ECERIFERUM3 are core components of a very-long-chain alkane synthesis complex . Plant Cell 24 : 3106 – 3118 Google Scholar Crossref Search ADS PubMed WorldCat Bessire M , Chassot C, Jacquat AC, Humphry M, Borel S, Petétot JMC, Métraux JP, Nawrath C ( 2007 ) A permeable cuticle in Arabidopsis leads to a strong resistance to Botrytis cinerea . EMBO J 26 : 2158 – 2168 Google Scholar Crossref Search ADS PubMed WorldCat Bhushan B , Jung YC ( 2006 ) Micro- and nanoscale characterization of hydrophobic and hydrophilic leaf surfaces . Nanotechnology 17 : 2758 – 2772 Google Scholar Crossref Search ADS WorldCat Bhushan B , Jung YC, Koch K ( 2009 ) Micro-, nano- and hierarchical structures for superhydrophobicity, self-cleaning and low adhesion . Philos Trans A Math Phys Eng Sci 367 : 1631 – 1672 Google Scholar PubMed OpenURL Placeholder Text WorldCat Bourdenx B , Bernard A, Domergue F, Pascal S, Léger A, Roby D, Pervent M, Vile D, Haslam RP, Napier JA, et al. ( 2011 ) Overexpression of Arabidopsis ECERIFERUM1 promotes wax very-long-chain alkane biosynthesis and influences plant response to biotic and abiotic stresses . Plant Physiol 156 : 29 – 45 Buschhaus C , Hager D, Jetter R ( 2014 ) Wax layers on Cosmos bipinnatus petals contribute unequally to the total petal water resistance . Plant Physiol 167 : 80 – 88 Delventhal R , Falter C, Strugala R, Zellerhoff N, Schaffrath U ( 2014 ) Ectoparasitic growth of Magnaporthe on barley triggers expression of the putative barley wax biosynthesis gene CYP96B22, which is involved in penetration resistance . BMC Plant Biol 14 : 26 Google Scholar Crossref Search ADS PubMed WorldCat Ding SY , Liu YS, Zeng Y, Himmel ME, Baker JO, Bayer EA ( 2012 ) How does plant cell wall nanoscale architecture correlate with enzymatic digestibility? Science 338 : 1055 – 1060 Google Scholar Crossref Search ADS PubMed WorldCat Freudiger CW , Min W, Saar BG, Lu S, Holtom GR, He C, Tsai JC, Kang JX, Xie XS ( 2008 ) Label-free biomedical imaging with high sensitivity by stimulated Raman scattering microscopy . Science 322 : 1857 – 1861 Google Scholar Crossref Search ADS PubMed WorldCat Gilbert RD , Johnson AM, Dean RA ( 1996 ) Chemical signals responsible for appressorium formation in the rice blast fungus Magnaporthe grisea . Physiol Mol Plant Pathol 48 : 335 – 346 Google Scholar Crossref Search ADS WorldCat Greene PR , Bain CD ( 2005 ) Total internal reflection Raman spectroscopy of barley leaf epicuticular waxes in vivo . Colloids Surf B Biointerfaces 45 : 174 – 180 Google Scholar Crossref Search ADS PubMed WorldCat Gierlinger N , Keplinger T, Harrington M ( 2012 ) Imaging of plant cell walls by confocal Raman microscopy . Nat Protoc 7 : 1694 – 1708 Gierlinger N , Luss S, König C, Konnerth J, Eder M, Fratzl P ( 2010 ) Cellulose microfibril orientation of Picea abies and its variability at the micron-level determined by Raman imaging . J Exp Bot 61 : 587 – 595 Google Scholar Crossref Search ADS PubMed WorldCat Grncarevic M , Radler F ( 1967 ) The effect of wax components on cuticular transpiration-model experiments . Planta 75 : 23 – 27 Google Scholar Crossref Search ADS PubMed WorldCat Heredia-Guerrero JA , Benítez JJ, Domínguez E, Bayer IS, Cingolani R, Athanassiou A, Heredia A ( 2014 ) Infrared and Raman spectroscopic features of plant cuticles: a review . Front Plant Sci 5 : 305 Google Scholar Crossref Search ADS PubMed WorldCat Hen-Avivi S , Lashbrooke J, Costa F, Aharoni A ( 2014 ) Scratching the surface: genetic regulation of cuticle assembly in fleshy fruit . J Exp Bot 65 : 4653 – 4664 Google Scholar Crossref Search ADS PubMed WorldCat Ho M , Pemberton JE ( 1998 ) Alkyl chain conformation of octadecylsilane stationary phases by Raman spectroscopy. 1. Temperature dependence . Anal Chem 70 : 4915 – 4920 Google Scholar Crossref Search ADS PubMed WorldCat Jetter R , Schäffer S, Riederer M ( 2000 ) Leaf cuticular waxes are arranged in chemically and mechanically distinct layers: evidence from Prunus laurocerasus L . Plant Cell Environ 23 : 619 – 628 Google Scholar Crossref Search ADS WorldCat Jun JH , Song Z, Liu Z, Nikolau BJ, Yeung ES, Lee YJ ( 2010 ) High-spatial and high-mass resolution imaging of surface metabolites of Arabidopsis thaliana by laser desorption-ionization mass spectrometry using colloidal silver . Anal Chem 82 : 3255 – 3265 Google Scholar Crossref Search ADS PubMed WorldCat Kim KW , Lee ST, Bae SW, Kim PG ( 2011 ) 3D surface profiling and high resolution imaging for refining the florin rings and epicuticular wax crystals of Pinus koraiensis needles . Microsc Res Tech 74 : 1166 – 1173 Google Scholar Crossref Search ADS PubMed WorldCat Koch K , Bhushan B, Barthlott W ( 2009 ) Multifunctional surface structures of plants: an inspiration for biomimetics . Prog Mater Sci 54 : 137 – 178 Google Scholar Crossref Search ADS WorldCat Koch K , Neinhuis C, Ensikat HJ, Barthlott W ( 2004 ) Self assembly of epicuticular waxes on living plant surfaces imaged by atomic force microscopy (AFM) . J Exp Bot 55 : 711 – 718 Google Scholar Crossref Search ADS PubMed WorldCat Kolattukudy PE ( 1981 ) Structure, biosynthesis, and biodegradation of cutin and suberin . Annu Rev Plant Physiol 32 : 539 – 567 Google Scholar Crossref Search ADS WorldCat Koornneef M , Hanhart CJ, Thiel F ( 1989 ) A genetic and phenotypic description of eceriferum (cer) mutants in Arabidopsis thaliana . J Hered 80 : 118 – 122 Google Scholar Crossref Search ADS WorldCat Krauss P , Markstadter C, Riederer M ( 1997 ) Attenuation of UV radiation by plant cuticles from woody species . Plant Cell Environ 20 : 1079 – 1085 Google Scholar Crossref Search ADS WorldCat Lee YH , Dean RA ( 1994 ) Hydrophobicity of contact surface induces appressorium formation in Magnaporthe grisea . FEMS Microbiol Lett 115 : 71 – 75 Google Scholar Crossref Search ADS WorldCat Littlejohn GR , Gouveia JD, Edner C, Smirnoff N, Love J ( 2010 ) Perfluorodecalin enhances in vivo confocal microscopy resolution of Arabidopsis thaliana mesophyll . New Phytol 186 : 1018 – 1025 Google Scholar Crossref Search ADS PubMed WorldCat Littlejohn GR , Mansfield JC, Christmas JT, Witterick E, Fricker MD, Grant MR, Smirnoff N, Everson RM, Moger J, Love J (April 23, 2014) An update: improvements in imaging perfluorocarbon-mounted plant leaves with implications for studies of plant pathology, physiology, development and cell biology. Front Plant Sci http://dx.doi.org/10.3389/fpls.2014.00140 Lu S , Min W, Chong S, Holtom GR, Xie XS ( 2010 ) Label-free imaging of heme proteins with two-photon excited photothermal lens microscopy . Appl Phys Lett 96 : 113701 Google Scholar Crossref Search ADS WorldCat Ly S , McNerney G, Fore S, Chan J, Huser T ( 2007 ) Time-gated single photon counting enables separation of CARS microscopy data from multiphoton-excited tissue autofluorescence . Opt Express 15 : 16839 – 16851 Google Scholar Crossref Search ADS PubMed WorldCat Mansfield JC , Littlejohn GR, Seymour MP, Lind RJ, Perfect S, Moger J ( 2013 ) Label-free chemically specific imaging in planta with stimulated Raman scattering microscopy . Anal Chem 85 : 5055 – 5063 Google Scholar Crossref Search ADS PubMed WorldCat Min W , Freudiger CW, Lu SJ, Xie XS ( 2011 ) Coherent nonlinear optical imaging: beyond fluorescence microscopy . Annu Rev Phys Chem 62 : 507 – 530 Google Scholar Crossref Search ADS PubMed WorldCat Moger J , Garrett NL, Begley D, Mihoreanu L, Lalatsa A, Lozano MV, Mazza M, Schatzlein A, Uchegbu I ( 2012 ) Imaging cortical vasculature with stimulated Raman scattering and two-photon photothermal lensing microscopy . J Raman Spectrosc 43 : 668 – 674 Google Scholar Crossref Search ADS WorldCat Pighin JA , Zheng H, Balakshin LJ, Goodman IP, Western TL, Jetter R, Kunst L, Samuels AL ( 2004 ) Plant cuticular lipid export requires an ABC transporter . Science 306 : 702 – 704 Google Scholar Crossref Search ADS PubMed WorldCat Raffaele S , Leger A, Roby D ( 2009 ) Very long chain fatty acid and lipid signaling in the response of plants to pathogens . Plant Signal Behav 4 : 94 – 99 Google Scholar Crossref Search ADS PubMed WorldCat Ristic Z , Jenks MA ( 2002 ) Leaf cuticle and water loss in maize lines differing in dehydration avoidance . J Plant Physiol 159 : 645 – 651 Google Scholar Crossref Search ADS WorldCat Samuels L , Kunst L, Jetter R ( 2008 ) Sealing plant surfaces: cuticular wax formation by epidermal cells . Annu Rev Plant Biol 59 : 683 – 707 Google Scholar Crossref Search ADS PubMed WorldCat Snyder R , Hsu S, Krimm S ( 1978 ) Vibrational spectra in the C—H stretching region and the structure of the polymethylene chain . Spectrochim Acta [A] 34 : 395 – 406 Google Scholar Crossref Search ADS WorldCat Teusink RS , Rahman M, Bressan RA, Jenks MA ( 2002 ) Cuticular waxes on Arabidopsis thaliana close relatives Thellungiella halophila and Thellungiella parvula . Int J Plant Sci 163 : 309 – 315 Google Scholar Crossref Search ADS WorldCat Veraverbeke EA , Van Bruaene N, Van Oostveldt P, Nicolaï BM ( 2001 ) Non destructive analysis of the wax layer of apple (Malus domestica Borkh.) by means of confocal laser scanning microscopy . Planta 213 : 525 – 533 Google Scholar Crossref Search ADS PubMed WorldCat Weissflog I , Vogler N, Akimov D, Dellith A, Schachtschabel D, Svatos A, Boland W, Dietzek B, Popp J ( 2010 ) Toward in vivo chemical imaging of epicuticular waxes . Plant Physiol 154 : 604 – 610 Google Scholar Crossref Search ADS PubMed WorldCat Xu X , Feng J, Lü S, Lohrey GT, An H, Zhou Y, Jenks MA ( 2014 ) Leaf cuticular lipids on the Shandong and Yukon ecotypes of saltwater cress, Eutrema salsugineum, and their response to water deficiency and impact on cuticle permeability . Physiol Plant 151 : 446 – 458 Google Scholar Crossref Search ADS PubMed WorldCat Ye T , Fu D, Warren WS ( 2009 ) Nonlinear absorption microscopy . Photochem Photobiol 85 : 631 – 645 Google Scholar Crossref Search ADS PubMed WorldCat Yeats TH , Rose JKC ( 2013 ) The formation and function of plant cuticles . Plant Physiol 163 : 5 – 20 Google Scholar Crossref Search ADS PubMed WorldCat Yu MML , Konorov SO, Schulze HG, Blades MW, Turner RFB, Jetter R ( 2008 ) In situ analysis by microspectroscopy reveals triterpenoid compositional patterns within leaf cuticles of Prunus laurocerasus . Planta 227 : 823 – 834 Google Scholar Crossref Search ADS PubMed WorldCat Yu Y , Ramachandran PV, Wang MC ( 2014 ) Shedding new light on lipid functions with CARS and SRS microscopy . Biochim Biophys Acta 1841 : 1120 – 1129 Google Scholar Crossref Search ADS PubMed WorldCat Zeng Y , Saar B, Friedrich M, Chen F, Liu YS, Dixon R, Himmel M, Xie X, Ding SY ( 2010 ) Imaging lignin-downregulated alfalfa using coherent anti-Stokes Raman scattering microscopy . BioEnergy Research 3 : 272 – 277 Google Scholar Crossref Search ADS WorldCat Zumbusch A , Holtom GR, Xie XS ( 1999 ) Three-dimensional vibrational imaging by coherent anti-Stokes Raman scattering . Phys Rev Lett 82 : 4142 – 4145 Google Scholar Crossref Search ADS WorldCat Author notes 1 This work was supported by the Biotechnology and Biological Sciences Research Council (grant to J.L. and J.M.), Shell Research Ltd. (grant to J.L.), and Syngenta (grant to J.M.). 2 These authors contributed equally to the article. * Address correspondence to [email protected] and [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Julian Moger ([email protected]). J.M. conceived the project; J.M., J.L., and N.S. designed the experiments; G.R.L. and J.C.M. performed the experiments; G.R.L., J.C.M., D.P., R.L., S.P., M.S., N.S., J.L., and J.M. analyzed the data; D.P., R.L., S.P., and M.S. contributed reagents and performed the analyses; G.R.L., J.C.M., N.S., J.L., and J.M. wrote the article. [OPEN] Articles can be viewed without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.15.00119 © 2015 American Society of Plant Biologists. All Rights Reserved. © The Author(s) 2015. Published by Oxford University Press on behalf of American Society of Plant Biologists. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Papanatsiou, Maria; Scuffi, Denisse; Blatt, Michael R.; García-Mata, Carlos

doi: 10.1104/pp.114.256057pmid: 25770153

Abstract Hydrogen sulfide (H2S) is the third biological gasotransmitter, and in animals, it affects many physiological processes by modulating ion channels. H2S has been reported to protect plants from oxidative stress in diverse physiological responses. H2S closes stomata, but the underlying mechanism remains elusive. Here, we report the selective inactivation of current carried by inward-rectifying K+ channels of tobacco (Nicotiana tabacum) guard cells and show its close parallel with stomatal closure evoked by submicromolar concentrations of H2S. Experiments to scavenge H2S suggested an effect that is separable from that of abscisic acid, which is associated with water stress. Thus, H2S seems to define a unique and unresolved signaling pathway that selectively targets inward-rectifying K+ channels. Hydrogen sulfide (H2S) is a small bioactive gas that has been known for centuries as an environmental pollutant (Reiffenstein et al., 1992). H2S is soluble in both polar and, especially, nonpolar solvents (Wang, 2002), and has recently come to be recognized as the third member of a group of so-called biological gasotransmitters. Most importantly, H2S shows both physical and functional similarities to the other gasotransmitters nitric oxide (NO) and carbon monoxide (Wang, 2002), and it has been shown to participate in diverse physiological processes in animals, including cardioprotection, neuromodulation, inflammation, apoptosis, and gastrointestinal functions among others (Kabil et al., 2014). Less is known about H2S molecular targets and its modes of action. H2S can directly modify specific targets through protein sulfhydration (the addition of an -SH group to thiol moiety of proteins; Mustafa et al., 2009) or reaction with metal centers (Li and Lancaster, 2013). It can also act indirectly, reacting with NO to form nitrosothiols (Whiteman et al., 2006; Li and Lancaster, 2013). Among its molecular targets, H2S has been reported to regulate ATP-dependent K+ channels (Yang et al., 2005), Ca2+-activated K+ channels, T- and L-type Ca2+ channels, and transient receptor potential channels (Tang et al., 2010; Peers et al., 2012), suggesting H2S as a key regulator of membrane ion transport. In plants, H2S is produced enzymatically by the desulfhydration of l-Cys to form H2S, pyruvate, and ammonia in a reaction catalyzed by the enzyme l-Cys desulfhydrase (Riemenschneider et al., 2005a, 2005b), DES1, that has been characterized in Arabidopsis (Arabidopsis thaliana; Alvarez et al., 2010). Alternatively, H2S can be produced from d-Cys by d-Cys desulfhydrase (Riemenschneider et al., 2005a, 2005b) and in cyanide metabolism by β-cyano-Ala synthase (García et al., 2010). H2S action was originally related to pathogenesis resistance (Bloem et al., 2004), but in the last decade it has been proven to have an active role in signaling, participating in key physiological processes, such as germination and root organogenesis (Zhang et al., 2008, 2009a), heat stress (Li et al., 2013a, 2013b), osmotic stress (Zhang et al., 2009b), and stomatal movement (García-Mata and Lamattina, 2010; Lisjak et al., 2010, 2011; Jin et al., 2013). Moreover, H2S was reported to participate in the signaling of plant hormones, including abscisic acid (ABA; García-Mata and Lamattina, 2010; Lisjak et al., 2010; Jin et al., 2013; Scuffi et al., 2014), ethylene (Hou et al., 2013), and auxin (Zhang et al., 2009a). ABA is an important player in plant physiology. Notably, upon water stress, ABA triggers a complex signaling network to restrict the loss of water through the transpiration stream, balancing these needs with those of CO2 for carbon assimilation. In the guard cells that surround the stomatal pore, ABA induces an increase of cytosolic-free Ca2+ concentration ([Ca2+]cyt), elevates cytosolic pH (pHi), and activates the efflux of anions, mainly chloride, through S- and R-type anion channels. The increase in [Ca2+]cyt inactivates inward-rectifying K+ channels (IKIN); anion efflux depolarizes the plasma membrane, and together with the rise in pHi, it activates K+ efflux through outward-rectifying K+ channels (IKOUT; Blatt, 2000; Schroeder et al., 2001). These changes in ion flux, in turn, generate an osmotically driven reduction in turgor and volume and closure of the stomatal pore. All three gasotransmitters have been implicated in regulating the activity of guard cell ion channels, but direct evidence is available only for NO (Garcia-Mata et al., 2003; Sokolovski et al., 2005). Here, we have used two-electrode voltage clamp measurements to study the role of H2S in the regulation of the guard cell K+ channels of tobacco (Nicotiana tabacum). Our results show that H2S selectively inactivates IKIN and that this action parallels that of stomatal closure. These results confirm H2S as a unique factor regulating guard cell ion transport and indicate that H2S acts in a manner separable from that of ABA. RESULTS AND DISCUSSION To address whether H2S-induced stomatal closure is mediated by changes in the activities of these channels, we recorded currents by two-electrode voltage clamp. IKIN and IKOUT currents were resolved on the basis of their voltage-dependent kinetics and challenging with H2S donor GYY4137 (for p-methoxyphenyl(morpholino)phosphinodithioic acid). Follow impalement current-voltage recordings were carried out to confirm a stable baseline of channel activities for 5 to 10 min before the H2S donor GYY4137 was added. We observed a complete response to the H2S donor within 2 to 3 min of additions, indicating a halftime for response to the donor of less than 120 s. Most impalements could be held for 20 to 30 min only and therefore, did not allow us to assess current recovery after washout of the H2S donor. Figure 1A shows current traces and the mean steady-state currents as a function of voltage (I-V curves) from guard cells before and after 5 min of exposure to 10 μm GYY4137. The I-V curves show the characteristic voltage dependence of both IKOUT and IKIN as previously described (Blatt, 1992; Gradmann et al., 1993; Garcia-Mata et al., 2003). In 10 mm KCl, voltages positive of −40 mV yielded IKOUT that increased in amplitude with the voltage. Mean IKOUT at +30 mV was +120 ± 28 and +91 ± 30 μA cm−2 before and after exposure to GYY 4137, respectively, indicating a small but not very significant effect on the IKOUT. Voltages negative of −100 mV were marked by a strong inward-directed current typical of IKIN, and the current amplitude increased with negative-going voltages. We found that IKIN at −220 mV was reduced by roughly 90% by H2S donor treatments, yielding a mean IKIN of −21 ± 8 μA cm−2 compared with −169 ± 12 μA cm−2 for the control. Exposure to the H2S donor also affected the halftimes for IKIN activation. Mean halftimes for IKIN activation at −220 mV were 710 ± 70 ms in the control and 1,230 ± 230 ms after exposure to 10 μm GYY4137, indicating a significant change in gating of IKIN (Fig. 1B). Figure 1. Open in new tabDownload slide H2S selectively affects IKIN. A, Currents through IKOUT and IKIN above and below the voltage axis, respectively, recorded under voltage clamp from tobacco guard cells. Voltages were clamped from a holding voltage of −100 mV in 6-s steps between −220 and +120 mV and 4-s steps between −80 and +40 mV. Data are means ± se of n = 5 guard cells bathed in 5 mm Ca2+-MES buffer (pH 6.1) with 10 mm KCl before (white circles) and 5 min after (black circles) adding 10 μm GYY4137. Insets, Current traces for IKOUT and IKIN for control and H2S treatment. Currents are cross referenced to the current-voltage curves by symbols. Bars = 2 s (horizontal); 50 μA cm−2 (vertical). B, Mean activation halftimes as a function of voltage derived from current traces, including those in A, and cross referenced by symbols. C, Steady-state conductance as a function of voltage normalized to G max in the control (black line) and with 10 μm GYY4137 (dashed line). Inset, Steady-state conductance as a function of voltage. Curves were jointly fitted to Boltzmann function (Eq. 1) with gating charge-δ held in common. Figure 1. Open in new tabDownload slide H2S selectively affects IKIN. A, Currents through IKOUT and IKIN above and below the voltage axis, respectively, recorded under voltage clamp from tobacco guard cells. Voltages were clamped from a holding voltage of −100 mV in 6-s steps between −220 and +120 mV and 4-s steps between −80 and +40 mV. Data are means ± se of n = 5 guard cells bathed in 5 mm Ca2+-MES buffer (pH 6.1) with 10 mm KCl before (white circles) and 5 min after (black circles) adding 10 μm GYY4137. Insets, Current traces for IKOUT and IKIN for control and H2S treatment. Currents are cross referenced to the current-voltage curves by symbols. Bars = 2 s (horizontal); 50 μA cm−2 (vertical). B, Mean activation halftimes as a function of voltage derived from current traces, including those in A, and cross referenced by symbols. C, Steady-state conductance as a function of voltage normalized to G max in the control (black line) and with 10 μm GYY4137 (dashed line). Inset, Steady-state conductance as a function of voltage. Curves were jointly fitted to Boltzmann function (Eq. 1) with gating charge-δ held in common. Steady-state current through any ion channel depends on the ensemble conductance (G), which is the product of the number of functional channels at the plasma membrane (N), the single-channel conductance for a given ion species (γX), and the gating characteristics of the channel that describe the open probability of the channel (P o). Plotting the conductance of IKIN before and after exposure to H2S as a function of voltage allowed a separation of the differences in the gating characteristics before and after H2S donor treatments (Fig. 1C). The G-V curves were jointly fitted to a modified Boltzmann function (Eq. 1) to determine the maximum conductance (G max) and the gating characteristics of IKIN (Table I). For joint fittings, δ was held in common, and it yielded statistically and visually satisfactory fittings with a value of −1.66 ± 0.04. As expected from the I-V data, the H2S donor suppressed the G max significantly by up to 90% relative to the control. We cannot distinguish from these data whether this effect was mediated through a change in the number of channels available for activation (N) or the single-channel conductance (γK). Such detail would require single-channel analysis. However, we noted that the H2S donor displaced V 1/2 by approximately −12 mV (Fig. 1C; Table I), indicating that the H2S not only resulted in a decrease of the maximum conductance but also affected the voltage dependence for gating of IKIN. An action on V 1/2 cannot be explained solely by an effect on N or γK. In short, H2S selectively inactivated IKIN. Fitted gating characteristics for IKIN Table I. Fitted gating characteristics for IKIN Statistical differences after ANOVA as determined by Student Newman-Keuls test. P values are indicated for each parameter comparing control and H2S treatments. . V 1/2 (P = 0.006) . δ . G KIN (P < 0.001) . mV μs cm−2 Control −183 ± 0.5 −1.67 ± 0.04 1.15 ± 0.01 +10 μm GYY4137 −195 ± 3 0.15 ± 0.01 . V 1/2 (P = 0.006) . δ . G KIN (P < 0.001) . mV μs cm−2 Control −183 ± 0.5 −1.67 ± 0.04 1.15 ± 0.01 +10 μm GYY4137 −195 ± 3 0.15 ± 0.01 Open in new tab Table I. Fitted gating characteristics for IKIN Statistical differences after ANOVA as determined by Student Newman-Keuls test. P values are indicated for each parameter comparing control and H2S treatments. . V 1/2 (P = 0.006) . δ . G KIN (P < 0.001) . mV μs cm−2 Control −183 ± 0.5 −1.67 ± 0.04 1.15 ± 0.01 +10 μm GYY4137 −195 ± 3 0.15 ± 0.01 . V 1/2 (P = 0.006) . δ . G KIN (P < 0.001) . mV μs cm−2 Control −183 ± 0.5 −1.67 ± 0.04 1.15 ± 0.01 +10 μm GYY4137 −195 ± 3 0.15 ± 0.01 Open in new tab The inactivation of the K+ current is consistent with GYY4137 action in suppressing K+ uptake and promoting stomatal closure, and it argues against earlier (and statistically undocumented) claims that H2S donors promote stomatal opening (Lisjak et al., 2010, 2011). To assess the action of H2S effect on stomatal movement, we measured apertures from stomata treated with different concentrations of the H2S donor. Epidermal peels were placed in opening buffer under light of 150 μmol m−2 s−1 for 2 h to open the stomata before transfer to 5 mm Ca2+-MES (pH 6.1) with 60 mm KCl supplemented with 0, 0.1, 1, or 10 μm GYY4137 for 90 min. Apertures were recorded immediately before and after H2S treatments, and the data were normalized to the controls (Fig. 2). Exposure to the control buffer alone yielded stomatal apertures of 6.6 ± 1.8 μm; treatments with H2S donor resulted in stomatal apertures ranging from 5.9 to 4.8 μm. Fitting the data to the hyperbolic decay function yielded an apparent K i of 160 ± 40 nΜ for GYY4137. We carried out parallel measurements of IKIN to determine its dose dependence after treating guard cells in buffer supplemented with 0, 0.1, 1, or 10 μm GYY4137. Figure 2 also shows the mean values for IKIN again normalized to the control treatment. Increasing the concentration of H2S donor, indeed, enhanced IKIN inactivation. Fitting these data to hyperbolic decay function gave a K i of 120 ± 70 nm for GYY4137, a value that did not significantly differ from that for stomatal closure compared with t test (P = 0.735). These results, thus, confirm the close kinetic relationship between IKIN inactivation and stomatal closure in H2S. Figure 2. Open in new tabDownload slide H2S affects stomatal aperture and K+ current in a dose-dependent manner with similar concentration dependencies. Mean stomatal apertures ± se (n = 50) recorded from epidermal peels of tobacco (triangles) and mean IKIN ± se (n = 5) at −200 mV (squares) normalized to the controls without GYY 4137 treatment. A hyperbolic decay function was fitted to each set of data, yielding a K i of 160 ± 40 nm GYY4137 for the aperture response (solid line) and 120 ± 70 nm GYY4137 for current inactivation (dashed line). Figure 2. Open in new tabDownload slide H2S affects stomatal aperture and K+ current in a dose-dependent manner with similar concentration dependencies. Mean stomatal apertures ± se (n = 50) recorded from epidermal peels of tobacco (triangles) and mean IKIN ± se (n = 5) at −200 mV (squares) normalized to the controls without GYY 4137 treatment. A hyperbolic decay function was fitted to each set of data, yielding a K i of 160 ± 40 nm GYY4137 for the aperture response (solid line) and 120 ± 70 nm GYY4137 for current inactivation (dashed line). The similar effects of H2S and ABA on IKIN and stomatal aperture prompted us to explore the connection between H2S and ABA signaling, which was suggested by García-Mata and Lamattina (2010). A similar set of protocols was used as above. Epidermal peels were pretreated for 2 h with opening buffer and light for 90 min before treatments in 5 mm Ca2+-MES (pH 6.1) with 10 mm KCl with and without supplement of five distinct combinations of stomatal effectors: 10 μm GYY4137, 10 μm hypotaurine (HT), 10 μm GYY4137 + 10 μm HT, 20 μm ABA, and 20 μm ABA + 10 μm HT. HT interacts with free sulfide to form thiotaurine, effectively scavenging free H2S in solution (Ortega et al., 2008). Figure 3A shows the percentage of stomatal closure induced by each treatment relative to the control. Measurements were carried out separately at the Consejo Nacional de Investigaciones Científicas y Técnicas and yielded starting apertures (4.9 ± 0.1 μm) that differed quantitatively from those recorded at the University of Glasgow. Qualitatively, however, the results were consistent between data sets. Exposure to 10 μm GYY4137 and 20 μm ABA resulted in 60% and 80% reductions in stomatal aperture, corresponding to apertures of 3.2 ± 0.1 and 2.2 ± 0.07 μm, respectively. Treatment with HT alone had no effect on stomatal aperture. When epidermal peels were treated with both 10 μm GYY4137 and HT, the effect of the H2S donor was alleviated, yielding apertures of 4.6 ± 0.1 μm, similar to those of the control. Treatment of epidermal peels with 20 μm ABA + 10 μm HT partially suppressed the effect of ABA on aperture, resulting in a reduction to 70% pore width compared with control treatment. Figure 3. Open in new tabDownload slide ABA and H2S affect aperture and IKIN in parallel. A, Mean stomatal apertures ± se (n > 190 per treatment), including the control (100%), after treatments with 10 μm GYY4137 (GYY; dark-gray bar), 10 μm HT (black bar), 10 μm GYY4137 + 10 μm HT (white striped bar), 20 μm ABA (light-gray bar), or 20 μm ABA + 10 μm HT (light-gray striped bar). Letters indicates statistical differences by ANOVA (P < 0.05) as determined by Student Newman-Keuls test. B, Mean current ± se (n = 5) for IKIN recorded under voltage clamp as in Figure 1 and normalized to IKIN at −200 mV in the control. Letters indicates statistical differences by ANOVA (P < 0.05) as determined by Student Newman-Keuls test. Figure 3. Open in new tabDownload slide ABA and H2S affect aperture and IKIN in parallel. A, Mean stomatal apertures ± se (n > 190 per treatment), including the control (100%), after treatments with 10 μm GYY4137 (GYY; dark-gray bar), 10 μm HT (black bar), 10 μm GYY4137 + 10 μm HT (white striped bar), 20 μm ABA (light-gray bar), or 20 μm ABA + 10 μm HT (light-gray striped bar). Letters indicates statistical differences by ANOVA (P < 0.05) as determined by Student Newman-Keuls test. B, Mean current ± se (n = 5) for IKIN recorded under voltage clamp as in Figure 1 and normalized to IKIN at −200 mV in the control. Letters indicates statistical differences by ANOVA (P < 0.05) as determined by Student Newman-Keuls test. Given the role of [Ca2+]cyt in ABA signaling and control of IKIN (Blatt, 2000; Garcia-Mata et al., 2003), we sought to test whether the H2S-induced effect on IKIN might be mediated by the Ca2+ intermediate. For this purpose, we loaded guard cells from the microelectrode with 50 mm EGTA, which chelates and buffers Ca2+ to suppress its elevation (Grabov and Blatt, 1998; Chen et al., 2010; Wang et al., 2012). After being impaled, guard cells were held for a period of 5 min to ensure loading with EGTA. Thereafter, the guard cells were either maintained in 5 mm Ca2+-MES (pH 6.1) with 10 mm KCl or challenged with 10 μm GYY4137 for a period of 10 min. In the absence of H2S donor, we observed no substantive effect on IKIN. The mean amplitude at −200 mV was −217 ± 29 μA cm−2. In the presence of H2S donor, IKIN was suppressed, yielding a mean current of −61 ± 11 μA cm−2 (Fig. 4A). EGTA did yield a small but not very significant recovery of IKIN in the presence of the H2S donor (Fig. 4B). These results indicate that H2S acts in a manner that is largely independent of [Ca2+]cyt. Figure 4. Open in new tabDownload slide H2S inactivates currents from IKIN in a Ca2+-independent manner. A, Current-voltage (I-V) curves for IKIN recorded under voltage clamp as in Figure 1. Guard cells were bathed in 10 mm KCl (white circles) or 10 mm KCl supplemented with 10 μm GYY4137 (gray circles) and loaded from the microelectrode with 50 mm EGTA. The I-V curve for guard cells treated with only 10 μm GYY4137 (black circles) from Figure 1 is included for visual reference. Data are means ± se of n = 5 guard cells for each data set. Curves were jointly fitted to Boltzmann function (lines), with V 1/2 and gating charge-δ held in common. Insets present current traces recorded under voltage clamp. Bars = 2 s (horizontal) and 200 μA cm−2 (vertical). B, Mean IKIN at −205 mV from the current recordings for control (white bars) and 10 μm GYY4137 (black bars) treatments in the presence of EGTA and 10 μm GYY4137 without EGTA (gray bars). Lettering indicates statistical differences by ANOVA (P < 0.001) as determined by Student Newman-Keuls test. Figure 4. Open in new tabDownload slide H2S inactivates currents from IKIN in a Ca2+-independent manner. A, Current-voltage (I-V) curves for IKIN recorded under voltage clamp as in Figure 1. Guard cells were bathed in 10 mm KCl (white circles) or 10 mm KCl supplemented with 10 μm GYY4137 (gray circles) and loaded from the microelectrode with 50 mm EGTA. The I-V curve for guard cells treated with only 10 μm GYY4137 (black circles) from Figure 1 is included for visual reference. Data are means ± se of n = 5 guard cells for each data set. Curves were jointly fitted to Boltzmann function (lines), with V 1/2 and gating charge-δ held in common. Insets present current traces recorded under voltage clamp. Bars = 2 s (horizontal) and 200 μA cm−2 (vertical). B, Mean IKIN at −205 mV from the current recordings for control (white bars) and 10 μm GYY4137 (black bars) treatments in the presence of EGTA and 10 μm GYY4137 without EGTA (gray bars). Lettering indicates statistical differences by ANOVA (P < 0.001) as determined by Student Newman-Keuls test. We also investigated the effect of the above compounds and their combinations on IKIN, again following the same set of protocols. Figure 3B displays the mean percentage reduction of the IKIN amplitude at −200 mV before and after the exposure to each of the treatments. H2S resulted in almost complete loss of IKIN, which is shown in Figure 1. ABA treatment reduced IKIN by 62%, resulting in a mean current of −170 ± 39 μA cm−2 at −200 mV. Interestingly, exposure of guard cells to 10 μm HT yielded IKIN of −292 ± 64 μA cm−2, marginally greater in amplitude compared with −241 ± 40 μA cm−2 for the control, although this difference was not very significant. Suppression of the current by H2S was blocked when guard cells were treated with the combination of H2S donor and scavenger, resulting in IKIN of similar amplitude as the control treatment. In contrast, the reduction of IKIN evoked by ABA was not prevented by adding HT, which yielded a mean IKIN of −174 ± 35 μA cm−2. Altogether, these data indicate that H2S acts in a manner paralleling that either of ABA or upstream of the hormone. Stomatal movement is a highly coordinated process that is generally recognized to engage several signaling networks leading to the regulation of K+ channels, anion channels, and H+-ATPases at the plasma membrane as well as a complementary assembly of transporters at the tonoplast (Blatt, 2000). For ABA-evoked stomatal closure, this process includes inactivation of IKIN through changes in [Ca2+]cyt and activation of the IKOUT mediated by a rise in pHi (Blatt, 1990; Blatt and Armstrong, 1993; Garcia-Mata et al., 2003; Siegel et al., 2009). Our findings that H2S differentially affects IKIN and IKOUT and that IKIN inactivation is dose dependent with an apparent K i in the low nanomolar range confirm a subcellular target for the action of this gasotransmitter. The timescale of the H2S-triggered changes in channel gating is entirely in keeping with posttranslational regulation, which is in contrast with the slower effects of ABA that, over timescales of many minutes or hours, clearly rely on the transcription regulation and trafficking of the channel proteins (Pilot et al., 2003; Sutter et al., 2007; Eisenach, et al., 2012, 2014). These findings together with evidence that H2S mediates stomatal closure and that its scavenging partially suppresses closure in ABA suggest a connection with the hormone, albeit a loose one. Notably, H2S scavenging failed to reverse ABA-evoked inactivation of IKIN (Fig. 3), and we found that stomatal closure was enhanced when treated with ABA and the H2S scavenger compared with treatment with ABA alone. These observations are difficult to reconcile with a role for H2S as an intermediate in ABA signaling per se and instead, suggest a partial overlap in signaling pathways. This interpretation is in agreement with recent studies showing an ABA dependency of H2S effect on stomatal movements (García-Mata and Lamattina, 2010; Scuffi et al., 2014). It also complements substantial evidence for a separate set of intermediates, especially ROS and NO, that trigger the elevation of [Ca2+]cyt and are important for ABA-mediated stomatal closure (Pei et al., 2000; García-Mata and Lamattina, 2002, 2003), and it agrees with our finding that Ca2+ buffering did not substantially rescue IKIN. Guard cells are thought to produce NO in response to ABA through the activity of nitrate reductases (Desikan et al., 2002), and NO action is also dependent on the secondary messengers cGMP and cADPR (Neill et al., 2002). Garcia-Mata et al. (2003) showed that NO promoted the inhibition of IKIN and activated anion efflux through an enhanced sensitivity of internal Ca2+ release to Ca2+ influx across the plasma membrane. Notably, the effects of [Ca2+]cyt on IKIN gating and especially, V 1/2 are much more pronounced than observed with H2S. Furthermore, NO is also able to modulate IKOUT by direct nitrosylation of the channel or an associated regulatory protein (Sokolovski and Blatt, 2004), but we observed little evidence of an effect on IKOUT current. Therefore, these observations implicate a parallel and as yet uncharacterized signaling pathway that acts on IKIN, thereby overlapping with the well-known pathways leading from ABA to IKIN and stomatal closure. How might H2S act to modulate IKIN? Nitrosylation of Cys sulfhydryl groups on either the channel itself or on closely associated regulatory proteins has been suggested to mediate the NO-induced block of IKOUT (Sokolovski and Blatt, 2004), and such modifications may be linked to ROS modification of residues within the voltage sensor domain (García et al., 2010). H2S is also capable of covalently modifying protein targets, and the mechanism is equally relevant to IKIN and the proteins that regulate these channels, including protein kinases and phosphatases (Thiel and Blatt, 1994; Li et al., 1998; Michard et al., 2005). In animals, for example, H2S activates ATP-dependent K+ channels through the sulfhydration of a Cys residue of the sulfonurea SUR protein (Babenko et al., 2000). Addition of the SUR inhibitor glibenclamide antagonized the H2S response and prevented the hypotensive effect of H2S (Zhao et al., 2001). In other cases, sulfhydration is suppressed by the reducing agent dithiothreitol and mutants defective in H2S production (Mustafa et al., 2009). Of interest, glibenclamide has also been shown to abolish stomatal closure triggered by ABA and external Ca2+ through the inhibition of anion and IKOUT (Leonhardt et al., 1999). More recent studies, however, have shown only partial suppression by glibenclamide of stomatal closure in ABA, whereas the response to H2S was completely abolished (García-Mata and Lamattina, 2010). These findings suggest that ABC proteins, a major target of glibenclamide, may contribute to channel regulation in guard cells upon H2S exposure. At present, however, there is not sufficient information from any system that would enable realistic predictions of the possible motifs for sulfhydration. No doubt, future studies with transgenic Arabidopsis lines defective in H2S, NO production, and ABA sensitivity should help clarify the role of H2S in these processes (Hetherington and Woodward, 2003). What is clear, however, is that H2S is active in selectively regulating IKIN of guard cells over timescales consistent with short-term posttranslational modification of specific target proteins. Furthermore, our evidence implicates H2S in a signaling pathway that is separable from that of ABA, although both ABA and H2S modulate stomatal behavior in parallel. MATERIALS AND METHODS Plant Material, Chemicals, and Stomatal Assays Tobacco (Nicotiana tabacum) plants were grown in Levington F2+S compost under long-day conditions (16-h-light/8-h-dark cycle; temperature approximately 26°C and 22°C for day and night, respectively; relative humidity of 60% and 70% for day and night, respectively) under 100 μmol m−2 s−1 of light. Epidermal peels were obtained from the abaxial side of tobacco leaves and placed in opening buffer comprised of 10 mm Na+-MES (pH 6.1; 10 mm MES titrated to pH 6.1 with NaOH) with 60 mm KCl under light of 150 μmol m−2 s−1 for 2 h before treatment with the H2S donor GYY4137 (Sigma) in the same buffer. Stomata were imaged before and after 90 min of H2S treatment using an LD Achroplan 40× Objective and an Axio-Cam HRc Digital Camera (Zeiss). Apertures were tracked for individual stomata and quantified using IMAGEJ version 1.48 (image.nih.gov/ij/). Guard Cell Electrophysiology Currents were recorded under two-electrode voltage clamp using Henry’s EP Software Suite (http://www.psrg.org.uk). Microelectrodes were constructed to give tip resistances of greater than 100 MΩ and filled with 200 mm K+-acetate (pH 7.5) to minimize interference arising from anion leakage from the microelectrode (Blatt and Slayman, 1983; Blatt, 1987; Wang and Blatt, 2011). Electrolyte filling solutions were equilibrated against the resin-bound Ca2+ buffer BAPTA [for 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid; Ca2+ sponge; Invitrogen] to prevent Ca2+ loading of the cytosol from the microelectrodes. K+ channel currents were recorded from guard cells bathed under continuous superfusion with 5 mm Ca2+-MES (pH 6.1; 5 mm MES titrated to pH 6.1 with Ca(OH)2; [Ca2+] = 1 mm) plus 10 mm KCl alone and supplemented with reagents as indicated. Recordings typically included a 2-s holding voltage at −100 mV and two to six steps to voltages between −220 and +40 mV. Surface areas of the impaled guard cells were calculated assuming a spheroid geometry (Blatt et al., 1987). Current voltage analysis and fittings were carried out using Henry’s EP Software Suite and SigmaPlot 11 (SPSS; Systat Software). Conductance-voltage curves were fitted by joint nonlinear least squares and the Marquardt-Levenberg algorithm using a modified Boltzmann function of the form (1) where G max is the maximum conductance, V is the membrane voltage, V 1/2 is the voltage at which half-maximum activation of channels occurs, δ is the apparent gating charge, and F, R, and T have their usual meanings. Statistical Analysis Results are reported as means ± se of n observations, with significance determined using Student’s t test and ANOVA at P < 0.05. ACKNOWLEDGMENTS We thank Amparo Ruiz-Prado for support in plant growth and maintenance. Glossary ABA abscisic acid [Ca2+]cyt cytosolic-free Ca2+ concentration HT hypotaurine H2S hydrogen sulfide IKIN inward-rectifying K+ channel IKOUT outward-rectifying K+ channel NO nitric oxide pHi cytosolic pH LITERATURE CITED Alvarez C , Calo L, Romero LC, García I, Gotor C ( 2010 ) An O-acetylserine(thiol)lyase homolog with l-cysteine desulfhydrase activity regulates cysteine homeostasis in Arabidopsis . Plant Physiol 152 : 656 – 669 Google Scholar Crossref Search ADS PubMed WorldCat Babenko AP , Gonzalez G, Bryan J ( 2000 ) Pharmaco-topology of sulfonylurea receptors. Separate domains of the regulatory subunits of K(ATP) channel isoforms are required for selective interaction with K(+) channel openers . J Biol Chem 275 : 717 – 720 Google Scholar Crossref Search ADS PubMed WorldCat Blatt MR ( 1987 ) Electrical characteristics of stomatal guard cells: the ionic basis of the membrane potential and the consequence of potassium chlorides leakage from microelectrodes . Planta 170 : 272 – 287 Google Scholar Crossref Search ADS PubMed WorldCat Blatt MR ( 1990 ) Potassium channel currents in intact stomatal guard cells: rapid enhancement by abscisic acid . Planta 180 : 445 – 455 Google Scholar Crossref Search ADS PubMed WorldCat Blatt MR ( 1992 ) K+ channels of stomatal guard cells. Characteristics of the inward rectifier and its control by pH . J Gen Physiol 99 : 615 – 644 Google Scholar Crossref Search ADS PubMed WorldCat Blatt MR ( 2000 ) Cellular signaling and volume control in stomatal movements in plants . Annu Rev Cell Dev Biol 16 : 221 – 241 Google Scholar Crossref Search ADS PubMed WorldCat Blatt MR , Armstrong F ( 1993 ) K+ channels of stomatal guard cells: abscisic-acid-evoked control of the outward rectifier mediated by cytoplasmic pH . Planta 191 : 330 – 341 Google Scholar Crossref Search ADS WorldCat Blatt MR , Rodriguez-Navarro A, Slayman CL ( 1987 ) Potassium-proton symport in Neurospora: kinetic control by pH and membrane potential . J Membr Biol 98 : 169 – 189 Google Scholar Crossref Search ADS PubMed WorldCat Blatt MR , Slayman CL ( 1983 ) KCl leakage from microelectrodes and its impact on the membrane parameters of a nonexcitable cell . J Membr Biol 72 : 223 – 234 Google Scholar Crossref Search ADS PubMed WorldCat Bloem E , Riemenschneider A, Volker J, Papenbrock J, Schmidt A, Salac I, Haneklaus S, Schnug E ( 2004 ) Sulphur supply and infection with Pyrenopeziza brassicae influence L-cysteine desulphydrase activity in Brassica napus L . J Exp Bot 55 : 2305 – 2312 Google Scholar Crossref Search ADS PubMed WorldCat Chen ZH , Hills A, Lim CK, Blatt MR ( 2010 ) Dynamic regulation of guard cell anion channels by cytosolic free Ca2+ concentration and protein phosphorylation . Plant J 61 : 816 – 825 Google Scholar Crossref Search ADS PubMed WorldCat Desikan R , Griffiths R, Hancock J, Neill S ( 2002 ) A new role for an old enzyme: nitrate reductase-mediated nitric oxide generation is required for abscisic acid-induced stomatal closure in Arabidopsis thaliana . Proc Natl Acad Sci USA 99 : 16314 – 16318 Google Scholar Crossref Search ADS PubMed WorldCat Eisenach C , Chen ZH, Grefen C, Blatt MR ( 2012 ) The trafficking protein SYP121 of Arabidopsis connects programmed stomatal closure and K⁺ channel activity with vegetative growth . Plant J 69 : 241 – 251 Google Scholar Crossref Search ADS PubMed WorldCat Eisenach C , Papanatsiou M, Hillert EK, Blatt MR ( 2014 ) Clustering of the K+ channel GORK of Arabidopsis parallels its gating by extracellular K+ . Plant J 78 : 203 – 214 Google Scholar Crossref Search ADS PubMed WorldCat García I , Castellano JM, Vioque B, Solano R, Gotor C, Romero LC ( 2010 ) Mitochondrial beta-cyanoalanine synthase is essential for root hair formation in Arabidopsis thaliana . Plant Cell 22 : 3268 – 3279 Google Scholar Crossref Search ADS PubMed WorldCat Garcia-Mata C , Gay R, Sokolovski S, Hills A, Lamattina L, Blatt MR ( 2003 ) Nitric oxide regulates K+ and Cl- channels in guard cells through a subset of abscisic acid-evoked signaling pathways . Proc Natl Acad Sci USA 100 : 11116 – 11121 Google Scholar Crossref Search ADS PubMed WorldCat García-Mata C , Lamattina L ( 2002 ) Nitric oxide and abscisic acid cross talk in guard cells . Plant Physiol 128 : 790 – 792 Google Scholar Crossref Search ADS PubMed WorldCat García-Mata C , Lamattina L ( 2003 ) Abscisic acid, nitric oxide and stomatal closure - is nitrate reductase one of the missing links? Trends Plant Sci 8 : 20 – 26 Google Scholar Crossref Search ADS PubMed WorldCat García-Mata C , Lamattina L ( 2010 ) Hydrogen sulphide, a novel gasotransmitter involved in guard cell signalling . New Phytol 188 : 977 – 984 Google Scholar Crossref Search ADS PubMed WorldCat Grabov A , Blatt MR ( 1998 ) Membrane voltage initiates Ca2+ waves and potentiates Ca2+ increases with abscisic acid in stomatal guard cells . Proc Natl Acad Sci USA 95 : 4778 – 4783 Google Scholar Crossref Search ADS PubMed WorldCat Gradmann D , Blatt MR, Thiel G ( 1993 ) Electrocoupling of ion transporters in plants . J Membr Biol 136 : 327 – 332 Google Scholar Crossref Search ADS PubMed WorldCat Hetherington AM , Woodward FI ( 2003 ) The role of stomata in sensing and driving environmental change . Nature 424 : 901 – 908 Google Scholar Crossref Search ADS PubMed WorldCat Hou Z , Wang L, Liu J, Hou L, Liu X ( 2013 ) Hydrogen sulfide regulates ethylene-induced stomatal closure in Arabidopsis thaliana . J Integr Plant Biol 55 : 277 – 289 Google Scholar Crossref Search ADS PubMed WorldCat Jin Z , Xue S, Luo Y, Tian B, Fang H, Li H, Pei Y ( 2013 ) Hydrogen sulfide interacting with abscisic acid in stomatal regulation responses to drought stress in Arabidopsis . Plant Physiol Biochem 62 : 41 – 46 Google Scholar Crossref Search ADS PubMed WorldCat Kabil O , Motl N, Banerjee R ( 2014 ) H2S and its role in redox signaling . Biochim Biophys Acta 1844 : 1355 – 1366 Google Scholar Crossref Search ADS PubMed WorldCat Leonhardt N , Vavasseur A, Forestier C ( 1999 ) ATP binding cassette modulators control abscisic acid-regulated slow anion channels in guard cells . Plant Cell 11 : 1141 – 1152 Google Scholar PubMed OpenURL Placeholder Text WorldCat Li J , Lee YR, Assmann SM ( 1998 ) Guard cells possess a calcium-dependent protein kinase that phosphorylates the KAT1 potassium channel . Plant Physiol 116 : 785 – 795 Google Scholar Crossref Search ADS PubMed WorldCat Li Q , Lancaster JR ( 2013 ) Chemical foundations of hydrogen sulfide biology . Nitric Oxide 35 : 21 – 34 Google Scholar Crossref Search ADS PubMed WorldCat Li ZG , Ding XJ, Du PF ( 2013 a ) Hydrogen sulfide donor sodium hydrosulfide-improved heat tolerance in maize and involvement of proline . J Plant Physiol 170 : 741 – 747 Google Scholar Crossref Search ADS PubMed WorldCat Li ZG , Yang SZ, Long WB, Yang GX, Shen ZZ ( 2013 b ) Hydrogen sulphide may be a novel downstream signal molecule in nitric oxide-induced heat tolerance of maize (Zea mays L.) seedlings . Plant Cell Environ 36 : 1564 – 1572 Google Scholar Crossref Search ADS PubMed WorldCat Lisjak M , Srivastava N, Teklic T, Civale L, Lewandowski K, Wilson I, Wood ME, Whiteman M, Hancock JT ( 2010 ) A novel hydrogen sulfide donor causes stomatal opening and reduces nitric oxide accumulation . Plant Physiol Biochem 48 : 931 – 935 Google Scholar Crossref Search ADS PubMed WorldCat Lisjak M , Teklić T, Wilson ID, Wood M, Whiteman M, Hancock JT ( 2011 ) Hydrogen sulfide effects on stomatal apertures . Plant Signal Behav 6 : 1444 – 1446 Google Scholar Crossref Search ADS PubMed WorldCat Michard E , Lacombe B, Porée F, Mueller-Roeber B, Sentenac H, Thibaud JB, Dreyer I ( 2005 ) A unique voltage sensor sensitizes the potassium channel AKT2 to phosphoregulation . J Gen Physiol 126 : 605 – 617 Google Scholar Crossref Search ADS PubMed WorldCat Mustafa AK , Gadalla MM, Sen N, Kim S, Mu W, Gazi SK, Barrow RK, Yang G, Wang R, Snyder SH ( 2009 ) H2S signals through protein S-sulfhydration . Sci Signal 2 : ra72 Google Scholar PubMed OpenURL Placeholder Text WorldCat Neill SJ , Desikan R, Clarke A, Hancock JT ( 2002 ) Nitric oxide is a novel component of abscisic acid signaling in stomatal guard cells . Plant Physiol 128 : 13 – 16 Google Scholar Crossref Search ADS PubMed WorldCat Ortega JA , Ortega JM, Julian D ( 2008 ) Hypotaurine and sulfhydryl-containing antioxidants reduce H2S toxicity in erythrocytes from a marine invertebrate . J Exp Biol 211 : 3816 – 3825 Google Scholar Crossref Search ADS PubMed WorldCat Peers C , Bauer CC, Boyle JP, Scragg JL, Dallas ML ( 2012 ) Modulation of ion channels by hydrogen sulfide . Antioxid Redox Signal 17 : 95 – 105 Google Scholar Crossref Search ADS PubMed WorldCat Pei ZM , Murata Y, Benning G, Thomine S, Klüsener B, Allen GJ, Grill E, Schroeder JI ( 2000 ) Calcium channels activated by hydrogen peroxide mediate abscisic acid signalling in guard cells . Nature 406 : 731 – 734 Google Scholar Crossref Search ADS PubMed WorldCat Pilot G , Gaymard F, Mouline K, Chérel I, Sentenac H ( 2003 ) Regulated expression of Arabidopsis shaker K+ channel genes involved in K+ uptake and distribution in the plant . Plant Mol Biol 51 : 773 – 787 Google Scholar Crossref Search ADS PubMed WorldCat Reiffenstein RJ , Hulbert WC, Roth SH ( 1992 ) Toxicology of hydrogen sulfide . Annu Rev Pharmacol Toxicol 32 : 109 – 134 Google Scholar Crossref Search ADS PubMed WorldCat Riemenschneider A , Nikiforova V, Hoefgen R, De Kok LJ, Papenbrock J ( 2005 a ) Impact of elevated H(2)S on metabolite levels, activity of enzymes and expression of genes involved in cysteine metabolism . Plant Physiol Biochem 43 : 473 – 483 Google Scholar Crossref Search ADS PubMed WorldCat Riemenschneider A , Wegele R, Schmidt A, Papenbrock J ( 2005 b ) Isolation and characterization of a D-cysteine desulfhydrase protein from Arabidopsis thaliana . FEBS J 272 : 1291 – 1304 Google Scholar Crossref Search ADS PubMed WorldCat Schroeder JI , Kwak JM, Allen GJ ( 2001 ) Guard cell abscisic acid signalling and engineering drought hardiness in plants . Nature 410 : 327 – 330 Google Scholar Crossref Search ADS PubMed WorldCat Scuffi D , Álvarez C, Laspina N, Gotor C, Lamattina L, García-Mata C ( 2014 ) Hydrogen sulfide generated by l-cysteine desulfhydrase acts upstream of nitric oxide to modulate abscisic acid-dependent stomatal closure . Plant Physiol 166 : 2065 – 2076 Google Scholar Crossref Search ADS PubMed WorldCat Siegel RS , Xue S, Murata Y, Yang Y, Nishimura N, Wang A, Schroeder JI ( 2009 ) Calcium elevation-dependent and attenuated resting calcium-dependent abscisic acid induction of stomatal closure and abscisic acid-induced enhancement of calcium sensitivities of S-type anion and inward-rectifying K channels in Arabidopsis guard cells . Plant J 59 : 207 – 220 Google Scholar Crossref Search ADS PubMed WorldCat Sokolovski S , Blatt MR ( 2004 ) Nitric oxide block of outward-rectifying K+ channels indicates direct control by protein nitrosylation in guard cells . Plant Physiol 136 : 4275 – 4284 Google Scholar Crossref Search ADS PubMed WorldCat Sokolovski S , Hills A, Gay R, Garcia-Mata C, Lamattina L, Blatt MR ( 2005 ) Protein phosphorylation is a prerequisite for intracellular Ca2+ release and ion channel control by nitric oxide and abscisic acid in guard cells . Plant J 43 : 520 – 529 Google Scholar Crossref Search ADS PubMed WorldCat Sutter JU , Sieben C, Hartel A, Eisenach C, Thiel G, Blatt MR ( 2007 ) Abscisic acid triggers the endocytosis of the Arabidopsis KAT1 K+ channel and its recycling to the plasma membrane . Curr Biol 17 : 1396 – 1402 Google Scholar Crossref Search ADS PubMed WorldCat Tang G , Wu L, Wang R ( 2010 ) Interaction of hydrogen sulfide with ion channels . Clin Exp Pharmacol Physiol 37 : 753 – 763 Google Scholar Crossref Search ADS PubMed WorldCat Thiel G , Blatt MR ( 1994 ) Phosphatase antagonist okadaic acid inhibits steady-state K+ currents in guard cells of Vicia faba . Plant J 5 : 727 – 733 Google Scholar Crossref Search ADS WorldCat Wang R ( 2002 ) Two’s company, three’s a crowd: can H2S be the third endogenous gaseous transmitter? FASEB J 16 : 1792 – 1798 Google Scholar Crossref Search ADS PubMed WorldCat Wang Y , Blatt MR ( 2011 ) Anion channel sensitivity to cytosolic organic acids implicates a central role for oxaloacetate in integrating ion flux with metabolism in stomatal guard cells . Biochem J 439 : 161 – 170 Google Scholar Crossref Search ADS PubMed WorldCat Wang Y , Papanatsiou M, Eisenach C, Karnik R, Williams M, Hills A, Lew VL, Blatt MR ( 2012 ) Systems dynamic modelling of a guard cell Cl− channel mutant uncovers an emergent homeostatic network regulating stomatal transpiration . Plant Physiol 160 : 1956 – 1972 Google Scholar Crossref Search ADS PubMed WorldCat Whiteman M , Li L, Kostetski I, Chu SH, Siau JL, Bhatia M, Moore PK ( 2006 ) Evidence for the formation of a novel nitrosothiol from the gaseous mediators nitric oxide and hydrogen sulphide . Biochem Biophys Res Commun 343 : 303 – 310 Google Scholar Crossref Search ADS PubMed WorldCat Yang W , Yang G, Jia X, Wu L, Wang R ( 2005 ) Activation of KATP channels by H2S in rat insulin-secreting cells and the underlying mechanisms . J Physiol 569 : 519 – 531 Google Scholar Crossref Search ADS PubMed WorldCat Zhang H , Hu LY, Hu KD, He YD, Wang SH, Luo JP ( 2008 ) Hydrogen sulfide promotes wheat seed germination and alleviates oxidative damage against copper stress . J Integr Plant Biol 50 : 1518 – 1529 Google Scholar Crossref Search ADS PubMed WorldCat Zhang H , Tang J, Liu XP, Wang Y, Yu W, Peng WY, Fang F, Ma DF, Wei ZJ, Hu LY ( 2009 a ) Hydrogen sulfide promotes root organogenesis in Ipomoea batatas, Salix matsudana and Glycine max . J Integr Plant Biol 51 : 1086 – 1094 Google Scholar Crossref Search ADS PubMed WorldCat Zhang H , Ye YK, Wang SH, Luo JP, Tang J, Ma DF ( 2009 b ) Hydrogen sulfide counteracts chlorophyll loss in sweetpotato seedling leaves and alleviates oxidative damage against osmotic stress . Plant Growth Regul 58 : 243 – 250 Google Scholar Crossref Search ADS WorldCat Zhao W , Zhang J, Lu Y, Wang R ( 2001 ) The vasorelaxant effect of H(2)S as a novel endogenous gaseous K(ATP) channel opener . EMBO J 20 : 6008 – 6016 Google Scholar Crossref Search ADS PubMed WorldCat Author notes 1 This work was supported by the United Kingdom Biotechnology and Biological Sciences Research Council (grant nos. BB/H0024867/1, BB/I024496/1, BB/K015893/1, and BBL001276/1), the Royal Society, London (travel grant no. IE120659), and the Begonia Trust (Studentship to M.P.). * Address correspondence to [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Michael R. Blatt ([email protected]). [OPEN] Articles can be viewed without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.114.256057 © 2015 American Society of Plant Biologists. All Rights Reserved. © The Author(s) 2015. Published by Oxford University Press on behalf of American Society of Plant Biologists. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

van Erp, Harrie; Shockey, Jay; Zhang, Meng; Adhikari, Neil D.; Browse, John

doi: 10.1104/pp.114.254110pmid: 25739701

Abstract One goal of green chemistry is the production of industrially useful fatty acids (FAs) in crop plants. We focus on hydroxy fatty acids (HFAs) and conjugated polyenoic FAs (α-eleostearic acids [ESAs]) using Arabidopsis (Arabidopsis thaliana) as a model. These FAs are found naturally in seed oils of castor (Ricinus communis) and tung tree (Vernicia fordii), respectively, and used for the production of lubricants, nylon, and paints. Transgenic oils typically contain less target FA than that produced in the source species. We hypothesized that competition between endogenous and transgenic isozymes for substrates limits accumulation of unique FAs in Arabidopsis seeds. This hypothesis was tested by introducing a mutation in Arabidopsis diacylglycerol acyltransferase1 (AtDGAT1) in a line expressing castor FA hydroxylase and acyl-Coenzyme A:RcDGAT2 in its seeds. This led to a 17% increase in the proportion of HFA in seed oil. Expression of castor phospholipid:diacylglycerol acyltransferase 1A in this line increased the proportion of HFA by an additional 12%. To determine if our observations are more widely applicable, we investigated if isozyme competition influenced production of ESA. Expression of tung tree FA conjugase/desaturase in Arabidopsis produced approximately 7.5% ESA in seed lipids. Coexpression of VfDGAT2 increased ESA levels to approximately 11%. Overexpression of VfDGAT2 combined with suppression of AtDGAT1 increased ESA accumulation to 14% to 15%. Our results indicate that isozyme competition is a limiting factor in the engineering of unusual FAs in heterologous plant systems and that reduction of competition through mutation and RNA suppression may be a useful component of seed metabolic engineering strategies. Production of vegetable oils in the form of triacylglycerols (TAGs) is of great importance for human nutrition and as a source of chemicals for industry. The vegetable oils produced in our food crops are composed of five major fatty acids (FAs; Bates et al., 2013). Other than these common FAs, numerous uncommon FAs are produced in nature, such as hydroxy fatty acids (HFAs), conjugated FAs, epoxy FAs, and short-chain FAs, that are or could be used for industrial purposes (Badami and Patil, 1980). However, the species producing these uncommon FAs are often not suitable for large-scale industrialized agriculture (Voelker and Kinney, 2001; Dyer et al., 2008). To solve this problem, attempts have been made to produce these uncommon FAs in seeds of crop plants. This has been a long-standing goal in the field of lipid research and seemed initially quite straightforward (Voelker and Kinney, 2001; Napier, 2007; Napier and Graham, 2010; Carlsson et al., 2011; Bates and Browse, 2012; Bates et al., 2013; Vanhercke et al., 2013). The approach taken was to identify the enzyme responsible for synthesis of the desired FA and express the corresponding gene in seeds of high-yielding crop plants. Unfortunately, in general, only low levels of the desired FAs were produced compared with levels in the native plant (Broun and Somerville, 1997; Cahoon et al., 2006). One reason for this discrepancy is that enzymes of TAG synthesis often lack proper substrate specificity and selectivity, leading to poor utilization of substrates containing these unusual FAs (Knutzon et al., 1999; Burgal et al., 2008; Li et al., 2010; Kim et al., 2011; van Erp et al., 2011). In this article, we focus on the engineering of HFAs (including ricinoleic acid) and α-eleostearic acid (ESA) in heterologous plant systems. HFAs are used in many industrial applications, including the production of nylon, plastics, and lubricants, and they are produced at high levels in the seeds of castor (Ricinus communis). However, castor is not suitable for industrialized agriculture and produces the toxic protein ricin as well as other proteins that can cause allergenic reactions in humans. Currently, most of the cultivation of castor for the production of HFAs occurs in China, India, and Brazil. ESA is used in industrial applications, such as inks, coatings, and resins, and produced in the seeds of tung tree (Vernicia fordii; formerly Aleurites fordii; Sonntag, 1979). Tung tree also has problematic agronomic characteristics and can only be grown in limited areas of the United States that are prone to damage from hurricanes. To create cheap and reliable sources of these FAs, their synthesis has been studied, and attempts have been made to produce them in heterologous plant systems. The gene responsible for the synthesis of HFA in castor is FATTY ACID HYDROXYLASE12 (RcFAH12). This enzyme hydroxylates the ∆12 position of oleic acid esterified to the stereospecifically numbered2 (sn-2) position of phosphatidylcholine (PC) and is a homolog of FATTY ACID DESATURASE2 (FAD2; Van de Loo et al., 1995). Tung tree Fatty Acid Conjugase/Desaturase (FADX; Dyer et al., 2002) is responsible for the synthesis of ESA, and also, it is an FAD2 homolog. It converts PC-bound linoleoyl groups to eleostearate (18:3Ɗ9cis, 11trans, and 13trans; Dyer et al., 2002). After synthesis on PC, these FAs can be incorporated into TAG by several different metabolic routes and enzymes (Fig. 1). Only the routes and enzymes relevant to this article will be discussed here (a more exhaustive description is in van Erp et al., 2011). The modified FAs can be hydrolyzed from the sn-2 position of PC by phospholipid:diacylglycerol acyltransferase (PDAT), which then esterifies these FAs to the sn-3 position of diacylglycerol (DAG) to generate TAG. The lysophosphatidylcholine generated by this reaction can be used by acyl-CoA:lysophosphatidylcholine acyltransferase to regenerate PC. PC can be converted into DAG by phospholipase C or phosphatidylcholine:diacylglycerol cholinephosphotransferase (PDCT). Phospholipase C is involved in the removal of the choline head group of PC to generate DAG with the same FA composition as PC. PDCT exchanges the choline head group between PC and DAG. The DAG formed by these enzymes can subsequently be used by PDAT or acyl-CoA:diacylglycerol acyltransferase (DGAT) to generate TAG. A reverse activity of lysophosphatidylcholine acyltransferase can hydrolyze the modified FAs from the sn-2 position of PC to generate acyl-CoA and lysophosphatidylcholine. These acyl-CoAs can subsequently be used by the acyltransferase enzymes of the Kennedy pathway (acyl-CoA:glycerol-3-P acyltransferase, acyl-CoA:lysophosphatidic acid acyltransferase, and DGAT) to generate TAG-containing modified FAs. Figure 1. Open in new tabDownload slide Overview of TAG biosynthesis in transgenic Arabidopsis seeds overexpressing FAH12 or FADX. These enzymes modify 18:1-PC and 18:2-PC, respectively, to generate HFA-PC or ESA-PC, respectively (m indicates FAs with these modifications). The modified FAs (as well as 18:2 and 18:3) are subsequently incorporated into TAG by the various enzymes involved in the synthesis of this storage lipid. GPAT, Acyl-CoA:glycerol-3-P acyltransferase; G3P, glycerol-3-P; LPA, lysophosphatidic acid; LPAT, acyl-CoA:lysophosphatidic acid acyltransferase; LPC, lysophosphatidylcholine; LPCAT, acyl-CoA:lysophosphatidylcholine acyltransferase; PA, phosphatidic acid; PAP, phosphatidic acid phosphatase; PLC, phospholipase C. Figure 1. Open in new tabDownload slide Overview of TAG biosynthesis in transgenic Arabidopsis seeds overexpressing FAH12 or FADX. These enzymes modify 18:1-PC and 18:2-PC, respectively, to generate HFA-PC or ESA-PC, respectively (m indicates FAs with these modifications). The modified FAs (as well as 18:2 and 18:3) are subsequently incorporated into TAG by the various enzymes involved in the synthesis of this storage lipid. GPAT, Acyl-CoA:glycerol-3-P acyltransferase; G3P, glycerol-3-P; LPA, lysophosphatidic acid; LPAT, acyl-CoA:lysophosphatidic acid acyltransferase; LPC, lysophosphatidylcholine; LPCAT, acyl-CoA:lysophosphatidylcholine acyltransferase; PA, phosphatidic acid; PAP, phosphatidic acid phosphatase; PLC, phospholipase C. Expression of an RcFAH12 complementary DNA (cDNA) in Arabidopsis (Arabidopsis thaliana) under control of several seed-specific promoters led to accumulation of HFA to only 17% of total seed FAs (Broun and Somerville, 1997; Smith et al., 2000, 2003; Lu et al., 2006) compared with approximately 90% in native castor seed oil (Badami and Patil, 1980). Engineering attempts to overcome this problem have focused on coexpression of castor acyltransferases, such as RcDGAT2 (Burgal et al., 2008), RcPDAT1A (Kim et al., 2011; van Erp et al., 2011), or the castor electron transfer system in Arabidopsis seeds expressing RcFAH12 (Wayne et al., 2013). Coexpression of RcDGAT2 and RcPDAT1A led to significant increases in HFA levels from 17% to 26% to 28% of total seed FAs, whereas coexpression of the electron transfer system did not result in an increase. Production of ESA in Arabidopsis has met similar challenges. Expression of the FADX genes from either tung tree or bitter gourd (Momordica charantia; another species containing high levels of ESA in seeds) in an fad3 fatty acid elongase1 (fae1) Arabidopsis line resulted in 7% to 13% ESA (Cahoon et al., 2006). Two types of DGAT from tung tree showed significantly different affinities toward ESA-containing substrates when expressed in yeast (Saccharomyces cerevisiae)-fed tung oil FAs (Shockey et al., 2006), but the effects of these and other tung tree TAG metabolic enzymes in planta have not previously been reported. The experiments reported here represent a valuable opportunity to investigate possible commonalities between metabolic engineering strategies for production of various types of value-added oils. Subsequent research into the underlying causes of suboptimal production of HFA revealed that RcFAH12 expression in Arabidopsis seeds caused metabolic perturbations, leading to the poor accumulation of HFA in TAG and a reduction in total FA content of seeds (Dauk et al., 2007; Bates and Browse, 2011; van Erp et al., 2011). Overexpression of castor PDAT and DGAT restored FA content in RcFAH12 transgenic seeds to nearly wild-type levels, while simultaneously increasing HFA content of the seeds (van Erp et al., 2011). Bates et al., 2014 elucidated the mechanism behind this decrease in FA content of seeds expressing RcFAH12 by showing that RcFAH12 expression resulted in feedback inhibition of FA synthesis. These results indicate that efficient transfer of HFA from their site of synthesis on PC into TAG is essential for the engineering of high levels of HFA in heterologous plant systems. In this article, we investigated whether it is possible to improve the flux of HFA out of PC and into TAG by reducing substrate competition between endogenous Arabidopsis isozymes and the transgenic counterparts from castor and tung tree. We focused on substrate competition between AtDGAT1 and/or AtPDAT1 and either RcDGAT2 and RcPDAT1A from castor or VfDGAT2 from tung tree. To increase synthesis of HFA, we attempted to replace AtDGAT1 and AtPDAT1 with RcDGAT2 and RcPDAT1A by introducing a null mutation in AtDGAT1 and suppressing the expression of AtPDAT1 with artificial microRNAs (amiRNAs). Introduction of an Atdgat1 null mutation in a plant line expressing RcFAH12 and RcDGAT2 resulted in a significant increase in HFA levels. Subsequent overexpression of RcPDAT1A resulted in a further increase in HFA levels. In the case of ESA, an established homozygous FADX line (producing approximately 8% ESA in total seed lipids) was retransformed with either VfDGAT2 alone or VfDGAT2 together with a seed-specific AtDGAT1 RNA interference (RNAi) construct. Analysis of seed lipids from homozygous double-transgenic versus parental single-transgenic FADX lines showed higher levels of ESA when tung tree DGAT2 was present, which were enhanced further when the AtDGAT1 RNAi was present. The similarities between these two data sets suggest that endogenous competition is a limitation common to many such engineering strategies. RESULTS Eliminating AtDGAT1 Increases HFA Accumulation in an FAH12 RcDGAT2 Transgenic Line AtDGAT1 and AtPDAT1 catalyze the final step in the synthesis of TAG in Arabidopsis seeds (Zhang et al., 2009). AtDGAT1 and AtPDAT1 preferentially incorporate common FAs into TAG, whereas RcDGAT2 and RcPDAT1A prefer HFA (Burgal et al., 2008; van Erp et al., 2011). We hypothesize that substrate competition for common FAs versus unusual FAs occurs between AtPDAT1/AtDGAT1 and the transgenic acyltransferases and that this competition limits the accumulation of unusual FAs in our FAH12 RcDGAT2 transgenic lines. To test this proposal, the Arabidopsis dgat1-2 mutant (Routaboul et al., 1999) was crossed with the Chaofu Lu7 (CL7) RcDGAT2 line 544 #5 (Burgal et al., 2008) followed by selfing of F1 plants. The CL7 line expresses RcFAH12 in seeds of the fae1 mutant (Kunst et al., 1992; Lu et al., 2006). Segregation of the four loci (FAH12, RcDGAT2, fae1, and dgat1) in an F2 population of 460 plants was monitored by PCR assays and gas chromatography of seed FA compositions. An F2 plant determined to be heterozygous for dgat1 and homozygous for fae1, RcFAH12, and RcDGAT2 was grown to maturity. A population of F3 progeny of this plant (n = 60) was grown, and sequencing of DGAT1 was used to identify nine wild-type and nine dgat1-2 segregants. Analysis of seed samples from these plants shows a 17% increase in the proportion of HFA from an average of 24.7% ± 0.61% in the DGAT1 wild-type plants to an average of 29.0% ± 0.9% in the dgat1 CL7 RcDGAT2 line (Fig. 2; Supplemental Fig. S1), and the amount of HFA per seed increased by 23% (Fig. 3). Introducing the dgat1 mutation also leads to a reduction in 18:1 and increases in 18:2 and 18:3 (Fig. 2). These changes are likely the result of the known increases in FAD2, FAD3, and PDAT1 expression that occur in dgat1-mutant seed (Xu et al., 2012). Mutations at the dgat1 locus cause a 45% reduction in seed FA content (Katavic et al., 1995; Routaboul et al., 1999; Zou et al., 1999). However, introducing a dgat1 mutation in the CL7 RcDGAT2 background did not lead to any additional reduction in total FA/seed or percentage of oil relative to CL7 RcDGAT2 controls (Fig. 3, A and D), indicating that the castor DGAT2 can maintain rates of oil synthesis in a dgat1 PDAT1 genetic background. Nevertheless, relative to the fae1 parental line, total FA/seed, seed weight, and percentage of oil are reduced by 14%, 10%, and 8%, respectively (Fig. 3). Results from previous studies have indicated that fae1 is comparable with the wild type in all of these parameters (Kunst et al., 1992; van Erp et al., 2011). Figure 2. Open in new tabDownload slide Seed FA composition of CL7 RcDGAT2 and dgat1 CL7 RcDGAT2 lines. The data represent the average of nine individual plants ± se. Two-tailed Student’s t test. *, P < 0.05; **, P < 0.01. Figure 2. Open in new tabDownload slide Seed FA composition of CL7 RcDGAT2 and dgat1 CL7 RcDGAT2 lines. The data represent the average of nine individual plants ± se. Two-tailed Student’s t test. *, P < 0.05; **, P < 0.01. Figure 3. Open in new tabDownload slide Results of extended analysis of mature seeds from the transgenic lines. A, Total FAs per seed. B, HFA per seed. C, Seed weight. D, Percentage of oil content. The data represent the average of five to 27 individual plants ± se. Two-tailed Student’s t test. *, P < 0.05; **, P < 0.01. Figure 3. Open in new tabDownload slide Results of extended analysis of mature seeds from the transgenic lines. A, Total FAs per seed. B, HFA per seed. C, Seed weight. D, Percentage of oil content. The data represent the average of five to 27 individual plants ± se. Two-tailed Student’s t test. *, P < 0.05; **, P < 0.01. Expression of RcPDAT1A in the dgat1 CL7 RcDGAT2 Background Further Increases HFA Levels Expression of RcPDAT1A in the dgat1 CL7 RcDGAT2 background could shift the balance in substrate competition between castor and Arabidopsis enzymes further in favor of incorporation of HFA into TAG. To test this hypothesis, RcPDAT1A was transformed in the dgat1 CL7 RcDGAT2 background using Discosoma spp. red fluorescent protein (DsRed) as a selection marker (Stuitje et al., 2003). Thirty-nine lines with independent transgene insertion sites were generated. From these, four lines with high levels of HFA segregating 1:3 for brown:red seeds, indicating a single functional insertion allele, were selected. To determine if the increase in HFA levels is caused by the presence of RcPDAT1A, a T2 population of 40 plants segregating for RcPDAT1A was planted. The four tested lines gave similar results, and Figure 4 shows data for one of these lines. HFA levels in T3 seeds increased from an average of 28.0% ± 0.72% in the dgat1 CL7 RcDGAT2 plants to an average of 31.4% ± 1.12% in the dgat1 CL7 RcDGAT2 RcPDAT1A plants. There were no significant changes in the amount of total FA per seed (Fig. 3A), but there was a 23% increase in the amount of HFA per seed (Fig. 3B) in the dgat1 CL7 RcDGAT2 RcPDAT1A plants compared with the dgat1 CL7 RcDGAT2 segregants. These increases are somewhat larger than observed previously (van Erp et al., 2011) when RcPDAT1A was expressed in the CL7 RcDGAT2 background (26.7% HFA). These improvements likely arise from the introduction of the dgat1 mutation, which shifts the balance in substrate competition further in favor of incorporation of HFA. Figure 4. Open in new tabDownload slide Seed FA composition of dgat1 CL7 RcDGAT2 and dgat1 CL7 RcDGAT2 RcPDAT1A lines. The data represent the average of five to 11 individual plants ± se. Two-tailed Student’s t test. *, P < 0.05. Figure 4. Open in new tabDownload slide Seed FA composition of dgat1 CL7 RcDGAT2 and dgat1 CL7 RcDGAT2 RcPDAT1A lines. The data represent the average of five to 11 individual plants ± se. Two-tailed Student’s t test. *, P < 0.05. Reduced Expression of AtPDAT1 in the dgat1 CL7 RcDGAT2 RcPDAT1A Background Does Not Further Increase HFA Levels In the dgat1 CL7 RcDGAT2 RcPDAT1A line, substrate competition between AtPDAT1 and RcPDAT1A/RcDGAT2 can still occur. The dgat1-1 pdat1 double mutant of Arabidopsis is not viable (Zhang et al., 2009); therefore, to determine if reducing AtPDAT1 activity in the dgat1 CL7 RcDGAT2 RcPDAT1A line might further increase HFA accumulation, we used an amiRNA approach. A 21-mer sequence targeting the 3′-untranslated region (UTR) of the AtPDAT1 mRNA was designed and used to replace the stem loops in the micro-RNA 319a (MIR319a) precursor (Palatnik et al., 2003; Ossowski et al., 2008). The pdat1-amiRNA was cloned in a multigene vector behind the glycinin seed-specific promoter, and RcPDAT1A was cloned behind the oleosin promoter in the same vector, which expresses the DsRed selectable marker. The resulting pdat1-amiRNA/RcPDAT1A construct was transformed into the dgat1 CL7 RcDGAT2 line. Most of the transgenic DsRed T1 seeds were wrinkled and had a low FA content. This resembles the observations made when AtPDAT1 was suppressed in the dgat1 background (Zhang et al., 2009). Germination of the T1 seeds was reduced to less than 50%, but 45 T1 plants were grown successfully. T2 seeds from these plants were analyzed, and lines with high levels of HFA showing Mendelian segregation patterns, indicating a single functional insertion allele, were selected for additional analysis. Two segregating populations of lines with the highest HFA levels were planted (50 plants of each line), and the HFA levels in T3 seeds were determined. Figure 5 shows the data for one line. The HFA level in the dgat1 CL7 RcDGAT2 pdat1-amiRNA RcPDAT1A segregants (32.2% ± 0.53% HFA) was not statistically different from that of the dgat1 CL7 RcDGAT2 RcPDAT1A line (31.4% ± 1.12% HFA). There were also no significant changes in the micrograms of FA and HFA per seed (Fig. 3, A and B). A possible explanation for this result is that T1 seeds with the strongest suppression of AtPDAT1 expression were inviable and did not germinate. Consistent with this possibility, quantitative reverse transcription (RT)-PCR results indicated that AtPDAT1 transcript levels in dgat1 CL7 RcDGAT2 pdat1-amiRNA RcPDAT1A plants were more than 40% of those measured in the dgat1 CL7 RcDGAT2 RcPDAT1A controls. These experiments indicate that RcPDAT1A and RcDGAT2 cannot fully replace the function of AtPDAT1 and AtDGAT1 during seed development. Numerical data for the seed FA compositions of four different transgene combinations investigated are included in Supplemental Table S1. Figure 5. Open in new tabDownload slide Seed FA composition of dgat1 CL7 RcDGAT2 and dgat1 CL7 RcDGAT2 pdat1-amiRNA RcPDAT1A lines. The data represent the average of 11 individual plants ± se. Two-tailed Student’s t test. *, P < 0.05; **, P < 0.01. Figure 5. Open in new tabDownload slide Seed FA composition of dgat1 CL7 RcDGAT2 and dgat1 CL7 RcDGAT2 pdat1-amiRNA RcPDAT1A lines. The data represent the average of 11 individual plants ± se. Two-tailed Student’s t test. *, P < 0.05; **, P < 0.01. Analysis of TAG Composition and Regiochemistry To determine the biochemical basis for the changes in HFA levels observed in our transgenic lines, the composition of seed TAG was investigated. TAG was extracted from the seeds of the transgenic plant lines and separated by thin-layer chromatography into molecular species with zero, one, two, or three HFAs. Quantitative FA analysis by gas chromatography was used to determine the relative amounts of four molecular species. Introduction of the dgat1 null mutation in the CL7 RcDGAT2 background increased the level of 2-HFA-TAG from 16.6% ± 0.60% to 26.3% ± 0.85% of total seed TAGs (a 59% increase), caused no change in 1-HFA-TAG, and decreased the amount of 0-HFA-TAG from 40.8% ± 1.62% to 28.8% ± 0.51% (a 29% decrease; Fig. 6A). This change in TAG species composition is most likely caused by a difference in substrate preference between AtDGAT1 and RcDGAT2. Burgal et al., 2008 showed that RcDGAT2 preferentially uses HFA-DAG as a substrate compared with 18:1- and 18:2-DAG. The decrease in 0-HFA-TAG is probably caused by acylation of 0-HFA-DAG with an HFA at the sn-3 position by RcDGAT2. The fact that this does not result in an increase in 1-HFA-TAG levels might be caused by acylation of sn-2 HFA-DAG with an HFA at the sn-3 position to produce 2-HFA-TAG (Fig. 6A). Figure 6. Open in new tabDownload slide Analysis of TAG compositions of mature seeds from the transgenic lines. A, TAG molecular species; 0-HFA to 3-HFA represent molecular species with zero to three HFAs, respectively. B, Percentage of HFAs at the sn-2 and sn-1/3 positions of TAG. The data represent the average of three replicates ± se. Two-tailed Student’s t test. *, P < 0.05; **, P < 0.01. Figure 6. Open in new tabDownload slide Analysis of TAG compositions of mature seeds from the transgenic lines. A, TAG molecular species; 0-HFA to 3-HFA represent molecular species with zero to three HFAs, respectively. B, Percentage of HFAs at the sn-2 and sn-1/3 positions of TAG. The data represent the average of three replicates ± se. Two-tailed Student’s t test. *, P < 0.05; **, P < 0.01. Expression of RcPDAT1A in the dgat1 CL7 RcDGAT2 background further increased 2-HFA-TAG from 26.3% ± 0.85% to 39.5% ± 0.21% (a 50% increase), decreased 1-HFA-TAG levels from 42.1% ± 1.05% to 32.4% ± 0.33% (a 30% decrease), and caused a small (10%) decrease from 28.8% ± 0.51% to 26.0% ± 0.14% in the 0-HFA-TAG levels (Fig. 6A). A possible explanation for these observations could be that RcPDAT1A might preferentially acylate HFA-DAG over normal DAG. The decrease in 1-HFA-TAG is probably caused by RcPDAT1A transferring HFA from the sn-2 position of PC to HFA-DAG. Consistent with this interpretation, regiochemical analysis of TAG by incubation with Rhizomucor miehei lipase (van Erp et al., 2011) indicated that increases in HFA seen in our lines was predominantly found at the sn-1/3 position of the TAG molecule (Fig. 6B). Suppression of AtPDAT1 in the dgat1 CL7 RcDGAT2 RcPDAT1A background did not lead to any significant additional change in TAG species composition, except that 1-HFA-TAG slightly decreased and 0-HFA-TAG slightly increased (Fig. 6A). This is consistent with the low level of suppression of expression of AtPDAT1 in the viable pdat1-amiRNA lines described above. Substrate Competition between AtDGAT1 and VfDGAT2 Limits the Accumulation of ESA To determine if substrate competition between endogenous acyltransferases and introduced transgenic enzymes limits the accumulation of other unusual FAs, we investigated the synthesis of ESA in Arabidopsis seeds. To examine how ESA accumulation was affected by competition between endogenous and transgenic enzymes, the fad3 fae1 double mutant of Arabidopsis (Smith et al., 2003) was first transformed with a construct containing the tung tree FA conjugase FADX (Dyer et al., 2002) driven by the strong seed-specific phaseolin promoter (Slightom et al., 1983). Using this line as a starting point, we explored the ability of VfDGAT2 to increase ESA content in Arabidopsis seeds when coexpressed with tung tree FADX. A DGAT2 was chosen for this role based on previous studies of VfDGAT2 in yeast (Shockey et al., 2006), which showed that this enzyme has a strong preference for ESA-containing substrates, and similar findings regarding the substrate selectivities of other related DGAT2 enzymes (Burgal et al., 2008; Li et al., 2010). T5 FADX plants grown from homozygous transgenic T4 plants producing 7.5% ± 0.1% ESA were retransformed with VfDGAT2 transgene under control of the Arabidopsis 2S seed storage protein3 (At2S-3) promoter (Guerche et al., 1990) either alone or combined with a seed-specific AtDGAT1 RNAi cassette driven by the soybean (Glycine max) β-conglycinin α′-subunit promoter (Doyle et al., 1986). Multiple transgenic T1 seeds were selected by observation of fluorescence from the DsRed marker included in the binary constructs. Segregating seed samples were harvested from mature T1 plants, and their lipids were analyzed by gas chromatography. Nineteen independent transgenic events yielded an average of 9.01% ESA at this stage, representing a 20% increase relative to the parental FADX plants. Twenty-two transgenic lines coexpressing the combination of VfDGAT2 and AtDGAT1 RNAi averaged 11.1% ± 1.29% ESA (a 48% increase relative to the parental lines; Fig. 7). Figure 7. Open in new tabDownload slide Distribution of ESA levels in transgenic plants expressing tung tree FADX alone or with tung tree DGAT2 or tung tree DGAT2 + AtDGAT1 RNAi. A parental FADX line was retransformed with VfDGAT2 or VfDGAT2 + AtDGAT1 RNAi. Seed pools from segregating T1 plants representing independent transgenic events were analyzed by gas chromatography; the corresponding ESA levels are shown. The data are plotted as a scatterplot, with the mean and se indicated. Differences between the three data sets are statistically significant (two-tailed Student’s t test; P < 0.001). Figure 7. Open in new tabDownload slide Distribution of ESA levels in transgenic plants expressing tung tree FADX alone or with tung tree DGAT2 or tung tree DGAT2 + AtDGAT1 RNAi. A parental FADX line was retransformed with VfDGAT2 or VfDGAT2 + AtDGAT1 RNAi. Seed pools from segregating T1 plants representing independent transgenic events were analyzed by gas chromatography; the corresponding ESA levels are shown. The data are plotted as a scatterplot, with the mean and se indicated. Differences between the three data sets are statistically significant (two-tailed Student’s t test; P < 0.001). To ascertain the full degree of change affected by expression of VfDGAT2 overexpression and AtDGAT1 silencing, two T2 lines for each of the double transformants were chosen for analysis in the T3 generation. Seeds were harvested from between 18 and 32 progeny from each of the chosen T2 lines. Several homozygous samples (consisting of 100% red seeds) and FADX parental revertants (displaying uniform brown seeds devoid of fluorescence) from each line were analyzed. All four sets of brown FADX parental revertant seeds averaged between 8.49% and 8.85% ESA. Individual FADX plants homozygous for VfDGAT2 averaged 11.6% ± 0.71% and 11.7% ± 0.62% ESA, respectively (Fig. 8), including individual plants with seeds that contained as much as 12.47% and 12.64% ESA, respectively. As in the T2 sample distributions, FADX plants coexpressing both VfDGAT2 and AtDGAT1 RNAi produced higher levels of the unique FA in terms of both plants with the highest seed ESA (14.9% and 16.1%, respectively) and average ESA across all homozygous plants (13.8% ± 0.69% and 14.7% ± 0.85%, respectively; Fig. 8). Complete FA analyses of seeds of these eight lines are included in Supplemental Table S2. Figure 8. Open in new tabDownload slide Determination of ESA levels in homozygous transgenic plants coexpressing FADX and VfDGAT2 with or without AtDGAT1I RNAi. Seeds from two high-performing lines from each of the transgenic genotypes were sown on soil and grown to maturity. Seed samples from homozygous transgenic (red) and nontransgenic (brown) plants were collected separately and analyzed by gas chromatography. Mean ESA levels ± se are shown. Lowercase letters indicate different statistically significant groups (Student’s t test; P < 0.01). Figure 8. Open in new tabDownload slide Determination of ESA levels in homozygous transgenic plants coexpressing FADX and VfDGAT2 with or without AtDGAT1I RNAi. Seeds from two high-performing lines from each of the transgenic genotypes were sown on soil and grown to maturity. Seed samples from homozygous transgenic (red) and nontransgenic (brown) plants were collected separately and analyzed by gas chromatography. Mean ESA levels ± se are shown. Lowercase letters indicate different statistically significant groups (Student’s t test; P < 0.01). To provide additional data on the seed oil composition of three different transgene combinations, we analyzed additional seed batches of line #5 red (homozygous VfFADX VfDGAT2) and line #9 red (homozygous VfFADX VfDGAT2 AtDGAT1 RNAi) along with brown segregants (homozygous VfFADX) as controls. The data obtained confirmed the increases in percentage of ESA in the oil conferred by expression of VfDGAT2 and VfDGAT2 + AtDGAT1 RNAi (Supplemental Fig. S2A). These increases were associated with increases in ESA per milligram of seed weight (Supplemental Fig. S2B), whereas there was no significant change in the total FA per milligram of seed (Supplemental Fig. S2C). DISCUSSION AtDGAT1 and AtPDAT1 are the two enzymes that are responsible for the acylation of the sn-3 position of DAG in seeds and some other Arabidopsis tissues. When expression of AtPDAT1 was suppressed in the atdgat1 background, most of the pollen was not viable, likely because of a severe reduction in pollen storage lipid, and the reduced number of seeds that did develop had very low oil content (Zhang et al., 2009). These results indicate that, in Arabidopsis, DGAT1 and PDAT1 have an essential, redundant role and that other enzymes, such as AtDGAT2 and AtDGAT3, do not contribute significantly to acylation of the sn-3 position of DAG in Arabidopsis seed lipids. AtDGAT1 has the quantitatively more significant role, which was indicated by the reduction in oil content in the atdgat1 background (Katavic et al., 1995; Routaboul et al., 1999; Zou et al., 1999). By contrast, characterization of an atpdat1 null mutant showed no reduction in seed oil content (Mhaske et al., 2005). The goal of our investigation was to determine if substrate competition between transgenic castor or tung tree enzymes and their endogenous Arabidopsis counterparts is a limiting factor for the accumulation of HFA or ESA in our transgenic plant lines. This possibility arises, because the endogenous Arabidopsis enzymes are known or proposed to favor substrates containing FA normally found in Arabidopsis seed TAG, whereas the castor and tung tree enzymes have been shown to have specificity for HFA- or ESA-containing substrates, respectively (Shockey et al., 2006; Burgal et al., 2008). We chose to focus on potential competition between AtDGAT1 and either RcDGAT2 or VfDGAT2, because these enzymes are known to play major roles in TAG synthesis in the respective plant species (Ståhl et al., 2004; Kroon et al., 2006; Shockey et al., 2006; Burgal et al., 2008; Zhang et al., 2009; van Erp et al., 2011). To test if substrate competition between AtDGAT1 and RcDGAT2 is limiting for accumulation of HFA, the atdgat1-2 mutation was crossed into the CL7 RcDGAT2 line. Homozygous dgat1 CL7 RcDGAT2 plants had significantly higher HFA levels than the CL7 RcDGAT2 segregants (Fig. 2). Subsequent analyses of the derived dgat1 and wild-type DGAT1 lines indicated that there was no reduction in seed oil content associated with the loss of the DGAT1 enzyme (Fig. 3). These results indicate that eliminating competition from the endogenous DGAT enzyme results in a 20% increase in HFA accumulation in the seed oil compared with the parental CL7 RcDGAT2 line. To determine if substrate competition is a more general problem for the engineering of unusual FAs, seed-specific overexpression of tung tree DGAT2 with or without accompanying silencing of AtDGAT1 expression by RNAi was also performed in an established homozygous transgenic line producing ESA. Analysis of T2 and T3 lines showed that VfDGAT2 expression resulted in expected increases in ESA levels compared with the parental lines. However, an additional 22% increase in ESA was observed when AtDGAT1 expression was reduced by RNAi. As with the results for castor DGAT2, the tung tree DGAT2 data clearly indicate that AtDGAT1 supports incorporation of normal FA into TAG and that AtDGAT1 and transgenic VfDGAT2 compete to mobilize different DAG and acyl-CoA substrates into TAG. Arabidopsis does not use the classic Kennedy pathway for the synthesis of TAG but instead, uses PC as an intermediate in TAG synthesis (Fig. 1). It is PC that is the substrate for FA desaturation and modification enzymes (including RcFAH12 and VfFADX). Thus, DAG and acyl-CoA derived from PC will contain HFA or ESA, and these, in turn, become available for TAG synthesis through DGAT and PDAT activities (Bates and Browse, 2011). In transgenic Arabidopsis seeds expressing RcFAH12, this pathway creates a metabolic bottleneck for the incorporation of HFA into HFA-TAG. In these plants, 1-HFA-DAG is synthesized but not efficiently converted into PC. Radiolabeling experiments indicate that 1-HFA-TAG is converted into 2-HFA-TAG and degraded (Bates and Browse, 2011), and the metabolic bottleneck also results in feedback inhibition of FA synthesis and reduced seed oil content (Bates et al., 2014). Both RcDGAT2 and RcPDAT1A expressions can increase incorporation of HFA into TAG and partially alleviate the reduction of seed oil in RcFAH transgenic lines (van Erp et al., 2011; Bates et al., 2014). Consistent with these previous findings, our dgat1 CL7 RcDGAT2 RcPDAT1A transgenics showed a significant additional increase in seed HFA content (Figs. 3B and 4). These plants also had a very substantial increase in the proportion of 2-HFA-TAG species relative to the dgat1 CL7 RcDGAT2 parental line (Fig. 6). However, the elimination of AtDGAT1 and the coexpression of RcDGAT2 and RcPDAT1A did not significantly change the proportion of 3-HFA-TAG. Our attempts to reduce competition from the endogenous AtPDAT1 isozyme were evidently unsuccessful. The pdat1-amiRNA/RcPDAT1A lines that we were able to recover had seed characteristics that are indistinguishable from the RcPDAT1A lines lacking the pdat1-amiRNA construct (Figs. 4 and 5) and contained levels of AtPDAT1 mRNA that were at least 40% of the levels in parental control lines. Many of the dgat1 CL7 RcDGAT2 pdat1-amiRNA RcPDAT1A T1 seeds were shrunken and did not germinate on either soil or agar medium supplemented with 1% (w/v) Suc. It is possible that these inviable seeds included transgenics with strong suppression of AtPDAT1 expression. Our results indicate that substrate competition between endogenous and transgenic acyltransferases may be a general problem in the engineering of unusual FAs in heterologous plant systems. The greater than 20% increases in HFA and ESA that we observed indicate that reducing enzyme competition can provide unique avenues for metabolic engineering. Hopefully this strategy will bring us a step closer to the engineering of crop plants with high levels of unusual FAs for industrial or health purposes. Although our results characterize examples of isozyme competition that lead to reduced HFA or ESA accumulation, it is also possible for endogenous and transgenic isozymes to act synergistically. The Reduced Oleate Desaturation1 (ROD1) gene encodes PDCT that interconverts PC and DAG (Lu et al., 2009). PDCT is required for efficient HFA incorporation into TAG in transgenic Arabidopsis expressing the castor hydroxylase, because the proportion of HFA in seed oil is reduced by >50% in rod1 mutants compared with ROD1 RcFAH12 controls (Hu et al., 2012). In additional experiments, Hu et al. (2012) showed that RcROD1 expression could compensate for the loss of the Arabidopsis enzyme when transformed into rod1 mutants expressing RcFAH12 or could lead to increased HFA when expressed in an RcFAH12 transgenic line that is the wild type at the ROD1 locus. Importantly, expression of RcROD1 also provided an increase in HFA (from 24.7% to 28.5% of seed FA) when expressed in the CL7 RcDGAT2 background used in our experiments (Hu et al., 2012). These results point to the potential for obtaining further increases in HFA accumulation in seeds through the production of higher order multitransgenic lines. MATERIALS AND METHODS Plant Growth Conditions and Transformation The Arabidopsis (Arabidopsis thaliana) lines and growth conditions were similar to those described in van Erp et al. (2011) unless otherwise mentioned. Plant transformation was achieved using the floral dip method (Clough and Bent, 1998). All lines are in the Columbia-0 wild-type background, except that dgat1-2 is in the Wassilewskija background. For experiments comparing genotypes, plants of the different lines were randomly distributed across the pots and trays that were used. Plants were grown in environmentally controlled chambers under continuous fluorescent illumination of 120 to 150 µmol quanta m−2 s−1 with 70% relative humidity at 22°C. Genotyping of Plants and RT-PCR Analysis For genotyping of AtDGAT1, plant material was ground (30 Hz for 30 s; Tissuelyser II; QIAGEN) followed by extraction of genomic DNA using a QIAcube (QIAGEN). A portion of AtDGAT1 was amplified with gene-specific primers (Supplemental Table S3) flanking both sides of the point mutation, and the amplified fragment was gel purified with a QIAcube. Sequencing was performed (Eurofins MWG Operon) to determine which plants were the wild type, heterozygous, or homozygous for the point mutation in AtDGAT1. To prepare RNA, young siliques were harvested approximately 8 to 12 d after flowering from plants, flash frozen in liquid nitrogen, and stored at −80°C. Developing seeds were scraped from each silique on a petri dish on dry ice and collected in 1.5-mL Eppendorf tubes kept in liquid nitrogen. For RT-PCR analysis, total RNA was extracted using the RNeasy Plant Mini Kit (Qiagen). Samples were treated with RNase-Free DNase (Qiagen) using the on-column DNase digestion method according to the manufacturer’s protocol. cDNA was synthesized using the SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen/Life Technologies). To confirm that the transgenic plants were expressing their respective transgenes, RT-PCR was performed using primers listed in Supplemental Table S3. This confirmed that the transgenic lines expressed the genes of interest. Design and Cloning of amiRNA against AtPDAT1 To suppress expression of AtPDAT1 in our transgenic plant lines, a 21-mer sequence targeting AtPDAT1 was designed (WMD3 Web MicroRNA Designer, Arabidopsis cDNA [The Arabidopsis Information Resource 9]; minimum no. of included targets, one; accepted off targets, zero; S. Ossowski, J. Fitz, R. Schwab, M. Riester, and D. Weigel, personal communication). To prevent suppression of expression of other Arabidopsis genes in our plant lines, a Target Search was performed against Arabidopsis cDNAs with the 21-mer sequences (The Arabidopsis Information Resource 9_cdna_20090619; no. of mismatches, five), and no targets were found. To prevent suppression of the castor (Ricinus communis) genes in our transgenic plant lines (RcFAH12, RcDGAT2, and RcPDAT1A), a Target Search was performed against castor mRNAs with the 21-mer sequences (castor PUT v163a [Plant Genome Database]; no. of mismatches, five), and no targets were found. The best 21 mer with no other targets in Arabidopsis and castor was selected for additional analysis (5′-TTGCGGGTTATACGTAGTGTA-3′; hybridization energy = −40.49 kcal mol−1). This 21 mer targets the AtPDAT1 mRNA in the 3′-UTR. Cloning of the pdat1-amiRNA and RcPDAT1A in a Multigene Vector The 21-mer sequence was used to replace the stem loops in the MIR319a precursor (Ossowski et al., 2008). Primer sequences used for cloning are described in Supplemental Table S1. The pdat1-amiRNA construct was cloned in the RS3GSeed DsRed vector in between the glycinin promoter and the glycinin 3′-UTR. The pdat1-amiRNA construct was digested with EcoRI and XbaI. The RS3GSeed DsRed vector was digested with EcoRI and XbaI and dephosphorylated with calf intestinal phosphatase (New England Biolabs). The EcoRI/XbaI pdat1-amiRNA fragment was ligated into the linearized vector using the Quick Ligation Kit (NEB). RcPDAT1A was cloned in the pdat1-amiRNA vector behind the oleosin promoter and followed by the oleosin 3′-UTR. The RcPDAT1A cDNA was PCR amplified with primers containing NotI sites (Supplemental Table S1) and digested with NotI HF (New England Biolabs). The PCR product was cloned in the NotI side of the calf intestinal phosphatase-treated pKMS2 vector behind the oleosin promoter. The oleosin promoter RcPDAT1A oleosin 3′-UTR construct was cut out of the pKMS2 vector using AscI. This construct was ligated into the AscI site of the pdat1-amiRNA vector to generate the pdat1-amiRNA RcPDAT1A vector. Construction of Tung Tree Gene Vectors and Expression in Plants The open reading frames (ORFs) for tung tree (Vernicia fordii) FADX and tung tree DGAT2 as well as the AtDGAT1 RNAi construct were assembled in various components of a flexible set of cloning vectors and plant binary plasmids before transformation into Agrobacterium tumefaciens. The cDNA for tung tree FADX (Dyer et al., 2002) was PCR amplified using a forward primer containing an NotI site and a reverse primer containing an SacII site adjacent to the stop codon. This product was digested with NotI and SacII and ligated into plasmid pK8 that had been similarly treated. pK8 contains the strong seed-specific promoter from the Phaseolus vulgaris gene (Slightom et al., 1983) and the cauliflower mosaic virus 35S transcriptional terminator; both are flanked on their respective distal ends by AscI sites. The resulting plasmid was named pB190. The AscI cassette from pB190 was transferred to the AscI site of the plant binary plasmid pB9, which carries a kanamycin resistance gene for bacterial selection and a gene for basta (glufosinate ammonium) herbicide resistance for selection in plants. The resulting VfFADX binary plasmid, designated pE181, was transformed into Agrobacterium spp. strain GV3101; kanamycin- and gentamycin-resistant colonies were cultured in liquid media and used to transform the fad3 fae1 double-mutant line of Arabidopsis (Smith et al., 2003) by floral dip. A seed-specific shuttle plasmid containing an N-terminal myelocytomatosis viral oncogene homolog (myc) epitope-tagged tung tree DGAT2 was generated by PCR amplification of the native DGAT2 ORF with a forward primer in which the initiator Met codon has been replaced by a KasI site and a reverse primer containing an SacII site adjacent to the stop codon. After KasI/SacII digestion, this product was ligated into similarly digested pB50, a shuttle plasmid containing the Arabidopsis 2S-3 promoter (Guerche et al., 1990), the soybean (Glycine max) glycinin G1 subunit transcriptional terminator (Sims and Goldberg, 1989), and a multiple cloning site containing sites that allow for production of N-terminal myc epitope fusions. The AscI fragment (representing the promoter-gene-terminator cassette) from this plasmid, pB240, was transferred into the corresponding site of the plant binary vector pB110 to form plasmid E278. In turn, pE278 was modified by MluI digestion and ligation of the AscI fragment of pJ6, which contains the soybean β-conglycinin promoter (Sato et al., 2004) driving the expression of an RNAi hairpin for AtDGAT1. The RNAi portion of this plasmid contains an intron from the 5′-UTR of AtFAD2 flanked by a 592-bp region of the AtDGAT1 ORF (base pairs 303–894) cloned in inverted orientations. The binary plasmid carrying both the tung tree DGAT2 overexpression cassette and the AtDGAT1 RNAi cassette is called pE290. The sequences of all primers used to generate these plasmid constructs are included in Supplemental Table S1. ESA production stabilized at approximately 8% in the T4 generation of E181 plants. T5 plants from this line were retransformed with Agrobacterium spp. bearing either pE278 or pE290. Red T1 seeds were sown on soil and grown to maturity followed by seed lipid extraction and analysis for determination of ESA content. Two lines, one representing the highest T2 18:3Ɗ9cis, 11trans, and 13trans producer and one representing a level near the average of the T2 population, were selected for each double transgenic and carried forward to the T3 generation. Seed samples from T3 plants producing either uniformly red or uniformly brown seeds were analyzed by gas chromatography. FA Analysis by Gas Chromatography Analysis of HFA was performed as described in van Erp et al., 2011. Lipids from Arabidopsis seed containing ESA were extracted as follows. Approximately 30 mg of seeds and three to four 2.3-mm chrome steel beads (BioSpec Products, Inc.) were added into a 2-mL Eppendorf tube with 500 µL of hexane followed by 5 min of agitation on a Bead Beater. The extract was centrifuged (13,000g for 2 min) to remove debris, and a portion (300 µL) of the extract was transferred to a 13- × 100-mm glass Corning culture tube. Hexane (700 µL) and sodium methoxide in methanol (400 µL) were added followed by 10 min of incubation at room temperature with intermittent shaking. The reactions were quenched by the addition of 2 mL of hexane and 2 mL of saturated NaCl solution, and the phases were separated by centrifugation in a tabletop swinging bucket centrifuge. Two milliliters of upper layer was transferred to vials, capped, and analyzed immediately or stored at −20° until needed. Gas chromatography was conducted as described in Shockey et al. (2011). Whenever possible, all samples containing ESA were prepared using amber-colored glassware to reduce exposure of ESA to light. Regiochemical analysis of TAG was performed as described (van Erp et al., 2011). Determination of Seed FA Content Seed FA content was determined according to the protocol in Li et al. (2006), except that 200 µL of toluene was added to each sample, butylated hydroxytoluene was omitted, and 20 seeds were used for each measurement. Seed weights were determined by counting seed numbers in samples of 1 to 2 mg of seeds. Lipid Extraction and Characterization of TAG Species Lipid extraction and characterization of TAG species were performed as described in van Erp et al. (2011). Supplemental Data The following supplemental materials are available. Supplemental Figure S1 . Expression of RcDGAT2 transcript in the parental CL7 RcDGAT2 line and the dgat1 CL7 RcDGAT2 line obtained by crossing. Supplemental Figure S2. Extended analysis of lines accumulating ESA. Supplemental Table S1. FA compositions of samples of T3 seeds of transgenic lines accumulating HFAs characterized in this study. Supplemental Table S2. FA compositions of samples of T3 seeds of transgenic lines accumulating ESA characterized in this study. Supplemental Table S3. Primers used. ACKNOWLEDGMENTS We thank all of the members of the laboratory of J.B. who contributed to this article, the greenhouse staff for help with growing plants, Dr. Jim Wallis (Institute of Biological Chemistry, Washington State University) for discussions and help with formatting the figures, and Catherine Mason (Southern Regional Research Center, U.S. Department of Agriculture-Agricultural Research Service) for technical assistance with seed lipid extraction and gas chromatography analysis. Glossary cDNA complementary DNA DAG diacylglycerol ESA α-eleostearic acid FA fatty acid HFA hydroxy fatty acid ORF open reading frame PC phosphatidylcholine PDCT phosphatidylcholine:diacylglycerol cholinephosphotransferase RT reverse transcription TAG triacylglycerol UTR untranslated region LITERATURE CITED Badami RC , Patil KB ( 1980 ) Structure and occurrence of unusual fatty acids in minor seed oils . Prog Lipid Res 19 : 119 – 153 Google Scholar Crossref Search ADS PubMed WorldCat Bates PD , Browse J ( 2011 ) The pathway of triacylglycerol synthesis through phosphatidylcholine in Arabidopsis produces a bottleneck for the accumulation of unusual fatty acids in transgenic seeds . Plant J 68 : 387 – 399 Google Scholar Crossref Search ADS PubMed WorldCat Bates PD , Browse J ( 2012 ) The significance of different diacylgycerol synthesis pathways on plant oil composition and bioengineering . Front Plant Sci 3 : 147 Google Scholar Crossref Search ADS PubMed WorldCat Bates PD , Johnson SR, Cao X, Li J, Nam JW, Jaworski JG, Ohlrogge JB, Browse J ( 2014 ) Fatty acid synthesis is inhibited by inefficient utilization of unusual fatty acids for glycerolipid assembly . Proc Natl Acad Sci USA 111 : 1204 – 1209 Google Scholar Crossref Search ADS PubMed WorldCat Bates PD , Stymne S, Ohlrogge J ( 2013 ) Biochemical pathways in seed oil synthesis . Curr Opin Plant Biol 16 : 358 – 364 Google Scholar Crossref Search ADS PubMed WorldCat Broun P , Somerville C ( 1997 ) Accumulation of ricinoleic, lesquerolic, and densipolic acids in seeds of transgenic Arabidopsis plants that express a fatty acyl hydroxylase cDNA from castor bean . Plant Physiol 113 : 933 – 942 Google Scholar Crossref Search ADS PubMed WorldCat Burgal J , Shockey J, Lu C, Dyer J, Larson T, Graham I, Browse J ( 2008 ) Metabolic engineering of hydroxy fatty acid production in plants: RcDGAT2 drives dramatic increases in ricinoleate levels in seed oil . Plant Biotechnol J 6 : 819 – 831 Google Scholar Crossref Search ADS PubMed WorldCat Cahoon EB , Dietrich CR, Meyer K, Damude HG, Dyer JM, Kinney AJ ( 2006 ) Conjugated fatty acids accumulate to high levels in phospholipids of metabolically engineered soybean and Arabidopsis seeds . Phytochemistry 67 : 1166 – 1176 Google Scholar Crossref Search ADS PubMed WorldCat Carlsson AS , Yilmaz JL, Green AG, Stymne S, Hofvander P ( 2011 ) Replacing fossil oil with fresh oil - with what and for what? Eur J Lipid Sci Technol 113 : 812 – 831 Google Scholar Crossref Search ADS PubMed WorldCat Clough SJ , Bent AF ( 1998 ) Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana . Plant J 16 : 735 – 743 Google Scholar Crossref Search ADS PubMed WorldCat Dauk M , Lam P, Kunst L, Smith MA ( 2007 ) A FAD2 homologue from Lesquerella lindheimeri has predominantly fatty acid hydroxylase activity . Plant Sci 173 : 43 – 49 Google Scholar Crossref Search ADS WorldCat Doyle JJ , Schuler MA, Godette WD, Zenger V, Beachy RN, Slightom JL ( 1986 ) The glycosylated seed storage proteins of Glycine max and Phaseolus vulgaris. Structural homologies of genes and proteins . J Biol Chem 261 : 9228 – 9238 Google Scholar Crossref Search ADS PubMed WorldCat Dyer JM , Chapital DC, Kuan JC, Mullen RT, Turner C, McKeon TA, Pepperman AB ( 2002 ) Molecular analysis of a bifunctional fatty acid conjugase/desaturase from tung: implications for the evolution of plant fatty acid diversity . Plant Physiol 130 : 2027 – 2038 Google Scholar Crossref Search ADS PubMed WorldCat Dyer JM , Stymne S, Green AG, Carlsson AS ( 2008 ) High-value oils from plants . Plant J 54 : 640 – 655 Google Scholar Crossref Search ADS PubMed WorldCat Guerche P , Tire C, De Sa FG, De Clercq A, Van Montagu M, Krebbers E ( 1990 ) Differential expression of the Arabidopsis 2S albumin genes and the effect of increasing gene family size . Plant Cell 2 : 469 – 478 Google Scholar Crossref Search ADS PubMed WorldCat Hu Z , Ren Z, Lu C ( 2012 ) The phosphatidylcholine diacylglycerol cholinephosphotransferase is required for efficient hydroxy fatty acid accumulation in transgenic Arabidopsis . Plant Physiol 158 : 1944 – 1954 Google Scholar Crossref Search ADS PubMed WorldCat Katavic V , Reed DW, Taylor DC, Giblin EM, Barton DL, Zou J, Mackenzie SL, Covello PS, Kunst L ( 1995 ) Alteration of seed fatty acid composition by an ethyl methanesulfonate-induced mutation in Arabidopsis thaliana affecting diacylglycerol acyltransferase activity . Plant Physiol 108 : 399 – 409 Google Scholar Crossref Search ADS PubMed WorldCat Kim HU , Lee KR, Go YS, Jung JH, Suh MC, Kim JB ( 2011 ) Endoplasmic reticulum-located PDAT1-2 from castor bean enhances hydroxy fatty acid accumulation in transgenic plants . Plant Cell Physiol 52 : 983 – 993 Google Scholar Crossref Search ADS PubMed WorldCat Knutzon DS , Hayes TR, Wyrick A, Xiong H, Maelor Davies H, Voelker TA ( 1999 ) Lysophosphatidic acid acyltransferase from coconut endosperm mediates the insertion of laurate at the sn-2 position of triacylglycerols in lauric rapeseed oil and can increase total laurate levels . Plant Physiol 120 : 739 – 746 Google Scholar Crossref Search ADS PubMed WorldCat Kroon JT , Wei W, Simon WJ, Slabas AR ( 2006 ) Identification and functional expression of a type 2 acyl-CoA:diacylglycerol acyltransferase (DGAT2) in developing castor bean seeds which has high homology to the major triglyceride biosynthetic enzyme of fungi and animals . Phytochemistry 67 : 2541 – 2549 Google Scholar Crossref Search ADS PubMed WorldCat Kunst L , Taylor DC, Underhill EW ( 1992 ) Fatty-acid elongation in developing seeds of Arabidopsis thaliana . Plant Physiol Biochem 30 : 425 – 434 Google Scholar OpenURL Placeholder Text WorldCat Li R , Yu K, Hatanaka T, Hildebrand DF ( 2010 ) Vernonia DGATs increase accumulation of epoxy fatty acids in oil . Plant Biotechnol J 8 : 184 – 195 Google Scholar Crossref Search ADS PubMed WorldCat Li Y , Beisson F, Pollard M, Ohlrogge J ( 2006 ) Oil content of Arabidopsis seeds: the influence of seed anatomy, light and plant-to-plant variation . Phytochemistry 67 : 904 – 915 Google Scholar Crossref Search ADS PubMed WorldCat Lu C , Fulda M, Wallis JG, Browse J ( 2006 ) A high-throughput screen for genes from castor that boost hydroxy fatty acid accumulation in seed oils of transgenic Arabidopsis . Plant J 45 : 847 – 856 Google Scholar Crossref Search ADS PubMed WorldCat Lu C , Xin Z, Ren Z, Miquel M, Browse J ( 2009 ) An enzyme regulating triacylglycerol composition is encoded by the ROD1 gene of Arabidopsis . Proc Natl Acad Sci USA 106 : 18837 – 18842 Google Scholar Crossref Search ADS PubMed WorldCat Mhaske V , Beldjilali K, Ohlrogge J, Pollard M ( 2005 ) Isolation and characterization of an Arabidopsis thaliana knockout line for phospholipid: diacylglycerol transacylase gene (At5g13640) . Plant Physiol Biochem 43 : 413 – 417 Google Scholar Crossref Search ADS PubMed WorldCat Napier JA ( 2007 ) The production of unusual fatty acids in transgenic plants . Annu Rev Plant Biol 58 : 295 – 319 Google Scholar Crossref Search ADS PubMed WorldCat Napier JA , Graham IA ( 2010 ) Tailoring plant lipid composition: designer oilseeds come of age . Curr Opin Plant Biol 13 : 330 – 337 Google Scholar Crossref Search ADS PubMed WorldCat Ossowski S , Schwab R, Weigel D ( 2008 ) Gene silencing in plants using artificial microRNAs and other small RNAs . Plant J 53 : 674 – 690 Google Scholar Crossref Search ADS PubMed WorldCat Palatnik JF , Allen E, Wu X, Schommer C, Schwab R, Carrington JC, Weigel D ( 2003 ) Control of leaf morphogenesis by microRNAs . Nature 425 : 257 – 263 Google Scholar Crossref Search ADS PubMed WorldCat Routaboul JM , Benning C, Bechtold N, Caboche M, Lepiniec L ( 1999 ) The TAG1 locus of Arabidopsis encodes for a diacylglycerol acyltransferase . Plant Physiol Biochem 37 : 831 – 840 Google Scholar Crossref Search ADS PubMed WorldCat Sato S , Xing AQ, Ye XG, Schweiger B, Kinney A, Graef G, Clemente T ( 2004 ) Production of gamma-linolenic acid and stearidonic acid in seeds of marker-free transgenic soybean . Crop Sci 44 : 646 – 652 Google Scholar OpenURL Placeholder Text WorldCat Shockey J , Chapital D, Gidda S, Mason C, Davis G, Klasson KT, Cao H, Mullen R, Dyer J ( 2011 ) Expression of a lipid-inducible, self-regulating form of Yarrowia lipolytica lipase LIP2 in Saccharomyces cerevisiae . Appl Microbiol Biotechnol 92 : 1207 – 1217 Google Scholar Crossref Search ADS PubMed WorldCat Shockey JM , Gidda SK, Chapital DC, Kuan JC, Dhanoa PK, Bland JM, Rothstein SJ, Mullen RT, Dyer JM ( 2006 ) Tung tree DGAT1 and DGAT2 have nonredundant functions in triacylglycerol biosynthesis and are localized to different subdomains of the endoplasmic reticulum . Plant Cell 18 : 2294 – 2313 Google Scholar Crossref Search ADS PubMed WorldCat Sims TL , Goldberg RB ( 1989 ) The glycinin Gy1 gene from soybean . Nucleic Acids Res 17 : 4386 Google Scholar Crossref Search ADS PubMed WorldCat Slightom JL , Sun SM, Hall TC ( 1983 ) Complete nucleotide sequence of a French bean storage protein gene: phaseolin . Proc Natl Acad Sci USA 80 : 1897 – 1901 Google Scholar Crossref Search ADS PubMed WorldCat Smith M , Moon H, Kunst L ( 2000 ) Production of hydroxy fatty acids in the seeds of Arabidopsis thaliana . Biochem Soc Trans 28 : 947 – 950 Google Scholar Crossref Search ADS PubMed WorldCat Smith MA , Moon H, Chowrira G, Kunst L ( 2003 ) Heterologous expression of a fatty acid hydroxylase gene in developing seeds of Arabidopsis thaliana . Planta 217 : 507 – 516 Google Scholar Crossref Search ADS PubMed WorldCat Sonntag NOV ( 1979 ) Composition and characteristics of individual fats and oils . In D Swern , ed, Bailey's Industrial Oil and Fat Products. John Wiley & Sons , New York , pp 289 – 477 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ståhl U , Carlsson AS, Lenman M, Dahlqvist A, Huang B, Banas W, Banas A, Stymne S ( 2004 ) Cloning and functional characterization of a phospholipid:diacylglycerol acyltransferase from Arabidopsis . Plant Physiol 135 : 1324 – 1335 Google Scholar Crossref Search ADS PubMed WorldCat Stuitje AR , Verbree EC, van der Linden KH, Mietkiewska EM, Nap JP, Kneppers TJ ( 2003 ) Seed-expressed fluorescent proteins as versatile tools for easy (co)transformation and high-throughput functional genomics in Arabidopsis . Plant Biotechnol J 1 : 301 – 309 Google Scholar Crossref Search ADS PubMed WorldCat van de Loo FJ , Broun P, Turner S, Somerville C ( 1995 ) An oleate 12-hydroxylase from Ricinus communis L. is a fatty acyl desaturase homolog . Proc Natl Acad Sci USA 92 : 6743 – 6747 Google Scholar Crossref Search ADS PubMed WorldCat van Erp H , Bates PD, Burgal J, Shockey J, Browse J ( 2011 ) Castor phospholipid:diacylglycerol acyltransferase facilitates efficient metabolism of hydroxy fatty acids in transgenic Arabidopsis . Plant Physiol 155 : 683 – 693 Google Scholar Crossref Search ADS PubMed WorldCat Vanhercke T , Wood CC, Stymne S, Singh SP, Green AG ( 2013 ) Metabolic engineering of plant oils and waxes for use as industrial feedstocks . Plant Biotechnol J 11 : 197 – 210 Google Scholar Crossref Search ADS PubMed WorldCat Voelker T , Kinney AJ ( 2001 ) Variations in the biosynthesis of seed-storage lipids . Annu Rev Plant Physiol Plant Mol Biol 52 : 335 – 361 Google Scholar Crossref Search ADS PubMed WorldCat Wayne LL , Wallis JG, Kumar R, Markham JE, Browse J ( 2013 ) Cytochrome b5 reductase encoded by CBR1 is essential for a functional male gametophyte in Arabidopsis . Plant Cell 25 : 3052 – 3066 Google Scholar Crossref Search ADS PubMed WorldCat Xu J , Carlsson AS, Francis T, Zhang M, Hoffman T, Giblin ME, Taylor DC ( 2012 ) Triacylglycerol synthesis by PDAT1 in the absence of DGAT1 activity is dependent on re-acylation of LPC by LPCAT2 . BMC Plant Biol 12 : 4 Google Scholar Crossref Search ADS PubMed WorldCat Zhang M , Fan J, Taylor DC, Ohlrogge JB ( 2009 ) DGAT1 and PDAT1 acyltransferases have overlapping functions in Arabidopsis triacylglycerol biosynthesis and are essential for normal pollen and seed development . Plant Cell 21 : 3885 – 3901 Google Scholar Crossref Search ADS PubMed WorldCat Zou J , Wei Y, Jako C, Kumar A, Selvaraj G, Taylor DC ( 1999 ) The Arabidopsis thaliana TAG1 mutant has a mutation in a diacylglycerol acyltransferase gene . Plant J 19 : 645 – 653 Google Scholar Crossref Search ADS PubMed WorldCat Author notes 1 This work was supported by the U.S. National Science Foundation Plant Genome Research Program (grant nos. DBI–0701919 and IOS–1339385) and the Agricultural Research Center at Washington State University. 2 Present address: Plant Biology and Crop Science Department, Rothamsted Research, Harpenden AL5 2JQ, UK. 3 Present address: College of Agronomy, Northwest A and F University, Number 3 Taicheng Road, 2216 Research Building, South Campus, Yangling, Shaanxi 712100, China. * Address correspondence to [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: John Browse ([email protected]). [OPEN] Articles can be viewed without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.114.254110 © 2015 American Society of Plant Biologists. All Rights Reserved. © The Author(s) 2015. Published by Oxford University Press on behalf of American Society of Plant Biologists. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Suzuki, Mami; Nakabayashi, Ryo; Ogata, Yoshiyuki; Sakurai, Nozomu; Tokimatsu, Toshiaki; Goto, Susumu; Suzuki, Makoto; Jasinski, Michal; Martinoia, Enrico; Otagaki, Shungo; Matsumoto, Shogo; Saito, Kazuki; Shiratake, Katsuhiro

Wada, Shinya; Hayashida, Yasukazu; Izumi, Masanori; Kurusu, Takamitsu; Hanamata, Shigeru; Kanno, Keiichi; Kojima, Soichi; Yamaya, Tomoyuki; Kuchitsu, Kazuyuki; Makino, Amane; Ishida, Hiroyuki

doi: 10.1104/pp.15.00242

Ishihara, Hirofumi; Obata, Toshihiro; Sulpice, Ronan; Fernie, Alisdair R.; Stitt, Mark

doi: 10.1104/pp.15.00209pmid: 25810096

Abstract Protein synthesis and degradation represent substantial costs during plant growth. To obtain a quantitative measure of the rate of protein synthesis and degradation, we supplied 13CO2 to intact Arabidopsis (Arabidopsis thaliana) Columbia-0 plants and analyzed enrichment in free amino acids and in amino acid residues in protein during a 24-h pulse and 4-d chase. While many free amino acids labeled slowly and incompletely, alanine showed a rapid rise in enrichment in the pulse and a decrease in the chase. Enrichment in free alanine was used to correct enrichment in alanine residues in protein and calculate the rate of protein synthesis. The latter was compared with the relative growth rate to estimate the rate of protein degradation. The relative growth rate was estimated from sequential determination of fresh weight, sequential images of rosette area, and labeling of glucose in the cell wall. In an 8-h photoperiod, protein synthesis and cell wall synthesis were 3-fold faster in the day than at night, protein degradation was slow (3%–4% d−1), and flux to growth and degradation resulted in a protein half-life of 3.5 d. In the starchless phosphoglucomutase mutant at night, protein synthesis was further decreased and protein degradation increased, while cell wall synthesis was totally inhibited, quantitatively accounting for the inhibition of growth in this mutant. We also investigated the rates of protein synthesis and degradation during leaf development, during growth at high temperature, and compared synthesis rates of Rubisco large and small subunits of in the light and dark. Protein synthesis accounts for a significant part of the energy required for plant growth (Penning de Vries et al., 1974; Penning de Vries, 1975; Amthor, 2000). The rate of protein synthesis can be qualitatively assessed by investigating the proportion of ribosomes loaded into polysomes (Bailey-Serres, 1999; Beilharz and Preiss, 2004). Polysome loading decreases in various stress treatments, including water deficit (Hsiao, 1970; Kawaguchi et al., 2003, 2004) and hypoxia (Branco-Price et al., 2005, 2008). In nonstressed plants in a light/dark cycle, polysome loading is high in the light and decreases in the dark (Piques et al., 2009; Pal et al., 2013). Polysome loading is increased by sugar addition to cell cultures (Nicolaï et al., 2006) and is closely correlated with Suc levels during diurnal cycles in Arabidopsis (Arabidopsis thaliana) rosettes (Pal et al., 2013). Information about polysome loading and ribosome abundance has been used to model the rate of protein synthesis in Arabidopsis rosettes (Piques et al., 2009; Pal et al., 2013). This approach indicated that the rate of protein synthesis is higher in the daytime than at night, when it is constrained by the rate of starch degradation and the resulting availability of carbon (C) and energy. It also indicated that the global rate of protein synthesis is only slightly higher than that needed for growth, indicating that there is only a low rate of protein degradation. Protein degradation is required to repair damaged proteins, to adjust the protein complement to environmental conditions, and to remobilize and recycle amino acids (Huffaker and Peterson, 1974; Nelson et al., 2014b). Protein degradation is thought to make a major impact on growth efficiency (Amthor, 2000). It has been estimated that protein degradation accounts for 20% to 30% of the ATP produced by root respiration (Scheurwater et al., 2000). Protein degradation is accelerated in low nitrogen (Lattanzi et al., 2005; Lehmeier et al., 2013) and by shading (Pons et al., 1993; Scheurwater et al., 2000); this may promote nitrogen remobilization to young leaves in more favorable light conditions. Protein degradation is accelerated under C starvation to provide an alternative source of C for respiration (Brouquisse et al., 1991; Araújo et al., 2011; Izumi et al., 2013; Pilkington et al., 2015). Protein degradation may also increase under other stresses, including high temperature in wheat (Triticum aestivum) roots (Ferguson et al., 1990) and osmotic stress in barley (Hordeum vulgare) leaves (Dungey and Davies, 1982). It may also change during leaf development, with various studies indicating that degradation is higher in expanding than in fully grown leaves (Schaefer et al., 1981; Barneix et al., 1988) and also increases in old leaves (Dungey and Davies, 1982; Bouma et al., 1994). Protein half-life depends on the rate of degradation and on the rate at which preexisting protein is diluted by growth. Global protein half-life was estimated to be 4 to 8 d in grass roots (Scheurwater et al., 2000). A similar protein half-life was predicted by modeling in Arabidopsis rosettes (Piques et al., 2009) and by experimental studies of 508 individual proteins in barley leaves (Nelson et al., 2014a). The rate of protein synthesis is usually measured by supplying a pulse of a labeled precursor, often followed by a chase in the absence of label, and monitoring the temporal kinetics of label incorporation and depletion (Huffaker and Peterson, 1974; Nelson et al., 2014b). While radioisotopes were used in the past, stable isotopes are easier to detect and quantify in specific compounds. Labeling experiments should meet several criteria. First, the labeled precursor must be supplied without surgical or other disturbance of the plant, as this may alter the rates of protein synthesis and degradation. Second, the supplied labeled precursor should not alter internal metabolite pools or fluxes. Third, movement of label per se does not provide information about flux. This requires information about the extent to which the incoming label is diluted by internal pools. Fourth, accurate measurements of the rate of growth are required to assess how much of the protein synthesis represents flux to growth and how much represents protein turnover (viz., the replacement of degraded proteins). Many studies in the past did not meet all these criteria; in particular, they rarely measured enrichment in the free amino acid precursor pools. Labeled amino acids have been widely used in microbes to study protein synthesis. To decrease dilution by internal pools, an amino acid is used that the microbe cannot synthesize. Similar approaches have been used in human cell cultures (Ong et al., 2002; Ong and Mann, 2006). While labeled amino acids have also been used in algae and plants, the resulting data are usually qualitative. In contrast to microbes and animals, plants are able to synthesize all proteinaceous amino acids. As a result, the supplied labeled amino acid is diluted by endogenously synthesized amino acid and may also be catabolized and converted to other amino acids that are incorporated into protein. Differential uptake and internal fluxes of amino acids can lead to complex temporal labeling kinetics and inhomogenous labeling of proteins, depending on which amino acid is supplied (He et al., 1991). In a comparison of [2H]Leu, [13C]Arg, and [2H]Lys labeling, Gruhler et al. (2005) concluded that all three substrates showed label dilution, and to varying extents. While stable isotopes of Lys are incorporated into the proteome with an enrichment of 83% to 91% in the dark, this falls to 58% in the light (Schütz et al., 2011). Another problem in plants is that labeled amino acids need to be added at high levels to generate a large increase in isotope abundance, which can alter metabolite pools and fluxes. Problems can also arise due to complex patterns of feedback regulation in amino acid biosynthesis pathways (Galili, 2002). In mammals, individual amino acids like Leu regulate and modify the rate of protein turnover (Buse and Reid, 1975; Pannemans et al., 1997). An additional complication in plants is that some amino acids are involved in further metabolic pathways. Examples include the use of Gly and Ser in photorespiration, Trp for auxin synthesis (Rapparini et al., 1999), and the aromatic amino acids Phe, Tyr, and Trp as precursors for the synthesis of phenylpropanoids and flavonoids (Fraser and Chapple, 2011). Protein synthesis can also be measured using stable isotopes of water. Movement of label from H2 18O into amino acids is rather slow (Zhou et al., 2012). Label moves rapidly from 2H2O into amino acids and proteins (Mitra et al., 1976; Yang et al., 2010), but discrimination against 2H2O can lead to changes in metabolite pools and fluxes, to changes in gene expression indicative of stress, and to inhibition of growth (Thomson et al., 1963; Sacchi and Cocucci, 1992; Kushner et al., 1999; Yang et al., 2010). Yang et al. (2010) showed that inhibitory effects on growth can be decreased, although not totally prevented, by decreasing enrichment to 30% and that the stress-related changes in transcript levels were less marked when a protocol was used in which seedlings were pulsed with 2H2O and the labeling kinetics were analyzed in a chase with water. However, while 2H2O can be readily supplied to seedlings, movement through a larger plant is likely to be slow and rather heterogenous. Another approach is to supply 15NO3 or 15NH4 to the medium of algae (Martin et al., 2012; Mastrobuoni et al., 2012) or the rooting medium of higher plants (Masclaux-Daubresse and Chardon, 2011; Nelson et al., 2014a, 2014b). Labeling with 15N is especially useful in studies that use liquid chromatography-tandem mass spectrometry to analyze the labeling kinetics of signature peptides of individual proteins (Li et al., 2012; Nelson et al., 2014a, 2014b). However, there are some disadvantages in higher plants, including the rather slow labeling kinetics of amino acids (Yang et al., 2010). This may be due to the time that elapses until the label is transported through the plant and assimilated as well as to dilution by large internal pools (Lattanzi et al., 2005; Lehmeier et al., 2013). This also makes it difficult to rapidly and completely remove 15NO3 or 15NH4 in a chase. Label can also be introduced as 13CO2 in the atmosphere. This has three potential advantages. First, 13CO2 is immediately incorporated into metabolism via photosynthesis, allowing the introduction of isotope into intact plants without any perturbation of metabolism and growth. Second, 13CO2 and unlabeled CO2 and be rapidly interchanged, allowing rapid commencement of a pulse and rapid and complete removal of external label at the start of a chase. Third, at least in principle, 13CO2 should label all amino acids. 13CO2 labeling has been used to investigate fluxes in photosynthetic metabolism in algae (Huege et al., 2007; Young et al., 2011) and higher plants (Szecowka et al., 2013; Ma et al., 2014). It has also been used to investigate the synthesis and degradation of a small set of abundant proteins (Chen et al., 2011). However, quantitative measurements are complicated by slow and incomplete changes in enrichment in amino acids. In a recent study, we showed that the rate and extent of labeling of amino acids by 13CO2 is highly variable; after a 60-min pulse, some amino acids like Ala were quite highly (more than 70%) labeled, others like Phe, Gly, and Ser were only moderately (20%–40%) labeled, and many, including Glu, Thr, Lys, Leu, Ile, Pro, and Asn, were very slowly and incompletely (less than 10%) labeled (Szecowka et al., 2013). Here, we present a relatively simple approach that uses 13CO2 pulse-chase analysis to measure the global rate of protein synthesis and degradation in intact soil-grown Arabidopsis plants. After labeling, gas chromatography time-of-flight mass spectrometry (GC-TOF-MS) is used to analyze enrichment in a large number of free amino acids and in their residues in protein. We identify which free amino acids show the cleanest labeling kinetics and use enrichment in these pools and the corresponding amino acid residues in protein to estimate the rates of protein synthesis and degradation. We also present a related method to quantify flux to cell wall polysaccharides. RESULTS Experimental Design Szecowka et al. (2013) applied a pulse of 13CO2 for various times up to 60 min, starting in the middle of the photoperiod after metabolic steady state had been achieved, to allow the computationally demanding modeling of fluxes in central metabolism. Based on the slow and incomplete labeling kinetics of amino acids by Szecowka et al. (2013) and estimates of the rate of protein synthesis based on ribosome abundance and usage (Piques et al., 2009), we reasoned that a longer pulse would be required to measure flux into protein. We decided to pulse for an entire day/night cycle and to chase for 4 d, taking samples at various times during this treatment. We also decided to commence labeling at dawn to minimize isotope dilution by internal pools, which are usually small at dawn (Gibon et al., 2006, 2009; Sulpice et al., 2014). In particular, starch is very low at dawn, accumulates in the light, and is remobilized to provide C for respiration and growth at night (Smith and Stitt, 2007; Stitt and Zeeman, 2012). By labeling from dawn onward, we hoped to achieve a high enrichment in starch that would allow high enrichment to be maintained in metabolites at night. To prevent dilution by dark fixation during the night, the 13CO2 pulse was continued through until dawn. Arabidopsis Columbia-0 (Col-0) plants were grown in a controlled climate chamber on soil under a short photoperiod (8 h of light/16 h of dark) for 18 d and transferred into two labeling chambers for 3 d before starting the labeling experiment. The labeling chambers were located inside the climate chamber, and each was large enough to hold 17 10-cm-diameter pots with five plants per pot. The chambers were flushed (5 L min−1) with humidified air mixtures (79% N2, 21% oxygen) containing 450 µL L−1 unlabeled or labeled CO2. The CO2 concentration reflected that measured in the climate chamber. The pulse was started by switching from unlabeled CO2 to 13CO2 just before dawn on day 21, and the chase was started by switching from 13CO2 to unlabeled CO2 at the following dawn. To prevent incorporation of label by plants that were being grown for subsequent experiments, the air mix exiting the labeling chamber was passed through a soda lime column. In routine experiments, plants were harvested at seven time points: at dawn just before starting the pulse (to measure natural enrichment), after 4 and 8 h in light, just before dawn at the end of the pulse, during the chase 4 h into first light period, and at dawn 1 and 4 d later. The extracts were separated into soluble and insoluble fractions and analyzed by GC-TOF-MS to determine enrichment in the individual free amino acids in the soluble fraction and, after chemical hydrolysis of the insoluble fraction, in the individual amino acids in protein. Although the analysis required only small amounts (20 mg fresh weight) of leaf material, for routine work, we typically pooled five rosettes per sample. In parallel, we measured the rate of rosette growth by weighing rosettes harvested at each sampled time point and by acquiring sequential images of the rosette area. Enrichment Kinetics of Free Amino Acids We were able to quantify enrichment in 16 free amino acids (Fig. 1; Supplemental Table S1A). The temporal kinetics of enrichment varies between amino acids and was often slow and incomplete. During the pulse, enrichment of individual amino acids ranged between 83% and 11% at the end of day and between 71% and 15% at the end of night. During the chase, enrichment decreased rapidly for some amino acids and slowly for others. This underlines the importance of information about enrichment in free amino acid pools for accurate estimation of the rate of protein synthesis. Figure 1. Open in new tabDownload slide Enrichment kinetics of free amino acids and amino acids in protein in Col-0 growing in an 8-h photoperiod. The temporal kinetics of enrichment in 16 free amino acids (black lines) are shown as well as, for 12 of these, the kinetics of enrichment of the amino acid residue in protein (red lines).The x axis depicts time during the pulse and the chase (depicted with red and blue bars at the top of each graph, respectively). The y axis shows the percentage enrichment in the free amino acid or the amino acid residue in protein. The light period and night are indicated by white and gray shading, respectively. Col-0 was grown for 21 d in an 8-h photoperiod and then pulsed with 13CO2 for 24 h, followed by a 4-d chase. Plants were harvested at dawn before the start of the pulse, after 4 and 8 h in the light, at the next dawn at the end of the pulse, at 4 h in the first light period of the chase, and at dawn at the end of days 1 and 4 of the chase. Each harvest comprised four pots, each containing five plants. The experiment was performed four times with separately grown batches of plants. The data are provided in Supplemental Data Set S1, the calculated enrichment in free amino acids and amino acids in protein in Supplemental Table S1, plots of individual isotopomers in Figure 2 and Supplemental Figure S2, and the calculated rates of protein synthesis and degradation in Table I. The plots show average (μ) of four biological replicates with sd (μ ± sd). Error bars are omitted when they are smaller than the symbol. DAHP, 3-Deoxy-d-arabino-heptulosonic acid 7-phosphate; PEP, phosphoenolpyruvate; 3PGA, 3-phosphoglyceraldehyde. Figure 1. Open in new tabDownload slide Enrichment kinetics of free amino acids and amino acids in protein in Col-0 growing in an 8-h photoperiod. The temporal kinetics of enrichment in 16 free amino acids (black lines) are shown as well as, for 12 of these, the kinetics of enrichment of the amino acid residue in protein (red lines).The x axis depicts time during the pulse and the chase (depicted with red and blue bars at the top of each graph, respectively). The y axis shows the percentage enrichment in the free amino acid or the amino acid residue in protein. The light period and night are indicated by white and gray shading, respectively. Col-0 was grown for 21 d in an 8-h photoperiod and then pulsed with 13CO2 for 24 h, followed by a 4-d chase. Plants were harvested at dawn before the start of the pulse, after 4 and 8 h in the light, at the next dawn at the end of the pulse, at 4 h in the first light period of the chase, and at dawn at the end of days 1 and 4 of the chase. Each harvest comprised four pots, each containing five plants. The experiment was performed four times with separately grown batches of plants. The data are provided in Supplemental Data Set S1, the calculated enrichment in free amino acids and amino acids in protein in Supplemental Table S1, plots of individual isotopomers in Figure 2 and Supplemental Figure S2, and the calculated rates of protein synthesis and degradation in Table I. The plots show average (μ) of four biological replicates with sd (μ ± sd). Error bars are omitted when they are smaller than the symbol. DAHP, 3-Deoxy-d-arabino-heptulosonic acid 7-phosphate; PEP, phosphoenolpyruvate; 3PGA, 3-phosphoglyceraldehyde. We were searching for amino acids that showed a rapid increase to high enrichment in the light, retained high enrichment during the pulse in the night, and underwent a rapid and nearly complete decrease in enrichment in the chase. Ala showed the fastest and most complete increase of enrichment in the pulse (79%, 81%, and 65% after 4 and 8 h of illumination and at the end of night, respectively) and a rapid decrease in enrichment in the chase. Asp showed slightly lower enrichment. Ala and Asp are synthesized via reversible aminotransferase reactions from the central metabolites pyruvate and oxaloacetate, respectively. Ser and Gly also showed a rapid increase in enrichment in the light, but their enrichment decreased during the night (from 83% and 76% at dusk to 53% and 15% at dawn). Their rapid labeling in the light reflects their formation during photorespiration; reasons for the decline during the night will be discussed later. Glu and Gln showed relatively slow labeling in the light (22% and 22% after 4 h of illumination and 57% and 53% after 8 h of illumination, respectively) and a further rise during the night (65% and 71% at dawn, respectively). The slow labeling in the light was unexpected, because these amino acids are derived via aminotransferase reactions from the central organic acid α-oxoglutarate and are also intermediates in the GOGAT pathway, which is very active in the light (see “Discussion”). Most other amino acids are synthesized via long biosynthetic pathways. Enrichment in the aromatic amino acids (Phe and Tyr) rose quite quickly in the light and decreased rapidly in the chase, but they were labeled more slowly than Ala and also showed a marked decrease at night during the pulse. Ile, Leu, Lys, Met, and Val showed a gradual increase in enrichment during the light period, incomplete enrichment at dusk (44%–58%), a decrease in enrichment at night during the pulse, and a slow decrease in enrichment during the chase. The slowest labeling kinetic was found for Pro. Slow and incomplete labeling kinetics can result for several reasons, including dilution of label by internal pools, synthesis of amino acids from unlabeled precursors, or the presence of compartmented pools with different labeling kinetics. We inspected our data to provide insights into the contribution of these different factors. Our GC-TOF-MS analysis provided information about enrichment in sugars and organic acids (Supplemental Fig. S1; Supplemental Table S1A). Labeling of malate was rapid and high in the pulse (73%, 79%, and 79% after 4 and 8 h of illumination and at the following dawn, respectively) and decreased rapidly in the chase. Enrichment was slightly lower in fumarate and succinate. Enrichment of pyruvate was high in the light (71% and 77% after 4 and 8 h, respectively), declined at night (35%), and declined rapidly in the chase. Enrichment in citrate was very low in the light (less than 1%), rose strongly in the night (52%), remained at this level for the first 4 h of the chase in the light, and decreased by the following dawn. The slow labeling of Glu and Gln in the light may be a consequence of the slow labeling of citrate (see “Discussion”). Labeling of Suc and reducing sugars was slow and incomplete in the pulse and declined slowly in the chase. They may provide a source of unlabeled C in the night. We inspected the detailed labeling pattern of each amino acid to learn whether there are strongly compartmented pools of any of the amino acids. If an amino acid occurs as a single pool or as multiple pools that are in relatively rapid exchange, during the pulse the unlabeled C0 isotopomer (i.e. all C atoms are unlabeled) will decrease to a low value, intermediate isotopomers with a low number of 13C atoms will rise transiently and then decrease to a low value, and heavily labeled isotopomers that consist predominantly or entirely of labeled atoms will rise to a high value. This pattern is seen in the light for Ala (Fig. 2A), Gly, Ser, and Asp (Supplemental Fig. S2) and, somewhat more slowly, for Met. If incomplete labeling of an amino acid is due to slow labeling of one pool, the same pattern will be followed, but more slowly. This pattern is seen for Glu (Fig. 2B) and Gln (Supplemental Fig. S2). If incomplete labeling is due to the presence of two or more pools, of which one or more is not labeled or is only very slowly labeled, the C0 isotopomer will decrease to an intermediate value, and the remainder of the pool will be present as heavily labeled isotopomers. This pattern was seen at dusk for Lys (Fig. 2C), Pro, Tyr, Phe, Val, Leu, and Ile (Supplemental Fig. S2). Figure 2. Open in new tabDownload slide Labeling kinetics for all detected isotopomers of the free amino acid pools. A, Ala. B, Glu. C, Lys. C0, C1, C2, C3, and CX correspond to the isotopomer with no 13C atoms (i.e. all atoms are unlabeled), with one 13C atom, with two 13C atoms, with three 13C atoms, and with x 13C atoms, respectively. The experimental design is described in the legend of Figure 1. The data are provided in Supplemental Data Set S1, and the calculated enrichment values are provided in Supplemental Table S1. Figure 2. Open in new tabDownload slide Labeling kinetics for all detected isotopomers of the free amino acid pools. A, Ala. B, Glu. C, Lys. C0, C1, C2, C3, and CX correspond to the isotopomer with no 13C atoms (i.e. all atoms are unlabeled), with one 13C atom, with two 13C atoms, with three 13C atoms, and with x 13C atoms, respectively. The experimental design is described in the legend of Figure 1. The data are provided in Supplemental Data Set S1, and the calculated enrichment values are provided in Supplemental Table S1. We also investigated whether some amino acids were present at a much higher level at dusk than at dawn. Such diurnal changes are due probably to synthesis of the amino acid from newly fixed C in the light and use of this pool at night. This will lead to any compartmented unlabeled pools making only a small contribution to the total pool at dusk but a large contribution at dawn. This analysis was only possible for the amino acids for which standards were included in the GC-TOF-MS analysis to allow quantification. There was an especially large increase for the photorespiratory intermediate Gly (37-fold) and a smaller increase for Ser (3.7-fold; Supplemental Table S2). Based on other studies, we can also expect a large diurnal change for Gln (Foyer et al., 2003; Fritz et al., 2006; Tschoep et al., 2009). Enrichment Kinetics of Amino Acids in Protein GC-TOF-MS analysis of the protein hydrolysate provided enrichment data for 12 amino acids (Fig. 1; Supplemental Table S1B; Gln, Asn, Met, and Tyr could not be reliably detected in the hydolysate). In total, 12 amino acids were detected in both the free and protein pools. Enrichment of amino acids in protein increased during the pulse in the light and rose further in the night, although more slowly than in the light. This provides qualitative evidence that protein synthesis continues at night. Enrichment at the end of the 24-h pulse varied more than 2-fold, depending on the amino acid. Like free amino acids, the highest enrichment was found for Ala, Gly, and Ser (20.7%, 20.1%, and 20.2%, respectively), with somewhat lower enrichment in Asp (15.4%). The lowest enrichment was for Pro (8.3%). We investigated whether these differences in enrichment in protein could be explained by the differences in average enrichment of the free amino acids. To do this, we plotted the increase in enrichment of the amino acid in protein during the light period against the enrichment in the free amino acid pool at dusk (Supplemental Fig. S3A). A similar plot was made comparing the increase in enrichment of the amino acid in protein during the night with the enrichment in the free amino acid pool at dawn (Supplemental Fig. S3B). A small group of amino acids showed high enrichment in the free pool and a high rate of incorporation into protein. This group included Ala in the light and dark, Ser in the light and dark, Gly in the light, and Glu in the dark. Another group, including Asp in the light and dark and Phe and Glu in the light, showed slower label incorporation into protein than expected from the enrichment in the free pool. This may reflect slow labeling kinetics, which would result in the average enrichment of the free pool at dawn or dusk being higher than the average during the preceding time interval. Another group, including Lys and Pro in the light and Gly, Pro, Leu, Ile, and Lys in the dark, showed higher label incorporation into protein than expected from enrichment in the free amino acid. This might be explained by the presence of a weakly labeled or unlabeled pool, with the result that the actual precursor pool has a higher enrichment than the average value. Another group, including Thr and Val in the light and dark, Leu and Lys in the light, and Phe in the dark, showed a similar relationship to Ala and Ser, but with lower enrichment and lower incorporation rates into protein. While this might indicate that there is one slowly labeled pool of these amino acids, inspection of the labeling kinetics of their isotopomers (see above; Supplemental Fig. S2) makes it more likely that the situation is actually more complex, with a combination of slow labeling kinetics for one pool and the presence of a weakly or unlabeled pool. Overall, this analysis and that of Figures 1 and 2 and Supplemental Figures S2 and S3 suggest that multiple factors contribute to the complex labeling kinetics of free amino acids. The washout kinetics in protein during the chase were analyzed with semilog plots (Supplemental Fig. S4). Ala, Ser, and Gly had the fastest washout (slopes of −0.23, −0.24, and −0.21), with smaller slopes for other amino acids (−0.18 to −0.11 and even slower for Pro). This shows that, for many amino acids, incomplete loss of label in free pools complicates protein labeling kinetics during the chase. Determination of the Relative Growth Rate Estimation of the rate of protein degradation requires information about the rate of growth. Growth rates are usually given as the relative growth rate (RGR), which is the gain in biomass per unit of existing biomass per day (Poorter and Nagel, 2000). We measured RGR by fitting a regression to a semilog plot of the fresh weight of the plants harvested during the pulse and chase. Such measurements are subject to experimental error because different plants are harvested at each time point and weight varies between individual plants. This could affect the accuracy of our calculation of protein degradation. We took two approaches to validate our estimates of RGR. First, we took sequential images of sets of plants grown in parallel with those that were used for destructive analyses. Destructive harvesting and time-lapse image analysis were in good agreement; in this experiment, they provided estimates for RGR of 0.224 and 0.215 mm2 mm−2 d−1, respectively (Supplemental Fig. S5). Second, we determined RGR by destructive harvesting in three separate large experiments performed on plants growing in the same conditions during the time when we were carrying out our labeling studies. The values were very similar (average [μ] ± sd = 0.221 ± 0.002; Supplemental Fig. S5). For routine estimates of protein degradation, we used the average estimate of RGR from all these replicated experiments. Estimation of the Rate of Protein Synthesis and Degradation The rate of protein synthesis, K s, was estimated from label incorporation into Ala in protein, with a correction for enrichment in free Ala (see “Discussion”; Pocrnjic et al., 1983) using the equation: where and represent average label incorporation into Ala in protein at the start and end of the analyzed time interval, is the average enrichment of free Ala at the end of the light period (see “Discussion”), and (t 2 − t 1) is the duration of the time interval in hours. The rate of synthesis is given as the percentage of the initial protein synthesized in the time interval. To calculate the rate of protein synthesis rate in the light period, t 2 was the end of the light period (8 h), and t 1 was the start of the pulse (0 h). To calculate protein synthesis rate in the dark, t 2 was the end of the 24-h pulse, and t 1 was the end of the light period (8 h). The estimated rate of protein synthesis in the light period (1.95% h−1) was about 3-fold higher than that at night (0.63% h−1). However, as the plants were in an 8-h photoperiod, the amount of protein synthesized at night was 61% of that synthesized in the light period. The protein synthesized during a 24-h cycle was equivalent to about 26% of the protein in the rosette. The rate of protein degradation, K d, was estimated in two ways. The first approach is based on the difference between the estimated rate of protein synthesis in the pulse, K s, and the rate of protein synthesis needed to sustain the observed rate of growth. It is calculated as: where P p is the fractional change in rosette protein content per day during the pulse. The second approach calculates the rate of protein degradation as the difference between the measured decrease in enrichment in Ala in protein during the chase and the decrease that would be expected due to dilution by growth (Yee et al., 2010) as: where k Loss is the rate constant for the loss of label in Ala in protein per day and P c is the fractional change in rosette protein content per day during the chase. k Loss was calculated for Ala by applying linear regression of the natural log-transformed enrichment value versus time over at least three time points. The very low enrichment in free Ala during the chase allowed us to ignore recycling of label into protein. The rosette protein concentration did not change significantly between days 19 and 26 after germination (Supplemental Data Set S1). This allowed us to set the protein concentration terms (P p and P c) at unity for the calculation of protein degradation in rosettes. The rate of protein degradation (i.e. protein synthesis in excess of that required for growth) per 24-h cycle was estimated at 3.5% d−1 from the pulse data and 3.1% d−1 from the chase data. Thus, the majority of the protein synthesis (26% d−1) represents flux to growth, rather than degradation. Combining the dilution of protein by growth with the rate of protein degradation yielded a global protein half-life of 3.1 to 3.5 d. Different Leaf Growth Stages We next applied this method to investigate the rate of protein synthesis and degradation at different stages of leaf development. Ribosome abundance (Dean and Leech, 1982) and, by implication, the rate of protein synthesis are high in young growing leaves and low in mature leaves. Leaves at six stages of development (Fig. 3A) were harvested at the beginning and end of a 24-h pulse and processed to provide information about the enrichment of free amino acids and of amino acids in protein (Supplemental Data Set S2). As large numbers of plants were required to provide enough material for the young leaves, we did not attempt to separate fluxes in the light period and night. We did not apply a chase over 4 d because in this time the leaves will change their developmental stage. Leaf relative growth rate (LRGR) was estimated from images of individual leaves on sequential days spanning the pulse (Fig. 3B). It should be noted that the oldest leaves are still growing slowly. The rate of protein synthesis was calculated from the average enrichment of free Ala (81%) in a subsample of whole rosettes harvested at the end of the light period. This was validated by checking that Ala enrichment at dawn at the end of the pulse was similar in all six leaf stages at dawn at the end of the pulse (Fig. 3C) but lower than at dusk, as was seen previously. The estimated rate of protein synthesis was fastest (42.8% d−1) in the youngest rapidly growing leaf (LRGR = 0.48) and lowest (15.6%–16% d−1) in the oldest leaves (LRGR = 0.12–0.13; Fig. 3D; Supplemental Table S3). Figure 3. Open in new tabDownload slide Leaf growth rates, protein synthesis rates, and protein degradation at different stages of leaf development. Col-0 was grown for 21 d in an 8-h photoperiod and then pulsed with 13CO2 for 24 h. Plants were harvested at dawn before the start of the pulse and at dawn at the end of the pulse and separated into leaf stages for further analysis. Each sample consisted of 20 leaves from 20 plants. A, Images of a typical plant at 21, 22, and 23 d after sowing (DAS). The numbers on the leaves correspond to leaf numbers on the x axes in B to D. B, LRGR at each leaf stage during the chase, estimated from sequential images of the rosette from three plants. C, Enrichment values of free Ala for each leaf stage at the end of the pulse from a sample of 20 pooled leaves. D, Comparison of the estimated rate of protein synthesis and the rate of protein synthesis that was required for growth. The complete bar (a + b) represents the rate of protein synthesis, estimated from enrichment in Ala in protein corrected for enrichment with free Ala. The gray bar (a) represents the rate of protein synthesis required for growth. This was calculated from LRGR corrected for the decrease in protein per day in a leaf at that stage. The rate of protein degradation (b) was calculated as the difference between the rate of protein synthesis and the amount of protein required for growth. The data, including estimated changes in leaf protein content with time, are provided in Supplemental Data Set S2. Calculated values for protein synthesis and degradation are provided in Supplemental Table S3. Figure 3. Open in new tabDownload slide Leaf growth rates, protein synthesis rates, and protein degradation at different stages of leaf development. Col-0 was grown for 21 d in an 8-h photoperiod and then pulsed with 13CO2 for 24 h. Plants were harvested at dawn before the start of the pulse and at dawn at the end of the pulse and separated into leaf stages for further analysis. Each sample consisted of 20 leaves from 20 plants. A, Images of a typical plant at 21, 22, and 23 d after sowing (DAS). The numbers on the leaves correspond to leaf numbers on the x axes in B to D. B, LRGR at each leaf stage during the chase, estimated from sequential images of the rosette from three plants. C, Enrichment values of free Ala for each leaf stage at the end of the pulse from a sample of 20 pooled leaves. D, Comparison of the estimated rate of protein synthesis and the rate of protein synthesis that was required for growth. The complete bar (a + b) represents the rate of protein synthesis, estimated from enrichment in Ala in protein corrected for enrichment with free Ala. The gray bar (a) represents the rate of protein synthesis required for growth. This was calculated from LRGR corrected for the decrease in protein per day in a leaf at that stage. The rate of protein degradation (b) was calculated as the difference between the rate of protein synthesis and the amount of protein required for growth. The data, including estimated changes in leaf protein content with time, are provided in Supplemental Data Set S2. Calculated values for protein synthesis and degradation are provided in Supplemental Table S3. To calculate protein degradation, we compared the rate of protein synthesis with the amount of protein required for growth. In this case, it was necessary to include information about the decrease in protein concentration during leaf development. The daily decrease in protein concentration at each leaf stage was estimated in a separate experiment, in which the rate of leaf initiation was scored and leaves 1 to 6 were harvested from 21-d-old plants to determine fresh weight and protein concentration at each leaf stage (Supplemental Data Set S2). The decrease in leaf protein was estimated as 12%, 15%, 1%, and 16% d−1 for leaves 6, 5, 4, and 3, respectively. The estimated requirement for protein for growth decreased from about 42% d−1 in the youngest leaves to 12% d−1 in the oldest leaves (Fig. 3D). At all leaf stages, most of the protein synthesis represented flux to growth (Fig. 3D). The estimated rate of protein degradation was negligible in the young leaves and about 8%, 3%, and 4% d−1 in leaves 3, 2, and 1, respectively. The estimates in young leaves are subject to error because they represent the difference between two large values, both of which are susceptible to experimental noise. The values found in the mature leaves resemble those seen in a mature rosette, as expected because mature leaves contribute most of the rosette biomass. Higher Growth Temperature High temperature results in an increase in respiration, especially maintenance respiration (Penning de Vries et al., 1979; Amthor, 2000; Pyl et al., 2012). While the main maintenance costs are thought to include protein turnover and the maintenance of electrical, pH, and ion gradients across membranes (Penning de Vries, 1975; Amthor, 2000), they are difficult to quantify from molecular information (Amthor, 2000; Cheung et al., 2013; Sweetlove et al., 2014). We applied our method to investigate whether higher temperature leads to a major increase in the rate of protein degradation. We grew Arabidopsis in an 8-h photoperiod at 20°C or 28°C for 21 d, applied a 13CO2 pulse for 24 h, harvested plants at dawn at the end of the pulse and after a 1-, 2-, 3-, or 4-d chase, determined the enrichment of free Ala and of Ala in protein, and estimated the rate of protein synthesis and, by comparison with parallel measurements of RGR, the rate of protein degradation (Table II; data provided in Supplemental Data Set S3). Protein synthesis was 15% higher at 28°C than at 20°C (Table II). As we did not collect samples at dusk, we used the enrichment data at dawn to estimate protein synthesis rates. This may result in a slight overestimation, because enrichment in free Ala decreases slightly during the night. For this reason, we did not attempt to use data from the pulse to estimate degradation rates. Degradation rate estimated from the chase showed a 17% increase at 28°C compared with 20°C (3.7% and 3.1% d−1, respectively). Thus, protein synthesis and protein degradation increase by approximately the same extent between 20°C and 28°C. Estimated rates of protein synthesis and degradation for Col-0 grown in an 8-h photoperiod Table I. Estimated rates of protein synthesis and degradation for Col-0 grown in an 8-h photoperiod Col-0 was grown for 21 d in an 8-h photoperiod and then pulsed with 13CO2 for 24 h, followed by a 4-d chase. The experimental design is described in the legend of Figure 1. The data are provided in Supplemental Data Set S1 and calculated enrichment values in Supplemental Table S1. The rate of protein synthesis was calculated by correcting enrichment in Ala in protein by the enrichment in free Ala at the end of the day. The calculations of protein degradation used an RGR of 0.221 mg fresh weight mg−1 fresh weight d−1, which was the average of three biological replicate experiments in these growth conditions (Supplemental Fig. S5). Results are estimated from the average of four biological replicates. RGR . Protein Synthesis . Protein Degradation . Half-Life . Average 24-h Cycle . Light (per Hour) . Dark (per Hour) . Light:Dark Ratio . Pulse . Chase . Pulse . Chase . mg mg−1 d−1 % total protein % total protein d−1 d 0.221 25.62 1.95 0.63 3.1 3.52 3.07 3.11 3.49 RGR . Protein Synthesis . Protein Degradation . Half-Life . Average 24-h Cycle . Light (per Hour) . Dark (per Hour) . Light:Dark Ratio . Pulse . Chase . Pulse . Chase . mg mg−1 d−1 % total protein % total protein d−1 d 0.221 25.62 1.95 0.63 3.1 3.52 3.07 3.11 3.49 Open in new tab Table I. Estimated rates of protein synthesis and degradation for Col-0 grown in an 8-h photoperiod Col-0 was grown for 21 d in an 8-h photoperiod and then pulsed with 13CO2 for 24 h, followed by a 4-d chase. The experimental design is described in the legend of Figure 1. The data are provided in Supplemental Data Set S1 and calculated enrichment values in Supplemental Table S1. The rate of protein synthesis was calculated by correcting enrichment in Ala in protein by the enrichment in free Ala at the end of the day. The calculations of protein degradation used an RGR of 0.221 mg fresh weight mg−1 fresh weight d−1, which was the average of three biological replicate experiments in these growth conditions (Supplemental Fig. S5). Results are estimated from the average of four biological replicates. RGR . Protein Synthesis . Protein Degradation . Half-Life . Average 24-h Cycle . Light (per Hour) . Dark (per Hour) . Light:Dark Ratio . Pulse . Chase . Pulse . Chase . mg mg−1 d−1 % total protein % total protein d−1 d 0.221 25.62 1.95 0.63 3.1 3.52 3.07 3.11 3.49 RGR . Protein Synthesis . Protein Degradation . Half-Life . Average 24-h Cycle . Light (per Hour) . Dark (per Hour) . Light:Dark Ratio . Pulse . Chase . Pulse . Chase . mg mg−1 d−1 % total protein % total protein d−1 d 0.221 25.62 1.95 0.63 3.1 3.52 3.07 3.11 3.49 Open in new tab Protein degradation in Arabidopsis grown at 20°C and 28°C Table II. Protein degradation in Arabidopsis grown at 20°C and 28°C Col-0 was grown for 21 d in an 8-h photoperiod at either 20°C or 28°C, then pulsed with 13CO2 for 24 h, followed by a 4-d chase. Plants were harvested at dawn before the start of the pulse, at dawn at the end of the pulse, and at dawn after 1, 2, 3, and 4 d of chase. The data are provided in Supplemental Data Set S3. Three or four samples of five plants were harvested at each time point for 28°C or 20°C, respectively. The rate of protein synthesis was estimated from the 13C enrichment in protein in Ala at the end of a 24-h pulse and normalized with the enrichment in free Ala at the end of the pulse. The rate of protein degradation was estimated by comparing the decrease in enrichment in Ala in protein during the chase with RGR. This normalization will lead to a slight overestimation of the protein synthesis. The RGR at 20°C is the mean of three biological experiments, and that at 28°C is the mean of three biological replicates. Results are estimated from the average of four biological replicates from 20°C-grown plants and the average of three biological replicates from 28°C-grown plants. Treatment . RGR . Protein Synthesis . Protein Degradation . Half-Life . mg mg−1 d−1 % d−1 d 20°C 0.221 31.7 3.07 3.49 28°C 0.244 37.3 3.70 2.99 Ratio 20°C:28°C 0.91 0.85 0.83 Treatment . RGR . Protein Synthesis . Protein Degradation . Half-Life . mg mg−1 d−1 % d−1 d 20°C 0.221 31.7 3.07 3.49 28°C 0.244 37.3 3.70 2.99 Ratio 20°C:28°C 0.91 0.85 0.83 Open in new tab Table II. Protein degradation in Arabidopsis grown at 20°C and 28°C Col-0 was grown for 21 d in an 8-h photoperiod at either 20°C or 28°C, then pulsed with 13CO2 for 24 h, followed by a 4-d chase. Plants were harvested at dawn before the start of the pulse, at dawn at the end of the pulse, and at dawn after 1, 2, 3, and 4 d of chase. The data are provided in Supplemental Data Set S3. Three or four samples of five plants were harvested at each time point for 28°C or 20°C, respectively. The rate of protein synthesis was estimated from the 13C enrichment in protein in Ala at the end of a 24-h pulse and normalized with the enrichment in free Ala at the end of the pulse. The rate of protein degradation was estimated by comparing the decrease in enrichment in Ala in protein during the chase with RGR. This normalization will lead to a slight overestimation of the protein synthesis. The RGR at 20°C is the mean of three biological experiments, and that at 28°C is the mean of three biological replicates. Results are estimated from the average of four biological replicates from 20°C-grown plants and the average of three biological replicates from 28°C-grown plants. Treatment . RGR . Protein Synthesis . Protein Degradation . Half-Life . mg mg−1 d−1 % d−1 d 20°C 0.221 31.7 3.07 3.49 28°C 0.244 37.3 3.70 2.99 Ratio 20°C:28°C 0.91 0.85 0.83 Treatment . RGR . Protein Synthesis . Protein Degradation . Half-Life . mg mg−1 d−1 % d−1 d 20°C 0.221 31.7 3.07 3.49 28°C 0.244 37.3 3.70 2.99 Ratio 20°C:28°C 0.91 0.85 0.83 Open in new tab Nucleus-Encoded Large Subunit and Plastid-Encoded Small Subunit of Rubisco We were interested to learn if our approach could be adapted to measure the synthesis and degradation of individual proteins. For this, we focused on the highly abundant protein Rubisco (Eckardt et al., 1997). Rubisco consists of a nucleus-encoded small subunit (RBCS) and a plastid-encoded large subunit (RBCL). Almost half of the ribosomes in leaves are located in the plastid (Detchon and Possingham, 1972; Dean and Leech, 1982; Piques et al., 2009). They are involved in the synthesis of plastid-encoded proteins, including RBCL and components of complexes in the thylakoid membrane. It is generally thought that plastid protein synthesis is inhibited in the dark (Marín-Navarro et al., 2007). In agreement, at night there is a larger decrease in loading of plastid ribosomes than cytosolic ribosomes into polysomes (Piques et al., 2009; Pal et al., 2013). Arabidopsis Col-0 was pulsed with 13CO2 for 24 h, harvested at dusk and dawn, and applied to SDS-PAGE, and the 52- and 18-kD bands were eluted and chemically digested (Allen et al., 2012) for analysis with GC-TOF-MS (Supplemental Data Set S4). Synthesis rates were calculated using enrichment in free Ala, assuming that there is similar enrichment in the cytosol and plastid. In the light, synthesis was faster for RBCL (1.2% h−1) than for RBCS (0.80% h−1), although the difference was not statistically significant (P = 0.13). Synthesis rates were lower at night but similar for RBCL and RBCS (0.37 and 0.36% h−1, respectively; Table III). The decrease in RBCL synthesis between light and dark (3.2-fold) resembled the decrease in the global rate of protein synthesis (Table I). Thus, RBCL translation is not preferentially inhibited in the dark. Synthesis rate of RBCL and RBCS in the light period and the night Table III. Synthesis rate of RBCL and RBCS in the light period and the night Col-0 was grown for 21 d in an 8-h photoperiod and then pulsed with 13CO2 for 24 h. Plants were harvested at dusk and at dawn at the end of the pulse. Protein was separated by PAGE, and bands corresponding to RBCL and RBCS were eluted and analyzed. The data are provided in Supplemental Data Set S4. The results are given as means ± sd (n = 3 biological replicates except for RBCS in the dark, where n = 2). Protein Synthesis Rate . RBCL . RBCS . Light:Dark Ratio . Light . Dark . Light . Dark . RBCL . RBCS . Per period (%) 9.27 ± 1.01 5.84 ± 1.05 6.40 ± 3.21 5.75 ± 2.70 1.59 1.11 Per hour (%) 1.16 ± 0.13 0.37 ± 0.07 0.80 ± 0.40 0.36 ± 0.17 3.17 2.23 Protein Synthesis Rate . RBCL . RBCS . Light:Dark Ratio . Light . Dark . Light . Dark . RBCL . RBCS . Per period (%) 9.27 ± 1.01 5.84 ± 1.05 6.40 ± 3.21 5.75 ± 2.70 1.59 1.11 Per hour (%) 1.16 ± 0.13 0.37 ± 0.07 0.80 ± 0.40 0.36 ± 0.17 3.17 2.23 Open in new tab Table III. Synthesis rate of RBCL and RBCS in the light period and the night Col-0 was grown for 21 d in an 8-h photoperiod and then pulsed with 13CO2 for 24 h. Plants were harvested at dusk and at dawn at the end of the pulse. Protein was separated by PAGE, and bands corresponding to RBCL and RBCS were eluted and analyzed. The data are provided in Supplemental Data Set S4. The results are given as means ± sd (n = 3 biological replicates except for RBCS in the dark, where n = 2). Protein Synthesis Rate . RBCL . RBCS . Light:Dark Ratio . Light . Dark . Light . Dark . RBCL . RBCS . Per period (%) 9.27 ± 1.01 5.84 ± 1.05 6.40 ± 3.21 5.75 ± 2.70 1.59 1.11 Per hour (%) 1.16 ± 0.13 0.37 ± 0.07 0.80 ± 0.40 0.36 ± 0.17 3.17 2.23 Protein Synthesis Rate . RBCL . RBCS . Light:Dark Ratio . Light . Dark . Light . Dark . RBCL . RBCS . Per period (%) 9.27 ± 1.01 5.84 ± 1.05 6.40 ± 3.21 5.75 ± 2.70 1.59 1.11 Per hour (%) 1.16 ± 0.13 0.37 ± 0.07 0.80 ± 0.40 0.36 ± 0.17 3.17 2.23 Open in new tab The Starchless phosphoglucomutase Mutant In a last application, we investigated protein synthesis and degradation in the starchless phosphoglucomutase (pgm) mutant. Analysis of transcript and metabolite profiles has revealed that pgm starves at night in short photoperiods (Gibon et al., 2004b, 2006; Usadel et al., 2008). We posed two questions. First, how strongly is protein synthesis inhibited in pgm at night? Analysis of polysome loading using density gradient centrifugation (Pal et al., 2013) showed that some polysomes are still present at night in pgm, although fewer than in Col-0. We wanted to distinguish between two explanations for this observation: that despite extreme C starvation, there is still a low rate of protein synthesis or that the ribosomal RNA retrieved in the polysome fraction in pgm at night is not in active polysomes. Second, based on an increase in the levels of minor amino acids in the night in pgm, it has been proposed that protein catabolism is activated (Gibon et al., 2004b, 2006; Izumi et al., 2013). This should be detectable as a higher rate of protein degradation. We would also expect a faster decay of the enrichment in free amino acids in the night, due to increased recycling of predominantly unlabeled amino acids from protein. The pgm mutant grows very poorly in short-day conditions. For this reason, we grew the mutant initially in a 12-h photoperiod before transferring it to an 8-h photoperiod, also increasing the duration of the growth period to obtain a rosette biomass similar to that in 21-d-old wild-type Col-0. Wild-type Col-0 and pgm were then exposed to a 24-h 13CO2 pulse and a 4-d chase, harvesting at dawn before the start of the pulse, at dusk, 4 h into the night, at dawn, and after a 4-d chase (for original data, see Supplemental Data Set S5; for enrichment data, see Supplemental Table S4). The time point at 4 h into the night was included because pgm accumulates high levels of sugars in the light and depletes them during the first 3 to 4 h of the night (Gibon et al., 2004b, 2006). C starvation is unlikely to develop until after this time; indeed, polysome loading decreases gradually in parallel with the depletion of sugars during the first hours of the night (Pal et al., 2013). Enrichment in free amino acids increased in a similar manner in Col-0 and pgm in the light but decayed about 2-fold more rapidly in pgm from 4 h onward in the night, pointing to rapid protein degradation in the night (for Ala, see Fig. 4; for other amino acids, see Supplemental Fig. S6). The only exception was Asn, whose enrichment increased at night in pgm (see “Discussion”). Figure 4. Open in new tabDownload slide Enrichment in free Ala and in Ala residues in protein in wild-type Col-0 and the starchless pgm mutant. The pgm mutant was grown for 25 d in a 12-h photoperiod and shifted to an 8-h photoperiod 3 d prior to the experiment. Col-0 was grown for 21 d in an 8-h photoperiod. Col-0 and pgm were pulsed with 13CO2 for 24 h, followed by a 4-d chase. Plants were harvested at five time points: at dawn before the start of the pulse, at dusk, 4 h into the night, at dawn at the end of the pulse, and at dawn after a 4-d chase (data not shown). The data are provided in Supplemental Data Set S5, enrichment in free amino acids and amino acids in protein in Supplemental Table S4, and estimated rates of protein synthesis and protein degradation in Table IV. The experiment was repeated three times, taking one sample of five plants at each time point in each experiment. The plot shows average (μ) of three biological replicates with sd (μ ± sd). The absence of error bars means that they were smaller than the symbols. The increase in enrichment in amino acids in protein in pgm between 4 and 16 h into the night was significant (P = 0.016). AA, Amino acid. Figure 4. Open in new tabDownload slide Enrichment in free Ala and in Ala residues in protein in wild-type Col-0 and the starchless pgm mutant. The pgm mutant was grown for 25 d in a 12-h photoperiod and shifted to an 8-h photoperiod 3 d prior to the experiment. Col-0 was grown for 21 d in an 8-h photoperiod. Col-0 and pgm were pulsed with 13CO2 for 24 h, followed by a 4-d chase. Plants were harvested at five time points: at dawn before the start of the pulse, at dusk, 4 h into the night, at dawn at the end of the pulse, and at dawn after a 4-d chase (data not shown). The data are provided in Supplemental Data Set S5, enrichment in free amino acids and amino acids in protein in Supplemental Table S4, and estimated rates of protein synthesis and protein degradation in Table IV. The experiment was repeated three times, taking one sample of five plants at each time point in each experiment. The plot shows average (μ) of three biological replicates with sd (μ ± sd). The absence of error bars means that they were smaller than the symbols. The increase in enrichment in amino acids in protein in pgm between 4 and 16 h into the night was significant (P = 0.016). AA, Amino acid. The estimated rates of protein synthesis and degradation are summarized in Table IV. For Col-0, the absolute rates and the changes between the light period and the night resemble those in the experiment of Table I. For pgm, the rate of protein synthesis in the light was slightly but not significantly lower than that of Col-0 (1.8% and 2.1% h−1, respectively). The rate of protein synthesis decreased in the first 4 h of the night but was similar to that in Col-0 (0.54% and 0.68% h−1, respectively). In the remainder of the night, the rate decreased further in pgm (0.29% h−1) but was maintained in Col-0 (0.7% h−1). The rate of protein synthesis over the entire 24-h cycle was 29% lower in pgm (18% d−1) than in Col-0 (25.3% d−1). Comparison of the average protein synthesis rate with RGR revealed a higher average rate of protein degradation in pgm than in Col-0 (4.3% compared with 3.2% d−1, respectively). Analysis of the chase yielded slightly higher degradation rates for both genotypes but confirmed that degradation was faster in pgm than in wild-type Col-0 (6.1% and 4.8% d−1, respectively). Comparison of the rates of protein synthesis and degradation in Col-0 and the starchless pgm mutant Table IV. Comparison of the rates of protein synthesis and degradation in Col-0 and the starchless pgm mutant Col-0 was grown for 21 d in an 8-h photoperiod. pgm was sown out 7 d earlier than Col-0 and initially grown in a 12-h photoperiod to allow the acquisition of biomass, before transfer to an 8-h photoperiod 3 d before the start of the pulse. Both genotypes were pulsed with 13CO2 for 24 h, followed by a 4-d chase. Plants were harvested at dawn before the start of the pulse, at dusk, 4 h into the night, at dawn at the end of the pulse, and at dawn after a 4-d chase. The experiment was repeated three times with separately grown batches of plants; in each experiment, one sample of five plants was harvested at each time point. Rates of synthesis were estimated for two time intervals in the night: between dusk and 4 h of darkness (T8–T12) and between 4 h of darkness and the end of the night (T12–T24). The data are provided in Supplemental Data Set S5, calculated enrichments in Supplemental Table S4, and plots of changes in the enrichment of selected amino acids in Figure 4. The rate of protein synthesis was calculated by correcting enrichment in Ala in protein by the enrichment in free Ala at dusk. Results are estimated from the average of three biological replicates. Plant . RGR . Protein Synthesis . Protein Degradation . Half-Life . Average per 24-h Cycle . Light (T0–T8) per Hour . Dark (T8–T12) per Hour . Dark (T12–T24) per Hour . Pulse . Chase . Pulse . Chase . mg mg−1 d−1 % total protein d−1 d Col-0 0.221 25.29 2.11 0.68 0.7 3.18 4.84 3.38 2.48 pgm 0.137 18.02 1.81 0.54 0.29 4.33 6.09 2.67 2.17 Ratio pgm:Col-0 0.62 0.71 0.86 0.8 0.42 1.36 1.26 Plant . RGR . Protein Synthesis . Protein Degradation . Half-Life . Average per 24-h Cycle . Light (T0–T8) per Hour . Dark (T8–T12) per Hour . Dark (T12–T24) per Hour . Pulse . Chase . Pulse . Chase . mg mg−1 d−1 % total protein d−1 d Col-0 0.221 25.29 2.11 0.68 0.7 3.18 4.84 3.38 2.48 pgm 0.137 18.02 1.81 0.54 0.29 4.33 6.09 2.67 2.17 Ratio pgm:Col-0 0.62 0.71 0.86 0.8 0.42 1.36 1.26 Open in new tab Table IV. Comparison of the rates of protein synthesis and degradation in Col-0 and the starchless pgm mutant Col-0 was grown for 21 d in an 8-h photoperiod. pgm was sown out 7 d earlier than Col-0 and initially grown in a 12-h photoperiod to allow the acquisition of biomass, before transfer to an 8-h photoperiod 3 d before the start of the pulse. Both genotypes were pulsed with 13CO2 for 24 h, followed by a 4-d chase. Plants were harvested at dawn before the start of the pulse, at dusk, 4 h into the night, at dawn at the end of the pulse, and at dawn after a 4-d chase. The experiment was repeated three times with separately grown batches of plants; in each experiment, one sample of five plants was harvested at each time point. Rates of synthesis were estimated for two time intervals in the night: between dusk and 4 h of darkness (T8–T12) and between 4 h of darkness and the end of the night (T12–T24). The data are provided in Supplemental Data Set S5, calculated enrichments in Supplemental Table S4, and plots of changes in the enrichment of selected amino acids in Figure 4. The rate of protein synthesis was calculated by correcting enrichment in Ala in protein by the enrichment in free Ala at dusk. Results are estimated from the average of three biological replicates. Plant . RGR . Protein Synthesis . Protein Degradation . Half-Life . Average per 24-h Cycle . Light (T0–T8) per Hour . Dark (T8–T12) per Hour . Dark (T12–T24) per Hour . Pulse . Chase . Pulse . Chase . mg mg−1 d−1 % total protein d−1 d Col-0 0.221 25.29 2.11 0.68 0.7 3.18 4.84 3.38 2.48 pgm 0.137 18.02 1.81 0.54 0.29 4.33 6.09 2.67 2.17 Ratio pgm:Col-0 0.62 0.71 0.86 0.8 0.42 1.36 1.26 Plant . RGR . Protein Synthesis . Protein Degradation . Half-Life . Average per 24-h Cycle . Light (T0–T8) per Hour . Dark (T8–T12) per Hour . Dark (T12–T24) per Hour . Pulse . Chase . Pulse . Chase . mg mg−1 d−1 % total protein d−1 d Col-0 0.221 25.29 2.11 0.68 0.7 3.18 4.84 3.38 2.48 pgm 0.137 18.02 1.81 0.54 0.29 4.33 6.09 2.67 2.17 Ratio pgm:Col-0 0.62 0.71 0.86 0.8 0.42 1.36 1.26 Open in new tab In summary, pgm has a 29% lower rate of protein synthesis per 24 h than Col-0, due to a lower rate of protein synthesis in the second part of the night. The rate of protein degradation over a 24-h cycle is higher, especially when related to the rate of protein synthesis; daily average protein degradation is equivalent to 24% to 34% of daily average protein synthesis in pgm, compared with 13% to 19% in Col-0. The lower rate of synthesis and higher rate of degradation result in an almost 2-fold decrease in the net gain of protein per day (11.9%–13.7% d−1 in pgm compared with 20.5%–22.1% d−1 in Col-0). These estimated net rates of protein synthesis are very similar to RGR estimated from changes in plant size (0.221 and 0.137 mg mg−1 d−1, respectively, in Col-0 and pgm). Estimation of Growth by Measuring Enrichment in Glc in the Cell Wall In a last set of experiments, we investigated whether 13CO2 labeling can be used to estimate the rate of cell wall synthesis. We did this for two reasons: first, to compare the rates of cell wall synthesis with those of protein synthesis; and second, as an alternative way of estimating RGR. Up to 55% of Arabidopsis rosette dry weight is cell wall (Williams et al., 2010), of which about 20% to 30% is cellulose, with further contributions from hemicellulose, pectin, and proteins (Somerville et al., 2004; Lodish et al., 2011; Somerville et al., 2004). Cellulose and the backbone of hemicellulose consist of β(1-4)-linked d-Glc. Assuming that there is little or no degradation of these polysaccharides, enrichment in Glc in the cell wall should be a reasonable proxy for the rate of growth. Using material from the experiment of Table IV, cell walls were isolated, thoroughly digested to remove starch, chemically hydrolyzed, and analyzed by GC-TOF-MS to determine the kinetics of enrichment in Glc in the cell wall (Tables V and VI). In this case, we did not apply a correction for incomplete labeling of precursor. Hexose phosphates and UDP-Glc show quite complex labeling kinetics due to the presence of compartmented pools (Szecowka et al., 2013). Suc-6-P, which lies downstream of these metabolites, is rapidly labeled to more than 70% enrichment, giving a minimum value for enrichment in the hexose phosphate and UDP-Glc precursor pools (Szecowka et al., 2013). Rate of cell wall synthesis in Col-0 and the starchless pgm mutant Table V. Rate of cell wall synthesis in Col-0 and the starchless pgm mutant Changes in enrichment in Glc in the cell wall fraction in different time intervals during a day-night cycle and during the entire 24-h 13CO2 pulse are shown. The data are for the same experiment as that of Table IV. The original data are provided in Supplemental Data Set S5. Time . Increase in Enrichment per Interval . Increase in Enrichment per Hour . Ratio Col-0:pgm . Col-0 . pgm . Col-0 . pgm . % T0 to T8 (light) 12.15 8.06 1.52 1.01 1.51 T8 to T12 (start of night) 1.79 5.12 0.45 1.28 0.35 T12 to T24 (later in night) 5.29 0.29 0.44 0.02 18.54 T0 to T24 19.23 13.47 Time . Increase in Enrichment per Interval . Increase in Enrichment per Hour . Ratio Col-0:pgm . Col-0 . pgm . Col-0 . pgm . % T0 to T8 (light) 12.15 8.06 1.52 1.01 1.51 T8 to T12 (start of night) 1.79 5.12 0.45 1.28 0.35 T12 to T24 (later in night) 5.29 0.29 0.44 0.02 18.54 T0 to T24 19.23 13.47 Open in new tab Table V. Rate of cell wall synthesis in Col-0 and the starchless pgm mutant Changes in enrichment in Glc in the cell wall fraction in different time intervals during a day-night cycle and during the entire 24-h 13CO2 pulse are shown. The data are for the same experiment as that of Table IV. The original data are provided in Supplemental Data Set S5. Time . Increase in Enrichment per Interval . Increase in Enrichment per Hour . Ratio Col-0:pgm . Col-0 . pgm . Col-0 . pgm . % T0 to T8 (light) 12.15 8.06 1.52 1.01 1.51 T8 to T12 (start of night) 1.79 5.12 0.45 1.28 0.35 T12 to T24 (later in night) 5.29 0.29 0.44 0.02 18.54 T0 to T24 19.23 13.47 Time . Increase in Enrichment per Interval . Increase in Enrichment per Hour . Ratio Col-0:pgm . Col-0 . pgm . Col-0 . pgm . % T0 to T8 (light) 12.15 8.06 1.52 1.01 1.51 T8 to T12 (start of night) 1.79 5.12 0.45 1.28 0.35 T12 to T24 (later in night) 5.29 0.29 0.44 0.02 18.54 T0 to T24 19.23 13.47 Open in new tab RGR and RGRSTR in Col-0 and the starchless pgm mutant Table VI. RGR and RGRSTR in Col-0 and the starchless pgm mutant Comparison of RGRSTR calculated from 13C enrichment in the chase and RGR calculated from fresh weight and rosette area is shown. The data are for the same experiment as that of Table IV. The original data are provided in Supplemental Data Set S5. Plant . RGRSTR . RGR . STR d−1 mg mg−1 d−1 Col-0 0.205 0.221 pgm 0.127 0.137 Plant . RGRSTR . RGR . STR d−1 mg mg−1 d−1 Col-0 0.205 0.221 pgm 0.127 0.137 Open in new tab Table VI. RGR and RGRSTR in Col-0 and the starchless pgm mutant Comparison of RGRSTR calculated from 13C enrichment in the chase and RGR calculated from fresh weight and rosette area is shown. The data are for the same experiment as that of Table IV. The original data are provided in Supplemental Data Set S5. Plant . RGRSTR . RGR . STR d−1 mg mg−1 d−1 Col-0 0.205 0.221 pgm 0.127 0.137 Plant . RGRSTR . RGR . STR d−1 mg mg−1 d−1 Col-0 0.205 0.221 pgm 0.127 0.137 Open in new tab In the light period, the rate of label incorporation into Glc in the cell wall is about 50% faster in Col-0 than in pgm (1.52% and 1.01% h−1, respectively). During the first 4 h of darkness, growth decreased in Col-0 but remained high in pgm (0.45% and 1.3% h−1, respectively). It should be noted that there were relatively small changes in enrichment in this short time interval, so the estimated rates are less reliable than for other time intervals. In the remainder of the night, Glc incorporation continued at the same rate in Col-0 but stopped in pgm (0.44% and 0.02% h−1, respectively). In wild-type Col-0, the decrease in the rate of cell wall synthesis in the night compared with the light period (about 3-fold) is similar to the decrease in the rate of protein synthesis. In contrast, in pgm, cell wall synthesis is almost completely inhibited in the last part of the night, whereas (see above) protein synthesis continues at a low rate. We used the labeling kinetics of Glc in the cell wall to estimate the relative rate of accumulation of structural dry matter (RGRSTR), assuming that there is no degradation of Glc in cell wall polysaccharides (Tables V and VI; Supplemental Data Set S5). Summing the rates of Glc incorporation estimated from the pulse data gave estimates of RGRSTR of 0.192 and 0.135 in Col-0 and pgm, respectively (Tables V and VI). The labeling kinetics in the chase allowed an independent estimate of RGRSTR as: where x and y are enrichment in Glc in cell wall at the end of the pulse and the end of the chase, respectively, and t is the duration of the chase in days. This approach gave estimates of RGRSTR of 0.205 in Col-0 and 0.127 in pgm (Tables V and VI). Both approaches yield estimates of RGRSTR that are slightly lower than RGR obtained by an analysis of plant size (0.221 and 0.137 mg mg−1 d−1 in Col-0 and pgm, respectively). The slightly lower RGRSTR obtained by an analysis of enrichment in cell wall Glc compared with RGR estimated from changes in plant size might be due to enrichment in metabolites in central metabolism saturating at about 90% (Szecowka et al., 2013). Despite this slight discrepancy, RGRSTR and RGR showed similar decreases (both about 38%) in pgm compared with Col-0. DISCUSSION A Simple Protocol to Measure the Global Rate of Protein Synthesis and Degradation in Whole Plants Growing in a Light/Dark Cycle We have established a simple method to quantify the global rate of protein synthesis and protein degradation. By providing 13CO2 in the atmosphere in the light, we rapidly introduce label in the pulse and remove it during a chase without perturbing metabolism and growth. Labeled C that accumulates in the light period in transitory C reserves like starch provides an internal source of label during the night. A key feature is our use of parallel information about enrichment in free amino acids and enrichment in amino acid residues in protein to calculate the absolute rate of protein synthesis. In principle, data for each amino acid could be used separately to provide many parallel estimates of the rate of protein synthesis. However, most amino acids show complex labeling kinetics, with a slow and incomplete increase in enrichment in the light, a partial decrease in enrichment at night during the pulse, and a slow decline in enrichment during the chase (Fig. 1). While the reasons probably vary from case to case, some general features can be distinguished. The most rapidly and completely labeled amino acids were Ala, Asp, Gly, and Ser. Ala and Asp are formed via reversible reactions from metabolites in central metabolism, and Gly and Ser are intermediates in the photorespiration pathway. Most other amino acids are synthesized via long dedicated pathways, which may contribute to their slow labeling kinetics. The aromatic amino acids showed quite rapid labeling kinetics. This may be explained by two factors: first, the immediate precursors for the shikimate pathway are the Calvin-Benson cycle intermediate erythrose 4-phosphate and pyruvate, and both show rapid increases in enrichment during a pulse (Szecowka et al., 2013); and, second, flux through the shikimate pathway will be higher than in other amino acid biosynthesis pathways, because the shikimate pathway also provides precursors for phenylpropanoid metabolism. Their labeling was nevertheless slower than for Ala, Asp, Gly, and Ser. Labeling kinetics are also complicated by the presence of multiple pools with different labeling kinetics. Many amino acids are located in both the cytoplasm and the vacuole (Riens et al., 1991; Weiner and Heldt, 1992; Winter et al., 1992; Krueger et al., 2011; Arrivault et al., 2014), and it is likely that the cytoplasmic pool is more rapidly labeled than the vacuolar pool (Szecowka et al., 2013). There may also be pools in nonphotosynthetic cells that are not rapidly labeled with 13CO2. As a consequence, for many amino acids, the average enrichment may underestimate the enrichment in the precursor pool for protein synthesis. Compartmentation may also contribute to the general decrease in average enrichment of free amino acids at night. Amino acid levels increase in the light and decrease at night, when they are used for protein synthesis (Gibon et al., 2009; Piques et al., 2009; Pal et al., 2013). The pools that are formed from newly fixed 13CO2 in light are probably preferentially utilized at night, resulting in an increase in the contribution of unlabeled or weakly labeled pools at dawn. This is probably the reason for the large decrease in the enrichment of Gly at dawn. Gly undergoes especially large diurnal changes, due to its accumulation during photorespiration. The pool that is involved in photorespiration will be highly enriched (Szecowka et al., 2013). The large decrease in enrichment at dawn would be explained if there is another small and weakly labeled pool of Gly, which is compartmented from the pool that is involved in photorespiration. This weakly labeled pool would make little contribution to total Gly at dusk but a large contribution at dawn. Incomplete labeling of free amino acid pools could also result from protein degradation, which will recycle unlabeled amino acids. However, turnover is generally quite low, so this is probably only a major factor in special cases, such as the pgm mutant at night. In some cases, the slow labeling kinetic reflects specific features of metabolic pathways. One example is the unexpectedly slow labeling of Glu and Gln in the light (Fig. 1). This resembles an earlier observation that enrichment in Glu remains high during the first light period of a chase when Arabidopsis plants are labeled for several days with 13CO2 and then chased with CO2 (Huege et al., 2007). It has been known for some time that citrate levels fall in the light and recover at night (Scheible et al., 2000; Urbanczyk-Wochniak et al., 2005). It was recently reported that 2-oxoglutarate is labeled only slowly in the light (Szecowka et al., 2013) and that most of the Glu formed in the light derives from a preexisting and thus unlabeled pool of citrate (Tcherkez et al., 2012a, 2012b). Our data confirm that there is strong restriction on the flux of newly fixed 13C to citrate, Glu, and Gln in the light and show that this restriction is relieved at night (Supplemental Fig. S1). This is presumably due either to inhibition in the light of pyruvate dehydrogenase activity (Tovar-Méndez et al., 2003) or of succinate dehydrogenase and fumarase by the mitochondrial thioredoxin system (Daloso et al., 2015). Comparison of the labeling kinetics after supplying 13CO2 to soil-grown plants and after supplying 2H2O to seedlings (Yang et al., 2010) reveals a qualitatively similar response for most amino acids. In both studies, Ala, Asp, Ser, and Gly labeled rapidly and reached high enrichment, while most minor amino acids labeled slowly and incompletely or showed biphasic kinetics, pointing to the presence of multiple pools. Ala and Asp labeled from 2H2O with half-times on the order of 0.5 to 0.7 h. In our study, half-times cannot be estimated, because the first sampling time point was at 4 h. However, by this time, maximum enrichment had been reached, indicating that the rate of labeling is not much slower than with 2H2O. This conclusion is supported by the rapid labeling of Ala reported by Szecowka et al. (2013). However, there were some differences between the labeling kinetics in our study and that of Yang et al. (2010). First, Ala and Asp labeled even more completely from 2H2O in seedlings than from 13CO2 in soil-grown plants. Second, relative to other amino acids, Gly and Ser labeled more slowly from 2H2O in seedlings (compared with Ala and Asp, the half-times of Gly and Ser were about 5 and 3 times slower, respectively) than from 13CO2 (labeling was not noticeably slower than for Ala and Asp; Szecowka et al., 2013). Third, Glu labeled much more rapidly from 2H2O in seedlings than from 13CO2 in soil-grown plants. These differences may be due in part to different pool sizes and compartmentation in seedlings and older plants. They may also reflect a lower contribution of photosynthesis in the seedling system. A significant part of the seedling is nonphotosynthetic tissue, and their leaves were submerged, which is likely to reduce CO2 access and the rate of photosynthesis. This might explain why, despite the 2H2O labeling of seedlings being performed in the light, labeling of Gly and Ser was relatively slow and labeling of Glu was much faster than in our experiments. It is also noteworthy that Yang et al. (2010) showed that the labeling kinetics of Ala and Asp are more than 50-fold slower from 15NO3 than from 2H2O. Overall, the study of Yang et al. (2010) and our study show that most individual amino acids have complex and incomplete labeling patterns and underline the importance of taking this into account in any study of the rate of protein synthesis. The reason for the complex labeling pattern of free amino acids will affect how this information should be used to calculate the rate of protein synthesis. For example, if low enrichment is due to the presence of multiple pools, use of the average value will result in overestimation of the rate of protein synthesis, because average enrichment will underestimate enrichment in the precursor pool. For this reason, we decided to use Ala, which showed the most rapid and complete labeling kinetics. The rate of protein synthesis will be slightly overestimated if the incomplete labeling of Ala is due to a separate unlabeled pool that is not involved in protein synthesis. However, analysis of the isotopomer labeling kinetics did not reveal any evidence for a small unlabeled pool of Ala. Any error introduced by the correction will be small anyway, because the overall enrichment in free Ala is high (typically, 80% in the light and 70% in the dark). Furthermore, our published analyses indicate that enrichment of Calvin-Benson cycle intermediates does not rise above 90%, presumably due to continual recycling of unlabeled C (Szecowka et al., 2013). In tissues or treatments where no amino acid shows a suitable enrichment pattern, it will be necessary to analyze the reasons for the incomplete enrichment in more detail to decide how to apply the correction. Other small errors can arise in our simple protocol. One is that we did not measure the precise rate at which 13C enrichment increased in Ala at the start of the pulse. Data from Szecowka et al. (2013) indicate that this occurs within 1 h even when labeling commences in the middle of the light period in the presence of a large unlabeled pool of Ala. Therefore, it is possible that our protocol results in a small but systematic underestimation of the rate of protein synthesis in the light. Another potential source of error is that some protein synthesis may occur in cell types that are not being labeled in the time frame of the 13CO2 pulse. The observation that Ala showed similar enrichment in very young leaves and in mature leaves (Fig. 2C) argues against this being a general problem, but we cannot exclude the possibility that some specific cell types are missing from our analysis. Our calculation will also underestimate the rate of protein synthesis in conditions where rapid protein degradation recycles substantial amounts of label out of protein; however, in our experiments, this error is negligible, because the protein pool was only labeled up to about 25% and the rate of protein degradation was low in almost all investigated conditions. It was necessary to perform pulse and chase experiments for a fairly long time, due to the relatively low rate of protein synthesis. Long pulses are also needed to analyze protein synthesis throughout the light/dark cycle. During this time, metabolism is moving through a series of quasisteady states. Our approach, which involves comparison of a single precursor pool and a single product, can be applied to nonsteady-state conditions. While there are more sophisticated computational methods that analyze dynamic labeling patterns in large metabolic networks, they can only be applied when the metabolism is in steady state (Antoniewicz et al., 2006; Yuan et al., 2006, 2008; Young et al., 2011) and, in higher plants, also require assumptions about the spatial compartmentation (Szecowka et al., 2013; Heise et al., 2014; Ma et al., 2014). Although recent theoretical developments allow the analysis of dynamic labeling in metabolic nonsteady state in microbes (Antoniewicz, 2013), it remains challenging to apply these methods in higher plants. While small errors will not greatly affect estimates of protein synthesis rates, they will affect the estimates of protein degradation rates. The latter are estimated from the difference between the rate of protein synthesis and the rate of growth. The rate of protein synthesis is often only slightly higher than the rate of growth, making our estimates of the rate of protein degradation sensitive to experimental and computational error. Measurement of Plant Growth Rate Estimation of the rate of protein degradation requires information about the growth rate. RGR was estimated by destructive harvesting of batches of plants during the pulse and chase. These values were confirmed by repeating these measurements in biologically replicated experiments with separately grown batches of plants in the same growth conditions (Supplemental Fig. S5C) and by time-series imaging of the leaf area of individual plants during the labeling experiments. However, changes in plant size may not always reflect changes in biomass, defined as structural dry mass. They will differ, for example, when vacuole expansion is leading to a change in plant size that is not tightly coupled to changes in structural biomass. Therefore, we explored whether the 13CO2 labeling could be used to monitor the enrichment of Glc in cell wall material and, hence, the rate of cell wall synthesis (Tables V and VI; Supplemental Data S5). This approach gave estimates of growth rate that agreed well with those obtained by measurements of plant size. It should be noted that this conclusion holds for an entire 24-h cycle but does not exclude the possibility that there may be temporal uncoupling of cell wall synthesis and increases in plant size at some times during the 24-h cycle (see below). A related problem will arise if expansion growth or other factors lead to a change in protein content during the experiment. In such cases, it is important to measure protein content and integrate this information into the calculation of protein degradation, as illustrated by the inclusion of information about changes in protein content in our analysis of the flux to growth and the rate of protein degradation in young leaves (Fig. 3) Rates of Protein Synthesis and Degradation during a Diurnal Cycle Our analysis reveals, for Arabidopsis Col-0 growing in short photoperiods, that the rate of protein synthesis in the light is about 3-fold higher than in the dark. This is a much larger decrease than for polysome loading, which only decreases by 25% to 30% (Pal et al., 2013; Sulpice et al., 2014). This comparison emphasizes that polysome loading only provides a qualitative proxy for the rate of protein synthesis. Several factors may lead to polysome loading underestimating the actual changes in the rate of protein synthesis, including changes in the rate of progression of ribosomes, ribosome stalling, or the presence of RNA in other high-mass structures. As the plants were growing in an 8-h photoperiod, the amount of protein synthesized at night was about 40% of the total protein synthesis in a 24-h cycle. Cell wall synthesis also continues in the night at about one-third of the rate in the light, which is equivalent to about 40% of the cell wall synthesis occurring at night. These results underline the importance of nocturnal metabolism for growth. In an 8-h photoperiod, Col-0 accumulates about 52% of the daily fixed C as starch and other metabolites to support respiration and growth at night (Sulpice et al., 2014), of which about 10% is respired, leaving about 40% for the synthesis of structural biomass. The rate of growth was about 22% d−1 in Col-0 in our growth conditions. Most of the synthesized protein, therefore, represents flux to growth. The estimated rate of protein degradation was 3.1% to 3.5% d−1, which is much lower than the flux to growth. A combination of dilution by growth and protein degradation results in a global protein half-life of 3.1 to 3.5 d. This is similar to that predicted for Arabidopsis rosettes using ribosome abundance and polysome loading to model the global rate of protein synthesis (Piques et al., 2009). Thus, our results validate this simple linear model, which used literature values for the rate of ribosome progression to link quantitative molecular data to whole-plant growth. Our estimates also lie in a similar range to those obtained in experimental studies of small sets of mitochondrial proteins in Arabidopsis cell cultures (Nelson et al., 2013) and Arabidopsis rosettes (Li et al., 2012) and a set of 508 proteins in barley leaves (Nelson et al., 2014b). Inappropriate Regulation of C Allocation Has Major Consequences for Protein Synthesis and Degradation Starch turnover buffers plants against the daily alternation of light and dark (Smith and Stitt, 2007; Stitt and Zeeman, 2012). We used the starchless pgm mutant to investigate the importance of starch for the rate and timing of protein synthesis and degradation. Protein synthesis in pgm occurred at a similar rate to wild-type plants in the light but was strongly inhibited in the dark. Nonetheless, even though pgm is severely C starved at night (Gibon et al., 2004b, 2006), some protein synthesis continued; the rate of protein synthesis in pgm in the last 12 h of the night was 40% of that in Col-0 in the night and 14% of that in Col-0 and pgm in the light. C starvation leads to strong changes in transcription, including the induction of many genes (Price et al., 2004; Bläsing et al., 2005; Usadel et al., 2008). Some of the residual protein synthesis in the night in pgm might be related to the synthesis of proteins that are required for C starvation responses. In agreement, C starvation-related enzymes like Glu dehydrogenase show an increase in abundance in pgm (Gibon et al., 2004a). The composition of ribosomes changes in response to C starvation (Hummel et al., 2012), although it is not yet known if this occurs rapidly enough to contribute to translational responses during the night in the pgm mutant. The inhibition of cell wall synthesis at night in pgm was stronger than the inhibition of protein synthesis. This is consistent with the idea that the residual protein synthesis may be related to adjustment to low C, rather than growth. The partial inhibition of cell wall synthesis in the night in wild-type Col-0 and the complete inhibition in pgm underlines that the synthesis of cellulose, and probably other cell wall components, is tightly regulated by C. Whereas amino acid levels decline during the night in wild-type Arabidopsis (Gibon et al., 2009; Pal et al., 2013; Sulpice et al., 2014), they increase in the pgm mutant (Gibon et al., 2006; Izumi et al., 2013), indicating that protein degradation is increased to provide amino acids as an alternative substrate for respiration. Our measurements provide direct evidence that protein degradation is increased at night in the pgm mutant. First, calculated rates of protein degradation are almost 2-fold higher in pgm than in wild-type plants. Second, there is a dramatic decrease in enrichment in almost all free amino acids at night in pgm, as expected if increased protein degradation is leading to rapid recycling of unlabeled amino acids. The only exception was Asn. C starvation leads to a strong increase in ASN1 transcript abundance (Lam et al., 1996, 1998; Bläsing et al., 2005; Usadel et al., 2008). The level of Asn typically increases strongly in C-starved plants, providing a store for amino acid residues that are recycled from respired amino acids (Brouquisse et al., 1991; Lam et al., 1996; Gibon et al., 2006). The increase in Asn enrichment during the night in pgm, at a time when enrichment in all other amino acids is falling, provides evidence for the synthesis of Asn from prelabeled C pools, which may include Asp and organic acids like malate and fumarate. The pgm mutant has an approximately 40% smaller daily net increment in protein than wild-type Col-0 (11.9%–13.7% d−1 in pgm compared with 20.5%–22.1% d−1 in Col-0). This closely matches the observed decrease in growth (38%) estimated either from size measurements or from Glc incorporation into the cell wall. This decrease in net protein synthesis is the result of a 29% reduction in the amount of protein synthesized per 24-h cycle, due to the lower rate of protein synthesis in the night, and a 26% to 35% increase in the rate of protein degradation (Table III). The decrease in growth-related protein synthesis may be even larger, because some of the protein synthesis at night in pgm may serve to produce proteins that are involved in C starvation responses (see above). Furthermore, this daily alternation between net protein synthesis and degradation results in a time offset and wasteful futile cycle in pgm, in which energy is used to synthesize proteins in the light period that are degraded in the dark. The low rate of protein synthesis during the night and the daily cycle of protein synthesis and degradation will also result in less efficient use of the translational apparatus in pgm than in wild-type Col-0. Many explanations have been advanced for the poor growth of pgm in short photoperiods, ranging from loss of C due to rapid respiration as sugars decline from high to low levels at the start of the night (Caspar et al., 1985) to disturbances in root metabolism (Brauner et al., 2014). However, they do not provide a quantitative explanation for the growth inhibition; indeed, the increase in respiration accounts for only a small proportion of the assimilated C. Furthermore, some of these effects may be secondary; for example, the high rate of respiration in the first part of the night can be at least partly explained as a consequence of the continuation of protein and cell wall synthesis at this time in the pgm mutant, rather than as wasteful stimulation of respiration by high sugar. Changes in Temperature Do Not Have a Major Impact on Protein Degradation High temperature results in an increase in respiration, in particular maintenance respiration, (Penning de Vries et al., 1979; Amthor, 2000). In Arabidopsis, maintenance respiration doubles between growth temperatures of 16°C and 24°C (Pyl et al., 2012). The major maintenance costs are thought to be protein turnover and the maintenance of gradients across membranes (Penning de Vries, 1975; Amthor, 2000). Our results show that increasing the growth temperature from 20°C to 28°C leads to a parallel 30% increase in both the rate of protein synthesis and the rate of protein degradation. This indicates that, although an increase in protein turnover could contribute to the rise in maintenance respiration at high temperature, it is unlikely to be the only factor. Two other recent studies also indicate that protein turnover is not the main factor underlying the increase in maintenance respiration at high temperature. Pilkington et al. (2015) estimated that the ATP provided by maintenance respiration in Arabidopsis growing in a short photoperiod would support the synthesis and degradation of 4% of the total protein per h, which is 10-fold higher than the measured rate of protein degradation. Cheung et al. (2013) used metabolic flux balance analysis to estimate the rate of consumption of ATP and NADPH during maintenance in Arabidopsis cell cultures and concluded that up to one-third of the output of maintenance respiration was NADPH, which is presumably used for processes other than protein synthesis. Darkness Does Not Lead to a Preferential Inhibition of all Protein Synthesis in the Plastid Although our protocol was designed to analyze global protein synthesis, we were interested to learn if it could be modified to investigate individual proteins. Using a procedure adapted from Allen et al. (2012), we found that synthesis of the plastid-encoded RBCL continues in the dark, being inhibited to approximately the same extent as the synthesis of the nucleus-encoded RBCS and the overall rate of protein synthesis. This was initially surprising, as it is thought that plastid translation is inhibited in the dark (Marín-Navarro et al., 2007). This idea is based on studies of protein synthesis in isolated chloroplasts and on studies of plastid-encoded proteins that are targeted to the thylakoids. However, isolated chloroplasts may not be a perfect model for studying plastid protein synthesis in whole leaves. It is also possible that the synthesis of a soluble protein like RBCL may be differently regulated from proteins that are assembled into protein complexes in the thylakoid membrane and whose targeting and assembly may be light dependent (Keegstra and Cline, 1999; Cline and Dabney-Smith, 2008). Furthermore, as Rubisco represents up to 40% of total leaf protein (Eckardt et al., 1997), restricting its synthesis to the light period would severely restrict the availability of ribosomes for the synthesis of other protein species. The question arises of whether our approach can be adapted to analyze the synthesis and degradation of less abundant proteins. One possibility might be to combine it with the approach used by Yang et al. (2010), who overexpressed tagged variants of selected proteins to aid their purification. This approach will probably be limited by the sensitivity of the detection of amino acids in gas chromatography-mass spectrometry. Alternatively, 13C-labeled protein might be converted to peptides for analysis, as was done by Yang et al. (2010). In this case, the major challenges will be deconvolution of the complex envelope and correction of the enrichment for all amino acids in the peptide, including many with very incomplete enrichment in the precursor free amino acid pool. These challenges are shared by strategies using 2H2O labeling. While the peptide envelope is simpler when 15N is used as a label (Li et al., 2012; Nelson et al., 2014a, 2014b), the slow labeling kinetic of amino acids in higher plants (Yang et al., 2010) means that this approach is probably unsuited for rapidly turning over proteins. The decision between labeling with 13CO2 and 2H2O may depend on the tissue and condition under investigation, as this will affect which isotope can be introduced with the least disturbance, the extent to which inhibitory effects of 2H2O can be minimized, and the speed and completeness of the labeling kinetics of free amino acid pools. In conclusion, we have developed a relatively simple method that introduces label via photosynthetic CO2 fixation and provides quantitative information on the global rate of protein synthesis and degradation. The key feature is parallel determination of isotope enrichment in the products and in the metabolite precursors from which they are synthesized. We have applied this approach to investigate the relationship between protein metabolism, the plant energy budget and growth, and the impact of development and environmental perturbations on protein synthesis and degradation. While it can also be used to study individual proteins, this application will probably be restricted to abundant proteins. We have shown that the approach can be extended to provide quantitative information about the rate of cell wall synthesis and expect that it will also be applicable to other classes of structural components as well as specialized metabolites. This would allow a comprehensive analysis of the rates of synthesis of all major structural components in the plant as well as major defense metabolites. An equally important challenge will be to adapt this approach to introduce 13CO2 via photosynthesis, allow label to move through the plant via endogenous transport routes, and analyze fluxes to growth in nonphotosynthetic organs such as roots and seeds. MATERIALS AND METHODS Plant Growth Conditions Arabidopsis (Arabidopsis thaliana) Col-0 seeds were germinated and grown in soil for 1 week with 16 h of light (20°C in the light and 6°C at night, 150 µmol m−2 s−1 fluorescent light, and 60%–70% relative humidity) before transfer to short days (8 h of light, 150 µmol m−2 s−1, 20°C/19°C in the light and dark, and 60%–70% relative humidity). Fourteen days after sowing, five seedlings were pricked into 10-cm pots and grown under the same conditions in a controlled-environment chamber (model E-36L; Percival Scientific; Thimm et al., 2004). In some experiments, Col-0 plants were transferred to short days with 28°C/28°C in the light and dark (8 h of light, 150 µmol m−2 s−1, and 60%–70% relative humidity) at 18 d after sowing and used for the labeling experiment at 21 d after sowing. The pgm mutant was germinated as above and grown in a 12-h photoperiod (20°C/19°C in the light and dark, 150 µmol m−2 s−1 fluorescent light, and 60%–70% relative humidity) until 25 d after sowing and then shifted to an 8-h photoperiod 3 d prior to the experiment at 28 d after sowing. 13CO2 Feeding Experiments 13CO2 feeding was carried in a Plexiglas box (internal dimensions, 60 × 31 × 17.4 cm, holding up to 17 10-cm pots, each containing five plants) in a Percival controlled-environment chamber. Three-week-old plants were transferred to the chamber 3 d before the experiment. The labeling chamber was supplied with a premixed air stream containing 450 µL L−1 CO2 or 13CO2, 21% (v/v) oxygen, and 79% (v/v) nitrogen. At a flow rate of 5 L min−1, the air in the chamber was completely replaced in 20 min. During 13CO2 feeding, the air exiting the box was passed through a gas-wash bottle filled with soda lime pellets with indicator (Merck Millipore) to prevent the release of 13CO2 into the growth chamber. Unless specified, 13CO2 feeding was performed in growth conditions with photon flux density and temperature inside the box of approximately 150 µmol m−2 s−1 and 20°C/18°C in light and dark, respectively. Labeling started 1 h before dawn and was continued through an entire light and dark period. Plants were harvested by opening the lid of the labeling chamber, removing the plants and submerging them in liquid N2, and reclosing the lid within 10 s. Immediately following harvest, leaf tissues were frozen in liquid nitrogen. Samples were ground to a fine powder at −70°C using a cryogenic grinding robot (http://www.labman.co.uk/portfolio-type/cryogenic-plant-grinder-and-dispensing-system/; Labman Automation) and stored until analysis at −80°C. Metabolite and Protein Analysis Homogenized frozen plant material (30 mg) was extracted with methanol followed by phase separation using chloroform-water as described by Heise et al. (2014). Ice-cold 100% (v/v) methanol and ribitol (1.31 mm) solution as an internal quantitative standard were added to the homogenized plant material and incubated for 10 min at 70°C. After the incubation, the samples were centrifuged. The pellet was kept for protein and cell wall analysis. The supernatant was used for phase separation using chloroform-water. The apolar (chloroform) phase was discarded. The upper polar phase was used for methoxysilylation of metabolites with N-methyl-N-trimethylsilyltrifluoroacetamide followed by GC-TOF-MS analysis using the same conditions and settings as described by Lisec et al. (2006). Total protein was extracted from the pellet using 6 m urea/2 m thiourea solution. Total protein was measured as described by Bradford (1976). Protein (about 50 µg) was precipitated with 15% (v/v) ice-cold TCA and washed with ice-cold 100% (v/v) acetone. The pellet was chemically hydrolyzed with 6 m hydrochloric acid at 100°C for 24 h at atmospheric pressure (Williams et al., 2010). The hydrolysate was dried and derivatized with N-methyl-N-trimethylsilyltrifluoroacetamide for GC-TOF-MS. The resulting chromatograms were processed as described by Heise et al. (2014). Briefly, chromatography peaks were manually assigned to parent metabolites, and ion intensity was determined using TagFinder (Luedemann et al., 2008) based on mass spectra and retention indices in a reference library derived from the Golm Metabolome Database (http://gmd.mpimp-golm.mpg.de; Kopka et al., 2005). Using the CORRECTOR software tool (Heuge et al., 2014), mass fragment intensities were adjusted to correct for the presence of naturally occurring stable isotopes. Fragmental isotope enrichment was calculated according to Equation 1: (1) where n represents the number of 13C atoms in the detected fragment, and m n is the corrected intensity of a mass fragment that contains n 13C atoms. In addition, relative isotope abundance (RIA) for each metabolite was calculated by following Equation 2: (2) Signal intensity was high enough to quantify the 13C enrichment in Glu, Asp, Ala, Ser, Gly, Ile, Val, Lys, Phe, Tyr, and Pro, but other amino acids could not be analyzed due to low abundance. Cell Wall Analysis The pellet remaining after protein extraction was used for cell wall analysis as described by Saeman et al. (1945) with modifications. After removing the urea/thiourea solution from the pellet by washing five times with distilled water, the pellet was resuspended in 0.3 mL of 0.1 m acetate/NaOH (pH 4.9). Starch in the pellet was removed in two sequential 16-h digestions with 0.2 mL of starch degradation mix (16.8 units mL−1 amyloglucosidase and 30 units mL−1 α-amylase) at 37°C. The remaining pellet was thoroughly washed three times by mixing it with 0.5 mL of distilled water. The sample was then incubated in 72% (v/v) sulfuric acid in 2-mL screw-cap microcentrifuge tubes at room temperature for 1 h, water was added to the sample to adjust sulfuric acid to 1 m, and incubation was continued for 3 h at 100°C to hydrolyze polysaccharides including cellulose. The hydolysate was neutralized with 1 mL of 20% (v/v) N,N-dioctylamine, and excess N,N-dioctylamine was then removed from the aqueous phase by washing three times with 1 mL of 100% (v/v) chloroform. The hydrolysate was dried and stored at −80°C until subsequent derivatization for GC-TOF-MS analysis to quantify 13C enrichment in Glc. Determination of Enrichment in RBCL and RBCS RBCL and RBCS were isolated from total protein and analyzed for 13C enrichment as described by Allen et al. (2012) with modifications. Protein was separated on 15% (v/v) SDS-PAGE Tris-HCl gels (PowerPac Universal Supply; Bio-Rad). The gel was electroblotted to a Bio-Rad immunoblot polyvinylidene difluoride membrane and stained with Ponceaus S solution (Sigma-Aldrich). Using M r markers, RBCL and RBCS bands were located, excised from the membrane, and destained. Protein amino acids were hydrolyzed on the membrane overnight with 6 m hydrochloric acid at 110°C. The hydrolysate was dried and stored at −80°C until subsequent derivatization for GC-TOF-MS analysis. RGR The RGR of rosette biomass (mg mg−1 d−1) was determined by estimating the slope of the natural logarithm-transformed plant weight measured for at least 3 d consecutively during the experiment (Hoffmann and Poorter, 2002). RGR from the rosette leaf area (mm2 mm−2 d−1) was estimated from images acquired using a Canon PowerShot SX500 IS digital camera. Area was extracted from the images using Adobe Photoshop CS5. Estimation of the Change in Protein Content per Day in Young Growing Leaves Calculation of the rate of protein degradation in young growing leaves (Fig. 3) required information about the rate at which the protein content decreased due to the relatively large technical error in measuring protein and the limited material in young leaf samples; this could not be accurately determined in the samples used for 13C enrichment analysis. In parallel experiments, Arabidopsis Col-0 was grown in identical conditions, and after 21 d, sequential leaves were pooled from 60 plants, and leaf area and protein were determined from three biological replicates (Supplemental Table S3, worksheet leaf protein calculation). The rate of leaf initiation (0.41 leaves d−1) was determined in the same material, allowing the data for leaf area and protein content for each leaf to be plotted on a time scale. These plots were inspected to determine the change in protein content per day: ƊP i = (p i − p i+1)/p i, where ƊP i is the decrease in protein per day in leaf i and p i and p i+1 are the leaf protein content in leaf i and the protein content of leaf i the next day, respectively. The protein required per day for growth was calculated as LRGRi × (1 – ƊP i), where LRGRi is the LRGR in leaf i obtained from sequential images of the rosettes from 10 plants. Supplemental Data The following supplemental materials are available. Supplemental Figure S1 . Enrichment kinetics of organic acids in Col-0 growing in an 8-h photoperiod. Supplemental Figure S2 . Labeling kinetics for all detected isotopomers of the free amino acid pools in a pulse and chase in Col-0 growing in an 8-h photoperiod. Supplemental Figure S3 . Comparison of enrichment in the free amino acid pool with the rate of incorporation of label into the corresponding amino acid residue in protein. Supplemental Figure S4 . Comparison of the 13C washout curve for different amino acid residues in protein during the chase. Supplemental Figure S5 . Determination of the RGR of Col-0 growing in an 8-h photoperiod. Supplemental Figure S6 . Kinetics of changes in enrichment in free amino acids during a 24-h pulse in wild-type Col-0 and the starchless pgm mutant growing in an 8-h photoperiod. Supplemental Table S1 . Enrichment in free amino acids, other metabolites, and amino acids in protein in Col-0 growing in an 8-h photoperiod. Supplemental Table S2 . Changes in the levels of amino acids and other metabolites between dawn and dusk in Col-0 growing in an 8-h photoperiod. Supplemental Table S3 . Protein synthesis and degradation rates estimated at different growth stages of leaf development. Supplemental Table S4 . Enrichment in free amino acids, other metabolites, and amino acids in protein in Col-0 and the starchless mutant pgm growing in an 8-h photoperiod. Supplemental Data Set S1 . Col-0 growing in an 8-h photoperiod at 20°C. Supplemental Data Set S2 . Six stages of leaf development in Col-0 growing in an 8-h photoperiod at 20°C. Supplemental Data Set S3 . Col-0 growing in an 8-h photoperiod at 20°C and 28°C. Supplemental Data Set S4 . Enrichment in RBCL and RBCS in the light and dark in Col-0 growing in an 8-h photoperiod at 20°C. Supplemental Data Set S5 . Col-0 and the starchless pgm mutant growing in an 8-h photoperiod at 20°C. ACKNOWLEDGMENTS We thank Christin Abel for help and Stéphanie Arrivault, John E. Lunn, Roosa A.E. Laitinen, and Carlos Figueroa for discussions. Glossary C carbon GC-TOF-MS gas chromatography time-of-flight mass spectrometry Col-0 Columbia-0 RGR relative growth rate RGRSTR relative rate of accumulation of structural dry matter LRGR leaf relative growth rate RBCS small subunit of Rubisco RBCL large subunit of Rubisco LITERATURE CITED Allen DK , Laclair RW, Ohlrogge JB, Shachar-Hill Y ( 2012 ) Isotope labelling of Rubisco subunits provides in vivo information on subcellular biosynthesis and exchange of amino acids between compartments . Plant Cell Environ 35 : 1232 – 1244 Google Scholar Crossref Search ADS PubMed WorldCat Amthor JS ( 2000 ) The McCree-de Wit-Penning de Vries-Thornley respiration paradigms: 30 years later . Ann Bot (Lond) 86 : 1 – 20 Google Scholar Crossref Search ADS WorldCat Antoniewicz MR ( 2013 ) Dynamic metabolic flux analysis: tools for probing transient states of metabolic networks . Curr Opin Biotechnol 24 : 973 – 978 Google Scholar Crossref Search ADS PubMed WorldCat Antoniewicz MR , Kelleher JK, Stephanopoulos G ( 2006 ) Determination of confidence intervals of metabolic fluxes estimated from stable isotope measurements . Metab Eng 8 : 324 – 337 Google Scholar Crossref Search ADS PubMed WorldCat Araújo WL , Tohge T, Ishizaki K, Leaver CJ, Fernie AR ( 2011 ) Protein degradation: an alternative respiratory substrate for stressed plants . Trends Plant Sci 16 : 489 – 498 Google Scholar PubMed OpenURL Placeholder Text WorldCat Arrivault S , Guenther M, Florian A, Encke B, Feil R, Vosloh D, Lunn JE, Sulpice R, Fernie AR, Stitt M, et al. ( 2014 ) Dissecting the subcellular compartmentation of proteins and metabolites in Arabidopsis leaves using non-aqueous fractionation . Mol Cell Proteomics 13 : 2246 – 2259 Google Scholar Crossref Search ADS PubMed WorldCat Bailey-Serres J ( 1999 ) Selective translation of cytoplasmic mRNAs in plants . Trends Plant Sci 4 : 142 – 148 Google Scholar Crossref Search ADS PubMed WorldCat Barneix AJ , Cooper HD, Stulen I, Lambers H ( 1988 ) Metabolism and translocation of nitrogen in two Lolium perenne populations with contrasting rates of mature leaf respiration and yield . Physiol Plant 72 : 631 – 636 Google Scholar Crossref Search ADS WorldCat Beilharz TH , Preiss T ( 2004 ) Translational profiling: the genome-wide measure of the nascent proteome . Brief Funct Genomics Proteomics 3 : 103 – 111 Google Scholar Crossref Search ADS WorldCat Bläsing OE , Gibon Y, Günther M, Höhne M, Morcuende R, Osuna D, Thimm O, Usadel B, Scheible WR, Stitt M ( 2005 ) Sugars and circadian regulation make major contributions to the global regulation of diurnal gene expression in Arabidopsis . Plant Cell 17 : 3257 – 3281 Google Scholar Crossref Search ADS PubMed WorldCat Bouma TJ , De Visser R, Janssen JHJA, De Kock MJ, Van Leeuwen PH, Lambers H ( 1994 ) Respiratory energy requirements and rate of protein turnover in vivo determined by the use of an inhibitor of protein synthesis and a probe to assess its effect . Physiol Plant 92 : 585 – 594 Google Scholar Crossref Search ADS WorldCat Bradford MM ( 1976 ) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding . Anal Biochem 72 : 248 – 254 Google Scholar Crossref Search ADS PubMed WorldCat Branco-Price C , Kaiser KA, Jang CJ, Larive CK, Bailey-Serres J ( 2008 ) Selective mRNA translation coordinates energetic and metabolic adjustments to cellular oxygen deprivation and reoxygenation in Arabidopsis thaliana . Plant J 56 : 743 – 755 Google Scholar Crossref Search ADS PubMed WorldCat Branco-Price C , Kawaguchi R, Ferreira RB, Bailey-Serres J ( 2005 ) Genome-wide analysis of transcript abundance and translation in Arabidopsis seedlings subjected to oxygen deprivation . Ann Bot (Lond) 96 : 647 – 660 Google Scholar Crossref Search ADS WorldCat Brauner K , Hörmiller I, Nägele T, Heyer AG ( 2014 ) Exaggerated root respiration accounts for growth retardation in a starchless mutant of Arabidopsis thaliana . Plant J 79 : 82 – 91 Google Scholar Crossref Search ADS PubMed WorldCat Brouquisse R , James F, Raymond P, Pradet A ( 1991 ) Study of glucose starvation in excised maize root tips . Plant Physiol 96 : 619 – 626 Google Scholar Crossref Search ADS PubMed WorldCat Buse MG , Reid SS ( 1975 ) Leucine: a possible regulator of protein turnover in muscle . J Clin Invest 56 : 1250 – 1261 Google Scholar Crossref Search ADS PubMed WorldCat Caspar T , Huber SC, Somerville CR ( 1985 ) Alterations in growth, photosynthesis, and respiration in a starchless mutant of Arabidopsis thaliana (L.) deficient in chloroplast phosphoglucomutase activity . Plant Physiol 79 : 11 – 17 Google Scholar Crossref Search ADS PubMed WorldCat Chen WP , Yang XY, Harms GL, Gray WM, Hegeman AD, Cohen JD ( 2011 ) An automated growth enclosure for metabolic labeling of Arabidopsis thaliana with 13C-carbon dioxide: an in vivo labeling system for proteomics and metabolomics research . Proteome Sci 9 : 9 Google Scholar Crossref Search ADS PubMed WorldCat Cheung CY , Williams TC, Poolman MG, Fell DA, Ratcliffe RG, Sweetlove LJ ( 2013 ) A method for accounting for maintenance costs in flux balance analysis improves the prediction of plant cell metabolic phenotypes under stress conditions . Plant J 75 : 1050 – 1061 Google Scholar Crossref Search ADS PubMed WorldCat Cline K , Dabney-Smith C ( 2008 ) Plastid protein import and sorting: different paths to the same compartments . Curr Opin Plant Biol 11 : 585 – 592 Google Scholar Crossref Search ADS PubMed WorldCat Daloso DM , Müller K, Obata T, Florian A, Tohge T, Bottcher A, Riondet C, Bariat L, Carrari F, Nunes-Nesi A, et al. ( 2015 ) Thioredoxin, a master regulator of the tricarboxylic acid cycle in plant mitochondria . Proc Natl Acad Sci USA 112 : E1392 – E1400 Google Scholar Crossref Search ADS PubMed WorldCat Dean C , Leech RM ( 1982 ) Genome expression during normal leaf development: 1. Cellular and chloroplast numbers and DNA, RNA, and protein levels in tissues of different ages within a seven-day-old wheat leaf . Plant Physiol 69 : 904 – 910 Google Scholar Crossref Search ADS PubMed WorldCat Detchon P , Possingham JV ( 1972 ) Ribosomal-RNA distribution during leaf development in spinach . Phytochemistry 11 : 943 – 947 Google Scholar Crossref Search ADS WorldCat Dungey NO , Davies DD ( 1982 ) Protein turnover in the attached leaves of non-stressed and stressed barley seedlings . Planta 154 : 435 – 440 Google Scholar Crossref Search ADS PubMed WorldCat Eckardt NA , Snyder GW, Portis AR Jr , Orgen WL ( 1997 ) Growth and photosynthesis under high and low irradiance of Arabidopsis thaliana antisense mutants with reduced ribulose-1,5-bisphosphate carboxylase/oxygenase activase content . Plant Physiol 113 : 575 – 586 Google Scholar Crossref Search ADS PubMed WorldCat Ferguson DL , Guikema JA, Paulsen GM ( 1990 ) Ubiquitin pool modulation and protein degradation in wheat roots during high temperature stress . Plant Physiol 92 : 740 – 746 Google Scholar Crossref Search ADS PubMed WorldCat Foyer CH , Parry M, Noctor G ( 2003 ) Markers and signals associated with nitrogen assimilation in higher plants . J Exp Bot 54 : 585 – 593 Google Scholar Crossref Search ADS PubMed WorldCat Fraser CM , Chapple C ( 2011 ) The phenylpropanoid pathway in Arabidopsis . The Arabidopsis Book 9 : e0152 , doi/10.1199/tab0152 Google Scholar Crossref Search ADS PubMed WorldCat Fritz C , Matt P, Müller C, Feil R, Stitt M ( 2006 ) Impact of the carbon-nitrogen status on the amino acid profile in tobacco leaves . Plant Cell Environ 29 : 2055 – 2076 Google Scholar Crossref Search ADS PubMed WorldCat Galili G ( 2002 ) New insights into the regulation and functional significance of lysine metabolism in plants . Annu Rev Plant Biol 53 : 27 – 43 Google Scholar Crossref Search ADS PubMed WorldCat Gibon Y , Bläsing O, Hannemann J, Carillo P, Höhne M, Cross J, Selbig J, Stitt M ( 2004 a ) A robot-based platform to measure multiple enzyme activities in Arabidopsis using a set of cycling assays: comparison of changes of enzyme activities and transcript levels during diurnal cycles and in prolonged darkness . Plant Cell 16 : 3304 – 3325 Google Scholar Crossref Search ADS PubMed WorldCat Gibon Y , Bläsing OE, Palacios-Rojas N, Pankovic D, Hendriks JHM, Fisahn J, Höhne M, Günther M, Stitt M ( 2004 b ) Adjustment of diurnal starch turnover to short days: depletion of sugar during the night leads to a temporary inhibition of carbohydrate utilization, accumulation of sugars and post-translational activation of ADP-glucose pyrophosphorylase in the following light period . Plant J 39 : 847 – 862 Google Scholar Crossref Search ADS PubMed WorldCat Gibon Y , Pyl ET, Sulpice R, Lunn JE, Höhne M, Günther M, Stitt M ( 2009 ) Adjustment of growth, starch turnover, protein content and central metabolism to a decrease of the carbon supply when Arabidopsis is grown in very short photoperiods . Plant Cell Environ 32 : 859 – 874 Google Scholar Crossref Search ADS PubMed WorldCat Gibon Y , Usadel B, Blaesing OE, Kamlage B, Hoehne M, Trethewey R, Stitt M ( 2006 ) Integration of metabolite with transcript and enzyme activity profiling during diurnal cycles in Arabidopsis rosettes . Genome Biol 7 : R76 Google Scholar Crossref Search ADS PubMed WorldCat Gruhler A , Schulze WX, Matthiesen R, Mann M, Jensen ON ( 2005 ) Stable isotope labeling of Arabidopsis thaliana cells and quantitative proteomics by mass spectrometry . Mol Cell Proteomics 4 : 1697 – 1709 Google Scholar Crossref Search ADS PubMed WorldCat He CY , Merrick BA, Mansfield BK, Hite MC, Daluge DR, Selkirk JK ( 1991 ) Comparison of 14C-amino acid mixture and [35S]methionine labeling of cellular proteins from mouse fibroblast C3H10T1/2 cells by two-dimensional gel electrophoresis . Electrophoresis 12 : 658 – 666 Google Scholar Crossref Search ADS PubMed WorldCat Heise R , Arrivault S, Szecowka M, Tohge T, Nunes-Nesi A, Stitt M, Nikoloski Z, Fernie AR ( 2014 ) Flux profiling of photosynthetic carbon metabolism in intact plants . Nat Protoc 9 : 1803 – 1824 Google Scholar Crossref Search ADS PubMed WorldCat Hoffmann WA , Poorter H ( 2002 ) Avoiding bias in calculations of relative growth rate . Ann Bot (Lond) 90 : 37 – 42 Google Scholar Crossref Search ADS WorldCat Hsiao TC ( 1970 ) Rapid changes in levels of polyribosomes in Zea mays in response to water stress . Plant Physiol 46 : 281 – 285 Google Scholar Crossref Search ADS PubMed WorldCat Huege J , Goetze J, Dethloff F, Junker B, Kopka J ( 2014 ) Quantification of stable isotope label in metabolites via mass spectrometry . Methods Mol Biol 1056 : 213 – 223 Google Scholar Crossref Search ADS PubMed WorldCat Huege J , Sulpice R, Gibon Y, Lisec J, Koehl K, Kopka J ( 2007 ) GC-EI-TOF-MS analysis of in vivo carbon-partitioning into soluble metabolite pools of higher plants by monitoring isotope dilution after 13CO2 labelling . Phytochemistry 68 : 2258 – 2272 Google Scholar Crossref Search ADS PubMed WorldCat Huffaker RC , Peterson LW ( 1974 ) Protein turnover in plants and possible means of its regulation . Annu Rev Plant Physiol 25 : 363 – 392 Google Scholar Crossref Search ADS WorldCat Hummel M , Cordewener JHG, de Groot JCM, Smeekens S, America AHP, Hanson J ( 2012 ) Dynamic protein composition of Arabidopsis thaliana cytosolic ribosomes in response to sucrose feeding as revealed by label free MSE proteomics . Proteomics 12 : 1024 – 1038 Google Scholar Crossref Search ADS PubMed WorldCat Izumi M , Hidema J, Makino A, Ishida H ( 2013 ) Autophagy contributes to nighttime energy availability for growth in Arabidopsis . Plant Physiol 161 : 1682 – 1693 Google Scholar Crossref Search ADS PubMed WorldCat Kawaguchi R , Girke T, Bray EA, Bailey-Serres J ( 2004 ) Differential mRNA translation contributes to gene regulation under non-stress and dehydration stress conditions in Arabidopsis thaliana . Plant J 38 : 823 – 839 Google Scholar Crossref Search ADS PubMed WorldCat Kawaguchi R , Williams AJ, Bray EA, Bailey-Serres J ( 2003 ) Water-deficit induced translational control in Nicotiana tabacum . Plant Cell Environ 26 : 221 – 229 Google Scholar Crossref Search ADS WorldCat Keegstra K , Cline K ( 1999 ) Protein import and routing systems of chloroplasts . Plant Cell 11 : 557 – 570 Google Scholar Crossref Search ADS PubMed WorldCat Kopka J , Schauer N, Krueger S, Birkemeyer C, Usadel B, Bergmüller E, Dörmann P, Weckwerth W, Gibon Y, Stitt M, et al. ( 2005 ) [email protected]: the Golm Metabolome Database . Bioinformatics 21 : 1635 – 1638 Google Scholar Crossref Search ADS PubMed WorldCat Krueger S , Giavalisco P, Krall L, Steinhauser MC, Büssis D, Usadel B, Flügge UI, Fernie AR, Willmitzer L, Steinhauser D ( 2011 ) A topological map of the compartmentalized Arabidopsis thaliana leaf metabolome . PLoS ONE 6 : e17806 Google Scholar Crossref Search ADS PubMed WorldCat Kushner DJ , Baker A, Dunstall TG ( 1999 ) Pharmacological uses and perspectives of heavy water and deuterated compounds . Can J Physiol Pharmacol 77 : 79 – 88 Google Scholar Crossref Search ADS PubMed WorldCat Lam HM , Coschigano KT, Oliveira IC, Melo-Oliveira R, Coruzzi GM ( 1996 ) The molecular genetics of N assimilation into amino acids in higher plants . Annu Rev Plant Physiol Plant Mol Biol 47 : 569 – 593 Google Scholar Crossref Search ADS PubMed WorldCat Lam HM , Hsieh MH, Coruzzi G ( 1998 ) Reciprocal regulation of distinct asparagine synthetase genes by light and metabolites in Arabidopsis thaliana . Plant J 16 : 345 – 353 Google Scholar Crossref Search ADS PubMed WorldCat Lattanzi FA , Schnyder H, Thornton B ( 2005 ) The sources of carbon and nitrogen supplying leaf growth: assessment of the role of stores with compartmental models . Plant Physiol 137 : 383 – 395 Google Scholar Crossref Search ADS PubMed WorldCat Lehmeier CA , Wild M, Schnyder H ( 2013 ) Nitrogen stress affects the turnover and size of nitrogen pools supplying leaf growth in a grass . Plant Physiol 162 : 2095 – 2105 Google Scholar Crossref Search ADS PubMed WorldCat Li L , Nelson CJ, Solheim C, Whelan J, Millar AH ( 2012 ) Determining degradation and synthesis rates of Arabidopsis proteins using the kinetics of progressive 15N labeling of two-dimensional gel-separated protein spots . Mol Cell Proteomics 11 : M111.010025 Google Scholar Crossref Search ADS PubMed WorldCat Lisec J , Schauer N, Kopka J, Willmitzer L, Fernie AR ( 2006 ) Gas chromatography mass spectrometry-based metabolite profiling in plants . Nat Protoc 1 : 387 – 396 Google Scholar Crossref Search ADS PubMed WorldCat Lodish H , Berk A, Zipursky SL, Matsudaira P, Baltimore D, Darnell J, editors (2011) The dynamic plant cell wall. In Molecular Cell Biology, Ed 4. WH Freeman, New York, section 22.5 http://www.ncbi.nlm.nih.gov/books/NBK21709/ (April 15, 2015) Luedemann A , Strassburg K, Erban A, Kopka J ( 2008 ) TagFinder for the quantitative analysis of gas chromatography-mass spectrometry (GC-MS)-based metabolite profiling experiments . Bioinformatics 24 : 732 – 737 Google Scholar Crossref Search ADS PubMed WorldCat Ma F , Jazmin LJ, Young JD, Allen DK ( 2014 ) Isotopically nonstationary 13C flux analysis of changes in Arabidopsis thaliana leaf metabolism due to high light acclimation . Proc Natl Acad Sci USA 111 : 16967 – 16972 Google Scholar Crossref Search ADS PubMed WorldCat Marín-Navarro J , Manuell AL, Wu J, P Mayfield S ( 2007 ) Chloroplast translation regulation . Photosynth Res 94 : 359 – 374 Google Scholar Crossref Search ADS PubMed WorldCat Martin SF , Munagapati VS, Salvo-Chirnside E, Kerr LE, Le Bihan T ( 2012 ) Proteome turnover in the green alga Ostreococcus tauri by time course 15N metabolic labeling mass spectrometry . J Proteome Res 11 : 476 – 486 Google Scholar Crossref Search ADS PubMed WorldCat Masclaux-Daubresse C , Chardon F ( 2011 ) Exploring nitrogen remobilization for seed filling using natural variation in Arabidopsis thaliana . J Exp Bot 62 : 2131 – 2142 Google Scholar Crossref Search ADS PubMed WorldCat Mastrobuoni G , Irgang S, Pietzke M, Assmus HE, Wenzel M, Schulze WX, Kempa S ( 2012 ) Proteome dynamics and early salt stress response of the photosynthetic organism Chlamydomonas reinhardtii . BMC Genomics 13 : 215 Google Scholar Crossref Search ADS PubMed WorldCat Mitra R , Burton J, Varner JE ( 1976 ) Deuterium oxide as a tool for the study of amino acid metabolism . Anal Biochem 70 : 1 – 17 Google Scholar Crossref Search ADS PubMed WorldCat Nelson CJ , Alexova R, Jacoby RP, Millar AH ( 2014 a ) Proteins with high turnover rate in barley leaves estimated by proteome analysis combined with in planta isotope labeling . Plant Physiol 166 : 91 – 108 Google Scholar Crossref Search ADS PubMed WorldCat Nelson CJ , Li L, Jacoby RP, Millar AH ( 2013 ) Degradation rate of mitochondrial proteins in Arabidopsis thaliana cells . J Proteome Res 12 : 3449 – 3459 Google Scholar Crossref Search ADS PubMed WorldCat Nelson CJ , Li L, Millar AH ( 2014 b ) Quantitative analysis of protein turnover in plants . Proteomics 14 : 579 – 592 Google Scholar Crossref Search ADS PubMed WorldCat Nicolaï M , Roncato MA, Canoy AS, Rouquié D, Sarda X, Freyssinet G, Robaglia C ( 2006 ) Large-scale analysis of mRNA translation states during sucrose starvation in Arabidopsis cells identifies cell proliferation and chromatin structure as targets of translational control . Plant Physiol 141 : 663 – 673 Google Scholar Crossref Search ADS PubMed WorldCat Ong SE , Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, Mann M ( 2002 ) Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics . Mol Cell Proteomics 1 : 376 – 386 Google Scholar Crossref Search ADS PubMed WorldCat Ong SE , Mann M ( 2006 ) A practical recipe for stable isotope labeling by amino acids in cell culture (SILAC) . Nat Protoc 1 : 2650 – 2660 Google Scholar Crossref Search ADS PubMed WorldCat Pal SK , Liput M, Piques M, Ishihara H, Obata T, Martins MC, Sulpice R, van Dongen JT, Fernie AR, Yadav UP, et al. ( 2013 ) Diurnal changes of polysome loading track sucrose content in the rosette of wild-type Arabidopsis and the starchless pgm mutant . Plant Physiol 162 : 1246 – 1265 Google Scholar Crossref Search ADS PubMed WorldCat Pannemans DL , Schaafsma G, Westerterp KR ( 1997 ) Calcium excretion, apparent calcium absorption and calcium balance in young and elderly subjects: influence of protein intake . Br J Nutr 77 : 721 – 729 Google Scholar Crossref Search ADS PubMed WorldCat Penning de Vries FWT ( 1975 ) The cost of maintenance processes in plant cells . Ann Bot (Lond) 39 : 77 – 92 Google Scholar Crossref Search ADS WorldCat Penning de Vries FWT , Brunsting AH, van Laar HH ( 1974 ) Products, requirements and efficiency of biosynthesis: a quantitative approach . J Theor Biol 45 : 339 – 377 Google Scholar Crossref Search ADS PubMed WorldCat Penning de Vries FWT , Witlage JM, Kremer D ( 1979 ) Rates of respiration and of increase in structural dry matter in young wheat, ryegrass and maize plants in relation to temperature, to water stress and to their sugar content . Ann Bot (Lond) 44 : 595 – 609 Google Scholar Crossref Search ADS WorldCat Pilkington SM , Encke B, Krohn N, Höhne M, Stitt M, Pyl ET ( 2015 ) Relationship between starch degradation and carbon demand for maintenance and growth in Arabidopsis thaliana in different irradiance and temperature regimes . Plant Cell Environ 38 : 157 – 171 Google Scholar Crossref Search ADS PubMed WorldCat Piques M , Schulze WX, Höhne M, Usadel B, Gibon Y, Rohwer J, Stitt M ( 2009 ) Ribosome and transcript copy numbers, polysome occupancy and enzyme dynamics in Arabidopsis . Mol Syst Biol 5 : 314 Google Scholar Crossref Search ADS PubMed WorldCat Pocrnjic Z , Mathews RW, Rappaport S, Haschemeyer AEV ( 1983 ) Quantitative protein synthetic rates in various tissues of a temperate fish in vivo by the method of phenylalanine swamping . Comp Biochem Physiol B 74 : 735 – 738 Google Scholar Crossref Search ADS PubMed WorldCat Pons T , van Rijnberk H, Scheurwater I, van der Werf A ( 1993 ) Importance of the gradient in photosynthetically active radiation in a vegetation stand for leaf nitrogen allocation in two monocotyledons . Oecologia 95 : 416 – 424 Google Scholar Crossref Search ADS PubMed WorldCat Poorter H , Nagel O ( 2000 ) The role of biomass allocation in the growth response of plants to different levels of light, CO2, nutrients and water: a quantitative review . Aust J Plant Physiol 27 : 595 – 607 Google Scholar OpenURL Placeholder Text WorldCat Price J , Laxmi A, St Martin SK, Jang JC ( 2004 ) Global transcription profiling reveals multiple sugar signal transduction mechanisms in Arabidopsis . Plant Cell 16 : 2128 – 2150 Google Scholar Crossref Search ADS PubMed WorldCat Pyl ET , Piques M, Ivakov A, Schulze W, Ishihara H, Stitt M, Sulpice R ( 2012 ) Metabolism and growth in Arabidopsis depend on the daytime temperature but are temperature-compensated against cool nights . Plant Cell 24 : 2443 – 2469 Google Scholar Crossref Search ADS PubMed WorldCat Rapparini F , Cohen J, Slovin J ( 1999 ) Indole-3-acetic acid biosynthesis in Lemna gibba studied using stable isotope labeled anthranilate and tryptophan . Plant Growth Regul 27 : 139 – 144 Google Scholar Crossref Search ADS WorldCat Riens B , Lohaus G, Heineke D, Heldt HW ( 1991 ) Amino acid and sucrose content determined in the cytosolic, chloroplastic, and vacuolar compartments and in the phloem sap of spinach leaves . Plant Physiol 97 : 227 – 233 Google Scholar Crossref Search ADS PubMed WorldCat Sacchi GA , Cocucci M ( 1992 ) Effects of deuterium oxide on growth, proton extrusion, potassium influx, and in vitro plasma membrane activities in maize root segments . Plant Physiol 100 : 1962 – 1967 Google Scholar Crossref Search ADS PubMed WorldCat Saeman JF , Bubl JL, Harris EE ( 1945 ) Quantitative saccharification of wood and cellulose . Ind Eng Chem 17 : 35 – 37 Google Scholar OpenURL Placeholder Text WorldCat Schaefer J , Skokut TA, Stejskal EO, McKay RA, Varner JE ( 1981 ) Estimation of protein turnover in soybean leaves using magic angle double cross-polarization nitrogen 15 nuclear magnetic resonance . J Biol Chem 256 : 11574 – 11579 Google Scholar Crossref Search ADS PubMed WorldCat Scheible WR , Krapp A, Stitt M ( 2000 ) Reciprocal changes of phosphenolpyruvate carboxylase and cytosolic pyruvate kinase, citrate synthase and NADP-isocitrate dehydrogenase expression regulate organic acid metabolism during nitrate assimilation in tobacco leaves . Plant Cell Environ 23 : 1155 – 1167 Google Scholar Crossref Search ADS WorldCat Scheurwater I , Dünnebacke M, Eising R, Lambers H ( 2000 ) Respiratory costs and rate of protein turnover in the roots of a fast-growing (Dactylis glomerata L.) and a slow-growing (Festuca ovina L.) grass species . J Exp Bot 51 : 1089 – 1097 Google Scholar PubMed OpenURL Placeholder Text WorldCat Schütz W , Hausmann N, Krug K, Hampp R, Macek B ( 2011 ) Extending SILAC to proteomics of plant cell lines . Plant Cell 23 : 1701 – 1705 Google Scholar Crossref Search ADS PubMed WorldCat Smith AM , Stitt M ( 2007 ) Coordination of carbon supply and plant growth . Plant Cell Environ 30 : 1126 – 1149 Google Scholar Crossref Search ADS PubMed WorldCat Somerville C , Bauer S, Brininstool G, Facette M, Hamann T, Milne J, Osborne E, Paredez A, Persson S, Raab T, et al. ( 2004 ) Toward a systems approach to understanding plant cell walls . Science 306 : 2206 – 2211 Google Scholar Crossref Search ADS PubMed WorldCat Stitt M , Zeeman SC ( 2012 ) Starch turnover: pathways, regulation and role in growth . Curr Opin Plant Biol 15 : 282 – 292 Google Scholar Crossref Search ADS PubMed WorldCat Sulpice R , Flis A, Ivakov AA, Apelt F, Krohn N, Encke B, Abel C, Feil R, Lunn JE, Stitt M ( 2014 ) Arabidopsis coordinates the diurnal regulation of carbon allocation and growth across a wide range of photoperiods . Mol Plant 7 : 137 – 155 Google Scholar Crossref Search ADS PubMed WorldCat Sweetlove LJ , Obata T, Fernie AR ( 2014 ) Systems analysis of metabolic phenotypes: what have we learnt? Trends Plant Sci 19 : 222 – 230 Google Scholar Crossref Search ADS PubMed WorldCat Szecowka M , Heise R, Tohge T, Nunes-Nesi A, Vosloh D, Huege J, Feil R, Lunn J, Nikoloski Z, Stitt M, et al. ( 2013 ) Metabolic fluxes in an illuminated Arabidopsis rosette . Plant Cell 25 : 694 – 714 Google Scholar Crossref Search ADS PubMed WorldCat Tcherkez G , Boex-Fontvieille E, Mahé A, Hodges M ( 2012 a ) Respiratory carbon fluxes in leaves . Curr Opin Plant Biol 15 : 308 – 314 Google Scholar Crossref Search ADS PubMed WorldCat Tcherkez G , Mahé A, Guérard F, Boex-Fontvieille ER, Gout E, Lamothe M, Barbour MM, Bligny R ( 2012 b ) Short-term effects of CO2 and O2 on citrate metabolism in illuminated leaves . Plant Cell Environ 35 : 2208 – 2220 Google Scholar Crossref Search ADS PubMed WorldCat Thimm O , Bläsing O, Gibon Y, Nagel A, Meyer S, Krüger P, Selbig J, Müller LA, Rhee SY, Stitt M ( 2004 ) MAPMAN: a user-driven tool to display genomics data sets onto diagrams of metabolic pathways and other biological processes . Plant J 37 : 914 – 939 Google Scholar Crossref Search ADS PubMed WorldCat Thomson JF ( 1963 ) Biological Effects of Deuterium. Pergamon Press , New York , pp 4 – 5 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tovar-Méndez A , Miernyk JA, Randall DD ( 2003 ) Regulation of pyruvate dehydrogenase complex activity in plant cells . Eur J Biochem 270 : 1043 – 1049 Google Scholar Crossref Search ADS PubMed WorldCat Tschoep H , Gibon Y, Carillo P, Armengaud P, Szecowka M, Nunes-Nesi A, Fernie AR, Koehl K, Stitt M ( 2009 ) Adjustment of growth and central metabolism to a mild but sustained nitrogen-limitation in Arabidopsis . Plant Cell Environ 32 : 300 – 318 Google Scholar Crossref Search ADS PubMed WorldCat Urbanczyk-Wochniak E , Baxter C, Kolbe A, Kopka J, Sweetlove LJ, Fernie AR ( 2005 ) Profiling of diurnal patterns of metabolite and transcript abundance in potato (Solanum tuberosum) leaves . Planta 221 : 891 – 903 Google Scholar Crossref Search ADS PubMed WorldCat Usadel B , Bläsing OE, Gibon Y, Höhne M, Günter M, Stitt M ( 2008 ) Global transcript levels respond to small changes of the carbon status during progressive exhaustion of carbohydrates in Arabidopsis rosettes . Plant Physiol 146 : 1834 – 1861 Google Scholar Crossref Search ADS PubMed WorldCat Weiner H , Heldt HW ( 1992 ) Inter- and intracellular distribution of amino acids and other metabolites in maize (Zea mays L.) leaves . Planta 187 : 242 – 246 Google Scholar Crossref Search ADS PubMed WorldCat Williams TC , Poolman MG, Howden AJ, Schwarzlander M, Fell DA, Ratcliffe RG, Sweetlove LJ ( 2010 ) A genome-scale metabolic model accurately predicts fluxes in central carbon metabolism under stress conditions . Plant Physiol 154 : 311 – 323 Google Scholar Crossref Search ADS PubMed WorldCat Winter H , Lohaus G, Heldt HW ( 1992 ) Phloem transport of amino acids in relation to their cytosolic levels in barley leaves . Plant Physiol 99 : 996 – 1004 Google Scholar Crossref Search ADS PubMed WorldCat Yang XY , Chen WP, Rendahl AK, Hegeman AD, Gray WM, Cohen JD ( 2010 ) Measuring the turnover rates of Arabidopsis proteins using deuterium oxide: an auxin signaling case study . Plant J 63 : 680 – 695 Google Scholar Crossref Search ADS PubMed WorldCat Yee JC , Jacob NM, Jayapal KP, Kok YJ, Philp R, Griffin TJ, Hu WS ( 2010 ) Global assessment of protein turnover in recombinant antibody producing myeloma cells . J Biotechnol 148 : 182 – 193 Google Scholar Crossref Search ADS PubMed WorldCat Young JD , Shastri AA, Stephanopoulos G, Morgan JA ( 2011 ) Mapping photoautotrophic metabolism with isotopically nonstationary 13C flux analysis . Metab Eng 13 : 656 – 665 Google Scholar Crossref Search ADS PubMed WorldCat Yuan J , Bennett BD, Rabinowitz JD ( 2008 ) Kinetic flux profiling for quantitation of cellular metabolic fluxes . Nat Protoc 3 : 1328 – 1340 Google Scholar Crossref Search ADS PubMed WorldCat Yuan J , Fowler WU, Kimball E, Lu W, Rabinowitz JD ( 2006 ) Kinetic flux profiling of nitrogen assimilation in Escherichia coli . Nat Chem Biol 2 : 529 – 530 Google Scholar Crossref Search ADS PubMed WorldCat Zhou H , Li W, Wang SP, Mendoza V, Rosa R, Hubert J, Herath K, McLaughlin T, Rohm RJ, Lassman ME, et al. ( 2012 ) Quantifying apoprotein synthesis in rodents: coupling LC-MS/MS analyses with the administration of labeled water . J Lipid Res 53 : 1223 – 1231 Google Scholar Crossref Search ADS PubMed WorldCat Author notes 1 This work was supported by the Max Planck Society and the European Union (collaborative project TiMet under contract no. 245143). 2 Present address: National University of Galway, Plant Systems Biology Laboratory, Plant and AgriBiosciences Research Centre, Botany and Plant Science, Aras de Brun, University Road, Galway, Republic of Ireland. * Address correspondence to [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Mark Stitt ([email protected]). [OPEN] Articles can be viewed without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.15.00209 © 2015 American Society of Plant Biologists. All Rights Reserved. © The Author(s) 2015. Published by Oxford University Press on behalf of American Society of Plant Biologists. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Gesell, Andreas; Blaukopf, Markus; Madilao, Lina; Yuen, Macaire M.S.; Withers, Stephen G.; Mattsson, Jim; Russell, John H.; Bohlmann, Jörg

doi: 10.1104/pp.15.00315pmid: 25829465

Abstract Western redcedar (WRC; Thuja plicata) produces high amounts of oxygenated thujone monoterpenoids associated with resistance against herbivore feeding, particularly ungulate browsing. Thujones and other monoterpenoids accumulate in glandular structures in the foliage of WRC. Thujones are produced from (+)-sabinene by sabinol and sabinone. Using metabolite analysis, enzyme assays with WRC tissue extracts, cloning, and functional characterization of cytochrome P450 monooxygenases, we established that trans-sabin-3-ol but not cis-sabin-3-ol is the intermediate in thujone biosynthesis in WRC. Based on transcriptome analysis, full-length complementary DNA cloning, and characterization of expressed P450 proteins, we identified CYP750B1 and CYP76AA25 as the enzymes that catalyze the hydroxylation of (+)-sabinene to trans-sabin-3-ol. Gene-specific transcript analysis in contrasting WRC genotypes producing high and low amounts of monoterpenoids, including a glandless low-terpenoid clone, as well as assays for substrate specificity supported a biological role of CYP750B1 in α- and β-thujone biosynthesis. This P450 belongs to the apparently gymnosperm-specific CYP750 family and is, to our knowledge, the first member of this family to be functionally characterized. In contrast, CYP76AA25 has a broader substrate spectrum, also converting the sesquiterpene farnesene and the herbicide isoproturon, and its transcript profiles are not well correlated with thujone accumulation. Western redcedar (WRC; Thuja plicata) is a Cupressaceae species native to the Pacific Northwest of North America. This evergreen gymnosperm tree has been used for thousands of years by First Nations people as a preferred building material for houses and boats and a weaving material for baskets, mats, and clothing as well as for carvings, totem poles, and various other practical and ceremonial purposes (Hebda and Mathewes, 1984). WRC is highly valued in the forest industry for its light, uniform, and durable wood. As a building material, WRC provides high-dimensional stability and good thermal insulation properties. Because of its high durability, WRC wood is widely used for exterior applications, such as roofing and sidings, decks, and exterior furniture. In addition, the essential oil of WRC, which is rich in mono- and sesquiterpenoids, is used in the cosmetics and fragrance industry because of its antimicrobial and scent qualities. To support sustainable reforestation and harvest of WRC, a breeding program was established (Russell and Ferguson, 2008; Russell and Yanchuk, 2012). Traits of particular interest in WRC breeding include heartwood durability, growth, and environmental adaptation as well as resistance to pathogens and herbivores. Decay resistance of WRC heartwood is attributed to the accumulation of terpenoid- and lignan-specialized (i.e. secondary) metabolites (Morris and Stirling, 2012), some of which are named after this species (e.g. thujaplicin terpenoids and plicatic acid and thujaplicatin lignans; Anderson and Gripenberg, 1948; Gardner et al., 1959; MacLean and Murakami, 1966). The oxidized monoterpenoids α- and β-thujone are characteristic specialized metabolites of the foliage and bark/phloem (Fig. 1). Figure 1. Open in new tabDownload slide Proposed biosynthesis of α- and β-thujone in WRC. The monoterpenenoids α- and β-thujone are proposed to derive from GDP through a four-step biosynthetic pathway starting with (+)-sabinene synthase (Foster et al., 2013). Sabinene may be oxidized to either (+)-cis- or (+)-trans-sabinol followed by oxidation to (+)-sabinone as the proposed direct precursor to α- and β-thujone. Here, we provide evidence for (+)-trans-sabinol but not (+)-cis sabinol as an intermediate in thujone biosynthesis in the gymnosperm WRC. In contrast, in the angiosperm common garden sage, thujone is formed by the (+)-cis-sabinol intermediate (Karp and Croteau, 1982). This article describes (+)-trans-sabinol dehydrogenase enzyme activity in WRC tissue extract and gene discovery and functional characterization of WRC (+)-sabinene-3-hydroxylase. The proposed pathway was redrawn based on Karp and Croteau (1982). Figure 1. Open in new tabDownload slide Proposed biosynthesis of α- and β-thujone in WRC. The monoterpenenoids α- and β-thujone are proposed to derive from GDP through a four-step biosynthetic pathway starting with (+)-sabinene synthase (Foster et al., 2013). Sabinene may be oxidized to either (+)-cis- or (+)-trans-sabinol followed by oxidation to (+)-sabinone as the proposed direct precursor to α- and β-thujone. Here, we provide evidence for (+)-trans-sabinol but not (+)-cis sabinol as an intermediate in thujone biosynthesis in the gymnosperm WRC. In contrast, in the angiosperm common garden sage, thujone is formed by the (+)-cis-sabinol intermediate (Karp and Croteau, 1982). This article describes (+)-trans-sabinol dehydrogenase enzyme activity in WRC tissue extract and gene discovery and functional characterization of WRC (+)-sabinene-3-hydroxylase. The proposed pathway was redrawn based on Karp and Croteau (1982). Herbivore feeding, specifically browsing by ungulate deer and elk, is a major ecological problem and economic cost factor for WRC reforestation. High foliage terpenoid content has been associated with reduced herbivore damage (Vourc’h et al., 2001, 2002; Burney and Jacobs, 2011; Kimball et al., 2012). The monoterpenes (+)-sabinene and α- and β-thujone are abundant in WRC foliage and account for 65% to 83% of terpenoids in WRC essential oil (Von Rudloff et al., 1988; Kimball et al., 2005; Tsiri et al., 2009; Tisserand and Young, 2014). These three monoterpenes are also found in other Cupressaceae species, such as the genera Juniperus spp. and Cupressus spp. They are also major monoterpenes in species of other plant families, including Asteraceae (e.g. species of wormwood [Artemisia spp.] or tansy [Tanacetum vulgare]) and Lamiaceae (e.g. common garden sage [Salvia officinalis]; Tisserand and Young, 2014). Based on characterization of purified enzyme preparations from common garden sage, Karp and Croteau (1982) proposed a biosynthetic pathway leading from GDP by (+)-sabinene to thujone (Fig. 1). Karp and Croteau (1982) showed that activity of cytochrome P450 catalyzed hydroxylation of (+)-sabinene synthesis in common garden sage. In recent work with WRC, we discovered and functionally characterized the monoterpene synthase gene encoding (+)-sabinene synthase (Foster et al., 2013). Terpene synthases are well characterized in gymnosperms and angiosperms (Keeling and Bohlmann, 2006; Chen et al., 2011). In contrast and to the best of our knowledge, a P450 gene involved in monoterpene biosynthesis has not yet been functionally characterized in a gymnosperm. The only functionally characterized terpene-modifying P450 genes of gymnosperms are CYP725A of taxol diterpene biosynthesis in Taxus spp. (Croteau et al., 2006; Rontein et al., 2008), CYP720B1 and CYP720B4 of diterpene resin acid biosynthesis in loblolly pine (Pinus taeda; Ro et al., 2005) and Sitka spruce (Picea sitchensis; Hamberger et al., 2011), and CYP706M1 of nootkatone sesquiterpene biosynthesis in yellow cypress (Callitropsis nootkatensis; Cankar et al., 2014). Building on a previous WRC transcriptome analysis (Foster et al., 2013), we describe here, to our knowledge, the first gene discovery and functional characterization of monoterpene-oxidizing P450s in a gymnosperm, specifically WRC (+)-sabinene hydroxylases CYP750B1 and CYP76AA25. A biosynthetic function of CYP750B1 in α- and β-thujone biosynthesis is supported by profiles of metabolite and transcript accumulation in contrasting genotypes of WRC with distinct patterns of terpenoid and thujone accumulation. WRC CYP750B1 is, to our knowledge, the first functionally characterized member of the apparently gymnosperm-specific CYP750 family. RESULTS Monoterpene Profiles of WRC Foliage and Bark Metabolite analysis by gas chromatography (GC)-mass spectrometry (MS) revealed similar monoterpenoid profiles in extracts of foliage and green bark of WRC saplings (genotype 5309) as shown with the results from three biological replicates (Table I). Genotype 5309 represents a phenotype with high levels of monoterpenoids and high resistance to deer browsing when tested against lower monoterpenoid genotypes (J. Russell, unpublished data). α-thujone was the major monoterpenoid in foliage and bark, accounting for two-thirds of total monoterpenoid content and 0.2% to 0.4% (w/w) of tissue dry weight. The second most abundant compound in both foliage and bark was (+)-sabinene followed by β-thujone. Quantification and identification of monoterpenoids in WRC foliage and bark extracts Table I. Quantification and identification of monoterpenoids in WRC foliage and bark extracts Monoterpene . RI DB1 . RI HP5 . Foliage . Bark . Determined . Reference . Determined . Reference . mg g−1 dry wt ± se a α-Thujeneb , c 932 932 924 924 0.007 ± 0.0018 0.021 ± 0.0078 α-Pineneb , d 938 936 929 932 0.019 ± 0.0038 0.086 ± 0.0287 Camphened , e 950 955 943 946 0.001 ± 0.0001 0.001 ± 0.0007 Sabinene b , c , f 970 973 970 969 0.507 ± 0.0320 f 0.481 ± 0.1523 f β-Pineneb , d 973 972 974 974 0.001 ± 0.0005 0.004 ± 0.0019 β-Myrceneb , c 986 987 988 988 0.056 ± 0.0071 0.083 ± 0.0364 α-Phellandreneb , d 996 1,002 1,002 1,002 0.188 ± 0.0211 0.039 ± 0.0341 α-Terpineneb , c 1,011 1,013 1,014 1,014 0.053 ± 0.0080 0.033 ± 0.0094 p-Cymeneb , d 1,015 1,015 1,022 1,022 0.009 ± 0.0038 0.023 ± 0.0133 β-Phellandreneb , d 1,021 1,023 1,028 1,025 0.010 ± 0.0017 0.001 ± 0.0210 (S)-Limoneneb , c 1,023 1,025 1,026 1,024 0.030 ± 0.0006 0.066 ± 0.0002 γ-Terpineneb , c 1,051 1,051 1,058 1,054 0.054 ± 0.0064 0.031 ± 0.0095 trans-Sabinene hydratee , g 1,054 1,053 1,064 1,065 0.018 ± 0.0088 0.017 ± 0.0073 p-Cymenened , e 1,075 1,075 1,088 1,082 0.001 ± 0.0004 0.003 ± 0.0024 Terpinoleneb , g 1,081 1,086 1,086 1,086 0.023 ± 0.0098 0.012 ± 0.0041 cis-Sabinene hydrateb , g 1,082 1,096 1,083 1,083 0.016 ± 0.0135 0.006 ± 0.0048 α-Thujone b , c , f 1,088 1,089 1,103 1,101 3.507 ± 0.6845 f 2.304 ± 0.9221 f β-Thujone b , h , f 1,098 1,103 1113 1,112 0.364 ± 0.0802 f 0.216 ± 0.0745 f Thujol/neothujole , g 1,134 1,136 1,148 1,149 0.159 ± 0.0372 0.003 ± 0.0022 Cymen-8-ole , g 1,162 1,169 1,183 1179 0.010 ± 0.0047 Terpinen-4-olb , c 1,164 1,164 1,174 1,174 0.014 ± 0.0026 0.019 ± 0.0100 trans-Sabinyl acetateb , g 1,275 1,278 1,292 1,289 0.002 ± 0.0004 0.006 ± 0.0026 Carvacrole , g 1,279 1,278 1,302 1,298 0.011 ± 0.0024 Monoterpene . RI DB1 . RI HP5 . Foliage . Bark . Determined . Reference . Determined . Reference . mg g−1 dry wt ± se a α-Thujeneb , c 932 932 924 924 0.007 ± 0.0018 0.021 ± 0.0078 α-Pineneb , d 938 936 929 932 0.019 ± 0.0038 0.086 ± 0.0287 Camphened , e 950 955 943 946 0.001 ± 0.0001 0.001 ± 0.0007 Sabinene b , c , f 970 973 970 969 0.507 ± 0.0320 f 0.481 ± 0.1523 f β-Pineneb , d 973 972 974 974 0.001 ± 0.0005 0.004 ± 0.0019 β-Myrceneb , c 986 987 988 988 0.056 ± 0.0071 0.083 ± 0.0364 α-Phellandreneb , d 996 1,002 1,002 1,002 0.188 ± 0.0211 0.039 ± 0.0341 α-Terpineneb , c 1,011 1,013 1,014 1,014 0.053 ± 0.0080 0.033 ± 0.0094 p-Cymeneb , d 1,015 1,015 1,022 1,022 0.009 ± 0.0038 0.023 ± 0.0133 β-Phellandreneb , d 1,021 1,023 1,028 1,025 0.010 ± 0.0017 0.001 ± 0.0210 (S)-Limoneneb , c 1,023 1,025 1,026 1,024 0.030 ± 0.0006 0.066 ± 0.0002 γ-Terpineneb , c 1,051 1,051 1,058 1,054 0.054 ± 0.0064 0.031 ± 0.0095 trans-Sabinene hydratee , g 1,054 1,053 1,064 1,065 0.018 ± 0.0088 0.017 ± 0.0073 p-Cymenened , e 1,075 1,075 1,088 1,082 0.001 ± 0.0004 0.003 ± 0.0024 Terpinoleneb , g 1,081 1,086 1,086 1,086 0.023 ± 0.0098 0.012 ± 0.0041 cis-Sabinene hydrateb , g 1,082 1,096 1,083 1,083 0.016 ± 0.0135 0.006 ± 0.0048 α-Thujone b , c , f 1,088 1,089 1,103 1,101 3.507 ± 0.6845 f 2.304 ± 0.9221 f β-Thujone b , h , f 1,098 1,103 1113 1,112 0.364 ± 0.0802 f 0.216 ± 0.0745 f Thujol/neothujole , g 1,134 1,136 1,148 1,149 0.159 ± 0.0372 0.003 ± 0.0022 Cymen-8-ole , g 1,162 1,169 1,183 1179 0.010 ± 0.0047 Terpinen-4-olb , c 1,164 1,164 1,174 1,174 0.014 ± 0.0026 0.019 ± 0.0100 trans-Sabinyl acetateb , g 1,275 1,278 1,292 1,289 0.002 ± 0.0004 0.006 ± 0.0026 Carvacrole , g 1,279 1,278 1,302 1,298 0.011 ± 0.0024 a se based on three biological replicates. b Identified by comparison with authentic standards. c Quantified using concentration series of authentic standard. d Quantity estimated using sabinene Rf (nonoxygenated compounds). e Identified based on comparison with NIST or Massfinder4 databases. f The three major monoterpene compounds are shown in boldface. g Quantity estimated using terpinene-4-ol Rf (oxygenated compounds). h Quantity estimated using α-thujone Rf. Open in new tab Table I. Quantification and identification of monoterpenoids in WRC foliage and bark extracts Monoterpene . RI DB1 . RI HP5 . Foliage . Bark . Determined . Reference . Determined . Reference . mg g−1 dry wt ± se a α-Thujeneb , c 932 932 924 924 0.007 ± 0.0018 0.021 ± 0.0078 α-Pineneb , d 938 936 929 932 0.019 ± 0.0038 0.086 ± 0.0287 Camphened , e 950 955 943 946 0.001 ± 0.0001 0.001 ± 0.0007 Sabinene b , c , f 970 973 970 969 0.507 ± 0.0320 f 0.481 ± 0.1523 f β-Pineneb , d 973 972 974 974 0.001 ± 0.0005 0.004 ± 0.0019 β-Myrceneb , c 986 987 988 988 0.056 ± 0.0071 0.083 ± 0.0364 α-Phellandreneb , d 996 1,002 1,002 1,002 0.188 ± 0.0211 0.039 ± 0.0341 α-Terpineneb , c 1,011 1,013 1,014 1,014 0.053 ± 0.0080 0.033 ± 0.0094 p-Cymeneb , d 1,015 1,015 1,022 1,022 0.009 ± 0.0038 0.023 ± 0.0133 β-Phellandreneb , d 1,021 1,023 1,028 1,025 0.010 ± 0.0017 0.001 ± 0.0210 (S)-Limoneneb , c 1,023 1,025 1,026 1,024 0.030 ± 0.0006 0.066 ± 0.0002 γ-Terpineneb , c 1,051 1,051 1,058 1,054 0.054 ± 0.0064 0.031 ± 0.0095 trans-Sabinene hydratee , g 1,054 1,053 1,064 1,065 0.018 ± 0.0088 0.017 ± 0.0073 p-Cymenened , e 1,075 1,075 1,088 1,082 0.001 ± 0.0004 0.003 ± 0.0024 Terpinoleneb , g 1,081 1,086 1,086 1,086 0.023 ± 0.0098 0.012 ± 0.0041 cis-Sabinene hydrateb , g 1,082 1,096 1,083 1,083 0.016 ± 0.0135 0.006 ± 0.0048 α-Thujone b , c , f 1,088 1,089 1,103 1,101 3.507 ± 0.6845 f 2.304 ± 0.9221 f β-Thujone b , h , f 1,098 1,103 1113 1,112 0.364 ± 0.0802 f 0.216 ± 0.0745 f Thujol/neothujole , g 1,134 1,136 1,148 1,149 0.159 ± 0.0372 0.003 ± 0.0022 Cymen-8-ole , g 1,162 1,169 1,183 1179 0.010 ± 0.0047 Terpinen-4-olb , c 1,164 1,164 1,174 1,174 0.014 ± 0.0026 0.019 ± 0.0100 trans-Sabinyl acetateb , g 1,275 1,278 1,292 1,289 0.002 ± 0.0004 0.006 ± 0.0026 Carvacrole , g 1,279 1,278 1,302 1,298 0.011 ± 0.0024 Monoterpene . RI DB1 . RI HP5 . Foliage . Bark . Determined . Reference . Determined . Reference . mg g−1 dry wt ± se a α-Thujeneb , c 932 932 924 924 0.007 ± 0.0018 0.021 ± 0.0078 α-Pineneb , d 938 936 929 932 0.019 ± 0.0038 0.086 ± 0.0287 Camphened , e 950 955 943 946 0.001 ± 0.0001 0.001 ± 0.0007 Sabinene b , c , f 970 973 970 969 0.507 ± 0.0320 f 0.481 ± 0.1523 f β-Pineneb , d 973 972 974 974 0.001 ± 0.0005 0.004 ± 0.0019 β-Myrceneb , c 986 987 988 988 0.056 ± 0.0071 0.083 ± 0.0364 α-Phellandreneb , d 996 1,002 1,002 1,002 0.188 ± 0.0211 0.039 ± 0.0341 α-Terpineneb , c 1,011 1,013 1,014 1,014 0.053 ± 0.0080 0.033 ± 0.0094 p-Cymeneb , d 1,015 1,015 1,022 1,022 0.009 ± 0.0038 0.023 ± 0.0133 β-Phellandreneb , d 1,021 1,023 1,028 1,025 0.010 ± 0.0017 0.001 ± 0.0210 (S)-Limoneneb , c 1,023 1,025 1,026 1,024 0.030 ± 0.0006 0.066 ± 0.0002 γ-Terpineneb , c 1,051 1,051 1,058 1,054 0.054 ± 0.0064 0.031 ± 0.0095 trans-Sabinene hydratee , g 1,054 1,053 1,064 1,065 0.018 ± 0.0088 0.017 ± 0.0073 p-Cymenened , e 1,075 1,075 1,088 1,082 0.001 ± 0.0004 0.003 ± 0.0024 Terpinoleneb , g 1,081 1,086 1,086 1,086 0.023 ± 0.0098 0.012 ± 0.0041 cis-Sabinene hydrateb , g 1,082 1,096 1,083 1,083 0.016 ± 0.0135 0.006 ± 0.0048 α-Thujone b , c , f 1,088 1,089 1,103 1,101 3.507 ± 0.6845 f 2.304 ± 0.9221 f β-Thujone b , h , f 1,098 1,103 1113 1,112 0.364 ± 0.0802 f 0.216 ± 0.0745 f Thujol/neothujole , g 1,134 1,136 1,148 1,149 0.159 ± 0.0372 0.003 ± 0.0022 Cymen-8-ole , g 1,162 1,169 1,183 1179 0.010 ± 0.0047 Terpinen-4-olb , c 1,164 1,164 1,174 1,174 0.014 ± 0.0026 0.019 ± 0.0100 trans-Sabinyl acetateb , g 1,275 1,278 1,292 1,289 0.002 ± 0.0004 0.006 ± 0.0026 Carvacrole , g 1,279 1,278 1,302 1,298 0.011 ± 0.0024 a se based on three biological replicates. b Identified by comparison with authentic standards. c Quantified using concentration series of authentic standard. d Quantity estimated using sabinene Rf (nonoxygenated compounds). e Identified based on comparison with NIST or Massfinder4 databases. f The three major monoterpene compounds are shown in boldface. g Quantity estimated using terpinene-4-ol Rf (oxygenated compounds). h Quantity estimated using α-thujone Rf. Open in new tab Detection of (+)-trans-Sabinol, (+)-cis-Sabinol, and (+)-Sabinone Thujone biosynthesis is proposed to start with (+)-sabinene synthase activity (Foster et al., 2013) and proceed through the intermediates (+)-trans- or (+)-cis-sabinol and (+)-sabinone (Fig. 1). We synthesized authentic standards of (+)-trans- and (+)-cis-sabinol and (+)-sabinone to support detection of these compounds in WRC. NMR spectra were recorded to determine the absolute configuration of the (+)-sabinol 3-hydroxyl group (Supplemental Fig. S1). (+)-trans- and (+)-cis-Sabinol and (+)-sabinone could be separated by GC, and their MS patterns were obtained (Supplemental Fig. S2). We detected (+)-trans-sabinol, but not (+)-cis-sabinol, and (+)-sabinone in WRC tissue extracts (Fig. 2). Figure 2. Open in new tabDownload slide Detection of (+)-trans sabinol and (+)-sabinone in WRC green bark and foliage tissue extracts. Shown are total ion chromatograms (TICs) of green bark (A) and foliage (B) extracts. Insets represent magnifications of TIC traces between 12.2 and 14 min. Extracts were separated on a DB1 column. (+)-Sabinene (peak 1), α-thujone (peak 2), and (+)-trans-sabinol (peak 3) were detected in bark and foliage tissue. In addition, (+)-sabinone (peak 4) was detected in bark tissue. (+)-cis-Sabinol was not detected. Figure 2. Open in new tabDownload slide Detection of (+)-trans sabinol and (+)-sabinone in WRC green bark and foliage tissue extracts. Shown are total ion chromatograms (TICs) of green bark (A) and foliage (B) extracts. Insets represent magnifications of TIC traces between 12.2 and 14 min. Extracts were separated on a DB1 column. (+)-Sabinene (peak 1), α-thujone (peak 2), and (+)-trans-sabinol (peak 3) were detected in bark and foliage tissue. In addition, (+)-sabinone (peak 4) was detected in bark tissue. (+)-cis-Sabinol was not detected. Identification of P450 Candidate Genes for Functional Characterization In previous work, we identified in the transcriptome of WRC several putative P450 candidates for thujone biosynthesis, including CYP750B1 (Foster et al., 2013). We reanalyzed the WRC transcriptome assembly, with focus on P450s and their annotation according to established clans, families, and nomenclature (Nelson, 2006). Our analysis identified WRC P450 sequences corresponding to most of the plant P450 clans; except, no WRC P450s were found for clans 710, 711, and 746. All currently known plant P450s with functions as monoterpene oxidases are members of the CYP71 clan (Hamberger and Bak, 2013). We, therefore, targeted WRC P450 contigs falling into families of this clan for additional analysis. We found a total of 26 different contigs belonging to the CYP71 clan (Fig. 3). Within the CYP71 clan, WRC sequences clustered with families representing P450s of well-defined functions in biological processes, such as phenylpropanoid biosynthesis (CYP73 and CYP98 families), flavonoid biosynthesis (CYP75 family), gibberellic acid biosynthesis (CYP701 family), chlorophyll oxidation (CYP89 family), and fatty acid oxidation (CYP77 family). Other WRC P450 sequences fell into families CYP76 (oxidation of terpenoids, fatty acids, lignans, and xenobiotics), CYP81 (oxidation of flavonoids, glucosinolates, and xenbiotics), CYP93 (oxidation of flavonoids and glucosinolates), and CYP736 (glucosinolate metabolism). We did not find WRC representatives of lineage-specific families, such as CYP705, CYP719, and CYP726, members of family CYP84, a family involved in S-lignin biosynthesis, and pollen-specific CYP703. Also, no WRC sequences were identified that cluster within the CYP71 family involved in terpenoid metabolism. Figure 3. Open in new tabDownload slide Phylogeny of WRC P450 contigs and full-length sequences of the CYP71 clan and their association with CYP450 families of the CYP71 clan. Nine different full-length P450 sequences were annotated according to P450 nomenclature standards (Nelson, 2006). Phylogenetic relationship was analyzed using the Parsimony method based on a ClustalW protein sequence alignment in MEGA. Quality of the tree was analyzed by bootstrapping. Branches with bootstrap support lower than 50% are collapsed in a strict consensus tree. Figure 3. Open in new tabDownload slide Phylogeny of WRC P450 contigs and full-length sequences of the CYP71 clan and their association with CYP450 families of the CYP71 clan. Nine different full-length P450 sequences were annotated according to P450 nomenclature standards (Nelson, 2006). Phylogenetic relationship was analyzed using the Parsimony method based on a ClustalW protein sequence alignment in MEGA. Quality of the tree was analyzed by bootstrapping. Branches with bootstrap support lower than 50% are collapsed in a strict consensus tree. Four WRC P450 sequences, including the full-length CYP750B1, fell into the gymnosperm-specific CYP750 family of unknown functions (Fig. 3). Transcripts abundance of CYP750B1 was previously shown to correlate with transcripts of (+)-sabinene synthase (Foster et al., 2013), which made this sequence an interesting candidate for functional characterization and testing of a possible role in thujone biosynthesis. In addition, we targeted eight WRC full-length CYP76 family members (Fig. 3) for functional characterization that represent the highest sequence similarity with known terpenoid-metabolizing enzymes, such as peppermint (Mentha × piperita) limonene-3-hydroxylase (CYP71D13; Lupien et al., 1999), strawberry (Fragaria vesca) pinene-10-hydroxylase (Aharoni et al., 2004), or Madagascar periwinkle (Catharanthus roseus) geraniol 10-hydroxylase (CYP76B6; Collu et al., 2001). These eight WRC CYP76 sequences belong to subfamily CYP76AA, with the sago palm (Cycas rumphii) expressed sequence tag CYP76AA1 as its first described member (National Center Biotechnology Information [NCBI] gene identification no. 27916177), and subfamily CYP76Z, with Sitka spruce (UniGene CYP76Z1) as its first described member (NCBI gene identification no. 1706841). To the best of knowledge, functions for members of the CYP76AA and CYP76Z subfamilies have not yet been identified in any species. Functional Characterization of CYP750B1 and CYP76AA25 as (+)-Sabinene 3-Hydroxylases Full-length coding sequences for CYP750B1, CYP76AA20, CYP76AA21, CYP76AA22, CYP76AA23, CYP76AA24, CYP76AA25, CYP76AA26, and CYP76Z2 were retrieved directly from the transcriptome assembly (for six of nine P450s) or obtained by additional 5′RACE and 3′RACE cloning based on partial contig sequences (for three of nine P450s). P450s were individually expressed from the pESC-LEU2d-u plasmid in yeast (Saccharomyces cerevisiae) BY4741 cells containing a genomically integrated plant cytochrome P450 reductase (lodgepole pine [Pinus contorta] cytochrome P450 reductase [LpCPR]). Assays were performed with microsomal membrane fractions of the transformed yeast cells containing 0.55 to 0.75 mg mL−1 of total protein, corresponding to 0.8 to 5.4 μg of P450 protein per assay, which was determined by the CO difference spectra analysis. LpCPR activity of the microsomal membrane preparations was confirmed by cytochrome c reduction assay. For example, LpCPR activity was 1.98 and 1.92 μmol min−1 mg−1 of protein for microsomal membrane preparations containing CYP750B1 and CYP76AA25, respectively. Each of the nine expressed P450 proteins was tested against a panel of 19 different substrates, including eight different monoterpenes, seven different sesquiterpenes, and four nonterpenoid substrates (Table II). Six of nine P450 proteins (CYP76AA20, CYP76AA21, CYP76AA22, CYP76AA23, CYP76AA25, and CYP750B1) were active with one or several of the substrates tested: CYP76AA20 was active with the herbicide isoproturon, CYP76AA21 and CYP76AA22 were active with the monoterpene (S)-limonene and isoproturon, and CYP76AA23 was active with the sesquiterpenes cadinene and cedrene and isoproturon. None of the terpenoid reaction products could be detected in WRC tissue extracts. Two P450 proteins, CYP76AA25 and CYP750B1, were active with (+)-sabinene (Table II). Relative catalytic activity of CYP76AA25 and CYP750B1 with different monoterpene, sesquiterpene, and nonterpenoid substrates Table II. Relative catalytic activity of CYP76AA25 and CYP750B1 with different monoterpene, sesquiterpene, and nonterpenoid substrates Substrate . CYP76AA25 k cat . CYP750B1 k cat . % Monoterpenes  (+)-Sabinene 100a 100b  (R)-Limonene 0 0  (S)-Limonene ndc ndc  3-Carene 0 0  Geraniol 0 0  Linalool 0 0  Nerol ndc 0  Terpinolene 0 0 Sesquiterpenes  Farnesene 29 0  Cadinene 0 0  Caryophyllene 0 0  Cedrene 0 0  Germacrene 0 0  Humulene 0 0  Longifolene 0 0 Nonterpenoid  Naringenine 0 0  Isoproturon 5 0  7-Methoxycoumarine 0 0  Matairesinol 0 0 Substrate . CYP76AA25 k cat . CYP750B1 k cat . % Monoterpenes  (+)-Sabinene 100a 100b  (R)-Limonene 0 0  (S)-Limonene ndc ndc  3-Carene 0 0  Geraniol 0 0  Linalool 0 0  Nerol ndc 0  Terpinolene 0 0 Sesquiterpenes  Farnesene 29 0  Cadinene 0 0  Caryophyllene 0 0  Cedrene 0 0  Germacrene 0 0  Humulene 0 0  Longifolene 0 0 Nonterpenoid  Naringenine 0 0  Isoproturon 5 0  7-Methoxycoumarine 0 0  Matairesinol 0 0 a 100% is 1.30 min−1. b 100% is 0.20 min−1. c Trace activity of <0.1% substrate conversion in standard assays (k cat) could not be determined. Open in new tab Table II. Relative catalytic activity of CYP76AA25 and CYP750B1 with different monoterpene, sesquiterpene, and nonterpenoid substrates Substrate . CYP76AA25 k cat . CYP750B1 k cat . % Monoterpenes  (+)-Sabinene 100a 100b  (R)-Limonene 0 0  (S)-Limonene ndc ndc  3-Carene 0 0  Geraniol 0 0  Linalool 0 0  Nerol ndc 0  Terpinolene 0 0 Sesquiterpenes  Farnesene 29 0  Cadinene 0 0  Caryophyllene 0 0  Cedrene 0 0  Germacrene 0 0  Humulene 0 0  Longifolene 0 0 Nonterpenoid  Naringenine 0 0  Isoproturon 5 0  7-Methoxycoumarine 0 0  Matairesinol 0 0 Substrate . CYP76AA25 k cat . CYP750B1 k cat . % Monoterpenes  (+)-Sabinene 100a 100b  (R)-Limonene 0 0  (S)-Limonene ndc ndc  3-Carene 0 0  Geraniol 0 0  Linalool 0 0  Nerol ndc 0  Terpinolene 0 0 Sesquiterpenes  Farnesene 29 0  Cadinene 0 0  Caryophyllene 0 0  Cedrene 0 0  Germacrene 0 0  Humulene 0 0  Longifolene 0 0 Nonterpenoid  Naringenine 0 0  Isoproturon 5 0  7-Methoxycoumarine 0 0  Matairesinol 0 0 a 100% is 1.30 min−1. b 100% is 0.20 min−1. c Trace activity of <0.1% substrate conversion in standard assays (k cat) could not be determined. Open in new tab Both CYP76AA25 and CYP750B1 catalyzed the 3-hydroxylation of (+)-sabinene and produced stereoselectively (+)-trans-sabin-3-ol but not (+)-cis-sabinol (Fig. 4). For CYP750B1, (+)-sabinene is the preferred substrate (K m = 110 ± 0.3 μm); a very minor conversion was detected with (S)-limonene. In addition to (+)-sabinene (K m = 299 ± 0.2 μm), CYP76AA25 also converted the sesquiterpene farnesene (K m = 10.3 ± 0.2 μm) and the herbicide isoproturon (K m = 30.5 ± 0.4 μm; Figs. 5 and 6). Minor conversion of nerol and (S)-limonene by CYP76AA25 was also detected. These results revealed CYP750B1 and CYP76AA25 as (+)-sabinene 3-hydroxylases producing stereoselectively (+)-trans-sabinol. In contrast to CYP76AA25, CYP750B1 seems to be more selective for (+)-sabinene as a substrate (Table II). Figure 4. Open in new tabDownload slide Conversion of (+)-sabinene (1) by CYP750B1 and CYP76AA25 producing stereoselectively (+)-trans-sabinol (2). Shown are total ion chromatograms of (+)-sabinene conversion by CYP750B1 (A), (+)-sabinene conversion by CYP76AA25 (B), empty vector control reactions (C), and authentic standards (D). Peaks correspond to (+)-sabinene (1), (+)-trans-sabinol (2), (+)-sabinone (3), and (+)-cis-sabinol (4). Insets show the mass fragmentation pattern of CYP750B1 assay product peak 2 (E) and CYP76AA25 assay product peak 2 (F) coeluting with the (+)-trans-sabinol authentic standard (G). Enzyme and control reactions were incubated for 1 h at 30°C. Pentane extracts of assays and standards were separated on a DB1 column. Rt, Retention time. Figure 4. Open in new tabDownload slide Conversion of (+)-sabinene (1) by CYP750B1 and CYP76AA25 producing stereoselectively (+)-trans-sabinol (2). Shown are total ion chromatograms of (+)-sabinene conversion by CYP750B1 (A), (+)-sabinene conversion by CYP76AA25 (B), empty vector control reactions (C), and authentic standards (D). Peaks correspond to (+)-sabinene (1), (+)-trans-sabinol (2), (+)-sabinone (3), and (+)-cis-sabinol (4). Insets show the mass fragmentation pattern of CYP750B1 assay product peak 2 (E) and CYP76AA25 assay product peak 2 (F) coeluting with the (+)-trans-sabinol authentic standard (G). Enzyme and control reactions were incubated for 1 h at 30°C. Pentane extracts of assays and standards were separated on a DB1 column. Rt, Retention time. Figure 5. Open in new tabDownload slide Farnesene conversion by CYP76AA25. Shown are total ion chromatograms of conversion of farnesene (1) in assays with CYP76AA25 (A) and lack of conversion of farnesene in negative control assays without CYP76AA25 after 1 h of incubation at 30°C and separation of the pentane extract on a DB1 column (B). The reaction product (2) formed could not be identified with any of our available authentic standards. Based on mass fragmentation pattern (C), this product is similar or identical to 2,6,10-trimethyldodeca-2,6,10-trienal. Rt, Retention time. Figure 5. Open in new tabDownload slide Farnesene conversion by CYP76AA25. Shown are total ion chromatograms of conversion of farnesene (1) in assays with CYP76AA25 (A) and lack of conversion of farnesene in negative control assays without CYP76AA25 after 1 h of incubation at 30°C and separation of the pentane extract on a DB1 column (B). The reaction product (2) formed could not be identified with any of our available authentic standards. Based on mass fragmentation pattern (C), this product is similar or identical to 2,6,10-trimethyldodeca-2,6,10-trienal. Rt, Retention time. Figure 6. Open in new tabDownload slide Isoproturon conversion by CYP76AA25. Shown are total ion chromatograms (TICs) and extracted ion chromatograms (EICs; 193) from liquid chromatography-MS analyses. A and B, Product (1) formed during conversion of isoproturon (2) by CYP76AA25. C and D, Negative control without CYP76AA25. E, Mass spectrum of the product with the major mass ion adducts 193 (M + H) and 215 (M + Na). Assays were performed with 1 h of incubation at 30°C. Reaction product (retention time = 9.32 min) of molecular mass 192 is in agreement with demethylation to N-(4-isopropylphenyl)-N'-methylurea, a known product for CYP450-catalyzed isoproturon conversion (Robineau et al., 1998). Figure 6. Open in new tabDownload slide Isoproturon conversion by CYP76AA25. Shown are total ion chromatograms (TICs) and extracted ion chromatograms (EICs; 193) from liquid chromatography-MS analyses. A and B, Product (1) formed during conversion of isoproturon (2) by CYP76AA25. C and D, Negative control without CYP76AA25. E, Mass spectrum of the product with the major mass ion adducts 193 (M + H) and 215 (M + Na). Assays were performed with 1 h of incubation at 30°C. Reaction product (retention time = 9.32 min) of molecular mass 192 is in agreement with demethylation to N-(4-isopropylphenyl)-N'-methylurea, a known product for CYP450-catalyzed isoproturon conversion (Robineau et al., 1998). Transcript Profiles of CYP750B1 But Not CYP76AA25 Correlate with Thujone Accumulation Transcript abundance of CYP750B1 and CYP76AA25 was measured first in two contrasting WRC genotypes, 5038 and 5313. Genotype 5038 represents a glandless, low-thujone phenotype that does not accumulate monoterpenes as shown in Foster et al. (2013). For comparison, genotype 5131 represents a typical phenotype with foliar glands for accumulation of thujone and other monoterpenes. Extreme differences in monoterpene content of foliage of 5038 and 5131 are shown in Figure 7. Patterns of high transcript abundance of (+)-sabinene synthase and CYP750B1 in 5131 and their very low levels in 5038 correlate well with the presence and the absence of thujone in 5131 and 5038, respectively (Fig. 7), confirming previous observations (Foster et al., 2013). In addition, we found that transcript abundance of CYP76AA25 did not correlate with differences in thujone content in 5131 and 5038 (Fig. 7). Figure 7. Open in new tabDownload slide Contrasting patterns of α-thujone accumulation and transcript accumulation in WRC glandless genotype 5038 foliage, which does not accumulate monoterpenes, and monoterpenoid-accumulating genotype 5131 foliage. A, Total monoterpene and α-thujone content in foliage of genotypes 5038 and 5131. Monoterpenes, including thujones, were not detectable (n.d.) in glandless 5038. B, Transcript accumulation for CYP76AA25, CYP750B1, and (+)-sabinene synthase quantified by qRT-PCR. Data are shown as normalized expression against EF-α expression. Error bars show se based on three biological replicates. *, Significant differences of transcript abundance between 5038 and 5131 as determined by t test (P < 0.05). Figure 7. Open in new tabDownload slide Contrasting patterns of α-thujone accumulation and transcript accumulation in WRC glandless genotype 5038 foliage, which does not accumulate monoterpenes, and monoterpenoid-accumulating genotype 5131 foliage. A, Total monoterpene and α-thujone content in foliage of genotypes 5038 and 5131. Monoterpenes, including thujones, were not detectable (n.d.) in glandless 5038. B, Transcript accumulation for CYP76AA25, CYP750B1, and (+)-sabinene synthase quantified by qRT-PCR. Data are shown as normalized expression against EF-α expression. Error bars show se based on three biological replicates. *, Significant differences of transcript abundance between 5038 and 5131 as determined by t test (P < 0.05). To extend this analysis, we compared thujone levels and transcript abundance in four different second generation inbred lines (S2), three of which (872, 873, and 875) had drastically reduced α-thujone levels compared with a reference line 8320 (Fig. 8). Comparison of transcript abundance in lines 872, 873, and 875 relative to line 8320 showed significantly reduced levels of CYP750B1 transcripts but not CYP76AA25 transcripts in the low α-thujone lines (Fig. 8). These results together with results from the characterization of P450 enzyme functions support a specific role of CYP750B1 in thujone biosynthesis in WRC. A function of CYP76AA25 in thujone biosynthesis is ambiguous. Figure 8. Open in new tabDownload slide Thujone, monoterpene, and transcript accumulation in WRC foliage of four different second generation inbred lines, including three lines with low-thujone phenotypes. A, Content of α-thujone and total content of other monoterpenes in three low-thujone lines (872, 873, and 875) and line 8,320. B, Transcript accumulation for CYP76AA25, CYP750B1, and sabinene synthase quantified by qRT-PCR and normalized to EF-α expression (shown as x fold difference in low-α-thujone lines [872, 873, and 875] relative to line 8,320). Error bars show se based on three biological replicates per line. SS, Sabinene synthase; *, significant differences (t test, P < 0.05) between low-α-thujone lines 872, 873, and 875 and line 8320. Figure 8. Open in new tabDownload slide Thujone, monoterpene, and transcript accumulation in WRC foliage of four different second generation inbred lines, including three lines with low-thujone phenotypes. A, Content of α-thujone and total content of other monoterpenes in three low-thujone lines (872, 873, and 875) and line 8,320. B, Transcript accumulation for CYP76AA25, CYP750B1, and sabinene synthase quantified by qRT-PCR and normalized to EF-α expression (shown as x fold difference in low-α-thujone lines [872, 873, and 875] relative to line 8,320). Error bars show se based on three biological replicates per line. SS, Sabinene synthase; *, significant differences (t test, P < 0.05) between low-α-thujone lines 872, 873, and 875 and line 8320. In Vitro Conversion of (+)-trans-Sabinol to (+)-Sabinone by WRC Tissue Extracts Given that (+)-trans-sabinol, but not (+)-cis-sabinol, was present in WRC tissue extracts (Fig. 2) and after the identification of (+)-trans-sabinol, but not (+)-cis-sabinol, as the product of CYP750B1 (Fig. 4), we tested if, indeed, (+)-trans-sabinol could serve as the substrate in the next step of thujone biosynthesis, presumably involving (+)-sabinol dehydrogenase activity to yield (+)-sabinone (Fig. 1). Using cell-free tissue extracts of soluble proteins, we tested the NADH-dependent conversion of both (+)-trans- and (+)-cis-sabinol. We detected NADH-dependent conversion of (+)-trans-sabinol but not (+)-cis-sabinol to sabinone (Fig. 9). This result together with lines of evidence from tissue metabolite analysis [presence of (+)-trans-sabinol but not (+)-cis-sabinol] and stereoselective formation of (+)-trans-sabinol by CYP750B1 strongly support a role of (+)-trans-sabinol as the biological intermediate in thujone biosynthesis in WRC. Figure 9. Open in new tabDownload slide NADH-dependent conversion of (+)-trans-sabinol but not (+)-cis-sabinol by cell-free protein extracts of WRC tissue. (+)-trans-Sabinol is converted to (+)-sabinone by cell-free protein extracts. Assays containing soluble protein extracts were incubated for 90 min at 30°C with either (+)-trans-sabinol or (+)-cis-sabinol with and without NADH. Assays products were extracted, separated on a DB1 column, and analyzed by GC-MS. Sabinone production was confirmed by comparing product peak retention time and mass fragmentation with authentic standards, and areas were calculated for ion 108 (mass to charge ratio). Shown are average values for quadruplicate experiments. *, Significant differences with and without NADH added in assays with (+)-trans-sabinol as determined by t test (P < 0.05). Figure 9. Open in new tabDownload slide NADH-dependent conversion of (+)-trans-sabinol but not (+)-cis-sabinol by cell-free protein extracts of WRC tissue. (+)-trans-Sabinol is converted to (+)-sabinone by cell-free protein extracts. Assays containing soluble protein extracts were incubated for 90 min at 30°C with either (+)-trans-sabinol or (+)-cis-sabinol with and without NADH. Assays products were extracted, separated on a DB1 column, and analyzed by GC-MS. Sabinone production was confirmed by comparing product peak retention time and mass fragmentation with authentic standards, and areas were calculated for ion 108 (mass to charge ratio). Shown are average values for quadruplicate experiments. *, Significant differences with and without NADH added in assays with (+)-trans-sabinol as determined by t test (P < 0.05). DISCUSSION Monoterpenoids and in particular, thujones have been shown to influence deer-browsing resistance in WRC (Vourc’h et al., 2001, 2002; Kimball et al., 2012). We recently identified a monoterpene synthase gene, (+)-sabinene synthase, encoding the first step in thujone biosynthesis in WRC (Foster et al., 2013). Here, we report the cloning and functional characterization of a gymnosperm-specific P450 enzyme CYP750B1, which functions as a (+)-sabinene 3-hydroxylase and catalyzes the second of four steps in the thujone pathway (Fig. 1). To the best of our knowledge, this is the first report of a functionally characterized P450 gene of monoterpene biosynthesis in any gymnosperm. CYP750B1 is also, to our knowledge, the first functionally characterized member of the apparently gymnosperm-specific CYP750 family. CYP750B1 produces stereoselectively (+)-trans-sabinol but not (+)-cis-sabinol, which was shown with a set of newly synthesized authentic standards. A role of (+)-trans-sabinol as a relevant intermediate in thujone biosynthesis was substantiated with enzyme assays using WRC tissue extracts, which selectively converted (+)-trans-sabinol but not (+)-cis-sabinol to (+)-sabinone. A P450 enzyme in thujone biosynthesis was originally shown by Karp and Croteau (1982) in common garden sage. Thujone biosynthesis in common garden sage, an angiosperm, and WRC, a gymnosperm, seems to involve different stereoisomers: (+)-cis- and (+)-trans-sabinol, respectively. A sabinene hydroxylase has not yet been cloned and characterized from common garden sage or any other angiosperm species, and it is not known to which P450 family such a gene would belong. Given that WRC CYP750B1 is a member of the gymnosperm-specific CYP750 family, it is quite possible that P450s of different stereospecificities in the formation (+)-trans- and (+)-cis-sabinol evolved independently in WRC and common garden sage. Our initial functional characterization of a set of nine P450 candidates revealed two P450s, CYP750B1 and CYP76AA25, that were able to convert (+)-sabinene to (+)-trans-sabinol. Several lines of evidence supported a role of CYP750B1 in thujone biosynthesis, whereas such a role was not supported for CYP76AA25. On the biochemical level, CYP750B1 seems to have narrow substrate specificity, with (+)-sabinene being the only substrate identified in a panel of 19 different compounds. On the molecular level, our work made use of two highly unique phenotype resources with regard to conifer monoterpene chemistry: (1) a clonally propagated glandless genotype (line 5038) and (2) a set of three different low-thujone S2 lines (872, 873, and 875). Foliar glands are the major site of monoterpene biosynthesis and accumulation in WRC. The glandless clone does not accumulate monoterpenes, consistent with a lack of expression of monoterpene biosynthesis (Russell and Ferguson, 2008; Russell and Yanchuk, 2012; Foster et al., 2013). The low-thujone S2 lines 872, 873, and 875 seem to be blocked more specifically in thujone biosynthesis. The phenotypes of these materials contrast with the high-monoterpene and high-thujone chemistry that is typical for WRC. Using these materials, we found a strong correlation of CYP750B1 transcript abundance with thujone accumulation in all of the different genotypes investigated. Transcript accumulation of both CYP750B1 and sabinene synthase was extremely low in the glandless, thujone-nonaccumulating line (Fig. 7). In all three low-thujone S2 lines, CYP750B1 transcript levels were significantly reduced compared with a reference line, whereas in this phenotype, sabinene synthase transcripts were not significantly reduced, which is in agreement with the presence of sabinene (Fig. 8). In contrast to CYP750B1, transcript abundance of CYP76AA25 did not correlate with thujone accumulation or transcript accumulation of sabinene synthase, and CYP76AA25 transcript abundance was unaffected in the glandless genotype and the low-thujone S2 lines (Figs. 7 and 8). Hence, in contrast to CYP750B1, there was no supporting molecular evidence for a role of CYP76AA25 in thujone biosynthesis. In the context of naturally occurring variation of monoterpenoid and thujone accumulation in WRC, this work and the work by Foster et al. (2013) suggest that differences in transcript abundance of sabinene synthase and CYP750B1 is an important and potentially causative component for this chemophenotypic variation. Transcript abundance of both the sabinene synthase and CYP750B1 genes correlates with thujone accumulation and the presence and absence of glands, and both of these traits contribute to deer-browsing resistance. These genes may be used as transcript biomarkers to screen large numbers of WRC accessions in a high-throughput and automated fashion in addition to using existing methods of screening for chemical and histology phenotype. In future work, it will be important to identify the factors that control expression of these two genes and possibly other genes in the thujone pathway as well. Using the glandless WRC phenotype, it might be possible to identify genes or mechanisms that control development of secretory structures for monoterpene biosynthesis and accumulation as well as gland-specific gene expression of thujone biosynthesis. Our results conclusively showed that the gymnosperm-specific cytochrome P450 enzyme CYP750B1 catalyzes the stereospecific monoterpene hydroxylation of (+)-sabinene, which is a critical step in the biosynthesis of α- and β-thujone, major defense compounds for deer-browsing resistance in WRC. To date, only a few gymnosperm draft genomes have been published (Birol et al., 2013; Nystedt et al., 2013; De La Torre et al., 2014; Neale et al., 2014), and a genome sequence for WRC is not yet available. Therefore, in the absence of RNA interference lines, it is not possible to conclude if other genes exist in WRC with similar or overlapping functions to those of CYP750B1. It is important to note that, in other gymnosperm trees, it has been shown that specific biochemical functions in terpenoid defense pathways are covered with several gene copies, such as multiple functionally similar monoterpene synthases (Hall et al., 2011; Roach et al., 2014) or multiple P450s of diterpene resin acid biosynthesis (Hamberger et al., 2011). MATERIALS AND METHODS Plant Material and Terpenes All WRC (Thuja plicata) plant material was from the breeding program of the British Columbia Ministry of Forests, Lands and Natural Resource Operations at Cowichan Lake Research Station (CLRS). Four second generation inbred seedling lines (S2; 872, 873, 875, and 8320) and three clones (genotypes 5309, 5131, and 5038) were used for this study. S2 reference line 8320 has normal accumulation of thujone and other monoterpenes, whereas S2 lines 872, 873, and 875 accumulate low amounts of thujone and normal amounts of sabinine and other monoterepenes. Reference genotypes 5309 and 5131 have above-normal thujone accumulation, whereas 5038, a glandless genotype, has no detectable accumulation of monoterpenoids, including thujone. S2 lines were sown in March of 2013 in a greenhouse, and foliage was harvested from three individual seedlings per line in November of 2014 at CLRS for metabolite and transcript analyses. The clones were propagated as rooted cuttings in 2009 in a greenhouse and transferred from CLRS to the University of British Columbia greenhouse facility in 2011. Sapling trees were moved outside in October of 2013, maintained under outside conditions in 2-gallon pots, and harvested in September of 2014 for metabolite and transcript analyses. Monoterpene standards were purchased from Chromadex and Sigma-Aldrich. Sabinyl acetate was from Extrasynthese. (+)-cis- and (+)-trans-sabinol and (+)-sabinone were synthesized as described below. Synthesis and Confirmation of (+)-cis-Sabinol, (+)-trans-Sabinol, and (+)-Sabinone (+)-cis-Sabinol, (1S,3S,5S)-1-isopropyl-4-methylidenebicyclo[3.1.0]hexan-3-ol, was synthesized following Umbreit and Sharpless (1977) and Sirisoma et al. (2001). In detail, (+)-sabinene (70 µL; 0.43 mmol) was dissolved in dichloromethane (5 mL); SeO2 (0.0027 mmol) and t-butylhydroperoxide (240 µL) were added, and the reaction was stirred at room temperature overnight. After the addition of more SeO2 (0.009 mmol), stirring was continued for another 2 h. Benzene (1 mL) was added, and stirring continued for 2 h; then, the reaction was evaporated to near dryness, diluted with diethyl ether (10 mL), washed with 10% KOH, saturated with NaCl, dried (MgSO4), filtered, and evaporated to near dryness again. The remaining liquid was applied to a silica gel column (10 g), the flask was rinsed with petroleum ether (500 μL), and the petroleum ether was loaded onto the same column. The column was washed with 8 × 8 mL of petroleum ether, 8 × 5 mL of 4% (v/v) ether/ethyl acetate, and 1 × 5 mL of 5% (v/v) petroleum ether/ethyl acetate. (+)-cis-Sabinol was eluted in 7 × 5 mL of 5% (v/v) petroleum ether/ethyl acetate. Eluates were pooled, and one-half of the eluate was evaporated to dryness to give 15 mg (0.1 mmol) of the target compound as one single isomer as determined by 1H and 13C NMR. The 1H (CDCl3, 400 MHz, reference 7.26 ppm) data are: δ 4.98 (d, 1H, J = 2.3 Hz, H-10a), 4.87 (d, 1H, J = 2 Hz, H-10b), 4.25 to 4.16 (bs, 1H, H-3), 2.26 (dd, 1H, J = 12.2 Hz, J = 7.6 Hz, H-2a), 1.70 (dd, 1H, J = 8.3 Hz, J = 3.4 Hz, H-5), 1.57 (ddd, J = 12.2 Hz, J = 8.3 Hz, J = 1.5 Hz, H-2b), 1.40 (m, 1H, H-7), 0.95 (d, 1H, J = 6.8 Hz, H-8), 0.89 (d, 1H, J = 6.8 Hz, H-9), 0.65 (ddd, J = 8.3 Hz, J = 4.9 Hz, J = 1 Hz, H6a), and 0.57 (dd, 1H, J = 4.8 Hz, J = 3.4 Hz, H-6b). 13C (CDCl3, 100 MHz, reference 77.0 ppm): δ 156.11 (C-4), 101.96 (C-10), 71.42 (C-3), 37.63 (C-2), 33.43 (C-1), 32.9 (C-7), 27.87 (C-5), 19.61 (C-8 or C-9), 19.45 (C-9 or C8), and 17.79 (C-6). (+)-trans-Sabinol, (1S,3R,5S)-1-isopropyl-4-methylidenebicyclo[3.1.0]hexan-3-ol, was obtained by basic hydrolysis of sabinyl acetate (40 mg) in ethanolic KOH (10%; Garside et al., 1969). After a 10-min incubation at room temperature, 3 parts of water was added to the solution, and the compound was extracted in an equal volume of pentene. The pentane phase was directly used or further purified by silica gel column chromatography (1.5 g). (+)-trans-Sabinol was purified by washing the column using pentene methyl tertiary-butyl ether (MTBE) gradient (3 mL of pentene, 1 mL of 90% pentene/10% MTBE, and 1 mL of 80% pentene/20% MTBE). (+)-trans-Sabinol was eluted in 70% pentene/30% MTBE, and the solvent was removed to give 27 mg (0.177 mmol; yield of 68%) of (+)-trans-sabinol, which was identified by 1H and 13C NMR. The 1H NMR (CDCl3, 400 MHz, reference 7.26 ppm) data are: δ 4.99 (s, 1H, H-10a), 4.93 (s, 1H, H-10b), 4.43 (d, 1H, J = 7.4 Hz, H-3), 2.05 (ddd, 1H, J = 13.9 Hz, J = 7.4 Hz, J = 2.1 Hz, H-2a), 1.72 (d, 1H, J = 13.8 Hz, H-2b), 1.64 (dd, 1H, J = 8.7 Hz, J = 3.4 Hz, H-5), 1.43 (hep, 1H, J = 6.8 Hz, H-7), 1.04 (t, 1H, J = 3.7 Hz, H-6a), 0.92 (d, 3 H, J = 6.8 Hz, H-8 or -9), 0.87 (d, J = 6.8 Hz, H-9 or -8), and 0.80 (ddd, 1H, J = 8.5 Hz, J = 4.2 Hz, J = 2.1 Hz, H-6b). 13C (CDCl3, 100 MHz, reference 77.0 ppm): 157.20 (C-4), 106.69 (C-10), 75.08 (C-3), 37.67 (C-1), 37.20 (C-2), 32.56 (C-7), 28.93 (C-5), 19.99 (C-6), 19.72 (C-8 or C-9), and 19.55 (C-9 or C-8). (+)-Sabinone, (1S,5S)-1-isopropyl-4 methylidenebicyclo[3.1.0]hexan-3-on, was obtained by oxidation of both (+)-cis- and (+)-trans-sabinol following Dess and Martin (1983). In detail, (+)-sabinol (15 mg; 0.1 mmol) was dissolved in dichloromethane (2 mL), a slight excess of Dess-Martin periodinane reagent (47 mg; 0.11 mmol) was added, and the reaction was stirred for 6 h at room temperature. The reaction was diluted with pentane (1 mL) and applied directly to a pentane prewashed silica gel column (3 g). The compound was eluted with a gradient from 0% to 4% acetone in pentane. Product-containing fractions were pooled, concentrated in volume, and then diluted with pentane or acetone for additional analysis. (+)-Sabinone was confirmed by NMR. The 1H (600 MHz, acetone-d6, reference 2.05 ppm) data are: δ 5.60 (s, 1H, H-10a), 5.24 (s, 1H, H-10b), 2.50 (dd, 1H, J = 18.9, J = 2.7, H-2a), 2.23 (d, 1H, J = 19, H-2b), 2.14 (dd, 1H, J = 8.5, J = 3.6, H-5), 1.52 (q, 1H, J = 7 Hz, H-6), 1.11 (ddd, 1H, J = 8.3, J = 5.2, J = 2.9, H-6a), 0.99 (d, 3H, J = 6.7 Hz, H-8), 0.95 (d, 3H, J = 6.8 Hz, H-9), and 0.42 (dd, 1H, J = 5 Hz, J = 3.7 Hz, H-6b). 13C (150.9 MHz, acetone-d6, reference 29.84 ppm): δ 205.1 (C-3), 149.4 (C-10), 112.6 (C-4), 41.5 (C-2), 33.1 (C-5), 30.1 (C-1), 26.8 (C-7), 22.0 (C-6), 19.6 (C-9), and 19.5 (C-8). Terpenoid Extraction from WRC Tissue Foliage tips (approximately 2 cm in length) were collected into GC-MS vials containing 1 mL of pentane spiked with isobutyl benzene (5 µg mL−1) as the internal standard. After overnight incubation, samples were centrifuged for 15 min at 1,000g, and the pentane phase was transferred into new vials for GC-MS analysis. GC-MS and Liquid Chromatography-MS Analysis Monoterpenoid analysis was done by GC-MS (Agilent 6890A/5975C and 7890A/7000A). GC conditions were as follows. Injections were done in pulsed splitless mode with inlet temperature at 250°C. Helium was used as a carrier gas. Terpenoid metabolites from enzyme assay extracts were separated on a DB-WAX Column (122-7032; J&W; 30 m × 250 µm; 0.25-µm film thickness) with oven temperature at 40°C for 1 min; temperature was increased by 15°C min−1 to 150°C and 30°C min−1 to 250°C and held for 8 min. Average carrier gas velocity was 33 cm s−1. Metabolites from tissue extracts were separated on a DB1 Capillary Column (122-0132; J&W; 30 m × 250 µm; 0.25-µm film thickness). The oven temperature started at 40°C followed by a 6°C min−1 increase to 160°C and a 40°C min−1 increase to 300°C and ended with a hold for 5 min; the average carrier gas velocity was set to 33 cm s−1. For validation of metabolite identification in tissue extracts, additional analysis was performed on a HP5 Capillary Column (19091S-433; Agilent; 30 m × 250 µm; 0.25-µm film thickness), with oven temperature starting at 40°, increasing by 3°C min−1 to 120°C and 15°C min−1 to 280°C, and finally holding for 5 min; the average carrier gas velocity was set to 38 cm s−1. Terpenoids were identified by (1) comparison with authentic standards, (2) comparison with retention indices (RIs) published for DB1 and HP5 capillary columns searching MassFinder4’s internal databases (massfinder.com/wiki/MassFinder_4) and the National Institute of Standards and Technology (NIST) data collection incorporated into W9N08L database (Wiley), or (3) accessing the NIST compound information at http://webbook.nist.gov/chemistry. For quantification, response factors (Rfs) were established with concentration series of authentic standards. Isobutylbenzene (Sigma-Aldrich) was used as an internal standard for all experiments. Enzymatic conversion of nonterpenoid compounds (isoproturon, naringenine, 7-methoxycoumarine, and matairesinol) was monitored by liquid chromatography-MS. For assays with isoproturon, extracts were separated on a C18 Column (Atlantis T3; Waters; 2.1 [i.d.] × 100 mm; 5-µm pore size) at a flow rate of 0.5 mL min−1 using a water-acetonitrile (ACN) gradient (0–1 min with 5% [v/v] ACN, 1–14 min with 5%–90% ACN, and 14–15 min with 90% ACN) containing 0.2% (v/v) formic acid in the mobile phase. Negative ion electrospray mass spectra of enzyme product and isoproturon were recorded by a coupled MSD-Trap-XCT_Plus (AgilentTechnologies Inc.). Assignment of (+)-Sabinol Stereochemistry Assignment of cis- and trans-configuration was based on the distance of C-6 methylene and C-3 hydroxyl groups (Supplemental Fig. S1). Because the 3-hydroxyl only in the cis isomer is close through space proximity to the C-6 CH2 group, the corresponding protons are shifted significantly to low field in the 1H NMR spectrum (cis, 0.65 ppm for 6a and 0.57 ppm for 6b; trans, 1.08 ppm for 6a and 0.85 ppm for 6b). The assignment agrees with data published by Ohloff et al. (1966) and Mamane et al. (2004). (+)-cis-Sabinol structure was determined by 1H and 13C NMR. A one-dimensional Nuclear Overhauser Effect difference spectroscopy revealed a correlation between H-3 and the syn proton at C-6 for the cis isomer, whereas no correlation between H-3 and any of the C-6 protons was observed for the trans isomer, confirming the trans and cis assignments. RNA Isolation and Measurement of Transcript Abundance RNA was isolated from WRC using Concert Plant Reagent (Invitrogen) following the manufacturer’s protocol for small-scale RNA isolation using 30 mg fresh weight of WRC tissue material. Sugar and other impurities as well as genomic DNA were removed using the Plant RNeasy RNA Mini Kit (Qiagen) and the RNAse Free DNAse Set (Qiagen); 1 µg of RNA was transcribed with Superscript III (Invitrogen) and Oligo(dT)20VN for 60 min at 42°C. The resulting complementary DNA (cDNA) was diluted to 3 ng μL−1. Quantitative PCR analysis was performed in a BioRad CFX96 Real-Time PCR Detection System following the SsoFast EvaGreen protocol (BioRad) and using primers as shown in Supplemental Table S1. Data were analyzed using the LinRegPCR Program (Ruijter et al., 2009) as described in Zifkin et al. (2012). Representative PCR products were purified and sequenced to confirm product specificity. We assessed four reference genes for quantitative reverse transcription (qRT)-PCR analysis in foliage of different WRC genotypes: actin, elongation factor-α (EF-α), glyceraldehyde 3-phosphate dehydrogenase, and RNA-Polymerase III. No variation was observed for EF-α and actin in the different genotypes. Relative transcription abundance of target genes was calculated by normalizing data against actin and EF-α expression. Reassembly of WRC Transcriptome Sequences Generation of 42 m pairs of 75-bp WRC transcriptome sequences using the Illumina Genome Analyzer IIx Platform was described by Foster et al. (2013). For quality control before assembly, we used FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Sequence trimming was performed with Trimmomatic (Lohse et al., 2012); 12 bp were trimmed from the 5′ end of reads. Additional trimming was done at the 3′ end of reads if they fell below a quality score of Q20. A minimum read length of 50 bp was used as a threshold. De novo transcriptomic assembly was performed on 25 m paired trimmed sequences using Trinity assembler (Grabherr et al., 2011). The assembly generated 75,507 contigs with an average length of 907 bp. Identification of P450 Candidate Sequences For the identification of candidate P450 sequences in the dataset of 75,507 transcriptome contigs, we used a set of 38 bait sequences (Supplemental Table S2) extracted from the NCBI Protein Database representing P450 families of all 11 plant P450 clades for a tBlastN search. The identified WRC contigs were consolidated and tested for redundancies and possible short overlapping regions, which were not recognized by the Trinity assembler, using the CLC Contig Assembler Tool. The obtained unique contigs were translated into protein sequences and analyzed for their phylogenetic relation to the bait sequences and each other. All sequence analyses were done using CLC Workbench, with the exception of phylogenetic reconstruction (see below) P450 Full-Length cDNA Cloning RNA was isolated from WRC foliage and reverse transcribed as described above. Amplicons were derived by PCR using gene-specific primer pairs (Supplemental Table S1) and Phusion Polymerase (NEB). Gel-purified (QIAquick; Qiagen) fragments were ligated into pJET2.1 vector (Fermentas), and the resulting plasmids were transformed into Alpha-Select Gold Efficiency Escherichia coli cells (Bioline). Plasmids were purified (QIAprep; Qiagen), and the nucleotide sequences were verified (3730 DNA Analyzer; Applied Biosystems) using Big Dye Reagent (Life Technologies) and vector-specific primer. Phylogenetic Sequence Analysis All phylogenetic analyses were conducted in MEGA5 (Nei and Kumar, 2000; Tamura et al., 2007). Protein sequences were aligned using ClustalW (Thompson et al., 1994). Phylogenetic relationships were reconstructed using maximum likelihood. Bootstrap values above 50% are shown as a percentage next to the branches. Expression of CYP76AA25 and CYP750B1 in Yeast A cassette for Uracil-Specific Excision Reagent (USER)-based cloning was inserted into multicloning site1 (MSC1) of the pESC-LEU2d vector (Ro et al., 2006) using the BamHI and HindIII site based on Hamann and Møller (2007). A PacI restriction site close to MCS2 was silenced by a 2-nucleotide exchange to enable USER cloning and obtain pESC-LEU2d-u (Supplemental Table S1). CYP76AA25 and CYP750B1 cDNAs were cloned into PacI, and Nb.BbvCI (NEB) digested pESC-LEU2d-u after amplification with T7 polymerase using USER-compatible primers (Supplemental Table S1) and treatment with USER enzyme (NEB). The resulting constructs were transformed into yeast (Saccharomyces cerevisiae) strain BY4741. The yeast BY4741 also contained a conifer (lodgepole pine [Pinus contorta]) CPR (accession no. KJ914574) integrated in the genome (BY4741:LpCPR). Proteins were expressed, and microsomal fractions containing the recombinant protein were isolated as described previously (Pompon et al., 1996; Ro et al., 2005; Hamberger et al., 2011). The amount of expressed P450 was calculated following the method by Omura and Sato (1964). P450 Enzyme Assays Microsomal preparations (30 µL) were added on ice to 270 µL of reaction mixture containing 50 mm potassium phosphate (pH 7.5), 0.8 mm NADPH, and 100 µm substrate in a GC-MS glass vial. Assays were incubated under constant shaking at 30°C for 1 h for qualitative analysis or variable times of 1 to 180 min for kinetic analysis. Reactions were stopped by adding 300 µL of pentane, rigorous mixing for 1 min, and subsequent freezing at −80°C. After phase separation by centrifugation at 500g for 10 min, the pentane layer was analyzed by GC-MS. Sabinol Dehydrogenase Assays Soluble protein extracts from WRC foliage were obtained using a modified protocol described for common garden sage (Salvia officinalis) monoterpene dehydrogenases (Dehal and Croteau, 1987). In detail, 1 g of fresh frozen foliage tissue (line 5309) was ground into a fine powder under liquid nitrogen and resuspended on ice in extraction buffer (pH 6.5; 250 mm Suc, 20 mm sodium pyrosulfite, 100 mm sodium phosphate, 10 mm sodium ascorbate, 5 mm dithiothreitol [DTT], 1 mm EDTA, and 0.0275 g mL−1 insoluble polyvinylpolypyrrolidone). After centrifugation for 20 min at 27,000g at 4°C, 2.5 mL of supernatant was buffer exchanged into assay buffer (Tris-HCl, pH 7.5 and 2 mm DTT) and cleared of low-M r metabolites on PD-10 Desalting Columns (GE-Healthcare). Enzyme assays were performed with 100 µL of desalted protein extract mixed with 200 µL of assay buffer (Tris-HCl, pH 7.5, and 2 mm DTT), 1 mm NADH, and 100 µm either (+)-trans- or (+)-cis-sabinol. Assays were incubated for 90 min at 30°C. Negative controls were done without NADH. Reactions were stopped by adding 300 µL of pentane, rigorous mixing for 1 min, and subsequent freezing at −80°C. After phase separation by centrifugation at 500g for 10 min, the pentane layer was analyzed by GC-MS. Sequence data from this article can be found in the National Center Biotechnology Information GenBank under accession numbers CYP750B1 (KP004988), CYP76AA20 (KP015848), CYP76AA21 (KP015849), CYP76AA22 (KP015850), CYP76AA23 (KP015851), CYP76AA24 (KP015852), CYP76AA25 (KP015853), CYP76AA26 (KP015854), and CYP76Z2 (KP015855). Supplemental Data The following supplemental materials are available. Supplemental Figure S1. Determination of (+)-sabinol stereochemistry. Supplemental Figure S2. GC-MS separation and detection of (+)-trans- and (+)-cis-sabinol and (+)-sabinone. Supplemental Table S1. Primers used for full-length cDNA isolation, qRT-PCR, and pESC-LEU2d modification. Supplemental Table S2. P450 sequences with accession numbers and P450 family and clan associations used as search baits against the WRC transcriptome. ACKNOWLEDGMENTS We thank Craig Ferguson (CLRS) for generous access to plant materials; Melina Biron, David Kaplan, Alfonso Lara Quesada, and Elizabeth Steves (University of British Columbia) for plant maintenance; David Nelson (University of Tennessee) for naming of P450 genes; Katrin Geissler (University of British Columbia) for the yeast expression cell line; and Karen Reid (University of British Columbia) for outstanding laboratory management. Glossary ACN acetonitrile cDNA complementary DNA CLRS Cowichan Lake Research Station DTT dithiothreitol GC gas chromatography MS mass spectrometry MTBE methyl tertiary-butyl ether NIST National Institute of Standards and Technology qRT quantitative reverse transcription Rf response factor RI retention index WRC Western red cedar LITERATURE CITED Aharoni A , Giri AP, Verstappen FWA, Bertea CM, Sevenier R, Sun Z, Jongsma MA, Schwab W, Bouwmeester HJ ( 2004 ) Gain and loss of fruit flavor compounds produced by wild and cultivated strawberry species . Plant Cell 16 : 3110 – 3131 Google Scholar Crossref Search ADS PubMed WorldCat Anderson AB , Gripenberg J ( 1948 ) Antibiotic substances from the heart wood of Thuja plicata D. Don; the constitution of β-thujaplicin . Acta Chem Scand 2 : 644 – 650 Google Scholar Crossref Search ADS PubMed WorldCat Birol I , Raymond A, Jackman SD, Pleasance S, Coope R, Taylor GA, Yuen MMS, Keeling CI, Brand D, Vandervalk BP, et al. ( 2013 ) Assembling the 20 Gb white spruce (Picea glauca) genome from whole-genome shotgun sequencing data . Bioinformatics 29 : 1492 – 1497 Google Scholar Crossref Search ADS PubMed WorldCat Burney OT , Jacobs DF ( 2011 ) Ungulate herbivory of regenerating conifers in relation to foliar nutrition and terpenoid production . For Ecol Manage 262 : 1834 – 1845 Cankar K , van Houwelingen A, Goedbloed M, Renirie R, de Jong RM, Bouwmeester H, Bosch D, Sonke T, Beekwilder J ( 2014 ) Valencene oxidase CYP706M1 from Alaska cedar (Callitropsis nootkatensis) . FEBS Lett 588 : 1001 – 1007 Google Scholar Crossref Search ADS PubMed WorldCat Chen F , Tholl D, Bohlmann J, Pichersky E ( 2011 ) The family of terpene synthases in plants: a mid-size family of genes for specialized metabolism that is highly diversified throughout the kingdom . Plant J 66 : 212 – 229 Google Scholar Crossref Search ADS PubMed WorldCat Collu G , Unver N, Peltenburg-Looman AMG, van der Heijden R, Verpoorte R, Memelink J ( 2001 ) Geraniol 10-hydroxylase, a cytochrome P450 enzyme involved in terpenoid indole alkaloid biosynthesis . FEBS Lett 508 : 215 – 220 Google Scholar Crossref Search ADS PubMed WorldCat Croteau R , Ketchum REB, Long RM, Kaspera R, Wildung MR ( 2006 ) Taxol biosynthesis and molecular genetics . Phytochem Rev 5 : 75 – 97 Google Scholar Crossref Search ADS PubMed WorldCat De La Torre AR , Birol I, Bousquet J, Ingvarsson PK, Jansson S, Jones SJN, Keeling CI, MacKay J, Nilsson O, Ritland K, et al. ( 2014 ) Insights into conifer giga-genomes . Plant Physiol 166 : 1724 – 1732 Google Scholar Crossref Search ADS PubMed WorldCat Dehal SS , Croteau R ( 1987 ) Metabolism of monoterpenes: specificity of the dehydrogenases responsible for the biosynthesis of camphor, 3-thujone, and 3-isothujone . Arch Biochem Biophys 258 : 287 – 291 Google Scholar Crossref Search ADS PubMed WorldCat Dess DB , Martin JC ( 1983 ) Readily accessible 12-I-5 oxidant for the conversion of primary and secondary alcohols to aldehydes and ketones . J Org Chem 48 : 4155 – 4156 Google Scholar Crossref Search ADS WorldCat Foster AJ , Hall DE, Mortimer L, Abercromby S, Gries R, Gries G, Bohlmann J, Russell J, Mattsson J ( 2013 ) Identification of genes in Thuja plicata foliar terpenoid defenses . Plant Physiol 161 : 1993 – 2004 Google Scholar Crossref Search ADS PubMed WorldCat Gardner JAF , Barton GM, MacLean H ( 1959 ) The polyoxyphenols of western red cedar (thuja plicata donn.) . Can J Chem 37 : 1703 – 1709 Google Scholar Crossref Search ADS WorldCat Garside P , Halsall TG, Hornby GM ( 1969 ) Action of peracetic acid on (+)-sabinol . J Chem Soc C 5 : 716 – 721 Google Scholar OpenURL Placeholder Text WorldCat Grabherr MG , Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, Adiconis X, Fan L, Raychowdhury R, Zeng Q, et al. ( 2011 ) Full-length transcriptome assembly from RNA-Seq data without a reference genome . Nat Biotechnol 29 : 644 – 652 Google Scholar Crossref Search ADS PubMed WorldCat Hall DE , Robert JA, Keeling CI, Domanski D, Quesada AL, Jancsik S, Kuzyk MA, Hamberger B, Borchers CH, Bohlmann J ( 2011 ) An integrated genomic, proteomic and biochemical analysis of (+)-3-carene biosynthesis in Sitka spruce (Picea sitchensis) genotypes that are resistant or susceptible to white pine weevil . Plant J 65 : 936 – 948 Google Scholar Crossref Search ADS PubMed WorldCat Hamann T , Møller BL ( 2007 ) Improved cloning and expression of cytochrome P450s and cytochrome P450 reductase in yeast . Protein Expr Purif 56 : 121 – 127 Google Scholar Crossref Search ADS PubMed WorldCat Hamberger B , Bak S ( 2013 ) Plant P450s as versatile drivers for evolution of species-specific chemical diversity . Philos Trans R Soc Lond B Biol Sci 368 : 20120426 Google Scholar Crossref Search ADS PubMed WorldCat Hamberger B , Ohnishi T, Hamberger B, Séguin A, Bohlmann J ( 2011 ) Evolution of diterpene metabolism: Sitka spruce CYP720B4 catalyzes multiple oxidations in resin acid biosynthesis of conifer defense against insects . Plant Physiol 157 : 1677 – 1695 Google Scholar Crossref Search ADS PubMed WorldCat Hebda RJ , Mathewes RW ( 1984 ) Holocene history of cedar and native Indian cultures of the north american pacific coast . Science 225 : 711 – 713 Google Scholar Crossref Search ADS PubMed WorldCat Karp F , Croteau R ( 1982 ) Evidence that sabinene is an essential precursor of C(3)-oxygenated thujane monoterpenes . Arch Biochem Biophys 216 : 616 – 624 Google Scholar Crossref Search ADS PubMed WorldCat Keeling CI , Bohlmann J ( 2006 ) Genes, enzymes and chemicals of terpenoid diversity in the constitutive and induced defence of conifers against insects and pathogens . New Phytol 170 : 657 – 675 Google Scholar Crossref Search ADS PubMed WorldCat Kimball BA , Russell JH, Griffin DL, Johnston JJ ( 2005 ) Response factor considerations for the quantitative analysis of western redcedar (Thuja plicata) foliar monoterpenes . J Chromatogr Sci 43 : 253 – 258 Google Scholar Crossref Search ADS PubMed WorldCat Kimball BA , Russell JH, Ott P ( 2012 ) Phytochemical variation within a single plant species influences foraging behavior of deer . Oikos 5 : 743 – 751 Google Scholar OpenURL Placeholder Text WorldCat Lohse M , Bolger AM, Nagel A, Fernie AR, Lunn JE, Stitt M, Usadel B ( 2012 ) RobiNA: a user-friendly, integrated software solution for RNA-Seq-based transcriptomics . Nucleic Acids Res 40 : W622 – W627 Google Scholar Crossref Search ADS PubMed WorldCat Lupien S , Karp F, Wildung M, Croteau R ( 1999 ) Regiospecific cytochrome P450 limonene hydroxylases from mint (Mentha) species: cDNA isolation, characterization, and functional expression of (-)-4S-limonene-3-hydroxylase and (-)-4S-limonene-6-hydroxylase . Arch Biochem Biophys 368 : 181 – 192 Google Scholar Crossref Search ADS PubMed WorldCat MacLean H , Murakami K ( 1966 ) Lignans of western red cedar (Thuja plicata Donn). V. Hydroxythujaplicatin methyl ether . Can J Chem 44 : 1827 – 1830 Google Scholar Crossref Search ADS WorldCat Mamane V , Gress T, Krause H, Fürstner A ( 2004 ) Platinum- and gold-catalyzed cycloisomerization reactions of hydroxylated enynes . J Am Chem Soc . 126 : 8654 – 8655 Google Scholar Crossref Search ADS PubMed WorldCat Morris PI , Stirling R ( 2012 ) Western red cedar extractives associated with durability in ground contact . Wood Sci Technol 46 : 991 – 1002 Google Scholar Crossref Search ADS WorldCat Neale DB , Wegrzyn JL, Stevens KA, Zimin AV, Puiu D, Crepeau MW, Cardeno C, Koriabine M, Holtz-Morris AE, Liechty JD, et al. ( 2014 ) Decoding the massive genome of loblolly pine using haploid DNA and novel assembly strategies . Genome Biol 15 : R59 Google Scholar Crossref Search ADS PubMed WorldCat Nei M , Kumar S ( 2000 ) Molecular Evolution and Phylogenetics , Ed 1. Oxford University Press , New York , pp 1 – 333 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Nelson DR ( 2006 ) Cytochrome P450 nomenclature, 2004 . Methods Mol Biol 320 : 1 – 10 Google Scholar PubMed OpenURL Placeholder Text WorldCat Nystedt B , Street NR, Wetterbom A, Zuccolo A, Lin YC, Scofield DG, Vezzi F, Delhomme N, Giacomello S, Alexeyenko A, et al. ( 2013 ) The Norway spruce genome sequence and conifer genome evolution . Nature 497 : 579 – 584 Google Scholar Crossref Search ADS PubMed WorldCat Ohloff G , Uhde G, Thomas AF, Kováts Esz ( 1966 ) The absolute configuration of thujane . Tetrahedron 22 : 309 – 320 Google Scholar Crossref Search ADS WorldCat Omura T , Sato R ( 1964 ) The carbon monoxide-binding pigment of liver microsomes. I. Evidence for its hemoprotein nature . J Biol Chem 239 : 2370 – 2378 Google Scholar Crossref Search ADS PubMed WorldCat Pompon D , Louerat B, Bronine A, Urban P ( 1996 ) Yeast expression of animal and plant P450s in optimized redox environments . Methods Enzymol 272 : 51 – 64 Google Scholar Crossref Search ADS PubMed WorldCat Ro DK , Arimura G, Lau SYW, Piers E, Bohlmann J ( 2005 ) Loblolly pine abietadienol/abietadienal oxidase PtAO (CYP720B1) is a multifunctional, multisubstrate cytochrome P450 monooxygenase . Proc Natl Acad Sci USA 102 : 8060 – 8065 Google Scholar Crossref Search ADS PubMed WorldCat Ro DK , Paradise EM, Ouellet M, Fisher KJ, Newman KL, Ndungu JM, Ho KA, Eachus RA, Ham TS, Kirby J, et al. ( 2006 ) Production of the antimalarial drug precursor artemisinic acid in engineered yeast . Nature 440 : 940 – 943 Google Scholar Crossref Search ADS PubMed WorldCat Roach CR , Hall DE, Zerbe P, Bohlmann J ( 2014 ) Plasticity and evolution of (+)-3-carene synthase and (-)-sabinene synthase functions of a sitka spruce monoterpene synthase gene family associated with weevil resistance . J Biol Chem 289 : 23859 – 23869 Google Scholar Crossref Search ADS PubMed WorldCat Robineau T , Batard Y, Nedelkina S, Cabello-Hurtado F, LeRet M, Sorokine O, Didierjean L, Werck-Reichhart D ( 1998 ) The chemically inducible plant cytochrome P450 CYP76B1 actively metabolizes phenylureas and other xenobiotics . Plant Physiol 118 : 1049 – 1056 Google Scholar Crossref Search ADS PubMed WorldCat Rontein D , Onillon S, Herbette G, Lesot A, Werck-Reichhart D, Sallaud C, Tissier A ( 2008 ) CYP725A4 from yew catalyzes complex structural rearrangement of taxa-4(5),11(12)-diene into the cyclic ether 5(12)-oxa-3(11)-cyclotaxane . J Biol Chem 283 : 6067 – 6075 Google Scholar Crossref Search ADS PubMed WorldCat Ruijter JM , Ramakers C, Hoogaars WMH, Karlen Y, Bakker O, van den Hoff MJB, Moorman AF ( 2009 ) Amplification efficiency: linking baseline and bias in the analysis of quantitative PCR data . Nucleic Acids Res 37 : e45 Google Scholar Crossref Search ADS PubMed WorldCat Russell JH , Ferguson CF ( 2008 ) Preliminary results from five generations of a western redcedar (Thuja plicata) selection study with self mating . Tree Genet Genomes 4 : 509 – 518 Google Scholar Crossref Search ADS WorldCat Russell JH , Yanchuk AD ( 2012 ) Breeding for growth improvement and resistance to multiple pests in Thuja plicata . Gen Tech Rep 240 : 40 – 44 Sirisoma NS , Höld KM, Casida JE ( 2001 ) alpha- and beta-Thujones (herbal medicines and food additives): synthesis and analysis of hydroxy and dehydro metabolites . J Agric Food Chem 49 : 1915 – 1921 Google Scholar Crossref Search ADS PubMed WorldCat Tamura K , Dudley J, Nei M, Kumar S ( 2007 ) MEGA4: Molecular Evolutionary Genetics Analysis (MEGA) software version 4.0 . Mol Biol Evol 24 : 1596 – 1599 Google Scholar Crossref Search ADS PubMed WorldCat Thompson JD , Higgins DG, Gibson TJ ( 1994 ) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice . Nucleic Acids Res 22 : 4673 – 4680 Google Scholar Crossref Search ADS PubMed WorldCat Tisserand R , Young R ( 2014 ) Essential Oil Safety , Ed 2. Churchill Livingstone , London , pp 1 – 784 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tsiri D , Graikou K, Pobłocka-Olech L, Krauze-Baranowska M, Spyropoulos C, Chinou I ( 2009 ) Chemosystematic value of the essential oil composition of Thuja species cultivated in Poland-antimicrobial activity . Molecules 14 : 4707 – 4715 Google Scholar Crossref Search ADS PubMed WorldCat Umbreit MA , Sharpless KB ( 1977 ) Allylic oxidation of olefins by catalytic and stoichiometric Selenium dioxide with tert-Butyl hydroperoxide . J Am Chem Soc 99 : 5526 – 5528 Google Scholar Crossref Search ADS WorldCat Von Rudloff E , Lapp MS, Yeh FC ( 1988 ) Chemosystematic study of Thuja plicata: multivariate analysis of leaf oil terpene composition . Biochem Syst Ecol 16 : 119 – 125 Google Scholar Crossref Search ADS WorldCat Vourc’h G , Russell J, Martin JL ( 2002 ) Linking deer browsing and terpene production among genetic identities in Chamaecyparis nootkatensis and Thuja plicata (Cupressaceae) . J Hered 93 : 370 – 376 Google Scholar Crossref Search ADS PubMed WorldCat Vourc’h G , Martin JL, Duncan P, Escarré J, Clausen TP ( 2001 ) Defensive adaptations of Thuja plicata to ungulate browsing: a comparative study between mainland and island populations . Oecologia 126 : 84 – 93 Google Scholar Crossref Search ADS PubMed WorldCat Zifkin M , Jin A, Ozga JA, Zaharia LI, Schernthaner JP, Gesell A, Abrams SR, Kennedy JA, Constabel CP ( 2012 ) Gene expression and metabolite profiling of developing highbush blueberry fruit indicates transcriptional regulation of flavonoid metabolism and activation of abscisic acid metabolism . Plant Physiol 158 : 200 – 224 Google Scholar Crossref Search ADS PubMed WorldCat Author notes 1 This work was supported by the Natural Sciences and Engineering Research Council of Canada (Strategic Grant to J.M. and J.B. and Discovery Grant to J.B.), Genome British Columbia (User Partnership Program Project funds for bioinformatics analysis to J.M., J.H.R., and J.B.), the Austrian Science Fund (Erwin Schroedinger Fellowship to M.B.), Michael Smith Laboratories and the University of British Columbia (funds for analytical work of metabolite profiling), and the University of British Columbia (Distinguished Scholar Award to J.B.). * Address correspondence to [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Jörg Bohlmann ([email protected]). [OPEN] Articles can be viewed without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.15.00315 © 2015 American Society of Plant Biologists. All Rights Reserved. © The Author(s) 2015. Published by Oxford University Press on behalf of American Society of Plant Biologists. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Üstün, Suayib; Börnke, Frederik

doi: 10.1104/pp.15.00132pmid: 25739698

Abstract Many animal and plant pathogenic bacteria inject type III effector (T3E) proteins into their eukaryotic host cells to suppress immunity. The Yersinia outer protein J (YopJ) family of T3Es is a widely distributed family of effector proteins found in both animal and plant pathogens, and its members are highly diversified in virulence functions. Some members have been shown to possess acetyltransferase activity; however, whether this is a general feature of YopJ family T3Es is currently unknown. The T3E Xanthomonas outer protein J (XopJ), a YopJ family effector from the plant pathogen Xanthomonas campestris pv vesicatoria, interacts with the proteasomal subunit Regulatory Particle AAA-ATPase6 (RPT6) in planta to suppress proteasome activity, resulting in the inhibition of salicylic acid-related immune responses. Here, we show that XopJ has protease activity to specifically degrade RPT6, leading to reduced proteasome activity in the cytoplasm as well as in the nucleus. Proteolytic degradation of RPT6 was dependent on the localization of XopJ to the plasma membrane as well as on its catalytic triad. Mutation of the Walker B motif of RPT6 prevented XopJ-mediated degradation of the protein but not XopJ interaction. This indicates that the interaction of RPT6 with XopJ is dependent on the ATP-binding activity of RPT6, but proteolytic cleavage additionally requires its ATPase activity. Inhibition of the proteasome impairs the proteasomal turnover of Nonexpressor of Pathogenesis-Related1 (NPR1), the master regulator of salicylic acid responses, leading to the accumulation of ubiquitinated NPR1, which likely interferes with the full induction of NPR1 target genes. Our results show that YopJ family T3Es are not only highly diversified in virulence function but also appear to possess different biochemical activities. Plants have been threatened by microbial infections throughout their phylogenetic history and thus evolved a sophisticated and multilayered immune system (Jones and Dangl, 2006). Phytohormonal signaling is instrumental to many aspects of plant defense, and salicylic acid (SA) has emerged as the central regulator of signaling pathways during induced immunity (Pieterse et al., 2012). SA signaling plays major roles during systemic acquired resistance (SAR), basal defense responses, and gene-for-gene resistance. Nonexpressor of PR1 (NPR1) represents an essential component of SA signaling that acts as a transcriptional coactivator to induce transcriptional reprogramming, including the induction of PR (for pathogenesis-related) genes and genes associated with secretion (Wang et al., 2005; Pieterse et al., 2012). Upon pathogen attack, SA levels increase within the cytoplasm of the host cell, which then leads to the monomerization of NPR1 and translocation of the protein into the nucleus. Once in the nucleus, NPR1 can interact with different TGACG SEQUENCE-SPECIFIC BINDING PROTEIN transcription factors and drives the activation of SA-responsive gene expression (Fu and Dong, 2013). Subsequent phosphorylation serves as a signal for the ubiquitination of NPR1, mediating its degradation via the 26S proteasome and thereby returning the target gene promoter to the initial state (Spoel et al., 2009). It is assumed that the rate of NPR1 degradation by the proteasome determines the rate of mRNA production from SA-responsive genes (Spoel et al., 2009). The significance of the proteasomal turnover of NPR1 during SA-mediated defense signaling highlights the important role of the proteasome in plant defense. Accordingly, Arabidopsis (Arabidopsis thaliana) mutant lines defective in proteasome function are impaired in certain immune responses (Yao et al., 2012), and particular proteasomal subunits have been proposed to act as a caspase-like enzyme during the induction of programmed cell death in response to avirulent bacterial strains (Hatsugai et al., 2009). Thus, the 26S proteasome seems to control defense mechanisms alongside its usual role as a multicatalytic protein complex that is essential for the degradation of ubiquitinated proteins. The 26S proteasome itself is composed of a 20S core particle (CP) that is capped on either end by a 19S regulatory particle (RP; Vierstra, 2009). The RP recognizes and binds ubiquitinated proteins, deubiquitinates and unfolds these substrates in an ATP-dependent manner, and controls the entry of unfolded target proteins to the proteolytic channel of the CP (Vierstra, 2009). The injection of type III effector (T3E) proteins or bacterial toxins into the host cell is an efficient mechanism employed by many bacterial pathogens to suppress plant immunity and to promote disease development (Galán and Wolf-Watz, 2006). In accordance with its central role in a wide array of cellular processes, including defense, the proteasome is targeted or exploited by bacterial toxins or T3E proteins (Dudler, 2013). Certain strains of Pseudomonas syringae pv syringae secrete syringolin A (SylA), which is a small nonribosomal peptide that inhibits proteasome activity through binding to the catalytic subunits of the 20S CP and is required for full virulence on host plants, such as bean (Phaseolus vulgaris; Groll et al., 2008). The initial discovery of the interaction between the bacterial T3E protein Xanthomonas outer protein J (XopJ) from Xanthomonas campestris pv vesicatoria (Xcv) and Regulatory Particle AAA-ATPase6 (RPT6), a subunit of the 19S RP, provided, to our knowledge, the first example of a bacterial T3E directly targeting the proteasome (Üstün et al., 2013). The interaction of XopJ with RPT6 leads to the inhibition of the proteasome, which subsequently interferes with SA-dependent defense responses to attenuate the onset of necrosis in infected tissue and to alter host transcription. The development of XopJ-associated phenotypes on susceptible pepper (Capsicum annuum) plants was shown to be dependent on NPR1 (Üstün et al., 2013). Another T3E effector targeting RPT6 to inhibit proteasome activity is Hypersensitivity and pathogenesis-dependent outer protein Z4 (HopZ4) from Pseudomonas syringae pv lachrymans (Üstün et al., 2014). Both T3Es belong to the widespread yersinia outer protein J (YopJ) family of effector proteins that are present among plant and animal pathogenic bacteria and were originally classified as Cys proteases (Lewis et al., 2011). However, recent studies indicate that these effectors, in particular the archetypal member YopJ from Yersinia pestis, HopZ1a from P. syringae pv syringae, and Avirulence protein AvrBsT from Xanthomonas euvesicatoria, act as acetyltransferases on their targets or decoys to suppress immunity in animals and plants (Mukherjee et al., 2006; Lee et al., 2012; Jiang et al., 2013; Lewis et al., 2013; Cheong et al., 2014). In some cases, these effectors also display weak protease activity and lead to the degradation of potential target proteins, opening the possibility that YopJ family effectors can also act proteolytically (Ma et al., 2005; Szczesny et al., 2010; Zhou et al., 2011). However, the mechanism through which the interaction of XopJ with RPT6 leads to the inhibition of the proteasome and the role of NPR1 in this context are currently unknown. Here, we show that XopJ acts as a protease to directly degrade its target protein, RPT6, in a process that is dependent on the localization of XopJ at the plasma membrane and on an intact catalytic triad of the effector. Interaction of RPT6 with XopJ is dependent on the ATP-binding activity of RPT6, but proteolytic cleavage additionally requires its ATPase activity. The malfunction of the proteasome impairs the proteasomal turnover of NPR1 and thus results in the accumulation of ubiquitinated NPR1. We conclude that XopJ-mediated degradation of RPT6 connected with the inhibition of the proteasome prevents the proteasomal turnover of NPR1 and thereby suppresses SA-mediated defense signaling. RESULTS XopJ Mediates the Destabilization of RPT6 Protein Levels Although various members of the YopJ-like effector family were already characterized as acetyltransferases (Lee et al., 2012; Cheong et al., 2014), recombinant XopJ does not possess measurable acetyltransferase activity in vitro (Supplemental Fig. S1). However, when XopJ-HA (for hemagglutinin) was transiently coexpressed with RPT6-GFP in leaves of Nicotiana benthamiana, GFP fluorescence was decreased dramatically as compared with N. benthamiana leaves coexpressing RPT6-GFP with the catalytically inactive XopJ variant C235A or the nonmyristoylated variant G2A (Fig. 1, A–D). To allow for the direct comparison of signal intensities, all confocal images shown in Figure 1 were recorded with the same microscope settings. However, Figure 1E illustrates that RPT6-GFP fluorescence becomes readily detectable in XopJ-coexpressing leaves when the microscope settings are adjusted to increase the sensitivity of detection. Quantification of the GFP intensity signal revealed that XopJ-HA caused a significant reduction of RPT6-GFP fluorescence dependent on its localization (G2A) or enzymatic activity (C235A; Fig. 1F). Analysis of the protein levels of RPT6-GFP confirmed the microscopic data, as the RPT6-GFP signal was substantially decreased in the presence of XopJ-HA but not in the presence of its mutant variants (Fig. 1G). To rule out any toxicity effects that could result from the overexpression of XopJ-HA in N. benthamiana, endogenous protein levels of the cytosolic Sucrose Phosphatase2 (SPP2) were examined in plants expressing XopJ and its variants. The data revealed that XopJ wild-type protein did not result in decreased SPP2 protein levels (Fig. 1G), indicating that XopJ specifically reduces RPT6-GFP protein levels. Reverse transcription (RT)-PCR analysis revealed that coexpression with XopJ does not affect RPT6-GFP mRNA levels (Fig. 1H). This suggests that XopJ interferes with RPT6 protein stability. Figure 1. Open in new tabDownload slide XopJ interferes with RPT6-GFP protein accumulation. A to D, Empty vector (EV; A), XopJ-HA (B), G2A-HA (C), and C235A-HA (D) were transiently coexpressed together with RPT6 GFP in N. benthamiana using A. tumefaciens infiltration. For confocal laser scanning microscopy (CLSM), samples were taken 48 h post inoculation (hpi), and images were generated with identical CLSM settings. E, Microscope image of an RPT6-GFP- and XopJ-HA-coexpressing leaf recorded with increased photomultiplier gain as compared with the images presented in A to D. GFP fluorescence is shown in green, and chlorophyll autofluorescence is shown in red. Bar = 20 μm. F, Quantification of GFP intensity after coexpression of RPT6-GFP together with XopJ and its variants. GFP quantification was determined by using the Leica software LAS_AF. Coexpression with the EV control was set to 100%. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks at P < 0.001. G, Total proteins were extracted 48 hpi with A. tumefaciens harboring the respective XopJ and RPT6 expression construct. RPT6-GFP protein levels were detected using an anti-GFP antibody. Expression of the XopJ variants was verified using an anti-HA antibody after stripping the same membrane. Analysis of the endogenous protein levels of cytosolic SPP2 (anti-NtSPP2) served as a control, and staining of the membrane with Amido Black showed equal loading. H, XopJ does not affect RPT6 expression levels. RT-PCR shows transgenic RPT6 mRNA in N. enthamiana leaves transiently coexpressing RPT6 with an EV control or XopJ-HA. Ubiquitin was used as an amplification control. All experiments were repeated at least three times with almost identical results. Figure 1. Open in new tabDownload slide XopJ interferes with RPT6-GFP protein accumulation. A to D, Empty vector (EV; A), XopJ-HA (B), G2A-HA (C), and C235A-HA (D) were transiently coexpressed together with RPT6 GFP in N. benthamiana using A. tumefaciens infiltration. For confocal laser scanning microscopy (CLSM), samples were taken 48 h post inoculation (hpi), and images were generated with identical CLSM settings. E, Microscope image of an RPT6-GFP- and XopJ-HA-coexpressing leaf recorded with increased photomultiplier gain as compared with the images presented in A to D. GFP fluorescence is shown in green, and chlorophyll autofluorescence is shown in red. Bar = 20 μm. F, Quantification of GFP intensity after coexpression of RPT6-GFP together with XopJ and its variants. GFP quantification was determined by using the Leica software LAS_AF. Coexpression with the EV control was set to 100%. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks at P < 0.001. G, Total proteins were extracted 48 hpi with A. tumefaciens harboring the respective XopJ and RPT6 expression construct. RPT6-GFP protein levels were detected using an anti-GFP antibody. Expression of the XopJ variants was verified using an anti-HA antibody after stripping the same membrane. Analysis of the endogenous protein levels of cytosolic SPP2 (anti-NtSPP2) served as a control, and staining of the membrane with Amido Black showed equal loading. H, XopJ does not affect RPT6 expression levels. RT-PCR shows transgenic RPT6 mRNA in N. enthamiana leaves transiently coexpressing RPT6 with an EV control or XopJ-HA. Ubiquitin was used as an amplification control. All experiments were repeated at least three times with almost identical results. In addition, RPT6-GFP and XopD from Xcv were transiently coexpressed in N. benthamiana to exclude that the reduction of RPT6-GFP protein levels is a general characteristic of effectors that share similarities to Cys proteases (Supplemental Fig. S2). To investigate whether XopJ would also affect RPT6 from other kingdoms of the phylogenetic tree, we tested its interaction with RPT6 from humans, as both proteins share high similarity on amino acid level (80%). Yeast two-hybrid analysis and in planta bimolecular fluorescence complementation (BiFC) showed that XopJ is not able to interact with human RPT6 (Supplemental Fig. S3, A and B). Thus, XopJ is not able to recruit HsRPT6 to the plasma membrane and also does not lead to the destabilization of HsRPT6-GFP in N. benthamiana (Supplemental Fig. S3, C and D). These data support the notion that XopJ specifically triggers the degradation of RPT6 from plants. The Degradation of RPT6 Is Protease Dependent To determine the mechanism underlying the XopJ-triggered reduction of RPT6 protein levels, coexpression of RPT6 and XopJ in the presence of different inhibitors was performed. The XopJ-dependent destabilization of RPT6-GFP in N. benthamiana was not affected by the proteasome inhibitor MG132 (Fig. 2A). In contrast, application of a commercially available mix of protease inhibitors (Ser, Cys, aspartic, and metalloproteases and aminopeptidases) abolished the XopJ-mediated decrease of RPT6 protein levels (Fig. 2A). Measurement of the proteasome activity in plants expressing XopJ in the presence or absence of protease inhibitors revealed that the mix of protease inhibitors abolished the proteasome-inhibiting ability of XopJ (Fig. 2B). These data indicate that XopJ might act as a protease with RPT6 as a substrate to inhibit proteasome activity. Figure 2. Open in new tabDownload slide Destabilization of RPT6 is protease dependent. A, RPT6-GFP was transiently coexpressed together with EV, XopJ-myc, and XopJ C235A-myc in N. benthamiana using agroinfiltration. At 42 hpi, 100 µm MG132 or a protease inhibitor mix was infiltrated into A. tumefaciens-inoculated leaves, and leaf material was collected 48 hpi. Expression of RPT6-GFP was detected using an anti-GFP antibody. After stripping of the membrane, protein levels of XopJ and XopJ C235A were analyzed with an anti-myc antibody. After immunodetection of proteins, the membrane was stained with Amido Black to control for equal protein loading. B, N. benthamiana leaves transiently expressing XopJ-myc were treated with a cocktail of protease inhibitors or water 42 hpi. At 48 hpi, relative proteasome activity in total protein extracts was determined by monitoring the breakdown of the fluorogenic peptide Suc-LLVY-AMC at 30°C in a fluorescence spectrophotometer. The EV control was set to 100%. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks at P < 0.01. The experiments in both A and B were repeated twice with similar results. Figure 2. Open in new tabDownload slide Destabilization of RPT6 is protease dependent. A, RPT6-GFP was transiently coexpressed together with EV, XopJ-myc, and XopJ C235A-myc in N. benthamiana using agroinfiltration. At 42 hpi, 100 µm MG132 or a protease inhibitor mix was infiltrated into A. tumefaciens-inoculated leaves, and leaf material was collected 48 hpi. Expression of RPT6-GFP was detected using an anti-GFP antibody. After stripping of the membrane, protein levels of XopJ and XopJ C235A were analyzed with an anti-myc antibody. After immunodetection of proteins, the membrane was stained with Amido Black to control for equal protein loading. B, N. benthamiana leaves transiently expressing XopJ-myc were treated with a cocktail of protease inhibitors or water 42 hpi. At 48 hpi, relative proteasome activity in total protein extracts was determined by monitoring the breakdown of the fluorogenic peptide Suc-LLVY-AMC at 30°C in a fluorescence spectrophotometer. The EV control was set to 100%. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks at P < 0.01. The experiments in both A and B were repeated twice with similar results. XopJ Possesses Protease Activity in Vitro and in Vivo To clarify whether XopJ acts as a protease itself, protease activity was monitored using a Förster resonance energy transfer-based protease detection kit containing a substrate peptide library of more than 2 × 106 peptide variants, with proteinase K serving as a positive control (Kapprell et al., 2011). Recombinant Maltose-Binding Protein (MBP)-XopJ, purified from Escherichia coli, showed a significant increase in protease activity when compared with MBP alone (Fig. 3A). This activity was dependent on an intact catalytic triad, because the MBP-XopJ C235A variant did not result in a significant rise in protease activity and remained at MBP background levels (Fig. 3A). Incubation of MBP-XopJ together with MBP-RPT6 diminished the protease activity, probably by outcompeting the binding of library peptides to XopJ (Fig. 3A). Application of a cocktail of protease inhibitors confirmed that XopJ indeed displays protease activity, as its activity was blocked similar to that of the positive control (Supplemental Fig. S4A). To further dissect this effect, specific protease inhibitor substances were tested in an in vitro assay. The results showed that the protease activity of XopJ is inhibited by the Cys protease inhibitor E-64 but not by inhibitors of proteases from other classes (Supplemental Fig. S4B), further corroborating the notion that XopJ acts as a Cys protease. Neither chymostatin, which is a specific inhibitor of α-, β-, γ-, and δ-chymotrypsin, nor the Ser protease inhibitor aprotinin was able to diminish the protease activity of recombinant XopJ (Supplemental Fig. S4B). The Ser and Cys protease inhibitor leupeptin also did not reduce the protease activity of XopJ, as it is highly specific toward the Cys proteases papain and cathepsin B (Umezawa, 1976). The reduction of the protease activity of XopJ by phenylmethylsulfonyl fluoride can be explained by the fact that phenylmethylsulfonyl fluoride also inhibits Cys proteases at high concentrations (van der Hoorn et al., 2004). To test the in vivo protease activity of XopJ, XopJ-HA, C235A-HA, and an EV control were transiently expressed in N. benthamiana and protease activity was measured in crude extracts 48 hpi. A significant increase in protease activity in plant extracts expressing XopJ-HA could be detected in comparison with the EV or the enzymatically inactive C235-HA variant that remained at background protease activity levels (Fig. 3B). Treatment with a mix of protease inhibitors diminished XopJ-dependent protease activity to EV or C235A levels (Supplemental Fig. S4C). Figure 3. Open in new tabDownload slide XopJ displays protease activity in vitro and in vivo. A, Five micrograms of MBP, MBP-XopJ, MBP-XopJ C235A, MBP-RPT6, and a mix of MBP-XopJ and MBP-RPT6 purified from E. coli was subjected to protease activity measurement using the P-CHECK protease detection kit. MBP and MBP-RPT6 served as negative controls. As a positive control, the Ser protease Proteinase K was included. A representative result of more than three repetitions with independent sets of purified proteins is shown. RFU, Relative fluorescence units. B, Protease activity in N. benthamiana leaves transiently expressing XopJ-HA proteins. XopJ protein variants along with an EV control were transiently expressed in leaves of N. benthamiana using agroinfiltration. After 48 h, protease activity in total protein extracts was measured using the P-CHECK protease detection kit. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks: *, P < 0.05; and **, P < 0.01. The experiment was repeated twice with similar results. RLU, Relative light units. Figure 3. Open in new tabDownload slide XopJ displays protease activity in vitro and in vivo. A, Five micrograms of MBP, MBP-XopJ, MBP-XopJ C235A, MBP-RPT6, and a mix of MBP-XopJ and MBP-RPT6 purified from E. coli was subjected to protease activity measurement using the P-CHECK protease detection kit. MBP and MBP-RPT6 served as negative controls. As a positive control, the Ser protease Proteinase K was included. A representative result of more than three repetitions with independent sets of purified proteins is shown. RFU, Relative fluorescence units. B, Protease activity in N. benthamiana leaves transiently expressing XopJ-HA proteins. XopJ protein variants along with an EV control were transiently expressed in leaves of N. benthamiana using agroinfiltration. After 48 h, protease activity in total protein extracts was measured using the P-CHECK protease detection kit. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks: *, P < 0.05; and **, P < 0.01. The experiment was repeated twice with similar results. RLU, Relative light units. XopJ Degrades RPT6 in a Protease-Dependent Manner To further investigate the proteolytic activity of XopJ toward RPT6, the same amounts of purified recombinant MBP-RPT6 (Supplemental Fig. S5) were incubated together with crude extracts from XopJ-expressing leaves. Subsequent western-blot analyses revealed that the amount of MBP-RPT6 was reduced in the presence of XopJ-expressing extracts (Fig. 4A). Mutation of the Cys-235 residue within the catalytic triad of XopJ as well as the addition of protease inhibitors abolished the degradation of MBP-RPT6 (Fig. 4B), supporting the notion that XopJ acts as a protease to degrade RPT6. As recombinant MBP-XopJ also has protease activity in vitro, we tested whether purified XopJ is able to degrade plant-expressed RPT6-GFP. After incubation of MBP-XopJ together with RPT6-GFP, RPT6 protein levels decreased in the presence of wild-type XopJ but not upon coincubation with the C235A variant (Supplemental Fig. S6A). Incubation of glutathione S-transferase (GST)-XopJ together with recombinant MBP-RPT6 purified from E. coli also resulted in a reduction of RPT6 protein abundance (Supplemental Fig. S6B), further indicating that the proteolytic activity of recombinant XopJ is sufficient to degrade RPT6 in vitro. Moreover, N. benthamiana extracts incubated with MBP-XopJ have significantly lower proteasome activities compared with MBP- or MBP-XopJ C235A-treated extracts (Supplemental Fig. S6C). Taken together, these data demonstrate that XopJ displays protease activity to degrade RPT6 and thereby inhibits proteasome activity in plant cells. Figure 4. Open in new tabDownload slide XopJ degrades RPT6 in a protease-dependent manner. XopJ has protease activity on recombinant MBP-RPT6 when transiently expressed in N. benthamiana leaves dependent on its catalytic activity. A, Extracts from N. benthamiana transiently expressing XopJ-HA or EV were incubated together with MBP-RPT6 purified from E. coli. RPT6 protein levels were monitored using an anti-MBP antibody, and XopJ protein accumulation was detected using an anti-HA antibody. Amido Black staining served as a loading control for the plant extracts used. B, Immunoblots using an anti-MBP antibody show MBP-RPT6 accumulation after incubation with crude extracts of N. benthamiana leaves expressing XopJ-HA or C235A-HA with (+) or without (−) protease inhibitors. Plants expressing the EV control served as a negative control. Anti-HA immunoblotting was performed to detect the proper expression of XopJ-HA proteins. Amido Black staining served as a loading control for the plant extracts used. The results in both A and B are representative of two independent experiments. Figure 4. Open in new tabDownload slide XopJ degrades RPT6 in a protease-dependent manner. XopJ has protease activity on recombinant MBP-RPT6 when transiently expressed in N. benthamiana leaves dependent on its catalytic activity. A, Extracts from N. benthamiana transiently expressing XopJ-HA or EV were incubated together with MBP-RPT6 purified from E. coli. RPT6 protein levels were monitored using an anti-MBP antibody, and XopJ protein accumulation was detected using an anti-HA antibody. Amido Black staining served as a loading control for the plant extracts used. B, Immunoblots using an anti-MBP antibody show MBP-RPT6 accumulation after incubation with crude extracts of N. benthamiana leaves expressing XopJ-HA or C235A-HA with (+) or without (−) protease inhibitors. Plants expressing the EV control served as a negative control. Anti-HA immunoblotting was performed to detect the proper expression of XopJ-HA proteins. Amido Black staining served as a loading control for the plant extracts used. The results in both A and B are representative of two independent experiments. The Walker A Motif of RPT6 Is Required for XopJ Binding RPT6 is one of six ATPases of the AAA protein family, being part of the RP of the 26S proteasome (Bar-Nun and Glickman, 2012). AAA proteins feature conserved Walker A and Walker B domains that bind and hydrolyze ATP, respectively, to generate a mechanical force that is used to unfold substrate proteins (Hanson and Whiteheart, 2005). Whereas ATP binding to the ATPase subunits of the proteasome is necessary for the association of the RP to the CP, substrate binding, and the opening of the proteolytic channel, ATP hydrolysis drives the unfolding of substrates that are destined to be degraded by the 26S proteolytic machinery (Bar-Nun and Glickman, 2012). In order to relate RPT6 functionality to its interaction with XopJ, a point mutation was generated in either of the two Walker motifs of RPT6 that rendered the protein defective in ATP binding (Walker A) or ATP hydrolysis (Walker B). Mutation of the Walker A motif in RPT6 by exchanging the Lys residue at position 206 to an Ala (K206A) abrogated the interaction with XopJ in yeast (Saccharomyces cerevisiae), while a point mutation in the Walker B motif (E260A) did not affect the binding of XopJ to RPT6 (Fig. 5A). Figure 5. Open in new tabDownload slide The RPT6 Walker A motif is required for XopJ binding. A, XopJ interacts with RPT6 E260A (Walker B) but not with RPT6 K206A (Walker A) in a yeast two-hybrid assay. XopJ fused to the GAL4 DNA-binding domain (pGBT9) was expressed in combination with RPT6 E260A or K206A fused to the GAL4 activation domain. NtRPT6, Nicotiana tabacum RPT6; –LT, yeast growth on medium without Leu and Trp; –HLT, yeast growth on medium lacking His, Leu, and Trp, indicating expression of the HIS3 reporter gene. B, BiFC in planta interaction studies of XopJ and RPT6 mutants. YFP confocal microscopy images show N. benthamiana leaf epidermal cells transiently expressing XopJ-VenusN in combination with RPT6 K206A or E260A-VenusC. A closeup of the same cells shows that the YFP fluorescence of XopJ-VenusN/RPT6 E260A-VenusC aligns with the plasma membrane. XopJ-VenusN and RPT6-VenusC and the dimerization of fructose-1,6-bisphosphatase (FBPase) within the cytosol served as positive controls. XopJ-VenusN with FBPase-VenusC or the RPT6 mutants together with FBPase-VenusN are included as negative controls. Bars = 20 μm, except for the closeup (5 µm). C, EV or XopJ-HA was transiently coexpressed together with RPT6-GFP, RPT6 K206A-GFP, or RPT6 E260A-GFP in N. benthamiana using agroinfiltration. Samples were taken 48 hpi, and images were generated with identical CLSM settings. GFP fluorescence is shown in green, and chlorophyll autoflourescence is shown in red. Arrows indicate cytosolic strands. Bars = 20 μm. D, Quantification of GFP intensity after coexpression of RPT6-GFP, RPT6 K206A, and RPT6 E260A-GFP together with XopJ. GFP quantification was determined by using the ZEN software of Zeiss. Coexpression with the EV control was set to 100%. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks at P < 0.001. E, XopJ does not degrade the RPT6 Walker B mutant. Total proteins were extracted 48 hpi with agrobacteria harboring the respective XopJ and RPT6 expression constructs. RPT6-GFP, RPT6 K206-GFP, and RPT6 E260A-GFP protein levels were detected using an anti-GFP antibody. Expression of XopJ was verified using an anti-HA antibody. All experiments were repeated three times with similar results. Figure 5. Open in new tabDownload slide The RPT6 Walker A motif is required for XopJ binding. A, XopJ interacts with RPT6 E260A (Walker B) but not with RPT6 K206A (Walker A) in a yeast two-hybrid assay. XopJ fused to the GAL4 DNA-binding domain (pGBT9) was expressed in combination with RPT6 E260A or K206A fused to the GAL4 activation domain. NtRPT6, Nicotiana tabacum RPT6; –LT, yeast growth on medium without Leu and Trp; –HLT, yeast growth on medium lacking His, Leu, and Trp, indicating expression of the HIS3 reporter gene. B, BiFC in planta interaction studies of XopJ and RPT6 mutants. YFP confocal microscopy images show N. benthamiana leaf epidermal cells transiently expressing XopJ-VenusN in combination with RPT6 K206A or E260A-VenusC. A closeup of the same cells shows that the YFP fluorescence of XopJ-VenusN/RPT6 E260A-VenusC aligns with the plasma membrane. XopJ-VenusN and RPT6-VenusC and the dimerization of fructose-1,6-bisphosphatase (FBPase) within the cytosol served as positive controls. XopJ-VenusN with FBPase-VenusC or the RPT6 mutants together with FBPase-VenusN are included as negative controls. Bars = 20 μm, except for the closeup (5 µm). C, EV or XopJ-HA was transiently coexpressed together with RPT6-GFP, RPT6 K206A-GFP, or RPT6 E260A-GFP in N. benthamiana using agroinfiltration. Samples were taken 48 hpi, and images were generated with identical CLSM settings. GFP fluorescence is shown in green, and chlorophyll autoflourescence is shown in red. Arrows indicate cytosolic strands. Bars = 20 μm. D, Quantification of GFP intensity after coexpression of RPT6-GFP, RPT6 K206A, and RPT6 E260A-GFP together with XopJ. GFP quantification was determined by using the ZEN software of Zeiss. Coexpression with the EV control was set to 100%. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks at P < 0.001. E, XopJ does not degrade the RPT6 Walker B mutant. Total proteins were extracted 48 hpi with agrobacteria harboring the respective XopJ and RPT6 expression constructs. RPT6-GFP, RPT6 K206-GFP, and RPT6 E260A-GFP protein levels were detected using an anti-GFP antibody. Expression of XopJ was verified using an anti-HA antibody. All experiments were repeated three times with similar results. To investigate in planta interactions, BiFC assays were performed using Agrobacterium tumefaciens-mediated transient expression in N. benthamiana. To this end, RPT6 (K206A) and RPT6 (E260A) mutants were each fused to the C-terminal part of the yellow fluorescent protein (YFP) derivative Venus (VenusC155) and transiently expressed in N. benthamiana leaves. Reconstitution of the fluorescence signal after transient coexpression with XopJ fused to the N-terminal 173 amino acids of Venus (VenusN173) indicated that XopJ interacts with the RPT6 Walker B mutant (E260A) at the plasma membrane but not with the RPT6 Walker A mutant (K206A) in planta (Fig. 5B). The interaction of XopJ and RPT6, as well as the homodimerization of the cytosolic FBPase, served as a positive control (Fig. 5B). Agroinfiltration of XopJ-VenusN173 together with FBPase-VenusC155 or FBPase-VenusN173 together with RPT6 (K206A) or (E260A)-VenusC155 served as negative controls and did not result in the reconstitution of fluorescence (Fig. 5B). Given the fact that the RPT6 Walker B mutant retained its ability to interact with XopJ, it was further investigated whether the interaction between both proteins leads to the XopJ-mediated degradation of RPT6 (E260A). Toward this end, coexpression studies in N. benthamiana of XopJ-HA together with the RPT6 wild type or its two mutant variants each fused to GFP were performed. A microscopic analysis revealed that XopJ was not able to recruit the RPT6 Walker A mutant to the plasma membrane and also did not lead to its degradation (Fig. 5C). This is likely owing to the lack of interaction between both proteins. In contrast, XopJ interacts with the RPT6 Walker B mutant and also clearly recruits it to the plasma membrane upon coexpression (Fig. 5C). Quantification of the GFP intensity showed that this variant is not destabilized by XopJ (Fig. 5D). Subsequent analysis of the protein levels by western blotting are consistent with the microscopic data (Fig. 5E) and indicate that, despite XopJ’s ability to interact with the RPT6 Walker B mutant, the effector is not able to trigger its degradation. XopJ Inhibits the Proteasome-Mediated Turnover of NPR1 Previous results demonstrated that XopJ-mediated inhibition of the proteasome interferes with SA-dependent defense responses and that this effect is dependent on NPR1 (Üstün et al., 2013). The transcriptional coactivator NPR1 has been shown to be continuously cleared by the proteasome in order to perpetuate SA-responsive defense signaling (Spoel et al., 2009). Transient expression of XopJ leads to the accumulation of ubiquitinated proteins, probably triggered by the inhibition of the proteasome (Üstün et al., 2013). Likewise, virus-induced gene silencing of RPT6 also causes a strong inhibition of the proteasome and the accumulation of ubiquitinated proteins and the transcriptional coactivator NPR1. Thus, the degradation of RPT6 likely interferes with the turnover of ubiquitinated proteins (Supplemental Fig. S7). Examination of endogenous NPR1 protein levels using an anti-NPR1 specific antibody revealed that transient expression of XopJ resulted in the accumulation of NPR1 protein comparable to that observed after treatment with the proteasome inhibitor MG132 (Fig. 6A). In contrast, N. benthamiana leaves transiently expressing the G2A or C235A mutant variant of XopJ did not show an increase in NPR1 protein abundance (Fig. 6A). To test whether XopJ triggers the accumulation of ubiquitinated NPR1, NPR1-GFP was transiently coexpressed with XopJ. This also led to an accumulation of the NPR1-GFP fusion protein (Supplemental Fig. S8). Subsequently, NPR-GFP was pulled down using GFP-trap beads, and western-blot analysis of the precipitates using an anti-GFP antibody showed that transient expression of XopJ, but not of C235A, caused an accumulation of NPR1-GFP similar to the MG132 positive control (Fig. 6B). Analysis of the ubiquitination status of the different precipitates revealed that transiently expressed XopJ enhanced NPR1 ubiquitination, proven by the accumulation of the characteristic ubiquitin smear (Fig. 6B). The XopJ C235A variant was not able to induce an accumulation of NPR1-GFP and hence did not enhance NPR1 protein ubiquitination (Fig. 6B). Figure 6. Open in new tabDownload slide XopJ inhibits the proteasome-mediated turnover of NPR1. A, XopJ-triggered accumulation of NPR1 protein levels. XopJ protein variants along with an EV control were transiently expressed in leaves of N. benthamiana using agroinfiltration. MG132 treatment for 6 h was included into the analysis as a positive control. The western blot was probed 48 hpi with an anti-NPR1 antibody directed against the N. tabacum NPR1 protein. The intensities of NPR1 bands were quantified by ImageJ and are shown at the bottom of the top gel. An anti-myc antibody was used to show the proper expression of XopJ proteins. Amido Black staining shows equal protein loading. B, XopJ leads to the accumulation of ubiquitinated NPR1. NPR1-GFP was transiently expressed together with EV, XopJ-HA, or XopJ C235A-HA in N. benthamiana. MG132 treatment served as a positive control and was performed as in A. Samples were taken 48 hpi, and total proteins (Input) were subjected to immunoprecipitation (IP) with GFP-Trap beads, followed by immunoblot analysis of the precipitates using either anti-GFP or anti-ubiquitin antibodies. The proper expression of XopJ proteins in the input fraction was monitored using an anti-HA antibody. C, XopJ affects NPR1 protein levels during Xcv infection in pepper. Xcv or XcvƊxopJ was inoculated at a bacterial density of 2 × 108 colony-forming units mL−1 into leaves of cv Early Cal Wonder plants. MG132 treatment served as a positive control. NPR1 protein levels were detected by an anti-NPR1 antibody. D, XopJ also comprises proteasome activity in the nucleus. Nuclear fractionation was performed after transient expression of XopJ-HA in N. benthamiana. Crude plant total extracts (TE) were separated into nucleus-depleted (ND) and nucleus-enriched (NE) fractions. Proteasome activity was measured in each fraction by monitoring the breakdown of the fluorogenic peptide Suc-LLVY-AMC at 30°C. The EV control was set to 100%. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks: *, P < 0.05; and **, P < 0.01. E, Anti-SPP2 and anti-histone H3 antibodies were used as markers for cytoplasmic and nuclear proteins, respectively. Expression of XopJ was confirmed by an anti-HA antibody. All experiments were repeated three times with similar results. Figure 6. Open in new tabDownload slide XopJ inhibits the proteasome-mediated turnover of NPR1. A, XopJ-triggered accumulation of NPR1 protein levels. XopJ protein variants along with an EV control were transiently expressed in leaves of N. benthamiana using agroinfiltration. MG132 treatment for 6 h was included into the analysis as a positive control. The western blot was probed 48 hpi with an anti-NPR1 antibody directed against the N. tabacum NPR1 protein. The intensities of NPR1 bands were quantified by ImageJ and are shown at the bottom of the top gel. An anti-myc antibody was used to show the proper expression of XopJ proteins. Amido Black staining shows equal protein loading. B, XopJ leads to the accumulation of ubiquitinated NPR1. NPR1-GFP was transiently expressed together with EV, XopJ-HA, or XopJ C235A-HA in N. benthamiana. MG132 treatment served as a positive control and was performed as in A. Samples were taken 48 hpi, and total proteins (Input) were subjected to immunoprecipitation (IP) with GFP-Trap beads, followed by immunoblot analysis of the precipitates using either anti-GFP or anti-ubiquitin antibodies. The proper expression of XopJ proteins in the input fraction was monitored using an anti-HA antibody. C, XopJ affects NPR1 protein levels during Xcv infection in pepper. Xcv or XcvƊxopJ was inoculated at a bacterial density of 2 × 108 colony-forming units mL−1 into leaves of cv Early Cal Wonder plants. MG132 treatment served as a positive control. NPR1 protein levels were detected by an anti-NPR1 antibody. D, XopJ also comprises proteasome activity in the nucleus. Nuclear fractionation was performed after transient expression of XopJ-HA in N. benthamiana. Crude plant total extracts (TE) were separated into nucleus-depleted (ND) and nucleus-enriched (NE) fractions. Proteasome activity was measured in each fraction by monitoring the breakdown of the fluorogenic peptide Suc-LLVY-AMC at 30°C. The EV control was set to 100%. Data represent means ± sd (n = 3). Significant differences were calculated using Student’s t test and are indicated by asterisks: *, P < 0.05; and **, P < 0.01. E, Anti-SPP2 and anti-histone H3 antibodies were used as markers for cytoplasmic and nuclear proteins, respectively. Expression of XopJ was confirmed by an anti-HA antibody. All experiments were repeated three times with similar results. In order to investigate whether XopJ can promote the stabilization of NPR1 during bacterial infection, susceptible pepper plants were infected with Xcv wild-type bacteria or an XcvƊxopJ deletion strain (Üstün et al., 2013). When NPR1 levels were monitored by immunoblotting at 3 dpi, leaves infected with wild-type Xcv showed a stronger NPR1 signal than those infected with XcvƊxopJ (Fig. 6C). This indicates that inhibition of the proteasome by XopJ impedes the turnover of NPR1 during Xcv infection of pepper plants. Degradation of NPR1 is supposed to take place in the nucleus (Spoel et al., 2009), while XopJ is localized at the plasma membrane (Bartetzko et al., 2009). A cellular fractionation experiment of N. benthamiana leaves transiently expressing XopJ was performed to assess the inhibition of the proteasome by XopJ at spatial resolution. Measurement of proteasome activity in total extracts and nucleus-depleted and nucleus-enriched fractions revealed that XopJ-mediated inhibition of the proteasome occurred in each fraction (Fig. 6D). To monitor the proper separation of nuclear and cytosolic fractions, blots were additionally probed with antibodies against the cytosolic Suc phosphatase (Chen et al., 2005) and the nucleus-specific histone H3 (Fig. 6E). These data demonstrate that XopJ affects the activity of the proteasome in the cytosol as well as in the nucleus. DISCUSSION Adapted phytopathogenic bacteria are able to suppress plant innate immunity and to reprogram cellular pathways to promote bacterial multiplication and thus cause disease in host plants. During infection, bacterial T3Es, translocated by the type III secretion system, play a central role in the manipulation of the host cellular machinery. In general, these T3Es are essential for pathogen virulence by interfering with plant processes involved in defense responses (Jones and Dangl, 2006). Although a range of studies have identified T3E target proteins, the enzymatic functions of many T3Es and their mode of action on their respective targets in plants are not well understood. We have previously shown that the Xcv T3E XopJ targets the proteasome subunit RPT6 to suppress SA-mediated defense signaling and hence promotes the survival of bacteria during infection (Üstün et al., 2013). However, the underlying mechanism of XopJ-triggered proteasome inhibition remained unknown. In this study, we show that XopJ acts as a protease to degrade the proteasome subunit RPT6 in plant cells. Destabilization of RPT6 negatively impacts on proteasome function, affecting the proteasomal turnover of NPR1. This provides a mechanistic link between XopJ and its effect on SA-dependent defense responses during the Xcv infection of pepper plants. XopJ from Xcv belongs to the YopJ superfamily of T3E proteins that are present in a wide range of animal and plant bacterial pathogens and symbionts, respectively (Lewis et al., 2011). YopJ-like effector proteins contain a conserved catalytic triad consisting of His, Glu, and Cys residues. Mutation of the Cys residue interferes with the virulence or avirulence activity of these effectors, indicating that the enzymatic function is necessary for effector function within the host cell (Lewis et al., 2011). Recent studies revealed that several YopJ-like effector proteins possess auto- or transacetyltransferase activity (Lee et al., 2012; Jiang et al., 2013; Cheong et al., 2014). However, based on their secondary structure, it was initially hypothesized that these effectors are Cys proteases (Lewis et al., 2011), although potential substrate proteins have not yet been identified. HopZ1a from P. syringae pv syringae as well as AvrBsT from Xcv have been shown to possess weak protease activity in vitro when casein was used as a generic substrate (Ma et al., 2005; Szczesny et al., 2010). In addition, HopZ1a is able to destabilize 2-hydroxy isoflavone dehydratase or Jasmonate-Zim-Domain Protein1 (JAZ1) in soybean (Glycine max) host plants, although the biochemical mechanism of this destabilization is not clear (Zhou et al., 2011; Jiang et al., 2013). XopJ has no detectable acetyltransferase activity in vitro but leads to the degradation of its target protein, RPT6, inside plant cells. The findings presented here apparently contradict our previous reports, where we could not observe differences in RPT6 protein levels upon coexpression with XopJ (Üstün et al., 2013). However, in contrast to our previous approach, we now recorded confocal images of fluorescently labeled RPT6, expressed either alone or together with XopJ, using the same microscope settings. Using the same sensitivity allows direct comparison between images and clearly brings out the differences in fluorescence signal intensity that reflect the reduction in RPT6 protein levels upon the coexpression of XopJ. Similarly, carefully controlling equal protein loading and shorter exposition times robustly reveal differences in RPT6 protein accumulation in transient coexpression experiments as compared with leaves expressing RPT6 and an EV control. Thus, the novel results presented here are in no way contradictory to the previous findings of Üstün et al. (2013) but represent a more differentiated reassessment of the effect of XopJ coexpression on the RPT6 protein level. Several lines of evidence support the notion that XopJ acts as a Cys protease to specifically degrade RPT6: (1) the degradation of RPT6 in planta is abolished in the presence of a protease inhibitor cocktail; (2) the recombinant E. coli-produced MBP-XopJ cleaves peptides within a generic substrate library in vitro, which is inhibited by the Cys protease inhibitor E-64; and (3) plant-expressed XopJ is able to degrade recombinant RPT6 and vice versa. In addition, the degradation of RPT6 was dependent on XopJ’s catalytic Cys residue (Cys-235), which is in line with previous results showing that the XopJ C235A mutant is not able to suppress proteasome activity and also does not delay the development of tissue necrosis in pepper plants when delivered by the type III secretion system of virulent Xcv (Üstün et al., 2013). Intriguingly, mutation in the myristoylation motif of XopJ by exchanging the Gly at position 2 to Ala affected XopJ’s ability to destabilize RPT6 in planta, indicating that the protease activity of in planta-expressed XopJ requires effector localization at the plant plasma membrane. This might indicate that a yet unidentified host cell factor or posttranslational modification of the effector is necessary for its activity. In contrast, measurement of the in vitro protease activity of E. coli-purified XopJ revealed that XopJ has detectable protease activity toward a peptide library as well as toward recombinantly produced RPT6 in the absence of a possible eukaryotic host factor. We currently have no clear explanation for this discrepancy; however, there could be quantitative differences in activity between XopJ in plant cells and recombinant protein in vitro that are associated with a modification of the effector by the host cell machinery. In order to elaborate this further, future experiments will aim to characterize the protease activity of XopJ purified from plants. Furthermore, the XopJ-triggered degradation of RPT6 did not produce any detectable cleavage products, as reported for other protease effectors such as AvrPphB or AvrRpt2 from P. syringae (Shao et al., 2003; Chisholm et al., 2005). It is possible that RPT6 degradation products escape detection by the antibody directed against the respective fusion partner of RPT6 or that multiple XopJ cleavage sites are present in RPT6 and the resulting small peptide fragments that are not visible in our degradation assays. Although the XopJ cleavage site within the RPT6 polypeptide chain is currently not known, a mutational analysis of RPT6 revealed some structural requirements for XopJ binding and degradation. RPT6 belongs to the AAA-ATPase family of proteins whose members are involved in a range of cellular processes and generally function by inducing conformational changes in substrate proteins during continuous cycles of nucleotide binding and hydrolysis (Hanson and Whiteheart, 2005). In concert with RPT1 to RPT5, RPT6 forms a hexameric ring that has direct contact with the 20S catalytic core of the proteasome and that is involved in substrate binding, opening of the gated channel in the 20S subunit, unfolding of proteins, and facilitating the translocation of the unfolded substrate through the AAA-ATPase ring into the 20S particle (Vierstra, 2009; Bar-Nun and Glickman, 2012). Substrate unfolding is the only process that requires the hydrolysis of ATP, while the other steps only depend on ATP binding (Benaroudj et al., 2003; Smith et al., 2004). Like other AAA-ATPase family members, RPT6 possesses so-called Walker A and Walker B motifs as integral parts of its ATP-binding site (Hanson and Whiteheart, 2005). The Walker A motif interacts directly with the ATP phosphates, and a mutation within this motif typically eliminates nucleotide binding and inactivates the AAA-ATPase protein (Hanson and Whiteheart, 2005). The Walker B motif is crucial for ATP hydrolysis, and a mutation within this motif blocks ATP hydrolysis but not binding (Babst et al., 1998; Weibezahn et al., 2003; Dalal et al., 2004). As ATP binding but not hydrolysis is required for substrate binding by most, if not all, AAA-ATPases, mutations in the Walker B motif have been used to create substrate traps that bind but cannot release substrates (Babst et al., 1998; Weibezahn et al., 2003; Dalal et al., 2004). Mutation of the Walker A motif of RPT6 abolishes is ability to interact with XopJ; thus, the Walker A RPT6 mutant protein is not degraded by the effector. This indicates that ATP binding to RPT6 is required for its recognition by XopJ, because it has been shown that ATP-bound RPT6 represents the active form of the protein. The observation that a mutation of the Walker B motif does not affect XopJ binding but prevents RPT6 from being proteolytically degraded suggests that XopJ binds RPT6 similar to substrate proteins during proteasomal degradation and, hence, could act as a substrate mimic. The conformational change imposed during ATP hydrolysis might be necessary for the full activation of XopJ’s protease activity or may be required to expose a possible cleavage site within the RPT6 protein that is recognized by the effector. Our experiments, including the destabilization of RPT6 by XopJ and virus-induced gene silencing of RPT6 in N. benthamiana, consistently suggest that XopJ triggers RPT6 degradation to reduce proteasome activity in plants, eventually leading to the accumulation of ubiquitinated proteins. Although most, if not all, RP subunits are essential for proteasome function in plants, analyses of weak mutant alleles for several RP subunits in Arabidopsis indicate that some have substrate-specific functions (Vierstra, 2009). Thus, it is conceivable that RPT6 has a specific function in proteasomal protein turnover during plant defense. In addition, RPT6 plays important roles in 26S proteasome assembly in other eukaryotes (Ehlinger et al., 2013; Park et al., 2013). XopJ-triggered destabilization of RPT6 would prevent one of the earliest steps in 19S RP assembly (Tomko and Hochstrasser, 2013). As a consequence of a defective 19S RP assembly, ubiquitinated proteins would not be recognized and not directed to enter the proteolytic channel of the 20S CP, leading to an accumulation of ubiquitinated proteins, as observed during transient expression of XopJ in N. benthamiana (Üstün et al., 2013) or in RPT6-silenced N. benthamiana plants. Loss of other AAA-ATPase proteasome subunits, such as RPT2 in Arabidopsis, also results in a decreased 26S complex stability and, hence, inhibition of the proteasome activity (Lee et al., 2011), indicating the overall importance of RP AAA-ATPases during proteasome assembly. Proteasome assembly itself is a complex process that is not fully understood in time and space (Tomko and Hochstrasser, 2013). Although it was shown that CP and RP components are imported separately into the nucleus, a recent study demonstrated that the 26S proteasome completes its assembly process in the cytoplasm and is able to enter the nucleus as a holocomplex (Pack et al., 2014). Thus, although XopJ itself is not a nuclear protein, its interference with proteasome assembly outside the nucleus eventually inhibits proteasome activity in the cytoplasm as well as in the nucleus, because it prevents the nuclear import of functional proteasome complexes. Previously, it was shown that NPR1, the master regulator of SA signaling, must be constitutively cleared by the proteasome in the nucleus to perpetuate SA-responsive gene expression (Spoel et al., 2009). Indeed, XopJ prevents the proteasomal turnover of NPR1 in a myristoylation- and catalytic triad-dependent manner. This also holds true during a compatible interaction of Xcv with pepper plants, because NPR1 accumulates in wild-type infected leaves but not in those inoculated with an XcvƊxopJ knockout mutant. This provides a mechanistic link to previous findings that XopJ inhibits downstream SA responses (Üstün et al., 2013). Emerging data suggest that phytopathogenic bacteria developed T3Es to interfere with SA-dependent defense signaling. Based on its ability to induce programmed cell death in host plants, SA is considered as the central regulator of plant immunity against biotrophic and hemibiotrophic pathogens (Pieterse et al., 2012). Additionally, basal defense mechanisms such as bacteria-induced closing of stomata and callose deposition at the cell wall are partially dependent on SA (DebRoy et al., 2004; Melotto et al., 2006). Consequently, some T3Es are able to manipulate SA signaling, either directly or indirectly. For instance, XopD from Xcv acts as a sumo-protease to impair ethylene signaling and indirectly affects SA signaling (Kim et al., 2013). Moreover, the P. syringae effector protein HopZ1a directly targets and acetylates JAZs, the negative regulators of jasmonic acid (JA) signaling, to antagonize SA-dependent defense as a consequence of the activation of JA signaling (Jiang et al., 2013). Another effector from P. syringae, HopX1, proteolytically degrades JAZ proteins to induce JA signaling and, hence, leads to the repression of SA signaling (Gimenez-Ibanez et al., 2014), providing further evidence that targeting hormonal signaling is an attractive strategy for plant pathogenic bacteria to inhibit plant immunity. Like other T3Es targeting hormonal signaling, the Xcv effector XopJ also manipulates SA signaling indirectly through the inhibition of the proteasome via the degradation of the proteasome subunit RPT6. As a consequence, proteasomal turnover of NPR1 is affected and leads to the accumulation of ubiquitinated NPR1. As the turnover of NPR1 is also required for the establishment of SAR, it is tempting to speculate that XopJ could play a role in the suppression of SAR during Xcv infection in pepper. A striking example for a bacterial proteasome-inhibiting toxin affecting SAR is the SylA peptide secreted by certain P. syringae strains (Misas-Villamil et al., 2013). Similar to XopJ, SylA inhibits the proteasome activity in plant cells, although through a different mechanism (Groll et al., 2008). SylA blocks SA signaling to suppress SAR and thus creates a zone of SA-insensitive tissue to promote spreading of the bacteria from infection sites (Misas-Villamil et al., 2013). Whether SylA is also able to prevent the proteasomal turnover of NPR1 has to be analyzed in future studies. However, recent data demonstrate that SylA promotes the accumulation of ubiquitinated proteins in Arabidopsis, indicating that SylA could also affect the turnover of regulators of SA signaling (Svozil et al., 2014). Manipulation of the ubiquitin proteasome system has emerged as a new virulence strategy of bacterial invaders to promote pathogenesis (Dudler, 2013). Recent advances in the plant immunity field revealed that several components of the ubiquitin proteasome system are required for plant immunity or exploited by pathogens (Marino et al., 2012). By showing that XopJ possesses protease activity resulting in the degradation of RPT6, we provide evidence for how XopJ disables the proteasome function. The general interference of XopJ with the turnover of ubiquitinated proteins also impedes the proteasomal turnover of NPR1, explaining our previous finding that XopJ interferes with SA-dependent defense responses (Üstün et al., 2013). MATERIALS AND METHODS Plant Material and Growth Conditions Pepper (Capsicum annuum ‘Early Cal Wonder’) and Nicotiana benthamiana plants were grown in soil in a greenhouse with daily watering and subjected to a 16-h-light/8-h-dark cycle (25°C/21°C) at 300 µmol m–2 s–1 light and 75% relative humidity. Infection of Pepper Plants Xanthomonas campestris pv vesicatoria infections for western-blot analysis of endogenous NPR1 protein levels were performed as described previously (Üstün et al., 2013). Site-Directed Mutagenesis Site-directed mutagenesis of RPT6 constructs was carried out using the Quick-Change site-directed mutagenesis kit (Stratagene) employing the primers listed in Supplemental Table S1. All base changes were verified by sequencing. Yeast Two-Hybrid Analysis Yeast two-hybrid techniques were performed according to the Yeast Protocols Handbook and the Matchmaker GAL4 Two-Hybrid System 3 manual (both Clontech). Point mutation variants of RPT6 were generated by site-directed mutagenesis in the vector pGAD424 (Clontech). For the generation of the HsRPT6 activation domain fusions, the coding region was amplified by PCR from complementary DNA (cDNA) derived from HeLa cells using the primers listed in Supplemental Table S1, inserted into the vector pGAD424 (Clontech), and sequence verified. Direct interaction of two proteins was investigated by cotransformation of the respective plasmids in the yeast strain AH109, followed by selection of transformants on medium lacking Leu and Trp at 30°C for 3 d and subsequent transfer to medium lacking Leu, Trp, and His for growth selection and β-galactosidase activity testing of interacting clones. Plasmid Construction for Transient Expression Experiments Construction of the binary vectors expressing XopJ and its mutant variants XopJ G2A and C235A was described previously (Bartetzko et al., 2009; Üstün et al., 2014). The RPT6 K206A and E260A-GFP constructs were generated by site-directed mutagenesis of the pENTR-D/TOPO (Invitrogen) clones. The HsRPT6-GFP construct was assembled by amplifying the entire coding region from cDNA from HeLa cells, while the NPR1-GFP construct was generated by amplifying the entire coding region from N. benthamiana cDNA, using the primers listed in Supplemental Table S1. The resulting PCR fragments were inserted in the pENTR-D/TOPO (Invitrogen) clones. Entry clones were subsequently recombined into pK7WGF2 (Karimi et al., 2002) using L/R Clonase (Invitrogen). BiFC Assay Entry clones of RPT6 (K206A), RPT6 (E260A), and HsRPT6 comprising the entire coding region of each cDNA were cloned in a Gateway system (Invitrogen)-compatible version of the BiFC vector pRB35S-GW-VenusC. Constructs were transformed into Agrobacterium tumefaciens C58C1 and transiently expressed by agroinfiltration in N. benthamiana. The BiFC-induced YFP fluorescence was detected by CLSM (LSM510; Zeiss) after 48 hpi. The specimens were examined using the LD LCI Plan-Apochromat 25×/0.8 water-immersion objective for detailed images with excitation using the argon laser (458- or 488-nm line for BiFC and chlorophyll autofluorescence). The emitted light passed the primary beam-splitting mirrors at 458/514 nm and was separated by a secondary beam splitter at 515 nm. Fluorescence was detected with filter sets as follows: on channel 3, 530 to 560 band pass; and on channel 1, for red autofluorescence of chlorophyll. Microscopic Analysis Coexpression studies were performed as described previously (Bartetzko et al., 2009) using a Leica TCS SP5II or Zeiss LSM510 confocal microscope. For quantification, images were generated with identical CLSM settings (i.e. detector gain, optical slice, scanning time, and magnification) using multiple infiltrated leaves with at least three independent repetitions of A. tumefaciens infiltration. GFP quantification was determined by using Leica software LAS_AF and Zeiss software ZEN. Transient Expression Assays and Inhibitory Studies For infiltration of N. benthamiana leaves, A. tumefaciens C58C1 was infiltrated into the abaxial airspace of 4- to 6-week-old plants using a needleless 2-mL syringe. Agrobacteria were cultivated overnight at 28°C in the presence of appropriate antibiotics. The cultures were harvested by centrifugation, and the pellet was resuspended in sterile water to a final optical density at 600 nm of 1. The cells were used for the infiltration directly after resuspension. Infiltrated plants were further cultivated in the greenhouse with daily watering and subjected to a 16-h-light/8-h-dark cycle (25°C/21°C) at 300 µmol m–2 s–1 light and 75% relative humidity. For inhibitory studies, 100 µm MG132 or 1% (v/v) ethanol, 1× cOmplete ULTRA Protease Inhibitors (Roche Applied Science), or water was infiltrated into A. tumefaciens-inoculated N. benthamiana leaves at 42 hpi, and the leaves were collected at 48 hpi. Western Blotting Leaf material was homogenized in SDS-PAGE loading buffer (100 mm Tris-HCl, pH 6.8, 9% [v/v] β-mercaptoethanol, 40% [v/v] glycerol, 0.0005% [w/v] Bromphenol Blue, and 4% [w/v] SDS) and, after heating for 10 min at 95°C, subjected to gel electrophoresis. Separated proteins were transferred onto nitrocellulose membranes (Porablot; Macherey-Nagel). Proteins were detected by an anti-HA-peroxidase high-affinity antibody (Roche), anti-myc-peroxidase antibody (Roche), anti-GFP antibody (Roche), anti-SPP2 antibody (Chen et al., 2005), anti-histone H3 antibody (Sigma), anti-NtNPR1 antibody, or anti-ubiquitin antibody (Agrisera) via chemiluminescence (GE Healthcare). The anti-NtNPR1 serum (kindly provided by Ursula Pfitzner, University of Hohenheim) was generated by immunizing rabbits with a recombinant GST fusion protein comprising the conserved SA-sensitive C terminus with amino acids 386 to 588 of Nicotiana tabacum NPR1 (Maier et al., 2011; GenBank accession no. AF480488_1). This portion of the NtNPR1 protein has 95% identity to the respective part of NPR1 from pepper (GenBank accession no. ABG38308.1). Measurement of Proteasome Activity Proteasome activity in crude plant extracts, or nucleus-depleted and -enriched fractions, was determined spectrofluorometrically using the fluorogenic substrate Suc-LLVY-AMC (Sigma) according to Üstün et al. (2013). For semi-in vitro proteasome inhibition assays, 5 µg of Escherichia coli-purified recombinant MBP-XopJ, MBP-C235A, or MBP was mixed with 50 µg of crude plant extract from N. benthamiana. The reaction was started after 1 h at 30°C by the addition of 0.2 mm Suc-LLVY-AMC. Released amino-methyl-coumarin was measured every 2 min between 0 and 120 min using a fluorescence spectrophotometer (Synergy HT; BioTek), with an excitation wavelength of 360 nm and an emission wavelength of 460 nm. Proteasome activity was calculated from the linear slope of the emission curve and is expressed as fluorescence units per minute or in percentage relative to controls. Protease Activity Assays For in vitro protease assays, MBP, MBP-XopJ, and MBP-XopJ C235A were expressed in E. coli M15 cells. Bacteria were lysed by sonication. After centrifugation, the recombinant proteins were purified using amylose resin (New England Biolabs) according to the manufacturer’s instructions. The purity of the proteins was approximately 90% as analyzed by SDS-PAGE and Coomassie Blue staining. For in vivo protease activity, EV, XopJ-HA, and XopJ C235A-HA were expressed in N. benthamiana by agroinfiltration, and harvested leaf material was homogenized in 200 µL of extraction buffer (50 mm HEPES-KOH, pH 7.2, 2 mm ATP, 2 mm dithiothreitol, and 250 mm Suc). After centrifugation, the protein concentration of the supernatant was adjusted to 1 mg mL−1 with extraction buffer. Measurement of in vitro and in vivo protease activity was performed according to the manufacturer’s instructions using the P-CHECK protease detection kit (Jena Bioscience). Briefly, 5 µg of E. coli purified recombinant protein or 50 µg of plant extract was mixed with working buffer I (pH 7.4) and 10 µL of P-CHECK substrate solution. Proteinase K served as a positive control. In control experiments, 1× cOmplete ULTRA Protease Inhibitors or specific inhibitor substances (Roche Applied Science) were added to the reaction and incubated for 1 h before the protease reaction was started by the addition of the P-CHECK substrate. Measurement was carried out at 37°C for 2 h using a fluorescence spectrophotometer (Synergy HT; BioTek), with an excitation wavelength of 320 nm and an emission wavelength of 405 nm. Protease activity was calculated from the linear slope of the emission curve and is expressed as fluorescence units per minute. For semi-in vitro degradation assays, 1 µg of E. coli purified recombinant proteins (MBP, MBP-RPT6, MBP-XopJ, and MBP-XopJ C235A) was mixed with crude plant extracts expressing EV, XopJ-HA, XopJ C235A-HA, or RPT6-GFP (with or without 1× cOmplete ULTRA Protease Inhibitors) and incubated overnight at 37°C with moderate shaking. Reactions were stopped by adding 4× SDS loading buffer and examined via western blotting. Protein Extraction and GFP Pull Down in N. benthamiana Approximately 1 g of leaf material was ground to a fine powder in liquid nitrogen, and 5 mL of extraction buffer (50 mm Tris-HCl, pH 7.5, 150 mm NaCl, 10% [v/v] glycerol, 10 mm dithiothreitol, 10 mm EDTA, 1 mm NaF, 1 mm Na2MoO4∙2H2O; 1% [w/v] polyvinylpolypyrrolidone, 1% [v/v] 1× cOmplete ULTRA Protease Inhibitor cocktail [Roche Applied Science], and 1% [v/v] Nonidet P-40) was added. Samples were cleared by centrifugation at 16,000g for 15 min at 4°C and adjusted to 2 mg mL−1 total protein concentration. Immunoprecipitation was performed on 1.5 mL of total protein by adding 20 µL of GFP Trap-M beads (Chromotek) and incubation at 4°C for 2 h. Beads were washed four times with Tris-buffered saline containing 0.5% (v/v) Nonidet P-40, and immunoprecipitates were eluted with 30 µL of 2× SDS loading buffer and heating at 70°C for 10 min. Nuclear Fractionation Plant cell fractionation was performed by using the CelLytic PN Isolation/Extraction Kit for plant leaves (Sigma) according to the manufacturer’s instructions. RNA Extraction and Expression Analysis Total RNA was isolated from leaf material and then treated with RNase-free DNase (Fermentas) to degrade any remaining DNA. First strand cDNA synthesis was performed from 2 µg of total RNA using Revert-Aid reverse transcriptase (Fermentas). For RT-PCR, cDNAs were amplified using Taq polymerase (New England Biolabs) and gene-specific primers (Supplemental Table S1). Virus-Induced Gene Silencing of N. benthamiana Virus-induced gene silencing was performed as described previously (Üstün et al., 2012). Briefly, A. tumefaciens strains with the pTRV1 vector and with pTRV2-GFPsil and PYL279-RPT6 (Üstün et al., 2013; optical density at 600 nm = 1) were mixed in a 1:1 ratio, and the mixture was infiltrated into a lower leaf of a 4-week-old N. benthamiana plant using a 1-mL sterile syringe without a needle. Silenced plants were analyzed 14 d post infiltration. Supplemental Data The following supplemental materials are available. Supplemental Figure S1 . XopJ does not display acetyltransferase activity. Supplemental Figure S2 . Cys protease XopD does not comprise RPT6 accumulation. Supplemental Figure S3 . XopJ does not interact with RPT6 from humans. Supplemental Figure S4 . XopJ exhibits protease activity in a protease-specific manner. Supplemental Figure S5 . Protein inputs used for semi-in vitro degradation assays to demonstrate comparable MBP-RPT6 protein amounts prior to the incubation with extracts from XopJ-HA expressing leaves. Supplemental Figure S6 . MBP-XopJ degrades RPT6 and inhibits proteasome activity. Supplemental Figure S7 . Virus-induced gene silencing of RPT6 in N. benthamiana mimics XopJ-triggered effects. Supplemental Figure S8 . XopJ triggers accumulation of NPR1-GFP. Supplemental Table S1 . Oligonucleotides used in this study. ACKNOWLEDGMENTS We thank Ursula Pfitzner for providing the anti-NtNPR1 antibody, and Susanne Jeserigk, Cornelia Symowski, and Patrick König for excellent technical assistance. Glossary SAR systemic acquired resistance SA salicylic acid CP core particle RP regulatory particle T3E type III effector SylA syringolin A RT reverse transcription Xcv Xanthomonas campestris pv vesicatoria BiFC bimolecular fluorescence complementation hpi hours post inoculation EV empty vector JA jasmonic acid cDNA complementary DNA YFP yellow fluorescent protein CLSM confocal laser scanning microscopy LITERATURE CITED Babst M , Wendland B, Estepa EJ, Emr SD ( 1998 ) The Vps4p AAA ATPase regulates membrane association of a Vps protein complex required for normal endosome function . EMBO J 17 : 2982 – 2993 Google Scholar Crossref Search ADS PubMed WorldCat Bar-Nun S , Glickman MH ( 2012 ) Proteasomal AAA-ATPases: structure and function . Biochim Biophys Acta 1823 : 67 – 82 Google Scholar Crossref Search ADS PubMed WorldCat Bartetzko V , Sonnewald S, Vogel F, Hartner K, Stadler R, Hammes UZ, Börnke F ( 2009 ) The Xanthomonas campestris pv. vesicatoria type III effector protein XopJ inhibits protein secretion: evidence for interference with cell wall-associated defense responses . Mol Plant Microbe Interact 22 : 655 – 664 Google Scholar Crossref Search ADS PubMed WorldCat Benaroudj N , Zwickl P, Seemüller E, Baumeister W, Goldberg AL ( 2003 ) ATP hydrolysis by the proteasome regulatory complex PAN serves multiple functions in protein degradation . Mol Cell 11 : 69 – 78 Google Scholar Crossref Search ADS PubMed WorldCat Chen S , Hajirezaei M, Peisker M, Tschiersch H, Sonnewald U, Börnke F ( 2005 ) Decreased sucrose-6-phosphate phosphatase level in transgenic tobacco inhibits photosynthesis, alters carbohydrate partitioning, and reduces growth . Planta 221 : 479 – 492 Google Scholar Crossref Search ADS PubMed WorldCat Cheong MS , Kirik A, Kim JG, Frame K, Kirik V, Mudgett MB ( 2014 ) AvrBsT acetylates Arabidopsis ACIP1, a protein that associates with microtubules and is required for immunity . PLoS Pathog 10 : e1003952 Google Scholar Crossref Search ADS PubMed WorldCat Chisholm ST , Dahlbeck D, Krishnamurthy N, Day B, Sjolander K, Staskawicz BJ ( 2005 ) Molecular characterization of proteolytic cleavage sites of the Pseudomonas syringae effector AvrRpt2 . Proc Natl Acad Sci USA 102 : 2087 – 2092 Google Scholar Crossref Search ADS PubMed WorldCat Dalal S , Rosser MF, Cyr DM, Hanson PI ( 2004 ) Distinct roles for the AAA ATPases NSF and p97 in the secretory pathway . Mol Biol Cell 15 : 637 – 648 Google Scholar Crossref Search ADS PubMed WorldCat DebRoy S , Thilmony R, Kwack YB, Nomura K, He SY ( 2004 ) A family of conserved bacterial effectors inhibits salicylic acid-mediated basal immunity and promotes disease necrosis in plants . Proc Natl Acad Sci USA 101 : 9927 – 9932 Google Scholar Crossref Search ADS PubMed WorldCat Dudler R ( 2013 ) Manipulation of host proteasomes as a virulence mechanism of plant pathogens . Annu Rev Phytopathol 51 : 521 – 542 Google Scholar Crossref Search ADS PubMed WorldCat Ehlinger A , Park S, Fahmy A, Lary JW, Cole JL, Finley D, Walters KJ ( 2013 ) Conformational dynamics of the Rpt6 ATPase in proteasome assembly and Rpn14 binding . Structure 21 : 753 – 765 Google Scholar Crossref Search ADS PubMed WorldCat Fu ZQ , Dong X ( 2013 ) Systemic acquired resistance: turning local infection into global defense . Annu Rev Plant Biol 64 : 839 – 863 Google Scholar Crossref Search ADS PubMed WorldCat Galán JE , Wolf-Watz H ( 2006 ) Protein delivery into eukaryotic cells by type III secretion machines . Nature 444 : 567 – 573 Google Scholar Crossref Search ADS PubMed WorldCat Gimenez-Ibanez S , Boter M, Fernández-Barbero G, Chini A, Rathjen JP, Solano R ( 2014 ) The bacterial effector HopX1 targets JAZ transcriptional repressors to activate jasmonate signaling and promote infection in Arabidopsis . PLoS Biol 12 : e1001792 Google Scholar Crossref Search ADS PubMed WorldCat Groll M , Schellenberg B, Bachmann AS, Archer CR, Huber R, Powell TK, Lindow S, Kaiser M, Dudler R ( 2008 ) A plant pathogen virulence factor inhibits the eukaryotic proteasome by a novel mechanism . Nature 452 : 755 – 758 Google Scholar Crossref Search ADS PubMed WorldCat Hanson PI , Whiteheart SW ( 2005 ) AAA+ proteins: have engine, will work . Nat Rev Mol Cell Biol 6 : 519 – 529 Google Scholar Crossref Search ADS PubMed WorldCat Hatsugai N , Iwasaki S, Tamura K, Kondo M, Fuji K, Ogasawara K, Nishimura M, Hara-Nishimura I ( 2009 ) A novel membrane fusion-mediated plant immunity against bacterial pathogens . Genes Dev 23 : 2496 – 2506 Google Scholar Crossref Search ADS PubMed WorldCat Jiang S , Yao J, Ma KW, Zhou H, Song J, He SY, Ma W ( 2013 ) Bacterial effector activates jasmonate signaling by directly targeting JAZ transcriptional repressors . PLoS Pathog 9 : e1003715 Google Scholar Crossref Search ADS PubMed WorldCat Jones JD , Dangl JL ( 2006 ) The plant immune system . Nature 444 : 323 – 329 Google Scholar Crossref Search ADS PubMed WorldCat Kapprell HP , Maurer A, Kramer F, Heinrich B, Buenning C, Narvaez A, Kalbacher H, Flad T ( 2011 ) Development of a fluorescence resonance energy transfer peptide library technology for detection of protease contaminants in protein-based raw materials used in diagnostic assays . Assay Drug Dev Technol 9 : 549 – 553 Google Scholar Crossref Search ADS PubMed WorldCat Karimi M , Inzé D, Depicker A ( 2002 ) GATEWAY vectors for Agrobacterium-mediated plant transformation . Trends Plant Sci 7 : 193 – 195 Google Scholar Crossref Search ADS PubMed WorldCat Kim JG , Stork W, Mudgett MB ( 2013 ) Xanthomonas type III effector XopD desumoylates tomato transcription factor SlERF4 to suppress ethylene responses and promote pathogen growth . Cell Host Microbe 13 : 143 – 154 Google Scholar Crossref Search ADS PubMed WorldCat Lee AH , Hurley B, Felsensteiner C, Yea C, Ckurshumova W, Bartetzko V, Wang PW, Quach V, Lewis JD, Liu YC, et al. ( 2012 ) A bacterial acetyltransferase destroys plant microtubule networks and blocks secretion . PLoS Pathog 8 : e1002523 Google Scholar Crossref Search ADS PubMed WorldCat Lee KH , Minami A, Marshall RS, Book AJ, Farmer LM, Walker JM, Vierstra RD ( 2011 ) The RPT2 subunit of the 26S proteasome directs complex assembly, histone dynamics, and gametophyte and sporophyte development in Arabidopsis . Plant Cell 23 : 4298 – 4317 Google Scholar Crossref Search ADS PubMed WorldCat Lewis JD , Lee A, Ma W, Zhou H, Guttman DS, Desveaux D ( 2011 ) The YopJ superfamily in plant-associated bacteria . Mol Plant Pathol 12 : 928 – 937 Google Scholar Crossref Search ADS PubMed WorldCat Lewis JD , Lee AH, Hassan JA, Wan J, Hurley B, Jhingree JR, Wang PW, Lo T, Youn JY, Guttman DS, et al. ( 2013 ) The Arabidopsis ZED1 pseudokinase is required for ZAR1-mediated immunity induced by the Pseudomonas syringae type III effector HopZ1a . Proc Natl Acad Sci USA 110 : 18722 – 18727 Google Scholar Crossref Search ADS PubMed WorldCat Ma W , Dong FFT, Stavrinides J, Guttman DS ( 2005 ) Diversification of a type III effector family via both pathoadaptation and horizontal transfer in response to a coevolutionary arms race . PLoS Genet 12 : e209 Google Scholar OpenURL Placeholder Text WorldCat Maier F , Zwicker S, Hückelhoven A, Meissner M, Funk J, Pfitzner AJ, Pfitzner UM ( 2011 ) NONEXPRESSOR OF PATHOGENESIS-RELATED PROTEINS1 (NPR1) and some NPR1-related proteins are sensitive to salicylic acid . Mol Plant Pathol 12 : 73 – 91 Google Scholar Crossref Search ADS PubMed WorldCat Marino D , Peeters N, Rivas S ( 2012 ) Ubiquitination during plant immune signaling . Plant Physiol 160 : 15 – 27 Google Scholar Crossref Search ADS PubMed WorldCat Melotto M , Underwood W, Koczan J, Nomura K, He SY ( 2006 ) Plant stomata function in innate immunity against bacterial invasion . Cell 126 : 969 – 980 Google Scholar Crossref Search ADS PubMed WorldCat Misas-Villamil JC , Kolodziejek I, Crabill E, Kaschani F, Niessen S, Shindo T, Kaiser M, Alfano JR, van der Hoorn RA ( 2013 ) Pseudomonas syringae pv. syringae uses proteasome inhibitor syringolin A to colonize from wound infection sites . PLoS Pathog 9 : e1003281 Google Scholar Crossref Search ADS PubMed WorldCat Mukherjee S , Keitany G, Li Y, Wang Y, Ball HL, Goldsmith EJ, Orth K ( 2006 ) Yersinia YopJ acetylates and inhibits kinase activation by blocking phosphorylation . Science 312 : 1211 – 1214 Google Scholar Crossref Search ADS PubMed WorldCat Pack CG , Yukii H, Toh-e A, Kudo T, Tsuchiya H, Kaiho A, Sakata E, Murata S, Yokosawa H, Sako Y, et al. ( 2014 ) Quantitative live-cell imaging reveals spatio-temporal dynamics and cytoplasmic assembly of the 26S proteasome . Nat Commun 5 : 3396 Google Scholar Crossref Search ADS PubMed WorldCat Park S , Li X, Kim HM, Singh CR, Tian G, Hoyt MA, Lovell S, Battaile KP, Zolkiewski M, Coffino P, et al. ( 2013 ) Reconfiguration of the proteasome during chaperone-mediated assembly . Nature 497 : 512 – 516 Google Scholar Crossref Search ADS PubMed WorldCat Pieterse CM , Van der Does D, Zamioudis C, Leon-Reyes A, Van Wees SC ( 2012 ) Hormonal modulation of plant immunity . Annu Rev Cell Dev Biol 28 : 489 – 521 Google Scholar Crossref Search ADS PubMed WorldCat Shao F , Golstein C, Ade J, Stoutemyer M, Dixon JE, Innes RW ( 2003 ) Cleavage of Arabidopsis PBS1 by a bacterial type III effector . Science 301 : 1230 – 1233 Google Scholar Crossref Search ADS PubMed WorldCat Smith GR , Contreras-Moreira B, Zhang X, Bates PA ( 2004 ) A link between sequence conservation and domain motion within the AAA+ family . J Struct Biol 146 : 189 – 204 Google Scholar Crossref Search ADS PubMed WorldCat Spoel SH , Mou Z, Tada Y, Spivey NW, Genschik P, Dong X ( 2009 ) Proteasome-mediated turnover of the transcription coactivator NPR1 plays dual roles in regulating plant immunity . Cell 137 : 860 – 872 Google Scholar Crossref Search ADS PubMed WorldCat Svozil J , Hirsch-Hoffmann M, Dudler R, Gruissem W, Baerenfaller K ( 2014 ) Protein abundance changes and ubiquitylation targets identified after inhibition of the proteasome with syringolin A . Mol Cell Proteomics 13 : 1523 – 1536 Google Scholar Crossref Search ADS PubMed WorldCat Szczesny R , Büttner D, Escolar L, Schulze S, Seiferth A, Bonas U ( 2010 ) Suppression of the AvrBs1-specific hypersensitive response by the YopJ effector homolog AvrBsT from Xanthomonas depends on a SNF1-related kinase . New Phytol 187 : 1058 – 1074 Google Scholar Crossref Search ADS PubMed WorldCat Tomko RJ Jr , Hochstrasser M ( 2013 ) Molecular architecture and assembly of the eukaryotic proteasome . Annu Rev Biochem 82 : 415 – 445 Google Scholar Crossref Search ADS PubMed WorldCat Umezawa H ( 1976 ) Structures and activities of protease inhibitors of microbial origin . Methods Enzymol 45 : 678 – 695 Google Scholar Crossref Search ADS PubMed WorldCat Üstün S , Bartetzko V, Börnke F ( 2013 ) The Xanthomonas campestris type III effector XopJ targets the host cell proteasome to suppress salicylic-acid mediated plant defence . PLoS Pathog 9 : e1003427 Google Scholar Crossref Search ADS PubMed WorldCat Üstün S , König P, Guttman DS, Börnke F ( 2014 ) HopZ4 from Pseudomonas syringae, a member of the HopZ type III effector family from the YopJ superfamily, inhibits the proteasome in plants . Mol Plant Microbe Interact 27 : 611 – 623 Google Scholar Crossref Search ADS PubMed WorldCat Üstün S , Müller P, Palmisano R, Hensel M, Börnke F ( 2012 ) SseF, a type III effector protein from the mammalian pathogen Salmonella enterica, requires resistance-gene-mediated signalling to activate cell death in the model plant Nicotiana benthamiana . New Phytol 194 : 1046 – 1060 Google Scholar Crossref Search ADS PubMed WorldCat van der Hoorn RA , Leeuwenburgh MA, Bogyo M, Joosten MH, Peck SC ( 2004 ) Activity profiling of papain-like cysteine proteases in plants . Plant Physiol 135 : 1170 – 1178 Google Scholar Crossref Search ADS PubMed WorldCat Vierstra RD ( 2009 ) The ubiquitin-26S proteasome system at the nexus of plant biology . Nat Rev Mol Cell Biol 10 : 385 – 397 Google Scholar Crossref Search ADS PubMed WorldCat Wang D , Weaver ND, Kesarwani M, Dong X ( 2005 ) Induction of protein secretory pathway is required for systemic acquired resistance . Science 308 : 1036 – 1040 Google Scholar Crossref Search ADS PubMed WorldCat Weibezahn J , Schlieker C, Bukau B, Mogk A ( 2003 ) Characterization of a trap mutant of the AAA+ chaperone ClpB . J Biol Chem 278 : 32608 – 32617 Google Scholar Crossref Search ADS PubMed WorldCat Yao C , Wu Y, Nie H, Tang D ( 2012 ) RPN1a, a 26S proteasome subunit, is required for innate immunity in Arabidopsis . Plant J 71 : 1015 – 1028 Google Scholar Crossref Search ADS PubMed WorldCat Zhou H , Lin J, Johnson A, Morgan RL, Zhong W, Ma W ( 2011 ) Pseudomonas syringae type III effector HopZ1 targets a host enzyme to suppress isoflavone biosynthesis and promote infection in soybean . Cell Host Microbe 9 : 177 – 186 Google Scholar Crossref Search ADS PubMed WorldCat Author notes 1 This work was supported by the Deutsche Forschungsgemeinschaft (grant nos. SFB 796 and BO 1916–5/1) and the Leibniz Institute of Vegetable and Ornamental Crops. * Address correspondence to [email protected] and [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Frederik Börnke ([email protected]). S.Ü. performed most of the experiments; S.Ü. and F.B. designed the experiments and analyzed the data; S.Ü. and F.B. conceived the project and wrote the article. [OPEN] Articles can be viewed without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.15.00132 © 2015 American Society of Plant Biologists. All Rights Reserved. © The Author(s) 2015. Published by Oxford University Press on behalf of American Society of Plant Biologists. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.