Journal of Experimental Botany

Lindermayr, Christian; Durner, Jörg

doi: 10.1093/jxb/ery193pmid: 29947809

Abiotic and biotic stress, electron paramagnetic resonance (EPR) spectroscopy, legume nodules, nitric oxide (NO), nitrosyl–leghemoglobin (Lb2+NO), NO detection/production, NO sensor proteins Nitric oxide (NO) is involved in regulation of plant growth and development, as well as the response to biotic and abiotic stressors. However, its instability makes NO methodology a complex and often controversial field. A new method fromCalvo-Begueria et al. (2018)to examine NO production in intact nodules uses electron paramagnetic resonance (EPR) spectroscopy to detect nitrosyl–leghemoglobin (Lb2+NO). NO sensor proteins are an optimal tool for NO detection/quantification in vivo and have the potential to revolutionize the field of plant NO research. Nitric oxide (NO) is an important redox molecule fulfilling a wide variety of signalling functions. These cover growth and development, as well as stress responses, in humans, animals, plants, fungi and bacteria (Besson-Bard et al., 2008; Sudhamsu and Crane, 2009; Murad, 2011; Arasimowicz-Jelonek and Floryszak-Wieczorek, 2016; Byung-Wook et al., 2016; Canovas et al., 2016). In plants, NO is involved in seed germination, root development, gravitropism, iron homeostasis, stomatal closure, flowering, and pollen tube growth (Bellin et al., 2013; Freschi et al., 2013; Simontacchi et al., 2015; Corpas et al., 2017). Moreover, programmed cell death, activation of defence genes and genes involved in UV, heat, drought and salinity stress tolerance require the function of NO (Groß et al., 2013; Kulik et al., 2015; Domingos et al., 2015; Simontacchi et al., 2015; Hu et al., 2017). As a diffusible gas it can be present in all extra- and intracellular spaces, where it easily interacts with the surrounding environment. NO can be produced by oxidative and reductive pathways (Moreau et al., 2010; Yu et al., 2014) and is sensed within the cell through redox modification of proteins, such as cysteine nitrosation, tyrosine nitration and metal nitrosylation (Astier et al., 2012; Astier and Lindermayr, 2012; Mata-Perez et al., 2016). One of its most important modes of action is protein S-nitrosation, the covalent attachment of NO to the thiol group of protein cysteine residues. Tyrosine nitration refers to the addition of a nitro group to susceptible tyrosine residues in the ortho position to the hydroxyl group yielding 3-nitrotyrosine. The main nitrating species is peroxynitrite which is produced in a diffusion-controlled reaction between NO and superoxide. In a direct metal nitrosylation reaction, NO (Lewis base) binds to the transition metal (Lewis acid) of metalloproteins yielding a metal–nitrosyl complex. Detection difficulties Understanding the ‘conduct’ of NO in biological systems is important. However, investigation of NO production and NO signalling is challenging because many available methods suffer from a lack of specificity and/or sensitivity, or are just unsuitable for the detection of NO in vivo. Additionally, in some cases, the production might be restricted to a few cells, such as guard cells or pollen (Corpas et al., 2004; Prado et al., 2004; Neil et al., 2008). NO is a reactive molecule with a lifetime in the order of seconds to minutes. Moreover, in physiological buffers, it diffuses rapidly with a diffusion coefficient approaching 3300 µm2 s–1(Malinski et al., 1993; Lancaster, 1997). Thus, any detection method must be very sensitive to be able to chase intraorganismic NO production. In sum, NO research requires a broad spectrum of complementary methods, which together allow an accurate identification of NO and its physiological function. Sensitive and specific analytical tools for measuring NO in vivo are rare. NO-specific fluorescent dyes, electrodes and sensor proteins are the only options for detecting and quantifying NO in living cells/tissues (Arasimowicz et al., 2009; Eroglu et al., 2016). Others, such as the Griess assay, oxyhemoglobin assay, electron paramagnetic resonance (EPR) spectroscopy, mass spectrometry or chemiluminescence, are used to detect/quantify NO or NO-derived metabolites in (plant) extracts or in the headspace of plants (Zeidler et al., 2004; Mur et al., 2011). However, these probably do not reflect the concentrations inside the intact plant cell. Breakthrough methodology The paper presented by Laura Calvo-Begueria and colleagues describes a method that enables detection of NO in vivo (Calvo-Begueria et al., 2018). They investigated the formation of the nitrosyl–leghemoglobin complex (Lb2+NO) and the production of NO in legume nodules using EPR spectroscopy and the fluorescent specific dye 4,5-diaminofluorescein diacetate (DAF-2 DA), respectively. The EPR method established by the authors allows the detection of Lb2+NO in the infection zone of intact nodules (Box 1). Moreover, their work demonstrates that Lb2+NO is generated as an artefact when nodules are not analysed immediately after detachment and hence quantification of Lb2+NO in nodule extracts is not valid. This confirms that analysis of such reactive compounds should be done using non-invasive methods or at least immediately after sample collection. Finally, their results indicate that EPR complemented by fluorometric methods does allow reliable conclusions about NO production in plants. Box 1. Monitoring NO production in planta using EPR spectroscopy NO binds to the Fe2+ of leghemoglobin and forms an Lb2+NO nitrosyl complex (left). Calvo-Begueria et al. (2018) have demonstrated that this complex can be detected by EPR spectroscopy in intact soybean nodules, allowing a direct monitoring of NO production. Nodules containing Lb2+NO show spectra with a clear diagnostic signal in the range of 320–345 mT (right; see Calvo-Begueria et al., 2018). Here, a numerical addition of the spectrum of intact soybean nodules and the spectra of authentic Lb2+NO at variable proportions is shown. Perspectives Although significant progress has been made in developing methods for NO research, future efforts should still concentrate on enhancing the sensitivity and specificity of these methods and focus on in vivo detection and quantification of NO. Although the method presented by Calvo-Begueria et al. (2018) is restricted to NO detection in the nodule-infected tissue containing leghemoglobin, it is certainly a very promising approach that can be further developed as a general NO sensing technique for analysing NO production in other biological systems. For example, transformation of the leghemoglobin coding sequence into other plant species would enable the use of this protein as an NO sensor and thereby the analysis of NO production/quantification via Lb2+NO detection. However, the availability of EPR spectroscopy might be a restriction for using this technology as a standard method in NO research. In general, an NO sensor protein is an optimal tool for NO detection/quantification in vivo. A fluorescence quenching-based NO probe was designed by Eroglu et al. (2016). Fusing a bacteria-derived NO-binding domain close to distinct fluorescent protein variants enables a direct observation and quantification of NO. Such genetically encoded NO probes (geNOps) provide a selective, specific and real-time read-out of cellular NO dynamics and, hence, open a new era for NO bio-imaging. Furnished with compartment-specific signal peptides, high-resolution, intracellular NO detection would be possible. Despite an increasing number of reports on the biological action of NO in plants, the validity of such work should be questioned depending on the manner in which NO has been measured and/or the solution composition used for NO quantification. Therefore, a re-evaluation of past findings is probably needed in some cases. The different measurement techniques that can be used for a given sample type are highlighted in Box 2. Ideally, methods for determination/quantification of NO should exhibit a high degree of sensitivity and specificity, and should in particular facilitate the detection of NO in planta. NO sensor proteins (Lb2+NO and geNOps) have the potential to fulfil all these ideal characteristics and could revolutionize the field of NO research in plants. Further development of such NO measurement approaches, including the use of appropriate signal peptides and spatiotemporal-specific promotor elements, will allow an accurate determination of NO production in different plant systems, tissues and cells, and help to reveal exactly how, when and where NO is produced. Such a method would provide robust results and assuage the controversial discussions on the detection of NO in plants. Box 2. Methods used for NO detection in headspace, in planta or in plant extracts The different methods available to detect and quantify NO are based on its particular physical and chemical properties. The method of choice depends on the biological question that needs to be answered. Some assays detect NO gas emitted from cells, whereas others allow measurement of NO and its derivatives (e.g. N2O3, NO2–) in liquid solutions. NO-sensitive dyes, electrodes and sensor proteins allow detection and quantification of NO in planta. NO sensor proteins (Lb2+NO) under the control of tissue- or cell-specific promotors are especially promising specific and sensitive tools for spatiotemporal detection of NO in planta. However, the use of NO sensor proteins requires suitable detection instruments, such as an EPR spectrometer, a confocal laser scanning microscope, or a chemiluminescence detector. The high cost of these pieces of equipment and the considerable expertise needed to work with them may limit their use for standard methods in NO research. References Arasimowicz-Jelonek M , Floryszak-Wieczorek J. 2016 . Nitric oxide in the offensive strategy of fungal and oomycete plant pathogens . Frontiers in Plant Science 7 , 252 . Google Scholar Crossref Search ADS PubMed WorldCat Arasimowicz M , Floryszak-Wieczorek J, Milczarek G, Jelonek T. 2009 . Nitric oxide, induced by wounding, mediates redox regulation in pelargonium leaves . Plant Biology 11 , 650 – 663 . Google Scholar Crossref Search ADS PubMed WorldCat Astier J , Kulik A, Koen E, Besson-Bard A, Bourque S, Jeandroz S, Lamotte O, Wendehenne D. 2012 . Protein S-nitrosylation: what’s going on in plants ? Free Radical Biology & Medicine 53 , 1101 – 1110 . Google Scholar Crossref Search ADS PubMed WorldCat Astier J , Lindermayr C. 2012 . Nitric oxide-dependent posttranslational modification in plants: an update . International Journal of Molecular Sciences 13 , 15193 – 15208 . Google Scholar Crossref Search ADS PubMed WorldCat Bellin D , Asai S, Delledonne M, Yoshioka H. 2013 . Nitric oxide as a mediator for defense responses . Molecular Plant-Microbe Interactions 26 , 271 – 277 . Google Scholar Crossref Search ADS PubMed WorldCat Besson-Bard A , Pugin A, Wendehenne D. 2008 . New insights into nitric oxide signaling in plants . Annual Review of Plant Biology 59 , 21 – 39 . Google Scholar Crossref Search ADS PubMed WorldCat Byung-Wook Y , Skelly MJ, Yin M, Yu M, Mun BG, Lee SU, Hussain A, Spoel SH, Loake GJ. 2016 . Nitric oxide and S-nitrosoglutathione function additively during plant immunity . New Phytologist 211 , 516 – 526 . Google Scholar Crossref Search ADS PubMed WorldCat Calvo-Begueria L , Rubio MC, Martínez JI, Pérez-Rontomé C, Delgado MJ, Bedmar EJ, Becana M. 2018 . Redefining nitric oxide production in legume nodules through complementary insights from electron paramagnetic resonance spectroscopy and specific fluorescent probes . Journal of Experimental Botany 69 , 3703–3714. Google Scholar OpenURL Placeholder Text WorldCat Cánovas D , Marcos JF, Marcos AT, Strauss J. 2016 . Nitric oxide in fungi: is there NO light at the end of the tunnel ? Current Genetics 62 , 513 – 518 . Google Scholar Crossref Search ADS PubMed WorldCat Corpas FJ , Barroso JB, Carreras A, et al. . 2004 . Cellular and subcellular localization of endogenous nitric oxide in young and senescent pea plants . Plant Physiology 136 , 2722 – 2733 . Google Scholar Crossref Search ADS PubMed WorldCat Corpas FJ , Freschi L, Rodríguez-Ruiz M, Mioto PT, González-Gordo S, Palma JM. 2017 . Nitro-oxidative metabolism during fruit ripening . Journal of Experimental Botany . doi: 10.1093/jxb/erx453 . Google Scholar OpenURL Placeholder Text WorldCat Crossref Domingos P , Prado AM, Wong A, Gehring C, Feijo JA. 2015 . Nitric oxide: a multitasked signaling gas in plants . Molecular Plant 8 , 506 – 520 . Google Scholar Crossref Search ADS PubMed WorldCat Eroglu E , Gottschalk B, Charoensin S, et al. . 2016 . Development of novel FP-based probes for live-cell imaging of nitric oxide dynamics . Nature Communications 7 , 10623 . Google Scholar Crossref Search ADS PubMed WorldCat Freschi L. 2013 . Nitric oxide and phytohormone interactions: current status and perspectives . Frontiers in Plant Science 4 , 398 . Google Scholar Crossref Search ADS PubMed WorldCat Groß F , Durner J, Gaupels F. 2013 . Nitric oxide, antioxidants and prooxidants in plant defence responses . Frontiers in Plant Science 4 , 419 . Google Scholar Crossref Search ADS PubMed WorldCat Hu J , Yang H, Mu J, et al. . 2017 . Nitric oxide regulates protein methylation during stress responses in plants . Molecular Cell 67 , 702 – 710.e4 . Google Scholar Crossref Search ADS PubMed WorldCat Kulik A , Noirot E, Grandperret V, et al. . 2015 . Interplays between nitric oxide and reactive oxygen species in cryptogein signalling . Plant, Cell & Environment 38 , 331 – 348 . Google Scholar Crossref Search ADS PubMed WorldCat Lancaster JR Jr. 1997 . A tutorial on the diffusibility and reactivity of free nitric oxide . Nitric Oxide 1 , 18 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat Malinski T , Taha Z, Grunfeld S, Patton S, Kapturczak M, Tomboulian P. 1993 . Diffusion of nitric oxide in the aorta wall monitored in situ by porphyrinic microsensors . Biochemical and Biophysical Research Communications 193 , 1076 – 1082 . Google Scholar Crossref Search ADS PubMed WorldCat Mata-Pérez C , Begara-Morales JC, Chaki M, Sánchez-Calvo B, Valderrama R, Padilla MN, Corpas FJ, Barroso JB. 2016 . Protein tyrosine nitration during development and abiotic stress response in plants . Frontiers in Plant Science 7 , 1699 . Google Scholar Crossref Search ADS PubMed WorldCat Moreau M , Lindermayr C, Durner J, Klessig DF. 2010 . NO synthesis and signaling in plants–where do we stand ? Physiologia Plantarum 138 , 372 – 383 . Google Scholar Crossref Search ADS PubMed WorldCat Mur LA , Mandon J, Cristescu SM, Harren FJ, Prats E. 2011 . Methods of nitric oxide detection in plants: a commentary . Plant Science 181 , 509 – 519 . Google Scholar Crossref Search ADS PubMed WorldCat Murad MD. 2011 . Nitric oxide: the coming of the second messenger . Rambam Maimonnides Medical Journal 2 , e0038 . Google Scholar OpenURL Placeholder Text WorldCat Neill S , Barros R, Bright J, Desikan R, Hancock J, Harrison J, Morris P, Ribeiro D, Wilson I. 2008 . Nitric oxide, stomatal closure, and abiotic stress . Journal of Experimental Botany 59 , 165 – 176 . Google Scholar Crossref Search ADS PubMed WorldCat Prado AM , Porterfield DM, Feijó JA. 2004 . Nitric oxide is involved in growth regulation and re-orientation of pollen tubes . Development 131 , 2707 – 2714 . Google Scholar Crossref Search ADS PubMed WorldCat Simontacchi M , Galatro A, Ramos-Artuso F, Santa-María GE. 2015 . Plant survival in a changing environment: the role of nitric oxide in plant responses to abiotic stress . Frontiers in Plant Science 6 , 977 . Google Scholar Crossref Search ADS PubMed WorldCat Sudhamsu J , Crane BR. 2009 . Bacterial nitric oxide synthases: what are they good for ? Trends in Microbiology 17 , 212 – 218 . Google Scholar Crossref Search ADS PubMed WorldCat Yu M , Lamattina L, Spoel SH, Loake GJ. 2014 . Nitric oxide function in plant biology: a redox cue in deconvolution . New Phytologist 202 , 1142 – 1156 . Google Scholar Crossref Search ADS PubMed WorldCat Zeidler D , Zähringer U, Gerber I, Dubery I, Hartung T, Bors W, Hutzler P, Durner J. 2004 . Innate immunity in Arabidopsis thaliana: lipopolysaccharides activate nitric oxide synthase (NOS) and induce defense genes . Proceedings of the National Academy of Sciences, USA 101 , 15811 – 15816 . Google Scholar Crossref Search ADS WorldCat © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.

Sabelleck, Björn; Panstruga, Ralph

doi: 10.1093/jxb/ery192pmid: 29947808

Barley, Blumeria graminis f.sp. hordei, effectors, HvMAGAP1, HvRACB, pathogenicity, peptide, powdery mildew, retrotransposon, ROP GTPase Phytopathogens deploy secreted effector proteins that interfere with different host cellular pathways to suppress plant immunity and redirect nutrient flow. Conventional effectors are small polypeptides with an N-terminal secretion signal, no transmembrane domain and often lacking any protein domain annotation. In their current work,Nottensteiner et al. (2018)provide experimental evidence for the existence of an unconventional, seemingly retrotransposon-derived effector of the barley powdery mildew fungus. This protein (ROPIP1) interacts with the barley susceptibility factor RACB, possibly leading to local destabilization of cortical microtubular network architecture, thereby supporting fungal host cell entry. Powdery mildews are filamentous phytopathogenic fungi that belong to the group of ascomycetes and thrive epiphytically on their plant hosts. They pursue an obligate biotrophic lifestyle, strictly depending on living host tissue for growth and propagation (Glawe, 2008). During infection, these fungi penetrate the cell wall of plant epidermal cells but leave the underlying plasma membrane intact. Inside host cells they form their feeding structure, the haustorium, which is covered by the so-called extrahaustorial membrane – a derivative of the plant plasma membrane (Box 1). Haustoria are probably the sites of fungal nutrient uptake and effector protein delivery, and their establishment is a strict prerequisite for subsequent hyphal growth and sporulation. The intimate association of powdery mildews with their plant hosts at the haustorial interface and their biotrophic lifestyle require an exquisite level of defence suppression, which is considered to be achieved largely via dedicated effector proteins (Hückelhoven and Panstruga, 2011). Box 1. Model illustrating the potential mode of action of ROPIP1 When the barley powdery mildew fungus Blumeria graminis f.sp. hordei (Bgh) successfully establishes its feeding structure (haustorium, grey flask-shaped structure) by penetrating the cell wall (dark green) of a host cell, the underlying plasma membrane (extrahaustorial membrane, red line), which is connected via the neck band (dark grey circles) to the regular host plasma membrane (light green), remains intact and the attacked plant cell stays alive. ROPIP1 (red pentagon) is encoded by a particular class of retrotransposons, so-called Eg-R1 SINEs (see Box 2), which occur in thousands of copies in the Bgh genome. Some of these copies acquired an N-terminal signal peptide by random in-frame insertions in the genome. Following its translation in the fungal pathogen, ROPIP1 is secreted and translocates into host cells via an unknown pathway. There it interacts with the susceptibility factor RACB (brown oval), a small monomeric Rho of plant (ROP) GTPase. RACB in turn interacts with the microtubule-associated ROP GTPase-activating protein (ROP-GAP) MAGAP1 (blue rectangle), which recruits the complex of ROPIP1 and RACB to cortical microtubules (purple lines). This association causes destabilization of the microtubular network via an uncharacterized mechanism. During the plant–powdery mildew interaction, the latter event might be spatially confined to the actual contact sites of host cells and fungal infection structures. Effectors are small proteins that are secreted by phytopathogens to suppress plant immunity and to promote host colonization (Dou and Zhou, 2012). They are often defined as proteins with an N-terminal secretion signal, no transmembrane domain and no recognizable functional protein domain(s). This group of proteins can be further subdivided, according to their destination in the host, into apoplastic and cytoplasmic effectors. Apoplastic effectors are typically cysteine-rich, likely to withstand the harsh oxidative conditions in the extracellular space and/or to resist plant proteases. Cytoplasmic effectors are delivered into or taken up by host cells via yet poorly resolved mechanisms (Lo Presti and Kahmann, 2017). Blumeria graminis f.sp. hordei (Bgh) and B. graminis f.sp. tritici (Bgt) are the powdery mildew fungi that colonize the cereals barley and wheat, respectively. Their genomes encode sets of several hundred conventional effector proteins called ‘Candidates for Secreted Effector Proteins’ (CSEPs) and/or ‘Blumeria Effector Candidates’ (BECs) (Pedersen et al., 2012; Wicker et al., 2013; Bindschedler et al., 2016; Frantzeskakis et al., 2018). A subgroup of these CSEPs classify based on their predicted distant structural similarities to ribonucleases and their high expression levels inside haustoria as ‘RNAse-Like Proteins associated with Haustoria’ (RALPHs). This set of effectors might derive from a common ancestral ribonuclease (Spanu, 2017). Plants evolved the ability to detect some pathogen effectors to trigger a boosted immune response (effector-triggered immunity, ETI). In this case a typically cytoplasmic sensor protein (R protein) recognizes a particular effector variant (then termed avirulence protein, AVR) from the pathogen (Jones and Dangl, 2006). Barley ‘Mildew Locus A’ (MLA) proteins are such R proteins, which belong to the class of intracellular nucleotide-binding domain and leucine-rich repeat proteins (NLRs). For MLA13 and MLA1, two allelic MLA versions, the corresponding Bgh AVRs, were identified as classical CSEPs [AVRa13 (CSEP0372) and AVRa1 (CSEP0008) (Lu et al., 2016)]. However, it is still unknown whether the interaction between R protein and effector is direct or indirect in these instances. So far, there are only a few documented examples of direct interactions between Bgh effector candidates and host plant proteins (Zhang et al., 2012; Schmidt et al., 2014; Ahmed et al., 2015; Pennington et al., 2016). Transposable elements – a source of new effectors? The genome of the barley powdery mildew pathogen Bgh is about four-times larger than the median genome size of other ascomycetes, but its gene number is smaller than in most other filamentous fungi. Around 75% of the genomic DNA consists of transposable elements (TEs), which are evenly distributed throughout the genome (Spanu et al., 2010; Amselem et al., 2015a ; Frantzeskakis et al., 2018). This unusually high amount probably reflects the absence of the repeat-induced point mutation (RIP) mechanism, which usually controls the spread of TEs, and this could account for the huge genome size (Spanu et al., 2010). TEs generally fall into class I (retrotransposons) or class II (DNA transposons). In powdery mildews, class I retrotransposons dominate (Spanu et al., 2010; Amselem et al., 2015a ). Retrotransposons are subdivided into those with long terminal repeats (LTRs) as well as long and short interspersed nuclear elements (LINEs and SINEs, respectively) that lack LTRs and are therefore collectively termed non-LTR retrotransposons (Box 2). While LTR retrotransposons and LINEs are autonomous, SINEs are non-autonomous, i.e. they cannot transpose on their own (Finnegan, 2012). Box 2. Classification and modular structure of retrotransposons Class I transposons (retrotransposons) are divided in two main groups, LTR retrotransposons (from a few hundred base pairs up to 25 kb in size) and non-LTR retrotransposons. The non-LTR retrotransposons can be further subdivided into LINEs and SINEs. LTR (dark blue) refers to ‘long terminal repeats’, which are a few hundred base pairs long and flank the central part of the TE as direct repeats. This kind of retrotransposon harbours two to three ORFs between the LTRs. The gag (group-specific antigen) gene (green) product forms a virus-like particle, while pol (orange) encodes a polyprotein that undergoes cleavage and comprises a combination of an integrase, a reverse transcriptase and an RNase H domain. Some LTR retrotransposons have also an env gene (grey), which codes for a (typically defective) viral envelope protein. LINEs are typically 6–8 kb in size and composed of two ORFs, where ORF1 (light blue) encodes an RNA-binding protein that is necessary for the LINE transposition intermediate and ORF2 (purple) codes for a nuclease and a reverse transcriptase. In some cases, an RNase H domain is also included in ORF2. At the beginning of a LINE retrotransposon is a promoter that regulates transcription of the ORFs, and at the 3′ end is a polyA signal. SINEs (~300 bp long) have no coding regions and harbour a Box A and a Box B, which show similarity with an internal RNA polymerase III promoter. These retrotransposons rely on LINE-encoded reverse trancriptases for their proliferation. They show sequence similarity at the 5′ ends to tRNAs and at the 3′ ends to the distal part of LINEs. Note that the various types of retrotransposons are not drawn to scale. Until now it was thought that TEs primarily generate an enormous playground for the birth and death of conventional effector variants, e.g. via non-homologous (illegitimate) recombination events (Raffaele and Kamoun, 2012; Gladieux et al., 2014; Grandaubert et al., 2014; Fouche et al., 2018). However, in their present study, Nottensteiner et al. (2018) show that a peptide, which apparently derives from a SINE-related TE termed Eg-R1 (Wei et al., 1996), acts as an unconventional effector of Bgh. This peptide was originally found in a yeast two-hybrid (Y2H) screen for interactors of RACB, a small monomeric Rho of plant (ROP)-type GTPase that serves as a susceptibility factor in barley (Schultheiss et al., 2002). Using RACB as a bait, the peptide was retrieved multiple times from a prey cDNA library that was established from barley leaves heavily infected with Bgh. The authors demonstrate that the peptide can interact with RACB in a targeted Y2H test and also in planta, as shown by ‘bimolecular fluorescence complementation’ (BiFC). In both assays a dominant-negative form of RACB, characterized by a single amino acid substitution (T20N), served as a specificity control (Kudla and Bock, 2016). Because of the interaction and its small size (74 amino acids), they named this peptide ‘ROP Interactive Peptide 1’ (ROPIP1). The respective open reading frame (ORF) is associated with the 5′ end of Eg-R1, which probably occurs in several thousand copies in the Bgh genome. At least some of these copies appear to have in-frame 5′ extensions that encode predicted canonical secretion signals. The group validated pathogen-induced ROPIP1 transcript accumulation using qRT-PCR analysis and detected infection-dependent protein accumulation in immunoblot analysis using a polyclonal antiserum raised against a synthetic epitope peptide located in the C-terminus of ROPIP1. Notably, the authors also performed in situ immunodetection of ROPIP1 via transmission electron microscopy-based immunogold labelling. This technique revealed the presence of ROPIP1-associated gold label in fungal infection structures (appressoria and haustoria), but also within the host cell wall, local cell wall appositions (papillae) and the host cytoplasm, suggesting secretion of ROPIP1 from the fungal pathogen into host cells. Results from transient overexpression and gene silencing experiments in single transformed barley epidermal cells hint at a role of ROPIP1 in fungal virulence since overexpression increased and silencing conversely decreased fungal penetration success. The specificity of the latter effect was confirmed with a silencing-resistant ‘RNAi rescue’ construct. The authors went on to demonstrate the fate of ROPIP1 in host cells by studying the subcellular localization of a fluorophore-tagged version. When co-expressed with the known RACB interactor MAGAP1, ROPIP1 localizes to cortical microtubules. RACB promotes the recruitment of ROPIP1 via MAGAP1 to microtubules, which leads to a dramatic alteration in microtubular network architecture, characterized by a substantial increase of cells with fragmented microtubules. The authors claim that this destabilization of the cortical microtubular network, which in the authentic interaction might be spatially confined to the site of attempted host cell entry, may aid fungal virulence (Box 1). ROPIP1 – a jack-in-the-TE-box? ROPIP1 is not the first effector nominee allegedly originating from a powdery mildew retrotransposon. In an earlier study, candidates for the avirulence proteins AVRa10 and AVRk1 were reported to derive from a particular class of LINE-1 retrotransposons present in the Bgh genome (Ridout et al., 2006). The elements encoding the respective family of proteins – termed EKAs (Effectors homologous to AVRk1 and AVRa10) – were similarly to Eg-R1 found to occur in >1000 copies in the Bgh genome (Sacristan et al., 2009; Amselem et al., 2015b ). However, the existence of EKA-related proteins was only indirectly shown by the occurrence of respective peptide fragments upon proteomic analysis of samples from infected plant epidermis and fungal hyphae (Amselem et al., 2015b ), and plant cellular targets and/or a potential mode of effector mechanism remained elusive. Moreover, the absence of prototypical N-terminal secretion signals in this protein family raised concerns regarding their potential for transfer into host cells. By contrast, the presence of prototypical secretion sequences in a subset of the ROPIP1 copies, the immunological detection of ROPIP1 in fungal and host cells, the identification of candidate host targets as well as a suggested cellular activity altogether make up a plausible scenario (Box 1). Accordingly, some Bgh Eg-R1 versions might have been neofunctionalized to support fungal virulence. It remains, however, puzzling that despite the many putative ROPIP1 copies in the Bgh genome, which show apparent length polymorphism regarding the encoded proteins, a discrete signal is seen in immunoblot analysis. This raises the possibility that only one or a few of the many potential ROPIP1 variants is/are produced by fungal cells. Although some copies seem to harbour potentially functional secretion signals, another formal possibility is that ROPIP1 is secreted via a non-conventional secretion pathway (Wang et al., 2018) as discussed by the authors. Given its non-canonical origin, future research activities are needed to corroborate (i) the actual existence of the ROPIP1 peptide (e.g. by protein mass spectrometry of immunoprecipitated protein), (ii) its transfer into host cells (via conventional or unconventional secretion pathways?) and (iii) its in planta interaction with the RACB/MAGAP1/microtubular network complex (e.g. via fluorescence resonance energy transfer (FRET) studies or by co-immunoprecipitation analysis). But even with these caveats in mind, the work by Nottensteiner and co-workers provides interesting food for thought. Acknowledgements This work is related to Deutsche Forschungsgemeinschaft (DFG) grants PA 861/13-1 and PA 861/14-1 awarded to R. P. We thank Stefan Kusch for critical reading of the manuscript. References Ahmed AA , Pedersen C, Schultz-Larsen T, Kwaaitaal M, Jørgensen HJ, Thordal-Christensen H. 2015 . The barley powdery mildew candidate secreted effector protein CSEP0105 inhibits the chaperone activity of a small heat shock protein . Plant Physiology 168 , 321 – 333 . Google Scholar Crossref Search ADS PubMed WorldCat Amselem J , Lebrun MH, Quesneville H. 2015a . Whole genome comparative analysis of transposable elements provides new insight into mechanisms of their inactivation in fungal genomes . BMC Genomics 16 , 141 . Google Scholar Crossref Search ADS WorldCat Amselem J , Vigouroux M, Oberhaensli S, Brown JK, Bindschedler LV, Skamnioti P, Wicker T, Spanu PD, Quesneville H, Sacristán S. 2015b . Evolution of the EKA family of powdery mildew avirulence-effector genes from the ORF 1 of a LINE retrotransposon . BMC Genomics 16 , 917 . Google Scholar Crossref Search ADS WorldCat Bindschedler LV , Panstruga R, Spanu PD. 2016 . Mildew-Omics: how global analyses aid the understanding of life and evolution of powdery mildews . Frontiers in Plant Science 7 , 123 . Google Scholar Crossref Search ADS PubMed WorldCat Dou D , Zhou JM. 2012 . Phytopathogen effectors subverting host immunity: different foes, similar battleground . Cell Host & Microbe 12 , 484 – 495 . Google Scholar Crossref Search ADS PubMed WorldCat Finnegan DJ. 2012 . Retrotransposons . Current Biology 22 , R432 – 437 . Google Scholar Crossref Search ADS PubMed WorldCat Fouché S , Plissonneau C, Croll D. 2018 . The birth and death of effectors in rapidly evolving filamentous pathogen genomes . Current Opinion in Microbiology 46 , 34 – 42 . Google Scholar Crossref Search ADS PubMed WorldCat Frantzeskakis L , Kracher B, Kusch S, Yoshikawa-Maekawa M, Bauer S, Pesersen C, Spanu PD, Maekawa T, Schulze-Lefert P, Panstruga R. 2018 . Signatures of host specialization and a recent transposable element burst in the dynamic one-speed genome of the fungal barley powdery mildew pathogen . bioRxiv . doi: 10.1101/246280 Google Scholar OpenURL Placeholder Text WorldCat Crossref Gladieux P , Ropars J, Badouin H, Branca A, Aguileta G, de Vienne DM, Rodríguez de la Vega RC, Branco S, Giraud T. 2014 . Fungal evolutionary genomics provides insight into the mechanisms of adaptive divergence in eukaryotes . Molecular Ecology 23 , 753 – 773 . Google Scholar Crossref Search ADS PubMed WorldCat Glawe DA. 2008 . The powdery mildews: a review of the world’s most familiar (yet poorly known) plant pathogens . Annual Review of Phytopathology 46 , 27 – 51 . Google Scholar Crossref Search ADS PubMed WorldCat Grandaubert J , Lowe RG, Soyer JL, et al. . 2014 . Transposable element-assisted evolution and adaptation to host plant within the Leptosphaeria maculans-Leptosphaeria biglobosa species complex of fungal pathogens . BMC Genomics 15 , 891 . Google Scholar Crossref Search ADS PubMed WorldCat Hückelhoven R , Panstruga R. 2011 . Cell biology of the plant-powdery mildew interaction . Current Opinion in Plant Biology 14 , 738 – 746 . 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 Kudla J , Bock R. 2016 . Lighting the way to protein-protein interactions: recommendations on best practices for bimolecular fluorescence complementation analyses . The Plant Cell 28 , 1002 – 1008 . Google Scholar Crossref Search ADS PubMed WorldCat Lo Presti L , Kahmann R. 2017 . How filamentous plant pathogen effectors are translocated to host cells . Current Opinion in Plant Biology 38 , 19 – 24 . Google Scholar Crossref Search ADS PubMed WorldCat Lu X , Kracher B, Saur IM, Bauer S, Ellwood SR, Wise R, Yaeno T, Maekawa T, Schulze-Lefert P. 2016 . Allelic barley MLA immune receptors recognize sequence-unrelated avirulence effectors of the powdery mildew pathogen . Proceedings of the National Academy of Sciences, USA 113 , E6486 – E6495 . Google Scholar Crossref Search ADS WorldCat Nottensteiner M , Zechmann B, McCollum C, Hückelhoven R. 2018 . A barley powdery mildew fungus non-autonomous retrotransposon encodes a peptide that supports penetration success on barley . Journal of Experimental Botany 69 , 3745 – 3758 . Google Scholar OpenURL Placeholder Text WorldCat Pedersen C , Ver Loren van Themaat E, McGuffin LJ, et al. . 2012 . Structure and evolution of barley powdery mildew effector candidates . BMC Genomics 13 , 694 . Google Scholar Crossref Search ADS PubMed WorldCat Pennington HG , Gheorghe DM, Damerum A, Pliego C, Spanu PD, Cramer R, Bindschedler LV. 2016 . Interactions between the powdery mildew effector BEC1054 and barley proteins identify candidate host targets . Journal of Proteome Research 15 , 826 – 839 . Google Scholar Crossref Search ADS PubMed WorldCat Raffaele S , Kamoun S. 2012 . Genome evolution in filamentous plant pathogens: why bigger can be better . Nature Reviews. Microbiology 10 , 417 – 430 . Google Scholar Crossref Search ADS PubMed WorldCat Ridout CJ , Skamnioti P, Porritt O, Sacristan S, Jones JD, Brown JK. 2006 . Multiple avirulence paralogues in cereal powdery mildew fungi may contribute to parasite fitness and defeat of plant resistance . The Plant Cell 18 , 2402 – 2414 . Google Scholar Crossref Search ADS PubMed WorldCat Sacristán S , Vigouroux M, Pedersen C, Skamnioti P, Thordal-Christensen H, Micali C, Brown JK, Ridout CJ. 2009 . Coevolution between a family of parasite virulence effectors and a class of LINE-1 retrotransposons . PLoS One 4 , e7463 . Google Scholar Crossref Search ADS PubMed WorldCat Schmidt SM , Kuhn H, Micali C, Liller C, Kwaaitaal M, Panstruga R. 2014 . Interaction of a Blumeria graminis f. sp. hordei effector candidate with a barley ARF-GAP suggests that host vesicle trafficking is a fungal pathogenicity target . Molecular Plant Pathology 15 , 535 – 549 . Google Scholar Crossref Search ADS PubMed WorldCat Schultheiss H , Dechert C, Kogel KH, Hückelhoven R. 2002 . A small GTP-binding host protein is required for entry of powdery mildew fungus into epidermal cells of barley . Plant Physiology 128 , 1447 – 1454 . Google Scholar Crossref Search ADS PubMed WorldCat Spanu PD. 2017 . Cereal immunity against powdery mildews targets RNase-Like Proteins associated with Haustoria (RALPH) effectors evolved from a common ancestral gene . New Phytologist 213 , 969 – 971 . Google Scholar Crossref Search ADS PubMed WorldCat Spanu PD , Abbott JC, Amselem J, et al. . 2010 . Genome expansion and gene loss in powdery mildew fungi reveal tradeoffs in extreme parasitism . Science 330 , 1543 – 1546 . Google Scholar Crossref Search ADS PubMed WorldCat Wang X , Chung KP, Lin W, Jiang L. 2018 . Protein secretion in plants: conventional and unconventional pathways and new techniques . Journal of Experimental Botany 69 , 21 – 37 . Google Scholar Crossref Search ADS WorldCat Wei YD , Collinge DB, Smedegaard-Petersen V, Thordal-Christensen H. 1996 . Characterization of the transcript of a new class of retroposon-type repetitive element cloned from the powdery mildew fungus, Erysiphe graminis . Molecular & General Genetics 250 , 477 – 482 . Google Scholar OpenURL Placeholder Text WorldCat Wicker T , Oberhaensli S, Parlange F, et al. . 2013 . The wheat powdery mildew genome shows the unique evolution of an obligate biotroph . Nature Genetics 45 , 1092 – 1096 . Google Scholar Crossref Search ADS PubMed WorldCat Zhang WJ , Pedersen C, Kwaaitaal M, Gregersen PL, Mørch SM, Hanisch S, Kristensen A, Fuglsang AT, Collinge DB, Thordal-Christensen H. 2012 . Interaction of barley powdery mildew effector candidate CSEP0055 with the defence protein PR17c . Molecular Plant Pathology 13 , 1110 – 1119 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.

Grosjean, Kevin; Der, Christophe; Robert, Franck; Thomas, Dominique; Mongrand, Sébastien; Simon-Plas, Françoise; Gerbeau-Pissot, Patricia

doi: 10.1093/jxb/ery152pmid: 29722895

Abstract The laterally heterogeneous plant plasma membrane (PM) is organized into finely controlled specialized areas that include membrane-ordered domains. Recently, the spatial distribution of such domains within the PM has been identified as playing a key role in cell responses to environmental challenges. To examine membrane order at a local level, BY-2 tobacco suspension cell PMs were labelled with an environment-sensitive probe (di-4-ANEPPDHQ). Four experimental models were compared to identify mechanisms and cell components involved in short-term (1 h) maintenance of the ordered domain organization in steady-state cell PMs: modulation of the cytoskeleton or the cell wall integrity of tobacco BY-2 cells; and formation of giant vesicles using either a lipid mixture of tobacco BY-2 cell PMs or the original lipid and protein combinations of the tobacco BY-2 cell PM. Whilst inhibiting phosphorylation or disrupting either the cytoskeleton or the cell wall had no observable effects, we found that lipids and proteins significantly modified both the abundance and spatial distribution of ordered domains. This indicates the involvement of intrinsic membrane components in the local physical state of the plant PM. Our findings support a major role for the ‘lipid raft’ model, defined as the sterol-dependent ordered assemblies of specific lipids and proteins in plant PM organization. di-4-ANEPPDHQ, cell wall, cytoskeleton, lipid raft, membrane organization, ordered domains, protein-lipid interactions Introduction Plant cells are delimited by a plasma membrane (PM) that protects them against the external environment, and that also regulates what (and how much) enters the cell. The PM defines the boundary between the intracellular and extracellular space and also plays a major role in transducing various signals into the appropriate adaptive responses. Since the advent of the classic ‘Fluid Mosaic Model’, consisting of a homogeneous lipid bilayer with embedded proteins arranged as mosaic-like structures (Singer and Nicolson, 1972), multiple lines of evidence have suggested the presence of a nanoscale lateral heterogeneity with regards to the composition and biophysical properties of plant PMs. One of the original concepts developed within this framework is the ‘lipid raft’ model (Simons and Ikonen, 1997; Simons and Gerl, 2010; Nicolson, 2014), which is based on the sub-division of membranes into regions further defined as ‘small (10–200 nm), heterogeneous, highly dynamic, sterol- and sphingolipid-enriched domains’ (Pike, 2006). According to this model, preferential interactions between cholesterol and sphingolipids generate liquid-ordered (Lo) phase separation (Phillips et al., 1970; Shimshick and McConnell, 1973; Grant et al., 1974; Lentz et al., 1976), in which the resulting membrane domains are highly ordered and tightly packed relative to the surrounding regions (Simons and Sampaio, 2011). The similar ability of plant-specific sterols to form an Lo phase has been reported in PM-purified fractions (Roche et al., 2008) as well as in living tobacco cells (Gerbeau-Pissot et al., 2014), with diffing capabilities to stabilize the lipid bilayer depending on the phytosterol structure (Rujanavech et al., 1986; Schuler et al., 1990, 1991; Hartmann, 1998; Halling and Slotte, 2004). This capacity of phytosterols to modulate the size and proportion of the Lo phase in the model membrane has been associated with their ability to interact with plant sphingolipids (Grosjean et al., 2015). Enrichment in sterols and sphingolipids has been identified for detergent-resistant membrane (DRM) fractions from the PMs of tobacco (Mongrand et al., 2004; Moscatelli et al., 2015), Arabidopsis thaliana (Borner et al., 2005; Minami et al., 2009), Medicago truncatula (Lefebvre et al., 2007), leek (Laloi et al., 2007), maize and bean (Carmona-Salazar et al., 2011), confirming their association within specific membrane regions. In the tobacco PM, use of immunogold electron microscopy has revealed localization of glycosyl inositol phosphorylceramide (GIPC), the major class of sphingolipids in plants, within 35-nm diameter domains (Cacas et al., 2016). Phosphatidylinositol 4,5-bisphosphate (PIP2) is similarly found in ~25-nm diameter clusters (Furt et al., 2010), suggesting that there is a heterogeneous spatial distribution of the lipids on a nanometre scale. Furthermore, various proteins have also shown a grouped distribution (about 70 nm in size) in tobacco leaf (Raffaele et al., 2009), tobacco BY-2 suspension (Noirot et al., 2014), and Arabidopsis (Li et al., 2012) cells. Plant cell PMs are thus covered with different types of domains enriched in specific membrane-resident proteins (Jarsch et al., 2014). Interestingly, the integrity of some of these protein clusters is sensitive to the amount of sterol, as observed after methyl-β-cyclodextrin (MβCD)-induced sterol depletion (Vereb et al., 2000; Raffaele et al., 2009), which suggests sterol-driven formation of these protein-enriched domains. Furthermore, a close relationship between the sterol amount and polar localization of auxin efflux carriers has been proposed, since sterol methyltransferase 1 (SMT1) and sterol-dependent endosomal recycling are essential in maintaining the polar localization of PIN1 and PIN3 (Willemsen et al., 2003), in addition to PIN2 (Men et al., 2008; Titapiwatanakun and Murphy, 2009). While lipid-driven segregation can be coupled with additional actin cytoskeleton-based processes to spatially organize the dynamic status of membrane proteins in animal cells (Lenne et al., 2006), no similar actin-dependent mechanisms have yet been fully described in plant cells. Nevertheless, interactions and crosstalk between microtubules and microfilaments have been reported in plant cells (Petrásek and Schwarzerova, 2009), revealing the role of microtubule organization in modulating plant protein motility (Szymanski et al., 2015; Lv et al., 2017). Furthermore, the pattern of cellulose deposition in the cell wall has been shown to strongly affect the trajectory and speed of PM protein diffusion in Nicotiana tabacum leaves (Martinière et al., 2012), suggesting the involvement of the cell wall in regulating plant PM lateral organization. These previous reports indicate that the plant PM consists of heterogeneously distributed components, in which grouped localization has been demonstrated for some lipid raft markers, including proteins and lipids. However, few studies have characterized the spatial distribution of these clusters at the whole-cell scale. Nonetheless, these data have independently described specific ways that, either alone or in combination, can account for PM spatial organization in plant cells (Ovecka et al., 2010; Li et al., 2012; Gerbeau-Pissot et al., 2014; Sandor et al., 2016), and they justify a comprehensive analysis of the global mechanisms responsible for the distribution of ordered domains. In order to obtain a snapshot evaluation of the respective influence of each of these different molecular and cellular elements on the features of PM lateral organization, we have compared the biophysical properties of several experimental systems with decreasing complexity, all of which correspond to tobacco BY-2 cells in a resting state. We used di-4-ANEPPDHQ, a lipid packing-sensitive dye capable of assessing different levels of membrane order, coupled with confocal microscopy. This approach allowed us to measure the global level of membrane order of the entire PM together with the spatial distribution of ordered domains in either intact tobacco cells, cells with a disrupted cytoskeleton, or cells devoid of a cell wall. The same parameters were also measured in model membranes including giant unilamellar vesicles (GUVs) composed of lipids extracted from the tobacco PM, and giant vesicles of native PMs (GVPMs) from tobacco cells (see Fig. 1 for rationale). Our comprehensive comparisons of the individual ability of membrane and cell elements to modify membrane order revealed a major role for lipids in promoting the formation of ordered domains, and the capacity of proteins to limit this formation. Our results suggest that PM components appear to act together to regulate plant PM heterogeneity, whilst neither the cytoskeleton nor the cell wall seem to play a significant role in the short-term control of ordered domain distribution within steady-state cell PMs. Phosphorylation events also failed to regulate PM organization of tobacco cells in a resting state. Finally, we discuss how our results may support the ‘lipid raft’ model in our understanding of the mechanisms that control the spatial distribution of ordered domains within the plant PM. Fig. 1. Open in new tabDownload slide Different models used to characterize the cellular determinants of plasma membrane order. (A) Intact whole BY-2 suspension cells in which the plasma membrane, composed of ordered domains (in orange), is tightly connected with the underlying filaments of the cytoskeleton network and/or the surrounding fibres of the cell wall meshwork. (B) Treatment of BY-2 suspension cells with chemicals disrupts the cytoskeletal components. (C) Protoplasts obtained by enzymatic digestion of BY-2 suspension cells. (D) Giant vesicles were either formed from the whole diversity of tobacco PM lipids (giant unilamellar vesicles, GUVs) or from the direct electrofusion of purified PM vesicles, thus containing both proteins and lipids in their original amounts (giant vesicles of native plasma membrane, GVPMs). Samples were observed by microscopy, using either differential interference contrast (DIC; in grey-scale), or fluorescence after di-4-ANEPPDHQ labelling (excitation at 488 nm; emission acquired in a 520–680 nm band-pass, in yellow). Scale bars are 5 µm. Materials and methods Cell growth conditions Wild-type BY-2 (Nicotiana tabacum cv. Bright Yellow-2) cells were grown in Murashige and Skoog (MS) modified medium (basal salt mixture, M0221, Duchefa) at pH 5.6, supplemented with 1 mg l–1 thiamine-HCl, 0.2 mg l–1 2,4 dichlorophenylacetic acid, 100 mg l–1 myo-inositol, 30 g l–1 sucrose, 200 mg l–1 KH2PO4, and 2 g l–1 MES. Cell suspensions were maintained under continuous light conditions (200 µE m–2 s–1) on a rotary shaker (140 rpm) and diluted (4:80) weekly into fresh medium. Chemicals treatments BY-2 cells were equilibrated according to Gerbeau-Pissot et al. (2014). After a 2-h cell incubation period, concentrated stock solutions (1000× in DMSO) of the cytoskeleton inhibitors cytochalasin D, latrunculin B, nocodazole, and oryzalin (Sigma-Aldrich), were individually added to cell suspensions at a final concentration of 50 µM, 10 µM, 20 µM, and 10 µM, respectively. Control cells were incubated with the same dilution of DMSO. Cells were treated for 1 h on a rotary shaker (120 rpm) at 25 °C before observation. Cells were subsequently plasmolysed in I2 (0.5 mM CaCl2, 0.5 mM K2SO4, and 2 mM MES, pH 5.9) containing 400 mM mannitol (instead of 175 mM used by Gerbeau-Pissot et al., 2014) for several minutes before observation. Staurosporine (Sigma-Aldrich) was added to the cell suspension from concentrated stock solutions in DMSO, taking care not to exceed the final DMSO concentration (0.5 % v/v). Protoplast preparation and cell wall regeneration The protoplast preparation protocol was adapted from (Zaban et al., 2013). All steps were performed in sterile conditions. The BY-2 cells were collected and centrifuged at 100 g. Cells were washed in 0.4 M mannitol at pH 5.5 and centrifuged again, then resuspended in an enzymatic solution (Pectolyase Y23 0.1 % w/v, cellulose Onozuka RS 1 % w/v in 0.4 M mannitol at pH 5.5) and digested for 4–5 h at 25 °C, under shaking at 65 rpm in Petri dishes. Protoplasts were harvested after centrifugation (500 g for 5 min) and washed three times in FMS wash medium (4.3 g l–1 MS salts, 100 mg l–1 myo-inositol, 0.5 mg l–1 nicotinic acid, 0.5 mg l–1 pyridoxine-HCl, 0.1 mg l–1 thiamine, 10 g l–1 sucrose in 0.25 M mannitol, pH 5.8). For cell wall regeneration, protoplasts were transferred to FMS-store medium (FMS with 0.1 mg l–1 1-naphthaleneacetic acid and 1 mg ml–1 benzylaminopurin) and incubated at 26 °C in the dark, with shaking in Petri dishes. Protoplasts were observed at 0, 24 h, 48 h, and 5 d after digestion. Preparation of GUVs Giant unilamellar vesicles (larger than 10µm) were prepared as follows. Tobacco PM isolation PM fractions were obtained from BY-2 cells by membrane partitioning in an aqueous polymer two-phase system with polyethylene glycol 3350/dextran T-500 (6.6% each), according to Mongrand et al. (2004). Protein content was quantified using the Bradford method, in order to obtain an aliquoted solution of 10 mg ml–1 final concentration. Purification and quantification of tobacco PM lipids Polar lipids were extracted according to three independent methods detailed in Cacas et al. (2016) and based on different extraction solvent mixtures, namely chloroform/methanol/HCl (200/100/1, v/v/v), methyl tert-butyl ether (MTBE)/methanol/water (100/30/25, v/v/v), or a lower phase of propan-2-ol/hexane/water (55/20/25, v/v/v). GIPCs were purified according to a method adapted from Carter and Koob (1969) to obtain sufficient amounts for GUV production and lipid quantification, as described in Buré et al. (2011). Extracted lipids were dissolved in chloroform/methanol/water (30/60/8, v/v/v) for storage and further quantified by GC-MS according to Buré et al. (2011). GUV production GUVs were prepared by electroformation in a flow chamber (ICP-25 Perfusion Imaging Chamber, Dagan) connected to a function generator (PCGU1000, Velleman) and a temperature controller (TC-10, Dagan). Tobacco PM fractions (2 µg of proteins) or a mixture of tobacco PM lipids corresponding to a final phospholipid/sphingolipid/sterol composition of 4/4/1.5 (w/w/w, 2 µg final) were deposited on two microscope slides (18 × 18 mm) coated with electrically conductive indium tin oxide (resistivity 8–12 ohms). Lipid-coated slides were placed under a vacuum and away from light for at least 2 h until a thin film was obtained. Cover slips were set up in the flow chamber, and the lipid layer was rehydrated with 200 µl of swelling solution (25 mM HEPES, 10 mM NaCl, and 100 mM sucrose) pre-heated to 40 °C for lipid GUVs. A voltage of 3.5 V (adjustable during the experiment) at 10 Hz and a temperature of 40 °C were applied for a 2-h minimum period in a light-protected environment. After lipid swelling, the temperature of the chamber was slowly cooled to 22 °C (2 h minimum cooling time). Fluorescence labelling To examine cytoskeleton integrity, rhodamine-phalloidin (Invitrogen, 0.1 mg ml–1, 30 min) and Tubulin TrackerTM (Invitrogen, 50 µM, 45 min) were used to detect actin filaments and microtubules, respectively. To determine whether the cell wall was present samples were examined after staining with calcofluor-white (Sigma-Aldrich, 0.01 %, w/v) for several minutes. The resulting fluorescence signal of this product reveals cellulose and chitin structures. To determine the membrane order, GUV suspensions or tobacco cells were labelled with the fluorescent probe 1-[2-Hydroxy-3-(N,N-di-methyl-N-hydroxyethyl)ammoniopropyl]-4-[β-[2-(di-n-butylamino)-6-napthyl]vinyl] pyridinium dibromide (di-4-ANEPPDHQ; Invitrogen, stock solution in DMSO, 3 µM final, 1–2 min). Confocal fluorescence microscopy Labelled cells were deposited between the slide and the cover slip and observed with a Leica TCS SP2-AOBS laser scanning microscope (Leica Microsystems) coupled to a HCPL Apochromat CS 63× (N.A. 1.40) oil immersion objective. Fluorescence excitations were obtained using either the 543-nm line of a helium-neon laser (rhodamine-phalloidin), the 488-nm line of an argon laser (tubulin tracker), or a 405-nm diode (calcofluor). Fluorescence emissions were recorded between 555–700 nm (rhodamine-phalloidin), 500–600 nm (tubulin tracker), and 410–480 nm (calcofluor). For di-4-ANEPPDHQ observation, after excitation at 488 nm, emission intensities were acquired between 540–560 nm (green image) and between 650–670 nm (red image). Ratiometric imaging was performed using the ImageJ software (http://imagej.nih.gov/ij/). Fluorescence spectroscopy Sample solutions (1 ml) were placed in a 10-mm special optic path glass cuvette filled in a thermoelectric cooler (24 °C, Wavelength Electronics, Inc.). Fluorescence measurements were performed using a Fluorolog-3 FL3-211 spectrometer (Jobin-Yvon, Horiba Group). The emission spectrum was monitored by one photomultiplier (520–700 nm). A xenon arc lamp (488 nm) was used as the light source. Data acquisitions were performed using the Datamax software (Jobin-Yvon/Thermo Galactic, Inc.). Statistical tests Statistical analyses were based on non-parametric tests (Mann–Whitney), since we observed that our data exhibited a non-Gaussian distribution. Results Phosphorylation does not regulate the level of PM order or the distribution of ordered domains We previously described the heterogeneity of tobacco PM lateral organization with respect to local membrane order by measuring the distribution of the liquid-ordered/liquid-disordered (Lo/Ld) phases (Gerbeau-Pissot et al., 2014). In order to identify key players that govern this PM lateral organization, we exploited the fluorescence properties of the environment-sensitive dye di-4-ANEPPDHQ (Jin et al., 2005), the red/green ratio of which is inversely correlated with the level of membrane order (Jin et al., 2006). BY-2 suspension cells were stained for 2 min with di-4-ANEPPDHQ (3 µM) and observed by confocal microscopy. We then measured the fluorescence intensities of the entire PM and evaluated the membrane order using the RGM parameter (red-to-green ratio of the membrane, RGM=I660/I550). To characterize the spatial distribution of PM ordered domains, we calculated the RGM value of the PM in tobacco cell regions, hereafter referred to as RGR for ‘red-to-green ratio of the ROI’, where ROI is a ‘region of interest’ corresponding to a 300 × 300 nm square. To test for the contribution of active pathways in the control of these parameters, tobacco suspension cells were treated with staurosporine, a broad-spectrum protein-serine/threonine kinase inhibitor (Suzuki and Shinshi, 1995). Addition of staurosporine (2.5 µM) failed to modify the RGM, regardless of the incubation time (from 10 min to 3 h, Fig. 2). In contrast, we were able to determine a significant modification in membrane order using the same experimental set-up under conditions known to induce an increase PM order (Supplementary Fig. S1 at JXB online). This indicates that the technique is sensitive enough to detect minor variations, and that the conditions are sufficient to inhibit signalling events (Bonneau et al., 2010; Sandor et al., 2016), demonstrating that the maintenance of membrane order is independent of phosphorylation events. These results suggested that the signalling pathway was not involved in the control of PM order during the course of the experiment. The results prompted us to address the issue of short-term regulation of ordered domain organization via other mechanisms, especially the contribution of physical interactions. Fig. 2. Open in new tabDownload slide Effect of a protein kinase inhibitor on membrane order of living tobacco cells. After incubation (for 10 min, 1 h, or 3 h) with the kinase inhibitor staurosporine (Stau; 2.5 µM), cells were labelled for 2 min with di-4-ANEPPDHQ (3 µM). Cells were subsequently analysed by spectrofluorimetry, and membrane order was quantified using the red-to-green membrane fluorescence ratio (RGM, =I660/I550). The RGM of PMs from treated cells is shown as a relative value compared to control cells (Ctl, with equivalent volume of DMSO). Data are means (±SD), n=26 independent experiments. Cytoskeleton remodelling does not modify the organization of PM-ordered domains To analyse the potential relationship between the cytoskeleton and BY-2 cell PM order, we measured the evolution of PM order in response to different compounds that affect cytoskeleton components. Brief incubations were performed to exclude any long-term metabolic regulation. Latrunculin B (Spector et al., 1983) and cytochalasin D (Cooper, 1987) bind to actin monomers and prevent their polymerization; their application led to the disruption of actin filaments (Fig. 3A), without any effect on plant cell viability during the 1-h experiment (Supplementary Fig. S2). Monitoring single cells by confocal microscopy showed that these treatments failed to significantly change BY-2 cell RGM (Fig. 3B). No significant differences in RGM were observed between control cells and cells treated with nocodazole (Samson et al., 1979) or oryzalin (Morejohn and Fosket, 1984) (Fig. 3B), both of which can interfere with microtubule polymerization and cause disturbance of the cytoskeleton (Fig. 3C) without affecting cell viability (Supplementary Fig. S2). Spectrofluorimetry measurements, which can assess PM order in batches of thousands of live tobacco cells, confirmed that PM order was independent of actin filament or microtubule polymerization (Supplementary Fig. S3). Simultaneous addition of latrunculin B and oryzalin also did not result in any significant differences between the RGM of control and treated cells (Supplementary Fig. S3), indicating the independence of this parameter with regards to cytoskeleton integrity in living tobacco cells. Fig. 3. Open in new tabDownload slide Influence of the cytoskeleton on the membrane order of tobacco cell PMs. (A, C) Effect of pharmacological treatments on cytoskeleton integrity. BY-2 cells were incubated for 1 h with either latrunculin B (Lat, 10 µM), cytochalasin D (Cyt, 50 µM), nocodazole (Noc, 20 µM), or oryzalin (Ory, 10 µM), or with the same concentration of DMSO (control, Ctl). The cytoskeleton was observed by fluorescence microscopy using (A) rhodamine-phalloidin for actin colouration (0.1 mg ml–1) and (C) tubulin tracker for microtubule staining (50 µM). After 1 h, patches were detected on treated cells, in comparison to the intact network of filaments observed on control cells. Scale bars are 20 µm. (B) Effect of cytoskeleton integrity on global membrane order. BY-2 cells were exposed to pharmacological treatments that disrupted the actin or tubulin meshwork (as detailed above), or both (Lat, 10 µM and Ory, 10 µM). After 1 h of incubation, cells were labelled for 2 min with di-4-ANEPPDHQ (3 µM). Cells were then observed by confocal microscopy, and membrane order was quantified using the red-to-green membrane fluorescence ratio (RGM, =I660/I550). The RGM values obtained for treated cells are shown relative to the values of untreated cells (Ctl, in DMSO). Data are means (SEM), n=96–428 from at least five independent experiments. (D) Effect of cytoskeleton integrity on the organization of ordered domains. The distribution of red-to-green membrane fluorescence values (RGR) of individual regions of interest (ROIs, 300 × 300-nm squares) of the PM (control conditions, Ctl) is not influenced by latrunculin B, 10 µM, 1 h). The x-axis corresponds to the class of RGR values, and only the maximal value of each class is indicated on the graph. The y-axis corresponds to the ROI percentage of each class. Data are means (±SEM), n=111–428 cells from at least five independent experiments. However, modifications to the lateral organization of membrane order could occur at the nanometre scale without affecting the global RGM. Indeed, the RGM represents the mean value of a multitude of small areas (ROIs) that exhibit different levels of local membrane order (RGR values): the distribution of individual values may be different even though the overall average remains the same. To test this possibility, the emission signals of fluorescently labelled cells were acquired on a tangential plane corresponding to membrane surfaces of 100–500 µm2, and the relative abundance of Lo/Ld phases was evaluated. We did not observe any significant difference in the distribution of RGR values between control and latrunculin B-treated cells (Fig. 3D). Furthermore, all of the microtubule- and actin-depolymerizing agents that we tested failed to modify the distribution of RGR values (Supplementary Fig. S4). A granulometric analysis characterizing the aggregation of the most ordered domains was then performed, with a focus on ROIs exhibiting a value in the first quartile of RGR values (Fig. 4–C). This approach did not reveal any significant difference between the size of ordered domains in cells treated with cytoskeletal disruptors and the control cells (Fig. 4D). Taken together, our data suggest that cytoskeleton integrity has no short-term effect on the level of membrane order, either globally (across the entire PM surface) or locally (within small areas at our scale of observation). Fig. 4. Open in new tabDownload slide Influence of the cytoskeleton on the spatial distribution of ordered domains. (A) Observation of di-4-ANEPPDHQ-labelled membrane surfaces (excitation at 488 nm; emission corresponds to a 520–680 nm band-pass) with the grey-scale representing fluorescence intensity. (B) The subsequent ratiometric image describing the red-to-green fluorescence ratio of the PM region of interest (RGR) of di-4-ANEPPDHQ within 300 × 300-nm membrane areas displayed in a grey-scale colour-coded representation. (C) A binarized image, representing a detail extracted from the membrane surface focused on the most ordered domains. Recorded regions of interest (ROIs) exhibiting an RGR value within the first quartile of lower RGR values (the most ordered ones) are represented as black pixels. (D) A granulometric approach was then used to compare the spatial distribution of ordered domains on the membrane surface under the different experimental conditions. Tobacco suspension cells were treated for 1 h with a cytoskeletal-active compound (latrunculin B, Lat, 10 µM; cytochalasin D, Cyt, 50 µM; nocodazole, Noc, 20 µM; or oryzalin, Ory, 10 µM) or a control (Ctl) before labelling with di-4-ANEPPDHQ (3 µM, 2 min). The mean area of the black ROI groups is shown. Data are means (±SD), n>27 cells from five independent experiments. No significant differences were observed (P>0.05). The cell wall does not affect the organization of PM-ordered domains To investigate the influence of the cell wall on PM order, we compared the RGM of freshly prepared protoplasts devoid of cell walls (1–3 h after enzymatic digestion) and after 24 h of cell wall regeneration (Fig. 5A). The presence of a newly synthesized cell wall, visualized by staining with calcofluor-white (Fig. 5A), did not modify the RGM value (Fig. 5B), suggesting cellulose deposition has no role in the control of PM order. Correspondingly, no modifications in RGM were measured between control and plasmolysed cells when plasmolysis was induced and protoplast shrinkage kept the PM away from the cell wall (Supplementary Fig. S5), suggesting no role in membrane packing for the bounded regions between the cell wall and the PM. Furthermore, the absence of a direct effect of the cell wall suggests that there is no direct contribution from cell wall–PM connections on the regulation of PM order. Fig. 5. Open in new tabDownload slide Influence of the cell wall on PM order of BY-2 protoplasts. (A) Cell morphology (differential interference contrast, DIC) and cellulose deposition (fluorescence imaging after calcofluor-white coloration) were analysed for protoplasts, either freshly prepared (3 h) or after cell wall regeneration (1 d after protoplast preparation). (B–D) Confocal microscopy was used to characterize di-4-ANEPPDHQ-labelled protoplasts (3 µM, 2 min), either freshly prepared (FP) or after cell wall regeneration (PCW). (B) Global membrane order was quantified using the red-to-green membrane fluorescence ratio (RGM, =I660/I550). (C) Local membrane order was estimated in each region of interest (ROI) of the fluorescent PM, and the distribution of RGM values of individual ROIs of corresponding PMs (RGR) is shown. The x-axis corresponds to the class of RGR values, and only the maximal value of each class is indicated on the graph. The y-axis corresponds to the ROI percentage of each class. (D) The size of ordered domains was measured as the mean area of groups of pixels corresponding to ROIs exhibiting an RGR value belonging to the first quartile of lower values. Data are means (±SD), n>66 cells from five independent experiments. A characterization of the abundance of ordered domains was then performed, although the spherical shape of protoplasts limited the size of the tangential area that could be analysed (10–100 µm2). No significant difference was observed between the freshly prepared protoplasts and protoplasts in which the cell wall had been regenerated (Fig. 5C), with both conditions displaying the same distribution of RGR values. Live ratiometric imaging also indicated a similar size for ordered domains in protoplasts, before and after cell wall regeneration (Fig. 5D). Thus, the cell wall does not seem to influence the spatial distribution of ordered domains within the tobacco PM. Production of giant vesicles from tobacco PMs reveals the involvement of lipids and proteins in the control of the spatial distribution of ordered domains The data presented above highlight the possibility of a crucial role for intrinsic PM components with regards to PM-ordered domains. To assess the involvement of a wide diversity of PM lipids in the regulation of membrane order, we prepared giant unilamellar vesicles (GUVs) using a mixture of the different classes of lipids in their relative proportions as found in native BY-2 cell PMs. A careful lipidomic analysis highlighted the large diversity of BY-2 PMs (Supplementary Fig. S6), with a phospholipid/sphingolipid/sterol ratio of 4/4/1.5 (w/w/w). The RGM value of these GUVs labelled with di-4-ANEPPDHQ was 0.89 ± 0.16 (±SD, n=29). Taking into account the vast array of lipids that comprise PMs, we then characterized the ability of fatty acid saturation to increase membrane order. The results showed that RGM significantly decreased from 2.47 ± 0.24 (n=38) to 1.49 ± 0.16 (n=20) for GUVs consisting of only 1,2 dioleoylphosphatidylcholine (DOPC) or DOPC/DPPC (DPPC: 1,2 dipalmitylphosphatidylcholine; 1/1, mol/mol), respectively (Supplementary Fig. S7). The complexity level of different lipid combinations (in terms of both number and composition of the different lipid families) similarly modified the GUV membrane order (Supplementary Fig. S7). Overall, this suggests that the diversity of lipid molecules represented in BY-2 cell PMs, together with their specific ability to organize the membrane (Grosjean et al., 2015), could be responsible for the high membrane order reported here for GUVs composed of tobacco PM lipids. In addition to lipid–lipid interactions, protein–lipid interactions are essential contributors to PM organization (van den Bogaart et al., 2011). Hence it is of primary interest to compare the characteristics of GUVs composed of tobacco PM lipids (as detailed above) with GUVs containing lipids and proteins from tobacco PMs in their native amounts. However, the molecular mechanisms involved in PM vesicle fusion induced by detergents cause significant artefacts, e.g. bilayer–micelle transition (Alonso et al., 1982). Furthermore, technical limitations restrict the protein levels included in these proteoliposomes to 5% (Kahya et al., 2005), far below the level of native PMs, which is estimated to be 50% by weight; consequently, the use of these approaches is restricted. Giant PM vesicles (GPMVs), corresponding to PM blebs detached from cells, have a protein and lipid diversity mirroring the native PM (Scott, 1976) and separate into co-existing Lo/Ld phases, which enables the investigation of the structural determinants of ordered domain association (Baumgart et al., 2007; Kaiser et al., 2009; Levental et al., 2011). However, the cell wall surrounding the plant cell prevents the use of this procedure for tobacco cells. We therefore developed a new protocol to produce giant vesicles of native PMs (GVPMs) by electrofusing small vesicles of purified PM fractions. However, GUV formation using purified PM fractions was extremely difficult due to the presence of proteins and the obstacle they presented to fusion, and so the electroformation method was modified by varying time, temperature, osmolarity, voltage, and frequency in order to enable a high yield of GVPM formation (Supplementary Fig. S8A). By comparing the different protocols, we were able to efficiently produce GVPMs (103 vesicles starting from 2 µg lipids) with a size amenable to observation by confocal microscopy (Supplementary Fig. S8B). Under optimized conditions, 20% of the GVPM population had a diameter larger than 15 µm. The procedure (described in detail the Methods section) can now be used routinely to efficiently prepare GVPMs directly from purified PMs. As has been previously reported (Takahashi et al., 2013), protein abundance can modify the contours of vesicles, allowing the formation of soft GVPMs with a non-spherical shape (Fig. 6A). Moreover, GVPMs containing proteins and lipids exhibited a higher RGM than GUVs formed with the same lipid mixture (Fig. 6B), indicating that proteins tend to limit the packing of the membrane. Fig. 6. Open in new tabDownload slide Influence of lipids and proteins on the global and local order of tobacco PMs. Giant unilamellar vesicles (GUVs) were produced by mixing extracts of purified lipid classes according to their relative amounts within the membrane (phospholipids/phytosphingolipids/phytosterols: 4/4/1.5 w/w/w) or by fusing isolated PM vesicles (giant vesicles of native plasma membrane, GVPMs). Vesicles were labelled with di-4-ANEPPDHQ (3 µM, 2 min). (A) Transverse (top) and tangential (below) views of fluorescent vesicles were made using confocal microscopy (excitation at 488 nm; emission corresponds to the sum of fluorescence intensities acquired in a 520–680 nm band-pass) of GUVs and GVPMs. (B) The red-to-green fluorescence ratio (RGM, =I660/I550) of fluorescent vesicles was measured using confocal microscopy (dividing the red, 545–565 nm, by the green, 635–655 nm, emission band-passes). Data are means (±SD), n>24 experiments; significant difference: *P<0.05. (C) Individual regions of interest (ROIs, 300 × 300 nm) of the surface of fluorescent vesicles were classified according to their RGM values (i.e. giving RGR values). The RGR distribution of a representative GUV composed of a PM lipid mixture is compared to the distribution of GVPMs. The x-axis represents the class of RGR values; only the maximal value of each class is indicated on the graph. The y-axis represents the percentage of each class of ROI values. (D) The size of ordered domains was measured as the mean area of groups of pixels corresponding to ROIs exhibiting an RGR value belonging to the first quartile of lower values, and is compared for giant vesicles composed of either PM lipids (GUV) or PM lipids and proteins (GVPM). Data are means (±SD), n>20 vesicles from five independent experiments. To compare local membrane order at the surface of these GUVs and GMPVs, the emission signals of fluorescently labelled vesicles were acquired on a tangential plane corresponding to membrane surfaces of 100–350 µm2 (Fig. 6A), and the distribution of RGR values was determined (Fig. 6C). The distribution for PM lipid GUVs was centred towards low values (from 0.6–1.1; Fig. 6C), consistent with the high membrane order of these GUVs (Fig. 6B). In contrast, the RGR value of GVPMs exhibited a more peaked distribution shifted towards higher RGR values (from 1.1–1.4; Fig. 6C), suggesting that the presence of protein induced a shift towards fewer ordered domains. To investigate a possible concomitant adjustment of the spatial organization of ordered domains, a granulometric analysis was performed by focusing on ROIs that exhibit an RGR value in the first quartile of values, in order to eliminate any density effects. We measured an ordered domain size of 0.342 µm2 for GVPMs (Fig. 6D); interestingly this was similar to the value for BY-2 cells (i.e. 0.350 µm2, Fig. 4D). Moreover, this size tended to be lower than the size of GUVs composed of a PM lipid mixture (Fig. 6D), suggesting a negative impact of the presence of protein on both the quantity and size of ordered domains at our scale of observation. To better understand the dissolution of the ordered domains in native PM conditions, ratiometric images were further segmented to only take into account ROIs exhibiting red/green ratios below a RGR value of 1.2, which corresponds to one population of ordered domains (Grosjean et al., 2015). This ordered domain fraction was calculated, and a significant decrease was observed for GVPMs in comparison to GUVs (Supplementary Fig. S9A), confirming that the reduction in ordered domain abundance was predominantly involved in the low membrane order reported for GVPMs (Fig. 6B). To determine the lateral organization of these ordered domains within PM vesicles, we analysed their degree of clustering. The group size of ordered domains revealed a protein-dependent decrease, with the presence of ordered domains exhibiting a mean size of ~0.8 µm2 in GUVs composed of only PM lipids and ~0.6 µm2 in GVPMs (Supplementary Fig. S9B). We observed a linear correlation between the size of ordered domains characterizing GUVs composed of different lipid mixtures and the membrane order (Supplementary Fig. S9C). This correlation disappeared in GVPMs (Supplementary Fig. S9C), suggesting that the presence of proteins decreases the number of ordered domains, but concomitantly induces another mechanism that can reduce the propensity of ordered domains to lie within clusters. Taken together, these results support the ability of PM lipids and proteins to finely govern PM order, and to subsequently control the tenuous organization of the PM. Discussion Lipids and proteins can account for the structuring of plant PM ordered domains In this study we determined a high level of membrane order for GUVs mimicking the native composition of tobacco BY-2 PM lipids. This included 38% GIPCs, 8% free phytosterols, and 10% conjugated phytosterols, which were associated with a large and continuous distribution of individual levels of membrane order exhibited by different membrane regions of these vesicles. One possible explanation is that the ‘lipid raft’ model is also applicable to plants. This model places the local interactions between lipid species (Ramstedt and Slotte, 2006; Quinn, 2010) as the first level of membrane organization (Lingwood and Simons, 2010). Consistent with this hypothesis, plant-specific conjugated sterols display a striking ability to induce ordered domain formation in the membrane that acts in synergy with the similar ability of free phytosterols (Grosjean et al., 2015). Indeed, phytosterols increase membrane stiffness and the inclusion of other lipids, depending on their structure and shape (Shahedi et al., 2006), resulting in macromolecular assemblies and lipid bilayer ordering (Mannock et al., 2015; Shaghaghi et al., 2016). Furthermore, formation of sterol-dependent membrane domains is modified by the addition of GIPCs (Grosjean et al., 2015). This adds weight to a model in which the combination of multiple molecular species of phytosphingolipids and phytosterols present in living tobacco cell PMs may induce membrane ordering and enhance the formation of various ordered domains. The high diversity of plant PM lipids could then increase opportunities for local interactions with different intensities, and consequently bring about heterogeneity at the local membrane compaction level, providing an explanation for the wide range of RGR values measured for GUVs consisting of tobacco PM lipid components. RGR values, which corresponded to average ratios of the smallest area optically possible to analyse (a 300 × 300 nm ROI), might reflect the mean membrane order of sub-population domains co-existing in varying proportions within these areas. Such a complex model assumes a multitude of nanodomains with levels of order between (rather than strictly corresponding to) the Lo and Ld phases (Bagatolli et al., 2010). In agreement with this new degree of complexity of PM organization, the diversity of mammalian lipid mixtures has been speculated to favour the formation of ultra-nanodomains (Pathak and London, 2015). In accordance with this, the formation of distinct nanodomains has been simulated in the outer leaflet of an idealized mammalian PM consisting of a complex mixture of 63 different lipid species (Ingólfsson et al., 2014). To investigate the influence of protein–lipid interactions on PM lateral organization, we characterized GVPMs produced from tobacco BY-2 cell PMs using a novel procedure. These GVPMs, which contain hundreds of integral membrane proteins and lipids (corresponding to native PM amounts), provide a very interesting system, and yet they are rarely used due to the difficulty in obtaining them. In contrast to GPMVs, which are isolated after chemical treatment of animal cells that induces formation of detachable PM blebs (Levental and Levental, 2015), the electrofusion procedure utilized here allows the initial characterization of isolated PM organization in a native resting state. In agreement with our observations on tobacco BY-2 cells, GPMVs from animal cells have shown a clear segregation into Lo/Ld phases, with a lateral distribution that depends on the overall protein content (Baumgart et al., 2007). Using a cell-swelling procedure to isolate PM spheres, the cholera toxin B subunit-dependent formation of ordered domains has also previously been reported (Lingwood et al., 2008). Here, by comparing the GUVs of PM lipids and the GVPMs, we further demonstrated the involvement of PM proteins in the modulation (especially loosening) of plant PM order. Indeed, when proteins were positioned between lipid molecules, they increased the membrane line tension and modified the mean size of the ordered domains in the tobacco PM, which has also been reported for lung surfactant monolayers (Dhar et al., 2012). Peptides with a short hydrophobic transmembrane domain accordingly decrease the affinity of sterols for neighbouring phospholipids (Nyström et al., 2010; Nyholm et al., 2011; Ijäs et al., 2013), suggesting that proteins could modify membrane lateral organization by excluding certain lipids from the closest surrounding bilayer. Moreover, lipids in direct interaction with proteins result in areas that are much less compact in the immediate vicinity of the proteins (Brannigan and Brown, 2006), since protein–lipid interactions depend on the size and the charge of chemical groups present at the protein surface (Honig et al., 1986). The anchoring of transmembrane proteins into ordered domains is also hypothesized to redistribute ordered domains (Epand et al., 2004; Epand, 2008). We are thus able to propose a final model in which lipid–lipid interactions may control the formation of ordered domains at the plant PM surface (whereas protein–lipid and/or protein–protein interactions contribute to drive the whole-membrane organization). With this model in mind, we noticed a remarkable limitation in the size of ordered domains within GVPMs, which could undoubtedly be attributed to the presence of proteins. Indeed, GVPMs and living cells both exhibit similar small ordered domains, and this small size originates from the ability of proteins to emulsify ordered domains (Bhatia et al., 2016). This PM organization should be based on hydrogen bonds and van der Waals interactions operating on a time scale representative of what we observed, and should show, in a coherent manner, an insensitivity to protein kinase inhibitors. The cell wall–PM–cytoskeleton continuum is not a common hallmark of domain assembly Besides local membrane composition, domain formation could also be related to a limited lateral diffusion of PM components. The cortical cytoskeleton in particular has been proposed to generate barriers that constrain movement in the membrane, as revealed by the restricted diffusion of PM proteins in certain membrane compartments (Kusumi et al., 2005; Umemura et al., 2008). Furthermore, the hindered diffusion of phospholipids and sphingolipids was similarly abolished in actin cytoskeleton-free cell-derived GPMVs, as measured using super-resolution stimulated emission depletion microscopy combined with fluorescence correlation spectroscopy (Schneider et al., 2017). These observations support the ‘picket fence’ model, in which transmembrane proteins, like pickets, are anchored to and lined up along a ‘fence’ of cytoskeletal proteins surrounding the confinement zones (Kusumi et al., 2005). In plants, single-particle tracking analysis has recently revealed that cytoskeleton integrity, especially microtubules, restricts the lateral mobility of plant innate immunity proteins such as AtHIR1 at the PM surface of Arabidopsis cells (Lv et al., 2017). However, quantifying protein diffusion using fluorescence recovery after photobleaching experiments has previously demonstrated that the cytoskeleton is not responsible for the relative immobility of plant PM proteins (Martinière et al., 2012). This apparent contradiction could be explained by different sensitivities to microtubule disturbance depending on the particular protein that is observed (Szymanski et al., 2015). This shows that multiple types of membrane domains (which are specifically enriched in one or the other of these proteins) co-exist at the same time in plants (Jarsch et al., 2014). Some of these domains may reflect the formation of specific areas enriched in PM components that are temporarily trapped by the cytoskeleton network within distinct PM sub-regions, as has been proposed in animal cells (Murase et al., 2004). Among these domains, the composition of DRM-associated domains has been shown to be under the control of the cytoskeleton network, since microtubules regulate the dynamics of DRM-marker proteins such as Arabidopsis Flot1 (Li et al., 2012) or Arabidopsis remorin (Szymanski et al., 2015). In plants, these domains may co-exist with other domains whose formation has been proposed to be directly subject to lipid and protein composition (Urbanus and Ott, 2012), and that have been revealed in our present study. Indeed, by comparing the membrane organization of living tobacco cells affected (or not) in the structure of their actin and/or tubulin networks, we have shown that the cytoskeleton does not impact on the short-term regulation of the distribution of PM-ordered domains in tobacco. To the best of our knowledge, only one previous study has reported that alteration of either actin content or its association with the PM affects the physical properties of the PM in animal cells, e.g. disruption of the cortical cytoskeleton coinciding with specific limitations of the ordered domain fraction (Dinic et al., 2013). However, this actin-dependent formation of ordered domains was demonstrated to occur at 37 °C. At this temperature, lipid–lipid interactions reduce the formation of ordered domains in GPMVs (Baumgart et al., 2007), giving prominence to interactions between the membrane and intracellular filaments. In line with this, it has been proposed that ordered domains are too small to be optically detected at high temperature, regardless of the lipid composition (Pathak and London, 2015). In order to be able to compare our different systems, we performed all of our experiments at room temperature (24 °C), as it is convenient for cultivating plant cells. At this temperature, we observed that the presence and characteristics of ordered domains essentially depended on the lipid and protein mixtures present within the membrane. In this study, we attempted to identify all the cellular elements underlying the heterogeneity of PM biophysical properties in plant cells at a specific time, using as a system a tobacco cell PM that had already been synthesized and was in a steady state. The results we obtained in characterizing the membrane order of protoplasts indicated that the cell wall does not play a key role in the maintenance of membrane order, or in the spatial organisation of ordered domains in resting tobacco cells. Even though direct interactions between PM proteins and cell wall components are known to constrain the lateral diffusion of PM proteins (Martinière and Runions, 2013), no involvement of cell wall organization in PM packing has been reported to date, and our results tend to exclude such a hypothesis. Furthermore, the absence of an effect of short-term cytoskeleton disorganization also questions the relevance of the ‘picket fence’ hypothesis in plant cells. Instead, our data suggest that lipids and proteins are the main determinants involved in the organization of membrane order in tobacco BY-2 cells. In agreement with our model, studies on the brassinosteroid binding receptor and its co-receptor have shown that the formation of nanoclusters within the PM of Arabidopsis seedlings is mainly subject to biophysical restraints, whereas cytoskeleton disruption does not have any effect on this parameter (Hutten et al., 2017). The complex mechanisms by which the PM is synthesized and renewed might be a way to control the targeting of the different membrane components (Zarský et al., 2009; Kleine-Vehn et al., 2011), and consequently to control the interactions between neighbouring lipids and proteins. In accordance with this, modifications of PM composition in an Arabidopsis mutant have been reported to alter membrane order at the resting state (Sena et al., 2017). According to the limited data available, such a metabolic turnover of membrane components might occur over a time frame of several hours (Shi et al., 2008; Stanislas et al., 2009; Chalbi et al., 2015), which far exceeds our experimental conditions. Ordered domains are part of PM lateral heterogeneity Our results highlight the possible co-existence of distinct domains: the first corresponds to PM domains, in which proteins or oligomers of proteic complexes exhibit restricted lateral diffusion, and the second corresponds to ordered domains with a sterol-enriched composition. Thus, microscale domains of restricted diffusion proteins might be distinct from nanoscale ordered domains (observed here using a lipid packing-sensitive probe), since the mechanisms responsible for their formation are most probably different. One hypothesis in animals posits that PM organization synthesizes a multiscale organization of: (i) plasma membrane compartments (i.e. microscale-sized domains) partitioned through membrane component entrapment that depends on the actin-based cytoskeleton; (ii) lipid raft domains (2–20 nm) created via sphingolipid–sterol interactions; and (iii) protein complexes (3–10 nm) composed of dimer/oligomer assemblies (Nicolson, 2014). In addition to the cytoskeleton-dependent formation of PM sub-regions with restricted lateral diffusion of their components mentioned above, immuno-electron microscopy experiments in plants have revealed that the lipid PIP2 is partitioned into 25-nm clusters (Furt et al., 2010). Furthermore, the protein remorin is similarly aggregated into 70-nm domains in the cytosolic leaflet of tobacco leaf PMs (Raffaele et al., 2009). These two lines of evidence argue in favour of the existence of integrated levels of PM organization in plants similar to that found in animals. In the PM of tobacco cells in a resting state, domains of restricted lateral diffusion may thus co-exist with ordered domains that are compatible with the size investigated here, and for which we have demonstrated that lipid–lipid and/or protein–lipid interactions are the major driving forces of their formation. Interestingly, both of these domains have recently been implicated in plant stress responses, and notably in plant immunity (Bucherl et al., 2017). Future work should aim to unravel the mechanisms used to regulate the number of ordered domains, the dynamics of which are known to be involved in the early stages of defense responses (Liu et al., 2009; Gerbeau-Pissot et al., 2014; Sandor et al., 2016). Supplementary data Supplementary data are available at JXB online. Fig. S1. Effect of staurosporine on RGM of control and cryptogein-elicited BY-2 cells. Fig. S2. Influence of cytoskeletal-active compounds on the viability of tobacco cells. Fig. S3. Influence of the cytoskeleton on the membrane order of BY-2 PMs measured by spectrofluorimetry. Fig. S4. Comparison of the distribution of RGR values between control cells and cells treated with cytoskeletal-active compounds. Fig. S5. Influence of cell wall–PM connections on the level of tobacco cell PM order. Fig. S6. Lipid composition of PMs isolated from tobacco suspension cells. Fig. S7. Effect of different lipid compositions on the membrane order of giant vesicles. Fig. S8. Influence of swelling-solution composition on GVPM size. Fig. S9. Formation of large ordered domains in giant vesicles made up of tobacco PM lipids and proteins. Abbreviations: Abbreviations: DOPC 1,2- dioleoyl -sn-glycero-3- phosphocholine DPPC 1,2- dipalmitoyl -sn-glycero-3- phosphocholine GUV giant unilamellar vesicle GVPM giant vesicles of native plasma membrane GPMV giant plasma membrane vesicles RGM red-to-green ratio of membrane fluorescence RGR red-to-green ratio for a specific region of interest. Acknowledgements This work was partially supported by grants from the ’Région Bourgogne’ and the Bordeaux Metabolome Facility-MetaboHUB (grant no. ANR–11–INBS–0010) to SM. We wish to thank the Microscopy Centre INRA/Université de Bourgogne Franche-Comté of the DImaCell facility for technical assistance in confocal microscopy. We also acknowledge the Metabolome-Fluxome-Lipidome facility of Bordeaux (http://www.biomemb.cnrs.fr) for their contribution to the lipid analysis. We thank Brandon Loveall of Improvence for for providing English language editing for this paper. References Alonso A , Sáez R, Goñi FM. 1982 . The interaction of detergents with phospholipid vesicles: a spectrofluorimetric study . FEBS Letters 137 , 141 – 145 . Google Scholar Crossref Search ADS PubMed WorldCat Bagatolli LA , Ipsen JH, Simonsen AC, Mouritsen OG. 2010 . An outlook on organization of lipids in membranes: searching for a realistic connection with the organization of biological membranes . Progress in Lipid Research 49 , 378 – 389 . Google Scholar Crossref Search ADS PubMed WorldCat Baumgart T , Hammond AT, Sengupta P, Hess ST, Holowka DA, Baird BA, Webb WW. 2007 . Large-scale fluid/fluid phase separation of proteins and lipids in giant plasma membrane vesicles . Proceedings of the National Academy of Science, USA 104 , 3165 – 3170 . Google Scholar Crossref Search ADS WorldCat Bhatia T , Cornelius F, Ipsen JH. 2016 . Exploring the raft-hypothesis by probing planar bilayer patches of free-standing giant vesicles at nanoscale resolution, with and without Na,K-ATPase . Biochimica et Biophysica Acta 1858 , 3041 – 3049 . Google Scholar Crossref Search ADS PubMed WorldCat Bonneau L , Gerbeau-Pissot P, Thomas D, Der C, Lherminier J, Bourque S, Roche Y, Simon-Plas F. 2010 . Plasma membrane sterol complexation, generated by filipin, triggers signaling responses in tobacco cells . Biochimica et Biophysica Acta 1798 , 2150 – 2159 . Google Scholar Crossref Search ADS PubMed WorldCat Borner GH , Sherrier DJ, Weimar T, et al. 2005 . Analysis of detergent-resistant membranes in Arabidopsis. Evidence for plasma membrane lipid rafts . Plant Physiology 137 , 104 – 116 . Google Scholar Crossref Search ADS PubMed WorldCat Brannigan G , Brown FL. 2006 . A consistent model for thermal fluctuations and protein-induced deformations in lipid bilayers . Biophysical Journal 90 , 1501 – 1520 . Google Scholar Crossref Search ADS PubMed WorldCat Bucherl CA , Jarsch IK, Schudoma C, Segonzac C, Mbengue M, Robatzek S, MacLean D, Ott T, Zipfel C. 2017 . Plant immune and growth receptors share common signalling components but localise to distinct plasma membrane nanodomains . eLIFE 6 . Google Scholar OpenURL Placeholder Text WorldCat Buré C , Cacas JL, Wang F, Gaudin K, Domergue F, Mongrand S, Schmitter JM. 2011 . Fast screening of highly glycosylated plant sphingolipids by tandem mass spectrometry . Rapid Communications in Mass Spectrometry 25 , 3131 – 3145 . Google Scholar Crossref Search ADS PubMed WorldCat Cacas JL , Buré C, Grosjean K, et al. 2016 . Revisiting plant plasma membrane lipids in tobacco: a focus on sphingolipids . Plant Physiology 170 , 367 – 384 . Google Scholar Crossref Search ADS PubMed WorldCat Carmona-Salazar L , El Hafidi M, Enríquez-Arredondo C, Vázquez-Vázquez C, González de la Vara LE, Gavilanes-Ruíz M. 2011 . Isolation of detergent-resistant membranes from plant photosynthetic and non-photosynthetic tissues . Analytical Biochemistry 417 , 220 – 227 . Google Scholar Crossref Search ADS PubMed WorldCat Carter HE , Koob JL. 1969 . Sphingolipids in bean leaves (Phaseolus vulgaris) . Journal of Lipid Research 10 , 363 – 369 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Chalbi N , Martínez-Ballesta MC, Youssef NB, Carvajal M. 2015 . Intrinsic stability of Brassicaceae plasma membrane in relation to changes in proteins and lipids as a response to salinity . Journal of Plant Physiology 175 , 148 – 156 . Google Scholar Crossref Search ADS PubMed WorldCat Cooper JA . 1987 . Effects of cytochalasin and phalloidin on actin . The Journal of Cell Biology 105 , 1473 – 1478 . Google Scholar Crossref Search ADS PubMed WorldCat Dhar P , Eck E, Israelachvili JN, Lee DW, Min Y, Ramachandran A, Waring AJ, Zasadzinski JA. 2012 . Lipid–protein interactions alter line tensions and domain size distributions in lung surfactant monolayers . Biophysical Journal 102 , 56 – 65 . Google Scholar Crossref Search ADS PubMed WorldCat Dinic J , Ashrafzadeh P, Parmryd I. 2013 . Actin filaments attachment at the plasma membrane in live cells cause the formation of ordered lipid domains . Biochimica et Biophysica Acta 1828 , 1102 – 1111 . Google Scholar Crossref Search ADS PubMed WorldCat Epand RM . 2008 . Proteins and cholesterol-rich domains . Biochimica et Biophysica Acta 1778 , 1576 – 1582 . Google Scholar Crossref Search ADS PubMed WorldCat Epand RM , Epand RF, Sayer BG, Melacini G, Palgulachari MN, Segrest JP, Anantharamaiah GM. 2004 . An apolipoprotein AI mimetic peptide: membrane interactions and the role of cholesterol . Biochemistry 43 , 5073 – 5083 . Google Scholar Crossref Search ADS PubMed WorldCat Furt F , König S, Bessoule JJ, et al. 2010 . Polyphosphoinositides are enriched in plant membrane rafts and form microdomains in the plasma membrane . Plant Physiology 152 , 2173 – 2187 . Google Scholar Crossref Search ADS PubMed WorldCat Gerbeau-Pissot P , Der C, Thomas D, Anca IA, Grosjean K, Roche Y, Perrier-Cornet JM, Mongrand S, Simon-Plas F. 2014 . Modification of plasma membrane organization in tobacco cells elicited by cryptogein . Plant Physiology 164 , 273 – 286 . Google Scholar Crossref Search ADS PubMed WorldCat Grant CW , Wu SH, McConnell HM. 1974 . Lateral phase separations in binary lipid mixtures: correlation between spin label and freeze–fracture electron microscopic studies . Biochimica et Biophysica Acta 363 , 151 – 158 . Google Scholar Crossref Search ADS PubMed WorldCat Grosjean K , Mongrand S, Beney L, Simon-Plas F, Gerbeau-Pissot P. 2015 . Differential effect of plant lipids on membrane organization: specificities of phytosphingolipids and phytosterols . The Journal of Biological Chemistry 290 , 5810 – 5825 . Google Scholar Crossref Search ADS PubMed WorldCat Halling KK , Slotte JP. 2004 . Membrane properties of plant sterols in phospholipid bilayers as determined by differential scanning calorimetry, resonance energy transfer and detergent-induced solubilization . Biochimica et Biophysica Acta 1664 , 161 – 171 . Google Scholar Crossref Search ADS PubMed WorldCat Hartmann MA . 1998 . Plant sterols and the membrane environment . Trends in Plant Science 3 , 170 – 175 . Google Scholar Crossref Search ADS WorldCat Honig BH , Hubbell WL, Flewelling RF. 1986 . Electrostatic interactions in membranes and proteins . Annual Review of Biophysics and Biophysical Chemistry 15 , 163 – 193 . Google Scholar Crossref Search ADS PubMed WorldCat Hutten SJ , Hamers DS, Aan den Toorn M, van Esse W, Nolles A, Bücherl CA, de Vries SC, Hohlbein J, Borst JW. 2017 . Visualization of BRI1 and SERK3/BAK1 nanoclusters in Arabidopsis roots . PLoS ONE 12 , e0169905 . Google Scholar Crossref Search ADS PubMed WorldCat Ijäs HK , Lönnfors M, Nyholm TK. 2013 . Sterol affinity for phospholipid bilayers is influenced by hydrophobic matching between lipids and transmembrane peptides . Biochimica et Biophysica Acta 1828 , 932 – 937 . Google Scholar Crossref Search ADS PubMed WorldCat Ingólfsson HI , Melo MN, van Eerden FJ, et al. 2014 . Lipid organization of the plasma membrane . Journal of the American Chemical Society 136 , 14554 – 14559 . Google Scholar Crossref Search ADS PubMed WorldCat Jarsch IK , Konrad SS, Stratil TF, Urbanus SL, Szymanski W, Braun P, Braun KH, Ott T. 2014 . Plasma membranes are subcompartmentalized into a plethora of coexisting and diverse microdomains in Arabidopsis and Nicotiana benthamiana . The Plant Cell 26 , 1698 – 1711 . Google Scholar Crossref Search ADS PubMed WorldCat Jin L , Millard AC, Wuskell JP, Clark HA, Loew LM. 2005 . Cholesterol-enriched lipid domains can be visualized by di-4-ANEPPDHQ with linear and nonlinear optics . Biophysical Journal 89 , L04 – L06 . Google Scholar Crossref Search ADS PubMed WorldCat Jin L , Millard AC, Wuskell JP, Dong X, Wu D, Clark HA, Loew LM. 2006 . Characterization and application of a new optical probe for membrane lipid domains . Biophysical Journal 90 , 2563 – 2575 . Google Scholar Crossref Search ADS PubMed WorldCat Kahya N , Brown DA, Schwille P. 2005 . Raft partitioning and dynamic behavior of human placental alkaline phosphatase in giant unilamellar vesicles . Biochemistry 44 , 7479 – 7489 . Google Scholar Crossref Search ADS PubMed WorldCat Kaiser HJ , Lingwood D, Levental I, Sampaio JL, Kalvodova L, Rajendran L, Simons K. 2009 . Order of lipid phases in model and plasma membranes . Proceedings of the National Academy of Science, USA 106 , 16645 – 16650 . Google Scholar Crossref Search ADS WorldCat Kleine-Vehn J , Wabnik K, Martinière A, et al. 2011 . Recycling, clustering, and endocytosis jointly maintain PIN auxin carrier polarity at the plasma membrane . Molecular Systems Biology 7 , 540 . Google Scholar Crossref Search ADS PubMed WorldCat Kusumi A , Nakada C, Ritchie K, Murase K, Suzuki K, Murakoshi H, Kasai RS, Kondo J, Fujiwara T. 2005 . Paradigm shift of the plasma membrane concept from the two-dimensional continuum fluid to the partitioned fluid: high-speed single-molecule tracking of membrane molecules . Annual Review of Biophysics and Biomolecular Structure 34 , 351 – 378 . Google Scholar Crossref Search ADS PubMed WorldCat Laloi M , Perret AM, Chatre L, et al. 2007 . Insights into the role of specific lipids in the formation and delivery of lipid microdomains to the plasma membrane of plant cells . Plant Physiology 143 , 461 – 472 . Google Scholar Crossref Search ADS PubMed WorldCat Lefebvre B , Furt F, Hartmann MA, et al. 2007 . Characterization of lipid rafts from Medicago truncatula root plasma membranes: a proteomic study reveals the presence of a raft-associated redox system . Plant Physiology 144 , 402 – 418 . Google Scholar Crossref Search ADS PubMed WorldCat Lenne PF , Wawrezinieck L, Conchonaud F, Wurtz O, Boned A, Guo XJ, Rigneault H, He HT, Marguet D. 2006 . Dynamic molecular confinement in the plasma membrane by microdomains and the cytoskeleton meshwork . The EMBO Journal 25 , 3245 – 3256 . Google Scholar Crossref Search ADS PubMed WorldCat Lentz BR , Barenholz Y, Thompson TE. 1976 . Fluorescence depolarization studies of phase transitions and fluidity in phospholipid bilayers. 2. Two-component phosphatidylcholine liposomes . Biochemistry 15 , 4529 – 4537 . Google Scholar Crossref Search ADS PubMed WorldCat Levental I , Grzybek M, Simons K. 2011 . Raft domains of variable properties and compositions in plasma membrane vesicles . Proceedings of the National Academy of Science, USA 108 , 11411 – 11416 . Google Scholar Crossref Search ADS WorldCat Levental KR , Levental I. 2015 . Giant plasma membrane vesicles: models for understanding membrane organization . Current Topics in Membranes 75 , 25 – 57 . Google Scholar Crossref Search ADS PubMed WorldCat Li R , Liu P, Wan Y, et al. 2012 . A membrane microdomain-associated protein, Arabidopsis Flot1, is involved in a clathrin-independent endocytic pathway and is required for seedling development . The Plant Cell 24 , 2105 – 2122 . Google Scholar Crossref Search ADS PubMed WorldCat Lingwood D , Ries J, Schwille P, Simons K. 2008 . Plasma membranes are poised for activation of raft phase coalescence at physiological temperature . Proceedings of the National Academy of Science, USA 105 , 10005 – 10010 . Google Scholar Crossref Search ADS WorldCat Lingwood D , Simons K. 2010 . Lipid rafts as a membrane-organizing principle . Science 327 , 46 – 50 . Google Scholar Crossref Search ADS PubMed WorldCat Liu P , Li RL, Zhang L, Wang QL, Niehaus K, Baluska F, Samaj J, Lin JX. 2009 . Lipid microdomain polarization is required for NADPH oxidase-dependent ROS signaling in Picea meyeri pollen tube tip growth . The Plant Journal 60 , 303 – 313 . Google Scholar Crossref Search ADS PubMed WorldCat Lv X , Jing Y, Xiao J, Zhang Y, Zhu Y, Julian R, Lin J. 2017 . Membrane microdomains and the cytoskeleton constrain AtHIR1 dynamics and facilitate the formation of an AtHIR1-associated immune complex . The Plant Journal 90 , 3 – 16 . Google Scholar Crossref Search ADS PubMed WorldCat Mannock DA , Benesch MG, Lewis RN, McElhaney RN. 2015 . A comparative calorimetric and spectroscopic study of the effects of cholesterol and of the plant sterols β-sitosterol and stigmasterol on the thermotropic phase behavior and organization of dipalmitoylphosphatidylcholine bilayer membranes . Biochimica et Biophysica Acta 1848 , 1629 – 1638 . Google Scholar Crossref Search ADS PubMed WorldCat Martinière A , Lavagi I, Nageswaran G, et al. 2012 . Cell wall constrains lateral diffusion of plant plasma-membrane proteins . Proceedings of the National Academy of Science, USA 109 , 12805 – 12810 . Google Scholar Crossref Search ADS WorldCat Martinière A , Runions J. 2013 . Protein diffusion in plant cell plasma membranes: the cell-wall corral . Frontiers in Plant Science 4 , 515 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Men S , Boutté Y, Ikeda Y, Li X, Palme K, Stierhof YD, Hartmann MA, Moritz T, Grebe M. 2008 . Sterol-dependent endocytosis mediates post-cytokinetic acquisition of PIN2 auxin efflux carrier polarity . Nature Cell Biology 10 , 237 – 244 . Google Scholar Crossref Search ADS PubMed WorldCat Minami A , Fujiwara M, Furuto A, Fukao Y, Yamashita T, Kamo M, Kawamura Y, Uemura M. 2009 . Alterations in detergent-resistant plasma membrane microdomains in Arabidopsis thaliana during cold acclimation . Plant & Cell Physiology 50 , 341 – 359 . Google Scholar Crossref Search ADS PubMed WorldCat Mongrand S , Morel J, Laroche J, et al. 2004 . Lipid rafts in higher plant cells: purification and characterization of Triton X-100-insoluble microdomains from tobacco plasma membrane . The Journal of Biological Chemistry 279 , 36277 – 36286 . Google Scholar Crossref Search ADS PubMed WorldCat Morejohn LC , Fosket DE. 1984 . Inhibition of plant microtubule polymerization in vitro by the phosphoric amide herbicide amiprophos-methyl . Science 224 , 874 – 876 . Google Scholar Crossref Search ADS PubMed WorldCat Moscatelli A , Gagliardi A, Maneta-Peyret L, et al. 2015 . Characterisation of detergent-insoluble membranes in pollen tubes of Nicotiana tabacum (L.) . Biology Open 4 , 378 – 399 . Google Scholar Crossref Search ADS PubMed WorldCat Murase K , Fujiwara T, Umemura Y, et al. 2004 . Ultrafine membrane compartments for molecular diffusion as revealed by single molecule techniques . Biophysical Journal 86 , 4075 – 4093 . Google Scholar Crossref Search ADS PubMed WorldCat Nicolson GL . 2014 . The Fluid–Mosaic Model of Membrane Structure: still relevant to understanding the structure, function and dynamics of biological membranes after more than 40 years . Biochimica et Biophysica Acta 1838 , 1451 – 1466 . Google Scholar Crossref Search ADS PubMed WorldCat Noirot E , Der C, Lherminier J, Robert F, Moricova P, Kiêu K, Leborgne-Castel N, Simon-Plas F, Bouhidel K. 2014 . Dynamic changes in the subcellular distribution of the tobacco ROS-producing enzyme RBOHD in response to the oomycete elicitor cryptogein . Journal of Experimental Botany 65 , 5011 – 5022 . Google Scholar Crossref Search ADS PubMed WorldCat Nyholm TK , van Duyl B, Rijkers DT, Liskamp RM, Killian JA. 2011 . Probing the lipid–protein interface using model transmembrane peptides with a covalently linked acyl chain . Biophysical Journal 101 , 1959 – 1967 . Google Scholar Crossref Search ADS PubMed WorldCat Nyström JH , Lönnfors M, Nyholm TK. 2010 . Transmembrane peptides influence the affinity of sterols for phospholipid bilayers . Biophysical Journal 99 , 526 – 533 . Google Scholar Crossref Search ADS PubMed WorldCat Ovecka M , Berson T, Beck M, Derksen J, Samaj J, Baluska F, Lichtscheidl IK. 2010 . Structural sterols are involved in both the initiation and tip growth of root hairs in Arabidopsis thaliana . The Plant Cell 22 , 2999 – 3019 . Google Scholar Crossref Search ADS PubMed WorldCat Pathak P , London E. 2015 . The effect of membrane lipid composition on the formation of lipid ultrananodomains . Biophysical Journal 109 , 1630 – 1638 . Google Scholar Crossref Search ADS PubMed WorldCat Petrásek J , Schwarzerová K. 2009 . Actin and microtubule cytoskeleton interactions . Current Opinion in Plant Biology 12 , 728 – 734 . Google Scholar Crossref Search ADS PubMed WorldCat Phillips MC , Kamat VB, Chapman D. 1970 . The interaction of cholesterol with the sterol free lipids of plasma membranes . Chemistry and Physics of Lipids 4 , 409 – 417 . Google Scholar Crossref Search ADS PubMed WorldCat Pike LJ . 2006 . Rafts defined: a report on the keystone symposium on lipid rafts and cell function . Journal of Lipid Research 47 , 1597 – 1598 . Google Scholar Crossref Search ADS PubMed WorldCat Quinn PJ . 2010 . A lipid matrix model of membrane raft structure . Progress in Lipid Research 49 , 390 – 406 . Google Scholar Crossref Search ADS PubMed WorldCat Raffaele S , Bayer E, Lafarge D, et al. 2009 . Remorin, a Solanaceae protein resident in membrane rafts and plasmodesmata, impairs Potato virus X movement . The Plant Cell 21 , 1541 – 1555 . Google Scholar Crossref Search ADS PubMed WorldCat Ramstedt B , Slotte JP. 2006 . Sphingolipids and the formation of sterol-enriched ordered membrane domains . Biochimica et Biophysica Acta 1758 , 1945 – 1956 . Google Scholar Crossref Search ADS PubMed WorldCat Roche Y , Gerbeau-Pissot P, Buhot B, Thomas D, Bonneau L, Gresti J, Mongrand S, Perrier-Cornet JM, Simon-Plas F. 2008 . Depletion of phytosterols from the plant plasma membrane provides evidence for disruption of lipid rafts . FASEB Journal 22 , 3980 – 3991 . Google Scholar Crossref Search ADS PubMed WorldCat Rujanavech C , Henderson PA, Silbert DF. 1986 . Influence of sterol structure on phospholipid phase behavior as detected by parinaric acid fluorescence spectroscopy . The Journal of Biological Chemistry 261 , 7204 – 7214 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Samson F , Donoso JA, Heller-Bettinger I, Watson D, Himes RH. 1979 . Nocodazole action on tubulin assembly, axonal ultrastructure and fast axoplasmic transport . The Journal of Pharmacology and Experimental Therapeutics 208 , 411 – 417 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Sandor R , Der C, Grosjean K, Anca I, Noirot E, Leborgne-Castel N, Lochman J, Simon-Plas F, Gerbeau-Pissot P. 2016 . Plasma membrane order and fluidity are diversely triggered by elicitors of plant defence . Journal of Experimental Botany 67 , 5173 – 5185 . Google Scholar Crossref Search ADS PubMed WorldCat Schneider F , Waithe D, Clausen MP, Galiani S, Koller T, Ozhan G, Eggeling C, Sezgin E. 2017 . Diffusion of lipids and GPI-anchored proteins in actin-free plasma membrane vesicles measured by STED-FCS . Molecular Biology of the Cell 28 , 1507 – 1518 . Google Scholar Crossref Search ADS PubMed WorldCat Schuler I , Duportail G, Glasser N, Benveniste P, Hartmann MA. 1990 . Soybean phosphatidylcholine vesicles containing plant sterols: a fluorescence anisotropy study . Biochimica et Biophysica Acta 1028 , 82 – 88 . Google Scholar Crossref Search ADS PubMed WorldCat Schuler I , Milon A, Nakatani Y, Ourisson G, Albrecht AM, Benveniste P, Hartman MA. 1991 . Differential effects of plant sterols on water permeability and on acyl chain ordering of soybean phosphatidylcholine bilayers . Proceedings of the National Academy of Science, USA 88 , 6926 – 6930 . Google Scholar Crossref Search ADS WorldCat Scott RE . 1976 . Plasma membrane vesiculation: a new technique for isolation of plasma membranes . Science 194 , 743 – 745 . Google Scholar Crossref Search ADS PubMed WorldCat Sena F , Sotelo-Silveira M, Astrada S, Botella MA, Malacrida L, Borsani O. 2017 . Spectral phasor analysis reveals altered membrane order and function of root hair cells in Arabidopsis dry2/sqe1-5 drought hypersensitive mutant . Plant Physiology and Biochemistry 119 , 224 – 231 . Google Scholar Crossref Search ADS PubMed WorldCat Shaghaghi M , Chen MT, Hsueh YW, Zuckermann MJ, Thewalt JL. 2016 . Effect of Sterol structure on the physical properties of 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine membranes determined using 2H nuclear magnetic resonance . Langmuir 32 , 7654 – 7663 . Google Scholar Crossref Search ADS PubMed WorldCat Shahedi V , Orädd G, Lindblom G. 2006 . Domain-formation in DOPC/SM bilayers studied by pfg-NMR: effect of sterol structure . Biophysical Journal 91 , 2501 – 2507 . Google Scholar Crossref Search ADS PubMed WorldCat Shi Y , An L, Zhang M, Huang C, Zhang H, Xu S. 2008 . Regulation of the plasma membrane during exposure to low temperatures in suspension-cultured cells from a cryophyte (Chorispora bungeana) . Protoplasma 232 , 173 – 181 . Google Scholar Crossref Search ADS PubMed WorldCat Shimshick EJ , McConnell HM. 1973 . Lateral phase separation in phospholipid membranes . Biochemistry 12 , 2351 – 2360 . Google Scholar Crossref Search ADS PubMed WorldCat Simons K , Gerl MJ. 2010 . Revitalizing membrane rafts: new tools and insights . Nature Reviews. Molecular Cell Biology 11 , 688 – 699 . Google Scholar Crossref Search ADS PubMed WorldCat Simons K , Ikonen E. 1997 . Functional rafts in cell membranes . Nature 387 , 569 – 572 . Google Scholar Crossref Search ADS PubMed WorldCat Simons K , Sampaio JL. 2011 . Membrane organization and lipid rafts . Cold Spring Harbor Perspectives in Biology 3 , a004697 . Google Scholar Crossref Search ADS PubMed WorldCat Singer SJ , Nicolson GL. 1972 . The fluid mosaic model of the structure of cell membranes . Science 175 , 720 – 731 . Google Scholar Crossref Search ADS PubMed WorldCat Spector I , Shochet NR, Kashman Y, Groweiss A. 1983 . Latrunculins: novel marine toxins that disrupt microfilament organization in cultured cells . Science 219 , 493 – 495 . Google Scholar Crossref Search ADS PubMed WorldCat Stanislas T , Bouyssie D, Rossignol M, Vesa S, Fromentin J, Morel J, Pichereaux C, Monsarrat B, Simon-Plas F. 2009 . Quantitative proteomics reveals a dynamic association of proteins to detergent-resistant membranes upon elicitor signaling in tobacco . Molecular & Cellular Proteomics 8 , 2186 – 2198 . Google Scholar Crossref Search ADS WorldCat Suzuki K , Shinshi H. 1995 . Transient activation and tyrosine phosphorylation of a protein kinase in tobacco cells treated with a fungal elicitor . The Plant Cell 7 , 639 – 647 . Google Scholar Crossref Search ADS PubMed WorldCat Szymanski WG , Zauber H, Erban A, Gorka M, Wu XN, Schulze WX. 2015 . Cytoskeletal components define protein location to membrane microdomains . Molecular & Cellular Proteomics 14 , 2493 – 2509 . Google Scholar Crossref Search ADS WorldCat Takahashi T , Nomura F, Yokoyama Y, Tanaka-Takiguchi Y, Homma M, Takiguchi K. 2013 . Multiple membrane interactions and versatile vesicle deformations elicited by melittin . Toxins 5 , 637 – 664 . Google Scholar Crossref Search ADS PubMed WorldCat Titapiwatanakun B , Murphy AS. 2009 . Post-transcriptional regulation of auxin transport proteins: cellular trafficking, protein phosphorylation, protein maturation, ubiquitination, and membrane composition . Journal of Experimental Botany 60 , 1093 – 1107 . Google Scholar Crossref Search ADS PubMed WorldCat Umemura YM , Vrljic M, Nishimura SY, Fujiwara TK, Suzuki KG, Kusumi A. 2008 . Both MHC class II and its GPI-anchored form undergo hop diffusion as observed by single-molecule tracking . Biophysical Journal 95 , 435 – 450 . Google Scholar Crossref Search ADS PubMed WorldCat Urbanus SL , Ott T. 2012 . Plasticity of plasma membrane compartmentalization during plant immune responses . Frontiers in Plant Science 3 , 181 . Google Scholar Crossref Search ADS PubMed WorldCat van den Bogaart G , Meyenberg K, Risselada HJ, et al. 2011 . Membrane protein sequestering by ionic protein–lipid interactions . Nature 479 , 552 – 555 . Google Scholar Crossref Search ADS PubMed WorldCat Vereb G , Matko J, Vamosi G, et al. 2000 . Cholesterol-dependent clustering of IL-2Rα and its colocalization with HLA and CD48 on T lymphoma cells suggest their functional association with lipid rafts . Proceedings of the National Academy of Science, USA 97 , 6013 – 6018 . Google Scholar Crossref Search ADS WorldCat Willemsen V , Friml J, Grebe M, van den Toorn A, Palme K, Scheres B. 2003 . Cell polarity and PIN protein positioning in Arabidopsis require STEROL METHYLTRANSFERASE1 function . The Plant Cell 15 , 612 – 625 . Google Scholar Crossref Search ADS PubMed WorldCat Zaban B , Maisch J, Nick P. 2013 . Dynamic actin controls polarity induction de novo in protoplasts . Journal of Integrative Plant Biology 55 , 142 – 159 . Google Scholar Crossref Search ADS PubMed WorldCat Zárský V , Cvrcková F, Potocký M, Hála M. 2009 . Exocytosis and cell polarity in plants – exocyst and recycling domains . New Phytologist 183 , 255 – 272 . Google Scholar Crossref Search ADS PubMed WorldCat Author notes These authors contributed equally to this work. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.

Ernesto Bianchetti, Ricardo; Silvestre Lira, Bruno; Santos Monteiro, Scarlet; Demarco, Diego; Purgatto, Eduardo; Rothan, Christophe; Rossi, Magdalena; Freschi, Luciano

doi: 10.1093/jxb/ery145pmid: 29912373

Abstract Light signaling has long been reported to influence fruit biology, although the regulatory impact of fruit-localized photoreceptors on fruit development and metabolism remains unclear. Studies performed in phytochrome (PHY)-deficient tomato (Solanum lycopersicum) mutants suggest that SlPHYA, SlPHYB2, and to a lesser extent SlPHYB1 influence fruit development and ripening. By employing fruit-specific RNAi-mediated silencing of SlPHY genes, we demonstrated that fruit-localized SlPHYA and SlPHYB2 play contrasting roles in regulating plastid biogenesis and maturation in tomato. Our data revealed that fruit-localized SlPHYA, rather than SlPHYB1 or SlPHYB2, positively influences tomato plastid differentiation and division machinery via changes in both light and cytokinin signaling-related gene expression. Fruit-localized SlPHYA and SlPHYB2 were also shown to modulate sugar metabolism in early developing fruits via overlapping, yet distinct, mechanisms involving the co-ordinated transcriptional regulation of genes related to sink strength and starch biosynthesis. Fruit-specific SlPHY silencing also drastically altered the transcriptional profile of genes encoding light-repressor proteins and carotenoid-biosynthesis regulators, leading to reduced carotenoid biosynthesis during fruit ripening. Together, our data reveal the existence of an intricate PHY–hormonal interplay during fruit development and ripening, and provide conclusive evidence on the regulation of tomato quality by fruit-localized phytochromes. Auxin, carotenoid, cytokinin, fleshy fruit, phytochrome, plastid division, tomato, Solanum lycopersicum, starch Introduction Fleshy fruit growth, maturation, and ripening are under strict developmental, hormonal, and epigenetic regulation, which in turn are fine-tuned by a plethora of environmental stimuli (Kumar et al., 2014; Giovannoni et al., 2017). Among environmental cues, light plays a significant role in determining fruit growth, pigmentation, and timing of ripening (Carvalho et al., 2011; Gupta et al., 2014; Llorente et al., 2016a). In tomato (Solanum lycopersicum), a major crop and important model species for fleshy fruits, several lines of evidence indicate that changes in light perception and signaling can lead to significant alterations in fruit development and quality traits (Giliberto et al., 2005; Schofield and Paliyath, 2005; Azari et al., 2010b; Bianchetti et al., 2017). One of the earliest pieces of evidence of the influence of light on tomato fruit biology dates back to 1954, when fruit pigmentation was shown to be regulated by red/far red (R/FR) light in a reversible manner (Piringer and Heinze, 1954). First isolated only a few years later, phytochromes (PHYs) act as molecular switches in response to R and FR light, existing as homodimers of two independently reversible subunits. Once activated by R light, PHYs are transported from the cytosol to the nucleus, where they counteract light-signaling repressor proteins, such as CONSTITUTIVE PHOTOMORPHOGENESIS1 (COP1), CULLIN4 (CUL4), DNA DAMAGE-BINDING PROTEIN 1 (DDB1), DETIOLATED1 (DET1), and PHYTOCHROME INTERACTION FACTOR (PIF) (Deng and Quail, 1992; Pepper et al., 1994; Schroeder et al., 2002; Duek and Fankhauser, 2005; Thomann et al., 2005). In line with their role as repressors of photomorphogenic responses, either the down-regulation or loss-of-function of tomato genes encoding COP1, CUL4, DDB1, DET1, and PIF1a profoundly alter tomato fruit physiology and nutritional composition (Cookson et al., 2003; Liu et al., 2004; Davuluri et al., 2005; Kolotilin et al., 2007; Wang et al., 2008; Azari et al., 2010b; Enfissi et al., 2010; Llorente et al., 2016b). In tomato, five PHY-encoding genes have been identified, namely SlPHYA, SlPHYB1, SlPHYB2, SlPHYE, and SlPHYF (Alba et al., 2000b). The paralogous SlPHYB1 and SlPHYB2, which originated during the Solanum whole-genome triplication event (Tomato Genome Consortium, 2012), display distinct expression profiles within tomato organs, pointing to functional diversification (Hauser et al., 1997; Weller et al., 2000). SlPHYB1 is more prominently expressed in vegetative tissues, whereas the highest SlPHYB2 expression levels are detected in fruits (Hauser et al., 1997; Bianchetti et al., 2017). Moreover, evidence also suggests a more direct involvement of SlPHYB1, rather than SlPHYB2, during early seedling photomorphogenic responses (van Tuinen et al., 1995a, 1995b; Weller et al., 2000). Very little is known about the influence of SlPHYE and SlPHYF on tomato vegetative and reproductive development (Schrager-Lavelle et al., 2016). Attempts to define the influence of fruit-localized PHYs on fruit development and ripening have been relatively limited. Brief R-light treatments of detached mature-green tomato fruits promote lycopene accumulation, a response reversed by subsequent treatment with FR light (Alba et al., 2000a), which is consistent with the hypothesis that fruit-localized PHYs play a regulatory role in controlling tomato fruit carotenogenesis. The marked accumulation of SlPHYA transcripts during fruit ripening (Alba et al., 2000a) associated with the reduced fruit lycopene levels observed in phyA tomato mutants (Gupta et al., 2014) raise the possibility that this PHY may be an important regulator of tomato fruit carotenoid biosynthesis. However, regardless of the development stage or tissue considered, SlPHYB2 is the most highly expressed PHY in tomato fruits (Bianchetti et al., 2017). Moreover, the phyB2 mutant also displays considerable changes in the fruit carotenoid profile (Gupta et al., 2014), suggesting that multiple PHYs are involved in regulating this metabolic process. Besides carotenogenesis, PHYs have also been found to control other aspects of tomato fruit development and metabolism, including chloroplast biogenesis, chlorophyll accumulation, sugar metabolism, sink activity, and hormonal signaling (Gupta et al., 2014; Bianchetti et al., 2017). However, as the existing evidence supporting these findings is exclusively based on studies performed in phy mutants, whether these responses are dependent on fruit-localized PHYs or are merely consequences of the collateral negative effects of PHY deficiency on vegetative plant growth remains to be elucidated. By employing fruit-specific RNAi-mediated silencing of SlPHY genes, we shed light on the functional specificity of fruit-localized SlPHYs in controlling developmental and metabolic processes associated with sugar and carotenoid accumulation, two essential nutritional quality traits of this edible fruit. Our data also reveal that an intricate light–hormonal signaling network involving key components of both auxin and cytokinin signal transduction pathways is implicated in the PHY-dependent regulation of fruit plastid biogenesis, sugar metabolism, and carotenoid accumulation. Materials and methods Plant material and growth conditions Tomato (Solanum lycopersicum L.) plants cv. Micro-Tom, which harbors the wild-type SlGLK2 allele (Carvalho et al., 2011), were grown under controlled conditions of 250 µmol m−2 s−1, a 12-h photoperiod, and air temperature of 27/22 °C day/ night. The fruit stages examined were immature green, mature green, breaker, and red ripe, which were harvested on average at 8, 25, 32, and 44 d post-anthesis. All fruits were harvested at the same time of the day with four biological replicates (each replicate was composed of a pool of at least five fruits from different plants). Columella, placenta, and seeds were immediately removed, and the remaining tissues were frozen in liquid nitrogen and stored at –80 °C until use. Generation of transgenic tomato plants Three fragments specific to the coding sequences of SlPHYA, SlPHYB2, and both SlPHYB1 and SlPHYB2 were selected using BLAST queries against the Sol Genomics Network database (https://solgenomics.net/, ITAG release 2.40) and the web-based computational tool pssRNAit (Dai and Zhao, 2011) was employed to avoid off-target silencing. Each fragment was independently cloned into pENTR D-TOPO plasmids (Invitrogen) using the primers listed in Supplementary Table S1 at JXB online. Subsequently, each fragment was recombined into the plant transformation vector pK8GWIWG (Fernandez et al., 2009). Transgenic Micro-Tom plants were generated by Agrobacterium-mediated transformation according to Pino et al. (2010), with minor changes: cotyledons from 5-d-old seedlings were used for the transformation, and the zeatin and kanamycin concentration were 5 µM and 70 mg l−1, respectively. All plants used in the study were from the T2 generation. Fruit color and pigment quantification Changes in fruit color (Hue angle) were determined using a Konica Minolta CR-400 colorimeter as described in Su et al. (2015). Chlorophyll extraction and quantification were carried out as described in Lira et al. (2016) with some modifications. Pericarp samples were weighed (typically 100 mg fresh weight, FW), ground in liquid nitrogen, immersed in a 10× excess volume of N, N-dimethylformamide, and incubated at room temperature for 24 h in absolute darkness and constant agitation (200 rpm). After centrifugation (9000 g, 5 min, 4 °C), the supernatant absorbance was recorded at 647 and 664 nm, and the total chlorophyll content was estimated using the equations given by Porra et al. (1989). For carotenoid extraction, approximately 200 mg FW of pericarp samples were ground in liquid nitrogen and sequentially homogenized with a solution of 100 µl of saturated NaCl, then 200 µl of dichloromethane, and finally 1 ml of hexane:diethyl ether (1:1, v/v). The supernatant was collected after centrifugation (5000 g, 10 min, 4 °C). The remaining carotenoids in the pellet were extracted three more times with 500 µl of hexane:diethyl ether (1:1, v/v). All supernatant fractions were combined, completely vacuum-dried, and suspended with 200 µl of acetonitrile. Lycopene, β-carotene, lutein, and neurosporene levels were determined by high-performance liquid chromatography (HPLC) with a photodiode array detector (PDA) as described by Lira et al. (2017). Starch and soluble sugar quantification Starch and soluble sugar extractions were performed as described in Bianchetti et al. (2017). Briefly, approximately 200 mg FW of pericarp samples was extracted with 1 ml of 80% (v/v) methanol for 10 min at 80 °C followed by the collection of the supernatants by centrifugation (13000 g, 10 min, 4 °C). The remaining pellets were re-extracted five times, and all supernatants were combined, completely vacuum-dried, and suspended in 200 µl distilled water. Soluble sugars (i.e. sucrose, fructose, and glucose) were measured using a HPLC system equipped with an amperometric detector (Dionex, Sunnyvale, USA) and a CarboPac PA1 (4 × 250 mm) column (Purgatto et al., 2002). Starch levels were determined from dried pellet as described in Suguiyama et al. (2014). Antioxidant capacity and total phenolics Hydrophilic and lipophilic Trolox equivalent antioxidant capacities (TEACs) were spectrophotometrically determined as described in Lira et al. (2016). Total phenolic content was determined in hydrophilic extracts by using the Folin–Ciocalteu method (Singleton and Rossi, 1965). Plastid ultrastructure and abundance Pericarp fragments taken from the pedicel region (green shoulder) of immature fruits were fixed at 4 °C in 2.5% (v/v) glutaraldehyde and 2% (v/v) paraformaldehyde in 0.1 M sodium phosphate buffer (pH 7.2). Subsequently, the samples were post-fixed in 1% osmium tetroxide in 0.1 M sodium phosphate buffer (pH 7.2), dehydrated in a graded acetone series, and embedded in Spurr’s resin. Ultrathin sections were stained with saturated uranyl acetate and lead citrate (Melo et al., 2016) and observed using a JEOL JEM1011 transmission electron microscope. Sections from three immature fruits picked from different plants were analysed per genotype. Plastid abundance was determined as described in Bianchetti et al. (2017). Briefly, small pieces (1 × 1 mm) of pericarp were fixed in 3.5% (v/v) glutaraldehyde for 1 h. Samples were washed twice and transferred to 0.1 M NaEDTA pH 9.5 solution for 4 h at 60 °C in complete darkness. Pieces were softly disrupted and transferred to microscope slides. Isolated cells were visualized using a Leica microscope. Plastid densities in individual cells were estimated using the ImageJ program (https://imagej.nih.gov/ij/). At least 40 individual cells were analysed per sample. Transcriptional profile Total RNA extraction, cDNA synthesis, primer design, and qPCR assays were performed as described by Quadrana et al. (2013). Primer sequences used are detailed in Supplementary Table S1. Quantitative real-time (qRT-)PCR reactions were performed in a StepOnePlus PCR Real-Time thermocycler (Applied Biosystems) in a final volume of 10 µl using 2× SYBR Green Master Mix reagent (Thermo Fisher Scientific). Melting curves were checked for unspecific amplifications and primer dimerization. Absolute fluorescence data were analysed using the LinRegPCR software package (Ruijter et al., 2009) to obtain quantitation cycle (Cq) values and to calculate primer efficiency. Transcript abundances were normalized against the geometric mean of two reference genes, CAC and EXPRESSED (Expósito-Rodriguez et al., 2008). Gene promoter analysis Gene promoter analysis was performed using the promotor sequences available at the Sol Genomics Network. Typically, 3 kb upstream of the initial ATG codon of each sequence was analysed using the PlantPAN 2.0 platform (http://plantpan2.itps.ncku.edu.tw/) (Chow et al., 2016) for the presence of PBE-box (CACATG), G-box (CACGTG), CA-hybrid (GACGTA), CG-hybrid (GACGTG), canonical AuxRE (TGTGTC), and degenerate AuxRE (TGTGNC) motifs (Martı́nez-Garcı́a et al., 2000; Song et al., 2008; Chaabouni et al., 2009). Statistical analysis ANOVA and Student’s t-test were performed using the JMP statistical software package (14th edition; http://jmp.com). Comparisons with P<0.05 were considered statistically significant. Data from wild-type and all independent transgenic lines were also compared with principal component analysis (PCA) using the InfoStat software (http://infostat.com.ar). Results Fruit-specific PHY knockdown in transgenic tomato plants To investigate the role played by distinct PHYs in tomato fruit development and ripening, we generated fruit-specific silenced tomato plants with reduced mRNA levels of SlPHYA, SlPHYB2, or both SlPHYB1 and SlPHYB2. This was achieved using a hairpin-mediated RNAi approach based on the expression of specific fragment sequences of these genes under the control of the fruit-specific PPC2 promoter (Fernandez et al., 2009). The transgenic plants obtained, hereafter designated as SlPHYARNAi, SlPHYB2RNAi, and SlPHYB1/B2RNAi (Fig. 1A), were generated in a Micro-Tom background homozygous for the wild-type GOLDEN2-LIKE-2 (SlGLK2) allele (Carvalho et al., 2011), which encodes a transcription factor critically important for chloroplast development in tomato fruits (Powell et al., 2012). Fig. 1. Open in new tabDownload slide Fruit-specific PHY knockdown in transgenic tomato plants. (A) Constructs designed for generation of the SlPHYARNAi, SlPHYB1/B2RNAi, and SlPHYB2RNAi transgenic lines. ‘A’ indicates the SlPHYA-specific fragment of the mRNA 5′ untranslated region (UTR). ‘B1/B2’ indicates the SlPHYB1/B2-specific fragment of the mRNA 5′ UTR. ‘B2’ indicates the SlPHYB2-specific fragment of the mRNA 5′ UTR. (B) Relative SlPHY mRNA levels in leaves, and immature green (IG), mature green (MG), and breaker (Bk) stages of fruits of the SlPHYARNAi, SlPHYB2RNAi, and SlPHYB1/B2RNAi lines. The first and second fully expanded leaves from the top of 2-month-old plants were harvested. Transcript abundance was normalized against the wild-type (WT) sample. Statistically significant differences compared with the WT genotype were determined using Student’s t-test: *P<0.05. Data are means (±SE) of at least three biological replicates. (This figure is available in color at JXB online.) Transcript abundance analysis revealed that SlPHYA, SlPHYB2, and both SlPHYB1 and SlPHYB2 were down-regulated in the SlPHYARNAi, SlPHYB2RNAi, and SlPHYB1/B2RNAi lines, respectively (Fig. 1B). A search for potential tomato off-targets via BLAST queries against the Sol Genomics Network database or via the public web-based computational tool pssRNAit (Dai and Zhao, 2011) failed to identify regions in the tomato coding that exhibited the 21-nucleotide perfect identity threshold reported to cause off-target silencing (Xu et al., 2006). The percentage of identity of the silencing fragments was below 60% with non-target tomato PHY genes (Supplementary Table S2). Moreover, the length of stretches with perfect identity between the RNAi fragments and non-target tomato PHY genes was ≤15 nucleotides (Supplementary Table S2). In line with this, no off-target SlPHY silencing was detected in the transgenic lines generated (Supplementary Fig. S1). In all the transgenic lines, PHY knockdown was restricted to the fruit tissues as no significant PHY silencing was observed in leaf samples (Fig. 1B). Transgenic lines exhibited normal plant growth and visual phenotypic features similar to those found in wild-type (WT) plants (Supplementary Fig. S2). Overall, fruit-specific PHY knockdown caused no marked changes in fruit size and ripening progression (Supplementary Fig. S3). Fruit-localized SlPHYA and SlPHYB2 differentially impact chloroplast biogenesis and differentiation during early fruit development The PHY-dependent regulation of chloroplast development has been extensively reported in leaf tissues of several species (Stephenson et al., 2009; Inagaki et al., 2015). Moreover, some recent reports have also indicated altered chlorophyll accumulation and chloroplast biogenesis in immature fruits of PHY-deficient tomato mutants (Gupta et al., 2014; Bianchetti et al., 2017). Compared to the WT, fruit-specific SlPHYA and SlPHYB2 knockdown reduced and increased the chlorophyll content in immature fruits, respectively (Fig. 2A). However, chlorophyll levels in immature fruits from SlPHYB1/B2RNAi plants were similar to WT counterparts. Fig. 2. Open in new tabDownload slide Fruit-localized SlPHYA and SlPHYB2 differentially impact on chloroplast biogenesis and differentiation during early fruit development. (A) Total chlorophyll content in immature fruits. (B) Plastid abundance per pericarp cell of immature fruits. (C) Relative mRNA levels of GOLDEN2-LIKE-2 (SlGLK2) normalized against the wild-type (WT) sample. Statistically significant differences compared with the WT sample were determined using Student’s t-test: *P<0.05. Chlorophyll content and transcript abundance data are means (±SE) of at least three biological replicates. For plastid density, three fruits of each genotype were randomly picked, and two technical replicates were taken at the pedicel region of each fruit. Plastid density was determined in at least 40 individual cells per sample. (D) Representative TEM images of plastids in the pedicel region of immature fruits. Arrows indicate plastoglobuli. G, granal thylakoid. Microscopy analysis of pericarp cells revealed that the reduced chlorophyll content detected in SlPHYARNAi immature fruits was associated with a reduction of up to 40% in the number of chloroplasts per pericarp cell compared to WT fruits (Fig. 2B). However, the higher chlorophyll content observed in SlPHYB2RNAi immature fruits was not accompanied by changes in plastid abundance but instead was linked to the up-regulation of the master regulator of chloroplast development and maintenance, SlGLK2 (Fig. 2C). SlPHYB1/B2 knockdown lines showed an intermediate impact on fruit chlorophyll content, plastid density, and SlGLK2 mRNA levels, exhibiting unaltered chlorophyll levels and chloroplast abundance in pericarp cells and slightly higher expression of SlGLK2 compared to the WT (Fig. 2). Plastids of WT, SlPHYB2RNAi, and SlPHYB1/B2RNAi immature fruits exhibited remarkably similar internal membranous structures, displaying well-developed grana and stroma thylakoids as well as numerous plastoglobuli (Fig. 2D, Supplementary Fig. S4). In contrast, fruit-specific SlPHYA knockdown resulted in the formation of chloroplasts with highly reduced grana, suggesting a promotive role of PHYA-mediated light perception on fruit plastid granal development. Plastoglobuli and starch grains were observed equally in fruit chloroplasts of the WT and all transgenic lines. As neither SlPHYB2 nor the SlPHYB1/B2 knockdown altered chloroplast density per cell or plastid ultrastructure (Fig. 2), fruit-localized SlPHYA seems to play a preponderant role in controlling chloroplast biogenesis and differentiation in early developing fruits. Transcript abundance analysis revealed that the reduced plastid abundance observed in SlPHYA-silenced fruits was most probably explained by a drastic reduction in mRNA levels of genes encoding key components of the plastid division machinery, such as FILAMENTOUS TEMPERATURE SENSITIVE-Z (FtsZs), ACCUMULATION AND REPLICATION OF CHLOROPLASTS (ARCs), and PLASTID DIVISION 2 (PDV2), compared to the WT genotype (Fig. 3A). Fig. 3. Open in new tabDownload slide SlPHYA-mediated regulation of chloroplast division machinery is associated with changes in the transcript abundance of light- and cytokinin-signaling genes. (A) Relative mRNA levels of genes encoding components of the plastid division machinery in immature fruits. (B) Relative mRNA levels of type-A TOMATO RESPONSE REGULATOR (TRR) genes in immature fruits. (C) Relative mRNA levels of CYTOKININ RESPONSE FACTOR (SlCRF) genes in immature fruits. (D) Relative mRNA levels of genes encoding light-signaling repressor proteins. Data are means (±SE) of at least three biological replicates. Transcript abundance was normalized against the wild-type (WT) sample. Statistically significant differences compared with the WT were determined using Student’s t-test: *P<0.05). FtsZ, filamentous temperature sensitive-Z; ARC, accumulation and replication of chloroplasts; PDV2, plastid division 2; COP1, constitutive photomorphogenic 1; CUL4, cullin 4; DDB1, UV-damaged DNA binding protein 1; DET1, de-etiolated1. Given the key role played by cytokinins in regulating plastid division and maturation in plants and the widely reported crosstalk between this hormonal class and PHY signaling (Okazaki et al., 2009; Cortleven and Schmülling, 2015), a transcriptional profiling of type-A TOMATO RESPONSE REGULATOR (TRR) was performed. Four out of the five type-A TRRs analysed were significantly down-regulated in immature fruits of SlPHYARNAi compared to the WT genotype (Fig. 3B). Moreover, among the five CYTOKININ RESPONSE FACTOR genes most highly expressed in tomato fruit tissues (Shi et al., 2012), SlCRF1, SlCRF2, and SlCRF5 were markedly down-regulated in SlPHYARNAi lines, whereas SlCRF3 and SlCRF9 mRNA levels remained unchanged (Fig. 3C). As AtCRF2 is responsible for inducing AtPDV2, subsequently increasing plastid division rates in Arabidopsis (Okazaki et al., 2009), the drastic down-regulation of both SlCRF2 and SlPDV2 in SlPHYA-silenced fruits suggests that a similar regulatory mechanism also takes place early in the development of tomato fruits. Alongside the down-regulation of cytokinin signaling genes, fruit-specific SlPHYA-silencing resulted in the up-regulation of tomato genes encoding light-signaling repressor proteins such as COP1, CUL4, DDB1, and DET1 (Fig. 3D), which are negative regulators of plastid division and maturation in tomato and other species (Chory and Peto, 1990; Kolotilin et al., 2007; Wang et al., 2008; Azari et al., 2010b). Collectively, these data suggest that fruit-localized PHYA positively influences tomato plastid division machinery via changes in the transcript abundance of both light- and cytokinin-signaling genes, whereas PHYB2 negatively regulates chlorophyll accumulation by controlling the expression of the master transcription factor of chloroplast development and maintenance, SlGLK2. Fruit-localized PHYs regulate starch metabolism during early fruit development Fruit-specific SlPHYA and SlPHYB2 knockdown promoted starch accumulation during early fruit development (Fig. 4A). In both the WT and transgenic lines, the highest starch content was observed in immature green (IG) fruits, followed by slightly more reduced levels at the mature green (MG) stage, and undetectable levels from the breaker (Bk) stage onwards (Supplementary Fig. S5). Fig. 4. Open in new tabDownload slide Fruit-localized phytochromes regulate sugar metabolism during early fruit development. (A) Schematic representation of the major steps of starch biosynthesis and graphs showing starch content and transcript abundance of starch biosynthesis-related genes in immature fruits. (B) Soluble sugar contents in immature fruits. (C) Summed values of the three soluble sugars analysed (i.e. sucrose + glucose + fructose). (D) Relative mRNA levels of tomato genes encoding invertases (SlLIN) in immature fruits. (E) Relative mRNA levels of AUXIN RESPONSE FACTOR 4 (SlARF4) in immature fruits. For simplicity, the mean of the three values for the transgenic lines is shown. Values for each transgenic line are presented in Supplementary Figs S5, S6. Data are means (±SE) of at least three biological replicates. Statistically significant differences compared with the wild-type (WT) sample were determined using Student’s t-test: *P<0.05. IG, immature green; MG, mature green; Bk, breaker; RR, red ripe; Glc-1-P, glucose 1-phosphate; ADPG, adenosine diphosphate glucose; AGPase, ADP-glucose pyrophosphorylase; STS, starch synthase; SBE, starch branching enzyme. Compared to the WT genotype, marked differences in the transcript profiles of starch biosynthesis genes were observed in both SlPHYA- and SlPHYB2-silenced fruits (Fig. 4A, Supplementary Fig. S6). Catalysing the first committed step in starch biosynthesis, ADP-glucose pyrophosphorylase (AGPase) is a heterotetramer comprising a pair of small/catalytic and a pair of large/regulatory subunits (Kim et al., 2007; Figueroa et al., 2013). Among the tomato genes encoding the large AGPase subunits, both SlAGPaseL1 and SlAGPaseL3 were up-regulated whereas SlAGPaseL2 mRNA levels remained unchanged in immature fruits of SlPHYARNAi plants. It is worth mentioning that SlAGPaseL1 was the large AGPase subunit most expressed in immature tomato fruits (Supplementary Table S3; Petreikov et al., 2006); therefore, the 3-fold increment in its mRNA levels correlates well with the higher starch levels and reduced soluble sugar levels detected in SlPHYARNAi immature fruits compared to the WT counterparts (Fig. 4, Supplementary Figs S5, S6). SlAGPaseS1, which encodes the small/catalytic AGPase subunit, was consistently down-regulated throughout fruit development and ripening in both the SlPHYARNAi and SlPHYB2RNAi lines. However, despite the negative impact of either SlPHYA- or SlPHYB2-silencing on SlAGPaseS1 expression, this gene exhibited higher expression levels than those encoding AGPase large subunits (Supplementary Table S3), suggesting that the catalytic AGPase subunit was not limiting for starch biosynthesis in tomato fruits. In both SlPHYARNAi and SlPHYB2RNAi immature fruits, the starch synthase (STS)-encoding genes SlSTS1 and SlSTS2 were markedly up-regulated compared to WT fruits, whereas SlSTS3 was slightly down-regulated. For SlSTS6, higher transcript accumulation was observed in SlPHYARNAi than in the WT throughout fruit development and ripening (i.e. IG to RR stage) (Supplementary Fig. S6). Finally, distinct expression patterns were observed for the starch branching enzyme (SBE)-encoding genes, as SlSBE1 was up-regulated in all the transgenic lines from MG to Bk stage whereas SlSBE2 was down-regulated in both SlPHYARNAi and SlPHYB2RNAi from IG to RR stage (Supplementary Fig. S6). The increased accumulation of starch in SlPHYARNAi fruits correlated well with higher mRNA levels of SlLIN5 and SlLIN6 (Fig. 4D), which encode cell-wall invertases critically important for sink activity in tomato (Fridman and Zamir, 2003; Kocal et al., 2008). By applying an unsupervised method (i.e. principal component analysis, PCA) to search for patterns in the expression profiles of genes related to sink- and starch-biosynthesis, we demonstrated a clear separation of the WT, SlPHYARNAi, and SlPHYB2RNAi groups (Supplementary Fig. S7). Previous findings have indicated that AUXIN RESPONSE FACTOR4 (SlARF4) is a major negative regulator of starch biosynthesis in early developing tomato fruits (Sagar et al., 2013; Bianchetti et al., 2017). Recent evidence also indicates that SlARF4 plays a repressor role in controlling the transcript abundance of sink-related genes, including SlLIN5 and SlLIN6 (Bianchetti et al., 2017). In accordance with this, fruit-specific SlPHYA and SlPHYB2 knockdown drastically reduced SlARF4 mRNA abundance in early developing tomato fruits (Fig. 4E). Although the direct transcriptional regulation of tomato AGPase, STS, and SBE genes by transcription factors associated with auxin- or light-signaling remains to be determined, the presence of PBE-box, G-box, CA-hybrid, and/or CG-hybrid motifs (Martı́nez-Garcı́a et al., 2000; Song et al., 2008) as well as canonical and/or degenerated ARF-binding Auxin Response Element (AuxRE) motifs within the 3-kb promoter sequence of these genes (Supplementary Fig. S8) is consistent with the hypothesis that light- and/or auxin-related transcription factors might directly control the expression of starch biosynthesis-related genes. Similarly, PIF, HY5, and/or ARF-binding motifs have also been identified within the promoter sequences of SlLIN5 and SlLIN6 genes (Bianchetti et al., 2017). PHY-dependent regulation of fruit carotenoid biosynthesis is associated with transcriptional changes in light- and auxin-signaling genes The very well-characterized PHY-mediated signaling networks controlling carotenogenesis in vegetative tissues (Toledo-Ortiz et al., 2010) contrasts with the considerably more limited information regarding the fruit-localized PHY-dependent signaling cascades regulating carotenoid biosynthesis in fleshy fruits (Llorente et al., 2016b, 2017). Carotenoid profiling revealed a significant reduction in lycopene content in red ripe (RR) fruits of both the SlPHYARNAi and SlPHYB2RNAi lines compared to the WT (Fig. 5A, Supplementary Table S4). In contrast, the content of all other carotenoids analysed (i.e. phytoene, phytofluene, β-carotene, and lutein) remained virtually unchanged in ripe fruits of the transgenic lines compared to WT counterparts. As lycopene is the main carotenoid accumulated in ripe tomato, fruit-specific SlPHYA- or SlPHYB2-knockdown led to a slight, yet significant, reduction in total carotenoid content compared to the WT genotype (Fig. 5A, Supplementary Table S4). In accordance with this, significantly lower mRNA levels of genes encoding carotenoid biosynthesis-related enzymes such as GERANYLGERANYL DIPHOSPHATE SYNTHASE (GGPS), PHYTOENE SYNTHASE 1 (PSY1), and PHYTOENE DESATURASE (PDS) were observed in ripe fruits of SlPHYA and SlPHYB2-silenced lines than in WT counterparts (Fig. 5B, Supplementary Fig. S9). In line with the relatively limited reduction in total carotenoids, no significant differences in lipophilic antioxidant activity were observed between ripe WT and transgenic fruits (Supplementary Table S5). Interestingly, however, red ripe SlPHYB2-down-regulated fruits exhibited increased hydrophilic antioxidant activity compared to the WT, which may be associated with the higher content of total phenolics also detected in SlPHYB2RNAi ripe fruits (Supplementary Table S5). Fig. 5. Open in new tabDownload slide Fruit-specific SlPHYA or SlPHYB2 knockdown represses carotenoid biosynthesis during tomato fruit ripening. (A) Lycopene, phytoene, phytofluene, β-carotene, lutein, and total carotenoid content in red ripe fruits. (B) Schematic representation of carotenoid biosynthetic pathway and graphs showing the transcript abundance of carotenoid biosynthesis genes in ripening fruits. Intermediate reactions are omitted. For simplicity, the mean of the three values for the transgenic lines is shown. Values for each transgenic line are presented in Supplementary Fig. S9, Supplementary Table S4. Data are means (±SE) of at least three biological replicates. Statistically significant differences compared with the wild-type (WT) sample were determined using Student’s t-test: *P<0.05. MG, mature green; Bk, breaker; RR, red ripe; MEP, methylerythritol 4-phosphate; GGDP, geranylgeranyl diphosphate; GGPS, GGDP synthase; PSY, phytoene synthase; PDS, phytoene desaturase; LCYβ, chloroplast-specific β-lycopene cyclase; CYCβ, chromoplast-specific β-lycopene cyclase. (This figure is available in color at JXB online.) Accumulating evidence indicates that light-signaling repressors such as SlPIF1a, SlCOP1, SlCUL4, SlDDB1, and SlDET1 negatively regulate carotenoid biosynthesis in tomato fruits (Azari et al., 2010b; Llorente et al., 2016b) whereas auxin response factors such as SlARF2a and SlARF2b play the opposite role (Hao et al., 2015). To gain insight into the potential role played by these signaling components during the PHY-dependent regulation of carotenoid biosynthesis in tomato fruits, the transcript abundance of their encoding genes was profiled in both SlPHYARNAi and SlPHYB2RNAi ripening fruits (Fig. 6, Supplementary Fig. S10). Among the four SlPIF genes most highly expressed in fruits (Rosado et al., 2016), SlPIF1a, SlPIF1b, and SlPIF4/5 mRNA levels were significantly higher in SlPHYB2-down-regulated fruits compared to the WT counterparts during fruit ripening (MG, Bk, and RR stages), whereas the opposite was observed for SlPIF3 transcripts. Although less pronounced, the overall impacts of fruit-specific SlPHYA knockdown on tomato PIF expression profiles were similar to those observed in the SlPHYB2RNAi lines (Fig. 6, Supplementary Fig. S10). Fig. 6. Open in new tabDownload slide PHY-dependent regulation of fruit carotenogenesis is associated with transcriptional changes in auxin- and light-signaling genes. (A) Transcript abundance of tomato genes encoding PHYTOCHROME INTERACTING FACTORs (SlPIFs). (B) Transcript abundance of tomato genes encoding the light-signaling repressors CONSTITUTIVE PHOTOMORPHOGENIC 1 (SlCOP1), CULLIN 4 (SlCUL4), UV-DAMAGED DNA BINDING PROTEIN 1 (SlDDB1), and DE-ETIOLATED1 (SlDET1). (C) Transcript abundance of the tomato AUXIN RESPONSIVE FACTOR 2a and 2b (SlARF2a and SlARF2b) genes. For simplicity, the mean of the three values for the transgenic lines is shown. Values for each transgenic line are presented in Supplementary Fig. S10. Data are means (±SE) of at least three biological replicates. Statistically significant differences compared with the wild-type (WT) sample were determined using Student’s t-test: *P<0.05. MG, mature green; Bk, breaker; RR, red ripe. Among the genes encoding light-signaling repressors, SlCUL4, SlDDB1, and SlDET1 exhibited significantly higher mRNA levels in SlPHYA-silenced fruits in comparison to the WT at all fruit development stages analysed (Fig. 6B, Supplementary Fig. S10). Moreover, strikingly higher SlDET1 transcript abundance was also detected in SlPHYB2-knockdown compared to WT fruits at all ripening stages (i.e. MG, Bk, and RR) whereas SlCOP1 and SlDDB1 mRNA levels were also up-regulated in SlPHYB2RNAi fruits exclusively at the MG stage. Transcript levels of the positive regulators of tomato fruit carotenogenesis SlARF2a and SlARF2b were considerably lower in SlPHYARNAi and SlPHYB2RNAi fruits, particularly at the Bk and RR stages (Fig. 6C, Supplementary Fig. S10). A PCA plot in which the expression profile of carotenoid biosynthesis-related genes as well as SlPIFs, SlCOP1, SlCUL4, SlDDB1, SlDET1, SlARF2a, and SlARF2b were represented revealed that the WT, SlPHYARNAi, and SlPHYB2RNAi groups clearly separated from each other at the red ripe stage (Supplementary Fig. S11). Altogether, these data suggest that both SlPHYA and SlPHYB2 play overlapping roles in promoting the paralogues SlARF2a and SlARF2b and repressing light-signaling repressors such as SlPIF1a, SlPIF1b, SlPIF4/5, SlCOP1, SlCUL4, SlDDB1, and SlDET1, which in turn mediate the PHY-dependent regulation of carotenoid biosynthesis in ripening tomato fruits. Discussion Studies performed on PHY-deficient mutants have suggested that PHY-dependent light perception participates in the regulation of several aspects of tomato fruit biology (Gupta et al., 2014; Bianchetti et al., 2017). Here, we applied a RNAi-mediated organ-specific silencing approach to investigate the impact of fruit-localized SlPHYs on tomato fruit physiology and quality traits. Differently from the pleiotropic phenotypical alterations observed in phy mutants (Gupta et al., 2014; Bianchetti et al., 2017), the fruit-specific silencing of the target SlPHY genes resulted in no obvious impacts on plant vegetative growth and overall yield. This suggests that the perturbation in fruit metabolism caused by the fruit-specific SlPHY manipulation does not propagate from fruits to the rest of the plant, which agrees with the limited transference of substances out of this predominantly sink organ. In a previous work, we demonstrated that a global deficiency in functional PHYs drastically reduces chlorophyll content and chloroplast abundance in tomato fruits (Bianchetti et al., 2017). Therefore, the PHY-mediated regulation of plastid biogenesis and maturation widely reported for leaf tissues (Stephenson et al., 2009; Oh and Montgomery, 2014; Melo et al., 2016) seems to be conserved early in the development of tomato fruits. In this current work, it is further demonstrated that fruit-localized SlPHYA and SlPHYB2 play distinct roles in controlling chloroplast biogenesis and activity during early stages of tomato fruit development. The results indicate that SlPHYA-mediated light perception promotes fruit chloroplast biogenesis and differentiation, as inferred from the reduced chlorophyll content, lower chloroplast abundance, and poorly-developed grana stacking detected in SlPHYARNAi immature fruits (Fig. 2). In line with this observation, an analysis of single and multiple phy mutants also suggested that SlPHYA is a major regulator of chlorophyll accumulation in tomato fruits (Gupta et al., 2014). In land plants, chloroplast division depends on nucleus-encoded proteins that form ring structures at the division site (Jarvis and López-Juez, 2013). Our findings clearly demonstrate that fruit-localized SlPHYA influences the transcript levels of genes derived from the ancestral prokaryotic cell-division machinery, such as SlFtsZ (i.e. SlFtsZ1, SlFtsZ2) and SlARCs (i.e. SlARC3 and SlARC6), as well as those encoding chloroplast division-related proteins specific to land plants, such as SlPDV2. In Arabidopsis, PDV2 determines the rate of chloroplast division and is positively regulated by cytokinins, being strongly promoted in transgenic plants overexpressing the cytokinin signaling-related transcription factor CRF2 (Okazaki et al., 2009; Cortleven and Schmülling, 2015). SlCRF2, along with other SlCRF and TRR genes, were drastically repressed in PHYA-down-regulated fruits, implying that changes in cytokinin signaling mediate the PHYA-dependent regulation of plastid division during early stages of tomato fruit development. In agreement with this, accumulating evidence indicates that there is an intensive crosstalk between the PHY and cytokinin signaling cascades, with particular involvement of CRF and type-A ARR proteins (Salomé et al., 2006; Oh et al., 2009). Fruit-specific SlPHYA-silencing also promoted the mRNA accumulation of genes encoding all the major light-signaling repressor proteins already described to negatively regulate chloroplast biogenesis in tomato fruits, i.e. SlCOP1, SlCUL4, SlDDB1, and SlDET1 (Liu et al., 2004; Kolotilin et al., 2007; Wang et al., 2008; Azari et al., 2010a). Defective mutants or transgenic lines with reduced levels of each of these genes are known to develop more chloroplasts containing more grana/thylakoids in both leaves and immature fruits (Cookson et al., 2003; Liu et al., 2004; Kolotilin et al., 2007; Wang et al., 2008; Azari et al., 2010a), which in some cases, such as in the SlDET1-knockout mutant, is associated with the up-regulation of plastid biogenesis-related genes (Kolotilin et al., 2007). Therefore, the presence of fewer chloroplasts with poorly developed or almost no grana in immature fruits of the SlPHYA-suppressed lines agrees with the higher transcript abundance of SlCOP1, SlDDB1, and particularly SlCUL4 and SlDET1 in these transgenic lines compared to the WT genotype. In contrast, fruit-localized SlPHYB2 was shown to play a negative role in chlorophyll accumulation, as evidenced by the increment in chlorophyll content in immature fruits of SlPHYB2RNAi plants with no impact in chloroplast number in pericarp cells. As SlPHYB2 fruit-specific silencing led to higher SlGLK2 mRNA levels compared to the WT genotype, it seems plausible to suggest that the effect of SlPHYB2 on fruit chloroplasts is mediated by SlGLK2, the master regulator of chloroplast development in tomato fruits (Powell et al., 2012). Further suggesting that the SlPHYB2-mediated regulation of SlGLK2 expression is essential for the consequent changes in fruit chlorophyll accumulation, no obvious changes in chlorophyll content were observed in phyb2 mutants from tomato varieties that lacked functional SlGLK2 proteins (Gupta et al., 2014). In agreement with these findings, PHY-dependent transcriptional regulation of GLK genes has been increasingly reported in vegetative tissues of other plant species (Oh and Montgomery, 2014; Song et al., 2014). Alterations in chloroplast number, internal structure, and size during the early development of tomato fruits significantly impact the abundance of metabolites associated with organoleptic and nutritional quality at the ripe stage (Galpaz et al., 2008; Cocaliadis et al., 2014). Intense starch synthesis and degradation take place in tomato fruit chloroplasts at the unripe and breaker stages, respectively (Schaffer and Petreikov, 1997). Whereas the global deficiency in PHYs significantly reduces the starch content in immature tomato fruits (Bianchetti et al., 2017), fruit-localized SlPHYA or SlPHYB2 suppression increased fruit starch levels and markedly altered the transcriptional profile of starch biosynthesis-related genes at the immature green stage (Fig. 4). AGPase, which catalyses the rate-limiting reaction in the starch synthesis pathway, is both transcriptionally and post-translational regulated by light (Harn et al., 2000; Geigenberger, 2011), although the role played by PHYs in this regulatory process remains elusive. During early fruit development, SlPHYA-suppressed fruits exhibited increased mRNA levels of both SlAGPaseL1 and SlAGPaseL3, which encode AGPase large subunits, and SlSTS1, SlSTS2, and SlSTS6, which encode starch synthase enzymes, along with an increase in starch accumulation and reduced soluble sugar content, thus indicating a repressor role for fruit-localized SlPHYA on the first steps of starch synthesis in tomato fruits. Whether the up-regulation of starch biosinthesis-related genes is a compensatory mechanism to cope with the fewer and poorly developed chloroplasts observed in SlPHYARNAi immature fruits remains to be investigated. In contrast, the increased starch accumulation detected in SlPHYB2-silenced immature fruits was not associated with increments in transcript abundance of AGPase-encoding genes nor with prominent reductions in soluble sugars, but instead were accompanied by increments in SlSTS1 and SlSTS2 mRNA levels. Furthermore, as no significant alterations in plastid abundance or internal structure were observed in SlPHYB2RNAi immature fruits, it seems likely that this genetic manipulation caused less prominent changes than SlPHYA-silencing on reactions taking place within fruit chloroplasts, including starch biosynthesis. Altogether, these findings suggest that SlPHYA and SlPHYB2 negatively regulate starch synthesis via overlapping, yet distinct, mechanisms. The influence of auxin on fruit sugar metabolism has been increasingly reported (Purgatto et al., 2002; Yuan and Carbaugh, 2007; Bianchetti et al., 2017). In tomato, SlARF4 has been described as a key negative regulator of starch synthesis during early fruit development via the transcriptional and post-transcriptional down-regulation of AGPase (Sagar et al., 2013). Recent findings have also indicated that PHYs strictly regulate the transcript abundance of this particular auxin response factor in both vegetative (Melo et al., 2016) and fruit tissues (Bianchetti et al., 2017). In line with this, the increased starch accumulation in pre-ripening SlPHYA- and SlPHYB2-silenced fruits correlated well with the down-regulation of SlARF4 in these transgenic lines (Fig. 4). In fact, SlPHYARNAi rather than SlPHYB2RNAi exhibited the most expressive decrease in SlARF4, and only the former displayed increased mRNA levels of AGPase-encoding genes in immature fruits. Together, these data strongly suggest that fruit-localized PHYA, and to some extent SlPHYB2, positively modulates SlARF4, which in turn represses starch biosynthetic enzymes, such as AGPase and STS, consequently limiting starch synthesis in pre-ripening tomato fruits. Previous findings indicated that a global deficiency in functional phytochromes transcriptionally represses both sink-related and starch biosynthesis-related enzymes in early developing tomato fruits, suggesting a promotive role of PHYs on the regulation of these processes (Bianchetti et al., 2017). However, it remained unclear whether these responses were dependent on fruit-localized PHYs or were the consequence of collateral negative effects of the global PHY deficiency on vegetative plant growth. Here, we shed light on this topic by showing that fruit-localized SlPHYA, and to some extent SlPHYB2, repress both starch metabolism and key determinants of tomato fruit sink strength, including SlLIN5 transcript accumulation (Fridman and Zamir, 2003; Kocal et al., 2008). Consequently, the down-regulation in starch synthesis and sink activity previously observed in fruits of the PHY-deficient mutant aurea (Bianchetti et al., 2017) may be due either to limitations in vegetative growth and metabolism or to the combinatory effect of the deficiency in all phytochromes instead of only in SlPHYA or SlPHYB2. Moreover, it also seems tempting to suggest that the fewer and poorly-developed chloroplasts detected in SlPHYARNAi immature fruits restrict photoassimilate production via fruit photosynthesis; therefore, the observed up-regulation of sink-related genes in transgenic fruits may represent a compensatory mechanism to maintain fruit growth and intense starch accumulation despite potential limitations in fruit-localized photoassimilation. The link between PHY-dependent light perception and carotenoid metabolism in both vegetative and fruit tissues has been highlighted by a number of studies (Alba et al., 2000a; Llorente et al., 2016b). Exposure of wild-type tomato fruits to red light (Alba et al., 2000a) or constitutively silencing of SlPIF1a (Llorente et al., 2016b) promotes tomato fruit lycopene accumulation, thereby implying a positive role of PHY-dependent signaling cascades in the fruit carotenoid biosynthetic pathway. Consistent with this, our findings indicate that fruit-localized SlPHYA and SlPHYB2 positively influence the transcript accumulation of all the major carotenoid biosynthesis-related genes, including SlGGPS, SlPSY1, SlPDS, SlCYCβ, and SlLYCβ, consequently modifying the lycopene and total carotenoid content in this fleshy fruit. Light-signaling repressor proteins such as SlDET1, SlDDB1, SlCOP1, SlCUL4, and more recently SlPIF1a have been identified as key negative regulators of tomato fruit carotenoid synthesis (Liu et al., 2004; Kolotilin et al., 2007; Wang et al., 2008; Azari et al., 2010a; Llorente et al., 2016b). Among these, the transcription factor SlPIF1a was shown to directly bind to the promoter of SlPSY1 to repress fruit carotenogenesis (Llorente et al., 2016b), thus resembling the action of its ortholog in Arabidopsis (AtPIF1) in controlling carotenoid biosynthesis in leaf tissues (Toledo-Ortiz et al., 2010). Therefore, the marked up-regulation of SlDET1, SlDDB1, SlCOP1, SlCUL4, SlPIF1a, and SlPIF1b together with the overall repression of carotenoid biosynthesis observed in both SlPHYA- and SlPHYB2-silenced fruits imply that light-signaling repressor proteins participate in SlPHYA- and SlPHYB2-mediated regulation of fruit carotenogenesis. In addition, it is becoming increasingly well established that auxin represses tomato ripening and down-regulates lycopene biosynthetic genes (Su et al., 2015). Among tomato ARF genes, two paralogs, SlARF2a and SlARF2b, have emerged as key positive regulators of tomato fruit ripening and lycopene accumulation (Hao et al., 2015). Either SlPHYA or SlPHYB2 fruit-specific silencing profoundly reduced both SlARF2a and SlARF2b, suggesting the involvement of these auxin signaling elements in the PHY-dependent regulation of carotenoid biosynthesis in tomato fruits. Overall, our results shed light on the specific role played by fruit-localized phytochromes and their downstream signaling cascades, showing that plastid division, as well as sugar and carotenoid metabolism, are profoundly regulated by SlPHYA- and SlPHYB2-mediated light perception. A model summarizing the influence of fruit-localized SlPHYs on tomato fruit physiology is presented in Fig. 7. According to this model, SlPHYA and SlPHYB2 play overlapping roles in regulating starch and carotenoid biosynthesis, whereas they differentially regulate distinct aspects of fruit plastid biogenesis and maturation. Compared to SlPHYB2, SlPHYA-dependent light perception seems to play a major role in promoting plastid division and differentiation as well as in controlling sink-related transcripts in tomato fruits. The data implicate cytokinin signaling-related proteins as mediators of the SlPHYA-dependent regulation of the plastid division machinery, and specific ARF genes as potential intermediates in the PHY-mediated regulation of fruit sugar and carotenoid metabolism. Altogether, these findings show that fruit-specific manipulation of PHY genes represents a promising approach to differentially regulate multiple biosynthetic pathways and, consequently, to modify the nutritional value of edible fleshy fruits. Fig. 7. Open in new tabDownload slide Proposed model for phytochrome-mediated signaling events controlling chloroplast biogenesis, and sugar and carotenoid metabolism in tomato fruits. (A) SlPHYA- and SlPHYB2-dependent light perception regulate fruit plastid division and maturation, respectively. By promoting key members of the cytokinin signaling-related CRF and TRR gene family, SlPHYA up-regulates SlPDV2, a rate-limiting component of the plastid division machinery. Moreover, the SlPHYA-mediated down-regulation of light-signaling repressors, such as SlCOP1, SlDET1, SlDDB1, and SlCUL4, induces other major components of the chloroplast division machinery, such as SlFTsZs and SlARCs. In contrast, Sl-PHYB2 represses the chloroplast differentiation transcription factor SlGLK2, consequently limiting chloroplast differentiation during early fruit development. (B) Fruit-localized SlPHYA and SlPHYB2 play overlapping roles in repressing and promoting starch and carotenoid biosynthesis, respectively. Both SlPHYA and SlPHYB2 induce SlARF4, a negative regulator of AGPase and starch accumulation in tomato fruits. In contrast, these same photoreceptors promote both SlARF2 paralogues and inhibit all the major genes encoding light-signaling repressor proteins, consequently up-regulating most components of the tomato carotenoid biosynthetic route. Arrows at the ends of lines indicate stimulatory effects, whereas bars indicate inhibitory effects. AGPase, ADP-glucose pyrophosphorylase; ARC, accumulation and replication of chloroplasts; ARF, auxin response factor; COP1, constitutive photomorphogenic 1; CRF, cytokinin response factor; CUL4, cullin 4; DDB1, UV-damaged DNA binding protein 1; DET1, de-etiolated1; FtsZ, filamentous temperature sensitive-Z; GGPS, geranylgeranyl pyrophosphate synthase; GLK2, golden2-like-2; PDS, phytoene desaturase; PDV2, plastid division 2; PIF, phytochrome interacting factor; PSY, phytoene synthase; TRR, tomato response regulator. Supplementary data Supplementary data are available at JXB online. Fig. S1. Transcriptional profile of tomato PHY genes in PHY-silenced fruits. Fig. S2. Vegetative phenotypes of the transgenic plants. Fig. S3. Visual phenotypes and color changes of PHY-silenced fruits. Fig. S4. Plastid structure in PHY-silenced fruits. Fig. S5. Carbohydrate profile in PHY-silenced fruits. Fig. S6. Transcript abundance of starch biosynthetic genes in PHY-silenced fruits. Fig. S7. PCA of the expression profile of sink-related and starch biosynthesis-related genes. Fig. S8. HY5-, PIF-, and ARF-binding motifs identified in the promoter regions of starch biosynthesis-related tomato genes. Fig. S9. Carotenoid metabolism during ripening in PHY-silenced fruits. Fig. S10. Transcript abundance of photomorphogenesis- and auxin-related genes in PHY-silenced fruits. Fig. S11. PCA of the expression profiles of photomorphogenesis-related, auxin-related, and carotenoid biosynthesis-related genes. Table S1. Primer sequences. Table S2. Homology of the RNAi fragments. Table S3. Relative transcript ratios of SlAGPase in immature fruits. Table S4. Carotenoid profiles in red ripe fruits. Table S5. Antioxidant activity and total phenolics in red ripe fruits. Acknowledgements The authors sincerely thank Prof. Lazaro E. P. Peres for providing the Micro-Tom GLK2 seeds. This work was supported by the CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico, grant no. 442045/2014-0) and the FAPESP (Fundação de Amparo à Pesquisa do Estado de São Paulo, grant nos. 2013/18056-2 and 2016/01128-9). References Alba R , Cordonnier-Pratt MM, Pratt LH. 2000a . Fruit-localized phytochromes regulate lycopene accumulation independently of ethylene production in tomato . Plant Physiology 123 , 363 – 370 . Google Scholar Crossref Search ADS WorldCat Alba R , Kelmenson PM, Cordonnier-Pratt MM, Pratt LH. 2000b . The phytochrome gene family in tomato and the rapid differential evolution of this family in angiosperms . Molecular Biology and Evolution 17 , 362 – 373 . Google Scholar Crossref Search ADS WorldCat Azari R , Reuveni M, Evenor D, Nahon S, Shlomo H, Chen L, Levin I. 2010a . Overexpression of UV-DAMAGED DNA BINDING PROTEIN 1 links plant development and phytonutrient accumulation in high pigment-1 tomato . Journal of Experimental Botany 61 , 3627 – 3637 . Google Scholar Crossref Search ADS WorldCat Azari R , Tadmor Y, Meir A, Reuveni M, Evenor D, Nahon S, Shlomo H, Chen L, Levin I. 2010b . Light signaling genes and their manipulation towards modulation of phytonutrient content in tomato fruits . Biotechnology Advances 28 , 108 – 118 . Google Scholar Crossref Search ADS WorldCat Bianchetti RE , Cruz AB, Oliveira BS, Demarco D, Purgatto E, Peres LEP, Rossi M, Freschi L. 2017 . Phytochromobilin deficiency impairs sugar metabolism through the regulation of cytokinin and auxin signaling in tomato fruits . Scientific Reports 7 , 7822 . Google Scholar Crossref Search ADS PubMed WorldCat Carvalho RF , Campos ML, Pino LE, Crestana SL, Zsögön A, Lima JE, Benedito VA, Peres LE. 2011 . Convergence of developmental mutants into a single tomato model system: ‘Micro-Tom’ as an effective toolkit for plant development research . Plant Methods 7 , 18 . Google Scholar Crossref Search ADS PubMed WorldCat Chaabouni S , Jones B, Delalande C, Wang H, Li Z, Mila I, Frasse P, Latché A, Pech JC, Bouzayen M. 2009 . Sl-IAA3, a tomato Aux/IAA at the crossroads of auxin and ethylene signalling involved in differential growth . Journal of Experimental Botany 60 , 1349 – 1362 . Google Scholar Crossref Search ADS PubMed WorldCat Chory J , Peto CA. 1990 . Mutations in the DET1 gene affect cell-type-specific expression of light-regulated genes and chloroplast development in Arabidopsis . Proceedings of the National Academy of Sciences, USA 87 , 8776 – 8780 . Google Scholar Crossref Search ADS WorldCat Chow CN , Zheng HQ, Wu NY, Chien CH, Huang HD, Lee TY, Chiang-Hsieh YF, Hou PF, Yang TY, Chang WC. 2016 . PlantPAN 2.0: an update of plant promoter analysis navigator for reconstructing transcriptional regulatory networks in plants . Nucleic Acids Research 44 , D1154 – D1160 . Google Scholar Crossref Search ADS PubMed WorldCat Cocaliadis MF , Fernández-Muñoz R, Pons C, Orzaez D, Granell A. 2014 . Increasing tomato fruit quality by enhancing fruit chloroplast function. A double-edged sword ? Journal of Experimental Botany 65 , 4589 – 4598 . Google Scholar Crossref Search ADS PubMed WorldCat Cookson PJ , Kiano JW, Shipton CA, Fraser PD, Romer S, Schuch W, Bramley PM, Pyke KA. 2003 . Increases in cell elongation, plastid compartment size and phytoene synthase activity underlie the phenotype of the high pigment-1 mutant of tomato . Planta 217 , 896 – 903 . Google Scholar Crossref Search ADS PubMed WorldCat Cortleven A , Schmülling T. 2015 . Regulation of chloroplast development and function by cytokinin . Journal of Experimental Botany 66 , 4999 – 5013 . Google Scholar Crossref Search ADS PubMed WorldCat Dai X , Zhao PX. 2011 . psRNATarget: a plant small RNA target analysis server . Nucleic Acids Research 39 , W155 – W159 . Google Scholar Crossref Search ADS PubMed WorldCat Davuluri GR , van Tuinen A, Fraser PD, et al. 2005 . Fruit-specific RNAi-mediated suppression of DET1 enhances carotenoid and flavonoid content in tomatoes . Nature Biotechnology 23 , 890 – 895 . Google Scholar Crossref Search ADS PubMed WorldCat Deng X-W , Quail PH. 1992 . Genetic and phenotypic characterization of cop1 mutants of Arabidopsis thaliana . The Plant Journal 2 , 83 – 95 . Google Scholar Crossref Search ADS WorldCat Duek PD , Fankhauser C. 2005 . bHLH class transcription factors take centre stage in phytochrome signalling . Trends in Plant Science 10 , 51 – 54 . Google Scholar Crossref Search ADS PubMed WorldCat Enfissi EM , Barneche F, Ahmed I, et al. 2010 . Integrative transcript and metabolite analysis of nutritionally enhanced DE-ETIOLATED1 downregulated tomato fruit . The Plant Cell 22 , 1190 – 1215 . Google Scholar Crossref Search ADS PubMed WorldCat Expósito-Rodríguez M , Borges AA, Borges-Pérez A, Pérez JA. 2008 . Selection of internal control genes for quantitative real-time RT-PCR studies during tomato development process . BMC Plant Biology 8 , 131 . Google Scholar Crossref Search ADS PubMed WorldCat Fernandez AI , Viron N, Alhagdow M, et al. 2009 . Flexible tools for gene expression and silencing in tomato . Plant Physiology 151 , 1729 – 1740 . Google Scholar Crossref Search ADS PubMed WorldCat Figueroa CM , Kuhn ML, Falaschetti CA, Solamen L, Olsen KW, Ballicora MA, Iglesias AA. 2013 . Unraveling the activation mechanism of the potato tuber ADP-glucose pyrophosphorylase . PLoS ONE 8 , e66824 . Google Scholar Crossref Search ADS PubMed WorldCat Fridman E , Zamir D. 2003 . Functional divergence of a syntenic invertase gene family in tomato, potato, and Arabidopsis . Plant Physiology 131 , 603 – 609 . Google Scholar Crossref Search ADS PubMed WorldCat Galpaz N , Wang Q, Menda N, Zamir D, Hirschberg J. 2008 . Abscisic acid deficiency in the tomato mutant high-pigment 3 leading to increased plastid number and higher fruit lycopene content . The Plant Journal 53 , 717 – 730 . Google Scholar Crossref Search ADS PubMed WorldCat Geigenberger P . 2011 . Regulation of starch biosynthesis in response to a fluctuating environment . Plant Physiology 155 , 1566 – 1577 . Google Scholar Crossref Search ADS PubMed WorldCat Giliberto L , Perrotta G, Pallara P, Weller JL, Fraser PD, Bramley PM, Fiore A, Tavazza M, Giuliano G. 2005 . Manipulation of the blue light photoreceptor cryptochrome 2 in tomato affects vegetative development, flowering time, and fruit antioxidant content . Plant Physiology 137 , 199 – 208 . Google Scholar Crossref Search ADS PubMed WorldCat Giovannoni J , Nguyen C, Ampofo B, Zhong S, Fei Z. 2017 . The epigenome and transcriptional dynamics of fruit ripening . Annual Review of Plant Biology 68 , 61 – 84 . Google Scholar Crossref Search ADS PubMed WorldCat Gupta SK , Sharma S, Santisree P, Kilambi HV, Appenroth K, Sreelakshmi Y, Sharma R. 2014 . Complex and shifting interactions of phytochromes regulate fruit development in tomato . Plant, Cell & Environment 37 , 1688 – 1702 . Google Scholar Crossref Search ADS PubMed WorldCat Hao Y , Hu G, Breitel D, Liu M, Mila I, Frasse P, Fu Y, Aharoni A, Bouzayen M, Zouine M. 2015 . Auxin response factor SlARF2 is an essential component of the regulatory mechanism controlling fruit ripening in tomato . PLoS Genetics 11 , e1005649 . Google Scholar Crossref Search ADS PubMed WorldCat Harn CH , Bae JM, Lee SS, Min SR, Liu JR. 2000 . Presence of multiple cDNAs encoding an isoform of ADP-glucose pyrophosphorylase large subunit from sweet potato and characterization of expression levels . Plant & Cell Physiology 41 , 1235 – 1242 . Google Scholar Crossref Search ADS PubMed WorldCat Hauser BA , Pratt LH, Cordonnier-Pratt MM. 1997 . Absolute quantification of five phytochrome transcripts in seedlings and mature plants of tomato (Solanum lycopersicum L.) . Planta 201 , 379 – 387 . Google Scholar Crossref Search ADS PubMed WorldCat Inagaki N , Kinoshita K, Kagawa T, Tanaka A, Ueno O, Shimada H, Takano M. 2015 . Phytochrome B mediates the regulation of chlorophyll biosynthesis through transcriptional regulation of ChlH and GUN4 in rice seedlings . PLoS ONE 10 , e0135408 . Google Scholar Crossref Search ADS PubMed WorldCat Jarvis P , López-Juez E. 2013 . Biogenesis and homeostasis of chloroplasts and other plastids . Nature Reviews Molecular Cell Biology 14 , 787 – 802 . Google Scholar Crossref Search ADS PubMed WorldCat Kim D , Hwang SK, Okita TW. 2007 . Subunit interactions specify the allosteric regulatory properties of the potato tuber ADP-glucose pyrophosphorylase . Biochemical and Biophysical Research Communications 362 , 301 – 306 . Google Scholar Crossref Search ADS PubMed WorldCat Kocal N , Sonnewald U, Sonnewald S. 2008 . Cell wall-bound invertase limits sucrose export and is involved in symptom development and inhibition of photosynthesis during compatible interaction between tomato and Xanthomonas campestris pv vesicatoria . Plant Physiology 148 , 1523 – 1536 . Google Scholar Crossref Search ADS PubMed WorldCat Kolotilin I , Koltai H, Tadmor Y, Bar-Or C, Reuveni M, Meir A, Nahon S, Shlomo H, Chen L, Levin I. 2007 . Transcriptional profiling of high pigment-2dg tomato mutant links early fruit plastid biogenesis with its overproduction of phytonutrients . Plant Physiology 145 , 389 – 401 . Google Scholar Crossref Search ADS PubMed WorldCat Kumar R , Khurana A, Sharma AK. 2014 . Role of plant hormones and their interplay in development and ripening of fleshy fruits . Journal of Experimental Botany 65 , 4561 – 4575 . Google Scholar Crossref Search ADS PubMed WorldCat Lira BS , Gramegna G, Trench BA, et al. 2017 . Manipulation of a senescence-associated gene improves fleshy fruit yield . Plant Physiology 175 , 77 – 91 . Google Scholar Crossref Search ADS PubMed WorldCat Lira BS , Rosado D, Almeida J, de Souza AP, Buckeridge MS, Purgatto E, Guyer L, Hörtensteiner S, Freschi L, Rossi M. 2016 . Pheophytinase knockdown impacts carbon metabolism and nutraceutical content under normal growth conditions in tomato . Plant & Cell Physiology 57 , 642 – 653 . Google Scholar Crossref Search ADS PubMed WorldCat Liu YS , Roof S, Ye ZB, Barry C, van Tuinen A, Vrebalov J, Bowler C, Giovannoni J. 2004 . Manipulation of light signal transduction as a means of modifying fruit nutritional quality in tomato . Proceedings of the National Academy of Sciences, USA 101 , 9897 – 9902 . Google Scholar Crossref Search ADS WorldCat Llorente B , D’Andrea L, Rodríguez-Concepción M. 2016a . Evolutionary recycling of light signaling components in fleshy fruits: new insights on the role of pigments to monitor ripening . Frontiers in Plant Science 7 , 263 . Google Scholar Crossref Search ADS WorldCat Llorente B , D’Andrea L, Ruiz-Sola MA, Botterweg E, Pulido P, Andilla J, Loza-Alvarez P, Rodriguez-Concepcion M. 2016b . Tomato fruit carotenoid biosynthesis is adjusted to actual ripening progression by a light-dependent mechanism . The Plant Journal 85 , 107 – 119 . Google Scholar Crossref Search ADS WorldCat Llorente B , Martinez-Garcia JF, Stange C, Rodriguez-Concepcion M. 2017 . Illuminating colors: regulation of carotenoid biosynthesis and accumulation by light . Current Opinion in Plant Biology 37 , 49 – 55 . Google Scholar Crossref Search ADS PubMed WorldCat Martínez-García JF , Huq E, Quail PH. 2000 . Direct targeting of light signals to a promoter element-bound transcription factor . Science 288 , 859 – 863 . Google Scholar Crossref Search ADS PubMed WorldCat Melo NK , Bianchetti RE, Lira BS, Oliveira PM, Zuccarelli R, Dias DL, Demarco D, Peres LE, Rossi M, Freschi L. 2016 . Nitric oxide, ethylene, and auxin cross talk mediates greening and plastid development in deetiolating tomato seedlings . Plant Physiology 170 , 2278 – 2294 . Google Scholar Crossref Search ADS PubMed WorldCat Oh E , Kang H, Yamaguchi S, Park J, Lee D, Kamiya Y, Choi G. 2009 . Genome-wide analysis of genes targeted by PHYTOCHROME INTERACTING FACTOR 3-LIKE5 during seed germination in Arabidopsis . The Plant Cell 21 , 403 – 419 . Google Scholar Crossref Search ADS PubMed WorldCat Oh S , Montgomery BL. 2014 . Phytochrome-dependent coordinate control of distinct aspects of nuclear and plastid gene expression during anterograde signaling and photomorphogenesis . Frontiers in Plant Science 5 , 171 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Okazaki K , Kabeya Y, Suzuki K, Mori T, Ichikawa T, Matsui M, Nakanishi H, Miyagishima SY. 2009 . The PLASTID DIVISION1 and 2 components of the chloroplast division machinery determine the rate of chloroplast division in land plant cell differentiation . The Plant Cell 21 , 1769 – 1780 . Google Scholar Crossref Search ADS PubMed WorldCat Pepper A , Delaney T, Washburn T, Poole D, Chory J. 1994 . DET1, a negative regulator of light-mediated development and gene expression in arabidopsis, encodes a novel nuclear-localized protein . Cell 78 , 109 – 116 . Google Scholar Crossref Search ADS PubMed WorldCat Petreikov M , Shen S, Yeselson Y, Levin I, Bar M, Schaffer AA. 2006 . Temporally extended gene expression of the ADP-Glc pyrophosphorylase large subunit (AgpL1) leads to increased enzyme activity in developing tomato fruit . Planta 224 , 1465 – 1479 . Google Scholar Crossref Search ADS PubMed WorldCat Pino LE , Lombardi-Crestana S, Azevedo MS, Scotton DC, Borgo L, Quecini V, Figueira A, Peres LE. 2010 . The Rg1 allele as a valuable tool for genetic transformation of the tomato ‘Micro-Tom’ model system . Plant Methods 6 , 23 . Google Scholar Crossref Search ADS PubMed WorldCat Piringer AA , Heinze PH. 1954 . Effect of light on the formation of a pigment in the tomato fruit cuticle . Plant Physiology 29 , 467 – 472 . Google Scholar Crossref Search ADS PubMed WorldCat Porra RJ , Thompson WA, Kriedemann PE. 1989 . Determination of accurate extinction coefficients and simultaneous equations for assaying chlorophylls a and b extracted with four different solvents: verification of the concentration of chlorophyll standards by atomic absorption spectroscopy . Biochimica et Biophysica Acta 975 , 384 – 394 . Google Scholar Crossref Search ADS WorldCat Powell AL , Nguyen CV, Hill T, et al. 2012 . Uniform ripening encodes a Golden 2-like transcription factor regulating tomato fruit chloroplast development . Science 336 , 1711 – 1715 . Google Scholar Crossref Search ADS PubMed WorldCat Purgatto E , Oliveira do Nascimento JR, Lajolo FM, Cordenunsi BR. 2002 . The onset of starch degradation during banana ripening is concomitant to changes in the content of free and conjugated forms of indole-3-acetic acid . Journal of Plant Physiology 159 , 1105 – 1111 . Google Scholar Crossref Search ADS WorldCat Quadrana L , Almeida J, Otaiza SN, et al. 2013 . Transcriptional regulation of tocopherol biosynthesis in tomato . Plant Molecular Biology 81 , 309 – 325 . Google Scholar Crossref Search ADS PubMed WorldCat Rosado D , Gramegna G, Cruz A, Lira BS, Freschi L, de Setta N, Rossi M. 2016 . Phytochrome interacting factors (PIFs) in Solanum lycopersicum: diversity, evolutionary history and expression profiling during different developmental processes . PLoS ONE 11 , e0165929 . Google Scholar Crossref Search ADS PubMed WorldCat Ruijter JM , Ramakers C, Hoogaars WMH, Karlen Y, Bakker O, van den Hoff MJB, Moorman AFM. 2009 . Amplification efficiency: linking baseline and bias in the analysis of quantitative PCR data . Nucleic Acids Research 37 , e45 . Google Scholar Crossref Search ADS PubMed WorldCat Sagar M , Chervin C, Mila I, et al. 2013 . SlARF4, an auxin response factor involved in the control of sugar metabolism during tomato fruit development . Plant Physiology 161 , 1362 – 1374 . Google Scholar Crossref Search ADS PubMed WorldCat Salomé PA , To JP, Kieber JJ, McClung CR. 2006 . Arabidopsis response regulators ARR3 and ARR4 play cytokinin-independent roles in the control of circadian period . The Plant Cell 18 , 55 – 69 . Google Scholar Crossref Search ADS PubMed WorldCat Schaffer AA , Petreikov M. 1997 . Sucrose-to-starch metabolism in tomato fruit undergoing transient starch accumulation . Plant Physiology 113 , 739 – 746 . Google Scholar Crossref Search ADS PubMed WorldCat Schofield A , Paliyath G. 2005 . Modulation of carotenoid biosynthesis during tomato fruit ripening through phytochrome regulation of phytoene synthase activity . Plant Physiology and Biochemistry 43 , 1052 – 1060 . Google Scholar Crossref Search ADS PubMed WorldCat Schrager-Lavelle A , Herrera LA, Maloof JN. 2016 . Tomato phyE is required for shade avoidance in the absence of phyB1 and phyB2 . Frontiers in Plant Science 7 , 1275 . Google Scholar Crossref Search ADS PubMed WorldCat Schroeder DF , Gahrtz M, Maxwell BB, Cook RK, Kan JM, Alonso JM, Ecker JR, Chory J. 2002 . De-etiolated 1 and damaged DNA binding protein 1 interact to regulate Arabidopsis photomorphogenesis . Current Biology 12 , 1462 – 1472 . Google Scholar Crossref Search ADS PubMed WorldCat Shi X , Gupta S, Rashotte AM. 2012 . Solanum lycopersicum cytokinin response factor (SlCRF) genes: characterization of CRF domain-containing ERF genes in tomato . Journal of Experimental Botany 63 , 973 – 982 . Google Scholar Crossref Search ADS PubMed WorldCat Singleton VL , Rossi JA. 1965 . Colorimetry of total phenolics with phosphomolybdic-phosphotungstic acid reagents . American Journal of Enology and Viticulture 16 , 144 – 158 . Google Scholar OpenURL Placeholder Text WorldCat Song Y , Yang C, Gao S, Zhang W, Li L, Kuai B. 2014 . Age-triggered and dark-induced leaf senescence require the bHLH transcription factors PIF3, 4, and 5 . Molecular Plant 7 , 1776 – 1787 . Google Scholar Crossref Search ADS PubMed WorldCat Song YH , Yoo CM, Hong AP, et al. 2008 . DNA-binding study identifies C-box and hybrid C/G-box or C/A-box motifs as high-affinity binding sites for STF1 and LONG HYPOCOTYL5 proteins . Plant Physiology 146 , 1862 – 1877 . Google Scholar Crossref Search ADS PubMed WorldCat Stephenson PG , Fankhauser C, Terry MJ. 2009 . PIF3 is a repressor of chloroplast development . Proceedings of the National Academy of Sciences, USA 106 , 7654 – 7659 . Google Scholar Crossref Search ADS WorldCat Su L , Diretto G, Purgatto E, Danoun S, Zouine M, Li Z, Roustan JP, Bouzayen M, Giuliano G, Chervin C. 2015 . Carotenoid accumulation during tomato fruit ripening is modulated by the auxin–ethylene balance . BMC Plant Biology 15 , 114 . Google Scholar Crossref Search ADS PubMed WorldCat Suguiyama VF , Silva EA, Meirelles ST, Centeno DC, Braga MR. 2014 . Leaf metabolite profile of the Brazilian resurrection plant Barbacenia purpurea Hook. (Velloziaceae) shows two time-dependent responses during desiccation and recovering . Frontiers in Plant Science 5 , 96 . Google Scholar Crossref Search ADS PubMed WorldCat Thomann A , Dieterle M, Genschik P. 2005 . Plant CULLIN-based E3s: phytohormones come first . FEBS Letters 579 , 3239 – 3245 . Google Scholar Crossref Search ADS PubMed WorldCat Toledo-Ortiz G , Huq E, Rodríguez-Concepción M. 2010 . Direct regulation of phytoene synthase gene expression and carotenoid biosynthesis by phytochrome-interacting factors . Proceedings of the National Academy of Sciences, USA 107 , 11626 – 11631 . Google Scholar Crossref Search ADS WorldCat Tomato Genome Consortium . 2012 . The tomato genome sequence provides insights into fleshy fruit evolution . Nature 485 , 635 – 641 . Crossref Search ADS PubMed WorldCat van Tuinen A , Kerckhoffs LH, Nagatani A, Kendrick RE, Koornneef M. 1995a . Far-red light-insensitive, phytochrome A-deficient mutants of tomato . Molecular & General Genetics 246 , 133 – 141 . Google Scholar Crossref Search ADS WorldCat van Tuinen A , Kerckhoffs L, Nagatani A, Kendrick RE, Koornneef M. 1995b . A temporarily red light-insensitive mutant of tomato lacks a light-stable, B-like phytochrome . Plant Physiology 108 , 939 – 947 . Google Scholar Crossref Search ADS WorldCat Wang S , Liu J, Feng Y, Niu X, Giovannoni J, Liu Y. 2008 . Altered plastid levels and potential for improved fruit nutrient content by downregulation of the tomato DDB1-interacting protein CUL4 . The Plant Journal 55 , 89 – 103 . Google Scholar Crossref Search ADS PubMed WorldCat Weller JL , Schreuder ME, Smith H, Koornneef M, Kendrick RE. 2000 . Physiological interactions of phytochromes A, B1 and B2 in the control of development in tomato . The Plant Journal 24 , 345 – 356 . Google Scholar Crossref Search ADS PubMed WorldCat Xu P , Zhang Y, Kang L, Roossinck MJ, Mysore KS. 2006 . Computational estimation and experimental verification of off-target silencing during posttranscriptional gene silencing in plants . Plant Physiology 142 , 429 – 440 . Google Scholar Crossref Search ADS PubMed WorldCat Yuan R , Carbaugh DH. 2007 . Effects of NAA, AVG, and 1-MCP on ethylene biosynthesis, preharvest fruit drop, fruit maturity, and quality of ‘Golden Supreme’ and ‘Golden Delicious’ apples . HortScience 42 , 101 – 105 . Google Scholar OpenURL Placeholder Text WorldCat © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.

Qu, Xiaoxiao; Yan, Min; Zou, Junjie; Jiang, Min; Yang, Kezhen; Le, Jie

doi: 10.1093/jxb/ery158pmid: 29701802

Abstract In rice, and other major cereal grass crops, stomata are arranged in linear files parallel to the long growth axis of leaves. Each stomatal unit comprises two dumbbell-shaped guard cells flanked by two subsidiary cells. These morphological and developmental characteristics enable grass stomata to respond to environmental changes more efficiently. Cyclin-dependent kinases (CDKs) and their cyclin partners co-ordinate cell proliferation and differentiation during the development of multicellular organisms. In contrast to animals, plants have many more types and members of cyclins. In Arabidopsis, four A2-type cyclins (CYCA2s) function redundantly in regulating CDKB1 activity to promote the asymmetric division for stomatal initiation and the symmetric division of guard mother cells (GMCs). In this study, we examine the function of the single A2-type cyclin in rice, OsCYCA2;1, as well the single B1-type CDK, OsCDKB1;1. Cross-species complementation tests demonstrated that OsCYCA2;1 and OsCDKB1;1 could complement the defective stomatal phenotypes of Arabidopsis cyca2 and cdkb1 mutants, but also could suppress DNA endoduplication and cell enlargement. The early asymmetric divisions that establish the stomatal lineages are often missing within the stomatal cell files of OsCYCA2;1-RNAi rice transgenic lines, leading to a significantly reduced stomatal production. However, GMC divisions are not disrupted either in OsCYCA2;1-RNAi or in OsCDKB1;1-RNAi rice transgenic lines as expected. Our results demonstrate a conserved but diverged function and behavior of rice A2-type cyclins, which might be associated with the distinct stomatal development pathways between rice and Arabidopsis. Cyclin-dependent kinases, cell differentiation, cell division, guard mother cells, cyclin, rice, stomata Introduction Stomata are microscopic valves on aerial surfaces of all land plants regulating the shoot–atmosphere gas exchange. Paleobotanical analyses revealed that stomata originated ~400 million years ago, a key evolutionary innovation formed in the early palaeozoic era (Raven, 2002). Despite the fact that the distribution pattern and morphology are highly diversified in different plants, stomata arise in the epidermis after a series of cell divisions, cell fate changes, and cell shape controls. In the past decades, results of molecular genetic studies demonstrated that stomatal development is an accessible system to reveal the evolution of genes and signals involved in plant development (Vatén and Bergmann, 2012; Ran et al., 2013; Chater et al., 2017; Qu et al., 2017). In the dicot model plant Arabidopsis, the earliest stomatal precursor, the meristemoid mother cell (MMC), divides asymmetrically (stomatal entry divisions) producing a smaller cell, the meristemoid, as well as a larger sister cell, the stomatal lineage ground cell (SLGC). Meristemoids normally undergo an additional 1–2 rounds of asymmetric divisions (amplifying divisions) to generate new meristemoids before converting into guard mother cells (GMCs). SLGCs can also undergo asymmetric divisions (spacing divisions) to generate satellite meristemoids. Meristemoids differentiate into GMCs after cell fate change. Then, GMCs divide symmetrically to produce paired young guard cells (GCs). During the final stage of stomatal development, GCs undergo cell differentiation, morphogenesis, and pore formation to form functional stomatal units (Bergmann and Sack, 2007). In contrast to the scattered pattern in dicot Arabidopsis leaves, monocot grass stomata are arranged within linear cell files that parallel the growth axis of the leaf. Stomatal lineage cells initiate at the base of the leaf, and divide asymmetrically to produce two daughter cells, a GMC, and a larger sister cell. At the final stage, GMCs divide symmetrically, producing paired dumbbell-shaped GCs. The stomatal subsidiary cells are produced from cell files flanking the stomatal lineage after asymmetric divisions (Franks and Farquhar, 2007; Liu et al., 2009; Serna, 2011; Raissig et al., 2016). Cyclins form complexes with specific cyclin-dependent kinases (CDKs) to co-ordinate the cell proliferation and differentiation during the development of multicellular organisms (Swenson et al., 1986; Obaya and Sedivy, 2002). Cyclins, acting as the regulator of CDK activity, contribute to the subcellular localization, substrate specificity, and protein stability of the CDK–cyclin complexes (Dewitte and Murray, 2003; Imai et al., 2006; Boudolf et al., 2009; Boruc et al., 2010). A-type cyclins, known as mitotic cyclins, are essential for the mitotic cell cycle. In contrast to animals, plants encode a large family of A-type cyclins that have been classed into A1, A2, and A3 groups (Vandepoele et al., 2002; Dewitte and Murray, 2003; Wang et al., 2004). The Arabidopsis genome has four genes encoding A2-type cyclins. AtCYCA2 genes display tissue- and cell type-specific and overlapping expression patterns, such as in vascular systems and stomatal lineage cells, which are associated with their redundant functions during plant development (Burssens et al., 2000; Imai et al., 2006; Vanneste et al., 2011; Donner and Scarpella, 2013). Mutants of AtCYCA2 genes frequently form unpaired single guard cells (SGCs), a similar defect of the terminal GMC division to that also observed in mutants of AtCDKB1 or AtCDKA;1 genes (Boudolf et al., 2004b; Vanneste et al., 2011; Yang et al., 2014). Overexpression of AtCYCA2:3 as well as AtCDKA;1 at the late stage of stomatal development induced excessive GC subdivisions (Yang et al., 2014). Arabidopsis cdkb1;1 1;2 double mutants and 35S:CDKB1;1.N161 dominant negative plants displayed decreased stomatal production and formation of SGCs, indicating that the activity of CDKB1 is required for both meristemoid asymmetric division and GMC symmetric division in stomatal development. In addition to the function in promoting mitosis, AtCYCA2s form functional complexes with CDKs to modulate the cell cycle transition from the mitotic cycle to the endocycle. Genetic suppression of AtCYCA2 or AtCDKB1 results in enhanced ploidy levels and enlarged pavement cells (PCs; Vanneste et al., 2011). Co-expression of CYCA2;3 and CDKB1;1 induces ectopic cell divisions, limits endoreduplication, and inhibits cell growth (Boudolf et al., 2009). There are at least 49 putative genes predicted to encode rice cyclins, which were classified into nine types based on evolutionary relationships. Eight of these nine types are common between rice and Arabidopsis (Umeda et al., 1999a; Cooper et al., 2003; La et al., 2006). The existence of numerous cyclins implies their diverse regulatory roles in modulating CDK activities during rice development and adaption in response to environmental changes (Cooper et al., 2003; Huang et al., 2008). For example, rice B2-type cyclins, OsCycB2;1 and OsCycB2;2, promote root cell divisions through an association with OsCDKB2;1 (Lee et al., 2003). OsCycH;1 specifically binds to R2 and positively controls CDK and CTD kinase activities to adjust the rate of cell proliferation (Fabian-Marwedel et al., 2002). In this study, we examine the function of the rice single A2-type cyclin OsCYCA2;1 and OsCDKB1;1 in rice development. Our results demonstrate a requirement for OsCYCA2;1 for stomatal and root development. In contrast to its homolog in Arabidopsis, OsCYCA2;1 is exclusively required for the asymmetric entry divisions to produce GMCs at the early stage of stomatal development. In addition, combined with phylogenetic analyses, we are providing new clues for further revealing the evolutionary correlation between cell cycle genes and developmental pathways. Materials and methods Plant materials and growth conditions The Col-0 ecotype of Arabidopsis thaliana L. was used as the wild-type control in the Arabidopsis study. The cdkb1;1 1;2 double mutants were confirmed by PCR-based approaches (Xie et al., 2010). The cyca2;34 double mutants were provided by Steffen Vanneste and Tom Beeckman (Vanneste et al., 2011). Seeds were surface sterilized (40 s) in an aqueous solution of 30% (w/v) hydrogen peroxide and 85% (v/v) ethanol in a volume ratio of 1:4 (v/v), and then sown on the surface of half-strength Murashige and Skoog (MS) medium supplemented with 0.8% agar and 1% sucrose. Plants were grown in a controlled temperature and photoperiod chamber at 22 ± 2 °C and 16 h/8 h light/dark illumination cycles. Oryza sativa L. spp. japonica cultivar Zhonghua 11 was used as the wild-type control and the transformation recipient in the rice study. Rice seeds were soaked in water at 28 °C for 2 d, and then grown in a controlled growth chamber with 30 °C/22 °C day/night temperature cycles, 12 h/12 h light/dark illumination cycles, and 60–70% relative humidity. Plasmid construction and generation of transgenic plants To obtain the construct of gene overexpression, cDNA of OsCDKB1;1 or OsCYCA2;1 was cloned into the pH7WG2D.1 vector by using gateway technology and LR Clonase™ II Enzyme Mix (Invitrogen). The recombinant plasmids were confirmed by DNA sequencing before the transformation into Arabidopsis wild type and mutants. To generate RNAi transgenic plants against OsCDKB1;1 and OsCYCA2;1, the conserved sequences from base pair 530 to 695 of OsCDKB1;1 cDNA and 747 to 979 of OsCYCA2;1 cDNA were amplified and cloned into pTCK303 vector, respectively. These constructs were electroporated into Agrobacterium tumefaciens EHA105 and transformed into rice Zhonghua 11 (Chen et al., 2011). T1 seeds were collected to screen positive transgenic plants by using 50 µg l−1 hygromycin B (Roche). Real-time quantitative PCR (RT-qPCR) was conducted to confirm the expression level of target genes in transgenic plants. The primer sequences used in this study are listed in Supplementary Table S1 at JXB online. Real-time quantitative PCR analysis Rice seedlings were harvested and immediately ground in liquid nitrogen, and the total RNA was isolated using TRNzol reagent (http://www.tiangen.com). The first-strand cDNA was synthesized using a Promega Reverse Transcription kit (http://www.promega.com). RT-qPCRs were performed by using SYBR Premix Ex Taq™ (TaKaRa) with a Corbett RG3000. The OsACTIN2 gene was used as an internal control. The primer sequences are listed in Supplementary Table S1. Yeast two-hybrid assay The full-length cDNA sequences of OsCYCA2;1 and OsCDKB1;1 were amplified using the primers listed in Supplementary Table S1 and cloned into pGBKT7 and pGADT7 vectors (Clontech), respectively. These constructs were transformed into Saccharomyces cerevisiae yeast strain AH109 and selected on SD/-Leu-Trp or SD/-Leu-Trp-His-Ade plates. X-Gal activity was then detected. Bimolecular fluorescence complementation assay For bimolecular fluorescence complementation (BiFC) assays, the full-length cDNAs of OsCYCA2;1 and OsCDKB1;1 were cloned into pSPYCE-35S and pSPYNE-35S vectors, respectively (Walter et al., 2004). These constructs were transformed into A. tumefaciens EH105 and co-injected into tobacco (Nicotiana benthamiana). Images were taken after 3 d using a laser scanning confocal microscope (FV1000-MPE, Olympus). Pull-down assay The OsCYCA2;1 and OsCDKB1;1 sequences were cloned into pET-28a and pGEX4T-1 vectors, respectively. The OsCYCA2;1-pET-28a and OsCDKB1;1-pGEX4T-1 constructs were transformed into the BL21 strain of Escherichia coli. The transformed strains were grown to OD600=0.5 under 37 °C and then placed at 18 °C for 30 min. Fusion proteins were induced with 0.4 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at 18 °C for 20 h. The harvested strains (5000 rpm, 10 min, 4 °C) were re-suspended with ice-cold phosphate-buffered saline (PBS) and lysed by sonication. The lysate was centrifuged at 10000 rpm for 60 min and the supernatant was collected. The glutathione S-transferase (GST)–OsCDKB1;1 supernatant was loaded on glutathione–Sepharose (GE Healthcare) and washed with PBS. The GST–OsCDKB1;1 fusion protein on glutathione–Sepharose was incubated with the His-OsCYCA2;1 supernatant at 4 °C for 2 h. Then the glutathione–Sepharose was washed with PBS and eluted with 10 mM reduced glutathione elution buffer. The samples were loaded on a 12% SDS–polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore) by using a semi-dry blotting system (Bio-Rad), and then incubated with anti-His6 monoclonal antibodies followed by horseradish peroxidase (HRP)-conjugated anti-mouse antibodies. The color reaction was performed using the Pro-Light HRP Kit (Tiangen). Signals were exposed to X-ray films and developed. DAPI staining and DNA content measurement Ten-day-old rice roots were fixed in a mixture of 3:1 (v/v) ethanol and acetic acid for 30 min, then rinsed with distilled water. After staining for 30 min with 2 µg ml−1 DAPI (Roche) in a staining solution (0.1 M sodium phosphate, 1 mM EDTA, 0.1% Triton X-100, pH 7.0), roots were photographed by a fluorescence microscope. The relative fluorescence intensities were measured using ImageJ software (http://imagej.nih.gov/ij/). Flow cytometric analysis About 20–50 mg of fresh tissue were cut into 2–4 mm fragments and then chopped immediately using a razor blade in 1 ml of Galbraith’s buffer (45 mM MgCl2·6H2O, 30 mM sodium citrate, 20 mM MOPS, 0.1% Triton X-100, pH 7.0). The cell culture was collected by gentle pipetting and filtered through a cell strainer. The samples were stained with 2 µg ml−1 DAPI in an ice bath for 30 min before the analysis using a MoFlo-XDP flow cytometer (Beckman) (Dolezel et al., 2007). A total of ~10000 nuclei were measured for each sample. Root semi-thin sections About 1 cm long primary root tips from 6-day-old rice seedlings were harvested and immersed in cold formaldehyde solution. The samples were subjected to a vacuum for 10 min and placed at 4 °C overnight. The materials were washed with 0.1 M PBS (pH 7.2) four times and were fixed in 1% osmic acid for 1 h, followed by a series of dehydration steps, which were performed by using 30, 50, 70, 80, 90, 100, and 100% ethanol (each step lasted 20 min). Ethanol was substituted with 1:1 acetone and ethanol (v/v) as well as pure acetone twice (each step lasted 20 min). Permeation was performed by using a series of 2:1, 1:1, and 1:2 (v/v) of acetone and epoxy (SPURR) mixture solution. Each step lasted for 3 h. After adding pure epoxy for 8 h, samples were embedded and polymerized at 60 °C for 24 h. Semi-thin sections (thickness 1 µm) were obtained by using a Leica UC7 microtome. Before imaging, sections were stained with 0.1% toluidine blue O. Results Evolutionary analysis of A2 cyclins in plants Phylogenetic analysis indicates that homologs of A2-type cyclin are found in lineages that diverged early in the evolution of land plants, before the appearance of stomata. For example, the unicellular green alga Coccomyxa subellipsoidea has a CYCA2 gene in its genome. In the non-vascular land plant moss Physcomitrella patens, stomata are exclusively found on the diploid sporophyte (Chater et al., 2016); there are six putative orthologs of CYCA2. In the vascular dicot plants Arabidopsis, soybean (Glycine max), and alfalfa (Medicago truncatula), the number of CYCA2 genes was four, six, and four, respectively (Fig. 1). In contrast, the rice genome contains only one copy of the CYCA2 gene, Os012g31810, which is predicted to encode OsCYCA2;1 protein consisting of 490 amino acid residues. Multiple sequence alignment reveals that OsCYCA2;1 shows 40.5% amino acid sequence identity with Arabidopsis CYCA2s, and contains a CDK-binding cyclin box, which is highly conserved among mitotic cyclins (Supplementary Fig. S1) (Umeda et al., 1999a). Interestingly, similar to rice, Brachypodium stacei, Brachypodium distachyon, Zea mays (Fig. 1), and many other monocot grasses, Hordeum vulgare, Oropetium thomaeum, Panicum hallii, Sorghum bicolor, Setaria italica, and Setaria viridis, only have 1–2 CYCA2 genes (Supplementary Fig. S2). Fig. 1. Open in new tabDownload slide Phylogenetic tree shows that A2-type cyclin-like proteins are conservatively present in green land plants. The phylogenetic tree was constructed using amino acid sequences of Arabidopsis CYCA2 family members based on Phytozome V12.1, using the Neighbor–Joining method in MEGA4. Bootstrap values for 1000 replicates are given in nodes as percentages. Amino acid sequences were used from Arabidopsis thaliana, Brachypodium distachyon, Brachypodium stacei, Coccomyxa subellipsoidea, Glycine max, Medicago truncatula, Oryza sativa, Physcomitrella patens, Zea mays, and Zostera marina. The seagrass Zostera marina belongs to basal monocots that returned to the sea. The absence of stomata in Z. marina is consistent with the evolutionary loss of entire genes that are required for stomatal development (Olsen et al., 2016). However, like the above grass plants, Z. marina possesses two CYCA2 genes, suggesting that the A2-type cyclin is fundamentally important for plant growth and development, and is not solely linked to stomatal development. The low number of CYCA2 genes in grasses indicates that CYCA2 gene duplication might not be necessary, which is associated with their unique developmental pathways and morphogenesis. Requirement of OsCYCA2;1 for stomatal initiation To elucidate the function of A2-type cyclin in rice development, RNAi transgenic rice lines targeting OsCYCA2;1 were generated. Transcript levels of OsCYCA2;1 in two lines, Ri1 and Ri3, were suppressed to 28% and 61%, respectively, in relation to the level in wild-type rice seedlings (Supplementary Fig. S3A). In rice leaf epidermis, stomata form within the stomatal lineage files following a gradual base to tip maturation pattern; the developing stomata can only be found at the proximal end (base) of the leaf. Unlike in Arabidopsis, GMCs in rice are produced directly by asymmetric entry divisions without the precursor stage of the meristemoid. Each undifferentiated cell close to the base of the leaf divides asymmetrically and generates one smaller GMC and one larger sister cell (Stage 2, upper panel of Fig. 2A). Subsidiary mother cells (SMCs) flanking the GMCs are produced by the cells in the neighboring cell files. The terminal division of GMC produces a pair of immature GCs (Stage 5, middle panel of Fig. 2A). Within the wild-type stomatal lineage cell files, stomatal complexes are spaced by one lobbed PC (Stage 6, lower panel of Fig. 2A). However, in OsCYCA2;1-RNAi transgenic plants, more than two spacing cells were often observed between two neighboring GMCs/stomata within the same cell file (Fig. 2B, C), leading to a decreased stomatal density and stomatal index (Fig. 2D, E). Mutations of Arabidopsis CYCA2 genes caused a failure of GMC division, leading to the formation of aberrant stomatal units (SGCs) (Vanneste et al., 2011). However, the structure and morphology of mature stomata in OsCYCA2;1-RNAi rice transgenic plants are indistinguishable from those of wild-type stomata, indicating that the subsequent GMC symmetric divisions as well the subsidiary cell asymmetric divisions are not interrupted by the suppression of OsCYCA2;1 (lower panels in Fig. 2A and B). Taken together, the above observations indicate that OsCYCA2;1 is essentially required for the asymmetric entry division during stomatal initiation at the early stage of stomatal development, but not for the terminal GMC symmetric divisions and subsidiary cell asymmetric divisions. Fig. 2. Open in new tabDownload slide Suppression of OsCYCA2;1 causes defective cell division in rice. (A, B) Differential interference contrast micrographs of epidermal cells from 6-day-old rice seedlings grown in darkness. Stomata were initiated at the proximal end (base) of young leaves. Asymmetric cell divisions produce a smaller GMC and one larger sister cell (Stage 2, upper panels). The terminal symmetric division of the GMC produces a pair of immature GCs (Stage 5, middle panels). Mature stomatal complexes (a pair of dumbbell-shaped GCs and two flanking SCs) are spaced by one pavement cell (PC) (Stage 6, lower panels). Arrowheads indicate the GMCs or stomatal complexes. Asterisks indicate the PCs that separate stomata in the same cell file. Scale bar=20 µm. (C) The numbers of PCs between two adjacent stomata within the same cell file are often increased in RNAi lines. (D, E) Leaf stomatal density and index of two OsCYCA2;1-RNAi lines and the wild type (WT; n=12). Data represent the mean ±SD. Asterisks indicate a significant difference from WT controls (Student’s t-test, **P<0.01). (F–H) Flow cytometric analysis of nuclei in shoot cells. (I) Quantitative analysis of DAPI fluorescence. OsCYCA2;1-RNAi transgenic lines have a higher average 4C DNA content than the WT. For each line, ~10000 cell nuclei were measured. It has been demonstrated that Arabidopsis CYCA2s not only promote cell proliferation but also negatively regulate endocycle onset (Imai et al., 2006; Yoshizumi et al., 2006; Vanneste et al., 2011). Flow cytometric analysis showed that in wild-type rice, only 6% of cells showed a 4C DNA content, whereas most cells were 2C (diploid). However, in OsCYCA2;1-RNAi transgenic rice lines Ri1 and Ri3, the fraction of 4C cells markedly increased to 36% and 18%, respectively. Higher ploidy levels, like in Arabidopsis cyca2 mutants (8C, 16C, 32C), were barely detectable in rice OsCYCA2;1-RNAi plants (Fig. 2F–I). OsCYCA2;1 complements epidermal defects of Arabidopsis cyca2 mutants The Arabidopsis epidermis is an ideal system to identify gene functions in plant development programs and morphogenetic patterns. To identify further the function of OsCYCA2;1 in epidermal development, OsCYCA2;1 coding sequences driven by 35S promoters were transformed into Arabidopsis cyca2;34 mutants. Compared with the wild type, cyca2;34 mutants display enlarged pavement cells and enhanced ploidy levels. Cross-species expression of OsCYCA2;1 (line #7) inhibits the abnormal PC enlargement in cyca2;34 epidermis, to a cell size even smaller than in the wild type (Fig. 3A–D). Moreover, revealed by flow cytometric analysis, expression of OsCYCA2;1 is able to inhibit the high DNA ploidy levels in cyca2;34 mutants (Fig. 3E; Supplementary Fig. S4). This result is consistent with the previous findings that overexpression of CYCA2 genes could restrain endoreduplication in Arabidopsis (Imai et al., 2006; Boudolf et al., 2009). Fig. 3. Open in new tabDownload slide Cross-species expression of OsCYCA2;1 complements the epidermal defects of Arabidopsis cyca2;34 mutants. (A–C) Differential interference contrast micrographs of cotyledon epidermal cells of 14-day-old Arabidopsis seedlings of the Col, cyca2;34, and cyca2;34 harboring 35S:OsCYCA2;1, Line #7. An arrow points to a single guard cell (SGC). Representative pavement cells (PCs) are traced with dashed lines. Scale bar=50 µm. (D) Comparison of PC area (n=30). (E) Proportions of cells with different ploidies. (F and G) Stomatal density and index. The diagonal line-filled box indicates the SGCs. Data in (D, F, G) represent the mean ±SD. Asterisks indicate a significant difference from Col wild-type controls (Student’s t-test, **P<0.01). In cyca2;34 mutants, ~10% of GMCs failed to divide symmetrically and formed into SGCs (Fig. 3B, arrow). Strikingly, OsCYCA2;1 expression can fully rescue the defective GMC division in cyca2;34 epidermis, suggesting that rice OsCYCA2;1 remains a conserved function in promoting the GMC symmetric divisions. In addition, expression of OsCYCA2;1 in the cyca2;34 mutant background induced formation of excessive stomata, reflected by an increased stomatal density and stomatal index (Fig. 3F, G). In another 35S:OsCYCA2;1 cyca2;34 transgenic line (line #8), the relative transcript level of OsCYCA2;1 is much lower than in line #7; the defective GMC division and reduced stomatal production are partially rescued, indicating that OsCYCA2;1 quantitatively promotes stomatal development depending on its expression level (Supplementary Fig. S5). Taken together, our results of cross-species complement tests demonstrate the conserved abilities of OsCYCA2;1 in limiting cell endoreduplication and PC size, as well as in rescuing cyca2;34 defective asymmetric entry divisions (for stomatal initiation) and symmetric GMC divisions (for guard cell formation), despite OsCYCA2;1 being functionally required only for stomatal entry divisions in rice. OsCYCA2;1 is also required for cell division and differentiation in roots OsCYCA2;1 is preferentially expressed in proliferating tissues. Besides in the base dividing zone (proximal end) of leaves, a higher transcript level of OsCYCA2;1 is found in rice root tips, implying that a similar OsCYCA2;1 regulatory mechanism exists in rice roots (Supplementary Fig. 3B). Therefore, we probed the impact of down-regulated expression of OsCYCA2;1 on root growth. As shown in Fig. 4, the overall growth of shoots and roots in 10-day-old OsCYCA2;1-RNAi transgenic seedlings is much less than in the wild type (Fig. 4A–C). To determine whether the root growth defects arose from a defective cell proliferation, we compared the root longitudinal sections of the wild type and the RNAi line. The shorter meristematic zone in OsCYCA2;1-RNAi is correlated with a considerably fewer number of cells within its meristematic zone (Fig. 4D–F). Similarly, Arabidopsis cyca2;34 mutants exhibit a short meristematic zone and fewer cells than Col wild type. Ectopic expression of OsCYCA2;1 restored the length of and cell number within the meristematic zone to the wild-type level (Fig. 4G–I), supporting that OsCYCA2;1 is an evolutionarily conserved regulator that is required for cell proliferation in roots. Fig. 4. Open in new tabDownload slide Suppression of CYCA2 expression causes defective cell proliferation within the meristematic zone of roots. (A) Ten-day-old OsCYCA2;1-RNAi and wild-type (WT) rice seedlings. Scale bar=1 cm. (B, C) The length of shoots and primary roots (n=24). (D) Longitudinal sections of the primary root tips. Scale bar=100 µm. (E, F) Length and cell number of the meristematic zone in roots (n=20). (G) Propidium iodide-stained images of Arabidopsis root tips. Scale bar=100 µm. (H, I) Length and cell number of the meristematic zone in Arabidopsis roots (n=20). Double-headed arrows in (D, G) indicate the extent of the meristematic zone. Data in (B, C, E, F, H, I) represent the mean ±SD. Asterisks indicate a significant difference from WT controls (Student’s t-test, **P<0.01, *P<0.05). (This figure is available in colour at JXB online.) By means of flow cytometry approaches, we found that, in contrast to the 6% of 4C cells in the wild type, the fractions of cells with 4C DNA content in OsCYCA2;1-RNAi lines Ri1 and Ri3 are dramatically increased to 32% and 15%, respectively (Fig. 5A–D). Moreover, the relative expression levels of an S-phase-specific gene, PCNA, and a M-phase cyclin gene, CYCB2;1, were suppressed in OsCYCA2;1-RNAi plants (Fig. 5E). Consistently, epidermal cells in the maturation zone of OsCYCA2;1-RNAi roots showed stronger DAPI fluorescent signals than in the wild type (Fig. 5F–H). Quantitative analysis of the DAPI fluorescence intensities further confirmed that a higher DNA level (~2-fold) is present in OsCYCA2;1-RNAi roots (Fig. 5I). The higher DNA content in OsCYCA2;1-RNAi root cells might be due to delayed or arrested G2 to M transition, a result supporting the idea that OsCYCA2;1 is required for cell mitosis. Fig. 5. Open in new tabDownload slide OsCYCA2;1 is required for rice root cell mitosis. (A–C) Profiles of distribution of cells with different DNA content after flow cytometric analysis. Roots of OsCYCA2;1-RNAi lines Ri1 (B) and Ri3 (C) have more 4C cells than wild-type (WT) roots (A). (D) Quantitative analysis of the cell DNA ploidy levels. (E) Relative expression levels of PCNA and CYCB2;1 in OsCYCA2;1-RNAi lines and WT roots. (F–H) DAPI staining of the epidermal cells in the maturation zone of the WT (F), OsCYCA2;1-RNAi line Ri1 (G), and Ri3 (H). Scale bar=50 µm. (I) Quantitative analysis of DAPI fluorescence revealed that OsCYCA2;1-RNAi transgenic lines have a higher average DNA content than the WT. Data represent the mean ±SD. Asterisks indicate a significant difference from WT controls (Student’s t-test, **P<0.01). (This figure is available in colour at JXB online.) OsCYCA2;1 conservatively interacts with OsCDKB1;1 CYCA2s play their regulatory roles through interacting with multiple CDKs, such as by forming CYCA2;3–CDKB1;1 or CYCA2;3–CDKA;1 protein complexes. Arabidopsis AtCYCA2;3 interacts with AtCDKB1;1 to form a functional complex which promotes the formation of a two-celled stoma and prevents entry into the endocycle program (Boudolf et al., 2009; Vanneste et al., 2011). According to the sequence blasting results in the rice genome, Os01g67160 encodes the putative OsCDKB1;1. The deduced amino acid sequence of OsCDKB1;1 shares 88.5% sequence identity with the Arabidopsis CDKB1s. A B1-type-specific cyclin interaction motif ‘PPTALRE’ is highly conserved in rice OsCDKB1;1 (Supplementary Fig. S6). Yeast two-hybrid assays showed that OsCYCA2;1 can interact with OsCDKB1;1 (Fig. 6A). Consistently, pull-down assays verified the direct protein interaction between OsCYCA2;1 and OsCDKB1;1 (Fig. 6B). To determine the subcellular localization, OsCYCA2;1 or OsCDKB1;1 fused with GFP were transiently expressed in tobacco (N. benthamiana) leaves. The fluorescent signals from OsCYCA2;1–GFP were exclusively detected in nuclei, while OsCDKB1;1–GFP was found in both the cytoplasm and nuclei (Fig. 6C). BiFC analysis confirmed that OsCYCA2;1 directly interacts with OsCDKB1;1 in nuclei (Fig. 6D). These results suggest that OsCYCA2;1 may act as a conserved activator regulating the activity of OsCDKB1;1 kinase in rice. Fig. 6. Open in new tabDownload slide OsCYCA2;1 directly interacts with OsCDKB1;1. (A) Yeast two-hybrid assay. (B) Protein pull-down assay. (C) Transient expression of OsCDKB1;1–GFP and OsCYCA2;1–GFP in tobacco leaves. Scale bar=20 µm. (D) Bimolecular fluorescence complementation assay shows that OsCDKB1;1 interacts with OsCYCA2;1 in nuclei. Scale bar=20 µm. (This figure is available in colour at JXB online.) Suppression of OsCDKB1;1 has no obvious effects on rice development To determine whether OsCDKB1;1, like its partner OsCYCA2;1, is required for rice development, we generated and selected two OsCDKB1;1-RNAi transgenic lines, Ri2 and Ri3, in which OsCDKB1;1 transcript levels were significantly suppressed (Supplementary Fig. S7A). However, the overall growth of these two transgenic lines is comparable with that of the untransformed controls (Supplementary Fig. S7B–D). Longitudinal sections of roots demonstrate that the suppression of OsCDKB1;1 has no significant impact on cell numbers of the root meristematic zone (Supplementary Fig. S7E–G). In addition, we found that neither the stomatal production (stomatal density) nor the stomatal complex morphology has been affected in OsCDKB1;1-RNAi transgenic lines (Supplementary Fig. S7H–J). Flow cytometric assays also indicate that DNA ploidy levels were not changed in either the roots or shoots of OsCDKB1;1-RNAi (Supplementary Fig. S8A–H). Consistent with this, the expression of S-phase PCNA and M-phase cyclin CYCB2;1 was not different between wild-type and OsCDKB1;1-RNAi transgenic plants (Supplementary Fig. S8I). Taken together, it seems that cell division was not interrupted by down-regulation of the OsCDKB1;1 transcript level in transgenic rice, though we could not exclude the possibility that the remaining activity of OsCDKB1;1 protein is sufficient to function. OsCYCA2;1 and OsCDKB1;1 enable complementation of Arabidopsis cdkb1;1 1;2 Arabidopsis cdkb1;1 1;2 mutants, like the cyca2;34 mutants, display a decreased stomatal production, formation of SGCs, enlarged PCs, and increased cell ploidy levels (Boudolf et al., 2004a; Xie et al., 2010). Introduction of OsCDKB1;1 fully complements the impaired GMC division in cdkb1;1 1;2, and restores stomatal production, indicating that OsCDKB1;1 has the ability to promote both symmetric and asymmetric division. Meanwhile, expression of OsCKDB1;1 could efficiently prevent the occurrence of enlarged PCs and increased DNA ploidy levels in cdkb1;1 1;2 mutants (Fig. 7; Supplementary Figs S9, S10). Fig. 7. Open in new tabDownload slide Ectopic expression of OsCYCA2;1 and OsCDKB1;1 complements Arabidopsis cdkb1;1 1;2 mutant phenotypes. (A–D) DIC images of the epidermis of 14-day-old Arabidopsis cotyledons. Arrows indicate the formation of SGCs. Representative PCs are traced with dashed lines. Scale bar=50 µm. (E–G) Comparison of stomatal density, stomatal index, and area of PCs from the cotyledons. The diagonal line-filled box indicates the fraction of SGCs. Data represent the mean ±SD (n=24). Asterisks indicate a significant difference from Col after Student’s t-test, **P<0.01. (H) Proportions of cells with different ploidies. (I–L) Results from the flow cytometric analysis; ~10000 cell nuclei were measured for each sample. Interestingly, the defective stomatal production, impaired GMC division, and abnormal cell enlargement and DNA levels in cdkb1;1 1;2 could be partially rescued by overexpression of OsCYCA2;1 (Fig. 7; Supplementary Figs S9, S10). It is therefore possible that OsCDKB1;1 and OsCYCA2;1 have evolved from the common ancestor genes with Arabidopsis AtCDKB1 genes and AtCYCA2 genes, even though the developmental pathways of the two species have been diverged. Discussion The control of cell division and differentiation is the core of the development and morphogenesis of multicellular organisms. Cyclins, known as conserved activators for the activity of CDKs, play a crucial regulatory role in cell cycle progression in diverse species. The functional pathway of CYCA2s and CDKB1s has been well investigated in the model plant Arabidopsis (Boudolf et al., 2004b, 2009; Imai et al., 2006; Xie et al., 2010; Vanneste et al., 2011; Yang et al., 2014). In this study, we generated RNAi transgenic rice lines and performed cross-species complement tests to explore the function of the single rice A2-type cyclin, OsCYCA2;1, as well the single rice B1-type CDK, OsCDKB1;1. Cross-species expression of OsCYCA2;1 or OsCDKB1;1 enables rescue of the defective asymmetric entry divisions for stomatal initiation and GMC symmetric divisions for GC production in Arabidopsis cyca2;34 and/or cdkb1;1 1;2 mutants, suggesting that both OsCYCA2;1 and OsCDKB1;1 might have evolved from the common ancestor genes with Arabidopsis. In Arabidopsis, asymmetric divisions generated the early stomatal precursor cells, meristemoids. Meristemoids then differentiate into GMCs after a cell fate change. In grasses, GMCs are created directly by stomata initiating asymmetric divisions (entry division) without the prior precursor stage of meristemoid. Orthologs of Arabidopsis stomatal transcriptional regulators SPCH, MUTE, FAMA, ICE1, and SCRM2 have been identified in grasses (Liu et al., 2009; Vatén and Bergmann, 2012; Ran et al., 2013; Chen et al., 2017). Instead of a single copy in Arabidopsis, the rice genome has duplicated SPCH genes, OsSPCH1 and OsSPCH2. Similar to the weak allele of Arabidopsis spch, the rice mutant osspch2 exhibits a reduced number of stomata (Liu et al., 2009). In Arabidopsis, SPCH heterodimerizes with SCRM/ICE1 or AtSCRM2 to promote stomatal lineage initiation (Kanaoka et al., 2008; Horst et al., 2015). In contrast, in the grass B. distachyon, BdICE1 and BdSCRM2 show a functional diversity in regulating stomatal pattern and morphology (Raissig et al., 2016). It is already known that Arabidopsis SPCH activity or stability is modulated by multiple kinases, including MPKs, GSK3/BIN2, and CDKs (Lampard et al., 2009; Gudesblat et al., 2012; Kim et al., 2012; Le et al., 2014). Phosphorylation of Ser186 of SPCH, which might be the target residue of CDKs, positively regulates stomata production (Yang et al., 2015). Thus, it will be interesting to establish if there is a conserved regulatory mechanism between CDK–cyclin and SPCH-ICE1/SCRM2 in grasses. Besides the involvement in stomatal initiation, Arabidopsis AtCYCA2s and CDKB1s are synergistically required for the GMC symmetric division that is a prerequisite for the final stomatal development (Vanneste et al., 2011). Arabidopsis and rice share common GMC–GC processing; GMCs divide symmetrically to produce the paired GCs of stomata, though the GC shapes are distinct. Thus, a role for OsCDKB1–OsCYCA2 in rice GMC divisions has been highly expected. However, suppression of OsCYCA2;1 transcription in rice by RNAi does not affect the rice GMC symmetric division. It has been identified that transcription of CDKB1;1, CYCA2;3, and CDKA;1 in Arabidopsis is repressed by FOUR LIPS(FLP)/MYB88 MYB transcription factors during the GMC–GC transition stages (Xie et al., 2010; Vanneste et al., 2011; Yang et al., 2014). FAMA, like FLP/MYB, also binds to the CDKB1;1 promoter (Hachez et al., 2011) to limit the GMC divisions to one. In contrast to the tumor-like phenotype in Arabidopsis fama-1 mutants, the loss-of-function rice allele osfama-1 did not undergo excessive division except the appearance of misshaped GCs and showing a lack of stomatal pores (Liu et al., 2009). These observations suggest that GMC–GC differentiation is uncoupled from GMC division, in which the putative downstream FAMA/FLP/MYB88, CDKB1;1, and CYCA2;1 are not essential. According to the phylogenetic analysis, CYCA2 and CDKB1 widely exist in diverse plant species, both in plants bearing stomata and in plants lacking stomata, indicating that CYCA2 and CDKB1 might function as fundamental regulators of the mitotic cell cycle, as well as outside stomatal development. High expression of OsCYCA2;1 is associated with a high activity of cell proliferation, such as in the proximal end of leaves or root tips (Supplementary Fig. S3B). Endoreduplication often occurs in cell types that undergo specialized differentiation. In Arabidopsis, the highly differentiated epidermal cells, such as mature PCs and trichomes, usually undergo multiple rounds of DNA replication without mitosis, resulting in polyploid cells (Burssens et al., 2000). In contrast to the differentiated cells in Arabidopsis, polyploid cells in rice can only be found in the endosperm (Sabelli and Larkins, 2009). In Drosophila, it has been reported that cyclin A is one of the key components of chromosomal DNA replication that prevents re-initiation of DNA replication. Overexpression of Drosophila cyclin A caused a reduction in ploidy levels and inhibition of the endocycle (Hayashi and Yamaguchi, 1999). Here we found that the fraction of 4C cells remarkably increased in OsCYCA2;1-RNAi transgenic plants, while most cells keep a 2C DNA content. However, the expression levels of an S-phase gene PCNA and a M-phase gene CYCB2;1 were suppressed in OsCYCA2;1-RNAi plants. Therefore, we speculated that the increase of 4C cells might be caused by the arrested G2 to M transition, similar to the observation in the rice knockdown lines of OsCDKB2;1 (Endo et al., 2012). However, the OsCDKB1;1-RNAi transgenic rice plants, in which OsCDKB1;1 transcript levels were significantly decreased, display phenotypes comparable with wild-type rice seedlings regarding the stomatal density, root cell division, and DNA content. CDKBs are plant-specific cyclin-dependent kinases that can be subdivided into two groups according to the different cyclin-binding motifs, namely ‘PPTALRE’ in CDKB1s and ‘PPTTLRE’ in CDKB2s (Joubès et al., 2001). In Arabidopsis, each CDKB1 and CDKB2 subgroup contains two members (Vandepoele et al., 2002). It has been predicted that the rice genome has a single CDKB1 gene and a single CDKB2 gene, encoding OsCDKB1;1 and OsCDKB2;1, respectively (Supplementary Fig. S11). However, the amino acid sequence alignment revealed that rice OsCDKB1;1 and OsCDKB2;1 share the same ‘PPTALRE’ cyclin-binding motif (Supplementary Fig. S12). Expression of OsCDKB2;1 has been detected in the dividing region of the rice root apex (Umeda et al., 1999a). Transcription of rice OsCDKB2;1 is abundant during the G2 to M phase. Knockdown of the OsCDKB2;1 gene in rice induces an increase of the 4C cell population (Umeda et al., 1999b; Endo et al., 2012). In addition, OsCDKB2;1 promotes cell division in the root meristem probably through the association with OsCYCB2s (Lee et al., 2003). Thus, we cannot rule out that OsCDKB1;1 might function redundantly with OsCDKB2;1, such as forming active CDK–cyclin complexes via binding to the same type cyclins (i.e. OsCYCA2;1). Previous in situ hybridization results showed that both OsCDKA;1 and OsCDKA;2 are expressed in dividing root cells of rice (Umeda et al., 1999b). Thus, further characterization of rice CDK–cyclin pairing and activity can help to reveal the regulatory mechanisms of cell division and differentiation during rice development. Supplementary data Supplementary data are available at JXB online. Table S1. List of primers used in this study. Fig. S1. Amino acid sequence comparison of A2-type cyclins from rice and Arabidopsis. Fig. S2. In contrast to dicot Arabidopsis, only one or two copies of genes encoding CYCA2 are found in monocot grasses. Fig. S3. Relative expression of OsCYCA2;1 in rice RNAi transgenic plants and in different tissues of wild-type plants. Fig. S4. Overexpression of rice OsCYCA2;1 suppresses the enhanced endoreduplication levels in Arabidopsis cyca2;34. Fig. S5. Correlation between stomatal phenotypes and OsCYCA2;1 expression levels in Arabidopsis cyca2;34 mutants harboring OsCYCA2;1. Fig. S6. Comparison of the amino acid sequence of OsCDKB1;1 with that of Arabidopsis CDKB1;1 and CDKB1;2. Fig. S7. Suppression of OsCDKB1;1 has no obvious impact on rice root and stomatal development. Fig. S8. Suppression of OsCDKB1;1 has no obvious impact on the distribution of DNA ploidy. Fig. S9. Expression analysis of OsCDKB1;1-OE and OsCYCA2;1-OE transgenic plants in Arabidopsis cdkb1 mutants. Fig. S10. Ploidy distribution analysis of Arabidopsis cdkb1;1 1;2 mutants carrying rice OsCDKB1;1 or OsCYCA2;1 genes. Fig. S11. Phylogenetic tree of CDKB1 and CDKB2 in monocots. Fig. S12. CDKB1 and CDKB2 contain the same cyclin-binding domain in most monocots. Abbreviations: Abbreviations: CYCA2 CYCLIN A2 CDKB1 CYCLIN-DEPENDENT KINASE B1 GC guard cell GMC guard mother cell PC pavement cell SGC single guard cell. Acknowledgements We thank Kang Chong for providing the pTCK303 vector. This work was supported by the Natural Science Foundation of China grants to JL (31271463, 31471362, and 31771515), to JZ (31470362), and to KY (31471285), and the National Transgenic Program of China to JL (2013ZX08009-003-002). References Bergmann DC , Sack FD. 2007 . Stomatal development . Annual Review of Plant Biology 58 , 163 – 181 . Google Scholar Crossref Search ADS PubMed WorldCat Boruc J , Van den Daele H, Hollunder J, Rombauts S, Mylle E, Hilson P, Inzé D, De Veylder L, Russinova E. 2010 . Functional modules in the Arabidopsis core cell cycle binary protein–protein interaction network . The Plant Cell 22 , 1264 – 1280 . Google Scholar Crossref Search ADS PubMed WorldCat Boudolf V , Barrôco R, Engler Jde A, Verkest A, Beeckman T, Naudts M, Inzé D, De Veylder L. 2004a. B1-type cyclin-dependent kinases are essential for the formation of stomatal complexes in Arabidopsis thaliana . The Plant Cell 16 , 945 – 955 . Google Scholar Crossref Search ADS PubMed WorldCat Boudolf V , Lammens T, Boruc J, et al. 2009 . CDKB1;1 forms a functional complex with CYCA2;3 to suppress endocycle onset . Plant Physiology 150 , 1482 – 1493 . Google Scholar Crossref Search ADS PubMed WorldCat Boudolf V , Vlieghe K, Beemster GT, Magyar Z, Torres Acosta JA, Maes S, Van Der Schueren E, Inzé D, De Veylder L. 2004b. The plant-specific cyclin-dependent kinase CDKB1;1 and transcription factor E2Fa-DPa control the balance of mitotically dividing and endoreduplicating cells in Arabidopsis . The Plant Cell 16 , 2683 – 2692 . Google Scholar Crossref Search ADS PubMed WorldCat Burssens S , de Almeida Engler J, Beeckman T, Richard C, Shaul O, Ferreira P, Van Montagu M, Inzé D. 2000 . Developmental expression of the Arabidopsis thaliana CycA2;1 gene . Planta 211 , 623 – 631 . Google Scholar Crossref Search ADS PubMed WorldCat Chater CC , Caine RS, Tomek M, et al. 2016 . Origin and function of stomata in the moss Physcomitrella patens . Nature Plants 2 , 16179 . Google Scholar Crossref Search ADS PubMed WorldCat Chater CCC , Caine RS, Fleming AJ, Gray JE. 2017 . Origins and evolution of stomatal development . Plant Physiology 174 , 624 – 638 . Google Scholar Crossref Search ADS PubMed WorldCat Chen N , Xu Y, Wang X, Du C, Du J, Yuan M, Xu Z, Chong K. 2011 . OsRAN2, essential for mitosis, enhances cold tolerance in rice by promoting export of intranuclear tubulin and maintaining cell division under cold stress . Plant, Cell and Environment 34 , 52 – 64 . Google Scholar Crossref Search ADS WorldCat Chen ZH , Chen G, Dai F, Wang Y, Hills A, Ruan YL, Zhang G, Franks PJ, Nevo E, Blatt MR. 2017 . Molecular evolution of grass stomata . Trends in Plant Science 22 , 124 – 139 . Google Scholar Crossref Search ADS PubMed WorldCat Cooper B , Hutchison D, Park S, Guimil S, Luginbühl P, Ellero C, Goff SA, Glazebrook J. 2003 . Identification of rice (Oryza sativa) proteins linked to the cyclin-mediated regulation of the cell cycle . Plant Molecular Biology 53 , 273 – 279 . Google Scholar Crossref Search ADS PubMed WorldCat Dewitte W , Murray JA. 2003 . The plant cell cycle . Annual Review of Plant Biology 54 , 235 – 264 . Google Scholar Crossref Search ADS PubMed WorldCat Dolezel J , Greilhuber J, Suda J. 2007 . Estimation of nuclear DNA content in plants using flow cytometry . Nature Protocols 2 , 2233 – 2244 . Google Scholar Crossref Search ADS PubMed WorldCat Donner TJ , Scarpella E. 2013 . Transcriptional control of early vein expression of CYCA2; 1 and CYCA2;4 in Arabidopsis leaves . Mechanisms of Development 130 , 14 – 24 . Google Scholar Crossref Search ADS PubMed WorldCat Endo M , Nakayama S, Umeda-Hara C, Ohtsuki N, Saika H, Umeda M, Toki S. 2012 . CDKB2 is involved in mitosis and DNA damage response in rice . The Plant Journal 69 , 967 – 977 . Google Scholar Crossref Search ADS PubMed WorldCat Fabian-Marwedel T , Umeda M, Sauter M. 2002 . The rice cyclin-dependent kinase-activating kinase R2 regulates S-phase progression . The Plant Cell 14 , 197 – 210 . Google Scholar Crossref Search ADS PubMed WorldCat Franks PJ , Farquhar GD. 2007 . The mechanical diversity of stomata and its significance in gas-exchange control . Plant Physiology 143 , 78 – 87 . Google Scholar Crossref Search ADS PubMed WorldCat Gudesblat GE , Schneider-Pizoń J, Betti C, et al. 2012 . SPEECHLESS integrates brassinosteroid and stomata signalling pathways . Nature Cell Biology 14 , 548 – 554 . Google Scholar Crossref Search ADS PubMed WorldCat Hachez C , Ohashi-Ito K, Dong J, Bergmann DC. 2011 . Differentiation of Arabidopsis guard cells: analysis of the networks incorporating the basic helix–loop–helix transcription factor, FAMA . Plant Physiology 155 , 1458 – 1472 . Google Scholar Crossref Search ADS PubMed WorldCat Hayashi S , Yamaguchi M. 1999 . Kinase-independent activity of Cdc2/cyclin A prevents the S phase in the Drosophila cell cycle . Genes to Cells 4 , 111 – 122 . Google Scholar Crossref Search ADS PubMed WorldCat Horst RJ , Fujita H, Lee JS, Rychel AL, Garrick JM, Kawaguchi M, Peterson KM, Torii KU. 2015 . Molecular framework of a regulatory circuit initiating two-dimensional spatial patterning of stomatal lineage . PLoS Genetics 11 , e1005374 . Google Scholar Crossref Search ADS PubMed WorldCat Huang YW , Tsay WS, Chen CC, Lin CW, Huang HJ. 2008 . Increased expression of the rice C-type cyclin-dependent protein kinase gene, Orysa;CDKC;1, in response to salt stress . Plant Physiology and Biochemistry 46 , 71 – 81 . Google Scholar Crossref Search ADS PubMed WorldCat Imai KK , Ohashi Y, Tsuge T, Yoshizumi T, Matsui M, Oka A, Aoyama T. 2006 . The A-type cyclin CYCA2;3 is a key regulator of ploidy levels in Arabidopsis endoreduplication . The Plant Cell 18 , 382 – 396 . Google Scholar Crossref Search ADS PubMed WorldCat Joubès J , Lemaire-Chamley M, Delmas F, Walter J, Hernould M, Mouras A, Raymond P, Chevalier C. 2001 . A new C-type cyclin-dependent kinase from tomato expressed in dividing tissues does not interact with mitotic and G1 cyclins . Plant Physiology 126 , 1403 – 1415 . Google Scholar Crossref Search ADS PubMed WorldCat Kanaoka MM , Pillitteri LJ, Fujii H, Yoshida Y, Bogenschutz NL, Takabayashi J, Zhu JK, Torii KU. 2008 . SCREAM/ICE1 and SCREAM2 specify three cell-state transitional steps leading to Arabidopsis stomatal differentiation . The Plant Cell 20 , 1775 – 1785 . Google Scholar Crossref Search ADS PubMed WorldCat Kim TW , Michniewicz M, Bergmann DC, Wang ZY. 2012 . Brassinosteroid regulates stomatal development by GSK3-mediated inhibition of a MAPK pathway . Nature 482 , 419 – 422 . Google Scholar Crossref Search ADS PubMed WorldCat La H , Li J, Ji Z, Cheng Y, Li X, Jiang S, Venkatesh PN, Ramachandran S. 2006 . Genome-wide analysis of cyclin family in rice (Oryza sativa L.) . Molecular Genetics and Genomics 275 , 374 – 386 . Google Scholar Crossref Search ADS PubMed WorldCat Lampard GR , Lukowitz W, Ellis BE, Bergmann DC. 2009 . Novel and expanded roles for MAPK signaling in Arabidopsis stomatal cell fate revealed by cell type-specific manipulations . The Plant Cell 21 , 3506 – 3517 . Google Scholar Crossref Search ADS PubMed WorldCat Le J , Zou J, Yang K, Wang M. 2014 . Signaling to stomatal initiation and cell division . Frontiers in Plant Science 5 , 297 . Google Scholar Crossref Search ADS PubMed WorldCat Lee J , Das A, Yamaguchi M, Hashimoto J, Tsutsumi N, Uchimiya H, Umeda M. 2003 . Cell cycle function of a rice B2-type cyclin interacting with a B-type cyclin-dependent kinase . The Plant Journal 34 , 417 – 425 . Google Scholar Crossref Search ADS PubMed WorldCat Liu T , Ohashi-Ito K, Bergmann DC. 2009 . Orthologs of Arabidopsis thaliana stomatal bHLH genes and regulation of stomatal development in grasses . Development 136 , 2265 – 2276 . Google Scholar Crossref Search ADS PubMed WorldCat Obaya AJ , Sedivy JM. 2002 . Regulation of cyclin–Cdk activity in mammalian cells . Cellular and Molecular Life Sciences 59 , 126 – 142 . Google Scholar Crossref Search ADS PubMed WorldCat Olsen JL , Rouzé P, Verhelst B, et al. 2016 . The genome of the seagrass Zostera marina reveals angiosperm adaptation to the sea . Nature 530 , 331 – 335 . Google Scholar Crossref Search ADS PubMed WorldCat Qu X , Peterson KM, Torii KU. 2017 . Stomatal development in time: the past and the future . Current Opinion in Genetics & Biology 45 , 1 – 9 . Google Scholar Crossref Search ADS WorldCat Raissig MT , Abrash E, Bettadapur A, Vogel JP, Bergmann DC. 2016 . Grasses use an alternatively wired bHLH transcription factor network to establish stomatal identity . Proceedings of the National Academy of Sciences, USA 113 , 8326 – 8331 . Google Scholar Crossref Search ADS WorldCat Ran JH , Shen TT, Liu WJ, Wang XQ. 2013 . Evolution of the bHLH genes involved in stomatal development: implications for the expansion of developmental complexity of stomata in land plants . PLoS One 8 , e78997 . Google Scholar Crossref Search ADS PubMed WorldCat Raven JA . 2002 . Selection pressures on stomatal evolution . New Phytologist 153 , 371 – 386 . Google Scholar Crossref Search ADS WorldCat Sabelli PA , Larkins BA. 2009 . The development of endosperm in grasses . Plant Physiology 149 , 14 – 26 . Google Scholar Crossref Search ADS PubMed WorldCat Serna L . 2011 . Stomatal development in Arabidopsis and grasses: differences and commonalities . International Journal of Developmental Biology 55 , 5 – 10 . Google Scholar Crossref Search ADS PubMed WorldCat Swenson KI , Farrell KM, Ruderman JV. 1986 . The clam embryo protein cyclin A induces entry into M phase and the resumption of meiosis in Xenopus oocytes . Cell 47 , 861 – 870 . Google Scholar Crossref Search ADS PubMed WorldCat Umeda M , Iwamoto N, Umeda-Hara C, Yamaguchi M, Hashimoto J, Uchimiya H. 1999a. Molecular characterization of mitotic cyclins in rice plants . Molecular Genetics and Genomics 262 , 230 – 238 . Google Scholar Crossref Search ADS WorldCat Umeda M , Umeda-Hara C, Yamaguchi M, Hashimoto J, Uchimiya H. 1999b. Differential expression of genes for cyclin-dependent protein kinases in rice plants . Plant Physiology 119 , 31 – 40 . Google Scholar Crossref Search ADS PubMed WorldCat Vandepoele K , Raes J, De Veylder L, Rouzé P, Rombauts S, Inzé D. 2002 . Genome-wide analysis of core cell cycle genes in Arabidopsis . The Plant Cell 14 , 903 – 916 . Google Scholar Crossref Search ADS PubMed WorldCat Vanneste S , Coppens F, Lee E, et al. 2011 . Developmental regulation of CYCA2s contributes to tissue-specific proliferation in Arabidopsis . EMBO Journal 30 , 3430 – 3441 . Google Scholar Crossref Search ADS PubMed WorldCat Vatén A , Bergmann DC. 2012 . Mechanisms of stomatal development: an evolutionary view . EvoDevo 3 , 11 . Google Scholar Crossref Search ADS PubMed WorldCat Walter M , Chaban C, Schütze K, et al. 2004 . Visualization of protein interactions in living plant cells using bimolecular fluorescence complementation . The Plant Journal 40 , 428 – 438 . Google Scholar Crossref Search ADS PubMed WorldCat Wang G , Kong H, Sun Y, Zhang X, Zhang W, Altman N, DePamphilis CW, Ma H. 2004 . Genome-wide analysis of the cyclin family in Arabidopsis and comparative phylogenetic analysis of plant cyclin-like proteins . Plant Physiology 135 , 1084 – 1099 . Google Scholar Crossref Search ADS PubMed WorldCat Xie Z , Lee E, Lucas JR, Morohashi K, Li D, Murray JA, Sack FD, Grotewold E. 2010 . Regulation of cell proliferation in the stomatal lineage by the Arabidopsis MYB FOUR LIPS via direct targeting of core cell cycle genes . The Plant Cell 22 , 2306 – 2321 . Google Scholar Crossref Search ADS PubMed WorldCat Yang K , Wang H, Xue S, Qu X, Zou J, Le J. 2014 . Requirement for A-type cyclin-dependent kinase and cyclins for the terminal division in the stomatal lineage of Arabidopsis . Journal of Experimental Botany 65 , 2449 – 2461 . Google Scholar Crossref Search ADS PubMed WorldCat Yang KZ , Jiang M, Wang M, Xue S, Zhu LL, Wang HZ, Zou JJ, Lee EK, Sack F, Le J. 2015 . Phosphorylation of serine 186 of bHLH transcription factor SPEECHLESS promotes stomatal development in Arabidopsis . Molecular Plant 8 , 783 – 795 . Google Scholar Crossref Search ADS PubMed WorldCat Yoshizumi T , Tsumoto Y, Takiguchi T, Nagata N, Yamamoto YY, Kawashima M, Ichikawa T, Nakazawa M, Yamamoto N, Matsui M. 2006 . Increased level of polyploidy1, a conserved repressor of CYCLINA2 transcription, controls endoreduplication in Arabidopsis . The Plant Cell 18 , 2452 – 2468 . Google Scholar Crossref Search ADS PubMed WorldCat Author notes Present address: Biotechnology Research Institute, Chinese Academy of Agricultural Sciences, Beijing 100081, China Present address: Shanghai Biotechnology Corporation, Shanghai 201203, China © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.

Buijs, Gonda; Kodde, Jan; Groot, Steven P C; Bentsink, Leónie

doi: 10.1093/jxb/ery156pmid: 29701795

Abstract Seed dormancy determines the timing of seed germination and may be released by dry storage, also referred to as after-ripening. Studies on dormancy-release mechanisms are often hampered by the long after-ripening requirements of seeds. After-ripening is thought to be mainly caused by oxidative processes during seed dry storage. These processes are also the main cause of seed ageing. Increasing partial oxygen pressure through the elevated partial pressure of oxygen (EPPO) system has been shown to mimic and accelerate dry seed ageing. In this study, we investigated whether the EPPO system may also release primary seed dormancy in Arabidopsis thaliana. EPPO mimics dry after-ripening at the genetic level, as quantitative trait locus (QTL) analysis after EPPO treatment identified the DELAY OF GERMINATION loci DOG1, DOG2, and DOG6 that were first described in a study using dry after-ripening to release seed dormancy. QTL analysis also showed that dormancy release by cold stratification (another common method to break seed dormancy) partly overlaps with release by after-ripening and EPPO treatment. We conclude that EPPO is an appropriate method to mimic and accelerate dormancy release and, as such, may have applications in both research and industry. After-ripening, Arabidopsis thaliana, Delay of Germination, DOG, elevated partial pressure of oxygen, EPPO, seed dormancy, quantitative trait loci Introduction The seed is the unit for propagation, dispersal, and survival of seed plants. Orthodox seeds can withstand drying and can thus survive over time and be dispersed over distance. The seed has to germinate and establish a seedling in order to grow and finally propagate. The timing of germination is essential for successful seedling establishment. As plants in temperate regions often disperse their seeds in autumn, immediate germination of the seed would cause it to grow in winter, decreasing the chance of seedling survival and propagation. Therefore, most species have dormancy mechanisms that control the timing of germination. A viable seed possesses dormancy when it is temporarily unable to germinate under favourable conditions. Dormancy induced during seed maturation is called primary dormancy, and this can be caused by physiological, physical, or developmental factors, or by a combination of these (Baskin and Baskin, 2004; Bewley et al., 2013). Arabidopsis thaliana displays coat-enhanced physiological dormancy in which the balance between the hormones abscisic acid (ABA; germination-inhibiting) and gibberellins (GAs; germination-promoting) is key in determining the germination status (Baskin and Baskin, 2004; Bewley et al., 2013). However, this hormonal balance is the result of multiple complex pathways and processes, of which the underlying mechanisms largely remain unknown (Nonogaki, 2014; Dekkers and Bentsink, 2015; Chahtane et al., 2017). Seed dormancy has been studied for both scientific reasons (e.g. the intriguing dormant stage in a diploid life stage) and for agricultural reasons (e.g. rapid germination for crops). In the laboratory, there are multiple methods to relieve seed dormancy and to assess and quantify the dormancy level of seed batches. One method is to expose the seeds to a period of cold imbibition (cold stratification, CS; Finch-Savage et al., 2007). In Arabidopsis, stratification induces the expression of gibberellic acid biosynthesis genes and the resulting high levels of GAs release dormancy (Yamauchi et al., 2004). Another common method to relieve and assess dormancy is to store the seeds under dry conditions, so-called dry after-ripening (AR). The period of dry storage that is required to release dormancy is often expressed as days of seed dry storage to reach 50% of germination (DSDS50; Alonso-Blanco et al., 2003). In true potato seeds, AR can be accelerated when the seeds are stored dry under an elevated temperature (37 °C; Alvarado and Bradford, 2005). Dry dormancy release is most likely caused by the formation and action of reactive oxygen species (ROS) (Oracz et al., 2007; El-Maarouf-Bouteau and Bailly, 2008; El-Maarouf-Bouteau et al., 2013; Morscher et al., 2015). The production of ROS has been shown to occur both in dry and imbibed seeds of Arabidopsis (Leymarie et al., 2012), barley (Ma et al., 2016), and sunflower (Oracz et al., 2007). ROS can potentially react with all molecules in a cell, such as lipids, DNA and RNA, proteins, and carbohydrates, and oxidation of a molecule may change its functioning. Oxidized proteins are damaged and degraded upon seed imbibition (reviewed by El-Maarouf-Bouteau et al., 2013; Morscher et al., 2015). The degradation of specific proteins might end the inhibition of germination, either directly by the removal of the proteins or indirectly by inducing germination-promoting signalling pathways. Oxidation of mRNAs has been reported to be important for dormancy release in sunflower seeds (Bazin et al., 2011); however, the underlying molecular mechanisms remain elusive (reviewed by Nonogaki, 2014). Oxidation not only results in dormancy release but also causes seed ageing. These processes are hard to separate (Morscher et al., 2015). Ageing probably starts directly after seed dispersal (or already during seed maturation). First it results in the release of seed dormancy and, later, in seed deterioration as ROS accumulate during seed ageing (Bailly et al., 2008). In accordance with this, seed storage under anoxia slows down seed ageing (Groot et al., 2015). Based on this role of oxygen, a method has been developed to mimic and accelerate dry seed ageing, namely elevated partial pressure of oxygen (EPPO) storage (Groot et al., 2012). In the EPPO method, seeds are stored dry under ambient air, but under increased pressure. This increases the absolute amount and partial pressure of oxygen (pO2) in the storage environment. During both dry ageing and EPPO storage seed tocopherol levels decrease, a process that does not occur under controlled deterioration [storage under high relative humidity (RH) and high temperatures; Groot et al., 2012]. Moreover, EPPO seems to mimic natural ageing better than controlled deterioration in barley, based on phenotypic (e.g. normal seedling formation) and quantitative trait locus (QTL) analyses (Nagel et al., 2016). To identify loci that affect dry after-ripening, genetic approaches have been used in, for example, weedy rice (Gu et al., 2004), barley (Sato et al., 2009), and sorghum (Guo et al., 2015). In Arabidopsis, QTL analysis for dry after-ripening requirement, expressed as DSDS50, resulted in the identification of eleven DELAY OF GERMINATION (DOG) loci in a combined mapping using six recombinant inbred line (RIL) populations (Bentsink et al., 2010). The RIL population of the Landsberg erecta and Cape Verde Island accessions (Ler/Cvi) has led to the identification of the most significant QTLs (Alonso-Blanco et al., 2003; Bentsink et al., 2010). The DOG QTLs identified in the Ler/Cvi RIL population have been confirmed by near-isogenic lines (NILs) in which the QTL regions from the Cvi or Kashmir-2 accession were introgressed into the Ler genotype (Alonso-Blanco et al., 2003; Bentsink et al., 2010). These NILs have been used to identify the genes underlying the DOG QTLs. Thus far, only the causal genes for DOG1 (Bentsink et al., 2006) and DOG18 have been identified (Xiang et al., 2016). These studies show the power of the use of genetic populations in understanding the regulation of quantitative traits. Here, we investigated whether EPPO may release primary seed dormancy. Dormancy release by AR often takes a long time, for example, more than a year for very dormant accessions of Arabidopsis (Vidigal et al., 2016). Accelerating this process is beneficial for both seed dormancy research and commercial applications. We show that EPPO accelerates dormancy release in Ler and the very dormant DOG1 NIL. EPPO mimics dry after-ripening very well, as shown by the large overlap of DSDS50 and EPPO QTLs in the Ler/Cvi RIL population. The identified QTLs were confirmed by testing a set of DOG NILs. Finally, the results are compared and discussed in relation to dormancy QTLs identified by cold stratification. Material and methods Plant material Seeds of the Arabidopsis thaliana Ler accession, NILDOG1-Cvi, NILDOG2-Cvi, NILDOG3-Cvi, and NILDOG6-Kas-2 in the Ler background, and the Ler/Cvi RIL population were used, previously described by Bentsink et al. (2010) and Alonso-Blanco et al. (1998), respectively. The NIL set was grown on Rockwool in 2016 and the Ler/Cvi RIL population was grown in soil in 2007 in a greenhouse under a 22 °C and a 16/8 h light/dark regime. All plants were grown with three biological replicates. After harvest the seeds were stored at –80 °C until use. Prior to the start of the experiments approximately 500 seeds were taken from the –80 °C freezer and placed in open 1.5-ml screw-cap tubes. Subsequently, the open tubes were placed at 20 °C and 35% RH for 3 d. During this acclimation period, the seeds were exposed to air and thus experienced a brief period of AR. Three biological replicates of the NIL set were used in all experiments. For the RIL population, only one biological replicate was used in the EPPO experiment, and another biological replicate was used in the stratification experiment. For the EPPO and the stratification experiments, respectively, 152 and 134 lines of the Ler/Cvi RIL population were used (Supplementary Table S1 at JXB online). The parental lines were also included. AR storage The AR data that we used originated from Alonso-Blanco et al. (2003). These seeds had been stored at ambient conditions with limited temperature control and no humidity control. EPPO storage During the 3-d acclimation period, sub-samples of approximately 50 seeds were taken and placed into 1.5-ml screw-cap tubes. Two holes were punctured in the screw-cap, the rubber ring was removed, and a piece of filter paper was placed inside the screw-cap to prevent the seeds from spilling through the holes. For each storage period one 1.5-l steel tank was used, into which the screw-cap tubes containing the 152 RILs and both parents were placed. A total of four tanks were used, and each was filled with compressed air as described by Groot et al. (2012). To set the relative humidity in the tanks to 35%, a nylon stocking with silica gel equilibrated at 35% RH was added to each one. All tanks were filled over 25 min with air to a pressure of 8 MPa. The tanks were placed at 20 °C for 34 d for the RIL population and for 29 d for the NIL set. For the NIL set, the pressure was subsequently increased to 20 MPa for 12 d. Control seed samples were stored in an air-tight jar at 20 °C and 35% RH. For the N2 treatment, tanks were flushed with N2 prior to filling in order to remove residual air. To test the effect of the rate of pressure build-up, two N2-filled tanks were filled either quickly (0.1 MPa to 8 MPa in 2.5 min) or slowly (0.1 MPa to 8 MPa in 25 min). Pressure release was controlled to prevent physical damage to the seeds that might be caused by a too-rapid expansion of gasses present in the intercellular spaces, using computer-controlled flow control equipment such that the relative pressure decline was maintained at 0.5% per minute. Germination and viability assays For all germination experiments, seeds were sown on blue germination paper in trays with 48 ml demineralised water and placed in a cabinet at 22 °C with continuous light. Each tray contained six samples of approximately 50 seeds. Seed germination was followed for 5 d using the Germinator system (Joosen et al., 2010). Viability of the non-germinated seeds was checked by placing the seeds in a new germination tray with 10 mM KNO3 added to the demineralised water. After 1 d in nitrate the seed coat was removed from the remaining non-germinated seeds. At 2 d after seed-coat removal, viability was assessed by checking for growth of the embryo (greening of the cotyledons and radicle elongation). Stratification experiment The seeds were sown as described for the standard germination experiment above. Seeds were taken from the –80° freezer and stored at 20 °C and 35% RH for 4 d prior to the stratification experiment. After sowing, the seeds were placed in a cold room at 4 °C for 10, 8, 6, 4, 2, or 0 d. Thus, prior to stratification the seeds were stored dry on the bench for 4, 6, 8, 10, 12, or 14 d, respectively. After cold storage all trays, including non-stratified seeds after 14 d of bench storage, were placed simultaneously in a germinator incubator at 22 °C as described. Analysis of DOxy50, DC50, and quantitative trait loci The days of seed EPPO storage to reach 50% germination (DOxy50), days of seed cold storage to reach 50% germination (DC50) and days of dry seed storage to reach 50% germination (DSDS50) were calculated using the statistical program R version 2.14 (R Development Core Team, 2009; www.r-project.org) according to He et al., 2014). QTL analyses were performed with the MapQTL program (version 6, www.kyazma.nl; van Ooijen, 1992). QTLs were identified with both interval mapping and rMQM mapping according to the manual. See Supplementary Table S1 for the phenotypic data that was used. Results Storage under EPPO conditions releases seed dormancy To investigate whether EPPO can release primary seed dormancy in Arabidopsis, a preliminary test was performed. Fresh seeds of the deeply dormant NILDOG1 genotype were stored for 2 weeks under 3, 6, 9, or 12 MPa of air, with ambient air (0.1 MPa) as a control (Supplementary Fig. S1). Storage for 2 weeks at 3 and 6 MPa did not significantly reduce dormancy levels in comparison to ambient air storage. Storage at 9 and 12 MPa resulted in dormancy release, but this but the resolution of the test was not sufficient to be able to identify temporal changes in low-dormant genotypes. To obtain better temporal resolution and a full dormancy release we tested EPPO at 8 MPa for 0–28 d followed by a short period (12 d) of EPPO at 20 MPa with dormant NILDOG1 and Ler seeds (Fig. 1). This protocol allowed a gradual dormancy release for both the low- (Fig. 1A) and deep-dormant genotypes (Fig. 1B) and eventually resulted in 100% germination. To investigate whether the effect of EPPO was due to the elevated pO2, a treatment with 8 MPa of pure nitrogen gas (N2) was performed (Fig. 1); similar to the EPPO air treatment, the pressure was increased to 20 MPa after 28 d of storage at 8 MPa. The N2 treatment with no free oxygen under 8 and 20 MPa showed no significant difference compared to ambient storage. To exclude potential effects of the pressure itself, we compared the effect of the rate of pressure build-up. Seeds were stored under 8 MPa of N2, with a fast (2.5 min) or slow (25 min) build-up of pressure Fig. S2A. There was no effect observed for the rate of pressure build-up itself compared to ambient storage (Fig. S2B). The EPPO treatments as performed in our experiments did not result in visually aged seeds, in that we did not observe morphologically aberrant seedlings, e.g. stunted root growth or discoloration of the cotyledons, which are the first signs of seed ageing in cruciferous species (ISTA, 2018). Fig. 1. Open in new tabDownload slide Dormancy release under elevated partial pressure of oxygen (EPPO) conditions. Germination percentages of Ler (A) and NILDOG1 (B). The seeds were germinated after 0, 6, 14, or 28 d of treatment: EPPO at 8 MPa, ambient conditions, N2 at 8 MPa (EPPN). After 28 d, the pressure was increased from 8 to 20 MPa for both the EPPO and EPPN treatments (indicated by the vertical dotted line) and the final germination assay was performed after 12 d of storage at 20 MPa. Error bars represent s.e.m., n=3. Fig. 1. Open in new tabDownload slide Dormancy release under elevated partial pressure of oxygen (EPPO) conditions. Germination percentages of Ler (A) and NILDOG1 (B). The seeds were germinated after 0, 6, 14, or 28 d of treatment: EPPO at 8 MPa, ambient conditions, N2 at 8 MPa (EPPN). After 28 d, the pressure was increased from 8 to 20 MPa for both the EPPO and EPPN treatments (indicated by the vertical dotted line) and the final germination assay was performed after 12 d of storage at 20 MPa. Error bars represent s.e.m., n=3. EPPO mimics dry seed after-ripening If EPPO dormancy release mimics dormancy release by dry AR, we would expect to identify the same loci when performing QTL analyses. The Ler/Cvi RIL population had previously been used to investigate the genetic basis of seed dormancy. In those experiments QTL mapping for after-ripening requirement was performed on the germination percentages after each storage period on the laboratory bench (1, 3, 6, 10, 15, and 21 weeks of dry AR) and on the DSDS50 value that was derived from these germination percentages (Fig. 2A; data from Alonso-Blanco et al., 2003). These analyses led to the identification of DOG1, DOG2, DOG5, and DOG6. We used this same population to investigate seed dormancy release after EPPO storage. Germination percentages after various intervals of EPPO storage at 8 MPa were determined (Supplementary Fig. S3B). With the same method used to calculate the DSDS50 (Alonso-Blanco et al., 2003), the days of EPPO storage to reach 50% of germination (DOxy50) were calculated (Fig. S3D). DOxy50 showed a strong correlation with DSDS50 (Pearson’s r=0.72, Fig. S3E). Furthermore, after correction for the high number of low-dormant lines in the Ler/Cvi RIL population (Fig. S3D) the correlation remained high (Pearson’s r=0.94, Fig. S3E). QTL analyses were performed for DOxy50 and for the germination percentages after each storage interval (0, 6, 12, 19, and 34 d, Fig. 2B). QTL analysis for the EPPO treatment (DOxy50) identified the DOG1, DOG2, and DOG6 QTLs, which explained 59.3% of the variance. The variance explained by the four DSDS50 QTLs was 62%. The DOG1, DOG2, and DOG6 regions overlapped with the previously identified DSDS50 QTL, which indicated that EPPO dormancy release mimicked dry after-ripening. Fig. 2. Open in new tabDownload slide QTL mapping of dormancy release during after-ripening (AR), elevated partial pressure of oxygen (EPPO) treatment, and cold stratification (CS). Logarithm of the odds ratio (LOD) score maps of dormancy release after AR (A), EPPO (B), and CS (C) for the five chromosomes of Arabidopsis. The graphs show interval mapping of the germination percentages after the different storage periods (grey and black lines) and rMQM mapping of (A) DSDS50, (B) DOxy50, and (C) DC50 (red lines). The horizontal black lines represent the LOD score threshold above which a QTL is significant (LOD=2.5, P<0.05). (D) Graphical representation of the Cvi introgression in the NILDOG genotypes. Dark grey represents the Ler background and black represents the Cvi introgression, flanked by light-grey regions of the introgression recombination breakpoints. Fig. 2. Open in new tabDownload slide QTL mapping of dormancy release during after-ripening (AR), elevated partial pressure of oxygen (EPPO) treatment, and cold stratification (CS). Logarithm of the odds ratio (LOD) score maps of dormancy release after AR (A), EPPO (B), and CS (C) for the five chromosomes of Arabidopsis. The graphs show interval mapping of the germination percentages after the different storage periods (grey and black lines) and rMQM mapping of (A) DSDS50, (B) DOxy50, and (C) DC50 (red lines). The horizontal black lines represent the LOD score threshold above which a QTL is significant (LOD=2.5, P<0.05). (D) Graphical representation of the Cvi introgression in the NILDOG genotypes. Dark grey represents the Ler background and black represents the Cvi introgression, flanked by light-grey regions of the introgression recombination breakpoints. The effect of the QTLs was confirmed by the use of the DOG NILs that contain Cvi introgression fragments in a Ler background at the position of the QTL (Fig. 2D). The NILs showed the same trend in dormancy release dynamics under both AR and EPPO conditions, but EPPO dormancy release was much quicker (Fig. 3). In both conditions the NILDOG2 genotype released dormancy most rapidly, closely followed by Ler. Dormancy was released at the lowest rate in both treatments in NILDOG1, followed by NILDOG3 and NILDOG6. Overall, EPPO dormancy release mimicked AR dormancy release on the genetic level. Fig. 3. Open in new tabDownload slide Dormancy release by elevated partial pressure of oxygen (EPPO) and after-ripening (AR). Germination percentages of dormant Ler and DOG NILs at different intervals during AR (A) and EPPO treatment (B). After 29 d of EPPO at 8 MPa the pressure was increased to 20 MPa (indicated with vertical dotted line), and the germination was assessed again after a further 12 d. Error bars represent s.e.m., n=3. Fig. 3. Open in new tabDownload slide Dormancy release by elevated partial pressure of oxygen (EPPO) and after-ripening (AR). Germination percentages of dormant Ler and DOG NILs at different intervals during AR (A) and EPPO treatment (B). After 29 d of EPPO at 8 MPa the pressure was increased to 20 MPa (indicated with vertical dotted line), and the germination was assessed again after a further 12 d. Error bars represent s.e.m., n=3. QTLs for dormancy release by cold stratification largely overlap with AR and EPPO QTLs Cold stratification (CS) is another method used to assess the dormancy level (Yamauchi et al., 2004; Penfield and Springthorpe, 2012). The Ler/Cvi population was sown and stratified for 0, 2, 4, 6, 8, or 10 d and the days of cold storage required to reach 50% of germination (DC50) was calculated. QTL mapping of dormancy release by CS in the Ler/Cvi RIL population was performed to investigate whether the same loci were identified as for dormancy release by AR. Based on the DC50, both DOG1 and DOG2 QTLs were identified (Fig. 2C). The DC50 QTL explained 56.4% of the total variance. The DC50 mapping lacked the DOG6 locus, which was identified in the DOxy50 and DSDS50 mappings. However, QTL mapping based on the germination percentages at the start of the CS treatment revealed that the DOG6 QTL was identified before stratification (Fig. 2C). Thus, the DOG6 locus is either very sensitive to stratification, indicating that DOG6 dormancy is efficiently removed by CS and therefore no allelic variation is detected, or there is no allelic variation for the response of DOG6 to stratification. To further investigate the stratification requirement of the DOG NILs, germination was determined after 0, 12, 24, 48, 96, or 144 h of cold stratification. The less-dormant genotypes Ler and NILDOG2 released dormancy quickly during CS treatment, whereas the more-dormant genotypes NILDOG1 and NILDOG3 released dormancy more slowly (Fig. 4). NILDOG6 showed an initial slow rate of dormancy release, similar to NILDOG1 and NILDOG3 (after 12 and 24 h of stratification). However, the rate of NILDOG6 dormancy release suddenly increased between 24 and 48 h of stratification, resulting in significantly different germination as compared with both the high- and low-dormant NILs. Fig. 4. Open in new tabDownload slide Germination behaviour of dormant DOG NIL and Ler seeds after different stratification periods at 4 °C. Different letters indicate genotypes that have significantly different germination percentages at each time-point (Student’s t-test, P<0.05). Error bars represent s.e.m., n=3. Fig. 4. Open in new tabDownload slide Germination behaviour of dormant DOG NIL and Ler seeds after different stratification periods at 4 °C. Different letters indicate genotypes that have significantly different germination percentages at each time-point (Student’s t-test, P<0.05). Error bars represent s.e.m., n=3. Discussion Although it has been the subject of considerable research interest, knowledge of the genetic and molecular mechanisms of seed dormancy is still limited (reviewed by Née et al., 2017). Genetic approaches using natural variation in Arabidopsis have identified the DOG loci (Alonso-Blanco et al., 2003); However, only a few genes underlying these QTLs have been identified, among which is the major seed dormancy regulator DOG1. The protein levels of DOG1 correspond to the primary dormancy levels (Nakabayashi et al., 2012). Interestingly, DOG1 protein levels do not decrease as dormancy is released, indicating that its activity is altered during dry storage. This alteration of DOG1 is thought to be caused by oxidation (Nakabayashi et al., 2012). To study dormancy and the role of oxidation, dormancy release can be monitored during seed dry storage (AR), but this can last a long time. A well-known example of an Arabidopsis accession with a high AR requirement for dormancy release is Cvi (Finch-Savage et al., 2007), but this is a characteristic that is not limited to just a few accessions. The Iberian Population, for example, is very dormant as a whole, requiring up to 559 d of dry storage to reach 50% germination (Vidigal et al., 2016). To be able to both accelerate and mimic dry AR, we used the EPPO method. During EPPO treatment, the relative amount of oxygen was the same as under ambient air pressure but the pO2 was increased: under EPPO treatment at 8 MPa the pO2 was 1.68 MPa as compared to 0.021 MPa under ambient conditions. Here, we showed that EPPO treatment released dormancy quickly and in a controlled manner, while the seeds remained dry. EPPO mimicked dormancy release under dry AR at the genetic level, as shown by comparing the QTLs identified for EPPO dormancy release (DOG1, DOG2, and DOG6) with those identified for AR requirement (DOG1, DOG2, DOG6, and DOG5; Alonso-Blanco et al., 2003). This supports the hypothesis that dormancy release by AR is mainly caused by oxidative processes (Oracz et al., 2007; El-Maarouf-Bouteau and Bailly, 2008; El-Maarouf-Bouteau et al., 2013; Morscher et al., 2015). This is certainly true for DOG1, the protein that underlies the DOG1 QTL (Nakabayashi et al., 2012). The DOG5 locus that was identified based on the DSDS50 analysis was the only QTL not identified in the DOxy50 QTL mapping (Fig. 2A, B). A possible explanation for this is the difference in RH between the AR and EPPO storage conditions. Dormancy release is known to be influenced by the moisture content (Probert, 2000). The DOG QTLs identified previously were identified based on dry AR in ambient conditions (estimated humidity fluctuated between 40 and 65%). During EPPO storage, the RH was constantly low (35%) and small differences in moisture content might have large effects during storage (Labuza, 1971). The lack of RH fluctuations during the EPPO treatment might also explain why the EPPO QTL mapping displayed such a high explained variance, even though only one biological replicate was used. Apart from the RH fluctuations that might occur under non-RH controlled AR storage, temperature fluctuations and other time-related factors were also eliminated in the EPPO treatment as compared to dry AR. DSDS50 and DOxy50 were calculated based on multiple germination assays after various storage intervals and thus provide a robust measure of dormancy level. However, the DOG loci also showed a temporal pattern during dormancy release (Alonso-Blanco et al., 2003). For example, the DOG2 locus could not be identified before 6 weeks of AR, but it was identified after all further AR storage periods (Fig. 2A). As these temporal patterns provide insights regarding the dormancy mechanisms underlying the different QTLs, we also studied and compared the DOG loci after each AR and EPPO storage period (Fig. 2A, B). To be able to know which AR storage periods were congruent with storage periods in EPPO, we analysed the Gmax frequency distributions from the different AR and EPPO storage periods and performed correlation analyses on the Gmax percentages (Supplementary Fig. S3A, B, F). These analyses indicated that 0 d in EPPO corresponded with 3 weeks AR, 6 and 12 d EPPO with 6 weeks AR, and 19 and 34 days EPPO with 10 weeks AR (Supplementary Fig. S3F). QTL analyses on these storage periods allowed a more detailed comparison of the DOG loci identified during AR and EPPO storage, and thus we could compare the genetic response to both treatments over storage time (Fig. 2, Supplementary Fig. S3G). The different temporal patterns indicated that the underlying mechanisms of dormancy release were different for the different loci. The DOG2 locus was first identified after 6–12 d of EPPO and 6 weeks of AR storage, and was identified similarly in both treatments thereafter. This indicated that the underlying mechanism of the DOG2 locus had a gradual response to oxidation. In contrast, the DOG3 locus responded more quickly to oxidation, as it was only identified at 3 weeks of AR storage and prior to EPPO treatment, and not after longer storage periods under either treatment. This indicated that the underlying mechanism of the DOG3 locus may be highly responsive to oxidation. With the exception of 1 week of AR, the DOG1 locus was identified following any period of AR and EPPO storage. This corresponds with the hypothesis that the DOG1 protein is oxidized gradually over storage time (Nakabayashi et al., 2012). Only the DOG6 locus showed a different pattern, as it was identified earlier during AR storage but only from 6 d EPPO treatment onward. We have already noted that the QTL for DOG6 is very sensitive to CS, as shown in Figs 2C and 4. This supports previous findings that dormancy release is regulated by multiple additive pathways (Bentsink et al., 2010). Identifying the precise location of the DOG2 locus has proved challenging and is probably because of its close vicinity of the DOG3 locus, which has an opposite effect on dormancy and which was not identified in the EPPO mapping. The EPPO storage showed this QTL to be slightly more to the middle of the chromosome compared to the AR analyses. However, the experiments with the NILDOG2 genotype confirmed the effect of EPPO on the DOG2 locus. All the QTL data combined indicate that EPPO mimics AR storage. We have not yet investigated how EPPO affects seed dormancy at the cellular or molecular level. However, a hypothesis is that proteins essential for the regulation of seed dormancy are either better protected against oxidation or more sensitive to oxidation, depending on their role. Protection could, for example, consist of cruciferin proteins, which have been suggested to buffer against oxidative stress (Nguyen et al., 2015). The EPPO method allows detailed studies of the effects of oxidation on seed dormancy and longevity in a very controlled way, and it provides a quick method to remove dormancy from seeds batches. We also compared EPPO treatment with CS, another dormancy-releasing method. A significant difference with AR or EPPO is that during CS the seeds imbibe and are metabolically active, which allows processes such as translation and transcription. During dry ageing, at least at 35% RH, there is no measurable metabolic activity and enzymatic repair processes cannot take place (Labuza, 1971). An example of the difference between CS and AR is that some accessions from a natural population collected in the Iberian Peninsula barely release dormancy during dry AR, but do release dormancy after CS (Vidigal et al., 2016). The difference in sensitivity to AR and CS might be explained by a difference in sensitivity to, or production of, GA. Physiological dormancy consists of multiple layers (e.g. sensitivity to nitrate, light, and temperature) and whether or not seeds will respond to GA depends on these layers. It is known that ABA levels first have to be low before GA can promote germination (Finch-Savage and Leubner-Metzger, 2006). The DOG1 and DOG2 loci were the only two that were identified in all three analyses (DSDS50, DOxy50, and DC50). QTL mapping for DC50 did not reveal additional QTLs, nor did the QTL analyses on the individual stratification time-points (Fig. 2C). This suggests that dormancy release through CS is (partly) different from dormancy release through oxidation; however, this requires further research. Multiple tests were performed in order to establish that EPPO functioned through oxidative processes and not through the high pressure itself. First, dormancy release was studied under slow and quick pressure build-up (Supplementary Fig. S2A), and no significant differences were observed. The tanks were filled with N2 but not flushed, so oxygen concentrations were comparable to ambient conditions. We chose to use this non-flushed N2 treatment because under EPPO air conditions dormancy release might have been too quick to be able to measure the differences. Second, we studied dormancy release without the presence of gaseous oxygen (tanks flushed and filled with N2, Supplementary Fig. S2B). This treatment was performed with air (EPPO) as well as N2 under the same pressure (ambient and 8 MPa, Fig. 1, Supplementary S2B). Under the N2 treatment at 8 MPa (subsequently increased to 20 MPa), the seeds released dormancy at the same rate as under AR ambient conditions. Remarkably, under ambient 0% oxygen conditions (0.1 MPa N2) dormancy release occurred at the same rate as under AR for the period tested. After 15 and 41 d of treatment, there was a significant difference between Ler stored under 0% oxygen at ambient and elevated pressure conditions (P<0.05). The fact that dormancy release did not stop completely without oxygen present in the tanks (flushed N2 treatments, Fig. 1, Supplementary S2B) can be explained by residual oxygen or ROS in the seeds. The pressure itself did not cause the dormancy release, as there was no observable effect when there was no gaseous oxygen present or when the rate of pressure build-up was increased. Abnormal seedling formation or rupture of the seed coat other than at the site of radicle protrusion was not observed after EPPO storage. Conclusions EPPO is the first method for which it has been proved genetically that it mimics and accelerates AR. It provides a quick and reliable method to assess dormancy levels or to remove dormancy altogether in seed batches, and it allows the mechanisms underlying the control of seed dormancy to be studied. A big advantage of the EPPO method is that the seeds do not imbibe during the treatment, and this allows subsequent treatments, storage, or experiments to be carried out. EPPO has been applied to other species, including lettuce, soybean, and barley, and they have responded well to the method when it has been used for seed ageing (Groot et al., 2012; Nagel et al., 2016). The method has to be adapted for each different species, with the optimal pressure, RH, and temperature needing to be determined. Finally, as seed ageing and seed dormancy are intertwined, the EPPO dormancy release method can be combined with the EPPO ageing method to enable the complete seed life span to be studied in one experiment: the pressure can easily be increased after dormancy release to produce accelerated ageing EPPO conditions. Furthermore, temperature and humidity during storage can be controlled and varied. Supplementary data Supplementary data are available at JXB online. Fig. S1. Germination of dormant NILDOG1 seeds after different EPPO storage treatments. Fig. S2. Effect of the rate of pressure build up on the dormancy release of Ler and NILDOG1 seeds. Fig. S3. Frequency distributions and correlation plots of the phenotypic data used for the QTL analyses. Table S1. Phenotypic data for the Ler/Cvi RILs used for the QTL analyses. Abbreviations: Abbreviations: AR after-ripening CS cold stratification Cvi Cape Verde Island accession DC50 days of cold stratification required to reach 50% of germination DOG delay of germination DOxy50 days of EPPO storage required to reach 50% of germination DSDS50 days of seed dry storage required to reach 50% of germination EPPO elevated partial pressure of oxygen Gmax maximum germination after 100 h of seed imbibition Ler Landsberg erecta accession NIL near-Isogenic Line pO2 partial pressure of oxygen QTL quantitative trait locus RH relative humidity RIL recombinant inbred line ROS reactive oxygen species. Acknowledgements We are grateful to Dr Henk Hilhorst for critically reading the manuscript and Arzu Gizem Kirici for practical support with the Ler/Cvi population experiment. This work was supported by the Netherlands Organisation for Scientific Research (NWO) and by the Dutch Technology Foundation (STW). The authors declare that they have no competing interests. References Alonso-Blanco C , Bentsink L, Hanhart CJ, Blankestijn-de Vries H, Koornneef M. 2003 . Analysis of natural allelic variation at seed dormancy loci of Arabidopsis thaliana . Genetics 164 , 711 – 729 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Alonso-Blanco C , El-Assal SE, Coupland G, Koornneef M. 1998 . Analysis of natural allelic variation at flowering time loci in the Landsberg erecta and Cape Verde Islands ecotypes of Arabidopsis thaliana . Genetics 149 , 749 – 764 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Alvarado V , Bradford KJ. 2005 . Hydrothermal time analysis of seed dormancy in true (botanical) potato seeds . Seed Science Research 15 , 77 – 88 . Google Scholar Crossref Search ADS WorldCat Bailly C , El-Maarouf-Bouteau H, Corbineau F. 2008 . From intracellular signaling networks to cell death: the dual role of reactive oxygen species in seed physiology . Comptes Rendus Biologies 331 , 806 – 814 . Google Scholar Crossref Search ADS PubMed WorldCat Baskin JM , Baskin CC. 2004 . A classification system for seed dormancy . Seed Science Research 14 , 1 – 16 . Google Scholar OpenURL Placeholder Text WorldCat Bazin J , Langlade N, Vincourt P, Arribat S, Balzergue S, El-Maarouf-Bouteau H, Bailly C. 2011 . Targeted mRNA oxidation regulates sunflower seed dormancy alleviation during dry after-ripening . The Plant Cell 23 , 2196 – 2208 . Google Scholar Crossref Search ADS PubMed WorldCat Bentsink L , Hanson J, Hanhart CJ, et al. 2010 . Natural variation for seed dormancy in Arabidopsis is regulated by additive genetic and molecular pathways . Proceedings of the National Academy of Sciences, USA 107 , 4264 – 4269 . Google Scholar Crossref Search ADS WorldCat Bentsink L , Jowett J, Hanhart CJ, Koornneef M. 2006 . Cloning of DOG1, a quantitative trait locus controlling seed dormancy in Arabidopsis . Proceedings of the National Academy of Sciences, USA 103 , 17042 – 17047 . Google Scholar Crossref Search ADS WorldCat Bewley JD , Bradford KJ, Hilhorst HW, Nonogaki H. 2013 . Seeds. Physiology of development, germination and dormancy , 3rd edn, New York : Springer . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Chahtane H , Kim W, Lopez-Molina L. 2017 . Primary seed dormancy: a temporally multilayered riddle waiting to be unlocked . Journal of Experimental Botany 68 , 857 – 869 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Dekkers BJW , Bentsink L. 2015 . Regulation of seed dormancy by abscisic acid and DELAY OF GERMINATION 1 . Seed Science Research 25 , 82 – 98 . Google Scholar Crossref Search ADS WorldCat El-Maarouf-Bouteau H , Bailly C. 2008 . Oxidative signaling in seed germination and dormancy . Plant Signaling & Behavior 3 , 175 – 182 . Google Scholar Crossref Search ADS PubMed WorldCat El-Maarouf-Bouteau H , Meimoun P, Job C, Job D, Bailly C. 2013 . Role of protein and mRNA oxidation in seed dormancy and germination . Frontiers in Plant Science 4 , 77 . Google Scholar Crossref Search ADS PubMed WorldCat Finch-Savage WE , Cadman CS, Toorop PE, Lynn JR, Hilhorst HW. 2007 . Seed dormancy release in Arabidopsis Cvi by dry after-ripening, low temperature, nitrate and light shows common quantitative patterns of gene expression directed by environmentally specific sensing . The Plant Journal 51 , 60 – 78 . Google Scholar Crossref Search ADS PubMed WorldCat Finch-Savage WE , Leubner-Metzger G. 2006 . Seed dormancy and the control of germination . New Phytologist 171 , 501 – 523 . Google Scholar Crossref Search ADS PubMed WorldCat Groot SPC , de Groot L, Kodde J, van Treuren R. 2015 . Prolonging the longevity of ex situ conserved seeds by storage under anoxia . Plant Genetic Resources-Characterization and Utilization 13 , 18 – 26 . Google Scholar Crossref Search ADS WorldCat Groot SPC , Surki AA, de Vos RC, Kodde J. 2012 . Seed storage at elevated partial pressure of oxygen, a fast method for analysing seed ageing under dry conditions . Annals of Botany 110 , 1149 – 1159 . Google Scholar Crossref Search ADS PubMed WorldCat Gu XY , Kianian SF, Foley ME. 2004 . Multiple loci and epistases control genetic variation for seed dormancy in weedy rice (Oryza sativa) . Genetics 166 , 1503 – 1516 . Google Scholar Crossref Search ADS PubMed WorldCat Guo Y , Li P, Yuyama N, Tan LB, Fu YC, Zhu ZF, Liu FX, Sun CQ, Cai HW. 2015 . Identification of quantitative trait locus for seed dormancy and expression analysis of four dormancy-related genes in sorghum . Tropical Plant Biology 8 , 9 – 18 . Google Scholar Crossref Search ADS WorldCat He H , de Souza Vidigal D, Snoek LB, Schnabel S, Nijveen H, Hilhorst H, Bentsink L. 2014 . Interaction between parental environment and genotype affects plant and seed performance in Arabidopsis . Journal of Experimental Botany 65 , 6603 – 6615 . Google Scholar Crossref Search ADS PubMed WorldCat ISTA . 2018 . The international rules for seed testing . Bassersdorf : International Seed Testing Association . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Joosen RV , Kodde J, Willems LA, Ligterink W, van der Plas LH, Hilhorst HW. 2010 . GERMINATOR: a software package for high-throughput scoring and curve fitting of Arabidopsis seed germination . The Plant Journal 62 , 148 – 159 . Google Scholar Crossref Search ADS PubMed WorldCat Labuza TP . 1971 . Kinetics of lipid oxidation in foods . Critical Reviews in Food Science and Nutrition 2 , 355 – 405 . Google Scholar OpenURL Placeholder Text WorldCat Leymarie J , Vitkauskaité G, Hoang HH, Gendreau E, Chazoule V, Meimoun P, Corbineau F, El-Maarouf-Bouteau H, Bailly C. 2012 . Role of reactive oxygen species in the regulation of Arabidopsis seed dormancy . Plant & Cell Physiology 53 , 96 – 106 . Google Scholar Crossref Search ADS PubMed WorldCat Ma Z , Marsolais F, Bykova NV, Igamberdiev AU. 2016 . Nitric oxide and reactive oxygen species mediate metabolic changes in barley seed embryo during germination . Frontiers in Plant Science 7 , 138 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Morscher F , Kranner I, Arc E, Bailly C, Roach T. 2015 . Glutathione redox state, tocochromanols, fatty acids, antioxidant enzymes and protein carbonylation in sunflower seed embryos associated with after-ripening and ageing . Annals of Botany 116 , 669 – 678 . Google Scholar Crossref Search ADS PubMed WorldCat Nagel M , Kodde J, Pistrick S, Mascher M, Börner A, Groot SP. 2016 . Barley seed aging: genetics behind the dry elevated pressure of oxygen aging and moist controlled deterioration . Frontiers in Plant Science 7 , 388 . Google Scholar Crossref Search ADS PubMed WorldCat Nakabayashi K , Bartsch M, Xiang Y, Miatton E, Pellengahr S, Yano R, Seo M, Soppe WJ. 2012 . The time required for dormancy release in Arabidopsis is determined by DELAY OF GERMINATION1 protein levels in freshly harvested seeds . The Plant Cell 24 , 2826 – 2838 . Google Scholar Crossref Search ADS PubMed WorldCat Née G , Xiang Y, Soppe WJ. 2017 . The release of dormancy, a wake-up call for seeds to germinate . Current Opinion in Plant Biology 35 , 8 – 14 . Google Scholar Crossref Search ADS PubMed WorldCat Nguyen TP , Cueff G, Hegedus DD, Rajjou L, Bentsink L. 2015 . A role for seed storage proteins in Arabidopsis seed longevity . Journal of Experimental Botany 66 , 6399 – 6413 . Google Scholar Crossref Search ADS PubMed WorldCat Nonogaki H . 2014 . Seed dormancy and germination—emerging mechanisms and new hypotheses . Frontiers in Plant Science 5 , 233 . Google Scholar Crossref Search ADS PubMed WorldCat Oracz K , El-Maarouf Bouteau H, Farrant JM, Cooper K, Belghazi M, Job C, Job D, Corbineau F, Bailly C. 2007 . ROS production and protein oxidation as a novel mechanism for seed dormancy alleviation . The Plant Journal 50 , 452 – 465 . Google Scholar Crossref Search ADS PubMed WorldCat Penfield S , Springthorpe V. 2012 . Understanding chilling responses in Arabidopsis seeds and their contribution to life history . Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 367 , 291 – 297 . Google Scholar Crossref Search ADS PubMed WorldCat Probert RJ . 2000 . The role of temperature in the regulation of seed dormancy and germination . In: Fenner M, ed. Seeds. The ecology of regeneration in plant communities . Wallingford, UK : CABI Publishing , 261 – 292 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC R Development Core Team . 2009 . R: A language and environment for statistical computing . Vienna, Austria : R Foundation for Statistical Computing . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sato K , Matsumoto T, Ooe N, Takeda K. 2009 . Genetic analysis of seed dormancy QTL in barley . Breeding Science 59 , 645 – 650 . Google Scholar Crossref Search ADS WorldCat van Ooijen JW . 1992 . Accuracy of mapping quantitative trait loci in autogamous species . Theoretical and Applied Genetics 84 , 803 – 811 . Google Scholar Crossref Search ADS PubMed WorldCat Vidigal DS , Marques AC, Willems LA, Buijs G, Méndez-Vigo B, Hilhorst HW, Bentsink L, Picó FX, Alonso-Blanco C. 2016 . Altitudinal and climatic associations of seed dormancy and flowering traits evidence adaptation of annual life cycle timing in Arabidopsis thaliana . Plant, Cell & Environment 39 , 1737 – 1748 . Google Scholar Crossref Search ADS PubMed WorldCat Xiang Y , Song B, Née G, Kramer K, Finkemeier I, Soppe WJ. 2016 . Sequence polymorphisms at the REDUCED DORMANCY5 pseudophosphatase underlie natural variation in Arabidopsis dormancy . Plant Physiology 171 , 2659 – 2670 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Yamauchi Y , Ogawa M, Kuwahara A, Hanada A, Kamiya Y, Yamaguchi S. 2004 . Activation of gibberellin biosynthesis and response pathways by low temperature during imbibition of Arabidopsis thaliana seeds . The Plant Cell 16 , 367 – 378 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.

Lee, Inhye; Kim, Kuglae; Lee, Sumin; Lee, Seungjun; Hwang, Eunjin; Shin, Kihye; Kim, Dayoung; Choi, Jungki; Choi, Hyunmo; Cha, Jeong Seok; Kim, Hoyoung; Lee, Rin-A; Jeong, Suyeong; Kim, Jeongsik; Kim, Yumi; Nam, Hong Gil; Park, Soon-Ki; Cho, Hyun-Soo; Soh, Moon-Soo

Cole, Rex A; Peremyslov, Valera V; Van Why, Savannah; Moussaoui, Ibrahim; Ketter, Ann; Cool, Renee; Moreno, Matthew Andres; Vejlupkova, Zuzana; Dolja, Valerian V; Fowler, John E

doi: 10.1093/jxb/ery162pmid: 29722827

Abstract The exocyst, a conserved, octameric protein complex, helps mediate secretion at the plasma membrane, facilitating specific developmental processes that include control of root meristem size, cell elongation, and tip growth. A genetic screen for second-site enhancers in Arabidopsis identified NEW ENHANCER of ROOT DWARFISM1 (NERD1) as an exocyst interactor. Mutations in NERD1 combined with weak exocyst mutations in SEC8 and EXO70A1 result in a synergistic reduction in root growth. Alone, nerd1 alleles modestly reduce primary root growth, both by shortening the root meristem and by reducing cell elongation, but also result in a slight increase in root hair length, bulging, and rupture. NERD1 was identified molecularly as At3g51050, which encodes a transmembrane protein of unknown function that is broadly conserved throughout the Archaeplastida. A functional NERD1–GFP fusion localizes to the Golgi, in a pattern distinct from the plasma membrane-localized exocyst, arguing against a direct NERD1–exocyst interaction. Structural modeling suggests the majority of the protein is positioned in the lumen, in a β-propeller-like structure that has some similarity to proteins that bind polysaccharides. We suggest that NERD1 interacts with the exocyst indirectly, possibly affecting polysaccharides destined for the cell wall, and influencing cell wall characteristics in a developmentally distinct manner. Arabidopsis, cell elongation, cell wall, exocyst, genetic interaction, root development, root hair, root meristem, secretory pathway, tip growth Introduction The secretory system in plants is a fundamental determinant of plasma membrane composition and cell wall formation (Luschnig and Vert, 2014; Ebine and Ueda, 2015; Kim and Brandizzi, 2016). Consequently, secretory events drive cell growth and morphogenesis, and influence plant development. For example, selective localization of secretion to specific regions of the cell periphery is essential for polarized growth of pollen tubes, enabling sperm cell delivery and sexual reproduction in flowering plants. Additionally, secretion of substances into the apoplast and delivery of receptors and transporters to the plasma membrane allow for intercellular communication and coordination. Ultimately, the secretory system’s intimate influence on the plasma membrane and extracellular activities facilitates responsiveness and survival within variable abiotic and biotic environments. However, it remains unclear how the secretory process in plants is spatially and temporally regulated to direct particular cargos to specific locations of the plasma membrane, cell wall, or apoplast at the appropriate time. Both the conventional secretory system, i.e. vesicular transport from endoplasmic reticulum to Golgi to plasma membrane, and non-conventional pathways are involved (Drakakaki and Dandekar, 2013; Robinson et al., 2016; van de Meene et al., 2017), but how these pathways are tailored to the dynamic requirements of different cell types is only beginning to be revealed. The exocyst, an evolutionarily conserved octameric complex, tethers secretory vesicles to specific sites on the plasma membrane prior to exocytosis and modulates secretory activity to achieve an array of specialized functions. In plants, components of the exocyst have been implicated in a range of processes including pollen tube germination and growth (Cole et al., 2005; Hála et al., 2008; Li et al., 2010), cytokinesis (Fendrych et al., 2010; Rybak et al., 2014), secondary cell wall deposition during tracheary element development (Li et al., 2013; Oda et al., 2015), hypocotyl elongation in etiolated seedlings (Hála et al., 2008), determination of meristem size and cell elongation during primary root growth (Cole et al., 2014), Casparian strip formation (Kalmbach et al., 2017), localized disposition of seed coat pectin (Kulich et al., 2010), callose deposition in trichomes (Kulich et al., 2015), and the polar growth of root hairs (Wen et al., 2005; Synek et al., 2006). The regulation, assembly, and functioning of the exocyst complex in non-plant eukaryotes has been linked to its interactions with small GTPases of the Rho, Ral, and Rab families (Mukherjee et al., 2014), membrane phospholipids (Thapa et al., 2012; Pleskot et al., 2015), plasma membrane scaffolding proteins (Liu and Novick, 2014), and the actin cytoskeleton (Jin et al., 2011; Liu et al., 2012). This interactive milieu helps define exocyst function in yeast and mammals, providing post-translational regulation of key secretory events (Wu and Guo, 2015; Pleskot et al., 2015). In plants, the molecular mechanisms that integrate the exocyst into distinct secretory processes are less well understood. One regulatory mechanism unique to plants is the proliferation and diversification of homologs of the exocyst subunit Exo70 (23 in Arabidopsis), which is hypothesized to allow for specification of particular exocyst functions (Synek et al., 2006; Li et al., 2010; Cvrčková et al., 2012; Vukašinović and Žárský, 2016). In support of this hypothesis, different Exo70 paralogs have been associated with specific cellular processes: Exo70B1 with autophagy (Kulich et al., 2013), Exo70I with arbuscular mycorrhizal symbiosis (Zhang et al., 2015), and Exo70E with the EXPO secretory pathway (Poulsen et al., 2014). Furthermore, in growing pollen tubes, members of the EXO70A, EXO70B and EXO70C subgroups show differential localization patterns and apparent activities (Sekereš et al., 2017; Synek et al., 2017). Other factors that help regulate the exocyst in specific developmental contexts are the scaffolding protein Interactor of Constitutive active ROPs 1 (ICR1) in roots (Lavy et al., 2007); the phosphoinositide PIP2 in pollen tubes (Bloch et al., 2016); ROP2 GTPase (with its effector RIC7) in stomata (Hong et al., 2016); and the combined activities of VETH1–VETH2–COG2 and cortical microtubules in xylem cells (Oda et al., 2015). However, given the breadth of functions known for the plant exocyst, other factors are likely to be involved, including cellular components that interact with the exocyst indirectly, e.g. by enhancing the activity of an exocyst-trafficked protein. To advance the investigation of exocyst-mediated secretory events in plants, we performed a mutagenesis screen to identify interactors linked to the exocyst’s role in Arabidopsis root growth. In this screen, we identified NEW ENHANCER OF ROOT DWARFISM (NERD1), a protein of unknown function that, based on genetic interaction data, acts with the exocyst to facilitate root and hypocotyl elongation and to influence root hair morphology. NERD1 is expressed throughout the plant, suggesting a potential role beyond the root and hypocotyl. NERD1 homologs are found throughout the plant kingdom and beyond. Interestingly, the functional interaction of the exocyst with NERD1 is likely to be indirect, and varies dependent on developmental context. We speculate that NERD1 is involved in the modification of cell wall polysaccharides that are important for cell wall expansion and are a cargo for exocyst-mediated transport to the apoplast. Materials and methods Plant materials and growth conditions Lines of Landsberg erecta-0 and Columbia-0 ecotype of Arabidopsis with T-DNA insertions were obtained from the SALK Institute (Alonso et al., 2003): nerd1-2 (At3g51050, SALK 018060C); nerd1-3 (At3g51050, SALK 051660); exo70A1-2 (At5g03540, SALK 135462); sec8-3 (At3g10380, SALK 026204); sec8-4 (At3g10380, SALK 118129); sec8-6 (At3g 10380, SALK 091118); and myo XI-K (At5g20490, SALK 067972). The exo84b-1 line was a GABI-Kat line (Rosso et al., 2003; Fendrych et al., 2010). The EXO84b-GFP and GFP-SEC8 lines were previously described (Fendrych et al., 2010). The nerd1-1 mutant was generated in an ethyl methanesulfonate (EMS) screen that treated ~5000 sec8-6 seeds with 0.2% EMS for 15 h. M2 generation seed from 4500 M1 plants was collected in pools derived from 16 plant lots. The effectiveness of the mutagenesis was verified by observing greater than 64% of M2 plants with one-quarter aberrant seed, with the gene mutation rate estimated at 1/3000. Arabidopsis seeds were surface-sterilized, stratified at 4 °C for 3–5 d, and planted on growth medium (1× MS, 2% (w/v) sucrose, and vitamins in 1% (w/v) Bacto-agar) or soil as previously described (Cole et al., 2005). Plants were grown in a climate chamber at 22 °C under long-day conditions (16 h of light per day; 7500 lx), with the exception of those used in hypocotyl elongation experiments. For these, seeds were placed in a lighted incubator at 228C for 2–4 h to stimulate germination and then wrapped in foil, oriented vertically, and placed in a dark box in a 22 °C incubator. After 5 d in the dark, digital images were captured and hypocotyl lengths measured. To evaluate the effect of Endosidin2, three groups were germinated on MS plates: seedlings that were homozygous for nerd1-2; nerd1-2 siblings complemented by NERD1–green fluorescent protein (GFP); and Col-0 plants. Plants of each genotype were transferred approximately 3 d after germination to plates containing 0, 20, or 40 μM Endosidin2 (ES2), and grown for an additional 4 d before imaging to determine primary root growth rates. DMSO-dissolved ES2 (or DMSO alone as a control, at 0.5% (v/v)) was added to media during plate preparation. High-throughput sequencing and analysis A plant homozygous for nerd1-1 and lacking a sec8 allele in a Col-0 background was backcrossed to Ler-0, and the progeny were self-crossed to generate an F2 population. Pooled genomic DNA from 150 F2 plants with the nerd phenotype (i.e. homozygotes) was sequenced via an Illumina HiSeq 2000 to generate 58 million paired-end reads. SHOREmap software (Schneeberger et al., 2009) was used to align the reads to the Arabidopsis genome and assess the frequency of Col (the mutagenized parent) and Ler single nucleotide polymorphisms (SNPs) across the population. All variant SNPs in the ~200 kb region of chromosome 3 harboring nerd1-1 (Supplementary Fig. S1 at JXB online) were then searched against genes to identify candidate mutations with likely deleterious effects. Genetic and molecular analyses DNA extraction from leaves and PCR genotyping for mutants containing T-DNA insertions was performed as previously described (Cole et al., 2005). Primers used in PCR and RT-PCR are shown in Supplementary Table S1. PCR-based genotyping to detect the EMS-generated nerd1-1 mutation required use of the restriction enzyme AvaII after amplification, as a target cleavage site in the At3g51050 genomic sequence was eliminated by the G→A transition in the mutant. To evaluate expression via RT-PCR, roots from approximately 50 10-day-old seedlings of each genotype were harvested from plates and frozen in liquid nitrogen. RNA was extracted from the pooled sample for each genotype using a phenol–chloroform procedure, followed by DNase treatment. First strand cDNA synthesis was performed using Superscript II as per the manufacturer’s specifications (Thermo Fisher Scientific), followed by removal of RNA with RNaseH. The cDNA was used as a template for PCR with primer pairs that amplified the sequence to the 5′ of the mutations, to the 3′ of the mutations, or spanning the sites of the mutations. Primers for ACTIN2 were included as an internal control. Generation of NERD1–GFP and imaging A 5597-bp-long genomic fragment encompassing the NERD1 gene along with its putative promoter was PCR amplified from the genomic DNA using KOD Hot Start high-fidelity DNA polymerase (Novagen) and cloned into a modified pMDC32 plasmid using SbfI and PacI sites. Enhanced GFP (EGFP; Clontech) cDNA was added downstream from the NERD1 open reading frame (ORF) to yield pMDC-NERD1-GFP. The plasmid was mobilized into Agrobacterium tumefaciens strain GV3101. Transient expression or coexpression of NERD1 construct and fluorescent Golgi markers in Nicotiana benthamiana leaf epidermal cells was performed by co-infiltrating with Agrobacterium strains carrying NERD1-GFP and either STtmd-YFP or NAG-mTurq (Peremyslov et al., 2012) at concentrations equal to 0.2 OD600. Imaging was conducted 2 d post-infiltration. Leaf fragments were immersed in water and observed using a Zeiss LSM 780 NLO confocal microscope equipped with a Plan-Apochromat ×63 1.4 NA lens. mTurquoise, GFP, and mCherry were excited with the 405 nm diode laser line, 488-nm argon laser line, or 561-nm He–Ne laser line, respectively. For the simultaneous visualization of two fluorophores, dual channel acquisition of signal for either GFP and mTurquoise or GFP and mCherry was performed sequentially to minimize crosstalk. For Brefeldin A (BFA) sensitivity, Arabidopsis seedlings expressing fluorophore-tagged proteins were treated with 50 μM BFA for 90 min, and the BFA-sensitive endomembrane compartments were imaged in root epidermal cells. Evaluation of cortical cell files and root growth parameters in nerd1 mutants using confocal microscopy was performed as previously described (Cole et al., 2014). Briefly, images of roots grown on vertical plates were captured on day 5 and day 7 after germination to determine root growth rates. The 7-day-old seedlings were stained with propidium iodine and then imaged with a Zeiss LSM 780 NLO confocal microscope system. Multiple digital images were taken to capture two cortical cell files for each root, one on each side of the longitudinal midline from the quiescent center to the differentiation zone. Cell widths and lengths were measured. The cortical cell length profile combined with the root growth rate allowed estimations of the number of cells in the meristem, the cell production rate, and the length of the cell cycle for each root. Measurements (e.g. root lengths, hypocotyl lengths, root hair dimensions, and root cortical cell lengths and widths) from confocal digital images were achieved using ImagePro analysis software (MediaCybernetics). Transmitted light images of hypocotyl and root hair specimens were captured with a Leica DFC 295 digital camera attached to a Zeiss Stemi SV 11 dissecting microscope, utilizing Leica Application Suite v3.8. Results Screening for exocyst interactors Mutations in genes encoding components of the exocyst complex in Arabidopsis result in root growth defects that vary from a mild decrease in growth rate in some mutants (e.g. sec8-6 and exo70A1) to severe dwarfism in others (e.g. exo84b-1 and sec8-3) (Cole et al., 2014). Additionally, mutations of some exocyst components reduce the length of root hairs (Synek et al., 2006). We reasoned that a protein that interacts with the exocyst could be revealed if its mutation accentuated the root growth defect of an exocyst mutation that by itself results in only a mild phenotype. Therefore, we screened an EMS-treated population of seedlings homozygous for the mild sec8-6 mutation in a Col-0 background to identify such second-site enhancers. Plants from M2 pools exhibiting both short roots and aberrant root hairs—dubbed the new enhancer of root dwarfism (nerd) phenotype—were outcrossed to a wild-type line, self-pollinated, and screened again for the phenotype in ~1/16 of the progeny, as expected for second site enhancers. After screening 45 pools, we recovered exactly one such mutation, nerd1-1, which, when combined with sec8-6, leads to profound dwarfism throughout the plant and shorter primary roots (~25% of wild-type length, with some demonstrating terminated growth). In addition, root hairs in nerd1 sec8-6 double mutants are occasionally misshapen (Supplementary Fig. S2). To verify that the genetic interaction was not specific to a particular SEC8 allele, the nerd1-1 mutation (isolated after a series of backcrosses to Col-0) was combined with another mild allele, sec8-4, yielding a similar result (Fig. 1A). Intriguingly, initial observations indicated that nerd1 plants, in the absence of a sec8-6 or sec8-4 mutation, have a similar, but less severe phenotype—e.g. root length ~75% of wild-type, and less frequently misshapen root hairs. Fig. 1. Open in new tabDownload slide NERD1 encodes a transmembrane domain protein important for wild-type root development. (A) The combination of nerd1-1 and sec8-4 mutations results in a more severe root growth defect than either single mutation. Scale bar: 1 cm. (B) Map showing exons of NERD1 (At3g51050), the site of the G→A point mutation at a splice junction in nerd1-1, and the sites of the T-DNA insertions nerd1-2 (triangle 2: SALK_018060) and nerd1-3 (triangle 3: SALK_051660). (C) Schematic representation of NERD1 primary sequence features, showing an N-terminal signaling peptide (green), FG–GAP domains (purple), transmembrane domain (aqua), cytoplasmic domain (yellow), and residues predicted to form a calcium-binding pocket by RaptorX (red). The blue diagonal line between (B) and (C) shows where the splice site mutation in nerd1-1 is predicted to affect the NERD1 polypeptide. (D) NERD1 tertiary structure predicted by RaptorX, showing a β-propeller in yellow and α-helices in magenta. (E) Side view of the model in (D) with locations of residues in the predicted calcium-binding pocket (green). Molecular identification of NERD1 To identify the nerd1-1 lesion, we used high-throughput sequencing of a pooled population of mutant plants to identify a ~200 kb region of chromosome 3 tightly linked to nerd1 (Supplementary Fig. S1). This region encompassed ~65 protein-coding genes, only two of which harbored putative EMS-generated G→A mutational differences from the Col-0 reference sequence linked to nerd1-1. Both mutations were validated via Sanger sequencing. The best candidate for nerd1-1 appeared to be a change in a conserved splice acceptor site at the ninth exon of At3g51050 (Fig. 1B). To confirm the molecular identity of NERD1, two independent T-DNA insertion alleles in At3g51050 (Fig. 1B) were obtained from the Salk mutant collection, both of which were associated with short root and root hair defects. Subsequent complementation tests between heterozygotes for all three alleles showed the nerd1 root growth and root hair phenotype appearing in approximately 25% of the progeny, verifying that the two insertion alleles (designated nerd1-2 and -3) were indeed inactivating this same locus affected in the original nerd1-1 line, and proving that At3g51050 corresponds to the NERD1 gene. This segregation pattern further shows that nerd1 mutants do not have a significant gametophyte-derived transmission defect. Notably, pollen is the developmental stage, across 105 stages assessed in the Genevestigator database (Grennan, 2006), associated with lowest expression of At3g51050. The absence of a significant transmission defect in nerd1 mutants, and the low level of expression, suggests that NERD1 function is not as central to polarized growth in pollen tubes as it is in root hairs. In addition to the similar phenotypic severity of each of the three alleles, RT-PCR assays suggested that all three were nulls, each generating aberrant transcripts that likely produce non-functional protein. NERD1 transcripts in nerd1-1 homozygotes were shorter than wild-type in the region spanning the point mutation, and thus were likely mis-spliced, whereas transcripts in nerd1-2 and -3 were detected upstream, but not downstream from their respective T-DNA insertions (Supplementary Fig. S3). The 698 amino acid-long NERD1 sequence is broadly conserved in plants, and homologs are detectable in non-plant species (Supplementary Dataset S1). The Gramene EnsemblPlants database (Kersey et al., 2016) identifies NERD1 homologs in 42 Viridiplantae species, primarily angiosperms, but also including more distantly related Archaeplastida species, including members of Bryophyta (Physcomitrella patens), Lycopodiophyta (Selaginella moellendorffii), and Chlorophyta (Chlamydomonas reinhardtii and Ostreococcus lucimarinus). In Arabidopsis and 26 other Viridiplantae species, NERD1 is identified as a single copy gene. Certain regions of NERD1 also show notable similarity to proteins in both Metazoan and Amoebozoan species (Supplmentary Dataset S1). Protein modeling software was used to predict potential structural characteristics of NERD1 (Fig. 1C; Supplementary Table S2; Supplementary Figs S4 and S5; Supplmentary Datasets S2 and S3). Arabidopsis NERD1 contains an N-terminal signaling peptide that is well conserved across all plant species and is indicative of association with the secretory system (TargetP 1.1, Emanuelsson et al., 2000). A single-pass transmembrane domain is identified near the C-terminus of the protein, leaving a short 18–23 amino acid cytoplasmic tail. Homology threading programs (3DLigandSite, Wass et al., 2010; Raptor-X, Källberg et al., 2012; SWISS-Model, Biasini et al., 2014; Phyre2, Kelley et al., 2015) that compared NERD1’s primary and secondary structure with proteins with tertiary structures solved by x-ray crystallography (Supplmentary Dataset S2) predict (with 90–98% confidence) that the non-cytoplasmic portion of NERD1 folds into a globular protein with a β-propeller structure: seven predicted β-sheets arranged radially and pseudosymmetrically around a central axis (Fig. 1D–E; Supplementary Fig. S5). β-Propellers are widely used as structural scaffold, providing a surface for ligand binding and enzymatic activity (Kopec and Lupas, 2013). In addition, NERD1’s β-propeller contains a putative calcium-binding pocket. The predicted tertiary structure resembles templates for some pyrroloquinoline quinone-dependent enzymes (e.g. alcohol dehydrogenases) and, intriguingly, shares similarity to proteins that interact with polysaccharides or glycoproteins, including lectins, integrins, carbohydrate binding proteins, and perhaps most notably, some pectin lyases and xyloglucanases (Supplmentary Dataset S2). Overall, however, the full length of NERD1 is not strictly homologous to members of any known protein family. Although the exact 3D structure and molecular function of NERD1 remain uncertain, these predictions raise the intriguing possibility that NERD1 is an integral membrane protein that interacts with polysaccharides, potentially in a calcium-dependent manner. Notably, NERD1 (At3g51050) mRNA is expressed throughout most of the Arabidopsis plant, with little change induced by developmental or environmental variables (Grennan, 2006), suggesting that NERD1 is a component of most plant cells. Localization of NERD1 Previous large-scale proteomic analyses detect NERD1 at the plasma membrane and/or Golgi (Mitra et al., 2009; Zhang and Peck, 2011; Parsons et al., 2012; Heard et al., 2015). To validate and refine these findings, a genomic clone harboring the native promoter and complete ORF of NERD1 was tagged with GFP at the 3′ end and expressed transiently in Nicotiana benthamiana or used to generate transgenic Arabidopsis plants. Confocal microscopy of leaf epidermis cells of N. benthamiana revealed that the tagged protein is present in small motile bodies of ca 1 μm in diameter that resembled Golgi stacks (Fig. 2A). To investigate the nature of these bodies, we examined cells of N. benthamiana co-expressing NERD1–GFP and one of two different Golgi markers: STtmd–Cherry or an N-acetylglucosaminyl transferase fused to fluorescent protein mTurquoise (NAG–mTurq) (Peremyslov et al., 2012). In all cells examined, the GFP signal co-localized with these Golgi markers, indicating that NERD1 is primarily present in Golgi (Fig. 2C–E; Supplementary Fig. S6). Fig. 2. Open in new tabDownload slide NERD1–GFP localizes to the Golgi. (A) NERD1–GFP in Nicotiana benthamiana leaf cells. (B) Root tip of a NERD1-GFP-complemented nerd1-2 mutant. (C–E) Co-localization of NERD1 with the Golgi marker: (C) STtmd::mCherry (sialyltransferase transmembrane domain), (D) NERD1–GFP, and (E) merged image. Scale bars: 5 μm (A, C, D, E) and 20 μm (B). The functionality of the NERD1–GFP fusion protein was validated by genetic complementation. To this end, the NERD1–GFP expression cassette was stably transformed into an Arabidopsis nerd1-2 heterozygote line, which was subsequently self-crossed. A progeny plant homozygous for the nerd1-2 mutation but phenotypically wild-type was identified and self-crossed. PCR genotyping verified that all the resultant seedlings were homozygous for the nerd1-2 mutation. One-fourth of these plants (13 of 52) exhibited the nerd root phenotype (shorter roots: 7.6 ± 1.3 mm versus wild-type 13.0 ± 2.6 mm, t-test P<10–12; and altered root-hair morphology); all plants exhibiting the nerd phenotype were negative for the cassette presence and NERD1–GFP expression. In contrast, the fusion cassette was present and expressed in all the seedlings that were phenotypically wild-type, indicating that it provides a functional NERD1 protein. Confocal microscopy revealed that the NERD1–GFP in a nerd1-2 mutant background was localized to mobile punctate structures in the cytoplasm (Fig. 2B), similar to the observations in N. benthamiana, and consistent with NERD1 localization in the Golgi. To further validate association of NERD1–GFP with Golgi, we investigated sensitivity of the fluorescent bodies to BFA, which disrupts Golgi architecture and induces formation of an endoplasmic reticulum (ER)–Golgi hybrid compartment (Ritzenthaler et al, 2002). Arabidopsis seedlings stably expressing either NERD1–GFP or, as a control, NAG–mTurq, were incubated with this drug. As expected, BFA treatment resulted in formation of a typical BFA compartment marked by either NAG–mTurq or NERD1–GFP in each line (Supplementary Fig. S7), strongly supporting the Golgi residence of NERD1–GFP. The localization of NERD1–GFP in the Golgi is notably distinct from that observed for components of the exocyst at the plasma membrane and in the cytoplasm (Fendrych et al., 2010, 2013; Li et al., 2013; Oda et al., 2015). This suggests that the interaction between NERD1 and the exocyst is not a direct interaction at the plasma membrane, as we had initially hypothesized. One possibility is that NERD1 in the Golgi is important for correct transit of exocyst components to the plasma membrane. Thus, we tested whether the nerd1 mutation causes a mislocalization of the exocyst by imaging GFP-labeled exocyst components. Both EXO84–GFP and SEC8–GFP localization patterns at the plasma membrane in nerd1 mutant roots were indistinguishable from their localization in wild-type controls (Fig. 3A and B, and 3C and D, respectively). Thus, the nerd1 mutant root phenotype is not explained by a mislocalization of the exocyst. The converse was also considered, i.e. do exocyst mutations result in altered localization of NERD1? Observation of NERD1–GFP in the roots of exo70A1 and exo84b mutants revealed that NERD1 localization (i.e. in punctate structures within the cytoplasm) was not altered by mutation of these exocyst components (Fig. 3E–H). These data further argue that the genetic interaction of NERD1 and exocyst mutants is indirect. Fig. 3. Open in new tabDownload slide Subcellular localizations of NERD1 and exocyst markers are independent of each other. The localization of exocyst markers to the outer surface of root epidermal cells of nerd1-2 mutants (A, C) is similar to that in wild-type siblings (B, D). Conversely, the predominant localization of NERD1–GFP in the cytoplasm of root epidermal cells of exocyst mutants (E, G) is similar to that of wild-type siblings (F, H). Shown are epidermal cells in the root transition zone (A–D, G, H) and meristem (E, F). Confocal images provide radial longitudinal sections through the center of the root (A, B, E, F) in which the upper portion of cells shown are on the root surface. In tangential sections (C, D, G, H) parallel with the root surface the lateral walls of the epidermal cells are shown. Scale bars: 20 µm. (This figure is available in colour at JXB online.) Genetic interactions with NERD1 mutants depends on developmental context The apparent absence of co-localization of NERD1 and the exocyst motivated a quantitative assessment of their phenotypes and genetic interactions in three distinct developmental contexts. Mutations of NERD1 combined with mutations of exocyst components were examined for their effects on primary root growth, cell elongation in etiolated hypocotyls, and the polarized growth of root hairs. Intriguingly, while all three developmental contexts involve some degree of cell expansion and are impacted by both NERD1 and the exocyst, as detailed below, the specific genetic interactions varied, depending upon context. Growth of primary root and etiolated hypocotyl Primary root growth was examined in plants harboring the nerd1-1 or nerd1-3 mutation in combination with a mutation in an exocyst component, exo70A1 or sec8-4. A comparison of sibling plants confirmed that the primary root growth defect was more severe in the double mutants than in either of the single mutants (Fig. 4A). Additive and multiplicative models were used to predict the severity of the root growth defect that would be observed if the mutations were non-interacting (Hála et al., 2008; Mani et al., 2008). The observed growth rate defect in the double mutants was much more severe than predicted by either model, verifying a synergistic interaction between NERD1 and exocyst in root growth. To further substantiate the functional interaction between NERD1 and the exocyst, nerd1-2 mutants were treated with the chemical Endosidin2 (ES2), which inhibits exocyst function by interacting with EXO70A1 (Zhang et al., 2016). As expected, root growth rates of control nerd1-2 plants harboring the NERD1–GFP construct were not significantly different from those of Col-0 seedlings grown on media containing 0, 20, or 40 μM ES2 (Fig. 4B). In contrast, nerd1-2 homozygotes are significantly more sensitive to the effect of ES2 on root growth rate, at both 20 and 40 μM ES2. Thus, similar to the genetic interaction, the effect of pharmacological inhibition of the exocyst on root growth in nerd1-2 mutants is more than would be predicted by multiplicative or additive models (Fig. 4B). Fig. 4. Open in new tabDownload slide NERD1 acts synergistically with exocyst components to affect primary root growth. (A) Mutation of exocyst components sec8-4 or exo70A1-2 results in a mild reduction in primary root growth rate compared with wild-type (blue bars), while mutation of nerd1 (nerd1-1 or nerd1-3, red bars) results in an approximately 50% reduction in growth rate. The combination of a mutation in an exocyst component with a nerd1 mutation (green bars) results in a severe root growth defect that is more severe than predicted by additive (P<0.001, z-test) or multiplicative models (P<0.0001, z-test), indicating a synergistic interaction. (Error bars: SE; n=19–67 roots for each genotype.) (B) Compared with DMSO-treated controls, root growth was significantly more reduced in nerd1-2 mutants when 20 or 40 μM Endosidin2, an exocyst inhibitor, was added to the growth medium, compared with either Col-0 or nerd1-2; NERD-GFP complemented seedlings. At both concentrations, the reduced growth rate in the nerd1-2 mutant treated with Endosidin2 was more than would be predicted by additive or multiplicative models (P<0.001, z-test). (Error bars: SE; n=13–17 roots for each genotype/Endosidin2 concentration.) The primary root growth defect in exocyst mutants is due to both a reduced number of cells dividing in a shorter meristem, and a slower rate of cell expansion in the elongation zone (outside the meristem) (Cole et al., 2014). Given the functional interaction between NERD1 and the exocyst in the primary root, we were curious to know if nerd1 mutant effects could be attributed to one or both of these underlying mechanisms. Consequently, cortical cell files in the root tips of nerd1 mutants were examined by confocal microscopy and compared with those in the root tips of Col-0 and exo84b (an exocyst mutant with a severe root growth defect) grown on the same vertical plates (Table 1). Similar to exocyst mutants with severe root growth defects (e.g. exo84b-1) (Cole et al., 2014), the reduced primary root growth in nerd1 mutants arises from both less cell elongation, leading to shorter mature cells, and a reduced number of cells dividing in shorter meristems. These defects were quantitatively similar for all three nerd1 alleles, and less severe than in exo84b-1. Notably, mature cortical cell widths in nerd1 mutants are similar to those in Col-0, indicating that the nerd1 defect is specific to cell elongation. This contrasts with exocyst mutant cells, in which overall mature cortical cell size, both length and width, is reduced. Thus, NERD1 appears to be more specifically involved in longitudinal expansion of lateral walls in the root elongation zone. Table 1. Primary root growth parameters of nerd1 mutants compared with Col-0 and exo84b-1 Genotype . Roots evaluated . Root growth(μm h−1) . Meristem size (no. of cells) . Mature cell length (μm) . Mature cell width (μm) . Cell production (cells h−1) . Estimated length of cell cycle (h) . Mean . SD . Mean . SD . na . Mean . SD . na . Mean . SD . Mean . SD . Mean . SD . Col-0 9 469.3 43.8 49.6 4.4 264 200.9 43.0 84 28.0 2.57 2.33 0.24 14.7 1.5 exo84b-1 10 51.2 6.6 10.0 1.9 278 86.6 21.8 84 13.2 1.53 0.59 0.08 11.4 1.1 nerd1-1 6 248.1 30.9 33.8 4.0 82 144.0 30.1 89 27.0 3.98 1.77 0.23 13.3 1.5 nerd1-2 6 231.2 38.8 32.4 3.0 112 153.3 38.5 87 28.4 3.34 1.50 0.17 14.5 0.8 nerd1-2 6 262.7 21.6 35.3 0.9 89 130.7 36.5 75 29.9 3.81 2.08 0.41 12.0 2.3 Genotype . Roots evaluated . Root growth(μm h−1) . Meristem size (no. of cells) . Mature cell length (μm) . Mature cell width (μm) . Cell production (cells h−1) . Estimated length of cell cycle (h) . Mean . SD . Mean . SD . na . Mean . SD . na . Mean . SD . Mean . SD . Mean . SD . Col-0 9 469.3 43.8 49.6 4.4 264 200.9 43.0 84 28.0 2.57 2.33 0.24 14.7 1.5 exo84b-1 10 51.2 6.6 10.0 1.9 278 86.6 21.8 84 13.2 1.53 0.59 0.08 11.4 1.1 nerd1-1 6 248.1 30.9 33.8 4.0 82 144.0 30.1 89 27.0 3.98 1.77 0.23 13.3 1.5 nerd1-2 6 231.2 38.8 32.4 3.0 112 153.3 38.5 87 28.4 3.34 1.50 0.17 14.5 0.8 nerd1-2 6 262.7 21.6 35.3 0.9 89 130.7 36.5 75 29.9 3.81 2.08 0.41 12.0 2.3 Values highlighted in bold are significantly different from Col-0 and exo84b-1 (P<0.001, t-test). Values highlighted in italic are significantly different from exo84b-1, but not Col-0 (P<0.001, t-test). an=number of cells measured. Open in new tab Table 1. Primary root growth parameters of nerd1 mutants compared with Col-0 and exo84b-1 Genotype . Roots evaluated . Root growth(μm h−1) . Meristem size (no. of cells) . Mature cell length (μm) . Mature cell width (μm) . Cell production (cells h−1) . Estimated length of cell cycle (h) . Mean . SD . Mean . SD . na . Mean . SD . na . Mean . SD . Mean . SD . Mean . SD . Col-0 9 469.3 43.8 49.6 4.4 264 200.9 43.0 84 28.0 2.57 2.33 0.24 14.7 1.5 exo84b-1 10 51.2 6.6 10.0 1.9 278 86.6 21.8 84 13.2 1.53 0.59 0.08 11.4 1.1 nerd1-1 6 248.1 30.9 33.8 4.0 82 144.0 30.1 89 27.0 3.98 1.77 0.23 13.3 1.5 nerd1-2 6 231.2 38.8 32.4 3.0 112 153.3 38.5 87 28.4 3.34 1.50 0.17 14.5 0.8 nerd1-2 6 262.7 21.6 35.3 0.9 89 130.7 36.5 75 29.9 3.81 2.08 0.41 12.0 2.3 Genotype . Roots evaluated . Root growth(μm h−1) . Meristem size (no. of cells) . Mature cell length (μm) . Mature cell width (μm) . Cell production (cells h−1) . Estimated length of cell cycle (h) . Mean . SD . Mean . SD . na . Mean . SD . na . Mean . SD . Mean . SD . Mean . SD . Col-0 9 469.3 43.8 49.6 4.4 264 200.9 43.0 84 28.0 2.57 2.33 0.24 14.7 1.5 exo84b-1 10 51.2 6.6 10.0 1.9 278 86.6 21.8 84 13.2 1.53 0.59 0.08 11.4 1.1 nerd1-1 6 248.1 30.9 33.8 4.0 82 144.0 30.1 89 27.0 3.98 1.77 0.23 13.3 1.5 nerd1-2 6 231.2 38.8 32.4 3.0 112 153.3 38.5 87 28.4 3.34 1.50 0.17 14.5 0.8 nerd1-2 6 262.7 21.6 35.3 0.9 89 130.7 36.5 75 29.9 3.81 2.08 0.41 12.0 2.3 Values highlighted in bold are significantly different from Col-0 and exo84b-1 (P<0.001, t-test). Values highlighted in italic are significantly different from exo84b-1, but not Col-0 (P<0.001, t-test). an=number of cells measured. Open in new tab Elongation of the hypocotyl in etiolated seedlings, in contrast to the more complex process of root growth, is due solely to cell elongation, thereby providing a second and more specific system to evaluate the genetic interaction of NERD1 and exocyst mutants in cell elongation. Hypocotyl lengths and epidermal cell lengths in the hypocotyls of 5-day-old dark grown nerd1-3 sec8-4 double mutant seedlings were evaluated and compared with similar measurements in single mutant and wild-type siblings (Supplementary Fig. S8). As in roots, nerd1-3 and sec8-4 interact synergistically to reduce hypocotyl lengthening (Supplementary Fig. S8B). Furthermore, the effect on the hypocotyl was associated with a synergistic defect in cell elongation (Supplementary Fig. S8C), again similar to the results in the primary root. These data are consistent with expression data pointing to a role for NERD1 and the exocyst in cell growth throughout the plant. Polarized growth of root hairs One phenotype leading to selection of the initial nerd1 allele was altered root hair morphology. Short root hairs are characteristic of several exocyst mutants (e.g. exo70A1 and exo84b) (Synek et al., 2006), whereas root hairs that are wild-type in length are observed in other exocyst mutants (e.g. sec8-4). As in the primary root, abrogating exocyst function does not alter NERD1 localization patterns in the root hair (Fig. 5A, B). Nevertheless, comparison of root hairs in exo70A1 nerd1-3 double mutants and sibling single mutants revealed a genetic interaction (Fig. 5C). Surprisingly, the average root hair length in the single nerd1 mutants was significantly longer than that of their wild-type siblings, indicating that NERD1 limits cell growth in this context. As a second surprise, in contrast to the synergistic interaction in primary root growth, exo70A1 nerd1-3 double mutants have short root hairs of similar size to those of exo70A1 single mutants. That is, the effect of the nerd1-3 mutation in increasing average root hair length is masked (epistasis), suggesting that, in root hairs, the exocyst is required for manifestation of NERD1’s root hair length limiting activity. To determine whether the epistatic interaction of NERD1 and the exocyst was specific, we tested for interactions with another mutation that affects the secretory pathway in root hairs, myosin xi-k. Mutation of myosin xi-k results in shorter root hairs, likely due to inhibition of cytoplasmic streaming that drives secretory vesicle transport (Peremyslov et al., 2008; Peremyslov et al., 2012; Park and Nebenführ, 2013; Peremyslov et al., 2015). The double nerd1-1 myo xi-k mutant demonstrates an additive phenotype: root hairs are longer than with the myo xi-k mutation alone, but not as long as wild-type (Fig. 5D). Thus, the epistatic interaction of nerd1 and exo70A1 mutants is specific, further arguing for a close functional relationship between NERD1 and the exocyst. Moreover, the differing outcomes of the interaction in root hairs versus primary roots (epistatic versus synergistic, respectively) argue that this relationship depends on cellular and developmental context. Fig. 5. Open in new tabDownload slide NERD1 is epistatic to the exocyst in the root hair. Root hair lengths were examined for plants growing on vertical plates so that their roots were on the surface of the growth medium, i.e. at the medium–air interface. NERD1–GFP is localized to mobile punctate structures in both wild-type (A) and exo70A1 (B) root hairs (scale bars for A and B: 20 μm). Mature root hairs of nerd1-3 and nerd1-1 mutants are longer than for wild-type siblings (P<0.0001, t-test). (C) Extremely short root hair lengths are observed in exo70A1 mutants compared with wild-type (P<0.0001), and this phenotype is not altered by the addition of a nerd1-3 mutation in the double mutant (P=0.3). (n=190 for WT, nerd1-3, and exo70A1, n =51 for double mutant.) (D) The myo xi-k mutant has short root hairs compared with wild-type (P<0.0001). In nerd1-1; myo xi-k double mutants, root hairs are longer than those of myo xi-k (P<<0.0001), but still far shorter than wild-type (P<0.001). (n>164 for WT, nerd1-1 and myo xi-k; n=79 for double mutant; error bars: SD.) (This figure is available in colour at JXB online.) Additional insight into the role of NERD1 in root hair growth was gained by a closer examination of the morphology of root hairs in nerd1 mutants, which exhibit branches, inflated bases, or bulbous shapes, morphologies that are rare in wild-type siblings (Supplementary Fig. S2). These deviant morphologies are more consistently observed in roots that are growing within agar medium, rather than on the agar surface (where root hairs predominantly extend into the air). Consequently, root hair morphology within the medium was evaluated in 18 roots for each of five genotypes: Col-0 (wild-type), exo84b-1, nerd1-1, nerd1-2, and nerd1-3 (Fig. 6; Supplementary Table S3). Root hairs in nerd1 mutants, as in wild-type, grow out of the apical end of the trichoblasts at a single location, thus indicating that root hair initiation per se is unaffected in nerd1 mutants. However, nerd1 root hairs are often more bulbous, with wider bases and shanks on average, compared with Col-0 (Fig. 6E). Because nerd1 roots exhibit both wild-type and bulbous root hair morphologies, high standard deviations are associated with nerd1 root hair measurements. Rupture of root hairs, evidenced by the extrusion of cytoplasmic contents into the medium from the root hair tip, is also notable in nerd1 mutants (Fig. 6C and D), occurring in 26–34% of root hairs evaluated, compared with a rare incidence (0–1.8%) in Col-0 or exo84b-1 (Fig. 6F; Supplementary Table S3). Average root hair length for nerd1 root hairs growing within the medium is similar to that in wild-type (nerd1-1: 227 μm; nerd1-3: 247 μm; WT: 230 μm), even though rupture presumably stopped growth in some of the mutant root hairs. Overall, nerd1 root hair morphology is consistent with a role for NERD1 in establishing the structural stability, and perhaps limiting compliance, of the cell wall in growing root hairs. Increased compliance of the cell wall upon loss of NERD1 function might allow for more rapid expansion, leading to increased root hair lengths. But such an effect might also make root hairs vulnerable to bulbous expansion and bursting, as is observed. Fig. 6. Open in new tabDownload slide NERD1 affects morphology of root hairs growing within agar medium. (A) Col-0 (wild-type), (B) exo84b-1, (C) nerd1-2, (D) nerd1-3. Arrowheads: ruptured root hair tips; scale bar: 200 µm. (E, F) Root hairs of nerd1 mutants (n=~200 root hairs for each allele) were significantly wider at the base and at the widest part of their shaft than either Col-0 (n=203) or exo84b-1 (n=79) (E), and showed a significantly higher incidence of root hair rupture (F). (Root hairs measured from 18 roots for each genotype, P<0.001, t-test; error bars: SD.) Discussion Secretory events upon which plant growth and development depend are manifested and regulated by a complex network of cellular components interacting both directly and indirectly. Facilitating secretion in many circumstances is an octameric protein complex, the exocyst (Cole and Fowler, 2006; Žárský et al., 2013; Kulich et al., 2015; Vukašinović and Žárský, 2016). To search for unknown components of the exocyst-mediated secretory network, we used a second-site enhancer screen, and identified nerd1 mutants via their genetic interaction with exocyst mutants, influencing primary root growth, root hair expansion, and hypocotyl elongation in Arabidopsis. Notably, the interaction between NERD1 and the exocyst appears to be indirect, and thus would not have been detected by other methods, e.g. yeast two-hybrid screening. Mutation of NERD1 leads to a shortened root meristem and reduced cell elongation in both the primary root and etiolated hypocotyls. These defects are also seen in plants with mutations affecting exocyst components, but are synergistically accentuated when both nerd1 and exocyst mutations are combined. Additionally, mutations of both NERD1 and components of the exocyst affect root hair morphology. The nearly ubiquitous expression of NERD1 throughout the plant (similar to that of most exocyst components) and its conservation throughout the plant kingdom underline its potential importance in a broader context. Notably, we were unable to generate a doubly homozygous nerd1/exocyst mutant plant from the self-cross of a double heterozygote combining nerd1-3 with severe exocyst mutants (i.e. exo84b-1 and sec8-3; 0 out of 58 and 0 out of 66 individuals genotyped from nerd/exocyst segregating populations, respectively; P<0.05 by chi-square test for each). This suggests that mutations of NERD1 combined with severe exocyst mutations lead to lethality due to very early developmental defects. We initially hypothesized that NERD1 directly interacts with the exocyst at the plasma membrane, where exocyst components are known to localize (Fendrych et al., 2010). However, the majority of proteomic studies identify NERD1 in the ER or Golgi (Parsons et al., 2012; Nikolovski et al., 2014; Heard et al., 2015), and not in the plasma membrane (Mitra et al., 2009). Direct examination of the functional NERD1–GFP fusion validated its prominent localization in the Golgi (Fig. 2; Supplementary Fig. S6), in contrast to the preferential association of exocyst components with the plasma membrane (Fig. 3), suggesting that NERD1–exocyst interaction is indirect. Although it remains possible that NERD1 is present at the plasma membrane transiently or at a low level, the most likely interpretation of our results is that the synergistic genetic interaction between NERD1 and exocyst components does not involve direct physical contact. One alternative hypothesis that does not rely on direct contact to explain the observed NERD1–exocyst genetic interaction is that NERD1 is required for correct exocyst localization; or vice versa, NERD1 is a cargo for exocyst-mediated trafficking. A few specific cargos requiring the plant exocyst for correct delivery have been identified (pectinacious mucilage in Arabidopsis seed coats (Kulich et al., 2010); callose in leaf trichomes (Kulich et al., 2015); and the integral plasma membrane proteins PEN3/ABCG36 and NIP5;1 (Mao et al., 2016)). However, no mislocalization of fluorescently tagged exocyst components in nerd1 mutants, or of NERD1–GFP in exocyst mutants, was observed (Fig. 3), arguing against this possibility. On the other hand, these experiments do not exclude genetic interaction via a currently unknown cargo that requires both NERD1 and exocyst-mediated vesicle transport for its proper function. The localization of NERD1 to the Golgi, the site of synthesis of non-cellulose polysaccharides incorporated into the cell wall matrix (e.g. pectin and hemicellulose; Driouich et al., 2012; Kim and Brandizzi, 2016), is tantalizing. NERD1 could be involved in the formation or function of trans-Golgi-localized protein complexes, such as the ECHIDNA/YIP4 complex that plays a role in post-Golgi secretion of pectin and hemicellulose to the cell wall, and which also influences cell elongation in roots and hypocotyls (Gendre et al., 2013). However, we currently favor a working hypothesis, based on predicted structure, in which NERD1 directly affects a cell wall matrix polysaccharide, glycoprotein or proteoglycan that is ultimately secreted via exocyst-mediated trafficking to influence cell wall growth and expansion. Most of NERD1 is predicted to be located in the Golgi lumen, folded into a β-propeller-like tertiary structure that could serve as a scaffold for interactions with polysaccharides. Homology modeling suggests that this lumenal portion of NERD1 resembles proteins that interact with polysaccharides, i.e. lectins, integrins, and carbohydrate binding proteins. Perhaps most notably, threading programs identify certain NERD1 regions as similar to bacterial RGI pectin lyases, and to a lesser extent xyloglucanases (Supplmentary Dataset S2). Interestingly, RGI pectin lyases are activated by calcium, consistent with the calcium-binding pocket predicted for NERD1. Examination of the cell wall in nerd1 roots by, for example, histochemical staining for specific components should help test the hypothesis that NERD1’s impact arises from a role influencing cell wall structure. It is noteworthy that the phenotypes observed in nerd1 mutants are consistent with cell wall pectins as a target of NERD1 activity. Altering the synthesis of pectic polysaccharides is known to cause a dwarfed phenotype with developmental defects that include shorter primary roots and reduced elongation of etiolated hypocotyls (Reboul et al., 2011), reminiscent of defects observed in nerd1 mutants. RGI pectins impact the same aspects of root hair morphology (e.g. swelling and branching) as those altered in nerd1 mutants (Diet et al., 2006; Reboul et al., 2011). The accelerated cell elongation phase observed in etiolated hypocotyls is the result of cell wall modification, and in particular has been associated with altered pectins (Derbyshire et al., 2007; Pelletier et al., 2010), possibly independent of cellulose synthesis. Thus, a role of NERD1 in the synthesis or modification of cell wall pectins might explain the range of phenotypes observed in nerd1 mutants and deserves further investigation. Intriguingly, the effect of NERD1 on cell expansion is not uniform throughout development: nerd1 mutations result in reduced cell elongation in the root tip and etiolated hypocotyl, but increased elongation of root hairs, along with increased likelihood of rupture at the growing root hair tip. It is also of note that a mutation altering Arabidopsis EXO70C2, another potential indirect exocyst interactor, leads to a similar phenotype in the tip-growing pollen tube: more rapid growth and increased tube rupture (Synek et al., 2017). The contrasting effects of nerd1 mutants on root hair tip growth versus primary root cell elongation could manifest because the cell wall matrices in the two cell types are structurally distinct from each other, generated by fundamentally different processes: delivery of non-cellulosic cell wall components to a narrowly focused region versus more broadly distributed modification of a preexisting cell wall matrix, respectively. Additional indirect evidence that cell walls are differentially altered is the increased incidence of bulging in nerd1 root hair shafts, although neither bulging nor rupture is characteristic of cells in mutant root meristematic and elongation zones. In the root, the composition and structure of the cell wall changes as cells progress through division, elongation, and differentiation zones. For example, the rhamnogalacturonan I pectin in cell walls is modified during the transition from cell proliferation to cell elongation in roots (Willats et al., 1999), and pectins in root hairs are structurally distinct from those in the lateral cell walls elsewhere in the primary root (Muszyński et al., 2015). Thus, the composition of cell wall components available for interaction with NERD1 within the Golgi likely varies as root cells progress from elongation to differentiation. Such differences could alter NERD1’s impact on cell wall extensibility, elongation, and fragility in a developmentally dependent manner. Revealing the molecular function of NERD1 should help define how it is integrated with secretory system to help determine developmentally stage-specific patterns of cell growth and expansion. Supplementary data Supplementary data are available at JXB online. Methods. In silico prediction of NERD1 tertiary structure, with references. Dataset S1. Alignment of NERD1 homologs. Dataset S2. Threading analysis of NERD1. Dataset S3. Alignment of NERD1 homologs, Jalview format. Fig. S1. SHOREmap identification of nerd1 candidate genes. Fig. S2. Images of root hair morphology in nerd1. Fig. S3. RT-PCR for nerd1 mutants. Fig. S4. Diagram of predicted domains in NERD1 protein. Fig. S5. Model depicting predicted tertiary structure of NERD1. Fig. S6. Localization of NERD1–GFP to Golgi using NAG::Turq. Fig. S7. Localization of NERD1–GFP changes in response to BFA. Fig. S8. NERD1 acts synergistically with SEC8 to affect hypocotyl elongation. Table S1. List of primers used for PCR and RT-PCR. Table S2. Domains within NERD1 protein. Table S3. Morphology of nerd1 root hairs. Acknowledgements We acknowledge the high school Apprenticeships in Science and Engineering (ASE) program (SVW, IM, MAM); the M. J. Murdock Charitable Trust’s Partners in Science program for high school teachers (RC); and the Center for Genome Research and Biocomputing (CGRB) Central Services Lab for sequencing and bioinformatics assistance. This work was supported by US National Science Foundation (NSF) grant MCB-1244633. Imaging was made possible by the Confocal Microscopy Facility of the CGRB, supported in part by award 1337774 from the NSF, MRI: Acquisition of Confocal and Two-Photon Excitation Microscope. The authors declare that they have no conflicts of interest. References Alonso JM , Stepanova AN, Leisse TJ, et al. 2003 . Genome-wide insertional mutagenesis of Arabidopsis thaliana . Science 301 , 653 – 657 . Google Scholar Crossref Search ADS PubMed WorldCat Biasini M , Bienert S, Waterhouse A, et al. 2014 . SWISS-MODEL: modelling protein tertiary and quaternary structure using evolutionary information . Nucleic Acids Research 42 , W252 – W258 . Google Scholar Crossref Search ADS PubMed WorldCat Bloch D , Pleskot R, Pejchar P, Potocký M, Trpkošová P, Cwiklik L, Vukašinović N, Sternberg H, Yalovsky S, Žárský V. 2016 . Exocyst SEC3 and phosphoinositides define sites of exocytosis in pollen tube initiation and growth . Plant Physiology 172 , 980 – 1002 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Cole RA , Fowler JE. 2006 . Polarized growth: maintaining focus on the tip . Current Opinion in Plant Biology 9 , 579 – 588 . Google Scholar Crossref Search ADS PubMed WorldCat Cole RA , McInally SA, Fowler JE. 2014 . Developmentally distinct activities of the exocyst enable rapid cell elongation and determine meristem size during primary root growth in Arabidopsis . BMC Plant Biology 14 , 386 . Google Scholar Crossref Search ADS PubMed WorldCat Cole RA , Synek L, Zarsky V, Fowler JE. 2005 . SEC8, a subunit of the putative Arabidopsis exocyst complex, facilitates pollen germination and competitive pollen tube growth . Plant Physiology 138 , 2005 – 2018 . Google Scholar Crossref Search ADS PubMed WorldCat Cvrčková F , Grunt M, Bezvoda R, Hála M, Kulich I, Rawat A, Žárský V. 2012 . Evolution of the land plant exocyst complexes . Frontiers in Plant Science 3 , 159 . Google Scholar Crossref Search ADS PubMed WorldCat Derbyshire P , McCann MC, Roberts K. 2007 . Restricted cell elongation in Arabidopsis hypocotyls is associated with a reduced average pectin esterification level . BMC Plant Biology 7 , 31 . Google Scholar Crossref Search ADS PubMed WorldCat Diet A , Link B, Seifert GJ, Schellenberg B, Wagner U, Pauly M, Reiter WD, Ringli C. 2006 . The Arabidopsis root hair cell wall formation mutant lrx1 is suppressed by mutations in the RHM1 gene encoding a UDP-L-rhamnose synthase . The Plant Cell 18 , 1630 – 1641 . Google Scholar Crossref Search ADS PubMed WorldCat Drakakaki G , Dandekar A. 2013 . Protein secretion: how many secretory routes does a plant cell have ? Plant Science 203–204 , 74 – 78 . Google Scholar Crossref Search ADS PubMed WorldCat Driouich A , Follet-Gueye ML, Bernard S, Kousar S, Chevalier L, Vicré-Gibouin M, Lerouxel O. 2012 . Golgi-mediated synthesis and secretion of matrix polysaccharides of the primary cell wall of higher plants . Frontiers in Plant Science 3 , 79 . Google Scholar Crossref Search ADS PubMed WorldCat Ebine K , Ueda T. 2015 . Roles of membrane trafficking in plant cell wall dynamics . Frontiers in Plant Science 6 , 878 . Google Scholar Crossref Search ADS PubMed WorldCat Emanuelsson O , Nielsen H, Brunak S, von Heijne G. 2000 . Predicting subcellular localization of proteins based on their N-terminal amino acid sequence . Journal of Molecular Biology 300 , 1005 – 1016 . Google Scholar Crossref Search ADS PubMed WorldCat Fendrych M , Synek L, Pecenková T, Drdová EJ, Sekeres J, de Rycke R, Nowack MK, Zársky V. 2013 . Visualization of the exocyst complex dynamics at the plasma membrane of Arabidopsis thaliana . Molecular Biology of the Cell 24 , 510 – 520 . Google Scholar Crossref Search ADS PubMed WorldCat Fendrych M , Synek L, Pecenková T, et al. 2010 . The Arabidopsis exocyst complex is involved in cytokinesis and cell plate maturation . The Plant Cell 22 , 3053 – 3065 . Google Scholar Crossref Search ADS PubMed WorldCat Gendre D , McFarlane HE, Johnson E, Mouille G, Sjödin A, Oh J, Levesque-Tremblay G, Watanabe Y, Samuels L, Bhalerao RP. 2013 . Trans-Golgi network localized ECHIDNA/Ypt interacting protein complex is required for the secretion of cell wall polysaccharides in Arabidopsis . The Plant Cell 25 , 2633 – 2646 . Google Scholar Crossref Search ADS PubMed WorldCat Grennan AK . 2006 . Genevestigator. Facilitating web-based gene-expression analysis . Plant Physiology 141 , 1164 – 1166 . Google Scholar Crossref Search ADS PubMed WorldCat Hála M , Cole R, Synek L, et al. 2008 . An exocyst complex functions in plant cell growth in Arabidopsis and tobacco . The Plant Cell 20 , 1330 – 1345 . Google Scholar Crossref Search ADS PubMed WorldCat Heard W , Sklenář J, Tomé DF, Robatzek S, Jones AM. 2015 . Identification of regulatory and cargo proteins of endosomal and secretory pathways in Arabidopsis thaliana by proteomic dissection . Molecular & Cellular Proteomics 14 , 1796 – 1813 . Google Scholar Crossref Search ADS WorldCat Hong D , Jeon BW, Kim SY, Hwang JU, Lee Y. 2016 . The ROP2-RIC7 pathway negatively regulates light-induced stomatal opening by inhibiting exocyst subunit Exo70B1 in Arabidopsis . New Phytologist 209 , 624 – 635 . Google Scholar Crossref Search ADS PubMed WorldCat Jin Y , Sultana A, Gandhi P, Franklin E, Hamamoto S, Khan AR, Munson M, Schekman R, Weisman LS. 2011 . Myosin V transports secretory vesicles via a Rab GTPase cascade and interaction with the exocyst complex . Developmental Cell 21 , 1156 – 1170 . Google Scholar Crossref Search ADS PubMed WorldCat Kalmbach L , Hématy K, De Bellis D, Barberon M, Fujita S, Ursache R, Daraspe J, Geldner N. 2017 . Transient cell-specific EXO70A1 activity in the CASP domain and Casparian strip localization . Nature Plants 3 , 17058 . Google Scholar Crossref Search ADS PubMed WorldCat Källberg M , Wang H, Wang S, Peng J, Wang Z, Lu H, Xu J. 2012 . Template-based protein structure modeling using the RaptorX web server . Nature Protocols 7 , 1511 – 1522 . Google Scholar Crossref Search ADS PubMed WorldCat Kelley LA , Mezulis S, Yates CM, Wass MN, Sternberg MJ. 2015 . The Phyre2 web portal for protein modeling, prediction and analysis . Nature Protocols 10 , 845 – 858 . Google Scholar Crossref Search ADS PubMed WorldCat Kersey PJ , Allen JE, Armean I, et al. 2016 . Ensembl genomes 2016: more genomes, more complexity . Nucleic Acids Research 44 , D574 – D580 . Google Scholar Crossref Search ADS PubMed WorldCat Kim SJ , Brandizzi F. 2016 . The plant secretory pathway for the trafficking of cell wall polysaccharides and glycoproteins . Glycobiology 26 , 940 – 949 . Google Scholar Crossref Search ADS PubMed WorldCat Kopec KO , Lupas AN. 2013 . β-Propeller blades as ancestral peptides in protein evolution . PLoS ONE 8 , e77074 . Google Scholar Crossref Search ADS PubMed WorldCat Kulich I , Cole R, Drdová E, Cvrcková F, Soukup A, Fowler J, Žárský V. 2010 . Arabidopsis exocyst subunits SEC8 and EXO70A1 and exocyst interactor ROH1 are involved in the localized deposition of seed coat pectin . New Phytologist 188 , 615 – 625 . Google Scholar Crossref Search ADS PubMed WorldCat Kulich I , Pečenková T, Sekereš J, Smetana O, Fendrych M, Foissner I, Höftberger M, Žárský V. 2013 . Arabidopsis exocyst subcomplex containing subunit EXO70B1 is involved in autophagy-related transport to the vacuole . Traffic 14 , 1155 – 1165 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Kulich I , Vojtíková Z, Glanc M, Ortmannová J, Rasmann S, Žárský V. 2015 . Cell wall maturation of Arabidopsis trichomes is dependent on exocyst subunit EXO70H4 and involves callose deposition . Plant Physiology 168 , 120 – 131 . Google Scholar Crossref Search ADS PubMed WorldCat Lavy M , Bloch D, Hazak O, Gutman I, Poraty L, Sorek N, Sternberg H, Yalovsky S. 2007 . A novel ROP/RAC effector links cell polarity, root-meristem maintenance, and vesicle trafficking . Current Biology 17 , 947 – 952 . Google Scholar Crossref Search ADS PubMed WorldCat Li S , Chen M, Yu D, Ren S, Sun S, Liu L, Ketelaar T, Emons AM, Liu CM. 2013 . EXO70A1-mediated vesicle trafficking is critical for tracheary element development in Arabidopsis . The Plant Cell 25 , 1774 – 1786 . Google Scholar Crossref Search ADS PubMed WorldCat Li S , van Os GM, Ren S, Yu D, Ketelaar T, Emons AM, Liu CM. 2010 . Expression and functional analyses of EXO70 genes in Arabidopsis implicate their roles in regulating cell type-specific exocytosis . Plant Physiology 154 , 1819 – 1830 . Google Scholar Crossref Search ADS PubMed WorldCat Liu D , Novick P. 2014 . Bem1p contributes to secretory pathway polarization through a direct interaction with Exo70p . The Journal of Cell Biology 207 , 59 – 72 . Google Scholar Crossref Search ADS PubMed WorldCat Liu J , Zhao Y, Sun Y, He B, Yang C, Svitkina T, Goldman YE, Guo W. 2012 . Exo70 stimulates the Arp2/3 complex for lamellipodia formation and directional cell migration . Current Biology 22 , 1510 – 1515 . Google Scholar Crossref Search ADS PubMed WorldCat Luschnig C , Vert G. 2014 . The dynamics of plant plasma membrane proteins: PINs and beyond . Development 141 , 2924 – 2938 . Google Scholar Crossref Search ADS PubMed WorldCat Mani R , Onge RPS, Hartman JL, Giaever G, Roth FP. 2008 . Defining genetic interaction . Proceedings of the National Academy of Sciences, USA 105 , 3461 – 3466 . Google Scholar Crossref Search ADS WorldCat Mao H , Nakamura M, Viotti C, Grebe M. 2016 . A framework for lateral membrane trafficking and polar tethering of the PEN3 ATP-binding cassette transporter . Plant Physiology 172 , 2245 – 2260 . Google Scholar Crossref Search ADS PubMed WorldCat Mitra SK , Walters BT, Clouse SD, Goshe MB. 2009 . An efficient organic solvent based extraction method for the proteomic analysis of Arabidopsis plasma membranes . Journal of Proteome Research 8 , 2752 – 2767 . Google Scholar Crossref Search ADS PubMed WorldCat Mukherjee D , Sen A, Aguilar RC. 2014 . RhoGTPase-binding proteins, the exocyst complex and polarized vesicle trafficking . Small GTPases 5 , e28453 . Google Scholar Crossref Search ADS PubMed WorldCat Muszyński A , O’Neill MA, Ramasamy E, et al. 2015 . Xyloglucan, galactomannan, glucuronoxylan, and rhamnogalacturonan I do not have identical structures in soybean root and root hair cell walls . Planta 242 , 1123 – 1138 . Google Scholar Crossref Search ADS PubMed WorldCat Nikolovski N , Shliaha PV, Gatto L, Dupree P, Lilley KS. 2014 . Label-free protein quantification for plant Golgi protein localization and abundance . Plant Physiology 166 , 1033 – 1043 . Google Scholar Crossref Search ADS PubMed WorldCat Oda Y , Iida Y, Nagashima Y, Sugiyama Y, Fukuda H. 2015 . Novel coiled-coil proteins regulate exocyst association with cortical microtubules in xylem cells via the conserved oligomeric golgi-complex 2 protein . Plant & Cell Physiology 56 , 277 – 286 . Google Scholar Crossref Search ADS PubMed WorldCat Park E , Nebenführ A. 2013 . Myosin XIK of Arabidopsis thaliana accumulates at the root hair tip and is required for fast root hair growth . PLoS ONE 8 , e76745 . Google Scholar Crossref Search ADS PubMed WorldCat Parsons HT , Christiansen K, Knierim B, et al. 2012 . Isolation and proteomic characterization of the Arabidopsis Golgi defines functional and novel components involved in plant cell wall biosynthesis . Plant Physiology 159 , 12 – 26 . Google Scholar Crossref Search ADS PubMed WorldCat Pelletier S , Van Orden J, Wolf S, et al. 2010 . A role for pectin de-methylesterification in a developmentally regulated growth acceleration in dark-grown Arabidopsis hypocotyls . New Phytologist 188 , 726 – 739 . Google Scholar Crossref Search ADS PubMed WorldCat Peremyslov VV , Cole RA, Fowler JE, Dolja VV. 2015 . Myosin-powered membrane compartment drives cytoplasmic streaming, cell expansion and plant development . PLoS ONE 10 , e0139331 . Google Scholar Crossref Search ADS PubMed WorldCat Peremyslov VV , Klocko AL, Fowler JE, Dolja VV. 2012 . Arabidopsis myosin XI-K localizes to the motile endomembrane vesicles associated with F-actin . Frontiers in Plant Science 3 , 184 . Google Scholar Crossref Search ADS PubMed WorldCat Peremyslov VV , Prokhnevsky AI, Avisar D, Dolja VV. 2008 . Two class XI myosins function in organelle trafficking and root hair development in Arabidopsis . Plant Physiology 146 , 1109 – 1116 . Google Scholar Crossref Search ADS PubMed WorldCat Pleskot R , Cwiklik L, Jungwirth P, Žárský V, Potocký M. 2015 . Membrane targeting of the yeast exocyst complex . Biochimica et Biophysica Acta 1848 , 1481 – 1489 . Google Scholar Crossref Search ADS PubMed WorldCat Poulsen CP , Dilokpimol A, Mouille G, Burow M, Geshi N. 2014 . Arabinogalactan glycosyltransferases target to a unique subcellular compartment that may function in unconventional secretion in plants . Traffic 15 , 1219 – 1234 . Google Scholar Crossref Search ADS PubMed WorldCat Reboul R , Geserick C, Pabst M, Frey B, Wittmann D, Lütz-Meindl U, Léonard R, Tenhaken R. 2011 . Down-regulation of UDP-glucuronic acid biosynthesis leads to swollen plant cell walls and severe developmental defects associated with changes in pectic polysaccharides . The Journal of Biological Chemistry 286 , 39982 – 39992 . Google Scholar Crossref Search ADS PubMed WorldCat Ritzenthaler C , Nebenführ A, Movafeghi A, Stussi-Garaud C, Behnia L, Pimpl P, Staehelin LA, Robinson DG. 2002 . Reevaluation of the effects of brefeldin A on plant cells using tobacco Bright Yellow 2 cells expressing Golgi-targeted green fluorescent protein and COPI antisera . The Plant Cell 14 , 237 – 261 . Google Scholar Crossref Search ADS PubMed WorldCat Robinson DG , Ding Y, Jiang L. 2016 . Unconventional protein secretion in plants: a critical assessment . Protoplasma 253 , 31 – 43 . Google Scholar Crossref Search ADS PubMed WorldCat Rosso MG , Li Y, Strizhov N, Reiss B, Dekker K, Weisshaar B. 2003 . An Arabidopsis thaliana T-DNA mutagenized population (GABI-Kat) for flanking sequence tag-based reverse genetics . Plant Molecular Biology 53 , 247 – 259 . Google Scholar Crossref Search ADS PubMed WorldCat Rybak K , Steiner A, Synek L, et al. 2014 . Plant cytokinesis is orchestrated by the sequential action of the TRAPPII and exocyst tethering complexes . Developmental Cell 29 , 607 – 620 . Google Scholar Crossref Search ADS PubMed WorldCat Schneeberger K , Ossowski S, Lanz C, Juul T, Petersen AH, Nielsen KL, Jørgensen JE, Weigel D, Andersen SU. 2009 . SHOREmap: simultaneous mapping and mutation identification by deep sequencing . Nature Methods 6 , 550 – 551 . Google Scholar Crossref Search ADS PubMed WorldCat Sekereš J , Pejchar P, Šantrůček J, Vukašinović N, Žárský V, Potocký M. 2017 . Analysis of exocyst subunit EXO70 family reveals distinct membrane polar domains in tobacco pollen tubes . Plant Physiology 173 , 1659 – 1675 . Google Scholar Crossref Search ADS PubMed WorldCat Synek L , Schlager N, Eliás M, Quentin M, Hauser MT, Žárský V. 2006 . AtEXO70A1, a member of a family of putative exocyst subunits specifically expanded in land plants, is important for polar growth and plant development . The Plant Journal 48 , 54 – 72 . Google Scholar Crossref Search ADS PubMed WorldCat Synek L , Vukašinović N, Kulich I, Hála M, Aldorfová K, Fendrych M, Žárský V. 2017 . EXO70C2 is a key regulatory factor for optimal tip growth of pollen . Plant Physiology 174 , 223 – 240 . Google Scholar Crossref Search ADS PubMed WorldCat Thapa N , Sun Y, Schramp M, Choi S, Ling K, Anderson RA. 2012 . Phosphoinositide signaling regulates the exocyst complex and polarized integrin trafficking in directionally migrating cells . Developmental Cell 22 , 116 – 130 . Google Scholar Crossref Search ADS PubMed WorldCat van de Meene AML , Doblin MS, Bacic A. 2017 . The plant secretory pathway seen through the lens of the cell wall . Protoplasma 254 , 75 – 94 . Google Scholar Crossref Search ADS PubMed WorldCat Vukašinović N , Žárský V. 2016 . Tethering complexes in the arabidopsis endomembrane system . Frontiers in Cell and Developmental Biology 4 , 46 . Google Scholar Crossref Search ADS PubMed WorldCat Wass MN , Kelley LA, Sternberg MJ. 2010 . 3DLigandSite: predicting ligand-binding sites using similar structures . Nucleic Acids Research 38 , W469 – W473 . Google Scholar Crossref Search ADS PubMed WorldCat Wen TJ , Hochholdinger F, Sauer M, Bruce W, Schnable PS. 2005 . The roothairless1 gene of maize encodes a homolog of sec3, which is involved in polar exocytosis . Plant Physiology 138 , 1637 – 1643 . Google Scholar Crossref Search ADS PubMed WorldCat Willats WG , Steele-King CG, Marcus SE, Knox JP. 1999 . Side chains of pectic polysaccharides are regulated in relation to cell proliferation and cell differentiation . The Plant Journal 20 , 619 – 628 . Google Scholar Crossref Search ADS PubMed WorldCat Wu B , Guo W. 2015 . The exocyst at a glance . Journal of Cell Science 128 , 2957 – 2964 . Google Scholar Crossref Search ADS PubMed WorldCat Žárský V , Kulich I, Fendrych M, Pečenková T. 2013 . Exocyst complexes multiple functions in plant cells secretory pathways . Current Opinion in Plant Biology 16 , 726 – 733 . Google Scholar Crossref Search ADS PubMed WorldCat Zhang C , Brown MQ, van de Ven W, et al. 2016 . Endosidin2 targets conserved exocyst complex subunit EXO70 to inhibit exocytosis . Proceedings of the National Academy of Sciences, USA 113 , E41 – 50 . Google Scholar Crossref Search ADS WorldCat Zhang X , Pumplin N, Ivanov S, Harrison MJ. 2015 . EXO70I is required for development of a sub-domain of the periarbuscular membrane during arbuscular mycorrhizal symbiosis . Current Biology 25 , 2189 – 2195 . Google Scholar Crossref Search ADS PubMed WorldCat Zhang ZJ , Peck SC. 2011 . Simplified enrichment of plasma membrane proteins for proteomic analyses in Arabidopsis thaliana . Proteomics 11 , 1780 – 1788 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.

Boatman, Tobias G; Mangan, Niall M; Lawson, Tracy; Geider, Richard J

doi: 10.1093/jxb/ery141pmid: 29659983

Abstract Increasing atmospheric CO2 concentrations are leading to increases in dissolved CO2 and HCO3– concentrations and decreases in pH and CO32– in the world’s oceans. There remain many uncertainties as to the magnitude of biological responses of key organisms to these chemical changes. In this study, we established the relationship between photosynthetic carbon fixation rates and pH, CO2, and HCO3– concentrations in the diazotroph, Trichodesmium erythraeum IMS101. Inorganic 14C-assimilation was measured in TRIS-buffered artificial seawater medium where the absolute and relative concentrations of CO2, pH, and HCO3– were manipulated. First, we varied the total dissolved inorganic carbon concentration (TIC) (<0 to ~5 mM) at constant pH, so that ratios of CO2 and HCO3– remained relatively constant. Second, we varied pH (~8.54 to 7.52) at constant TIC, so that CO2 increased whilst HCO3– declined. We found that 14C-assimilation could be described by the same function of CO2 for both approaches, but it showed different dependencies on HCO3– when pH was varied at constant TIC than when TIC was varied at constant pH. A numerical model of the carbon-concentrating mechanism (CCM) of Trichodesmium showed that carboxylation rates are modulated by HCO3– and pH. The decrease in assimilation of inorganic carbon (Ci) at low CO2, when TIC was varied, was due to HCO3– uptake limitation of the carboxylation rate. Conversely, when pH was varied, Ci assimilation declined due to a high-pH mediated increase in HCO3– and CO2 leakage rates, potentially coupled to other processes (uncharacterised within the CCM model) that restrict Ci assimilation rates under high-pH conditions. Carbon acquisition, carbon concentrating mechanism (CCM), CO2, Cyanobacteria, gross photosynthesis, net photosynthesis, ocean acidification, Trichodesmium Introduction Over the past 150 years, atmospheric CO2 concentrations have increased from pre-industrial levels (i.e. 280 µmol mol–1) to a current value of about 400 µmol mol–1, and are predicted to increase further to 650 µmol mol–1 by mid-century, and to 750–1000 µmol mol–1 by the end of this century (Raven et al., 2005). Equilibration of CO2 between the atmosphere and the oceans is leading to increases in dissolved CO2 and HCO3– and to decreases in pH and CO32–. This process of ocean acidification is predicted to reduce the pH from average pre-industrial levels of 8.2 to about 7.9 by the end of the century (Zeebe et al., 1999; Zeebe and Wolf-Gladrow, 2001). To date, there are still many uncertainties as to the magnitude of biological responses of key organisms to these chemical changes. One group of organisms of particular importance are the diazotrophic cyanobacteria (photosynthetic dinitrogen-fixers), notably because of their significant contribution to marine primary productivity by converting N2 into NH4+, thus providing ‘new’ nitrogen to the oceans. The filamentous cyanobacterium Trichodesmium is a colony-forming species that fixes nitrogen in an area corresponding to half the Earth’s surface (Davis and McGillicuddy, 2006), and is estimated to account for more than half of the new (combined) nitrogen production in many parts of the oligotrophic tropical and sub-tropical oceans (Capone et al., 2005). Cyanobacteria, Trichodesmium species included, achieve high photosynthetic rates despite (i) the slow diffusion of CO2 in water (104 times slower than in air), (ii) a slow chemical equilibrium between HCO3– and CO2 within the 7–8.5 pH range, and (iii) a low affinity of Rubisco for CO2 relative to ambient CO2 concentrations. Cyanobacteria employ an intracellular carbon-concentrating mechanism (CCM) (Badger and Price, 2003; Badger et al., 2006; Kranz et al., 2010), where enhanced primary productivity significantly outweighs the metabolic costs of CCM activity (Price et al., 2008). The CCM benefits cyanobacteria by reducing photorespiration (Schwarz et al., 1995; Kaplan and Reinhold, 1999), aiding in the dissipation of excess light energy, and by maintaining an optimal intracellular pH (Badger et al., 1994; Kaplan and Reinhold, 1999). The general consensus is that up-regulation of CCM activity in response to a low-CO2 environment involves two components. Firstly, an increase in the transport of inorganic carbon (Ci) from the environment into the cell via a suite of Ci transporters, which could involve using ATP (BCT1 HCO3– transporter), NADPH, or reduced ferredoxin (CO2 conversion from passive diffusion) or coupling to an electrochemical Na+ gradient (SbtA or BicA HCO3– transport) to provide the energy for Ci uptake (Badger et al., 2002; Badger and Price, 2003). Secondly, an increased ability to reduce CO2 leakage from around the site of carboxylation, achieved via arrangement of the molecular components of the carboxysome structure and a CO2-uptake system located on the thylakoid layer, preventing the efflux of leaked CO2 to the outer cytosolic layer (Price et al., 2008). Both 14C isotope disequilibrium experiments and simultaneous measurements of CO2 and O2 exchanges during sequential light–dark transitions indicate that HCO3– contributes >90% of the Ci assimilation by T. erythraeum IMS101 (Kranz et al., 2009; Eichner et al., 2015). This preference for HCO3– is consistent with the evidence that Trichodesmium lacks a plasma membrane-bound extracellular carbonic anhydrase (eCA) (Badger et al., 2006; Price et al., 2008). Furthermore, the T. erythraeum genome indicates the presence of both a plasma membrane HCO3– transporter (BicA) and an intracellular system for conversion of CO2 to HCO3– (NDH-I4) (Price et al., 2008). These two modes of the CCM result in the accumulation of HCO3– in the cytosol, which diffuses to the carboxysome. Inorganic carbon uptake by Trichodesmium involves the uptake of HCO3– by the BicA transporter. This transporter has a half-saturation constant, Km, of 40–100 µM HCO3–, which is well below the typical concentration of HCO3– in seawater (~2000 µM) (Badger et al., 2006). Following transport into the cell, C-fixation in Trichodesmium, like other cyanobacteria species, occurs within carboxysomes where HCO3– is converted to CO2 via a carbonic anhydrase, followed by fixation of CO2 by Rubisco. Carboxysomes provide micro-environments where CO2 is elevated to compensate for the low affinity of cyanobacterial Rubiscos for CO2 (KmCO2>150 mM) (Badger and Andrews, 1987). In Trichodesmium, CO2 that leaks from carboxysomes can be converted to HCO3– by the plasma membrane-bound NDH-I4 protein, thus reducing the efflux of CO2 from the cell, but at a cost of consuming reducing equivalents (NADPH or reduced Fd) (Price et al., 2008). Despite having a mechanism for intracellular recycling of CO2, efflux is reported to account for the loss of up to 50% of HCO3– uptake in Trichodesmium (Kranz et al., 2010; Eichner et al., 2015). As reviewed in Boatman et al. (2017), the majority of previous studies have shown an increase (albeit not all statistically significant) in T. erythraeum IMS101 growth under predicted future CO2 concentrations (~750–1000 µmol mol–1), although the magnitudes of these responses differ between studies (see Supplementary Table S1 at JXB online). The increased growth and productivity of T. erythraeum IMS101 with increased CO2 is probably attributable to a decrease in the energy required for operation of the CCM, allowing more energy (ATP) and reductant (NADPH) to be reallocated to N2 fixation, CO2 fixation, and biosynthesis (Kranz et al., 2011). Given the significant contribution of Trichodesmium to carbon and nitrogen biogeochemical cycles, and the predicted changes to Ci speciation over the coming decades due to ocean acidification, we performed a systematic study to assess how the kinetics of Ci assimilation of T. erythraeum IMS101 were affected by acclimation to varying CO2 concentrations. We ensured that the Ci chemistry and all other growth conditions were well defined, with cultures fully acclimated over long time periods to achieve balanced growth. We assessed how the rate of Ci assimilation was related to CO2 or HCO3– concentrations in experiments where Ci speciation was modulated by varying pH and total dissolved inorganic carbon concentration (TIC). These assays of photosynthetic performance showed that Trichodesmium productivity was influenced by high pH when TIC was held at a saturating concentration, indirectly making the rate of Ci assimilation a saturating function of CO2 concentration, and that maximum rates of CO2 fixation declined and affinity for CO2 increased when Trichodesmium was acclimated to a low-CO2 concentration. We discuss how these responses can be attributed to decreases in the cost of operating a CCM under future CO2 conditions. Materials and methods Trichodesmium erythraeum IMS101 was semi-continuously cultured to achieve fully acclimated balanced growth at three target CO2 concentrations (180, 380, and 720 µmol mol–1) under saturating light intensity (400 µmol photons m–2 s–1), a 12/12 h light/dark (L/D) cycle, and an optimum growth temperature (26 ± 0.7 °C) for ~5 months (~80 generations). Experimental set-up Cultures of T. erythraeum IMS101 were grown using YBCII medium (Chen et al., 1996) at 1.5-l volumes in 2-l Pyrex bottles that had been acid-washed and autoclaved prior to culturing. Daily growth rates were quantified from changes in baseline fluorescence (Fo) measured between 09.00 to 10.30 h on dark-adapted cultures (20 min) using a FRRfII FastAct Fluorometer System (Chelsea Technologies Group Ltd, UK). Cultures were kept at the upper section of the exponential growth phase through periodic dilution with new growth media at 3–5 d intervals. They were deemed fully acclimated and in balanced growth when both the slope of the linear regression of ln(Fo) and the ratio of live-cell to acetone-extracted (method detailed below) baseline fluorescence were constant following every dilution with fresh YBCII medium. Illumination was provided side-on by fluorescent tubes (Sylvania Luxline Plus FHQ49/T5/840). Cultures were constantly mixed using magnetic PTFE stirrer bars and aerated with a filtered (0.2-µm pore) air mixture at a rate of ~200 ml s–1. The CO2 concentration was regulated (±2 µmol mol–1) by mass-flow controllers (Bronkhorst, Newmarket, UK) and CO2-free air was supplied by an oil-free compressor (Bambi Air, UK) via a soda-lime gas-tight column that was mixed with a 10% CO2-in-air mixture from a gas cylinder (BOC Industrial Gases, UK). The CO2 concentration in the gas phase was continuously monitored and recorded by an infra-red gas analyser (Li-Cor Li-820, Nebraska USA), calibrated weekly against a standard gas (BOC Industrial Gases). The Ci chemistry was measured prior to the dilution of each culture with fresh media; pH and TIC were measured directly, while HCO3–, CO32–, and CO2 concentrations were calculated using CO2SYS with the same constants as described in Boatman et al. (2017) (see Supplementary Information SI). Elemental stoichiometry Samples for elemental composition and CO2-response curves were collected at the same time of day between 4 and 6 h into the light period of the L/D cycle. Samples for determination of particulate organic carbon (POC), particulate nitrogen (PN), and particulate phosphorus (PP) were collected together with each CO2-response curve, where each sample was a biological replicate culture. Three 100-ml aliquots from each culture were vacuum-filtered onto pre-combusted 25-mm (0.45-µm pore) glass-fibre filters for measurements of POC, PN, and PP. The POC and PN filters were placed in 1.8-ml cryovials (lids off) and dried at 60 °C. The PP filters were rinsed with 2 ml of sodium sulphate (0.1 M), placed in a 20-ml glass scintillation vial, 2 ml of magnesium sulphate (0.017 M) added, and then dried at 60 °C. POC was quantified using a TC analyser (Shimadzu TOC-V Analyser & SSM-5000A Solid Sample Combustion Unit), PN by the method of Bronk and Ward (2000), and PP by the method of Solorzano and Sharp (1980). Inorganic carbon fixation-response curves The dependencies of CO2 fixation on CO2 and HCO3– were determined from experiments that involved varied TIC with fixed pH and varied pH with fixed TIC (see Supplementary Information SII, SIII) in TRIS-buffered YBCII medium using the 14C uptake technique (Steemann Nielsen and Jensen, 1957). Prior to each experiment, 1 l of bicarbonate-free YBCII medium was aerated overnight with CO2-free air (soda-lime column). A 200-ml sample from each culture was gravity-filtered onto a 47-mm cyclopore filter (1-µm pore; Whatman 60750) and gently re-suspended into 50 ml of the CO2-free YBCII medium. Exactly 5 ml of concentrated culture was pipetted into each tube of the TIC or pH gradients (35 ml total volume per tube) and gently inverted to evenly distribute the trichomes. The remaining culture was used for measurement of initial activity, T0. Three replicate cultures were used per treatment. During sample preparation, test-tubes were maintained at growth temperature (26 °C) and a low light intensity (<10 µmol photons m–2 s–1). To characterise the Ci chemistry, exactly 20 ml of culture from each treatment was filtered through a Swinnex filter (25 mm, 0.45-µm pore, glass-fibre filter): 15 ml into a plastic centrifuge tube (no headspace) for TIC analysis (Shimadzu TOC-V Analyser & ASI-V Autosampler), and 5 ml into a plastic cryogenic vial (Sigma-Aldrich V5257-250EA; no headspace) for pH analysis. To measure chlorophyll a concentrations, a 1-ml sample from each treatment was pipetted into 9 ml of 100% acetone and left in a freezer (–20 °C) overnight (Welschmeyer, 1994). The sample was vortex-mixed and left in the dark (~30 min) to allow cell debris to precipitate and the solution to equilibrate to room temperature. A 2-ml aliquot was used to measure Fo using a FRRfII FastAct Fluorometer System (Chelsea Technologies Group Ltd, UK) with the same parameters as used for live cultures. Chlorophyll a concentrations were calculated from a calibration curve derived from a dilution series measured using a chlorophyll a standard (Sigma-Aldrich C5753). To assess whether cells had been affected by concentration via filtration and re-suspension and exposure to the range of TIC and pH gradients over the course of the 14C incubations, 2-ml aliquots of culture from each treatment were dark-acclimated (~20 min) and the photosynthetic efficiency of PSII (Fv/Fm) was determined using a FRRfII FastAct Fluorometer System (Chelsea Technologies Group Ltd, UK) (see Supplementary Fig. S1). Finally, 10 ml of culture from each treatment was pipetted into 12-ml glass (PTFE-capped) test-tubes and used for 14C incubations. A 14C spike solution was prepared by pipetting 45 µl of a 14C-labelled sodium bicarbonate solution (NaH14CO3) with a specific activity of 52 mCi mmol–1 (Perkin Elmer, USA) into 8 ml of bicarbonate-free YBCII media. Exactly 250 µl of the spike was added to each tube culture. The T0 tubes were immediately filtered through Swinnex filters containing 25-mm diameter (0.45-µm pore) glass-fibre filters, placed in scintillation vials, and acidified (500 µl of 3 M HCl). To determine the total activity (TC), 20 µl of the spike was added into three scintillation vials already containing 4.5 ml of scintillation cocktail (Gold LLT) and 200 µl of phenylethylamine. The TC vial caps were screwed tight immediately. The spiked test-tubes were placed within a custom-made water-jacketed incubator and maintained at 26 °C under saturating light intensity (400 ± 6 µmol photons m–2 s–1) (The Optoelectronic Manufacturing Corporation Ltd. 1ft T5 Daylight, UK). The incubations lasted between 60 and 90 min and took place between 4 to 6 h into the light period of the L/D cycle. The 14C incubations were repeated in the dark, using black-coated (Plasti-Kote paint) test-tubes. Dark 14C uptake rates were 8.25% (±0.46) and 7.05% (±0.25) of the maximum light-saturated 14C uptake rates for the TIC and pH response curves, respectively. Dark 14C uptake rates exhibited no response to varying TIC or pH and were used to correct the light-dependent rates of photosynthesis (Li and Dickie, 1991). To terminate 14C uptake, samples were filtered through 25-mm (0.45-µm pore) glass-fibre filters (Fisherbrand FB59451, UK) using a bespoke 30-funnel filtration manifold. Test-tubes and filters were rinsed twice with 5 ml of YBCII media, before the filters were placed into scintillation vials. The vials were acidified (500 µl of 3 M HCl) overnight along with the T0 samples. Exactly 4.5 ml of scintillation cocktail (Gold LLT) was added to the acidified vials and the caps tightened. Ensuring that the scintillation cocktail and filtered samples were well mixed, the vials were placed within a scintillation counter and the disintegrations per minute (DPM) of each vial were measured (20 min per vial). The CO2 fixation rates were calculated using the following equation: C−fixation=(DPMS−DPMT0DPMTC)×(VolTCVols)×(TICt)×1.05(1) where DPMS, DPMT0, and DPMTC are the measurements for the sample, initial activity, and total activity vials, respectively; TIC (mmol l–1) is the mean concentration of total dissolved inorganic carbon within the sample over the course of the incubation (inclusive of the NaH14CO3 spike); VolTC and VolS are the volumes of the sample and TC vials, respectively; t is the experimental incubation time (h); and 1.05 is the radioisotope discrimination factor (12C:14C). Note that mean T0 and TC values were used when calculating the C-fixation rates (n=3). Inorganic carbon fixation rates were normalised to a POC basis and the CO2 response curves were fitted to a Michaelis–Menten function: VC=(VC,max ⋅ [CO2])(Km + [CO2])(2) where VC is the organic C-specific rate of CO2 fixation, VC,max is the maximum rate of CO2 fixation, and Km is the half-saturation constant. Curve-fitting was performed on individual replicates to calculate mean (±SE) curve-fit parameters (Sigmaplot 11.0), as well on the combined data where all replicates of the varied TIC (fixed pH) and varied pH (fixed TIC) data were combined per CO2 treatment. Spectrophotometric chlorophyll a analysis Samples for spectrophotometric determination of chlorophyll a were collected together with each CO2-response curve and were used to normalise productivity rates as well as to calculate the ratio of Chl a:C (i.e. total C). A 100-ml sample from each culture was vacuum-filtered onto a 25-mm (0.45-µm pore) glass-fibre filter (Fisherbrand FB59451, UK) and extracted in 5 ml of 100% methanol. The filters were homogenised and extracted overnight at –20 °C before being centrifuged at 12 000 g for 10 min and a 3-ml aliquot of the supernatant was transferred to a quartz cuvette. The absorption spectrum (400–800 nm) was measured using a spectrophotometer (Hitachi U-3000, Japan) and the Chl a concentration (µg l–1) was calculated using the following equation (Ritchie, 2008); Chl a=(12.9447×(Abs665−Abs750)×VolEVolF)×1000(3) where Abs665 and Abs750 are the baseline-corrected optical densities of the methanol extracted sample at 665 and 750 nm, respectively; VolE is the volume of the solvent used for extraction (i.e. 5 ml); VolF is the volume of culture that was filtered (i.e. 100 ml); and 12.9447 is a cyanobacteria-specific Chl a coefficient for 100% methanol extraction. Modelling the CCM The CO2 and HCO3– fluxes and concentrations in an idealised Trichodesmium cell were calculated using the numerical model from Mangan et al. (2016) and Mangan and Brenner (2014). The aim was to provide a qualitatively informative view of the CCM system, without attempting to match carboxylation rates or fluxes to the experimental system or to rescale the results from the idealised cell to what would be expected from the experimental data. With the exception of a few key parameter values (Table 1), the model used was equivalent to that reported in Mangan et al. (2016). The main changes between the idealised Trichodesmium cell and previous models were an increase in cell and carboxysome size to be consistent with reported values for T. erythraeum, changes to the Rubisco kinetic constants, use of pH and external CO2 and HCO3– concentrations similar to those in the 14C incubations, updating the pKaeff for HCO3– to CO2 to match that used in the CO2SYS calculation, and re-calculating the HCO3– uptake rate to support internal inorganic carbon concentrations of ~30 mM. We scaled the Rubisco concentration by the carboxysome volume, so that the activity per volume remained the same. Similarly, we scaled the amount of carbonic anhydrase by the carboxysome surface area, so that the activity per area remained the same. The carbonic anhydrase activity was sufficient to equilibrate CO2 and HCO3– to K´eq=[HCO3−/[CO2]=10–pKaeff+pH ⁠. We set the carbonic anhydrase Kca value to preserve the correct equilibrium value for the internal pH. Table 1. Key parameter values used in the numerical simulation of the CCM in Trichodesmium Variable . Units . Model value . Cell radius, Rb µm 3 Carboxysome radius, Rc µm 0.15 Rubisco reaction rate, kRub s–1 per active site 1.92 Rubisco KCO2 µM 145 Rubisco KO2 µM 600 Rubisco specificity, S – 45 Number of Rubisco active sites – 54000 Number of carbonic anhydrase active sites – 900 Carbonic anhydrase half-maximum constant for CO2, Kca µM 104.7 Internal pH – 8.3 pKaeff for HCO3–:CO2 – 5.84 Carboxysome permeability cm s–1 3 × 10–5 HCO3– uptake velocity, jc cm s–1 2.4 × 10–7 CO2 to HCO3– conversion at membrane cm s–1 0.6 × 10–7 Variable . Units . Model value . Cell radius, Rb µm 3 Carboxysome radius, Rc µm 0.15 Rubisco reaction rate, kRub s–1 per active site 1.92 Rubisco KCO2 µM 145 Rubisco KO2 µM 600 Rubisco specificity, S – 45 Number of Rubisco active sites – 54000 Number of carbonic anhydrase active sites – 900 Carbonic anhydrase half-maximum constant for CO2, Kca µM 104.7 Internal pH – 8.3 pKaeff for HCO3–:CO2 – 5.84 Carboxysome permeability cm s–1 3 × 10–5 HCO3– uptake velocity, jc cm s–1 2.4 × 10–7 CO2 to HCO3– conversion at membrane cm s–1 0.6 × 10–7 The cell radius was measured from a bioimage collected using fluorescence microscopy (Supplementary Fig. S12). Kinetic constants of Rubisco carboxylation (KCO2), oxygenation (KO2), and the specificity factor (S) for a form 1B cyanobacteria were taken from Badger et al. (1998). Open in new tab Table 1. Key parameter values used in the numerical simulation of the CCM in Trichodesmium Variable . Units . Model value . Cell radius, Rb µm 3 Carboxysome radius, Rc µm 0.15 Rubisco reaction rate, kRub s–1 per active site 1.92 Rubisco KCO2 µM 145 Rubisco KO2 µM 600 Rubisco specificity, S – 45 Number of Rubisco active sites – 54000 Number of carbonic anhydrase active sites – 900 Carbonic anhydrase half-maximum constant for CO2, Kca µM 104.7 Internal pH – 8.3 pKaeff for HCO3–:CO2 – 5.84 Carboxysome permeability cm s–1 3 × 10–5 HCO3– uptake velocity, jc cm s–1 2.4 × 10–7 CO2 to HCO3– conversion at membrane cm s–1 0.6 × 10–7 Variable . Units . Model value . Cell radius, Rb µm 3 Carboxysome radius, Rc µm 0.15 Rubisco reaction rate, kRub s–1 per active site 1.92 Rubisco KCO2 µM 145 Rubisco KO2 µM 600 Rubisco specificity, S – 45 Number of Rubisco active sites – 54000 Number of carbonic anhydrase active sites – 900 Carbonic anhydrase half-maximum constant for CO2, Kca µM 104.7 Internal pH – 8.3 pKaeff for HCO3–:CO2 – 5.84 Carboxysome permeability cm s–1 3 × 10–5 HCO3– uptake velocity, jc cm s–1 2.4 × 10–7 CO2 to HCO3– conversion at membrane cm s–1 0.6 × 10–7 The cell radius was measured from a bioimage collected using fluorescence microscopy (Supplementary Fig. S12). Kinetic constants of Rubisco carboxylation (KCO2), oxygenation (KO2), and the specificity factor (S) for a form 1B cyanobacteria were taken from Badger et al. (1998). Open in new tab Results Inorganic carbon chemistry, growth rate, and cell composition Overall, the CO2 drawdown in the cultures ranged between 57–78 µmol mol–1 for all CO2 treatments (Table 2) and exhibited a negligible CO2 drift over a diurnal cycle (see Supplementary Fig. S2). Dissolved inorganic NH4+ concentrations in the growth medium were ~1 µM, while NO3– concentrations were ~0.3 µM, which was below the 1 µM detection limit. Table 2. The growth conditions (±SE) achieved for T. erythraeum IMS101 when cultured at three target gas-phase CO2 concentrations (Low=180 µmol mol–1, Mid=380 µmol mol–1, and High=720 µmol mol–1), saturating light intensity (400 µmol photons m–2 s–1), and optimal temperature (26 °C) Variable . Units . Low CO2 . Mid CO2 . High CO2 . pH – 8.458 8.174 7.906 H+ nM 3.5 (0.20) 6.7 (0.13) 12.4 (0.28) AT µM 2431 (70) 2447 (54) 2442 (56) TIC µM 1800 (69) 2039 (46) 2201 (50) HCO3– µM 1362 (67) 1743 (39) 2005 (44) CO32– µM 435 (16) 289 (9) 179 (6) CO2 µM 3.3 (0.3) 8.1 (0.2) 17.3 (0.5) NH4+ µM 1.03 (0.14) 1.00 (0.08) 1.08 (0.06) NO3– µM 0.34 (0.05) 0.32 (0.03) 0.30 (0.02) n 89 67 39 Variable . Units . Low CO2 . Mid CO2 . High CO2 . pH – 8.458 8.174 7.906 H+ nM 3.5 (0.20) 6.7 (0.13) 12.4 (0.28) AT µM 2431 (70) 2447 (54) 2442 (56) TIC µM 1800 (69) 2039 (46) 2201 (50) HCO3– µM 1362 (67) 1743 (39) 2005 (44) CO32– µM 435 (16) 289 (9) 179 (6) CO2 µM 3.3 (0.3) 8.1 (0.2) 17.3 (0.5) NH4+ µM 1.03 (0.14) 1.00 (0.08) 1.08 (0.06) NO3– µM 0.34 (0.05) 0.32 (0.03) 0.30 (0.02) n 89 67 39 Individual pH values were converted to a H+ concentration, allowing a mean pH value (Total scale) to be calculated. Dissolved inorganic NH4+ was determined using the phenol-hypochlorite method as described by Solorzano (1969), while dissolved inorganic NO3– was determined using the spectrophotometric method as described by Collos et al. (1999). Open in new tab Table 2. The growth conditions (±SE) achieved for T. erythraeum IMS101 when cultured at three target gas-phase CO2 concentrations (Low=180 µmol mol–1, Mid=380 µmol mol–1, and High=720 µmol mol–1), saturating light intensity (400 µmol photons m–2 s–1), and optimal temperature (26 °C) Variable . Units . Low CO2 . Mid CO2 . High CO2 . pH – 8.458 8.174 7.906 H+ nM 3.5 (0.20) 6.7 (0.13) 12.4 (0.28) AT µM 2431 (70) 2447 (54) 2442 (56) TIC µM 1800 (69) 2039 (46) 2201 (50) HCO3– µM 1362 (67) 1743 (39) 2005 (44) CO32– µM 435 (16) 289 (9) 179 (6) CO2 µM 3.3 (0.3) 8.1 (0.2) 17.3 (0.5) NH4+ µM 1.03 (0.14) 1.00 (0.08) 1.08 (0.06) NO3– µM 0.34 (0.05) 0.32 (0.03) 0.30 (0.02) n 89 67 39 Variable . Units . Low CO2 . Mid CO2 . High CO2 . pH – 8.458 8.174 7.906 H+ nM 3.5 (0.20) 6.7 (0.13) 12.4 (0.28) AT µM 2431 (70) 2447 (54) 2442 (56) TIC µM 1800 (69) 2039 (46) 2201 (50) HCO3– µM 1362 (67) 1743 (39) 2005 (44) CO32– µM 435 (16) 289 (9) 179 (6) CO2 µM 3.3 (0.3) 8.1 (0.2) 17.3 (0.5) NH4+ µM 1.03 (0.14) 1.00 (0.08) 1.08 (0.06) NO3– µM 0.34 (0.05) 0.32 (0.03) 0.30 (0.02) n 89 67 39 Individual pH values were converted to a H+ concentration, allowing a mean pH value (Total scale) to be calculated. Dissolved inorganic NH4+ was determined using the phenol-hypochlorite method as described by Solorzano (1969), while dissolved inorganic NO3– was determined using the spectrophotometric method as described by Collos et al. (1999). Open in new tab Balanced growth rates increased from ~0.2 d–1 at low CO2 to ~0.34 d–1 at mid-CO2 and ~0.36 d–1 at high CO2 (Table 3). The dark-adapted photochemical efficiencies of PSII (Fv/Fm) were proportionate to the CO2 treatment, increasing from 0.27 at low CO2 to ~0.31 at mid-CO2 and ~0.34 at high CO2 (Table 3). The particulate C:N ratio was independent of CO2, while the C:P and N:P ratios increased with increasing CO2 (Table 3). Both Chl a:C and Chl a:N ratios were about 30–40% higher at mid-CO2 than at low or high CO2. Table 3. The mean (±SE) balanced growth rate, dark-adapted photochemical efficiency of PSII (Fv/Fm), elemental stoichiometry, and chlorophyll a to C and N ratios for T. erythraeum IMS101 when acclimated to three target CO2 concentrations (Low=180 µmol mol–1, Mid=380 µmol mol–1, and High=720 µmol mol–1), saturating light intensity (400 µmol photons m–2 s–1), and optimal temperature (26 °C) Variable . Units . Low CO2 . Mid CO2 . High CO2 . Growth rate d–1 0.198 (0.027)A 0.336 (0.026)B 0.361 (0.020)B Fv/Fm dimensionless 0.274 (0.025)A 0.305 (0.020)B 0.342 (0.037)C Elemental stoichiometry C:N mol:mol 7.9 (0.8) 7.8 (0.3) 7.3 (0.8) C:P mol:mol 91.9 (6.3)A 143.6 (6.3)B 155.5 (13.5)B N:P mol:mol 11.9 (0.6)A 18.4 (0.7)B 21.8 (1.7)B Chl a:C g:mol 0.052 (0.003)A 0.089 (0.003)C 0.066 (0.003)B Chl a:N g:mol 0.401 (0.037)A 0.693 (0.035)B 0.474 (0.043)A Variable . Units . Low CO2 . Mid CO2 . High CO2 . Growth rate d–1 0.198 (0.027)A 0.336 (0.026)B 0.361 (0.020)B Fv/Fm dimensionless 0.274 (0.025)A 0.305 (0.020)B 0.342 (0.037)C Elemental stoichiometry C:N mol:mol 7.9 (0.8) 7.8 (0.3) 7.3 (0.8) C:P mol:mol 91.9 (6.3)A 143.6 (6.3)B 155.5 (13.5)B N:P mol:mol 11.9 (0.6)A 18.4 (0.7)B 21.8 (1.7)B Chl a:C g:mol 0.052 (0.003)A 0.089 (0.003)C 0.066 (0.003)B Chl a:N g:mol 0.401 (0.037)A 0.693 (0.035)B 0.474 (0.043)A Replicates comprised n=9 at low CO2, n=6 at mid- and high CO2. Letters indicate significant differences between CO2 treatments (one-way ANOVA, Tukey post hoc test; P<0.05); where B is significantly greater than A, and C is significantly greater than B and A. Open in new tab Table 3. The mean (±SE) balanced growth rate, dark-adapted photochemical efficiency of PSII (Fv/Fm), elemental stoichiometry, and chlorophyll a to C and N ratios for T. erythraeum IMS101 when acclimated to three target CO2 concentrations (Low=180 µmol mol–1, Mid=380 µmol mol–1, and High=720 µmol mol–1), saturating light intensity (400 µmol photons m–2 s–1), and optimal temperature (26 °C) Variable . Units . Low CO2 . Mid CO2 . High CO2 . Growth rate d–1 0.198 (0.027)A 0.336 (0.026)B 0.361 (0.020)B Fv/Fm dimensionless 0.274 (0.025)A 0.305 (0.020)B 0.342 (0.037)C Elemental stoichiometry C:N mol:mol 7.9 (0.8) 7.8 (0.3) 7.3 (0.8) C:P mol:mol 91.9 (6.3)A 143.6 (6.3)B 155.5 (13.5)B N:P mol:mol 11.9 (0.6)A 18.4 (0.7)B 21.8 (1.7)B Chl a:C g:mol 0.052 (0.003)A 0.089 (0.003)C 0.066 (0.003)B Chl a:N g:mol 0.401 (0.037)A 0.693 (0.035)B 0.474 (0.043)A Variable . Units . Low CO2 . Mid CO2 . High CO2 . Growth rate d–1 0.198 (0.027)A 0.336 (0.026)B 0.361 (0.020)B Fv/Fm dimensionless 0.274 (0.025)A 0.305 (0.020)B 0.342 (0.037)C Elemental stoichiometry C:N mol:mol 7.9 (0.8) 7.8 (0.3) 7.3 (0.8) C:P mol:mol 91.9 (6.3)A 143.6 (6.3)B 155.5 (13.5)B N:P mol:mol 11.9 (0.6)A 18.4 (0.7)B 21.8 (1.7)B Chl a:C g:mol 0.052 (0.003)A 0.089 (0.003)C 0.066 (0.003)B Chl a:N g:mol 0.401 (0.037)A 0.693 (0.035)B 0.474 (0.043)A Replicates comprised n=9 at low CO2, n=6 at mid- and high CO2. Letters indicate significant differences between CO2 treatments (one-way ANOVA, Tukey post hoc test; P<0.05); where B is significantly greater than A, and C is significantly greater than B and A. Open in new tab CO2-response curves Based on the shape of the response curves, the inorganic carbon (14C) fixation rate was fitted to a saturating function of the dissolved CO2 concentration in both the pH gradient and TIC gradient experiments (Fig. 1). Although a saturating function of HCO3– concentration was observed when TIC was varied at constant pH (Fig. 1A–C), Ci assimilation could not be described by the same kinetic constants when pH was varied at constant TIC (Fig. 1D–F). Fig. 1. Open in new tabDownload slide (A–C) CO2- and (D–F) HCO3–-response curves for inorganic C-fixation by T. erythraeum IMS101. C-fixation rates are normalised to a carbon h–1 basis. Filled circles indicate data obtained by varying TIC and HCO3– at a fixed pH of ~8.15. Open circles indicate data obtained by varying pH (~7.52–8.54) at a fixed TIC. Differences in the range of HCO3– and CO2 gradients between CO2 treatments were due to variability in pipetting and not from instability in the Ci chemistry. For the CO2 response, curve-fitting was performed using all replicates from both the TIC and pH gradients. For the HCO3– response, curve-fitting was performed using data from the TIC gradient only. The CO2- and HCO3–- response curves for individual experiments are shown in Supplementary Figs S6–S11. The Km for photosynthetic C-fixation increased from 0.8 µM in cultures acclimated to low CO2 to 2.2 µM and 3.2 µM in cultures acclimated to mid- and high CO2, respectively, and were approximately 4- to 5-fold lower than the ambient CO2 concentrations in the cultures. The maximum organic carbon-specific rate of C-fixation (VC,max) was also higher in cells grown at mid-CO2 than at low CO2, although the rates at mid- and high CO2 did not differ significantly (Table 4). The affinity for CO2 (VC,max/Km) declined by about 40% between the low- and high-CO2 treatments (Table 4). Table 4. The physiological parameters (±SE) of the C-specific C-fixation versus CO2 concentration response curves for T. erythraeum IMS101, fitted using the Michaelis–Menten model to obtain estimates using the combined data from all replicates from both experiments employing varied TIC at fixed pH and varied pH at fixed TIC for each CO2 treatment Parameter . Units . Low CO2 . Mid CO2 . High CO2 . VC,max h–1 0.011 (0.0002) 0.024 (0.0007) 0.026 (0.0008) Km µM CO2 0.8 (0.1) 2.2 (0.3) 3.2 (0.4) Affinity mM (CO2)–1 h–1 13.3 (1.7) 10.9 (1.5) 8.0 (1.0) Parameter . Units . Low CO2 . Mid CO2 . High CO2 . VC,max h–1 0.011 (0.0002) 0.024 (0.0007) 0.026 (0.0008) Km µM CO2 0.8 (0.1) 2.2 (0.3) 3.2 (0.4) Affinity mM (CO2)–1 h–1 13.3 (1.7) 10.9 (1.5) 8.0 (1.0) VC,max, the C-specific maximum C-fixation rate; Km, the half saturation constant; Affinity, the C-specific initial slope of the VC,max versus CO2-response curve. Open in new tab Table 4. The physiological parameters (±SE) of the C-specific C-fixation versus CO2 concentration response curves for T. erythraeum IMS101, fitted using the Michaelis–Menten model to obtain estimates using the combined data from all replicates from both experiments employing varied TIC at fixed pH and varied pH at fixed TIC for each CO2 treatment Parameter . Units . Low CO2 . Mid CO2 . High CO2 . VC,max h–1 0.011 (0.0002) 0.024 (0.0007) 0.026 (0.0008) Km µM CO2 0.8 (0.1) 2.2 (0.3) 3.2 (0.4) Affinity mM (CO2)–1 h–1 13.3 (1.7) 10.9 (1.5) 8.0 (1.0) Parameter . Units . Low CO2 . Mid CO2 . High CO2 . VC,max h–1 0.011 (0.0002) 0.024 (0.0007) 0.026 (0.0008) Km µM CO2 0.8 (0.1) 2.2 (0.3) 3.2 (0.4) Affinity mM (CO2)–1 h–1 13.3 (1.7) 10.9 (1.5) 8.0 (1.0) VC,max, the C-specific maximum C-fixation rate; Km, the half saturation constant; Affinity, the C-specific initial slope of the VC,max versus CO2-response curve. Open in new tab Modelled response curves Without parameter-fitting, the CCM model of Trichodesmium produced behaviors consistent with the experimental data when either external TIC (i.e HCO3–) was varied at a fixed pH or when pH was varied at a fixed TIC (Fig. 2A, B). Assuming HCO3– is the dominant form of inorganic carbon taken up by the cell (Kranz et al., 2009; Eichner et al., 2015), Trichodesmium exhibited a significant response to changes in external pH and CO2 concentrations. The decrease in carboxylation rate with decreasing external CO2 was due to a decrease in HCO3– uptake (when TIC was varied) or an increase in HCO3– and CO2 leakage out of the cell (when pH was varied) (Supplementary Fig. S3). Modelled carboxylation rates from both numerical simulations exhibited a smooth function of HCO3– uptake, HCO3– leakage, and CO2 leakage (Fig. 2C). Fig. 2. Open in new tabDownload slide Calculated carboxylation rates obtained from model simulations for T. erythraeum IMS101 as a function of external CO2 (A) and HCO3– (B) concentrations, with TIC (i.e. HCO3–) varied at a fixed pH=8.15 (dashed lines) and pH varied at a fixed HCO3–=1.9 mM (solid lines). Carboxylation rates are also plotted against the net HCO3– uptake rate (C), where HCO3– and CO2 leakage rates were subtracted from the rate of gross HCO3– transport. The VC,max of the pH gradient and TIC gradient experiments were not significantly different (Supplementary Table S3). However, the maximum carboxylation rates from the simulations were significantly different (Fig. 2); principally because the external HCO3– concentration used in the pH-dependent simulation (chosen to be the same as the experiment) was not sufficient to saturate Rubisco. It is possible that the Km value assumed for Rubisco was set too high, or the internal pH, geometry, or HCO3– uptake values were substantially different. Note that we were simulating values beyond the range of those in the experiments, so such a discrepancy is magnified. Discussion The key findings of our study were as follows. The acclimated growth rate increased from low- to mid-CO2 treatments but did not increase significantly between mid- and high-CO2, suggesting that the positive effect of elevated CO2 on Trichodesmium carbon assimilation over the coming decades may only be slight. The maximum rate (VC,max) and the half-saturation constant (Km) for C-fixation increased with increasing CO2 treatment, but the affinity for CO2 (VC,max/Km) declined, which is probably attributable to the activity of the CCM in Trichodesmium. The measured inorganic C-fixation rate in Trichodesmium could be described as a saturating function of CO2, both when CO2 was manipulated by varying pH at constant TIC and when CO2 was manipulated by varying TIC at constant pH. A mechanistic model of the CCM in Trichodesmium indicated that the former was due to HCO3– uptake limitation of carboxylation rate, whereas the latter was due to a high-pH-mediated increases in HCO3– and CO2 leakage, potentially coupled to other unknown processes operating outside of the paramaterised model that were restricting Ci assimilation rates at high pH. Such processes may involve the direct effect of pH on membrane conformation, membrane transport processes, or metabolic functions. Effect of acclimation to variations in inorganic chemistry on growth rates and elemental stoichiometry The increased growth rates that we observed from low- (180 µmol mol–1) to mid- (380 µmol mol–1) and high-CO2 treatments (720 µmol mol–1) were similar to previous findings (Barcelos e Ramos et al., 2007; Boatman et al., 2017, 2018b). The growth rate at high CO2 was 8% greater than at mid-CO2, but this difference was not statistically significant. The magnitude of this increase at high CO2 was comparable to several recent studies, which report growth rate increases of 7–26% with increases of CO2 beyond 400 µmol mol–1 (Barcelos e Ramos et al., 2007; Hutchins et al., 2007; Levitan et al., 2007; Kranz et al., 2010; Garcia et al., 2011; Boatman et al., 2017). The observed increases in C:P and N:P were consistent with previous findings (Barcelos e Ramos et al., 2007; Kranz et al., 2010; Levitan et al., 2010), with changes that can be ascribed to increases in cellular N and C incorporation, with P content relatively unaffected by CO2 (Hutchins et al., 2007; Kranz et al., 2010). In contrast, the C:N ratio and thus the balance between CO2 fixation and N2 fixation was not significantly affected by the CO2 treatment. Similarly, Levitan et al. (2007) found that C:N varied only slightly (from 6.5 to 7.0) across growth CO2 concentrations ranging from 250 to 900 µmol mol–1. We report C-specific rates here as these are most directly related to changes in specific growth rate because both rates can be expressed in equivalent units of inverse time (e.g. h–1 or d–1). However, we note that due to differences in the Chl a:C ratio, chlorophyll a-specific rates showed a different pattern, increasing progressively from low through mid- to high CO2 (see Supplementary Table S2, Supplementary Fig. S4). A reduction in Chl a:C decreases the energy demands associated with synthesis of the photosynthetic apparatus and is dictated by the total demands for reductant (NADPH) and high-energy phosphate bonds (ATP) (Geider et al., 2009), the minimum turnover times for PSII (τPIIʹ) and PSI (τPIʹ), and the minimum pigment content required for effective light absorption and energy transfer (aminʹ) (Behrenfeld et al., 2008). We suggest that the reduced Chl a:C at low CO2 relative to mid-CO2 was probably due to the cost of up-regulating the CCM, whereas the reduced Chl a:C at high CO2 may have been due to an increase in carbohydrate storage granules relative to the mid-CO2 treatment (Table 3). CO2-response curves The growth rates reported here were comparable to the 2-µM EDTA, iron-replete (unchelated) treatments in Boatman et al. (2017), as well as 20-µM EDTA, iron-replete (chelated) cultures (Boatman et al., 2018b), which suggests that our cultures were not exposed to toxic concentrations of certain trace metals (e.g. copper) caused from low trace metal buffering capacity, as reported by Hong et al. (2017). Furthermore, dissolved inorganic NH4+ concentrations were consistently around 1.0 µM (Table 2). We are therefore confident that the observed positive effect of ocean acidification on growth and primary productivity is driven by the increased CO2 concentration, rather than being a consequence of a pH-induced shift of the NH3/NH4+ equilibrium. We determined CO2-response curves at one time of day (4–6 h into the photoperiod of a 12/12 h L/D cycle) and as such cannot extrapolate to a diel response given the reports of temporal separation of photosynthesis and N2 fixation in Trichodesmium (Berman-Frank et al., 2001). The mechanistic model of Mangan et al. (2016) indicates that the CO2 response we observed when the TIC was varied (pH fixed) was caused by HCO3– limitation, where HCO3– uptake limits the rate of carboxylation. Conversely, the CO2 response we observed when pH was varied (TIC fixed) was a function of the pH dependency of HCO3– and CO2 leakage, which in turn could lead to CO2 limitation of C-fixation and/or diversion of reducing equivalents from powering CO2 fixation via the Calvin cycle to powering the conversion of CO2 to HCO3– by the NDH-I4 complex. The model of the CCM in Trichodesmium showed the relative importance of leakage, which is notably sensitive to certain parameters in the system such as internal pH, Rubisco activity, cell size, and carboxysome size. Previous studies have shown a notable response in CCM activity to changes in CO2; for example, a two-fold lower dissolved inorganic carbon half-saturation concentration in cells acclimated to 150 µmol mol–1 (pH 8.56) compared with 370 µmol mol–1 (pH 8.26) (Kranz et al., 2009). Our experimental observations indicated that Ci assimilation (VC) was well described by a CO2-response curve, but not by a single HCO3–-response curve (Fig. 1). We now offer an explanation as to the response of VC to HCO3– concentration in the experiments where we varied pH from 7.65 to 8.5 at constant TIC. Based on the numerical simulations, carboxylation rates across an external pH gradient ranging from 7.5 to 8.5 exhibited a clear linear response, which could not be ascribed to a Michaelis–Menten function (see Supplementary Fig. S3). Conversely, our experimental data showed a clear and significant decrease in Ci assimilation rates at low external CO2/high pH (Fig. 1). In addition, the Ci assimilation rates for the pH-gradient and TIC-gradient experiments, for all replicates of all three CO2 treatments, exhibited similar inflection points to external CO2 (Supplementary Fig. S5). In order for the simulated system to exhibit a rate-saturating response to external CO2, CO2 would have to be the dominant source of inorganic carbon. This would contradict all previous research showing that HCO3– accounts for >90% of inorganic carbon uptake (Kranz et al., 2009, 2010) and the currently accepted mechanism of Ci assimilation in T. erythraeum IMS101 (Badger and Price, 2003). Given how well the numerical simulations modelled carboxylation rates as a smooth function of HCO3– uptake, HCO3– leakage, and CO2 leakage (Fig. 2C), we propose that the linear pH-dependency of carboxylation rate predicted by the model is mechanistically correct, but that processes not captured by the model are contributing to the decrease in Ci assimilation rate at high pH. Such factors could include a direct effect of high pH on cell membrane properties and alteration in membrane conformation (Myklestad and Swift, 1998), or the influence of pH on membrane transport processes and metabolic functions involved in cellular pH regulation (Raven, 1981). Interestingly, for the mid- and high-CO2 treatments, a Michaelis–Menten function provided a better fit for the pH-varied (TIC fixed) data than a linear regression. However, there was no significant difference between a linear or Michaelis–Menten function for the low-CO2 data, which suggests that full acclimation to a high-pH environment prior to the 14C incubations lessened the negative effect that high pH had on Ci assimilation. Based on our simulation, the actual carboxylation rate of Trichodesmium should be modelled as a function of HCO3– and pH. This is because the CO2 concentration in a saturated HCO3–/high-pH environment (i.e. 3.8 mM HCO3–, pH=8.4) could be equivalent to a limited HCO3–/present-day pH environment (i.e. 1.9 mM HCO3–, pH=8.1); which for the aforementioned reasons will impose different constraints on leakage/uptake rates. That said, our experimental data clearly suggested that high-pH-induced processes operating outside of the CCM were contributing to decrease Ci assimilation. Overall, this may allow the Ci assimilation rates of Trichodesmium to be ascribed as a function of CO2 (Fig. 1, see Supplementary Fig. S4), which would be considerably simpler to implement in biogeochemical models of Trichodesmium growth and photosynthesis (Hutchins et al., 2013) than a HCO3–-response curve in which the kinetic constants (Km and Vm) are pH-dependent. Further experimental work is needed to assess whether a CO2 parameterisation is consistent across a more extended range of pH and HCO3– conditions than those used in our experiments. Conclusions Climate change is driving ocean acidification, which results in higher CO2 and HCO3– concentrations and a decrease in pH. We observed systematic changes in the kinetics of inorganic carbon assimilation of T. erythraeum IMS101 in response to acclimation to increasing CO2 concentrations ranging from low CO2 (levels at the last glacial maximum) through mid-CO2 (levels at the end of the 20th century), to high CO2 (levels predicted for 2050–2100). Extrapolating these responses to future scenarios of the natural environment should take into account the fact that our findings were obtained using acclimation experiments whereas Trichodesmium may adapt to future conditions (Hutchins et al., 2015), that variability may exist between strains and clades (Hutchins et al., 2013), and that there will be additional effects of integrated abiotic variables (e.g. light and temperature) and nutrients (e.g. P and Fe) on Trichodesmium productivity (Walworth et al., 2016; Boatman et al., 2018a, 2018b). In the context of the open oceans, our results indicate that nutrient-replete net photosynthesis and growth rates of T. erythraeum IMS101 would have been severely CO2-limited at the last glacial maximum relative to current conditions. However, future increases in CO2 (i.e. 720 µmol mol–1) may not significantly increase its growth and productivity, although we note that other studies have reported a stimulation of growth and photosynthesis by increasing CO2 beyond current ambient concentrations (Hutchins et al., 2007; Levitan et al., 2007, 2010). On the other hand, we did observe that growth under high CO2 will increase key stoichiometric ratios (N:P and C:P). Increases of N:P and C:P in Trichodesmium-dominated oceanic regimes may affect bacterial and zooplankton metabolism, the pool of bioavailable nitrogen, the depth at which sinking organic matter is remineralised, and consequently carbon sequestration via the biological carbon pump (Mulholland et al., 2004; McGillicuddy, 2014). These responses could serve as a negative feedback to climate change by increasing new N and C production and thereby increasing the organic carbon sinking to the deep ocean. Supplementary data Supplementary data are available at JXB online. Information SI. Calculation of inorganic carbon speciation. Information SII. Preparation of medium for CO2-response curves where TIC was varied at fixed pH. Information SIII. Preparation of medium for CO2-response curves where pH was varied at fixed TIC. Table S1. Recent literature on the C- and N2-fixation rates and elemental stoichiometry of T. erythraeum IMS101 in response to CO2, temperature, and light. Table S2. The Chl a-specific curve-fitting parameter values of the carbon assimilation–CO2-response curves when the ‘TIC varied/pH fixed’ and ‘pH varied/TIC fixed’ data were modelled separately. Table S3. The carbon-specific curve-fitting parameter values of the carbon assimilation–CO2-response curves when the ‘TIC varied/pH fixed’ and ‘pH varied/TIC fixed’ data were modelled separately. Fig. S1. The effect of filtration/re-suspension and incubation on photosynthetic efficiency. Fig. S2. The inorganic carbon chemistry of the culture vessels over a diurnal period. Fig. S3. The modelled rates of carboxylation, CO2 leakage, HCO3–, uptake and HCO3– leakage for a Trichodesmium cell. Fig. S4. The Chl a-specific curve fits of the carbon assimilation–CO2-response curves when the ‘TIC varied/pH fixed’ and ‘pH varied/TIC fixed’ data were modelled together. Fig. S5. The Chl a- and carbon-specific curve fits of the carbon assimilation–CO2-response curves when the ‘TIC varied/pH fixed’ and ‘pH varied/TIC fixed’ data were modelled seperately. Fig. S6. The Chl a- and carbon-specific curve fits of the carbon assimilation–CO2-response curves of the low-CO2 treatment for the ‘TIC varied/pH fixed’ data. Fig. S7. The Chl a- and carbon-specific curve fits of the carbon assimilation–CO2-response curves of the low-CO2 treatment for the ‘pH varied/TIC fixed’ data. Fig. S8. The Chl a- and carbon-specific curve fits of the carbon assimilation–CO2-response curves of the mid-CO2 treatment for the ‘TIC varied/pH fixed’ data. Fig. S9. The Chl a- and carbon-specific curve fits of the carbon assimilation–CO2-response curves of the mid-CO2 treatment for the ‘pH varied/TIC fixed’ data. Fig. S10. The Chl a- and carbon-specific curve fits of the carbon assimilation–CO2-response curves of the high-CO2 treatment for the ‘TIC varied/pH fixed’ data. Fig. S11. The Chl a- and carbon-specific curve fits of the carbon assimilation–CO2-response curves of the high-CO2 treatment for the ‘pH varied/TIC fixed’ data. Fig. S12. A bioimage of T. erythraeum IMS101 filaments cultured at mid-CO2, saturating light, and optimal temperature. Acknowledgements Tobias G. Boatman was supported by a UK Natural Environment Research Council PhD studentship (NE/J500379/1 DTB). References Badger MR , Andrews TJ. 1987 . Co-evolution of Rubisco and CO2 concentrating mechanisms. In: Biggins J, ed . Progress in photosynthesis research . Dordrecht: Springer , 601 – 609 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Badger MR , Andrews TJ, Whitney S, Ludwig M, Yellowlees DC, Leggat W, Price GD. 1998 . The diversity and coevolution of Rubisco, plastids, pyrenoids, and chloroplast-based CO2-concentrating mechanisms in algae . Canadian Journal of Botany 76 , 1052 – 1071 . Google Scholar Crossref Search ADS WorldCat Badger MR , Hanson D, Price GD. 2002 . Evolution and diversity of CO2 concentrating mechanisms in cyanobacteria . Functional Plant Biology 29 , 161 – 173 . Google Scholar Crossref Search ADS WorldCat Badger MR , Palmqvist K, Yu JW. 1994 . Measurement of CO2 and HCO3− fluxes in cyanobacteria and microalgae during steady-state photosynthesis . Physiologia Plantarum 90 , 529 – 536 . Google Scholar Crossref Search ADS WorldCat Badger MR , Price GD. 2003 . CO2 concentrating mechanisms in cyanobacteria: molecular components, their diversity and evolution . Journal of Experimental Botany 54 , 609 – 622 . Google Scholar Crossref Search ADS PubMed WorldCat Badger MR , Price GD, Long BM, Woodger FJ. 2006 . The environmental plasticity and ecological genomics of the cyanobacterial CO2 concentrating mechanism . Journal of Experimental Botany 57 , 249 – 265 . Google Scholar Crossref Search ADS PubMed WorldCat Barcelos e Ramos J , Biswas H, Schulz KG, LaRoche J, Riebesell U. 2007 . Effect of rising atmospheric carbon dioxide on the marine nitrogen fixer Trichodesmium . Global Biogeochemical Cycles 21 , GB2028 . Google Scholar Crossref Search ADS WorldCat Behrenfeld MJ , Halsey KH, Milligan AJ. 2008 . Evolved physiological responses of phytoplankton to their integrated growth environment . Philosophical Transactions of the Royal Society of London. Series B, Biological sciences 363 , 2687 – 2703 . Google Scholar Crossref Search ADS PubMed WorldCat Berman-Frank I , Lundgren P, Chen YB, Küpper H, Kolber Z, Bergman B, Falkowski P. 2001 . Segregation of nitrogen fixation and oxygenic photosynthesis in the marine cyanobacterium Trichodesmium . Science 294 , 1534 – 1537 . Google Scholar Crossref Search ADS PubMed WorldCat Boatman TG , Lawson T, Geider RJ. 2017 . A key marine diazotroph in a changing ocean: the interacting effects of temperature, CO2 and light on the growth of Trichodesmium erythraeum IMS101 . PLoS ONE 12 , e0168796 . Google Scholar Crossref Search ADS PubMed WorldCat Boatman TG , Davey PA, Lawson T, Geider RJ. 2018a . The physiological cost of diazotrophy for Trichodesmium erythraeum IMS101 . PLoS ONE 13 , e0195638 . Google Scholar Crossref Search ADS WorldCat Boatman TG , Oxborough K, Gledhill M, Lawson T, Geider RJ. 2018b . An integrated response of Trichodesmium erythraeum IMS101 growth and photo-physiology to iron, CO2, and light intensity . Frontiers in Microbiology 9 , 624 . Google Scholar Crossref Search ADS WorldCat Bronk DA , Ward BB. 2000 . Magnitude of dissolved organic nitrogen release relative to gross nitrogen uptake in marine systems . Limnology and Oceanography 45 , 1879 – 1883 . Google Scholar Crossref Search ADS WorldCat Capone DG , Burns JA, Montoya JP, Subramaniam A, Mahaffey C, Gunderson T, Michaels AF, Carpenter EJ. 2005 . Nitrogen fixation by Trichodesmium spp.: an important source of new nitrogen to the tropical and subtropical North Atlantic Ocean . Global Biogeochemical Cycles 19 , GB2024 . Google Scholar Crossref Search ADS WorldCat Chen YB , Zehr JP, Mellon M. 1996 . Growth and nitrogen fixation of the diazotrophic filamentous nonheterocystous cyanobacterium Trichodesmium sp. IMS 101 in defined media: evidence for a circadian rhythm . Journal of Phycology 32 , 916 – 923 . Google Scholar Crossref Search ADS WorldCat Collos Y , Mornet F, Sciandra A, Waser N, Larson A, Harrison P. 1999 . An optical method for the rapid measurement of micromolar concentrations of nitrate in marine phytoplankton cultures . Journal of Applied Phycology 11 , 179 – 184 . Google Scholar Crossref Search ADS WorldCat Davis CS , McGillicuddy DJ Jr. 2006 . Transatlantic abundance of the N2-fixing colonial cyanobacterium Trichodesmium . Science 312 , 1517 – 1520 . Google Scholar Crossref Search ADS PubMed WorldCat Eichner M , Thoms S, Kranz SA, Rost B. 2015 . Cellular inorganic carbon fluxes in Trichodesmium: a combined approach using measurements and modelling . Journal of Experimental Botany 66 , 749 – 759 . Google Scholar Crossref Search ADS PubMed WorldCat Garcia NS , Fu FX, Breene CL, Bernhardt PW, Mulholland MR, Sohm JA, Hutchins DA. 2011 . Interactive effects of irradiance and CO2 on CO2 fixation and N2 fixation in the diazotroph Trichodesmium erythraeum (Cyanobacteria) . Journal of Phycology 47 , 1292 – 1303 . Google Scholar Crossref Search ADS PubMed WorldCat Geider RJ , Moore CM, Ross ON. 2009 . The role of cost–benefit analysis in models of phytoplankton growth and acclimation . Plant Ecology and Diversity 2 , 165 – 178 . Google Scholar Crossref Search ADS WorldCat Hong H , Shen R, Zhang F, et al. 2017 . The complex effects of ocean acidification on the prominent N2-fixing cyanobacterium Trichodesmium . Science 356 , 527 – 531 . Google Scholar Crossref Search ADS PubMed WorldCat Hutchins DA , Fu F-X, Webb EA, Walworth N, Tagliabue A. 2013 . Taxon-specific response of marine nitrogen fixers to elevated carbon dioxide concentrations . Nature Geoscience 6 , 790 – 795 . Google Scholar Crossref Search ADS WorldCat Hutchins DA , Fu F-X, Zhang Y, Warner M, Feng Y, Portune K, Bernhardt P, Mulholland M. 2007 . CO2 control of Trichodesmium N2 fixation, photosynthesis, growth rates, and elemental ratios: implications for past, present, and future ocean biogeochemistry . Limnology and Oceanography 52 , 1293 – 1304 . Google Scholar Crossref Search ADS WorldCat Hutchins DA , Walworth NG, Webb EA, Saito MA, Moran D, McIlvin MR, Gale J, Fu F-X. 2015 . Irreversibly increased nitrogen fixation in Trichodesmium experimentally adapted to elevated carbon dioxide . Nature Communications 6 , 8155 . Google Scholar Crossref Search ADS PubMed WorldCat Kaplan A , Reinhold L. 1999 . CO2 concentrating mechanisms in photosynthetic microorganisms . Annual Review of Plant Physiology and Plant Molecular Biology 50 , 539 – 570 . Google Scholar Crossref Search ADS PubMed WorldCat Kranz SA , Eichner M, Rost B. 2011 . Interactions between CCM and N2 fixation in Trichodesmium . Photosynthesis Research 109 , 73 – 84 . Google Scholar Crossref Search ADS PubMed WorldCat Kranz SA , Levitan O, Richter KU, Prásil O, Berman-Frank I, Rost B. 2010 . Combined effects of CO2 and light on the N2-fixing cyanobacterium Trichodesmium IMS101: physiological responses . Plant Physiology 154 , 334 – 345 . Google Scholar Crossref Search ADS PubMed WorldCat Kranz SA , Sültemeyer D, Richter KU, Rost B. 2009 . Carbon acquisition in Trichodesmium: the effect of pCO2 and diurnal changes . Limnology and Oceanography 54 , 548 – 559 . Google Scholar Crossref Search ADS WorldCat Levitan O , Brown CM, Sudhaus S, Campbell D, LaRoche J, Berman-Frank I. 2010 . Regulation of nitrogen metabolism in the marine diazotroph Trichodesmium IMS101 under varying temperatures and atmospheric CO2 concentrations . Environmental Microbiology 12 , 1899 – 1912 . Google Scholar Crossref Search ADS PubMed WorldCat Levitan O , Rosenberg G, Setlik I, Setlikova E, Grigel J, Klepetar J, Prasil O, Berman-Frank I. 2007 . Elevated CO2 enhances nitrogen fixation and growth in the marine cyanobacterium Trichodesmium . Global Change Biology 13 , 531 – 538 . Google Scholar Crossref Search ADS WorldCat Li W , Dickie P. 1991 . Light and dark 14C uptake in dimly-lit oligotrophic waters: relation to bacterial activity . Journal of Plankton Research 13 , 29 – 44 . Google Scholar OpenURL Placeholder Text WorldCat Mangan NM , Brenner MP. 2014 . Systems analysis of the CO2 concentrating mechanism in cyanobacteria . eLIFE 3 , e02043 . Google Scholar Crossref Search ADS WorldCat Mangan NM , Flamholz A, Hood RD, Milo R, Savage DF. 2016 . pH determines the energetic efficiency of the cyanobacterial CO2 concentrating mechanism . Proceedings of the National Academy of Sciences, USA 113 , E5354 – E5362 . Google Scholar Crossref Search ADS WorldCat McGillicuddy DJ . 2014 . Do Trichodesmium spp. populations in the North Atlantic export most of the nitrogen they fix ? Global Biogeochemical Cycles 28 , 103 – 114 . Google Scholar Crossref Search ADS WorldCat Mulholland MR , Bronk DA, Capone DG. 2004 . Dinitrogen fixation and release of ammonium and dissolved organic nitrogen by Trichodesmium IMS101 . Aquatic Microbial Ecology 37 , 85 – 94 . Google Scholar Crossref Search ADS WorldCat Myklestad S , Swift E. 1998 . A new method for measuring soluble cellular organic content and a membrane property, Tm, of planktonic algae . European Journal of Phycology 33 , 333 – 336 . Google Scholar Crossref Search ADS WorldCat Price GD , Badger MR, Woodger FJ, Long BM. 2008 . Advances in understanding the cyanobacterial CO2-concentrating-mechanism (CCM): functional components, Ci transporters, diversity, genetic regulation and prospects for engineering into plants . Journal of Experimental Botany 59 , 1441 – 1461 . Google Scholar Crossref Search ADS PubMed WorldCat Raven JA . 1981 . Nutrient transport in microalgae . Advances in Microbial Physiology 21 , 47 – 226 . Google Scholar Crossref Search ADS WorldCat Raven J , Caldeira K, Elderfield H, Hoegh-Guldberg O, Liss P, Riebesell U, Shepherd J, Turley C, Watson A. 2005 . Ocean acidification due to increasing atmospheric carbon dioxide. Policy Document 12/05. London: The Royal Society . Google Scholar OpenURL Placeholder Text WorldCat Ritchie R . 2008 . Universal chlorophyll equations for estimating chlorophylls a, b, c, and d and total chlorophylls in natural assemblages of photosynthetic organisms using acetone, methanol, or ethanol solvents . Photosynthetica 46 , 115 – 126 . Google Scholar Crossref Search ADS WorldCat Schwarz R , Reinhold L, Kaplan A. 1995 . Low activation state of ribulose-1,5-bisphosphate carboxylase/oxygenase in carboxysome-defective Synechococcus mutants . Plant Physiology 108 , 183 – 190 . Google Scholar Crossref Search ADS PubMed WorldCat Solorzano L . 1969 . Determination of ammonia in natural waters by the phenol hypochlorite method . Limnology and Oceanography 14 , 799 – 801 . Google Scholar Crossref Search ADS WorldCat Solorzano L , Sharp JH. 1980 . Determination of total dissolved phosphorus and particulate phosphorus in natural waters . Limnology and Oceanography 25 , 754 – 758 . Google Scholar Crossref Search ADS WorldCat Steemann Nielsen E , Jensen EA. 1957 . Primary oceanic production. The autotrophic production of organic matter in the oceans. In: Bruun AF, Greve SV, Spärck R, eds. Galathea Report, vol. 1. Copenhagen: The Galathea Committee , 49 – 135 . Walworth NG , Fu FX, Webb EA, Saito MA, Moran D, Mcllvin MR, Lee MD, Hutchins DA. 2016 . Mechanisms of increased Trichodesmium fitness under iron and phosphorus co-limitation in the present and future ocean . Nature Communications 7 , 12081 . Google Scholar Crossref Search ADS PubMed WorldCat Welschmeyer NA . 1994 . Fluorometric analysis of chlorophyll a in the presence of chlorophyll b and pheopigments . Limnology and Oceanography 39 , 1985 – 1992 . Google Scholar Crossref Search ADS WorldCat Zeebe RE , Wolf-Gladrow D, Jansen H. 1999 . On the time required to establish chemical and isotopic equilibrium in the carbon dioxide system in seawater . Marine Chemistry 65 , 135 – 153 . Google Scholar Crossref Search ADS WorldCat Zeebe RE , Wolf-Gladrow DA. 2001 . CO2 in seawater: equilibrium, kinetics, isotopes . Elsevier Oceanography Series 65 . Elsevier. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Author notes Chelsea Technologies Group Ltd, 55 Central Avenue, West Molesey, Surrey KT8 2QZ, UK © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.

Fernández-San Millán, Alicia; Aranjuelo, Iker; Douthe, Cyril; Nadal, Miquel; Ancín, María; Larraya, Luis; Farran, Inmaculada; Flexas, Jaume; Veramendi, Jon

doi: 10.1093/jxb/ery148pmid: 29912355

Abstract The leaf mesophyll CO2 conductance and the concentration of CO2 within the chloroplast are major factors affecting photosynthetic performance. Previous studies have shown that the aquaporin NtAQP1 (which localizes to the plasma membrane and chloroplast inner envelope membrane) is involved in CO2 permeability in the chloroplast. Levels of NtAQP1 in plants genetically engineered to overexpress the protein correlated positively with leaf mesophyll CO2 conductance and photosynthetic rate. In these studies, the nuclear transformation method used led to changes in NtAQP1 levels in the plasma membrane and the chloroplast inner envelope membrane. In the present work, NtAQP1 levels were increased up to 16-fold in the chloroplast membranes alone by the overexpression of NtAQP1 from the plastid genome. Despite the high NtAQP1 levels achieved, transplastomic plants showed lower photosynthetic rates than wild-type plants. This result was associated with lower Rubisco maximum carboxylation rate and ribulose 1,5-bisphosphate regeneration. Transplastomic plants showed reduced mesophyll CO2 conductance but no changes in chloroplast CO2 concentration. The absence of differences in chloroplast CO2 concentration was associated with the lower CO2 fixation activity of the transplastomic plants. These findings suggest that non-functional pores of recombinant NtAQP1 may be produced in the chloroplast inner envelope membrane. Aquaporin, chloroplast envelope, CO2 permeability, plastid transformation, protein targeting, tobacco Introduction It is predicted that future increases in the human population will require a 30% increase in crop yield rates (Edgerton, 2009). Improving the photosynthetic performance of crops is one way in which plant production might be increased (Parry et al., 2011; Reynolds et al., 2011; Parry et al., 2013; Flexas et al., 2013), and a number of strategies have been identified that, either individually or in combination, might achieve this (Long et al., 2006; Flexas et al., 2006; Parry et al., 2013; Flexas et al., 2012). Photosynthetic performance is affected by two major factors: the concentration of CO2 within the chloroplast and the efficiency of the carboxylation biochemistry. Availability of CO2 at the carboxylation site in the chloroplast can be limited by its diffusion into the substomatal cavities, referred to as stomatal conductance (gs), and by the conductance of CO2 from the substomatal cavity to the chloroplast, referred to as mesophyll conductance (gm). Classically, gm has been described not to limit photosynthesis, and the CO2 concentration was thought to be similar in the substomatal cavity (Ci) and in the chloroplast stroma (Cc). However, over the past decade, a number of studies (Flexas et al., 2006; Scafaro et al., 2011; Kaldenhoff, 2012; Evans and von Caemmerer, 2013; Flexas et al., 2013) have shown that gm has a major influence on CO2 diffusion into the chloroplast, with a consequent impact on the photosynthetic rate. At the cellular level, atmospheric CO2 has to pass through the cell wall and three membranes (the plasma membrane and the two membranes of the chloroplast envelope) to reach the chloroplast stroma. The CO2 permeability of the chloroplast envelope is low, probably due to its relatively large protein content (Priestley and Woolhouse, 1980); indeed, it was estimated that it may account for almost half of the internal leaf resistance to CO2 (Uehlein et al., 2008). As a result, under light-saturated conditions, photosynthesis is limited by the availability of CO2 within the chloroplast. Other studies have shown that the gm can change quickly in response to varying environmental conditions, such as leaf temperature (Bernacchi et al., 2002), water stress (Galmés et al., 2007), blue light (Loreto et al., 2009), and the external CO2 concentration (Flexas et al., 2007a). This rapid modification of gm points to the existence of additional components, some of them probably proteins, controlling the conductance of the mesophyll to CO2 diffusion. Proteins forming pore-like structures, such as aquaporins (AQPs), might help explain how these rapid variations in gm occur. AQPs are small proteins that increase the permeability of cell membranes to water and certain small, neutral molecules, including CO2 (Maurel et al., 2008; Gomes et al., 2009; Chaumont and Tyerman, 2014; Kaldenhoff et al., 2014; Groszmann et al., 2017). AQPs were discovered for the first time in plants in the vacuolar tonoplast of Arabidopsis (Maurel et al., 1993), and are present in the whole plant kingdom. AQPs are located in the plasma membrane and also in most of the intracellular membranes. Many isoforms of AQPs exist, which can be classified according to their sequence homologies and subcellular localization. The plasma membrane intrinsic protein (PIP) class includes isoforms that are most abundant in the plasma membrane. This class can be subdivided into subclasses PIP1 and PIP2 according to sequence similarity. Investigations on the mesophyll cells of tobacco leaves have shown that the plasma membrane protein NtAQP1 (a PIP1 member) facilitates CO2 transport, and that it has important functions in photosynthesis and stomatal opening (Uehlein et al., 2003). Further studies revealed a dual localization of NtAQP1 in the plasma membrane and the inner envelope membrane (IEM) of the chloroplast (Uehlein et al., 2008). A mutation in the Arabidopsis thaliana AtPIP1;2 gene was found to be associated with reduced gm and a reduction in the rate of photosynthesis (Heckwolf et al., 2011). Genetic engineering to modify NtAQP1 expression levels confirmed these results, revealing a function for NtAQP1 in CO2 conductance. Antisense or RNA interference-mediated downregulation of NtAQP1 resulted in a reduction of IEM CO2 permeability, Cc values, and photosynthetic performance (Uehlein et al., 2003; Flexas et al., 2006; Uehlein et al., 2008). Overexpression of NtAQP1 in tobacco and Arabidopsis, however, increased chloroplast membrane CO2 permeability, the rate of photosynthesis, and plant growth (Aharon et al., 2003; Uehlein et al., 2003; Flexas et al., 2006). A similar positive effect on CO2 permeability, plus an increase in leaf net photosynthesis, was observed in tomato and rice plants overexpressing AQP (Hanba et al., 2004; Sade et al., 2010). It was eventually suggested that the function of NtAQP1 might depend on its localization in the cell, and that it might provide a water channel in the plasma membrane and a CO2 channel in the chloroplast envelope (Uehlein et al., 2008). In the present study, the hypothesis that higher levels of AQPs in the chloroplast would increase CO2 transport and the rate of photosynthesis was tested by overexpressing NtAQP1 from the chloroplast genome of tobacco. Compared with nuclear transformation, plastid transformation provides the advantage of high transgene expression levels (Bock, 2015). In addition, the recombinant protein is confined to the chloroplast, eliminating the effect of AQP1 modification in the plasma membrane. Therefore, the main objective of the present study was to evaluate the role of NtAQP1 overexpression specifically in the chloroplast membranes on CO2 permeability and photosynthetic performance. Materials and methods Production of plants overexpressing NtAQP1 in the chloroplast Total RNA from Nicotiana tabacum L. (cv. Petite Havana SR1) leaves was extracted using the Ultraspec RNA kit (Biotecx Laboratories, Houston, TX, USA), and cDNA was synthesized using the SuperScript III system (Invitrogen, Carlsbad, CA, USA). The NtAQP1 gene (GenBank Accession AJ001416) was amplified by PCR with the primers NTAQP1for: AAGCTTTTGCAAGTATATT TTCCATGGCAGAAAACAAAGAA GAAGATGTTAAGCTCGG and NTAQP1rev: GCGGCCGCTTAA GACGACTTG TGGAATGGAATGGCTCTG. The full-length cDNA was then cloned into the pGEMTeasy vector (Promega, Madison, WI, USA) and sequenced. The tobacco NtAQP1 gene was subsequently cloned into the pAF chloroplast transformation vector (Fernández-San Millán et al., 2008) under the control of the psbA promoter and 5ʹ-untranslated region, to obtain the expression vector pAF-AQP1. The Tic40 transit peptide sequence (240 bp) was amplified by PCR using cDNA from A. thaliana using the primers AtTic40TPfor: CCATGGAGAACCTTACCCTAGTTTC and AtTic40TPrev: GCGGCCGCAAGCTTTGCTTCTCTGTTTC. It was then fused with NtAQP1 at an NcoI restriction site to produce the expression vector pAF-TicAQP1. N. tabacum L. (cv. Petite Havana SR1) was also used in plastid transformations. The PDS-1000/He biolistic system (Bio-Rad, Hercules, CA, USA) was used for the integration of transgenes as previously described (Daniell, 1997). The aadA gene, conferring resistance to spectinomycin, was used as a selectable marker gene. Two rounds of selection and shoot development on RMOP medium containing 500 mg/l spectinomycin were performed. The transplastomic plants produced were named AQP1 and TicAQP1. Southern and northern blotting Southern and northern blotting experiments were performed as previously described (Sanz-Barrio et al., 2013). For Southern blotting, the flanking sequence P1 probe generated by PCR was used. After Southern blot confirmation of the T0 generation, selected plants were transplanted into pots and grown in the greenhouse for seed production. T1 plants were used in further experiments. Northern blotting was performed using the AQP1-specific P2 probe (515 bp) obtained by NcoI digestion of AQP1. Protein extraction, separation, and western blotting Leaf samples (100 mg) from transformed and untransformed 70-day-old plants were ground in liquid nitrogen, homogenized in 300 μl of 2× Laemmli buffer (0.5 M Tris–HCl, pH 6.5, 4% SDS, 20% glycerol, and 10% β-mercaptoethanol) and heated at 95 °C for 5 min. After 5 min of centrifugation at 20000 g, the supernatant was deemed to represent the total protein (TP) content. TP was quantified using the RC-DC protein assay (Bio-Rad) with BSA as a standard. Proteins were separated by SDS-PAGE on 12% polyacrylamide gels and transferred to a polyvinylidene fluoride (PVDF) membrane for immunoblotting. The primary antibodies used were anti-PIP1 (Agrisera, Vännäs, Sweden) and anti-NtAQP1 (kindly provided by R. Kaldenhoff) (dilution 1:3000). Peroxidase-conjugated goat anti-rabbit or anti-chicken immunoglobulin G (Sigma-Aldrich, St Louis, MO, USA) (both at a dilution of 1:3000) were used as secondary antibodies with the anti-PIP1 and anti-NtAQP1 primary antibodies, respectively. Detection was performed using the chemiluminescence ECL western blotting system (GE Healthcare, Fairfield, CT, USA). Relative quantification of NtAQP1 monomers and oligomers was performed by comparing dilution series of TP from wild-type (WT) plants and both types of transplastomic plant (three replicates were analysed). For each line, adequate amounts were loaded on to an SDS-PAGE gel, electrophoretically separated, and then analysed by western blotting. Immunoblots were quantified using GeneTools Analyzer software (SynGene, Cambridge, UK). Plasma membrane isolation Plasma membranes from 50-day-old WT and transplastomic plants grown in a growth chamber [16 h light/8 h dark; 200 µmol m−2 s−1 photosynthetic photon flux density (PPFD) and a day/night temperature regime of 28 °C/25 °C] were obtained as previously described (Santoni, 2007). Chloroplast isolation, fractionation, and immunoblotting For chloroplast isolation, leaves from 50-day-old tobacco plants were cut into 1–3 cm2 pieces and homogenized in a blender. The isolation buffer (330 mM sorbitol, 20 mM MOPS, 13 mM Tris, 3 mM MgCl2, 0.1% BSA) was six times (v/w) the fresh mass weight of the leaf samples. The homogenate was passed through a filter mesh and centrifuged for 5 min at 1000 g and 4 °C. The pellet fraction was resuspended in isolation buffer and chloroplasts were isolated by 80–40% Percoll gradient fractionation after centrifugation for 10 min at 7700 g and 4 °C. Isolated chloroplasts were washed in 3 volumes of washing buffer (330 mM sorbitol, 50 mM HEPES/KOH, pH 7.6, 3 mM MgCl2). For fractionation, the chloroplasts were lysed by freeze-thawing in hypotonic TE buffer [10 mM Tris, 2 mM EDTA, pH 7.5, including a cocktail of protease inhibitors from Roche (Mannheim, Germany)]. Stroma, envelopes, and thylakoids were separated by using discontinuous sucrose gradients (0.93/0.6/0.3 M) after 2 h of centrifugation at 20000 rpm in a swing-out rotor. Stroma was collected from the upper fractions and one volume of extracts was combined with one volume of 2× Laemmli buffer. The thylakoid membranes sedimented out and were resuspended in 10 mM TE buffer, to which one volume of 2× Laemmli buffer was added. The chloroplast envelopes were collected at the interface between 0.9 and 0.6 M sucrose. The envelope proteins were concentrated by methanol/chloroform extraction (Ferro et al., 2002) and resuspended in 10 mM TE buffer with one volume of 2× Laemmli buffer added. All three fractions were heated for 1 h at 37 °C. Protein concentrations were determined by the Bradford method. All samples were resolved by 9% SDS-PAGE and transferred to PVDF membranes for western blotting. Antisera to ADP-glucose pyrophosphorylase (AGPase), LHC chlorophyll a/b binding protein 1 (Lhcb1), and Tic40 (Agrisera) proteins were used at dilutions of 1:1000. Plants used for gas exchange Plants were grown in 2 litre pots (organic soil/perlite, 70/30 v/v) in two places, Pamplona and Mallorca (Spain). In Pamplona, plants were grown in a greenhouse at 24–28 °C and relative humidity ~40%. Plants were watered with water by drip irrigation and twice a week with 50% diluted Hoagland’s solution. In Mallorca, plants were grown in a growth chamber at PPFD ~350 µmol m−2 s−1 at the top of the plants, daily temperature of 24–26 °C, relative humidity ~40%, and watered twice a week with water and once a week with 50% diluted Hoagland’s solution. Gas exchange and chlorophyll fluorescence analyses Gas exchange measurements were performed with a calibrated Li-6400 XT portable gas analyser (Licor, Lincoln, NE, USA) equipped with the 2 cm2 Li-6400-40 Leaf Chamber Fluorometer. Determinations were conducted in apical fully developed leaves. Three independent experiments (two in Pamplona and one in Mallorca) were performed. For the measurements made in Pamplona, plants were transferred to a growth chamber with similar environmental conditions to those of the Mallorca site. For each plant, the same procedure was followed: first, stabilization until a steady state of stomatal conductance was reached (typically ~20–30 minutes) in ambient conditions (CO2 concentration=400 µmol mol−1, 1500 PPFD, and 25 °C). After stabilization, the AN/Ci curve was performed by changing the concentration of CO2 entering the leaf chamber with, the following steps: 400, 300, 250, 200, 150, 100, 50, 400, 400, 500, 600, 700, 800, 1000, 1200 and 1500 µmol mol−1, with typically 2–3 minutes between each step. Each AN/Ci curve was corrected for leaks by following the protocol described by Flexas et al., (2007b). In all three experiments, the results of net CO2 assimilation and stomatal conductance were very consistent (data not shown). In the third experiment, performed in Mallorca, chlorophyll fluorescence was measured together with gas exchange to estimate the mesophyll conductance to CO2. Therefore, all the results shown in the present paper correspond to this latter experiment. After performing the AN/Ci curve, the leaf was kept in the chamber and N2 from a tank (Air Liquide) was piped into the Li-6400 inlet to remove O2 from the entering air in the leaf chamber, to allow measurements to be made in non-photorespiratory conditions (Valentini et al., 1995). We then performed a light curve at ambient CO2 concentration (400 µmol mol−1) with the following PPFD steps: 1500, 2000, 1750, 1500, 1250, 1000, 800, 550, 300, 150, 100, 75, 50, 25 and 0 µmol m−2 s−1. These measurements were used to estimate the product of leaf absorption (α) and the partitioning of absorbed quanta between photosystems I and II (β) (see Valentini et al., 1995 and Pons et al., 2009 for details). We used only the first points of the curve, with PPFD >400 µmol m−2 s−1, to estimate α*β (Martins et al., 2013), avoiding non-linearity of ΦPSII versus ΦCO2 due to changes/higher influence of leaf respiration at low PPFD. Values of (α*β) were 0.36 ± 0.04 (TicAQP1), 0.35 ± 0.03 (AQP1), and 0.38 ± 0.03 (WT), with no significant differences between genotypes (see Supplementary Fig. S1 at JXB online). Night respiration rate (Rd) was estimated by measuring leaf gas exchange in darkness, 1 h after the lights of the growing chamber were turned off (night). Mesophyll conductance (gm) was estimated by the method developed by Harley et al. (1992), as follows: gm=AN/(Ci−{Γ*[ETR+8(AN+Rl)]/[ETR–4(AN+Rl)]}) where AN is the net CO2 assimilation rate, Ci is the CO2 concentration in the substomatal cavity, Γ* is the CO2 compensation point in the absence of Rd (assumed to be 40 µmol mol−1, from Walker et al., 2013), Rl is the respiration rate in light (estimated as Rd/2), and ETR is the electron transport rate, estimated as follows: ETR=α×β×ΦPSII×PPFD where ΦPSII is the yield of photosystem 2. ΦPSII was estimated using the ‘Multiphase Flash’ method described by Loriaux et al. (2013). Discrimination against 13CO2 The 13C isotope discrimination (Δ, ‰) was calculated as: Δ=δ13Catm−δ13Csampleδ13Csample+1 where δ13Catm is the carbon isotope composition in atmospheric CO2 in the greenhouse (–10.8‰) and δ13Csample is the carbon isotope composition of leaf total organic matter (TOM). The 13C/12C ratio (R) in plant material was expressed in δ notation (δ13C) with respect to Vienna Pee Dee Belemnite calcium carbonate (V-PDB), and measured with an analytical precision of 0.1‰: δ13C=(RsampleRstandard)−1 δ13C accuracy was monitored using international secondary standards of known 13C/12C ratios (IAEA-CH7 polyethylene foil, IAEA-CH6 sucrose, and USGS-40 glutamic acid, IAEA, Austria). TOM and gas δ13C determinations were conducted at the Serveis Cientifico-Tecnics of the University of Barcelona. For TOM analyses, 1 mg of dry ground leaf material was analysed using an elemental analyser (EA1108, Series 1, Carbo Erba Instrumentazione, Milan, Italy) coupled to an isotope ratio mass spectrometer (Delta C, Finnigan MAT, Bremen, Germany) operating in continuous flow mode. Air δ13C samples were analysed by gas chromatography (Agilent 6890 Gas Chromatograph, Agilent Technologies, Spain) coupled to an isotope ratio mass spectrometer Deltaplus via a GC-C Combustion III interphase (ThermoFinnigan, Thermo, Barcelona, Spain). Determination of Rubisco, starch, and chlorophyll contents Samples from the same leaves used for gas exchange and chlorophyll fluorescence measurements were analysed for their Rubisco, starch, and chlorophyll contents. Three leaf discs (2.1 cm2) per plant from WT, AQP1, and TicAQP1 plants were frozen in liquid nitrogen and ground in a Mikro-dismembrator (Braun, Melsungen, Germany). The volume of the extraction buffer (phosphate buffer, pH 7.0, 100 mM) was three times the fresh weight (v/w) of the powdered leaf sample obtained from the three leaf discs. Samples were mixed in a vortex and, after 15 min on ice, centrifuged for 5 min at 20000 g at 4 °C. Protein fractions recovered from the supernatants were quantified by the Bradford method. For the separation of these proteins, one volume of the protein fraction was combined with one volume of 2× Laemmli buffer, boiled for 5 min, and then centrifuged at 20000 g for 5 min. Samples (15 μg) of the proteins in these supernatants were separated by SDS-PAGE (10%) and the gels were stained with Coomassie brilliant blue G-250. The Rubisco levels of the transplastomic plants were compared with those of the WT plants using GeneTools Analyzer software (SynGene). Starch was determined using an amyloglucosidase-based test kit (R-Biopharm AG, Darmstadt, Germany). The leaf chlorophyll contents of the transplastomic and WT plants [measured with a SPAD 502 chlorophyll meter (Minolta Optics Inc, Tokyo, Japan)] were recorded in the same leaves used for photosynthetic rate measurements. Leaf area of flowering plants grown in a growth chamber was determined after scanning with ImageJ. Electron microscopy Samples from the same leaves used to measure the photosynthetic rate were fixed in Karnovsky fixative (4% formaldehyde and 5% glutaraldehyde in 0.025 M cacodylate buffer, pH 6.7) by vacuum infiltration and further prepared for examination by transmission electron microscopy at the Microscopy Service of the University of Navarre, Spain. Statistical analysis One-way ANOVA was used to analyse differences in the measured variables between the control and transplastomic plants. Differences among means were analysed using the Tukey test (P<0.05). All calculations were performed using SPSS 10.0 software. Results Generation of tobacco transplastomic plants and determination of homoplasmy Tobacco plants expressing NtAQP1 from the plastid genome were obtained by biolistic bombardment of the leaves with the engineered pAF vector (Fernández-San Millán et al., 2008), which inserted the transgenes between the trnI and trnA regions of the plastid genome (Fig. 1A). Two different transformation vectors were designed, both with the transgene controlled by the promoter and the 5ʹ-untranslated region of the psbA gene. The pAF-AQP1 vector included the full coding sequence of NtAQP1. In the pAF-TicAQP1 vector, the 76 amino acid sequence transit peptide of A. thaliana Tic40 protein was fused to the N-terminus of NtAQP1. Two independent transplastomic lines for each construction, developed after two rounds of selection on spectinomycin, were analysed by Southern blotting (Fig. 1B). As predicted for the correct homologous recombination of the transgenes, the flanking region P1 probe hybridized to a 10.4 or 10.7 kb BamHI DNA fragment in the AQP1 and TicAQP1 transplastomic plants, respectively. A 7.1 kb band was detected only in the WT plants, indicating that all four lines were homoplasmic. Fig. 1. Open in new tabDownload slide Integration of Nicotiana tabacum AQP1 into the tobacco chloroplast genome. (A) Map of the wild-type (WT) and AQP1-transformed plastid (pt) genomes. The transgenes were targeted to the intergenic region between trnI and trnA. The selectable marker gene aadA (encoding aminoglycoside 3ʹ-adenylyltransferase) was driven by the 16S ribosomal RNA operon promoter (Prrn). AQP1 was driven by the psbA promoter and 5ʹ-untranslated region (PpsbA). Arrows within boxes show the direction of transcription. Numbers below each ptDNA indicate the predicted size of hybridizing fragments when total DNA was digested with BamHI. A 0.8 kb fragment of the targeting region for homologous recombination was used as a probe (P1) for Southern blot analysis. TpsbA, 3ʹ-untranslated region of the psbA gene. (B) Southern blot analysis of two independent lines (1 and 2) for each transformation cassette. Expression of aquaporin in the chloroplast Analysis at the transcriptional level in the transplastomic plants revealed transcripts of the expected size in both cases (Fig. 2A). Monocistrons of 1.4 and 1.7 kb derived from the psbA promoter were detected in the AQP1 and TicAQP1 plants, respectively. Dicistrons transcribed from the upstream rrn promoter were also present. A greater abundance of transcripts was observed in the AQP1 plants than in the TicAQP1 plants, probably owing to the TicAQP1 transcripts being less stable. The endogenous AQP1 mRNA was below the detection limit. Fig. 2. Open in new tabDownload slide Analysis of AQP1 expression in wild-type (WT) and AQP1 and TicAQP1 transplastomic plants. (A) Northern blot analysis of leaf samples. The expected transcript sizes of the mono- and dicistrons originating from different promoters are indicated below the map of the transformed plastid genome. The 515 bp AQP1 sequence (P2) was used as a probe. A 10 μg aliquot of total RNA was loaded per well. Ethidium bromide-stained rRNA was used to assess loading. (B) Western blot analysis of total protein from leaf samples (two independent lines for each construction). The lower panel was overexposed to show the 30 kDa AQP1 monomer, which was not detected in the upper panel. A 30 μg aliquot of protein was loaded per well. (C) Western blot analysis of proteins extracted from the plasma membrane. A 3 μg aliquot of protein was loaded per well. The positions and sizes of molecular weight protein standards are indicated. The blots in B and C were detected using anti-NtAQP1 as the primary antibody. The overexpression of AQP1 protein in the transformed chloroplasts was confirmed by immunoblotting (Fig. 2B). A faint 30 kDa band of the expected size for the AQP1 protein monomer was observed in the WT plants. A stronger signal of the same electrophoretic mobility was detected in the AQP1 and TicAQP1 plants. Thus, the TicAQP1 recombinant protein was correctly processed in the stroma of the chloroplast following cleavage of the Tic40 transit peptide. Higher molecular weight signals were mainly present in transplastomic plants, indicating the presence of abundant oligomeric structures despite the denaturing conditions used during sample preparation and electrophoresis. It is also possible that the increased AQP1 protein production or inadequate post-translational modifications caused misfolding or the formation of non-specific aggregates with other proteins that resulted in abnormal migration patterns in the gel. The putative non-specific protein aggregates of high molecular weight could indicate that only a proportion of the recombinant protein equates to functional AQP1 complexes. Relative AQP1 protein levels were estimated by densitometry of different protein extract dilutions in western blots, analysing both monomeric and multimeric signals. The expression level of AQP1 protein in the AQP1 and TicAQP1 plants was approximately 10-fold and 16-fold greater, respectively, than in the WT plants. As expected, analysis of purified plasma membranes by immunoblotting indicated that the AQP1 protein levels in the WT and transplastomic plants were similar (Fig. 2C), confirming that the AQP1 protein synthesized in the chloroplast was not exported out of the chloroplast. NtAQP1 was mainly localized to the chloroplast envelope The distribution of AQP1 in the chloroplast of the AQP1 and TicAQP1 transplastomic plants was examined by chloroplast purification and suborganellar fractionation followed by immunoblotting (Fig. 3). The equal loading of gels for each fraction and the purity of the fractions were assessed using specific antibodies: anti-ADP-glucose pyrophosphorylase (AGPase) for the stroma (ap Rees, 1995), anti-Tic40 for the envelope, and anti-LHC chlorophyll a/b binding protein 1 (Lhcb1) for the thylakoid membrane (Farmaki et al., 2007). AQP1 was detected in the envelope and thylakoid membrane fractions but no signal was seen for the stroma soluble fraction (Fig. 3). As expected, a stronger signal was detected in the AQP1 and TicAQP1 transplastomic plants than in the WT plants. Monomeric and oligomeric forms were detected in the envelope and thylakoid fractions of both types of transplastomic plants. The AQP1 signal was most intense in the envelope fraction; note that 10-fold more thylakoid fraction protein was loaded to enable the detection of AQP1. Monomers and dimers were present mainly in the envelope fraction, with faint signals detected in the thylakoid fraction. Trimers and tetramers were detected in the envelope fraction but not in the thylakoid fraction. A higher proportion of high molecular weight aggregates was observed in the thylakoid fraction of TicAPQ1 plants relative to the envelope fraction. Immunoblotting could not provide an accurate estimate of the relative distribution of NtAQP1 in each membrane fraction owing to the disproportionate method by which each membrane fraction was purified. However, the higher NtAQP1 expression level in TicAQP1 transplastomic plants correlated with the higher AQP1 content in the thylakoid and envelope membranes of these plants relative to AQP1 transplastomic plants (Fig. 3). Fig. 3. Open in new tabDownload slide Localization of AQP1 in the thylakoid and envelope membranes. Envelope, thylakoid, and stroma fractions were isolated from wild-type (WT), AQP1, and TicAQP1 leaves, and separated by SDS-PAGE. Samples of 2, 20, and 30 μg of protein from the envelope, thylakoid, and stroma, respectively, were loaded per well. Representative western blots performed with antibodies to AQP1, the inner-membrane Tic40 protein, the thylakoid membrane-specific LHC chlorophyll a/b binding protein 1 (Lhcb1), and the stroma-specific ADP-glucose pyrophosphorylase (AGPase) are shown. Asterisks indicate the positions of monomer (*), dimer (**), trimer (***), and tetramer (****) AQP1. Photosynthetic performance and protein and starch metabolism were impaired by chloroplast NtAQP1 overexpression The transplastomic plants, particularly TicAQP1 plants, showed retarded growth in comparison to the WT plants during the first 3 weeks following transplantation into pots, but thereafter they caught up and reached a similar height under standard growth conditions (data not shown). Net photosynthesis (AN) analyses showed both the AQP1 and the TicAQP1 plants to have lower CO2 fixation rates than the WT plants (Fig. 4A), with no differences between the transplastomic plants. As expected, no significant differences were observed between genotypes for leaf stomatal conductance (gs) (Fig. 4D). In contrast, gm was diminished by 50% in AQP1 and TicAQP1 plants relative to WT plants (Fig. 4E). AN/Ci curve determinations highlighted that the Rubisco maximum carboxylation capacity (VCmax) and the maximum electron transport rate contributing to ribulose 1,5-bisphosphate regeneration (Jmax) values measured in both transplastomic plants were lower than those in WT plants (Fig. 4B, C). Gas exchange analyses also showed that the transplastomic plants had a higher Ci value (Fig. 5A), while no significant differences in Cc were detected (Fig. 5B). The Cc/Ci ratio was not altered in the transplastomic plants relative to the WT plants (Fig. 5C). In contrast, compared with the WT plants, AQP1 and TicAQP1 plants showed higher 13CO2 discrimination (Δ) values (Fig. 4F). Transplastomic plants had lower soluble protein levels than their WT counterparts (Table 1). Moreover, the Rubisco levels were strongly reduced in the AQP1 and TicAQP1 plants (reductions of 23% and 41%, respectively). A 4-fold reduction in the leaf starch content, and reduced chlorophyll levels, were detected in the transplastomic plants relative to the WT plants (Table 1). In addition, the leaf mass area was reduced in the transplastomic plants (12.5% in AQP1 and 15.9% in TicAQP1 plants). Fig. 4. Open in new tabDownload slide (A) Net photosynthesis, (AN), (B) maximum carboxylation velocity of Rubisco (VCmax), (C) maximum electron transport rate contributing to ribulose 1,5-bisphosphate regeneration (Jmax), (D) stomatal conductance (gs), (E) mesophyll conductance (gm), and (F) 13C isotope discrimination (Δ) of wild-type (WT) and AQP1 and TicAQP1 transplastomic plants. Representative data from two independent experiments are presented. Values are means ±SE (n=7). Different letters indicate significantly different values (ANOVA, P<0.05). Fig. 5. Open in new tabDownload slide (A) Intercellular CO2 concentration (Ci), (B) chloroplast CO2 concentration (Cc), (C) and Cc/Ci ratio of wild-type (WT) and AQP1 and TicAQP1 transplastomic plants. Representative data from two independent experiments are presented. Values are means ±SE (n=7). Different letters indicate statistically different values (ANOVA, P<0.05). Table 1. Biochemical variables measured in young leaves from wild-type (WT) and transplastomic plants grown in a growth chamber . WT . AQ1 . TicAQP1 . Starch (μmol Glu g FW−1) 21.7 ± 2.9a 5.5 ± 0.2b 5.7 ± 0.4b Soluble protein (mg g FW−1) 16.1 ± 0.7a 13.3 ± 1.0ab 12.3 ± 0.4b Insoluble protein (mg g FW−1) 13.0 ± 1.7a 10.5 ± 1.3a 10.2 ± 2.4a Rubisco (% relative to WT) 100 ± 3.4a 76.7 ± 3.9b 58.8 ± 5.4c Chlorophyll content (SPAD) 42.9 ± 1.0a 34.2 ± 0.9b 30.0 ± 0.6c . WT . AQ1 . TicAQP1 . Starch (μmol Glu g FW−1) 21.7 ± 2.9a 5.5 ± 0.2b 5.7 ± 0.4b Soluble protein (mg g FW−1) 16.1 ± 0.7a 13.3 ± 1.0ab 12.3 ± 0.4b Insoluble protein (mg g FW−1) 13.0 ± 1.7a 10.5 ± 1.3a 10.2 ± 2.4a Rubisco (% relative to WT) 100 ± 3.4a 76.7 ± 3.9b 58.8 ± 5.4c Chlorophyll content (SPAD) 42.9 ± 1.0a 34.2 ± 0.9b 30.0 ± 0.6c Values are means±SE (n=5–7). Different superscript letters denote significant differences (ANOVA, P<0.05). The chlorophyll content was measured using a Minolta SPAD 502 chlorophyll meter. Open in new tab Table 1. Biochemical variables measured in young leaves from wild-type (WT) and transplastomic plants grown in a growth chamber . WT . AQ1 . TicAQP1 . Starch (μmol Glu g FW−1) 21.7 ± 2.9a 5.5 ± 0.2b 5.7 ± 0.4b Soluble protein (mg g FW−1) 16.1 ± 0.7a 13.3 ± 1.0ab 12.3 ± 0.4b Insoluble protein (mg g FW−1) 13.0 ± 1.7a 10.5 ± 1.3a 10.2 ± 2.4a Rubisco (% relative to WT) 100 ± 3.4a 76.7 ± 3.9b 58.8 ± 5.4c Chlorophyll content (SPAD) 42.9 ± 1.0a 34.2 ± 0.9b 30.0 ± 0.6c . WT . AQ1 . TicAQP1 . Starch (μmol Glu g FW−1) 21.7 ± 2.9a 5.5 ± 0.2b 5.7 ± 0.4b Soluble protein (mg g FW−1) 16.1 ± 0.7a 13.3 ± 1.0ab 12.3 ± 0.4b Insoluble protein (mg g FW−1) 13.0 ± 1.7a 10.5 ± 1.3a 10.2 ± 2.4a Rubisco (% relative to WT) 100 ± 3.4a 76.7 ± 3.9b 58.8 ± 5.4c Chlorophyll content (SPAD) 42.9 ± 1.0a 34.2 ± 0.9b 30.0 ± 0.6c Values are means±SE (n=5–7). Different superscript letters denote significant differences (ANOVA, P<0.05). The chlorophyll content was measured using a Minolta SPAD 502 chlorophyll meter. Open in new tab The ultrastructure of the mesophyll cell chloroplasts was analysed by transmission electron microscopy. Major differences were observed between the WT and transplastomic plants (Fig. 6). The WT plants showed the normal architecture of the thylakoid network, arranged in grana and lamellae (Fig. 6A), while the AQP1 and TicAQP1 plants showed abnormal granal stacking with a reduced number of appressed thylakoids (Fig. 6B, C). In comparison to the normal granal structure of the WT plants (Fig. 6D), the transplastomic plants (especially the TicAQP1 plants) also showed defective grana and swelling of the thylakoid lumen (Fig. 6E, F). Large protein aggregates detected by western blot (Figs 2B and 3), particularly in TicAQP1 plants, could have affected the thylakoid membrane integrity, resulting in abnormal thylakoid architecture. No differences were observed in the envelope membranes (Fig. 6G–I). Fig. 6. Open in new tabDownload slide Changes in chloroplast ultrastructure due to AQP1 overexpression from the plastid genome. Transmission electron microscopic images of chloroplasts from (A) wild-type (WT), (B) AQP1, and (C) TicAQP1 plants. (D–F) Detail of the thylakoids from (D) WT, (E) AQP1, and (F) TicAQP1 plants. (G–I) Detail of the chloroplast envelope (delineated by two arrowheads) in (G) WT, (H) AQP1, and (I) TicAQP1 plants. C, cytosol; s, starch granule; st, stroma; v, vacuole. Scale bar=1 μm. The integrity of the thylakoid membranes was assessed by treating isolated chloroplasts with SDS, a product commonly used in cell permeation assays in bacteria (Griffith and Wolf, 2002). For the chloroplasts of the AQP1 and especially the TicAQP1 plants, total chlorophyll solubilization was obtained with lower SDS concentrations than those required to achieve the same effect with WT plant chloroplasts (Supplementary Fig. S2A, C). The same pattern was observed for the solubilization of AQP1 and the thylakoid membrane Lhcb1 proteins, but not for the stromal AGPase or thylakoid-lumen-associated TL29 proteins (Supplementary Fig. S2B). Discussion Targeting of recombinant AQP1 to chloroplast membranes The present study shows that NtAQP1 can be overexpressed from the plastid genome and that it localizes to the chloroplast membranes. Native NtAQP1 localizes to both the plasma membrane and the chloroplast IEM. No transit peptide for its chloroplast targeting has been identified, and the sorting mechanisms responsible for this dual localization are unknown (Luu and Maurel, 2013). The expression of NtAQP1 from the plastid genome results in the AQP1 polypeptide being synthesized in the stroma and not in the cytosol as in WT plants, conditioning its subsequent import pathway to the IEM. Hence, in addition to the AQP1 coding sequence, and given the uncertainty that the recombinant AQP1 from the chloroplast stroma would reach its target location correctly, a second construct with the transit peptide of the IEM Tic40 protein fused to AQP1 was prepared for plastid transformation. Tic40 is an integral IEM protein involved in protein translocation across this membrane (Chou et al., 2003) that follows the post-import pathway for IEM targeting (Li and Schnell, 2006). Both constructs resulted in the incorporation of AQP1 in the chloroplast envelope and the thylakoid membranes. This result shows that the topology-determining sequence information within NtAQP1 is sufficient for its integration into the envelope membranes from the stroma (Fig. 3). Chloroplast IEM proteins that follow the post-import pathway are first imported from the cytoplasm into the chloroplast stroma in the form of a soluble, processed, intermediate product, and subsequently reinserted from the stroma into the IEM. NtAQP1 expressed from the plastid genome could putatively use the second part of this pathway from the stroma to the IEM. It has been shown that conserved proline residues in the N-terminal region of IEM proteins such as Tic40 and Tic110 are required for stromal reinsertion (Chiu and Li, 2008). The six proline residues present at positions 21, 35, 36, 37, 39, and 43 of NtAQP1 suggest that it uses a common import mechanism from the stroma. Native IEM proteins utilize two different pathways for their targeting. It is unknown whether the native AQP1, with six membrane-spanning alpha helices, uses the post-import pathway or the stop-transfer pathway. Very little is known about the insertion of polytopic IEM proteins. For instance, the Cor413im1 membrane protein, with five or six transmembrane domains, is incorporated in the IEM via the stop-transfer pathway (Okawa et al., 2014), while Tic110, which has six transmembrane domains, utilizes the post-import pathway (Lübeck et al., 1997; Li and Schnell, 2006). If native and plastidial AQP1 use different import pathways, this could result in an improper location within the IEM that negatively affects its functionality. Tic40, a component of the TIC complex, is involved in the reinsertion process for proteins that use the post-import pathway (Chiu and Li, 2008). It could be expected that the recombinant TicAQP1 protein, which includes the Tic40 transit peptide, is inserted close to and potentially interacts with the TIC complex. It remains to be elucidated whether the IEM import pathways of recombinant TicAQP1 and AQP1 proteins are the same, but the physiological performance of both transplastomic plants was similar. The putative drawback related to IEM import seems to be equivalent in both transplastomic plants, irrespective of the presence of the Tic40 transit peptide. However, integration of NtAQP1 into the thylakoid membranes was unexpected, and the mechanism of protein sorting remains unknown. Plastid transformation has also allowed the successful integration of other foreign proteins, with or without signal peptides, to the thylakoid membranes (Henig et al., 2007; De Marchis et al., 2011; Ahmad et al., 2012; Shanmugabalaji et al., 2013; Scotti et al., 2015), indicating that different import mechanisms might be used. The expression of the Synechococcus BicA bicarbonate transporter in tobacco plastids unexpectedly resulted in dual targeting of the protein to the thylakoid membranes and, in a smaller proportion, to the chloroplast envelope (Pengelly et al., 2014). A model for contact zones between plasma and thylakoid membranes, allowing protein trafficking in short-lived connection assemblies, has been proposed for the cyanobacterium Synechocystis (Pisareva et al., 2011). In addition, functional thylakoid membranes were developed in association with the chloroplast envelope in Chlamydomonas under certain conditions (Hoober et al., 1991). These and other investigations have suggested a role of the IEM for thylakoid biogenesis in vascular plants (Celedon and Cline, 2013). This mechanism might tentatively explain the dual targeting of recombinant AQP1 to the envelope and thylakoid membranes. Physiological performance of transplastomic plants The present study sought to determine whether CO2 transport to the chloroplast could be boosted by increasing the amount of AQP1 in the chloroplast membranes. Much higher AQP1 protein levels (up to 16-fold higher than in the WT) were obtained in this study than in another study using nuclear transformation, in which double the levels in WT were obtained (Flexas et al., 2006); this difference was probably due to the plastid transformation method. Despite the integration of AQP1 into the chloroplast envelope membranes, the transplastomic plants overexpressing NtAQP1 showed lower photosynthetic rates than the WT plants. Associated with the low values of AN, transplastomic plants showed both a reduction in CO2 diffusion capacity (associated with gm but not gs) and a lower photosynthetic capacity (VCmax and Jmax). Because the different lines differed in their Ci, and gm responds to Ci (Flexas et al., 2007a), the observed differences in gm could be attributable to differences in Ci. However, gm/Ci ratios (obtained from data shown in Figs 4E and 5A) were 0.3 × 10–3 for the two transplastomic lines, compared with ~0.7 × 10–3 for the WT, suggesting that gm is reduced in the transplastomic lines regardless of their Ci. Considering that between 200 and 400 μmol CO2 mol−1 air gm decreases at an approximate rate of 0.1% per μmol CO2 mol−1 air (shown for different species, including tobacco, by Flexas et al., 2008), if WT plants had had the same Ci as the transplastomic plants, their gm would have decreased by ~5%, that is, it would have still been significantly higher. The realized photosynthesis achieved by a plant can be manipulated by two means (Gago et al., 2014; Flexas et al., 2016): either by modifying the photosynthetic capacity of the plant (i.e. changing the rate of photosynthesis for a given substrate availability) or by changing the diffusion capacity of the leaf (which, in turn, will modify the quantity of substrate available for photosynthesis). In this study, increasing AQP1 expression dramatically decreased the photosynthesis of the transplastomic plants. The reduced AN was an unexpected result, especially given the role of AQP1 in CO2 diffusion, as described in tobacco (Uehlein et al., 2003; Flexas et al., 2006) and in other species (Hanba et al., 2004; Sade et al., 2010). In a previous study, Flexas et al. (2006) showed that, compared with the corresponding WT plants, photosynthetic rates were lower in NtAQP1-deficient plants and higher in NtAQP1-overexpressing plants, suggesting that variations in the photosynthetic rate were certainly linked to changes in Cc (Flexas et al., 2006). In the present study, the higher Ci detected in the transplastomic (compared with the WT) plants indicates that stomatal opening was not involved in the lower AN of the AQP1 and TicAQP1 plants. Indeed, the same stomatal conductance in the three genotypes, along with a clearly lower CO2 fixation by the photosynthetic machinery in the transplastomic plants, might be related to the higher Ci in AQP1 and TicAQP1 plants. Following from this observation, it could be tentatively concluded that the lower AN in the transplastomic plants was caused by their lower gm (by ~50%) compared with the WT plants, since gm is now recognized to play a major role in CO2 diffusion into the chloroplast (Flexas et al., 2006; Flexas et al., 2007b; Scafaro et al., 2011; Kaldenhoff, 2012; Evans and von Caemmerer, 2013; Flexas et al., 2013). Nevertheless, the hypothesis on limited diffusion (reduced gm) in AQP1 and TicAQP1 plants cannot explain alone their lower photosynthetic rates. In fact, Cc was not different between WT and transplastomic plants, and nor was the Cc/Ci ratio, despite very different rates of photosynthesis (Figs 4 and 5). Although gs was not affected by the overexpression of AQP1 in transplastomic plants, the overall CO2 supply was reduced as a consequence of the reduction of gm. However, because the CO2 demand was also reduced due to impaired VCmax and Jmax, the overall result was a relatively higher Cc in the transplastomic plants compared with WT plants; the difference was more evident according to the Δ measurements (Fig. 4F) than according to chlorophyll fluorescence-based estimates (Fig. 5B). If Cc (i.e. the substrate for photosynthesis) was the same between the WT, AQP1, and TicAQP1 plants, then differences in AN were more likely related to differences in photosynthetic capacity. Evidence that reduced VCmax and Jmax are the true factors responsible for decreased AN in the two transplastomic lines arises from reverse photosynthesis modelling and from carbon isotope discrimination. Using reverse modelling, it is possible to estimate how large Cc and gm should be for the transplastomic lines to reach WT AN if their VCmax and Jmax is reduced. It turns out from this simulation that Cc should increase from the estimated ~230 μmol mol−1 to ~330 μmol mol−1, which would require gm values of 8.8 and 4.3 μmol CO2 m−2 s−1 for AQP1 and TicAQP1, respectively; these are unrealistically large values, far out of the range of estimates for any plant species (Flexas et al., 2012). On the other hand, stable isotopes, such as δ13C, have been proposed as indicators of stomatal opening and CO2 diffusion (Farquhar et al., 1989; Araus et al., 2003). Changes in Δ have been linked to changes in CO2 availability and/or Rubisco carboxylation activity (Farquhar et al. 1989; Brugnoli and Farquhar, 2000; Ghashghaie et al., 2003). The fact that CO2 supply was reduced in transplastomic plants due to the reduction of gm, but Δ was still larger in these plants than in WT plants, suggests a reduction of carboxylase activity relative to CO2 supply around Rubisco. The localization of AQP1 to the thylakoid membranes might negatively affect their functionality, perhaps via interaction with proteins of the photosynthetic apparatus that affects the protein dynamics within the thylakoid, with a consequent impact on photosynthetic capacity. Indeed, the reduced membrane resistance to SDS (Supplementary Fig. S2) suggests that the thylakoid membrane was damaged in some way, possibly related to the presence of recombinant AQP1. It has been reported that targeting foreign membrane proteins to thylakoid membranes by chloroplast transformation causes mutant phenotypes with reduced growth and photosynthetic capacity, altered thylakoid ultrastructure, and impairment of the integrity of the photosystems (Henig et al., 2007; Gnanasekaran et al., 2016). A similar response could be produced by the recombinant AQP1 located in the thylakoid membranes, particularly considering the presence of large protein aggregates (Fig. 3). However, despite the higher content of recombinant AQP1 in the thylakoid membranes of TicAQP1 plants, the photosynthetic parameters were similar in both AQP1 and TicAQP1 plants (Fig. 4). Perhaps a threshold of chloroplast damage was reached in the AQP1 plants and the greater content of the recombinant protein in TicAQP1 plants was inconsequential. Another possibility is that other processes, in addition to the functionality of the thylakoids, might be involved in the photosynthetic impairment of these plants. Other factors affecting recombinant AQP1 functionality The gas exchange data highlight the fact that, in contrast to what was expected, NtAQP1 overexpression from the chloroplast genome constrains CO2 diffusion from the substomatal cavity to the chloroplast. An explanation for this harmful effect is not easy to provide, but other factors, in addition to the above-mentioned factors related to the import pathway and thylakoid targeting, could be involved. The regulation of the function of AQP proteins depends on several processes, including post-translational modifications and protein interactions, that affect both their activity and their subcellular localization (Hove and Bhave, 2011; Chaumont and Tyerman, 2014; Verdoucq et al., 2014). It is possible that differences between the stromal and cytosolic environments prevent the required AQP post-translational modifications and/or interactions in the stroma. Another important aspect to be considered is that AQPs assemble as homo- and/or heterotetramers in the membranes. The AQP monomer is the functional unit for water transport, but the tetramer and its composition may be important for CO2-related transport. Further, in tobacco, CO2 diffusion is greater when tetramers consist only of NtAQP1 from the PIP1 family (Otto et al., 2010), and different proportions of PIP1 and PIP2 subunits in the tetramer may promote either water or CO2 transfer, or both (Flexas et al., 2012). If improper AQP1 monomers (due to incorrect post-translational modifications) synthesized in the chloroplast stroma interact with normal WT AQP1 synthesized in the cytosol, non-functional or partially functional homotetramers might find their way to the IEM. It must be noted that unusual oligomeric and possible non-specific protein aggregates were detected (Figs 2B and 3). Given that the transplastomic plants in this study had 10–16 times the AQP1 content of the WT plants, defective homotetramers might have prevailed over any homotetramers composed exclusively of WT AQP1 monomers, thus compromising AQP1-mediated CO2 transport in the IEM. Finally, despite the engineered modification to AQP1 expression levels being able to alter gm (suggesting a role for AQPs in CO2 transport), the molecular bases controlling gm remain unclear and are affected by components such as the cell walls, plasma and chloroplastic membranes, and carbonic anhydrases. Thus, other regulatory mechanisms that control gm might be unmasked in plants with altered levels of AQPs (Verdoucq et al., 2014), such as those analysed in the present study. In conclusion, in an attempt to increase CO2 diffusion across the chloroplast membranes we engineered tobacco plants with substantially increased levels of the CO2-permeable NtAQP1 protein within the chloroplast membranes. However, there was no improvement in intracellular CO2 transfer capacity in the transplastomic plants. In fact, we observed impairment in photosynthetic capacity, which could be partially attributed to changes in the thylakoid ultrastructure, and reductions in VCmax and Jmax, which were the main limiting factors. This study serves to highlight obstacles that need to be overcome in work towards engineering improved CO2 transfer capacity for enhanced photosynthesis. Supplementary data Supplementary data are available at JXB online. Fig. S1. Relationship between ΦPSII versus ΦCO2 used to correct ETR. Fig. S2. Membrane disruption of chloroplasts from wild-type (WT), AQP1, and TicAQP1 plants with increasing concentrations of SDS. Acknowledgements The authors thank Ralf Kaldenhoff (Darmstadt University of Technology, Germany) for providing the NtAQP1 antibody and Iván Jáuregui for technical assistance. MA is a recipient of a Formación de Profesorado Universitario fellowship from the Ministerio de Educación, Cultura y Deporte, Spain. MN was supported by a predoctoral fellowship BES-2015-072578 from the Ministerio de Economía y Competitividad (MINECO), Spain, co-financed by the European Science Foundation. References Aharon R , Shahak Y, Wininger S, Bendov R, Kapulnik Y, Galili G. 2003 . Overexpression of a plasma membrane aquaporin in transgenic tobacco improves plant vigor under favorable growth conditions but not under drought or salt stress . The Plant Cell 15 , 439 – 447 . Google Scholar Crossref Search ADS PubMed WorldCat Ahmad N , Michoux F, McCarthy J, Nixon PJ. 2012 . Expression of the affinity tags, glutathione-S-transferase and maltose-binding protein, in tobacco chloroplasts . Planta 235 , 863 – 871 . Google Scholar Crossref Search ADS PubMed WorldCat ap Rees T . 1995 . Where do plants make ADP-Glc ? In: Pontis HG, Salerno G, Echeverría E, eds. Sucrose metabolism, biochemistry, physiology and molecular biology . Rockville : American Society of Plant Physiologists , 143 – 155 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Araus JL , Villegas D, Aparicio N, del Moral L, Hani El S, Rharrabti Y, Ferrio JP, Royo C. 2003 . Environmental factors determining carbon isotope discrimination and yield in durum wheat under Mediterranean conditions . Crop Science 43 , 170 – 180 . Google Scholar Crossref Search ADS WorldCat Bernacchi CJ , Portis AR, Nakano H, von Caemmerer S, Long SP. 2002 . Temperature response of mesophyll conductance. Implications for the determination of Rubisco enzyme kinetics and for limitations to photosynthesis in vivo . Plant Physiology 130 , 1992 – 1998 . Google Scholar Crossref Search ADS PubMed WorldCat Bock R . 2015 . Engineering plastid genomes: methods, tools, and applications in basic research and biotechnology . Annual Review of Plant Biology 66 , 211 – 241 . Google Scholar Crossref Search ADS PubMed WorldCat Brugnoli E , Farquhar GD. 2000 . Photosynthetic fractionation of carbon isotopes . In: Leegood RC, Sharkey TD, von Caemmerer S, eds. Photosynthesis: physiology and metabolism . Dordrecht : Kluwer Academic Publishers , 399 – 434 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Celedon JM , Cline K. 2013 . Intra-plastid protein trafficking: how plant cells adapted prokaryotic mechanisms to the eukaryotic condition . Biochimica et Biophysica Acta 1833 , 341 – 351 . Google Scholar Crossref Search ADS PubMed WorldCat Chaumont F , Tyerman SD. 2014 . Aquaporins: highly regulated channels controlling plant water relations . Plant Physiology 164 , 1600 – 1618 . Google Scholar Crossref Search ADS PubMed WorldCat Chiu CC , Li HM. 2008 . Tic40 is important for reinsertion of proteins from the chloroplast stroma into the inner membrane . The Plant Journal 56 , 793 – 801 . Google Scholar Crossref Search ADS PubMed WorldCat Chou ML , Fitzpatrick LM, Tu SL, Budziszewski G, Potter-Lewis S, Akita M, Levin JZ, Keegstra K, Li HM. 2003 . Tic40, a membrane-anchored co-chaperone homolog in the chloroplast protein translocon . The EMBO Journal 22 , 2970 – 2980 . Google Scholar Crossref Search ADS PubMed WorldCat Daniell H . 1997 . Transformation and foreign gene expression in plants by microprojectile bombardment . Methods in Molecular Biology 62 , 463 – 489 . Google Scholar PubMed OpenURL Placeholder Text WorldCat De Marchis F , Pompa A, Mannucci R, Morosinotto T, Bellucci M. 2011 . A plant secretory signal peptide targets plastome-encoded recombinant proteins to the thylakoid membrane . Plant Molecular Biology 76 , 427 – 441 . Google Scholar Crossref Search ADS PubMed WorldCat Edgerton MD . 2009 . Increasing crop productivity to meet global needs for feed, food, and fuel . Plant Physiology 149 , 7 – 13 . Google Scholar Crossref Search ADS PubMed WorldCat Evans JR , von Caemmerer S. 2013 . Temperature response of carbon isotope discrimination and mesophyll conductance in tobacco . Plant, Cell & Environment 36 , 745 – 756 . Google Scholar Crossref Search ADS PubMed WorldCat Farmaki T , Sanmartín M, Jiménez P, Paneque M, Sanz C, Vancanneyt G, León J, Sánchez-Serrano JJ. 2007 . Differential distribution of the lipoxygenase pathway enzymes within potato chloroplasts . Journal of Experimental Botany 58 , 555 – 568 . Google Scholar Crossref Search ADS PubMed WorldCat Farquhar GD , Ehleringer JR, Hubick KT. 1989 . Carbon isotope discrimination and photosynthesis . Annual Review of Plant Physiology and Plant Molecular Biology 40 , 503 – 537 . Google Scholar Crossref Search ADS WorldCat Fernández-San Millán A , Ortigosa SM, Hervás-Stubbs S, Corral-Martínez P, Seguí-Simarro JM, Gaétan J, Coursaget P, Veramendi J. 2008 . Human papillomavirus L1 protein expressed in tobacco chloroplasts self-assembles into virus-like particles that are highly immunogenic . Plant Biotechnology Journal 6 , 427 – 441 . Google Scholar Crossref Search ADS PubMed WorldCat Ferro M , Salvi D, Riviere-Rolland H, Vermat T, Seigneurin-Berny D, Grunwald D, Garin J, Joyard J, Rolland N. 2002 . Integral membrane proteins of the chloroplast envelope: identification and subcellular localization of new transporters . Proceedings of the National Academy of Sciences, USA 99 , 11487 – 11492 . Google Scholar Crossref Search ADS WorldCat Flexas J , Barbour MM, Brendel Oet al. 2012 . Mesophyll diffusion conductance to CO2: an unappreciated central player in photosynthesis . Plant Science 193-194 , 70 – 84 . Google Scholar Crossref Search ADS PubMed WorldCat Flexas J , Díaz-Espejo A, Berry JA, Cifre J, Galmés J, Kaldenhoff R, Medrano H, Ribas-Carbó M. 2007b . Analysis of leakage in IRGA’s leaf chambers of open gas exchange systems: quantification and its effects in photosynthesis parameterization . Journal of Experimental Botany 58 , 1533 – 1543 . Google Scholar Crossref Search ADS WorldCat Flexas J , Díaz-Espejo A, Conesa MAet al. 2016 . Mesophyll conductance to CO2 and Rubisco as targets for improving intrinsic water use efficiency in C3 plants . Plant, Cell & Environment 39 , 965 – 982 . Google Scholar Crossref Search ADS PubMed WorldCat Flexas J , Diaz-Espejo A, Galmés J, Kaldenhoff R, Medrano H, Ribas-Carbo M. 2007a . Rapid variations of mesophyll conductance in response to changes in CO2 concentration around leaves . Plant, Cell & Environment 30 , 1284 – 1298 . Google Scholar Crossref Search ADS WorldCat Flexas J , Niinemets U, Gallé Aet al. 2013 . Diffusional conductances to CO2 as a target for increasing photosynthesis and photosynthetic water-use efficiency . Photosynthesis Research 117 , 45 – 59 . Google Scholar Crossref Search ADS PubMed WorldCat Flexas J , Ribas-Carbó M, Diaz-Espejo A, Galmés J, Medrano H. 2008 . Mesophyll conductance to CO2: current knowledge and future prospects . Plant, Cell & Environment 31 , 602 – 621 . Google Scholar Crossref Search ADS PubMed WorldCat Flexas J , Ribas-Carbó M, Hanson DT, Bota J, Otto B, Cifre J, McDowell N, Medrano H, Kaldenhoff R. 2006 . Tobacco aquaporin NtAQP1 is involved in mesophyll conductance to CO2 in vivo . The Plant Journal 48 , 427 – 439 . Google Scholar Crossref Search ADS PubMed WorldCat Gago J , Douthe C, Florez-Sarasa I, Escalona JM, Galmes J, Fernie AR, Flexas J, Medrano H. 2014 . Opportunities for improving leaf water use efficiency under climate change conditions . Plant Science 226 , 108 – 119 . Google Scholar Crossref Search ADS PubMed WorldCat Galmés J , Medrano H, Flexas J. 2007 . Photosynthetic limitations in response to water stress and recovery in Mediterranean plants with different growth forms . New Phytologist 175 , 81 – 93 . Google Scholar Crossref Search ADS PubMed WorldCat Ghashghaie J , Badeck FW, Lanigan G, Nogués S, Tcherkez G, Deléens E, Cornic G, Griffiths H. 2003 . Carbon isotope fractionation during dark respiration and photorespiration in C3 plants . Phytochemistry Reviews 2 , 145 – 161 . Google Scholar Crossref Search ADS WorldCat Gnanasekaran T , Karcher D, Nielsen AZet al. 2016 . Transfer of the cytochrome P450-dependent dhurrin pathway from Sorghum bicolor into Nicotiana tabacum chloroplasts for light-driven synthesis . Journal of Experimental Botany 67 , 2495 – 2506 . Google Scholar Crossref Search ADS PubMed WorldCat Gomes D , Agasse A, Thiébaud P, Delrot S, Gerós H, Chaumont F. 2009 . Aquaporins are multifunctional water and solute transporters highly divergent in living organisms . Biochimica et Biophysica Acta 1788 , 1213 – 1228 . Google Scholar Crossref Search ADS PubMed WorldCat Griffith KL , Wolf RE Jr. 2002 . Measuring β-galactosidase activity in bacteria: cell growth, permeabilization, and enzyme assays in 96-well arrays . Biochemical and Biophysical Research Communications 290 , 397 – 402 . Google Scholar Crossref Search ADS PubMed WorldCat Groszmann M , Osborn HL, Evans JR. 2017 . Carbon dioxide and water transport through plant aquaporins . Plant, Cell & Environment 40 , 938 – 961 . Google Scholar Crossref Search ADS PubMed WorldCat Hanba YT , Shibasaka M, Hayashi Y, Hayakawa T, Kasamo K, Terashima I, Katsuhara M. 2004 . Overexpression of the barley aquaporin HvPIP2;1 increases internal CO2 conductance and CO2 assimilation in the leaves of transgenic rice plants . Plant & Cell Physiology 45 , 521 – 529 . Google Scholar Crossref Search ADS PubMed WorldCat Harley PC , Loreto F, Di Marco G, Sharkey TD. 1992 . Theoretical considerations when estimating the mesophyll conductance to CO2 flux by analysis of the response of photosynthesis to CO2 . Plant Physiology 98 , 1429 – 1436 . Google Scholar Crossref Search ADS PubMed WorldCat Heckwolf M , Pater D, Hanson DT, Kaldenhoff R. 2011 . The Arabidopsis thaliana aquaporin AtPIP1;2 is a physiologically relevant CO2 transport facilitator . The Plant Journal 67 , 795 – 804 . Google Scholar Crossref Search ADS PubMed WorldCat Henig A , Bonfig K, Roitsch T, Warzecha H. 2007 . Expression of the recombinant bacterial outer surface protein A in tobacco chloroplasts leads to thylakoid localization and loss of photosynthesis . The FEBS Journal 274 , 5749 – 5758 . Google Scholar Crossref Search ADS PubMed WorldCat Hoober JK , Boyd CO, Paavola LG. 1991 . Origin of thylakoid membranes in Chlamydomonas reinhardtii y-1 at 38°C . Plant Physiology 96 , 1321 – 1328 . Google Scholar Crossref Search ADS PubMed WorldCat Hove RM , Bhave M. 2011 . Plant aquaporins with non-aqua functions: deciphering the signature sequences . Plant Molecular Biology 75 , 413 – 430 . Google Scholar Crossref Search ADS PubMed WorldCat Kaldenhoff R . 2012 . Mechanisms underlying CO2 diffusion in leaves . Current Opinion in Plant Biology 15 , 276 – 281 . Google Scholar Crossref Search ADS PubMed WorldCat Kaldenhoff R , Kai L, Uehlein N. 2014 . Aquaporins and membrane diffusion of CO2 in living organisms . Biochimica et Biophysica Acta 1840 , 1592 – 1595 . Google Scholar Crossref Search ADS PubMed WorldCat Li M , Schnell DJ. 2006 . Reconstitution of protein targeting to the inner envelope membrane of chloroplasts . The Journal of Cell Biology 175 , 249 – 259 . Google Scholar Crossref Search ADS PubMed WorldCat Long SP , Ainsworth EA, Leakey AD, Nösberger J, Ort DR. 2006 . Food for thought: lower-than-expected crop yield stimulation with rising CO2 concentrations . Science 312 , 1918 – 1921 . Google Scholar Crossref Search ADS PubMed WorldCat Loreto F , Tsonev T, Centritto M. 2009 . The impact of blue light on leaf mesophyll conductance . Journal of Experimental Botany 60 , 2283 – 2290 . Google Scholar Crossref Search ADS PubMed WorldCat Loriaux SD , Avenson TJ, Welles JM, McDermitt DK, Eckles RD, Riensche B, Genty B. 2013 . Closing in on maximum yield of chlorophyll fluorescence using a single multiphase flash of sub-saturating intensity . Plant, Cell & Environment 36 , 1755 – 1770 . Google Scholar Crossref Search ADS PubMed WorldCat Luu DT , Maurel C. 2013 . Aquaporin trafficking in plant cells: an emerging membrane-protein model . Traffic 14 , 629 – 635 . Google Scholar Crossref Search ADS PubMed WorldCat Lübeck J , Heins L, Soll J. 1997 . A nuclear-coded chloroplastic inner envelope membrane protein uses a soluble sorting intermediate upon import into the organelle . The Journal of Cell Biology 137 , 1279 – 1286 . Google Scholar Crossref Search ADS PubMed WorldCat Martins SC , Galmés J, Molins A, DaMatta FM. 2013 . Improving the estimation of mesophyll conductance to CO2: on the role of electron transport rate correction and respiration . Journal of Experimental Botany 64 , 3285 – 3298 . Google Scholar Crossref Search ADS PubMed WorldCat Maurel C , Reizer J, Schroeder JI, Chrispeels MJ. 1993 . The vacuolar membrane protein gamma-TIP creates water specific channels in Xenopus oocytes . The EMBO Journal 12 , 2241 – 2247 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Maurel C , Verdoucq L, Luu DT, Santoni V. 2008 . Plant aquaporins: membrane channels with multiple integrated functions . Annual Review of Plant Biology 59 , 595 – 624 . Google Scholar Crossref Search ADS PubMed WorldCat Okawa K , Inoue H, Adachi F, Nakayama K, Ito-Inaba Y, Schnell DJ, Uehara S, Inaba T. 2014 . Targeting of a polytopic membrane protein to the inner envelope membrane of chloroplasts in vivo involves multiple transmembrane segments . Journal of Experimental Botany 65 , 5257 – 5265 . Google Scholar Crossref Search ADS PubMed WorldCat Otto B , Uehlein N, Sdorra Set al. 2010 . Aquaporin tetramer composition modifies the function of tobacco aquaporins . The Journal of Biological Chemistry 285 , 31253 – 31260 . Google Scholar Crossref Search ADS PubMed WorldCat Parry MA , Andralojc PJ, Scales JC, Salvucci ME, Carmo-Silva AE, Alonso H, Whitney SM. 2013 . Rubisco activity and regulation as targets for crop improvement . Journal of Experimental Botany 64 , 717 – 730 . Google Scholar Crossref Search ADS PubMed WorldCat Parry MA , Reynolds M, Salvucci ME, Raines C, Andralojc PJ, Zhu XG, Price GD, Condon AG, Furbank RT. 2011 . Raising yield potential of wheat. II. Increasing photosynthetic capacity and efficiency . Journal of Experimental Botany 62 , 453 – 467 . Google Scholar Crossref Search ADS PubMed WorldCat Pengelly JJ , Förster B, von Caemmerer S, Badger MR, Price GD, Whitney SM. 2014 . Transplastomic integration of a cyanobacterial bicarbonate transporter into tobacco chloroplasts . Journal of Experimental Botany 65 , 3071 – 3080 . Google Scholar Crossref Search ADS PubMed WorldCat Pisareva T , Kwon J, Oh J, Kim S, Ge C, Wieslander A, Choi JS, Norling B. 2011 . Model for membrane organization and protein sorting in the cyanobacterium Synechocystis sp. PCC 6803 inferred from proteomics and multivariate sequence analyses . Journal of Proteome Research 10 , 3617 – 3631 . Google Scholar Crossref Search ADS PubMed WorldCat Pons TL , Flexas J, von Caemmerer S, Evans JR, Genty B, Ribas-Carbo M, Brugnoli E. 2009 . Estimating mesophyll conductance to CO2: methodology, potential errors, and recommendations . Journal of Experimental Botany 60 , 2217 – 2234 . Google Scholar Crossref Search ADS PubMed WorldCat Priestley DA , Woolhouse HW. 1980 . The chloroplast envelope of Phaseolus vulgaris L .1. Isolation and compositional characteristics . Journal of Experimental Botany 31 , 437 – 447 . Google Scholar Crossref Search ADS WorldCat Reynolds M , Bonnett D, Chapman SC, Furbank RT, Manès Y, Mather DE, Parry MA. 2011 . Raising yield potential of wheat. I. Overview of a consortium approach and breeding strategies . Journal of Experimental Botany 62 , 439 – 452 . Google Scholar Crossref Search ADS PubMed WorldCat Sade N , Gebretsadik M, Seligmann R, Schwartz A, Wallach R, Moshelion M. 2010 . The role of tobacco Aquaporin1 in improving water use efficiency, hydraulic conductivity, and yield production under salt stress . Plant Physiology 152 , 245 – 254 . Google Scholar Crossref Search ADS PubMed WorldCat Santoni V . 2007 . Plant plasma membrane protein extraction and solubilization for proteomic analysis . Methods in Molecular Biology 355 , 93 – 109 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Sanz-Barrio R , Corral-Martinez P, Ancin M, Segui-Simarro JM, Farran I. 2013 . Overexpression of plastidial thioredoxin f leads to enhanced starch accumulation in tobacco leaves . Plant Biotechnology Journal 11 , 618 – 627 . Google Scholar Crossref Search ADS PubMed WorldCat Scafaro AP , von Caemmerer S, Evans JR, Atwell BJ. 2011 . Temperature response of mesophyll conductance in cultivated and wild Oryza species with contrasting mesophyll cell wall thickness . Plant, Cell & Environment 34 , 1999 – 2008 . Google Scholar Crossref Search ADS PubMed WorldCat Scotti N , Sannino L, Idoine A, Hamman P, De Stradis A, Giorio P, Maréchal-Drouard L, Bock R, Cardi T. 2015 . The HIV-1 Pr55 gag polyprotein binds to plastidial membranes and leads to severe impairment of chloroplast biogenesis and seedling lethality in transplastomic tobacco plants . Transgenic Research 24 , 319 – 331 . Google Scholar Crossref Search ADS PubMed WorldCat Shanmugabalaji V , Besagni C, Piller LE, Douet V, Ruf S, Bock R, Kessler F. 2013 . Dual targeting of a mature plastoglobulin/fibrillin fusion protein to chloroplast plastoglobules and thylakoids in transplastomic tobacco plants . Plant Molecular Biology 81 , 13 – 25 . Google Scholar Crossref Search ADS PubMed WorldCat Uehlein N , Lovisolo C, Siefritz F, Kaldenhoff R. 2003 . The tobacco aquaporin NtAQP1 is a membrane CO2 pore with physiological functions . Nature 425 , 734 – 737 . Google Scholar Crossref Search ADS PubMed WorldCat Uehlein N , Otto B, Hanson DT, Fischer M, McDowell N, Kaldenhoff R. 2008 . Function of Nicotiana tabacum aquaporins as chloroplast gas pores challenges the concept of membrane CO2 permeability . The Plant Cell 20 , 648 – 657 . Google Scholar Crossref Search ADS PubMed WorldCat Valentini R , Epron D, De Angelis P, Matteucci G, Dreyer E. 1995 . In situ estimation of net CO2 assimilation, photosynthetic electron flow and photorespiration in Turkey oak (Quercus cerris L.) leaves: diurnal cycles under different levels of water supply . Plant Cell and Environment 18 , 631 – 640 . Google Scholar Crossref Search ADS WorldCat Verdoucq L , Rodrigues O, Martinière A, Luu DT, Maurel C. 2014 . Plant aquaporins on the move: reversible phosphorylation, lateral motion and cycling . Current Opinion in Plant Biology 22 , 101 – 107 . Google Scholar Crossref Search ADS PubMed WorldCat Walker B , Ariza LS, Kaines S, Badger MR, Cousins AB. 2013 . Temperature response of in vivo Rubisco kinetics and mesophyll conductance in Arabidopsis thaliana: comparisons to Nicotiana tabacum . Plant, Cell & Environment 36 , 2108 – 2119 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology.