ERECTA signaling controls Arabidopsis inflorescence architecture through chromatin-mediated activation of PRE1 expression

Transcription

1 Research ERECTA signaling controls Arabidopsis inflorescence architecture through chromatin-mediated activation of PRE expression Hanyang Cai, Lihua Zhao, Lulu Wang, Man Zhang, Zhenxia Su, Yan Cheng, Heming Zhao and Yuan Qin State Key Laboratory of Ecological Pest Control for Fujian and Taiwan Crops & Key Laboratory of Genetics, Breeding and Multiple Utilization of Crops, Ministry of Education & Fujian Provincial Key Laboratory of Haixia Applied Plant Systems Biology, Center for Genomics and Biotechnology, College of Life Science, Fujian Agriculture and Forestry University, Fuzhou, 35 Fujian Province, China Author for correspondence: Yuan Qin Tel: yuanqin@fafu.edu.cn Received: January 7 Accepted: 9 February 7 New Phytologist (7) : doi:./nph.5 Key words: Arabidopsis, ERECTA, HA.Z, inflorescence architecture, PRE, SWR. Summary Flowering plants display a remarkable diversity in inflorescence architecture, and pedicel length is one of the key contributors to this diversity. In Arabidopsis thaliana, the receptor-like kinase ERECTA (ER) mediated signaling pathway plays important roles in regulating inflorescence architecture by promoting cell proliferation. However, the regulating mechanism remains elusive in the pedicel. Genetic interactions between ERECTA signaling and the chromatin remodeling complex SWR in the control of inflorescence architecture were studied. Comparative transcriptome analysis was applied to identify downstream components. Chromatin immunoprecipitation and nucleosome occupancy was further investigated. The results indicated that the chromatin remodeler SWR coordinates with ERECTA signaling in regulating inflorescence architecture by activating the expression of PRE family genes and promoting pedicel elongation. It was found that SWR is required for the incorporation of the HA.Z histone variant into nucleosomes of the whole PRE gene family and the ERECTA controlled expression of PRE gene family through regulating nucleosome dynamics. We propose that utilization of a chromatin remodeling complex to regulate gene expression is a common theme in developmental control across kingdoms. These findings shed light on the mechanisms through which chromatin remodelers orchestrate complex transcriptional regulation of gene expression in coordination with a developmental cue. Introduction Flowering plants have diverse inflorescence architectures that play a key role in determination of plant reproductive success by affecting the fruit set and plant interaction with pollinators and wind (Evers et al., ; Iwata et al., ). Inflorescence architecture can be broadly grouped into two types, the simple type (inflorescences without branching) and the compound type (inflorescences with branching). Inflorescence branching largely depends on the activity of the inflorescence meristems, which either end in a terminal flower or continue to produce branches and flowers, resulting in three inflorescence types observed in nature: racemes, panicles and cymes (Prusinkiewicz et al., 7; Kellogg et al., 3; Han et al., b). Arabidopsis develops the raceme-type inflorescence with the characteristics of simple and indeterminate inflorescence architecture comprising a main inflorescence meristem that generates multiple floral meristems, each of which differentiates a flower at the tip and a pedicel at the base, whereas the inflorescence meristem itself grows indefinitely (Teo et al., ). These authors contributed equally to this work. Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust Several pathways involved in the control of Arabidopsis inflorescence architecture have been identified. An antagonistic interaction between the shoot identity gene TERMINAL FLOWER (TFL) and floral meristem identity genes, such as APETALA (AP) and LEAFY (LFY), regulates inflorescence architecture through affecting branching pattern by modulating meristem identity (Bradley et al., 997; Liljegren et al., 999; Ratcliffe et al., 999; Kaufmann et al., ). A set of MADS-box transcription factors involved in flowering time regulation, namely SUPPRESSOR OF OVEREXPRESSION OF CONSTANS (SOC), SHORT VEGETATIVE PHASE (SVP), AGAMOUS- LIKE (AGL) and SEPALLATA (SEP), recently was shown to act redundantly in control of inflorescence architecture by directly suppressing TFL in association with AP (Liu et al., 3). This TFL-mediated regulation in inflorescence architecture appears to be conserved among flowering plants as TFL orthologs in various plant species show similar functions in control of inflorescence branching (Nakagawa et al., ; Carmona et al., 7; Danilevskaya et al., ; Liu et al., 3). The class-i KNOTTED-like homeobox (KNOX) transcription factor family, which has four members including SHOOT MERISTEMLESS (STM), BREVIPEDICELLUS (BP, also New Phytologist (7) :

2 5 Research New Phytologist called KNAT), KNAT and KNAT (Hake et al., ), contributes to inflorescence architecture through affecting the internode and pedicel growth (Hamant & Pautot, ). BP is known to regulate inflorescence architecture in association with the BELL-like (BELL) transcription factor PENNYWISE (PNY) and the SWI/SNF chromatin remodeling ATPase BRAHMA (BRM) through restricting the expression of KNAT and KNAT (Venglat et al., ; Smith & Hake, 3; Bhatt et al., ; Ragni et al., ; Zhao et al., 5). Plant hormone auxin and cytokinin also play indispensable roles in inflorescence architecture through mediating floral meristem initiation and specification (Vernoux et al., ; Cheng et al., ; Yamaguchi et al., 3) or promoting activity of inflorescence meristem and branch meristem (Miyawaki et al., ; Han et al., a,c; Wang et al., ). ERECTA (ER), a leucine-rich repeat receptor-like kinase, was shown to regulate inflorescence architecture by promoting localized cell proliferation. Mutation of the ERECTA gene results in shortened internodes and pedicels leading to corymb-like clustered inflorescences due to reduced cell proliferation (Shpak et al., 3, ; Woodward et al., 5). ER has two closely related homologs, ER-LIKE (ERL) and ERL, which are partially redundant with ER in regulating several developmental processes including inflorescence architecture, stomatal formation and patterning, and ovule development (Shpak et al., ; van Zanten et al., 9; Lee et al., ; Pillitteri & Torii, ). Two secreted cysteine-rich peptides, EPIDERMAL PATTERNING FACTOR-LIKE (EPFL) and EPFL, were identified as ligands of ER receptor and act redundantly as upstream components of ER-mediated inflorescence growth (Abrash et al., ; Uchida et al., ). As in the stomatal development system, ER transmits signals for inflorescence architecture using the same mitogen-activated protein kinase (MAPK or MPK) cascade composed of YODA (YDA), MPKK/5 and MPK/3 (Meng et al., ; Shpak, 3). However, the downstream components of the ER MAPK signaling pathway in inflorescence architecture control are as yet unidentified. The molecular mechanism underlying ER MAPK signaling mediated regulation of downstream gene expression during inflorescence development remains elusive. The ATP-dependent chromatin remodeling complex SWR plays key roles in the regulation of gene expression in multiple biological processes by modifying the chromatin structure via exchanging histone HA-HB dimers with HA.Z-HB dimers (Krogan et al., 3; Mizuguchi et al., ; Draker & Cheung, 9; March-Diaz & Reyes, 9). The HA.Z variant is highly conserved and its deposition into nucleosomes can regulate gene transcription either positively or negatively by affecting the accessibility of gene to transcription factors (Marques et al., ). The yeast SWR complex contains subunits, and homologs for most of the yeast SWR subunits have been identified in Arabidopsis thaliana including PHOTOPERIOD INDEPENDENT EARLY FLOWERING (PIE), ACTIN-RELATED PROTEIN (ARP) and SERRATED LEAVESANDEARLY FLOWERING (SEF) (March-Diaz & Reyes, 9). Mutants for the components of Arabidopsis SWR New Phytologist (7) : complex, such as pie, arp, sef and hta9 hta double mutant (HTA, HTA9 and HTA are the three HA.Z variants encoding genes in Arabidopsis) display pleiotropic vegetative and reproductive phenotypes including dwarf and bushy plants with shortened inflorescence internodes (Choi et al., 5, 7; Deal et al., 5; March-Diaz et al., 7; Lazaro et al., ). However, whether SWR complex is involved in the ER-signalingregulated inflorescence development is yet unknown. In the present study we showed that ER MAPK signaling regulates inflorescence architecture through activating the expression of a basic helix-loop-helix (bhlh) transcription factor, PACLOBUTRAZOL RESISTANCE (PRE), which is required for pedicel cell proliferation and pedicel elongation. We further demonstrated that SWR maintains tight nucleosome structure and plays a critical role in the control of the ER MAPK signaling promoted transcription of PRE. Our studies revealed the involvement of chromatin remodeling regulation in an important signaling pathway that regulates multiple developmental processes including inflorescence architecture, stomatal formation and patterning, and ovule development. Utilization of chromatin remodeling complex to regulate gene expression is likely a common strategy in developmental control across kingdoms. Materials and Methods Materials and growth conditions Wild-type () Arabidopsis thaliana Columbia- (Col-) ecotype, arp (Garlic_599_G3), sef (CS79), er-5, er-3 (Torii et al., 99), mpk (Salk_7397), mpk3 (SALK_559), pre-amir (Oh et al., ) and p35s:pre-myc (Zhang et al., 9a) were grown under h : h, light : dark conditions at C. Vector constructs A 73-bp segment of ERECTA (ER) genomic DNA was amplified from genomic DNA using the primers listed in the Supporting Information Table S. The PCR fragments were verified by DNA sequencing and then cloned into the pentr/d-topo vector (Invitrogen). pentr/d-topo clones were then recombined into the destination vector pgwb using LR Clonase II (Invitrogen). ppre:gus was generated by amplifying bp of PRE promoter sequence ( to ) from genomic DNA using the primers listed in Table S. The PCR fragments were then cloned into the pentr/d-topo vector (Invitrogen). pentr/d-topo clones were then recombined into the destination vector pgwb533 using LR Clonase II (Invitrogen). GUS staining Inflorescence samples were fixed in prechilled 9% acetone for min and washed with distilled water. After brief vacuum infiltration, the inflorescences were incubated in b-glucuronidase (GUS) staining buffer overnight at 37 C. After being cleared in Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust

3 New Phytologist Research 5 % lactic acid/% glycerol solution, the inflorescences were observed under a Leica (M5 FA) microscope. gel-purified and used in qpcr. Relative nucleosome occupancy was represented as fractions of uncut chromatin DNA. Histological analysis, cell length and number measurement Pedicel tissue samples were fixed overnight in formalin-acetic acid-alcohol solution at C, dehydrated through a graded series of ethanol, and infiltrated with Eponate resin (TED Pella, Inc.) followed by embedding and polymerization. Twomicrometer sections were prepared using a Leica (RM55) microtome. The tissue sections were stained with.% toluidine blue and observed with an Olympus (BX3) microscope. Three regions of each sectioned pedicel were photographed. The number of cells in a middle longitudinal cortex row was determined. This number was used to calculate the total number and average length of cells in the cortex row of each pedicel. The number of cells was counted using 5 sectioned pedicels for each genotype. The cell length was measured directly on the photographic images of plastic sections. Quantitative real-time PCR In order to determine the relative transcript levels of selected genes, real-time quantitative polymerase chain reaction (qpcr) was performed with specific primers (Table S) according to manufacturer instructions using the BIO-RAD real-time PCR system and the SYBR Premix Ex Taq II system (TaKaRa). Chromatin immunoprecipitation For each chromatin immunoprecipitation (ChIP) experiment, g offloralbudsatstage5oryoungerwasused.floralbudswere formaldehyde cross-linked as described (Bowler et al., ). Crosslinked chromatin was fragmented with. units of micrococcal nuclease (Sigma) in ml of MNase digestion buffer ( mm Tris-HCl (ph.), 5 mm NaCl, mm-mercaptoethanol,.% NP, mm CaCl,and9protease inhibitor cocktail, Roche). Digestion was stopped using 5 mm EDTA. ChIP was performed using a HA.Z polyclonal antibody (from R. Deal, Emory University), or polyclonal antibody against H3K7me3 (Millipore, 7-9), polyclonal antibody against H3Kme3 (Millipore, 7-73), or a monoclonal antibody against the RNA polymerase II CTD repeat YSPTSPS (Abcam ab7). of associated DNA fragments was analyzed by qpcr. All oligonucleotide sequences used in the ChIP experiments are given in Table S. Each ChIP experiment was repeated twice and the presented data are from one representative experiment. Nucleosome occupancy Nucleosome positioning was performed as described previously (Petesch & Lis, ). Micrococcal nuclease (MNase) digestion of enriched chromatin was followed by qpcr using primers used for the ChIP experiments. For nucleosome occupancy analysis, chromatin from floral buds at stage 5 or younger were digested with MNase, and mononucleosome-sized fragments were Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust RNA-seq and data analysis RNA was extracted from ovules dissected out from flowers at stage (Smyth et al., 99) using the Qiagen RNeasy kit. One microgram of RNA from each sample for three independent biological replicates was used for Illumina Sequencing, performed as previously described (Zhao et al., ). The raw reads were first filtered by removing the adapter sequences and lowquality sequences. We used the Arabidopsis thaliana genome from TAIR as reference. The clean reads were aligned to the reference genome with STAR v..5.. Alignment results were then processed using the SOURCEFORGE SUBREAD package FEATURECOUNT v..5. for gene quantification. Furthermore, EDGER v.3.. was used to identify the differentially expressed genes (fold change, FDR.5) between each sample. Results ARP is involved in the ERECTA-controlled inflorescence architecture We showed previously that SWR subunit ARP controls female fertility through affecting interhomolog interaction during female meiosis (Qin et al., ). To identify genes of the ARP pathways controlling female fertility, we performed an ethyl methanesulfonate (EMS) mutant screen for enhancers of the reduced fertility mutant arp in the Col-. From this screen, we isolated an enhancer mutation aeh that dramatically enhanced the reduced fertility defects in arp plants. The arp aeh plants displayed extremely low fertility phenotype compared with the, arp and aeh plants (Fig. a,b). To assess whether the enhanced phenotype was caused by a single mutation or not, we crossed this mutant with the plant. Among the 339 F plants, 93 plants exhibited the -like phenotype, plants exhibited the arp-like phenotype and plants exhibited the arp aeh-like extremely reduced silique phenotype. The remaining aeh plants were shorter than. This phenotype segregation was not significantly deviated from the expected 9 : 3 : 3 : ( : arp : aeh : arp aeh) ratio (P >.5, Chi-square test). This genetic study suggested that aeh was a single, recessive mutation. The aeh mutant displayed short internode, short pedicel, compact inflorescence and blunt siliques, reminiscent of the er mutant phenotypes (Torii et al., 99; Shpak et al., ). Sequencing of the ER genomic DNA in aeh revealed a C-to-T mutation resulting in the conversion of 397th amino acid in the LRR domain from Arg to a premature stop codon. The aeh mutation was subsequently designated as er-9. The expression of ER in the er- 9 inflorescence detected by quantitative reverse transcriptioncoupled polymerase chain reaction (qrt-pcr) was significantly reduced (Fig. c), suggesting that er-9 is a knock-down mutant. The er-9 single mutant did not exhibit fertility defects (Fig. b), similar to the reported er mutants, because of New Phytologist (7) :

4 5 Research New Phytologist (a) arp aeh aeh arp (b) Seed-set (%) arp aeh arp aeh (c).5.5 Relative expression level of ER arp er-9 er-5 (d) Percent (%) arp er-9 arp er-9 (e) arp er-9 arp er-9 arp er-9 per:er Internode length between flowers (mm) (f) (g) Pedicel length (mm) (h) arp er-9 arp er-9 (i) Cortex cell number 5 5 (j) Cortex cell length (um) Ep Co Ep Co Ep Co Ep Co Fig. Fertility and inflorescence phenotypes in Arabidopsis arp, er and arp er mutants. (a) Siliques of plants with genotype as indicated. (b) Quantification of seed-set percentage. Data are means SD (n = siliques per genotype). Asterisks above the columns indicate significant differences compared with wild-type () (P <. by Student s t-test). (c) Relative ERECTA (ER) mrna levels in inflorescences of, arp, er-9, and er-5 flowers tested by quantitative real time polymerase chain reaction (qrt-pcr). Data are means SD (n = 3 three biological replicates. Each biological replicate contains two technical repeats). Asterisks above the columns indicate significant differences compared with (P <. by Student s t-test). (d) Distribution of the internode length between two successive siliques. Ten internodes between the st and th siliques were analyzed for pedicels per genotype. (e) Inflorescence stem apices of, arp, er-9, arp er-9 and arp er-9 per:er plants. (f) Fully open mature flowers and attached pedicels of respective genotypes. (g) Lengths of mature pedicels on the main stems of fully open flower - to 5-wk-old plants. Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences compared with or in the marked comparisons (P <., Student s t-test). (h) Longitudinal sections of mature pedicels from fully open flower of, arp, er-9, and arp er-9 plants. Co, cortex; Ep, epidermis. (i) Cell number in the longitudinal cortex file of a mature pedicel from a fully open flower (n = pedicels per genotype). Asterisks above the columns indicate significant differences in the marked comparisons or compared with (P <., Student s t-test). (j) Quantitative analysis of cortex cell length. Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences in the marked comparisons or compared with (P <., Student s t-test). Bars: (a, e, f) mm; (h) lm. the functional redundancy among the three ER family members in fertility control (Pillitteri et al., 7). In addition to the extremely low fertility defects in the arp er- 9 double mutant, we found that arp er-9 double mutant plants exhibited more severe inflorescence architecture defects compared with er-9 and arp single mutants, with shorter internodes between siliques (Fig. d) and more clustered inflorescences (Fig. e). Quantitative analysis of pedicel length indicated that the more clustered inflorescence phenotype in the arp er-9 double mutant was associated with a reduced pedicel length compared with the er-9 single mutant (Fig. f,g). Previous studies showed that the defective pedicel elongation in the er- 5 null allele was due to reduced cell proliferation of the pedicel cortex, accompanied by compensatory cortex cell expansion New Phytologist (7) : Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust

5 New Phytologist Research 53 (Shpak et al., ; Woodward et al., 5; Uchida et al., ). We performed longitudinal sectioning of pedicel tissues in the and mutants and observed similar pedicel cell proliferation defects indicated by reduced cortex cell numbers in er-9 and further reduced cortex cell number in arp er-9 compared with (Fig. h,i), as well as increased cell length in the er-9 pedicel cortex and even longer cortex cells in the arp er-9 double mutant (Fig. j). In order to determine whether ER mutation is the cause of the clustered inflorescence phenotype in the EMS mutated arp plant, we introduced a genomic ER sequence containing the promoter region into arp er-9, resulting in the rescuing of the compact inflorescence and short pedicel phenotypes (Fig. e g). In addition, we crossed arp with other er alleles, er-5 and er-3 (Torii et al., 99), and observed the more clustered inflorescence associated with reduced pedicel length and shorter internodes in the arp er-5 and arp er-3 double mutants compared with the er-5 and er-3 single mutants (Fig.a d). These results together suggest that ARP and ER interact genetically in controlling inflorescence architecture by promoting cell proliferation in the pedicel. SWR complex genetically interacts with the ER mitogenactivated protein kinase (MAPK) signaling pathway in regulating inflorescence architecture In order to investigate if control of inflorescence architecture is a specific function of ARP or is broadly applicable to the SWR complex, we examined other components of the SWR complex. SEF is a subunit of the SWR complex and interacts physically with ARP (March-Diaz et al., 7). We generated sef er-5 and sef er-3 double mutants and found that similar to arp er, the sef er-5 and sef er-3 double mutants had more compact inflorescences, reduced pedicel lengths and shorter internodes than the er-5 and er-3 mutants (Fig. a d). (a) arp sef er-5 (b) arp er-5 arp er-3 sef er-5 sef er-3 Fig. Synergistic effect of SWR subunits, ERECTA (ER) and mitogen-activated protein kinase (MPK) in regulating inflorescence architecture and pedicel elongation in Arabidopsis. (a) Inflorescence stem apices of wild-type (), arp, sef, er-5, arp er- 5, arp er-3, sef er-5 and sef er-3 plants. (b, f) Fully open mature flowers and attached pedicels of respective genotypes. (c, g) Lengths of mature pedicels on the main stems of fully open flowers from - to 5-wkold plants. Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences compared with or in the marked comparisons (P <., Student s t-test). (e) Inflorescence stem apices of, mpk, arp mpk, and er-9 mpk plants. (d, h) Distribution of the internode length between two successive siliques. Ten internodes between the st and th siliques were analyzed for pedicels per genotype. Bars, mm. Pedicel length (mm) (c) (e) (g) Pedicel length (mm) mpk arp mpk arp mpk (d) Percent (%) arp mpk (h) 5 Percent (%) < er-9 mpk 3 arp sef er-5 er-3 arp er-5 arp er-3 sef er-5 sef er-3 Internode length between flowers (mm) (f) < Internode length between flowers (mm) arp mpk arp mpk Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust New Phytologist (7) :

6 5 Research New Phytologist MPK and MPK3 have been shown to share overlapping functions downstream of ER in regulating inflorescence architecture (Meng et al., ). To investigate if ARP also interacts genetically with MPK or MPK3 in regulating inflorescence architecture, we crossed arp with mpk and mpk3, respectively. Similar to arp er and er mpk plants, the arp mpk double mutant showed further reduced pedicels and internode lengths, and more compact inflorescences than the arp and mpk single mutants (Fig. e h). We did not obtain arp mpk3 double mutants from > F plants likely due to embryo lethality. Collectively, these results suggest that the SWR complex interacts genetically with the ER MAPK signaling pathway in regulating inflorescence architecture by promoting cell proliferation in the pedicel. SWR coordinates with ER MAPK to promote PRE gene expression Because the female fertility was dramatically reduced in arp er- 9 plants, in a separate study we performed comparative mrna profiling with RNAs isolated from ovules of, arp, er-5 and arp er-9 when flowers were at stages (Christensen et al., 997) to identify the downstream components of the ARP and ER pathways involved in regulating female fertility. Because the compact inflorescence phenotype was exhibited in the er-5 single mutant, we used the profiling data to identify candidate genes downstream of ER by screening for genes that are downregulated in er-5. We found that except for ER, genes were downregulated in er-5 in comparison with (fold change, P.5; Table S). We were able to obtain T-DNA insertion mutants for 7 of the genes (Table S) from Joseph Ecker s SALK-confirmed T-DNA collection (CS79, CS79, CS793, CS79) (ABRC, However, none of the mutants showed the compact inflorescence phenotype observed in the er mutants. Among the remaining 33 genes, we noticed that the expression of PACLOBUTRAZOL RESISTANCE (PRE), a bhlh transcription factor playing a role in hypocotyl elongation in association with brassinosteroid (BR), gibberellin (GA) and auxin plant hormones (Lee et al., ; Zhang et al., 9a; Bai et al., ), was slightly downregulated in arp er-9 compared with er-5 (Fig. 3a), in correlation with the enhanced inflorescence architecture defects in arp er double mutants than er single mutants. Using qrt-pcr, we showed that PRE was expressed in different tissues including roots, stems, leaves, pedicels, inflorescences and siliques (Fig. S). We confirmed the decreased expression of PRE in er-9 and more reduced expression in arp er-9 in comparison with by qrt-pcr using pedicel tissues of developing flowers at stages with growing pedicels (Smyth et al., 99) (Fig. 3b). In addition, the expression of PRE in arp mpk young floral pedicels was also significantly reduced below that in, arp and mpk (Fig. 3b). Moreover, we constructed the PRE promoter GUS fusion (ppre:gus) and detected the expression of (a) (b). RPKM value 5 5 arp er-5 arp er-9 Relative expression level arp er-9 arp er-9 mpk er-9 mpk arp mpk (c) PRE PRE PRE PRE5 PRE arp er-9 er-9 arp er-9 mpk mpk Fig. 3 The expression of PRE is downregulated in Arabidopsis arp er-9 floral buds. (a) Transcript levels of PRE in wild-type (), arp, er-5 and arp er- 9 ovules detected by RNA-seq. Data are means SD (n = 3 biological replicates). (b) Relative PREs mrna levels in mature pedicels of, arp, er-9, arp er-9, mpk and er-9 mpk flower tested by quantitative real-time polymerase chain reaction (qrt- PCR). Bars represent average values SD from three biological replicates. (c) ppre: GUS expression pattern in floral buds. Bars, mm. New Phytologist (7) : Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust

7 New Phytologist Research 55 ppre:gus in floral buds and the attached pedicels in the background (Fig. 3c). More importantly, the expression of ppre:gus was reduced in er-9 floral buds and further reduced in er-9 mpk and arp er-9 (Fig. 3c). The expression of ppre:gus in arp and mpk single mutants was comparable to that in, consistent with the comparable expression level of native PRE in, arp and mpk. These data indicate that ARP coordinates with the ER MAPK pathway to promote PRE expression in floral pedicels and regulate inflorescence architecture. Suppression of PREs expression resulted in reduced pedicel length and phenocopied the compact inflorescence architecture phenotype in er-9 mpk and arp er-9 PRE belongs to a PRE gene family comprising six members in Arabidopsis, which have redundant roles in various aspects of plant growth and development (Lee et al., ; Oh et al., ). The expression of the four other PRE genes (PRE//5/) was also detected by qrt-pcr in various tissues including pedicels (Fig. S). We assayed the expression of PRE//5/ genes in the floral bud pedicel of, single and double mutants of ARP and ER MPK pathways by qrt-pcr. The results showed that in contrast to the reduced PRE expression level in the er-9 pedicel, the expression of PRE//5/ genes in the er-9 pedicel was comparable to that in (Fig. 3b). However, the expression level of all of the five detected PRE genes was dramatically reduced in the arp er-9, arp mpk and er-9 mpk double mutants that displayed more compact inflorescence architecture than, arp, er-9, mpk (Fig. 3b). The transgenic line (Oh et al., ), in which four PREs (PRE//5/) were knocked down by artificial microrna (pre-amir) (Fig. a), showed clustered inflorescence architecture (Fig. b), short pedicel lengths (Fig. c) and reduced pedicel cell number in the cortex (Fig. d, e), similar to the arp er-9 and er-9 mpk double mutants (Fig. d,e). In contrast with the large and expanded cortex cells in arp er-9 and er-9 mpk, the cell length of the pedicel cortex of pre-amir plants was similar to that of plants (Fig. f), Fig. Suppression of PREs caused compact inflorescence architecture in Arabidopsis. (a) Relative PREs mrna levels in mature pedicels of pre-amir transgenic line tested by quantitative real-time polymerase chain reaction (qrt-pcr). Bars represent average values SD from three biological replicates. (b) Inflorescence stem apices and fully open mature flowers and attached pedicels of wild-type (), and pre-amir plants. (c) Lengths of mature pedicels on the main stems of fully open flowers from - to 5-wkold plants. Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences compared with or in the marked comparisons (P <., Student s t-test). (d) Longitudinal sections of mature pedicels from fully open flower of, arp, er-9, and arp er-9 plants. Ep, epidermis; Co, cortex. (e) Cell number in the longitudinal cortex file of a mature pedicel from fully open flower. Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences in the marked comparisons or compared with (P <., Student s t-test). (f) Quantitative analysis of cortex cell length. Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences in the marked comparisons or compared with (P <., Student s t-test). (g) Distribution of the internode length between two successive siliques. Ten internodes between the st and th siliques were analyzed for pedicels per genotype. Bars: (b).5 cm; (d) lm. Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust (b) (d) Cortex cell length (a) Relative expression level (f) pre-amir pre-amir pre-amir PRE PRE PRE PRE5 PRE arp er-9 Ep Co Ep Co Ep Co Ep Co er-9 mpk (g) Percent (%) (c) Pedicel length (mm) 5 5 Cortex cell number (e) 5 5 pre-amir < Internode length between flowers (mm) pre-amir New Phytologist (7) :

8 5 Research New Phytologist indicating that the compensatory cortex cell expansion in arp er-9 and er-9 mpk is not associated with the reduced expression of PREs. Knockdown of PREs also caused dwarf and short internode phenotypes (Fig. g) similar to the arp er-9 double mutants. These results demonstrated that PREs are likely a key component downstream of the ARP and ER MPK pathways in the control of inflorescence architecture by promoting the growth of internodes and pedicels. Overexpression of PRE partially complemented the compact inflorescence phenotype in er-9 mpk and arp er- 9 In order to further establish that PREs function downstream of ARP and ER MPK in regulating plant inflorescence architecture, we crossed er-9 mpk with p35s:pre-myc (Zhang et al., 9a). In the F generation, individual er-9 mpk plants carrying the p35s:pre-myc transgene were obtained and all showed a less compact inflorescence with elongated pedicels compared with er-9 mpk plants (Fig. 5a c). Moreover, the overexpression of PRE also partially rescued the short pedicel and compact inflorescence phenotype in arp er-9 (Fig. 5d f). Longitudinal sectioning of pedicel tissues and statistical analysis showed that significantly increased cell number in the pedicel cortex in er-9 mpk p35s:pre-myc and arp er-9 p35s: PRE-myc lines compared with er-9 mpk and arp er-9, respectively (Fig. 5g,h). However, in contrast to the expected reduced cell size along with the increased cell number, overexpression of PRE in er-9 mpk and arp er-9 caused further increased cell length in pedicel cortex (Fig. 5i). These results indicate that the elongated pedicel length by overexpression of PRE in er-9 mpk and arp er-9 is due to promoted pedicel cortex cell proliferation, as well as cell expansion. Although knockdown of PREs does not cause compensatory cell expansion along with reduced cell proliferation in the pedicel cortex (Fig. e,f), overexpression of PRE in er-9 mpk and arp er-9 is sufficient to promote both cell expansion and cell proliferation in the cortex (Fig. 5h,i). Moreover, we observed that the pedicel cortex cell length in the p35s:pre-myc plants was significantly longer than those in (Fig. 5i) and the cell number of the p35s: PRE-myc plants pedicel cortex is reduced in compensation (Fig. 5h), given that the pedicel length of p35s:pre-myc plant is comparable to (Fig. 5b,c,e,f). Overexpression of PRE results in increased cell length in the pedicel cortex, indicating the role of PRE in promoting cell expansion in the cortex, reminiscent previous report that PRE is required for hypocotyl elongation by promoting epidermal cell expansion (Zhang et al., 9a; Bai et al., ). It has been shown that the short pedicel phenotype of er-5 and er-5 mpk is due primarily to a reduced number of cortex cells accompanied by compensatory cell expansion (Shpak et al., 3; Meng et al., ). We found that the severity of the short pedicel phenotype in er-9, er-9 mpk and arp er-9 mutants correlated with a decrease in both the cell number and the cell size of the epidermal pavement (Figs S, S3). In contrast to the reduced cell number and similar cell size of the pedicel New Phytologist (7) : cortex of pre-amir plants relative to plants (Fig. e,f), the pedicel epidermis cell number of pre-amir plants was comparable to that of plants (Fig. S3), whereas the pedicel epidermis cell length of pre-amir plants was significantly reduced (Fig. S3). These data suggest that PRE is required for pedicel epidermal cell expansion and cortex cell proliferation. Similar to that in pedicel cortex, overexpression of PRE in er-9 mpk and arp er-9 resulted in both increased cell number and cell length in pedicel epidermis (Fig. S). Similar to that in cortex, the p35s: PRE-myc pedicels also exhibited increased epidermal cell length and reduced cell number in epidermis as a consequence (Fig. S). Taken together, we concluded that PRE functions downstream of the ARP and ER MPK pathways in regulating inflorescence architecture through promoting pedicel cortex cell proliferation, but also plays a role in stimulating pedicel epidermal cell expansion. Cell cycle gene expression is globally changed in PREs knocked-down mutant pedicels In order to gain insight into the molecular mechanisms underlying the regulation of pedicel cell proliferation by PREs, we analyzed the expression of cell cycle genes, which are expressed in developing floral buds including pedicels (Fig. S5), in pedicel tissues of and pre-amir, as well as in the arp, er-9 and mpk single and double mutant flowers at stages by qrt- PCR. The expression of all detected CYCLIN genes and 5 CYCLIN-DEPENDENT KINASEs (CDKs) genes was dramatically reduced in pre-amir and arp er-9, er-9 mpk and arp mpk mutant pedicel compared with and the arp, er-9 and mpk single mutant (Fig. Sa,b), consistent with the reduced cell number in the pedicel cortex of the pre-amir and arp er-9, er-9 mpk and arp mpk mutant plants. The ICK/KRP cyclin-dependent kinase (CDK) inhibitors are other important plant cell cycle factors because of their ability to inhibit CDK activity (Cheng et al., 3). Overexpressing ICK/ KRP caused smaller plant size, compact inflorescence architecture and reduced cell number (Wang et al., ). We found that the expression of seven ICK/KRP genes all increased in pre-amir and arp er-9, er-9 mpk and arp mpk mutant pedicel compared with and the arp, er-9 and mpk single mutant (Fig. Sc). To determine whether the increased cell number in er-9 mpk and arp er-9 pedicel cortex from overexpression of PRE is associated with complemented cell cycle regulation, we analyzed the expression of cell cycle genes in pedicel tissues of er-9 mpk p35s:pre-myc and arp er-9 p35s:pre-myc lines. Consistent with increased cell numberw and elongated pedicel lengthw in er-9 mpk p35s:pre-myc and arp er-9 p35s:pre-myc lines, the expression of all detected CYCLIN genes and five CDKs genes increased and the expression of seven ICK/KRP genes all decreased in pedicel tissues of er-9 mpk p35s:pre-myc and arp er-9 p35s:pre-myc lines compared with er-9 mpk and arp er-9, respectively (Fig. Sa c). These results together suggested that PRE functions downstream of ARP and ER MPK pathways in promoting pedicel cell proliferation through affecting cell cycle gene expression. Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust

9 New Phytologist Research 57 (a) (b) (c) Pedicel length (mm) er-9 mpk er-9 mpk p35s:pre-myc (d) (e) (f) Pedicel length (mm) arp er-9 arp er-9 p35s:pre-myc (g) er-9 mpk er-9 mpk arp er-9 p35s:pre-myc arp er-9 p35s:pre-myc p35s:pre-myc Ep Co Ep Co Ep Co Ep Co Ep Co Ep Co (h) Cortex cell number (i) 35 Cortex cell length (µm) Fig. 5 Overexpression of PRE partially rescued the compact inflorescence defects of er-9 mpk and arp er-9 inflorescences by promoting pedicel cortex cell proliferation and cell elongation in Arabidopsis. (a, d) Inflorescence stem apices of plants of respective genotypes. (b, e) Fully open mature flowers and attached pedicels of respective genotypes. (c, f) Lengths of mature pedicels on the main stems of fully open flower - to 5-wk-old plants. Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences compared with wild-type () or in the marked comparisons (P <., Student s t-test). (g) Longitudinal sections of mature pedicels from fully open flowers of plants of respective genotypes. Ep, epidermis; Co, cortex. (h) Cell number in the longitudinal cortex file of a mature pedicel from fully open flower (n = pedicels per genotype). Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences in the marked comparisons or compared with (P <., Student s t-test). (i) Quantitative analysis of cortex cell length. Bars represent average values SD (n = pedicels per genotype). Asterisks above the columns indicate significant differences in the marked comparisons or compared with (P <., Student s t-test). Bars: (a, d).5 cm; (b, e) cm; (g) lm. Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust New Phytologist (7) :

10 5 Research New Phytologist HA.Z deposition at PRE locus requires ARP The ATP-dependent chromatin remodeling complex SWR catalyzes the replacement of HAZ with its variant HA.Z, which acts as a molecular rheostat for transcriptional control (Maze et al., ), leading to both positive and negative effects on transcription (Mizuguchi et al., ). The expression of PRE was downregulated in the er single mutant and further downregulated in the arp er double mutant (Fig. 3b). We therefore hypothesized that ARP is involved in the regulation of PRE expression by modulating deposition of HA.Z at the PRE chromatin, similar to its role in regulating the expression of other genes (Qin et al., ). To determine whether the SWR complex modulates the deposition of the histone variant HA.Z at PRE and controls its transcription, we performed a ChIP assay using a HA.Z antibody in floral buds of and arp. In floral buds, we detected the enrichment of HA.Z in the region of transcriptional starting site (TSS) and the gene body of PRE (Fig. a,b). This was consistent with previous findings that in most eukaryotic genes, HA.Z deposition is usually limited to gene bodies and the immediate vicinity of TSS (Whittle et al., ; Coleman-Derr & Zilberman, ). In arp floral buds, the enrichment of HA.Z in these regions was greatly depleted (Fig. b). The enrichment of HA.Z in the region of TSS and the gene body of PRE was also significantly reduced in arp er-9 and arp mpk floral buds, but not in er-9, mpk and er-9 mpk floral buds (Fig. b). These data suggested that the deposition of HA.Z on PRE gene loci is dependent on ARP but not ER MPK. To test if lack of HA.Z signal in these assays correctly reflects the absence of HA.Z, we checked for HA.Z deposition in ATG77 (Fig. c), a gene previously shown to contain HA.Z-free nucleosomes. Indeed, in and mutant floral buds, deposition of HA.Z was not detected in any of the regions assayed in ATG77. For additional HA.Z abundance controls, we analyzed HA.Z amounts at two genes previously shown to be marked by HA.Z (Atg7 and Atg39), but the enrichment of HA.Z and the expression of these two genes are independent of SWR (Zilberman et al., ; Nutzmann & Osbourn, 5). Indeed, we found the deposition of HA.Z at TSS of these two genes in, and that the amount of HA.Z at these two genes is not affected in arp and other mutants (Fig. c). HA.Z deposition modulated PRE nucleosome structure HA.Z deposition may facilitate transcriptional regulation by regulating chromatin structure and accessibility of genomic DNA (Coleman-Derr & Zilberman, ). We therefore evaluated the PRE nucleosome accessibility in the presence or absence of ARP using micrococcal nuclease (MNase) accessibility assays. MNase digests the nucleosome-free DNA and the linker regions between nucleosomes (Petesch & Lis, ). Quantification of undigested DNA after MNase treatment enables the identification of nucleosome structure and the evaluation of the nucleosome stability at specific sites. The MNase-qPCR analysis showed that nucleosome occupancy at + nucleosome of PRE, New Phytologist (7) : the first nucleosome downstream of TSS, greatly decreased in arp mutants compared with (Fig. d), indicating that HA.Z deposition on the + nucleosome resulted in tightened nucleosome structure, which may prevent the access of a transcriptional repressor and thus promote the transcription of PRE. The nucleosome occupancy at + nucleosome of PRE was slightly reduced in er-9 and nearly completely abolished in arp er-9, arp mpk and er-9 mpk floral buds (Fig. d), suggesting that ARP and ER MPK are both required to maintain the tighten nucleosome structure at the + nucleosome region, which may be important for active transcription of PRE. In er-9 mpk, HA.Z enrichment at PRE is not affected (Fig. b), but the + nucleosome occupancy is significantly reduced (Fig. d), suggesting that HA.Z independent factors downstream of ER MPK may also be involved in regulating the nucleosome structure. In the control gene (Atg77), nucleosome occupancy is independent of HA.Z enrichment (Fig. e). The H3Kme3 and H3K7me3 levels of PRE were altered in er-9 mpk and arp er-9 Because RNA Pol II is required for active transcription, we therefore evaluated the enrichment of RNA Pol II in the PRE nucleosome through carrying out floral bud ChIP experiments using an RNA Pol II antibody. The results showed that RNA Pol II occupancy at PRE was comparable in arp, mpk and, but significantly reduced in arp er-9 and arp mpk (Fig. f). These data indicated that the transcription of PRE in arp er-9 and arp mpk was repressed. The RNA Pol II occupancy at PRE in er-9 and er-9 mpk was comparable to that in (Fig. f), whereas the transcripts of PRE in er-9 and er-9 mpk were significantly reduced (Fig. 3a), suggesting that other factors besides RNA Pol II are involved in determining the level of PRE transcripts. Histone modification H3Kme3 has been shown to be positively correlated with transcription (Zhang et al., 9b), whereas H3K7me3 is negatively correlated with transcription (Zhang et al., 7). HA.Z could influence H3Kme3 and H3K7me3 by affecting the binding of the MLL and PRC complexes mediating the two histone modifications, respectively (Hu et al., 3). To understand the transcriptional regulation of PRE by ARP and ER MPK, we next determined the levels of the active H3Kme3 marker and the repressive H3K7me3 marker of PRE in, and the single and double mutants of arp, er- 9 and mpk by floral bud ChIP assay. The results showed that the level of H3Kme3 of PRE in arp and mpk was comparable to (Fig. h); however, reduced levels of H3Kme3 were detected in the gene body of PRE in er-9, arp er-9, arp mpk and er-9 mpk (Fig. h), consistent with the reduced expression level of PRE in these mutants. In contrast to the comparable level of H3Kme3 of PRE in arp and, a significantly increased level of H3K7me3 of PRE was detected in the promoter, TSS and gene body of PRE in arp (Fig. j). A significantly elevated level of H3K7me3 of PRE was also detected in er-9, arp er-9 and arp mpk, and a slightly increased level of H3K7me3 of PRE was detected in er-9 Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust

11 New Phytologist Research 59 (a) PRE TSS bp 3 5 (b) Enrichment of HA.Z at PRE arp er-9 mpk arp er-9 arp mpk er-9 mpk (c) Enrichment of HA.Z at control genes.5.5 (d) MNase-qPCR at PRE (e) MNase-qPCR at control genes (f)..3.. Enrichment of Pol II at PRE (g) Enrichment of Pol II at control genes (h).... Enrichment of H3Kme3 at PRE (i) Enrichment of H3Kme3 at control genes.5.5 (j) Enrichment of H3K7me3 at PRE (k) Enrichment of H3K7me3 at control genes Fig. ARP is required for the deposition of HA.Z at PRE in Arabidopsis. (a) Diagram of PRE gene with exons indicated as yellow boxes and the promoter indicated as blue boxes. The transcription start site (TSS) is shown as an arrowhead. PCR primer sets are shown as black bars below the diagram. Primer set numbers correspond to the numbers on the x-axis of the graphs in (b f). (b) Chromatin immunoprecipitation (ChIP) analysis for the enrichment of HA.Z at PRE in wild-type (), arp, er-9, arp er-9, mpk, arp mpk and er-9 mpk flower buds. (c) Nucleosome occupancy at PRE in floral buds as measured by MNase treatment followed by quantitative polymerase chain reaction (qpcr). (d) ChIP analysis for enrichment of RNA Pol II at PRE. (e) ChIP-qPCR analysis of relative H3Kme3 levels of PRE chromatin. (f) ChIP-qPCR analysis of relative H3K7me3 levels of PRE chromatin. (b f) Values are mean SD from two biological replicates. Each biological replicate contains three technical repeats. Asterisks above the columns indicate significant differences compared with (P <., Student s t-test). Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust New Phytologist (7) :

12 59 Research New Phytologist mpk (Fig. j). Similar to H3Kme3, the level of H3K7me3 of PRE in mpk was comparable to (Fig. j). These results together showed that consistent with the reduced expression level of PRE in er-9, arp er-9, arp mpk and er-9 mpk, either the level of the activation maker H3Kme3 of PRE was reduced or the level of the repression maker H3K7me3 of PRE was increased in these mutants. Expression of PRE gene family was under the regulation of HAZ deposition and histone modification Given the high sequence homology and functional redundancy among the PRE genes, and their similarly reduced expression pattern in the arp er-9 double mutant than the er-9 single mutant, we hypothesized that the SWR complex controls the transcription of the whole PRE gene family through modulating the deposition of histone variant HA.Z and nucleosome structure. To test this hypothesis, we performed HA.Z ChIP analysis using and mutant floral buds for PRE/5/. Similar to that in PRE, we detected the enrichment of HA.Z in the region of TSS and + nucleosome of PRE/5/ in floral buds, and reduced HA.Z enrichment in these regions in arp, arp er-9 and arp mpk floral buds (Fig. 7a,b), but not in er-9, mpk and er-9 mpk floral buds (Fig. 7b). These data suggested that in addition to the PRE gene, the deposition of HA.Z into the PRE gene family was also dependent on ARP. We then performed MNase-qPCR analysis and found that nucleosome occupancy at + nucleosome of PRE/5/ was comparable in and arp floral buds (Fig. 7c), in contrast with the reduced nucleosome occupancy at the + nucleosome of PRE in arp. These results suggested that although ARP was required for HA.Z deposition on PREs nucleosome, HA.Z deposition on different PRE genes caused different nucleosome structure change. Although the enrichment of HA.Z on the promoter region of PRE5 and PRE was relatively low (Fig. 7b), nucleosome occupancy at these regions was greatly increased in arp compared with (Fig. 7c), suggesting that other factors may also be implicated in the nucleosome structure change. Interestingly, we found that nucleosome occupancy at PRE/5/ in the er-9 and arp mpk gene bodies was also significantly reduced (Fig. 7c), suggesting that ER MPK may also be involved in the regulation of ARP-controlled nucleosome dynamics. RNA Pol II occupancy at PRE/5/ was comparable in arp, mpk, and, but significantly reduced in arp er-9, arp mpk and er-9 mpk (Fig. 7d), in correlation with the reduced expression level of PRE/5/ in arp er-9, arp mpk and er-9 mpk compared with that in and the single mutants (Fig. 3b). We also found that the enrichment of the active marker H3Kme3 on PRE/5/ was significantly reduced in arp er-9, arp mpk and er-9 mpk (Fig. 7e), and that the enrichment of the repressive marker H3K7me3 on PRE/5/ was significantly increased in arp er-9, arp mpk and er-9 mpk (Fig. 7f). Taken together, our data indicated that the expression of the PRE gene family is under the coordinated regulation of the SWR chromatin remodeling complex and ER MAPK pathways (Fig. a,b). New Phytologist (7) : Discussion SWR complex is implicated in the ERECTA mitogenactivated protein kinase (ER MAPK) pathway controlling inflorescence architecture The roles of the SWR complex in plant development and environment response have been described (March-Diaz et al., 7). In addition to the defects in stress and immunity response, mutants of the Arabidopsis SWR subunits displayed pleiotropic vegetative and reproductive defects including dwarf and bushy plants with shortening inflorescence internodes, serrated leaves, extra petals, reduced fertility and defective meiosis (Choi et al., 5, 7; Deal et al., 5; Martin-Trillo et al., ; March- Diaz et al., 7; Lazaro et al., ; Qin et al., ). Here we showed that er mutation significantly enhanced the plant fertility and growth defects in the arp mutant. arp er double mutant plants exhibited extremely low seed set, shortened internodes and more severely clustered inflorescence architecture in comparison with the arp and er single mutants. It has been determined that the ER-YDA-MKK/MKK5-MPK3/MPK cascade plays a key role in control of inflorescence architecture through promoting localized cell proliferation (Shpak et al., ; Woodward et al., 5; Meng et al., ; Uchida et al., ), but the downstream components were not identified and the regulating mechanism remains unknown. In this study, we showed that the pedicel cell number in a longitudinal cortical file in arp er-9 were significantly decreased in comparison with er-9, with a concomitant decrease in the final pedicel length and a more compact inflorescence. We also found significantly reduced pedicel length and a more compact flower phenotype in arp mpk and sef er double mutant than that in the single mutants. Taken together, we concluded that the SWR complex and ER MAPK signaling pathways interact genetically in regulating localized cell proliferation during plant growth and development, which shapes inflorescence architecture (Fig. a). PREs are required for inflorescence architecture The bhlh transcription factor genes PRE and its homologous genes (PRE/3//5/) are known to redundantly play a role in hypocotyl elongation by promoting epidermal cell expansion in association with brassinosteroid (BR), gibberellin (GA) and auxin plant hormones (Zhang et al., 9a; Bai et al., ). A single knockout mutant of PRE or PRE displayed a normal phenotype (Lee et al., ), whereas the pre-amir plants, in which the expression of four PREs (PRE//5/) are suppressed, had reduced hypocotyls with retarded epidermal cell elongation (Bai et al., ; Ikeda et al., ). Here we showed that repression of PRE//5/ also caused shortened pedicel, shortened internode and clustered inflorescence architecture. Longitudinal sectioning of pedicel tissues and statistical analysis revealed that cell number in the pedicel cortex tissue in pre-amir transgenic lines was significantly reduced compared with, but the cell size of the pedicel cortex was not obviously changed, suggesting that PREs are involved in pedicel elongation mainly through promoting Ó 7 The Authors New Phytologist Ó 7 New Phytologist Trust