Polycomb Group Complexes Mediate Developmental Transitions in Plants

The modulation of transcription by chromatin modifications participates in the coordination of gene networks regulating development. Chromatin marks deposited by Polycomb group (PcG) complexes induce a repressive state of the transcription, which is propagated through cell division. Here, we focus on the implications of this epigenetic regulation in the development of flowering plants like Arabidopsis. We present the mechanism of chromatin modification by PcG and its modulation by other factors. We discuss in detail the mechanisms leading to flowering and illustrate how PcG controls major progressions through the life cycle. The life cycle of eukaryotes is marked by developmental phases characterized by specific spatial and temporal regulation of genome expression. Dynamic regulation of chromatin state is crucial to ensure proper regulation of gene expression. The mechanisms involved in this regulation comprise nuclear localization, DNA methylation, histone variant replacement, and histone posttranslational modifications that recruit or release various chromatin remodeling factors and govern nucleosome occupancy on the DNA (Margueron and Reinberg, 2010). Among the regulators of chromatin state and transcription, Polycomb group (PcG) genes were first discovered in Drosophila melanogaster as repressors of the homeotic Hox genes involved in embryo segmentation (Lewis, 1978; Jürgens, 1985). Four PcG proteins form the core Polycomb Repressive Complex2 (PRC2): Enhancer of Zeste [E(z)], Suppressor of Zeste12 [Su(z)12], Extra sex combs (Esc), and p55. When properly assembled in a protein complex by p55 and Esc, E(z) trimethylates histone H3 Lys-27 (H3K27me3) at target loci (Nekrasov et al., 2005). This mark induces a durable transcriptional silencing (Fig. 1). In the absence of PcG activity, the initial pattern of Hox genes breaks down, causing embryo developmental arrest. Action of Polycomb and Trithorax complexes. PRC2 complexes trimethylate the Lys-27 residue of H3K27me3 and induce a silenced state of the chromatin at the target locus. This mark is recognized by the PRC1 complex. In animals, this complex then monoubiquitinates H2A and triggers a compaction of the chromatin into a heterochromatin state that stably represses expression. The components of PRC1 complex are not fully described in plants so far. PRC2 activity of repression is antagonized by the function of Trithorax (TRX). Trithorax trimethylates the histone H3 Lys-4 residue (H3K4me3) and induces the expression of target loci. Although Trithorax function has been found in plants, the complex has not been identified yet. PcG proteins were likely present in the last common ancestor of eukaryotes and are conserved from unicellular organisms to metazoans, and plants but were lost in yeast (Shaver et al., 2010; Margueron and Reinberg, 2011). Here, we summarize the nature of plant PcG complexes and their mode of action. We describe how PcG complexes interact with other chromatin-modifying complexes and influence gene expression. Furthermore, we detail the specific role of plant PcG complexes in developmental transitions that require coordinated regulation of gene networks. In Arabidopsis (Arabidopsis thaliana), PcG proteins form a family of eight homologs of PRC2 components. The homologs of E(z) are MEDEA, CURLY LEAF (CLF), and SWINGER. Su(z)12 also has three homologs, EMBRYONIC FLOWER2 (EMF2), FERTILIZATION INDEPENDENT SEED2 (FIS2), and VERNALIZATION2 (VRN2). Esc and p55 have the single homologs FERTILIZATION INDEPENDENT ENDOSPERM (FIE) and MULTICOPY SUPPRESSOR OF IRA1 (MSI1), respectively (Fig. 2). Components of complexes associated with PRC2, PRC1, or Trithorax functions in Arabidopsis. While PRC2 complexes are well established, the other complexes are currently hypothetical. FIE and MSI1 are expressed in all cell types, but the expression of other PRC2 homologs is more restricted, and molecular and genetic evidences support the idea of at least three forms of PRC2 in plants: FIS, VRN, and EMF, as per the different Su(z)12 homologs (Spillane et al., 2000; Yoshida et al., 2001; Köhler et al., 2003; Chanvivattana et al., 2004; De Lucia et al., 2008). The FIS complex, for instance, is active in the female gametophyte and developing seed endosperm (Köhler et al., 2003; Guitton et al., 2004), whereas the EMF complex acts in the embryo and during subsequent sporophytic development (Goodrich et al., 1997; Yoshida et al., 2001). The VRN complex activity is triggered upon association with PHD proteins VERNALIZATION INSENSITIVE3 (VIN3), VERNALIZATION LIKE1 (VEL1; also known as VIN3-LIKE2 [VIL2]), and VERNALIZATION5 (VRN5; also known as VIL1) and represses targets of vernalization, the prolonged cold period that allows annual plants to flower in spring (Fig. 2). A comparable association of PRC2 with PHD proteins was observed in animals (Nekrasov et al., 2007; Sarma et al., 2008). In Drosophila, stable transcriptional repression by the PRC2 complex requires the activity of other proteins complexes: PRC1, Pleiohomeotic Repressive Complex, and a newly identified module named PR-DUB (Zheng and Chen, 2011). PRC1, in animals, binds to H3K27me3 marks and ubiquitinates H2A Lys-119. As a consequence, chromatin is compacted in a stable heterochromatin state, and transcription is durably repressed (Sawarkar and Paro, 2010). In plants, homologs have been identified for the PRC1 components BMI1 and RING1. Monoubiquitination of H2A Lys-121 in Arabidopsis by AtBMI1A and AtBMI1B is implicated in repression of embryonic and stem cell regulators (Bratzel et al., 2010). AtBMI1C physically interacts with AtRING1A/B and may be involved in flowering regulation (Li et al., 2011). AtRING1A and AtRING1B also associate with LIKE HETEROCHROMATIN PROTEIN1 (LHP1) to form a complex similar to the animal PRC1 (Xu and Shen, 2008; Fig. 1). There are no clear homologs of other members of PRC1 defined in animals. However, proteins with an activity comparable to PRC1 have been identified in Arabidopsis. LHP1 binds H3K27me3 (Turck et al., 2007) and shares a subset of PRC2 targets in a manner similar to Pc, a PRC1 component in animals (Bracken et al., 2006). With the exception of an impact on flowering time (Mylne et al., 2006; Sung et al., 2006), the lhp1 phenotype is quite distinct from the phenotypes of mutants affecting PRC2 (Gaudin et al., 2001; Takada and Goto, 2003). Two other proteins, VRN1 and EMF1, which bind and act together with LHP1 and AtBMI1A/B (Bratzel et al., 2010), have been proposed to be involved in PRC1-like functions (Mylne et al., 2006; Calonje et al., 2008). Hence, several proteins with an activity similar to PRC1 components are specific to plants. Because PRC1 homologs are not conserved between animal species, it may not be surprising that PRC1 function in plants is mediated by mechanisms distinct from that described in mammals. As opposed to PRC2, Drosophila Trithorax group proteins cause trimethylation of H3 Lys-4 and positively regulate transcription. The Arabidopsis genome encodes five ARABIDOPSIS TRITHORAX proteins (ATX1 to ATX5), with a SET domain and a PHD domain, and seven ARABIDOPSIS TRITHORAX-RELATED proteins (ATXR1 to ATXR7; Tamada et al., 2009). ATXR5 and ATXR6 do not appear to be directly related to ATXR function defined as antagonists to PRC2 and rather cause H3K27 monomethylation associated with constitutive heterochromatin (Jacob et al., 2009; Roudier et al., 2011). ATX1 and ATXR7 are required for the expression of homeotic genes involved in flower organogenesis and the floral repressor FLOWERING LOCUS C (FLC; Pien et al., 2008; Tamada et al., 2009). Besides its H3K4 methylation activity, ATX1 influences transcription by recruiting the TATA binding protein and RNA Polymerase II at promoters (Ding et al., 2011). Although ULTRAPETALA1 (ULT1) has no homology with animal Trithorax group components, it functions as a Trithorax factor regulating in an opposed manner to PRC2 a common set of loci silenced by CLF. However, ULT1 does not have a methlytransferase activity but rather helps recruiting ATX members on the DNA (Carles and Fletcher, 2009). Hence, it is possible that ATX1, ATXR7, and ULT1 associate to form a Trithorax complex that may contain other proteins (Fig. 1). Similarly, the chromatin remodeling factors PICKLE and PICKLE RELATED2 antagonize CLF for the regulation of root meristematic growth (Aichinger et al., 2011). In conclusion, PRC2 is conserved in plants, but the regulatory PRC1 and TRX activities also are mediated by nonconserved protein complexes that evolved independently from their functional counterparts in metazoans. As many other annual species, Arabidopsis flowers in spring after an obligate period of cold in winter (vernalization). However, some natural accessions of Arabidopsis flower late in summer and do not require vernalization. Such natural variations rely primarily on the regulation of two factors, FRIGIDA and FLC (Koornneef et al., 1994; Lee et al., 1994). FLC encodes a MADS box transcription factor that represses flowering. At the FLC locus, H3K27me3 marks accumulate during vernalization, and FLC expression remains repressed throughout the life cycle until seed development initiates. Then, FLC becomes expressed again during early embryogenesis (Sheldon et al., 2008; Choi et al., 2009; Kim et al., 2009). The mode of deposition and removal of H3K27 methylation that accompanies the cycle of FLC expression have been studied in much detail and provide a good study case. The cycle can be divided in four steps: deposition, spreading, maintenance, and removal. How PRC2 is recruited on specific targets to deposit the H3K27me3 mark is still largely unknown. Some DNA sequences of the Hox loci in Drosophila are recognized by PRC2 complexes and are called Polycomb response elements (PREs). However, the degree of conservation of these elements is low across animal species, and no PRE has been found in plants to date. Nonetheless, at the FLC locus, some cis-sequences in the promoter and intron 1 are required for the establishment of repression (Sheldon et al., 2002). Some of these sequences play a role in assisting the recruitment of the noncoding RNA COLDAIR (for COLD ASSISTED INTRONIC NONCODING RNA) that is required for establishing stable repressive chromatin by recruiting PRC2 (Heo and Sung, 2011). In addition, COOLAIR (for COLD INDUCED LONG ANTISENSE INTRAGENIC RNA) is an antisense transcript initiated after the polyadenylation site of FLC that is induced by cold early in the vernalization process and silences FLC (Swiezewski et al., 2009). The role of long noncoding RNAs is reminiscent of the silencing by the XIST noncoding RNA and PRC2 in mammals (Margueron and Reinberg, 2011), and further analyses may show a generalization of this mechanism (Spitale et al., 2011). Hence, it is likely that PRE represents only the sequence complementary to a specific element of the long noncoding RNA that forms a complex with PRC2, and if true, this hypothesis would explain the lack of conservation of PREs. It is also possible that other mechanisms participate to PRC2 recruitment. In animals, the JumonjiC (JmjC) domain protein Jarid2 anchors PRC2 but inhibits its repressive action. It remains to be discovered whether plant homologs of Jarid2 fulfills the same role (Zheng and Chen, 2011). In addition, a combination of histone modifications could also be read by PRC2 complexes or associated proteins to specify a target. Analyses of transgenes carrying FLC and a fusion protein have shown that PRC2 can spread from an initial entry site to methylate histones at adjacent sequences (Schubert et al., 2006). However, genome-wide studies showed that H3K27me3 marks do not extend over large regions in plants as it does in flies and mammals but are restricted to discrete domains (Turck et al., 2007; Zhang et al., 2007). The PHD proteins VIN3, VRN5 (VIL1), and VEL1 (VIL2; Sung et al., 2006) that enhance the VNR complex activity at vernalization were shown to play a role in the spreading of H3K27me3 marks on FLC locus upon return of warm temperature (De Lucia et al., 2008). At the tissue level, spreading of H3K27me3 marks takes place only in dividing cells (Finnegan and Dennis, 2007). In root meristems, a pFLC-GUS reporter is expressed in a salt and pepper pattern after a limited period of vernalization. This is interpreted as two populations of cells, one with FLC still expressed and another with FLC expression suppressed. The gradual reduction of FLC expression would result from an increased proportion of cells with FLC expression turned off, and a mathematical model has been proposed to back up this hypothesis. The model proposes a switching mechanism involving the local nucleation of opposing histone modifications (Angel et al., 2011). The mark H3K27me3 is generally heritable through cell division for a stable repression. The PRC2 complex may directly bind to H3K27me3, most likely through the WD40 domain of FIE (Xu et al., 2010), thus ensuring a semiconservative mode of maintaining the PRC2 mark (Hansen et al., 2008). It was hypothesized that the spreading of H3K27me3 also plays a role in this mitotic inheritability, as a critical number of modified nucleosomes at one locus may be necessary to ensure the fidelity of the epigenetic information being passed along mitosis (Schubert et al., 2006). Possible additional mechanisms for this controlled inheritance have been proposed in mammals where it was shown that PcG proteins have the ability to stay bound on replicating DNA. As noncoding RNAs have been observed to play a role in PRC2 recruitment, RNA could be the molecule passing along the information (Sawarkar and Paro, 2010). In addition, H3K27me3 recognition by PRC1 induces chromatin compaction, thus enforcing a stable repression and cellular memory of PRC2 marks. But in the case of FLC, none of these mechanisms has been identified and the maintenance of H3K27me3 at FLC locus remains to be understood. FLC expression resumes after flowering, during the early stages of seed development in the embryo but not in the endosperm (Sheldon et al., 2008; Choi et al., 2009). This period coincides with the time when histones H3 transmitted by the gametes are actively removed after fertilization and replaced by newly synthesized histone H3 variants in the zygote but not in the endosperm (Ingouff et al., 2010). Thus, resetting the H3 proteins provides a potential global mechanism to discard H3K27me3 as well as other marks carried by the nucleosomes at the FLC locus (Fig. 3). This model is supported by the association between histones and FLC reactivation (Yun et al., 2011). Possible mechanisms causing H3K27me3 removal. A, Dilution by successive cell divisions: At each DNA replication cycle, the H3K27me3 mark is lost if not added actively on the newly formed DNA strand. B, Removal in a transcription factor-dependent manner. For flowering after vernalization, the transcription factor AG binds on the KNU locus, and H3K27me3 marks are gradually removed in an AG-dependent manner. The precise mechanism of this removal is still unknown. C, Demethylation by an H3K27me3 demethylase, such as REF6. D, Replacement of the H3 histone that bears the H3K27me3 mark by a variant H3 (H3v) that doesn’t carry the H3K27me3 mark. A second possible mechanism of removal of H3K27m3 is biochemical demethylation by a bone fide demethylase (Fig. 3). In animals, H3K27me3 is removed by KDM6a and KDM6b, two Jumonji domain-containing Lys demethylases. In Arabidopsis, RELATIVE OF EARLY FLOWERING6 (REF6), also known as JUMONJI12 (JMJ12), demethylates H3K27me3 and H3K27me2, whereas its metazoan counterparts, the KDM4 proteins (homologous in the JmjN and JmjC domains), are H3K9 and H3K36 demethylases (Lu et al., 2011). However, other proteins might be required for the demethylation of all H3K27me3 targets as there are at least two close homologs of REF6 in Arabidopsis, EARLY FLOWERING6 and JMJ13, that could act redundantly with REF6. The reactivation of FLC remains to be understood at the mechanistic level. The transition toward flowering represents one of the many sharp transitions between different phases of development that mark the plant life cycle. All plant species alternate a haploid life form (gametophyte) and a diploid form (sporophyte). The gametophyte produces gametes, which undergo fertilization resulting in a zygote. Zygotic division initiates a sporophyte that produces meiotic haploid spores. In flowering plants, the sporophytic life is divided into further transitions, including, for example, the differentiation of meristem cells into vegetative organs and the switch between an indeterminate vegetative meristem to a determinate inflorescence meristem. Here, we detail how PcG is involved in several phase developmental transitions (Fig. 4). PcG complexes mediate transitions from one stage of development to another. PRC2 complexes have been found to play an essential role in major transitions from one phase to another of plant development. The inner cycle marks the transition between the to the sporophytic mark transitions that place during the sporophytic development. In the life is the vegetative and the sporophyte is a in and In the mutants were for the homologs of FIE and CLF et al., 2009; et al., 2009). In the absence of and and are to or the of cells that sporophytic that PRC2 a transition from the to the sporophytic in Arabidopsis, mutants for the FIS complex an cell in the gametophyte et al., et al., This phenotype was interpreted as a of fertilization or endosperm development by the FIS complex. However, in the of the impact of PRC2 on transition in it is more likely that in Arabidopsis, FIS the of Hence, the tissue from the cell would This an role of PRC2 in the of the transition between the and sporophytic development requires the function of the second of the development is initiated by a of leading to a large a specific number of nuclear takes place and cellular endosperm is formed and 2009). accumulate in the cellular endosperm on the species, are in the endosperm or to the The transition to cellular endosperm is controlled by PRC2 of FIS function and The of a endosperm in mutants is by an expression of molecular of endosperm and the absence of expression of genes expressed in cellular endosperm (Ingouff et al., 2005). This is further by the of H3K27me3 on genes involved in cell and other of late endosperm development et al., 2010). Hence, PRC2 controls the transition from to cellular When the of the FIS complex for development and endosperm is it is possible to study the function of PRC2 during plant embryogenesis et al., 2001; et al., 2011). but and that PRC2 not only the transition between early seed to seed but also represses networks of genes involved in the are from plants, that the initial does not rely on PRC2 but rather does the maintenance of this pattern et al., 2011). The meristem initiates that the stem cell and toward development (Fig. 4). The genome-wide of H3K27me3 marks and the were between meristematic cells and to into the function of PRC2 in differentiation et al., 2011). H3K27me3 marks are removed at several of loci. A large of these loci transcription factors, some of which are identified as regulators of development. many of the transcriptional networks identified are also by which also during For instance, genes involved in the of the regulation are by H3K27me3, response factors. The mechanism leading to coordinated removal of the H3K27me3 mark remains The meristem until the transition to flowering. Then, the meristem becomes determinate and becomes an inflorescence meristem in produces floral (Fig. 4). We the role of PRC2 in the of flowering in response to vernalization. This role also to the of meristem other major targets of and the transcription factors and is essential for the transition to an inflorescence meristem. expression is repressed by PRC2 during embryo development et al., 2001). The MADS box transcription factor AG and the domain transcriptional repressor KNU ensure a determinate number of organs is by the floral meristem. expression is repressed by H3K27me3 in a temporal manner et al., 2009). AG expression is initiated and AG binds to the promoter of which is still repressed by the repressive mark is removed in a cell cycle in an AG-dependent the transcriptional to the KNU locus and induce transcription. This mechanism provides a that is to ensure that a proper cell number is by the floral meristem the flower PcG complexes PRC2 are However, there is limited conservation of the activities that read the H3K27me3 marks As shown in animals, long noncoding RNAs play a role in PRC2 et al., 2010), but it remains whether this is also the case in plants where long noncoding RNAs have been so far. other mechanisms might recruit PRC2 and explain PRC2 in animals, PRC2 in plants does not appear to play a major role in but is essential in the of developmental phase transitions (Fig. 4). This of maintenance and and the nature of these mechanisms is still not understood. phases are long (for example, the sporophytic phase in flowering and contain of more specific transitions (for example, vegetative to inflorescence to floral This that developmental transitions controlled by PRC2 gradually more do not rely on a of repressive marks. resetting mechanisms and how plants marks specific for developmental other PRC2 marks are during a transition this period are major of this We for with the