Understanding the genetic and epigenetic architecture in complex network of rice flowering pathways.

Although the molecular basis of flowering time control is well dissected in the long day (LD) plant Arabidopsis, it is still largely unknown in the short day (SD) plant rice. Rice flowering time (heading date) is an important agronomic trait for season adaption and grain yield, which is affected by both genetic and environmental factors. During the last decade, as the nature of florigen was identified, notable progress has been made on exploration how florigen gene expression is genetically controlled. In Arabidopsis expression of certain key flowering integrators such as FLOWERING LOCUS C (FLC) and FLOWERING LOCUS T (FT) are also epigenetically regulated by various chromatin modifications, however, very little is known in rice on this aspect until very recently. This review summarized the advances of both genetic networks and chromatin modifications in rice flowering time control, attempting to give a complete view of the genetic and epigenetic architecture in complex network of rice flowering pathways.

INTRODUCTION

Rice flowering time (heading date), which is affected by both endogenous and exogenous factors, is an important agronomic trait for regional and seasonal adaption. Heading on a proper time is the most critical step for grain production. Precocious flowering reduces the vegetative phase and leads to reduction of biological yield. On the other hand, delayed flowering could cause low seed setting percentage in cold late autumn or delay next planting season, which both results in production loss.

Florigen is produced in the leaf under inductive day length conditions and transported to the shoot apex where it triggers flowering transition (Cajlachjan, 1937; Corbesier et al., 2007; Tamaki et al., 2007). Unlike only one florigen gene FLOWERING LOCUS T (FT) in Arabidopsis, rice evolves two florigen genes, Heading date 3a (Hd3a) and RICE FLOWERING LOCUS T 1 (RFT1), and at least two flowering pathways are developed to control the expression of florigens, the Heading date 1 (Hd1) pathway which is conserved between rice and Arabidopsis, and the Early heading date 1 (Ehd1) pathway which is unique to rice (Doi et al., 2004) (Fig. 1). Numerous studies reveal that a large number of rice genes regulate flowering time through the two flowering integrators.

Figure 1

Comparison of core-flowering-pathways in rice and Arabidopsis. OsGI/GI-Hd1/CO-Hd3a/FT pathway is conserved between rice and Arabidopsis. CO. accelerates flowering under LD, however, Hd1 promotes flowering under SD and represses it under LD. Besides, Ghd7 and Ehd1 in rice, and FLC in Arabidopsis are unique flowering integrators, respectively. FLC and Ghd7 are major flowering suppressors, while Ehd1 acts as a flowering promoter

In Arabidopsis, some flowering regulators such as FLC and FT are reported to be regulated by various chromatin modifications (He, 2009; Liu et al., 2010). However, little is known in rice in this field. Recently, we characterized a major histone methyltransferase (HMTase) gene SET DOMAIN GENE 724 (SDG724), which is required for Histone H3 lysine 36 (H3K36) methylation, promotes rice heading, indicating that rice flowering could also be regulated by chromatin modifications (Sun et al., 2012). In the past two years, more and more molecular genetic studies gave the clues on the chromatin modification mechanism in rice flowering pathways, we summarize here most recent advances towards understanding of genetic networks and epigenetic chromatin modifications in rice flowering time control.

TWO FLORIGEN GENES Hd3a AND RFT1 IN RICE

Florigen, which has been hypothesized by many physiological studies, is believed to be produced in leaves by the inductive photoperiod, then moves to the shoot apical meristem (SAM) and triggers flowering transition. But this florigen has been eluded identification since it was first proposed for 70 years (Cajlachjan, 1937). In 2007, it was firstly revealed that FT encoded protein in Arabidopsis, is a leaf-derived long-distance signal directed to floral transition (Corbesier et al., 2007).

In rice, there are 13 FT homologs in the genome (Chardon and Damerval, 2005), Hd3a and RFT1 are two of them which were confirmed to act as florigen genes (Komiya et al., 2008; Komiya et al., 2009; Tamaki et al., 2007). By fusing Hd3a or RFT1 with GFP, it was demonstrated that Hd3a or RFT1 protein was expressed in vascular tissue of leaves, and could be moved to SAM where they started flowering induction. As Hd3a-GFP was only detected in the SAM of plants grown under short day conditions (SD), RFT1-GFP was merely detected under long day conditions (LD) (Komiya et al., 2009; Tamaki et al., 2007). On the other hand, Hd3a-RNAi (RNA interference) plants significantly delayed heading date under SD but not LD, RFT1-RNAi plants flowering was obviously delayed under LD but not SD oppositely. Furthermore, rice with knockout of both Hd3a and RFT1 caused at least 300 days late flowering under both SD and LD (Komiya et al., 2009). All these data demonstrated that, unlike Arabidopsis, rice has two florigen genes, Hd3a and RFT1, Hd3a is responsible for flowering under inductive SD, whereas RFT1 is responsible for flowering transition under non-inductive LD. Although Hd3a and RFT1 are located in some chromosome and separated by only 11.5 kb in the genome, the fine-tuning of long day flowering by the H3K36me2/3 level of RFT1 but not Hd3a via SDG724, therefore, RFT1 and Hd3a which have functionally diverged to control flowering time under LD and SD conditions are partly due to a fine-tuned epigenetic mechanism (Sun et al., 2012).

FLORIGEN REGULATED NETWORK

How flowering pathways are regulated differs in plants. In Arabidopsis, flowering is controlled by a small number of large-effect genes such as FLC (Salome et al., 2011), whereas in maize is controlled by many additive small-effect quantitative trait loci (QTLs) (Buckler et al., 2009). Interestingly, rice combines both regulatory manners, including a few large-effect factors, such as Hd1, Ehd1, and Grain number, plant height andheading date 7 (Ghd7), in addition to some small-effect QTLs and genes (Ebana et al., 2011; Tsuji et al., 2013) (Table 1).

Table 1

Flowering regulators in rice

So far, quite a number of QTLs controlling rice heading date (Hd) were identified and characterized using different segregating populations derived from crossing a japonica cultivar (Nipponbare) and an indica cultivar (Kasalath) (Lin et al., 1998; Yano et al., 1997). These QTLs include the major loci controlling photoperiodic flowering responses, Hd1 (Yano et al., 1997; Yano et al., 2000), Hd2/Ghd7.1/OsPRR37 (Oryza sativa Pseudo-Response Regulator 37) (Koo et al., 2013; Liu et al., 2013; Shibaya et al., 2011; Yamamoto et al., 2000; Yan et al., 2013), Hd3a (Kojima et al., 2002), Hd4/Ghd7 (Ghd7 for short) (Koo et al., 2013; Xue et al., 2008), Hd5/Days to heading 8/Grain number, Plant height, and Heading date 8/Late Heading Date 1 (Hd5/DTH8/Ghd8/LHD1) (Dai et al., 2012; Fujino et al., 2013; Yan et al., 2011). Furthermore, backcross progenies derived from the same original cross allowed identification of other QTLs, such as Hd6/CK2 (CASEIN KINASE 2) (Ogiso et al., 2010; Takahashi et al., 2001; Yamamoto et al., 2000), Hd14/Ehd1 (Doi et al., 2004), Hd16/EL1 (Early flowering 1) (Dai and Xue, 2010; Hori et al., 2013; Shibaya et al., 2011), Hd17/OsELF3/EF7/OsEF3/Hd3b (Hd17/Oryza sativa Early Flowering 3/Early Flowering 7/Oryza sativa Early Flowering 3/Hd3b, OsELF3 for short) (Hori et al., 2012; Matsubara et al., 2012; Saito et al., 2012; Yang et al., 2013; Zhao et al., 2012). Additionally, using rice near isogenic lines and mutants, more genes implicated in controlling flowering time have been identified and positioned into a regulatory network (Brambilla and Fornara, 2013; Itoh and Izawa, 2013; Tsuji et al., 2011, 2013) (Fig. 2).

Figure 2

Complex flowering time control network in rice. Rice flowering network is formed by two florigen genes Hd3a and RFT1, and four regulation modules, including Hd1-dependent pathway, Ehd1-dependent pathway, crosstalk between Hd1 and Ehd1 pathway, and flowering regulators independent of Hd1 and Ehd1. The first three signals come together to regulate Hd1 and Ehd1 and affect florigen gene expression; the last module may directly control the expression of florigen genes independent of Hd1 and Ehd1. Besides, expressions of Ehd3, RID1, OsMADS50, Hd3a, and RFT1 can be affected by H3K36me2/3; OsLFL1 and OsLF transcriptions can be mediated by H3K27me3. Finally, all those florigen signals are transported from leaves to SAM and trigger flowering transition there. All the gene names for short are showed in Table 1

Hd1-DEPENDENT PATHWAY

There is a similar molecular system for florigen control in Arabidopsis and rice (Izawa, 2007; Tsuji et al., 2011). Hd1 and Hd3a in rice are homologs of CONSTANS (CO) and FT in Arabidopsis, respectively. As in Arabidopsis, Hd1 acts upstream of Hd3a (Kojima et al., 2002; Yano et al., 2000), and overexpression of a rice ortholog of Arabidopsis GIGANTEA (GI) which acts upstream of CO, namely OsGI, increased the expression of Hd1 in the transgenic plants, followed by suppressing Hd3a expression, resulting in late flowering under both SD and LD (Hayama et al., 2003). Differently, CO merely promotes FT expression, Hd1 plays a more enigmatic role in rice, which promotes flowering under SD, but represses flowering under LD (Hayama et al., 2003; Komiya et al., 2008; Lin et al., 2000; Tamaki et al., 2007). These results indicate that the core photoperiodic pathway composed of the three key flowering genes OsGI/GI-Hd1/CO-Hd3a/FT is conserved between rice and Arabidopsis, but its function has diverged during evolution to produce opposite flowering responses. While the photoperiodic pathway in Arabidopsis merely accelerates flowering under LD, in rice, it promotes flowering under SD and represses flowering under LD (Takahashi and Shimamoto, 2011).

The reversible mechanism that Hd1 functions as either an activator or suppressor of Hd3a involves the action of the red-light photoreceptor phytochrome B (phyB), since mutations in phyB or phytochrome chromophore synthesis, such as photoperiod sensitivity 5 (se5), attenuate this conversion and maintain Hd1 as an activator under any photoperiodic conditions. On the other hand, though Hd1-overexpressing plants delay flowering, Hd1 protein levels in these plants are not significantly altered (Andres et al., 2009; Ishikawa et al., 2011; Izawa et al., 2002), thus it is speculated that LD light signals may modify the protein of Hd1 or Hd1 complex through phytochrome but not its expression levels, and convert it into a repressor of flowering. Therefore, uncovering of the biochemical function of Hd1 protein and the molecular nature of its dual activity will provide exciting insight into the control of photoperiodic flowering in rice.

Recently, it is deduced that Hd1 protein activity is possible affected by an additional posttranslational regulatory factor, Hd6, which encodes a CK2 α-subunit (Ogiso et al., 2010; Takahashi et al., 2001). The delay flowering effect of Hd6 is observed only when Hd1 is functional, however, Hd1 is not phosphorylated by Hd6 in vitro (Ogiso et al., 2010), suggesting that Hd6 phosphorylates unknown substrates that cooperate with Hd1 in the LD floral suppression pathway.

Ehd1 DEPENDENT PATHWAY

In 2004, a novel regulatory Ehd1-pathway which is not presented in Arabidopsis, is discovered in rice (Doi et al., 2004). Ehd1, encoding a B-type response regulator, is a floral promoter, and rice variety Taichung 65 (T65) without functional Ehd1 allele delays flowering under both LD and SD (Doi et al., 2004). As it has been shown that Ehd1 contributes to flowering time by its expression levels (Takahashi et al., 2009), thus fine-tuning of Ehd1 expression is crucial for rice flowering at suitable time, and several flowering regulators have been identified to participate in this regulation.

Ghd7, which is important for increasing rice productivity and adaptability, is a major regulator of Ehd1 and could delay flowering by repressing Ehd1 under LD (Takahashi et al., 2009; Xue et al., 2008). As Ghd7 encodes a CCT (CONSTANS, CO-like, and TOC1) domain protein, which shows very low homology to Arabidopsis genome, the Ghd7-Ehd1 may be a unique pathway in rice (Koo et al., 2013; Xue et al., 2008). Further study shows that Ghd7 and Ehd1 can respectively set a daylength threshold for Hd3a expression, which is usually observed in SD plants but not in LD plants (Itoh et al., 2010; Takimoto and Ikeda, 1961), and this capacity of discernment in critical day length in rice greatly enriches the daylength-dependent regulated mechanism of florigen gene expression.

Until now, at least three genes, Early heading date 3 (Ehd3), ELF3, and Hd16/EL1, were identified to control Ghd7 expression in Ehd1-pathway. Ehd3 encodes a plant homeodomain (PHD) finger protein and is identified as one repressor of Ghd7. Generally, Ghd7 transcript reaches its highest level after seeding for two weeks, and then the expression is gradually reduced to a basal level, but in ehd3 mutants, Ghd7 expression level is always higher and delays heading date for more than one year under LD. Interestingly, under SD, Ehd3 could promote Ehd1 expression regardless of Ghd7, suggesting a perplexed role of Ehd3 (Matsubara et al., 2011).

ELF3 in Arabidopsis is responsible for generating circadian rhythm and regulating photoperiodic flowering, consistently, its homolog in rice OsELF3 is also required to sustain the robust oscillation, and lesions in OsELF3 delay flowering under both SD and LD (Saito et al., 2012; Yang et al., 2013; Zhao et al., 2012). Under SD, OsELF3 promotes flowering mainly by repressing Ghd7, because late flowering of oself3 mutants can be rescued if Ghd7 but not Hd1 is mutated. Under LD, oself3 mutants increase OsGI and Ghd7 expression, thus up-regulate Hd1 and repress Ehd1 expression, respectively, indicating that OsELF3 influences photoperiodic flowering in both Hd1 and Ehd1 pathways (Brambilla and Fornara, 2013; Saito et al., 2012).

Hd16/EL1, encoding a casein kinase I protein, is associated with the gibberellin-mediated flowering transition (Dai and Xue, 2010). Deficient in Hd16 weakens rice photoperiod sensitivity, but increases Ehd1, Hd3a, and RFT1 expression under LD. Though the expression level of Ghd7 is not significantly altered in el1 mutants, the biochemical data indicate that Hd16 acts as a flowering repressor by phosphorylation of Ghd7 (Dai and Xue, 2010; Kwon et al., 2013).

OsLFL1 (Oryza sativa LEC2 and FUSCA3 Like 1) encodes a putative B3 transcription factor, knockdown of OsLFL1 does not affect flowering time, while ectopic overexpression of OsLFL1 decreases Ehd1 expression and results in late flowering (Peng et al. 2007, 2008). OsLFL1 is controlled by two members of MIKC-type MADS-box family, OsMADS50 and OsMADS56. Both osmads50 mutants and OsMADS56-overexpressing plants, which produce increased OsLFL1 expression, show late flowering phenotype (Lee et al., 2004; Ryu et al., 2009). Interestingly, OsMADS56 can interact with OsMADS50 in vitro, suggesting that the two MADS-box members tend to form a heterodimer complex and function antagonistically through OsLFL1-Ehd1 pathway under LD (Ryu et al., 2009).

As mentioned in Hd1-pathway, phytochrome is probably a primary cause of Hd1-dependent suppression of rice flowering, but underlying molecular mechanism of phytochrome in Ehd1-pathway is not well understood. Recent studies showed that SE5 and phyB also suppress Ehd1 expression, and the phyB-mediated suppression of Ehd1 is confirmed to be repressed by a CONSTANS-like (COL) gene OsCOL4 (Oryza sativaCONSTANS-like4) (Andres et al., 2009; Komiya et al., 2009; Lee et al., 2010). OsCOL4 expression is decreased in osphyB mutants, and osphyB oscol4 double mutants flower is similar to osphyB single mutants, indicating that OsCOL4 functions downstream of OsphyB (Lee et al., 2010).

Besides the above regulators, Ehd1 expression is also modulated by other four flowering factors independently. Indeterminate 1 (ID1) is one of them, which expresses in leaf but induces flowering in the shoot meristem. ID1 has been once thought to be involved in the florigen synthesis in maize (Colasanti et al., 2006; Colasanti et al., 1998), and its regulated mechanism has been exhibited in rice. Lesions in rice RID1 (Early heading date 2/OsINDETERMINATE 1/Rice INDETERMINATE 1, Ehd2/OsID1/RID1, RID1 for short) lead to extremely late flowering phenotype, as well as decreased expression of Ehd1 and downstream florigen genes under both SD and LD (Matsubara et al., 2008; Park et al., 2008; Wu et al., 2008).

Ehd4 (Early heading date 4), encoding a CCCH-type zinc finger transcriptional regulator, is expressed mostly in immature leaves and shows a similar diurnal expression pattern of Ehd1 under both SD and LD. Ehd4 up-regulates the expression of the florigen genes Hd3a and RFT1 through Ehd1. Strikingly, Ehd4 is highly conserved in both wild rice and cultivated rice, but homologs cannot be found in other species, suggesting that Ehd4 is unique flowering regulator in Oryza genus differed from other grass members during evolution (Gao et al., 2013).

OsMADS51 is another MADS box gene, other than OsMADS50 and OsMADS56, it acts downstream of OsGI, transmits a promotion signal from OsGI to Ehd1 under SD. Though its null mutants showed late flowering phenotype followed by decreased expression of Ehd1 and Hd3a, ectopic expression of OsMADS51 causes early flowering, accompanying with increased expression Ehd1 and Hd3a (Kim et al., 2007).

Hd5/DTH8/Ghd8/LHD1 encodes a putative HEMEACTIVATOR PROTEIN 3 (HAP3) subunit of a CCAAT-box binding protein (HAP complex) that binds to CCAAT boxes in yeast and animals. Similar to Hd1, Hd5/DTH8/Ghd8/LHD1 delayed flowering in rice under LD and promotes flowering under SD, but by regulating expression of Ehd1 (Dai et al., 2012; Lin et al., 2003; Wei et al., 2010; Yan et al., 2011).

Most interestingly, though Hd5/DTH8/Ghd8/LHD1 suppresses rice heading though Ehd1, genetic analysis implies that Hd5 requires functional Hd1 to repress flowering under LD (Nonoue et al., 2008), rising a question what is the relationship between Hd1 and Ehd1. Recent findings indicate that transcript level of Ehd1 is down-regulated in Hd1-overexpression transgenic lines, suggesting that, to some degree, Hd1 is an upstream regulator of Ehd1 expression, but how this crosstalk works is still undefined (Ishikawa et al., 2011).

FLOWERING REGULATORS INDEPENDENT OF Hd1 AND Ehd1

Besides Ehd1, T65 also bears a loss-of-function allele of Hd1, but it could still flower in time and serves as a commercial rice variety, so there are must some other regulators independent of Hd1 and Ehd1 in rice flowering network (Doi et al., 2004). OsCO3 and DTH2 are two of them, and promote flowering by regulating florigen genes. Though both of them are COL genes, they function under different photoperiodic conditions. Expressions of Hd3a and FT-like genes are decreased in the OsCO3-overexpressing plants under SD without altered expression of other florigen upstream regulators, suggesting that OsCO3 primarily controls flowering time under SD by negatively regulating the expression of florigen genes, independent of other known SD-promotion pathways (Kim et al., 2008). For DTH2, both association analysis and transgenic experiments indicate that two functional nucleotide polymorphisms that correlated with early heading and increased reproductive fitness under natural LD in northern Asia. Further combined population genetics and network analyses suggest that DTH2 probably represents a target of artificial selection for adaptation to LD during rice domestication and improvement, demonstrating an important role of minor effect quantitative trait loci in crop adaptation and breeding (Wu et al., 2013).

Although some PRR genes are major components of the circadian oscillator, a rice PRR gene Hd2/Ghd7.1/OsPRR37 may down-regulate Hd3a expression independent of any known pathways to suppress flowering under LD. As lesions in Hd2/Ghd7.1/OsPRR37 cause early flowering phenotype, the japonica varieties harboring nonfunctional alleles of both Ghd7 and Hd2/Ghd7.1/OsPRR37 flower extremely early under natural LD, and make these varieties adapt to the northernmost rice cultivation regions. Further study implied that natural variations in Hd2/Ghd7.1/OsPRR37 have contributed to the expansion of rice cultivation to temperate and cooler regions (Koo et al., 2013; Liu et al., 2013; Yan et al., 2013).

Different from Hd2/Ghd7.1/OsPRR37, OsDof12 is LD-specific flowering repressor and encodes a DNA-binding with one finger (Dof) transcription factor which is involved in a variety of biological processes of plants. The transcriptions of OsDof12 can express at different development stages, but strongly inhibited by dark treatment. OsDof12-overexpressing plants flower earlier in consistent with the up-regulation of Hd3a independent of other flowering genes under LD but not SD (Li et al., 2009).

CHROMATIN MODIFICATIONS REGULATE FLOWERING IN RICE

Chromatin, which is composed by complexing DNA with histone, carries not only genetic, but also epigenetic information. In Arabidopsis, the expression of a major flowering repressor FLC is regulated by a number of active and repressive chromatin modifications, such as histone tails methylation, acetylation, ubiquitination etc. In addition, histone modifications can also directly regulate the expression of florigen gene FT, and the regulation manner of FLC and FT provides a paradigm for control of developmental regulators through chromatin modifications (He, 2009). Currently, not so many data are available about that in rice, but molecular genetic studies indicated that rice flowering control also undergoes the complex chromatin modifications (Table 2).

Table 2

Chromatin modification regulators in rice

Gene name	Pathways	Gene ID	Note	Modifications	Target genes
SDG724	Ehd1	Os09g0307800	SET domain group protein	H3K36me2/3	RFT1, OsMADS50
SDG725	Ehd1	Os02g0554000	SET domain group protein	H3K36me2/3	Ehd3, Ehd2, OsMADS50,Hd3a, RFT1
LC2/OsVIL3	Hd1	Os02g0152500	Plant homeo domain (PHD) finger	H3K27me3	OsLF
OsVIL2	Ehd1	Os12g0533500	Plant homeo domain (PHD) finger	H3K27me3	OsLFL1
OsEMF2b	Ehd1	Os04g0162100	C2H2 zinc-finger protein, interact with OsVIL2	H3K27me3	OsLFL1
OsTrx1	Ehd1	Os09g0134500	SET domain group protein, interact with Ehd3	Unkown	Unkown

ACTIVE CHROMATIN MODIFICATIONS AND RICE FLOWERING

S-Adenosyl-l-methionine is a universal methyl group donor involved in numerous transmethylation reactions, including histone methylation. Knockdown of rice S-Adenosyl-l-methionine synthetase (SAMS) 1, 2, and 3 greatly reduced the expression of Ehd1, Hd3a, RFT1 and led to a late flowering phenotype. Moreover, the histone H3K4me3 and symmetric DNA methylation at these genes was significantly reduced, suggesting an association between epigenetic modification and flowering in rice, but more research are required on this relationship (Li et al., 2011).

We have demonstrated that SDG724, a histone methyltransferase gene which belongs to SET domain family Class II (Ng et al., 2007), affected flowering time by mediating H3K36 methylation in rice. SDG724 loss-of-function mutant lvp1 showed a late flowering phenotype under both LD and SD, which was associated with the suppressed expression of RFT1 and Hd3a. Interestingly, only the chromosomal region of RFT1, but not Hd3a, reduced the level of H3K36me2/3 modifications which associated with the transcriptionally active chromatin state, although the two florigenic genes are closely linked in the genome and separated by only 11.5 kb (Sun et al., 2012). This similar regulated way in RFT1 is also found in a previous report that RFT1 expression can be promoted through another active histone modification H3K9 acetylation around the transcriptional start site of its chromatin in Hd3a-RNAi transgenic plants (Komiya et al., 2008). In conclusion, both of the two findings suggest an epigenetic regulation mechanism through RFT1. In addition, SDG724 also affects the histone modification state at OsMADS50 chromosomal region, thus all the results suggest a LD floral promotion pathway mediated by H3K36me2/3 deposition through OsMADS50-Ehd1-RFT1 pathways in rice (Sun et al., 2012).

Coincidentally, another member of Class II in SET domain family (Ng et al., 2007), SDG725, is also involved in promoting rice flowering through H3K36me2/3. In SDG725 knockdown plants, the expression levels of Ehd3, RID1, OsMADS50, OsMADS51, Ehd1, Hd3a, and RFT1 were all drastically reduced, but the Ghd7 expression was increased, under either SD or LD. Different from SDG724, SDG725 is required for deposition of H3K36me2/3 at more flowering gene loci, such as Ehd3, RID1, OsMADS50, Hd3a, and RFT1. Thus, SDG724 and SDG725 regulate both overlapped and specific flowering genes by mediating H3K36me2/3 deposition and promote rice flowering, which are different to the previously known function of these epigenetic marks in Arabidopsis flowering (Sui et al., 2012; Xu et al., 2008; Zhao et al., 2005).

Very recently, another homolog of SDG724, OsTrx1, which belongs to SET domain family Class III (Ng et al., 2007), might activate or maintain the active transcribed states of target genes, and was reported to delay flowering time under LD through Ghd7 pathway but not OsMADS50 and Hd1 pathways (Choi et al., 2014). Though expression of Ehd3 that functions upstream of Ghd7 is unchanged in ostrx1 mutants, it was proved that OsTrx1 could bind to Ehd3 in vitro. Further study showed that PHD motif of OsTrx1 could bind to native histone H3 and the C-terminal end of SET domain of OsTrx1 has histone H3 methyltransferase activity, thus OsTrx1 and Ehd3 tend to form a complex to methylate downstream genes, but further studies are needed to illuminate its function in detail (Choi et al., 2014).

REPRESSIVE CHROMATIN MODIFICATIONS AND RICE FLOWERING

Arabidopsis VILs (VIL, VERNALIZATION INSENSITIVE), VIN3, and VRN5 are components of PRC2 (Polycomb Repressive Complex 2), mediating the H3K27 trimethylation at the FLC locus to repress its expression and hence to induce flowering. In rice, a VIL homolog gene LC2/OsVIL3 is considered as a possible component of PRC2 complex, and lc2 mutants display late flowering along with the reduced expression of Hd1 and Hd3a under SD. Furthermore, consistent with the result that OsLF (Oryza sativaLate Flowering) directly repressed Hd1 expression (Zhao et al., 2011), LC2/OsVIL3 binds to the promoter region of OsLF and represses the OsLF expression via H3K27me3 methylation, thus eventually promotes flowering (Wang et al., 2012).

OsVIL2 may be another VILs member in rice PRC2 complex, and mutations in OsVIL2 cause late flowering under both SD and LD. Different from LC2/OsVIL3, the late flowering phenotype is associated with increased OsLFL1 and reduced Ehd1, Hd3a, RFT1 expression. Furthermore, OsVIL2 can bind to native histone H3 in vitro and is directly associated with OsLFL1 chromatin in vivo, and H3K27me3 is significantly reduced on OsLFL1 chromatin in osvil2 mutants compared to the wild type, indicating that OsVIL2 epigenetically represses OsLFL1 expression to promote flowering in rice. Besides, OsVIL2 can physically interact with OsEMF2b, which may be also a member of PRC2. Similar to osvil2, a null mutation of OsEMF2b caused late flowering by increasing OsLFL1 and decreasing Ehd1 expression (Wang et al., 2012; Yang et al., 2012).

In short, similar to Arabidopsis, LC2/OsVIL3, OsVIL2, and OsEMF2b may function together with PRC2 to induce flowering by affecting histone modification H3K27me3, but their target flowering genes are different, indicating that a diverse flowering pathway regulated by PRC2 in rice flowering.

CONCLUSION AND PERSPECTIVES

Heading date is an important agronomic trait that determining rice to grow in different regions and seasons. In last two decades, tremendous progress has been made by the study of QTLs and genes controlling rice flowering, which not only identified the nature of the mobile signal florigen, but also unveiled a complex genetic network that controls florigen in rice. Hd3a and RFT1, two florigens regulated respectively rice flowering under SD and LD, are mainly controlled through Hd1 and Ehd1 pathways. However, as mentioned in various T65 with lack of both, rice also develops some additional pathways that could induce rice flowering.

Histone modification is very important for defining transcriptional regulation expression, thus plays a fundamental role in plant growth and development, as well as responding to various environmental conditions. These modification marks are dynamically “written” and “erased”, and then specifically recognized by the “readers” and instruct specific biological process, such as flowering. Very recently, a large number of studies have revealed that various ‘active’ histone modifications, H3K4 methylation, H2B monoubiquitination, H3K36me2/me3, histone deacetylation, and ‘repressive’ chromatin modifications, H3K4 demethylation, H3K9 methylation, H3K27 methylation, histone arginine methylation, are involved in modulating FLC expression in Arabidopsis. Though the regulation of FLC expression via chromatin modification provides a paradigm in flowering gene expression, whether there exists a major flowering regulator such as FLC in rice is still unknown (Fig. 1). Rice possibly has some new routes in its flowering control. In rice, a number of studies revealed the difference in chromatin modification mechanism in the past two years, ‘active’ H3K36me2/3, H3K4me3, H3K9 acetylation and ‘repressive’ H3K27me3 modifications mediate flowering time through Hd1 and Ehd1 dependent pathways, and our finding about SDG724 also suggests a LD floral promotion pathway that could be mediated via an epigenetic regulation of florigen RFT1 itself. All these data suggest that the target flowering genes of chromatin modifications are dispersed in both conserved and unique flowering pathways in rice. Taken together, all the progress in rice, along with Arabidopsis, provides a complete evolutionarily comparative view of genetic and epigenetic flowering mechanisms in plants until now.

Furthermore, in rice, some histone modification participators tend to function under SD and LD, but others like to function mainly under SD or LD, thus unveiling of histone modification mechanism in rice flowering might set a solution to verify the relationships between particular histone modifications and photoperiod environments. On the other hand, as a LD plant, Arabidopsis flowering is accelerated by LD, but SD plant rice flowers earlier under SD than under LD, further study will be helpful to distinguish the function and evolutionary process of histone modification in various photoperiodic plants. Thus, it would be of great interest to identify more chromatin modification regulators and their target genes in rice flowering in future.