Resetting and regulation of FLOWERING LOCUS C expression during Arabidopsis reproductive development

Authors

  • Jean Choi,

    1. Department of Biological Sciences, Seoul National University, Seoul, 151-742, Korea,
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Youbong Hyun,

    1. Department of Biological Sciences, Seoul National University, Seoul, 151-742, Korea,
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Min-Jeong Kang,

    1. Department of Biological Sciences, Seoul National University, Seoul, 151-742, Korea,
    2. Global Research Laboratory for Floral Regulatory Signaling at SNU & UW, Seoul National University, Seoul 151-742, Korea,
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Hye In Yun,

    1. Department of Biological Sciences, Seoul National University, Seoul, 151-742, Korea,
    Search for more papers by this author
  • Jae-Young Yun,

    1. Department of Biochemistry, University of Wisconsin, Madison, Wisconsin 53706, USA,
    Search for more papers by this author
  • Clare Lister,

    1. Department of Cell and Developmental Biology, John Innes Centre, Norwich, NR4 7UH, UK, and
    Search for more papers by this author
  • Caroline Dean,

    1. Department of Cell and Developmental Biology, John Innes Centre, Norwich, NR4 7UH, UK, and
    Search for more papers by this author
  • Richard M. Amasino,

    1. Global Research Laboratory for Floral Regulatory Signaling at SNU & UW, Seoul National University, Seoul 151-742, Korea,
    2. Department of Biochemistry, University of Wisconsin, Madison, Wisconsin 53706, USA,
    Search for more papers by this author
  • Bosl Noh,

    Corresponding author
    1. Global Research Laboratory for Floral Regulatory Signaling at SNU & UW, Seoul National University, Seoul 151-742, Korea,
    2. Environmental Biotechnology National Core Research Center, Gyeongsang National University, Jinju, 660-701, Korea
      (fax +82 2 871 6673; e-mail yhc@snu.ac.kr, ysnoh@snu.ac.kr, or bnoh2003@yahoo.co.kr).
    Search for more papers by this author
  • Yoo-Sun Noh,

    Corresponding author
    1. Department of Biological Sciences, Seoul National University, Seoul, 151-742, Korea,
    2. Global Research Laboratory for Floral Regulatory Signaling at SNU & UW, Seoul National University, Seoul 151-742, Korea,
      (fax +82 2 871 6673; e-mail yhc@snu.ac.kr, ysnoh@snu.ac.kr, or bnoh2003@yahoo.co.kr).
    Search for more papers by this author
  • Yeonhee Choi

    Corresponding author
    1. Department of Biological Sciences, Seoul National University, Seoul, 151-742, Korea,
      (fax +82 2 871 6673; e-mail yhc@snu.ac.kr, ysnoh@snu.ac.kr, or bnoh2003@yahoo.co.kr).
    Search for more papers by this author

(fax +82 2 871 6673; e-mail yhc@snu.ac.kr, ysnoh@snu.ac.kr, or bnoh2003@yahoo.co.kr).

Summary

The epigenetic regulation of the floral repressor FLOWERING LOCUS C (FLC) is one of the critical factors that determine flowering time in Arabidopsis thaliana. Although many FLC regulators, and their effects on FLC chromatin, have been extensively studied, the epigenetic resetting of FLC has not yet been thoroughly characterized. Here, we investigate the FLC expression during gametogenesis and embryogenesis using FLC::GUS transgenic plants and RNA analysis. Regardless of the epigenetic state in adult plants, FLC expression disappeared in gametophytes. Subsequently, FLC expression was reactivated after fertilization in embryos, but not in the endosperm. Both parental alleles contributed equally to the expression of FLC in embryos. Surprisingly, the reactivation of FLC in early embryos was independent of FRIGIDA (FRI) and SUPPRESSOR OF FRIGIDA 4 (SUF4) activities. Instead, FRI, SUF4 and autonomous-pathway genes determined the level of FLC expression only in late embryogenesis. Many FLC regulators exhibited expression patterns similar to that of FLC, indicating potential roles in FLC reprogramming. An FVE mutation caused ectopic expression of FLC in the endosperm. A mutation in PHOTOPERIOD-INDEPENDENT EARLY FLOWERING 1 caused defects in FLC reactivation in early embryogenesis, and maintenance of full FLC expression in late embryogenesis. We also show that the polycomb group complex components, Fertilization-Independent endosperm and MEDEA, which mediate epigenetic regulation in seeds, are not relevant for FLC reprogramming. Based on our results, we propose that FLC reprogramming is composed of three phases: (i) repression in gametogenesis, (ii) reactivation in early embryogenesis and (iii) maintenance in late embryogenesis.

Introduction

The correct timing of flowering is essential for the survival of plant species. Plants have evolved a complex regulatory network that adjusts flowering time in response to various environmental and endogenous signals. FLOWERING LOCUS C (FLC), a floral repressor, is one of the central regulators of flowering in Arabidopsis (Michaels and Amasino, 1999; Sheldon et al., 1999). FLC encodes a MADS domain-containing transcription factor that inhibits the transcription of downstream floral activators. Expression of FLC is promoted by FRIGIDA (FRI), and is repressed by sets of genes in the autonomous and vernalization pathways (Sheldon et al., 1999, 2000; Michaels and Amasino, 2001).

The autonomous pathway is composed of a group of genes that repress FLC expression in the absence of functional FRI (Sheldon et al., 1999, 2000; Michaels and Amasino, 2001). Among them, FLOWERING LOCUS D (FLD; He et al., 2003), FVE (Ausin et al., 2004) and RELATIVE OF EARLY FLOWERING 6 (REF6; Noh et al., 2004) encode a lysine-specific demethylase 1 (LSD1) class putative histone demethylase, a homolog of a retinoblastoma-associated protein and a Jumonji domain-containing putative histone demethylase, respectively. These might repress FLC transcription via chromatin modification. FCA (Macknight et al., 1997), FPA (Schomburg et al., 2001), FY (Simpson et al., 2003) and FLOWERING LOCUS K (FLK; Lim et al., 2004) encode RNA-binding or RNA-processing proteins. Although a recent study suggested that FCA and FPA play a role in RNA-dependent chromatin modification (Baurle et al., 2007), the molecular mechanisms by which these proteins repress FLC are largely unknown. The putative homeodomain protein LUMINIDEPENDENS (LD; Lee et al., 1994) and the Arabidopsis CREB-binding protein (CBP) homologs HISTONE ACETYLTRANSFERASEs OF THE CBP FAMILY (HACs; Han et al., 2007) are also categorized as autonomous-pathway members. However, the biochemical roles of these proteins in FLC repression are not understood.

The late-flowering habit of winter annual Arabidopsis is conferred by dominant alleles of FRI and FLC (Gazzani et al., 2003; Michaels et al., 2003). FRI, a protein with two coiled-coil domains, elevates FLC expression, even in the presence of the autonomous-pathway repressors (Johanson et al., 2000). SUPPRESSOR OF FRIGIDA 4 (SUF4), a C2H2-type zinc-finger protein, was recently characterized as an interacting partner of FRI (Kim et al., 2006). SUF4 binds to the FLC promoter, and might recruit a protein complex containing FRI to activate FLC. Arabidopsis homologs of the members of yeast RNA polymerase II-associated factor 1 (PAF1) complex, EARLY FLOWERING 7 (ELF7), also known as VERNALIZATION INDEPENDENCE 2 (VIP2), ELF8 (VIP6), VIP4 and VIP5, are required for elevated FLC expression (He et al., 2004; Oh et al., 2004; Kim et al., 2005). EARLY FLOWERING IN SHORT DAYS (EFS), also known as SET DOMAIN GROUP 8 (SDG8), a SET-domain containing putative histone methyltransferase, is also required for FLC activation, and either trimethylation at histone H3 lysine 4 (H3K4; Kim et al., 2005) or dimethylation at histone H3 lysine 36 (H3K36; Zhao et al., 2005) in FLC chromatin. Thus, mutations in these genes prevent the expression of FLC in both FRI-containing winter annuals and in autonomous-pathway mutants. In addition, PHOTOPERIOD-INDEPENDENT EARLY FLOWERING 1 (PIE1; Noh and Amasino, 2003), SUPPRESSOR OF FRIGIDA 3/ACTIN-RELATED PROTEIN 6/EARLY IN SHORT DAYS 1 (SUF3/ARP6/ESD1; Choi et al., 2005; Deal et al., 2005; Martin-Trillo et al., 2006) and AtSWC6/SERRATED LEAVES AND EARLY FLOWERING (SEF; Choi et al., 2007; March-Diaz et al., 2007) have been isolated as Arabidopsis homologs of members of the yeast SWR1 complex. This complex mediates the exchange of histone protein H2A with its variant H2A.Z, and this process is required for the full activation of FLC (Choi et al., 2007).

Vernalization establishes competence for flowering in winter annuals after the prolonged cold of winter (Sung and Amasino, 2005). Vernalization leads to a series of repressive modifications in FLC chromatin, and this repression of FLC permits the photoperiod pathway to accelerate flowering. The transcriptional induction of VERNALIZATION INSENSITIVE 3 (VIN3) is an initial step in vernalization-induced FLC repression (Sung and Amasino, 2004). Expression of the PHD domain protein VIN3 is necessary for the deacetylation of histone H3, and the methylation at histone H3 lysine 9 (H3K9) and histone H3 lysine 27 (H3K27) within FLC chromatin during cold treatment. After this, the B3-domain protein VERNALIZATION 1 (VRN1) and the polycomb group protein VERNALIZATION 2 (VRN2) maintain the repressed state of FLC chromatin (Gendall et al., 2001; Levy et al., 2002; Bastow et al., 2004; Sung and Amasino, 2004). Subsequently, vernalization-mediated FLC repression is stably maintained under warm conditions. However, this ‘memory of winter’ is reset in the next generation, and this reprogramming is critical to reestablish the vernalization requirement each generation.

Various studies have focused on isolating FLC regulators and understanding how they regulate FLC transcription in the post-embryonic vegetative developmental stages. In contrast, less is known about the resetting of FLC during reproductive development. In this work, we have studied the expression patterns of FLC and a variety of FLC regulators, and have determined the effects of FLC regulators on FLC expression during reproductive development. Our results indicate the existence of an epigenetic reprogramming of gene expression, which takes place during gametogenesis and embryogenesis, in flowering plants that is analogous to that in mammals.

Results

Reprogramming of FLC expression during gametogenesis and embryogenesis

To explore the resetting of FLC, the spatiotemporal expression pattern of FLC::GUS was analyzed in gametophytes and developing embryos of FLC::GUS FRI flc-3 plants in the Columbia-0 (Col-0) ecotype background. In the FLC::GUS construct, the GUS gene was inserted in frame into an NheI site located in the sixth exon of a 16-kb genomic clone, spanning 5.4-kb upstream of the FLC start site and 5-kb downstream of the stop codon (Michaels et al., 2005). Before fertilization, FLC::GUS expression was detected in ovules, but not in stamens, of non-vernalized plants (Figure 1a,b). The region showing GUS staining in the ovule was restricted to the central cell of the embryo sac, and the part of the integument that originates from sporophytic maternal tissue. To examine whether the GUS signal in the ovule was a result of gametophytic expression per se, or whether it was of sporophytic origin, we generated FLC::GUS hemizygous plants by reciprocal crosses between FLC::GUS homozygotes and wild-type (WT) plants. Given that only half of the female gametophytes of the FLC::GUS hemizygous plants contained the transgene, if the GUS signal originated from the female gametophytes per se, it would be detected in half of the ovules. However, all of the ovules in the hemizygous plants exhibited a GUS signal (Figure S1), indicating that FLC::GUS expression in the ovules was derived from diploid maternal tissues, and not from gametophytic embryo sacs.

Figure 1.

 Reprogramming of FLC during gametogenesis and embryogenesis.
(a and b) FLC::GUS expression in female and male gametophytes of non-vernalized (NV) plants prior to fertilization. (c and d) FLC::GUS expression in female and male gametophytes of vernalized (V) plants prior to fertilization. (e–i) FLC::GUS expression in developing seeds of non-vernalized plants. (j–n) FLC::GUS expression in developing seeds of vernalized plants. (a and c) Ovules; (b and d) stamens; (e and j) early globular stage; (f and k) globular stage; (g and l) heart stage; (h and m) torpedo stage; (i and n) walking-stick stage. Plants used for GUS analysis are of the FLC::GUS FRI flc-3 genotype in the Col-0 background. (o) Endogenous expression of FLC mRNA in various tissues, including gametophytes and developing seeds, from non-vernalized and vernalized plants. RT-PCR analysis was performed with total RNA extracted from different stages and tissues of FRI FLC plants in the Col-0 background. To analyze FLC expression in developing seeds, seeds containing embryos at each stage were independently harvested for total RNA extraction. Actin2 (ACT2) was used as a quantitative expression control. Ovules include integuments and placentas. (p) Quantitative comparison of FLC expression between seeds from plants with or without vernalization treatment. Real-time qPCR was performed using the same cDNAs as in (o). The vertical axis represents the normalized relative expression of FLC in comparison with that of Tubulin2 (TUB2). See Experimental procedures for the details. Scale bars: ovules, 20 μm; stamens, 100 μm; seeds, 100 μm.

After fertilization, the GUS signal began to appear in embryos from the early globular stage, and was sustained throughout the rest of embryonic development (Figure 1e–i). A weak GUS signal in the endosperm was also detected immediately after fertilization (Figure 1e), but this is likely to have resulted from the residual expression of FLC::GUS in the maternal tissues of the ovule, because all of the seeds from FLC::GUS hemizygous plants displayed this expression pattern in the endosperm immediately after fertilization (data not shown).

To determine which parental allele of FLC contributes to expression in embryos, and to test for the possibility of imprinting, we introduced the FLC::GUS transgene uniparentally by reciprocal crosses between FLC::GUS and WT plants, and then compared the resulting GUS expression patterns (Figure 2). Paternally and maternally inherited FLC::GUS transgenes showed the same expression patterns in developing embryos. Thus, FLC is not imprinted, and both parental alleles contribute equally to expression in embryos.

Figure 2.

FLC::GUS expression in embryos produced by reciprocal crosses between wild-type (WT) and FLC::GUS plants.
(a–e) FLC::GUS expression in F1 seeds of FLC::GUS (♀) × WT (♂). (f–j) FLC::GUS expression in F1 seeds of WT (♀) × FLC::GUS (♂). (a and f) One day after pollination (DAP); (b and g) 3 DAP; (c and h) 5 DAP; (d and i) 7 DAP; (e and j) 9 DAP. Scale bars: 100 μm.

We also performed semi-quantitative reverse transcription-polymerase chain reaction (RT-PCR) analysis to examine whether the expression of endogenous FLC has the same pattern as that of the FLC::GUS transgene (Figure 1o). We observed ubiquitous expression of FLC in vegetative tissues, and its expression reached the highest levels in inflorescence meristem, including floral buds. Consistent with the GUS staining results, endogenous FLC mRNA was detected in ovules prior to fertilization, and was expressed throughout embryogenesis. However, the expression of FLC mRNA in unfertilized ovules should originate from maternal diploid cells, as described above. Taken together, these results indicate that FLC expression is repressed during gametogenesis, and is then reactivated after fertilization in embryos.

Reprogramming of the vernalization-induced silencing of FLC during reproductive development

To characterize the resetting of FLC after vernalization, we studied the expression pattern of FLC by RT-PCR, as well as by using the FLC::GUS transgene. FLC mRNA was not detected by RT-PCR in ovules or pollens of vernalized plants (Figure 1o), which is consistent with FLC::GUS expression (Figure 1c,d). Seeds from vernalized plants exhibited similar FLC expression pattern as those from non-vernalized plants throughout the embryonic stages, as analyzed by RT-PCR (Figure 1o) or by using the FLC::GUS transgene (Figure 1j–n). We observed a lower FLC mRNA level in seeds with globular-stage embryos after vernalization, which is presumably to the result of vernalization-induced repression of FLC in maternal tissues within the ovules. Consistent with this hypothesis, seeds from vernalized FLC::GUS plants did not show residual GUS expression at early embryonic stages (compare Figure 1j with 1e). When we used random decamer primers instead of an oligo-dT primer for RT, we observed a similar expression pattern of FLC mRNA throughout the embryonic stages in vernalized seeds (data not shown), indicating that FLC mRNA is not subject to poly(A)-tail-mediated stability control during embryogenesis. Real-time quantitative RT-PCR (real-time qPCR) was also employed to compare FLC expression between seeds from plants with or without vernalization (Figure 1p). Similar to the above results, globular stage seeds from vernalized plants exhibited a lower FLC expression level than those from non-vernalized plants. FLC expression in seeds was similarly increased in both samples after the globular stage. In summary, our results demonstrate that, regardless of the epigenetic state of maternal tissues, FLC expression is repressed in gametophytes and is then reactivated in embryos, but not in endosperms, after fertilization. This epigenetic reprogramming of FLC expression is similar to that of mammalian systems, in which epigenetic markers are erased and reset during reproductive development (Reik et al., 2001).

Expression of FLC activators during reproductive development

FLC is regulated by a number of factors in various floral regulatory pathways, such as FRI, vernalization and the autonomous pathway (Sheldon et al., 1999, 2000; Michaels and Amasino, 2001; Baurle and Dean, 2006). To evaluate whether these factors might also play roles in FLC reprogramming during reproductive development, we examined the mRNA expression patterns of several FLC activators, including FRI, EFS, PIE1 and ELF7, by RT-PCR. As shown in Figure 3a, these genes displayed ovule-specific expression prior to fertilization. After fertilization, these FLC activators were constitutively expressed from the globular to the mature embryonic stages. Vernalization had little effect on the expression of these genes during gametogenesis and embryogenesis (Figure 3a).

Figure 3.

 Expression of FLC activators during gametogenesis and embryogenesis.
(a) Endogenous mRNA expression of FLC activators. Total RNA was isolated as described in Figure 1o. ACT2 was used as a quantitative expression control.
(b) GUS expression in gametophytes and developing seeds of transgenic plants containing GUS fusions for each FLC activator. Either the promoters (PIE1 and EFS) or the coding region along with the promoter (FRI) were used to drive GUS expression. All transgenic plants except for PIE1::GUS (Ws) were in the Col-0 (fri FLC) genetic background. Left to right: ovule, stamen, walking-stick to mature embryo, and whole seed with endosperm and embryo. Scale bars: ovules, 20 μm; stamens, 100 μm; embryos, 100 μm; seeds, 100 μm.

The expression patterns of FRI, PIE1 and EFS were further analyzed using transgenic plants with GUS fusion constructs (Figure 3b). Like the FLC promoter, the FRI, PIE1 and EFS promoters also drove GUS expression in ovules, but not in stamens. The ovule-specific expression of these genes arose from maternal diploid cells, and not from gametophytic cells, because all of the ovules of hemizygous transgenic plants containing the GUS fusion constructs were stained, as they were for FLC::GUS hemizygous plants (data not shown). There was minor variation in the ovule-specific expression of these genes: FRIpro::FRI:GUS and EFS::GUS were expressed preferentially in the central region of ovule, whereas PIE1::GUS was expressed in the chalazal end of ovule, which is connected to the funiculus (Figure 3b). Despite these differences during gametogenesis, all of these genes exhibited embryo-specific promoter activity during embryogenesis. Taken together, these results suggest that genes acting as FLC activators in vegetative development are expressed similarly to FLC during embryogenesis.

Expression of FLC repressors during reproductive development

To gain insight into the potential roles of FLC repressors in the reprogramming of FLC, their expression patterns during gametogenesis and embryogenesis were analyzed. We first performed RT-PCR analysis of the expression of the autonomous-pathway genes FVE, FLD, LD, FCA, FY, FPA, HAC1, HAC12, FLK and REF6, either with or without vernalization treatment (Figure 4a). Expression of FVE and FLD mRNA was detected neither in ovules nor in pollen. After fertilization, FVE was strongly reactivated during embryogenesis, whereas FLD was expressed at very low levels in developing seeds. Expression of LD mRNA was low and ovule-specific prior to fertilization, but increased strongly in developing seeds. In addition to expression in ovules and seeds, FCA, FY, FPA, HAC1 and HAC12 showed weak but significant expressions in pollen. In contrast, FLK and REF6 did not show expression patterns that were different from FLC. We also analyzed the expression of the vernalization pathway component VIN3 (Figure 4a). Interestingly, VIN3 expression was weak in ovules and seeds, but strong in pollen grains. VIN3 mRNA was also present in the non-vernalized seeds. Because VIN3 expression is barely detectable in vegetative tissues without vernalization (Sung and Amasino, 2004), this result indicates the existence of a mechanism for VIN3 regulation in reproductive tissues, that differs from that in vegetative tissues.

Figure 4.

 Expression of FLC repressors during gametogenesis and embryogenesis.
(a) Endogenous mRNA expression of FLC repressors. Total RNA was isolated as described in Figure 1o. ACT2 was used as a quantitative expression control.
(b) GUS expression in gametophytes and developing seeds of transgenic plants containing GUS fusions of each FLC repressor. Either the promoters (HAC1, HAC5, HAC12 and VIN3) or the coding regions along with the promoters (REF6, VRN1 and VRN2) were used to drive GUS expression. All transgenic plants except for VRN1pro::VRN1:GUS and VRN2pro::VRN2:GUS (Ler) were in the Col-0 (fri FLC) genetic background. Left to right: ovule, stamen, walking-stick to mature embryo, whole seed with endosperm and embryo. Scale bars: ovules, 20 μm; stamens, 100 μm; embryos, 100 μm; seeds, 100 μm.

The expression levels and patterns of most of the autonomous-pathway genes examined above, during reproductive and embryonic development, were not affected by vernalization treatment during vegetative growth (Figure 4a). Interestingly, unlike other FLC repressors in the autonomous pathway, FLD and VIN3 expression levels in the seeds of vernalized plants were slightly higher and lower than those in the seeds of non-vernalized plants, respectively.

The expression patterns of REF6, HAC1, HAC5, HAC12, VRN1, VRN2 and VIN3 were further studied by histochemical GUS assays with transgenic plants harboring transcriptional or translational GUS fusion constructs (Figure 4b). Like FLC::GUS, REF6pro::REF6:GUS was specifically expressed in ovules and embryos. HAC1::GUS, HAC5::GUS and HAC12::GUS were expressed in pollen grains. These genes were also expressed in ovules, but the regions expressing GUS were somewhat different from each other. Whereas HAC1::GUS was expressed in the entire ovule, expression of HAC5::GUS or HAC12::GUS was more concentrated in regions containing the egg cell and the central cell. Interestingly, HAC5::GUS and HAC12::GUS, but not HAC1::GUS, were expressed not only in embryos but also in endosperms.

We also studied the expression pattern of some key FLC repressors acting in the vernalization pathway. The VRN1pro::VRN1:GUS and VRN2pro::VRN2:GUS constructs, which nearly completely rescue the vrn1 and vrn2 mutant phenotypes, respectively (CL and CD, unpublished data), showed expression patterns similar to FLC::GUS during gametogenesis and embryogenesis (Figure 4b). However, the pattern of VIN3 expression was different from those of FLC, VRN1 and VRN2. VIN3::GUS was expressed both in ovules and pollen prior to fertilization. In addition, unlike most of the FLC regulators tested in this study, which are expressed in the entire embryo, VIN3::GUS expression was restricted to the shoot apical meristem region of the embryo (Figure 4b). Taken together, our results suggest that the majority of FLC repressors have expression patterns similar to FLC with a few exceptions, namely, HAC1, HAC5, HAC12 and VIN3, which are also expressed in pollen and/or in the endosperm, where FLC is not expressed. These results suggest the possibility that repression of FLC in gametophytes and the endosperm might be mediated by the FLC repressors that are expressed in those tissues.

Reprogramming of FLC in the mutant backgrounds of FLC regulators

The expression analysis of FLC regulators suggested their potential roles in the reprogramming of FLC during gametogenesis and embryogenesis. To identify factors mediating FLC reprogramming, we studied the expression of FLC in mutants of various FLC regulators. First, we introduced the FLC::GUS transgene into fld, ld and fve mutants by genetic crosses, and analyzed the resulting GUS expression patterns (Figure 5a). The fld and ld mutations did not alter the expression pattern of FLC::GUS during gametogenesis and embryogenesis, indicating that the reprogramming of FLC is independent of the functions of FLD and LD. In contrast, FLC::GUS was expressed in the endosperm as well as in the embryo in fve mutants (Figure 5a). The ectopic expression of FLC in the endosperm was not apparent immediately after fertilization, but began to be detectable in seeds containing torpedo-stage embryos (Figure S2).

Figure 5.

FLC::GUS expression in mutants of FLC regulators.
(a) FLC::GUS expression in mutants of FLC repressors and FLC activators. Left to right: ovule, stamen, walking-stick to mature embryo, and whole seed with endosperm and embryo.
(b) FLC::GUS expression in developing embryos in the FRI and fri backgrounds. Left to right: heart, torpedo and mature stage embryos, and seedlings at 1 day after germination. Scale bars: ovules, 20 μm; stamens, 100 μm; embryos, 100 μm; seeds, 100 μm.

HAC1, HAC5 and HAC12 are expressed in pollen, and HAC5 and HAC12 are also expressed in the endosperm (Figure 4). Therefore, we tested whether FLC is ectopically expressed in pollen or the endosperm in single or double hac mutant backgrounds by RT-PCR analysis. FLC expression was higher in hac1 single and hac1 hac12 double mutants in open flowers than in controls (Figure S3a). However, there was no ectopic expression of FLC in these hac mutants in pollen. FLC expression was not increased in the pollen, endosperm or embryos of hac5 hac12 double mutants (Figure S3b).

Vernalization induces the transcriptional activation of VIN3 (Sung and Amasino, 2004), and increased VIN3 expression results in the epigenetic repression of FLC in vegetative tissues. Our results showing the strong expression of VIN3 in pollen grains (Figure 4a,b) also suggest a possible role for VIN3 in the repression of FLC in male gametophytes. However, in vin3 mutants, FLC::GUS expression was not detected in pollen, and the expression pattern of FLC during gametogenesis and embryogenesis was unchanged (Figure 5a). The results of FLC::GUS expression analysis in the hac and vin3 mutants indicate that these genes are not involved in FLC repression in pollen and the endosperm, nor are they involved in the reprogramming of FLC during gametogenesis and embryogenesis.

Because FRI, a transcriptional activator of FLC, is expressed in a similar pattern as FLC during gametogenesis and embryogenesis (Figure 3), we tested whether FRI functions in FLC reactivation upon fertilization. To achieve this, we generated FLC::GUS fri flc plants, and compared their GUS expression patterns with those of FLC::GUS FRI flc plants during embryogenesis (Figure 5b). Interestingly, the early stage embryos of the fri plants showed strong GUS expression, similar to GUS expression in the same stage embryos of FRI plants. However, after the late-torpedo embryonic stages, GUS expression in the fri plants decreased gradually, and eventually was almost fully repressed in mature embryos, with minor expression in the vasculature. In contrast, GUS was strongly expressed in the FRI plants until embryonic maturation, and was then maintained throughout germination (Figure 5b). Therefore, FRI might be dispensable for the reactivation of FLC in early embryogenesis, although it is required to maintain high levels of FLC expression in later embryonic and vegetative development. To further confirm the role of FRI in FLC reactivation during embryogenesis, we also analyzed the expression of FLC::GUS in the suf4 mutant background. A loss of SUF4 activity has been reported to cause the decreased expression of FLC, as observed in fri plants, and the SUF4 protein has been reported to physically interact with FRI, and might recruit FRI to the FLC promoter (Kim et al., 2006). Consistent with our results in fri plants, FLC reactivation in early-stage embryogenesis was not affected by the suf4 mutation (Figure 5a). In summary, our results from the fri and suf4 plants demonstrate that FRI and SUF4 are not required for FLC reactivation, but are required for the maintenance of high levels of FLC expression in late embryogenesis and vegetative development.

Exchange of the histone variant H2A.Z with H2A has been proposed to play a critical role in epigenetic reprogramming in animals (Hajkova et al., 2008). In Arabidopsis, a yeast SWR1-like PIE1-containing complex is involved in the H2A to H2A.Z exchange, and is required for the full activation of FLC in vegetative tissues (Deal et al., 2007; Choi et al., 2007; March-Diaz et al., 2008). To test whether the PIE1 complex is also required for the reprogramming of FLC, FLC::GUS expression was studied in pie1 mutants (Figure 5a). FLC::GUS was not expressed in the ovules and pollens of pie1: this might result from the suppression of FLC in diploid maternal tissues of pie1, as has been reported previously (Noh and Amasino, 2003). After fertilization, globular-stage embryos did not exhibit a detectable GUS signal (Figure S2e). Torpedo-stage embryos of the pie1 mutant exhibited a weak GUS staining only in the basal region (Figure 5a). FLC::GUS was expressed strongly in root, but was expressed weakly in the shoot apex and vasculature of hypocotyl and cotyledons of fully matured pie1 embryos. Therefore, these results indicate that the PIE1 complex is not relevant to the repression of FLC in gametophytes, but instead plays a pivotal role in the reactivation of FLC in early embryos, as well as in the maintenance of full activation of FLC in late embryos.

Polycomb group (PcG) complexes repress various sets of genes (Pien and Grossniklaus, 2007). In Arabidopsis, the vernalization-induced repression of FLC expression is maintained by the VRN2 PcG complex during vegetative development (Gendall et al., 2001). Another PcG complex, the MEDEA (MEA)-Fertilization Independent Endosperm (FIE) PcG complex, acts in the endosperm, and represses genes that might cause endosperm overproliferation (Grossniklaus et al., 1998; Kinoshita et al., 1999; Kiyosue et al., 1999; Kohler et al., 2003, 2005; Gehring et al., 2006). A recent study revealed that FIE also interacts with the VRN2 PcG complex, and is required for vernalization responses (Wood et al., 2006). Hence, we hypothesized that the repression of FLC in the endosperm might be mediated by the MEA-FIE PcG complex. If the MEA-FIE PcG complex is responsible for the repression of FLC in the endosperm, we should be able to detect FLC::GUS expression in mea and fie mutant seeds. As discussed earlier, FLC is capable of being expressed in the endosperm, as shown by the fve mutant results in Figure 5a. Seeds of mea (Figure 6a,b) and fie mutants (Figure 6d,e) showed arrested embryos and an enlarged endosperm phenotype. However, the pattern of FLC::GUS expression was not altered by these PcG mutations during embryogenesis. We also examined FLC::GUS expression in emasculated fie mutant ovules (Figure 6c). We emasculated fie heterozygous flowers to determine whether FLC::GUS was derepressed in fie mutant ovules, because the fie mutation is embryonic lethal. As the fie mutation allows the division of diploid central cells without fertilization, we could easily distinguish fie mutant ovules from WT ovules in emasculated fie heterozygous plants. Again, we could not detect the derepression of FLC::GUS expression in the fie seed-like structures. We also examined the expression of FLC mRNA in the pollen of fie heterozygous plants, but were unable to detect expression by RT-PCR (Figure S3c). Therefore, the above data indicate that the repression of FLC in pollen, and in the endosperm of developing seeds, is not mediated by the MEA-FIE PcG complex, and should be regulated by a mechanism that is distinct from the one acting in vernalization in vegetative tissues.

Figure 6.

FLC::GUS expression in mea and fie mutants.
(a and b) FLC::GUS expression in mea-3 mutant seeds. (a) Seven days after pollination (DAP); (b) 9 DAP; (c–e) FLC::GUS expression in fie-1 heterozygous mutants. (c) Ovules at 6 days after emasculation. Seed-like structures containing 2n central-cell replication without embryos can be seen in fie mutant ovules (arrows); (d) 7 DAP; (e) 9 DAP. Scale bars: seeds, 100 μm; pistil, 50 μm.

Discussion

FLC expression during gametogenesis

Before fertilization, FLC expression was detected in ovules, but not in the pollen of non-vernalized plants (Figure 1a,b). Using a genetic test, we demonstrated that FLC expression in the ovule originates from the diploid maternal tissues that enclose female gametophytes (Figure S1). The expression pattern of FLC after vernalization treatment further supports this conclusion: when vernalization suppressed FLC expression, we could not detect any FLC expression in ovules (Figure 1c,d). Therefore, we conclude that FLC expression is fully repressed before gamete formation.

Recently, Sheldon et al. (2008) reported that FLC is reactivated temporarily in the developing somatic and sporogenous tissues of anthers, but is re-repressed in mature anthers. This temporary reactivation was observed using two independent FLC::GUS transgenic lines in either C24 or Ler backgrounds. In this study, we did not observe the temporary reactivation of FLC::GUS throughout anther development in the Col-0 background (Figure 1 and data not shown). However, similar to the observation made by Sheldon et al. (2008), we observed FLC::GUS expression in the pollen sacs of the hybrid progeny of crosses between Col and Ler plants (data not shown). The expression was restricted to somatic tissues such as tapeta, and was not observed in pollen grains. Therefore, the anther-specific temporary reactivation of FLC seems to vary depending on genetic background. In both studies, FLC expression is fully repressed in mature male gametophytes, as well as in female gametophytes.

Biological roles of autonomous-pathway genes in FLC resetting

The autonomous pathway represses FLC expression in the vegetative tissues of many summer-annual Arabidopsis accessions (Sheldon et al., 1999, 2000; Michaels and Amasino, 2001). However, the repression of FLC by the autonomous pathway is fully suppressed by the transcriptional activating role of a functional FRI allele in winter-annual accessions (Johanson et al., 2000). Because many summer-annual Arabidopsis accessions have arisen from loss-of-function mutations in FRI (Johanson et al., 2000), the ancestral genetic composition of Arabidopsis should contain functional FRI alleles. Hence, the function of autonomous-pathway members in the vegetative tissues of FRI-containing ancestral or winter-annual genetic backgrounds is likely to balance the effects of FRI, to achieve a proper level of FLC expression. Because we observed the repression of FLC expression in gametophytes and the endosperm, we tested the possible repressive role of autonomous-pathway members on FLC in those reproductive tissues. We also observed that many of the autonomous-pathway genes are expressed in patterns similar to FLC in reproductive tissues, with a few exceptions (HAC1, HAC5 and HAC12) that are also expressed at significant levels in pollen and the endosperm, where FLC is not expressed (Figure 4). However, our tests using several autonomous-pathway mutants revealed that none of these are involved in the repression of FLC in gametophytes and the endosperm (Figures S3 and 5), although we could not rule out the possibility that other autonomous-pathway genes, such as FCA, FY, FPA, FLK and REF6, that were not tested in this study might also repress FLC.

A recent study reported defective seed production in fca fpa double mutants (Baurle et al., 2007), suggesting that FCA and FPA might also function in reproductive development. In this study, we show that the repression of FLC in the endosperm is mediated, at least in part, by FVE (Figures S2 and 5a). FVE was also reported to regulate cold responses (Kim et al., 2004). Therefore, autonomous-pathway members might have multiple roles in various aspects of Arabidopsis development, as well as in FLC regulation, although the details of these roles have yet to be elucidated.

Roles of FRI in the reactivation of FLC during embryogenesis

When a dominant-active allele of FRI exists, FLC expression is activated even in the presence of functional autonomous-pathway repressors, and the plant shows a late-flowering phenotype. RT-PCR and the GUS fusion analyses in this study revealed that like FLC (Figure 1), FRI is expressed in the ovule and the embryo (Figure 3). However, the results presented in Figure 5 clearly show that an active FRI allele is dispensable for the initial reactivation of FLC in the embryo. We have also found that SUF4, an interacting partner of FRI (Kim et al., 2006), has no role in FLC reactivation (Figure 5). These results, taken together, indicate that FRI and SUF4 are required for the activation of FLC after late embryogenesis, but not during early embryogenesis. Accordingly, FLC resetting in reproductive tissues should be initiated by a different mechanism from that which regulates FLC in vegetative tissues. It is possible that some of the factors isolated as FLC activators might be responsible for the initiation of FLC reactivation during early embryogenesis. To address this possibility, FLC expression must be analyzed during embryogenesis in mutant backgrounds of various FLC activators. Interestingly, the FLC transcript was not detected in atx1-1 mutant embryos in a recent study (Pien et al., 2008). Therefore, ARABIDOPSIS TRITHORAX 1 (ATX1) might be the factor required for FLC reactivation. ATX1 directly interacts with FLC chromatin, and is required for trimethylating H3K4 in the FLC locus of rapidly flowering accessions. These results indicate a FRI-independent function of ATX1 in FLC activation. Therefore, it is worthwhile to further address the functional relationship between FLC resetting and ATX1, or other FLC activators.

On the other hand, the high level of FLC expression in the early-stage embryos of fri plants was no longer maintained in late embryogenesis (Figure 5). As the expression of FLC::GUS remained high until late embryogenesis in autonomous-pathway mutants such as fld and ld (Figure 5), the repression of FLC expression in mature embryos of fri plants might be mediated by autonomous-pathway genes. The fact that high FLC::GUS expression is maintained in the mature embryos of FRI plants (Figure 5) means that the hierarchy between autonomous-pathway genes and FRI also exists in late embryogenesis. Taken together, our results support the idea that FRI and autonomous-pathway genes determine the transcriptional activity of FLC during late embryogenesis, and this is important for the initial establishment of flowering competence.

Possible mechanisms for the reprogramming of FLC during reproductive development

Based on our tests of the role of FLC repressors and the MEA-FIE PcG complex in silencing FLC in gametophytes, none of these repressors or the components of the PcG complex were responsible for the repression of FLC in gametophytes (Figures 5 and 6). Although we cannot exclude the possibility of activity by other FLC repressors not tested in this study, it is possible that the silencing of FLC in gametophytes is established by the canonical process of epigenetic reprogramming, rather than by specific FLC repressors.

A dynamic exchange of histone proteins in mouse germ cells was reported recently (Hajkova et al., 2008). The authors demonstrated that the dynamic exchange of histone H2A with its variant H2A.Z occurs before the production of totipotent germ cells, and suggested the importance of this exchange in the erasure of epigenetic modifications that are pivotal for genomic reprogramming. In Arabidopsis, a yeast SWR1-like PIE1-containing complex mediates the exchange of H2A with H2A.Z at the FLC locus in vegetative tissues (Choi et al., 2007). Our results in Figure 5a demonstrate that the PIE1 complex is not required for the repression of FLC in gametophytes, but is essential for the reactivation and maintenance of FLC expression in early and late embryogenesis, respectively. Therefore, the H2A to H2A.Z exchange is likely to play a critical role in epigenetic reprogramming in Arabidopsis.

Changes in genome-wide DNA methylation are involved in genomic reprogramming in mammals (Reik et al., 2001). Interestingly, FLC transcription is low in the vegetative tissues of hypomethylated Arabidopsis mutants, such as ddm1 or antisense MET1 transgenic plants (Jean Finnegan et al., 2005). Although changes in FLC expression in both mutants were suggested as an indirect effect of the changes in genomic DNA methylation, the relationship between FLC resetting and DNA methylation during reproductive development has yet to be determined. In this study, we have addressed the reprogramming of FLC and the expression patterns of a number of its regulators during gametogenesis and embryogenesis. The model in Figure 7 summarizes the three phases of FLC reprogramming, and the role of some FLC regulators during the process, as revealed from our study. These results provide new insights into FLC reprogramming and the mechanisms for epigenetic reprogramming in general, in flowering plants.

Figure 7.

 Schematic model for FLC reprogramming during reproductive development.
The process of FLC reprogramming during reproductive development might be divided into three phases: (i) a phase of repression in gametogenesis, (ii) a phase of reactivation in early embryogenesis, and (iii) a phase of maintenance in late embryogenesis. Regardless of the epigenetic state in adult plants, FLC expression is repressed in male and female gametophytes by unknown mechanisms. Then, FLC expression is reactivated after fertilization in embryos, but not in the endosperm. The present study demonstrates that the reactivation process is independent of FRI and SUF4, but requires PIE1 function. Although Pien et al. (2008) reported the failure of FLC reactivation in early-globular stage embryos of atx1 mutants through an in situ hybridization study, the role of ATX1 in FLC reactivation may need to be re-evaluated in an FRI-containing genetic background. At stages around maturation, the level of FLC expression in embryos is determined by genetic interactions between FRI and autonomous pathway (AP) genes. PIE1 is also necessary for the full activation of FLC at these stages, as well as for the initial reactivation. Interestingly, the repression of FLC in endosperm is mediated, at least in part, by an autonomous pathway gene FVE. The FLC expression in non-vernalized (red line) or vernalized (blue line) FRI-containing WT plants is depicted. Maternally transmitted FLC expression in ovules or early stage seeds is not considered in this model, to avoid misconception.

Experimental procedures

Plant materials and growth conditions

All plants used in this study are in the Col-0 background, except for vrn1, vrn2, mea-3 and fie-1 (Ler background) or pie1-1 (Ws background). Seeds were stratified on 0.65% phytoagar containing half-strength MS (Plantmedia, http://www.plantmedia.com) salts for 3 days at 4°C. All plants were grown in long-day photoperiodic conditions (16-h light/8-h dark) under cool, white fluorescence light (100 μmole m−2 s−1) at 22°C, with 60% relative humidity. For vernalization, seedlings germinated on MS plates were incubated for 4 weeks at 4°C under short-day conditions (8-h light/16-h dark). Afterwards, vernalized seedlings were further grown on soil with long days.

Generation ofFRI::GUS, VRN1pro::VRN1:GUSandVRN2pro::VRN2:GUSplants

The procedures used to generate these GUS fusion constructs and transgenic plants are described in Appendix S1.

Histochemical GUS imaging

The GUS activity in gametophytes and seeds was analyzed by incubation in 50 mm NaPO4 (pH 7.0), 1 mm X-Gluc (Sigma-Aldrich, http://www.sigmaaldrich.com), 10 mm K3Fe(CN)6, 10 mm K4Fe(CN)6, 10 mm EDTA and 0.2% Triton X-100 at 37°C for 8–10 h. After staining, tissues were cleared by incubation in 70% EtOH for several hours. Stained tissues were photographed using an Axio Imager A1 microscope (Carl Zeiss, http://www.zeiss.com), with an AxioCam HRc camera.

Analysis of gene expression

Methods for pollen collection were described previously by Choi et al. (2002). To analyze gene expression in ovules, pistils were dissected before fertilization, and exposed ovules and placentas were harvested for total RNA extraction. Total RNA extraction from developing seeds was performed using seeds independently harvested according to the stage of embryo development.

Total RNA from various vegetative tissues and gametophytes were extracted with Trizol reagent (Invitrogen, http://www.invitrogen.com) following the manufacturer’s protocol. For RNA extraction from seeds, the Ambion RNAqueous™ Kit (Ambion, http://www.ambion.com) was used. A total of 1–2 μg of RNA was reverse transcribed using M-MLV reverse transcriptase (Ambion) and an oligo-dT primer after RNase-free DNase treatment (TaKaRa Bio, http://www.takara-bio.com). PCR amplification was performed using gene-specific primers (see Table S1 for primers).

Real-time qPCR

Real-time qPCR was performed on 96-well optical reaction plates (Bio-Rad, http://www.bio-rad.com). All PCR mixtures contained 10 μl of iQ™ SYBR green Supermix (Bio-Rad), 0.5 μl of forward primer (10 μm), 0.5 μl of reverse primer (10 μm) and 5 μl of each diluted RT product per well. PCR amplification of the TUB2 housekeeping gene was performed as a control for sample loading, and for normalization. Negative controls were treated the same way as the samples, but without reverse transcriptase. All of the templates were run in triplicate, and the threshold cycle (Ct) was determined using iQ™optical system Software (Bio-Rad). Gene-specific transcripts were quantified using the ddCt method (ddCt = Cgene of interest – CTUB2). Real-time SYBR-green dissociation curves showed one species of amplicon for each primer combination.

Acknowledgements

We are grateful to S. Michaels and I. Lee for providing FLC::GUS and suf4 plants, respectively. This work was supported by grants from the Brain Korea 21 project to JC and M-JK, and by the Korea Research Foundation funded by the Korean Government (MOEHRD, Basic Research Promotion Fund; KRF-2005-070-C00129) to YH and YC, and (KRF-2006-312-C00672) to YC and BN. BN was also supported by a grant from the MEST/KOSEF to the Environmental Biotechnology National Core Research Center (R15-2003-012-01001-0). Work in Y-SN’s lab was supported by the GRL Program from the MEST/KICOS, by the Plant Diversity Research Center and by the BioGreen21 Program from the RDA. Work in RMA’s laboratory was supported by the College of Agricultural and Life Sciences and the Graduate School of the University of Wisconsin, National Institutes of Health Grant 1R01GM079525, National Science Foundation Grant 0209786, and by the GRL Program from the MEST/KICOS.

Ancillary