To investigate the mechanisms regulating the initiation of floral development in Arabidopsis, a construct containing β-glucuronidase (GUS) gene driven by APETALA1 promoter (AP1::GUS) was introduced into emf fwa and emf ft double mutants. GUS activity was strongly detected on shoot meristem of emf1-1 single mutants harboring AP1::GUS construct just 5 d after germination. By contrast, GUS activity was undetectable on emf1-1 fwa-1, emf1-1 ft-1, emf2-1 fwa-1, emf2-3 fwa-1 and emf2-3 ft-1 double mutants harboring AP1::GUS construct 10 d after germination. GUS activity was only weakly detected on the apical meristem of 20-day-old emf1-1 fwa-1 and emf2-1 fwa-1 seedlings. During this time, only sessile leaves were produced. Further analysis indicated that AP1 was strongly expressed in 10-day-old emf1-1 and emf2-1 single mutants. Its expression was significantly reduced in all emf1-1 or emf2-1 late-flowering double mutants tested. Similar to AP1, the expression of LEAFY (LFY) was also high in emf1-1 and emf2-1 single mutants and reduced in emf1-1 or emf2-1 late-flowering double mutants. Our results indicate that the precocious expression of AP1 and LFY is dependent not only on the low EMF and FWA activities but also on the expression of most of the late-flowering genes such as FT, FCA, FE, CO and GI. These data also reveal that most late-flowering genes may function downstream of EMF or in pathways distinct from EMF to activate genes specified floral meristem identity during shoot maturation in Arabidopsis.
(Received September 25, 2000; Accepted February 24, 2001).
In Arabidopsis thaliana, flowers start to initiate during late inflorescence development after production of the cauline leaves and coflorescence branches. Genes such as TERMINALFLOWER1 (TFL1), LEAFY (LFY), APETALA1 (AP1) and CAULIFLOWER (CAL) have been identified to be involved in the regulation of this floral initiation process (Shannon and Meeks-Wagner 1991, Shannon and Meeks-Wagner 1993, Bradley et al. 1997, Weigel et al. 1992, Huala and Sussex 1992, Irish and Sussex 1990, Schultz and Haughn 1993, Bowman et al. 1993, Gustafson-Brown et al. 1994, Liljegren et al. 1999).
In addition to these meristem identity genes, late-flowering genes such as FVE, CONSTANS (CO), FPA, FT, FWA, FCA, FLD and LD also affect inflorescence and flower development based on a number of molecular and genetic analyses (Martinez-Zapater et al. 1995, Putterill et al. 1995, Simon et al. 1996, Madueno et al. 1996, Ruiz-Garcia et al. 1997, Piñeiro and Coupland 1998, Chou and Yang 1998, Nilsson et al. 1998, Page et al. 1999, Yang and Chou 1999, Aukerman et al. 1999). These late-flowering mutations all cause an enhancement of ap1 or lfy phenotype in late-flowering ap1/lfy double mutants although a much stronger enhancement was caused by ft, fwa and fld. The expression of CO has been reported to activate the transcription of LFY and AP1 (Simon et al. 1996). The presence of FT activity has been shown to be required for the AP1 expression in lfy mutant background (Ruiz-Garcia et al. 1997). FCA and FVE have been proposed to be involved in LFY transcriptional activation whereas FT, FE, and FD were most likely to act in response to LFY activity (Page et al. 1999, Nilsson et al. 1998). Part of the LD function has been thought to participate in the regulation of LFY (Aukerman et al. 1999). Therefore it has been proposed that most late-flowering genes is likely to have a role in the activation of genes that specify flower meristem identity such as LFY, AP1, and CAL to promote the inflorescence-to-flower transition. Different from other late-flowering mutants, it has been reported recently that dominant mutation fwa was caused by the expression of the gene that is normally inactive in wild-type plant (Koornneef et al. 1998, Levy and Dean 1998, Soppe et al. 2000). Therefore, the FWA gene should act as a repressor of flowering and has been thought to act in parallel with LFY during flower initiation (Ruiz-Garcia et al. 1997, Nilsson et al. 1998).
emf mutants are the most extreme examples of early-flowering mutants. Mutations in EMF1 or EMF2 genes caused the direct production of a small inflorescence shoot that contained 4–5 sessile leaves and 1–3 flowers with incomplete floral organ development in the absence of rosette growth (Sung et al. 1992, Yang et al. 1995). Double mutant analysis between emf1, emf2 and tfl1, lfy, ap1, ap2 support the hypothesis that EMF genes are not only regulating the rosette-to-inflorescence transition but also involving in inflorescence and flower development (Yang et al. 1995, Chen et al. 1997). It has been suggested that changing levels of EMF activity during shoot maturation activates floral initiation genes and initiates various developmental phase transitions in Arabidopsis (Yang et al. 1995, Chen et al. 1997). The fact that AP1 promoter is activated in emf1 seedling indicated that AP1 was expressed in the low or absence of EMF1 activity (Chen et al. 1997). Therefore, AP1 is normally suppressed by high EMF1 activity in wild-type plants before floral initiation.
Since the mutant phenotypes are completely opposite between emf and late-flowering mutants, late-flowering genes have been thought to be antagonistic partners of EMF in the regulation of flowering time and floral initiation. Double mutant analysis has indicated that emf are epistatic to some but not to all late-flowering mutations in the regulation of floral initiation (Haung and Yang 1998). This suggests that different late-flowering genes should function in parallel pathways to interact with EMF, directly or indirectly, during shoot development.
In order to further specifically determine the interaction of EMF and late-flowering genes in regulating floral meristem identity genes, the expression of AP1 and LFY was analyzed in double mutants between emf and various late-flowering mutants by quantitative reverse transcription (RT)-PCR and promoter assay during different developmental stages. Our data clearly indicates that floral meristem identity genes AP1 and LFY are not only activated by the low EMF activity but also require the presence of the activity from most of late-flowering genes. Furthermore, late-flowering genes tested here are not likely functioning in the upstream of EMF in regulating shoot maturation in Arabidopsis.
Materials and Methods
Late-flowering mutant lines (fwa-1, ft-1, fca-1, fe-1, co-1 and gi-1) used in this study were obtained from the Arabidopsis Biological Resource Center, Ohio State University, Columbus, OH, U.S.A. One emf1 mutant, emf1-1, and two emf2 mutants, emf2-1 and emf2-3 used in this research were isolated after ethyl methanesulfonate (EMS) or r-ray mutagenesis as described previously (Sung et al. 1992, Yang et al. 1995). Transgenic plants harboring a chimeric construct of AP1 promoter fused to a β-glucuronidase gene (AP1::GUS) were kindly supplied by Dr. M. Yanofsky (University of California at San Diego, La Jolla, CA, U.S.A.). emf1-1 single mutants contained AP1::GUS were kindly supplied by Dr. R. Sung (University of California at Berkeley, Berkeley, CA, U.S.A.). Double mutants between emf and various late-flowering mutants were generated as described previously (Haung and Yang 1998).
Plant growth conditions
Seeds were sterilized and plated on agar plates containing 1/2 MS medium (Murashige and Skoog 1962), kept at 4°C for 2 d, and then germinated in growth chambers at 22°C under long-day conditions (16 h light/8 h dark) for 10 d before being transplanted to the greenhouses. The light intensity of the growth chambers was 150 µE m–2 s–1. The greenhouses were maintained at 22°C with 16 h of light for long-day conditions.
To introduce AP1::GUS gene into emf fwa and emf ft double mutant background, plants homozygous for AP1::GUS were selected to cross with plants homozygous for fwa or ft and heterozygous for emf. F1 plants heterozygous for the emf locus were self-pollinated and used to generate F2 plants. F2 plants homozygous for late-flowering mutant allele and AP1::GUS and heterozygous for emf allele were used to generate F3 plants carrying the double mutations and AP1::GUS (one-fourth in F3 progeny).
Whole seedling at various days after germination was subjected for GUS assay. Histochemical staining was performed under standard method described previously (Jefferson et al. 1987, Chen et al. 1997). The sample was incubated in 2 mM 5-bromo-4-chloro-3-indolyl β-d-glucuronide (X-Gluc) solution (0.05 mM potassium ferricyanide, 0.05 mM potassium ferrocyanide, 100 mM phosphate buffer, pH 7.0) for several hours at 37°C. The sample was examined under a dissecting microscope.
Ten mg of fresh tissue from whole seedlings of 10-day-old wild-type Columbia, emf single mutants or emf late-flowering double mutants was homogenized with 0.1 ml homogenization solution (ULTRASPEC RNA isolation system, BIOTECX Company, Houston, TX, U.S.A.). During this stage, emf single mutants and emf late-flowering double mutants were phenotypically indistinguishable by producing only few sessile leaves. The homogenate was stored at 4°C for 5 min and 0.2 ml of chloroform per 1 ml homogenate was added. The samples were covered tightly and shaken vigorously for 15 s and kept on ice for 5 min. After centrifugation at 12,000×g (4°C) for 15 min, an equal volume of isopropanol was added to the aqueous phase and samples were stored for 10 min at 4°C. RNA was precipitated by centrifugation for 10 min at 12,000×g (4°C), washed twice with 75% ethanol and centrifuged 5 min at 7,500×g (4°C). RNA was dried under vacuum and resuspended in diethylpyrocarbonate (DEPC)-treated water.
RT-PCR and DNA gel blot analysis
For cDNA synthesis, total RNA (1 µg) was reverse-transcribed in a 20 µl reaction mixture using the BcaBESTTM RNA PCR system (TaKaRa Shuzo, Shiga, Japan). Five µl of cDNA sample from RT reaction was used for a 30 cycles of PCR as follows: denaturation at 94°C (45 s), annealing at 60°C (45 s), and extension at 72°C (90 s). Final 5 min at 72°C was performed as extension. Total PCR product (25 µl) in each reaction was analyzed by electrophoresis in 1.5% agarose gels and transferred to Hybond N+ membranes (Amersham International, Buckinghamshire, U.K.). For highly stringent hybridization, the membranes were prehybridized for 30 min and hybridized with 32P-labeled DNA probes overnight at 65°C in the same solution (0.25 M Na2HPO4, pH 7.2, 7% SDS), and then washed twice each in solution I (20 mM Na2HPO4, pH 7.2, 5% SDS) and solution II (20 mM Na2HPO4, pH 7.2, 1% SDS) at 65°C for 30 min per wash. The blots were then air dried, covered with plastic wrap, and autoradiographed. Primers specific for AP1, LFY and UBIQUITIN 5 (UBQ5) used in RT-PCR and in the generation of DNA probes were listed below. Primers for AP1: Atap1-3A (5′-GCTCCAAAAAAAGGAGAAGGC-3′) and Atap1-3B (5′-GCCAAAATATATTAATTGGATGAAA-3′). Primers for LFY: Atley-1A (5′-TCATTTGCTACTCTCCGCCGCT-3′) and Atley-1B (5′-CATTTTTCGCCACGGTCTTTAG-3′). Primers for UBQ5: UBQ5-1 (5′-GTGGTGCTAAGAAGAGGAAGA-3′) and UBQ5-2 (5′-TCAAGCTTCAACTCCTTCTTT-3′). DNA fragments in 413 bp, 465 bp and 300 bp were amplified from RT-PCR for AP1, LFY and UBQ5 respectively.
AP1::GUS expression in wild-type plants and emf1-1 mutants
In wild-type Arabidopsis, AP1 is undetectable in both vegetative leaves and inflorescence meristem. AP1 is first expressed during the formation of the floral primordia and its expression extends to the developing sepals and petals during early flower development (Mandel et al. 1992). AP1::GUS wild-type plant developed rosette leaves normally (Fig. 1A, left) whereas AP1::GUSemf1-1 mutant produced sessile leaves and reproductive organs 10 d after germination (Fig. 1A, right). In this 10-day-old wild-type Arabidopsis plant harboring AP1::GUS, no GUS activity was observed in either leaves or shoot apical meristem as expected (Fig. 1B, C). Later on, GUS was strongly detected in the young floral buds produced from both primary and secondary inflorescences (Fig. 1D). The wild-type plant containing no AP1::GUS was GUS-negative in floral buds during the same stage (Fig. 1E). This result indicated that GUS was correctly regulated by AP1 promoter and was detected in the location AP1 is normally expressed. In contrast to AP1::GUS wild-type plant, GUS activity was detected in both shoot apical meristem and the primordia for sessile leaves in AP1::GUSemf1-1 seedling just 5 d after germination (Fig. 1F). The GUS activity in leaves disappeared when they expanded in a 10-day-old seedling (Fig. 1G). At this stage, shoot meristem and young sessile leaves were still stained GUS positively (Fig. 1G). This result revealed that AP1 was ectopically expressed in emf1-1 mutant seedlings. We therefore concluded that the inability of AP1 expression in wild-type seedlings was due to the suppression by high level of EMF1 activity.
AP1::GUS expression in emf1-1 fwa-1 and emf1-1 ft-1 double mutants
Double mutants between emf1-1 and different late-flowering mutations have been constructed and analyzed previously (Haung and Yang 1998). The result indicated that different late-flowering mutations caused different effects on emf1-1 mutant background (Haung and Yang 1998). Among those double mutants, emf1-1 fwa-1 double mutants produced significantly more sessile leaves without further inflorescence or flower organs formation whereas severe alteration of the flower formation was observed in emf1-1 ft-1 double mutants (Haung and Yang 1998). Since the early appearance of inflorescence and flower structures in emf1-1 mutants was greatly influenced by the presence of the late-flowering mutations fwa-1 and ft-1, the investigation of the expression of AP1 in emf1-1 fwa-1 and emf1-1 ft-1 is interesting.
For this purpose, AP1::GUS was introduced into emf1-1 fwa-1 and emf1-1 ft-1 double mutants and GUS activity was analyzed. In contrast to emf1-1AP1::GUS mutant plant (Fig. 1F, G), GUS activity was undetectable in either sessile leaves or shoot apical meristem in both 10-day-old emf1-1 fwa-1 (Fig. 1H) and emf1-1 ft-1 double mutant seedlings (Fig. 1I). Interestingly, GUS was stained positively on the tips of the cotyledons in emf1-1 fwa-1 double mutant seedling (Fig. 1H). This was not observed in either emf1-1AP1::GUS single mutants or emf1-1 ft-1AP1::GUS double mutants. Later on, weak GUS activity was detected in the primordia of young sessile leaves in 20-day-old emf1-1 fwa-1AP1::GUS double mutants (Fig. 1J). No GUS activity was observed in either expanded sessile leaves or cotyledons in this emf1-1 fwa-1 double mutant seedling (Fig. 1J). In 20-day-old emf1-1 ft-1AP1::GUS double mutants, GUS was undetectable in whole seedling (Fig. 1K). This result clearly indicated that the expression of AP1 was either delayed or reduced in emf1-1 fwa-1 and emf1-1 ft-1 double mutants. Since the dominant fwa mutation was caused by the expression of FWA gene (Soppe et al. 2000), the proper AP1 expression therefore relies on the presence of late-flowering gene FT and absence of late-flowering gene FWA.
AP1::GUS expression in emf2-1 fwa-1, emf2-3 fwa-1 and emf2-3 ft-1 double mutants
When double mutants between emf2 and different late-flowering mutations were analyzed, fwa-1 and ft-1 caused the production of more sessile leaves and the formation of sepal-leaf flower structures in double mutants (Haung and Yang 1998). To examine whether the expression of AP1 in emf2 fwa-1 and emf2 ft-1 double mutants was also influenced by the presence of the late-flowering mutations fwa-1 and ft-1, AP1::GUS was introduced into emf2-1 fwa-1, emf2-3 fwa-1 and emf2-3 ft-1 double mutants and GUS activity was analyzed.
Similar to the result observed in emf1-1 fwa-1 double mutants, GUS activity was undetectable in either sessile leaves or shoot apical meristem on 10-day-old emf2-1 fwa-1AP1::GUS seedling (Fig. 1L). Weak GUS staining was observed in the shoot apical meristem in 20-day-old emf2-1 fwa-1AP1::GUS double mutants (Fig. 1M, N). When GUS activity was examined in a 20-day-old emf2-3 fwa-1AP1::GUS double mutant, no GUS activity was detected in shoot apical meristem in expanded sessile leaves or cotyledons (Fig. 1O). The difference observed between 20-day-old emf2-1 fwa-1AP1::GUS and emf2-3 fwa-1AP1::GUS double mutants may reflect the leaky nature of emf2-3 mutant allele in which the residual activity of EMF2 prohibited or delayed the expression of AP1. A similar result was obtained in 20-day-old emf2-3 ft-1AP1::GUS double mutants in which GUS was also undetectable in whole seedling (Fig. 1P). Similar to those observed in emf1-1 fwa or emf1-1ft double mutants, the results obtained here supported the notion that the expression of AP1 was activated by late-flowering gene FT and was suppressed by late-flowering gene FWA.
AP1 expression was reduced in emf1-1 late-flowering double mutants
To further confirm the results obtained by AP1::GUS transgenic plants, a combination of RT-PCR and Southern analysis was performed to detect the AP1 expression in emf1-1 late-flowering double mutants. As shown in Fig. 2, strong signal for AP1 was observed in emf1-1 single mutants 10 d after germination. This was correlated with the strong GUS activity observed in emf1-1AP1::GUS seedlings (Fig. 1F, G). Since different late-flowering mutations caused different effects on emf1-1 mutant background (Haung and Yang 1998), we therefore examined the AP1 expression in the six representative emf1-1 late-flowering (fwa-1, ft-1, fca-1, fe-1, co-1 and gi-1) double mutants. The result indicated that the AP1 expression was detected in all six emf1-1 late-flowering double mutants 10 d after germination (Fig. 2A, B). The level of AP1 signal was however reduced significantly in these six double mutants (about 10 to 20% strength of that observed in emf1-1 single mutant seedlings) (Fig. 2B). In 10-day-old wild-type Columbia seedling, no AP1 signal was detected (Fig. 2A, B). This result clearly indicated that AP1 was expressed earlier and more abundantly in emf1-1 plants than in emf1-1 late-flowering double mutant plants. Different from the negative staining of GUS in 10-day-old emf1-1 fwa-1 or emf1-1 ft-1AP1::GUS seedlings (Fig. 1H, I), the detection of low AP1 activity in emf1-1 fwa-1 and emf1-1 ft-1 double mutants might be due to the high sensitivity of RT-PCR strategy. The reduction of AP1 signal was clearly caused by the mutations of these late-flowering genes. We therefore concluded that AP1 activity was negatively regulated by FWA and positively activated by other late-flowering genes.
LFY expression was reduced in emf1-1 late-flowering double mutants
AP1 expression has been reported to be activated by another flower meristem identity gene LFY (Ruiz-Garcia et al. 1997, Liljegren et al. 1999, Wagner et al. 1999). In addition, AP1 has also been thought to be able to positively regulate LFY activity (Ruiz-Garcia et al. 1997, Liljegren et al. 1999). To examine whether the reduction of AP1 expression in emf1-1 late-flowering double mutants was correlated with the changes of LFY activity, RT-PCR combined with Southern analysis was also performed. As shown in Fig. 2A and C, strong LFY signal was detected in 10-day-old emf1-1 single mutants. This result matched well with the high AP1 activity observed in emf1-1 single mutants (Fig. 2A, B). AP1 and LFY were therefore co-expressed at high levels in the absence of EMF1 activity and resulted in the early initiation of flower organs in emf1-1 mutants. When six emf1-1 late-flowering double mutants were analyzed, LFY expression was detected in all different double mutants 10 d after germination (Fig. 2A, C). The strength of the LFY signal was however reduced to about 10 to 40% of that observed in emf1-1 single mutant seedlings (Fig. 2C). Similar to the result obtained for AP1, LFY signal was undetectable in wild-type Columbia seedlings (Fig. 2A, C). Our result clearly demonstrated that in the absence of EMF1 activity, similar to AP1, LFY was also activated by most late-flowering genes tested here. Therefore, late-flowering genes should play critical roles in the regulation of LFY activity during floral transition in Arabidopsis.
The expression of AP1 and LFY was reduced in emf2-1 late-flowering double mutants
To explore the question that if the expression of AP1 and LFY was also influenced by late-flowering mutation in emf2-1 mutants, the expression of AP1 and LFY was also examined in emf2-1 late-flowering double mutants by RT-PCR and Southern analysis. AP1 and LFY activity were examined in five representative emf2-1 late-flowering (fwa-1, ft-1, fe-1, co-1 and gi-1) double mutants (Haung and Yang 1998). Similar to that observed in emf1-1 single mutants, the expression of both AP1 and LFY were strongly detected in emf2-1 single mutants and were reduced in all five emf2-1 late-flowering double mutants 10 d after germination (Fig. 3). Signals for AP1 and LFY were absent in 10-day-old wild-type Columbia seedlings. As was the case in emf1-1, the results indicated that the early expression of AP1 and LFY in emf2-1 mutants was due to the absence of EMF2 activity. Their precocious expression was however suppressed by the mutation in either one of the late-flowering genes tested.
In addition to controlling flowering time, early-flowering genes EMF and most late-flowering genes have been thought to affect inflorescence and flower development (Yang et al. 1995, Chen et al. 1997, Piñeiro and Coupland 1998, Yang and Chou 1999, Aukerman et al. 1999). Although double mutant analyses indicated that different late-flowering mutations caused different effects on emf mutation (Haung and Yang 1998), the question of how late-flowering genes in different flowering pathways interact with EMF in regulating floral meristem identity genes is less clear. Since AP1 promoter was activated early in emf seedlings, EMF genes were thought to be suppressors for AP1 in wild-type plants (Chen et al. 1997). This assumption was confirmed by the result presented here that AP1 expression was strongly detected in emf mutants just 10 d after germination. Since the function is opposite between EMF and most late-flowering genes except FWA which was functionally similar to EMF as repressor for flower transition (Koornneef et al. 1998, Levy and Dean 1998, Soppe et al. 2000), it was reasonable to propose that the early AP1 activity in emf mutants may be partially due to the activation of late-flowering genes (or without the repression by FWA). If this is true, AP1 activity would be reduced or diminished in emf mutants once late-flowering genes were mutated. By contrast, AP1 activity will not be influenced in an emf late-flowering double mutant if either this late-flowering gene is not involved in the activation (or repression by FWA) of AP1 or it is an upstream gene of EMF.
To test the possible relationships between EMF and late-flowering genes in regulating AP1, AP1::GUS activity as well as AP1 expression were examined in double mutants constructed between different late-flowering and emf mutants. Our results clearly indicated that all different late-flowering mutations tested here caused a reduction on the expression of AP1 in both emf1 and emf2 mutants no matter if they are in the same or different flowering pathways. The significant reduction of GUS activity in emf fwa or emf ft double mutants indicated that AP1 expression was reduced once these two late-flowering genes are mutated. In addition, double mutants in all combinations between late-flowering mutants (gi-1, co-1, ft-1, fe-1, fwa-1, or fca-1) and emf mutants (emf1-1, emf2-1) displayed a clear reduction of AP1 expression. These results suggest that AP1 is likely a downstream target gene for late-flowering genes tested here and may provide evidence to support the notion that most late-flowering genes function to activate AP1, whereas FWA functions to repress AP1, in regulating the floral initiation in Arabidopsis (Fig. 4). This result also indicates that late-flowering genes tested here are not likely regulating AP1 through acting as an upstream gene for EMF. These late-flowering genes are possibly functioning in the downstream of EMF or in a pathway independent from EMF (Fig. 4).
The high expression of LFY was strongly correlated with the early AP1 expression in emf mutant seedlings (Fig. 2, 3). This result indicated that LFY was also likely a downstream gene for EMF and was suppressed by EMF in wild-type plants. LFY has been thought to positively regulate AP1 directly (Mandel and Yanofsky 1995, Weigel and Nilsson 1995). This assumption was supported by the observation that AP1 expression was delayed in lfy mutants (Liljegren et al. 1999); LFY protein was able to bind to the AP1 promoter (Parcy et al. 1998), and direct activation of AP1 by LFY was confirmed (Wagner et al. 1999). If AP1 is indeed regulated by LFY, then high AP1 expression in emf mutants can be simply explained due to the activation of LFY. Therefore LFY seems to have redundant function with late-flowering genes in regulating AP1. Based on the data reported previously, LFY may interact with late-flowering genes differently. For example, LFY expression has been thought to be activated by CO, GI, FCA, FVE and LD (Simon et al. 1996, Nilsson et al. 1998, Aukerman et al. 1999, Reeves and Coupland 2000) whereas its activation was independent from FT, FWA and FE (Ruiz-Garcia et al. 1997, Nilsson et al. 1998, Reeves and Coupland 2000). Therefore, it is interesting to explore the question of how LFY interacts with EMF and different late-flowering genes during floral initiation.
If a late-flowering gene functions upstream of LFY in regulating AP1, LFY activity should be reduced or abolished in emf mutants once this late-flowering gene is mutated. By contrast, LFY activity will not be influenced in an emf late-flowering double mutant if this late-flowering gene functions in a pathway independent from LFY. Similar to that observed for AP1, our results indicated that six different late-flowering mutations (gi-1, co-1, ft-1, fe-1, fwa-1 or fca-1), no matter if they are in the same or different flowering pathways, all caused reduction on the expression of LFY in both emf1 and emf2 mutant backgrounds. This result suggests that these six late-flowering genes are likely functioning in the upstream of LFY in regulating AP1 activity (Fig. 4). This result also indicates that these six late-flowering genes are not upstream genes for EMF in regulating LFY activity (Fig. 4).
This result was however different from those described for FT, FWA and FE in which the activation of LFY has been thought to be independent from these three genes (Ruiz-Garcia et al. 1997, Nilsson et al. 1998). One possible explanation is that FT, FWA and FE may be involved in the regulation of LFY through activating AP1 (Fig. 4) (Blázquez 2000). In this case, AP1 but not LFY is the target gene for these late-flowering genes. It has been shown that AP1 and LFY function redundantly and can regulate each other positively (Liljegren et al. 1999, Blázquez 2000). Therefore, the high expression of LFY in emf mutant may be partially due to the activation of AP1 by these late-flowering genes. The reduction of LFY expression in double mutants between emf and late-flowering mutations such as ft, fwa or fe may be caused by the lack of AP1 activity.
Another possibility is that FT, FWA and FE may be involved in regulation of LFY through positive response to LFY activity whereas genes such as FCA, CO and GI activated LFY through transcriptional regulation (Nilsson et al. 1998, Reeves and Coupland 2000) (Fig. 4). Based on this assumption, mutation in any one of these six late-flowering genes will cause a similar reduction of LFY activity. Therefore, overexpression of LFY in ft, fwa or fe mutant will not correct the delay in flowering in these mutants as seen in Nilsson et al. (1998) since LFY activity is not increased. This assumption is also supported by the fact that mutations in FT, FWA and FE have only a small effect on the activity of the LFY promoter (Nilsson et al. 1998). Although our results were unable to distinguish these two possibilities, they however provided direct evidence to support the notion that late-flowering genes are involved in floral initiation by regulating AP1 and LFY genes (Fig. 4).
We previously reported that the reproductive defects in emf co-1 and emf gi-1 double mutants is less sever than that in emf fwa-1 and emf ft-1 (Yang et al. 1995, Haung and Yang 1998). This result suggested that LFY and AP1 activity in emf co-1 and emf gi-1 double mutants should be similar to that in emf single mutants and higher than in emf fwa-1 and emfft-1 double mutants. Our result, however, showed that both LFY and AP1 expression were also reduced in emf co-1 and emf gi-1 double mutants. One explanation for this result is that CO and GI are involved in more than one pathway to regulate LFY and AP1 (Fig. 4). Recently, CO has been reported to directly activate two separate flowering pathways in regulating AP1 and LFY, one involving FT/FWA and another involving SUPPRESSOR OF OVEREXPRESSION OF CO1 (SOC1) (Onouchi et al. 2000, Samach et al. 2000, Blázquez 2000). Therefore, it is reasonable to believe that the expression of AP1 and LFY was lower in 10-day-old emf co-1 and emf gi-1 double mutant seedlings than in emf single mutants as seen in our result due to the lack of activation by GI, CO, FT and SOC1 during this early developmental stage.
Since the expression of FT was detected in co mutants during late development, it has been thought that at least one pathway other than CO/GI pathway was involved in the activation of FT during late development (Kardailsky et al. 1999, Kobayashi et al. 1999). This result suggested that FT can be activated by such a pathway late on and lead to the activation of AP1 and LFY (Fig. 4) and caused the early appearance of reproductive organs in emf co-1 or emf gi-1 double mutants as described previously (Haung and Yang 1998). By contrast, the FT activity was constantly reduced in emf ft-1 and emf fwa-1 double mutants due to the ft mutation or the negative regulation by fwa respectively (Onouchi et al. 2000, Samach et al. 2000, Blázquez 2000). This caused the low expression of AP1 and LFY all the time and resulted in the much severe reproductive defects in emf ft-1 and emf fwa-1 double mutants (Haung and Yang 1998).
In summary, EMF genes may regulate the floral initiation by acting as upstream genes to negatively regulate late-flowering genes in controlling LFY and AP1 activity. EMF genes may also prohibit the expression of LFY and AP1 through distinct pathway from most late-flowering genes. In emf late-flowering double mutants, the expression of AP1 and LFY was delayed or reduced due to the loss of the promotion or maintenance by late-flowering genes. The incomplete loss of AP1 and LFY activity supported that at least two pathways existed among these late-flowering genes in regulating flower initiation.
We thank Dr. M. Yanofsky for providing transgenic plants harboring an AP1::GUS chimeric construct. We also thank Dr. R. Sung for providing emf1-1 single mutants contained AP1::GUS as controls. This work was supported by a grant to C-H Y from the National Science Council, Taiwan, ROC, grant number: NSC89–2311-B-005–025.
Corresponding author: E-mail, email@example.com; Fax, +886-4-2285-3126.