LEAFY expression and flower initiation in Arabidopsis

Post-embryonic life of flowering plants is divided into two distinct phases: an initial vegetative phase, during which leaves with associated lateral shoots or paraclades are produced, and a subsequent reproductive phase, during which flowers are produced. The transition between the two phases is caused by a complex process termed floral induction, which is controlled by both endogenous and environmental signals. In Arabidopsis, a facultative long-day plant, flowering is promoted by long photoperiods, vernalization (transient exposure to cold), and higher growth temperatures (Napp-Zinn, 1985; Martínez-Zapater and Somerville, 1990; Koornneef et al., 1995).

Physiological approaches to study the molecular details of floral induction have recently been complemented by genetic studies, which have led to the identification of a large group of flowering-time mutations that either reduce or lengthen the time to flowering (Koornneef et al., 1991; Zagotta et al., 1992; Martínez-Zapater et al., 1994; Peeters and Koornneef, 1996). The genes inactivated in two late-flowering mutants, luminidependens (ld) and constans (co), encode putative transcription factors (Lee et al., 1994; Putterill et al., 1995), each of which functions in a different floral induction pathway. Mutations in LD cause late flowering in both long and short days, and, consistent with this, the abundance of LD mRNA is not affected by photoperiod (Lee et al., 1994). In contrast, co mutants are only delayed in long days, and CO mRNA is more abundant in long than in short days (Putterill et al., 1995). A causal relation between CO RNA levels and flowering time has been confirmed by Simon and coworkers (1996), who showed that overexpression of CO causes photoperiod-independent early flowering.

Another group of genes involved in the floral transition are the flower-meristem-identity (or floral-initiation-process) genes. Among these, the LEAFY (LFY) gene stands out, because its expression precedes that of other meristem-identity genes with flower-specific expression, and because lfy loss-of-function mutations have the strongest effect on meristem identity (Mandel et al., 1992; Weigel et al., 1992; Jofuku et al., 1994; Ingram et al., 1995; Kempin et al., 1995). These observations suggest that LFY is responsible for the initial steps in flower initiation.

At the time when wild-type plants begin to produce flowers, lfy mutants continue to produce leaves and associated lateral shoots. In contrast to early-arising flowers, later-arising flowers have only a partial requirement for LFY activity due to the redundant activity of other genes such as APETALA1 (AP1), and later flowers are replaced by structures that combine characteristics of both flowers and shoots. One indication of this mixed identity is that lfy mutant flowers are subtended by leaf-like organs, or bracts (Schultz and Haughn, 1991; Huala and Sussex, 1992; Weigel et al., 1992). Conversely, while inactivation of LFY causes the transformation of flowers into shoots or shoot-like structures, constitutive expression of LFY from the cauliflower-mosaic-virus 35S promoter causes a transformation of all lateral shoots into solitary flowers and a reduction in the number of leaves on the primary axis, particularly in short days (Weigel and Nilsson, 1995). In addition to its role in establishing flower-meristem identity, LFY functions also in maintaining meristem identity, and plants heterozygous for a lfy null mutation show a high rate of floral reversion when grown in short days (Okamuro et al., 1996).

The analysis of 35S::LFY plants also revealed a second level of control of flower initiation, namely regulation of the competence of a meristem to respond to LFY activity, as all 35S::LFY transformants make at least a few leaves before flowers without subtending leaves are produced (Nilsson and Weigel, unpublished results). That the 35S::LFY phenotype is attenuated in short days indicates that acquisition of competence is at least partially dependent on environmental factors (Weigel and Nilsson, 1995).

To gain new insights into the molecular basis of flower initiation, we have studied in detail the activity of the promoter of the LFY gene, which has been previously reported to be expressed at high levels in flowers (Weigel et al., 1992). This study focuses on LFY expression during the entire life cycle of the plant, while the accompanying study by Hempel et al. (1997) describes in more detail the relationship between LFY expression and determination of floral fate during the transition to flowering. The two major conclusions from our studies are that the profiles of LFY expression during the Arabidopsis life cycle differ under different environmental conditions, and that the relative levels of wild-type LFY affect the timing of the transition to flowering.

All transgenes were constructed in the pCGN1547 transformation vector (McBride and Summerfelt, 1990). pDW130 contains a 5.3 kb genomic LFY fragment (nucleotides 464 to 5957; GenBank accession no. M91208; Weigel et al., 1992). pLS41 contains a 2.3 kb BamHI promoter fragment fused at the downstream site (nucleotide 2755) to the LFY cDNA excised from pDW123, representing a longer splice form, but otherwise identical to pDW122 (Weigel et al., 1992). 3′ of the LFY cDNA is a 300 bp EcoICR-EcoRI fragment from pBI221 (Clontech; Jefferson et al., 1987), encompassing the nos transcriptional terminator from Agrobacterium. pDW150 is derived from pDW137, a pCGN1547 derivative containing the HindIII-EcoRI fragment of pBI101.2, which encompasses the β-glucuronidase (GUS) coding region and the nos terminator (Clontech; Jefferson et al., 1987). The 2.3 kb BamHI fragment spanning the LFY promoter was inserted in front of GUS in pDW137, such that the fusion gene uses the LFY initiation codon, which overlaps the downstream BamHI site (ATGGATCC). The authentic GUS initiation codon is seven codons downstream from the LFY initiation codon.

The lfy-12 and lfy-26 mutant alleles have been described by Huala and Sussex (1992) and Lee et al., (1997); see also URL: http://www.salk.edu/LABS/pbio-w/leafyseq.html. Transgenic plants were generated with the vacuum-infiltration method (Bechtold et al., 1993). pDW130 was introduced into the Nossen ecotype; pLS41 into Nossen and Columbia ecotypes as indicated. pDW150 was introduced into the Columbia, Landsberg erecta and Nossen ecotypes, and lines used here were homozygous for the transgene.

Plants were grown at 23°C in long-day (16 hours light/8 hours dark) or short-day cycles (10 hours light/14 hours dark), under a mixture of Cool White and Gro-Lux/WS fluorescent lights (Osram, Sylvania). For GA treatments, plants were liberally sprayed twice weekly with a solution of 100 μM GA3 (Sigma) in 0.02% Tween-20.

For histochemical analyses, dissected apices were incubated in 50 mM X-gluc staining solution (50 mM sodium phosphate buffer pH 7, 0.2% Triton X-100, 3 mM potassium ferricyanide, 3 mM potassium ferro-cyanide, 20% methanol), first on ice under vacuum for 1-2 hours, then at 30°C for 16 hours. For whole-mount photographs, tissue was dehydrated in 70% ethanol and photographed under differential interference contrast optics on a Zeiss Axioskop microscope. For photographs of sections, tissue was dehydrated and embedded in Paraplast (Sherwood Medical). 12 μm thick sections were prepared on a Leica microtome, mounted, and, after removal of Paraplast with xylenes, photographed either under bright-field or dark-field illumination.

For quantitative measurements, we refined an assay using 4-methyl umbelliferyl glucuronide (MUG), which is converted by GUS enzyme into the fluorescent product 4-methyl umbelliferone (4-MU). Apices were incubated at 37°C for 16 hours in 1 mM MUG assay solution (50 mM sodium phosphate buffer pH 7, 10 mM EDTA, 0.1% SDS, 0.1% Triton X-100), in individual wells of a microtiter plate. After the reaction had been stopped by the addition of 0.3 M Na2CO3, fluorescence at 430 nm was measured on a luminescence spectrophotometer equipped with an ELISA plate reader (Perkin Elmer, model LS50B).

In situ hybridization using antisense RNA probes labeled with [S]-UTP as well as photography and image processing of hybridized sections were as described by Drews et al. (1991) and Lee et al. (1997). Template for the LFY probe was pDW124 (Weigel et al., 1992; URL: http://www.salk.edu/LABS/pbio-w/pDW124.html); for GUS it was pLS27, which contains the GUS coding region from pBI221 (Jefferson et al., 1987) subcloned into pBstKS+ (Stratagene).

DW130 or LS41 transformants in Nossen ecotype were crossed to lfy-26 in Landsberg erecta. To genotype F2 plants, we developed a CAPS (cleaved amplified polymorphic sequence; Konieczny and Ausubel, 1993) marker that distinguishes between the Landsberg erecta and Nossen alleles of LFY. The two oligonucleotide primers used for polymerase chain reaction (PCR) were 5′-GCT GAT ATC GTT TAA CTA TCT TAA GAT ACA TGG-3′ and 5′-CGC TCA GTT GGT TGA CTC CGA CTC-3′. Since the upstream primer is located outside the region present in the transgenes, only endogenous LFY sequences are amplified. Amplification of Landsberg erecta DNA yields a 1.9 kb product that is digested by RsaI into fragments of 970, 813 and 156 bp. The 970 bp fragment contains an additional RsaI site in the Nossen ecotype, yielding two fragments of about 600 and 370 bp.

In the experiments with reduced copies of LFY, the lfy-12 mutation was detected with a CAPS marker based on a MaeIII site created by the lfy-12 nucleotide change, which is the same as in lfy-1 (Weigel et al., 1992; URL: http://www.salk.edu/LABS/pbio-w/caps.html).

lfy null mutations interfere with the formation of normal flowers. In positions where the first flowers are found on the inflorescence shoot of a wild-type plant, additional leaves and lateral shoots are present in lfy mutants, while later-arising flowers are replaced by structures that have only partial shoot character. One indication of this mixed identity is that lfy mutant flowers are subtended by small leaves called bracts (Schultz and Haughn, 1991; Huala and Sussex, 1992; Weigel et al., 1992).

To define the extent of genomic sequences required for in vivo function, we crossed four independent transgenic lines containing a 5.3 kb fragment of the LFY region (DW130) to plants carrying the strong lfy-26 allele, and analyzed the phenotype of the F₂ progeny. In all cases, a single copy of the transgene was able to complement all defects of the lfy mutant, both in short and in long days. To further test whether regulatory sequences contained within the LFY promoter were sufficient for in vivo function, we constructed a mini-gene consisting of the LFY cDNA fused to 2.3 kb of LFY 5′ upstream sequences. This minigene (LS41) complemented the inflorescence and flower defects caused by the lfy-26 mutation in two of three independent lines tested in both long and short days. One line showed no rescue in short days, but almost complete rescue in long days (Table 1). This result indicates that the 2.3 kb promoter can confer a pattern of LFY expression that is sufficient to support normal flower development, and that this promoter fragment is suitable for reporter gene studies.

We chose the reporter gene uidA encoding E. coli β-glu-curonidase (GUS; Jefferson et al., 1987), because of the ease with which its expression is quantified, and generated 65 lines containing a LFY::GUS transgene (DW150) in the ecotypes Columbia, Landsberg erecta and Nossen. GUS expression was easily detectable by X-gluc staining in inflorescence apices of many lines, and several representative ones were chosen for further analysis after the transgene insertions had been made homozygous. We compared, in more detail, the pattern of GUS RNA expression to that of LFY RNA in inflorescence apices by in situ hybridization, and found GUS and LFY RNA expression to be similar (Fig. 1). Strong GUS expression is apparent in floral anlagen and in flowers of stages 1 and 2, in which LFY is thought to execute its role in promoting floral identity (Weigel et al., 1992). The pattern of GUS RNA from stage 3 on, however, deviates from that of endogenous LFY, with GUS RNA levels being strongest in prospective whorls two and three, whereas LFY RNA levels are highest in emerging sepal primordia. Although differences in RNA stability might contribute to the differences seen in RNA accumulation, it is likely that other elements in addition to the 2.3 kb promoter are needed to completely reproduce the LFY RNA pattern.

To determine the time course of LFY promoter activity under different photoperiods, we grew three independent Columbia and two Landsberg erecta lines homozygous for the LFY::GUS insertion in long and short days, and analyzed samples taken at different time points from germination until flower buds became visible to the naked eye. GUS activity was quantitatively measured in individual dissected apices, using the substrate 4-methyl umbelliferyl glucuronide (MUG), which is converted by GUS into the fluorescent product 4-MU. The procedure we used does not involve mechanical protein extraction, and GUS activity was determined on a per-apex basis. The quantitative data obtained with the MUG assays were complemented qualitatively by histochemical X-gluc analyses and by in situ hybridization to endogenous LFY RNA.

For all lines tested, the profiles of GUS activity were similar to the one shown in Fig. 2. In long days, LFY::GUS expression was sharply upregulated beginning after about 9 days, which correlates with the short time to flowering under this light regime (21 days to visible flower buds in Columbia). We expected a similar pattern in short days, that is, initial low activity, with a sharp increase at the time of the transition to flowering. The observed profile, however, differed strikingly from the expected one (Fig. 2). Starting from initially low levels, LFY::GUS expression increased gradually several weeks before flowers were initiated (56 days to visible flower buds).

Since plant hormones of the gibberellin (GA) class are known to accelerate flowering in short days (Langridge, 1957; Wilson et al., 1992), we also tested the effects of exogenously applied GA₃ on LFY::GUS expression. GUS activity in shortday grown plants that were continuously treated with GA₃ increased more rapidly than in control plants (Fig. 2). Since this was paralleled by a reduction in flowering time, from 56 to 43 days, the effects of GA on flowering time correlate with those on LFY promoter induction.

The quantitative measurements of whole apices did not indicate whether the observed changes in GUS activity were caused by a change in the number of LFY::GUS expressing cells, or an absolute change of LFY::GUS activity in a constant number of cells, or by a combination of both. To distinguish between these possibilities, and to determine the tissue in which the LFY promoter was active, we localized LFY::GUS expression by staining plants in the vegetative phase with the chromogenic substrate X-gluc (Fig. 3). GUS activity was already detectable in 4-day-old seedlings, both in short and long days. At this and later time points, GUS activity was present in the newly emerging leaf primordia, but absent from older leaves. In sections of 3-week-old plants, we detected weak GUS activity even in leaf anlagen (Fig. 3D).

Examination of X-gluc-stained apices showed also that the number of GUS-positive leaf primordia increased during the first 10 days under long-day conditions and the first 20 days under short day conditions (results not shown). It remained constant afterwards, suggesting that the overall increase in LFY::GUS expression after these time points primarily reflected increased activity of the LFY promoter, at least on a per-primordium basis.

To evaluate the validity of the LFY::GUS analyses, we monitored expression of endogenous LFY RNA by in situ hybridization. Wild-type Columbia plants were grown in short days, and harvested for in situ hybridization at 2, 4, 6 and 8 weeks after germination (Fig. 4). Detection of LFY RNA in apices of 2-week-old plants was close to the detection limit, but silver grain density was consistently higher over leaf primordia than over the shoot apical meristem (Fig. 4A). Expression of LFY RNA in emerging leaf primordia was readily apparent in 4- and 6-week-old plants, with higher levels in the older plants (Fig. 4B,C). While it is difficult to quantify expression levels by in situ hybridization, these qualitative data together with the quantitative MUG assays indicate that the relative levels of LFY RNA in emerging leaf primordia increased gradually during the first 6 weeks of vegetative growth under short-day conditions. Although we have not analyzed long-day-grown plants in detail, it has recently been reported that low levels of endogenous LFY RNA are found in 3-day-old long-day grown seedlings (Bradley et al., 1997).

We have detected GUS activity, but not LFY RNA, in leaf anlagen, and we detect GUS activity in more primordia than we detect LFY RNA. These differences likely reflect that detection of GUS activity is more sensitive than detection of LFY RNA by in situ hybridization, and that GUS protein is more stable than most RNAs. While higher sensitivity in the detection assay is not of concern, the higher stability of GUS protein means that the gradual increase in LFY::GUS activity should be viewed as a qualitative result.

The more rapid increase of LFY::GUS expression in long days than in short days suggested that the LFY promoter is a target of photoperiodic regulation. Therefore, we measured GUS activity in LFY::GUS plants that were grown under short-day conditions for either 14 or 28 days, after which they were still in the vegetative phase, and then transferred to inductive long-day conditions. The inductive treatments had similar effects on 14- and 28-day-old plants, and upregulation of LFY::GUS was apparent after three inductive photoperiods (Fig. 5). A similar set of experiments was performed with 28-day-old wild-type Columbia plants grown in short days, and LFY expression was studied by in situ hybridization. Upregulation of LFY RNA upon transfer to long days occurred in the newly emerging primordia at the flanks of the shoot apex, as well as in already existing young leaf primordia (Fig. 6). We also studied the induction of LFY::GUS expression upon GA treatment of 21-day-old short-day grown plants. In this case, LFY::GUS expression after the beginning of the treatment increased faster than in untreated plants (not shown), but was not as dramatic as observed for plants transferred to long days.

Previous studies have shown that transient exposure to long days can stimulate flowering in plants grown in short days. The efficiency with which transient exposure to long days reduces flowering time depends on the age of the plants and on the number of long days before plants are returned to short days (Mozley and Thomas, 1995; Corbesier et al., 1996). To investigate whether transient inductive photoperiods affect the level of LFY expression only temporarily or, alternatively, promote an irreversible activation of LFY, we studied the effect of transient long-day treatments on LFY::GUS expression. LFY::GUS plants were grown in short days for 21 days and then exposed to 2, 4 or 6 consecutive long days before being returned to short-day conditions. In our conditions, 2 long days were ineffective in promoting flower initiation or increasing LFY::GUS expression. In contrast, both four and six inductive photoperiods led to a temporary increase in the rate of LFY::GUS expression for the duration of the treatment (Fig. 7). Both treatments reduced flowering time as measured by the number of leaves produced (37 and 35 leaves, respectively) when compared to control plants left in short days (39 leaves). Thus, there was a correlation between the level of LFY promoter expression and the lengths of the transient inductive treatments.

A more detailed study of the behavior of the LFY promoter upon transfer from non-inductive to inductive light conditions, and comparison of LFY upregulation with the time point of commitment to flowering, can be found in the accompanying manuscript (Hempel et al., 1997). The light conditions used here were different from those used by Hempel et al. (1997), who observed a more rapid increase in LFY::GUS activity upon induction by extended photoperiods than reported here. In addition, those authors observed a decrease in LFY::GUS activity in short-day grown plants during advanced stages of vegetative growth. Similarly, we have recently found that when plants are grown under light with an increased red-to-far-red ratio, total LFY::GUS activity increases only during the first 6 weeks of vegetative growth, with a substantial drop after this time period (Lee and Weigel, unpublished results).

The gradual change in LFY expression during the vegetative phase suggested that the level of LFY expression might be an important determinant in flower initiation. This idea is partially supported by the precocious flowering of 35S::LFY plants, which have also a reduced numbers of leaves, particularly under short days (Weigel and Nilsson, 1995). However, in the case of 35S::LFY it is difficult to discern whether the effect on flowering time is a direct one or whether it is an indirect consequence of the formation of ectopic lateral flowers, which might in turn influence the initiation of leaves on the main shoot. To investigate whether in wild-type plants there is a critical level of LFY expression that turns a newly emerging primordium into a bract-less flower instead of a leaf with associated shoot, we attempted to change levels of endogenous LFY without changing the spatial expression pattern by varying the number of LFY wild-type copies from one to four. A standard indicator for flowering time is the number of leaves produced on the primary shoot before the first flower is initiated (Koornneef et al., 1991), and we counted leaf number in a segregating population of lfy-12 mutants, which provided individuals with one or two copies of LFY, as well as in three transgenic lines segregating for a LFY transgene that contains the LFY cDNA under the control of the LFY promoter, which provided individuals with two, three or four copies of LFY. Homozygous lfy mutants (zero copies) were also included in the experiment, although in this case normal flowers were never initiated and leaf counts are not immediately comparable.

In both long and short days, we observed statistically significant differences between wild-type plants and plants with reduced or increased copy number of LFY (Table 2). Plants with one copy of LFY flowered later than wild-type plants, while plants with one or two extra copies flowered earlier. Varying the copy number of LFY affected primarily the number of cauline leaves (bracts), although the variation in rosette-leaf number was also significant. The presence of two extra copies of the LFY cDNA driven by the LFY promoter caused precocious flower initiation in all three transgenic lines tested, but with varying efficiency, most likely due to position effects of the integration sites.

Our results with lfy-12 plants are at variance with those of Schultz and Haughn (1991), who reported for lfy-1, which carries the same mutation as lfy-12 and was induced in the same ecotype, that there is a small, statistically insignificant reduction in rosette leaf number. However, Mizukami and Ma (1997) recently reported an increase in rosette leaf number in plants homozygous for the weak lfy-5 allele.

The switch from the formation of leaves with associated lateral shoots to the formation of flowers is but one example of the many changes that occur during the Arabidopsis life cycle, and which are collectively known as phase change (Poethig, 1990). Several of these changes, which include altered patterns of trichome distribution on leaves, an increase in the rate of cell division in the apex, the release of lateral shoots, and internode elongation on the main stem, are closely associated with the transition between the vegetative and reproductive phases of growth (Napp-Zinn, 1985; Hempel and Feldman, 1994; Chien and Sussex, 1996; Telfer et al., 1997). These processes, along with flowering, are normally synchronized and are triggered by the same environmental signals, but can be uncoupled under specific circumstances (Chien and Sussex, 1996; Telfer et al., 1997). To understand the control of phase change, we have to learn how the genes that execute the different processes are regulated. Here, we have analyzed how the expression level of a gene involved in flower initiation, LFY, is controlled by environmental and endogenous factors, and how levels of LFY activity in turn determine floral fate.

Two models have been used to describe the transition to flowering in Arabidopsis. The traditional view has been that the apical meristem dictates the fate of lateral meristems, and that in order for the shoot meristem to produce flowers, it has to switch from a vegetative to an inflorescence phase (Sussex, 1989). The other model postulates that phase change affects lateral primordia directly, causing primordia that would otherwise develop into leaves with associated shoots (paraclades) to adopt a floral fate, with the shoot meristem proper being of little, if any, importance in this transition (Hempel and Feldman, 1994, 1995). Morphological changes in the central, undifferentiated zone of the shoot apical meristem, as well as the rapid induction of genes expressed in the center of the shoot apical meristem support the first model (Miksche and Brown, 1965; Mandel and Yanofsky, 1995; Menzel et al., 1996). Our observation that LFY expression in lateral primordia is continuous from the vegetative to reproductive phase and merely changes in intensity appears to fit well with the second model, although it does not exclude the former. This observation is also compatible with the previous suggestion that the activity of upstream-acting genes that control phase switching would increase gradually during the Arabidopsis life cycle (Schultz and Haughn, 1993; Yang et al., 1995). However, the connection between changes in activity of these upstream factors and LFY expression might be complex, as indicated by the observation that the initial increase of LFY::GUS activity during the vegetative phase is not maintained when plants are grown under light with a high red-to-far-red ratio (Hempel et al., 1997; Lee and Weigel, unpublished observations).

We do not believe that LFY has a positive role in leaf development, based on the absence of leaf defects in lfy mutants. However, LFY suppresses the outgrowth of leaves during the reproductive phase, since all lateral structures on the inflorescence shoot of a strong lfy mutant are subtended by small leaves called bracts (Schultz and Haughn, 1991; Weigel et al., 1992). Therefore, we propose that LFY is expressed in emerging leaf/paraclade primordia because this is the place where floral induction is effective, as these primordia have the potential to adopt the alternative floral fate, once LFY activity reaches a critical level.

One of the questions that still remains to be solved concerns the molecular mechanism by which LFY exerts its role and initiates flower formation at the shoot apex. Previous work has demonstrated that flower initiation is promoted by a combination of LFY expression and competence to respond to LFY activity (Weigel and Nilsson, 1995). We have extended this observation and shown that the time to flowering is critically affected by levels of LFY expression in its normal pattern, as determined with plants that carry one, two, three or four copies of wild-type LFY. Since altering copy number of wild-type LFY changes flowering time, LFY can be said to have characteristics of both flowering-time and flower-meristem-identity genes, which is the behavior expected for a gene that integrates the signals from the environment and then promotes a switch in meristem identity. A related finding is that FLO, the LFY ortholog in Antirrhinum, affects development of both individual flowers and of full inflorescence traits (Bradley et al., 1996b).

Our observations also provide additional support for the idea that LFY is a rather direct target of floral-induction signals, because upregulation of LFY in long days is not restricted to future flower primordia, but includes previously initiated leaf primordia (Fig. 6H). Apparently, there is no need for precise spatial regulation of LFY, because more advanced leaf primordia are insensitive to LFY activity. Furthermore, the different responses of the LFY promoter to long and short days suggest that the pathways defined by these two conditions are separable at the LFY promoter. Simon et al. (1996) have recently obtained at least circumstantial evidence that the signal transduction path from one specific element in the long-day pathway, the transcription factor encoded by the flowering-time gene CO, to the LFY promoter might indeed be very short. They showed that, irrespective of day length, induction of CO activity in transgenic plants triggers both immediate flowering and upregulation of LFY, while upregulation of other meristem-identity genes lagged, when compared to induction by long days.

Expression of LFY and its orthologs has now been studied in detail in wild-type plants of three species, Nicotiana (Kelly et al., 1995), Antirrhinum (Bradley et al., 1996b), and Arabidopsis (this work), and all three species have been found to behave differently.

In Nicotiana, the apparent LFY ortholog NFL is expressed constitutively, with expression in the vegetative phase being confined to defined regions of the apical meristem including emerging leaf primordia (Kelly et al., 1995). Although a NFL mutant is not available, it is likely that NFL/LFY plays a role in flower initiation, as overexpression of the Arabidopsis LFY gene in Nicotiana induces early flowering (Weigel and Nilsson, 1995). It thus appears that the main check point controlling the transition to flowering in Nicotiana is the competence to respond to LFY activity. One interpretation for the phenotype of 35S::LFY Nicotiana plants is that high levels of LFY activity can compensate for low competence at a young age.

In Antirrhinum, expression of the LFY ortholog FLO, whose loss-of-function phenotype is similar to that of LFY in Arabidopsis (Coen et al., 1990), is tightly correlated with floral induction, as measured by very sensitive polymerase-chain-reaction assays (Bradley et al., 1996b). FLO is not detected at all during the early vegetative phase under either inductive or non-inductive photoperiods, and its expression after floral induction is confined to newly emerging floral primordia. However, in older plants, weak FLO expression is occasionally detected at a low level in vegetative tissue (Bradley et al., 1996b). Since the effects of FLO overexpression in Antirrhinum are unknown, we cannot evaluate the importance of competence to respond to FLO in this species, but it is not unlikely that competence plays only a minor role in Antirrhinum.

Arabidopsis, then, appears to be a case that lies in between the two more extreme cases represented by Nicotiana and Antirrhinum. While LFY is expressed at low levels in leaf primordia, its RNA expression is regulated in response to environmental cues and plant age. This is paralleled by the finding that increasing LFY expression levels by increasing the copy number of wild-type LFY accelerates flowering. These observations confirm that transcriptional regulation of LFY is an important determinant of flowering time. However, the phenotype of plants that express LFY constitutively is still affected by environmental conditions, demonstrating that competence to respond to LFY activity represents a second level for the control of flowering time (Weigel and Nilsson, 1995; Fig. 8).

It has been proposed that the indeterminate architecture of the Arabidopsis inflorescence is the result of the interplay between genes that confer floral identity, including LFY, and genes that promote shoot identity, including TERMINAL FLOWER 1 (TFL1), such that TFL1 represses LFY expression in shoots and LFY represses TFL1 expression in flowers (Shannon and Meeks-Wagner, 1991; Alvarez et al., 1992; Weigel et al., 1992; Bradley et al., 1997). Similarly to LFY, TFL1 is expressed not only in the reproductive phase, but also during the vegetative phase, and tfl1 mutants flower early (Shannon and Meeks-Wagner, 1991; Bradley et al., 1997). These observations are in agreement with a role of TFL1 in the regulation of LFY expression during the vegetative phase. In Antirrhinum, on the other hand, expression of the LFY ortholog FLO and of the TFL1 ortholog CENTRORADIALIS (CEN) is largely restricted to the reproductive phase (Coen et al., 1990; Bradley et al., 1996a). It will be interesting to determine what constitutes the evolutionarily derived condition – vegetative and reproductive, or exclusively reproductive expression of LFY/FLO and TFL1/CEN.

We thank Fred Hempel, Ove Nilsson, François Parcy and Greg Gocal for discussions and comments on the manuscript, and Diana Wolfe for help with vacuum infiltration. Thanks also to Elliot Meyerowitz (Caltech), in whose lab D. W. constructed two of the vectors used in this study. This work was supported by grants from the Samuel Roberts Noble Foundation, the National Science Foundation (IBN-9406948) and the US Department of Agriculture (95-37301-2038) to D. W., by fellowships from the Spanish Ministry of Education and the Human Frontiers Science Program Organization (M. A. B.), and by NIH training grant GM07240 (L. N. S.). D. W. is a National Science Foundation Young Investigator.