Phytochrome control of flowering is temperature sensitive and correlates with expression of the floral integrator FT

Authors


Summary

In Arabidopsis flowering is accelerated by reduced red:far-red (R:FR) ratio which signals the presence of neighbouring vegetation. Hastened flowering is one component of the shade-avoidance syndrome of responses, which alter many aspects of development in response to the threat of potential competition. Of the red/far-red-absorbing photoreceptors it is phyB that plays the most prominent role in shade-avoidance, although other related phytochromes act redundantly with phyB. It is well established that the phyB mutant has a constitutively early flowering phenotype. However, we have shown that the early flowering phenotype of phyB is temperature-dependent. We have established that this temperature-sensitive flowering response defines a pathway that appears to be independent of the autonomous-FLC pathway. Furthermore, we have demonstrated that the phytochromes control the expression of the floral promoter FT. We have also shown that other phyB-controlled responses, including petiole elongation, are not sensitive to the same temperature change. This suggests that discrete pathways control flowering and petiole elongation, components of the shade-avoidance response. This work provides an insight into the phytochrome and temperature interactions that maintain flowering control.

Introduction

The onset of flowering in plants is often determined by the interaction of environmental cues with endogenous developmental signals. Temperature and light signals are amongst the most important and well-characterised environmental factors that regulate flowering time (Samach and Coupland, 2000; Simpson and Dean, 2002; Simpson et al., 1999). In particular, changes in day length and exposure to extended periods of cold temperature, both of which are predictable and reliable indicators of seasonal progression, control floral transition in many plants. In Arabidopsis, the acceleration of flowering following an extensive period of cold treatment (vernalization) is observed in many ecotypes and can be mapped as monogenic trait, with a vernalization requirement conferred by dominant alleles of the FRI gene (see Johanson et al., 2000). The product of the FRI gene promotes the accumulation of FLC, a MADS box transcription factor that acts as a floral repressor (e.g. Michaels and Amasino, 1999). A related gene, FLM/MAF1 also acts as a floral repressor though it appears to do so in a FRI-independent manner (Ratcliffe et al., 2001; Scortecci et al., 2001). FLC is also controlled by genes that act in the autonomous pathway and so represents an important convergence point for flowering pathways (Michaels and Amasino, 2001; Sheldon et al., 1999). Downstream targets for FLC are the floral integrators FT and SOC1/AGL20 (Lee et al., 2000; Rouse et al., 2002).

Like temperature, photoperiod is recognised as a major environmental determinant of flowering time in many plants. The acceleration of flowering by long days (LDs) in Arabidopsis involves the interaction of photoreceptors, principally cryptochrome 2 (cry2) and phytochrome A (phyA), with the endogenous circadian oscillator (Yanovsky and Kay, 2002). These light-signalling pathways converge at the circadian-regulated transcriptional regulator CO, providing a mechanism for the photoperiodic regulation of flowering in Arabidopsis (Suarez-Lopez et al., 2001; Yanovsky and Kay, 2002). CO has been shown to be a positive regulator of FT and SOC1 (Samach et al., 2000; Yanovsky and Kay, 2002). Indeed, the levels of FT and SOC1 mRNA appear to be determined by a balance of CO and FLC activity (Hepworth et al., 2002; Samach et al., 2000). Thus, FT and SOC1 represent important integration points between temperature- and light-signalling pathways.

In addition to photoperiod, other light signals also have a marked effect of flowering time. Light quality, specifically the relative proportions of red and far-red light (R:FR ratio), provides a powerful signal that regulates both vegetative development and the transition to flowering. The light reflected from green vegetation is attenuated in the blue and red regions, but relatively rich in the green and far-red regions. This alteration in R:FR ratio is detected by the reversibly photochromic red/far-red-absorbing phytochromes and in many plants leads to the initiation of a syndrome of responses called shade avoidance (Smith and Whitelam, 1997). Shade-avoidance responses include increased stem and petiole elongation and the acceleration of the transition to flowering. In rosette plants, such as Arabidopsis, acceleration of the onset of flowering is perhaps the most dramatic response to low R:FR ratio signals (e.g. Bagnall, 1993; Halliday et al., 1994). The precocious flowering induced by exposure to low R:FR ratio light, has possible fitness implications, as it may increase the likelihood of survival to reproduction under the otherwise unfavourable conditions of vegetational shade (Botto and Smith, 2002; Donohue et al., 2001; Dudley and Schmitt, 1995).

Of the five phytochromes in Arabidopsis, at least three (phyB, phyD and phyE) play a role in the perception of the R:FR ratio signal and the initiation of shade-avoidance responses (see Whitelam et al., 1998). From the physiological analysis of null mutants, it is apparent that phyB is the principal photoreceptor involved in R:FR ratio signal perception. Thus, compared with wild-type plants phyB mutants are constitutively elongated and early flowering and display attenuated responses to low R:FR ratio. Mutants that are null for either phyD or phyE have less obvious phenotypes, although their deficiency is far more evident when phyB is also absent (Aukerman et al., 1997; Devlin et al., 1998; 1999). Thus, phyBphyD and phyBphyE double mutants are more elongated, earlier flowering than monogenic phyB mutants. These observations suggest that phyD and phyE contribute to control of the shade-avoidance response.

Relatively little is known about the genes that act downstream of the phytochromes in the low R:FR ratio-induced acceleration of flowering. Phytochrome B-deficiency has been reported to regulate expression of the floral meristem identity gene LFY, independent of CO and FT, suggesting that the pathway is separate from the photoperiodic pathway (Blázquez and Weigel, 1999). The low R:FR ratio signal, or phyB deficiency can correct the late flowering of several induced mutations in components of the autonomous pathway that confer a vernalization requirement (Bagnall, 1993; Halliday et al., 1994). As the normal function of these autonomous pathway components is to limit the expression of FLC (e.g. Sheldon et al., 1999) and the induced mutations have elevated FLC expression, it is possible that phytochromes may also control FLC levels.

In the present study, we have investigated the roles of different phytochromes and the interaction between phytochrome status and ambient temperature in the regulation of the transition to flowering. Using mutants that are null for single or multiple phytochromes, we reveal an important role for temperature in phytochrome-regulated flowering. We demonstrate that the flowering time gene FT is regulated by the phytochromes. Furthermore, we show that other phytochrome responses are not subject to the same temperature control. This suggests the phytochromes have evolved different mechanisms to regulate specific responses.

Results

The early flowering phenotype of the phyB mutant is temperature dependent

It is now well established that phyB plays a major role in the perception of alterations in R:FR ratio, a sensitive means of detecting the presence of potential competitors. Low R:FR light ratio triggers the shade-avoidance syndrome of responses which includes increased petiole and internode elongation, reduced apical dominance and early flowering (e.g. Devlin et al., 1996; Halliday et al., 1994). This syndrome of responses is thought to confer competitive advantage when resources are limited.

Consistent with previous reports, we found that when grown under a constant temperature of 22°C, the phyB mutant displayed a constitutive elongated petiole and early flowering phenotype. Under short-day (SD) growth conditions (8-h photoperiods), phyB mutants flowered with about 10 fewer rosette leaves than wild-type plants (Figure 1). However, quite remarkably the early flowering phenotype of phyB was completely abolished when plants were grown at the slightly lower temperature of 16°C (Figures 1 and 2). Wild-type plants flowered slightly later at 16°C compared with 22°C, producing more than four additional leaves. The phyA mutant behaved in a similar manner to wild-type plants under both temperature regimes. However, the 6°C difference in temperature had a huge impact on phyB mutants, which produced 16 more leaves at bolting under the cooler conditions than phyB mutants grown at 22°C. Similar temperature-dependent effects on flowering time were observed in the phyAphyB double mutant. Whilst phyAphyB flowered slightly earlier than the monogenic phyB mutant under all test conditions, growth at 16°C invoked a delay in flowering of a similar magnitude to the monogenic phyB mutant. These data suggest that under our experimental conditions, whilst the effects of the phyA mutation on flowering were largely independent of temperature, the impact of the phyB mutation on flowering time was strongly dependent on the ambient temperature.

Figure 1.

The effect of temperature on flowering time of wild type, phyA, phyB, phyAphyB, phyAphyBphyD, and phyAphyBphyDphyE mutants. Flowering time was measured as the number of primary rosette leaves produced at bolting in plants grown in 8-h photoperiods (photon irradiance, 400–700 nm, 180 µmol m−2 sec−1) at 22°C (±1) or 16°C (±1). Standard errors are shown. Nomenclature: A = phyA, B = phyB, D = phyD and E = phyE null mutations.

Figure 2.

Phenotypes of wild type and phyB mutant in 16°C short days. Plants were grown for 78 days in 16°C (±1), 8-h photoperiods, photon irradiance, 400–700 nm, 180 µmol m−2 sec−1.

To establish if these effects were specific to plants grown under SD conditions we tested whether the phyB mutant retained its early flowering phenotype following growth under long days (LDs: 16-h photoperiods) at 16°C (Figure 3). We observed that flowering was also delayed in phyB under LDs at 16°C. In fact, phyB mutants flowered slightly later than the wild type under these conditions. Thus, it appears that the early flowering phenotype of phyB is temperature-dependent and this conditional phenotype operates under different photoperiodic conditions.

Figure 3.

Flowering time of wild type, phyA, phyB, phyAphyB, phyAphyBphyD, and phyAphyBphyDphyE mutants grown in long days. Flowering time was measured as the number of primary rosette leaves produced at bolting in plants grown at 16°C (±1) in 16-h photoperiods, photon irradiance, 400–700 nm, 180 µmol m−2 sec−1. Standard errors are shown.

Although growth at 16°C abolished the early flowering of phyB, the elongated petiole phenotype was still evident in phyB mutants grown at this temperature (Figure 2). This indicates that the effects imposed by a change in temperature on phyB-mediated flowering responses are not observed for this phyB-controlled elongation response. Thus, these different facets of phyB action may represent different branches of the phyB-signalling network.

As phyB plays a predominant role in R:FR ratio perception and the initiation of shade avoidance, the observation that the flowering time phenotype of the phyB mutant is temperature-conditional raises the question whether the flowering response of wild-type plants to R:FR ratio is also temperature-dependent. The data in Table 1 show that wild-type seedlings grown under low R:FR ratio at 16°C display a classical acceleration of flowering. These results suggest that although the phyB-mediated flowering response to shade may be perturbed at 16°C other phytochromes are capable of fully compensating for the apparent loss of phyB action under these conditions.

Table 1. The effects of temperature on the wild-type response to low R:FR ratio light
Temperature (°C)Number of rosette leaves at flowering
High R:FR ratioLow R:FR ratio
  1. Rosette leaf number was counted at flowering time for plants grown in high or low R:FR ratio light at 22°C (±1) or 16°C (±1). Standard errors are shown.

226.90 ± 0.295.60 ± 0.22
1610.00 ± 0.566.80 ± 0.25

PhyE severely delays flowering in the phyAphyBphyD mutant

The effects of growth temperature on the flowering phenotype were particularly striking in the phyAphyBphyD triple mutant. In line with previous reports, when grown under short days at 22°C the phyAphyBphyD mutant was very early flowering, producing only approximately nine leaves at bolting (Figure 1). In comparison, wild-type plants flowered with approximately 35 leaves under the same conditions. However, when grown under identical photoperiods, but at 16°C, the phyAphyBphyD mutant flowered with approximately 40 rosette leaves, a similar leaf number to the wild type. Thus, a change of just 6°C completely abolished the extreme early flowering phenotype of the phyAphyBphyD mutant. These data suggest a possible role for phyA and phyD, in addition to phyB, in the temperature-regulated control of flowering. Substantial delays in flowering for plants grown at 16°C compared with plants grown at 22°C were also observed for phyAphyBphyD mutants grown under LDs (Figure 3). Indeed, like phyB, the phyAphyBphyD mutant flowered slightly later than the wild type under these conditions.

When grown at 16°C the phyAphyBphyDphyE mutant flowered considerably earlier than the phyAphyBphyD mutant. In SDs, the quadruple mutant flowered with approximately 16 leaves, 25 fewer leaves than phyAphyBphyD (Figure 1). In LDs, phyAphyBphyDphyE flowered with approximately 5 leaves, that is, 19 fewer leaves than the phyAphyBphyD mutant (Figure 3). Thus, the absence of phyE substantially reduced the effects of temperature on flowering in phyAphyBphyD. These data suggest that in phyAphyBphyD the delay in flowering observed following growth at 16°C is largely mediated by phyE, and therefore suggest a role for phyE in the temperature-dependent control of flowering.

It was a possibility that the comparatively late-flowering phenotype of phyAphyBphyD plants grown at 16°C was a consequence of elevated phyE and/or phyC levels. This was tested by Western blot analysis. Figure 4 shows that for both phyC and phyE, protein levels were very similar in wild-type and phyAphyBphyD mutant seedlings. Previous work showed that phyC levels were reduced in a phyB mutant background (Hirschfield et al., 1998). In our experiments, phyC levels are similar in phyAphyBphyD mutant and wild-type seedlings. This suggests that the effect of phyB deficiency on phyC levels is overcome when phyA and phyD are removed in addition to phyB. Figure 4 also shows that phyC and phyE protein levels were identical in extracts from phyAphyBphyD mutant seedlings grown at either 16 or 22°C. These data suggest that the delayed flowering of the phyAphyBphyD triple mutant grown at 16°C, mediated by phyE, may be a result of down-stream signalling events.

Figure 4.

Immunoblots of phytochrome protein levels in wild-type and phyAphyBphyD mutant seedlings. Bands were detected by monoclonal antibodies selective for phyC and phyE. Protein extracts were made from 12-day-old seedlings grown in 8-h photoperiods at 22°C (±1) or 16°C (±1).

The extraordinary plasticity of the flowering response in the phyAphyBphyD mutant is accompanied by equally dramatic changes in the vegetative phenotype (Figure 5a). The prolonged vegetative phase of phyAphyBphyD observed in plants grown at 16°C was characterised by a loss of apical dominance and the consequent development of numerous axillary rosette leaves, which formed in a basal–apical direction. This is typical of development during prolonged vegetative growth and is seen in several late-flowering mutants, e.g. fca and fwa (Page et al., 1999; Soppe et al., 2000). The first of the axillary leaves was visible to the eye in both wild-type plants and phyAphyBphyD mutants between days 52 and 57, approximately half way through the vegetative phase. However, the phyAphyBphyD vegetative period was so long that total leaf number (rosette + axillary leaves) at flowering time was far in excess of the wild type (Figure 5a,b). Indeed, the mature adult phenotype of the phyAphyBphyD mutant was quite striking, with plants becoming quite bushy in appearance.

Figure 5.

(a) Phenotypes of the wild type and phyAphyBphyD triple mutant grown in 22 or 16°C.

(b) Total leaf number (rosette + axillary) of 16°C-grown wild type and phyAphyBphyD mutant.

All plants were grown in 8-h photoperiods (photon irradiance, 400–700 nm, 180 µmol m−2 sec−1), at 22°C (±1) or 16°C (±1). Standard errors are shown.

The development of axillary leaf primordia is thought to be under the control of auxin. Indeed, rapid lateral shoot development in the axr1, max1 and max2 mutants (Stirnberg et al., 1999; 2002) provides genetic evidence that auxin negatively regulates this process. This appears to be the case for phyAphyBphyD as we were able to completely inhibit axillary shoot development by exogenously applying synthetic auxin (data not shown).

The phytochromes regulate FT levels

The signalling pathways via which the phytochromes control flowering in the shade-avoidance pathway are not yet known. We were therefore interested in establishing if the floral integrators FT and SOC1 were regulated by the phytochromes and temperature in this response. We also wanted to ascertain whether the phytochromes signalled through the CO-dependent photoperiodic pathway, or through the FLM/FLC branches of the temperature-sensitive pathway. These components of the flowering pathways are controlled mainly at a transcriptional level, we therefore used quantitative PCR to measure transcript levels in 22-day-old wild-type, phyB and phyAphyBphyD seedlings grown either at 22 or 16°C (Figure 6).

Figure 6.

Expression levels of the flowering genes FLC, FLM, CO, SOC1 and FT.

Gene expression was determined for phyB and phyAphyBphyD relative to WT levels. Seedlings were grown for 22 days at 22°C (±1) or 16°C (±1) and transcript levels were measured by quantitative PCR. All plants were grown in 8-h photoperiods, photon irradiance, 400–700 nm, 180 µmol m−2 sec−1.

Expression levels of CO were slightly elevated in the phyB mutant relative to the wild type and phyAphyBphyD under 22 or 16°C. However, this relatively small increase in CO mRNA did not correlate with accelerated flowering in phyB and so is unlikely to be significant. For the vernalization-responsive genes FLC and FLM, expression levels were low in all genotypes at both temperatures. This suggests that FLC and FLM expression does not change in response to changes in ambient temperature or reduced phytochrome status. As the floral integrator SOC1 is a convergence point for the photoperiod and autonomous pathways, we wanted to check whether it was also a target for phytochrome. SOC1 was expressed to a similar low level in the wild type and the phytochrome mutants at 16°C. Under 22°C levels were slightly, but sequentially elevated in the phyB and phyAphyBphyD mutants, respectively. These modest rises in SOC1 expression correlate with the degree of early flowering observed in the mutants. Therefore, it is possible that SOC1 levels contribute to the early flowering phenotype at 22°C. The most impressive temperature-dependent response to phytochrome status was observed in FT transcript levels, where for plants grown at 22°C transcript abundance was 20-fold higher in phyB and 40-fold higher in phyAphyBphyD, compared with the wild type. These changes in FT expression correlate with the phyB and phyAphyBphyD early flowering phenotypes providing indirect evidence that phyB, and other phytochromes, delay flowering through repression of FT expression at 22°C. This contrasts with the situation seen for plants grown at 16°C, where FT transcript levels were generally lower and more similar among the different genotypes. Thus, at 16°C the observed repression of FT expression may have occurred as a direct consequence of temperature. Alternatively, FT transcription may have been repressed by photoreceptor action in the wild-type and mutant plants. Our results suggest such a role for phyE. When grown at 16°C, the early flowering phyAphyBphyDphyE quadruple mutant had sixfold higher levels of FT expression than phyAphyBphyD (Figure 7). Hence, there is a positive correlation between phyE loss, acceleration of flowering, and FT levels. This provides indirect evidence that at 16°C phyE inhibits flowering in the phyAphyBphyD mutant via repression of FT.

Figure 7.

Expression levels of FT in phyAphyBphyDphyE relative to phyAphyBphyD at 16°C. Plants were harvested at 22 days, and expression levels were measured by quantitative PCR. All plants were grown at 16°C (±1) in 8-h photoperiods, photon irradiance, 400–700 nm, 180 µmol m−2 sec−1.

PhyB and phyD control the rate of vegetative development

In wild-type seedlings, development through the juvenile phase, during which the first four leaves are formed, is slow. Thereafter, leaves are produced at an increasing rate until approximately two-thirds of the way through the vegetative phase when leaf production slows prior to bolt emergence (Figure 8). Although the overall rate of development is slower in plants grown at the cooler temperatures, this pattern of development is consistent for plants grown in 22 or 16°C.

Figure 8.

Rosette leaf production rate in wild type, phyA, phyB, phyAphyB, phyAphyBphyD, and phyAphyBphyDphyE mutants.

Rosette leaf number was counted at time intervals (days) until flowering time in plants grown in 8-h photoperiods (photon irradiance, 400–700 nm, 180 µmol m−2 sec−1), at either 22°C (±1) or 16°C (±1). Standard errors are shown. Nomenclature: A = phyA; B = phyB; D = phyD and E = phyE null mutations.

In accordance with previous observations (Mazzella et al., 2001) we demonstrate that loss of phyB significantly slowed the rate of rosette leaf production (Figure 8). In addition, we have demonstrated that this effect was not dependent on temperature as phyB mutants produced rosette leaves more slowly than the wild type when grown at either 22 or 16°C (Figure 8). However, when phyB was grown at the cooler temperature its leaf production rate markedly accelerated just prior to bolting. It was as a result of this final burst of rapid development, prior to bolting, that the otherwise slowly developing phyB mutant flowered with a similar number of leaves as the wild type. The phyAphyB double mutant produced leaves at an identical rate to the phyB mutant. This and the lack of a phyA mutant phenotype in this respect suggest that phyA does not contribute significantly to the control of leaf production rate.

When compared with wild-type plants, leaf production was also slowed in the phyAphyBphyD mutant following growth at either 22 or 16°C (Figure 8). However, growth at 16°C revealed a role for phyD in the control of leaf formation rate. At the cooler temperature, the phyAphyBphyD mutant exhibited a reduced rate of development throughout the vegetative phase. This, coupled with the abolition of the early flowering phenotype of phyAphyBphyD means that these mutants flowered extremely late (at 90 days) at 16°C. In a similar fashion to plants grown at 22°C, phyAphyBphyD mutants grown at 16°C produced leaves at the same rate as phyB and phyAphyB, but just for the first half of the vegetative phase. During the second half of the vegetative phase the rate of leaf production in the triple mutant was slower than that of the phyB and phyAphyB mutants. Indeed, although phyB and phyAphyBphyD flowered with a similar number of primary rosette leaves, phyAphyBphyD flowered some approximately 17 days later than phyB. This delayed development must be due to the action of the phyD mutation acting to repress leaf production rate in the second half of the vegetative phase. As the phyAphyBphyDphyE quadruple mutant produced leaves at an even slower rate than the phyAphyBphyD mutant, this reveals redundant actions of phyA and/or phyE in control of leaf formation.

Discussion

As the reproductive success of the plant hinges on the decision of when to flower, it is no surprise that the flowering-signalling network is intricate. However, even though this developmental decision is subject to a high level of control, these pathways converge at common components: the key regulators of flowering (Simpson and Dean, 2002). We have shown that the phytochrome-controlled flowering pathway is subject to temperature regulation and that it appears to act independently of the autonomous-FLC pathway.

In the natural environment it is the explicit role of the phytochromes to adjust the plants developmental programme to maximise reproductive success in response to changes in the local environment. The phytochromes can regulate flowering, elongation growth and apical dominance in response to the R:FR ratio of ambient light. This complex of responses, which comprise components of the shade-avoidance response, enables the plant to adjust its developmental programme in response to potential competitors. Of the phytochromes, it is phyB that exerts the greatest control on flowering, although phyA, phyD and phyE each contribute to varying extents (Devlin et al., 1998; 1999; Halliday et al., 1994).

We have shown that the much reported early flowering phenotype of the phyB mutant is temperature conditional. The dramatic acceleration in flowering observed in phyB at 22°C was completely abolished at 16°C where the phyB mutant flowered at a similar time as the wild type. Comparable observations were also made for the phyAphyBphyD triple mutant, which flowered very early at 22°C, but with a similar number of rosette leaves as the wild type at 16°C. In contrast, the elongated petiole phenotype of phyB and phyAphyBphyD was observed in plants grown at either 22 or 16°C. This suggests that control of this aspect of the shade-avoidance phenotype is independent of temperature, consistent with the notion that discrete pathways control the elongation and flowering components of the shade-avoidance response.

Little is known of how the phytochromes control flowering in context with the established flowering pathways. By focussing on key components and integrators of the flowering network, we have gained valuable insights into how the phytochromes influence flowering. For many genes that regulate flowering their activity strongly correlates with their levels of expression. One such gene is FLC, an important vernalization-responsive gene and floral integrator (Simpson and Dean, 2002).

Earlier work by Bagnall (1993) demonstrated that either vernalization or reduced R:FR ratio conditions could abolish the late-flowering phenotype of the autonomous pathway mutants fca, and fve. Furthermore, a low temperature treatment could eliminate the response to light quality, suggesting one or more convergence points for these signalling pathways. As FLC is an important common component of the vernalization and autonomous pathways it was possible that FLC was also an integration point for phytochrome-signalling (Michaels and Amasino, 1999; 2001; Sheldon et al., 1999). Furthermore, as vernalization results in the reduction of FLC mRNA, it was possible that FLC may also be subject to control by persistent changes in ambient temperature. Another potential target for phytochrome action in this response was FLM, which appears to regulate flowering independently of the autonomous-FLC pathway (Ratcliffe et al., 2001; Scortecci et al., 2001).

Our results showed the differential affects on flowering at 22 and 16°C in phyB and phyAphyBphyD did not result from changes in FLC or FLM expression. FLC and FLM transcript levels were low in all genotypes at 22 and 16°C. These data suggest that FLC and FLM expression do not contribute to this phytochrome-controlled flowering response. In support of these findings, we have recently shown that FLC levels do not respond to changes in R:FR ratio (data not shown). Thus, it appears that this temperature-sensitive phytochrome flowering pathway defines a pathway that acts independently of the autonomous-FLC pathways.

CO is an important link between the circadian clock and light, promoting flowering in inductive photoperiods (Putterill et al., 1995; Suarez-Lopez et al., 2001; Yanovsky and Kay, 2002). Under LDs the peak of CO expression is broader than under SDs with the highest levels of CO mRNA coinciding with dawn and dusk (Suarez-Lopez et al., 2001). This photoperiodic adjustment of CO mRNA, which results from the coincidence of light and the circadian phase, is important for induction of flowering under LDs (Yanovsky and Kay, 2002). Growth in photoperiods with incandescent light extensions revealed a role for phyA in controlling the waveform and levels of CO mRNA (Yanovsky and Kay, 2002). We therefore wanted to establish if there were differences in CO expression in our temperature-sensitive phytochrome mutants. Our data indicate that whilst CO transcript levels were slightly elevated in phyB, they were similar in phyAphyBphyD mutant and wild-type seedlings under 22 and 16°C. Furthermore, the moderately raised CO mRNA levels observed in phyB did not correlate with the flowering behaviour of the mutant. These observations suggest that this temperature-sensitive phytochrome pathway does not operate via control of CO transcription. However, our data do not exclude the possibility that CO has a post-transcriptional role in these flowering pathways. Indeed, such a role is proposed in the phyA- and cry2-mediated photoperiodic flowering response (Yanovsky and Kay, 2002).

The myriad of pathways that control flowering time and flower meristem identity converge on floral integrators such as FT and SOC1/AGL20 (see Simpson and Dean, 2002). Indeed, both FT and SOC1 are common targets for the automomous-FLC and the photoperiodic pathways (Lee et al., 2000; Rouse et al., 2002; Samach et al., 2000; Yanovsky and Kay, 2002). Recently, the mechanism via which the automomous-FLC and the photoperiodic-CO pathways control SOC1 has been demonstrated (Hepworth et al., 2002). We set out to test whether SOC1 and/or FT were also targets for the temperature-controlled phytochrome-mediated flowering pathway. In our experiments, SOC1 expression was elevated 3- and 4.4-fold in the phyB and phyAphyBphyD mutants, respectively, at 22°C. These data do not support a prominent role for SOC1 in this response, although they do not eliminate a post-transcriptional role. In contrast, FT expression was greatly enhanced in phyB and phyAphyBphyD at 22°C. Relative to the wild type, FT levels were 19-fold higher in phyB and 40-fold higher in phyAphyBphyD. This rise in FT transcript correlates with the observed early flowering phenotypes of the mutants. Previous work has demonstrated that FT mRNA accumulation correlates with early flowering (Kardailsky et al., 1999; Kobayashi et al., 1999). Thus, our data suggest that the early flowering of these phytochrome mutants under 22°C is due, at least in part, to reduced FT repression.

When the ambient temperature was reduced to 16°C, the phyB and phyAphyBphyD mutants no longer flowered early and FT expression was low. Loss of phyE accelerated both phyAphyBphyD mutant flowering and triggered an elevation in FT transcript. This demonstrates that the suppression of FT at cooler temperatures was not solely due to the effects of temperature, as FT repression was relieved by loss of phyE. These data provide evidence in support of a role for FT in the temperature-dependent induction of flowering mediated by the phytochromes. Our observations are interesting in context with the recent findings that cry2 controls flowering, at least partly, by the repression of phyB action (Mockler et al., 1999). As cry2 regulates FT expression in a CO-dependent fashion, it is quite conceivable that these photoreceptor pathways converge at CO and/or FT.

The early flowering phenotypes of the phyB and phyAphyBphyD mutants when grown at 22°C suggests that phyB, and to a lesser extent phyA and phyD, antagonise flowering promoted by warm temperatures. This appears to occur, at least partly, via repression of FT. Under cooler temperatures, the phyB and phyAphyBphyD mutants are no longer early flowering and FT expression is low. Thus, at 16°C, in these genetic backgrounds, the phyB mutation had no impact on flowering. As removal of phyA and phyD in addition to phyB did not alter flowering this suggests that phyA and phyD are either inactive at 16°C, or they have a functional requirement for phyB and/or each other in this response. The removal of phyE in addition to phyA, phyB and phyD relieves the repression of FT and accelerates flowering. Furthermore, the delayed flowering imposed by phyE in the phyAphyBphyD mutant was not simply a result of altered phyE or phyC levels, as levels were identical in the triple mutant at 16 and 22°C. These data reveal a new and potentially important role for phyE signalling in the control of flowering at 16°C, conditions where phyB is less active. They are also in keeping with our observations that wild-type plants grown at 16°C show a normal accelerated flowering response to low red/far-red ratio light. Further analysis will establish whether other phytochromes in addition to phyE have a role in this temperature-regulated response. These findings provide an environmental context for ‘redundant’ actions of the phytochromes in the control of flowering. It appears that at 22°C phyB is the major regulator of low R:FR ratio-induced flowering, whilst at lower temperatures, phyE has the greater role. This type of accommodative behaviour exhibited by the phytochromes is not unusual in pathways that have been subject to strong selection pressure, such as flowering (Stearns, 2002).

A reduction in ambient growth temperature is accompanied by a slowing of plant development. However, we have demonstrated that phytochromes also regulate developmental rate, but between 16 and 22°C this effect is independent of temperature. In agreement with previous observations (Mazzella et al., 2001), we showed that a lack of phyB slowed the rate of leaf production, suggesting a prominent role for phyB in this respect. At 22°C roles for additional phytochromes were not apparent, as flowering was so rapid in plants lacking multiple phytochromes. At 16°C, the loss of the early flowering phenotype in mutants null for phyB allowed us to establish a role for phyD in regulating the rate of leaf production. The removal of phyD slowed leaf production rate in the phyAphyB mutant background, but only in the second half of the vegetative phase. Indeed, the reduced rate of vegetative development imposed by the phyD mutation in the phyAphyB background was sufficient to delay flowering by a further 20 days. These data reveal roles for phyB and phyD in the control of growth rate where they operate in overlapping phases of the vegetative development. The even more slowly growing phyAphyBphyDphyE quadruple mutant revealed minor roles for either phyA and/or phyE in this regulation. This represents the first evidence that the phytochromes act collectively to regulate the rate at which development proceeds.

Conclusion

Analysis of mutants null for one or more phytochrome species under different ambient growth temperatures has revealed that the phyB monogenic mutant early flowering phenotype is temperature dependent. Moreover, we have demonstrated that at the cooler temperature of 16°C phyE has a prominent role in the regulation of flowering. This temperature-sensitive flowering pathway appears to operate independently of the autonomous-FLC pathway. We have correlative data that suggest this newly defined thermosensory pathway functions at least partly through the regulation of FT. Thus, FT appears to be an important integration point for the photoperiod, the autonomous-FLC and the phytochrome-regulated flowering pathways. Not all the phytochrome-mediated responses are subject to temperature modification. Within the same temperature range (16–22°C) phytochrome control of leaf elongation and rosette leaf-formation rate was not substantially altered. This provides an insight into the divergent strategies that have evolved in phytochrome-signalling pathways controlling different responses.

Experimental procedures

Plant material and growth conditions

All the experiments described were performed with Arabidopsis Heynh, ecotype La-er. The phytochrome mutant alleles used were phyA-2 (Whitelam et al., 1993), phyB-1 (Koornneef et al., 1980), phyD-1 (Aukerman et al., 1997) and phyE (Devlin et al., 1998). As the phyD-1 mutation is a naturally occurring allele found in the Ws ecotype, near-isogenic La-er phyD-1 mutant lines were created by introgression of the phyD-1 mutation into the La-er ecotype as described by Aukerman et al. (1997).

In all experiments seeds were sown on 0.8% Lehle medium (Lehle Seeds, Round Rok, TX), and stratified in darkness at 4°C for 5 days before transfer to different photoperiods or continuous white light (for high/low R:FR ratio experiments) at 22°C (±1) or 16°C (±1). After a further 5 days, uniform seedlings were transplanted to 5 cm × 5 cm × 5 cm pots containing a 3 : 1 compost–horticultural silver–sand mix. In photoperiod experiments, plants were grown at either 22°C (±1) or 16°C (±1) under short days (8 h : 16 h light:dark cycles), or long days (16 h : 8 h light:dark). In high/low R:FR ratio experiments, seedlings were also grown at either 22°C (±1) or 16°C (±1). R:FR ratio treatments began after 1 day of adaptation. Light was provided by Osram (Osram Ltd., St. Helens, UK) L65/80 W/30 warm white fluorescent tubes, photon irradiance 400–700 nm, 180 µmol m−2 sec−1. Light conditions for plants grown under continuous light were as follows: high R:FR ratio at 22°C (±1), R:FR ratio = 6.1, photon irradiance 400–700 nm, 91 µmol m−2 sec−1; at 16°C (±1), R:FR ratio = 5.8, photon irradiance 400–700 nm, 99 µmol m−2 sec−1; low R:FR ratio at 22°C (±1), R:FR ratio = 0.11, photon irradiance 400–700 nm, 92 µmol m−2 sec−1, at 16°C (±1), R:FR ratio = 0.12, photon irradiance 400–700 nm, 99 µmol m−2 sec−1.

Plant growth assays

For plants grown under short-day photoperiods, rosette leaf counts were carried out twice a week. Leaves were counted only when the petiole was visible to the naked eye. Flowering time was recorded as primary rosette leaf number at inflorescence production. Rosette leaves were distinguished from axillary leaves on the basis of morphological differences.

Protein extraction and immunoblotting

Proteins were extracted as described previously by Devlin et al. (1992). After resolution on 8% SDS–polyacrylamide gels, proteins were electroblotted to an Imobilon-P polyvinylidene difluoride membrane (Millipore Corp., Bedford, MA). Membranes were probed with the following monoclonal antibodies: C1 and C13, specific for phyC (Somers et al., 1991); and 2C1, which is selective for phyD (Hirschfield et al., 1998). Protein bands were visualized by secondary incubation with horseradish peroxidase–anti-mouse immunoglobulin antibodies and chemiluminescence (Amersham International, Bucks., UK).

Quantitative RT-PCR

Plants were grown under short-day conditions and RNA samples were prepared from shoot tissue harvested 22 days post-germination. Tissue was harvested in either the middle of the photoperiod (FLC and FLM) or the middle of the dark period (CO, FT and SOC1). RNA was extracted using RNeasy miniprep kits (Qiagen) as per manufacturer's instruction and DNA removed with DNA-free (Ambion). RNA quantification was performed using a TD-360 minifluorometer (Turner Designs) with ribo-green RNA quantification system (Molecular Probes). A volume of 2 µg total RNA was placed in a 40-µl RT-PCR reaction (Qiagen Omniscript), two reactions per treatment.

Quantitative PCR was performed using Brilliant QPCR Core Reagent kit (Stratagene) and sybr-green (Molecular Probes) on the MX4000 Multiplex Quantitative PCR System (Stratagene). For each gene PCR was optimised for MgCl2 concentration, relative primer concentration and annealing temperature to give linearity using a standard curve from a dilution series as per manufacturer's instructions. Samples from the PCR reactions were gel analysed using an Agilent Bioanalyser to ensure that only single PCR products were produced. For each gene MgCl2 concentration at 2 mm, a primer concentration of 150 nm and an annealing temperature of 54°C proved optimal PCR conditions. Actin 2 was used as a control gene (Ratcliffe et al., 2001).

Gene Primer sequence
ACTIN2 FTCAGATGCCCAGAAGTCTTGTTCC
RCCGTACAGATCCTTCCTGATATCC
FLC FCTTGTGGATAGCAAGCTTGT GGG
RCATGAGTTCGGTCTTCTTGGCTC
FT FTACGAAAATCCAAGTCCCACTG
RAAACTCGCGAGTGTTGAAGTTC
FLM FCTTGAGACTGCTCTGTCCGTAAG
RCCAGAACCTGGTTCTCTTCTCTC
SOC1 FCGAGCAAGAAAGACTCAAGTGTTTAAGG
RTTCATGAGATCCCCACTTTTCAGAGAG
CO FGACCACTCTACTCACCACCAAAG
RCAACCTCCTTGGCATCCTTATC

Relative expression levels were determined using the Comparative CT method (User Bulletin 2, ABI PRISM Sequence Detection System, pp. 11–15, 1997, PE Applied Biosystems).

Acknowledgements

We thank Wendy Stoddart for technical assistance and the BBSRC for financial support. We also wish to thank John Butler, Eugene Halligan and Joe Lunec (Oxidative Stress Group, Department of Clinical Biochemistry, University of Leicester) for helpful assistance with QPCR.

Ancillary