Interaction of phytochromes A and B in the control of de-etiolation and flowering in pea

Authors


For correspondence (fax +31 317 484740; e-mail Jim.Weller@algem.pf.wau.nl, or fax +61 362 262698; e-mail Jim.Reid@utas.edu.au).

Summary

The interactions of phytochrome A (phyA) and phytochrome B (phyB) in the photocontrol of vegetative and reproductive development in pea have been investigated using null mutants for each phytochrome. White-light-grown phyA phyB double mutant plants show severely impaired de-etiolation both at the seedling stage and later in development, with a reduced rate of leaf production and swollen, twisted internodes, and enlarged cells in all stem tissues. PhyA and phyB act in a highly redundant manner to control de-etiolation under continuous, high-irradiance red light. The phyA phyB double mutant shows no significant residual phytochrome responses for either de-etiolation or shade-avoidance, but undergoes partial de-etiolation in blue light. PhyB is shown to inhibit flowering under both long and short photoperiods and this inhibition is required for expression of the promotive effect of phyA. PhyA is solely responsible for the promotion of flowering by night-breaks with white light, whereas phyB appears to play a major role in detection of light quality in end-of-day light treatments, night breaks and day extensions. Finally, the inhibitory effect of phyB is not graft-transmissible, suggesting that phyB acts in a different manner and after phyA in the control of flower induction.

Introduction

Photoperiod length is an important environmental determinant of flowering time in many plant species. Work in several species has shown that photoperiod is detected in leaves, and under the appropriate inductive conditions a signal is transmitted to the apex where flowering is initiated (Thomas and Vince-Prue, 1997). Photoperiodic flower induction is therefore a complex process, and there is considerable interest in how light is perceived, how time is measured, and how flowering signals are transmitted within the plant.

Members of the phytochrome family of red-(R) and far-red (FR)-light-absorbing photoreceptors play a dominant role in the regulation of flowering in many species (Thomas and Vince-Prue, 1997). In fact, induction of flowering was one of the first responses for which the phenomenon of R/FR reversibility, diagnostic for phytochrome action, was demonstrated (Borthwick et al., 1952). Classical physiological studies of photoperiodism have attempted to dissect the complex relationship between the wavelength, duration, or timing of light treatments and their effectiveness for floral induction. For example, rhythms in light responsiveness indicate that photoperiodic time measurement is achieved by the interaction of photoreceptor signaling pathways with the circadian clock, while the existence of distinct rhythms in response to R and FR light has suggested that more than one phytochrome may contribute to the regulation of flowering (Thomas and Vince-Prue, 1997).

The roles of specific phytochromes throughout plant development have been explored by a mutational approach in several species, including Arabidopsis, tomato and pea (Weller et al., 1997a; Kendrick et al., 1997; Lin, 2000; Neff et al., 2000; Weller et al., 1995). These studies have shown that phytochrome B (phyB)-type phytochromes act as R sensors which can be reversibly inactivated by FR, and play a dominant role in the regulation of seedling de-etiolation under high-irradiance R. PhyA is also active in mediating de-etiolation responses to low-irradiance R, and in both Arabidopsis and tomato there are at least three phytochromes (phyA and two phyB-type phy) that contribute to de-etiolation under continuous R (Aukerman et al., 1997; Mazzella et al., 1997; Weller et al., 2000a). In contrast, the photomorphogenic effects of FR are uniquely mediated by phyA (van Tuinen et al., 1995; Weller et al., 1997a; Whitelam et al., 1993).

The role of specific phytochromes in flowering has been mainly investigated in Arabidopsis and in pea, since tomato is not suitable for studies of photoperiodism. Both Arabidopsis and pea are long-day plants (LDP) and flower earlier in long photoperiods (LD) than under short photoperiods (SD). In both species, the effects of phyA and phyB on flowering have been examined using mutants specifically deficient in these phytochromes. Results from these studies agree in identifying a promotive function for phyA (Johnson et al., 1994; Weller et al., 1997a), and inhibitory function for phyB (Goto et al., 1991; Weller et al., 1995). The main difference between the two species is that in pea, phyA is the main photoreceptor controlling the promotion of flowering under LD, whereas in Arabidopsis, phyA plays a relatively minor role, and the blue-light photoreceptor cryptochrome 2 is much more dominant (Guo et al., 1998).

Downstream processes in photoperiodic flower induction have also been explored by mutational approaches. In pea, mutants affecting photoperiod responsiveness have been characterised primarily at the physiological level, in terms of their effect on a physiologically defined graft-transmissible inhibitor of flowering. Several mutants which confer earlier flowering and reduced photoperiod sensitivity are impaired in the production or transport of this inhibitor (Murfet, 1977; Weller et al., 1997b). In addition, the late-flowering phenotype of the phyA-deficient mutant was shown to result from loss of ability to down-regulate inhibitor under inductive conditions (Weller et al., 1997a). The effect of phyB on mobile flowering signals in pea is not yet known. Genes required for normal photoperiod responsiveness have also been identified in Arabidopsis (Koornneef et al., 1998; Simpson et al., 1999). Several of these genes have now been cloned, and while this work has begun to outline a signalling pathway (Samach et al., 2000; Simpson et al., 1999), it has so far provided no clear evidence about the molecular or biochemical function of individual genes, or the mechanisms of leaf-to-apex signalling.

Although there are clear differences in the utility of pea and Arabidopsis for a genetic analysis of photoperiodism, the existence of phyA- and phyB-deficient mutants in both species provides the possibility for a direct comparison between the two systems. In this study we have identified a null mutant for the pea PHYB gene. We have used this mutant in combination with a null mutant for PHYA to examine in more detail the roles and interaction of these two phytochromes in the control of various aspects of light-regulated development, including de-etiolation, shade-avoidance and photoperiodic flower induction. We have also investigated whether the influence of phyB on flowering is transmissible across a graft junction.

Results

Isolation of the PHYB gene from pea

Mutants at the lv locus are deficient in a PHYB apoprotein detected by monoclonal antibody mAP11, which was raised against pea phytochrome II (Abe et al., 1989; Weller et al., 1995). To address the possibility that LV is a structural gene for PHYB, we first screened a cDNA library prepared from light-grown pea shoot apices with a fragment of the Arabidopsis PHYB gene. One of the resulting clones showed a high degree of sequence similarity to other PHYB genes but was truncated at the 5′ end, and the missing sequence was obtained by genome walking. Conceptual translation of the 3.46 kb open reading frame yielded an apoprotein sequence of 1152 amino acids in length and a predicted molecular mass of 129 kDa, which showed highest identity with PHYB sequences from soybean (84%), tobacco (82%), tomato (PHYB1; 80%), potato (80%) and Arabidopsis (76%). A neighbour-joining cluster analysis grouped the pea and soybean sequences separately from PHYB pairs in Arabidopsis and tomato (Figure 1a). Perfect agreement of the pea PHYB sequence with the sequence of phy II peptide fragments (Abe et al., 1989) supported the conclusion that this gene encodes the PHYB apoprotein detected by mAP11.

Figure 1.

Molecular characterisation of phyB-deficient lv mutants of pea.

(a) Dendrogram showing relationship of pea PHYB with other phytochromes.

(b) 10-day-old white-light-grown seedlings of wild-type (WT) cv. Torsdag and the lv-5 and lv-6 mutants.

(c) Immunoblot detection of PHYB in WT cv. Torsdag, lv-5 and lv-6. PHYB was detected using monoclonal antibody mAP11. Lanes were loaded on an equivalent fresh weight basis.

(d) Location of lv mutations within the PHYB cDNA. Nucleotide substitutions and predicted amino acid changes are indicated. Vertical divisions represent exon boundaries.

GenBank accession numbers for other sequences are as follows, Glycine max (Gm) PHYB (L34843), Solanum lycopersicum (Sl) PHYB1 (AJ002281) and PHYB2 (AF122901), Arabidopsis thaliana (At) PHYB (X17342), PHYD (X76609), PHYE (X76610) and PHYA (X17341).

LV is the structural gene for phyB

Of four lv mutants described previously, three are strongly deficient in the apoprotein detected by mAP11, whereas the original lv-1 allele has WT levels of the protein (Nagatani et al., 1990; Weller et al., 1995). However, none of these alleles are in the standard WT cv. Torsdag background, and an EMS-mutagenised M2 population of cv. Torsdag was therefore screened under continuous R to identify additional lv alleles. We selected two new alleles, lv-5 and lv-6, with a strong and a weaker phenotype, respectively (Figure 1b). The lv-5 mutant is completely deficient in PHYB, whereas lv-6, like lv-1, had WT levels of the apoprotein (Figure 1c). In all three alleles (lv-1, lv-5 and lv-6) single G-to-A nucleotide substitutions were identified within the PHYB coding sequence (Figure 1d), and verified by resequencing from both cDNA and genomic DNA. The lv-5 molecular lesion created a novel MnlI restriction site that showed perfect cosegregation with the elongated phyB-deficient phenotype in the F2 of a cross between lv-5 and WT cv. Torsdag (not shown). These results show that the LV gene encodes PHYB, and lv mutants will subsequently be referred to as phyB.

The phyB-5 mutation introduces a stop codon at position 1674 and is predicted to result in a truncation of the PHYB protein at residue 558 (Figure 1d). This would eliminate the entire C-terminal domain which has been shown to be critical for phyB regulatory activity in Arabidopsis (Wagner et al., 1996). In addition, the Arabidopsis phyB-5 mutation (hy3–8.36) introduces a premature stop codon at residue 552 (residue 547 in PsPHYB), and is widely used as a null allele (Cantón and Quail, 1999; Devlin et al., 1999; Neff and Chory, 1998; Poppe et al., 1998). These comparisons suggest that the phyB-5 mutant is effectively null for phyB. The phyB-1 and phyB-6 mutations are both located at the 5′ end of the PHYB gene, and are, respectively, predicted to result in the substitution of asparagine 57 with aspartic acid and arginine 100 with histidine (Figure 1d). Asn57 is located in the first conserved section of the PHYB apoprotein after the highly divergent N-terminal extension, and is the second residue after Gln53 that is invariant among all available full-length phytochrome sequences (not shown). Arg100 is not invariant but is conserved among phyB-type phytochromes.

Isolation of a double mutant deficient in phyA and phyB

In order to explore the interaction of these two photoreceptors, we constructed a double mutant between phyB-5 and the phyA-deficient mutant fun1-1 (Weller et al., 1997a). This mutant carries a lesion in the PHYA gene that predicts complete loss of phyA function (S. L. Batge, unpublished data), and fun1 mutants will subsequently be referred to here as phyA. The double mutant was identified as a discrete class of segregant in the F2 of a cross between phyA-1 and phyB-5 mutants, and its identity was confirmed in F3 progeny. Double mutant seedlings showed severely reduced de-etiolation compared with either single mutant parent. They retained a partial apical hook in WL for more than a week after emergence, had reduced levels of chlorophyll in the stem and showed greatly retarded opening and expansion of leaflets and stipules (Figure 2a). Interestingly, internodes in the double mutant were somewhat shorter than in the monogenic phyB mutant (Figure 2a). The apical bud was smaller and less vigorous and the rate of node production was slower in the double mutant than in WT or either single mutant (Figure 2a). The etiolated characteristics seen in double mutant seedlings persisted in older plants (Figure 2b). Later internodes and petioles remained almost white and very brittle, and began to thicken and distort several nodes below the node of flower initiation, with severe twisting and splitting of the stem seen in a majority of plants (Figure 2b). Double mutant plants flowered relatively normally, although thickening and twisting were also seen in peduncles, pedicels and pods. In some cases, root-like outgrowths began to form on the lower surface of peduncles. The double mutant yielded fewer than 10 seeds per plant, compared with approximately 30 in WT and phyB, and more than 50 in phyA (Weller et al., 1997a).

Figure 2.

Phenotype of a phyA phyB double mutant.

(a) 12-day-old seedlings grown in continuous white light.

(b) Shoot apex and upper internodes of eight-week-old phyA and phyA phyB plants.

(c) Size of individual epidermis, cortex and stele cells (estimated as cell length × breadth) in transverse sections taken from the internode above node 16 of WT, phyA, phyB and and phyA phyB double mutant plants.

Values represent mean ± SE of the means of measurements from 10 cells from each of four plants.

The cellular basis for the morphological abnormalities in the double mutant was examined in transverse stem sections taken from internode 16, in which the phenotype was best developed. Both epidermal cells and parenchyma cells of the stele and cortex were greatly enlarged in the phyA phyB double mutant relative to the WT or either single mutant (Figure 2c). This was associated with a reduction in the degree of secondary thickening. Vascular tissue in the double mutant was also poorly developed, with vascular bundles containing fewer xylem elements and a much reduced fiber cap (not shown). Chloroplasts were abundant in WT cortical cells but were almost completely lacking in comparable cells in the double mutant (not shown).

Roles of phyA and phyB in de-etiolation and shade-avoidance

We next investigated the interaction of phyA and phyB in the control of de-etiolation under monochromatic light. As expected, both phyA and phyA phyB mutants failed to de-etiolate under cFR while the phyB mutant de-etiolated normally (Figure 3). It was shown previously that the loss of phyA or phyB has only a relatively small effect on de-etiolation under R, and both the phyA and phyB mutants retained strong R responses (1997a; Weller et al., 1995). However, the double mutant showed a dramatic reduction in responsiveness to R for both leaf expansion and inhibition of stem elongation (Figure 3) and there was no clear evidence for residual responsiveness to R for either character in the double mutant. Under BL, both phytochromes contributed to the stimulation of leaf expansion, although the effect of phyA was stronger than that of phyB. The double mutant showed an additive phenotype, with smaller leaflets than either single mutant. The loss of phyB resulted in a slight increase in internode length, suggesting that phyB acts to inhibit elongation under BL. However, the loss of phyA in either a WT or a phyB background caused a reduction in elongation, confirming the previous observation that phyA in pea acts to promote elongation under BL (Weller et al., 1997a).

Figure 3.

De-etiolation of the phyA phyB double mutant under continuous monochromatic light.

Internode length and leaf area in wild type (WT) cv. Torsdag, phyA, phyB and phyA phyB double mutant seedlings kept in complete darkness (D) or grown under continuous far-red (FR; 12 µmol m−2 s−1), red (R; 12 µmol m−2 s−1), blue (B; 12 µmol m−2 s−1) or white (WL; 100 µmol m−2 s−1) irradiation for two weeks after sowing. Leaflet area was estimated as length × width of the largest leaflet from leaf 3. Values represent mean ± SE, n = 12.

WT pea plants were previously shown to have a relatively small shade-avoidance response, with an increase in internode length of around 20% upon lowering of the R:FR ratio from 5.4 to 0.7 (Weller et al., 1995). In contrast, two strong alleles phyB-1 and phyB-4 exhibited a negative response to this change in R:FR ratio, which was attributed to an inhibitory action of phyA under the lower R:FR ratio (Weller et al., 1995). To test this conclusion, we conducted a similar experiment with phyA-1 and phyB-5 mutants and the double mutant. The results confirmed that WT cv. Torsdag pea seedlings show only a small shade-avoidance response (Figure 4). The loss of phyA conferred a substantial increase in the shade-avoidance response, while the loss of phyB again resulted in a negative response, for both stem elongation and leaf expansion (Figure 4). The double mutant showed poor leaf expansion under both high and low R:FR, comparable to that seen in phyB plants under high R:FR, and a minimal negative response for internode elongation.

Figure 4.

Shade-avoidance responses in the phyA phyB double mutant.

Plants were grown for three weeks from sowing under an 18-h photoperiod of white light (150 µmol m-2 sec-1) either with (WL + FR; R:FR = 0.08) or without (WL; R:FR = 6.28) supplementary far-red light given for the duration of each daily photoperiod. Leaflet area was estimated as length × width of the largest leaflet from leaf 4. Values represent mean ± SE, n = 7.

Roles of phyA and phyB in the control of flower initiation

PhyA in pea has been shown to promote flowering in response to photoperiod extensions, whereas phyB inhibits flowering under non-inductive SD (Weller and Reid, 1993; Weller et al., 1997a). We examined their interaction by growing the phyA phyB double mutant under both SD and LD conditions in a phytotron, where an 8-h SD of natural daylight was extended with 16 h of low-irradiance white light from a combination of fluorescent and incandescent sources, to provide the LD. The results in Figure 5 show that the double mutant flowers relatively early and at the same node under both photoperiods. Thus, in the absence of the promotive action of phyA, a strong inhibitory effect of phyB is revealed under LD as well as SD, suggesting that the inhibitory action of phyB is independent both of photoperiod and of phyA. In contrast, in the phyB-deficient background, phyA has only a minimal effect, which is slightly larger in LD than SD (Figure 5).

Figure 5.

Photoperiod response of the phyA phyB double mutant. Node of flower initiation in plants grown from sowing under an 8-h photoperiod of natural daylight (SD) or under SD extended with 16 h of low-irradiance (10 µmol m−2s−1) tungsten-filament light (SD + 16WI). The early day-neutral mutant dne is included for comparison. Values are mean ± se for n = 10–12.

It is of note that the phyA phyB double mutant, although early flowering and effectively day-neutral, still flowers substantially later than the inhibitor-deficient day-neutral mutant dne(Figure 5; King and Murfet, 1985). However, triple mutant phyA phyB dne plants flowered as early as the dne single mutant (not shown), suggesting that the flower inhibitor is still active in the phyA phyB double mutant. Introduction of the dne mutation was previously shown to overcome many of the effects of the phyA mutant in older plants (Weller et al., 1997a), but introduction of dne had no effect on the etiolated phenotype of the phyA phyB mutant (not shown).

We next examined the contribution of phyA and phyB to specific flowering responses previously identified in studies of LDP photoperiodism. For these experiments WT, phyA and phyB plants were grown in growth cabinets under an 8-h photoperiod of cool-white fluorescent light, extended with various exposures to low-irradiance, FR-rich tungsten-filament light (WI) or far-red-free white light from fluorescent tubes (WF). We also included the phyA phyB double mutant in these experiments. However, these plants were extremely weak, and only survived to flower in one of the experimental conditions (16 h WF day-extension). The irradiance of the main photoperiod given in these experiments was considerably lower than in the phytotron experiments (Figure 5), where the 8 h photoperiod was given as natural daylight. It is therefore probable that the poor growth of the double mutant in the growth cabinets resulted from photosynthetic limitation, although a light-quality effect cannot be excluded. Figure 6(a) shows that WT plants did not respond to a 30-min end-of-day (EOD)-WI treatment, but responded similarly to a 16-h extension given as either WI or WF. PhyA-deficient phyA mutants flowered much later than WT in the 8 h WF SD conditions. This is in contrast to experiments in which the SD were given with daylight and the phyA mutant flowered only slightly later than or at the same node as WT (Figure 5; Weller et al., 1997a). The loss of phyA revealed a strong promotion in response to the 30 min EOD-WI treatment, and a large difference in effectiveness of the 16 h WI and WF day extensions (Figure 6a). Consistent with previous reports, the phyB mutant flowered early and showed only minimal response to different light treatments. Figure 6(b) shows that WT plants exhibited a slightly stronger promotion of flowering in response to the daily interruption of a 16-h night with 1 h of WF than with WI light. In the absence of phyA, plants no longer responded to the WF NB, but still showed a clear promotion in response to the WI NB (Figure 6b). Relative to WT, phyB-deficient plants showed a greatly reduced but still significant response to both NB treatments.

Figure 6.

Effect of phyA and phyB on the promotion of flowering in response to day-extensions, night breaks and end-of-day light treatments.

Node of flower initiation was recorded from plants grown in growth cabinets under an 8-h photoperiod of cool-white fluorescent light (SD; 100 mol m−2 s−1) and subjected to additional daily light treatments with low-irradiance (10 µmol m−2 s−1) tungsten-filament (WI; R:FR = 0.6), or cool-white fluorescent light (WF; R:FR = 4.8). Values are mean ± se for n = 10–12.

(a) Response to 30 min end-of-day WI (0.5WI EOD), and 16 h day-extensions with either WI (16WI) or WF (16WF).

(b) Response to a 1-h night-break with either WI (1WI NB) or WF (1WF NB) given in the middle of the 16-h night. The 16WF treatment was included for comparison.

The inhibition of flowering by PHYB in SD is not graft-transmissible

In pea, the promotion of flowering by phyA in LD is mediated by a reduction in the production or transport of a graft-transmissible inhibitor of flowering (Weller et al., 1997a). The effect of phyB with respect to graft-transmissible signals has not yet been examined. One possibility could be that the early flowering of the phyB-deficient mutants in SD might result from a reduction in inhibitor production in leaves or transport to the apex. Alternatively, phyB might act in the apex, by modifying perception of mobile flowering signals or by directly regulating inflorescence identity genes. We therefore tested whether phyB deficiency caused any alteration in graft-transmissible inhibition of flowering.

As in previous experiments with the phyA mutant (Weller et al., 1997a), inhibitor-deficient dne mutant shoots were used as ‘reporter’ scions for assessing the effect of mobile signals from the stock. Because phyB inhibits flowering we looked for a reduction in the ability of phyB stocks to delay flowering under non-inductive conditions (that is SD), in contrast to the experiments with the phyA mutant where we looked for an increased ability of phyA stocks to delay flowering under inductive (LD) conditions. The previous experiments with the phyA mutant could be performed using stocks with only a few leaves, because the WT control stocks had no inhibitory ability. However, in the present experiments it was necessary to maximise the inhibitory effect of the WT stocks in order to improve the chance of detecting any reduction in the phyB mutant by comparison. Older plants possessing 8 or 9 leaves were therefore used as stocks.

Figure 7 shows that in SD conditions (8 h natural daylight) ungrafted phyB plants flowered five nodes earlier than WT, at around node 17. Ungrafted dne plants flowered at node 11 or 12. Self-grafting of dne scions to dne stocks had no significant effect on flowering. WT stocks delayed flowering of dne scions by 2–3 nodes, and phyB stocks were at least as effective as WT in delaying flowering of the dne scions (Figure 7). We repeated this experiment three times, with small variations in the age of the stocks, but in all cases the results were extremely similar and in no case were phyB stocks less effective than WT in delaying flowering in dne scions. This result suggests that phyB, in contrast to phyA, does not affect flowering by altering the level of mobile flowering signals.

Figure 7.

The inhibitory effect of phyB on flowering is not graft-transmissible.

Epicotyls of 6-d-old dne seedlings (scions) were wedge-grafted into the ninth internode of 26-d-old wild-type (WT), phyB or dne mutant seedlings grown from sowing under an 8-h photoperiod of natural daylight. Values represent mean ± SE for n = 12–18.

Discussion

Novel missense mutations in the N-terminal region of pea phyB

Two PHYB apoprotein-positive phyB mutants both carry missense mutations near the N-terminal end of the phyB molecule. The phyB-1 mutation, which has a phenotype as strong as the phyB-5 null mutation, does not alter the level or spectral activity of phyB (Nagatani et al., 1990), whereas the phyB-6 mutation has a weaker phenotype and its effect on phyB spectral activity is not known. Missense regulatory mutations have been identified in both Arabidopsis phyB and phyA (Wagner and Quail, 1995; Xu et al., 1995). However, these cluster mainly in the C-terminal domain, and none have so far been reported for the extreme N-terminal part of either molecule. The strong phenotype conferred by the substitution of Asn57 in the phyB-1 mutant is somewhat surprising given that deletion of a small N-terminal region of Arabidopsis phyB containing the corresponding residue (Asn64) only resulted in a mild reduction in activity (Wagner et al., 1996). However, the conservation of this residue in the available full-length phytochrome sequences from both higher and lower plants does suggest that it may be essential for an aspect of phytochrome function common to all phytochromes. In addition, the region between residues 58 and 90 of Arabidopsis PHYB includes 15 residues invariant across available full-length PHYB sequences, further emphasizing the likelihood that this region is important for phyB function.

Roles of phyA and phyB in de-etiolation and vegetative development

PhyB-deficient mutants of pea exhibit a strong residual response to R (Weller et al., 1995). The present results confirm that phyA also contributes significantly to R responses. This has also been reported for Arabidopsis and tomato (Mazzella et al., 1997; Weller et al., 2000a), but the relative contribution of phyA in pea is clearly greater than in these other species. A strong mutual compensation was observed for phyA and phyB under R, with either phytochrome alone being able to induce 80% of the WT response for inhibition of stem elongation. This is not seen in tomato, where the effects of phyA on hypocotyl elongation are relatively small and essentially independent of phyB1 and phyB2 (Weller et al., 2000a). This difference may in part reflect the longer half-life for light-induced degradation of phyA in pea (100 min; J. L. Weller, unpublished data) compared with tomato and Arabidopsis (40 min; Hennig et al., 1999; Peters et al., 1992). Greater stability could also contribute to the stronger effect of pea phyA in light-grown plants compared with Arabidopsis (Johnson et al., 1994; Weller et al., 1997a).

A second phyB-type phytochrome also contributes to R sensing in both Arabidopsis (phyD) and tomato (phyB2), particularly when the dominant phyB is missing (Aukerman et al., 1997; Weller et al., 2000a). Also, in both species triple mutants deficient in phyA and both phyB-type phytochromes retain a strong shade-avoidance response (Devlin et al., 1999; Weller et al., 2000a), which in Arabidopsis is largely mediated by phyE (Devlin et al., 1998). However, our results show no clear evidence for residual phytochrome responses in the phyA phyB double null mutant, either in seedlings de-etiolated under continuous R, or in WL-grown seedlings exposed to differential R:FR. PCR-based surveys of phytochrome-related sequences in the Leguminosae have indicated that only four PHY sequences are generally present; two PHYA, one PHYB and one PHYE (Mathews and Sharrock, 1997). Thus it is possible that the PHYB duplication seen in other species is not represented in pea. PhyE may well be present in pea, but it is clear that if so, it does not play a substantial role in shade-avoidance. The presence of a second PHYA-related sequence in pea has been reported, but this is almost certainly a pseudogene (Sato, 1990).

Under BL, phyA makes a clear contribution to leaf expansion, which is again consistent with the role for phyA in BL reported for both Arabidopsis and tomato (Neff and Chory, 1998; Weller et al., 2001). The increased effectiveness of BL for the inhibition of stem elongation in the absence of phyA confirms a previous report (Weller et al., 1997a), but has not been described in other species. However, loss of phyA does result in an increased effectiveness of R for inhibition of hypocotyl elongation in Arabidopsis and for anthocyanin biosynthesis in tomato (Mazzella et al., 1997; Weller et al., 2000a), which in both cases is due to an interference of phyA with a phyB-type phytochrome. The residual leaf expansion response seen in the phyA phyB double mutant of pea clearly implies the action of a cryptochrome, and it may be that for stem elongation phyA may interfere in a similar way with this photoreceptor. Isolation of cryptochrome mutants in pea will help address this question.

The morphological abnormalities in WL-grown plants of the phyA phyB double mutant are more severe than those reported for the phyA phyB1 phyB2 mutant of tomato or the phyA phyB phyD mutant of Arabidopsis (Devlin et al., 1999; Weller et al., 2000a). The reduction in length of double mutant internodes is different in nature from that seen in plants suffering photosynthetic limitation, and suggests that when phytochrome signalling is sufficiently impaired, some other factor becomes limiting for elongation. The twisting of double mutant stems suggests a loss of co-ordination between cell elongation in the epidermis and in the cortex. A similar swelling and twisting of stem tissue is commonly seen when dark-grown pea seedlings are exposed to ethylene (for example Stewart et al., 1974), and several lines of evidence implicate ethylene in light responses. For example, phytochrome-mediated opening of the apical hook in pea has been associated with a reduction in ethylene evolution (Goeschl et al., 1967), and constitutive activation of ethylene signalling impairs the light induction of hook opening in Arabidopsis (Kieber et al., 1993). Also, the loss of phyB in Sorghum causes substantially elevated levels of ethylene due to an increase in the expression level of a circadially regulated ACC oxidase gene (Finlayson et al., 1999). It will be of interest to quantify ethylene and other interacting hormones such as auxin in the pea phyA phyB double mutant.

Roles of phyA and phyB in photoperiodic flower induction

The flowering response of the phyA phyB double mutant shows that the previously described inhibitory effect of phyB in SD is clearly apparent in the absence of phyA, and that phyB can also actively inhibit flowering under LD. In contrast, the promotive effect of phyA is greatly reduced in a phyB-deficient background. This differs somewhat from the essentially additive interaction seen for phyA and phyB in the control of flowering in Arabidopsis under LD (Neff and Chory, 1998). However, in Arabidopsis, phyD and phyE are both known to influence flowering in a similar way to phyB (Devlin et al., 1998, 1999), and the residual inhibition of flowering by these phytochromes in the phyB mutant may be sufficient to allow expression of the promotive effect of phyA.

Poor growth of the phyA phyB double mutant in the present study precluded a complete analysis of the roles of phyA and phyB in night-break and day-extension responses. Nevertheless, some useful information can be gained from results with the single mutants. Although flowering of WT plants was promoted to a similar extent by WI and WF, the loss of phyA revealed WI to be more effective than WF, whether given as an EOD treatment, NB, or day-extension. These results imply the presence of a phytochrome response, normally masked in the WT by the promotive effect of phyA, for which Pfr is inhibitory to flowering. The early flowering phenotype of the phyB mutant suggests that phyB is likely to mediate these responses. This predicts that it may be possible to detect a residual FR-induced, R-reversible NB effect in the phyA mutant, quite different in nature from the NB response previously reported by Reid (1982), which was induced by short exposures to R and partially reversed by FR. The complete loss of responsiveness to a 1-h WF NB in the phyA mutant suggests that no other phytochrome can promote flowering in response to short exposures to high R:FR light, and we suggest that the previously described R/FR-reversible NB response is likely to be a response to phyA. The small residual photoperiod response in the phyB mutant makes it difficult to test this directly, but it may be possible after transfer of the phyB mutant into the more responsive HR background (Reid, 1982).

These results are consistent with a model in which the promotion of flowering in pea results from the superimposition of two separate photoresponses – a promotion mediated by phyA predominantly under high R:FR light, and a promotion resulting from a reduction in PfrB-mediated inhibition under low R:FR light. In such a situation the most effective light for promotion of flowering would balance the activities of the two phytochromes, and could consist either of a combination of R and FR light or of monochromatic light at an intermediate wavelength. Action spectra for promotion of flowering by day extensions in LDP are somewhat variable, consisting most commonly as a single FR peak in the range 700–720 nm, and in other cases as distinct peaks in both the R and FR wavebands (Thomas and Vince-Prue, 1997). It is conceivable that this variation could reflect underlying differences in the inherent activity, relative abundance or spectral or dynamic characteristics of phyA and phyB. Additional complexity clearly derives from circadian clock-controlled rhythms in light-responsiveness (Thomas and Vince-Prue, 1997).

Although phyA appears to be the sole photoreceptor for a 30-min WF NB in WT pea, it is clearly not the only photoreceptor capable of promoting flowering in response to longer WF exposures, because a strong promotion by a 16-h WF day-extension was still seen in the phyA mutant. This response is unlikely to be due to phyB, because the loss of phyB results in a promotion of flowering under these conditions. It must therefore be due either to an additional phytochrome promotive as Pfr, or to the BL content of the WF. Day-extensions with monochromatic BL are effective for promotion of flowering in WT pea (Reid, 1982), and it will be interesting to examine their effect in the phyA mutant. Similar findings have been reported from Arabidopsis, where the loss of phyA resulted in a substantial reduction in response to a 1-h WL NB (Reed et al., 1994), but had little effect on the response to day-extensions with high R:FR WL (Johnson et al., 1994).

Basis for interaction of photoreceptors in the control of flowering

In Arabidopsis, the BL receptor cry2 is the main photoreceptor responsible for promotion of flowering under LD (Guo et al., 1998; Mockler et al., 1999), and a role for phyA is detectable only under certain specific conditions (Johnson et al., 1994). In pea, the photoperiod-dependent pathway has been primarily distinguished in terms of a graft-transmissible inhibitor of flowering (Weller et al., 1997b), and phyA clearly promotes flowering by reducing the level or transport of this inhibitor (Weller et al., 1997b). PhyB mutants in pea, although much earlier-flowering than WT in SD and near day-neutral for flower initiation, still show clear photoperiod responsiveness for vegetative characteristics (Weller and Reid, 1993), showing that they retain a functional photoperiod detection mechanism. In the present study, the epistasis of phyB to phyA and the lack of a graft-transmissible effect of phyB further support this conclusion and suggest that phyB may act predominantly at the apex.

This relatively late action proposed for phyB in pea is consistent with other recent results. In Arabidopsis, cry2 is epistatic to phyB for flowering under mixed R + BL, and acts at least partly through phyB (Mockler et al., 1999). In addition, expression of the photoperiod pathway gene CO, which is increased in response to long photoperiods (Putterill et al., 1995) and constitutively low in the cry2 mutant (Guo et al., 1998), is not markedly altered in the phyB mutant, whereas expression of the floral regulatory gene LFY is substantially increased (Blázquez and Weigel, 1999). However, phyB is known to be involved in entrainment of the circadian clock (Somers et al., 1998), and in the regulation of phyA expression (Cantón and Quail, 1999) and may thus also be capable of modifying the photoperiod pathway. In any case, the availability of pea PHYA (Sato, 1988) and PHYB genes, together with homologues of LFY (Hofer et al., 1997) and CO (Macknight et al., 1999) will allow comparisons of phytochrome function in pea and Arabidopsis flowering to be extended to the molecular level.

Experimental procedures

Plant material, mutagenesis and mutant screening

The isolation and characterisation of the phyB-1, phyA-1 and dne mutants has been previously reported (King and Murfet, 1985; Reid and Ross, 1988; Weller et al., 1997a). For sequencing of the PHYB cDNA, the original phyB-1 mutant was used together with its progenitor cv Sparkle. Mutant screening employed an EMS-mutagenised M2 population of WT cv. Torsdag described previously (Weller et al., 1997a). The phyB-5 and phyB-6 plant material used for molecular analysis was derived from the second back-cross of each mutant to cv. Torsdag, unless otherwise stated.

Growth conditions and light sources

All plants were grown from sowing in either a 1 : 1 mix of dolerite chips and vermiculite topped with potting mix or a 4 : 1 peat/sand mix. Light sources used for mutant screening, de-etiolation experiments (Figure 3) and growth-cabinet flowering experiments (Figure 5) have been described previously, as has the phytotron used in photoperiod and grafting experiments (Figures 6 and 7) (1997c; Weller et al., 1997a). The shade-avoidance experiment (Figure 4) was conducted in custom-built growth cabinets at the University of Leicester (UK) and these have also been previously described (McCormac et al., 1991). The spectral distribution and irradiance of all light sources was measured using a spectroradiometer (LiCor Corp. Lincoln, NE, USA).

Tissue sectioning

Stem segments were fixed in a solution of FAA (50 ml 36.5–38% formaldehyde, 25 ml glacial acetic acid, 204 ml 96% ethanol made up to 500 ml with milliQ) for 2 h and transferred to a graded ethanol dehydration series consisting of 2 × 30-min rinses each of 70% and 100% ethanol. Segments were then infiltrated in catalysed LR White resin (London Resin Company Ltd, Woking, Surrey, UK) and polymerised according to the manufacturer's instructions. Sections 3 µm thick were cut using a Microm Heidelberg HM 340 E rotary microtome. Sections were stained with toluidine blue, mounted in euparal, and viewed under a light microscope.

Molecular analysis of PHYB

A lambda Uni-ZAP XR cDNA library (Stratagene, La Jolla, CA, USA) prepared from light grown shoot apices of pea line JI1813 was screened with a fragment of the Arabidopsis PHYB gene spanning amino acids 185–385 (provided by J. Chory and T. Elich, Salk Institute, San Diego, CA, USA). Positive clones were identified after hybridization at 42°C in 2xSSC (1xSSC is 0.15 m NaCl and 0.015 m sodium citrate), 1%SDS, 20% formamide and 10% dextran sulfate, and washed at 42–50°C in 2xSSC and 0.1%SDS. Genomic 5′ sequence was obtained using the Universal Genome Walking Kit (ClonTech, Palo Alto, CA, USA) with gene-specific nested primers.

Both genomic and cDNA were used as template in the PCR amplification of PHYB for sequencing from WT lines cv. Torsdag and cv. Sparkle, and the phyB mutants. Genomic DNA was prepared from the leaves of light grown pea seedlings by the method of Ellis (1994). Total RNA was extracted using the RNEasy Plant Mini Kit (Qiagen, Valencia, CA, USA), and used to produce cDNA using the Superscript Preamplification System for First Strand cDNA Synthesis Kit (Life Technologies, Rockville, MD, USA). Amplified products were cloned in pGEM-T (Promega, Madison, WI, USA) and sequenced using dye-terminator cycle sequencing methods and an automated sequencer (model Prism 377; Applied Biosystems, Foster City, CA, USA) at the CSIRO Marine Laboratories (Hobart, Australia). Sequence alignments and comparisons were made using the BLAST program facility at the National Centre for Biotechnology Information (NCBI) (Bethesda, MD, USA). Sequence differences between WT and phyB mutants were verified by sequencing surrounding regions using cDNA from two additional, independent RNA extractions, and confirmed using genomic DNA. The phyB-5 mutation introduced a novel MnlI restriction site that was converted into a CAPS marker with primers LV5F and LV5R and used to follow the cosegregation of the molecular lesion with the phyB-5 phenotype.

Immunoblot detection of PHYB was performed as described previously (Weller et al., 1995), using anti-PHYB antibody mAP11 kindly supplied by Akira Nagatani (Kyoto University, Japan).

Grafting

Scions were prepared from 6-day-old glasshouse-grown seedlings as previously described (Weller et al., 1997b), and grafted into the internode between leaves eight and nine of 24-day-old plants. The scion-stock junction was secured using small sections of rubber tubing, and a plastic zip-lock bag was placed over the scion and sealed below the site of the graft to maintain humidity during establishment of the graft union. Node of flower initiation in the scion was counted from the first scale leaf as node 1.

Primer sequences

The primer sequences are listed below, together with the nucleotide position to which they correspond in the coding strand (cs) or non-coding strand (ncs) of the PHYB cDNA sequence (GenBank accession AF069305).

ABF, cs 515–540, 5′-CTTTTTCAGCTCGGGAGATTAGTTTA-3′;

ABR, ncs 3713–3741, 5′- ACATTATCAGCAAGGTTTTAGGTTAC ATT-3′;

LV5F, cs 1553–1582, 5′-ATGCTGGTTATCCTGGGGCTGCTTCTC TTG-3′;

LV5R, ncs 1696–1725, 5′-ATGCATTTTCTGCCCGTCATCCTTATC CCC-3′.

An additional primer (5′-ACAACTTCACTCTCACATTCCATTCT TAC-3′) was used for amplification of sequence at the 5′ end of the gene, corresponding to bases 98 to126 in the PHYB 5′ genomic sequence (Genbank accession AF323519) and bases −88 to −60 relative to the start ATG codon.

Acknowledgements

We thank Ian Cummings and Tracey Jackson for plant maintenance, Akira Nagatani for providing monoclonal antibodies, and Harry Smith for access to R:FR cabinets at the University of Leicester. This work was supported in part by Australian Postgraduate Research Awards to J. L. W. and N. B., an Australian Postdoctoral Fellowship to L. H. J. K, and a grant to J. B. R from the Australian Research Council.

Ancillary