Phytochrome-mediated inhibition of shade avoidance involves degradation of growth-promoting bHLH transcription factors

Authors

  • Séverine Lorrain,

    1. Center for Integrative Genomics, University of Lausanne, Genopode Building, CH-1015 Lausanne, Switzerland
    Search for more papers by this author
    • These authors contributed equally to this paper.

  • Trudie Allen,

    1. Department of Biology, University of Leicester, Leicester LE1 7RH, UK
    Search for more papers by this author
    • These authors contributed equally to this paper.

  • Paula D. Duek,

    1. Center for Integrative Genomics, University of Lausanne, Genopode Building, CH-1015 Lausanne, Switzerland
    Search for more papers by this author
    • These authors contributed equally to this paper.

    • Present address: Swiss Institute of Bioinformatics, 1 rue Michel Servet, CH-1211 Geneva 4, Switzerland.

  • Garry C. Whitelam,

    1. Department of Biology, University of Leicester, Leicester LE1 7RH, UK
    Search for more papers by this author
  • Christian Fankhauser

    Corresponding author
    1. Center for Integrative Genomics, University of Lausanne, Genopode Building, CH-1015 Lausanne, Switzerland
    Search for more papers by this author

*(fax +41 21 692 3925; e-mail christian.fankhauser@unil.ch).

Summary

Plant growth and development are particularly sensitive to changes in the light environment and especially to vegetational shading. The shade-avoidance response is mainly controlled by the phytochrome photoreceptors. In Arabidopsis, recent studies have identified several related bHLH class transcription factors (PIF, for phytochrome-interacting factors) as important components in phytochrome signaling. In addition to a related bHLH domain, most of the PIFs contain an active phytochrome binding (APB) domain that mediates their interaction with light-activated phytochrome B (phyB). Here we show that PIF4 and PIF5 act early in the phytochrome signaling pathways to promote the shade-avoidance response. PIF4 and PIF5 accumulate to high levels in the dark, are selectively degraded in response to red light, and remain at high levels under shade-mimicking conditions. Degradation of these transcription factors is preceded by phosphorylation, requires the APB domain and is sensitive to inhibitors of the proteasome, suggesting that PIF4 and PIF5 are degraded upon interaction with light-activated phyB. Our data suggest that, in dense vegetation, which is rich in far-red light, shade avoidance is triggered, at least partially, as a consequence of reduced phytochrome-mediated degradation of transcription factors such as PIF4 and PIF5. Consistent with this idea, the constitutive shade-avoidance phenotype of phyB mutants partially reverts in the absence of PIF4 and PIF5.

Introduction

Light signals trigger important adaptive responses such as the shade-avoidance syndrome, which enable plants to respond to depletion of photosynthetically active light (Franklin and Whitelam, 2005; Smith, 2000; Vandenbussche et al., 2005). A reduction of the red to far-red (R/FR) ratio is a major feature of shaded environments. It is a consequence of the depletion of red (R) light absorbed by the competing plants as well as an enrichment in far-red (FR) light reflected by neighboring plants. This important response is controlled by phytochromes, which are R/FR photoreceptors (phyA–phyE in Arabidopsis) (Franklin and Whitelam, 2005; Smith, 2000; Vandenbussche et al., 2005). In Arabidopsis, phyB is the primary phytochrome mediating R/FR-reversible responses (known as low-fluence responses), but there is a clear degree of redundancy among members of this family (Franklin and Whitelam, 2004; Smith, 2000). At the physiological level, the shade-avoidance response is characterized by a reallocation of resources to elongation-growth responses in order for the plant to reach unfiltered sunlight (Franklin and Whitelam, 2004; Smith, 2000; Vandenbussche et al., 2005). At the molecular level, low R/FR ratio results in extremely rapid expression of genes encoding transcription factors such as ATHB-2, PIF3-LIKE 1 (PIL1) and LONG HYPOCOTYL IN FAR-RED 1 (HFR1), which are required for the appropriate growth responses (Salter et al., 2003; Sessa et al., 2005).

The phytochromes are synthesized in the red light-absorbing state known as Pr. Upon excitation by red light, they are converted into the far-red light-absorbing (Pfr state) (Rockwell et al., 2006). The active Pfr conformer can be rapidly inactivated upon FR light absorption, or slowly inactivated in the absence of light in a process known as dark reversion (Rockwell et al., 2006). Prior to photo-excitation, phyB is a soluble cytoplasmic protein, but, in its Pfr conformation, phyB accumulates in the nucleus where it has been suggested to regulate gene expression (Jiao et al., 2007). In its light-activated state (Pfr), phyB can physically interact with a group of related bHLH transcription factors known as phytochrome interacting proteins (PIFs), suggesting a short signal transduction cascade from light sensing to regulated gene expression (Jiao et al., 2007; Khanna et al., 2004; Martinez-Garcia et al., 2000). Members of the PIF family control phytochrome responses including seed germination, seedling de-etiolation and shade avoidance (Huq and Quail, 2002; Huq et al., 2004; Kim et al., 2003; Oh et al., 2004; Penfield et al., 2005; Salter et al., 2003; Shin et al., 2007). Interaction between phyB and the PIFs is mediated by the active phytochrome binding (APB) domain that is present at the N-terminus of these proteins (Khanna et al., 2004). Some PIF proteins, such as PIF1, interact with both phyB and phyA, while others, such as PIF3, PIF4 and PIF5, interact preferentially with phyB. These interaction data, determined in vitro, correlate with the physiological functions of these bHLH class transcription factors. For instance, PIF1 is involved in phyA and phyB signaling, while PIF4 and PIF5 are selectively involved in phyB signaling (Fujimori et al., 2004; Huq and Quail, 2002; Huq et al., 2004; Oh et al., 2004).

Somewhat surprisingly, most PIF proteins act as negative regulators of the phyB pathway, leading to questions regarding the function of the PIF–phy interaction (Duek and Fankhauser, 2005). It has been proposed that the interaction of phyB with PIF4 is essential for the function of this transcription factor (Khanna et al., 2004). On the other hand, PIF1 inhibits several light responses, and phytochrome activation of these responses occurs by inhibition of this negative regulator (Huq et al., 2004; Oh et al., 2004). This phytochrome-mediated inhibition of PIF1 appears to involve light-induced degradation of PIF1 (Oh et al., 2006; Shen et al., 2005). A similar mechanism operates for PIF3; moreover, PIF3 degradation requires the interaction with phyB through a functional APB domain (Al-Sady et al., 2006). In both cases, the degradation of these transcription factors is proteasome-mediated (Al-Sady et al., 2006; Bauer et al., 2004; Monte et al., 2004; Oh et al., 2006; Park et al., 2004; Shen et al., 2005; Viczian et al., 2005).

Here we show that PIF4 and PIF5 positively regulate low R/FR ratio-mediated shade avoidance, using both morphological criteria and marker gene expression. These proteins are abundant in the dark and shade-mimicking conditions, but are rapidly degraded in response to high R/FR ratio. Our data indicate that changes in light quality perceived by the phytochromes rapidly regulate the abundance of positive regulators of shade avoidance such as PIF4 and PIF5.

Results

PIF4 and PIF5 promote the shade-avoidance response

During seedling de-etiolation, PIF4 and PIF5 act as negative regulators of the phyB pathway, but their function in adult plants remains poorly characterized (Fujimori et al., 2004; Huq and Quail, 2002). In order to gain a better understanding of their function, we generated pif4 pif5 double mutants and wild-type Col-0 plants expressing HA epitope-tagged PIF4 or PIF5 under the control of the constitutive CaMV 35S promoter. Such plants had considerably elongated hypocotyls and displayed what may be regarded as a constitutive shade-avoidance phenotype (Supplementary Figure S1). This phenotype correlated with strong expression of the HA-tagged PIF proteins, and was similar but often more extreme than the phenotype reported for PIF4- or PIF5-over-expressing plants (Fujimori et al., 2004; Khanna et al., 2004). We therefore analyzed lines with moderate phenotypes, and concentrated our analysis on PIF5-expressing plants because plants expressing high levels of PIF4 were difficult to maintain.

It has previously been proposed that the APB domain of PIF4 is essential for function in vivo (Khanna et al., 2004). It was thus surprising to notice that a high proportion of primary transformants containing 35S:ΔN-PIF4-HA and 35S:ΔN-PIF5-HA also displayed a strong constitutive shade-avoidance phenotype (Supplementary Figure S1, and data not shown). We note that the two experiments may not be directly comparable. In the previous experiment, there were point mutations in the APB domain, while in our study the APB domain is deleted. Moreover, the promoters driving these mutated forms of PIF4 also differ (Khanna et al., 2004). Importantly, this gain-of-function phenotype was also observed when 35S:ΔN-PIF4-HA was transformed into pif4 mutants (data not shown). These data indicate that over-expression of these transcription factors inhibits phytochrome responses in adult plants irrespective of the presence of the N-terminus of the protein containing the APB domain. These results correlate with data showing that over-expression of full-length PIF4 and PIF5 or PIF4 and PIF5 lacking the APB domain led to an etiolated seedling phenotype in R light (data not shown).

Contrary to the constitutive shade-avoidance phenotype displayed by over-expressing lines, mutants null for PIF4 and PIF5 exhibited hypersensitivity to light (Figure 1 and Supplementary Figure S1) (Fujimori et al., 2004; Huq and Quail, 2002). For these studies, we analyzed the previously characterized pil6-1 allele (pif5 mutant) (Fujimori et al., 2004) and a new T-DNA insertion allele in PIF4 (pif4-101) presumably resulting in a null allele (see Experimental procedures). Reduced hypocotyl elongation was observed in seedlings, an effect that was most pronounced in the pif4 pif5 double mutant (Supplementary Figure S1). In adult plants, these mutants displayed a compact rosette phenotype (Supplementary Figure S1). Together, these data suggest a role for PIF4 and PIF5 as positive regulators of the elongation-growth responses that are typical of the shade-avoidance syndrome.

Figure 1.

 Gain and loss-of-function studies indicate that phytochrome-interacting factors (PIF)4 and PIF5 are required for normal responses to low red to far-red (R/FR).
(a) Hypocotyl elongation of pif mutants. Hypocotyl lengths of seedlings grown for 7 days under high R/FR (black bars), or for 4 days under high R/FR and 3 days under low R/FR (white bars) were measured. Error bars represent SE. Asterisks indicate a significant difference from Col-0 using Student’s t test (≤ 0.05).
(b) Hypocotyl elongation of PIF5-over-expressing plants. Growth conditions and measurements were performed as in (a).
(c) Petiole length of leaf 3 in pif mutants and PIF5-over-expressing lines. Plants were grown for 13 days under high R/FR (black bars), or for 3 days under high R/FR and 10 days under low R/FR (white bars). Error bars represent SE. Asterisks indicate a significant difference from Col-0 using Student’s t test (≤ 0.05).
(d) Representative leaf 3 phenotypes from 13-day-old-plants grown under high or low R/FR.
(e) PIF4 and PIF5 regulate the expression of shade-avoidance markers. Expression of low-R/FR-regulated genes in pif mutants and PIF5-over-expressing lines was determined by quantitative PCR in seedlings grown for 9 days under continuous high R/FR (black bars) or for 5 days under high R/FR and 4 days under low R/FR (white bars). Three biological repeats were performed, with two technical replicates for each. Relative expression was determined according to the method described by Pfaffl (2001), comparing each sample to one Col-0 sample under high R/FR; ACTIN2 was used as a reference gene. Each column represents the mean relative expression for the three biological repeats; error bars represent the SE.

We thus characterized the role of PIF4 and PIF5 during shade avoidance by analyzing growth under high and low R/FR ratios. The hypocotyls of pif4 and pif4pif5 mutants grown under constant high R/FR conditions were slightly less elongated than those of the wild-type (WT). Importantly, pif4 mutants displayed a significantly reduced elongation response to low R/FR (Figure 1a). The response to low R/FR was also diminished in the pif5 mutant, and the pif4 pif5 double mutant behaved similarly to the pif4 mutant. In contrast, over-expression of full-length or truncated PIF5 resulted in an elongated-hypocotyl phenotype under high R/FR that is typical of constitutive shade avoidance. In all cases, low R/FR led to a significant growth-promoting effect (Figure 1b). In high R/FR, petiole elongation of pif4 and pif5 was normal, while the pif4 pif5 double mutant showed reduced elongation (Figure 1c,d). Again, pif4, pif5 and pif4 pif5 mutants demonstrated elongation of petiole length in response to low R/FR, but this response was attenuated in pif5 (Figure 1c,d). The petioles of lines over-expressing PIF5 showed constitutive elongation under high R/FR compared to wild-type (Figure 1c,d).

The identification of genes displaying rapid R/FR-dependent transcript accumulation has provided molecular markers for the analysis of shade-avoidance responses. Such genes include PIL1, HFR1 and ATHB-2, which encode two bHLH transcription factors and a homeodomain ZIP transcription factor, respectively (Carabelli et al., 1993; Salter et al., 2003; Sessa et al., 2005). The relative expression of these three markers was therefore determined in null mutants and over-expressing lines. Seedlings were grown for 9 days under high R/FR, or 5 days under high R/FR and 4 days under low R/FR. Consistent with published observations, WT plants displayed elevated PIL1, HFR1 and ATHB-2 transcript levels under low R/FR (Figure 1e) (Carabelli et al., 1993; Salter et al., 2003; Sessa et al., 2005). Interestingly, although pif4, pif5 and particularly pif4 pif5 mutants respond to low R/FR, this induction was reduced compared to the WT (Figure 1e). Over-expression of PIF5 resulted in higher transcript levels for all genes under high R/FR; however, low R/FR still led to robust gene induction (Figure 1e). To examine shade-induced gene regulation in more detail, transcript levels of PIL1 and ATHB-2 were followed over a 24 h time course in seedlings that were entrained in day–night cycles and released at dawn into either high or low R/FR (Figure 2). These data confirmed the reduced response to low R/FR of pif4, pif5 and particularly pif4 pif5. Moreover, the PIF5-over-expressing plants had high levels of these transcripts at the end of the night, and these remained at a high level in low R/FR, but declined significantly faster in response to high R/FR. Overall, these data correlate well with the morphological phenotypes of null and over-expression lines, and confirm the involvement of PIF4 and PIF5 in shade avoidance.

Figure 2.

 Time courses of red to far-red (R/FR)-regulated genes after transfer to continuous light.
Seedlings were grown for 12 days under high R/FR (12 h light, 12 h dark), then transferred to continuous high or low R/FR at dawn on day 13. The expression of all samples relative to Col-0 at dawn was determined according to the method described by Pfaffl (2001), using ACTIN2 as a reference gene.

The simplest interpretation of our data is that PIF4 and PIF5 are required for the full growth responses that occur when plants grow in the shade. It was thus of interest to test whether pif mutants could at least partially suppress the constitutive shade-avoidance phenotype displayed by phyB mutants (Franklin et al., 2003). As the pif4 and pif4 pif5 mutants displayed the strongest phenotypes, we generated phyB pif4 and phyB pif4 pif5 mutants (Figure 3). phyB mutants displayed the characteristic constitutive shade-avoidance phenotype, with elongated hypocotyls, long petioles and elevated levels of shade marker genes under high R/FR (Figure 3, and data not shown). Interestingly, this phenotype was suppressed in the phyB pif4 double mutants and even more significantly in the phyB pif4 pif5 triple mutants in terms of hypocotyl elongation and expression of shade marker genes (Figure 3). These results are thus consistent with the notion that PIF4 and PIF5 are necessary for the full constitutive shade-avoidance phenotype observed in phyB mutants.

Figure 3.

 The pif4 and pif5 mutations partially suppress the constitutive shade-avoidance phenotype of phyB.
(a) Hypocotyl elongation of pif4, phyB, phyB pif4 and phyB pif4 pif5 mutants. Hypocotyl lengths of seedlings grown for 7 days under high R/FR. Error bars represent SE.
(b) Expression of PIL1 in pif4, phyB, phyB pif4 and phyB pif4 pif5 mutants. Expression of low-R/FR-regulated genes was determined by quantitative PCR in seedlings grown for 7 days under continuous high R/FR. Three biological repeats were performed, with two technical replicates for each. Relative expression was determined according to the method described by Pfaffl (2001), comparing each sample to one Col-0 sample; ACTIN2 was used as a reference gene. Each column represents the mean relative expression for the three biological repeats; error bars represent the SE.
(c) Expression of HFR1 in pif4, phyB, phyB pif4 and phyB pif4 pif5 mutants. The experiment was performed as described in (b).
(d) Expression of ATHB2 in pif4, phyB, phyBpif4 and phyBpif4pif5 mutants. The experiment was performed as described in (b).

PIF4 and PIF5 protein levels are specifically reduced in response to R light

Recently, it has been shown that PIF4 and PIF5 protein stability depends on light (Nozue et al., 2007), with an accumulation of both proteins in the dark and a strong reduction in their level in the light. In order to determine which light quality controls PIF4 and PIF5 accumulation, we analyzed the regulation of these proteins under various light conditions. The results obtained for PIF4-HA and PIF5-HA are similar. For simplicity, the data for PIF5-HA are presented here, and the data for PIF4-HA are shown in Supplementary Figures S2 and S3. Etiolated seedlings were transferred to various light conditions, and proteins were extracted and analyzed by western blotting. PIF5 accumulated in the dark and was detected as multiple isoforms (Figure 4a). A short period (5 min) of R light induced the appearance of additional slower-migrating forms on SDS–PAGE gels (Figure 4a). After 1 h of R light, the level of proteins reached a minimum, with the disappearance of the higher-molecular-weight isoforms. Interestingly, PIF5 re-accumulated after prolonged exposure to light, indicating that the light-induced decrease of PIF5 levels was transient (Figure 4a). When seedlings treated with R light were transferred back into darkness, the proteins re-accumulated to a high level at a fast rate (Figure 4b). This pattern of light regulation was specifically induced by R but not FR light (Figure 4c). The light-regulated protein abundance of PIF4-HA and PIF5-HA was thus similar to that reported for their close relatives PIF1 and PIF3, except that PIF4 and PIF5 re-accumulated much faster than PIF1 after return into darkness and FR light did not induce a significant decrease in PIF4 and PIF5 (Figure 4c and Supplementary Figure S2c, and data not shown) (Al-Sady et al., 2006; Bauer et al., 2004; Monte et al., 2004; Park et al., 2004; Shen et al., 2005; Viczian et al., 2005).

Figure 4.

 PIF5 is stable in the dark and rapidly and specifically degraded in response to red light through a low-fluence response. Total proteins were extracted at the indicated times and PIF5-HA accumulation was analyzed by Western blotting using anti-HA antibodies. Ponceau S staining or DET3 accumulation are shown as a loading control.
(a) PIF5-HA accumulation in response to red light. Three-day-old etiolated seedling over-expressing HA-tagged PIF5 proteins (35S:PIF5) were exposed to 50 μmol m−2 sec−1 constant red light.
(b) Reaccumulation of PIF5-HA in darkness after 30 min red light treatment as in (a).
(c) PIF5-HA accumulation in response to far-red light. Seedlings were treated with 50 μmol m−2 sec−1 constant far-red light.
(d) PIF5-HA accumulation depends on a low fluence response. Etiolated seedlings were treated with a 2 min red pulse and returned to darkness. The effect of the 2 min red pulse was reversed when it was immediately followed by a 2 min far-red pulse.

Phytochromes regulate PIF4 and PIF5 protein accumulation

The decrease in PIF4 and PIF5 levels in response to R but not FR light suggested that PIF4 and PIF5 protein levels may be controlled by a phytochrome low-fluence response (Quail, 2002; Smith, 2000). To test this hypothesis, etiolated seedlings were subjected to a 2 min R light pulse and transferred back into darkness. This short light treatment was sufficient to trigger the initial appearance of slower-migrating forms, followed by a strong decrease in PIF4 and PIF5 protein abundance (Figure 4d and Supplementary Figure S2d). A hallmark of a phytochrome low-fluence response is the reversibility of the R-light-induced response by a subsequent FR light treatment (Quail, 2002; Smith, 2000). Both PIF4 and PIF5 proteins levels remained unchanged when a FR light pulse followed the R light treatment (Figure 4d and Supplementary Figure S2d). In control experiments, we verified that, in etiolated seedlings treated with a FR light pulse, the levels and isoforms of PIF4 and PIF5 remained unchanged (data not shown). Taken together, these data indicate that a phytochrome low-fluence response controls PIF4 and PIF5 protein abundance.

PIF5 protein accumulation is controlled by R/FR ratio

The role of PIF5 in shade avoidance (Figures 1–3) and the involvement of phytochromes in the control of its accumulation (Figure 4) prompted us to test whether different R/FR ratios could regulate PIF5 stability. Seedlings expressing PIF5-HA were grown under short-day conditions for 5 days, and PIF5 accumulation was monitored on day 6 at around dawn and dusk in response to low or high R/FR ratios. At these time points, maximal responses to the low R/FR signal occur for gene expression and elongation growth, respectively (Salter et al., 2003). PIF5 protein levels were high when seedlings were transferred into low R/FR at dawn, and significantly lower when the seedlings were treated with high R/FR at that time of the day (Figure 5a). Furthermore, when seedlings grown in high R/FR were transferred into low R/FR towards the end of the day, PIF5 re-accumulation was observed as rapidly as 15 min after the change in R/FR ratio (Figure 5b). These experiments demonstrate that, in light-grown plants, R/FR ratios have a rapid impact on PIF5 protein levels. PIF4 responded similarly to PIF5 at dawn but showed little response to light quality around dusk (data not shown).

Figure 5.

 PIF5 protein level is regulated by red to far-red (R/FR) ratio.
Seedlings over-expressing HA-tagged PIF5 (35S:PIF5) were grown under short-day conditions (8 h of high R/FR light/16 h dark) for 5 days.
(a) At dawn on the 6th day (sample D, prior to the light treatment), seedlings were exposed to a high (18) or low (0.1) R/FR ratio for 2 h, and proteins were extracted at the indicated times.
(b) After 6 h under a high R/FR ratio (2 h before dusk, sample T2), seedlings were transferred to low R/FR ratio, and proteins were extracted at the indicated times. Seedlings kept under a high R/FR ratio were used as a control.
In both cases, PIF5-HA protein accumulation was analyzed by western blotting using anti-HA and anti-DET3 (control) antibodies.

PIF4 and PIF5 degradation is proteasome-mediated, requires the N-terminus of the protein, and is regulated by phosphorylation

The rapid decrease in PIF4 and PIF5 abundance in response to a R-light treatment, and the precedent for PIF1 and PIF3 light regulation, suggested involvement of the proteasome in the regulation of PIF4 and PIF5 (Bauer et al., 2004; Monte et al., 2004; Park et al., 2004; Shen et al., 2005). To test this directly, etiolated seedlings either kept in the dark or transferred into R light were treated with proteasome inhibitors. The level of PIF4 and PIF5 did not decrease in response to light in the presence of these inhibitors (Figure 6b and Supplementary Figure S3), thus supporting the hypothesis of proteasome-mediated degradation of these bHLHs.

Figure 6.

 Degradation of the PIF5 phosphoprotein is mediated by the proteasome and requires the APB domain.
(a) Deletion of the first 68 amino acids of PIF5 stabilizes the protein. Three-day-old etiolated transgenic seedlings over-expressing full-length (35S:PIF5) or truncated (35S:ΔN-PIF5) HA-tagged PIF5 proteins were exposed to 50 μmol m−2 sec−1 constant red light or kept in the dark.
(b) The use of proteasome inhibitors abolishes PIF5 degradation. Three-day-old etiolated transgenic seedlings expressing PIF5-HA or ΔN-PIF5 were incubated for 1 h with proteasome inhibitors and exposed to 50 μmol m−2 sec−1 constant red light for 1 h. Control dark-grown seedlings were extracted after 1 h proteasome treatment in the dark (D). Control seedlings were treated with 3.5% DMSO.
(c) Slower-migrating forms of PIF5 correspond to phosphorylated forms. Proteins were extracted from 3-day-old etiolated seedlings expressing PIF5-HA or after an additional 1 h exposure to red light (50 μmol m−2 sec−1). Phosphatase treatments were performed on total protein extract for 30 min in the absence (−) or presence (+) of phosphatase inhibitors (50 mm EDTA). As a control, protein extracts were incubated for 30 min in the presence of inhibitors without phosphatase. In all cases, total proteins were extracted at the indicated times, separated on SDS–PAGE, western blotted, and probed with anti-HA and anti-DET3 (control) antibodies.

In Arabidopsis, phyB is the principal phytochrome mediating the low-fluence response (Quail, 2002). It has previously been shown that the phyB–PIF interaction selectively occurs with the Pfr conformer of phyB via the N-terminal APB domain of the PIF proteins (Khanna et al., 2004). Moreover, the phytochrome-mediated degradation of PIF3 requires direct interaction between PIF3 and the phytochromes through its phytochrome-binding domains (Al-Sady et al., 2006). We thus tested whether the phytochrome-mediated degradation of PIF4 and PIF5 required interaction between the photoreceptor and the PIF proteins by generating transgenic lines expressing epitope-tagged PIF4 and PIF5 lacking the first 65 amino acids of the protein containing the APB domain (ΔN lines). Such transgenic lines were grown in the dark, moved into R light, and protein extracts were analyzed by western blotting. ΔN-PIF4 and ΔN-PIF5 migrated as multiple isoforms, and, in the absence of the APB domain, the PIF4 and PIF5 protein isoforms were no longer significantly affected by light, while their accumulation only showed a weak light response (Figure 6a, Supplementary Figure S3). These results suggest that light-induced interaction between phytochrome and these two PIF proteins triggers their subsequent modification and degradation.

PIF4 and PIF5 exist as multiple isoforms in the dark (Figures 4 and 5, and Supplementary Figures S2 and S3). Moreover, in response to a light treatment, additional more slowly-migrating forms appear very transiently (Figures 4–6, and Supplementary Figures S2 and S3). As proteasome-mediated degradation is often regulated by specific phosphorylation events (Al-Sady et al., 2006; Collins and Tansey, 2006; Hardtke et al., 2000; Richardson and Zundel, 2005), we tested whether PIF4 and PIF5 are phosphoproteins. Extracts of dark-grown and etiolated seedlings exposed for 1 h to R light were treated with phosphatase. In both cases, this treatment caused disappearance of the slower-migrating isoforms, indicating that PIF4 and PIF5 are phosphorylated in the dark and that light induces the appearance of additional phosphorylation events (Figure 6c and Supplementary Figure S3). These additional phosphorylations were very rapidly induced by light and are R/FR-reversible, indicating phytochrome regulation (Figure 4d and Supplementary Figure S2d). The slowest-migrating isoforms were strongly stabilized in light-treated seedlings in the presence of proteasome inhibitors (Figure 6b and Supplementary Figure S3), indicating that these isoforms are particularly labile. The absence of these isoforms in dark-grown seedlings treated with proteasome inhibitors (Figure 6b and Supplementary Figure S3) further confirmed the light dependency of specific phosphorylation events.

Discussion

Competition for light is a rate-limiting factor determining the success of individual plants in dense vegetation. Plants possess sophisticated light-sensing mechanisms that enable them to detect neighbors and rapidly initiate appropriate growth responses (Franklin and Whitelam, 2004; Smith, 2000; Vandenbussche et al., 2005). A hallmark of light reflected by, or transmitted through, vegetation is its low R/FR ratio, detected by the phytochromes (Franklin and Whitelam, 2004; Smith, 2000; Vandenbussche et al., 2005). Analysis of phytochrome mutants has demonstrated that, in Arabidopsis, phyB, phyD and phyE are the major phytochromes inhibiting shade avoidance (Devlin et al., 1998; Franklin et al., 2003). The search for genes up-regulated in response to shade led to the identification of several transcription factors that are very rapidly induced in response to low R/FR ratios (Carabelli et al., 1993; Devlin et al., 2003; Roig-Villanova et al., 2006; Salter et al., 2003; Sessa et al., 2005). Functional characterization of these genes indicated that, while ATHB-2 promotes growth responses associated with shade, HFR1 is part of a negative branch preventing ‘over-reaction’ in response to canopy light (Roig-Villanova et al., 2006; Salter et al., 2003; Sessa et al., 2005; Steindler et al., 1999). As ATHB-2 and PIL1 expression is constitutively high in phy mutants, especially in the phyB phyD phyE triple mutants, the high level of their mRNA during shade conditions may result from a phytochrome-dependent de-repression mechanism (Salter et al., 2003). However, the molecular mechanism explaining phytochrome-mediated repression of growth-promoting genes is still unknown.

PIF4 and PIF5 have recently been shown to be involved in the control of diurnal growth, as the integration point of circadian and light signals. In particular, they promote hypocotyl growth during the night while their protein abundance is high. At dawn, growth is stopped as the result of their light-triggered degradation (Nozue et al., 2007). Taking these results into account, we performed our shade-avoidance experiments in continuous light to distinguish between the effects of diurnal cycles versus light quality. Our data indicate that PIF4 and PIF5 growth-promoting functions are recruited during shade-avoidance responses (Figures 1–3). Plants over-expressing PIF4 or PIF5 constitutively express high levels of HFR1, PIL1 and ATHB-2 (Figures 1 and 2, and data not shown) and display a constitutive shade-avoidance phenotype (Supplementary Figure S1). Additionally, pif4, pif5 and especially pif4 pif5 mutants show reduced low R/FR-mediated expression of HFR1, PIL1 and ATHB-2, and pif4, in particular, displays reduced physiological responses to shade, indicating that PIF4 and PIF5 are required for normal responses to the low R/FR signal (Figures 1 and 2). Finally, pif4 and pif4 pif5 partially suppress the shade-mimicking phenotype of phyB (Figure 3). All three shade marker genes tested are regulated by PIF4 and PIF5 (Figures 1–3), and their promoters all contain E or G boxes, which are the binding sites of bHLH transcription factors. This suggests a global role for those two PIF proteins as early intermediates in the control of shade-regulated gene expression. They represent a direct link between the light-controlled changes in phytochrome conformation and the regulation of early shade-responsive genes. However, it should be noted that pif4, pif5 and pif4 pif5 still display shade-avoidance responses (Figure 1). These data suggest that multiple molecular mechanisms couple R/FR ratio perception with shade avoidance. Functional redundancy between the various members of the PIF family is one possibility. Another possibility is that several distinct mechanisms coordinately control shade avoidance (Djakovic-Petrovic et al., 2007).

In parallel to their involvement in elongation responses to the shade signal, we found that PIF4 and PIF5 protein are specifically degraded in responses to R but not FR (Figure 4 and Supplementary Figure S2). Furthermore, their level depends on the R/FR ratio (Figure 5), providing an explanation of how the R/FR ratio can control elongation responses through phytochromes. We thus propose that, in addition to controlling growth during day/night transition, PIF4 and PIF5 are also recruited for elongation growth in response to shade. Perception of the R/FR ratio by the phytochromes controls the PIF4 and PIF5 protein level, which ultimately modulates elongation responses.

There is a good correlation between the phy–PIF interaction data and the requirement of phyA and/or phyB for the degradation of PIF protein in vivo (Al-Sady et al., 2006). PIF1 interacts significantly with phyA and phyB in vitro, and is degraded in response to continuous FR and R light, conditions under which phyA and phyB, respectively, are the predominant phytochromes controlling the light response (Huq et al., 2004; Oh et al., 2006; Shen et al., 2005). In contrast, PIF4 and PIF5 have only been reported to interact strongly with phyB (Huq and Quail, 2002; Khanna et al., 2004), and they are not degraded in response to continuous FR light (Figure 4 and Supplementary Figure S2). A recent study of PIF3 protein stability also indicates that phy–PIF interaction targets PIF3 to degradation; however, the functional consequences of expressing light-stable PIF3 have not been reported (Al-Sady et al., 2006). Consistent with the phy–PIF interaction data, truncated forms of PIF4 and PIF5 lacking the first 65 amino acids including the APB domain show strongly impaired degradation in response to light (Figure 6). Remarkably, the constitutive shade-avoidance phenotype displayed by plants over-expressing either ΔN-PIF5 or ΔN-PIF4 (Figure 1, and data not shown) suggests that these proteins may still function in the absence of the APB domain. This finding is not easy to reconcile with previous models, which have suggested a requirement for phytochrome activation for PIF activity (Khanna et al., 2004). However, we cannot rule out the possibility that the APB domain does more than regulate phytochrome-controlled PIF stability. This is suggested by the surprisingly strong response of line ΔN-PIF5-OX1 to low R/FR (Figure 1b).

It has been proposed that the PIF proteins act as negative regulators of phyB activity (Khanna et al., 2004). However, the partial suppression of the constitutive shade-avoidance phenotype of phyB by pif mutations (Figure 3) is difficult to reconcile with this model. Our data indicate that the phytochromes are negative regulators of PIF activity, although it is possible that the PIFs have an additional indirect effect on phytochrome activity. We thus propose that a major function of the PIF–phy interaction is to modulate PIF activity by targeting PIF proteins to degradation (Figure 7). Our model predicts that PIF proteins may act in darkness, without phytochrome activation. This is in accordance with a report showing that PIF4 and PIF5 regulate diurnal hypocotyl growth (Nozue et al., 2007), with the role of PIL5/PIF1 during seed germination (Oh et al., 2006;Shen et al., 2005), and with a recent report showing a function for PIL5/PIF1, PIF3 and PIF4 in etiolated seedlings (Alabadi et al., 2007).

Figure 7.

 Speculative model for the control of the shade avoidance through PIF4 and PIF5 stability.
In sunlight, phyB is mainly in the Pfr form. The Pfr form accumulates in the nucleus where it interacts with PIFs via the active phytochrome binding domain. This interaction targets PIF proteins to degradation through the proteasome, leading to deactivation of genes induced by the shade. Under a canopy, light filtered or reflected by the neighboring vegetation is enriched in FR light, and phyB is converted into its Pr form that no longer interacts with the PIF proteins. These proteins will thus rapidly re-accumulate, promoting the expression of genes such as PIL1, ATHB2 and HFR1 that direct shade-avoidance responses. Although this model suggests that shade marker genes are direct targets of PIF4 and PIF5, this has not yet been demonstrated.

PIF4 and PIF5 are phosphoproteins in the dark and in the light; however, light triggers the appearance of novel phosphorylation forms (Figure 6 and Supplementary Figure S3). This light-specific modification appears within min and is R/FR-reversible, indicating phytochrome involvement (Figure 4 and Supplementary Figure S2). Given the speed of this reaction, a tempting scenario is direct involvement of the proposed kinase activity of higher-plant phytochromes (Yeh and Lagarias, 1998). Light-induced PIF3 degradation is also preceded by phytochrome-dependent phosphorylation (Al-Sady et al., 2006). As for PIF3, the light-specific phosphorylation forms of PIF4 and PIF5 appear to be the shortest-lived ones (Figures 4 and 6). Regulation of proteasome-mediated degradation of transcription factors is not uncommon (Al-Sady et al., 2006; Collins and Tansey, 2006; Hardtke et al., 2000; Richardson and Zundel, 2005). In mammalian systems, activation and degradation of transcription factors are often coupled, presumably to ensure tight control of the activation mechanism (Collins and Tansey, 2006). The phytochrome-mediated degradation of PIF4 and PIF5 may thus also be coupled to phytochrome-induced phosphorylation, leading to activation prior to degradation, as previously suggested for PIF3 (Al-Sady et al., 2006).

Taken together, our data suggest a mechanism contributing to the control of shade avoidance by light-stable phytochromes (Figure 7). In sunlight (high R/FR), the phytochromes are converted into their Pfr conformation, interact with PIF4 and PIF5 (and other members of the PIF family), and trigger additional phosphorylation and the degradation of these bHLH class transcription factors. This leads to reduced levels of expression of low R/FR-induced genes such as PIL1, ATHB-2 and HFR1, resulting in suppression of the shade-avoidance syndrome. In low R/FR, the phytochrome photo-equilibrium is shifted towards the inactive Pr conformation, which no longer interacts with PIF4 and PIF5. This leads to stabilization of these proteins, which in turn allows a high level of expression of shade-induced genes, resulting in stem and petiole growth.

Experimental procedures

Plant material and growth conditions

Seedlings were grown as described previously (Duek et al., 2004). All mutants are in the ecotype Columbia (Col-0). The mutant in PIF5/PIL6, pil6-1 (SALK-087012), has been described previously (Fujimori et al., 2004); we refer to it as pif5. We obtained a single T-DNA insertion line in the PIF4 gene (based on segregation of the resistance marker) from the Garlic collection (Garlic_114_G06). The T-DNA is inserted in the 5th exon (confirmed by sequencing). No full-length PIF4 mRNA was detected in this mutant. The mutant was backcrossed to Col-0 and has a phenotype that is similar to that described for pif4-1 (Huq and Quail, 2002) (J. Maloof, UC Davis, unpublished observation). It is thus likely that both alleles of pif4 are null. We named this allele pif4-101. pif4 pif5 double mutants were obtained by crossing. pif4 and pif5 were genotyped using a pair of primers to test for the presence of the T-DNA, and a second pair of primers spanning the T-DNA insertion to test for the homozygous state of the mutation. phyB pif4 and phyB pif4 pif5 mutants were obtained by crossing phyB-9 with pif4-101 and phyB pif4 with pif4 pif5, respectively. phyB-9 genotyping was performed as described previously (Neff et al., 1998).

Full-length and ΔN versions of PIF4 and PIF5 in binary plant transformation vectors were generated by PCR using cDNA clones from Kasuza APZ34f03 (for PIF5) (Seki et al., 2002) and M64N03 (for PIF4) as templates, with the primers CF314 (CTAGTCTAGAAAAATGGAACACCAAGGTTGGA) and CF315 (TAGGAGATCTGAGTGGTCCAAACGAGAACC) for PIF4, CF315 and CF316 (CTAGTCTAGAAAAATGCTTGAAGATCAAGAAACTGTC) for ΔN-PIF4, CF319 (CCGGGATCCAAAATGGAACAAGTGTTTGCTGA) and CF320 (CCGGGATCCGCCTATTTTACCCATATGAAG) for PIF5, and CF220 and CF221 (CGGGATCCAAAATGCTTGACAACCAAGAAACAGTA) for ΔN-PIF5. ΔN-PIF4 lacks the first 65 amino acids and ΔN-PIF5 lacks the first 68 amino acids. The PCR products were digested by XbaI (PIF4 and ΔN-PIF4) or BamHI (PIF5 and ΔN-PIF5), and introduced into pFP101 (Bensmihen et al., 2004) with a C-terminal triple-HA tag (Tyers et al., 1992), to generate pCF402 (35S:PIF4-HA), pCF403 (35S:ΔN-PIF4-HA), pCF404 (35S:PIF5-HA) and pCF405 (35S:ΔN-PIF5-HA). Equivalent clones in pPZP-derived vectors (Hajdukiewicz et al., 1994) were also generated in order to use kanamycin resistance rather than GFP fluorescence of the seed as a selectable marker. These constructs were transformed into Arabidopsis Col-0 plants by the Agrobacterium tumefaciens spray method (Weigel et al., 2000). Transformants with a 3:1 segregation ratio were self-fertilized, and homozygous progeny were selected. Over-expressing lines were selected by seed GFP fluorescence or kanamycin resistance. Multiple independent single insertion lines were generated and studied for each construct, and a representative result is shown in each case.

Protein extraction, treatment with proteasome inhibitors, phosphatase treatment and Western blotting

These experiments were performed as described in previously (Duek et al., 2004), except that Roche anti-HA antibodies (3F10; http://www.roche.com) directly conjugated with peroxidase were used to detect HA-tagged proteins. For phosphatase treatments, seedlings were homogenized with 110 μl of ice-cold extraction buffer [50 mm Tris–HCl, pH 7.5, 150 mm NaCl, 1% Na deoxycholate, 0.5% Triton X-100, 1 mm DTT, 10 μl ml–1 protease inhibitor cocktail (Sigma, http://www.sigmaaldrich.com/), 50 mm MG132, 50 mm MG115, 50 mm ALLN1, 50 mm PS1). After centrifugation at 14 000 g for 10 min at 4°C, the supernatant was used for phosphatase treatment with 400 units of lambda protein phosphatase (400 units/μl, NEB; http://www.neb.com) in a final volume of 50 μl for 30 min at 30°C. The reaction was stopped with 25 μl of 4× SDS-PAGE loading buffer. As a control, samples were treated with the phosphatase buffer alone or with the phosphatase in the presence of phosphatase inhibitors (50 mm EDTA).

Shade-avoidance phenotypes and quantitative PCR

Seeds were sown directly onto soil in 5 cm Petri dishes and stored in the dark for 3 days at 4°C. A germinating white-light pulse of 1 h duration was given at 22°C, and the plates returned to dark at 22°C for a further 24 h. Plates were then transferred 12 h light/12 h dark or continuous light in controlled-environment growth chambers maintained at 22°C. High-R/FR-grown plants received a photon irradiance of 400–700 nm at 120 μmol m−2 sec−1 (R/FR = 5). Low R/FR was provided by supplementary FR LEDs (λmax 735 nm, Shinkoh Electronics; http://www.shinkoh-elecs.com) positioned overhead (R/FR = 0.03). Light measurements were performed using a Stellarnet EPP2000 fiber optic spectrometer with a planar sensor (Stellarnet; http://www.stellarnet-inc.com).

Measurement of hypocotyl length and petiole length was achieved using imagej software (http://www.rsb.info.nih.gov/ij). The number of experimental replicates was at least 10, and is representative of at least three biological repeats.

For quantitative PCR experiments, seedlings were grown under high R/FR for 9 days, or 5 days of high R/FR (PAR of approximately 100 μmol m−2 sec−1, R/FR = 2.3) followed by 4 days of low R/FR (R/FR = 0.13). To determine the time course of gene expression, seedlings were grown under 12 h light/12 h dark cycles at high R/FR for 12 days, then transferred to continuous high or low R/FR at dawn of day 13. Samples were taken at specific time points during the next 26 h period. RNA extraction and quantitative PCR were performed as described by Allen et al. (2006). The primers used were PIL1f (AAATTGCTCTCAGCCATTCGTGG) and PIL1r (TTCTAAGTTTGAGGCGGACGCAG), ATHB2f (GAGGTAGACTGCGAGTTCTTACG) and ATHB2r (GCATGTAGAACTGAGGAGAGAGC), HFR1f (TAAATTGGCCATTACCACCGTTTA) and HFR1r (ACCGTGAAGAGACTGAGGAGAAGA), and ACTINf (TCAGATGCCCAGAAGTGTTGTTCC) and ACTINr (CCGTACAGATCCTTCCTGATATCC).

Acknowledgements

We thank Nicolas Roggli (University of Geneva) for artwork, Martine Trévisan, Arianne Honsberger and Dany Rifat for technical support, and Karin Schumacher (Universität Tübingen) for antibodies against DET3. We are grateful to Keara Franklin for carefully reading this manuscript. We thank the SAIL project (Syngenta) and the Arabidopsis Stock Center for providing GARLIC and SALK T-DNA insertion lines, respectively, and the RIKEN GSC Arabidopsis full-length cDNA program and the Arabidopsis Stock Center for cDNA clones. This work was supported by grants from the Swiss National Science Foundation (PP00A-103 005 and 3100A0-112 638) and the HFSP program (to C.F.), a Roche post-doctoral fellowship (to S.L.) and by the UK Biotechnology and Biological Sciences Research Council (to G.C.W.).

Ancillary