The HY5-PIF Regulatory Module Coordinates Light and Temperature Control of Photosynthetic Gene Transcription

The ability to interpret daily and seasonal alterations in light and temperature signals is essential for plant survival. This is particularly important during seedling establishment when the phytochrome photoreceptors activate photosynthetic pigment production for photoautotrophic growth. Phytochromes accomplish this partly through the suppression of PHYTOCHROME INTERACTING FACTORS (PIFs), negative regulators of chlorophyll and carotenoid biosynthesis. While the bZIP transcription factor LONG HYPOCOTYL 5 (HY5), a potent PIF antagonist, promotes photosynthetic pigment accumulation in response to light. Here we demonstrate that by directly targeting a common promoter cis-element (G-box), HY5 and PIFs form a dynamic activation-suppression transcriptional module responsive to light and temperature cues. This antagonistic regulatory module provides a simple, direct mechanism through which environmental change can redirect transcriptional control of genes required for photosynthesis and photoprotection. In the regulation of photopigment biosynthesis genes, HY5 and PIFs do not operate alone, but with the circadian clock. However, sudden changes in light or temperature conditions can trigger changes in HY5 and PIFs abundance that adjust the expression of common target genes to optimise photosynthetic performance and growth.


Introduction
Light and temperature are prominent cues that signify seasonal and climatic change, as well as the phase of the daily light/dark cycle. The ability to sense and integrate these external signals is essential for plant life cycle progression and ultimately, survival. Central to this process is organization of photosynthetic machinery that captures energy from light. This vital process provides energy to fix carbon into organic matter that is essential for plant growth and development.
Light quality and quantity is sensed by a suite of specialized photoreceptors. An established class of these receptors are the red (R) and far-red (FR) light absorbing phytochromes that have an important role driving the switch from skotomorphogenic to photoautotrophic growth during seedling de-etiolation (reviewed by [1]). The phytochromes exist in two interconvertible forms. When the inactive-Pr form absorbs R light, it converts to the active, Pfr form and relocates from the cytoplasm into the nucleus. Once in the nucleus, phytochromes induce major changes in gene expression that modify the developmental program [2]. Amongst the phytochrome-mediated changes during this critical time is the promotion of the photosynthetic apparatus assembly, and the production of the photosynthetic pigments, chlorophylls and carotenoids [3][4][5] While chlorophyll production is required for photoautotrophic growth, excessive accumulation of the chlorophyll precursor protochlorophyllide can lead to harmful photooxidative damage [6][7][8]. It is therefore critical that this process is stringently regulated. Carotenoids are integral accessory pigments in the light absorbing antenna. In the greening process, carotenoids play an essential photoprotective role minimizing the potentially damaging effects of light on the photosynthetic machinery [9][10][11]. Therefore, to maximize light capture but curtail the potentially damaging effects of light over the emerging photosynthetic machinery, the production of carotenoids and chlorophylls has to take place in a tightly controlled and interdependent manner. This might be achieved in part by the circadian gating of isoprenoid pathway genes, that produce both carotenoids and chlorophylls, to ensure coordinated pathway gene expression [12]. However, we do not currently understand how this regulation is influenced by either the regular daily or unexpected variations in external light and temperature.
Earlier studies have shown that bHLH-Phytochrome Interacting Factors (PIFs) not only prevent the over-accumulation of protochlorophyllide in etiolated seedlings [8] but have a broader role in the seedling as repressors of photomorphogenic development [1,13]. In accordance, a PIF quadruple mutant (pif1-1;pif3-3;pif4-2;pif5-3) (pifQ) exhibits a constitutive photomorphogenic phenotype in darkness and a global gene expression pattern that resembles R light-grown seedlings [1,[14][15][16]. Concurring with these observations, we previously showed that PIFs repress carotenoid accumulation by down-regulating the expression of the Arabidopsis thaliana gene encoding PHYTOENE SYNTHASE (PSY), the main rate-determining enzyme in the carotenoid pathway [17]. PIF1, a potent PSY gene repressor, effects control through direct binding to at least one of the the two G-box (CACGTG) elements found in the PSY promoter. Indeed, PIFs have been shown to preferentially target G-box motifs in a range of genes [8,[18][19][20][21]. Suppression of PSY transcription is strongest in etiolated seedlings when PIF are abundant, while exposure to light leads to a depletion in PIF levels and concomitant de-repression in PSY expression [17].
PIFs are strongly regulated by light and therefore their activity is conditional on the external light environment. In de-etiolated seedlings growing in light/dark cycles phytochromes control the degradation and re-accumulation of PIF proteins (PIF3, PIF4 and PIF5) generating a daily alternation in the abundance of these transcription factors [22][23][24]. For PIF4 and PIF5, additional temporal control is delivered by the circadian clock that regulates the phase of expression [23,[25][26][27]. As these PIFs are potent regulators of cell expansion in the seedling hypocotyl, this translates to a diurnal rhythm in hypocotyl elongation rate [23,25]). Recent studies have shown that PIFs also participate in temperature signalling. PIF4, and to a lesser extent PIF5, are required for the promotion of growth in warmer conditions [28,29]. Elevated temperatures promote the accumulation of phosphorylated PIF4 and enhance PIF4 binding at target promoters [30][31][32]. Indeed, PIF4 occupation at the FLOWERING TIME (FT) promoter increases with the eviction of H2A.Z nucleosomes by temperature [32].
For a broad range of responses PIFs act antagonistically with the bZIP transcription factor, HY5 [1,33,34]. HY5 drives photomorphogenic development by activating genes that promote photosynthetic machinery assembly, photopigment production, chloroplast development, and seedling cotyledon expansion [21,35], while PIF suppression of these responses is required to maintain skotomorphogenesis [1,33]. HY5 has been shown to regulate gene expression directly through interaction with ''ACE'' motifs (ACGT containing elements) that include the Z-box (ATACGTGT), C-boxes (GTCANN), G-boxes, and hybrid C/G (G) and C/A boxes [35][36][37][38]. Like PIFs, HY5 protein levels are light regulated -but in contrast to PIFs that are light labile, HY5 protein is stabilized by light [1,39,40]. Multiple photoreceptors participate in the accumulation of HY5 protein in the light, in part by reducing the nuclear levels of CONSTITUTIVE PHOTO-MORPHOGENIC 1 (COP1), an E3 ubiquitin ligase that targets HY5 for proteasome mediated degradation in the dark [39][40][41]. The differential regulation of HY5 and PIF proteins by light means that they exhibit opposing diurnal expression patterns [23,24,35,42].
Whereas PIF4 mediates warm temperature signalling, HY5 has been shown to have a role in cold temperature responses [43]. Here HY5 has been implicated in cold acclimation where it promotes anthocyanin accumulation, ROS production, and regulates a large swathe (,10%) of cold inducible genes [43][44][45]. HY5 abundance is regulated by temperature. A shift in temperature from 20uC to 4uC elevates HY5 transcript levels and stabilises the HY5 protein through nuclear depletion of COP1 [43].
This study examines the impact of light and temperature on the control of photopigment production by HY5 and PIFs. ChIP, EMSA and transcript analysis established that HY5 and PIF1/ PIF4 impart antagonistic regulation to common gene targets through direct binding to the same G-box cis element. This dynamic activation-suppression transcriptional module does not act alone, but with the circadian clock to modify the level of rhythmic gene expression. However, abrupt changes in either light or temperature adjusts the equilibrium of HY5 and PIF bound to target promoters altering the transcriptional response. In this way the expression of photopigment biosynthesis genes can be revised by external signals.

HY5 is required for light induction of PSY expression, and photopigment synthesis
Our earlier work showed that PIFs restrict the accumulation of carotenoids during de-etiolation, in part by negatively regulating PSY, a rate limiting step in the carotenoid biosynthetic pathway [17]. We demonstrated that sequential removal of PIFs leads to incremental rises in PSY mRNA levels, illustrating that PIFs act redundantly to suppress PSY expression. This data also suggested the existence of a PSY activator that becomes more effective with PIF depletion. As HY5 is known to act in opposition to PIFs for a number of photomorphogenic responses [7,14,37,46], we tested whether HY5 fulfilled this role by quantifying PSY gene expression in the hy5 mutant (hy5-215) during red light-triggered de-etiolation ( Figure 1A). Our data confirmed that hy5 perturbs PSY accumulation following exposure to light. Likewise, red light induction of carotenoid and chlorophyll levels is severely attenuated in the hy5 mutant ( Figure 1B-C). This response contrasts with that in pif1 and pifQ mutants, where PSY mRNA and photopigment levels are significantly elevated in dark and red light illuminated seedlings [17]. Because mutants defective in the HY5 homologue HYH did not exhibit photopigment accumulation defects under our conditions, we therefore focussed our analysis on HY5 ( Figure  S1). Our data illustrate that HY5 and PIFs have opposing roles in the regulation of PSY expression and carotenoid and chlorophyll biosynthesis during seedling de-etiolation.

Author Summary
Plants, as sessile and photosynthetic organisms, have to constantly adjust their growth and development in response to the environment. While light and temperature are recognized as the most prominent environmental factors modulating plant photosynthetic metabolism, how the seasonal and daily adjustments are achieved is not understood. Global climate alterations will bring together the combination of light and temperature changes and will require an understanding of signal convergence. If we are to mitigate the impact of variable weather patterns on agriculture, it is critical to advance our understanding of the basis of plant responses to environmental variations. In our study we show that the antagonistic activity of key plant transcription factors involved in phytochrome red light photoreceptors signaling (PIFs and HY5) optimize photosynthetic pigment production in response to environmental cues. These light and temperature responsive transcription factors operate in cooperation with the circadian clock to regulate photosynthetic pigment production through a common gene promoter element.

HY5 binds to the PSY promoter through a G-box element
Since HY5 and PIFs confer antagonistic regulation to PSY, we wanted to establish whether this was mediated at the PSY promoter. Previous chromatin immunoprecipitation (ChIP)-chip analysis indicated that HY5 binds to promoter regions that carry different types of the generic ACE motif (CACGT) including CA hybrid (GACGTA), CG hybrid (GACGTG), Z-boxes (ATACTGTGT), as well as G-boxes (CACGTG), the main recognition motif for PIFs [17,21,35,37]. The PSY promoter contains two G-boxes (I and II), separated by 80 base pairs, but not CG, CA or Z type motifs ( Figure S2 and Figure 1D). We therefore reasoned that the opposing actions of HY5 and PIF1 on PSY expression could be mediated through the same regulatory Gbox element. This we tested using an in-vitro electrophoretic mobility shift assay (EMSA). Previously we showed that PIF1 did not bind in-vitro to G-box II [17], therefore our analysis is focussed exclusively on G-box I. Consistent with our previous results we detected PIF1 binding to a template encompassing the PSY G-box I ( Figure 1D, E) [17]. In the same assay, we were able to demonstrate that like PIF1, HY5 bound to a labelled probe containing this G-box, but not a mutated version (G-mut I) where the G-box had been disrupted ( Figure 1E).
To establish if we could verify our findings in-vivo we conducted a ChIP assay using the 35S::HA-HY5 lines used in a previous study that identified genome wide the targets of HY5 regulation [35]. After immunoprecipitation of protein-DNA complexes using the HA antibody, enriched DNA sequences were amplified by qPCR using primers bordering the PSY promoter G-box region. For the assay, radiolabelled probes were incubated with TnT produced PIF1 and HY5 proteins. G-box I or G-mut I probes were used as indicated. Asterisk (*) corresponds to a TnT non-specific band and FP stands for free probe. (F) HY5 binds in-vivo in Chromatin immunoprecipitation (ChIP) assays to the G-box region in the PSY promoter. ChIP assays were carried out using a 35S:: HA-HY5 line with antibodies against the HA-tag. A 35S::GUS-HA line was used as a control (+HA antibody samples). A second set of samples were processed without antibody and used as negative controls (-HA antibody). Immunoprecipitated DNA was amplified by qPCR using primers against the G-box region in the PSY promoter (see Table S1). As shown in Figure 1F, in contrast to the 35S::GUS-HA and the no-antibody controls, we detected significant enrichment of PSY promoter sequences containing G-box I in 35S::HA-HY5 samples. These results together with our in vitro analysis suggest that HY5 regulates PSY transcription through direct binding to the G-box region of the promoter.
HY5 enhances carotenoid and chlorophyll pigment synthesis at cooler ambient temperatures As PIFs (e.g. PIF4 and PIF5) and HY5 have been implicated in temperature signalling we wanted to assess whether this action extended to carotenoid and chlorophyll regulation [28][29][30][31][43][44][45][47][48][49][50][51]. We found that in both etiolated and red light exposed wild type seedlings carotenoid and chlorophyll levels rose incrementally with temperature (17uC, 22uC and 27uC) (Figure 2A-C). Compared to the wild type, the pifQ mutant had constitutively elevated carotenoid and chlorophyll levels under all conditions, but still remained responsive to temperature (Figure 2A, B). This suggests that while specific PIFs (e.g. PIF4) may operate at warmer temperatures, the collective action of PIFs maintain control of carotenoid and chlorophyll levels over a temperature range [28,[30][31][32]47,49] In contrast, hy5 suppressed red light-induction of carotenoid and chlorophyll levels ( Figure 2A-C). While the effects of hy5 were apparent across temperatures, its impact was most marked in the cooler 17uC conditions. We next measured HY5 protein levels in etiolated and deetiolating seedlings at 17uC and at 27uC. Concurring with published work, our data show that HY5 transcript and protein levels increase steadily following exposure to 1 h then 6 h red light ( Figure 2D, Figure S3) [39]. In the dark, HY5 transcript and protein levels were low and not particularly affected by temperature in our study range. However, we detected higher levels of HY5 transcripts and protein at 17uC when compared to 27uC, after exposure to -red light ( Figure 2D, Figure S3). Thus, HY5abundance correlates well with both the light and the temperature requirement for HY5 control of photopigment levels (Figure 2A-C).

HY5 and PIFs regulate a subset of photosynthetic pigment genes through a common mechanism
To establish whether G-box element convergence represented a generic mechanism through which HY5 and PIF operate, we tested the impact of pifQ and hy5 mutations on the expression of other genes central to carotenoid or chlorophyll biosynthesis and function with G-box elements in their promoters. These genes were chosen from genome wide transcriptional analyses specific to PIFs or HY5 [14,21,35,44]. To rule out binding to other promoter elements, the selected genes did not possess any of the alternative high affinity binding sites for HY5: C, Z or C/A, G/A or G/A boxes ( Figure S1). For carotenoid biosynthesis, in addition to PSY, we selected VIOLAXANTIN DE-EPOXIDASE (VDE) that is involved in the conversion of violaxanthin to zeaxanthin for optimal photoprotection in the xanthophyll cycle [52]. Chlorophyll biosynthesis genes are represented by PROTOCHLORO-PHYLLIDE OXIDOREDUCTASE C (PORC), that phototransforms endogenous protochlorophyllide to chlorophyllide, GENOMES UNCOUPLED 5 (GUN5) that encodes the ChlH subunit of Mgchelatase, a key enzyme in the chlorophyll branch of the tetrapyrrole pathway [53,54], and the Photosystem I LIGHT-HARVESTING COMPLEX 4 (LHCA4) [55,56].
The transcript levels of these genes tightly correlated with the carotenoid and chlorophyll accumulation data (compare Figure 3 with Figure 2A, B). Consistently, mRNA levels were elevated in dark or light-grown pifQ at 17uC and 27uC, while hy5 led to reduced red light induction of each target gene, particularly at 17uC. To establish whether HY5 and PIFs regulate this suite of genes through common G-box elements we conducted ChIP analysis with lines expressing 35S::HA-HY5, 35S::PIF1-TAP and 35S::PIF4-TAP. In these experiments seedlings were grown in more natural 12L:12D cycles sampling at 2 or 3 h post dawn (T2, T3) or 8 h post dusk (T20) (Figures 4 and Figure S2). Levels of binding to the VDE promoter, were very low which prevented detailed analysis (data not shown). However, we detected enrichment of HY5, PIF1 and PIF4 relative to controls, at G-box containing regions of PSY, LHCA4, PORC and GUN5 promoters. G-box cis element convergence therefore appears to represent a common mechanism through which HY5 and PIFs regulate some carotenoid and chlorophyll biosynthetic genes (Figure 4).

PIF1 action is temperature regulated
Consistent with the more prominent role for HY5 at cooler temperatures ( Figure 2) we detected an enrichment of HY5 at each promoter at 17uC compared to 27uC (Figures 4, and Figure S4). This may reflect a change in HY5 protein abundance that is detectable after exposure to 6 h of red light ( Figure 2D). Previously we showed that heat leads to an accumulation of phosphorylated PIF4. This process appears to be extremely temperature sensitive as stepped increases in temperature lead to incremental rises of modified forms of PIF4 (Figure S5 A). We also observed increased binding of PIF4 to PSY, LHCA4 and GUN5 at 27uC compared to 17uC, but this was relatively modest [30] (Figures 4 and Figure S4). Interestingly, we detect slightly elevated protein levels and enhanced PIF1 promoter binding at 27uC suggesting that PIF1 is also temperature-regulated (Figures 4, Figure S4 B and Figure  S5). In support of this notion our genetic data illustrates that like pif4pif5, pif1 has a greater impact on carotenoid and chlorophyll accumulation at 27uC compared to 17uC ( Figure S6). Our results show the relative binding, particularly of HY5 and PIF1, at the Gbox motifs in the promoters of several photopigment genes varies according to the ambient temperature regime.

HY5 and PIFs exhibit opposite diurnal shifts in promoter binding
Analysing binding during the daytime compared to the night allowed us to establish whether the documented diurnal changes in HY5, PIF4 and PIF1 abundance [16,[22][23][24]35,42] led to corresponding changes in binding. The levels and the proportion of HY5 bound to PSY, LHCA4, PORC and GUN5 promoters was greater at T2 than T20 (Figures 4, Figure S4 and Figure S5 C). PIF1 had the converse response with less bound at T2 (when PIF1 is less abundant) than T20 (Figures 4, Figure S4 and Figure S5 B). This trend was also evident, but less marked, for PIF4. This may reflect less dramatic diurnal changes in 35S::PIF4, the line used in this study, compared to 35S::HY5 and 35S::PIF1 (Figures 4, Figure S5 D and Figure S4). We then conducted EMSA to test whether a change in protein abundance is sufficient to drive a switch in binding to the G-box promoter fragment. In this assay the ''primary'' protein (either HY5 or PIF1) was incubated with a fixed amount of the G-box I probe and then the ''challenge'' protein was added at increasing concentrations ( Figure S7). We could not detect a switch in binding when an equimolar amount of either ''challenge'' protein was added. However, binding was detected with increased concentrations of ''challenge'' protein, with a correlative decrease in ''primary'' protein binding. This indicates that when provided in excess HY5 or PIF1 can indeed prevent the other protein from binding.

HY5 and PIFs regulate photopigment biosynthetic genes in cooperation with the circadian clock
Through a red/dark cycle at 17uC, wild type expression profiles of PSY, LHCA4, GUN5 and PORC are rhythmic with differing phases of expression, suggesting underlying regulation by the circadian oscillator ( Figure 5). The hy5 and pifQ mRNA profiles appear to approach or on occasion converge with the wild type at distinct phases in the cycle, suggesting that HY5 and PIF control of this gene set is gated by the clock (Figure 5). The contrasting day/ night shifts in HY5 vs PIF1 binding to target promoters (Figure 4, Figure S4) inferred that the hy5 mutation should be more effective during the daytime and pifQ more potent at night. Unexpectedly, we found that this is not the case, as both hy5 and pifQ cause a shift in transcript levels during the day and the night ( Figure 5). This indicates that during a diurnal cycle there is not a simple relationship between our recorded changes in promoter binding and the transcriptional response. It also indicates that HY5 and PIFs operate through the light/dark cycle to regulate this gene set. Interestingly, HY5 transcript levels are elevated in the pifQ mutant, while PIF4 and PIF5 (but not PIF1 or PIF3) mRNA levels are raised in hy5 at T2 and T20 ( Figure S8). Such cross-regulation is predicted to increase the amplitude of expression in pifQ and conversely, reduce the amplitude in hy5 ( Figure 5). In support of this notion, at least for PSY, LHCA4 and GUN5 the data show amplitude differences in pifQ compared to hy5 ( Figure 5).
The data presented suggest that HY5 and PIFs regulate gene expression in cooperation with the clock. This is further supported by the observation that the cross-regulated genes, HY5, PIF4 and PIF5 exhibit strong diurnal regulation, whereas PIF1 and PIF3 are not rhythmic and are not subject to cross-control (Figure S8 B) [22,24]. A non-antibody control sample was processed in parallel in each case (white bars). Immunoprecipitated DNA was analysed by qPCR using specific primers covering the G-box containing region in the promoters for the indicated genes (see Table S1 for primer information). The assay was carried out in triplicates. Error Bars represent 6SE. Collectively, our analysis shows that HY5 and PIFs do not exhibit strong diurnal shifts in activity, but operate with the circadian clock to moderate rhythmic gene expression through the light/dark cycle.

Abrupt changes in HY5 and PIF protein abundance control binding to target gene promoters
The observation that the relative action of HY5 and PIF did not change during a diurnal cycle appeared to be at odds with our finding that light or temperature stabilisation of HY5 led to concomitant rises in binding activity and gene regulation (Figures 1,  3, 4 and Figure S4). This suggested that a sudden or sizeable increase in HY5 levels relative to PIFs may activate gene expression.
To test this hypothesis we next manipulated HY5 and PIF1 levels in vivo to establish whether we observed corresponding changes in promoter binding and gene regulation. In these experiments plants were grown in 17uC diurnal cycles and protein levels were measured 3 h into the darkness in controls or following an end-of-day FR (EOD-FR) pulse which deactivates phytochrome [57]. This treatment led to a concurrent rise in PIF1 protein levels and fall in HY5 protein levels (Figure 6 A). In response to EOD-FR we observed correlative alterations in HY5 and PIF1 binding to the PSY, LHCA4, GUN5 and PORC promoters (Figure 6 B-E). We then tested whether EOD-FR led to predictable changes in gene regulation. PSY and GUN5 were unresponsive to the EOD-FR treatment ( Figure S9). This result was unsurprising for PSY, as the EOD-FR timing coincided with the phase in which PIFs and HY5 are not fully engaged in transcriptional regulation. Likewise, the contribution of PIFs and HY5 to GUN5 transcriptional regulation was relatively low when compared to the other genes ( Figure 5). Nonetheless, for LHCA4, PORC and VDE, the EOD-FR treatment elicited a decrease in transcript levels (Figure 7), a response that is consistent with the observed changes in PIF vs HY5 protein abundance and promoter binding.

HY5 regulates photosynthetic capacity at cooler temperatures
Our data illustrate that HY5 acts antagonistically with PIFs to control photosynthetic pigment genes through a common cis  element, particularly at cooler temperatures. As this work was conducted in seedlings we wanted to establish the significance of this regulation in older plants. We tested this by measuring photosynthetic capacity in WT and hy5 adult plants (grown in 12L:12D) shifted to either 17uC or 27uC from 22uC. The hy5 mutant had lower chlorophyll levels and CO 2 net flux m 22 s 21 compared to WT, and this difference is more marked at 17uC vs 27uC (Figure 8 A-B). Our previous studies of pifQ demonstrated that PIFs also influence the accumulation of carotenoids and chlorophylls under photoperiodic conditions [17]. It therefore appears that the HY5 and PIF driven transcriptional switch is important for controlling the production of photosynthetic pigments beyond the seedling stage. The temperature dependence of the hy5 mutation on chlorophyll levels and CO 2 uptake was evident seven days following the transfer from 22uC to either 17uC or 27uC. This indicates the process is dynamic and that HY5 is required to actively maintain chlorophyll content in response to changing external light and temperature signals. The impact of hy5 on carbon assimilation may have consequences for growth as fresh weight is markedly reduced in hy5 mutants at 17uC but not at 27uC (Figures 8 C and D).

HY5 and PIFs co-regulate photosynthetic genes at common G-box motifs
Despite the knowledge that light and temperature are among the most relevant environmental signals modulating photosynthetic pigment production, there is no integrative view on how these adjustments are achieved. We have shown that by responding to external light and temperature signals, HY5 opposes PIF action to deliver environmental control of common target genes involved in photosynthesis and photoprotection. This dual control system appears to operate in conjunction with the circadian oscillator to adjust levels of rhythmic photosynthetic gene expression.
Our previous work, and that of others, identified the PIFs as negative regulators of chlorophyll and carotenoid accumulation [8,17,51,58]. However, optimization of these essential responses would require an additional regulator that acts antagonistically to PIFs. Here we have shown that HY5 fulfils this role as a positive regulator of photosynthetic pigment synthesis (Figure 1). Our earlier research demonstrated that PIFs control the expression of PSY, a rate limiting step in carotenogenesis. This is achieved by directly binding to G-box motifs in the PSY promoter [17]. Previous genome wide gene expression and ChIP-CHIP analyses highlighted an overlap in PIFs and HY5 gene targets and identified G-boxes as potential HY5 targets [7,35,36,44,59]. We have shown in-vitro and in-vivo that HY5 binds to the same G-box containing region targeted by PIF1 in the PSY promoter ( Figure 1). Our data indicates that HY5 and PIFs act antagonistically to control PSY gene expression via a common cis element. This regulation is not confined to PSY, as we presented evidence that other central carotenoid and chlorophyll pathway genes are regulated through the same dual input mechanism. HY5 and PIFs do not always act antagonistically. In the case of anthocyanin biosynthesis, PIF3 and HY5 both positively regulate the pathway by activating transcription of the same biosynthetic genes by binding to distinct cis-promoter elements (G-box and another ACE motif respectively) [19]. In this instance PIF3 binding is facilitated by the presence of HY5, suggesting cooperativity of action. However, our analysis of photopigment gene regulation, together with a recent study focussing on REACTIVE OXYGEN SPECIES genes, indicates that when HY5 and PIF act in opposition, G-box convergence is a common mechanism [7].

Coaction of HY5 and PIF antagonists fine tune rhythmic photopigment gene expression
Light has opposing effects on HY5 and PIF levels and/or activity [1,34,60,61]. HY5 protein accumulates in the light and is degraded in the dark [39,41]. In contrast, PIFs accumulate in the dark to promote dark development, and light induces rapid phytochrome dependent phosphorylation, degradation and deactivation [22,60,[62][63][64]. Accordingly we established that HY5 promotes the expression of PSY, VDE, PORC, LHCA4 and GUN5 genes following exposure to light, while PIFs strongly suppress the transcription of these genes in etiolated seedlings (Figure 3).
In diurnal cycles we showed that the levels of HY5, PIF1 and PIF4 bound to common G-box elements correlated with times that these proteins are abundant (Figures 4, Figure S4 and Figure S5). Furthermore, in vitro EMSA assays demonstrated that binding of either HY5 or PIF1 to the G-box region of the PSY promoter was reduced when the opposing challenge protein was provided in excess ( Figure S7). This, analysis illustrated that the change in relative HY5:PIF promoter binding could at least partly be driven by alterations in protein abundance. However, when we analysed the transcript profiles of wild type plants compared to hy5 and pifQ mutants, there was no obvious correlation between the shift in binding dominance and the relative effectiveness of HY5 and PIFs through the light/dark cycle ( Figure 5). A potential explanation is that cross-regulation between HY5 and PIFs may dampen the diurnal swings in response. Such a mechanism has been recently reported by Chen and coworkers [7]. This study showed that PIF1 and PIF3 physically interact with HY5 and its homologue HYH. Through this interaction these antagonistic pairs of transcription factors, PIF1/PIF3 and HY5/HYH, moderate each other's control of ROS-responsive genes. It is therefore possible that this cross-control is a more broadly utilised ''buffering'' mechanism that dampens antagonistic HY5-PIF responses. Alternatively, an independent regulator such as the circadian clock may be modulating the level or activity of HY5 or PIFs at the promoter. Indeed, we showed that HY5/PIF control of PSY, PORC, LHCA4 and GUN5 expression is gated by the circadian clock ( Figure 5). These findings are consistent with earlier work that shows that the HY5 protein physically interacts with CCA1, which in turn enhances the binding of HY5 to the LHCB1*1 promoter [65]. It is not yet known how PIFs interface with the oscillator. However, an in vitro interaction between PIF3 and TOC1/PRR1 has been reported [66,67], and GUN5, a gene identified as a HY5/PIF target in this study, was previously shown to be directly and negatively regulated by TOC1 [68][69][70]. PRR9, PRR7 and PRR5 are also reported to be negative regulators of chlorophyll and carotenoid biosynthesis [71]. Thus, PIFs may regulate specific gene targets in concert with TOC1/PRR1, other PRRs or their regulators. Further analysis will be required to evaluate these and other possible links between HY5, PIFs and the clock.

Changes in protein abundance alter HY5/PIF promoter binding
The diurnal switch between HY5-and PIF-dominant promoter binding did not lead to correlative diurnal shifts in transcriptional regulation (Figures 4, 5 and Figure S4). However, a sudden change in HY5 or PIF protein abundance in-vivo did elicit a matching alteration in gene expression in LHCA4, PORC and VDE (Figure 7). Supplying an EOD-FR pulse, which deactivates phyB, induced a simultaneous fall in HY5 and rise in PIF1 protein abundance, with correlative alterations in promoter binding and transcript regulation (Figures 6 and 7). These results illustrate that abrupt changes in HY5 or PIF levels can lead to corresponding adjustments in signalling. Sizeable adjustments in HY5 and PIF levels and activity are known to occur in etiolated seedlings following exposure to light, and are predicted in plants exposed to abrupt changes in temperature [1,28,29,35,41,43,61,62,72,73].
Environmental temperature signals are delivered through HY5 and PIF1 Earlier work illustrated that HY5 is important for cold acclimation responses and enhancement of freezing tolerance [43][44][45]. At 4uC HY5 protein levels were shown to stabilise, particularly in the dark [43]. We have shown that HY5 also has a prominent role at the more moderate 17uC, but here its action depends on light. Exposure of etiolated seedlings to light leads to a rapid depletion in PIF levels and increase in HY5 ( Figure 2D) [1,16,35,41,62]. The sudden change from PIF to HY5 dominance induces a switch from transcriptional repression to activation of target photopigment genes (Figure 3). This switch is more robust at 17uC compared to 27uC.
PIF4 is known to activate genes in a temperature dependent manner [30][31][32]47,50,74]. PIF4 levels and its binding to target promoters are boosted by warm temperature [30][31][32]. In line with these studies we observed enhanced PIF4 enrichment at carotenoid and photosynthetic gene promoters under warmer conditions. Interestingly, PIF1 levels and binding to target promoters is also increased at 27uC compared to 17uC (Figures 4 and Figure S4). Our genetic data also show that PIF1 has a greater influence on carotenoid and chlorophyll levels at 27uC in etiolated seedlings ( Figure S6). Thus, in the regulation of photosynthetic pigment synthesis, PIF1 action appears to be temperature dependent. The temperature response is particularly evident in etiolated seedlings as PIF protein levels are very high compared to HY5, and therefore PIF signalling predominates. Our data also show that the pifQ mutant has perturbed responses at 17uC and 27uC, suggesting that collectively, PIF1, PIF3, PIF4 and PIF5 operate over a broad temperature range (Figures 2 and 3).

HY5 maintains photosynthetic capacity at lower temperatures through development
We have shown that HY5 is not only important in developing seedling, but as for PIFs, it continues to regulate photopigment levels in adult plants (Figure 8) [17]. Our data demonstrate that HY5 plays a role in maintaining chlorophyll levels and CO 2 uptake. Interestingly, our analysis indicates that HY5 has a greater impact in plants transferred (from 22uC) to 17uC compared to 27uC (Figure 8 A-C). This illustrates that maintenance of chlorophyll pool is very dynamic and that HY5 is required to augment chlorophyll synthesis and carbon uptake, particularly when temperatures fall. Furthermore at 17uC the fresh weight of hy5 is markedly reduced -compared to wild type plants at 17uC but mot at 27uC (Figure 8 C-D).HY5 has been shown to modulate several aspects of plant hormones pathways including auxins, giberellins and absicic acid signal transduction. Therefore, it is likely that alterations in hormone signalling will contribute to hy5 adult phenotype [34] However, the reduction in photosynthetic capacity observed in hy5 at 17uC (Figure 8) may also compromise growth.
This paper illustrates that signal convergence of antagonistic regulators HY5 and PIFs at a shared cis regulatory element provides an effective mechanism to integrate light and temperature signals. A similar type of control has been reported for the endoreduplication E2Fb and E2Fc transcription factors which antagonistically regulate DEL1 expression through a common ciselement [75,76]. Differential regulation of E2Fb vs E2Fc protein levels by light, determines relative binding capacity and the level of DEL1 activation [77]. It is therefore possible that the single cis element activation-inactivation module is a prevalent signalling mechanism through which external signals can change or finetune transcriptional responses.

Plant material and growth conditions
Columbia-0 (Col-0) wild type and mutants were used for experiments. The mutant alleles corresponded to: hy5-215, hyh (GK-57200610-N323769), pif4pif5 (pif4-2, pif5-2), pif1-1, pifQ (pif1-2, pif3-3, pif4-2, pif5-2). Over expressing plants included 35S::PIF4-HA, 35S::HA-HY5 and 35S::TAP-PIF1. All have been previously described [14,35,58,78,79]. Seeds were surface sterilized, sown in GM-agar media and stratified in darkness for 3 days at 4uC before given a 3 h white light pulse to induce germination. For deetiolation experiments (qPCR and photosynthetic pigment measurements), 3 d-old seedlings were used. Seedlings were kept in the dark for 2 days at 22uC and transferred to the indicated temperature for 1d before exposure to red light (40 mmol m 22 s 21 ). For Red Diurnal experiments plants were grown for one week under white diurnals (12 h light/12 h dark, 100 mmol m 22 s 21 ) at 22uC and then were transferred for 7 days to Red diurnal cycles at 17uC before sample collection. In the case of ChIP experiments, samples were collected at T2 for PIF1 and HY5 and T3 for PIF4. Times were selected based on moments where the proteins have started to recover following phyB degradation (in the case of the PIFs, that starts at T0) and when levels were comparable among all proteins (in the case of HY5) ( [80], [81], [22]).

Carotenoid and chlorophyll measurements
For the experiments, 3 d-old etiolated seedlings were used and treated as indicated above. Total carotenoids and chlorophylls were extracted and quantified spectrophotometrically as described by [17]. Concentration was expressed per sample fresh weight and measured in biological triplicates.

RNA isolation and transcript levels analysis by qPCR
For deetiolating experiments quantitative qPCR seedlings were prepared and sown as previously described, and grown for 2 days in the dark at 22uC and 1 day at the indicated temperature before redlight illumination (as indicated above). For diurnal expression analyses, 2 week old seedlings were used. Seedlings were grown under white light 12L:12D cycles (80 mmol m 22 s 21 ) at 22uC for one week before being transferred to 17uC red light 12:12 dark/light cycles (40 mmol m 22 s 21 ) for one week and harvesting at the indicated time points. Samples were collected in RNAlater (Sigma). RNA was extracted with RNeasy Plant Mini kit (Qiagen). cDNA synthesis was performed using the SuperScript VILO cDNA synthesis kit (Invitrogen). The qPCR was set up with a liquid handling robot (Tecan Freedom EVO) and qPCR performed with a LightCycler 480 (Roche). All samples were processed in biological triplicates.

Electrophoretic Mobility Shift Assays (EMSA)
EMSA were conducted as described in [18]. In brief HY5 and PIF1 proteins were produced in an in-vitro transcription and translation system (TnT) (Promega) and incubated with a PSY promoter fragment generated by annealed oligonucleotides containing the G-box motif labelled with 32 P-dCTP.

Chromatin Immunoprecipitation assays
ChIP assays were conducted according to [58]except that 2 week old plants were used for the assay. Unless otherwise stated, plants were grown for one week in 12:12 white light diurnal cycles at 22uC and moved to one week growth in red 12:12 diurnal cycles at the testing temperature. Samples were harvested at the indicated time points during the diurnal cycle. For the end-of-day FR experiment, a saturating FR pulse (3000 mmol total) was given at the end of the day and samples harvested 3 h after the pulse. A non-FR treated sample was harvested at the same time point. The sequence of the primers used in these experiments to amplify G-box containing promoter regions of individual genes is shown in Table S1.

Immunoblots
Total proteins were extracted from 100 mg of tissue in a buffer containing 100 mM Tris-HCl pH 8, 50 mM EDTA, 0.25 M NaCl, 0.7% SDS and 1 mM DTT. Samples were heated at 65uC for 10 minutes and then centrifuged at maximum speed to remove debris. Total protein was quantified by Bradford. 30 mg of total protein were loaded. Samples were run in a 10% SDS gel, followed by wet transfer to nitrocellulose. The HA-or MYC-tag were detected by probing with rat-anti HA horseradish peroxidase HRP coupled HA-antibody (3F10 Roche) or an Anti-MYC mouse antibody (mAb 9E10, Calbiochem) at a 1:5000 dilution followed by HRP-conjugated anti-mouse antibody at 1:10,000 dilution. Loading was confirmed by reprobing the membranes with an anti-goat UGP-ase antibody (Agrisera) at 1:1000 dilution followed by a HRP-conjugated sheep anti-goat antibody (Biorad) at a 1:5000 dilution. Signal was detected with Amersham ECL kit (GE Health care) according to the manufacturer's protocol. Quantification was performed using Image-J software.

CO 2 assimilation measurements and chlorophyll measurements
For adult plant analysis seedlings were sown as indicated in plant material and grown for 1 week under white light 12 h L/12 h D regimes at 22uC. Then plants were transferred to soil and kept under the same photoperiod and temperature for 3 weeks. On week 3, plants were transferred to 12 h L/12 h D regimes at 17uC or 27uC for 7 days before measurements were taken. CO 2 assimilation was measured over a 1 minute period for 48 plants using a EGM-4 machine (PP Systems) for 1 minute. Readings were taken in ppm and plants photographed and weighted to calculate area, fresh weigh and total CO 2 flux. The same set of plants was used for fresh weight determination and chlorophyll measurements. For chlorophyll extraction 300 mg of fresh tissue from twelve randomly selected plants were used for individual extractions conducted in triplicates. The whole rosette fresh weight was estimated and chlorophyll extracted by the protocol: http://www. nature.com/protocolexchange/protocols/521. Chlorophyll content was measured spectrophotometrically. Figure S1 hyh does not affect photosynthetic pigment accumulation during Red controlled deetiolation. (A-C) Carotenoid and chlorophyll accumulation in Col-0 and hyh 3 day-old seedlings grown at different temperatures (17,22 and 27uC respectively) in the dark (black columns) or after 6 h Red light illumination (40 mmol m 22 s 21 ) (yellow or green columns). For measurements seedlings were kept for two days at 22uC and 1 day at the indicated temperature in darkness. On day 3 they were subjected to red light treatment (for 6 h). The control set was kept in darkness (0 h time point).  Figure S5 (A) Higher mobility PIF4 forms accumulate as temperature increases. Immunoblot of 35S::PIF4-HA protein extracted from 6 day old seedlings kept in the dark. Seedlings were grown at the indicated temperature. Immunoblots were carried out with an HA antibody. Loading control indicated by UGPase signal. (B-D) Quantification of protein abundance for samples grown in the same conditions as the ones used for ChIP at T2 (or T3 for 35S::PIF4-HA) and T20 time points in Figure 4. Protein was quantified by immunoblots for 35S::PIF1-TAP (B), 35S::HA-HY5 (C) and 35S::PIF4-HA (D) at 17 and 27uC during the morning (at T2/T3, black bars) and the evening (time T20, grey bars). Antibodies against the tag were used for signal detection (anti -Myc or anti-HA) and relative quantification was carried out against UGPase signal. Error bars represent 6SE of biological triplicate samples. (TIF) Figure S6 Carotenoid and chlorophyll accumulation for the pif4pif5 (pif4-2 pif5-3) double mutant compared to pif1(pif1-1) and Col-0 at 17 and 27uC. Plants were grown and processed for pigment extraction as indicated in Figure 2. Error bars represent 6SE of biological triplicate sets. (TIF) Figure S7 Simultaneous binding of PIF1 and HY5 to the PSY promoter in EMSA. The probe preparation and assay was carried out as described in Figure 1E, except that one protein was incubated first with the probe for 30 min and then the second one added in 1X, 2X,10X and 20X excess (illustrated by the increasing triangle slope). Asterisk (*) indicates a non-specific band from the TnT. FP stands for Free probe. (TIF) Figure S8 (A) Expression levels by qPCR of HY5, PIF4 and PIF5 at T2 and T20 of a 17uC Red diurnal cycle in hy5-215, Col-0 and pifQ (pif1-2 pif3-3 pif4-2 pif5-3) backgrounds. Plants were grown and sampled as in Figure 4. (B) Expression levels for PIF1 and PIF3 under the same conditions as (A). Samples were normalized against ACT7 expression. Error bars represent 6SE of biological triplicates. (TIF) Figure S9 Expression levels of PSY and GUN5 in Col-0 seedlings treated with or without an EOD-FR light treatment (T12) at 17uC. Samples were obtained and processed as described in Figure 7 and Figure 6. In brief, expression levels were measured for (A) PSY and (B) GUN5 by qPCR on day 14 th in plants grown in 17uC Red diurnal cycles. Samples were treated with (+FR) or without (-FR) a saturating EOD-FR (T12) light pulse on day 14 th and harvested 3 h after treatment (T15). Levels are expressed relative to Col-0 (-FR) T15 sample. Error bars represent 6SE of biological triplicates. (TIF)