Abstract
DE-ETIOLATED1 (DET1) is a negative regulator of plant photomorphogenesis acting as a component of the C3D complex, which can further associate to CULLIN4 to form a CRL4C3D E3 ubiquitin ligase. CRL4C3D is thought to act together with CRL4COP1SPA ubiquitin ligase, to promote the ubiquitin-mediated degradation of the master regulatory transcription factor ELONGATED HYPOCOTYL5 (HY5), thereby controlling photomorphogenic gene regulatory networks. Yet, functional links between COP1 and DET1 have long remained elusive. Here, upon mass spectrometry identification of DET1 and COP1-associated proteins, we provide in vivo evidence that DET1 associates with COP1 to promote its destabilization, a process necessary to dampen HY5 protein abundance. By regulating HY5 over-accumulation, DET1 is critical to avoid its association to second-site loci, including many PIF3 target genes. Accordingly, excessive HY5 levels result in an increased HY5 repressive activity and are sufficient to trigger fusca-like phenotypes otherwise observed typically in COP1 and COP9 signalosome mutant seedlings. This study therefore identifies that DET1-mediated regulation of COP1 stability tunes down HY5 cistrome and avoids hyper-photomorphogenic responses that might compromise plant viability.
Introduction
Light fuels plant life and is an essential cue that modulates growth and development throughout all the plant life cycle. Initial exposure of a young or germinating seedling to light sensed by photoreceptors initiates a light signalling cascade that triggers the passage from skotomorphogenic to photomorphogenic growth (Von Arnim and Deng, 1996).
To identify genes controlling plant photomorphogenesis, independent screenings were performed in the 90’s, which led to the isolation of the det (deetiolated) and cop (constitutive photomorphogenic) mutant families (Chory et al., 1989; Deng et al., 1991). Many of these mutants are allelic to the fusca (fus) mutants that excessively accumulate anthocyanin in embryos and seedlings in both light and dark (Castle and Meinke, 1994; Misera et al., 1994; Pepper et al., 1994). COP/DET/FUS group of proteins contain components of CULLIN4 based E3 ubiquitin ligases (CRL4) that mediate protein polyubiquitination marking for proteasomal degradation, and members of the signalosome (CSN). CSN is a multimeric complex (CSN1-8) structurally similar to the proteasome lid that enables deconjugation of NEDD8/RUB1 from CRLs, a process essential for CRL inactivation and recycling (Wei et al., 1994; Lyapina et al., 2001; Serino and Deng, 2003; Lau and Deng, 2012; Qin et al., 2020).
In Arabidopsis, COP1 is a central repressor of photomorphogenesis that works in complex with SUPPRESSOR OF PHYA-105 (SPA1-SPA4) proteins (Laubinger and Hoecker, 2003; Saijo et al., 2003; Seo et al., 2003; Laubinger et al., 2004). Light exposure reduces COP1 activity by controlling its transcript accumulation (Zhu et al., 2008; Huang et al., 2012); its exclusion from the nucleus (von Arnim and Deng, 1994; von Arnim et al., 1997; Pacín et al., 2014); degradation of its partner SPA2 (Balcerowicz et al., 2011; Chen et al., 2015) and by promoting its inactivation through association with photoreceptors (Huang et al., 2014; Podolec and Ulm, 2018). Though COP1 itself displays E3 ubiquitin ligase activity in vitro, which might promote its degradation according to results obtained for human COP1, it further associates in vivo with CUL4 to form CRL4COP1SPA E3 ubiquitin ligase complexes (Osterlund et al., 2000; Saijo et al., 2003; Seo et al., 2003; Dornan et al., 2006; Chen et al., 2010). COP1 is highly active in darkness and targets numerous transcription factors (TFs) for proteasomal degradation, many of them being positive regulators of photomorphogenesis, such as ELONGATED HYPOCOTYL5 (HY5) and its homolog HYH, HIGH IN FAR RED 1 (HFR1) and LONG AFTER FAR-RED LIGHT 1 (LAF1) (Hardtke and Deng, 2000; Osterlund et al., 2000; Holm et al., 2002; Seo et al., 2003; Jang et al., 2005; Yang et al., 2005; Hoecker, 2017). Hence, a key function of COP1 is to regulate the availability of photo/skoto-morphogenic TFs including HY5.
HY5 is a basic leucine zipper (bZIP) TF that promotes photomorphogenesis by regulating, directly or indirectly, the expression of as much as one-third of Arabidopsis genes involved in diverse hormonal and metabolic pathways (Koornneef et al., 1980; Oyama et al., 1997; Jakoby et al., 2002; Lee et al., 2007; Zhang et al., 2011). In darkness, HY5 is a substrate of CRL4COP1SPA, and upon COP1 inactivation by light, HY5 accumulates in a proportional manner with light intensity (Osterlund et al., 2000). HY5 also positively regulates its own gene expression and directly binds to the promoters of anthocyanin, carotenoid and chlorophyll biosynthetic genes (Abbas et al., 2014; Binkert et al., 2014; Gangappa and Botto, 2016). Although HY5 does not bear an activation or repression domain (Ang et al., 1998), it mainly behaves in vivo as a transcriptional activator (Burko et al., 2020). Like other TFs regulating photomorphogenesis such as the PIFs, HY5 selectively binds to G-box sequence motifs (CACGTG) and variants (Gangappa and Botto, 2016). Attempts to determine the genes directly targeted by HY5 using different approaches (ChIP-chip, ChIP-seq), antibodies and transgene-driven expression levels gave rise to highly variable results ranging from 297 to 11797 genes (Lee et al., 2007; Zhang et al., 2011; Kurihara et al., 2014; Hajdu et al., 2018; Burko et al., 2020). HY5 acts in concert with other TFs to activate specific targets in specific tissues and conditions (Shin et al., 2007; Gangappa and Botto, 2016), while in other cases a competition for binding to promoter sequences was described (Toledo-Ortiz et al., 2014; Xu et al., 2014; Li and He, 2016; Gangappa and Kumar, 2017; Nawkar et al., 2017). Hence, several evidences indicate that regulation of HY5 stability is important, as rate-limited or excess amounts of HY5 may differently influence its regulatory activity.
HY5 down-regulation has also been shown to rely on DE-ETIOLATED1 (DET1), a chromatin-associated protein that is also key for photomorphogenesis repression in darkness (Chory et al., 1989; Chory and Peto, 1990; Pepper et al., 1994; Benvenuto et al., 2002). Both copl and det1 mutant alleles accumulate high protein levels of HY5, supporting the idea that DET1 is necessary for COP1 function on HY5 degradation (Osterlund et al., 2000). Accordingly, hy5 mutations partially suppress det1 phenotypes (Pepper and Chory, 1997) but, still, the possibility of direct relationship between DET1 and HY5 remain elusive.
Together with DAMAGED DNA BINDING PROTEIN1 (DDB1), COP10, and DDB1-ASSOCIATED1 (DDA1), DET1 forms a C3D adaptor module (also termed CDDD) of CRL4C3D E3 ubiquitin ligases to potentially target proteins for proteasomal degradation, including DDB2 (damaged DNA sensor), two subunits of a histone H2B deubiquitination module, the PYL8 abscisic acid receptor, and the HFR1 TF (Schroeder et al., 2002; Chen et al., 2006; Castells et al., 2011; Irigoyen et al., 2014; Nassrallah et al., 2018). Oppositely to its negative influence on HY5, DET1 contributes to the stabilization of PHYTOCHROME INTERACTIONG FACTORs (PIFs) TFs in the dark and can act as a transcriptional co-repressor (Maxwell et al., 2003; Lau et al., 2011; Dong et al., 2014; Shi et al., 2015). The repressive role of DET1 on transcription relies in part to its ability to regulate chromatin states. Plant DET1 shows high affinity for histone H2B and controls H2B ubiquitination levels on most genes mainly through proteolytic degradation of the DUBm (Benvenuto et al., 2002; Nassrallah et al., 2018). More generally, DET1 and COP1 are evolutionarily conserved in animals and plants (Schwechheimer and Deng, 2000; Yi and Deng, 2005; Olma et al., 2009). In humans however, the tumor-suppressor COP1 serves as an adaptor protein for DET1 to form the CRL4DET1COP1 complex. The latter recruits TFs such as c-Jun, ETS1/2 and ETV1/4/5, and target them for proteasomal degradation (Wertz et al., 2004; Vitari et al., 2011; Marine, 2012) Arabidopsis COP10 and COP1 were found to associate together (Yanagawa et al., 2004; Chen et al., 2010), but their functional relationship is plant systems remain elusive. Lack of evidence of physical association between DET1 and COP1 proteins conducted to the idea that they are part of distinct CRL4 complexes that cooperate in plant photomorphogenesis through an unidentified mechanism (Lau and Deng, 2012; Huang et al., 2014).
Here, we report that DET1 and COP1 can associate in the same complexes in vivo, DET1 being required for decreasing COP1 level in a light-independent manner. Despite this observation being seemingly paradoxical with HY5 over-accumulation in det1-1 mutant seedlings, our observations suggest that DET1-mediated COP1 destabilization is necessary for HY5 degradation. By profiling HY5 chromatin landscape in wild-type and in det1-1 and transgenic lines that over-accumulate HY5 protein, we further showed that uncontrolled HY5 levels lead to an aberrant enrichment over the promoter of second-site gene targets shared by light-regulated TFs such as PIF3, and frequently induce fusca-like phenotypes. Collectively, these observations led us to propose a direct role for DET1 in the regulation of COP1-mediated HY5 protein regulation to tune its association to a potentially very large gene repertoire. This study further identifies dynamic interaction of HY5 with hundreds’ genomic loci that may underlie its interplay with other TFs such as PIF3.
Results
COP1 and DET1 co-exist in one or more protein complexes
In search for DET1 interactors, we carried out Tandem Affinity Purification (TAP) coupled to mass spectrometry (MS) analysis of DET1 constitutively expressed in Arabidopsis cell cultures grown under dark conditions. In total, five independent TAP assays in which cells were collected in dark or after a 24 h white light treatment, allowed identifying proteins that co-purified with DET1. In all experiments, a large number of peptides representative of all known components of the C3D complex (DET1, DDB1, DDA1, COP10) was detected (Fig. 1A, Table S1 and S2), suggesting that the C3D complex is stable under these conditions as previously reported (Schroeder et al., 2002; Yanagawa et al., 2004; Olma et al., 2009; Irigoyen et al., 2014). Remarkably, the scores obtained for DDB1 (either DDB1a or DDB1b) recovery are in some experiments above those obtained for DET1, indicating that the large majority of DET1 proteins exist in association with DDB1. We also detected CUL4 peptides in high abundance, as well as all CSN subunits (CSN1 to 8) (Fig. 1A, Table S1) confirming that the C3D complex forms a CUL4 based E3 ligase (CRL4C3D) and that the C3D complex works in close association with the signalosome similarly to mammalian systems (Wertz et al., 2004; Chen et al., 2006; Lau and Deng, 2012). Among CSN subunits, CSN1 displays higher protein scores that might be due to its larger molecular size (easier to detect by MS) or because it represents a conserved direct contact point with DDB1 as previously reported in the structural analysis of human CSN-CRL4 complexes (Fig. 1A) (Cavadini et al., 2016).
Strikingly, we detected COP1 and SPA1 peptides in three of the five DET1 TAP experiments (Fig. 1A, Table S1 and S2). Since a COP1-DET1 interaction was discarded long ago in Arabidopsis (Chen et al., 2010), we aimed to confirm this result by performing TAP assays for COP1. Expectedly, MS analysis of proteins co-purified with COP1 under dark and light conditions allowed the detection of a high number of peptides for all four SPA proteins (SPA1 to 4), and also from CRY1 and CRY2, confirming that COP1 predominantly associates with these proteins (Fig. 1A; Table S1, S3; Wang et al., 2001a; Yang et al., 2001; Saijo et al., 2003; Laubinger et al., 2004). Importantly, we also recovered peptides, at a relatively lower number, for DET1 and DDB1, indicating association of COP1 with the C3D complex. However, no peptides from C3D subunits COP10 and DDA1 were found, potentially owing to their smaller molecular size. Neither CSN nor CUL4 peptides were detected (Fig. 1, Table S1), which is noteworthy considering the proposal that COP1 binds DDB1 to form a stable CRL4COP1SPA E3 ligase (Chen et al., 2010). From our data, we cannot discard the existence of these complexes, however, COP1 preferentially associates with SPA and CRY proteins instead of forming a stable CRL4DDB1-COP1 in our Arabidopsis cell suspensions, meaning that COP1 association with CUL4 is potentially transitory or cell-type specific.
Altogether, our results indicate that COP1 and DET1 associate transiently in vivo. This association was confirmed by semi-in vivo pull-down assays where bacteria-purified MBP-COP1 and MBP-HY5, but not the MBP alone, could pull-down MYC-DET1 from Arabidopsis protein extracts (Fig. 1B). These analyses support the existence of an association between DET1 and the COP1-HY5 module.
DET1 represses COP1 accumulation in a light independent manner
Considering this association and the established role of DET1 in ubiquitin-mediated protein degradation, we hypothesized that DET1 could affect the accumulation of COP1. To test this, we analysed the accumulation of endogenous COP1 in wild-type (WT) and det1-1 dark-grown seedlings and 3 or 24 hours upon exposure to light, as well as in plants germinated directly under light. We detected very low COP1 levels in dark-grown plants and slightly higher COP1 levels in light-grown plants, suggesting that low levels of COP1 in darkness are sufficient for its activity, including HY5 targeting for degradation (Fig. 2A).
Surprisingly, COP1 protein accumulation in det1-1 was much higher than in WT plants independently of the light condition used (Fig. 2A). Abundance of COP1 in GFP-DET1/det1-1 complemented lines (Pepper and Chory, 1997) was similar to that of WT plants in both dark and light conditions, confirming that elevated COP1 levels are linked to DET1 loss-of-function (Fig. 2B). It has been recently proposed that HY5 activity can induce COP1 accumulation (Burko et al., 2020). Since det1-1 mutants display high levels of HY5 (Osterlund et al., 2000), we analysed COP1 accumulation in hy5 and det1hy5 mutants to discard a possible direct effect of HY5 over-accumulation on COP1 abundance. We found no difference between hy5 and WT or between det1hy5 and det1-1, demonstrating that COP1 over-accumulation in det1-1 does not rely on HY5 (Fig. 2A). In line with this observation, RT-qPCR analysis of hy5 mutant seedlings confirmed that HY5 does not affect COP1 gene expression in dark or light conditions (Fig. 2C; Fig. S1). In these transcript analyses, we observed a slight, yet significant increase in COP1 RNA level in det1-1 mutants, suggesting a primary negative influence of DET1 on COP1 transcription. To test whether DET1 also controls COP1 accumulation at the post-transcriptional level, we treated Arabidopsis plants with the translation inhibitor cycloheximide (CHX) (Fig. 2D-F). We found that COP1 level strongly decreases in WT (Fig. 2D) within 6 hours after treatment while it remained mostly unaffected in det1-1 mutant plants (Fig. 2E). This indicated that DET1 plays also a role in controlling COP1 protein stability. Similar results were obtained using dark-grown seedlings (Fig. 2F). Thus, DET1-mediated COP1 destabilization seems to be light independent.
Further considering the possibility that our native protein extraction may favour the extraction of the cytoplasmic pool of COP1 and that det1-1 might alter COP1 nuclear accumulation in the dark (Chamovitz et al., 1996; von Arnim et al., 1997; Wang et al., 2009), we performed a denaturing protein extraction by adding 4M Urea to the extraction buffer and, in both cases, we analysed the insoluble (pellet) and soluble (supernatant) fractions. COP1 over-accumulated in det1-1 in all the conditions tested (dark and light) even when the majority of proteins were recovered from the insoluble fraction (Fig. 2G). Since DET1 is a nuclear protein playing a role in the regulation of chromatin (Nassrallah et al., 2018), we performed a nuclear extraction to analyse COP1 nuclear level. We found that COP1 nuclear pool is much higher in det1-1 background and is unaffected in hy5 seedlings (Fig. 2H). Collectively, these analyses unveil a role for DET1 in moderating COP1 abundance at multiple levels, including protein destabilization.
COP1 destabilization depends on the proteasome, C3D and CSN
Further supporting the idea that COP1 is being degraded by the proteasome independently of light, we observed that treatment with the proteasome inhibitor bortezomib (Bor) results in a moderate increase in COP1 protein accumulation in WT seedlings both under light and dark conditions (Fig. 3A and B). This was not the case in det1-1 seedlings, Bor treatment having little or no effect on COP1 accumulation, suggesting that DET1 is mediating COP1 proteasomal degradation (Fig. 3A and B).
DET1 and COP10 being part of the CUL4 based C3D complex, we further tested COP1 accumulation in cop10 and cul4 mutant seedlings. Both cop10-4 (weak allele), cop10-1 (null allele) and cul4-1 seedlings displayed COP1 over-accumulation (Fig. 3C,D). As in TAP assays DET1 and the CSN associate together, we tested COP1 accumulation in cop9-1 homozygous mutant for CSN8 (Fig. 3D). In this mutant, COP1 accumulated to levels similar to those in det1-1, supporting the idea that CSN, through the canonical mechanism of CUL4 deneddylation, is required for COP1 rapid destabilization, likely by facilitating the function of a CRL4C3D.
DET1 facilitates HY5 degradation
A role for DET1 in COP1 protein regulation represents a new hint on the early hypothesis that DET1 facilitates COP1 function in promoting HY5 degradation (Osterlund et al., 2000). Still, over-accumulation of HY5 in det1-1 mutant plants having never been analysed in detail, we monitored HY5 accumulation in WT, det1-1 and cop1-4 seedlings for 24 h during de-etiolation. As reported earlier, in WT seedlings HY5 protein was detectable by immunoblot 3 h after illumination. HY5 also accumulates at high levels in dark-grown det1-1 seedlings, and this accumulation is even increased after light exposure in a similar way than in cop1 seedlings (Fig. 4A).
To test for a potential transcriptional regulation of HY5 by DET1, we analysed HY5 transcript levels in this experimental setup by qRT-PCR. As previously reported, HY5 transcript level increased after exposure to light (Fig. 4B; Osterlund et al., 2000). In det1-1 and cop1-4 seedlings, HY5 transcript accumulation shows little variations in response to light. Hence, based on these observations, as previously suggested (Osterlund et al., 2000), the discrepancy in the kinetics and low increase of HY5 transcripts accumulation in these mutants does not appear to be on its own a major determinant of high HY5 protein over-accumulation in det1-1 mutant plants.
Collectively, these findings shed light on intricate links between DET1 and COP1-HY5 protein abundance, suggesting that DET1 promotes COP1 destabilization and, yet, is further necessary for COP1 activity in HY5 degradation.
HY5 over-accumulation increases its chromatin occupancy and expands the repertoire of target genes
As mentioned earlier, HY5 over-expression is a major determinant of det1-1 photomorphogenic phenotype (Pepper and Chory, 1997). Hence, to test whether HY5 excessive accumulation impacts on its activity at the chromatin level, we conducted ChIP-seq assays in WT and det1-1 mutant seedlings grown under light conditions. To identify endogenous HY5 targets, we used an anti-HY5 antibody that gave no significant background (Fig. S2A) and was recently used for ChIP in Bellegarde et al., (2019). Additionally, 2×35S::GFP-HY5/hy5 (hereafter called GFP-HY5) functionally complemented seedlings were also used for an anti-GFP driven ChIP experiment to assess HY5 targets when the protein is overexpressed. GFP-HY5 level was higher than endogenous HY5 levels of WT and of det1-1 seedlings (Fig. S2B). In our analysis, only genes with significant HY5 peaks in each of two independent biological replicates but not in the mock IP controls (hy5-215 for the anti-HY5 ChIP and 2×35S::GFP for anti-GFP) were considered as true HY5 binding genes for further analyses (Fig. S3A and B). This resulted in defining 422 HY5 target genes in WT seedlings (Fig. 5A), representing a smaller subset than in other studies reporting thousands of target genes (Kurihara et al., 2014; Hajdu et al., 2018). Among them, 372 genes (88%) were also identified in Hajdu et al., (2018) and/or by DAP-seq in O’Malley et al., (2016) (Fig. S3C) and 64 were common with the 297 HY5 so-called “directly-regulated genes” defined in Burko et al., (2020) (Fig. S3D).
Confirming the robustness of these endogenous HY5 genomic profiles, this HY5 target gene repertoire was also almost entirely conserved in the det1-1 samples (only 10/422 target genes were not found in det1-1; Fig. 5A). In this mutant line, the HY5 targeted genes was extended to 1,297 (a 3-fold increase respect to WT). In line with a possible effect of HY5 over-accumulation in targeting extra sites, the number of HY5 associated genes further increased in GFP-HY5 plants to reach 2,753 genes (a 6.5-fold increase respect to WT).
We classified all gene sets in four groups, A representing the genes commonly bound in all three plant lines, B representing the genes significantly occupied in either det1 or GFP-HY5 (where HY5 over-accumulates), while C and D represent the genes specifically targeted in det1-1 or in the GFP-HY5 line, respectively (Fig. 5A). Remarkably, Class B represented a large set of 619 genes, indicating a robust tendency of HY5 binding on additional targets upon over-accumulation. Observation of these extra peaks in the WT sample unveils their initial pre-existence as low (but not statistically significant) ChIP signals, which get enriched to form robust peaks in the two plant lines with high HY5 global level (Fig. 5B, C and D). In line with these observations, HY5 occupancy also increases on class A (WT targets) in det1-1 plants (Fig. 5C and D). These findings suggest that HY5 over-accumulation expands the HY5 cistrome, presumably as a consequence of TF rate-limiting availability that allows binding only to primary targets where its affinity is stronger or has better chromatin accessibility. Interestingly, HY5 global level on its own might not be the unique determinant of its DNA binding specificity, as HY5 enrichment over Class A genes is normal in GFP-HY5 plants (Fig. 5C).
We therefore undertook a de novo sequence motif search, which identified that a CACGT motif is present under most of HY5 peaks in all gene classes but Class C (Fig. 5E). In this det1-1 specific gene class, only 41% (versus 76-88% for A/B/D classes) of the peaks contain a G-box while, instead, several other DNA sequence motifs are over-represented in the extra-peaks identified in det1-1 plants (Fig. S4).
To further test whether HY5 chromatin association is linked to gene expression, we first analysed the expression of the genes from classes A-D in hy5-215 loss-of-function mutant plants using previously published RNA-seq (Myers et al., 2016) (Fig. S3E). Around 20% of the genes in each class were found to be misregulated, suggesting that only a minor proportion of HY5 target genes directly respond to HY5 occupancy. A majority of HY5 target genes may therefore be regulated by redundant transcription factors, as for example with the related HYH. As expected, gene classes A-B-C had a strong tendency for down-regulation in hy5, in agreement with HY5 being an activator of gene expression (Burko et al., 2020). Noteworthy, this was not the case for Class D, i.e., genes bound by HY5 only in the over-expressing GFP-HY5 line (Fig. S3E).
To assess the effect of HY5 second-site binding on a genome-wide scale, we first determined the gene ontology (GO) of all different HY5 target gene sets. Extra-binding genes were found to be involved in the control of photosynthesis and pigment accumulation as well as to the response to stress responses as high light, temperature, UV-B, oxidative stress, hypoxia, water deprivation and nutrient deficit (Fig. 5F). Then, we analysed the expression of HY5-target genes in det1-1 mutant seedlings using previously published RNA-seq (Nassrallah et al., 2018). Genes bound by HY5 in det1-1 (Classes A, B and C) tend to be upregulated in det1-1 mutant with respect to WT (Fig. 5G). Accordingly, HY5 target genes misregulated in det1-1 are almost exclusively upregulated (Fig. S3F). On the contrary, class D genes (occupied by HY5 only in the overexpressed GFP-HY5 line) displayed tend to be equally expressed in WT and det1-1 plants, confirming that this class of genes is specifically affected in the overexpressing line and not in det1-1 (Fig. C). We therefore analysed the expression of a non-biased selection of class D genes in GFP-HY5 seedlings, which showed that most of them are slightly down-regulated as compared to WT levels (Fig. S5). These observations support the possibility of an increasing repressive activity of HY5 when it over-accumulates. Misregulation of Class D genes and the identification of hundreds of secondary target genes indicate that Arabidopsis photomorphogenic seedlings require a tight modulation of HY5 levels, potentially avoiding out of context responses affecting multiple gene ontologies.
HY5 and PIF3 targets overlap
DET1 has been found to associate with PIFs (Dong et al., 2014), that are thought to play antagonistic roles to HY5 in many aspects of plant development (Gangappa and Botto, 2016). This antagonism might rely on one hand on their differential regulation by light and on a potential competitive binding over a common gene repertoire bearing G-Box motifs as proposed earlier (Toledo-Ortiz et al., 2014; Gangappa and Kumar, 2017). To assess whether HY5 primary (WT) and secondary site (upon overexpression) binding may overlap the PIFs chromatin landscape, we performed PIF3 ChIP-seq experiments using a pif3::eYFP:PIF3/pif3-3 Arabidopsis line (Al-Sady et al., 2006) intended to mimic the endogenous PIF3 levels. Determination of PIF3 binding sites as done previously with HY5 identified 958 in dark-grown seedlings. In accordance with PIF3 being much more active under dark than under light conditions (Soy et al., 2012), most of the peaks called in light-grown plants were low intensity and correspond to genomic positions generating a high level of noise in the corresponding mock IPs and were therefore not considered as being robust for further analyses (Fig. 6A and B). PIF3-associated genes in darkness significantly overlap with a previous ChIP-seq experiment done on 2-day-old seedlings from an overexpression line and also with the repertoire of genes downregulated in the quadruple pifQ mutant line (Leivar et al., 2012; Zhang et al., 2013) impaired in the function of PIF1, PIF3, PIF4 and PIF5 partially redundant TFs (Fig. S6B and C). Confirming previous studies (Zhang et al., 2013), we detected a sequence motif matching the canonical motifs G-box (CACGTG) and PBE-box (CACATG) under half of PIF3 peaks (Fig. 6C). Two other over-represented motifs were also identified, with homology to TEOSINTE BRANCHED 1/CYCLOIDEA/PCF (TCPs) and, interestingly, FUS3 recognition motifs determined by DAP-seq (Fig. S6D; O’Malley et al., 2016). Altogether, these three motifs cover ~96% of the PIF3 peaks from our ChIP-seq, suggesting a high level of sequence specificity.
Comparison of PIF3 with HY5 target genes restricts the shared repertoire to 39 loci that may consequently be simultaneously or alternatively occupied by both TFs in wild-type plants, representing ~4% of all PIF3 associated genes in darkness or ~9% of all HY5 associated genes in the light (Fig. 6D). Increased HY5 levels in det1-1 and in the GFP-HY5 lines resulted in an increased overlap with PIF3 target genes, with 79, 18 and 148 additional target genes found among classes B, C and D, respectively (Fig. 6D-F). This shows that HY5 secondary sites in the light tend to span loci occupied by PIF3 in darkness. Accordingly, analysis of PIF3 enrichment at HY5 peaks showed that HY5 and PIF3 peaks are centred around the same position, certainly corresponding to the G-box motif. A second PIF3 enrichment at positions neighbouring the main peak is detected for genes in class C, indicating that genes gaining HY5 in det1-1 are frequently targeted by PIF3 at multiple positions in the promoter (Fig. 6F). Collectively, these analyses unveiled that an initial tendency for HY5 and PIF3 to target similar gene positions is exacerbated in det1-1 and GFP-HY5 plant lines, suggesting that moderation of HY5 accumulation in both light and dark conditions is required to avoid indirect effects on both HY5 and PIF3 regulatory networks.
HY5 over-expression causes fusca-like phenotype
Aiming at identifying potential out-of-context responses to HY5 over-expression, we analyzed the GFP-HY5 complemented line used in our ChIP-seq analysis. Interestingly, attempts to obtain homogeneous complemented transgenic lines throughout generations were unsuccessful. Besides a variable number of non-germinated seeds, we obtained a segregating population of phenotypically “non-complemented” and “complemented” plants, with another group of small seedlings exhibiting typical “fusca-like” phenotypes of underdeveloped and purple plants with multiple growth defects including seedling lethality (Fig. 7A and B; Fig. S2B).
Considering that appearance of this exaggerated photomorphogenic phenotype in several independent transgenic lines might represent the phenotypic translation of HY5 binding to an extended number of genes, we pursued the characterization of fusca-like plants. These seedlings accumulate greater levels of GFP-HY5 fusion protein than phenotypically complemented plants (Fig. 7C). Similarly to fusca mutants described in landmark studies from Misera et al., (1994) and Castle and Meinke genetic screenings (1994), HY5-based fusca-like plants display reduced growth, limited cotyledon expansion and high anthocyanin content especially in the cotyledons (Fig. 7E and F; S3C). They also present stronger defects in the aerial part of the plant as the root develops as in weak fusca mutants, perhaps correlating with a higher impact of HY5 activity in aerial tissues. Increased accumulation of anthocyanins occurs at early stages and relies on sugar as they do not accumulate when plants are grown in MS media without sucrose (Fig. S2D). This characteristic is also shared with original fusca mutant plants (Castle and Meinke, 1994) and further suggests an implication of HY5 in sugar induced anthocyanin accumulation.
Unlike original fusca mutant seedlings, HY5-based fusca-like plants turn pale across the days and present a strong defect on chlorophyll accumulation in the light, while behaving without significant phenotype when grown in darkness (Fig. 7G; Castle and Meinke, 1994; Misera et al., 1994). This correlates with GFP-HY5 protein level being undetectable by immunoblot analysis of wild-type etiolated seedlings and accumulating from 3 hours after transfer to light (Fig. 7H). This means that COP1 in the dark displays an enormous capacity to degrade HY5 protein present at much higher levels than those occurring endogenously and still, this capacity is completely impaired in a det1-1 background (Fig. 7H).
Last, we examined expression of a selection of genes from clases A to D in light-grown fusca-like seedlings to test whether such phenotype can be linked to HY5 association to chromatin secondary targets. This unveiled stronger misregulation in this category of plants than in functionally complemented GFP-HY5 plants. While in some cases the fusca-like phenotype can be associated to higher gene upregulation (as in the case of CHS, F3H, CRF6, CUC1 or WOX1), in the majority of genes tested they were associated to a repressive effect suggesting that HY5 extra occupancy contributes to triggering fusca-like phenotypes. Interestingly, this feature is true for the majority of PIF3 direct targets that are also targeted by HY5 (e.g., BBX28, TCP2, RGA2/GAI, GASA6, BBX27, XTH7).
Altogether, these results provide evidence that exaggerated HY5 accumulation by itself is sufficient to generate fusca-like phenotypes and render plants unviable under light conditions. At several loci, HY5-derived gene repression occurs through ectopic enrichment over secondary sites, potentially at the expenses of PIF3 occupancy over a set of commonly targeted genes.
Discussion
DET1 and COP1 were identified more than 30 years ago as repressors of photomorphogenesis and of multiple light responses. Impairment of any of these proteins results in the induction of deetiolation in darkness and in hyper-photomorphogenic or so-called fusca phenotypes in the light. Because they form CRL4 based E3 ligases involved in HY5 degradation, COP1 and DET1 were expected to work in a close relationship whose nature was never elucidated. We identified that COP1 and DET1 can associate–together, DET1 controlling COP1 stability by promoting its proteasomal degradation which seems to be essential for COP1 activation.
DET1-mediated regulation of COP1 is of prime importance given its necessity for the regulation of HY5 abundance, a feature that seemingly look paradoxical as det1 mutant plants over-accumulate both COP1 and HY5 proteins. In view of these observations, we propose that DET1-mediated COP1 destabilization is necessary to maintain COP1 turnover and activity towards its downstream targets. We also unveiled that, by down-regulating HY5 levels, DET1 restrains HY5 binding to primary targets. Over-expression of HY5 can largely extend its cistrome, not only by inducing HY5 enrichment over second-sites (normally poorly bound) target genes but also by occupying PIF3 target genes. As detailed below, HY5 second-site occupancy correspond to multiple categories of light-regulated genes and was further found to be linked in cis to gene misregulation, a property that presumably underlies the frequent occurrence of fusca-like phenotype upon artificial HY5 over-expression (Fig. 7I).
DET1 conforms CRL4C3D complexes and associates with CSN and with COP1
Together with DDB1, COP10 and DDA1, DET1 forms a stable C3D complex that associates with the CUL4 scaffold as well as with the CSN (Schroeder et al., 2002; Wertz et al., 2004; Yanagawa et al., 2004; Chen et al., 2006; Olma et al., 2009; Lau and Deng, 2012; Irigoyen et al., 2014). Our TAP assays showed that most of DET1 engages in the formation of the C3D complex and only part of this complex associates with CUL4. A smaller fraction of DET1, perhaps in the form of CRL4C3D complexes, associated with a fully assembled CSN. CSN-associated CRL4C3D might represent the substrate-free fraction, since substrate binding to CRL4 is expected to displace the CSN (Cavadini et al., 2016). Moreover a set of WD40 domain containing proteins, known as typical CRL4-associated target receptors, also co-purified with DET1 (Table S3; Fonseca and Rubio, 2019). Among them, we found COP1 and SPA1 proteins and reciprocally, DET1 peptides were found in TAP assays when COP1 was used as a bait. By means of co-immunoprecipitation assays, previous reports discarded a DET1-COP1 association in Arabidopsis even though COP1 and the C3D subunit COP10 were found to interact, presumably independently of the C3D complex (Chen et al., 2010). This supported the idea that DET1 and COP1 exist in distinct complexes, each of them conforming CUL4 based E3 ligases, CRLC3D and CRL4COP1SPA (Schroeder et al., 2002; Yanagawa et al., 2004; Chen et al., 2010) and that they may work together, repressing photomorphogenesis in an unsolved way (Lau and Deng, 2012). For the first time, we demonstrate here that, similar to human cells (Wertz et al., 2004), DET1 and COP1-HY5 module can associate in Arabidopsis.
As previously reported, we also detected that COP1 further associates with all four SPA proteins (specially with SPA4) and with CRY2 (and CRY1 with less affinity) (Wang et al., 2001; Yang et al., 2001; Saijo et al., 2003; Laubinger et al., 2004; Liu et al., 2011). It was however surprising that we could not recover CUL4 or CSN peptides from COP1 TAPs, as COP1 has been described to associate with CUL4 to stably form CRL4COP1SPA complexes (Chen et al., 2010). This might be due to transient association of COP1 with CUL4 or to limited resolution capacity of our MS analysis; which might be insufficient to capture the whole catalogue of complexes conformed by COP1.
DET1 promotes COP1 degradation and activity
Following the identification of a DET1-COP1 association, our study sheds light on the nature of this relationship, DET1 being necessary for COP1 protein destabilization. COP1 accumulates at higher levels in the light than under dark and is a short-lived protein with a fast turnover rate in both conditions (estimated half-life of 4,5 hours, Fig. S1B). DET1-mediated COP1 degradation is light independent and depends at least in part on the proteasome. Our mutants’ analyses indicate that COP1 degradation mechanism seems to rely on a canonical CSN-mediated CUL4 recycling that is mediated by a full C3D complex.
Our biochemical findings seemingly enter in conflict with genetics, because det1-1 seedlings, as well as cop1-4, are deetiolated in the dark, meaning that DET1 and COP1 generally repress light signalling. Accordingly, for this reason, we wished to confirm that, as previously reported, HY5 levels are higher in both mutant backgrounds (Fig. 4A; Osterlund et al., 2000). In 1994, Ang and Deng analysed the epistatic relationships between cop1 and det1 hypomorphic mutations. They found cop1-6 is epistatic to det1-1, with respect to light control of seed germination and dark-induced gene expression, suggesting that DET1 and COP1 may act in the same pathway, with COP1 being downstream, which fully supports our findings.
In our study, COP1 protein levels do not simply correlate with its activity when using its capacity to degrade HY5 as readout. Indeed, low levels of COP1 in darkness are sufficient to degrade both physiological and over-accumulated HY5 protein levels, as those displayed by wild-type and GFP-HY5 overexpressing lines, respectively (Fig. 7H). Counter-intuitively, higher COP1 protein levels following exposure to light or in the det1-1 mutant are associated to diminished COP1 function. This effect might result from COP1 activity being regulated by its association with SPA proteins or photoreceptors and also through nucleo-cytoplasmic partitioning (Wang et al., 2001; Yang et al., 2001; Laubinger and Hoecker, 2003; Seo et al., 2003; Laubinger et al., 2004; Fankhauser and Ulm, 2011; Lian et al., 2011; Liu et al., 2011; Rizzini et al., 2011; Zuo et al., 2011; Ponnu et al., 2019). We found a large amount of COP1 in the nuclei of det1-1 mutants, precluding any inactivation of the COP1 protein pool through nuclear exclusion. Alternative molecular mechanisms should be envisaged, as for example DET1 promotion of COP1 association with SPA proteins or any other member of CRL4COP1-SPA complexes. As well, DET1 could promote CSN-mediated cycles of CUL4 neddylation/deneddyation required for CRL4COP1-SPA activity. Finally, following on our results showing DET1-HY5 coprecipitation, DET1 could facilitate substrate recognition and ubiquitination by COP1. All these processes might be necessary for efficient ubiquitination and proteasomal degradation of the COP1 substrates but also of COP1 itself, as a feedback mechanism to limit the extent of its activity.
Higher HY5 accumulation increases binding to extra targets including PIF3 target genes
HY5 is a pivotal TF in light signalling with a strong effect on plant morphogenesis, whose levels are tightly regulated (Osterlund et al., 2000). Therefore, the identification of its direct genomic targets to identify genes directly regulated by HY5 fundamentally needs to be considered in the context of dynamic changes in HY5 global level in the nucleus. In other words, as proposed for PIF transcription factors (Pfeiffer et al., 2014), HY5 chromatin association and its sets of targeted genes need to be envisaged as a potential continuum that varies with HY5 protein availability, chromatin accessibility and the abundance of other TFs potentially binding competitively to the same loci. Adding to technical variability, this concept might be central in the large variations of HY5 target genes from previous studies that reported lists reaching ~12,000 HY5 targets genes (Lee et al., 2007; Zhang et al., 2011; Kurihara et al., 2014; Hajdu et al., 2018). Probing endogenous HY5 protein, our ChIP analysis led to the identification of a moderate number of 422 targets that largely overlap the repertoire of 297 high-confidence HY5-activated genes reported by Burko and co-workers (2020) using an elegant strategy combining transcriptional and ChIP analyses of constitutive activator and repressor HY5 fusion proteins. Still in line with Burko et al., (2020), we found that in WT in light conditions HY5 behaves mainly as a transcriptional activator. Among HY5 targets in light-grown seedlings, we found previously described light-regulated HY5-bound genes such as the HY5 gene itself and many other genes with the capacity to trigger downstream transcriptional cascades influencing a range of light-regulated processes: light stress (ELIP1), pigment biosynthesis (CHS, F3H, FLS1), signalling proteins (SPA1, SPA3 and SPA4) as well as a high number of transcription factors (Table S4) (Oyama et al., 1997; Gangappa and Botto, 2016). HY5 peak summits were positioned on a typical or a related G-box sequence motif in the vast majority of these genes, but traces of HY5 binding could also be found over many other loci, potentially secondary or cell-specific. When identifying that many of these second-site binding loci are increasingly occupied by HY5 in det1-1 and in the GFP-HY5 over-expression line hints at the necessity for HY5 level fine-tuning to restrict the activity of this TF over a specific set of loci. This also hints at potential variations of the HY5 target gene repertoire during dark-to-light or light-to-dark transitions when HY5 abundance (Osterlund et al., 2000) and chromatin properties (Bourbousse et al., 2020) are subjected to strong variations.
Considering the preponderant role of HY5 over-accumulation in det1-1 photomorphogenic phenotypes (Pepper and Chory, 1997), HY5 enrichment over second-site target genes likely contributes to gene misregulation induced by DET1 loss of function. For example, among the genes significantly upregulated and targeted by HY5 specifically in det1-1 plants (class C) (Fig. S3F), we identified four subunits of the chloroplast FtsH protease complex (FTSH1, 2, 5 and 8 found in the GO category “PSII associated LHCII catabolic process”; Fig. 5F) involved in the quality control of the photosynthetic electron transfer chain during photo-oxidative stress (Kato and Sakamoto, 2018). Misregulation of such stress response pathways must have a high cost impact on plant growth and performance as observed in det1 plants.
Gene misregulation of HY5 secondary targets in det1-1 and GFP-HY5 plant lines might result from a combination of multiple mechanisms, ranging from HY5 capacity to activate transcriptional, ectopic recruitment of chromatin machineries such as the GCN5 acetyltransferase (Benhamed et al., 2006) and, among other effects, competitive binding with other TFs. Comparison of our HY5 and PIF3 ChIP experiments indicate that HY5 enrichment over additional targets leads to a large increase in the overlap with PIF target genes, as for example 245 genes occupied by PIF3 in the dark are newly bound by HY5 when in det1-1 and/or GFP-HY5 overexpressor (Fig. 6D). Reduced abundance of PIFs in det1-1 mutant plants (Dong et al., 2014) might facilitate HY5 binding to common targets with PIF3, but this is presumably not the case in the GFP-HY5 overexpressing line. Expression analyses indicate genes downregulated in det1-1 and GFP-HY5 lines are PIF3 targets, thereby suggesting that HY5 enrichment is translated into a higher repressive activity (Fig. S5).
Taken together, these interplays indicate that TFs availability and balanced levels are key to account for differential binding of HY5 and PIF3 proteins. It has been previously proposed that HY5 could compete with PIFs for binding sites on DNA for specific gene targets. For instance, HY5 and PIF4 proteins bind with different intensities to common targets genes at different day-times and temperatures (Toledo-Ortiz et al., 2014; Gangappa and Kumar, 2017). This idea is fully supported by our data in a genome-wide context. In future studies, it will be interesting to test HY5 binding specificity in pif higher-order mutant plants and, vice-versa, to assess the influence of HY5 on PIF chromatin landscape when their respective levels are balanced during dark-light transitions.
Uncontrolled HY5 accumulation triggers fusca-like phenotypes
At high levels, HY5 chromatin association exceeds its primary target genes to an increased number of target sites, with consequent transcriptional changes. In line with the concept of fine-tuning TF abundance, appearance of fusca-like phenotypes in GFP-HY5 plants show the pleiotropic effects derived from the uncontrolled accumulation of a single transcription factor, especially when, like HY5, it regulates many signalling pathways and other TFs. The fusca mutant plants isolated in the 90’s (Castle and Meinke, 1994; Misera et al., 1994) accumulate high levels of anthocyanin in the seeds and seedlings, display light independent (constitutive) seed development and compromised viability. These fusca mutants were shown to be impaired in COP1, DET1 or CSN activity, which all contribute to moderate HY5 accumulation, a process probably enhanced by a cis-acting positive feedback loop linked to HY5 autoactivation (Chory et al., 1989; Deng et al., 1991; Mayer et al., 1996). Across the years however, studies based on the transcriptional analysis of the fusca mutants suggested that these phenotypes could not be supported uniquely by altered light responses and should be due to a general defect in developmental programming, because several other signal transduction pathways were affected. These pathways described by Mayer et al., (1996), widely overlap with those present in the GO analysis of HY5 target gene classes B, C and D (Fig. 5F), showing that they are spanned by HY5 action. Gene expression analysis of key transcription factors involved in developmental processes such as cytokinin signalling (CRF6), meristem maintenance and initial organ development (WOX1, CUC1) and circadian clock (TOC1) showed these genes are upregulated in fusca-like plants (Fig. S5; Takada et al., 2001; Dolzblasz et al., 2016; Kim, 2016; Fung-Uceda et al., 2018). Reciprocally, a number of TFs (e.g. CBF3, BBX28, TCP2, HFR1, BBX27) and photosynthesis related genes (LHCA1, LHCB7, FTSH1, FTSH5, PAP2, PSAE) are downregulated in fusca-like seedlings. For the latter ones, HY5 apparently behaves as a transcriptional repressor by occupying extra target sites shared with PIF3 in darkness (Fig. 6D). Through this mechanism, HY5 may control numerous processes necessary for plant viability, including meristem activity, cell cycle, pigment accumulation and photoautotrophy.
Thus, in line with the finding that it lacks its own activation or repression domains (Ang et al., 1998), HY5 transcriptional regulatory activity over a limited gene repertoire might be regulated by titration of its availability. COP1 and DET1 activities are part of this regulatory mechanism to keep HY5 transcriptional activation sharp and responsive to light perception in a dynamic system.
Material and Methods
Plant Materials
All plant lines are in the Columbia-0 ecotype background. The det1-1 (Chory et al., 1989), hy5-215 (Osterlund et al., 2000), det1-1hy5-215 was kindly provided by Prof. Roman Ulm (Geneva, Switzerland); cop1-4 (McNellis et al., 1994), pif3::eYFP:PIF3/pif3-3 (Al-Sady et al., 2006); kindly provided by Drs Lot Gommers and Elena Monte, CRAG Barcelona, Spain). The 2×35S::GFP-HY5/hy5-215 was generated by Agrobacterium tumefaciens (GV3101) and floral dip (Clough and Bent, 1998) transformation of hy5-215 mutants with a GFP-HY5 expressing plasmid based on the pVR TAP Nt plasmid where the TAP tag cassette was substituted by the GFP reporter gene (Rubio and Deng, 2008).
Plant growing conditions
Arabidopsis seedlings were sterilized with a solution of 75% sodium hypochlorite and 0.1% Tween-20 and stratified at 4°C during 3 days in darkness. Seedlings were grown in Murashige and Skoog (MS) medium with 1% sucrose and 0.7% agar at 22°C for 7 days (unless otherwise specified) under fluorescent white light (100 μmol m-2 s-1) in a 16-h light/8-h dark period (LD).
Hypocotyl measurements
For hypocotyl measurements 7 days-old plants were disposed on agar plates, photographed and hypocotyls measured using ImageJ software (http://www.imagej.net). Three biological replicates, each consisting of measurements for at least 30 seedlings grown at different times, were analyzed with similar results.
Protein extraction and immunoblotting
In the indicated experiments, 6 to 7 day old light or dark grown seedlings were pre-treated with 50 μM cycloheximide (CHX, Sigma Aldrich) or with 50 μM proteasome inhibitor Bortezomib (Selleckchem). For COP1 detection, extraction of plant soluble protein extracts was performed in 4 M Urea, 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10 mM MgCl2, 1 mM phenylmethylsulfonyl fluoride, 0.1% Nonidet P-40 and cOmplete EDTA-free protease inhibitor cocktail (Roche) supplemented with followed by centrifugation twice 10 min at 16,000 g at 4°C. Protein concentration in the final supernatants was determined using the Bio-Rad Protein Assay kit. For HY5 detection, an equal number of plants were collected for each condition, directly denatured in Laemmli buffer and separated by a 15% blotted with antibodies described below. Chromatin-enriched protein fractions were obtained as previously described (Nassrallah et al., 2018).
Pull-down assays
MBP recombinant protein fusions were expressed in the Escherichia coli BL21 (DE3) strain carrying the corresponding coding sequence cloned into the pKM596 plasmid, a gift from David Waugh (Addgene plasmid # 8837). Recombinant proteins were purified and pull-down assays were performed according to Fonseca and Solano, (2013). MBP-tagged fusions were purified using amylose agarose beads. Equal amounts of seedling protein extracts were combined with 10 mg MBP-tagged fusion or MBP protein alone, bound to amylose resin for 1 hr at 4°C with rotation, washed three times with 1 ml of extraction buffer, eluted and denatured in sample buffer before immunoblot analysis.
TAP Assays
Cloning of a GSRhino-TAP–tagged DET1 and COP1 fusion under the control of the constitutive cauliflower mosaic virus 35S promoter, transformation of PSB-D Arabidopsis cell suspension cultures and TAP purifications were performed as described previously (Van Leene et al., 2015; García-León et al., 2018). For the protocols of proteolysis and peptide isolation, acquisition of mass spectra by a 4800 MALDI TOF/TOF Proteomics Analyzer (AB SCIEX), and mass spectrometry–based protein homology identification based on the TAIR10 genomic database, we referred to Van Leene et al., (2010) and García-León et al., (2018). Experimental background proteins were subtracted based on 40 TAP experiments on wild-type cultures and cultures expressing TAP-tagged mock proteins GUS, RFP, and GFP (Van Leene et al., 2010).
ChIP analyses
HY5 ChIPs were performed on 5 day-old seedlings grown under LD conditions. PIF3 ChIPs were performed on 5-day-old seedlings grown in darkness or LD conditions. ChIP experiments were performed as in Fiorucci, et al. (2019), using anti-HY5 (Agrisera #AS121867) or anti-GFP (Life Technologies #11122).
Library preparation, sequencing, and analysis
Libraries were prepared using the NEBNext Ultra II DNA Library Prep Kit (New England Biolabs E7645). Sequencing was performed on an Illumina HiSEq 4000 in 150-bp paired-end mode. Reads were trimmed using trim_galore (https://github.com/FelixKrueger/TrimGalore) with options “--phred33 --paired -q 20 --stringency 1 --length 35” and then mapped to the TAIR10 genome using bowtie2 (Langmead and Salzberg, 2012) with options “--very-sensitive -I 150 -X 2000 -p 20 --no-mixed”. Duplicated reads were marked using picard-tools (https://github.com/broadinstitute/picard). Reads were then filtered using samtools (https://github.com/samtools/samtools.git) with options “view -hb -F 1804 –L selected_TAIR10_genome_Chr.bed”, which correspon ds to the TAIR10 genome after filtering out genomic regions with aberrant coverage or low sequence complexity (Quadrana et al., 2016). Browser tracks were generated using deeptools (Ramírez et al., 2016) function bamCoverage with options “--binSize 20 --normalizeUsingRPKM --extendReads --centerReads --ignoreForNormalization ChrC ChrM”. Peaks were called using MACS2 (McNellis et al., 1994) with options “callpeak -f BAMPE --bdg -q 0.01 -g 120e6 --bw 300” and using the input bam files as control. The peaks present in the mock IPs were removed for further analysis using bedtools (Quinlan and Hall, 2010) with options “subtract -A -f 0.2” and only the remaining peaks with a score above 60 were kept. A second filtering step consisted in keeping only the peaks present in both biological replicates using bedtools with options “intersect -f 0.2 -r”. Peaks were then annotated to the closest TSS using HOMER annotatePeaks.pl (Heinz et al., 2010) providing Araport11 gtf annotation file (Cheng et al., 2017). Heatmap and metaprofiles were generated using deeptools computeMatrix, plotHeatmap and plotProfile functions.
Motif and GO enrichment search
Motif search was performed using meme from the MEME suite (Bailey et al., 2009) with options “-dna -mod anr -revcomp -maxsize 25000000 -nmotifs 10 -minw 6 -maxw 12 -maxsites 10000 -brief 3000 -p 5”. All found motifs were then compared with the DAP-seq database of motifs (O’Malley et al., 2016) using Tomtom (Gupta et al., 2007) with default options. The Gene Ontology enrichment analyses were performed using GO-TermFinder (Boyle et al., 2004) via the Princeton GO-TermFinder interface (https://go.princeton.edu/cgi-bin/GOTermFinder), and then simplified using REVIGO (Supek et al., 2011; Langmead and Salzberg, 2012) and visualized as an unclustered heatmap using pheatmap (https://cran.r-project.org/package=pheatmap).
Pigment quantification
For anthocyanin quantification the aerial parts of 15 to 20 5-day-old seedlings, collected from different plates, were pooled for each replicate. Anthocyanin quantification was performed as described in Hillis and Swain, 1959. Six to 10 7-day-old seedlings were pooled for chlorophyll measurements. Acetone 80% (V/V) was used for extraction and A645 and A663 was measured in a spectrophotometer Data analysis was done according to Arnon, 1949. Three independent replicates (seedling pools) were measured for each sample. Values represent mean ± SD.
RNA analyses
For RT-qPCR assays, 2 ug total RNA extracted from 7 day old seedlings with the Favorprep Plant Total RNA Purification Mini kit (Favorgen) was used for cDNAs synthesis with using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems) with DNAse I treatment (Roche). Quantitative PCR was carried out using 5x PyroTaq qPCR mix Plus EvaGreen (CMB Cultek Molecular Bioline) in a QuantStudio5 machine (Applied Biosystems). Transcripts were amplified and results were normalized to PP2A transcript levels. Primers used for QPCR are represented in Table S5.
Antibodies
Antibodies used for immunoblot experiments: anti-GFP-HRP (Milteny Biotec #130-091-833); anti-H2B (Millipore #07–371); anti-MBP (Abcam #9084); anti-Actin (Sigma #A04080); anti-RPT5 (Enzo Life Sciences# BML-PW8245); anti-HY5 (Agrisera #AS121867 or Abiocode #R1245-1b); anti-COP1 (kindly provided by Xing Wang Deng); anti-mouse (ThermoFisher Scientific #A11001) or anti-rabbit (ThermoFisher Scientific #A11008) secondary antibodies.
GEO accession
ChIP-seq data generated in this work are accessible through GEO Series accession number GSE155147.
Author Contributions
C.B., F.B., V.R and S.F., designed the research and conceived the study. E.C., C.B., M.G-L., L.W., C. G-B., V.R. and S.F. performed experiments. C.B. analysed ChIP-seq data. C.B., F.B., V.R. and S.F. discussed the results; C.B. and S.F. wrote the initial version of the paper and V.R. and F.B. edited the manuscript.
Funding
This work was supported by a Ramon y Cajal (RYC-2014-16308) grant funded by the Ministerio de Economia y Competitividad to S.F.. Work by S.F. in F.B. lab was supported by the COST Action CA16212 INDEPTH (E.U.). Work in V.R.’s laboratory was funded by the Agencia Estatal de Investigación/Fondo Europeo de Desarollo Regional/European Union (BIO2016-80551-R and PID2019-105495GB-I00). Work in F.B.’s lab was supported by CNRS EPIPLANT Action (France) and funded by Agence Nationale de la Recherche (ANR) grants ANR-10-LABX-54, ANR-18-CE13-0004-01, ANR-17-CE12-0026-02 (France) and by Velux Stiftung (Switzerland).
Supplemental Information
Supplemental files contain:
Figure S1. COP1 expression levels in different mutants.
Figure S2. GFP-HY5/hy5 line analysis.
Figure S3. Analysis of HY5 targets in comparison with previous published binding and expression data.
Figure S4. De novo motif search under HY5 peaks annotated A to D gene classes.
Figure S5. Gene expression analysis of HY5 bound genes.
Figure S6. Analysis of PIF3 targets expression and binding sites and overlapping with HY5 binding classes.
Table S1. Interactomics of DET1 and COP1 proteins.
Table S2. List of DET1 and COP1 associated proteins found in the different replicates of TAP assays.
Table S3. List of DWD proteins that associate with DET1.
Table S4. Transcription factors that are direct targets of HY5 in wild-type.
Table S5. Oligonucleotides used in this study.
Supplemental Data 1 - HY5 targeted genes.
Supplemental Data 2 - PIF3 targeted genes in the dark.
Acknowledgements
We are grateful to Roberto Solano and Salomé Prat for the critical reading and suggestions on the manuscript.
Footnotes
↵3 Co-first authors
Contact: sfonseca{at}cnb.csic.es; (+34) 91 585 4681