Plant photoreceptors and their signaling components compete for binding to the ubiquitin ligase COP1 using their VP-peptide motifs

Plants sense different parts of the sun’s light spectrum using specialized photoreceptors, many of which signal through the E3 ubiquitin ligase COP1. Photoreceptor binding modulates COP1’s ubiquitin ligase activity towards transcription factors. Here we analyze why many COP1-interacting transcription factors and photoreceptors harbor sequence-divergent Val-Pro (VP) peptide motifs. We demonstrate that VP motifs enable different light signaling components to bind to the WD40 domain of COP1 with various binding affinities. Crystal structures of the VP motifs of the UV-B photoreceptor UVR8 and the transcription factor HY5 in complex with COP1, quantitative binding assays and reverse genetic experiments together suggest that UVR8 and HY5 compete for the COP1 WD40 domain. Photoactivation of UVR8 leads to high-affinity cooperative binding of its VP domain and its photosensing core to COP1, interfering with the binding of COP1 to its substrate HY5. Functional UVR8 – VP motif chimeras suggest that UV-B signaling specificity resides in the UVR8 photoreceptor core, not its VP motif. Crystal structures of different COP1 – VP peptide complexes highlight sequence fingerprints required for COP1 targeting. The functionally distinct blue light receptors CRY1 and CRY2 also compete with downstream transcription factors for COP1 binding using similar VP-peptide motifs. Together, our work reveals that photoreceptors and their components compete for COP1 using a conserved displacement mechanism to control different light signaling cascades in plants.

their components compete for COP1 using a conserved displacement mechanism to control different light signaling cascades in plants. ( COP1 is thus a crucial repressor of photomorphogenesis (Deng et al., 1991). COP1 contains an Nterminal zinc-finger, a central coiled-coil, and a C-terminal WD40 domain, which is essential for proper COP1 function (Deng et al., 1992;McNellis et al., 1994). Light-activated phytochrome, cryptochrome and UVR8 photoreceptors inhibit COP1's activity Hoecker, 2017;Podolec and Ulm, 2018). Although COP1 can act as a stand-alone E3 ubiquitin ligase in vitro (Seo et al., 2003;Saijo et al., 2003), it forms higher-order complexes in vivo, for example with SUPPRESSOR OF PHYA-105 (SPA) proteins (Hoecker and Quail, 2001;Zhu et al., 2008;Ordoñez-Herrera et al., 2015). COP1 can also act as a substrate adaptor in CULLIN4 -DAMAGED DNA BINDING PROTEIN 1 (CUL4-DDB1)-based heteromeric E3 ubiquitin ligase complexes . These different complexes may modulate COP1's activity towards different substrates (Ren et al., 2019). COP1 regulates gene expression and plays a central role as a repressor of photomorphogenesis by directly modulating the stability of transcription factors that control the expression of light-regulated genes (Lau and Deng, 2012;Podolec and Ulm, 2018). For example, the bZIP transcription factor ELONGATED HYPOCOTYL 5 (HY5) acts antagonistically with COP1 (Ang et al., 1998). COP1 binding to HY5 leads to its subsequent degradation via the 26S proteasome in darkness, a process that is inhibited by light (Osterlund et al., 2000).
UVR8 itself contains a conserved C-terminal VP-peptide motif that is critical for UV-B signaling (Cloix et al., 2012;Yin et al., 2015). Moreover, overexpression of the UVR8 C-terminal 44 amino acids results in a cop-like phenotype (Yin et al., 2015). A similar phenotype has been observed when overexpressing the COP1-interacting CRY1 and CRY2 C-terminal domains (CCT) (Yang et al., 2000(Yang et al., , 2001. Indeed, CRY1 and CRY2 also contain potential VP-peptide motifs within their CCT domains, but their function in blue-light signaling has not been established (Lin and Shalitin, 2003;Müller and Bouly, 2015). The presence of VP-peptide motifs in different light signaling components suggests that COP1 may use a common targeting mechanism to interact with downstream transcription factors and upstream photoreceptors. Here we present structural, quantitative biochemical and genetic evidence for a VP-peptide-based competition mechanism, enabling COP1 to play a crucial role in different photoreceptor pathways in plants.
We next assessed the impact of COP1 VP-peptide binding pocket mutants in UV-B signaling assays in planta. The seedling-lethal cop1-5 null mutant can be complemented by expression of YFP-COP1 driven by the CaMV 35S promoter. We introduced COP1 mutations into this construct and isolated transgenic lines in the cop1-5 background. All lines expressed comparable levels of the YFP-fusion proteins and complemented the seedling lethality of cop1-5 ( Figures 1F, 1G and S4).

High-affinity, cooperative binding of photoactivated UVR8
HY5 levels are stabilized in a UVR8-dependent manner under UV-B light (Favory et al., 2009;Huang et al., 2013). We hypothesized that COP1 is inactivated under UV-B light, by activated UVR8 preventing HY5 from interacting with COP1. Our analysis of the isolated VP-peptide motifs of UVR8 and HY5 suggests that UVR8 cannot efficiently compete with HY5 for COP1 binding.
However, it has been previously found that the UVR8 β-propeller core can interact with the COP1 WD40 domain independent of its VP motif (Yin et al., 2015). We thus quantified the interaction of UV-B activated full-length UVR8 with the COP1 WD40 domain. Recombinant UVR8 expressed in insect cells was purified to homogeneity, monomerized under UV-B, and analyzed in ITC and grating-coupled interferometry (GCI) binding assays. We found that UV-B-activated full-length UVR8 binds COP1 with a dissociation constant (K d ) of ~150 nM in both quantitative assays (Figures 2A and 2B) and ~10 times stronger than non-photoactivated UVR8 ( Figure S6A). This 1,000 fold increase in binding affinity compared to the UVR8 406-413 peptide indicates cooperative binding of the UVR8 β-propeller core and the VP-peptide motif. In line with this, UV-B-activated UVR8 monomers interact with the COP1 WD40 domain in analytical size-exclusion chromatography experiments, while the non-activated UVR8 dimer shows no interaction in this assay ( Figure S7A).
As the interaction of full-length UVR8 is markedly stronger than the isolated UVR8 VP-peptide, we next dissected the individual contributions of the individual UVR8 domains to COP1 binding ( Figure 2C). We find that the UV-B activated UVR8 β-propeller core (UVR8 12-381 ) binds COP1 with We could not detect sufficient binding enthalpies to monitor the binding of UVR8 ValPro/AlaAla to COP1 in ITC assays nor detectable signal in GCI experiments in the absence of UV-B ( Figure S8). The COP1 Lys422Ala mutant binds UV-B-activated full-length UVR8 with wild-type affinity, while COP1 Trp467Ala binds ~5 times more weakly ( Figures S9A and S9B). Mutations targeting both COP1 and the UVR8 C-terminal VP-peptide motif decreases their binding affinity even further ( Figure   S9C). Thus full-length UVR8 uses both its β-propeller photoreceptor core and its C-terminal VPpeptide to cooperatively bind the COP1 WD40 domain when activated by UV-B light.
We next asked if UV-B-activated full-length UVR8 could compete with HY5 for binding to COP1.
We produced the full-length HY5 protein in insect cells and found that it binds the COP1 WD40 domain with a K d of ~1 μM in GCI assays ( Figure 2F). For comparison, the isolated HY5 VPpeptide binds COP1 with a K d of ~20 μM ( Figure 1B). This would indicate that only the UV-Bactivated UVR8 and not ground-state UVR8 (K d ~ 150 nM vs ~ 1 μM, see above) can efficiently compete with HY5 for COP1 binding. We tested this hypothesis in yeast 3-hybrid experiments. We confirmed that HY5 interacts with COP1 in the absence of UVR8 and that this interaction is specifically abolished in the presence of UVR8 and UV-B light ( Figure 2G). We conclude that UV-B-activated UVR8 efficiently competes with HY5 for COP1 binding in yeast cells, thereby impairing the COP1 -HY5 interaction under UV-B. The UVR8 ValPro/AlaAla and UVR8 1-396 mutants cannot interfere with the COP1 -HY5 interaction in yeast cells ( Figure 2G), suggesting that a functional UVR8 VP-peptide motif is required to compete off HY5 from COP1, in agreement with our biochemical assays.

UVR8 -VP peptide chimeras trigger UV-B signaling in planta
Our findings suggest that UVR8 requires both its UV-B-sensing core and its VP-peptide motif for high affinity COP1 binding and that the UVR8 VP-peptide can inhibit the interaction of HY5 with COP1 ( Figures 1 and 2) (Yin et al., 2015). This led us to speculate that any VP-peptide with sufficient binding affinity for COP1 could functionally replace the endogenous VP motif in the UVR8 C-terminus in vivo. We generated chimeric proteins in which the UVR8 core domain is fused to VP-containing sequences from plant and human COP1 substrates, namely HY5 and TRIB1 ( Figure 3A). Arabidopsis uvr8-7 null mutants expressing these chimeric proteins show complementation of the hypocotyl and anthocyanin phenotypes under UV-B, suggesting that all tested UVR8 chimeras are functional (Figures 3B-D, and S10). Early UV-B marker genes are also up-regulated in the lines after UV-B exposure, demonstrating that these UVR8 chimeras are functional photoreceptors, although to different levels ( Figure 3E). In line with this, the UVR8 HY5C44 chimera can displace HY5 from COP1 in yeast 3-hybrid assays ( Figure 3F), can bind COP1 affinities comparable to wild-type (Figures 3G and S10) and are dimers in vitro that monomerize under UV-B ( Figure 3H). Together, these experiments reinforce the notion that divergent VPpeptide motifs compete with each other for binding to the COP1 WD40 domain.

Sequence-divergent VP-peptide motifs are recognized by the COP1 WD40 domain
Our protein engineering experiments prompted us to map core VP-peptide motifs in other plant light signaling components, including the COP1-interacting blue-light photoreceptors CRY1 and CRY2 (Yang et al., 2000;Wang et al., 2001;Yu et al., 2007;Yang et al., 2018) and the transcription factors HYH, CO/BBX1, COL3/BBX4, SALT TOLERANCE (STO/BBX24) (Holm et al., 2002;Datta et al., 2006;Jang et al., 2008;Yan et al., 2011), andHFR1 (Duek et al., 2004;Yang et al., 2005;Jang et al., 2005). We mapped putative VP-motifs in all these proteins and assessed their binding affinities to the COP1 WD40 domain ( Figures 4A and 4B). We could detect binding for most of the peptide motifs in ITC assays, with dissociation constants in the midmicromolar range ( Figure 4A and S11). Next, we obtained crystal structures for the different peptides bound to COP1 (1.3 -2.0 Å resolution, see Tables 1 and 2, Figures 4E, 4F and S11) to compare their peptide binding modes ( Figure 4C). We found that all peptides bind in a similar configuration with the VP forming the center of the binding site (r.m.s.d.'s between the different peptides range from ~0.3 Å to 1.5 Å, comparing 5 or 6 corresponding C α atoms). Chemically diverse amino-acids (Tyr/Arg/Gln) map to the -3 and -2 position and often deeply insert into the COP1 binding cleft, acting as anchor residues ( Figure 4C). This suggests that the COP1 WD40 domain has high structural plasticity, being able to accommodate sequence-divergent VP-containing peptides.
To experimentally investigate this property of COP1, we quantified the interaction of different VP peptides with our COP1 Lys422Ala mutant protein. As for UVR8 ( Figures 1B and 1E), COP1 Lys422Ala showed increased binding affinity for some peptides such as those representing the COL3 287-294 and CO 366-373 VP-motifs, while it reduced binding to others, such as to CRY1 544-552 and CRY2 [527][528][529][530][531][532][533][534][535] ( Figures 4A and 4D). These observations may be rationalized by an enlarged VP-binding pocket in the COP1 Lys422Ala mutant, increasing accessibility for the COL3 Phe288 anchor residue, and potentially abolishing interactions with CRY1 Asp545 (Figures 4E and 4F). In yeast 3-hybrid assays we find that, similar to HY5 ( Figure 2G), UV-B-activated UVR8 can efficiently compete with HYH, an Nterminal fragment of HFR1 and the CCT domain of CRY1 for binding to COP1 ( Figure S12). Taken together, VP-peptide motifs of cryptochrome photoreceptors and diverse COP1 transcription factor targets all bind to the COP1 WD40 domain and UVR8 is able to compete with COP1 partners for binding.

CRY2 and CONSTANS compete for COP1 binding
The structural plasticity of the COP1 WD40 domain is illustrated by the variable modes of binding for sequence-divergent VP motifs found in different plant light signaling components. The COP1 Lys422Ala mutation can modulate the interaction with different VP-peptides ( Figure 4A). We noted that the cop1-5/Pro 35S :YFP-COP1 Lys422Ala but not other COP1 mutants show delayed flowering when grown in long days ( Figures 5A-D). This phenotype has been previously associated with mutant plants that lack the COP1 substrate CO ( Figure 5B-D) (Putterill et al., 1995;Jang et al., 2008;. We thus hypothesized that in COP1 Lys422Ala plants, binding and subsequent degradation of CO may be altered under long day conditions. In vitro, we found that the CO VP-peptide binds COP1 Lys422Alã 4 times stronger than wild-type COP1 ( Figure 5F). The same mutation in COP1 strongly reduces (~30 times) binding of the CRY2 VP-peptide in vitro ( Figure 5F). It is of note that, in contrast to UVR8 ( Figure 1F), CRY2 levels are not altered in the COP1 Lys422Ala background ( Figure 5E). Thus, the late flowering phenotype of the COP1 Lys422Ala mutant suggests that CRY2 and CO compete for COP1 binding, and that this competition is altered in the COP1 Lys422Ala mutant background: reduced affinity to CRY2, enhanced binding to CO -both consistent with the late flowering phenotype. In line with this, we find that recombinant light-activated full-length CRY2 binds wildtype COP1 with nanomolar affinity in quantitative GCI experiments ( Figure 5G). This ~200 fold increase in binding affinity over the isolated CRY2 VP-peptide strongly suggests, that UVR8 and CRY2 both use a cooperative binding mechanism to target COP1. As a control, we tested a fragment of the CRY2 C-terminus containing the VP motif, the NC80 domain (CRY2 486-565 ) (Yu et al., 2007). We found that NC80 binds COP1 with an affinity comparable to the isolated CRY2 527-535 VP-peptide assayed by ITC ( Figures 5F and 5H). Together, the COP1 Lys422Ala phenotypes and our biochemical assays suggest that different plant photoreceptors may use a light-induced cooperative binding mechanism, preventing COP1 from targeting downstream light signaling partners for degradation.

DISCUSSION
The COP1 E3 ubiquitin ligase is a central hub in plant light sensing and signaling. There is strong evidence that the UV-B-sensing photoreceptor UVR8, the blue-light receptors CRY1 and CRY2 and the red/far-red discriminating phytochromes all regulate COP1 activity (Hoecker, 2017;Podolec and Ulm, 2018). The regulation of COP1 by photoreceptors enables a broad range of photomorphogenic responses, including de-etiolation, cotyledon expansion and transition to flowering, as well as UV-B light acclimation (Lau and Deng, 2012;Jenkins, 2017;Yin and Ulm, 2017;Gommers and Monte, 2018). Here we have dissected at the structural, biochemical and genetic level how the activated UVR8 and cryptochrome photoreceptors impinge on COP1 activity, by interacting with its central WD40 domain, resulting in the stabilization of COP1 substrate transcription factors. For both types of photoreceptors, interaction through a linear VP-peptide motif and a folded, light-regulated interaction domain leads to cooperative, high-affinity binding of the activated photoreceptor to COP1. We propose that in response to UV-B light, UVR8 dimers monomerize, exposing a new interaction surface that binds to the COP1 WD40 domain and releases the UVR8 C-terminal VP motif from structural restraints that prevent its interaction with COP1 in the absence of UV-B (Yin et al., 2015;Heilmann et al., 2016;Wu et al., 2019;Camacho et al., 2019). Similarly, the VP motif in the CCT domain of cryptochromes may become exposed and available for interaction upon blue-light activation of the photoreceptor (Müller and Bouly, 2015;Wang et al., 2018). Because UVR8 and CRY2 are very different in structure and domain 296 composition, they likely use distinct interaction surfaces to target the COP1 WD40 domain, in addition to the VP-peptide motifs. The cooperative, high-affinity mode of binding enables UVR8 and cryptochromes to efficiently displace downstream signaling components such as HY5, HYH, HFR1 and CO in a light-dependent manner. Structure-guided mutations in the COP1 WD40 binding cleft resulted in the identification of the COP1 Lys422Ala mutant, which displays flowering phenotypes, and COP1 Tyr441Ala and COP1 Trp467Ala , which display UV-B signaling phenotypes, that are all consistent with our competition model. Similar mutations have previously been shown to affect hypocotyl elongation in white light (Holm et al., 2001). Unexpectedly, COP1 Lys422Ala rendered the UVR8 protein unstable, preventing conclusive analysis of the effect of this COP1 mutant on UV-B signaling in vivo. Moreover, the mechanism behind UVR8 protein instability remains to be determined. Independent of this, it is interesting to note that the hy4-9 mutant, which replaces the proline in the CRY1 VP-peptide motif with leucine, does not show inhibition of hypocotyl elongation under blue light (Ahmad et al., 1995). Similarly, mutations of the UVR8 VP-peptide motif or C-terminal truncations (including the uvr8-2 allele, which has a premature stop codon at Trp400) all strongly impair UV-B signaling (Brown et al., 2005;Cloix et al., 2012;Yin et al., 2015).
We now report quantitative biochemical and crystallographic analyses that reveal that UVR8 and cryptochrome photoreceptors and their downstream transcription factors all make use of VPcontaining peptide motifs to target a central binding cleft in the COP1 WD40 domain. VPcontaining peptides were previously identified based upon a core signature motif E-S-D-E-x-x-x-V-P-[E/D]-Φ-G, where Φ designated a hydrophobic residue (Holm et al., 2001;Uljon et al., 2016).
Our structural analyses of a diverse set of VP-containing peptides now reveal that COP1 has evolved a highly plastic VP-binding pocket, which enables sequence-divergent VP motifs from different plant light signaling components to compete with each other for COP1 binding. It is reasonable to assume that many more bona fide VP-motifs may exist and our structures now provide sequence fingerprints to enable their bioinformatic discovery.
Interestingly, although we predict that at least some of our COP1 mutant variants (e.g. Trp467Ala) completely disrupt the interaction with VP-motif harboring COP1 targets, all COP1 variants can complement the cop1-5 seedling lethal phenotype and largely the cop phenotype in darkness (Holm et al., 2001;and this work). This could imply that a significant part of COP1 activity is independent from the VP-mediated destabilization of photomorphogenesis-promoting transcription factors. It has been recently suggested that part of the cop1 phenotype could be explained by COP1-mediated stabilization of PIFs (Pham et al., 2018). Our COP1 lines could be used to gain further insight into this aspect of COP1 activity. Human COP1 prefers to bind phosphorylated substrates and their post-translational regulation may also be relevant in plants (Hardtke et al., 2000;Uljon et al., 2016). In this respect it is noteworthy that the full-length COP1 protein may exist as an oligomer as well as in complex with other light signaling proteins, such as SPA proteins (Seo et al., 2003;Huang et al., 2013;Sheerin et al., 2015;Holtkotte et al., 2017). The four SPA protein family members share a similar domain architecture with COP1, consisting of an N-terminal kinase-like domain, a central coiled-coil domain and a Cterminal WD40 domain (~ 45 % amino-acid identity with the COP1 WD40 domain) and are partially redundant in their activities (Yang and Wang, 2006;Ordoñez-Herrera et al., 2015).
Mutations in the SPA1 WD40 domain residues Lys767 and Trp812, which correspond to COP1 residues Lys422 and Trp467, cannot complement the spa1-3 mutant (Yang and Wang, 2006). These higher-order complexes are known to be part of some but not all light signaling pathways and could thus encode additional determinants for signaling specificity (Hoecker, 2017;Podolec and Ulm, 2018). In addition to the competition mechanism presented here, it has been observed that active cryptochrome and phytochrome receptors directly interact with SPA proteins and thereby separate COP1 from SPA proteins, which results in COP1 inactivation (Lian et al., 2011;Liu et al., 2011;Zuo et al., 2011;Lu et al., 2015;Sheerin et al., 2015). However, early UVR8 signaling is independent of SPA proteins (Oravecz et al., 2006), and may thus rely exclusively on the competition mechanism described here. For cryptochrome signaling, the VP-mediated competition and COP1-SPA disruption mechanisms are obviously not mutually exclusive but likely function in parallel in vivo to reinforce COP1-SPA E3 ligase inactivation in blue light signaling. Reconstitution of a photoreceptor -COP1/SPA signaling complex may offer new insights into these different targeting mechanism in the future.

RMS deviations (angles) #
Average B-factor # # as defined by phenix.table_one and phenix.model_vs_data. *Data were collected from one crystal. $ Data were collected from two crystals and scaled between three datasets. @ Data were collected from two crystals and scaled between two datasets. & Data were collected from one crystal and scaled between two datasets.

CONTACT FOR REAGENT AND RESOURCE SHARING
Further information and requests for resources and reagents should be directed to, and will be fulfilled by the Lead Contact, Michael Hothorn (michael.hothorn@unige.ch)

Protein expression and purification
All COP1, UVR8, HY5 and CRY2 proteins were produced as follows: The desired coding sequence was PCR amplified (see Table S1 for primers) or NcoI/NotI digested from codon-optimized genes All protein concentrations were measured by absorption at 280 nm and calculated from their molar extinction coefficients. Molecular weights of all proteins were confirmed by MALDI-TOF mass spectrometry. SDS-PAGE gels to assess protein purity are shown in Figure S13.
For UVR8 monomerization and activation by UV-B, proteins were diluted to their final assay concentrations (as indicated in the figure legends) in Eppendorf tubes and exposed to 60 minutes at max intensity (69 mA) under UV-B LEDs (Roithner Lasertechnik GmbH) on ice.

Analytical size-exclusion chromatography
Gel filtration experiments were performed using a Superdex 200 Increase 10/300 GL column (GE Healthcare) pre-equilibrated in 150 mM NaCl, 20 mM HEPES 7.4, 2 mM BME. 500 μl of the respective protein solution or a mixture (~4 μM per protein) was loaded sequentially onto the column and elution at 0.75 ml/min was monitored by UV absorbance at 280 nm.

Isothermal titration calorimetry (ITC)
All experiments were performed in a buffer containing 150 mM NaCl, 20 mM HEPES 7.4, 2 mM BME. Peptides were synthesized and delivered as lyophilized powder (Peptide Specialty Labs GmbH) and dissolved directly in buffer. The peptides were centrifuged at 14000 x g for 10 minutes and only the supernatant was used. The dissolved peptide concentrations were calculated based upon their absorbance at 280 nm and their corresponding molar extinction coefficient. Typical experiments consisted of titrations of 20 injections of 2 μL of titrant (peptides) into the cell containing COP1 at a 10-fold lower concentration. Typical concentrations for the titrant were between 500 and 3000 μM for experiments depending on the affinity. Experiments were performed at 25°C and a stirring speed of 1000 rpm on an ITC200 instrument (GE Healthcare). All data were processed using Origin 7.0 and fit to a one-site binding model after background buffer subtraction.
Crystals were harvested and cryoprotected in mother liquor supplemented with 25% glycerol and frozen under liquid nitrogen.
All datasets were collected at beam line PX-III of the Swiss Light Source, Villigen, Switzerland.
Native datasets were collected with λ=1.03 Å. All datasets were processed with XDS (Kabsch, 1993) and scaled with AIMLESS as implemented in the CCP4 suite (Winn et al., 2011).

Crystallographic structure solution and refinement
The structures of all the peptide -COP1 WD40 complexes were solved by molecular replacement as implemented in the program Phaser (McCoy et al., 2007), using PDB-ID 5IGO as the initial search model. The final structures were determined after iterative rounds of model-building in COOT (Emsley and Cowtan, 2004), followed by refinement in REFMAC5 (Murshudov et al., 2011) as implemented in CCP4 and phenix.refine (Adams et al., 2010). Polder omit maps were generated for the UVR8 406-413 -COP1 structure by omitting residue Tyr407 of the bound peptide as implemented in phenix.polder. Final statistics were generated as implemented in phenix.table_one.

Plant transformation
To generate the cop1-5/Pro 35S :YFP-COP1 line, COP1 cloned into pENTR207C was introduced into the Gateway-compatible binary vector pB7WGY2 (Karimi et al., 2002). COP1 mutated versions were generated by PCR-based site-directed mutagenesis, cloned into pDONR207 and then introduced in pB7WGY2 ( Karimi et al., 2002). The wild-type version of the construct contains an additional Gateway-cloning related 14 amino acids linker sequence between the YFP and COP1.
cop1-5 heterozygous plants (kan R ) were transformed using the floral dip method (Clough and Bent, 1998). Lines homozygous for the cop1-5 mutation and for single locus insertions of the Pro 35S :YFP-COP1 transgene were selected.
To generate lines expressing chimeric UVR8 receptors, the HY5 and TRIB1 sequences were introduced by PCR to the UVR8 coding sequences as indicated, and the chimeras were cloned into the Gateway-compatible binary vector pB2GW7 (Karimi et al., 2002) for transformation into the uvr8-7 mutant background. Lines homozygous with single genetic locus transgene insertions were selected.
To generate a co mutant in the Ws background, designated co-11, plants were transformed with the CRISPR/Cas9 binary vector pHEE401E (Wang et al., 2015) in which an sgRNA specific to the CO CDS was inserted (see Table S1). A plant was isolated in T2 and propagated, harboring a 1 base-pair insertion after the codon for residue Asp137 leading to a frameshift and a premature stop codon after four altered amino acids (DPRGR*; D representing Asp137 in CO, * representing the premature stop).

Plant growth conditions
For experiments at seedling stage, Arabidopsis seeds were surface-sterilized and sown on half-

Hypocotyl length assays
For hypocotyl length measurements, at least 60 seedlings were randomly chosen, aligned and scanned. Measurements were performed using the NeuronJ plugin of ImageJ (Meijering et al., 2004). Violin and box plots were generated using the ggplot2 library in R (Wickham, 2009).

Anthocyanin quantification
Accumulation of anthocyanin pigments was assayed as described previously (Yin et al., 2012). In brief, 40 to 60 mg of seedlings were harvested, frozen and grinded before adding 250 μl acidic methanol (1% HCl). Samples were incubated on a rotary shaker for 1 hour and the supernatant was collected and absorbances at 530 and 655 nm were recorded using a spectrophotometer.
Anthocyanin concentration was calculated as (A 530 -2.5 * A 655 ) / mg, where mg is the fresh weight of the sample.
Proteins were separated by electrophoresis in 8% (w/v) SDS-polyacrylamide gels and transferred to PVDF membranes (Roth) according to the manufacturer's instructions (iBlot dry blotting system, ThermoFisher Scientific), except for CRY2 immunoblots, which were transferred on nitrocellulose membranes (Bio-Rad). followed by addition of DEPC-treated EDTA for inactivation at 65°C for 10 min. Reverse transcription was performed using Taqman Reverse Transcription reagents (Applied Biosystems), using a 1:1 mixture of oligo dT and random hexamer primers. Quantitative real-time PCR was performed on a QuantStudio 5 Real-Time PCR system (ThermoFisher Scientific) using PowerUp SYBR Green Master Mix reagents (Applied Biosystems). Gene-specific primers for CHS, COP1, ELIP2, HY5, RUP2, and UVR8 were described before (Favory et al., 2009;Gruber et al., 2010;Heijde et al., 2013) and 18S expression was used as reference gene (Vandenbussche et al., 2014), expression values were calculated using the ΔΔCt method (Livak and Schmittgen, 2001) and normalized to the wild-type. Each reaction was performed in technical triplicates; data shown are from three biological repetitions.

QUANTIFICATION AND STATISTICAL ANALYSIS
Data of ITC and GCI binding assays are reported with errors as indicated in their Figure legends.

DATA AND SOFTWARE AVAILABILITY
The atomic coordinates of complexes have been deposited with the following Protein Data Bank    (C) Superposition of the X-ray crystal structures of the HY5 and UVR8 peptides in the VP-peptide binding site of the COP1 WD40 domain. COP1 is depicted in surface representation and belongs to the HY5 -COP1 complex. The HY5 peptide is depicted in green in ball-and-stick representation (with Arg41 labelled). The UVR8 peptide from the UVR8 -COP1 complex is superimposed on top in purple (with Tyr407 labelled), depicted in ball-and-stick representation. The surface of COP1 has been clipped to better visualize the anchor residue in the COP1 WD40 domain.    Figure 2: High-affinity co-operative binding of activated full-length UVR8 to the COP1 WD40 domain is mediated by its UV-Bactivated β-propeller core and its C-terminal VP-peptide motif.
(A) Sequence alignment of the N-and C-termini of UVR8, chimera UVR8 HY5VP (replacing the UVR8 VP motif with the corresponding sequence from HY5), chimera UVR8 HY5C44 (replacing the C44 domain of UVR8 with the corresponding sequence from HY5), and chimera UVR8 TRIB1VP (replacing the UVR8 VP motif with the TRIB1 VP motif and a truncation of the rest of the UVR8 C-terminus). The black box indicates the core VP motif of UVR8. The VP is colored in red, the anchor residues in orange, the divergent residues of the HY5 VP core sequence in green, the HY5 sequence replacing the UVR8 C-terminal 44 amino acids (C44) in cyan, and the divergent residues of the TRIB1 VP core sequence in blue. Asterisks represent amino acids identical in all constructs, amino acids 21-390 of UVR8 are not shown. The previously crystallized UVR8 core domain (PDB: 4D9S) is highlighted with a gray bar.      Lys422 / HY5 peptide / COP1 HY5 peptide / COP1 Lys422Ala F Figure S3: X-ray crystal structures of COP1 wild-type and COP1 Lys422Ala WD40 domains bound to the UVR8 VP-peptide. (A) The crystal structure of the UVR8 VP-peptide depicted in purple in stick representation bound to the COP1 WD40 domain depicted in white also in stick representation. Only selected residues and water molecules in red are shown. The white mesh represents the 2mFo-DFc electron density map contoured around all atoms depicted at a level of 1 σ. The orange mesh represents the polder omit map depicted at a level of 2.5 σ and contoured only around Tyr407 of the UVR8 VP-peptide. Two different conformers of Tyr407 were visible in the electron density and were modeled as shown. (B) The crystal structure of the UVR8 VP-peptide depicted in purple in stick representation bound to the COP1 Lys422Ala WD40 domain depicted in white in stick representation. Only selected residues and water molecules in red are shown. The white mesh represents the 2mFo-DFc electron density map contoured around all atoms depicted at a level of 1 σ. (C,D) A surface representation of the UVR8 VP-peptide binding site of (C) wild-type COP1 and (D) COP1 Lys422Ala . COP1 is depicted in surface representation, the UVR8 peptide is depicted in green in ball-and-stick representation. Lys422 or Ala422 is highlighted in magenta along with their corresponding accessible surface area. (E,F) Superposition of the X-ray structures of the (E) UVR8 and (F) HY5 VP-peptides bound to the COP1 WD40 domain versus COP1 Lys422Ala . The UVR8 VP-peptides are depicted in ball-and-stick representation. Selected residues from COP1 are depicted stick representation. The wild-type structure is gray. In the COP1 Lys422Ala structure, the peptide is highlighted in purple and the residues in blue.   (A) Quantitative real-time PCR analysis of ELIP2 and CHS expression. Four-day-old seedlings grown in white light were exposed to narrowband UV-B for 2 hours (+UV-B), or not (-UV-B). Error bars represent SEM of 3 biological replicates. (B,C) Images of representative individuals (B) and quantification of hypocotyl lengths (C) of 4-day-old seedlings grown in darkness. Violin and box plots are shown for n > 60 seedlings.
(D) Proteins used in Figure 3 and S10.
(E) Proteins used in Figure 5.