SUMMARY
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.
INTRODUCTION
Flowering plants etiolate in darkness, manifested by the rapid elongation of the embryonic stem, the hypocotyl, and closed and underdeveloped embryonic leaves, the cotyledons. Under light and upon photoreceptor activation, seedlings de-etiolate and display a photomorphogenic phenotype, characterized by a short hypocotyl and open green cotyledons, enabling a photosynthetic lifestyle (Gommers and Monte, 2018). The constitutively photomorphogenic 1 (cop1) mutant displays a light-grown phenotype in the dark, including a short hypocotyl, and open and expanded cotyledons. COP1 is thus a crucial repressor of photomorphogenesis (Deng et al., 1991). COP1 contains an N-terminal 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 (von Arnim and Deng, 1994; 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 (Chen et al., 2010). 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).
In addition to HY5, other COP1 targets have been identified including transcriptional regulators, such as the HY5 homolog HYH (Holm et al., 2002), CONSTANS (CO) and other members of the BBX protein family (Jang et al., 2008; Liu et al., 2008; Khanna et al., 2009; Xu et al., 2016; Lin et al., 2018; Ordoñez-Herrera et al., 2018), and others such as LONG HYPOCOTYL IN FAR-RED (HFR1) (Jang et al., 2005; Yang et al., 2005) and SHI-RELATED SEQUENCE5 (SRS5) (Yuan et al., 2018). It has been suggested that specific Val-Pro (VP)-peptide motifs with a core sequence V-P-E/D-Φ-G, where Φ designated a hydrophobic residue, are able to bind the COP1 WD40 domain (Holm et al., 2001; Uljon et al., 2016). Deletion of regions containing the VP-peptide motifs result in loss of interaction of COP1 substrates with the COP1 WD40 domain (Holm et al., 2001; Jang et al., 2005; Datta et al., 2006). Recently, it has been reported that the human COP1 WD40 domain directly binds VP-motifs, such as the one from the TRIB1 pseudokinase that acts as a scaffold facilitating ubiquitination of human COP1 substrates (Uljon et al., 2016; Durzynska et al., 2017; Newton et al., 2018; Kung and Jura, 2019).
Arabidopsis photoreceptors for UV-B radiation (UV RESISTANCE LOCUS 8, UVR8), for blue light (cryptochrome 1 and 2, CRY1/CRY2) and for red/far-red light (phytochromes A-E), are known to repress COP1 activity in a light-dependent fashion (Yang et al., 2000; Wang et al., 2001; Yu et al., 2007; Favory et al., 2009; Jang et al., 2010; Lian et al., 2011; Liu et al., 2011; Zuo et al., 2011; Viczián et al., 2012; Huang et al., 2013; Lu et al., 2015; Sheerin et al., 2015; Yang et al., 2018). 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, 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.
RESULTS
The COP1 WD40 domain binds VP motifs from UVR8 and HY5
The WD40 domains of human and Arabidopsis COP1 can directly sense VP-containing peptides (Uljon et al., 2016). Such a VP-peptide motif can be found in the UVR8 C-terminus that is not part of the UV-B-sensing β-propeller domain (Figure 1A) (Kliebenstein et al., 2002; Rizzini et al., 2011; Christie et al., 2012; Wu et al., 2012), but is essential for UV-B signaling (Cloix et al., 2012; Yin et al., 2015). HY5 (Oyama et al., 1997), which is a COP1 target acting downstream of UVR8 in the UV-B signaling pathway (Ulm et al., 2004; Brown et al., 2005; Oravecz et al., 2006; Binkert et al., 2014), also contains a VP-peptide motif (Figure 1A) (Holm et al., 2001). UV-B absorption leads to UVR8 monomerization, COP1 binding and subsequent stabilization of HY5 (Favory et al., 2009; Rizzini et al., 2011; Huang et al., 2013). Mutation of the HY5 VP pair to alanine (AA) stabilizes the HY5 protein (Holm et al., 2001).
In order to compare how the VP-peptide motifs from different plant light signaling components bind COP1, we quantified the interaction of the UVR8 and HY5 VP peptides with the recombinant Arabidopsis COP1349-675 WD40 domain (termed COP1 thereafter) using isothermal titration calorimetry (ITC). We find that both peptides bind COP1 with micromolar affinity, with HY539-48 binding ~ 8 times stronger than UVR8406-413 (Figure 1B). Next, we solved crystal structures of the COP1 WD40 domain – VP-peptide complexes representing UVR8406-413 – COP1 and HY539-48 – COP1 interactions to 1.3 Å resolution (Figure 1C). Structural superposition of the two complexes (r.m.s.d. is ~0.2 Å comparing 149 corresponding Cα atoms) reveals an overall conserved mode of VP-peptide binding (r.m.s.d is ~ 1.2 Å comparing 6 corresponding Cα atoms), with the central VP residues making hydrophobic interactions with COP1Trp467 and COP1Phe595 (buried surface area is ~500 Å2 in COP1) (Figures 1D and S1). COP1Lys422 and COP1Tyr441 form hydrogen bonds and salt bridges with either UVR8Tyr407 or HY5Arg41, both being anchored to the COP1 WD40 core (Figures 1C, 1D, and S1), as previously seen for the corresponding TRIB1 residues in the COP1 – TRIB1 peptide complex (Uljon et al., 2016). In our HY539-48 – COP1 structure, an additional salt bridge is formed between HY5Glu45 and COP1His528 (Figure 1D). In the peptides, the residues surrounding the VP core adopt different conformations in UVR8 and HY5, which may rationalize their different binding affinities (Figures 1B and 1C). We tested this by mutating residues Lys422, Tyr441 and Trp467 in the VP-peptide binding pocket of COP1. Mutation of COP1Trp467 to alanine disrupts binding of COP1 to either UVR8 or HY5 derived peptides (Figures 1B and 1E). Mutation of COP1Tyr441 to alanine abolishes binding of COP1 to the UVR8 peptide and greatly reduces binding to the HY5 peptide (Figures 1B and 1E), in good agreement with our structures (Figure 1D). The COP1Lys422Ala mutant binds HY539-48 as wild-type, but increases the binding affinity of UVR8406-413 ~10-fold (Figures 1B and 1E). Interestingly, COP1Lys422Ala interacts with full-length UVR8 also in the absence of UV-B in yeast two-hybrid assays, which is not detectable for wild-type COP1 (Figure S2A) (Rizzini et al., 2011). Moreover, COP1Lys422Ala also interacts more strongly with the constitutively interacting UVR8C44 fragment (corresponding to the C-terminal UVR8 tail containing the VP motif) when compared to wild-type COP1 in yeast two-hybrid assays (Figure S2B). In contrast, COP1Tyr441Ala and COP1Trp467Ala show reduced interaction to both UVR8 and HY5 (Figure S2). A UVR8406-413 – COP1Lys422Ala complex structure reveals the UVR8 VP-peptide in a different conformation, with UVR8Tyr407 binding at the surface of the VP-binding pocket (Figures S3A-S3E). In contrast, a structure of HY539-48 – COP1Lys422Ala closely resembles the wild-type complex (Figure S3F).
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). We found that cop1-5/Pro35S:YFP-COP1Trp467Ala and cop1-5/Pro35S:YFP-COP1Lys422Ala transgenic lines have constitutively shorter hypocotyls when compared to wild-type or cop1-5/Pro35S:YFP-COP1 control plants (Figures 1G and 1H), in agreement with previous work (Holm et al., 2001), suggesting partially impaired COP1 activity. This is similar to the phenotype of cop1-4 (Figures 1G and 1H), a weak cop1 allele that is viable but fully impaired in UVR8-mediated UV-B signaling (McNellis et al., 1994; Oravecz et al., 2006; Favory et al., 2009). In contrast, cop1-5/Pro35S:YFP- COP1Tyr441Ala showed an elongated hypocotyl phenotype when compared to wild-type (Figures 1G and 1H), suggesting enhanced COP1 activity. However, in contrast to YFP-COP1, none of the YFP-COP1Lys422Ala, YFP-COP1Tyr441Ala or YFP-COP1Trp467Ala restored UV-B-induced marker gene activation like HY5, RUP2, ELIP2 and CHS to wild-type level (Figures 1I, 1J and S4A). Surprisingly, however, the YFP-COP1Lys422Ala line showed strongly reduced UVR8 levels (Figure 1F), despite showing normal UVR8 transcript levels (Figure S5), precluding any conclusion of the mutation’s effect on UV-B signaling per se. In contrast, YFP-COP1Tyr441Ala and YFP-COP1Trp467Ala were impaired in UV-B signaling, despite showing wild-type UVR8 protein levels (Figure 1F). This indicates strongly reduced UVR8 signaling, in agreement with the reduced affinity of the COP1 mutant proteins vs. UVR8406-413 in vitro (Figure 1E). Together, our crystallographic, quantitative biochemical and functional assays suggest that UVR8 and HY5 can specifically interact with the COP1 WD40 domain using sequence-divergent VP motifs, and that mutations in the COP1 VP-binding site can modulate these interactions and impair UVR8 signaling.
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 (Kd) 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 UVR8406-413 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 (UVR812-381) binds COP1 with a Kd of ~0.5 μM and interacts with the COP1 WD40 domain in size-exclusion chromatography experiments (Figures S6B and S7B). The interaction is strengthened when the C-terminus is extended to include the VP-peptide motif (UVR812-415) (Figures S6B and S6C). Mutation of the UVR8 VP-peptide motif to alanines results in ~20 fold reduced binding affinity when compared to the wild-type protein (Figures 2D). However, the mutant photoreceptor is still able to form complexes with the COP1 WD40 domain in size exclusion chromatography assays (Figures 2E). We could not detect sufficient binding enthalpies to monitor the binding of UVR8ValPro/AlaAla to COP1 in ITC assays nor detectable signal in GCI experiments in the absence of UV-B (Figure S8). The COP1Lys422Ala mutant binds UV-B-activated full-length UVR8 with wild-type affinity, while COP1Trp467Ala 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 VP-peptide 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 Kd of ~1 μM in GCI assays (Figure 2F). For comparison, the isolated HY5 VP-peptide binds COP1 with a Kd of ~20 μM (Figure 1B). This would indicate that only the UV-B-activated UVR8 and not ground-state UVR8 (Kd ~ 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 UVR8ValPro/AlaAla and UVR81-396 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 UVR8HY5C44 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 VP-peptide 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; Liu et al., 2008; Yan et al., 2011), and HFR1 (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 Cα 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 COP1Lys422Ala mutant protein. As for UVR8 (Figures 1B and 1E), COP1Lys422Ala showed increased binding affinity for some peptides such as those representing the COL3287-294 and CO366-373 VP-motifs, while it reduced binding to others, such as to CRY1544-552 and CRY2527-535 (Figures 4A and 4D). These observations may be rationalized by an enlarged VP-binding pocket in the COP1Lys422Ala mutant, increasing accessibility for the COL3Phe288 anchor residue, and potentially abolishing interactions with CRY1Asp545 (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 N-terminal 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 COP1Lys422Ala mutation can modulate the interaction with different VP-peptides (Figure 4A). We noted that the cop1-5/Pro35S:YFP-COP1Lys422Ala 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; Liu et al., 2008).
We thus hypothesized that in COP1Lys422Ala plants, binding and subsequent degradation of CO may be altered under long day conditions. In vitro, we found that the CO VP-peptide binds COP1Lys422Ala ~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 COP1Lys422Ala background (Figure 5E).
Thus, the late flowering phenotype of the COP1Lys422Ala mutant suggests that CRY2 and CO compete for COP1 binding, and that this competition is altered in the COP1Lys422Ala 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 wild-type 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 (CRY2486-565) (Yu et al., 2007). We found that NC80 binds COP1 with an affinity comparable to the isolated CRY2527-535 VP-peptide assayed by ITC (Figures 5F and 5H). Together, the COP1Lys422Ala 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 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 COP1Lys422Ala mutant, which displays flowering phenotypes, and COP1Tyr441Ala and COP1Trp467Ala, 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, COP1Lys422Ala 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 VP-containing peptide motifs to target a central binding cleft in the COP1 WD40 domain. VP-containing 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 C-terminal 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.
STAR METHODS 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{at}unige.ch)
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Sf9 Cell Culture
Spodoptera frugiperda Sf9 cells (Thermofisher) were cultured in Sf-4 Baculo Express insect cell medium (Bioconcept, Switzerland).
Yeast Strains
The following Saccharomyces cerevisiae reporter strains were used: L40 (MATa trp1 leu2 his3 ade2 LYS2::lexA-HIS3 URA3::lexA-lacZ GAL4) (Vojtek and Hollenberg, 1995), Y190 (MATa ura3-52 his3-200 lys2-801 ade2-101 trp1-901 leu2-3 112 gal4Δ gal80Δ cyhr2 LYS2::GAL1UAS-HIS3TATA-HIS3 MEL1 URA3::GAL1UAS-GAL1τATA-lacZ), Y187 (MATa ura3-52 his3-200 ade2-101 trp1-901 leu2-3 112 gal4Δ met− gal80Δ URA3::GAL1UAS-GAL1TATA-lacZ MEL1) (Yeast Protocols Handbook, Clontech).
Plants
cop1-4 (Oravecz et al., 2006) cop1-5 (McNellis et al., 1994), cop1-5/Pro35S:YFP-COP1, cop1-5/Pro35S:YFP-COP1Lys422Ala, cop1-5/Pro35S:YFP-COP1Tyr441Ala, cop1-5/Pro35S:YFP-COP1Trp467Ala (this work), uvr8-7 (Favory et al., 2009), uvr8-7/Pro35S:UVR8HY5C44, uvr8-7/Pro35S:UVR8HY5VP, and uvr8-7/Pro35S:UVR8TRIB1 (this work) are in the Arabidopsis thaliana Wassilewskija (Ws) accession. The cry2-1 (Guo et al., 1998) mutant is in the Columbia accession. The co-11 allele was generated in the Ws accession (this work) using CRISPR/Cas9 technology (Wang et al., 2015).
METHOD DETAILS
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 (Geneart or Twist Biosciences) for expression in Sf9 cells. Chimeric UVR8 constructs were PCR amplified directly from vectors used for yeast 3-hybrid assays (see below). All constructs except CRY2NC80 were cloned into a modified pFastBac (Geneva Biotech) insect cell expression vector, via NcoI/NotI restriction enzyme sites or by Gibson assembly (Gibson et al., 2009). The modified pFastBac vector contains a tandem N-terminal His10-Twin-Strep-tags followed by a TEV (tobacco etch virus protease) cleavage site. CRY2NC80 was cloned into a modified pET-28 a (+) vector (Novagen) containing a tandem N-terminal His10-Twin-Strep-tags followed by a TEV (tobacco etch virus protease) cleavage site by Gibson assembly. Mutagenesis was performed using an enhanced plasmid mutagenesis protocol (Liu and Naismith, 2008).
pFastBac constructs were transformed into DH10MultiBac cells (Geneva Biotech), white colonies indicating successful recombination were selected and bacmids were purified by the alkaline lysis method. Sf9 cells were transfected with the desired bacmid with Profectin (AB Vector). eYFP-positive cells were observed after 1 week and subjected to one round of viral amplification. Amplified, untitred P2 virus (between 5 – 10 % culture volume) was used to infect Sf9 cells at a density between 1-2 x 106 cells/mL. Cells were incubated for 72 h at 28°C before the cell pellet was harvested by centrifugation at 2000 x g for 20 minutes and stored at −20°C.
CRY2NC80 was produced in transformed Rosetta (DE3) pLysS cells. E. coli were grown in 2xYT broth. 1 L of broth was inoculated with 20 mL of a saturated overnight preculture, grown at 37°C until OD600 ~ 0.5, induced with IPTG at a final concentration of 0.2 mM, and then shaken for another 16 h at 18°C. The cell pellet was harvested by centrifugation at 2000 x g for 20 minutes and stored at −20°C.
Every 1 L of Sf9 or bacterial cell culture was dissolved in 25 mL of Buffer A (300 mM NaCl, 20 mM HEPES 7.4, 2 mM BME), supplemented with glycerol (10% v/v), 5 μL Turbonuclease and 1 Roche cOmplete Protease inhibitor tablets. Dissolved pellets were lysed by sonication and insoluble materials were separated by centrifugation at 60000 x g for 1 h at 4°C. The supernatant was filtered through tandem 1 μm and 0.45 μm filters before Ni2+-affinity purification (HisTrap excel, GE Healthcare). Ni2+-bound proteins were washed with Buffer A and eluted directly into a coupled Strep-Tactin Superflow XT column (IBA) by Buffer B (500 mM NaCl, 500 mM imidazole pH 7.4, 20 mM HEPES pH 7.4). Twin-Strep-tagged-bound proteins on the Strep-Tactin column were washed with Buffer A and eluted with 1X Buffer BXT (IBA). Proteins were cleaved overnight at 4°C with TEV protease. Cleaved proteins were subsequently purified from the protease and affinity tag by a second Ni2+-affinity column or by gel filtration on a Superdex 200 Increase 10/300 GL column (GE Healthcare). Proteins were concentrated to 3 – 10 mg/mL and either used immediately or aliquoted and frozen directly at −80°C. Typical purifications were from 2 to 5 L of cell pellet.
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.
Grating-coupled interferometry (GCI)
The Creoptix WAVE system (Creoptix AG), a label-free surface biosensor was used to perform GCI experiments. All experiments were performed on 2PCH or 4PCH WAVEchips (quasi-planar polycarboxylate surface; Creoptix AG). After a borate buffer conditioning (100 mM sodium borate pH 9.0, 1 M NaCl; Xantec) COP1 (ligand) was immobilized on the chip surface using standard amine-coupling: 7min activation (1:1 mix of 400mM N-(3-dimethylaminopropyl)-N’-ethylcarbodiimide hydrochloride and 100 mM N-hydroxysuccinimide (both Xantec)), injection of COP1 (10 μg/mL) in 10 mM sodium acetate pH 5.0 (Sigma) until the desired density was reached, and final quenching with 1 M ethanolamine pH 8.0 for 7 min (Xantec). Since the analyte CRY2 showed nonspecific binding on the surface, BSA (0.5% in 10 mM sodium acetate, pH 5.0; BSA from Roche) was used to passivate the surface between the injection of COP1 and ethanolamine quenching. For a typical experiment, the analyte (UVR8/CRY2) was injected in a 1:3 dilution series (highest concentrations as indicated in the figure legends) in 150 mM NaCl, 20 mM HEPES 7.4, 2 mM BME at 25°C. Blank injections were used for double referencing and a dimethylsulfoxide (DMSO) calibration curve for bulk correction. Analysis and correction of the obtained data was performed using the Creoptix WAVEcontrol software (applied corrections: X and Y offset; DMSO calibration; double referencing) and a one-to-one binding model with bulk correction was used to fit all experiments.
Protein crystallization and data collection
Crystals of co-complexes of HY539-48 – COP1349-675, UVR8406-413 – COP1349-675, HY539-48 – COP1349-675, Lys422Ala, UVR8406-413 – COP1349-675, Lys422Ala and COL3287-294 – COP1349-675 were grown in sitting drops and appeared after several days at 20°C when 5 mg/mL of COP1 supplemented with 3 to 10 fold molar excess in peptide was mixed with two-fold (v/v) more mother liquor (1:2 ratio; protein:buffer) containing 2 M (NH4)2SO4 and 0.1 M HEPES pH 7.4 or 0.1 M Tris pH 8.5. Crystals were harvested and cryoprotected in mother liquor supplemented with 25% glycerol and frozen under liquid nitrogen.
Crystals of complexes of HYH27-34 – COP1349-675, HFR157-64 – COP1349-675, STO240-247 – COP1349-675 and CRY1544-552 – COP1349-675 were grown in sitting drops and appeared after several days at 20°C when 5 mg/mL of COP1 supplemented with 3 to 10 fold molar excess in peptide was mixed with two-fold (v/v) more mother liquor (1:2 ratio; protein:buffer) containing 1.25 M sodium malonate pH 7.5. 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 UVR8406-413 - 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. All figures were rendered in UCSF Chimera (Pettersen et al., 2004).
Plant transformation
To generate the cop1-5/Pro35S: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 (kanR) 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 Pro35S: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 halfstrength MS medium (Duchefa), stratified in the dark at 4°C for 48 h, and grown under aseptic conditions in controlled light conditions at 21°C. For hypocotyl length and anthocyanin measurements, the MS medium was supplemented with 1% sucrose (AppliChem). For flowering experiments, Arabidopsis plants were grown on soil in long day (16 h / 8 h; light / dark cycles) growth chambers at 21°C.
UV-B treatments were performed as described before, using Osram L18W/30 tubes, supplemented with narrowband UV-B from Philips TL20W/01RS tubes (Oravecz et al., 2006; Favory et al., 2009).
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 (A530 - 2.5 * A655) / mg, where mg is the fresh weight of the sample.
Protein extraction and immunoblotting
For total protein extraction, plant material was grinded and incubated with an extraction buffer composed of 50 mM Na-phosphate pH 7.4, 150 mM NaCl, 10% (v/v) glycerol, 5 mM EDTA, 0.1% (v/v) Triton X-100, 1 mM DTT, 2 mM Na3VO4, 2 mM NaF, 1% (v/v) Protease Inhibitor Cocktail (Sigma) and 50 μM MG132, as previously described (Arongaus et al., 2018).
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).
For protein gel blot analyses, anti-UVR8(426-440) (Favory et al., 2009), anti-UVR8(1-15) (Yin et al., 2015), anti-UVR8(410-424) (Heijde and Ulm, 2013), anti-CHS (sc-12620; Santa Cruz Biotechnology), anti-GFP (Living Colors® A.v. Monoclonal Antibody, JL-8; Clontech), anti-actin (A0480; Sigma-Aldrich) and anti-CRY2(588-602) (Eurogentec, raised against the peptide N’-CEGKNLEGIQDSSDQI-C’ and affinity purified) were used as primary antibodies. Horseradish peroxidase-conjugated antirabbit and anti-mouse immunoglobulins (Dako) were used as secondary antibodies. Signal detection was performed using the ECL Select Western Blotting Detection Reagent (GE Healthcare) and an Amersham Imager 680 camera system (GE Healthcare).
Quantitative real-time PCR
RNA was extracted from seedlings using the RNeasy Plant Mini kit (Qiagen) following the manufacturer’s instructions. RNA samples were treated for 20 min with RNA-free DNAse (Qiagen) 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.
Flowering time assays
For quantitative flowering time measurements, the number of days to flowering was determined at bolting, and rosette and cauline leaf numbers were counted when the inflorescence reached approximately 1 cm in length (Möller-Steinbach et al., 2010).
Yeast 2-hybrid and 3-hybrid assays
For yeast 2-hybrid assay, COP1 and its mutated variants were introduced into pGADT7-GW (Marrocco et al., 2006; Yin et al., 2015) and HY5, UVR8 and UVR8C44 were introduced into pBTM116-D9-GW (Stelzl et al., 2005; Yin et al., 2015; Binkert et al., 2016). Vectors were cotransformed into the L40 strain (Vojtek and Hollenberg, 1995) using the lithium acetate-based transformation protocol (Gietz, 2014). Transformants were selected and grown on SD/-Trp/-Leu medium (Formedium). For analysis of β-galactosidase activity, enzymatic assays using chlorophenol red-β-D-galactopyranoside (Roche Applied Science) as substrate were performed as described (Yeast Protocols Handbook, Clontech).
For yeast 3-hybrid analysis, pGADT7-GW-COP1 was transformed into the Y190 strain (Harper et al., 1993). HY5, HYH and HFR1N186 were cloned into the BamHI/EcoRI site of pBridge (Clontech), and UVR8, UVR8ValPro/AlaAla, UVR81-396 and UVR8HY5C44 were cloned into the BglII/PstI cloning site, followed by transformation into the Y187 strain (Harper et al., 1993). Transformants were mated, selected and grown on SD/-Trp/-Leu/-Met medium (Formedium). For analysis of β-galactosidase activity, filter-lift assays were performed as described (Yeast Protocols Handbook, Clontech). Enzymatic assays using chlorophenol red-β-D-galactopyranoside (Roche Applied Science) were performed as described (Yeast Protocols Handbook, Clontech). For repression of ProMet25:UVR8 expression, SD/-Trp/-Leu/-Met medium was supplemented with 1 mM L-methionine (Fisher Scientific).
For assays, yeast cells were grown for 2 days at 30°C in darkness or under narrow-band UV-B (Philips TL20W/01RS; 1.5 μmol m−2 s−1), as indicated.
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 accession codes: HY539-48 – COP1349-675 : 6QTO, UVR8406-413 – COP1349-675 : 6QTQ, HY5 – COP1349-675, Lys422Ala: 6QTR UVR8 – COP1349-675, Lys422Ala: 6QTS HYH27-34 – COP1349-675: 6QTT STO240-247 – COP1349-675 : 6QTU, HFR157-64 – COP1349-675 : 6QTV, CRY1544-552 – COP1349-675 : 6QTW and COL3287-294 – COP1349-675 : 6QTX.
Acknowledgments
This work was supported by an HHMI International Research Scholar Award to M.H., the Swiss National Science Foundation (grant number 31003A_175774 to R.U.), and the European Research Council (ERC) under the European Union’s Seventh Framework Programme (grant no. 310539 to R.U.). R.P. was supported by an iGE3 PhD Salary Award, K.L. was supported by an EMBO Longterm Fellowship (ALTF 493-2015). We thank Luis Lopez-Molina for technical assistance with the UVB LEDs.
Footnotes
↵4 Lead Contact