Flavonoid pathway intermediates implicate UVR8 in functions beyond canonical UV-B signaling

Abstract

Plants integrate light signals to regulate development and produce protective compounds such as flavonoids under excessive light stress. While these metabolites protect against light damage, their roles in modulating photoreceptor responses remain unclear. Here, we show that naringenin chalcone (NGC), a flavonoid precursor that accumulates in the Arabidopsis chalcone isomerase mutant tt5 but which is also present in wild type plants under solar light, promotes UVR8 monomer accumulation beyond its canonical UV-B response. A genetic suppressor screen reveals that loss of UVR8 restores normal growth of tt5 mutants grown under high light, irrespective of the presence of UV-B. Biochemical and transcriptomic analyses show that NGC binds to monomeric UVR8, stabilizing its active form and triggering gene expression changes, even in the absence of UV-B. This work identifies a mechanism for stabilizing activated UVR8 and reveals crosstalk between light-induced metabolic intermediates and photoreceptor-mediated developmental regulation.

Whether excessive light activated photo-protective compounds accumulation functions as sunscreen or participating in modulating photoreceptor responses in plants is still unclear. Here, the authors provide evidences to support the role of flavonoid pathway intermediates in integration of light signals into plant development.

Introduction

Plants obtain their energy from sunlight through photosynthesis, and photosynthetically active radiation is normally considered in the 400–700 nm range¹. Plants sense light composition through a cadre of photoreceptors, which ultimately integrate light information with other environmental signals (such as temperature) into precise developmental and metabolic responses^2–5. Plant photoreceptors include light-oxygen voltage sensors, phytochromes, flavin-binding blue-light sensors, and cryptochromes. Each photoreceptor absorbs a portion of the solar radiation spectrum that reaches Earth (280–750 nm), often through associated chromophores that serve as photon absorption sites³. For example, phytochromes perceive red/far red (R/FR) light (600–750 nm); cryptochromes, phototropins and flavin-binding F-box proteins perceive blue/ultraviolet-A (UV-A) light (320–500 nm); and UV RESISTANCE LOCUS8 (UVR8) perceives UV-A (315–400 nm) and UV-B light (280–315 nm)^3,5. The photon contribution of UV light is about 10% of the photosynthetically active radiation, with 95% or more corresponding to UV-A, although geographical location, time of day, cloud and canopy coverage significantly affect the relative influence of UV-B⁶.

UV-B, a minor yet unavoidable component of solar light, can have detrimental effects on plants, including DNA damage primarily through the formation of cyclobutane-pyrimidine dimers, inhibition of the photosynthetic machinery inhibition (e.g., degradation of photosystem II components), and the induction of reactive oxygen species², which oxidize lipids and other cellular components^7–9. The mechanisms evolved by plants to minimize the deleterious UV-B effects include the photo repair of UV-induced pyrimidine dimers through a process known as photoreactivation^6,10; the induction of antioxidant enzymes and compounds¹¹; and the accumulation of UV-absorbing compounds including hydroxycinnamate esters and flavonoids, phenolic compounds that accumulate primarily in epidermal cells in response to UV-B radiation and absorb a significant fraction of the UV light, minimizing damage to deeper tissue layers^7,9,12. Indeed, Arabidopsis mutants lacking sinapate esters because of a mutation in the ferulic acid hydroxylase (fah1) gene, or flavonoids resulting from loss of CHALCONE ISOMERASE (CHI, encoded by TT5, TRANSPARENT TESTA5) are significantly more sensitive to UV-B radiation¹³. CHI is responsible for the conversion of naringenin chalcone (NGC) produced by CHALCONE SYNTHASE (CHS) encoded by TT4 in Arabidopsis, to the flavanone naringenin, a central intermediate in the biosynthesis of flavonols and anthocyanin pigments¹⁴. Arabidopsis tt5 mutants grown in conditions that mimic solar light showed growth defects that were partially ameliorated by filtering UV-B light¹⁵. The sensitivity of tt5 to UV-B was used for a screen of mutants that displayed enhanced UV-B sensitivity, resulting in the identification of UVR8¹⁶. UVR8 not only controls UV-B stimulated protective responses, including flavonoid biosynthesis¹⁷, but it functions as a UV-B photoreceptor participating in a number of photomorphogenic processes^18,19. Under normal growth conditions, UVR8 is majorly present as an inactive dimer in the cytoplasm²⁰. Upon UV-B absorption, UVR8 rapidly monomerizes and translocates to the nucleus, where it interacts with several proteins, including CONSTITUTIVELY PHOTOMORPHOGENIC1 (COP1), a component of the E3 ligase complex, inhibiting the COP1-mediated degradation of the ELONGATED HYPOCOTYL5 (HY5) transcription factor, in part responsible for the induction of various phenolic compounds and anthocyanins^18,21–24. Two related WD40 domain proteins, REPRESSOR OF UV-B PHOTOMORPHOGENESIS1 (RUP1) and RUP2, negatively and redundantly regulate UVR8 activity by promoting reversion to the dimer^20,25. Disruption of this UVR8 photo-equilibrium, for example, by strong overexpression of a constitutive monomeric version of UVR8, results in strong photomorphogenic phenotypes, even in the dark²⁶. Thus, maintaining the appropriate monomeric/dimeric UVR8 balance is central for proper plant development.

Here, we describe a mechanism by which this balance is disrupted by flavonoid pathway intermediates that accumulate in tt5 mutants when exposed to moderate high light (HL, ~ 180 μmol cm⁻²s⁻¹) in a growth chamber, resulting in a plethora of developmental defects. Our studies are prompted by the finding that tt5 mutants grown in a chamber lacking FR light exhibit stunted growth and several phenotypes, besides the absence of anthocyanin pigments. Under these conditions, wild-type Col-0 plants strongly induce anthocyanin accumulation, and we determine that the tt5 developmental phenotypes are recapitulated in HL but disappear when we cross tt4 into the tt5 mutant, indicating that NGC or its derivatives are responsible. Transcriptome analyses identify many genes specifically affected by HL in tt5, most significantly genes involved in the biosynthesis and response to salicylic acid (SA). High levels of SA and SA glucoside are confirmed by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). To identify molecular factors impacted by NGC, we screen for genetic suppressors that restore the developmental phenotypes of tt5 mutants under HL conditions. We identify and characterize five suppressors, two corresponding to tt4, and three to loss-of-function alleles of UVR8. The development defects of tt5 in HL continue to be present even when UV-B light is eliminated, suggesting a function of UVR8 in plant development that extends beyond the effects promoted by UV-B. We show that tt5 plants grown under HL have significantly increased levels of monomeric UVR8 protein, and that the stabilization of the monomer is recapitulated in vitro using pure NGC or methanol extracts of tt5 plants grown in HL. Using purified recombinant UVR8, nano-differential scanning fluorimetry (nanoDSF) and microscale thermophoresis (MST), we show that NGC, but not other flavonoids, binds to UVR8 monomers, but not to dimers. Gene expression analyses by RNA-sequencing (RNA-seq) comparing mutant and wild-type Arabidopsis plants identify UV-B independent UVR8-controlled genes that overlap with those induced by HL in tt5, indicating that NGC (or derivatives) modulate an UVR8 activity that is unrelated to UV-B. Though nearly undetectable under artificial growth conditions, substantial levels of NGC derivatives are observed in Arabidopsis wild-type plants exposed to sunlight, highlighting the potential role of flavonoid pathway intermediates in modulating UVR8 activity-likely as part of light signal integration.

Results

Developmental defects of Arabidopsis tt5 mutants in high light

Serendipitously, we grew wild-type (WT, Col-0) and tt5 mutant plants in two different growth chambers under similar low light intensities (70–80 μmol cm⁻²s⁻¹). After approximately five weeks of growth, we noticed that, while in one chamber, tt5 mutants were indistinguishable from WT (Fig. 1a), in the other, tt5 mutant plants (two independent alleles, tt5-2 and tt5-4) were stunted and showed a variety of abnormal phenotypes (Fig. 1b and Supplementary Table 1). The abnormal growth phenotypes were not observed in tt4 (tt4-11) mutant plants lacking the CHS activity that produces NGC, the substrate of TT5/CHI (Supplementary Fig. 1). When we analyzed the light spectra of both chambers, we determined that the second chamber had less red light and almost no far-red light (Fig. 1c, d, compare 600–780 nm wavelength profiles). This resulted from the absence of incandescent bulbs (see “Methods”), highlighting the importance of checking light spectra and intensity before conducting experiments. We call the first chamber the “High R + FR chamber” (Fig. 1a, c), and the second as “Low R + FR chamber” (Fig. 1b, d). Our results show that the R/FR imbalance present in the second chamber (Fig. 1b) efficiently promotes anthocyanin accumulation in WT plants. Both R and FR lights were shown to induce anthocyanin accumulation in Arabidopsis²⁷; the inhibitory effect of FR on the anthocyanin-inductive effect of R light was not anticipated. Alterations in the R/FR ratio are also known to cause several growth and developmental phenotypes in Arabidopsis²⁸; the fact that we only see the effects in tt5-2 and tt5-4 indicates that this is not a general phytochrome response, but rather that something is unique to these mutants.

Fig. 1. Arabidopsistt5 mutants exhibit morphological phenotypes.

a Wild type (Col-0), tt4-11, tt5-2, and tt5-4 plants were grown for 5 weeks in one chamber. b Images of plants with the same genotypes as (a), recorded at 5 weeks under another chamber. tt5 mutants demonstrate development defects. Scale bar: 1 cm. c, d Light spectra of the growth conditions in (a) and (b), respectively. PFD-near UV near-UV (380–400 nm); PFD-B blue wavelength (400–500 nm); PFD-G green wavelength (500–600 nm); PFD-R red wavelength (600–700 nm); PFD-FR far-red wavelength (700–780 nm); PFD photon flux density; R red wavelength; FR far-red wavelength; PPFD photosynthetic photon flux density. Source data are provided as a Source Data file.

Because WT plants in the Low R + FR chamber accumulated significant amounts of anthocyanins (Fig. 1b), we reasoned that high light conditions, known to induce anthocyanin accumulation²², could phenocopy the observed effect of low R + FR light. When we grew plants in full-spectrum moderate-high light (HL, ~ 180 μmol cm⁻² s⁻¹, Supplementary Fig. 2), we found that tt5-2 plants displayed identical phenotypes (Supplementary Fig. 3) as observed in the Low R + FR chamber. Expression in tt5-2 of the TT5 coding region fused to mCherry, driven by the native TT5 promoter²⁹, fully complemented the developmental phenotypes of tt5-2 in HL (Supplementary Fig. 3), indicating that the observed phenotypes are due to tt5 and not another mutation. Moreover, these phenotypes were not observed in tt4-11 plants, and they were suppressed by crossing tt4-11 into tt5-2 (Supplementary Fig. 3). These results demonstrate that the phenotypes observed for tt5 under HL, which included early senescence, reduced inflorescence height and internode length, altered phyllotaxis of flowers and siliques, and reduced pedicel length (Supplementary Fig. 4 and Supplementary Table 1), result from the accumulation of a CHS (TT4) product that cannot be further metabolized in the tt5 mutants. To further test this, we expressed the Antirrhinum majus chalcone 4’-O-glucosyltransferase (4’CGT, an activity normally not present in Arabidopsis, Supplementary Fig. 1), which uses NGC as substrate³⁰, in tt5-2 plants under the constitutive CaMV 35S promoter (p35S::Am4’CGT) and found that the developmental phenotypes characteristic of tt5 in HL disappeared (Supplementary Fig. 5). We conclude that NGC or a derivative that accumulates in tt5 mutants is responsible for the developmental phenotypes observed in tt5 under HL.

HL induces major transcriptome changes in tt5

To determine the gene expression changes potentially responsible for the developmental phenotypes of tt5 under HL, we conducted RNA-seq on 3- and 4-week-old Col-0, tt4-11, tt5-2, and tt4-11 tt5-2 plants under both moderate high light (HL, ~ 180 μmol cm⁻²s⁻¹) and normal light (NL, ~ 80 μmol cm⁻²s⁻¹). Plants grown three weeks in HL conditions show none of the developmental defects associated with tt5 (Supplementary Fig. 6a) which start becoming evident four weeks in HL conditions (Supplementary Fig. 6b). We first conducted differentially expressed gene (DEG) analyses between each genotype grown in HL versus NL for three (Fig. 2a) and four weeks (Fig. 2b). The majority of the DEGs were shared between the three genotypes, yet this is more pronounced at three weeks (2287 + 2206 = 4493 genes; Fig. 2a) with about half that number (516 + 1397 = 1913; Fig. 2b) at four weeks.

Fig. 2. Arabidopsis plants were subjected to RNA-seq analysis and to salicylic acid accumulation measurements under both high light (HL) and normal light (NL) conditions.

Differential expressed genes (DEGs) and gene ontology (GO) enrichment analysis were conducted on RNA-seq results. The Venn diagrams illustrate shared and unique DEGs among Col-0 HL vs Col-0 NL, tt4-11 HL vs tt4-11 NL, and tt5-2 HL vs tt5-2 NL for 3 (a) or 4 weeks (b). We conducted GO enrichment analysis of the DEGs unique to tt5-2. The x-axis indicates the fold enrichment for each biological process GO term. The top ten most significantly enriched GO terms are displayed. Red bars represent GO-term enrichment of up-regulated genes, while blue bars represent GO-term enrichment of down-regulated genes. c DEGs and GO enrichment analysis. Comparisons were made between unique DEGs in tt5-2 at 3 and 4 weeks. d Salicylic acid accumulation. Quantification of salicylic acid (SA) was performed using targeted LC-MS/MS. The amount of SA (left panel, nmol g⁻¹ dry weight) was assessed under normal light (gray bars) and high light (red bars) conditions. Experiments were conducted in biological triplicate, with each replicate corresponding to compounds extracted from four plants of the same genetic background. Error bars represent the standard deviation of the average, and p-values were calculated using a two-sided t-test. P-values between Col-0 NL vs Col-0 HL, tt4-11 NL vs tt4-11 HL, tt5-2 NL vs tt5-2 HL, and tt4-11 tt5-2 NL vs tt4-11 tt5-2 HL are 0.115502, <0.00001, 0.000021, and 0.000541, respectively. Source data are provided as a Source Data file.

Unique sets of DEGs characterized the response of Col-0, tt4-11 and tt5-2 to HL (Fig. 2a, b and Supplementary Data 1 and 2). One group of genes that was notably induced solely in tt5-2 under HL conditions both at three weeks (DEG3, Fig. 2a) and four weeks (DEG7, Fig. 2b) with a notable overlap (Fig. 2c and Supplementary Data 3) between both time points corresponded to those related to the immune response, and particularly the response to salicylic acid (SA)³¹. These DEGs were differentially expressed in tt5-2 by HL compared to DEGs in the tt4-11 tt5-2 double mutant (Supplementary Figs. 7a, b and Supplementary Data 4 and 5), suggesting that they are induced by a pathway metabolic product, and not by the absence of the CHI enzyme or reduced flux through the flavonol/anthocyanin pathway. To determine whether 3-week-old tt5-2 plants under HL conditions had increased amounts of SA and SA glucoside (SAG), we quantified levels by conducting targeted LC-MS/MS on extracts obtained from the samples used for RNA-seq. Our results show that HL had no significant effect in the accumulation of SA or SAG in WT (P > 0.01, two-sided t-test), but significantly reduced SA and SAG accumulation in tt4-11 (P < 0.01, two-sided t-test; Fig. 2d and Supplementary Fig. 7c). Consistent with the RNA-seq results, the levels of both SA and SAG were significantly increased (P < 0.01, two-sided t-test) in tt5-2. SA and SAG levels in the tt4-11 tt5-2 double mutant were comparable to those in tt4-11, providing evidence that this increase in SA and SAG is a consequence of the increased accumulation of NGC or its derivative (Fig. 2d and Supplementary Fig. 7c). Overaccumulation of SA to levels comparable to those present in tt5 under HL were shown to induce cell death and dwarfism^32–34, like the phenotypes observed in the tt5-2 mutants grown under HL (Supplementary Fig. 3). Thus, at this point we could not rule out the possibility that the growth and developmental phenotypes observed in tt5-2 under HL could result from the high SA levels particular to this mutant.

UVR8 mutations suppress the tt5 HL phenotypes

We reasoned that NGC or its derivative interacts with one or more cellular components, resulting in the observed developmental phenotypes. To identify such cellular components, we embarked on identifying suppressors that can prevent the early leaf senescence phenotype of tt5-2 in HL. We mutagenized tt5-2 seeds with ethyl methyl sulfate (EMS) and screened ~ 10,000 M₂ plants for the absence of early leaf senescence by growing plants in HL conditions for five weeks (Supplementary Fig. 8). We identified 15 plants that phenotypically looked like Col-0 yet lacked anthocyanin pigmentation. We predicted that some suppressors would abolish the accumulation of NGC. To establish this, we extracted phenolic compounds from control (Col-0, tt4-11, and tt5-2) and M₂ suppressor lines grown for four weeks in HL, acid-hydrolyzed the extracts to simplify interpretation since NGC and most its derivatives are converted to naringenin after hydrolysis^35–37, and analyzed them by untargeted LC-MS/MS using a commercial standard of naringenin as reference (Supplementary Fig. 9a, b). As expected, tt5-2 accumulates significant levels of the m/z 273.08 peak characteristic of naringenin in hydrolyzed extracts (derived from the conversion of NGC to naringenin during the hydrolysis), which is absent in tt4-11 and very low in Col-0 (Supplementary Fig. 9c). From the 15 suppressor lines, ten lines showed no naringenin accumulation after hydrolysis. From the remaining five lines accumulating naringenin after hydrolysis, three (S1, S20 and S32) showed naringenin levels comparable to tt5-2, while two (S17 and S33) showed naringenin levels higher than Col-0, but lower than tt5-2 (Supplementary Fig. 9c). We back-crossed these five suppressor lines to tt5-2 to perform bulked segregant analysis by whole genome re-sequencing (Supplementary Fig. 8) and determined that the causative mutations were in two distinct chromosomal regions (Supplementary Fig. 10). S17 and S33 showed missense mutations in At5g13930 (TT4) (Supplementary Fig. 11), consistent with these two lines showing reduced naringenin levels after hydrolysis (Supplementary Fig. 9c) and suggesting that the low levels of NGC or its derivative present in these two suppressor lines is insufficient to cause the developmental defects observed in tt5-2 under HL. The other three suppressor lines (S1, S20 and S32) contained mutations in At5g63860 (UVR8, UV RESISTANCE LOCUS8), with the S1 and S20 mutations predicted to cause splicing defects at the 3’ end of introns 4 and 2 respectively, in which the normal splicing acceptor sites were mutated, and S32 harboring replacement of R55 with a stop codon (Fig. 3a). Thus, we named the uvr8 alleles uvr8-31 (S1), uvr8-32 (S20), and uvr8-33 (S32). These results demonstrate that uvr8 mutations can abolish the developmental phenotypes of tt5 mutants grown in HL (Fig. 3b) and suggest that UVR8 is a potential target (direct or indirect) of NGC or its derivatives.

Fig. 3. Identification of mutations in the UV-B receptor UVR8 as tt5-2 suppressors.

a Structure of the UVR8 gene indicating the position and nature of the EMS-induced mutations for the S1, S20 and S32 suppressors. The S1 and S20 mutations cause splicing defects at the 3’ end of introns 4 and 2, respectively. The S32 suppressor harbors a replacement of R55 with a stop codon. b Pictures of tt5-2 plants and the three tt5 suppressors lines, tt5 S1 (tt5-2 uvr8-31), tt5 S20 (tt5-2 uvr8-32), and tt5 S32 (tt5-2 uvr8-33), under high light conditions for 5 weeks. The adaxial and abaxial surfaces of the rosette leaves are observed. Scale bar: 1 cm.

UVR8 is primarily recognized as an UV-A and UV-B receptor⁵, hence we asked whether UV-A/B is necessary for this activity of UVR8. We determined that our initial HL conditions (Supplementary Fig. 2) contained low but significant quantities of UV-A/B (280–400 nm; ~ 50 μW/cm⁻²). Thus, we grew Col-0, tt4-11, tt5-2, and tt4-11 tt5-2 plants for five weeks in identical HL conditions with and without a filter that absorbs all UV-A/B radiation (Supplementary Fig. 12). No difference in the phenotypes of the plants with and without UV-A/B was observed (Supplementary Fig. 12) indicating that the participation of UVR8 in the effects of HL on tt5 are independent of UV-A/B.

Given that uvr8 mutations suppress the early senescence and growth phenotype of tt5 plants grown in HL, we asked whether the elevated SA and SAG levels remained high in the suppressor lines, or whether they returned to normal levels. Thus, we analyzed SA and SAG by LC-MS/MS and determined that the suppressor lines grown in HL had SA and SAG levels comparable to those of tt5 plants (Supplementary Fig. 13). These results indicate that the developmental phenotypes tt5 plants in HL conditions are not a direct consequence of the enhanced accumulation of SA or SAG but rather depend on UVR8 activity.

Increased monomeric UVR8 in tt5 plants

To explore the connections between UVR8 function, NGC or its derivative accumulation in tt5 grown in HL, and the tt5 phenotypes, we collected leaf samples from Col-0, tt4-11, tt5-2, tt4-11 tt5-2, tt5-2 uvr8-31, tt5-2 uvr8-32, and tt5-2 uvr8-33 grown for 3, 4, or 5 weeks in HL and NL without UV-A/B (Supplementary Fig. 14) for UVR8 analysis. UVR8 monomeric and dimeric forms were analyzed using a UVR8 antibody (α-UVR8) that recognizes the UVR8 dimer only if the polyacrylamide gel is treated with UV-B prior to transfer to the nylon membrane¹⁸. This commercial UVR8 antibody recognizes the C-terminal region of the protein (see Methods). Consistent with the molecular analysis of the uvr8-31, uvr8-32, and uvr8-33 mutants (Fig. 4a), no protein was detected in Arabidopsis plants harboring these alleles (lanes 5–7, 12–14 and 19–21, Fig. 4a). We found that the amount of total UVR8 protein did not change between Col-0, tt4-11, tt5-2, tt4-11 tt5-2, regardless of whether plants were grown for 3, 4, or 5 weeks in HL (Fig. 4a, denaturing). When the proteins were separated in non-denaturing conditions and the polyacrylamide gel was treated with UV-B to expose the epitope in the dimer (Non-denaturing, Gel: + UV-B, Fig. 4a), the antibody detected the UVR8 dimer (indicated by D in Fig. 4a), one band likely non-specific as it was present also in the mutants (indicated with open arrow, Fig. 4a), and two bands of increased mobility (indicated by M₁ and M₂, Fig. 4a), specifically present in tt5-2 plants after 4 weeks of HL exposure (line 10, Fig. 4a), and with intensities that further increased at 5 weeks of HL (lines 17, Fig. 4a). Based on our results and the literature, M₁ and M₂ correspond to monomeric UVR8^18,25,38 which can adopt different conformations³⁹. The UVR8 monomer was barely detectable in WT plants grown under HL (lines 8 and 15, Fig. 4a; see Supplementary Fig. 15a for longer exposure), and it was undetectable in WT or tt5-2 plants grown under NL (Fig. 4a and Supplementary Fig. 15a). These results suggest that NGC or its derivative promotes UVR8 monomer formation by destabilizing the dimer, stabilizing the monomer, preventing it from reforming dimers, or both.

Fig. 4. Distribution of dimeric and monomeric UVR8s.

a Total proteins were extracted from Arabidopsis plants grown under high light (HL) or normal light (NL) conditions for 3, 4 or 5 weeks. UVR8 dimers were separated by non-denaturing PAGE, and the gel was irradiated with UV-B (30 min, 300 µW/cm²) before transfer to a nylon membrane (non-denaturing, upper blots in each panel). Denatured samples (denaturing, center blots in each panel) were run by SDS-PAGE and transferred to nylon. Blots were incubated with the antibody specific to Arabidopsis UVR8 (α-UVR8), and the blot corresponding to the gel run in denaturing conditions was incubated with an antibody specific to Arabidopsis actin (α-Actin, lower blots in each panel) for protein loading normalization. b Col-0, tt5-2, and tt4-11 tt5-2 mutant plants grown for 3 weeks under HL (T₀) were treated with UV-B (200 µW/cm², 60 min, T₁) and transferred back to HL conditions for 2 h (T₃), 5 h (T₆), 11 h (T₁₂), 23 h (T₂₄), 47 h (T₄₈), and 71 h (T₇₂). Samples were collected in a time series, and total proteins were extracted to determine the distribution of dimeric and monomeric UVR8s. Upper panel, non-denatured condition with the irradiation of protein gels to detect both dimeric and monomeric UVR8s. Middle panel, denatured samples indicating the amounts of total UVR8s. Actin levels were used as a loading control (lower panel). D indicates the dimeric form of UVR8, M₁ and M₂ correspond to two UVR8 monomer (M) conformers, and the open white arrow marks the position of a non-specific protein recognized by α-UVR8. The experiments were independently carried out twice with similar results. Source data are provided as a Source Data file.

UV-B rapidly and reversibly induces the formation of UVR8 monomers in vivo^18,21. To gain a better understanding of the mechanisms associated with the increase of UVR8 monomer in tt5 mutants, we investigated the response to UV-B of UVR8 in tt5-2, compared to Col-0 and the tt4-11 tt5-2 double mutant. We treated plants (HL grown without UV-B, T₀) with UV-B (200 µW/cm²) for 60 min (T₁) and then we transferred plants back to HL conditions for 2 h (T₃), 5 h (T₆), 11 h (T₁₂), 23 h (T₂₄), 47 hours (T₄₈), and 71 h (T₇₂) (Fig. 4b). We extracted protein and analyzed by Western blot the accumulation of monomeric and dimeric UVR8 as described above. The UV-B treatment slightly induced total UVR8 levels in all lines (compare denaturing conditions signals with α-UVR8 and α-Actin, Fig. 4b), but most prominently induced the formation of UVR8 monomers (M₁ and M₂) in Col-0 (lane 2, Fig. 4b), tt5-2 (lane 10, Fig. 4b) and tt4-11 tt5-2 (lane 18, Fig. 4b). Upon transfer back to HL conditions, the monomer rapidly disappeared in Col-0 and tt4-11 tt5-2, but although reduced, did not completely vanish in tt5-2 (compare lanes 3, 11 and 19, Fig. 4b). However, the levels of UVR8 monomer dramatically increased 11 h after recovery from UV-B (lane 13, Fig. 4b), but only in tt5-2, and continued high for 72 h, the duration of the experiment (lane 16, Fig. 4b). Noticeable was that, while UV-B significantly induced the formation of monomer M₂ in both Col-0 and tt5-2 (lanes 2 and 10, Fig. 4b), the appearance of M₁ at longer times of recovery in HL was unique to tt5-2 (lanes 14–16, Fig. 4b). These results suggest that UV-B and NGC or its derivative promote formation of different UVR8 monomer conformers.

Compared to other Arabidopsis flavonoid mutants, tt5 is hypersensitive to UV-B⁴⁰. This increased sensitivity of tt5 to UV-B is evident, for example, in the appearance of a shiny leaf surface as early as T₁₂ and the development of more severe phenotypes including pale and curled leaves at later times after UV-B, phenotypes absent in the double tt4-11 tt5-2 mutant (Supplementary Fig. 16). Based on our results, it is possible that NGC or its derivative directly or indirectly increases the levels of monomeric UVR8, sustaining a response to UV-B even after UV-B is gone, and that this is responsible for the observed UV-B sensitivity of tt5.

Accumulation of NGC derivatives in tt5 mutants in HL

To understand the accumulation of NGC or its derivative in tt5 under HL potentially responsible for the increase in monomeric UVR8, we analyzed the plant samples described in the previous section (Supplementary Fig. 14), by performing untargeted metabolic profiling of methanol extracts by LC-MS/MS, focusing on compounds exhibiting a fragmentation product of m/z 273.08, characteristic of NGC or its derivatives. In extracts of tt5-2 mutant with 4 weeks in HL, four major peaks absent in Col-0 and corresponding to NGC derivatives were identified (Fig. 5). Remarkably, NGC itself was not found at significant levels in any of the samples, indicating that the naringenin identified in hydrolyzed samples (Supplementary Fig. 9) corresponded most likely to NGC derived from O-linked derivative(s) easily hydrolysable in hot acid. Subsequent tandem MS analysis of these four peaks indicated that Peak 1 (m/z 614.21; retention time 4.04 min) is consistent with a double glycosylation of NGC forming an NH₄ adduct during ionization (Hexose-NGC-Hexose_1, 273.08 + 162.05 + 162.05 + 17.03 = 614.21); Peak 2 (m/z 597.18; retention time 4.51 min) agrees with another NGC-diglycoside (Hexose-NGC-Hexose_2, 273.08 + 162.05 + 162.05 = 597.18); Peak 3 (m/z 435.13; retention time 5.02 min) is likely a single glycoside (hexose) linked to NGC (NGC-Hexose, 273.08 + 162.05 = 435.13); and Peak 4 (m/z 521.13; retention time 5.28 min) corresponds most likely to NGC conjugated to a glycoside and a malonyl group (NGC-Hexose-Mal, 273.08 + 162.05 + 86.01 = 521.13) (Fig. 5).

Fig. 5. Identification of naringenin chalcone (NGC) derivatives in tt5-2 plants.

NGC derivatives in tt5-2 plants. Four major peaks, resulting in the m/z 273.08 daughter ion from LC-MS/MS, were detected in the tt5-2 mutant (black line) under high light (HL) conditions for 4 weeks, but not in Col-0 (red line). Tandem MS profiles of these peaks were annotated as follows: Peak 1 (m/z 614.21, Hex-NGC-Hex_1), Peak 2 (m/z 597.18, Hex-NGC-Hex_2), Peak 3 (m/z 435.13, NGC-Hex), and Peak 4 (m/z 521.13, NGC-Hex-Mal). Hex, hexose group; Mal, malonyl moiety. The experiments were independently repeated twice with similar results, each time using different sets of plants. Source data are provided as a Source Data file.

Using the diagnostic m/z peaks, we determined that all the NGC derivatives accumulate at very low levels, if at all, in tt5-2 under NL (Supplementary Fig. 17a). When grown under HL conditions, Peaks 1 and 2 continued to increase during development in the tt5-2 mutant and in the tt5 uvr8 suppressor lines, while Peaks 3 and 4 accumulated at a maximum after four weeks of HL exposure, to then decrease at five weeks in the tt5-2 and tt5 uvr8 suppressor lines under HL. These compounds were present at low levels in Col-0 plants grown in HL and were absent in tt4-11 or tt4-11 tt5-2 plants (Supplementary Fig. 17b). The absence of NGC in all the samples indicates that all the NGC that accumulates in tt5-2 in HL gets rapidly glycosylated and/or malonylated.

We then analyzed by LC-MS/MS the accumulation of NGC derivatives in plants treated with UV-B to determine whether a particular compound was associated with the sustained levels of UVR8 monomer observed in tt5-2 following recovery from UV-B treatment (Fig. 4b). All four NGC derivatives are present in tt5-2 plants grown in HL for three weeks (T₀, Supplementary Fig. 18) with UV-B significantly enhancing the accumulation of all four compounds at ~ 5 h after treatment but only in tt5-2 (T₆, Supplementary Fig. 18). This is consistent with previous studies demonstrating the induction of Arabidopsis flavonoid biosynthesis by UV light after 7 h of treatment⁴¹. At subsequent recovery times, the levels of the four NGC derivatives returned to levels comparable to those present before the UV-B treatment (Supplementary Fig. 18, compare T₀ and T₁₂).

These results show that UV-B induces UVR8 monomer formation irrespective of whether NGC derivatives are present or not, although the monomer levels in the flavonoid mutants might be higher than in Col-0 because of the absence of UV-protecting flavonoid sunscreen in tt5-2 and tt4-11 tt5-2 (compare lanes 2, 10 and 18 in Fig. 4b). Once the UV-B treatment is finished, UVR8 rapidly reverts to the dimeric form in Col-0 and tt4-11 tt5-2 (lanes 3 and 19, Fig. 4b), but the presence of NGC derivatives in tt5-2 sustains a low but significant amount of UVR8 monomer (lanes 11 and 12, Fig. 4b). The striking increase of monomeric UVR8 in tt5-2 about 11 h after the UV-B treatment (Lanes 13–16, Fig. 4b) could be a consequence of the UV-B induced increase in NGC derivatives in tt5-2, which can be observed at around 5 h following the UV-B treatment (T₆, Supplementary Fig. 18). It is interesting that all four NGC derivatives return to the levels found prior to the UV-B treatment by 11 h of recovery (T₁₂, Supplementary Fig. 18), while the amount of monomeric UVR8 continues to increase (M, lanes 1–16, Fig. 4b).

To assess the significance of our findings under more natural Arabidopsis growth conditions and their physiological roles in wild-type plants, we cultivated Col-0 and mutant lines (tt4-11, tt5-2, tt4-11 tt5-2, tt5-2 uvr8-31, PAP1-D, tt6-2, and tt19-8) outdoors in a courtyard under natural Hawai’i September-October sunlight and temperature conditions, with and without a UV-A/B filter to block natural UV-A/B, for five weeks (Supplementary Fig. 19). PAP1-D corresponds to a line in which the PAP1 anthocyanin regulator is constitutively expressed derived from an activation tagging screen⁴². After five weeks outside, total proteins were extracted from the plants and analyzed via Western blot to examine the distribution of monomeric and dimeric UVR8 (Supplementary Fig. 20a). Our analysis showed that the total UVR8 protein levels were comparable across Col-0, tt4-11, tt5-2, tt4-11 tt5-2, PAP1-D, tt6-2, and tt19-8, regardless of whether the plants were grown with or without UV-A/B filtration (Supplementary Fig. 20a, α-UVR8 Denaturing). Under non-denaturing conditions, when the polyacrylamide gels were exposed to UV-B, the UVR8 dimer levels were detected and showed no significant differences between the samples (Supplementary Fig. 20a, α-UVR8 Non-denaturing, +UV-B). Monomeric UVR8 levels were evaluated under non-denaturing conditions without UV-B exposure of the polyacrylamide gel, where UVR8 dimers are barely detectable (Supplementary Fig. 20a, α-UVR8 Non-denaturing, -UV-B). Comparing all samples, we observed accumulation of UVR8 monomer in Col-0 at moderate levels and tt5-2 at the highest levels, regardless of whether UV-A/B radiation was eliminated or not during the growth of the plants (Supplementary Fig. 20a, α-UVR8 Non-denaturing, -UV-B, lanes 1, 3, 9, and 11).

To determine the impact of sunlight conditions on the levels of NGC and derivatives, we analyzed by LC-MS/MS the accumulation of NGC derivatives in Col-0, tt4-11 tt5-2, tt5-2, and tt5-2 uvr8-31 plants grown with and without a UV-A/B filter for 5 weeks (Supplementary Fig. 20b). The results showed that Peak 1 (Hex-NGC-Hex), Peak 3 (NGC-Hex), and Peak 4 (NGC-Hex-Mal) accumulated in Col-0 plants at meaningful levels irrespective of whether UV-A/B was present or not, were significantly (~ 10-fold) higher in tt5-2 with a modest positive impact of UV-A/B, and absent in tt4-11 tt5-2 (Supplementary Fig. 20b). Notably, the levels of Peaks 1 – 3 were indistinguishable in tt5-2 and tt5-2 uvr8-31 in the presence of natural UV-A/B radiation (i.e., -UV-A/B filter, Supplementary Fig. 20b) suggesting that UVR8 under sunlight conditions has a modest effect on flavonoid accumulation. Overall, our findings indicate that while NGC derivatives are undetectable in Col-0 growth chamber conditions, even in HL, their accumulation is significant when plants grow outdoors, regardless of UV-A/B exposure, and correlates with increased levels of monomeric UVR8.

NGC and derivatives increase UVR8 monomer formation in vitro

Our results so far indicate that the presence of NGC derivatives correlates with the increase of UVR8 monomer accumulation. To uncover the mechanism driving this correlation, we incubated protein extracts obtained from Col-0 plants grown under HL for 4 weeks with methanol extracts of tt5-2 plants grown under HL for 4 weeks (lanes 1 and 2, Supplementary Fig. 21a), pure commercial NGC (lanes 3 and 4, Supplementary Fig. 21a), acid hydrolyzed methanol extracts of tt5-2 plants grown under HL for 4 weeks (lanes 5 and 6, Supplementary Fig. 21a) and compared with the intensity of the UVR8 monomer band in the absence of any exogenous small molecule (lane 7, Supplementary Fig. 21a). Because we are focusing specifically on the UVR8 monomer (M), we did not treat the gel with UV prior to transfer, hence the UVR8 dimer (D) band is barely visible under non-denaturing conditions, but the total amount of UVR8 is comparable between all the samples (α-UVR8 Denaturing, Supplementary Fig. 21a). The methanol extracts visibly increased the amount of UVR8 monomer, an increase that could be observed also with pure NGC, but not with hydrolyzed extracts (keep in mind that acid hydrolysis converts NGC into naringenin). To quantify this effect, we repeated this experiment with three different protein and methanol extracts (Supplementary Fig. 21b), and determined that NGC or methanol extracts, but not hydrolyzed extracts, increased the level of monomeric UVR8 by about 2-fold (Supplementary Fig. 21b). This is only a fraction of the UVR8 capable of forming monomer, evidenced by the finding that a 60 min brief treatment with UV-B (300 µW/cm²) of the protein extracts increased the amount of monomer by about 7.5-fold (lane 9 in Supplementary Fig. 21a; arrow in Supplementary Fig. 21c). Taken together, our results demonstrate that NGC (and NGC derivatives) enhance UVR8 monomer accumulation.

NGC physically interacts with UVR8 monomer

To determine whether the correlation between NGC and UVR8 monomer results from a direct interaction rather than an indirect effect, for example requiring other cellular factors, we purified recombinant wild-type UVR8, UVR8^D96N/D107N (a constitutive monomer⁴³), and UVR8^W285F (a constitutive dimer⁴⁴) and validated their in vitro UV-B response properties by size exclusion chromatography (SEC). As expected, wild-type UVR8 dimers dissociated into monomers upon UV-B treatment, while UVR8^D96N/D107N remained monomeric and UVR8^W285F remained dimeric, regardless of UV-B exposure (Supplementary Fig. 22). These functionally validated UVR8 variants were then used for protein-ligand binding assays. In nano differential scanning fluorimetry (nanoDSF) experiments, NGC and naringenin were incubated with various UVR8 forms: wild-type UVR8 (dimer), wild-type UVR8 + UV-B (UVR8 monomer), UVR8^D96N/D107N (UVR8 monomer), UVR8^D96N/D107N + UV-B (UVR8 monomer), UVR8^W285F (UVR8 dimer), and UVR8^W285F + UV-B (UVR8 dimer). The thermal denaturation temperature (T_m) was measured to assess protein conformational stability in each condition. The results showed no significant differences in T_m among the three dimeric forms (wild-type UVR8, UVR8^W285F, and UVR8^W285F + UV-B) when incubated with DMSO (control), NGC, or naringenin (Supplementary Fig. 23 and Supplementary Table 2). However, a significant increase in T_m was observed when NGC was incubated with the three monomeric forms (wild-type UVR8 + UV-B, UVR8^D96N/D107N, and UVR8^D96N/D107N + UV-B) compared to incubation with DMSO or naringenin (Supplementary Fig. 23 and Supplementary Table 2). These findings indicate that NGC, but not naringenin, physically interacts with monomeric UVR8 but not with dimeric UVR8. To further investigate the specificity of UVR8 monomer interactions with small molecules, we tested a collection of flavonoids, including NGC, naringenin, kaempferol, dihydrokaempferol, dihydroquercetin, leucocyanidin, and cyanidin. Among these, only NGC incubation resulted in a significantly increased T_m for monomeric UVR8 (Supplementary Fig. 24 and Supplementary Table 3). These results demonstrate that the UVR8 monomer specifically interacts with NGC, but not with other flavonoids or flavonoid pathway intermediates.

To further characterize the interaction, we conducted microscale thermophoresis (MST) experiments to produce binding curves by incubating UVR8 in its dimeric or monomeric form with NGC or naringenin (Supplementary Fig. 25). These binding curves were then used to calculate the apparent equilibrium dissociation constant (K_d). The results showed no binding for dimeric UVR8 incubated with either NGC or naringenin. In contrast, a calculated K_d value (in triplicate) for NGC binding to UVR8 monomer was estimated to be ~ 12 ± 9.2 μM, demonstrating a strong interaction. As anticipated from our previous results, naringenin exhibited a much weaker interaction with UVR8 monomer (apparent K_d > 1.6 mM). These findings demonstrate that NGC physically interacts with UVR8 in its monomeric form with high affinity. Unfortunately, we have not yet succeeded in purifying the NGC derivatives that are found in vivo in quantities large enough to test their affinity for UVR8 monomers.

To predict the potential binding site(s) of NGC on the UVR8 monomer, we utilized Protenix⁴⁵, a simulation tool that models protein-ligand interactions through a comprehensive AlphaFold3 reproduction. We tested different molar ratios of NGC to UVR8 monomer (1:1, 2:1, 3:1, and 4:1) and found that the 1:1 ratio produced the highest interface predicted template modeling (ipTM) score (the only score exceeding 0.9, indicating a confident, high-quality prediction; Supplementary Table 4). These results suggest that one NGC molecule binding to one UVR8 monomer is the most probable interaction model. We could not recover a UVR8-naringenin model in which the solved structure of UVR8 was preserved^26,46, consistent with the experimental results that show no naringenin high-affinity binding to monomeric UVR8 (Supplementary Fig. 25c, d). To visualize the potential NGC binding site on the UVR8 monomer, we analyzed the highest-scoring model using PyMOL⁴⁷, which revealed four candidate amino acids in UVR8 that may interact with NGC: G283, C335, W337, and R439 (Supplementary Fig. 26). The variation in free energy (ΔG) of the binding of NGC to the UVR8 monomer was estimated by PyFepRestr⁴⁸ at ~ 30 kJ/mol. These analyses predict a specific and high-affinity interaction between NGC and the UVR8 monomer and identify the potential key UVR8 residues involved.

UVR8 may contribute to processes beyond UV-B signaling

Our results suggest that NGC (and its derivatives) modulate a UVR8 function that does not require UV-A/B. To gain insight into this function, we analyzed mRNA accumulation changes in plants with the null uvr8-6 allele grown for 4 weeks under NL and HL conditions with a UV-A/B filter (Supplementary Fig. 6b), comparing them to Col-0. DEG analyses between Col-0 and uvr8-6 under HL (with UV-A/B filter) identified 43 genes higher in Col-0 (potentially positively regulated by UVR8 independently of UV-B) and 54 genes higher in uvr8-6 (potentially negatively regulated by UVR8 independently of UV-B, Fig. 6 and Supplementary Data 6). These results were further compared to those from tt5-2, tt4-11, and tt4-11 tt5-2 mutants grown under the same conditions (Fig. 2 and Supplementary Fig. 7). We found that more than half of the UVR8-dependent UV-B independent DEGs were also unique to the response of tt5 to four weeks of HL (DEG7, Fig. 2b), yet mRNAs that accumulated lower in uvr8-6 (presumably under positive UVB8 regulation) accumulated higher in tt5-2 (Fig. 6). The opposite trend is observed for genes that are higher in uvr8-6 (i.e., they are lower in tt5 grown for four weeks in HL), but the effect is less marked (Supplementary Data 7). An analysis of the potential function of these genes revealed that the vast majority are associated with plant defense or immunity responses, including genes encoding pathogen-associated molecular patterns, receptor-like proteins, chitinases, and disease resistance proteins (highlighted with asterisks in Fig. 6). These results reveal that NGC and its derivatives modulate UVR8 functions beyond UV-B signaling, influencing stress responses and immunity. Moreover, they uncover unrecognized roles for UVR8, broadening our understanding of its versatility in plant biology.

Fig. 6. Identification of UVR8-dependent genes positively affected by the accumulation of naringenin chalcone (NGC) derivatives.

Differentially expressed gene (DEG) analysis of Col-0 HL versus uvr8-6 HL and tt5-2 HL versus tt5-2 NL were conducted on RNA-seq data with resulting in the identification of 97 and 5251 DEGs, respectively. The Venn diagram illustrates 55 shared genes between DEG9 and DEG7. The gene IDs, fold change, and description for the 27 genes upregulated in tt5-2 HL are indicated. Color gradient indicates the log₂Fold Change. The asterisks indicate 10 genes associated with plant defense or immune responses. Source data are provided as a Source Data file.

Discussion

Mutations of flavonoid biosynthesis genes have not conventionally been associated with pleiotropic phenotypic effects, like those observed here for tt5 under HL conditions (Supplementary Fig. 3). The finding that the pathway intermediate NGC and likely NGC derivatives that accumulate in WT plants under normal solar light, physically interact with the UVR8 photoreceptor increasing the monomeric form of UVR8, even in the absence of detectable UV-A/B radiation, provides an unexpected link between light-induced metabolites and photoreceptor-dependent developmental programs. Indeed, many of the developmental effects observed for tt5 grown in HL resemble moderate manifestations of the phenotypes described for plants overexpressing a constitutively monomeric UVR8 protein²⁶. While we initially considered the possibility that the developmental phenotypes of tt5 under HL resulting from the accumulation of NGC derivatives resemble a constitutive response to UV-B (even in the absence of UV-B), our results indicate that this is not the case. When we compared the UV-B responsive UVR8-dependent genes from the literature⁴⁹ with those uniquely accumulating in tt5 under HL (Fig. 2), we found no statistically significant overlap (Supplementary Fig. 27). However, when we investigated UVR8 regulated genes under HL conditions in the absence of UV-B, we found that they significantly overlap with those uniquely induced in tt5 under HL (Fig. 6). In fact, genes expressed uniquely higher in tt5-2 under HL show lower mRNA accumulation in the uvr8-6 mutant under HL (Fig. 6), indicating that their expression is dependent on active UVR8. The enrichment of these genes in functions related to stress responses and immunity was unexpected and indicates a UVR8 function that goes beyond UV-B perception and integration of light signals. In this regard, it is noteworthy the induction of SA and SAG in tt5 plants under HL conditions, with the concomitant changes in gene expression (Fig. 2d and Supplementary Fig. 13a). Indeed, about a third (642/2111) of the genes induced in tt5-2 under HL (DEG7, Fig. 2b) correspond to genes described as induced by one hour of SA treatment³¹ indicating that most likely they result from the increased SA accumulation in tt5-2 under HL. This increased SA accumulation under HL may provide tt5 enhanced resistance to plant pathogens⁵⁰, providing an additional piece to the puzzle of how light influences plant-pathogen interactions⁵¹. Our results, however, show that this increase in SA and SAG is independent of UVR8 (Supplementary Fig. 13b), indicating that the induction by light of NGC and derivatives targets the pathogen response/immunity axis by UVR8-dependent and UVR8-independent mechanisms. Given that the induction of flavonoids by UV light is conserved across the plant kingdom¹⁷, our results may help explain the elusive mechanisms by which UV-A/B light contributes to reducing disease and pest incidence in many horticultural settings⁵². Constitutively high SA accumulation results in stunted Arabidopsis growth^34,53,54, but the lack of similar defects in the tt5 uvr8 double mutant suppressor lines (Supplementary Fig. 12) indicates that either the growth defect of tt5 in HL is independent of high SA accumulation, or that UVR8 function is somehow required for the SA hyperaccumulation phenotypes. A mutation in the CDK8 component of the Mediator complex similarly offset SA hyperaccumulation phenotypes⁵⁵, perhaps implying a relationship between the Mediator complex and UVR8.

Under normal growth chamber conditions, most UVR8 is dimeric (believed to correspond to the cytoplasmic ground state) and upon UV-B irradiation, a significant fraction of UVR8 can be found in the monomeric form (believed to correspond to the nuclear signaling state) where it interacts with COP1 and various regulatory proteins^{18,23–25,56,57} (Fig. 7a). Consistent with a role of NGC (and derivatives) in stabilizing monomeric UVR8, the observed phenotypes of tt5 under HL conditions resemble to some extent those of cop1 mutants^58,59.

Fig. 7. Proposed mechanism for stabilization of monomeric UVR8 by naringenin chalcone (NGC) derivatives and its effects on UVR8-mediated gene regulation.

a Under growth chamber light conditions, UVR8 predominantly exists in the cytosol as an inactive homodimer. A minor fraction of UVR8 is present in the nucleus in its monomeric form, where it interacts with COP1 (E3 ubiquitin ligase) to inhibit COP1-mediated degradation of HY5. Defense-related genes I are activated by endogenous SA. b Under natural sunlight conditions, accumulated NGC derivatives elevate SA levels and directly bind to UVR8 monomers. This SA accumulation leads to the activation of defense-related genes II. The formation of monomeric UVR8-NGC derivative complex promotes the activation of a subset of UVR8-regulated genes that don’t require UV-B signaling. COP1, CONSTITUTIVELY PHOTOMORPHOGENIC1; HY5, ELONGATED HYPOCOTYL5; NGC, naringenin chalcone; SA, salicylic acid; UVR8, UV RESISTANCE LOCUS8.

The importance of UVR8 under growth chamber conditions likely lies in its long-term effects, such as mitigating UV-B induced DNA damage^19,60. Solar light clearly induces monomeric UVR8, an induction that is observed even when UV-A/B light is filtered (Supplementary Fig. 20). Our results suggest that this is a consequence of the increase in NGC derivatives that physically interact with monomeric UVR8 (Supplementary Figs. 23–26) and presumably results in higher levels of nuclear stabilized monomeric UVR8 (Fig. 7b). How are such UVR8 monomers formed in the absence of UV radiation? Putting aside the possibility that there is some UV fluence that is below detection levels (yet could very slowly contribute to UVR8 monomer formation, as is the case in the experiments here described, Supplementary Fig. 14), it is important to remember that even though the dissociation constant for UVR8 dimer (without UV) is low²⁰ (yet has not been measured), monomer and dimer forms are in equilibrium, which would be displaced towards monomer formation when the monomer is stabilized by NGC (or derivatives). Because the accumulation of NGC derivatives in tt5 grown in HL does not resemble the gene expression changes associated with UV-B exposure⁶¹, we argue that the interaction of UVR8 with NGC derivatives is important for a subset of the UVR8 functions that don’t require UV-A/B radiation, directly implicating UVR8 with stress responses and immunity (Fig. 7b). The interaction of UVR8 monomers with NGC and presumably one or more of the NGC derivatives identified in this study (Fig. 5) is both robust (low micromolar range K_d, Supplementary Fig. 25) and specific, at least within the flavonoid pathway intermediates available (Supplementary Fig. 24). The possibility that NGC interacts in the UVR8 monomer with some of the same residues that are involved in dimerization^62,63 (Supplementary Fig. 26) is consistent with the model that it stabilizes the UVR8 monomer (Fig. 7b). This discovery potentially identifies an interesting interaction pocket for small molecules, presenting exciting opportunities to engineer UVR8 mutants with tailored responsiveness to NGC and its derivatives. Our findings also highlight the need for caution when interpreting the phenotypes of mutants in metabolic pathways. Accumulation of pathway intermediates at significantly elevated levels in these mutants may exhibit unforeseen biological activities, potentially confounding phenotype analysis and revealing unexpected roles for these intermediates. Taken together, our results identify developmental functions for UVR8 that extend beyond just UV-A/B sensing and can be modulated by flavonoid pathway intermediates, which themselves are to some extent controlled by UVR8.

Methods

Chemicals and instruments

Chemical standards naringenin chalcone (HY-N3007), naringenin (HY-N0100), kaempferol (HY-14590), dihydrokaempferol (HY-N2897), leucocyanidin (HY-119580), and dihydroquercetin (HY-N0136) were purchased from MedChemExpress (Monmouth Junction, NJ). Cyanidin (chloride) No. 14803 was purchased from Cayman Chemical (Ann Arbor, MI). Salicylic acid (Product No. 247588), MWGF1000 Gel Filtration Markers Kit, and HPLC-grade solvents were obtained from MilliporeSigma (Burlington, MA). Salicylic acid glucoside (Catalog No. sc-473013) and [¹³C₆]-salicylic acid (Catalog No. sc-220088) were sourced from Santa Cruz Biotechnology (Dallas, TX). Antibodies to Arabidopsis thaliana UVR8 (PHY0809A), Actin (AT3G12110, PHY0001), goat anti-rabbit IgG H&L (HRP, PHY6000), and goat anti-mouse IgG H&L (HRP, PHY6006) were acquired from PhytoAB (San Jose, CA). The anti-plant actin monoclonal antibody (A17309) was purchased from Antibodies.com LLC (St. Louis, MO). The IRDye^® 680RD goat anti-mouse IgG secondary antibody was obtained from LI-COR Biotechnology (Lincoln, NE). Platinum Taq DNA Polymerase High Fidelity and pCR Blunt II TOPO vector were purchased from ThermoFisher Scientific (Waltham, MA). BamHI, HindIII, and T4 DNA ligase were purchased from Bio-Rad Laboratories (Hercules, CA). Ni-NTA resin for His-tag purification was purchased from QIAGEN (Hilden, Germany).

Arabidopsis plant pictures were captured using a Canon PowerShot G12 camera. Light spectra were measured with a LI-180 Spectrometer from LI-COR Biosciences (Lincoln, NE). Detection of sinapoylmalate was carried out using Portable UVP 3UV Lamps from Analytik Jena (Upland, CA). To block UV-B light, an Acrylic UV Filter Plexiglass Sheet OP3/UF-5 from ePlastics (San Diego, CA) was employed. The amount of UV light was quantified using a Digital UVA/UVB Meter #UV513AB (280–400 nm) from General Tools (New York City, NY). Digital Ultraviolet Radiometer Model 6.2 UVB from Solar Light Company (Glenside, PA) was used to measure UV-B intensity. An additional source of UV-B was provided by Philips UVB Broadband TL 40 W/12 RS SLV/25 tubes. Western blot results were processed using the Mini-Med 90 X-Ray film processor from AFP manufacturing (Peachtree City, GA) with GeneMate blue basic autorad film, the iBright^TM CL1500 Imaging System from Invitrogen (Waltham, MA), or the Sapphire Biomolecular Imager from Azure Biosystems (Dublin, CA). Induced proteins were solubilized using a Misonix S-4000 sonicator ultrasonic liquid processor (Newtown, CT). Size-exclusion chromatography (SEC) was conducted on a Cytiva ÄKTA GO protein purification system from Danaher Corporation (Washington, D.C.). NanoDSF and MST were performed on Monolith NT.Automated instrument from NanoTemper Technologies (San Francisco, CA). Chemical structures were created using ChemDraw 19.0, and graphs were generated using Prism 10.1.1.

Plant materials and growth conditions

Arabidopsis thaliana (L.) accession Columbia-0 (Col-0), along with tt4-11 (SALK_020583), tt5-2 (GK_176H03), tt5-4 (SAIL_641_E10)⁶⁴, tt8-6 (SALK_033468), PAP1-D (CS3884), tt6-2 (GK_292E08), and tt19-8 (SALK_105779) lines, were obtained from the Arabidopsis Biological Resource Center (Columbus, OH). All plants were cultivated in Suremix growth medium (Michigan Grower Products Inc, MI) under controlled conditions (22 °C and a 16 h light/8 h dark photoperiod) unless otherwise specified. For experiments involving HL or natural Sunlight, seedlings were grown for 7 days in MS media and then transferred to soil under the appropriate light conditions. The combination of light bulbs in High R + FR chamber consists of four fluorescent tubes (S-19983, 48” T8 Daylight, SYLVANIA^®, Wilmington, MA) and two incandescent bulbs (S3810, A15 40 Watt-Clear-Medium Base, Satco, Brentwood, NY); whereas, the Low R + FR chamber has five fluorescent tubes but no incandescent bulbs. High-light conditions were obtained by the action of eight fluorescent tubes and four incandescent bulbs.

Constructs, plasmid generation, and plant transformation

To generate the TT5pro::TT5-mCherry complemented lines, three constructs containing the TT5 promoter region (161 nt upstream of TSS), the TT5 CDS (741 nt), and mCherry CDS (711 nt) were synthesized by GeneUniversal (Dewark, DE). These constructs were integrated into the pGWB1 binary vector, resulting in the TT5pro::TT5-mCherry construct through a MultiSite Gateway LR reaction. For the Am4’CGT overexpression lines, the 4’CGT CDS from Antirrhinum majus (AB198665.1, 1374 bp) were synthesized by GeneUniversal and integrated into the pGWB502 binary vector. This resulted in the 35S::Am4’CGT construct through a Gateway LR reaction. The resulting TT5pro::TT5-mCherry and 35S::Am4’CGT constructs were introduced into Agrobacterium tumefaciens (GV3101) and used for Arabidopsis plant transformation via the floral dip method⁶⁵. Transformed lines were selected by culturing surface-sterilized seeds on Murashige and Skoog (MS) solid media with 5% sucrose and hygromycin (50 mg/L). Primers for confirming T-DNA insertions (listed in Supplementary Table 5) were synthesized by Integrated DNA Technologies (Skokie, Il).

For His-SUMO-UVR8/UVR8^D96N/D107N/UVR8^W285F purification, the wild-type UVR8 CDS (1323 nt) with restriction enzyme recognition sites (BamHI and HindIII) was amplified from the U22080 construct that obtained from the ABRC and then subcloned into the pCR Blunt II TOPO vector. The constitutive monomer UVR8^D96N/D107N and constitutive dimer UVR8^W285F constructs were synthesized by GeneUniversal with flanking BamHI and HindIII sites into the pUC57 vector. Restriction enzyme-released UVR8 fragments and linearized vector (modified pET28b vector with 6xHis-SUMO tag at N-terminus) were ligated and then transformed into E. coli (BL21) for subsequent protein induction and purification.

Salicylic acid and salicylic acid glucoside quantification

Arabidopsis tissues were ground in liquid nitrogen (N₂) and lyophilized before storage. In brief, 5 mg of dried samples were soaked in 100 μL of extraction buffer (80:20 v/v methanol:water, 0.1% formic acid, and 100 nM of [¹³C₆]-SA as an internal standard) at 4 °C for 16 h. The sample was then diluted with an equal volume (100 μL) of water containing 0.1% formic acid. After centrifugation at 12,000 x g for 10 min, the supernatant was analyzed by LC-MS/MS. The amounts of SA and SAG in each sample were detected by multiple reaction monitoring (MRM) and quantified by comparing with SA and SAG standards. For creating standard curves of SA and SAG, the SA and SAG standards were dissolved in 80% of methanol with 0.1% formic acid. The UHPLC-MS/MS run lasted 5 min with a flow rate of 0.4 ml/min at 40 °C. The gradient was as follows with solvent A (0.1% formic acid in water) and solvent B (methanol): 0-0.5 min, 98% A and 2% B; 0.5-3.0 min, 30% A and 70% B; 3.0-4.0 min, 100% B; 4.0-5.0 min, 98% A and 2% B.

Untargeted metabolic profiling and LC-MS/MS data analysis

For acid hydrolysis, 10 mg of dry weight tissue was resuspended in 100 μL of 2 M HCl containing 1 μM telmisartan (used as an internal standard) and incubated at 100 °C for 20 min. Aglycone flavonoids were then extracted in 50 μL isoamyl alcohol. To extract total metabolites, lyophilized Arabidopsis tissue was extracted in methanol/water (1:1, v/v) containing 0.1% formic acid and 1 μM telmisartan at a ratio of 1 mg dry tissue in 20 μL extraction buffer overnight at room temperature. After centrifugation at 10,000 x g for 5 min, the supernatant was diluted ten times with 80% of methanol containing 0.1% formic acid for LC-MS/MS analysis.

LC-MS/MS analyses were conducted using an Acquity UHPLC system with a HSS T3 C18 column (100 × 2.1 mm, 1.8 μM, Waters, Milford, MA) connected to the mass spectrometer (Xevo G2-XS QTOF, Waters). Runs were performed at a flow rate of 0.3 mL/min and column temperature of 40 °C. The run involved a 10 min gradient with solvent A (0.1% formic acid in water) and solvent B (acetonitrile): 0-0.5 min, 100% A; 0.5-6 min, 50% A and 50% B; 6.01-6.5 min, 1% A and 99% B; 6.51-8.5 min, 1% A and 99% B; and 8.51-10 min 99% A and 1% B. Mass spectra acquisition was in positive mode with an m/z scan range of 50– 2000. Absorbance was recorded by a diode array detector (DAD) with a range of 200–800 nm. All MS raw data files (.raw) were processed using Waters Progenesis QI v3.0. NGC (m/z 273.08_6.31 min) and naringenin (m/z 273.08_6.43 min) were confirmed by tandem MS profiles, and MS intensity was recorded by Progenesis. The identification of NGC derivatives was achieved by analyzing the mass spectra using Waters MassLynx v4.2 SCN966. The MS raw data files were scanned by the mass chromatogram function with description m/z 273.08.

EMS mutagenesis and whole genome re-sequencing to identify tt5 mutant suppressors

EMS-treated tt5-2 seeds (2.0 g, ~ 100,000) were planted on soil, grown to maturity, and collected into ~ 100 pools of M₁ seeds. Recessive suppressors that restored the leaf early senescence phenotype were isolated from 4-week-old M₂ plants under high light conditions. Genomic DNA from pooled leaves of ~ 50 Arabidopsis BC₁F₃ tt5 suppressor plants and ~ 50 tt5-like plants were extracted using the Wizard Genomic DNA Purification Kit (Promega, WI). The pooled gDNA samples were then submitted to BGI genomic services (http://www.genomics.cn/en/index) for 100 bp paired-end Illumina sequencing. Raw data quality was assessed using FastQC (v0.11.5)⁶⁶. Reads were aligned to the Arabidopsis reference genome (TAIR10) using Bowtie2 (v2.3.4)⁶⁷. In total, ~ 236 million paired-end reads were obtained. The alignment files were sorted and converted to BAM files⁶⁸. SNP calling was performed using the Single Sample Variants Detector module of NGSEP⁶⁹, and the conversion into a VCF matrix was carried out using bcftools (v1.2)⁶⁸. Variants were parsed using vcftools (v0.1.15)⁶⁸ to include only high-quality single SNPs. High-quality SNPs were defined as those with a minimum read depth of 5, a maximum read depth of 200, and a minimum mapping quality greater than 20 (--minDP 5; --MaxDP 200; --MinMQ 20; --remove-indels). SNP frequency was then calculated and used to filter each of the suppressor lines. The filtering criteria were based on common SNPs in each suppressor line, with an allele frequency greater than 0.8 in the samples from tt5 suppressor plants and less than 0.4 in the corresponding samples from tt5-like plants.

Extraction of total proteins and immunoblot analysis

Tissue samples were ground into powder with liquid nitrogen (N₂) and resuspended in protein extraction buffer (150 mM NaCl, 2 mM EDTA, 10% Glycerol, 50 mM Tris-HCl, pH 7.6, and cOmplete™, EDTA-free Protease Inhibitor Cocktail). After 10 min of incubation on ice, samples were centrifuged at 13,000 x g for 10 min at 4 °C. Protein concentration in the supernatant was determined using the Bradford assay⁷⁰.

For NGC incubation, 50 or 200 µM of NGC was mixed with total protein extracts (~ 4 µg) at 4 °C for 16 h. For incubation with compound extracts, compounds extracted from ~ 0.25 or ~ 1.0 mg dry weight of tt5-2 mutant under HL with or without acid hydrolysis were lyophilized and then resuspended with total protein extracts (~ 4 µg) at 4 °C for 16 h.

Protein extracts (~ 4 µg total protein) were loaded on 10% polyacrylamide gels without heating and without a reducing agent, and the protein gel was exposed to 300 µW/cm² UV-B for 30 min before transfer to a membrane to detect UVR8 dimers¹⁸ or under denatured conditions (samples were boiled for 10 min) to detect total UVR8s using anti-UVR8 (PhytoAB Inc, CA, catalog # PHY0809A) (1:4000) and goat anti-rabbit IgG H&L (HRP, 1:5000) with ECL Prime Western Blotting Detection Reagents (Cytiva, Marlborough, MA). Membranes were probed for plant Actin levels using anti-plant Actin monoclonal antibody (1:5000) and IRDye 680RD goat anti-mouse IgG secondary antibody (1:10,000) with the Sapphire Biomolecular Imager.

Size-exclusion chromatography, NanoDSF, and MST

Recombinant proteins His-SUMO-UVR8/UVR8^D96N/D107N/UVR8^W285F were induced by 0.2 mM IPTG at 18 °C for 16 h. Proteins were resuspended by sonication with lysis buffer (150 mM NaCl, 25 mM Tris, pH 8.0, 2 mM β-Mercaptoethanol, and 10 mM Imidazole) and then purified through Ni-NTA resin following the manufacturer’s recommendations with elution buffer (150 mM NaCl, 25 mM Tris, pH 8.0, 2 mM 2-Mercaptoethanol, and 250 mM Imidazole). The eluted samples were buffer-changed (150 mM NaCl, 25 mM Tris, pH 8.0, and 5 mM DTT) by size exclusion chromatography using Superdex 200 Increase 10/300 GL (flow rate at 0.4 mL/min) and collected as 0.25 mL fractions. The fractions containing His-SUMO-UVR8/UVR8^D96N/D107N/UVR8^W285F were combined and then concentrated through Amicon Ultra-15 centrifugal filters (MilliporeSigma). Protein concentration was determined by Bradford and adjusted to ~ 0.6 mg/mL. Purified His-SUMO-UVR8/UVR8^D96N/D107N/UVR8^W285F were treated with UV-B (300 µW/cm², 60 min) on ice. For nanoDSF and MST experiments, 100 µM of flavonoid compounds (NGC, naringenin, kaempferol, dihydrokaempferol, leucocyanidin, dihydroquercetin, or cyanidin) were incubated with purified UVR8 proteins (~ 1.5 µg) at 4 °C for an hour.

Prediction the binding sites of NGC on UVR8 proteins

The Protenix Server can be accessed at http://101.126.11.40:8000. To generate the predicted interactions between NGC and UVR8, we enabled both Multiple Sequence Alignment (MSA) and Evolutionary Scale Modeling (ESM) by selecting the “true” options. The input was uploaded as a CIF file, with MSA features activated to enhance folding accuracy by incorporating evolutionary information. The sequences for UVR8 and NGC were added along with the desired number of copies for the interaction prediction model. For molecular classification, UVR8 was designated as a protein, while NGC was entered using its Ligand Smiles representation (C1 = CC( = CC = C1/C = C/C( = O)C2 = C(C = C(C = C2O)O)O)O). Once all components were included, the interaction prediction request was submitted, and the results were received via email. The server provided five samples of predicted interactions in CIF and JSON file formats. CIF files were used to visualize molecular binding in PyMOL, while JSON files contained additional information such as the interface predicted template modeling (ipTM) score for each model.

To analyze and visualize the NGC-UVR8 interaction and binding sites, we used PyMOL. The NGC molecule was selected, and by navigating to Action > Modify > Around > residues within 4 Å of the NGC molecule were identified. Additionally, all polar contacts were located by selecting Action > Find > Polar Contact > to any atoms, where binding sites were highlighted using red dashed lines.

For binding free energy calculations, PyFepRestr was utilized as a PyMOL plug-in. The input parameters, including temperature (300 Kelvin) and force constant values (K raA, K θa, K θA, K φba, K φaA, and K φAB), were set to 10 kcal/mol/Ų and 10 kcal/mol/rad², respectively. Using the “Select 6 Atoms” function within the plug-in, three atoms from binding amino acids in UVR8 and three atoms from NGC were chosen. The plug-in then computed free energy values (ΔG_off and ΔG_on) in kJ/mol, providing insights into the interaction strength between NGC and UVR8.

RNA-seq analysis

Total RNA was extracted from Arabidopsis plants using QIAGEN RNeasy Plant Mini Kit following the manufacturer’s instructions. Approximately 1 μg of RNA was utilized for constructing the RNA-seq libraries, which were then sequenced with 100 bp paired-ends on an Illumina sequencing platform by BGI. Each sample yielded ~ 24 million paired-end reads. RNA-seq reads underwent quality assessment using FastQC (v0.11.5)⁶⁶. Subsequently, the reads were aligned to the Arabidopsis thaliana reference genome (TAIR 10) using Hisat2 (v2.1)⁷¹. Raw read counts for gene features were quantified from these alignments using FeatureCounts (v1.5)⁷². Transcript levels were calculated in transcripts per million (TPM) with StringTie (v2.1.1)⁷³. The R package DESeq2 (v1.36.0)⁷⁴ was employed to test for differential expression (DE) through paired contrasts between HL and NL. The resulting P-values were adjusted using Benjamini and Hochberg’s method⁷⁵. Genes with an adjusted P-value < 0.05, as determined by DESeq2, were considered differentially expressed, regardless of the fold change value, unless otherwise specified. Venn diagrams were generated using InteractiVenn (http://www.interactivenn.net/index.html#)⁷⁶. Gene ontology (GO) enrichment analyses were performed using the online tool ShinyGO v0.76: Gene Ontology Enrichment Analysis (http://bioinformatics.sdstate.edu/goc/)⁷⁷.

Statistical analyses

Statistical analyses were conducted using GraphPad Prism 10.1.1 (San Diego, California).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

N.J. and E.G. designed the research; N.J. performed most of the research and analyses; T.G.N. contributed with the sequencing of the mutants and RNA-seq analysis; Y.S.L. performed nanoDSF and MST experiments; E.P. conducted protein expression and purification; S.P. contributed with Arabidopsis growth in Hawai’i and simulation between protein and compound; N.J. and E.G. wrote the manuscript.

Peer review

Peer review information

Nature Communications thanks Hongtao Liu, Andreas Richter and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Data supporting the findings of this work are available within the paper and the respective Supplementary Information files. A reporting summary for this article is available as a Supplementary Information file. The raw RNA-seq reads and whole genome re-sequencing data generated in this study have been deposited in the NCBI BioProject database under accession codes PRJNA1065785, PRJNA1231375, and PRJNA1065513. The metabolomics data have been deposited to MetaboLights repository with the study identifier MTBLS12792. Source data are provided in this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-63010-3.