The Arabidopsis Transcriptome Responds Specifically and Dynamically to High Light Stress

Experimental Design of High Light and Recovery Treatments

To capture the dynamic transcriptional response of Arabidopsis plants to HL stress and recovery, we performed a time course RNA sequencing (RNA-seq) study. Arabidopsis seedlings were treated with high light (HL: 1,200 μmol m⁻² s⁻¹, leaf temperature: 22°C) for 0.5, 6, 12, 24, 48, and 72 h with corresponding growth light (GL: 60 μmol m⁻² s⁻¹, leaf temperature: 22°C) treatments as control. We also collected the recovery (R) samples (Figure 1A). To avoid stress memory, we used continuous light conditions. The spectrum of HL is shown in Figure 1B. In HL treatment, the light lamps are located at the top of the chamber and the cool air comes from the bottom. By changing the settings, we were able to maintain the temperature of the growth medium and leaf to match the control conditions (Figure 1C). To investigate the light intensity and leaf temperature during winter in the temperate zone in nature, we monitored the light intensity and plant leaf temperature around the Salk Institute for Biological Studies (coordinates: 32.887579° N, −117.244914° W). We found that for most of the day, the light intensity was more than 1,000 μmol m⁻² s⁻¹ and the leaf temperature was between 22°C and 25°C (Figure 1D), which suggested that in nature, plants could be exposed to high-intensity light with no accompanying heat stress. A preliminary experiment indicated that when the chamber was set to 22°C, under 1,200 μmol m⁻² s⁻¹ of HL, the leaf temperature rose to 28°C in 5 min, 32°C in 10 min, and 35°C in 30 min, which would generate heat stress on plants (Figure 1E). By setting the chamber to 11.5°C, the temperature of the medium and leaf was maintained at the normal growth temperature (22°C) (Figure 1E). Similar experimental settings have been employed to avoid leaf heating under HL in previous studies (Ramel et al., 2013). Thus, we excluded the effect of heat from our study.

We confirmed that the HL treatment was successful by examining the induction of HL marker genes. Consistent with previous studies (Casazza et al., 2005; Heddad et al., 2006), we found that ELIP1 and ELIP2 had low expression levels under GL and were highly induced by HL (Figure 1F). However, the expression of APX2 was only slightly upregulated under our HL treatment (Figure 1F). Previous studies showed that APX2 was highly induced by warm HL(Jung et al., 2013; Crisp et al., 2017). We inferred that APX2 mainly responded to heat or HL/heat-combined stresses. Thus, we examined the expression of ELIP1, ELIP2, and APX2 under HL and warm HL. The results showed that ELIP1 and ELIP2 were strongly upregulated by HL and moderately upregulated by warm HL. However, APX2 was only highly induced by warm HL (Figure 1G), suggesting that APX2 induction was primarily heat driven, but not light driven. These results emphasize one of the important aspects of our study design: separation of heat from HL to unravel the independent action and biological consequences of a high-intensity light-driven response.

Dynamic Transcriptome Responses of Plants to Short-, Middle-, and Long-Term HL

We used pairwise comparisons (e.g., 0.5 h of HL [HL0.5h] versus 0.5 h of GL [GL0.5h]) to identify differentially expressed genes (DEGs) of each time point. The average reads per kilobase of transcript per million mapped reads (RPKM) of each gene at all time points are listed in Table S1. In total, 5,403 DEGs were identified for at least one time point of HL (Table S2). In detail, there were 1,300 DEGs for 0.5 h, 1,780 DEGs for 6h, 2,021 DEGs for 12 h, 3,081 DEGs for 24 h, 2,681 DEGs for 48 h, and 2,885 DEGs for 72 h (Figure 2A). The recovery sample had 2,156 DEGs compared with 72 h of HL (HL72h) and only had 604 DEGs compared with 72 h of GL (GL72h) (Figure 2A). The expression profiles of all DEGs fell into significant short-term (0.5 h), middle-term (6 to 12 h), and long-term (more than 24 h) HL response patterns (Figure 2B). cis-element analysis showed that at most time points, the G box (CACGTG) was significantly enriched in the promoters of DEGs (Figure 2C). We identified nine significant expression clusters from all DEGs in response to different durations of HL, which includes two main clusters: gradually downregulated (1,227 genes) and gradually upregulated (560 genes) (Figure 2D; Table S3). GPT2, required for dynamic acclimation of photosynthetic capacity to increase light (Athanasiou et al., 2010), was in cluster 31 (Figure S1A). MAPKKK18, which was upregulated by HL (Ramel et al., 2013), was in cluster 33 (Figure S1A). ERF6 displayed significant upregulation at 10 and 30 min after the low light to HL shift, followed by a strong decline (Vogel et al., 2014). A similar expression pattern was observed for ERF6, which was in cluster 35 (Figure S1A).

After recovery, about 57% of HL upregulated genes were downregulated and 66% of HL downregulated genes were upregulated (Figure S1B; Table S2). When using 2-fold change as the criterion, 79% of HL72h DEGs restored their expression to the non-stress level while the rest (21%) could not (Figure S1C). For example, the expression of RHL41 restored to the non-stress level after recovery. AT1G12030 did not express under GL, was highly induced by HL, and recovered to the moderate expression level after recovery. The opposite expression pattern was observed for LHCB2.2 (Figure S1D).

To our knowledge, most previous studies describing the HL-related transcriptomes only had one or two time points and did not exclude the interference of heat. We compared our DEGs with those identified in previous studies. We found that 85% of DEGs in Kleine et al. (2007), 36% of DEGs in Jung et al. (2013), 59% of DEGs in Suzuki et al. (2015), and 42% of DEGs in Crisp et al. (2017) were observed in our study. We also found that more than 70% of our HL DEGs were not identified in previous studies and some previously reported DEGs did not appear in our data (Figure S2A). Previous studies showed that many heat shock factor (HSF) and HSP genes, including HSFA2, HSP70, HSP17.6A, HSP18.2, and HSP21, were highly induced by short-time HL (Crisp et al., 2017). However, in our data, HSP18.2 did not show differential expression under HL, while HSP70, HSP17.6A, and HSP21 were uninduced or lowly induced under 0.5 and 6 h of HL treatment (Figure 2E; Table S2). We confirmed this observation by qPCR (Figures 2F-2J) and found that most of these HSF and HSP genes were only highly induced by warm HL. These results indicated that the induction of these HSF and HSP genes under short-time HL treatment was heat driven.

We also compared our data with transcriptomes of other abiotic stresses, including heat (Higashi et al., 2015) and drought (Crisp et al., 2017). We found that only 16% of the genes were common between HL and heat and 30% of the genes were shared by HL and drought (Figure S2B). For these three stresses, 226 genes were common DEGs, and their enriched Gene Ontology (GO) terms were mainly related to stress response (Figure S2C).

The Set of Core HL Stress-Responsive Genes

Of the 5,403 DEGs, 250 genes were differentially expressed across all six HL time points (Figure 3A; Table S4A). These 250 common genes appeared to be the core HL-responsive genes. The expression pattern of these genes clustered into three groups. Genes in group I had dramatically decreased expression, expression of group II genes showed a moderate increase of less than 5-fold, and expression of group III genes increased dramatically (Figure 3B). Among the 250 common genes, in addition to known HL-responsive genes (e.g., ELIP1, ELIP2, GPT2, and RHL41), we found some highly induced genes (e.g., ELI3-2, AT1G71000, and BAM5) and severely repressed genes (e.g., BBX17, SAUR3, and SAUR27). Moreover, five genes were from the B box (BBX) gene family, which has 32 members (Khanna et al., 2009), and nine genes were from the SMALL AUXIN UPREGULATED RNA (SAUR) gene family, which has 80 members (Markakis et al., 2013). BBXs can act as negative or positive regulators of light signaling (Huang et al., 2012). Four of the five BBX genes were downregulated at all six time points, while BBX31 was upregulated (Figure 3C). We also found that BBX31 was the upstream regulator of ELIP2 based on the Arabidopsis cistrome data (O’Malley et al., 2016) (Figure 3D). Our data also showed that 18 BBX genes were regulated by HL (Figure 3C; Table S4B). All nine SAUR genes were downregulated from 0.5 to 72 h of HL. Moreover, 46 SAUR genes were differentially expressed for at least one HL time point (Figure 3E; Table S4B).

We then constructed a co-expression network of the 250 common DEGs (Figure S3A). We first extracted the sub-network using RHL41 as the seed gene, which was involved in the HL stress response (lida et al., 2000; Rossel et al., 2007). RHL41 had 67 connections in the co-expression network. Besides ELIP1 and ELIP2, BBX17, BBX27, SAUR15, SAUR27, and SAUR29 co-expressed with RHL41 (Figure S3B). The top three hub genes that had the highest degree and betweenness centrality in the co-expression network were AT1G27210, AT4G01330, and BBX14 (Figure S3A). AT1G27210, a member of the ARM repeat superfamily, had 129 connections (Figure 4A). AT4G01330 is a protein kinase gene and had 124 connections (Figure 4B). BBX14 had 112 connections (Figure 4C). All three hub genes were co-expressed with ELIP1, ELIP2, and GPT2, suggesting that these genes were potentially important HL response regulators. We acquired the mutants of AT1G27210 and AT4G01330 and assayed their phenotypes under HL. We found that after about 20 h of HL, AT1G27210 mutants had more plants with bleached, paler cotyledons and less chlorophyll content compared with the wild type, suggesting that this hub gene might play an important role in HL stress response (Figures S3C and S3D).

Gene Ontology (GO) Term Enrichment of HL Differentially Expressed Genes

To gain insight into biological processes in response to HL, we analyzed the GO term enrichment of DEGs of different HL time points. For HL upregulated DEGs, GO terms related to stress and stimulus response were significantly enriched for all time points. GO terms related to high light intensity, light intensity, response to light stimulus, UV, and UV-B were also enriched. Hormone-related GO terms, e.g., response to abscisic acid (ABA), jasmonic acid (JA), auxin, salicylic acid (SA) stimulus, or mediated signaling pathway, were enriched. In addition, anthocyanin and flavonoid metabolic process were enriched after 6 h of HL (Figure S4A; Table S5). For downregulated DEGs, stress and light response-related GO terms were enriched, and photosynthesis-related terms appeared after 6 h of HL. The GO term enrichments for long-term HL (24, 48, and 72 h) were similar and included development-related GO terms, e.g., cell differentiation, cell wall modification, trichoblast differentiation, epidermis, ectoderm, and root development. Moreover, terms related to growth and development hormones brassinosteroids (BRs), auxin, and cytokinin showed up (Figure S4B; Table S5). The enriched GO terms for the 0.5 h DEGs were different from those of other time points; e.g., GO terms related to gene expression, RNA biosynthetic, RNA elongation, and RNA metabolic only appeared for 0.5 h. Altogether, this analysis suggests that HL significantly regulates the expression of genes involved in stress stimulus, light response, hormone, anthocyanin metabolism, photosynthesis, and development. Of interest, the top GO terms for upregulated DEGs of recovery were the GO terms for the downregulated DEGs of 24, 48, and 72 h, and vice versa (Figure S4; Table S5).

Hormone Biosynthetic and Signaling Pathways Are Involved in HL Stress Response

To inspect the role of hormones in different durations of HL stress response, we analyzed the expression changes of biosynthetic genes for different phytohormones. Most genes in ABA and JA biosynthetic pathways showed increased expression at all time points. SA, ethylene, and gibberellic acid (GA) biosynthetic genes only showed significant expression changes at a few time points (Figure 5A). We identified nine DEGs from the ABA biosynthetic genes, including NCED3, NCED5, and NCED9, which were upregulated after 0.5 h of HL, and NCED2, which had increased expression only after 24 h of HL. Four CYP707 genes involved in the oxidative catabolism of ABA were upregulated under HL (Table S6). In addition, ABA signaling pathway genes were regulated by HL (Figure 5B). Six of the 14 ABA receptor genes were HL DEGs, and most of them were downregulated after HL. Eight of the nine clade A PP2C genes were persistently upregulated by HL (Figure 5B). We measured the ABA levels in plants under GL or HL at each time point. The ABA level increased slightly after 0.5 h of HL, dramatically increased by more than 2-fold after 6 h of HL, and then maintained at a high level for the entire duration of HL. The ABA level decreased during recovery but was still higher than GL control (Figure 5C). To validate the role of ABA in HL, and because NCED3 was differentially expressed across all six HL time points, we investigated the phenotype of ABA biosynthesis-defective mutant nced3nced5. After 24 h of HL, the nced3nced5 mutants were hypersensitive to HL; more mutant plants had bleached cotyledons with larger bleached areas compared with the wild type (Figure 5D). To quantify the phenotype, we found that nced3nced5 mutants had less chlorophyll content compared with the wild type under HL (Figure 5E), which suggests that ABA is required for HL stress response.

Twelve of 17 JA biosynthetic genes were differentially expressed. Seven of nine SA biosynthetic genes were DEGs, and nine of 12 ethylene biosynthetic genes were differentially expressed for at least one time point (Table S6). Of the 24 GA biosynthetic genes, 11 were differentially expressed for at least one time point (Table S6). BRs, auxin, and cytokinin are known to regulate many aspects of plant development, and many of their biosynthetic genes showed changes in transcript abundance in response to HL (Figure 5A). BR biosynthetic pathway genes showed repressed expression at all time points, while genes involved in auxin and cytokinin biosynthesis showed decreased expression under long-term HL. Three of the BR biosynthetic genes were downregulated from short- to long-term HL, while UGT73C5, which is responsible for BL-23-O-glucosylation, was upregulated across all HL time points (Table S6). Fourteen of the 32 genes involved in auxin biosynthesis were DEGs, of which 10 were downregulated for at least one time point (Table S6). Seven of 10 cytokinin biosynthetic genes were downregulated for at least one time point. The downregulation of genes involved in hormones related to plant growth and development correlates with the observation that plant growth is retarded under HL stress. In addition, JA, GA, BR, auxin, and cytokinin signaling pathway genes were regulated by HL (Figure S5).

The Genes for Blue/UV-A Receptors and PIFs Are Responsive to HL Stress

We asked whether genes for photoreceptors and phytochrome-interacting factors (PIFs) were responsive to high-intensity light-driven signaling. We analyzed the expression of genes for red/far-red photoreceptors (PHYA, PHYB, PHYC, PHYD, and PHYE), blue/UV-A photoreceptors (CRY1, CRY2, PHOT1, PHOT2, ZTL, FKF1, and LKP2), the UV-B receptor (UVR8), and PIFs (including PIF1, PIF3, PIF4, PIF5, and PIF7). The expression levels of PHYA to PHYE were similar between GL and HL (Table S1). However, CRY1 and PHOT1 were downregulated after HL, and FKF1 showed rapid upregulation followed by downregulation (Figure 6A). The similar expression pattern for blue/UV-A photoreceptor genes under HL may explain the synergistic relationships of different receptors in acclimation to HL. Cryptochromes interact with UVR8 and are required for plant survival in natural and simulated sunlight (Rai et al., 2019). We then compared the HL DEGs with UV-B-regulated genes and found that 58% (778 of 1,349) of UV-B-regulated genes (Favory et al., 2009) were also differentially expressed under our HL treatment (Figure S6A).

We found that PIF1, PIF4, PIF5, and PIF7 genes were all downregulated by HL (Figure 6B). To investigate the role of PIF genes in HL stress response, we examined the phenotype of pif1, pif4, pif5, pif7, pif4,5, and the quadruple mutant pif1,3,4,5 (pifq). After 24 h of HL, pif4, pif5, pif4,5, and pifq mutants showed HL-hypersensitive phenotypes; the mutants had more plants with bleached cotyledons, and the bleached areas in cotyledons of mutant plants were larger compared with the wild type (Figure 6C). We found that pif4, pif5, pif4,5, and pifq mutants had less chlorophyll content compared with the wild type under HL (Figure 6D). We compared the HL DEGs with PIFQ-regulated genes (Zhang et al., 2013) and found that 44% (891 of 2,025) of pifq-regulated genes were differentially expressed under HL (Figure S6B), which supported the important role of PIFs in HL stress response.

PIFs are the key regulators in shade-avoidance responses (Lorrain et al., 2008; Li et al., 2012). Shade also increases the endogenous ABA level, probably by enhancing the transcript level of ABA biosynthetic genes NCED3 and NCED5 (Kohnen et al., 2016). After we observed that NCED3/5 and PIFs were required for HL stress response, and to determine the link between NCED3/5 and PIFs under HL, we examined the expression of NCED genes in pifq mutants before and after HL treatment. The results showed that the induction of NCED3 and NCED5 by HL were similar between wild-type and pifq mutants (Figure S6C), suggesting that the induction of NCED3/5 by HL was independent of PIFs.

Genes Encoding the Photosynthetic Apparatus Respond to Middle- to Long-Term HL

Our data showed that the expression of most photosystem genes changed little after 0.5 h of HL. However, these genes were dramatically repressed after 6 h of HL, indicating an adjustment in the light harvesting system and reaction centers to capture less light energy for photosynthesis. Eleven nucleus-encoded genes in Photosystem I were downregulated from 6 to 72 h of HL and upregulated after recovery (Figure 7A). In Photosystem II, the expression of PsbP1, PsbO1, PsbQ1, and PsbQ2 was downregulated (Figure 7A). The expression of genes encoding Light harvesting complex I (LHCI) proteins, including Lhca1, Lhca2.1, Lhca3, and Lhca4, was also repressed after HL (Figure 7A). Transcript levels of 13 of the 16 genes encoding Light harvesting complex II (LHCII) proteins were reduced under HL (Figure 7A). The expression of PPL1 and PnsL4 was downregulated more than 2-fold at 48 h of HL. The transcript levels of other component genes in the electron flow chain, including the cytochrome b6f complex, ferredoxin, and the cyclic electron flow switch, were similar to control, except petM and ferredoxin 2 were suppressed since 48 h of HL, PC1 was downregulated from 6 to 48 h, and PC2 was downregulated at 24 and 48 h of HL (Table S6). In contrast, most genes involved in the Calvin-Benson cycle did not change their expression. This suggests that HL is more harmful to light reaction than the Calvin-Benson cycle (Table S6).

Consistent with the downregulation of plant growth-related hormone genes and photosynthetic genes, after 24 h of HL, the fresh weight of HL-treated plants was significantly lower compared with control, indicating that the growth of plants was inhibited by long-term HL (Figure 7B).

Anthocyanin Biosynthetic Genes Are Activated after 6 h of HL Treatment

Under HL, Arabidopsis plants accumulate anthocyanin in vegetative tissues, which can act as sunscreen, protecting cells from photoinhibition and damage by absorbing blue-green and UV light. We analyzed the expression profile of anthocyanin biosynthetic genes (Solfanelli et al., 2006). Anthocyanin biosynthesis can be divided into three phases: the general phenylpropanoid pathway, the flavonoid pathway, and the anthocyanin-specific pathway (Shi and Xie, 2014). PAL1, C4H, and 4CL proteins are involved in the beginning steps of these pathways, and six of their corresponding genes were DEGs. PAL3 was downregulated since 0.5 h of HL, and the rest of the genes were upregulated after 6 h of HL (Figure 7C; Table S6). Chalcone synthase, chalcone isomerase, and flavanone 3-hydroxylase (F3H) carry out the initial steps of the flavonoid pathway. Five genes encoding these enzymes were upregulated after 6 h of HL and maintained a high expression level in long-term HL. Three MYB transcription factor genes (MYB11, MYB12, and MYB111) were upregulated across all six time points (Figure 7D; Table S6). The late steps transform dihydroflavonols into anthocyanidins and modify the anthocyanidins. Eighteen genes involved in these steps were differentially expressed (Figure 7E; Table S6). The anthocyanin content significantly increased after 12 h of HL and continuously accumulated with HL treatment time lapse. After recovery, the anthocyanin content decreased but was still higher than that of non-treatment control (Figure 7F).

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Joanne Chory (chory@salk.edu). This study did not generate new unique reagents.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

METHOD DETAILS

QUANTIFICATION AND STATISTICAL ANALYSIS

Significant differences between two samples were determined with Student’s t test. Significant differences for multiple comparisons were determined by one-way or two-way ANOVA as indicated in figure legends. The plots and heatmaps were generated using R (https://www.r-project.org/). Technical and biological replicate experiments were performed as indicated.

DATA AND CODE AVAILABILITY

The RNA-seq raw data in this paper were deposited into the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus database under accession number GEO: GSE111062.