Photosynthetic signalling during high light stress and recovery: targets and dynamics

Abstract

The photosynthetic apparatus is one of the major primary sensors of the plant's external environment. Changes in environmental conditions affect the balance between harvested light energy and the capacity to deal with excited electrons in the stroma, which alters the redox homeostasis of the photosynthetic electron transport chain. Disturbances to redox balance activate photosynthetic regulation mechanisms and trigger signalling cascades that can modify the transcription of nuclear genes. H₂O₂ and oxylipins have been identified as especially prominent regulators of gene expression in response to excess light stress. This paper explores the hypothesis that photosynthetic imbalance triggers specific signals that target discrete gene profiles and biological processes. Analysis of the major retrograde signalling pathways engaged during high light stress and recovery demonstrates both specificity and overlap in gene targets. This work reveals distinct, time-resolved profiles of gene expression that suggest a regulatory interaction between rapidly activated abiotic stress response and induction of secondary metabolism and detoxification processes during recovery. The findings of this study show that photosynthetic electron transport provides a finely tuned sensor for detecting and responding to the environment through chloroplast retrograde signalling.

This article is part of the theme issue ‘Retrograde signalling from endosymbiotic organelles’

1. Photosynthetic redox balance and the impact of light

Photosynthesis converts light energy into the stored chemical energy of carbon–carbon bonds through a series of sequential redox reactions. Light energy is collected and funnelled to the reaction centres of photosystem II (PSII) and photosystem I (PSI), inducing the oxidation of water molecules by PSII and the reduction of stromal acceptors by PSI. Photosynthetic redox balance can be defined as an equilibrium between the amount of light energy used for photochemistry and the capacity of metabolism to use the excited electrons. In a balanced state, redox cofactors are sufficiently oxidized and the movement of electrons through the photosynthetic electron transport chain (PETC) from water to stromal sinks is fluent. Sudden increases in available energy within the PETC, or decreases in stromal acceptor capacity, both of which frequently occur amidst changes in the natural growth environment, can disrupt the fluency of photosynthetic electron transport. Under these circumstances, energy and electrons find alternative targets, including nearby oxygen molecules that leads to the formation of distinct reactive oxygen species (ROS). Excessive ROS levels are associated with oxidative damage to proteins, lipids and nucleotides, and are controlled by antioxidant scavenging systems. However, chloroplast ROS induce long-range signalling networks that modify the nuclear transcriptional program of the cell (reviewed in [1,2]), linking the redox status of the photosynthetic light reactions to the communication of, and acclimation to, changes in the environment (see reviews [1,3–5]). In this way, the photosynthetic light reactions operate as both a sensor of environmental conditions such as light intensity, temperature and water availability, and a transmitter of signals important for acclimation to changes in the environment.

Light poses a fundamental conundrum for redox balance in the PETC, mostly due to rapid and enormous fluctuations in light intensity in the natural environment. In plants, a large, energetically connected system of light-harvesting antennae concentrates light energy towards PSI and PSII, which is ideal for maximizing photosynthesis under low light. However, a sudden increase in light intensity can quickly increase absorbed energy, causing the upregulation of PSII and PSI activities that leads to the saturation of stromal electron acceptors, over-reduction of intersystem electron carriers and excessive ROS production [2]. Photosynthetic redox homeostasis is protected and/or restored by several independent mechanisms that regulate light-harvesting and electron transport, including non-photochemical quenching (NPQ), control of cytochrome b₆f and reversible phosphorylation of light-harvesting complex II (LHCII; reviewed in [6]). However, regulation does not completely prevent disruption of redox balance. This is clearly evident in changes in chlorophyll a fluorescence at room temperature, which reflects changes in the redox state of the Q_A site in PSII and can also infer reduction pressure from the reduced plastoquinone (PQ) pool or changes in energy dissipation by NPQ. Increased light intensity induces a rapid, transient increase in chlorophyll a fluorescence that indicates reduced Q_A and hence ‘closed’ PSII (reviewed in [7]). Furthermore, the steady-state level of chlorophyll a fluorescence increases with increasing light intensity, despite efficient photoprotection [8], demonstrating that the steady-state redox balance of the PETC continually responds to the environment. Changes in the redox status of the PSI reaction centre, as well as the PSI electron donor (lumenal) and acceptor (stromal) sides, are also evident during fluctuations in light intensity. Increased irradiance upregulates both PSI electron transport and stromal metabolism, but also increases electron pressure at the PSI donor side and activates regulation mechanisms described above [6,9]. Although changes in environmental conditions may be minor or transient, they have an immediate effect on the energy balance within the PETC; therefore, even small environmental fluctuations may be sensed and signalled through photosynthesis by production of metabolic side-products. Although many metabolites have been implicated in chloroplast retrograde signalling (see [10] for a comprehensive review), the current analysis has identified ROS and oxylipins as the primary instigators of operational signalling related to photosynthetic redox balance, as discussed below.

2. Photosynthetic redox balance and chloroplast signalling

In order to explore the effect of disturbed photosynthetic homeostasis on nuclear gene transcription, and to high-light the specific pathways employed by photosynthetic signalling, we analysed a publicly available set of Arabidopsis thaliana transcriptomics data collected during a light stress and recovery regime [11] (data retrieved from NCBI Sequence Read Archive SRP109986). The dataset derives from plants exposed to 10-fold increases, and subsequent decreases, in light intensity, resembling many light stress experiments in the database; however, the experiment of Crisp et al. [11] was selected due to their implementation of both sustained and repetitive stress treatments (see experiment description in figure 1). Approximately 10 000 genes underwent a greater than or equal to twofold change (FC) in expression during the course of the experiment, relative to the unstressed starting condition (time-point 1 in figure 1).

Figure 1.

Expression profiles of light stress-responsive genes. Clustered heatmaps showing expression of Arabidopsis thaliana genes responsive to (a) hydrogen peroxide (H₂O₂; GO:0042542 and NCBI GEO accessions GSE41136 and GSE5530), (b) jasmonic acid (JA; GO:0009753 and EBI ArrayExpress accession E-MEXP-883), (c) 12-oxophytodienoic acid (OPDA; NCBI GEO accession GSE10732) and (d) phytoprostane A₁ (PPA₁; NCBI GEO accession GSE10732). Heatmaps were constructed from genes that underwent ≥twofold change in expression within the high light (HL) stress and recovery experiment of Crisp et al. [11]. Transcriptomics data were downloaded from NCBI Sequence Read Archive accession SRP109986. Colours indicate the normalized expression value of each gene, according to the scale shown in the bottom left of the figure. Numbers in parenthesis indicate the percentage of the respective gene list that underwent ≥twofold change in expression. The treatment regimen comprised successive exposures of plants to 100 µmol photons m^–2 s⁻¹ (GL, growth light) and 1000 µmol photons m⁻² s⁻¹ (HL; see arrows at top of heatmaps) for the time periods indicated in (e) (in minutes or hours, as indicated). Transcripts were sampled at time-points indicated by numbers 1–11. Distinct temporal expression profiles identified in the current analysis of clustered heatmaps are indicated on the right-hand side; genes of profile S were induced under HL stress; profile R was induced during recovery and profile G was induced after long-term recovery from HL stress in normal GL conditions.

Gene ontology (GO) term enrichment analysis of this large group of differentially expressed genes showed significant over-representation of the GO terms ‘response to hydrogen peroxide’ (GO:0042542) and ‘response to jasmonic acid’ (GO:0009753). Production of hydrogen peroxide (H₂O₂) in the plant cell is well known to be rapidly induced under conditions of photosynthetic redox imbalance [12–17]. H₂O₂ formed in the chloroplast by photoreduction of O₂ is linked to limited availability of oxidized electron acceptors at the stromal side of PSI (reviewed in [2,18]). In addition, H₂O₂ production in the peroxisome during glycolate turnover has been co-opted as a proxy chloroplast signalling mechanism connected to photorespiration activity [14,19–22]. Chloroplast- and peroxisome-specific H₂O₂ signals are known to induce partially distinct gene expression profiles, with chloroplastic H₂O₂ linked to detoxification and secondary metabolism, and peroxisomal H₂O₂ linked to protein damage and unfolding [22]. The oxylipin hormone jasmonic acid (JA) is most commonly associated with biotic stress response and regulation of growth and development (reviewed in [23]); however, biosynthesis and signalling activities of JA and its precursor 12-oxophytodienoic acid (OPDA) are connected to photosynthetic production of singlet oxygen (¹O₂; see below), which is generated during excitation/electron imbalance at PSII (reviewed in [24]). The ¹O₂ signalling intermediate β-cyclocitral (β-CC) was shown to upregulate enzymatic lipid oxidation and oxylipin synthesis [25–27], as did ¹O₂ over-production in the ch1 and flu mutants (see reviews [28–30]). Upregulated ¹O₂ and OPDA/JA production and signalling in npq mutants [31–35] with inhibited dissipation of excess LHCII excitation, and in the stn7 mutant [36] that lacks LHCII phosphorylation, further underscore the connection between PSII excitation/electron balance and oxylipin signalling.

To study the activities of these chloroplast signalling pathways more closely, we retrieved lists of genes belonging to the above GO terms and supplemented them with genes found to be upregulated in experimental treatments of Arabidopsis plants with either H₂O₂ (NCBI GEO accessions GSE41136 and GSE5530) or methyl JA (EBI ArrayExpress accession E-MEXP-883). In addition, genes found to be upregulated in response to exogenous OPDA treatment (NCBI GEO accession GSE10732) were analysed in the current study. OPDA is synthesized in the chloroplast through enzymatic processing of oxidized linolenic acid (18:3) and provides the precursor for JA synthesis in the peroxisome. Both OPDA and JA are therefore referred to as ‘enzymatic’ oxylipins (for recent reviews, see [37,38]). OPDA is also considered a ‘reactive electrophile species' (RES) due to an α,β-unsaturated carbonyl, while JA is a non-RES oxylipin due to lack of this reactive feature [39,40]. The impact of OPDA on nuclear gene expression is known to be partially independent of JA signalling, although there is substantial overlap between the gene targets of the two oxylipins (see [41–43] and results described below). In the current analysis, we also included a list of genes upregulated by phytoprostane A₁ (PPA₁; NCBI GEO accession GSE10732), which represents radical-derived oxylipins [39]. Hydroxyl radical (^•OH) and lipid peroxide radicals (LOO^•) are highly reactive ROS that can form in the chloroplast through ‘Fenton-like’ reactions between peroxides (H₂O₂ and LOOH, respectively) and reduced transition metals (e.g. Fe²⁺, reviewed in [44,45]). Both ^•OH and LOO^• react with unsaturated lipids to propagate the formation of lipid radicals (L^•) that can spontaneously cyclize to form a series of phytoprostane-type oxylipins, or fragment to form aldehydes such as malondialdehyde (MDA) and 4-hydroxy-2-nonenal [39,46]. In addition to enzymatically derived RES such as OPDA, many products of radical-induced lipid oxidation are also RES that can form adducts through reaction with nucleophilic thiols in amino acids and glutathione (GSH; reviewed in [40]).

The expression dynamics of the H₂O₂-, JA-, OPDA- and PPA₁-induced genes within the transcriptomics data of Crisp et al. [11] were analysed in order to assess the activity of these signalling pathways during high light (HL) stress and recovery. This analysis revealed three distinct, recurring profiles of nuclear gene expression that were induced during HL stress, during recovery or after acclimation to growth light (GL) conditions, which were dubbed profile S, profile R and profile G, respectively (figure 1). Strong induction of profile S genes in the ‘H₂O₂-responsive genes' occurred after 30 min in HL (figure 1, time-point 2). This group predominantly contained large and small heat shock protein (HSP) chaperones and heat shock-responsive transcription factors (HSFs). Rapid and intense upregulation of these so-called ‘heat shock’ genes comprises a typical nuclear transcriptional response to excess light and other abiotic stresses [47–50]. H₂O₂ profile S genes remained upregulated, compared to non-stress conditions, throughout the HL treatment (time-points 2–4) and during the initial minutes of recovery in GL (time-point 5), and were downregulated 15 min after the transfer back to GL for recovery (time-point 6). The low basal expression level of H₂O₂ profile S was largely restored after 60 min recovery in GL conditions (time-point 8), while further HL treatments given after 60 min or 24 h recovery under GL (time-points 9 and 11, respectively) upregulated these genes in a similar manner to the initial HL stress. A second expression-based cluster of H₂O₂-responsive nuclear genes was denoted ‘profile R’ for ‘recovery-induced profile’ (figure 1), which contained factors involved in stress-related detoxification, including glutathione S-transferase (GST) and UDP-glycosyltransferase (UGT) enzymes. Expression of H₂O₂ profile R genes was largely downregulated during HL stress and upregulated under recovery conditions (time-points 5–8 and 10).

The majority of ‘JA-responsive genes’ followed the expression profile R, i.e. expression remained low during the HL stress periods, while strong induction occurred after 15–60 min of recovery under GL (time-points 6 and 7). The most responsive JA profile R genes were enzymes of the JA biosynthesis pathway: lipoxygenases (LOXs), allene oxide synthase (AOS) and cyclase (AOC), OPDA reductase (OPR3) and OPC-8:0 CoA ligase (OPCL1), as well as many JA-ZIM domain (JAZ) proteins that negatively regulate JA-induced genes (reviewed in [51]). Numerous MYB-type transcription factors were also strongly expressed within the JA gene set, as well as several cytochrome P450 monooxygenase (CYP450) enzymes associated with metabolic and detoxification processes related to stress response. A minor cluster of JA-induced genes differed from profile R in its higher expression level under non-stress conditions and after longer recovery periods (figure 1, time-points 1, 8, 10). This set, denoted ‘profile G’ due to high expression after long-term acclimation to growth conditions, included many factors involved in biosynthesis and transport of secondary metabolites, as well as several MYB transcription factors and defence-related genes PDF1.2a and PDF1.2b, and the classical JA marker genes VSP1 and VSP2.

Most ‘OPDA-responsive genes' followed typical profile R expression, with low expression levels during the HL treatment and strongest induction during recovery periods, especially after 15–60 min in GL (figure 1). Approximately 250 genes belonged to OPDA profile R in the current experiment, encoding many enzymes related to secondary metabolism including numerous members of the CYP450 and UGT families involved in flavonoid and amino acid metabolism, as well as several GSTs associated with detoxification of electrophilic compounds. Also included in OPDA profile R were enzymes for OPDA and JA synthesis listed above. Although OPDA and JA profile R genes showed strong expression correlation in the current treatment, comparison of the gene contents of respective JA and OPDA profile R sets highlighted both common and specific gene targets for these two oxylipins (figure 2, discussed below). Similarly, the contents of expression ‘profile G’ found in OPDA-induced genes in response to HL and recovery (figure 1) contained some overlap with JA profile G, but largely comprised unique entities including many GST and CYP450 genes, suggesting OPDA-specific signalling after long-term acclimation to growth conditions independent from JA.

Figure 2.

Light-responsive chloroplast signalling pathways have both specific and common impacts on the expression of target genes. The Venn diagram contains genes responsive to hydrogen peroxide (H₂O₂), jasmonic acid (JA), 12-oxophytodienoic acid (OPDA) and phytoprostane A₁ (PPA₁) that underwent a ≥twofold change in expression within the HL stress and recovery experiment of Crisp et al. [11]. Numbers in parenthesis indicate the total number of genes in each list. Circled numbers represent the number of genes found exclusively in a single gene list (filled circles), or common to two or more gene lists (open circles). A selection of genes identified in each section of the Venn diagram, which are mentioned in the text or otherwise relevant to the current analysis, are presented in boxes. The reader is encouraged to consult the full gene lists from each section of the Venn diagram, which can be found in electronic supplementary material, File S1. The Venn diagram was constructed using Venny 2.1 (http://bioinfogp.cnb.csic.es/tools/venny/index.html). Gene descriptions were extracted using g:Convert (https://biit.cs.ut.ee/gprofiler/convert).

Expression of another set of OPDA-responsive genes clearly resembled H₂O₂ profile S identified above (figure 1), comprising about 40 genes that were strongly induced after 30 min HL treatment and downregulated during recovery in GL (figure 1). Most members of OPDA profile S also appeared in H₂O₂ profile S, underscoring the overlap between the OPDA and H₂O₂ signalling pathways that has already been identified [52,53].

Expression of nuclear genes responsive to the so-called ‘non-enzymatic’ oxylipin PPA₁ also clustered into temporal profiles S, R and G in the current analysis (figure 1). The contents of PPA₁ profiles R and G overlapped considerably with the respective OPDA profiles R and G, although the PPA₁ profiles contained substantially fewer genes and notable differences between the two profiles were identified (discussed below). The constituent genes in the typical profile S found in PPA₁-induced genes predominantly encode HSPs and HSFs, overlapping considerably with OPDA and H₂O₂ profile S.

3. Common and specific regulation targets of photosynthetic signalling

A set comparison of all H₂O₂-, PPA₁-, OPDA- and JA-responsive genes that underwent differential regulation in the analysed HL treatment and recovery experiment [11] revealed both common and specific regulation targets of each signalling pathway during light stress and recovery (figure 2). Full lists of genes used in this analysis can be inspected in the electronic supplementary material, File S1. Based on the analysis described above, nuclear gene targets of light stress- and recovery-responsive signalling were grouped into six general categories (numbered 1–6, see below) that have been used to describe the specific and overlapping cellular effects of photosynthetic signalling in figure 3.

Figure 3.

Hypothetical scheme of photosynthetic signalling based on analysis of gene expression. Changes in environmental conditions disturb photosynthetic redox balance due to over-supply of light energy or over-reduction at the acceptor side of PSI and/or PSII. Red arrows depict excitation pressure from light-harvesting complex LHCII (at PSII) and insufficient acceptor capacity (at both PSII and PSI) that lead to formation of distinct reactive oxygen species (ROS; written in red). Production of singlet oxygen (¹O₂) at PSII upregulates β-cyclocitral (β-CC) and/or executor (EX1) signalling pathways, leading to upregulated enzymatic oxylipin synthesis. ¹O₂ also induces non-enzymatic oxidation of chloroplast lipids (LOOH). Hydrogen peroxide (H₂O₂) production at PSI triggers H₂O₂ signalling pathways. H₂O₂ and LOOH can react with ferrous iron (Fe²⁺), causing the formation of lipid radicals (L^•) and subsequent non-enzymatic production of oxylipins, such as the phytoprostane (phytoP) series that includes PPA₁. The gene targets of chloroplast signals that were analysed in the current work have been grouped into general categories 1–6. A classical abiotic stress response comprising upregulation of protein chaperones and heat stress proteins (category 1) was found to be induced by H₂O₂, OPDA and PPA₁. Genes encoding enzymes involved in ROS and RES detoxification (category 2), especially glutathione S-transferases and UDP-glycosyltransferases, were upregulated by H₂O₂, OPDA and PPA₁, as well as β-CC [25]. Notably, enzymatic detoxification strategies presumably antagonize the signalling activities of ROS and RES, as depicted by orange bars. Secondary metabolite biosynthesis enzymes (category 3), most prominently cytochrome P450 monooxygenases, were upregulated by OPDA, JA and PPA₁. Oxylipin biosynthesis enzymes (category 4), including lipoxygenases and allene oxide synthase/cyclases, were induced by OPDA and JA, as well as β-CC [25], presumably propagating enzymatic oxylipin signalling pathways (indicated by blue arrows). Factors involved in regulation of JA signalling (category 5, indicated by pink bars), including JAZ-type regulators, were induced by OPDA and JA. Genes encoding MYB transcription factors known to regulate expression of secondary metabolite enzymes (category 6) were upregulated by OPDA and JA.

Light-induced genes exclusively responsive to H₂O₂ signalling included peroxisomal catalase (CAT2) and four chloroplastic ferritin iron chaperones (Fer1– 4). Heat stress transcription factors HSFA2 and HSFB2A, along with several HSPs, were upregulated by both H₂O₂ and PPA₁, while still further HSPs and other protein chaperones were common to H₂O₂ and both RES (PPA₁ and OPDA), clustering within profile S. Expression of HSP genes was notably insensitive to JA. Similarly, GST and UGT enzymes (profile R) were abundant among RES-responsive genes and were, to a lesser extent, also induced by H₂O₂, while very few were JA targets. A majority of the light stress-responsive UGTs induced by OPDA and PPA₁ are linked to secondary metabolism involved in detoxification and defence [54–56].

These results show that a general abiotic stress response (figure 3, category 1), characterized by strong upregulation of molecular chaperones and heat shock factors, can be induced by both H₂O₂ and RES-type oxylipins, although the induction of iron chaperones appears to be H₂O₂ specific. A more targeted stress response through enzymatic detoxification pathways (figure 3, category 2) is also triggered by both ROS and RES through upregulation of GST and UGT gene expression, with RES apparently being far more effective inducers of these pathways. Partial overlap in gene expression profiles of ROS and RES has been attributed to their shared tendency to react with thiol groups, which can have a similar impact on redox-sensitive proteins and the cellular redox buffer GSH [52]. On the other hand, the unique characteristics of RES oxylipins, such as the unsaturated ketone moiety and the tendency to form covalent adducts with nucleophiles [57–60], are in line with induction by PPA₁ and OPDA of a potent detoxification response comprising GST and UGT enzymes. Notably, the ANAC102 transcription factor was found in both the PPA₁ and OPDA sets of HL-induced genes, but was not among H₂O₂ or JA sets. ANAC102 was recently implicated in HL-responsive signalling induced by β-CC, which is also a RES [61], perhaps indicating a common RES signalling pathway involving ANAC102 and the redox-active type II TGA transcription factors [52,59,61]. However, substantial variation in gene targets of different RES [25,42,59,62–65] demonstrates a considerable degree of specificity between individual RES signalling pathways.

Seventeen CYP450 enzymes were differentially expressed in the current study and were distributed across profiles R and G in the PPA₁, OPDA and/or JA sets. CYP450 genes were particularly abundant in the OPDA-responsive set, comprising almost 5% of the total OPDA-induced genes. Conversely, the H₂O₂-responsive set contained only a single CYP450 gene (CYP81D8; profile R). CYP450s identified in the current study are associated with metabolism of terpenoids, fatty acids, glucosinolates and various hormones (figure 3, category 3). CYP94B1, B3 and C1, found here to be OPDA-specific in light stress/recovery, have been linked to catabolism of JA-isoleucine, the active form of JA [66,67], suggesting a role for OPDA in suppression of JA signalling. As discussed above, most of the genes encoding OPDA/JA biosynthesis enzymes (category 4) were upregulated in the current experiment, and these were found in the overlap of JA- and OPDA-responsive genes (figure 2). The MYC2/JIN1 transcription factor, which is a central regulator of JA-induced genes, as well as most of the JAZ-type repressors of JA-responsive transcription (category 5; reviewed in [51,68]), was also common to both OPDA and JA sets. This observation indicates that JA response is simultaneously propagated and repressed in response to HL stress, through concomitant upregulation of both the biosynthesis pathway and antagonistic transcriptional regulators. Numerous R2R3-type MYB transcription factors were also induced by OPDA and/or JA; however, they were most prominent in the JA set, making up over 10% of the JA-responsive genes found to be sensitive to HL and recovery in the current experiment. MYBs identified in both OPDA and JA sets are involved in the regulation of anthocyanin and glucosinolate biosynthesis (category 6), while many JA-specific MYBs have been implicated in cross-talk between JA and other hormones such as gibberellin (GA), auxin and brassinosteroids [69,70]. Similarly, DELLA transcription factors RGL1, RGL2 and RGL3, which repress GA signalling (reviewed in [23]), occurred in the OPDA (RGL3) and JA (RGL1–3) sets. The TPS03 and TPS10 enzymes involved in monoterpene synthesis (reviewed in [71]) were specific to JA.

These results suggest that signalling via both RES and non-RES oxylipins regulate a variety of secondary metabolite biosynthesis pathways (category 3) in response to light stress and recovery. Flavonoid-related enzymes appeared to be more responsive to RES-type oxylipins (i.e. OPDA and PPA₁), while glucosinolate, terpene and brassinosteroid metabolism were found to be regulated by enzymatic oxylipins (i.e. JA and OPDA). Similarly, enzymes and transcription factors involved in OPDA and JA biosynthesis and signalling regulation were conspicuously insensitive to the radical-derived PPA₁, demonstrating specificity to the enzymatic oxylipins. The overlapping signalling effects of exogenously applied JA and OPDA are difficult to separate because OPDA is a precursor for JA. Another complicating factor is that both hormones promote the expression of the entire OPDA/JA synthesis pathway (see results above), theoretically meaning that genes induced after application of exogenous JA may in fact be responding to upregulated endogenous biosynthesis of OPDA. However, the current work identified OPDA- and JA-specific genes and expression profiles. The profile S expression cluster was prominent in the OPDA-responsive set of genes and missing from the JA-responsive set (figure 1), which shows that JA does not activate abiotic stress-responsive gene expression, and also indicates that any endogenous OPDA induced in response to exogenous methyl JA does not activate profile S expression. Conversely, many light stress-induced genes found in the current study to be JA-specific encode factors involved in developmental and metabolic processes that are known to be responsive to JA and independent of OPDA [43,72–75].

4. Interactions between photosynthesis signalling pathways

The expression kinetics of different gene profiles identified here during HL stress and recovery (figure 1) illustrates the complex nature of chloroplast signalling pathways and dynamic regulation of target processes, both during and after stress. Rapid induction of so-called ‘abiotic stress response’ genes (profile S) upon HL exposure can be attributed to increased concentrations of either H₂O₂ [13,17] or RES [16], or both, and the subsequent impact on the redox state of the chloroplast and the wider cell [52,59,76]. Downregulation of these genes upon cessation of stress occurs through targeted transcript decay [11].

In contrast with the rapidly responding profile S genes, profile R was most strongly expressed during stress recovery. Profile R genes, which are largely involved in secondary metabolism, glutathionylation, and transcriptional and hormone regulation, have been shown in numerous studies to be upregulated under prolonged HL stress [16,22,77,78]. Nonetheless, the current study shows that the expression of this profile was not simply time dependent, exemplified by the repression of profile R genes at time-point 4 and induction at time-point 8, both of which correspond to 2 h after initiation of HL stress (figure 1). Instead, profile R expression appears to be truly recovery responsive. The prevalence of profile R expression in JA- and OPDA-responsive genes suggests a relationship with ¹O₂-type signalling that may be triggered by an increase in excitation and/or reduction pressure on PSII after the relaxation of NPQ (reviewed in [79]), and/or the downregulation of redox-regulated stromal metabolism (reviewed in [80]) upon HL-to-GL transition. However, profile R-type expression was also evident, to a lesser extent, in PPA₁ and H₂O₂-responsive gene sets (figure 1), which have no clear links with ¹O₂-type signalling or enzymatic oxylipin metabolism. This overlap may be due to partially overlapping impacts of ROS and RES on redox-sensitive signalling networks [52].

An apparent inverse correlation between profile R and profile S expression may further indicate an interaction between these signalling pathways, potentially suggesting that upregulation of heat shock factors, chaperones, ROS scavengers and/or ferritins can actually suppress secondary metabolism and hormone signalling. Indeed, increased cellular H₂O₂ correlates with repressed expression of secondary metabolite biosynthesis genes and decreased abundance of secondary metabolites [15,81,82]. Conceivably, stress-induced protein chaperone, protease and antioxidant activities may neutralize toxic RES and RES-modified proteins, thereby inhibiting oxylipin signalling during stress, but not during recovery when these general protective processes are rapidly downregulated [11]. An additional facet of the interaction between stress- and recovery-responsive signalling might involve detoxification of RES by glutathionylation, following which, the slow release from GSH adducts may activate oxylipin signalling pathways [52]. Notably, oxylipin signalling is well known to interact with other signalling pathways, especially with GA signalling through antagonistic interaction between the JAZ and DELLA transcription suppressors [23]. Although JAZ and DELLA genes were upregulated by HL stress (see above), a prominent role for GA signalling was not evident in analysis of the genes induced by HL and recovery.

5. Concluding remarks

Photosynthetic redox homeostasis operates as a sensing and signalling mechanism allowing a plant to rapidly process and respond to changes in environmental conditions. Analysis of the Arabidopsis genes affected by light stress and recovery indicates that H₂O₂ and oxylipins are prominent primary signalling factors in abiotic stress response. Enzymatic oxylipin synthesis induced by ¹O₂ production can be traced to redox balance at PSII, while H₂O₂ production is associated with PSI acceptor side limitation. However, substantial overlap between the targets of H₂O₂ and radical-induced oxylipin signalling indicates that these pathways may not be entirely mutually exclusive (e.g. see [53,83]), and the contribution of H₂O₂ to ^•OH radical production should be considered in this context. Common targets of ROS and RES signalling identified in the current study support a pathway operating through changes in the cellular redox state, paradoxically activating processes that mitigate the toxicity of these reactive species (figure 3). On the other hand, JA signalling is quite specific to regulation of metabolic processes, with substantial signalling overlap with its precursor OPDA. Importantly, the results of this study do not exclude other metabolites, hormones and pathways as vital components of chloroplast retrograde signalling. While redox imbalance may be the primary cause of photosynthetic ROS formation, many other retrograde signalling factors are also sensitive to ROS from the chloroplast or other sources.

Rapid upregulation of so-called ‘abiotic stress response’ genes upon changes in photosynthetic redox balance is intuitively sensible; however, the activation of numerous secondary metabolic pathways during recovery is more difficult to understand. Upregulated metabolite production can provide a sink for carbon assimilated during periods of high photosynthetic activity [84]. Although JA is most commonly associated with reproductive development and response to herbivory, an important role for JA in regulating plant growth and abiotic stress response continues to emerge [85,86]. The current work shows that oxylipins are major components of chloroplast retrograde signalling during stress recovery, implying an important role in allocation of resources between defence and growth.

Data accessibility

This article has no additional data.

Authors' contributions

P.J.G. performed analyses and wrote the manuscript. E.M.A. contributed to discussions and writing.

Competing interests

We declare we have no competing interests.

Funding

This study was funded by Suomen Akatemia project 26080431, grant nos. 303757 and 307335.