From wild-type indica variety Longtepu B (LTB), we previously developed a xantha mutant line, Huangyu B (HYB) via gamma ray mutagenesis (Zhou et al., 2006). HYB carries an epi-mutant allele of OsGUN4 (hereafter gun4^epi), which downregulates the expression of OsGUN4 to ∼ 0.5% that of LTB (Li et al., 2014a). Further, through gamma ray irradiation, our previous study also succeeded in generating and identifying a few OsGUN4 mutant alleles in the background of a japonica line GS113 by crossing HYB to GS113 M₂ plants (Li et al., 2014a). Among them, the gun4-1 allele is a complete knockout mutation because its translation initiation codon (ATG) is completely deleted (and thus no OsGUN4 protein is expected to be translated) (Li et al., 2014a). In the present study, the F₂ seeds of the F₁ hybrid plant of GS113 (gun4-1)/HYB (gun4^epi) were used for various investigations, together with LTB and HYB (Supplementary Figure 1).

For easy management, germinated seeds were first grown in plastic trays with soils taken from paddy fields (surface, not sterilized) of Zhejiang University Experimental Farm at 30°C for 7 days in growth rooms with LED (light emitting diode) under 16 h/8 h light/dark, at low light (LL, 100 μmol photons m^–2 s^–1) or high light (HL, 1000 μmol photons m^–2 s^–1) conditions. For assessing the effect of light intensity on seedling growth and gene expression, seven-day old seedlings grown on soil were transferred to 1× MS liquid medium (Murashige and Skoog, 1962) and grown under either LL or HL conditions for another 7–28 days. For assessment of H₂O₂ effect, seedlings that had been grown on soil (7 days) and 1× MS medium (21 days) under LL were transferred to 1× MS liquid medium with or without 1 mM H₂O₂ and grown for another 7 days under LL. Liquid medium was replaced every 3 days.

For genotyping of the F₂ seedlings, genomic DNA was extracted from leaf segments and subject to high resolution melting (HRM) curve analysis with primers of 5′-TTACCGG CAGGCCGACGAGA-3′ and 5′-TGCCCAGGAGCTGTGTCC CT-3′ according to Tan et al. (2016).

For investigation of diurnal expression of OsGUN4, seedlings of LTB and Nipponbare (japonica) were grown in soil under HL conditions (16 h/8 h light/dark) for 35 days after germination (DAG), and their leaves were sampled every 2 h for RNA extraction within a 24-h period.

The concentrations of chlorophylls were determined using 0.3 g fresh leaf tissues following a published method (Peter and Grimm, 2009). For measurement of PPIX, Mg-PPIX and protochlorophyllide (pchlide), 0.3 g fine powder of liquid nitrogen-ground leaves were immediately mixed with pre-cold alkaline acetone containing 0.1 N NH₄OH (9:1; v/v), and centrifugated at 16,000 g for 5 min. Subsequently, the supernatants were collected for extraction of PPIX (Papenbrock and Grimm, 2001), Mg-PPIX (Papenbrock and Grimm, 2001), and pchlide (Koski and Smith, 1948). The contents of PPIX, Mg-PPIX and pchlide were determined with commercial enzyme-linked immunosorbent assay (ELISA; Engvall and Perlmann, 1971) kits (Jiangsu Jingmei Biotechnology Co., Ltd., Yancheng, China) following the manufacturer’s instructions.

The concentration of singlet oxygen (¹O₂) was determined using the SOSG (singlet oxygen sensor green) method (Hideg et al., 2002). Briefly, the liquid nitrogen homogenized power of 0.3 g leaf samples were suspended with pre-cold lysis buffer (GENMED Scientifics Inc., Shanghai, China), and then immediately centrifuged at 4°C for 10 min (5000 g). Subsequently, SOSG (GENMED Scientifics Inc.) was added to the supernatant to a final concentration of 10 μM. The fluorescence spectra were detected at excitation of 485 nm and emission of 520 nm using a fluorescence spectrophotometer (359S, Lengguang Tech., Shanghai, China).

The hydrogen peroxide (H₂O₂) content was measured according to the method of Rokebul Anower et al. (2017) with minor modifications. 0.3 g leaf tissues were ground in liquid nitrogen and then immediately mixed with pre-chilled Krebs–Ringer phosphate (145 mM NaCl, 5.7 mM sodium phosphate, 4.86 mM KCl, 0.54 mM CaCl₂, 1.22 mM MgSO₄, 5.5 mM glucose, pH 7.35). Subsequently, the homogenate was centrifuged at 12,000 g for 20 min at 4°C, and then the supernatants were incubated with 50 μM Amplex^® Red reagent (10-acetyl-3,7-dihydrophenoxazine, Thermo Fisher Scientific, Shanghai, China) and 0.1 U/mL horseradish peroxidase (HRP, Thermo Fisher Scientific, Shanghai China) at 37°C for 10 min away from light. The fluorescence was measured at 530 nm under irradiation of 590 nm light for excitation by using a VersaFluor Fluorometer (Bio-Rad, Hercules, CA, United States) and calculated based on a standard curve constructed using known concentrations of H₂O₂.

The total ROS concentration was determined according to the method of Joo et al. (2005) with modifications. In brief, for assessment of ROS in seedlings, the above-ground part of seedlings was weighed, quickly frozen and ground into powder in liquid nitrogen, and immediately suspended in 10 mM Tris-HCl buffer (pH 7.3) for extraction. The mixture was centrifuged at 16,000 g for 5 min, and after collecting the supernatant, the pellet was suspended with buffer and centrifuged for another 5 min. For assessment of ROS in protoplasts, the transfected protoplasts containing 4 × 10⁴ protoplasts/μL were added with 10 mM H₂DCFDA (Thermo Fisher Scientific, Waltham, MA, United States) at 4°C in DMSO.

The supernatant was used for ROS measurement by adding 10 mM H₂DCFDA (Thermo Fisher Scientific, Waltham, MA, United States) at 4°C in DMSO. The fluorescence was measured at 530 nm under irradiation of 488 nm light for excitation by using a VersaFluor fluorometer (Bio-Rad, Hercules, CA, United States) with six biological replicates. The ROS content, expressed as relative fluorescence units, was obtained from six successive measurements and is given as units per milligram of fresh weight.

Total RNA was extracted from fresh leaves or protoplasts using the Qiagen Spin Plant RNA Mini Kit (Qiagen, Hilden, Germany). cDNA was reverse transcribed from 1 μg total RNA with oligo-dT₁₈ Primer using GoScript^TM Reverse Transcription System Kit (Promega, WI, United States) according to the manufacturer’s instructions. Quantitative real-time PCR (qRT-PCR) was performed using a SYBR Green GoTaq^® qPCR Master Mix containing ROX as internal control (Promega, WI, United States). All qRT-PCRs used the following cycling conditions: 10 min at 95°C, 40 cycles of 30 s at 94°C, 30 s at 55°C and 60 s at 72°C. Relative mRNA abundance was analyzed with the rice Ubiquitin gene as internal reference using the 2^–ΔΔCt method (Livak and Schmittgen, 2001). Primers used for qRT-PCR are listed in Supplementary Table 1.

Total RNA of LTB and HYB was extracted from fresh leaf tissues of 35 DAG seedlings using the Qiagen Spin Plant RNA Mini Kit (Qiagen, Hilden, Germany). cDNA libraries were constructed with oligo-dT₁₈ Primer using GoScript^TM Reverse Transcription System Kit (Promega, WI, United States) according to the manufacturer’s instructions, and were subsequently sequenced on an Illumina Hiseq 2000 platform (Beijing Novogene Bioinformatics Technology Co., Ltd., Beijing, China). For mapping, the raw reads were cleaned by removing adapter sequences using the Cutadapt (v1.11) software and then aligned to the reference genome sequences (IRGSP-1.0¹, 09/17/2016) by using the Tophat v2.0.9 program with E-value < 10^–5 as cut-off point (Trapnell et al., 2012). For detection of differentially expressed genes (DEGs), the DESeq package (ver 2.1.0) was used with a false discovery rate (FDR) ≤ 0.05 and the absolute value of the log₂ (fold change ≥ 2) with RPKM ≥ 1 as the threshold to determine significant differences of gene expression. Gene ontology (GO) enrichment was conducted according to PANTHER Overrepresentation Test (released 20210224) by using GO Ontology database (released 02/01/2021) using Oryza sativa database as a reference list, and was visualized with ggplot2 R package. KEGG pathways were analyzed with a FDR ≤ 0.05 as significant levels of differential expression. DEGs potentially related to ROS homeostasis were identified and subjected to RT-q PCR analysis using gene-specific primers (Supplementary Table 1).

For construction of OsGUN4 complementary expression vector, a DNA fragment encompassing the full open reading frame of OsGUN4 and its upstream (2447 bp) and downstream (518 bp) sequences were amplified from genomic DNA of Nipponbare using primers OGUN4F and OGUN4R (Supplementary Table 1), which was cloned into a pUCm-T vector (Sangon Biotech, Shanghai, China). The recombinant pUCm-T vector was then digested with NcoI/HindIII and the target fragment was cloned into binary vector pCAMBIA 1301 using the same restriction enzymes. The ultimate vector is named pOsGUN4-C.

For construction of OsPRX39 overexpression vector, the full open reading frame of OsPRX39 (999 bp) sequences were amplified from genomic DNA of Nipponbare with the primers OPRX39F/OPRX39R (Supplementary Table 1), which was cloned into a pUCm-T vector (Sangon Biotech, Shanghai, China), and the recombinant vector was digested with NcoI/EcoRI. Subsequently, the target fragment was simultaneously cloned into the digested binary vector of pCAMBIA 1301 (by using NcoI/EcoRI), the OsPRX39 overexpression vector was thus constructed and named p1301-Ubi-PRX39.

For silencing of OsGUN4, the hairpin RNA (hpRNA) down-regulating system was used and a silencing vector was constructed according to Li et al. (2014b). An OsGUN4 fragment (+562 to +761 bp downstream of the ⁺¹ATG), containing part of the coding sequence and a partial 3′ UTR was amplified using primers RGUN4F and RGUN4R (Supplementary Table 1) and cloned into a pUCm-T vector. Two fragments containing the OsGUN4 fragment were cloned from the recombinant pUCm-T vector by double restriction with NotI/BamHI and XhoI/KpnI, respectively. They were then sequentially ligated into pBSSK-IN in the sense and antisense orientations. The OsGUN4 hairpin structure in the recombinant pBSSK-IN was introduced into pCAMBIA1301-35S using restriction enzymes KpnI and SacI. The recombinant pCAMBIA1301-35S vector is named pOsGUN4-I.

Using the slightly modified enzymolysis method (Zhang et al., 2011), protoplasts of LTB and HYB were extracted from 35 DAG seedlings grown under LL conditions and used for transient expression analysis. They were transformed with the OsGUN4 complementation/overexpression or silencing vectors with PEG-mediated transformation according to Bart et al. (2006). Briefly, 100 μL solution containing 4 × 10⁴ protoplasts/μL was transformed with 10 μg plasmids and incubated in W5 solution (2.4 × 10⁴ protoplasts/μL) in darkness for about 1 h. The transformed protoplasts were then cultured at low light or high light for another 6 h before RNA extraction.

All experiments were performed with six biological replicates. Values are expressed as means ± standard deviations. Comparisons of data from different groups were analyzed using ANOVA test followed by the Tukey’s Multiple Comparison Test with P < 0.05.

To detect the daily variation of OsGUN4 expression, we first we investigated its relative expression in seedlings of indica variety LTB and japonica variety Nipponbare at different times during a 24-h period (Figure 1A). We observed that the expression of OsGUN4 was lower at night and quickly increased after day-break at 06 00 h. Two peaks of expression were observed at 10 00 and 14 00 h for both varieties. The expression gradually decreased in the afternoon and remained at a low level at night. These results suggest that the expression of OsGUN4 had the nature of diurnal oscillation, and the relatively reduced expression at the mid-day suggests that its expression is also subject to internal physiological changes.

To examine the responsiveness of OsGUN4 to light intensity, the abundance of OsGUN4 mRNA transcripts was examined for seedlings grown under LL and HL conditions at 7, 21, and 35 days after germination (DAG) (Figure 1B). When grown under LL conditions, the expression of OsGUN4 was low and hardly increased with plant age (Figure 1B and Supplementary Table 2). Under HL conditions, the OsGUN4 expression level significantly increased with seedling age, and was much higher than that of seedlings grown under LL conditions at all ages (Figure 1B). For example, at 7 DAG, the OsGUN4 expression level in seedlings under HL conditions was 1.53 that of seedlings under LL, while at 35 DAG, the relative value increased to 17.49. Thus, OsGUN4 expression was responsive to light intensity and increased with seedling age.

To examine its responsiveness to oxidative stress, the expression of OsGUN4 was examined in seedlings grown for 7 days in medium with 1 mM H₂O₂. The OsGUN4 expression level was significantly higher (8.8-fold) in seedlings treated with H₂O₂ than in the control plants (Figure 1C and Supplementary Table 2), indicating a response to oxidative stress.

Down-regulation of OsGUN4 is known to result in chlorophyll deficiency and reduced plant growth, but it was not known whether rice could survive a complete OsGUN4 knockout under different light conditions. During the germination stage, we observed that a high percentage of F₂ seeds derived from the gun4-1/gun4^epi F₁ plants could not develop into complete seedlings and died within 7 DAG (Figure 2 and Supplementary Figure 2A).

To distinguish the effect of the gun4^epi and gun4-1 mutations on seed germination and seedling growth, we performed an HRM-based genotyping (Supplementary Figure 1B). It was revealed that all the early dying germinated seeds belonged to the gun4-1/gun4-1 genotype, suggesting OsGUN4 is indispensable for rice seedling growth. The gun4^epi/gun4^epi and gun4^epi/gun4-1 F₂ seedlings were further grown under different conditions and their growth and ROS content were measured.

Most F₂ seedlings with the gun4^epi/gun4-1 genotype survived, but their growth was significantly slower and weaker than the WT and gun4^epi/gun4^epi ones, particularly under the HL conditions (Figure 2A; Supplementary Figure 2B; and Supplementary Table 3). Under HL conditions, the gun4^epi/gun4-1 F₂ seedlings were significantly shorter than the gun4^epi/gun4^epi ones (Figure 2B), suggesting that there is a dosage effect of the gun4^epi allele on seedling growth.

To assess the effect of OsGUN4 mutations on tetrapyrrole metabolism, the concentrations of chlorophyll (Chl) and three key tetrapyrrole metabolites was measured in 35 DAG seedlings grown under LL and HL conditions (Figure 3 and Supplementary Table 4). Under LL conditions, in consistent with phenotypical changes, although the concentration of Chl and tetrapyrrole metabolites was significant lower in the OsGUN4 mutants than the WT, the differences were very limited between mutants and the WT, with the maximum difference being observed for Chl (<30%, Figure 3D); No significantly differences were detected for any of the metabolites between the F₂ seedlings of gun4^epi/gun4^epi and gun4^epi/gun4-1 (Figure 3).

Under HL condition, the metabolite concentration differences became much more pronounced between the two mutants and the WT (Figure 3). PPIX is the tetrapyrrole metabolite that can be diverted into both heme (through Fe-branch) and Chl (through Mg-branch), its concentration was significantly higher (+60%) in the two mutants than in the WT (Figure 3A). However, the metabolites in the Mg-branch were significantly reduced in both mutants (Figures 3B–D). The two mutants had 89% of the Mg-PPIX concentration in the WT (Figure 3B). The reduction of pchlide and Chl contents was more dramatic; the F₂ seedlings of gun4^epi/gun4^epi and gun4^epi/gun4-1 had pchlide content only 63 and 56% that of the WT, and Chl content only 65 and 41% that of the WT, respectively (Figures 3C,D).

When WT seedlings were moved from LL to HL conditions, the concentration of PPIX, Mg-PPIX and Chl was significantly reduced, while that of pchlide significantly increased. In the two OsGUN4 mutants, the concentrations of all metabolites except PPIX significantly decreased (Figure 3). Thus, OsGUN4 mutations also changed the responsiveness of tetrapyrrole metabolism to light intensity.

To further examine whether the repressive effect of OsGUN4 mutations on seedling growth and tetrapyrrole metabolism was ROS-dependent, we measured the concentrations of total ROS, H₂O₂ and ¹O₂ in the F₂ and WT seedlings at different growth stages under LL and HL conditions. Significantly greater total ROS and H₂O₂ contents, but lower ¹O₂ content, was observed in the mutant F₂ seedlings than the WT line, irrespective of the growth conditions and mutant genotype (Figure 4 and Supplementary Table 5).

For 7 DAG seedlings, we were able to measure the total ROS level in all three types of mutant seedlings (Figure 4A) and observed that the total ROS level in the gun4-1/gun4-1 seedlings, irrespective of growth condition, was much higher than those in the WT and other two mutants. For example, under HL conditions, gun4-1/gun4-1 F₂ seedlings accumulated 9.18, 3.54, and 2.97 times as much total ROS than WT, gun4^epi/gun4^epi and gun4^epi/gun4-1, respectively (Figure 4A). Because gun4-1/gun4-1 F₂ seedlings did not survive beyond 7 DAG, ROS content could only be measured on seedlings of the other two mutants at 21 and 35 DAG. The gun4^epi/gun4-1 F₂ seedlings accumulated significantly higher levels of total ROS than the gun4^epi/gun4^epi ones. Comparatively, both types of mutant seedlings contained far more ROS than the WT at all time point, i.e., the gun4^epi/gun4^epi F₂ seedlings accumulated 2.7–3.3 times, under the LL conditions, and 2.6–3.5 times under the HL conditions (Figure 4A) more ROS than the WT (Figure 4A), respectively.

The total ROS level in the gun4^epi/gun4-1 F₂ seedlings increased with seedling age under both the LL and HL conditions. A similar trend was observed in gun4^epi/gun4^epi F₂ seedlings grown under the HL conditions (Figure 4A). At a given growth stage, significantly more ROS accumulated in seedlings grown under HL conditions than under LL conditions irrespective of genotype (Figure 4A).

To differentiate different types of ROS, the seedlings at 35 DAG were further assessed for the level of H₂O₂ and ¹O₂. Significantly higher levels of H₂O₂ were observed in mutant seedlings than that in the WT; the differences became more significant under HL conditions. The levels of H₂O₂ in gun4^epi/gun4^epi and gun4^epi/gun4-1 F₂ seedlings were 2.5- and 3.2-fold those of the WT, respectively (Figure 4B and Supplementary Table 6). Both gun4^epi/gun4^epi and gun4^epi/gun4-1 F₂ seedlings accumulated significantly lower levels of ¹O₂ (66–87%) compared to the WT, irrespective of growth conditions (Figure 4C and Supplementary Table 6). Furthermore, HL treatment induced an increase in ¹O₂ content by 25% in WT, while no significant change was detected in the mutants under any condition (Figure 4C).

To examine the effect of the gun4^epi mutation on overall gene expression, the transcriptomes of LTB and its epigenetic mutant HYB at 35 DAG were sequenced (Supplementary Tables 7, 8). Besides, a total of 468 differentially expressed genes (DEGs) were identified between HYB and LTB, with 203 genes being up-regulated and 265 down-regulated in HYB. GO analysis showed genes involved in peroxidase activity, response to oxidative stress were down-regulated (Figure 5). Further examination revealed that eight peroxidase (PRX)-encoding genes were among the down-regulated genes.

We already demonstrated that transcription of OsGUN4 is responsive to high HL conditions and H₂O₂ supplement (Figures 1B,C), this type of response of OsGUN4 expression is no longer present in HYB seedlings: OsGUN4 expression is extremely low irrespective of treatment, and no significant differences were observed for OsGUN4 transcript abundance in HYB and the two F₂ mutant seedlings (Figure 6A). These results indicated that the OsGUN4 mutations not only dramatically reduced its transcription but also abolished its response to oxidative stresses caused by HL and H₂O₂. Similar performance of the mutant F₂ seedlings to HYB indicated that the mutation effects were independent of genetic background.

qRT-PCR analyses demonstrated that the 8 PRX genes downregulated in HYB were all responsive to HL and exogenous H₂O₂ treatment in the WT (Figure 6 and Supplementary Table 9). The shift from LL to HL resulted in significant increases of transcripts abundance, i.e., from 2.96-times (OsPRX89, Figure 6H) to 8.54-times (OsPRX86, Figure 6G); Exogenous H₂O₂ treatment also generated similar increases (but slightly less than HL shift).

The transcriptional response of the 8 PRX genes to HL and H₂O₂ treatment seemed to be either significantly reduced or abolished in HYB and the mutant F₂ seedlings (Figures 6B–I). Three genes (OsPRX-L, OsPRX 11 and OsPRX39) were no longer responsive to HL and H₂O₂ treatment (Figures 6B–D), and the remaining genes became significantly less responsive (Figures 6E–I).

To test whether the expression differences of the 8 PRX genes in WT and mutant seedlings are truly due to OsGUN4 mutation, LTB protoplasts were transformed with the OsGUN4 interference vector pOsGUN4-I and HYB protoplasts were transformed with the OsGUN4 complementary vector pOsGUN4-C. The transformed protoplasts were subjected to LL and HL treatment and transcription of OsGUN4 and the 8 PRX genes were then analyzed by q RT-PCR.

The expression pattern of OsGUN4 in protoplasts seemed similar to that in seedlings, i.e., its expression was extremely low in HYB and did not increase after HL treatment, in sharp contrast to what was observed in WT (Figure 7A). Importantly, we succeeded in repressing OsGUN4 expression by transforming LTB protoplasts with a silencing vector (GUN4_RNAi) and restoring its expression by transforming HYB protoplasts with a complementation vector (GUN4^CE; Figure 7A).

The transcription of OsGUN4 and the 8 PRX genes was significantly affected by light intensity and OsGUN4 genotype (Supplementary Table 10). The native WT OsGUN4 promoter was used for the construct of both vectors, hence OsGUN4 expression in HYB protoplasts transformed with the complementation vector was also responsive to HL. In WT protoplasts transformed with OsGUN4-RNAi, transcription of OsGUN4 was downregulated to the level similar to that in HYB at both LL and HL, while HYB protoplasts transformed with GUN4^CE, the expression level of OsGUN4 was restored to nearly the same level of WT at LL, and to higher levels at HL (Figure 7A).

In both WT and HYB_GUN4^CE protoplasts, transcription of all the 8 PRX genes was significantly increased after HL treatment. In HYB protoplasts transformed with GUN4^CE, the transcription level of OsPRX genes was similar to or significantly higher than that of WT at LL, but did not restore to the same level as in the WT except OsPRX11 at HL (Figures 7B–I). Thus, the transcriptional response of PRX genes to light fluence shift was restored to different degrees in OsGUN4-complemented HYB protoplasts.

In both HYB and GUN4_RNAi protoplasts, either no such increase (OsPRX-L, OsPRX11 and OsPRX39) or limited increase of transcription was observed for the other genes, similar to the pattern observed in seedlings. In LTB protoplasts transformed with the interference vector, the expression level of OsPRXs was reduced to 10–44% that of untransformed protoplasts and was very similar to that observed in HYB, their response to HL was either abolished (OsPRX-L, OsPRX 11 and OsPRX39, Figures 7B–D) or significantly repressed (OsPRX65, OsPRX 81 and OsPRX86, OsPRX 89 and OsPRX125, Figures 7E–I).

These results demonstrate that the protoplast expression system worked well for interference and complementation of OsGUN4 mutation, and that there is a cause-and-effect relationship between OsGUN4 mutation and altered expression and response of the 8 PRX genes to a light fluence shift and H₂O₂ treatment.

To examine whether OsGUN4 complementation and silencing affect ROS accumulation, the total ROS level in various types of protoplast was measured. Under LL conditions, no significant differences were observed among protoplasts with different OsGUN4 genotypes, however, significant differences were observed after transferred to HL conditions (Figure 8A and Supplementary Table 10). Consistent with the results in seedlings (Figure 4A), a significantly greater total ROS level was observed in HYB protoplasts than its WT parent (Figure 8A). Importantly, the total ROS level was increased in GUN4_RNAi LTB protoplasts (to a level even significantly higher than that of HYB), while the total ROS level was reduced to the level of the WT in GUN4 complemented HYB protoplasts (gun4^epi_GUN4^CE).

To further examine the effect of repression of PRX genes on ROS accumulation, we overexpressed OsPRX39 in HYB protoplasts (gun4^epi-PRX39^OX). The OsPRX39 expression level in gun4^epi_PRX39^OX protoplasts was high under both LL and HL conditions because it was driven by the Ubi promoter, and it showed a 2.73-fold increase as compared to that of WT under HL (Figure 8B and Supplementary Table 10). The total ROS level in gun4^epi_PRX39^OX protoplasts was significantly reduced compared to the HYB protoplasts, and was even significantly lower that the WT LTB (Figure 8B).

We have made three significant observations:

(i) The expression of OsGUN4 is sensitive to light intensity and oxidative stress (Figures 1, 7A). In photosynthetic organisms, the major sources of ROS production are PSI and PSII reaction centers (Asada, 2006; Møller et al., 2007), which is consistent with our observation that significantly more ROS is accumulated in WT seedlings grown under HL than LL conditions (Figure 7A). As we previously reported, there is an anti-oxidative element (ARE) in the promoter of OsGUN4 (Li et al., 2014a), hence the observation that its transcription is responsive to oxidative stress was expected. Here, we further argue that the response of OsGUN4 transcription to light intensity and oxidative stress might be governed by the same mechanism or by overlapping mechanisms.

(ii) OsGUN4 transcription was low at night and increased rapidly at daybreak, and showed oscillation during the daytime (Figure 1A). The low expression at night is understandable because, at night, no photosynthesis takes place hence the ROS level also is low, while photosynthesis is active during the light period and the ROS level consequently increases. It is well-known that photosynthesis is reduced at midday in many crop plants including wheat (Sawada, 1978), rice (Ishihara and Saito, 1987), cotton (Pettigrew et al., 1990), maize (Tadashi and Theodore, 1999), and citrus (Hu et al., 2007). In a previous study, we observed the midday photoinhibition under natural field condition in the WT variety LTB (but not in HYB), with significantly reduced values of Fv/Fm and ϕPSIIaround 12 00 h (Wu et al., 2007). The underlying mechanism leading to midday depression of photosynthesis has been investigated from various aspects and at different levels, such as its relation to light-induced chlorophyll fluorescence changes (Wang et al., 2021) and chloroplast positioning (Maai et al., 2020), however, it is not yet fully understood. We observed a decrease in OsGUN4 expression around midday in both WT varieties Nipponbare and LTB grown under HL conditions (Figure 1A), surprisingly similar to the reported midday depression of photosynthesis in LTB (Wu et al., 2007) and other varieties (Ishihara and Saito, 1987). It would be interesting to study whether depressed photosynthesis leads to a reduction in ROS content and subsequently to a downregulation of OsGUN4 transcription, or vice versa.

(iii) We observed a significant increase in total ROS content in OsGUN4 mutants. It is very interesting to note that while expression of OsGUN4 is responsive to ROS in WT LTB, its mutation caused elevated ROS accumulation in HYB. Elevated accumulation could result from either increased production or reduced scavenging of ROS. Because the overall photosynthetic activity in LTB is known to be higher than in HTB than in HYB, particularly before the midday depression of photosynthesis (Wu et al., 2007), and there was indeed a lower ¹O₂ content, particularly under HL conditions (Figure 1C), therefore we argue that the it is very likely that the ROS scavenging capacity is impaired due to OsGUN4 mutations.

Plant cells deploy a delicate system to prevent or limit ROS production (Holt et al., 2004) and exquisite antioxidant system to scavenge excessive ROS in chloroplast (Gupta et al., 2015) via non-enzymatic (Havaux et al., 2007) and enzymatic reactions (Dietz, 2016).

In the present study, we identified 8 peroxidase encoding genes that were downregulated in HYB as compared with LTB, and further demonstrated that their transcriptional response to oxidative stress was abolished or repressed in OsGUN4 mutant seedlings (Figure 6). Of particular importance is that the transcriptional repression of the 8 PRX genes was also observed in the mutant F₂ seedlings, because it demonstrates that the differences between HYB and LTB is very likely due to the OsGUN4 mutation. We further established the cause-and-effect relationship between OsGUN4 mutation and its repression effect on transcription by silencing and complementing OsGUN4 in protoplasts and determination of expression of the 8 PRX genes (Figure 7).

Peroxidases are important ROS scavengers in plants. It is therefore predictable that OsGUN4 mutations such as downregulation (gun4^epi) and knockout (gun4-1) would significantly impair the ROS scavenging system and result in ROS accumulation. Consistent with this prediction, the content of total ROS and H₂O₂ was significantly higher in the mutants than the WT, particularly under HL conditions (Figures 4A,B, 8).

The gun4^epi allele dramatically downregulates its transcription, but would still retain a very low level of OsGUN4 protein, while the gun4-1 allele is expected to completely abolish its translation, which means that it would produce no OsGUN4 protein. The expected different mutational effects on OsGUN4 expression was indeed further manifested at both the biochemical and phenotypical level. The gun4-1/gun4-1 seeds could not germinate normally and conferred early death (Supplementary Figure 2), which is very likely due to the very high ROS accumulation t (Figure 4A). Moreover, significantly more ROS and H₂O₂ were accumulated in gun4^epi/gun4-1 seedlings than in gun4^epi/gun4^epi ones (Figures 4A,B), further supports the more severe effect of gun4-1. The different effects on ROS accumulation were well manifested on seedling growth: gun4^epi/gun4^epi grew much better than gun4^epi/gun4-1 seedlings (Figure 2).

Genomes Uncoupled 4 was identified as a regulator of chlorophyll synthesis and intracellular signaling in Arabidopsis (Larkin et al., 2003; Larkin, 2016). However, most of yet limited number of studies on GUN4 and its orthologs have been focusing on its regulatory function in tetrapyrrole biosynthesis, with only a few studies examined its role and mechanism in retrograde signaling.

In Chlamydomonas reinhardtii, Brzezowski et al. (2014) demonstrated that light-or dark-grown gun4 mutant accumulates high levels of PPIX, and fails to downregulate mRNA levels of the tetrapyrrole biosynthesis and the photosynthesis-associated nuclear genes (PhANGs). While they proposed that GUN4 functions in ‘shielding’ PPIX, and most likely Mg-PPIX, by reducing reactivity with O₂, and further suggested that “GUN4 seems to be involved in sensing elevated levels of these photoreactive tetrapyrrole intermediates, and contributing to ¹O₂-mediated retrograde signaling.” However, Tabrizi et al. (2016) argued for the opposite mechanism in that GUN4-PPIX is a ¹O₂ generator and proposed that the light-dependent ¹O₂ generation from GUN4-PPIX is the first step in retrograde signaling from the chloroplast to the nucleus. While the two studies suggested quite different roles of GUN4 in ¹O₂ generation, they suggested the same route of retrograde signaling, that is, EXECUTER 1 and/or EXECUTER 2 sense ¹O₂ level and regulate transcription of PhANGs.

In rice, we previously observed that downregulation of OsGUN4 results in deregulated transcription of PhANGs and altered tetrapyrrole biosynthesis, as well as decreased ¹O₂ production and down-regulated ¹O₂-dependent retrograde signaling (Li et al., 2017). In the present study, we observed the same concentration changes of PPIX, Mg-PPIX and pchlide (Figures 3A–C) and ¹O₂ (Figure 4D) in both of gun4^epi/gun4^epi and gun4^epi/gun4-1 seedlings as in HYB. These results are consistent with the findings that OsGUN4 plays a role by promoting ¹O₂ generation and its deficiency is expected to impair ¹O₂-dependent retrograde signaling.

Regarding the role of GUN4 in retrograde signaling, it has so far limited to the ¹O₂-dependent, EXECUTER 1 and/or EXECUTER 2 mediated transcription regulation of PhANGs. Our present study for the first time extended it to the regulation of PRX genes (Figure 9). While further studies are needed to establish the regulatory chain from GUN4 to PRX genes, we here propose that regulation of genes responsive for ROS scavenging, such as the PRX genes studied in the present study, is an essential function of the retrograde signaling pathway. We further envision the following working module: In WT plants, the ¹O₂ level increases with the progress of photosynthesis, and activate EXECUTER 1 and/or EXECUTER 2 sense ¹O₂ level and other transcription factor (TF) genes, such as BAP1. Some of the activated TFs repress the transcription of PhANGs after photosynthesis reaches to a high level, for example late morning, such that the photosynthesis is prohibited at midday. On the other side, some of the TFs activate genes including PRX genes, and the peroxidases encoded by them together help scavenge ROS thus maintain a low ROS level. However, in gun4 mutants, the ¹O₂ level remains low due to lack of functional GUN4 hence the ¹O₂-mediated retrograde signaling pathway is not activated, and consequently, (1) a relatively high level of photosynthesis is maintained, which continuously generates ROS, (2) PRX genes are not activated hence ROS scavenging capacity remains low, which together result in the accumulation of excessive ROS (Figure 9). More studies are needed to validate our hypothesis and reveal the players leading to the activation of PRX genes downstream the ¹O₂-mediated signaling pathway.