In plants, reactive oxygen species (ROS) are produced continuously as byproducts of multiple metabolic pathways and are localized in multiple cellular compartments. Under normal conditions, ROS are scavenged by various antioxidative defense systems1,2. The equilibrium between the generation and the scavenging of ROS may be perturbed by many environmental stresses. To date, research on the physiological activities of ROS in plant cells has mainly been restricted to hydrogen peroxide and superoxide, which are released upon abiotic and biotic stress and work as signaling molecules to regulate various processes, such as stomatal behavior, programmed cell death, and pathogen defense1,2.
In photosynthetic organisms, excited chlorophylls or tetrapyrrole intermediates can stimulate the formation of singlet oxygen (1O2) upon illumination; a highly toxic molecule that in addition to its damaging nature acts as a crucial signaling molecule. 1O2 signaling has been shown to interact with the signal cascades of other ROS, lipid hydroperoxide-derived reactive electrophile species, and oxidized carotenoids, which may induce programmed cell death3. 1O2 is responsible for more than 80% of non-enzymatic lipid peroxidation4. The FLU (FLUORESCENT) protein is a nuclear-encoded plastid protein that plays a key role in the negative-feedback control of chlorophyll biosynthesis5. Inactivation of this protein in the flu mutant leads to the over-accumulation of free protochlorophyllide (Pchlide), which may work as a potent photo-sensitizer. Thus, the flu mutant generates 1O2 in plastids in a controlled but noninvasive manner. Immediately after the release of 1O2, mature flu plants stop growing, whereas young seedlings bleach and die, when grown under dark / light cycles5. Inactivation of the plastid proteins EXECUTER1 (EX1) and EXECUTER2 (EX2) attenuate the extent of 1O2-induced up-regulation of nuclear gene expression, thereby reversing the growth arrest and seedling lethality of the flu mutant under light / dark cycles6,7. Despite this research progress, little is known about how and where the 1O2 signals are sensed and transmitted to the nucleus, and no nuclear signaling factor has been identified. In this report, we found that the inactivation of a master Apetala 2 (AP2)–type transcription factor ABI4 (ABSCISIC ACID INSENSITIVE 4) was sufficient to abrogate 1O2-mediated signaling, resulting in flu mutant survival.
ABI4 mutation rescued the lethal phenotype of flu
The 1O2 signal is a type of plastid retrograde signal2,3,6,7, in which plastid GUN1 (GENOMES UNCOUPLED 1) protein and nuclear ABI4 may be involved8. A previous study demonstrated that the gun1 mutation did not prevent the 1O2-mediated bleaching and cell death response of the flu mutant grown under light / dark cycles9. Thus, the possible role of ABI4 in 1O2 signaling was investigated in this study. Two abi4 mutants were used to perform genetic crossingwith the flu mutant: abi4-104 mutant (CS3839; single nucleotide substitution at codon 69 leading to missense E to K) and abi4-t mutant (SALK_080095; T-DNA insertion at codon 152). In contrast to wild-type plants, all flu mutants (flu, flu/abi4-t, flu/abi4-104, and flu/ex1/ex2) accumulated five times higher levels of free Pchlide in the dark (Fig. 1a, b and Extended Data Fig. 1a, b). After transfer to the light, flu/abi4-t, flu/abi4-104, and flu/ex1/ex2 generated singlet oxygen in amounts similar to that of flu (Fig. 1c and Extended Data Fig. 1c). Despite their high Pchlide levels, both flu/abi4-t and flu/abi4-104 rescued the lethal phenotype of flu (Fig. 1d and Extended Data Fig. 1d). To rule out the possibility that other proteins may interact with ABI4 and function in 1O2 signaling, the single-point mutant abi4-104 with the full-length ABI4 ORF preserved was selected for the following experiments.
When grown under light / dark cycles, the flu seedlings ceased growth, whereas the flu/abi4 and flu/ex1/ex2 seedlings continued to grow at similar levels to the wild-type, except that their growth was slightly reduced and flu/abi4 flowered a little earlier than the wild-type plants (Fig. 1e, f). Under continuous light, all four lines grew equally well and finally reached the same flowering stage (Extended Data Fig. 2). The 35S-ABI4-GFP/flu/abi4-104 complemented lines showed growth arrest and photobleaching under light / dark cycles (Extended Data Fig. 3), confirming the irreplaceable role of ABI4 in 1O2 signaling.
Singlet oxygen regulates gene expression depending on ABI4
The binding strength of ABI4 with 1O2-responsive gene promoters was then evaluated using the chromatin immunoprecipitation (ChIP)-PCR method. For generating the antibody used for the ChIP essay, five epitopes of the ABI4 protein were synthesized chemically. The monoclonal antibodies against these epitopes were generated by inoculation in a mouse. The immune specificity of each antibody was verified by Western blotting (Extended Data Fig. 4a, b), and the antibody against epitope#2 was selected for the following experiments.
Previous studies indicated that the CCAC motif is a core element required for ABI4 binding8,10. This core binding element was found to be present at high frequencies in the promoters of five representative 1O2-responsive genes, including ZP (a putative C2H2 zinc finger transcription factor; At5g04340)11, WRKY33 (a transcription factor; At2g38470), WRKY46 (a transcription factor; At2g46400), DRP (a disease resistance protein; At1g66090), and ACS6 [1-amino-cyclopropane-1 carboxylic acid (ACC) synthase 6; At1g11280]7. As shown in Fig. 2a–e, the dark-to-light shift significantly induced ABI4 binding to these gene promoters as well as their expression in the flu mutant, whereas these levels were only slightly increased or down-regulated in flu/abi4, flu/ex1/ex2, or wild-type seedlings, also implying the indispensable roles of EXECUTER1/2 and ABI4 in 1O2 signaling.
Nevertheless, the induction rates of ABI4 mRNA levels were much lower than the increasing rates of ABI4 binding (Extended Data Fig. 5), and no significant changes in ABI4 protein levels were observed after the dark-to-light shift in all four lines (Extended Data Fig. 4c), indicating some other regulatory mechanism besides transcriptional or translational enhancement. Previous studies8,10 showed similar results whereby ABI4 expression was barely affected by plastid signals. ABI4 might transmit the signals (bind to the target promoters) through a conformational change (a post-transcriptional regulation)12.
It is interesting to note that Pchlide / 1O2 accumulation levels in the dark were almost the same in the flu mutant, flu/abi4 double-mutant, and flu/ex1/ex2 triple-mutant (Fig. 1a–c); however, only the flu mutant showed a lethal phenotype after illumination (Fig. 1d). This demonstrated that the growth inhibition and seedling lethality do not result from the physicochemical damage caused by 1O2 after the dark-to-light shift but are rather caused by the activation of a genetically determined stress response program6,7.
The 1O2-dependent nuclear gene expression changes were clarified by RNA-seq. Genes with a 2-fold or greater transcript level than the control were considered to be significantly upregulated. After 30 min of reillumination, a total of 896 genes had been upregulated in flu or flu/abi4 or flu/ex1/ex2 relative to the wild-type (Fig. 3a). Among them, 200 transcripts were 1O2-induced genes in flu specifically, including 30 genes regulating cell wall organization (29 genes) or lignin catabolic process (1 gene), 22 phytohormone-regulated genes (6 cytokinin-regulated genes; 5 abscisic-acid-regulated genes; 4 gibberellin-regulated genes; 3 brassinosteroid-regulated genes; 1 ethylene-regulated gene; 1 auxin-regulated gene; 1 salicylic-acid-regulated gene; one jasmonic-acid-regulated gene), 20 stress-responsive genes (11 genes responsive to water deprivation; 5 genes responsive to cold; 2 genes responsive to salt stress; 2 genes responsive to oxidative stress), 15 photosynthesis-related genes, 11 lipid metabolic process genes, 2 genes related to stomatal development, and 2 genes related to cell death (Supplementary data sheet).
Interestingly, 95% (521/548) of upregulated genes in flu/ex1/ex2 relative to the wild-type were also upregulated in flu/abi4 relative to the wild-type (Fig. 3a), suggesting that ABI4 functions downstream of EXECUTER1/2. The expression levels of six representative genes were detected by quantitative real-time PCR, including two cell-death genes XCP1 (XYLEM CYSTEINE PEPTIDASE 1; At4g35350)13 and AED3 [APOPLASTIC EDS1 (ENHANCED DISEASE SUSCEPTIBILITY 1)-DEPENDENT 3; At1g09750]14, two cell wall organization genes F8H (FRAGILE FIBER 8 HOMOLOG; At5g22940)15 and LRX2 (LEUCINE-RICH REPEAT/EXTENSIN 2; At1g62440)16, and two stomatal development genes TMM (TOO MANY MOUTHS; At1g80080)17 and BCA1 (BETA CARBONIC ANHYDRASE 1; At3g01500)18. These genes were all induced more than three times in flu after the dark-to-light shift but were non-significantly or less induced in the wild-typeor flu/abi4 or flu/ex1/ex2 seedlings (Fig. 3b–g), further confirming the key roles of EXECUTER1/2 and ABI4 in 1O2 signaling.
Singlet oxygen propagates a lethal phenotype under light / dark cycles
Consistent with the RNA-seq data, under the condition of light / dark cycles, abrupt ROS accumulation, lipid peroxidation, and cell death were observed in the flu mutant, but not in the flu/abi4 or flu/ex1/ex2 mutants (Extended Data Fig. 6). Besides oxidative damage, interestingly, no stomata have been observed on the flu cotyledons grown under light / dark cycles (Fig. 4a–d and Extended Data Fig. 7b). Given that two negative stomatal development regulators TMM and BCA1 were exponentially up-regulated by 1O2 signals (Fig. 3f, g), the stomatal development defect in the flu cotyledons may be partly explained. Furthermore, apparent cell wall thickening (Fig. 4a and Extended Data Fig. 7b) and increases in pectin (uronic acid) and cellulose contents of cell walls (Fig. 4e, f) were also observed in the flu mutant grown under light / dark cycles, but not in the other lines. As an integrated effect, following the dark-to-light shifts, flu stopped growing, whereas flu/abi4 and flu/ex1/ex2 grew normally.
Although all flu mutants accumulated the same high level of Pchlide in the dark (Fig. 1a, b), the Pchlidewas converted into chlorophylls in flu/abi4 and flu/ex1/ex2 under light, and thus their Pchlide declined to low levels similar to that of the wild-type seedlings (Extended Data Fig. 8b, c). By contrast, although a certain level of chlorophylls (about 1/4 of the wild type; Extended Data Fig. 8f) was synthesized in the flu mutant grown under 7-day light / dark cycles and then transferred to reillumination for 1 h, the plants became bleached (the steady-state chlorophyll level declined to 1/25 of the wild-type; Extended Data Fig. 8e), and their Pchlidecould not be converted into chlorophylls and therefore accumulated (Extended Data Fig. 8c), which then induced 1O2 accumulation and 1O2-responsive gene expression that propagated the lethal phenotype.
Singlet oxygen induces cell wall thickening and stomatal reduction under normal light / dim light cycles
When culturing, we noticed that, if the complete darkness was replaced by a dim light, the flu mutant could grow up. Although about 50% higher Pchlide (Extended Data Fig. 9a) and 120% higher 1O2 (Extended Data Fig. 9b) accumulated, no apparent growth arrest was observed for the flu mutant under 16-h normal light (100 μmol ∙ m-2 ∙ s-1) / 8-h dim light (10 μmol ∙ m-2 ∙ s-1) cycles (Fig. 5a and Extended Data Fig. 9c–e). Nevertheless, stomatal reduction (by 2/3; Fig. 5b, d, e) and cell wall thickening (doubled; Fig. 5c, f, g) and still occurred under normal light / dim light cycles.