Edinburgh Research Explorer The arabidopsis zinc finger proteins SRG2 and SRG3 are positive regulators of plant immunity and are differentially regulated by nitric oxide

Nitric oxide (NO) regulates the deployment of a phalanx of immune responses, chief among which is the activation of a constellation of defence-related genes. However, the underlying molecular mechanisms remain largely unknown. The Arabidopsis thaliana zinc ﬁnger transcription factor (ZF-TF), S -nitrosothiol (SNO) Regulated 1 (SRG1), is a central target of NO bioactivity during plant immunity. Here we characterize the remaining members of the SRG gene family. (cid:1) Both SRG2 and, especially, SRG3 were positive regulators of salicylic acid-dependent plant immunity. Analysis of SRG single, double and triple mutants implied that SRG family members have additive functions in plant immunity and, surprisingly, are under reciprocal regulation. (cid:1) SRG2 and SRG3 localized to the nucleus and functioned as ethylene-responsive element binding factor-associated amphiphilic repression (EAR) domain-dependent transcriptional repressors: NO abolished this activity for SRG3 but not for SRG2. Consistently, loss of GSNOR function, resulting in increased (S)NO concentrations, fully suppressed the disease resistance phenotype established from SRG3 but not SRG2 overexpression. Remarkably, SRG3 but not SRG2 was S -nitrosylated in vitro and in vivo . (cid:1) Our ﬁndings suggest that the SRG family has separable functions in plant immunity, and, surprisingly, these ZF-TFs exhibit reciprocal regulation. It is remarkable that, through neofunctionalization, the SRG family has evolved to become differentially regulated by the key immune-related redox cue, NO. repress the transcription of a repressor of salicylic acid (SA) signalling and reactive oxygen species (ROS) accumulation, contributing to the establishment of plant immunity. As the concentration of NO accumulates over time, SRG3 becomes S -nitrosylated, resulting in a structural change precluding binding of SRG3 to its cognate cis -element. Thus, the transcriptional repressive activity of SRG3 is abolished, enabling expression of immune repressor(s) that contribute to the damping of the defence response. By contrast, NO does not drive the expression of SRG2 , which is induced by other defence-related cues. Remarkably, SRG2 is not sensitive to direct NO-mediated S -nitrosylation. Thus, the SRG family of C2H2 ZnTFs, despite their high similarity, are differentially regulated by NO bioactivity.

In addition to the indirect SNO-reductase activity of GSNOR, Thioredoxin h5 may also function as a denitrosylase by directly reducing SNO groups present in some S-nitrosylated proteins (Kneeshaw et al., 2014), providing an additional layer of regulation. Interestingly, NO and GSNO have also been shown to have separable and overlapping functions in the development of plant immunity, possibly because they redox-modulate the activity of different target proteins (Yun et al., 2016).
The accumulating data suggest that NO production following the pathogen-triggered nitrosative burst contributes to the reprogramming of plant immune-response genes (Parani et al., 2004;Zago et al., 2006;Palmieri et al., 2008;Bellin et al., 2013;Xu et al., 2013). However, the molecular mechanism(s) responsible remain largely opaque. To date, NO has been proposed to modulate the translocation of the transcriptional coactivator NPR1 into the nucleus (Tada et al., 2008;Lindermayr et al., 2010) and the specific DNA-binding activity of its protein interactor, the basic leucine-zipper transcription factor, TGA1 (Lindermayr et al., 2010). Recently, the zinc finger transcription factor (ZF-TF), SNO Regulated Gene 1 (SRG1), was shown to be a positive regulator of plant immune responses, by acting as a transcriptional repressor, presumably by suppressing the transcription of an immune repressor (Cui et al., 2018). Significantly, SRG1 function during the plant immune response was shown to be modulated by S-nitrosylation of cysteine (Cys) 87, a highly evolutionary conserved residue, leading to both compromised DNA binding and, by extension, transcriptional repression activity.
Here we characterize the remaining members of the SRG gene family. Both SRG2 and, especially, SRG3 function as positive regulators of SA-dependent plant immunity. Genetic analysis of SRG single, double and triple mutants revealed that SRG family members have additive functions in plant immunity and, unexpectedly, are reciprocally regulated. Both SRG2 and SRG3 localized to the nucleus and acted as transcriptional repressors. Significantly, NO abolished this activity for SRG3 but not for SRG2. The absence of GSNOR function, leading to increased (S)NO concentrations, fully suppressed the disease resistance phenotype established from SRG3 overexpression but this was not found to be the case for overexpression of SRG2. SRG3 but not SRG2 was S-nitrosylated in vitro and in vivo, further highlighting the differences between these two SRG proteins. Our findings collectively suggest that the SRG family has separable functions in plant immunity and, surprisingly, these ZF-TFs exhibit reciprocal regulation. It is remarkable that through neofunctionalization the SRG family has evolved to become differentially regulated by the key immune-related redox cue, NO.

Plant materials
Arabidopsis thaliana (Arabidopsis) seeds were placed in ½ Murashige & Skoog (MS) medium, and subsequently 12-d-old plants were transferred to soil and grown at 22°C either under short-day conditions of 8 h : 16 h, light : dark (employed for pathogen infiltration experiments and quantitative PCR (qPCR) analysis) or under 16 h : 8 h, light : dark conditions (utilized for seed collection and plant transformation).
For the construction of SRG2 or SRG3 overexpression lines, the open reading frame (ORF) of SRG2 or SRG3 was fused to the CaMV35S promoter and C-terminal FLAG tag within the binary vector pGWB11 (Nakagawa et al., 2007), employing the Gateway cloning system. To generate constructs for the conditional expression of either SRG2 or SRG3, the coding sequence of these genes was inserted into the XhoI/SpeI sites of the boestradiol-inducible vector pER8 (Zuo et al., 2000). The resulting constructs were subsequently transferred into Arabidopsis Col-0 plants by Agrobacterium-mediated plant transformation (Clough & Bent, 1998;Zhang et al., 2006). Plant transformants were confirmed by both antibiotic resistance and genotyping. To conditionally induce SRG2/ SRG3 gene expression, 6-wk-old plants were sprayed with 100 lM b-oestradiol (Sigma) or dimethyl sulphoxide (DMSO; Invitrogen) and incubated for 48 h before further experiments.
The double mutant srg2 srg3 was obtained by crossing the individual mutant lines. In a similar fashion, the SRG2 and SRG3 overexpression lines (SRG2-OX and SRG3-OX, respectively) were crossed with the salicylic acid induction deficient (sid) 2 and gsnor1-3 mutants, employing the transgenic lines as pollen donors, to create the SRG2-OX sid2-2, SRG3-OX sid2-2, SRG2-OX gsnor1-3 and SRG3-OX gsnor1-3 lines. To generate the srg1 srg2 srg3 triple mutant line, a sgr1-specific DNA sequence (14-34 bp) of the corresponding ORF was cloned into the CRISPR-Cas9 system vector pDe-CAS9 as previously described (Pyott et al., 2016). The resulting construct was subsequently transformed into the srg2 srg3 double mutant line by Agrobacteriummediated transformation. Plants carrying the transgenic construct were selected in the T1 generation by spraying with a 120 mg l À1 solution of BASTA. The resulting transgenic plants were confirmed as srg1 srg2 srg3 triple mutants by PCR genotyping.

Pathogen inoculation
The bacteria pathogen Pseudomonas syringae pv. tomato (Pst) DC3000 expressing the avrRpm1 avirulence gene (Pst DC3000 (avrRpm1)) was cultured in low-salt LB medium (10 g l À1 tryptone, 5 g l À1 yeast extract and 5 g l À1 NaCl, pH 7.0) containing appropriate antibiotics and grown overnight in the dark at 28°C. Bacteria were collected by centrifugation and washed twice with 10 mM MgCl 2 , then finally suspended in 10 mM MgCl 2 at a final concentration of 10 5 cells ml À1 . This bacterial solution was pressure-infiltrated into the leaves of 5-wk-old Arabidopsis plants of the indicated genotypes. At least 10 leaves from five different plants of a given genotype were inoculated for each experiment. Bacterial titres were determined at the time points indicated. The experiments were repeated at least three times.
To assay cell death development, Arabidopsis leaves were stained using trypan blue (10 g phenol, 10 ml glycerol, 10 ml lactic acid, 10 ml water and 10 mg Trypan blue) by boiling for 2 min. Leaves were then destained with 2.5 g ml À1 chloral hydrate after cooling to room temperature. Images were taken following successful destaining. Subsequently, the relative intensity of staining was quantified by IMAGE J (v.1.51j8, Java 1.8.0_261, Wayen Rasband, National Institutes of Health, Bethesda, MD, USA).
To explore the ROS burst, leaves were immersed in 0.5 mg ml À1 nitro blue tertrazolium (NBT; Sigma) for 3 h to detect superoxide formation or 1 mg ml À1 3-3 0 -diaminobenzidine (DAB; Sigma) to score for hydrogen peroxide formation, for 8 h at room temperature in the dark. Following clearance of Chl, stained leaf discs were submerged in ethanol until clear and then photographed.

Quantification of cell death and ROS
Quantification of hydrogen peroxide was as previously described (Chen et al., 2013) with minor modifications. Leaves were stained with freshly prepared 1 mg ml À1 DAB (Sigma) solution for 8 h, and Chl was then removed with ethanol. The resulting leaves were ground in liquid nitrogen after their weight was recorded and then solubilized in 0.2 M HClO 4 , followed by centrifugation at 10 000 g for 15 min to remove debris. The absorbance of the resulting supernatants was measured immediately at A 450 and quantified by comparison with a standard curve generated with known concentrations of H 2 O 2 in 0.2 M HClO 4 -DAB.
Quantification of superoxide was adapted as described previously (Chen et al., 2013) with minor modifications. The leaf was stained with fresh NBT (0.5 mg ml À1 ; Sigma) for 3 h, and then Chl was removed by addition of ethanol. The NBT-stained leaves were ground in liquid nitrogen, solubilized in 2 M KOH-DMSO, and then centrifuged at 10 000 g for 15 min to remove debris. The samples were evaluated at A 630 and compared with a standard curve, generated with known amounts of NBT in the KOH-DMSO mix.
To measure electrolyte leakage induced by cell death, leaves from 5-wk-old plants were cut into 5-mm-diameter slices and soaked in water for 6 h, the conductivity of the solution was measured with a DiST WP conductivity meter (Hanna Instruments, Woonsocket, RI, USA) as previously described (Cui et al., 2018). The units for this assay are microsvedbergs cm À1 (lS cm À1 ), where the distance refers to that between the electrodes.
Reactive oxygen species production among wild-type Col-0 and mutant Arabidopsis plants was determined by luminol-based assay (G omez-G omez et al., 1999;Smith & Heese, 2014). Leaves were cut into 5-mm-diameter slices and floated overnight on water. Subsequently, the water was removed and 0.1 ml of H 2 O supplied containing 20 lM luminol, 1 lg horseradish peroxidase (Fluka, Gillingham, UK) and flg22 (Sigma) was added. Luminescence was measured in a Multimode Plate Reader SpectraMax M5 (Molecular Devices, Gillingham, UK) for 35 min; 1 lM flg22 was used in this experiment.

SA determination
Plant material (0.1 g) was collected and ground in liquid nitrogen. Samples were extracted using 95% ethanol and the resulting liquid was analysed by high-performance liquid chromatography-MS after centrifugation (Kim et al., 2013).

Fluorescence microscopy and dual luciferase assays
For the localization of SRG2 and SRG3, the SRG2 or SRG3 ORFs were cloned into the binary vector pEarleyGate103, using the Gateway system, so their expression was driven by the CaMV35S promoter and a C-terminal green fluorescent protein (GFP) tag was added (Earley et al., 2006). The truncation of the ethylene-responsive element binding factor-associated amphiphilic repression (EAR) motif of SRG proteins was undertaken utilizing a QuikChange II Site-Directed Mutagenesis Kit (Stratagene, Agilent Technologies, Cheadle, UK). The resulting constructs were transiently transformed into tobacco leaves or Arabidopsis protoplasts. A Leica TCS SP5 II confocal microscope was employed for GFP imaging (excitation 488 nm, emission 500-600 nm). Also, protein extracts were obtained from transformed leaves or protoplasts and subjected to Western blotting using an anti-GFP antibody (Sigma).
To score SRG2 and SRG3 for potential transcriptional repressive activity, 10 µl of the indicated DNA (4 µg of effector plasmid, 5 µg of reporter plasmid and 5 µg of internal plasmid) was transformed into Arabidopsis protoplasts and incubated for 16 h. The resulting cells were then collected and extracted for luciferase activity assay using a Dual-Luciferase Report Assay System (Promega) with a SpectraMax M5 Multimode Plate Reader (Molecular Devices). Ten replicates were measured for each experiment. Each experiment was repeated at least three times.
qPCR Total RNA was extracted from 0.1 g plant tissue using a RNA isolation mini-kit (Agilent Technologies) and cDNA was synthesized from 2 lg total RNA employing oligo (dT) primers and reverse transcriptase (First-Strand cDNA Synthesis Kit; Invitrogen). PCR was performed in a 20 µl reaction containing SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA), cDNA and primers (listed in Table S1) utilizing the LightCycler ® 480 Real-Time PCR System (Roche). UBQ10 and Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust New Phytologist (2020) www.newphytologist.com UBC9 were used as internal controls. Mean values and standard deviations were obtained from at least three biological replicates.

Protein expression and S-nitrosylation
The ORFs of either SRG2 or SRG3 were cloned into the expression vector pMAL-c5X with a maltose-binding protein (MBP) tag at the N-terminus. The construct was transformed into Escherichia coli strain BL21(DE3) for protein expression, then purified using amylose magnetic beads (New England Biolabs, Ipswich, MA, USA). The purified protein was utilized for further experiments.
S-nitrosylation assays employing purified recombinant protein or proteins extracted directly from plant tissue were interrogated by the biotin switch assay (BSA) (Jaffrey & Snyder, 2001). CysNO was synthesized by dissolving 13 mg of reduced free L-cysteine in 0.5 ml of 0.1 M HCl, and then added to 0.5 ml of 220 mM NaNO 2 to obtain 110 mM CysNO. The CysNO was maintained in the dark for 20 min and then diluted to working concentration. Freshly prepared CysNO was always utilized in any experiments performed.

Statistics
Data are expressed as means AE SD from a minimum of three independent experiments. Statistical analysis of the data was carried out using ANOVA analysis followed by Dunnett's test unless otherwise specified. Differences were considered significant at P < 0.05 (*) and highly significant at P < 0.01 (**).

Results
The expression of SRG1 homologues, SRG2 and SRG3, are induced by pathogens and NO Previously we identified an Arabidopsis C2H2 type ZF-TF, SRG1, which positively regulates plant immunity (Cui et al., 2018). SRG1 function is regulated at the transcriptional and post-transcriptional level by NO (Cui et al., 2018). Phylogenetic analysis showed that four C2H2-type ZF-TFs were classified into a small group: SRG1, SRG2 (At3g46090), SRG3 (At5g59820) and SRG4 (At3g46070) (Fig. S1a). In order to examine whether these Arabidopsis SRG1 paralogues are also involved in plant immunity, quantitative reverse transcription polymerase chain reaction (qRT-PCR) was performed to determine if the expression of these genes was transcriptionally activated by the plant immune activator SA, flg22, a pathogen-associated molecular pattern (PAMP) derived from bacterial flagellin or Pst DC3000, a virulent bacterial pathogen on the Col-0 accession of Arabidopsis (Whalen et al., 1991). Expression of SRG2 and SRG3 was induced by all three of these immune-related stimuli (Figs 1a-d, S1b,c), implying that these genes might participate in the establishment of plant immunity. By contrast, transcripts of SRG4 did not accumulate in response to these cues (Fig. S1d), so this gene was not investigated further. Interestingly, the expression of both SRG2 and SRG3 could also be induced by the NO donors, SNP and GSNO (Figs 1e,f, S1e-i). Either SNP-or GSNO-induced SRG3 expression was also significantly reduced in the presence of the NO scavenger, cPTIO. Conversely, SNPand GSNO-induced SRG2 expression was only weakly reduced by this NO scavenger (Figs 1e,f, S1g). Thus, while SRG3 expression is responsive to NO, the accumulation of SRG2 transcripts appears to be strikingly less sensitive to this redox signalling cue.

Constitutive expression of SRG2 and SRG3 enhances plant immunity
To gain further insights into the possible role(s) of SRG2 and SRG3 in plant immunity, plants containing 35S::SRG2-FLAG (SRG2-OX) or 35S::SRG3-FLAG (SRG3-OX) transgenes were generated. These transgenic lines exhibited reduced stature compared with wild-type Col-0 ( Fig. 2a,b). Further, the FW of these lines was directly correlated with the strength of SRG2 or SRG3 expression ( Fig. 2c-f). Thus, SRG2 and SRG3 expression negatively impacts Arabidopsis stature.
To examine the effect of SRG2 and SRG3 on basal immunity, Pst DC3000 was inoculated into SRG2-OX and SRG3-OX transgenic plants and the bacterial titre determined over time. The amount of infiltrated Pst DC3000 in the SRG2-OX and SRG3-OX transgenic lines was comparable to wild-type Col-0 at 0 d post-inoculation (dpi) (Fig. 2g,h), which suggests that SRG2-OX and SRG3-OX transgenic plants can be infiltrated to similar extents as the wild-type. Hence, SRG2 and SRG3 overexpression does not reduce the amount of infiltrated bacteria. A reduced titre of Pst DC3000 was detected in SRG2-OX and SRG3-OX transgenic plants compared with the wild-type at 3 dpi ( Fig. 2g,h). Further, our data suggest that the level of SRG2 and SRG3 expression is directly related to the extent of resistance against Pst DC3000, demonstrating that SRG2 and SRG3 act as positive regulators of plant basal disease resistance.
We next examined the impact of SRG2 and SRG3 expression on R gene-mediated disease resistance. Pst DC3000 expressing the avirulence gene avrRpm1 (Pst DC3000(avrRpm1)) is recognized by the R protein, RPM1, in the Col-0 accession of Arabidopsis (Grant et al., 1995). The titre of Pst DC3000 (avrRpm1) in SRG-OX plants was significantly less than that in the wild-type at 3 dpi ( Fig. 2i,j), indicating that overexpression of SRG2 or SRG3 leads to increased resistance against Pst DC3000 (avrRpm1). These results imply that overexpression of SRG2 or SRG3 in Arabidopsis enhances both basal defence and R gene-mediated resistance. Collectively, our findings therefore suggest that SRG2 and SRG3 are positive regulators of plant immunity.
SRG2 and SRG3 overexpression lines exhibit elevated cell death, accumulation of ROS and constitutive PR1 expression High levels of SRG2 and SRG3 overexpression resulted in the formation of microlesions, which was confirmed by trypan blue (TB) staining and associated microscopy (Fig. 3a). The relative intensity of cell death staining was quantified by IMAGEJ, which indicated that microlesion formation increased with increasing New Phytologist (2020) Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust www.newphytologist.com
ROS formation is a key early defence response Torres et al., 2006). Therefore, we treated the leaves of SRG2-OX and SRG3-OX lines with either DAB, which stains hydrogen peroxide (H 2 O 2 ), or NBT, which stains superoxide (O 2 À ) (Jabs et al., 1996;Thordal-Christensen et al., 1997;, to determine potential accumulation of these molecules. Both SRG2-OX and SRG3-OX lines exhibited increased DAB and NBT staining compared with that of wildtype Col-0 ( Fig. 3c-f). Further, increasing SRG2 or SRG3 expression resulted in enhanced DAB and NBT staining (Fig. 3d,f). We also determined the expression of the SA marker gene, Pathogenesis-Related1 (PR1). As expected, PR1 expression was significantly increased in SRG2-OX and SRG3-OX lines compared with wild-type Col-0 ( Fig. 3g) and was positively correlated to SRG transcript accumulation. Taken together, our results show that overexpression of SRG2 or SRG3 activates a number of immune responses, including cell death development, ROS production and PR1 expression.
SRG2 and SRG3 promote increased basal and R gene-mediated immunity To further investigate the biological contribution of SRG2 and SRG3 in plant immunity, T-DNA insertion lines for these genes were obtained and homozygous loss-of-function mutants generated utilizing PCR genotyping (Fig. S2a). The expression level of SRG1, SRG2 and SRG3 in these T-DNA loss-of-function Fig. 1 Expression of Arabidopsis SRG2 and SRG3 is nitric oxide-and pathogen-inducible. (a, b, e, f) Relative gene expression of SRG2 (a, e) and SRG3 (b, f) in response to flg22, salicylic acid (SA), sodium nitroprusside (SNP) or SNP with 2-(4-carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (cPTIO) treatment. Ten-day-old seedlings were used for these experiments and water was employed as a control. UBQ10 was utilized as an internal control. Error bars represent means AE SD. hpi, h post-inoculation. (c, d) Transcript abundances of SRG2 (c) and SRG3 (d) were determined following treatment with Pseudomonas syringae pv. tomato (Pst) DC3000 or Pst DC3000 + cPTIO. UBQ10 was utilized as an internal control. Experiments were repeated at least three times. Error bars represent means AE SD. One-way ANOVA assays were performed to determine significant differences relative to mock treatments or indicated lines, **, P < 0.01; *, P < 0.05; ns, not significant.
Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust New Phytologist (2020) www.newphytologist.com New Phytologist insertion lines was determined by qPCR, which confirmed that these lines were null mutants for the relevant genes and also indicated a complex transcriptional relationship among them (Fig. 4a). Surprisingly, loss-of-function mutations in either SRG1 or SRG3 strikingly reduced SRG2 expression. SRG3 expression, by contrast, was not reduced in loss-of-function srg1 plants and was only reduced by c. 30% in a loss-of-function srg2 line. SRG1 transcript accumulation was reduced c. 60 % in either srg2 or srg3 Error bars indicate means AE SD from three to six biological replicates. **, P < 0.01. (f) Western blot analysis of either SRG2 or SRG3 protein expression in SRG overexpression lines using an anti-FLAG antibody. Total protein extracted from fresh leaves of indicated lines was used in these experiments. Wild-type Col-0 plants served as negative controls. Coomassie Brilliant Blue (CBB) stain was employed as loading control. (g-j) Bacterial titre of Pseudomonas syringae pv. tomato (Pst) DC3000 (g, h) and Pst DC3000 carrying avrRpm1 (i, j) in the indicated plant genotypes. Error bar indicates mean AE SD (n = 7). One-way ANOVA assay: **, P < 0.01. All experiments were repeated at least three times. The relative intensity of TB staining was performed with IMAGEJ software. Error bars indicate means AE SD, n ≥ 10. **, P < 0.01. (c) Accumulation of H 2 O 2 in 6-wk-old plant leaves was determined by 3,3 0 -diaminobenzidine (DAB) staining. Bar, 5 mm. (d) Quantification of H 2 O 2 in 10-d-old Arabidopsis seedlings of the indicated genotypes; 0.1 g of seedlings from each line were grouped as one sample. Error bars indicate AE SD (n = 5). **, P < 0.01. (e) Accumulation of superoxide in 6-wk-old plant leaves was detected by nitro blue tetrazolium (NBT) staining. Bar, 5 mm. (f) Quantification of superoxide production in 10-d-old seedlings of the indicated genotypes. Error bars indicate AE SD (n = 5). **, P < 0.01. (g) mRNA level of PR1 in the stated Arabidopsis lines. Error bars indicate means AE SD. Experiments were repeated three times with similar results. **, P < 0.01. (2020) Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust www.newphytologist.com

Research
New Phytologist loss-of-function mutants. These three ZF-TFs are therefore under reciprocal regulation. The expression of SRG genes in the SRG overexpression lines also revealed a complex relationship between these transcription factors (Fig. S2b).
Of these C2H2 ZF TFs, SRG2, SRG3 and SRG1 share high similarities in DNA sequence and might function redundantly. Therefore, a srg2 srg3 double loss-of-function mutant was obtained by crossing the associated single mutations, followed by PCR analysis of F 2 plants. We also generated a srg1 srg2 srg3 triple loss-of-function mutant. As SRG1 and SRG2 are closely linked, we employed CRISPR/Cas9 gene editing technology to mutate SRG1 in a srg2 srg3 double mutant. Interestingly, phenotypic analysis indicated there was a small increase of FW in the srg1, srg2 and srg3 single SRG loss-of-function mutants compared with that of the wild-type and a significant increase in FW in the triple loss-of-function mutant (Fig. 4b,c). Thus, loss-of-function mutations in SRG1, SRG2 and SRG3 exhibit partially overlapping impacts on Arabidopsis growth. The expression of these ZF-TFs was then determined in the associated double and triple mutants (Fig. S2c-e). SRG2 expression was abolished in the srg2 srg3 double mutant and in the srg1 srg2 srg3 triple loss-of-function mutant. In a similar fashion, SRG3 expression was abolished in the srg2 srg3 double loss-of-function mutant and the srg1 srg2 srg3 triple loss-of-function mutant.
To examine if SRG2, SRG3 and SRG1 have redundant functions in plant immunity, Pst DC3000 (avrRpm1) was infiltrated into the associated loss-of-function mutants and the bacterial titre was recorded over time (Fig. 4d). Both single loss-of-function mutants support more Pst DC3000(avrRpm1) compared with wild-type Col-0 and the titre in the srg2 srg3 loss-of-function line was significantly higher than that in the respective single loss-offunction mutants, indicating disruption of SRG2 or SRG3 leads to an increased titre of Pst DC3000(avrRpm1). Significantly, statistical analysis revealed that the titre of PstDC3000(avrRpm1) in the srg1 srg2 srg3 triple loss-of-function mutant is significantly higher than that in the srg2 srg3 double mutant and srg1, srg2 and srg3 single loss-of-function mutants. Also, the srg2 srg3 double mutant is more susceptible than srg2 and srg3 single loss-of-function mutants. The impact of these mutations on basal disease resistance was tested. Similar results were observed following inoculation of Pst DC3000 (Fig. 4e).
Taken together, our findings suggest that SRG1, SRG2 and SRG3 are required for R gene-mediated protection and basal resistance and there is functional redundancy between these ZF-TFs.
Loss of SRG2, SRG3 and SRG2 SRG3 function reduces cell death development, ROS production and SA accumulation As SRG2-OX and SRG3-OX lines exhibited elevated cell death development, we assessed the biological consequence of a loss of either SRG2 or SRG3 function on development of the hypersensitive response (HR) during the development of immunity. Leaves were stained by TB after Pst DC3000(avrRpm1) infiltration, and TB staining was then quantified. Cell death development was diminished in these mutants relative to wild-type Col-0 at 12 h post-inoculation (hpi; Fig. S3a,b). Further, this result was confirmed by transient overexpression of either SRG2 or SRG3 in the respective SRG loss-of-function lines using an b-estradiol-inducible expression system (Zuo et al., 2000) (Fig. S3c,d). b-estradiol-treated plants showed higher SRG2 or SRG3 expression respectively and increased TB staining in leaves, whereas mock treatment did not induce either SRG2 or SRG3 expression, and cell death development was similar to that scored in the srg2 and srg3 loss-of-function mutants (Fig. S3c,d). To further determine the extent of HR cell death in these lines, electrolyte leakage was quantified (Fig. 5a,b). Again, the amount of electrolyte leakage was reduced in these single, double and triple SRG loss-of-function mutants compared with wild-type Col-0, with that in the srg1 srg2 srg3 triple loss-of-function mutant most pronounced (Fig. 5a), implying that SRG2, SRG3 and SRG1 play important roles in HR development during immunity. Further, we examined the effects of transient conditional SRG2 or SRG3 overexpression in srg2 or srg3 loss-of-function lines, respectively, on New Phytologist electrolyte leakage. Following challenge with Pst DC3000 (avrRpm1), b-estradiol cued SRG2 or SRG3 transient overexpression resulted in higher electrolyte leakage relative to mock-treated plants (Fig. 5b).
We determined ROS production upon Pst DC3000 challenge in the single, double and triple SRG loss-of-function mutants by a luminol-based assay (G omez-G omez et al., 1999). ROS production was slightly reduced in srg1, srg2 and srg3 single loss-offunction mutants and also the srg2 srg3 double loss-of-function mutant in response to Pst DC3000 (Fig. 5c). Statistical analysis showed that total ROS accumulation was significantly reduced in the srg1 srg2 srg3 triple loss-of-function mutant compared with the related double or single loss-of-function mutants (Fig. S3d). Further, we used flg22 to examine PAMP-induced ROS production in these lines. The ROS burst was faster and stronger in wild-type Col-0 after treatment with flg22, whereas it was reduced in the related double and triple loss-of-function mutants (Figs 5d, S3e). Further, the srg1 srg2 srg3 triple loss-of-function mutant exhibited reduced total ROS production following flg22 treatment relative to the srg1, srg2, srg3 single loss-of-function mutants. Moreover, increased ROS accumulation was observed following conditional transient overexpression of either SRG2 or SRG3 relative to the mock control (Figs 5e, S3f).
Overexpression of either SRG2 or SRG3 induced Respiratory Burst Oxidase Homolog D (RBOHD) expression, which encodes the key source of pathogen-triggered apoplastic ROS Torres et al., 2006) (Fig. S3g). Together, these results further support the suggestion that SRG2, SRG3 and SRG1 play important and redundant functions required for immune-related ROS production.
We also examined the expression of the SA-associated marker, PR1, in response to inoculation of Pst DC3000 (Fig. 6a). Compared with the wild-type, PR1 expression was reduced in all SRG loss-of-function mutants. However, the srg2 srg3 double loss-offunction mutant exhibited a reduction greater than that in the single srg1, srg2 and srg3 loss-of-function mutants, and the triple srg1 srg2 srg3 loss-of-function mutant showed lower PR1 expression than the double mutant at both 12 and 24 hpi. Further, SA concentrations in these lines were also analysed upon Pst DC3000 infection. The concentration of SA was similar in all tested plant lines without infection (Fig. 6b), suggesting that the size of srg mutants may not be associated with SA accumulation. A significant increase of SA in wild-type Col-0 plants was observed upon pathogen inoculation at both 12 and 24 hpi, whereas SA concentration in the single, double and triple SRG loss-of-function mutants, especially the triple mutant, was significantly decreased compared with wild-type Col-0 (Fig. 6b), linking the pathogen susceptibility of these SRG loss-of-function mutants to a decrease in SA accumulation. Therefore, we also determined the expression level of CBP60g, SARD1 and WRKY62 genes implicated in the regulation of SA synthesis in this collection of mutant lines in response to Pst DC3000 inoculation. Expression of CBP60g, SARD1 and WRKY62 was reduced upon Pst DC3000 infection in all of these mutants and this was especially pronounced in the srg1 srg2 srg3 triple loss-of-function mutant (Fig. S4), further implicating SRG2, SRG3 and SRG1 in Electrolyte leakage triggered by Pseudomonas syringae pv. tomato (Pst) DC3000 in the indicated lines was measured with a conductivity meter. Error bars indicate means AE SD (n = 8). One-way ANOVA assays were performed to determine significant differences at 36 h post-inoculation (hpi) relative to wildtype Col-0 (a) or wild-type Col-0 dimethyl sulphoxide (DMSO) (b). **, P < 0.01. (c-e) ROS production in response to Pst DC3000 (c) or flg22 (d, e) treatment was determined by a luminol-based assay. Error bars indicate means AE SD (n = 6). (2020) Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust www.newphytologist.com

Research
New Phytologist SA-associated immune responses. We also crossed our SRG2-OX and SRG3-OX lines with the SA biosynthesis-deficient mutant, salicylic acid induction deficient 2-2 (sid2-2) (Oide et al., 2013). In SRG2-OX sid2-2 and SRG3-OX sid2-2 lines the reduced physical stature resulting from overexpression of either SRG2 or SRG3 was only partially recovered (Fig. 6c,d), implying that the impact of either SRG2 or SRG3 overexpression on Arabidopsis growth is not totally dependent on SA accumulation. However, the titre of PstDC3000 in either SRG2-OX sid2-2 or SRG3-OX sid2-2 lines was similar to that in wild-type Col-0 plants (Fig. 6e,f). Similar results were observed following inoculation of Pst DC3000 (avrRpm1) (Fig. 6g,h). Collectively, these results suggest that SRG2 and SRG3 regulate plant immunity through the SA pathway.
SRG2 and SRG3 encode NO-regulated transcriptional repressors SRG2 and SRG3 are plant C2H2 ZF TFs containing a predicted nuclear localization signal (NLS) (Fig. 7a). To test the functionality of this domain, SRG2 and SRG3 were fused to the C-terminus of GFP downstream of the CaMV35S promoter. Each of the resulting constructs was then transiently transformed into tobacco leaves (Fig. S5a) or Arabidopsis protoplast (Fig. S5c). The integrity of each of these fusion proteins was confirmed by Western blot analysis (Fig. S5b,d). In both tobacco and Arabidopsis, SRG2-GFP and SRG3-GFP localized to the nucleus.
Similar to SRG1, SRG2 and SRG3 both contain a leucine-rich EAR motif-like sequence within their C-terminus (Fig. 7a). This motif has previously been reported to function as a transcriptional repressor (Kagale & Rozwadowski, 2011;Cui et al., 2018). Therefore, we carried out a transcriptional activity assay in Arabidopsis protoplasts with either a Galactose 4 DNA Binding Domain fused with either SRG2 or SRG3 (GAL4-DB-SRG2 or GAL4-DB-SRG3, respectively) together with a reporter gene comprising five copies of the GAL4 DNA-binding site fused to the firefly Luciferase (LUC) reporter gene (Fig. 7b). Both SRG2 and SRG3 exhibited transcriptional repression activity and this activity was abolished by truncation of the EAR motif (Fig. 7c), which did not influence their localization (Fig. S5e-g).
It has been previously reported that NO may modulate the transcriptional repressive activity of SRG1 (Cui et al., 2018). Therefore, we tested if NO might impact the transcriptional repressive activity of either SRG2 or SRG3. As expected, application of the NO donor, SNP, strikingly reduced the ability of SRG3 to operate as a transcriptional repressor. Surprisingly, and in complete contrast, NO was not found to blunt the transcriptional repressive activity of SRG2 following the addition of two distinct NO donors, SNP (Fig. 7d) and GSNO (Fig. 7e). In a gsnor1-3 genetic background, which chronically accumulates (S) NO, SRG3-dependent transcriptional repression was abolished (Fig. 7f), which was not the case for SRG2-dependent transcriptional repression, which was slightly diminished. However, the chronic, long-term accumulation of (S)NO in the gsnor1-3 line may result in the indirect action of these molecules on this activity. To determine if SRG function might be impacted by NO in SRG overexpression plants, we determined the NO concentration in these lines. The concentration of NO accumulation in these plants was similar to that of the wild-type line (Fig. S6). Collectively, our findings show that, remarkably, SRG2 and SRG3, despite their high degree of sequence similarity, were differentially regulated by direct NO function.

NO selectively S-nitrosylates SRG proteins
S-nitrosylation is a key mechanism to convey NO bioactivity (Astier et al., 2011;Yu et al., 2012). As our data suggest that NO can directly inhibit SG3 but not SRG2 transcriptional repression activity, we determined if either SRG2 or SRG3 could be directly modified by NO. We generated recombinant SRG2 and SRG3 proteins using a MBP fusion protein system (MBP-SRG2 and MBP-SRG3, respectively) and exposed these proteins to the natural NO donors, GSNO (Fig. 8a) or Cys-NO (Fig. 8b) and monitored their possible S-nitrosylation by the BSA. No SRG2-SNO formation was detected by either GSNO or Cys-NO treatment in vitro (Fig. 8a,b), even following long film exposure times to detect weak signals (Fig. S7). By contrast, MBP-SRG3 was S-nitrosylated strongly in response to both GSNO and Cys-NO (Fig. 8a,b). In order to confirm and extend these findings, MBP-SRG2 and MBP-SRG3 were subjected to a range of GSNO concentrations and possible SNO formation determined. In agreement with our previous data, SRG3 was S-nitrosylated and, further, this redox-based PTM occurred in a GSNO concentration-dependent fashion (Fig. 8d). However, formation of SRG2-SNO could not be detected, even over a range of GSNO concentrations (Fig. 8c). Together, these data suggest that NO selectively modifies SRG proteins in vitro.
To determine if either SRG2 or SRG3 could be S-nitrosylated in vivo, we generated C-terminal FLAG-tagged SRG2 (SRG2-FLAG) and SRG3 (SRG3-FLAG), respectively. Subsequently, Arabidopsis protoplasts expressing the indicated transgene were exposed to GSNO and endogenous proteins subjected to the BSA. Subsequently, biotinylated proteins were purified with streptavidin beads. These proteins were then immunoblotted with an anti-FLAG antibody. SRG3 was found to be S-nitrosylated in vivo. By contrast, no S-nitrosylation of SRG2 could be detected (Fig. 8e). We next tested possible S-nitrosylation of SRG2 and SRG3 in gsnor1-3 plants, which show increased concentrations of GSNO accumulation and, by extension, elevated levels of global SNO-protein formation. The BSA assay revealed the formation of SRG3-SNO. Conversely, no SRG2-SNO was detected (Fig. 8f). These data suggest that SRG3 but not SRG2 can be S-nitrosylated in vivo in a gsnor1-3 genetic background.
We next investigated if SRG3-SNO formation or S-nitrosylation of SRG2 occurred during attempted pathogen infection. Following Pst DC3000 inoculation, SRG3-SNO formation was detected from 6 hpi (Fig. 8h). By contrast, SRG2-SNO could not be detected at any of the time points tested (Fig. 8g,h). Collectively, these findings imply that SRG3 but not SRG2 is the target of in vitro and in vivo S-nitrosylation.
To further explore the possible impact of increased SNO concentrations on phenotypes resulting from SRG overexpression, Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust New Phytologist (2020) www.newphytologist.com Fig. 6 Arabidopsis SRG2 and SRG3 function in salicylic acid (SA)-dependent signalling and immunity. (a) mRNA level of PR1 in the given plant genotypes following pathogen challenge. Six-week-old plants were inoculated as indicated, followed by quantitative PCR. Pseudomonas syringae pv. tomato (Pst) DC3000 was resuspended in 10 mM MgCl 2 and MgCl 2 was used as mock. Error bars indicate means AE SD from three independent biological replicates. **, P < 0.01 (significant difference compared with wild-type Col-0 by ANOVA assay). (b) Total salicylic acid (SA) concentrations in the indicated Arabidopsis lines in response to pathogen challenge. Six-week-old plants were challenged as indicated, and the SA concentration was subsequently analysed by highperformance liquid chromatography. Error bars indicate AE SD (n = 3). ANOVA assays were performed to determine significant differences compared with wild-type Col-0 at 24 h post inoculation (hpi). **, P < 0.01. (c) Morphological phenotypes of the indicated plant genotypes at 6 wk old under short-day conditions. Bar, 2 cm. (d) FW in the stated Arabidopsis lines at 6 wk old under short-day conditions (8 h : 16 h, light : dark). Error bars indicate means AE SD (n = 8). **, P < 0.01 (ANOVA assay). (e-h) The titre of Pst DC3000 (e, f) or Pst DC3000(avrRpm1) (g, h) was determined at 0 and 3 d post-inoculation (dpi) following infiltration of virulent Pst DC3000 (e, f) or avirulent Pst DC3000 carrying avrRpm1 (g, h) (1 9 10 5 CFU ml À1 ). Error bars indicate means AE SD (n = 7). **, P < 0.01 (ANOVA assay). All experiments were repeated at least three times. (2020) Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust www.newphytologist.com

Discussion
Our findings show that SRG2 and SRG3 function as transcriptional repressors, presumably through the recruitment of the corepressor, TOPLESS, via interaction with their EAR domain, in a similar fashion to SRG1 (Cui et al., 2018). Further, SRG3 was S-nitrosylated in vitro by both Cys-NO and GSNO, two natural NO donors, and in vivo in response to attempted pathogen infection. Also, following (S)NO accumulation and subsequent SRG3 S-nitrosylation, the transcriptional repressive activity of SRG3 was abolished. In addition, gsnor1-3, which results in increased total cellular (S)NO accumulation, suppressed the growth and immunity phenotypes associated with ectopic overexpression of SRG3 in SRG3-OX gsnor1-3 lines. Remarkably, SRG2 was not S-nitrosylated either in vitro or in vivo following attempted pathogen infection. It was also surprising that (S)NO accumulation failed to directly abolish the transcriptional repressive activity of SRG2. Moreover, gsnor1-3 did not fully suppress the growth and immunity phenotypes associated with ectopic overexpression of SRG2. In aggregate, our results show that SRG C2H2 ZnTFs exhibit differential capacities to act as substrates for NO-mediated S-nitrosylation. Thus, while SRG1 and SRG3 undergo SNO formation, which regulates their biological (c) Transient transcriptional repression assay. Indicated effector constructs were cotransformed into Arabidopsis protoplasts with reporter and internal control plasmids. Relative luciferase activity was quantified. Relative luciferase activities of fusion constructs compared with GAL4-DB control. Error bars represent means AE SD (n = 10). Experiments were repeated three times with similar results. **, P < 0.01. (d, e) Transient transcriptional repression assay after addition of NO donors, 10 mM sodium nitroprusside (SNP) (d) or 0.5 mM S-nitrosoglutathione (GSNO). Relative luciferase activity of GAL4-DB control was normalized as 1. Error bars represent means AE SD (n = 10). Experiments were repeated three times with similar results. **, P < 0.01; ns, not significant. (f) Transient transcriptional repression assay in wild-type Col-0 and gsnor1-3 protoplasts. Relative luciferase activity of the GAL4-DB control was normalized as 1. Error bars represent means AE SD (n = 10). Experiments were repeated three times with similar results. **, P < 0.01 (significant difference compared to GAL4-DB by ANOVA assay).
Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust New Phytologist (2020) www.newphytologist.com function, the closely related family member, SRG2, is not a substrate for this redox-based PTM and is therefore, by extension, not regulated by this modification. Thus, the SRG family of C2H2 ZnTFs, despite their high similarity, are, remarkably, differentially regulated by NO bioactivity. Collectively, our findings support a model that is presented and described in Fig. 10.
By employing a molecular modelling strategy, we established that in both SRG1 and SRG3, Cys87, a highly conserved target residue for SNO formation (Cui et al., 2018), is completely solvent exposed and fully accessible for modification by NO. Conversely, in SRG2, Cys87 is significantly less accessible to NOdriven S-nitrosylation (Fig. S8a-j). Thus, our data suggest that the observed differential S-nitrosylation of SRG proteins is facilitated by the structural location of Cys87 within SRG2 relative to the position of this residue within SRG1 and SRG3. To the extent of our knowledge, this is the first report of a closely related protein family being differentially regulated by this redox-based modification in either plants or animals.
Differential regulation of a plant TF gene family by a given PTM has, for example, also been demonstrated for the basic leucine zipper TGA transcriptional activators which function in both plant immunity and development (Pontier et al., 2002). This TF family consisting of 10 members has been proposed to bind TGACG-motifs in SA-regulated promoters from Recombinant SRG2 and SRG3 proteins were generated using the myelin basic protein (MBP) expression system (MBP-SRG2 and MBP-SRG3, respectively). (a, b) The resulting SRG proteins were exposed to either 0.5 mM S-nitrosoglutathione (GSNO) (a) or 0.5 mM Cys-NO (b) with or without ascorbate (ASC) for monitoring S-nitrosylation of these proteins by the biotin switch assay (BSA). (c, d) Dose-dependent GSNO-dependent S-nitrosylation. PS, Ponceau staining; SNO, detection of S-nitrosothiol formation; -MBP, detection of MBP antibody as loading control. (e, f) S-nitrosylation of SRG2 and SRG3 in vivo; total protein extracts from SRG2-FLAG or SRG3-FLAG constructs in Col-0 plants (e) or gsnor1-3 plants (f) were subjected to the BSA. Ten percent of total protein before BSA was used as input and detected by an anti-FLAG antibody; 0.5 mM GSNO was used in (e). (g, h) Pseudomonas syringae pv. tomato (Pst) DC3000 induced S-nitrosylation of SRG3 (h) but not SRG2 (g). Total protein extracts from Arabidopsis protoplast expressing SRG-FLAG constructs were subjected to the BSA after Pst DC3000 inoculation. Ascorbate was employed as indicated to control for SNO formation. Ten percent of total protein before BSA was used as input and detected by anti-FLAG.

New Phytologist
Arabidopsis and tobacco, which have been shown to be functionally relevant (Zhang et al., 1999;Despr es et al., 2000). Two TGA TFs are specifically targeted for developmental stage-specific proteolysis by the 26S proteasome, presumably following ubiquitination of a target lysine residue (Pontier et al., 2002). Thus, TGA TFs may be regulated by differential targeted proteolysis, serving to modulate the contribution of specific members of this multigene TF family to complex developmental pathways.
To explore the differential S-nitrosylation of SRG proteins further, we undertook phylogenetic analysis to determine the possible selective pressure on SRG1, SRG2 and SRG3 (Methods S1, S2). This analysis suggested more nonsynonymous to synonymous substitutions, indicating the presence of positive selection (Table S2). Thus, the duplication of SRG genes may have resulted in subfunctionalization: the resulting gene copies post-duplication specialize to perform different functions within the same genetic pathway. Alternatively, neofunctionalization might have ensued, where one gene copy maintains the ancestral function while the additional copies are selected to perform novel activities outside the original genetic pathway. Our SRG functional data suggest that loss-of-function mutations in SRG1, SRG2 or SRG3 result in decreased disease resistance. Further, the negative impact on disease resistance of the srg2 srg3 double loss-of-function mutant is additive relative to the srg2 and srg3 single loss-of-function mutants, implying that these genes may have evolved, at least partially, separate functions in the regulation of plant immunity. Further, the srg1 srg2 srg3 triple loss-of-function mutant exhibits greater pathogen susceptibility relative to the srg2 srg3 double loss-of-function mutant, indicating that all three SRG genes may have related but separable functions associated with plant immunity. Null loss-offunction mutations within the same genetic pathway would not be expected to be additive. These findings are therefore consistent with neofunctionalization of the SRG gene family, where one gene copy maintains the ancestral function, while the additional copies are selected to perform novel functions outside the original genetic pathway. Our experimental data also demonstrate that SRG1 and SRG3 are actively S-nitrosylated and this redox-based PTM regulates the activity of these proteins. By contrast, SRG2 is not subject to SNO formation. This feature also clearly differentiates SRG proteins at the molecular level. Western blot analysis was carried out to detect SRG proteins in the indicated plant lines using an anti-FLAG antibody. Coomasie Brilliant Blue (CBB) stain was employed as a loading control. (e, f) Titre of Pseudomonas syringae pv. tomato (Pst) DC3000(avrRpm1) (e) or Pst DC3000 (f) was determined at 0 and 3 d post-inoculation (dpi) in each indicated line. The concentration of bacteria used in these experiments was 1 9 10 5 CFU ml À1 . Error bars indicate means AE SD (n = 7). Experiments were repeated three times with similar results. **, P < 0.01; ns, not significant.
Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust New Phytologist (2020) www.newphytologist.com The evolution of transcriptional regulatory networks has been proposed to occur predominantly through variation in cis-regulatory elements located within the promoters of target genes . This is because mutations in transcription factors would potentially result in widespread consequences relative to mutations within the regulatory elements themselves. Conversely, the accumulating data imply that variations in transcriptional regulators themselves can also drive the evolution of complex transcriptional regulatory networks, supported through gene duplication events of cognate transcriptional activators (Force et al., 1999;Innan & Kondrashov, 2010). We suggest that the plant SRG family of C2H2 ZnTFs represents a quintessential example of this type of transcriptional-network evolution.
Interestingly, our results suggest that all members of the SRG family of ZnTFs reciprocally regulate the transcription of other members of this gene family. Thus, loss-of-function mutations in SRG1, SRG2 or SRG3 impact the associated transcript accumulation of the other two SRG genes. For example, direct comparison of SRG transcript abundances shows that loss of SRG3 function results in c. 80% reduction in transcript accumulation for SRG2 and a 70% reduction for SRG1. These data establish SRG3 as the SRG family member that has the greatest influence on the expression of other SRG genes. Analysis of the promoter sequences associated with SRG genes reveals, in each case, multiple potential binding motifs for C2H2 ZnTFs. Therefore, it is possible that the observed reciprocal regulation of mRNA abundance amongst SRG family members is mediated through the direct binding of a given SRG TF to cognate cis-elements within the promoter sequences of other SRG genes. However, our data suggest that SRG1, SRG2 and SRG3 function as transcriptional repressors and hence mutations in these TFs would be expected to increase rather than decrease the transcript abundance of other SRG family members. Thus, the observed reciprocal regulation of gene expression amongst SRG family members may not occur through direct promoter binding and subsequent transcriptional regulation. More likely, our data infer that SRG reciprocal regulation occurs indirectly, perhaps by the given SRG repressing the transcription of a transcriptional repressor, which would ordinarily target other SRG family members.
Reciprocal regulation of gene expression is typically linked to feedback loops. A classic example is the circadian clock where the basic molecular architecture consists of negative-feedback loops where positive and negative components control each other's expression to generate oscillations with an approximate 24 h period (Bell-Pedersen et al., 2005). In this context, the first described feedback loop of the Arabidopsis circadian clock was based on the reciprocal regulation between TIMING OF CAB EXPRESSION 1 (TOC1) and CIRCADIAN CLOCK-ASSOCIATED 1 (CCA1)/LATE ELONGATED HYPOCOTYL (LHY). CCA1 and LHY are Myb TFs that bind directly to the TOC1 promoter to negatively regulate its expression, while TOC1 binds directly to the CCA1 and LHY promoters negatively regulating their expression (Gendron et al., 2012). The molecular mechanism underpinning SRG reciprocal regulation would therefore be an interesting target for future investigation, especially because, unlike the plant circadian clock example, SRG reciprocal regulation is likely to be indirect and also there is a general paucity of information on similar molecular interactions in plants. Model depicting the role of SRG proteins in plant immunity. Attempted pathogen infection rapidly triggers the nitrosative burst leading to nitric oxide (NO) production. The accumulation of NO drives the expression of selected SRG family members. This regulatory feature may occur through an indirect mechanism: SRG transcriptional repressors repress the transcription of a transcriptional repressor that targets SRG genes. SRG2 and SRG3 transcriptional repressors subsequently repress the transcription of a repressor of salicylic acid (SA) signalling and reactive oxygen species (ROS) accumulation, contributing to the establishment of plant immunity. As the concentration of NO accumulates over time, SRG3 becomes S-nitrosylated, resulting in a structural change precluding binding of SRG3 to its cognate cis-element. Thus, the transcriptional repressive activity of SRG3 is abolished, enabling expression of immune repressor(s) that contribute to the damping of the defence response. By contrast, NO does not drive the expression of SRG2, which is induced by other defence-related cues. Remarkably, SRG2 is not sensitive to direct NO-mediated Snitrosylation. Thus, the SRG family of C2H2 ZnTFs, despite their high similarity, are differentially regulated by NO bioactivity.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.        Methods S1 SRG protein structure analysis.
Methods S2 Evolutionary analysis of SRG genes.