ChIP-seq reveals broad roles of SARD1 and CBP60g in regulating plant immunity

Recognition of pathogens by host plants leads to rapid transcriptional reprogramming and activation of defence responses. The expression of many defence regulators is induced in this process, but the mechanisms of how they are controlled transcriptionally are largely unknown. Here we use chromatin immunoprecipitation sequencing to show that the transcription factors SARD1 and CBP60g bind to the promoter regions of a large number of genes encoding key regulators of plant immunity. Among them are positive regulators of systemic immunity and signalling components for effector-triggered immunity and PAMP-triggered immunity, which is consistent with the critical roles of SARD1 and CBP60g in these processes. In addition, SARD1 and CBP60g target a number of genes encoding negative regulators of plant immunity, suggesting that they are also involved in negative feedback regulation of defence responses. Based on these findings we propose that SARD1 and CBP60g function as master regulators of plant immune responses.

P lants use a multilayered defence system to combat microbial pathogens. At the front line, pattern recognition receptors on the plasma membrane recognize conserved features of microbes, collectively known as microbe-associated molecular patterns or pathogen-associated molecular patterns (PAMPs), to activate PAMP-triggered immunity (PTI) 1 . Most PAMP receptors belong to the receptor-like kinase and the receptorlike protein families. A second line of plant defence called effector-triggered immunity (ETI) relies on resistance (R) proteins that detect effector proteins secreted by pathogens to inhibit PTI (ref. 2). The majority of plant R proteins belong to the intracellular nucleotide-binding site (NB) leucine-rich repeats (LRR) protein family. Recognition of pathogens and activation of local defence responses further induce a secondary immune response in the distal part of plants termed systemic acquired resistance (SAR) 3 .
Salicylic acid (SA) is a signal molecule that plays key roles in local defence and SAR (ref. 4). SALICYLIC ACID INDUCTION-DEFICIENT 2 (SID2) and ENHANCED DISEASE SUSCEPTIBILITY 5 (EDS5) are required for pathogen-induced SA accumulation 5,6 . Mutations in SID2 or EDS5 block the accumulation of SA, resulting in enhanced susceptibility to pathogens and loss of SAR (refs 5-7). SID2 encodes Isochorismate Synthase 1 (ICS1), which is a key enzyme in pathogen-induced SA synthesis 6 . EDS5 encodes a transporter involved in exporting SA from chloroplast to cytoplasm 8,9 . Activation of defence gene expression and pathogen resistance by SA depends on the downstream component NON-EXPRESSOR OF PATHOGENESIS RELATED GENES1 (NPR1) 10 . Recent studies showed that NPR1 and its paralogs, NPR3 and NPR4, bind to SA and may function as SA receptors 11,12 .
Two pathogen-induced transcription factors, SAR DEFICIENT1 (SARD1) and CAM-BINDING PROTEIN 60-LIKE G (CBP60g), regulate the expression of ICS1 and are required for pathogen induction of SA synthesis [23][24][25] . Following pathogen infection, SARD1 and CBP60g are recruited to the promoter of ICS1 (ref. 24). In the sard1 cbp60g double mutant, induction of ICS1 expression and SA synthesis is blocked 24,25 . SARD1 and CBP60g belong to the same protein family but are regulated differently, suggesting that they function in two parallel pathways to activate ICS1 expression 23,24 . CBP60g, but not SARD1, can bind calmodulin. On the other hand, overexpression of SARD1, but not CBP60g, leads to constitutive activation of defence responses.
Arabidopsis SNC2 encodes an receptor-like protein that is required for resistance against pathogenic bacteria Pseudomonas syringae pv tomato (P.s.t.) DC3000 and non-pathogenic bacteria P. syringae pv tomato DC3000 hrcC (refs 26,27). A gain-of-function mutation in snc2-1D leads to constitutive activation of both SA-dependent and SA-independent defence pathways 26 . The snc2-1D mutant has small stature, accumulates high levels of salicylic acid, constitutively expresses PATHOGENESIS-RELATED (PR) genes, and exhibits enhanced pathogen resistance. From a suppressor screen of snc2-1D npr1-1, WRKY DNA-BINDING PROTEIN 70 (WRKY70) was identified as an essential regulator of the SA-independent pathway downstream of snc2-1D (ref. 26).
Here we report that SARD1 and CBP60g regulate not only the expression of ICS1 and SA synthesis, but also the expression of WRKY70 and the SA-independent defence pathway in snc2-1D. Chromatin immunoprecipitation (ChIP) analysis revealed that a large number of plant defence regulators including WRKY70 are direct binding targets of SARD1 and CBP60g suggesting that SARD1 and CBP60g function as master regulators of plant defence responses.

Results
SARD1 and CBP60g are required for autoimmunity in snc2-1D.
To determine whether the increased SA synthesis in snc2-1D mutant plants is dependent on SARD1 and CBP60g, we crossed sard1-1 and cbp60g-1 into snc2-1D to obtain the sard1-1 snc2-1D and cbp60g-1 snc2-1D double mutants and the sard1-1 cbp60g-1 snc2-1D triple mutant. Quantitative reverse transcription PCR (RT-PCR) analysis showed that the expression of ICS1 in snc2-1D is much higher than in wild type, but the increased expression of ICS1 is blocked in the sard1-1 cbp60g-1 snc2-1D triple mutant (Fig. 1a). Consistent with the expression levels of ICS1, increased accumulation of SA in snc2-1D is also suppressed in the triple mutant (Fig. 1b).
The sard1-1 snc2-1D and cbp60g-1 snc2-1D double mutants have similar morphology as snc2-1D and are only slightly bigger than snc2-1D (Fig. 1c). Surprisingly, the mutant morphology of snc2-1D is almost completely suppressed in the sard1-1 cbp60g-1 snc2-1D triple mutant (Fig. 1d). Quantitative RT-PCR analysis showed that the expression levels of defence marker genes PR1 and PR2 are slightly lower in the double mutants but are markedly reduced in the triple mutant compared with snc2-1D (Fig. 1d,e). In addition, the enhanced resistance to Hyaloperonospora arabidopsidis Noco2 in snc2-1D is partially reduced in the double mutants and almost completely lost in the triple mutant (Fig. 1f). As blocking SA accumulation by eds5-3 has very little effect on the morphology, PR2 expression and resistance to H. arabidopsidis Noco2 in snc2-1D (ref. 26), these data suggest that SARD1 and CBP60g also regulate SA-independent pathways in snc2-1D. SARD1 and CBP60g regulate the expression of WRKY70. In sard1 cpb60g mutant plants expressing the SARD1-HA fusion protein under its native promoter, pathogen-induced ICS1 expression was restored to similar level as in the cbp60g single mutant, suggesting that SARD1-HA functions similarly as wildtype SARD1 protein ( Supplementary Fig. 1). To identify genes targeted by SARD1, ChIP was carried out on transgenic plants expressing a SARD1-HA fusion protein under its own promoter using an anti-HA antibody. The immunoprecipitated DNA was sequenced by Illumina sequencing. Analysis of the ChIPsequencing (ChIP-seq) data showed a Â 20 genome coverage.
Sequence coverage at each position on the genome was plotted to identify peaks in the Arabidopsis genome. Analysis of peaks in the genic region showed that most sequence peaks are located in the 1.5 kb region upstream of the translation start site, which includes the 5'-UTRs and promoter regions. After removing genes that showed similar sequence peaks in the negative control, peaks with heights of 90 or greater were found in the introns of 84 genes, the 3'-UTRs of 60 genes and the 1.5 kb region upstream of the translation start sites of 1,902 genes. We focused our analysis on the group containing peaks with heights of 90 or greater in the 1.5 kb region upstream of the translation start sites (Supplementary Data 1), because it contains many genes encoding known regulators of plant defence that are strongly induced by pathogen infection ( Table 1). Distribution of sequence reads in the promoter and coding regions of these known defence regulators are shown in Supplementary Fig. 2.
One of the candidate target genes of SARD1 identified by ChIP-seq is WRKY70 (Table 1), which is known to regulate SAindependent defence responses in snc2-1D (ref. 26). Quantitative PCR analysis of the DNA immunoprecipitated by the anti-HA antibody confirmed that WRKY70 is a binding target of SARD1 (Fig. 2a). In sard1 cpb60g mutant plants expressing the CBP60g-HA fusion protein under its native promoter, pathogen-induced ICS1 expression was restored to similar level as in the sard1 single mutant, suggesting that CBP60g-HA functions similarly as wildtype protein ( Supplementary Fig. 1). To determine whether WRKY70 is also a binding target of CBP60g, we carried out ChIP-PCR experiments on transgenic plants expressing a CBP60g-HA fusion protein under its own promoter using the anti-HA antibody. As shown in Fig. 2b, CBP60g is also targeted to the promoter region of WRKY70.
Next we analysed the expression of WRKY70 in snc2-1D, sard1-1 snc2-1D, cbp60g-1 snc2-1D and sard1-1 cbp60g-1 snc2-1D mutant plants. As shown in Fig. 2c, WRKY70 is expressed at a considerably higher level in snc2-1D than in wild type. The expression of WRKY70 is slightly lower in cbp60g-1 snc2-1D and clearly reduced in sard1-1 snc2-1D compared with snc2-D. However, it is further reduced to below wild-type level in the sard1-1 cbp60g-1 snc2-1D triple mutant (Fig. 2c). These data suggest that SARD1 and CBP60g have overlapping functions in regulating the expression of WRKY70 and that reduced expression of WRKY70 is at least partly responsible for the suppression of the snc2-1D-mediated SA-independent constitutive defence responses in the sard1-1 cbp60g-1 snc2-1D triple mutant. SARD1 and CBP60g regulate the expression of EDS5 and NPR1. EDS5 is involved in pathogen-induced SA synthesis 5,7 . Analysis of the SARD1 ChIP-seq data revealed that EDS5 is a potential target gene of SARD1 as well (Table 1). A peak with a height of 110 was identified B700 bp upstream of the translation start site of EDS5. ChIP-PCR experiments confirmed that SARD1 is targeted to the promoter region of EDS5 (Fig. 3a). Further ChIP-PCR analysis showed that CBP60g also binds to the promoter region of EDS5 (Fig. 3b). To determine whether SARD1 and CBP60g are required for the induction of EDS5 by Pseudomonas syringae pv. maculicola (P.s.m.) ES4326, we compared the expression levels of EDS5 in wild type and sard1-1 cbp60g-1 plants. As shown in Fig. 3c, induction of EDS5 by P.s.m. ES4326 is greatly reduced in the sard1-1 cbp60g-1 double mutant. These data suggest that SARD1 and CBP60g directly regulate pathogen-induced expression of EDS5.
Another candidate target gene of SARD1 identified by ChIP-seq is NPR1, which encodes a putative SA receptor 11 . A peak with a height of 163 was identified B100 bp upstream of the translation start site of NPR1 ( Table 1). Binding of SARD1 to the promoter region of NPR1 was confirmed by ChIP-PCR (Fig. 3d). As shown in Fig. 3e, CBP60g is also targeted to the promoter region of NPR1. Analysis of the expression levels of NPR1 in wild type and sard1-1 cbp60g-1 plants showed that induction of NPR1 by P.s.m. ES4326 is compromised in the sard1-1 cbp60g-1 double mutant (Fig. 3f). These data suggest that SARD1 and CBP60g also regulate pathogen-induced expression of NPR1.
Multiple SAR regulators are targets of SARD1 and CBP60g. In addition to EDS5 and NPR1, three other genes required for SAR, FMO1, ALD1 and PBS3, were identified as candidate target genes of SARD1 from the ChIP-seq data. The height of the peaks identified in the promoter regions of FMO1, ALD1 and PBS3 are 99, 138 and 199, respectively ( Table 1). Binding of SARD1 to the promoters of these three genes was confirmed by ChIP-PCR experiments (Fig. 4a). Further ChIP-PCR analysis showed that CBP60g also binds to the promoters of these genes (Fig. 4b). Consistent with data from previous gene expression studies 25,28 , we also observed dramatic reduction in bacteria-induced expression of FMO1, ALD1 and PBS3 in the sard1-1 cbp60g-1 double mutant (Fig. 4c). These data suggest that SARD1 and CBP60g directly regulate the expression of FMO1, ALD1 and PBS3 in plant defence responses. SARD1 and CBP60g target positive regulators of ETI. ENHANCED DISEASE SUSCEPTIBILITY 1 (EDS1) and PHY-TOALEXIN DEFICIENT 4 (PAD4) encode positive regulators of defence responses activated by TIR-NB-LRR R proteins [29][30][31][32][33] . NDR1 is required for defence responses activated by CC-NB-LRR R proteins 29,34 . EDS1 and PAD4, but not NDR1, were identified as candidate target genes of SARD1 by ChIP-seq ( Table 1). The height of the peaks identified in the promoter regions of EDS1 and PAD4 are 258 and 137, respectively. ChIP-PCR experiments showed that SARD1 was targeted to the promoter regions of EDS1 and PAD4, but not NDR1 (Fig. 5a). In addition, CBP60g is also targeted to the promoters of EDS1 and PAD4, but not NDR1 (Fig. 5b). Quantitative RT-PCR was subsequently carried out to determine whether induction of the expression of EDS1 and PAD4 by bacterial infections is dependent on SARD1 and CBP60g. As shown in Fig. 5c, induction of EDS1 and PAD4 by P.s.m. ES4326 is markedly reduced in the sard1-1 cbp60g-1 double mutant. These data suggest that induction of EDS1 and PAD4 following pathogen infection is directly regulated by SARD1 and CBP60g.
ADR1, ADR-L1 and ADR-L2 encode three closely related CC-NB-LRR proteins required for immunity mediated by TIR-NB-LRR R proteins RPP2 and RPP4 (ref. 35). They were also identified as candidate target genes of SARD1 by ChIP-seq ( Table 1). The heights of the peaks identified in the promoter regions of ADR1, ADR-L1 and ADR-L2 are 117, 324 and 230, PLANT U respectively. ChIP-PCR analysis confirmed that SARD1 binds to the promoter regions of these three genes (Fig. 5d). ChIP-PCR experiments showed that CBP60g is also targeted to the promoter regions of ADR1, ADR-L1 and ADR-L2 (Fig. 5e). As shown in Fig. 5f, the expression of ADR1, ADR-L1 and ADR-L2 is induced by P.s.m. ES4326 and the induction is partially dependent on SARD1 and CBP60g. These data suggest that SARD1 and CBP60g are likely to directly regulate the expression of ADR1, ADR-L1 and ADR-L2 in plant defence.
SARD1 and CBP60g play critical roles in PTI. Among the candidate target genes of SARD1 identified by ChIP-seq, eight genes including BAK1, BKK1, AGB1, BIK1, MEKK1, MKK4, MPK3 and CPK4 (Table 1) were previously shown to encode positive regulators of PTI (refs 36-48). Binding of SARD1 to the promoter regions of these genes was further confirmed by ChIP-PCR (Fig. 6a). In addition, CBP60g is also targeted to the promoter regions of these genes (Fig. 6b). As shown in Fig. 6c, expression of BAK1, BKK1, AGB1, BIK1, MEKK1, MKK4, MPK3 and CPK4 is induced P.s.m. ES4326 and the induction is reduced in the sard1-1 cbp60g-1 double mutant, suggesting that SARD1 and CBP60g may directly regulate their expression in plant defence responses.
To test whether SARD1 and CBP60g are required for PTI, we analysed bacterial growth in wild type, sard1-1, cbp60g-1 and sard1-1 cbp60g-1 plants pretreated with flg22, a peptide from bacterial flagellin that is recognized by FLAGELLIN-SENSITIVE 2 (FLS2) (ref. 49). As shown in Fig. 6d, flg22-induced resistance to P. syringae pv tomato DC3000 is not obviously affected in the sard1-1 and cbp60g-1 single mutants, but clearly reduced in the sard1-1 cbp60g-1 double mutant, suggesting that SARD1 and CBP60g contribute to PTI. SARD1 and CBP60g target negative regulators of defence. Analysis of the SARD1 ChIP-seq data also identified a number of negative regulators of plant immunity including PUB13, WRKY40, WRKY60, NUDT6, NUDT7, MLO2, BON1, BAP1 and BAP2 as candidate target genes of SARD1 ( Table 1). Binding of SARD1 to the promoter regions of PUB13, WRKY40, WRKY60, NUDT6, NUDT7, MLO2, BON1, BAP1 and BAP2 was confirmed by ChIP-PCR analysis (Fig. 7a). In addition, CBP60g was also found to target the promoter regions of these nine genes (Fig. 7b). Quantitative RT-PCR analysis showed that the expression of PUB13, WRKY40, WRKY60, NUDT6, NUDT7, MLO2, BON1, BAP1 and BAP2 are all induced by P.s.m. ES4326 and the induction is either reduced or blocked in the sard1-1 cbp60g-1 double mutant (Fig. 7c). These data suggest that SARD1 and CBP60g regulate the expression of these negative regulators of plant immunity during plant defence.
SARD1 regulates gene expression through the GAAATTT element. Previously, we showed that SARD1 and CBP60g bind preferentially to the oligonucleotide probe GAAATTTTGG (ref. 24). Bioinformatics analysis showed that the GAAATTT motif within this probe is over-represented in the promoters of the genes with SARD1 and CBP60g-dependent expression 25 . Analysis of the 1,902 candidate target genes of SARD1 in Supplementary Data 1 showed that the GAAATTT motif is also over-represented in the promoter regions of this group of genes (Po10 À 15 , Fisher's exact test). This motif is over-represented in the promoter regions of 29 confirmed target genes of SARD1 and CBP60g listed in Table 1 (Po0.005, Fisher's exact test) as well. However, not every gene in this group contains this motif in their promoter region. It is likely SARD1 and CBP60g can also bind to certain variants of the GAAATTT motif. Interestingly, a closely related sequence motif, G(A/T)AATT(T/G), was identified as a conserved motif (Po10 À 25 , Fisher's exact test) among the sequence peaks of genes in Supplementary Data 1 using the motif discovery algorithm DREME.
To test whether SARD1 activates its target gene expression through the GAAATTT motif, we made a construct expressing the luciferase reporter gene under the control of a 56 bp fragment from the ChIP-Seq peak region in the promoter of ICS1, which contains a GAAATTT and a related GAAATT motif (Fig. 8a). Two additional constructs containing mutations in these two motifs were also created to determine whether they are required for activation of reporter gene expression by the 56 bp fragment. These reporter gene constructs were transformed into Arabidopsis protoplasts to exam the luciferase reporter expression levels. As shown in Fig. 8b, all three constructs expressed similar levels of luciferase as the original NOS101-Luc vector, suggesting that the 56 bp promoter fragment cannot activate luciferase expression on its own in protoplast transient assays. However, when the  luciferase reporter gene constructs were co-transformed together with a plasmid expressing the SARD1 protein into protoplasts, luciferase expression was much higher in samples transformed with the construct containing the wild-type 56 bp promoter fragment compared with samples transformed with the NOS101-Luc vector (Fig. 8c). In comparison, samples transformed with the construct carrying mutations in the GAAATTT motif exhibited significantly reduced luciferase activity. When both the GAAATTT and GAAATT motifs were mutated, the luciferase activity was further reduced to a level similar to that in the NOS101-Luc vector control. Together, these data suggest that SARD1 activates gene expression through GAAATTT or similar DNA sequence elements.

Discussion
SA functions as a key signalling molecule in SAR. SARD1 and CBP60g have previously been shown to regulate pathogen-induced SA synthesis [23][24][25] . In this study, we showed that, in addition to ICS1, the expression of another regulator of SA synthesis, EDS5, is also likely controlled by SARD1 and CBP60g. NPR1, a gene required for the perception of SA by plants, is a target of SARD1 and CBP60g as well. Moreover, SARD1 and CBP60g also regulate pathogen-induced expression of several other genes required for SAR. Both SARD1 and CBP60g are targeted to the promoter regions of FMO1, ALD1 and PBS3 and induction of these genes by P.s.m. ES4326 is markedly reduced in the sard1 cbp60g double mutant. These data suggest that SARD1 and CBP60g function in coordinating the induction of SAR regulators during plant defence.
Several defence regulators that function upstream of SA synthesis are also regulated by SARD1 and CBP60g. Both PAD4 and EDS1 are required for pathogen-induced SA synthesis 32,50 . SARD1 and CBP60g are targeted to their promoters and are required for their induction by P.s.m. ES4326. In addition, ADR1, ADR1-L1 and ADR1-L2, three helper R genes required for pathogen-induced SA synthesis 35 , are also targets of SARD1 and CBP60g. Regulation of the induction of PAD4, EDS1, ADR1, ADR1-L1 and ADR1-L2 by SARD1 and CBP60g may play critical roles in promoting SA synthesis during pathogen infection.    We also found that SARD1 and CBP60g function downstream of the receptor-like protein SNC2 to regulate both SA-dependent and SA-independent defence pathways. SARD1 and CBP60g are required for the increased expression of ICS1 and SA synthesis in snc2-1D. Regulation of the SA-independent defence pathway by SARD1 and CBP60g appears to be at least partly through their control of WRKY70 expression, a key regulator of the SA-independent defence responses in snc2-1D (ref. 26   CBP60g in PTI was further confirmed by the attenuation of flg22induced pathogen resistance in the sard1 cbp60g double mutant. In addition to upregulation of positive regulators, negative regulators are often induced during plant defence as well. Induction of negative regulators is critical for feedback inhibition of defence responses to prevent uncontrolled activation, which may lead to autoimmunity. We showed that a number of negative regulators of plant immunity including PUB13, WRKY40, WRKY60, NUDT6, NUDT7, MLO2, BON1, BAP1 and BAP2 are also targets of SARD1 and CBP60g. Among them, PUB13 is a U-box/ARM E3 ubiquitin ligase that regulates cell death as well as degradation of FLS2 after flagellin induction 55,56 ; WRKY40 and WRKY60 function redundantly with their close homologue, WRKY18, to repress basal defence 57,58 ; NUDT6 and NUDT7 are two Nudix domain-containing proteins that negatively regulate EDS1-dependent immune responses 13,59,60 ; MLO2 functions as a negative regulator of resistance to powdery mildew 61 ; BON1 functions as a negative regulator of immunity mediated by the TIR-NB-LRR R protein SNC1 (ref. 62); BAP1 and BAP2 encode two C2 domain-containing proteins that negatively regulate programmed cell death 63,64 . All these genes are induced following infection by P.s.m. ES4326 and their induction requires SARD1 and CBP60g, suggesting that SARD1 and CBP60g also play an important role in the negative feedback regulation of plant defence.
Bioinformatics analysis has previously been used to analyze genes that are co-expressed with a group of SARD1/CBP60gdependent genes 28 . Four genes including AGP5, At5g52760, CML46 and CML47 that form a small cluster with SARD1 and ICS1 were identified as candidate target genes of SARD1 and CBP60g. These genes are also identified as binding targets of SARD1 in our ChIP-seq data (Supplementary Data 1). EDS1 and PAD4 were also found to cluster with ICS1 in the co-expression analysis. They were placed upstream of SARD1 and CBP60g. Interestingly, both EDS1 and PAD4 have been shown to be targets of SARD1 and CBP60g in our ChIP studies. The commonly used defence marker genes PR1 and PR2 were also found in one of the clusters co-expressed with SARD1/CBP60g-dependent genes. However, both of them were not identified as binding targets of SARD1 in our ChIP-seq data, suggesting that they are not directly regulated by SARD1 and CBP60g. It is likely that genes co-expressed with SARD1/CBP60g-dependent genes include   genes that are either directly or indirectly regulated by SARD1 and CBP60g.
In summary, a large number of genes encoding key regulators of plant immunity are direct binding targets of SARD1 and CBP60g and their expression is modulated by SARD1 and CBP60g during plant defence. This is consistent with the functions of these two transcription factors in PTI, ETI and SAR. Based on this data we suggest that SARD1 and CBP60g orchestrate the induction of plant defence regulators in plant immunity (Fig. 9).
SA was extracted and measured using a protocol modified from Li et al 65 . About 0.1 g of leaf tissue from 3-week-old soil-grown plants was collected and grounded in liquid nitrogen. Four samples for each genotype were collected and analysed. A volume of 0.6 ml of 90% methanol was added to each sample and the samples were subsequently vortexed and sonicated for 20 min. After spinning at 16,000g for 20 min, the supernatant was collected and the pellet was re-extracted with 0.5 ml of 100% methanol as above. The supernatants from the two extractions were combined and dried by vacuum. A volume of 0.5 ml of 5% trichloroacetic acid was then added to the dry samples and the samples were vortexed, sonicated for 5 min, and centrifuged at 16,000g for 15 min. The supernatant was collected and extracted three times with 0.5 ml of extraction medium (ethylacetate/cyclopentane/ isopropanol:100/99/1 by volume). After spinning at 16,000g for 1 min, the top organic phases were collected, combined and dried by vacuum. The samples were then re-suspended in 250 ml of mobile phase (0.2 M KAc, 0.5 mM EDTA (pH 5)) by vortexing and sonicating for 5 min. After spinning at 16,000g for 5 min, the supernatants were kept and analysed by high-performance liquid chromatography to determine the amount of SA.
H. arabidopsidis Noco2 infection assays were carried out on 3-week-old soilgrown plants by spraying plants with H. arabidopsidis Noco2 spore suspension at a concentration of 50,000 per ml water. Afterwards, plants were covered with a clean dome and grown at 18°C under 12 h light per 12 h dark cycle in a growth chamber. H. arabidopsidis Noco2 sporulation was scored 7 days later as previously described 66 .
To assay for flg22-induced pathogen resistance, leaves of 4-week-old plants were infiltrated with 1 mM of flg22 or ddH 2 O as control. After 24 h, the same leaves were inoculated with P. syringae pv tomato DC3000 (OD 600 ¼ 0.001) in 10 mM MgCl 2 . Three days post inoculation, a leaf disc was taken from each infected leaf and two leaf discs from the same plant were collected as one sample. The samples were ground, diluted serially in 10 mM MgCl 2 , and plated on Lysogeny broth agar plates with 25 mg ml À 1 rifampicin and 50 mg ml À 1 kanamycin. After incubation at 28°C for 36 h, bacterial colonies were counted from selected dilutions and the colony numbers were used to calculate colony forming units.
ChIP analysis. For ChIP experiments, two to three fully expanded leaves of 25day-old plants grown under short day condition were infiltrated with P.s.m. ES4326 (OD 600 ¼ 0.001). The inoculated leaves were collected after 24 h. About 4 g of leaf tissue was cross-linked in 75 ml of 1% formaldehyde solution plus 0.01% Silwet L-77 under vacuum for 20 min. Glycine (2 M) was added to a final concentration of 0.125 M and the sample was vacuumed for an additional 5 min to stop crosslinking. The tissue was rinsed three times with 60 ml of cold ddH 2 0 and dried with blotting paper. The nuclei were prepared as previously described 67 and resuspended in 300 ml of nuclei lysis buffer (50 mM Tris-HCl (pH 8.0), 10 mM EDTA, 1% SDS, 0.1 mM PMSF, 1xPI). The nuclei suspensions were subsequently sonicated to shear the DNA to an average size of 0.3-1 kb.
The sonicated chromatin suspension was spun at 12,000g for 5 min at 4°C to pellet debris. The supernatant was moved to a new 15 ml tube. An aliquot of 5 ml from each sample was moved into a clean 1.5 ml Eppendorf tube and set aside  at -20°C as 'input'. ChIP dilution buffer (3 ml; 1.1% Triton X-10, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.0, 167 mM NaCl) was then added to the 15 ml tube. For pre-clearing, 100 ml of Protein A agarose beads balanced with the ChIP dilution buffer was added to the chromatin samples and kept at 4°C for 1 h with rotation. The beads were pelleted at 2,400g for 2 min and the supernatants were divided equally into two samples. A volume of 5 ml (0.4 mg ml -1 ) of anti-HA antibody (Roche) was added to one sample for immunoprecipitation and immunoglobin G was added to the other sample as control. The samples were incubated overnight at 4°C with gentle agitation. Subsequently, 100 ml of Protein A agarose beads balanced with ChIP dilution buffer was added to each sample and kept at 4°C for 2 h with gentle agitation.
After the final wash, the samples were pelleted for an additional 2 min at 2,400g to remove the supernatant thoroughly. To elute the immune complexes, 250 ml of Elution Buffer (1% SDS, 0.1 M NaHCO 3 ) was added to the beads. The samples were vortexed briefly and incubated at 65°C for 15 min with gentle agitation. After spinning at 3,800g for 2 min, the supernatant was carefully transferred to a fresh tube. The pellet was eluted one more time with 250 ml of Elution Buffer and the two eluates were combined (a total of B500 ml). At the same time, 500 ml of Elution Buffer was added to the input samples collected before immunoprecipitation. A volume of 1 ml of 10 mg ml -1 DNase-free RNase A was added to each sample. After incubation at 37°C for 1 h, 10 ml of 0.5 M EDTA, 20 ml 1 M Tris-HCl (pH 6.5) and 2 ml of 10 mg ml -1 proteinase K were added to each sample. The samples were incubated at 45°C for 1 h and extracted with the same volume of Tris saturated phenol:chloroform:isoamyl alcohol (25:24:1v/v) twice. DNA was then precipitated by adding 0.7 volume isoproponal, 1/10 volume of 3 M NaOAc and 1 ml 2M glycogen and incubating at room temperature for 30 min. DNA was pelleted by spinning for 20 min at 16,700g. The DNA pellets were washed with 80% ethanol, dried at room temperature, re-suspended in 50 ml TE buffer, and stored at À 20°C for further use.
For ChIP-sequencing, DNA sequencing libraries were prepared from chromatin immunoprecipitated DNA using a library preparation kit (E6000s, New England Biolabs) according to the manufacturer's instructions and sequenced using an Illumina Genome Analyzer. Tissue from untreated SARD1-HA transgenic plants was used as the negative control. Sequence reads were mapped to Arabidopsis genome sequence using Bowtie 0.12.8 (ref. 68). Sequence coverage at each position on the genome was scored by Samtools 69 and used to identify peaks in the genome. The 500 bp sequences centred on the peak summits shown in Supplementary Data 1 were used to identify conserved SARD-binding motifs using DREME (ref. 70). DREME was run with default settings and sequences from the promoter regions of randomly chosen genes were used as background control. Confirmation of ChIPseq results was carried out with three independent ChIP experiments. Independently grown plants were used in each repeat and immunoprecipitated DNA was quantified by real-time PCR using gene-specific primers. The primers used to amplify the promoter regions of the target genes are listed in Supplementary Table 1. Real-time PCR was performed in 96-well format using Bio-Rad CFX connect Real-Time PCR systems and the SYBR Premix Ex Taq II (TAKARA).
Promoter activity assay. The NOS101-Luciferase reporter vector was created by modifying pGreen0229 to include a firefly luciferase gene driven by a basal promoter of the nopaline synthase gene ( À 101 to þ 4, designated NOS101). The wild type and mutant versions of the 56 bp promoter fragment of ICS1 were synthesized and inserted upstream of the NOS101 basal promoter in the reporter vector. Promoter activity assays were performed by expressing the reporter constructs with the 35S-SARD1 construct or empty vector in Arabidopsis protoplasts. A 35S-driven Renilla luciferase reporter was included in the assays as internal transfection controls. Transformed protoplasts were incubated for 16-20 h before the activities of the luciferases were measured using a Dual-Luciferase Reporter Assay (Promega).