Bifurcation of Arabidopsis NLR Immune Signaling via Ca2+-Dependent Protein Kinases

Nucleotide-binding domain leucine-rich repeat (NLR) protein complexes sense infections and trigger robust immune responses in plants and humans. Activation of plant NLR resistance (R) proteins by pathogen effectors launches convergent immune responses, including programmed cell death (PCD), reactive oxygen species (ROS) production and transcriptional reprogramming with elusive mechanisms. Functional genomic and biochemical genetic screens identified six closely related Arabidopsis Ca2+-dependent protein kinases (CPKs) in mediating bifurcate immune responses activated by NLR proteins, RPS2 and RPM1. The dynamics of differential CPK1/2 activation by pathogen effectors controls the onset of cell death. Sustained CPK4/5/6/11 activation directly phosphorylates a specific subgroup of WRKY transcription factors, WRKY8/28/48, to synergistically regulate transcriptional reprogramming crucial for NLR-dependent restriction of pathogen growth, whereas CPK1/2/4/11 phosphorylate plasma membrane-resident NADPH oxidases for ROS production. Our studies delineate bifurcation of complex signaling mechanisms downstream of NLR immune sensors mediated by the myriad action of CPKs with distinct substrate specificity and subcellular dynamics.


Introduction
The first line of nonself recognition and immune responses in multicellular organisms is triggered by conserved pathogen-or microbe-associated molecular patterns (PAMPs/MAMPs) through pattern recognition receptors (PRRs). MAMPs, such as bacterial flagellin and peptidoglycan (PGN) or fungal chitin, are perceived by cell-surface receptors to mount PAMP/MAMP-triggered immunity (PTI) for broad-spectrum microbial resistance in plants [1,2]. Successful pathogens acquired virulence effectors to suppress PTI. To confine or eliminate pathogens, plants further evolved polymorphic R proteins to directly or indirectly recognize effectors and initiate effector-trigger immunity (ETI) accompanied with localized PCD and systemic defense signaling [3,4,5,6,7]. The most common R proteins are intracellular immune sensors with the nucleotide-binding domain (NB) and leucine-rich repeat (LRR), a structural feature shared by mammalian NOD-like receptors that perceive intracellular MAMPs and danger signals to initiate inflammation and immunity [6,8,9,10,11,12]. Whether and how distinct intracellular and cell-surface immune sensors trigger overlapping or/and differential primary immune signaling responses are still largely open questions.
In Arabidopsis thaliana, NLR protein RPS2 initiates resistance upon recognition of Pseudomonas syringae effector AvrRpt2, whereas RPM1 recognizes two sequence-unrelated effectors, AvrRpm1 and AvrB. With a few exceptions, NLR proteins do not interact directly with pathogen effectors, but instead monitor perturbation of host proteins by pathogen effectors to mount defense responses [3,4,5,6,7,8,9,10]. For instance, AvrRpt2 degrades Arabidopsis RIN4 protein to activate RPS2 signaling, whereas AvrRpm1 and AvrB induce RIN4 phosphorylation via host kinases to initiate RPM1 signaling [13,14,15,16]. Although several plant NLR proteins, such as barley MLA10 [17], tobacco N [18] and Arabidopsis RPS4 [19,20], require effector-induced nuclear translocation for immune signaling, RPS2 and RPM1 are anchored to the plasma membrane to elicit immune responses [15,21]. Potato Rx protein requires both nuclear and cytoplasmic localizations for full immunity [22,23]. Apparently, different NLR proteins deploy distinct mechanisms in multiple subcellular compartments to activate complex downstream signaling. The molecular link between the activated NLR proteins and the diverse downstream signaling events that lead to PCD activation, ROS production and transcriptional reprogramming has remained elusive.
Ca 2+ is an essential and conserved second messenger in nearly every aspect of cellular signaling programs. Ca 2+ influx is a prerequisite for PCD triggered by AvrRpm1/AvrB-RPM1 and AvrRpt2-RPS2 interactions [24,25,26]. How the Ca 2+ signal is sensed and transduced upon NLR protein activation has remained obscure. There are three major types of Ca 2+ sensors in plants, including calmodulin (CAM), calcineurin B-like proteins and calcium-dependent protein kinases (CPKs) [27,28,29]. It has been shown that Arabidopsis CAM-like protein CML24 is required for nitric oxide (NO) production and AvrRpt2-mediated PCD [26]. CPKs have been identified ubiquitously throughout the plant kingdom and share a protein kinase domain with high sequence homology to the mammalian multifunctional CAM-dependent protein kinases, suggesting their dual function as Ca 2+ sensors and signal transducers [27,29]. Tobacco CPKs play essential roles in PCD induced by Avr9-Cf9 interaction, in which Cf9 encodes a cell-surface receptor with an N-terminal LRR domain [30,31]. Potato StCPK4 and StCPK5 directly phosphorylate and activate NADPH oxidase RBOHB (Respiratory Burst Oxidase Homologue B) [32]. There are 34 CPKs in Arabidopsis genome, which can be classified into four groups (I-IV) based on sequence similarity [27]. Recently, four Arabidopsis CPKs (CPK4/5/6/11) have been identified to play important roles, together with the MAPK cascades, in relaying primary MAMP immune signaling [33]. Distinct from the rapid and transient increase of cytosolic Ca 2+ concentration induced by MAMPs [34,35,36], inoculation with bacteria carrying avrRpm1, avrB or avrRpt2 triggered a much prolonged and sustained increase of cytosolic Ca 2+ concentration accompanied with PCD in Arabidopsis leaves [25,26]. It remains enigmatic how the distinct calcium signatures are sensed and relayed for differential and overlapping immune responses in ETI and PTI signaling.
In the present study, we have identified six Arabidopsis CPKs in sensing and transducing Ca 2+ signatures dynamically activated by RPS2 and RPM1 upon AvrRpt2 and AvrRpm1/AvrB elicitation, respectively. The specificity and redundancy of individual CPKs in NLR signaling events, including CPK4/5/6/11 in orchestrating immune gene expression, CPK1/2/4/11 in ROS production, and CPK1/2/5/6/ in PCD, were revealed by integrative biochemical, cellular, functional genomic and genetic analyses. Apparently, specific CPKs are engaged in diverse immune responses via phosphorylation and activation of different substrates in distinct subcellular compartments. Functional genomic screens identified a specific subgroup of WRKY transcription factors that act synergistically with CPKs in primary NLR signaling. Sustained activation of CPK4/5/6/11 phosphorylates WRKY8/28/48 for transcriptional reprogramming of immune genes, whereas CPK1/ 2/4/11 phosphorylate NADPH oxidases for ROS production and contribute to PCD. Our results reveal bifurcate NLR signaling mechanisms through specific, overlapping and prolonged actions of CPKs in concert with distinct substrates in multiple subcellular compartments.

PCD and immune gene activation triggered by bacterial effectors
To elucidate early signaling events in plant ETI, we have deployed an Arabidopsis mesophyll protoplast system in which pathogen-encoded individual effector genes are expressed to monitor specific and temporal responses. The cell-autonomous and synchronized elicitation in a homogeneous cell population by a single pathogen effector circumvents the complex responses simultaneously activated or/and repressed by a large array of MAMPs and effectors in intact plant-pathogen interactions [37,38]. Expression of effector gene, avrRpm1, avrB or avrRpt2, in protoplasts triggered distinct kinetics of PCD as detected by Evan's blue staining ( Figure 1A). The PCD induced by AvrRpm1 or AvrB was observed as early as 2 hr post-transfection (hpt), whereas the PCD induced by AvrRpt2 was evident at 16 hpt, reminiscent of observations with the actual plant-pathogen interactions ( Figure  S1A) [39]. PCD was not detected in the corresponding NLR mutants rpm1 and rps2 ( Figure 1A). Effector-induced PCD was accompanied by enhanced nuclear fragmentation visualized by fluorescent YO-PRO-1 iodide staining ( Figure S1B), consistent with a previous report based on direct effector protein delivery [38].
We performed a genome-wide transcriptome analysis of protoplasts expressing avrRpm1 or avrRpt2, and identified WRKY46 as an early marker gene in convergent ETI signaling (data not shown). The WRKY46 transcript was strongly induced in protoplasts expressing avrRpm1, avrB or avrRpt2 in an RPM1 or RPS2 dependent manner ( Figure 1B and S1C). The induction of WRKY46 by effectors was further confirmed with plants infected by P. syringae DC3000 (Pst) carrying avrRpm1 or avrB ( Figure 1C) and in dexamethasone-inducible avrRpt2 transgenic plants ( Figure 1D and S1D). Similar to the endogenous gene, the promoter of WRKY46 fused to a firefly luciferase reporter gene (LUC) was strongly activated by AvrRpm1, AvrB or AvrRpt2 in protoplasts ( Figure 1E). Notably, unlike PCD, effector-induced WRKY46 activation was observed to follow with similar kinetics, as early as 2 hpt, suggesting distinct mechanisms governing PCD and immune gene activation.

Differential CPK activation in ETI signaling
To elucidate the signaling mechanisms underlying PCD and gene activation triggered by different bacterial effectors, we first explored chemical inhibitors affecting various Ca 2+ channels. Consistent with previous reports, the calcium-channel blocker, LaCl 3 , suppressed effector-mediated PCD in Arabidopsis leaves inoculated with Pst avrRpm1 or avrRpt2 ( Figure S2A) [24,25]. Interestingly, effector-mediated PCD was also significantly diminished in the presence of ruthenium red (RR), which inhibits Ca 2+ release from internal stores ( Figure S2A). The similar effects of calcium-channel blockers were observed in protoplasts expressing AvrRpm1, AvrB or AvrRpt2 ( Figure 1F), validating the responses Author Summary Distinguishing self from non-self is the fundamental principle of immunity. Nucleotide-binding leucine-rich repeat (NLR) proteins were first identified in plants as disease resistance proteins and were recently found to play essential roles in mammalian innate immunity and inflammation. NLR protein complexes sense intracellular pathogenic effectors in plants and microbial patterns and danger signals in humans, but the signaling mechanisms upon NLR activation remain elusive. Using the Arabidopsis-Pseudomonas interaction as a model system, we discovered the molecular link between NLR immune sensors and the convergent immune responses triggered by distinct pathogen effectors. Integrated functional genomic and biochemical genetic screens identified six closely related Ca 2+ -dependent protein kinases (CPKs) that orchestrate bifurcate NLR immune signaling via distinct substrate specificity and subcellular dynamics. The CPK1/2 regulate the onset of programmed cell death; CPK4/5/6/11 phosphorylate specific WRKY transcription factors to regulate immune gene expression crucial for NLR-dependent restriction of pathogen growth, whereas CPK1/2/4/11 phosphorylate NADPH oxidases for the production of reactive oxygen species. Our studies decode the complex signaling mechanisms via the myriad action of CPKs downstream of NLR immune sensors.
in whole leaves and mesophyll single-cell system. These Ca 2+ inhibitors also suppressed effector-mediated WRKY46 promoter activation ( Figure 1G). Thus, both external and internal sources of Ca 2+ are essential in ETI signaling.
To investigate the potential involvement of CPKs in ETI signaling, we developed an in-gel kinase assay using histone as a general substrate. Interestingly, different effectors activated two major groups of putative endogenous CPKs with distinct molecular masses and kinetics in a Ca 2+ dependent manner ( Figure 2A). The activation of 72-kDa CPKs by AvrRpm1 or AvrB appeared stronger and occurred earlier (2 hpt) than the corresponding responses induced by AvrRpt2 (3 hpt), whereas the activation of 60-kDa CPKs displayed similar kinetics triggered by three effectors (Figure 2A). The differential CPK activation is unlikely due to the differences in the expression levels and timing of effector expression ( Figure S2B). In light of the observation that AvrRpm1/AvrB-RPM1 interaction triggers a more rapid cell death than the AvrRpt2-RPS2 interaction ( Figure 1A and S1A), we hypothesized that the 72-kDa CPKs were likely involved in regulating PCD. Importantly, effector-mediated kinase activation was not observed in the corresponding rpm1 and rps2 mutants ( Figure 2B and S2C), reinforcing the requirement for host immune sensors in transducing Ca 2+ signaling. The weak response mediated by AvrB-TAO1 [40] and AvrRpm1-RPS2 [41] might be below the threshold of detection for CPK activation. The activation of CPKs by bacterial effectors was further confirmed in Arabidopsis plants inoculated with Pst, Pst avrRpm1 or avrRpt2 ( Figure 2C). Notably, bacterial flagellin-mediated CPK activation is rather transient and peaks within 5-15 min [33]. In contrast, coincident with sustained cytoplasmic Ca 2+ elevation, effectortriggered CPK activation lasted for hours ( Figure 2A) [25,26]. In addition, unlike flagellin, AvrRpm1 and AvrRpt2 did not induce strong MAPK activation ( Figure 2D and S2D), indicating differential early signaling events in PTI and ETI. Kinase inhibitor K252a and Ca 2+ channel blockers, LaCl 3 and RR, substantially abolished the activation of putative CPKs ( Figure 2E), further confirming the requirement of Ca 2+ signaling in the kinase activation. Catalase, a decomposer of H 2 O 2 , or NO scavenger CPTIO and NO synthase inhibitor L-NNA had no effects on the kinase activation ( Figure 2E), implying that the CPK activation likely occurs upstream or independently of ROS and NO signaling, which are induced upon Pst avrRpm1 or avrRpt2 infection in Arabidopsis leaves [24,26,42].

Functional genomic screen of CPKs in ETI signaling
The predicted molecular mass of CPK1 and CPK2 in group I matches the putative 72-kDa CPKs whose activation kinetics was coincident with the onset of effector-triggered PCD, whereas the majority of the remaining CPKs falls into the range of molecular mass of 60-kDa [27]. We reasoned that if any specific CPK The data are shown as the mean 6 SE from three independent biological replicates. (D) Induction of WRKY46 in dexamethasone (DEX)-inducible avrRpt2 transgenic plants and protoplasts. The WRKY46 expression was detected 6 hr after DEX treatment with real-time RT-PCR analysis. The expression of WRKY46 was normalized to the expression of UBQ10. The data are shown as the mean 6 SE from three independent biological replicates. (E) AvrRpm1, AvrB and AvrRpt2 activated WRKY46 promoter in protoplasts. The pWRKY46-LUC was co-transfected with avrRpm1, avrB, or avrRpt2, or a vector control in protoplasts and samples were collected at indicated time points. The UBQ-GUS was included as an internal transfection control. The relative luciferase activity was normalized with GUS activity. (F) AvrRpm1, AvrB and AvrRpt2-induced cell death was suppressed by calcium inhibitors in Arabidopsis protoplasts. The avrRpm1, avrB, or avrRpt2 was co-transfected with UBQ-GUS and incubated with 1 mM LaCl 3 , 1 mM GdCl 3 or 10 mM RR. The samples were collected 16 hpt, and the cell death ratio was presented as the percentage of GUS activity repression in effector-transfected cells compared to control-transfected cells. (G) Effector-induced WRKY46 promoter activity was suppressed by calcium inhibitors in protoplasts. The samples were collected 6 hpt. The above experiments were repeated at least four times with similar results. doi:10.1371/journal.ppat.1003127.g001 functions in ETI signaling, its constitutively active (CPKac) form lacking the autoinhibitory domain [33] would likely activate ETI marker gene WRKY46 in the absence of effectors. We performed a functional genomic screen by co-expressing individual CPKac with pWRKY46-LUC in protoplasts. Among the 23 CPKs that are well expressed in Arabidopsis leaves [33], only specific CPKacs, CPKac3, 4, 5, 6, 10, 11 and 30, induced pWRKY46-LUC expression two to four fold ( Figure 2F). The expression level and kinase activity of CPKac3 are relatively higher than the other CPKacs [33]. Notably, CPKac4, 5, 6 and 11 belong to a closely related clade in subgroup I [27]. The molecular mass of CPK4, 5, 6, and 11 is around 60 kDa [33], which matches 60-kDa CPKs activated by effectors. Thus, CPK4, 5, 6, and 11 were chosen for the further studies. The kinase-dead mutants of CPKac4, 5 and 11 did not activate pWRKY46-LUC expression ( Figure 2G). CPKac1 and 2, which are likely involved in PCD regulation, did not significantly induce pWRKY46-LUC ( Figure 2F).

WRKY transcription factors act synergistically with CPKs in ETI signaling
Compared to the strong activation by effectors ( Figure 1E), CPKacs only moderately activated the WRKY46 promoter. We hypothesized that additional factors may be involved to act synergistically with CPKs for WRKY46 promoter activation in ETI signaling. Bioinformatics analysis of the putative promoter region (1.5 Kb upstream of the translational start codon) of WRKY46 identified four W-box elements that are recognized by WRKY transcription factors ( Figure 3A) [43]. Compared to the wild-type reporter, the mutation of W1 or W4 attenuated AvrRpt2mediated activation of pWRKY46-LUC ( Figure 3A), suggesting the involvement of WRKYs in ETI signaling.
The 75 Arabidopsis WRKY genes were classified into three groups with group II further divided into five subgroups [44]. We carried out a second functional genomic screen to identify WRKY candidates that could function synergistically with specific CPKs in ETI signaling. Representative WRKYs induced by Pst avrRpt2 from each WRKY group ( Figure S3A) [43] were co-expressed with CPKac5 in protoplasts for the activation of pWRKY46-LUC reporter. Remarkably, co-expression of CPKac5 and WRKY48 in subgroup IIc strongly induced the WRKY46 promoter to the same extent as that activated by effectors ( Figure 3B). Consistently, CPKac4, 6 and 11, close family members of CPKac5, but not CPKac1 and 2 that were unable to activate WRKY46 promoter ( Figure 2F), also exhibited synergistic activity with WRKY48 to induce pWRKY46-LUC ( Figure 3C and S3B). WRKY8 and 28, closely related to WRKY48 in subgroup IIc, also strongly activated pWRKY46-LUC when co-expressed with CPKac4, 5, 6

Direct phosphorylation of WRKYs by CPKs
To determine whether CPKs could directly phosphorylate WRKYs for their functional synergism, we purified full-length CPKs as Glutathione-S-Transferase (GST) and WRKYs as Maltose-Binding Protein (MBP) fusion proteins from E. coli and carried out in vitro kinase assays. Significantly, CPK4, 5 and 11, but not the kinase-dead mutants, were able to phosphorylate WRKY8, 28 and 48 in a Ca 2+ dependent manner ( Figure 4A, 4B and S4A). The conserved DNA-binding WRKY domain of WRKY8, 28 and 48 could be directly phosphorylated by CPK4, 5 and 11, but not by 10 and 30 ( Figure 4C, 4D and data not shown). The amino acid sequence surrounding T247 and T248 of WRKY48 [basic-X-T-T-X-X-X-X-hydrophobic (h)-basic] closely matches an optimal phosphorylation substrate target of CPKs (basic-h-X-basic-X-X-S/T-X-X-X-h-basic) [27]. Indeed, both T247 and T248 were phosphorylated by CPKs in vitro with mass spectrometry (MS) analysis ( Figure 4E and S4B). Interestingly, T248A, but not T247A, abolished the phosphorylation of the WRKY48 DNA binding domain by CPK4 and 5 ( Figure 4D), suggesting the functional importance of T248 in WRKY48. T248 in WRKY48 is conserved in WRKY8 and 28 ( Figure S3A). Importantly, T199 in WRKY28, corresponding to WRKY48 T248, was also phosphorylated by CPK5 with MS analysis ( Figure S4C).

Phosphorylation of NADPH oxidases by CPKs
ETI signaling is often associated with a rapid production of ROS generated by plasma membrane-resident NADPH oxidases encoded by RBOH genes in plants. Arabidopsis rbohD rbohF double mutants showed decreased ROS production and PCD in response to Pst avrRpm1 infection [45]. Potato StCPK4 and 5 phosphorylated StRBOHB and activated ROS production in tobacco leaves [32]. Surprisingly, CPKac5 and 6, the closest orthologs of StCPK4 and 5, only displayed weak phosphorylation activity on the cytoplasmic N-terminus of RBOHD or RBOHF ( Figure 4F). However, CPKac1, 2, 4 and 11, but not the kinase-dead mutants, strongly phosphorylated the cytoplasmic N-terminus of RBOHD and RBOHF in an immunocomplex kinase assay ( Figure 4F). The weak phosphorylation activity of CPKac5 and 6 on RBOHD and RBOHF was unlikely due to their overall kinase activities ( Figure  S4D). This finding was further substantiated by the full-length CPK11 phosphorylation of RBOHD and RBOHF in a Ca 2+dependent manner with an in vitro kinase assay ( Figure S4E). StCPKs phosphorylated StRBOHB at residues Ser-82 and Ser-97 [32], corresponding to Ser-133 and Ser-148 in Arabidopsis RBOHD. Mutation of Ser-148, but not Ser-133, to alanine reduced the RBOHD phosphorylation by CPK2, 4 and 11 ( Figure 4G), indicating Ser-148 as an important phosphorylation site of RBOHD by CPKs. The data suggest that specific Arabidopsis CPKs play an important role in ROS production by phosphorylating NADPH oxidases.

CPK phosphorylation enhances WRKY binding to W-boxes
The GFP fusions of CPK4, 5, 6 and 11 were observed in both cytoplasm and nucleus [33], whereas WRKY8 and 48 were mainly detected in the nucleus [46,47]. Since CPKs directly phosphorylated WRKYs, we examined the localization of CPK-GFP upon effector elicitation. Interestingly, co-expressing with AvrRpt2 enriched the strong and bright nuclear CPK5-GFP signals ( Figure 5A). The enriched nuclear GFP signal was not due to the cleavage of CPK-GFP by AvrRpt2 ( Figure S5A). Similarly, expression of AvrRpt2 under the control of a dexamethasoneinducible promoter within 2 hr was able to stimulate both CPK4-GFP ( Figure S5B) and CPK5-GFP ( Figure S5C) nuclear localization. Subcellular fractionation further confirmed a quantitative increase of CPK5-HA protein in the nucleus in the presence of AvrRpt2 ( Figure 5B). The purity of subcellular fractionations was confirmed with a-histone H3 antibody for nuclear proteins and coomassie blue staining of rubisco carboxylase (RBC) for proteins excluded from the nucleus ( Figure 5B). The data suggest that AvrRpt2 stimulates CPK5 nuclear translocation, where CPK5 phosphorylates specific WRKYs to regulate target gene transcription. The biological importance of phosphorylation was reinforced by that mutation of T248, a CPK phosphorylation residue in the DNA binding domain of WRKY48, partially compromised its ability to activate pWRKY46-LUC in the presence of CPKac4, 5 or 11 ( Figure 5C).
WRKYs bind to the W-boxes of target genes to regulate transcription. We show that WRKY48 proteins bound to the DNA oligos consisting of four W-boxes from WRKY46 promoter in a gel mobility shift assay ( Figure 5D and S6A) and quantitative chromatin immunoprecipitation-polymerase chain reaction (ChIP-PCR) assay ( Figure 5E). The binding appears specific as WRKY48 proteins did not bind to the mutated W-boxes ( Figure 5D), and the binding was largely reduced with the addition of unlabeled specific oligos, but not with nonspecific oligos ( Figure S6B). Importantly, phosphorylation of WRKY48 or 28 by CPK5 further enhanced its binding to the W-boxes ( Figure 5D and S6C). Apparently, phosphorylation is essential for Sequencing of a doubly charged peptide ion at m/z 531.22 that matches to CTpTVGCGVK of WRKY48. The confident b2 and b3 ions as well as y7 ion provide strong evidence for phosphorylation of the third Thr residue. (F) CPKacs phosphorylated RBOHD and RBOHF with an immunocomplex kinase assay. The FLAG-tagged CPKacs or the kinase-dead mutants (m) were expressed in protoplasts, and immunoprecipitated with an a-FLAG antibody for an in vitro kinase assay using GST-RBOHD or GST-RBOHF as a substrate. The proteins of RBOHD and RBOHF were shown, and the expression of individual CPKacs was detected by Western blot (bottom panel). (G) S148 is an essential phosphorylation site of RBOHD by CPKs in vitro. * indicates phosphorylated RBOHD. The numbers below indicate the relative phosphorylation level compared to WT RBOHD (set as 1) as quantified by Image J. The above experiments were repeated three times with similar results. The MS analysis was repeated twice. doi:10.1371/journal.ppat.1003127.g004 the enhanced binding activity since the kinase-dead mutant CPK5m did not potentiate WRKY28 binding to the W-boxes ( Figure S6C). Consistently, an in vitro assay revealed that CPK5 directly pulled down WRKY8 or 48, suggesting a physical interaction between specific WRKYs and CPKs ( Figure 5F). Together, the data support the synergistic roles of specific CPKs and WRKYs in WRKY46 activation in ETI signaling.

Compromised ETI signaling and pathogen resistance in cpk mutants
To examine the genetic importance of specific CPKs in ETI signaling, we characterized Arabidopsis loss-of-function cpk mutants. In addition to our previously identified cpk5, cpk6 and cpk11 single mutants and the cpk5,6 double mutants [33], we isolated cpk1 (Salk_096452) and cpk2 (Salk_059237) single mutants from the Salk T-DNA insertion collection ( Figure S7A). RT-PCR analysis confirmed that both cpk1 and cpk2 were null mutants with undetectable full-length transcripts ( Figure S7A). We did not observe overt phenotypes for any single mutants (cpk1, 2, 5, 6 and 11) in response to Pst avrRpm1 or avrRpt2 infections (data not shown). We further generated the cpk1,2 double mutants and the cpk1,2,5,6 quadruple mutants by genetic crosses. These mutants did not display any obvious growth defects under normal growth conditions. Importantly, AvrRpm1-stimulated WRKY28 phosphorylation by endogenous CPKs was reduced in the cpk5,6 mutants with WRKY28 fusion protein as a substrate in an in-gel kinase assay ( Figure 6A).
The in planta bacterial multiplication of Pst avrRpm1 or avrRpt2 increased about five to ten fold in the cpk5,6 and cpk1,2,5,6, but not cpk1,2 mutants, compared to that in WT plants ( Figure 6B). The disease symptom was also more severe in the cpk5,6 and cpk1,2,5,6 mutants than that in WT and cpk1,2 mutants ( Figure S7B). The increased susceptibility of the cpk5,6 mutants to Pst avrRpm1 or avrRpt2 was not due to a general defect in basal defense ( Figure  S7C). NLR proteins were divided into TIR (Toll-interleukin 1 receptor)-domain-containing and CC (coiled-coil)-domain-containing classes. Interestingly, the cpk5,6 and cpk1,2,5,6 mutants were also more susceptible to the infection by Pst avrRps4, mediated by TIR-type NLR RPS4 ( Figure S7D). Consistently, AvrRps4 activated expression of WRKY46 promoter ( Figure S7E). The data suggested the involvement of CPK5 and 6 in disease resistance mediated by both CC-and TIR-type NLRs. However, the cell death triggered by Pst avrRpm1 and avrRpt2 was partially compromised only in the cpk1,2,5,6, but not in the cpk1,2 or cpk5,6 mutants ( Figure S7F). We further quantified PCD using an electrolyte leakage assay. Consistently, compared to WT plants, cpk1,2,5,6 mutants showed a diminished increase in conductance, due to the release of electrolytes during cell death upon Pst avrRpm1 infection ( Figure 6C). Thus, CPK5 and 6 play roles in pathogen resistance, whereas CPK1 and 2 together with CPK5 and 6 are likely involved in the control of PCD in ETI signaling.
To obtain further genetic evidence of specific CPKs in ETImediated transcriptional reprogramming, we examined immune gene expression by pathogen effectors in cpk mutants. The WRKY46 induction by Pst avrRpm1, avrB, or avrRpt2 was abolished in the cpk5,6 mutants, but not cpk1,2 mutants ( Figure 6D), consistent with the role of CPK5 and 6 in phosphorylating specific WRKYs. Similarly, the WRKY46 transcripts induced by AvrRpm1 or AvrB in protoplasts were reduced in cpk5,6 mutants ( Figure  S7G). Infection of plants with Pst avrRpm1, avrB, or avrRpt2 also induced strong induction of SID2 gene, which was diminished in cpk5,6 mutants ( Figure 6E). Consistent with CPK1 and 2 phosphorylating RBOHD and RBOHF in vitro ( Figure 4F), the ROS production induced by Pst avrRpm1 or avrRpt2 was reduced in cpk1,2 double mutants ( Figure 6F). Together, these data provide genetic evidence that Ca 2+ signaling via specific CPKs plays pivotal roles in the diverse downstream signaling and pathogen resistance mediated by distinct intracellular NLR immune sensors.

WRKY 8 and WRKY48 as positive regulators in convergent ETI signaling
To reveal the function of WRKYs in ETI signaling, we characterized the loss-of-function wrky mutants. The wrky8-1 (Salk_107668), wrky8-2 (Salk_050194) and wrky48 (Salk_066438) mutants are null alleles with undetectable full-length transcripts ( Figure S8A) [46,47], whereas the available T-DNA insertion lines of wrky28 (Salk_007497 and Salk_092786) mutants did not significantly reduce its transcript level (data not shown). Significantly, the wrky8-1, wrky8-2 and wrky48 mutants were partially immunocompromised to Pst avrRpm1, avrRpt2 and avrB infection. The bacterial population in the wrky mutants was about five to ten fold more than that in WT plants 4 days post infection (dpi) ( Figure 7A and S8B). The disease symptom was also more pronounced in the wrky mutants than that in WT plants ( Figure  S8C). The wrky8-1, wrky8-2 and wrky48 mutants did not affect the PCD induced by Pst avrRpm1 or avrB ( Figure S8D). Our results suggest that WRKY8 and 48 play positive roles in plant ETImediated disease resistance. These findings are in contrast to the negative regulation of WRKY8 and 48 in plant basal defense to Pst infection ( Figure S8E) [46,47]. Apparently, the same transcription factors may serve distinct functions in plant PTI and ETI signaling or in response to different pathogens.
We further examined immune gene expression by pathogen effectors in wrky mutants. The WRKY46 and SID2 induction by Pst avrRpm1, avrB, or avrRpt2 was diminished in the wrky8-1 and wrky48 plants ( Figure 7B and 7C). Similarly, the effector-mediated activation of WRKY46 transcripts was reduced in the wrky8-1 and wrky48 protoplasts ( Figure S8F). The physiological and genetic analyses with cpk and wrky mutants thus substantiate the specific and overlapping functions of CPKs in phosphorylating distinct substrates for the bifurcate control of immune gene activation, PCD and ROS production ( Figure 7D).

Discussion
Plants have evolved sophisticated innate immune systems to effectively defend pathogen attacks without specialized immune cells and the adaptive immune system. Polymorphic plant NLR R proteins are intracellular immune sensors that recognize pathogenencoded effectors to initiate complex immune responses, including a sustained increase in cytosolic Ca 2+ concentration, transcriptional reprogramming, production of ROS, and PCD. Recent studies have advanced our understanding of NLR protein functions in terms of effector recognition, subcellular localization and structural determination, but the molecular mechanisms leading to the convergent immune responses upon NLR activation remain enigmatic [8,9,11,48]. In this study, we uncovered the molecular consequences of sustained Ca 2+ elevation, which leads to bifurcate signaling events controlled by specific and overlapping CPKs through phosphorylation of distinct substrates upon NLR protein activation. Two major groups of CPKs were dynamically activated by bacterial effectors AvrRpm1, AvrB and AvrRpt2. Functional genomic and biochemical analyses revealed that CPK4, 5, 6 and 11 were involved in immune gene activation, whereas CPK1 and 2, and likely 4 and 11 played key roles in the control of ROS generation, and CPK1, 2, 4, 5, 6 and 11 together contributed to PCD. CPK4, 5, 6 and 11 phosphorylated WRKY8, 28 and 48, leading to enhanced WRKY protein binding to the Wboxes of specific target gene promoters for transcriptional regulation, whereas CPK1, 2, 4 and 11 in vitro phosphorylated RBOHD and RBOHF for ROS production. Genetic and physiological characterization of multiple knockout mutants substantiated the biochemical data as cpk5,6, wrky8 and wrky48 mutants were compromised in immune gene activation and disease resistance, cpk1,2 mutants were impaired in effectorinduced oxidative burst and cpk1,2,5,6 mutants were defective in PCD. Taken together, our studies decode the specific functions of individual CPKs in the control of differential ETI responses ( Figure 7D). Our findings offer a potential molecular link for the uncoupled PCD and restriction of pathogen growth upon NLR activation [11,19,20,49].
The rapid increase of cytosolic Ca 2+ concentration has been observed in plants response to MAMPs or pathogen effectors [50]. Apparently, each signal elicits a specific calcium signature with unique kinetics, magnitude, duration and cellular compartment distribution. MAMPs, such as flagellin and PGN, activate Ca 2+ increase for 5-15 min [34], coincident with transient CPK activation [33]. However, Pst avrRpm1 or avrB elicited a Ca 2+ transient increase with a maximum about 10 min followed by a sustained increase peaked around 2 hr after infection [25]. Treatment of La 3+ , Gd 3+ and RR significantly suppressed AvrRpm1-and AvrRpt2-mediated gene activation and cell death ( Figure 1F, 1G and S2A), indicating that both extracellular and intracellular Ca 2+ release contributes to ETI signaling. It has been suggested that cyclic nucleotide-gated channels (CNGCs) function in conducting Ca 2+ to mediate PCD [24,26]. Interestingly, Arabidopsis dnd (defense no death) and hlm1 (hr-like lesion mimic) mutants, carrying mutations in CNGC2 and CNGC4 genes, exhibited aberrant PCD depending on genetic backgrounds and growth conditions [51,52,53]. The constitutive PR1 activation and enhanced pathogen resistance in the dnd and hml1 mutants may be a consequence of low intrinsic Ca 2+ levels due to CNGC mutations. It will be interesting to determine whether specific CNGCs are responsible for CPK-WRKY activation and the immune gene induction. Future studies may elucidate the precise functions of various CNGCs and other Ca 2+ channels in mediating distinct Ca 2+ signatures of extracellular and internal origins upon NLR activation.
WRKYs are a group of plant specific transcription factors involved in transcriptional reprogramming during various biological processes, in particular plant defense responses [44]. A large number of the Arabidopsis WRKY genes are transcriptionally activated upon pathogen infection [43]. Genetic analyses have indicated many WRKYs function as negative regulators in plant defense. For example, WRKY11 or 17 loss-of-function rendered plant more resistant to Pst infection [54]. Similarly, wrky8 or 48 mutants were more resistant, while overexpressors were more susceptible to Pst infection [46,47]. Despite unclear molecular mechanism of WRKY8 and 48 in plant basal defense, it is likely that WRKY8 and 48 act as repressors of plant PTI signaling. Surprisingly, our results suggest that WRKY8 and 48 play positive roles in ETI signaling since the wrky8 or 48 mutants were compromised in effector-mediated disease resistance and defense gene activation ( Figure 7A, 7B and 7C). Consistently, WRKY8, 28 and 48 were quickly and strongly activated upon Pst avrRpt2 infection independently or upstream of SA signaling [43]. The distinct functions of WRKYs in PTI and ETI signaling could be regulated at transcriptional, translational and post-translational levels in response to different stimuli. Alternatively, differential phosphorylation events mediated by distinct kinases could modulate the different immune responses in PTI and ETI signaling.
It has been suggested that PTI and ETI share downstream signaling machineries and hormonal networks [55]. Genome-wide gene expression profiling suggests that CPK4, 5, 6 and 11 mediate convergent signaling triggered by multiple MAMPs [33]. Our current study also revealed the involvement of these CPKs in ETI signaling. However, a transient Ca 2+ increase and CPK activation were observed upon MAMP treatment, whereas effectors induced sustained CPK activation (Figure 2A) [33]. Thus, the timing, amplitude and duration of differential CPK activities appear to dictate their substrate specificity and differential transcriptional reprogramming in ETI and PTI signaling. MAPK activation is a convergent MAMP signaling event [2]. MAPKs play pivotal roles and also act in parallel or synergistically with CPKs in the control of early MAMP responsive genes [33]. However, the role of MAPK cascade in ETI signaling remains unclear. We observed a strong activation of CPKs but little MAPK activation by bacterial effectors in a gene-for-gene dependent and cell-autonomous manner (Figure 2A, 2B, 2C and 2D), suggesting a predominant role of CPKs in ETI signaling mediated by RPM1 and RPS2 in Arabidopsis. It is possible that elevated CPK signaling may compromise MAPK activation in ETI signaling [56]. Nevertheless, the current data imply that activation of distinct PRRs, namely cell-surface receptor kinases recognizing MAMPs and intracellular NLR proteins recognizing pathogen-encoded effectors, initiates differential early signaling events, which trigger both overlapping and specific immune responses to maximize plant defense against pathogen attacks.

Plant growth conditions, chemical treatments and bacterial inoculation
Arabidopsis wild-type (Col-0), cpk and wrky mutant plants were grown in pots containing soil (Metro Mix 360 ) in a growth room at 23uC, 60% relative humidity and 75 mE m 22 s 21 light with a 12 hr photoperiod for approximately 4 weeks before protoplast isolation or bacterial inoculation. T-DNA insertion mutants cpk1 (Salk_096452), cpk2 (Salk_059237), wrky8-1 (Salk_107668), wrky8-2 (Salk_050194) and wrky48 (Salk_066438) were obtained from Arabidopsis Biological Resource Center (ABRC), and confirmed by PCR and RT-PCR analyses. The higher order cpk mutants were generated by genetic crosses.
Different Pst DC3000 strains were grown overnight at 28uC in the KB medium containing rifamycin (50 mg ml 21 ) or in combination with kanamycin (50 mg ml 21 ). Bacteria were pelleted by centrifugation, washed, and diluted to the desired density. The leaves were hand-inoculated with bacteria using a needleless syringe, collected at the indicated time for bacterial counting or for RNA isolation. To measure bacterial growth, two leaf discs were ground in 100 ml H 2 O and serial dilutions were plated on KB medium with appropriate antibiotics. Bacterial colony forming units (cfu) were counted 0, 2 or 4 days post incubation (dpi) at 28uC. Each data point is shown as triplicates.
At least three independent repeats were performed for all experiments. The representative data with similar results were shown. The statistic analysis was performed using the general linear model of SAS (SAS Institute, Inc., Cary, NC) with mean separations by least significant difference (LSD).

Protoplast transient assay and identification of WRKY46 as a marker gene in ETI signaling
Protoplast isolation and transient expression assay were conducted as described [37]. In general, protoplasts were collected 6 hpt for promoter activity, protein expression and kinase assays. For reporter assay, UBQ10-GUS was co-transfected as an internal transfection control, and the promoter activity was presented as LUC/GUS ratio. Protoplasts transfected with empty vector were used as effector controls.
To identify early immune genes in ETI signaling, 5 ml protoplasts at a density of 2610 5 /ml were transfected with 500 ul AvrRpm1, AvrRpt2 or a control vector (2 ug/ul). The protoplasts were collected 3 hrs after transfection for RNA isolation, cDNA and cRNA synthesis. The cRNA was fragmented for Affymetrix GeneChip (ATH1) hybridization, washing, staining and scanning at Partners HealthCare Center for Personalized Genetic Medicine (Boston, MA). Data analyses with Affymetrix GeneChip Operating Software (GCOS) and GeneSpring identified WRKY46 as one of the highest induced genes by avrRpm1 and avrRpt2 in two independent biological repeats.

Plasmid construction, recombinant protein isolation and kinase assays
Arabidopsis CPK and WRKY genes were amplified by PCR from Col-0 cDNA, and introduced into a plant expression vector with an HA or FLAG epitope-tag at the C terminus. Point mutations of pWRKY46-LUC, WRKY8, WRKY28 and WRKY48 were generated by a site-directed mutagenesis kit (Stratagene). The primer sequences for cloning and point mutations are listed in Table S1.
Different CPKs and WRKY constructs were sub-cloned into a modified GST pGEX4T-1 (Pharmacia) or MBP fusion protein expression vector pMAL-C2 (New England BioLabs) with BamHI and StuI digestion and transformed into E. coli strain BL21 (DE3). Expression of GST and MBP fusion proteins and affinity purification were performed with standard protocol, and in vitro kinase assay was carried out as described [57]. Immunocomplex kinase assay was conducted as described [37].

MS analysis
The in vitro phosphorylation for MS analysis was performed in a 10 mL reaction containing 20 mM Tris?HCl, pH 7.5, 10 mM MgCl 2 ,100 mM NaCl, 3 mM CaCl 2 , 1 mM DTT and 0.1 mM ATP. The fusion proteins of 1 mg CPK4 and 1 mg CPK5 were used to phosphorylate 10 mg of GST fusion proteins of WRKY48 DNA binding domain, and 1 mg CPK5 was used to phosphorylate 10 mg of MBP fusion proteins of WRKY28 DNA binding domain. The reaction was performed for 3 hr at room temperature with gentle shaking, and stopped by adding 46SDS loading buffer. Six individual reactions were combined and separated by 10% SDS-PAGE gel. The gel was stained with Thermo GelCode Blue Safe Protein Stain and distained with dH 2 O. The corresponding bands were cut for MS analysis, which was performed according to Avila et al. [58]. Briefly, gel bands were in-gel digested with trypsin overnight, and phosphopeptides were enriched for liquid chromatography-MS/MS analysis with a LTQ Orbitrap XL mass spectrometer (Thermo Scientific). The MS/MS spectra were analyzed with Mascot (Matrix Science; version 2.2.2), and the identified phosphorylated peptides were manually inspected to ensure confidence in phosphorylation site assignment.

Plant cell death assays
For hypersensitive response (HR) assays, the leaves of 4-weekold plants were hand-inoculated with different bacteria at 1610 8 cfu/ml, and the cell death for each genotype was calculated as the percentage of leaves showing typical HR response to total leaves inoculated.
For trypan blue staining, leaves were collected 8 hpi for Pst avrRpm1 and 16 hpi for Pst avrRpt2, and stained with trypan blue in lactophenol (Lactic acid: glycerol: liquid phenol:distilled water = 1:1:1:1) solution. The stained leaves were destained with 95% ethanol/lactophenol solution, and washed with 50% ethanol. For electrolyte leakage assays, eight leaf discs (0.5 cm diameter) were excised from the WT or cpk mutants infiltrated with bacteria and pre-floated in 10 ml of ddH 2 O for 10-15 min to eliminate wounding effect. The ddH 2 O was then exchanged and electrolyte leakage was measured using a conductivity meter (VWR; Traceable Conductivity Meter) with three replicates per time point per sample (n = 8). The YO-PRO-1 iodide was purchased from Molecular Probes/Invitrogen.

Electrophoretic mobility shift assay
Electrophoretic mobility shift assay (EMSA) was conducted as described [47] with modifications. Briefly, a pair of complementary single-stranded synthetic oligonucleotides (1.25 mM each) was end-labeled at 37uC with [c-32 P] ATP for 1 hr using T4 DNA polynucleotide kinase. The labeled oligonucleotides were mixed and annealed in TE buffer (pH 7.5) with 0.1 M NaCl at 65uC for 15 min, followed by gradual cooling to room temperature. After annealing, the double-stranded oligonucleotide probes were purified with QIAquick Nucleotide Removal kit (Qiagen). Binding reaction contains 1 ml of poly-dIdC (Roche) at 1 mg/ml, 2 ml of 56 Binding buffer (4% glycerol, 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM DTT and 10 mM Tris-HCl, pH 7.5), 1 ml of labeled probe (approximately 20,000 cpm), 1 ml cold competitor (if needed), 0.1 ml 1006 BSA (10 mg/ml) and 2.5 mg recombinant proteins. DNA-protein complexes were allowed to form at room temperature for 30 min and resolved on a 5% native polyacrylamide gel in 0.56 TBE. The gel was dried and exposed on X-ray. For the effect of CPK phosphorylation on WRKY binding activity, the MBP-WRKY proteins were subjected to the phosphorylation assay by CPKs for 1 hr prior to EMSA.

Detection of ROS production
Histological H 2 O 2 production in WT and cpk mutants upon infection with different Pst strains was examined according to the DAB staining method [59] with modifications. Briefly, WT and cpk mutant leaves were hand-inoculated with different Pst strains at 5610 7 cfu/ml for 24 hr. The leaves were excised and subsequently immersed in 1 mg/ml DAB (3,39-diaminobenzidine, Sigma) (pH 3.8) solution with low vacuum pressure for 30 min, followed by an overnight incubation at room temperature in the dark. The stained leaves were fixed and cleared in alcoholic lacto-phenol (95% ethanol : lactic acid : phenol = 2 : 1 : 1) at 65uC, rinsed once with 50% ethanol, and twice with H 2 O. The destained leaves were stored in 50% glycerol or subjected to microscope observation.

Subcellular localization and nuclear fractionation
C-terminal GFP fusion of CPK5 was co-transfected with a vector control or avrRpt2. Protein localization was observed 12 hpt with a confocal microscopy. The nucleus was indicated with a cotransfected nuclear-localized RFP.
The immunoprecipitated proteins were analyzed by Western blot with an a-HA antibody.

Real-time RT-PCR
Total RNA was isolated from leaves or protoplasts after treatment with TRIzol Reagent (Invitrogen). Complementary DNA was synthesized from 1 mg of total RNA with 0.1 mg oligo (dT) primer and reverse transcriptase (New England BioLabs). Real-time RT-PCR analysis was carried out using iTaq SYBR green Supermix (Bio-Rad) supplemented with ROX in an ABI GeneAmp PCR System 9700. The expression of immune genes was normalized to the expression of UBQ10. The primer sequences of different effectors and RT-PCR are listed in Table  S1.
Protoplast ChIP assays 5 ml of protoplasts were transfected with WRKY48-HA or WRKY8-HA and incubated for 4 hrs. Cells were crosslinked with 1% formaldehyde for 20 min and quenched by glycine for 5 min. Nuclei were extracted freshly as described [60] and the rest of ChIP was performed as described (http://sites.bio.indiana.edu/ ,pikaardlab/Protocols%20page.html) with some modifications. Bioruptor (Diagenode) was used for sonication and DNA was eluted with 1% SDS and 0.1 M NaHCO 3 at 65uC for overnight. Anti-HA antibody (Roche) was used. The quantitative PCR primers have similar efficiency. The relative enrichment fold changes were calculated by normalizing % input of each primer pair against the control gene primer (CAB1). Specificity of WRKY48 binding to the W-boxes in vitro. The recombinant WRKY48 protein was incubated with 32 P-labeled W-boxes in a gel mobility shift assay. Specific competitor (S. C.) was non-labeled W-boxes oligonucleotide. Non-specific competitor (N.C.) was a random oligonucleotide. (C) Kinase activity is required for CPK-enhanced WRKY28 binding to the W-boxes in vitro. CPK phosphorylation of WRKY28 was conducted prior to DNA binding assay. (TIF) Figure S7 Analysis of cpk mutants. (A) T-DNA insertion sites and RT-PCR analysis in cpk1 and cpk2 mutants. (B) The disease phenotype of WT and cpk mutant plant by Pst avrRpm1 or avrRpt2 infection. Plant leaves were hand-inoculated with bacteria at 5610 5 cfu/ml. The picture was taken at 5 dpi. (C) The cpk5,6 mutant plants were compromised in avrRpm1and avrRpt2mediated disease resistance. Plant leaves were hand-inoculated with Pst, Pst avrRpm1 or Pst avrRpt2 at 5610 5 cfu/ml. The bacterial growth was measured 2 dpi. The data are shown as mean 6 SE of three repeats, and the asterisk (*) indicates a significant difference with p,0.05 when compared with data from WT plants. (D) The cpk5,6 mutant plants were compromised in avrRps4-mediated disease resistance. Plant leaves were hand-inoculated with Pst avrRps4 at 5610 5 cfu/ml. The bacterial growth was measured 3 dpi. The data are shown as mean 6 SE of three repeats, and the asterisk (*) indicates a significant difference with p,0.05 when compared with data from WT plants. (E) AvrRps4 activated WRKY46 promoter in protoplasts. The pWRKY46-LUC was cotransfected with AvrRpm1, AvrRps4 or a vector control in protoplasts and samples were collected at 6 hpt. The UBQ-GUS was included as an internal transfection control. The relative luciferase activity was normalized with GUS activity. (F) The cpk1,2,5,6 mutant plants diminished effector-mediated cell death. Plant leaves were hand-inoculated with Pst avrRpm1 or avrRpt2 at 1610 8 cfu/ml. The cell death ratio was recorded for avrRpm1 at 8 hpi and avrRpt2 at 16 hpi. The leaves were further stained with trypan blue to detect cell death. (G) Effector-induced WRKY46 expression was reduced in cpk mutant protoplasts. WRKY46 expression was detected in protoplasts 3 hpt by real-time RT-PCR analysis. The expression of WRKY46 was normalized to the expression of UBQ10. The data are shown as the mean 6 SE from three independent biological replicates. (TIF) Figure S8 Analysis of wrky mutants. (A) T-DNA insertion sites and RT-PCR analysis in wrky8 and wrky48 mutants. (B) The bacterial growth of Pst avrB in wrky mutant plants. Plant leaves were hand-inoculated with Pst avrB at 5610 5 cfu/ml. The bacterial growth was measured at 3 dpi. The data are shown as mean 6 SE of three repeats, and the asterisk (*) indicates a significant difference with p,0.05 when compared with data from WT plants. (C) The disease phenotype of WT and wrky mutant plants by Pst avrRpm1 or avrRpt2 infection. Plant leaves were handinoculated with different bacteria at 5610 5 cfu/ml and the pictures were taken at 6 dpi. (D) The cell death of wrky mutant plants. Plant leaves were hand-inoculated with Pst avrRpm1 or avrB at 1610 8 cfu/ml. The cell death ratio was recorded at 10 hpi, and indicated with the percentage (%) of wilting leaves of total inoculated leaves. (E) The wrky mutant plants are resistant to Pst infection. Plant leaves were hand-inoculated with Pst at 5610 5 cfu/ml. The bacterial growth was measured at 3 dpi. The data are shown as mean 6 SE of three repeats, and the asterisk (*) indicates a significant difference with p,0.05 when compared with data from WT plants. (F) Effector-induced WRKY46 expression was reduced in wrky mutant protoplasts. WRKY46 expression was detected in protoplasts 3 hpt by real-time RT-PCR analysis. The expression of WRKY46 was normalized to the expression of UBQ10. The data are shown as the mean 6 SE from three independent biological replicates. (TIF)