Pseudomonas syringae effector HopZ3 suppresses the bacterial AvrPto1–tomato PTO immune complex via acetylation

The plant pathogen Pseudomonas syringae secretes multiple effectors that modulate plant defenses. Some effectors trigger defenses due to specific recognition by plant immune complexes, whereas others can suppress the resulting immune responses. The HopZ3 effector of P. syringae pv. syringae B728a (PsyB728a) is an acetyltransferase that modifies not only components of plant immune complexes, but also the Psy effectors that activate these complexes. In Arabidopsis, HopZ3 acetylates the host RPM1 complex and the Psy effectors AvrRpm1 and AvrB3. This study focuses on the role of HopZ3 during tomato infection. In Psy-resistant tomato, the main immune complex includes PRF and PTO, a RIPK-family kinase that recognizes the AvrPto effector. HopZ3 acts as a virulence factor on tomato by suppressing AvrPto1Psy-triggered immunity. HopZ3 acetylates AvrPto1Psy and the host proteins PTO, SlRIPK and SlRIN4s. Biochemical reconstruction and site-directed mutagenesis experiments suggest that acetylation acts in multiple ways to suppress immune signaling in tomato. First, acetylation disrupts the critical AvrPto1Psy-PTO interaction needed to initiate the immune response. Unmodified residues at the binding interface of both proteins and at other residues needed for binding are acetylated. Second, acetylation occurs at residues important for AvrPto1Psy function but not for binding to PTO. Finally, acetylation reduces specific phosphorylations needed for promoting the immune-inducing activity of HopZ3’s targets such as AvrPto1Psy and PTO. In some cases, acetylation competes with phosphorylation. HopZ3-mediated acetylation suppresses the kinase activity of SlRIPK and the phosphorylation of its SlRIN4 substrate previously implicated in PTO-signaling. Thus, HopZ3 disrupts the functions of multiple immune components and the effectors that trigger them, leading to increased susceptibility to infection. Finally, mass spectrometry used to map specific acetylated residues confirmed HopZ3’s unusual capacity to modify histidine in addition to serine, threonine and lysine residues.


Introduction
The plant pathogen Pseudomonas syringae uses type III-secreted proteins to promote its growth during infection of plants. These effector proteins are injected into plant cells, where they often interfere with plant defense signaling either through binding, post-translational modifications (PTMs) and/or destabilization of host factors [1,2]. A major mechanism to suppress P. syringae growth is signaling mediated by plant immune receptors that monitor specific perturbations caused by effectors. A well-studied example of such a receptor is Arabidopsis RESISTANCE TO P. SYRINGAE MACULICOLA 1 (RPM1), a member of the NUCLEOTIDE BINDING-LEUCINE RICH REPEAT (NB-LRR) protein family. Recognition and signaling occur when RPM1 senses a specific phosphorylation (mainly p-T166) of RPM1-INTERACT-ING PROTEIN 4 (RIN4), an intrinsically disordered hub protein [3]. Two unrelated effectors, AvrB or AvrRpm1, from different P. syringae strains can strongly trigger RPM1 signaling and are thus considered avirulence factors. These effectors cause the cytoplasmic RIN4-INDUCED PROTEIN KINASE (RIPK) and probably additional kinases to phosphorylate RIN4. RIN4 is also involved in promoting defense signaling in response to conserved microbial patterns. Immune responses are induced by phosphorylations of specific RIN4 residues that are triggered by recognition of effectors or microbial patterns [3][4][5][6].
Pseudomonas syringae pv. syringae B728a (PsyB728a) is a bean pathogen that can also grow to moderate levels on Arabidopsis and tomato without causing overt disease symptoms [7,8]. In Arabidopsis, PsyB728a with a deletion of the type III secreted effector HopZ3 (PsyΔHopZ3) causes the activation of RPM1 signaling. This occurs via two interacting effectors with homology to AvrB and AvrRpm1: AvrB3 Psy and AvrRpm1 Psy . In the context of PsyΔHopZ3 infection, both effectors are needed to activate signaling [9]. HopZ3 belongs to the YopJ acetyltransferase family that comprises several effectors from animal and plant pathogens. The acetyltransferase activity of HopZ3 is necessary for suppression of RPM1 activation in Arabidopsis and several components of the RPM1 immune-effector complex are substrates of HopZ3 [9]. HopZ3 acetylates the activation loop and active site residues of RIPK, which inhibits its ability to phosphorylate RIN4. Additionally, acetylation of RIN4 prevents its phosphorylation by RIPK. HopZ3 also acetylates residues in AvrB3 that are predicted to disrupt hydrogen bonds at the key interaction sites with RIN4. Thus, HopZ3 suppresses plant immunity through modification of both Arabidopsis and bacterial proteins that act in the same complex.
Interestingly, in a large screen for interactions between effectors and plant immune signaling proteins ( [9], https://charge.ucdavis.edu/charge_db/interaction/Y2H/Y2H_interaction. php), we found that HopZ3 interacted with the resistance-inducing effector AvrPto1 Psy and its tomato targets, PTO-like proteins. Moreover, HopZ3 suppressed AvrPto1 Psy -induced cell death in Nicotiana benthamiana [8]. That suggested that HopZ3 may affect tomato immunity. The interaction between PsyB728a and tomato has not been well characterized; however, resistance to P. syringae pv. tomato has been studied in great detail. Resistant tomato lacks RPM1 but contains PSEUDOMONAS RESISTANCE AND FENTHION SENSITIVITY (PRF), an NB-LRR protein that forms complexes with the kinases PSEUDOMONAS SYRINGAE PV TOMATO RESISTANCE (PTO) and FENTHION SENSITIVITY (FEN) and recognizes effectors AvrPto and AvrPtoB from P. syringae pv. tomato and other pathovars [10]. PTO, FEN and related cytoplasmic protein kinases in the same family as RIPK show natural variation that affects their functional specificity in promoting immunity in different tomato accessions [11]. PTO and FEN interact differently with AvrPto and AvrPtoB. Both effectors can bind to PTO and elicit PRF-dependent immune signaling [12][13][14][15]. In contrast, FEN can bind and be activated by AvrPto if the key residue N202 (that corresponds to T204 in PTO) is substituted with threonine [16]. Truncated versions of AvrPtoB (e.g., AvrPtoB 1-387 ) bind to FEN and stimulate immunity; however, due to the C-terminal E3 ubiquitin ligase domain, full-length AvrPtoB causes proteasome-dependent FEN degradation and does not trigger FEN/PRF immunity [14]. Structure-based biochemical analysis has indicated that AvrPto-PTO binding is a key step that leads to activation of PRF signaling [17]. The kinase activity of PTO is important for disease resistance triggered by AvrPto [18][19][20][21][22]. PTO acts as a dimer or higher order complex together with PRF [17,22,23]. Although AvrPto can inhibit PTO and other kinases [17], transphosphorylation between unbound PTO molecules and those bound to AvrPto is thought to be needed for downstream signaling [17,22,23].
Another potential player in PTO/PRF-conferred immunity is SlRIN4-1, one of three RIN4-related proteins in tomato. Infection with P. syringae pv. tomato strain T1 engineered to express AvrPto causes reduction of SlRIN4 protein levels. Downregulation of SlRIN4-1 using RNAi decreases the growth of strain T1 carrying AvrPto but not the growth of strain T1 alone [24]. Thus, downregulation of SlRIN4-1 seems to specifically enhance PTO-dependent resistance. Moreover, N. benthamina homologue of RIN4 was found in a search for proteins proximal to AvrPto, suggesting their interaction [25].
PsyB728a has AvrPto and AvrPtoB homologues (AvrPto1 Psy and AvrPtoB Psy /HopAB1, hereafter called AvrPtoB Psy ) that induce resistance in tomato. Transfer of a plasmid carrying AvrP-to1 Psy to a P. syringae pv. syringae strain that lacks AvrPto and AvrPtoB (Psy61) confers PTOdependent recognition, whereas plasmid-borne AvrPtoB Psy confers some PTO-independent recognition that involves other members of PTO family [26]. AvrPto1 Psy is 88% identical at the amino acid level with AvrPto Pto while AvrPtoB alleles share 52% identity. Both AvrPto1 Psy and AvrPtoB Psy can interact with PTO in a yeast two-hybrid assay [26]. Consistent with these findings, PRF is a major factor that restricts the growth of PsyB728a on tomato [10,26].
We previously found that deletion of HopZ3 decreased the growth of Psy on tomato with functional PTO [7], raising the possibility that HopZ3 normally suppresses effector-triggered immunity in tomato. In this study, we investigated this hypothesis. Through genetics and biochemical reconstruction, our data point to a mechanism that involves immune suppression via acetylation of AvrPto1 Psy , PTO and other immunity factors.

HopZ3 suppresses PTO/PRF defenses triggered by AvrPto1 Psy
PsyB728a has a strong epiphytic growth phase modulated by effectors [7]. P. syringae effectors, including AvrPto Pto , are predominantly expressed by bacteria on a leaf surface and delivered to epidermal cells during infection, where they can induce and suppress defenses [7,27]. Deletion of HopZ3 reduced epiphytic growth of PsyB728a in a resistant tomato PtoR (76R), which has a functional PTO [7]. In a transient expression assay in N. benthamiana, HopZ3 suppressed AvrPto1 Psy -induced cell death, a proxy for immune activation [7,8]. Therefore, it seemed plausible that the effect of HopZ3 on the growth of PsyB728a in tomato is dependent on PTO and PRF proteins needed for recognition and resistance triggered by AvrPto1 Psy . Bacterial growth of PsyB728a and PsyΔHopZ3 was indistinguishable in pto11 and prf3 plants lacking functional PTO and PRF, respectively, indicating that the PTO/PRF pathway is needed for the effect of HopZ3 ( Fig 1A). As expected, deletion of HopZ3 similarly restricted total (epiphytic + endophytic, Fig 1A and 1C) and epiphytic (Fig 1B and 1D) populations of PsyB728a in PtoR tomato and we tested these populations interchangeably in further experiments. The growth defect of PsyΔHopZ3 was restored only when a plasmid carrying wild-type HopZ3 but not a catalytically inactive version (HopZ3_C300A) was introduced (Fig 1B). HopZ3 and HopZ3_C300A proteins in these strains are produced at the same level in PsyΔHopZ3 [7]. These results suggest that enzymatically active HopZ3 suppresses PTO-mediated plant immunity in tomato.
A possible explanation for why PTO is needed to observe HopZ3's effect on promoting PsyB728a growth is that HopZ3 suppresses AvrPto1 Psy recognition. If this is true, the effect of deleting HopZ3 should be reversed when AvrPto1 Psy is also deleted. To test this hypothesis, we assessed the growth of a double mutant of PsyB728a that lacks both HopZ3 and AvrPto1 Psy in PtoR tomato. Both total ( Fig 1C) and epiphytic ( Fig 1D) populations of PsyΔHopZ3ΔAvrPto1 Psy were increased relative to PsyΔHopZ3 to levels similar to WT PsyB728a. The effect of deleting AvrPto1 Psy was complemented when the double mutant was transformed with a plasmid carrying AvrPto1 Psy (Fig 1D). Deletion of AvrPto1 Psy in PsyB728a with intact HopZ3 had no effect on the growth of PsyB728a in PtoR tomato (Fig 1E), as previously reported [28]. AvrPto1 Psy did not confer resistance in pto11 plants due to lack of functional PTO, regardless of the presence of HopZ3 ( Fig 1F). Altogether, our genetic analysis indicates that HopZ3 suppresses AvrPto1 Psy -triggered immunity during PsyB728a infections.

HopZ3 interacts with SlRIN4s, tomato kinases PTO, FEN, SlRIPK and effectors that target PTO
To investigate the molecular mechanisms of HopZ3 suppression of tomato immunity, we performed a screen for HopZ3 and AvrPto1 Psy interacting proteins using a semi-automated yeast two-hybrid analysis ( [9], https://charge.ucdavis.edu/charge_db/interaction/Y2H/Y2H_ interaction.php). Initial yeast experiments indicated interactions of HopZ3 with SlRIN4-1, SlRIN4-2, PTO homologous protein2 (PTH2), PTO homologous protein4 (PTH4), FEN, AvrPto1 Psy and AvrPtoB Psy . We followed up on a subset of these proteins and also tested additional candidate proteins (S1 Fig and Table 1). Although HopZ3 and PTO did not show an interaction in the yeast two-hybrid assays ( [7]; S1 Fig), they interacted in an in vitro pull-down assay and in planta bimolecular fluorescence complementation (BIFC) analysis (Table 1 and  Figs 2A and S2). In addition, HopZ3 interacted with FEN, tomato RIN4 homologues (SlRIN4-1, -2 and -3), the bacterial effectors AvrPto1 Psy and AvrPtoB Psy in in vitro pull-downs and in planta and with SlRIPK in yeast and in planta (Figs 2 and S1 and S2 and Table 1).
HopZ3 and AvrPto1 Psy displayed similar protein-protein interaction profiles. AvrPto1 Psy directly interacted with the same tomato kinases and SlRIN4s as HopZ3 in at least one of the assays (Table 1 and S1-S3 Figs), which suggests these proteins are common targets for both effectors. As expected, recombinant AvrPto1 Psy could directly bind to PTO in vitro (Table 1 and S3A Fig), similarly to what was shown for AvrPto Pto [17]. We also detected a weak signal  (Table 1 and S2 and S3B Figs). In addition to HopZ3, AvrPto1 Psy also interacted with AvrPtoB Psy in yeast two-hybrid and in vitro pull-down assays (Table 1 and S1 and S3F Figs). Many of HopZ3 interacting proteins interacted with each other (S1 and S2 Figs). These data show that HopZ3 directly targets the AvrPto-PTO defense pathway in tomato.

HopZ3 acetylates a subset of interacting proteins
Since HopZ3 has acetyltransferase activity [9], we tested whether several interacting proteins were its substrates in vitro, in reactions with 14 C-acetyl-CoA and the cofactor inositol hexakisphospate (IP6). Recombinant HopZ3, but not the catalytically inactive variant HopZ3_C300A, acetylated AvrPto1 Psy and its target PTO, SlRIPK, SlRIN4-1, SlRIN4-2 and SlRIN4-3 ( Fig 3A  and 3B). There was no detectable acetylation of FEN by HopZ3 (Fig 3B). Although AvrPtoB Psy was capable of binding to HopZ3, it was not a good substrate for acetylation ( Fig 3C). Despite diversity of substrates, HopZ3 activity is specific, as the enzyme does not acetylate interacting proteins MPK4 [9], FEN and AvrPtoB Psy or non-interacting HopI Psy [9].

HopZ3 acetylates AvrPto1 Psy residues essential for interaction with PTO and decreases phosphorylation of residues involved in defense activation
To gain further insight into molecular mechanisms of immune suppression by HopZ3, we analyzed post-translational modifications of AvrPto1 Psy produced in E. coli and N. benthamiana by LC-MS/MS. By comparing acetylation sites found in E. coli-produced AvrPto1 Psy after in vitro acetylation reactions with 13 C-acetyl-CoA, IP6 and HopZ3 or HopZ3_C300A, we found that H125 and H130 were specifically acetylated by HopZ3 (S1 Table). These histidine residues were also specifically acetylated in planta, when AvrPto1 Psy and HopZ3 were coexpressed in N. benthamiana. Several other AvrPto1 Psy residues were acetylated in vitro and in planta to higher levels in the presence of HopZ3 compared to HopZ3_C300A (S1 Table and Figs 4 and S4). T91 and S94 in the AvrPto1 Psy GINP Ω loop that is essential for interaction with PTO [15,17,29,30] were consistently found to be the most highly acetylated in several experiments (S1 Table). S46, which is also important for interaction with PTO [15,29,30] and the virulence function of AvrPto Pto [31], was also acetylated by HopZ3. This residue is not in the binding interface, but likely stabilizes the protein fold [30].

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation Many residues in AvrPto1 Psy produced in E. coli or in N. benthamiana were phosphorylated (S1 Table and Figs 4 and S4). Interestingly, S136 was very highly phosphorylated in planta (regardless of the presence of HopZ3), but it was not phosphorylated in the recombinant

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation protein. This plant modification of AvrPto has not been reported previously; its functional significance is unknown and was not further explored. Since HopZ3 also targets serines and threonines, the same residues may also be phosphorylated. S147 and S149 of AvrPto1 Psy were phosphorylated in vitro and in planta, and HopZ3 acetylated a fraction of these residues as well. Importantly, in N. benthamiana expressing HopZ3, phosphorylation of S147 and/or S149 was significantly reduced (S1 Table). These residues were previously shown to be phosphorylated and contribute to the avirulence activity of AvrPto Pto during interactions with resistant tomato [32] and Nicotiana sp. [33], as well as to virulence during susceptible tomato infection [32]. In our LC-MS/MS analysis, we also directly detected myristoylation of G2, a modification that enables membrane localization of AvrPto [32] (S1 Table and Figs 4

and S4).
Acetylation of residues in the AvrPto1 Psy Ω loop that interacts with PTO and decreased phosphorylation of residue(s) involved in signaling likely contribute to the mechanism by which HopZ3 reduces the immune response to AvrPto1 Psy .

Residues acetylated by HopZ3 are important for AvrPto1 Psy avirulence during tomato infection
Many residues acetylated by HopZ3 are important for the ability of AvrPto1 Psy to trigger a defense response in resistant tomato. For example, S94 and S147/S149 in AvrPto Pto were shown to contribute to triggering PTO-mediated disease resistance and were extensively studied, as discussed above. Although T91 in the GINP O loop was not found to affect interaction  Fig 5A). Importantly, AvrPto1 Psy variants were expressed in PsyB728a to similar levels as wild-type AvrPto1 Psy (Fig 5C). Therefore, the residues acetylated by HopZ3 are important for the ability of AvrPto1 Psy to trigger a defense response in resistant tomato. planta experiment that are important for immune signaling. Models were developed using the iTASSER modeling server and algorithm. Major acetylation sites dependent on HopZ3 are shown in red, important phosphorylation sites in blue, sites either acetylated or phosphorylated in purple, known sites of interaction between AvrPto Pto and PTO in yellow, acetylated interaction sites in orange and G2 myristoylation site in green. See also S1 and S2 Tables and S4 and S5 Figs. HopZ3 acetylates sites essential for interaction (orange) and decreases phosphorylation of residue(s) involved in signaling (blue box). (C) Model of HopZ3 acetylation sites in the crystal structure of PTO:AvrPto Pto contact site [17]. AvrPto is shown in green with residues acetylated by HopZ3 in red, and PTO is shown in orange with sites acetylated by HopZ3 in blue. Modifications on either protein are in the known interaction area of the two proteins. https://doi.org/10.1371/journal.ppat.1010017.g004

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation

HopZ3 acetylates key sites in the activation loop and other residues important for the immune function of PTO and reduces their phosphorylation
We used an LC-MS/MS analysis of PTO to gain insight into what specific effect acetylation might have. By comparing acetylation sites found in the presence of HopZ3 and HopZ3_C300A after in vitro acetylation reactions with 13 C-acetyl-CoA, we identified T204 in the P+1 activation loop/region of PTO as a specific HopZ3-mediated acetylation site (S2 Table  and S5 Fig). T204 is a cognate of T257 in Arabidopsis RIPK, another member of this kinase family that we found to be acetylated by HopZ3 [9].
T204 and T199 were the major acetylation sites in planta in PTO immunoprecipitated from N. benthamiana that also expressed functional HopZ3 (S2 Table and Figs 4 and S5). Both of these residues in the P+1 loop are important for interaction with AvrPto [16,17,20,22]. In addition, the structurally proximal residue K123 was acetylated in PTO coexpressed with HopZ3 in planta. Moreover, phosphorylation of S198/T199 (and T190) was reduced in the presence of HopZ3 compared to HopZ3_C300A (S2 Table and Figs 4 and S5). Since phosphorylation of S198 and T199 is necessary for immune signaling [17,22,23], this may be a part of the mechanism by which HopZ3 reduces the plant defense response to AvrPto1 Psy .

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation

Acetylation of AvrPto1 Psy and PTO affect their binding
A key step in the activation of AvrPto Pto -triggered immunity requires its binding to PTO [19]. We hypothesized that modification by HopZ3 may affect the AvrPto1 Psy -PTO interaction because HopZ3 targets several residues in the binding interface (Fig 4 and S1 and S2 Tables). Therefore, we assayed the impact of AvrPto1 Psy or PTO acetylation on their interaction by performing in vitro acetylation reactions with HopZ3 followed by binding experiments. We found that binding was reduced when either AvrPto1 Psy or PTO was acetylated (Fig 6). Thus, part of the HopZ3 mechanism of immune suppression involves inhibition of the formation of the AvrPto1 Psy -PTO complex through their modification.

Amino acid substitutions in PTO and FEN alter their acetylation specificity
FEN has an asparagine (N202) at the cognate position to T204 in PTO. Conversion of T204 to N in PTO abolished the acetylation of the protein by HopZ3 in vitro ( Fig 7A). Conversely, mutating N202 to T in FEN rendered it susceptible to acetylation by HopZ3 (Fig 7B). The same amino acid substitutions switched the signaling specificity of PTO and FEN in response to AvrPto Pto as assessed by cell death induction in transient expression experiments in N. benthamiana [16]. The loss of in vitro acetylation of PTO_T204N by HopZ3 is consistent with our finding of only one in vitro acetylation site in PTO by LC-MS/MS (S2 Table).
Amino acid substitutions at position 204/202 greatly affected kinase activities of PTO and FEN, respectively. PTO and FEN variants with the T at 204/202 had higher kinase activity and showed more autophosphorylation than the N or R versions (Fig 7C and 7D; [17]). Together our data suggest that HopZ3 targets an essential residue in PTO that differentiates it from FEN in immune activation ability.

HopZ3 acetylates multiple sites in SlRIN4s and SlRIPK
We analyzed modifications of tomato RIN4s and RIPK acetylated in vitro by HopZ3 using 13 C-acetyl-CoA and found many residues to be acetylated by HopZ3 (S3 and S4 Tables). We did not observe common modified sites among all three SlRIN4 paralogues and AtRIN4; however, these proteins are not highly conserved ( [9], S7 Fig). The lack of conserved acetylations may also result from the intrinsically unstructured nature of RIN4s. We found one residue that is acetylated in tomato and Arabidopsis: S88 in SlRIN4-1/S79 in AtRIN4, respectively. This residue is conserved among RIN4s from many species [9,34]. The main regulatory phosphorylation sites corresponding to AtRIN4, T166 and S141 [6] were not acetylated by HopZ3 in tomato or Arabidopsis.
The major acetylation sites in AtRIPK [9] were acetylated by HopZ3 in the tomato ortho-logue. Similar to Arabidopsis, these sites could often be also phosphorylated (S8 Fig). In particular, SlRIPK K120 (K122 in AtRIPK) in the ATP binding site, S219 (S221 in At) near the ATP binding site, SlRIPK S249/T250 (At S251/T252) in the activation loop and T255/H256 (T257 in At) were specifically acetylated by HopZ3 in both species; in addition, the serines/ threonines were phosphorylation sites. K122 and S251/T252 in AtRIPK are necessary for RIPK activity [9] and S251/T252 are uridylated by the Xanthomonas effector AvrAC leading to RIPK inhibition [35]. Moreover, SlRIPK S249/T250 (At S251/T252) correspond to PTO S198/T199, whose phosphorylation is important for immunity [17,22,23] and is decreased by HopZ3 (S2 Table). The highest acetylation by HopZ3 was observed for SlRIPK T255, which corresponds to acetylated T257 in Arabidopsis RIPK and T204 in the PTO activation loop. Therefore, HopZ3 targets important residues conserved in SlRIPK, AtRIPK and PTO and directly acetylates SlRIPK residues necessary for kinase activity, acetylation of which may compete with phosphorylation.

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation

PTO, FEN and SlRIPK phosphorylate HopZ3 and SlRIN4s, and are differentially affected by HopZ3 acetylation
We tested whether kinases from the RIPK family that interact with HopZ3 can phosphorylate HopZ3 and its putative targets, SlRIN4s. Indeed, PTO, FEN and SlRIPK phosphorylated HopZ3 and SlRIN4s in vitro (Figs 7C, 7D and S9 and 8).
Next, we performed acetylation reactions with HopZ3 or HopZ3_C300A followed by kinase reactions. This permitted us to test the effect of acetylation on kinase activities. Acetylation of

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation SlRIPK greatly reduced its kinase activity and phosphorylation of SlRIN4s and HopZ3 ( Fig  8A-C), similar to what we observed with Arabidopsis RIPK [9]. These results confirm that HopZ3 targets SlRIPK sites that are important for activity (S8 Fig). As expected, incubation of FEN with HopZ3 in the acetylation reaction did not affect the autophosphorylation activity of FEN ( Fig 8D); however, HopZ3 phosphorylation was lower than HopZ3_C300A, possibly due to autoacetylation of HopZ3. We expected that PTO activity may be suppressed by acetylation because an R substitution at T204, the residue acetylated by HopZ3, reduced its activity ( Fig  7C). However, PTO kinase activity was not strongly affected by acetylation (Fig 8E and 8F). These data show a complex network of reciprocal modifications of HopZ3 and its substrates and suggest that HopZ3 does not exert its immune-suppressing effect by direct inhibition of PTO kinase activity.

Discussion
In this study, we explored the hypothesis that the HopZ3-dependent mechanism of suppressing effector immune induction is conserved in diverse plant species, even when the effectors triggering defenses and components of the plant immune complexes are different. In resistant tomato, phosphorylation plays a prominent role in immune activation, with phosphorylated residues in effector and plant proteins promoting signaling [17,22,23,32,33]. The PTO kinase binds to the AvrPto effector, leading to the robust PRF-dependent restriction of bacterial growth. This study points to several mechanisms by which HopZ3 disrupts the PTO pathway,

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation as outlined in the model in Fig 9. In one mechanism, acetylation of residues at the binding interface of AvrPto1 Psy (T91, S94) and PTO (T199, T204) or other residues needed for binding (S46 in AvrPto1 Psy ), disrupt the AvrPto1 Psy -PTO interaction and subsequent immune

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation responses. Acetylation can also directly compete at other sites for phosphorylation events that promote activity/signaling of the targets. For example, S147/S149 residues in AvrPto1 Psy and T199 in PTO are acetylated in planta, and phosphorylation of these residues is decreased in the presence of active HopZ3. An additional mechanism could be inactivation of kinases by acetylation; HopZ3 may also inhibit the unknown plant kinase(s) that phosphorylates AvrPto1 Psy . It is also possible that acetylated AvrPto1 Psy is a poor kinase substrate. Although we did not observe in vitro suppression of PTO kinase activity by acetylation, it might be affected in planta, where more residues in the activation domain are acetylated.  [23]; it is unknown whether RIN4 and/or RIPK are in a complex. AvrPto and RIN4 were shown to be associated with the membrane, but it is not known where the interactions occur. Upon infection with P. syringae containing AvrPto1 Psy but not HopZ3, AvrPto1 Psy becomes phosphorylated and binds to a PTO molecule, inhibiting its activity. Another molecule of PTO can autophosphorylate and transphosphorylate PTO bound to AvPto1 Psy . The AvrPto1 Psy -PTO interaction and phosphorylations cause PRF activation and initiation of effector-triggered immunity [23]. RIN4 interacts with AvrPto1 Psy and may be phosphorylated by RIPK and/or PTO during infection and contribute to signaling. In the presence of HopZ3, acetylation of AvrPto1 Psy , PTO, RIPK and RIN4 leads to reduced phosphorylation and suppression of AvrPto1 Psy -PTO complex formation, ultimately resulting in disruption of effector-triggered immunity. https://doi.org/10.1371/journal.ppat.1010017.g009

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation In addition to acetylation at serine, lysine or threonine typically seen with YopJ family acetyltransferases, HopZ3 can also modify histidine [9]. Here we confirmed this unusual activity of HopZ3, as several histidines in AvrPto1 Psy and SlRIPK were acetylated. In AvrPto1 Psy , H125/H130 residues are targets of HopZ3 acetylation and are required for the immune-inducing activity of AvrPto1 Psy in tomato. A similar observation was made in AvrB3, where substitution of H221 mitigated defense activation [9]. Although AvrPto1 Psy histidine substitution did not alter protein stability or binding to PTO, these sites might facilitate other protein dynamics or binding to different immune components.
Residues corresponding to T204, S198 and T199 in PTO were also acetylated by HopZ3 in RIPK from Arabidopsis [9] and tomato, interfering with phosphorylation and decreasing RIPK activity. Both PTO and SlRIPK (like AtRIPK [9]) could phosphorylate HopZ3 and three tomato RIN4 homologues. SlRIN4-1 is involved in PTO-PRF immunity triggered by several effectors, including AvrPto Pto and AvrPtoB Pto , that lead to its degradation [24]. In Arabidopsis, both RIN4 degradation by AvrRpt2 and phosphorylation by RIPK triggered by AvrRpm1 and AvrB, induce immunity. This phosphorylation is prevented by HopZ3, which modifies Arabidopsis RIPK, RIN4, AvrRpm1 and AvrB3 [9]. In tomato, HopZ3 also modifies the aforementioned proteins and reduces SlRIPK activity in vitro, resulting in the reduced phosphorylation of SlRIN4s. The significance of the phosphorylation of SlRIN4s in tomato is unknown, but their perturbations may be guarded by R proteins and involved in immunity via a mechanism similar to that in Arabidopsis.
Tomato kinases and SlRIN4s targeted by HopZ3 also interact with AvrPto1 Psy . Moreover, AvrPto1 Psy , AvrPtoB Psy and several other HopZ3 targets interact with each other. Many effectors target the same host hub proteins essential for immunity [36], including multiple kinases involved in defense [37]. Interestingly, in bean the epistatic relationship between AvrPto1 Psy and HopZ3 is reversed from that seen in tomato such that AvrPto1 Psy suppresses defenses induced by HopZ3 [38]. Epistatic interactions of the effector repertoire (effectome) are revealed in the context of the host immune repertoire (targetome) [39]. It is possible that bacterial effectors act as multi-effector anti-immune complexes, similar to plant immune complexes. Further research is needed to reveal the dynamics of these mixed plant-effector complexes. HopZ3 modification of multiple components of host defense pathways and bacterial effectors themselves may provide ways to balance the suppression of immune responses in different plants while maintaining the virulence functions of effectors.
A survey of public databases suggests that HopZ3 homologues are not present in P. syringae pv. tomato strains sequenced to date. However, many P. syringae strains contain HopZ3 and we do not know if they can infect tomato. Pathogens constantly evolve, acquire (or lose) new effectors and this may enable infection of new plant species. It is plausible that tomato pathovars could acquire HopZ3 and overcome PTO/PRF-mediated disease resistance in the future, or a HopZ3-containing strain could become adapted to tomato. Epistatic interactions between effectors determine host range and effector loss and gain allow changes in host range.
Remarkably, some of HopZ3 immune modulations mirror those of other YopJ family acetyltransferases. Effectors in human and animal pathogens, such as YopJ in Yersinia sp., AvrA in Salmonella and VopA in Vibrio, acetylate residues in activation loops and ATP binding sites of kinases in MAPK and IKK pathways, blocking their phosphorylation [40]. Plant pathogen YopJ family effectors from Pseudomonas, Ralstonia and Xanthomonas are much more diverse and are known to have a large spectrum of unrelated substrates [40]. So far, HopZ3 is unique in its strategy to modify other bacterial effectors in addition to their plant targets to reduce immune responses. The ability to post-translationally modify its own effectors adds another layer to the bacterial arsenal, in addition to the acquisition of effectors suppressing PAMP-or effector-triggered defenses and the evolution of multiple effector alleles that can avoid recognition.

Plant growth and bacterial infection
Tomato (Solanum lycopersicum) plants had the Rio Grande-PtoR (76R) background that has Pto/Prf locus introgressed from resistant S. pimpinellifolium; pto11 and prf3 are lines with mutated, nonfunctional Pto and Prf genes, respectively [41].  [7,8]. Samples were serially diluted and plated on LB medium containing appropriate antibiotics. Bacterial growth experiments were performed at least three times. Results obtained with total and epiphytic bacteria counts were very similar and these experiments were used interchangeably. Transient transformation of N. benthamiana leaves using Agrobacterium was performed as previously described [7]. Bacterial strains are listed in S5 Table.

Plasmid construction
For Gateway cloning vectors, the open reading frame (ORF) of each gene was amplified without a stop codon using Pfu-DNA polymerase (Agilent Technologies) and the entire region was cloned into pDONR207 by Gateway BP reaction (Life Technologies) and then recombined by Gateway LR reaction (Life Technologies) into the destination vectors (pG005/pG006 for BiFC, pLaw vectors for yeast two-hybrid assay, pBAV226 for expression in PsyB728a). Point mutations were introduced by PCR using overlapping primers with mutated codons. The E. coli protein expression vectors used in this study (S6 Table) are not Gateway compatible. The ORFs were amplified using gene-specific primers with restriction enzyme sites at the 5'-end or 3'-ends. PCR products were digested with specific restriction enzymes and ligated into expression vectors. All constructs were verified by sequencing. Details of primers, vectors, bacterial and yeast strains are provided in S5-S7 Tables.

Effector deletion strains and complementation
Unmarked deletions of AvrPto1 in PsyB728a and PsyB728aΔHopZ3 [8] were created as described [8,9]. Briefly, regions upstream and downstream of AvrPto1were amplified with 5'and 3'primers (S7 Table) and linked together in pMTN1907 that has SacB cassette for negative selection. Colonies with integrated plasmid were selected on kanamycin, and subsequently deletion strains were selected on 10% sucrose. Deletion strains were complemented with effectors expressed from the nptII promoter in the low-copy pBAV226 plasmid as previously described [8]. Details of vectors and primers are provided in S6 and S7 Tables.

Yeast two-hybrid assay
The yeast two-hybrid screen was a part of a large scale effector-plant immune signaling protein interaction screen ( [9], https://charge.ucdavis.edu/charge_db/interaction/Y2H/Y2H_ interaction.php), and identified interactions were confirmed as previously described [9]. Briefly, the corresponding cells of the bait and prey were mated as shown in S1 Fig. Mated yeast strains (S5 Table) expressing the bait and prey constructs were grown on the selective minimal SD media (SD-Leu/-Trp/-His supplemented with 2.0 mM 3-aminotriazole (3-AT and SD-Leu/-Trp/+X-gal) for 4-6 days. Experiments were performed at least twice.

In vitro pull-down assay
In vitro pull-down assays were performed between purified recombinant GST-tagged SlPTO, -SlFEN, -SlRIN4-2, -3 or SlRIN4-1-MBP and His-tagged HopZ3; between His-tagged AvrPto1 Psy or AvrPtoB Psy and GST-tagged HopZ3, -PTO, or -FEN or PTO-MBP as described [9]. Mixed proteins were incubated at 4˚C for 1-2 h. Protein bound to the glutathione-sepharose beads (GE Healthcare or Promega), Ni-NTA agarose (QIAGEN) or amylose beads (NE BioLabs) was washed three to four times, separated on SDS-PAGE and stained with Coomassie blue or immunoblotted with anti-GST, anti-MBP and anti-His antibodies, respectively. All experiments were performed at least twice.
To assess a protein-protein interaction after acetylation by HopZ3, beads with immobilized AvrPto1 Psy -His or PTO-GST were incubated with 1 mM Acetyl-CoA, 5 μM IP6 and 1 μg HopZ3 or HopZ3_C300A for 2 h at room temperature (RT), washed three times, then the second interacting protein was added and pull down was performed as described above. Relative band intensities (interacting protein relative to immobilized protein) were quantified from at least four independent experiments using Image Lab software (Bio-Rad). To compare different experiments, interaction after acetylation with HopZ3 was set to 1.

BIFC assay and confocal microscopy
For BIFC analysis, protein-coding sequences were cloned into expression plasmids pG005 to create protein fused to the N-terminal half of YFP (protein:nYFP fusions) and into pG006 to create protein fused to the C-terminal half of YFP (protein:cYFP fusions), as previously described [9]. N. benthamiana leaves were co-infiltrated with mixtures of Agrobacteria harboring indicated combinations of BIFC constructs and YFP fluorescence was imaged 2 days after agroinfiltration. A LSM710 confocal laser scanning microscope (Zeiss Microsystems) equipped with a 40X water-immersion objective was used to examine protein subcellular localization or protein-protein interaction in BIFC assays with N. benthamiana epidermal cells. GFP or YFP imaging was performed by excitation with 488 nm argon laser and emission at 496-544 nm for GFP and 494-573 nm for YFP. YFP fluorescence indicated interaction. Experiments were repeated two to three times.

In vitro kinase assay
In vitro kinase assays were performed as previously described [9]. Briefly, 0.2, 0.4 and 0.6 μg of purified GST-tagged PTO or -FEN or 0.5 μg of purified GST-tagged SlRIPK were incubated with a buffer containing 100 mM Tris 6.8, 10 mM MgCl 2 , 10 mM MnCl 2 , 10 μM ATP and 1 μl of γ-32 P-ATP and adding 2 μg of His-tagged SlRIN4-1 or HopZ3 at RT for 60 min. The reaction was stopped by adding 5x Laemmli buffer. Proteins were separated by 12 or 15% SDS-PAGE, and signals were visualized by autoradiography.
To determine kinase activity after acetylation by HopZ3, 1 μg of SlRIPK-GST, FEN-GST or PTO-His were incubated with 1 mM Acetyl-CoA, 5 μM IP6 and 1 μg of His-Tagged HopZ3 or HopZ3_C300A for 2 h at RT and then washed with PBS. The kinase activity of SlRIPK, FEN or PTO was initiated by adding ATP, γ-32 P-ATP and MgSO 4 with or without SlRIN4s and incubated for 30 min at RT. All experiments were performed two to three times.

In vitro PTM mapping
For in vitro acetylation mapping, reactions were performed with 13 C-acetyl-CoA (Acetyl-1,2-13 C coenzyme A lithium salt, Sigma-Aldrich) to differentiate between background 12 Cacetylation that occurred in E. coli during the synthesis of recombinant protein and HopZ3-mediated acetylation in vitro. Substrates were mixed with either HopZ3 or the catalytically inactive HopZ3_C300A to distinguish any background acetylation that could be chemically caused by the presence of 13 C-acetyl-CoA. Briefly, 1 μg of purified His-tagged HopZ3 or HopZ3_C300A were mixed with 3 μg bead-bound substrate to which the acetylation reaction cocktail (50 mM HEPES (pH 8.0), 10% glycerol, 5 μM IP6 and 50 μM of 13 C-acetyl-CoA (Sigma-Aldrich)) was added in a total volume of 20 μl. Subsequently, beads were washed twice with washing buffer (50 mM HEPES pH 8.0, 50 mM NaCl, 10% glycerol), boiled in Laemmli loading buffer and processed for LC-MS/MS analysis. Data from the mass spectrometry of treated samples were analyzed for the presence of 13 C-acetylated peptides in the substrate (AvrPto1 Psy , PTO, SlRIN4s, SlRIPK).

Immunoprecipitation and in planta PTM mapping
For in planta acetylation mapping, Dex-AvrPto1 Psy -HA or Dex-PTO-HA were transiently coexpressed with Dex-HopZ3-GFP or Dex-HopZ3_C300A-GFP constructs in N. benthamiana. Plants were treated with 30 μM dexamethasone for 16 h to induce protein production. Proteins were extracted with lysis buffer containing 50 mM Tris pH 8.0, 150 mM NaCl, 10% glycerol,

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation 1% NP40, 0.5% sodium deoxycholate, phosphatase inhibitor (Thermo Fisher Scientific), 2 μM sodium butyrate (TOCRIS Bioscience) and 3 μM trichostatin A (Sigma-Aldrich). Clarified total protein lysate was incubated for 3 h with anti-HA magnetic beads (Medical and Biological Laboratories Co., LTD) at 4˚C. After washing the beads three times with the lysis buffer, proteins were eluted by boiling with Laemmli loading buffer. Samples were analyzed by LC-MS/ MS. PTM mapping of AvrPto1 Psy and PTO was repeated with independent experiments.

LC-MS/MS analysis
Trypsin digestion and HPLC were performed as described [9]. Mass spectrometry was performed at the Medical Genome Facility Proteomics Core at Mayo Clinic, Rochester, MN, US. Samples were analyzed via liquid chromatography-electrospray tandem mass spectrometry (LC-MS/MS) on a Q-Exactive (Thermo Fisher Scientific) mass spectrometer, using a 60,000 RP survey scan, m/z 375-1950, with lockmasses, followed by 15 HCD (higher energy collisional dissociation) CID (collision-induced dissociation) scans on only doubly and triply charged precursors between 375 and 1950 Da and ions selected for MS/MS were placed on an exclusion list for 60 seconds. Inclusion lists were applied to enhance the detection of acetylated or phosphorylated peptides from specific targets. Briefly, using in house software to process the FASTA sequence file for AvrPto1 Psy , PTO, tomato RIN4_1-3 and SlRIPK, we performed in silico trypsinization and modeled the following modifications: (formyl n-term, oxidation (M), acetyl (K, H, S, T), 13 C heavy acetyl (K, H, S, T), phospho (S, T), myristoylation (N-terminal G)), calculated m/z for doubly and triply charged ions, and combined the results into a � . csv file that was applied to the QE instrumentation method to enhance selection of the PTMbearing ions for fragmentation. The MS data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) with the dataset identifier PXD022953. Database searching of the 160610_Greenberg_db9 database (3412 entries) and protein identification and PTM quantification were performed as described in [9] and [42]. All acetylated, phosphorylated and myristoylated peptide spectra were manually validated [9]. The second in planta experiment was quantified by TIC (total ion current) using Scaffold [43]. PTMs above 5% are shown in S1-S4 Tables.

Structural modeling
To assess the relevance of the acetylated residues found by mass spectrometry, we modeled the structure of the HopZ3 substrates using the iTASSER (Iterative Threading Assembly Refinement) structural prediction software as previously described [9]. The best possible model was selected based on confidence score (C-score) calculated based on the significance of threading to the template alignments and convergence to the parameters of the structural assembly simulations. Model visualizations were generated using PyMOL software. PTM and interaction sites were labeled using the stick setting and coloring (Fig 4); however, the sites in the model are shown without PTMs.
Supporting information S1 Fig. Yeast two-hybrid assay. Positive interactions are indicated by the growth on the selection medium without Trp, Leu and His (SD-WLH+5mM 3-AT) for the reporter gene HIS3 or by blue color on medium containing X-gal [9]. A schematic overview of a subset of tested combinations is represented in Table 1. SlRIN4-3 trunc was used as a negative control; it has a deletion of nucleotide 14 that caused a frameshift mutation and early stop. FEN as a bait caused auto-activation (false positive). (PDF)

PLOS PATHOGENS
PTO defense pathway in tomato is suppressed by acetylation

S2 Fig. Interaction between HopZ3, AvrPto1 Psy and their potential interactors in planta.
Interactions of HopZ3, AvrPto1 Psy , AvrPtoB Psy , PTO, FEN, SlRIPK, SlRIN4-1, -2 and -3 were tested by BiFC. YFP fluorescence was imaged by confocal microscopy in epidermal N. benthamiana cells co-infiltrated with mixtures of Agrobacteria harboring expression plasmids pG005 (protein:nYFP fusions) and pG006 (protein:cYFP fusions). Bar = 20 μm. Schematic overview of a subset of tested combinations is represented in Table 1: +, fluorescence detected; -, fluorescence not detected; weak, weak signal, as determined from images of several experimental samples.  [9] (in vitro and in planta). Residues acetylated by HopZ3 are bold and highlighted in yellow; phosphorylation sites are underlined; known phosphorylation sites important for signaling (S141, T166) in AtRIN4 [6] are highlighted blue; residues phosphorylated by RIPK in AtRIN4 (T21, S160, T166) [3] are circled in red. � (asterisk)-fully conserved residues,: (colon)-conservation between groups of strongly similar properties,. (period)-conservation between groups of weakly similar properties. (PDF) S8 Fig. HopZ3 acetylates SlRIPK residues important for activity. Modifications in SlRIPK were determined in vitro, modifications in AtRIPK are from [9] (in vitro and in planta). Residues acetylated by HopZ3 are bold and highlighted in yellow; phosphorylation sites are underlined; known sites in AtRIPK important for activity (K122; S251/T252 which correspond to S198/T199 in PTO) [9] are circled in red; sites corresponding to T204 in PTO are circled in blue. � (asterisk)-fully conserved residues,: (colon)-conservation between groups of strongly similar properties,. (period)-conservation between groups of weakly similar properties. (PDF)  Table. AvrPto1 Psy PTMs in vitro and in planta. PTMs were determined either in vitro, using purified recombinant AvrPto1 Psy after 13 C-acetylation by HopZ3/HopZ3_C300A, or in planta, by co-expressing AvrPto1 Psy and HopZ3/HopZ3_C300A in N. benthamiana, followed by immunoprecipitation. Numbers indicate enrichment (fold change) of acetylation in the presence of HopZ3 vs. HopZ3_C300A. Red shading: significant (>50%) increase of acetylation with HopZ3. Blue shading: significant decrease of phosphorylation in planta in the presence of HopZ3. Residues known to be important for AvrPto signaling or interaction with PTO are in bold. + indicates phosphorylation found in a recombinant protein (in vitro) or in planta. Z3: acetylation found only in AvrPto1 Psy treated or co-expressed with HopZ3 and not HopZ3_C300A. Ac: acetylation; Phos: phosphorylation; Myr: myristoylation; exp: experiment. � Some spectra do not distinguish these 2 close residues. # In planta sites with acetylation above 25% in the presence of HopZ3. (PDF) S2 Table. PTO PTMs in vitro and in planta. PTMs were determined either in vitro, using purified recombinant PTO after 13 C-acetylation by HopZ3/HopZ3_C300A, or in planta, by co-expressing PTO and HopZ3/HopZ3_C300A in N. benthamiana, followed by immunoprecipitation. Numbers indicate enrichment (fold change) of acetylation in the presence of HopZ3 vs. HopZ3_C300A. Red shading: significant (>50%) increase of acetylation with HopZ3. Blue shading: significant decrease of phosphorylation in planta in the presence of HopZ3. Residues important for PTO signaling or interaction with AvrPto are in bold. + indicates phosphorylation found in a recombinant protein (in vitro) or in planta. Z3: acetylation found only in PTO treated or co-expressed with HopZ3 and not HopZ33_C300A.; Ac: acetylation; Phos: phosphorylation; exp: experiment. � Some spectra do not distinguish these 2 close residues. # In planta sites with acetylation above 25% in the presence of HopZ3. (PDF) S3 Table. SlRIN4s PTMs in vitro. PTMs were determined using purified recombinant SlRIN4s after in vitro 13 C-acetylation by HopZ3/HopZ3_C300A. Numbers indicate enrichment (fold change) of 13