Paladin, a tyrosine phosphatase-like protein, is required for XA21-mediated immunity in rice

XA21 encodes a rice immune receptor that confers robust resistance to most strains of the Gram-negative bacterium Xanthomonas oryzae pv. oryzae (Xoo). XA21-mediated immunity is triggered by recognition of a small protein called RaxX-sY (required for activation of XA21-mediated immunity X, tyrosine-sulfated) secreted by Xoo. To identify components regulating XA21-mediated immunity, we generated and screened a mutant population of fast-neutron-mutagenized rice expressing Ubi:Myc-XA21 for those susceptible to Xoo. Here, we report the characterization of one of these rice mutants, named sxi2 (suppressor of XA21-mediated immunity-2). Whole-genome sequencing revealed that sxi2 carries a deletion of the PALADIN (PALD) gene encoding a protein with three putative protein tyrosine phosphatase-like domains (PTP-A, -B, and -C). Expression of PALD in the sxi2 genetic background was sufficient to complement the susceptible phenotype, which requires the catalytic cysteine of the PTP-A active site to restore resistance. PALD co-immunoprecipitated with the full-length XA21 protein, whose levels are positively regulated by the presence of the PALD transgene. Furthermore, we foundd that sxi2 retains many hallmarks of XA21-mediated immunity, similar to the wild type. These results reveal that PALD, a previously uncharacterized class of phosphatase, functions in rice innate immunity, and suggest that the conserved cysteine in the PTP-A domain of PALD is required for its immune function.


INTRODUCTION
Plants and animals utilize immune receptors to recognize and respond to microbes. Plants deploy immune receptors at the cell surface that detect conserved immunogens produced by pathogens. These immune receptors may form protein complexes with co-receptors and receptor-like cytoplasmic kinases upon recognition of cognate molecular pattern ligands to activate plant defense responses. Research in rice, wheat, Arabidopsis, and tomato suggest that the molecular mechanisms of immune signaling downstream of these immune receptors are highly conserved (Boutrot and Zipfel, 2017). Understanding these underlying signaling pathways advances our knowledge of the mechanisms of plant immunity and can guide efforts to engineer enhanced resistance in agricultural crops.
In Arabidopsis and rice, recognition of cognate ligands by immune receptors triggers rapid immune responses across several time frames. Within minutes, bursts of reactive oxygen species (ROS) production and a cascade of mitogen-activated protein kinases phosphorylation convey the immune signal generated from membrane-bound immune receptors to nuclear components (Couto and Zipfel, 2016;Pruitt et al., 2015). Late immune responses, triggered hours after ligand perception, include the production of ethylene and accumulation of pathogenesisrelated protein gene transcripts (Kim et al., 2016;Pruitt et al., 2015). These immune responses contribute to disease resistance in many plant-pathogen interactions (Chen et al., 2014;Macho and Zipfel, 2014;Pruitt et al., 2015).
called RaxX (RaxX-sY) (required for activation of XA21-mediated immunity X, tyrosine-sulfated) (Luu et al., 2019;Pruitt et al., 2015). XA21 is a receptor-like kinase consisting of an extracellular leucine-rich repeat (LRR) domain, a transmembrane region, a juxtamembrane motif, and an intracellular non-arginine-aspartate (non-RD) kinase domain (Park et al., 2012). Transgenic rice plants overexpressing an N-terminal Myc-tagged XA21 display robust resistance to Xoo strain PXO99; in contrast, wild-type rice that lacks XA21 is susceptible to Xoo strain PXO99, reflected by the formation of long disease lesions from the point of inoculation along the leaf (Park et al., 2010a). The XA21 LRR domain directly binds a synthetic, 21 amino acid, tyrosine-sulfated RaxX (RaxX21-sY) with high affinity (K d = 16 nM) (Luu et al., 2019). This binding triggers a ROS burst and ethylene production, and induces expression of defense marker genes, such as PR10b (LOC_Os12g36850) and LOC_Os04g10010 in XA21 plants (Luu et al., 2019;Pruitt et al., 2015). Despite these advances, the detailed molecular and physiological mechanisms underlying XA21 resistance have yet to be fully determined.
In this study, we characterized a rice mutant called sxi2 (suppressor of XA21-mediated immunity 2) that was identified in a forward genetic screen by direct inoculation with PXO99. We identified a 20kb deletion in chromosome 3 of sxi2 that co-segregates with the susceptible phenotype and found that genetic complementation with one of the deleted genes, PALADIN (PALD), rescues the susceptible phenotype of sxi2. PALD encodes a protein that carries three putative protein tyrosine phosphatase (PTP) domains. We further found that the first PTP domain is critical for PALD to com- (B) Lesion length and PCR genotyping results of the F2 progeny (n = 4-7), derived from a cross of sxi2 with the XA21 parental genotype, inoculated with Xoo. Controls plants include Kitaake, XA21, and sxi2. Lesion lengths were measured 14 days post-inoculation. Each dot represents a single leaf inoculation. Lower panel: genotyping results of three different sets of primers targeting three different mutations on chromosome 3 of each F2 individual. Primer 1 targets a deletion region in chromosome 3 of sxi2 (see A). Primer pairs 2 and 3 target two different single-base pair substitutions nearby the deletion region in LO-C_Os03g10990. The absence of a PCR product indicates a homozygous deletion or mutation; the presence of a PCR product indicates that the plant carries at least one copy of the wild-type allele. Primer sequences are listed in Supplemental Table 1. *P < 0.05 compared with sxi2 using Dunnett's test. plement the sxi2 mutant. PALD interacts with XA21 in rice protein co-immunoprecipitation and in yeast two-hybrid assays. In contrast, the ROS burst and expression of defense marker genes, such as LOC_Os12g36830 and LOC_Os06g37224, are not affected in sxi2. Together, these results demonstrate a role for the phosphatase PALD, a previously undescribed class of PTPs, in XA21-mediated immunity.

RESULTS
The susceptible phenotype of sxi2 co-segregates with a deletion locus in chromosome 3 To study the molecular mechanism underlying XA21-mediated immunity, we carried out a forward genetic screen of the rice variety Kitaake expressing the XA21 gene to isolate Xoo-susceptible mutants. This Kitaake transgenic line expresses Ubi:Myc-XA21 (called the XA21 line in this manuscript) and displays robust resistance to Xoo strain PXO99 (Pruitt et al., 2015). This XA21 line was mutagenized by fast-neutron treatment, generating a population containing over 6000 M1 lines (Li et al., 2017). Approximately 21 000 M2 lines (derived from 3000 M1) were inoculated with Xoo, and 10 mutants that carry compromised XA21-mediated immunity were isolated and named suppressors of XA21mediated immunity (sxi). Here, we focus on the mutant sxi2.
To identify the underlying mutation associated with the susceptibility phenotype, sxi2 was backcrossed with its XA21 parent to generate a segregating F2 population. Inoculation of 40 F2 plants revealed that 7 progenies are fully susceptible to Xoo and 33 are resistant (Figure 1), suggesting that a recessive mutation in sxi2 plants is causing the susceptible phenotype. To identify the responsible mutation, we sequenced the genome of sxi2 and identified three large deletions distributed on chromosomes 1, 3, and 8, respectively. Using 3 pairs of primers each targeting one of these 3 deletions to PCR genotype the sxi2 mutant in 40 segregating F2 progeny derived from a backcross with the Ubi:Xa21 parent, we found that only the primer pair targeting the 20-kb deletion on chromosome 3 of sxi2 (shown as primer 1 in Figure 1B) co-segregated with the susceptible phenotype. This result suggests that one or more of the genes deleted in this region is required for XA21-mediated immunity. In addition to this deletion, two non-synonymous single-base substitution mutations are found nearby LOC_Os03g10990. We also PCR genotyped these two single-base substitutions in the same F2 population, and the result confirmed that the susceptible phenotype is totally linked to the 20-kb deleted region but not to the two single-base substitutions ( Figure 1B).
We also generated transgenic plants that express the PALD coding sequence tagged with a hemagglutinin (HA) epitope at its N terminus under control of the maize ubiquitin promoter (Ubi:HA-PALD) in the sxi2 background. The Myc-XA21 expression construct used in these experiments is driven by the maize ubiquitin-1 promoter, which extends XA21 protein expression to all cell types in the rice leaf. We obtained 38 independent T0 Ubi:HA-PALD/sxi2 lines, which were assessed for PALD RNA and protein levels, and resistance to Xoo. Only three T0 lines (nos. 3, 25, and 26) gave rise to T1 progeny that express sufficient HA-PALD protein to confer heritable resistance to Xoo ( Figure 2C and Supplemental Figure 2). We found that detectable HA-PALD protein did not correlate with HA-PALD RNA quantity, because many T0 lines that showed undetectable protein levels displayed similar PALD transcript levels as line no. 3 (Supplemental Figure 3). Despite the low frequency of detectable HA-PALD pro-tein, the heritability of the complemented phenotype in line nos. 3, 25, and 26 further supports that the PALD protein is required for XA21-mediated immunity and sufficient to complement sxi2.
ROS production and expression defense marker genes are not altered in sxi2 upon recognition of RaxX21-sY by XA21 In our previous work, we observed that application of RaxX21-sY synthetic peptide to XA21 rice leaves triggers hallmarks of the innate immune response, including ROS production via plasma membrane RBOH proteins and induction of several defense marker genes (Pruitt et al., 2015). Using reagents that convert ROS to luminescence, we observed clear bursts of ROS production in XA21, sxi2, and HA-PALD/sxi2 plants that were similar in both amplitude over time and cumulative chemiluminescence ( Figure 3). Chemiluminescence was not significantly elevated in water or non-sulfated RaxX21 treatments, nor in Kitaake that lack XA21. RNA expression analysis of leaf clippings treated with RaxX21-sY for 8 h also does not show a change in expression of established marker genes in sxi2 relative to XA21 (Supplemental Figure 4). These results demonstrate that ROS bursts are specifically elicited by RaxX21-sY, and that both ROS and marker gene expression in sxi2 respond like the wild type.
PALD positively regulates XA21 protein levels and interacts with XA21 in vivo LRR receptor-like kinases often function with an assortment of interacting proteins that modulate the activity or accumulation of the receptor (Lu et al., 2011). As loss of XA21-mediated resistance in sxi2 could be due to a direct effect on the receptor, we performed immunoblot analysis to assess whether the knockout of PALD affects XA21 gene expression or protein levels. XA21 protein levels in the two PALD native expression complementation lines (pPALD:PALD/sxi2 nos. 1-4-1 and 2-1-1) and the overexpression complementation line HA-PALD/sxi2 were compared with wild-type XA21 and sxi2 genotypes. We found that XA21 protein is significantly reduced in sxi2 relative to wild-type XA21 plants, and that the XA21 protein level is restored in all PALD complementation lines ( Figure 4). These data suggest that PALD positively affects XA21 protein abundance. This effect is not due to altered transcript levels of the XA21 gene, as transcript levels remain similar between complemented and mutant lines (Supplemental Figure 5).
To assess whether PALD associates with the XA21 protein in vivo, putative PALD interacting proteins were coprecipitated from leaf protein extracts of Ubi:HA-PALD/sxi2 nos. 3-14-9 (hereafter referred to as HA-PALD/sxi2) using anti-HA beads. XA21 plants, which do not carry any HA-tagged proteins, served as a negative control. Immunoblotting with anti-HA antibodies revealed two prominent bands near 140 kDa and one band around 100 kDa present in the input sample and enriched in the immunoprecipitation (IP) samples of HA-PALD/sxi2 ( Figure 5). These bands were not present in the XA21 samples, indicating that the HA antibody is specific to HA-PALD. The two bands around 140 kDa are near the expected size of full-length PALD (142 kDa). Full-length Myc-XA21 protein ($140 kDa), detected by an anti-Myc antibody, is enriched in the IP samples of HA-PALD/sxi2. This result was further confirmed in an independent series of co-immunoprecipitation experiments in which XA21 peptide fragments were detected in HA-PALD-enriched samples using liquid chromatography-tandem mass spectrometry (LC-MS/MS) (Supplemental Figure 6). These results demonstrate that HA-PALD interacts with XA21 in vivo.
sequences and the XA21 cytoplasmic kinase domain (XA21K668) in yeast two-hybrid vectors (Chen et al., 2010a). We observed that full-length PALD does not interact with XA21K668 in yeast two-hybrid, whereas the N-terminal half (PALD-AB) interacts with XA21K668 ( Figure 6A and 6B). To determine if PTP-A or PTP-B is required for the interaction between PALD-AB and XA21K668, we further truncated PALD-AB from the C terminus ( Figure 6C; from Del-1 to Del-5, A2, and A1) and N terminus ( Figure 6D; from Del-6 to Del-10, and B). PALD-Del-3 and PALD-Del-4 show robust interactions with XA21K668, indicating that PTP-B domain is not required for the interaction. In contrast, all PALD N-terminal truncations affecting the PTP-A domain abolished the interaction between PALD and XA21K688, suggesting that the N terminus of PALD encompassing the PTP-A domain is required for interaction with XA21K668 ( Figure 6D).
Plant PALD proteins contain three PTP domains (PTP-A, PTP-B, and PTP-C); in contrast, mammalian PALD proteins contain only two PTP domains (PTP-A and PTP-B). To study whether the PTP-C domain located at the C terminus may play a role in regulating plant PALD function, we first investigate if the PTP-C domain interacts PALD-AB and if it inhibits the interaction between PALD-AB and XA21K668. We tested whether the C-terminal end of the protein (from amino acid 701 to 1256) containing PTP-C domain interacts with PTP-AB protein truncation. The results suggest that the PTP-C domain interacts with PALD-AB; the interaction remains when PTP-B domain is gradually truncated from the C terminus ( Figure 6E). In contrast, PALD-AB N-terminal truncations (A) ROS production in leaves of Kitaake, XA21, sxi2, and HA-PALD/sxi2 treated with water, RaxX21-Y, or RaxX21-sY (500 nM). Data points depict mean ROS production ±SE (n = 6). (B) Total chemiluminescence relative light units detected between 0 and 120 min of ROS assays shown in (B). This experiment was performed three times with similar results. *P < 0.05 compared with H 2 O samples in each genotype using t-test.
( Figure 6F) abolished interaction with PTP-C. Immunoblot analysis was performed to verify that all PALD truncations are expressed in the yeast two-hybrid assay (Supplemental Figure 7). The results show that full-length PALD is expressed at a lower level, while all the other PALD truncation forms are expressed at similar, higher levels. Together, these data show that the N terminus of PALD encompassing PTP-A is critical for interaction with the XA21 kinase domain and that the PTP-C domain interacts with the same PALD N terminus in the yeast two-hybrid.

XA21 kinase does not phosphorylate PALD in vitro
Structural-based homology modeling suggests that PALD possesses an iterative series of three subunits connected by linker regions (Supplemental Figure 8), each with one PTP domain. LC-MS/ MS analysis of immunoprecipitated HA-PALD also identified three amino acid residues of the protein that can be phosphorylated in vivo (Supplemental Figure 8). Each PTP domain possesses one such residue, and Ser641 lies just outside the PTP-B predicted catalytic domain. Phosphorylated residues were identified in both mock-and RaxX21-sY-treated samples. Previously, we observed bidirectional phosphorylation between the XA21 kinase domain and OsSERK2, which is a positive regulator of XA21-mediated immunity that interacts with XA21 in vivo (Chen et al., 2014). Because PALD likely interacts with the XA21 intracellular kinase domain in vivo and can be phosphorylated, we hypothesized that XA21 might phosphorylate PALD. We assessed XA21 in vitro kinase activity using purified His-tagged XA21 kinase domain and full-length PALD recombinant proteins and radiolabeled [ 32 P]g-ATP. Whereas the XA21 kinase autophosphorylates under the assay conditions, we found no evidence that the XA21 kinase transphosphorylates PALD when the proteins are incubated together (Supplemental Figure 9).

An active site mutation of PALD PTP-A abolishes complementation of resistance, while PTP-B is dispensable
Experimental analysis of PALD homologs in animal systems do not reach consensus whether PALD is a functional enzyme. While plant homologs possess three PTP domains (Supplemental Figure 10), only PTP-A and PTP-B possess the conserved cysteine residue in the CXGXGRT active site motif typical of a phosphatase; the catalytic cysteine is of particular importance for initiating the enzymatic reaction (Bellomo et al., 2018). Our efforts to determine the phosphatase activity of PALD recombinant protein in vitro using the standard method of para-nitrophenyl phosphate as a substrate were unsuccessful (data not shown). To assess whether the putative phosphatase function of PALD is required in planta for resistance to Xoo strain PXO99, we introduced into sxi2 a complementation construct containing native promoter-driven HA-tagged PALD either with the wild type, or with a cysteine-toserine point mutation in PTP-A (C228S, abbreviated as CS1) or PTP-B (C647S, shown as CS2). Pathogen challenge of numerous independent T0 transgenic lines show that those with PALD mutated in PTP-A are unable to confer resistance to PXO99, while those with mutated PTP-B show restoration to XA21 disease resistance similar to wild-type PALD (Figure 7). T1 segregation analyses confirmed the T0 results; null-segregant plants exhibit lesion lengths similar to sxi2. HA-PALD protein levels were assessed in the T0 generation, and detectable protein was present in these parent lines, in particular the PTP-A lines, indicating that loss of the ability to complement sxi2 by the PTP-A protein is not due to the absence of the mutant PTP-A PALD protein. This result suggests that PALD possesses phosphatase activity in vivo, and that disruption of that activity by mutation of PTP-A abolishes the ability of PALD to complement sxi2. PTP-B appears dispensable for XA21-mediated immunity.

DISCUSSION
In this study, we report the identification and characterization of a fast-neutron-induced rice mutant, called sxi2 (suppressor of XA21-mediated immunity-2). Through whole-genome resequencing, co-segregation analysis, and genetic complementation we show that a deletion of the PALD gene is responsible for the susceptible phenotype observed in the sxi2 mutant. Introduction of PALD rescues the susceptible phenotype of sxi2.
What is the underlying cause of the sxi2 susceptibility phenotype?
Although the XA21 protein level is reduced in sxi2, RaxX21-sY still induces many hallmarks of the immune response in sxi2, such as rapid ROS production and induction of defense marker genes (Figure 3 and Supplemental Figure 4). Our results suggest that, although the amount of XA21 protein in sxi2 is sufficient to recognize RaxX21-sY, these responses are insufficient to confer resistance to Xoo strain PXO99. Together these data do not give a clear conclusion as to how mutations in PALD affect XA21mediated signal transduction or XA21 protein accumulation.

How does PALD regulate XA21 immunity?
In vertebrates, PALD was reported to regulate chicken neural crest cell formation and migration (Adams et al., 2008;Gammill and Bronner-Fraser, 2002;Roffers-Agarwal et al., 2012). Because mutation of both catalytic cysteine residues within the predicted phosphatase active site of PALD did not abolish the function of PALD, the authors hypothesized that PALD may function as an anti-phosphatase that protects phosphosites from dephosphorylation (Roffers-Agarwal et al., 2012). In contrast, the expression levels of human PALD in an African green monkey kidney fibroblast-like cell line negatively correlated with insulin-stimulated AKT phosphorylation, suggesting that human PALD possesses bona fide phosphatase activity (Huang et al., 2009). Here, we report that the catalytic cysteine C228 in the PTP-A domain is required to restore resistance in sxi2 (Figure 7), whereas C647 in the PTP-B domain is not essential. These results suggest that, in plants, the putative phosphatase activity of PALD is critical for function. PALD possesses two conserved putative PTP domains, and interacts with XA21 in vivo in both rice and yeast (Figures 6 and 7). One possible explanation for the lower abundance of XA21 in sxi2 is that PALD alters XA21 phosphorylation status directly. It has been shown that XA21 autophosphorylates Ser686, Thr688, and Ser689 residues localized in the juxtamembrane domain, and that alanine substitution of these residues specifically destabilizes the XA21 protein in vitro and in vivo (Xu et al., 2006). Transgenic rice carrying the same alanine substitutions in XA21 are slightly more susceptible to Xoo Philippine race 6 (PXO99), suggesting that autophosphorylation of these serine/threonine residues of XA21 positively regulate XA21 stability and immunity (Xu et al., 2006). In contrast, XA21 is capable of autophosphorylation of Tyr698, Tyr786, Tyr907, and Tyr909. Rice plants expressing phosphomimetic XA21 variants with these tyrosine residues mutated to aspartate (XA21 YD -GFP) are susceptible to Xoo. These phosphomimetic mutations do not affect XA21 protein levels, suggesting that these mutations directly compromise XA21 function. Conversely, rice plants expressing XA21-GFP fusion proteins with these tyrosine residues individually mutated to Myc-XA21 detection in total leaf protein extracts of independently transformed PALD complementation lines (pPALD:PALD/sxi2 nos.1 and 2) and one PALD overexpression line (HA-PALD/sxi2). Each genotype carries the XA21 gene except for Kitaake, which is included as a negative control. Western blotting analysis with the anti-Myc antibody was used to detect Myc-XA21 protein levels. The XA21 protein level was quantitated and normalized to the Rubisco protein level; the value of the first XA21 sample was set as 1. The membrane was stained with Coomassie brilliant blue (CBB) for equal loading of the protein samples. The western blots were repeated three times with similar results. phenylalanine (XA21 YF -GFP), which prevents phosphorylation at these sites, maintain resistance to Xoo (Caddell et al., 2018). These results suggest that constitutive phosphorylation on these tyrosine residues of XA21 may negatively regulate Xoo resistance. Alternately, the effect of PALD may be indirect, influencing XA21 immunity through interactions with not yet described components.
To distinguish between the direct versus indirect models, we carried out in vitro kinase assays to assess whether PALD is a direct phosphorylation target of XA21 or if the putative PALD phosphatase reduces XA21 phosphorylation. Although we did not observe PALD phosphorylation or XA21 dephosphorylation in these experiments (Supplemental Figure 9), this negative result is not conclusive as in vitro assays can give false negatives when essential components are missing.

PALD is widely conserved in plants and animals
Despite numerous studies of protein tyrosine kinases (PTKs) and PTPs in animal models, the biological functions of PTKs and PTPs in plants have only recently begun to be explored (Shankar et al., 2015). Human PTPs can be divided into three different families based on their critical catalytic amino acids: cysteine, aspartate, and histidine. The cysteine-based PTPs can be further categorized into three different classes, PTP family (class I), lowmolecular-weight PTPs (class II), and cell division cycle 25s (class III) (Alonso and Pulido, 2016). Human PALADIN belongs to the class I PTP family with two PTP domains, and both domains carry the PTP signature motif CxxGxGR at their C terminus. PALD is the only gene in subclass IV of the class I PTP family (Alonso and Pulido, 2016). PALD is present as a single copy gene in rice and Arabidopsis, and homologs are found from vertebrates to plants (Supplemental Figure 10). PALD has not been previously studied in plants, where single to multiple homologs have been observed.
Using the structural homology predictive tool RaptorX, we generated a potential model for the three-dimensional PALD protein (Supplemental Figure 8). This analysis indicates that Protein samples were extracted from leaves of XA21 and HA-PALD/sxi2. Myc-XA21 protein is expressed in both plants, while HA-PALD is only expressed in HA-PALD/sxi2. The protein crude extract was incubated with anti-HA beads and centrifuged to co-immunoprecipitate PALD. Proteins in the input, non-bound (NB) and immunoprecipitated (IP) samples were analyzed with SDS-PAGE and western blotting. One major band enriched in the immunoprecipitated samples around 100 kDa (HA-PALD-CP?) is a putative cleavage products of HA-PALD. Immunodetection of PEP carboxylase (PEPC) serves as a loading control. Note: the anti-Myc antibody detects a non-specific band around 120 kDa as reported previously (Park et al., 2008). Similar results were observed in two other independent experiments. PALD is composed of three repeating subunits that each contains a PTP active site. Both PTP-A and PTP-B possess the proper corresponding amino acid residues on both the phosphate binding loop and acid loop found in common PTP enzymes. Despite this observed conservation, our genetic complementation of cysteine mutants in sxi2 finds that only the cysteine of PTP-A is necessary for complementation (Figure 7). It may be that the PTP-A and PTP-B domains in rice PALD have a distinct function. Support for this hypothesis comes from a study of the two PTP domains of mammalian bi-domain PTP Leukocyte Common Antigen-Related (LAR). In this case, the conserved cysteine in the PTP-D1 domain of LAR is important for phosphatase activity, whereas the cysteine in the PTP-D2 domain of LAR functions in substrate specificity (Ahuja and Gopal, 2014;Tsujikawa et al., 2001).
PALD homologs in land plants carry a third PTP-like (PTP-C) subunit near the C terminus that is lacking in animal isoforms; PTP-C carries a serine instead of the typical cysteine residues in PTP-A and PTP-B, suggesting that the PTP-C domain does not have a tyrosine phosphatase activity. Instead, the PTP-C domain may inhibit the interaction between PALD-AB and XA21 by sequestering the PALD-AB domain ( Figure 6). In support of this hypothesis, we observed three major bands of PALD in the immunoblots of HA-PALD/sxi2. The first two bands of 140 kDa are close to the predicted size of full-length PALD. The top band may be phosphorylated PALD, as we observed three phosphorylated residues on PALD in vivo (Supplemental Figure 6). The third band of 100 kDa might be a cleaved product of PALD carrying the first two PTP domains ( Figure 5). This observation is reminiscent of the cleavage of human PTPa at the linker sequence between its two PTP domains (Sonnenburg et al., 2003). The linker region between the two PTP domains of PTPa is highly flexible and proteolytically labile, which may be important for the intra-and/or intermolecular interactions that are critical for its function. Our results showing that the N terminus of PALD is critical for the interactions with both XA21K668 and the C-terminal end of PALD and that deletion of PTP-C domain enhances the interaction between PALD and XA21 suggest that the PTP-C domain may inhibit the interaction between PALD-AB and XA21 by sequestering the PALD-AB domain ( Figure 6).
In vertebrates, PALD has been reported to regulate insulin signaling (Huang et al., 2009), neural crest formation, and the The XA21 intracellular domain (XA21K668) and C-terminal end of PALD (PALD-C, carry 701-1256AA) were cloned in pLexA vector. b-Glucuronidase was cloned into the pLexA vector as negative control (pLexA-GUS). Yeast two-hybrid assays were performed to test the interaction between truncated PALD and XA21K668 (B-D), and between truncated PALD and PALD-C (E and F). The results of yeast two-hybrid are scored by the degree of blue color on colonies (-, no interaction; +, weak interaction; ++, moderate interaction; +++, strong interaction; NT, not tested). apoptotic response (Egana et al., 2017). In a large-scale proteomic study, human PALD was found to interact with Toll-like receptor 9 (TLR9), DNA-dependent activator of IFN-regulatory factors (DAI), and interferon regulatory factor 7 (IRF7), all important regulators functioning in the human innate immune response (Li et al., 2011). The TLR9 receptor triggers the proinflammatory cytokine response upon the binding of DNA from bacteria and viruses. DAI and IRF7 are transcription factors involved in DNA-mediated activation of innate immune responses (Ning et al., 2011;Takaoka et al., 2007). Our finding that PALD plays a role in the XA21mediated immune response provides a possible, but as yet undemonstrated, linkage between plant and animal defense pathways.

Assessment of gene expression
Detached rice leaves were treated with RaxX21-sY following a previous publication with minor modifications (Pruitt et al., 2015). In brief, rice seeds of Kitaake, XA21, sxi2, and HA-PALD/sxi2 were first germinated in water and grown in hydroponic system using A-OK Starter Plugs (Grodan) and watered with Hoagland's solution (pH 5.7) once a week. Chemically synthesized RaxX21-sY (aa 35-55, called RaxX-sY in this study) is used as described in a previous publication (Pruitt et al., 2015). For the gene expression analysis, rice leaf strips were cut into 2-cm-long strips and incubated in distilled water overnight to reduce wounding responses. The incubation water was then replaced with 500 nM RaxX21-sY peptide or water (mock treatment) to start peptide treatment. Leaf samples were then collected and snap frozen in liquid N 2 1 and 8 h after the treatment for gene expression assays. For gene expression assays, total RNAs were extracted following lab protocol (Pruitt et al., 2015).

ROS assay
ROS assays were performed as described previously with minor differences (Pruitt et al., 2015;Wei et al., 2016). Leaves of 4-5-week-old hydroponically grown rice plants were cut longitudinally along the mid vein and then transversely cut into 1-to 1.5-mm-thick leaf pieces. These leaf pieces were incubated on autoclaved ddH 2 O in regular Petri dishes overnight with constitutive light (between 5 and 10 mmol/(m2*s)) at room temperature. The next morning, two leaf pieces were transferred into each well of a 96-well white plate containing 95 ml of excitation solution (5-20 mM L-012 [Wako] and horseradish peroxidase [0.5-2 mg/ml; Sigma]). Five mL of 10 mM sulfated (RaxX21-sY), non-sulfated (RaxX21-Y) peptides, or ddH 2 O were added into each well right before the start of the ROS assay. Chemiluminescence was measured for 3 h with 0.5 s per reading with a high-sensitivity plate reader (TriStar, Berthold).

RNA extraction and qRT-PCR analysis
Rice leaves were homogenized to powder using a tissue lyser and RNA were extracted with 1 ml of TRIzol Reagent (Thermo Fisher) following the manufacturer's protocol. RNA samples were treated with Turbo DNase (Ambion) at 37 C for 60 min to remove genomic DNA contamination. RNA samples were purified using a NucleoSpin RNA column (Macherey-Nagel) and stored in À20 C refrigerator. For each RNA sample, 2 mg of total RNA were used for cDNA synthesis using the High Capacity cDNA Reverse Transcription Kit (Thermo Fisher). Gene expression changes were determined using the DDCt method and normalized to actin (LOC_Os03g50885) and compared with mock-treated samples (Schmittgen and Livak, 2008).

Protein extraction and western blotting
Total proteins extracted from rice leaves were analyzed by western blotting as described previously (Wei et al., 2016). In brief, around 70 mg of leaf tissues were collected in a 2-ml homogenization tube and snap frozen in liquid N 2 . Leaf samples were homogenized into powder and incubated with 200 ml of protein extraction buffer on ice for 30 min (150 mM NaCl, 10 mM sodium phosphate buffer [pH 7.2], 2 mM EDTA, 1% Triton X-100, 1 mM PMSF, 0.07% b-mercaptoethanol, Protease Inhibitor Cocktail P-9599 [Sigma], and 20 mM NaF). Cell debris was removed by centrifugation with 16 100 g, 20 min at 4 C twice. Protein concentrations were measured using the Bradford method and normalized to the lowest sample concentrations in each experiment. Equal amounts of total proteins were mixed with 53 protein sample buffer and incubated at 95 C for 10 min before SDS-PAGE. For each well, around 40 mg of proteins were loaded and separated by SDS-PAGE. Proteins were transferred from the gel to polyvinylidene difluoride (PVDF) membrane for western blotting analysis. Three different antibodies were used in this study: anti-HA (1:3000 dilution, Covance catalog no. MMS-101R); anti-Myc (1:1000 dilution, Santa Cruz sc-47694); and anti-PEPC (1:2000 dilution, Abcam catalog no. ab34793)

Co-immunoprecipitation assay
Total proteins were extracted from rice leaves of 5-week-old XA21 and HA-PALD/sxi2 plants in the absence of Xoo or RaxX21-sY. Leaves of XA21 and HA-PALD/sxi2 plants were collected in liquid N 2 and kept at -80 C until ready. Around 500 mg of leaves were grounded in N 2 by mortar and pestle. Leaf powders were incubated with 1.5 ml of protein extraction buffer on ice for 30 min (150 mM NaCl, 10 mM sodium phosphate buffer [pH 7.2], 2 mM EDTA, 1% Triton X-100, 1 mM PMSF, 0.07% b-mercaptoethanol, Protease Inhibitor Cocktail P-9599 [Sigma], and 20 mM NaF). Cell debris was removed by centrifugation with 16 100 g, 20 min at 4 C twice. PALD proteins were precipitated by adding 30 ml of Roche Anti-HA Affinity Matrix (Roche). The protein extracts were incubated with anti-HA affinity resin to retain HA-PALD and its associated proteins. Protein samples were separated via SDS-PAGE and probed by anti-HA and anti-Myc antibodies sequentially, which specifically targets HA-PALD and Myc-XA21 proteins, respectively. The protein samples were incubated at 4 C overnight on an end-to end rotator. The samples were centrifuged at 1500 g for 3 min at 4 C to precipitated HA-PALD and protein complex. The pellets were washed five times with cold protein extraction buffer twice to remove unbounded proteins. SDS-PAGE sample buffer was added to the pellet and incubated at 95 C for 10 min. After centrifugation at 16 100 g for 5 min, 20 ml of the supernatant were loaded into 8% acrylamide gel for SDS-PAGE. Proteins were transferred to PVDF membrane at 23 V overnight at 4 C. Myc-XA21 proteins were detected by western blotting analysis with c-Myc antibody (Santa Cruz, sc-47694, 1:1000 dilution) and goat antimouse IgG-HRP (Santa Cruz, sc-2005, 1:2000 dilution). The PVDF membrane was incubated with 1 ml of Pierce ECL 2 western blotting substrate (Thermo Scientific) and visualized with X-ray film.

Purification of recombinant proteins and in vitro kinase assay
Purification of His-Nus fusion proteins and in vitro kinase assays were performed as described previously (Schwessinger et al., 2011). In brief, the full length of PALD CDS was cloned in pDEST17 vector with a His-tag in the N terminus. The His-Nus-XA21JK plasmid was used in a previous study (Chen et al., 2014). His-Nus-XA21JK and His-PALD plasmids were transformed into BL21 cells for protein expressions. After being induced by isopropyl b-d-1-thiogalactopyranoside, proteins were purified from BL21 cells using Ni-NTA agarose beads following the manufacturer's protocol. For in vitro kinase assay, 3 mg of XA21K668 and PALD-AB proteins were incubated in 20 ml kinase buffer (50 mM Tris [pH 7.5], 10 mM MgCl 2 , 10 mM MnCl 2 , 1 mM DTT) in the presence of 1 mM unlabeled ATP or 10 mCi of [ 32 P]g-ATP for 30 min at 30 C with shaking at 900 rpm. The reac-tions were stopped by adding 53 protein sample buffer and boiled for 5 min. The phosphorylation status of fusion proteins was analyzed by autoradiography after separation of the in vitro kinase assay by 8% SDS-PAGE.

Yeast two-hybrid assay
Yeast two-hybrid experiments were carried out using the Matchmaker LexA yeast two-hybrid system (Clontech) as described in a previous study (Chen et al., 2014). The cDNA of PALD was first cloned in pENTR vector and sequenced. The full-length PALD cDNA sequence was used as a template to clone the truncation forms of PALD construct. PALD-AB and PALD-C are subcloned constructs that carry the N-terminal half (PALD-AB, aa 1-704) and the C-terminal half (PALD-C, aa 701-1256) of PALD, respectively. The other 13 truncation forms of PALD were subcloned from PALD-AB. XA21K668 was cloned and used in a previous study (Chen et al., 2014). PALD and XA21K668 cDNA were then subcloned into pB42 (PALD-pB42) and pLexA (XA21K668-pLexA) vectors, respectively. Both of PALD-pB42 and XA21K668-pLexA vectors were co-transformed into yeast EGY48 strain (Clontech) by using the Frozen-EZ Yeast Transformation II Kit (Zymo Research) and spread on an appropriate auxotrophic medium. For negative controls, the pLexA-XA21K668 was replaced by pLexA-GUS plasmid.

SUPPLEMENTAL INFORMATION
Supplemental information can be found online at Plant Communications Online.

AUTHOR CONTRIBUTIONS
T.-C.C., M.C., M.S., and P.R. conceived and designed the experiments, analyzed the data, and wrote the paper. T.-C.C., M.C., R.R., M.S., Y.W., and A.I. performed the experiments, analyzed the data, and/or prepared figures and tables. All authors reviewed drafts of the paper.