Tomato Wall-Associated Kinase SlWak1 Depends on Fls2/Fls3 to Promote Apoplastic Immune Responses to Pseudomonas syringae 1[OPEN]

Wall-associated kinases (Waks) are important components of plant immunity against various pathogens, including the bacterium Pseudomonas syringae pv. tomato ( Pst ). However, the molecular mechanisms of their role(s) in plant immunity are largely unknown. In tomato ( Solanum lycopersicum ), wall-associated kinase 1 (SlWak1), has been implicated in pattern recognition receptor (PRR)-triggered immunity (PTI) because its transcript abundance increases signi ﬁ cantly after treatment with the ﬂ agellin-derived, microbe-associated molecular patterns ﬂ g22 and ﬂ gII-28, which activate the PRRs Fls2 and Fls3, respectively. We generated two SlWak1 tomato mutants ( D wak1) using CRISPR/Cas9 gene editing technology and investigated the role of SlWak1 in tomato – Pst interactions. Late PTI responses activated in the apoplast by ﬂ g22 or ﬂ gII-28 were compromised in D wak1 plants, but PTI at the leaf surface was unaffected. The D wak1 plants developed fewer callose deposits than wild-type plants, but retained early PTI responses such as generation of reactive oxygen species and activation of mitogen-activated protein kinases upon exposure to ﬂ g22 and ﬂ gII-28. Induction of Wak1 gene expression by ﬂ g22 and ﬂ gII-28 was greatly reduced in a tomato mutant lacking Fls2 and Fls3, but induction of Fls3 gene expression by ﬂ gII-28 was unaffected in D wak1 plants. After Pst inoculation, D wak1 plants developed disease symptoms more slowly than D ﬂ s2.1/2.2/3 mutant plants, although ultimately, both plants were similarly susceptible. SlWak1 coimmunoprecipitated with both Fls2 and Fls3, independently of ﬂ g22/ ﬂ gII-28 or of BRASSINOSTEROID INSENSITIVE1-ASSOCIATED RECEPTOR KINASE1. These observations suggest that SlWak1 acts in a complex with Fls2/Fls3 and is important at later stages of PTI in the apoplast.

To overcome PTI, pathogens deliver virulence proteins (effectors) into the plant cells to interfere with MAMP detection or PTI signaling and promote disease development (Toruño et al., 2016;Wei et al., 2018).
AvrPto and AvrPtoB, two effectors from Pseudomonas syringae pv. tomato (Pst), suppress the early PTI response by interfering with the interaction of Fls2 with BAK1 (Xiang et al., 2008;Martin, 2012;Hind et al., 2016). In response to bacterial effectors, plants have evolved genes encoding NLRs, which recognize specific effectors and activate NTI Cui et al., 2015;Jubic et al., 2019;Tamborski and Krasileva, 2020). In tomato, the Pto kinase protein interacts with AvrPto or AvrPtoB and forms a complex with the NLR protein Prf resulting in the induction of NTI and inhibition of pathogen growth (Martin et al., 1993;Salmeron et al., 1996;Pedley and Martin, 2003).
Although Wak proteins have been identified as important contributors to disease resistance against various pathogens (Hu et al., 2017;Bacete et al., 2018), much remains to be learned about the molecular mechanisms they use to activate immune responses. The beststudied Wak protein, the Arabidopsis AtWAK1, recognizes cell-wall-derived oligogalacturonides (OGs) and activates OG-mediated defense responses against both fungal and bacterial pathogens (Brutus et al., 2010;Gramegna et al., 2016). In maize, the ZmWAK-RLK1 protein (encoded by Htn1) confers quantitative resistance to northern corn leaf blight by inhibiting the biosynthesis of secondary metabolites, benzoxazinoids, that suppress pathogen penetration into host tissues . Another ZmWAK protein encoded in a major head smut quantitative resistance locus qHSR1 enhances maize resistance to Sporisorium reilianum by arresting the fungal pathogen in the mesocotyl (Zuo et al., 2015). One wheat Wak protein encoded by the Stb6 gene recognizes an apoplastic effector (AvrStb6) from Zymoseptoria tritici and confers resistance to the fungal pathogen without a hypersensitive response (Saintenac et al., 2018). In rice, three OsWAKs act as positive regulators in resistance to the rice blast fungus by eliciting ROS production, activating defense gene expression, and recognizing chitin by being partially associated with the chitin receptor Chitin elicitor-binding protein (Delteil et al., 2016). Wak proteins therefore appear to exhibit extensive functional diversity and have different mechanisms to defend against pathogen infection in different plant species. The functional characterization of Wak proteins in tomato has not been reported and their possible contributions to PTI or NTI are not well understood in this species.
Tomato is an economically important vegetable crop throughout the world and its production is threatened by many pathogens including Pst, which causes bacterial speck disease and can result in severe crop losses (Jones, 1991;Kimura and Sinha, 2008). Understanding the functions of Wak proteins in tomato could therefore provide fundamental information for breeding tomato cultivars that are resistant to various pathogens. Tomato contains seven Wak and 16 Wak-like kinase genes (Zheng et al., 2016). The SlWak1 (Solyc09g014720) gene is clustered together with another three SlWak genes (Solyc09g014710, Solyc09g014730, and Sol-yc09g014740) on chromosome 9; however, the expression of only the SlWak1 gene (hereafter Wak1) is significantly induced after MAMP treatment or Pst inoculation (Rosli et al., 2013). Reduction of Wak1 gene expression in N. benthamiana leaves using virusinduced gene silencing (VIGS) compromised resistance to the bacterial pathogen Pst. However, three closely-related NbWak genes were simultaneously silenced in these experiments, making it unclear if one or a combination of NbWak genes contributed to the enhanced susceptibility to Pst (Rosli et al., 2013). To gain deeper insight into the role of Wak1 in tomato-Pst interactions, we generated two homozygous Wak1 mutant lines (Dwak1) in tomato using CRISPR/Cas9. Characterization of these Dwak1 mutants indicated that Wak1 protein acts as an important positive regulator in later stages of flagellin-mediated PTI response in the apoplast and associates in a complex with Fls2 and Fls3 to trigger immune signaling.

Generation of Wak1 Mutants in Tomato by CRISPR/Cas9
We reported previously that VIGS of three homologs of Wak1 in N. benthamiana led to enhanced susceptibility to Pst (Rosli et al., 2013). In tomato leaves, transcript abundance of the Wak1 gene (Solyc09g014720) is significantly increased after treatment with flg22, flgII-28, or csp22, suggesting Wak1 might play a role in tomato-Pst interactions (Rosli et al., 2013;Pombo et al., 2017). To study the possible role of Wak1 in plant immunity, we generated mutations in Wak1 using CRISPR/Cas9 with a guide RNA (gRNA), Wak1-gRNA1 (GTTAAGATT AGCATAAAACA; Fig. 1A), which targets the first exon of the Wak1 gene. After transformation of the cultivar Rio Grande-PtoR (RG-PtoR, which has the Pto and Prf genes), we obtained a biallelic mutant (Dwak1 4) from which two Wak1 homozygous mutant lines (Dwak1 4-1; Dwak1 4-2) were derived (Fig. 1A). Line 4-1 has a 10-bp deletion in Wak1, resulting in a premature stop codon at the 17th amino acid of the protein, whereas line 4-2 has a 1-bp deletion in Wak1, causing a premature stop codon at the 18th amino acid (Fig. 1A). The growth, development, and overall morphology of both Dwak1 mutants were indistinguishable from wildtype RG-PtoR plants (Supplemental Fig. S1).
To determine if the gRNA designed for Wak1 editing inadvertently caused mutations in other genomic regions of the Dwak1 plants, we selected seven putative sites with the highest off-target scores using Geneious R11 and Cas-OFFinder, although all of these sites had at least three mismatches compared with the spacer sequence of the Wak1 gRNA (Fig. 1B). Of the seven potential offtarget sites, two are located in the coding region of a gene, three are in the untranslated region of genes, and another two are in intergenic regions. For each site, we tested 10 to 20 independent T1 or T2 plants, with or without Cas9, and did not detect any offtarget mutations. This is not unexpected, as the gRNA we designed for Wak1 was highly specific, with little possibility to target Wak1 homologs or other genes in tomato, considering that even one mismatch in the seed sequence (the last 12 nucleotides of a gRNA spacer sequence) can severely impair or completely abrogate the editing ability of the Cas9/gRNA complex (Jiang and Doudna, 2017 To test whether the response to Pst is affected in the Dwak1 plants, we vacuum-infiltrated Dwak1 and wildtype RG-PtoR plants with the Pst strain DC3000ΔavrP-toΔavrPtoB (DC3000DD), in which avrPto and avrPtoB have been deleted and therefore cannot activate NTI. Both Dwak1 lines showed enhanced disease symptoms compared to wild-type plants 4 d after bacterial inoculation, with ;6-fold more bacterial growth compared to the wild-type plants ( Fig. 2A). No differences in symptoms or bacterial populations were observed between the Dwak1 and wild-type plants when they were inoculated with DC3000ΔavrPtoΔavrPtoBDfliC (DC3000DDD; Fig. 2B), which lacks avrPto and avrPtoB and the flagellin-encoding gene fliC. This result indicates that Wak1 plays a role in flagellin-mediated PTI.
To test whether Wak1 contributes to NTI, the Dwak1, RG-PtoR, and Rio Grande-prf3 plants (RG-prf3, which contains a mutation in Prf that makes the Pto pathway nonfunctional) were inoculated with DC3000. Six days post inoculation (dpi), the Dwak1 and RG-PtoR plants had no disease symptoms, whereas the RG-prf3 control showed severe disease symptoms (Fig. 2C). Bacterial populations were ;30-fold less in the Dwak1 and RG-PtoR plants compared to RG-prf3. Wak1 therefore has no observable role in NTI.
The two Dwak1 mutant lines were derived from the same primary transformant and it was formally possible that another mutation induced during tissue culture is responsible for the enhanced susceptibility to Pst. We therefore developed F1 hybrids by crossing the Dwak1 plants to RG-PtoR plants (Supplemental Fig. S2). Sequencing confirmed that all F1 hybrids were heterozygous for the Wak1 mutation. F1 hybrids that were vacuum-infiltrated with DC3000DD developed disease symptoms and supported bacterial populations similar to RG-PtoR plants (Supplemental Fig. S2A), indicating Wak1 is a dominant allele. Four F1 plants (two were 210 bp/ wild type and two were 21 bp/wild type) were selfed to develop F2 populations. After inoculation of 117 F2 plants with DC3000DD, we observed a segregation ratio of 3 (resistant) to 1 (susceptible; Supplemental Fig. S2B). Sequencing revealed all resistant plants were either homozygous wild type or heterozygous, while the susceptible plants were homozygous for the wak1 mutation (Supplemental Fig. S2C). Combined with the lack of off-target mutations, these disease assays with F2 populations strongly support that the susceptibility to Pst of Dwak1 plants is due to the CRISPR/Cas9-induced loss-of-function mutations in the Wak1 gene. Figure 2. The Dwak1 tomato plants are compromised in flagellin-mediated PTI but unaffected in NTI. A to C, Four-weekold Dwak1 plants and wild-type RG-PtoR plants were vacuum-infiltrated with 5 3 10 4 cfu mL 21 DC3000DavrPtoDavrPtoB (DC3000DD; Α), 5 3 10 4 cfu mL 21 D C 3 0 0 0 Da v r P t o Da v r P t o B Dfl i C (DC3000DDD; Β), and 1 3 10 6 cfu mL 21 DC3000 (C). Photographs of disease symptoms were taken 4 (A and B) and 6 (C) dpi. Bacterial populations were measured at 3 h (day 0) and 2 d (day 2) after infiltration. Bars show means 6 SD (SD). Different letters indicate significant differences based on a one-way ANOVA followed by Tukey's Honest Significant Difference post hoc test (P , 0.05). ns, No significant difference. Three or four plants for each genotype were tested per experiment. The experiment was performed three times with similar results.

Wak1 Mutant Plants Are Compromised in PTI Induced by flg22 and flgII-28
The observation that Dwak1 plants are more susceptible to DC3000DD but show no differences compared to wild-type plants for their response to DC3000DDD that lacks flagellin, suggests that Wak1 is involved in immune responses mediated by flg22 and/or flgII-28. To further test this, we performed a "PTI protection" assay using a heat-killed Pst strain lacking flagellin and three type-III effectors (DC3000DavrPtoDavrPtoBDhopQ1-1DfliC; DC3000DDDD) complemented with a construct expressing fliC from either DC3000 (which has active flg22 and flgII-28) or Pseudomonas cannabina pv. alisalensis ES4326 (only flgII-28 is active; Hind et al., 2016), or an empty vector (EV) as a control (Fig. 3A). Because both of the Dwak1 lines were similarly susceptible to DC3000DD, most subsequent experiments were focused on the 4-1 line. Dwak1 4-1 plants were first infiltrated with the various suspensions of heat-killed bacteria to induce PTI and then challenged with DC3000DDD 16 h later. Wild-type plants pretreated with Pst DC3000DDDD with an EV supported a significantly higher bacterial population than plants pretreated with the heat-killed bacterial suspensions containing either DC3000 fliC or ES4326 fliC (7.5-fold and 3.3-fold, respectively), indicating that pretreatment of wild-type plants activated PTI defenses due to recognition of flg22 and/or flgII-28. The Dwak1 plants, however, supported higher bacterial populations regardless of the pretreatment, indicating the PTI response was compromised (Fig. 3A).
We next performed the PTI protection assay using the synthetic peptides flg22 and flgII-28. Plants were first syringe-infiltrated with buffer alone, 1 mM of flg22, or 1 mM of flgII-28, and then challenged with DC3000DDD 16 h later as described above (Fig. 3B). Two days later, wild-type plants that were pretreated with either flg22 or flgII-28 had significantly lower bacterial populations compared to the buffer-only treatment. In contrast, no significant differences in bacterial populations regardless of pretreatment were observed in Dwak1 plants. Collectively, these experiments demonstrate that Wak1 plays an important role in PTI that is activated by two flagellin-derived MAMPs.

Dwak1 Plants Are Not Compromised in PTI Responses on the Leaf Surface, or in Stomatal Numbers or Conductance
Pst inoculation experiments using vacuum infiltration assess PTI responses primarily in the apoplast. To test if Wak1-mediated immunity also plays a role in PTI on the leaf surface, we spray-inoculated Dwak1 and wild-type RG-PtoR plants with DC3000DD. This inoculation method requires the pathogen to enter the apoplastic space through stomata or natural openings. Interestingly, in contrast to experiments using vacuum infiltration, both wild-type and Dwak1 plants developed disease symptoms after spray inoculation that were indistinguishable both in the amount of time until they developed and in their ultimate severity (Fig. 4A). Consistent with this observation, there was no significant difference in DC3000DD populations in any of the lines after spray inoculation (Fig. 4B). Thus, Wak1 does not appear to play an important role in PTI responses on the leaf surface. Measurements of stomatal numbers and of stomatal conductance as an Figure 3. The Dwak1 plants are compromised in two PTI induction assays. A, Four-week-old Dwak1 plants (4-1) and wild-type RG-PtoR plants were first syringe-infiltrated with 1 3 10 8 cfu mL 21 of heatkilled DC3000DavrPtoDavrPtoBDhopQ1-1DfliC (DC3000DDDD) complemented with a fliC gene from DC3000 or ES4326, or no fliC (EV). Sixteen hours later, the whole plants were vacuum-infiltrated with DC3000DavrPtoDavrPtoBDfliC (DC3000DDD) at 5 3 10 4 cfu mL 21 . Bacterial populations were measured 2 d after the infiltration. B, Plants (Dwak1 4-1 and wild type) were first syringe-infiltrated with buffer only (mock; 10 mM of MgCl 2 ), 1 mM of flg22, or 1 mM of flgII-28, respectively. Sixteen hours later, plants were vacuum-infiltrated with DC3000DDD at 5 3 10 4 cfu mL 21 . Bacterial populations were measured 2 d later. Bars in A and B represent means 6 SD. Different letters indicate significant differences based on a one-way ANOVA followed by Tukey's Honest Significant Difference post hoc test (P , 0.05).
indicator of stomatal activity revealed no differences between wild-type and Dwak1 plants, further indicating that Wak1 does not play a role at the leaf surface (Fig. 4, C  Generation of ROS and activation of MAPK cascades are two typical early PTI-associated responses in plants (Nguyen et al., 2010;Zipfel, 2014). To investigate whether Wak1 participates in these responses, we first performed ROS assays using flg22, flgII-28, or csp22. We observed no differences in ROS production in Dwak1 plants compared to wild-type plants when treated with any of these MAMPs (Fig. 5, A and B; Supplemental Figs. S3 and S4). We also observed no difference between wild-type and Dwak1 plants for their ability to activate MAPKs in response to flg22 and flgII-28 (Fig. 5C).
Callose deposition is a response associated with later stages of PTI, and one that is regulated independently or downstream of MAPK activation (Li et al., 2016). We measured callose deposition by challenging Dwak1 and wild-type plants using a nonpathogenic bacterial strain, Pseudomonas fluorescens 55, a strong inducer of PTI (Rosli et al., 2013). Compared to wild-type plants, Dwak1 plants showed significantly reduced callose deposition 1 d after vacuum infiltration of Pf 55 (Fig. 5D). These observations therefore indicate that Wak1 functions at a later stage of the PTI response in a flagellin-induced process independent of ROS production and MAPK activation.
The Increase in Wak1 Transcript Abundance upon flgII-28 Treatment Is Fls3-Dependent In tomato, the transcript abundance of Wak1 is low in unchallenged conditions, but is significantly higher after Pst inoculation (Rosli et al., 2013). To gain insight into the transcriptional regulation of Wak1 and Fls3 during the immune response, we used reverse-transcription quantitative PCR (RT-qPCR) to measure Wak1 and Fls3 transcript abundance after treatment of wild-type leaves with flgII-28 (Fig. 6A). The relative abundance of Wak1 or Fls3 transcripts at various time points after syringe-infiltrating 1 mM of flgII-28 was compared to a mock treatment (10 mM of MgCl 2 ). Both Wak1 and Fls3 transcript abundance increased significantly at 6 and 8 h after syringeinfiltrating flgII-28 compared to the mock control (Fig. 6A).
To investigate possible codependence of Wak1 and  DC3000DavrPtoDavrPtoB. Photographs of disease symptoms were taken 6 dpi. Photographs show a representative plant and leaflet from each line. B, Bacterial populations were measured at 3 h (day 0) and 2 d (day 2) after spray inoculation. Bars show means 6 SD. ns, No significant difference using a one-way ANOVA followed by Tukey's Honest Significant Difference post hoc test (P , 0.05). C, Stomatal index taken from wild-type and Dwak1 4-1 plants. Photographs from the abaxial epidermis of the leaves were taken using an epifluorescence microscope and the number of cells and both closed and open stomata were counted manually. The stomatal index was calculated as the percentage of stomata number per total number of cells (stomata plus epidermal cells). Five photographs per biological replicate were analyzed. Bars represent the mean of four biological replicates with their corresponding SD. D, Stomatal conductance (millimoles water) was measured on the abaxial side of leaflets on the third leaf. Data correspond to the average of two leaflets from at least four biological replicates per line, with 6SD. ns, No significant difference using Student's t test (P , 0.05).
unpublished data) and the Fls3 transcript abundance in Dwak1 plants after treatment with flgII-28. The abundance of Wak1 transcripts was greatly reduced in the Dfls2.1/2.2/3 plants compared to wild-type plants, whereas Fls3 abundance was not significantly different in Dwak1 or wild-type plants (Fig. 6, B and C). These results indicate that Wak1 gene expression is regulated by the Fls3 pathway and its function likely occurs downstream of the mechanism inducing Fls3 gene expression. , or water only, and relative light units (RLUs) were measured over 45 min. One-way ANOVA followed by Tukey's Honest Significant Difference post hoc test (P , 0.05) was performed at 24 min (peak readout) and 45 min after treatment with flg22 or flgII-28. No significant difference was observed between Dwak1 and wild-type plants in either treatment. C, Leaf discs from Dwak1 (4-1) or wild-type RG-PtoR plants were treated with water, 10 nM of flg22, or 25 nM of flgII-28 for 10 min. Proteins were extracted from a pool of discs from three plants and subjected to immunoblotting using an antipMAPK antibody that detects phosphorylated MAPKs. The photographs shown are derived from the same immunoblot with identical exposure times. Ponceau staining and the nonspecific band (*) indicate equal loading of protein. This experiment was performed three times with similar results. D, Wild-type and Dwak1 plants (4-1) were vacuum-infiltrated with 1 3 10 8 cfu mL 21 P. fluorescens 55. Leaf samples were taken 24 h after infiltration, de-stained with 96% ethanol, and stained with aniline blue for 1 h. Callose deposits were analyzed using an epifluorescence microscope. Top, Representative photographs of wild-type and Dwak1 plants taken for callose deposition estimation. Red spots indicate the callose deposits observed and used for quantification. Scale bars 5 100 mm. Bottom, Total number of callose deposits per mm 2 quantified in each group of plants. Fifteen photographs per biological replicate were analyzed. Bars represent the mean of four biological replicates with their corresponding SD. The asterisks represent a significant difference using Student's t test (**P , 0.01).

Dwak1 Plants Develop Bacterial Speck Disease Symptoms
More Slowly than Dfls2.1/2.2/3 Plants To determine the relative contributions of Wak1 and Fls2/Fls3 to PTI, we next compared the response of Dwak1 and Dfls2.1/2.2/3 plants to DC3000DD (Fig. 7). Three days after inoculation, the Dfls2.1/2.2/3 plants showed more severe disease symptoms than Dwak1 plants or wild-type plants, but by 4 dpi both the Dwak1 and Dfls2.1/2.2/3 plants developed more disease symptoms than the wild-type plants (Fig. 7A). There was no visible difference in disease symptoms between the Dwak1 and Dfls2.1/2.2/3 plants 4 to 10 dpi (Fig. 7A). Two days after inoculation, the bacterial population in the Dfls2.1/2.2/3 and Δwak1 plants was 6-fold and 4-fold higher than the wild-type plants, respectively, with no statistically significant difference in bacterial populations between the Dwak1 and Dfls2.1/ 2.2/3 plants (Fig. 7B). Thus, although there was a delay in disease progression at the whole plant level, this delay was not reflected in a difference in bacterial populations at 2 dpi. A delay in disease progression would be expected if the fls2.1/2.2/3 mutations result in the loss of both early and later-stage PTI whereas the wak1 mutation compromises primarily later-stage PTI responses.
Wak1 Occurs in a Complex with Fls2 and Fls3 Independent of flg22, flgII-28, or BAK1 The results above indicate that Wak1 plays a major role in flg22and flgII-28-induced processes that occur in the apoplast later in the PTI response. We considered Figure 6. Transcript abundance changes of Wak1 are dependent on the presence of Fls3 in tomato. A, Transcript abundance of Wak1 and Fls3 genes measured by RT-qPCR at the times shown after treatment with 1 mM of flgII-28 compared to a bufferonly control (10 mM of MgCl 2 ; mock treatment). Each treatment included three biological replicates and three technical replicates. SlArd2 (Solyc01g104170) was used as the reference gene for quantification. Bars represent means 6 SD. B, RT-qPCR was used to measure transcript abundance of Wak1 6 h after treatment of Dfls2.1/2.2/3 or wild-type leaves with 1 mM of flgII-28. Bars represent the mean 6 SD. Asterisks indicate a significant difference using Student's t test (**P , 0.01). C, RT-qPCR was used to measure transcript abundance of Fls3 6 h after treatment of Dwak1 (4-1) or wild-type leaves with 1 mM of flgII-28. Bars represent the mean 6 SD. ns, No significant difference using Student's t test (P , 0.05).
the possibility that Wak1 acts in a complex with Fls2 and Fls3 similar to what has been reported for FLS2 and FERONIA (FER) in Arabidopsis (Stegmann et al., 2017). We therefore used transient expression of proteins in N. benthamiana leaves and coimmunoprecipitation to investigate if Wak1 physically associates with Fls2, Fls3, or the coreceptor BAK1 and, if so, whether the interaction is affected by the presence of flg22 or flgII-28. We observed a weak, but reproducible and specific, interaction of Wak1 with both Fls2 and Fls3 with the interactions occurring independently of flg22, flgII-28, or the presence of BAK1( Fig. 8A; Supplemental Fig. S5). As expected, Fls3 and Fls2 each interacted strongly with BAK1 only in the presence of flgII-28 or flg22, respectively. No interaction was observed between Wak1 and BAK1 proteins (Fig. 8B). Additionally, Wak1 did not affect the accumulation of the Fls2, Fls3, or BAK1 proteins or vice versa ( Fig. 8; Supplemental Fig. S5).

DISCUSSION
The tomato Wak1 gene was first identified as a flagellin-induced, repressed-by-effectors gene in the immune response against P. syringae (Rosli et al., 2013). When its expression was reduced by VIGS in N. benthamiana, the morphology of the plants was unaffected but their ability to activate PTI was compromised, leading to more severe disease symptoms and enhanced growth of a virulent Pst strain (Rosli et al., 2013). The interpretation of these experiments was limited somewhat by the fact that three N. benthamiana Wak1 homologs were silenced by the tomato Wak1 VIGS construct and, as is typical for VIGS, their transcripts were not completely eliminated (they were reduced by ;50%). Thus, whether one, or more, of the Wak1 homologs in N. benthamiana play a role in PTI was unclear as was the degree to which a complete knockout of the Wak1 genes might affect PTI or affect plant morphology. Here we have addressed these limitations by developing two CRISPR/Cas9-mediated Wak1 mutants in tomato and using them to investigate the contributions of Wak1 to several PTI-associated responses and to resistance to P. syringae. As elaborated upon below, our results indicate Wak1 gene expression is induced by the Fls2 and Fls3 pathways in tomato; the Wak1 protein associates in a complex with Fls2 and Fls3; and Wak1 plays an important role in later stages of flagellininduced PTI.
Consistent with our earlier observations of Wak1silenced N. benthamiana plants, the Dwak1 tomato plants developed more severe disease symptoms compared to wild-type plants and supported larger populations of Pst; they also had wild-type morphology. Interestingly, the differences in pathogen responses were abolished when the Pst strain used for inoculation lacked flagellin, suggesting that either flg22 and/or flgII-28 and their corresponding receptors Fls2 and Fls3 play a key role in activating Wak1-mediated responses. In fact, subsequent experiments using Pst strains with Figure 7. The Dwak1 plants develop disease symptoms more slowly than Dfls2.1/2.2/3 plants. Four-week-old Dwak1 (4-1), Dfls2.1/2.2/3 or wild-type RG-PtoR plants were vacuum-infiltrated with 5 3 10 4 cfu mL 21 DC3000DavrPtoDavrPtoB. A, Photographs were taken at 3, 4, 5, or 10 dpi. The red arrow points to more extensive disease on the Dfls2.1/ 2.2/3 plant at 3 dpi. B, Bacterial populations were measured 3 h (day 0) and 2 dpi (day 2). Bars represent means 6 SD. Different letters indicate significant differences based on a one-way ANOVA followed by Tukey's Honest Significant Difference post hoc test (P , 0.05). ns, No significant difference. Three or four plants for each genotype were tested per experiment. This experiment was performed twice with similar results. variant FliC proteins, or using synthetic flg22 and flgII-28 peptides, confirmed that either one of these MAMPs is sufficient to induce Wak1-dependent PTI. At this stage of the work this dependence potentially could be explained simply by the fact that both of these MAMPs are able to significantly upregulate expression of the Wak1 gene.
Several observations support the hypothesis that Wak1 acts at a later stage of the PTI response in tomato. First, the Dwak1 plants showed no difference from wild-type plants when Pst was spray-inoculated, a method that assays for PTI responses at the leaf surface. The importance of PTI on the leaf surface has been extensively documented in Arabidopsis where a major regulator of this response is the activity of FLS2 in the stomata (Melotto et al., 2006(Melotto et al., , 2008(Melotto et al., , 2017. Our observations suggest that Wak1 does not act in PTI on the leaf surface but instead exerts its function at a later stage, after Pst enters the apoplastic space as simulated by vacuum infiltration. Second, Dwak1 plants showed no defects in their ability to produce ROS upon exposure to csp22, flg22, or flgII-28 or activate MAPKs in response to flg22 and flgII-28. Both of these responses occur early (within minutes) in leaves that are exposed to MAMPs. Third, Fls3 gene expression induced by flgII-28 was the same in Dwak1 plants as it was in wild-type plants. Transcriptional changes also occur rapidly (within 1 h) of MAMP treatment (Pombo et al., 2017). As expected, the induction of Wak1 gene expression by flgII-28 was compromised in Dfls2.1/2.2/3 plants. Fourth, the Dwak1 plants produced just 25% of the callose deposits observed in wild-type plants in response to P. fluorescens, a source of flagellin and other MAMPs. Callose deposition occurs later than ROS production and MAPK activation, and contributes to cell-wall strengthening that may inhibit the infection process (Nguyen et al., 2010;Voigt, 2014). Finally, the Dwak1 plants developed disease symptoms more slowly than did Dfls2.1/2.2/3 plants. This would be expected if the Dfls2.1/2.2/3 mutations result in the loss of both early (e.g. ROS, MAPK activation, transcriptional reprogramming) and later-stage PTI (callose deposition), whereas the Wak1 mutation compromises primarily later-stage PTI responses. Importantly, however, both Dwak1 and Dfls2.1/2.2/3 plants ultimately developed the same severe disease symptoms that demonstrate the critical role that Wak1 plays in the host response to Pst.
The dependence of Wak1-mediated PTI on Fls2 and Fls3 activity could be explained, in part, by the induction of Wak1 gene expression by the Fls2 and Fls3 pathways. However, our observations also raised the possibility that Wak1 resides in a complex that contains Fls2 and Fls3 and its function involves these receptors. We tested this hypothesis and found that Wak1 does coimmunoprecipitate with Fls2 and Fls3 in a MAMPindependent manner and it does not affect accumulation of Fls2/Fls3 proteins. This is reminiscent of the Arabidopsis malectin-like receptor kinase, FER, which was found to weakly associate with FLS2 independent of flg22 treatment and also had no effect on FLS2 accumulation (Stegmann et al., 2017). It is possible that Wak1, like FER, may act as an important cell-wall-associated scaffold to regulate immune receptor-complex formation. Tomato Wak1 did not coimmunoprecipitate with BAK1, and BAK1 was not required for the Wak1-Fls2/Fls3 interactions. In contrast, FER weakly associates with BAK1 and the interaction is enhanced upon flg22 treatment, but whether BAK1 is required for the weak association of FER-FLS2 was not investigated (Stegmann et al., 2017).
Based on our observations, we propose a model for the role of Wak1 in PTI (Fig. 9). In this model, Wak1 transcript abundance is greatly increased upon Figure 8. Wak1 associates with Fls3 independently of flgII-28 and BAK1, and Wak1 does not associate with BAK1. Proteins were extracted from N. benthamiana leaves expressing Fls3-GFP in combination with AtBAK1-Myc and/or Wak1-HA after treatment with or without 1 mM of flgII-28 for 2 min and were used for immunoprecipitation using anti-GFP magnetic agarose beads (A) or anti-Myc magnetic beads (B). Wak1 was pulled down with Fls3 (A) but not with BAK1 (B) after treatment with or without flgII-28. Wak1, BAK1, Fls3, GFP, and Yellow Fluorescent Proteins (YFP) were detected by immunoblotting with ⍺-HA, ⍺-Myc, or ⍺-GFP antibodies. This experiment was repeated three times with similar results.
activation of the PRRs Fls2 and Fls3. We hypothesize this gene expression occurs primarily when Pst enters the apoplastic space and that Wak1 is not expressed in leaf surface or stomatal cells. Increased transcript abundance leads to increased Wak1 protein accumulation and subsequent localization to a cellwall-associated protein complex that contains Fls2 and Fls3 and possibly other PRRs. Wak1 might act as a receptor of a damage-associated molecular pattern (DAMP), such as OGs. Binding of such a DAMP might impact the association of Wak1 with the Fls2/ Fls3 complex to promote stabilization and accumulation of the PRRs, enhance the interaction of Wak1 with PRRs, or possibly stimulate PRR kinase activity. Whatever the mechanism, the presence of Wak1 in this Wak plays a critical role in later stages of PTI, including callose deposition and other processes that ultimately inhibit growth of virulent Pst.
This model gives rise to several questions that will need to be addressed in the future. First, why is Wak1 not active in plant cells on the leaf surface, including stomata, but only functions when Pst enters the apoplastic space? This could be due to lack of Wak1 gene expression, protein accumulation, association with the Fls2/Fls3 complex, or kinase activity in leaf surface cells. Second, how does Wak1 affect the cellwall-associated Fls2/Fls3 complex and is its activity in this complex influenced by perception of a DAMP? In Arabidopsis, AtWAK1 was demonstrated to bind pectin and OGs in vitro (Kohorn et al., 2009), and was identified as the receptor for OGs in vivo (Brutus et al., 2010). Does Wak1 bind OGs and, if so, do OGs impact the way Wak1 associates with Fls2/Fls3 and its role in PTI? Finally, it will be interesting to investigate possible differences in the transcriptome, metabolome, and proteome of the Dwak1 mutants in comparison with wild-type plants to understand what are the later PTI responses to which Wak1 contributes. CONCLUSION We generated two SlWak1 tomato mutants (Dwak1s) using CRISPR/Cas9 and investigated the role of SlWak1 using various Pst strains, immune response assays, RT-qPCR, and protein biochemistry. We discovered that late PTI responses activated by flg22 or flgII-28 are compromised in the apoplast but not on the leaf surface in Dwak1 plants. Dwak1 plants developed fewer callose deposits than wild-type plants but retained the ability to activate early PTI responses such as generation of ROS and activation of MAPKs upon exposure to flg22 and flgII-28. After Pst inoculation, Dwak1 plants developed disease symptoms more slowly than Dfls2.1/2.2/3 mutant plants, although both plants ultimately were similarly susceptible. SlWak1 coimmunoprecipitated with both Fls2 and Fls3 independently of flg22/flgII-28 or BAK1. These observations suggest that SlWak1 acts in a complex with Fls2/ Fls3, and plays an important role at later stages of the PTI in the apoplast.

Generation of Wak1 Tomato Mutants Using CRISPR/Cas9
To mutate the Wak1 gene in tomato (Solanum lycopersicum), we designed two gRNAs (Wak1-gRNA1: GTTAAGATTAGCATAAAACA and Wak1-gRNA2: GGGGCGGTGGCATTCGTTGG; Supplemental Table S1) targeting the first exon of Wak1 using the software Geneious R11 (Kearse et al., 2012). Each gRNA Figure 9. A model for the role of SlWak1 in PTI. Transcript abundance of Wak1 increases in mesophyll cells upon activation of the Fls2 or Fls3 pathways. This leads to increased accumulation of Wak1 protein, which is then localized to the cell wall and here it joins a complex containing Fls2 and Fls3. Wak1 might act as a scaffold and could be a receptor for a DAMP. Wak1 functions to promote deposition of callose at the cell wall and other immune responses that inhibit multiplication of the pathogen.
cassette was cloned into a Cas9-expressing binary vector (p201N:Cas9) by Gibson assembly as described in Jacobs et al. (2017). Tomato transformation was performed at the biotechnology facility at the Boyce Thompson Institute. Agrobacterium cells containing each gRNA/Cas9 construct were pooled together and used for transformation into the tomato cultivar RG-PtoR, which has the Pto and Prf genes. To determine the mutation type, genomic DNA was extracted from cotyledons or young leaves of each transgenic plant using a modified cetyl trimethylammonium bromide method (Murray and Thompson, 1980). Genomic regions spanning the target site of the Wak1 gene were amplified with specific primers (Supplemental Table S1) and sequenced at the Biotechnology Resource Center at Cornell University. Geneious R11 and the web-based tool called Tracking of Indels by Decomposition (https:// tide.deskgen.com; Brinkman et al., 2014) were used to determine the mutation type and frequency using the sequencing files (ab1. format) as described in Zhang et al. (2020).

Off-Target Evaluation
To investigate potential off-target mutations caused by gRNAs in the Dwak1 plants, Wak1-gRNA1, which induced target mutations in Wak1 in the transgenic plants, was used as a query to search putative off-target sites across the tomato genome with up to four nucleotide mismatches by Geneious R11 or with up to three nucleotide mismatches by Cas-OFFinder (Bae et al., 2014). Seven potential off-target sites with the highest similarity to the spacer sequence of Wak1-gRNA1 were chosen for evaluation. Genomic regions spanning the putative off-target sites were amplified with specific primers (Supplemental Table  S1) and PCR amplicons were sequenced to determine if off-target mutations were induced at those sites.

Bacterial Inoculation Assay
Four-week-old Dwak1 and wild-type plants were vacuum-infiltrated with various Pst DC3000 strains at different titers, including DC3000ΔavrPtoΔavrP-toB (DC3000DD) or DC3000ΔavrPtoΔavrPtoBΔfliC (DC3000DDD) at 5 3 10 4 cfu mL 21 or DC3000 at 1 3 10 6 cfu mL 21 . Three to four plants per line were tested with each bacterial strain. Bacterial populations were measured at 3 h and 2 dpi. Disease symptoms were photographed 4 or 5 d after bacterial infection. Dwak1 and wild-type plants were also spray-inoculated with DC3000DD at 1 3 10 8 cfu mL 21 . In this case, the leaf surfaces were sterilized with 15% H 2 O 2 for 5 min and then rinsed thoroughly with sterile water before sampling for measuring bacterial populations. Photographs of disease symptoms were taken at 6 dpi.

PTI Protection Assay
Four leaflets on the third leaf of 4-week-old plants were first syringe-infiltrated with 1 3 10 8 cfu mL 21 of heat-killed DC3000DavrPtoDavrPtoBDhopQ1-1DfliC (DC3000DDDD) complemented with a fliC allele from DC3000 or ES4326, or no fliC (EV). Note that hopQ1 was deleted from this strain to allow its use on Nicotiana benthamiana where HopQ1 activates NTI, but HopQ1 is not relevant to the experiment shown in Figure 3A. Sixteen hours later, whole plants were vacuuminoculated with DC3000DavrPtoDavrPtoBDfliC (DC3000DDD) at 5 3 10 4 cfu mL 21 . Bacterial populations were measured at 2 dpi. Alternatively, plants were first syringe-infiltrated with 1 mM of flg22 (GenScript), 1 mM of flgII-28 (EZBiolab), or buffer alone (10 mM of MgCl 2 ), respectively. Plants were inoculated with DC3000DDD 16 h later and bacterial populations were measured at 2 dpi as described above.

Measurement of Stomata Number and Stomata Conductance
Leaf samples were taken from Dwak1 and wild-type plants. Photographs from the abaxial epidermis of the leaves were taken using a model no. BX51 epifluorescence microscope (Olympus) and the number of cells and both closed and open stomata were counted manually. The stomata index was calculated as the percentage of stomata number per total number of cells (stomata plus epidermal cells). Stomatal conductance was measured at 2 PM (8 h after lights went on), using a leaf porometer (SC1 Decagon Devices) on the abaxial side of two leaflets of the third leaf from four plants per line.

ROS Assay
ROS production was measured as described in Hind et al. (2016). In brief, leaf discs were collected and floated in water overnight (;16 h). Water was then removed and replaced with a solution containing either 50 nM of flg22 (QRLSTGSRINSAKDDAAGLQIA) or 50 nM of flgII-28 (ESTNILQRMRE-LAVQSRNDSNSSTDRDA), in combination with 34 mg mL 21 of luminol (Sigma-Aldrich) and 20 mg mL 21 of horseradish peroxidase. ROS production was then measured over 45 min using a Synergy-2 microplate reader (BioTek). Three to four plants per line and three discs per plant were collected for each experiment.

Callose Deposition
Four-week-old plants were vacuum-infiltrated with 1 3 10 8 cfu mL 21 Pseudomonas fluorescens 55, a strong inducer of PTI (Rosli et al., 2013). Leaf samples were taken 24-h post infiltration, cleared with 96% ethanol, and stained with aniline blue for 1 h. Callose deposits were visualized with an epifluorescence microscope (BX51; Olympus). For each of the images analyzed, callose deposits were first selected manually to avoid counting trichomes or stomata. Valid deposits consisted of spots ranging between 57 and 5,674 mm 2 (8.5-85mm diameter) as described in Nguyen et al. (2010). Quantification was performed using the Fiji package in the software ImageJ (https://imagej.nih. gov/ij/). Fifteen photographs per biological replicate were analyzed using four plants per line.

Coimmunoprecipitation
Agrobacterium tumefaciens strains (GV31011 pMP90) carrying a Gateway binary vector with Fls2, Fls3, BAK1, Wak1, or GFP/YFP) were infiltrated into leaves of 4-week-old N. benthamiana. Leaves were treated with either 1 mM of flg22, 1 mM of flgII-28, or buffer alone for 2 min before harvesting. Total protein was extracted from 500 mg of N. benthamiana leaves in 1.5 mL of extraction buffer consisting of 50 mM of Tris-HCl at pH 7.5, 150 mM of NaCl, 0.5% Triton X-100, 1% (v/v) plant protease inhibitor cocktail (Sigma-Aldrich), 1 mM of Na3VO4, 1 mM of NaF, and 20 mM of b-glycerophosphate. Soluble proteins were incubated with 20 mL of GFP-Trap_MA slurry (Chromotek) or anti-Myc magnetic beads (Thermo Fisher Scientific) per sample for 2 h at 4°C, followed by washing three times with cold extraction buffer, and one more wash with cold 50 mM of Tris-HCl at pH 7.5. Proteins were eluted with 40-mL 23 Laemmli sample buffer and boiled at 95°C for 5 min. For input samples, 8-mL soluble protein mixed with 23 sample buffer was loaded for gel electrophoresis. Proteins were loaded on 8% SDS-PAGE gel, blotted on PVDF membrane (Merck Millipore), treated with appropriate antibodies, and detected by Immobilon Forte western HRP substrate (Millipore Sigma).

RT-qPCR
Four leaflets from the third leaf of 5-week-old plants were first syringeinfiltrated with 1 mM of flgII-28 or buffer. Three plants were used for each treatment and two biological replicates were performed. Leaf tissues were collected 0.5, 1, 2, 4, 6, and 8 h after infiltration, immediately frozen in liquid N 2 and stored at 280°C until used. Total RNA was isolated using an RNeasy Plant Mini Kit (Qiagen). RNA (4 mg) was treated with TURBO DNA-free DNase (Thermo Fisher Scientific) twice, each for 30 min at 37°C. First-strand complementary DNA was synthesized from 2 mg of RNA using SuperScript III (Thermo Fisher Scientific). Quantitative PCR was performed with specific primers (Supplemental Table S1) using the QuantStudio 6 Flex Real-Time PCR System (Thermo Fisher Scientific) and cycling conditions for PCR were 50°C for 2 min, 95°C for 10 min, and then 40 cycles of 95°C for 30 s, 56°C for 30 s, and 72°C for 30 s.

Accession Numbers
Sequence data from this article can be found in the Plant Genome Editing Database (http://plantcrispr.org) under Solyc09g014720.

Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. The growth, development, and morphology of Wak1 plants was indistinguishable from wild-type RG-PtoR tomato plants.
Supplemental Figure S2. Enhanced susceptibility to DC3000DD cosegregates with the Wak1 mutations.
Supplemental Figure S3. The Dwak1 plants are not affected in flg22 or flgII-28-induced ROS production.
Supplemental Figure S4. The Dwak1 plants are not affected in csp22induced ROS production.
Supplemental Table S1. Primers used in this study.