The C. michiganensis type strain LMG7333, its mutant strains, Tn::pat-1_Cm and Tn::celA_Cm generated by transposon mutagenesis, and the complemented strains were used in this study (Hwang et al., 2019). All bacterial strains were grown on KB medium (20 g of protease peptone no. 3, 1.5 g of K₂HPO₄, 6 ml of 1 M MgSO₄, and 16 ml of 50% glycerol per liter) supplemented with appropriate antibiotics: kanamycin (100 μg ml⁻¹), neomycin (100 μg ml⁻¹), and chloramphenicol (10 μg ml⁻¹) at 26°C for 24–48 h. For plant inoculation, single colonies of cultured C. michiganensis strains were incubated in KB broth overnight with shaking at 140 rpm, and then cells were collected by centrifugation and resuspended in 10 mM of MgCl₂.

Tomato (Solanum lycopersicum L., cv. “Betatini”) and N. tabacum (cv. “Samsun”) plants were grown in a growth chamber at 26°C with a 14:10-h light:dark photoperiod condition. Then, 2- or 3-week-old tomato plants and 6-week-old N. tabacum plants were used for disease and HR assays, respectively.

To select a mutant with a transposon insertion in the pat-1_Cm gene, we screened approximately 1,400 C. michiganensis strain LMG7333 mutants generated with a transposon, Tn1409Cβ, in the vector pKGT452Cβ (Kirchner et al., 2001; Hwang et al., 2019) by virulence assay in tomato and by PCR using specific primer set (Hwang et al., 2019). The transposon insertion site was determined by whole genome sequencing of the mutant using Illumina sequencing and genome comparison with the genome sequence of the WT C. michiganensis strain LMG7333 (GenBank accession nos. CP080437, CP080438, and CP080439). Furthermore, the single transposon insertion was confirmed by southern hybridization using the chloramphenicol resistance gene as a probe.

To remove the large plasmid (pCM2) carrying the pat-1_Cm gene, the WT C. michiganensis strain LMG7333 was incubated at 26°C for 2 days and moved to 37–42°C for 3 days. This temperature change was repeated several times in a fresh medium, and then the bacterial culture was spread on KB plates. To check the presence of pCM2, colonies grown on KB plate were screened by PCR using primer sets targeting plasmid backbone genes and the pat-1_Cm gene (Supplementary Table 1). The pCM2-cured strain, LMG7333ΔpCM2, was selected and used for this study.

Two-week-old tomato seedlings were inoculated with C. michiganensis strains by the root-dipping inoculation method described by Hwang et al. (2019). Inoculums were prepared from freshly cultured bacteria, and their concentration was adjusted to OD₆₀₀ = 2.0 [1 × 10⁹ colony forming units (CFU) ml⁻¹] with 10 mM MgCl₂. Whole tomato seedlings were pulled out of the soil, and, if necessary, trimmed with sterile scissors. Then, the prepared seedlings were submerged into tubes containing 1 ml of bacterial inoculum for 30 min. As a negative control (mock), seedlings were inoculated with 10 mM MgCl₂. Inoculated seedlings were transplanted into soil again in mini pots and grown in a growth chamber at 26°C. Approximately 10–14 days after inoculation (dai), the disease severity of above-ground disease symptoms of all seedlings was evaluated. Wilting symptom severity for each seedling was rated from 0 to 5 scales based on disease index defined by Hwang et al. (2019), which is as follows: 0, no visible symptoms; 1, one or two leaves mildly wilted; 2, more than two leaves mildly wilted, but less than two leaves severely wilted; 3, 25–50% of leaves severely wilted; 4, 50–75% of leaves severely wilted or dead; and 5, all leaves severely wilted or dead. All experiments were performed at least three times with 10 plants per treatment (n = 10). For stem inoculation, the stems of 3-week-old tomato plants were wounded above the cotyledons, the wounds were inoculated with 10 μl of C. michiganensis inoculum, and then the infected plants were transferred to a growth chamber for 3 weeks for canker development.

For the HR assay, leaves of 6-week-old N. tabacum, a nonhost plant, were used. Each C. michiganensis strain inoculum, adjusted to a concentration of OD₆₀₀ = 0.05, was infiltrated by needless syringe into 6-week-old N. tabacum plant leaves. HR development in the infiltrated leaves was observed for 48 h.

The ORFs of pat-1_Cm, chpC_Cm, chpE_Cm, chpF_Cm, and chpG_Cm genes without the stop codon were amplified from genomic DNA of C. michiganensis strain LMG7333 (Supplementary Table 1). The amplified DNA fragments were cloned into the pENTR/SD/D-TOPO vector (Invitrogen, CA, United States) according to the manufacturer’s instructions. After cloning, a gene fragment encoding a signal peptide of tobacco PR1b (GenBank accession no. X03465.1) was inserted into their N-terminal sites to generate in-frame fusion proteins and was verified by DNA sequencing. The resulting entry clones were recombined with the Gateway pGWB417 destination vector by LR reactions (Nakagawa et al., 2007). All cloned genes were also fused with c-Myc tag in their C-termini and expressed under the control of the 35S promoter. Recombinant plasmid constructs were transformed into Agrobacterium tumefaciens strain GV3101 for further assay.

For transient expression assay in N. tabacum leaves, A. tumefaciens strains with target genes were grown overnight in YEP medium (10 g of yeast extract, 1 0 g of peptone, and 5 g of NaCl per liter) with rifampicin (50 μg ml⁻¹) and spectinomycin (50 μg ml⁻¹) at 26°C. Agrobacterium cells were collected by centrifugation, washed with infiltration buffer (10 mM MES and 10 mM MgCl₂), and resuspended in the same buffer. Bacterial suspension with 100 μM acetosyringone was incubated at 25°C for more than 3 h and centrifuged again. After adding the infiltration buffer with 100 μM acetosyringone, the bacterial suspension was adjusted to an OD₆₀₀ of 0.4. Leaves of 6-week-old N. tabacum plants were infiltrated with an Agrobacterium suspension using a needleless syringe, and the plants were kept in the light to dry the leaves and subsequently incubated at 25°C.

To measure electrolyte leakage to quantify the degree of HR development, a total of six leaf disks (9.2 mm diameter) were harvested from infiltrated leaves at 0 and 24 hai. Detached leaf disks were washed with 20 ml of deionized water for 30 min and then placed in a new tube containing 20 ml of deionized water and shaken at 160 rpm for 2 h. Ion conductivity was measured at the indicated time points with a conductometer (CON6 portable conductivity meter; Oakton, IL, United States).

The pat-1_Cm gene and its variants, including an SP-deleted mutant and six alanine-substituted mutants with 0.4-kb native promoter region and a FLAG tag on their C-termini were amplified from C. michiganensis strain LMG7333 by PCR and cloned into the pTOP blunt V2 vector (Enzynomics, Daejeon, Korea), named pTOP-pat-1_Cm. Moreover, to generate Tn::pat-1_Cm strains in which pat-1_Cm orthologous genes present in C. capsici and C. sepedonicus were expressed under the native pat-1_Cm gene promoter region, pat-1_Cc was amplified from C. capsici strain PF008 by PCR. The pat-1_Cs and the paralogous chp-7_Cs gene from C. sepedonicus were obtained by custom gene synthesis with the promoter region of the pat-1_Cm gene (Bioneer, Daejeon, Korea). The DNA fragments of all pat-1 genes digested by SpeI/HindIII restriction enzymes were ligated into the linearized pK2-22 vector with the same enzymes.

To generate Pat-1_Cm variants with alanine substitutions at specific amino acid residues, amino acid substitutions of pat-1_Cm were performed by site-directed mutagenesis. Using the construct pTOP-pat-1_Cm as a template, PCR was performed with two mutagenic primers and then treated with DpnI to digest the methylated template. The final product was transformed into Escherichia coli strain DH5α and then confirmed by PCR and DNA sequencing. These variant genes were cloned into the pK2-22 vector.

Bacterial complementation using the pK2-22 vector was performed as previously described (Hwang et al., 2018). Briefly, the mutant Tn::pat-1_Cm was transformed with each gene construct, in which a FLAG tag was fused at the C-terminus, into the pK2-22 vector. After transformation by electroporation with up to 4–5 μg of plasmid DNA, the transformants were grown over 3 days at 26°C, and then the single transformed bacterial colony was selected with neomycin (50 μg ml⁻¹) as a selectable marker.

Bacterial strains containing C-terminal FLAG-tagged gene constructs were grown in half-strength KB media with 0.4% CMC. Incubated bacterial cells were harvested by centrifugation, and its supernatant was precipitated with 10% w/v of chilled trichloroacetic acid (TCA)/acetone at a ratio of 4:1 after filtration with a 0.22 μm pore size sterile filter. Proteins collected in the supernatant by centrifugation were washed twice with ice-cold acetone and resuspended in 0.1 ml distilled water. Proteins from the cell pellets were lysed at 25°C for 20 min using B-PER™ Bacterial Protein Extraction Reagent (Thermo Scientific, Rockford, IL, United States) supplemented with 100 μg ml⁻¹ lysozyme. After centrifugation, the supernatant extracted from the cell pellet was used for western blot analysis.

For total protein extraction from N. tabacum leaves, inoculated leaves were harvested at 36 hai and ground into a fine powder in liquid nitrogen. Samples were homogenized in protein extraction buffer [10% glycerol, 150 mM Tris–HCl (pH 7.5), 1 mM EDTA, 150 mM NaCl, 5 mM dithiothreitol, protease inhibitor cocktail (Sigma, St. Louis, MO, United States), and 0.2% Triton X-100 (Sigma)]. The homogenates were centrifuged at 12,000 g for 20 min at 4°C, and supernatants were collected for further study.

For western blot analysis, protein samples in sodium dodecyl sulfate (SDS) sample buffer [100 ml of 1.5 M Tris (pH 6.8), 60 ml of 20% SDS, 300 ml of glycerol, 150 ml of β-mercaptoethanol, and 18 mg of bromophenol blue per liter] were denatured by boiling and separated in a 12% SDS-polyacrylamide gel at 100 V for 2–3 h. After electrophoresis, separated proteins were transferred onto a polyvinylidene difluoride membrane (Millipore, Burlington, MA, United States). Subsequently, blots were blocked for 1 h with 5% w/v nonfat milk in phosphate-buffered saline (PBS) with 0.1% w/v Tween 20 and immunoblotted with horseradish peroxidase-conjugated anti-FLAG antibody (Sigma-Aldrich, St. Louis, MO, United States) or anti-c-Myc antibody (Santa Cruz Biotechnology, Santa Cruz, CA, United States), followed by anti-mouse IgG-HRP secondary antibody (Santa Cruz Biotechnology). Proteins were detected by the HRP activity using enhanced chemiluminescence plus western blotting detection reagent (GE Healthcare, Little Chalfont, United Kingdom).

For 2D gel electrophoresis, proteins were precipitated from cell-free supernatant of C. michiganensis type strain LMG7333 by 10% TCA-acetone precipitation. Isoelectric focusing was performed using 24 cm immobilized pH gradient strips with a non-linear pH 4–10 gradient. 2D gel electrophoresis and MALDI-TOF MS analysis were conducted at GenoMine (Pohang, Gyeongbuk, Korea).

A tertiary structure of the Pat-1_Cm protein was predicted using phyre2¹ based on the principles of homology-based modeling (Kelley et al., 2015).

To obtain a large amount of Pat-1_Cm proteins tagged with C-terminal FLAG, C. michiganensis Tn::pat-1_Cm expressing pat-1_cm or its variants was cultured in 1/5 strength KB media with 0.2% CMC for 2 days at 26°C with constant shaking. The large-scale cultures (500 ml) were centrifuged to collect the supernatant, and then the supernatant was filtered through a 0.22 μm sterile filter (Sartorius Stedim Biotech GmbH, Göttingen, Germany). To immune-precipitate Pat-1_Cm protein, 300 μl of washed Pierce Anti-DYKDDDDK Magnetic Agarose (A36797; Thermo Scientific) was added to 50 ml of filtered supernatant and incubated with rotation for 1 h. The tube with the anti-FLAG magnetic beads/protein was placed on a magnetic stand, and the supernatant was removed. This procedure was repeated several times to collect enough protein for the experiment. After removing the supernatant, bound beads were washed twice with PBS (pH 7.4) and once with distilled water. Washed beads were eluted with Pierce IgG Elution buffer (pH 2.8, 21004; Thermo Scientific) or SDS-PAGE sample buffer.

To determine the N-terminal amino acids of mature and secreted Pat-1_Cm, purified proteins extracted from cell-free supernatant were separated by 1D SDS-PAGE and blotted onto PVDF (Immobilon-P membrane, Merck Millipore, MA, United States). After electro-blotting, the membrane was stained with Coomassie Blue R-250 for 30 min, de-stained multiple times with de-staining solution (10% acetone and 45% methanol–water per liter), and then rinsed with distilled water to remove the high concentration of other buffers, including transfer buffers. This membrane was dried at 25°C, and the targeted protein bands were excised with a blade. For N-terminal sequencing, the cut membrane was subjected to automated Edman degradation using a Procise 492 Protein Sequencing System (Applied Biosystems, CA, United States) at PROTEINWORKS (Daejeon, Korea).

Statistical analysis of disease severity data was done by applying a non-parametric Kruskal-Wallis test with Dunnett’s multiple comparisons (p < 0.01) using the software statistiXL version 2.0 (statistiXL, Broadway, Australia). Duncan’s multiple range test was performed to analyze other results for comparisons between independent groups (p < 0.05).

Previously, the curing of plasmid pCM2 and complementation analysis with the pat-1_Cm gene in pCM2 showed that the pat-1_Cm gene is critical for wilting caused by C. michiganensis strain NCPPB382 in tomato (Meletzus et al., 1993; Dreier et al., 1997). To further study the role of the pat-1_Cm gene as a pathogenicity factor, we screened a C. michiganensis type strain LMG7333 mutant library generated by transposon (Tn1409Cβ) insertion (Hwang et al., 2019) and selected the LMG733-Tn::pat-1_Cm mutant strain (hereafter Tn::pat-1_Cm strain; Supplementary Figure 1). The transposon’s position was determined by whole genome sequencing using Illumina MiSeq. The transposon was located at a distance of 542 bp from the translation start site (ATG) of pat-1_Cm (Supplementary Figure 1A), and a single insertion was confirmed by Southern blot analysis (Supplementary Figure 1B). Simultaneously, we generated the pCM2-curred strain (7333ΔpCM2) by plasmid curing. Transposon insertion and pCM2 absence were confirmed by PCR with primer pairs targeting pat-1_Cm and the plasmid backbone gene within pCM2 (Supplementary Figure 1C).

We next determined the pathogenicity of the Tn::pat-1_Cm strain and LMG7333ΔpCM2 for wilting symptoms in 2-week-old tomato plants via the root-dipping inoculation method and for bacterial canker symptom in 3-week-old tomato plants via the stem-inoculation method. The wild-type (WT) LMG7333 strain caused severe wilting that led to the death of inoculated plants (Supplementary Figure 1D). However, tomato plants inoculated with either Tn::pat-1_Cm strain or 7333ΔpCM2 strain exhibited only very mild or no wilting symptoms (Supplementary Figure 1D), although overall plant growth was slightly reduced after infection with both mutant strains. Interestingly, the C. michiganensis WT and its two mutants caused a similar degree of canker symptoms around the inoculation sites (Supplementary Figure 1E). When the stems of the WT-inoculated plants showing canker symptoms were cut lengthwise and examined, the brown discoloration in the vascular tissues and the tissue collapse typical of severe wilting symptoms were observed (Supplementary Figure 2). In contrast, the decrease in discoloration of vascular tissues was only observed in the absence of tissue collapse and of wilting symptoms in plants inoculated with the Tn::pat-1_Cm strain (Supplementary Figure 2).

The mutant strains were complemented by transformation with the intact pat-1_Cm gene controlled by its native promoter and fused with a FLAG tag at its 3′-terminus, named Tn::pat-1_Cm (pat-1_Cm) and 7333ΔpCM2 (pat-1_Cm). The ability to cause wilting in tomato as much as the WT was restored in both complemented strains (Figure 1; Supplementary Figure 3), indicating that the pat-1_Cm gene is a critical factor for wilting in the host plant, tomato. Nevertheless, all strains, including mutants and the complementary strains, were consistently able to cause bacterial canker on plant stems (Supplementary Figures 1E, 2). Overall, these results indicate that pat-1_Cm in the plasmid pCM2 is critical for C. michiganensis to cause wilting, but not canker, in tomato.

The pat-1_Cm gene of the C. michiganensis type strain LMG7333 encodes a protein consisting of 280 amino acids that harbors a putative SP in the N-terminal region (Supplementary Figure 4A). Additional protein comparison revealed that Pat-1 proteins from various Clavibacter species contain a conserved protein sequence, including three amino acids for a catalytic triad of putative serine proteases and two cysteines. This holds true even with the Pat-1_Cs protein from C. sepedonicus, which lacks an SP. Since Pat-1 was predicted to have an SP in its N-terminal region, we hypothesized that Pat-1_Cm is a secreted protein with a functional SP. To ensure the importance of Pat-1_Cm secretion for C. michiganensis pathogenicity, the first 33 amino acids, predicted as an SP, were removed, and the resulting truncated form, pat-1_CmΔSP was tagged with FLAG at its C-terminus and transformed into the Tn::pat-1_Cm strain. This complementary strain did not cause wilting in tomato (Figures 1A,B). Immunoblot with a polyclonal FLAG antibody detected the FLAG-tagged full-length of Pat-1_Cm from both the supernatant and pellet after growth of both complemented strains in King’s B (KB) medium with 0.4% carboxymethyl cellulose (CMC) as a substrate (Figure 2A). However, the Pat-1_CmΔSP in the Tn::pat-1_Cm strain was not detected in either fraction. These results indicate that Pat-1_Cm protein is expressed in the complemented strains and that the predicted SP is required for secretion and might be critical for protein stability.

To reconfirm that Pat-1_Cm proteins are secreted, C. michiganensis strain LMG7333 was first grown in KB medium with 0.4% CMC and pelleted, and then the cell-free supernatant was collected. The total protein content of the cell-free supernatant was separated by 2D gel electrophoresis, and distinct protein spots were analyzed by MALDI TOF MS/MS. As a result, Pat-1_Cm protein was detected (Figure 2A) and possessed a 100% match to five peptide sequences (Figure 2B).

To determine the SP cleavage site of this protein, we analyzed the N-terminal sequences of mature and secreted Pat-1_Cm using Edman degradation. The first five amino acid residues (V-D-R-I-A) were obtained by N-terminal sequencing (Figure 2C; Supplementary Figure 5). These results indicate that Pat-1_Cm is a genuine secreted protein with the functional SP consisting of the first 33 amino acids.

Previously, it was shown that Clavibacter species induced HR in nonhost plant species, and that the chp-7_Cs gene, a homolog of pat-1_Cm that encodes a putative serine protease in C. sepedonicus, is involved in HR induction in N. tabacum (Nissinen et al., 2009). To determine if Pat-1_Cm plays a role in C. michiganensis LMG7333 HR induction in a nonhost plant, we performed the HR assay in N. tabacum (cv. Samsun) plants. Infiltration of WT C. michiganensis into N. tabacum leaves induced a typical HR within 24 h after infiltration (hai); however, an HR was not induced by the mutant strains, Tn::pat-1_Cm (Figure 3A) and 7333ΔpCM2 (Supplementary Figure 6). Moreover, the HR-eliciting ability of both mutants was restored by complementation with an intact pat-1_Cm gene. To quantify C. michiganensis-triggered HR, we measured electrolyte leakage during HR induction in tobacco leaves. Compared to the WT, electrolyte leakage after infiltration with the Tn::pat-1_Cm strain was significantly and consistently reduced. Furthermore, complementing with the intact pat-1_Cm restored electrolyte leakage in tobacco leaves to normal levels (Figure 3B). Our finding that the expression of pat-1_Cm is directly associated with HR development in a nonhost plant suggests that the pat-1_Cm gene in C. michiganensis is required not only for wilting development in a host plant, but also for HR induction in a nonhost plant.

The pat-1_Cm orthologs were found in several species, including C. michiganensis, C. capsici, and C. sepedonicus. Those genes, pat-1_Cc, pat-1_Cs, and chp-7_Cs, are conserved in their plasmids with amino acid sequences that are approximately 77% identical to pat-1_Cm (Supplementary Figure 4A). The chp-7_Cs, which is a chromosomal putative serine protease of C. sepedonicus, was also classified as a pat-1_Cm ortholog. A functional SP was predicted in Pat-1_Cc and Chp-7_Cs, but not in Pat-1_Cs (Supplementary Figure 4A).

Subsequently, we determined if the pat-1_Cm orthologs confer virulence of by complementing C. michiganensis Tn::pat-1_Cm with each of pat-1_Cc, pat-1_Cs, and chp-7_Cs genes under control of a 0.4-kb native promoter of pat-1_Cm (Supplementary Figure 4B). The Tn::pat-1_Cm strain with chp-7_Cs caused wilting in tomato, but disease severity was significantly less than with the WT C. michiganensis (Figures 4A,B). The amount of expressed and secreted Chp-7_Cs proteins in Tn::pat-1_Cm was significantly lower than that of Pat-1_Cm (Figure 4C). In contrast, the strains with either pat-1_Cc or pat-1_Cs failed to recover virulence activity (Figures 4A,B). When their expression and secretion levels were examined, Pat-1_Cc was expressed and secreted, but less than Chp-7_Cs, and was only detected after immunoprecipitation with FLAG affinity beads (Figure 4C). As expected, due to the lack of the N-terminal SP on Pat-1_Cs, it was not detected. Next, the HR induction was examined with complemented strains. Complementation with pat-1_Cm induced a strong HR similar to the WT, but none of the other three complemented strains could induce HR in N. tabacum leaves (Figure 4D). These results indicate that the virulence of only chp-7_Cs is conserved with pat-1_Cm and that the amount of secreted proteins might be directly correlated with disease severity.

In the chromosome of C. michiganensis strain LMG7333, there are seven genes (chpA_Cm to chpG_Cm) known as chromosomal homology to pat-1 (chp; Stork et al., 2008). Three (chpA_Cm, chpB_Cm, and chpD_Cm) are pseudogenes containing internal stop codons, whereas the remaining four genes encode mature forms of putative serine proteases. These latter four proteins have a putative SP sequence and were functionally classified as a serine protease family. However, protein sequence identities between these four proteins and Pat-1_Cm were lower than 40% (Supplementary Figures 7A,B). To determine if any of these chp genes could functionally complement pat-1_Cm, full length (FL) chpC_Cm, chpE_Cm, chpG_Cm, and chpF_Cm genes tagged with FLAG on their C-termini were transformed into both Tn::pat-1_Cm and 7333ΔpCM2, and then strains’ ability to cause wilting in tomato and to elicit HR in N. tabacum was examined. Intriguingly, none of the four chp genes enabled either mutant to cause wilting in tomato (Figures 5A,B; Supplementary Figures 8A,B) or to elicit HR in N. tabacum (Figure 5C; Supplementary Figure 8C), even though they were expressed and secreted similar to Pat-1_Cm proteins (Figure 5D). These results indicate that the manner in which Pat-1_Cm interacts with plants is distinct from other homologous Chp proteases.

Pat-1_Cm is considered a putative serine protease due to its C-terminal conserved motif, Gly-ASP-Ser-Gly-Gly (GDSGG; Krem et al., 1999; Burger et al., 2005; Nissinen et al., 2009). Here, Pat-1_Cm protein structures were predicted using the Phyre2 program to determine the key functional amino acid residues. Protein modeling showed that Pat-1_Cm protein contains three residues in the following order histidine (His, H96), aspartate (Asp, D122), and serine (Ser, S231). This sequence is consistent with the catalytic triad of well-characterized serine proteases (Figure 6A; Supplementary Figure 9). The Ser/His/Asp catalytic triad is highly conserved in all Pat-1 proteins from Clavibacter species (Supplementary Figure 4A) and the Chp protein family in the chromosome of LMG7333 (Supplementary Figure 7A). In addition, protein modeling showed that the Pat-1_Cm protein has two cysteine residues at positions 189 and 199 that are connected by disulfide bond to form a proper tertiary structure (Supplementary Figure 9). Similar to the catalytic triad, two cysteine residues are found in Pat-1 proteins and the Chp protein family except for ChpC (Supplementary Figures 4A, 7A).

To investigate if the five residues in the catalytic triad (H96, D122, and S231) and the disulfide bond (C189, C199) play crucial roles in the function of Pat-1_Cm, the individual residues were substituted with alanine by site-directed mutagenesis (Figure 6A) and introduced into the mutant strain, Tn:: pat-1_Cm. Additionally, the serine residue at position 223 (Ser, S223) located in the un-conserved region was substituted with alanine as an internal control to rule out the possibility that alanine substitution itself affects the protein structure. All alanine substituted proteins were expressed and secreted similar to the WT Pat-1_Cm, and the S223A substituent was expressed the least (Figure 6B). The Tn:: pat-1_Cm mutant strain carrying all alanine substitutions except for at S223A failed to restore the mutant’s abilities to cause wilting in tomato (Figures 6C,D) or to elicit HR in N. tabacum (Figure 6E). These results indicate that the catalytic triad for putative serine protease and the specific disulfide bond for proper tertiary structure formation is essential for Pat-1_Cm functioning in plants.

We showed that the mutant Tn::pat-1_Cm lost its ability to elicit HR in a nonhost plant (Figure 3). To determine whether the Pat-1_Cm protein alone elicits HR, the SP of tobacco PR1a was fused to the N-terminus of full-length Pat-1_Cm to mimic Pat-1_Cm secretion and translocation to the apoplast of N. tabacum leaves. In addition, a c-Myc tag was fused to the C-terminus of Pat-1_Cm to check protein expression. The final construct, 35S-SP::Pat-1_Cm under control of the 35S promoter, was transformed into Agrobacterium bacterium. Agrobacterium-mediated transient expression of Pat-1_Cm in N. tabacum leaves, but not in a host plant tomato, elicited HR within 48 h, whereas the empty vector (EV) did not (Figure 7A; Supplementary Figure 10). Since the catalytic triad and disulfide bond are important for Pat-1_Cm functioning, we generated two more constructs carrying Pat-1_Cm–C189A and Pat-1_Cm–S231A and expressed in N. tabacum leaves. These two variants failed to elicit HR, unlike the WT Pat-1_Cm (Figure 7A), although they were expressed at the same level as the WT Pat-1_Cm (Figure 7B). These results indicate that the Pat-1_Cm protein alone, when located in plant apoplast, can elicit HR in a nonhost plant. In addition, the results suggest that both the catalytic triad for putative serine protease and a disulfide bond dependent structure in Pat-1_Cm are required for HR induction in a nonhost plant.

Although four intact Chp proteins in C. michiganensis, ChpC_Cm, ChpE_Cm, ChpF_Cm, and ChpG_Cm, could not complement the Tn::pat-1_Cm mutant to restore its HR-eliciting ability (Figure 5C), we checked whether any of these four Chp proteins alone could elicit HR when ectopically expressed on N. tabacum leaves. An HR was induced by expression of ChpG_Cm, but not ChpC_Cm, ChpE_Cm, or ChpF_Cm (Figure 7C). All four proteins were expressed at higher levels than Pat-1_Cm (Figure 7D). These results indicate that, although ChpG_Cm protein alone can elicit HR, it cannot substitute the ability of Pat-1_Cm in C. michiganensis to elicit HR in N. tabacum.

Previously, we reported that celA, another plasmid-borne pathogenicity gene in C. michiganensis that encodes a secreted cellulase, is critical for the development of wilting in tomato (Hwang et al., 2019). Because both Pat-1_Cm and CelA encode secreted proteins with different enzymatic activities, we hypothesized that both proteins originated from different mutant bacteria present in the same site can functionally cooperate to cause wilting in tomato. To test this hypothesis, mutants Tn::celA and Tn::pat-1_Cm were inoculated individually or co-inoculated on 2-week-old tomato plants by the root-dipping inoculation method. The individually inoculated plants did not show wilting, whereas the co-inoculated plants showed severe wilting, similar to WT-inoculated plants (Figures 8A,B). Moreover, cellulase activity similar to that of the WT strain was observed in Tn::pat-1_Cm, whereas Tn::celA lost its ability to produce cellulase (Figure 8C). These results indicate that two different but important pathogenicity proteins, CelA and Pat-1_Cm, function cooperatively to cause wilting in a host plant.