Single gene enables plant pathogenic Pectobacterium to overcome host‐specific chemical defence

Abstract Plants of the Brassicales order, including Arabidopsis and many common vegetables, produce toxic isothiocyanates to defend themselves against pathogens. Despite this defence, plant pathogenic microorganisms like Pectobacterium cause large yield losses in fields and during storage of crops. The bacterial gene saxA was previously found to encode isothiocyanate hydrolase that degrades isothiocyanates in vitro. Here we demonstrate in planta that saxA is a virulence factor that can overcome the chemical defence system of Brassicales plants. Analysis of the distribution of saxA genes in Pectobacterium suggests that saxA from three different phylogenetic origins are present within this genus. Deletion of saxA genes representing two of the most common classes from P. odoriferum and P. versatile resulted in significantly reduced virulence on Arabidopsis thaliana and Brassica oleracea. Furthermore, expressing saxA from a plasmid in a potato‐specific P. parmentieri strain that does not naturally harbour this gene significantly increased the ability of the strain to macerate Arabidopsis. These findings suggest that a single gene may have a significant role in defining the host range of a plant pathogen.

distribution of single genes or gene clusters suggests that some virulence factors only contribute to virulence depending on the context of the infection (Li et al., 2018).
Certain plant secondary metabolites are characteristic for specific groups of plants and are mostly known for their antiherbivorous properties. Much less is known about how secondary metabolites affect plant pathogenic bacteria and about ways in which these pathogens may overcome the host-specific defences. It was only recently shown that caffeine has antibacterial properties against multiple plant pathogenic bacteria, including Pectobacterium, Pseudomonas, and Dickeya (Sledz et al., 2015). Bacterial resistance to such plant antimicrobials can be considered a context-dependent virulence factor. One example of a context-dependent virulence factor in Pectobacterium is tolC, a gene that decreases the susceptibility of Pectobacterium brasiliense PCC21 to the antibacterial plant chemicals berberine, rhein, and genistein (Lee et al., 2013).
Most research about Pectobacterium and Dickeya has focused on potato pathogenesis as it is a staple food with high economic relevance (Motyka et al., 2017). Certain Pectobacterium spp., however, have a wide host range or are specific pathogens to other crops, which is also reflected in the high genomic variability of these bacteria (Li et al., 2018;Zoledowska et al., 2018). Here we approached the identification of a novel virulence factor bottom-up by directly targeting a gene with a recently elucidated function that was hypothesized to aid specifically in the colonization of plants that produce the secondary metabolite isothiocyanate (ITC). Among the host plants of pectobacteria, those of the Brassicaceae family are some of the most produced vegetables worldwide (FAO, 2012). Plants of this family use a unique form of chemical defence to protect against herbivory and disease. Mustard oil glycosides (or glucosinolates, GSN) and the enzyme myrosinase are stored separately within plant compartments and toxic ITCs are produced when these two come into contact after tissue damage by herbivory or lesion caused during harvesting (Zhu et al., 2013). Coincidentally, soft rot commonly establishes in crops after initial tissue damage.
Phytopathogenic bacteria have been shown to be particularly susceptible to allylisothiocyanate (A-ITC), benzylisothiocyanate (B-ITC), 2-phenylethylisothiocyanate (2PE-ITC), and sulforaphane (SFN) in vitro (Aires et al., 2009). In nature, the composition of different ITCs that are produced in a plant varies greatly by host species, age, and a variety of other factors (Van Dam et al., 2009). The model plant Arabidopsis thaliana, also a member of the Brassicaceae family, produces a range of mostly aliphatic ITCs (Hogge et al., 1988). Insects feeding on mutant Arabidopsis with abnormally low levels of GSN grow to larger sizes and do more damage to the leaves than those feeding on Arabidopsis with natural GSN concentrations (Beekwilder et al., 2008). A 2001 study determined that Pectobacterium carotovorum (then Erwinia carotovora) and Pseudomonas syringae were equally infectious to Arabidopsis gsm1-1 mutants that contain fewer GSNs as they were to wild-type plants (Tierens et al., 2001). Only the fungus Fusarium oxysporum was found to be significantly more aggressive on A. thaliana gsm1-1 than on wild-type plants.
Recently we found that the bacterial gene saxA encodes ITC hydrolase (ITCase) that is capable of hydrolysing ITCs in vitro (van den Bosch et al., 2018). The protein catalytic site activates one water molecule to subsequently hydrolyse the ITC, resulting in the release of the ITC's corresponding amine and a molecule of carbonyl sulfoxide (Equation 1).
The saxA gene is widely distributed among the Enterobacterales and the saxA gene family could be categorized into five phylogenetic clusters. The saxA gene was originally identified in plant pathogenic P. syringae pv. tomato DC3000, where it was shown to be one of six genes increasing survival on Arabidopsis extracts (Fan et al., 2011).
The lifestyle of plant pathogenic P. syringae differs markedly from that of soft rot pectobacteria in that it does not rely on tissue maceration in order to utilize nutrients from dead host cells (Davidsson et al., 2013). It is currently unknown whether ITCs are involved in defence against the latent phase of phytopathogenic infections, but it is clear that tissue damage, which the PCWDE arsenal of pectobacteria is known to cause, activates the production of ITCs.
In this study we investigated the effect of deleting the saxA genes in two Pectobacterium soft rot strains (P. odoriferum NCPPB3841 and P. versatile SCC1) on their ability to infect Arabidopsis. A third strain, P. polaris NCPPB3395, also containing a genomic saxA gene, could not infect Arabidopsis leaves and was therefore tested in medium with Arabidopsis leaf extracts. A fourth Pectobacterium strain, P. parmentieri SCC3193, is a known potato pathogen that naturally lacks a genomic saxA gene and we assessed how its virulence was affected by appending it with a saxA gene. Based on the known function of SaxA in vitro, we hypothesized that deleting the saxA gene in Pectobacterium strains results in reduced virulence on plants that use the GSN-myrosinase system as a defence, whereas adding it to a strain that naturally lacks saxA would achieve the opposite. We also compared the distribution of the saxA gene in most known Pectobacterium whole genomes to see whether presence of the saxA gene is restricted to strains isolated from hydrolase that may mitigate the plant defence of Brassica plants (van den Bosch et al., 2018). In order to assess whether saxA was required for infection in planta and would therefore be a new virulence factor, we deleted the saxA gene in the genomes of P. odoriferum (Po) NCPPB3841 and P. versatile (Pv) SCC1. The saxA genes of Po NCPPB3841 and Pv SCC1 are phylogenetically distinct (44.9% amino acid identity) and cluster in different subfamilies (CL2 and CL5, respectively, Figure 1).
In order to assess the virulence, A. thaliana Col-0 plant leaves were infected with Po NCPPB3841 wild-type and saxA deletion mutant (ΔsaxA) (Figure 2a,b) or with Pv SCC1 wild-type and ΔsaxA ( Figure 2d,e), and incubated for 24 hr before the area of maceration was measured. The macerated area of plants infected with ΔsaxA was significantly smaller than the area of maceration in plants infected with the wild-type (p Po = 3.0e−06, 7.4e−07, 7.4e−07, and p Pv = 7.4e−07, 7.4e−07, 7.4e-07, respectively, for three independent experiments, Wilcoxon rank sum test). In order to verify that the reduction of virulence in the mutant is indeed due to the loss of saxA, a complementation of the ΔsaxA strains was performed (Figure 2b,e).
Transforming Po NCPPB3841 ΔsaxA with a plasmid containing the deleted gene (pMW119-NCPPB3841_saxA) resulted in significantly more maceration than transformation with the empty plasmid (p = 3.0e−06, 7.4−07, and 7.4e−07, respectively, for three independent experiments, Wilcoxon rank sum test). Furthermore, a plasmid containing the saxA gene from Pv SCC1 was also able to restore the virulence of Po NCPPB3841 ΔsaxA (p = 1.0e−04, 7.4e−07, and 7.4e−07, respectively, for three independent experiments, Wilcoxon rank sum test). Similarly, complementing Pv SCC1 ΔsaxA using either pMW119 with the saxA gene from SCC1 (p = 7.4e−07 for all three independent experiments, Wilcoxon rank sum test) or NCPPB3841 (p = 7.4e−07 for all three independent experiments, Wilcoxon rank sum test) restored the virulence phenotype ( Figure 2e). In order to show that the effect of saxA on virulence is dependent on plantproduced ITCs, A. thaliana myb28 myb29 was used, which is impaired in aliphatic GSN production and therefore does not release the derived ITCs (Sønderby et al., 2007). Figure 2c,f shows that F I G U R E 1 Phylogenetic tree adapted from Li et al. (2018) with arbitrary branch lengths. The presence of a saxA gene in the strain genome is overlaid in branch colours. The colour of the strain identifier corresponds with isolation sources from either potato (Solanum tuberosum) or Brassica vegetables F I G U R E 2 Isothiocyanate hydrolase SaxA is required for full virulence of Pectobacterium on Arabidopsis thaliana. Arabidopsis leaves were inoculated locally with approximately 10 4 cfu and maceration symptoms were documented 24 hr after inoculation. Representative images of symptoms on Arabidopsis Col-0 caused by P. odoriferum NCPPB3841 (a) or P. versatile SCC1 (d). Quantification of Arabidopsis Col-0 leaf maceration caused by P. odoriferum NCPPB3841 (b) or P. versatile SCC1 (e), the respective saxA mutants, and the saxA mutants complemented with two saxA genes of different origins. Quantification of Arabidopsis Col-0 leaf maceration caused by P. odoriferum NCPPB3841 (c) or P. versatile SCC1 (f) and the respective saxA mutants on wild-type and myb28 myb29 double mutants with impaired aliphatic glucosinolate accumulation. Three leaves per plant were inoculated and one dot represents the combined area of maceration per plant. Different symbols represent experimental replicates on different dates. For each of the three experiments, N = 12 in (b) and (e), and N = 10 in (c) and (f). Each individual experiment was analysed separately and asterisks indicating statistically significant differences according to the Wilcoxon rank sum test are shown for the largest p value obtained (in black, ***p < .001 in all three experiments, ns: p > .05 in all three experiments) or for each experiment separately in cases where differing results were obtained (in grey, circles/squares/triangles; **p < .01, *p < .05, ns: p > .05)  Figure S2d) led us to conclude that P. polaris NCPPB3395 is not fully virulent on these three hosts and thus a possible in planta effect of saxA could not be tested in these systems. In order to better assess the role of saxA genes in P. polaris NCPPB3395 we conducted growth experiments of P. polaris NCPPB3395 wild type and three different deletion mutants with medium containing A. thaliana extract ( Figure 4a). The single deletion mutants P. polaris ΔsaxA1 and P. polaris ΔsaxA2 did not show growth inhibition on medium containing A. thaliana Col-0 extract as compared to the wild-type whereas P. polaris ΔsaxA1 + ΔsaxA2 showed severely impaired growth. Medium containing myb28 myb29 extract devoid of aliphatic GSN/ITCs enabled all deletion strains to grow equally well as the wild-type. This suggests that the two copies of saxA in P. polaris are either functionally redundant in detoxifying a blend of naturally occurring ITCs in Arabidopsis or that the differential function of saxA genes within different phylogenetic clusters remains to be elucidated.

| In trans expression of saxA enables potato pathogen P. parmentieri SCC3193 to cause maceration on Arabidopsis
P. parmentieri SCC3193 was isolated from diseased potato stem and does not naturally contain a saxA gene in its genome. A. thaliana Col-0 extract has been shown to be toxic to SCC3193 in a F I G U R E 3 Isothiocyanate hydrolase SaxA is required for full virulence of Pectobacterium on cabbage. Brassica oleracea var. capitata leaves were inoculated locally with approximately 5 × 10 4 cfu of P. odoriferum NCPPB3841 or P. versatile SCC1 and the respective saxA mutants, and maceration symptoms were documented 48 hr after inoculation. Representative images of symptoms (a) and quantification of maceration (b). One dot represents one cabbage leaf disc. Different symbols represent experimental replicates on different dates. N = 12 for each experiment. Each individual experiment was analysed separately and asterisks indicating statistically significant differences according to Wilcoxon rank sum test are shown for the largest p value obtained (***p < .001 in all three experiments)

(B) (A)
myrosinase-dependent manner (Brader et al., 2001) and SCC3193 does not readily macerate Col-0 leaves. We observed that even an inoculum 100 times the one used with Po NCPPB3841 and Pv SCC1 rarely resulted in initiation of maceration. Ten out of 30 plants showed maceration on one out of five inoculated leaves after 24 hr. Furthermore, the maceration caused by wild-type SCC3193 never seemed to spread significantly ( Figure S1). As a result of our previous experiments, we wondered whether transforming this strain with the previously constructed saxA-containing vector would impart it with the ability to tolerate ITC-rich Col-0 extract and macerate Col-0 leaves. In fact, the P. parmentieri SCC3193 with the saxA gene sourced from the genome of P. versatile SCC1 was able to grow on Col-0 extract, unlike the wild-type strain ( Figure 4b).
Furthermore, it caused maceration in all five inoculated leaves in 12 out of 30 tested plants after 24 hr. Thirteen plants showed maceration in four out of five leaves, four plants showed maceration in three out of five leaves and one plant showed maceration in two out of five inoculated leaves. The maceration was often able to spread to cover large parts of the leaf (Figures 4c and S1). These results indicate that saxA could increase the virulence of some Pectobacterium strains on non-host brassica plants. However, only a high inoculum led to maceration even when SCC3193 was amended with saxA, suggesting that there are also other factors contributing to the poor ability of SCC3193 to macerate Col-0. This conclusion is consistent with the result of cabbage inoculations where neither wild-type SCC3193 nor SCC3193 with saxA were able to cause spreading disease symptoms ( Figure S2).

| Distribution of saxA virulence factor in the Pectobacterium genus
The saxA gene is widely distributed among Enterobacterales, and a phylogenetic analysis of a subset of enterobacterial saxA genes showed that the saxA gene family can be divided into five separate phylogenetic clusters, numbered CL1 to CL5 (van den Bosch et al., 2018). We analysed the distribution of saxA genes in 84 Pectobacterium and three Dickeya genomes that were previously analysed by Li et al. (2018) regarding virulence factors. The results of the BLAST analysis of the remaining 10 strains are discussed separately as the exact phylogeny is currently insufficiently supported.
saxA from CL5 is widely distributed in the Pectobacterium genus, as it is represented in most strains of P. brasiliense, P. versatile, P. polaris, and P. atrosepticum (Figure 1). Notable exceptions are P. polaris SS28 and a large branch of P. brasiliense that has lost this gene. saxA from CL2 seems to have been acquired only once in the Pectobacterium genus F I G U R E 4 Effect of saxA on Pectobacterium growth in the presence of plant-derived isothiocyanates. Growth of P. polaris NCPPB3395 and its saxA (double) deletion mutants (a) or P. parmentieri SCC3193 carrying an empty vector (SCC3193) or a vector containing the saxA gene of P. versatile SCC1 (SCC3193 + SCC1 saxA) (b) in M9 salt solution amended with Arabidopsis leaf extracts from Col-0 wild-type or myb28 myb29 plants measured as attenuation of light on a 96-well plate. One by a common ancestor of both P. odoriferum and P. carotovorum but not P. brasiliense. Five out of 97 genomes contained a saxA gene from both CL2 and CL5. These are distributed among the P. carotovorum, P. odoriferum, and P. polaris species. Pectobacterium saxA genes are restricted to phylogenetic clusters CL2 and CL5, with the exception of P. wasabiae. The P. wasabiae species is unique in that all three genomes (CP015750.1, JQHP00000000.1, and JQOH00000000.1), besides a CL2-and a CL5-related saxA gene, each contain a third copy of a saxA gene that is related to CL4. Neither of the two P. betavasculorum strains (NCPPB2795 and NCPPB2793) contains saxA. All four of the P. peruviense strains (IFB5232, IFB5229, A97-S13-F16, and A350-S18-N16) contain saxA from CL5, as well as P. punjabense

| D ISCUSS I ON
A range of virulence factors and conditions are required for a pathogenic bacterium to successfully colonize and cause disease in a host plant. Many factors are important for virulence irrespective of host-specific defence mechanisms. In this study we focused on the chemical plant defence mechanism of the Brassicales order.
We investigated whether a single enzyme that impairs this specific chemical host defence contributes to the virulence of several Pectobacterium spp. As a model for this, we chose A. thaliana.
Brassicales plants employ the GSN-myrosinase defence system which, on tissue damage, releases a blend of toxic volatiles, of which ITCs are particularly toxic to insects and bacteria (Wittstock et al., 2003;Dufour et al., 2015). We have shown previously that the bacterial saxA gene encodes a hydrolase that can detoxify ITCs in vitro and that the gene is widespread among Enterobacterales van den Bosch et al., 2018). In this study, we demon-  , suggesting that this method for overcoming host plant resistance could spread quickly in a microbial community. The data presented here may help us understand why Pectobacterium is capable of infecting a broad spectrum of hosts.
The trial experiments on important cabbage crops indicate that our findings can be extrapolated to real-life scenarios, therefore potentially aiding research to improve food security.

| Mutant generation
saxA deletion mutants of P. polaris NCPPB3395, P. odoriferum NCPPB3841, and P. versatile SCC1 were generated by λ Red recombination as described by Datsenko and Wanner (2000) with a protocol optimized for Pectobacterium (Nykyri et al., 2012). Overnight cultures were diluted 1:50 in fresh medium and grown to an optical density (OD 600 ) value of approximately 0.5 before being chilled on ice for at least 30 min. Cells were then collected from 2 ml of culture in microcentrifuge tubes by centrifugation at 7,740 g for 1 min at 4 °C, washed once with ice-cold sterile water and resuspended in 40 μl of ice-cold 10% glycerol. The cells were transformed in 0.2 cm cuvettes with 100 ng pKD46 plasmid by electroporation at 2.5 kV and then allowed a recovery period of 2 hr in super optimal broth with catabolite repression (SOC) medium (2% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl 2 , 10 mM MgSO 4 , 20 mM glucose) at 28 °C before being plated on LB plates containing 100 μg/ml ampicillin. In case of naturally ampicillin-resistant P. versatile SCC1, pKD46-tetA, where the tetA gene conferring resistance to tetracycline has been introduced to pKD46, was used instead of pKD46, and plating was done on LB plates containing 10 μg/ml tetracycline. Linear saxA-targeting cat cassettes providing resistance to chloramphenicol were constructed by Polymerase Chain Reaction (PCR) for each target gene by prepending the 50 bp up-and downstream regions of each gene (SCC1_2268, KU75_11850, KP17_16520, and KP17_11025) to the forward and reverse primers, respectively, targeting the pKD3 template plasmid (Table S1). Pectobacterium strains harbouring pKD46 or pKD46-tetA were grown overnight in LB medium supplemented with 100 μg/ml ampicillin or 10 μg/ml tetracycline, respectively. Cultures were diluted 1:50 in fresh medium containing 0.1 M l-arabinose and grown to an OD 600 value of approximately 0.5. Competent cells were prepared as described above and transformed with 200-300 ng of cassette fragment followed by a 2-hr recovery period in SOC medium at 28 °C. The cells were then plated on LB plates containing 15 μg/ml chloramphenicol and grown over one (P. odoriferum NCPPB3841) or two (P. polaris NCPPB3395 and P. versatile SCC1) nights. Colonies were streaked on new plates and correct recombination of the cassettes to the genomes was confirmed with two PCRs, one using a primer pair targeting genomic sequences flanking the cassette and the other using one flanking primer (Pc_SCC1_saxA_500up, Pco_3841_500up, Pcc_3395_A5_500up, or Pcc_3395_B2_500up) with the cat cassette targeting primer C2 (Table S1). The mutants were grown on LB plates without selection after which streak testing on LB plates containing 100 μg/m ampicillin or 10 μg/ml tetracycline was used to identify colonies where cells no longer harboured pKD46 or pKD46-tetA, respectively. In order to generate a saxA double-deletion mutant of P. polaris NCPPB3395, the cat cassette was excised from P. polaris NCPPB3395 ΔsaxA2 with the help of the Flp-FRT recombination system plasmid pFLP2 (Hoang et al., 1998). Then 200 ng of pFLP2 was transformed to competent cells followed by a 2-hr recovery period in SOC medium at 28 °C before plating on LB plates containing 100 μg/mL ampicillin and growth over two nights at 28 °C. Colonies were streaked on new plates and grown overnight. Streak testing on LB plates containing 15 μg/mL chloramphenicol was used to identify colonies where excision of the cat cassette had been successful. In addition, PCR using primers targeting genomic sequences flanking the insertion site of the cat cassette was used to confirm excision (Pcc_3395_B2_500up/ Pcc_3395_B2_300down, Table S1). Finally, pFLP2 was cured from the chloramphenicol sensitive cells by sucrose selection on modified LB plates devoid of sodium chloride but supplemented with 5% sucrose.
The P. polaris NCPPB3395 ΔsaxA2 without the cat cassette was subsequently used for another round of mutagenesis to delete saxA1. The cat cassette was also excised from P. versatile SCC1 ΔsaxA as described for P. polaris NCPPB3395 except that after pFLP2 transformation LB plates containing 600 μg/ml ampicillin were used due to the strain's natural resistance to ampicillin. Excision of the cat cassette from P. odoriferum NCPPB3841 ΔsaxA was attempted but did not succeed.

| Complementation of mutants
P. odoriferum NCPPB3841 ΔsaxA, P. versatile SCC1 ΔsaxA, and P. parmentieri SCC3193 were complemented by producing SaxA in trans from the low-copy-number plasmid pMW119 (Nippon Gene, GenBank: AB005476.2). Selection of pMW119 is based on ampicillin, to which P. versatile SCC1 is naturally resistant. Thus, the npt gene conferring resistance to kanamycin was amplified by PCR from Entranceposon (Kan R -3) F-779 (Thermo Scientific) using primers Kan(SacI)_F and Kan(EcoRI)_R (Table S1). The resulting fragment was digested with SacI and EcoRI and ligated into SacI-and EcoRI-digested pMW119. Subsequently, the saxA genes were inserted into the multiple cloning site of pMW119-npt in the same orientation as lacZ in the plasmid. saxA genes were amplified by PCR from genomic DNA of P. odoriferum NCPPB3841 and P. versatile SCC1 using the primer pairs Pco_3841_US/Pco_3841_DS and SCC1_saxA(BamHI)_F/SCC1_saxA(SacI)_R (Table S1), respectively.
The P. odoriferum NCPPB3841 saxA fragment was blunt-end cloned into SmaI-digested cloning vector pUC19, excised from pUC19 by digestion with SacI taking advantage of a naturally occurring SacI site located 320 bp upstream of the start codon within the saxA fragment in addition to the SacI site located within the multiple cloning site of pUC19, gel purified and ligated into SacI-digested pMW119npt. The P. versatile SCC1 saxA fragment was digested with BamHI and SacI and ligated into BamHI-and SacI-digested pMW119-npt.
Correctness of insert sequences and orientations was verified by PCR and sequencing (Table S1). The complementation plasmids and pMW119-npt without an insert were electroporated into competent cells of P. odoriferum NCPPB3841 ΔsaxA, P. versatile SCC1 ΔsaxA as well as P. parmentieri SCC3193 as described above for plasmid pKD46 and plated on LB plates containing 25 μg/ml kanamycin.

| Arabidopsis leaf maceration assay
A. thaliana ecotype Col-0 wild-type and myb28-1 myb29-1 (Sønderby et al., 2007)  with tissue paper, and placed between transparent films for scanning with resolution of 600 dots per inch (dpi). The macerated leaf area was manually measured from the scanned images using Fiji ImageJ v.
1.51. All experiments were conducted so that bacterial suspensions and plant lines were colour-coded and the person performing the infections and processing the images was unaware of the identities of the bacteria or plants until after all data were gathered. Each experiment was conducted three times. Leaves inoculated with P. polaris NCPPB3395 did not show maceration even when a higher inoculum of OD 600 of 0.2 was used. The conditions for the Arabidopsis leaf maceration experiment using P. parmentieri SCC3193 were slightly altered as described here. The cells were diluted to an OD 600 of 0.2 instead of 0.002 and not three but five leaves per plant (leaves 5 to 9) were wounded. The maceration of a leaf after 24 hr was qualified manually to be either positive or negative in four experiments totalling 30 plants for P. parmentieri SCC3193 transformed with empty pMW119 vector and P. parmentieri SCC3193 transformed with pMW119 + SCC1_saxA.

| B. oleracea var. capitata leaf maceration assay
Leaves of store-bought cabbages (B. oleracea var. capitata) were surface-sterilized with 70% ethanol and discs 3 cm in diameter were cut and placed on top of wet tissue paper on six-well plates.
Overnight-grown P. odoriferum NCPPB3841 and P. versatile SCC1 cells were washed with 10 mM MgSO 4 and diluted to an OD 600 of 0.01 corresponding to approximately 10 7 cfu/ml. A small hole was punctured with a pipette tip at the centre of each disc and a 5 μl drop of bacterial suspension was pipetted on top of the hole. The plates were sealed with Parafilm, placed in a closed container, and incubated in the dark at room temperature. After 48 hr, the discs were gently dried with tissue paper and placed between transparent film for scanning with resolution of 600 dpi. The macerated leaf area was manually measured from the scanned images using Fiji ImageJ v. 1.51. Twelve discs were inoculated with wild-type bacteria and 12 discs with the saxA mutant in each experiment. The experiment was conducted three times and one cabbage was used per experiment.

| Statistical analysis of maceration assays
Statistical analysis was performed in the R programming environment and figures were prepared using the ggplot2 package in R (Wickham, 2016;R Core Team, 2018). Due to non-normal distributions and/or unequal variances, the Kruskal-Wallis rank sum test followed by pairwise Wilcoxon rank sum test was used to address the statistical significance of differences between groups. p values were not adjusted to correct for multiple comparisons.

| P. polaris and P. parmentieri growth on Arabidopsis extracts
Plant extracts were prepared from 6-week-old A. thaliana ecotype Col-0 wild-type and myb28-1 myb29-1 (Sønderby et al., 2007) plants with a protocol modified from Fan et al. (2011). Rosette leaves were collected and 2 ml M9 salt solution (40 mM Na 2 HPO 4 , 22 mM KH 2 PO 4 , 8.5 mM NaCl, 18.7 mM NH 4 Cl) was added for every 1 g of plant tissue followed by homogenization with a Polytron PT 2100 homogenizer (Kinematica AG). Cell debris was pelleted by centrifugation at 3,230 g for 15 min and the supernatant was first filtered through a 0.45 μm filter followed by filtering through a 0.20 μm filter to produce the final extract. A 96-well plate growth assay was used to assess the effect of saxA on the ability of P. polaris NCPPB3395 and P. parmentieri SCC3193 to grow in the presence of plant extract. Fresh colonies of P. polaris NCPPB3395 wild-type, ΔsaxA1, ΔsaxA2, and the ΔsaxA1 + ΔsaxA2 double mutant, as well as P. parmentieri SCC3193 and P. parmentieri SCC3193 transformed with pMW119 + SCC1_saxA were inoculated into the wells of 48-well plates containing 750 μl of liquid LB medium. Six inoculations were performed per strain and in addition six control wells with medium were left uninoculated. The plate was incubated overnight at 28 °C with shaking (120 rpm).
The contents of the wells were then diluted 1:5 in M9 salt solution supplemented with 10 mM sucrose on a 96-well plate and attenuation of light at 600 nm was measured using an Enspire 2300 plate reader (Perkin Elmer) to ensure approximately equal overnight growth of all cultures. Then 20 μl of the dilutions were used to inoculate wells of another 96-well plate containing 180 μl of M9 salt solution supplemented with 10 mM sucrose or 100 μl of plant extract (wild-type or myb28 myb29) and 80 μl of M9 salt solution supplemented with 10 mM sucrose. Attenuation at 600 nm was measured immediately after inoculation and subsequently every 2 hr for a period of 10 hr to detect bacterial growth. Between measurements the 96-well plate was incubated with a lid on at 28 °C with shaking (120 rpm).

| saxA distribution in the Pectobacterium genus
In order to understand the evolutionary history of saxA genes in the