Bacterial strains and plasmids used in this study are listed in Supplementary Table 1. Bacteria were cultured in Luria-Bertani (LB) agar or broth medium (Sambrook et al., 1989). Acidovorax citrulli and Escherichia coli were cultured at 28 and 37°C, respectively. When required, the appropriate dose of antibiotics were added to media at the following final concentrations: rifamycin (Rif) 100 μg/mL, kanamycin (Km) 50 μg/mL, gentamicin (Gm) 50 μg/mL, chloramphenicol (Cm) 20 μg/mL and ampicillin (Amp), 100 μg/mL. The turbidity of the cell suspensions was measured by optical density at 600 nm using a spectrophotometer (Thermo Scientific, Waltham, MA, United States). For pathogenicity assays, melon (cv. Huanghou) seeds were planted in soil mixed with 50% vermiculite and grown under standard greenhouse conditions including 24°C and 12 h of natural light until inoculated. The inoculated seedlings were incubated in growth chamber with 28°C, 85% relative humidity (RH) and 12 h of fluorescent light.

To obtain a high efficiency, random insertion mutant library, triparental mating was conducted. In brief, cells of wild-type A. citrulli strain XjL12, and E. coli containing pUTKm and pRK600, respectively, were mixed and cultured on nitrocellulose membrane (NCM) on an LB agar plate. After 48 h, the lawn was harvested, washed and plated on LB agar supplemented with Rif and Km. The resulting mutants were confirmed by A. citrulli-specific primers WFB1/WFB2 (Walcott and Gitaitis, 2000) and KMF/KMR for the Tn5 transposon. Mutants were grown overnight and bacterial cell suspensions were adjusted to OD₆₀₀ = 0.3 (3 × 10⁸ CFU/mL). Approximately 10 μL of each cell suspension was infiltrated into tobacco leaves to test for HR induction. The inoculated tobacco leaves were grown at 28°C and observed per 8 h.

To identify the disrupted gene of the Tn5 insertion mutant of A. citrulli, a plasmid rescue method was employed as described previously (Bahar et al., 2009b) with some modifications. Briefly, the genomic DNA of the A. citrulli mutant was extracted and digested with PstI restriction enzyme that is unable to digest the Tn5 region, but is able to excise the genome into fragments. The digestion products were ligated into pUC19 digested with the same restriction enzyme, and the recombinant vector was introduced into DH5α. Due to the existence of a kanamycin (KM) cassette in Tn5, the flanking regions of Tn5 were identified by backward sequencing using the specific primers Tn5-F/Tn5-R for KM cassette. The sequence was then blast searched against the A. citrulli AAC00-1 genome in NCBI using Blastn. All primer sequences used in this study are listed in Supplementary Table 2.

The ΔyggS mutant of A. citrulli was generated through homologous recombination as described previously (Johnson et al., 2011). The yggS gene (locus tag: Aave_0638) is located in region 688,056 to 688,781 in the A. citrulli AAC00-1 genome (GenBank NC_008752). The upstream and downstream fragments of yggS were amplified using primer pairs (yggS-upF/yggS-upR, yggS-downF/yggS-downR) designed using Primer3 online¹, and then digested with restriction enzymes (HindIII and XbaI, BamHI and KpnI, respectively). The Km fragment was cloned from pET30 using Km cassette primer pair Km-F/Km-R and digested with XbaI and BamHI simultaneously. Three fragments were ligated with suicide vector pEX18 (digested with HindIII and KpnI), and the recombinant vector pEX18yggSKm was introduced into E. coli BW20676 for biparental mating with wild-type XjL12. The resulting A. citrulli mutant ΔyggS was confirmed by PCR assay using primers yggS-upF/yggS-downR.

To construct complemented strains, the yggS expression vector was first constructed using pBBR1-MCS-5. The yggS promoter was predicted using the online promoter prediction program². The sequence containing the promoter and open reading frame of yggS, was generated using primers yggS-F/yggS-R and digested with KpnI and BamHI and then ligated with pBBR-MCS-5 digested with the same enzymes. Afterwards, the recombinant vector pBBR-yggS verified by Sanger sequencing was introduced into E. coli BW20676 for biparental mating with the A. citrulli mutant ΔyggS. All primer sequences used in this assay are listed in Supplementary Table 2.

Acidovorax citrulli growth was measured in LB broth. After culturing overnight, the strains were harvested and adjusted to an OD₆₀₀ of 0.1 with sterilized water. The bacterial suspensions were added to fresh LB broth at 1:100 (vol/vol) and then incubated at 28°C with shaking at 220 rpm. A growth curve was investigated by measuring bacterial cell turbidity using a spectrophotometer (BioPhotometer, Eppendorf) at OD₆₀₀ at 2 h intervals until the cultures reached the plateau phase. In this assay, three biological replicates were performed and the experiment was repeated three times independently.

The ability of biofilm formation of A. citrulli was measured as described previously (Bahar et al., 2009b). Briefly, all strains cultured overnight were harvested and washed twice with sterilized water. Forty microliters of each bacterial suspension at OD₆₀₀ of 1.0 were added to 4 mL LB broth, in 12-well polyvinyl chloride (PVC) plates and incubated at 28°C. After 48 h, the cell suspension in each well was removed and the plate was dried at 80°C for 20 min. The biofilms attached to the plate well walls were stained with 1% methyl violet for 50 min and then solubilized in absolute ethyl alcohol. The biofilms were quantified by measuring solutions at OD₅₉₀ with a Microplate Reader (Synergy H1, Biotek). Three replicates for each strain were performed per experiment and the experiment was repeated three times.

Swimming assay was conducted as a previous protocol (Wang et al., 2016) with some modifications. For the swimming assay, 5 μL of cell suspensions at OD₆₀₀ of 0.3 (3 × 10⁸ CFU/mL) were deposited in the center of oligotrophic medium containing 0.3% agar and incubated at 28°C. The diameters of three colonies of each strain were measured after 48 h. Each strain was tested three times per experiment and this experiment was conducted three times.

To determine the ability of A. citrulli to induce hypersensitive response (HR), cell suspensions were injected into the leaves of Nicotiana tabacum as described previously (Liu et al., 2019). In brief, A. citrulli strains grown in LB overnight were washed with sterilized water and adjusted to OD₆₀₀ = 0.3. About 10 μL of cell suspensions were syringe-infiltrated into the leaves of N. tabacum grown at 28°C and HR was observed after 12 h. Three leaves at the same leaf position on the different stem were inoculated by tested strains. The experiment was conducted three times.

To evaluate A. citrulli sensitivity to some chemical compounds, the inhibition zone method was used. For sensitivity to H₂O₂ and antibiotics, 1 mL of cell suspensions at OD₆₀₀ of 0.3 were added to 50 mL LB agar medium and poured into sterilized petri dishes. A sterilized paper disk, approximately 4 mm in diameter, was placed in the center of each plate containing bacteria. Ten microliters of H₂O₂ (5% and 10%) or antibiotics were dropped on the disk. After incubating at 28°C for 48 h, the diameters of inhibition zones were measured. Similarly, sensitivity to 1 M PN was tested as described above except for the concentration of cell suspension. To make the inhibition zone more easily visible, cell suspension added to LB agar medium was at an OD₆₀₀ = 0.1. In this assay, the final concentrations of tested antibiotics were as follows (antibiotic/concentration in mg/mL): ampicillin/10, β-chloro-D-alanine (BCDA)/100, chloramphenicol/4, D-cycloserine (DCS)/10, gentamicin/12.5, spectinomycin/50, tetracycline/5. Each strain was tested three times per experiment and the experiment was conducted three times.

In order to evaluate the effect of yggS on A. citrulli virulence, three inoculation methods were performed as described previously (Liu et al., 2019).

A previous established protocol (Tian et al., 2015) was used to assess the role of yggS in A. citrulli colonization of melon seedlings with slight modification. For cotyledon colonization assay, cell suspensions of each A. citrulli strain (approximately 1 × 10³ CFU/mL) were injected into at least twenty-five cotyledons of melon (cv. Huanghou) seedlings per experiment. Five 5-mm disks were collected from cotyledons injected with each strain at 0, 1, 2, 3, 4 dpi and triturated in 1 mL of buffer in the sterilized 1.5-mL centrifuge tubes. Homogenate was 10-fold serially diluted with sterile water and 100 μL of homogenate was spread on LB plates with appropriate antibiotics. Resulting A. citrulli colonies were counted after two days. For seed colonization assays, the seeds (cv. Huanghou) in the assay were disinfected with 5% H₂O₂ for 20 min before germination to prevent microbe contamination. The front end of the germinating seeds were opened gently and five microliters of cell suspensions (approximately 1 × 10⁵ CFU/mL) were inoculated into the melon seeds. Each strain was inoculated into at least thirty-five melon seeds per experiment. The inoculated seeds were placed on moist filter paper. Five seeds inoculated with each strain were collected at 0, 1, 2, 3, 4, 5, 6, 7 dpi, and each seed was shaken for 10 min in a sterilized 2-mL centrifuge tube containing 1 mL sterilized water. Seed homogenate was 10-fold serially diluted with sterilized water and 100 μL was spread on LB plates with appropriate antibiotics.

Bacterial colonization was quantified by calculating the area under population dynamics curve (AUPDC) as follows: AUPDC = ∑i=1n[(Yi+Yi+1)/2]×(Xi+1-Xi) (Bjarko and Line, 1988). (X_i: the value of horizontal coordinates at ith observation; Y_i: the value of vertical coordinates at the ith observation; n: the total number of observation). This assay was repeated three times.

To determine the regulatory mechanism of YggS in A. citrulli, RNA-Seq was conducted commercially by Beijing Allwegene Technology Company Limited (Beijing, China). Briefly, total RNA were extracted from A. citrulli strains that were cultured in LB broth to OD₆₀₀ = 1.0 using the TRIzol method (TIANGEN BIOTECH, Bejing). RNA was quantified by Agilent 2100 (Agilent Technologies, CA, USA), and the quality and integrity were detected by NanoDrop spectrophotometer (IMPLEN, CA, United States). Ribosomal RNA (rRNA) was removed from qualified RNA sample using Vazyme Ribo-off rRNA depletion kit (Bacteria) (Vazyme biotech, United States). Subsequently, the sequencing libraries were generated using NEBNext ULtraTM RNA library Prep Kit (NEB, United States). Library quality was assessed on the Agilent Bioanalyzer 2100 system. The qualified library was sequenced by Illumina Hiseq 4000 platform. The reads cleaned by Trimmomatic V0.33 were mapped to A. citrulli AAC00-1 genome (GenBank NC_008752) by Bowtie2 V2.2.6. Resulting data were subjected to DESeq R package (1.10.1) for analyzing differential expression genes between mutant and wild-type strain. Genes with an adjusted P-value < 0.05 found by DESeq were assigned as differentially expressed. Gene Ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were performed by GOSeq and KOBAS, respectively. GO term and KEGG pathway with corrected P value < 0.05 were defined as significantly enriched by the DEGs. Each strain was analyzed in three biological repetitions.

To validate the result of RNA-Seq, quantitative real-time PCR analysis (qPCR) was conducted. The culture condition for bacteria growth was the same as described in the transcriptome sequencing and data analysis. Total A. citrulli RNA was extracted from bacteria using the bacterial RNA kit (OMEGA), and the concentration of RNA was measured by spectrophotometer (Nanodrop One, Thermo Scientific). cDNA was synthesized and purified using HiScript III RT SuperMix kit (Vazyme, Nanjing, China) and diluted to 100 ng/μL for qPCR with ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China). The qPCR assay was carried out in a real-time PCR machine (7,500, Applied Biosystems) as the following program: 95°C for 30 s (1 cycle); 95°C for 10 s, 60°C for 30 s (40 cycles); melting curve profiled from 60°C to 95°C to check the specialty of reaction. The primers of selected genes used in the assay are listed in Supplementary Table 2. 16s ribosomal RNA gene was used as a reference gene. Each sample was tested four times per experiment and experiments were conducted three times independently. Relative gene expression was calculated in the method of 2^−ΔΔct as described previously (Livak and Schmittgen, 2001).

All data were analyzed by SPSS Statistics 26. The one-way analysis of variance (ANOVA) and least Significant Difference (LSD) test were used to determine the significant difference in disease index, biofilm assay, motility assay and AUPDC. Differences with p < 0.05 were considered significant.

Through library screening, an A. citrulli mutant named sk24 that was unable to induce a HR on tobacco at 12 h after inoculation was obtained (Supplementary Figure 1). By subcloning and sequence analysis, we identified the gene disrupted by transposon as Aave_0638 (Supplementary File 1). This gene putatively encodes YggS family pyridoxal phosphate-dependent enzyme, which exhibits structual similarity to the N-terminal domain of alanine racemase (EC 5.1.1.1). Multiple sequence alignment shows that the amino acid sequence of YggS from A. citrulli has high identity among the tested bacteria including E. coli (Figure 1A). By BLASTP sequence homology analysis, YggS displays 53.94, 38.97, 49.57, 53.78, 55.46, and 55.14% sequence identity with homologs from E. coil MG1655, S. elongates PCC7942, X. oryzae PXO99A, P. syringae DC3000, E. amylovora CFBP1430 and P. carotovorum PC1, respectively. Furthermore, previous studies have shown that yggS clusters strongly with genes such as those related to metabolism, including cell division and cell wall (Prunetti et al., 2016). We observed the position of yggS on genome and found that yggS and pilT, encoding ATPase-mediating pilus retraction and disassembly, are clustered together. This genetic organization appears to be conserved for serval phytopathogenic bacteria (Figure 1B).

To investigate the effect of yggS on A. citrulli virulence, we compared the pathogenicity of the wild-type strain XjL12, the mutant strain ΔyggS and complemented strain CyggS by conducting cotyledon injection and spray inoculation of melon seedling euphylla. The results of the spray inoculation assay showed that mutation of yggS impaired the virulence of A.citrulli (Figure 2A). Seven days after inoculation, necrotic lesions developed on the euphylla sprayed with A. cirulli strains XjL12 and CyggS, while the euphylla sprayed with ΔyggS showed mild necrotic symptoms. The disease indices (DI) caused by XjL12, Δyggs and CyggS were 47.87, 5.15, and 40.78, respectively. The DI of ΔyggS was significantly lower than the wild-type and complemented strains (P < 0.05) (Figure 2B). As shown in Figure 3A, the cotyledons injected with ΔyggS showed no symptoms, similar to the negative control (NC) at 5 days after inoculation. Meanwhile, melon cotyledons injected with XjL12 and CyggS showed typical BFB symptoms, including seedling blight/collapse. By 7 dpi, there were no visible BFB symptoms on melon cotyledons injected with the mutant strain. In agreement with the results of cotyledon injection, ΔyggS impaired its ability to colonize melon cotyledons. The bacterial populations of XjL12, ΔyggS and CyggS were approximately 2.94 × 10⁸, 1.69 × 10⁶, 1.95 × 10⁸ CFU/cm² by 4 dpi, respectively (Figure 3B). As expected, AUPDC data showed that the population of ΔyggS was significantly less than that of XjL12 and CyggS in cotyledons (p < 0.05) (Figure 3C).

In the seed-to-seedling transmission assay, melon seeds infiltrated with A. cirtulli strains were planted at 20 seeds per pot and the percentage of dead seedlings was calculated one week after planting. Most seedlings from seeds inoculated with wild-type strain XjL12 and complemented strain CyggS were died, while all the seedlings inoculated with ΔyggS survived without visible BFB symptoms (Figure 4A). The reduced seed-to-seedling transmission indicated that it is likely that the loss of yggS impaired the colonization ability in melon seeds. So, we assayed the population dynamics of A. citrulli in germinating seeds and the bacterial population in inoculated seeds were quantified per 24 h after inoculation (Figure 4B). The average populations of XjL12, ΔyggS and CyggS were approximately 1.07 × 10⁸, 5.12 × 10⁵, 5.37 × 10⁷ CFU/g of seed by one week after inoculation, respectively. AUPDC data showed that the populations of ΔyggS was significantly less than XjL12 and CyggS in seeds (p < 0.05) (Figure 4C). These results suggest that yggS plays an important role in seed-to-seedling transmission of BFB and seed colonization by A. citrulli.

For most organism but Pseudomonas aeruginosa, YggS is dispensable for growth (Rusmini et al., 2014; Vu et al., 2020). To investigate the role of YggS in A. citrulli growth, the ability of the A. citrulli strains ΔyggS, wild-type XjL12 and CyggS complement to grow in LB medium was compared by measuring the optical density of cell suspensions. ΔyggS reached the exponential growth phase after 14 h and reached the stationary phase at 26 h. On the other hand, the wild-type and complemented strains reached the exponential and stationary phases approximately at 12 h and 24 h, respectively (Figure 5A). The date indicated that yggS also plays a dispensable role for A. citrulli despite of the slight reduction in growth of ΔyggS. In addition, ΔyggS cultured in the LB broth amended with 1 μM PL returned to wild-type growth (Figure 5A), while PL was unable to enhance the growth of wild-type or complemented strain (Supplementary Figure 2).

To determine if PL could restore compromised virulence, ΔyggS was syringe-inoculated into melon cotyledons along with 1 μM PL. Although PL restored growth of ΔyggS, it did not restore pathogenicity on melon cotyledons (Figure 5B). This result showed that the effect of yggS on bacterial growth may not be related to virulence attenuation.

Previous studies have confirmed that bacterial motility plays a key role in pathogenicity of A. citrulli (Bahar et al., 2009b; Bahar et al., 2011). A. citrulli strains produced near circular colonies via swimming motility using the soft agar plate assay (Figure 6A). The diameters of XjL12, ΔyggS, CyggS, and XjL12 (pYggS) were 2.75 ± 0.17, 2.25 ± 0.19, 2.20 ± 0.14, and 2.05 ± 0.09 cm, respectively (Figure 6B). The diameters of the swimming motility colony produced by ΔyggS and overexpressed strain XjL12 (pYggS) were significantly smaller than the colony produced by wild-type strain (p < 0.05) (Figure 6B). However, the complemented strain with pBBRMCS-5, containing the yggS gene, failed to restore the swimming motility phenotype.

In E. coli, excess PN led to a zone of inhibition with ΔyggS, but not with the wild-type strain. We observed a similar phenomenon with A. citrulli (Figure 7A). The mean diameter of the inhibition zones produced by ΔyggS (2.42 ± 0.12 cm) was greater (p < 0.05) than the diameters of the wild type XjL12 (1.50 ± 0.08 cm) and the complementation strain CyggS (1.41 ± 0.06 cm) (Figure 7B). This result suggested that the deletion of yggS increases the sensitivity to PN. However, the sensitivity to PN could not be rescued by exogenous addition of PL, and even with a range of PL concentrations; this was different to the observations reported for E. coli (Figure 7A; Prunetti et al., 2016). To determine if YggS is involved in oxidative stress tolerance in A. citrulli, we measured the sensitivity to H₂O₂ (Figure 7C). Compared to the wild-type and complemented strains, the diameters of inhibition zone for the mutant were greater (p < 0.05) at 5 and 10% H₂O₂; conversely, the diameter of CyggS was not significantly different to that of XjL12 (Figure 7D). The results suggest that deletion of yggS increases sensitivity to H₂O₂ and that yggS contributes to oxidative stress tolerance of A.citrulli. In order to determine whether yggS inactivation affects the sensitivity of A. citrulli to antibiotics, we tested 7 common antibiotics, including D-cycloserine (DCS) and β-chloro-D-alanine (BCDA) targeting alanine racemase. As shown in Figures 7E,F, the yggS mutant was more sensitive than XjL12 and CyggS to BCDA, but not to the other antibiotics (Supplementary Figure 3). Further, the sensitivity of mutant to BCDA was reduced by exogenous addition of L-alanine or D-alanine (Figure 7E). These results suggest that yggS contributes to the resistance of A. citrulli to BCDA.

To determine whether yggS contributes to the A. citrulli type III secretion system (T3SS), we examined HR induction by wild-type XjL12 and yggS mutant ΔyggS strains on N. tabacum (Figure 8). At 12 hpi, XjL12 and CyggS induced HR, while ΔyggS showed no cell death (similar to the negative control). However, HR induction was observed for ΔyggS at 24 hpi. These observations suggest that the deletion of yggS delays HR induction on N. tabacum and the deletion of yggS potentially impairs the T3SS function. In addition, the effect of reduced growth in ΔyggS on HR induction is inevitable.

To investigate the cause of attenuated virulence observed for ΔyggS, the differences in the transcriptomes between ΔyggS and XjL12 were analyzed by RNA-Seq. A list containing the expression of all genes including the differentially expressed genes (DEGs) is provided in Supplementary File 2. The RNA-Seq results were validated by qPCR assay with ten chosen DEGs (Supplementary Figure 4). Compared with wild-type strain, there were 971 DEGs in the mutant strain: 506 genes were upregulated and 465 were downregulated (Supplementary Figure 5). In agreement with the previous abovementioned phenotypes, the DEGs were involved with A. citrulli motility, T3SS and anti-oxidative stress (Table 1).

Among these DEGs listed in Table 1, in vitro condition, a variety of T3SS genes including that encode indispensable apparatus subunit HrcC, HrcJ, et al., transcriptional regulator HrpX directly mediating the expression of most T3SS and T3Es genes significantly lowly-expressed. Correspondingly, the expressing level of many T3Es genes was decreased (Table 2). In addition to the T3Es revealed by annotation of genome, we also summarized the genes with high similarity to known T3Es from other plant pathogens based on the results reported by Jiménez-Guerrero et al. (2020). The homologues of Aave_2177, Aave_1555, Aave_1244, and Aave_3960 have been validated as T3Es in A. citrulli strain M6 by translocation assay (Jiménez-Guerrero et al., 2020). Gene ontology (GO) analysis showed that down-regulated DEGs were enriched in terms related to secretion system (e.g., protein secretion, peptide secretion, secretion by cell) and regulations of these pathways were downregulated significantly in ΔyggS (Supplementary Figure 6A). Based on these factors, it is possible that YggS regulates a variety of pathogenic factors, and that regulation of secretion systems, especially T3SS, contributes to attenuated virulence on melon.

During infection, pathogens encounter the large generation of reactive oxygen species as part of the oxidative burst associated with plant defense response, making response to and protection against oxidative stress an important aspect for infection (Apel and Hirt, 2004; Burbank and Roper, 2014). The DEGs related to oxidative stress response partly listed in Table 1 indicated the perturbation in antioxidative response as a result of yggS mutation. FlgM and PilA have been conformed as a key factor responsible for flagellum and TFP assembly, respectively, in A. citrulli (Rosenberg et al., 2018; Yang et al., 2018), and both were upregulated in mutant (Table 1). Surprisingly, a gene Aave_3811, accounting for synthesizing acyl homoserine lactones-QS signal molecules (Fan et al., 2011), were negatively regulated. The expressing level of genes related to T2SS and T6SS was also changed in ΔyggS.

Ribosomes carrying out protein synthesis is required for cell growth (Lempiäinen and Shore, 2009). However, the biosynthesis of ribosomes is energy-consuming, thus the overproduction of ribosomal proteins is detrimental to cell proliferation (Jorgensen et al., 2004). GO analysis and KEGG pathway analysis revealed that the ribosome was most affected in yggS mutation (Supplementary Figures 6, 7). Supplementary Table 3 displayed the expression profile of the DEGs encoding 30S or 50S ribosomal proteins and all the genes were upregulated, indicating that inactive yggS caused increasing ribosomal biosynthesis, which may result in the growth defect. Previous reports have demonstrated the yggS-deficient E. coil impacts amino acids homeostasis, such as valine and isoleucine (Ito et al., 2013, 2019). According to KEGG pathway analysis, the pathway in valine, leucine and isoleucine degradation, β-Alanine metabolism was significantly regulated by yggS. In addition to both pathways, Supplementary Table 4 displayed the expression profile of the DEGs related to arginine and proline metabolism, alanine, aspartate and glutamate metabolism, glycine, serine and threonine metabolism, indicating that yggS also was involved in amino acids homeostasis in A. citrulli.

In addition, among the most GO enriched terms, only up-regulated DEGs were divided into terms belonging to cellular component domain, such as cellular anatomical entity, intracellular organelle, et al. (Supplementary Figure 6B). YggS appears to negatively affect pathway terms related to cellular component in A. citrulli.