pilA Gene Contributes to Virulence, Motility, Biofilm Formation, and Interspecific Competition of Bacteria in Acidovorax citrulli

Acidovorax citrulli, the causative agent of bacterial fruit blotch, can be divided into two main groups based on factors such as pathogenicity and host species preference. PilA is an important structural and functional component of type IV pili (T4P). Previous studies have found significant differences in pilA DNA sequences between group I and group II strains of A. citrulli. In this study, we characterized pilA in the group I strain pslb65 and the group II strain Aac5. pilA mutants, complementation strains, and cross-complementation strains were generated, and their biological phenotypes were analyzed to identify functional differences between pilA in the two groups. pilA deletion mutants (pslb65-ΔpilA and Aac5-ΔpilA) showed significantly reduced pathogenicity compared with the wild-type (WT) strains; pslb65-ΔpilA also completely lost twitching motility, whereas Aac5-ΔpilA only partially lost motility. In King’s B medium, there were no significant differences in biofilm formation between pslb65-ΔpilA and WT pslb65, but Aac5-ΔpilA showed significantly reduced biofilm formation compared to WT Aac5. In M9 minimal medium, both mutants showed significantly lower biofilm formation compared to the corresponding WT strains, although biofilm formation was recovered in the complementation strains. The biofilm formation capacity was somewhat recovered in the cross-complementation strains but remained significantly lower than in the WT strains. The interspecies competitive abilities of pslb65-ΔpilA and Aac5-ΔpilA were significantly lower than in the WT strains; Aac5-ΔpilA was more strongly competitive than pslb65-ΔpilA, and the complementation strains recovered competitiveness to WT levels. Furthermore, the cross-complementation strains showed stronger competitive abilities than the corresponding WT strains. The relative expression levels of genes related to T4P and the type VI secretion system were then assessed in the pilA mutants via quantitative PCR. The results showed significant differences in the relative expression levels of multiple genes in pslb65-ΔpilA and Aac5-ΔpilA compared to the corresponding WT stains. This indicated the presence of specific differences in pilA function between the two A. citrulli groups, but the regulatory mechanisms involved require further study.


Introduction
Bacterial fruit blotch (BFB) is caused by Acidovorax citrulli [1] and is one of the most important seed-borne bacterial diseases of cucurbit crops (such as watermelon and melon). The disease causes seedling blight and fruit rot in susceptible cucurbits [2]. Since it was

Bacterial Strains, Plasmids, and Growth Conditions
The bacterial strains and plasmids used in this study are listed in Table S1. A. citrulli strain Aac5 was isolated from watermelon collected in Taiwan, China, and strain pslb65 was isolated from melon collected in Xinjiang, China. A. citrulli strains were grown in King's B (KB) broth or on KB plates at 28 • C. Escherichia coli strains were grown in Luria Bertani (LB) medium at 37 • C [32]. Liquid cultures were grown with 200 rpm shaking on a rotary shaker (DDHZ-300; Taicang Experimental Instrument Factory, Jiangsu, Taicang, China). Growth media contained 100 µg/mL ampicillin (Amp), 25 µg/mL chloramphenicol (Cm), 25 µg/mL gentamicin (Gm), and/or 50 µg/mL kanamycin (Km), as specified for each experiment. All antibiotics were purchased from Sigma (Shanghai, China).

Construction of pilA Gene Deletion Mutants and Complementation Lines
Markerless pilA deletion mutants were generated in the A. citrulli strains Aac5 and pslb65 using homologous double recombination [34]. The sequences upstream and downstream of pilA were amplified from the wild-type (WT) pslb65 and Aac5 strains using the 65AL-F/R and 65AR-F/R primers (for pslb65) and the A5AL-F/R and A5AR-F/R primers (for Aac5) ( Table S2). The sequences of the resulting fragments were confirmed via sequencing. The upstream and downstream fragments were concatenated with overlap PCR and then ligated into the pk18mobsacB plasmid using the ClonExpress II One Step Cloning Kit (Vazyme Biotech, Nanjing, China) to generate the pK18-pslb65pilA and pK18-Aac5pilA constructs in E. coli DH5α. pK18-pslb65pilA and pK18-Aac5pilA were then introduced to the A. citrulli strains pslb65 and Aac5, respectively, through triparental conjugation using pRK600 as a helper plasmid. Single-exchange colonies were selected on KB plates containing Amp and Km. Individual transformants were continuously sub-cultured in liquid KB medium containing Amp alone and then screened on KB supplemented with Amp and 10% sucrose. The pilA deletion mutants for pslb65 and Aac5, pslb65-∆pilA and Aac5-∆pilA, respectively, were confirmed via PCR using the 65-∆pilA-L/R, A5AL-L/A5AR-R, Km-F/R, and WFB1/2 primers (Table S2). Information on the primer position, product length, and gene knockout position during the construction of pilA gene deletion mutants and the PCR validation process can be seen in Figure S1.

Spray-Inoculation Assays
The virulence levels of each strain (namely, pslb65, pslb65-∆pilA, pslb65-∆pilAcomp1, pslb65-∆pilAcomp2, Aac5, Aac5-∆pilA, Aac5-∆pilAcomp1, and Aac5-∆pilAcomp2) were Microorganisms 2023, 11,1806 4 of 17 measured in 3-week-old watermelon seedlings (Citrullus lanatus cv. 'Jingxin#3') and melon seedlings (Cucumis melo cv. 'TVF192'). Briefly, A. citrulli strains were cultured overnight in KB broth to an OD 600 of 0.3 (corresponding to~3 × 10 8 colony-forming units (CFU)/mL). The leaves of the melon and watermelon seedlings were spray-inoculated with the bacterial suspension, with sterile water serving as the negative control. Each treatment had 15-20 melon or watermelon seedlings, and 200mL of bacterial suspension was evenly sprayed onto the front and back of the leaves of seedlings in each treatment. At 15 d after inoculation (DAI), the disease severity was determined by measuring the symptoms and calculating the disease index using previously described methods with slight modifications [14]. Seedlings were incubated in a light incubator with a relative humidity of 80% under a 12/12 h light/dark cycle at 26/20 • C.

Stem-Inoculation Experiments
All strains used in spray-inoculation experiments were also used in stem-inoculation experiments, which were performed as previously described [29] with some modifications. Briefly, at 7 d after sowing, the base of the hypocotyl of each seedling (~1 cm above the soil) was inoculated with 10 8 CFU/mL bacterial suspension using a needle. Seedlings were cultured in a light incubator under the conditions described above for the spray-inoculation assays, and the number of dead seedlings was recorded each day from 3 to 11 DAI. There were 30 seedlings per treatment group (i.e., bacterial genotype) and three independent replicates of the experiment.

HR Assays
To evaluate the effects of pilA deletion on HR induction, bacterial cell suspensions (~10 8 CFU/mL) were injected into the leaves of 3-week-old tobacco (Nicotiana tabacum var. samsun) plants as previously described [34]. Leaves were visually examined for the HR (tissue collapse) at 24 h after infiltration. Control plants were injected with sterile water. There were three independent replicates of this experiment.

Twitching Motility Assays
Twitching motility was measured in each A. citrulli strain using the method described by Bahar et al. [28]. Briefly, A. citrulli strains were grown on KB plates containing Amp at 28 • C for 72 h. Each strain was considered to have twitching motility if a thin, light halo was visible around the colonies via light microscopy. An IX83 microscope (Olympus, Japan) was used. Nine plates were inoculated per strain in each replicate experiment, and there were three independent replicates.

Biofilm Formation Assays
Biofilm formation was assessed as previously described [35]. Briefly, 24-well polystyrene cell culture plates (Costar 3524, Corning, NY, USA) were filled with KB broth and then inoculated with a 1:1000 dilution of each A. citrulli strain at a concentration of 3 × 10 8 CFU/mL. The plates were incubated at 28 • C for 48 h without shaking. After incubation, 0.1% crystal violet was added to each well, and the plate was incubated for 30 min at room temperature. Each well was then gently washed three times with distilled water. To quantify biofilm formation, the stained biofilms were solubilized in 95% ethanol for 2 h, and then the OD 575 value of each stained cell suspension was measured with an Evolution 300 UV/VIS spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). There were three biological replicates for each A. citrulli strain in each experiment, and there were three independent replicates of this experiment.

Interspecies Bacterial Competition Assays
Interspecies competitions were conducted, with E. coli DH5α serving as the prey strain. The killer strains (WT, mutant, and complementation A. citrulli strains) were cultured in KB broth to an OD 600 of 1.5, and the prey culture was grown in LB broth to an OD 600 of 2.0. The killer and prey cultures were each centrifuged and resuspended in sterile water and then mixed at a 20:1 killer-prey ratio. After co-culture, the mixed cultures were 10-fold serially diluted and spotted on LB plates containing 25 µg/mL Gm. After 3 h, the number of surviving E. coli DH5A was quantified in log 10 (CFU/mL) [18].

Quantitative Reverse Transcription (qRT)-PCR Assays
To assess the effects of pilA expression on T3SS genes, the WT A. citrulli strains pslb65 and Aac5 and the pilA mutants pslb65-∆pilA and Aac5-∆pilA were cultured in the T3SSinducing broth XVM2 [16]. To assess the effects of pilA expression on T4P genes, the same strains were cultured in KB broth. To assess the effects of pilA expression on T6SS-related genes, the same strains were also cultured in KB broth to OD 600 values of 1.0 and 1.5. Total RNA was extracted from each culture using Trizol reagent (Invitrogen, Waltham, MA, USA). cDNA fragments were synthesized using a HiScript II RT SuperMix kit (Vazyme Biotech, Najing, China). rpoB was used as the internal reference gene for expression normalization [35]. qRT-PCR was performed on a 7500 Sequence Detection System (ABI, Waltham, MA, USA) with SYBR Green SuperReal PreMix Plus (TIANGEN, Beijing, China) following the manufacturer's instructions. Relative gene expression was calculated using the 2 −∆∆Ct method [36]. The primers used for qRT-PCR are listed in Table S3. There were three technical replicates per sample and three independent replicates of the experiment.

Statistical Analysis
Statistical analyses were conducted using SPSS version 22.0 (SPSS, Chicago, IL, USA) and GraphPad Prism 7.0 software (GraphPad Software, La Jolla, CA, USA). Significant differences between treatment groups were assessed with Student's t-test (for sample pairs) or analysis of variance (ANOVA) and post hoc Duncan's multiple range test (for multiple samples). Differences were considered statistically significant at p < 0.05.

pilA Sequence Analysis in A. citrulli Strains pslb65 and Aac5
The pilA gene of the group I strain pslb65 was 522 bp, and the pilA gene of the group II strain Aac5 was 507 bp. DNA sequence analysis revealed that the two pilA genes were highly similar from positions 1 to 107 bp, after which the sequences diverged significantly. The overall similarity between the two genes was only 61.64% (Figure 1a). The alignment of the corresponding amino acid sequences showed similar results; amino acids 1-44 were identical, whereas the C-terminal sequences had extensive variation ( Figure 1b). SMART (http://smart.embl-heidelberg.de/, accessed on 7 February 2023) did not predict any typical protein domains in either of the two PilA proteins.

Knockout Mutant Verification
PCR amplification using the primer pair WFB1/WFB yielded 360 bp amplicons from each of the two putative mutants (pslb65-∆pilA and Aac5-∆pilA), but there was no amplification when the primer pairs Km-F/Km-R, 65-∆pilA-L/65-∆pilA-R, or A5AL-L/A5AR-R were used ( Figure S2a,b). The two new lines were therefore confirmed to be A. citrulli without the pilA gene, indicating successful mutant construction for each strain.

Knockout Mutant Verification
PCR amplification using the primer pair WFB1/WFB yielded 360 bp amplicons from each of the two putative mutants (pslb65-ΔpilA and Aac5-ΔpilA), but there was no amplification when the primer pairs Km-F/Km-R, 65-ΔpilA-L/65-ΔpilA-R, or A5AL-L/A5AR-R were used ( Figure S2a,b). The two new lines were therefore confirmed to be A. citrulli without the pilA gene, indicating successful mutant construction for each strain.

Complementation Line Verification
PCR was performed on the putative complementation lines using the primer pairs WFB1/WFB and Km-F/Km-R. WFB1/WFB2 yielded 360 bp amplicons, and Km-F/Km-R yielded 634 bp amplicons from the four complementation lines (pslb65-ΔpilAcomp1, pslb65-ΔpilAcomp2, Aac5-ΔpilAcomp1, and Aac5-ΔpilAcomp2) ( Figure S2c,d,e,f). These results indicated that the four complementation strains were A. citrulli, and each contained the pBBR-pslb65pilA or pBBR-Aac5pilA plasmid. Furthermore, using the primer pair 65-ΔpilA-L/65-ΔpilA-R, pslb65-ΔpilAcomp1 and Aac5-ΔpilAcomp2 yielded identical products, and the primer pair HBA5-L/HBA5-R yielded a separate amplicon that was identical between Aac5-ΔpilAcomp1 and pslb65-ΔpilAcomp2. The PCR products were then analyzed via sequencing. Both PCR and sequencing showed that the complementation and cross-complementation lines had been successfully constructed.

The Absence of pilA Did Not Affect the HR in Nonhost Plants
The WT, mutant, and complementation strains of pslb65 and Aac5 were injected into tobacco leaves and cultivated for 24 h. The results showed that ∆pilA-pslb65 and ∆pilA-Aac5 could cause the HR in a nonhost plant species ( Figure S3).

pilA Was Required for A. citrulli Virulence
To reveal the role of pilA in A. citrulli pathogenicity, we carried out spray-and steminoculation assays.

Spray-Inoculation Assays
Watermelon and melon seedlings were inoculated with the group I strain pslb65 and the corresponding mutant and complementation strains. Leaf symptoms were observed, and the disease indices were calculated at 15 DAI. The results showed that the average disease index was significantly lower in watermelon inoculated with pslb65-∆pilA compared to WT pslb65. The average disease index was significantly higher in watermelon inoculated with pslb65-∆pilAcomp1 compared to the knockout mutant but did not reach WT levels. However, the disease index of watermelon plants inoculated with the pslb65 cross-complementation line (pslb65-∆pilAcomp2) was statistically comparable to that inoculated with the WT strain. In melon, the average disease index of seedlings inoculated with pslb65-∆pilA was significantly lower than in those inoculated with WT pslb65. The average disease indices were significantly higher for both pslb65-∆pilAcomp1 and pslb65-∆pilAcomp2 compared to the knockout mutant but did not return to WT values ( Figure 2a).
Watermelon and melon seedlings were inoculated with the group I strain pslb65 and the corresponding mutant and complementation strains. Leaf symptoms were observed, and the disease indices were calculated at 15 DAI. The results showed that the average disease index was significantly lower in watermelon inoculated with pslb65-ΔpilA compared to WT pslb65. The average disease index was significantly higher in watermelon inoculated with pslb65-ΔpilAcomp1 compared to the knockout mutant but did not reach WT levels. However, the disease index of watermelon plants inoculated with the pslb65 cross-complementation line (pslb65-ΔpilAcomp2) was statistically comparable to that inoculated with the WT strain. In melon, the average disease index of seedlings inoculated with pslb65-ΔpilA was significantly lower than in those inoculated with WT pslb65. The average disease indices were significantly higher for both pslb65-ΔpilAcomp1 and pslb65-ΔpilAcomp2 compared to the knockout mutant but did not return to WT values ( Figure 2a).  Watermelon and melon seedlings were also inoculated with the group II strain Aac5 and the corresponding mutant and complementation strains. The average disease index was significantly lower in watermelon inoculated with Aac5-∆pilA compared to those inoculated with WT Aac5. The average disease indices of watermelon treated with Aac5-∆pilAcomp1 and Aac5-∆pilAcomp2 were significantly higher than those inoculated with the knockout mutant but did not reach WT levels. The average disease index of melon inoculated with Aac5-∆pilA was comparable to the results seen in watermelon (Figure 2b).

Stem-Inoculation Assays
In order to further reveal the role of differences in pilA between the two groups of strains in A. citrulli pathogenicity, watermelon seedlings were inoculated with the group I strain pslb65 and the group II strain Aac5 and the associated knockout mutant and complementation strains. We then measured the number of diseased seedlings from 3 to 11 DAI. The number of diseased seedlings gradually increased over time, and trends were similar between the two groups of strains (Figure 3). At 11 DAI, an average of 21.6 seedlings inoculated with WT pslb65 were diseased (incidence rate = 72.0%). In plants inoculated with pslb65-∆pilA, an average of 6.3 seedlings were diseased (incidence rate = 21.0%). For plants inoculated with the complementation and cross-complementation lines pslb65-∆pilAcomp1 and pslb65-∆pilAcomp2, the average number of diseased seedlings increased again to 24.0 and 18.3, respectively, representing incidence rates of 80.0% and 61.0%, respectively. Seedlings inoculated with WT Aac5 averaged 23.6 diseased seedlings (incidence rate = 78.6%); this decreased to an average of 9.6 diseased seedlings (incidence rate = 32.0%) among those inoculated with the Aac5-∆pilA mutant. In seedlings inoculated with the complementation line (Aac5-∆pilAcomp1), there were an average of 22.3 diseased seedlings (incidence rate = 74.3%), which was comparable to the rate among those inoculated with the cross-complementation line Aac5-∆pilAcomp2 (20 seedlings; incidence rate = 66.6%). These results demonstrated that the pilA mutants of both pslb65 and Aac5 exhibited low virulence in watermelon seedlings, and that the functional pilA genes of group I and group II strains had similar effects on watermelon seedling disease rates. strains in A. citrulli pathogenicity, watermelon seedlings were inoculated with the group I strain pslb65 and the group II strain Aac5 and the associated knockout mutant and complementation strains. We then measured the number of diseased seedlings from 3 to 11 DAI. The number of diseased seedlings gradually increased over time, and trends were similar between the two groups of strains (Figure 3). At 11 DAI, an average of 21.6 seedlings inoculated with WT pslb65 were diseased (incidence rate = 72.0%). In plants inoculated with pslb65-ΔpilA, an average of 6.3 seedlings were diseased (incidence rate = 21.0%). For plants inoculated with the complementation and cross-complementation lines pslb65-ΔpilAcomp1 and pslb65-ΔpilAcomp2, the average number of diseased seedlings increased again to 24.0 and 18.3, respectively, representing incidence rates of 80.0% and 61.0%, respectively. Seedlings inoculated with WT Aac5 averaged 23.6 diseased seedlings (incidence rate = 78.6%); this decreased to an average of 9.6 diseased seedlings (incidence rate = 32.0%) among those inoculated with the Aac5-ΔpilA mutant. In seedlings inoculated with the complementation line (Aac5-ΔpilAcomp1), there were an average of 22.3 diseased seedlings (incidence rate = 74.3%), which was comparable to the rate among those inoculated with the cross-complementation line Aac5-ΔpilAcomp2 (20 seedlings; incidence rate = 66.6%). These results demonstrated that the pilA mutants of both pslb65 and Aac5 exhibited low virulence in watermelon seedlings, and that the functional pilA genes of group I and group II strains had similar effects on watermelon seedling disease rates.

Twitching Motility Was Impaired in the pilA Mutants of Group I and Group II A. citrulli Strains
The pslb65-∆pilA strain showed no outer halo when it was grown on KB plates, meaning it had completely lost the capacity for twitching motility. In comparison, the capacity for twitching motility was significantly reduced in the group II strain Aac5-∆pilA, but not completely lost. All four complementation lines (pslb65-∆pilAcomp1, pslb65-∆pilAcomp2, Aac5-∆pilAcomp1, and Aac5-∆pilAcomp2) showed the restoration of twitching motility to WT levels ( Figure 4).

Biofilm Formation Was Reduced in the pilA Mutants of Strains from Both A. citrulli Groups
The effect of the pilA gene on the capacity of group I and group II strains to form biofilms was next evaluated in KB and M9 media. In KB medium, pslb65-∆pilA showed decreased biofilm formation compared to the WT strain, but the difference was not statistically significant. However, Aac5-∆pilA showed significantly lower biofilm formation than the WT Aac5 (p < 0.05). For both strains, both complementation lines restored biofilm formation to WT levels (Figure 5a). In M9 medium, both pslb65-∆pilA and Aac5-∆pilA showed significantly lower biofilm formation than the corresponding WT strains (p < 0.05); all four complementation strains somewhat restored biofilm formation, but only Aac5-∆pilAcomp1 reached levels comparable to the WT (Figure 5b).

Twitching Motility Was Impaired in the pilA Mutants of Group I and Group II A. citrulli Strains
The pslb65-∆pilA strain showed no outer halo when it was grown on KB plates, meaning it had completely lost the capacity for twitching motility. In comparison, the capacity for twitching motility was significantly reduced in the group II strain Aac5-∆pilA, but not completely lost. All four complementation lines (pslb65-ΔpilAcomp1, pslb65-ΔpilAcomp2, Aac5-ΔpilAcomp1, and Aac5-ΔpilAcomp2) showed the restoration of twitching motility to WT levels ( Figure 4).

Figure 4.
Effects of pilA on the capacity for twitching motility in Acidovorax citrulli. The group I strain pslb65 and corresponding knockout, complementation, and cross-complementation lines (pslb65-ΔpilA, pslb65-ΔpilAcomp1, and pslb65-ΔpilAcomp2, respectively) and the group II strain Aac5 and the corresponding knockout, complementation, and cross-complementation lines (Aac5-ΔpilA, Aac5-ΔpilAcomp1, and Aac5-ΔpilAcomp2, respectively) were grown on King's B plates with ampicillin at 28 °C for 72 h. Twitching motility was assessed based on the visible presence of a thin, light halo around each colony under light microscopy.

Biofilm Formation Was Reduced in the pilA Mutants of Strains from Both A. citrulli Groups
The effect of the pilA gene on the capacity of group I and group II strains to form biofilms was next evaluated in KB and M9 media. In KB medium, pslb65-ΔpilA showed decreased biofilm formation compared to the WT strain, but the difference was not statistically significant. However, Aac5-ΔpilA showed significantly lower biofilm formation than the WT Aac5 (p < 0.05). For both strains, both complementation lines restored biofilm formation to WT levels ( Figure 5a). In M9 medium, both pslb65-ΔpilA and Aac5-ΔpilA showed significantly lower biofilm formation than the corresponding WT strains (p < 0.05); all four complementation strains somewhat restored biofilm formation, but only Aac5-ΔpilAcomp1 reached levels comparable to the WT (Figure 5b).

Figure 4.
Effects of pilA on the capacity for twitching motility in Acidovorax citrulli. The group I strain pslb65 and corresponding knockout, complementation, and cross-complementation lines (pslb65-∆pilA, pslb65-∆pilAcomp1, and pslb65-∆pilAcomp2, respectively) and the group II strain Aac5 and the corresponding knockout, complementation, and cross-complementation lines (Aac5-∆pilA, Aac5-∆pilAcomp1, and Aac5-∆pilAcomp2, respectively) were grown on King's B plates with ampicillin at 28 • C for 72 h. Twitching motility was assessed based on the visible presence of a thin, light halo around each colony under light microscopy.

pilA Was Involved in Interspecies Competition
Interspecific competition is one of the important biological functions of T6SS in bacteria. When the T4P gene pilA of A. citrulli was deleted, the expression of T6SS genes hcp and vgrG was affected [18]. To verify the impact of pilA gene deletion on the biological function of T6SS, bacterial interspecific competition experiments were conducted. As a weak competitive strain and model strain, E coli was used as the prey strain in this test [18].
E. coli DH5α was co-cultured with each of the eight A. citrulli lines assessed in this

pilA Was Involved in Interspecies Competition
Interspecific competition is one of the important biological functions of T6SS in bacteria. When the T4P gene pilA of A. citrulli was deleted, the expression of T6SS genes hcp and vgrG was affected [18]. To verify the impact of pilA gene deletion on the biological function of T6SS, bacterial interspecific competition experiments were conducted. As a weak competitive strain and model strain, E coli was used as the prey strain in this test [18].
E. coli DH5α was co-cultured with each of the eight A. citrulli lines assessed in this study, namely, the WT, mutant, complementation, and cross-complementation lines from group I (pslb65, ∆pilA-pslb65, pslb65-∆pilAcomp1, and pslb65-∆pilAcomp2, respectively) and group II (Aac5, ∆pilA-Aac5, Aac5-∆pilAcomp1, and Aac5-∆pilAcomp2, respectively). The E. coli and A. citrulli colonies were then counted. E. coli was significantly more abundant (p < 0.05) when co-cultured with pslb65-∆pilA compared to WT pslb65. The complementation line (pslb65-∆pilAcomp1) completely recovered interspecific competitiveness with E. coli, whereas colony counts of the cross-complementation line (pslb65-∆pilAcomp2) were significantly lower than those of the WT or pslb65-∆pilAcomp1 after co-culture with E. coli. The results for Aac5 and derivative strains were similar. E. coli abundance increased significantly (p < 0.05) when it was co-cultured with Aac5-∆pilA compared with the WT; Aac5-∆pilAcomp1 completely recovered interspecific competitiveness with E. coli; and Aac5-∆pilAcomp2 competitiveness was significantly decreased compared to both the WT and Aac5-∆pilAcomp1 (p < 0.05) ( Figure 6). The E. coli and A. citrulli colonies were then counted. E. coli was significantly more abundant (p < 0.05) when co-cultured with pslb65-∆pilA compared to WT pslb65. The complementation line (pslb65-ΔpilAcomp1) completely recovered interspecific competitiveness with E. coli, whereas colony counts of the cross-complementation line (pslb65-Δpi-lAcomp2) were significantly lower than those of the WT or pslb65-ΔpilAcomp1 after coculture with E. coli. The results for Aac5 and derivative strains were similar. E. coli abundance increased significantly (p < 0.05) when it was co-cultured with Aac5-∆pilA compared with the WT; Aac5-ΔpilAcomp1 completely recovered interspecific competitiveness with E. coli; and Aac5-ΔpilAcomp2 competitiveness was significantly decreased compared to both the WT and Aac5-ΔpilAcomp1 (p < 0.05) ( Figure 6).

Effects of pilA on the Expression of T3SS-Related Genes
The relative expression levels of six T3SS core genes (hrpG, hrpX, hrpE, hrcJ, hrcQ, and hrcR) were measured in pslb65-ΔpilA and Aac5-ΔpilA to evaluate the effects of pilA on T3SS by comparing differences between the two groups of strains. In the group I strain pslb65, pilA deletion did not cause significant differences in the expression of six T3SS genes ( Figure 7a); for the group II strain Aac5, hrcJ and hrcR were significantly upregulated in the pilA mutant, whereas the other four genes showed no significant differences (Figure 7b).  The relative expression levels of six T3SS core genes (hrpG, hrpX, hrpE, hrcJ, hrcQ, and hrcR) were measured in pslb65-∆pilA and Aac5-∆pilA to evaluate the effects of pilA on T3SS by comparing differences between the two groups of strains. In the group I strain pslb65, pilA deletion did not cause significant differences in the expression of six T3SS genes ( Figure 7a); for the group II strain Aac5, hrcJ and hrcR were significantly upregulated in the pilA mutant, whereas the other four genes showed no significant differences (Figure 7b).

Effects of pilA on the Expression of T4P-Related Genes
Expression levels of seven T4P-related genes were next analyzed in WT A. citrulli and pilA knockout mutants. Three genes (pilO, pilP, pilZ) were upregulated to varying degrees in pslb65-ΔpilA compared to WT pslb65, while the expression levels of the other four genes showed no significant difference (Figure 8a).  Student's t-test). "ns" indicates that there was no significant difference.
Two genes (pilM and pilE) were significantly downregulated to varying degrees in Aac5-ΔpilA compared to WT Aac5 (p < 0.05), and the other five genes showed no significant differences (Figure 8b).

Effects of pilA on the Expression of T4P-Related Genes
Expression levels of seven T4P-related genes were next analyzed in WT A. citrulli and pilA knockout mutants. Three genes (pilO, pilP, pilZ) were upregulated to varying degrees in pslb65-∆pilA compared to WT pslb65, while the expression levels of the other four genes showed no significant difference (Figure 8a).

Effects of pilA on the Expression of T4P-Related Genes
Expression levels of seven T4P-related genes were next analyzed in WT A. citrulli and pilA knockout mutants. Three genes (pilO, pilP, pilZ) were upregulated to varying degrees in pslb65-ΔpilA compared to WT pslb65, while the expression levels of the other four genes showed no significant difference (Figure 8a).  (a,b). Relative expression levels in (a) the group I strain pslb65-ΔpilA and (b) the group II strain Aac5-ΔpilA compared to the corresponding WT strains. Bars show the mean values from three technical replicates per gene in three independent experiments. Error bars represent standard deviations. * p < 0.05, ** p < 0.01, *** p < 0.001 (Student's t-test). "ns" indicates that there was no significant difference.
Two genes (pilM and pilE) were significantly downregulated to varying degrees in Aac5-ΔpilA compared to WT Aac5 (p < 0.05), and the other five genes showed no significant differences (Figure 8b). Error bars represent standard deviations. * p < 0.05, ** p < 0.01, *** p < 0.001 (Student's t-test). "ns" indicates that there was no significant difference. Two genes (pilM and pilE) were significantly downregulated to varying degrees in Aac5-∆pilA compared to WT Aac5 (p < 0.05), and the other five genes showed no significant differences (Figure 8b).

Effects of pilA on the Expression of T6SS-Related Genes
To evaluate the effects of pilA deletion on the T6SS in A. citrulli, relative expression levels were assessed for 29 T6SS-related genes in pslb65-∆pilA compared to pslb65 and in Aac5-∆pilA compared to Aac5. The tested genes comprised 17 T6SS genes and 12 valineglycine repeat protein G (VgrG)-encoding genes (vgrGs). Because T6SS activation is affected by the concentration of a bacterial suspension, we tested the relative expression of T6SS genes in cultures at an OD 600 value of 1.5.

Discussion
Previous studies have found that group I and group II strains of A. citrulli exhibit differences in pathogenicity, host preference, and genome sequences [5,8,9]. However, there have been no previous reports of the functional differences in a single gene between the two A. citrulli groups. In the present study, the sequence and function of pilA were analyzed in the two groups of strains. Significant differences in the sequences between the two groups were found to correspond to differences in biofilm formation ability, twitching motility, and interspecies competition.
Biofilm formation is a process that allows microorganisms to irreversibly adhere to and grow on a surface by producing extracellular polymers that promote adhesion and matrix formation; this enhances the ability of the bacteria to withstand harsh environmental conditions [37]. The results of this study showed inconsistent effects of pilA deletion on biofilm formation during growth in KB and M9 media. In M9 medium, pilA deletion significantly reduced the capacity for biofilm formation in both strains, consistent with previous reports [28,29]. However, in the nutrient-rich KB medium, the effects of pilA deletion differed between strains. In Aac5, the absence of pilA significantly reduced biofilm formation, but this was not the case for pslb65. Thus, for the group I strain pslb65, nutrient conditions determined the effects of pilA deletion on the capacity for biofilm formation. The relationship between nutrient metabolism and biofilm formation is largely unknown, but it has been suggested that carbon molecules may act as simple nutrients for biofilm-forming bacteria [38]. Alterations in nutrient availability, oxygen depletion, and other environmental factors are also known to affect biofilm dispersion [39]. These factors may be the cause of the observed significant differences in biofilm formation under different nutritional conditions. However, the differences in biofilm formation between pilA deletion mutants in the group I and the group II strain under consistent nutrient conditions may have been caused by group-specific differences in the regulatory pathways associated with pilA.
Previous research found that the pilA and pilO mutant strains of P. syringae pv tabaci 6605 reduced the HR in nonhost Arabidopsis leaves, and qPCR results showed that multiple T3SS-related genes of pilA and pilO deletion mutants were significantly downregulated in expression compared to the WT [27]. However, in this study, we found that there was no significant effect on the HR in nonhost plant tobacco leaves between the pilA deletion mutant of A citrulli and the WT. The qPCR results showed that the expression of T3SSrelated genes in pslb65-∆pilA did not significantly differ compared to the WT, and in Aac5-∆pilA strains, hrcJ and hrcR were significantly upregulated, whereas the other four genes showed no significant differences. This is different from the research results on P. syringae pv tabaci 6605. This indicates that in A citrulli, the effect of pilA on virulence may not be mediated through T3SS, but through alternative pathways. Vfr is a cAMPbinding protein and a transcriptional regulator that can regulate many virulence factors, such as quorum sensing, T4P biogenesis, and T3SS-related genes [40]. Taguchi (2011) believes that there is a global regulatory system between cAMP, Vfr, and T4P that controls bacterial virulence [27]. Through protein sequence alignment, we found that there is a gene homologous to P. aeruginosa Vfr in A. citrulli. Next, we will further investigate the regulatory relationship between vfr and T4P by constructing vfr mutants.
The group I strain pslb65 completely lost its capacity for twitching motility when pilA was knocked out, consistent with previously published results for the group I strain M6 [29]. Surprisingly, the group II strain Aac5 only partially lost its twitching motility in the pilA mutant. To understand the cause of the loss in motility, we identified seven genes that are predicted to have a close regulatory relationship with pilA in the group II strain AAC00-1 using the STRING website (http://version10.string-db.org, accessed on 3 May 2022). We then measured the relative expression levels of those seven genes in pslb65-∆pilA and Aac5-∆pilA compared to the corresponding WT strains. Three genes (pilO, pilP, and pilZ) were upregulated to varying degrees in pslb65-∆pilA, while in Aac5-∆pilA, no genes were upregulated, and two genes (pilM and pilE) were significantly downregulated to varying degrees compared to WT Aac5. This indicates that there were certain differences in the regulation of T4P by the pilA gene between the two groups of A. citrulli.
The T6SS plays important roles in pathogenicity, biofilm formation, bacterial interactions, and other biological functions [41]. There is reportedly a T6SS gene cluster in A. citrulli that contains 17 genes [17]. Hcp is not only a protein that plays a structural role in the T6SS but also an indispensable special effector in the T6SS [42]. VgrG is a versatile protein that is essential to T6SS function and is involved in bacterial competition. There is some evidence that Hcp and T6SS expression is only activated in the A. citrulli strain Aac5 at a cell density of OD 600 = 1.5 [18]. At that cell density, we found that the interspecific competition experiments with E. coli showed no significant differences between the group I strain pslb65 and the group II strain Aac5; however, after pilA deletion, the interspecific competition ability of Aac5-∆pilA was significantly higher than that of pslb65-∆pilA, and complementation with endogenous pilA returned their competitive abilities to WT levels, whereas cross-complementation significantly enhanced the interspecific competition ability of both strains. To further analyze the effects of pilA deletion on T6SS functioning, we analyzed the relative expression of 17 T6SS genes and 12 vgrG genes in pslb65-∆pilA and Aac5-∆pilA at cell densities of OD 600 = 1.5. Interestingly, the experimental results showed that the expression level of the hcp gene in Aac5-∆pilA was significantly downregulated (p < 0.0001), while there was no significant difference in pslb65-∆pilA. Aave_3347 (vgrG gene) also caught our attention, as its relative expression levels were significantly downregulated (p < 0.001) in both pslb65-∆pilA and Aac5-∆pilA. This may be one of the reasons for the decrease in interspecific competition with E. coli. The regulatory relationship between the pilA gene and Aave_3347 needs further research.
T6SS activation in Neisseria cinerea, a commensal of the human respiratory tract, can limit the growth of related pathogens. Custodio et al. (2020) believe that T4P affects T6SSmediated antagonism by constructing and distributing bacteria in mixed microcolonies in this process; in other words, spatial segregation driven by type IV pili dictates prey survival against T6SS assault. In the process of bacterial interspecific competition, the pilA gene is more sensitive to attack by other bacterial T6SSs and can mount an earlier response [43]. Based on comprehensive analysis, we speculate that the main reason for the decrease in pslb65-∆pilA and Aac5-∆pilA in the interspecific competition of E. coli is the loss of spatial isolation caused by the loss of T4P; in addition, key differences in the pilA-mediated regulation of T6SS and vgrG between pslb65 and Aac5 play a secondary role.

Conclusions
In the present study, we characterized the function of pilA in the A. citrulli group I strain pslb65 and in the group II strain Aac5. There were significant differences in phenotype between the two strains when pilA was deleted. Specifically, twitching motility was completely lost in pslb65-∆pilA, whereas Aac5-∆pilA only partially lost motility. Biofilm formation was significantly lower for Aac5-∆pilA than for Aac5, but there were no significant differences in biofilm formation between pslb65-∆pilA and the WT pslb65. The deletion of pilA also significantly reduced the interspecific competitiveness of pslb65-∆pilA and Aac5-∆pilA compared to the corresponding WT strains. qRT-PCR results showed that significantly more genes related to T4P and the T6SS were upregulated, and significantly fewer were downregulated in Aac5-∆pilA compared with pslb65-∆pilA. These results revealed critical differences in pilA function between group I and group II strains, indicating significant intraspecies divergence in T4P regulatory mechanisms.
Supplementary Materials: The following supporting information can be downloaded at https://www. mdpi.com/article/10.3390/microorganisms11071806/s1: Data S1: The sequences of three types of pilA genes in Acidovorax citrulli; Figure S1: Information on primer position, product length, and gene knockout position during the construction of pilA gene deletion mutants and PCR validation process; Figure S2: PCR validation of the mutants and complementary strains; Figure S3: Effects of pilA in inducing hypersensitive responses in nonhost tobacco; Table S1: The bacterial strains and plasmids in this study; Table S2: Sequences and related information of primers used for construction of mutant and complemen-tary strains; Table S3: Information about primers used for qPCR in this study. References [44][45][46] are cited in the supplementary materials.
Author Contributions: Y.Y. designed the study. Y.Y., N.F., P.Q. and W.J. performed the experiments. N.F., L.Y., D.L. and W.G. performed data analyses. Y.Y. and N.F. wrote the manuscript. T.Z. critically reviewed the manuscript. All authors have read and agreed to the published version of the manuscript.