The bacterial strains and plasmids used in this study are described in Supplementary Table S1. B. subtilis strains were routinely grown at 37°C in Luria-Bertani (LB) medium. For assays of biofilm formation, MSgg medium was used. The recipe for MSgg is as follows: 5 mM potassium phosphate (pH 7.0), 100 mM MOPS (morpholine propane sulfonic acid) (pH 7.0), 2 mM MgCl₂, 700 μM CaCl₂, 50 μM MnCl₂, 50 μM FeCl₃, 1 μM ZnCl₂, 2 μM thiamine, 0.5% glycerol, 0.5% glutamic acid, 50 μg/mL tryptophan, 50 μg/mL threonine, and 50 μg/mL phenylalanine (Branda et al., 2001). Escherichia coli DH5α was used as a host for molecular cloning and grown at 37°C in LB medium. A. citrulli MH21 was grown at 28°C in LB medium (Ren et al., 2014). When required, antibiotics were added at the following concentrations for growth of B. subtilis: 5 μg/mL of chloramphenicol, 10 μg/mL of tetracycline. For growth of E. coli and A. citrulli MH21, antibiotics were added at the following concentrations: 100 μg/mL of ampicillin and 50 μg/mL of ampicillin, respectively.

The general methods for molecular cloning were performed according to standard procedures (Sambrook and Russell, 2001). Restriction enzymes and other enzymes for molecular cloning were all purchased from Takara Co., Ltd. (Dalian, China). PCR products were purified with a PCR purification kit (Tiangen Biotech Co., Ltd., Beijing, China) and used according to the manufacturer’s instructions. The plasmids were isolated using a plasmid mini-prep kit purchased from Bioteke Bio Solutions Co., Ltd. (Beijing, China). E. coli DH5α cells were transformed by heat shock transformation. B. subtilis cells were transformed by electroporation using a Gene-Pulser (Bio-Rad, MicroPulser, United States), as recommended by the manufacturer. Oligonucleotide primer synthesis and DNA sequencing services were performed at Tsingke Biological Technology Co., Ltd. (Beijing, China).

The antagonistic activity of B. subtilis 9407 against A. citrulli MH21 was roughly analyzed as previously described (Wu et al., 2015), with some modifications. Fresh A. citrulli MH21 plates were prepared for the assay. When the concentration of A. citrulli MH21 grown in LB medium at 28°C was up to 4 × 10⁷ CFU/mL, 5 mL bacteria suspension was mixed with 300 mL melting LB agar and cooled below 60°C to prepare the plates. The 1 μL of an overnight culture of B. subtilis 9407 was spotted onto the surface of the plate which was then incubated at 28°C for 48 h to observe the growth inhibition effect. The diameters of inhibition zones were then measured and recorded. The other plant pathogens tested in this study were supplied by the Department of Plant Pathology in the College of Plant Protection at China Agricultural University, China.

To analyze the antibacterial activity of the extraction of lipopeptides from B. subtilis 9407, 10 μL of the appropriate dilutions of lipopeptide crude extracts were loaded into wells punched in A. citrulli MH21 plates prepared by the method described above. The plates were incubated at 28°C for 48 h and the antibacterial abilities of the lipopeptide crude extracts were assessed by observing the resulting inhibition zones.

The inhibiting activity of ΔsrfAB and ΔppsB against A. citrulli MH21 was also tested by spotting bacterium on A. citrulli MH21 plates prepared by the method described above. B. subtilis 9407 used as a control was also spotted. Then the plates were incubated at 28°C for 48 h to observe the growth inhibition effect.

The inhibiting activity of the extraction of lipopeptides from wild-type strain 9407, ΔsrfAB and ΔppsB were also tested by spotting 10 μL of lipopeptide crude extracts into wells punched in A. citrulli MH21 plates according to the method described above.

The minimum inhibitory concentration (MIC) of lipopeptide crude extracts from B. subtilis 9407 against A. citrulli MH21 was determined as previously described (Bais et al., 2004). In brief, A. citrulli MH21 was first grown in 5 mL LB broth to 4 × 10⁵ CFU/mL, and 190 μL of culture was added to 96-well microtiter plates. Then, 10 mL of serial 2-fold dilutions of lipopeptide crude extracts was mixed with culture in 96-well plates. Ten microliters of the methanol was used as control. The MIC was visually defined as the lowest concentration of an antibiotic that completely inhibited cell growth after incubation for 24 h at 28°C. All susceptibility trials were conducted in triplicate.

The MIC of surfactin standard for A. citrulli MH21 was determined according to the method described above. Four microliters of the serial 2-fold dilutions of commercial surfactin was mixed with 196 mL of A. citrulli MH21 culture in 96-well plates. Commercial surfactin (Sigma–Aldrich, St. Louis, MO, United States) was dissolved in dimethyl sulfoxide (DMSO). Four microliters of the DMSO was used as control.

The srfAB marked deletion mutant was constructed using the temperature-sensitive suicide plasmid pMAD as described previously (Arnaud et al., 2004). A tetracycline resistance gene (Tet) was amplified by polymerase chain reaction (PCR) from plasmid pGFP78 (Gao et al., 2015) using primer pair Tet-F and Tet-R (All primer sequences are shown in Supplementary Table S2). The resulting fragment was digested with SpeI/PstI and cloned into pEBS (Wang et al., 2007), which had also been digested with SpeI/PstI, generating pEBST. Regions that were upstream and downstream of the srfAB gene were amplified from B. subtilis 9407 genomic DNA using the primer pairs srfAB-Up-F/srfAB-Up-R and srfAB-Dn-F/srfAB-Dn-R, respectively. These fragments were digested with SalI/PstI and SpeI/SacI, respectively, and were cloned into the pEBST plasmid to create pEBST-srfAB. The Up-Tet-Dn fragment was amplified from pEBST-srfAB using the primer pair srfAB-F and srfAB-R. The resulting fragment was digested with BglII/MluI and cloned into pMAD, which had also been digested with BglII/MluI, generating pMAD-srfAB. The pMAD-srfAB plasmid was subsequently mobilized into B. subtilis 9407 by electroporation. Transformants were obtained after incubation at 30°C for 2 days on LB plates containing erythromycin and X-Gal (40 μg/mL). The allelic replacement of pMAD-srfAB in B. subtilis 9407 was performed according to the published protocol (Arnaud et al., 2004). Erythromycin-sensitive clones were isolated, and the mutants were identified by PCR with primer pair srfAB-F and srfAB-R and subsequently confirmed by Sanger sequencing.

Lipopeptide extraction was performed as previously described (Chen et al., 2016). In brief, after cultivating the cells in 50 mL Landy medium at 30°C for 72 h, the cell-free supernatant was obtained by centrifugation at 6,000 × g for 10 min and filtration using a bacterial filter (φ = 0.22 μm). The column (Sigma Amberlite) containing 6 g of XAD16 adsorbent resin (Sigma–Aldrich, St. Louis, MO, United States) was washed with 50 mL of deionized water to remove salts. The cell-free supernatant was passed through the XAD16 resin column, washed with deionized water and eluted with 14 mL of 100% methanol. The lipopeptide crude extracts were dissolved in 1 mL of methanol, followed by drying with a rotary evaporator.

The samples were run on an RP-HPLC equipped with Waters Sunfire C18 columns (5 μm, 4.6 × 150 mm) at room temperature, with a flow rate of 1.0 mL/min. The mobile phase consisted of 0.1% trifluoroacetic acid (TFA) in HPLC grade water (solvent A) and 0.1% TFA in HPLC grade acetonitrile (solvent B). The elution was monitored by determining the absorbance at 214 nm. Commercial surfactin (Sigma–Aldrich, St. Louis, MO, United States) was used as standards.

Quattro Premier XE tandem quadrupole mass spectrometer (Waters) was used for LC-MS/MS analysis. The Waters Sunfire C18 column (4.6 × 150 mm, 5 μm) was used for liquid chromatography at 1.0 mL/min flow rate, at 25°C. The mobile phase consisted of solvent A and solvent B. Solvent A was water, eluent B was acetonitrile (ACN), both containing 0.1% formic acid (FA). The injection volume was 10 μL of surfactin standard (1 mg/mL) or lipopeptide crude extracts from B. subtilis 9407. The elution was monitored by determining the absorbance at 214 nm. The mass spectral was analyzed in both the negative ion and positive ion mode (ESI+/ESI-), the other parameters were as follows: 120°C source temperature; voltages were 3.2 kV for the capillary and 30 V for the cone voltage, 350°C desolvation temperature. Surfactin standard was purchased from Surfactin (Sigma–Aldrich, St. Louis, MO, United States).

The biofilm formation was analyzed in MSgg medium by using the method described previously (Yan et al., 2016). The wild-type and ΔsrfAB strains were first grown in 5 mL LB broth to late exponential growth phase (OD₆₀₀ = 1.0), and 4 μL of culture was added to 4 mL of MSgg medium (a 1000-fold dilution) in 12-well St. microtiter plates and incubated statically at 25°C for 72 h.

In the experiments for the restoration of biofilm formation, 2 μL of the appropriate dilutions of commercial surfactin was mixed with 4 mL of MSgg medium in 12-well plates prior to inoculation with ΔsrfAB. Commercial surfactin (Sigma–Aldrich, St. Louis, MO, United States) was dissolved in DMSO.

Swarming motility assays of wild-type B. subtilis 9407 and ΔsrfAB were performed according to Chen et al. (2012). In brief, 5 mL LB liquid cultures were prepared with shaking (200 rpm) at 37°C to mid-log phase, 1 mL of cells were collected by centrifugation at 6,000 × g for 5 min, washed with phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 2 mM KH₂PO₄), and resuspended in 100 μL PBS. The swarming agar plates (LB solidified with 0.7% agar) were dried for 20 min in a laminar flow hood, centrally inoculated with 3 μL of the cell suspension, dried for another 10 min and incubated at 30°C for 6.5–7 h (the surface of the plates inoculated with wild-type cells had almost been fully covered by the swarming cells). Afterward, the swarming plates were removed to a laminar flow hood, dried for 1 h and incubated at room temperature for another 12 h to allow cell growth in order to clearly visualize the swarming zone. The diameter of the swarming zone was measured.

For extracellular complementation experiments, 2 μL of the appropriate dilutions of commercial surfactin (dissolved in DMSO) was spotted at the center of swarming agar plates and dried in a laminar flow hood before the motility assay.

B. subtilis 9407 and ΔsrfAB strains were marked with resistance gene using the shuttle vector pC-1 containing a chloramphenicol resistance gene. The plasmid pC-1 was transformed into B. subtilis 9407 and ΔsrfAB by electroporation. Transformants were selected with chloramphenicol and then identified by PCR with primer pair pC-1-F and pC-1-R.

The melon (Cucumis melo) was used for colonization assays. Melon seeds were soaked in the 55°C water for 30 min, placed between the wet gauze, and incubated at 28°C for 36 h to germinate. The wild-type and ΔsrfAB strains were grown in nutrient broth (NB) in a shaker at 30°C with 160 rpm for 48 h. The cells were harvested by centrifugation at 6,000 × g for 10 min and adjusted with PBS buffer (pH 7.0) to obtain the desired bacterial concentration (10⁷ CFU/mL). The germinated melon seeds were soaked in the bacterial suspension at room temperature for 30 min with gentle agitation. Seeds soaked in PBS buffer alone were used as the controls. After that the treated seeds were air-dried and sown in 600 mL black plastic pots (six seeds per pot) filled with sterile soil and vermiculite in a ratio of 2:1. Three pots were used for each replicate and three replicates were used for each bacterial strain. The pots were placed in a growth chamber at 28°C with a 16 h photoperiod and 60% humidity. Samples were collected at 0, 5, 10, 15, and 20 days post inoculation. To assays of cell colonization on roots and leaves, 0.1 g fresh weight of the melon roots and leaves were collected. Samples were surface sterilized by 3 min in sodium hypochloride (10% active chlorine) and washed three times with sterile water for at least 5 min each. The surface-sterilized samples were then disrupted in a sterile mortar and pestle. The suspensions were diluted by 10-fold serial dilutions in sterile water, and 100 μL of each diluted suspension was plated on LB agar plates supplemented with chloramphenicol (5 μg/mL) and then incubated at 37°C for 12 h. Bacterial colony-forming units on each plate were counted. The experiments were repeated three times.

Melon was used to evaluate the biocontrol activity of wild-type strain 9407 and ΔsrfAB against A. citrulli MH21. Melon seeds were pre-germinated as described above. The bacterial suspensions of B. subtilis 9407 and ΔsrfAB were prepared by the method described in Section “Colonization Assay.” The germinated seeds were soaked in the bacterial suspension at room temperature for 30 min with gentle agitation and then air-dried. Seeds soaked in PBS buffer alone were used as the controls. Eight treated seeds were sown in 600 mL black plastic pots filled with soil and vermiculite in a ratio of 2:1. The pots were placed in a greenhouse with the following conditions: 28–30°C, 60% humidity, and 16 h of light alternating with 8 h of darkness. A. citrulli MH21 was cultured at 28°C for 48 h in LB broth. The cells were harvested by centrifugation at 5,000 × g for 15 min and the bacterial suspension was adjusted with PBS to 10⁸ CFU/mL. After sown for 3 days, both sides of the leaves were sprayed with bacterial suspension of A. citrulli MH21.

The seedlings were evaluated for BFB severity daily based on the disease index, as described previously (Bahar et al., 2009). The disease index of each leaf was rated using a scale of 0–6, where 0, no symptom; 1, 10% or less necrotic lesions on the leaves; 2–5, 11–25%; 26–50%; 51–75% and 76–90% necrotic lesions on the leaves, respectively; and 6, >90% necrosis of the leaves. The disease severity and biocontrol efficacy were calculated as follows:

Disease severity (%) = Σ the number of diseased leaves in each grade × grade/(total number of leaves investigated × the highest disease index) × 100.

Biocontrol efficacy (%) = (incidence rate in the control - incidence rate in the Bacillus-treated group)/ incidence rate in the control × 100.

Three pots were used for each replicate, and the values were recorded as the means of three replicates for each treatment. The experiments were repeated three times.

The data were statistically analyzed using SPSS software 20.0. Student’s t-tests were used to determine the significant differences.

To investigate the antibacterial activity of B. subtilis 9407 against A. citrulli MH21, a dual culture assay was conducted. Two days after inoculation, the inhibition zone of B. subtilis 9407 was 18.1 mm, suggesting that B. subtilis 9407 could observably inhibit the growth of A. citrulli MH21 (Figure 1). Moreover, B. subtilis 9407 showed significant antagonistic activity in vitro toward other pathogens that cause plant diseases in tomato, oilseed rape, Chinese cabbage, potato and other plants (Table 1).

Next, we evaluated the biocontrol ability of B. subtilis 9407 against BFB caused by A. citrulli MH21 on melon under greenhouse conditions. We performed biocontrol assays using the B. subtilis 9407 strain pretreated seeds and inoculated A. citrulli MH21 3 days after planted (see “Materials and Methods”). Four days after A. citrulli MH21 inoculation, the disease severity of melon seedlings that were pre-treated with PBS (control) was 43.52% (Table 2). The disease severity of melon seedlings that were pre-treated with B. subtilis 9407 was 16.67%, for a noticeable reduction compared with the PBS treated control. Accordingly, the efficacy of B. subtilis 9407 in controlling the BFB caused by A. citrulli MH21 up reached 61.7%. Seven days post inoculation, the biocontrol efficacy of B. subtilis 9407 against BFB was 57.8% (Table 2). These results indicated that B. subtilis 9407 is a potential biological control agent for efficiently controlling BFB.

To identify the primary antibacterial compound of B. subtilis 9407 against A. citrulli, the antibacterial activity of lipopeptide crude extracts from B. subtilis 9407 was tested against A. citrulli MH21 as described in Section “Materials and Methods.” Lipopeptide crude extracts from B. subtilis 9407 showed antibacterial activity against A. citrulli MH21, at the dilution rate from 0- to 10-fold (Figure 2). The MIC of lipopeptide crude extracts from B. subtilis 9407 for A. citrulli MH21 was determined to be 40 times dilution of lipopeptide crude extracts. These results indicated that lipopeptide crude extracts from B. subtilis 9407 showed strong antibacterial activity against A. citrulli MH21.

To further identify the primary antibacterial compound of B. subtilis 9407 against A. citrulli, we decided to mutate candidate genes for biosynthesis of selected lipopeptide, which were reported to show antibacterial activity. The selected candidate genes were srfAB and ppsB, responsible for the synthesis of surfactin and fengycin, respectively (Zeriouh et al., 2014; Fan et al., 2017). To generate a srfAB marked deletion mutant, the temperature-sensitive suicide plasmid pMAD-srfAB was constructed (see “Materials and Methods”) to disrupt the open reading frame of the srfAB gene, abrogating the production of surfactin. The surfactin synthesis abilities of the wild-type B. subtilis 9407 and ΔsrfAB were further detected by RP-HPLC (Figure 3). It was found that wild-type B. subtilis 9407 produced surfactin, while ΔsrfAB completely lost the ability to produce surfactin. In our previous work, we confirmed that the ΔppsB was phenotypically characterized as non-fengycin producer (Fan et al., 2017). We tested the antibacterial activity of the mutants against A. citrulli MH21. Compared with the wild-type B. subtilis 9407, ΔppsB showed a reduced antibacterial activity against A. citrulli MH21. However, ΔsrfAB showed no clear zone of inhibition of the growth of A. citrulli MH21 (Figure 4). After that, lipopeptide crude extracts were subjected to analyze the antibacterial activity in vitro. Consistent with our hypothesis, lipopeptide crude extracts from ΔsrfAB showed almost no inhibition of A. citrulli MH21 (Figure 5). These results suggested that surfactin is the primary active compound to exert the inhibitory effect of B. subtilis 9407 against A. citrulli MH21.

To further characterize surfactin produced by B. subtilis 9407, the complete surfactin gene cluster of B. subtilis 9407 was aligned with that in B. subtilis subsp. subtilis str. 168, B. subtilis subsp. subtilis 6051-HG and B. subtilis SG6, which were reported to produced surfactin A (Aleti et al., 2015, 2016). It was found that the complete surfactin gene cluster of B. subtilis 9407 displays 98.99% sequence identity to B. subtilis subsp. subtilis str. 168 and B. subtilis subsp. subtilis 6051-HG and 98.57% to B. subtilis SG6. AntiSMASH analysis of the surfactin cluster of B. subtilis 9407 predicted a lipopeptide sequence of _L-Glu-_L-Leu-_D-Leu-_L-Val-_L-Asp-_D-Leu-_L-Ile, namely surfacrin A (Figure 6). Moreover, lipopeptide crude extracts from B. subtilis 9407 and commercial standard for surfactin, which is composed of a macrolide containing the heptapeptide Glu-Leu-Leu-Val-Asp-Leu-Leu (Lim et al., 2005; Aleti et al., 2016) were subjected to LC-MS/MS analysis. The LC-MS/MS analysis of B. subtilis 9407 lipopeptide crude extracts showed a series of four peaks with identical mass to surfactin standard (Figure 7). For each of the four compounds, the LC-MS/MS spectra with the same precursor m/z exhibited the quite similar fragmentation behavior was observed in the surfactin standard (Figure 8). These results indicated that the surfactin produced by B. subtilis 9407 is a mixture of C13- to C16-surfactin A.

In order to determine the antibacterial activity of surfacrin, the MIC of surfactin standard was tested. It was found that the MIC of surfactin standard against A. citrulli MH21 was 100 μg/mL. Surfactin standard also showed antibacterial activity against A. citrulli MH21 on LB agar (Figure 9). These results suggested that surfactin standard has antibacterial activity against A. citrulli MH21.

Previous works have shown that surfactin triggers biofilm formation in B. subtilis (Romero, 2013; Zeriouh et al., 2014; Luo et al., 2015). We therefore asked whether this lipopeptide would also have a similar role in biofilm formation of B. subtilis 9407. We compared the biofilm formation phenotype of ΔsrfAB and wild-type. The results showed that wild-type B. subtilis 9407 formed wrinkly floating pellicles at the liquid-air interface of liquid cultures in MSgg (Figure 10). In contrast, the ΔsrfAB formed thinner pellicles in MSgg (Figure 10). Next, we tested whether the commercial surfactin could rescue the biofilm formation. We found that biofilm formation of ΔsrfAB was restored in the presence of exogenously supplemented surfactin at 10 μg/mL (Figure 10). Therefore, our evidence suggested that surfactin production is important for biofilm formation in B. subtilis 9407.

Swarming motility is an important mechanism for bacterial colonization and a type of bacterial social movement related to surfactin production (Kearns and Losick, 2003). To analyze swarming motility, cell suspension was spotted at the center of the swarming agar plates, and the diameter of the swarming zone was measured. The wild-type strain showed excellent swarming motility and the plate almost fully covered by the swarming cells. While, the ΔsrfAB showed a clear defect in swarming motility, with a 22.5 mm diameter of the swarming zone (Figures 11A,B). Based on the results of our external complementation experiments, where commercial surfactin rescued biofilm formation (Figure 10), we decided to analyze the effect of surfactin in the restoration of swarming motility by ΔsrfAB. It was found that 10 μg/mL of commercial surfactin was necessary to restore swarming motility of ΔsrfAB (Figures 11A,B). The above results indicated a significant role of the surfactin in swarming motility of B. subtilis 9407.

The implication of surfactin in the biofilm formation and swarming motility of B. subtilis suggested that this lipopeptide might have a role in the efficient colonization of melon leaves and roots. To test this hypothesis, the population dynamics of the wild-type and ΔsrfAB on melon leaves and roots were evaluated over time. The results showed that wild-type and ΔsrfAB could be isolated from the seedling tissues at 0, 5, 10, 15, and 20 days post inoculation, although the cell population of tested strains in the leaves and roots decreased continuously post inoculation (Tables 3, 4). The population of wild-type strain was higher in roots and leaves at each testing time point, where the highest level was reached at 5 days post inoculation, and the population was at a level of 3.56 × 10⁴ CFU/g (fresh weight) in the leaves (Table 3) and 7.44 × 10⁴ CFU/g (fresh weight) in the roots (Table 4). In contrast, the bacterial numbers of ΔsrfAB were significantly lower than wild-type strain, and at 5 days post inoculation, the population was 1.67 × 10⁴ CFU/g (fresh weight) in the leaves (Table 3), 1.89 × 10⁴ CFU/g (fresh weight) in the roots (Table 4). Similar results were obtained at 10, 15, 20 days post inoculation (Tables 3, 4). In comparison with the wild type, ΔsrfAB showed a two- to four-fold reduction and three- to ten-fold reduction in the number of leaves-colonizing cells and root-colonizing cells after inoculation, respectively. In general, the results strongly suggested that surfactin produced by B. subtilis 9407 affects the efficient colonization on melon leaves and roots.

To examine whether surfactin is responsible for the biocontrol of B. subtilis 9407 against BFB in vivo, the wild-type B. subtilis 9407 and ΔsrfAB were compared with the disease severity to evaluate the biocontrol activity. Three days after A. citrulli MH21 inoculation, the disease severity of melon seedlings that were pre-treated with PBS (control) was 58.52% (Figure 12). Treatment of melon seedlings with cells of the wild-type strain reduced the severity of BFB low to 20.37%. However, treatment with same amount of the ΔsrfAB strain demonstrated a disease severity as high as 56.67%, showed no obvious differences compared with controls (Figure 12). Accordingly, ΔsrfAB lost the capability of controlling BFB. Similar results were obtained 5 and 7 days post inoculation (Figure 12). These results demonstrated that surfactin produced by B. subtilis 9407 plays a major role in suppressing A. citrulli-induced BFB.