Nutrient Availability and Biofilm Polysaccharide Shape the Bacillaene-Dependent Antagonism of Bacillus subtilis against Salmonella Typhimurium

We aimed to test the competition between the pathogen S. Typhimurium SL1344 and the potential probiotic strain B. subtilis PS-216 in mixed-species biofilms. We tested the effect of nutrient availability on competition in static coculture by utilizing nutrient-rich (tryptic soy broth [TSB]), nutrient-restricted (1/5 TSB), and nutrient-depleted (1/20 TSB) media. Both species were labeled with an antibiotic resistance gene, which allowed their quantification by CFU counting, directly upon mixing at a 1:1 ratio and after 24 h of incubation in TSB medium. The results showed that B. subtilis inhibited S. Typhimurium growth at 1.3 log in nutrient-rich conditions, while no growth inhibition of B. subtilis was observed in coculture (Fig. 1). Nutrient-restricted conditions reduced the growth of both species as well as the antagonistic potential of B. subtilis. Under nutrient-depleted conditions S. Typhimurium fully escaped growth inhibition and even significantly inhibited the growth of B. subtilis (Fig. 1).

Since the spatial organization of a community can strongly influence the level of competition (35), we determined the cell distribution of a fluorescently labeled B. subtilis PS-216 WT (mKate2) and an S. Typhimurium SL1344 (GFP) strain by confocal laser scanning microscopy (CLSM) using the whole-volume 3D microscopy approach. Moreover, the effect of medium dilution on B. subtilis biofilm architecture has not yet been studied. In nutrient-rich monoculture conditions, S. Typhimurium formed submerged biofilms, whereas the B. subtilis strain formed both a submerged biofilm and a pellicle at the air-liquid interface. The spatial organization did not change in coculture conditions, and the submerged B. subtilis biofilm was localized just above the Salmonella biofilm (Fig. 2A). The localization of S. Typhimurium did not change in media with lower nutrient concentrations; however, B. subtilis no longer formed a pellicle and only produced a submerged biofilm, which in coculture was again localized just above S. Typhimurium (Fig. 2B, C). Macroscopic biofilm images support these observations (Fig. S1 in the supplemental material).

Moreover, in nutrient-depleted conditions, the interspecies mixing was more intimate, and a few B. subtilis clusters were even interspersed between S. Typhimurium cell patches, as seen in the orthogonal view of the biofilm representing the first 50 μm of the z-stack (Fig. 2C). This increased mixing is associated with a lower degree of inhibition by B. subtilis and increased inhibition by S. Typhimurium (Fig. 1). Consistently, the S. Typhimurium biofilm was most robust in the nutrient-rich environment, with an average thickness of 37 μm as shown by confocal microscopy (Fig. 2D). S. Typhimurium biofilm thickness decreased in coculture. The greatest effect of interspecies interactions was observed in nutrient-rich medium, whereas in nutrient-depleted conditions, the reduction of S. Typhimurium biofilm thickness in coculture compared to that in monoculture was not significant (Fig. 2D). These results are in line with the reduced inhibition of S. Typhimurium under nutrient-depleted conditions (Fig. 1).

Adhesion to surfaces is the first step in surface colonization and the formation of a mature, resistant biofilm (41). By controlling adhesion, we can control the maturation and dispersal of biofilms. To further test the potential of B. subtilis to disrupt formation of S. Typhimurium biofilms, we cocultured both strains in nutrient-rich conditions and determined the adhesion of S. Typhimurium to the polystyrene surface. This assay confirmed that the B. subtilis strain was able to significantly reduce S. Typhimurium adhesion, probably due to growth inhibition (Fig. 2E).

To test whether the production of antimicrobial compounds underlies the strong inhibition of Salmonella under nutrient-rich conditions, we compared levels of S. Typhimurium inhibition in coculture between B. subtilis strains PS-216 and 168. B. subtilis PS-216 is a natural soil isolate (33), while B. subtilis 168 is a laboratory strain defective in the sfp gene responsible for synthesis of different antimicrobial compounds (42–44). B. subtilis 168 did not inhibit the growth of S. Typhimurium (Fig. 3A), indicating that the antagonism of B. subtilis PS-216 is indeed due to the production of antimicrobial compounds. We then tested several PS-216 mutants defective in either surfactin (ΔsrfAA), plipastatin (ΔppsB), or bacilysin (ΔbacA). However, all tested mutants showed similar levels of Salmonella inhibition as the B. subtilis PS-216 WT (wild-type) strain, suggesting that these antimicrobial compounds do not contribute to S. Typhimurium growth inhibition (Fig. 3A). In contrast, the B. subtilis PS-216 Δpks mutant, which does not produce bacillaene, completely lost its inhibitory effect against the pathogen (Fig. 3A) and consequently its ability to interfere with the adhesion of Salmonella to the polystyrene surface (Fig. 3B). Moreover, when the ΔpksC deletion was complemented by the pksC gene integrated into the sacA locus, the mutant was restored to the wild-type phenotype and could again inhibit S. Typhimurium (Fig. S2). This confirms that bacillaene is the major antagonist of B. subtilis against S. Typhimurium in nutrient-rich conditions.

Next, we explored whether a difference in bacillaene production underlies the reduced inhibition by B. subtilis in nutrient-depleted conditions. Here, we measured the expression of the pks operon via a P_pksC-yfp reporter fusion in nutrient-rich and nutrient-depleted conditions, since it has been previously established that activity of the P_pksC promoter correlates well with bacillaene synthesis (45, 46). In addition, because it has been shown that bacteria can alter the production of antimicrobials in the presence of competitors (45, 47, 48), we compared the expression between monoculture and coculture conditions (Fig. 3C). Overall, the pks operon showed a much stronger activity in nutrient-rich than in nutrient-depleted conditions, possibly explaining the lack of Salmonella inhibition in nutrient-depleted conditions. In nutrient-rich conditions, the level of P_pksC-yfp activity was initially comparable between monoculture and coculture. However, at the 8- to 10-h time period, a dramatic increase in P_pksC-yfp activity was observed in coculture with S. Typhimurium, which was absent in the monoculture. Consistently, we also observed the increase in P_pksC-yfp promoter activity in cocultures at the 16-h time point using flow cytometry; however, P_pksC-yfp activity was no longer increased at 24 h. Possibly, partial lysis of the Bacillus population in stationary phase (49, 50) resulted in the release of yellow fluorescent protein (YFP) into the surrounding medium, which could still be detected by the microplate reader method but not by flow cytometry (Fig. S3). In nutrient-depleted conditions, P_pksC-yfp promoter activity dropped after 12 h in monoculture conditions, whereas in coculture activity remained constant (Fig. 3C). Similarly, P_pksC-yfp promoter activity remained uninduced when the promoter activity under monoculture and coculture conditions was monitored by flow cytometry (Fig. S3). Interestingly, conditioned medium isolated from S. Typhimurium monoculture or coculture grown in nutrient-rich conditions did not induce P_pksC-yfp promoter activity (Fig. S4). These results underscore the importance of interspecies interactions and nutrient availability in regulating bacillaene production.

It was previously reported that biofilm formation could alter the production of bacillaene and mediate the outcome of interspecies interactions (46). The extracellular matrix of the B. subtilis biofilm is composed of polysaccharides, proteins, and nucleic acids (27, 40, 51). The polysaccharide component of the extracellular matrix contributes to pellicle formation and biofilm structure and is synthesized by proteins made from the epsA-O operon (52). To determine how the production of polysaccharides influences competition with Salmonella, we utilized a B. subtilis PS-216 ΔepsA-O mutant which does not produce extracellular polysaccharides and is defective in pellicle formation (32, 53). This PS-216 ΔepsA-O mutant showed deficient pellicle formation in nutrient-rich conditions and only formed a submerged biofilm at the bottom of the well in both monoculture and coculture, resulting in increased mixing with the pathogen (Fig. 4A, Fig. S1). Next, growth inhibition of S. Typhimurium by B. subtilis ΔepsA-O mutant was tested in nutrient-rich and nutrient-depleted media. Although the ΔepsA-O mutant had a slightly lower inhibitory effect on S. Typhimurium growth than the WT strain in the nutrient-rich medium, it was a significantly better inhibitor of Salmonella under nutrient-depleted conditions (Fig. 4B). Inhibition by the PS-216 ΔepsA-O mutant was also dependent on bacillaene, as the B. subtilis PS-216 ΔepsA-OΔpks double mutant completely lost its antagonism in both nutrient-rich and nutrient-depleted conditions (Fig. 4B).

The matrix deficiency of the B. subtilis ΔepsA-O mutant could potentially confer a metabolic advantage to B. subtilis in nutrient-depleted conditions and consequently increase its competitive strength with Salmonella. Therefore, we determined the growth rates of the B. subtilis ΔepsA-O mutant and WT strain. However, there was no difference in growth rates between these two strains in either nutrient-rich or nutrient-depleted conditions (Fig. S5). Thus, increased growth of the B. subtilis ΔepsA-O mutant could not explain the enhanced inhibition of Salmonella in nutrient-depleted conditions.

Subsequently, we compared bacillaene production between the B. subtilis WT strain and the ΔepsA-O mutant by monitoring the activity of the P_pksC-yfp reporter fusion. Overall, the ΔepsA-O mutant showed slightly higher levels of P_pksC activity in nutrient-depleted conditions than the WT strain. The ΔepsA-O mutant did not significantly induce P_pksC activity in the presence of Salmonella in either nutrient-rich or nutrient-depleted conditions. After 15 h in nutrient-depleted conditions, the P_pksC activity in monoculture conditions strongly declined, whereas it remained stable in coculture (Fig. 4C). However, these observations could not be validated by single-cell measurements (Fig. S6). Nevertheless, the fact that the ΔepsA-OΔpks double mutant behaves similarly to the Δpks single mutant in terms of Salmonella inhibition still suggests that the effect of eliminating the extracellular polysaccharide is at least partially mediated through bacillaene. In addition, other factors such as delayed sporulation (Fig. S7) in the ΔepsA-O mutant can further contribute to the mutant’s improved inhibitory efficacy of Salmonella under nutrient-depleted conditions.

In this study, B. subtilis PS-216- and B. subtilis 168-derived strains were used, labeled with a red fluorescent protein, mKate2, the gene for which is linked to a constitutive promoter P_hyspank or P₄₃ integrated in two different loci, namely, amyE::P_hyspank-mKate2 and sacA::P₄₃-mKate2 (Table 1). The B. subtilis mutant strains were obtained by a standard transformation protocol in which the B. subtilis PS-216 or 168 recipient strains were transformed by added DNA (68). Briefly, the recipient strains were grown in modified competence (MC) medium at 37°C shaken at 200 rpm for 5 h, and donor DNA (plasmid or PCR product) was added to the cultures. After 1.5 h of culture incubation under the same incubation conditions, transformants were plated on LB agar plates with the appropriate antibiotics: 10 μg/mL chloramphenicol (Cm), 50 μg/mL kanamycin (Kn), 10 μg/mL tetracycline (Tc), 100 μg/mL or spectinomycin (Sp). The B. subtilis PS-216 Δpks::spec mutant was constructed by amplifying the spectinomycin-inactivated pks gene cluster from the genomic DNA of B. subtilis PSK0178 using the pksX1 and pksX4 primer pair (Table S1) (17). The PCR product was then transformed in different recipient strains such as B. subtilis PS-216 WT, B. subtilis PS-216 amyE::P_hyspank-mKate2 (BM1097), and B. subtilis PS-216 epsA-O::tet amyE::P_hyspank-mKate2 (BM1310), producing the B. subtilis mutant strains BM1875, BM1876, and BM1906, respectively. The B. subtilis PS-216 ΔpksC mutant strain was generated by amplifying the erythromycin-inactivated pksC gene from the genomic DNA of B. subtilis BKE17100 using the 5pL-pksC and 3pR-pksC primers (69) (Table S1). The PCR product of the erythromycin-inactivated pksC gene was then transformed into the recipient strain of B. subtilis PS-216 WT, resulting in B. subtilis BM1957. The B. subtilis pksC complementation mutant was generated by amplifying and the pksC gene from the genomic DNA of B. subtilis PS-216 WT using the primer pair pksC compl-F (HindIII) and pksC compl-R (BamHI) (Table S1). The PCR product was digested with HindIII and BamHI restriction enzymes and ligated into the previously digested plasmid pSac-Cm (70), which allows integration of a gene into the sacA locus. The integration was then verified by sequencing. The obtained plasmid pEM1112 (Table S2) was then transformed into BM1957 to create the B. subtilis BM1959 strain.

To assay bacillaene biosynthetic gene expression, we generated the plasmid pEM1108 (Table S2) carrying a P_pksC-yfp reporter fusion. Briefly, the P_pksC promoter region was PCR-amplified from B. subtilis PS-216 WT using the pC-F(EcoRI) and pC-R(HindIII) primer pair (Table S1). The PCR product was then digested with EcoRI and HindIII restriction enzymes and ligated into the previously digested plasmid pKM3 (71) to obtain the plasmid pEM1108 (Table S2). The plasmid pKM3 was digested with EcoRI and HindIII restriction enzymes to remove the P_spoIIQ promoter region from the original vector. Plasmid pEM1108 was further transformed into B. subtilis PS-216 sacA::P₄₃-mKate2 (BM1629) and PS-216 epsA-O::tet (BM1070) strains, producing the strains B. subtilis BM1884 and BM1899, respectively. The latter strain was then transformed with plasmid pMS17 (32) with a kanamycin resistance cassette to generate the B. subtilis BM1901 strain labeled with mKate2, the gene for which was linked to a constitutive promoter P₄₃ and integrated in the sacA locus. S. Typhimurium SL1344 (WT) (34, 47) was fluorescently labeled via the gfpmut3 gene expressed from a plasmid (72, 73) (Table 1).

To prepare overnight cultures, bacterial strains were grown in tryptic soy broth (Conda, Spain) supplemented with the appropriate antibiotics at 37°C and shaken at 200 rpm for 16 h. The antibiotic concentrations in the medium were as follows: Cm 10 μg/mL, Kn 50 μg/mL, Tc 10 μg/mL, Sp 100 μg/mL, and Amp 100 μg/mL.

The growth of the B. subtilis PS-216 WT strain and PS-216 ΔepsA-O mutant was monitored by measuring the optical density at 650 nm (OD₆₅₀) on a spectrophotometer Spectroquant Prove 100 (Merck, Germany) at 30-minute intervals for up to 8 h. Briefly, overnight cultures of bacterial strains were prepared in TSB and 20-times diluted (1/20) TSB medium and incubated with shaking at 200 rpm for 16 h at 37°C. A total of 1% (V/V) of overnight cultures was transferred to fresh TSB and 1/20 TSB medium and incubated with shaking at 200 rpm and 37°C for 8 h.

To prepare the inoculum, overnight cultures were centrifuged at 10,000 × g for 10 min, supernatants were discharged, and pellets were resuspended in fresh undiluted, 5-times diluted (1/5), or 1/20 TSB medium. S. Typhimurium suspensions were then diluted to OD₆₅₀ ~ 0.1 absorbance units (AU), while suspensions of the different B. subtilis strains were diluted to OD₆₅₀ ~ 0.2 AU to obtain approximately 10⁷ cells/mL.

S. Typhimurium was mixed with different B. subtilis strains at a 1:1 ratio and properly vortexed, and 100 μL of each coculture sample was transferred into the wells of a 96-well microtiter plate with an F-bottom (Cellstar, Greiner Bio-One, Austria) and incubated further for 24 h at 37°C under static conditions. Monocultures were prepared at a 1:1 ratio, with fresh undiluted, 1/5, or 1/20 TSB medium, representing the controls. To estimate the number of cells at the beginning and the end of the experiment in monocultures and cocultures, the complete samples were disrupted by vigorous pipetting and vortexing. Samples were then diluted and plated on LB agar plates with the appropriate antibiotics to determine CFU/mL.

To calculate the inhibition of S. Typhimurium SL1344 GFP by the different B. subtilis strains, cell counts after 24 h of S. Typhimurium growth were used to calculate the log₁₀ of obtained CFU/mL values for monoculture and coculture. The inhibition of S. Typhimurium was then calculated by subtracting the coculture values from the monoculture values for each biological replicate (n = 3). Average and standard deviation are shown in Fig. 4B and Fig. S2.

S. Typhimurium SL1344 GFP was tested for its adhesive potential to abiotic polystyrene surfaces by determining the CFU/mL of cells that adhered to the surface in monoculture and coculture with different B. subtilis strains. The inoculum was prepared as described above for static cultures and the protocol was carried out as described previously (21). Briefly, after 24 h of incubation at 37°C in a 96-well microtiter plate, the whole sample was removed from each well by pipetting and washed 3 times with 100 μL sterile phosphate-buffered saline (PBS). At the end, 100 μL PBS was left in the well and cells were detached from the surface by sonication in a water bath for 10 min at low frequency (33 kHz) and 2 min at high frequency (40 kHz, 120 W power; Asonic, Slovenia) at room temperature and plated on LB agar plates with ampicillin to determine the CFU/mL of attached S. Typhimurium cells.

Monocultures and cocultures of B. subtilis (labeled with mKate2) and S. Typhimurium (labeled with GFPmut3) were prepared as described previously and analyzed after 24 h of incubation using CLSM with a slightly modified protocol described by Erega et al. (21). Briefly, an inverted confocal laser scanning microscope (Axio Observer Z1, LSM800; Zeiss, Germany) was used to investigate the spatial distribution of the two species and the architecture of the mixed- and monospecies biofilms. Excitation of red fluorescent protein mKate2 was performed at 589 nm with an argon laser and emission was recorded between 580 and 700 nm. Excitation of green fluorescent protein GFPmut3 was performed at 488 nm and emission was recorded between 400 and 580 nm. The GaAsP detector gain and laser intensities were 800 V and 4.5% for mKate2 and 650 V and 0.7% for GFP, respectively. The pinhole size was 55 μm for both lasers used. For imaging the whole sample in the well in monocultures and cocultures, the thickness of one slice in the z-stack was 10 μm, while the thickness of one slice in the z-stack was 3.27 μm when determining the thickness of S. Typhimurium biofilms. The wells in the microtiter plate were scanned using a 20×/0.4 NA objective and the captured images were 1,180 × 1,180 pixels in size with 8-bit color depth.

To determine spatial distribution and biofilm thickness, analyses of monocultures and cocultures were performed in the same biological replicate with at least two technical repetitions. Three biological replicates were performed time-independently.

For image processing, Zen 2.3 Software (Carl Zeiss) was used, and the image noise was reduced by applying a single pixel filter (threshold = 1.5). After applying the single-pixel filter, the Zen imaging software was used to assemble the 3D and orthogonal views and the results were presented as one of the most representable images. The thickness of the submerged S. Typhimurium biofilms was determined as the total sum of slices with visible GFP signal. To account for technical error, we deducted the first- and last-detected slices from the total sum of slices.

Strains were prepared as described above for static cultures. Here, 100-μL volumes of monocultures and cocultures in TSB and 1/20 TSB medium were allocated to 3 technical replicates in the wells of a 96-well, black transparent-bottomed microtiter plate (Cellstar, Greiner Bio-One, Austria). The microtiter plate was sealed with micropore tape to minimize evaporation during incubation and then incubated statically for 24 h at 37°C in a Cytation 3 imaging reader (BioTek, USA). To monitor P_pksC-yfp expression, we measured the YFP fluorescence intensity with excitation at 500 nm, emission at 530 nm, and the gain set at 100. The fluorescence intensity of mKate2 (red fluorescent protein) with excitation at 570 nm, emission at 620 nm, and the gain set at 100 was determined to monitor expression of the constitutively expressed P₄₃-mKate2. Both fluorescence intensities (YFP and mKate2) were measured every half-hour for 24 h. To calculate the promoter activity in each biological replicate, the fluorescence intensity of the unmarked monoculture was deducted from that of the monoculture carrying the fluorescent reporter YFP or mKate2, and the fluorescence intensity of the unmarked coculture was deducted from that of the coculture carrying the fluorescent reporter YFP or mKate2. Next, the mKate2 fluorescence intensity of the monoculture, representing the maximum, was set at 1. The remaining mKate2 fluorescence intensities, measured at different time points of each experimental setting, were divided by the highest mKate2 fluorescence intensity value to obtain relative constitutive fluorescence intensity for monocultures and cocultures. To compare the P_pksC promoter activity between monocultures and cocultures, the deducted YFP fluorescence intensity was divided by the relative fluorescence intensity of constitutively expressed mKate2 to normalize the promoter activity per biomass.

To measure the expression of pksC at the single-cell level, static monocultures and cocultures were prepared in TSB and 1/20 TSB medial as described above. Three biological repeats were prepared per condition, and each biological repeat consisted of two duplicate cultures. pksC expression was measured after 16 h in the first duplicate and after 24 h in the second duplicate. Here, biofilms were first disrupted by vigorous pipetting and vortexing. Afterwards, 30,000 cells per sample were measured using a CytoFLEX Flow Cytometer (Beckman Coulter). Gene expression was analyzed using the CytExpert software. First, singlets were identified based on forward and side scatter. Second, the B. subtilis cells were identified based on the red fluorescence provided by the constitutively expressed P₄₃-mKate2. Next, the expression of the pksC was quantified by measuring the YFP fluorescence intensity (excitation 488 nm, emission 530 nm). To account for any background fluorescence, the measured fluorescence level was subtracted by the average yellow fluorescence intensity measured in B. subtilis strains which only carried P₄₃-mKate2.

To study the effect of S. Typhimurium monoculture and coculture conditioned medium (CM) on P_pksC-yfp expression, CM was obtained by first preparing S. Typhimurium and B. subtilis strains in TSB medium as described above and transferring 2.82-mL samples into a 6-well plate (Brand, Germany) to achieve the same surface-to-volume ratio as in the 96-well microtiter plate. After 24 h of static incubation at 37°C, samples were centrifuged at 10,000 × g for 10 min and the isolated supernatants were passed through filters with 0.2-μm pores to obtain CM. CM was then mixed with B. subtilis at a 1:1 (V/V) ratio, transferred into the wells of the 96-well, black transparent-bottomed microtiter plate, and incubated further for 24 h at 37°C under static conditions in a Cytation 3 imaging reader to monitor P_pksC-yfp expression.

All experiments were performed in at least three independent biological replicates, each represented by at least 3 technical repetitions. All the results are shown as mean values and error bars represent standard deviation. Statistical analysis of the data using a two-sample Student’s t test was performed in OriginPro.