S. cerevisiae single-gene deletion strain and BY4743 parent strain were obtained from YKOC, which was purchased from Invitrogen in 2014. All strains were stored at −80°C in a preservation solution containing 20% glycerol (v/v) (Sigma-Aldrich). Before the experiment, the strains were inoculated twice on Yeast Extract Peptone Dextrose (YPD)ager plates (yeast extract 1%, peptone 2%, glucose 2%, and agar 2%) (Qingdao Hope Bio-technology Company) and incubated at 30°C. Single colonies were inoculated into the YPD broth (2% tryptone, 1% yeast extract, and 2% glucose) and cultured overnight at 30°C, 200 r/min. S. boulardii CNCM I-745 (imported drug registration No. s5288500) was ordered in French Encyclopedia Pharmaceutical Factory; E. coli ATCC25922, E. coli (mcr-1) 12-2, S. aureus ATCC29213, S. aureus ATCC25923, MRSA 1668, Pseudomonas aeruginosa ATCC27853, P. aeruginosa 1554, Klebsiella pneumoniae ATCC700603, K. pneumoniae ATCC1706, K. pneumoniae ATCC1705, K. pneumoniae 2118, Acinetobacter baumannii 21-1 and Salmonella Typhimurium SL1344 were donated by the Laboratory of Affiliated Hospital of Xuzhou Medical University. Lactobacillus johnsonii D-SM 10553 freeze-dried powder (SHBCC, China) was solubilised with 0.5 mL of de Man Rogosa and Sharpe (Falsen et al., 1999) (MRS) medium (Qingdao Hope Bio-technology Company), then the bacterial solution was applied to MRS solid plates, and single bacterial colonies appeared after about 24 h. Subsequently frozen stocks of L. johnsonii (in MRS medium with 20% glycerol) were prepared, stored at -80°C for further experiments. E. coli, S. aureus, P. aeruginosa, K. pneumoniae, A. baumannii and S. Typhimurium were incubated in Luria-Bertani (LB)broth (Tryptone 10.0 g, Yeast Extract 5.0 g, NaCl 10.0 g per litre) (Shanghai Sangon Biotech Company) at 37°C overnight. L. johnsonii was incubated in MRS broth overnight at 37°C under anaerobic conditions.

The S. cerevisiae single-gene deleted strain and the BY4743 parent strain were inoculated into YPD broth and cultured at 30°C for 16 h in a shaker at 200 rpm. After incubation, the samples were centrifuged at 4000×g at 4°C for 10 min. Then, the supernatants were filtered by a syringe filter (0.22-μm pore size), and then stored at -80°C for later use, in which CFS of BY4743 was used as a control.

CFS with antibacterial activity was screened by 96-well plate method. A volume of 90 μL of CFS of deletion strain was mixed with 10 μL of bacterial liquid (1×10⁸ CFU/mL) in each well of 96-well plate, and CFS of BY4743 was used as control group. The pathogenic bacteria used in this experiment are E. coli ATCC25922, E. coli (mcr-1) 12-2, S. aureus ATCC29213 and MRSA 1668. The well plates were cultured at 37°C for 18-24 h, and the absorbance at 600 nm was measured by enzyme-labelled instrument. The ratio of OD value of each well divided by OD value of the control group was regarded as its growth rate, and the strains with inhibition rate (1- growth rate) greater than 80% were regarded as having antibacterial activity.

The MIC method was employed to assess the antibacterial activity of CFS of BTS1-KO S. cerevisiae. Pathogenic bacteria cultured overnight were diluted in the Mueller Hinton (MH) broth medium (Qingdao Hope Bio-technology Company). Various concentrations of CFS and bacterial suspensions were added to 96-well plates, resulting in a final volume of 100 μL per well and a bacterial concentration of 1×10⁶ CFU/mL, compared to wells without CFS. The pathogenic bacteria included E. coli ATCC25922, E. coli (mcr-1) 12-2, S. aureus ATCC29213 and MRSA 1668. Then the plates were cultured at 35°C for 24 h, and the lowest CFS concentration to inhibit bacterial growth or inhibit 80% of bacterial cells as MIC.

The antimicrobial activity of CFS against various bacterial strains was determined according to the screening method, and the inhibitory activity of CFS of S. boulardii and L. johnsonii s was used as a control.

Overnight cultures of pathogenic bacteria were diluted in LB broth. Experimental groups were prepared by mixing bacterial suspensions with 90% CFS, 50% CFS or 25% CFS, while the control group was mixed with YPD broth, and a blank control group was mixed with sterile water. The pathogenic bacteria were E. coli ATCC25922, E. coli (mcr-1) 12-2, S. aureus ATCC29213 and MRSA 1668. Initial optical density at 600 nm (OD₆₀₀) was measured using an enzyme-labelled instrument (Thermo Fisher Scientific) and then cultured at 37°C and 200 rpm. Samples of 100 μL were taken every 2 h to determine OD₆₀₀ and construct growth curves.

The inhibition of biofilm formation by CFS was evaluated using the crystal violet (CV) method (Xu et al., 2016). Then, 90%, 50% or 25% of CFS concentrations were mixed with bacterial suspensions in 96-well plates (experimental group), while control groups were mixed with YPD broth and incubated at 37°C for 24 h. The pathogenic bacteria were E. coli ATCC25922, E. coli (mcr-1) 12-2, S. aureus ATCC29213 and MRSA 1668. After incubation, the cell suspension was washed twice with 150 μL sterile water. Biofilm cells were dried at 37°C for 30 min and then stained with 0.1% 150 μL CV solution for 30 min. Subsequently, the excess CV solution was gently washed away with sterile distilled water. and 150 μL of solution (30% methanol and 10% acetic acid) was added to each well to dissolve the CV solution. Finally, the OD₅₇₀ was measured using an enzyme-labelled instrument. The following equation was used to calculate the biofilm inhibition rate (%):

Suspensions of pathogenic bacteria were co-cultured with varying amounts of CFS in 24-well plates, with sterile polystyrene discs added to wells. Control wells contained bacteria without CFS. After 24 h of incubation at 37°C, planktonic bacteria were removed, washed thrice with phosphate buffer solution (PBS) (Shanghai Sangon Biotech Company), fixed with 1% osmium acid solution at 4°C for 2 h and dehydrated with an ethanol gradient (50%, 70%, 80%, 90% and 95%) for 10-20 min. Then, add tert-butyl alcohol to infiltrate for 2 h. Finally, the film was fixed on the surface of the wafer after carbon dioxide critical drying and ion sputtering, revealing a golden yellow surface, and the film was observed under an SEM (American FEI Company).

Overnight cultures of E. coli and S. aureus were diluted in LB broth. Bacterial suspensions were mixed with 50% CFS and incubated in glass tubes at 37°C for 24 h. The control group was mixed with the same amount of YPD broth. The glass tube was placed at an angle of 30 and incubated at 37°C for 24 h. After incubation, the floating cells were retained, and the adhered cells were collected, washed with PBS, resuspended in PBS buffer and their OD₆₀₀ was measured (OD _adhesion). Then, the planktonic cells were mixed with the adherent cells, washed with PBS and resuspended and the OD₆₀₀ of the mixed cells was measured (OD _mixture). Adhesion capacity (%) was calculated using the following formula:

Overnight cultures of E. coli and S. aureus were prepared in LB broth. The bacterial suspension was mixed with 50% CFS, while the control group received an equivalent volume of YPD broth and was cultured at 37°C for 24 h. Following incubation, it was centrifuged at 8000 ×g for 10 min at 4°C, washed twice with PBS, resuspended with 0.9% saline (1 mL) and then mixed with 5% phenol and 5% sulfuric acid in equal volumes and incubated in darkness for 1 h. The OD₄₉₀ was measured to quantify EPS production using the following equation:

Overnight cultures of E. coli and S. aureus were prepared in LB broth. The experimental group received 50% CFS mixed with bacterial suspension, while the control group received an equal volume of YPD broth and was cultured at 37°C for 24 h. After incubation, the cell suspension was centrifuged at 17,709× g at 4°C for 5 min, washed twice with PBS and resuspended in fresh PBS. The OD₆₀₀ was adjusted to 0.5 ± 0.05 (OD _initial). Then, 1.2 mL of the cell suspension was transferred to a glass test tube, mixed with n-octane (0.3 mL) and vortexed for 3 minutes before standing for 15 minutes. The lower water phase was collected, and the OD₆₀₀ (OD _treatment) was measured.

To assess the inhibitory effort of live BTS1-KO S. cerevisiae on E. coli and S. aureus cells, bacteria-S. cerevisiae co-culture model was established. Overnight cultures of S. cerevisiae and pathogenic bacteria were washed twice with PBS and diluted to 10⁷ CFU/mL in YPD and LB broth medium, respectively. Co-culture groups were prepared by mixing S. cerevisiae and pathogenic bacteria at a 1: 1 ratio (1 mL each). The pathogen culture was mixed with YPD broth (1 mL each) as the single culture group, while the blank group consisted of pathogenic bacteria mixed with sterile water (1 mL each). Cultures were incubated at 37°C and 200 rpm. The cells in the co-culture group and pathogen single culture group were counted on an LB agar plate supplemented with 10 mg/mL amphotericin B (Sigma-Aldrich). The experiment was performed in triplicate.

Precipitated BTS1-KO S. cerevisiae cultured at 30°C for 16 h was washed with PBS three times and resuspended in PBS (Alive S. cerevisiae). For heat-inactivated probiotic suspension, a portion (10 mL) of the above suspension was autoclaved at 121°C for 15 min, then centrifuged at 8000×g for 10 min, washed and resuspended with PBS (Heat-inactivated S. cerevisiae). The G. mellonella larvae were randomly divided into eight groups: one blank control group (PBS group); one control group (Pathogenic bacteria group); three toxicity test groups (CFS group, Alive group and Heat-inactivated group); three treatment groups (CFS + Pathogenic bacteria, Alive + Pathogenic bacteria, Heat-inactivated + Pathogenic bacteria).

The antibacterial activity of farnesene was assessed using a standardised broth microdilution method following the Clinical and Laboratory Standards Institute (CLSI) guidelines (M27-A3) (CLSI, 2008). The inhibitory activity of farnesene against pathogenic bacteria was tested in 96-well plates.

Organic acid types and concentrations in CFS were analysed using a Waters Acquity I class UPLC coupled to a Waters XEVO TQD mass spectrometer. The CFS of BTS1-KO S. cerevisiae, CFS of BY4743 and YPD medium were thawed at 4°C. Then, 5 μL of the supernatant of the sample to be tested was taken into a centrifuge tube, and 35 μL of the configured derivatisation reagent (4.8 μL of AQB and 2 μL of DIPEA dissolved in 2 mL of acetonitrile) and 35 μL of the condensation agent (15.2 mg of HATU dissolved in 2 mL of acetonitrile) were added. After vortexing for 20 min, 100 μL of PBS and 300 μL of methyl tert-butyl ether were added, vortexed for 10 min, and then centrifuged at 6000 rpm at 4°C for 5 min. A total of 50 μL of supernatant was collected in a new centrifuge tube and concentrated by centrifugation at 40°C for 20 min in a vacuum concentrator. The sample was then dried and reconstituted by adding 100 μL of 10% acetonitrile in water; After vortexing for 5 min and centrifuging at 12000 rpm for 15 min at 4°C, 40 μL of supernatant was transferred to the injection vial, and the type and concentration of organic acids were analysed using UPLC-MS/MS.

All experiments were set up in three replicate groups and repeated independently three times. Statistical analysis and plotting were performed using GraphPad Prism software. Experimental data were statistically tested using Student’s t-test or one-way ANOVA. Statistical differences were determined based on P values of *P<0.05, **P<0.01, and ***P<0.005.

Screening of 1800 S. cerevisiae mutants, BTS1-KO S. cerevisiae exhibited potent antibacterial activity. The MIC₈₀ of CFS against E. coli ATCC25922, E. coli -mcr1 12-2, S. aureus ATCC29213 and MRSA 1668 was 90%, 50%, 50% and 50%, respectively, compared to the control group ( Figure 1A ). As the concentration of CFS decreased, the antibacterial activity gradually diminished. Furthermore, BTS1-KO S. cerevisiae demonstrated effectiveness in inhibiting the proliferation of various bacteria, including gram-negative, gram-positive and drug-resistant bacteria. Specifically, CFS inhibited E. coli (mcr-1) 12-2 and S. aureus ATCC29213 by 98%; S. aureus ATCC25923 and S. Typhimurium SL1344 by 96%; E. coli ATCC25922, MRSA 1668, P. aeruginosa ATCC27853 and K. pneumoniae ATCC1705 by 95%; P. aeruginosa 1554 by 94%; K. pneumoniae ATCC700603, K. pneumoniae ATCC1706 and A. baumannii 21-1 by 93%; and K. pneumoniae 2118 by 92%. Importantly, BTS1-KO S. cerevisiae exhibited comparable activity to well-known probiotics like L. rhamnosus and surpassed S. boulardii ( Table 1 ). These results underscored the potent antibacterial activity of BTS1-KO S. cerevisiae CFS.

The growth curve analysis of E. coli and S. aureus treated with BTS1-KO S. cerevisiae CFS revealed significant inhibition compared to controls. H₂O was used instead of YPD broth to examine the effect of nutritional differences on the experimental group. Differences in bacterial growth between the blank control (representing the minimum amount of additional nutrients) and the YPD broth control (representing the maximum amount of additional nutrients) were not significant, suggesting that the bacterial cultures were nutrient-rich and leading to the hypothesis that nutrient depletion did not contribute to the changes in growth. When treated with 90% and 50% CFS, the growth of E. coli ATCC25922 was significantly inhibited, with no significant change in bacterial absorbance within 24 h ( Figure 1B ). Similarly, treatment with 25% CFS reduced the growth rate of E. coli (mcr-1) 12-2, with a lag period from 2 to 4 h, and the total bacterial content at 24 h reduced to approximately 85% of the control. Notably, 90% and 50% CFS effectively inhibited the growth of E. coli (mcr-1) 12-2 within 24 h ( Figure 1C ). Treatment with 50% CFS significantly altered the growth curve of S. aureus ATCC29213, delaying the logarithmic growth phase until 10 h, with the total bacterial content at 24 h reduced to about 53% of the control ( Figure 1D ). Similarly, MRSA growth was significantly delayed, entering the logarithmic growth phase only after 16 h of growth in the presence of 50% CFS, while hardly growing in the presence of 90% CFS throughout the growth phase ( Figure 1E ). These findings indicated that BTS1-KO S. cerevisiae CFS effectively inhibited the growth of both E. coli and S. aureus, with the inhibitory effect showing dose dependence.

Biofilm formation confers strong resistance to pathogenic bacteria (Roy et al., 2018). The inhibitory effect of BTS1-KO S. cerevisiae CFS on E. coli and S. aureus biofilm formation was quantitatively assessed using the CV reduction assay. Results demonstrated that CFS effectively inhibited the biofilm formation of E. coli and S. aureus in a dose-dependent manner. Specifically, treatment with 25% CFS significantly inhibited biofilm formation. Under 90% CFS treatment, the inhibition rates of E. coli ATCC25922, E. coli (mcr-1) 12-2, S. aureus ATCC29213 and MRSA 1668 were 94.7%, 94.7%, 94.0%, and 93.7%, respectively( Figure 2A ). Additionally, SEM images of biofilms revealed a significant reduction in bacterial density and more dispersed arrangements in CFS-treated groups compared to untreated groups, which showed a higher bacterial density and tight arrangement ( Figure 2B ). These findings underscored the ability of BTS1-KO S. cerevisiae CFS to effectively inhibit biofilm formation in both E. coli and S. aureus.

Adhesion is the initial step in biofilm formation, facilitated by the secretion of extracellular polymer matrix, such as EPS produced by pathogens., which in turn can be used to resist antibiotics and other compounds (Lourenço et al., 2012). The results showed that the adhesion abilities of E. coli and S. aureus were reduced by 53.24% and 63.56%, respectively, in the 50% CFS-treated group compared to the CFS-untreated group ( Figure 3A ). Moreover, EPS production by E. coli and S. aureus decreased by 20.18% and 15.69%, respectively, after CFS treatment ( Figure 3B ), indicating a reduction in adhesion ability and consequently in biofilm formation. Additionally, surface hydrophobicity, indirectly reflecting adhesion ability (Yang et al., 2022), was reduced by 18.09% in CFS-treated E. coli, although no significant effect was observed in S. aureus ( Figure 3C ).

Using the bacteria-S. cerevisiae co-culture model, we verified the ability of live BTS1-KO S. cerevisiae to inhibit the growth of pathogenic bacteria. Results indicated a significant reduction in E. coli cell numbers after co-culture with live BTS1-KO S. cerevisiae compared to YPD medium and H₂O controls. At 16 and 24 h, E. coli growth inhibition rates were 91.20% (1.1×10¹⁵ CFU/mL) and 91.43% (1.2×10¹⁵ CFU/mL), respectively ( Figure 4A ). Moreover, a significantly lower cell number of the S. aureus co-culture group was determined compared to the single culture groups, and inhibition rates of 80.43% (1.5×10¹⁵ CFU/mL) and 85.16% (1.58×10¹⁵ CFU/mL) were reported at 16 h and 24 h, respectively ( Figure 4B ). Additionally, YPD broth was replaced with sterile water as a blank control group. The indicator bacteria culture was revealed to be nutrient-rich, with nutrient depletion not contributing to growth changes. These results confirmed the inhibitory effect of BTS1-KO S. cerevisiae cells on the growth of both E. coli and S. aureus.

The G. mellonella larvae model, resembling mammalian systems, was employed to evaluate the in vivo antibacterial activity of BTS1-KO S. cerevisiae (Ménard et al., 2021). Compared with the PBS control group, a single injection of live S. cerevisiae, heat-inactivated S. cerevisiae and CFS had no significant effect on the survival of healthy larvae ( Figure 7A ). According to the survival curve, the survival time of larvae infected with E. coli or S. aureus in the CFS or live cells treated groups was greater than the control group. In the E. coli-infected larvae model, larvae in the CFS-treated group survived longer than those in the control group, with a survival rate of 60% ( Figure 7B ). The survival time of the infected S. aureus larvae treated with CFS was also greater than the control groups, with a survival rate of 50% ( Figure 7C ). Additionally, live BTS1-KO S. cerevisiae prolonged the survival time of E. coli-infected G. mellonella, with a survival rate of 60% observed on the first day ( Figure 7B ). Live BTS1-KO S. cerevisiae also prolonged the survival time of S. aureus-infected G. mellonella, which was infected for 4 days before all died ( Figure 7C ). However, heat-inactivated BTS1-KO S. cerevisiae showed no therapeutic effect on G. mellonella, suggesting that live BTS1-KO S. cerevisiae is required for effective protection in vivo ( Figures 7B, C ). These results underscored the potential of BTS1-KO S. cerevisiae as a therapeutic agent against pathogenic bacterial infections in vivo.

BTS1 encodes geranylgeranyl diphosphate synthase (GGPPS) gene, an enzyme involved in terpenoid biosynthesis (Jiang et al., 1995). Experiments have confirmed that the content of farnesene increased after BTS1 was knocked out (Wang et al., 2022). In this study, the antibacterial activity of farnesene was assessed, revealing that farnesene exerted no antibacterial activity, which indicated that increased farnesene levels were not responsible for the observed antibacterial effects of BTS1-KO S. cerevisiae ( Figure 8A ).

S. boulardii produces organic acids such as lactic acid, acetic acid and formic acid to exert its antibacterial activity (Pais et al., 2020). Therefore, we determined the pH of the CFS of BTS1-KO S. cerevisiae and CFS of BY4743, which were 3.75 and 5.12, respectively ( Figure 8B ). Then, 2 mol/L NaOH solution was used to adjust the pH of CFS to 6.5, which resulted in the loss of the antibacterial activity of CFS ( Figure 8C ). UPLC-MS/MS analysis revealed a significantly higher content of lactic acid in BTS1-KO S. cerevisiae CFS (35-fold increase) compared to S. cerevisiae BY4743, while the content of acetic acid, propionic acid, isobutyric acid and isovaleric acid were not significant, suggesting lactic acid production as a major contributor to its antimicrobial activity ( Figure 8D ). Thus, increased lactic acid production and the consequent decrease in pH likely contribute to the antimicrobial effects of BTS1-KO S. cerevisiae.