The bacterial strains and plasmids used in this study are listed in Table 1. E. coli K-12 strain BW25113 and its derived mutants were cultured in Luria-Bertani (LB) medium. L. plantarum BM-1 was cultured in de Man, Rogosa, and Sharpe (MRS) medium. Unless otherwise stated, all bacterial strains were cultured for 12 h at 37°C with a rotational speed of 180 rpm in a horizontal shaker.

Plantaricin BM-1 was isolated and purified as described previously (Zhang et al., 2013). L. plantarum BM-1 was cultured for 12 h, as described in Section 2.1. Cell-free supernatants (CFS) were collected through centrifugation (9,000 rpm, 15 min) at 4°C. Crude plantaricin BM-1 in the CFS was separated using saturated ammonium sulfate (516 g/L CFS) and centrifuged (9,000 rpm, 15 min, 4°C). Crude plantaricin was dissolved in sterile double-distilled water (ddH₂O) and purified by dialysis, cation-exchange chromatography, and desalting. The purified plantaricin solution was lyophilized using a FreeZone 6 freeze dryer (Labconco, Corp., Kansas City, MO, United States).

Lyophilized plantaricin BM-1 was re-dissolved in ddH₂O. The protein concentration of the re-dissolved solution was determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States). Before use, plantaricin BM-1 solution was sterilized by passing through a Millex^®-GP filter (0.22 μm; Merck Millipore, Germany).

To further investigate the consequences of fliA deletion, fliA-complemented E. coli strain JW1907 was constructed using the lambda red homologous recombination method (Datsenko and Wanner, 2000). E. coli BW25113 and JW1907 were cultured for 12 h, as described in section 2.1. Genomic DNA was extracted from E. coli BW25113 cultures using a TIANamp bacterial DNA kit (Tiangen Biotech, Beijing, China). The fliA fragment (760 bp) was amplified from the E. coli BW25113 genome using the fliA-Red-F (5′-AAGTGCGGCATTTACTGACGTTATAACTTACCCAGTTTAG-3′) and fliA-Red-R (5′-TATTGCGTCCCGACAAATAGGTGAATT CACTCTATACCGC-3′) primers and purified through a TIANquick midi purification kit (Tiangen Biotech). Plasmid pKD46 was heat-shocked into E. coli JW1907 competent cells, prepared with 0.1 mol/L calcium chloride. Subsequently, 0.01 mol/L L-arabinose was added and incubated with pKD46-harboring E. coli JW1907 at 30°C, inducing the full expression of recombinases. The fliA fragment was then transformed into E. coli JW1907, spread on LB agar plates (containing 50 μg/mL kanamycin), and grown overnight at 37°C. PCR products of E. coli BW25113, JW1907, and the positive transformant genomes amplified with the fliA-Red-F/R primer pair were analyzed by 1% agarose gel electrophoresis, and a Takara DL2,000 DNA Marker (Cat# 3427, Takara Bio, Japan) was used. The complementary transformant was renamed E. coli ReJW1907.

E. coli strains were cultured in LB broth for 12 h. The viable cell concentration in each E. coli culture was adjusted to 10⁴ CFU/mL. One milliliter of E. coli culture was added to a 48-well plate, and plantaricin BM-1 was added at a final concentration of 3.75 mg/mL, according to our previous study by Chen et al. (2021). The plate was subjected to a MicroScreen real-time microbial growth analysis system (Jieling Instrument Manufacturing Co., Ltd, Tianjin, China) for 12-h at 37°C with a rotational speed of 500 rpm. The optical density at 600 nm (OD_{600 nm}) of each well was obtained hourly. E. coli strains without plantaricin treatment were used as negative controls, and sterile LB broth was used as the blank control. Three biological replicates were performed for all E. coli groups, and the average value was used to plot the bacterial growth curves.

The inhibition rate (I%) of E. coli by plantaricin BM-1 was defined as the reduction in bacterial growth according to Luo et al. (2014), and the I% at the indicated point was calculated using the following formula:

where OD_CT,t represents the OD_{600 nm} of untreated E. coli at the indicated time point; OD_CT,0 represents the OD_{600 nm} of untreated E. coli at the initial time point; OD_PT,t represents the OD_{600 nm} of plantaricin-treated E. coli at the indicated time point; OD_PT,0 represents the OD_{600 nm} of plantaricin-treated E. coli at the initial time point; I% = 100% is defined as no bacterial growth; I% > 100% represents bacterial death.

Scanning (SEM) and transmission electron microscopy (TEM) were used as previously described (Bian et al., 2022; Chen et al., 2021). E. coli BW25113 and JW1907 were incubated with plantaricin BM-1 (3.75 mg/mL) for 12 h at 37°C, as described in section 2.4. The untreated E. coli strains were used as controls. E. coli cells were collected at 8 and 12 h by centrifugation (8,000 rpm, 4°C, 10 min). After being washed with 0.1 mol/L PBS buffer thrice, E. coli cells were fixed with 2.5% glutaraldehyde at 4°C for 24 h. Fixed samples were rinsed with 0.1 M PBS thrice, dehydrated at 25 ± 2°C using a graded ethanol series (30, 50, 70, 80, 90, 95, and 100%), and preserved in 100% ethanol.

For SEM, dehydrated samples were dried using a Quorom k580 critical point dryer (Quorom, UK). The dried samples were coated with conductive carbon blue and fixed on the sample stage; platinum was sprayed using an MC1000 ion sputtering instrument (Hitachi, Japan). Cell samples were analyzed using SEM (Region 8100, Hitachi, Japan).

For TEM, the dehydrated samples were treated twice with 100% acetone for 10 min each and soaked in different proportions of acetone and EMBed 812 resin mixtures in turn (3:1 for 1 h; 1:1 for 2 h; 1:3 for 2 h; pure EMBed 812 for 8 h) at 37°C. The samples were inserted into the embedding models containing pure EMBed 812 resin for polymerization, which was performed for 24 h at 65°C. The polymerized samples were sliced to create 70–90 nm sections using a UC7 ultramicrotome (Leica, Germany). The sections were transferred to copper grids and stained with 2% uranyl ethanol and lead citrate. After being washed with deionized water, the samples were dried overnight at 25 ± 2°C and analyzed using TEM (HT7800, Hitachi, Japan).

Flow cytometry was performed to assess the damage caused by plantaricin BM-1 to the cell membrane integrity of E. coli cells according to the method described by Li et al. (2013), with several modifications. E. coli BW25113 and JW1907 were cultured in LB broth for 12 h. The concentration of viable cells in each E. coli culture was adjusted to 10⁶ CFU/mL. Next, 1 mL of each culture sample was transferred to a sterile 1.5-mL Eppendorf tube, and plantaricin BM-1 was added to a final concentration of 3.75 mg/mL and incubated at 37°C for 8 h and 12 h. Untreated E. coli cells were the control group. E. coli cells were collected by centrifugation (8,000 rpm, 4°C, 10 min), washed with PBS buffer thrice, and stained with propidium iodine (PI) at a final concentration of 10 μg/mL for 10 min. The samples were then subjected to a Guava^® EasyCyte™ system (Luminex, Corp., Austin, TX, US) for flow cytometric analysis. PI fluorescence was recorded using the RED-B channel (695/50 nm).

To investigate the effect of fliA deletion on biofilm formation by E. coli BW25113, crystal violet (CV) staining and confocal laser scanning microscopy (CLSM) were performed on E. coli BW25113, JW1907, and ReJW1907.

To investigate the regulatory role of FliA in the antibacterial mechanism of plantaricin BM-1 against E. coli, RNA-sequencing (RNA-seq) was performed on E. coli BW25113 and JW1907. E. coli strains were cultured in LB broth for 12 h, as described in section 2.1. Three biological replicates were used for both E. coli strains. Bacterial samples were collected through centrifugation (8,000 rpm, 4°C, 10 min), washed thrice with 0.1 M PBS, subjected to liquid nitrogen for 10 min, and preserved at −80°C.

Samples were sent to Shanghai Majorbio Bio-pharm Technology Co., Ltd for subsequent RNA-seq. Library preparation was performed using a TruSeq™ Stranded Total RNA Library Prep Kit (Illumina, Inc., San Diego, CA, US). Briefly, total RNA was extracted from both E. coli cells, and ribosomal RNA was removed. The enriched mRNA was randomly fragmented into approximately 200 bp pieces, which served as templates for reverse transcription to synthesize cDNAs. The sticky ends of double-stranded cDNAs were supplemented by End repair Mix to become blunt ends, and the adapter was added to the 3′ end of cDNA. Before PCR enrichment, the second strand was digested with uracil-N-glycosylase (UNG). The library was enriched and subjected to Illumina NovaSeq X Plus platform (Illumina, Inc., San Diego, CA, US) for sequencing. Quality control (QC) was performed on raw data files, and the sequences of the clean reads from each sample were aligned with the reference genome of E. coli K-12 BW25113 (NCBI Assembly No. GCF_000750555.1). The RSEM software (Version 1.1.3) was used to quantitatively analyze gene expression levels. Normalization of gene expression was performed using transcripts per million (TPM), according to the following formula:

where TPM_i represents the calculated TPM value of gene (i); Counts_i represents the mapped read counts for gene (i); Length_i represents the length (kb) of gene (i); ∑imCountsiLengthi represents the sum of CountsiLengthi for all genes.

Differential analysis was performed using DESeq2 software (Version 1.42.0), and differentially expressed genes (DEGs) were screened with the following thresholds: p-adjust < 0.05; |log₂Fold Change (FC)| ≥ 1. Functional annotation analysis of the DEGs was conducted using the Gene Ontology (GO) database. The Kyoto Encyclopedia of Genes and Genomes (KEGG) database was used to analyze the functions of biological pathways associated with the DEGs.

Supplementary Table S1 lists all primers used for reverse transcription quantitative PCR (RT-qPCR), and the housekeeping gene GADPH (gapA) was used as the internal control. Total RNA of E. coli BW25113 and JW1907 was extracted using a RNAprep Pure Cell/Bacteria Kit (Tiangen Biotech, China). An RT-qPCR system (25 μL) was established by a Fastking one step RT-qPCR kit (Tiangen Biotech, China): 2 × Fastking RT-qPCR buffer (SYBR Green), 12.5 μL; 25 × Enzyme mix, 1 μL; forward and reverse primers (10 μM), 0.62 μL; RNA template, 0.5 μL; RNase-free water to a total volume of 25 μL. Amplification was performed using the LightCycler^® 96 real-time PCR system (Roche, Swiss); the RT-qPCR conditions are presented in Supplementary Table S2. Relative gene expression was calculated using the 2^–ΔΔCt method. All experiments were performed in triplicate.

All statistical analyses were performed using SPSS Statistics software V27.0.1.0 (IBM, corp., Armonk, NY, United States). Data are presented as mean ± standard deviation (SD). Mean comparisons were performed by one-way analysis of variance (ANOVA), followed by Dunnett’s t-test or Duncan’s test. The significance level was set to 0.05. All growth curves and bar charts were plotted using GraphPad Prism software version 10.4.1 (627) (GraphPad Software, Boston, MA, United States).

The fliA-Red-F/R primer pair was used to amplify the genomes of E. coli K12, JW1907, and positive transformants using PCR. The electrophoretogram shown in Figure 1 confirms that the fliA fragment (760 bp) was amplified from the genomes of E. coli BW25113 and the positive transformant, not E. coli JW1907, showing the successful transformation of the fliA gene into E. coli JW1907. The fliA-complemented transformant was renamed E. coli ReJW1907.

The untreated E.coli BW25113, ReJW1907, and JW1907 cells in the control group exhibited similar growth curves (Figure 2A), entering the exponential phase within 5 h of cultures, with OD_{600 nm} values reaching 4.619 ± 0.086, 4.659 ± 0.049, and 4.660 ± 0.085 by 12 h, respectively. Significant delays in entering the exponential phase were observed in all three E. coli strains following plantaricin BM-1 treatment (3.75 mg/mL), demonstrating its antibacterial activity. Plantaricin-treated E. coli BW25113 and ReJW1907 entered the exponential phase after 7 h, with OD_{600 nm} values reaching 4.738 ± 0.108 and 4.728 ± 0.098, respectively. In comparison, plantaricin-treated E. coli JW1907 entered the exponential phase later, by 8 h, with a peak OD_{600 nm} of 2.810 ± 0.196 at 12 h.

The inhibition rate (I%) after 8 h of culture in plantaricin BM-1 (3.75 mg/mL)-treated E. coli JW1907 (89.22 ± 1.13%) was significantly higher (p < 0.01) than that of E. coli BW25133 (70.36 ± 6.30%) and E. coli ReJW1907 (74.75 ± 4.99%; Figure 2B). No significant difference was observed between the I% of E. coli BW25133 (70.36 ± 6.30%) and ReJW1907 (74.75 ± 4.99%; Figure 2B).

To investigate the effect of plantaricin BM-1 on E. coli cell morphology, SEM and TEM analyses were performed on E. coli BW25113 and JW1907 following treatment with plantaricin BM-1 (3.75 mg/mL) for 8 or 12 h. E. coli BW25113 and JW1907 cells in the untreated groups appeared complete, with a full cylindrical shape with blunt ends (Figure 3Aa,d) and a dense and uniform internal structure with clear, intact edges (Figure 3Ba,d). This indicated that fliA deletion alone did not lead to visible cell damage in E. coli. However, after 8 or 12 h of plantaricin BM-1 treatment, the external shape of E. coli BW25113 and JW1907 cells shrunk significantly with distortion and collapse (Figure 3Ab,c,e,f); slight plasmolysis was observed inside both E. coli strains (Figure 3Bb,c,e,f). Notably, no visible fractures or pore formation were observed on the cell surfaces.

PI is a fluorescent molecule that only penetrates impaired cells and binds DNA (Li et al., 2013; Crowley et al., 2016). Histograms of cell counts versus PI fluorescence intensity for the untreated E. coli BW25113 (Figure 4A) and JW1907 (Figure 4D) exhibited single peaks with cell counts corresponding to fluorescence distributed below 10¹. The PI uptakes by E. coli BW25113 (Figures 4A–C) and JW1907 (Figures 4D–F) did not increase following 8 h or 12 h of plantaricin BM-1 (3.75 mg/mL) treatment compared to the untreated control group. This showed that the membrane integrity of both treated E. coli strains was not significantly damaged.

The biofilm production of E. coli BW25113, JW1907, and ReJW1907 was quantified by CV staining after 24-h of incubation. All three E. coli strains formed biofilms, with biomasses significantly higher (p < 0.05) than that of the blank control (OD_{595 nm} = 0.130 ± 0.005; Figure 5A). The biofilm biomass produced by E. coli JW1907 (OD₅₉₅ = 0.227 ± 0.04) was significantly lower (p < 0.05) than those of E. coli BW25113 (OD_{595 nm} = 0.343 ± 0.056) and ReJW1907 (OD_{595 nm} 0.358 ± 0.027); however, no significant difference was observed between E. coli BW25113 and ReJW1907.

The 3D and 2D CLSM images of the biofilms revealed that after 24 h, E. coli BW25113 (Figure 5Ba) and ReJW1907 (Figure 5Bc) formed dense, evenly distributed, multilayered biofilms on the glass bottom. In contrast, E. coli JW1907 did not form a significant biofilm, with degeneration of the biofilm structure and cell adhesion observed, resulting in reduced biofilm biomass and thickness (Figure 5Bb).

To further evaluate the expression levels of the eight DEGs (lsrKRBDCAFG) associated with the LuxS/AI-2 QS system, RT-qPCR was performed on the total RNA extracted from E. coli BW25113 and JW1907. The relative expressions of lsrB, lsrD, lsrG, and lsrF were significantly up-regulated by 3.693 ± 1. 641-, 2.924 ± 0. 578-, 3.456 ± 0. 570-, and 2.676 ± 0.320-fold, compared with gapA, respectively (p-adjust < 0.05) (Figure 7). The relative expressions of lsrC, lsrA, lsrK, and lsrR were also up-regulated by 2.529 ± 0. 391-, 1.315 ± 0. 444-, 2.587 ± 0. 402-, and 1.628 ± 0.188-fold compared with gapA; however, statistical significance was not reached.

To determine whether the DEGs associated with the LuxS/AI-2 QS system contributed to the antibacterial mechanism of plantaricin BM-1 against E. coli BW25113, bacterial growth curves were constructed following treatment of E. coli BW25113 and 8 single-gene knockout mutants (E. coli JW1509, JW1508, JW1507, JW1506, JW1511, JW1510, JW1505, and JW1504) with plantaricin BM-1 (3.75 mg/mL) for 12 h (Figure 8A).

Untreated E. coli BW25113 and the eight mutants in the control group entered the exponential phase after 5 h of culture, with OD_{600 nm} values peaking between 4.275 ± 0.079 and 4.749 ± 0.069 after 12 h. The logarithmic growth of the nine E. coli strains treated with plantaricin BM-1 (3.75 mg/mL) was delayed, with the OD_{600 nm} of treated E. coli BW25113 exhibiting a steeper increase than those of the eight mutants after 8 h of culturing. Moreover, the I% of plantaricin BM-1 (3.75 mg/mL) for E. coli BW25113 (70.77 ± 7.01%) was significantly lower (p < 0.05) than that for any mutant (from 80.68 ± 3.50% to 90.07 ± 1.08%) after 8 h (Figure 8B).

To further investigate the relationship between the LuxS/AI-2 QS system and biofilm formation in E. coli K-12, CV staining was performed on E. coli BW25113 and eight LuxS/AI-2 QS system-related gene mutants, including E. coli JW1509, JW1508, JW1507, JW1506, JW1511, JW1510, JW1505, and JW1504.

All E. coli strains formed biofilms, with biomasses that were significantly higher (p < 0.05) than that of the blank control (OD_{595 nm} = 0.116 ± 0.003; Figure 9). After 24-h incubation, E. coli BW25113 produced significantly more (p < 0.05) biofilm biomass (OD_{595 nm} = 0.254 ± 0.014) than the eight LuxS/AI-2 QS system-related gene mutants (from 0.135 ± 0.007 to 0.188 ± 0.015), showing a regulatory role of the LuxS/AI-2 QS system in E. coli biofilm formation.