The Coexistence of Bacterial Species Restructures Biofilm Architecture and Increases Tolerance to Antimicrobial Agents

Bacterial biofilms possess distinct community properties that affect various aspects of our daily lives. In particular, biofilms exhibit increased tolerance to chemical disinfectants, antimicrobial agents, and host immune responses. ABSTRACT Chronic infections caused by polymicrobial biofilms are often difficult to treat effectively, partially due to the elevated tolerance of polymicrobial biofilms to antimicrobial treatments. It is known that interspecific interactions influence polymicrobial biofilm formation. However, the underlying role of the coexistence of bacterial species in polymicrobial biofilm formation is not fully understood. Here, we investigated the effect of the coexistence of Enterococcus faecalis, Escherichia coli O157:H7, and Salmonella enteritidis on triple-species biofilm formation. Our results demonstrated that the coexistence of these three species enhanced the biofilm biomass and led to restructuring of the biofilm into a tower-like architecture. Furthermore, the proportions of polysaccharides, proteins, and eDNAs in the extracellular matrix (ECM) composition of the triple-species biofilm were significantly changed compared to those in the E. faecalis mono-species biofilm. Finally, we analyzed the transcriptomic profile of E. faecalis in response to coexistence with E. coli and S. enteritidis in the triple-species biofilm. The results suggested that E. faecalis established dominance and restructured the triple-species biofilm by enhancing nutrient transport and biosynthesis of amino acids, upregulating central carbon metabolism, manipulating the microenvironment through “biological weapons,” and activating versatile stress response regulators. Together, the results of this pilot study reveal the nature of E. faecalis-harboring triple-species biofilms with a static biofilm model and provide novel insights for further understanding interspecies interactions and the clinical treatment of polymicrobial biofilms. IMPORTANCE Bacterial biofilms possess distinct community properties that affect various aspects of our daily lives. In particular, biofilms exhibit increased tolerance to chemical disinfectants, antimicrobial agents, and host immune responses. Multispecies biofilms are undoubtedly the dominant form of biofilms in nature. Thus, there is a pressing need for more research directed at delineating the nature of multispecies biofilms and the effects of the properties on the development and survival of the biofilm community. Here, we address the effects of the coexistence of Enterococcus faecalis, Escherichia coli, and Salmonella enteritidis on triple-species biofilm formation with a static model. In combination with transcriptomic analyses, this pilot study explores the potential underlying mechanisms that lead to the dominance of E. faecalis in triple-species biofilms. Our findings provide novel insights into the nature of triple-species biofilms and indicate that the composition of multispecies biofilms should be a key consideration when determining antimicrobial treatments.

biofilms at 24 h (Fig. 1A). Strikingly, the triple-species biofilm biomass was also increased by relatively low proportions of E. coli O157:H7 and S. enteritidis (20:1:1) at 24 h. In addition, the biomass of the triple-species biofilm with the three species presented at a ratio of 1:1:1 was still significantly higher than the total biomass of the three mono-species biofilms at 48 h. However, the biomass of the triple-species biofilm with an initial ratio of 2:1:1, 5:1:1, or 10:1:1 was not significantly different from the total biomass of the three mono-species biofilms, while the triple-species biofilm with an initial ratio of 20:1:1 showed reduced biomass (Fig. 1A). The results strongly suggest that the coexistence of E. faecalis, E. coli, and S. enteritidis enhanced triple-species biofilm formation and that the initial ratio affected triple-species biofilm formation. Accordingly, we chose 1:1:1 as the ratio for the following assays to further investigate the triple-species biofilm.
E. faecalis biofilm formation occurs in two phases, namely, the initial phase (24 h) and mature phase (48 h). To confirm the effect of the coexistence of these bacteria on biofilm formation, we further measured biofilm formation with 0.1% crystal violet staining and confocal laser scanning microscopy (CLSM) imaging separately. The crystal violet staining results showed that the triple-species biofilm biomass increased approximately 1.7-and 1.45-fold compared to the biomass of the E. faecalis mono-species biofilm at 24 h and 48 h, respectively (Fig. 1A). CLSM imaging clearly showed that the triple-species biofilm was denser than the E. faecalis mono-species biofilm in both phases (Fig. 1B). In addition, based on the live/dead cell staining results from CLSM imaging, the triple-species biofilm at 48 h contained more live and dead cells compared to the other biofilms (Fig. 1B). Again, these results confirmed that the coexistence of these bacteria enhanced triple-species biofilm formation.
Interestingly, coculturing E. faecalis with either E. coli O157:H7 or S. enteritidis also The Coexistence of Bacterial Restructures Biofilm Microbiology Spectrum significantly enhanced double-species biofilm formation (Fig. S1). However, the biomass of the double-species biofilm formed by E. coli O157:H7 and S. enteritidis was lower (Fig. S1). These pilot results indicated that E. faecalis plays a key role in interspecies interactions and multispecies biofilm formation. The triple-species biofilm forms a distinct tower-like architecture. To further dissect the architecture of the triple-species biofilm, we imaged it using scanning electron microscopy (SEM). Surprisingly, the triple-species biofilm formed a complex tower-like structure with tight cell-cell contact, while the E. faecalis mono-species biofilm formed a flat structure ( Fig. 2A). The tower-like structure was dominated by E. faecalis, while E. coli O157:H7 and S. enteritidis were observed in very small amounts ( Fig. 2A). These results suggested that E. faecalis-harboring triple-species biofilms form a tower-like architecture.
Biofilm formation is known to be closely associated with tolerance to antimicrobials in chronic and device-related infections. To examine the tolerance of E. faecalis-harboring triple-species biofilms to antimicrobial agents, we treated the triple-species biofilm cultured for 24 h or 48 h with various concentrations of the commonly used disinfectant sodium hypochlorite and antibiotic ampicillin. The results showed that the triple-species biofilm exhibited much greater tolerance to both sodium hypochlorite and ampicillin than the E. faecalis mono-species biofilm ( Fig. 2B and C). Together, these observations revealed that the coexistence of E. faecalis, E. coli O157:H7 and S. enteritidis results in the formation of triple-species biofilms with a tower-like architecture and enhanced tolerance to antimicrobial agents.
E. faecalis dominates triple-species biofilms. Bacterial cells generally account for 10% of the biofilm biomass. Our CLSM imaging results showed that the triple-species biofilm had a higher density of both live and dead cells than the E. faecalis mono-species biofilm (Fig. 1C). Furthermore, the SEM results indicated that spherical E. faecalis cells dominated the triple-species biofilm ( Fig. 2A). To further determine the proportion of live bacterial cells in the community in the triple-species biofilm, we enumerated the colony-forming unit (CFU) of E. faecalis, E. coli O157:H7, and S. enteritidis on corresponding selection media on which only one of these bacterial species could grow. In the triple-species biofilm with a starting coculture ratio of 1:1:1 for E. faecalis, E. coli O157:H7, and S. enteritidis, E. faecalis dominated the community, with a proportion of 93.18% at 24 h and 99.18% at 48 h (Fig. 3). To exclude the potential effect of the starting coculture ratio on the variation in live cell proportion, we verified the dominance of E. faecalis at 24 h and 48 h in the triple-species biofilm with a starting coculture ratio of 0.5:1:1 for E. faecalis, E. coli O157:H7, and S. enteritidis. The results consistently demonstrated that E. faecalis dominated the triple-species biofilm independent of the starting coculture ratio (Fig. S2).
ECM components of the triple-species biofilm show phase-dependent variation. The extracellular matrix (ECM) consists primarily of polysaccharides, proteins, and nucleic acids, which account for approximately 90% of the biofilm biomass. SEM imaging of the triple-species biofilm showed the restructuring of the biofilm and tightened contact among cells ( Fig. 2A). These observations indicated that the ECM components of the triple-species biofilm may have changed compared to those of the E. faecalis mono-species biofilm. To address this question, we analyzed the amounts of polysaccharides, proteins, and nucleic acids in triple-species biofilm. Our results showed that all three kinds of ECM components were significantly enriched in the triple-species biofilm compared to the E. faecalis mono-species biofilm at 24 h and 48 h, respectively (Fig. 4A). In the E. faecalis mono-species biofilm, all three kinds of ECM components were enriched from the initial phase (24 h) to the mature phase (48 h) of biofilm formation, indicating the accumulation of ECM components (Fig. 4A). Interestingly, the protein and polysaccharide levels were significantly decreased in the 48 h triple-species biofilm compared to the 24 h triple-species biofilm, while the extracellular DNA (eDNA) level was dramatically increased (Fig. 4A). These observations were confirmed by staining of the three ECM components (Fig. S3).
To further investigate the effects of changes in ECM components on biofilm  formation, we treated the triple-species biofilm with proteinase K (proteins), sodium periodate (polysaccharides), and DNase I (eDNA). The results showed that all three treatments significantly reduced the triple-species biofilm biomass (Fig. 4B). Interestingly, treatment with proteinase K or sodium periodate had more dramatic effects on the biomass of the 24 h triple-species biofilm than on that of the 48 h biofilm, while DNase I treatment had a more dramatic effect on the 48 h triple-species biofilm than on the 24 h biofilm (Fig. 4B). Taken together, these results demonstrated that the ECM components changed significantly in the triple-species biofilm and that the changes may have contributed to biomass accumulation in the triple-species biofilm.   Transcriptomic profiling reveals genes involved in the dominance of E. faecalis in triple-species biofilms. Given the dominance of E. faecalis in the triple-species biofilm, we decided to systematically analyze the transcriptomic changes of E. faecalis in response to coexistence with E. coli and S. enteritidis by RNA sequencing (RNA-seq). We used a static biofilm to mimic a nutritionally unsustainable environment and performed RNA-seq on E. faecalis. The RNA-seq results showed that there were 592 upregulated genes and 520 downregulated genes in the 24 h triple-species biofilm and 788 upregulated genes and 848 downregulated genes in the 48 h triple-species biofilm ( Fig. 5A and B; Table S1). We next conducted qPCR analyses on the transcript levels of differentially expressed genes (DEGs), which are known to be related to biofilm formation and E. faecalis sugar metabolism (28,29), and obtained consistent differential expression patterns with the RNA-seq data (Fig. S4). Of these DEGs from RNA-seq analyses, 274 were upregulated at both 24 h and 48 h, with the major functions involving various response regulator transcription factors, bacterial cell division, glycosyl hydrolase, PTS transport, ABC transport, WXL domain-containing protein, type II toxin-  Table S2). Notably, at 48 h in the triple-species biofilm, approximately 60.3% of the upregulated genes and 61.9% of the downregulated genes showed a more than 4-fold change in expression (Table S1). The Gene Ontology (GO) functional enrichment analysis showed that the functions of the E. faecalis DEGs in the 24 h triple-species biofilm were mainly annotated to cellular macromolecule metabolic process, ribosome, nucleic acid binding, and heterocyclic compound binding, while the functions of the E. faecalis DEGs in the 48 h triple-species biofilm were mainly annotated to establishment of localization, organic substance transport, transmembrane transport, catalytic complex, transporter activity, and transmembrane transport activity ( Fig. 5D and E; Table S3). According to the Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis, more DEGs of E. faecalis primarily involved in membrane transport, glycan biosynthesis and metabolism, and carbohydrate metabolism were upregulated in 24 h triple-species biofilm, consistent with the accumulation of biomass in the triple-species biofilm in this initial phase ( Fig. 5F; Table S4). In comparison, the number of DEGs of E. faecalis involved in membrane transport, glycan biosynthesis and metabolism, and carbohydrate metabolism was reduced in the 48 h triple-species biofilm, which may be related to nutrient depletion in the mature phase ( Fig. 5G and Table S4). In summary, the GO and KEGG analyses strongly suggest that E. faecalis establishes dominance in triple-species biofilms by upregulating the nutrient acquisition and carbohydrate metabolism pathways.
In terms of nutrient transport, the expression of 2 DEGs (ef0892 and ef0893) involved in amino acid transport was upregulated 6.3-fold at 24 h ( Table 1). The expression of 7 DEGs (gltA, glnA, gdhA, glmS, purF, carB, and pyraA) involved in the metabolism and catalysis in the glutamine biosynthesis pathway increased 2.5 to 6.4-fold at 24 h (Table 1). In addition, the expression of 2 DEGs (ef2180 and ef2181) encoding glycosyltransferase 2 family proteins increased 5.3 and 4.5-fold, respectively (Table 1). Furthermore, the expression of some genes encoding glycosyl hydrolases, metallohydrolases, a/b hydrolases, and HAD family hydrolases was upregulated 1.6 to 27.6fold at 24 h (Table S5). At 48 h, 9 genes involved in sugar transport by the phosphatase transfer system (PTS) were upregulated, especially the genes involved in the transport of mannose (ef0815, ef0816, and ef0817), galactose (gatA and gatC), and glucose (ef2438 and ef0541), which were upregulated 2.3 to 44.2-fold (Table 1).
In addition, in terms of central carbon metabolism, 47 and 75 genes involved in carbohydrate metabolic process pathways were detected at 24 h and 48 h, respectively ( Fig. 6 and Fig. S5; Table S3 and S6). Of these, the zwf, ef1918, and gnd genes involved in the oxidative branch phase of the pentose phosphate pathway (PPP) were upregulated 1.5 to 10.5-fold and 1.3 to 3.8-fold at 24 h and 48 h, respectively. Six genes (glpF, gldA, glpK, dhaM, dhak, and glpO) involved in glycerol metabolism were upregulated 2.7 to 4.0-fold at 24 h and 1.3 to 2.8-fold at 48 h (Table S6).
In addition, our transcriptomic analyses showed that the majority of virulence genes, such as the genes gelE encoding gelatinase and entV encoding enterocin, were not differentially expressed or were downregulated in the triple-species biofilm (Table S7). Interestingly, the expression of the E. faecalis gene ldh-1 encoding L-lactate dehydrogenase (LDH) was upregulated 1.9-fold at both 24 h and 48 h in the triple-species biofilm ( Table 1). LDH generates lactic acid during the fermentation process. It has been reported that E. faecalis reduces the pH of the polymicrobial environment by generating lactic acid as a "biological weapon" to inhibit the growth of other species and facilitate its own growth (30)(31)(32)(33). To test the effect of E. faecalis ldh-1 upregulation on the pH of the microenvironment in the triple-species biofilm, we determined the pH by pHrodo staining. The results showed that the pH of the microenvironment in the triple-species biofilm decreased with increasing coculture time (Fig. S6).

DISCUSSION
Multispecies biofilms are generally more tolerant to antimicrobial treatments than the corresponding planktonic cells. Understanding the nature of multispecies biofilms The Coexistence of Bacterial Restructures Biofilm Microbiology Spectrum will contribute to the development of novel strategies to treat multispecies biofilm infections. In this pilot study, we used a static biofilm model to investigate the effect of the coexistence of E. faecalis, E. coli O157:H7, and S. enteritidis on triple-species biofilm formation. Given that static biofilms induce a nutrient stress response, they resemble the microenvironment observed during coinfection in vivo and in biofilm studies in other environments. We found that the coexistence of E. faecalis with E. coli O157:H7 and S. enteritidis dramatically enhanced the triple-species biofilm biomass, altered its ECM composition, and restructured the biofilm with a tower-like architecture. The triple-species biofilm was dominated by E. faecalis and showed elevated tolerance to disinfectants and antibiotics compared to the E. faecalis mono-species biofilm. Our transcriptome analyses indicated that E. faecalis reprograms its metabolic and regulatory pathways in response to coexistence with E. coli and S. enteritidis. Generally, bacteria in polybacterial biofilm communities interact with each other synergistically or competitively, resulting in enhanced or decreased biofilm formation, respectively (34,35). Our results demonstrated that the biofilm biomass was significantly enhanced by different bacterial mixing ratios in the triple-species biofilm model. The biofilm biomass was also significantly increased in the double-species biofilm formed by E. faecalis with E. coli O157:H7 or S. enteritidis, while coculture of E. coli O157: H7 and S. enteritidis dramatically decreased the biofilm biomass. In addition, the triplespecies biofilm formed a more complex tower-like architecture than the E. faecalis monospecies biofilm. These findings strongly indicate that intricate interspecies interactions occur when E. faecalis, E. coli O157:H7, and S. enteritidis coexist with each other. Further interactome studies will guide the future investigation of the roles of individual bacterial species in mediating the interspecies interactions. The principle of competitive exclusion asserts that the adaptive capacity of species to a niche defines its reproduction rates and consequently leads to the dominance or exclusion of the species in multispecies biofilms (8). E. faecalis has a strong adaptive capacity, characterized by persistence in different environments and typical metabolic flexibility (36). We found that E. faecalis dominated the triple-species biofilm, accounting for 93.18% of the bacterial population in the 24 h triple-species biofilm and up to 99.93% in the 48 h triple-species biofilm. This dominance of E. faecalis has also been previously observed in other E. faecalis-harboring multispecies biofilms (37). More importantly, the coexistence of a small portion of E. coli O157:H7 and S. enteritidis with E. faecalis leads to restructuring of the triple-species biofilm into a tower-like form with elevated tolerance to disinfectant and antibiotic treatments. These findings highlight the challenges in the treatment of clinical biofilm infections and the removal of biofilm contamination in food processing.
The formation of a tower-like architecture may promote the stability of triple-species biofilms. Similar tower-like structures biofilms have been observed in E. faecalis biofilms under antibiotic stress and in S. aureus biofilms (38). Additionally, in tower-like biofilms, localization in the top biofilm layers is believed to provide direct growth advantages via continuous supply of and access to resources, especially glucose (11,39). Nutrient availability is critical for biofilm structure (40). Therefore, access to more nutrients and space for growth may be the top requirement for the survival of each bacterium in resource-limited triple-species biofilms. The ECM plays a scaffolding role in biofilms (41). The ECM components changed in different phases of triple-species biofilm formation. At 24 h, proteins and polysaccharides dominated the ECM in the triple-species biofilm, while at 48 h, the levels of proteins and polysaccharides decreased, and the eDNA level increased significantly. Similar enhancements in biofilm formation through augmentation of bacterial numbers and eDNA levels were also observed in mixed-species biofilms formed by Staphylococcus epidermidis and Candida albicans (42), as well as Streptococcus mutans and C. albicans (43). Additionally, ECM components can also function as a nutrient pool (44). Under nutritional stress, E. faecalis may utilize ECM components, especially polysaccharides and proteins, as major nutrient sources to acquire fitness benefits (44), consequently leading to a decrease in polysaccharide and protein levels in the ECM of the 48 h triple-species biofilm compared to that of the 24 h biofilm (Fig. 4B). The transcriptomic analyses, showing the upregulation of genes for sugar and amino acid transport in E. faecalis, also indicated the enhanced utilization of polysaccharides and proteins (Table 1). In contrast, the eDNA level in the 48 h triple-species biofilm increased in comparison with that in the 24 h triple-species biofilm (Fig. 4B), consistent with the CLSM results (Fig. 1C). These observations support the previous findings indicating that eDNA functions as a structural component within biofilms, providing stability to the entire structure (45). Furthermore, changes in the ECM can alter the interactions between cells in biofilms, especially between cells of different species (46). These social interactions can lead to changes in the composition and structure of microcolonies in biofilms, thus shaping their overall function, as well as their virulence in the presence of pathogens (9). Thus, such changes in ECM components are important for the clinical treatment of biofilmrelated infections.
Competitors in biofilms can utilize the lysate of neighboring cells and the ECM for growth and survival by producing exoenzymes (9,10). In the triple-species biofilm, E. faecalis used mannose, galactose, and glucose in the ECM as the major carbon sources to engage in central carbon metabolism by upregulating a large number of genes encoding hydrolases, particularly polysaccharide lyases, and genes related to the PTS and ABC transport systems involved in the transport of extracellular sugars and amino acids (47). In addition, upregulation of glycosyltransferase genes may lead to increased catalysis of nucleotide sugars acting as activating monosaccharide donors in E. faecalis, resulting in the production of oligosaccharides/polysaccharides and their derivatives (48). This is similar to the increase in a glucan production caused by the glycosyltransferase GtfB secreted by S. mutans in the enhanced mixed-species biofilms formed by S. mutans and C. albicans (49). Moreover, RNA-seq of E. faecalis in dual-species biofilms (E. faecalis1 E. coli O157:H7) showed increased the expression of genes related to the PTS system and ABC transport system at 24 h (Table S8), indicating that E. faecalis may acquire increased access to more nutrients in dual/multispecies biofilms by regulating the expression of genes related to transport systems. Furthermore, E. faecalis may convert ammonia generated from basic amino acids, including arginine, lysine, and histidine, to active glutamine during the formation of the triple-species biofilm by upregulating the expression of the genes gltA, glnA, and gdhA, and the glutamine is then further catalyzed and processed for aminosugar, purine, and pyrimidine metabolism by upregulating the expression of the genes glmS, purF, carB, and pyraA (11). These observations strongly suggest that E. faecalis shows metabolic flexibility and enhanced environmental tolerance.
Competitive interactions between microorganisms in multispecies biofilms are closely linked with their metabolism (50). Nutrient limitations, such as the low glucose levels encountered by species in a multispecies biofilm microenvironment, can lead to changes in the expression of enzyme genes in central carbon metabolic pathways (51). Thus, the capacity of central carbon metabolism under stress conditions defines the fitness and dominance of species in a polymicrobial community. During the formation of the triplespecies biofilm (24 h) and in response to nutrient limitation, E. faecalis exhibited rapid upregulation of the expression of enzyme genes (pgm and pgi) in glycolytic phase I as well as enzyme genes (zwf, ef1918, and gnd) in the oxidative branch phase of the PPP, thereby shifting the carbon flux of sugar metabolism to the PPP, which may reflect the need for E. faecalis in the triple-species biofilm to generate more reducing power for biosynthesis (29). This is similar to the high expression of genes in the PPP of C. albicans in mixed biofilms formed with Pseudomonas aeruginosa (52). Furthermore, E. faecalis can obtain additional carbon sources by using enhanced glycerol metabolic pathways (deoxygenation pathway GldA and phosphorylation pathway GlpK) (33).
Additionally, E. faecalis may establish its dominance by manipulating the microenvironment via the secretion of "biological weapons." During triple-species biofilm formation, E. faecalis exhibited upregulated expression of ldh-1, encoding LDH, which generates lactic acid during the fermentation process. It is known that E. faecalis acidifies the polymicrobial environment by generating lactic acid as a "biological weapon" to inhibit the growth of other species and facilitate its own growth (30,31,53). This interference competition with other microorganisms mediated by factors such as secreted metabolites has also been reported in previous studies (4,(54)(55)(56). Notably, while acquiring growth advantages, E. faecalis exhibited downregulated expression of virulence genes, including gelE encoding a gelatinase with hydrolase function, entV encoding an enterocin with bacteriocin function (57), and other virulence genes involved in biofilm formation (Table S7) (1,58,59). Acidifying the environment through metabolites and economizing unnecessary and expensive costs, such as the expression of virulence genes, may be beneficial for the dominance of E. faecalis during triple-species biofilm formation.
E. faecalis continuously activates versatile stress response regulators during biofilm formation, enabling its rapid adaptation to its environment (60). During triple-species biofilm formation, genes encoding transcriptional regulators involved in the response, such as GntR, MerR, PadR, ArgR, PerR, AbrB, and LacI, were consistently upregulated in E. faecalis. The key functions of these regulators include the regulation of general metabolism, resistance and detoxification, carbon and nitrogen metabolism, carbon source utilization, arginine metabolism, and oxidative stress (61). These transcriptional responses to a multibacterial environment to enhance the stress response and adaptation capabilities have been described previously (62).
Finally, it has been reported that quorum sensing (QS) can regulate mono-and multispecies biofilm formation by diverse behaviors, such as by controlling the production of matrix components and increasing cell cooperation (63,64). In E. faecalis, QS systems control major virulence determinants that cause nosocomial infections, including several virulence factors, such as the cytolysin operon and the Fsr system (65). The Fsr system indirectly regulates genes that play roles in surface adhesion, autolysis, and biofilm development, including fsrA, fsrB, fsrC, fsrD, gelE, sprE, and ef1097 (65). However, in our RNA-seq results, we did not observe a significant correlation between the changes in the expression of the Fsr system and the enhancement in biofilm biomass. Although it is possible that QS molecules regulate E. faecalis biofilm formation in an as-yetunknown indirect manner, this is beyond the focus of our current study. Taken together, the results of our transcriptome analyses indicated that E. faecalis established its dominance in the triple-species biofilm in four possible ways: (i) by enhancing nutrient transport and biosynthesis of amino acids; (ii) by enhancing central carbon metabolism; (iii) by manipulating the microenvironment through "biological weapons"; and (iv) by activating versatile stress response regulators.
In summary, the coexistence of E. faecalis with E. coli and S. enteritidis enhanced triple-species biofilm formation and led to restructuring of the biofilm into a tower-like architecture (Fig. 7). The enhanced biomass production in the triple-species biofilm was an intrinsic community property, with E. faecalis playing a key role in stabilizing interspecies relationships. This pilot study revealing this characteristic of E. faecalis in triple-species biofilms has the potential to reveal amenable targets for the design of targeted antibiofilm inhibitors and to provide novel insights into the treatment of associated biofilm infections and the removal of environmental contamination in food processing. Further study, including testing other bacterial strains with better mono-species biofilm formation capability, performing a time course assay to monitor multispecies biofilm formation, and dissecting the interspecies interaction in triple-species biofilms at the molecular level, will be performed in the future.

MATERIALS AND METHODS
Strains and growth conditions. E. faecalis, E. coli O157:H7, and S. enteritidis were inoculated in brain heart infusion (BHI) broth overnight at 37°C. E. faecalis and S. enteritidis strains were isolated from an intensive swine farm by our laboratory. E. coli O157:H7, which is a common foodborne pathogen, was purchased from the National Center for Medical Culture Collections (CMCC44939).
Biofilm biomass assay. The cell density of the original "microbial stock solution" for each of the three bacteria was standardized to 1Â 10 8 CFU/mL in BHI medium containing 2% glucose. Then, for mono-species biofilms, 100 mL of the "microbial stock solution" for each of the three species was separately added into the wells of 96-well microplates. In this pilot study, for dual-species biofilms, 100 mL of "microbial stock solution" for each of the two strains was mixed into the 96-well microplates, resulting in a total volume of 200 mL of solution per well. For the triple-species biofilms with different mixing ratios (1:1:1, 2:1:1, 5:1:1, 10:1:1, and 20:1:1), 100 mL of E. faecalis solution was added into the wells of the 96well microplates, and then the corresponding volumes of E. coli and S. enteritidis solutions were added into the wells containing 100 mL of E. faecalis solution, resulting in a series of mixed solutions corresponding to the above-mentioned proportions (300, 200, 140, 120, or 110 mL). Biofilms for all the tests in this study were prepared by culturing the corresponding single or mixed microbial solutions for 24 h or 48 h. The wells were gently washed 3 times with phosphate-buffered saline (PBS, pH 7.2). Prior to biofilm staining, the plates were dried at room temperature in an inverted position overnight. Biofilm quantification assays were performed as described previously by Christensen (66). The wells were stained with 200 mL of 0.1% crystal violet solution for 15 min at room temperature. After washing and airdrying the plates, the biofilm-associated dye was solubilized in 200 mL of destaining solution consisting of ethanol and acetone at a ratio of 4:1 (vol/vol) and quantified by measuring the absorbance at 570 nm (OD 570 ). An equal volume of BHI containing only 2% glucose was used as the negative control to determine the Then, the biofilms were washed, dried, stained, and quantified by measuring the OD 570 . Biofilm biomass assays were performed in four biological replicates and three technical replicates for a total of 12 readings. CLSM. One milliliter of each microbial stock solution was added and mixed into a glass-bottom cell culture dish (NEST, China), resulting in 3 mL of culture for triple-species biofilm formation. One milliliter of E. faecalis stock solution was cultured to form the mono-species biofilm. After incubation for either 24 h or 48 h, the wells were gently washed 3 times with PBS and stained as described below. Live/dead staining of the biofilms was performed with the LIVE/DEAD BacLight Bacterial Viability kit (Invitrogen, USA). The determination of biofilm compositions was carried out according to the study of Tee et al. (67) with minor modifications. First, 1 mL of 1ÂFilmTracer Sypro Ruby biofilm matrix stain (Thermo Fisher Scientific, USA) was added, and the plates were incubated for 20 min. Sypro Ruby was removed, and the wells were washed twice with filtered PBS. Next, 1 mL of 500 mM propidium iodide (Thermo Fisher Scientific, USA) was added, and the plates were incubated for 5 min. The wells were then washed 3 times with filtered PBS. Finally, 1 mL of 0.1% calcofluor white (Sigma-Aldrich) was added for 30 min. After the final incubation, the wells were washed 3 times with distilled water, and fixation was performed with Vectashield Antifade Mounting Medium (Vector Laboratories, USA). Microscopic observation and image acquisition were performed using a Zeiss LSM800 confocal microscope (Zeiss, Germany). Images were analyzed with Zen Black software (Zeiss, Germany). At least three independent experiments were performed on different days, and the images displayed are representative images.
Visualization of biofilm pH with fluorescence microscopy. The biofilm used for visualization of biofilm pH was prepared following the same protocol as that used for CLSM and then stained using the method described in the study by Katherine et al. (68) with minor modifications. After incubating for 12, 24, 36, and 48 h, 1 mM pHrodo Red Dextran (Molecular Probes, Invitrogen, USA) was added to the wells, and the mixture was incubated at room temperature in the dark for 3 h. The biofilms were then washed with PBS three times and stained with SYTO9 (Molecular Probes, Invitrogen, USA) diluted in PBS at room temperature in the dark for 15 min. After the final incubation, the wells were washed 3 times with distilled water and then affixed with Vectashield Antifade Mounting Medium (Vector Laboratories, USA). Fluorescence microscopy was performed using 10Â objective magnification on an Invitrogen EVOS M5000 fluorescence microscopic imaging system. At least three independent experiments were performed on different days, and the images displayed are representative images.
SEM. Then, 0.5 mL of each microbial stock solution was added and mixed into 24-well plates with Aclar coupons (Electron Microscopy Sciences, USA), resulting in 1.5 mL of cultures for triple-species biofilm formation. E. faecalis stock solution (0.5 mL) was cultured under the same conditions to form the mono-species biofilm. The biofilm cultured for either 24 h or 48 h was used for SEM preparation and observation following the standard protocol for SEM. Briefly, the coupons were washed 3 times with PBS and then subjected to fixation in 2.5% glutaraldehyde overnight. Next, fixed samples were washed 3 times with distilled water and then dehydrated using a graded ethanol series (25%, 50%, 75%, 85%, 95% [2Â] and 100% [2Â]), processed for 15 min each time. After drying at room temperature, the samples were sputter-coated with a layer of gold/platinum (CRESSINGTON, the UK). Specimens were evaluated with an FEI Q45 SEM (Thermo Fisher Scientific, USA). At least three independent experiments were performed on different days, and the images displayed are representative images.
CFU enumeration. Biofilms were prepared in the same manner as the sample for the biofilm biomass assay and washed gently 3 times with PBS to remove any unadhered bacterial cells. The bacterial cell aggregates were collected in PBS, vortex shaken for 5 min, and sonicated for 10 min to thoroughly break them up. Then, bacterial cells were diluted and plated onto BHI medium and the corresponding selective media, including Pfizer (E. faecalis), Hektoen Enteric (S. enteritidis), and CT-SMAC (E. coli O157: H7) media, after which CFU were enumerated. The CFU enumeration assay was performed in four biological replicates and three technical replicates for a total of 12 readings.
Extraction and quantification of ECM components. Biofilm ECM crude samples were extracted as previously described by Ramirez et al. (69) with minor modifications. One and a half milliliters of each microbial stock solution was mixed together into a 6-well plate, resulting in a 4.5-mL culture for triple-species biofilm formation. One milliliter of E. faecalis stock solution was cultured to form a mono-species biofilm for either 24 h or 48 h. Biofilms were washed 3 times with PBS, collected and suspended in 10 mL of PBS, and vortexed for 5 min. Then, the samples were sonicated in an ultrasonic ice-water bath for 45 min, including three rounds of 15 min of ultrasonication (5 sec ultrasonic pulses with a 20% amplitude and 5 sec intervals) with interval of 5 min, followed by vortexing for 5 min and centrifugation at 3500 rpm for 15 min (Heraeus Multifuge X1R, Thermo Fisher Scientific, USA). The supernatant fraction was recovered and filtered through 0.2 mm acrodisc syringe filters with Supor Membrane (Pall Life Sciences, USA). Next, total exopolysaccharides and eDNA were obtained and quantified in accordance with the method described by T Ramirez et al. The Bradford Protein assay kit (Solarbio, China) was used to quantify the concentration of proteins in the supernatant filtrate. ECM component quantification was performed in four biological replicates and three technical replicates, resulting in a total of 12 readings.
Transcriptome analysis using RNA-seq. RNA-seq was performed by Personal Biotechnology Company (Shanghai, China) using the Pacific Biosciences platform and the Illumina NovaSeq platform. Three biological replicate biofilms were prepared in the same way as sample preparation for ECM component extraction and quantification. RNA from E. faecalis in the mono-species biofilm and triple-species biofilm samples was extracted using TRIzol Reagent (Invitrogen Life Technologies, USA). RNA purity was determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific, USA), and RNA integrity was assessed using the RNA Nano 6000 assay kit for the Bioanalyzer 2100 system (Agilent Technologies, USA). A portion of the RNA was used for qPCR. Library preparation was completed using the NEBNext Ultra II RNA Library Prep kit for Illumina (NEB, USA) following the manufacturer's recommendations. The library was then sequenced on the NovaSeq 6000 platform (Illumina). Cutadapt software v1.15 was used to filter the sequencing data to obtain high-quality sequences for further analysis. The filtered reads were mapped to the reference genome (E. faecalis V583) using HISAT2 v2.0.5, followed by quantification of the gene expression level. HTSeq statistics were used to compare the read count values for each gene as the original expression level of the gene. Then, FPKM was used to standardize the expression levels. Differential expression analysis of three biological replicates was performed using DESeq (1.30.0) with screened conditions as follows: expression difference multiple jlog2FoldChangej $1 and corrected P value , 0.05. GO enrichment analysis of DEGs was performed by topGO. ClusterProfiler (3.4.4) software was used to test the statistical enrichment of DEGs in the KEGG pathways.
Verification of DEGs by qPCR. A subset of genes that were differentially regulated in the triple-species biofilm (i.e., ebpA, ebpB, gnd, gap-1, and gpm) was verified by qRT-PCR using ChamQ Universal SYBR qPCR Master Mix (number Q711-02, Vazyme Biotech Co., Ltd., China). Partial primers were designed using Primer v5.0 software (Table 2), and melt curve analysis was performed at the end of each amplification run to verify signal specificity. The results are presented as the relative expression levels normalized to the level of the housekeeping gene 16S rRNA. The assay was performed in three biological replicates and three technical replicates.
Data accessibility. The accession number for the RNA-seq and related metadata reported in this paper is NCBI: PRJNA755819. Other data that support the findings of this study are available from the corresponding author upon reasonable request.

ACKNOWLEDGMENTS
This study was supported by funds to Yabin Wang from the Key Laboratory for Animal-derived Food Safety of Henan Province, and the startup grant to Youbao Zhao from Henan Agricultural University (30500946). The funders had no role in study design, data collection or analysis, decision to publish, or preparation of the manuscript.
We declare that we have no conflicts of interest.