The MPB strains included in this study are listed in Supplementary Table S1. E. coli O157:H7 strain 1934 was originally recovered from beef and kindly provided by Dr. Alexander Gill (Health Canada, Ottawa, ON, Canada) and a non-biofilm former (Yang et al., 2023). Generic E. coli were recovered from beef cuts and trimmings as well as beef fabrication equipment at processing plants in previous studies and genotyped using multiple-locus variable-number tandem-repeat analysis (MLVA) (Yang et al., 2017a; Yang et al., 2015; Yang et al., 2017b). Genotypes that were recovered only from a single sampling visit were regarded as non-persistent, while those that were recovered from more than three sampling visits at the same facility were regarded as persistent. A total of 10 generic E. coli strains consisting of 5 genotypes selected at random from each of these 2 groups were included. In a previous study, 567 MPB were recovered from conveyor belts at a Canadian beef packing plant, which belonged to 40 genera (Wang et al., 2018). From these isolates, an isolate from each genus was randomly selected, consisting of 18 Gram-negative aerobic bacteria (GNA), 8 Gram-positive aerobic bacteria (GPA), 5 lactic acid bacteria (LAB), and 9 Enterobacteriaceae (ENT). In total, 50 bacterial isolates were included in the study, with 40 MPB and 10 generic E. coli.

All bacterial isolates were stored at −80°C in half-strength Brain Heart Infusion broth (BHI; Oxoid, Mississauga, ON, Canada) containing 15% (v/v) glycerol (Fisher Scientific, Edmonton, AB, Canada), and working cultures were maintained on Trypticase Soy Agar (TSA; Oxoid) at 4°C with the monthly transfer. Lennox broth without salt (LB-NS, 10 g/L of tryptone and 5 g/L of yeast extract; Oxoid) was used for cultivating all bacterial isolates (Visvalingam et al., 2019), except for Acinetobacter haemolyticus, Flavobacterium columnare, Pedobacter ginsengisoli, and Aerococcus urinaeequi. These four isolates did not grow well in LB-NS (optical density at 600 nm < 0.1 after 120 h incubation at 25°C), so they were grown in BHI. A single colony from an agar plate was picked and inoculated into 10 mL of LB-NS or BHI and incubated in a shaking incubator operated at 80 rpm and 25°C until the stationary phase. Each bacterial culture grown in LB-NS was diluted 100-fold in the same media to obtain a bacterial suspension containing approximately 10⁷ CFU/mL of cells. For bacterial strains grown in BHI, a 1 mL portion of each bacterial suspension was centrifuged at 10,000 × g for 10 min at 4°C to pellet cells. The resulting pellet was re-suspended in 1 mL of LB-NS and diluted 100-fold in LB-NS. A total of 51 diluted suspensions were used as inoculum for the biofilm experiments below.

Biofilms were grown using a device consisting of a 96-peg lid (Nunc Immuno TSP lid; Fisher) fitted to a 96-well round bottom microtiter plate (Nunc; Fisher). Inocula of 50 co-cultures were prepared by mixing an equal volume of E. coli O157:H7 and each of the 50 MPB inocula suspensions. A 160 μL of aliquot from each of the 51 strains or each of the 50 co-culture inocula was added to duplicate wells. Duplicate wells containing 160 μL of non-inoculated LB-NS were included as blank. Inoculated plates were fitted with pegged lids and incubated at 15°C for up to 6 days.

Biofilms formed on pegs were quantified using the previously described crystal violet (CV) staining method (Visvalingam et al., 2019). Briefly, after 2, 4, or 6 days of incubation, loosely attached planktonic cells were removed by successively placing pegged lid into two microtiter plates containing 160 μL of phosphate-buffered saline (PBS; Hardy Diagnostics, Santa Maria, CA, United States) per well. For each wash, pegs were incubated in PBS for 1 min at ambient temperature. Then, biofilms formed on pegs were stained for 20 min by placing the pegged lid into a microtiter plate with 160 μL of 0.1% (w/v) aqueous CV solution in each well, followed by rinsing off excess CV using two successive PBS washes as described above. The rinsed pegged lid was placed into a microtiter plate containing 180 μL of 80% (v/v) ethanol (Azer Scientific Inc., Morgantown, PA, USA) in each well for de-staining. After 20 min of de-staining, the pegged lid was replaced with a regular lid without pegs (Nunc microwell lid, Fisher) and absorbance of CV was determined at 570 nm (A₅₇₀) using a POLARstar Omega microplate reader (BMG LABTECH GmbH, Ortenberg, Germany). Blank-corrected values generated by microplate data analysis software (BMG LABTECH) were used for analysis. Three independent experiments were performed for each monoculture and for each co-culture combination.

To understand how differences in biofilm-forming ability as determined by CV staining in comparison with that determined by enumerating viable bacterial cell numbers, a strong biofilm-forming E. coli genotype 136 (EC136), moderate biofilm-forming Acinetobacter haemolyticus; strong biofilm-forming Sphingopyxis bauzanensis and weak biofilm-forming Carnobacterium maltaromaticum were selected. Interaction between these strains and E. coli O157:H7 was also investigated. To avoid disrupting biofilm while removing pegs from lids, aseptically removed pegs from lids (from Nunc Immuno TSP lid) were used. For monoculture biofilm, 160 μL of EC136, E. coli O157:H7, A. haemolyticus, S. bauzanensis, or C. maltaromaticum inoculum was added into a 2-ml micro-centrifuge tube. Subsequently, an aseptically removed peg was placed into the inoculum to cover two-thirds of its length and incubated for 4 days as described above. To obtain co-culture biofilms, a total volume of 160 μL of EC136, A. haemolyticus, S. bauzanensis, or C. maltaromaticum and E. coli O157:H7 (80 μL each) were added to a 2-ml micro-centrifuge tube with a peg and incubated as described above. On day 4, the pegs were removed and each was washed twice using 160 μL of PBS. The washed pegs were each transferred to 15-ml centrifuge tubes containing 3 mL of 0.1% (w/v) peptone water and 0.3 g of glass beads (500 μm; BioSpec Products, Bartlesville, OK, USA) and vortexed at the maximum speed for 1 min. The resulting suspension was serially diluted in 0.1% peptone water and appropriate dilutions were spread-plated on selective agar plates. For the E. coli genotype 136 and E. coli O157:H7 combination, MacConkey agar (MAC; Oxoid,) and Sorbitol MacConkey agar supplemented with Cefixime-Tellurite (CT-SMAC; Oxoid) plates were used for enumeration of the total bacterial number and E. coli O157:H7, respectively. The C. maltaromaticum and E. coli O157:H7 combination was enumerated using All Purpose TWEEN^® agar (Oxoid) supplemented with 20 mg/L of nalidixic acid (APT-NA; Sigma-Aldrich, Oakville, ON, Canada) and CT-SMAC, respectively (Edima et al., 2007). For A. haemolyticus, S. bauzanensis, and E. coli O157:H7 combinations, TSA and CT-SMAC plates were used for enumeration of the total bacterial numbers and E. coli O157:H7, respectively. Inoculated MAC and CT-SMAC plates were incubated at 35°C for 24 h, while inoculated APT-NA and TSA plates were incubated at 25°C for up to 72 h. Then, plates containing 30 to 300 colonies were counted. Experiments were independently conducted twice, and three pegs were analyzed in each experiment.

Meat-processing plants generally conduct cleaning and disinfection of the surfaces of the facility after daily operation. A commonly used chlorinated alkaline cleaner (Powerfoam Plus, Epsilon Chemicals Ltd., Edmonton, AB, Canada) and quaternary ammonium compound (QAC)-based sanitizer (E-San, Epsilon Chemicals Ltd), were selected to investigate the potential effects of the relevant chemicals used in this procedure on removing bacteria in biofilms. Dilutions of Powerfoam Plus (final concentration, 2.5%, v/v) or E-San (final concentration, 200 ppm) were prepared by following the manufacture’s instructions right before use. The combination of E. coli O157:H7 and Acinetobacter haemolyticus, which showed consistent synergistic interactions in biofilms, was included in the test.

First, the microscopy method was used to examine the effects on biofilms. For this, biofilms were grown at 15°C for 4 days on cover glasses (18*18 mm; VWR, Edmonton, AB, Canada) placed in wells of a 12-well tissue culture plate (VWR) with each well containing 2 mL of mono- or co-culture of E. coli O157:H7 and/or A. haemolyticus on day 0. The biofilm-bearing cover glass was treated by being consecutively soaked in 2 mL of water, Powerfoam Plus/0.85% NaCl, water, and E-San/0.85% NaCl for 1 min (Figure 1a). Saline water (0.85% NaCl) was used as a control for Powerfoam Plus and E-San treatments. The treated cover glasses were washed twice using Difco™ Neutralizing Buffer (VWR, Edmonton, AB, Canada) and/or saline water to remove loosely attached bacterial cells (Figure 1a). The remained bacteria/biofilms on the cover glasses were examined using an Olympus light microscope BX53 equipped with a digital camera DP80 and cellSens imaging software at 1,000 × magnification.

In addition, viable bacterial cells were further enumerated for biofilms treated by E-San, Powerfoam Plus, and saline water (control), respectively. For this purpose, biofilms were developed at 15°C for 4 days on detached pegs in a 96-well plate with each well containing 160 μL of mono- or co-culture. The biofilms were treated by following the procedure shown in Figure 2a. The number of E. coli O157:H7, A. haemolyticus, or total bacteria in treated biofilm was enumerated as described above.

Mean A₅₇₀ from CV staining was calculated. The cutoff value for biofilm formation (A_570C) was calculated using the equation: A_570C = Mean A₅₇₀ of blank +3 × standard deviation of A₅₇₀ for blank (Stepanović et al., 2007). Based on A₅₇₀ values, strains were classified into four categories: non-biofilm former, weak, moderate, or strong biofilm former (Table 1). The interactions between two bacterial strains in a biofilm were regarded as synergistic, with no effect or antagonistic if the A₅₇₀ value of dual culture is larger than, equals to, or smaller than the higher of the A₅₇₀ values of the two relevant monocultures (Ren et al., 2015; Ren et al., 2014). Bacterial counts and A₅₇₀ values were analyzed using a one-way ANOVA with the help of GraphPad Prism 10 (GraphPad Software, Boston, MA, United States). The Tukey’s test was used to assess pairwise differences between means, with a significance level of p ≤ 0.05 and mean values were presented with standard error of means. Two-way ANOVA was performed to analyze the effect of treatment and culture combinations on the bacterial numbers in the biofilm treated with different cleaning and sanitizing agents.

Among 18 GNA and 8 GPA, 61.1% and 25–37.5% of isolates formed monoculture biofilm between days 2 and 6, respectively (Table 2). The majority of those GNA and GPA strains formed weak or moderate biofilm with the exception of the GNA stains Brevundimonas staleyi, Massilia aurea, and Stenotrophomonas maltophilia, and the GPA strain Macrococcus caseolyticus, all of which formed strong biofilm at one or more sampling points (Figures 3, 4a). The E. coli O157:H7 strain was a non-biofilm former with A₅₇₀ values below the cutoff value of 0.21. When it was co-cultured with GNA, more GNA formed biofilm on days 2 and 4 than on day 6 (Table 2). Unlike GNA, fewer GPA strains formed co-culture biofilm on days 2 and 6 than on day 4.

Of the five LAB isolates tested, only C. maltaromaticum formed measurable, but weak biofilms (Figure 4b). When co-cultured with E. coli O157:H7, only C. maltaromaticum and Vagococcus fluvialis formed biofilms (Figure 4b). At day 2, 66.7% of ENT strains formed biofilms which reached 88.9% by day 4 with no further increase by day 6 (Table 2), with Citrobacter gillenii, Buttiauxella noackiae, and Salmonella sp. being strong biofilm formers (Figure 4c). When co-cultured with E. coli O157:H7, all strains that formed monoculture biofilms also formed dual-species biofilms by day 4 (Figure 4b). As previously described (Visvalingam et al., 2017), biofilm formation of GEC strains remained at 80% between days 2 and 6 (Table 2). All persistent GEC genotypes and three of the non-persistent GEC genotypes formed biofilms. In general, persistent genotypes formed stronger biofilm than non-persistent ones, more so on day 2 than on days 4 and 6 (Figures 5a, b). All persistent and non-persistent strains that formed mono-culture biofilms also formed dual-culture biofilms with E. coli O157:H7.

Biofilm-forming interaction was classified as no-effect, synergy, and antagonism based on A₅₇₀ values observed between monoculture and dual-culture biofilms. Overall, the interactions between MPB and E. coli O157:H7 in their respective dual-culture biofilms were mostly no-effect (Figure 6). Consistent interactions with E. coli O157:H7 in co-culture biofilms were observed for the GNA isolates B. staleyi and S. bauzanensis (antagonistic; #14 and #26), and A. haemolyticus (synergistic; #12) during the study period (Figures 3, 6). GPA isolates Microbacterium sp., (#29) and Macrococcus caseolyticus (#32) formed antagonistic and synergistic relationship, respectively (Figures 4a, 6). Interestingly, with the commonly regarded biocontrol agents, lactic acid bacteria, only synergistic effects were observed (C. maltaromaticum, #38; V. fluvialis, #41). None of the ENT isolates had interactions in co-culture biofilms except for Salmonella sp. (#50) which was antagonistic. Persistent E. coli isolates mainly showed synergistic interactions when forming biofilms with E. coli O157:H7, while non-persistent isolates showed no effect.

EC136, A. haemolyticus, S. bauzanensis, and C. maltaromaticum selected based on their biofilm-forming ability and interactions with E. coli O157:H7 were further assessed by enumeration in single and dual-culture biofilms with E. coli O157:H7 (Table 3). In monoculture, the number of viable E. coli O157:H7 cells was 7 log CFU/peg. The viable cell numbers of E. coli O157:H7 did not differ significantly (p > 0.05) when it formed dual-species biofilms with A. haemolyticus, S. bauzanensis, or C. maltaromaticum. In contrast, viable numbers of E. coli O157:H7 were significantly lower (p < 0.05) when it formed biofilm with EC136. Viable numbers of EC136 and C. maltaromaticum did not differ between monoculture and dual-culture biofilms. Viable numbers of A. haemolyticus were significantly higher in dual-species biofilm containing E. coli O157:H7 than its monoculture biofilm while that of S. bauzanensis was lower (p < 0.05) in dual species than in its monoculture biofilms.

Due to the synergistic effect between A. haemolyticus and E. coli O157:H7, this combination was included to test the effects of a commonly used cleaner and sanitizer in removing biofilms or bacteria in biofilms. Microscopic observation of dual- (Figure 1b) and mono-culture biofilms (Supplementary Figure S1) showed an unnoticeable effect by E-San treatment alone. In contrast, Powerfoam Plus treatment alone removed a large portion of bacteria in both types of biofilms. The sequential treatment of Powerfoam Plus and E-San was more effective than Powerfoam Plus alone for mono-culture biofilms (Supplementary Figure S1), however, no difference was noticed for dual-culture biofilms (Figure 1b). Further comparison of Powerfoam Plus with E-San by enumerating bacteria in treated biofilms showed that non-significant but numerically stronger (total bacteria in co-culture biofilms; E. coli O157:H7 in mono-culture biofilms) or statistically significant stronger effects (E. coli O157:H7 in dual-culture biofilms; A. haemolyticus in mono-culture biofilms) by Powerfoam Plus in removing bacteria (1–3 log CFU) in biofilms than E-San (Figure 2b). No significant difference was observed between E-San and control (saline water).