A total of 9 STEC strains including 7 serogroups, 12 SP, and 12 LAB were assessed for their suitability to use in this study (Table 1). STEC strains were cultured on MacConkey agar plates (Hardy Diagnostics Inc., Santa Maria, CA, United States), while SP and LAB bacteria were cultured on Trypticase Soy Agar (TSA; Becton, Dickinson and Company, MD, United States) at 25°C. A single colony of each STEC, LAB, and SP strain was transferred from each plate into individual 10 ml Lennox broth with no salt (LB-NS; Tryptone 10 g/L and yeast extract 5 g/L) and grown to a cell density of 10⁸ CFU/ml. The incubation time required for the strains to reach early stationary phase varied from 24 to 72 h (Visvalingam et al., 2019a). Cultures were subsequently diluted to 10⁶ CFU/ml for use in biofilm formation assessment assays.

To simulate the nutrient profile within beef fabrication plants, LB-NS broth was supplemented with sterile beef purge (Pang and Yuk, 2018) that originated from vacuum-packed beef product (i.e., the eye of round). The vacuum package was opened, the beef purge was collected and distilled water was added at a ratio of 1:6. The aqueous solution was then passed through a 0.45 μm sterile filter (Midelet and Carpentier, 2002), protein content was determined using a Bradford kit (Thermo Scientific, Rockford, IL, United States) and the filtrate was stored at −20°C. The filtrate was mixed with LB-NS broth (10% v/v; mLB-NS) for use in biofilm formation assays.

Scanning electron microscopy (SEM) was performed at the Manitoba Institute for Materials (MIM) to visualize dry biofilm formation on SS and TPU surfaces as described previously (Adator et al., 2018). The TPU and SS coupons were fixed (neutral buffered 10% formalin solution, Sigma–Aldrich) for 2 h and then washed with BPB for 30 min. The coupons were dried for 4 h at room temperature in a BSL2 cabinet, and the TPU surface was Gold–Palladium coated (Denton Vacuum Desk II, Moorestown, NJ, United States) in the high-vacuum mode on the following day. Biofilm structures were observed using a Quanta 650 FEG scanning electron microscope (FEI Co., Hillsboro, OR, United States) in the high-vacuum mode at 5 KV.

All experiments were performed three times. The Proc Mixed procedure of the Statistical Analysis System (Cary, NC, United States) was used to analyze the data with the least mean separation accomplished using the PDIFF option. For biofilm-forming ability and multispecies biofilm microplate assays, a factorial model was applied to analyze the main effects of bacterial strain, temperature, and their two-way interaction. For beef contaminated by O103:H2 on food contact surfaces, effects of species, contact surface, storage time, and humidity along with the appropriate interactions were tested. For all statistical analysis, significance was declared at P ≤ 0.05.

Isolates varied substantially in their biofilm-forming ability (Figure 1). A bacterial isolate by temperature interaction was identified (P < 0.001) as some isolates more readily formed biofilms at 25°C than others. For example, at 25°C many strains showed strong biofilm-forming abilities, which included all of the STEC strains except O145:H2 (weak) and O157:H7 1934 (intermediate), (Figure 1A). Within the LAB, L. bulgaricus, L. curvatus, Lactobacillus sakei S206, and Aerococcus viridans were all classified as strong biofilm formers at 25°C. Likewise, P. aeruginosa, Rahnella sp., R. terrigena, and E. coli (8_77) also formed strong biofilms at this temperature. In contrast, Serratia sp. and R. terrigena isolates were able to form strong biofilms at 10°C, while other isolates were intermediate or weak biofilm formers at this temperature (Figure 1B).

Evaluation of curli and cellulose indicated that strains O26: H11 (00-3941), O103: H2 (99-2076), O111: NM (CFS3), O121: H19 (03-2832), O157: H7 (R508), and E. coli (8_77) showed both curli- and cellulose-producing ability at 25°C (Table 4). Based on these data, L. sakei S19, Serratia sp., C. koreensis, R. terrigena, L. bulgaricus, L. curvatus, C. divergens B1, C. piscicola, P. aeruginosa, and O103:H2 were selected for further investigation in multi-species biofilms.

At 25°C, spoilage of P. aeruginosa + C. koreensis (T3) biofilms reduced (P < 0.001) the transfer of O103:H2 to beef by 2.54 log₁₀ CFU/g (Figures 3A–C). Mixed species C. piscicola + L. bulgaricus (T1) and C. koreensis + R. terrigena (T2) biofilms did not alter (P > 0.05) the transfer of O103:H2 cells to beef. Overall transfer of O103:H2 to beef from biofilms formed on TPU (2.14 log₁₀ CFU/g) was greater (P < 0.001) than that from SS (1.40 log₁₀ CFU/g). Transfer of O103:H2 to beef decreased (P < 0.001) with biofilm aging, from 3.17 log on Day 6 to 1.52 log₁₀ CFU/g on Day 30 and 0.62 log₁₀ CFU/g on Day 60. Reductions in the transfer of O103:H2 to beef were highest for 6 days T3-mixed biofilms grown on TPU. Overall moist biofilms transferred more O103:H2 (P < 0.001) to beef (2.93 log₁₀ CFU/g) than dry biofilms (0.61 log₁₀ CFU/g) regardless of the surface type.

At 10°C, none (P > 0.05) of the multispecies biofilms reduced or enhanced the transfer of O103:H2 to beef as compared to the control positive (Figures 3D–F). Transfer of O103:H2 to beef was also higher (P < 0.001) from TPU (1.14 log₁₀ CFU/g) than from SS (0.55 log₁₀ CFU/g). Results also showed that moist biofilms were more (P < 0.001) likely to contaminate beef (1.58 log₁₀ CFU/g) than dry biofilms (0.10 log₁₀ CFU/g). At 10°C, contamination of beef with O103:H2 decreased (P < 0.001) as biofilms aged, from 1.38 log₁₀ CFU/g after 6 days to 0.47 log₁₀ CFU/g after 60 days.

At 25°C, no O103:H2 was recovered (0%) from 60 days dry biofilms on SS, even after enrichment (Table 5). On TPU, O103:H2 was recovered from approximately 33% of dry 60 days multispecies of C. koreensis + R. terrigena (T2) and control positive (T4) biofilms. However, O103:H2 was not recovered from the following multispecies biofilms combinations C. piscicola + L. bulgaricus (T1) and P. aeruginosa + C. koreensis (T3).

At 10°C after enrichment, O103:H2 was most often recovered from mixed biofilms of T2 formed on TPU/SS and stored for 60 days (Table 5). Recoveries of O103:H2 from dry biofilms on TPU ranked as T2 (89%) > T4 (44%) > T1 (33%) > T3 (22%). On SS surfaces, the highest O103:H2 recovery rate from dry biofilm was as follows T2 (33%) > T4 (11%) = T1 (11%) > T3 (0%).

After 60 days storage, dry multispecies biofilms composed of C. koreensis + R. terrigena (T2) were observed as a multilayer structure with rod-shaped bacteria covered by extensive EPS at both 25°C (Figure 4) and 10°C (Figure 5). Interestingly, control positive (T4) dry biofilm was displayed differently after 60 days storage at 25°C (Figure 4) and 10°C (Figure 5). At 25°C, a well-developed multilayer T4 biofilm extensively covered the whole TPU surface. However, no individual bacteria and EPS were observed when the T4 biofilm was stored at 10°C for 60 days.

In this study, strains isolated from beef-processing facilities (e.g., L. sakei S19, C. maltaromaticum and R. terrigena) and meat products (Serratia sp.) were found to be stronger biofilm formers at 10°C than those (e.g., L. bulgaricus and L. curvatus) isolated from the fermented product. This is showing that isolates from beef production facilities are likely adapted to growth and form biofilms at low temperatures (Visvalingam et al., 2019a). In Canada, beef fabrication facilities operate at temperatures below 10°C to limit the growth of enteric pathogens and spoilage bacteria (Visvalingam et al., 2017; Yang et al., 2017a,b). However, temperatures at the beef fabrication facility vary and can reach up to 15°C during non-operational hours (Visvalingam et al., 2017). In addition, the equipment used during beef fabrication and other interventions can generate higher temperatures, such as frictional heat produced by conveyor belts or the transfer of body heat to gloves, and the application of high-pressure hot water (40–50°C) during sanitation. Thus temperature variation and microenvironments created in the beef-processing facility attributed to different factors could promote the formation of biofilms within the beef-processing environment (Yang et al., 2015, 2017a; Visvalingam et al., 2016).

Curli fimbriae and the production of cellulose have been reported to play a significant role in STEC biofilm formation and persistence (Gualdi et al., 2008; Iibuchi et al., 2010; Adator et al., 2018). Curli play a significant role in mediating surface and cell-to-cell contact in E. coli and Salmonella biofilms (Barnhart and Chapman, 2006). Uhlich et al. (2014) demonstrated that curli and cellulose formation are influenced by temperature and media composition (Uhlich et al., 2014). Results demonstrate STEC strain O145:H2 (75–83), and O157:H7 (1934) showed curli-producing ability at 37°C instead of at 25°C, while only STEC O157:H7 (R508) and E. coli (8_77) kept cellulose-producing capacity at 37°C (Supplementary Table 1). Except for the strong biofilm former O157:H7 (1931), STEC lacked curli or did not produce cellulose and were only able to form weak or intermediate biofilms at 25°C (Table 4). Wang et al. (2012), also reported that the occasional strain of STEC that lacked curli could still produce strong biofilms at room temperature (Wang et al., 2012). Stanford et al. (2021) also found that some STEC strains that produced curli did not form strong biofilms. This indicates that traits other than just curli and cellulose production are likely to mediate biofilm formation in some STEC strains (Wang et al., 2012; Visvalingam et al., 2017). For instance, some research has shown that biofilm formation accompanied by massive EPS production is highly reliant on E. coli metabolic energy support (Landini, 2009). However, energy metabolisms among the top six STEC serogroups have been reported to be highly diverse, which might be attributed to the genetic diversity among serogroups (Chen et al., 2020).

Some studies have shown that E. coli found on the surface of fabrication equipment can survive sanitation procedures and contaminate meat (Yang et al., 2015; Visvalingam et al., 2016). STEC attached to food contact surfaces may also interact with pre-established multispecies biofilm (Wang, 2019). In the present study, well-structured multilayer multispecies biofilms were developed (Figures 4, 5) and O103:H2 within these biofilms was able to transfer to beef (Figure 3), Furthermore, Visvalingam et al. (2019a) demonstrated that STEC O157:H7 readily integrated (3.8 log₁₀ CFU/cm²) into multispecies biofilms containing 41 different bacterial strain isolated from beef-packing plant (Visvalingam et al., 2019a). In the same study, the relative abundance of participating strains (n = 42) did vary; as some strains such as Carnobacterium sp. accounted for (10%) of the biofilm membership, whereas O157:H7 accounted for only 0.04% of the community (Visvalingam et al., 2019a).

Contamination of beef by O103:H2 was substantially reduced after interaction with P. aeruginosa + C. koreensis (T3) biofilms. Both Pseudomonas sp. (Pang et al., 2017; Pang and Yuk, 2018) and Comamonas sp. (Carpentier and Chassaing, 2004) have been shown to form robust biofilms on stainless steel surfaces. In the present study, pre-existing biofilms of P. aeruginosa + C. koreensis on TPU and SS may have inhibited O103:H2 integration into biofilms. Wang et al. (2015) investigated mixed biofilms of STEC serotypes O157:H7 and O111:H8 and found that the STEC serotype that was inoculated onto the surface first, exhibited the dominant membership within mature biofilms (Wang et al., 2015). Most biofilm studies have inoculated similar numbers of different bacterial species onto food contact surfaces (Wang R. et al., 2013; Liu et al., 2014; Pang et al., 2017). Within beef fabrication facilities, probably multispecies biofilms are already established on contact surfaces (Wang, 2019). These mature biofilms may preclude the integration of STEC due to a lack of adhesion sites or available nutrients (Pang and Yuk, 2018). Hence, developing biofilms composed of pre-selected innocuous bacterial species may inhibit STEC biofilm formation and have merit as a biocontrol strategy (Alegre et al., 2013). On the other hand, C. piscicola + L. bulgaricus (T1) and C. koreensis + R. terrigena. (T2) biofilms did not affect the extent that O103:H2 contaminated beef. The impact of mixed-species biofilms on the contamination of meat is likely species dependent and possibly strain dependent (Wang, 2019). For example, Pseudomonas sp. have been shown to inhibit the formation of E. coli O157:H7 biofilms on SS (Kim et al., 2018) and Salmonella biofilms on TPU (Wang H.-H. et al., 2013). Pseudomonas sp. are known to produce antimicrobials such as pyocyanin, pyoluteorin, and siderophores, which may inhibit the integration of foreign bacteria into biofilms (Hernández-León et al., 2015; Collazo et al., 2017; Kim et al., 2018). Moreover, P. aeruginosa may also produce acyl homoserine-lactones (AHL) (Lee et al., 2007), which have been shown to inhibit E. coli biofilm formation by altering gene expression (Van Houdt et al., 2006).

Certain groups of LAB such as Lactobacillus sp. are commonly used as probiotics, as many of these isolates produce bacteriocins and organic acids (e.g., lactic acid) (Schrezenmeir and De Vrese, 2001; Imani Fooladi et al., 2014). Previous research indicates that some Lactobacillus spp. can reduce the shedding of E. coli O157:H7 in cattle feces (Brashears et al., 2003; Younts-Dahl et al., 2004). If added to ground beef at 5°C, they can also reduce the prevalence of E. coli O157:H7 and Salmonella (Smith et al., 2005). Carnobacterium sp. are frequently isolated from beef fabrication facilities, and can persist on non–food-contact surfaces after sanitization (Wang et al., 2018). Interestingly, multiple studies indicate that Carnobacterium sp. can inhibit the growth of Listeria monocytogenes on meat by producing bacteriocins, but this species can also cause food spoilage (Leisner et al., 2007). In the present study, no synergistic or antagonistic interactions on beef contamination were observed between O103:H2 and C. piscicola + L. bulgaricus biofilms. Furthermore, extracts of C. piscicola and L. bulgaricus did not exhibit activity against O103:H2 in clearing zone assays (data not shown). Most LAB bacteriocins target a narrow range of bacteria and primarily target Gram-positive bacteria such as Listeria (Jones et al., 2008). Jones et al. (2008) tested 75 meat-related LAB strains and none of them showed antimicrobial activity against E. coli O157:H7 (Jones et al., 2008). Others have shown that Lactobacillus sp. and Lactococcus sp. bacteriocin activity against E. coli is strain dependent (Gómez et al., 2016). Lactobacillus sp. can also produce lactic, acetic, and propionic acids, which can inhibit the growth of pathogenic bacteria (Abedi et al., 2013; Jalilsood et al., 2015). However, after 6 days the alkaline extract (pH > 8) was produced by C. piscicola + L. bulgaricus (T1) biofilms, suggesting that the production of ammonia from amino acid metabolism may have neutralized any antimicrobial activity of organic acids.

Stainless steel and thermoplastics are two of the most common food contact surfaces used in the food industry (Chia et al., 2009; Sofos and Geornaras, 2010). In this research, the transfer of O103:H2 to beef from biofilms on TPU was greater than that from SS (Figure 3). In previous studies, E. coli O157:H7 and L. monocytogenes were able to form stronger monoculture biofilms (higher bacteria count) on polyurethane plastic than on stainless steel (SS-304) (Midelet and Carpentier, 2002; Graziella et al., 2006; Sofos and Geornaras, 2010). The stronger biofilms on polyurethane surfaces may be related to its greater hydrophobicity than stainless steel (Donlan, 2002). It has been hypothesized that as bacteria cells irreversibly attach to solid surfaces, hydrophobic surfaces may have less electrostatic repulsive forces (Loosecht et al., 1987; Sinde and Carballo, 2000; Donlan, 2002). Other bacteria such as Salmonella and Listeria have also been found to more readily attach and form biofilms on surfaces that are more hydrophobic (Sinde and Carballo, 2000; Donlan, 2002). Apart from beef contamination, the different food contact surfaces also affected O103:H2 recovery from 60 days dry biofilms (Table 5), with O103:H2 being more readily recovered from multispecies on TPU than SS at both 10 and 25°C. A similar result has been reported by Adator et al. (2018), as recovery rates from STECs biofilms were higher on polystyrene than SS (Adator et al., 2018). Previous studies have demonstrated that conveyor belts are often linked to the contamination of beef trimmings and cuts with E. coli in beef fabrication facilities (Youssef et al., 2013; Yang et al., 2015; Visvalingam et al., 2016). Regarding the efficiency of routine commercial sanitation processes in beef-processing facilities, published work shows that the sanitation process cannot completely remove E. coli from the conveyor belt (Yang et al., 2015, 2017b; Visvalingam et al., 2016). This limited E. coli removal has been attributed to meat residues, which can reduce the efficacy of sanitizers and the impact of desiccation on the viability of E. coli (Yang et al., 2015, 2017b; Visvalingam et al., 2016). Our study did find that O103:H2 within 60 days dry biofilms did not transfer to beef (Figure 3). In contrast, O103:H2 bacteria within moist biofilm readily transferred to beef even after 60 days of storage at 10 or 25°C (Figure 3). As indicated by Gill and Landers (2004), others have shown that desiccation can reduce the transfer of E. coli from fabrication equipment to beef (Gill and Landers, 2004; Youssef et al., 2013). However, humid conditions are prevalent in the beef industry due to condensation originating from the routine use of hot water during the sanitation processes (Møretrø et al., 2010), which could re-hydrate dry biofilms allowing bacteria within the biofilm to thrive, persist, and spread. Typically, the relative humidity within beef-processing plants is high, varying from 40 to 97% during the day with peak humidity reached during sanitation (Møretrø et al., 2010). Meat residues on the conveyer belts, combined with the high relative humidity are factors that undoubtedly contribute to the formation of robust surface biofilms (Yang et al., 2015, 2017b; Visvalingam et al., 2016) or maintenance of dry old biofilms. Results obtained in this research showed that prolonged storage time was associated with a decrease in transfer to O103:H2 from biofilms to beef at both 10 and 25°C. During prolonged storage, nutrients within biofilms may become limiting and a buildup of waste products may also reduce cell viability within biofilms (Rendueles and Ghigo, 2015). However, dry biofilms can pose a cross-contamination risk if those dry biofilms come in contact with meat juices and water since STEC can still be viable within the biofilm.

Although O103:H2 associated with 60 days dry biofilms did not transfer to beef, viable O103:H2 could still be recovered from dry biofilm after enrichment (Table 5). Others have also found that dormant STEC cells can be recovered from dry biofilms via enrichment, which mimics the conditions at the meat plants where dry biofilms could rehydrate with water and beef juices allowing bacteria to recover (Adator et al., 2018). After enrichment, O103:H2 recovery from C. koreensis + R. terrigena (T2)-mixed biofilm at 10°C was always higher than from positive control (T4) 60 days dry biofilms (Table 3). Furthermore, SEM images indicated that 10°C. koreensis + R. terrigena (T2) 60 days dry biofilm displayed well-structured multilayered biofilm on TPU and SS surfaces (Figure 5). Instead, 10°C positive control (T4) 60 days dry biofilm was non-observed on both surfaces (Figure 5). This is showing that spoilage bacteria naturally occurring in beef-processing environments, that could survive the sanitation process and are adapted to lower temperatures, could shelter STEC and allow its persistence. Other researchers have made similar observations for Pseudomonas–Salmonella mixed biofilms, where the presence of P. aeruginosa enhanced Salmonella typhimurium and Salmonella enteritidis resistance to sanitizers (Pang et al., 2017). P. aeruginosa has been shown to produce more EPS (e.g., glycoconjugates) in mixed-species biofilms, thicker EPS hinders sanitizer penetration and thus protects Salmonella against sanitizers (Pang et al., 2017; Pang and Yuk, 2018). During biofilm development, EPS are secreted by the bacterial community, complex matrixes are formed which in turn embed bacterial cells (Flemming and Wingender, 2010) and protect them from desiccation while trapping nutrients (Kumar and Anand, 1998; Stewart and Franklin, 2008). Therefore, it is possible that C. koreensis + R. terrigena (T2) produced a more complex EPS matrix at 10°C (Figure 5) that enhanced the ability of O103:H2 to persist in desiccated multispecies biofilms. In beef-processing facilities, dry biofilms on beef fabrication equipment that come in contact with beef purge or water may result in conditions that promote cell viability similar to enrichment (Skandamis et al., 2009; Adator et al., 2018). If this is the case, even old dry biofilms on beef fabrication equipment could continuously pose a risk of beef contamination (Skandamis et al., 2009; Yang et al., 2017a; Adator et al., 2018).

To limit enteric pathogen growth on food contact surfaces and the product low environmental temperatures (5–15°C) is maintained in the beef industry (Ma et al., 2019, 2020). Our research shows that a temperature of 10°C reduced O103:H2 biofilm formation and thus cell transfer to beef when compared with biofilms formed and kept at 25°C. However, a higher O103:H2 recovery rate was observed in 60 days dry biofilms formed and maintained at 10°C instead of 25°C, the highest O103:H2 recovery was obtained from C. koreensis + R. terrigena (T2)-mixed biofilm at 10°C. Interestingly, both strains, C. koreensis and R. terrigena can form biofilms at 10°C. This finding suggests that despite reducing STEC biofilm formation in beef fabrication facilities through temperature control, STEC can still pose a cross-contamination risk to beef, especially under the assistance of psychotrophic spoilage bacteria in the beef-processing environment.

Scanning electron microscope imaging was used to verify the biofilm formation on the food contact surface, important to mention that one of the disadvantages of using SEM to research biofilms is that biofilm structure could be altered due to the fixation procedures, mainly when studying wet biofilms, however, in this research biofilm structure was not part of the objectives. The present study focused on STEC, spoilage, and LAB interactions, mainly to assess STEC cross-contamination potential. Multispecies biofilm structure or composition was not part of our research objectives. Further research is needed to investigate the dynamic spatial distribution of STEC within the multispecies biofilm, for which different microscopy techniques are needed, such as confocal laser scanning microscopy (CLSM) (Chen et al., 2015). In addition, multiple latest studies indicate that the metabolic diversity among STEC serogroups might lead to varied stress responses (Chen et al., 2020; Wang et al., 2022; Zhao et al., 2022), suggesting the interaction in multispecies biofilm may be varied according to different STEC serogroups. Hence, further studies could attempt to investigate the general interaction between SP or LAB multispecies biofilm with STEC (e.g., top 7 STEC) and also could look into STEC genetic makeup. Such work could facilitate the development of biofilm management strategies for STEC in beef-processing environments.