Biofilm formation by heat-resistant dairy bacteria: multispecies biofilm model under static and dynamic conditions

ABSTRACT

In the food industry, especially dairy, biofilms can be formed by heat-resistant spoilage and pathogenic bacteria from the farm. Such biofilms may persist throughout the processing chain and contaminate milk and dairy products continuously, increasing equipment cleaning, maintenance costs, and product recalls. Most biofilms are multispecies, yet most studies focus on single-species models. A multispecies model of dairy biofilm was developed under static and dynamic conditions using heat-resistant Bacillus licheniformis, Pseudomonas aeruginosa, Clostridium tyrobutyricum, Enterococcus faecalis, Streptococcus thermophilus, and Rothia kristinae isolated from dairies. C. tyrobutiricum and R. kristinae were weak producers of biofilm, whereas the other four were moderate to strong producers. Based on cross-streaking on agar, P. aeruginosa was found to inhibit B. licheniformis and E. faecalis. In multispecies biofilm formed on stainless steel in a CDC reactor fed microfiltered milk, the strong biofilm producers were dominant while the weak producers were barely detectable. All biofilm matrices were dispersed easily by proteinase K treatment but were less sensitive to DNase or carbohydrases. Further studies are needed to deepen our understanding of multispecies biofilms and interactions within to develop improved preventive strategies to control the proliferation of spoilage and pathogenic bacteria in dairies and other food processing environments.

IMPORTANCE

A model of multispecies biofilm was created to study biofilm formation by heat-resistant bacteria in the dairy industry. The biofilm formation potential was evaluated under static conditions. A continuous flow version was then developed to study multispecies biofilm formed on stainless steel in microfiltered milk under dynamic conditions encountered in dairy processing equipment. The study of biofilm composition and bacterial interactions therein will lead to more effective means of suppressing bacterial growth on food processing equipment and contamination of products with spoilage and pathogenic bacteria, which represent considerable economic loss.

INTRODUCTION

Raw milk is a highly nutritious medium with a near neutral pH and supports vigorous growth of a wide variety of microorganisms, especially bacteria (1). Most microbial contamination of raw milk occurs on the farm, where any part of the environment may be involved, for example, soil, bedding, manure, udders, milking equipment, the milk cooling and storage tank, and so on (1, 2). Bacteria such as Bacillus licheniformis can contaminate and spoil milk (3). Pseudomonas aeruginosa is at fault for color defects in dairies (4). Clostridium tyrobutyricum, with a production of butyric acid, causes late blowing defect in hard and semi-hard cheese (5). Enterococcus faecalis is a contaminant in milk and an opportunistic pathogen (6). Streptococcus thermophilus is responsible for milk acidification (7). Rothia kristinae is ubiquitous in the dairy environment (8). To limit the growth of these species as well as other spoilage and pathogenic bacteria, raw milk is kept at temperatures between 4°C and 7°C until processing so that thermal treatments such as pasteurization can inactivate vegetative microbial cells (1, 9). However, spore-formers and other heat-resistant bacteria can survive pasteurization and remain viable in milk products, thus exposing consumers to danger and constituting a liability for the dairy industry (10, 11). Some of these bacteria form biofilms on milk-processing surfaces including stainless steel. Biofilms are multicellular communities of bacteria living in and multiplying in a matrix made of polysaccharides or other polymers of bacterial origin (12). Once formed, a biofilm allows the bacteria residing therein to resist the cleaning and disinfecting processes usually applied to dairy processing equipment (13), thus creating an inexhaustible supply of live bacterial cells that become contaminants in milk and dairy products. The resulting increased frequency of food spoilage and reduced product shelf life, not to mention more laborious cleaning procedures imposed on the dairy farm or processing plant, lead to considerable economic losses for dairy producers (14).

Although environmental biofilms are usually multispecies, most studies are based on single species biofilms, making our current understanding of the complexity of biofilms insufficient (15, 16). Multispecies biofilm appears to be more stable than single-species biofilm (17), and little is understood about the interactions between species in biofilms (18). These interactions may be competitive (e.g., for nutrients or space) or supportive (e.g., protection against chemical attack or other antimicrobials). Communication within a biofilm is an important factor in the evolution of the biofilm mass (18, 19). Lessening the burden of spoilage biofilms in the food industry and especially in dairies will require much deeper knowledge of how multispecies biofilms form and how microbial cells within them interact, especially heat-resistant bacteria.

The aim of this study was to examine the biofilm-forming capacity of six dairy bacterial strains in a dynamic system mimicking the dairy setting using a Centers for Disease Control and Prevention (CDC) biofilm reactor that applies continuous stirring and shear forces.

MATERIALS AND METHODS

Bacterial strains and growth conditions

Four heat-resistant bacteria isolated previously from Quebec dairy farm bulk tank raw milk warmed to 63°C for 10 min were later identified as Bacillus licheniformis B149, Rothia kristinae basonym Kocuria kristinae M57, Streptococcus thermophilus M388, and Enterococcus faecalis M458 (20). Pseudomonas aeruginosa QF2-56 was isolated from milk tank piping on a Quebec dairy farm (Gagnon et al., unpublished). Clostridium tyrobutyricum isolate MK183, provided by a local dairy processing plant, was identified presumptively by MALDI-TOF mass spectrometry. The P. aeruginosa and C. tyrobutyricum strains were identified also using partial 16S rRNA gene sequencing as described previously (20). Their heat resistance was evaluated in this study. Phylogenetic affiliation was determined using the Nucleotide Basic Local Alignment Search Tool (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The 16S rRNA gene sequences are available in GenBank under accession numbers OP584935 to OP584940. The strains were grown at 30°C for 24 h in tryptic soy broth (TSB, BD Bacto, Franklin Lakes, NJ, USA), de Man, Rogosa and Sharpe broth (MRS, Sigma-Aldrich), or reinforced Clostridium medium (RCM, EMD Millipore, Darmstadt, Germany) according to the assignments shown in Table 1.

TABLE 1.

Microbial strains used in this this study

Isolate	Identity	Origin	Factor of concern	Culture conditions
B149	Bacillus licheniformis	Bulk tank	Spore formation, proteolytic enzymes	Tryptic soy broth (TSB, BD Bacto, Franklin Lakes, NJ, USA), 24 h, 30°C
M388	Streptococcus thermophilus	Bulk tank	Milk acidification	de Man, Rogosa and Sharpe (MRS) medium (Sigma-Aldrich), 24 h, 30°C
M458	Enterococcus faecalis	Bulk tank	Proteolytic enzymes	MRS, 24 h, 30°C
M57	Rothia kristinae	Bulk tank	Contamination of raw milk	TSB, 24 h, 30°C
MK183	Clostridium tyrobutyricum	Local dairy plant	Spore former, butyric acid production, late blowing defect in cheese	Reinforced Clostridium medium (EMD Millipore, Darmstadt, Germany), 96 h, 30°C, anaerobic
QF2-56	Pseudomonas aeruginosa	Milk pipeline	Proteolytic enzymes, milk spoilage	TSB, 24 h, 30°C

Heat resistance

The ability of the six strains to withstand High Temperature, Short Time (HTST) pasteurization was evaluated previously (21) with modifications. After two washes with phosphate-buffered saline (PBS 1X: 0.137 M sodium chloride, 0.01 M sodium phosphate dibasic, 0.0027 M potassium chloride, 0.0018 M potassium phosphate monobasic, all Thermo Fisher Scientific, Ottawa, ON, Canada), cultures in microfiltered milk (Lactantia PurFiltre brand, Lactalis Canada, Toronto, ON, Canada) were pasteurized for 15 s at 72°C. Colonies were plated and counted on MacConkey Agar without crystal violet (MCWCV, Criterion, Hardy Diagnostics, Santa Maria, CA, USA) before and after pasteurization under anaerobic conditions for C. tyrobutyricum and under aerobic conditions for the other strains. For each isolate, heat resistance was based on a single criterion: cell death <1 log cfu/mL.

Interaction between bacteria

Antagonistic interactions between pairs of strains were assessed using the cross-streak method as previously described (22). A single isolate cultured in microfiltered milk was streaked onto MCWCV and incubated for 24 h at 30°C. A second culture was then streaked onto the same plate perpendicular to the first streak, followed by an additional 24 h of incubation. Interactions involving C. tyrobutyricum were made under anaerobic and aerobic conditions whereas the interactions with other bacteria were made under aerobic conditions only. Decreasing outgrowth from the second streak as it approached the first streak indicated antagonistic interaction. Both strains in each pairing were tested as the first streak and as the second streak.

Biofilm formation under static conditions

In microtiter plates (96 well)

Single-species and multispecies bacterial biofilms were formed in wells of Corning Costar #3,595 plates (Thermo Fisher Scientific) as described previously (23) with modifications. Briefly, 24 h culture was diluted to 2 × 10⁶ cfu/mL in TSB, MRS, RCM, or microfiltered milk. Three wells each were loaded with 200 µL of each diluted culture, and the plates were incubated at 30°C for 24 h. The wells were then washed three times with PBS 1X. For biofilm grown in microfiltered milk only, a fourth wash was necessary and the wells were then filled with tris HCl buffer pH 7.5 (50 mM tris HCl, 1 mM CaCl₂, Sigma-Aldrich Darmstadt, Germany) and held at room temperature for 1 h and washed twice with PBS 1X. The plates were dried at 30°C for 30 min, and the biofilm was stained with 0.1% (wt/vol) crystal violet (Thermo Fisher Scientific) for 15 min, rinsed three times with PBS 1X, dried again at 30°C for 30 min, and then decolorized with 200 µL of 50% (vol/vol) ethanol (95%, Greenfield, Boucherville, QC, Canada) +acetic acid (99.8% Sigma-Aldrich) to release the crystal violet, which was quantified by measuring absorbance at 570 nm (Tecan Infinite 200 PRO spectrometer, Männedorf, Switzerland). The ability to form biofilm was classified as weak (Absorbance at 570 nm, A₅₇₀ ≤0.5), moderate (0.5 > A₅₇₀ ≤1.5), or strong (A₅₇₀ ˃ 1.5) using an arbitrary scale with Staphylococcus epidermidis as a control (23).

Dispersion of biofilm by enzymatic treatment

Biofilm matrix composition of the single species biofilm and the multi-species biofilm in microfiltered milk was studied using enzymatic treatments described previously (24) with modifications. Biofilm grown for 24 h in microtiter plate wells was washed three times with PBS 1X buffer and then 200 µL of enzyme solution was added separately in each well: DNase I [500 µg/L in 150 mM NaCl (Thermo Fisher Scientific), 1 mM CaCl2 (Sigma-Aldrich)], proteinase K [500 µg/L in 50 mM tris HCl (Sigma-Aldrich) and 1 mM CaCl2 (Sigma-Aldrich)], pectinase [1 mg/mL in 0.1 M potassium phosphate (Sigma-Aldrich)], 200 µL of amylase [1 mg/mL in 0.1 M potassium phosphate (Sigma-Aldrich)], alginate lyase [1 mg/mL in 0.1 M potassium phosphate (Sigma-Aldrich)], and cellulase [1 mg/mL in 0.1 MOPS (Sigma-Aldrich)]. All enzyme solutions were from Sigma-Aldrich. For each enzyme, control wells received 200 µL of PBS to compare the values of untreated biofilms against those treated with enzymes. All wells were then held at 37°C for 1 h and then washed twice with PBS 1X and dried. The biofilm was tested with crystal violet as described above, and the difference between A₅₇₀ of the treated and untreated biofilms was used to evaluate the effect of each enzymatic treatment.

Biofilm formation under dynamic conditions

In a CDC biofilm reactor

Biofilm was formed on stainless steel slides (CBR 2128–316, 76 mm × 15 mm, BioSurfaces Technologies Corporation, Bozeman, MN, USA) in a CDC reactor (CBR 90, BioSurfaces Technologies Corporation) according to the protocol described in ASTM standard E2562-17 (25) with modifications. Biofilm formation by each bacterial isolate in microfiltered milk was assessed first, followed by co-culture of all strains. The inoculum was 10⁶ cfu/mL in all cases. Eight CBR 2203 Gl slide holder rods (BioSurfaces Technologies Corporation) holding the slides were immersed in 340 mL of microfiltered milk inoculated with 1 mL of inoculum. Stirring was set at 130 rpm, and the temperature was maintained at 30°C using a hotplate (VWR International, NJ, USA). After 24 h of batch mode growth, the reactor was emptied, and fresh microfiltered milk was introduced using a peristaltic pump (HV-77913–70 Masterflex, Cole-Parmer Canada, Laval, QC, Canada) at a flow rate of 11.8 mL/min for 24 h. The effluent from the reactor was collected in a carboy. Viable bacteria in the biofilm were counted on MCWCV and quantified by qPCR (gene copy number) based on cells treated with propidium monoazide (26).

Viable counts of bacteria in biofilm

Biofilm formed on the stainless-steel slides was recovered in 40 mL of PBS 1X buffer in 50 mL screw-cap tubes (Sarstedt, Montreal, Québec), which were vortex agitated for 30 s followed by 30 s in an ultrasonic bath, both repeated three times. Suspensions were diluted in PBS 1X for viable counts on MCWCV incubated at 30°C for 72 h in an anaerobic jar containing an oxygen-free gas generator (Mitsubishi AnaeroPack-Anaero, Thermo Fisher Scientific, Ottawa, Ontario) for C. tyrobutyricum MK183 or for 48 h under aerobic conditions for the other strains.

Propidium monoazide (PMA) treatment

Samples from the CDC biofilm reactor were treated with PMA to account for cell viability. The suspension in PBS 1X was centrifuged at 12,000 × g for 10 min at 4°C. The cell pellets were then mixed with 500 µL of Tris-EDTA 2X pH 8 (20 mM Tris-HCl +2 mM EDTA, both Sigma-Aldrich) and 1.25 µL of PMA solution (Biotium, Fremont, CA, USA). Samples were then incubated and shaken for 5 min in the dark at 150 rpm. The PMA-treated samples were placed on ice 20 cm from a halogen lamp and then centrifuged for 10 min at 12,000 × g. The samples were kept at −80°C until DNA extraction.

DNA extraction

DNA was extracted using the MasterPure Gram-positive DNA purification kit protocol (Lucigen, Madison, WI, USA) with modifications. Briefly, the PMA-treated samples were mixed with 150 µL of Tris-EDTA buffer (Lucigen) containing 30 U/mL of lysozyme (Lucigen) and 10 U/mL of mutanolysin (Sigma-Aldrich) for enzymatic lysis. The remaining steps were followed according to the recommendations of the kit manufacturer. The extracted DNA was stored at −20°C until quantification.

Quantitative PCR

The primer sets used for amplification are listed in Table 2. All PCR amplification solutions (10 µL) were composed of 3.6 µL UltraPure DNAse and RNAse free water, 5 µL of PowerUp SYBR Green master mix (Thermo Fisher Scientific), 0.2 µL each of forward and reverse primer (10 nM), and 1 µL of DNA sample, except for E. faecalis M458, in which case these solution proportions were, respectively, 3.8 µL, 5 µL, 0.1 µL of each primer and 1 µL of sample. Amplification was run on a ViiA7 real-time PCR system (Thermo Fisher Scientific) with one cycle at 50°C for 2 min, one at 95°C for 10 min, 40 cycles with denaturation at 95°C for 15 s and primer annealing, and DNA extension at 60°C for 1 min, followed by a melting curve step. Standard curves made from single-species biofilms were used to quantify gene copy numbers using droplet digital PCR (ddPCR) on the IBIS (Institut de biologie intégrative et des systems at Université Laval) sequencing platform as described previously (27). Standard curve reactions were performed in triplicate. All standard curve efficiencies were between 90% and 110% with R² values superior to 0.98. Biofilm bacterial biomass, expressed in gene copy number per cm² of stainless steel slide, was calculated from the corresponding standard curve regression equation. Analysis was run using QX Manager Software Standard Edition version 1.2 for the QX200 Droplet Digital PCR System.

TABLE 2.

Primers used for sequencing

Isolate	Primer	Sequence	Gene	Reference
Bacillus licheniformis	Forward	CCTACGGGAGGCAGCAGTAG	rRNA 16S	(28)
	Reverse	GCGTTGCTCCGTCAGACTTT
Streptococcus thermophilus	Forward	TAGTTCGCTTTGGAAACTGTCAAC	rRNA 16S	This study
	Reverse	CTCCATATATCTACGCATTTCACCG
Enterococcus faecalis	Forward	CCCTTATTGTTAGTTGCCATCATT	rRNA 16S	(29)
	Reverse	ACTCGTTGTACTTCCCATTGT
Rothia kristinae	Forward	GGCATCACCATCAACATCGC	tuf	This study
	Reverse	GTTCTTCACGTAGTCCGCGT
Clostridium tyrobutyricum	Forward	AAGGGAAGTGCACAACATGA	rRNA 16S	This study
	Reverse	ACTACCAGGTGCTTTTAAATTTGC
Pseudomonas aeruginosa	Forward	ACTTTAAGTTGGGAGGAAGGG	rRNA 16S	(30)
	Reverse	ACACAGGAAATTCCACCACCC

Protein assay

The protein content of biofilms formed in the CDC biofilm reactor was measured using the colorimetric DC Protein Assay (Bio-Rad, Mississauga, Ontario), as recommended by the manufacturer, on samples recovered from stainless steel slides in PBS buffer as described above. Bovine serum albumin (BSA, 500–0112, Bio-Rad, Mississauga, Ontario) in PBS buffer (five dilutions, starting from 1.5 mg/mL) was used as the protein standard.

Carbohydrate assay

The carbohydrate content of the biofilms formed in the CDC biofilm reactor was measured using a colorimetric assay performed in 96-well microtiter plates as described previously (31) with modifications. Dextran standard (Sigma-Aldrich) in PBS 1X buffer, with 1:2 serial dilution starting from 2 mg/mL to 0.0625 mg/mL, was used as the standard. Briefly, 125 µL of 0.5% (vol/vol) periodic acid diluted in 5% (vol/vol) acetic acid (both Sigma-Aldrich) were deposited in each well. Sample or standard (25 µL) was then added, and the plate was held at 37°C for 1 h. Schiff’s reagent (100 µL) was then added, and the plate was incubated for an additional 4 h. Absorbance at 570 nm was measured using a Tecan Infinite 200 PRO spectrometer.

Scanning electron microscopy

Scanning electron microscopy (SEM) was used to confirm the three-dimension conformation of the biofilms. Biofilm grown on 0.5-inch stainless steel coupons (RD128-316, BioSurfaces Technologies Corporation) held in CBR 2203 coupon holder rods (BioSurfaces Technologies Corporation) in the CDC biofilm reactor (BioSurfaces Technologies Corporation) with microfiltered milk as described above was fixed with 2.5% glutaraldehyde (Electron Microscopy Sciences, Hatfield, PA, USA) and 4% formaldehyde (Electron Microscopy Sciences) in 0.1 M Na cacodylate buffer pH 7.4 (Electron Microscopy Sciences) then dried in increasing gradients of ethanol (30%, 50%, 70%, 90%, 95%, 100%, 100%, and 100% for 10 min at each concentration) at room temperature (standard protocol for the IBIS imaging platform). The fixed biofilms were coated with epoxy resin and visualized using a scanning electron microscope (JSM-6360LV model, JEOL, Tokyo, Japan). Images were acquired using the software provided by the manufacturer. Those shown are representative of several observations. Only biofilms with an extracellular matrix were observed.

Statistical analysis

All experiments were conducted in triplicate. Mean and standard deviation (SD) were calculated for each count or colorimetric assay. One-way analysis of variance (ANOVA) followed by the Tukey multiple comparisons test (GraphPad Prism version 8 for Windows, GraphPad Software, San Diego, CA, USA) was used to compare the six strains against each other as well as multispecies samples in all experiments involving biofilm formation under static conditions. Two-way ANOVA followed by a Tukey multiple comparisons test was performed to compare cfu or gene copy numbers between single-species and multispecies biofilms. Differences were considered statistically significant at P < 0.05.

RESULTS

Heat resistance

Based on the reductions in viable cell counts after pasteurization, all six strains qualified as heat-resistant (Fig. 1). Bacterial cell death ranged from 0.02 log to 0.6 log cfu/mL, with B. licheniformis B149 and C. tyrobutyricum MK183 showing significantly more resistance than S. thermophilus M388, indicating their greater ability to withstand the heat treatment applied.

Fig 1.

Difference between viable counts (log cfu/mL) of the six bacterial species before and after pasteurization (72°C, 15 s) expressed as ∆log cell death. Results are presented as mean ± SD (error bars). * =P ≤ 0.05 indicates that the difference is significantly different.

Interaction between bacteria

The only antagonistic activity noted was that of P. aeruginosa QF2-56 against B. licheniformis B149 and E. faecalis M458 when streaked first. All other strains were able to grow next to each other within the confluence area (see example of an inhibition zone in Fig. S1 in Supplemental material).

Biofilm formation under static conditions

C. tyrobutyricum MK183 and R. kristinae M57 were shown to be weak biofilm producers (A₅₇₀ ≤0.5) in TSB (Fig. 2A). No significant difference was found between the biofilm-forming abilities of these two strains or between B. licheniformis B149, P. aeruginosa QF2-56, and S. thermophilus M388, which were strong producers (A₅₇₀ >1.5). E. faecalis M458 was shown to be a moderate biofilm producer (0.5 > A₅₇₀ ≤1.5), significantly weaker than B. licheniformis B149, P. aeruginosa QF2-56, and S. thermophilus M388. Results in MRS broth were similar (Fig. 2B) with P. aeruginosa QF2-56 and S. thermophilus M388 both giving significantly lower absorbance than the multispecies mixture. In RCM broth (Fig. 2C), only B. licheniformis B149 was shown to be a strong biofilm producer, with no significant difference compared to the multispecies mixture. The other strains were weak producers (C. tyrobutyricum MK183, R. kristinae M57, and S. thermophilus M388) or moderate producers (P. aeruginosa QF2-56 and E. faecalis M458). In milk (Fig. 2D), C. tyrobutyricum MK183 was still a weak biofilm producer, while R. kristinae M57 was a moderate producer but significantly weaker than the remaining strains, which were strong biofilm producers, like the multispecies mixture. In all culture media, the multispecies biofilms showed the same absorbance as the strongest single-species biofilm.

Fig 2.

Biofilm formation by the six bacterial species or mixture grown in (A) tryptic soy broth, (B) de Man, Rogosa and Sharpe broth, (C) reinforced Clostridium medium, and (D) microfiltered milk. Results (see Materials and Methods for details of the optical absorbance measurement) are presented as mean ± SD. The asterisks indicate statistically significant differences.

Enzymatic breakdown of biofilms

With a mean of 27% of remaining biofilm, B. licheniformis B149 biofilm was the most affected by proteinase K (Fig. 3A). Statistically, no difference was noted with QF2-56, the second most affected by proteinase K (Fig. 3A). The remaining biofilm of B149 was significantly very low compared to M57 and M388 and significantly low compared to MK183, M458, and the multispecies biofilm (Fig. 3A). The other enzymes had minimal effect on the biofilms, with nearly 100% of the biofilms remaining intact (Fig. 3B through F). However, additional testing of amylase (Fig. 3D) will be necessary, as M388 and M458 appeared to be slightly affected by this enzyme.

Fig 3.

Tenacity of biofilm after treatment with (A) proteinase K, (B) DNase I, (C) pectinase, (D) amylase, (E) alginate lyase, and (F) cellulase. Results are presented as mean ± SD. * =P ≤ .05 indicates that the difference is significantly different; ** =P ≤ .01 indicates that the difference is very significantly different.

Biofilm formation under dynamic conditions

Based on viable counts, B. licheniformis B149 was more viable by 4 log cycles in multispecies biofilm than in single-species growth (Fig. 4A). However, slightly more gene copies (1 log cycle) were found in the single-species biofilm. E. faecalis M458 counts differed by 1 log cycle in favor of multispecies biofilm and were high (12–13 log cfu/cm²) in either case (Fig. 4B). Also, P. aeruginosa QF2-56 growth was of the same order (12 log cfu and 7 log gene copies per cm²) in both situations (Fig. 4C). Streptococcus thermophilus M388 viable counts were favored significantly in the multispecies biofilm (11 log versus 8 log cfu/cm²) whereas gene copies appeared to number about 8 log/cm² in both cases (Fig. 4D). Based on viable counts, R. kristinae M57 appeared to be a relatively weak (5 log (cfu/cm²). However, they could not be quantified within the multi-species biofilm as a high dilution factor was needed to differentiate bacteria within the multi-species biofilm and M57 got diluted too much. The qPCR data showed a positive amplification of R. kristinae M57 biofilm both as a single species and in the multispecies biofilms averaging 3 log (gene copies/cm²). Regarding C. tyrobutyricum MK183, no viable cells (cfu/cm²) were quantified in a single-species nor in the multispecies biofilms but using qPCR, 1.2–1.7 log gene copies per cm² were detected in both biofilms (Fig. 4F).

Fig 4.

Counts of viable bacterial cells in single-species and multispecies biofilms grown in a CDC reactor. (A) B. licheniformis B149, (B) E. faecalis M458, (C) P. aeruginosa QF2-56, (D) S. thermophilus M388, (E) R. kristinae M57, and (F) C. tyrobutyricum MK183. The cell concentrations were determined by viable counts in cfu/cm² and by PMA-qPCR in gene copy number (gc) /cm². Absent values were below the detection limit. Results are presented as mean ± SD. Asterisks indicate statistically significant differences. * =P ≤ .05 indicates that the difference is significantly different; ** =P ≤ .01 indicates that the difference is significantly very different.; *** =P ≤ .001 and **** =P ≤ .0001 both indicate that the results are significantly highly different.

Scanning electron microscopy

Examination of biofilm microstructure (Fig. 5) revealed that in all cases the cells formed a 3D network with depth and very few planktonic cells. P. aeruginosa QF2-56 (Fig. 5A) formed a more organized multilayered surface than did B. licheniformis B149, E. faecalis M458, and R. kristinae M57, which lacked this sponge-like appearance (Fig. 5B, D and E). In these latter cases, the cells were difficult to distinguish from the matrix. The C. tyrobutyricum MK183 biofilm appeared to consist mostly of cells embedded deeply in matrix (Fig. 5C), whereas S. thermophilus M388 cells were abundant on the surface of the matrix (Fig. 5F). The multispecies biofilm had a rich 3D structure in which bacilli (perhaps mostly B. licheniformis B149 and/or P. aeruginosa QF2-56) and cocci were abundant (Fig. 5G).

Fig 5.

Scanning electron micrographs (magnification 2000X) of biofilms formed in microfiltered milk by (A) P. aeruginosa QF2-56, (B) B. licheniformis B149, (C) C. tyrobutyricum MK183, (D) E. faecalis M458, (E) R. kristinae M57, (F) S. thermophilus M388, and (G) the multispecies mixture. Images are representative of a large number of observations.

Protein and polysaccharide contents

The protein content of single-species and multispecies biofilms (Fig. 6A) ranged from 0.4 mg/mL to 1 mg/mL and was greatest in those formed by P. aeruginosa QF2-56, E. faecalis M458, and the multispecies mixture, three sample groups that did not differ significantly. Carbohydrate contents appeared to differ, but this was not statistically significant because of the large variance (Fig. 6B). They ranged from about 0.2 mg/mL to slightly over 0.5 mg/mL.

Fig 6.

Protein (A) and carbohydrate (B) contents of single-species and multispecies biofilms grown in the CDC reactor. Values represent mean ± SD. Asterisks indicate statistically significant differences. * =P ≤ .05 indicates that the difference is significantly different; ** =P ≤ .01 indicates that the difference is very significantly different.

DISCUSSION

We have studied here the biofilm-forming ability of six reportedly heat-resistant bacterial strains from dairy farms or processing plants, in pure culture and co-culture. Resistance to the pasteurization condition known as high temperature/short time was confirmed. The largest drop in viable count was 1 log, and B. licheniformis B149 and C. tyrobutyricum MK183 counts fell by, respectively, 0.02 and 0.04 log. These two strains form endospores, which are well known to survive pasteurization and become vegetative cells under permissive conditions (32). The heat resistance of the other strains is believed to be due in part to a protective effect of milk or milk proteins (33). Resistance of the six bacterial strains from this study to pasteurization is a major cause for concern since their growth in milk can pose a health hazard in addition to causing economic losses due to product spoilage.

The formation of biofilms in milk processing equipment depends largely on the ability of bacteria to attach to surfaces (9). The bacteria in this study possessed strong, weak, or intermediate biofilm-forming capacities. For example, B. licheniformis B149 formed strong biofilms under static conditions on polystyrene and under dynamic conditions on stainless steel. This is consistent with previous observations of strains of B. licheniformis on stainless steel (12, 34). P. aeruginosa isolate QF2-56 also produced strong biofilms under both static and dynamic conditions. This was observed on polystyrene in tryptic soy broth, in contrast with a previous report that P. aeruginosa was either a non-producer (A₅₉₀ <0.1) or a weak producer (0.1 < A₅₉₀ <1) in this culture medium (35). Strain, therefore, must not be overlooked when characterizing the biofilm production capability of a bacterial species. We found S. thermophilus M388 to be a strong producer, which has been noted in another study (7). Our results for E. faecalis M458 on both polystyrene and stainless steel at 30°C are consistent with reports of abundant production by E. faecalis strains on polystyrene at 25°C and 37°C (9) and to some extent on stainless steel at 27°C (36). Temperatures in the 25–37°C range thus appear to be optimal for biofilm formation by E. faecalis, and dairies experiencing quality problems due to this thermophile should check for gaps in their cold chain (7). On the other hand, R. kristinae M57 and C. tyrobutyricum MK183 were weak producers on polystyrene and on stainless steel. Although C. tyrobutyricum is a well-known spore-forming anaerobe found in raw milk and cheese (37) and R. kristinae M57 is also found in raw milk (38), few studies on their biofilm-forming properties in dairies have been reported. In fact, the weak biofilm formation by C. tyrobutyricum MK183 may be a result of the aerobic conditions in which the biofilm growth was performed or it may be due to a formation of spores as a way of protection instead of a biofilm. Though less likely to form a tenacious biofilm on their own, they may infiltrate a strong biofilm and thereby persist in a dairy production environment. This would still raise concern as C. tyrobutyricum causes late blowing defects during cheese ripening (2) and R. kristinae, a common species on dairy farms, can find its way into pasteurized milk (20, 39). An important finding in our study is that all the strains were able to attach to polystyrene and stainless steel and are, therefore, quite polyvalent. In addition, richer culture media supported stronger biofilms. It has been demonstrated that biofilms are stronger when nutrients are more abundant or supplied continuously (40).

In dairies and other non-laboratory settings, biofilms are usually communities of bacterial species (19). Interactions within the biofilm determine its proprieties. Understanding these interactions should suggest ways to prevent biofilm formation or maturation (19). For example, P. aeruginosa QF2-56 was antagonistic toward B. licheniformis B149 and E. faecalis M458. It is known that P. aeruginosa can detect the presence of other species and respond by secreting molecules that give it a competitive advantage (41). It would be interesting to identify the P. aeruginosa metabolites that prevented the other two bacterial strains from growing in the zone of confluence. Such metabolites could be useful in new strategies of biofilm control. Given that P. aeruginosa QF2-56 had antagonistic effects toward B. licheniformis B149 and E. faecalis M458, one might expect B149 and M458 viable cell counts to be affected in the multispecies biofilm; however, this is not the case. One explanation would be that the antagonistic effect appears only when P. aeruginosa QF2-56 grows before B. licheniformis B149 and E. faecalis M458, whereas in the mixed biofilm all the bacteria are inoculated at the same time. This hypothesis could be studied by growing a P. aeruginosa QF2-56 biofilm first and then challenging it with other strains. We also found that in multispecies biofilms formed on stainless steel in the CDC biofilm reactor, the four species that could be quantified by viable counts were the strong producers of biofilm under static conditions, namely, B. licheniformis B149, P. aeruginosa QF2-56, S. thermophilus M388, and E. faecalis M458. B. licheniformis is one of the most common species found in multispecies biofilms (12) and P. aeruginosa has also been described as the most dominant in multispecies biofilms formed in milk (42). Although all six strains were detected using qPCR or ddPCR, C. tyrobutyricum MK183 and R. kristinae M57 were present in low gene copy numbers suggesting that they were overwhelmed by the other bacteria or were mainly viable but not culturable. Other studies suggest that the growth rate of a bacterial strain or isolate in milk may affect the biofilm composition (19).

In biofilms, bacteria are embedded in an extracellular matrix composed mainly of bacterial polysaccharides. Other matrix components may include DNA, proteins, lipids, and materials gathered from the biofilm environment (14). In the present study, single-species and multispecies biofilms obtained under static conditions were dispersed readily by proteinase K but not by the other enzymes used. This suggests that protein was a major component of the matrices or that the polysaccharide portion was not an alpha or beta 1,4 glucoside or an alginate. Similar results were obtained with biofilm formed in the CDC biofilm reactor. The composition of the matrix usually reflects the environment and the bacterial species present (14). Since milk is an excellent source of protein, biofilm matrices produced by bacteria growing therein may be expected to contain polypeptides in larger proportions (33). It has also been reported that strong producers of biofilm cause caseins to coagulate and that protein clumps could be important stabilizers of biofilms (43).

Scanning electron micrography confirmed that biofilm structure and the distribution of bacteria in the matrix varied with the environment. For example, in the biofilm of C. tyrobutyricum MK183, the matrix is the predominant component observed, along with what appear to be extracellular materials. This is probably due to the strict anaerobic nature of this species, which requires protection from oxygen. The single-species biofilms of B. licheniformis B149, R. kristinae M57, E. faecalis M458, and S. thermophilus M388 were similar in organization. Matrix structure can be the result of adherence of bacterial cells to milk proteins (7). Our observations are consistent with this, except for the wide differences in the cell density, notably between E. faecalis M458 and R. kristinae M57. Confocal microscopy could be used to study bacterial cell localization and dispersion in greater detail as well as biofilm overall composition and distribution of the different matrix components.

To conclude, the heat-resistant bacteria in this study formed biofilms as single species and as part of multispecies biofilms under static and dynamic conditions on polystyrene and on stainless steel. C. tyrobutyricum MK183 and R. kristinae M57 were weak biofilm producers, while B. licheniformis B149, P. aeruginosa QF2-56, S. thermophilus M388, and E. faecalis M458 were generally strong biofilm producers. In multispecies biofilm, the weak producers were barely detectable, perhaps overwhelmed by the strong producers. It was also found that P. aeruginosa QF2-56 inhibited the growth of B. licheniformis B149 and E. faecalis M458 on agar.

A model of multispecies biofilm production by bacteria isolated from dairies provides a new and valuable tool for understanding biofilms as a bacterial community. This should allow improvement of biofilm prevention measures in the dairy industry and in the food industry in general by developing new control strategies based on knowledge of bacterial interactions within multispecies biofilms.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.00713-23.

Fig. S1 and S2. aem.00713-23-s0001.docx.

S1 Antagonistic reactions and S2 Absence of antagonistic reactions.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.