Source of Enterococci in a Farmhouse Raw-Milk Cheese

Abstract

Enterococci are widely distributed in raw-milk cheeses and are generally thought to positively affect flavor development. Their natural habitats are the human and animal intestinal tracts, but they are also found in soil, on plants, and in the intestines of insects and birds. The source of enterococci in raw-milk cheese is unknown. In the present study, an epidemiological approach with pulsed-field gel electrophoresis (PFGE) was used to type 646 Enterococcus strains which were isolated from a Cheddar-type cheese, the milk it was made from, the feces of cows and humans associated with the cheese-making unit, and the environment, including the milking equipment, the water used on the farm, and the cows' teats. Nine different PFGE patterns, three of Enterococcus casseliflavus, five of Enterococcus faecalis, and one of Enterococcus durans, were found. The same three clones, one of E. faecalis and two of E. casseliflavus, dominated almost all of the milk, cheese, and human fecal samples. The two E. casseliflavus clones were also found in the bulk tank and the milking machine even after chlorination, suggesting that a niche where enterococci could grow was present and that contamination with enterococci begins with the milking equipment. It is likely but unproven that the enterococci present in the human feces are due to consumption of the cheese. Cow feces were not considered the source of enterococci in the cheese, as Enterococcus faecium and Streptococcus bovis, which largely dominated the cows' intestinal tracts, were not found in either the milk or the cheese.

Enterococci are found in a variety of artisanal cheeses made from raw or pasteurized milk from goats, sheep, water buffaloes, and cows (6, 14). Their role in flavor development in cheese is not clear; some workers have reported that they cause deterioration in the flavor (21, 38), while others believe they play a major role in improving flavor development and cheese quality (5, 7, 18, 29; H. E. Walter, A. M. Sadler, J. P. Malkames, and C. D. Mitchell, abstr., J. Dairy Sci. 39:917, 1956). The most common enterococci in cheese are Enterococcus faecalis and Enterococcus faecium (1, 4, 6, 8, 34, 36).

Enterococci have become important over the past decade because they are the most frequently encountered nosocomial pathogens and appear to have increasing antimicrobial resistance (30). An even greater threat is the transfer of vancomycin resistance from vancomycin-resistant enterococci to methicillin-resistant Staphylococcus aureus (25). Enterococci in food usually indicate poor bacteriological quality and poor hygiene during manufacture (21). Therefore, there is a need to determine their source in food.

The natural habitat of enterococci is the mammalian intestinal tract (14). E. faecalis is often the dominant species found in human feces (13, 14, 28); however, E. faecium, Enterococcus hirae, Enterococcus avium, and Enterococcus durans have also been found (14, 16, 28). Some enterococci are also found in the lower and upper urogenital tracts and in the oral cavity (27, 28). Although enterococci are prevalent in the human intestinal tract, they are less frequently isolated from the animal intestinal tract. E. faecium is the most frequently occurring species in dairy cows (12, 19, 22, 24, 26), but E. faecalis, E. hirae, and Enterococcus casseliflavus are also found (10, 12). In addition, enterococci are found in water, soil, plants, vegetables, birds, and insects (13, 14, 16, 22).

In a previous study (15), enterococci were isolated from a Cheddar-type cheese during manufacture and ripening, from the milk it was made from, from the feces of the personnel involved in cheese making, and from the feces of the dairy cows present on the farm. In addition, strains were isolated from the environment and the milking equipment, the tap water, the milking machine, and the cows' teats. All these isolates were identified to species level by random amplified polymorphic DNA (RAPD). The dominant species in milk, cheese, and human feces was E. casseliflavus, followed by smaller numbers of E. faecalis. The cows' feces contained mainly E. faecium. The enterococci isolated from the environmental sources were mainly E. casseliflavus and E. faecalis. As the RAPD technique was not sensitive enough to distinguish between strains, no conclusion could be drawn about the clonal relationships of the enterococci in the cheese and their putative source.

Pulsed-field gel electrophoresis (PFGE) has proven to be highly discriminatory and is generally accepted as the method of choice for typing strains of enterococci (17, 23, 35, 40). The aim of this study was to determine the sources of enterococci in the Cheddar cheese by comparing the strains isolated from the raw milk and the cheese samples with the isolates from the human fecal samples and the isolates from the milking equipment and the environment by PFGE. As a result of this study, valuable information has been obtained about the biodiversity of enterococci in a ripening cheese and the personnel associated with one cheese over a period of 15 months.

MATERIALS AND METHODS

Isolation of strains.

Six-hundred forty-six enterococcal strains isolated from the three trials described by Gelsomino et al. (15) were used. These included 120 E. casseliflavus and 20 E. faecalis isolated from the milk, 304 E. casseliflavus and 56 E. faecalis isolated from the curd and the cheese, 61 E. casseliflavus, 49 E. faecalis, and 9 E. durans isolated from the human fecal samples, 10 E. faecalis isolated from cows' feces, and 12 E. casseliflavus and 5 E. faecalis isolated from the milking equipment and the environment.

Kanamycin-esculin-azide agar (Merck, Darmstadt, Germany) (KAA) containing 6.5% (wt/vol) salt (KAAS) was tested as a selective medium for enterococci. Growth of type strains of enterococci (E. avium, Enterococcus cecorum, E. casseliflavus, Enterococcus columbae, Enterococcus dispar, E. durans, E. faecalis, E. faecium, Enterococcus flavescens, Enterococcus gallinarum, E. hirae, Enterococcus malodoratus, Enterococcus mundtii, Enterococcus pseudoavium, Enterococcus saccharolyticus, Enterococcus solitarius, and Enterococcus sulfurous) (BCCM/LMG) and Streptococcus bovis (R-16101 and R-16102; Research Collection of the Laboratory of Microbiology, University of Ghent, Ghent, Belgium) were tested on both KAA and KAAS.

Two additional experiments were undertaken during the present study. The first consisted in mixing the first few squirts of milk (first milk) from each cows' teat, which are normally disposed of, and plating them on KAA and KAAS.

The second additional experiment consisted of collecting fecal samples by manual rectal retrieval and plating them on KAAS as well as on KAA. Where possible, 10 putative enterococci were isolated per cow.

In both of these experiments, when all 27 cows in the herd were milked, samples were taken from the bulk milk and the cheese at day 1 after manufacture. Samples were emulsified and plated, and individual colonies were isolated as described previously (15). In this way, 131 isolates were obtained from the milk (n = 30), cheese (n = 30), and cows' feces (n = 71) from both KAA and KAAS plates.

RAPD typing.

DNA was prepared by the rapid procedure described by Pitcher et al. (31). Random primer D11344 was used to identify the new isolates from the milk, cheese, cows' feces, and samples of first milk to the species level in the two additional experiments as described previously (15). DNA amplification was performed by the procedure described by Descheemaeker et al. (9).

PFGE.

The strains were grown overnight at 37°C on BM (2% Tryptose [Oxoid, Basingstoke, England], 0.5% NaCl [Merck, Darmstadt, Germany], 0.5% yeast extract [Merck, Darmstadt, Germany], 0.5% glucose [Merck]; pH 6.85) agar. All reagents were obtained from Sigma (St. Louis, Mo.) unless otherwise stated. One loopful of cells from an overnight culture was washed three times in 1 ml of EET buffer (100 mM EDTA, 10 mM EGTA, 10 mM Tris-HCl [pH 8.0]). After centrifugation, the pellet was resuspended in EC buffer (6 mM Tris-HCl [pH 7.6], 1 M NaCl [Merck], 100 mM EDTA [pH 8.0], 0.5% polyoxyethylene 20 cetylether [Brij 58], 0.2% deoxycholate, 0.5% N-laurylsarcosyl) and mixed with an equal volume of 1.6% (wt/vol) low-melting-point agarose (Bio-Rad, Richmond, Calif.) in EC buffer and pipetted into plug molds. The solidified plugs were incubated overnight at 37°C in 1 ml of EC buffer-lysozyme solution (2.88 mg of lysozyme per ml of EC buffer). The lysis buffer was replaced with 1 ml of protein digestion solution (3.3 mg of pronase E in 1 ml of EET buffer containing 1.6% [wt/vol] sodium dodecyl sulfate [SDS]), and the plugs were incubated again overnight at 37°C.

The agarose plugs were washed three times for 1 h in EET buffer, twice for 1 h in Milli-Q water, and once for 1 h in the appropriate restriction buffer (buffer Y⁺/Tango; MBI Fermentas, St. Leon-Rot, Germany) at room temperature. The restriction was carried out overnight at 27.5°C in 300 μl of restriction buffer containing 30 U of SmaI (MBI Fermentas). The digestion was stopped by adding 0.5 ml of 0.5 M EDTA (pH 8.0), and the plugs were stored at 4°C. The restriction fragments were separated by PFGE in a contour-clamped homogeneous electric field MAPPER system (Bio-Rad) by loading pieces of the plugs in a 1% (wt/vol) pulsed-field-certified agarose (Bio-Rad) gel prepared with 0.5× TBE buffer (45 mM Tris-HCl, 45 mM boric acid, 1 mM EDTA). Electrophoresis of E. casseliflavus was performed in 2 liters of 0.5× TBE at 14°C for 22 h at 6 V/cm and an angle of 120°, with pulse times ramping linearly from 0.41 to 15.11 s. For all the other Enterococcus strains, pulse times ramping linearly from 5 to 30 s were chosen.

A Staphylococcus aureus strain (R-6314; Department for Medical Microbiology, University of Antwerp) was used as a molecular size marker. The genome was prepared as described above except that 500 U of mutanolysin was added to the lysozyme solution. The gels were stained with ethidium bromide. Conversion, normalization, and analysis of the band patterns were performed with GelCompar software (Applied Maths, Ghent, Belgium). Correlation coefficients and levels of similarity were calculated by using the Dice coefficient and cluster analysis with the unweighted pair group method.

Growth of S. bovis in milk.

The two strains of S. bovis were grown overnight in BM broth and in BM broth containing lactose instead of glucose, and 1 ml of each culture was centrifuged and resuspended in 1 ml of buffered peptone water (Oxoid). After washing, the cells were diluted in order to obtain an initial number of cells of between 10³ and 10⁴ per ml of milk. Fifty milliliters of fresh raw milk and 50 ml of heat-treated milk (20 min at 95°C) were inoculated and incubated on a stirring plate at 4, 15, and 37°C. Samples taken periodically during incubation for 3 days were plated on KAA and incubated for 24 h at 37°C. Control samples were also plated on plate count agar (Oxoid), which was incubated at 30°C for 3 days.

RESULTS

Selectivity of the isolation medium.

The selectivity of KAA for the isolation of enterococci from cows' feces is very low, as the vast majority of isolates from the medium were found to be S. bovis (15). The ability of 2 S. bovis and 17 BCCM/LMG enterococcus type strains to grow on KAA and KAAS was determined. E. avium, E. cecorum, E. columbae, E. pseudoavium, and E. malodoratus did not grow or grew very poorly on KAAS (data not shown). All the other type strains grew as well on KAAS as on KAA, while the two S. bovis strains did not grow at all on KAAS.

Seventy-one enterococci were isolated from the feces of 12 of the 27 cows present on the farm on KAAS at dilutions of 10⁻¹ to 10⁻³. No enterococci were found in the feces of the other 15 cows. In contrast, all samples plated on KAA showed growth up to a dilution of 10⁻⁴, indicating the presence of S. bovis.

No enterococci were found in the first samples of milk from all the teats of the cows on either KAA or KAAS. This corroborates the results of the previous study (15).

An average of 7.3 × 10³ enterococci/ml was found in the bulk milk ∼3 h after the cows' teats and feces were sampled on both KAA and KAAS, while an average of 2 × 10⁵ enterococci/ml was found in the cheese on both KAA and KAAS.

Species identification.

In the two additional experiments, the isolates on both KAA and KAAS from the milk (n = 30) and the cheese (n = 30), which was made shortly after the cows were milked, were shown to be E. casseliflavus (50%) and E. faecalis (50%) by RAPD. All isolates from the cows' feces (n = 71) were shown to be E. faecium by RAPD and were not categorized any further.

Strain typing.

PFGE was performed on the 646 isolates of E. casseliflavus, E. faecalis, and E. durans isolated previously (15). All 646 PFGE band patterns were clustered and compared visually and with the GelCompar software. Only nine different clones were found among the 646 isolates tested. An example of each of these is shown in Fig. 1. In this figure, E. casseliflavus patterns are marked with a C, E. faecalis patterns are marked with an F, and E. durans patterns are marked with a D. The patterns marked F1a, F1b, and F1c and the patterns marked C1a and C1b differed by only one band and were, in accordance with the definition of Tenover et al. (37), considered to represent the same clone.

FIG. 1.

Agarose gel showing the PFGE band patterns of the nine clones found and digested with SmaI. A Staphylococcus aureus strain was used for molecular size markers. Sc and Sf are S. aureus strains used as a molecular size marker run with the PFGE parameters for E. casseliflavus and E. faecalis, respectively. C, F, and D signify E. casseliflavus, E. faecalis, and E. durans, respectively, while C1a and C1b and F1a, F1b, and F1c are considered the same clones.

Table 1 shows the distribution of all the clones found in the milk, cheese, feces, and environment, and Fig. 2 shows the percentages of the most common clones in each of the three trials. Three of the nine clones, C1, C2, and F1, largely dominated in the milk, cheese, and human feces. These three clones, two E. casseliflavus and one E. faecalis, were found throughout the three trials, and two of them, E. casseliflavus C1 and C2, were also present in the environmental samples, i.e., the bulk tank and the milking machine rinses. E. casseliflavus C3 appeared less abundantly in the cheeses and in adult 1 but was not evenly distributed over the three trials.

TABLE 1.

Strain distribution of isolates found in milk, cheese, feces, and the environment

Sample	No. of isolates
	E. casseliflavus			E. faecalis					E. durans (D)	Total
	C1	C2	C3	F1	F2	F3	F4	F6
Milk
Bulk tank	46	23		7	4					80
Cheese vat	32	19		9						60
Cheese
Before molding	29	22		9						60
At 1 day	16	29		15						60
At 11 days	35	14	2	8		1				60
At 25 days	19	23	3	15						60
At 39 days	35	20	2	3						60
At 53 days	22	30	3	5						60
Feces
Adult 1	8	7	4	10	1					30
Adult 2	15	2		3	4				6	30
Teenager 1	5	5		20						30
Teenager 2	6	9		10	1				3	29
Cow 27, trial 2					10					10
Environment
Water							2			2
Rinse from empty bulk tank	2	2			1					5
Rinse from milking machine before chlorination		5								5
Rinse from milking machine after chlorination		3						2		5
Total										646

FIG. 2.

Distribution of clones found in each of the three trials (July 1998, September 1998, and October 1999). Numbers in brackets give the total number of isolates in that particular trial. C1, C2, and C3 are different clones of E. casseliflavus, and F1 is a clone of E. faecalis.

Other clones were occasionally found. Four E. faecalis isolates of clone F2 were found in the milk from the bulk tank, and one, four, and one isolate of this clone were found in the feces of adults 1 and 2 and teenager 2, respectively, in trial 1. Ten isolates in the cows' feces during trial 2 and one isolate from the rinse of the bulk tank were found to be clone F2. Only one isolate of clone F3 (E. faecalis) was found in the cheese of trial 1, ripened for 11 days. In addition, six and three isolates of E. durans were found in the feces of adult 2 and teenager 2, respectively, and they all showed the same band pattern (D). The two F4 clones were found in the tap water, while the two F6 clones were found in the milking machine (Fig. 2 and Table 1).

A representative isolate of each clone has been deposited in the BCCM/LMG Bacteria Collection with the following numbers: LMG 19846 (F1), LMG 20228 (F2), LMG 20229 (F3), LMG 20230 (F4), LMG 20234 (F6), LMG 20231 (D), LMG 19844 (C1), LMG 19845 (C2), and LMG 20235 (C3).

Growth of S. bovis in milk.

Large numbers of S. bovis were found in cows' feces when plated on KAA, but no S. bovis isolates were found in any milk sample (15). This raised the question of whether S. bovis grows in milk.

Both glucose-grown and lactose-grown cells of the two strains of S. bovis survived for at least 3 days at 4°C and grew at 15 and 37°C from 1.5 × 10³ CFU/ml to 9.6 × 10⁷ CFU/ml in 2 days in the raw and heat-treated milk samples, indicating that the natural inhibitors present in the raw milk did not prevent the growth of S. bovis. In 3 days, the total bacterial counts on plate count agar increased from 4.6 × 10⁴ to 7.0 × 10⁶ CFU/ml in the raw milk at 4°C and from 3.2 × 10⁴ to 3.5 × 10⁸ CFU/ml in the raw milk at 37°C (data not shown); S. bovis was able to compete with these bacteria.

No colonies were detected on KAA or plate count agar in the uninoculated heat-treated milk, indicating that the heat treatment was sufficient to inactivate any contaminating bacteria present.

DISCUSSION

In this study, an epidemiological approach was used to identify the sources of enterococci in cheese by using PFGE. Nine different PFGE band patterns, which were considered to be distinct clones, were found among 646 isolates from the milk, cheese, human feces, equipment, and environment (Fig. 1). Of these, five were produced by E. faecalis (F1 to F4 and F6), one was produced by E. durans (D), and three were produced by E. casseliflavus (C1 to C3).

The dominant species found in the cheese and milk was E. casseliflavus (15), although E. faecalis and E. faecium have been reported to be the most common species of Enterococcus in cheese (1, 6, 11, 36, 39), whereas E. casseliflavus is more often associated with plants (12, 20). Two E. casseliflavus clones, C1 and C2, accounted for 47 and 36% of the 500 isolates from the milk and cheese, respectively. E. faecalis clone F1 was present in lower numbers (14%). Except for the milk and cheese in trial 2, where clone F1 was not present (Fig. 2), these three clones were found in all samples of the milk and cheese throughout the three trials (Table 1) and constituted 97% of all isolates found in the milk and cheese. E. casseliflavus clone C3 was found quite often in the ripening cheese but not in the milk. Of the other E. faecalis isolates, only four isolates of clone F2 were found in the bulk tank milk, while clone F3 was found only once, in the cheese ripened for 11 days.

The dominant species in the human feces was E. casseliflavus, followed by smaller numbers of E. faecalis (15). This result disagrees with previous studies which showed that E. faecalis is the most common species in the human intestine (13, 20, 32). Strain identification with PFGE showed that the same clones occurred in the milk, cheese, and human feces. Within the human feces, E. faecalis clone F1 was found in high numbers (36% of isolates), followed by E. casseliflavus clones C1 (28% of isolates) and C2 (19% of isolates). Clone C2 was found only in trial 1, while clones F1 and C1 were found in all trials (Fig. 2). E. casseliflavus clone C3 was found in adult 1 in trial 3, while E. faecalis clone F2 was distributed among three of the four members of the family. The nine isolates of E. durans gave identical patterns and were found in adult 2 and teenager 2 in trial 3 (D).

As both the milk and cheese samples and the human fecal samples contained the same Enterococcus species and, more importantly, the same three dominant clones, human contamination could be regarded as the possible source of enterococci in cheese. Another possibility is that the enterococci in the human feces are due to ingestion of the milk and cheese. The latter is more likely, as the family eat their own cheese.

In the previous study (15), only a few enterococci were found in the cows' feces. All of them were E. faecium except for 10 isolates from cow 27 in trial 2. A more selective medium (KAAS) was used in the present study to isolate putative enterococci from the bovine feces, and E. faecium was still the only species found, again in very low numbers; some samples of cows' feces showed no colonies even at a dilution of 10⁻¹. In contrast, growth on KAA without NaCl occurred up to a dilution of 10⁻⁴, indicating the presence of S. bovis (2, 15).

As indicated, E. faecalis was isolated only from cow 27 in trial 2 (this cow was present in the herd of trials 1 and 2 but was culled in trial 3), and all isolates belonged to clone F2. The same clone was found in the milk in trial 3 and also in human feces in trial 1. We have no adequate explanation for this result, but we believe that cows' feces could not have contaminated the milk, as F2 was already present in the human feces in trial 1. Enterococci were not found on the teats (15) or in the udder (this study). In addition, no clones of E. casseliflavus or E. faecalis found in the milk or cheese could be traced back to cows except for the one already discussed. Furthermore, the only Enterococcus species found in cows' feces, except for the one cow mentioned above, was E. faecium, which was also not found in the milk or cheese. Therefore, we conclude that the cows' feces were not the source of enterococcal contamination of the cheese and milk. This is corroborated by the findings that S. bovis, which is the most dominant species in cows' feces and grows quite well on KAA, can grow in milk but was not found in the milk or cheese.

Both the milking machine and bulk tank are normally sterilized with Hydrosan before chlorination. Both were rinsed with sterile water, and the rinses were examined for enterococci. Ten isolates of E. casseliflavus clone C2 were found in the rinses of the milking machine and bulk tank, and two isolates of clone C1 were found in the rinse of the bulk tank. Clone F2 was found in the rinse of the bulk tank and in the milk of the bulk tank. E. faecalis clones F4 and F6 were also found in the tap water and in the milking machine. The fact that the dominant clones in the milk, cheese, and human feces are also found in the milking equipment strongly suggests that enterococcal contamination starts there. Although the milking machine had been rinsed and sterilized with a 0.024% sodium hypochlorite solution (15) in order to maintain good hygiene, disinfection might not be fully effective, so that milk residues and bacteria are not completely eliminated from the equipment. Crevices, joints, dead ends, and fittings are danger points where bacteria tend to accumulate and grow, infecting the milk directly (3, 33). The possibility that a biofilm is also present (41) cannot be ruled out. This might explain why enterococci were found even after chlorination. It is also possible that E. casseliflavus survives better in the milking equipment than E. faecalis.

In conclusion, data from the present study demonstrate that enterococci survive and grow in the hidden corners of the milking machine and the bulk tank, thus infecting the milk directly. From the milk, the enterococci are transferred into the cheese and from there probably into the human intestinal tract by ingestion. Whether these strains were brought into the equipment via the air or by poor hygienic conditions is not obvious. The original source of the enterococci in the milking machine is not clear.