|Home | About | Journals | Submit | Contact Us | Français|
The microbial composition of smear-ripened cheeses is not very clear. A total of 194 bacterial isolates and 187 yeast isolates from the surfaces of four Irish farmhouse smear-ripened cheeses were identified at the midpoint of ripening using pulsed-field gel electrophoresis (PFGE), repetitive sequence-based PCR, and 16S rRNA gene sequencing for identifying and typing the bacteria and Fourier transform infrared spectroscopy and mitochondrial DNA restriction fragment length polymorphism (mtDNA RFLP) analysis for identifying and typing the yeast. The yeast microflora was very uniform, and Debaryomyces hansenii was the dominant species in the four cheeses. Yarrowia lipolytica was also isolated in low numbers from one cheese. The bacteria were highly diverse, and 14 different species, Corynebacterium casei, Corynebacterium variabile, Arthrobacter arilaitensis, Arthrobacter sp., Microbacterium gubbeenense, Agrococcus sp. nov., Brevibacterium linens, Staphylococcus epidermidis, Staphylococcus equorum, Staphylococcus saprophyticus, Micrococcus luteus, Halomonas venusta, Vibrio sp., and Bacillus sp., were identified on the four cheeses. Each cheese had a more or less unique microflora with four to nine species on its surface. However, two bacteria, C. casei and A. arilaitensis, were found on each cheese. Diversity at the strain level was also observed, based on the different PFGE patterns and mtDNA RFLP profiles of the dominant bacterial and yeast species. None of the ripening cultures deliberately inoculated onto the surface were reisolated from the cheeses. This study confirms the importance of the adventitious, resident microflora in the ripening of smear cheeses.
Surface-ripened cheeses can be divided into mold-ripened cheeses, such as Camembert and Brie, and bacterium-ripened cheeses, such as Reblochon, Tilsit, and brick. The latter cheeses are also called smear or red smear cheeses because of the development of viscous, red-orange smears on their surfaces during ripening. The smear is a microbial mat composed of bacteria and yeast, and these microorganisms are mainly responsible for the development of the flavor characteristics of the cheeses (5, 27). The ripening process starts with the development of yeast cells, which metabolize lactate to CO2 and H2O and form alkaline metabolites, such as ammonia (5, 30), that lead to deacidification of the cheese surface, enabling the growth of salt-tolerant but less acid-tolerant gram-positive catalase-positive bacteria, such as Micrococcaceae and coryneform bacteria.
The microbiology of these cheeses is poorly understood. In the past, Brevibacterium linens was considered to be the major organism found on the cheese surface. However, more recent investigations show that other bacteria are also important. Valdés-Stauber et al. (27) found that Arthrobacter nicotianae, B. linens, Corynebacterium ammoniagenes, Corynebacterium variabile, and Rhodococcus fascians were the dominant organisms in 21 brick cheeses from six German dairies, while Eliskases-Lechner and Ginzinger (7) found that although B. linens accounted for a large share of the flora (~30%) in Tilsit cheese deliberately inoculated with B. linens, Arthrobacter globiformis and Brevibacterium ammoniagenes were also dominant. In both of these studies the bacteria were identified by phenotypic analysis, including the assimilation of numerous sugars and amino and fatty acids. Bacteria in the genera cited above have high G+C contents and are difficult to identify categorically unless the type of peptidoglycan (diamino acid present and the presence or absence of N-glycolyl residues), the major fatty acids, and the major menaquinones present in their cell walls are determined. Thus, the bacteria may have been misidentified. Recently, Brennan et al. (2), using a polyphasic approach combining phenotypic, chemotaxonomic, and genotypic analyses, showed that single clones of novel strains of Corynebacterium casei, Corynebacterium mooreparkense, and Microbacterium gubbeenense dominated the surface of Gubbeen cheese during the ripening period, although the cheese had been deliberately inoculated with Brevibacterium aurantiacum BL2. No microbial succession of bacterial species was observed during ripening. C. mooreparkense has since been shown to be a junior subjective synonym for C. variabile (14). More recently, Feurer et al. (11) confirmed that C. casei was the dominant bacterium in a French farmhouse cheese and that B. linens and M. gubbeenense were also important. Other novel species, including Staphylococcus equorum subsp. linens, Staphylococcus succinus subsp. casei (24, 25), Arthrobacter arilaitensis, and Arthrobacter bergerei (11, 18), have been identified in smear-ripened cheeses, while B. linens ATCC 9175 has been shown to be B. aurantiacum (13). The commercial ripening culture B. linens BL2 (Chr. Hansen) has recently been shown to be B. aurantiacum (Vancanneyt et al., unpublished).
There appears to be much less diversity in the dominant yeasts found on the surface of smear-ripened cheeses. Debaryomyces hansenii and Galactomyces geotrichum have been reported to be present in rennet cheeses, and Kluyveromyces marxianus and Pichia membranifaciens have been reported to be present in acid-curd cheeses (8, 23, 27).
In some countries, particularly Germany, freshly made cheeses are inoculated with a complex, undefined flora taken from the already ripened cheese surface. Despite the fact that this “old-young smearing” technique ensures the transfer of desirable microorganisms for the ripening process, it can also lead to the transfer of spoilage and pathogenic microorganisms, such as Listeria monocytogenes or Staphylococcus aureus, to the young cheese (27). Surface-ripening cultures, mainly comprising B. linens, D. hansenii, and Geotrichum candidum, are often used to inoculate the young cheeses (1, 3).
The aim of this study was to identify and compare the cheese surface microfloras of four Irish smear-ripened cheeses, three of which were deliberately smeared with surface ripening cultures and one of which was not, using a polyphasic approach.
Gubbeen, Durrus, Ardrahan, and Milleens cheeses were studied. These are all semisoft, surface-ripened cheeses, similar to Reblochon, Limburger, and Tilsit, and are made on farms in southwest Ireland. All are made with mesophilic lactic starters and, except for Durrus, are made from pasteurized milk. After brining, Gubbeen cheese is smeared with a commercial culture, OFR9 (Visby), Ardrahan cheese is smeared with B. aurantiacum BL2 (Chr. Hansen), and Durrus cheese is smeared with OFR9 and BL2. No commercial ripening cultures are used for Milleens cheese. The surface microflora was examined in the middle of ripening, after approximately 2 weeks.
The surface microflora of each cheese was enumerated as described previously by Brennan et al. (2) on milk plate count agar (Merck) containing 5% salt for the bacteria and yeast glucose chloramphenicol agar (Merck) for the yeast. Fifty bacterial colonies and 44 to 49 yeasts were selected from countable plates from each cheese and were purified by restreaking twice on the appropriate media. The organisms were stored at −80°C in a 1:1 mixture of Trypticase soy broth (Difco) and glycerol until characterization.
Bacterial cultures were examined by phase-contrast microscopy, Gram stained, and tested for the presence of catalase. The cell morphology of the isolates was determined on mineral base E, yeast extract, glucose agar after incubation at 30°C for 8 h and 1, 3, and 7 days, as described by Cure and Keddie (6).
Gram-positive, catalase-positive, irregularly rod-shaped isolates or isolates that underwent a rod-coccus transformation were considered to be coryneforms and were subjected to pulsed-field gel electrophoresis (PFGE) analysis as described below.
Gram-positive, catalase-positive cocci were considered members of the Micrococcaceae family and were tested for (i) the ability to grow anaerobically and/or aerobically in a glucose-containing medium, (ii) glycerol acidification, and (iii) susceptibility to lysostaphin (200 μg/ml), lysozyme (400 μg/ml), and furazolidone (100-μg diffusion disk) (9, 12, 19). The organisms that were facultative anaerobes, able to use glycerol aerobically in the presence of erythromycin, and susceptible to lysozyme and furazolidone were considered to be Staphylococcus sp. and were subjected to PFGE analysis as described below.
PFGE was performed using a modification of the method described previously by Brennan et al. (2). Lysis of all isolates except coccal isolates was performed by one treatment with 10 mg/ml lysozyme. Staphylococcus sp. isolates were lysed first in 10 mg/ml lysozyme and 50 μg/ml lysostaphin overnight at 37°C and then by overnight incubation at 37°C in the presence of 10 mg/ml lysozyme. The restriction enzymes used were SmaI for the Staphylococcus sp., SfiI for the spore formers, AscI for Halomonas sp., and SpeI for the coryneform isolates. Coryneform strains that produced PFGE patterns consisting of mainly low-molecular-weight fragments were also digested with AscI. Gels were stained using ethidium bromide (0.5 μg/ml) for 2 h, destained in water, and photographed using a digital camera. The digitized patterns were normalized and analyzed numerically using Bionumerics software, version 2.0 (Applied Maths). Similarities among band patterns were calculated based on Pearson's similarity coefficient, and dendrograms were constructed by the unweighted-pair group method using average linkages (UPGMA).
The reference strains used for repetitive sequence-based PCR (rep-PCR) genomic fingerprinting are listed in Tables Tables11 and and2.2. One to three representatives of the strains of each PFGE cluster were also subjected to this analysis. Biomass, scraped from Trypticase soy agar cultures which had been incubated for 1 day, was suspended in 1 ml of TE 10/1 (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) and centrifuged for 10 min at 2,700 × g at 4°C. Cells were then kept for at least 1 h at −20°C. Total DNA was extracted using a modified method described by Gevers et al. (15). The lysis buffer contained 1,330 U/ml mutanolysin and 40 mg/ml lysozyme for rod-shaped isolates and 1,330 U/ml lysostaphin and 40 mg/ml lysozyme for coccal isolates. The oligonucleotide primers used were BOXA1R (5′-TACGGCAAGGCGACGCTACG-3′) for the gram-positive, catalase-positive rods and (GTG)5 (5′-GTGGTGGTGGTGGTG-3′) for the gram-positive, catalase-positive cocci, and each primer was used with the appropriate PCR program (29). PCR amplifications were performed with a DNA thermal cycler (model 9600; Perkin-Elmer) as described previously (29), using Goldstar DNA polymerase (Eurogentec, Belgium). The PCR products were run in a 1.5% (wt/vol) agarose gel for 16 h at a constant voltage of 55 V in 1× TAE (40 mM Tris-acetate, 1 mM EDTA, pH 8.0) at 4°C. The rep-PCR patterns were visualized after staining with an ethidium bromide solution (1 μg/ml) under UV light and were digitized using a digital camera. The resulting fingerprints were analyzed using the Bionumerics software, version 2.0. Similarities among band patterns were calculated based on Pearson's similarity coefficient, and dendrograms were constructed using UPGMA.
Partial sequencing (~700 bp) of the 16S rRNA gene as described by Coeyne et al. (4) was performed for the strains that remained unidentified after rep-PCR fingerprinting. Sample preparation was assisted using a Tecan Genesis 200 workstation (Tecan, Switzerland). Sequence assembly was performed using AutoAssembler (Perkin-Elmer). The most closely related sequences were found using the FASTA program.
Fourier transform infrared (FTIR) spectroscopy was performed to identify yeast isolates as described previously by Kümmerle et al. (20). To record and evaluate the spectra, an IFS-28B FTIR spectrometer (Bruker, Ettlingen, Germany) and the software OPUS for Windows, version 3.17 (Bruker, Ettlingen, Germany), were used. The yeast database comprised around 2,500 reference spectra.
Total DNA of the yeasts was extracted using a modification of the methods described previously by Romano et al. (26) and Petersen et al. (23). Yeast cells that had been grown in 5 ml of YEPD (1% yeast extract, 2% peptone, 2% glucose, pH 6.0) at 25°C overnight were harvested, washed with 1 ml of distilled water, harvested again, and resuspended in 0.5 ml of solution A (0.9 M sorbitol, 0.1 M EDTA [pH 7.5], 200 μg/ml lyticase). The mixture was incubated for 60 min at 37°C to produce spheroplasts. The spheroplasts were harvested by centrifugation at 5,200 × g for 5 min, resuspended in 0.5 ml of solution B (50 mM Tris-HCl, 20 mM EDTA, 1% sodium dodecyl sulfate [pH 7.5], 100 μg/ml proteinase K), and incubated for 30 min at 65°C. Then 200 μl of 5 M sodium acetate was added, and the mixture was placed on ice for 30 min. After centrifugation at 10,600 × g for 15 min, the supernatant was precipitated with 1 volume of isopropanol for 15 min at room temperature and centrifuged for 10 min at 20,800 × g. The supernatant was decanted, and the DNA was washed with 70% ethanol and centrifuged for 5 min at 20,800 × g. The residual ethanol was aspirated with a pipette, and the pellets were dried on a heating block at 37°C for 45 min. The dried DNA was dissolved in 25 μl TE 10/1. Ten to fifteen microliters of the purified DNA was digested with 5 U of HaeIII overnight at 37°C. The restriction fragments were analyzed by electrophoresis on a 1% (wt/vol) agarose gel in 1× TAE buffer at 100 V for 3 h. λ DNA cut with HindIII was used as a marker. The restriction fragments were visualized by ethidium bromide staining and UV transillumination. After photography using a digital camera, the resulting fingerprints were analyzed using the Bionumerics software, version 2.0. Similarities among band patterns were calculated based on Dice's similarity coefficient, and a dendrogram was constructed using UPGMA.
Moisture and salt contents were determined by standard methods (21). pH was determined by placing the electrodes directly into grated cheese.
The following representatives of some bacterial species isolated in this study have been deposited in the BCCM/TM/LMG Bacteria Collection (Laboratory of Microbiology, Ghent University, Belgium): Agrococcus sp. strain LMG 22965 (Gubbeen cheese), A. arilaitensis strain LMG 22967 (Gubbeen cheese), C. casei strain LMG 22963 (Milleens cheese), C. variabile strain LMG 22966 (Milleens cheese), M. gubbeenense strain LMG 22964 (Gubbeen cheese), Staphylococcus saprophyticus strain LMG 22969 (Gubbeen cheese), and S. equorum strain LMG 22968 (Milleens cheese).
The moisture contents, salt contents, salt-in-moisture contents, and pHs of the surfaces of the four cheeses are shown in Fig. 1A and B. The moisture content varied from 43.4 to 48.7 g/100 g, and the salt content ranged from 0.78 to 1.97 g/100 g. A difference in the salt-in-moisture contents was found; Ardrahan and Gubbeen cheeses showed greater salt-in-moisture contents (4.1%) than Milleens and Durrus cheeses, which had salt-in-moisture contents of 1.6% and 3%, respectively. The surface pHs ranged from 5.9 for Gubbeen cheese to 6.8 for Ardrahan cheese.
The counts of yeasts and salt-tolerant bacteria on the cheese surfaces are shown in Fig. Fig.1C.1C. There was little difference between cheeses in terms of yeast or bacterial counts, but the yeast counts were consistently lower than the bacterial counts.
The salt-tolerant microflora was composed mainly of gram-positive, irregularly shaped rods (122 of 200 isolates) that were considered to be coryneform bacteria. The next most important group was the gram-negative, catalase-positive rods (31 isolates), followed by Bacillus sp. (25 isolates), gram-positive, catalase-positive cocci (14 isolates), and lactic acid bacteria (two isolates). Thirteen of the 14 gram-positive, catalase-positive cocci were found to be Staphylococcus sp., and the other isolate was a Micrococcus sp.
The commercial culture OFR9 contained coryneform bacteria that produced two different PFGE patterns, designated OFR9.1 and OFR9.2, while the commercial culture B. aurantiacum BL2 contained strains that produced only one PFGE pattern. A dendrogram of the different SpeI restriction patterns of chromosomal DNA of the coryneform isolates and the commercial isolates is shown in Fig. Fig.2.2. Twenty-two of the 122 coryneform strains from the cheese surface produced good profiles and were differentiated into six clusters; all except one of these strains were from Gubbeen cheese. The remaining coryneform isolates produced mainly low-molecular-weight fragments when SpeI was used and were therefore difficult to distinguish from each other (data not shown). The chromosomal DNA of these isolates was subsequently digested with AscI, which yielded much better differentiation into 27 clusters (Fig. (Fig.33).
One to three representatives of each cluster were subjected to rep-PCR analysis using the BOX primer, and within each cluster, all analyses resulted in the same identification (Fig. (Fig.4).4). Six different species, Agrococcus sp. nov., M. gubbeenense, B. linens, C. variabile, C. casei, and A. arilaitensis, were identified using rep-PCR, and one member of the genus Arthrobacter was also found (Fig. (Fig.3).3). This organism was identified as Arthrobacter mysorens by partial 16S rRNA gene sequencing (Table (Table3).3). Most PFGE clusters contained strains from a specific cheese. The exceptions were cluster 8 (Fig. (Fig.2)2) (C. variabile) and cluster 18 (Fig. (Fig.3)3) (C. casei); for these clusters the same clone was found in both Gubbeen and Milleens cheeses and in both Durrus and Ardrahan cheeses, respectively.
Different strains of the same species were isolated, as demonstrated by the PFGE patterns (Fig. (Fig.22 and and3).3). Three different clones of C. variabile (clusters 7, 8, and 9) were found (Fig. (Fig.2),2), and 17 different clones of C. casei (clusters 10 to 26), and 9 different clones of A. arilaitensis (clusters 1 to 5, 7 to 9, and 27) were found (Fig. (Fig.33).
The dendrogram generated from the SmaI restriction patterns of the Staphylococcus isolates and the results of rep-PCR identification using the (GTG)5 primer are shown in Fig. Fig.5.5. Among the 13 staphylococcal isolates, three different species were found; S. equorum and S. saprophyticus were identified using rep-PCR (Fig. (Fig.6),6), and members of the genus Staphylococcus were also found. Partial 16S rRNA gene sequencing of these three organisms revealed 100% similarity with Staphylococcus epidermidis (Table (Table3).3). Each cheese contained only one staphylococcal species. The other coccal isolate, which was isolated from Ardrahan cheese, was identified as Micrococcus luteus on the basis of partial 16S rRNA gene sequencing (Table (Table3).3). Three different clones of S. epidermidis and S. equorum were found, while a single clone of S. saprophyticus was isolated.
The dendrogram generated by using the AscI restriction patterns of the gram-negative, catalase-positive rods and their identities are shown in Fig. Fig.7.7. These organisms were identified as Halomonas venusta (30 strains) or Vibrio sp. (five strains) by partial sequencing of the 16S rRNA gene (Table (Table3).3). Single clones of H. venusta were found in Milleens, Gubbeen, and Ardrahan cheeses. A single clone of Vibrio sp., which exhibited less than 97% similarity with other Vibrio sp. on the basis of 16S rRNA gene partial sequencing, was found in Milleens cheese (data not shown). This strain did not show any hemolytic activity on blood agar (data not shown).
Twenty-five isolates were spore formers. Two different clones of Bacillus sp. were found on Ardrahan and Gubbeen cheeses, and one clone was found on Milleens cheese (data not shown). These isolates were not identified further.
Only two species of yeasts were identified using FTIR spectroscopy; these species were D. hansenii (180 isolates) and Yarrowia lipolytica (7 isolates). One isolate from Milleens cheese was not identified. The RFLP profiles of the mtDNA of the D. hansenii isolates are shown in Fig. Fig.8.8. Most of the isolates from the different cheeses had specific RFLP profiles; the exceptions were the dominant strains from Durrus and Ardrahan cheeses, which had similar profiles. One to three different D. hansenii strains were found on each cheese, and there was one highly dominant RFLP profile (≥84% of the D. hansenii isolates). A single RFLP profile was found for the Y. lipolytica strains (data not shown). The yeast species from the OFR9 commercial culture, which was used to smear Gubbeen and Durrus cheeses, had an RFLP pattern different from that of the D. hansenii isolates and could not be identified using the current version of the FTIR database of >2,500 strains. In a similarity analysis of the FTIR spectra, the 20 isolates of the OFR9 culture which were analyzed clustered together but did not cluster with any other yeast spectra recorded in this study (data not shown). This result was confirmed using mtDNA RFLP (Fig. (Fig.8),8), which showed that none of the different RFLP profiles of the yeasts isolated from Gubbeen and Durrus cheeses were similar to the OFR9 yeast RFLP profile.
An overall view of the surface microflora in terms of yeast, bacterial, and strain diversity is shown in Table Table4.4. Two to six species of coryneform bacteria were isolated from each cheese. A. arilaitensis and C. casei dominated Ardrahan, Durrus, and Milleens cheeses, while C. casei and C. variabile dominated Gubbeen cheese. M. gubbeenense, Agrococcus sp. nov., and B. linens were isolated only from Gubbeen cheese, and A. mysorens was isolated only from Ardrahan cheese. One isolate of C. variabile was identified on the surface of Milleens cheese. One species of Staphylococcus was isolated from each cheese. S. epidermidis was found only in Durrus cheese, and S. saprophyticus was found only in Gubbeen cheese, while S. equorum was found in Ardrahan and Milleens cheeses. H. venusta was isolated from Milleens, Durrus, and Ardrahan cheeses, and each isolate had a different PFGE pattern. Two clones of Bacillus sp. were isolated from Gubbeen and Ardrahan cheeses, and one clone of Bacillus sp. was isolated from Milleens cheese. The other bacteria were isolated sporadically. D. hansenii and Y. lipolytica dominated Milleens cheese, while D. hansenii dominated Gubbeen, Durrus, and Ardrahan cheeses.
In this study, a polyphasic approach combining phenotypic and genotypic techniques was used to identify 198 bacterial isolates and 187 yeast isolates from the surfaces of four Irish smear-ripened cheeses; two bacteria and one yeast were not identified. The yeast and salt-tolerant bacterial counts were similar to those normally found on red smear cheese (2, 7, 8). The yeast microflora was very uniform, and D. hansenii was the dominant species in the four cheeses. Y. lipolytica was also isolated but only at relatively low levels from one cheese, Milleens cheese. These results are in agreement with those obtained for similar types of cheeses, including Tilsit, Limburger, and Danbo (5, 8, 23, 27). In contrast, the bacterial microflora showed considerable diversity, with 14 different species identified on the four cheeses (Table (Table4).4). Gubbeen cheese showed the greatest bacterial diversity (nine species), and Durrus cheese showed the lowest bacterial diversity (four species), while Ardrahan and Milleens cheeses contained six and seven species, respectively. Only two bacteria, C. casei and A. arilaitensis, were found in all cheeses, but the numbers of isolates of each of these species differed significantly in the four cheeses. These two species were much more important in Durrus, Ardrahan, and Milleens cheeses than in Gubbeen cheese. The other coryneform bacteria identified were either unique to a particular cheese or present in two or three of the cheeses. For example, Agrococcus sp. nov., B. linens, and M. gubbeenense were found only in Gubbeen cheese. A. arilaitensis and C. casei dominated Ardrahan, Durrus, and Milleens cheeses, while C. casei and C. variabile dominated Gubbeen cheese. C. casei and C. variabile have been isolated previously from Gubbeen cheese (2), and A. arilaitensis has recently been described by Irlinger et al. (18). C. casei and M. gubbeenense have also been isolated from a French farmhouse cheese (11). All these data suggest that the species mentioned above may also be important in the surface microflora of other smear-ripened cheeses.
Three Staphylococcus species were isolated, and each species was unique to a particular cheese; S. equorum was unique to Ardrahan and Milleens cheeses, although only one and two isolates, respectively, were found, S. saprophyticus was unique to Gubbeen cheese, and S. epidermidis was unique to Durrus cheese. S. equorum and S. saprophyticus have been isolated previously from the surfaces of traditional French cheeses (17). It is not surprising to find S. epidermidis as it is a component of the human skin microflora and many smear-ripened cheeses are made in small units, where they receive considerable manual handling. M. luteus was isolated only once (from Ardrahan cheese). This is surprising since micrococci are ubiquitous and quite salt tolerant and therefore would be expected to occur on the cheese surface. Perhaps they are less acid tolerant than staphylococci.
H. venusta accounted for 44% of the bacterial isolates from Milleens cheese and 3% of the bacterial isolates from Ardrahan cheese. However, PFGE patterns showed that these isolates were single clones. This is not the first time that Halomonas strains have been found in smear-ripened cheese. Maoz et al. (22) reported the presence of Halomonas variabilis, H. venusta, and two unidentified members of the genus Halomonas at relatively high levels (106 to 107 CFU/g) in a German red smear soft cheese. Maoz et al. (22) also felt that the presence of Halomonas may be an indicator of hygienic problems. Halomonas spp. have the ability to produce exopolysaccharides and to survive exposure to and grow in the presence of very high NaCl concentrations (up to 20%); they also occur in saline environments (e.g., seawater and salterns) (31). In this regard, the brining solution may constitute a niche for this organism, and the salt used is a possible source of contamination. It is not clear what contribution Halomonas species make to the ripening process.
Only two bacteria, C. casei and A. arilaitensis, were common to the four cheeses examined, and overall, we concluded that each cheese has a more or less unique microflora. Other species may have been present at lower levels, but the sampling and isolation methods used in the present study could lead to identification of only the dominant species. Recently, culture-dependent (isolation) and culture-independent (rRNA gene cloning and single-strand conformation polymorphism techniques) methods were used to compare the bacterial floras of French raw and pasteurized milk cheeses (11). The results showed that the dominant species were detected by all three methods but that greater numbers of species were detected using the molecular techniques. The species not detected by the culture-dependent technique are likely to have been subdominant populations. The dominant bacteria on the surfaces of red smear cheeses are corynebacteria, which are difficult to identify using only classical phenotypic analyses. The novelty of the present study is that both molecular and classical techniques were used to identify the dominant yeast and bacterial species from four different smear-ripened cheeses, thus ensuring that the identifications were correct. Strain differences within several species were also found. Culture-independent methods were not used, but the results show that the four cheeses had different microfloras, despite the fact that their methods of manufacture are quite similar. DNA-DNA hybridization may be necessary to ensure categorical identification of the strains that 16S rRNA gene sequencing was used for since sequencing of the 16S rRNA gene may not be sufficient to guarantee species identity (28).
Diversity at the strain level was also observed, based on the different PFGE patterns of the dominant species. Three different clones of C. variabile were found in the 18 isolates from Gubbeen cheese, 2, 3, 3, and 10 clones were found in the 76 isolates of C. casei from Ardrahan, Milleens, Gubbeen, and Durrus cheeses, respectively, and 1, 2, 3, and 4 clones were found among the 23 isolates of A. arilaitensis from Gubbeen, Milleens, Ardrahan, and Durrus cheeses, respectively. A. arilaitensis has been described only recently and was isolated from the French smear cheese Reblochon (18). Most PFGE clusters contained strains from a specific cheese; the exceptions were one cluster of C. variabile and one cluster of C. casei, which contained the same clone from two different cheeses (Gubbeen and Milleens cheeses and Durrus and Ardrahan cheeses, respectively). We have no explanation for this result. Brennan et al. (2), in a previous study of Gubbeen cheese, reported that a single clone of C. casei dominated the cheese surface, as determined by SpeI restriction analysis. Here, three distinct PFGE patterns were found. This may have been due to the use of AscI, which yielded better separation of higher-molecular-weight fragments and subsequently better differentiation between patterns. A combination of restriction enzymes AscI and SpeI with species of Arthrobacter and Microbacterium barkeri has also been reported to be a valuable tool for strain typing by PFGE (16). The present study showed that these two restriction enzymes were valuable for typing other genera of coryneform bacteria, including Corynebacterium and Brevibacterium, and for other species of Arthrobacter and Microbacterium.
Interestingly, none of the bacterial ripening cultures deliberately inoculated onto the surface (i.e., Brevibacterium sp. and Brevibacterium helvolum [OFR9, Visby] in the case of Gubbeen and Durrus cheeses and B. aurantiacum [BL2, Chr. Hansen] in the case of Ardrahan and Durrus cheeses) were reisolated. This indicates that they are not significant members of the dominant cheese surface microflora at the middle stage of ripening. However, they may be present at lower levels since only the dominant strains were isolated with the methodology used in this study. In a previous study of Gubbeen cheese, Brennan et al. (2) showed that the ripening starter used (BL2) was not isolated at any time during the ripening, and this was most likely due to inhibition by staphylococci at the early stages of ripening. In investigations on the microflora of a German smear-ripened cheese, the bacterial smear cultures were also not detectable in the isolates collected from cheeses during different stages of ripening (S. Goerges, J. Mounier, M. Rea, R. Gelsomino, T. M. Cogan, J. Swings, M. Vancanneyt and S. Scherer, unpublished data). Feurer et al. (10) showed that B. linens, which was heavily inoculated at the initial stages of ripening, had almost disappeared from the surface of a French smear-ripened cheese after 21 days and that after 31 days of ripening a species of Arthrobacter dominated the bacterial flora. In our study, the dominant organisms of the cheese surface were found after a shorter ripening time, 10 days, and at high levels (107 CFU/cm2), even if they were not deliberately inoculated onto the cheese surface. Several hypotheses could explain these results. The strains isolated were probably members of a resident adventitious microflora in the cheese-making environment (e.g., on the skin of the workers, on the shelves, and/or in the brine). The environmental conditions during ripening (e.g., the surface pH and the ripening temperature) could have been more favorable for the growth of the adventitious strains than for the ripening culture. These conditions would have allowed them to grow faster than the bacterial ripening culture, which would have been rapidly outnumbered on the cheese surface. Negative interactions between yeast and bacteria or between bacteria could have occurred. However, many of the same species were found on Milleens cheese, in which no commercial ripening culture was used, and in the three other cheeses. This confirms the importance of the adventitious, resident microflora in the dairy ripening process of red smear cheeses.
The commercial yeast starter used to smear Gubbeen and Durrus cheeses could not be identified using the current version of the FTIR database. When we performed a similarity analysis of all yeast spectra analyzed by FTIR spectroscopy in this study, it became apparent that the 20 isolates of the OFR9 culture formed a separate cluster. This result implies that the starter yeast was also not recovered in the cheese. This result was confirmed by mtDNA RFLP analysis. mtDNA restriction profiles were obtained from total DNA by use of a restriction enzyme with a recognition site rich in G and C, which resulted in overdigestion of the nuclear DNA and thereby gave specific bands for mtDNA. Therefore, it is likely that this strain was not a significant member of the cheese surface flora at the mid-ripening stage. Instead, each cheese was dominated by one to three strains of D. hansenii. These findings are in agreement with the results for a Danbo-type cheese (23), which showed that after 3 days, the commercial yeast culture was not part of the dominant yeast microflora and was outnumbered by another strain, which was isolated from the smear of previously ripened cheeses. Similarly, the results of a study on a German red smear cheese indicate that the commercial yeast starter strains used were not detected on the cheese surface during several different ripening stages (Goerges et al., unpublished). No yeast starter was used to smear Ardrahan and Milleens cheeses. Therefore, it is probable that the yeast which grew on the cheese surfaces largely occurred in the dairy environment (e.g., the brine and shelves). The two dominant strains on Ardrahan and Durrus cheeses had similar mtDNA RFLP profiles, which suggested that they are closely related, but PFGE analysis would be necessary to confirm that these strains are identical.
In this study, each cheese was sampled only once. This raises the question of whether the data are representative. The data are likely to be representative since the same species of bacteria were isolated from six different batches of Gubbeen cheese (unpublished data). Furthermore, there was little difference in the identities of isolates at several times during ripening (unpublished data).
This is the first time that the yeast and bacteria in smear-type cheeses have been identified using a combination of classical and molecular methods for categorical identification. This study provided significant understanding of the surface microflora of these cheeses and should aid in the development of more defined ripening starter systems for them. Inoculating bacterial or yeast species or even strains that are known to colonize the cheese surface may be valuable for preventing the growth of undesirable organisms, particularly pathogens and spoilage microorganisms, and the unnecessary cost of inadequate ripening. Our results also raise questions concerning the sources of these organisms in the cheese-making environment and how they are transferred to the cheese surface.
J.M. thanks Teagasc for awarding a Walsh Fellowship to him. This project was partially financed by the European Commission under project QLK1-CT-2001-02228.