Three molecular techniques (RAPD-polymerase chain reaction analysis, plasmid profile and pulsed-field gel electrophoresis) were used for a preliminary approach to type, at strain level, enterococci isolated from a 24-h-old home-made Pecorino Sardo (protected designation of origin) cheese. A high genetic polymorphism was found. Clusters obtained by the RAPD technique and plasmid profile analysis contained different strains. Pulsed-field gel electrophoresis proved to be an efficient and highly reproducible typing method. In addition, by combining the results from plasmid profile analysis and pulsed-field gel electrophoresis, it was possible to identify closely related strains probably belonging to the same clonal lineage.
Home-made Pecorino Sardo ‘maturo’ is a semi-cooked (42–45°C) 2-month or more ripened, protected designation of origin (PDO) cheese manufactured in Sardinia by using raw whole ewe's milk and without the addition of a selected or natural starter culture. The inoculum is provided by bacterial cells present in the milk. In previous studies [1, 2], the microflora of Pecorino Sardo was studied at the genus and species level. Phenotypic tests suggested that after 24 h of clotting, the microflora of this cheese, which reaches 109 CFU g−1, is formed by only two genera of bacteria, namely Lactococcus and Enterococcus. After 1 month of ripening, a third bacterial component, mesophilic rods, probably belonging to facultatively heterofermentative lactobacilli, is also isolated among the dominant bacterial population.
Lactococci are the large majority of the isolates at 24 h and enterococci represent 1–10% of the total microflora, but, during ripening, the level of lactococci decreases, while enterococci maintain a constant number of CFU g−1.
It is well known that enterococci are a relevant part of the bacterial population of several artisanal cheeses produced by using raw ewes' milk and they are not only under study for their technological application, but also for their potential risks . From this relevant presence of enterococci in Pecorino Sardo arises the need for in-depth characterisation of this group of starter bacteria.
Unfortunately, genetic typing techniques, such as plasmid profiling [4, 5], analysis of chromosomal DNA macro-restriction patterns by pulsed-field gel electrophoresis (PFGE) [4–9], RAPD-PCR [4, 8, 10] and ribotyping [4, 5, 7], which are nowadays widely used to characterise clinical isolates of enterococci, have not yet been used, as far as we know, to type bacterial strains composing this natural dairy microflora.
The aim of this work was to obtain, by means of three different molecular approaches, preliminary typing data of isolates coming from this home-made cheese and to develop a strategy for a further investigation about the genetic diversity of this natural microbial ecosystem. We analysed the enterococcal isolates from one home-made Pecorino Sardo cheese loaf, after 24 h of ripening, by means of three different molecular methods (RAPD-PCR, plasmid profile and PFGE).
Materials and methods
Bacterial strains and culture conditions
A total of 49 isolates, phenotypically identified as enterococci, coming from a single 24-h-ripened loaf of Pecorino Sardo manufactured in Sardinia were analysed. All the isolates were maintained at −80°C and subcultured in M17 medium (Oxoid).
The DNA extraction for the RAPD-PCR analysis was performed by microwave oven treatment using a synthetic resin (Gene Releaser, Bioventure, TN, USA) as described by Cocconcelli et al. . DNA amplification was made in a Thermal Cycler 2400 Perkin Elmer, using the following primers of arbitrary nucleotide sequence: M13 (5′-GAG GGT GGC GGT TCT-3′) and OPB7 (5′-GGT GAC GCAG-3′) . The reaction mixture consisted of 2 µl of sample, 13 µl of Genereleaser, 2.5 U of Taq polymerase (Boehringer), 2.5 µl of 10×PCR buffer, 3.5 mmol l−1 of MgCl2, 200 µmol l−1 of each dNTP (Boehringer), 1 µmol l−1 of primer (Pharmacia), in a final volume of 25 µl, overlaid with two drops of mineral oil. The amplification reactions were all preceded by one cycle at 90°C for 10 min; when primer M13 was used, 40 cycles were performed, each consisting of 94°C for 1 min, 45°C for 20 s, 72°C for 54 s, followed by one cycle at 72°C for 2 min. With primer OPB7, the amplification process consisted of 35 cycles of 94°C for 1 min, 25°C for 20 s, 72°C for 54 s and an extension cycle at 72°C for 2 min. The PCR products were separated by electrophoresis on 1.5% agarose gel in Tris-acetate buffer at 85 V and then stained in ethidium bromide solution (0.5 µg ml−1).
The extrachromosomal DNA was extracted for plasmid screening by alkaline lysis according to Vescovo et al.  and separated by agarose gel electrophoresis (0.6–0.8% agarose) in 1×TAE buffer. Plasmid sizes were determined by comparing the bands with the Supercoiled DNA molecular weight marker (Gibco-BRL).
Chromosomal DNA for pulsed-field gel electrophoresis was prepared in situ in ‘low melting point’ (Sigma) agarose blocks according to Morelli et al. , and then digested with SmaI restriction endonuclease (Boehringer) according to the supplier's instructions. DNA fragments were separated in 0.8% (w/v) Chromosomal grade agarose (Bio-Rad) gel in 0.5×TBE buffer. Electrophoresis was performed in a CHEF-DRIII (Bio-Rad), containing 0.5×TBE buffer equilibrated at 12°C, at a constant voltage of 6 V cm−1, with different pulse times for a total running time of 16 h:1–20 s for 5 h; 1–5 s for 5 h; 10–40 s for 6 h. The gels were then stained in ethidium bromide (0.5 µg ml−1). PFGE marker I from Boehringer was used as a molecular size standard.
The carbohydrate fermentation profile was assayed by API 50 CHL system (Biomerieux). The acidifying activity was evaluated by measuring the pH reached in reconstituted skimmed milk (Oxoid) from a standard inoculum after 6 and 24 h of growth at 42°C. The proteolytic activity of milk-grown cultures from a standard inoculum was measured by the OPA method  after 24 h of growth at 42°C.
Results and discussion
The available information on the microbial composition and genetic diversity of the bacterial population present in home-made raw milk cheeses is limited by the lack of use of reliable methods for comparing strains. A strain is an isolate or group of isolates that can be distinguished from other isolates of the same species by phenotypic or genotypic characteristics or both . Methods based on phenotypic characteristics are not effective enough in distinguishing strains, they are often time-consuming, tedious, not reproducible and difficult to perform.
In the present work, different molecular typing techniques, RAPD-PCR analysis, plasmid profile and PFGE have been applied in order to assess the strain and clone composition of the natural and complex microbial ecosystem found in home-made Pecorino Sardo cheese.
We analysed, by RAPD-PCR, analysis a total number of 49 Enterococcus isolates from one single 24-h-old cheese by using, in separate amplifications, M13 and OPB7 primers. The visual inspection analysis of the band patterns allowed us to divide the 49 isolates examined into 7 groups, each containing two or more isolates showing the same profile with both primers, while 23 isolates gave unique profiles, different from the others when amplified with one or both primers. All the isolates belonging to a cluster could then be considered, according to RAPD, as duplicates of a single strain isolated several times, because the profiles from the individual isolates in each group were identical for each of the primers.
Plasmid analysis, performed on all the 49 isolates, grouped 24 isolates into 7 clusters consisting of two or more isolates showing the same plasmid pattern, while 16 isolates were found to have unique plasmid profiles. Nine isolates did not fall into any of these groups by this technique because they lacked plasmids detectable by the isolation and separation conditions used by us. None of the 7 clusters fully overlapped those obtained by RAPD technique as at least one of the isolates clustered in a RAPD group was found to belong to a different group obtained by plasmid analysis (see below). On the contrary, eight isolates showing unique plasmid patterns were found to be unique also when assayed by RAPD.
The analysis of the extrachromosomal DNA (Fig. 1) allowed us to highlight the presence of isolates that seemed to represent a clonal lineage of an ancestral strain. In particular, we grouped 22 enterococcal isolates into 11 closely related groups, each containing one or more isolates with the same plasmid profiles. The PSS 119 strain seemed to be the wild-type, in fact it carried the highest number of plasmids (six plasmids with approximate sizes from 2.7 to 12 kb).
It is tempting to speculate that these strains are the dominant biotypes among the enterococci isolated from the 24-h cheese and that it would be possible to assume that they are partially cured derivatives of the same parent.
Looking through the RAPD patterns of these closely related isolates (Figs. 2 and 3), we were able to observe some disagreements between the results obtained from this technique and the plasmid content analysis. The PSS 119, 52, 118, 133, 135 and 34 isolates showed the same RAPD patterns with both the M13 primer and the OPB7 (Fig. 3) but, as regards the plasmid profile, only the two isolates PSS 133 and PSS 135 carried the same plasmids, while the others differed from one another in lacking one or more plasmids and therefore we can suppose that they are all clones of the same strain (Fig. 1). Another example of isolates showing the same RAPD profile (Fig. 2) but different plasmid content (Fig. 1) is shown by the isolates PSS 28, 34 and 53. These results could suggest that RAPD patterns are not influenced by the presence of plasmids, but also that this technique is unable to detect variations in the genetic extrachromosomal complements. On the contrary, it should also be noted that the PSS 47 and the PSS 53, showing exactly the same plasmid profile (Fig. 1), differed in their RAPD patterns with primer M13 for the presence of one band (Fig. 2). The isolates PSS 34 and PSS 38 behaved in a similar way, differing in one band by RAPD (Fig. 2) but harbouring the same plasmids (data not shown).
These results point out the need for a careful selection of the typing technique, as they could provide a different clustering of strains, and different insight into the genetic composition of isolates.
Once isolates were recognised as having identical plasmid profiles, a representative isolate of each putative clonal group was compared by PFGE in order to see if this technique could actually show that they are derivatives of the same strain. All the isolates analysed revealed the same DNA fragment macrorestriction pattern when digested with SmaI restriction endonuclease, with the only exception of the PSS 118 strain that lacked a 60–65-kb band, probably the linear form of one large plasmid (Fig. 4). These results confirmed they were clones indistinguishable as regards their chromosomal DNA, but differing in plasmid content.
Phenotypic tests performed on all the 22 closely related isolates did not reveal any significant difference in their phenotypic behaviour; the acidifying and proteolytic activity in milk were almost the same for all clones. The only differences found were related to carbohydrate fermentation: the PSS 119 (the wild-type) seemed to be the only one able to ferment the sucrose. As regards the other sugars, no difference between the clones was observed.
As for the nine isolates indistinguishable by plasmid analysis, PFGE grouped them into one group containing six isolates with the same DNA restriction profile and three strains showing unique PFGE patterns. These three strains also gave unique profiles when analysed by RAPD.
Our results indicate a high heterogeneity in the biotypes among enterococci present in this cheese 24 h after its manufacture. RAPD analysis made it possible to identify 30 different biotypes, while by combining the results obtained from plasmid profile analysis and PFGE, we observed the presence of a total of 27 different genotypes among the 49 enterococci isolates.
It should also be noted that RAPD and plasmid analyses are not always in agreement in discriminating strains. Indeed strains with identical RAPD patterns did not necessarily have identical plasmid content. The opposite is also true: strains that showed different RAPD patterns with one or both the primers could carry the same plasmids. We found quite a high level of agreement between the two techniques, only for isolates showing unique patterns.
This work clearly suggests that only a combination of different techniques is able to provide a complete picture of the ecology of a natural dairy microbial population. The need for a careful selection of the typing technique used for enterococci was also pointed out by Tomayo and Murray  and by Chiew and Hall .
We can also conclude, in accordance with Clark et al. , that, for ecological studies, it is impossible to rely on a single method and that ‘a progression from plasmid analysis to PFGE is more effective’.