E. Litopoulou-Tzanetaki Laboratory of Food Microbiology and Hygiene, Faculty of Agriculture, Aristotle University of Thessaloniki, 54006, Thessaloniki, Greece.
Aims: The aim of the study was to characterize isolates of lactic acid bacteria from Beyaz cheese and to study some of their technologically important properties.
Methods and Results: Seventy-seven lactic acid bacteria isolated from Beyaz, a Turkish white-brined cheese made from raw ewes’ milk without any starter, were classified by phenotypic criteria and sodium dodecyl sulphate-polyacrylamide gel electrophoresis of whole-cell proteins. Whole cells of Lactococcus lactis subsp. lactis and enterococci showed lipolytic and proteolytic activities. Strains were found that differed in terms of their acidifying and caseinolytic activity. Most of the enterococci isolates showed tyrosine decarboxylase activity; moreover, lactobacilli exhibited weak antibacterial activities against foodborne pathogens.
Conclusions: Strains of lactic acid bacteria with interesting biotechnologically important properties may be found in Beyaz cheese.
Significance and Impact of the Study: Lactic acid bacteria isolates with interesting biotechnological profiles may influence the quality and variety of dairy products, if they are used as starters, and their cheese-making characteristics seem promising.
In the present investigation LAB isolated from Beyaz cheese were characterized by phenotypic criteria and sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) of whole-cell proteins and some of their biotechnologically relevant properties were screened in order to be evaluated as starters in controlled fermentations.
MATERIALS AND METHODS
Source and maintenance of the strains
The LAB strains used in this study were isolated from 17 samples of Beyaz cheese, made from raw ewes’ milk, purchased from retail shops in the city of Ankara, Turkey. The LAB were grown on MRS agar (Lactobacillus agar according to Deman, Rogosa and Sharpe; Merck, Darmstadt, Germany) plates at 30°C for 5 d and colonies taken at random were grown in MRS broth, streaked on MRS agar plates and kept in litmus milk at −18°C.
Phenotypic characterization of the strains
The Gram-positive, catalase-negative rods and cocci isolated were characterized at the University of Ankara according to the methods and criteria of Sharpe and Fryer (1965) and Sharpe (1979). Additional tests were also performed on the strains at the Laboratory of Food Microbiology and Hygiene (Aristotle University of Thessaloniki) and criteria suggested by Schleifer and Kilpper-Balz (1984), Devriese et al. (1991), Schleifer et al. (1985), Collins et al. (1989) and Balows et al. (1991) were taken into consideration for taxonomy. Multiwell plates (Corning S.A., NY, USA) were used for sugar fermentation tests, according to the methodology suggested by API (API 50CHL; Bio-Merieux SA, Marcy l’Etoile, France). Tests for phenotypic characterization were conducted twice for each strain. The isolates were subjected to two successive subcultures in MRS (rods) or M17 (cocci) broth for activation before use.
Characterization by sodium dodecyl sulphate-polyacrylamide gel electrophoresis of whole-cell proteins
Activated cultures were inoculated in 20 ml broth and cells were collected from the growth medium after 24 h growth by centrifugation (12 000 g, 10 min, 4°C). The cell pellet was washed three times at ambient temperature with 0·01 mol 1−1 potassium-phosphate buffer, pH 7·0. The washed cells were disrupted by grinding with alumina (type 305; Sigma Chemical, St Louis, MO, USA) using a pestle and mortar. The extracts were suspended in 1 ml 0·01 mol 1−1 potassium-phosphate buffer, pH 7·0, and the alumina and cells separated by centrifugation (12 000 g, 10 min, 4°C). The supernatant fluids obtained were designated as crude cell-free extracts and had been previously submitted to PAGE, in order to check the effectiveness of the pressure technique. The appropriate extracts were subjected to SDS-PAGE according to Laemmli (1970). Registration of the protein electrophoretic patterns, normalization of the densitometric traces, grouping of strains by the Pearson product moment correlation coefficient (r) and UPGMA (Unweighted Pair Group Method using Average linkages) cluster analysis were performed by the techniques described by Pot et al. (1994), using the software package Gel Compar (version 4.0, Applied Maths, Belgium). Identification of the isolates was performed by comparison of their protein patterns to the fingerprints of reference strains of LAB. Reference strains of Lactobacillus paracasei subsp. paracasei (LMG 151), Lact. plantarum (LMG 14917), Lactococcus lactis subsp. lactis (LMG 0065), Enterococcus faecalis (LMG 19433), E. faecium (LMG 19434), E. hirae (LMG 6399), E. durans (LMG 19432), E. gallinarum (LMG 13129) and E. casseliflavus (LMG 10746) were kindly provided by Dr E. Tsakalidou (Agricultural University of Athens, Greece).
The API ZYM (Bio-Merieux SA) tests were applied on eight L. lactis subsp. lactis strains and 26 strains characterized phenotypically as E. faecium. Each strain tested was grown on slopes of MRS agar containing 0·5% glucose for 24 h. The growth was then removed from the surface and resuspended in 2 ml distilled water to produce a dense suspension. Two drops of the suspensions were added to each of the 20 cupules in the API ZYM strip. The strips were then incubated under a moist atmosphere at 37°C for 4 h. The reaction was terminated by the addition of one drop of each of the API reagents A and B. Enzyme activity was graded from 0 to 5 by comparing the colour developed within 5 min with the API ZYM colour reaction chart.
The strains were initially grown in MRS (rods) or M17 (cocci) broth and then in sterile reconstituted skim milk supplemented with yeast extract (0·3%) and glucose (0·2%) for two successive subcultures. Sterile reconstituted skim milk (100 ml) was inoculated with 1% of a 24-h activated culture and pH changes were determined using pH meters (glass electrode; HANNA Instruments, Padova, Italy) during incubation at 30°C after 6 h for lactococci and lactobacilli and after 5 h for enterococci. The pH of the cultures was also measured after 18 h for lactococci and/or 24 h for lactobacilli and enterococci and their proteolytic activity then estimated.
The proteolytic activity of the strains grown in milk was measured by the tyrosine method (International Dairy Federation Standard 149A, 1997; Brussels, Belgium). The proteolytic activity of the cultures results in liberation of the amino acids tyrosine and tryptophan from the milk substrate, which then react with the phenol reagent to form a blue colour which is measured at 650 nm (results expressed as μg tyrosine ml−1).
The ability of the test strains to decarboxylate histidine, tyrosine, lysine, ornithine, phenylalanine and tryptophane was detected as suggested by Joosten and Northolt (1989).
The well diffusion assay was used to study the antibacterial activity (Ahn and Stiles 1990). To rule out the possibility that inhibition was related to acid production, supernatant fluids were adjusted to pH 4·5, 5·5 and 6·5 and uninoculated broth at the same pH served as the control. Possible antibacterial activity by H2O2 formation was also checked by treatment with catalase. Supernatant fluids were filtered through a 0·45-μm filter for sterilization before use. Target strains were the test strains (each one against the others) and the pathogens Yersinia enterocolitica 0: 9/4360, a strain of Bacillus cereus, Escherichia coli 0: 44 NCTC 9702, Staphylococcus aureus NCTC 6571 and a Listeria monocytogenes strain which were obtained from the collection of the Laboratory of Food Microbiology and Hygiene (Aristotle University of Thessaloniki).
RESULTS AND DISCUSSION
Characterization of the isolates by phenotypic criteria and the sodium dodecyl sulphate-polyacrylamide gel electrophoresis of whole-cell proteins
All strains were Gram-positive, catalase-negative cocci, which produced no gas from glucose (60 isolates), and/or rods (17 isolates). Among the cocci, 12 isolates were able to grow at 10 and 40°C, but not at 45°C, and in pH 9·2 broth, except for two (SL2 and SL3). In addition, these strains did not grow in pH 9·6 broth, with the exception of two strains (SL7 and SL15), and they did not survive at 60°C for 30 min. All of the 12 strains could grow in 4%, but not in 6·5%, NaCl broth, produced NH3 from arginine and hydrolysed esculin. The strains formed acid from lactose, maltose, salicine and ribose but acid production from mannitol, sucrose and D-xylose was strain dependent. The phenotypic characteristics of the strains (Table 1) suggest their close resemblance to L. lactis subsp. lactis (Sharpe 1979; Schleifer et al. 1985; Balows et al. 1991). Forty-eight isolates of cocci were able to grow at 10 and 45°C, in 6·5% NaCl and pH 9·6 broth; they also formed NH3 from arginine but not CO2 from glucose and were characterized as enterococci. Six of them seemed to be E. faecalis, as suggested by their ability to grow in the presence of 0·4% potassium tellurite, reduce 0·01% triphenyl tetrazolium chloride (TTC) and ferment sorbitol (Sharpe 1979; Schleifer and Kilpper-Balz 1984; Devriese et al. 1991). Forty-two enterococci strains, unable to either grow in the presence of tellurite or reduce tetrazolium, were differentiated by their ability to form acid from sugars. Thus, 26 strains producing acid from mannitol and arabinose were characterized as E. faecium, six strains that were mannitol and arabinose negative but fermented melibiose and sucrose seemed to be E. hirae and 10 strains characterized by their inability to ferment melibioze and sucrose and unable, in general, to ferment sugars were classified as E. durans (Schleifer & Kilpper-Balz 1984; Devriese et al. 1991).
Table 1. Phenotypic characteristics of the lactic acid bacteria isolates from Beyaz cheese
The 17 isolates of Gram-positive rods grew at 15°C and did not form either CO2 from glucose or NH3 from arginine. These characteristics suggest their classification as facultatively heterofermentative lactobacilli (Sharpe 1979; Balows et al. 1991). Three of 17 isolates did not form acid from arabinose, melibiose, raffinose, rhamnose and sorbose and were characterized as Lact. paracasei subsp. paracasei (Collins et al. 1989; Balows et al. 1991). The 14 isolates of rods were classified as Lact. plantarum, as suggested by their sugar fermentation patterns. All strains fermented arabinose, cellobiose, lactose, maltose, melibiose, raffinose, ribose, salicine, sucrose and trehalose (Sharpe 1979; Balows et al. 1991). The strains did not form acid from rhamnose and acid production from sorbose, sorbitol and xylose was variable and strain dependent.
Using the Gel Compar software package, all 77 protein patterns were compared with protein fingerprints of reference strains and the resulting dendrogram is shown in Fig. 1. According to the SDS-PAGE results, the phenotypic characterization was confirmed for eight L. lactis subsp. lactis strains. In agreement with the visual interpretation, strains SL7, SL32 and SL15 were removed and clustered separately (at a correlation level of r=0·83); these strains were not assigned to any species but their cluster analysis show similarities to the Enterococcus group. In addition, strain SL33 was classified as E. hirae by SDS-PAGE (Table 2).
Table 2. Identification of lactic acid bacteria isolates from Beyaz cheese according to phenotypic characterization and sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) of whole-cell proteins
Table 3. Table 2 (Continued)
Four strains phenotypically characterized as E. faecalis were clustered together with the E. faecalis reference strain (r=0·66). In the E. faecium group, 43 strains were allocated and clustered (r=0·68) with reference strains of E. hirae (27 strains at a mean correlation level of r=0·73), E. faecium (14 strains, r=0·77) and E. durans (two strains, r=0·83). The phenotypic characterization was confirmed for nine E. faecium, two E. durans and four E. hirae strains. Fifteen strains phenotypically characterized as E. faecium and seven strains phenotypically characterized as E. durans were clustered with E. hirae. In the E. hirae cluster, four subclusters delineated at correlation levels of r=0·81, 0·83, 0·92 and 0·93 were recognized. Similarly, in the E. faecium cluster, two subclusters were separated at r=0·83. In this cluster, two E. faecalis (FL29 and FL30), two E. hirae (H21 and H17) and one E. durans strain (DR23) were also allocated. Two strains, phenotypically characterized as E. faecium (FC1 and FC2), did not resemble each other or any of the reference strains and they were removed separately. The genus Enterococcus has recently undergone considerable taxonomic revision and at least 18 species grouped into ‘species groups’ have been described to date (Devriese et al. 1991). The E. faecium species group consists of species, such as E. durans or E. hirae, considered potentially pathogenic while E. faecium is usually considered to be part of the normal intestinal flora. In addition, the genus Enterococcus is not a phylogenetically coherent and homogeneous genus and includes atypical enterococcal strains (Giraffa et al. 1997), as possibly were the strains SL7, SL15, SL32 and SL33, phenotypically characterized as lactococci. Enterococci, present in raw milk, may develop during cheese manufacture and ripening and may represent the predominant microflora found in cheese made from raw milk (Neviani et al. 1982), as also observed for Beyaz cheese.
Two Lact. paracasei subsp. paracasei strains (LC6 and LC11) and 13 strains phenotypically characterized as Lact. plantarum were identified by SDS-PAGE as Lact. paracasei subsp. paracasei (similarity 81%). In the Lact. plantarum cluster only Lact. plantarum strain LP21 and Lact. paracasei subsp. paracasei strain LC1 were allocated.
The high similarities of the test strains with the reference strains confirm the close phylogenetic relationship between them (Kandler and Weiss 1986; Collins et al. 1991; Pot et al. 1993). The fact that profiles of our isolates did not perfectly correspond with those of the reference strains did not significantly affect identification and dissimilarities may be due to the different origin of the reference strains.
Hydrolysis profiles of whole cells
Hydrolysis profiles from whole cells show (Table 3) that L. lactis subsp. lactis strains were able to hydrolyse the eight-carbon (β-naphthyl caprylate) and the 14-carbon substrate (β-naphthyl myristate), showing esterase/lipase or lipase activity, respectively, but the mean activity was low. It is well known that LAB exhibit low activity on milk fat and that esterases from lactococci preferentially degrade short-chain fatty acids (Kamaly et al. 1988; Tsakalidou et al. 1992). The Leu-aminopeptidase activity exhibited by all of the strains was also low, while 50% of them showed weak Val-aminopeptidase activity. Strong acid phosphatase and phosphoamidase activity was displayed by all L. lactis subsp. lactis strains. The strains of E. faecium exhibited weak activities (< 2·0) of esterase (78·8% of the strains), esterase/lipase, lipase (96·2%) and Leu- and Val-aminopeptidase (60·9%).
Table 3. Mean hydrolase activities* of Lactococcus lactis subsp. lactis and Enterococcus faecium strains from Beyaz cheese
Enzymes from L. lactis and E. faecium may, therefore, contribute to fat hydrolysis during cheese ripening, if sufficient cells are present. It seems likely that the role of LAB in fat hydrolysis starts after the milk fat has been partially degraded by milk lipase or other microbial lipases. In addition, strains of L. lactis and E. faecium that were found to produce the exopeptidases Leu- and Val-aminopeptidase may have some effect on milk proteins. It is believed that enterococci contribute to the flavour development of cheeses (Thompson and Marth 1986; Litopoulou-Tzanetaki et al. 1993; Tsakalidou et al. 1993).
Acidification ability and proteolytic activity
With respect to the acidifying activity of the strains (Fig. 2), it seems that none of the L. lactis subsp. lactis strains can be characterized as fast, as they did not reach a pH of 5·0 ± 0·2 in 6 h at 30°C (Huggins and Sandine 1984). Strains SL47, SL24 and SL2 were slow initially (6 h) but acid production from strains SL24 and SL2 was enhanced later and they finally (18 h) accumulated enough acid (ΔpH(18 h) 1·70 and 1·95 units, respectively) in the milk. Strains SL1, SL35, SL34 and SL46 were faster initially and the ΔpH(18 h) ranged between 2·10 and 2·45 pH units. Plasmid linkage of lactose fermentation in lactococci has been demonstrated and slow variants defective in lactose metabolism may be found (McKay et al. 1976).
Lactobacillus strains differed in their ability to reduce the pH of milk initially and there were strains that did not change the pH of milk at 6 h. Nevertheless, after 24 h incubation the ΔpH(24 h) of the strains were similar and ranged between 1·00 and 1·40, except for strain LC1, which had a ΔpH(24 h) of 1·60. Lactobacillus casei and Lact. plantarum may ferment lactose through a β-galactosidase activity, but some strains also show a β-phospho-galactosidase activity (Herrero et al. 1996).
The acidifying abilities of enterococci at 30°C were, in general, low and only eight cultures lowered the pH of milk to < 5·0 after 24 h incubation. The ΔpH(24 h) was < 1·00 for 71% of the strains. Nevertheless, the ΔpH(5 h) of enterococci was, in general, higher than that of lactobacilli at 6 h, but there was a tendency for the strains to become slow after 5 h.
Our test strains were characterized by a different caseinolytic breakdown ability (Fig. 3). The range of proteolytic activity was 25·5–77·2 μg tyrosine ml−1 for L. lactis subsp. lactis strains. The proteolytic activity of the enterococci strains was measured at levels between 29·3 (E. faecium strain FC10) and 121·1 μg ml−1 (E. faecium strain FC11) and ˜36% of the strains had activity > 60 μg tyrosine ml−1. The proteolytic activity of lactobacilli ranged between 29·4 (strains LP21 and LP22) and 57·3 μg tyrosine ml−1 (seven strains). The data reported here on proteolytic activity suggest that there was no relationship between the proteolytic and acidifying activities of the strains, as also suggested by Bottazzi (1962) and Fontina et al. (1998) for strains of lactobacilli. Thus, strains with the strongest acidifying abilities (lactobacilli strains LC1, LP12 and LP13 and enterococci strains DR36, FC9, FC10 and FC24) did not exhibit the highest proteolytic activities and there were strains with very low acidifying but high proteolytic activity (e.g. E. faecium strain FC17) and strains with high acidifying and proteolytic activity (for example, L. lactis subsp. lactis strain SL1). The proteinase system of Lactococcus has been studied for several years and consists of cell wall-bound proteinases and several peptidases (Bockelmann 1995). Enzymes formed by Lactobacillus strains were studied in detail and many authors have described enzymes that were biochemically similar to those of Lactococcus and their importance for cheese ripening is obvious (Bockelmann 1995). The proteolytic activity and acid production of enterococci during growth in milk are sometimes comparable to those of Strep. thermophilus (Gatti et al. 1994).
Lactococci, as well as lactobacilli, isolates did not degrade histidine, tyrosine, lysine, ornithine, phenylalanine and threonine. All enterococci strains, except SL33 and DR31, decarboxylated tyrosine. Decarboxylating bacteria can find suitable conditions to proliferate and produce biogenic amines during ripening of cheeses. Tyramine is the only biogenic amine produced after growth in milk by E. faecalis and E. faecium in the presence of a pool of free amino acids as precursor (Giraffa et al. 1995).
Six strains of L. lactis subsp. lactis and 12 enterococci strains formed small (2 mm) inhibition zones against Listeria monocytogenes but the zones disappeared after treatment with catalase. The strains of lactobacilli showed an inhibitory effect against Y. enterocolitica, Escherichia coli and B. cereus at pH 4·5 and the zones of inhibition remained almost the same after the addition of catalase except for strains LP12, LP13, LP15 and LP18, for which no inhibitory effect against Escherichia coli was detected after treatment with catalase. The inhibition zones were small (Table 4), which possibly suggests antibacterial activity other than bacteriocin.
Table 4. Antibacterial activity* of lactobacilli isolates from Beyaz cheese
In conclusion, the results obtained suggest that LAB found in Beyaz cheese may contribute to ripening changes of the cheese by their acidifying, proteolytic, lipolytic and inhibitory activities. Our results have also shown that a rich source of strain variability exists in the microflora present in Beyaz cheese made from raw milk without any commercial starter. Cheese may, therefore, constitute a source of strains with physiological properties of biotechnological interest. Our results seem to indicate that lactococci and lactobacilli strains with marked proteolytic and acidifying abilities can be found in cheese and this product may be the origin of industrial developments in the future.