INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

There is an expanding range of bacterial cultures being used in food products and increased interest in probiotics and health benefits. A major issue of concern is the safety of cultures that can be consumed live and in large quantities. Several species of lactic acid bacteria are used for this purpose (1, 15), and the most controversial are the enterococci. Enterococci are used as probiotics to improve the microbial balance of the intestine and to treat gastroenteritis in humans and animals (2, 27). They are found as a component of the natural flora of certain foods, where they may have beneficial effects. In certain cheeses, they are significant in ripening and the development of flavor (3, 41, 42). Some enterococci also produce bacteriocins which may have anti-Listeria activity (17). Enterococci occur as food contaminants and may be implicated in food-borne illnesses. The implications of enterococci for food safety has recently been reviewed (13).

The genus Enterococcus is of particular medical relevance because of its increased incidence as a cause of disease, notably in nosocomial infections, and because the available antibiotic therapies are being compromised by evolving antibiotic resistance (22, 37). Enterococcal cultures have featured in dairy fermentations for decades, and isolates with histories of safe use are being promoted as probiotic cultures. These bacteria probably represent the largest risk to human health of any species currently used in this way.

The differences between an enterococcal pathogen and an apparently safe food use strain is unclear, and the potential for the latter to acquire virulence factors by gene transfer has not been investigated. It is already established that the molecular taxonomy of enterococci does not lead to a distinction between these two types of strains. Our knowledge of virulence in enterococci is incomplete, in part due to the fact that they are normal human commensals and, as such, have subtle virulence traits that are not easily identified. Virulence traits include adherence to host tissue, invasion and abscess formation, modulation of host inflammatory responses, and secretion of toxic products. Several genes for virulence factors in Enterococcus faecalis have been characterized (see Table 2), and their effects have been demonstrated in animal models (4, 19, 21, 23, 34) and cultured cells (25, 31). In this study, the incidence of known virulence factors in medical, food, and dairy starter Enterococcus strains was investigated.

Enterococci are noted for their capacity to exchange genetic information by conjugation (5), and these processes are known to take place in the gastrointestinal tract (19). As well as transmissible antibiotic resistance plasmids, virulence factors such as hemolysin-cytolysin production and the capacity for adhesion are known to be transmissible by highly efficient gene transfer mechanisms (4, 16, 25, 43). Thus, this study also investigated the possibility that apparently safe starter or probiotic cultures might acquire virulence factors by conjugation. In addition, the potential for conjugation to take place was investigated by screening for the presence of newly identified pheromone-encoding genes whose products are involved in initiating characterized conjugation processes (7; The Institute for Genomic Research E. faecalis genome database [http://www.tigr.org/tdb/mdb/mdb.hmtl]). Sex pheromones are also thought to be involved in eliciting an inflammatory response. These pheromones are chemotactic for human and rat polymorphonuclear leukocytes in vitro and induce superoxide production and secretion of lysosomal enzymes (11, 24, 33). They may therefore be considered virulence factors.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and media. Escherichia coli JM109 was obtained from Promega. L broth (26) was used for growth of E. coli at 37°C and was supplemented with ampicillin at 100 µg ml¹ when appropriate. The enterococcal strains and plasmids used in this study are listed in Table 1. Enterococcal strains were grown in brain heart infusion (BHI) broth and agar (Oxoid) at 37°C. For enterococcal mating experiments, the medium was supplemented with erythromycin at 50 µg ml¹, kanamycin at 500 µg ml¹, streptomycin at 1,000 µg ml¹, tetracycline at 10 µg ml¹, rifampin at 25 µg ml¹, and fusidic acid at 25 µg ml¹ when appropriate.

Enterococcal isolates. Enterococcus isolates were collected from National collections and various companies and educational establishments, including the American Type Culture Collection, the Deutsche Sammlung von Mikroorganismen, the National Collections of Industrial and Marine Bacteria, CIT INIA Spain, the Danish Veterinary Laboratory, the University of Bath, the University of Michigan, and the Public Health Laboratories, Collindale. Strains were identified to the species level with API ID32-STREP kits (bioMerieux Vitek Inc., Hazelwood, Mo.). One starter strain (F41) was isolated from a Danish fermented-milk product by diluting the culture in one-quarter-strength Ringer solution and plating dilutions onto BHI agar. Isolated colonies were then subjected to 16S rRNA sequence analysis and biochemical testing. The results indicated that this was an E. faecalis strain. Forty-nine strains were used for testing and comprised panel A. Panel A consisted of 9 dairy starter, 22 food, and 18 medical strains representing different source groups (milk, cheese, meat, blood, pus, urine, feces, hospital environment).

Isolation of enterococcal DNA. Total DNA was extracted from overnight Enterococcus cultures by the method of Lewington et al. (28).

DNA manipulations. DNA manipulations, including restriction endonuclease digestion and Southern blotting, were carried out using standard methods (31). Southern hybridizations were carried out using the ECL direct nucleic acid labeling and detection system (Amersham) under standard-stringency conditions, except that 0.25 M NaCl was used in the hybridization buffer. For Southern hybridizations using the efaA probe, high-stringency conditions were used, with the primary wash buffer containing 0.1× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). Restriction enzymes, 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal), and isopropyl--D-thiogalactopyranoside (IPTG) were purchased from Promega and used in accordance with the manufacturer's instructions. Plasmid DNA from E. coli used for sequence analysis was purified using a QIAprep spin miniprep kit (QIAGEN). Sequencing was performed with an Applied Biosystems 373A automated sequencer and a BigDye Terminator Cycle Sequencing Kit (PE Applied Biosystems) in accordance with the manufacturer's instructions.

Generation of primers and probes. PCR primers for the efaA, gelE, cyl, and agg virulence genes described in Table 2 were designed on the basis of published sequences and EMBL database sequences. PCR primers for esp were originally developed on the basis of primer sequences supplied by V. Shankar (University of Oklahoma), and those for sex pheromone determinants were based on sequences in the E. faecalis genome database at The Institute for Genomic Research. A degenerate primer was designed for the cAD1 sex pheromone determinant on the basis of the amino acid sequence and enterococcal codon bias. The Wisconsin Genetics Computer Group sequence analysis software package, version 8, was used for sequence analysis and comparisons. Multiple-sequence alignments using sex pheromone plasmids pAD1, pPD1, and pCF10 were performed, and regions of high conservation were used to generate primers for aggregation substance (agg). Strain P4 containing pAD1 was used to generate the agg-specific probe. No suitable primers were produced for the structural cytolysin genes (cylL_L and cylL_S) due to the extensive sequence homology between the two genes. All PCR-generated fragments were cloned into pGEM-T vectors and sequenced to confirm identity (data not shown). The target DNA fragments amplified by PCR served as the DNA probes for Southern hybridization experiments.

Production of gelatinase and hemolysin. Production of gelatinase was determined on Todd-Hewitt agar containing 30 g of gelatin (Difco) per liter. Single colonies were streaked onto plates, grown overnight at 37°C, and placed at 4°C for 5 h before examination for zones of turbidity around the colonies, indicating hydrolysis. For investigation of hemolysin production, strains were streaked onto layered fresh horse blood agar plates and grown for 1 to 2 days at 37°C. Zones of clearing around colonies indicated hemolysin production (Table 4).

Generation of transconjugants. Broth matings were carried out by the method of Clewell et al. (8) using overnight cultures and overnight matings. Filter matings were performed using a 1:1 donor-recipient mixture. Briefly, 0.5 ml of each overnight culture was mixed and harvested using 0.45 µm-pore-size filters and incubated on BHI agar plates at 37°C overnight. Cells were harvested and diluted in BHI broth and grown for 2 days at 37°C on BHI agar plates.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Three different groups of enterococcal isolates were screened for the presence of characterized virulence determinants. These groups were intentionally added starter strains, strains found to occur naturally in food, and strains isolated from human clinical infections.

Distribution of virulence determinants in food and medical isolates. PCR amplifications and Southern hybridizations of enterococcal DNA with the specific primers and probes for each gene described (Tables 2 and 4) revealed distinct trends in the occurrence of virulence determinants in the food and medical strains. A summary of the pattern of virulence gene incidence in E. faecalis and E. faecium strains in the three groups is presented in Table 5.

(i) E. faecalis. E. faecalis strains harbor significantly more virulence determinants than do E. faecium strains, and a distinct trend occurs within the three groups. All of the E. faecalis strains tested, including the two starter strains, possessed multiple virulence determinants (between 6 and 11). Three E. faecalis strains (two medical strains and one food strain) possessed all of the virulence genes. The sex pheromone determinants were detected exclusively in the E. faecalis strains. They occurred in all of the strains tested, sometimes with and sometimes without the agg determinant. The agg virulence determinant was always associated with the presence of pheromone determinants. No other pairs of determinants occurred together exclusively. The majority of strains (>89%) possessed the efaAfs determinant. For the gelE, cylMBA, and esp determinants, a distinct trend was evident. Starter strains had fewer virulence determinants than did food strains (10 Versus 49%), which, in turn, had fewer than medical strains (58%).

(ii) E. faecium and E. durans. A significantly different pattern was seen in the E. faecium strains. All of the strains were clear of the agg, cylMBA, and sex pheromone determinants. All of the E. faecium starter strains were clear of virulence determinants, except for efaAfm. For some food and medical E. faecium strains, weakly hybridizing bands were obtained under standard conditions which were absent under high-stringency conditions, using the efaAfs-specific probe. PCR amplification products were obtained using the efaAfs-specific primers for these strains. However, for E. faecium starter strains, no signal was obtained with the efaAfs probe in Southern hybridizations and no products were obtained using the efaAfs-specific primers. Of the three E. durans strains tested, starter strain F25 did not hybridize with either of the efaA probes or produce amplification products with the efaAfs-or efaAfm-specific primers. The E. durans F24 and F26 food strains produced the most strongly hybridizing bands with efaAfm-and efaAfs-specific probes, respectively (data not shown).

Silent genes are present in starter, food, and medical strains. Silent gelE genes occurred in E. faecalis strains from all groups. Gelatinase activity could not be detected in E. faecalis F41, F1, F4, P21, P26, P31, and P36, although these strains carried the gelE gene. One silent gene was identified in an E. faecium strain, P20. In medical strains E. faecalis P14 and E. faecium P20, gelatinase activity was lost during subculture from the original stock and no gelatinase activity was detected. Two other medical strains. E. faecalis P21 and P36, appeared to have no hemolytic activity but carried cyl genes. A trend was observed in which starters had fewer silent genes than did food strains, which, in turn, had fewer than did medical strains.

Transfer of virulence plasmids from medical strains into food starter strains by conjugation. Transfer of pAD1 into E. faecalis F41 by filter mating was achieved at a frequency of 2.1 × 10⁶ per donor. Broth matings used to generate FI9530 allowed transfer of pAD1 into E. faecalis FA2-2 at a frequency of 3.6 × 10⁵. From this donor, pAD1 was introduced into E. faecalis F27 by filter mating at a frequency of 5.7 × 10³. E. faecalis FI9530 was used directly in matings with starter strain E. faecium F12 using lacZ selection. Several attempts at both broth and filter matings using various donor-recipient ratios (data not shown) did not result in the generation of any transconjugants.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Enterococci are known to cause serious disease in humans and are becoming a major and increasing problem in nosocomial infections due to the acquisition of antibiotic resistance determinants. Acquired antibiotic resistance, combined with natural resistance to several major classes of antibiotic and the natural resistance of these organisms to low pH, high salt concentrations, and high temperatures, contributes to their survival. Enterococci are found in foods as contaminants, and they are intentionally added to foods as starter cultures. Enterococci are opportunistic pathogens, and the immune status of the host is a factor in their ability to cause disease. Despite their significance, little is known about the enterococcal virulence mechanisms that contribute to pathogenesis; however, several factors have been implicated as potential virulence determinants. By using a DNA screening approach to compare the incidence of these virulence determinants in starter, food, and medical strains, we have identified distinct trends in the occurrence of these determinants.

E. faecalis and E. faecium strains show significantly different patterns in the incidence of virulence determinants. The E. faecalis strains tested all harbor multiple virulence determinants. The incidence is greater in medical strains than in food or starter strains. All E. faecium strains are generally clear of virulence determinants, but there are notable exceptions. In this study, although all of the E. faecium strains tested lacked the cylMBA and agg determinants, one medical strain had gelE and the majority of the E. faecium strains and the E. faecium medical strains had the efaAfm and esp determinants, respectively. This contrasts with other work in which gelE and esp were found exclusively in E. faecalis strains and not in any of the E. faecium strains tested (10, 35). Regional variation may account for these differences. Alternatively, the emergence of esp⁺ E. faecium strains capable of producing this novel cell wall protein may be related to the increasing occurrence of pathogenic E. faecium strains. Further work is in progress to investigate these possibilities.

An efaA determinant is found at similar frequencies in the E. faecalis and E. faecium strains and appears in all three groups. However, differing degrees of homology among efaA genes in the different enterococcal species were demonstrated in the Southern blots. The E. faecium starter strain efaA gene appears to be less similar than the E. faecium food or medical strain efaA gene to the E. faecalis efaA gene. This may indicate significant sequence divergence among the E. faecium starter strains. E. durans strains produced variable or no bands on Southern blots using the efaA probes. Assuming that E. durans contains efaA homologues, then this could indicate further sequence divergence. So far, only the efaAfs gene has been shown to influence pathogenicity in animal models (38). The role of efaAfm has not yet been demonstrated. Significant sequence variation in the E. faecium and E. durans strains may result in functional differences in the EfaA adhesin that affect pathogenicity. Alternatively, in these strains, other virulence factors may be required in conjunction with EfaA. PCR amplification products were obtained using the efaAfs-specific primers for some E. faecium food and medical strains. The sequence analysis of these fragments should be useful in discerning the presence of the adhesin in these strains.

Several strains, including a starter strain, contained apparently silent gelE and cyl determinants. In the case of E. faecalis P21, the lack of hemolytic activity is explained by the absence of cylA, which is required for activation of cylL_L/S. For the other strains, the lack of phenotypic activity may be explained by low levels or down regulation of gene expression or an inactive gene product. Environmental factors are known to influence gene expression (12), and conditions used for phenotypic testing are different from those found in the human body. Silent genes may also become activated by temporal factors, such as conditions found in the gastro-intestinal tract, the balance of organisms in the intestinal flora, and effects of bacterial synergism, as well as the presence and persistence of large numbers of viable enterococci. Recently, a pheromone-mediated quorum-sensing system has been described. (J. Nakayama, Y. Cao, A. D. L. Akkermans, W. M. DeVos, and H. Nagasawa, Abstr. 1st Int. ASM Conf. Enterococci: Pathog. Biol. Antibiot. Resist. abstr. 21, p. 30). In this system, the gelE gene is part of the same operon as the pheromone-responsive histidine kinase genes. Expression of gelE is triggered in late exponential phase, at high cell densities, when this may be an advantage. There was a higher incidence of silent genes in the medical strains than in the food and starter strains. This would correlate with a higher potential for pathogenic effects in these strains. Thus, the presence of apparently silent genes in both food and starter strains is relevant to their safe use in foods.

Until recently, the majority of infection-derived isolates were E. faecalis strains and it is therefore regarded as the most pathogenic enterococcal species. The role of a number of virulence factors in pathogenesis, including gelE, esp, and originally efaA, was supported by the fact that these determinants were only found in E. faecalis strains. In this study, the majority of characterized virulence determinants were found in E. faecalis strains, providing further supporting evidence for their role in pathogenesis. However, the incidence of E. faecium-derived infections is increasing (B. Murray, Abstr. 1st Int. ASM Conf. Enterococci: Pathog. Biol. Antibiot. Resist. abstr. 516, p. 16). E. faecium strains containing these virulence determinants may be involved in the evolution of pathogenic E. faecium strains.

Enterococci possess highly effective gene transfer mechanisms, including conjugation and conjugative transposition. Virulence genes are known to be associated with some highly transmissible plasmids. Thus, there is a risk that a safe strain that lacks genes for known virulence factors could acquire such genes by conjugation. In the context of starter and probiotic strains, it is possible that very large numbers of viable bacteria might be consumed. This would provide a large recipient population into which transmissible plasmids carrying virulence genes might spread. The possibility of such an event was investigated using starter strains E. faecium F12 and E. faecalis F27 and F41. The potential for transfer of virulence genes into the starter strains was tested using the pheromone-inducible conjugation system of E. faecalis DS16.

Transfer of virulence determinants from E. faecalis DS16 via sex pheromone plasmid pAD1 into the E. faecalis starter strains was achieved at frequencies similar to those obtained in broth matings using plasmid-free recipient strains (40). Thus, it was possible to demonstrate that starter strains can acquire known virulence genes by the natural-conjugation gene transfer process. It will be interesting to determine whether the acquisition of pAD1 by E. faecalis strains F27 and F41 increases their virulence. Animal challenge tests to address this question are being planned.

The process of conjugation used by plasmids, such as pAD1, is highly evolved and involves a sophisticated signaling process. Potential recipient strains secrete small peptide signal molecules, known as sex pheromones. These pheromones induce genes leading to the production of aggregation substance in potential donor cells, resulting in a cell clumping phenotype. This process enhances the transfer frequency of the sex pheromone plasmid into plasmid-free recipient cells. Each sex pheromone plasmid responds specifically to a corresponding sex pheromone determinant. The pAD1 sex pheromone plasmid is found exclusively in E. faecalis strains (6) and responds specifically to the cAD1 sex pheromone. E. faecium F12 does not contain the cad sex pheromone determinant, and so, filter matings were used in an attempt to enhance the transfer frequency in the absence of induced cell clumping. Despite repeated attempts to transfer pAD1 into the E. faecium F12 starter strain via matings with E. faecalis FI9530, no transfer was achieved.

The cpd, cob, ccf, and cad sex pheromone determinants were found in all of the E. faecalis strains tested. In some of these strains, the agg gene was also present. This situation corresponds to recipient strains that have acquired a sex pheromone plasmid. The strains with the sex pheromone determinants have the potential to acquire the respective sex pheromone plasmids and, hence, the associated virulence determinants. In this study, all of the E. faecalis food strains tested had the ability to acquire several sex pheromone plasmids. For E. faecium, the lack of sex pheromone genes may reflect sequence divergence rather than the absence of gene transfer pheromones. Indeed, sex pheromone cross talk between E. faecium and E. faecalis has been established (18).

Enterococcus gene transfer and pheromone-responsive systems are especially relevant to the problems associated with acquired glycopeptide resistance. Acquired glycopeptide resistance is associated mostly with the VanA phenotype and is located on large conjugative plasmids. Some VanA conjugative plasmids are also pheromone responsive (30). Glycopeptide-resistant enterococci have become a major health problem in intensive care units in Europe and the United States; where these antibiotics are used as reserve antibiotics against multiresistant gram-positive pathogens. Heaton et al. (18) showed that a clinical isolate of E. faecium contained a conjugative plasmid highly related to pCF10 from E. faecalis. This plasmid conferred high-level vancomycin resistance that could be transferred by conjugation to plasmid-free E. faecalis recipients. The linked virulence determinants agg and cyl may be cotransferred and will be selected for by antibiotic resistance. In addition, the presence of agg in these strains may lead to increased colonization ability. Thus, sex pheromone production by E. faecalis food strains may promote the acquisition of vancomycin resistance and other linked traits from E. faecium strains and lead to increased virulence.

In conclusion, we have identified patterns in the presence of both expressed and silent virulence genes in starter, food, and medical enterococcal strains. The presence of these virulence determinants in E. faecium strains may be significant in the evolution of pathogenic E. faecium strains. In addition, the transfer of virulence determinants to starter strains via natural conjugation mechanisms has been demonstrated. These results reinforce the concern expressed elsewhere about the safety of enterococcal strains used in foods. In particular, the introduction of food products or probiotics based on the use of new enterococcal strains merits careful premarket safety evaluation.