760
I
Journal of Food Protection, Vol. 71, No. 4, 2008, Pages 760-769
Antimicrobial Resistance and Virulence of Enterococcijs faecalis Isolated from Retail Foodt LORI L. MCGOWAN-SPICER, 12 PAULA J. FEDORKA-CRAY,' JONATHAN G. FRYE,' RICHARD J. MEINERSMANN,' JOHN B. BARRETT,' AND CHARLENE R. JACKSON'* Bacterial Epidemiology and Antimicrobial Resistance Research Unit, U.S. Department of Agriculture, Agricultural Research Service, Russell Research Center, Athens, Georgia 30605; and 2 Department of Biology, Gottwald Science Center, University of Richmond, Richmond, Virginia 23173, USA
MS 07-251: Received 10 May 2007/Accepted 15 December 2007 ABSTRACT Although enterococci are considered opportunistic nosocomial pathogens, their contribution to foodborne illnesses via dissemination through retail food remains undefined. In this study, prevalence and association of antimicrobial resistance and virulence factors of 80 Enterococcus faecalis isolates from retail food items were investigated. The highest rates of resistance
were observed for lincomycin (73 of 80 isolates, 91%) and bacitracin (57 of 80 isolates, 71%), and lower rates of resistance (:!^40%) were found for chloramphenicol, ciprofioxacin, erythromycin, flavomycin, gentamicin, kanamycin, nitrofurantoin, penicillin, and tylosin. Overall resistance to antimicrobials was low for most isolates tested. Of the virulence factors tested, the majority of isolates were positive for ccf (78 of 80 isolates, 98%), efaAfs (77 of 80, 96%), and cpd (74 of 80, 93%). Isolates also commonly contained cob (72 of 80 isolates, 90%) and gelE (68 of 80, 85%). Very few isolates contained cy1MBA (12 of 80 isolates [l5%] for cyiM and 9 of 80 isolates [11%] for both cyiB and cylA) and efaAfln (2 of 80 isolates, 3%). Positive statistical associations (significance level of 0.05) were found .between agg and tetracycline resistance, cyiM and erythromycin resistance, and gelE and efaAfs and lincomycin resistance. The presence of the cyiB and cylA alleles also was positively correlated with bacitracin and tetracycline resistance. Negative correlations were observed between many of the virulence attributes and resistance to ciprofloxacin, erythromycin, flavomycin, gentamicin, kanamycin, and tylosin. These data suggest that both positive and negative associations exist between antimicrobial resistance genes and virulence factors in E. faecalis isolates from foods commonly purchased from grocery stores.
Enterococci have been found in a wide range of habitats ranging from the intestinal tract of mammals to soil, water, plants, insects, and food items (10, 24, 32, 40). Although their role as opportunistic nosocomial pathogens has been well documented (8, 11, 14, 16, 18, 20, 26), their contribution to foodborne illnesses is not well defined. Because they harbor specific biochemical traits that are essential in manufacturing fermented milk products such as cheeses, these bacteria are desirable in certain foods, but their presence also can indicate spoilage for fermented meats or unsanitary conditions in other food industries (4, 10). Enterococci also have been implicated as a cause of food intoxication, which results in symptoms such as vomiting and headaches due to the production of biogenic amines in fermented foods (7, 10, 36). Problems associated with enterococci on foodstuffs are exacerbated by their potential for harboring antimicrobial resistance genes and putative virulence determinants, and these altered enterococci may be passed to humans. Enterococci also can transfer antimicrobial resistance genes and some virulence factors to other members of the intestinal microflora and to more *
Author for correspondence. Tel: 706-546-3604; Fax: 706-546-3616; E-mail:
[email protected]. t The mention of trade names or commercial products in this manuscript is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture.
pathogenic bacteria, increasing their threat as nosocomial pathogens (2, 9, 10, 19, 26, 39).
Acquired antimicrobial resistance in enterococci is cause for concern in treatment of enterococcal infections. Resistance to a number of antimicrobials, including gentamicin, quinupristin-dalfopristin, tetracycline, and vancomycin, has been described in enterococci isolated from clinical infections and from food (4, 12, 17, 18), although none of the vancomycin-resistant enterococci from food have been identified from U.S. food. In addition to antimicrobial resistance, potential virulence factors such as toxins, adhesins, surface proteins, cytolysin, and pheromones also have been isolated from enterococcj from different sources (3, 4, 10, 25, 32). Even though clinical enterococcal isolates contain more virulence determinants than do enterococci recovered from food, increased vigilance for antimicrobial resistance and virulence factors in enterococcj from food is necessary because of the potential for dissemination to the consumer (3). To date, the hypothesis that increased antimicrobial resistance is directly associated with increased virulence has not been overwhelmingly supported, and the contribution of potential virulence factors to the pathogenesis of enterococcal infections has not been established (25). While the prevalence of antimicrobial resistance or virulence in enterococci has been investigated, few studies have characterized the antimicrobial and virulence gene
J. Food Prot., Vol. 71, No. 4
ANTIMICROBIAL RESISTANCE AND VIRULENCE IN ENTEROCOCCI 761
content of enterococci isolated from food. The contribution of EnterocoCCUS faecalis from food to cases of human infection is unknown. Although nosocomial infections by Enterococcus faecium are increasing, primarily because of vancomycin-resistant E. faecium, E. faecalis accounts for the majority of nosocomial enterococcal infections (5, 16, 21, 25). Because antimicrobial resistance often is considered a virulence trait in enterococci, increasing rates of antimicrobial resistance in E. faecalis and the presence of putative virulence determinants and linkage of the two traits could possibly predict overall increasing virulence in this species. In this study, the prevalence of antimicrobial resistance and virulence determinants in E. faecalis isolates from retail food items was determined to evaluate the possible role of retail food in dissemination of pathogenic enterococci. MATERIALS AND METHODS Sample collection, isolation, and identification of enterococci. E. faecalis isolates used in this study were described previously (23). During 2000 and 2001, fresh fruits and vegetables and prepackaged ground and whole meat were purchased from six retail food stores in the Athens, Ga., area. Approximately 2 lb (0.9 kg) of each food product was purchased, and each food was collected in a way to ensure that cross-contamination from lab personnel to the product did not occur. Foods were bagged separately and kept refrigerated until processed. Fruits and vegetables were placed in a sterile bag to which 50 ml of phosphate-buffered saline (lx PBS) was added, and 100 ml of PBS was added to each ground or whole meat sample bag. Bags were vigorously shaken for 2 min to remove bacteria from the surface, and I ml of each rinsate was then transferred to 9 ml of Enterococcosel broth (Becton Dickinson, Sparks, Md.) and incubated for 24 h at 37°C. A swab was used to transfer broth from positive cultures to Enterococcosel agar (Becton Dickinson) for isolation of enterococci. Plates were incubated overnight at 37°C. One presumptive positive colony was inoculated onto blood agar, and the resulting clones were identified as E. faecalis by multiplex PCR assay as previously described (15). Antimicrobial susceptibility. MICs for enterococci were determined by broth microdilution using the Sensititre semiautomated antimicrobial susceptibility system (Trek Diagnostic Systems, Inc., Cleveland, Ohio) and the Sensititre Gram-positive custom plate CMV 1AGPF (Trek Diagnostic) according to the manufacturer's directions. Results were interpreted according to CLSI guidelines when defined (27). Breakpoints for bacitracin, flavomycin, gentamicin, kanamycin, lincomycin, salinomycin, streptomycin, and tylosin were those defined by the U.S. Department of Agriculture Agricultural Research Service: ^!128 p.g/ml bacitracin, ^!32 jig/ml chloramphenicol, ^!4 jig/ml ciprofloxacin, ^:-8 i.g/ml erythromycin, ^l6 i.g/ml flavomycin, ^!500 p.g/ml gentamicin, ^!500 .g/ml kanamycin, ^!4 p.g/ml lincomycin, ^!8 jig/ ml linezolid, ^!128 Vg/ml nitrofurantoin, ^-16 p.gIml penicillin, ^! 16 p.g/ml salinomycin, ^: 1,000 p.g/ml streptomycin, ^!4 g/ml quinupristin-dalfopristin, ^!16 jig/ml tetracycline, ^!20 jig/ml tylosin, and ^!32jig/ml vancomycin. E. faecalis ATCC 29212 was the positive control for determination of MICs. Microarray and PCR of virulence factors. A microarray was constructed and used for the detection of putative virulence factors using the methods described by Frye et al. (6). Sequences for 12 genes chosen from the published literature were download-
ed from GenBank (National Center for Biotechnology Information, Bethesda, Md.), and 70-mer oligonucleotide probes were designed using OligoWiz 2. 1.0 (Table 1) (28, 38). Microarrays were printed in triplicate with negative and positive controls as previously described. Hybridizations, washing, scanning, image processing, scoring, and data analysis were done as previously described (6). Hybridization signals greater than 2 standard deviations above the signals for the negative controls for all three probes were scored as detected in the isolates hybridized. To confirm results of the microarray, amplification of agg, gelE, cyIMBA, esp, efaAfs, efaAfm, cpd, cob, and ccf was performed as previously described (3). Results from the two methods were compared, and discrepancies were resolved by consensus after repeating the PCR analysis. Results for cad were determined using microarray analysis only because PCR primers were not available. Suitable microarray probes were not designed for cob and cpd, and only PCR results were available for those two genes. PFGE. Pulsed-field gel electrophoresis (PFGE) was performed as previously described (37). Cells from a 5-ml overnight culture were pelleted, embedded in agarose plugs, and lysed. Plugs were digested overnight with 20 U of SmaI (Roche, Indi-
anapolis, Ind.), and digested DNA was separated on a 1.2% SeaKem agarose gel using a CHEF-DRII pulsed-field electrophoresis system (Bio-Rad, Hercules, Calif.). Electrophoresis was carried out at 6 V for 21 h with a ramped pulse time of 5 to 30 s in 0.5X Tris-borate-EDTA buffer (14°C). Data analysis. Cluster analysis of PFGE results was conducted using the Bionumerics software program (Applied Maths, Sint-Martens-Latem, Belgium) with the Dice coefficient and the
unweighted pair group method. Optimization settings for the dendrogram were 3% with a band tolerance of 2.1%. Test of linkage disequilibrium (nonrandom association of alleles at different loci) for all pairs of antimicrobial resistance and virulence alleles for all pairs of loci was determined with a Markov chain (34, 35). Significant nonrandom association between alleles was determined using Fisher's exact test of contingency tables. RESULTS Antimicrobial resistance profiles of E. faecalis.
Twenty-three E. faecalis isolates were obtained from fruit and vegetable samples, and 57 were obtained from meat samples. From the fruit and vegetables, the majority (13 of 23, 56.5%) of the E. faecalis isolates were from potatoes, and more E. faecalis isolates were obtained from turkey (17 of 57, 29.8%) than from any other meat product (Table 2). Overall, the highest rates of resistance (>50%) were
observed for lincomycin (73 of 80 isolates, 91.3%), followed by bacitracin (57 of 80, 71.3%) (Table 2). High rates of tetracycline resistance also were observed in E. faecalis isolates from retail meat (37 of 57 isolates, 64.9%) but not from produce (3 of 23 isolates, 13%). Low rates of resis-
tance (
