Introduction

S. aureus (SA) is involved in food poisoning outbreaks and systemic human diseases. Antibiotic resistance in SA isolated from food of animal origins has been reported. The European Food Safety Authority (EFSA) has identified animals as a possible reservoir of SA antimicrobial-resistant (AMR) strains. Transmission through foods of AMR strains involved in human outbreaks has been reported in sporadic cases (EFSA 2008). Several mechanisms are involved in the antibiotic resistance of SA. Plasmid-encoded β-lactamase inactivates penicillin. Methicillin and other semi-synthetic penicillins have been effective against β-lactamase producer bacterial strains. However, the emergence of methicillin-resistant SA (MRSA) strains have been reported. MRSA strains harbor the mecA gene, which encodes a low-affinity penicillin-binding protein (PBP2a). Nosocomial infections caused by MRSA can sometimes lead to death (Moreillon and Que 2004). Macrolide resistance in SA can be caused by a target modification of a 23S rRNA methylase, encoded by the erm genes. SA can also be resistant to vancomycin, which has been used successfully to treat MRSA strains. Vancomycin resistance is encoded by several genes and appears to have originated from mutations under antibiotic-selective pressure. Antibiotic resistance is frequently acquired by exchange of mobile DNA elements during food processing between pathogenic and non-pathogenic microorganisms. In the dairy sheep industry, SA AMR strains can be recovered from raw milk, secretions from animals with mastitis, and dairy products. The aim of the present study was to assess a protocol to characterize SA isolates and their resistance to antibiotics using phenotypic and molecular techniques. In SA strains isolated from raw sheep’s milk cheese collected at early stages of the cheese making process, minimum inhibitory concentrations (MICs) and susceptibility to antibiotics used in human and veterinary practice were determined. The presence of the ermA, blaZ, mecA, and vanA and B genes was also investigated. A population study of the isolated microorganisms was carried out by pulsed-field gel electrophoresis (PFGE).

Material and methods

SA strains were recovered from cheese made from raw sheep’s milk collected 6 hours after molding, a phase generally associated with the highest levels of SA. Twelve cheese wheels were collected from six household plants (A-F) located on the island of Sardinia (Italy). Each sample was representative of a different production batch, two from each plant. Samples were analyzed according to UNI EN ISO 6888–2:2004. Thirty-six strains were isolated from typical colonies and identified according to Scarano et al. (2005). For each strain, the MIC was tested by the broth microdilution method (CLSI 2006a, b) for the following antibiotics: ampicillin (AM), cefalotin (CF), cefoperazone (CFP), cloxacillin (OB), erythromycin (E), gentamycin (GM), kanamycin (K), novobiocin (NV), ofloxacin (OFX), oxacillin (OX), penicillin (P), rifampicin (RA), streptomicin (S), tetracycline (TE), tobramycin (TM), and vancomycin (VA). Twelve serial dilutions of each antibiotic were made in cation-adjusted Mueller Hinton Broth (CAMHB; Oxoid LTD Basingstoke, UK) containing NaCl (2%), with the final concentrations of antibiotics ranging from 0.06 to 128 μg/mL (0.006-12.8 μg/mL for RA). The MIC mode, range, MIC50, and MIC90 of the antimicrobial agents were calculated, and antibiotic susceptibility was defined comparing the MICs with reference values (CLSI 2006b). After DNA extraction, multiplex PCR was used (Rosato et al. 2003; Vancraeynest et al. 2004) to detect gyrA encoding the DNA gyrase, agr (I–IV), an accessory gene regulator, blaZ encoding a penicillinase, ermA, a gene encoding resistance to macrolide E, mecA encoding PBP2a, and vanA-B, genes encoding VA resistance. The templates were loaded onto an agarose gel, and electrophoresis was performed. The population study was conducted by PFGE on SA strains according to Prevost et al. (1992) after SmaI restriction digest (Invitrogen, CA, USA) of bacterial DNA. Correlations between the isolates were determined by an unweighted pair grouping by mathematical averaging (UPGMA) model with a similarity greater or equal to 80% of the Pearson correlation coefficient (Gel Compar II, Applied Math).

Results

Coagulase-positive staphylococci (CPS) in cheese samples ranged from 3.9 to 5.8 Log10 CFU/mL. Thirty-six isolates were identified as SA based on their biochemical and metabolic profiles and on the presence of the gyrA gene. All isolates produced TDNase and were coagulase positive. The isolates were characterized by PFGE clusters (P1–P10) and by the presence of agr and antibiotic resistance genes (Table 1). None of the tested isolates showed resistance to CF, CFP, E, GM, K, NV, OFX, RA, S, and TM. Only two isolates showed intermediate levels of resistance to K (5.6%), and one to OFX (2.8%). Four SA strains could not be typed by PFGE. The SA isolates exhibited resistance to AM (36.1%), P (33.3%), TE (11.1%), and OB (2.8%), and the MIC90 for these antibiotics was, respectively, 16.0, 32.0, 64.0, and 32.0 μg/mL. MIC90 of the isolates was below the sensitivity range of 13 out of the 16 tested antibiotics. Regarding AM and P resistance, the results of broth microdilution and molecular analysis (i.e., blaZ gene detection) were in agreement, showing a concordance of 94.4%. The only disagreements between results obtained with the two methods applied to an SA blaZ+ isolate that showed resistance only to AM and one blaZ- isolate with resistance to AM, P, and OB.

Table 1 Genotypic characterization and antibiotic sensitivity
Full size table

Discussion

The SA strains isolated during cheese processing were resistant only to AM, P, OB, and TE, which are among the most commonly used antibiotics on dairy sheep farms. In SA isolated from sheep, the agrIII allele is the most frequently observed (De Santis et al. 2005). Fifty percent of the SA strains isolated from four different dairies belonged to the agrIII group, while 27.8% belonged to the agrII group and were isolated from three dairies (C, D, and F). The strains belonging to the agrI group (dairies C and F) and to the agrIV group (dairy B) showed low prevalence. A dendrogram showed a specific distribution of isolates from each dairy. Only isolates included in the P1 cluster were observed at two different dairies (A and B). Each of the isolates of the P3, P4, and P10 clusters was recovered at only one dairy (D, E, or F, respectively), but in both production batches. Our results show a good correspondence between PFGE clusters and antibiotic resistance profiles of the isolates. At two dairies (A and D), all the isolates were susceptible to the tested antibiotics. Four strains belonging to the agrI group (dairy C) were untypeable by PFGE and exhibited resistance to TE. In agreement with previous studies on SA strains isolated from raw sheep’s milk (De Santis et al. 2005), the mecA gene was not detected in any of the tested strains. The recovery of MRSA strains from half udder milk samples is sporadic and commonly of human origin (De Santis et al. 2005). The agreement between the antibiotic resistance profiles and detection of the corresponding genes was 100% for E, OX, and VA, and 94.4% for AM and P. The determination of the antibiotic resistance profile allowed an understanding of the source of contamination and a trace back to each step of the cheese making process. This study permitted the identification of the spreading of strains with the same profile and their persistence within the cheese plant.