Received 25 April 2006/Returned for modification 16 July 2006/Accepted 17 September 2006

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, Dec. 2006, p Vol. 44, No /06/$ doi: /jcm Copyright 2006, American Society for Microbiology. All Rights Reserved. Presence of New meca and mph(c) Variants Conferring Antibiotic Resistance in Staphylococcus spp. Isolated from the Skin of Horses before and after Clinic Admission Christina Schnellmann, 1 Vinzenz Gerber, 2 Alexandra Rossano, 1 Valentine Jaquier, 1 Yann Panchaud, 2 Marcus G. Doherr, 3 Andreas Thomann, 1 Reto Straub, 2 and Vincent Perreten 1 * Institute of Veterinary Bacteriology 1 and Equine Clinic, 2 Department of Clinical Veterinary Medicine, 3 Vetsuisse Faculty, University of Berne, CH-3001 Bern, Switzerland Received 25 April 2006/Returned for modification 16 July 2006/Accepted 17 September 2006 Because of the frequency of multiple antibiotic resistance, Staphylococcus species often represent a challenge in incisional infections of horses undergoing colic surgery. To investigate the evolution of antibiotic resistance patterns before and after preventative peri- and postoperative penicillin treatment, staphylococci were isolated from skin and wound samples at different times during hospitalization. Most staphylococci were normal skin commensals and belonged to the common coagulase-negative group. In some cases they turned out to be opportunistic pathogens present in wound infections. MICs were determined for 12 antibiotics, and antibiotic resistance genes were detected by microarray. At hospital admission, horses harbored staphylococci that were susceptible to antibiotics or resistant to one group of drugs, mainly due to the presence of new variants of the methicillin and macrolide resistance genes meca and mph(c), respectively. After 3 days, the percentage of Staphylococcus isolates displaying antibiotic resistance, as well as the number of resistance genes per isolate, increased moderately in hospitalized horses without surgery or penicillin treatment but dramatically in hospitalized horses after colic surgery as well as penicillin treatment. Staphylococcus species displaying multiple resistance were found to harbor mainly genes conferring resistance to -lactams (meca and blaz), aminoglycosides [str and aac(6 )-Ie aph(2 )-Ia], and trimethoprim [dfr(a) and dfr(d)]. Additional genes conferring resistance to macrolides [mph(c), erm(c), and erm(b)], tetracycline [tet(k) and tet(m)], chloramphenicol [cat(pc221) and cat(pc223)], and streptothricin (sat4) appeared in several strains. Hospitalization and preventive penicillin use were shown to act as selection agents for multidrug-resistant commensal staphylococcal flora. Postoperative incisional infection is a major complication after colic surgery in horses and can lead to intensive postoperative care, prolonged hospitalization, and increased costs (15). All equine patients undergoing colic surgery at the equine clinic of the Vetsuisse Faculty of the University of Berne receive preventative treatment of 30,000 IU penicillin G per kg body weight once preoperatively and then daily for at least 3 days postoperatively. Despite antimicrobial prevention, the rates of postoperative wound infection after colic surgery reached 58% in 1992 (17), 42% in 1995/1996 (22), 62% in 1998 to 2000 (32), and 38% in 2004/2005 (unpublished data). In other equine clinics in Australia, Canada, the United Kingdom, and the United States, postoperative wound infections are less frequent, with rates varying from 16% to 38% (23, 27, 34, 39, 43). The routine use of penicillin to prevent incisional infections may lead to selection for specific penicillin-resistant gram-positive flora, including commensals and pathogens such as methicillin-resistant Staphylococcus aureus (MRSA). MRSA strains are characterized by the presence of a meca gene encoding low-affinity penicillin binding protein (PBP2 ), which mediates resistance to all classes of -lactam antibiotics (20, 52). MRSA strains are also often resistant to other commonly used antibiotics, making them difficult to treat. As a result, they represent one of the major causes of nosocomial infections in humans (29). Over the last decade, MRSA strains have emerged in community-acquired infections and caused serious morbidity and even death in previously healthy children and adults (16, 28). Animals were reported to act as a reservoir for MRSA strains (2, 55), which have been isolated from the milk of mastitis-afflicted cows (11, 12), household pets (5, 13, 33, 35, 37, 49, 51, 54), food-producing animals (31), and horses (21, 37, 46, 47, 56, 57). The presence and origin of multidrug-resistant staphylococci in hospitalized animals in Switzerland and the source of contamination remain to be determined. Resistant bacteria can be acquired during hospitalization (nosocomial infection) or may already be present in the normal flora of the patient before hospital admission. To test these hypotheses, the evolution of the staphylococcal population and its resistance profile were followed in horses from the time of admission to the equine clinic of the University of Berne to the development of postoperative wound infection. * Corresponding author. Mailing address: Institute of Veterinary Bacteriology, University of Berne, Länggass-Strasse 122, Postfach, CH-3001 Bern, Switzerland. Phone: Fax: vincent.perreten@vbi.unibe.ch. Published ahead of print on 27 September MATERIALS AND METHODS Sample collection. Between August 2004 and May 2005, 12 of 37 horses that underwent colic surgery at the equine clinic of the Vetsuisse Faculty at the University of Berne were sampled for bacterial analysis. The samples were taken from different parts of the horse body at different times during hospitalization. At 4444

2 VOL. 44, 2006 ANTIBIOTIC RESISTANCE IN STAPHYLOCOCCI FROM HORSES 4445 TABLE 1. Oligonucleotides used for PCR, RT-PCR, and sequence analysis Gene Primer name Sequence (5 33 ) aac(6 )-Ie aph(2 )-Ia aac6-aph2-f CAGAGCCTTGGGAAGATGAAG 53 aac6-aph2-r CCTCGTGTAATTCATGTTCTGGC 53 Primer design reference or source blaz blaz-f CAGTTCACATGCCAAAGAG This study blaz-r TACACTCTTGGCGGTTTC This study cat(pc194) catpc194-f CGACTTTTAGTATAACCACAGA This study catpc194-r GCCAGTCATTAGGCCTAT This study cat(pc221) catpc221-f ATTTATGCAATTATGGAAGTTG This study catpc221-r TGAAGCATGGTAACCATCAC This study cat(pc223) catpc223-f1 GAATCAAATGCTAGTTTTAACTC This study catpc223-r ACATGGTAACCATCACATAC This study dfr(a) dfra-f CCTTGGCACTTACCAAATG This study dfra-r CTGAAGATTCGACTTCCC This study dfr(d) dfrd-f TTCTTTAATTGTTGCGATGG This study dfrd-r TTAACGAATTCTCTCATATATATG This study meca meca-f ATGAAAAAGATAAAAATTGTTCCAC This study meca-r TTATTCATCTATATCGTATTTTTTATTAC This study mph(c) mphc-f1 ATGACTCGACATAATGAAAT This study mphc-r1 CTACTCTTTCATACCTAACTC This study mphc-r4 CTATTCCGTGTTCATCTTCTCC This study sat4 sat4-f CGATAAACCCAGCGAACC This study sat4-r ATAACATAGTATCGACGG This study str str-ps194-f TATTGCTCTCGAGGGTTC This study str-ps194-r CTTTCTATATCCATTCATCTC This study tet(k) tet(k)-1 TTAGGTGAAGGGTTAGGTCC This study tet(k)-2 GCAAACTCATTCCAGAAGCA This study tet(m) tet1 GCTCAYGTTGAYGCAGGAA 3 tet2 AGGATTTGGCGGSACTTCKA 3 each designated sampling site, a field of approximately 2 by 2 cm was shaved using a single-use sterile razor (P. J. Dahlhausen & Co. GmbH, Köln, Germany). A superficial scrape was then taken with a single-use 7-mm-diameter sterile curette (Stiefel Laboratorium GmbH, Offenbach am Main, Germany) and stored in glycerol-salt solution (25% glycerol, 0.8% NaCl) at 20 C until analyzed. The first sample (sample I) was taken from the abdomen near the incision area upon clinic admission before the horse was stabled and before penicillin injection and abdomen disinfection, which was performed as previously described (32). Two second samples were taken at least 72 h after surgery and at the beginning of penicillin administration. One of the two (sample IIa) was taken approximately 3 cm from the incision, and the other (sample IIb) was taken from a position cranial to the surgical incision, where the skin had not been disinfected prior to surgery. In the event of wound secretion, some liquid was collected with a transport swab (Oxoid Ltd., Basingstoke, England) and analyzed within 48 h (sample III). A group of seven horses that came to the clinic with colic symptoms that could be treated without surgery and, therefore, without penicillin was analyzed as a control group. For this group, the first sample (sample I) was taken at the time of admission to the clinic and the second sample (sample IIc) was taken at least 72 h after arrival, both from the same areas and using the same techniques employed for the group of horses undergoing surgery. Isolation and identification of staphylococci. A 10-fold serial dilution of skin samples (samples I and II) or wound secretion swabs (sample III) was spread on tryptone soy agar containing 5% sheep blood (Oxoid Ltd., Basingstoke, England) and incubated for 24 to 48 h at 37 C. In order to avoid selecting for identical isolates (clones), one representative colony was selected from each distinct morphological group (shape, size, color, and shine) per sample and subcultured on tryptone soy agar. Staphylococci were identified by hemolysis, catalase activity, Gram staining, and API ID32 STAPH galleries (biomérieux, Marcy l Etoile, France). The API ID32 STAPH galleries were automatically read using a miniapi instrument and associated software (biomérieux). When the probability of species identification with API materials was low, strains were identified by sequencing species-specific 16S rrna, as previously described (30). Antimicrobial susceptibility testing. MICs were determined in Mueller- Hinton broth by use of custom Sensititre susceptibility plates (Trek Diagnostics Systems, East Grinstead, England, and MCS Diagnostics BV, Swalmen, The Netherlands) according to CLSI guidelines (8, 9) plus an informational supplement (10). Detection and characterization of antibiotic resistance genes. Genomic DNA was isolated using a high pure template preparation kit (Roche Diagnostics, Basel, Switzerland) in accordance with the manufacturer s recommendations. Antibiotic resistance genes were detected using a microarray capable of detecting more than 90 antibiotic resistance genes known to be present in gram-positive bacteria (38). Some antibiotic resistance genes were verified by PCR and sequence analysis using the primers listed in Table 1. The presence of the bifunctional gene aac(6 )-Ie aph(2 )-Ia was verified by PCR as previously described (53). The complete open reading frames of the meca and mph(c) genes were amplified using the primer pairs meca-f and meca-r and mphc-f1 and mphc- R1, respectively. PCR was performed using Taq DNA polymerase according to the supplier s directions (Roche Diagnostics, Basel, Switzerland), and an annealing temperature of 50 C was used, except with meca and mph(c), for which an anneal-

3 4446 SCHNELLMANN ET AL. J. CLIN. MICROBIOL. ing temperature of 45 C was used. The genes were sequenced on an ABI Prism 3100 genetic analyzer (Applied Biosystems, Foster City, CA) with drhodamine-labeled terminators and edited using Sequencer software (Gene Codes, Ann Arbor, MI). The DNA sequences were aligned using ClustalW ( and compared to GenBank sequences using BLAST ( Measurement of antibiotic resistance gene expression. In addition to gene detection by DNA microarray, all isolates were tested for the production of PBP2, encoded by meca, as well as -lactamase, encoded by blaz. PBP2 was detected with an Oxoid penicillin binding protein (PBP2 ) latex agglutination test (Oxoid Ltd., Basingstoke, England) in accordance with the supplier s instructions. The production of -lactamase was tested on nitrocefin dry slides (Becton Dickinson Microbiology Systems, Cockeysville, MD) using colonies grown on Mueller-Hinton agar for 18 h at 37 C with 0.05 g penicillin per ml to induce -lactamase production. S. aureus strains MIL1971 and D1003 (laboratory collection) were used as -lactamase- and meca-positive controls, respectively. Following CLSI recommendations, isolates that carried the meca gene or produced PBP2 were reported as oxacillin resistant and isolates that were meca negative or did not produce PBP2 were considered oxacillin resistant only if the MIC was 4 g/ml. Isolates were reported resistant to penicillin if they were resistant to oxacillin and expressed meca, even though they had penicillin MICs of 0.12 g/ml (9). For strains which contained the gentamicin resistance gene aac(6 )-Ie aph(2 )-Ia or the erythromycin resistance gene mph(c) but whose resistance phenotype was below the CLSI resistance breakpoint, gene expression was verified using reverse transcription-pcr (RT-PCR). RNA was isolated using a QIAGEN RNeasy mini kit (QIAGEN, Inc., Valencia, CA). RNA quality was verified on a 1.2% denaturing formaldehyde agarose gel (45), and quantity was measured using a Nanodrop ND-1000 spectrophotometer (Nanodrop Technologies, Inc., Wilmington, DE). DNA residues were digested using RQ1 RNasefree DNase (Promega, Madison, WI). Single-stranded RNA was reverse transcribed into cdna at 37 C for 1 h using murine leukemia virus reverse transcriptase (Roche Diagnostics, Basel, Switzerland) and primers mphc-f1, specific to mph(c), and aac6-aph2-f, specific to aac(6 )-Ie aph(2 )-Ia. Obtained cdna was used for second-round PCR amplification using primers mphc-f1 and mphc-r4 for mph(c) and aac6-aph2-f and aac6-aph2-r for aac(6 )-Ie aph(2 )-Ia and an annealing temperature of 54 C. Controls without reverse transcriptase were included to ensure that samples did not contain genomic DNA. PFGE. Pulsed-field gel electrophoresis (PFGE) analysis of SmaI-digested chromosomal DNA was performed on the basis of PulseNet s Listeria monocytogenes PFGE protocol with some modifications. Cells were resuspended in 10 mm Tris, 100 mm EDTA, ph 8.0 (TE buffer), to obtain an optical density at 600 nm of 15. Gel plugs were prepared in 1.5% SeakemGold agarose (Cambrex Bio Science Rockland, Inc., Rockland, ME) with 0.5 Tris-borate-EDTA buffer (TBE buffer) (1 TBE buffer is 89 mm Tris, 89 mm boric acid, 2.5 mm EDTA, ph 8.0) and incubated for 5 h at 37 C in 1.2 ml TE buffer plus 10 U of lysostaphin. They were then incubated overnight at 50 C in 1.5 ml 0.5 M EDTA, 1% N-lauroyl sarcosin, 2 mg/ml proteinase K, ph 8.0, and washed five times for 30 min at room temperature in TE buffer. Plug slices (2 by 5 by 10 mm) were digested with 50 U of SmaI (Roche Diagnostics, Basel, Switzerland) overnight at 25 C. DNA fragments were separated in a 1% agarose gel slab using a contourclamped homogeneous electric field DRIII device (Bio-Rad Laboratories, Inc., Richmond, Calif.). A lambda ladder PFG marker (New England BioLabs, Beverly, Mass.) was used as a reference marker. Electrophoresis was performed in 0.5 TBE buffer containing 9 g/ml thiourea (61) with a ramped pulse time of 2 to 30 s at 6 V/cm for 16 h at 12 C. The gel was stained for 20 min in water containing 1 g/ml ethidium bromide and destained twice for 15 min in water at room temperature. The digital PFGE pattern images were analyzed with Bio- Numerics software (Applied Maths, Kortrijk, Belgium). Band-based dendrograms were constructed by use of the unweighted pair group method using the arithmetic averages algorithm, based on Dice similarity coefficients. The PFGE profiles were defined on the basis of DNA banding patterns in accordance with the criteria of Tenover et al. (50) for bacterial strain typing. Statistical analysis. Data were described as frequencies and proportions of antibiotic resistance by strain and antibiotic. The proportions of resistant strains for each antibiotic among the three sampling sites were compared using simple logistic regression (with resistance y/n designated as the binary outcome). Odds ratios for resistance with sample I (no penicillin and no stay at the clinic) as the baseline category were derived and are presented with 95% confidence limits (95% CI) and P values, whenever statistically significant (P 0.05). RESULTS Molecular diversity of the strains. Out of 12 horses undergoing colic surgery and seven control horses, 121 purified strains of randomly chosen colonies with different morphologies were identified. Among them, 75 (62%) were identified as Staphylococcus species: 23 S. xylosus strains from 14 horses, 15 S. capitis strains from 8 horses, 14 S. equorum strains from 8 horses, 9 S. vitulinus strains from 8 horses, 6 S. kloosii strains from 3 horses, 3 S. sciuri strains from 3 horses, and 3 S. aureus strains from 3 horses, as well as 1 S. epidermidis strain and 1 S. cohnii subsp. cohnii strain. S. xylosus, S. capitis, S. vitulinus, S. kloosii, and S. equorum strains were all found before and after hospitalization in similar proportions. New species (S. sciuri, S. aureus, S. epidermidis, and S. cohnii subsp. cohnii) were found to emerge after hospitalization (Table 2). The genetic relationship among 71 isolates (68 coagulase-negative staphylococci [CoNS] and three S. aureus isolates) was determined by PFGE (data not shown). Only the two S. xylosus strains CSRE21 (horse no. 13) and CSGO26 (horse no. 14), both isolated upon arrival to the clinic (sample I), displayed identical PFGE patterns and were, therefore, considered indistinguishable. These two horses came from different geographic regions and arrived at the clinic on the same day. All other Staphylococcus species showed different PFGE band patterns. Different profiles were also found for strains isolated from the same horse and belonging to the same species (S. capitis strains CSLA5 and CSLA6 from horse no. 6 [sample III], S. xylosus strains CSKR32 and CSKR33 from horse no. 16 [sample I], and S. xylosus CSKR35 and CSKR36 also from horse no. 16 [sample IIc]), revealing that diverse strains of the same species coexist on the same animal. Overall analysis of antibiotic resistance. Out of 75 Staphylococcus isolates analyzed in this study, 18 (24%) were susceptible to all 12 antibiotics tested. The others displayed phenotypic resistance to antibiotics such as -lactams, combination -lactam -lactamase inhibitors, aminoglycosides, tetracycline, chloramphenicol, macrolides, lincosamides, and/or streptogramins. No strains were resistant to enrofloxacin or vancomycin. MICs are shown in Table 2. The presence and type of antibiotic resistance genes were determined by microarray. In general, the genotype corresponded to the phenotype of the strains. Of note, the MICs of erythromycin and gentamicin for strains harboring the respective resistance genes mph(c) and aac(6 )-Ie aph(2 )-Ia often displayed intermediate resistance, with MICs below the determining resistance breakpoint (Table 2), although these genes were found to be expressed as demonstrated by reverse transcription-pcr (see below). The presence of antibiotic resistance genes in strains with an MIC below the CLSI resistance breakpoint may lead to false interpretation of susceptibility profiles. This indicates the importance of resistance gene detection in addition to phenotypic determination (28). Strains displaying decreased susceptibility to erythromycin were found to contain mainly the macrolide phosphotransferase gene mph(c), with the exception of seven strains (S. equorum CSCA4, CSRE23, and CSJAE49 and S. xylosus CSCA13, CSFA7, CSNO46, and CSKR33). These strains were resistant to the macrolide erythromycin but not to the lincosamide clindamycin and did not contain any known macrolide resistance

4 VOL. 44, 2006 ANTIBIOTIC RESISTANCE IN STAPHYLOCOCCI FROM HORSES 4447 TABLE 2. Antibiotic resistance profiles of 72 CoNS strains and three S. aureus strains a Horse no. Strain Species Sample b MIC ( g/ml) for antibiotic c Antibiotic resistance gene(s) detected d AMC CHL CLI ENR ERY GEN OXA PEN Q-D TET VAN XNL 1 CSCA1 S. capitis I 2/ CSCA4 S. equorum I 2/ * CSCA7 S. kloosii I 2/ * meca CSCA13 S. xylosus I 2/ * CSHO1 S. equorum I 2/ * 0.25* mph(c) CSHO3 S. vitulinus I 2/ * meca CSHO5 S. xylosus I 2/ CSFA7 S. xylosus I 2/ * CSFO9 S. capitis I 2/ * * tet(k) 7 CSWE7 S. xylosus I 2/ * 0.25* CSBO8 S. vitulinus I 2/ * meca CSBO9 S. equorum I 2/ mph(c) 9 CSFAL12 S. xylosus I 2/ CSFAL13 S. vitulinus I 2/ CSKO17 S. vitulinus I 2/ * aac(6 )-Ie aph(2 )-Ia CSKO18 S. equorum I 2/ CSEV19 S. equorum I 2/ * mph(c) 13 CSRE21 S. xylosus I 2/ CSGO25 S. capitis I 2/ * 0.5* CSGO26 S. xylosus I 2/ CSTA28 S. capitis I 2/ * meca CSTA29 S. xylosus I 2/ CSKR32 S. xylosus I 2/ (hemol.) e CSKR33 S. xylosus I 2/ * 1 1 1* (white) CSKR34 S. kloosii I 2/ CSJAE47 S. capitis I 2/ * meca, aac(6 )-Ie aph(2 )-Ia 8 CSBO10 S. capitis IIc 2/ * * 1 2 meca, aac(6 )-Ie aph(2 )-Ia, str, tet(k) CSBO11 S. xylosus IIc 2/ * blaz, aac(6 )-Ie aph(2 )-Ia, ant(4 )-Ia, dfr(d), mph(c) 10 CSWY14 S. cohnii IIc 2/ * 0.5* subsp. cohnii CSWY15 S. vitulinus IIc 2/ * 1* * blaz, meca, aac(6 )-Ie aph(2 )-Ia, str, tet(k) CSWY16 S. capitis IIc 2/ * 1* aph(2 )-Ic 15 CSTA30 S. xylosus IIc 2/ CSTA31 S. capitis IIc 2/ CSKR35 S. xylosus IIc 2/ (hemol.) CSKR36 S. xylosus IIc 2/ (white) CSKR37 S. kloosii IIc 2/ * 16* 16* blaz, meca, aac(6 )-Ie aph(2 )-Ia, str 1 CSCA2 S. capitis IIa 2/ * 0.5* CSCA5 S. equorum IIa 2/ * blaz, aac(6 )-Ie aph(2 )-Ia, str, dfr(a) CSCA8 S. kloosii IIa 2/1 32* * meca, str, dfr(d) CSCA10 S. xylosus IIa 2/ * dfr(d), mph(c) 2 CSHO4 S. vitulinus IIa 2/ * * aac(6 )-Ie aph(2 )-Ia, str, dfr(d), tet(k) 3 CSFA8 S. xylosus IIa 2/ * CSFO10 S. capitis IIa 2/ * meca 5 CSMA1 S. sciuri IIa 16/8* 64* * 32* 16* 2 32* 1 32* meca, aac(6 )-Ie aph(2 )-Ia, ant(6)-ia, aph(3 )-III, str, tet(m), cat(pc221), sat4 13 CSRE22 S. sciuri IIa 8/4* 64* * 32* 16* 1 64* 1 32* meca, aac(6 )-Ie aph(2 )-Ia, ant(6)-ia, str, dfr(a), tet(m), cat(pc221) CSRE23 S. equorum IIa 2/ * CSGO27 S. vitulinus IIa 2/ * 0.5 4* blaz, aac(6 )-Ie aph(2 )-Ia, str 17 CSNO40 S. xylosus IIa 2/ * * blaz, aac(6 )-Ie aph(2 )-Ia, str, dfr(a), mph(c) CSNO41 S. equorum IIa 2/ * blaz, aac(6 )-Ie aph(2 )-Ia, str, dfr(d), mph(c) CSNO42 S. vitulinus IIa 2/ * meca, aac(6 )-Ie aph(2 )-Ia, str, dfr(d) CSNO43 S. aureus IIa 2/ CSJAE48 S. capitis IIa 2/ * 32* 8* blaz, meca, aac(6 )-Ie aph(2 )-Ia, str Continued on following page

5 4448 SCHNELLMANN ET AL. J. CLIN. MICROBIOL. TABLE 2 Continued Horse no. Strain Species Sample b MIC ( g/ml) for antibiotic c Antibiotic resistance gene(s) detected d AMC CHL CLI ENR ERY GEN OXA PEN Q-D TET VAN XNL CSJAE49 S. equorum IIa 2/ * * blaz, aac(6 )-Ie aph(2 )-Ia, str, dfr(a), dfr(d) 1 CSCA12 S. equorum IIb 2/ * mph(c) CSCA9 S. kloosii IIb 2/ * meca, aac(6 )-Ie aph(2 )-Ia, aph(2 )-Ic CSCA11 S. xylosus IIb 2/ * tet(k) 2 CSHO2 S. equorum IIb 2/ * 4 1 8* blaz, aac(6 )-Ie aph(2 )-Ia, str, dfr(a), mph(c) CSHO6 S. xylosus IIb 2/1 64* * * 2 2 blaz, aac(6 )-Ie aph(2 )-Ia, dfr(a), tet(k), cat(pc223) 4 CSFO11 S. capitis IIb 2/ str 6 CSLA3 S. sciuri IIb 2/1 32* 32* * 32* 32* 0.25* * meca, aac(6 )-Ie aph(2 )-Ia, ant(4 )-Ia, str, dfr(d), cat(pc221), erm(b) 12 CSEV20 S. xylosus IIb 2/ CSRE24 S. equorum IIb 2/ * CSNO44 S. equorum IIb 2/ * * blaz, aac(6 )-Ie aph(2 )-Ia, str, dfr(d), mph(c) CSNO45 S. kloosii IIb 2/ * 0.25* meca, aac(6 )-Ie aph(2 )-Ia, str, dfr(d) CSNO46 S. xylosus IIb 2/ * * blaz, aac(6 )-Ie aph(2 )-Ia, ant(4 )-Ia, str, dfr(d) 18 CSJAE50 S. vitulinus IIb 2/ * 1* blaz, meca, aac(6 )-Ie aph(2 )-Ia, dfr(a) 1 CSCA3 S. capitis III 2/ * 0.5* 0.25* blaz, meca, aac(6 )-Ie aph(2 )-Ia, str CSCA6 S. equorum III 2/ * * blaz, aac(6 )-Ie aph(2 )-Ia, str, dfr(a), mph(c) 5 CSMA2 S. xylosus III 2/ * blaz, aac(6 )-Ie aph(2 )-Ia, ant(4 )-Ia, dfr(a) 6 CSLA4 S. xylosus III 2/ CSLA5 S. capitis (white) III 2/ * * 1 1 meca, aac(6 )-Ie aph(2 )-Ia, str, dfr(d), tet(k) CSLA6 S. capitis III 2/ * 1* * 1 1 (gray) 17 CSNO38 S. epidermidis III 2/1 4 32* * 16* 8* 16* * 2 4 blaz, meca, aac(6 )-Ie aph(2 )-Ia, ant(6)-ia, aph(3 )-III, str, tet(k), sat4, erm(c) 18 CSJAE39 S. aureus III 2/1 8 32* * 16* 16* 8* * blaz, aac(6 )-Ie aph(2 )-Ia, dfr(a) 19 CSRI51 S. aureus III 2/ * 1 16* blaz, aac(6 )-Ie aph(2 )-Ia, dfr(a) a Strains were isolated, before and after admission to the clinic and before and after penicillin treatment, from 19 horses seen at the equine clinic of the Vetsuisse Faculty of the University of Berne between August 2004 and May b Sample I includes strains isolated from horses before hospitalization, sample IIc includes strains isolated from hospitalized horses that did not receive penicillin (control), samples IIa and IIb include strains isolated from hospitalized horses that received penicillin, and sample III includes strains isolated from wound secretion. c AMC, amoxicillin-clavulanic acid (tested in a concentration ratio of 2:1); CHL, chloramphenicol; CLI, clindamycin; ENR, enrofloxacin; ERY, erythromycin; GEN, gentamicin; OXA, oxacillin supplemented with 2% NaCl; PEN, penicillin; Q-D, quinupristin-dalfopristin; TET, tetracycline; VAN, vancomycin; XNL, ceftiofur. Asterisks denote MICs above the resistance breakpoint, and daggers denote MICs below the resistance breakpoint but with the presence of a respective antibiotic resistance gene. d Shown are resistance genes to -lactams (meca and blaz), macrolides [mph(c), erm(b), and erm(c)], tetracycline [tet(k) and tet(m)], chloramphenicol [cat(pc223) and cat(pc221)], trimethoprim [dfr(a) and dfr(d)], streptothricin (sat4), and aminoglycosides [aac(6 )-Ie aph(2 )-Ia, ant(4 )-Ia, ant(6)-ia, aph(3 )-III, aph(2 )-Ic, and str]. Resistance genes shown in bold have been sequenced. e hemol., hemolytic on sheep blood agar plates. genes. Two macrolide-resistant strains (S. epidermidis CSNO38 and S. sciuri CSLA3), which also displayed resistance to clindamycin, were found to contain the methyltransferase genes erm(c) and erm(b), respectively. No lincosamide resistance genes were detected in the clindamycin-resistant and erythromycin-susceptible strain S. aureus CSJAE39. Resistance to -lactam antibiotics resulted mainly from the presence of the blaz or the meca gene (see below). Decreased susceptibility to gentamicin could be explained by the presence of the bifunctional aminoglycoside acetyltransferase-phosphotransferase gene aac(6 )-Ie aph(2 )-Ia (31 strains) or the aminoglycoside phosphotransferase genes aph(2 )-Ic (two strains) and aph(2 )-III (two strains). Resistance to chloramphenicol could be attributed to chloramphenicol acetyltransferase genes of the cat(pc221) and cat(pc223) family, except for S. kloosii strain CSCA8, in which no cat gene was detected. Resistance to tetracycline could be attributed mainly to the efflux gene tet(k) and, to a lesser extent, to ribosomal protection encoded by the tet(m) gene. Additional resistance genes with phenotypes that were not determined by MIC were detected by microarray. Thus, the streptomycin adenylyltransferase genes str and ant(6 )-Ia, the tobramycin and amikacin adenylyltransferase gene ant(4 )-Ia, the trimethoprim dihydrofolate reductase genes

6 VOL. 44, 2006 ANTIBIOTIC RESISTANCE IN STAPHYLOCOCCI FROM HORSES 4449 FIG. 1. Distribution of antibiotic resistance (based on phenotypic and/or genotypic testing) in 72 CoNS strains and three S. aureus strains isolated, before and after admission to the clinic and before and after penicillin treatment, from 19 horses seen at the equine clinic of the Vetsuisse Faculty of the University of Berne between August 2004 and May Antibiotics: AMC, amoxicillin-clavulanic acid; CHL, chloramphenicol; CLI, clindamycin; ENR, enrofloxacin; ERY, erythromycin; GEN, gentamicin; OXA, oxacillin; PEN, penicillin; STH, streptothricin; STR, streptomycin; SYN, quinupristin-dalfopristin; TET, tetracycline; TMP, trimethoprim; VAN, vancomycin; XNL, ceftiofur. dfr(a) and dfr(d), and the streptothricin acetyltransferase gene sat4 were detected in several strains. S. aureus strain CSJAE39 showed phenotypic resistance to the streptogramin combination quinupristin-dalfopristin, but no corresponding resistance genes were found (Table 2). Emergence of antibiotic resistance after hospitalization. Most susceptible strains were isolated before hospitalization (10 strains from nine horses) or from hospitalized horses that did not receive penicillin (four strains from two horses). In contrast, only four strains (from four horses), each isolated from hospitalized horses that received penicillin, displayed no antibiotic resistance. The prevalence of antibiotic resistance in the analyzed Staphylococcus population (based on MIC and/or microarray results) increased after administration of penicillin (Fig. 1). Specifically, resistance to penicillin increased from 38.5% to 76.9%, oxacillin from 26.9% to 41.0%, erythromycin from 26.9% to 30.8%, gentamicin from 7.7% to 66.7%, streptomycin from 0% to 53.8%, trimethoprim from 0% to 51.3%, tetracycline from 7.7% to 20.5%, chloramphenicol from 0% to 12.8%, ceftiofur from 0% to 10.3%, clindamycin from 0% to 7.7%, the combination amoxicillin-clavulanic acid from 0% to 5.1%, streptothricin from 0% to 5.1%, and the combination quinupristin-dalfopristin from 0% to 2.6%. Horses that had received at least 72 h of penicillin treatment had a 5.8-timeshigher chance (odds) of harboring penicillin-resistant staphylococci than horses that were not admitted to the clinic and did not receive penicillin treatment (95% CI, 2.0 to 16.6; P 0.001). Control horses with a 72-h stay at the clinic but no penicillin treatment still had a 2.4-times-higher chance (odds) of harboring penicillin-resistant strains (95% CI, 0.5 to 10.7). No difference was observed between the resistance profiles of staphylococci isolated from a previously disinfected skin area (sample IIa) and staphylococci isolated from skin that had not been disinfected (sample IIb) (Table 2). Before hospitalization, only a few isolated genes that confer resistance to oxacillin (meca), macrolides [mph(c)], aminoglycosides [aac(6 )-Ie aph(2 )-Ia], and tetracycline [tet(k)] were found (Table 2 and Fig. 2). The number of resistance genes subsequently increased after hospitalization, even in the absence of penicillin treatment. In the control group, genes conferring resistance to -lactams (blaz and meca), aminoglycosides [aac(6 )-Ie aph(2 )-Ia, aph(2 )-Ic, ant(4 )-Ia, and str], tetracycline [tet(k)], macrolides [mph(c)], and trimethoprim [dfr(d)] appeared. The same trend, but even more accentuated, occurred in strains isolated from horses that received penicillin. Besides resistance to -lactams, which was attributed mainly to the meca and/or the blaz gene, these isolates harbored additional genes, such as those conferring resistance to aminoglycosides [aac(6 )-Ie aph(2 )-Ia, ant(4 )-Ia, ant(6)- Ia, aph(3 )-III, aph(2 )-Ic, and str], macrolides [mph(c), erm(b), and erm(c)], trimethoprim [dfr(a) and dfr(d)], tetracycline [tet(k) and tet(m)], chloramphenicol [cat(pc221) and cat(pc223)], and streptothricin (sat4). Of a total of nine strains found in wound infections, all strains except one (S. xylosus CSLA4) displayed phenotypic resistance to -lactam antibiotics. The -lactam-resistant strains harbored either or both the blaz and the meca gene and demonstrated multiresistance (Table 2). The aminoglycoside resistance tandem gene aac(6 )-Ie aph(2 )-Ia was present in all wound isolates harboring meca or blaz. In addition to these two resistance traits, additional genes conferring resistance to trimethoprim [dfr(a) and dfr(d)], tetracycline [tet(k)], macrolides [mph(c)], or streptothricin (sat4) were found in some isolates. The S. capitis CSLA6 strain showed clear phenotypic resistance to penicillin, oxacillin, and tetracycline, but no respective genes could be detected. Resistance linkage. Samples from horses with at least 72 h of penicillin treatment had a 22.5-times-higher chance (odds) of carrying the resistance gene to aminoglycosides [aac(6 )-Ie aph(2 )-Ia] than horses without admission to the clinic and without penicillin treatment (95% CI, 4.70 to ; P 0.001). Horses with a 72-h stay at the clinic but without penicillin treatment still had an 8.0-times-higher chance (odds) of carrying this resistance gene (95% CI, 1.17 to 54.50; P 0.034). The blaz gene was always associated with the aminoglycoside resistance gene aac(6 )-Ie aph(2 )-Ia. In the absence of the blaz gene, the aac(6 )-Ie aph(2 )-Ia gene was associated with a meca gene. Only one strain (S. vitulinus CSHO4) carrying aac(6 )-Ie aph(2 )-Ia isolated after hospitalization did not carry a -lactam resistance gene. Additionally, the streptomycin resistance gene str appeared in 21/24 cases together with the aminoglycoside resistance gene aac(6 )-Ie aph(2 )-Ia and the -lactam resistance genes. The trimethoprim resistance genes dfr(a) and dfr(d) were always linked with a -lactam resistance gene and the aac(6 )-Ie aph(2 )-Ia gene, except in strains CSCA10 and CSHO4.

7 4450 SCHNELLMANN ET AL. J. CLIN. MICROBIOL. FIG. 2. Occurrence of resistance genes in 72 CoNS strains and three S. aureus strains isolated, before and after admission to the clinic and before and after penicillin treatment, from 19 horses seen at the equine clinic of the Vetsuisse Faculty of the University of Berne between August 2004 and May The detected antibiotic resistance genes confer resistance to -lactams (meca and blaz), macrolides [mph(c), erm(b), and erm(c)], tetracycline [tet(k) and tet(m)], chloramphenicol [cat(pc221) and cat(pc223)], trimethoprim [dfr(a) and dfr(d)], streptothricin (sat4), and aminoglycosides [aac(6 )-Ie aph(2 )-Ia, ant(4 )-Ia, ant(6)-ia, aph(3 )-III, aph(2 )-Ic, and str]. Sequence analysis of antibiotic resistance genes. In order to determine the percentages of similarity between resistance genes detected by microarray and corresponding genes from GenBank, a certain number of genes were chosen based on their occurrence and sequenced (Table 2). The sequence analysis of internal fragments of the genes blaz, aac(6 )-Ie aph(2 )-Ia, str, dfr(a), dfr(d), tet(k), tet(m), sat4, and cat(pc221) showed sequence identity greater than 99% with those from GenBank. The cat(pc223) gene showed 97% identity to that from GenBank. However, the mph(c) and meca sequences displayed only 91% identity with similar genes from GenBank and were, as a result, further characterized. Characterization of a new meca variant. Sequence alignment of complete genes revealed three types of meca genes. The first type displayed 100% DNA identity with that from S. aureus (GenBank accession no. DQ106887) and was found only in strains isolated after hospitalization (S. sciuri CSRE22, CSLA3, and CSMA1, S. epidermidis CSNO38, S. vitulinus CSNO42 [GenBank accession no. AM048802], and S. kloosii CSNO45 [GenBank accession no. AM048803]). The second type of meca gene showed 99.8% DNA identity to meca of S. aureus (GenBank accession no. DQ106887) and was also found only in strains isolated after hospitalization (S. kloosii CSCA9 [GenBank accession no. AM048804] and CSKR37 and S. capitis CSLA5 [GenBank accession no. AM048805], CSBO10, and CSJAE48). The third type of meca gene represents a new variant. It displayed 91.0% DNA identity to the meca gene of S. aureus (GenBank accession no. DQ106887) and 90.0% identity to the amino acid sequence. This new meca variant was found in different Staphylococcus species isolated from five different horses before hospitalization (S. kloosii CSCA7 [GenBank accession no. AM048807], S. vitulinus CSBO8 [GenBank accession no. AM048810] and CSHO3, and S. capitis CSJAE47 [GenBank accession no. AM048806] and CSTA28), as well as in staphylococci which had been isolated from two horses after hospitalization (S. kloosii CSCA8, S. vitulinus CSWY15 [GenBank accession no. AM048811] and CSJAE50, and S. capitis CSCA3 [GenBank accession no. AM048809] and CSFO10). This variant appeared mainly alone and in strains isolated before hospitalization, but it always appeared in multidrug-resistant strains after hospitalization. Characterization of new mph(c) variants. Eleven isolates of S. xylosus and S. equorum were found to harbor an mph(c) gene. The mph(c) genes differed slightly from each other, with identities between 98.1% and 98.9%. Alignment of the complete nucleotide sequence of these genes with those of the known mph(c) gene from S. aureus (GenBank accession no. AF167161) and Stenotrophomonas maltophilia (GenBank accession no. AJ251015) revealed them to be new mph(c) variants. The most frequently occurring mph(c) variant (GenBank accession no. AM180066) was found in strains isolated before hospitalization (S. equorum CSBO9) as well as in strains isolated after hospitalization (S. equorum CSNO41, CSCA6, and CSCA12 and S. xylosus CSNO40). This variant showed 93.0% and 92.7% overall DNA identities to the mph(c) gene of S. aureus and S. maltophilia, respectively, and 90% overall identity to the Mph(C) amino acid sequence. The second most

8 VOL. 44, 2006 ANTIBIOTIC RESISTANCE IN STAPHYLOCOCCI FROM HORSES 4451 frequently occurring mph(c) variant (GenBank accession no. AM180067) was found in one strain isolated before hospitalization (S. xylosus CSBO11) and another isolated after hospitalization (S. equorum CSHO1). This variant showed 92.7% and 92.4% overall DNA identities to the mph(c) gene of S. aureus and S. maltophilia, respectively, and 90% overall identity to the Mph(C) amino acid sequence. Each of the four other mph(c) variants had a distinct sequence. The overall DNA identity of these mph(c) genes to those of S. aureus and Stenotrophomonas maltophilia was between 92.8% and 93.4%, and the amino acid identity was between 90% and 91.3%. These different mph(c) variants were found in strains isolated before hospitalization (S. equorum CSEV19 [GenBank accession no. AM397631]) and after hospitalization (S. xylosus CSCA10 [GenBank accession no. AM180068] and S. equorum CSNO44 [GenBank accession no. AM397632] and CSHO2). As was already observed with the meca gene, the mph(c) gene appeared alone in strains isolated before hospitalization and in combination with other genes in strains isolated after hospitalization. Strains containing the mph(c) gene displayed different levels of erythromycin resistance, some having an MIC below the CLSI resistance breakpoint independent of the isolation stage. Expression of meca, blaz, mph(c), and aac(6 )-Ie aph(2 )-Ia resistance genes. All strains containing meca genes were found to be oxacillin resistant and able to produce PBP2 protein. All strains carrying blaz were resistant to penicillin and produced a -lactamase. Many of the meca-expressing strains displayed no resistance to penicillin in vitro, having an MIC value of 0.12 g/ml (S. capitis CSLA5, CSBO10, CSTA28, CSJAE47, and CSFO10; S. vitulinus CSHO3, CSBO8, and CSNO42; and S. kloosii CSCA7, CSCA8, and CSCA9). This is probably due to the fact that penicillin was not tested under inducing conditions, in contrast to oxacillin, which was tested with 2% NaCl to maximize expression of the meca gene. Six strains (S. xylosus CSWE7, S. capitis CSGO25, CSWY16, CSCA2, and CSLA6, and S. cohnii subsp. cohnii CSWY14) displayed oxacillin resistance with an MIC of 4 g/ml but did not contain either the meca or the blaz gene and did not produce the PBP2 protein or a -lactamase. Five strains (S. capitis CSFO9, S. equorum CSHO1 and CSRE24, and S. xylosus CSKR33 and CSFA8) without blaz and meca genes were found to be resistant to penicillin but susceptible to oxacillin. These strains also did not produce PBP2 protein or a -lactamase. S. aureus strain CS- JAE39 was clearly oxacillin resistant, having an MIC of 16 g/ml, and contained blaz but not meca. The expression of the erythromycin resistance gene mph(c) and the gentamicin resistance gene aac(6 )-Ie aph(2 )-Ia, which displayed intermediate resistance in several Staphylococcus strains, was verified by RT-PCR. The aac(6 )-Ie aph(2 )-Ia gene was transcribed in strains S. xylosus CSBO11 and CSNO40 and S. equorum CSCA6 and CSNO41, which displayed intermediate resistance to gentamicin (MIC, 2 to 4 g/ml). Gentamicin-resistant S. kloosii CSKR37 (MIC, 16 g/ ml) was used as a control. Similarly, transcription of the mph(c) gene also occurred in strains S. xylosus CSBO11 and S. equorum CSBO9, CSHO1, and CSNO41, which displayed intermediate resistance to erythromycin (MIC, 2 to 4 g/ml). S. equorum CSCA6 and CSCA12 (MICs of 8 and 16 g/ml, respectively) were used as positive controls. This confirmed that the expression of the aac(6 )-Ie aph(2 )-Ia and mph(c) genes did not confer high-level resistance in every Staphylococcus strain. Decreased susceptibility to gentamicin in CoNS containing aac(6 )-Ie aph(2 )-Ia has already been reported for blood culture isolates (26). A recent study demonstrated that the induction of the aac(6 )-Ie aph(2 )-Ia gene with a -lactam antibiotic increased levels of resistance to aminoglycosides (24). DISCUSSION The predominant Staphylococcus species isolated during this study from the skin and wounds of horses were from the group of coagulase-negative staphylococci found in normal skin flora. The few S. aureus isolates obtained were found mainly in wound infections, with the exception of S. aureus strain CSNO43, which was isolated from a skin sample. Some CoNS, such as S. epidermidis, S. capitis, and S. equorum, were found to be opportunistic pathogens recovered from wound secretions as single agents causing infections (horses no. 1 and 17). In two cases, S. xylosus and S. capitis were found to be present in the wound with other bacteria, like Escherichia coli and Pasteurella caballi (horse no. 5) and E. coli, Aerococcus viridans, and Kocuria rosea (horse no. 6). The PFGE-based molecular fingerprinting of the staphylococcal isolates revealed a high diversity among CoNS, even among the same species isolated from the same horse. Such diversity within members of the same species on the same animal has been observed previously with equine skin isolates (59). This high heterogeneity within species did not enable us to trace strains back to their point of origin in order to determine whether those involved in infections were present before hospitalization. The ability to trace strains was also limited by the fact that only one representative of each colony type was picked from each skin culture. This strategy allowed us to characterize a large variety of species and avoid selecting identical isolates from the same samples. At the same time, it decreased the chance of isolating the same strain at different stages of sampling, i.e., upon arrival of the horses to the hospital, after penicillin treatment, and in wound secretions. However, there was a clear trend toward an increasing amount of resistance in staphylococci after hospitalization, both in horses that received penicillin and in horses that did not receive penicillin. Among the latter, two of four horses (no. 8 and 16) from which staphylococci with no resistance or only one antibiotic resistance gene were isolated before hospitalization were found to harbor staphylococci containing multiple antibiotic resistance after hospitalization. In this control group, the amount of resistance remained low only in isolates from horse no. 15. Antibiotic resistance genes detected in staphylococci before hospitalization were mainly meca and mph(c) genes and, to a lesser extent, aac(6 )-Ie aph(2 )-Ia and tet(k) genes. Each meca gene detected in staphylococci isolated before hospitalization belonged to a group of new variants, suggesting that horses bring new antibiotic resistance genes into the hospital. After application of penicillin, the number of phenotypic resistances as well as the number of resistance genes increased dramatically. Only 4 of 39 strains isolated from animals after penicillin treatment were susceptible to all antibiotics. The majority (30/39 strains) were resistant to -lactam antibiotics,

9 4452 SCHNELLMANN ET AL. J. CLIN. MICROBIOL. indicating that penicillin application had selected for a penicillin-resistant population. The few isolates that were susceptible to -lactams may have been acquired by subsequent contamination after the bactericidal efficacy of penicillin had diminished. Importantly, selected penicillin-resistant bacteria were found to harbor additional antibiotic resistance genes. A clear link exists among resistance to -lactam, gentamicin, and streptomycin, but the locations of the corresponding genes on the same genetic element remain to be determined. The aminoglycoside resistance gene aac(6 )-Ie aph(2 )-Ia has already been found to be linked with blaz on transposon Tn5385 (44) and with meca after insertion of Tn4001 into the staphylococcal cassette chromosome mec cassette (6) (R. Heusser and B. Berger-Bächi, personal communication). The increase of aminoglycoside resistance genes after hospitalization may also be due to the fact that all horses undergoing surgery (other than colic surgery) at the equine clinic of the University of Berne are routinely given the combination penicillin-gentamicin as an antimicrobial therapy. This common practice may have selected for a -lactam- and aminoglycosideresistant bacterial flora in the hospital. In a recent study, the administration of -lactam antibiotics and aminoglycosides during hospitalization has been considered a risk factor in the colonization of horses with MRSA (58). This is also true for other staphylococci, since multidrug-resistant CoNS are found on horses shortly after hospitalization. The hospital environment harbors a multidrug-resistant flora that rapidly colonizes the skin of horses and represses natural skin flora. The repression of original and antibiotic-susceptible flora is, in addition, enhanced by the use of preventive penicillin. The rapid colonization of horses with nosocomial flora may be caused either through direct contact with horses already hospitalized or via vectors like brushes, dust, flies, straw, and personnel of the clinic. This multidrug-resistant nosocomial flora has certainly acquired and exchanged, over the years, a collection of individual antibiotic resistance genes or clusters of genes via mobile genetic elements. These mobile elements may then be transferred to bacteria that were previously susceptible to antibiotics. Although macrolide antibiotics are rarely used in equine medicine in Switzerland, due to the risk of fatal diarrhea in adult horses (40, 42) and the infrequency of Rhodococcus equi infections in foals (18), examined horses were found to harbor staphylococci with macrolide resistance genes, including new variants of mph(c) genes. The DNA sequence variations found in the different mph(c) genes may result from gene transfer between different Staphylococcus species. Additionally, the presence of new mph(c) variants in Staphylococcus species that are specific to animals is an indication that resistance has been selected for in animals rather than humans. Similarly, meca has most likely been selected for by -lactam administration in horses. This gene is known to be widespread in CoNS from healthy horses and has already been detected in S. epidermidis, S. saprophyticus, S. xylosus, S. lentus, S. haemolyticus, and S. sciuri in Japan (59, 60) and in S. sciuri, S. lentus, S. capitis, S. cohnii subsp. cohnii, S. epidermidis, S. kloosii, S. haemolyticus, S. caprae, and S. warneri in The Netherlands (4). In our study, three different meca genes were found, including a new variant identified in S. kloosii, S. vitulinus, and S. capitis. Nevertheless, -lactam resistance in six strains could not be explained, suggesting the presence of additional variants or of an as-yetunknown mechanism of resistance in these strains. CoNS of both human (14, 19, 36, 41, 48) and animal (4, 25, 60) origin represent a large reservoir of meca for S. aureus, and MRSA may emerge under the selective pressure of -lactam antibiotics. Evidence supports horizontal transfer of meca between different Staphylococcus species, and the direction of gene flow seems to go from CoNS to S. aureus (1). In Canada, MRSA strains have already emerged in horses, causing therapeutic problems and representing a public health threat if transmitted to humans. Indeed, MRSA strains of equine origin were previously involved in an outbreak of skin infections in the personnel at a veterinary hospital (55). As a consequence, all horses entering the clinic of the University of Guelph s Ontario Veterinary College are now screened for MRSA to reduce the risk of spreading this zoonotic multiantibiotic-resistant pathogen (7). None of the isolated S. aureus strains in this study contained meca, even though one of them was clearly resistant to oxacillin (CSJAE39). It might, therefore, only be a question of time until the first MRSA strain emerges at the equine clinic of the University of Berne because -lactam antibiotics are routinely used. This study has demonstrated that horses entering the hospital harbor staphylococci containing antibiotic resistance genes, including new variants of meca and mph(c) genes. Shortly after hospitalization, horses acquired a specific multidrug-resistant skin flora that was presumably selected for and maintained in the hospital by the use of penicillin. Hence, antibiotics should be limited to the treatment of infection and not used for infection prevention. Such prudent use could help prevent selection for multidrug-resistant strains such as MRSA strains in animals. ACKNOWLEDGMENTS We thank the personnel from the equine clinic of the Vetsuisse Faculty of the University of Berne for assistance in collecting skin samples, Isabelle Brodard, Anita Jaussi-Holzer, Jan Keller, and Yvonne Schlatter for technical assistance, and Sarah Burr and Edy Vilei for helpful advice. REFERENCES 1. Archer, G. L., J. A. Thanassi, D. M. Niemeyer, and M. J. Pucci Characterization of IS1272, an insertion sequence-like element from Staphylococcus haemolyticus. Antimicrob. Agents Chemother. 40: Baptiste, K. E., K. Williams, N. J. Williams, A. Wattret, P. D. Clegg, S. Dawson, J. E. Corkill, T. O Neill, and C. A. Hart Methicillin-resistant staphylococci in companion animals. Emerg. Infect. Dis. 11: Barbosa, T. M., K. P. Scott, and H. J. Flint Evidence for recent intergeneric transfer of a new tetracycline resistance gene, tet(w), isolated from Butyrivibrio fibrisolvens, and the occurrence of tet(o) in ruminal bacteria. Environ. Microbiol. 1: Busscher, J. F., E. van Duijkeren, and M. M. Sloet van Oldruitenborgh- Oosterbaan The prevalence of methicillin-resistant staphylococci in healthy horses in the Netherlands. Vet. Microbiol. 113: Cefai, C., S. Ashurst, and C. Owens Human carriage of methicillinresistant Staphylococcus aureus linked with pet dog. Lancet 344: Chongtrakool, P., T. Ito, X. X. Ma, Y. Kondo, S. Trakulsomboon, C. Tiensasitorn, M. Jamklang, T. Chavalit, J. H. Song, and K. Hiramatsu Staphylococcal cassette chromosome mec (SCCmec) typing of methicillin-resistant Staphylococcus aureus strains isolated in 11 Asian countries: a proposal for a new nomenclature for SCCmec elements. Antimicrob. Agents Chemother. 50: Church, S. L MRSA surveillance in horses at a hospital. The Horse 2005(August): Clinical and Laboratory Standards Institute/NCCLS Performance standards for antimicrobial disk and dilution susceptibility tests for bacteria isolated from animals, 2nd ed., vol. 22, no. 6. Approved standard M31-A2. NCCLS, Wayne, Pa.