Diversity of Staphylococcal Cassette Chromosome mec Structures in Methicillin-Resistant Staphylococcus epidermidis and Staphylococcus haemolyticus Strains among Outpatients from Four Countries

Abstract

In staphylococci, methicillin (meticillin) resistance (MR) is mediated by the acquisition of the mecA gene, which is carried on the size and composition variable staphylococcal cassette chromosome mec (SCCmec). MR has been extensively studied in Staphylococcus aureus, but little is known about MR coagulase-negative staphylococci (MR-CoNS). Here, we describe the diversity of SCCmec structures in MR-CoNS from outpatients living in countries with contrasting environments: Algeria, Mali, Moldova, and Cambodia. Their MR-CoNS nasal carriage rates were 29, 17, 11, and 31%, respectively. Ninety-six MR-CoNS strains, comprising 75 (78%) Staphylococcus epidermidis strains, 19 (20%) Staphylococcus haemolyticus strains, 1 (1%) Staphylococcus hominis strain, and 1 (1%) Staphylococcus cohnii strain, were analyzed. Eighteen different SCCmec types were observed, with 28 identified as type IV (29%), 25 as type V (26%), and 1 as type III (1%). Fifteen strains (44%) were untypeable for their SCCmec. Thirty-four percent of MR-CoNS strains contained multiple ccr copies. Type IV and V SCCmec were preferentially associated with S. epidermidis and S. haemolyticus, respectively. MR-CoNS constitute a widespread and highly diversified MR reservoir in the community.

Methicillin (meticillin)-resistant (MR) staphylococci cause a wide variety of infections and raise high concerns, because often few therapeutic options are available. Among the staphylococci, Staphylococcus aureus is much more virulent than the other species, which are grouped together under the generic name of coagulase-negative staphylococci (CoNS). CoNS usually are much more resistant to antibiotics than S. aureus (11), but in most cases they cause infections only in patients who carry indwelling devices and/or are immunocompromised (20). CoNS are believed to constitute a reservoir of resistance genes for S. aureus (16). The nares are an ecological niche for staphylococci (8), as 20 to 30% of humans are colonized by S. aureus (31, 48), although the few available data on CoNS carriage range from 46 to 65% of hospital patients (1, 42). The horizontal transfer of resistance genes from CoNS to S. aureus has been clearly demonstrated for fusidic acid, gentamicin, and mupirocin (2, 13, 21, 45, 46) and at least once for MR (47). MR in staphylococci is driven by the acquisition of the mecA gene, which encodes PBP2A, a transpeptidase with a low affinity for β-lactams (18, 33). mecA is part of the mec complex, which includes its repressor genes mecI and mecR1. Ten mec complex subclasses, divided into six main classes (A to E), have been described so far on the basis of the polymorphism of mecI and mecR1 (25, 28, 29, 44). The mec complex is carried by a mobile genetic element called the staphylococcal cassette chromosome mec (SCCmec) (26), a genomic island of variable size (range, 21 to 67 kb). This island is integrated at the 3′ extremity of orfX (23), a gene of unknown function located near the chromosomal origin of replication. In addition to mecA, SCCmec also carries a set of cassette chromosome recombinase (ccr) genes encoding recombinases responsible for both its chromosomal integration and excision (26). Six ccr allotypes have been identified so far (10, 22, 24, 26, 38). The definition of an SCCmec type is based on the combination of a mec complex class and a ccr allotype (22). In S. aureus, six main types of SCCmec (I to VI) and several subtypes have been described (9, 22, 24, 32, 38). Much less is known about the genetics of MR in CoNS. Recent reports suggest that in CoNS, SCCmec structures are more diverse and include either mec-ccr combinations as-yet undescribed for S. aureus (35) or more than one ccr allotype (17). However, the epidemiological magnitude of these combinations has not yet been assessed. Country-to-country variations in the antibiotic susceptibility of staphylococci have been observed (1, 7, 27, 31, 50), but they may result from differences in the epidemiological and microbiological methods of investigation used. Here, we undertook a multiple-country study based on the same protocol designed to describe the nasal MR-CoNS reservoir in outpatients. We found that the prevalences of MR-CoNS carriage significantly differed according to geographical area, and that their SCCmec elements were much more diverse than those so far described for S. aureus.

MATERIALS AND METHODS

Populations and bacterial strains.

The present study was performed on a subset of nasal samples gathered during a large epidemiological study devoted to the analysis of S. aureus nasal carriage in a community of countries with different environments (41 and S. Mesli, L. Armand-Lefevre, S. Benchouk, H. Hassaine, K. Megueni, M. Benkalfat, J. C. Lucet, A. Andremont, and R. Ruimy, presented at the 26ème Réunion Interdisciplinaire de Chimiothérapie Anti-Infectieuse (RICAI), Paris, France, 7 to 8 December 2006). Nasal swabs were obtained using a standard procedure by an investigator trained in the coordinating center of the study. Swabs were taken within 8 h of the admission of 330, 338, 448, and 442 consecutive patients into the respective emergency wards of the major hospitals in the following four towns: Tlemcen, Algeria (June to October 2005); Chisinau, Moldova (June to October 2005); Bamako, Mali (March to August 2005); and Phnom Penh, Cambodia (June to October 2006). Swabs were rapidly discharged into 1.5 ml of brain heart infusion broth with 10% glycerol, stored at −80°C, and transported frozen to the coordinating center of the study, where 70 samples per site were randomly selected. Fifty microliters of the selected samples were plated on Chapman agar (Oxoid, Basingstoke, United Kingdom) for 48 h at 37°C. Four different mannitol-negative colonies were randomly chosen from each plate, subcultured, and screened by triplex real-time PCR (RT-PCR) using primers hybridizing a specific rrs region (located in the 16S RNA gene) for the Staphylococcus genus, the femA gene (specific for S. aureus) and the mecA gene (MR), as described previously (40). mecA gene-positive and femA gene-negative isolates (i.e., putative MR-CoNS) were further characterized by the disc diffusion method for susceptibility to 18 antibiotics: benzylpenicillin, oxacillin, cefoxitin, moxalactam, kanamycin, tobramycin, gentamicin, erythromycin, lincomycin, pristinamycin, levofloxacin, vancomycin, teicoplanin, tetracyline, cotrimoxazole, rifampin (rifampicin), fusidic acid, and fosfomycin, as recommended by the French Society for Microbiology (www.sfm.asso.fr). In the same isolates, species were identified by sequencing 1,300 bp located within the rrs gene, as described previously (39). The sequences were aligned using BioEdit 5.0.6 (http://www.mbio.ncsu.edu/BioEdit/bioedit) with those of 38 Staphylococcus species obtained from GenBank and corresponding to the type strain of each species when it was available. Phylogenetic analysis was carried out using the neighbor-joining algorithm (Kimura 2-parameter distance estimation) as implemented in MEGA 4.0. This identification targeting rrs was sufficiently accurate in terms of Staphylococcus species found here, as described previously (3, 14). Coresistances to non-β-lactam antibiotics were scored in each strain, as described previously (36). One phenotype per patient was considered for further analysis.

SCCmec fragments were characterized in all nonreplicate isolates by typing ccr and mec complexes. ccr complexes were typed by multiplex PCR (M-PCR), as described previously (30), except that three primers were changed as follows: (i) γR was replaced a new primer, ccrCU1 (Table 1), which generates amplicons of 607 bp with the primer γF (517 bp for the original pair γF/γR), thus allowing the detection of all known ccrC allotypes; and (ii) primers α4.2 and β4.2 were replaced by two new primers, α4U and β4U (Table 1), respectively, which amplify a 1,304-bp fragment and target ccrAB4, to detect recently published ccrAB4 alleles (12). mec complexes were typed by M-PCR as described previously (30), except that a fifth newly designed primer, IS2L (Table 1), was added in order to detect the class C1 mec complexes described for Staphylococcus haemolyticus (28). All PCRs were performed in 50-μl mixtures containing 1× Taq DNA polymerase buffer (Roche), 2.5 mM MgCl₂, 0.5 pmol of each primer, 200 μM of each deoxynucleoside triphosphate, 2.5 U of DNA Taq polymerase (Roche), and 2 μl of bacterial extract, and the mixtures were subjected to a denaturation step of 4 min at 94°C; 30 cycles of 30 s at 94°C, 1 min of annealing at either 57°C (ccr M-PCR) or 60°C (mec M-PCR), and 2 min of extension at 72°C; and a final elongation step of 2 min at 72°C in the GeneAmp PCR system 2700 (Applied Biosystem, Courtaboeuf, France). PCR products were visualized after migration in 1.7% agar 0.5× Tris-acetate-EDTA (TAE) gel using SybrSafe (Invitrogen, Cergy-Pontoise, France) as the double-stranded DNA marker.

TABLE 1.

Primers used in this study

PCR	Primer name	5′-3′ Sequence	Gene(s)	RT-PCR probe	Reference or source
16S rRNA for species identification	A2	AGAGTTTGATCATGGCTCAG	rrs		39
	A10	AAACTCAAATGAATTGACGG	rrs		39
	S15	GGGCGGTGTGTACAAGGCC	rrs		39
	S10	CCGTCAATTCATTTGAGTTT	rrs		39
mec complex typing	mA7	ATATACCAAACCCGACAACTACA	mecA		30
	mI6	CATAACTTCCCATTCTGCAGATG	mecI		30
	IS7	ATGCTTAATGATAGCATCCGAATG	IS1272		30
	IS2	TGAGGTTATTCAGATATTTCGATGT	IS431R		30
	IS2L	GAACCGCAGGTCTCTTCAGATC	IS431L		This study
ccr complex typing	mA1	TGCTATCCACCCTCAAACAGG	mecA		30
	mA2	AACGTTGTAACCACCCCAAGA	mecA		30
	α1	AACCTATATCATCAATCAGTACGT	ccrA1		30
	α2	TAAAGGCATCAATGCACAAACACT	ccrA2		30
	α3	AGCTCAAAAGCAAGCAATAGAAT	ccrA3		30
	βc	ATTGCCTTGATAATAGCCTTCT	ccrB1, ccrB2, ccrB3		30
	γF	CGTCTATTACAAGATGTTAAGGATAAT	ccrC		30
	ccrCU1	TTACCTTTGACCAATATCACATC	ccrC		This study
	α4U	GCGACGAATCAAATGTCCTTACTG	ccrA4		This study
	β4U	ATCGCTCCAGTGTCTATACTTCGC	ccrB4		This study

From the seven strains harboring three ccr allotypes (see below), we selected three strains (A186-1, M327-2, and C327-2) whose mec-ccr contents were representative of the combinations found in the four others. These contents were as follows: class A mec complex, ccrAB3, ccrAB4, and ccrC; class B mec complex, ccrAB2, ccrAB4, and ccrC; and class C2 mec complex, ccrAB2, ccrAB4, and ccrC). In the three strains, we PCR amplified the long fragments located between mecA and ccr to ascertain whether or not they were located in the mecA environment by following the protocol given below. The 25 strains harboring SCCmec type V also were tested for ccrC positioning toward mecA. Primer mA1 or mA2 was used as an anchor on mecA, together with each of the ccr primers described in Table 1. The inferred location of all ccr genes, upstream/downstream and sense/antisense in relation to mecA, then was assessed in each strain by the size of the amplicons resulting from the above-described PCRs. The size of the fragments joining the ccr allotypes in a given strain also was tested. Long-range PCR experiments were performed using the GeneAmp XL PCR kit (Applied Biosystems). Reaction mixtures contained 1× XL buffer II, 1 mM MgOAc₂, 20 pmol of each selected primer (mA1 or mA2), one ccr primer (to test all of the putative positions of ccr toward mecA), 1 U of rTth DNA polymerase XL, 0.8 mM of each deoxynucleoside triphosphate, and 2 μl of bacterial extract in a final volume of 50 μl. The mixtures were subjected to a first denaturation step of 4 min at 94°C; 10 cycles of 15 s of denaturation at 94°C, 30 s of annealing at 55°C, and 8 min of extension at 68°C; 20 cycles of 15 s of denaturation at 94°C, 30 s of annealing at 55°C, and an 8-min (with an increment of 15 s at each cycle) extension at 68°C; and a final elongation step of 10 min at 72°C. PCR products were visualized by a migration step in a 0.8% agar 0.5× TAE gel with SybrSafe as the double-stranded DNA marker. Control strains for long-range PCR are described below.

Reference strains.

The following strains were used as references: S. aureus strains COL (SCCmec Type I), BK2464 (SCCmec Type II), ANS46c (SCCmec Type III), HU25 (SCCmec Type IIIA), and HDE288 (SCCmec Type VI), kindly provided by Herminia de Lencastre, and S. aureus strain WCH100 (SCCmec Type V), kindly provided by Michele Bes.

Statistical analysis.

Epi-Info v3.2.2 (Centers for Disease Control and Prevention, Atlanta, GA) was used for statistical analysis. Associations between species, antibiotic profiles, country, mec complex class, ccr complex type, and SCCmec type were investigated using the chi-squared test. Continuous variables were compared by the analysis of variance test. P < 0.05 was considered significant.

RESULTS

The mean (range) ages of the subjects from Algeria, Moldova, Mali, and Cambodia were 42 (17 to 79), 39 (16 to 69), 52 (17 to 80), and 32 (16 to 71) years (P < 0.0001), respectively, and their respective male/female ratios were 1.1, 1.2, 1.8, and 1.8 (Table 2). The overall prevalence of MR-CoNS carriage (a patient was considered an MR-CoNS carrier when at least one MR-CoNS isolate was isolated from a nasal sample) was 22%, and it ranged from 31% (Cambodia), 29% (Algeria), and 17% (Mali) to 11% (Moldova, P < 0.05). This prevalence was higher in men than women (27% for men and 18% for women; P < 0.05). The mean (range) age was not significantly different between noncarriers and carriers (39 [4 to 80] and 42 [14 to 79] years, respectively).

TABLE 2.

Characteristics of the populations studied and prevalence of nasal carriage of CoNS and MR-CoNS

Characteristic	Result by country				P value
	Algeria	Mali	Moldova	Cambodia
No. of subjects	70	70	70	70
Avg age (yr)	42	52	39	32	<0.0001
Sex ratio (male/female)	1.1	1.8	1.2	1.8	NSa
MR-CoNS carriage prevalence (%)	28.6	17.1	11.4	31.0	<0.05
No. of MR-CoNS strains	33	19	12	32
Antibiotic resistance prevalence (%)
Kanamycin	59	79	58	56	NS
Tobramycin	47	58	58	44	NS
Gentamicin	29	58	58	38	NS
Erythromycin	32	53	67	53	NS
Lincomycin	9	0	58	37	<0.0001
Pristinamycin	0	0	33	9	<0.001
Ofloxacin	18	58	58	41	<0.05
Fusidic acid	50	16	0	6	<0.001
Fosfomycin	6	0	0	0	NS
Rifampin	12	5	32	12	NS
Cotrimoxazole	35	74	76	65	NS
Tetracycline	41	63	83	56	NS
Coresistance score	28	39	42	33	<0.05

^aNS, not significant.

Antibiotic susceptibility testing.

In all, 1,120 mannitol-negative strains were screened, of which 120 were mecA-CoNS positive by RT-PCR. After duplicates had been excluded on the basis of species identification and antibiotic susceptibility patterns, there were 96 (8.6%) separate strains from 62 subjects (20, 8, 12, and 21 from Algeria, Moldova, Mali, and Cambodia, respectively). All of these strains expressed phenotypic MR. Ninety-two (96%) also were resistant to at least one of the non-β-lactam antibiotics tested, as follows: 57 (62%) were resistant to kanamycin, 38 (41%) to tobramycin and gentamicin, 44 (48%) to erythromycin, 45 (49%) to cotrimoxazole, 51 (55%) to tetracycline, 16 (17%) to rifampin, and 2 (2%) to fosfomycin. There were no significant differences in this respect between countries. In contrast, significant between-country differences (Table 2) were observed for resistance to fluoroquinolones (the lowest resistance was in Algeria), lincomycin (the highest resistance was in Moldova), cotrimoxazole (the highest resistances was in Cambodia and Mali), and fusidic acid (the highest resistance was in Algeria). Coresistance scores also were significantly different for the four countries, with mean scores of 24.2, 29.7, 32.7, and 39.9% for Algeria, Cambodia, Mali, and Moldova, respectively (P < 0.05) (Table 2). However, coresistances were not significantly different between S. haemolyticus and S. epidermidis, the two most frequently identified species (34.2 and 28.9%, respectively).

Species identification.

The phylogenetic position of isolates (data not shown) within the genus Staphylococcus showed that 75 (78%) of the 96 strains isolated were identified as S. epidermidis, 19 (20%) as S. haemolyticus, 1 as Staphylococcus cohnii, and 1 as Staphylococcus hominis (1%). Differences in species distribution between countries were not significant (data not shown).

mec complex typing.

Ten (10.4%), 47 (49.0%), 4 (4.2%), and 35 (36.4%) strains exhibited class A, B, C1, and C2 mec complexes, respectively. S. haemolyticus was combined with class C mec complexes significantly more frequently than S. epidermidis (P < 0.001), as classes C1 and C2 were found in 16/19 S. haemolyticus strains. However, there were no significant differences in mec complex distribution between countries (data not shown).

ccr complex typing.

As many as 120 ccr complexes were detected in 86 out of 96 MR-CoNS strains (90%), including 1 ccrAB1, 52 ccrAB2, 10 ccrAB3, 7 ccrAB4, and 50 ccrC complexes (mean, 1.4 ccr complexes per strain; 34% of the strains had more than one complex). Four different combinations of two ccr complexes were observed (ccrAB2 and ccrAB3, ccrAB2 and ccrAB4, ccrAB3 and ccrC, and ccrAB4 and ccrC), and two combinations of three complexes were observed (ccrAB2, ccrAB4, and ccrC and ccrAB3, ccrAB4, and ccrC) (Fig. 1). ccrAB2 was significantly more prevalent in S. epidermidis than in S. haemolyticus (51/75 and 1/18, respectively; P < 0.0001). There were no significant between-country differences in ccr complex distribution (data not shown). No ccr gene was detected with the primers used in 10 (10%) of the confirmed positive mecA strains.

FIG. 1.

Migration of PCR products obtained by the ccr typing by M-PCR used in this study (30) for the strains harboring three distinct ccr allotypes. Strains A186-1, A186-3, and A186-4 harbored three ccr allotypes: ccrAB3, ccrAB4, and ccrC. Strains A201-4, BB66-2, M327-2, and C327-1 harbored ccrAB4, ccrAB2, and ccrC. All strains were mecA positive. MMW, mass molecular weight (1 kb plus DNA ladder; Invitrogen).

SCCmec typing.

Only 54 (56%) of the 96 strains could be assigned to known SCCmec types (Table 3), including 28 to type IV (27 S. epidermidis and 1 S. haemolyticus), 25 to type V (13 S. epidermidis and 12 S. haemolyticus), and 1 to type III (S. epidermidis). The remaining 42 strains (44%) had mec-ccr combinations that did not fit into the current classification scheme. They included 28 strains (29%) with more than one ccr complex (21 of them [20.8%] with two complexes and 7 [7.3%] with three complexes), 10 (10%) with untypeable ccr, 2 (2%) with already observed but not yet assigned mec-single-ccr combinations (class A mec and ccrAB1 and class A mec and ccrC [17, 35]), and 2 (2%) new class B mec and ccrC combinations.

TABLE 3.

Types of SCCmec according to Ito et al. (22)

mec complex class	ccr complex(es)	SCCmec type	Species (no. of isolates)	Country (no. of isolates)
A	AB1	NT	S. cohnii (1)	Algeria (1)
A	C	NT	S. epidermidis (1)	Mali (1)
A	AB2, AB3	NT	S. epidermidis (1)	Cambodia (1)
A	AB3, C	III	S. epidermidis (1)	Mali (1)
A	AB3, AB4, C	NT	S. epidermidis (3)	Algeria (3)
A	NTa	NT	S. epidermidis (2), S. haemolyticus (1)	Algeria (3)
B	AB2	IV	S. epidermidis (27), S. haemolyticus (1)	Algeria (11), Cambodia (10), Moldova (4), Mali (3)
B	C	NT	S. epidermidis (1), S. hominis (1)	Algeria (1), Mali (1)
B	AB2, AB3	NT	S. epidermidis (6)	Cambodia (5), Moldova (1)
B	AB2, C	NT	S. epidermidis (5)	Mali (3), Cambodia (1), Moldova (1)
B	AB2, AB4, C	NT	S. epidermidis (3)	Algeria (1), Cambodia (1), Moldova (1)
B	NT	NT	S. epidermidis (2), S. haemolyticus (1)	Algeria (3)
C1	NT	NT	S. haemolyticus (4)	Algeria (2), Cambodia (1), Mali (1)
C2	C	V	S. epidermidis (9), S. haemolyticus (4)	Cambodia (7), Mali (3), Moldova (2), Algeria (1)
C2	AB2, C	NT	S. epidermidis (8)	Algeria (5), Moldova (2), Cambodia (1)
C2	AB4, C	NT	S. epidermidis (1)	Algeria (1)
C2	C, C	NT	S. haemolyticus (8), S. epidermidis (4)	Cambodia (5), Mali (5), Algeria (1), Moldova (1)
C2	AB2, AB4, C	NT	S. epidermidis (1)	Cambodia (1)

^aNT, not typeable.

ccr allotype positions relative to mecA.

The following strains were tested for ccr allotype positions: A186-1 (class A mec complex; ccrAB3, ccrAB4, and ccrC), M327-2 (class B mec complex; ccrAB2, ccrAB4, and ccrC), and C327-1 (class C2 mec complex; ccrAB2, ccrAB4, and ccrC). The full positioning of ccr allotypes in relation to mecA could be determined only in strain A186-1 (Fig. 2). A fragment amplified between primers α3 and mA2 was around ∼20 kb, so that the deduced distance between the end of ccrB3 and the beginning of mecA was ∼16 kb, with ccrAB3 being located upstream of mecA. A fragment amplified between primers mA1 and γF was ∼10 kb, so that ccrC was deduced to be located ∼8 kb downstream from mecA. A fragment amplified between mA1 and β4U was ∼20 kb, so that the ccrAB4 complex was deduced to be located ∼17 kb downstream of mecA. In strain M327-2, ccrAB2 was deduced to be located ∼5 kb upstream from mecA, as described for type IV SCCmec (32) and ccrC, ∼8 kb downstream from mecA. ccrAB4 could not be located. In strain C327-1, ccrAB2 and ccrAB4 could not be located either, but surprisingly, we found two copies of ccrC located ∼8 kb upstream from mecA (as described for type V SCCmec [24]) and ∼6 kb downstream from mecA, respectively. Therefore, to assess the proportion of MR-CoNS strains harboring several copies of ccrC, we tested the 25 type V SCCmec strains harboring MR-CoNS in our collection and found that 9 had ccrC copies ∼8 kb upstream from mecA, 2 had ccrC copies ∼6 kb downstream from mecA, and 12 had 2 ccrC copies both ∼8 kb upstream and ∼6 kb downstream of mecA. We were unable to amplify the remaining two strains.

FIG. 2.

Schematic representation of structural variations in type V SCCmec deduced from long-range PCR results. (A) Strain A186-1, harboring ccrAB3 ∼16 kb upstream from mecA, ccrC ∼8 kb downstream from mecA, and ccrAB4 ∼17 kb downstream from mecA; (B) ccrC located ∼8 kb upstream from mecA, as described for type V SCCmec (24); (C) ccrC located ∼6 kb downstream from mecA; and (D) two ccrC allotypes located ∼8 kb upstream and ∼6 kb downstream from mecA.

DISCUSSION

As far as we know, the present study is the first multicountry report on the epidemiology and SCCmec characterization of nasal MR-CoNS in outpatients. Rates of MR-CoNS carriage ranged from 11 to 31%, with significant variations between the four countries studied. However, the species distribution, mostly limited to S. epidermidis and S. haemolyticus, the most prevalent species colonizing the human nares (4, 8), did not change significantly with the geographical origin of the subjects. Our results showed that nasal MR-CoNS in the community may constitute a diversified reservoir of resistance genes not only for MR but also for many other antibiotics, as shown by the fact that many other resistance traits were present in MR-CoNS.

We did not find any difference between S. epidermidis and S. haemolyticus in terms of coresistances, and these two species were homogeneously distributed among the four countries. In pathogens such as Streptococcus pneumoniae, country-to-country variations in resistance rates are closely related to overall antibiotic use (6). Our results suggest that this also applies to commensals such as CoNS. However, since the extent of antibiotic use is not known in the countries studied here, this hypothesis could not be tested.

The variations we observed also might be due to differences in the populations of patients attending each hospital. Although the male/female ratios of patients were similar among the four countries, their mean ages varied significantly. As strictly the same sampling and analysis methods were used in each country, we are confident that the differences and similarities observed between them were meaningful. The study was designed to ensure that the patients included were representative of community populations, as they were hospitalized under emergency conditions and sampled very soon afterwards to keep the risk of the hospital acquisition of resistant strains to a minimum. In addition, the local investigator in each country had been trained in the main investigator's laboratory to guarantee as much homogeneity as possible and to enable comparisons. Nevertheless, we cannot exclude the possibility that different recruitment biases existed between countries, thus explaining some of the differences observed. Also, we cannot exclude that some MR-CoNS strains had been acquired during a previous healthcare stay.

The structural biodiversity of the SCCmec regions was striking. We found, mostly in S. epidermidis, 15 mec-ccr combinations, some of which had already been observed in this species (17, 35) but not in S. aureus. We also found that 34 MR-CoNS strains (35.4%) harbored two distinct ccr allotypes, including 12 with two copies of ccrC. In addition, seven (7.3%) strains had three ccr allotypes. S. aureus strains with multiple ccr genes have seldom been described (5, 12, 19, 22), possibly because the methods used to type MR S. aureus (34, 37, 51) could not detect ccr duplication. In MR-CoNS strains, recent studies (10, 17) showed that multiple ccr-carrying strains were not infrequent; nevertheless, their prevalence was not evaluated, and to the best of our knowledge it was assessed here for the first time.

The presence of multiple ccr genes in some MR-CoNS strains suggested the presence of non-mec SCC elements in addition to SCCmec. Composite SCC already have been described; for instance, type III SCCmec was found to carry what is known as an SCC mercury element driven by the ccrC3 allotype (22). A composite SCCmec, type V(T), harboring ccrAB2 and ccrC in community-acquired MR S. aureus (5) has been described, as well as a new SCCmec also harboring ccrAB2 and ccrC together with structures similar to those found in various known SCC elements (19). Our results suggest that SCC structures are composites, because SCCmec results from intra- and interspecies SCC exchanges probably mediated by the expression of ccr genes.

Long-range PCR experiments disclosed further information on the SCCmec backbone. The three strains tested carried partly structured SCCmec elements like SCCmec types III and IV. In strain A186-1, ccrAB3 was located ∼16 kb upstream from mecA, as was the case for type III SCCmec (22), and in strain M327-2, ccrAB2 was located ∼5 kb upstream from mecA, as was the case for all type IV SCCmec sequenced to date. Similarities between SCC elements found in S. epidermidis and S. aureus had been observed already (15, 49) and support the hypothesis of SCC transfers between these species.

Furthermore, we found that ccrAB2 complexes were more prevalent in S. epidermidis, and ccrC complexes were more prevalent in S. haemolyticus. Compared to ccr complexes, mec complexes were much less diverse, but we observed some species specificity, for instance, class C mec complexes (C1 and C2) predominated in S. haemolyticus, whereas class B mec complexes predominated in S. epidermidis. Class B mec-ccrAB2 (i.e., type IV SCCmec) complexes were found preferentially in S. epidermidis, and class C2 mec-ccrC (i.e., type V SCCmec) complexes were found in S. haemolyticus. Thus, S. epidermidis and S. haemolyticus appeared to be major reservoirs of type IV and V SCCmec, respectively, whatever the country.

We found a high prevalence of new combinations of previously known mec complexes and ccr allotypes that formed SCCmec structures that were not typeable by the current classification scheme (22). The structural diversity of SCCmec also was recently described by others (12, 17). Since SCCmec was first described (26) and its classification introduced (22), as many as 10 mec complexes (19, 22, 25, 28, 43, 44) and six ccr allotypes (9, 10, 22, 24, 38) have been reported from staphylococci. However, our observation in the present study, that one mec complex was combined with one, two, or three different ccr allotypes, shows that mec and ccr combinations are much more diverse than was previously thought. In the future, descriptions of new types of SCCmec, in both S. aureus and CoNS, should include both quantitative data (e.g., the number of ccr copies) and qualitative data on their mec and ccr content. Our results underscore the importance, and also the complexity, of CoNS as a reservoir for MR genes.