Recently, MRSA infections have increasingly been reported among groups of patients with no apparent connection to hospitals (4). Those strains, designated community-acquired MRSA (C-MRSA) strains, have been reported in various countries such as Australia (16, 18), New Zealand (19), the United Kingdom (20), Canada (5), and the United States (6, 8). The death of four children caused by C-MRSA strains has alerted us to the threat of latent dissemination of highly virulent C-MRSA strains in the community in the United States (3).

In contrast to hospital-acquired MRSA (H-MRSA), C-MRSA is characteristically susceptible to many antibiotics (3, 21), but it remains unclear whether C-MRSA is a descendant of H-MRSA or was born and evolved independently of the hospital environment (4). In order to understand the evolutionary relationship between C-MRSA and H-MRSA, we determined the entire nucleotide sequences of the SCCmec elements integrated into the chromosomes of two C-MRSA clinical strains. Strain CA05 (JCSC1968) was isolated from the joint fluid of a patient with septic arthritis and osteomyelitis, and strain 8/6-3P (JCSC1978) was isolated from the perineum of another patient (10).

We amplified DNAs encompassing the entire SCCmec sequence by long-range PCR with several sets of primers, as follows. The region from the left extremity to the ccr genes (L-C region) was covered by primer sets α5 and cLs1 (CA05) or CL2b (8/6-3P) (Fig. ​(Fig.1).1). Primers α6 and mcR8 were used to cover the middle part (from the region just upstream of ccrA to mecR1; the C-M region). Two overlapping primer sets (primers is4 and mA2 and primers mA3 and cR1) were used to cover the right extremities (from IS431mec to orfX; the I-R region) (Fig. ​(Fig.1)1) (11, 12, 14). The rest of the element was amplified and sequenced with primers as described previously (11, 12, 14). The PCR products were purified with a High Pure PCR product purification kit (Roche Diagnostics GmbH, Mannheim, Germany), and their nucleotide sequences were determined as described previously (9).

Figure ​Figure11 illustrates the genomic organizations of the SCCmec elements identified from the two C-MRSA strains in comparison with the three extant types of SCCmec. The SCCmec elements of strains CA05 and 8/6-3P were 24,248 and 20,920 bp, respectively, and were much smaller (34 to 67 kb) than three types of SCCmec elements identified from H-MRSA strains (12). Both elements were found to be integrated at the integration site for SCCmec, attBscc, which is found inside orfX, which has an unknown function and is located near the origin of replication of the S. aureus chromosome (12, 15). The SCCmec elements shared a pair of 15-bp direct repeat sequences: one (DRscc-R) at the right extremities and the other (DRscc-L) on the chromosome region abutting the left termini of the elements (shown as thick arrows in Fig. ​Fig.2).2). Degenerate inverted repeats were also found in the extremities of the two SCCmec elements (shown by thin arrows in Fig. ​Fig.2).2). The authenticity of the SCCmec boundaries was vindicated by using a previously described method (14); i.e., by confirming that the element was precisely cut out from the chromosomes of the two C-MRSA strains, regenerating attBscc in the chromosomes of the cell populations from which it was excised.

The nucleotide sequences of the L-C regions of the two SCCmec elements differed from each other, but otherwise, they shared the same features: type 2 ccr gene complexes, the class B mec gene complexes (IS1272-ΔmecR1-mecA-IS431), and the I-R region that was 99.9% identical to that of type II SCCmec elements (Fig. ​(Fig.1).1). This combination of ccr and mec gene complexes was a novel one. That is, the class B mec gene complex has been found to be in close linkage with the type 1 ccr gene complex in type I SCCmec (12). However, the ccr gene complexes of the novel SCCmec elements were more similar to the type 2 gene complex than to the type 1 ccr gene complex: the amino acid identities between the novel CcrA proteins and the type 2 CcrA (CcrA2) and type 1 CcrA (CcrA1) proteins were about 98 and 75%, respectively; and those between the novel CcrB proteins and the CcrB2 and CcrB1 proteins were 97 to 99 and 72 to 81%, respectively (Table ​(Table1).1). Accordingly, we classified the genes as variants of ccrA2 and ccrB2 by designating them the ccrA2.1 and ccrA2.2 genes and the ccrB2.1 and ccrB2.2 genes, respectively (Table ​(Table1).1). As shown in Table ​Table1,1, the predicted proteins encoded by the open reading frames (ORFs) surrounding the ccrA and ccrB genes also had significantly higher degrees of similarity to those of the type 2 gene complex than to those of the type 1 ccr gene complex. On the basis of this unique combination of the two complexes, we named the novel SCCmec elements type IV SCCmec and assigned the designations subtypes IVa and IVb to the individual elements of CA05 and 8/6-3P, respectively, on the basis of their unique nucleotide sequences in the L-C region (Fig. ​(Fig.11).

The BLAST and MOTIF programs failed to assign any biological functions to the hypothetical ORFs in the L-C regions of type IV SCCmec elements (1, 17). With their simple genetic organizations, neither a virulence factor nor antibiotic resistance other than that encoded by mecA was found to be encoded by the entire regions of the novel SCCmec elements. This lack of genes encoding resistance to non-β-lactam antibiotics in the type IV SCCmec was consistent with the notable characteristic of C-MRSA; i.e., its susceptibility to various antibiotics except β-lactams (3, 6). In fact, as shown in Table ​Table2,2, the two strains were susceptible to all the non-β-lactam antibiotics tested.

The lack of function of type IV SCCmec other than those for the movement of the element (ccr genes) and methicillin resistance (mecA), together with its small size, may lead to the view that the element has gone through evolutionary refinement as a specific carrier of DNA for methicillin resistance. Furthermore, the lack of superfluous functions may make the element more fit as a mobile genetic element for S. aureus in the community than any other type of SCCmec, for the following reasons. A number of ORFs carried by type I and type II SCCmec elements in their long L-C regions are considered unnecessary for the benefit of the host cells that carry the elements (11, 12). Actually, many of them are mutated or partially deleted and do not seem to be active. Moreover, an ORF for type I SCCmec encoding plasmin-sensitive surface protein may even be hazardous to the host cell because it interferes with the fibrinogen- and fibronectin-binding properties of the host cell (23). Although the type III SCCmec has a short L-C region comparable in size to those of type IV SCCmec elements, the size of the element is extremely large (68 kb) because of the accumulation of multiple genes for resistance to various antibiotics and heavy metals (12). The determinants for resistance to multiple antibiotics carried by the previously studied types of SCCmec elements (type II and type III elements) may be suited for the survival of H-MRSA in the hospital environment, where various antibiotics as well as antiseptics provide selective pressure, but their large sizes and potentially hazardous arrays of exogenous genes may not be suited to MRSA strains in the community, where selective advantage would make strains more inclined to have a higher growth rate and to be better able to colonize humans than to have a multidrug resistance phenotype. From this viewpoint, the type IV SCCmec may be one of the fit SCCmec types that can confer β-lactam resistance to community strains of S. aureus without greatly compromising their competitiveness among the natural flora of humans. Future analyses of many community-acquired strains will be required to test if this hypothesis holds true.