ABSTRACT
Cluster analysis of the SmaI patterns, generated by pulsed-field gel electrophoresis, of 44 methicillin-resistant (MRSA) and 118 methicillin-sensitive (MSSA) Staphylococcus aureus strains isolated in various French hospitals and 61 MRSA and 48 MSSA strains from 20 other countries revealed 20 genomic groups distributed into four distantly related phylogenic branches. Eighty-three of the 105 MRSA strains (79%) were clustered in the six genomic groups of phylogenic branch I; and 154 of the 166 MSSA strains (92.8%) were clustered in the 14 genomic groups of phylogenic branches II, III, and IV. Agreement between genomic group and two other markers, esterase type and phage group, was obtained, emphasizing the clonal structure of the population. The genomic groups were delineated by esterase type. The distribution of the strains within the genomic groups was independent of their geographical origin; French strains were clustered with strains from other countries. The three types of the staphylococcal cassette chromosome mec (SCCmec) complex were distributed according to genomic groups. Most of the time, type I and type II SCCmec complexes were found in the MRSA strains belonging to the same genomic groups. In contrast, the type III SCCmec complex was specific to the MRSA strains belonging to the three genomic groups characterized by a common esterase type.
Staphylococcus aureus is a major human pathogen responsible for a wide spectrum of diseases. Soon after methicillin was introduced into clinical use, methicillin-resistant S. aureus (MRSA) strains were reported. Since then, outbreaks of MRSA have been recorded worldwide (1, 9, 13) and the control of MRSA has become a significant problem in hospitals (15, 26).
The methicillin resistance gene (mecA) is carried on a mobile genetic element, the staphylococcal cassette chromosome mec (SCCmec), which differs in size and genetic composition among strains (17, 18). For movement, SCCmec carries two specific genes (ccrA and ccrB) which encode recombinases. Three SCCmec complex elements, carrying a characteristic combination of the ccr and mec gene complex, have been identified in hospital-acquired MRSA strains (17). SCCmec is integrated in the chromosome of MRSA at a unique site (attBscc), which is found inside an open reading frame (orfX) of unknown function.
Knowledge of the underlying genetic structure is important in understanding the epidemiological relationship between strains. Therefore, a number of epidemiological methods have been used to identify and differentiate strains of S. aureus, including phage typing, enzyme electrophoretic typing, pulsed-field gel electrophoresis (PFGE), randomly amplified polymorphic DNA analysis-PCR, ribotyping, and multilocus sequence typing (11, 14, 16, 19, 23, 27).
A high degree of diversity has been observed among methicillin-sensitive S. aureus (MSSA) strains (3, 6, 10). However, Schlichting et al. (27) noticed that epidemiologically unrelated MSSA strains with similar PFGE patterns were found in different geographical regions, suggesting that some genotypes are ubiquitous. Multilocus enzyme electrophoresis demonstrated the presence of clones in natural populations of S. aureus and, in particular, the presence of a clone responsible for most cases of toxic shock syndrome (12, 22).
In contrast to MSSA, several studies showed that MRSA strains were represented by a limited number of genotypes (3, 10, 16, 23). There has been considerable speculation about the origin and evolution of the MRSA strains (19). According to Fitzgerald et al. (12), the mec gene has been horizontally transferred into distinct S. aureus chromosomal backgrounds, demonstrating that methicillin-resistant strains have evolved several times independently.
In this study, the relationships between the genetic structures of MRSA and MSSA strains isolated from various hospitals in France and of MRSA and MSSA strains from international sources are described. We used PFGE, as it is an interesting tool not only for strain identification but also for estimation of interstrain genetic relatedness (7, 13, 21, 28). The Dice similarity coefficient and Ward's algorithm (25, 30) were used for cluster analysis to define the degree of DNA relatedness between the strains. The correlation of the genomic group with two other markers, the esterase electrophoretic type and the phage type, was established. We also analyzed the distribution of the type of SCCmec complex among MRSA strains.
MATERIALS AND METHODS
Bacterial strains.A total of 271 strains of S. aureus originating from diverse geographical origins were examined (Table 1). Two hundred twenty-six of the strains came from collections that have previously been analyzed by antibiotic typing, esterase electrophoretic typing, and phage typing (3-7, 27). Forty-five new strains were examined. Twenty-five strains isolated from bronchopulmonary infections were kindly provided by N. Hoiby (University of Copenhagen, Copenhagen, Denmark), 11 strains were obtained from the collection of M. Fournier (Institut Pasteur, Paris, France), and 6 strains were isolated at our hospital between 1991 and 1998. Three reference strains were also included. The strains were selected according to their susceptibilities to methicillin and according to their geographical origins to maximize diversity. The strains were also selected to represent the most frequently encountered esterase electrophoretic types, as described in previous studies (4, 6). One hundred sixty-two strains were recovered from French hospitals between 1980 and 1998, and 109 strains were recovered from 20 other countries. Except for strains of esterase type 1, each esterase type included strains from 2 to 16 countries (Table 1); esterase type 1 included only strains from France. MRSA strains were represented by 44 French strains and 61 strains from other countries. Resistance to methicillin was confirmed by detection of mecA by PCR (24).
Repartition of the strains studied according to their geographical origin, esterase type, and resistance to methicillin
Esterase electrophoretic typing and phage typing.The electrophoretic mobility patterns of three different esterases were investigated as described previously (3). Phage typing was performed at the routine test dilution (RTD) and 100 times the RTD with the international basic set at the Centre National de Reference des Staphylocoques, Institut Pasteur, Paris, France (N. El Solh).
PFGE of macrorestriction fragments and analysis of DNA relatedness.DNA fragment patterns were obtained as described previously (27) after digestion of genomic DNA with the restriction endonuclease SmaI and subsequent PFGE.
The DNA patterns obtained by PFGE were analyzed with Gel Compar software (Applied Maths, Kortrijk, Belgium). A similarity matrix was created by using the band-based Dice similarity coefficient, and Ward's algorithm was used to cluster the strains (25, 30).
SCCmec complex.The SCCmec complex type was determined by using PCR to detect a sequence overlapping the right SCCmec-chromosome junction (18). The primers used to detect type I and type II SCCmec complexes were designed on the basis of the type II SCCmec sequence (GenBank accession no. D86934) They were MDR1 (5′-TGCGTAAAATCTTCCAG-3′) and OR3 (5′-GTCTTACAACGCAGTAACTA-3′). An amplification fragment of 877 bp was expected for strains with complex type I, and a fragment of 979 bp was expected for strains with complex type II. The primers used to detect strains with the type III SCCmec complex were designed on the basis of the type III SCCmec sequence (GenBank accession no. AB047089). They were MRTc3 (5′-CCGTCGA TATCAATTGCTC-3′) and OR3. An amplification fragment of 562 bp was expected. The PCR mixture was as described previously (24). The following cycling conditions were used: 10 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min, followed by 25 cycles of 94°C for 30 s, 56°C for 30 s, and 72°C for 1 min.
RESULTS
PFGE pattern analysis.Among the 271 strains studied, 226 PFGE patterns that differed by at least one band were identified. One hundred thirty-eight different patterns were obtained for MSSA strains, and 88 different patterns were obtained for MRSA strains. Eleven patterns were represented by MSSA strains recovered from more than one country, and two patterns were represented by MRSA strains recovered from more than one country. For example, esterase type 11 MSSA strains from Denmark and France had the same SmaI pattern (data not shown).
Cluster analysis.After comparison of the Dice coefficients, Ward's algorithm was used to discern groups of isolates among the strains that have a defined level of similarity and to display relatedness between the groups. When clusters with more than 90% relatedness were considered, 20 genomic groups were defined. A dendrogram of all the genomic groups was constructed on the basis of the levels of similarity (Fig. 1). Four major branches were identified. These branches were very divergent, with only 15 to 45% of relatedness. Branch I included six groups, branch II included seven groups, branch III included only two groups, and branch IV included five groups.
Dendrogram showing the estimated genetic relationships among 271 S. aureus strains clustered in 20 genomic groups. The dendrogram was generated by applying the Ward's algorithm to SmaI patterns. R, resistant; S, sensitive.
Correlation between genomic group and esterase type.In most cases, a good correlation was obtained between genomic group and esterase type: strains within a given genomic group tended to be of the same esterase type (Table 2). This was true of all groups, except groups 1 and 2, which contained strains of two esterase types (type 5 and type 6). Strains of a given esterase type could be distributed into two or three groups. For example, strains of esterase type 11 were distributed in genomic groups 6a and 6b, but these groups are closely related and can be considered subgroups of a major group. This was also the case for strains of esterase type 7, which were distributed in groups 12a and 12b. In contrast, strains of esterase type 14 were distributed, on the one hand, in the closely related genomic groups 5a and 5b and, on the other, in a very divergent genomic group, group 17, which shares only 15% relatedness with groups 5a and 5b.
Correlation between esterase type, phage type, and genomic group
Correlation between genomic group and phage type.When the strains were typeable, a correlation was observed between the genomic group and the phage type (Table 2). Most of the MRSA strains (99 of 104) were lysed by phages of lytic group III. These strains belonged to genomic groups 1 to 6. Among the MSSA strains, phages of lytic group III preferentially lysed strains of genomic group 11. Strains of genomic groups 7, 9, 12, and 16 were lysed only by phages of lytic group II. Phages of lytic group I seemed to lyse preferentially strains of genomic group 17, and phages of lytic group V seemed to lyse preferentially strains of genomic group 14. Phage 95 lysed only strains of genomic group 13. Most of the strains from genomic groups 8, 10, and 15 were nontypeable.
Distribution of MRSA strains.Most of the MRSA strains (83 of 105 strains) clustered in the six genomic groups of branch I (groups 1, 2, 3, 4, 5a, and 5b) (Fig. 1). Thus, these groups were called MRSA genomic groups. The groups were related by at least 68%. However, genomic groups 1, 2, and 3 also included 12 methicillin-sensitive strains, which did not carry the mecA gene and which had none of the additional antibiotic resistance markers usually associated with methicillin resistance. These 12 strains were fully sensitive to aminoglycosides, tetracyclines, sulfonamides, fluoroquinolones, and macrolides. In contrast, most of the methicillin-sensitive strains (154 of 166) were clustered in the 14 other genomic groups included in branches II, III, and IV (Fig. 1). Thus, these groups were called MSSA genomic groups. Four of these MSSA genomic groups included 22 MRSA strains. Sixteen of these MRSA strains were found in branch II, 11 were found in genomic group 6a, 5 were found in genomic group 6b, 5 were found in branch IV, 2 were found in genomic group 13, and 4 were found in genomic group 17. The MSSA genomic groups and the MRSA genomic groups were distantly related, as the coefficients of relatedness did not exceed 45%.
SCCmec complex analysis.The SCCmec complex types were determined for 91 of the 105 MRSA strains. Type I and type II SCCmec complexes share a common sequence at their right ends, next to the junction with the chromosome. The type II SCCmec complex differs from the type I SCCmec complex by an additional 102 bp (17, 18). The type II SCCmec complex includes SCCmec type II and type IV (20). The right sequence of the type III SCCmec complex is different from those of the type I and type II SCCmec complexes (18).
Primers specific for the type III SCCmec complex gave an amplification product of about 562 bp for 37 strains (Table 3). All of these strains were of the same esterase type (type 14) and were distributed in the three corresponding genomic groups (groups 5a, 5b, and 17). Primers specific for the type I and type II SCCmec complexes gave an amplification product of about 877 bp, corresponding to the type I SCCmec complex, for 33 strains and an amplification product of about 979 bp, corresponding to the type II SCCmec complex, for 21 strains. These strains belonged to genomic groups 1, 2, 3, 4, 6, and 13. Strains of esterase type 5 and type 6 from genomic groups 1, 2, 3, and 4 contained either the type I or the type II SCCmec complex. Genomic groups 2 and 4 more frequently contained the type I SCCmec complex (Table 3). Strains from genomic groups 6a and 6b (esterase type 11) could be differentiated by their SCCmec complex types: all group 6a strains were of type I, and all group 6b strains were of type II.
Distribution of the SCCmec complex among the genomic groups containing MRSA strains
Comparison of the French strains with strains from other countries.When the esterase type 1 strains, which clustered in genomic group 15, were excluded (because only French strains of this esterase type were included in our collection), strains from France were found in all of the genomic groups, along with strains from 1 to 11 other countries. Strains from different geographical sources displayed similar or closely related patterns, differing by one to four bands, although there was no evidence of contact (Fig. 2).
PFGE of SmaI-digested DNA from S. aureus strains of genomic group 13. Lanes 1, 8, 15, and 22, control strains; lanes 3, 4, 6, 7, 9 to 13, 16, 17, 21, and 23, strains from France; lanes 2, 5, and 18 to 20, strains from Denmark; lane 14, a strain from Germany; lane 24, a strain from England. The numbers on the left are in kilobases.
The French MRSA strains were recovered in all the genomic groups that contained MRSA strains. However, 11 of the 19 French strains of esterase type 6, which was the most common type among the French MRSA strains, were clustered in genomic group 4, along with three strains from other countries (Ireland, England, and the United States). These strains originated from eight French hospitals in three distant areas and were isolated between 1984 and 1994. Two of the other French strains of esterase type 6 were found in genomic group 1, and six were found in genomic group 2. Strains of esterase type 14, which was the second largest group of French MRSA strains by esterase type, were found in genomic group 5a (4 strains), genomic group 5b (9 strains), and the distantly related genomic group 17 (2 strains).
DISCUSSION
In this study, we compared the genomic diversity of French strains of MSSA and MRSA with that of strains isolated in 21 countries. In previous studies, typing of S. aureus isolates recovered from patients with cystic fibrosis showed concordance between the genome type obtained by PFGE and the types obtained by other typing methods such as esterase typing and phage typing (7, 27). As esterase electrophoretic typing explores only a small part of the genome, it was less powerful than PFGE for distinguishing between strains of S. aureus. However, genotypes of the same esterase type were closely related and clustered in the same phylogenic branch (7). In contrast to PFGE, which detects events that occur randomly during chronic or recurrent infections, esterase electrophoretic typing detects variations that accumulate slowly and that therefore could represent major phylogenic branches. To analyze the clonal structure of S. aureus further, a larger collection of strains from various genetic backgrounds, as defined by esterase electrophoretic typing, and from diverse geographical origins was studied. To obtain significant results, only strains of the most common esterase types were analyzed. PFGE has been used to track the genetic development of natural S. aureus populations (7, 21, 28). However, when strains from various origins are compared, the genetic variations indexed by PFGE led to considerable differences in SmaI patterns. Therefore, certain investigators do not consider this technique to be well suited for global epidemiological studies (11, 17). To define strain relatedness accurately, computer analysis with Ward's algorithm was chosen, because it minimizes overall variations. It discerned groups of strains that had some level of similarity and that were related to some extent. Twenty groups of strains were defined. In most cases, strains from the same genomic group or from a major group that included two subgroups were represented by a unique esterase type, which identified the lineage. This was particularly true for the MSSA genomic groups. Conversely, strains of different esterase types (types 6 and 5) were found in two MRSA genomic groups (groups 1 and 2). It is noteworthy that these MRSA genomic groups were closely related to each other (90% similarity). Only strains of esterase type 14 were found in two very divergent groups, one with only MRSA strains and one with MSSA strains and a few MRSA strains. Thus, in some cases, strains of the same esterase type can be divergent. Either these two divergent lineages shared this trait or the technical conditions used were not able to reveal differences in the electrophoretic mobilities of the three enzymes tested.
Bacteriophage typing used to be the reference method for epidemiological typing for S. aureus, but because of problems with reproducibility and the nontypeability of strains, it was replaced by molecular typing methods, notably, PFGE (2, 29). As for esterase type, a correlation had previously been observed with PFGE type (7, 27). Our results revealed a relationship between the genetic background of the strains and the lytic attack of the phages belonging to a given phage group. This was particularly noticeable for phage 95, which lysed only strains of genomic group 13. Group V phages (phages 94 and 95) almost exclusively lysed strains of genomic group 14. Witte (32) previously described strains lysed by group V phages as a complex containing strains that are possibly evolutionarily related clones that differ from other S. aureus strains by independent characteristics. In our study, although they presented a certain degree of diversity in their SmaI patterns, most of the strains lysed by group V phages were closely related and very divergent from the strains in other genomic groups. As observed many times during the characterization of epidemic MRSA strains, most of the MRSA strains tested were lysed by group III phages (8, 13). Epidemic MRSA strains that are lysed by phage groups other than group III appear to be very rare. Only one report has described an epidemic strain of phage group I (31). In our study, MRSA strains of esterase type 14 and genomic group 17 were lysed by group I phages.
MSSA strains were clustered in 17 different genomic groups. Although some of these groups were related, such as groups 7 to 11 and groups 15 and 16, the degree of interrelatedness was often very low (only 15%). These results are in agreement with previous estimates of genetic diversity (3, 6, 23); MSSA strains are polymorphic but are distributed in a limited number of major phylogenic branches. In contrast, MRSA strains were found to be clustered in a smaller number of genomic groups. Most of these groups, the MRSA groups, were closely related and constituted the major phylogenic branch, branch I. Only a few groups included in what we called MSSA groups were distantly related to the latter groups and to each other. Among these MSSA groups, only genomic group 6a included a large number of MRSA strains. MRSA strains appeared to be uncommon in the other MSSA groups. Several studies have provided evidence that MRSA strains evolved in a relatively small number of lineages that are clonally related and that some MRSA strains are present in distant lineages. This led some investigators to the conclusion that the presence of the mec gene in such widely divergent lineages is certainly the consequence of the horizontal transfer of the mec region into distantly related S. aureus strains (11, 12).
The distribution of the SCCmec complex types among the MRSA strains studied was interesting. The type I and type II SCCmec complexes, on the one hand, and the type III SCCmec complex, on the other, were found among strains from distinct genetic backgrounds. The type III SCCmec complex was recovered exclusively from esterase type 14 strains clustered in major genomic group 5, and in distantly related genomic group 17. No other type of SCCmec complex was found in these groups of strains. Although genomic groups 5 and 17 are distantly related according to the PFGE results, they seem to have a common trait linked to the acquisition of the type III SCCmec complex element. Hiramatsu et al. (17) found that the type III SCCmec complex is associated with a cluster of strains defined by their ribotype, called clonotype III-A. Some of the strains of this clonotype were included in our study, such as the strains from Portugal, Hong Kong, Saudi Arabia, Yugoslavia, and Norway, and were clustered in major genomic group 5. In contrast, the type I and type II SCCmec complexes were found in strains belonging to the same genomic group, such as the strains of genomic groups 1, 2, 3, and 4. Enright et al. (11) observed different SCCmec complex types among MRSA isolates of the same multilocus sequence type and suggested that MRSA clones emerged on more than one occasion in the same genetic background. However, within genomic group 6 (esterase type 11), the type of SCCmec complex was clearly associated with the subgroup: strains from group 6a were of type I, and strains from group 6b were of type II.
S. aureus strains from France did not display specific genomic traits when they were compared to strains from other countries: the strains clustered in genomic groups with strains from other countries. As observed before (7, 27), MRSA and MSSA strains from different countries or from different areas of France were found to have identical or similar genome types, confirming that the degree of geographical spread is not correlated with the extent of genetic change detected by PFGE. This feature, which may be due to the slow genetic development of S. aureus, complicates epidemiological studies involving the analysis of hospital isolates.
In conclusion, cluster analysis of the PFGE patterns of populations of S. aureus strains and correlation of the genomic group with the esterase type and the phage type delineated 20 clonal groups. The distribution of the strains among the groups was independent of the geographical origin. The type III SCCmec complex was strongly associated with three genomic groups characterized by the same esterase type (type 14). Identification of well-defined groups of strains with a common genetic background gives a basis for further understanding of the distribution of virulence genes or the dissemination of particular clones in the hospital environment.
ACKNOWLEDGMENTS
We gratefully acknowledge N. El Solh and A. Morvan from the Centre de Réference des Staphylocoques (Institut Pasteur) for help with phage typing. We thank N. Hoiby (University of Copenhagen) and M. Fournier (Institut Pasteur) for providing the strains, P. Ferron for technical assistance, and P. Pame for help typing the manuscript.
FOOTNOTES
- Received 17 January 2003.
- Returned for modification 6 March 2003.
- Accepted 7 April 2003.
- American Society for Microbiology
