Evidence for clonal evolution among highly polymorphic genes in methicillin-resistant Staphylococcus aureus

Abstract

The evolution of Staphylococcus aureus has been described as predominantly clonal, based on evidence from seven housekeeping genes. We aimed to test if this was also true for more polymorphic genes. In a collection of 60 isolates including major European epidemic methicillin-resistant S. aureus (MRSA) and sporadic MRSA strains, we compared the partial gene sequences of seven housekeeping genes (arcC, aroE, glpF, gmk, pta, tpi, and yqiL), six core adhesion genes (present in all strains) (clfA, clfB, fnbA, map, sdrC, and spa), and four accessory adhesion genes (not present in all strains) (ebpS, fnbB, sdrD, and sdrE). Nucleotide diversity of adhesion genes was 2- to 10-fold higher than genes used for multilocus sequence typing. All genes showed evidence for purifying selection with a weakly reduced level among accessory adhesion genes. Among these highly variable genes, there was no evidence for a difference in molecular evolution between epidemic and sporadic strains. Gene trees constructed from concatenated sequences of housekeeping, core adhesion, and accessory adhesion genes were highly congruent, indicating clonality, despite some evidence for homologous exchange. Further evidence for clonality was found with an overall positive correlation of allelic and nucleotidic divergence for both seven housekeeping genes and six core adhesion genes. However, for small allelic differences that fit the demarcations of clonal complexes (CCs) there was no such correlation, suggesting that recombination occurred. Therefore, despite an overall clonal population structure, recombination between related isolates within CCs might have contributed to S. aureus evolution.

FIG. 1.

Purifying selection. The aligned sequence data of each gene were taken to calculate the proportion of synonymous (ds) and nonsynonymous (dn) substitutions. The plotted values of each gene are shown. The ratio of dn/ds indicates purifying selection (positive Darwinian selection) if values are <1, diversifying selection if values are >1, and balancing selection or neutral evolution if values are close to 1. Three classes of genes are represented: housekeeping genes arcC, aroE, glpF, gmk, pta, tpi, and yqiL (circles); core adhesion genes (triangles); and accessory adhesion genes (diamonds). Gene names of the latter two classes are shown in the graph. Data points of all genes are located in the area of purifying selection. The dotted line shows the regression line calculated on the basis of all data points (R² = 0.9151).

FIG. 2.

Genetic relationship between isolates. The sequences obtained from each isolate for each gene were taken to produce alignments without gaps. We then concatenated separately the gene sequences to produce data sets for seven MLST housekeeping genes (A), six core adhesion genes (B), and four accessory adhesion genes (C). Maximum parsimony trees were constructed for all three data sets with MEGA 3.0. For the data shown in panels A and B, the branch lengths were also calculated and their scales are indicated at the lower left of each panel. Bootstrap values were calculated for 1,000 repetitions and are indicated at branches. In addition, STs were determined for all isolates based on housekeeping genes using the MLST database, and these were used to assign each ST to published CCs. CCs are indicated in all three panels. We observed that all groups of CCs were resolved in each concatenated gene tree.

FIG. 3.

Sequence diversity versus allelic diversity. Two data sets were analyzed separately: MLST housekeeping genes (A) and core adhesion genes (B). They comprised 24 and 34 STs, respectively. Pairwise comparisons of all STs were performed. They were grouped according to differences at one, two, three, four, five, six, and seven loci (up to six for core adhesion genes). Within each group, the number of nucleotide differences was then calculated for each locus separately; the average was afterwards determined across loci. In this way, a graph was produced where each group (x axis) was plotted against the average number of nucleotide differences across loci (y axis). The calculations were performed using the BLUND program that was kindly provided by D. A. Robinson. We observed a positive trend for MLST housekeeping genes, indicating that diversification was mainly due to accumulation of mutations. There was also a positive trend from three to six different loci among core adhesion genes. However, there was no positive correlation for up to three different loci among core adhesion genes; this could be explained by recombination.