Genome Diversification in Staphylococcus aureus: Molecular Evolution of a Highly Variable Chromosomal Region Encoding the Staphylococcal Exotoxin-Like Family of Proteins

RD13 PCR-RFLP analysis.

PCR primers (forward primer, 5′-CAA AGC TAA ACA AGA CGG CTT TGA TG-3′; reverse primer, 5′-TCC GCG CCA ATC TTC TGG AAC-3′) specific for conserved regions flanking RD13 were used to amplify the intervening sequence with the Advantage genomic PCR kit (Clontech, Palo Alto, Calif.) according to the manufacturer's instructions. The PCR products were digested with restriction endonuclease HinP1I at 37°C for 2 h, and the DNA restriction fragments were resolved by electrophoresis in a 1.2% agarose gel.

Analysis of size variation in Vr in RD13.

The size of the RD13 variable region (Vr) in each strain was inferred on the basis of amplification by PCR with primers (forward primer, 5′-ATA GAA CTC GCC TGC TTT TTT ACC-3′; reverse primer, 5′-CCA AAG CCT TTA GGT TCA TCA TAC A-3′) specific for conserved flanking regions (the mod and orf3 genes, respectively) (Fig. ​(Fig.11)

DNA sequence analysis.

Sequence data obtained from both DNA strands with an Applied Biosystems model 3700 automated sequencer were analyzed by DNASTAR (Madison, Wis.). Multiple-sequence alignment of the inferred amino acid sequences was performed with Clustal W version 1.8 (18). Statistics of nucleotide and amino acid content were determined with MEGA version 2.1 (9).

Population and molecular evolutionary genetic analyses.

The Nei-Gojobori method (13) was used to calculate the proportion of synonymous sites in nucleotide sequence data. Phylogenies were constructed with the neighbor-joining algorithm by using MEGA version 2.1 (9). A concatenated sequence consisting of orf2, set1, res, mod, set2, set3, set9, set10, and set11 was used for the analysis of the conserved architecture of RD13. Due to strain-to-strain variation in the content of certain set genes, set4, set5, set6, set7, and set8 were studied independently. To examine variation in the patterns of nucleotide substitution, the proportion of synonymous (p_S) and the proportion of nonsynonymous (p_N) nucleotide substitutions were calculated by sliding-window analysis of 30 codons along each gene with the program PSWIN (15) (PSWIN is available from S. D. Reid [sreid@niaid.nih.gov]). Estimates of the sampling variance of these statistics were made by Monte Carlo simulation or by bootstrapping.

To identify the putative end points of past recombination events, a computer program (MAXCHI) that implements the maximum chi-square method was used (16). Happlot analysis was used to diagram the locations of polymorphic sites.

Real-time reverse transcriptase PCR (TaqMan assays).

PCR primers and probes were designed with the software package Primer Express (Perkin-Elmer) on the basis of the genomic DNA sequence available for S. aureus strain COL (http://www.tigr.org) and purchased from PE Applied Biosytems. Total RNA was isolated from bacteria cultured for 3 h (mid-exponential phase) or 12 h (stationary phase), and reverse transcription and PCRs were performed as described previously (3). To confirm that the era gene encoding the GTP-binding protein Era was constitutively expressed and suitable for use as an internal control, the amount of era mRNA was measured in three separate reverse transcriptase PCR experiments with two independent RNA samples isolated from mid-exponential and stationary-phase cultures of strain COL. The results confirmed that era is expressed at similar levels (0.9 to 1.3 relative expression levels) in both phases of the growth cycle (data not shown).

set gene cloning and production of recombinant SET proteins.

The primers, vectors, and restriction enzymes used for PCR amplification and cloning of the set genes of S. aureus strains COL and 8325 are listed in Table ​Table2.2. The primers were designed on the basis of genome sequences available for S. aureus strains COL (http://www.tigr.org) and 8325 (http://www.genome.ou.edu), and restriction enzyme sites for cloning were included in each primer. PCR products were amplified with genomic DNA isolated from strain COL or strain 8325, digested with the appropriate restriction enzymes, and cloned into vector pET21b or pET15. The cloned genes were sequenced to ensure that spurious mutations had not been introduced, and the plasmids were moved into Escherichia coli BL21 for expression of recombinant proteins.

To assess the production of recombinant proteins, BL21 cells with recombinant plasmids were grown at 37°C for 10 h in 3 ml of Luria-Bertani broth supplemented with 100 mg of ampicillin per liter. Cells were pelleted by centrifugation and suspended in 1× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer at a ratio of 100 μl of buffer per 1 unit of optical density 600 nm per ml. The samples were boiled for 10 min and analyzed by SDS-PAGE.

Purification of recombinant SET protein.

To purify recombinant SET6 protein, BL21 cells with recombinant plasmid were grown overnight at 37°C in Luria-Bertani broth. The culture was treated with 4 volumes of absolute ethanol for 48 h, and the precipitated proteins were collected by centrifugation (500 × g, 10 min). Five liters of total culture volume was combined in a single culture toxin preparation. The precipitate was air dried after centrifugation and resuspended in 100 to 150 ml of pyrogen-free water. The resuspended precipitate was centrifuged at 10,000 × g for 30 min; the concentrated supernatant was removed, placed in dialysis tubing with a 12,000- to 14,000-molecular-weight cutoff (Spectrum Laboratories, Inc., Miami, Fla.), and dialyzed overnight against 4 liters of distilled water. The dialyzed supernatant was subjected to preparative isoelectric focusing. Successive gradients of pH 3.5 to 10 and pH 6 to 8 were used to isolate highly purified recombinant SET6 protein. Final purification of proteins was accomplished with a gel filtration column (Bio-Rad Laboratories, Hercules, Calif.) containing Sephadex G-75 (Sigma, St. Louis, Mo.). Purity was verified by SDS-PAGE in which 10 μg gave a homogenous band of the appropriate molecular weight. The concentration of the purified protein was assessed with the Bradford protein assay (Bio-Rad) (2a), and the protein was lyophilized and stored until used in biological and biochemical assays.

Assay for superantigen activity.

Rabbit splenocytes or human peripheral blood mononuclear cells (PBMCs) were seeded into the wells of a 96-well microtiter plate at a concentration of 2 × 10⁵ cells per well. Serial 10-fold dilutions of SET or toxic shock syndrome toxin 1 (TSST-1; positive control) were added to each well in quadruplicate, starting with 1 μg/well and with dilution to 10⁻⁸ μg/well. The assay results for these dilutions were compared to those for cells incubated in the presence of phosphate-buffered saline alone. The splenocytes were grown at 37°C for 3 days and pulsed with 1 μCi of [³H]thymidine overnight. The cells were harvested the next day, and cell proliferation (incorporation of ³H into DNA) was measured with a scintillation counter (Beckman Instruments, Fullerton, Calif.).

Pyrogenicity and endotoxin enhancement.

American Dutch belted rabbits were injected intravenously with recombinant SET proteins at a maximal dose of 10 μg/kg of body weight per ml. The temperature of each rabbit (n = 3) was measured with a rectal thermometer at 0 and 4 h postinjection. After 4 h, each rabbit was injected intravenously with 10 μg of lipopolysaccharide from Salmonella enterica serovar Typhimurium (1/50 of the 50% lethal dose of endotoxin alone). The lethality of this toxin regimen was assessed over a 48-h period.

Western immunoblot analysis.

Proteins separated by SDS-PAGE were transferred to nitrocellulose membranes (Immobilon-NC; Millipore Corporation) with Towbin transfer buffer with a Trans-Blot SD semidry transfer cell (Bio-Rad Laboratories) at 15 V for 40 min. The membrane was incubated with 10 ml of block solution (5% powdered milk in 150 mM NaCl and 100 mM Tris-HCl, pH 7.4) for 1 h, incubated for 1 h with a 1:500 dilution of human patient serum in block solution, rinsed twice, and washed three times for 15 min each time with 0.1% Tween 20 in phosphate-buffered saline. The membrane was incubated with goat anti-human immunoglobulin G horseradish peroxidase-conjugated secondary antibody (Sigma) for 1 h and rinsed and washed as described above. Immunoreactivity was visualized by enhanced chemiluminescence. The human serum samples were obtained during a countrywide study of invasive episodes of S. aureus infection in Finland in 1999 and 2000. Patients had infections with or without deep infection foci, including abscesses, cellulitis, pneumonia, and purulent arthritis. Sera were obtained 2 to 7 days (i.e., the acute phase) and 20 to 30 days (i.e., the convalescent phase) after the first positive blood culture. Control serum samples were obtained from four female and four male healthy volunteers, including Caucasian, African-American, Hispanic, and Asian individuals (n = 2 each).

PCR-RFLP analysis of RD13.

To analyze structural variation in chromosomal region RD13 among S. aureus strains representing the breadth of diversity within the species, a PCR-RFLP method was developed. We analyzed 44 strains that represent each of the major clonal complexes identified in an MLEE population genetics study of more than 2,000 isolates from diverse infection types, localities, and host species (12) (Table ​(Table1).1). Twenty-seven distinct PCR-RFLP types were identified (Fig. ​(Fig.2;2; Table ​Table1),1), indicating that substantial variation exists at this chromosomal region in natural populations. Each strain had 7 to 12 restriction fragments in total, which were between 0.5 and 4.5 kb in length, and resulted in restriction profiles that were readily distinguishable (Fig. ​(Fig.2).2). Most of the 27 PCR-RFLP types comprised relatively few isolates (range, one to six strains per type) (Table ​(Table1),1), which is indicative of considerable genetic diversity in this region. Multiple PCR-RFLP types were identified among isolates assigned to electrophoretic type (ET) clusters A1, F2, F4, F5, F6, and H1. There was little sharing of PCR-RFLP types between ETs. The exceptions to this were strain MSA 565 (lineage F8/ET 178), which shared PCR-RFLP type 3 with strains MSA 3426 (F2/89), MSA 1006 (F3/91), and MSA 1007 (F3/1006); and strain MSA 1134 (K1/249), which had an identical PCR-RFLP type to strains MSA 820 and MSA 3400 (both F2/91). In addition, MSA 2766 (F5/161) had an identical PCR-RFLP type 32 to MSA 1183 of the I1 cluster. This PCR-RFLP typing method had a very high index of discrimination of 0.962 (8). (An index of 1.0 would indicate that a typing method was able to distinguish each member of a strain population from all other members, whereas an index of 0.0 would indicate that all members of the population were of an identical type.) Taken together, these data clearly indicate a high degree of genetic diversity at this chromosomal region.

Open in a separate window

FIG. 2.

(A) Schematic representation of 34 RD13 PCR-RFLP types identified among 44 S. aureus strains of divergent clonal lineage and 19 strains isolated from patients with invasive disease in Finland; (B) Vr PCR product size variants. Agarose gel electrophoresis of representative PCR products amplified with primers specific for regions flanking the RD13 variable region.

We also analyzed the RD13 PCR-RFLP patterns of 19 S. aureus isolates recovered from patients with invasive infections whose sera were used in Western immunoblot analysis (see below). Among these 19 organisms, 7 had RD13 PCR-RFLP profiles which were identical to type 13 or differed by the presence of one or two restriction sites (types 33 and 34). Two isolates were PCR-RFLP type 16; there were single isolates of type 2, type 6, type 8, and type 15; and five isolates had unique RD13 PCR-RFLP profiles (types 27 to 31). A PCR product could not be generated for one isolate.

PCR analysis of the Vr of RD13.

PCR amplification was used to determine the size of the Vr among the 63 strains used in this study (Fig. ​(Fig.2).2). The primers used were specific for nucleotides located approximately 700 and 800 bp upstream and downstream, respectively, of the Vr. The estimated size of the Vr was calculated by subtracting 1.5 kb from the size of the PCR product. Three strains had Vrs of 7 kb, 5 strains had Vrs of 8 kb, 19 strains had Vrs of 9 kb, 17 strains had Vrs of 10 kb, and 19 strains had Vrs of 11 kb (Table ​(Table1).1). Based on the set gene content of the Vrs in the six sequenced strains, these results are consistent with the presence of 6, 7, 8, 9, or 10 set genes in this chromosomal region in S. aureus strains.

Phylogenetic analysis of RD13.

Phylogenetic analysis was used to reconstruct the evolutionary history of the RD13 locus and to determine the extent that recombination contributed to the divergence of the region (Fig. ​(Fig.3).3). To add to the publicly available genome sequences of S. aureus strains 8325, Sanger MSSA, Sanger MRSA, N315, Mu50, and COL, DNA sequencing of variably present set genes (Vr) of 12 selected strains was carried out. Strains were selected based on Vr size variation and clonal diversity (determined by MLEE) (strains 1006, 1260, 141, 143, 145, 147, 232, 240, 3400, 3402, 537, and 554). The set4, set5, set6, set7, and set8 genes were found to be variably absent in the strains examined. A phylogenetic tree was constructed from the concatenated nucleotide sequences of orf2, set1, res, mod, set2, set3, set9, set10, and set11 (i.e., genes in RD13 that are common to all strains) from the previously sequenced strains 8325, Sanger MSSA, Sanger MRSA, N315, Mu50, and COL to represent the conserved architecture of the RD13 locus. The topology of this tree was compared to the topologies of the individual gene trees of set4 (strains MSA 1006, MSA 1260, MuSA 145, MSA 3400, MSA 3402, 8325, Sanger MSSA, Mu50, and N315), set5 (strains MSA 1006, MSA 1260, MuSA 141, MuSA 143, MuSA 145, MSA 3400, MSA 3402, MSA 537, 8325, Sanger MRSA, Sanger MSSA, Mu50, N315, and NCTC6571), set6 (strains MSA 1006, MSA 3402, 8325, and Sanger MSSA), and set7 (strains MSA 1006, MSA 1260, MuSA 141, MuSA 143, MuSA 145, MuSA 147, MuSA 232, MuSA 240, MSA 3400, MSA 3402, MSA 537, MSA 554, 8325, Sanger MRSA, Sanger MSSA, Mu50, N315, and NCTC6571). Extensive recombination would result in set genes composed of segments of DNA with different evolutionary histories and conflicting phylogenetic topologies. However, we found that the topology representing the conserved regions of RD13 and each of the individual topologies constructed for genes set4 to set7 were consistent, suggesting that recombination had not occurred at a sufficiently high frequency to distort the phylogenetic signal.

Open in a separate window

FIG. 3.

Two models for the evolution of the RD13 chromosomal region and the Vr. The trees were constructed by the neighbor-joining algorithm based on the number of synonymous substitutions per synonymous site (d_S). A comparison of phylogenetic trees constructed from individual set gene nucleotide data and a concatenated sequence representing the conserved genes of RD13 indicated that the topologies were cognate. Thus, the extent of recombination present was insufficient to disrupt the underlying phylogenetic signal. The set9 gene tree is shown. Bootstrap confidence limits are shown under the major nodes. The proposed ancestral state of the Vr is indicated at the hypothetical root. The gain and loss of set genes necessary to explain the extant state of each strain are indicated in red. (A) Evolution of RD13 and the Vr explained solely by the loss of set genes over time. This (preferred) model is supported by the observation that the proportion of silent mutations (synonymous substitution) within each set gene is the same. (B) Evolution of RD13 explained by both the gain and loss of set genes over time (alternative model). This model is not supported by an analysis of the proportion of polymorphic synonymous sites present in each set gene. While other models are possible, they require a larger number of evolutionary events (i.e., gain or loss), making them less likely. (C) Recombination has occurred within RD13. A Happlot analysis of the orf2, set1, and res genes in six representative strains is shown. Polymorphic nucleotide sites based upon pairwise comparisons are represented by vertical lines. Putative areas of recombination, regions of similar nucleotide sequence in different genetic backgrounds, are represented in red.

The analysis suggested two models for the evolution of the RD13 locus (Fig. ​(Fig.3).3). One model requires a common ancestor which has only set5 and set7. The present-day complement of set genes in each lineage would have arisen through multiple acquisition and deletion events. In this case, one would expect less variation in the form of synonymous (silent) nucleotide substitutions in the recently acquired set genes. However, our analysis indicated that the proportions of polymorphic synonymous sites in set4 to set7 are very similar, a result inconsistent with this model. Alternatively, the ancestral state could be represented by a complete complement of set genes in the Vr. In this model, extant states are explained by the loss of set genes in parallel in separate lineages of pathogenic S. aureus (Fig. ​(Fig.3).3). The occurrence of similar proportions of synonymous sites strongly supports this idea. Taken together, these analyses suggest that the evolution of the RD13 locus is explained by a model in which the loss of set genes has occurred several times independently in separate lineages of pathogenic S. aureus (Fig. ​(Fig.3).3). While the extent of horizontal gene transfer has been insufficient to mask the phylogenetic signal present, recombination may have contributed to chromosomal diversification at the RD13 locus. To investigate this possibility, the maximum χ² method was used to identify putative end points of past recombination. Multiple end points flanking small regions of the sequence throughout the RD13 locus were identified (data not shown). The largest of these regions was identified in the res gene based on a comparison of six strains (Fig. ​(Fig.3).3). Restricted allelic variation within the recombined segments of the res gene suggests that recombination has occurred recently.

Analysis of the level of selective constraint acting on set genes.

Considering the high level of predicted homology among SET proteins produced by a single strain, it is conceivable that SETs may have some redundancy of function. Accordingly, we examined the possibility that one or more set genes may have become silent. Silent genes are free to accumulate synonymous (silent) mutations and corresponding amino acid replacements, as selective constraint is no longer acting to maintain protein function by limiting deleterious amino acid replacements. To determine if the level of selective constraint varies across RD13, we calculated p_N and p_S for subsets of 30 codons in a sliding window for the length of each set gene (Fig. ​(Fig.4).4). The difference, p_N − p_S, is a measure of the degree of selective constraint. The more negative the value, the less the contribution of amino acid replacements and the greater the contribution of synonymous nucleotide substitutions. A difference of zero indicates selectively neutral variation, where the per-site rates of synonymous and nonsynonymous substitutions are equal. A positive difference (i.e., amino acid replacements exceeding silent substitutions) suggests the action of diversifying (positive) selection. Overall, the p_N − p_S difference is consistently negative for each of the set genes, a value indicative of purifying selection (Fig. ​(Fig.4).4). The set9 gene did possess a single region 53 codons in length, stretching from position 17 to position 69, in which the value for p_N − p_S was ≥0. In this region, the rate of substitution per 100 sites is expressed by the equations d_N = 9.3 ± 2.1 and d_S = 10.1 ± 3.8. The results suggest that if any of the set genes were silenced, it was a very recent event given that the number of corresponding amino acid replacements has not had sufficient time to accumulate substantially.

Open in a separate window

FIG. 4.

Graphic display of the degree of selective constraint across a representative portion of RD13. The p_N and p_S nucleotide substitutions are indicated by the red and blue lines, accordingly. The black line represents the difference (p_N − p_S), which is a measure of the degree of selective constraint. The more negative the value, the less the contribution of amino acid replacements and the greater the contribution of synonymousnucleotide substitutions. Overall, the difference (p_N − p_S) is consistently negative for each of the set genes, indicative of purifying selection. Polymorphic nucleotide sites based upon pairwise comparisons are represented by vertical lines.

Expression of genes located in RD13 in S. aureus strain COL.

To test the hypothesis that the genes in RD13 are transcribed, TaqMan assays were used to determine relative gene-specific mRNA levels present in mid-exponential and stationary-phase S. aureus cells (Fig. ​(Fig.5)5) by using the oligonucleotides listed in Table ​Table4.4. No transcript was detected in either exponential- or stationary-phase cells for set1, hypothetical orf5, and the putative transposase (tra) genes, a result indicating that these three genes are not transcribed in vitro at these time points under the conditions studied. In contrast, all other genes examined were either constitutively expressed during the growth cycle (res, mod, and set2) or were up-regulated in the stationary phase of growth (orf1 to orf3, orf5, set3, set8, set9, set10, and set11) (the cutoff was a 1.5-fold increase in the normalized transcript level).

Open in a separate window

FIG. 5.

(A) Growth curve of S. aureus strain COL. Total RNA was isolated from S. aureus cells grown in tryptic soy broth at 37°C to the A₆₀₀ values and time points indicated by the dashed lines. (B) Relative quantities of RD13 reverse-transcribed mRNA normalized to the internal control era, determined by Taqman assays.

Western blot analysis of recombinant SET proteins with sera from patients with invasive S. aureus infections.

To determine if SET proteins are expressed during the course of human infection and stimulate a humoral immune response, Western immunoblot analysis of recombinant SET proteins was carried out with paired acute- and convalescent-phase serum samples from 19 patients with S. aureus bacteremia. Of the panel of 11 recombinant SET proteins, 6 were immunoreactive with sera obtained from at least one patient, indicating expression of these proteins during human infection (Table ​(Table5).5). SET2, SET4, SET5, SET8, SET9, and SET10 were reactive with sera from 13, 1, 4, 12, 18, and 3 of the 19 patients examined, respectively. SET1, SET3, SET6, SET7, and SET11 were not reactive with any of the 19 patient sera. Of the 11 recombinant proteins, only SET9 was reactive with any of the eight control sera tested (n = 2) (data not shown). A representative Western blot is shown in Fig. ​Fig.66.

Open in a separate window

FIG. 6.

(A) Immunogenicity of recombinant SET proteins with human patient sera. SDS-PAGE gel of lysates of E. coli expressing recombinant SET protein variants. Lanes 1 to 11, SET1 to SET11, respectively; lane 12, lysate of E. coli containing the expression vector only. Note that SET8 and SET9 resolve at approximately twice their expected molecular weight, thus indicating a dimeric form of the protein (see Western blot analysis in panel B). (B) Representative Western blot analysis of recombinant SET proteins with convalescent-phase serum from human patient 2. Lanes 1 to 11, SET1 to SET11, respectively.

We thank J. Voyich for technical assistance and M. Otto for critical review of the manuscript.