Use of Multiplex PCR To Identify Staphylococcus aureus Adhesins Involved in Human Hematogenous Infections

Staphylococcus aureus is both a commensal and an extremely versatile pathogen. It causes superficial lesions, deep-seated and systemic infections, and toxemic syndromes (10). With the exception of toxemic syndromes, which are directly due to toxin production, S. aureus pathogenesis results from the combined action of a variety of factors (1, 17). S. aureus infection begins with bacterial adhesion to host tissues. S. aureus adhesins are grouped into a single family named microbial surface components recognizing adhesive matrix molecules (MSCRAMMs) (5). MSCRAMM involvement in S. aureus pathogenicity in human beings is poorly documented. Results obtained in experimental models suggest that S. aureus strains that specifically express receptors for fibrinogen and fibronectin are associated with infective endocarditis, while strains that produce receptors for bone sialoprotein, collagen, and fibronectin are associated with osteomyelitis and arthritis (2, 3, 9, 18, 20, 21, 23). However, these associations have not been confirmed in all relevant studies.

Staphylococcal MSCRAMMs can be detected with binding assays using purified matrix molecules (11), but several adhesins are redundant. For example, there are two fibronectin binding proteins (FnbpA and -B) and three receptors for fibrinogen (clumping factors A and B [ClfA and -B] and fibrinogen binding protein [Fib]). In addition, some MSCRAMMs bind to more than one matrix molecule (for example, FnbpA binds to both fibronectin and fibrinogen) (22). Finally, binding assays are expensive, owing to the high cost of purified matrix molecules. PCR has been used to detect staphylococcal adhesin genes (12, 13, 16), but multiplex PCR technology has not yet been used to study adhesin distribution in S. aureus isolates of various origins. We therefore developed a multiplex PCR procedure to examine the distribution of nine MSCRAMM genes in S. aureus isolates from patients with nasal colonization and with invasive infections associated with hematogenous seeding.

The nucleotide sequences of ebpS (encoding elastin binding protein), eno (encoding laminin binding protein), cna (encoding collagen binding protein), fnbA and fnbB (encoding fibronectin binding proteins A and B), fib (encoding fibrinogen binding protein), clfA and clfB (encoding clumping factors A and B), and bbp (encoding bone sialoprotein binding protein) obtained from GenBank (accession numbers U48826, AF065394, M81736, X95848, X62992, X72014, Z18852, AJ224764, and Y18653, respectively) were compared and evaluated by using Blast and ClustalX softwares to identify regions unique to each gene, with similar annealing temperatures. Genomic DNA was extracted from staphylococcal cultures (2) and used as an amplification template with the primers (Eurogentec, Seraing, Belgium) described in Table 1 in both simplex and multiplex PCRs. Two primer sets were prepared for multiplex PCR: PCR1 to amplify bbp, cna, ebpS, and eno and PCR2 to amplify fnbA, fnbB, fib, clfA, and clfB. The thermal cycling conditions included an initial denaturation step (5 min at 94°C) followed by 25 cycles of amplification (denaturation for 1 min at 94°C, annealing for 1 min at 55°C, and extension for 1 min at 72°C). The reaction was terminated with a 10-min incubation step at 72°C. PCR products were analyzed by electrophoresis through 1% agarose gels (Sigma, Saint Quentin Fallavier, France).

A subset of 157 S. aureus isolates, identified as S. aureus by their abilities to coagulate citrated rabbit plasma and to produce a clumping factor, were selected for PCRs. These strains were collected from nose swabs of asymptomatic carriers or from clinical specimens of patients with native-valve endocarditis or hematogenous osteomyelitis and/or arthritis (7, 8); types of infection were defined in accordance with published criteria (4, 6).

We first examined the specificity of the newly designed primers by means of simplex PCRs. As no control strains were available for the nine genes, we randomly selected 17 S. aureus strains and tested each with the nine simplex PCRs. All amplicons were consistent with the predicted sizes (Table 1). Two randomly selected amplicons from each of the nine simplex PCRs were subjected to DNA sequencing to confirm their specificity and were 98 to 100% identical to the corresponding GenBank sequences. The 17 strains of S. aureus tested by simplex PCR were then tested by multiplex PCR and yielded the expected band patterns; no discrepancies between the two methods were observed.

All 157 clinical isolates tested by multiplex PCR were positive for eno and clfB. The other MSCRAMM genes were heterogeneously detected: 99% of the strains were positive for clfA, 80% were positive for fib, 58% were positive for ebpS, 43% were positive for fnbB, 36% were positive for cna, 28% were positive for fnbA, and 22% were positive for bbp (Table 2). One isolate was positive for eight of the nine MSCRAMM genes, 20 isolates were positive for seven genes, 67 were positive for six genes, 63 were positive for five genes, 5 were positive for four genes, and 1 was positive for three genes (median, six genes per isolate). Eighty-six of the 157 isolates harbored none of the fibronectin binding protein genes, while all of the isolates harbored at least two of the three fibrinogen receptor genes (fib, clfaA, and clfB). The minimal MSCRAMM gene equipment in this panel of 157 isolates comprised two genes encoding fibrinogen receptor (including clumping factor B in all strains) and the gene encoding laminin binding protein. This confirms previous reports that most S. aureus isolates harbor functionally redundant MSCRAMMs (16).

The MSCRAMM gene distribution did not differ between isolates associated with nasal colonization and those associated with invasive infections (endocarditis and osteomyelitis/arthritis; P > 0.05 for each gene comparison [chi square test]; Table 2). In contrast, two MSCRAMM genes were significantly associated with specific invasive infections: bbp was associated more with osteomyelitis/arthritis than with endocarditis (38 versus 11%; odds ratio = 5.2; P = 0.006), and fnbB was associated more with endocarditis than with osteomyelitis/arthritis (55 versus 32%; odds ratio = 0.4; P = 0.046) (Table 2). Previous studies have shown that strains expressing receptors for bone sialoprotein are associated with osteomyelitis and arthritis (19) and that strains bearing receptors for fibronectin (but not specifically FnbB) are associated with infective endocarditis (18). However, in our study, not all of the isolates associated with osteoarthritis and endocarditis harbored the bbp or fnbB gene, respectively (Table 2). Thus, in addition to possible host factors, this tropism of S. aureus may involve additional MSCRAMMs, such as receptors for fibrinogen (in endocarditis) and for fibronectin and/or collagen (in osteomyelitis and arthritis) or other virulence factors involved in bacterial persistence at the site of infection after tissue binding (3, 15, 21, 23). Indeed, inactivation of a single MSCRAMM gene is not sufficient to abolish strain virulence in an experimental model of hematogenous S. aureus infections (14, 15, 21). Combined investigation of MSCRAMM and toxin gene distribution in S. aureus isolates would permit better understanding of the complex process of tissue seeding during bacteremia.

• American Society for Microbiology