ABSTRACT
Virulence mechanisms of the leading nosocomial pathogen Staphylococcus epidermidis are poorly understood. We used microarray-based genome-wide comparison of clinical and commensal S. epidermidis strains to identify putative virulence determinants. Our study revealed high genetic variability of the S. epidermidis genome, new markers for invasiveness of S. epidermidis, and potential targets for drug development against S. epidermidis infections.
Staphylococcus epidermidis is the most prevalent cause of nosocomial infections, costing the public health system ∼$1 billion/year in the United States alone (18). Usually an innocuous commensal microorganism on human skin, S. epidermidis can cause severe infection after penetration of epidermal and mucosal barriers, which frequently occurs in the hospital during the insertion of indwelling medical devices (18). S. epidermidis mostly lacks components that are easily recognized as virulence factors, such as toxins or aggressive degradative exoenzymes (18). Furthermore, genetic manipulation of S. epidermidis is very difficult. For these reasons we have a serious lack of knowledge about the basis of S. epidermidis virulence. However, discovering the genes that determine success of S. epidermidis as an opportunistic pathogen is a crucial prerequisite step for designing therapeutic interventions directed to control S. epidermidis infections.
In contrast to studies performed with some other pathogenic bacteria, approaches using in vivo expression technology to identify virulence genes in Staphylococcus have proven problematical (3, 9). Low infectivity and the resulting difficulty to establish reproducible animal infection models further complicate the use of this technology with S. epidermidis. Therefore, we used comparative genomics of clinical and benign strains as an alternative approach to identify S. epidermidis virulence determinants. This approach has been used to characterize virulence factors in the opportunistic pathogen Pseudomonas aeruginosa (20). In our study, 22 strains isolated from prostheses infections and 20 strains isolated from the skin of healthy individuals were analyzed by DNA/DNA hybridization of genomic DNA on a whole-genome S. epidermidis microarray. The microarray contained a 70mer oligonucleotide of every gene found in the genome of S. epidermidis RP62A (sequence available at www.tigr.org ). We have described synthesis and characterization of the microarray previously (21). Specifically, we have verified that all oligonucleotides hybridize with control DNA isolated from strain RP62A. The strains used in the present study were a subset of an essentially nonclonal collection (5) and were further screened to exclude related strains that were sometimes found in the same patient. We determined the degree of relatedness of the strains by microarray analysis (Fig. 1). As anticipated from the preselection, microarray data confirmed that the strains were not clonal. Results of the distribution of individual genes were analyzed by Fisher's exact test. A P value of <0.05 was considered significant.
A total of 939 (36%) genes in the control strains and 425 (16%) genes among clinical strains lacked a hybridization signal in at least one strain, indicating absence or significant mutation. These data reveal considerable, previously unknown genetic variability of the S. epidermidis genome. A total of 59 genes showed a significantly disproportionate distribution between the two groups (Table 1 and Table 2). Also, 39 genes were found to be more frequent among clinical strains than among commensal strains (Table 1). Importantly, these genes included the ica locus, which encodes the biosynthetic machinery for the exopolysaccharide PIA (6), and genes related to the insertion sequence IS256. ica and IS256 are among the very few factors that have been described as determinants of virulence and markers for invasiveness of S. epidermidis (6, 7, 15, 23). Notably, these findings validated our approach and further confirmed the use of ica and IS256 as markers for invasiveness of S. epidermidis. However, many other genes revealed an even more significant difference. Particularly, a gene encoding a 190-kDa cell surface protein with similarity to a streptococcal hemagglutinin binding protein showed the most pronounced difference. It was present in 16 of 22 (73%) clinical strains but in only 5 of 20 (25%) control strains (P = 0.0026). We validated the distribution of the gene coding for this protein by analytical PCR, using a different strain collection containing strains from different infections and skin strains from Shanghai, China (Table 3). The results confirmed that the gene occurred significantly more frequently among clinical isolates than among isolates from healthy individuals (P = 0.03).
So-called MSCRAMMs (for “microbial surface components recognizing adhesive matrix molecules”) are believed to play an eminent role in bacterial pathogenesis during the establishment of infection (13). S. epidermidis MSCRAMMs, e.g., the fibrinogen-binding protein Fbe, are presently under intense investigation for use as drug targets or antigens for vaccine development (12, 14). Many but not all MSCRAMMs have an LPXTG motif for linkage to the bacterial cell surface (10). The 190-kDa protein gene lacked a clearly distinguishable LPXTG motif but revealed repeat regions and a putative cell wall binding domain that are typical for MSCRAMMs (13). We detected 10 putative MSCRAMM genes with an LPXTG motif in the S. epidermidis genome. Remarkably, 8 of these 10 putative MSCRAMM genes were absent from at least one strain in our study, as previously shown for the sdrF gene (11), indicating high genetic variability for this class of surface proteins (data not shown). The two MSCRAMMs found to be present in all strains were SE0828 and SE1682, two yet uncharacterized proteins. However, the 190-kDa protein gene was the only putative MSCRAMM that appeared significantly more frequently among invasive strains, suggesting a crucial role for this protein in S. epidermidis pathogenesis.
Several genes with a previously proposed role in virulence were more frequently present in invasive strains. For example, the lipoprotein signal peptidase LspA is required for the secretion of lipoproteins, which represent surface-attached extracellular proteins that may be involved in various virulence mechanisms (2, 19). Furthermore, SsaA is an abundant extracellular antigenic protein in S. epidermidis for which a role in pathogenesis has been proposed (8). Moreover, we detected two genes encoding murein synthesis enzymes in this group, a phosphomuraminic acid pentapeptide translocase-encoding gene and the murF gene. murF and other murein synthesis genes are critical for bacterial survival, and their gene products are under current investigation as potential novel drug targets (4). The absence of a DNA hybridization signal for an essential gene like murF is likely due to significant gene mutation. A mutated murF has been shown to influence methicillin resistance in Staphylococcus aureus (17). A similar role of murF in S. epidermidis might cause the observed lower gene frequency in control strains compared to clinical strain results. Taken together, our data help to emphasize the importance of specific genes among a variety of proposed virulence factors for further investigation of S. epidermidis pathogenesis.
The genes that were more frequent in invasive strains comprised several resistance genes such as genes coding for an antibiotic transport-associated protein and the arsenic efflux pump ArsB (1). In accordance with the latter finding, the gene coding for ArsD, the trans-acting repressor of the arsenic resistance operon, was less common among invasive strains. On the other hand, some putative antibiotic resistance genes showed a significantly higher frequency in the control group. Further, a proline-betain transporter homolog gene was more frequent among invasive strains than in commensal strains, which is in contrast to the assumption that an osmoprotective factor is required during life on the skin rather than during infection. Moreover, recent findings indicate that the arsenic resistance operon might also be involved in osmoprotection (16). These data suggest interesting, previously unexpected roles for antibiotic resistance genes and osmoprotection factors in S. epidermidis.
In conclusion, our study revealed high genetic variability of S. epidermidis as a species. We identified several markers for S. epidermidis invasiveness, which included proposed virulence factors, confirming the validity of our approach and the role in pathogenesis of these factors. Most importantly, our study also identified genes with unknown function for use as potential novel drug targets. We are presently investigating the contribution to virulence of several of the detected putative novel virulence factors.
Relatedness of strains used for microarray-based investigation of gene distribution in this study. Strain names for strains from infections are according to Galdbart et al. (5), by whom the strains were first described. Control strains are numbered from c1 to c22 and are also from that study. Absence of genes is shown in blue; presence of genes is shown in yellow and red. A dendrogram is shown on the right. The analysis was performed with GeneSpring software.
Genes more frequent among invasive strains of S. epidermidis
Genes more frequent among commensal strains of S. epidermidis
Distribution of gene SE2251 in strains from Huashan hospital, Shanghai, determined by analytical PCR
ACKNOWLEDGMENTS
We thank Donald J. Meyer, Partek Inc., for statistical analysis, Nevine El Solh, Institut Pasteur, for S. epidermidis strains, and Frank R. DeLeo for critically reading the manuscript.
FOOTNOTES
- Received 27 July 2004.
- Returned for modification 10 September 2004.
- Accepted 17 October 2004.
- Copyright © 2005 American Society for Microbiology