|Home | About | Journals | Submit | Contact Us | Français|
Staphylococcal enterotoxins (SE) can cause toxin-mediated disease, and those that function as superantigens are implicated in the pathogenesis of allergic diseases. The prevalence of 19 enterotoxin genes was determined by PCR in clinical S. aureus strains derived from wounds (108) and blood (99). We performed spa typing and multilocus sequence typing (MLST) to determine clonal origin, and for selected strains staphylococcal enterotoxin B (SEB) production was measured by enzyme-linked immunosorbent assay. Strains carried a median of five SE genes. For most SE genes, the prevalence rates among methicillin-resistant and methicillin-sensitive S. aureus isolates, as well as wound- and blood-derived isolates, did not differ. At least one SE gene was detected in all except two S. aureus isolates (>99%). Complete egc clusters were found in only 11% of S. aureus isolates, whereas the combination of sed, sej, and ser was detected in 24% of clinical strains. S. aureus strains exhibited distinct combinations of SE genes, even if their pulsed-field gel electrophoresis and MLST patterns demonstrated clonality. USA300 strains also showed considerable variability in SE content, although they contained a lower number of SE genes (mean, 3). By contrast, SE content was unchanged in five pairs of serial isolates. SEB production by individual strains varied up to 200-fold, and even up to 15-fold in a pair of serial isolates. In conclusion, our results illustrate the genetic diversity of S. aureus strains with respect to enterotoxin genes and suggest that horizontal transfer of mobile genetic elements encoding virulence genes occurs frequently.
As a commensal, Staphylococcus aureus colonizes the nasal mucosa of 20 to 40% of humans (54), and as a pathogen it causes pyogenic diseases and toxin-mediated diseases (38). S. aureus produces many different virulence factors, including enterotoxins (SEs), which can cause defined toxic shock syndromes (4). The characterization of some of these toxins led to the discovery of superantigens (41), which bind to major histocompatibility complex class II molecules and Vβ chains of T-cell receptors, resulting in the activation of large numbers of T cells (20 to 30%) and massive cytokine production (10, 18). These superantigen-induced “cytokine storms” are responsible for the toxic effects seen in staphylococcal entertoxin B (SEB)- and toxic shock syndrome toxin (TSST)-associated shock syndromes in S. aureus infections (13, 40, 47). To date, 19 SEs have been identified based on sequence homologies, and studies have reported enterotoxin genes in up to 80% of all S. aureus strains (4, 21). Although many new enterotoxins have been identified, i.e., seg ser and seu (33, 37, 44, 49), their precise functions have not been characterized yet. The majority of experimental work with SEs is still done with SEB, toxic shock syndrome toxin 1, and SEA (27, 31), because these toxins are commercially available. Most SEs are located on mobile elements in bacterial genomes such as plasmids or pathogenicity islands and can thus be easily transferred horizontally between strains (5, 34, 35). Certain SE genes are grouped together. For instance seg, sei, sem, sen, and seo are commonly found in a gene cluster (egc) on genomic island νSAβ (34), and sel and sek are often found together with seb or sec on S. aureus pathogenicity islands. Other staphylococcal superantigen genes are encoded on plasmids (sed, sej, and ser) or are linked to the antibiotic resistance cassette SCCmec (seh) (44, 55). Phage 3 carries either sea (strain Mu50), sep (N315), or sea sek seq (MW2) (1, 29).
Although a few clinical studies have attempted to correlate shock and outcome with the presence of certain SEs in patients with S. aureus infections (17, 28), the contribution of these toxins to outcome is still unclear. Recent papers have proposed the SEs are immunomodulators and that colonization with S. aureus strains that produce SEB may contribute to the pathogenesis of asthma, chronic rhinitis, and dermatitis (2, 36, 46, 48, 56). The superantigen function of SEs in supernatants of S. aureus cultures can be neutralized by serum of colonized patients (21, 23). With new data emerging implicating SEs in the pathogenesis of chronic allergic syndromes, production of monoclonal antibodies and or vaccine strategies targeting SEs may be considered (6, 24, 26, 30) in the future. It is therefore important to characterize the prevalence of SE genes in clinical S. aureus strains.
In this study, we analyzed SE content in both methicillin-resistant S. aureus (MRSA) and methicillin-sensitive S. aureus (MSSA) strains that were cultured from wounds (including USA300) and bloodstream infections of patients from a defined geographical area. In addition, SEB production was quantified by enzyme-linked immunosorbent assay (ELISA) in S. aureus strains carrying the seb gene, and spa typing confirmed clonal diversity among S. aureus isolates from different patients, as well as clonal stability in serial isolates, and multilocus sequence typing (MLST) done on a subset of less common spa types. We conclude that SE genes are abundant in S. aureus strains, albeit less abundant in USA300. SE content and combination are highly diverse and therefore more discriminatory than pulsed-field gel electrophoresis (PFGE) and MLST typing, albeit stable in serial isolates. Quantification of SEB production demonstrates that enterotoxin secretion can vary greatly among strains, even if they belong to the same S. aureus lineage. Given the complexities of SE prevalence, regulation, and possible function, we propose that the association of these toxins with chronic allergic diseases or outcome may be oversimplified at present. Precise characterizations of SE function and secretion patterns in individual S. aureus clones are warranted.
A total of 207 S. aureus isolates were obtained from three hospitals in the Bronx, NY. Identification of clinical isolates and antibiotic susceptibility testing were performed in the microbiology laboratory of Montefiore Medical Center in accordance with NCCLS guidelines (43). Isolates were further subcultured in Bacto brain heart infusion medium overnight at 37°C. Of these, 108 S. aureus strains were isolated from wounds, and 99 came from blood cultures. A total of 105 wound isolates were skin and soft tissue infections, and 3 were other types (one each derived from organ graft, placenta, and vagina). Isolates were intermittently collected in two blocks of 2 months each. The blood isolates reflect about 50% of all S. aureus isolates sent to the laboratory during that time. For wound isolates, the denominator cannot be accurately determined because many isolates logged in as wound isolates are not necessarily derived from wounds, e.g., catheter tips. We asked the laboratory to hold all S. aureus isolates derived from abscesses. Wound isolates were more commonly derived from males and younger patients. The collection of strains also included five pairs of serial isolates from the same patient (four from blood and one from a wound). They were collected 6, 56, 62, and 150 days apart. Retrospective chart review of 207 patients with wound and bloodstream infections was approved by the Institutional Review Board. Information on demographic characteristics, clinical presentation, antibiotic therapy, complications, and outcomes was obtained. Fever was defined as greater than 100.4°F, and leukocytosis was defined as ≥12,000/μl. Hospital care-associated risk factors included end-stage renal disease (for which patients were on dialysis), hospitalization 6 months prior to admission, prior history of S. aureus infection, and nursing home residence.
Genomic S. aureus DNA was purified using a Qiagen DNeasy tissue kit according to the manufacturer's protocol and then used for PCR amplification of enterotoxin genes. The following 19 toxin genes were examined using PCR assays: sea, seb, sec, sed, see, seg, seh, sei, sej, sek, sel, sem, sen, seo, sep, seq, ser, seu, and tsst. The nucleotide sequences of all PCR primers used in this study and their respective amplified products are listed in Table Table1.1. All PCR assays were performed with individual primer sets. PCR amplifications were performed in a Mastercycler (Eppendorf, Germany) with Platinum PCR SuperMix (Invitrogen). Genomic DNA was added to 23 μl of PCR mixture with 10 pmol of forward and reverse primers. PCR products were visualized by 1% agarose gel electrophoresis. Positive and negative controls were included in each PCR run. Amplicon sequences were verified against S. aureus database sequences to ensure specificity. Control strains for the PCR-based assays included the following: strain FRI-361 for sec2, sed, seg, sei, sel, sem, sen, seo, sej and ser genes; strain FR1472 for sed and sej genes; strain MNDON (NRS113) for sec1, seg, seh, and sei genes; strain FR1722 for sea, sed, sej, and ser genes.
PFGE was performed for all USA300 strains and SEB-positive S. aureus isolates, as well as for 10 serial isolates. Briefly, plugs were generated from overnight cultures of S. aureus and digested with SmaI prior to performing PFGE as described elsewhere (39). Electrophoresis was performed in 0.5× Tris-buffered EDTA buffer using a CHEF-DR III electrophoresis cell (Bio-Rad, Melville, NY), under the following conditions: initial pulse of 5 s, final pulse of 40 s, voltage at 6 V/cm, 14°C, for 20 h. The gel was stained with ethidium bromide for 20 min, destained with distilled water for 30 min, and visualized using a UV transilluminator.
S. aureus isolates were grown overnight at 37°C on Luria-Bertani agar, and DNA was isolated by boiling lysis. PCR was performed using oligonucleotide primers for the seven MLST targets and spa (15, 50), and DNA sequencing was performed commercially. MLST sequences for each primer set were assembled using MEGA version 4.0 (52) and analyzed using the MLST website (http://www.mlst.net) and eBURST v.3 (16). spa typing (50) analysis was performed using eGenomics software (http://www.egenomics.com), and Ridom spa types were assigned using the SpaServer website (http://spaserver2.ridom.de).
All isolates were screened by real-time PCR with molecular beacons targeting the spa and mecA genes (51); isolates positive for both were classified as MRSA, whereas those positive for spa alone were classified as MSSA. All mecA-positive isolates were subjected to SCCmec typing using a novel two-panel multiplex real-time PCR procedure (9). The first panel targets mecA, ccrB2, mecI, and IS1272J and can identify SCCmec types II (ccrB2, mecI) and IV (ccrB2, IS1272J). The second panel targets ccrC, ccrB1, ccrB3, ccrB4, and mecC2, and in combination with the first panel, can identify SCCmec types I (ccrB1, IS1272J), III (ccrB3, mecI), V (ccrC, mecC2), VI (ccrB4, IS1272J), and VIII (ccrB4, mecI). All real-time PCR assays were performed on an Mx4000 multiplex quantitative PCR system (Stratagene, La Jolla, CA).
Isolates were also screened for the presence of arginine catabolic mobile element (ACME), which is known to be associated with USA300 (12). Detection of Panton-Valentine leukocidin in these strains was performed by real-time PCR with molecular beacons targeting the lukF-PV gene (51). Detection of ACME was accomplished by conventional PCR targeting two genes from the ACME operon, arcA and opp3A.
Routine culture of all S. aureus strains was performed using brain heart infusion broth. For analysis of toxin production, S. aureus strains from overnight cultures were diluted to 1 × 106 to 3 × 106 CFU/ml in fresh culture medium and grown at 37°C in 5% CO2 with shaking (200 rpm) for 24 h, at which time all culture specimens reached a plateau of 1 × 109 to 4 × 109 CFU/ml. Bacterial culture specimens were rendered free of bacteria by centrifugation, and supernatant was frozen at −70°C until used for SEB quantification by ELISA as previously described (20).
Categorical variables between groups were assessed using the chi-square test. P values of <0.05 were considered statistically significant. Contingency tables were used to compare the prevalence rates of particular SE genes.
In this study, PCR-based typing of 19 SE genes was performed in 207 blood- (99) and wound- (108) derived S. aureus isolates from three Bronx hospitals. The demographic and clinical characteristics are summarized in Table Table2.2. The distributions of MRSA and MSSA were similar between blood (53% MRSA) and wound (43% MRSA) isolates (P = 0.2). Wound isolates were more commonly derived from male patients, whereas blood isolates were derived from patients exhibiting comorbidities such as diabetes mellitus, end-stage renal disease, and human immunodeficiency virus disease. The latter were typically older and infirm and thus more likely to experience fever, longer hospital stays, and expiration.
A total of 16 unique MLST clonal complexes (CC) were identified in 207 S. aureus isolates tested (Fig. (Fig.1).1). MRSA strains were less diverse and only included 6 CCs, whereas MSSA strains included 15 CCs, in addition to one unknown type (nearest match was singleton ST883). The same genotypes (as per MLST and spa typing) were found in both MRSA and MSSA isolates, including CC1 (spa type 131), CC5 (spa types 2 and 23), CC8 (spa types 1, 7, and 363), and CC59 (spa type 17). MLST type CC5 was more common in blood isolates, whereas CC8 was predominant in wound isolates (CC5, 37% versus 19%; CC8, 32% versus 49% for blood and wound isolates, respectively; P = 0.003 and P = 0.015 by chi-square, respectively). The difference in CC8 prevalence was largely due to the higher prevalence of USA300 strains in wound isolates (58%) compared to blood isolates (25%).
The prevalence of SE genes was determined in all S. aureus isolates (see the table in the supplemental material). At least one SE gene was detected in all except two S. aureus isolates (>99%). Most (198 of 207) S. aureus strains contained more than one SE gene. The median number of SE genes in S. aureus isolates was 5; however, up to 12 distinct SE genes could be identified in a given S. aureus strain. The prevalence of individual SE genes was variable and was as follows: ser > sep > sek > sem > sei > sen > seg > seu > sej > sed > seo > sec > sel > seq > seb > tsst > sea > seh > see. The most abundant SE was ser, which was found in 82% of wound and 96% of blood isolates. Of note is that the see gene was not found in any of the wound and blood isolates, despite using three independent primer sets. Prevalences in blood and wound isolates were comparable for most SE genes and differed for in only three genes, namely, sed, sej, and ser (Fig. (Fig.2A).2A). Although earlier data had suggested that certain SE genes are restricted to either MRSA or MSSA strains, we detected similar distributions of most SE genes in both MRSA and MSSA, except for sek and sen genes (P < 0.02) and the egc cluster (sem, sen, and seo) (P = 0.006) (Fig. (Fig.2B).2B). Notably, except for sek, differences were alleviated when USA300 isolates (see below) were excluded from the comparison. We also explored if multiresistant strains carried more or less SE genes. The results are summarized in Table Table33 and demonstrate no correlation between the number of SE genes and degree of antimicrobial resistance.
Combinations of SE genes were highly conserved in S. aureus strains. We examined if certain combinations of SE genes remained stable among clinical strains (n = 207). For instance, seg and sei alone was found in 74 and 79 isolates, respectively, while coexistence of seg and sei occurred in 71 and 77% of the isolates, respectively. Complete egc clusters (seg, sei, sem, sen, and seo) were found in only 10.5% of S. aureus isolates, and the combination of sed sej ser was detected in 23.9% of clinical strains. Two new SE combinations, namely ser sep seu (18.6%) and seg ser sej (13.3%) (both partial and incomplete), were commonly found in this study (Fig. (Fig.3A),3A), but have not been previously reported to our knowledge.
Several mechanisms of antimicrobial resistance have been described in S. aureus, mostly involving the acquisition of plasmids. We also explored if multiresistant strains carried more or fewer SE genes. The results are summarized in Table Table33 and demonstrate no correlation between number of SE genes and degree of antimicrobial resistance.
We then analyzed SE frequencies of S. aureus isolates belonging to the same genotypic background. Diverse toxin patterns were found in identical MLST clonal complexes, as well as within similar spa types (Table (Table4).4). Toxin frequencies were highly diverse within individual clonal complexes (Fig. (Fig.3B),3B), ranging from 2 to 12 toxins. The only two S. aureus strains possessing no SE genes were distinct clones (CC8/spa type 59 and CC88/spa type 166), which were only represented once. MLST typing identified an unknown MSSA strain with spa repeat pattern Q2-new-JMMMMJJJMK, whose closest MLST database match was ST883. This strain carried an unusually large number (n = 10) of toxins. In this collection, both USA400 (n = 4), and USA500 (n = 13) strains carried more toxins on average (means, 6.4 and 5.8; range, 4 to 9) than USA300 strains (mean, 4.4; range, 0 to 11), while USA700 strains (CC8/72) carried an unexpectedly high number of toxins (mean, 7.1; range, 6 to 9). Despite identical MLST, spa, and even PFGE patterns, USA300 isolates (n = 38), which represented the most homogenous group, exhibited 28 distinct combinations of SE genes. The largest clonal complex, CC8 (n = 90), included both MRSA and MSSA strains which carried comparable numbers of SE genes regardless of methicillin sensitivity, but within which many different combinations of SEs were identified. Taken together, these data demonstrate that SE content is not strain related and may be more discriminatory than PFGE, spa typing, or MLST, as virtually every strain (with the exception of some USA300 strains) contained a different combination of SE genes. By contrast, MLST, spa type, PFGE, and SE content patterns were shown to be stable in five pairs of serial isolates derived from the same patient (data not shown, except for the serial seb-positive isolates discussed below).
The seb+ strains (n = 19) were further characterized insofar as this SE is of particular interest since it has been proposed to be an immunomodulator and is implicated in the pathogenesis of allergenic diseases. Notably, seb was only identified in four clonal complexes, CC8 being the most common (14/19), followed by CC59 (3/19), CC20 (1/19), and the previously mentioned unknown strain (1/19). It is noteworthy that out of 13 spa type 7 strains (USA500), 12 carried the seb gene. Serial seb+ isolates derived from the same patient had identical PFGE and MLST types, whereas all other seb+ strains contained different combinations of SE genes, regardless of genotypic similarity (Fig. (Fig.3A).3A). Interestingly, none of the seb+ S. aureus strains possessed the egc cluster, while two seb+ blood isolates also carried the tsst-1 gene (Fig. (Fig.4B).4B). SEB production by individual strains was also quantified by capture ELISA (11). These data demonstrated that although all seb+ strains produced SEB, their levels of toxin production varied more than 200-fold (Fig. (Fig.4C).4C). SEB production varied in clones with the same PFGE pattern and even varied 15-fold in a pair of serial seb+ isolates derived 6 days apart from the same patient (isolates 4a and 4b).
S. aureus is one of the most frequently encountered bacterial pathogens and is responsible for a variety of life-threatening infections. Recent studies have investigated the prevalence of SE genes in S. aureus strains from geographically diverse locations (4, 7, 22, 53). In this study, we used a collection of clinically well-characterized wound and blood isolates from patients in three different hospitals from the Bronx, in order to evaluate SE gene content in different S. aureus strains. We determined the presence of 19 different SE genes in strains isolated from 99 bloodstream infection and 108 skin and soft tissue infection cultures. Although the collection does not include all S. aureus isolates captured by the hospital laboratory during the period in question, no apparent sampling bias was introduced that we could ascertain; consequently, we believe that these data are representative of the prevalence of different S. aureus strains in our community. Overall, >99% of all isolates contained at least one SE gene. Except for sed, sej, and ser, we detected no differences in SE prevalence rates between wound and blood isolates or with respect to methicillin resistance. It is noteworthy that the prevalence of SEs is both abundant and diverse among clinical strains, such that any correlation with clinical presentation or outcome would require examination of very large cohorts. In addition, based on quantification of SEB, one cannot assume that strains harboring the same SE genes also produce similar quantities of toxins, which further complicates correlation of strain typing data or clinical outcome with enterotoxin content.
In this study, ser was the most prevalent SE in both blood (96%) and wound (82%) isolates. The higher prevalence rates of seg and sei in blood (44%, 46%) compared to wound (30% and 33%) isolates were consistent with previous studies (4, 14, 25, 42, 45). These SEs commonly coexist in the SE gene cluster (egc) along with three other genes (sem, sen, and seo). In our study, 11% of wound and 10% of blood isolates carried a complete egc cluster. The coexistence of sed and sej genes has been reported by many investigators, as they are encoded by a plasmid pIB485 (3, 19, 55). In the present study, we found that the frequency of sed and sej in combination with ser was significantly higher in blood (35%) than in wound isolates (13.7%). This finding is in agreement with a previous observation reporting a high prevalence of sed-sej combinations in blood isolates (4). In addition, Lehn and colleagues also reported that five out of six sed-positive isolates originated from blood cultures (32).
Strain diversification in S. aureus is an ongoing process, an observation that was borne out by our study. Our data demonstrate that MLST, spa typing, and PFGE were all less discriminatory than SE content in seb+ strains, as every strain except the serial pair exhibited a different combination of SE genes. Likewise, 28 distinct combinations of SE content were found in S. aureus strain USA300, the predominant cause of community-acquired MRSA infections in the United States (39). These strains were further distinguished insofar as they harbored a relatively low number of SE genes, with the exception of a single blood isolate harboring 11 SEs. Given that most strains contained a distinct SE signature even if their MLST backgrounds or PFGE patterns demonstrated clonality, horizontal transfer or loss of mobile genetic elements such as SE genes may occur frequently within individual lineages. In some cases, possession of other mobile elements may interfere with acquisition of certain SE genes, suggested by the fact that seb+ S. aureus strains did not carry the egc cluster and that this SE was only found in a few distinct clonal complexes. The limited data from the seb+ serial isolates suggest that SE gene content does not change during the course of infection within a human host. Given that finding, recent experiments have demonstrated that mobile elements can be exchanged among bacteria of different species (8) and that polymicrobial infections are not uncommon in wounds and the gastrointestinal tract. This conclusion would need to be strengthened further by examining larger numbers of serial isolates, which were not available to us.
In clinical studies, sea, seb, and tsst-1 have been thoroughly investigated, since they are all superantigens associated with toxic shock syndromes. However, most of the clinical studies have been hampered by the fact that low concentrations of SE cannot be quantified accurately in vivo. This is especially true for SEB, which is classified as a biological warfare agent due to its toxicity at very low concentrations. This study demonstrated that SEB production varies greatly among clinical strains grown to the same density under identical conditions and that toxin production may even vary among serial isolates.
In conclusion, our results illustrate the genetic diversity of S. aureus strains with respect to SE genes. They suggest that horizontal transfer of mobile genetic elements encoding virulence genes probably occurs frequently in settings where patients are coinfected or cocolonized with more than one strain. Whether certain strains (e.g., USA300) always have lower numbers of SE genes, and also whether seb is always predominantly found in specific clonal complexes, is not known. It is conceivable that factors in the core genome of the recipient strain, or else prior presence of accessory genetic elements, may restrict the transfer of mobile elements. The abundant presence of SEs in most clinical strains, combined with the fact that the majority of SEs have not been functionally characterized, underlies the complexity of SE content analysis with respect to clinical relevance. In addition, our data suggest that clinical S. aureus strains can regulate toxin production differentially, thereby further complicating analyses. Taken together, investigations that attempt to associate the presence of SEs, or even the presence of IgEs that recognize specific SEs with clinical outcome, should take into account the complexity of the findings reported in this study. Further functional characterization of individual SEs may be warranted before additional conclusions can be drawn.
We thank Emily Cook for technical help. We thank Marilou Corpuz for assistance with obtaining IRB permission.
This work was funded by the Northeast Biodefense Center (U54-AI5718-Lipkin).
Published ahead of print on 11 September 2009.
†Supplemental material for this article may be found at http://aem.asm.org/.