Clonal Complex 398 Methicillin Susceptible Staphylococcus aureus: A Frequent Unspecialized Human Pathogen with Specific Phenotypic and Genotypic Characteristics

Clonal complex 398 livestok-associated-MRSA (CC398 LA-MRSA) clone is described as a major animal pathogen that can also colonize and infect humans. CC398 methicillin susceptible Staphylococcus aureus (CC398 MSSA) is less described. We identified 126 CC398 MSSA strains of human origin within 6380 S. aureus isolates gathered between 2009 and 2011, from the French National Reference Centre for Staphylococci. They were characterized using antimicrobial susceptibility testing, spa typing, DNA microarrays (Identibac S. aureus Genotyping ®, Alere), CC398-specific sequence PCR, ermT (encoding macrolides résistance) PCR. Fifty-three CC398 LA-MRSA collected from French pigs and veal were used as comparators, and phylogenetic relations between human CC398 MSSA and animal CC398 MRSA populations were explored on the basis of spa-typing and DNA microarrays. CC398 MSSA were able to induce a large spectrum of infections (especially skin, bloodstream, and pneumonias). The prevalence rate of this clone was high in MSSA population, i.e., 24.7% in a local prospective study on nasal colonization, and 7.5% in a national prospective study on infective endocarditis. CC398 MSSA isolates were frequently (89%) erythromycin resistant, due to the presence of the ermT gene, a gene not detected in erythromycin resistant CC398 LA-MRSA strains. Expression of staphylococcal complement inhibitor (scn) and the chemotaxis inhibitory protein (chp), was also specific to this population. The CC398 MRSA signature included also a panel of antibiotic resistance genes, especially a type IV or V cassette mec and tetM. CC398 MSSA and CC398 LA-MRSA populations were closely related based on spa-typing and DNA microarrays, with the MRSA strains forming the most derived lineage in phylogenic trees. Both MSSA and MRSA populations may come from common ancestors, which would have evolved in the settings of different selective pressures, explaining the acquisition of ermT, chp and scn for MSSA, and antibiotic resistance genes for MRSA.


Introduction
Staphylococcus aureus is both a human commensal organism and a pathogen [1] and inter-human transmission is the main route of its dissemination. Wild strains are susceptible to methicillin (MSSA), and methicillin resistance acquisition can be healthcare associated (HA-MRSA), or community associated (CA-MRSA) [2,3]. Since the 2000s, livestock animals, especially pigs, have been described to be a third reservoir of MRSA (called livestock-associated MRSA (LA-MRSA)). Particularly, MRSA belonging to sequence type 398 (ST398) or related, clustering in clonal complex 398 (CC398), have emerged world-wide and have been described to be able to colonize a large panel of farm animals. The transmission of these MRSA to humans, leading to colonization but also to infection, has been now reported in many countries [4,5].
If animal and human CC398 MRSA strains have been extensively explored, their human MSSA counterparts have been poorly studied. Herein, we investigated clinical, phenotypic, genotypic features of the largest collection of such isolates collected in France, country where MRSA CC398 have been initially described in 2005 [17] and where emergence of CC398 MSSA have been first reported [18]. Molecular data were compared to those obtained from French CC398 LA-MRSA isolates.

Results
CC398 MSSA was found to be associated with colonization as well as with a wide range of infections ( Figure 1). Skin and soft tissue infections, bacteraemiae, and pneumoniae were the three most frequent infection types (63/89, i.e. 70.7%). On the basis of a French national prospective study, the prevalence rate of CC398 MSSA within a collection of 132 MSSA strains isolated in 2008 from human infective endocarditis was 7.5% (10/132), thus representing the sixth most frequent CCs after CC45, CC5, CC15, CC30 and CC8. In the same way, upon MSSA strains collected from nasal carriage in a French healthy population, CC398 MSSA was highly prevalent and represented 24.7% of all isolates (21/85).
Erythromycin resistance was the most common within CC398 MSSA (93/105, i.e. 88.6%), whereas resistance to other antibiotics (except penicillin) were detected in less than 6% of isolates ( Table 1). A low diversity of antibiotic resistance profiles (namely 11) was observed ( Table 2), but resistance to erythromycin alone, or associated to penicillin was the two most frequent, representing more than 80% of all profiles. The data from DNA microarrays permitted to genetically explain erythromycin resistance (n = 93) only in 7 cases: 3 strains were positive for ermA and ermC, 2 for ermC alone, 1 for ermA alone, and 1 for msrA (efflux pump). Conversely, the presence of the ermT gene, screened using dedicated PCR, explained erythromycin resistance for all of the 86 remaining strains (92.5% of the 105 isolates tested).
The genetic background of the 105 CC398 MSSA isolates was explored using spa typing and microarrays, and was found to be highly homogenous. If 25 different spa-types were identified, 85% (89/105) of the strains were clustered in a unique spaCC (spaCC 571) and the phylogenic analysis showed 80% homology of spa repeats between all strains (see Figure S1). This homogeneous structure of the MSSA CC398 population was confirmed by DNA microarrays data, with constant results for 73.8% of the genes (127 out of 172: 49 always positive and 78 always negative) but also differences (positive versus negative) impacting less than 3% of the 105 isolates, and concerning 27 other genes. The clustered analysis based on microarray data divided the strains into 2 subgroups ( Figure S2), matching presence/absence of penicillinase (blaZ).
The same phenotypic and genetic data was compared to those obtained with CC398 MRSA population collected from colonized animals in France (n = 53 isolates). Unlike human MSSA CC398, CC398 MRSA presented highly diverse multiresistant antibiotic phenotypes, confirmed by the identification of more than 25 different resistance genes profiles. Conversely, a homogenous background was highlighted by spa typing; all isolates were distributed in 6 spa-types that all belonged to the same spaCC571, as CC398 MRSA ( Figure S3). In the same way, DNA microarrays revealed constant results for more than 68% of the genes within CC398 MRSA collection. The clustered analysis based on microarray data divided the strains into 2 subgroups, related to the acquisition of a different SCCmec cassette, namely type IV or type V ( Figure S4). Finally, it must be noticed that ermT gene was absent within this collection and especially in erythromycin resistant strains (unlike their MSSA counterparts).
The compared analysis of the 105 human CC398 MSSA and the 53 animal CC398 MRSA strains ( Figure 2) showed a higher diversity of spa-types within CC398 MSSA compared to CC398 MRSA (15 versus 6, Simpson diversity index = 88.9% (95%CI, 86.8%-90.9%)). In the same way, DNA microarrays showed that a major part of the constant genes were shared by the two MSSA and MRSA groups (n = 86, i.e. 82.7% of the constant MSSA genes, 73.5% of the constant MRSA genes, and 50% of all the studied genes). The clustered analysis based on microarray data divided the strains into the 4 subgroups described before ( Figure 3): two subgroups within MSSA matched presence/ absence of penicillinase (blaZ) and two within MRSA related to the presence of SCCmec IV or SCCmec V. Interestingly, the exclusion of the 45 antibiotic resistance genes of this analysis lead to a representation with an unique cluster (Figure 4). When a global statistical analysis of DNA microarray data was performed, eighteen genes appeared significantly associated with either the livestock CC398 MRSA or the human CC398 MSSA in univariate analysis ( Table 3). The majority of these genes (16/ 18) had functions related to antimicrobial or antiseptic resistances and was significantly associated to livestock MRSA isolates, with tetM being constant; however, two genes, namely scn and chp, were significantly associated with the human CC398 MSSA. In both methods (spa-typing and DNA microarrays), the corresponding phylogeny made appear MRSA strains as the most derived lineage (Figures 2 and 5).
Noteworthy, the characterization of 26 CC398 MSSA strains collected from human abroad or outside of mainland France showed genetic and phenotypic features matching those observed in isolates from our French collection (data not shown) demonstrating the worldwide representativity of the CC398 MSSA phenomenon described here.

Discussion
In this study, we described the largest published collection (in number and in period of time) of human CC398 MSSA isolates. Our data highlighted that CC398 MSSA are unspecialized pathogens able to colonize nares as well as to induce a large spectrum of infections. This population of S. aureus has specific phenotypic and genotypic signatures including resistance to erythromycin encoded by the ermT gene, and expression of staphylococcal complement inhibitor and the chemotaxis inhibitory protein, respectively encoded by scn and chp genes (major staphylococcal defense/evasion mechanism against the human inanimate immune system) [19]. On the other hand, the CC398 MRSA signature included a panel of antibiotic resistance genes, especially tetM. The genetic background of both populations was homogenous, that may come from common ancestress, which may have evolved in the settings of different selective pressures, explaining the acquisition of ermT, chp and scn in the first case, and antibiotic resistance genes in the second.
On the basis of our large strain collection, particular MSSA CC398 characters have been highlighted. Since they do not possess any clone-specific virulence factor, MSSA CC398 strains are able to induce a large spectrum of diseases, unlike ''profes-  sionalized'' S. aureus clones (such as TSST-1 producing clones involved in toxic shock syndrome [20] or PVL producing clones, involved in primary skin and soft tissue infection or necrotizing pneumonia). In our study, 89 MSSA strains were involved in 10 different types of infections, even if the 3 more frequent types (skin and soft tissue infections, bacteremia, and pneumonia) represented more than 70% of the overall. The presence of a specific toxin encoding gene appears to be scarce, all the more that a bias of recruitment surely exists in our collection, since FNRCS is recognized for its expertise in staphylococcal toxins, that probably prompts French labs to send to FNRCS isolates involved in apparent human staphylococcal toxinic syndromes. Nevertheless, acquisition of the PVL genes were found to be possible (in only 6 out of 105 MSSA strains), as already reported by Rasigade et al. in one case of necrotizing pneumonia [12], but also by Uhlemann et al. in skin infections [21]. On the other hand, the recent description of a prevalence of 64.3% (18/28) for lukS-PV and lukF-PV in community-acquired ST398 MSSA involved in SSTIs in China demonstrated that the acquisition of such virulence factor and/or spread of such isolates are a source of concern in view of the high fitness of ST398 MSSA for humans. Indeed, epidemiological data collected here revealed a high frequency of CC398 MSSA in MSSA population involved in human colonization and infections. The prevalence rate of CC398 MSSA was 24.7% in a local prospective study on nasal colonization, and reached 7.5% in a national prospective study on infective endocarditis. Such a high prevalence has been reported in some studies elsewhere over the world: in China, where two studies [15,22] [11]; in agreement with Uhlemann et al. reporting that ST398 MSSA was involved in 2.5% of bloodstream infections in Northern Manhattan [21]. Another interesting feature of MSSA CC398 is its frequent exclusive erythromycin resistant phenotype, encoded by the ermT gene: 86 of the 93 erythromycin-resistant MSSA CC398 strains were positive using a specific PCR. This gene has initially been detected in lactobacilli, streptococci and enterococci [24], and has recently been found in MRSA ST398 of animal origin, carried by a plasmid [24][25][26] (with a perfect sequence homology with the ermT gene isolated from Streptococcus pyogenes [24,25]); but also in MSSA ST398 strains isolated from humans in Spain, Belgium, and Northern Manhattan [10,14,21]. Vandendrissche et al., studying the restriction pattern of 5 MSSA CC398 isolates showed that ermT was integrated in the chromosome [14]. Intriguingly, in our collection of 53 animal CC398 MRSA strains, ermT was completely absent, unlike other erythromycin-resistance conferring genes. Since the antibiotic selective pressure may be present in both cases, this difference might traduce a different microbiological environment between human CC398 MSSA and animal CC398 MRSA (and thus different horizontal transfer possibilities), or different genetic (or epigenetic) backgrounds in these two groups (and thus different integration capacities for the same resistance genes). The close genetic background observed for the two MSSA and MRSA CC398 populations using microarray and spa-typing combined with the presence of human virulence factors (toxins, scn, chp) but poverty of antimicrobial resistance gene profiles (except ermT) in MSSA, and with the absence of specific human virulence factors but a high diversity of antimicrobial resistance gene profiles in MRSA, is suggestive of a different and independent evolutive history of the two populations. The level of diversity and segregation of spa-type and the specific gene content observed for the two populations strongly suggest that CC398 lineage originated in humans and then has undergone a radiation driven by the jump from humans to animals. The change of niches would have induced the loss of useless human virulence factors (such as scn or chp, that have no impact and confer no advantage in animals), and the acquisition of some others, such as various antibiotics resistance genes, including tetracyclin and methicillin resistance in connection to the strong and specific antimicrobial selection associated with livestock production. Recently, results of whole-genome sequencing approaches, strongly reinforced by the fact that CC398 MRSA showed decreased ability for human colonization transmission and virulence, support this point of view [21,[27][28][29]. Nevertheless, the lack of some ''missing links''(like for instance the unexplained different distribution of the ermT gene between MSSA and MRSA CC398 populations) could suggest another evolutive history. An external and undescribed third MSSA ST398 reservoir would have alternatively been at the source of the human niche and the animal niche, with respective specific selective adaptation to host and antibiotic pressure. Genetic recombination likely involving genes from coagulase-negative staphyphylococci or other S. aureus clones that beforehand colonized each of these niches, might have played a pivotal role.
Whole-genome analysis has been applied to the study of CC398 MSSA and MRSA and data published recently by Price et al. [27] and Uhlemann et al. [21]. Such approaches are relevant for studying genetic content and diversity among populations but cannot be applied to routine analysis, since they are expensive, and require bioinformatic allocated resources and time-consuming analysis. Although microarrays cover only a part of the S. aureus genome, we showed that data matched those obtained by wholegenome approaches and are easily obtained. Thus, microarrays allow the rapid and cheap identification of animal or human origin The Z branch groups 7 strains presenting a more frequent equipment in virulence genes and alleles than the others (lukE, sak, splA, splE, ssl07/set1 (MRSA252), hysA2 (All Other Than COL+USA300+NCTC)), but also 2 hyaluronate lyase genes (hysA2 (AllOtherThan COL+USA300+NCTC)). The Y branch gathers 8 strains with some of the virulence genes (lukS (ST22+ST45), lukX, chp, scn, set6-var4_11, ssl07/set1, ssl11/set2 (MRSA252)) and one of an adhesine (clfA (MRSA252)) less frequent. doi:10.1371/journal.pone.0068462.g004 of CC398 isolates. A detailed analysis of data shows, that a multiplex PCR targetting scn, chp, ermT could be useful for simply and efficiently discriminate animal and human origin of such isolates.
Because of a retrospective design, this work presents selection bias. Indeed, the strains described here were mainly collected passively by the CNRS, or within the framework of research projects: consequently, they may not be as representative of the global MSSA CC398 population as would be strains prospectively and randomly selected. Nevertheless, we described the tendency that CC398 MSSA strains might present in terms of epidemiology and characteristics.
On the basis of this work, we assume that MSSA CC398 is a frequent cause of human colonization and infections, suggesting a well-adapted fitness of this clone to humans. This contrasts with the behavior of CC398 MRSA, which seems to be more adapted to animal than to human beings [30,31]. We detected no patent differences in terms of adherence factors between these two populations, but found a more frequent distribution of scn and chp (but not sak neither entA that are often clustered on the same phages) in the MSSA group. These two genes are parts of the immune evasion cluster located on the conserved 39 end of betahemolysin converting bacteriophages, easily horizontally transferable between different S. aureus strains [19]. They have already been shown to be associated to S. aureus human host specificity for the ST398 lineage [11,21,27,32], but not only [33]. Since the microarrays covered a small part of the S. aureus genome and were only based on small probes designed to target conserved genetic regions, their use gives us only a partial view of the gene content and the gene polymorphism. Whole genome sequencing approach would allow us to open new possible lines of reflection. Besides, Uhleman et al., using preliminary comparison of three fully sequenced ST398 MSSA and MRSA genomes, identified mutations and deletions in genes encoding cell surface-bound proteins [21]. These differences must be now correlated with phenotypic features, confirmed by the construction of isogenic couples to demonstrate causality and clinical relevancy. Moreover, the study of the CC398 S. aureus at the phenotypic or regulation/ dysregulation system levels could give us other keys.

Materials and Methods
One hundred and five CC398 MSSA human isolates were included from the collection of the French National Center for Staphylococci ( Table 4). This collection included isolates voluntarily sent by French microbiology laboratories for further characterization because of particular antibiotic resistance phenotype or unusual infections, or in the setting of targeted studies. Thus, between March 2009 and May 2011, DNA microarrays (Identibac S. aureus Genotyping H, Alere) [25,34], used in routine, allowed the prospective identification of 94 isolates as members of CC398 on the basis of clonal assignment provided by the software of the kit. Eleven additional strains were included retrospectively after being tested using the same DNA microarrays for various purposes. Eighty-nine isolates were involved in infections and 11 in colonization (5 unknown). Standardized clinical and demographic information was collected for all CC398 MSSA strains that caused infections (n = 89): age and gender of each patient, infection type, date and site of isolation. In addition, to evaluate the prevalence of CC398 MSSA in specific populations, a collection of 85 MSSA isolated from human healthy carriers (nasal colonization) and a collection of 132 MSSA collected during a French national prospective study of human infective endocarditis (infections diagnosed in 2008) were screened for CC398. Fifty-three CC398 Table 3. Genes significantly associated with animal and human isolates belonging to the ST398 lineage. MRSA strains from animal origin were used as comparators. Finally, 26 CC398 MSSA strains collected from human abroad or outside of mainland France (Switzerland, n = 6; China, n = 5; US, n = 3; Algeria, n = 2; Denmark, n = 3; Martinique, n = 2; Dominican Republic, n = 2; India, n = 1; Madagascar, n = 1; La Reunion island, n = 1; French Guyana, n = 1) were included and characterized. Susceptibility to 16 antibiotic drugs was determined with the standard agar diffusion technique, as recommended by the French Society for Microbiology [35]. Genomic DNA was extracted using a standard procedure on Qiacube [20]. agr alleles and the presence of mecA gene were determined by specific PCR [20]. ermT gene was detected using specific primers and amplification protocol based on the description by Kadlec et al. [25,36], with modifications (see the Appendix). spa-typing was performed with the Ridom Staph Type standard protocol [37] and by using Ridom Staph Type softwareH (version 1.5) [38] (http://spaserver.ridom.de/), which automatically analyses spa repeats and assigns spa types. spa types were clustered into clonal complexes CCs (ie, spaCCs) using the integrated BURP (Based Upon Repeat Patterns) algorithm [37]. spa phylogenic trees were constructed based on this analysis using the UPGMA reconstruction algorithm. DNA microarray analyses (Identibac S. aureus Genotyping H, Alere) were performed according to the protocol described by Monecke et al. [34], in aim to characterize the S. aureus isolates. These microarrays detected 330 target sequences corresponding to 172 genes and their allelic variants. The data were interpreted based on the algorithm previously described [34]. A split network tree was generated using the modified Parcimony method and the SeaView program version 4 [39]. Distributions of the 172 genes were compared between the MSSA and MRSA groups. Univariate statistical analysis was performed using a two-tailed Fisher's exact test. p- values were corrected for multiple testing using the Holm-Bonferroni method. For each gene, isolates exhibiting ambiguous microarray results were removed from the analysis. The statistical significance threshold for all tests was set to 0.05. Because of the close genetic relationships between isolates, strong colinearity occurred in gene distributions; multivariate analysis was thus deemed irrelevant and was not performed. All statistical analyses were performed by means of the R software version 14.0 (The R Foundation for Statistical Computing, Vienna, Austria).
Appendix ermT gene specific PCR protocol. The ermT gene was detected using specific primers and amplification protocol based on the description by Kadlec et al. [25,36], but with modifications. For each probe, the mix was composed of 1,5 UI of Taqpolymerase EurobioH, 5 ml of reaction buffer EurobioH without MgCl2 (106), 2 ml of MgCl2 EurobioH (50 mM), 5 ml of dNTP (20 mM), 5 ml of each one of the 2 primers (2 pmol/ml), and 5 ml of DNA. The amplification stage was composed of 30 cycles, with a phase of denaturation of 30 sec at 94uC, of hybridation of 30 sec at 57uC, and elongation of 40 sec at 72uC. Amplification products were obtained at the expected size; one of them was sequenced, and the result was blasted with a 99% homology with the reference ermT gene found in GenBank (http://www.ncbi.nlm.nih.gov/ blast/Blast.cgi). Figure S1 Dendrogram (UPGMA method, squared representation) based on the homology degree of the spatypes of the 105 MSSA CC398 strains of human origin isolated in France. The homology between the spa repeats is of 80%. Of note, 6 other strains (6%) corresponding to 3 spa types (t3625 (n = 3), t4316 (n = 1), t8698 (n = 2)) were excluded from the analysis since it had a repeated sequence shorter than 5 repeats. And 1 strain were non typable by spa typing. (TIF) Figure S2 Phylogenic tree with 105 MSSA CC398 strains (modified Parsimony method, circular representation), based on the analysis of 319 genes and alleles by DNA microarrays. Since the PCR analysis of the agr alleles confirmed that all the strains were agr 1, the analysis of the results of DNA microarrays concerned only 319 genes and alleles. bla-negative strains are grouped in the bla2 cluster (light grey), bla-positive strains in the bla+ cluster (dark grey). (n: number of strains) The Z branch is composed of strains more frequently positive for some virulence factors. The green strain is positive for tsst1; the one in red for pvl; these over lined in blue for edinB; and these underlined for etd. Within the bla+ population 7 strains are isolated in a branch Z, presenting a more frequent equipment in virulence genes and alleles than the others (lukE, sak, splA, splE, ssl07/set1 (MRSA252), hysA2 (All Other Than COL+USA300+NCTC)), but also 2 hyaluronate lyase genes (hysA2 (AllOtherThan COL+USA300+NCTC)). (TIF) Figure S3 Dendrogram (UPGMA method, squared representation) based on the homology degree of the spatypes of the 53 MRSA CC398 strains of animal origin isolated in France. The homology between the spa repeats is of 86%. Of note, the spa type t1456, gathering 4 strains, was excluded of the analysis, having a too short repeated sequence. spaCC: spa clonal complex. (TIF) Figure S4 Phylogenic tree with 53 MRSA CC398 strains (modified Parsimony method, circular representation), based on the analysis of 319 genes and alleles by DNA microarrays. The strains carrying the type IV SCCmec cassette are grouped in a first cluster (light grey), and strains carrying the type V SCCmec cassette in a second one (dark grey). The 8 strains of cattle origin are separated from the others. Since the PCR analysis of the agr alleles confirmed that all the strains were agr 1, the analysis of the results of DNA microarrays concerned only 319 genes and alleles. (TIF)