Genetic Diversity of Arginine Catabolic Mobile Element in Staphylococcus epidermidis

Background The methicillin-resistant Staphylococcus aureus clone USA300 contains a novel mobile genetic element, arginine catabolic mobile element (ACME), that contributes to its enhanced capacity to grow and survive within the host. Although ACME appears to have been transferred into USA300 from S. epidermidis, the genetic diversity of ACME in the latter species remains poorly characterized. Methodology/Principal Findings To assess the prevalence and genetic diversity of ACME, 127 geographically diverse S. epidermidis isolates representing 86 different multilocus sequence types (STs) were characterized. ACME was found in 51% (65/127) of S. epidermidis isolates. The vast majority (57/65) of ACME-containing isolates belonged to the predominant S. epidermidis clonal complex CC2. ACME was often found in association with different allotypes of staphylococcal chromosome cassette mec (SCCmec) which also encodes the recombinase function that facilities mobilization ACME from the S. epidermidis chromosome. Restriction fragment length polymorphism, PCR scanning and DNA sequencing allowed for identification of 39 distinct ACME genetic variants that differ from one another in gene content, thereby revealing a hitherto uncharacterized genetic diversity within ACME. All but one ACME variants were represented by a single S. epidermidis isolate; the singular variant, termed ACME-I.02, was found in 27 isolates, all of which belonged to the CC2 lineage. An evolutionary model constructed based on the eBURST algorithm revealed that ACME-I.02 was acquired at least on 15 different occasions by strains belonging to the CC2 lineage. Conclusions/Significance ACME-I.02 in diverse S. epidermidis isolates were nearly identical in sequence to the prototypical ACME found in USA300 MRSA clone, providing further evidence for the interspecies transfer of ACME from S. epidermidis into USA300.


Introduction
Staphylococcus epidermidis is a ubiquitous commensal of the human skin and mucosal surfaces and a major cause of indwelling medical device infections. This organism is notorious for its capacity to accumulate antibiotic resistance determinants and to produce biofilm, making infections caused by this opportunistic pathogen particularly difficult to treat. The large gene pool of antibiotic resistance and virulence determinants in S. epidermidis is shared with other more pathogenic species such as S. aureus. In particular, multidrug-resistant conjugative plasmids and staphylococcal chromosome cassette mec (SCCmec) elements conferring b-lactam resistance are transferred frequently between S. epidermidis and S. aureus, enabling rapid evolution and adaptation against antibiotic selection pressure [1,2,3].
Many S. epidermidis strains also carry the arginine catabolic mobile element (ACME), a novel genomic island that may contribute to enhanced capacity of this species to colonize the human skin and mucosal surfaces [4]. The horizontal transfer of ACME from S. epidermidis to S. aureus is thought to be central in the evolution of the highly transmissible community-associated methicillin resistant S. aureus (CA-MRSA) clone USA300 [4]. In USA300, ACME is integrated at the orfX site downstream of SCCmec, and is flanked by repeat sequences typical of SCC cassettes [4,5]. Mobilization of ACME is believed to be mediated by the cassette chromosome recombinases (ccrAB) encoded by SCC element [5,6]. This element has been found also in diverse S. aureus genetic backgrounds, suggesting frequent horizontal dissemination [4,5,6,7]. ACME contains two gene clusters (arc encoding a secondary arginine deiminase system and opp-3 encoding an ABC transporter) that are homologs of virulence determinants in other bacterial species [5]. However, ACME was shown not to contribute to the severity of necrotizing pneumonia or skin abscess using rat infection models [8], a finding consistent with the presence of this element among various commensal Staphylococcus species [4,9]. Using highly sensitive in vivo competition assays, ACME was shown to confer bacterial survival advantage in a rabbit bacteremia model and a mouse gastrointestinal colonization model [5,10]. Taken together, the data suggest that ACME may play a role in bacterial transmission among susceptible hosts by contributing to bacterial growth and survival.
Although ACME was found in diverse S. epidermidis [4], the genetic diversity of ACME among S. epidermidis has not yet been characterized. It is also not clear the extent to which S. epidermidis serves as a reservoir of ACME for horizontal transfer to related pathogenic species such as S. aureus. We report herein an analysis of 127 genotypically diverse isolates of S. epidermidis from worldwide sources. The results show extensive genetic variations within ACME. ACME allotypes are often found in association with diverse allotypes of SCCmec. The most prevalent allotype of ACME among S. epidermidis is nearly identical to prototypical ACME found in USA300, suggesting that ACME transfers from S. epidermidis into the epidemic USA300 clone.

Restriction Fragment Length Polymorphism (RFLP) Analysis of Arginine Deiminase (arc) Gene Cluster
Chromosomal DNA of S. epidermidis strains was digested with ClaI and the resulting fragments were separated by a electrophoresis in a 1% agarose gel in 1x Tris-acetate-EDTA buffer at 30 volts for 17 h. DNA fragments were transferred by vacuum blotting to nitrocellulose membranes as previously described [11] and hybridized with DNA probes for a region encompassing ACME-encoded arcC and arcB genes using primers arcC-1, 59-AATTTATCAGCCTGCTCTTTTGT-39, and arcB-1, 59-AA-AACAGGTAATCCACATACA-39. High stringency hybridization was performed with ECL direct prime labelling and detection systems (Amersham Biosciences, Buckinghamshire, U.K.), using a wash buffer with 0.1 X Standard Sodium Citrate (SSC). The different ClaI-arcCB hybridization band patterns found in this study were identified using arabic numbers (e.g., ClaI-arcCB types 1-7).

ACME PCR Scan
S. epidermidis isolates were screened for the presence of ACME using PCR-based assays using the primer pairs AIPS.27-AIPS.28 (arcA) and AIPS.45-AIPS.46 (opp3A), as previously described [5]. The arcA and opp-3A genes are surrogate markers of the arc gene cluster encoding for an arginine deiminase system and opp-3 encoding for an ABC transporter system, respectively. S. epidermidis isolates containing arcA and/or opp-3 clusters were further characterized by a PCR-based scanning/tiling method that allows for identification of variations in ACME gene content and gene synteny, as previously described [5]. This method is based on 31 individual PCR reactions using distinct primer pairs designed to generate 1-2 kb PCR fragments that overlap with one another at both ends for complete scanning coverage of the prototype ACME found in USA300. PCR scan patterns were classified into three ACME allotypes: (1) ACME-I contains both the arc and opp-3 gene clusters; (2) ACME-II contains arc but not opp-3; and (3) ACME-III contains opp-3 but not arc. Distinct PCR scan patterns within each ACME allotype are given subtype designations (e.g. ACME-I.02). Sequencing of the PCR amplicons from the PCR scan among 5 ACME-I.02-positive S. epidermidis isolates from different countries were performed using primers as previously described [5].

SCCmec Typing
S. epidermidis isolates were characterized for the two central elements of the staphylococcal cassette chromosome mec (SCCmec), namely, the ccr complex encoding for recombinases and the mec complex encoding for broad spectrum b-lactam resistance. The multiplex PCR strategy, M-PCR 1, was used to identify the 5 types of ccr gene complex, and M-PCR 2 to identify class A to class C mec complex, as previously described [12].

Ccr-mediated Excision of ACME
A tetracycline-selectable temperature-sensitive plasmid, pSR2, containing the ccrAB2 gene complex, was electroporated into S. epidermidis strain 1457. Se1457(pSR2) was passaged for three days in tryptic soy broth (TSB) supplemented with 10 mg/ml of tetracycline at 30uC. Growth at the non-permissive temperature of 42uC in antibiotic-free TSB resulted in loss of pSR2 in the excision mutants. Individual colonies were screened for excision and loss of ACME by assaying for the loss of the arcA gene. Confirmation of ACME excision was performed by Southern hybridization of SmaI DNA restriction fragments after pulsed-field gel electrophoresis with the probe encompassing arcC and arcB as described above [13]. In vitro growth rate of S. epidermidis 1457 and its isogenic ACME-excision mutant was determined in tryptic soy broth as measured by OD 600 .

MLST and eBURST
Multilocus sequence typing (MLST) was performed, based on the sequencing of internal fragments of seven housekeeping genes, and using the revised scheme described by Thomas et al. [14]. The most likely patterns of evolutionary descent in the collection were assessed using the eBURST algorithm (http://eburst.mlst.net), using previously validated parameters [15]. Clonal complexes were represented by the abbreviation CC, and singletons were represented by the abbreviation S. CC2 was subdivided into clusters I and II, and cluster II was further separated into subclusters as previously described [15,16].

Construction of Evolutionary Models and Estimation of Independent ACME Acquisitions
An evolutionary model illustrating the number of ACME acquisitions was constructed, based on the evolutionary relation-ships as defined by eBURST and ACME typing as defined by the PCR scanning strategy (see above). The number of independent ACME acquisitions was estimated based on the following assumptions: (i) there is a low probability of ACME excision, (ii) there is a low probability that the exact same mutation occurs twice; and (iii) for ACME acquisition/excision to occur, a gene coding for a recombinase (ccrAB or ccrC) must be present in the chromosome. The number of acquisition of SCC elements (SCCmec and SCC non-mec) was estimated using the same methods.

Statistical Analysis
Two-sided chi-square test statistics were used for between group comparisons (Stata, version 9, College Station, Texas).

Distribution of ACME and SCCmec among S. epidermidis Lineages
Of the 127 S. epidermidis isolates selected to represent the broad genetic and geographic diversity of the species, 52% (65/127) contained either the ACME-encoded arcA and/or opp-3 gene clusters. Presence of arcA and/or opp-3 gene clusters did not correlate with isolates recovered from infection sites or colonization sites. Using a revised MLST scheme [14,16], 86 distinct sequence types (ST) were identified among the 127 isolates ( Figure 1 and Table 1). The majority of the STs (50 of 86) were closely related and clustered into a single clonal complex, CC2. Of note, 65% (57/88) of isolates belonging to CC2 contained ACME-encoded arcA or opp-3, whereas only 21% (8/39) of isolates belonging to non-CC2 clonal complexes contained these genes (P,0.001).

Excision of ACME by CcrAB in S. epidermidis
Horizontal transfer of ACME in S. aureus is mediated by SCCencoded cassette recombinases (ccr), which catalyze the site-specific recombination between repeat sequences flanking the element and an attB site within orfX [5]. To test whether SCC-encoded ccr could mobilize ACME in S. epidermidis, we provided in trans ccrAB2 via plasmid pSR2 in a clinical isolate S. epidermidis 1457 and assayed for excision of ACME by pulsed-field gel electrophoresis of SmaIdigested chromosomal DNA and hybridization with ACME-specific arcCB probe ( Figure 2). A 240-kb SmaI fragment hybridized with the

DEN112
Denmark   arcCB probe in the S. epidermidis 1457 parental strain, whereas a corresponding 180-kb fragment did not hybridize with the arcCB probe in the ACME-excision mutant. This corresponds to a deletion of approximately 60-kb DNA fragment, which probably contains another SCC-like element mobilizable by CcrAB in addition to ACME (typically 30-kb in size). This is reminiscent of a CcrABmediated mobilization of 55-kb of DNA encompassing both SCCmec and ACME in S. aureus clone USA300 [5]. There was no difference in the in vitro growth rate of S. epidermidis 1457 and its ACME excision mutant, confirming the finding in S. aureus that carriage of ACME does not engender a biological fitness cost [5].

Genetic Diversity of ACME in S. epidermidis
For initial characterization of the genetic diversity of ACME found in S. epidermidis isolates, chromosomal DNA were digested with a frequent-cutting restriction enzyme ClaI and screened for restriction fragment length polymorphisms near the arc gene cluster by hybridization with a probe encompassing the ACME-encoded arcC and arcB genes. We selected 39 (64%) isolates positive for ACME-encoded arcA for this analysis. Since a single restriction site for ClaI is observed within arcCB fragment (inside arcC), the hybridization band patterns obtained typically contained two bands, a constant 1.2 kb band, corresponding to a ClaI site downstream of arcC, and the other band varying in size between 7 and 9 kb, corresponding to a variable ClaI site upstream of this gene (Figure 3). A total of seven different ClaI-arcCB DNA restriction band patterns were identified among the 39 S. epidermidis strains carrying ACME (Table 1). Of these, ClaI-arcCB pattern-6 was the most common (n = 19), followed by pattern-4 (n = 8), pattern-7 (n = 6), patterns -2 and -5 (n = 2 each) and patterns -1 and -3 (n = 1 each).  MLST, multilocus sequence typing [14]; 7-loci allelic profile listed in parenthesis (arcC-aroE-gtr-mutS-pyrR-tpiA-yqiL); 4 ACME PCR scan method [5] defines genetic variants within 3 ACME allotypes: ACME-I contains arc and opp-3 gene clusters; ACME-II contains arc but not opp-3; and ACME-III contains opp-3 but not arc. ACME-neg is negative for both arc and opp-3. Amplicons from the ACME PCR scan for 5 underlined ACME-I.02 variants were used for DNA sequencing and construction of 25-kb contigs. 5 NH, no hybridization with arcCB probe; ClaI-arcCB banding patterns shown in Figure 3; 6 SCCmec typing to identify class A, B, C, and other non-typeable (NT) mec gene complex, and 5 types of ccr gene complex, ccrAB1, ccrAB2, ccrAB3, ccrAB4, ccrC. SCCmec type I contains B/ccrAB1; type II A/ccrAB2; type III A/ccrAB3; type IV B/ccrAB2; and type V C/ccrC. doi:10.1371/journal.pone.0007722.t001 Table 1. Cont.
To further characterize genetic diversity of ACME among the S. epidermidis isolates, we used a PCR-based scanning method for amplification of 30 overlapping segments of 1-2 kb in length spanning to the entire archetypal ACME found in USA300; this method allows for a comprehensive assessment of gene content, gene synteny and other structural features of ACME [5]. Among the 65 isolates containing either ACME-encoded arcA or opp-3 genes, 39 distinct PCR scan patterns or variants were identified (Table 1). Of these, 66% (43/65) were classified as ACME-I allotype because they contained both arc and opp-3 gene clusters. Only a single subtype of ACME-I, designated ACME-I variant 02 (abbreviated ACME-I.02), was found in 42% (27/65) of the isolates; the remaining ACME-I subtypes (i.e. subtypes ACME-I.03 to ACME-I.18) were represented by one isolate each. Additionally, there were 18 distinct PCR scan patterns represented by one isolate each that were classified as ACME-II (containing arc but not opp-3 gene cluster) and 4 patterns that were classified as ACME-III (containing opp-3 but not arc gene cluster). ACME-I.02 was found in S. epidermidis isolates recovered from diverse locations, including Argentina, Denmark, Iceland, Hungary, Portugal, Poland and United States. Sequencing of the amplicons that resulted from the PCR scan of five ACME-I.02positive S. epidermidis isolates from different countries yielded a 24,605-bp contig encompassing both the arc and opp-3 gene clusters (Table 1). This ACME-I.02 contig from diverse S. epidermidis differed from the archetypal ACME type I variant 01 (abbreviated ACME-I.01) found in USA300 in only 11 nucleotides that corresponded to the open reading frames SAUSA300_0048 to SAUSA300_0077 (GenBank accession number NC007793). From the eleven variant sites found, 10 were single nucleotide polymorphisms (6 non-synonymous mutations, 3 synonymous mutations, 1 mutation in non-coding region), and one site involved an inframe insertion/deletion of a 6-bp within a transposaseencoding sequence (SAUSA300_0060). Altogether, these results showed that the prevalent ACME-I.02 type in S. epidermidis is nearly identical to the ACME-I.01 found in USA300, indicating a recent common origin.
Estimated Frequency of Horizontal Acquisition of ACME-I.02 in the CC2 Lineage All 27 S. epidermidis isolates containing ACME-I.02 belonged to the prevalent CC2 lineage (Table 1). ACME-I.02 was distributed unevenly between the two clusters that comprised CC2: cluster I of CC2 (abbreviated CC2-I) contains 21 (78%) isolates, and cluster II of CC2 (CC2-II) contains 6 (22%) isolates. To estimate the  number of independent horizontal acquisitions of ACME-I.02 within CC2, an evolutionary model was constructed based on the genetic relationships revealed by the eBURST when applied to MLST data (see Figure 4 and Table 1). According to the model proposed, ACME-I.02 was estimated to have been acquired at least on 15 different occasions by strains belonging to CC2 lineage, suggesting frequent mobility of ACME-I.02 within but not beyond this S. epidermidis lineage.

Discussion
In the present study we found that 52% (65/127) of S. epidermidis isolates representing the broad genetic and geographic diversity of the species contained one of three ACME allotypes. There were extensive genetic diversity found in ACME islands of S. epidermidis, with 39 distinct variants identified by a PCR-based scanning method. Only one of these variants was represented by more than one isolate in the S. epidermidis population; this variant, ACME-I.02, contained both the arc and opp-3 gene clusters. All the other variants of ACME are likely to derive from the ancestral ACME-I.02 variant. ACME-I.02 was found in 21% (27/127) of the isolates recovered from seven countries. Importantly, a 24-kb DNA fragment of ACME-I.02 in five S. epidermidis isolates was virtually identical to a homologous contig of the ACME-I.01 variant found in USA300, suggesting the interspecies transfer of ACME from S. epidermidis into USA300. A similar observation was made for the interspecies transfer of SCCmec type IV from S. epidermidis strains to S. aureus, indicating that S. epidermidis provides a reservoir for genetic exchange with S. aureus [3].
The observation that the nearly identical ACME-I.01 and ACME-I.02 variants are prevalent among the most widely disseminated lineages of S. aureus (i.e. USA300) and S. epidermidis (i.e. CC2) suggests that these specific ACME allotypes may confer a particularly high biological fitness advantage. Several lines of evidence suggest that this fitness advantage is not associated to a higher capacity of causing disease. The high prevalence of ACME among Staphylococcus species that are common commensals of the human skin, e.g. S. epidermidis, S. capitis and S. haemolyticus [4,9], together with the fact that ACME was not found specifically associated with disease-causing isolates when compared to  I, CC2-II6, CC2-II5, CC2-II85,  CC2-II89). Each dot represents a strain with specific characteristics with respect to ST and content of the different mobile genetic elements, namely, SCCmec (represented by types I-VI), SCC non-mec (represented by the allotype of ccr) and ACME types (represented by Roman numbers followed by Arabic numbers). The occurrence of genetic events involving a single MLST locus variation and/or SCC and ACME acquisition/deletion are indicated by arrows and the elements involved in the event are shown next to the arrow. Blue and white dots represent strains found within the collection studied, and black dots represent hypothetical S. epidermidis strains. ACME I.02 acquisition is represented in blue. doi:10.1371/journal.pone.0007722.g004 colonizing isolates of S. epidermidis and S. haemolyticus [9], suggest that this element is unlikely to contribute to the capacity of coagulase-negative staphylococci to cause disease in humans. Moreover, ACME was found not to contribute to the capacity of USA300 to cause skin abscess and necrotizing pneumonia in rat infection models [8]. An often overlooked feature of bacterial pathogenicity is the capacity to grow and survive within the host, thereby allowing for enhanced transmission. In this regard, ACME was shown to contribute to the growth and survival of USA300 in the rabbit and in the gastrointestinal tract of the mouse [5,10]. Furthermore, USA300 was found to be frequently recovered from axilla, inguinal, perineum and rectum [17,18], which are not common sites of colonization for S. aureus. These body sites are usually colonized ubiquitously by S. epidermidis, S. capitis and S. haemolyticus, which exhibit a high frequency of ACME carriage [19,20]. The acquisition of ACME by S. aureus might have allowed for the expansion of its typical colonization niches, providing new opportunities for transmission and dissemination. Altogether, these findings point to a potential role of ACME in conferring a fitness advantage for colonization and transmission rather than an enhanced capacity for infection.
These data also provide evidence for extensive intraspecies transfer of ACME, SCCmec, and other SCC elements among S. epidermidis, perhaps owing to the fact that 81% of the S. epidermidis population carry ccr gene complexes (Table 1 and Figure 1). Particularly, a high rate of intraspecies transfer of ACME-I.02 variant within the CC2 lineage was observed, which could be explained not only by the multiple ccr gene complexes frequently carried by these strains, but also to an enhanced capacity to accommodate multiple mobile elements, including ACME, SCCmec and other SCC elements, within the orfX [13]. This together with a high rate of recombination events, previously observed to occur frequently within CC2 lineage [15], may allow for the generation of the extensive genetic diversity among ACME islands afforded by strains belonging to this clonal lineage.
Although the role of S. epidermidis species as reservoir and donor of virulence and antibiotic resistance determinants to S. aureus is becoming unequivocal, the circumstances that favor the transfer of SCC elements between these two species is not completely understood and should be the focus of future studies. The understanding of the mechanisms and physiological conditions in which such transfer occur would provide us with fundamental tools to help to prevent the emergence of epidemic MRSA strains such as USA300.