Comparative Genomics of Community-Acquired ST59 Methicillin-Resistant Staphylococcus aureus in Taiwan: Novel Mobile Resistance Structures with IS1216V

Methicillin-resistant Staphylococcus aureus (MRSA) with ST59/SCCmecV and Panton-Valentine leukocidin gene is a major community-acquired MRSA (CA-MRSA) lineage in Taiwan and has been multidrug-resistant since its initial isolation. In this study, we studied the acquisition mechanism of multidrug resistance in an ST59 CA-MRSA strain (PM1) by comparative genomics. PM1’s non-β-lactam resistance was encoded by two unique genetic traits. One was a 21,832-bp composite mobile element structure (MESPM1), which was flanked by direct repeats of enterococcal IS1216V and was inserted into the chromosomal sasK gene; the target sequence (att) was 8 bp long and was duplicated at both ends of MESPM1. MESPM1 consisted of two regions: the 5′-end side 12.4-kb region carrying Tn551 (with ermB) and Tn5405-like (with aph[3′]-IIIa and aadE), similar to an Enterococcus faecalis plasmid, and the 3′-end side 6,587-bp region (MEScat) that carries cat and is flanked by inverted repeats of IS1216V. MEScat possessed att duplication at both ends and additional two copies of IS1216V inside. MESPM1 represents the first enterococcal IS1216V-mediated composite transposon emerged in MRSA. IS1216V-mediated deletion likely occurred in IS1216V-rich MESPM1, resulting in distinct resistance patterns in PM1-derivative strains. Another structure was a 6,025-bp tet-carrying element (MEStet) on a 25,961-bp novel mosaic penicillinase plasmid (pPM1); MEStet was flanked by direct repeats of IS431, but with no target sequence repeats. Moreover, the PM1 genome was deficient in a copy of the restriction and modification genes (hsdM and hsdS), which might have contributed to the acquisition of enterococcal multidrug resistance.


Introduction
Methicillin-resistant Staphylococcus aureus (MRSA) is a major public health problem worldwide. It was first isolated in 1961 as a nosocomial pathogen [1,2], now known as hospital-acquired MRSA (HA-MRSA). Another class of MRSA, designated community-acquired MRSA (CA-MRSA), emerged in the community from 1997 to 1999, posing a novel threat worldwide [1][2][3][4]. CA-MRSA infection usually occurs in children and young adults, or even the elderly, without exposure to hospital environments, and includes common skin and soft tissue infections (SSTIs) and occasionally life-threatening invasive infections, such as sepsis, necrotizing pneumonia and osteomyelitis [1][2][3][4][5].
A major CA-MRSA lineage in Taiwan is ST59 with SCCmecV and luk PV SF genes encoding PVL. Wang et al. [16] and Boyle-Vavra et al. [17] demonstrated that ST59 CA-MRSA in Taiwan was already resistant to multidrugs (including erythromycin, clindamycin, and chloramphenicol) when it began emerging in 1997, in contrast to the general understanding of CA-MRSA. Boyle-Vavra et al. [17] also characterized the SCCmec of Taiwanese ST59 CA-MRSA (representative strain, TSGH17) as SCCmec V T . The ST59 CA-MRSA lineage (with SCCmec type V T and luk PV SF) has also been reported in Singapore [18], Hong Kong [19], Japan [20,21], and Western Australia [22].
We further characterized the Taiwanese ST59 CA-MRSA lineage and found that it possessed a novel SCCmec with two dictinct ccrC genes (ccrC1 allele 2 and ccrC1 allele 8) [23][24][25]. We determined the entire SCCmec sequence of the ST59 CA-MRSA lineage (strain PM1) and tentatively designated it as SCCmecVII [24]; later, the SCCmec of the Taiwanese ST59 CA-MRSA lineage (strains TSGH17 and PM1) was reclassified as SCCmecV by the International Working Group on the Classification of Staphylococcal Cassette Chromosome Elements [6]. As for non-b-lactam resistance, three resistance determinants, ermB (encoding erythromycin/clindamycin resistance), aph(39)-IIIa (encoding kanamycin resistance), and aadE (encoding streptomycin resistance), were clustered in a region, previously named the drug resistance gene cluster (rgc) [23].
IS1216V is an enterococcal insertion sequence [26] and is rarely found in S. aureus. IS1216V-positive S. aureus cases include vancomycin-resistant MRSA (VRSA), in which IS1216V is inserted in a vancomycin resistance transposon (Tn1546) [27]. The rgc region of strain PM1 also carried IS1216V [23], raising the question of whether the rgc region of the Taiwanese ST59 CA-MRSA lineage (strain PM1) originated in enterococci through an IS1216V function.
In order to further understand the mechanism of multidrug resistance acquisition by PVL-positive ST59 CA-MRSA in Taiwan, we performed comparative genomics of strain PM1 and found that strain PM1 became multidrug resistant by acquiring two mobile genetic elements, an enterococcal IS1216V-mediated composite mobile element structure (MES PM1 ) on the chromosome and an IS431-mediated element (MES tet ) on a plasmid, besides b-lactam resistance by SCCmecV. We also searched for the genetic background of the ST59 CA-MRSA lineage, which allowed (or stimulated) the acquisition of multidrug resistance.

Results
Susceptibility to Non-b-lactam Agents of PVL-positive ST59/SCCmecV CA-MRSA Strains in Taiwan Drug resistance patterns of ST59/SCCmecV CA-MRSA strains isolated in Taiwan, compared to an ST59 MRSA-type strain USA1000, are summarized in Table 1. Half of the strains (including PM1) were resistant to erythromycin/clindamycin, kanamycin, streptomycin, chloramphenicol, and tetracycline, in addition to b-lactam agents, showing marked multidrug resistance (in contrast to USA1000). The rest showed five distinct patterns in terms of resistance to non-b-lactam agents: resistance to erythromycin/clindamycin, kanamycin, streptomycin, and chloramphenicol; resistance to erythromycin/clindamycin, kanamycin, streptomycin, and tetracycline; resistance to erythromycin/clindamycin, kanamycin, and streptomycin; resistance to chloramphenicol and tetracycline; and non-resistance (to any non-b-lactam agent).
Comparative Genomics of PVL-positive Multidrugresistant ST59/SCCmecV CA-MRSA (Strain PM1) The PM1 genome was analyzed by pyrosequencing (GenBank Accession number BAFA01000000) and compared with the genomes of ST398 MRSA human strain S0385 and PVLpositive ST8 CA-MRSA USA300, since ST398 MRSA carried SCCmecV [28], similar to PM1, and showed the highest similarity to ST59 MRSA [29] in terms of the seven housekeeping gene sequences used for multilocus sequence typing, and USA300 was one of the best characterized CA-MRSA [30]. The data are summarized in Figure 1. The alignment of PM1 contigs on the reference MRSA genome was the same in both cases ( Figure 1A); the similarities of the PM1 genome to the ST398 and ST8 genomes were both approximately 98%, albeit with divergence in, for example, the carriage of mobile elements ( Figure 1A). The PM1 genome was estimated to be at least 2.8 Mb. Strain PM1 carried a 25,961-bp plasmid (pPM1), encoding tetracycline resistance; plasmid DNA analysis of strain PM1 also demonstrated a single species of plasmid of 26.0 kb (Table 1).
When PM1 was compared with S0385 and USA300, the four regions were unique to PM1: i) SCCmecV with two ccrC genes (ccrC1 allele 2 and ccrC1 allele 8); ii) wSA1 PM1 , which was 70.2% homologous to wETA3, a member of the wSA1 family (wETA3 is eta-positive, while wSA1 PM1 was negative); iii) MES PM1 , a novel mobile element structure on the chromosome, encoding multidrug resistance; and iv) a tetracycline resistance plasmid pPM1. Moreover, nSAb on the PM1 genome had a large deletion.

Structure of MES PM1
The entire sequence of MES PM1 and its surrounding region was determined (GenBank Accession number AB699882); the structure is summarized in Figure 2. MES PM1 was 21,832 bp long and flanked by direct repeats of IS1216V at both ends. It was inserted into the att site (8-bp target sequence) within the sasK gene; the 8bp att sequence was duplicated at both ends of MES PM1 , indicating that MES PM1 is a large transposon. The target gene (sasK) with the 8-bp att sequence was present on the genome of ST59 MRSA USA1000, but not in strains S0385 and USA300 ( Figure 1A). Interestingly, MES PM1 was IS1216V-rich and contained three additional copies of IS1216V (a total of five copies of IS1216V), as shown Figure 2. The left side of MES PM1 was a composite transposon region consisting of Tn551 carrying ermB and Tn5405-like carrying aph(39)-IIIa and aadE. The composite transposon region was identical to the corresponding region of a plasmid (pLG2) of Enterococcus faecalis ( Figure 2); the entire sequence of pLG2 has not been reported.
Analysis of the right-side region of MES PM1 revealed an independent transposon (MES cat ) carrying cat (encoding chloramphenicol resistance). MES cat was 6,587 bp long and flanked by the inverted repeats of IS1216V; moreover, the 8-bp att sequence was duplicated at both ends of MES cat ( Figure 2). MES cat contained two additional copies of IS1216V. The central cat region in MES cat was highly homologous (99.5%) to the cat region of SAP084A plasmid in S. aureus.
Based on the data, we conclude that MES PM1 is a highly composite transposon, which originated in enterococci, but acquired an additional drug resistance gene (cat) in S. aureus. Another multidrug-resistant strain PM22 (Table 1) also possessed an MES structure (MES PM22 ) very similar to MES PM1 , although there was a small deletion within the structure (data not shown).

Structure of MES PM1 Segregants with Different Resistance Patterns
The MES PM1 -corresponding region of ST59 CA-MRSA strains with different resistance patterns (Table 1) was determined by sequencing ( Figure 3). MES PM1 seems genetically unstable, and recombination likely takes place between two direct repeats of IS1216V, resulting in deletion, which could explain the distinct resistance patterns of ST59 CA-MRSA.
For instance, strain PM9, which was resistant to erythromycin/ clindamycin, kanamycin, and streptomycin, but susceptible to chloramphenicol, seemed to be a segregant, generated through deletion (1) in Figure 3 (A, B). Likewise, strain PM8, which was resistant to chloramphenicol only, seemed to be a segregant, generated through deletion (2) in Figure 3 (A, C). Strain PM18, which was susceptible to all the non-b-lactam agents, seemed to be a segregant, generated through deletion (3) in Figure 3 (A, D).
MES PM1 segregants, MES PM9 ( Figure 3B) and MES PM8 ( Figure 3C), may also behave as transposons, since they were flanked by direct repeats of IS1216V at both ends with att sequence duplication and inserted into the sasK gene, just like MES PM1 . MES PM18 (corresponding to IS1216V), which was inserted into the Figure 1. Genome comparisons of ST59 MRSA strain PM1 with two other S. aureus genomes. In A, two gray circles show the complete genomes of ST8 CA-MRSA strain USA300 FPR3757 (2,872,769 bp) and ST398 MRSA strain S0385 (2,872,582 bp); gaps in the genome circles indicate USA300 or S0385 sequences not present in PM1. The sasK gene with the MES PM1 integration site (att) is shown outside the gray circles (on the USA1000 genome circle), since the USA300 and S0385 MRSA strains do not possess the sasK gene. The outermost (white) circle represents the alignment of PM1 contigs and provides PM1 genome information, including drug resistance traits, pathogenetic islands, bacteriophages, and genomic islands nSAa and nSAb. Plasmid pPM1 is shown in the right bottom of the figure. Color region in the genome: orange, orfX with the att sequence for SCCmec; green, genetic elements with diversity; brown, genetic elements present in USA300 or S0385 (but not in PM1); pink, genetic elements present in PM1 (but not in USA300 or S0385). In B, relevant genotypes of PM1 is summarized. Drug resistance (r): Ery/Cli, erythromycin/ clindamycin; Kan, kanamycin; Stm, streptomycin; Chl, chloramphenicol; Oxa, oxacillin; Tet, tetracycline. Virulence factors: SEB, staphylococcal enterotoxin B; SEK, staphylococcal enterotoxin K; SEQ, staphylococcal enterotoxin Q; PVL, Panton-Valentine leukocidin. doi:10.1371/journal.pone.0046987.g001

Structure of pPM1
The entire sequence of pPM1 was determined (GenBank Accession number AB699881) and its structure is summarized in Figure 4A. pPM1 was a novel composite plasmid, consisting of a replication (rep) region (homologous to pSK156 plasmid), cadmium resistance (cad) region (homologous to SAP063A plasmid), ampicillin resistance (bla) region (homologous to SAP063A plasmid), and tetracycline resistance (tetK) region (homologous to SAP014A plasmid). The tetK-carrying region was 6,025 bp long and was flanked by direct repeats of IS431 at both ends; this structure was designated MES tet . There was no att sequence duplication at either end of MES tet .

Analysis of Tetracycline-susceptible Strains
Of three tetracycline-susceptible strains (Table 1) examined, two strains, PM11 and PM18, carried a pPM1-related plasmid (pPM11 and pPM18, respectively). Entire sequence analysis revealed that the two plasmids carried one copy of IS431 only at the MES tet insertion site on pPM1, as shown in Figure 4B; therefore, pPM11 and pPM18 could be a segregant of pPM1 generated through a recombination/deletion event between two direct repeats of IS431. The remaining strain (PM12) lacked pPM1 plasmid.

Conjugative Transfer of Drug Resistance
Conjugative transfer of drug resistance was performed between PM1 (donor strain) and S. aureus RN2677 (recipient strain) by filter mating. As shown in Table 2, resistance to ampicillin, tetracycline, and cadmium was transferred to S. aureus RN2677 with equal frequency; all transconjugants obtained carried a 26-kp penicillinase plasmid (pPM1), as expected. However, no MES PM1 -associated transconjugants (exhibiting resistance to erythromycin, clindamycin, chloramphenicol or streptomycin) were obtained.

Restriction and Modification Genes (hsdS and hsdM) in Strain PM1
Genomic island vSAb of MRSA strains, e.g. USA300, carried intact hsdS and hsdM, while vSAb of PM1 carried truncated hsdS and lacked hsdM ( Figure 5), although genomic island vSAa of PM1 carried an intact set of hsdS and hsdM, similar to USA300 (Figure 1). Lack of a set of hsdS and hsdM in PM1 may have contributed, in part, to PM1's potential to acquire foreign DNA.    originate in enterococci (such as E. faecium and E. faecalis) [26,33,34], strongly suggesting the origin of MES PM1 in enterococci.
In E. faecium, IS1216V is considered to be part of a large mobile element containing Tn1546 with vanA (encoding vancomycin resistance) [35], as shown in Figure 6B (a); however, a possible mobile structure with att duplication has not been reported. Moreover, in E. faecium, a large mobile element (Tn5506), which contains Tn1546 and is flanked by hetero terminal sequences, IS1216V and IS1252, with 8-bp att duplications at both ends, has been described [32], as shown in Figure 6B (b); however, the att of Tn5506 was different from the att of MES PM1 ( Figure 6A and B [b]), and in the case of Tn5506, another copy of IS1216V was located adjacent to IS1252 ( Figure 6B [b]).
Transfer of the van genes (such as vanA) from enterococci to MRSA would be a great public health concern and threat, because vancomycin is the first-line agent for MRSA infections [3,36,37]. Unfortunately, Tn1546 has already emerged in MRSA [27], as shown in Figure 6C; in this case, IS1216V is present near the van genes (although possible IS1216V repeats and att duplication have not been reported).
Therefore, this study demonstrates the first IS1216V-mediated mobile element structure with terminal direct repeats of IS1216V with att duplication at both ends. Interestingly, MES PM1 also contained another mobile element structure (MES cat ) within the structure, which was flanked by inverted repeats of IS1216V with att duplication at both ends. Chloramphenicol resistance in MRSA is usually encoded by a plasmid [31,38]; however, in this case, it was encoded by MES cat (as a part of MES PM1 ) on the chromosome. The chloramphenicol resistance region seemed to originate in S. aureus; therefore, MES PM1 is a highly composite structure with mobile elements originating in Enterococcus and Staphylococcus species.
This study clearly concludes that the acquisition mechanism of multidrug resistance in ST59 CA-MRSA in Taiwan is quite different from many other MRSA strains, but shows some similarity to VRSA cases, and that the MES PM1 acquisition event occurred most probably around 2007 or earlier in Taiwan, and then ST59 CA-MRSA carrying MES PM1 spread in Taiwan, followed by segregation events generating different resistance patterns.
The reason why ST59 CA-MRSA in Taiwan acquired MES PM1 is not known. The type I restriction-modification system, composed of HsdR (restriction), HsdM (modification), and HsdS (sequence specificity), is known to block horizontal gene transfer from other species into S. aureus, as well as to limit DNA exchange between the different lineages of S. aureus [39,40]. Deficiency in the type I restriction-modification system would allow foreign DNA acquisition with high frequency. For example, a bovine pathogenic S. aureus strain, which is naturally mutated in both hsdS genes, has been reported displaying a ''hyper-recipient'' phenotype to acquire DNA from Enterococcus species [41]. In the case of ST59 CA-MRSA (strain PM1) in Taiwan, a copy of hsdM and hsdS in nSAb was truncated, which might contribute to the acquisition of foreign DNA from enterococci such as IS1216V and MES PM1 ; further study is needed for elucidation.
In this study, a conjugative penicillinase plasmid (pPM1) in strain PM1 was successfully transferred by filter mating, while MES PM1 was not, suggesting that MES PM1 may not be a conjugative transposon.
In strain PM1, we also identified a novel mosaic penicillinase plasmid encoding tetracycline resistance (pPM1). The tetracycline resistance region (MES tet ) could be a novel transposon flanked by IS431, which most probably originated in S. aureus. This structure also generated segregants (tetracycline-susceptible penicillinase plasmid), resulting in strains PM11 and PM18.
Regarding other genetic structures in ST59 CA-MRSA in Taiwan, we confirmed the sequence of SCCmecV of strain PM1 (formerly SCCmecVII) [24] and the sequence of PVL phage (w5967PVL; wSA2 PM1 in this study), which was originally reported with another strain (JCSC5967) by Zhang et al. [42]; they were 99.9% homologous.
During the course of this study, Huang et al. described the genome sequence of ST59 CA-MRSA strain M013 in Taiwan [43], although they did not describe the plasmid(s). According to their data, their strain (M013) is not multidrug-resistant; most probably it corresponds to strain PM18 in this study, a minor derivative of PM1 carrying only IS1216V at the MES PM1 insertion site and showing no resistance to non-b-lactam agents. M013 also does not carry wSA1 PM1 . In addition, three genes (sasA, fnbA, and sdrC) in M013 have more than 24-bp deletions or insertions. The comparison between the PM1 and M013 genomes is summarized in Table S1.
In conclusion, an IS1216V-mediated mobile element structure (MES PM1 ), encoding multidrug resistance and originating in enterococci, has emerged in PVL-positive ST59 CA-MRSA (e.g. strain PM1) in Taiwan. Although IS1216V, related to vancomycin resistance transposon Tn1546, has been reported in MRSA, MES PM1 represents the first mobile element flanked by IS1216V direct repeats with att duplication at both ends. MES PM1 also includes additional MES cat flanked by IS1216V inverted repeats with att duplication at both ends, although the cat region originated in S. aureus, indicating a highly composite formation in its evolution. The deficiency of a copy of hsdM and hsdS (in nSAb) in strain PM1 might contribute to the acquisition of foreign DNA, such as IS1216V and MES PM1 , from enterococci.

Bacterial Strains
Twenty-four PVL-positive ST59/SCCmecV CA-MRSA strains were collected from 2000 to 2006 in the Bacteriology Laboratory, National Taiwan University Hospital. These strains were isolated from patients with SSTI, bacteremia, respiratory tract infection, and surgical wounds [23]. USA1000 is an ST59 MRSA-type strain (of the SmaI PFGE type) isolated in the United States, and was kindly provided by L. K. McDougal and L.L. McDonald.

Pyrosequence Analysis of the Genome
The PM1 genome was analyzed by pyrosequencing using a genome sequencer FLX system (Roche Diagnostics, Branford, CT, USA). It yielded 93 Mb raw sequences by 236,943 reads, corresponding to approximately 33.2-fold of the genome size. Contigs were then assembled according to the two genomes of ST8 MRSA USA300 FPR3757 (GenBank accession number NC_007793) and ST398 MRSA human strain S0385 (GenBank accession number AM990992). Open reading frames were searched for using the software in silico MolecularCloning (version 4.2) (In Silico Biology, Yokohama, Japan) or DNAman software package (Version 6) (Lynnon Biosoft, Quebec, Canada).

Entire Sequencing of Mobile Elements and a Plasmid
The gaps between contigs, related to mobile elements, were filled by PCR and sequencing. We also assembled contigs using an LA PCR in vitro cloning kit (Takara Bio, Otsu, Japan) according to the manufacturer's instructions. In brief, after digestion with suitable restriction enzymes and ligation with the corresponding cassette adapters, amplification was performed with cassette primers and target-specific primers.
PCR Analysis of the MES PM1 Structure in Other PVLpositive ST59 CA-MRSA Strains Based on the determined entire sequence of MES PM1 , a series of primer sets was designed (Table 3) and used to screen the genetic constitution of MES PM1 . In PCR reactions, 10% glycerol or 4% DMSO was added in order to dissolve the secondary structure of IS1216V.

Susceptibility Testing
Antimicrobial susceptibility tests were performed by the agar dilution method with Hueller-Hinton agar (Difco, Sparks, MD, USA) according to the guidelines of the Clinical and Laboratory Standards Institute [44].

Conjugative Transfer
Donor strains were mated with S. aureus RN2677 (recipient strain, which is restriction-negative and resistant to rifampicin and novobiocin) on membrane filters on tryptic soy agar (Difco) (filter mating), as previously described [45]; RN2677 was used as a recipient because it carried no plasmids and had non-transmissible drug resistance (recipient) markers. The resistance genes of transconjugants were examined by PCR as previously described [23].

Plasmid Analysis
Plasmid DNA was isolated using a plasmid midi kit (Qiagen, Hilden, Germany) and lysostaphin (Wako Pure Chemicals, Osaka, Japan), following the manufacturer's instructions.