ABSTRACT
We examined 97 strains of methicillin-resistant Staphylococcus aureus (MRSA) isolated between 1979 and 1985, the period of time when the appearance of MRSA strains increased, and we determined that these strains are distinct from the MRSA clones predominating in today's Japanese hospitals. Type IV staphylococcal cassette chromosome mec (SCCmec) strains were the most frequent, comprising 53.6% of all strains, followed by type I (22.7%) and type II (21.6%) SCCmec strains. Among the type IV SCCmec strains, the frequencies of two new subtypes, type IV.3 (IVc) and type IV.4 (IVd), were very high, comprising 38.1 and 10.3% of all strains, respectively. Forty-four of the 97 strains (45.3%) were Panton-Valentine leukocidin (PVL) positive. Among the PVL-positive strains, sequence type 30 (ST30)-SCCmec type IV strains producing type 4 coagulase were the most frequent. This is in striking contrast to the MRSA strains isolated in the 1990s, most of which were ST5-SCCmec type II strains producing type 2 coagulase and positive for the toxic shock syndrome toxin 1 gene. We also identified a new PVL-carrying prophage lysogenized in a type IV.3 SCCmec strain, 81/108. φ108PVL was distinct from the three extant PVL-carrying phages and was presumed to be carried by ST30-type IV.3 SCCmec strains isolated in Japan. These results provide genetic bases for the transition of MRSA clones in Japan that is commonly considered the transition from coagulase type 4 MRSA strains to coagulase type 2 MRSA strains. The results also suggested that MRSA strains that predominated between 1979 and 1985 were generated from PVL-positive methicillin-susceptible S. aureus strains through the integration of SCCmec elements.
Methicillin-resistant Staphylococcus aureus (MRSA) is one of the major human pathogens and causes a wide range of infections in health care settings and community environments (5, 11, 36). The first MRSA strain was reported in England in 1961, after which it appeared worldwide (25). In Japan, S. aureus strains exhibiting low-level resistance to methicillin were reported in the early 1960s, albeit at very low frequencies of less than 3% (20, 35). However, from the late 1970s to the early 1980s, a period when third-generation cephems with a wide spectrum of antimicrobial activity were introduced in the clinical field, MRSA strains began to dominate in Japanese hospitals (8). Since some of the third-generation cephems had weak antibacterial activities toward S. aureus, it was suspected that the excessive use of these antibiotics was responsible for the expansion of MRSA strains in hospitals throughout Japan during this time. The frequency of isolation of MRSA strains increased drastically in the early 1980s from 34% (from 1982 to 1983) to 43% (from 1986 to 1987) and from 18.6% (from July to December 1981) to 35.9% (from January to June 1982) (34, 38, 43). From the mid-1980s to the early 1990s, β-lactam antibiotics with improved antibacterial activities towards S. aureus were introduced, but by 1990, these antibiotics had become less effective against clinical isolates.
MRSA strains have been characterized by susceptibility testing and coagulase isotyping. Coagulase typing is a method developed in Japan which can classify S. aureus strains based on the antigenic difference in coagulase, the protein causing coagulation of plasma (47). In the early 1980s, coagulase type 4 MRSA strains, which showed a heterogeneous profile of oxacillin resistance, disseminated in Japanese hospitals. But from the mid 1980s to 1990s, the majority of hospital MRSA isolates were coagulase type 2 MRSA strains, most of which exhibit high resistance to oxacillin as well as to other many antibiotics. It was suggested that MRSA strains disseminated in Japanese hospitals had undergone a transition (9, 29, 31, 46). We have investigated the genotypes by conducting ribotyping and the carriage of toxin genes and reported that a shift of MRSA strains from coagulase type 4 MRSA strains to coagulase type 2 MRSA strains should be regarded as a shift of MRSA clones. The coagulase type 2 MRSA strains, which were characterized by the production of enterotoxin A and by their specific ribotyping patterns, were prevalent in the early 1980s but declined drastically in the 1990s. On the other hand, coagulase type 4 MRSA strains were characterized by the production of enterotoxin C and toxic shock syndrome toxin 1 (TSST-1) and by having their specific ribotyping patterns (16, 46).
We now know that MRSA strains have evolved from methicillin-susceptible Staphylococcus aureus (MSSA) by the acquisition of the staphylococcal cassette chromosome mec (SCCmec) carrying the mecA gene (24, 30). There are several types of SCCmec elements that differ in their genomic organizations and structures (22, 23, 33, 40, 44), and it is generally accepted that MRSA clones could be defined by SCCmec type and genotype. The classification of MRSA clones is important for epidemiological study to distinguish MRSA clones that have predominated through a hospital setting and a community setting.
In the current study, we conducted a retrospective analysis of MRSA strains isolated from Japanese hospitals between 1979 and 1985 and during the early 1990s using several different molecular typing methods: SCCmec element typing, virulence gene repertoire, and multilocus sequence typing (MLST) (13). Our results provided genetic proof for the transition of MRSA clones predominating in Japanese hospitals. Furthermore, we found that approximately one-half of MRSA strains disseminated in the early 1980s in Japan were Panton-Valentine leukocidin (PVL)-positive clones, possibly harboring a novel PVL-carrying phage.
MATERIALS AND METHODS
MRSA strains used in this study.MRSA strains isolated during three different time periods were tested: 97 MRSA strains isolated from 1979 to 1985 in Japanese hospitals (24 strains isolated at Tokyo University Hospital in 1982, 49 strains isolated at Gunma University Hospital from 1981 to 1985, 22 strains isolated at Jikei University Hospital from 1979 to 1981, and two strains isolated at other hospitals in 1981), 22 strains isolated in 1992 at Tokyo University Hospital, and 138 strains isolated in 1999 in 14 Japanese hospitals (Teikyo University, Fukushima Prefectural Medical College, Iwate Medical College, Kobe University, Akita University, Saga Medical College, Tokyo University, Hiroshima University, Fukuoka University, Mie University, Kurume Medical College, Shiga University, Jikei Medical College, and National Defense Medical College).
Susceptibility testing.MICs were determined using the agar dilution method according to the procedure recommended by the Clinical and Laboratory Standards Institute. The antibiotics tested were oxacillin, tetracycline, erythromycin, and gentamicin (Sigma Chemical Co., St. Louis, Mo.), ceftizoxime (Fujisawa Pharmacy Co., Osaka, Japan), imipenem (Banyu Pharmaceutical Co., Tokyo, Japan), and levofloxacin (Daiichi Pharmaceutical Co., Tokyo, Japan).
SCCmec typing and identification of virulence genes.SCCmec typing was performed using PCR as described previously (21, 22, 39). The presence of four virulence-related genes, lukS-PV-lukF-PV, tsst-1, seh, and cna, was investigated by PCR using the primers listed in Table 1.
Primers used in this study
Nucleotide sequencing of type IV.3 and type IV.4 SCCmec elements.Several DNA fragments spanning the entire nucleotide sequence of the SCCmec element and an SCC-like element of strain 81/108 were amplified by long-range PCR with the five primer sets (Fig. 1): the region spanning orfX (an open reading frame [ORF] that works as a port of the gene cassettes SCC; all SCC elements are integrated into the 3′ end of orfX or into the regions having the specific nucleotide sequence that is very similar to the nucleotide sequence of the 3′ end of orfX) to mecA was amplified using primers cR1 and mA1; the region spanning mecA to IS1272 was amplified using primers mA2 and IS4; the region spanning mecR1 to the ccr complex was amplified using primers mcR6 and α6; the region spanning the ccr complex to IE25923 was amplified using primers α5 and IE2 (5′-TCCACAAAATTTACATATACTCTCT-3′); the region spanning IE25923 to left-flanking chromosomal region flanked to left end of SCCmec was amplified by PCR with primers IE1(5′-AGAAATTTTGTAGCGAATGATGA-3′) and cLt2. The DNA fragments encompassing the region from the J1 region to the mec gene complex of the SCCmec element of strain JCSC4469 were amplified by long-range PCR using the following three sets of primers, mA2 and IS4, mcR6 and α6, and α5 and cL1, as indicated in Fig. 1. The PCR products were purified using the QIAquick PCR purification kit (QIAGEN, Hilden, Germany), and their nucleotide sequencing was carried out by fluorescence dideoxy chain termination chemistry using the BigDye Terminator version 1.1 cycle sequencing kit (Applied Biosystems, CA) and an ABI Prism 3100 genetic analyzer (Applied Biosystems, CA). PCR and long-range PCR were performed as described previously (24).
Structural comparison of different subtypes of type IV SCCmec elements. The structures of four subtypes of type IV SCCmec elements are illustrated based on the nucleotide sequences deposited in the DDBJ, EMBL, and GenBank databases under accession no. AB063172 (type IV.1 SCCmec), AB063173 (type IV.2 SCCmec), AB096217 (type IV.3 SCCmec), and AB097677 (IV.4 SCCmec). The SCCmec element is composed of two essential gene complexes, the ccr gene complex (pink) and the mec gene complex (light blue). The ccr gene complex consists of ccr genes which are responsible for the mobility of the SCCmec element and surrounding ORFs. The mec gene complex is responsible for methicillin-cephem resistance. Other areas (light gray) of SCCmec are nonessential and are divided into three regions, J1 to J3. Direct repeats containing integration site sequences for SCCmec elements are indicated by red arrowheads. The locations of the primer sets used for PCR amplification of an entire SCCmec element of strain 81/108 and a part of SCCmec element of strain JCSC4469 are indicated by arrows.
PCR amplification and nucleotide sequencing of PVL-carrying prophage φ108PVL.The DNA fragments encompassing the entire phage-specific region in the bacterial chromosome, along with the primers used to amplify them by long-range PCR, were as follows: the phage-flanking chromosomal region to integrase, primers phiMW-DN and intR; integrase to antirepressor, primers int-F and anti-R; anti-repressor to terminase large subunit, primers anti-F and termi-R; terminase large subunit to portal gene, primers termi-F and portal-R; portal gene to tail gene, primers, portal-F and tail-R; tail gene to the lukS-PV gene, primers tail-F and LukS-R; and lukS-PV gene to flanking chromosomal region, primers, LukS-F and phiMW-UP.
The nucleotide sequences of the primers used in these studies are listed in Table 1, and their locations in the phage region of the genome are illustrated in Fig. 2. PCR products were purified using a QIAquick PCR purification kit (QIAGEN, Hilden, Germany), and their nucleotide sequences were determined by primer walking.
(A) Essential structural and functional components of φ108PVL element are illustrated. Black arrowheads indicate the locations of primers used to amplify the entire φ108PVL genome. The two red arrowheads flanking the core sequence indicate the att sites on the phage element. (B) The locations of the primers used to amplify regions specific to φPVL, φSLT, and φSa2MW. (C) The ORFs in and around the φ108PVL element are illustrated as arrows in six possible reading frames. The direction of the arrows indicates the transcriptional direction for each ORF. Color codes are as follows: dark green, ORFs (or the parts of ORFs) that are well conserved among all other three PVL-carrying phages, φPVL, φSLT, and φSa2MW; red, ORFs that are highly homologous to φPVL, which is lysogenized in the S. aureus ATCC 49775 strain; yellow, ORFs that are highly homologous to φSLT; blue, ORFs that are highly homologous to φSa2mw; orange, ORFs that are highly homologous to both φPVL and φSLT; green, ORFs that are highly homologous to both φSLT and φSa2mw; white, ORFs that are unique in φ108PVL.
Coagulase isotyping.Coagulase type was determined by using an inhibition test for the coagulation of plasma with eight specific antisera (Denka Seiken, Niigata, Japan) according to the method of Ushioda et al. (47). Briefly, bacterial strains were grown overnight in brain heart infusion broth and culture supernatants were collected by centrifugation. An appropriately diluted 0.1-ml aliquot of the supernatant was mixed with 0.1 ml of a solution containing each antiserum and incubated at 37°C for 1 h. Diluted rabbit plasma (0.2 ml) was added to each tube, followed by incubation at 37°C for 1 h or more until coagulation of the plasma was observed by visual inspection. The serotype of cogulase produced by a given strain was determined by the specificity of the antiserum which inhibited coagulation.
MLST.Genotypes of representative strains were determined by MLST according to the procedure of Enright et al. (13). Alleles of the seven loci were assigned by a comparison of their sequences to the corresponding loci in the S. aureus MLST database (www.mlst.net). Sequence types were determined according to the combined pattern of the seven alleles, and clonal complexes were defined by the BURST (based upon related sequence types) program available on the MLST website.
Nucleotide sequence accession numbers.The entire nucleotide sequence of φ108PVL has been deposited in the DDBJ, EMBL, and GenBank databases under accession no. AB243556. The sequences of the type IV.3 (VIc) SCCmec of strain 81/108 and the type IV.4 (IVd) SCCmec of strain JCSC4469 have been deposited in the DDBJ, EMBL, and GenBank databases under accession no. AB096217 and AB097677, respectively.
RESULTS
SCCmec typing of MRSA strains isolated between 1979 and 1985 and the 1990s.We characterized the SCCmec elements carried by 97 MRSA strains isolated between 1979 and 1985 (1979-1985 strains) and 22 MRSA strains isolated in 1992 and compared them to the types of SCCmec elements of MRSA strains isolated in 1999, which have previously been reported (6). As shown in Table 2, 95 of the 1979-1985 strains (97.9%) could be classified into one of three types of SCCmec elements, judging from the combinations of mec gene complex and ccr gene complex identified by PCR. Type IV SCCmec strains (53.6%) were the most frequent overall (53.6%), followed by type I SCCmec strains (22.7%) and type II SCCmec strains (21.6%). The frequencies of each SCCmec strain, grouped according to hospital and ranked in order from highest to lowest, were as follows: Tokyo University Hospital, type IV, 17/24 (70.8%), type I, 6/24 (25.0%), and type II, 1/24 (4.2%); Gunma University Hospital, type IV, 32/49 (65.3%), type II, 8/49 (16.3%), type I, 7/49 (14.3%), and nontypeable, 2/49 (4.1%); and Jikei University Hospital, type II, 12/22 (54.5%), type I, 9/22 (4.1%), and type IV, 1/22 (4.5%). Although the frequency of each SCCmec strain at the three hospitals was different, it was clear that three types of SCCmec strains were disseminated and over three-quarters of the 1979-1985 strains (76%) carried either a type IV or a type I SCCmec element. In contrast, 126 out of 138 (93%) MRSA strains isolated in 1999 in 14 hospitals carried a type II SCCmec element. When we looked at the MRSA strains isolated in Tokyo University Hospital in particular, 91% of the strains isolated in 1992 and 100% of the strains isolated in 1999 carried a type II SCCmec element, indicating that type II SCCmec strains predominated in the early 1990s, at least at Tokyo University Hospital.
Presence of virulence determinants in MRSA strains and genotypes of the strains representing each combinationa
New subtypes of type IV MRSA strains, IV.3 (IVc) and IV.4 (IVd), predominated in the early 1980s in Japan.We further classified the SCCmec type II and type IV elements based on the nucleotide sequences of their J1 regions. The majority of SCCmec type II strains were classified as subtype type II.1 (type IIa), while six strains isolated in 1999 could not be classified into subtype 1 (type IIa) or 2 (type IIb). When we examined the type IV SCCmec elements by PCR experiments using primers that would amplify type IV.1- and type IV.2-specific J1 sequences, we identified one SCCmec type IV.1 strain among 58 SCCmec type IV strains examined. This result suggested that most type IV strains carried SCCmec type IV of unknown subtype. Therefore, we amplified and sequenced several large DNA fragments spanning the entire SCCmec element from MRSA strain 81/108. The nucleotide sequence of the J1 region of the SCCmec element carried by 81/108 was not homologous to any previously reported type IV.1 or type IV.2 SCCmec elements, so we have designated this element type IV.3 SCCmec (Fig. 1). We then designed specific primers to amplify type IV.3-specific J1 region sequences and conducted PCR experiments using chromosomal DNA from strains whose subtypes, based on the sequence of the J1 region, were unclassifiable. According to this type of analysis, 40 of 55 SCCmec type IV strains were type IV.3 SCCmec strains and 15 strains still remained unclassifiable. We then amplified and sequenced DNA fragments from MRSA strain JCSC4469 that spanned the region of the SCCmec element from mecA to J1. The J1 region of the SCCmec element of strain JCSC4469 was not homologous to that of a type IV.1, IV.2, or IV.3 SCCmec element, so we have designated it type IV.4 (Fig. 1). We then carried out PCR experiments using primers that amplified type IV.4-specific J1 sequences and found that 10 of 15 previously unclassified strains belonged to this type, leaving 5 strains still nontypeable. These results indicated that type IV.3 SCCmec strains, which are infrequent among recent isolates, predominated in Japan between 1979 and 1985 (Table 2).
Antibiotic susceptibilities.We determined the MICs of eight antibiotics (oxacillin, imipenem, ampicillin, gentamicin, tobramycin, erythromycin, tetracycline, and levofloxacin) for MRSA strains isolated between 1979 and 1985 and compared them to MICs of MRSA strains isolated in 1992 and 1999 (Table 3). Most of the strains isolated between 1979 and 1985 were susceptible to tetracycline, levofloxacin, and imipenem and showed low-level resistance to oxacillin. In contrast, the majority of MRSA strains isolated in 1999 were highly resistant to all of the antibiotics tested. The MICs of 1992 isolates for nine antibiotics reported by Tanaka et al. (46) are very similar to those of the 1999 isolates, indicating that MRSA strains became highly resistant to many antibiotics in the early 1990s.
Comparison among MICs to 10 antibiotics of MRSA strains isolated in each period
Distribution of virulence-related genes.The prevalence of virulence-related genes in all 257 MRSA strains was examined by PCR amplification of four gene loci: lukS-PV-lukF-PV, encoding Panton-Valentine leukocidin; cna, encoding collagen adhesion protein; seh, encoding staphylococcal enterotoxin H; and tsst-1, encoding toxic shock syndrome toxin-1. Three of these genes, lukS-PV-lukF-PV, cna, and seh, were identified in the highly virulent community-acquired MRSA (C-MRSA) strain MW2. The results are summarized in Table 2. Forty-four of 97 MRSA strains (45.3%) isolated between 1979 and 1985 carried lukS-PV-lukF-PV genes. In contrast, none of the MRSA strains isolated in 1992 and 1999 carried these genes. We detected cna in 44 of 97 MRSA strains isolated between 1979 and 1985, whereas only a few MRSA strains isolated in 1992 and 1999 carried this gene. The prevalence of the seh gene was very low, being identified in only four strains isolated in 1999, indicating that this gene was rarely carried by hospital-associated MRSA (H-MRSA) strains in Japan. There was a low prevalence of TSST-1-positive strains among those isolated between 1979 and 1985; however, the majority of the MRSA strains isolated in both the early and late 1990s carried tsst-1 (86.4% [19 of 22] and 92.0% [127 of 138], respectively).
Comparison of genotypes of strains chosen from different periods.To identify the genotypes of MRSA strains isolated at different periods of time, MLST and coagulase typing were conducted on strains representing each combination of SCCmec type and lukS-PV-lukF-PV- or tsst-1-positive gene profiles (Table 2). Among MRSA strains isolated between 1979 and 1985, ST30-coagulase type 4 strains were the most dominant, followed by ST5-coagulase type 2 strains and ST8-coagulase type 3 strains. All PVL-positive and TSST-1-negative strains were ST30-type 4 coagulase producers and carried a type I or type IV SCCmec element. Curiously, we found a PVL-positive/TSST-1-positive strain that carried a type II SCCmec and was a ST5-coagulase type 2 producer. Among the MRSA strains isolated in the 1990s, ST5-coagulase-type 2 strains were the most dominant, followed by ST81-coagulase type 7 strains. All tested type II.1 SCCmec strains, which were PVL negative/TSST-1 positive, were ST5-coagulase type 2 producers.
Identification of a novel PVL-carrying phage φ108PVL.It has been well established that lukS-PV and lukF-PV are encoded by a prophage that integrated into the S. aureus chromosome. To date, the nucleotide sequences of three temperate phages carrying the lukS-PV-lukF-PV genes, φPVL, φSLT, and φSa2mw, have been reported (2, 28, 37). To determine whether MRSA strains isolated in the early 1980s harbored one of the three extant PVL-carrying phages, we designed two primer sets to identify three phages specifically: a primer pair composed of a primer specific for integrases that are common to three phages and three primers specific for the repressor gene of each PVL-carrying phage, φPVL, φSLT, and φSa2mw; a primer pair composed of a primer specific for the lukS gene that is common to three phages and another two primers, one is specific for an ORF, P052, in φPVL and the other is specific for ORFs of unknown function that are conserved in two phages, φSLT and φSa2mw. PCR was carried out using chromosomal DNAs from 34 of the 44 PVL-positive MRSA strains. The results from selected strains, representing different combinations of SCCmec type and exotoxin repertoire, are presented in Table 4. There were no strains that carry either one of three phages. A strain showed a positive result by PCR for identifying the gene lineage int to rep, and a strain showed a positive result by PCR for identifying the gene lineage lukS to an ORF of unknown function, However, no strain showed a positive result with both primer sets, indicating that these strains did not carry one of three extant PVL-carrying phages.
PCR identification for four integrated PVL-carrying genesa
To determine whether these strains carried a prophage different from those of the three extant phages, DNA fragments spanning the entire phage genome were amplified by long PCR and sequenced using the chromosomal DNA of MRSA strain 81/108, a ST30-type IV.3 SCCmec strain. The characteristic 25-bp sequences attP left and attP right, which are located at both ends of the phage genome, were present. Judging from the locations of attP left and attP right, we estimated that the size of the prophage carried by MRSA strain 81/108, which we termed φ108PVL, was 44,107 bp in length. This was comparable to the sizes of the three extant PVL phages, φPVL, φSLT, and φSa2mw, which are 41,421, 42,942, and 45,924 bp in size, respectively. Two 29-bp conserved core sequences were identified adjacent to the attP left and attP right sequences. The core sequences in φ108PVL were identical to the corresponding sequences in φPVL but differed by 2 bp from the conserved core sequences in φSLT and φSa2mw.
Figure 2A illustrates the genomic organization of φ108PVL. Using BLAST to search for homologies to the three PVL-carrying phages (φPVL, DDBJ/EMBL/GenBank accession no. AB009866; φSLT, DDBJ/EMBL/GenBank accession no. NC_002661; and φSa2mw, DDBJ/EMBL/GenBank accession no. BA000033) (Fig. 2C), a total of 59 predicted ORFs larger than 99 bp were identified in and around φ108PVL. The organization of the genome of φ108PVL was very similar to that of the extant PVL-carrying phages and contained regions related to lysogeny, DNA replication/transcriptional regulation, packaging/head, tail, and lysis as well as lukS-PV-lukF-PV (Fig. 3; Table 5). Although all four phages were similar in their genomic organizations, not all of the ORFs encoded by the four phages were homologous. There were complexes of ORFs that were highly homologous, and those that appeared to be distantly related.
Alignment of the four PVL-carrying phages. Structures of φ108PVL, φPVL, φSLT, and φSa2mw are indicated based on the following nucleotide sequences: φ108PVL (DDBJ, EMBL, and GenBank databases under accession no. AB243556), φPVL (DDBJ, EMBL, and GenBank accession no. AB009866), φSLT (DDBJ, EMBL, and GenBank accession no. NC_002661), and φSa2mw (DDBJ, EMBL, and GenBank accession no. BA000033). Genes having sequence identities of more than 90% are linked by light blue shading. Known functions of ORFs are colored as follows: lysogeny, blue; DNA replication, red; recombination, pink; DNA packaging and head, yellow; tail, green; lysis, dark blue; lukS-PV-lukF-PV, black.
ORFs in and around 108PVL and their similarities to three extant PVL-carrying phages
Among the four phages, lukS-PV and lukF-PV genes were conserved with a predicted amino acid identity of greater than 99.8%. In addition, two genes encoding holin (hol) and amidase (ami), which are located upstream of lukS-PV-lukF-PV, were also well conserved, with predicted amino acid identities of more than 94.6%. We also identified 49 nucleotides located upstream of hol that were conserved among all four phages, with 98.0% nucleotide identity, while 555 nucleotides upstream of hol were highly homologous between φPVL and φ108PVL. The 204 nucleotides located between lukS-PV-lukF-PV and attP right were also highly conserved among the four phages, with nucleotide identities of more than 99.5%.
Integrases from four phages were identical with a ratio of more than 98% amino acid identity, suggesting that all four phages had integrated at the same position on the staphylococcal chromosome. In addition, 58 nucleotides between attP left and int were highly conserved with 98.3 to 100% nucleotide identities. A region located upstream of int and an additional sequence of 886 nucleotides in φ108PVL were identical to the corresponding regions in φSa2mw, and a 489-nucleotide region in φ108PVL had very high nucleotide identity to φSLT and φPVL (90.9 to 91.1%).
Other highly conserved ORFs included a putative DNA binding protein and a hypothetical protein, P028.
The ORFs that were conserved among at least three of the phages were as follows: P004 and P005, conserved among φ108PVL, φSLT, and φSa2mw with predicted amino acid identities of more than 99.5% (Fig. 2C and 3), and P027, P030, and P031, conserved among φ108PVL, φPVL and φSLT with predicted amino acid identities of more than 85.7% (Fig. 2C).
Twenty-four ORFs in φ108PVL were highly homologous to only φPVL (Fig. 2C and 3): P006 (cI-like repressor), P008 (antirepressor), P009 (transcriptional regulator), and P033 to P053 (DNA packaging and head and tail formation). The high identities between φ108PVL and φPVL among genes involved in morphogenesis suggested that φ108PVL belonged to Siphoviridae, the same group as φPVL, based on the taxonomy criteria based on the genetic organization of the structural gene cluster (4).
Six ORFs in φ108PVL related to DNA replication, P015, P020, P022, P023, P025, and P032, were homologous to only the corresponding ORFs in φSLT (Fig. 2). It is notable that only a small ORF, P003, located downstream of int, was homologous to the corresponding ORF in φSa2mw (Fig. 2). The remaining ORFs were unique to φ108PVL (Fig. 2). Among them, ORF P056, which encodes a putative transposase, was located just upstream of lukS-PV. BLAST searches of this ORF revealed that it was identical to transposase encoded by IS1272-SA, which was identified in a strain, MRSA 252, and a phage, PV83.
Dissemination of φ108PVL among Japanese MRSA strains.To determine whether PVL-positive MRSA strains isolated between 1979 and 1985 carried φ108PVL, we performed PCR experiments using chromosomal DNA from selected strains and six primer sets designed to amplify the following six genes or gene alleles in φ108PVL: set A, integrase; set B, terminase large subunit; set C, portal protein; set D, tail-length tape measure protein; set E, tail-length tape measure protein; and set F, transposase and the region at the rightmost end of φ108PVL (Fig. 2A and Table 1). All strains were positive in PCR with primer set A, indicating that they shared the same integrase and that the integration site was located at the same position on the bacterial chromosome. We found a strain (ST5-SCCmec type II.1) negative by PCR with the remaining five primer sets, indicating that it might carry a PVL-carrying phage which was distantly related to φ108PVL. Three ST30-SCCmec type I strains and an ST30-type IV.n SCCmec (a type IV SCCmec strain of which the J1 region was not classified into extant J1 regions of type IV SCCmec elements) were positive by PCR with primer sets B, C, and E and negative by PCR with primer sets D and F. A SCCmec type IV.1 strain and five SCCmec type IV.3 strains, all of which belonged to ST30, gave positive results with all other five sets of primers, indicating that they carried φ108PVL. A SCCmec type IV.3 strain was positive using four sets of primers, indicating that it carried a phage similar to φ108PVL, differing only in the presence of transposase encoded by IS1272-SA, which is found in MRSA strain 252 in its intact form.
DISCUSSION
The changing epidemiology of MRSA strains disseminated in Japanese hospitals.We conducted a retrospective study of MRSA strains disseminated in Japanese hospitals between 1979 and 1980 using molecular epidemiological methods. We showed that MRSA clones predominating in Japanese hospitals have changed drastically in 20 years. In contrast to MRSA clones predominated in 1999, most of which were ST5-SCCmec type II strains, the majority of MRSA strains isolated between 1979 and 1980 belonged to ST30 and carried either a type IV or a type I SCCmec element. In contrast to MRSA strains isolated in 1999, which were highly resistant against all tested antibiotics, MRSA strains isolated between 1979 and 1980 showed heterogeneous resistance to oxacillin and were susceptible to carbapenems, new quinolones, and tetracycline. They were different in toxin repertoire, too. The lukS-PV-lukF-PV genes, which are identified in the majority of C-MRSA strains, were identified in only MRSA strains isolated between 1979 and 1985. Since 22 MRSA strains isolated in Tokyo University Hospital in 1992 showed characteristics similar to those of MRSA strains isolated in 1999, we suppose that the shift of MRSA clones might have occurred in the early 1990s, at least in the case of Tokyo University Hospital. The type of coagulase correlated very well to the MLST genotype.
The change of MRSA clinical isolates, noticed in many facilities as the change of antibiotic susceptibility patterns and coagulase types, can be regarded as the change of MRSA clones from ST30-type I SCCmec or ST30-type IV SCCmec to ST5-type II SCCmec. In Japan, carbapenems, new quinolones, and minocyclines have been used for the treatment of MRSA infections since the mid-1980s to the late 1980s. Extensive use of antibiotics might exert selective pressures on bacteria, and only those strains that carry or acquire resistance genes or that acquire resistance through mutation are able to adapt and survive. Under the selective pressure caused by the extensive use of antibiotics, MRSA clones carrying type II SCCmec, which carries several resistance genes, might have replaced to the MRSA strains predominating in early 1980s, which could to adapt to the environmental change.
The historical shift of MRSA clones is not limited to Japanese hospitals. In 1960s, MRSA clones represented by strain COL or NCTC10442 (ST250-type 1 SCCmec) predominated in the United Kingdom. But, other clones, e.g., epidemic MRSA 16, represented by MRSA 252 (ST36-type II SCCmec), began to predominate in the United Kingdom in 1990s (15, 17, 26). In a Greece hospital, a change from ST30-type IV SCCmec strains to ST239-type III or type IIIA SCCmec IV had occurred (1). The ancient MRSA clones might have been replaced by the highly resistant MRSA clone, which has a strong capacity to spread or to survive under the selective pressure of antibiotics.
Characteristics of MRSA strains isolated between 1979 and 1985.It is interesting that approximately three-fourths of MRSA strains isolated between 1979 and 1985 carried either a type IV SCCmec element, which was identified primarily in C-MRSA strains, or a type I SCCmec element, which was identified in the first reported MRSA strains in England. Since we used strains that were stocked 20 years ago, it was very difficult to classify them into hospital-associated MRSA strains or community-associated MRSA strains due to the lack of detailed information that is required for the definition, e.g., the date of hospitalization or previous association to medical facilities. Using the records from two hospitals, we could classify the MRSA strains into two groups, strains obtained from outpatients and strains obtained from inpatients. They were as follows (the type is followed by the number obtained from inpatients and outpatients, respectively): isolates from Gunma Hospital, type I, 2 and 5; type II.1, 3 and 5; type IV.1, 0 and 1; type IV.3, 14 and 5; type IV.4, 8 and 2; type IV.n, 2 and 0; and nontypeable, 1 and 1; isolates from Tokyo University Hospital, type I, 2 and 4; type II.1, 0 and 1; type IV.3, 11 and 4; and type IV.n, 2 and 0. This analysis indicated that there was no significant difference among SCCmec types carried by outpatients and those carried by inpatients. It could be presumed that MRSA strains, which were similar to hospital-associated MRSA strains, predominated in the community as well. In the case of a large outbreak of S. aureus in Uruguay, which was caused by a highly virulent Uruguay clone represented by UR6 (ST30-type IV.3 SCCmec), the Uruguay clones were identified not only from community isolates but also from hospital isolates (32). These data suggested that both C-MRSA strains and H-MRSA strains could not be defined by their genotypes and SCCmec types, although representative C-MRSA clones and H-MRSA clones have been identified.
MRSA strains isolated between 1979 and 1985 carried lukS-PV-lukF-PV genes in a ratio of 45.3%. Since none of the tested MRSA strains isolated in 1992 and 1999 carried lukS-PV-lukF-PV genes, it is a remarkable characteristic of MRSA strains isolated between 1979 and 1985. Oka et al. reported that mortality associated with bacteremia was very high at the ratio of 47.8% (93 cases total) at Tokyo Metropolitan Geriatric Hospital from 1973 through 1984 (38). Although lukS-PV-lukF-PV genes were identified in highly virulent C-MRSA strains, e.g., MW2 that caused the death of healthy infants, we do not have the data to conclude whether the cause of high mortality is due to the presence of PVL-positive MRSA strains. Further retrospective studies of MRSA strains isolated from patients with well-documented clinical histories will help answer that question.
Evolution of PVL-carrying phages.It is well known that PVL-positive S. aureus strains harbor a bacteriophage carrying lukS-PV-lukF-PV. All extant PVL-carrying phages belonged to the class of Shi21-like Siphoviridae. Canchaya et al. classified the Staphylococcus prophages into five groups based on the similarities in genomic structures, predominantly of the genes related to morphogenesis, such as those specifying the phage head and tail (4). φSLT and φSa2mw were determined to be part of one group based on nucleotide sequence identity in the region encoding the head and tail genes. φSLT has an elongated shape, and φSa2mw is presumed to have a similar morphology. Similarly, φPV83 and φPVL were determined to belong to a second group and exhibit an icosaheadral head morphology (27). The novel phage that we identified in the current study, which we designated φ108PVL, shared basic structural components with all three extant PVL-carrying phages and showed the highest similarity to φPVL.
The four PVL-carrying phages had two highly homologous regions, the regions in and around the gene encoding integrase and the regions in and around the lysis-related genes and lukS-PV-lukF-PV. These observations strongly suggested that PVL-carrying phages evolved from non-PVL-carrying phage by acquiring the region containing lukS-PV-lukF-PV, although it is not certain whether the acquisition of the region containing lukS-PV-lukF-PV was earlier than the acquisition of the region carrying the gene for integrase. Discrepancies in the lengths of the homologous regions among these phages suggested that the acquisition of these two regions might have occurred independently by one or more illegitimate recombination events.
Characteristics of φ108PVL.The φ108PVL carried a transposase which is not present in the three extant PVL phages. The noncoding region between the genes for amidase and transposase had very low similarity to other phages, but some noncoding regions between transposase and lukS-lukF showed rather high similarity, although the length of this region differed for each, i.e., the lengths were 339 bp (φPVL, 98.2%), 340 bp (φSLT, 97.4%), and 362 bp (φSa2mw, 99.5%). These results suggested that φ108PVL acquired transposase, along with the noncoding region between the genes for amidase and transposase, as encoded by IS1272-SA. It seems that IS1272 integrated into the phage, as is the case for φPV83.
Kaneko and Kamio found no evidence for a tail structure on the φPVL phage particle using electron microscopy, although the nucleotide sequence of this phage suggests that it carries tail-related genes and that φPVL was defective in its ability to infect any S. aureus strains experimentally (27). Since most of the ST30 MRSA strains examined to date carry φ108PVL by PCR, it can be presumed that the phage was responsible for disseminating the PVL gene among MRSA strains isolated in the early 1980s in Japan. However, when we examined the genes encoding tail-length tape measure protein in φ108PVL, we found that the coding region was split into two ORFs, P049 and P050, similar to what is observed in φPVL. In contrast, φSLT, which is able to infect S. aureus strains experimentally, encodes a large tail-length tape measure protein of 2,067 amino acids. It therefore seemed that φ108PVL and φPVL are both inactive and that the spread of phages among the S. aureus strains occurred in an ancient time, and since which time, lysogenized phages have become inactive. Our data revealed that the genome of φ108PVL is a mosaic, suggesting that φ108PVL was produced as a result of lateral gene transfer and illegitimate recombination events. Thus, it seemed reasonable to assume that φ108PVL and φPVL originated from a relatively close ancestor during their revolution.
The origin of PVL-positive ST30 MRSA strains isolated in Japan.Infections caused by C-MRSA have been reported worldwide, and their genetic background and the carriage of virulence factors have been examined. Vandenesch et al. reported that all tested C-MRSA strains shared type IV SCCmec and lukS-PV-lukF-PV genes (48). Although it has been pointed out that the presence of lukS-PV-lukF-PV genes could not be used for defining C-MRSA strains, lukS-PV-lukF-PV genes have been identified in many C-MRSA strains. They reported that the genotypes of PVL-carrying C-MRSA strains were classified into six strains, ST1, -8, -30, -59, -80, and -93. Enright et al. reported eight major international epidemic MRSA strains, ST5, -8, -22, -36, -45, -289, -247, and -250 (12). Pandemic clones were also reported as New York/Japan clone (ST5-type II SCCmec), pediatric clone (ST5-type IV SCCmec), archaic clone (ST250-type I SCCmec), Iberian clone (ST247-SCCmec type 1A), Hungarian clone (ST239-type III SCCmec), and Brazilian clone (ST239-SCCmec type IIIA) by Oliveira et al. (40). The genotypes of PVL-carrying C-MRSA strains were different from those of pandemic clones other than ST8, which was identified mostly in MSSA strains (12). Subsequent study revealed that most PVL-carrying MRSA strains isolated worldwide belonged to one of six genotypes reported by Vandenesch et al. (48): ST1, from the United States; ST30, from the United States, Australia, England, Singapore, Belgium, Uruguay, and Japan; ST80, from England, The Netherlands, Denmark, Greece, Belgium, France, Switzerland, and Germany; ST8, from Belgium and the United States; and ST59, from the United States and Taiwan (3, 7, 10, 14, 18, 19, 32, 41, 45, 49, 50).
It is noteworthy that Japanese PVL-positive MRSA strains are mostly ST30, known as the southwest Pacific clone after the region where it was first described (48).
But when we investigated the subtypes of type IV SCCmec elements carried by ST30 strains, we found that different subtypes of SCCmec elements were carried by these strains. Australian strains carried type IV.1 SCCmec elements, whereas the Uruguay and Japanese PVL-positive ST30 MRSA clones carried a type IV.3 SCCmec similar to those of Japanese MRSA strains. However, their banding patterns under pulsed-field gel electrophoresis were closely related (32).
Robinson et al. reported that phage type 80/81 MSSA strains, which have been spread throughout the world, were PVL positive and belonged to ST30 and suggested that these MSSA strains might have changed to MRSA strains through the acquisition of different types of SCCmec elements independently (42). It seems likely that ST30 MSSA strains carrying lukS-PV-lukF-PV existed in the Japanese community as well and evolved into MRSA strains through the acquisition of a type of SCCmec element. This is partly supported by the observation that 38% (15 of 39) of MSSA strains isolated in the 1960s carried lukS-PV-lukF-PV (X. X. Ma et al., unpublished data). The diversity in the types of SCCmec elements identified in these isolates indicates that integration of SCCmec elements occurred independently in several different time periods.
In this paper, we introduced a method to distinguish PVL-carrying phage by PCR. By adopting the method, we found that most Japanese MRSA strains disseminated in 1979 to 1985 carried φ108PVL and that a ST5-type II SCCmec MRSA strain carried a phage which was different from three extant PVL phages as well as φ108PVL (X. X. Ma et al., unpublished). By using our methods, we will be able to classify PVL phages to know which phage was integrated in MSSA strains and whether ST30 MRSA strains isolated in Australia and Uruguay carried a phage homologous to φ108PVL. Further studies are awaited to learn the origin or spread of PVL-positive MRSA strains.
ACKNOWLEDGMENTS
We thank Toyoji Okubo and Shizuko Iyobe of Gunma University for providing us with MRSA strains.
This work was supported by a grant-in-aid for 21st Century COE Research and a grant-in-aid for Scientific Research on Priority Areas from the Ministry of Education, Science, Sports, Culture and Technology of Japan.
FOOTNOTES
- Received 11 May 2006.
- Returned for modification 3 July 2006.
- Accepted 5 October 2006.
↵▿ Published ahead of print on 18 October 2006.
- American Society for Microbiology