Unexpected Diversity of Staphylococcal Cassette Chromosome mec Type IV in Methicillin-Resistant Staphylococcus aureus Strains

Bacterial isolates. (i) Reference strains.Fifty-two well-characterized MRSA reference strains from previously described collections (3, 4, 7), five strains with published whole-genome sequences not containing SCCmec type IV, and one strain with a published whole-genome sequence containing SCCmec type IV (MW2) were used in this study. The six sequenced strains and their corresponding whole-genome sequence accession numbers were as follows: COL (SCCmec type I; GenBank accession number CP000046), MU3 (SCCmec type II; GenBank accession number AP009324), MU50 (SCCmec type II; GenBank accession number BA000017), NCTC8325 (S. aureus; GenBank accession number CP000253), ATCC 12228 (S. epidermidis; GenBank accession number AE015929), and MW2 (SCCmec type IV; GenBank accession number BA000033).

(ii) Clinical isolates.Forty-one unique clinical isolates, collected from different patients in three tertiary hospitals in Sydney, Australia, were used. In addition, another 38 paired isolates from 19 MRSA-colonized or -infected patients, collected at intervals of 1 to 30 months (median, 5 months), were used to assess in vivo stability.

Details of the reference strains and clinical isolates are provided in Fig. 2 to 4. Isolates were stored at −70°C until DNA extracts were prepared as previously described (3). These MRSA strains were all shown to possess SCCmec type IV by a previously described method (2, 21, 31).

Primer and probe design for mPCR/RLB assay and in silico experiments.Primers were designed to amplify each of the known ORFs of SCCmec type IV, based on the sequences of six type IV subtypes available in GenBank (http://www.ncbi.nlm.nih.gov ) (accession numbers AB063172 [SCCmec type IVa], AB063173 [SCCmec type IVb], AB096217 [SCCmec type IVc], AB097677 [SCCmec type IVd], DQ106887 [SCCmec type IVg], and AF411936 [SCCmec type IVh]). One primer from each pair was labeled with biotin. Immediate downstream or upstream sequences were used to design antisense or sense probes for mPCR/RLB assay as described previously (15). Some primers were used to amplify more than one target sequence due to homology between ORFs. Two probes for ORF R009 were used because of its sequence diversity; other ORFs were detected with one probe. In all, 62 primer pairs were utilized in two mPCRs, with product detection relying on 63 probes distributed across two RLB membranes (see Table S1 in the supplemental material). In silico experiments were conducted with these primers and probes to ensure oligonucleotide specificity, using 21 SCCmec region sequences published in NCBI GenBank, including all 16 SCCmec type IV sequences that were of adequate length and from MRSA strains. The GenBank accession numbers of these sequences were as follows: AB063172, EF596937, AB266531, EU437549, and EU437550 for SCCmec type IVa; AB063173 for SCCmec type IVb; AB266532, AB096217, and AY271717 for SCCmec type IVc; AB097677 for SCCmec type IVd; AJ810121 for SCCmec type IVe; DQ106887 for SCCmec type IVg; AF411936 for SCCmec type IVh; AB425824 for SCCmec type IVj; CP000730 for SCCmec type IV USA300_TCH1516; CP000253 for methicillin-susceptible S. aureus (MSSA) NCTC8325 whole-genome sequence; BA000033 for SCCmec type IV; AE015929 for methicillin-susceptible Staphylococcus epidermidis ATCC 12228 whole-genome sequence; AP009324 for SCCmec type II; BA000017 for SCCmec type II; and CP000255 for SCCmec type I. GenBank sequence searches and alignments and design of primers and probes were conducted using BioManager (Sydney Bioinformatics [http://biomanager.info/ ]) and Sigma DNA calculator browsers (Sigma-Genosys).

uPCR and mPCR.PCR amplifications were performed in a thermal cycler with HotStar Taq DNA polymerase (Qiagen, Valencia, CA). A 25-μl uniplex PCR (uPCR) mixture was prepared as follows: 2.5 μl of 10× PCR buffer with a final MgCl₂ concentration of 1.5 mM, a 0.2 mM concentration of each deoxynucleoside triphosphate (dNTP), 12.5 μM (each) forward and reverse primers, 2 μl template DNA (∼43 μg/ml; equivalent to five MRSA colonies), 0.2 μl Qiagen Hotstar Taq polymerase (5 U/μl), and molecular biology-grade H₂O (Eppendorf) were added to a total volume of 25 μl.

Two mPCR mixes were prepared: the first contained 32 primer pairs, and the second contained 31 primer pairs. Each mPCR mix consisted of 2 μl of DNA template and 3 μl of 10× PCR buffer with 1.5 mM (final concentration) MgCl₂ (Qiagen), a 0.2 mM concentration of each dNTP, 1.5 U HotStar Taq DNA polymerase, and 12.5 μM (each) forward and reverse primers. Molecular biology-grade H₂O (Eppendorf) was added to a total volume of 30 μl. PCR was performed according to the Qiagen HotStar Taq polymerase kit instructions, as follows: 95°C for 15 min; 35 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 45 s; and 72°C for 10 min. PCR products were stained with SYBR Safe DNA gel stain and visualized in a 2% agarose gel.

RLB hybridization.The RLB hybridization assay was performed as described previously (3, 15), using two membranes. The hybridization temperature was 60°C, the washed membrane was incubated in peroxidase-labeled streptavidin conjugate (Roche, Mannheim, Germany) at 42°C for 60 min, and the time of exposure to X-ray film (Hyperfilm; Amersham) was 5 min. RLB results were regarded as positive when a hybridization dot signal was clearly visible. Additional uPCR testing was conducted to resolve any weak signals produced by a test isolate and an individual probe. Weak mPCR/RLB results confirmed by the positive uPCR were treated as positive results.

Quality control of mPCR/RLB results.To ensure the reproducibility of mPCR/RLB results, positive and negative controls were run on each membrane. Positive controls were constructed by mixing extracted DNAs from selected strains to produce a sample expected to be positive for each probe on the membrane, with the exception of the CM001SP, CM002SP, and PK01SP probes, for which no strains in our collection were positive. The negative (no DNA) control was master mix only. A signal produced by the negative control was assumed to be due to contamination, and the assay was repeated in such a case.

PFGE.Pulsed-field gel electrophoresis (PFGE) analysis of SmaI-digested DNA was performed using the protocol described by McDougal et al. (20). Gels were stained with ethidium bromide and photographed under UV light with a charge-coupled device (CCD) camera. PFGE patterns were analyzed and compared using BioNumerics software (version 4.61; Applied Maths). The Dice coefficient was used for pairwise comparisons of patterns, and the unweighted-pair group method using average linkages (UPGMA) was used for pattern groupings. Position tolerance and optimization were both set at 1%.

spa typing. spa typing was performed as previously described (3), and types were assigned by consulting the Ridom SpaServer (8; http://spaserver.ridom.de ).

Statistical analysis.The mPCR/RLB results were recorded as binary data and exported into BioNumerics software (version 4.61; Applied Maths). Dendrograms were generated using the categorical coefficient and clustering by the UPGMA algorithm. Selection of discriminatory targets was performed using the AuSeTTS program (http://www.cidmpublichealth.org/pages/ausetts.html ).

SCCmec type IV subtyping by mPCR/RLB assay.The RLB patterns on both membranes are shown in Fig. 1 for 40 of the 52 MRSA reference strains. Figure 2 shows the relationships between mPCR/RLB results for the MRSA SCCmec type IV reference strains (labeled R) and those of in silico alignments of primers and probes against the sequences of 16 SCCmec type IV strains from GenBank (labeled G). Overall, 41 distinct subtypes were identified among the 52 reference strains. Cluster analysis revealed that isolates with the same SCCmec subtype, as defined by Milheiriço et al. (21), did not always group together and that none of the 52 reference strains corresponded exactly with the six type strains, suggesting that existing SCCmec type IV subtyping methods may not be able to represent complex phylogenetic relationships for this diverse MRSA type. Weak signals were noted for some probes with a few isolates, which were likely due to minor sequence variations in the probe regions, as reported previously (28).

• Open in new tab
• Download powerpoint

FIG. 1.

RLB patterns of 40 MRSA reference strains. The two membranes are aligned side by side and show 32 and 33 probes, including one control probe (mecASP) (for a total of 63 subtype-specific probes). Rows: P, positive control; N, negative control; 1, PAH1; 2, E804531; 3, F829549; 4, MW2; 5, 13792-4492; 6, IP01M1081; 7, WA01; 8, WA02; 9, FH43; 10, SJOG 30; 11, RHH58; 12, RPH85; 13, B8-10; 14, CH 16; 15, WA08; 16, RHH10; 17, WA15; 18, WA17; 19, WA19; 20, WA20; 21, CH97; 22, DEN2988; 23, WA13; 24, WA23; 25, PC8; 26, IP01M2046; 27, WA24; 28, WA26; 29, WA29; 30, WA30; 31, WA31; 32, WA33; 33, WA37; 34, WA39; 35, WA41; 36, WA42; 37, RBH98; 38, WA47; 39, WA48; 40, WA54.

• Open in new tab
• Download powerpoint

FIG. 2.

Dendrogram showing in vitro mPCR/RLB results for 52 MRSA reference strains (labeled class R), aligned with simulated in silico mPCR/RLB results for 16 SCCmec subtype IV gene cluster sequences published in GenBank (labeled class G), based on BLASTn searches against mPCR/RLB probe sequences. For the GenBank SCCmec sequences (class G), black squares represent perfect matches (>99.9% identity between sequences) with probe sequences and gray squares represent probes for which there was no matching GenBank sequence. For the 52 MRSA reference strains (class R), the black and gray squares represent positive and negative signals, respectively, on the RLB membrane. Abbreviations for geographic sources of Australian reference strains: WA, Western Australia; NSW, New South Wales; TAS, Tasmania; QLD, Queensland.

The mPCR/RLB assay classified 79 clinical isolates into 38 subtypes (Fig. 3). Isolates with similar spa types were clustered together, suggesting that they have epidemiologic relationships. In contrast, there was a poor correlation between mPCR/RLB clusters and spa types of the reference strains, which are epidemiologically unrelated.

• Open in new tab
• Download powerpoint

FIG. 3.

Dendrogram showing mPCR/RLB results for 79 MRSA clinical isolates. The black and gray squares represent positive and negative signals, respectively, on the RLB membrane. Isolates labeled with the same number and either “a” or “b” represent paired isolates from the same patient. Seventeen of 19 pairs had the same mPCR/RLB and spa typing results; for the 56a-56b pair, mPCR/RLB profiles were different but spa types were the same between its members, and for the 22a-22b pair, both mPCR/RLB and spa typing results were different (suggesting that the patients' colonizing strains had changed over time). Abbreviations for sources (three public hospitals in metropolitan Sydney, Australia): A, Westmead; B, North Shore; and C, Ryde.

In the study of in vivo stability, members of 12 of 19 isolate pairs produced identical results by PFGE, mPCR/RLB assay, and spa typing. The members of seven pairs differed from each other by at least one band in PFGE gels (data not shown), but among these, five pairs had identical mPCR/RLB and spa typing results between their members. Assuming that the members of these pairs represented essentially the same strains, carried for various periods by the same patients, these results indicate that the SCCmec type IV loci remain stable for at least several months. They imply that despite its high discriminatory power, the mPCR/RLB typing method is more stable than PFGE over time for the same or epidemiologically related strains, which is important in the investigation of prolonged outbreaks. Two isolate pairs had different mPCR/RLB patterns between members (22a-22b and 56a-56b) (Fig. 3); isolates 22a and 22b also had different spa types. The major differences in mPCR/RLB patterns between the members of these two pairs suggest that the patients' original colonizing strains were replaced by different ones. This was supported by the PFGE results for these pairs, which showed eight and three band differences between 22a and 22b and between 56a and 56b, respectively.

A total of 79 SCCmec type IV subtypes in the whole set of 131 isolates were identified using mPCR/RLB assay (with 63 probes and targets), with a Simpson's index of diversity (D) of 0.975. Using an in-house computer program, AuSeTTS (automated selection of typing target subsets [available at http://www.cidmpublichealth.org/ ]), we determined that using a subset of only 20 probes could achieve the same level of discrimination as the whole (see Table S1 in the supplemental material).

Comparison of in vitro and in silico analyses.Results of the in vitro mPCR/RLB assay for six reference strains and of in silico analysis of their corresponding whole-genome sequences against mPCR/RLB probes are shown in Fig. 4. As predicted by in silico analysis of whole-genome sequences, a number of probes hybridized with five strains that do not contain SCCmec type IV as well as with the one SCCmec type IV-containing strain (10, 22). This was most apparent for SCCmec type II strains MU3 and MU50, since SCCmec type II has considerable homology with SCCmec type IV, as shown by positive results with the IVhSP1, IVhSP2, IVhSP3, IVhSP4, IVSP5, PK02SP, and aacA-aphDSP probes. No corresponding sequences were found for probes PK02SP and aacA-aphDSP in the whole-genome sequences of these two SCCmec type II strains, suggesting that the reference strain cultures tested had acquired mobile (e.g., phage or plasmid) DNA, possibly as a result of subculture. Three of the cross-reacting targets (aacA-aphD-IV3, IVhSP1, and IVhSP3) were in the subset of 20 which contributed to the discriminatory power of the assay, but the other 4 could be omitted from the final assay configuration without a loss of discrimination. The cross-reactions are unlikely to be confusing in practice, since the system is designed for rapid subtyping of SCCmec type IV strains rather than identification of SCCmec types.

• Open in new tab
• Download powerpoint

FIG. 4.

Comparison of in silico (G) and in vitro (I) results for five isolates with published whole-genome sequences that do not contain SCCmec type IV and one that contains SCCmec type IV. The reference strains and corresponding GenBank accession numbers are as follows: rows 1 and 2, SCCmec type I strain COL (class I) and GenBank accession number CP000046 (class G); rows 3 and 4, SCCmec type IV GenBank accession number BA000033 (class G) and strain MW2 (class I); rows 5 to 8, SCCmec type II strains MU3 (class I) and MU50 (class I) and GenBank accession numbers AP009324 (class I) and BA000017 (class G); rows 9 and 10, methicillin-susceptible S. aureus GenBank accession number CP000253 (class G) and strain NCTC8325 (class I); and rows 11 and 12, S. epidermidis GenBank accession number AE015929 (class G) and strain ATCC 12228 (class I). For GenBank SCCmec sequences (class G), the black squares represent perfect matches (>99.9% identity between sequences) of probe sequences with the GenBank SCCmec sequences, and the gray squares represent probes for which there were no matching GenBank sequences. For the six reference strain isolates (class I), the black and gray squares represent positive and negative signals, respectively, on the RLB membrane.

Relative efficiency of mPCR/RLB assay.In contrast to traditional approaches to SCCmec subtyping based on gel electrophoresis, the use of the mPCR/RLB assay enables the simultaneous screening of up to 43 isolates, using two multiplex PCRs with a short turnaround time. Culture of isolates, DNA extraction, mPCR setup and running time, and RLB hybridization require no more than two working days. The preparation of the RLB membrane takes less than 2 h, and the membrane can be reused at least 20 times, reducing the cost of consumables to approximately AU$10 per isolate. The mPCR/RLB assay can easily be transferred to other laboratories, and the results are represented as binary data, which can be shared between laboratories. The reproducibility of the mPCR/RLB assay between laboratories should be excellent, provided that the same primers, probes, reagents, PCR conditions, and positive control are used (15). Contamination between samples is rare and is easily avoided by careful technique and the use of controls. The method is as discriminatory as other MRSA typing methods and is stable over time.

In combination, these features make mPCR/RLB assays very suitable for rapid epidemiological studies of large numbers of MRSA SCCmec type IV isolates for investigations of potential outbreaks. The MRSA SCCmec type IV genomic region appears to be more variable than previously thought, but our method discriminates among currently recognized major SCCmec type IV subtypes and can easily be modified to accommodate probes for newly identified SCCmec elements. Selective sequencing of the 20 most discriminatory markers could allow development of an even more accurate and reproducible sequence-based typing system, which would be applicable to testing of small numbers of isolates.

In conclusion, we have shown that SCCmec type IV has a much higher level of diversity than previously described. The mPCR/RLB-based SCCmec typing system improves our capacity to monitor the molecular evolution and spread of MRSA SCCmec type IV strains and contributes to effective strategies for infection control. The application of the mPCR/RLB hybridization assay to MRSA SCCmec typing enhances the specificity, discriminatory power, and throughput of the typing procedure. The simultaneous detection of up to 43 mPCR products in a single hybridization assay makes this assay a practicable tool for rapid infection control surveillance and MRSA outbreak detection.