ABSTRACT
We compared three novel methicillin-resistant Staphylococcus aureus (MRSA) genotyping methods with multilocus sequence typing (MLST) and spa typing to assess their utility for routine strain typing. The new methods were femA and nuc sequence typing and toxin gene profiling (TGP), using a multiplex-PCR-based reverse line blot assay to detect 13 pyrogenic superantigen and exfoliative toxin genes. Forty-two well-characterized MRSA strains, representing 15 MLSTs or 9 clonal clusters (CCs), were genotyped by all methods. Twenty-two spa, nine femA, and seven nuc sequence types were identified. The femA sequence types correlated exactly with CCs; nuc sequences types were less discriminatory but generally correlated well with femA types and CCs. Ten isolates contained none of 13 toxin genes; TGPs of the remainder comprised 1 to 5 toxin genes. The combination of spa typing and TGPs identified 26 genotypes among the 42 strains studied. A combination of two or three rapid, inexpensive genotyping methods could potentially provide rapid MRSA strain typing as well as useful information about clonal origin and virulence.
Methicillin-resistant Staphylococcus aureus (MRSA) genotyping is used to study its evolution and epidemiology and to assist in infection control (39). Different typing methods provide different information. Multilocus sequence typing (MLST) reveals slowly accumulating changes in conserved genes that reflect long-term evolutionary changes and can identify global spread of the relatively small number of successful clones (13). It has limited discriminatory power and is unsuitable for outbreak investigation, whereas pulsed-field gel electrophoresis is highly discriminatory and can identify recent changes. It is most widely used for outbreak investigation and infection control (9, 28). Both methods are relatively expensive and slow, and a number of rapid, inexpensive typing methods, based on sequence or length polymorphisms of variable genes or loci, have been described that are objective and relatively inexpensive. These include coa (41) and spa (38) sequence typing and the multilocus variable number tandem repeat assay (12, 29).
spa typing, which depends on differences in the number and sequence of tandem repeats in region X of the protein A gene (44), is discriminatory, rapid, inexpensive, and objective (25, 37, 41). The development of a shareable web-based database (www.spaServer.Ridom.de) (15) and the utility of spa typing for early-warning systems (31) have contributed to the rapid uptake of MRSA spa typing by diagnostic and public health laboratories.
In this study, we investigated the potential utility of two additional S. aureus gene polymorphisms for strain typing, namely, femA, one of several genes involved in the synthesis of the branched-peptide structure of S. aureus peptidoglycan (4), and nuc, which encodes an extracellular thermostable nuclease of S. aureus (5). Both are species-specific S. aureus genes; they have been widely used as PCR targets for identification (21), but their polymorphisms have not been widely investigated (14).
S. aureus produces numerous toxins, including enterotoxins or pyrogenic superantigens and exfoliative toxins, some of which are encoded by genes carried on staphylococcal pathogenicity islands and associated with certain clonal complexes (CCs), whereas genes encoding others, such as the Panton-Valentine leucocidin (PVL), are carried on bacteriophages and readily transferred between different lineages (26, 27). This suggests that a toxin gene profile (TGP) could help identify S. aureus CCs as well as providing information about virulence. Various molecular methods have been described for studying the distribution of staphylococcal toxins (2, 10).
We used 42 well-characterized MRSA strains to compare sequence polymorphisms of femA and nuc and TGPs, based on a multiplex PCR-based reverse line blot assay (mPCR/RLB) (22), with two established typing methods—namely, spa typing and MLST—to determine their potential utility for MRSA genotyping.
MATERIALS AND METHODS
S. aureus isolates.We used 42 well-characterized reference and clinical S. aureus isolates in this study, as shown in Table 1, including 35 from various parts of Australia, provided by Philip Giffard, Cooperative Research Centre for Diagnostics, Queensland University of Technology, Brisbane, and Graeme Nimmo, Queensland Health Pathology Services, Princess Alexandra Hospital, Brisbane, Australia. Some have been used in several previous studies (40). Seven strains were provided by Herminia de Lencastre, Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa, Oeiras, Portugal, and also have been used in previous studies (33, 34) Two of these strains (MW2 and COL) have been fully sequenced (26). MLST and SCCmec typing results were provided by the donors of the strains (33, 34, 40).
Genotypes and spa types of 42 well-characterized methicillin-resistant S. aureus isolates used in this study
DNA extraction.DNA extraction was performed as described previously (23).
Toxin gene detection.A well-established mPCR/RLB protocol developed in our laboratory (22) was used to detect 13 Staphylococcus aureus toxin genes. Target genes, primer and probe sequences, physical characteristics, and locations within selected GenBank sequences are shown in Table 2. All primers and probes had similar physical characteristics to allow simultaneous amplification and hybridization, respectively, in a multiplex reaction (22). Two gene-specific PCR primer pairs and two gene-specific probes were designed for each of 13 toxin genes. All primers were 5′ end biotinylated to allow detection of hybridization with a streptavidin peroxidase substrate. The probes were labeled with a 5′-end amine group to facilitate covalent linkage to the nylon membrane and to allow membranes to be stripped and reused repeatedly (22). Each multiplex reaction included nuc primers as the positive control for S. aureus and for quality control of DNA extraction and mPCR/RLB. All primers and probes were synthesized by Sigma-Aldrich (Sydney, Australia).
Primers and probes used in mPCR/RLB for detection of 13 toxin genes
The mPCR/RLB was performed as previously described (22) with the following modifications: each 25-μl reaction mixture contained 0.5 U Hotstar Taq polymerase (QIAGEN, Melbourne, Australia), and the mPCR annealing temperature was optimized to 55°C.
Sequencing, sequence analysis, and phylogenetic tree.femA, nuc, and spa PCR primers were based on the published GenBank sequences using BioManager (http://biomanager.angis.org.au/). Sequencing was performed as described previously (24). For most targets, outer primers were used for amplification and inner primers for sequencing (Table 3).
Primers used for PCR sequencing of nuc, femA, and spa genes
The spa types were defined by reference to the shareable web-based database (www.spaServer.Ridom.de) (15). All spa repeat regions were submitted to the database, and spa types were assigned by the database by combining the sequences of all repeat regions.
Data obtained from different typing methods were recorded and stored in an Access file, which was imported into the BioNumerics software program (Applied Maths) with appropriate formatting. A phylogenetic tree was generated by using the categorical coefficient and clustered by the Ward algorithm.
Calculation of index of diversity.Simpson's index of diversity was calculated for each individual genotyping method and for combinations of methods, as described by Hunter and Gaston (16).
Nucleotide sequence accession numbers.The nearly full-length sequences (see below) of selected femA and nuc genes and partial spa sequences were deposited in GenBank with the following accession numbers: for femA, DQ103589 and DQ352456 to DQ352463; for nuc, DQ507377 to DQ507382; and for spa, EF094507 to EF094528. Eight S. aureus genome sequences were used for reference: AJ938182 (RF122), NC_002952 (MRSA252), AF144661 (Staphylococcus aureus subsp. anaerobius), NC_002745(N315), NC_002758 (Mu50), NC_003923 (MW2), NC_002951 (COL), and NC_002953 (MSSA476).
RESULTS AND DISCUSSION
Sequence polymorphisms of femA and nuc.Nine femA sequence types with lengths of approximately 1,215 bp were identified, of which five were newly identified and four had been previously deposited in GenBank. A total of 39 polymorphism sites were found among those sequences. Seven nuc sequence types of approximately 700 bp were identified, with 28 polymorphisms sites. Four sequence types had not been previously identified.
spa types and TGPs.Twenty-two spa types were identified, and sequences of each new type were submitted to GenBank with accession numbers shown in Table 1. Nineteen of the spa types were already recorded in the spa database (www.spaServer.Ridom.de) (15), and three were new types, first identified in Australian strains (Table 1; Fig. 1). There were 14 different TGPs (Fig. 1 and 2).
Relatedness of 42 MRSA strains between different typing methods. *, strains CH 69, IPO1M2046, and 14176-5710 belonged to spa types t1963, t1958, and t1959, respectively, which have not been previously deposited in the database.
The 13 toxin gene profiles of the 41 strains. Lanes 1 to 42 show results for the following isolates, in order (see Table 1): FH43, SJOG 30, RPH85, B827549, SN39, RHH58, RHH10, FH53, B8-10, RPH2, IMVS67, CH16, CH69, PAH58, PAH1, E822485, J710566, RPH74, E804531, CH97, IP01M2046, C801535, F829549, RBH98, 13792-4492, IP01M1081, 14176-5710, K704540, K711532, AH13, RDH81, AH1, RPAH18, RPAH15, COL, MW2, DEN2988, BK2464, HDG2, HU25, a control strain, and ANS46.
Comparison of MLST, femA and nuc sequence types, spa types, and TGPs.MLST are based on sequences of seven housekeeping genes (http://www.mlst.net/). Isolates with identical sequences for all seven genes are considered to be clonal and those with five or six matching genes to belong to the same CC (27). There were 15 MLST and nine CCs among the 42 strains studied (Table 1). The latter correlated exactly with femA sequence types, suggesting that femA sequencing may be a useful “shorthand” single-locus surrogate for MLST (Fig. 1). In future, informative single nucleotide polymorphisms in the femA sequence may be able to predict femA type and CC. One of the candidate methods is rolling-circle amplification, which has been used successfully in our laboratory (42).
There were seven nuc sequence types, which were therefore less informative. One sequence type was represented among three femA sequence types (and corresponding CCs).
The relationships between MLST and TGP varied (Fig. 1). No toxin genes were found in 10 isolates belonging to four sequence types (STs) (ST-8, -78, -88, and -239). One to five toxin genes were found in various combinations in the remaining isolates. Some STs included more than one TGP, e.g., ST-5, -22, -45, and -239 each included isolates with two different TGPs, which reflects the ability of mobile genetic elements on which toxin genes are carried to transfer laterally between clones.
However, some toxin genes are transferred vertically within specific CCs (32, 36). For example, all 10 ST-1 isolates (but none belonging to other STs) contained sea and seh; sea alone was found in another eight isolates belonging to CC 8/239. seg and sei, which are part of the enterotoxin gene complex (egc), were always present together, in 10 isolates spread among four CCs (ST-5/73, -22, -45, and -30/36). This is consistent with a previous report that egc is preferentially distributed among CCs 5 and 30/36 (19). In addition, we identified mutations in regions of seg and sei probes in isolates belonging to CC 30/36.
The Panton-Valentine leukocidin gene (pvl) was identified, with a variety of other toxin genes (depending on ST/CC), in eight isolates belonging to ST-1 (three isolates), ST-93 (two isolates), or CC 30/36 (three isolates); seven of eight PVL-containing isolates belonged to SCCmec type IV, which is generally associated with community-acquired MRSA. PVL is associated with necrotic skin and soft tissue lesions and, more recently, with life-threatening necrotizing pneumonia and sepsis due to community-acquired MRSA (8, 44).
There were 22 spa types among the 42 strains tested. When combined with TGP, some spa types were further subdivided, making a total of 26 genotypes. For example, isolates belonging to spa type t002 contained two TGPs (seg-sei and sed-seg-sei), and those belonging to t008 contained three (sea, seb, and none). The combination of these two methods thus provides a high level of discrimination, using relatively inexpensive, rapid methods.
Comparison of discriminatory powers of each genotyping method and various combinations (Table 4) showed that spa typing is the most discriminatory. The addition of TGPs alone or TGP plus SCCmec typing increases the discriminatory power, but there is little additional increase from additional femA or nuc sequence typing.
Comparison of discriminatory powers of each genotyping method and various combinations of methods for 42 MRSA strains using Simpson index of diversity
Significance of sequence polymorphisms.These results indicate that femA and to some extent nuc sequence types correlate closely with MLST in this set of MRSA isolates, suggesting that these genes evolve at a rate similar to that of housekeeping genes within CCs (20, 27). spa types were more discriminatory for strain typing but correlated less well with MLST or CCs.
Significant sequence variation in femA and nuc also has potential implications for their use as species-specific PCR targets for identification of S. aureus. The possibility of mutations needs to be considered in the design of probes and primers to avoid false-negative or inaccurate quantitative PCR results. Our results show that some mutations occurred in the region of primers used as species-specific primers (1).
There were significant sequence polymorphisms in femA and nuc genes, which have not been previously well studied, which has potential implications for their use as species-specific PCR targets for identification of S. aureus. Both correlated well with each other, and femA sequence types correlated with MLST/CCs. TGPs provide useful information about potential virulence and the evolutionary history of S. aureus strains and can increase the discriminatory power of femA and spa sequence typing. Prospective testing of unselected clinical isolates will be needed to adequately determine the optimal combination of methods for MRSA surveillance.
ACKNOWLEDGMENTS
We sincerely thank the following colleagues for allowing us to study their isolates: Herminia de Lencastre, Philip Giffard, and Graeme Nimmo.
Fanrong Kong, Qinning Wang, and Yongwei Cai made similar contributions to this work and so would be seen as co-first authors.
FOOTNOTES
- Received 28 May 2007.
- Returned for modification 2 August 2007.
- Accepted 15 August 2007.
↵▿ Published ahead of print on 22 August 2007.
- American Society for Microbiology
