|Home | About | Journals | Submit | Contact Us | Français|
One of the major concerns in global public health and the dairy industry is the emergence of host-specific virulent Staphylococcus aureus strains. The high degree of stability of the species genome renders detection of genetic microvariations difficult. Thus, approaches for the rapid tracking of specialized lineages are urgently needed. We used clumping factor A (clfA) to profile 87 bovine mastitis isolates from four regions in Canada and compared the results to those obtained by pulsed-field gel electrophoresis (PFGE) and spa typing. Twenty-five pulsotypes were obtained by PFGE with an index of discrimination of 0.91. These were assigned to six PFGE lineage groups A to F and seven spa types, including two novel ones. Group A had 48.3% of the isolates and group D had 43.7% of the isolates, while only 8% of the isolates were variable. The results of antimicrobial susceptibility testing indicated that all isolates were sensitive to methicillin and the non-beta-lactam antibiotics, while three isolates were resistant to penicillin and one isolate was resistant to tetracycline. All isolates had the clfA gene and belonged to 20 clfA repeat types with an index of discrimination of 0.90. The dominant clfA types, types X, Q, C, and Z, formed 82% and 43% of PFGE groups A and D, respectively, and had copy numbers that varied only within a narrow range of between 46 and 52 copies, implying clonal selection. The rest were variable and region specific. Furthermore, the dominant groups contained subpopulations in different regions across Canada. Sequence information confirmed the relatedness obtained by the use of clfA repeat copy numbers and other methods and further revealed the occurrence of full-repeat deletions and conserved host-specific codon-triplet position biases at 18-bp units. Thus, concordant with the results of PFGE and spa typing, clfA typing proved useful for revealing the clonal nature of the mastitis isolate lineage and for the rapid profiling of subpopulations with comparable discriminatory powers.
Infections with Staphylococcus aureus have been a public health concern and a significant economic burden globally. In the dairy industry, S. aureus bovine mastitis is one of the most important diseases and results in considerable economic losses (52). Enterotoxigenic S. aureus strains in dairy products have been reported to cause food-borne diseases (2, 67). Although a limited number of dominant clones are responsible for the majority of infections (11, 27), there has been a rise in the number of subtypes with elevated virulence and epidemicity in both hosts. A devastating example is the ongoing clonal expansion and diversification of a subset of community-associated methicillin-resistant S. aureus (CA-MRSA) isolates classified as USA300 (28, 29). Likewise, there is the emergence of a hypervirulent strain, labeled ET3-1, a subtype of the common mastitis ET3 clone (17). A potential risk would therefore be the parallel emergence of zoonotic infections (26, 63); in particular, the ET3-1 clone is highly prone to the acquisition of resistance determinants (56). The factors and mechanism(s) underlying the specificity and evolution of mastitis and CA-MRSA lineages have been quite elusive. Therefore, there is an urgent need for objective strain-subtyping tools that couple straightforward epidemiological investigations with pathogenesis and colonization potentials.
Multilocus sequence typing (MLST) grouped S. aureus in five main clonal complexes (CCs), CC8, CC30, CC5, CC22, and CC45 (8, 9, 10, 47), which were represented in three major and two minor clusters of amplified fragment length polymorphisms (AFLPs) (38). High-throughput genotyping of human and animal isolates has consistently shown a common background and virulence gene content, but mastitis-specialized S. aureus isolates belong to distinct clusters (62). Subsequent genomic analysis revealed 10 dominant lineages in each host (35, 57). Similarly, mastitis lineages were uniquely caused by sequence type (ST) 151 (ST151), ST771, and ST97; and unique genes conserved in all human isolates were variable or missing in animal lineages, including the lineage-specific genes fnbA, fnbB, and coa. It was suggested that a handful of genes account for host adaptation (57). Hence, although stable core genome markers, such as those obtained by MLST, are useful in identifying S. aureus common ancestors, the phenotypes of isolates of the same sequence type could be quite different. For instance, both strains MW2 and MSSA467 belong to ST1, but the former strain caused a fatal case of MRSA bacteremia in the United States, and the latter was isolated from a case of invasive methicillin-susceptible S. aureus (MSSA) osteomylitis in the United Kingdom (22).
The species-specific variable genomic domain (the oriC-containing region) is responsible for staphylococcal species determination (7, 35, 58) and contains genes for adherence and invasion mediated by specialized structures called microbial-surface-component-recognizing adhesive matrix molecules (MSCRAMMs) (13), such as spa and clfA. These MSCRAMMs mediate host-pathogen interactions (13, 45) and therefore could be used to identify strains with common host- and tissue-specific properties. For instance, spa has been one of the most successful single genetic markers used for the typing of S. aureus strains (30). Recently, when spa typing was examined as an alternative to pulsed-field gel electrophoresis (PFGE) typing of the Canadian MRSA epidemic clones, a high degree of concordance was found between the results obtained by the methods, demonstrating the feasibility of the use of spa typing as a more rapid and simple but powerful alternative method (14). However, the usefulness of spa typing for bovine isolates is questionable. The SpA interaction in cows is not clear; it formed insoluble complexes with serum IgGs only from the guinea pig and mouse and not from the cow, goat, sheep, horse, or chicken (3), implying a different mechanism in hoofed animals.
In addition to spa, another major MSCRAMM class consists of clumping factors ClfB (clfB) and ClfA (clfA). ClfB is expressed early in growth under increased oxygenation and is later digested by the stationary-phase proteases (36, 42). This brings to question its role in deep-seated oxygen-limited microenvironments, such as intramammary infections (IMIs), which are known to induce the emergence of persistent variants such as the small-colony types, in which significant ClfA activity has been shown by IMI challenge of mice immunized with anti-ClfA antibody (60). On the other hand, several properties of the latter make it an ideal tool for specialized clones. ClfA is constitutive, independent of the Agr system, and abundant in deep infections (25, 68); and it mediates virulence even in the absence of fibrinogen (43). Furthermore, the number of repeats in clfA affected the adherence and clumping titers of the cocci in vitro (19, 46). Moreover, ClfA and not fibronectin binding proteins was responsible for the intracellular counts of the cocci (1) and for the bacterial load and dissemination leading to abscess formation (5, 54, 41). Thus, clfA is a useful strain-specific marker that is reasonable for use for the screening of specialized lineages. We have recently evaluated the usefulness and stability of this marker using isolates and strains from human infections and cows with bovine mastitis (50). In addition, we have further provided new perspectives on the organ and host specificity of clfA from temporal and geographically independent clinical isolates from different human organs and cows with bovine mastitis (51).
The last study on the genetic structure and antimicrobial susceptibility of S. aureus recovered from cows in Canada was carried out with isolates recovered in 1999 and 2000 in Ontario and Quebec, Canada (48). Of these isolates tested for their resistance to antimicrobials, 24.5% were resistant to at least one antimicrobial, with resistance to penicillin being the most common (9.9%), followed by resistance to sulfadimethoxine (7.5%). Isolates collected in the province of Ontario exhibited the highest proportion of resistant isolates (30.2%). In Quebec, only resistance to penicillin, tetracycline, and sulfadimethoxine was found. Resistance to the penicillin-novobiocin combination, cephalothin, or ceftiofur was not found in any isolate (48). Despite the in vitro susceptibility, it is still difficult to eradicate the pathogen due to its invasiveness and its development of intracellular dormant variants.
Many recent comparative genomic analyses have consistently reached the same conclusion: S. aureus isolates from humans and animals have the same genetic background and gene content (35, 62) and that the differences in virulence in the seemingly identical emerging clones are due to a few subtle changes rather than to the large-scale acquisition of virulence factor genes (21, 28, 57). The rapidly changing epidemiology and evolution necessitate constant strain-specific regional profiling. Thus, the objective of this study was to exploit clfA for use in the determination of strain diversity in the western, central, and eastern Canadian provinces and to compare the results obtained by clfA typing with those obtained by PFGE and spa typing as well as to determine the antimicrobial susceptibilities of those isolates.
A total of 87 S. aureus isolates (Table (Table1;1; strains CO15 and CE22 were examined only by clfA typing) from 78 cows with clinical mastitis housed in 24 different farms from different regions across Canada were studied. The geographic distribution of the isolates, which were collected from January 2007 to April 2008, was as follows: six farms located in western Canada, five farms in Ontario, six farms in Quebec, and seven farms in eastern Canadian provinces. Thus, except for nine cows that had two independent isolations at two different times, all isolates were from different cows. These were obtained from the Canadian Bovine Mastitis Research Network (CBMRN) database, kindly provided by Grant Tomita. They were retested by use of a Staph latex kit (Pro-Lab Diagnostics Canada, Ontario, Canada) and were further confirmed by the identification of the nuc gene. In addition, published sequences of six strains (strains RF122, Newman, COL, MSSA 476, and the USA300 strains Staphylococcus aureus subsp. aureus USA300 PR3757 [abbreviated FPR] and Staphylococcus aureus subsp. aureus USA300 TCH1516 [abbreviated TCH]) were used as reference strains for sequence analysis. Glycerol stocks of isolates in Trypticase soy broth (TSB) were stored at −80°C.
PFGE was carried out with SmaI at the Laboratoire de Santé Publique du Québec (LSPQ), according to the Canadian Standardized Protocol for PFGE of S. aureus (39) (SOP number ARNI-PR-001, National Microbiology Laboratory [NML], Winnipeg, Manitoba, Canada).
Macrorestriction fingerprint patterns were analyzed with Bionumerics software (version 5.0; Applied Maths, Austin, TX), and dendrograms were created by using the Dice similarity coefficient and the unweighted-pair group method using average linkages. A band position tolerance of 1.0% was used, and the cluster cutoff was set at an 80% similarity level. Identical restriction patterns were assigned to the same type, whereas types that differed from the common type by one to six bands or less were assigned to the same lineage group, according to the criteria of Tenover et al. (59). Lineage groups were named alphabetically, and types within those groups were assigned numbers.
Susceptibility to seven antibiotics was examined at LSPQ in 96-well round-bottom U plates (Fisher Scientific, Ottawa, Ontario, Canada). These were oxacillin, cephalothin, tetracycline, erythromycin, penicillin, trimethoprim-sulfamethoxazole, and penicillin-novobiocin. The wells of the plates were inoculated with Mueller-Hinton broth cultures of the test isolates, as recommended by the manufacturer and in accordance with Clinical and Laboratory Standards Institute (CLSI) guide M31-A3 (5a) for veterinary isolates. After incubation, the plates were read by eye. The isolates were scored as antibiotic sensitive, intermediate, or resistant on the basis of growth or no growth at the appropriate breakpoint MIC for a given antibiotic on the basis of the CLSI guidelines; the breakpoints used are given in the Results. Reference strain S. aureus ATCC 29213 served as the assay control.
PCR amplification of clfA R-domain typing was carried out as described previously (51). Differences in R-domain copy numbers were obtained from the PCR product sizes on gels by using the information for S. aureus COL. Repeat types (RTs) were assigned on the basis of differences in the 18-bp copy numbers; isolates with a difference of a single 18-bp copy were considered to have different RTs. It is widely accepted that isolates with one PFGE pattern are genetically identical; on the basis of this, we first determined the dominant PFGE pattern groups, and then with the assumption that variant isolates have evolved from the dominant ones, we identified the variant RTs for clfA within these groups.
The copy numbers were confirmed by sequencing the clfA R-domain PCR amplicons at McGill University and the Quebec Genome Innovation Center. clfA nucleotide sequence analysis and phylogenetic relationships were carried out as described by Said et al., (50, 51). We used the whole repeated region of the sequences along with the conserved 3′ and 5′ regions to determine the alignments and the phylogenetic relationships as measures of relatedness. Global multiple-sequence alignments of the 26 pulsotypes were carried out with the Geneious Bioinformatics package (version 4.6), with the most common transversions being highlighted. The sequences were further checked visually. The sequences were also checked for agreement, disagreement, transitions, and transversions by use of the Geneious Bioinformatics package and by use of a 100% nucleotide identity against a common consensus sequence. Phylogenetic grouping was carried out with the help of the neighbor-joining method built in the Geneious Bioinformatics package (version 4.6). The phylograms were based on the most likely groupings that reflect genetic relationships in circular trees.
The determination of RTs was based on differences in the number of the 18-bp copies of tandem repeats contained in the R domain of the clfA gene. The discriminatory power calculator (http://biophp.org/stats/discriminatory_power/demo.php) was used to calculate the index of discrimination (ID), in which a value of 1 indicates the ability to differentiate each isolate and a value of 0 indicates that all isolates are identical. The numerical index of discriminatory power was used to give numerical estimates for strain differentiation, and the values (defined as the average probability that the typing system will assign a different type to two unrelated strains randomly sampled in the microbial population of a given taxon) were estimated as described by Hunter (23). The different types of organization of a repeat region, termed “repeat profiles,” was not used in this study, as we attempted to correlate relationships on the basis of repeat unit differences as well as sequence information with the ultimate goal of introducing a rapid primary screening tool with repeats without sequencing, as applied to tandem-repeat typing systems (61).
Real-time PCR assays for mecA, nuc, and lukPV were performed with the isolates of the different PFGE genotypes at NML by using the primers, protocols, and instruments described previously (37).
PCR amplification of the spa repeat region was performed as described previously (18). Amplicons were sequenced in-house by the DNA Core Facility at NML. The DNA sequences of the spa repeat region in both directions were imported as ABI or SCF files and were analyzed by using the spa typing program provided with Bionumerics software (version 5.0; Applied Maths). DNA sequences were compared through the use of the spa typing websites http://tools.egenomics.com/public/login.aspx and http://www.spaserver.ridom.de, the latter of which was developed by Ridom GmbH and is curated by SeqNet.org (http://www.SeqNet.org/).
The SmaI macrorestriction fragments of 87 isolates produced 25 patterns (Fig. (Fig.1),1), and the ID was 0.91. On the basis of the cluster cutoff, which was set at an 80% similarity level, these were assigned to six lineage groups designated A to F. Lineage group A was the most common, and 42 (48.3%) of the 87 isolates belonged to that lineage. Subgroup A1 comprised about half (20 isolates) of the isolates that were, with the exception of 6 isolates, from the Quebec region. All except 4 of the other isolates (22 isolates) were from eastern Canada and formed distinct region-specific clonal lines, namely, A3 (8 isolates), A6, A7, A8, and A9. Sublineage A4 included four isolates (two from eastern Canada and two from western Canada). Sublineage A5 isolates were restricted to Ontario. Of the remaining 45 isolates, 38 (43.7%) made up the second-largest lineage, labeled lineage group D, to which all except 4 isolates from western Canada belonged. Sublineage D1 was the largest and had two isolates from each region but was dominated by isolates from western Canada, followed by sublineage D7, which was unique only to western Canada. However, sublineage D10 was unique to the eastern region. The remaining seven isolates belonged to group B, which contained one isolate from the east and another from the west, and group C, which contained all three isolates from Ontario, in addition to unique lineages E and F from eastern Canada and Quebec, respectively. Thus, each region had certain sublineage patterns either with a high frequency, such as PFGE patterns 19 and 18 in the west, pattern 3 in the east, and pattern 1 in Quebec, or with a moderate frequency, such as pattern 20 in Ontario. Some types were also found in all regions, such as PFGE patterns 1 and 19. Genetic heterogeneity within farms was also found. As shown in Table Table1,1, the majority of the isolates produced both complete and incomplete zones of hemolysis on blood agar (BA).
All of the 87 isolates were also analyzed for the clfA repeat region, and all had the gene. clfA revealed 20 different RTs, designated A to R, X, and Z (Table (Table1),1), and the ID was 0.9. There was concordance between the groups of major clusters and independent lineage assignments obtained by clfA typing and PFGE. However, clfA further distinguished new types within the PFGE groups on many occasions, except for a few instances in which PFGE was superior at distinguishing new types, as shown in Table Table2.2. An important observation is that 82% of the isolates in PFGE lineage group A belonged to the dominant RTs, RTs X, Q, C, and Z, further subdividing group A into 10 different RTs (Table (Table1).1). Of the 42 PFGE group A isolates, 14 were RT X and all except 2 had PFGE pattern 1. Nine were of type RT Q, and all except two had pattern 3; seven were of type C, and five were type Z. Isolates of the last two RTs had mixed PFGE patterns; and the rest of the RTs, RTs F, R, G, H, I, and N, were underrepresented. As indicated in Table Table1,1, except for four isolates, isolates of RT X were from independent cows at four different farms; the exceptions were strains CQ14 and CQ15 and strains CE20 and CE21, each pair of which was from the same cow, but the strains were independently isolated from each cow at two different times and from different quarters or teats. The hemolytic pattern of the first pair of isolates was 4/2, as shown in Table Table1.1. Except for four isolates, repeat types Q, C, and Z were also from different cows located at different regions. The exceptions were strains CQ9 and CQ10, which were independently isolated from a single cow, and strains CE4 and CE5, which were isolated on the same date from different quarters of a single cow. The same major RTs, RTs X, Q, C, and Z, were also present in PFGE group D (43% of 38 isolates) but at a lower frequency than in PFGE group A. The rest of the isolates in PFGE group D mostly had unique RTs. The remaining seven PFGE groups were also represented by unique RTs. The major RTs in group D were characterized by representations of isolates from wider geographic regions; an exception was new RT A, the isolates of which were restricted to a farm in the eastern region. Each region had one or two minor RTs, each with two to three isolates that were from different cows in the same herd and that showed different PFGE patterns.
Because of the widely established notion that S. aureus is clonal both in core and in polymorphic genes, diversifications are assumed to occur more by point mutation than by recombination. To confirm the copy numbers, to establish evolutionary relationships, and to examine which type of mutations (point mutations or recombination) would be the most likely driving force for repeat variations, we analyzed the whole R-domain sequences with 3′ and 5′ conserved regions of the 25 PFGE patterns obtained. The clfA copy numbers were confirmed by sequencing. Repeat types with identical or similar copy numbers had identical or very similar nucleotide sequence alignment patterns (Fig. (Fig.2).2). These groupings were highly consistent with the groupings made by both other methods. Many isolates that had either identical or similar repeat copies and that belonged to the same or similar PFGE lineage groups showed the same nucleotide alignment patterns along the whole R domain. For instance, PFGE lineage group A isolates CQ2, CE3, CE21, CE20, CO5, and CE5 were of type X and had same sequence type, but the last two isolates had the same RT, RT C (Fig. (Fig.2).2). Similarly, the following two groups of isolates had two distinct patterns: isolates CW4, CW6, and CW7 were RT Q (PFGE group D) and isolates CW9, CW13, and CW14 were RT L for the first isolate and RT K for the last two isolates (lineage groups B and D, respectively). Finally, USA300 subsets FPR and TCH were closely related to strains COL and Newman, and MSSA isolate 476 showed a relatively evolutionary distance within the group (Fig. (Fig.22 and and3).3). All of these strains except MSSA 476 had 51 copies of repeats; MSSA 476 had 49 copies.
Figure Figure22 also shows that the overall patterns were highly conserved within related groups except in the middle of the R-domain sequence, where extensive full repeat-unit deletions were very common. This figure also shows that related groups had very closely related deletion patterns. In addition, all of the sequences analyzed (human or animal) had the same 18-bp repeat-unit structures that make up the six codon motifs with highly conserved positions, suggesting a strong codon usage preference at those sites. Furthermore, there were also a very few incomplete repeat units. Additionally, human and animal isolates uniquely selected different nucleotide codon triplets for the same amino acid at particular positions. One of these regions is shown in Fig. Fig.2B,2B, in which for animal strains the repeat unit number 1 (TCA GAC TCA GAC AGC GAC) shows the less conserved form of the Ser codon in the 1st, 3rd, and 5th positions (indicated in boldface). In subsequent units 2, 3, 4, and 5, the more conserved form of the Ser codon (AGT) was always in the 5th position, while the GAC form of aspartate was in the 2nd, 4th, and 6th positions. For human strains (TCC GAC TCC GAC AGT GAC) there was a codon bias toward the more conserved primordial forms of Ser (indicated in boldface). This is supported by an additional U-rich bias in the aspartate codon (GAT).
The combination of PFGE and spa typing has been proved to be useful, and in many typing instances these methods, particularly spa typing, also identify the most common ancestor lineages. Thus, in this study the 26 isolates representing the 25 PFGE patterns were further typed by spa typing to examine the distribution of the major lineage lines. These belonged to seven spa types, as identified by use of the Ridom and Kreiswirth systems: four types were identified in PFGE lineage group A (t267/spa type 105 [which indicates the Ridom system type/Kreiswirth system type]), t3380/spa type 106, t529/spa type 102, and t2445/new spa type [Kreiswirth repeat succession designation UK]), and three other spa types belonged to PFGE lineage group D (t529/spa type 102, t521/spa type 88, and t359/spa type 92). A novel type was found in lineage group B2 by the Ridom system (repeat succession designation r07r23r12r12r21r17r34r34r34r34r33r34) and the Kreiswirth system (repeat succession designation UJGGFMBBBBPB). In addition, Ridom types t605 and t529 were same as Kreiswirth spa type 102. Thus, two new spa types were identified, as shown in Tables Tables11 and and3.3. spa type t529/spa type 102 was predominant in group D; however, lineage groups B1, C, E, and F were indistinguishable by spa typing, as they were all type t529/spa type 102. In PFGE lineage group A, major clfA types were mostly associated with only one spa type. Similarly, in the more variable PFGE lineage group D, potentially recombinogenic clfA types Q and X had more than one spa type each. These groups also had the most variable PFGE pattern types.
All of the 87 S. aureus isolates used in this study were tested for their resistance to seven antimicrobials (Table (Table4).4). Three isolates, isolates CO12, CO13, and CO14 (all from Ontario), were resistant to penicillin; and one isolate, CE18 (from the eastern Canada region), was resistant to tetracycline. Erythromycin intermediate resistance was common in isolates from the eastern region. In addition, the multiplex reverse transcription-PCR assay for mecA and lukPV revealed that the lukPV and mecA genes were absent in all of the isolates tested (Table (Table3).3). Furthermore, screening of the isolates from animals with IMIs from different Canadian provinces for hemolysis revealed that the overwhelming majority mainly produced both types of hemolysis, clear and incomplete rings, on BA. This was irrespective of the region or subregion that was the source of the isolate, as shown in Table Table11.
Knowledge of the evolution of S. aureus strains with higher levels of epidemicity and virulence has been quite elusive. Combinations of typing approaches that couple epidemiological data and strain phenotype might serve as a paradigm strategy in both basic and applied research as a means of studying the basis for specialization and evolution. One such method would be the coding of tandem repeats, particularly in places where S. aureus infection is endemic. Recently, we have provided new perspectives on the use of the clfA locus for the organ- and host-specific grouping of clinical isolates from different host sites (51). In the present study, we have used clfA typing to show for the first time the distinct subpopulations of repeat profiles of the mastitis-specific S. aureus lineage in different regions in Canada, and the groupings were in concordance with those obtained by PFGE and spa typing. The general agreement among three typing methods suggests that clfA typing is a useful tool for the screening of strains in both basic and applied research.
In this study, concordance between the PFGE lineage groupings, the spa types, and clfA types was obtained. The simplicity, reproducibility, and high discriminatory power of clfA typing make it a valuable tool for the screening of clonal groups of the mastitis-specialized lineage before the application of major methods. The dominance of the major RTs within PFGE lineage group A and the lower frequency of the same RTs in completely different PFGE lineage group D and other groups (Table (Table1)1) are potentially interesting indications of the recombinogenic nature of the locus in a clonal background, consistent with the findings of Koreen et al. (31) for the clfB locus. Despite the fact that in clonal diversification point mutations within MLST loci give rise to new alleles at least 15-fold more frequently than recombination does (9), the latter mechanism might be occurring in the defined regions of the core variable genome responsible for determination of species and strains (7, 35, 58), such as the coding repeat region of clfA, and has been suggested as the basis for the evolution of S. aureus (32).
A general concordance in the regional distributions of genotypes and host specificity was obtained by clfA typing and other methods. The dominant isolates in each region mostly clustered together, and their sequences showed similar repeat deletion patterns. In agreement with the findings of Highlander et al. (21), the sequence alignment pattern revealed that USA300 subsets FPR and TCH were closely related to strains COL and Newman. This is further confirmed by the phylogenetic relationships of the isolates shown in Fig. Fig.3.3. A recent study also confirmed the relationships described above and further revealed the relative evolutionary distance between the group described above and CA-MSSA strain 476 within the cluster (4). Interestingly, this is also consistent with the results of clfA typing; all four strains except MSSA 476 were found to be identical on the basis of their copy number (51 copies); MSSA 476 had 49 copies. Altogether, these findings further confirm the view that CA-MRSA lineages are distinct and may descend from compatible endemic MSSA lineages and not from the hospital-acquired (HA) MRSA lineages (16, 40, 64). The similarities of the repeat profiles of the strains described above might be due to their geographical proximity, as follows: COL is from Colindale, England; Newman is from Hampshire, United Kingdom; MSSA 476 is from the United Kingdom; and USA300 was identified in Denmark in 2000 (33) and then in Europe, the United States, Canada, and worldwide, although confirmation of this assumption would require the use of a comprehensive screening strategy. In fact, typing by another repeat-based method, that with the coagulase gene, was useful in revealing the predominant S. aureus types in different geographical regions (49, 55). Thus, the geographic variation in genotypes often correlates with virulence properties. For instance, in dairy cows, herd- and environment-specific subtypes showed significantly different epidemicities and virulence in different regions (53, 24, 15), where the predominant types had higher levels of virulence and fitness (11, 55).
Typing by use of the core variable genes responsible for adaptations, such as the genes coding tandem repeats (66), would potentially help close the gap between clonality and the evolution of host-specific hypervirulent phenotypes. clfA repeats showed codon triplet position bias in individual repeat units between human and animal isolates. The majority of the point mutations detected were transversions (Fig. (Fig.2).2). Furthermore, the high degree of sequence pattern similarity in related groups, the prevalence of full repeat-unit deletions in the center of the sequence, and the occurrence of a few incomplete repeat units would imply a strong positive selection at the full repeat level and not at the point mutation level to change amino acids individually, supporting the notion that tandem repeats mediate adaptation by changing the repeat lengths and showing the importance of the repeat size over point mutations (66). This is further supported by the fact that human and animal isolates uniquely showed different nucleotide codon triplets for the same amino acid at particular positions. One of these regions is shown in Fig. Fig.2B,2B, which shows less conserved forms of amino acid codons in animal strains and a strong bias toward the more conserved primordial forms of the Ser codon (6) in human strains, in addition to the U-rich mRNA bias in aspartate codons (GAT). Thus, altogether, these findings strongly indicate that there are three types of selection pressure: one at the amino acid level not to change the dipeptides, another at the position level to preserve the amino acid locations in the unit structure, and a third one at the overall repeat unit level, as indicated by the bias for U-rich mRNAs in aspartate codons (44). This is an interesting form of adaptive variation at the whole repeat level rather than at the point mutation level that would alter amino acids to change or modulate the function while preserving the sequence.
spa typing revealed seven types for the 25 PFGE pattern types: four types in PFGE lineage group A and three in lineage group D. In group A, each major clfA type was mostly associated with one spa type. Given the temporal and regional diversity of the isolates, this suggests that in isolates from cows with mastitis, the primary differentiation of spa is followed by polymorphism of clfA. It follows that the inability of SpA to interact with cow serum (3) might be compensated for by ClfA's antiphagocytic property, which parallels that of the former (12, 20); in particular, Clfa was more prevalent than Spa in the strains tested. Finally, similar to the clfA types, certain spa types, particularly the predominant type, t529/spa type 102, were present in different lineage groups, as determined by other methods. In comparison to the Canadian database for MRSA strains, types t267, t359, and t521 were associated with ST97; the last two belonged to group D and the first one belonged to group A (corresponding to RTs X and C, respectively), indicating diversification within common types (57).
The antimicrobial susceptibility testing results obtained in the present study compared with the earlier findings for S. aureus isolates from dairy herds in the provinces of Ontario and Quebec (48) reflect a significant change in resistance patterns after a single decade. None of the 87 S. aureus isolates used in this study were MRSA, and only three isolates (isolates CO12, CO14, and CO13), all from a single farm in Ontario province, were penicillin resistant (Table (Table4).4). These could perhaps be the same strain; however, the last isolate had a PFGE pattern distant from the patterns of the other two isolates, suggesting that these isolates may have ancestral differences. In addition, the resistance patterns of these three strains appeared to be due to beta-lactamase production rather than the development of small-colony phenotypes (65), as indicated by their growth properties. There was also a single case of tetracycline resistance and many cases of erythromycin intermediate resistance in the eastern provinces (Table (Table4).4). These results, coupled with the negative results of multiplex PCR amplifications for the mecA and pvl genes, as well as possession of beta-type hemolysis, are properties resembling those of S. aureus strains with the methicillin-susceptible phenotype and strains of the mastitis-specialized lineage. This is consistent with the finding of a highly significant regional variation (2% to 65%) in the prevalence of the superantigen toxin gene in mastitis isolates from different countries, suggesting their limited role in the pathogenesis of bovine mastitis, in contrast to the prevalence of beta-hemolysin and its role (34).
In summary, we found a good correlation between the results of clfA repeat typing and those of PFGE and spa typing and found that the methods have comparable discriminatory powers. Our results were in agreement with the results of other investigators in overall strain clustering and in the clonality of the mastitis lineage of isolates, as shown by clfA and the other techniques used. In addition, we have found that the distribution and the diversity of mastitis-specific S. aureus genotypes across Canada were unique and region specific. Two major PFGE lineage groups, lineage groups A and D, mainly represented eastern and western Canada strain types, respectively. These groups, 82% of the first and 43% of the latter, were further differentiated by the same four major clfA subtypes. Thus, we have further provided a new perspective on the possible recombinogenic nature of the clfA locus in a clonal background of the species that implies selective evolution, potentially by recombination, into subclonal populations in different geographical regions across Canada. These results would show the usefulness of clfA in the typing and tracing of S. aureus strains and the coupling of the findings of straightforward epidemiological investigations with pathogenesis and colonization potential. The concordance between copy number typing and the whole R-domain sequence indicates sequence conservation within repeat units, suggesting that the former method can be used for screening purposes even without extensive sequencing.
We are grateful to Michel Couillard and the PFGE staff at LSPQ, in particular, Joanne Prévost, Guylaine Aubin, and Isabelle Robillard, for the opportunity to train on the Canadian standards for the PFGE protocol; to Brigitte Lefebvre and Luc Massicotte and their staff for help with the antibiotic susceptibility testing of isolates; and to Louis Bryden and members of Antimicrobial Resistance and Nosocomial Infections program at the National Microbiology Laboratory for training and expert advice on the use of Bionumerics software for analysis.
This study was supported by funding from a discovery grant from the Natural Science and Engineering Research Council of Canada (NSERC) and a grant from Alberta Milk; Dairy Farmers of New Brunswick, Nova Scotia, Ontario, and Prince Edward Island; Novalait Inc.; Dairy Farmers of Canada; the Canadian Dairy Network; AAFC; PHAC; Technology PEI Inc.; the Universite de Montreal; and the University of Prince Edward Island through the Canadian Mastitis Research Network. Additional support was provided by a grant from the National Nature Science Foundation of China (NSFC grant 30828026).
Published ahead of print on 2 December 2009.