Direct repeat unit (dru) typing and antimicrobial resistance of methicillin-resistant Staphylococcus pseudintermedius isolated from dogs in Atlantic Canada

Abstract

There are few reports investigating the characterization of methicillin-resistant Staphylococcus pseudintermedius (MRSP) in dogs in Canada and none from Atlantic Canada. The objectives of this study were to strain type MRSP isolates cultured at a regional diagnostic laboratory using direct repeat unit (dru) typing and to describe their antimicrobial resistance profiles. Ninety-four isolates recovered from dogs between 2010 and 2012 had dru typing, cluster analysis, and antimicrobial susceptibility testing done. The majority of isolates belonged to type dt11a (30.9%), dt10h (24.5%), dt9a (18.1%), and dt11af (10.6%) with the remaining 15.9% of isolates distributed among 13 dru types. The predominant dru types identified were similar in Ontario; however, cluster 9a appears to be less common in Atlantic Canada. A significant difference in the distribution of clusters among Atlantic provinces was detected (P = 0.01). Resistance to ≥ 2 non-β-lactam antimicrobials was observed in 71.4% of the isolates. The MRSP isolates from this study were notably less resistant than those reported in the literature. A more comprehensive study of the MRSP dru types could help further elucidate the distribution of this pathogen in Canada.

Introduction

Over the last decade, methicillin-resistant Staphylococcus pseudintermedius (MRSP) has emerged as a major opportunistic pathogen in dogs, and as the leading cause of skin disease, otitis, surgical site infections, and wound infections (1,2). Methicillin-resistance in MRSP is mediated by the mecA gene, which codes for the expression of a modified penicillin-binding protein, PBP2′, and resistance to most β-lactam antimicrobials (2). Increasing antimicrobial resistance to antimicrobials other than β-lactams, has been reported in MRSP isolates, leaving few therapeutic options available in many instances (1–3). Methicillin-resistant Staphylococcus pseudintermedius has been implicated in hospital-acquired infections within veterinary hospitals, and the potential of hospital-wide outbreaks should not be underestimated (1,4). Not only has MRSP been reported in the canine population, but also there have been several reports of humans infected or colonized with MRSP (5–11). Understanding the diversity and dissemination of MRSP is essential for developing mitigation and control strategies and for future epidemiologic studies. Currently, there is no study characterizing MRSP isolates in Atlantic Canada.

A number of molecular tools have been reported to genetically discriminate among staphylococci. Most of these tools were first developed and standardized for methicillin-resistant Staphylococcus aureus (MRSA), but few have been further refined for discriminating MRSP (1,12–16). While pulsed-field gel electrophoresis (PFGE) remains the gold standard for MRSA outbreak investigations, it has been shown to be less useful for MRSP investigations (12,14,17,18). Historically, multi-locus sequence typing (MLST) has been the primary tool for investigating the population genetic structure of MRSP, and 2 major sequence types (ST) have been identified (1,16,19). Both PFGE and MLST are laborious, costly, and require the use of reference strains, thus making them impractical tools for implementation in smaller diagnostic laboratories.

The use of a single-locus marker for isolate discrimination has also been explored for MRSA, and provides a less expensive and less laborious approach to strain typing (20,21). Sequence analysis of the staphylococcal protein A region (spa typing) has been shown to have greater discriminatory power than MLST and PFGE for MRSA, but it is less successful for strain typing all MRSP isolates (14,20–22). The mec-associated direct repeat unit (dru) typing was first demonstrated to be useful in discriminating highly-clonal MRSA isolates in Scotland (13), and has more recently been successfully applied to MRSP isolates from Canada, the United States (US), Europe, and Australia (22–24). In a previous study, significant associations between dru cluster and MLST were established, in which cluster 9a was associated with ST71 (the “International clone”) and cluster 11a was associated with ST68 (the “North American clone”) (1,22). This single-locus sequence-based method is rapid, standardized, and cost-effective, making it an ideal candidate for use in small-scale laboratories (22,23).

In a previous study, significant differences have been reported in the distribution of the 2 main dru clusters, 9a and 11a, between Canada, the US, and Europe (22). The report also inferred that variation within a country may exist, as some dru types were detected in some US states, but not the others included in the study. To date, there has been only one study investigating dru types in Canada, which was limited to Ontario; therefore, genetic information on MRSP isolates from Canadian regions other than Ontario is needed. Thus, the primary objective of this current study was to explore the strain type diversity of MRSP isolates from dogs using the dru typing method from submissions to a regional diagnostic laboratory in Atlantic Canada. A secondary objective was to describe the antimicrobial resistance (AMR) pattern of these MRSP isolates.

Materials and methods

Isolate screening and collection

Isolates were recovered from canine specimens submitted to the Atlantic Veterinary College (AVC) Diagnostic Services Bacteriology Laboratory for routine culture and susceptibility testing. Staphylococci were identified by colony morphology, including hemolysis, a positive tube coagulase test and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry. Antimicrobial susceptibilities to the following drugs were determined using the Kirby-Bauer disk diffusion method, following Clinical Laboratory Standards Institute (CLSI) standard M31-A3 (25): ampicillin (10 μg), amikacin (30 μg), amoxicillin-clavulanic acid (30 μg), cephalexin (30 μg), cefovecin (30 μg), chloramphenicol (30 μg), clindamycin (30 μg), doxycycline (30 μg), enrofloxacin (5 μg), erythromycin (15 μg), fusidic acid (10 μg), gentamicin (10 μg), penicillin (10 μg), and trimethoprim-sulfamethoxazole (SXT; 25 μg). Methicillin resistance was detected by oxacillin (1 μg) disk diffusion, and confirmed by mecA expression using the PBP2′ latex agglutination test (PBP2′ Latex Agglutination Test; Oxoid Company, Nepean, Ontario). All isolates identified presumptively as MRSP were placed in a medium containing 15% glycerol and frozen at −80°C.

Patient province of residence, submitting clinic, anatomical source/site of specimen, and antimicrobial susceptibility data were extracted from the laboratory data management system and descriptive statistics were computed. Only 1 isolate from each patient was included for analysis in most patients. If multiple MRSP isolates on repeated lab submissions were recovered from a patient but had the same dru type, only the first isolate was included for analysis. If different dru types were isolated from the same patient on different lab submissions, then both isolates were retained for analysis.

Molecular identification and typing

Genomic material was extracted using a resin matrix (InstaGene Matrix; Bio-Rad Laboratories, Montreal, Quebec). Manufacturer’s guidelines were followed except a large loopful of bacterial colonies were suspended in 1.0 mL of polymerase chain reaction (PCR) water, and the incubation time at 56°C was increased from 30 min to 1 h. Isolated DNA was frozen at −20°C until use. A multiplex PCR reaction was used to identify coagulase-positive staphylococci to the species level based on partial amplification of the nuc gene locus (26). Staphylococcus pseudintermedius isolates were characterized using the dru typing method, as previously described (17). A 40-μL reaction was prepared containing 4 μL of template DNA, 0.8 U of DNA polymerase, buffer mixture containing 1.5 mM MgCl₂ and 200 μM of each dNTP (KAPA2G Fast Hot Start DNA polymerase; KAPA Biosystems, Boston, Massachusetts, USA), additional 0.3 mM MgCl₂, and 0.8 μM of the forward primer druGF (5′-GTTAGCATATTACCTCTCCTTGC-3′) and the reverse primer druGR (5′-GCCGATTGTGCTTGATGAG-3′). Reaction mixtures were thermally cycled for an initial denaturation step of 94°C for 2 min followed by 30 cycles at 94°C for 1 min, 52°C for 1 min, and 72°C for 1 min (17). The PCR products were purified using a kit (QIAquick PCR Purification Kit; Qiagen, Mississauga, Ontario) prior to sequencing.

The dru repeats and types were assigned using a website (www.dru-typing.org) following the previously described nomenclature (17). Cluster analysis was used to compare the relatedness among dru types, and a minimum spanning tree (MST) was generated. Distance intervals (or similarity values) were created using a bin distance of 1.0%, in which dru types separated by an MST distance of ≤ 2 repeats (> 98.5% similarity) were considered closely related and assigned to the same cluster. Root nodes were assigned to the dru type with the greatest number of isolates. Cluster analysis was completed using the TRST plug-in tool (BioNumerics version 6.6; Applied Maths, Austin, Texas, USA), as described previously for MRSP dru typing (22–24,27).

Statistical analysis

Multi-drug resistant (MDR) MRSP were defined as being resistant to ≥ 2 antimicrobials classes in addition to β-lactams. Dru clusters were used for comparisons to decrease the number of tested groups. Unconditional associations between dru cluster and patient data and dru cluster and resistance to the non-β-lactam antimicrobials were assessed using Chi-squared or Fisher’s exact tests, where appropriate, with significance set at P ≤ 0.05. Subtables were explored for any significant association to determine where the differences were. All statistical computations were done using computer software (Stata/IC 13.1 for Mac; StataCorp, College Station, Texas, USA).

Results

Isolate collection

The diagnostic laboratory collected 129 isolates from 90 dogs between January 2010 and December 2012. Of those, 98 isolates were retained for analysis. Twenty-three patients had multiple submissions. Isolates were collected from various specimens (n = 98): skin (n = 52, 53.1%), ears (n = 18, 18.4%), wounds (n = 8, 8.2%), surgical sites (n = 6, 6.1%), urine (n = 5, 5.1%), abscesses (n = 2, 2.0%), and other (n = 7, 7.1%). Most of the patients were seen by private veterinary clinics in the region (71.4%) compared to being seen by the AVC Veterinary Teaching Hospital (AVC-VTH) (28.6%).

Antimicrobial resistance patterns

Only 8.2% (n = 8) of isolates were susceptible to all tested drugs with the exception of β-lactams (Table I). Another 20.4% (n = 20) of isolates were resistant to 1 non-β-lactam antimicrobial class, while 71.4% (n = 70) were resistant to ≥ 2 non-β-lactam antimicrobial classes. The most common resistance was to SXT (74.5%), followed by erythromycin (68.4%), clindamycin (55.1%), enrofloxacin (46.9%), gentamicin (34.7%), chloramphenicol (23.5%), doxycycline (15.3%), and fusidic acid (3.1%). In isolates that displayed multi-drug resistance (n = 70), 68.6% of isolates were resistant to SXT, erythromycin, and clindamycin, while 55.7% (n = 70) of those were also resistant to enrofloxacin. Amikacin resistance was not detected in any of the isolates.

Table I.

Number (and percent) of antimicrobial resistant methicillin-resistant Staphylococcus pseudintermedius (MRSP) isolates overall and by dru cluster, following Clinical Laboratory Standards Institute (CLSI) guidelines. Multidrug resistance (MDR) in each cluster is also reported

Antimicrobial	Overall (n = 94)	9a (n = 18)	10h (n = 27)	11a (n = 45)	No cluster (n = 4)
Amikacin	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0%)
Chloramphenicol	23 (24.5%)	3 (16.7%)	0 (0%)	19 (42.2%)	1 (25%)
Clindamycin	54 (57.5%)	16 (88.9%)	7 (25.9%)	29 (64.4%)	2 (50%)
Doxycycline	14 (14.9%)	1 (5.6%)	7 (25.9%)	4 (8.9%)	2 (50%)
Erythromycin	63 (67.0%)	17 (94.4%)	11 (40.7%)	32 (71.1%)	3 (75%)
Enrofloxacin	46 (48.9%)	16 (88.9%)	0 (0%)	29 (64.4%)	1 (25%)
Fusidic acid	2 (2.1%)	1 (5.6%)	1 (3.7%)	0 (0%)	0 (0%)
Gentamicin	34 (36.2%)	10 (55.6%)	2 (7.4%)	20 (44.4%)	2 (50%)
SXT	70 (74.5%)	18 (100%)	23 (85.2%)	27 (60.0%)	2 (50%)
MDR	66 (70.2%)	17 (94.4%)	15 (55.6%)	31 (68.9%)	3 (75%)

SXT — trimethoprim-sulfamethoxazole.

Direct repeat unit typing

The dru types were determined for 94/98 isolates, because 4 isolates were not available for typing. From the 94 isolates, 18 dru types were recovered with 3 predominant dru types contributing more than 70% of the distribution: dt11a (30.9%), dt10h (24.5%), and dt9a (18.1%). The frequency of each dru type can be found in Table II. Nine of the dru types were previously unidentified at the time of analysis and are represented by one isolate each: dt5k, dt6t, dt8ag, dt9bd, dt10cc, dt10cj, dt11ca, and dt11cm.

Table II.

Frequency of methicillin-resistant Staphylococcus pseudintermedius dru types from 94 isolates collected at a regional diagnostic laboratory between 2010 and 2012

dru type	Frequency (n = 94)	Cluster
dt11a	29 (30.9%)	11a
dt10h	23 (24.5%)	10h
dt9a	17 (18.1%)	9a
dt11af	10 (10.6%)	11a
dt10as	3 (3.2%)	10h
dt10cca	1 (1.1%)	10h
dt10cja	1 (1.1%)	11a
dt11bn	1 (1.1%)	11a
dt11caa	1 (1.1%)	11a
dt11v	1 (1.1%)	11a
dt11y	1 (1.1%)	11a
dt5ka	1 (1.1%)	—
dt6ta	1 (1.1%)	—
dt8aga	1 (1.1%)	—
dt9baa	1 (1.1%)	9a
dt9bda	1 (1.1%)	—
dt11cma	1 (1.1%)	11a

^aNovel dru types identified in this study.

A minimum spanning tree (MST) was constructed to identify relatedness among dru types, and to determine the predominate clusters (Figure 1). The cluster analysis identified 3 main dru clusters that included 90/94 typed isolates (Figure 1). Cluster 11a contained the most diversity, comprising 9 dru types and 47.9% of the typed isolates. Cluster 10h contained 3 dru types and 28.7% of the isolates, while cluster 9a contained 2 dru types and 19.2% of the isolates. Table II contains the dru type and cluster association.

Figure 1.

Minimum spanning tree (MST) of methicillin-resistant Staphylococcus pseudintermedius in Atlantic Canada. Numerical values on the branches indicate the similarity (MST distance) among different dru types. The dru types separated by an MST distance of ≤ 2 (≥ 98.5% similar) were considered closely related and assigned to the same cluster.

Associations among dru clusters, patient demographics, and AMR

Significant differences were determined among the distributions of dru clusters in the 4 Canadian Atlantic provinces (P = 0.01; Table III). In Newfoundland and Labrador (NL), cluster 9a was over-represented with 5/7 strains belonging to dt9a. Cluster 10h was present in Nova Scotia (NS; 15/52) and New Brunswick (NB; 5/23), but not NL, and was the predominant cluster from Prince Edward Island (PEI; 7/12) isolates. Both mainland provinces, NB and NS, shared the same predominant cluster, 11a, which comprised over 50% of the isolates from each province.

Table III.

Number (and percent) of methicillin-resistant Staphylococcus pseudintermedius (MRSP) isolates in each province overall and by dru cluster

dru cluster	NB	NL	NS	PEI
Overall (n = 94)	23 (24.5%)	7 (7.4%)	52 (55.3%)	12 (12.8%)
9a (n = 18)	6 (33.3%)	5 (27.8%)	6 (33.3%)	1 (5.6%)
10h (n = 27)	5 (18.5%)	0 (0.0%)	15 (55.6%)	7 (25.9%)
11a (n = 45)	12 (26.7%)	2 (4.4%)	28 (62.2%)	3 (6.7%)
No cluster (n = 4)	0 (0.0%)	0 (0.0%)	3 (75.0%)	1 (25.0%)

NB — New Brunswick; NL — Newfoundland and Labrador; NS — Nova Scotia; PEI — Prince Edward Island.

The dru cluster was not significantly different among the AVC-VTH and private practice (P = 0.18) or specimen types (P = 0.91). Significant associations among AMR and dru clusters were detected for all antimicrobials except fusidic acid (P = 0.31; Table I). For clindamycin, erythromycin, enrofloxacin, and SXT, cluster 9a had a significantly higher proportion of resistant isolates (P ≤ 0.01). Cluster 10h had significantly higher proportion of isolates resistant to doxycycline (P = 0.04), while cluster 11a had a significantly higher proportion of chloramphenicol resistance (P < 0.001). The proportion of MDR isolates was significantly different among dru clusters (P = 0.03), with cluster 9a having the highest proportion of MDR (94.4%), followed by 10h (55.6%) and 11a (68.8%).

Discussion

The majority of isolates (73.5%) collected as part of this convenience sampling at the AVC Diagnostic Services Bacteriology Laboratory were represented by dru types dt11a, dt10h, and dt9a, which is similar to the findings in a previous multi-national study (22). However, in the current study, clusters 11a and 10h comprised more than 75% of the isolates collected, whereas in the Ontario study, cluster 11a (including dt10h) and 9a were evenly distributed (22). The distribution of MRSP dru clusters in Atlantic Canada was more similar to what has been reported in California, Illinois, North Carolina, Tennessee, and Texas (22). Differences among geographic regions within the US were also noted in the previous study (22).

Significant differences in the diversity of strains were detected among provinces of Atlantic Canada (Table III). It should be noted that this study was not designed for prevalence estimation because of the convenience sampling approach. Although it may appear that the incidence of MRSP is higher in certain provinces, these numbers are reflective of the sample submission demographics from these regions (i.e., the laboratory receives approximately 45% of their canine samples from NS, but only 5% are from NL). These provincial differences in the distribution of dru types could be attributed to geographic separation, since differences among Newfoundland and Labrador and the Maritime Provinces (NB, NS, and PEI) exist, as well as population density, since provinces with larger populations have greater dru type diversity. The small sample size of isolates from Newfoundland and Labrador may bias its estimate. It is possible that the true diversity of the isolates from some provinces is underestimated, since each province except PEI has its own animal health microbiology laboratory, decreasing submissions from these provinces to the AVC diagnostic laboratory.

This study detected 9 dru types that had not been previously identified (Table II), and required entry into the dru database. These new dru types are results of a random single-nucleotide polymorphism in one of the 40bp dru repeat sequences, subsequently creating a new dru repeat and thus new dru type. Novel dru types have also been identified in a previous study (22). Although there were no significant differences in the distribution of the new dru types, most of the novel dru types were isolated from Nova Scotia (5/9), and from skin samples (5/9).

More than 70% of the isolates recovered in this study were resistant to ≥ 2 non-β-lactam antimicrobial classes, highlighting the clinical concerns regarding management of MRSP infections. Multi-drug resistance has been frequently reported in MRSP isolated from canine specimens (1,28). The isolates recovered in this study were from submissions to a regional diagnostic bacteriology laboratory that also is the diagnostic laboratory for the AVC-VTH, thus a selection bias is possible, based on submission of specimens from complicated referral cases from the AVC-VTH. However, most samples in this study were from primary care practices. Information about prior antimicrobial exposure was not always available, but submissions to a diagnostic laboratory for bacterial culture and susceptibility testing can be from cases proving difficult to treat, with the primary reason being non-response to antimicrobial therapy. It is likely that diagnostic laboratory submissions are biased to be more resistant.

Interestingly, cluster 9a isolates had the highest proportion of MDR at 94.4%, and also had a significantly higher proportion of resistant isolates for each antimicrobial except fusidic acid, chloramphenicol, and doxycycline. Strong associations between cluster 9a and ST71 have been shown (22), and reports have also shown ST71 to have an increased antimicrobial resistance compared with other STs (1,29). MLST was not done on the isolates in this study, but the cluster 9a isolates in this study likely belong to ST71. Although this study was not designed for prevalence estimation, the ST71 clone associated dru cluster 9a seems to have not disseminated into Atlantic Canada (18.1%), with the possible exception of Newfoundland, to the degree that it has in Europe (90%), the US (66%), and Ontario, Canada (47%) (22).

Resistance of the MRSP isolates to the non-β-lactam antimicrobials in this study was lower than in previous reports of multi-drug resistance in MRSP, which could be explained by the low prevalence of cluster 9a isolates. However, caution should be taken when comparing studies because of differences in testing methodology. Specifically, it was reported that 62% (n = 107) of study isolates were resistant to doxycycline in Ontario (24), while in our study doxycycline resistance of isolates was 15%. One study in Australia and another in the United Kingdom (UK) found similarly high tetracycline resistance at over 50% and 35%, respectively (23,30). Since the time of this study, canine specific doxycycline breakpoints for S. pseudintermedius were proposed and implemented by CLSI (31,32). This change in zone diameter breakpoints would likely increase the number of doxycycline resistant isolates in both this study and the Ontario study. Similar trends can be observed where the proportion of resistant isolates in Atlantic Canada is much lower than those reported in Australia and the UK. A recent study completed at a Texas (USA) Veterinary Medical Teaching hospital reported amikacin resistance in 36% of their MRSP isolates from dogs, whereas in our study amikacin resistance was not detected (33). A systematic literature review of antimicrobial resistance in MRSP isolates reported individual antimicrobial resistance ranging between 0% and 100%, with most studies reporting AMR estimates > 50% for most antimicrobials, except for chloramphenicol and amikacin (28). Thus, it can be inferred that the MRSP isolates in Atlantic Canada are typically less resistant, specifically to doxycycline and amikacin, than their counterparts in some other regions, even though multi-drug resistance is still common. A possible explanation for this low-level resistance could be the low population density of the region as a whole, when compared with larger, more densely populated regions, which could mean less exposure to AMR organisms and less total antimicrobial use.

The overall picture of MRSP in Atlantic Canada is similar to reports elsewhere in Canada and the world. The distribution of the dru types reported in this study is similar to other reports from North America; however, the distribution is more similar to what was observed in the US versus Ontario, Canada. The international MRSP clone, ST71, which is disseminated throughout Europe and made up almost half the isolates in Ontario, Canada, was less common in Atlantic Canada. This confirms that dru type distributions can vary significantly across the same country, and a larger, more comprehensive study of the dru types in Canada could help further clarify the dissemination of this pathogen. Multi-drug resistance in these isolates is common, especially within cluster 9a isolates, but resistance to the non-β-lactam antimicrobials is still considerably lower than has been previously reported.