Methicillin-resistance S. pseudintermedius isolated from skin and soft tissue infections in dogs were collected from the veterinary laboratories of Resapath, the network for surveillance of antimicrobial resistance in clinical pathogens in France¹. Two collections were constituted: the first one presented 95 isolates sampled in 2012–2013, and the second one 87 isolates collected in 2015–2016. Each clinical isolate corresponded to a unique animal. All isolates were sent to the Anses laboratory in Lyon, France, for further analysis. Species identification was performed by PCR-RFLP using TaqI restriction analysis on the kat gene (Blaiotta et al., 2010). The presence of the mecA and 16S rRNA genes were detected by PCR (Maes et al., 2002) and used to confirm methicillin resistance and as an internal quality control, respectively.

All S. pseudintermedius isolates were tested for their antimicrobial susceptibility by disk diffusion method according to the guidelines of the French Society for Microbiology (CA-SFM ²). Antibiotics tested were cefovecin 30 μg (which was used as the phenotypic marker of methicillin resistance), penicillin 6 μg, fusidic acid 10 μg, kanamycin 30 μg, gentamicin 15 μg, tobramycin 10 μg, erythromycin 15 μg, spiramycin 100 μg, lincomycin 15 μg, tetracycline 30 μg, chloramphenicol 30 μg, florfenicol 30 μg, enrofloxacin 5 μg, vancomycin 30 μg, and teicoplanin 30 μg (Mast Diagnostics, Amiens, France). Bacteria were classified as susceptible, intermediate or resistant according to the clinical breakpoints approved by the veterinary part of the CA-SFM. Staphylococcus aureus ATCC 25923 was used as the quality control strain.

Multi-locus sequence typing was performed for all MRSP isolates according to the 7-loci scheme (Solyman et al., 2013). STs were determined using the MLST website³ and new STs were assigned by the curator Vincent Perreten (vincent.perreten@vbi.unibe.ch). In addition, for each collection (2012–2013 and 2015–2016), STs were assigned to CCs by Phyloviz v 2.0 (Nascimento et al., 2017), using goeBURST (Francisco et al., 2009) algorithm with default parameters. A clonal complex is composed by all strains sharing 5 or more identical profile alleles (double-loci variants – DVL parameter).

Whole genome sequencing (WGS) was performed on a subset of 23 isolates belonging to the three most prevalent clones in both surveillance studies: all ST258 (n = 6), ST496 (n = 9), and a subset of ST71 (n = 8, among the 110 ST71 identified) were whole-genome sequenced. WGS was performed on genomic DNA extracted using the Microbial DNA extraction kit (Macherey Nagel, Hoerdt, France). The genomic library was prepared using the Nextera XT DNA Library Preparation kit (Illumina Inc., Cambridge, United Kingdom), according to the manufacturer’s protocol, and 2x 250-bp paired-end reads were generated with Illumina sequencing technology (San Diego, CA, United States).

We used the Nullarbor pipeline (T. Seemann⁴) to perform paired-end reads trimming using Trimmomatic v 0.33 (Bolger et al., 2014), and the cleaned reads were then assembled using SPAdes v 3.5 (Bankevich et al., 2012). Quality of genome assemblies was assessed with QUAST v 4.6.3 (Mikheenko et al., 2016). Always in the Nullarbor pipeline, the genomes annotation was then performed using Prokka v 1.10 (Seemann, 2014).

All publicly available genomes (NCBI genomes database, accession date 14.02.2018) S. pseudintermedius (n = 19) were in silico MLST typed using the MLST pipeline⁵.

The resistome and virulome of all sequenced ST71, ST258, and ST496 genomes were characterized using the ABRicate v 0.8.1 software (Seeman T⁶) against the ResFinder database as updated on February 20, 2018, and VFDB database (Chen et al., 2005; Zankari et al., 2012). Two curated in-house databases (DB) were subsequently built and the genomes were screened against these databases, using ABRicate v 0.8.1, to search for homologous genes. As resistance genes of S. pseudintermedius are poorly described, and because S. aureus share common resistance genes with S. pseudintermedius, we built the first database (DB_SA) using 336 markers used in the S. aureus DNA microarray Genotyping kit v2 (Monecke et al., 2008). Furthermore, we performed a state of the art of reported virulence-associated genes in S. pseudintermedius and built the second database (DB_SP) using a total of 49 genes : spsA-R, clpP, siet, speta, se-int, lukF-I, lukS-I, sec-canine, exi, expB, agrA-D, icaA-D, nanB, coa, clpX, saeR, saeS, htrA, nucC, hlb, sigB, srrA, sarA, rot, traP, expA, and epbS (Ben Zakour et al., 2011; Bannoehr et al., 2012; Garbacz et al., 2013; Pietrocola et al., 2015; Couto et al., 2016; Ruzauskas et al., 2016). All gene sequences used in DB_SP, except the sec-canine gene, were obtained from the complete genome of S. pseudintermedius strain ED99 (GenBank accession number: CP002478.1).

Because sec-canine gene was not available for S. pseudintermedius, we used the sec-canine gene of the closest phylogenetic species S. intermedius (strain 95-011195, GenBank accession number: U91526.1) and we included it in DB_SP.

Presence of both virulence and resistance-associated genes was initially detected: (i) if ≥50% of the gene length and ≥50% nucleotide identity were detected when using the DB_SA; (ii) if ≥90% of the gene length and ≥70% nucleotide identity were detected when using the DB_SP. For both databases (DB_SP and DB_SA), we used ABRicate to identify the presence of virulence and resistance-associated genes. As DB_SA was built based on S. aureus genes, we used a lower identity threshold (i.e., 50%) than the one used for DB_SP (i.e., 90%) to be able to detect homologous genes. Then, for the DB_SA, a second validation of positive hits was performed manually using a threshold of at least 80% of the query sequence length. We also screened for the presence of putative CRISPR regions, using CRISPRcasfinder (Grissa et al., 2007).

Mutations in topoisomerase II (gyrA) and IV (grlA) are known to confer fluoroquinolone resistance in S. pseudintermedius (Descloux et al., 2008). Detection of mutations Ser84Leu in gyrA and Ser80Ile in grlA sequences was performed using Snippy v 3.1⁶. The fluoroquinolone-susceptible S. pseudintermedius strain CCUG49543^T (also known as LMG 22219^T) gyrA and grlA sequences as reference (accession numbers AM262968 and AM262971, respectively) (Devriese et al., 2005). Visual confirmation of non-synonymous mutations was performed with Seaview (Gouy et al., 2010).

SNP detection was performed with Snippy v.3⁷ to construct a SNP based phylogeny for ST71, ST258, and ST496. Strain ED99 (ST25) was used as a reference outgroup/external ST to root the tree. A raw phylogenetic tree using the approximately-maximum-likelihood (ML) model and generalized time-reversible (GTR) models of nucleotide evolution were built using FastTree2 (Price et al., 2010). Strain ED99 was used to root the phylogenetic tree. To visualize and annotate the trees we used the bioconductor package ggtree (Yu et al., 2017). In addition, a pairwise SNP distance matrix was built with snp-dists v 0.6⁸, as implemented in Snippy v.3.

Proportions of antimicrobial resistance in ST71 isolates vs. non-ST71 isolates were compared using the chi-squared test. Differences were considered significant if P < 0.05.

No ethical approval was needed since this study did not involve any experimentation on animals. Only bacteria collected from clinical samples were used in the frame of this study.

The data sets supporting the results of this article are available from the Sequence Read Archive (SRA) under the BioProject no. PRJNA498009.

ST71 MRSP remains the dominant methicillin-resistant clone in France (Table 1) but its prevalence over the years revealed a decreasing trend, even not statistically significant (p = 0.1643), from 65.3% in the 2012–2013 collection to 55.2% in the 2015–2016 collection. The non-ST71 MRSP group is largely polyclonal, with most STs occurring only once. Moreover, seventeen STs were newly identified and assigned to the following numbers: 406, 407, 408, 410, 413, 414, 415, 416, 419, 420, 421, 424, 425, 1026, 1027, 1028, and 1029. The ST258 MRSP lineage was detected in one isolate in 2012–2013 and in five isolates in 2015–2016, which were all isolated in different counties. Conversely, ST261 was identified in eight isolates from eight different counties in 2012–2013 but only in one isolate in 2015–2016. Finally, we observed the emergence of the ST496 MRSP clone, which was absent in 2012–2013 and was found in nine isolates from four different counties in 2015–2016.

ST71 MRSP was highly resistant to veterinary-licensed antibiotics, with chloramphenicol, tetracyclines, and fusidic acid being the last antibiotics with resistance rates lower than 90% in 2015–2016 (Tables 2, 3). Between 2012–2013 and 2015–2016, we observed a significant increase in the resistance to tobramycin, kanamycin, and chloramphenicol. In non-ST71 MRSP, we observed a higher rate of resistance to gentamicin (97.4% in 2015–2016), macrolides-lincosamides (84.8 and 63.6%, respectively, due to the presence of the inducible MLS_B phenotype). Compared to ST71 MRSP isolates, the rate of resistance to kanamycin, tobramycin, and enrofloxacin (30.3% vs. >90%) was lower in non-ST71 MRSP but higher for tetracyclines (78.8% vs. 37.1%).

Whole genome sequencing was performed on all ST258 and ST496, as well as on a subset of eight ST71 selected from the two collections (four isolates from each collection) and originating from eight different districts. The phylogenetic tree of the French strains and the ones available on public databases shows that the ST71 and ST496 clades are highly homogenous (Figure 1). On the contrary, the ST258 cluster is more diverse. To support these observations, we used the pairwise SNP distance matrix (Supplementary Table S1) which showed that ST71 is a highly homogenous cluster with an average of 149 SNPs, while ST258 is much more heterogeneous with an average of 1919 SNPs, about thirteen time more than ST71, and twice more than ST496.

Multiple resistance genes were screened (reported in Figure 2), which perfectly correlated with the phenotypic characterization. Each ST presented a specific pattern. For ST496, all isolates displayed an identical pattern, with a higher number of resistance genes when compared to ST71 and ST258. Concerning the ST258 MRSP group, isolates presented a conserved pattern of resistance genes (blaZ, tet(M), aphA3, truncated sat genes) with sporadic variations, while ST71 was more heterogeneous. ST258 and ST496 displayed the tet(M) gene, whereas ST71 (when present) displayed the tet(K) gene. ST258 only rarely displayed the aacA-aphD gene conferring resistance to gentamicin, kanamycin, and tobramycin. All ST496 and ST71 isolates were resistant to fluoroquinolones, and they all presented the prototypical mutations Ser84Leu and Ser80Ile in the gyrA and grlA genes, respectively. On the contrary, all ST258 isolates were susceptible and all isolates presented the intact gyrA and grlA genes, except one isolate presenting the mutated grlA gene.

CRIPSR sequences were detected in all isolates belonging to the ST496, but were absent in ST71 and ST258 isolates. Typing of the CRISPR elements found in ST496 isolates revealed that they all harbored the same subtype of CRISPR: the subtype IIIA, which is located within the SCCmec element found in this lineage.

All MLST types shared a core virulence profile, with a group of 24 genes (including the biofilm-associated ica operon – except for two ST496 isolates in which icaB was absent – and sps surface proteins) being systematically present in isolates of all three STs (Figure 3). ST496 isolates were characterized by the presence of the spsI gene. ST258 isolates were devoid of the spsR gene, but were the only ones to present the nanB gene coding for a potential sialidase toxin.

In this study we showed a decreasing trend of ST71 MRSP, with a proportion falling from 65.3% in 2012–2013 to 55.2% in 2015–2016, among the French MRSP associated with canine SSTIs. This clear trend is nevertheless not statistically significant (p = 0.1643), due to the number of isolates included, but also mainly to the temporal proximity of the observed populations. A previous French study reported that the proportion of ST71 isolates within MRSP was as high as 82.9% in 2010 (Haenni et al., 2014). The decrease in the ST71 clone between the 2010 study and the 2015–2016 collection is statistically significant (p = 0.0023), even though the number of isolates in the 2010 study was low (n = 41). Of note, the sampling scheme was not identical since all pathologies were included in the first surveillance study; however, SSTIs represented 63.4% of the reported infections. Taken together, our results corroborate the downward trend of ST71 isolates within the French population of MRSP over the last decade. The decrease of the ST71 MRSP lineage, which has already been reported in several countries from Northern Europe (Duim et al., 2016; Gronthal et al., 2017), thus most likely reflects of a global evolution of the population structure of MRSP in Europe. The number of isolates included in the present study may be considered a limitation. However, all MRSP isolates collected from SSTIs over the two periods were included and the same sampling effort was made in both periods. Considering the robustness of the collection and the number of other studies reporting the same trend in Northern countries, the tendencies detected here can be considered reliable in France.

In both the 2012–2013 and 2015–2016 collections, as many as 21 distinct non-ST71 STs were identified, the vast majority being found only once. Analysis of the MRSP population structure based on the clonal complexes (CC) represented in our sampling revealed 2 different CCs and 12 singletons in the 2012–2013 collection (Supplementary Figure S1), and 1 CC and 14 singletons in the 2015–2016 collection (Supplementary Figure S2). The presence of many singletons and only a few CCs points out the coexistence of distinct and diverse MRSP population lineages. Identical non-ST71 lineages were rarely identified in both collections. The exceptions are: (i) ST265, which was identified eight times in 2012–2013 and three times in 2015–2016 and thus seems to be on a decreasing trend; (ii) ST258, which was identified once in 2012–2013 and five times in 2015–2016 and, thus, seems to be expanding. Interestingly, a concomitant decrease of ST71 MRSP with the emergence of ST258 MRSP has been previously observed in Finland, Denmark, and Netherlands (Damborg et al., 2016; Duim et al., 2016; Gronthal et al., 2017).

Our data show that, despite a higher tetracycline resistance, French ST258 MRSP isolates present a higher susceptibility to gentamicin and sulfonamides (absence of the aacA-aphD and cat genes), but also to fluoroquinolones. These results are consistent with the potential dissemination of a ST258 MRSP in Europe, which may be good news for veterinarians since this lineage is more susceptible than ST71 MRSP to veterinary-licensed antibiotics (Damborg et al., 2016; Duim et al., 2016; Gronthal et al., 2017). The SNP based phylogeny of the French ST71, ST258, ST496 and publicly available ST71, ST258, and ST496 genomes shows a very low diversity within the ST71 and ST496 populations and a more heterogeneous ST258 population. The success of this newly emerging ST258 MRSP lineage remains unknown and cannot be explained by a difference in virulence factors since only nanB, a putative sialidase toxin, was identified to be exclusively present in the ST258 isolates. However, as only a few virulence factors have been reported and validated in S. pseudintermedius, one cannot exclude the presence or absence of other, yet undescribed, virulence associated genes within this lineage.

In parallel with the emergence of a ST258 MRSP lineage, we also observed the emergence of a novel ST496 MRSP lineage in France in the collection of 2015–2016. All isolates belonging to this new clone were systematically resistant to all veterinary-licensed antibiotics. Of note, over the 87 isolates from the 2015–2016 collection, only the nine ST496 and the three ST85 isolates presented such a large multi-resistance phenotype, with susceptibility present only for florfenicol and fusidic acid. Interestingly, this clone has very recently been reported in Australia (Worthing et al., 2018), where it is described as one of the major MRSP clones, especially in Sydney. Whole-genome comparison of French and Australian isolates shows a very low level of population diversity within this clade, the same resistome and a very close virulome. Compared to isolates of the other two most prominent MRSP lineages in France – ST71 and ST258 – ST496 isolates were the only ones that systematically carried cell wall associated (CWA) genes spsI and, except for one isolate, spsF genes, which are hypothesized to play a role in pathogenesis as they code for proteins capable of adhesion to extracellular matrix (Bannoehr et al., 2011).