Streptococcus canis is a β-hemolytic Streptococcus species from the Lancefield Group G, typically colonizing the skin and mucous membranes of asymptomatic dogs and cats (1, 2). S. canis can cause a variety of infections in these animals, including skin and soft tissue infections, and, although rare, more severe diseases such as ulcerative keratitis, necrotizing fasciitis, septicemia, endocarditis, respiratory disease, genital, and urinary infections (3–8). Less frequently, S. canis has been found in other wild and domestic mammalian hosts: in cattle is a rare but contagious agent of mastitis (3, 9). S. canis is also a zoonotic agent, with the first infection in humans described in 1998. Since then, a growing number of cases have been reported in humans, including severe cases of bacteremia, osteomyelitis, endocarditis, and pneumonia (10–12). Human infections are generally a consequence of exposure to dogs, or, less frequently, to cats, and occur after bite wounds, superficial ulcers, or cellulitis (12). Molecular epidemiological investigations have revealed significant genomic overlap between animal and human isolates, suggesting direct interspecies transmission (5, 13). Moreover, S. canis infection is rare in the setting of neonatal sepsis; however, it can lead to high morbidity and mortality as reported recently (14).

Therapy against S. canis infection is based on antimicrobials, with penicillins identified as the first-line antibiotic class in animals and humans (15). The development of AMR in S. canis has been reported, but the mechanisms of resistance are not yet fully characterized (16–18). S. canis has been reported to show low resistance rates to quinolones (7.0%) and from 5.6 to 39.7% for tetracyclines (15). Conversely, the species is considered highly susceptible to beta-lactams (15). This assumption has recently been challenged by reports of beta-lactam-resistant isolates from animals emerging in Japan (19).

Despite some research efforts, S. canis remains less studied compared to other streptococcal species, and many aspects of its epidemiology and virulence are still not well understood (3–5, 20). Two genotyping methods, a multi-locus sequence typing (MLST) scheme (21) and a scheme based on the allelic variations of the SCM protein gene (17), were used to investigate the diversity of S. canis population. Nevertheless, a consensus on the preferred method for S. canis typing was not reached, complicating the understanding of the population structure of this pathogen (13, 21–23). More recently, these two systems were compared to core genome typing based on data from WGS (5). This comparison highlighted that both MLST and SCM typing schemes lack in describing the diversity within the S. canis population, probably because they both analyze small fragments of the bacterial genome. By contrast, core genome analysis based on WGS provides a more comprehensive understanding of the epidemiology of this pathogen (5, 13). The application of WGS could potentially address many of the current knowledge gaps regarding S. canis, particularly in terms of its population structure, evolution, and host specificity.

Here we use whole genome sequencing of S. canis isolates collected between 2004 and 2021 from clinical cases in domestic animals in Italy to investigate genetic diversity, antimicrobial resistance, and pathogenicity.

S. canis is a significant pathogen in both canines and cattle and is increasingly recognized as an emerging zoonotic agent. Despite its clinical importance, there remains a scarcity of comprehensive data regarding the epidemiology, antibiotic susceptibility, and virulence mechanisms of this bacterium. Our study addresses this gap by presenting data on S. canis isolates collected from various animal hosts over a span of 17 years in Italy. Additionally, we conducted comparative genomic analyses between our S. canis isolates and publicly available genomes from human cases and different geographical regions, thereby providing new insights into the genetic diversity of this microorganism.

Regarding the analysis of virulence characteristics, we identified 18 out of the 19 virulence genes previously described in the literature (5). Several genes such as eno, fbp54, hasC, hyl, plr, rfbA, rfbB, which were identified in our collection, are recognized as components of the S. pyogenes core genome, which further supports the close evolutionary relationship between S. canis and S. pyogenes (35). hasC is part of the has operon, which is responsible for the synthesis of hyaluronic acid. In fact, in S. pyogenes, the capsule composed of hyaluronic acid has a composition analogous to that of hyaluronic acid found in human connective tissue, which contributes to the low immunogenicity of the bacterium in the host (36). In the genus Streptococcus, fbp54 encodes a surface protein capable of binding to fibrinogen and fibronectin, thus being involved in adhesion mechanisms (37). Forty-three isolates were positive for the ssp-5 gene, being one of the most prevalent virulence genes in our collection. The gene ssp-5 codes for an agglutinin receptor and is responsible for adhesion and colonization of Streptococcus to different substrates inside the host (38). Notably, ssp-5 has been identified in other Streptococcus species, such as S. suis and S. canis, both of which are associated with zoonotic transmission from animals to humans. Its presence is strongly correlated with increased pathogenicity, making it a critical factor in cross-species infections and a valuable target for surveillance and therapeutic interventions (5, 38).

Moreover, S. canis, like the majority of species within this genus, is generally susceptible to the beta-lactam class of antibiotics (4, 39, 40). A recent study has reported reduced susceptibility to penicillin-G in S. canis isolates from dogs in Japan, with this resistance attributed to amino acid substitutions in penicillin-binding proteins (15). In contrast, all isolates in our collection exhibited full susceptibility to beta-lactams, including penicillin, ampicillin, and ceftiofur. This finding reinforces the continued efficacy of beta-lactams as the first-line treatment for S. canis infections in Italy. Our isolates frequently exhibited resistance to lincosamides, tetracyclines, and macrolides consistent with previous reports (4, 5, 41–43). Resistance to tetracyclines was detected in approximately one-third of our collection, a proportion aligning with other studies, where it ranges from 30–40% (4, 21, 44, 45). This resistance was associated with the presence of tet genes, with tet(O) being the most prevalent, followed by tet(M), tet(S), and tet(T). While the presence of tet(O) and tet(M) is well documented in the literature, the detection of tet(S) and tet(T) is relatively rare (5, 46). Notably, all Italian ST15 isolates harbored tet(S), and all ST3 was positive for tet(O).

Resistance to macrolides and lincosamides, likely attributable to the MLSB phenotype (macrolides, lincosamides, and streptogramin B group), was detected in six isolates. This resistance was generally associated with the presence of the lmrP and ermB determinants (4, 42). Additionally, up to 80% of the isolates were categorized as non-susceptible to enrofloxacin according to the CLSI breakpoints for veterinary pathogens, which classify MIC values of 1–2 μg/mL as intermediate. In S. canis, resistance to quinolones is generally associated with substitutions in the parC (Ser67/Asp71), gyrA (Ser81/Glu85) and parE (Asp438) sequences (47). None of these substitutions were detected in our collection (Supplementary Table S3). It is noteworthy that the aforementioned substitutions are typically coupled with MIC values for quinolones higher than 2 μg/mL, while our isolates exhibited MIC values equal or below 2 μg/mL (47).

A total of 6 out of 55 S. canis isolates (10.9%) obtained from the bacterial collection between 2020 and 2021 exhibited a multi-drug resistant phenotype. Of these, four isolates originated from canine hosts, three of which belonged to ST2, while the remaining two were derived from cats. The emergence of MDR strains within S. canis is a growing concern, as resistance to multiple antibiotic classes can severely limit therapeutic options for treating infections in companion animals. The observed association of MDR with ST2 strains may indicate clonal expansion or selective pressure within this lineage. From a clinical standpoint, MDR S. canis infections may result in prolonged illness, treatment failure, or increased reliance on last-resort antimicrobials. Moreover, the potential zoonotic transmission of resistant S. canis strains from pets to humans—particularly immunocompromised individuals—poses a notable public health risk, as companion animals can act as reservoirs and vectors for antimicrobial-resistant bacteria (13). These findings underscore the importance of routine antimicrobial susceptibility monitoring and the implementation of prudent antibiotic use policies in veterinary practice.

There is no standard reference technique for the phylogenetic analysis of S. canis; therefore, three methods were utilized in parallel: core genome analysis, MLST and SCM sequences analysis. For the phylogenetic analysis, publicly available genomes of S. canis were also included, even though their number was quite limited. The data confirm that ST9 of S. canis is a dominant sequence type in Italy, consistent with previous studies in other European countries, such as Portugal and Germany (48). ST9 is characterized by the presence of allele 1 of the S. canis M-like protein (SCM1), a recognized virulence factor of this bacterium. Similarly, ST21 was associated with the production of SCM allele 10, as previously reported by Fukushima et al. (49), but this association is not exclusive, as the same variant was found in ST15. The phylogenetic analysis of the core genome revealed that the isolates did not cluster based on their species of origin. For instance, isolates belonging to ST9 originated from diverse sources, including dogs, cats, cattle, hedgehogs, and seals, yet were placed within the same clusters in the phylogenetic tree. Human-origin isolates did not form separate clusters but were included within the same clusters as canine, feline, and bovine isolates. The comparison among the three methods highlights an incomplete agreement between MLST typing and the core genome analysis, as shown by the presence of multiple STs in the same cluster. Genomic analyses, including multilocus sequence typing, confirm the zoonotic origin of these infections and illustrate genetic recombination events with Streptococcus dysgalactiae, enhancing its virulence and adaptability (21). ST9 was isolated from a patient with bacteremia, while previous animal studies had consistently identified ST9 in dogs suffering from dermatitis and wound infections (22). This suggests that certain STs are predisposed to cross-species infection and may possess enhanced virulence factors, such as the scm gene.

The SCM classification system groups scm alleles based on sequence similarity into three major categories: Group I (alleles 1–7), Group II (alleles 8–15), and SCM-NT (non-typeable). Group I includes classical alleles typically associated with less invasive strains, while Group II encompasses novel variants such as allele 10, which has been linked to increased intracellular invasion and potential virulence, particularly in strains belonging to ST21 and ST15 (49). SCM-NT strains either lack detectable scm sequences or express untypeable variants, and their role in pathogenesis is still under investigation. This grouping provides a molecular framework for epidemiological and virulence profiling of S. canis, particularly in zoonotic contexts (49, 50). We did not observe a clustering between SCM group 1 isolates and SCM group 2 isolates in the phylogenetic tree. As already noted by Pagnossin et al. (5), the analysis of data from WGS offers higher discrimination as compared to SCM or MLST analysis. Our study provides S. canis genomes from novel geographical regions, periods, and hosts, thereby offering new opportunities to compare this pathogen diversity across various ecological niches.

The high genetic similarity of S. canis isolates from different hosts and tissues confirms the generalist nature of this pathogen and its lack of adaptation to specific host species, in agreement with findings by other authors (5, 13). A limitation of this study is the relatively small number of S. canis isolates analyzed. However, the strains were collected over a broad temporal span (2004–2021) and from diverse host species and geographic regions, which enhances the relevance of the observed phylogenetic patterns and allows for insights into potential long-term and cross-host transmission dynamics. Regardless of genetic lineage, S. canis seems capable of cross-species transmission. This genomic evidence is supported by the nature of infections reported in cattle, where before spreading among lactating cows, S. canis infection usually originates from cats or dogs having access to the barn (9). More importantly, S. canis infections in humans are primarily attributed to a close contact, often through bites, with dogs and cats (4).

In conclusion, our study underscores the susceptibility of Italian S. canis isolates to beta-lactams antibiotics, which remain the first line of defense in managing infections. In Italy, ST9 represents the predominant clone of this pathogen. Despite the diversity in species of origin and the various sequence types identified, our findings confirm that S. canis has not adapted to different ecological niches, corroborating the accidental pathogenic nature of human cases.