Non-typhoidal Salmonella enterica (NTS) is a significant cause of human and animal illness (Centers for Disease Control, US, 2019). Zoonotic transmission of AMR Salmonella to humans occurs through 1) contact with or consumption of contaminated food products or 2) via contact with animals and animal waste. Though Salmonella infections primarily arise from contaminated food, their broad host infectivity and dissemination potential necessitates a One-Health approach to account for all transmission points, including companion animals like dogs (Chantharothaiphaichit et al., 2022; Cherry et al., 2004; Dróżdż et al., 2021; Leahy et al., 2016; Wright et al., 2005).

Salmonella infections in dogs can be clinical or asymptomatic with numerous studies reporting Salmonella isolation from clinically healthy dogs (Carter & Quinn, 2000). In many of these instances, isolates were resistant to ≥1 antibiotics with 45.2% of isolates being multidrug resistant (MDR; resistance to ≥1 agent in >3 antimicrobial classes) (Chantharothaiphaichit et al., 2022; Kiflu et al., 2017; Seepersadsingh et al., 2004; Sweeney et al., 2018). Concerningly, proximity of humans and pet dogs creates ample opportunity for Salmonella zoonosis, and pet management decisions involving food contamination, improper food handling, or both can increase infection likelihood (Behravesh et al., 2010; Finley et al., 2006, 2007; Leonard et al., 2012).

AMR biosurveillance is increasingly applying whole-genome sequencing (WGS), enabling foodborne outbreak investigation and monitoring of AMR potential in relevant pathogens. The National Antimicrobial Resistance Monitoring System (NARMS) uses WGS to understand AMR trends in enteric bacteria with a One-Health approach (Karp et al., 2017). In 2017, the FDA began monitoring AMR in pathogens, like NTS isolated from dogs, collected through the Veterinary Laboratory Investigation and Response Network (Vet-LIRN) (Ceric et al., 2019).

Given the proximity of companion animals to humans, and the human and veterinary significance of AMR emergence, we aimed to leverage available WGS data, collected via Vet-LIRN and NARMS activities, to investigate the AMR and zoonotic potential of NTS isolated from dogs and humans.

Using NCBI Sequence Read Archive BioProjects specific to Vet-LIRN Salmonella surveillance (PRJNA481328, PRJNA324575, PRJNA324570, PRJNA316457, PRJNA316452, PRJN284276, PRJNA230403), we identified all NTS strains collected from domestic dogs (Canis lupus familiaris, N = 87) within the United States (U.S.) in the NCBI Pathogen Isolate Browser (National Library of Medicine (US), National Center for Biotechnology Information, 2016). With the collection states and date range of the 87 dog-associated strains as inclusion criteria, we queried NCBI’s Pathogen Isolate Browser and identified all available NTS strains collected from humans (N = 77). These 164 isolates, collected between May 2017 and March 2023 from 17 states in the U.S., formed the basis of our analysis (Table 1; Supplementary Table 1).

Raw sequence reads were trimmed with Trimmomatic (Bolger et al., 2014). Following de novo genome assembly with Unicycler (Wick et al., 2017), assembly quality and contamination were assessed with QUAST (Gurevich et al., 2013) and CheckM (Parks et al., 2015), respectively. Genome assemblies were in silico serotyped with both SeqSero2 (Zhang et al., 2019) and SISTR (Yoshida et al., 2016) for concordance. Assembly contigs were screened for AMR- and virulence-associated genes using Abricate (Seemann, 2020) with MEGARes v2.0 (Doster et al., 2020) and Virulence Factor Databases (Chen et al., 2016), respectively.

We performed a reference-based phylogenetic analysis using PhaME (Shakya et al., 2020) and FastTree2 (Price et al., 2010) with Salmonella enterica ser. Typhimurium LT2 (GCA_000006945.2) as the reference genome. Using a matrix of pairwise single-nucleotide polymorphism (SNP) differences, hierarchical clustering was performed. All results were visualized using ape (v5.7), dendextend (1.17.1), ggtree (v3.9.0), and ggpubr (v0.6.0) packages in R version 4.3.0.

Of the 87 dog-associated isolates, 82 were serotyped as Salmonella enterica subspecies I. Among the remaining five, four were typed as Salmonella enterica subspecies IV and one was typed as Salmonella bongori. Though these strains represented 31 different serotypes, the five most common were Newport (N=16), Typhimurium (N=9), Infantis (N=8), Enteritidis (N=6), and Anatum (N=5) (Table 2). All 77 human-associated isolates were typed as Salmonella enterica subspecies I serotypes. Eleven different serotypes were identified, and the five most common were Enteritidis (N=36), Newport (N=9), Typhimurium (N=8), Oranienburg (N=6), and Infantis (N=5) (Table 2).

A total of 55 different antimicrobial resistance genes (ARGs) were identified in dog-associated strains. These ARGs are associated with specific drug class resistance (N=26), multi-drug resistance (N=4), and multi-compound resistance (N=25). (Table 3; Figure 1A). In human-associated isolates, 53 different ARGs were identified. These ARGs are associated with specific drug class resistance (N=23), multi-drug resistance (N=4), and multi-compound resistance (N=26). (Table 3; Figure 1B).

Across dog- and human-associated strains respectively, 139 and 123 unique virulence genes primarily associated with invasion were identified. Additional functional categories include adherence, colonization, extraintestinal invasion, immune modulation, iron sequestration, and toxicity (Table 4; Figure 2A–B). The virulence gene shdA, associated with prolonged fecal shedding (Kingsley et al., 2000), was identified in 34 dog-associated isolates.

Relative to the Salmonella Typhimurium LT2 reference genome, 446,396 SNPs were identified across the 164 strains. The SNP-based phylogeny is shown in Figure 3. Strain clustering was driven by Salmonella subspecies and serotype, with no apparent trends in location or time proximity. Within a clade of S. Enteritidis, two dog-associated strains were each closely related (<20 SNP differences) to six different human strains (Table 5). In a clade of S. Infantis, one dog-associated strain was closely related to five human-associated strains. Single dog-associated strains of S. Newport and S. I 4,[5],12:i:- were each closely related to a human-associated strain of the same serotype (Table 5).

Non-typhoidal Salmonella (NTS) are recognized for their broad host range and zoonotic potential as animal and human pathogens. As NTS infections in companion animals are under-investigated, we leveraged surveillance activities of the FDA’s Vet-LIRN to characterize NTS isolated from dogs. We identified a diverse population of serotypes including those clinically relevant to human health and all isolates in this study, independent of host association, possessed AMR determinants ranging from general efflux capacities to drug-specific associations. Our examination of spatiotemporally matched strains from human clinical cases identified sixteen NTS isolates related to at least one of six dog-associated strains.

Given the absence of a dog-adapted serotype (Carter & Quinn, 2000; Ceric et al., 2019; Leahy et al., 2016; Leonard et al., 2012; Reimschuessel et al., 2017), the serotype diversity revealed by our study was expected. Of the non-S. enterica spp. enterica strains identified, all have been previously isolated from a domestic dog (Andruzzi et al., 2020; Giammanco et al., 2002). The most common dog-associated serotypes in this study overlapped with those most represented among human-associated strains, and align with the Laboratory-based Enteric Disease Surveillance 2016 report regarding those most frequently reported in human Salmonella infections (National Center for Emerging and Zoonotic Infectious Disease, 2016), suggesting transmission or possible common Salmonella exposures.

As companion animals, dogs represent an understudied transmission point in Salmonella zoonoses (Hoelzer et al., 2011). We identified six dog-associated strains with 10–20 SNP differences relative to human strains, suggesting strain relatedness between Salmonella in human and dog reservoirs – which could indicate zoonotic, reverse-zoonotic, and common source as potential transmission points. Probable transmission of NTS between companion animals and humans has been reported primarily in susceptible populations, such as infants, young children and the elderly; however, probable transmission to healthy adults has also been reported in veterinary clinic and animal shelter outbreaks (Cherry et al., 2004; Morse et al., 1976; Sato et al., 2000; Wright et al., 2005). As with animal products for human consumption, improper handling of contaminated pet food products, including dry food and raw food, can also introduce Salmonella into household environments, increasing transmission risk (Arsevska et al., 2017; Behravesh et al., 2010; Davies et al., 2019; Finley et al., 2006, 2007; Hellgren et al., 2019; Schotte et al., 2007).

While the isolates studied here were from clinical cases, most dog-associated NTS infections do not yield clinical disease, apparently healthy dogs may shed Salmonella in their feces (Demirbilek, 2018; Leahy et al., 2016; Reimschuessel et al., 2017; Schotte et al., 2007). Notably, we identified shdA, a gene associated with prolonged fecal shedding, in 39% of strains (Kingsley et al., 2000). Thus, seemingly healthy dogs may be carriers of Salmonella, disseminating the bacteria for longer durations.

Though antibiotic treatment of salmonellosis is reserved for severe and invasive disease, drug resistant NTS are nonetheless considered a serious threat. In our study, we identified strains with resistance determinants to World Health Organization-defined, critically and highly important antibiotic classes. Although antibiotics are not typically prescribed for diarrheal Salmonella infections diagnosed in dogs, there are no universally accepted guidelines for antibiotic treatment of Salmonella gastroenteritis or more complicated infections in companion animal medicine, and little regulation exists regarding the use of antimicrobials in companion animals (Andruzzi et al., 2020).

Despite this study’s sampling limitations and selection bias which preclude any direct causal or epidemiological conclusions related to source attribution, we provide documentation of AMR in NTS from dogs, supporting the rationale for ongoing surveillance via Vet-LIRN and other entities, and reporting of AMR pathogens in companion animals as noted in the NARMS 2021–2025 Strategic Plan (The National Antimicrobial Resistance Monitoring System Strategic Plan 2021–2025, 2020). Our findings highlight the need to continue AMR surveillance within the One-Health paradigm.