Phylodynamics Uncovers the Transmission of Antibiotic-Resistant Escherichia coli between Canines and Humans in an Urban Environment

Abstract

The role of canines in transmitting antibiotic resistant bacteria to humans in the urban environment is poorly understood. To elucidate this role, we utilized genomic sequencing and phylogenetics to characterize the burden and transmission dynamics of antibiotic resistant Escherichia coli (ABR-Ec) cultured from canine and human feces present on urban sidewalks in San Francisco, California. We collected a total of fifty-nine ABR-Ec from human (n=12) and canine (n=47) fecal samples from the Tenderloin and South of Market (SoMa) neighborhoods of San Francisco. We then analyzed phenotypic and genotypic antibiotic resistance (ABR) of the isolates, as well as clonal relationships based on cgMLST and single nucleotide polymorphisms (SNPs) of the core genomes. Using Bayesian inference, we reconstructed the transmission dynamics between humans and canines from multiple local outbreak clusters using the marginal structured coalescent approximation (MASCOT). Overall, we found human and canine samples to carry similar amounts and profiles of ABR genes. Our results provide evidence for multiple transmission events of ABR-Ec between humans and canines. In particular, we found one instance of likely transmission from canines to humans as well as an additional local outbreak cluster consisting of one canine and one human sample. Based on this analysis, it appears that canine feces act as an important reservoir of clinically relevant ABR-Ec within the urban environment. Our findings support that public health measures should continue to emphasize proper canine feces disposal practices, access to public toilets and sidewalk and street cleaning. Importance: Antibiotic resistance in E. coli is a growing public health concern with global attributable deaths projected to reach millions annually. Current research has focused heavily on clinical routes of antibiotic resistance transmission to design interventions while the role of alternative reservoirs such as domesticated animals remain less well understood. Our results suggest canines are part of the transmission network that disseminates high-risk multidrug resistance in E. coli within the urban San Francisco community. As such, this study highlights the need to consider canines, and potentially domesticated animals more broadly, when designing interventions to reduce the prevalence of antibiotic resistance in the community. Additionally, it showcases the utility of genomic epidemiology to reconstruct the pathways by which antimicrobial resistance spreads.

Keywords: Antibiotic Resistance; Canines; ESBL; Environment; Genomic Epidemiology; Phylodynamics.

Figure 1:

Fifty-nine fecal samples were collected over a one-mile range in the Tenderloin and SoMa neighborhoods of San Francisco, CA, USA. Samples were collected over one month period on four collection dates. Spatial distribution of sample sites was visualized using QGIS in a Pseudo-Mercator projection.

Figure 2:

Isolates from humans and dogs were selectively plated on MacConkey, MacConkey and Ampicillin, or MacConkey and Ceftriaxone, and assessed for phenotypic resistance (A). Sequences of isolates determined by whole genome sequencing were analyzed using the ResFinder database with ABRicate (B).

Figure 2:

Isolates from humans and dogs were selectively plated on MacConkey, MacConkey and Ampicillin, or MacConkey and Ceftriaxone, and assessed for phenotypic resistance (A). Sequences of isolates determined by whole genome sequencing were analyzed using the ResFinder database with ABRicate (B).

Figure 3:

Contigs containing ARGs determined by ResFinder were assigned as plasmid or chromosomal according to mlplasmids and MOB-suite in human (A) and dog (B) isolates. Replicon types of the contigs were determined by MOB-suite. Distribution of localization and replicon types was visualized using ggsankey in R studio.

Figure 4:

Maximum likelihood phylogenetic analysis. Phylogram depicting the best estimate of the phylogenetic relationships was computed with RAxML using 200473 SNP sites among the core genome of E. coli isolates with bootstrapping of 100 replicates. ClonalFrameML was used to correct the branch lengths of the tree to account for recombination. Fecal sample source, phylogroups, sequence type and ARG carriage of each isolate is indicated. Tree clades are indicated by gray shading.

Figure 5.

Transmission of E. coli between canine and humans. (A) Inferred number of host jumps. Host jumps are computed as the number of edges for which parent and child nodes are in different states. The total number denotes the posterior distribution of host jump events in either direction. (B) Posterior distribution of typed phylogenetic trees inferred from local outbreak clusters using MASCOT. The trees show the densitree representation of the local outbreak clusters plotted using ggtree. Resistance gene profile for isolates are depicted on the right, with orange tiles representing a resistance gene to be chromosomally located, blue tiles representing a resistance gene to be plasmid located and green tiles representing an undetermined resistance gene location.