Pathogenomics and clinical recurrence influence biofilm capacity of Escherichia coli isolated from canine urinary tract infections

Abstract

Biofilm formation enhances bacteria's ability to colonize unique niches while protecting themselves from environmental stressors. Escherichia coli that colonize the urinary tract can protect themselves from the harsh bladder environment by forming biofilms. These biofilms promote persistence that can lead to chronic and recurrent urinary tract infections (UTI). While biofilm formation is frequently studied among urinary E. coli, its association with other pathogenic mechanisms and adaptations in certain host populations remains poorly understood. Here we utilized whole genome sequencing and retrospective medical record analysis to investigate associations between the population structure, phenotypic resistance, resistome, virulome, and patient demographic and clinical findings of 104 unique urinary E. coli and their capacity to form biofilms. We show that population structure including multilocus sequence typing and Clermont phylogrouping had no association with biofilm capacity. Among clinical factors, exposure to multiple antibiotics within that past 30 days and a clinical history of recurrent UTIs were positively associated with biofilm formation. In contrast, phenotypic antimicrobial reduced susceptibility and corresponding acquired resistance genes were negatively associated with biofilm formation. While biofilm formation was associated with increased virulence genes within the cumulative virulome, individual virulence genes did not influence biofilm capacity. We identified unique virulotypes among different strata of biofilm formation and associated the presence of the tosA/R-ibeA gene combination with moderate to strong biofilm formation. Our findings suggest that E. coli causing UTI in dogs utilize a heterogenous mixture of virulence genes to reach a biofilm phenotype, some of which may promote robust biofilm capacity. Antimicrobial use may select for two populations, non-biofilm formers that maintain an arsenal of antimicrobial resistance genes to nullify treatment and a second that forms durable biofilms to avoid therapeutic insults.

Fig 1. Maximum likelihood tree of 104 canine UPEC isolates.

Maximum likelihood phylogenetic tree of core SNP variation among 104 canine UPEC isolates and their associated phylogroup (innermost ring), most frequent MLST sequence types (middle ring), and biofilm capacity (outermost ring). Biofilm formation was categorized (none, weak, moderate/strong) based on a qualitative scale separated by specific optical density value cut-offs.

Fig 2. Phenotypic reduced susceptibility of UPEC isolates to 15 antimicrobials.

Percentage of isolates with reduced susceptibility (RS) to the 15 antimicrobials tested stratified by biofilm capacity. Asterisk (*) indicate RS phenotypes that were negatively associated with biofilm formation (P<0.10) prior to correcting for multiple comparisons. Amox = Amoxicillin-clavulanic acid; Amp = Ampicillin; Faz = Cefazolin; Pod = Cefpodoxime; Taz = Ceftazidime; Chlor = Chloramphenicol; Tet = Tetracycline; Doxy = Doxycycline; Enro = Enrofloxacin; Marbo = Marbofloxacin; Prado = Pradofloxacin; Gent = Gentamicin; Amik = Amikacin; PipTaz = Piperacillin Tazobactam; SMZ-TMP = Trimethoprim Sulfamethoxazole; RS = Reduced susceptibility; MDRS = Multidrug reduced susceptibility.

Fig 3. The acquired resistome profile of the 104 canine UPEC isolates.

The resistome profiles of UPEC isolates stratified by biofilm formation and 30-day history of antibiotic exposure. Sample prevalence of dogs exposed to antibiotics and resistance genes is presented at the bottom of the figure. The cumulative number of acquired resistance genes in the resistome of each isolate is presented in the furthest right column.

Fig 4. Co-occurrence networks of the unique virulence gene relationships among different levels of biofilm production.

(A) Virulence co-occurrence network of unique gene relationships among isolates forming any type of biofilm. Only significant co-occurrences with a Spearman’s rho ≥0.6 are shown. (B) Virulence co-occurrence network of unique gene relationships among isolates forming moderate/strong biofilms. Only significant co-occurrences with a Spearman’s rho ≥0.6 are shown. Nodes are color coded to represent virulotype clusters based on community partitioning analysis.

Fig 5. Maximum-likelihood tree of isolates from dogs that submitted two or more UPEC during the study period.

Maximum likelihood phylogenetic tree of core SNP of 19 separate UPEC isolates. Metadata of the samples are present to the left of the tree including within-patient isolate clusters, isolate phylogroup, biofilm capacity and a heat map of between-isolate whole genome similarity.