Abstract
Since their identification in 2008, E. coli ST131 isolates have become a major cause of cystitis and pyelonephritis. However, these strains are not known to encode different, or increased numbers of virulence factors compared to other sequence types of pathogenic E. coli. To date, no data in the literature can entirely explain the global dominance of ST131 E. coli. The purpose of this study was threefold: 1) to evaluate infectivity among ST131 E. coli compared to non-ST131 clinical isolates using a murine model of urinary tract infection, 2) to evaluate the correlation between in vitro assay results and in vivo infectivity using the same mouse model, and 3) to evaluate the transcriptome of ST131 E. coli strains in culture conditions and compare those results to the transcriptome of ST131 E. coli directly from an infected kidney. In vitro assays were conducted according to published protocols and included: motility assays, biofilm formation, epithelial cell adhesion and invasion, and curli production. Twenty UPEC clinical isolates of E. coli ST131 and non-ST131 were used for both in vivo and in vitro studies. Six mice per isolate were inoculated via urethral catheterization. CFUs were determined from bladder and kidneys. In vitro and in vivo correlations were evaluated by multiple linear regression analysis. Transcriptome analysis was evaluated by RNAseq on five clinical isolates (W15, XQ12, FHM6, CUMC247 (ST131s) and C15 (ST405)). Metabolomic evaluations were conducted using three clinical ST131 isolates (W15, XQ12, FHM6) and the lab strain BW25113.
Both ST131 and non-ST131 strains could infect the bladder and kidneys; however, a greater number of ST131 strains were able to infect individual mice than the non-131 isolates (p=0.005). Non-ST131 isolates were able to cause bladder or kidney infections with relatively similar CFUs, indicating a similar fitness among these organisms in the bladder and/or kidney environments, but a potential difference in the ability to get to the bladder and/or kidney to establish infections. In vitro assays were poor predictors of in vivo infectivity, with R2 Pearson Correlation values less than ±0.5 for any pair, indicating little to no statistical associations.
Transcriptomic and metabolomic data are invaluable for providing insight into the fitness and virulence mechanisms of pathogenic E. coli at the time and site of infection. In this study we saw important expression differences of outer membrane proteins including OmpA and OmpN in establishing infection while evading host immune responses. The two component systems for regulation of formate hydrogenylase and potassium metabolism were also increased in pyelonephritic infections. Regulation of SN-glycerol 3-phosphate, and the associated transporters encoded by glp and ugp genes also appear to be involved in the pathogenesis of ST131 pyelonephritis and warrant further study. By elucidating genetic changes occurring during infection through transcriptomic analysis and comparing those data with metabolomic analyses, we can gain valuable insight into the systems required for E. coli to colonize bladder or kidney tissues.