Experimental mouse lethality of Escherichia coli isolates, in relation to accessory traits, phylogenetic group, and ecological source.

BACKGROUND Whether accessory traits, phylogenetic background, or ecological source best predicts extraintestinal virulence within Escherichia coli is undefined. METHODS A total of 90 E. coli strains (18 fecal isolates and 72 extraintestinal-infection isolates) were characterized for 55 accessory traits and phylogenetic group (A, B1, B2, or D). Bacterial traits and ecological source were compared with experimental mouse lethality. RESULTS Of the 90 strains, 41% were "killers" (i.e., killed > or =90% of mice). By univariate analysis, multiple group B2-associated traits (including malX [pathogenicity-island marker], pap [P fimbriae] elements, usp [uropathogenic-specific protein], and fyuA [yersiniabactin system]) were most closely associated with killer status, followed by group B2 (or non-group A) status and then by nonfecal origin. Stepwise multivariate analysis identified pap, malX, usp, fyuA, and B2 (all of which were positive predictors) and ireA (which was a negative predictor) as significant predictors of killer status. Killer strains segregated significantly from nonkiller strains, according to accessory-trait profiles. Factorial analysis of correspondence placed group B2 among the traits most closely associated with killer status, but not as the closest. CONCLUSIONS Specific group B2-associated accessory traits are more potent predictors of experimental virulence among E. coli isolates than is either phylogenetic background or ecological source. Molecular typing can estimate an E. coli isolate's extraintestinal virulence potential, regardless of source.