Urinary tract infections (UTIs) reduce the quality of life of millions and is a considerable cause of disease and death in vulnerable individuals1. Importantly, UTI is a major driver of antibiotic consumption and antibiotic resistance, thus emphasizing the need to understand UTI pathogenesis and develop new treatment strategies2. The mouse has for decades been regarded as a critical experimental model for deciphering basic UTI pathogenesis, due to its small size, ease of maintenance, low cost and amenability to genetic manipulation3. Nevertheless, it also has shortcomings as a model of human disease. In health science, the emphasis on mice is believed to be a contributor to the failure of drug and vaccine clinical trials4,5. Awareness of the limitations of this model is therefore critical in translational medical research.
Of relevance to the use of mice in UTI research is the unique capacity of mice to produce highly concentrated urine. Mice produce urine at concentrations far outside the range of humans and most other mammals6. It has remained unexplored how this fundamental physiological trait of mice impacts the transferability of UTI research to humans.
We hypothesized that the highly concentrated urine of mice affects bacterial growth fitness and furthermore is an inducer of bacterial filamentation, a key step in UTI pathogenesis that has been extensively studied in mice where filamentation is abundant3,7,8. Uropathogenic Escherichia coli (UPEC) elongates and forms filaments to some extent during growth in highly concentrated human urine, and during experimental infection of human bladder epithelial cells exposed to human urine9,10. Induction of the phenotype requires human urine at concentrations above a certain threshold and is moreover affected by surface growth and pH9,10,11. UPEC filaments have been observed in human urine from patients with UTI12, although our experience from routine inspection of human patient urine samples is that they are relatively rare, unless the patient is treated with filament-inducing antibiotics.
To assess the effect of the high osmolarity of mouse urine on UPEC morphology and infectious potential, we established a natural diuresis mouse model by allowing animals access to sweetened water (tap water supplemented with 20% (w/v) glucose) 30 h before infection. Increased water intake reduces the urine concentration of mice—measured as urine specific gravity (USG, that is, density)—to levels comparable to those of humans (normal range 1.005–1.030) (Fig. 1a). Increased water intake did not significantly affect urine glucose levels, pH, the urine albumin-to-creatinine ratio or total urinary protein excretion; however, wide variability was observed in the urine albumin-to-creatinine ratio in diuretic mice (Supplementary Fig. 1).
a, Promoting water intake by feeding mice with sweetened water significantly reduced USG (equals density) from day 0 (baseline) to day 1 (24 h after giving sweetened water) (n = 64, P b–d, Significantly higher level of UPEC filamentation was observed in mice with access to regular tap water (control; b) (n = 10) compared with mice with access to sweetened water (diuretic mice; c) (n = 27); results were quantified by manual counting using confocal laser scanning microscopy (d), with bars showing mean ± s.d. (P = 0.0003, Mann–Whitney test). e–g, The low USG in diuretic mice increased UPEC survival after 18 h of infection in the bladder (e) (P n = 12), diuretic (n = 20)); in the urine (g) (P = 0.01, Mann–Whitney test; control (n = 17), diuretic (n = 12)); but CFUs were not significantly different in kidneys (f) (P = 0.14; control (n = 24), diuretic (n = 31)). h, Osmo-responsive genes proV and ompC, monitored by RT–qPCR were induced in control mice (n = 12) relative to diuretic mice (n = 10), (P i, Compared with control mice, diuretic mice were significantly more susceptible to infection at reduced infectious doses of 106 CFU ml−1 (P 4 CFU ml−1 (P P P P P
Diuresis and control mice were then experimentally infected with the prototypical UPEC isolate UTI89 transformed with the green fluorescent protein-encoding plasmid pMAN01. The mice were euthanized after 18 h—a sweet spot for observing filaments in this model (which occur 12–20 h after infection)—and bacterial morphology was examined on splayed bladders by microscopy7,13. In control mice, extensive filamentation was observed, whereas bacteria remained mainly rod-shaped in diuresis mice (mean filamentation score of 2.80 and 1.07, respectively; P = 0.0003, Mann–Whitney test) (Fig. 1b–d). In 41% of diuresis mice, rod-shaped bacteria were exclusively observed without any detectable filamentation (Fig. 1b–d). Bacterial colony-forming units (CFUs) were significantly higher in the urine (P = 0.01, Mann–Whitney) and bladders (P 1e–g). Reverse-transcription quantitative polymerase chain reaction (RT–qPCR) analysis indicated osmotic stress in bacteria from control mice compared with diuresis mice, based on increased expression of the central osmo-responsive gene proV from the osmo-inducible proU operon and ompC (Fig. 1h, mean log2 fold change (FC) of 1.90 and 0.85, respectively)14,15.
The higher bacterial burden in the diuresis mice led us to speculate that USG might also affect susceptibility to infection. Mice are intrinsically resistant to UTI and require high inocula of 108–109 CFU ml−1 for successful infection3. This dose is approximately a million times higher than that used in pigs, an alternative model animal that is naturally susceptible to UTI and produces urine with USG levels comparable to those of humans16. To test this hypothesis, groups of diuresis and control mice were inoculated with 106 and 104 CFU ml−1. At 106 CFU ml−1, only one of six control animals (17%) became infected compared with seven of seven (100%) diuretic mice (P 1i). Similarly, for 104 CFU ml−1, all control animals (n = 6) resisted infection whereas diuresis mice remained highly susceptible, with five of six (83%) becoming successfully infected (P = 0.015). Of note, in the previous experiment, all mice from both groups inoculated with 109 CFU ml−1 became infected (Fig. 1e). Taken together, these results suggest that lowering the USG of mice to values in the human range decreases the necessary infectious dose by 10,000-fold or more.
A plausible explanation for UPEC’s attenuated infectious potential in mice is that hyperosmolar urine inhibits growth. To test this hypothesis, urine specimens were collected from control mice, inoculated with UTI89 and cultured overnight. Growth experiments in urine specimens from human volunteers and pigs were conducted in parallel. Mouse urine osmolarity was outside the range of both humans and pigs (Fig. 2a,b). While UTI89 proliferated in human and pig urine by 1,000- to 10,000-fold (no significant difference between these two species), the bacterial population was significantly reduced in mouse urine by an average of 10-fold (P = 0.002, one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test) (Fig. 2c). Two other UPEC strains (NU14 and CFT073) were also growth-inhibited in mouse urine, thus demonstrating that UPEC strains are substantially challenged in the urinary tract of mice, contrary to humans and pigs (Fig. 2d). Furthermore, when incubating the three strains in urine collected from diuretic mice, we found growth rates equal to growth in human urine (Fig. 2d); proV expression was also downregulated in diuretic mouse urine compared with control mouse urine (Fig. 2e).
a,b, Urine osmolarity (a) and USG (b) of mice (n = 18), humans (n = 19) and pigs (n = 15). c, CFUs per milliliter after overnight incubation of UTI89 in urine from mice, humans and pigs. d, The UPEC strains UTI89, CFT073 and NU14 were all significantly reduced in colony counts after 18 h incubation in pooled mouse urine (n = 3 for each strain) compared with inoculum (P P e, All strains significantly upregulated the osmo-responsive gene proV during growth in mouse urine compared with growth in urine from diuretic mice (P a–c), two-way ANOVA (d) and Mann–Whitney test (e). ****P
Our study demonstrates that the hyperosmolar urine of the mouse promotes UPEC filamentation, a widely accepted hallmark of UTI in humans3,7,8. This result aligns with earlier studies using human urine, which showed that filamentation is dependent on—or induced by—urine at the higher end of the normal range of osmolarity9,10,11,17. Reducing mouse urine osmolarity to levels corresponding to intermediate human urine osmolarity almost completely abolished bacterial filamentation, leading to increased bacterial infectious potential and survival in the mouse urinary tract.
This association aligns with findings from experimental studies in pigs, an animal with human-equivalent urine osmolality, where UPEC does not exhibit a filamentous growth response18. Furthermore, in contrast to results in the mouse model, recent genome-wide analyses of UPEC have revealed that osmotic stress-related genes were not fitness factors in pigs19. This pattern is also reflected in transcriptomic comparisons between human and mouse UTIs, which, despite reporting an overall correlation in bacterial gene expression profiles, identified hundreds of genes that were differentially expressed, including increased expression of proV during mouse infection20.
A potential limitation of the mouse diuresis model is that urine osmolarity may not be the only factor impacted by the intake of water with 20% glucose. Although we found that pH, glucose, albumin and total protein excretion were unaffected by this procedure (Supplementary Fig. 1), we cannot rule out that other urine properties such as metabolites or metal ions could be affected. In this regard, it should be mentioned that, although filamentation correlates with high overall urine solute concentration, studies with human urine indicate that specific low-molecular constituents are responsible for this phenotype rather than general osmolarity, and that pH also influences filamentation9,11. These specific factors were not further investigated in the current study. Furthermore, if filamentation occurs in the diuretic mice far outside the typical filamentation window in mice (12–20 h), we would not have captured it by relying on one time point (18 h). Nevertheless, our findings highlight the problem that the unique properties of mouse urine strongly influence UTI pathogenesis and UPEC survival.
The influence of hyperosmolar mouse urine on infectious potential and UTI pathogenesis highlights the need for vigilance among researchers working in the UTI field. To our knowledge, this factor has not been accounted for in previous mouse UTI studies and may have affected the interpretation of results, including the modeling of the widely accepted UTI pathogenic cascade8. In addition, the inhibitory effect of mouse urine on UPEC should be accounted for in preclinical evaluation of drugs and vaccines against UTI.
UTI vaccines, which are currently pursued by several companies worldwide, may show an overestimated efficacy in mice, as shown here, as bacteria are already challenged by a toxic urine environment. Regarding antibiotic susceptibility, β-lactam antibiotics—typically used against UTIs—target bacteria in a state of active cell division and may therefore have reduced activity against filamentous UPEC in mice. Indeed, studies have shown that UPEC survive β-lactam treatment in mice, but not in pigs18; however, the exact influence of urine osmolarity on the antibiotic activity remains to be explored.
In conclusion, the discrepancies between human and mouse urine need to be carefully considered when designing basic research studies and preclinical challenges in the murine UTI model. The model can be useful for certain studies, and its limitations can be partly circumvented by using diuretic mice, as demonstrated here. However, studies on complex UTI pathogenesis and drug activity may require the use of animals more similar to humans.
