EHEC (and enterobacteria in general) do not normally colonize the mouse intestine, which is largely colonized by gram-positive anaerobes (4
). When conventional or SPF mice are inoculated with live EHEC, bacteria are detectable only briefly in the feces, and bacterial counts fall to undetectable levels within a few days (7
). Generally, signs of disease and lesions are minimal or not detectable (29
). Conlan et al. (7
) reported death in mice inoculated with one of nine EHEC strains tested, but neither clinical signs nor lesions were reported in their study. A second study reported rapid onset of renal and colonic lesions, neurologic signs, and death in SPF mice that were orally inoculated with several strains of EHEC O157, but the mice in that study became bacteremic and intestinal colonization was not documented (19
The most widely used infection model is the streptomycin-treated mouse model, in which a subset of the normal enteric microbiota is suppressed by streptomycin treatment and mice are orally inoculated with streptomycin-resistant EHEC strains (3
). These mice are highly susceptible to colonization by less than 100 CFU of EHEC, and colonization persists for as long as the mice are treated with streptomycin. Signs of disease and lesions vary according to the bacterial strain, laboratory, and other (unknown) factors and range from no disease or lesions (12
) to acute renal tubular necrosis and death (53
). A few reports have also included descriptions of intestinal epithelial necrosis (44
), cerebral hemorrhage (12
), or brain edema (22
). The residual microbiota in streptomycin-treated mice have not been described and likely differ between laboratories (and even between individual mice), possibly accounting for some of the differences in the outcomes of the experiments.
Several reports have described disease due to EHEC in germ-free mice, as described in this study (1
). Like streptomycin-treated mice, germ-free mice are exquisitely susceptible to colonization by EHEC, and the colonization is indefinite. Descriptions of signs and lesions are inconsistent. Lesions that have been reported include renal tubular necrosis (48
), necrosis of colonic epithelial cells (16
), and neurologic signs or lesions (16
). Most previous studies with germ-free mice, however, did not investigate or report clinical signs or lesions (1
). The role of age and sex in the susceptibility of mice to EHEC has not been investigated previously with any model.
The results of this study are consistent with the results of previous reports for streptomycin-treated and germ-free mice in several respects. First, we demonstrated that germ-free mice are exquisitely susceptible to colonization by EHEC and that inoculation of as few as 100 bacteria results in a bacterial density of 109
CFU/g of feces or more by 1 day PI and persistence until death of the mouse or (in the case of nonpathogenic strains) for the duration of the experiment. We also demonstrated that the principal outcome (and probably the cause of death) of pathogenic EHEC infection in mice is acute renal tubular necrosis and that vascular lesions are subtle in mice and are confined to multifocal glomerular thrombosis and red blood cell sludging. Previous studies in which kidneys were examined generally demonstrated that there was renal tubular necrosis in infected mice, although the reported severity of disease varied between studies (3
). A few studies showed that there were consistent vascular or glomerular lesions (12
), and the descriptions of these lesions varied between studies, confirming our findings that glomerular disease is present but mild in EHEC-infected mice. Several studies that used an injection model, in which Stx or Stx in combination with lipopolysaccharide was injected parenterally into mice, described glomerular lesions (11
), supporting our finding that virulent EHEC is capable of inducing glomerular disease in mice.
In addition to renal lesions, we report several new findings concerning intestinal colonization of infected mice here. First, we showed that bacteria are present throughout the lower intestine but adhere to the cecal and ileal mucosa and not to the colonic mucosa. Culture of the ileum, cecal contents, cecal wall, and colon detected bacteria at all sites and demonstrated that bacterial populations adhered to the cecal wall. Adherence was morphologically detectable only in histologic sections of the ileum and cecum of the mouse and was largely absent in the colon. This distribution of colonization in mice has not been described previously and could explain the failure of some previous studies to detect adherence, because only the colon was examined (29
), while other studies reported adherence in the ceca of infected mice (30
). We also showed that although mice do not develop diarrhea due to EHEC, chronic infection is accompanied by marked luminal fluid accumulation in the cecum. The restriction of fluid accumulation to the ceca without diarrhea in these mice could be due to the fact that bacterial adherence occurs largely in the cecum, inducing fluid loss there but not in the colon. Mice are desert animals with very efficient fluid reabsorption mechanisms (46
). In EHEC-infected mice, since bacterial adherence is confined to the cecum, the fluid lost into the cecum is likely reabsorbed in the colon, resulting in the absence of diarrhea. Thus, germ-free mice do develop enteric disease due to EHEC, although the lesions develop more slowly and the distribution is different than that in humans, likely due to differences in the pattern of adherence of EHEC and in the physiology of mice and humans.
In this study we showed that while sex does not affect susceptibility to EHEC-associated disease, age is an important factor. Mice of all ages were susceptible to infection and disease, but infant mice were the most susceptible to both clinical disease and renal lesions. This finding correlates with the increased incidence of EHEC-associated HUS in children compared to adults (2
). We also showed that while (as previously described for mice [37
]) BUN is an insensitive marker of renal disease, urine specific gravity is a sensitive indicator of the onset of renal failure and correlates well with histologic evidence of renal disease. This has not been shown previously in a mouse model of EHEC.
Our investigations demonstrated that Stx2 is necessary but not sufficient to induce both renal disease and cecal fluid accumulation in mice. Deletion of Stx2 abrogated pathogenicity both in 86-24, which produces only Stx2, and in EDL933, which produces both Stx1 and Stx2. In the latter case, Stx1 alone was insufficient to induce disease in mice colonized by EDL933::Δstx2
. The results of previous studies performed by other investigators support our findings, although to our knowledge, no one has used Stx2 deletion mutants to test directly the role of Stx2 in renal disease. Several studies have demonstrated that plasmid-mediated expression of Stx2 confers pathogenicity on laboratory strains of E. coli
), and a number of studies have demonstrated that antibodies directed against Stx2 provide protection (24
); both of these findings provide indirect evidence of the role of Stx2 in renal disease. Donohue-Rolf et al. (10
) showed that isogenic Stx2-negative mutants failed to cause neurologic lesions or signs in gnotobiotic piglets, but renal lesions were not described. Finally, several studies have demonstrated the nephrotoxic effect of parenterally injected Stx1 or Stx2 (11
), again providing indirect evidence that Stx is at least partially responsible for renal disease.
Because the only strain in this study that expressed Stx1 alone, DEC10B, did not cause disease in mice, we could not evaluate Stx1 directly. EDL933::Δstx2
did not cause disease in mice in spite of the presence of Stx1 in cecal contents, suggesting that Stx1 does not contribute to disease in this model, but the evidence remains indirect. Our results are compatible with previous studies cited above showing that Stx2 is necessary for disease in orally infected mice, but studies performed with other models suggest a role for Stx1 in disease. For example, Sjogren et al. (47
) observed enhanced severity of enteric disease in rabbits infected with RDEC that was engineered to express Stx1. Also, direct injection studies with mice have shown that parenteral Stx1 alone, as well as Stx2, causes renal disease (20
). Definitive evidence of the in vivo pathogenicity of Stx1 awaits identification of a pathogenic strain that produces Stx1 alone.
Unlike a previous report (40
), we found no evidence of a role for Stx2 in colonization. In our hands, both Δstx2
mutants colonized as well as the wild-type parental strains. This finding is compatible with one report which demonstrated that Stx2 did not influence colonization in a rabbit model (39
), but it differs from the results of the study of Robinson et al. (40
), in which challenge of conventional SPF mice with mutants deficient in Stx2 resulted in less colonization than challenge with the wild-type parent strain. However, because Robinson et al. used SPF mice with a full complement of intestinal microbiota, our study may not be directly comparable. EHEC did not cause disease in SPF mice, and the colonization density in SPF mice was low and decreased rapidly over the course of the experiment, indicating that neither the wild type nor the Stx mutant colonized well. It is possible that in the presence of normal murine microbiota, Stx2 provides a competitive advantage to EHEC strains. If this is the case, we would not expect differences in colonization between mice monocolonized with an Stx deletion mutant and mice colonized by the wild-type parent. However, in the presence of other intestinal microbiota, the Stx deletion mutant might be cleared faster than the wild type. Confirmation of this hypothesis awaits direct competition studies with Δstx
and wild-type bacteria and/or studies with mice having a defined flora.
In spite of the dependence of disease on the presence of Stx2 in the two strains examined in this study, several other EHEC strains that did produce Stx2 failed to cause disease in mice, demonstrating that while Stx2 is necessary to cause disease, it is not sufficient. In addition, although the severity of disease in mice varied for the bacterial strains, the concentration of Stx2 in the cecum of infected mice did not correlate with the severity of disease. These results suggest that strain-specific factors in addition to toxin production contribute to disease due to EHEC in mice. Disease did correlate with O157 serotype in this study, consistent with the prominence of O157 serotypes in clinical disease (18
), but only two non-O157 strains were examined in this study, precluding definitive interpretation.
Stx-independent differences in pathogenicity of EHEC strains have been described previously. The manifestations of clinical outbreaks associated with Stx2-producing strains vary (26
), suggesting that, as we have shown in mice, production of Stx2 alone does not completely account for the variation in pathogenicity among strains. A few studies have compared the pathogenicities of different EHEC strains in streptomycin-treated mice (24
), and the results have been similar to our results, showing differences in disease outcome between EHEC strains, all of which produce Stx2. One possible explanation for these differences is genetic diversity among pathogenic strains. Marked genomic diversity among Stx2-producing O157 strains has been described by several authors (33
), and a recent study suggested that differences in disease outcome may be due to genotypic diversity among strains, regardless of Stx2 (26
). Genetic comparison of strains that differ in virulence is likely to reveal specific factors that suppress or enhance Stx-associated disease.