New Insights Into Shiga Toxigenic Escherichia coli Pathogenesis: When Less Is More

(See the major article by Russo et al on pages 1271–9.)

Hemorrhagic colitis due to Shiga toxin–producing Escherichia coli (STEC) is perhaps the most feared foodborne infection in the industrialized world. The Centers for Disease Control estimates that there are 265 000 cases of STEC infection in the United States each year [1]. Whereas hemorrhagic colitis is a painful and often severe illness, the greatest concern with STEC infections is hemolytic-uremic syndrome (HUS), which develops in 5%–10% of patients and can cause permanent and occasionally fatal renal or neurologic sequelae. Some of the greatest frustrations with STEC infections are the lack of demonstrated effectiveness of antibiotics or other targeted treatments and their well-demonstrated ability to cause outbreaks because of contamination of seemingly innocuous foods (like sprouts, ready-to-eat salads, and tree nuts), in addition to ground beef, where these infections were first identified.

The first reported serotype of STEC, and still the most common cause of hemorrhagic colitis and HUS, is O157:H7. Most O157:H7 isolates carry a specific complement of virulence factors within a locus of enterocyte effacement (LEE). The proteins encoded on the LEE allow O157:H7 to adhere intimately to intestinal epithelial cells and produce cytoskeletal changes in the cells, which greatly enhance its pathogenicity. However, the deadliest virulence factors of O157:H7 are its Shiga toxins, Stx1 and Stx2 (formerly called Shiga-like toxins). For E. coli to be called EHEC, it must possess the LEE and Stx1 and/or Stx2; those that express Stx1 and/or Stx2 without the LEE are referred to as STEC. In general, it is believed that non-EHEC STEC are less virulent than O157:H7, but some (like the O104:H4 enteroaggregative STEC that caused a large European outbreak in 2011) have alternative virulence factors that lead to severe illness [2].

Both Stx1 and Stx2 are AB₅ toxins with a similar mechanism of action. After the B subunit binds to its glycolipid receptor, Gb3, the toxin is internalized by endocytosis and transported retrograde to the cytosol, where the A subunit enzymatically damages ribosomal RNA, leading to cell death. In isolated hemorrhagic colitis, the toxin-mediated necrosis remains largely confined to the gut, but when toxins enter the circulation, they can lead to HUS. The high expression of Gb3 on vascular endothelium and epithelial cells in the colon, central nervous system, and kidneys is believed to explain why STEC-induced HUS affects these organs so specifically [3–5].

Although Stx1 and Stx2 are very similar mechanistically, there is a clear epidemiologic association of Stx2 with severe human illness (hemorrhagic colitis and HUS), whereas many bovine isolates express only Stx1 [6]. Strains expressing Stx1 without Stx2 are reported to cause human disease, but they are less likely to cause severe colitis or HUS [7, 8]. These and other epidemiologic studies, curiously, have shown a higher association with severe illness and HUS in STEC strains expressing Stx2 without Stx1, compared with strains expressing both toxins, even in the presence of other virulence factors [9, 10]. The reasons for this difference are not clear, given the similarity between the toxins. In this issue of The Journal of Infectious Diseases, Russo et al attempt to understand this phenomenon, using an oral mouse intoxication model. Previous publications have shown that intraperitoneal Stx1 and Stx2 are both highly toxigenic in mice, with 50% lethal doses of around 125 ng and 1 ng, respectively [11]. The authors of the current study extend these findings to show that oral Stx2 intoxication is also highly lethal (albeit with reduced potency) but, curiously, that oral Stx1 is well tolerated at doses of >150 µg per mouse [12].

These findings led to a reasonable hypothesis that Stx1 has impaired ability to cross the intestinal barrier, which the authors chose to study in more detail in the current study. To accomplish this, they covalently labeled the toxins with a fluorochrome (Alexa Fluor 750) and used ex vivo fluorescent imaging to localize distribution of toxins after oral gavage. Contrary to what was expected, they found basically equivalent dissemination of Stx1 and Stx2 to internal organs, with the strongest fluorescent signal in the kidneys for both toxins. They next examined the effect of coadministration of Stx1 and Stx2 on mouse lethality and found that Stx1 given with Stx2 reduced its toxicity, as measured by weight change and overall survival. This effect was not linear but showed peak protection at a Stx1 to Stx2 ratio of 5:1, with higher doses of Stx1 (10:1 ratio) being apparently less protective. The changes in weight loss and mortality corresponded with biochemical markers of nephrotoxicity, including creatinine, blood urea nitrogen, sodium, and neutrophil gelatinase-associated lipocalin; renal pathology scores were also done but failed to show significant differences.

Russo et al did additional experiments to explore the mechanisms of this protection. They substituted an Stx1 toxoid (SW09) for Stx1 and found that it offered similar protection against Stx2. They further showed that treatment with Stx1 3 hours prior to Stx2 improved the protection. There were some curious additional findings that may merit further study. One is that coadministered Stx1 and Stx2 did not colocalize in the kidneys (based on immunofluorescent microscopy findings) but appeared in different renal tubules. Another is that neutralization of Stx1 with a monoclonal antibody only partly reversed the protection against weight loss but did not reduce survival. Finally, the authors showed that EHEC O157:H7 strains 933 and 2812 produced different ratios of Stx1 to Stx2 in vitro, with the former producing about 5-fold more Stx1 and the latter producing only 2.7-fold more. Interestingly, these 2 outbreak strains caused different rates of HUS; 2812 was more virulent, and it is intriguing to speculate that the lower Stx1 to Stx2 ratio could be part of the reason.

Overall, Russo et al rightly conclude from their data that, at least in the mouse oral inoculation model, Stx1 protects against Stx2 toxicity. They provide some plausible mechanistic explanations for this, based on their data. The protection using Stx1 toxoid suggests that the B subunit, rather than the enzymatic A subunit, is responsible; this, in turn, would imply that Stx1 somehow interferes with binding and/or internalization of Stx2. This finding could have to do with unique effects of the 2 toxins on Gb3 lipid raft formation, but further study is needed to confirm this.

There are a few limitations to the present study that merit discussion. One is the discrepancy between weight loss curves and survival in several of the experiments. The cause of mortality in the mice was not well described, although the authors note that mice did not show signs of morbidity or humane end points requiring euthanasia prior to death. Statistical differences among the various groups were largely absent, although type II error could easily be responsible, and it is reasonable to accept this limitation rather than perform the requisite numbers of mouse experiments to facilitate multiple comparisons. Another confusing point is the fact that oral Stx1 does appear to reach the kidneys, yet lacks toxicity in this model. Even odder is the fact that Stx1 appears to be less protective at a 10:1 ratio to Stx2 than a 5:1 ratio—this difference would make sense if Stx1 itself were toxic but is hard to understand otherwise.

The final limitation is the question of how relevant this model is to human enterohemorrhagic E. coli/STEC infection, given observed differences between mouse and human pathogenicity in other models. The nature of the toxins, however, would suggest that their effect is broadly similar among mammalian species. Applying this study to an STEC infection model (perhaps using strains engineered to overexpress Stx1 vs Stx2) would be an important next step. Nevertheless, this intriguing study goes a long way toward understanding the relative contributions of Stx1 and Stx2 to hemorrhagic colitis and HUS, and it helps to provide a mechanistic explanation for long-standing epidemiologic observations. If the mechanism by which Stx1 protects against Stx2 can be fully elucidated, this understanding may open the door to a sorely needed therapy for STEC infection, based on that mechanism.

Note

Potentialconflict of interest. T. S. S. received speaker honoraria from Merck and Bristol-Meyers-Squibb; received research funding from Merck, Sanofi-Pasteur, and Rebiotix; and served as an advisor for Merck and Pendopharm. The author has submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References