A Toxic Environment: a Growing Understanding of How Microbial Communities Affect Escherichia coli O157:H7 Shiga Toxin Expression

Enterohemorrhagic Escherichia coli (EHEC) strains, including E. coli O157:H7, cause severe illness in humans due to the production of Shiga toxin (Stx) and other virulence factors. Because Stx is coregulated with lambdoid prophage induction, its expression is especially susceptible to environmental cues. Infections with Stx-producing E. coli can be difficult to model due to the wide range of disease outcomes: some infections are relatively mild, while others have serious complications.

ABSTRACT

Enterohemorrhagic Escherichia coli (EHEC) strains, including E. coli O157:H7, cause severe illness in humans due to the production of Shiga toxin (Stx) and other virulence factors. Because Stx is coregulated with lambdoid prophage induction, its expression is especially susceptible to environmental cues. Infections with Stx-producing E. coli can be difficult to model due to the wide range of disease outcomes: some infections are relatively mild, while others have serious complications. Probiotic organisms, members of the gut microbiome, and organic acids can depress Stx production, in many cases by inhibiting the growth of EHEC strains. On the other hand, the factors currently known to amplify Stx act via their effect on the stx-converting phage. Here, we characterize two interactive mechanisms that increase Stx production by O157:H7 strains: first, direct interactions with phage-susceptible E. coli, and second, indirect amplification by secreted factors. Infection of susceptible strains by the stx-converting phage can expand the Stx-producing population in a human or animal host, and phage infection has been shown to modulate virulence in vitro and in vivo. Acellular factors, particularly colicins and microcins, can kill O157:H7 cells but may also trigger Stx expression in the process. Colicins, microcins, and other bacteriocins have diverse cellular targets, and many such molecules remain uncharacterized. The identification of additional Stx-amplifying microbial interactions will improve our understanding of E. coli O157:H7 infections and help elucidate the intricate regulation of pathogenicity in EHEC strains.

ESCHERICHIA COLI

The species Escherichia coli encompasses a vast collection of organisms, including pathogens (1, 2), probiotics (3, 4), and common gut bacteria (5). E. coli bacteria are Gram-negative facultative anaerobes that are primarily found in the colon and cecum of vertebrates. They reside in the mucus layer, are shed into the intestinal lumen, and are excreted into the feces (6). The concentration of E. coli in human feces ranges from 10⁷ to 10⁹ CFU per g (7, 8). Multiple E. coli genotypes colonize a particular host at any given time (9, 10). These E. coli strains are present early in human development and have a mutual interaction with humans (11), benefiting their hosts through the production of menaquinone (vitamin K) (12) and riboflavin (vitamin B₂) (13).

Pathogenic E. coli strains are broadly classified based on their virulence factors and disease characteristics. The two major distinctions of pathogenic E. coli strains are nondiarrheagenic and diarrheagenic. Nondiarrheagenic E. coli strains are referred to as extraintestinal pathogenic E. coli (ExPEC) when they cause disease outside the host intestinal tract (14–16). ExPEC strains include uropathogenic E. coli (UPEC), sepsis-associated E. coli (SEPEC), neonatal meningitis-associated E. coli (NMEC), and others (17). Diarrheagenic E. coli strains cause diarrhea, as the name suggests, and include the Shiga toxin-producing E. coli (STEC) and enterohemorrhagic E. coli (EHEC) groups that are the focus of this review (11).

STEC strains, also known as verotoxigenic E. coli (VTEC) strains for their cytotoxicity to Vero cells, are a growing public health concern, first described in 1977 by Konowalchuk et al. (18). While all STEC strains produce Shiga toxin (Stx), not all are known human pathogens. The term EHEC is generally reserved for a subset of STEC strains that cause attaching and effacing (A/E) lesions and can lead to severe clinical complications, including hemorrhagic colitis (HC) and hemolytic-uremic syndrome (HUS) (2, 19, 20).

The most notorious serotype of EHEC is O157:H7, which emerged in the United States in the 1980s (21). Since then, there have been many outbreaks associated with undercooked hamburger (21, 22), various contaminated fruits (23) and vegetables (24, 25), and contaminated water (26, 27). Cattle are the primary reservoir for E. coli O157:H7 (28, 29); however, this serotype and non-O157:H7 STEC are frequently found in sheep, pigs, and other animals (30). E. coli O157:H7 has a low infectious dose, estimated to be between 10 and 100 cells (31). The time to onset of disease is typically 3 to 4 days, and the most common illness is hemorrhagic colitis, which is characterized by abdominal cramps and bloody diarrhea (31). HUS is a severe clinical sequela that is characterized by red blood cell death, thrombocytopenia, and acute renal failure (32). HUS occurs in approximately 18% of EHEC infections and is more common in children and the elderly (33, 34). STEC strains are estimated to cause more than 265,000 illnesses annually in the United States (35).

SHIGA TOXIN AND ITS TRANSCRIPTION

Shiga toxin is an AB₅ toxin, composed of one A subunit and five B subunits. It is produced by all EHEC strains and is the main virulence factor associated with hemorrhagic colitis. The 32-kDa A subunit is proteolytically nicked into two peptides, a 28-kDa peptide (A₁) and a 4-kDa peptide (A₂), by the cellular protease furin (36). Cleavage results in an activated A₁ fragment connected by a disulfide bond to the A₂-B complex (37). The B subunits are responsible for binding to the host receptor globotriaosylceramide (Gb₃) (38, 39). Gb₃ is found on various cell types but primarily on renal endothelial cells (40, 41). After binding, endocytosis occurs through clathrin-coated pits, and the toxin is transported to the Golgi membrane and endoplasmic reticulum, where A₁ is released from A₂-B after reduction of the disulfide bond (37). The A₁ subunit is then translocated to the cytoplasm, where it acts upon the 60S ribosomal subunit. Specifically, the A₁ peptide is an N-glycosidase and removes one adenine at position 4324 from the 28S rRNA (42). This cleavage inhibits protein synthesis, triggering the ribotoxic stress response and subsequently apoptosis (43). An additional mechanism has been proposed to explain the entry of Stx into the bloodstream: upon infection, neutrophils are recruited into the intestinal lumen and move across the intestinal epithelial layer. Neutrophil transmigration transiently increases the permeability of this layer, allowing Stx to traffic across Gb₃-negative intestinal epithelial cells and access the underlying tissue (44).

There are two distinct Shiga toxin types, referred to as Stx1 and Stx2 (45). Stx1 (also known as Stx1a) is almost identical to the Stx produced by Shigella, differing by one amino acid (46). Two additional Stx1 subtypes, Stx1c and Stx1d, have also been identified (47). Stx2 has many alleles, including Stx2a, Stx2b, Stx2c, Stx2d, Stx2e, Stx2f, Stx2g (47), Stx2h (48), and Stx2i (49, 50). Recent reports suggest that there are at least two additional subtypes, Stx2j and Stx2k, that have yet to be broadly accepted (51, 52). In general, infections with Stx1-producing STEC cause less severe disease than those with Stx2-producing STEC (53). This holds true even for strains producing both toxin types: strains that are Stx1⁺ Stx2⁺ are often less toxic in vitro and in vivo than strains that are Stx2⁺ only (54, 55). Indeed, epidemiological data suggest that the presence of Stx2 alone is associated with higher rates of HUS (53, 56–58). Of the Stx2 subtypes studied, Stx2a causes the most cases of HUS (59, 60), and therefore, most of the discussion that follows focuses on this subtype.

Stx1 and Stx2 are encoded on temperate lambdoid bacteriophages (61). Temperate lambdoid bacteriophages are viruses that infect certain bacteria; upon infection, the phage genome circularizes and inserts into the host chromosome using site-specific recombination. In this state, known as lysogeny, the phage DNA is referred to as a prophage and remains integrated into the bacterial chromosome. Alternatively, the phage may enter the lytic cycle, a switch that can occur upon the induction of the SOS response (62). The SOS response is mainly associated with DNA damage but can be induced by other stressors as well (63). When induced, the prophage recombines out of the genome, and assembly of new phage particles begins. As stx genes are transcribed late in the phage’s lytic cycle, Shiga toxin production occurs after prophage excision and before host cell lysis (64). An STEC strain’s Stx production can therefore be modulated by various chemical and environmental factors that influence the propagation of stx-carrying prophages (65).

REGULATING Stx EXPRESSION

Because Stx is correlated with the development of HUS (53, 56), treatment of EHEC infections must take care to minimize the expression of stx genes. For example, antibiotics are generally not a recommended therapy, as certain classes can induce stx phage and increase the amount of Stx released upon cell lysis (66). It is important to consider both of these parameters, the effect on the EHEC strain and on its phage, when evaluating current clinical approaches and designing alternatives.

While several groups have identified external factors that decrease Stx production by EHEC strains, not much is known about the mechanisms through which these factors act. Often, absolute Stx levels are reduced because the growth or colonization of EHEC is inhibited. For instance, coculture of the E. coli O157:H7 strain EDL933 with the probiotic E. coli strain Nissle 1917 (67, 68) or with the probiotic yeast Saccharomyces cerevisiae var. boulardii (69) yields less Stx than monoculture. Similarly, when EDL933 is cocultured with strains from the common probiotic species Lactobacillus, Bifidobacterium, and Pediococcus, stx_2a transcription (as measured by quantitative PCR [qPCR]) is downregulated in each coculture (70). Lactic acid, acetic acid, butyric acid, and formic acid are among the probiotic products that can decrease the expression of stx (70–73). Protection by probiotic bacteria has also been observed in infection models. In one study, mice colonized with the Bifidobacterium breve strain Yakult had significantly lower concentrations of Stx1 and Stx2 in their cecal contents than mice without B. breve (74). This effect was transient, however, and Stx expression was artificially induced by the administration of mitomycin C (74). In another example, precolonization of germfree mice with Bacteroides fragilis suppressed Stx production by EHEC and prevented EHEC-mediated inflammation and lethality (75). Beyond direct cellular interactions, the gut microbiota can produce soluble signals that dampen Stx expression, as demonstrated by spent supernatants of Bacteroides thetaiotaomicron (76, 77). Cultured medium from B. thetaiotaomicron apparently contains small molecules that inhibit stx₂ synthesis by repressing recA transcription (76).

Conversely, there are known small-molecule factors that promote stx expression. Host-derived epinephrine/norepinephrine and microbial autoinducer-3 are sensed by QseC, a bacterial histidine kinase that activates the transcription of stx genes via the induction of recA (78, 79). Ethanolamine, a membrane component that is released into the intestine as a result of cell turnover, can increase stx_2a expression during early- and mid-log-phase growth (80). One study links the stx-modulating ability of B. thetaiotaomicron discussed above to its vitamin B₁₂ uptake, suggesting that the availability of vitamin B₁₂ in the environment positively impacts Stx production (81). Vitamin B₁₂ may in fact be a cofactor for ethanolamine activation of stx genes, although further investigation of their roles is necessary (81). These and other molecular activators of stx have uncovered additional layers of virulence regulation in EHEC strains and will undoubtedly illuminate new disease dynamics.

In an effort to better classify the stimuli that may increase toxin production, this review characterizes several intraspecies mechanisms of Stx amplification. The remainder of the review examines two axes of Stx amplification by E. coli strains: cellular interactions, which promote phage infection, and acellular interactions, which promote phage induction.

DIRECT INTERACTIONS: CELLULAR MECHANISMS OF Stx AMPLIFICATION

During the course of an E. coli O157:H7 infection, the pathogen will come into contact with other bacteria. If these non-O157:H7 strains are infected by the stx-converting phage, they become new potential producers of Stx, and toxin levels can increase accordingly (82–85). Conversely, strains that are resistant or immune to stx phage lysis do not allow increased toxin production and can in fact decrease Stx levels in coculture (86, 87). This phenomenon was also observed in a streptomycin-treated mouse model, where coinfection with phage-resistant E. coli and stx-carrying E. coli resulted in lower fecal toxin loads than infection with stx-carrying E. coli alone (82).

The sheer abundance of bacteria in the human microbiome would suggest that interactions among E. coli strains in an O157:H7 infection are numerous. Several studies have modeled these interactions to understand the mechanisms involved. First, to estimate the frequency of stx-converting phage infection in the human body, Gamage et al. isolated E. coli strains from 37 healthy volunteers, incubated each strain with phage induced from a clinical strain of E. coli O157:H7, and found that about 10% of the strains were susceptible to the phage (82). The same group cocolonized mice with O157:H7 and either phage-susceptible or phage-resistant nonpathogenic E. coli and recovered higher levels of toxin in feces from mice with phage-susceptible strains (88). However, the presence of additional bacteria reduced O157:H7 colonization, potentially confounding Stx levels (88).

Despite these trends, generalizing from a single pathogenic strain is not possible because not all O157:H7 isolates produce the same baseline levels of Stx. Just as commensal isolates vary in their ability to amplify Stx when cocultured with E. coli O157:H7, there are diverse outcomes when different O157:H7 strains are grown with an identical commensal strain (84). In other words, Stx amplification is strain dependent with respect to both pathogenic and nonpathogenic E. coli strains. When several E. coli O157:H7 strains were grown in pairwise cocultures with the commensal strain C600, the isolate O157:H7 PA2 had the greatest increase in Stx2a production, while the O157:H7 strain Sakai showed no increase (84). Infection by the majority of stx₂-converting phages occurs via adsorption to the BamA receptor, an essential outer membrane protein (89, 90). No Stx2a amplification was observed in C600 strains expressing heterologous BamA variants that prohibit phage adsorption (85, 91). Although phage infection through BamA is perhaps the most direct cellular interaction that amplifies Stx, others exist and remain to be characterized. This observation springs from the fact that unlike C600, not all strains are lysed in coculture with O157:H7 (84), and yet some unlysed strains still amplify Stx. Xiaoli et al. (85) identified a case of Stx2a amplification independent of phage-mediated lysis in an E. coli isolate designated 1.1954. Amplification by this nonpathogenic strain did not require BamA and was not due to a secreted molecule (85). Therefore, it seems that Stx2a amplification by strain 1.1954 represents a novel mechanism of increasing toxin production through direct cell-to-cell contact.

Sources differ with respect to the significance of cellular Stx amplification in vivo. Gamage et al. (88) and Goswami et al. (84) each identified an increase in Stx levels when mice were cocolonized with Stx-producing strains and phage-susceptible strains. These studies were conducted in streptomycin-treated and germfree mice, respectively. Subsequent work with a Φstx_2dact lysogen of Citrobacter rodentium, a murine pathogen that replicates many of the features of human EHEC infection, indicates that phage infection of commensal bacteria is negligible in a complex microbiome (92). In this model, while RecA-dependent phage induction is necessary for lethal disease, the production of infectious phage particles and secondary lysogeny are in fact dispensable (92). This is consistent with the observation that in mice, phage induction directly contributes to disease caused by EDL933 (93). Consequently, triggers in the intestinal environment known to induce the stx phage, including the bacteriocins described below, may be important contributors to pathogenesis.

INDIRECT INTERACTIONS: ACELLULAR MECHANISMS OF Stx AMPLIFICATION

Cell-to-cell contact is not required for commensal bacteria to alter Stx production by E. coli O157:H7. Secreted molecules, particularly those produced to kill competitor strains, can also induce bacteriophage in their target. Bacteriocins, which are proteinaceous toxins that target closely related bacteria, are the primary example of factors in the acellular fraction that increase Stx levels in E. coli O157:H7.

Bacteriocins are produced by both Gram-positive and Gram-negative bacteria and are named according to their producing species: colicins for E. coli, pyocins for Pseudomonas aeruginosa, and pesticins for Yersinia pestis, etc. (94). They employ multiple mechanisms for release and import and have various modes of action. Bacteriocin-related genes are found in clusters or operons, which typically consist of a structural gene encoding the bacteriocin and an immunity gene that protects the producing cell from its lethal effects (95). Some bacteriocin operons harbor additional genes for the export or lysis of the host cell (96–100). Because of their ability to damage DNA or otherwise antagonize target strains, bacteriocins have been previously associated with increased Stx production of O157:H7 (85, 101, 102).

Colicins.

Colicin genes are widespread in E. coli, with an estimated 30% to 59% of isolates thought to produce at least one colicin (103–105). The molecules are between 40 and 80 kDa and are composed of three domains: the N-terminal domain is responsible for membrane translocation, the central domain is responsible for receptor binding, and the C-terminal domain is responsible for activity (95, 106). There are various modes of action that colicins use against target cells. Many are endonucleases; among them are colicin E3, an RNase that cleaves 16S rRNA (107); colicins E2, E7, E8, and E9, which cleave DNA (98, 108); and colicin D, which cleaves tRNA (109). Colicins A, B, and E1 are pore formers that destabilize the membrane (110, 111). Finally, colicin M and colicin Z act on peptidoglycan (112, 113).

Because some colicins induce the SOS response, they may also cause phage induction, cell lysis, and Stx release (85, 101, 114). Colicins with DNase function (colicins E8 and E9) were therefore tested for their effect on E. coli O157:H7. Coculture with colicinogenic strains and the addition of colicin extracts both increased the Stx production of various E. coli O157:H7 strains (101). A similar study looked at 18 different colicinogenic E. coli strains, 15 of which inhibited at least one O157:H7 strain (114). In an infection setting, DNA damage to E. coli O157:H7 would presumably cause cell lysis and the subsequent release of phage and Stx. Thus, evidence suggests that colicinogenic bacteria in the microflora could exacerbate EHEC infections (101).

Microcins.

Comparatively few studies have investigated the impact of microcins and their producing strains on Stx production in E. coli O157:H7. Microcins are smaller than colicins, and fewer have been characterized. In brief, microcins are the low-molecular-weight category of bacteriocins and can be found on either the chromosome or plasmids (115). They are separated into two classes: class I microcins are less than 5 kDa, and class II microcins are between 5 and 10 kDa (116). Unlike colicins, microcin production is not induced by DNA-damaging agents, and microcins are secreted by intact cells (117). There are three known mechanisms of action for microcins: inhibition of metabolic enzymes, inhibition of DNA replication, and impairment of a cell’s energy-generating system (118).

Although they are still understudied, microcins have a notable ability to influence the virulence of their target cells. Recent studies characterized the effects of two microcins, microcin B17 (MccB17) and the novel Mcc1229, on Stx production in E. coli O157:H7. Microcin B17 is a posttranslationally modified class I microcin; it is a DNA gyrase inhibitor that induces the SOS response because it destabilizes gyrase-dependent DNA cleavage, resulting in cell death (119, 120). When E. coli O157:H7 strain PA2 is grown with ZK1526, an MccB17-encoding strain, its Stx2a production is greater than that in monoculture (85). MccB17 was also identified in the E. coli O18:H1 strain 0.1229 and again found to increase Stx2a production in PA2 (102). However, an additional microcin in this strain also contributes to Stx2a amplification of O157:H7. The putative Mcc1229 is encoded on a 12.8-kb plasmid and was bioinformatically identified in two additional E. coli strains that also amplify Stx2a (102). The amplification phenotype is dependent on an ABC transporter and the efflux protein TolC in the producing cell and on TonB in the target cell (102). The TolC/TonB dependency of Mcc1229 could indicate that it is a class IIa microcin (116, 121). The mechanism of action of Mcc1229 is under investigation, but it appears to induce the SOS response. Overall, this work characterized two SOS-inducing microcins in a phylogroup B2 E. coli strain, both of which are able to increase Stx2a production (102).

CONCLUSION

While several E. coli strains and their metabolites are known to increase Stx2 production in E. coli O157:H7 in vitro, the broad implications for EHEC infections are not yet clear. For one thing, the expression of colicins and microcins in vivo is largely unknown, and so the frequency with which stx-inducing interactions occur in an infection remains to be seen. Furthermore, differences in induction and activity among toxin subtypes mean that E. coli O157:H7 strains are not equally susceptible to these interactions. For instance, the stx₁-converting phage is typically not amplified to the same extent as the stx₂-converting phage (122, 123). In addition to the late phage promoter pR′, stx₁ is also transcriptionally controlled by an iron-regulated promoter (124). Under high-iron conditions, the Fe-Fur complex represses stx₁ transcription; when iron is limited, the complex dissociates, and stx₁ is transcribed (125, 126). Epidemiological data overwhelmingly suggest that Stx2 is primarily correlated with severe human illness (127–131), but the distinct regulatory mechanisms of each stx-converting phage may contribute to variation in disease outcomes. MccB17- and Mcc1229-dependent induction was specifically examined in an Stx2a-encoding strain, but subtypes Stx2c and Stx2d are also relevant to human disease, and their toxin amplification should be similarly characterized. The emergence of subtypes, including Stx2e and Stx2f, and their potential to cause illness in humans (51, 132, 133) and animal reservoirs (134) merit consideration as well.

Understanding the impact of these interactions on the non-Stx virulence factors of EHEC is also essential. The LEE (locus of enterocyte effacement) is a pathogenicity island in the EHEC genome encoding a type III secretion system (T3SS), intimin (eae), and the translocated intimin receptor (tir), among other effectors (135, 136). LEE genes are positively regulated by quorum sensing (137), 1,2-propanediol (138), pyruvate (139), and many other biotic and abiotic factors. There is cross talk between LEE genes and stx-converting phage: lysogeny has been shown to either repress type III secretion through the phage cII regulator (140) or activate the T3SS through the phage transcription factor Cro (141). The T3SS and Stx also exploit some of the same host factors in order to cause infection (142), and Stx itself promotes colonization by EHEC strains (143). The regulation of pathogenicity in E. coli O157:H7 is clearly intricate and multifactorial, and future work should seek to address Stx amplification in concert with other virulence factors.

Nevertheless, the fundamental observation remains that individuals infected with the same strain of O157:H7, encoding the same virulence factors, manifest varying degrees of disease severity (144). Among other elements like age, general health, and immune status, a person’s microbiome could account for some of this variation. The population structure of E. coli in the human microbiome is often dominated by phylogroup B2 strains (5, 145–148). Relative to other phylogroups, B2 has both the highest prevalence of bacteriocin-producing strains (149, 150) and the greatest capacity for persistence in the intestinal environment (151). It is conceivable that the colicins and microcins produced by group B2 E. coli strains enable these strains to first establish a niche among resident bacteria and then act upon transient pathogenic E. coli. Bearing this in mind, efforts to link microbial community composition with health and disease will benefit from incorporating subspecies-level differences such as E. coli phylogroups and bacteriocinogeny into their models (152, 153).

Here, we conclude that E. coli intraspecies interactions contribute to EHEC infections along at least two axes, namely, phage infection and phage induction (Fig. 1). First, when nonpathogenic strains produce bacteriocins, including SOS-inducing colicins and microcins, these molecules stimulate the induction of the stx-converting phage in E. coli O157:H7. Second, when nonpathogenic strains are susceptible to stx-converting phage, they can be infected, and excess Stx can be released when cells lyse. At present, our knowledge of these interactions is limited, and there are likely additional SOS-inducing bacteriocins that await discovery. The identification of novel secreted factors and further investigation of their Stx-amplifying ability in vitro, in vivo, and at epidemiological scale are paramount to fully understanding the potential interplay between E. coli strains.

FIG 1.

Two mechanisms of Stx amplification. Bacteriocins (e.g., colicins and microcins) released by antagonistic strains can induce the SOS response in E. coli O157:H7, stimulating phage induction and thereby increasing Stx production. Susceptible bacteria can be infected by the stx-converting bacteriophage from E. coli O157:H7 and produce Shiga toxin upon cell lysis. Infection typically occurs through the BamA receptor.

Biographies

Erin M. Nawrocki majored in Biology at Allegheny College, receiving a B.S. in 2013. She then attended the University of Wisconsin—Madison, where she studied the transfer of neurotoxin-encoding plasmids in Clostridium botulinum and completed a Ph.D. in Microbiology in 2018. In 2019, she joined the Department of Food Science at Pennsylvania State University as a postdoctoral scholar. Her interests in bacterial genetics and molecular biology are reflected in her current research, which seeks to characterize mechanisms of Shiga toxin amplification in interactions between E. coli strains.

Hillary M. Mosso studied Microbiology at the University of Rochester, obtaining her B.S. in 2014. She then received her Ph.D. in Immunology and Infectious Disease from the Pennsylvania State University in 2019, as part of the Huck Institutes of the Life Sciences. Her doctoral work focused on the interaction of E. coli O157:H7 and nonpathogenic isolates capable of amplifying Shiga toxin production. Since graduating, she has transitioned into the data science field and is currently working as a Curation Scientist at Rancho BioSciences.

Edward G. Dudley received an M.S. in Food Science from the University of Wisconsin and a Ph.D. in Bacteriology from the same institution, studying the genetics and physiology of lactic acid bacteria. He next moved to the University of Maryland at Baltimore as a postdoctoral fellow, examining virulence gene regulation in the human pathogen enteroaggregative Escherichia coli. He joined the Department of Food Science at Penn State in 2007 and is currently Professor and Director of Penn State’s E. coli Reference Center. His lab investigates the biology of E. coli O157:H7 focusing on Shiga toxin production and the application of genomics to diagnostics and food safety. Despite E. coli’s position as the best-studied bacterium, he remains fascinated by its genetic malleability, permitting it to evolve into a wide range of pathogens, probiotics, and commensals.