One’s success or failure in any endeavor depends on what is believed (e.g., our assumptions). For example, if we do not believe an organism can be found in a certain location or product, then we will not sample it or exercise appropriate responses to such findings. Consequently, when one considers that the U.S. Centers for Disease Control and Prevention (CDC) found that Shiga toxin-producing Escherichia coli (STEC) is estimated to cause more than 265,000 illnesses each year in the United States, with more than 3,600 hospitalizations and 30 deaths,[1,2] assumptions about these organisms are worth a second look.

In the past, I have heard several assumptions that can have a deleterious effect on the food industry’s ability to control pathogenic STEC should they prove to be wrong. A number of these questionable paradigms follow.

Paradigms Worth Challenging through Research and Investigation
In my experience, I have been made aware of various assumptions about STEC worth research and investigation as to their veracity. These include the following:

    “Among all the STECs, one need only worry about E. coli O157:H7 in food.”

    “All pathogenic STECs must have the eae gene.”

    “E. coli O157:H7 is an indicator of other STECs.”

    “If my assay is negative for E. coli, then E. coli O157:H7 cannot be present.”

    “E. coli O157:H7 cannot form niches in food production facilities.”

    “Other pathogenic STECs cannot form niches in food production facilities.”

“One Need Only Worry about E. coli O157:H7”
There are at least six general types of pathogenic E. coli that cause illness from foods. These include enteroinvasive (EIEC), enterotoxigenic (ETEC), and classical or typical enteropathogenic serotypes (EPEC), as described in an early review by Kornacki and Marth.[3] Pathogenic Shiga toxin-producing (EHEC), enteroaggregative (EAEC), and diffusely adherent E. coli (DAEC) were later found to be foodborne pathogens.[4] High populations of EIEC and ETEC are needed to cause illness. In this article, Shiga toxin-producing strains are called STECs. If pathogenic, they are called EHECs or “pathogenic STECs.”

In 1982, the first low-dose strain of pathogenic E. coli, an EHEC, called E. coli O157:H7, was recognized.[5] E. coli O157:H7 can attach to the colon, produce toxins, and cause intestinal lesions. This activity results in gastroenteritis and sometimes bloody diarrhea, hemolytic uremic syndrome (HUS), and brain lesions. Not surprisingly, this strain was the predominant E. coli of concern in the food industry for some years thereafter. However, recognition of the number of pathogenic STECs causing illness has continued to increase. All STECs must produce a Shiga toxin, sometimes abbreviated Stx, but formerly called SLT toxins. There are also subtypes within Stx, called Stx1, Stx2, and their subtypes (e.g., Stx2a, Stx2b). Pathogenic E. coli are typically characterized by their cell wall and pili (fimbriae) chemistries, reported as “O” and “H” antigens, respectively. These antigens are detected by the ability of antisera to specific O and H groups to cause a population of the organisms to visibly clump or “agglutinate.”

The U.S. Department of Agriculture (USDA) regards certain raw meats to be adulterated if they contain seven serogroups of these STECs, namely, serotype O157:H7, and serogroups O26, 0103, 0111, 0121, O45, and O145,[6] provided that those strains also contain the virulence genes stx and eae.

“All Pathogenic STECs Must Have the eae Gene”
The eae gene encodes a protein called intimin. Intimin, along with a protein called Tir, is responsible for attachment to the human colonic epithelium. The eae gene and the gene for production of Tir are on a region of DNA in pathogenic STECs called the “locus of enterocyte effacement (LEE).” The stx gene(s) encode Shiga toxin(s), or Stx. Thus, many of my colleagues previously believed that an E. coli strain needed to be able to produce both a Shiga toxin(s) and the eae gene to cause illness.[7,8]

This seemed a reasonable assumption, until 2011, when a severe foodborne outbreak occurred in Germany from ingestion of sprouts contaminated with E. coli O104:H4.[9,10] In this outbreak, there were 3,128 cases of acute gastroenteritis, 782 cases of HUS, a case of severe kidney disease, and 46 deaths. This organism produced a Shiga toxin (Stx2) but did not have the eae gene for attachment.

The German strain seemed unique in terms of virulence factors. Rather than possessing a copy of an adhesin gene (eae) and the STEC autoagglutination adhesin gene (saa), the strain possessed both strong adherence fimbriae (AAF/I) and a very strong enteroaggregative factor known as IrgA homologue adhesin (iha). These factors allowed the strain to effectively adhere to the colonic epithelium and, combined with several other factors, such as a high level of antibiotic resistance, resulted in a remarkably virulent strain.[11] Atypical EHEC strains (such as the 2011 German outbreak strain) that lack LEE are more often found in adult cases of HUS. These strains rely on some other virulence factors for their pathogenicity.[12] Hence, it is not necessary for an EHEC to have the toxin and the eae gene. The presence of both the toxin and an attachment factor appears to be required.

Bielaszewska et al.[13] stated that, “Shiga toxin (Stx)-producing Escherichia coli (STEC) strains of serogroup O91 are the most common human pathogenic eae-negative STEC strains.” These strains have been reported to be the fourth most common STEC group isolated in Germany, behind O157, O26, and O103, which cause disease, mostly among children.

E. coli O91 serovars producing the Stx2 toxin have been isolated from patients suffering from HUS, including a fatal case involving a comorbid Clostridium difficile infection.[14] It should be noted that Stx2 toxins are highly associated with HUS; hence, these are more dangerous than Stx1. For example, Stx2 is 1,000 times more lethal to human kidney macrovascular endothelial cells than Stx1.[6] The most common subtypes of the stx2 gene associated with hemorrhagic colitis and HUS are stx2a, stx2c, and stx2dact (more commonly referred to as stx2d). Again, we see with these O91 strains that eae is not necessary for pathogenicity. Nevertheless, Paton[15] stated, “The capacity of Shiga toxigenic Escherichia coli (STEC) to adhere to the intestinal mucosa undoubtedly contributes to the pathogenesis of human disease.” Hence, attachment, including by means other than eae, is also needed for pathogenicity and occurs with the O91 serotype, which has been isolated from patients suffering from a range of symptoms, including HUS and bloody diarrhea.[13] Certain O91:H8 and O91:H21 strains are also known to possess the autoagglutination adhesin gene saa, which Feng et al.[16] stated may be related to the pathogenicity of these strains. It is noteworthy that a rare strain of E. coli O91:H9 was shown to have IrgA and several strong adherence factors (saa, sab) and a strong fimbrial adhesin gene (lfpA₀₂₆) as well as stx1a, stx2a, and stx2d. In addition, it had the gene for production of cytolethal distending toxin (cdt-V), which is lethal to human microvascular endothelial cells,[16] suggesting, in my view, that this strain may also have high pathogenicity.

Therefore, the literature supports a conclusion that the presence of an stx gene and a few adhesive factors is enough to identify a dangerous strain of E. coli. This view fits the general paradigm that USDA used when it indicated that among the top seven isolates, any possessing the Shiga toxin gene (stx1 or stx2) and the eae gene would be regarded as adulterants. USDA may have assumed that only eae could be used by an organism to attach to the colon, but it was also only looking for the most common serotypes associated with illness in the products that the agency regulates. It is important to note that in a 2017 study, an O91:H10 serogroup strain (CB12624) was able to cause HUS in a human, despite lacking copies of both the lpfA₀₂₆ fimbrial gene and the saa autoagglutination adhesin gene, but it was positive for the iha adhesin factor gene,[16] as is the O91:H9 strain mentioned above. Thus, current research supports a view that for an STEC to be clinically pathogenic, it requires the following: an stx1 or stx2 gene, or both, as well as a strong adhesive factor.  

“E. coli O157:H7 Is an Indicator of Other STECs; if My Assay Is Negative for E. coli, Then O157:H7 Cannot Be Present”
The concept of indicators is fraught with confusion, as the term may mean several things (e.g., index organism group, microbial indicators of sanitation efficacy, hygienic conditions, wholesomeness, or spoilage potential). Sometimes, even surrogate microorganisms that may mimic the behavior of a pathogen subjected to a treatment have been called indicators.[17,18] However, I am unaware of a situation wherein one strain, serotype, or subtype of an organism is used to represent a broader group of organisms, except in the case of surrogate microorganisms. In the case of indicator organisms, one uses a larger group to represent a smaller group of organisms.

Furthermore, E. coli O157:H7 is physiologically divergent from generic E. coli, as it lacks the β-D-glucuronidase enzyme used for the MUG (4-methylumbelliferyl-beta-D-glucuronide) reaction used in some media for identification of E. coli and is not expected to grow at elevated temperatures associated with classical MPN (most probable number) methods.[18] If E. coli O157:H7 is not found in a sample, it does not mean that other E. coli including STEC cannot be present. It is important to note that CDC found 137 different STECs that caused illness in a 10-year period in the United States.[19] What basis is there to assume that other STECs would behave in this unusual manner? It should be pointed out that as of this writing, the current U.S. Food and Drug Administration (FDA) method screens isolates for the top 10 STECs plus all others that can produce Shiga toxin and have two major attachment factors, that is, AggR and eae.[20]

“E. coli O157:H7 Cannot Form Growth Niches in Food Production Facilities”
If we assume that these organisms cannot grow in a food processing facility and thus move from growth niche to growth niche, then assumptions of direct fecal-oral transmission would follow.

The perception that E. coli O157:H7 does not form growth niches in food production facilities may be based upon several factors:

1.    In laboratory studies, some STEC strains have been shown to be poor biofilm formers in monoclonal biofilms.[21] However, E. coli O157 biofilms have been formed in the presence of multispecies biofilms,[22] and in one study, biofilm formation by E. coli O157:H7 was stimulated by the presence of Acinetobacter calcoaceticus.[23] In nature and in food processing plants, all biofilms would be expected to be multispecies. Furthermore, E. coli O157 have been shown to attach to industrial surfaces and have formed biofilms in the laboratory.[21,22,24,25]They have also been recovered from industrial food contact surfaces after cleaning and before operations, in a recent experience I have had in a food production plant, and also in the literature.[26]

2.    Effective methods for cultural recovery of E. coli O157 for confirmation have lagged molecular approaches for detecting signals consistent with STECs, in my experience. Hence, the presence of so-called false-positive results may reveal higher-than-reported incidence as determined by cultural confirmations.

3.    The likely false perception that these organisms do not form biofilms leads to a conclusion that the only source is the fecal matter of infected animals. In my view, this assumption can lead to overly simplistic approaches to contamination investigations.

“Other Pathogenic STECs Cannot Form Growth Niches in Food Processing Facilities”
Even if E. coli O157:H7 cannot form growth niches in food processing facilities, does that mean other STECs cannot? Many times, in my experience in food processing facilities, sample testing resulted in a finding of generic E. coli, independent of the likelihood of direct fecal contamination.[27] These tests were based on typical phenotypic (physiological) approaches. Is it more likely that the 136 pathogenic non-O157:H7 STEC strains found in the United States would behave physiologically like generic E. coli or like the physiologically unusual E. coli O157:H7?

Vogeleer[22] wrote, “Several studies demonstrated that STEC can persist as a biofilm on fresh produce, in water, and in processing plants. In some cases, factors contributing to biofilm formation have been identified.”

Conclusions
STECs can be in production facilities as potential sources of product contamination. Paradigms that attribute STEC contamination of food to only fecal matter (usually from contaminated cattle or grain directly contaminated with animal feces) are likely to be wrong.

Food production companies that have a quantitative specification for allowable levels of generic E. coli should be aware that these organisms could be STECs and take appropriate actions and adjust their specification appropriately. If E. coli is isolated from ready-to-eat foods, it should be tested for pathogenic potential.

Companies should evaluate whether they should add generic E. coli or STECs to their environmental monitoring programs. This approach should be especially considered by companies handling raw agricultural commodities. Those that find STECs should seek to eliminate them from their environments.

Ready-to-eat, raw agricultural products, in which generic E. coli is likely to be found fairly frequently, should be considered for screening for pathogenic strains in accordance with FDA’s current Bacteriological Analytical Manual.[28]   
 

Jeffrey L. Kornacki, Ph.D., is president and senior technical director of Kornacki Microbiology Solutions Inc. He is on the editorial advisory board of Food Safety Magazine.