E. coli outbreak and biofilms

Now that the outbreak has been firmly traced back to sprouts, attention is turning to how the sprouts were contaminated in the first place. The initial assumption was that this would be a classic case of contamination with E. coli via feces from cows or other livestock, however in this case it is more likely the original source was humans. Some clues to this:

  • Serotype O104:H4 has not been reported in animals, only in humans
  • Enteroagreggative E. coli (EAEC) is usually isolated from humans, not animals
  • EAEC is not always symptomatic in humans, it can be carried asymptomatically
  • There are no animals kept on the sprout farm linked to the outbreak, nor is there an obvious route by which contamination with animal feces could have occurred

So it could be somewhat similar to the situation with Salmonella Typhi, where people can become colonized but never develop typhoid fever symptoms and so don’t know they are carriers… but they are shedding the bacteria in their feces, which can cause illness in other people. Thus localised outbreaks are sometimes linked to food handlers (in EAEC as well as typhoid)… the most famous example being “Typhoid Mary”, a cook who was a typhoid carrier and spread typhoid fever to dozens, probably hundreds of people. Water contaminated with human feces can also transmit the bacteria. In this outbreak a lot of cases are linked to eating sprouts, so the logic would be that the sprouts have become contaminated. However there are reports of people attending the same function, but not eating the vegetables, also getting sick, which is consistent with secondary transmission via human carriers.

The European Food Safety Authority report summarizes the evidence for a human source of EAEC:

On 21st May 2011, Germany reported an ongoing outbreak of Shiga-toxin producing Escherichia coli- bacteria (STEC4), serotype O104:H4. […] In the past STEC O104:H4 had been isolated in humans twice in Germany in 2001 (Mellmann et al., 2008) and once in Korea in 2005 (Bae et al., 2006). In addition, according to the information reported to ECDC, a total of 10 persons were infected with STEC O104 in the EU Member States from 2004 to 2009.


The German outbreak strain seems to share virulence characteristics of STEC and EAEC strains. STEC strains usually have an animal reservoir, while EAEC have a human reservoir.


Outbreaks of diarrhoeal illness due to EAEC have been reported and linked
to the ingestion of food which was contaminated by food handlers. In
addition, it has been shown that EAEC carriage by humans is possible (Huang
et al., 2003
; Huang et al., 2006).
EAEC have rarely been identified in animals, suggesting that they are not
zoonotic, but exclusive to humans as a pathogen (Cassar et al., 2004).


Outbreaks may have more than one exposure route involved. For example, primary human infection may originate from consumption of contaminated food or direct contact with an animal carrying STEC, while secondary infection may occur by the faecal-oral route, after contamination of food through handling by an infected person shedding the bacteria. As a result, especially during the late stages of an outbreak multiple exposure routes are likely.

One of their recommendations is:

Since there is evidence of asymptomatic carriers of STEC in humans, screening of humans involved in food handling is relevant. The monitoring and/or exclusion of STEC carriers from food handling should be considered as a mitigation option.

So what can the genomes tell us?

It’s likely that the combination of Shiga-toxin production in an EAEC strain is particularly dangerous because EAEC are particularly ‘sticky’ or adhesive. The cells autoaggregate, forming biofilms, and they are also good at sticking to human cells (in fact EAEC is defined by HEp-2 cell-adherence assay). They can also form mixed biofilms with other bacteria. These abilities are probably what make EAEC good at establishing long-term colonization of humans, which can sometimes result in chronic diarrhea or long-term asymptomatic carriage. Similar properties could aid transmission via sticking to plants.

Several gene families are known to be involved in biofilm formation in E. coli, see this review.


the direct contribution of adhesive organelles of the fimbrial family to the irreversible attachment of bacteria to surfaces has been amply demonstrated. Three classes of fimbriae have a role in strengthening the bacteria-to-surface interactions: type 1 fimbriae, curli, and conjugative pili.

The outbreak genomes have swapped their aggregative adherence fimbriae relative to their closest known relatives. Both the African diarrheic strain Ec55989 (the closest related strain to have its complete genome sequenced) and the 2001 German O104:H4 strain, HUSEC O41, expressed type III fimbriae (AAF/III+). [see report for HUSEC O41, here for Ec55989]. AAF/I fimbriae are not uncommon, but they are quite different to AAF/III and may be relevant. They are plasmid borne, and the plasmid they are carried on in the outbreak strain is a bit different to those that have been sequenced previously from EAEC (see this post for details).

The outbreak genome shares with Ec55989 a fimbrial adhesion operon yehABC, however it has also acquired, adjacent to this operon, an insertion of genes yehIyehQ which are present in E. coli O157:H7 and include proteins that are probable regulators…could they regulate the fimbriae?:

It also shares with Ec55989 several other fimbrial clusters including lpf (long polar fimbriae), but I can’t find any other fimbriae-related genes in the annotation of the HPA outbreak strain that are not in Ec55989.

Conjugative pili:

most tested conjugative plasmids directly contribute, upon derepression of their conjugative function, to bacterial host capacity to form a biofilm (Ghigo 2001)

The outbreak strain contains a conjugative plasmid of the IncI type, carrying the ESBL (extended spectrum beta-lactamase) gene CTX-M. This encodes a type IV pilus, could it be contributing to the fitness of the strain by enhancing biofilm production of the host bacterium? And/or promoting horizontal transfer of DNA? This paper suggests that a type IV pilus encoded on an IncI plasmid enhances biofilm formation in E. coli, although the sequences of the type IV pili are different to the IncI plasmid in the outbreak strain.

Type V secretion system and autotransporters:

…the type V secretion pathway enables a family of proteins to reach the surface with a very limited number of accessory secretion factors because most information necessary to the translocation process is contained within the secreted protein itself. These proteins, which can therefore carry out their own transport to the outer membrane, are called autotransported or autotransporter proteins.


The flu gene encodes antigen 43 (Ag43), a major outer membrane protein found in most commensal and pathogenic E. coli. Although E. coli K-12 has only one copy of flu, most other strains of E. coli have several copies of this gene.

Ag43 is a self-recognizing surface autotransporter protein that does not seem to be involved in non-specific initial adhesion to abiotic surfaces, but rather, promotes cell-to-cell adhesion (Kjaergaard et al. 2000a). While, in liquid culture, this property leads to autoaggregation and clump formation rapidly followed by bacterial sedimentation, it also facilitates bacteria–bacteria adhesion and leads to the three-dimensional development of the biofilm (Owen et al. 1996; Henderson et al. 1997a; Hasman et al. 1999; Kjaergaard et al. 2000a; Schembri et al. 2003a). When expressed in different species, Ag43 can also be used to promote mixed biofilm formation between different bacteria, for example, between E. coli and Pseudomonas aeruginosa (Kjaergaard et al. 2000a, 2000b).

There are three Flu/Ag43 in the outbreak strain (using HPA assembly and ERA7 annotation). One is novel compared to Ec55989, and has been acquired via an integrase-mediated insertion in the same locus as the multidrug resistance genes (see post here for details). Another appears to present in the same prophage as the Shiga-toxin, although it is incomplete (662984-664255 in HPA assembly). A third is conserved in Ec55989 (EC55989_3357). AidA adhesin proteins appear to be conserved between the outbreak stain and Ec55989.

So the differences so far between the outbreak strain and Ec55989 (EAEC diarrhea), with respect to known biofilm-associated adhesins are:

  • replaced AAF/III with AAF/I (EAEC plasmid)
  • acquired novel Ag43 gene via integrase-mediated insertion
  • type IV conjugative pili in IncI plasmid
  • acquired partial Ag43 in same phage as Shiga-toxin
  • acquired cluster of genes adjacent to yehABC fimbrial cluster, yehIyehQ, which are present in E. coli O157:H7 and include regulators…maybe regulators of fimbriae?

Pic proteins

I also had a look at the pic genes in the outbreak strain. These are mucinases that have recently been shown to promote mucous secretion in the gut, responsible for mucoid diarrhea that is a classic symptom of Shigella and EAEC infection. Ec55989 contains three of these, two on the chromosome (intact, EC55989_4682, EC55989_3279) and one on the EAEC plasmid 55989p (truncated), all of which are conserved in the German outbreak strain. In addition, there is a fourth pic gene present in the EAEC plasmid of the outbreak strain (but missing from Ec55989), which seems to be intact. An NCBI blastp search turned up the same protein sequence in multiple Shigella genomes, and also the EAEC plasmid pO86A1:

>Novel pic gene from HPA assembly