OUP user menu

Fim operon variation in the emergence of Enterohemorrhagic Escherichia coli: an evolutionary and functional analysis

Nurmohammad Shaikh, Nicholas J. Holt, James R. Johnson, Phillip I. Tarr
DOI: http://dx.doi.org/10.1111/j.1574-6968.2007.00781.x 58-63 First published online: 1 August 2007


Fim operons were examined to illuminate the emergence of Escherichia coli O157:H7 from the less-virulent E. coli O55:H7. A fim invertible element deletion occurred only after O157:H7 descended from O55:H7, and after sorbitol nonfermenting O157 diverged. Type 1 pili nonexpression correlates with this deletion in all enterohemorrhagic E. coli (EHEC) tested. An N135K FimH mutation in the two most evolved O157:H7 clusters is not found in other EHEC. These data refine the evolutionary history of an emerging pathogen.

  • FimH
  • pili
  • evolution


Enterohemorrhagic Escherichia coli (EHEC) are E. coli that contain genes encoding Shiga toxins (Stxs), and which are also pathogenic to humans. EHEC usually belong to one of two groups. The best studied is the EHEC 1 group, consisting of E. coli O157:H7 and related organisms. Members of the EHEC 1 group possess the intimin-γ allele and express either the O55 or the O157 lipopolysaccharide antigen. In its descent from its less-pathogenic progenitor E. coli O55:H7, E. coli O157:H7 underwent a series of step-wise genetic changes (Feng et al., 1998). Figure 1 portrays a recent iteration of this model, taking into account bacteriophage insertion sites, and is modified from Shaikh & Tarr (2003). In this model, subgroup A consists of nontoxigenic E. coli O55:H7, subgroup B consists of sorbitol fermenting EHEC O157:H, which is a pathogenic clone found mostly in Europe (Karch & Bielaszewska, 2001), and subgroup C consists of the sorbitol-nonfermenting worldwide pathogen EHEC O157:H7. Subgroup C can be further parsed into sequentially evolved genotype clusters 1, 2, and 3 based on bacteriophage structure and insertion sites.

Figure 1

Proposed evolutionary scenario for EHEC 1 group. Boxes represent organisms known to exist today. Circle represents proposed intermediate organism between Escherichia coli O55:H7 and E. coli O157:H7. Sites of occurrence of fimA switch and FimH pocket mutations are noted. Modified from Shaikh N, Tarr PI. Escherichia coli O157:H7 Shiga toxin-encoding bacteriophages: integrations, excisions, truncations, and evolutionary implications (J Bacteriol. 2003; 185:3596-3605. Erratum in: J Bacteriol. 2003; 185:6495).

Non-EHEC 1 Stx-producing E. coli are best exemplified by members of the EHEC 2 group, which contain the intimin-β allele and express non-O157 lipopolysaccharide antigens (notably O26 and O111). EHEC 2 usually express Stx1 but not Stx2. At least one member of the EHEC 2 group, RDEC-1, does not encode any Stx. Additional EHEC are outside these phylogenetic groups, such as E. coli expressing the O103 (Tarr et al., 1996) and O121 lipopolysaccharide antigens (Tarr et al., 2002).

The E. coli fim operon encodes type 1 pili, which, among other functions, enable E. coli to colonize urinary and other epithelial surfaces. However, E. coli O157:H7 almost never express type 1 pili despite containing the fim operon (Durno et al., 1989), and the analysis of loci within the fim operon provides an opportunity to discern evolutionary and adaptive events as the EHEC 1 group emerged. Here, an analysis of type 1 pili operon variants on a recently refined map of the evolution of the EHEC 1 group (Shaikh & Tarr, 2003) is overlaid, to confirm or refute the current model of descent by elucidating the type 1 pili genotypes as these organisms emerged. The interrogated parts of the operon consist of a 16 bp deletion 5′ to fimA (Li et al., 1997) and a C→A mutation at position 467 in fimH in E. coli O157:H7, which results in replacement of an asparagine (N) by lysine (K) at amino acid residue 135 (i.e. the N135K mutation) in the expressed FimH protein (Hung et al., 2002). The expression or nonexpression of type 1 pili in genotypically diverse EHEC is also reported.

Materials and methods

Bacteria examined

As portrayed in Table 1, isolates representing E. coli O55:H7 (subgroup A); sorbitol-fermenting EHEC O157:H (subgroup B), and each of the three clusters of sorbitol nonfermenting EHEC O157:H7 (subgroup C) in the EHEC 1 group were analyzed. Members of the EHEC 2 group, which belong to several different serogroups, as well as EHEC that phylogenetically most closely resemble the EPEC 2 group, including urinary pathogen EHEC O103:H2, (Tarr et al., 1996) were also studied. EHEC O121:H19 was also studied, which has not yet been assigned phylogenetically (Tarr et al., 2002).

View this table:
Table 1

Strains studied, and fim genotypes and type 1 pili phenotypes

SerotypeStrain(s)N/Kfim switch Δ16 bpAggReferences/source
EHEC 1 group
    Subgroup A (nontoxigenic progenitors to E. coli O157)
        O55:H7TB156A, TB182A, DEC 5a, DEC 5b, DEC 5c, DEC 5d, DEC 5eNIntact+Shaikh & Tarr (2003)
    Subgroup B (sorbitol fermenting nonmotile toxin-producing E. coli O157)
        O157:HCB2755, 493/89NIntact+Shaikh & Tarr (2003)
    Subgroup C (classic pathogenic E. coli O157:H7)
        Cluster 1
            O157:H787-14, EK1, EK18NDeletedShaikh & Tarr (2003)
        Cluster 2
            O157:H786-24, 86-17, 87-07, EK15, P8KDeletedShaikh & Tarr (2003)
        Cluster 3
            O157:H70157 Sakai, EDL933, 84-01, 87-01, 87-20, 93-111, EK4, EK5KDeletedShaikh & Tarr (2003)
EHEC 2 group
    ON:NMD89-4377NIntactT.S. Whittam
TB285DNIntact+Bokete et al. (1997)
    O85:NMTB334CNIntact+Bokete et al. (1997)
    O70:H1188-817NIntact+T.S. Whittam
    O69:NM90-0106NIntact+T.S. Whittam
    O6:H28BH206C-1NIntact+T.S. Whittam
    O55:H8C96-53NIntact+T.S. Whittam
    O45:NM4309-65NIntact+T.S. Whittam
    O45:HN2566-58NIntact+T.S. Whittam
    O2r:H1BH61-3NIntact+T.S. Whittam
    O26:H2TB285ANIntact+Bokete et al. (1997)
    O26:HNTB206ANIntact+Bokete et al. (1997)
    O26:H113047-86CL-5, H30, 1557-77, 3157-74, 3360-70, 88-0493, LTO26-12, H19NIntact+T.S. Whittam
3323-61NIntactT.S. Whittam
EK29NIntact+Klein et al. (2002)
    O26:HLTO26-19, TB352ANIntact+T.S. Whittam
    O26VP30NIntact+T.S. Whittam
    O15:NMRDEC-1NIntact+T.S. Whittam
    O145:H1687-1713NIntact+T.S. Whittam
    O128:H83435-72NIntact+T.S. Whittam
    O118:H8TB216ANIntact+Bokete et al. (1997)
    O118:H16EK36, EK37NIntact+Klein et al. (2002)
    O111:NM3007-85, 2198-77,NIntact+T.S. Whittam
C412, 86-10049NIntactT.S. Whittam
    O111:H8CL-37NIntactT.S. Whittam
    O111:H78/92, TB226ANIntact+T.S. Whittam
ED-31NIntactT.S. Whittam
    Orough:H11EK39NIntact+Klein et al. (2002)
EPEC 2 group
    O103:H2UTINIntact+Tarr et al. (1996)
EK30, EK31, EK32, EK33NIntact+Klein et al. (2002)
    O103:H6TB154ANIntact+Bokete et al. (1997)
Phylogenetically indeterminate group
    O121:H19EK38NIntact+Klein et al. (2002)
  • N or K refers to the amino acid in the FimH mannose-binding pocket at position 135 of the mature FimH peptide if the nucleotide at position 467 in fimH is a C or an A, respectively.

  • Two amplicons are produced by primer pairs A and B. See text for details.

Fim polymorphisms examined

In this study, ‘fim polymorphisms’ refer to the interrogated variations in the fim promoter region as well as in the fimH structural gene, ‘switch deletion’ refer to the 16 bp deletion in the fim promoter, and ‘FimH polymorphisms’ refers specifically to amino acid variations in the predicted FimH protein. Primers (5′-ACGAGTTATTACCCTGTTTGCTG-3′ and 5′-GCCAGTAGGCACTACCACATC-3′) were used to amplify a portion of fimH that encodes the FimH mannose-binding pocket, and the resulting 535 bp amplicons were sequenced. To determine if the switch deletion is present, primers A [based on Sts-1 from Li (1997)], B [Con-2 from Li (1997)], and C [Sts-1 from Li (1997)] (Fig. 2) were used. Primers A and B produce a 952 bp amplicon if the switch is in the intact (i.e. ancestral) configuration, but no amplicon if the E. coli O157:H7 switch deletion is present. In contrast, Primers C and B produce a 936 bp amplicon if the switch is in a deleted (i.e. the E. coli O157:H7 or variant) configuration, but no amplicon if the switch is intact (Fig. 2).

Figure 2

Schematic depiction of Fim switch region in the intact and deleted configurations. Primers are the oligonucleotides used to elicit amplicons that differentiate the two configurations. Primers A (bolded nucleotides) and B (underlined nucleotides) produce an amplicon that signifies an intact fimA switch. Primers C (italicized nucleotides) and B produce an amplicon that signifies a deleted switch region. Relative to the ATG start codon of fimA, the boundaries of the invertible switch element are 119 and 433 nucleotides upstream, and there are SNPs 179 and 181 nucleotides upstream (Escherichia coli O55:H7 has an A and a T, while E. coli O157:H7 has a G and a C at these loci, respectively). The boundaries of the 16 bp deletion are 352 and 368 nucleotides upstream of the start of fimA.

Type 1 pili expression

Expression of type 1 pili was assessed by mannose-sensitive agglutination of baker's yeast in microscope slide assays, using bacterial suspensions grown overnight in static broth at 37°C (Johnson et al., 1987, 1988). Strain P678-54, and P678-54 transformed with pSH2 (a recombinant plasmid encoding type 1 fimbriae from E. coli pyelonephritis isolate J96) (gifts of Sheila and Richard Hull), served as a positive and negative control, respectively (Hull et al., 1981). Reactions were considered positive if an isolate agglutinated yeast with ≥2+ intensity (0–4+ scale), and if that agglutination intensity was reduced by ≥2 levels in the presence of 10%α-methyl-mannoside. Isolates were tested daily for 3 days during serial passage in static broth before they were considered negative for type 1 pili expression.


The FimH pocket mutation is found only in most evolved E. coli O157:H7

The N135K FimH mannose-binding pocket mutation, which could potentially attenuate urovirulence, was found only in clusters 2 and 3 of EHEC 1 subgroup C, and not in the other EHEC studied (Table 1, Fig. 1).

Switch deletion analysis correlates with Type 1 pili phenotypes

The Fim switch deletion was sought in all isolates in this study. It was found in all E. coli O157:H7 clusters, but was absent from E. coli O55:H7, sorbitol fermenting E. coli O157:H, and EHEC outside the EHEC 1 group. Type 1 pili were expressed by each of the nine EHEC 1 isolates tested that had an intact Fim switch, but by none of the 16 E. coli O157:H7 with an ‘off’ Fim switch configuration (Table 1). Each of 48 isolates outside the EHEC 1 group had an intact Fim switch, and 42 of these isolates expressed type 1 pili. It is noteworthy that of the six strains that did not express type 1 pili despite having an intact fim switch, five expressed the O111 lipopolysaccharide antigen. Results for both pairs of fimH switch region primers concurred, although two E. coli O45 isolates expressed type 1 pili but yielded two PCR products when amplified with primers A and B. One of these PCR products was of the size expected for an intact Fim switch region (0.9 kb), while the second was 2.3 kb. The latter amplicon was sequenced from one of the isolates, and contained an intra-fimA IS2 element (deposited in GenBank as accession number AY627636).


In E. coli O157:H7, a 16 bp deletion 5′ to the start of the fimA gene, in the region that determines the phase-variable expression of type 1 pili in most E. coli (Abraham et al., 1985), appears relevant to nonexpression of type 1 pili. Specifically, its repair in a laboratory strain engineered to contain the O157:H7 deletion resulted in expression of the downstream fimA, as evidenced by expression of a fimAlacZYA transcriptional fusion (Roe et al., 2001). Hung (2002) analyzed the 0157 Sakai and EDL933 strains, the two genomically sequenced E. coli O157:H7 (Hayashi et al., 2001; Perna et al., 2001), and determined that fimH, which encodes the adhesin that recognizes mannose, contains an unusual C→A mutation at position 467, which replaces an asparagine by lysine at amino acid residue 135 (i.e. N135K) in the expressed FimH peptide. They also demonstrated that replacing the asparagine with alanine or aspartate ablated the ability of FimH to bind to bladder epithelial cells, though they did not test the effect of substituting a lysine. This finding suggests that even without the switch mutation, E. coli O157:H7 type 1 pili, had they been expressed, probably would not be functional for epithelial cell adherence. This mutation in the FimH mannose-binding pocket in a subset of E. coli O157:H7 might limit the urovirulence potential of those isolates even if the switch configuration permitted the expression of FimH. However, it is notable that EHEC as a group, including those that can express type 1 pili, are almost never found among collections of urinary tract isolates (Beutin et al., 1994; Johnson et al., 2002). Perhaps they have not sustained a variety of pathoadaptive mutations in FimH that enable urovirulence (Sokurenko et al., 1999, 2004; Hommais et al., 2003; Weissman et al., 2006), and have not acquired a sufficient repertoire of additional virulence loci that augment uropathogenicity (Bahrani-Mougeot & Gunter, 2002; Johnson & Russo, 2004, 2005).

Data of this study confirm and extend the findings of Li (1997), (Iida et al., 2001; Roe et al., 2001). Like Li et al., it was found that the German sorbitol nonfermenting, nonmotile strain of E. coli O157 had an intact fim switch. Escherichia coli O157:H strain E32511 did not express type 1 pili, and has a variant fim switch, as reported by Li et al. and Iida et al. However, it should be noted that strain E32511 belongs not to the sorbitol fermenting nonmotile clone of E. coli O157, which have been found largely in Europe (Eklund et al., 2006), but rather is a member of the subset of North American sorbitol nonfermenting isolates that are nonmotile (Slutsker et al., 1997). Like Iida et al. and Roe et al., it was found that EHEC O26 and O111 express type 1 pili, and it was demonstrated additionally that a broad sample of non-O157:H7 EHEC express this phenotype.

The recent occurrence of the N135K FimH mannose-binding pocket mutation in the O157 lineage has evolutionary implications. Specifically, allelic conservation should not be inferred among less-evolved E. coli O157:H7 on the basis of analyses of the most evolved clusters within the O157:H7 serotype, because ancestral organisms (such as those belonging to cluster 1) might not have sustained mutations identified in the more evolved, sequenced organisms (cluster 3).

In summary, the 16 bp Fim switch deletion, which underlies nonexpression of type 1 pili among E. coli O157:H7, appears to have occurred after sorbitol fermenting E. coli O157:H diverged during the descent of the EHEC 1 group. This mutation is not found in other EHEC, which, unlike E. coli O157:H7, usually do express type 1 pili. The N135K mutation in the FimH mannose-binding pocket occurred after E. coli O157:H7 appeared, and cluster 1 E. coli O157:H7 have not sustained this mutation. These data support the current model of descent of EHEC 1 organisms, and refine the understanding of the sequence of events that underlie the evolution of this important group of pathogens.


Elizabeth Wolf is thanked for assistance in preparing this manuscript, Drs Thomas Whittam and David Lacher for providing sequences and helpful discussions on the evolutionary assignment of isolates in this communication, and the reviewers of this manuscript for making important comments on the text. Connie Clabots (Minneapolis Veterans Affairs Medical Center) provided expert technical assistance. The STEC Center at Michigan State University is thanked for supplying strains. The STEC Center is supported with Federal funds under NIH Research Contract # N01-AI-30058. This work was supported by NIH grants R01 AI47499 (P.I.T.) and Office of Research and Development, Medical Research Service, Department of Veterans Affairs (J.R.J.).


  • Editor: Rob Delahay


View Abstract