INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Attaching and effacing Escherichia coli (AEEC) induce a unique intestinal histopathological phenotype termed attaching and effacing (A/E), which is characterized by intimate bacterial attachment to the epithelial cells with effacement of microvilli and rearrangement of the host cell cytoskeleton to form a pedestal-like structure that cups individual bacterial cells (35). The A/E phenotype is encoded by the locus of enterocyte effacement (LEE) pathogenicity island (PAI) (32). The LEE in human enteropathogenic E. coli (EPEC) strain E2348/69 is a 36-kb cluster of genes containing five major polycistronic operons: LEE1, LEE2, LEE3, tir, and LEE4 (14). The LEE1, LEE2, and LEE3 operons contain genes which encode components of the type III secretion apparatus (esc and sep genes) and the ler (LEE-encoded regulator) gene. Ler is a positive regulator for the genes located inside the LEE and also regulates a number of genes located outside the LEE (16). The tir operon encodes intimin, Tir, and CesT (the chaperone for Tir) (1, 15). Intimin is a bacterial outer membrane adhesin involved in the intimate attachment of bacteria to host epithelial cells (22). Tir is translocated from the bacteria to the host epithelial cells, where it serves as a receptor for intimin (27). The LEE4 operon encodes the secreted proteins EspA, EspD, and EspB, which are involved in delivering Tir and other proteins to host cells (26, 29). LEE4 also encodes the type III secreted effector protein EspF (33), which is translocated into host cells and alters the tight-junction permeability of epithelial cells (G. Hecht, B. McNara, A. Koutsouris, and M. S. Donnenberg, Abstr. Dig. Dis. Week and 101st Annu. Meet. Am. Gastroenterol. Assoc., abstr. 2370, 2000).

The LEE is also found in a diverse range of A/E pathogens with differing host specificity and evolutionary history. These include Citrobacter rodentium, associated with colonic hyperplasia in mice; certain diarrheagenic strains of Hafnia alvei, now classified as Escherichia (20, 23); and diverse E. coli strains associated with diarrhea and other enteric infections in rabbits, pigs, calves, and dogs (8, 10, 21, 48). Among human diarrheal AEEC isolates, multilocus enzyme electrophoresis has identified four phylogenetic groups which reflect divergent evolutionary histories and follow biotype and serotype (46). Strains within each group also have different pathogenic properties and different virulence factors. EPEC strains colonize the small intestine and cause severe infant diarrhea, and they have been divided into two groups, EPEC 1 and EPEC 2. EPEC 1 strains include the well-characterized strain E2348/69 (O127:H6), as well as isolates of serotypes O55:H6, O142, and O86. The EPEC 2 group, which is more commonly isolated in developing nations, includes the well-characterized strain B171 (O111:H2), as well as other isolates of O111:H2, O126:H2, and O128:H2. Enterohemorrhagic E. coli (EHEC) strains differ from EPEC strains in that they colonize the large intestine and cause diarrhea, bloody diarrhea, colitis, and hemolytic-uremic syndrome due to the production of Shiga toxin (Stx) in these strains. The EHEC 1 group includes the O157:H7 and O55:H7 strains, while EHEC 2 includes the O26:H11 and O111:H8 serotypes. Some AEEC strains which lack Stx and are therefore EPEC are more closely grouped with EHEC than with EPEC. For example, O55:H7 EPEC strains are related to EHEC 1, while the rabbit pathogen RDEC-1 is closely related to O26:H11 strains of EHEC 2 (46).

Electrophoretic grouping is also reflected in differences in the LEE. Among electrophoretic groups, there can be significant differences in the site of LEE insertion in the chromosome. In strains of the EPEC 1 group, the LEE is inserted at a selenocystyl (selC) tRNA gene at min 82 on the E. coli K-12 chromosome (31, 43, 46). In the related EHEC 1 group, the LEE is also inserted into the selC locus, but the LEEs from these strains contain 13 additional open reading frames (ORFs) within a putative P4 family prophage (40, 43, 47). LEEs of some strains from the EHEC 2 and EPEC 2 groups are inserted at the pheU site at min 94 of the K-12 chromosome, while other strains are inserted at a third site, as yet unknown (43). Different insertion sites are believed to represent distinct evolutionary events.

Within the LEE, there is also significant divergence among related genes. This was first observed in comparing EPEC 1 strain E2348/69 (14) and EHEC 1 strain EDL933 (40). Each LEE had a similar genetic organization, and genes encoding the Esc, Sep and Ces proteins, which are involved in assembling the type III machinery, were highly conserved. In contrast, more divergence was observed in proteins (and protein motifs) involved in host-pathogen interactions, such as intimin, Esp, and Tir, as well as in a number of genes with cryptic function (40). This variation has formed the basis for several typing schemes. Attempts have been made to classify intimin by means of sequence homology (3, 38), antigenic relatedness (2, 4), and PCR amplification (3, 38, 41, 42). There are at least five genetically and serologically distinct intimin types that generally correspond to electrophoretic groupings. A recent report extended a typing scheme to Tir, EspA, and EspB and demonstrated that A/E strains in different groups and from different animal isolates produce divergent intimin, Tir, and Esp proteins and that these were linked to pathogenesis in cattle (11).

Comparative analyses of complete LEE sequences have been limited since full sequence data are known only for the LEEs from E2348/69 (EPEC 1) and EDL933 (EHEC 1). These two groups are relatively close in evolutionary terms, and both are predominantly human pathogens, although EHEC 1 strains also colonize cattle. Relatively less information is available for animal AEEC. To better understand the contributions of different evolutionary lineage and host specificity, we sequenced the LEE from a divergent pathogen. RDEC-1 (O15:H) is a rabbit enteropathogenic E. coli (REPEC) strain that produces characteristic A/E lesions in rabbit intestine and appears to be restricted to that host (10). RDEC-1 is most closely related to EHEC 2, especially the O26:H11 strains and to a lesser extent the O111:H8 strains, although RDEC-1 lacks the Stx-producing phage (46). We have previously described the construction and characterization of a plasmid (pDK5) containing a SmaI fragment from RDEC-1 which contains the LEE sequences and confers the A/E phenotype to E. coli K-12 in vitro (25). In this report, we describe the sequence of the RDEC-1 LEE and compare it with those of the EPEC 1 and EHEC 1 LEEs as well as with sequences of individual genes from a variety of A/E animal pathogens.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Cloning, sequencing, and analysis of the LEE. Plasmid and cosmid DNA for direct sequencing and for the PCR template was prepared using the QIAquick Miniprep kit (Qiagen). The nucleotide sequence of pDK5 was determined by the Taq Dye-terminator methods using an automated 373A DNA sequencer (Applied Biosystems). Electropherograms were analyzed and edited by using Sequencher 4.0 (Gene Codes Corp., Inc., Ann Arbor, Mich.). Comparisons of nucleotide and amino acid sequences were performed with BLAST programs offered by the National Center for Biotechnology Information (NCBI) (National Institutes of Health, Bethesda, Md.).

PCR. PCR to determine the insertion site of the LEE in RDEC-1 was performed as previously described (44). For the selC gene, chromosomal DNA template prepared from bacterial cells was amplified with primers K261 (5'-CCTGCAAATAAACACGGCGCAT) and K260 (5'-GAGCGAATATTCCGATATCTGGTT), giving a 402-bp amplification product for the intact selC gene. Primers K913 (5'-CATCGGCTGGCGGAAGATAT) and K914 (5'-CGCTTAAATCGTGGCGTC) were used to amplify the pheU gene, giving a 300-bp amplification product for the intact pheU gene. Two pair of primers, K260 plus K255 (K255, 5'-CGTTGAGTCGATTGATCTCTGG) and K296 plus K295 (K296, CATTCTGAAACAAAC TGCTC; K295, 5'-CGCCGATTTTTCTTAGCCCA), for amplifying the right and left junctions for the EPEC E2348/69 LEE, respectively, were used to further analyze both ends of the RDEC-1 LEE. These two pairs of primers both generate 418-bp products from strain E2348/69 (32). For PCR amplification, SUPERMIX high-fidelity mixture (Gibco BRL, Rockville, Md.) was mixed with template DNA (bacterial suspension in distilled water; 94°C for 10 min) and the respective pair of primers, while amplification was performed on a Peltier thermal cycler (MJ Research, Waltham, Mass.). PCR amplification products were separated on a 1.2% agarose gel and analyzed.

Nucleotide sequence accession number. The nucleotide sequence for the RDEC-1 LEE region and flanking sequences determined in this study has been assigned GenBank accession number AF200363.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

In this report we describe the RDEC-1 LEE region, the first LEE sequenced from an AEEC of animal origin and the first sequenced from the EHEC 2 evolutionary group. This enables us to examine the structure of a divergent LEE and to compare genes and proteins from AEEC strains of differing host specificity and evolutionary history.

Defining the RDEC-1 LEE. The physical map resulting from the complete nucleotide sequence of the 37,889-bp SmaI fragment of RDEC-1 in pDK5 is shown in Fig. 1. The G+C content of this 38-kb region is 41.3%, which is well below the E. coli K-12 average (50.8%) and is similar to that of the LEEs of E2348/69 (38.3%) (14) and EDL933 (41.2%) (40). This 38-kb DNA region contains the complete LEE of RDEC-1. Because it is difficult to determine the boundaries of the RDEC-1 LEE, we define an internal 34-kb (nucleotides [nt] 1032 to 35232) "core region" that is highly homologous to the nt-510-to-34592 region of E2348/69 and the nt-43749-to-9556 region of EDL933. The overall homology at the nucleotide level of this 34-kb core region between RDEC-1 and E2348/69 or EDL933 is 89.3%, compared with 92.2% homology between E2348/69 and EDL933. The 34-kb core region of the RDEC-1 LEE contains 40 ORFs, all of which correspond to the genes located on the LEEs from E2348/69 and EDL933 (14, 40). It is noteworthy that the LEE core regions of RDEC-1, EPEC (E2348/69), and EHEC (EDL933) are of virtually the same size and genetic organization, indicating that they have a common ancestor.

View larger version (25K): [in this window] [in a new window]   FIG. 1.   (A) Physical map of the RDEC-1 LEE. The orientation of each individual gene is indicated by the direction of the arrow. (B) Amino acid identity of conserved genes between RDEC-1 and E2348/69 (GenBank accession no. AF022236) and between RDEC-1 and EDL933 (GenBank accession no. AF071034). Arrows under the bars indicate the major polycistronic operons corresponding to those in panel A.

View larger version (24K): [in this window] [in a new window]   FIG. 2.   Schematic representation of homologous and unique sequences in the LEE of RDEC-1, E2348/69, EDL933, and the lifA gene of E2348/69 (GenBank accession no. AJ133705). Homologous sequences are indicated by boxes with the same pattern. The dashed box indicates the 34-kb conserved core region of LEE. The flanking regions of the LEE are divergent among RDEC-1, E2348/69, and EDL933.

Overview of the structure of the RDEC-1 LEE core region. We have identified a total of 40 potential ORFs in the RDEC-1 LEE core region, all of which are found in the previously described LEEs of EPEC (E2348/69) and EHEC (EDL933) (14, 40). The RDEC-1 LEE genes are oriented and annotated according to the LEE of E2348/69 (14), starting with rorf1 and continuing through to espF. The arrangement of the RDEC-1 ORFs is also identical to that of the ORFs in EPEC, and so we predict the presence of at least five major polycistronic operons within the RDEC-1 LEE based on the experimentally determined operon structure of the EPEC LEE (Fig. 1).

Promoter regions and regulation. The promoter regions for several LEE operons in E2348/69 have been experimentally determined (34, 44), and we compared the known promoter regions of E2348/69 with regions upstream of the RDEC-1 operons. We found that the RDEC-1, EPEC, and EHEC LEEs have very similar predicted promoters, with similar spacing from the initiation codon (Table 2), and that the LEE2, LEE3, tir, and LEE4 promotors and putative regulatory regions are highly conserved. Since both LEE2 and LEE3 are directly regulated by Ler (34), which is highly conserved among AEEC strains, a conserved binding region would be expected. In contrast, examination of the DNA surrounding the predicted promoters revealed significant differences in LEE1 and tir. Upstream of the LEE1 promoter, there is substantial divergence, and the ERIC sequence is entirely lacking in the RDEC-1 LEE, as noted above. These findings may indicate fundamental differences in the regulation of ler, which is itself a key event in pathogenesis.

Type III secretion components (Esc, Sep, and Ces). The proteins that constitute the type III secretion system include Esc proteins, which are conserved among Yersinia species and other pathogens containing the type III secretion system; Sep proteins, which are more often LEE specific or may be found in the SPI-2 island of Salmonella; and the Ces secretion chaperones. In general, proteins necessary for type III secretion, including Esc, Sep, and Ces proteins, are conserved among strains, reflecting conservation of function and a lack of strong evolutionary pressure on cytoplasmic proteins. The Ces proteins are highly conserved (96% identical) among the three LEEs but are unique to the LEE and are not not found in other type III secretion systems. The Esc proteins and SepL also diverge by less than 5% and so are highly conserved among RDEC-1, E2348/69, and EDL933. Among the remaining Sep proteins, SepQ is 86% identical among the three strains but SepZ is hypervariable and is among the most divergent proteins encoded by the LEE (Table 1). This has been previously reported on the basis of a comparison of EPEC and EHEC (40) and of other SepZ entries in GenBank (S. J. Elliott and J. B. Kaper, unpublished data). The function of SepZ is unknown, and the significance of this observation is not clear.

Tir operon. The Tir operon encodes three proteins that are both genetically and functionally linked: Tir, CesT, and intimin. Tir varies in length from 538 amino acids (aa) for RDEC-1 to 550 aa for E2348/69 and 558 aa for EDL933. Both N-terminal and C-terminal regions of Tir demonstrated significant heterogeneity between RDEC-1 and the two human strains. These domains are located within the host cell and have host cell-signaling functions (27). In contrast, the central 107-aa region of Tir (aa 245 to 351) is more conserved and has been demonstrated to contain the intimin-binding domain of Tir (18). Overall, Tir is the second most divergent LEE-encoded factor, with only 66% identity to Tir from E2348/69 and EDL933.

Esp operon. Half of the genes on the LEE4 operon are conserved between RDEC-1 and other AEEC, including sepL and escF, which encode components of the secretion system, and two genes of unknown function, orf27 and orf29. In striking contrast, the genes encoding the secreted proteins EspADBF in the LEE4 operon show up to 30% divergence between RDEC-1 and E2348/69 or EDL933 (Fig. 1). A more complete analysis of the heterogeneity of the esp genes is discussed below and in Table 1.

EspF. The most variable protein encoded by the LEE is the secreted effector protein EspF (Fig. 3), which differs in both amino acid sequence and in the size of each protein. RDEC-1 EspF is 160 aa, while E2348/69 EspF is 206 aa and EDL933 EspF is 248 aa. Alignment of the different EspF sequences demonstrates the existence of an absolutely conserved proline-rich APPPPT motif that is part of a larger (35-aa) motif (Fig. 3A) repeated within the EspF sequence and also conserved among EspF proteins from different isolates. The number of the larger repeats differs among the EspF proteins, from four repeats in EHEC to three in EPEC and two in RDEC-1, resulting in great differences in the lengths of EspF among strains. Interestingly, EspF from REPEC strain B10 (O103:H) is identical to that of RDEC-1, except that the B10 EspF contains an additional 47-aa stretch (between aa 116 to 164 of B10 EspF). This may suggests that B10 diverged from RDEC-1 after deletion (or addition) of a repeat unit (Fig. 3B). The EspF of RDEC-1 is 71% identical to the EspF of another EPEC 2 strain of the O111 serogroup. The latter strain contains an additional 12-aa stretch in the EspF protein. We do not yet know the function of these repeats, although they have been observed to contain a proline-rich SH3 binding domain (13, 14, 33), implying that EspF interacts with host signaling molecules. A recent study demonstrated that EspF of EPEC is required for alteration of host intestinal epithelial barrier function (Hecht et al., Abstr. Dig. Dis. Week). We have demonstrated that EspF proteins from different isolates vary in size and that the number of repeat regions ranges from four in EHEC to two in RDEC-1. The presence of different numbers of repeats, as well as differences within each repeat, suggests different signaling functions, and such an ability could be consistent with differences in pathogenesis and host specificity among these AEEC strains.

View larger version (42K): [in this window] [in a new window]   FIG. 3.   (A) Alignment of EspF of EHEC (O157:H7, strain EDL933), EPEC (O126:H7, strain E2348/69), REPEC O103:H2 (strain B10, GenBank accession no. AF116900), and RDEC-1, showing repeated regions (repeats 1, 2, 3, and 4). A 31- to 35-aa conserved region is indicated in bold. Identical nucleotides are indicated by dots. Dashes indicate absent nucleotides. Numbers on the right side of the sequence indicate the amino acid number. (B) Comparison of EspF of RDEC-1 and B10. Identical amino acids are indicated by shaded bars. Numbers above the box indicate the positions of the amino acids. EspF of RDEC-1 is 47 aa (position indicated by the hatched box) shorter than EspF of B10. The sequence of the conserved 35-aa region is indicated in bold.

Other ORFs. The proteins rOrf1 and rOrf2 of RDEC have moderate homology (83 to 87%) to those of E2348/69 and EDL933. This level of divergence is intermediate between that of the highly conserved intracellular components of the secretion system and that of the intimin, Esp, and Tir proteins, which are all highly divergent, are exposed on the surface of the cell or secreted into the medium, and interact with the host. This suggests that the rOrf1 and rOrf2 may be surface exposed and/or interact with the host. rOrf1 is predicted to be an outer membrane protein with homology to an outer membrane protein from Salmonella (14), and rOrf2 has recently been demonstrated to be a type III secreted protein that is able to interact with host cells (16a).

Comparison of highly variable RDEC-1 proteins with those from other AEEC strains. The LEE clearly contains a number of highly divergent genes and predicted protein products. In general, the highly variable proteins encoded by the LEE are those that are exposed on the surface of the cell and/or directly interact with the host. These include Tir, intimin, the Esp proteins, and the cryptic proteins SepZ and Orf19. Greater divergence is thought to reflect the greater evolutionary pressure on these proteins both from the host immune system, which selects for variation in surface-exposed proteins, and from differences among hosts, which select for differences in the functional domains. Such divergence has been used to classify AEEC isolates and to study differences associated with different host species, pathogenicities, and evolutionary lineages. In this context, we examined the relationship of highly variable RDEC-1 LEE proteins with those from isolates from different animal species and of different lineages, using previously reported sequences (Table 1).