INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Enterotoxigenic Escherichia coli (ETEC) strains are a major cause of secretory diarrhea in both humans and animals (4, 19). The basic pathogenic repertoire of ETEC includes species-specific surface adhesins that promote small intestinal colonization and enterotoxins that stimulate intestinal cell secretion (13, 24). These virulence determinants are often encoded on transmissible genetic elements (23, 40). Two types of enterotoxins are elaborated by ETEC, a heat-labile (LT) and a heat-stable (ST) toxin. STI is the more predominant of two different ETEC heat-stable enterotoxins (21) and causes fluid secretion by activating membrane-bound guanylyl cyclase C (38).

Enteroaggregative E. coli (EAEC) heat-stable enterotoxin 1 (EAST1) is a genetically distinct toxin that is structurally related to STI and also elevates intestinal cGMP levels (35, 36). Little is known about the pathogenic significance of EAST1 in diarrhea. In one case-control study, E. coli isolates that were genotypically positive for EAST1 were highly associated with diarrhea in Spanish children (43). It was notable that very few of the EAST1 E. coli isolates from this study hybridized with an aggregative adherence DNA probe. Three separate outbreaks of diarrhea linked to EAST1 E. coli have been reported. The index strain in a Minnesota outbreak was an O39:NM E. coli that expressed EAST1 and had the enteropathogenic E. coli gene locus for enterocyte effacement (17). In one Japanese report, an O nontypeable:H10 EAEC with a plasmid-borne EAST1 gene was associated with an extensive school outbreak of severe diarrhea (20). In a second Japanese outbreak, the implicated pathogen was an O166 E. coli that had no other identifiable E. coli virulence genes (31). In a small volunteer challenge study investigating the pathogenicity of EAEC, one of two EAST1-positive EAEC strains caused diarrhea (30). These reports suggest the potential importance of EAST1 and indicate that it may be found in association with different virulence factors.

The gene, astA that encodes EAST1 has been found in ETEC of both human and animal origin (37, 45, 46). The nucleotide sequence of a plasmid allele of astA in the prototype ETEC strain H10407 was found to be nearly identical to that of EAEC strain 17-2, although expression of enterotoxic activity was not reported (36, 45). That multiple genomic fragments of H10407 hybridized with an astA probe prompted speculation that the gene may be carried on a transposon (45). In this study, we isolated and characterized the EAST1 coding region from a virulence plasmid in human-derived ETEC strain 27D. We present evidence that this astA allele indeed expresses enterotoxic activity. Moreover, we show that it is located on an insertion sequence (IS) element lying entirely within a transposase-like gene.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

The experiments reported herein were conducted according to the principles set forth in Guide for the Care and Use of Laboratory Animals (6a).

Bacterial strains, plasmids and growth conditions. The wild-type ETEC and derivative strains, host E. coli, and plasmid vectors used in this study are described in Table 1. E. coli MC4100LAC was made by conjugation of E. coli HfrH (1) with E. coli MC4100 (6), with selection for lactose fermentation, and was kindly provided by Colin Manoil. E. coli strains were routinely grown at 37°C on Luria-Bertani medium (33) except where noted otherwise. Growth medium was supplemented with ampicillin (50 to 100 µg ml¹) when appropriate for selection.

DNA preparation and Southern blot hybridizations. Plasmid DNA was isolated as described by Birnboim and Doly (5) or using plasmid extraction kits (Qiagen, Chatsworth, Calif.). Total genomic DNA was isolated by sodium dodecyl sulfate (SDS) lysis and subjected to proteinase K digestion, phenol-chloroform extraction, and ethanol precipitation. Restriction endonucleases and other DNA-modifying enzymes (Boehringer Mannheim, Indianapolis, Ind.) were used according to the manufacturer's instructions. Transformations were carried out by the procedure of Mandel and Higa (26), and electroporation was performed using a Bio-Rad Gene Pulser (Bio-Rad Laboratories, Hercules, Calif.) with electrocompetent cells prepared according to the manufacturer's protocol.

Construction of an astA mutant. A point mutation in astA was made by altering codon 16 from TCG (serine) to TAG (amber), using a QuickChange mutagenesis kit (Stratagene) and mutagenic primers. Plasmid pAM2 served as the template, and the mutant plasmid was designated pAM13. Presence of the mutation was confirmed by nucleotide sequence analysis of the astA gene.

Sequencing of astA coding region and analysis of deduced products. The DNA sequence of the astA coding region was determined by the dideoxynucleotide chain termination method (34) on double-stranded pAM1 plasmid DNA template by using TaqI cycle sequencing kits (Applied Biosystems) in a thermal cycler (Perkin-Elmer/Cetus). Both strands of DNA were sequenced using custom oligonucleotide primers that were synthesized on an ABI model 392 automated DNA-RNA synthesizer. Automated sequencing was performed with an ABI model 373 DNA sequencer.

IS designation. The Plasmid Reference Center Prefix Registry (E. M. Lederberg, Stanford University School of Medicine, Stanford, Calif.) made the IS1414 assignment.

Transposition experiments. To detect transposition of IS1414, plasmid pAM2 was transformed into E. coli strain MC4100LAC and grown on tetrazolium agar with ampicillin. Lactose-nonfermenting colonies, identified by red coloration, were isolated and analyzed for insertions into lacZ-lacY by PCR analysis with two pairs of overlapping primers used in separate reactions. Primers SJ7 (5'-GCGTACTGTGAGCCAGATGCCCGGC-3') and SJ8 (5'-GGTTTCCCGACTGGAAAGCGGGC-3') amplify a 2.25-kb region that encodes the 5' end of lacZ. Primers SJ5 (5'-GCGCCGCATCCGACATTGATTGC-3') and SJ6 (5'-CCTGAACTACCGCAGCCGGAGAGC-3') amplify a 2.35-kb region encoding the 3' region of lacZ and all of lacY. For mutants with a lac insertion in the size range of IS1414, the lacZ-lacY amplicon containing the insertion was hybridized in Southern blots with an oligonucleotide probe for IS1414, SJ10 (5'-CTGGCAATCCAGTCTGCG-3).

In vitro transcription-translation. In vitro transcription-translation was performed with the E. coli S30 extract system as instructed by the manufacturer (Promega, Madison, Wis.). Equivalent amounts of template DNA were used in each reaction. [³⁵S]methionine (NEN Dupont, Boston, Mass.) was used for labeling in each reaction. Proteins were analyzed on a sodium dodecyl sulfate-10% polyacrylamide gel. After drying, the gel was exposed to X-Omat AR film overnight at 70°C.

Enterotoxin assays. The Ussing chamber enterotoxin assay was performed using rabbit ileal tissue mounted in Lucite Ussing chambers (1.12-cm² aperture) as previously described (37). Adult New Zealand White rabbits (2 to 3 kg) were sacrificed by cervical dislocation. Potential difference was measured, and short-circuit current (I_sc) and tissue conductance were calculated. A rise in I_sc unaccompanied by a change in tissue conductance indicates an anionic secretory effect.

Nucleotide sequence accession number. The IS1414 nucleotide sequence is deposited in the GenBank DNA sequence data library under accession number AF143819.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Copy number of astA in ETEC strains. It was previously reported that prototype ETEC strain H10407 carries multiple genomic copies of astA, including one on its 92-kb virulence plasmid (29, 45). We examined three additional astA-positive ETEC strains to determine if the number of toxin gene copies varied from strain to strain. Similar to H10407, strain 27D had a single plasmid and multiple chromosomal restriction fragments that hybridized with the astA probe (Fig. 1). For the other two strains, several plasmid and chromosomal bands hybridized with the probe.

View larger version (76K): [in this window] [in a new window]   FIG. 1.   Southern analysis of plasmid DNA (odd-numbered lanes) and total cellular DNA (even-numbered lanes) of representative ETEC strains digested with PstI and hybridized with a 32P-labeled astA probe. Lanes: 1 and 2, strain 27D; 3 and 4, H10407; 5 and 6, H10407P; 7 and 8, strain 83/7; 9 and 10, strain 12-1; 11, negative control HB101 (total cellular DNA). The relative positions of DNA standards are shown in kilobases on the left.

Isolation of the astA plasmid allele from ETEC strain 27D. ETEC strain 27D was selected for further characterization. Similar to H10407, astA as well as the genes for STIb (estA) and CFA/I (cfaB) were localized to a 92-MDa plasmid (data not shown). The astA probe hybridized with a 5.5-kb PstI fragment of the large plasmid from both strains (Fig. 1). We purified the astA-hybridizing PstI plasmid fragment from strain 27D and ligated it into pUC18. The recombinant plasmid was designated pAM1 (Fig. 2). An STIb DNA probe did not hybridize to pAM1, indicating that we had cloned the astA allele separately from STIb. A 2.4-kb fragment of the pAM1 insert was obtained by digestion with SmaI and SphI (from the pUC18 polylinker region) which also hybridized with the astA probe (data not shown). This fragment was ligated back into pUC18 to yield plasmid pAM2 (Fig. 2).

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Restriction map of the 5.5-kb PstI insert of pAM1, below which is shown an expanded map of a 1.3-kb segment comprising IS1414. The two open triangles in the expanded area represent 30-bp imperfect inverted repeats, and the flanking direct repeats are indicated by filled vertical bars. Locations of the astA gene and a putative transposase gene (ORF1) are indicated by open-headed arrows that show their extent and direction of transcription. The astA ORF is in the +1 reading frame relative to ORF1. Extent of the pAM2 insert is shown above the restriction map of pAM1. Restriction sites are indicated by P (PstI), Nr (NruI), E (EcoRV), K (KpnI), S (SmaI), Sp (SphI), H (HindIII), B (BamHI), Nd (NdeI), and X (XmnI).

Characterization of the 1.3-kb region containing astA. The nucleotide sequence of the 1.3-kb region including the astA-hybridizing allele was determined (Fig. 3). Analysis revealed that this region contains a 117-nucleotide (nt)-long ORF predicted to encode a polypeptide of 38 amino acids. Comparison of this sequence with that of the EAST1 structural gene of EAEC strain 17-2 (36) revealed a single nonsynonymous nucleotide sequence difference in codon 21 that resulted in a conserved amino acid substitution (threonine to alanine) (Fig. 3). The astA gene of ETEC strain 83/7 was amplified using primers that anneal to the 5' and 3' ends of the EAEC 17-2 astA coding sequence. Direct DNA sequence analysis of this PCR product showed that it was identical to astA of ETEC 27D (data not shown).

View larger version (54K): [in this window] [in a new window]   FIG. 3.   Nucleotide sequence of IS1414 and deduced amino acid sequence of astA and ORF1 of pAM1. The deduced amino acid sequence of ORF1 (starting at bp 81) is shown above and that of EAST1 (starting at bp 796) is shown below the DNA sequence in one-letter code, with the latter sequence highlighted in gray. The ATG start codon of each gene is boxed, and the purported ribosome binding site for each is underscored by a dotted line. The single nucleotide difference between this astA allele and that of EAEC 17-2 (AG) is marked by a thick underline. The 30-base imperfect IR sequences are underlined, and the flanking 8-base direct repeats are double underlined.

EAST1 phenotypic analysis. Culture filtrates from ETEC strains 27D and HB101(pAM1) were assayed for enterotoxic activity by both the suckling mouse bioassay and the in vitro Ussing chamber assay. ETEC 27D culture filtrates induced fluid accumulation in suckling mice (mean G/B ratio of 0.133 ± 0.009 [standard error of the mean], n = 4 animals) and also increased I_sc in the Ussing chamber assay (Table 2). Culture filtrates from HB101(pAM1) exhibited activity in the Ussing chamber assay (data not shown) but failed to elicit a response in the suckling mouse assay (mean G/B ratio of 0.059 ± 0.002, n = 4 animals). Culture filtrates from HB101(pAM2), a subclone of pAM1, also exhibited enterotoxin activity in the Ussing chamber assay (Table 2). Hence, the presence of astA correlated with detection of enterotoxic activity in Ussing chamber but not in the suckling mouse assay, consistent with earlier findings for the astA allele of EAEC strain 17-2 (35).

In vitro expression of the ORF1 gene product. By in vitro transcription-translation, pAM1 and pAM2, both containing the entirety of ORF1, expressed a distinct protein of M_r ~46,000, which corresponds closely in size to the predicted product of ORF1 (M_r 45,800) (data not shown).

Attempted demonstration of IS1414 transposition. E. coli strain MC4100LAC was transformed with pAM2, and transformants were selected on tetrazolium agar with ampicillin. About 1.5 × 10⁶ transformants were screened, and 7 independent lactose nonfermenting transformants were obtained. These were isolated and analyzed for insertions in lacZ-lacY by PCR. Two of the transformants had an insertion of about 1.2 kb in the 3' lacZ-lacY region. On Southern blot analysis, the amplified lacZ-lacY region containing each of these insertions failed to hybridize with an IS1414 DNA probe, while the homologous sequence in the pAM2 positive control did hybridize with the probe. Accordingly, we concluded that neither insertion into the lac operon was IS1414. We were thus unable to demonstrate transposition of IS1414.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we have extended the observations of Yamamoto and Echeverria (45) showing that human ETEC often carry multiple genomic copies of astA. We determined the nucleotide sequence of astA from two ETEC strains of human origin, 27D and 83/7, and showed that the plasmid allele from one strain, ETEC 27D, expressed toxic activity. The EAST1 coding sequences of both strains were identical to each other and to that reported previously for prototype ETEC H10407 (45) and were nearly identical to that originally reported for EAEC 17-2 (36).

We isolated and further characterized the astA allele from the CFA/I-, STIb-bearing plasmid of one strain, ETEC 27D. Recombinant plasmids containing astA on a 5.5-kb insertion fragment and a 2.4-kb derivative fragment both expressed enterotoxic activity, as evidenced by electrogenic stimulation of rabbit ileal tissue mounted in Ussing chambers. Mutation of the toxin allele, whereby serine 16 was replaced with an amber codon, resulted in a loss of enterotoxin expression. This mutation was silent with respect to ORF1, an overlapping gene on the same strand but in an alternate reading frame to astA. This provides incontrovertible evidence that enterotoxicity is attributable to EAST1 and not to the product of ORF1.

These findings add to the evidence that some ETEC strains have the potential to express EAST1, a factor distinct from other members of its more firmly established arsenal of enterotoxins. With regard to secretogenic capacity, ETEC strains apparently approach the host-pathogen interface with a considerable degree of redundancy. Whether the ability to produce multiple enterotoxins confers pathogenic advantage to ETEC is not entirely clear. Some epidemiological data suggest that strains producing both LT and STI are more likely to cause disease or more severe disease than strains producing only one of these toxins (15, 25). In light of the findings present herein and elsewhere (37, 45, 46), assessment of the incremental effects of ETEC enterotoxins on virulence deserves renewed attention. The independent and additive role of EAST1 in the virulence of ETEC or other diarrheogenic E. coli, if any, may best be addressed by carefully designed epidemiological studies supplemented by studies in the human challenge model with well-defined wild-type strains and EAST1 isogenic derivatives.

The ETEC 27D plasmid allele of astA is part of IS1414, an element with features highly characteristic of an insertion sequence. IS1414 is 1,330 bp long, having one ORF (ORF1) that spans its core region with 30-bp imperfect IR sequences at its ends. It is flanked by 8-bp direct repeat sequences that typically arise from duplication of target DNA upon IS insertion. ORF1 encodes a protein that is highly similar to the predicted transposase proteins of IS285 of Y. pestis, IS1356 of B. cepacia, and ISRm3 of R. meliloti. A protein of the same size as that predicted for the product of ORF1 was expressed in vitro from IS1414 DNA template. We were, however, unable to demonstrate transposition of IS1414. Our lack of success may be explained by a relatively low frequency of transposition into the selected target DNA, the requirement for additional structural elements not present on pAM2, or the presence of mutations within IS1414 which render it nonfunctional.

In the core region of IS1414, the entirety of the astA gene is embedded in the coding region of ORF1, but in the +1 frame relative to the larger gene. This configuration suggests a remarkably economic arrangement for the storage and dissemination of genetic information of importance to bacterial virulence or survival. It represents the rarest type of overlapping gene and one that, to our knowledge, has not been found in a bacterial system. Pairs of completely overlapping genes have been described for several viruses (2, 10, 48). One pertinent example of this arrangement is found in the highly compact hepatitis B virus genome, where the gene encoding hepatitis B surface antigen is contained completely within the DNA polymerase gene (27). A different overlapping gene arrangement has recently been reported for EAEC. Two tandem genes identical to those encoding Shigella enterotoxin 1 were found within and on the complementary noncoding strand of a gene encoding a protein involved in intestinal colonization (Pic) (18). Other arrangements, including partially overlapping genes and intron-containing genes, have been more frequently observed (3, 32).

It is interesting to speculate how IS1414 came to acquire these two overlapping genes. The existence of related IS elements that do not contain the astA sequence implies that acquisition of the toxin gene occurred some time after divergence of the ancestrally related transposases. It is possible that EAST1 then arose by overprinting, de novo translation of the ORF1 transposase sequence in a different frame (22). Origination of new coding sequences by overprinting predicts unusual codon usage for the new gene. Consistent with this prediction, astA has an atypically low CAI (0.132), indicative of a deviation away from preferred E. coli codon usage, while ORF1 shows a greater degree of synonymous codon bias toward E. coli (CAI, 0.362). Bestowal of some advantage to these progeny would favor maintenance and dissemination of the toxin gene, in this case facilitated by IS1414 transposition.

The presence of two overlapping genes poses constraints on evolution, such that a tendency of the individual genes to separate after duplication may be expected (22). Bearing this in mind, we compared the nucleotide sequence of IS1414 ORF1 to the corresponding sequence in the region of astA on the virulence plasmid of EAEC strain 17-2 (36) (S. J. Savarino, unpublished data). In the region of ORF1 5' to nucleotide 667, there is no similarity between the two sequences. In the 3' end of ORF1 (nt 668 to 1209), the span in which the EAST1 coding sequence is also found, the two sequences share 99.2% nucleotide identity. It seems likely that a recombination event in EAEC strain 17-2 at this transition point led to disruption of ORF1 and loss of integrity of IS1414, without affecting the EAST1 coding sequence.

If astA expression confers a particular advantage on survival of E. coli expressing additional virulence factors, such as may be the case with EAEC, degradation of IS1414 function by deletion within the transposase gene might ensure the continued association of astA with other factors encoded by the EAEC virulence plasmid (7). If, on the other hand, astA expression is neutral in a particular background, such as may be the case in ETEC where EAST1 activity may be redundant to other enterotoxins, maintenance of IS1414 function might ensure a reservoir for continued dissemination of the gene into new host strains. Additional studies of IS1414 stability, function, and association with astA in diverse E. coli isolates are necessary to further our understanding of this novel mechanism of evolution of bacterial virulence.