INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Enteroaggregative Escherichia coli (EAEC) is an emerging enteropathogen associated with infantile diarrhea in developing countries (2, 8, 27, 47) and has more recently been associated with food-borne diarrhea outbreaks in developed countries (21, 43). This E. coli pathotype is defined by aggregative adherence (AA) to HEp-2 cells, where bacteria display adherence to the cell surface and also to the intervening substratum in a stacked-brick configuration (27). Potential virulence factors, such as adhesins and toxins, have been identified in EAEC strains (32). However, the complete picture of EAEC pathogenesis remains unclear.

The AA phenotype is associated with the presence of a plasmid 60 to 65 MDa in size and with the expression of one of two distinct aggregative adherence fimbria (AAF/I and AAF/II). AAF/I is expressed in EAEC strain 17-2, is 2 to 3 nm in diameter, and is arranged in flexible bundles (28). The AAF/I biogenesis genes are organized as two unlinked plasmid-borne regions separated by 9 kb (29). Region 1 comprises a contiguous cluster of fimbrial structural genes, in the order chaperone-usher-putative invasin-pilin genes (41). These roles have been assigned on the basis of nucleotide and organizational homology with members of the Dr family of adhesins, a group of fimbrial and nonfimbrial adherence factors which recognize the Dr^a blood group antigen and which have been associated with uropathogenic and possibly diarrheagenic E. coli (18, 22, 36, 50). AAF/I region 2 encodes a regulator termed AggR, which is a member of the AraC family of DNA binding proteins and which is necessary for AAF/I fimbrial expression in strain 17-2 (30). Other Dr adhesins have not been shown to require AraC-homologous regulators.

We have recently described a second EAEC fimbrial antigen, designated AAF/II, in prototype EAEC strain 042 (9), a strain which caused diarrhea in a volunteer study (31). The AAF/II fimbriae are 5 nm in diameter and are arranged in semirigid bundles of filaments. The plasmid-borne fimbrial subunit of AAF/II (aafA) has been cloned, sequenced, and mutated; an aafA mutant lost the ability to adhere to human intestinal tissue in culture, suggesting a role for AAF/II as a human intestinal colonization factor (9). The AAF/I and AAF/II fimbrial subunits are 25% identical and are related to those of the Dr family.

In this work we describe the complete AAF/II fimbrial gene organization in strain 042 and show that the genes define a plasmid-borne virulence gene cluster. We characterize the genes essential for AAF/II biogenesis and demonstrate the role of the AggR transcriptional activator in AAF/II expression. Our data reveal a novel fimbrial genetic organization.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. EAEC strain 042 (O44:H18) used in this study was originally isolated from a child with diarrhea in Lima, Peru. 042 adheres to HEp-2 cells in an aggregative pattern and harbors a 65-MDa adherence-encoding plasmid designated pAA2 (9). 042 expresses AAF/II but not AAF/I. Also used in this study was HB101 transformed with plasmid pAA2 [designated HB101(pAA2)], which is able to adhere to HEp-2 cells in an AA pattern (9).

1

1

1

1

1

1

1

1

Recombinant DNA techniques. DNA analysis and cloning were performed as described by Sambrook et al. (39). A plasmid Midi kit (Qiagen, Inc., Chatsworth, Calif.) was used for large-scale DNA preparation. Restriction fragments were isolated by using Prep-A-Gene (Bio-Rad Laboratories, Hercules, Calif.). Plasmid DNA was introduced into host cells by the calcium chloride method (19) or by electroporation (2.5 kV, 25 µF, and 200 ). DNA fragments used in hybridization were labeled with [-³²P]dATP by random priming using a Prime-It kit and protocols recommended by the manufacturer (Boehringer Mannheim, Indianapolis, Ind.).

Transposon mutagenesis. Strain 042 was mutagenized by using transposon TnphoA delivered on the suicide vector pRT733 (44), following the methodology described by Nataro et al. (28). Alkaline phosphatase-producing mutants were selected on L-agar plates containing kanamycin and X-P. Mutants lacking HEp-2 adherence were selected for further cloning and sequencing analysis. The sites of TnphoA insertion were determined by cloning of genomic fragments conferring kanamycin resistance and subsequent nucleotide sequencing of the flanking DNA.

Cosmid library construction. Plasmid DNA from strain 042 was purified and was partially digested with Sau3a to yield 15- to 30-kb fragments. After separation on a 0.7% agarose gel, fragments were extracted and were ligated with BamHI-digested cosmid vector pCVD301. Ligated DNA was packaged into phage by using the Gigapack II packaging extract (Stratagene, La Jolla, Calif.) and then transfected into strain HB101, and cosmid clones were selected on L-agar plates containing tetracycline. The cosmid clones were screened for the AA pattern in the HEp-2 cell adherence assay. Clones exhibiting HEp-2 adherence were selected for further analysis.

pBluescript pAA2 library construction. Purified plasmid DNA from strain 042 was sonicated to generate fragments 1 to 5 kb in length. After separation in a 0.7% agarose gel, 1.5- to 2.5-kb fragments were eluted from the gel and were blunt ended by using Pfu DNA polymerase as instructed by the manufacturer (Stratagene). Fragments were ligated into SmaI-digested pBluescript. Ligated DNA was transformed into strain DH5, selecting for ampicillin resistance.

DNA sequencing and analysis. DNA sequencing was performed in the Biopolymer Laboratory, Department of Microbiology and Immunology, University of Maryland School of Medicine. Double-stranded plasmid DNA was sequenced by the chain terminator method (40) with an Applied Biosystem model 373A sequencer. Sequence analysis was performed through Genepro sequence analysis software (version 5.00; Riverside Scientific, Bainbridge Island, Wash.), DNASIS version 2.10, and the Wisconsin sequence analysis package (Genetics Computer Group), available through the Center of Marine Biotechnology, University of Maryland, and programs available through the National Center for Biotechnology Information.

Directed insertional mutagenesis. To identify open reading frames (ORFs) related to AAF/II fimbrial biogenesis, the following strategy was used. Internal segments of the target genes were prepared by PCR, and primers were designed such that KpnI (5')-SacI (3') restriction sites were engineered at the ends of each fragment. PCR products were digested with KpnI and SacI and were cloned into the corresponding sites in suicide vector pJP5603. PCRs were performed with an Opti-Prime PCR optimization kit (Stratagene), plasmid pAA2 as the DNA template, and the PCR primers listed in Table 2. For cloning, PCR products were resolved by agarose gel electrophoresis and were eluted by using a Prep-A-Gene kit (Bio-Rad). After ligation and transformation into DH5(pir), transformants were selected on L-agar plates containing kanamycin and X-Gal; plasmid DNA of transformants harboring the correct insert was transformed into strain S17-1(pir) for mobilization into strain 042 (nalidixic acid resistant) via filter mating on cellulose nitrate membranes. Transconjugants were selected on L-agar plates containing kanamycin and nalidixic acid, and their identity was confirmed by agglutination with anti-O44 antisera (Unipath, Inc., Ogdensburg, N.Y.). The site of integration was confirmed by restriction mapping. To terminate transcription, the 2.0-kb  interposon (38) with BamHI ends was cloned into the corresponding site of pJP5603 downstream of the PCR product internal to aafD'. The construction was mobilized into strain 042, and transconjugants were selected by using the technique described above.

RNA analysis. Total RNA was extracted from log-phase L-broth cultures of strains 042, 042 aggR, and HB101, using the Ultraspec RNA isolation system (Biotecx Laboratories, Houston, Tex.) according to the manufacturer's protocol. 042 aggR, also called 042::pWPE20, was constructed as described above by insertion of an aggR internal fragment into pJP5603 and integration into the aggR gene of pAA2. After extraction, RNA samples were treated with DNase I-amplification grade (Gibco BRL, Gaithersburg, Md.), and 20-µg aliquots of total RNA were denatured and blotted onto nylon transfer membranes (Schleicher & Schuell, Keene, N.H.). As a control, each sample was blotted in duplicate, with one blot treated with 80 U of RNase (Promega, Inc., Madison, Wis.) for 30 min at 37°C. Membranes were hybridized with aafA, aafB, aafC, aafD, or aafD' -³²P-labeled DNA probes (obtained by PCR amplification using the primers described in Table 2). Membranes were hybridized, washed, and visualized by autoradiography as described by Sambrook et al. (39). Only autoradiographic signals that were abolished by RNase treatment were considered to represent mRNA transcripts.

HEp-2 cell adherence assay. The HEp-2 adherence assay was performed as described by Vial et al. (45). Briefly, HEp-2 cells were grown to 50% confluent monolayers on glass coverslips in a 24-well tissue culture plate. After a gentle wash with phosphate-buffered saline (PBS), 1 ml of fresh Eagle's minimal essential medium with 1% D-mannose was added to each well with 40 µl of an overnight L-broth bacterial culture. The assay mixture was incubated for 3 h (for 042 and its derivatives) or 6 h (for HB101 and derivatives) at 37°C in 5% CO₂. After the incubation period, cells were washed several times with PBS, fixed with methanol, stained with 10% Giemsa stain for 15 min, and examined by light microscopy. The tests were considered positive for the AA phenotype when bacteria displayed the stacked-brick aggregates attached to HEp-2 cells and to the glass substratum between cells.

Polystyrene adherence. Bacterial strains were grown overnight at 37°C in L broth in 24-well polystyrene cell culture dishes (Falcon Industries, Becton-Dickinson, Lincoln Park, N.J.). Plates were then gently washed twice with PBS, dried, and stained with 10% Giemsa stain. The plates were analyzed visually for adherent bacteria present as a film on the plastic surface (20).

Electron microscopy. Strains were grown statically in L broth in 24-well cell culture dishes. Cultures were passaged twice for 24 h each time at 37°C. Transmission electron microscopy of negatively stained specimens was performed by standard methodology (25). Grids were examined with a JEOL JEM 1200 EX II transmission electron microscope. AAF/II was detected by immunogold electron microscopy as previously described (9).

Nucleotide sequence accession numbers. The sequences derived from the analyses performed in this study are deposited in GenBank under accession no. AF114827 and AF114828.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

TnphoA mutagenesis of strain 042. To localize genes required for the biogenesis of AAF/II, we performed transposon TnphoA mutagenesis on strain 042 and screened 480 insertion mutants for adherence to polystyrene culture dishes and to HEp-2 cells; these two phenotypes have been shown to correlate with AAF/II expression (9). Twelve nonsibling mutants that were deficient in adherence to both cells and plastic were identified. By nucleotide sequence and Southern blot analysis, 11 of these insertions were localized to the previously described AAF/II fimbrial subunit, aafA, which is located on the pAA2 plasmid (9) (GenBank accession no. AF012835). The remaining site identified by sequence analysis was found to lie within the chromosomal dsbA gene, which encodes a periplasmic enzyme catalyzing disulfide bond formation (1) (Genbank accession no. M77746). Like many fimbrial subunits, the aafA gene features two cysteine residues with the potential to form a disulfide-stabilized loop.

Localization of AAF/II biogenesis genes on plasmid pAA2. In a further effort to define the complete AAF/II biogenesis cluster, we constructed a cosmid pCVD301 library of Sau3a partially digested fragments derived from plasmid pAA2. This approach was used because the biogenesis cluster of AAF/I fimbriae had been shown to comprise a primary biogenesis cluster regulated by an unlinked transcription activator (AggR) encoded approximately 9 kb from the other required genes (29, 30). The cosmid library was screened for clones that expressed polystyrene adhesion and aggregative adherence to HEp-2 cells. Five cosmid clones (designated C9, D6, F9, E6, and I10) were found to confer adhesion in these assays. The insert DNA from each of these cosmids was analyzed by restriction analysis, and the cosmids were found to have in common a region of approximately 23 kb. This region featured a 13-kb MluI fragment, which we have shown previously to encode the enterotoxins Pet and EAST, and the AafA fimbrial subunit (14).

Sequence analysis of the fimbrial biogenesis loci. The nucleotide sequence of the DNA corresponding to the inserts of each of the adherence-positive cosmid clones was determined by use of a pBluescript II SK library. Selected restriction fragments derived from the pCVD301 cosmid clones were used to hybridize a pAA2 gene bank constructed in pBluescript II SK; the inserts of these pBluescript clones were then sequenced from universal primers. These experiments yielded 25 kb of contiguous sequence that comprised the complete adherence-encoding region.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Map of the virulence region of pAA2. Black arrows represent AAF-related reading frames discussed in this paper. The pet and astA genes are enterotoxins previously published. IS-like elements are indicated by boxes; the boxes with horizontal lines represent directly repeated IS-like elements that are homologous to IS1353. Boxes with vertical stripes represent complete IS1 elements. Line, 1 kb. The inset at the top shows the upstream region of the aafD gene. Indicated are the predicted ATG start (italics), ribosomal binding site, and 10 and 35 promoter regions. Upstream from the predicted aafD promoter lies a region with high homology to the araI1 half-site, shown to be the binding site of the AraC protein (35). This region is thus a candidate binding site of the AraC homolog AggR, under which the aafD gene is positively regulated (see text).

Insertional mutagenesis of fimbrial biogenesis genes. Sequence analysis of cosmid D6 suggested that the AAF/II biogenesis genes may be arranged as two unlinked clusters with a unique organization. Of particular interest was the presence of the two chaperone-homologous regions and the apparently incomplete nature of each cluster. We undertook an insertional mutagenesis strategy to determine conclusively which of the sequenced ORFs was required for expression of the AAF/II fimbriae. An internal fragment of each of the putative biogenesis ORFs (aafB, aafC, aafD, aafD', and aggR [Table 2]) was cloned into the multiple cloning site of the suicide vector pJP5603. These constructions were then independently mobilized into parent strain 042, and integration into the homologous genes was confirmed. An aafA mutant had been obtained previously by TnphoA mutagenesis (9). All mutants were screened for adherence to plastic and to HEp-2 cells and for the presence of AAF/II fimbriae by electron microscopy. The latter was determined by transforming the mutated pAA2 plasmids into HB101 to avoid confusion with other surface fimbriae of 042. Phenotypic analyses showed perfect correlation between adherence and fimbrial expression. Insertional inactivation of aafA, aafC, aafD, and aggR resulted in loss of fimbrial expression, whereas inactivation of the aafD' and aafB reading frames did not result in loss of these phenotypes.

Transcription of AAF/II-related genes. Our data suggest that the aafD' gene is a pseudogene derived from a fimbrial chaperone homolog. Data from Bilge et al. (3) have suggested that Dr fimbrial clusters, to which the AAF/II region 2 bears organizational and nucleotide homology, are transcribed from a single major promoter upstream from the first gene in the polycistronic message. We therefore asked whether the promoter of the region 2 cluster was located upstream of the silent aafD' gene or whether instead it was located immediately upstream of the aafC gene. We constructed a mutation in the aafD' locus by integration of pJP5603 (in strain 042) exactly as described above except that in addition to the internal aafD' fragment, we also inserted an  interposon to terminate transcription at the site of integration. As for the previous mutation in the aafD' gene, this construction was able to express functional fimbriae indistinguishable from those of the wild-type parent. These data suggest that the promoter for the region 2 cluster lies upstream from aafC and not from aafD' as would be expected from published observations of other Dr fimbriae (5).

View larger version (23K): [in this window] [in a new window]   FIG. 2.   RNA dot blots of aafD, aafA, and aafC transcripts in the presence and absence of a functional aggR gene product. Bacterial RNA prepared from L-broth cultures of 042 or 042 aggR was blotted to nitrocellulose and hybridized with probes generated by PCR. Details are presented in Materials and Methods. aafA and aafD transcripts were clearly increased in the presence of AggR; the aafC transcript was barely detectable but did not exhibit AggR dependence.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   Predicted secondary structure of the aggA downstream region. Nucleotide 1 is the first base of the aafA stop codon. The sequence was analyzed by using the Zuker-Stiegler algorithm processed with DNASIS version 2.10 software.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The mechanism of EAEC pathogenicity is still not completely elucidated, but several potential virulence factors have been described (32). These factors include several putative fimbrial and nonfimbrial adhesins (12, 24, 28, 46). The fimbrial adhesin AAF/II, expressed in human-virulent strain 042, has been shown to confer aggregative adherence and adherence to the human colonic mucosa (9). Although the fimbrial subunit gene (aafA) has been cloned and sequenced, the complete genetic organization of AAF/II has not been described.

In this work we have characterized the genes necessary for AAF/II expression. Our mutagenesis data demonstrate that dsbA and two plasmid-borne loci are required for expression of this adherence factor. The periplasmic thiol:disulfide oxidoreductase enzyme encoded by dsbA has been demonstrated to be necessary for the expression of other fimbrial antigens (49), presumably by catalyzing the formation of a disulfide loop in the fimbrial subunit protein.

We report here that the plasmid-borne AAF/II biogenesis genes feature a unique organization. The fimbrial genes are organized in two clusters separated by a 12-kb segment; the latter contains the genes encoding two enterotoxins, Pet and EAST, and a series of IS elements (9, 14). Analysis of AAF/II genes indicate that each is closely related to genes encoding the Dr family of adhesins (36). This family includes the nonfimbrial uropathogenic E. coli adhesins AFA-I, AFA-III, and Dr hemagglutinin (49); the F1845 fimbriae of DAEC (4); and more distantly AAF/I, whose fimbrial subunit is closest to that of AAF/II (9). However, all members of this family thus far described feature a conserved genetic arrangement, chaperone-usher-invasin-pilin, thought to be transcribed as a polycistronic message (3). Despite being a relative of this family, the AAF/II fimbria displays a markedly different organization.

In a genetic segment that we have designated AAF/II region 1, we localized the aafD, aafA, and aggR genes, encoding the chaperone, the pilin, and the transcriptional activator, respectively. Homology analysis demonstrates high similarity between each of the genes of this region and the AAF/I biogenesis genes. We have previously shown that AafA is the fimbrial subunit and that its closest homolog is AggA, the pilin subunit of AAF/I (9). The closest homolog of AafD is AggD of AAF/I (41); lesser homology is seen with other periplasmic chaperone proteins of Dr adhesins implicated in adhesin assembly. The aggR gene downstream of aafA displays almost complete homology at the nucleotide level to aggR of the AAF/I operon (30). aggR encodes a protein which belongs to the AraC binding-protein family and is required for AAF/I expression in strain 17-2 (30). We have found that for AAF/II, this gene is located only 956 bp downstream from aafA, while in AAF/I it is located 9 kb away from the fimbrial gene cluster.

The remaining AAF/II biogenesis genes are situated in region 2, comprising aafD' (a silent chaperone-like locus), aafC, and aafB, which apparently encode an usher and the putative Dr invasin (16), respectively. Notably, the cryohemagglutinin described by Yamamoto et al. (48) appears to be AAF/II, as the inactivating insertion site reported by those authors maps to the aafC sequence (33). Our data do not allow us to draw inferences regarding a possible role for the aafB gene, since it does not appear to be required for the expression of functional AAF/II fimbriae or for wild-type adherence to HEp-2 cells.

The most notable feature of the AAF/II organization is the separation of the usher and chaperone genes in unlinked clusters. Nucleotide homology studies have allowed us to postulate one possible mechanism for the origin of this unique arrangement. Both the region 1 and region 2 clusters have some overall similarities in organization to the highly conserved Dr family biogenesis operons: region 1 begins with a chaperone homolog and ends with a subunit gene; region 2 begins with a chaperone locus (though silent) and then progresses, like other Dr adhesins, to the usher and then the putative invasin gene before truncating short of the expected pilin gene. Interestingly, whereas the closest homologs of the region 1 genes are the AAF/I genes, the closest homologs of the region 2 genes are genes reported for the Dr adhesin AFA-III. These data suggest that the two regions originated independently and not as the result of duplication. Moreover, it is notable that both region 1 and region 2 are flanked on one side by apparently complete IS1 transposase genes (region 1 downstream only and region 2 upstream only). It has been reported that the AFA-III cluster is flanked by intact IS1 elements which are capable of mediating transposition of the entire afa3 operon (17). Our data are thus most compatible with the following model: region 1 is likely to have been the original adhesin gene cluster, since it is encoded upon a plasmid that, overall, is highly homologous to that encoding AAF/I (33). We hypothesize that a Dr adhesin transposed onto the plasmid, which then would have harbored more than one copy of the chaperone, usher, invasin, and, most likely, a second pilin. Since this redundancy was apparently not adaptive, the organism evolved to lose one copy of each gene: the pilin from region 2 by deletion and the usher and invasin from region 1, also by deletion. The redundant chaperone in region 2 was apparently lost by point mutations. Analysis of G+C content of region 1 and region 2 genes supports this model of AAF/II evolution: G+C content of the region 1 genes aafA (41.2%) and aafD (37.5%) averages 37.6%, while that of the region 2 genes aafC (53.7%) and aafB (44.7%) and the pseudogene aafD' (45.4%) averages a substantially higher 51.5%. The G+C contents of AAF/I genes aggA (37.6%), aggB (39.7%), aggC (41.7%), aggD (36.4%), and aggR (30.8%) are markedly lower than that of afa genes, which are in the range of 56 to 61% (accession no. X76688). Despite these differences, our data suggest that the accessory biogenesis genes of the two regions act in concert, and the ability of Dr accessory proteins to substitute for each other in the expression of Dr adhesin subunits has been demonstrated previously (7).

The presence of IS elements may play an important role in dissemination of Dr fimbrial genes. The presence of these elements apparently not only is conserved among Dr adhesins but also extends to their relatives, the AAFs, as well. On their sides lacking the IS1 elements, both AAF/II region 1 and region 2 have evidence of a gene that is also found upstream of the CS1 and CS5 fimbrial clusters and which also displays features of an insertion sequence. Whether this gene plays a role in the dissemination of fimbrial gene clusters remains to be determined.

We demonstrate that AggR regulates transcription of aafA and aafD in region 1. AggR apparently acts directly to regulate the aafD promoter; however, the role of AggR in aafA transcription is not clear. Bilge et al. have suggested that the daaE (major adhesin) transcript of the Dr fimbria F1845 is present in greater abundance than those of upstream accessory genes, despite polycistronic transcription from a common upstream promoter (3). The mechanism of this effect was reported to be enhanced stability of the daaE transcript mediated by a downstream hairpin structure. We have identified a candidate hairpin downstream of aafA, but in addition, our prior work complementing an aafA mutant suggests that there may be an independent, perhaps AggR-regulated, promoter upstream of aafA (9). Given that AggR was shown to be required for aggA expression in AAF/I-producing strain 17-2 (30) and that AraC homologs are not known to be involved in AFA regulation, these data support our hypothetical model of AAF/II evolution.

Virulence in EAEC strains has been associated with the presence of a high-molecular-weight plasmid which seems to be conserved among different strains (33). The 23-kb DNA sequence of pAA2 described here defines a plasmid-borne virulence gene cluster. In this large plasmid region are localized the genes encoding the Pet and EAST enterotoxins and the proteins related to AAF/II biogenesis. This gene cluster seems to be an important virulence-related element in pAA2, since both AAF/II and Pet have been associated with EAEC effects on intestinal tissue (9, 34). Further studies will test whether these genes are both necessary and sufficient for EAEC pathogenesis.