Binding to sulfatide and enterotoxicity of various Escherichia coli STb mutants.

Binding of the 48 amino acid polypeptide of the mature heat-stable Escherichia coli enterotoxin b (STb) to the functional receptor sulfatide (SFT) constitutes the first step in inducing secretory diarrhoea in the intestinal lumen of animals. The NMR structure of this toxin dictated the choice of amino acids for site-directed mutagenesis to delineate the binding site of STb to SFT. Amino acids facing the solvent either in the loop or the hydrophobic alpha-helix were selected. Seventeen site-specific mutants of STb toxin were produced and purified by high-pressure liquid chromatography. Enterotoxicity of the 17 mutants was determined using a rat loop assay and binding was evaluated using a microtitre plate binding assay. Both hydrophobic and electrostatic interactions are important for STb attachment. When mutations (F37K, I41S and M42S) were introduced into the hydrophobic alpha-helix to lessen hydrophobicity, binding activity and enterotoxicity decreased by more than sixfold. The loop defined by C21 and C36 also made specific contributions. Mutants generated at basic residues (K22, K23 and R29) within this region exhibited both reduced binding activities and reduced toxic activities. For all STb mutants constructed and analysed, when binding to SFT was reduced, a reduction in toxicity equivalent or greater was noted, indicating that binding to SFT is a step that precedes the toxic effect observed for STb toxin. Significantly, when the negatively charged D30 was substituted for either alanine or valine, the binding to SFT was about twice that of native STb, whereas the enterotoxicity was reduced by half.

Oligomerization of Escherichia coli enterotoxin b through its C-terminal hydrophobic alpha-helix.

Using a chemical cross-linker and gel electrophoresis or a dot blot overlay assay, we studied protein-protein interaction of STb toxin, a 48-residue amphiphilic polypeptide causing intestinal disorders. For the first time, we report on the oligomerization property of STb. This enterotoxin forms hexamers and heptamers in a temperature-independent fashion in presence or absence of its receptor (sulfatide) anchored in a 50-nm liposome or as a free molecule. Full STb structure integrity is necessary for its oligomerization as this process is not observed under reducing conditions in the presence of beta-mercaptoethanol. STb treatment with tetramethylurea (TMU) and different detergents prevented oligomerization. Site-directed mutagenesis decreasing overall STb hydrophobicity in the hydrophobic alpha-helix resulted in the incapacity to form oligomers. Taken together, these data suggest that the C-terminal hydrophobic alpha-helix corresponds to the domain of STb-STb inter-binding where hydrophobic interaction is involved.

Cloning and characterization of the esp region from a dog attaching and effacing Escherichia coli strain 4221 and detection of EspB protein-binding to HEp-2 cells.

The espA, espB and espD genes from enteropathogenic Escherichia coli were previously shown to be essential for triggering the signal transduction in infected host cells. We have cloned and determined the nucleotide sequences of the espA, espB and espD homologues from an E. coli strain (4221) isolated from a dog which manifested the attaching and effacing lesions in the small intestine. This strain is designated as a dog enteropathogenic E. coli. When comparing predicted amino acid sequences to those of the corresponding proteins from enteropathogenic E. coli O127, enterohemorrhagic E. coli serotype O26, enterohemorrhagic E. coli O157 and rabbit enteropathogenic E. coli, the EspADEPEC protein showed the same level of similarity (75% identity) with EspA of enteropathogenic E. coli O127 and rabbit enteropathogenic E. coli. The EspBDEPEC protein showed the highest similarity with the EspB of enteropathogenic E. coli O127 (99% identity). The EspDDEPEC protein showed 88% identity with the EspDEPEC. We constructed and purified a maltose-binding fusion protein containing the product of the entire espBDEPEC gene of the dog enteropathogenic E. coli strain 4221. Purified maltose-binding protein-EspBDEPEC fusion protein was shown to bind efficiently to HEp-2 cells in a localized fashion as shown by immunofluorescence microscopy. In addition, when the dog enteropathogenic E. coli strain 4221 was grown in tissue culture medium (DMEM) supplemented with serum, a secreted 36-kDa protein was identified by immunoblot analysis using a polyclonal antiserum against the maltose-binding protein-EspBDEPEC fusion protein.

Production of Escherichia coli STb enterotoxin is subject to catabolite repression.

Enterotoxigenic Escherichia coli are known to secrete several types of toxins including STb, a heat-stable enterotoxin. STb enterotoxin production was studied in wild-type E. coli strains. Using a quantitative STb-specific inhibition ELISA, the amount of toxin present in the culture supernatant fractions of various E. coli strains was determined. Variation in the production of STb toxin was observed for the wild-type strains. For E. coli strain 82-4247 grown in trypticase soy broth, the toxin was produced after 4 h of growth and was maximal after about 57 h of growth. The amount of toxin in the culture supernatant fraction increased concomitantly with bacterial growth. Using the rat loop assay, the biological activity of STb was retained even after the logarithmic phase of growth when STb production levelled off (i.e. from 24 to 74 h). STb production by E. coli strain 82-4247 varied with the culture medium used. In particular, addition of 1.0% (w/v) compared to 0.1% glucose to Davis minimal medium decreased STb production, whereas addition of 1.0% (w/v) glycerol did not affect STb production. Addition of exogenous cAMP reversed the repressive effect of glucose. Using mutant strains, STb production was shown to be subject to catabolite repression.

Fusion of the genes encoding Escherichia coli heat-stable enterotoxin b (STb) and the maltose-binding protein to obtain mature STb enterotoxin.

The heat-stable enterotoxin b gene (estB) of Escherichia coli was fused to the gene for maltose-binding protein (malE). The estB gene was cloned into the pMAL-p vector using PCR. The construct consists of the signal sequence of maltose-binding protein, which directs the export of the fusion protein to the periplasm, and the maltose-binding protein fused to the STb polypeptide. A sequence encoding a factor Xa cleavage site is present between malE and estB. The fused genes are controlled by Ptac, a strong inducible promoter. Following IPTG induction, the recombinant strain expressed a 47 kDa protein, which was easily purified from osmotic shock fluid by using preparative electrophoresis and electroelution. Cleavage of the fusion protein with factor Xa generated the maltose-binding protein (42 kDa) and a polypeptide of approximately 5 kDa, corresponding to the molecular mass of mature STb. A monospecific polyclonal rabbit antiserum raised against purified STb reacted in immunoblot with the fusion protein and the cleaved-off peptide. A positive response was observed when testing the osmotic shock fluid containing the fusion protein in a rat intestinal loop assay. On average, 3-4 mg of MBP-STb protein was recovered per litre of induced recombinant strain.

Antigenic differences among Campylobacter fetus S-layer proteins.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of S-layer proteins extracted from Campylobacter fetus strains by using acid glycine buffer showed that the predominant S-layer proteins of different strains had subunit molecular weights in the range of 90,000 to 140,000. Electron microscopy revealed oblique S-layer lattices with a spacing of approximately 5.6 nm (gamma = 75 degrees) on wild-type strains VC1, VC119, VC202, and VC203. Three variants of C. fetus VC119 producing a predominant S-layer subunit protein of different molecular weight (Mr) from that of the parent were also examined. Each variant produced an oblique lattice morphologically indistinguishable from that of the parent. Amino-terminal sequence analysis showed that the S-layer proteins of the VC119 parent and variants were identical up to residue 18 and that this sequence differed from but was related to the first 16 N-terminal residues shared by the S-layer proteins of the three other wild-type C. fetus isolates. Western immunoblot analysis with an antiserum prepared to the VC119 protein and an antiserum prepared to C. fetus 84-40 LP (Z. Pei, R. T. Ellison, R. V. Lewis, and M. J. Blaser, J. Biol. Chem. 263:6416-6420, 1988) showed that strains of C. fetus were capable of producing S-layer proteins with at least four different antigenic specificities. Immunoelectron microscopy with antiserum to the VC119 S-layer protein showed that C. fetus cultures contained cells with immunoreactive oblique S-layer lattices as well as cells with oblique S-layer lattices which did not bind antibody. This suggests that C. fetus S-layer proteins undergo antigenic variation. Thermal denaturation experiments indicated that the antigenicity conferred by the surface-exposed C. fetus S-layer epitopes was unusually resistant to heat, and the thermal stability appeared to be due to the highly organized lattice structure of the S. layer. Protease digestion of purified VC119 S-layer protein revealed a trypsin-, chymotrypsin-, and endoproteinase Glu-C-resistant domain with an apparent Mr of 110,000, which carried the majority of the epitopes of the S-layer protein, and a small enzyme-sensitive domain. The trypsin- and chymotrypsin-resistant polypeptides shared an overlapping sequence which differed from the N-terminal sequence of the intact S-layer protein.

Purification, characterization, and localization of a protein antigen shared by thermophilic campylobacters.

A protein antigen with an apparent molecular weight (Mr) of 31,000 was isolated from 0.2 M glycine hydrochloride (pH 2.2) extracts of a typical human fecal isolate, Campylobacter jejuni VC74. The protein was purified to homogeneity on a preparative scale by immunoaffinity chromatography followed by molecular sieving with a Superose 12 column. Isoelectric focusing under nondenaturing conditions indicated a pI of 9.3, and amino acid composition analysis showed that the protein was unusually rich in lysine, containing 14.9 mol% of this basic amino acid. Cysteine and tryptophan were absent. The protein also contained approximately 35% hydrophobic amino acid residues, and N-terminal amino acid analysis showed that 17 of the first 38 residues were hydrophobic. This amino-terminal sequence to residue 22 was virtually identical to that of an antigenically cross-reactive 31,000-Mr protein isolated from another C. jejuni strain belonging to a different heat-labile serogroup. Western blotting (immunoblotting) of glycine extracts of other C. jejuni, Campylobacter coli, and Campylobacter laridis strains belonging to different thermolabile and thermostable serotypes, as well as Campylobacter fetus, with a rabbit polyclonal antiserum raised against the purified C. jejuni VC74 protein showed that all C. jejuni, C. coli, and C. laridis strains tested contained a 31,000-Mr protein with epitopes which were antigenically cross-reactive with the C. jejuni VC74 protein. The antigenically cross-reactive epitopes of this protein were also readily detected by immunodot blot assay of glycine extracts of C. jejuni, C. coli, and C. laridis with monospecific polyclonal antisera to the 31,000-Mr protein, suggesting that this serological test could be a useful addition to those currently employed in the rapid identification of these important pathogens. Slide agglutination reactions, immunofluorescence assay, and immunogold electron microscopy with antisera to purified 31,000-Mr protein and trypsin treatment of whole cells indicated that the cross-reactive epitopes of the 31,000-Mr protein were not exposed on the cell surface. Cell fractionation analysis and immunogold electron microscopy located the protein on the outer surface of the cytoplasmic membrane. This finding suggests that the 31,000-Mr protein is not a good candidate for inclusion in a monovalent subunit Campylobacter vaccine.

Structural and biochemical analyses of a surface array protein of Campylobacter fetus.

Electron microscopic examination of ultrathin sections and freeze-etched and shadow cast preparations of a bovine prepuce isolate of Campylobacter fetus VC119 showed an S layer with subunits in an apparent linear arrangement. Surface radioiodination, enzyme digestion, low-pH extraction, and Western immunoblotting showed that the layer was composed mainly of one protein which is the predominant protein antigen of C. fetus. This protein was purified to homogeneity by gel filtration, ion-exchange chromatography, and high-performance liquid chromatography. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis showed an apparent molecular weight of 131,000 for this protein with a pI of 6.35, and no carbohydrate could be detected by a variety of techniques. Amino acid composition analysis showed that the protein contained approximately 1,304 residues per molecule, 41.2% of which were hydrophobic and approximately 22% of which were acidic. Cysteine and histidine were absent. Circular dichroism spectra showed that the prominent structure of the S layer protein was a beta-pleated sheet (36%) with aperiodic foldings (31%); a moderate amount of alpha-helix (28%) and a low amount of beta-turn (5%) were also present. The N-terminal amino acid sequence was determined for the first 18 residues. No sequence homology with other S layer proteins was found.