LT-IIc, a New Member of the Type II Heat-Labile Enterotoxin Family Encoded by an Escherichia coli Strain Obtained from a Nonmammalian Host ▿
Two families of bacterial heat-labile enterotoxins (HLTs) have been described: the type I HLTs are comprised of cholera toxin (CT) of Vibrio cholerae, LT-I of Escherichia coli, and several related HLTs; the type II HLTs are comprised of LT-IIa and LT-IIb. Herein, we report LT-IIc, a new type II HLT encoded from an enterotoxigenic E. coli (ETEC) strain isolated from an avian host. Using a mouse Y1 adrenal cell bioassay, LT-IIc was shown to be less cytotoxic than CT, LT-IIa, or LT-IIb. Cytotoxicity of LT-IIc was partially neutralized by antisera recognizing LT-IIa or LT-IIb but not by anti-CT antiserum. Genes encoding putative A polypeptide and B polypeptides of LT-IIc were arranged in an operon which was flanked by potential prophage sequences. Analysis of the nucleotide and predicted amino acid sequences demonstrated that the A polypeptide of LT-IIc has moderate homology to the A polypeptides of CT and LT-I and high homology to the A polypeptides of LT-IIa and LT-IIb. The B polypeptide of LT-IIc exhibited no significant homology to the B polypeptides of CT and LT-I and only moderate homology to the B polypeptides of LT-IIa and LT-IIb. The binding pattern of LT-IIc for gangliosides was distinctive from that of either LT-IIa or LT-IIb. The data suggest that other types of the type II HLT subfamily are circulating in the environment and that host specificity of type II HLT is likely governed by changes in the B polypeptide which mediate binding to receptors.
Infections caused by enterotoxigenic Escherichia coli (ETEC) are the leading cause of traveler's diarrhea and the major cause of diarrheal disease in underdeveloped nations, especially among children. ETEC, which is usually transmitted by food or water contaminated with animal or human feces, is estimated to be responsible annually for more than 650 million cases of enteric infections and nearly 800,000 deaths (29). Infection begins with ingestion of bacteria, followed by elaboration of enterotoxin and bacterial colonization of the gut, and presents as a profuse watery diarrhea which disseminates the bacteria back into the environment (10).
ETEC strains are lactose-fermenting E. coli strains that produce a heat-labile enterotoxin (LT, hereafter referred to as LT-I), heat-stable enterotoxins (ST), or both and colonization factors which enable ETEC to colonize the small intestine (22). The pathogenesis of ETEC is dependent on the strains' capacity to produce LT-I and/or ST (10, 29). LT-I is closely related functionally, antigenically, and structurally to cholera toxin (CT), the heat-labile enterotoxin produced by Vibrio cholerae. Antiserum against CT neutralizes the toxicity of LT-I, and antiserum against LT-I neutralizes the toxicity of CT (15). Structurally, LT-I and CT are oligomeric proteins composed of an A polypeptide which is noncovalently coupled to a pentameric array of B polypeptides (15). The A polypeptide of LT-I and CT is enzymatically active and catalyzes an ADP-ribosylation of the Gsα regulatory protein in the intoxicated cell. Ribosylation of this regulatory protein constitutively activates adenylate cyclase, the enzyme which catalyzes production of cyclic AMP (cAMP) (3, 20). Accumulation of cAMP induces the intoxicated cell to secrete electrolytes and chloride ions, thus generating the watery diarrhea, which is symptomatic of intoxication. Intracellular accumulation of cAMP modulates other cellular processes such as protein kinase activity, activation of calcium channels, etc. (15). Binding of LT-I and CT to ganglioside receptors is mediated by the B polypeptides. Gangliosides are members of a heterogeneous family of sialylated glycosphingolipids expressed on the surface of eukaryotic cells (9). Based on these characteristics, LT-I and CT have been designated as members of the large family of toxins known as the A1B5 ADP-ribosylating heat-labile enterotoxins (HLTs).
LT-IIa and LT-IIb, two new members of the A1B5 family of HLTs produced by E. coli, were recently described (11, 12, 27). While it is clear that LT-IIa and LT-IIb are evolutionarily related to LT-I and CT, there are major differences between the two groups of enterotoxins. LT-IIa and LT-IIb are antigenically distinguishable from LT-I and CT and from each other (12). These antigenic differences are reflected in the low amino acid sequence similarity of the A polypeptides and the virtual absence of amino acid sequence homology of the B polypeptides between the two groups (LT-I and CT versus LT-IIa and LT-IIb) (35). To distinguish between CT and LT-I and LT-IIa and LT-IIb, the HLTs were catalogued into two subfamilies. The division, based upon the genetic, biochemical, and immunological characteristics of the various enterotoxins, assigned CT and LT-I to the type I subfamily, while LT-IIa and LT-IIb were assigned to the type II subfamily (11, 12, 15).
Strains producing type II HLT have been isolated from various sources. Type II HLT-producing strains of E. coli have been isolated in Thailand from water buffalo and from cooked beef, which had been submitted to a food microbiology laboratory in São Paulo, Brazil (11, 27). ETEC strains producing HLT with apparent homology to type II enterotoxins have also been isolated from human patients and from other mammals which were exhibiting symptoms of diarrhea (11). Recently, four strains of E. coli obtained from the feces of diarrheic ostriches were identified in an animal clinic in Brazil (21). In vitro cytotoxicity assays using CHO and Vero cells indicated that these strains were toxigenic (21). The toxic activities produced by these strains could not be neutralized by anti-CT antisera. It was hypothesized, therefore, that the strains likely produced a non-type I HLT.
In this study, structural, nucleotide sequence, and functional experiments confirmed that one of these strains of E. coli expressed LT-IIc, a new type II HLT. It is likely, therefore, that additional members of the type II HLT subfamily are circulating in the environment.
MATERIALS AND METHODS
Bacterial strains and culture media.Four isolates of enterotoxigenic E. coli (OS-1, OS-2, OS-3, and OS-4), which were isolated from ostriches suffering from diarrhea, were a gift from Tomomasa Yano (Universidade Estadual de Campinas, São Paulo, Brazil). Enterotoxigenic E. coli strains expressing LT-IIa (SA53) and LT-IIb (Ec41) were obtained from the laboratory of Randall K. Holmes (11, 27). All strains were maintained on LB agar or LB broth without antibiotics, unless stated otherwise.
Restriction enzymes.All restriction and DNA-modifying enzymes were purchased from Fermentas, Inc. (Glen Burnie, MD).
Southern hybridizations.Genomic DNA (gDNA) was isolated from strains of enterotoxigenic E. coli by a cetyltrimethylammonium bromide (CTAB) method (37). gDNA (25 μg) was digested overnight with HindIII, DNA fragments were resolved on agarose gels, the fragments were transferred to Nytran (Whatmann, Piscataway, NJ), and the filters were hybridized with 32P-labeled DNA sequences encoding LT-IIa holotoxin (PCR amplified from pHN4 , with synthetic oligonucleotides 5′-CTGTGTTTAAGTTTTAATAT-3′ and 5′-TGACTCTCTATCTAATTCCA-3′), LT-IIb holotoxin (PCR amplified from pHN1  with the synthetic oligonucleotides 5′-CGGGATCCATGCTCAGGTGAG-3′ and 5′-TTCTGCCTCTAACTCGA-3′), the B polypeptide of LT-IIa (gel-purified SacI/BcuI fragment of pHN15 [14, 23]), or the B polypeptide of LT-IIb (gel-purified HindIII/BcuI fragment of pHN16 ). Hybridizations were performed at 60°C in 0.1× SSPE (1× SSPE is 0.18 M NaCl, 10 mM NaH2PO4, and 1 mM EDTA [pH 7.7]) to produce moderately high-stringency conditions (30).
Serotyping.Serotyping of the ostrich isolates, SA53, and Ec41 was performed by the Gastroenteric Disease Center, Pennsylvania State University.
Cloning of genes encoding LT-IIc.To clone the genes encoding the nucleotide sequences of the OS-1 enterotoxin, gDNA isolated from the bacterium was double digested with HindIII and EcoRI, the fragments were ligated into pBluescript II SK− (Stratagene, La Jolla, CA), and the ligation mixture was transformed into E. coli DH5αF′Kan. Colony blot hybridizations using a 32P-labeled nucleotide sequence of the LT-IIb operon was used to identify hybridization-positive clones. Sequence analysis of a recombinant plasmid (pRAY1) purified from a positive clone revealed a truncated OS-1 enterotoxin locus due to an EcoRI restriction site located at the 3′ end of the gene encoding the A polypeptide of LT-IIc. To clone the entire LT-IIc locus, DNA fragments obtained by digesting OS-1 gDNA with HincII were resolved in a 1% agarose gel. Fragments encompassing a range of 3.5 to 5 kbp were excised from the gel and ligated into pJAZZ-OC with the BigEasy v2.0 linear cloning kit (Lucigen Corporation, Middleton, WI). Colonies were screened by colony blot hybridization (31) using a 32P-labeled EcoRI/HindIII fragment of pRAY1 that encoded essentially the A polypeptide of the OS-1 enterotoxin. Nucleotide sequencing of the insert of a plasmid (pJCH1) from a positive clone confirming the successful cloning of the entire LT-IIc operon.
Cloning and purification of LT-IIc.To facilitate purification of LT-IIc, a His-tagged version of LT-IIc was engineered by PCR using pJCH1 as a template and the synthetic oligonucleotides 5′-GGATCCAAGGAGATATACATATGATTAAGCATGTATTGTTGTTTTTT-3′ (BamHI site underlined) and 5′-CTCGAGTTAGTGGTGGTGGTGGTGGTGTGGTGCTAATTCAATTGCC-3′ (XhoI site underlined and with His6 codons double underlined) as primers for the reaction. The PCR conditions employed were 95°C for 45 s, 51°C for 45 s, and 72°C for 2 min for 30 cycles. After digestion with BamHI and XhoI, the PCR fragment was ligated into pBluescript II SK− (Stratagene) to produce pJCH6.2, which was introduced into E. coli DH5αF′Kan (Life Technologies, Inc., Gaithersburg, MD).
Expression of recombinant LT-IIc holotoxin by DH5αF′Kan(pJCH6.2) was induced by addition of 1.0 mM isopropyl-β-d-thiogalactoside to mid-log-phase cultures. After 6 h of induction, recombinant enterotoxin was extracted from the periplasmic space and purified to homogeneity using a combination of nickel affinity and gel filtration chromatography (Sephacryl-100; Pharmacia, Piscataway, NJ) (23). Recombinant enterotoxin was examined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis to confirm that each recombinant protein was purified to apparent homogeneity.
Rabbit anti-LT-IIc hyperimmune antiserum.Rabbit polyclonal anti-LT-IIc hyperimmune antiserum was commercially produced using purified LT-IIc as an immunogen (LAMPIRE Biological Laboratories, Inc., Pipersville, PA). The antiserum exhibited strong immunoreactivity to LT-IIc.
Comparisons of immunological cross-reactivity.Cross-reactivity of polyclonal antiserum to each of the enterotoxins (LT-IIc, LT-IIa, LT-IIb, and CT) was measured by enzyme-linked immunosorbent assay (ELISA). Polystyrene microtiter plates (96-well; Nunc, Roskilde, Denmark) were coated overnight at 4°C with CT, LT-IIa, LT-IIb, or LT-IIc (1 μg/ml). After washing to remove unbound enterotoxin, plates were blocked for 2 h at room temperature (RT) with 1% bovine serum albumin (BSA)-phosphate-buffered saline (PBS) before addition of 100 μl of a diluted (1:5,000) antiserum to the appropriate wells. After overnight incubation at 4°C, plates were washed with PBS. One hundred microliters of alkaline phosphatase-conjugated goat anti-rabbit IgG (Southern Biotechnology Associates, Birmingham, AL) (diluted 1:3,000 in PBS plus 1% BSA) was added to each well, and the plates were incubated for 4 h at RT. Plates were developed by adding 100 μl to each well of 1.0 mg/ml of nitrophenyl phosphate (Amresco, Solon, OH) diluted in diethanolamine buffer (100 ml of diethanolamine, 1 mM MgCl2, and sufficient deionized H2O to bring the volume to 1 liter; pH 9.8). Color reactions were terminated by the addition to each well of 100 μl of a 2.0 M solution of NaOH. The optical density at 405 nm (OD405) of each reaction was measured with a Versamax microplate reader (Molecular Devices, Sunnyvale, CA).
Toxicity bioassay.The toxicity of purified enterotoxins was measured using a standard Y1 adrenal cell (ATCC #CCL-79) bioassay. Y1 cells are acutely sensitive to HLTs and respond by morphological changes in the cell, which are visible by phase-contrast microscopy (23, 24). Briefly, 105 mouse Y1 adrenal cells were cultured to 75% confluence in 96-well tissue culture plates in F-12 medium supplemented with 30% horse serum and 10% fetal bovine serum at 37°C and in an atmosphere of 5% CO2. One microgram per well of LT-IIc, LT-IIa, LT-IIb, or CT was added to the Y1 cell cultures and diluted in a 2-fold dilution series across the 12-well rows of the plate. Plates were incubated at 37°C in an atmosphere of 5% CO2 and examined after 4 h to monitor for rounding of the cells, which is an indicator of toxicity. One unit of toxicity was defined as the smallest concentration of enterotoxin that induced rounding of 75 to 100% of the cultured mouse Y1 adrenal cells (23, 24).
Detection of cAMP.To measure intracellular accumulation of cAMP, peritoneal macrophage cells (25) were treated for 4 h with 1 μg/ml of LT-IIa, LT-IIb, or LT-IIc. Untreated cells were employed as negative controls. After incubation, cells were extracted for 20 min at RT with 200 μl of 0.1 M HCl, scraped from the culture wells, and centrifuged to clear the extracts of cell debris. Levels of cAMP in the extracts were measured with a cAMP enzyme immunoassay kit (Cayman Chemical Co., Ann Arbor, MI).
Neutralization assay.Different concentrations of LT-IIc, LT-IIa, LT-IIb, and CT were incubated for 2 h in 100 μl of F-12-supplemented medium containing 1 μl of rabbit anti-CT, anti-LT-IIa, anti-LT-IIb, or anti-LT-IIc antiserum. After incubation, the solutions were transferred to Y1 cell monolayers for toxicity determinations. The relative unit of toxicity for each enterotoxin was calculated as defined above.
Ganglioside-dependent ELISA.Binding of LT-IIc, LT-IIa, and LT-IIb to commercially available gangliosides (Matreya, State College, PA) was measured by ELISA (24).
Genetic screening for type II genes in ostrich E. coli isolates.Strains expressing LT-I enterotoxins have been isolated from human, pigs, chickens, and turkeys (7, 22, 26, 34). To better define the enterotoxins producing disease in commercial ostriches, Nardi et al. (21) reported isolation of four strains of E. coli from the feces of diarrheic ostriches. These strains appeared to be enterotoxigenic. Using antigenic neutralization assays and PCR-based genetic screens, Nardi et al. (21) concluded that these strains likely produced a type II HLT. While Nardi's results were consistent with that hypothesis, the data were not of sufficient resolution to determine if those strains encoded either LT-IIa or LT-IIb. To begin to resolve that question, genomic DNAs from the four strains were analyzed by high-stringency Southern hybridizations. Nucleotide sequences encoding the A polypeptide genes of LT-IIa and LT-IIb are ∼74% identical; the genes encoding the B polypeptides, however, are less conserved (∼64% identity) (35). Moderately high-stringency hybridizations revealed that all four ostrich strains (OS-1, OS-2, OS-3, and OS-4) exhibited strong homology to LT-IIa (Fig. 1 A) but only weak homology to LT-IIb (Fig. 1B). Hybridizations were also performed with DNA fragments as probes which encoded only the B polypeptides of LT-IIa (LT-IIaB) or LT-IIb (LT-IIbB). Notably, none of the ostrich strains hybridized at high stringency to the B polypeptide-specific probes (Fig. 1C and D). These results indicated that the nucleotide sequences in the ostrich strains likely had high homology to nucleotide sequences encoding the more conserved A polypeptides of LT-IIa and LT-IIb but had low or negligible homology to nucleotide sequences encoding the less-conserved B polypeptides of the two enterotoxins. No hybridizations to the genomic DNAs of the ostrich strains were evident when probes specific for LT-I were employed (data not shown).
These results were consistent with a model in which the ostrich isolates expressed a new type II HLT which was genetically distinguishable from LT-IIa and LT-IIb. To confirm that hypothesis, it was decided to clone the chromosomal segment having homology to LT-IIa and LT-IIb from one of the ostrich strains.
Virulence factors of OS-1.To determine which of the strains to use for cloning, a general analysis was performed by the E. coli Reference Center at The Pennsylvania State University to evaluate any differences between the four ostrich strains. All four strains exhibited the same phenotypes. The original serotyping of the four ETEC isolates from ostriches had been limited to the flagellar (H) and lipopolysaccharide (O) antigens. One of the ostrich strains, OS-1, was identified as an H8:O15 strain (21). Genotypic screening by the E coli Reference Center confirmed that OS-1 was an H8 serotype (Table 1). H8 serotypes have rarely been isolated from birds, but they have been isolated from humans, sheep, and goats (2). In contrast to the original report (21), however, no reactivity to an O15 antigen was detected in OS-1. Surprisingly, nucleotide sequences encoding major virulence factors found in ETEC strains that infect humans and other mammals were not detected in OS-1. It is feasible, therefore, that OS-1 is a commensal strain of E. coli that only recently acquired the genes for LT-IIc.
Nucleotide and amino acid sequence comparisons.Colony blot screening of a genomic library constructed with DNA isolated from OS-1 was employed to identify a clone with homology to LT-IIb holotoxin. Nucleotide sequence analysis of the cloned genomic fragment revealed two open reading frames which overlapped by 12 nucleotides, a genetic arrangement which was very similar to the genetic architecture of the LT-IIa and LT-IIb operons (28). Nucleotide sequences were consistent with a typical σ70-dependent promoter harboring a −35 box (5′-TTTAGA-3′) and a −10 box (5′-TTTAATAAT-3′) located 32 bp in a 5′ direction from the initiation codon of the upstream open reading frame. No nucleotide sequences consistent with a typical σ70-dependent promoter were evident immediately upstream of the second open reading frame. These results suggested an operon structure for the two open reading frames.
The upstream open reading frame, comprised of 777 nucleotides, had the coding capacity for a polypeptide of 259 amino acids. Structural analysis of the predicted amino acid sequences of the predicted polypeptide indicated the presence of a signal peptide and a potential signal peptidase II cleavage site located at amino acid position 18, which, after cleavage, would produce a predicted mature polypeptide of 241 amino acids with a molecular mass of 27.31 kDa. The downstream open reading frame, with a length of 363 nucleotides, coded for a prospective polypeptide of 121 amino acids. A signal peptide region was also predicted at the NH2 terminus of this polypeptide. Upon cleavage by signal peptidase II at amino acid position 23, a mature 98-amino-acid mature polypeptide would be produced with a predicted molecular mass of 10.66 kDa. These predicted molecular masses were confirmed by SDS-PAGE after purification of the recombinant LT-IIc (Fig. 2).
By continuing to sequence in the 5′ and 3′ directions from the two open reading frames, additional open reading frames in the cloned DNA fragment of OS-1 were identified. Database comparisons suggested that these additional open reading frames had significant homology to genes encoding lysozyme (33) and a ParB-like protein (4) found in lambdoid bacteriophages. These data suggested that the genes for the new enterotoxin might reside on a lysogenized prophage, similar to the situation observed for the genes encoding cholera toxin in the genome of V. cholerae (36) (Fig. 3).
Homology comparisons.Amino acid sequences of the mature 27.31-kDa polypeptide of OS-1 encoded by the upstream open reading frame were compared with the amino acid sequences of both type I and type II HLTs. Alignments revealed that the 27.31-kDa polypeptide of OS-1 was homologous to the A polypeptides of type II HLTs and to the A polypeptides of type I HLTs (Table 2). Amino acids Arg7, Asp9, His44, Ser61, Val97, Tyr104, Pro106, His107, Glu110, and Glu112, which are crucial for the ADP-ribosylating activity of CT and LT-I (35) and LT-IIb (19), were identified in the 27.31-kDa polypeptide, albeit shifted by two amino acid positions (Fig. 4). In CT, LT-I, LT-IIa, and LT-IIb, a (K/R)DEL signal motif at the carboxyl terminus of the A polypeptides which mediates retrograde trafficking of the HLT to the endoplasmic reticulum (ER) (13) was also conserved at the carboxyl terminus of the OS-1 27.31-kDa polypeptide, suggesting that the 27.31-kDa polypeptide also has the capacity to retrograde traffic to the ER.
Amino acid sequence comparisons also indicated that the 10.6-kDa polypeptide of OS-1 exhibited greater than 50% identity to the B polypeptides of the type II HLT but little or no similarity to the B polypeptides of the type I HLT (CT or LT-I) (Table 2). Amino acids Thr13, Thr14, and Trp92, but not Thr34 of the mature B polypeptide of LT-IIa and LT-IIb, which are required for binding of those HLT to their respective ganglioside receptors (5, 6, 32, 35), were also conserved in the 10.66-kDa polypeptide of OS-1 (Fig. 5). In addition, Met69, Ala70, Leu73, and Ser74, which are components of a motif [N2-AMAA(I/V)LS-COOH] that mediates binding of the B polypeptides of LT-IIa and LT-IIb to Toll-like receptor 2 (TLR2) (18), were conserved in the 10.66-kDa polypeptide of OS-1. This motif is absent in the B polypeptides of type I HLT (e.g., CT and LT-I) (18).
From these genetic data, it was determined that the locus cloned from OS-1 likely encoded the new type II HLT. To distinguish the new type II HLT from LT-IIa and LT-IIb, the putative enterotoxin was denoted LT-IIc. To confirm that LT-IIc had the appropriate characteristics, the purified recombinant enterotoxin (Fig. 2) was employed in various assays which are routinely used to characterize type I and type II HLTs.
Immunological cross-reactivity of LT-IIc.The type I and type II HLTs are distinguishable by the failure of the enterotoxins to be recognized by heterologous antisera. Rabbit polyclonal antibodies against CT (or LT-I) only weakly recognize LT-IIa or LT-IIb by reactivity to the more conserved A polypeptide (15). Furthermore, anti-CT antisera (and anti-LT-I antisera) cannot neutralize the cytotoxicity of either LT-IIa or LT-IIb (12). Conversely, antisera against either LT-IIa or LT-IIb will not recognize either CT or LT-I in Western immunoblots or by ELISA, and neither type II HLT antiserum neutralizes the cytotoxic effects of CT or LT-I (11, 12, 21). To further analyze the structural and antigenic characteristics of LT-IIc, the immunological cross-reactivities between LT-IIc, LT-IIa, LT-IIb, and CT were compared.
As expected, each of the HLTs was strongly recognized by its cognate rabbit antiserum (Table 3). However, each of the HLTs, including LT-IIc, was only weakly recognized by the heterologous rabbit antisera. This pattern of reactivity was mirrored in ELISAs using mouse antitoxin antisera. Mouse polyclonal antiserum against CT and mouse polyclonal antiserum against LT-IIb cross-reacted with LT-IIc. In contrast, LT-IIc was not recognized by mouse antisera against LT-IIa. These results suggested that LT-IIc shares some limited antigenic similarity to LT-IIa, LT-IIb, and CT, a point which is consistent with the limited conservation in amino acid homologies between LT-IIc and the various type I and type II HLTs.
Cytotoxicity of LT-IIc.To compare the cytotoxic activity of purified LT-IIc to those of CT, LT-IIa, and LT-IIb, a highly sensitive Y1 cell bioassay was employed (23, 24). In comparison to other HLTs, LT-IIc exhibited the lowest cytotoxic activity. To elicit the same cytotoxic effect on Y1 cells as a unit of LT-IIc required 32-fold the amount of CT, 4-fold the amount of LT-IIa, or 2-fold the amount of LT-IIb (Table 4). Cytotoxic activity of LT-IIc was partially neutralized by rabbit anti-LT-IIa and rabbit anti-LT-IIb, but not by anti-CT (Table 4).
Type I and type II HLTs promulgate their cytotoxic activities by ADP-ribosylating the Gsα regulatory protein of the adenylate cyclase sytem. This ribosylation constitutively activates adenylate cyclase in the cell, thus dramatically elevating the intracellular concentrations of cAMP. To determine whether LT-IIc exhibited this property, peritoneal macrophages from naïve mice were treated with LT-IIc and measured for production of cAMP. Concentrations of cAMP in the LT-IIc-treated cells increased by 4.9-fold over the concentration of cAMP in untreated cells (Fig. 6), a level similar to the levels of cAMP produced by intoxicating the cells with either LT-IIa (5.26-fold) or LT-IIb (5.08-fold).
Ganglioside binding activity of LT-IIc.Binding of LT-IIa, LT-IIb, and CT to different cell types is mediated by the enterotoxins' capacities to interact and avidly bind to different gangliosides (1, 9). Each of the HLTs binds to a different array of gangliosides (5, 6). LT-IIa binds specifically, in descending order of avidity, to gangliosides GD1b, GM1, GT1b, GQ1b, GM2, GD1a, and GM3 (9). LT-IIb binds most avidly only to GD1a and, at much lower activities, to GM2 and GM3 (9). CT binds tightly to ganglioside GM1 and has some limited binding activity for GM2, GD1a, GM3, GT1b, GD1b, and asialo-GM1 (17).
Ganglioside-specific ELISA (23, 24) using commercially available gangliosides demonstrated that LT-IIc bound to GM1, GM2, GM3, and GD1a (Fig. 7). Yet, LT-IIc lacked detectable binding activity for GD1b, GQ1b, and GT1b, three gangliosides which are avidly bound by LT-IIa. LT-IIc bound to GM1, a ganglioside also bound by LT-IIa, but not by LT-IIb. These data confirmed that the pattern of binding of LT-IIc to gangliosides is distinguishable from the binding patterns observed for LT-IIa, LT-IIb, and CT.
ETEC is a major health burden in developing countries, where infections by ETEC are the second most frequent cause of mortality in children and infants living in those countries (16). In the developed world, ETEC is one of the major etiological agents for traveler's diarrhea (8). Following the initial discovery of ETEC in humans, there was an intensive effort to further characterize the mechanisms of pathogenesis and to find the means to identify different strains of ETEC strains and the enterotoxins expressed by those strains. In most cases, ETEC strains causing human disease encoded LT-I, ST, or both enterotoxins. To date, most type II HLTs have been isolated from domestic animals (29). Although several reports indicated that type II HLTs may be expressed by human strains of ETEC (11, 29), few efforts have been expended to rigorously screen for type II HLTs in human isolates. Nucleotide sequencing, however, revealed that the operons encoding LT-IIa and the new type II HLT described herein are flanked by genes having significant homology to genes usually associated with lysogenic bacteriophage. This observation raises the possibility that both operons may have the capacity to horizontally transfer to other strains of E. coli or to other enteric species by lysogeny, thus enabling these two type II HLTs to be expressed in pathogens which colonize humans or other hosts. Experiments are under way to determine if the prophage segment in OS-1 and SA53 which encodes LT-IIa and the new type II HLT is capable of excision and of lysogenizing other strains of E. coli and if this prophage system is conserved in all strains of E. coli which produce type II HLTs.
OS-1 encodes a type II HLT which exhibits characteristics distinguishable from either LT-IIa or LT-IIb. Analysis of the predicted amino acid sequences of the three HLTs confirmed that the new HLT is a member of the type II subfamily. Minor amino acid sequence divergence is observed between the A polypeptides of the new HLT, LT-IIa and LT-IIb, the type II HLT, and CT and LT-I, the two members of the type I HLTs. Limitations in the range of changes which would be tolerable for maintaining ADP-ribosylation likely constrain significant divergence in the A polypeptides. Homologies significantly decrease when the amino acid sequences of the five HLTs are compared. The B polypeptide of the new HLT is more similar to the B polypeptides of LT-IIa and LT-IIb than to the B polypeptides of CT and LT-I. Yet, sufficient differences occur between the amino acid sequences of LT-IIa and LT-IIb and the new HLT to engender different immunoreactivities to the respective antisera. These data support the decision to denote the new HLT as LT-IIc, a third member of the type II HLT subfamily.
Divergence in the amino acid sequences in the B polypeptides of LT-IIa, LT-IIb, and LT-IIc was driven by evolutionary forces that required substitutions in the amino acid sequences of the B polypeptides to enable the three HLTs to bind to particular sets of ganglioside receptors. ELISA data demonstrated that LT-IIa, LT-IIb, and LT-IIc have distinctive patterns of binding to commercial gangliosides. Different ganglioside-binding patterns likely enable the three HLTs to intoxicate cells of different hosts or to intoxicate different types of cells within a single host. For example, LT-IIc has a lower cytotoxic activity for mouse Y1 adrenal cells than do LT-IIa, and LT-IIb. This lower toxicity could be explained by the possibility that LT-IIc does not bind (or binds less well) to the gangliosides that are required in mouse cells to promote maximal cytotoxicity. Rather, the lower cytotoxic activity of LT-IIc in mice likely reflects the changes in the B polypeptide that enable this HLT to optimally intoxicate nonmammalian (avian?) cells. Currently, experiments using gangliosides isolated from human, mouse, and avian cells are being designed to define the unique binding patterns of the three type II HLTs at higher resolution and with more relevant substrates.
The discovery of new type II HLT has been hampered by the lack of high-throughput methods to screen for strains of E. coli encoding the HLT which would not cross-react with type II HLT. Serological methods are not useful since none of the three type II HLTs are recognized by the heterologous antisera. Hybridization techniques using low-stringency conditions and fragments of LT-IIa, LT-IIb, and LT-IIc as probes might have utility. However, low-stringency hybridization methods would fail to recognize variants which had significantly higher divergence in nucleotide sequences than those observed between LT-IIa, LT-IIb, and LT-IIc. By using short nucleotide sequences which are highly conserved in the three type II HLTs, however, it should be feasible to design oligonucleotide primers for use in PCR to identify strains encoding new type II HLTs. For example, the TLR2-interacting domain in the gene encoding the B polypeptide is reasonably well conserved in LT-IIa, LT-IIb, and LT-IIc and would likely be conserved in new type II HLTs. The regions encoding amino acids 37 to 47 are highly conserved in the A polypeptides of the three type II HLTs, suggesting potentially some constraints on the structures involved in catalytic activity. Using such a universal screening method, it will be interesting to determine if type II HLTs other than LT-IIa, LT-IIb, and LT-IIc are circulating in the environment and if any of the strains encoding type II HLT have the requisite determinants for infecting humans. In addition, universal screening methods will be crucial for determining the prevalence of LT-IIa, LT-IIb, LT-IIc, and other type II HLTs in nature.
We thank Tomomasa Yano for providing us with the ostrich isolates.
This work was supported by National Institutes of Health research grants DE013833 and DE014357 (T.D.C.).
- Received 7 July 2010.
- Returned for modification 31 July 2010.
- Accepted 5 August 2010.
- ↵*Corresponding author. Mailing address: Department of Microbiology & Immunology, School of Medicine and Biomedical Sciences, The University at Buffalo, 138 Farber Hall, 3435 Main St., Buffalo, NY 14214. Phone: (716) 829-3364. Fax: (716) 829-2158. E-mail:
↵▿ Published ahead of print on 16 August 2010.
Editor: S. M. Payne
- American Society for Microbiology