Heat-Labile Enterotoxin Promotes Escherichia coli Adherence to Intestinal Epithelial Cells

Abstract

Given recent evidence suggesting that the heat-labile enterotoxin (LT) provides a colonization advantage for enterotoxigenic Escherichia coli (ETEC) in vivo, we hypothesized that LT preconditions the host intestinal epithelium for ETEC adherence. To test this hypothesis, we used an in vitro model of ETEC adherence to examine the role of LT in promoting bacterium-host interactions. We present data demonstrating that elaboration of LT promotes a significant increase in E. coli adherence. This phenotype is primarily dependent on the inherent ADP-ribosylation activity of this toxin, with a secondary role observed for the receptor-binding LT-B subunit. Rp-3′,5′-cyclic AMP (cAMP), an inhibitor of protein kinase A, was sufficient to abrogate LT's ability to promote subsequent bacterial adherence. Increased adherence was not due to changes in the surface expression of the host receptor for the K88ac adhesin. Evidence is also presented for a role for bacterial sensing of host-derived cAMP in promoting adherence to host cells.

Enterotoxigenic Escherichia coli (ETEC) is an important cause of human and porcine morbidity and mortality (5). ETEC is the principal etiology of travelers' diarrhea (36) and a leading cause of mortality due to infectious diarrhea in young children in developing nations (39). Similarly, porcine ETEC causes severe intestinal disease in pigs, manifested clinically by a frequently fatal secretory diarrhea (29). Despite the global prevalence and importance of these enteric pathogens, the full complement of virulence factors responsible for intestinal colonization, a critical step in diarrheal pathogenesis, remains undetermined.

ETEC expresses one or more enterotoxins, including the heat-labile (LT) and heat-stable (ST) enterotoxins (30). The enzymatic activities of these enterotoxins cause diarrhea by inducing water and electrolyte loss from the intestines of infected subjects (30). LT is a heterohexameric A-B subunit toxin, structurally and functionally similar to cholera toxin (CT), in which the homopentameric B subunit is responsible for binding to the host GM1 ganglioside receptors (42). The A subunit possesses an ADP-ribosylation activity that covalently modifies the α subunit of the Gs GTP-binding protein, resulting in constitutive activation of adenylate cyclase and production of 3′,5′-cyclic AMP (cAMP). Intracellular increases in cAMP lead to activation of the cAMP-dependent protein kinase A (PKA), which phosphorylates the R domain of the cystic fibrosis transmembrane conductance regulator. Ensuing chloride and water efflux into the intestinal lumen leads to significant volumes of watery diarrhea.

Given that cAMP is an important secondary messenger in numerous essential host pathways, it is intriguing to speculate that, in addition to their proven roles in secretory diarrhea, bacterial enterotoxins may also provide a competitive advantage by modulating host gene expression or protein function in target intestinal epithelial cells. Recent in vivo experiments with both human and porcine ETEC isolates are consistent with this hypothesis. Preliminary characterization of a mouse model of ETEC infection indicated that production of LT provides an advantage in colonization of the murine small intestine (1). A porcine ETEC strain deficient in LT expression colonized piglets less efficiently than either the parent strain or a strain in which the mutation was complemented (5). Seminal studies of mucosal colonization of rabbits by Vibrio cholerae strains possessing or lacking CT subunits established that toxigenic strains colonized more efficiently than nontoxigenic mutants, independent of the receptor-binding CT-B subunit (31). Additionally, coadministration of CT with V. cholerae complemented the colonization ability of the nontoxigenic mutant (31).

Taken together, these data suggest both that expression of LT plays a role in bacterial virulence beyond provoking diarrhea and that modulating the host intestinal epithelium may provide a competitive advantage for these organisms in early stages of adherence to host cells. The in vitro studies described here were designed to examine this hypothesis in more detail.

MATERIALS AND METHODS

Oligonucleotides, bacterial strains, and plasmids.

The oligonucleotides and bacterial plasmids and strains used in these studies are given in Tables 1 and 2, respectively.

TABLE 1.

Oligonucleotides

Oligonucleotide	DNA sequence (5′→3′)
eltAB-f-EcoRV	CG2AT2GTCT2CT2GTATGAT
eltAB-r-EcoRV	GATCG2TAT2GC2TC2TCTAC
eltA-f-BamHI	CTCTG2ATC2A2TG2CGACAGAT2ATAC2GTGC
eltA-r-XhoI	CTC3TCGAGCT3AT2CATA2T2CATC3GA2T2CTG
eltB-f-BamHI	CGCG3ATC2GA2TGA2TA3GTA4TGT2ATGT4AT3ACG2
eltB-r-BamHI	ATATG2ATC2TC2TG2TC2GT5CATACTGAT2GC2GC
A72R-f	GA2GTGCTCACT2ACGTG2ACAGTCTATAT2ATCAG2
A72R-r	C2TGATA2TATAGACTGTC2ACGTA2GTGAGCACT2C
R192G-f	G2T2GTG2A3T2CATCAG2TACA2TCACAG2TG
R192G-r	CAC2TGTGAT2GTAC2TGATGA2T3C2ACA2C2
STa-f-BamHI	G2ATC2AG2AG3TATAT2ATGA5GCTA2TGT2G2C
STa-r-XhoI	CTCGAGATA2CATC2AGCACAG2CAG2
STb-f-BamHI	G2ATC2AG2AG3TATAT2ATGA5GCTA2TGT2G2C
STb-r-XhoI	CTCGAG2CATC2T4GCTGCA2C2
K88ac-CAT-f-BamHI	G2ATC2G2CACTCAGTGC2AG2CAGCAG2
K88ac-CAT-r-HindIII	A2GCT2CGAC2GA2C2AT2GA3TCAC

TABLE 2.

Bacterial plasmids and strains

Strain	Relevant genotype or phenotype	Source or reference
Wild-type ETEC 2534-86	WAM2317 O8:K87:NM:F4ac LT-I+ STb+	5
2534-86ΔeltAB(ΔeltAB171-1081)	ΔeltAB mutant lacking 692 bp of eltA and 220 bp of eltB in WAM2317	5
2534-86ΔeltAB/pLT	Complemented version of 2534-86 ΔeltAB containing an eltAB fragment in pMUN287	5
2534-86/pKK232-8	Promoterless CAT plasmid	37
2534-86/pK88ac(−400-+100)-CAT	Regulatory region upstream of faeBCDEFGHIJ-CAT fusion	This study; 4
ETEC H10407	078:H11 CFA/I LT+ ST+	ATCC
ETEC H10407ΔeltA	ΔeltA mutant of H10407	13
ETEC 3030-2	K88ac LT STb	18
ETEC B41M	F41 K99 STa	M. Zhao; 28
E. coli C600	E. coli K-12 derivative containing pEWD299	10
E. coli G58-1	O101:K28:NM	18
E. coli G58-1/pBR322	Empty cloning vector	This study
E. coli G58-1/pLT	LT holotoxin in pBR322	This study
E. coli G58-1/pLT-A	LT-A subunit in pBR322	This study
E. coli G58-1/pLT-B	LT-B subunit in pBR322	This study
E. coli G58-1/pLT(R192G)	LT (R192G) site-directed mutant of pBR322	This study; 12
E. coli G58-1/pLT(A72R)	LT (A72R) site-directed mutant of pBR322	This study; 19
E. coli G58-1/pSTa	STa in pBR322	This study
E. coli G58-1/pSTb	STb in pBR322	This study
E. coli G58-1/pK88ac	K88ac fimbriae in pBR322	2
ETEC 91.1283	O6:STa+ LT+	C. DebRoy
ETEC 91.1033	O27:STa+ LT−	C. DebRoy
ETEC 91.1626	O6:STa+ LT+	C. DebRoy
ETEC 91.1261	O27:STa+ LT−	C. DebRoy
ETEC 91.1614	O27:STa+ LT−	C. DebRoy

Chemicals.

All chemicals were obtained from Sigma-Aldrich unless otherwise indicated. GM1, cAMP, and Rp-cAMP were applied to host cells in their sodium salt, monophosphate Tris salt, and triethylammonium salt hydrate forms, respectively.

Molecular biology.

The genes encoding the enterotoxin subunits LT-AB (eltAB), LT-A (eltA), LT-B (eltB), and STb (estB) were amplified from ETEC 2534-86 genomic DNA, and the estA gene, encoding STa, was amplified from ETEC B41M genomic DNA by PCR using the primers eltAB-f/eltAB-r, eltAB-f/eltA-r, eltB-f/eltAB-r, STb-f/STa-r, and STa-f/STa-r, respectively (Table 1) and cloned into complementary restriction sites in pBR322. These PCR products lack the native gene promoters. The fae regulatory region upstream of the K88 operon (23) was amplified from ETEC 3030-2 genomic DNA by PCR using the primers K88ac-CAT-f/K88ac-CAT-r and cloned into the BamHI-HindIII sites of pKK232-8, a vector containing a promoterless chloramphenicol acetyltransferase (CAT) gene (37). Overlap extension-PCR (46) was used to construct LT A72R and R192G site-directed mutants, using primers A72R-f/A72R-r and R192G-f/R192G-r, respectively. Restriction digestion and DNA sequencing using an ABI BigDye Terminator cycle sequencing ready reaction kit and an ABI3730 DNA analyzer (Nevada Genomics Center) were used to confirm plasmid constructs.

Mammalian cell culture.

IPEC-J2 cells were undifferentiated porcine intestinal epithelial cells derived from the small intestines of day-old piglets (32) and were maintained in a humidified incubator in an atmosphere of 5% CO₂ at 37°C and cultured in Dulbecco's modified Eagle's medium (DMEM)-F-12 medium supplemented with 5% fetal bovine serum (FBS; Atlanta Biologicals), insulin (5 μg/ml), transferrin (5 μg/ml), selenium (5 ng/ml), and epidermal growth factor (5 ng/ml). Caco-2 cells (human colorectal adenocarcinoma cells [27]) were cultured in DMEM supplemented with 10% FBS supplemented with 1% nonessential amino acids.

Quantitative bacterial adherence assays.

IPEC-J2 and Caco-2 cells were seeded in 24-well plates (∼1.0 × 10⁴ cells/well) and grown to >90% confluence. Bacteria were grown overnight at 37°C in 3 ml of LB supplemented with appropriate antibiotics in 17- by 100-mm glass culture tubes, with shaking at 200 rpm; diluted 1:100; grown for 2 h; and then added to wells of the tissue culture plates, without centrifugation, in triplicate at a multiplicity of infection of approximately 10:1. After incubation for various times, samples were processed for enumeration of adherent bacteria by washing them three times with phosphate-buffered saline (PBS) (5 min/wash) to remove nonadherent bacteria, treated with 0.25% trypsin (37°C for 5 min) to dislodge mammalian cells from tissue culture plates (11), centrifuged (1,000 × g for 5 min), resuspended in 1 ml PBS, serially diluted, plated on LB, and incubated overnight at 37°C. The number of CFU was measured, and data were normalized against minor differences in bacterial inocula. The growth rates among bacterial strains were not significantly different. Control experiments were performed with tissue culture plates lacking mammalian cells to verify that bacteria were cell associated and did not adhere nonspecifically to the plastic.

For experiments involving pretreatment of ETEC prior to adherence assays, bacteria were grown overnight, subcultured 1:20, and grown for 2 h in the presence of 20 μM adenosine or 50 μM cAMP.

For experiments involving pretreatment of IPEC-J2 cells prior to ETEC infection, donor cells were pre- or cotreated with 100 μM 2′5′-dideoxyadenosine (DDA; BIOMOL International), 1 μM guanylin (9), 20 μM adenosine, 1 μg/ml pertussis toxin (PT), 100 nM ATP, 50 μM cAMP, 200 μM Rp-cAMP, or 100 ng/ml LT for 1 h. In some experiments, cell supernatants obtained after infection or incubation with toxins were removed and passaged through a 0.2-μm filter before dilution (1:100) into naive bacterial or mammalian cell cultures.

Purification of LT.

LT was extracted from E. coli C600, a derivative of E. coli K-12 containing the EWD299 plasmid (10), and purified by one-step chromatography with an immobilized d-galactose column as previously described (44). Fractions containing LT, as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and an enzyme-linked immunosorbent assay, were pooled, dialyzed against distilled water, concentrated to 1 μg/ml, flash frozen, and stored at −80°C until use.

LT binding to IPEC-J2 cells.

IPEC-J2 cells (1 × 10⁵) were resuspended in DMEM containing 1 μg/ml LT in the presence or absence of 1 μg/ml GM1 (Fisher). After 1 h of incubation at 37°C, cells were washed three times with PBS, fixed in 4% paraformaldehyde, rewashed, and incubated with 1% goat serum in PBS for 20 min. Fixed cells were incubated with rabbit anti-CT antibody (1:100 in PBS) for 30 min and with goat anti-rabbit Alexa 488 (1:500) for 30 min. LT binding was visualized by immunofluorescence microscopy after cell resuspension in 1% paraformaldehyde.

Purification of outer membrane vesicles (OMVs).

Vesicles were harvested from culture supernatants of bacteria grown overnight in 50 ml LB at 37°C with shaking (150 rpm), as previously described (25). Bacteria were pelleted by centrifugation (10,000 × g for 10 min at 4°C), and the supernatant was decanted and passed through a 0.2-μm filter. Vesicles were collected by ultracentrifugation (150,000 × g for 3 h at 4°C) and resuspended to 1.0 mg/ml in water.

Quantification of cAMP concentration.

The concentrations of cAMP in IPEC-J2 cell lysates and cell supernatants were quantified using a cAMP (direct) enzyme immunoassay kit according to the manufacturer's instructions (Assay Designs).

CAT assays.

CAT assays were performed as described previously (37), using ETEC 2534-86 grown to an optical density at 600 nm of 0.4 to 0.6 in CFA medium (1% Casamino Acids, pH 7.4, 0.08% yeast extract, 0.4 mM MgSO₄, 0.04 mM MnCl₂) supplemented where indicated with 1 mM glucose, 1 mM cAMP, or cell-free supernatants (3%, vol/vol) derived from donor cells treated with 100 ng/ml LT for 1 h or infected with wild-type (wt) or ΔeltAB ETEC 2534-86 for 3 h.

FaeG immunoblots.

K88ac adhesins were extracted from ETEC 2534-86 by using a previously described method (15). Bacteria were collected by centrifugation, resuspended in PBS, heated to 65°C for 30 min, and blended for 2 min in a Waring blender. The supernatant was concentrated by dialysis against solid polyethylene glycol and precipitated by adding 2.5% citric acid to give a final pH of 4.0. The precipitate was resuspended in PBS and analyzed by immunoblotting using an anti-FaeG monoclonal antibody (43).

K88ac receptor distribution.

The surface expression of the intestinal mucin-type sialoglycoprotein (IMTGP) was quantified by measuring binding of biotinylated K88ac adhesin as previously described (15). Streptavidin-fluorescein isothiocyanate fluorescence intensity was measured by flow cytometry, with analysis of at least 20,000 events on a FACScan flow cytometer (Becton Dickinson).

Statistical analysis.

The experiments were performed in triplicate on at least three separate occasions. The results for the bacterial adherence assays were analyzed statistically by calculating the median number of CFU in each treatment group and compared using the Mann-Whitney test to determine significant differences between sample medians. cAMP and CAT assays were analyzed with unpaired Student t tests. P values of <0.05 were considered significant.

RESULTS

Recent in vivo observations suggesting a potential role for bacterial enterotoxins in initial stages of host colonization (1, 5) prompted us to quantify, in vitro, potential contributions of LT in modulating early stages of pathogen-host interactions. We utilized an IPEC-J2 cell culture system to quantify the bacterial adherence levels of porcine ETEC isolates differing in their expression of enterotoxins. IPEC-J2 cells provide a relevant model of pig intestinal enterocytes, as they produce glycoprotein receptors for bacterial adhesins and glycocalyx-bound mucin, form microvilli, and form tight junctions in vitro (26, 38).

LT enhances porcine ETEC adherence to intestinal cells.

To examine the contribution of LT to the interaction of ETEC with intestinal cells, we first quantified the adherence levels of a wt ETEC strain (2534-86) (5), an isogenic mutant deficient in LT expression (ΔeltAB), and the ΔeltAB mutant complemented by plasmid-based eltAB expression (ΔeltAB/pLT). IPEC-J2 cells were infected with ETEC (multiplicity of infection, ∼10) and subsequently processed to enumerate the adherent bacteria. The wt strain was significantly enhanced in adherence relative to the ΔeltAB mutant (Fig. 1A) (P = 0.02; Mann-Whitney test). Complementation of ΔeltAB with pLT restored adherence to near-wt levels (P = 0.82).

FIG. 1.

LT promotes porcine ETEC adherence to IPEC-J2 cells. (A) Quantification of adherence (CFU/ml) to IPEC-J2 cells by ETEC 2534-86 strains possessing or lacking eltAB versus time of infection. (B) Quantification of ETEC 2534-86 adherence to IPEC-J2 cells (CFU/ml) after a 2-h infection in the absence (white bars) or presence of 100 ng/ml LT, either added at 1 h preinfection (gray bars) or coinfected (black bars). The inset depicts electrophoretic separation of purified LT by sodium dodecyl sulfate-polyacrylamide gel electrophoresis in a 15% resolving gel.

Exogenously supplied LT promotes ETEC adherence.

Given our observation that endogenous expression of LT provides an advantage to ETEC adherence to intestinal cells, we hypothesized that exogenously supplied soluble LT might also promote adherence to strains otherwise lacking LT. To test this hypothesis, we either pre- (1 h) or cotreated intestinal cells with 100 ng/ml LT (Fig. 1B, inset) and then infected these cells with ETEC possessing or lacking eltAB. Pretreatment of cells with LT significantly increased subsequent adherence of ΔeltAB (P = 0.03) (Fig. 1B). Cotreatment with LT upon bacterial inoculation also increased adherence of ΔeltAB, but to a lesser extent (P = 0.05). Conversely, addition of exogenous LT did not significantly alter the adherence levels of strains that express eltAB (wt 2534-86 and ΔeltAB/pLT) (Fig. 1B). Similar data were obtained with as little as 15 min of host cell pretreatment (data not shown). We verified by immunofluorescence microscopy that purified LT bound to the surfaces of IPEC-J2 cells and that such binding was inhibited by coincubation with GM1 (data not shown).

To determine whether these data were unique to porcine ETEC or general to other ETEC strains, we quantified the adherence levels of the prototypical human ETEC isolate H10407 and an isogenic mutant in which eltA was deleted (ΔeltA) (13) to Caco-2 cells. Again, both pre- and cotreatment of Caco-2 cells with 100 ng/ml LT significantly increased the adherence of the ΔeltA mutant (P = 0.02) (Fig. 2A), whereas supply of exogenous LT to wt H10407 had no effect on adherence (Fig. 2A).

FIG. 2.

LT promotes human ETEC adherence to Caco-2 cells. (A) Quantification of adherence (CFU/ml) to Caco-2 cells by ETEC H10407 possessing or lacking eltA after a 2-h infection in the absence or presence of 100 ng/ml LT, either added at 1 h preinfection (gray bars) or coinfected (black bars). (B) Quantification of adherence (CFU/ml) to Caco-2 cells by ETEC field isolates in the absence (white bars) or presence (gray bars) of 100 ng/ml LT.

We also sought to determine the extent to which the adherence levels of ETEC isolates not routinely passaged in the laboratory would be altered by exogenous LT. We therefore obtained several human diarrheal disease isolates from the Middle East (C. DebRoy, personal communication) and quantified their adherence to Caco-2 cells in the absence or presence of LT. Several ETEC strains (91.1033, 91.1261, and 91.1614) deficient in LT expression displayed increased adherence in the presence of 100 ng/ml LT. In contrast, the magnitude of adherence by LT-positive (LT⁺) strains (91.1283 and 91.1626) did not change significantly (Fig. 2B).

Endogenous expression of enterotoxin subunits in G58-1.

We examined the extent to which the adherence of E. coli G58-1, a porcine E. coli strain isolated from swine feces that lacks any known plasmids or known enterotoxins (17), responded to LT. Addition of exogenous LT during infection of intestinal cells by G58-1 increased adherence in a concentration-dependent manner (Fig. 3A) (P = 0.01).

FIG. 3.

Exogenous and endogenous LTs promote E. coli G58-1 adherence. (A) Quantification of G58-1 adherence (CFU/ml) to IPEC-J2 cells after a 4-h infection as a function of LT concentration. (B) Change in adherence (relative to the level for unmodified G58-1) as a function of endogenous expression of the indicated enterotoxin subunits or fimbrial adhesins.

We evaluated the extent to which endogenous expression of different enterotoxin subunits in G58-1 would promote bacterial adherence. Genes encoding several enterotoxin subunits were amplified by PCR, cloned into pBR322, and expressed in G58-1. Where available, enzyme-linked immunosorbent assays were used to confirm the expression of the relevant enterotoxin subunit (data not shown). The adherence levels of these modified strains were quantified and compared to the level for the parental, unmodified G58-1 strain. The expression levels of genes encoding the enterotoxin subunits LT-A, STa, and STb did not significantly increase G58-1 adherence (Fig. 3B). In contrast, expression of the LT holotoxin significantly increased adherence compared to the parental strain (28.1 ± 9.3-fold; P = 0.003). Expression of the receptor-binding subunit LT-B modestly increased G58-1 adherence (6.3 ± 2.7-fold; P = 0.03).

To determine the extent to which the secretory activity of LT is critical to its ability to enhance bacterial adherence, we also constructed two well-characterized variants of LT, each bearing mutations in the A subunit (A72R [19] and R192G [12]) which render the toxin inactive. In comparison to the increased adherence provided by expression of the LT holotoxin, expression of A72R (3.9 ± 1.6-fold) or R192G (4.3 ± 1.5-fold) promoted G58-1 adherence only to approximately the same magnitude as the LT-B subunit construct, suggesting that ADP-ribosylation activity is necessary to effect changes in bacterial adherence (Fig. 3B).

Elevation of host cAMP.

We evaluated the responsiveness of our in vitro culture system to LT by measuring the concentrations of cAMP secreted into host cell supernatants. Administration of 100 ng/ml LT yielded a 19.7 ± 2.6-fold increase in secreted cAMP levels, in general agreement with other experiments (21), whereas administration of the adenylate cyclase antagonist DDA resulted in a marked decrease in cAMP, as predicted (Fig. 4A).

FIG. 4.

Profiling of G58-1 adherence following stimulation with pharmacological agents. (A) Change in concentration of secreted cAMP ([cAMP]) following infection or intoxication with indicated bacterial strains or chemicals. (B) Change in G58-1 adherence to IPEC-J2 cells after a 4-h infection as a function of coincubation with 1.0 μg/ml OMVs purified from the indicated bacterial strains. (C) Change in G58-1 adherence to IPEC-J2 cells after a 4-h infection following 1 h of pretreatment of IPEC-J2 cells with 50 μM cAMP, 20 μM adenosine, 100 nM ATP, 1 μg/ml PT, or 1 μM guanylin. (D) Change in G58-1 adherence to IPEC-J2 cells after a 4-h infection, with cotreatment of host cells with 100 ng/ml LT, 200 μM Rp-cAMP, or LT plus Rp-cAMP.

Infection of IPEC-J2 cells with wt ETEC increased cAMP secretion to levels (11.7 ± 3.1-fold increase) not significantly different from those induced by LT (Fig. 4A). In contrast, infection with ΔeltAB elevated cAMP secretion only slightly (3.2 ± 1.4-fold), whereas complementation (ΔeltAB/pLT) restored secretion to near-wt levels (14.6 ± 4.7-fold). As expected, these effects were partially abrogated by cotreatment with DDA.

OMVs containing LT promote adherence.

When ETEC is grown to high density in liquid culture, LT is highly enriched in OMVs (22). When LT is added to the surfaces of tissue culture cells, vesicle endocytosis is dependent on cholesterol-rich lipid rafts (25). We determined the extent to which application of OMVs purified from ETEC possessing or lacking eltAB to host cells could reproduce previous results obtained with soluble LT. OMVs (1.0 μg/ml) purified from wt ETEC and ΔeltAB/pLT, but not ΔeltAB, significantly raised the concentrations of cAMP secreted into the cell supernatants. Similar to the effect of adding purified LT, addition of OMVs from wt and ΔeltAB/pLT ETEC promoted increases in G58-1 adherence (Fig. 4B). Conversely, OMVs purified from G58-1 or ΔeltAB had no effect on bacterial adherence.

Rp-cAMP blocks LT adherence promotion.

To determine the extent to which host-derived nucleotides and other bacterial toxins might also promote adherence, we quantified G58-1 adherence as a function of pretreatment of host cells with ATP, adenosine, cAMP, pertussis toxin (PT), and guanylin. These compounds were studied in our assays because they have previously been characterized for their roles in host-pathogen interactions. ETEC toxins potentiate ATP release from cultured T84 cells (7), which is subsequently degraded to adenosine, which can stimulate host Cl⁻ secretion through interaction with adenosine A₂ receptors (8). Intracellular increases in cAMP lead to activation of PKA. PT catalyzes ADP ribosylation of heterotrimeric G proteins to prevent inhibition of adenylate cyclase (6). Guanylin is a peptide secreted by goblet cells that also increases Cl⁻ secretion (9). Under the employed assay conditions, none of these chemicals altered G58-1 adherence to IPEC-J2 cells (Fig. 4C).

In contrast, whereas LT effected a significant increase in G58-1 adherence (Fig. 4D, left), cotreatment with Rp-cAMP, an inhibitor of PKA (35), blocked the ability of LT to increase G58-1 adherence (Fig. 4D). These data suggest that LT's ability to promote adherence may be mediated through the alteration of host signaling via PKA, whose activity is induced by increased cAMP concentrations.

Cell-free supernatants derived from intoxicated IPEC-J2 cells promote adherence.

To begin to determine if a change in host cell physiology induced by LT may be responsible for increased E. coli adherence, we performed experiments in which host cells were either treated with LT or infected with wt or ΔeltAB ETEC 2534-86. After 1 h, cell supernatants were removed, centrifuged, filtered, and then applied to an independent set of naive cells that were then subjected to infection with G58-1. Transfer of cell-free IPEC-J2 cell supernatants previously stimulated with LT or infected with wt ETEC 2534-86 to naive cells significantly increased G58-1, relative to that of supernatants obtained from either uninfected cells or cells infected with ΔeltAB ETEC (Fig. 5A). Cotreatment of donor cells with DDA during original infections or LT treatments abrogated the effect of these conditioned supernatants on target cells (Fig. 5A).

FIG. 5.

Stimulation of G58-1 adherence with cell-free supernatants. (A) Change in G58-1 adherence to IPEC-J2 cells after a 4-h infection as a function of coincubation of naive IPEC-J2 cells with cell-free supernatants (supt) obtained after stimulation of donor cells with the indicated bacterial strains or chemicals. Where indicated, donor cells were cotreated with 100 μM DDA (supt+DDA). (B) Change in G58-1 adherence in the presence of either purified LT with or without GM1 (soluble toxin) or supernatants obtained from LT-stimulated cells with or without GM1 (supernatant). (C) Change in G58-1 (4 h) and ETEC 2534-86 (1 h) strain adherence following pretreatment (2 h) of bacterial inocula with 20 μM adenosine (open bars) or 50 μM cAMP (gray bars). The inset depicts the influence of coincubation with tetracycline on 2534-86 adherence in the presence of cAMP. (D) Differential expression of K88ac-CAT. CAT assays were performed using ETEC 2534-86, grown to an optical density at 600 nm of 0.4 to 0.6, in CFA medium supplemented where indicated with 1 mM glucose, 1 mM cAMP, or cell-free supernatants (3%, vol/vol) derived from donor cells treated with 100 ng/ml LT for 1 h or infected with wt or ΔeltAB ETEC 2534-86 for 3 h. Data are plotted as relative CAT activities versus the bacterial culture additive. (E) FaeG adhesin expression. K88ac adhesins were extracted from ETEC 2534-86 and analyzed by immunoblotting using an anti-FaeG (α-FaeG) monoclonal antibody (43).

To verify that the increased G58-1 adherence observed after incubation with cell-free supernatants was due to a secreted host factor rather than to residual LT potentially transferred to naive cells, we performed control experiments in which supernatants obtained from donor cells treated with LT were incubated with GM1. As expected, while GM1 could neutralize the effect of LT added directly to host cells (Fig. 5B, soluble toxin), GM1 did not significantly reduce the activity of cell-free supernatants from donor cells previously treated with LT (Fig. 5B, supernatants). These data suggest that any LT remaining in transferred cell supernatants is unlikely to be present at a concentration sufficient to account for the observed proadherence phenotype.

Incubation of G58-1 with cAMP promotes adherence to IPEC-J2 cells.

To determine if secreted host factors might be sensed by E. coli to promote physiological changes that promote adherence, we quantified changes in bacterial adherence to intestinal cells following pretreatment of bacterial inocula with cAMP. Pretreatment of G58-1 with 50 μM cAMP (Fig. 5C) significantly increased the ability of this strain to adhere to IPEC-J2 cells (35.0 ± 13.1-fold increase; P = 0.01). In contrast, pretreatment with 20 μM adenosine (Fig. 5C) did not significantly increase adherence (2.2 ± 1.2-fold; P = 0.14). Similar trends, though reduced in magnitude, were also observed with both wt and ΔeltAB ETEC 2534-86. De novo bacterial protein synthesis appeared to be required, as treatment of ETEC 2534-86 with tetracycline reduced the ability of cAMP to promote subsequent adherence (Fig. 5C, inset).

We also interrogated the transcriptional activity of the fae regulatory region upstream of the K88ac operon by constructing a fusion to CAT and performing CAT activity assays with ETEC 2534-86 grown in CFA medium supplemented with glucose, cAMP, or supernatants derived from IPEC-J2 cells intoxicated with LT or infected with wt or ΔeltAB ETEC. Addition of glucose slightly repressed CAT activity, whereas cAMP increased CAT activity approximately 5-fold (Fig. 5D). Transfer of supernatants (3%, vol/vol) from IPEC-J2 cells intoxicated with LT or infected with wt ETEC, but not ΔeltAB ETEC, also increased CAT activity (Fig. 5D) (P = 0.03). Consistent with these data, we also observed increased expression of FaeG, the major K88ac fimbrial subunit (3), following incubation of ETEC 2534-86 with host cell supernatants derived from wt, but not ΔeltAB, ETEC infection (Fig. 5E).

LT does not alter surface localization of the host K88ac receptor.

The host coreceptor nucleolin migrates to the cell surface in response to Stx2 to promote Shiga toxin-producing E. coli adherence (34). We hypothesized that LT might alter the surface localization of the K88ac receptor, a high-molecular-weight IMTGP (15). To test this hypothesis, we treated host cells with OMVs purified from ETEC possessing or lacking LT or with purified LT and subsequently measured the surface expression of IMTGP by using flow cytometry. Binding of purified, biotinylated K88ac fimbriae was detected in 71.2% of untreated IPEC-J2 cells, whereas significant binding was not detected in human Caco-2 cell lines lacking IMTGP (data not shown). However, pretreatment of IPEC-J2 cells with OMVs purified from ETEC possessing or lacking LT or with purified LT did not significantly alter the binding of biotinylated K88ac fimbriae, thus suggesting that K88ac receptor redistribution may not be a significant contributor to the observed ability of LT to promote ETEC adherence. The mean expression levels (relative to that for untreated cells) ± standard deviations for three independent experiments using this method were as follows: for IPEC-J2 cells treated with 2534-86 OMVs, 97.6% ± 4.3%; for those treated with 2534-86 ΔeltAB OMVs, 92.5% ± 7.5%; for those treated with 2534-86 ΔeltAB/pLT OMVs, 93.3% ± 3.2%; and for those treated with 1 μg/ml LT, 96.4% ± 7.7%.

DISCUSSION

Several groups have studied the role of bacterial toxins in promoting bacterial adherence in vitro and host colonization in vivo. The E. coli O157:H7-encoded Shiga-like toxin 2 (Stx2) protein contributes to increased adherence to epithelial cells in vitro, possibly by mediating an increase in surface expression of the nucleolin protein, which acts as a receptor for E. coli outer membrane proteins (34). However, other experiments have indicated that Stx2 plays no significant role in promoting initial adherence to bovine terminal rectal mucosa (40) and that Stx2 has no role in intestinal colonization following intragastric inoculation of rabbits (33). Bordetella pertussis adenylate cyclase toxin mutants suffer a colonization defect compared to wt strains when administered intranasally at low doses in infant mice (20). Toxigenic V. cholerae colonizes rabbits more efficiently than nontoxigenic mutants, independent of the receptor-binding CT-B subunit (31).

Our data suggest that LT promotes the adherence of ETEC to intestinal epithelial cells in vitro. While a role for LT in promoting host colonization in vivo had already been established (1, 31), the molecular basis by which LT promotes ETEC adherence had yet to be investigated. Deletion of eltAB from ETEC 2534-86 and eltA from ETEC H10407 significantly reduced adherence to host cells (Fig. 1A and 2A). These data parallel prior observations that the eltA mutant was markedly defective (∼10-fold) in intestinal colonization relative to wt ETEC (1). Coadministration of LT with ETEC inocula was sufficient to increase adherence of ETEC lacking eltAB to near-wt levels (Fig. 1B and 2A). In our in vitro studies, pretreatment of host cells prior to addition of ETEC was more effective in promoting adherence than coadministration of the toxin, consistent with the previously reported time needed for translocation of CT across the cell membrane before adenylate cyclase activation (30).

Our utilization of an E. coli field isolate (G58-1) lacking characterized enterotoxins permitted clarification of the relative contribution of ADP ribosylation activity to promotion of bacterial adherence. This subject has been of interest following the seminal observations that LT bound to the surface of vesicles can bind host GM1 and bacterial LPS simultaneously (21). Thus, an “LT-induced bridge” (21) might reasonably be expected to be the primary determinant of LT-mediated adherence promotion.

Expression of the LT holotoxin increased G58-1 adherence ∼28-fold, whereas expression of the LT-B subunit increased G58-1 adherence ∼6-fold (Fig. 3B). Two site-directed mutants (A72R [19] and R192G [12]) previously shown to abolish the enzymatic activity of LT failed to promote adherence more than LT-B. These data suggest that the ADP ribosylation activity of LT, and its subsequent influence on host cell physiology, may be the primary contributor to LT's role in bacterial adherence. These data are consistent with earlier findings that association of CT-B with the monosialosylganglioside GM1 was insufficient for promoting V. cholerae colonization (31).

Much of LT's contribution to E. coli adherence appeared dependent upon the ability to mediate an increase in host cAMP concentration. Experiments in which donor host cells were intoxicated and supernatants were transferred to naive cells prior to infection suggest that a factor is secreted into host cell supernatants that subsequently improves bacterial adherence (Fig. 5A). Coincubation of donor cells with DDA to abrogate increases in cAMP levels was sufficient to block supernatant activity. Cotreatment with the PKA inhibitor Rp-cAMP blocked the ability of LT to increase G58-1 adherence (Fig. 4D). While such a PKA-mediated activity might be due to induction of structural alterations in target cells (1), we did not observe a significant change in IMTGP abundance at the plasma membrane, as implied by K88ac fimbria binding assays (see Materials and Methods).

Pretreatment of E. coli with cAMP, but not adenosine, promoted subsequent adherence to host cells (Fig. 5C). Adenosine has been studied in detail in the context of enteropathogenic E. coli (EPEC) infections, in which EPEC-mediated killing of host cells causes release of ATP (8). ATP is subsequently degraded to adenosine, which is able to stimulate a chloride secretory response (8). LT (and STa) had previously been shown to potentiate the release of ATP from EPEC-infected cells (7) and may increase disease severity in mixed EPEC/ETEC infections (7).

Addition of supernatants from IPEC-J2 cells intoxicated with LT significantly increased the activity of an fae(K88)-CAT transcriptional fusion (Fig. 5D). There is significant precedence for the role of cAMP in regulation of bacterial virulence factors. Studies of the ETEC CFA/I (24) and CFA/II (16) adhesins demonstrated that glucose inhibits the promoters of the genes encoding these adhesins and further showed that addition of cAMP could relieve this repression, suggesting regulation by catabolite repression. The cAMP-cAMP receptor protein complex also regulates expression of fasH, a transcriptional activator of 987p fimbriae (14). Expression of the V. cholerae toxin-coregulated pilus (41) and the EPEC bundle-forming pilus are also regulated by carbon status (45). Therefore, carbon gradients in the host intestine were proposed to provide a mechanism for preferential bacterial colonization of different intestinal microenvironments (14). It will be of significant interest to elucidate the molecular mechanisms governing enterotoxin-mediated enhancement of bacterial adherence.