Inhibition of Binding of the AB5-Type Enterotoxins LT-I and Cholera Toxin to Ganglioside GM1 by Galactose-Rich Dietary Components

Abstract

Cholera, travelers' diarrhea, or colibacillosis in pigs can possibly be prevented or attenuated by dietary provision of competitive inhibitors that react with the GM1-binding sites of the enterotoxins cholera toxin (CT), human Escherichia coli heat-labile enterotoxin of serogroup I (LTh-I), and porcine LT-I (LTp-I). The interfering efficiency of natural substances with binding of the toxins to the gangliosid receptor GM1 was tested using a specially adapted GM1-coated-microtiter-well enzyme-linked immunosorbent assay. The substances tested for their GM1 displacing capacity were galactose-containing or -related saccharides from bovine milk, skim milk powder, galactan from gum arabic, food stabilizers as well as ground fenugreek seed and soy bean constituents that contain galactomannans, the galactopolysaccharides agar and agarose, and larch wood and other plant materials that contain arabinogalactans. Skim milk powder, compared with the pure milk saccharides tested, interfered to a higher extent with LTh-I (65–66% inhibition at 5 mg test substance/mL) and CT binding (63–67% inhibition at 5 mg test substance/mL) when supplied before or simultaneously with the toxins in the GM1-enzyme-linked immunosorbent assay. Ground fenugreek seed counteracted GM1 binding of 5 ng LTh-I/mL as well as 5 ng and 1 μg LTp-I/mL (43–65% inhibition at 5 mg test substance/mL), and 4 ng CT/mL (61–92% inhibition at 5 mg test substance/mL) very efficiently when supplied before the toxin–GM1 complex had formed. With 50 mg/mL fenugreek seed, inhibition percentages of even 92–99% were reached for LTh-I and CT binding. Efforts to resolve already bound toxin from GM1 with the test substances were less effective than preincubations and concurrent incubations.

Introduction

I n humans, Escherichia coli heat-labile enterotoxins of serogroup I (LT-I) are the most common causes of traveler's diarrhea worldwide, and are also responsible for the death of hundreds of thousands of children per year in developing countries (Qadri et al., 2005). Likewise, in animal husbandry LT-I is a major cause of severe porcine colibacillosis (Zhang et al., 2006).

E. coli LT-I is very similar to cholera toxin (CT) in amino acid sequence, in mode of action, and in biological properties (Sixma et al., 1993; Hol et al., 1995; Nataro and Kaper, 1998). Both toxins have an AB5 composition, in which the B pentamer has a membrane-binding function and the A subunit exerts enzymatic activity after cleavage (Hol et al., 1995). The five B subunits of LT-I are arranged in a plane donut shape, mediating strong host cell attachment via ganglioside GM1 and, much weaker, via GD1b and some intestinal glycoproteins (Teneberg et al., 1994).

Group I heat-labile enterotoxins of E. coli comprise two closely related antigenic variants, called LTp-I and LTh-I, after their initial discovery in strains isolated from pigs and humans, respectively (Tsuji et al., 1982). Group II contains type II heat-labile enterotoxins, which have been associated with diarrhea, but have not been positively identified as disease-causing agents in animals or humans so far (Van den Akker et al., 1996; Nardi et al., 2005).

The genes for the diarrhea-inducing LT-I are located on plasmids in E. coli (Gyles et al., 1974; Smith, 1977) and are hence transmissible. Therefore, it is not surprising that LT-I production has been associated with enterotoxigenic E. coli (ETEC) carrying different adhesion factors or belonging to different serogroups (Willshaw et al., 1980; Nagy et al., 1999; Do et al., 2006; Khac et al., 2006). As a consequence, counteracting specific E. coli might not be the straightest way toward an inhibition of LT activity in the intestine. Hence, this article focuses on the inhibition of LT-I binding to GM1 to prevent the toxin from causing cell damage. Svennerholm and Holmgren (1978) were the first who described a GM1-enzyme-linked immunosorbent assay (ELISA) procedure for the identification of E. coli LT-I. More recently, Dawson (2005) gave an overview of different adaptations of this ELISA technique in his paper about a GM1-ELISA assay for CT. Based on Dawson's revised concept, we designed our GM1-ELISA assay for LT-I.

On a molecular level, Merritt et al. (1994a) compared the binding mode of GM1 for LT-I and CT with a two-fingered grip. GM1 puts forth a galactose residue as tip of a disaccharide forefinger, and a sialic acid residue as thumb to grasp the toxins (Merritt et al., 1994b; Minke et al., 1999). Of the two terminal sugar residues, the galactose gets more deeply buried into the toxin and is crucial for the toxin–GM1 binding. Hence, the galactose-binding site of LT-I was supposed to offer great promise as a target for blocking receptor binding (Merritt et al., 1994b, 1997; Hol et al., 1995). In view of this potential Achilles' heel, the aim of this study was to determine the interfering efficiency of galactose-rich substances with LT-I and CT binding to GM1. We chose to test natural substances, for it is especially the low-priced inhibitors that might have prospects to reach those who need them most in the developing world or help protect pigs against diarrhea.

Materials and Methods

Toxins and primary antibodies

Purified type I human E. coli heat-labile enterotoxin (LTh-I) was derived from the strain HC22/TP235km (Dertzbaugh and Cox, 1998) and was purchased from Sigma-Aldrich Chemie B.V. (E-8015; Zwijndrecht, The Netherlands). Purified porcine E. coli heat-labile enterotoxin (LTp-I) was derived from strain C600/P307 (Neill et al., 1983) and was acquired from Lactive B.V. (Lelystad, The Netherlands). CT was obtained from Sigma-Aldrich Chemie B.V. (C8052).

Polyvinylidene difluoride/LT-ELISA

Disks with a diameter of 5.5 mm were punched out of a polyvinylidene difluoride (PVDF) membrane (Boehringer, Mannheim, Germany). The PVDF disks were immersed in methanol (106009; Merck Chemicals, Darmstadt, Germany) for a few seconds until the membranes became translucent, and then equilibrated in tris-buffered saline (TBS: 0.05 M Tris-base and 0.15 M NaCl, pH 7.5) until the disks had sunken into the buffer, employing Petri dishes at room temperature. Subsequently, the PVDF disks were submerged in 100 μL of 2 μg LTp-I/mL TBS or phosphate-buffered saline (PBS: 0.01 M phosphate buffer and 0.15 M NaCl, pH 7.2) or TBS/PBS containing 0.05% Tween 20 (7374; Mallinckrodt Baker B.V., Deventer, The Netherlands), respectively, at room temperature for 1 h. This LT-binding step and all further steps were performed in the wells of a microplate (762070; Greiner Bio-One B.V., Alphen a/d Rijn, The Netherlands) that had been precoated with 1% bovine serum albumin (BSA; A7906; Sigma-Aldrich Chemie B.V.) in TBS. Then, the disks were washed three times with 200 μL TBS/PBS or TBS/PBS with Tween 20 for 5 min each. Afterward, the PVDF membranes were blocked with 1% BSA in TBS/PBS or TBS/PBS with Tween 20. Next, the PVDF disks were incubated with primary antibodies (Table 1; 1:200–1000 diluted in TBS/PBS or TBS/PBS with Tween 20 with 1% BSA), or solely in TBS/PBS or TBS/PBS with Tween 20 with 1% BSA for the negative control, at RT for 1 h, followed by washing as described above. For recognition of unlabeled antibodies, secondary antibodies were added as totals of 100 μL of horseradish peroxidase–conjugated bovine anti-goat IgG (1:2000 diluted in TBS/PBS or TBS/PBS with 1% BSA and 0.05% Tween 20; 805-035-180; Jackson ImmunoResearch Europe Ltd., Newmarket, UK) or peroxidase-conjugated goat anti-mouse IgG (1:1000 diluted in TBS with 1% BSA; 115-035-062; Jackson ImmunoResearch Europe Ltd.), followed by incubation at RT for 1 h. After washing as described above, the disks were incubated with 100 μL diaminobenzidine solution (10 mg DAB per 20 mL TBS; D8001; Sigma-Aldrich Chemie B.V.) and 4 μL 30% H₂O₂ (108597; Merck Chemicals) until the brown color reaction was fully developed. The color reaction was stopped by washing with TBS. The end-point absorbance of the brown-colored PVDF membranes was measured at a wavelength of 460 nm in a spectrophotometric plate reader in comparison with an untreated control membrane (SpectraMax 340; Molecular Devices Ltd., Wokingham, UK).

Table 1.

Primary Anti-Escherichia coli LT Antibodies Tested

Designation	Supplier	Catalog code
Serum Polycl. Goat anti-LT	AbD Serotec (Oxford, UK)	4330-1104
Mab Mouse anti-LT	AbD Serotec	4330-1058
Mab Mouse AE4	HyTest Ltd. (Turku, Finland)	2LTA2 AE4
Mab Mouse AB3 I	HyTest Ltd.	2LTA2 AB3
Mab Mouse AB3 II	Abcam plc (Cambridge, UK)	ab37199
Mab LT-4/PO	CVI (Lelystad, The Netherlands)	Mu-MoAb-α-LT-4-HRPO
Mab Mouse BG12	HyTest Ltd.	2LTB2 BG12
Mab Mouse BB2	Abcam plc	ab37203
Mab Mouse BB12	Abcam plc	ab37204

LT, Escherichia coli heat-labile enterotoxin.

Polystyrene/LT-ELISA

The wells of a high-binding polystyrene (PS) microplate (762071; Greiner Bio-One B.V.) were coated with 100 μL of 2 μg LTp-I/mL TBS or PBS or TBS/PBS, respectively, at 4°C overnight. To remove nonbinding LTp-I, plates were washed three times with 200 μL PBS containing 0.05% Tween 20. Uncovered nonspecific binding sites were blocked by the addition of 200 μL PBS, containing 1% BSA and 0.05% Tween 20, during incubation at RT for 1 h. Subsequently, plates were emptied and washed again as described above. Then, 100 μL of the antibodies each (Table 1) in PBS containing 1% BSA and 0.05% Tween 20 was added to the wells. Plates were incubated at RT for 1 h, followed by washing as described above. For recognition of unlabeled antibodies, secondary antibodies were added and incubated as described above. After washing as described above, 100 μL of ready-to-use TMB solution (Ceditest^® 3,3′,5,5′-Tetramethyl benzidine; Prionics Lelystad B.V., Lelystad, The Netherlands) was added to all wells, followed by incubation at RT for 10 min. The reaction was stopped by addition of 50 μL 1.8 N H₂SO₄ per well. The end-point absorbance of the solutions was measured at a wavelength of 450 nm in the SpectraMax 340 plate reader.

GM1/LT-ELISA

Flat-bottom wells of medium- and high-binding PS strip plates mounted in frame (Microlon^® 200 [762070] and Microlon 600 [762071]; Greiner Bio-One B.V.) were coated with 100 μL of GM1 monosialoganglioside (G7641; Sigma-Aldrich Chemie B.V.) in PBS at 37°C overnight. To remove nonbinding GM1, plates were washed three times with 200 μL PBS containing 0.05% Tween 20. Uncovered nonspecific binding sites were blocked by the addition of 200 μL ELISA buffer (0.01 M phosphate buffer, 0.5 M NaCl, and 0.003 M KCl, pH 7.0), containing 1% BSA and 0.05% Tween 20, during incubation at 37°C for 1 h.

Subsequently, plates were emptied and washed again as described above. Then, 100 μL of LTp-I in ELISA buffer was added to the wells. ELISA buffer without toxin served as blanks. Plates were incubated on a platform shaker (Polymax; Heidolph Instruments, Schwabach, Germany) with 25 rpm at 37°C for 4 h, followed by washing as described above. Antibody detection of bound LTp-I was performed as described for the PS/LT-ELISA.

Interference of test substances with toxin binding to GM1

Flat-bottom wells of medium-binding PS strip plates mounted in frame (Microlon 200) were coated with 100 μL of 1 μg GM1/mL PBS at 37°C overnight. To remove nonbinding GM1, plates were washed three times with 200 μL PBS containing 0.05% Tween 20. Uncovered nonspecific binding sites were blocked as described for the PS-ELISA. Subsequently, plates were emptied and washed again as described above.

Three different protocols (A, B, and C) were followed. In protocol A, the toxin and the test substance were simultaneously incubated in a GM1-coated plate; in protocol B, the toxin was preincubated with the test substance before the transfer of the mixture into a GM1-coated plate; and in protocol C, the test substance was postincubated in a GM1 and toxin–treated plate to resolve bound toxin from GM1.

For protocol A, a total of 50 μL test substance solution (double concentrated in ELISA buffer containing 1% BSA and 0.05% Tween 20) and 50 μL toxin solution (double concentrated in the same buffer) were pipetted into the wells of the GM1-coated and blocked PS strip plate. Plates were incubated on a platform shaker (Polymax; Heidolph Instruments, Schwabach, Germany) with 25 rpm at 37°C for 4 h, followed by washing as described for the GM1/LT-ELISA. Antibody detection of bound LTp-I was performed with anti-LT Mab LT-4/PO (Table 1) as described for the PS/LT-ELISA, employing an incubation time with TMB solution of 40 min. Antibody detection of bound LTh-I was performed with Serum Polycl. Goat anti-LT (Table 1) as described for the PS/LT-ELISA. Antibody detection of bound CT was performed using polyclonal IgG Rabbit anti-CT/HRP (2060-0040; AbD Serotec, Oxford, UK) in PBS containing 1% BSA and 0.05% Tween 20. Plates were incubated and treated as described for the PS/LT-ELISA.

For protocol B, a total of 70 μL test substance solution was incubated with 70 μL toxin solution (see protocol A) for 4 h at 37°C in an inert PCR plate (AB-0600; Abgene, Epsom, UK). Then, 100 μL of the preincubated mixture was pipetted into the wells of the GM1-coated and blocked PS strip plate. The subsequent steps were performed as described in protocol A.

For protocol C, a total of 100 μL of 5 ng or 1 μg LTp-I, 5 ng LTh-I or 4 ng CT per mL ELISA buffer containing 1% BSA and 0.05% Tween 20 was pipetted into the wells of the GM1-coated and blocked PS strip plate. Plates were incubated on a platform shaker (Polymax; Heidolph Instruments) with 25 rpm at 37°C for 4 h, followed by washing as described for the GM1/LT-ELISA. Then, 50 μL ELISA buffer containing 1% BSA and 0.05% Tween 20 and 50 μL test substance solution (see protocol A) were added to the wells. Plates were again incubated on the platform shaker with 25 rpm at 37°C for 4 h, followed by washing as described for the GM1/LT-ELISA. The subsequent steps were performed as described in protocol A.

The predominantly galactose-rich substances tested for their interfering effect with LTp-I binding to GM1 are shown in Table 2. Those substances that most efficiently counteracted LTp-I binding were also tested with LTh-I and CT.

Table 2.

Substances Tested for Their Interfering Efficiency with Toxin Binding to GM1

Designation	Supplier	Catalog code
D-(+)-Galactose	Sigma-Aldrich (St. Louis, MO)	G6404
D-(+)-Fucose (6-deoxy-D-galactose)	Sigma-Aldrich	F8150
Lactose	BD Difco (Franklin Lakes, NJ)	215620
Lactulose	Sigma-Aldrich	61360
N-Acetyl-D-galactosamine	Sigma-Aldrich	A2795
3′-Sialyllactose (3′-N-Acetylneuraminyl-D-lactose)	Sigma-Aldrich	A8681
6′-Sialyllactose (6′-N-Acetylneuraminyl-D-lactose)	Sigma-Aldrich	A8556
Skim milk	BD Difco	232100
Galactan, from gum Arabic	Sigma-Aldrich	851396
Guar gum	Sigma-Aldrich	G4129
Locust bean gum	Sigma-Aldrich	G0753
Tara gum	Water Ingredients b.v. (Zeist, The Netherlands)	—
Konjac gum	Water Ingredients b.v.	—
Fenugreek I	Ethnic food store (whole seeds, ground in analytical grinder [mill type A10S; IKA-Werke, Staufen, Germany] before use)	—
Fenugreek II	Sera Food (Istanbul, Turkey) (çemen; seed powder)	PN: 2618
Soybean hulls, CF >360	(Feedstuff sample, crude fiber >360 g/kg DM)	—
Soybean hulls, CF = 360	(Feedstuff sample, crude fiber = 360 g/kg DM)	—
Dehulled soybean meal, CF <45	(Feedstuff sample, crude fiber <45 g/kg DM)	—
Agar	BD Bacto (Franklin Lakes, NJ)	214010
Agarose	Sigma-Aldrich	A6013
(+)-Arabinogalactan	Sigma-Aldrich (from harvested larch trees)	851361
Larch wood (Larix decidua Mill.)	Lärchenholz Sägewerk Jannach GmbH & Co. KG (Thalheim, Styria, Austria) (dried sawdust^a)	—
Willow wood (Salix alba L.)	Bionoria AG (Neumarkt, Germany) (cortex^a)	—
Carrot	Ernteband Fruchtsaft GmbH (Winnenden, Germany) (root pomace^a)	—
Tomato	Local supermarket (chopped freeze-dried fruit)	—

See Becker and Galletti (2008) for more details.

CF, crude fiber; DM, dry matter.

Data analysis

The raw data obtained with the SpectraMax 340 plate reader were converted into inhibition percentages of toxin-binding to GM1 by means of sigmoidal calibration curves for different concentrations of the toxins using the SoftMax Pro 5.2 software of the reader. Results were considered valid if they fell within the 95% confidence interval of the calibration curve, and the intra-assay variation coefficient was <10%. The inhibition percentages were subjected to a two-way analysis of variance using the GenStat 11 statistical software package (VSN International Ltd., Hemel Hempstead, UK).

Results

Antibody recognition of LTp-I

A dot blot PVDF/LT-ELISA was performed to check for recognition of LTp-I by various primary antibodies (Tables 1 and 3). The employment of Tween 20 in the LTp-I solution ensured an effectual reduction of the background activity for subsequent LT detection. TBS buffer with Tween 20 performed slightly better than PBS buffer with Tween 20 for LTp-I, except with the Mab Mouse BG12 antibody. All antibodies tested were shown to react with LTp-I in this assay (Table 3). In the PS/LT-ELISA, the employment of TBS buffer with Tween 20 gave the best results for LTp-I. In the GM1/LT-ELISA, PBS with Tween 20 as solvent for LTp-I delivered the best results.

Table 3.

Recognition of LTp-I by Primary Antibodies (cf. Table 1) in the PVDF/LT-ELISA, PS/LT-ELISA, and GM1/LT-ELISA

Antibody designation	Specificity	PVDF/LT-ELISA	PS/LT-ELISA	GM1/LT-ELISA
Serum Polycl. Goat anti-LT	LT	+	+	+
Mab Mouse anti-LT	LT	+	+	+
Mab Mouse AE4	LT A	+	+	+
Mab Mouse AB3 I	LT A	+	+	+
Mab Mouse AB3 II	LT A	+	+	+
Mab LT-4/PO	LT B	+	+	+
Mab Mouse BG12	LT B	+	−	−
Mab Mouse BB12	LT B	+	−	−
Mab Mouse BB2	LT B	+	−	−

PVDF, polyvinylidene difluoride; LT, Escherichia coli heat-labile enterotoxin; ELISA, enzyme-linked immunosorbent assay; PS, polystyrene.

Compared to the PVDF/LT-ELISA, the PS/LT-ELISA and the GM1/LT-ELISA resulted in a lower detection incidence of LTp-I (Table 3).

GM1 concentration dependency of LTp-I binding

Figures 1 and 2 show the GM1 concentration dependency of LTp-I binding for different microplate materials (Microlon 200 vs. Microlon 600). With medium-binding microplates (Microlon 200), more LTp-I was bound at lower GM1 concentrations than with high-binding plates, indicating higher capacities for LTp-I recognition by GM1 in medium-binding plates.

FIG. 1.

GM1 concentration dependency of LTp-I binding: ▪, 1 μg/mL LTp-I; ▴, 0.5 μg/mL LTp-I; ▾, 0.25 μg/mL LTp-I; ∙, 0.125 μg/mL LTp-I. Absorption results obtained after Mab LT-4/PO detection from six experiments using Microlon 200 microplates. Error bars indicate SEM.

FIG. 2.

Time dependency of LTp-I binding to GM1

Binding of LTp-I to GM1 was completed after about 4 h (Fig. 3).

FIG. 3.

Time dependency of binding of LTp-I to 1 μg/mL GM1 coating: ▪, 1 μg/mL LTp-I; ▴, 0.5 μg/mL LTp-I; ▾, 0.25 μg/mL LTp-I; ∙, 0.125 μg/mL LTp-I. Absorption results obtained after Mab LT-4/PO detection from six experiments using Microlon 200 microplates. Error bars indicate SEM.

Interference of test substances with LT-I and CT binding to GM1

The interfering potential of different test substances with binding of LTp-I to GM1 was determined with the Mab LT-4/PO antibody (Table 4). At 1 μg/mL LTp-I, instantaneous mixing of toxin and substance sample (treatment A) resulted in comparable findings to those obtained by preincubation of the toxin with a substance sample (treatment B). The most efficient substance in terms of a reduction of toxin binding to GM1 was ground fenugreek seed. Skim milk powder, at a concentration of 5 mg/mL, also performed very well and did better than any of the individual milk saccharides lactose, lactulose, N-acetyl-D-galactosamine, 3′-sialyllactose, or 6′-sialyllactose. At 50 mg/mL, D-(+)-galactose and lactulose scored quite well. Further promising substances were tomato and willow wood. In contrast to the pretreatment and simultaneous treatment A and B, treatment C—after the toxin had already bound to GM1—was, in general, much less effective. At 50 mg/mL, in addition to fenugreek, larch and willow wood scored highest in treatment C. At 5 mg/mL, tara gum was the most potent one among all the substances tested in the after-treatment (Table 4).

Table 4.

Percentage of Inhibition of Binding to GM1 of LTp-I at 1 mg/mL by Different Test Substances

Test substance (mg/mL)	50			5			0.5
Treatment	A	B	C	A	B	C	A	B	C
D-(+)-Galactose	77.8	77.3	40.2	21.0	11.9	4.7	5.9	0.0	−2.7
D-(+)-Fucose^*	17.0	23.8	−2.0	7.1	9.7	−6.0	5.1	8.6	−1.4
Lactose	62.6	59.1	31.7	17.2	11.3	8.9	3.3	1.3	5.1
Lactulose	84.8	82.6	43.8	32.2	24.5	11.0	7.8	4.0	5.2
N-Acetyl-D-galactosamine^*	7.8	22.0	1.1	−5.6	7.9	−0.6	−6.8	1.3	1.8
3′-Sialyllactose^*	—	—	—	—	—	—	−10.7	0.5	−5.8
6′-Sialyllactose^*	—	—	—	—	—	—	2.3	2.8	−2.9
Skim milk	67.3	75.4	6.1	42.3	39.3	4.1	13.2	3.0	−0.7
Galactan^*	42.9	45.2	2.7	21.0	24.6	−11.0	7.9	6.8	−6.4
Guar gum	—	—	—	13.8	12.0	2.6	5.0	1.6	1.0
Tara gum	—	—	—	41.3	29.6	30.4	10.4	7.0	0.5
Locust bean gum	—	—	—	17.8	20.6	15.5	8.6	1.1	3.2
Konjac gum	—	—	—	5.8	5.6	3.0	2.7	6.1	1.1
Fenugreek I^*	86.4	91.0	64.9	63.3	59.3	20.9	36.4	53.9	6.7
Fenugreek II (Çemen)	87.8	91.2	68.6	65.0	58.9	6.8	48.5	29.1	−13.6
Soybean hulls, CF >360^*	52.2	35.7	16.1	4.6	9.1	4.5	−3.4	−1.2	4.8
Soybean hulls, CF = 360	37.3	31.1	11.8	8.4	14.0	1.8	0.1	−1.7	1.5
Dehulled soybean meal, CF <45^*	39.0	31.0	2.2	21.8	13.1	−1.2	5.0	5.2	−3.7
Agar^*	—	—	—	20.8	21.4	−1.1	10.2	6.6	0.3
Agarose^*	—	—	—	−3.6	−10.4	−3.7	0.4	−10.0	−4.4
D-(+)-Arabino-galactan	53.6	53.1	15.2	13.2	11.2	−0.5	1.4	1.0	0.1
Larch wood^*	62.1	63.6	61.8	20.8	18.5	21.1	1.5	9.1	−3.3
Willow wood	68.9	28.1	66.9	21.9	12.4	19.8	15.9	8.1	1.8
Carrot^*	43.6	39.9	12.3	11.7	14.1	6.6	2.3	5.2	0.9
Tomato^*	73.7	78.9	55.2	27.4	15.0	10.2	3.0	1.5	0.6
LSD		10.3			10.5			8.1

LTp-I was determined with Mab LT-4/PO antibody.

Data are means of two independent experiments with duplicate measurements, except for data of substances marked by an asterisk (*), which are means of duplicate measurements without replication. Missing data in the table are due to costliness or solubility issues.

A, simultaneous incubation of LTp-I and test substance in GM1-coated plate; B, preincubation of toxin with test substance before transfer of mixture into GM1-coated plate; C, postincubation of test substance in GM1 and toxin–treated plate to resolve bound LTp-I from GM1.

When testing concentrations of 5 ng/mL LTh-I and 4 ng/mL CT, fenugreek I most efficiently blocked GM1-binding not only of LTh-I (Table 5) but also of CT (Table 6). Testing the effect of the substances lactose, lactulose, skim milk powder, tara gum, and fenugreek I on binding of 5 ng/mL LTp-I to GM1 with the Serum Polyclonal Goat anti-LT antibody resulted in comparable findings to the ones reported in Table 4 (data not shown). However, because these measurements were close to the detection limit of the assay, they were not as reliable as the ones shown in Tables 4 –6. With the ng-per-mL toxin concentrations (Tables 5 and 6), preincubation with fenugreek I resulted in slightly higher inhibition percentages than simultaneous incubation for LTh-I and CT.

Table 5.

Percentage of Inhibition of Binding to GM1 of LTh-I at 5 ng/mL by Selected Test Substances

Test substance (mg/mL)	50			5			0.5
Treatment	A	B	C	A	B	C	A	B	C
Lactose	53.7	52.8	11.7	8.8	14.7	7.7	6.8	4.8	3.8
Lactulose	77.5	81.5	14.5	14.7	22.5	4.2	13.4	17.1	3.7
Skim milk	43.7	79.5	9.7	64.9	66.0	29.2	20.7	14.1	10.0
Tara gum	—	—	—	19.0	32.5	3.6	7.3	5.0	−2.6
Fenugreek I	92.2	91.5	59.3	42.5	59.2	−1.0	9.3	35.7	−0.8
LSD		10.8			9.4			13.9

LTh-I was determined with serum polyclonal goat anti-LT antibody.

Data are means of two independent experiments with duplicate measurements.

Table 6.

Percentage of Inhibition of Binding to GM1 of Cholera Toxin at 4 ng/mL by Selected Test Substances

Test substance (mg/mL)	50			5			0.5
Treatment	A	B	C	A	B	C	A	B	C
Lactose	39.9	48.5	10.7	4.8	4.1	11.1	7.9	13.4	7.8
Lactulose	56.2	63.7	19.5	6.3	5.9	6.3	3.2	6.5	2.2
Skim milk	69.9	97.3	8.1	62.6	67.3	23.5	12.4	5.0	2.2
Tara gum	—	—	—	33.4	29.3	16.4	3.9	2.9	1.7
Fenugreek I	93.0	98.8	55.4	61.4	91.8	8.4	32.5	75.3	4.6
LSD		8.5			9.2			11.2

CT was determined with polyclonal IgG rabbit anti-CT/HRP.

Data are means of two independent experiments with duplicate measurements.

CT, cholera toxin; A, simultaneous incubation of CT and test substance in GMl-coated plate; B, preincubation of toxin with test substance before transfer of mixture into GM1-coated plate; C, postincubation of test substance in GM1 and toxin–treated plate to resolve bound CT from GM1.

Discussion

ELISA assays

While all primary antibodies were shown to recognize LTp-I when attached to the hydrophobic PDVF membrane material, significantly less did so when LTp-I was bound to the hydrophilic, high-binding PS material or to GM1 (Table 3). Probably, most antibodies with a specificity to the B-subunit of the toxin gave no results in the PS/LT- and the GM1/LT-ELISA, because LTp-I attached with the B-subunit to the PS plate material and to GM1, thus turning distinct B-sites inaccessible for antibody recognition.

When comparing high-binding with medium-binding microplate materials for coating with GM1, the medium-binding Microlon 200 proved superior (Figs. 1 and 2). To be able to expose its hydrophilic LT-I receptor, GM1 needs to attach to the carrier material with its hydrophobic ceramide moiety. Obviously, Microlon 200 offers more hydrophobic and thus more adequate binding conditions for that purpose than Microlon 600. Conversely, the higher hydrophilicity of Microlon 600 compared to Microlon 200 seems to cause a higher proportion of GM1 to bind upside down, resulting in the effect that on Microlon 600, bound GM1 is less reactive toward LT-I (Figs. 1 and 2).

The time dependency of LTp-I binding to GM1 (Fig. 3) was in agreement with the finding reported by Dawson (2005) for CT. Hence, the GM1-ELISA protocol adapted for LTp-I was also used for testing LTh-I and CT binding, with the exception that different antibodies were employed for toxin detection based on our own findings in terms of the antibodies' sensitivities or the supplier's recommendation in the case of CT.

Interference of test substances with LT-I and CT binding to GM1

Because the galactose binding site of LT-I was reported to be a promising target for blocking GM1 receptor binding (Merritt et al., 1994b), we tested the intervening efficiency with toxin adherence of several galactose-containing substances (Table 2).

Ground fenugreek seed performed best among all substances for all toxins in treatments A and B (Tables 4 –6). Fenugreek seeds contain about 20% fenugreek gum, consisting of galactomannans (Brummer et al., 2003), which might be the active principle of fenugreek. The spice fenugreek (seeds of Trigonella foenum-graecum L.; Gal/Man = 0.8–0.9), soybean hulls (Glycine max L. Merr.; Gal/Man = 0.7), and food stabilizers such as guar gum (E412; Gal/Man = 0.6–0.7), locust bean gum (E410; Gal/Man = 0.3), and tara gum (E417; Gal/Man = 0.3) all contain galactomannans, which in the food stabilizers are the predominant components (Daas et al., 2000; Srivastava and Kapoor, 2005). Guar gum is extracted from the seed of the leguminous shrub Cyamopsis tetragonoloba L., locust or carob bean gum is obtained from the seeds of the carob tree (Ceratonia siliqua L.), and tara gum is the endosperm of the seeds of the tara tree (Caesalpinia spinosa [Molina] Kuntze). Other galactomannan sources than fenugreek performed inferior in our tests, which might be due to the plant-specific characteristics of the respective galactomannans. Tara gum might offer some potential to resolve bound LTp-I from GM1, but not LTh-I or CT (Tables 4 –6). Guar and locust bean gum were less efficient in this respect with LTp-I, despite having higher or similar ratios of galactose/mannose in their mannans. Konjac gum contains glucomannan as its main component and is obtained from the corm of the konjac plant (Amorphophallus konjac K. Koch). Konjac gum was included in this study for the purpose of comparison. In line with its composition, the results obtained with konjac gum were much poorer than the ones obtained with the other gums.

Agar (E406), derived from red seaweeds, consists of agarose ([4)-3,6-An-α-L-Galp-(1 → 3)-β-D-Galp-(1 → ]_n) and agaropectin. In agaropectin, some sugar units are replaced by 4,6-O-(1-carboxyethylidene)-D-galactopyranose or by methylated or sulfated units (Labropoulos et al., 2002). Agar inhibited LTp-I binding to a higher extent than agarose (Table 4). However, compared to some non-agar products, the agar-related substances performed mediocre at best.

Arabinogalactans are composed of β-D-(1 → 3)-galactan main chains, with side chains made up of D-galactose and L-arabinose residues. Not only larch wood contains arabinogalactan (Gal/Ara = 6), but also carrots, tomatoes, and herbs (D'Adamo, 1990). Willow wood as a reference substance, despite containing only minor amounts of galactose (Skrigan et al., 1967), performed similar to larch wood in the LTp-I inhibition test A (Table 4).

Among mammalian substances, especially human milk is rich in the disaccharide lactose (55–70 mg/mL; β-D-Gal-(1 → 4)-D-Glc) and galactose-containing oligosaccharides (Kunz et al., 2000; Boehm and Stahl, 2007). Bovine milk contains about 40–50 mg lactose/mL (Kunz et al., 2000). During thermal processing of milk, lactose isomerizes to lactulose (β-D-Gal-(1 → 4)-D-Fru). Crystal structures have been described for LT-galactose, LT-lactose, and LT-lactulose complexes (Sixma et al., 1992; Hol et al., 1995; Merritt et al., 1997; Minke et al., 1999). Hol et al. (1995) had already pointed out that the galactose moiety of the sugars seemed more important for saccharide binding by LT than the nature of a second β1-linked hexose. However, knowledge about the displacing capacity of natural galactose-rich or -related compounds for the GM1 pentasaccharide remained sparse. Griffiths and Critchley (1991) reported that binding of LT-I to brush borders was effectively inhibited by galactose and fucose (6-deoxy-D-galactose). In our GM1/LTp-I inhibition test, however, galactose performed much better than fucose (Table 4). Minke et al. (1999) reported inhibition percentages for LT-binding to GD1b of 22% for galactose (20 mM; 4 mg/mL) and lactose (20 mM; 7 mg/mL) and 54% for lactulose (20 mM; 7 mg/mL) in a coated plate test similar to the one employed by us. GD1b has the same galactose–forefinger structure as GM1, but an extra N-acetylneuraminic acid (NANA) linked to the glycerol moiety of the NANA thumb of the GM1 pentasaccharide. Our findings agree with the ones of Minke et al. (1999) in terms of the superior performance of lactulose in the GM1/LTp-I ELISA. Apart from that, galactose did slightly better than lactose (Table 4). In contrast to LT binding, CT binding is quite specific for GM1 (Hol et al., 1995). Dawson (2005) observed practically no inhibition of CT binding to GM1 when trying to displace bound CT or preincubate CT with galactose. Likewise, we found somewhat lower inhibition potentials of lactose and lactulose for 4 ng/mL CT binding than for 5 ng/mL LTh-I binding (Tables 5 and 6). Free saccharides other than lactose occur only in trace amounts in bovine milk (Kunz et al., 2000). Low-level saccharides are, for example, modified lactoses, such as 3′-sialyllactose ( = 3′-N-acetylneuraminyl-D-lactose) and 6′-sialyllactose ( = 6′-N-acetylneuraminyl-D-lactose) (Urashima et al., 2001). These sialyllactoses amount to about 0.03–0.06 mg/mL milk (Kunz et al., 2000) and were also employed as test compounds in this study. Due to their costliness, the latter saccharides were only tested in low concentrations, which did not result in a noticeable inhibition (Table 4). Probably the linkage of NANA to the galactose moiety in these saccharides obstructed toxin binding. Comparably, the N-acetyl-group in N-acetyl-D-galactosamine most likely impeded the reaction with LTp-I. Unexpectedly, skim milk powder at a concentration of 5 mg/mL performed better in simultaneous (A) and in preincubation (B) treatments with 5 ng and 1 μg/mL LTp-I and 5 ng/mL LTh-I than all the single milk compounds tested, plus the galactose polymer galactan from gum arabic (Table 4). With 4 ng/mL CT, skim milk powder also did very well, especially when preincubated (treatment B) (Table 6). In skim milk, apart from saccharides, glycopeptides and glycolipids could be responsible for the observed effect (Otnæss et al., 1983; Shida et al., 1994).

Footnotes

Acknowledgments

This study was funded by the Knowledge Basis Research program of the Dutch Ministry of Agriculture, Nature, and Food Quality and by the project SAFEWASTES (no. 513949) of the European Commission FP6 Framework Programme. The authors are grateful to Dr. F.G. van Zijderveld (CVI, Lelystad, The Netherlands) for his kind provision of the B-specific antibody Mu-MoAb-α-LT4-HRPO, and to Monique Verveer for excellent technical assistance.

Disclosure Statement

No competing financial interests exist.

References

Becker

, Galletti

. Food and feed components for gut health-promoting adhesion of E. coli and Salmonella enterica. J Sci Food Agric, 2008; 88:2026–2035.

Boehm

, Stahl

. Oligosaccharides from milk. J Nutr, 2007; 137:847S–849S.

Brummer

, Cui

, Wang

. Extraction, purification and physicochemical characterization of fenugreek gum. Food Hydrocoll, 2003; 17:229–236.

Daas

PJH

, Schols

, de Jongh

HHJ

. On the galactosyl distribution of commercial galactomannans. Carbohydr Res, 2000; 329:609–619.

D'Adamo

. Larch arabinogalactan. J Naturopath Med, 1990; 6:33–37.

Dawson

. Characterization of the binding of cholera toxin to ganglioside G_M1 immobilized onto microtitre plates. J Appl Toxicol, 2005; 25:30–38.

Dertzbaugh

, Cox

. The affinity of cholera toxin for Ni²⁺ ion. Protein Eng, 1998; 11:577–581.

, Cu

, Nguyen

et al. Pathotypes and serogroups of enterotoxigenic Escherichia coli isolated from pre-weaning pigs in north Vietnam. J Med Microbiol, 2006; 55:93–99.

Griffiths

, Critchley

. Characterisation of the binding sites for Escherichia coli heat-labile toxin type I in intestinal brush borders. Biochim Biophys Acta, 1991; 1075:154–161.

10.

Gyles

, So

, Falkow

. The enterotoxin plasmids of Escherichia coli. J Infect Dis, 1974; 130:40–49.

11.

Hol

WGJ

, Sixma

, Merritt

. Structure and function of E. coli heat-labile enterotoxin and cholera toxin B pentamer. Bacterial Toxins and Virulence Factors in Disease. Handbook of Natural Toxins, 8 Moss

, Iglewski

, Vaughan

, Tu

. New York: Marcel Dekker, 1995; 185–223.

12.

Khac

, Holoda

, Pilipcinec

et al. Serotypes, virulence genes, and PFGE profiles of Escherichia coli isolated from pigs with postweaning diarrhoea in Slovakia. BMC Vet Res, 2006; 2:10(Online.)

13.

Kunz

, Rudloff

, Baier

et al. Oligosaccharides in human milk: structural, functional, and metabolic aspects. Annu Rev Nutr, 2000; 20:699–722.

14.

Labropoulos

, Niesz

, Danforth

, Kevrekidis

. Dynamic rheology of agar gels: theory and experiments. Part I. Development of a rheological model. Carbohydr Polym, 2002; 50:393–406.

15.

Merritt

, Sarfaty

, Akker

FVD

et al. Crystal structure of cholera toxin B-pentamer bound to receptor G(M1) pentasaccharide. Protein Sci, 1994a. 3:166–175.

16.

Merritt

, Sarfaty

, Feil

, Hol

WGJ

. Structural foundation for the design of receptor antagonists targeting Escherichia coli heat-labile enterotoxin. Structure, 1997; 5:1485–1499.

17.

Merritt

, Sixma

, Kalk

et al.

Galactose-binding site in Escherichia coli heat-labile enterotoxin (LT) and cholera toxin (CT)

Mol Microbiol, 1994b. 13:745–753.

18.

Minke

, Roach

, Hol

WGJ

, Verlinde

CLMJ

. Structure-based exploration of the ganglioside GM1 binding sites of Escherichia coli heat-labile enterotoxin and cholera toxin for the discovery of receptor antagonists. Biochemistry, 1999; 38:5684–5692.

19.

Nagy

, Wilson

, Whittam

. Genetic diversity among Escherichia coli isolates carrying f18 genes from pigs with porcine postweaning diarrhea and edema disease. J Clin Microbiol, 1999; 37:1642–1645.

20.

Nardi

ARM

, Salvadori

, Coswig

et al. Type 2 heat-labile enterotoxin (LT-II)-producing Escherichia coli isolated from ostriches with diarrhea. Vet Microbiol, 2005; 105:245–249.

21.

Nataro

, Kaper

. Diarrheagenic Escherichia coli. Clin Microbiol Rev, 1998; 11:142–201.

22.

Neill

, Twiddy

, Holmes

. Synthesis of plasmid-coded heat-labile enterotoxin in wild-type and hypertoxinogenic strains of Escherichia coli and in other genera of Enterobacteriaceae. Infect Immun, 1983; 41:1056–1061.

23.

Otnæss

, Lægreid

, Ertresvåg

. Inhibition of enterotoxin from Escherichia coli and Vibrio cholerae by gangliosides from human milk. Infect Immun, 1983; 40:563–569.

24.

Qadri

, Svennerholm

A-M

, Faruque

ASG

, Sack

. Enterotoxigenic Escherichia coli in developing countries: epidemiology, microbiology, clinical features, treatment, and prevention. Clin Microbiol Rev, 2005; 18:465–483.

25.

Shida

, Takamizawa

, Nagaoka

et al. Enterotoxin-binding glycoproteins in a proteose-peptone fraction of heated bovine milk. J Dairy Sci, 1994; 77:930–939.

26.

Sixma

, Kalk

, van Zanten

BAM

et al. Refined structure of Escherichia coli heat-labile enterotoxin, a close relative of cholera toxin. J Mol Biol, 1993; 230:890–918.

27.

Sixma

, Pronk

, Kalk

et al. Lactose binding to heat-labile enterotoxin revealed by X-ray crystallography. Nature, 1992; 355:561–564.

28.

Skrigan

, Murashkevich

, Pershina

et al. The chemical composition of the wood of some species of willow. Chem Nat Compd, 1967; 3:328–330.

29.

Smith

. The enterotoxin and other plasmids of enteropathogenic Escherichia coli. Vet Res Comm, 1977; 1:213–224.

30.

Srivastava

, Kapoor

. Seed galactomannans: an overview. Chem Biodivers, 2005; 2:295–317.

31.

Svennerholm

, Holmgren

. Identification of Escherichia coli heat-labile enterotoxin by means of a ganglioside immunosorbent assay (GM1-ELISA) procedure. Curr Microbiol, 1978; 1:19–23.

32.

Teneberg

, Hirst

, Angstrom

, Karlsson

. Comparison of the glycolipid-binding specificities of cholera toxin and porcine Escherichia coli heat-labile enterotoxin: identification of a receptor-active non-ganglioside glycolipid for the heat-labile toxin in infant rabbit small intestine. Glycoconj J, 1994; 11:533–540.

33.

Tsuji

, Taga

, Honda

et al. Molecular heterogeneity of heat-labile enterotoxins from human and porcine enterotoxigenic Escherichia coli. Infect Immun, 1982; 38:444–448.

34.

Urashima

, Saito

, Nakamura

, Messer

. Oligosaccharides of milk and colostrum in non-human mammals. Glycoconj J, 2001; 18:357–371.

35.

Van den Akker

, Sarfaty

, Twiddy

et al. Crystal structure of a new heat-labile enterotoxin, LT-IIb. Structure, 1996; 4:665–678.

36.

Willshaw

, Barclay

, Smith

et al. Molecular comparison of plasmids encoding heat-labile enterotoxin isolated from Escherichia coli strains of human origin. J Bacteriol, 1980; 143:168–175.

37.

Zhang

, Berberov

, Freeling

et al. Significance of heat-stable and heat-labile enterotoxins in porcine colibacillosis in an additive model for pathogenicity studies. Infect Immun, 2006; 74:3107–3114.