Double-sandwich enzyme-linked immunosorbent assay for determination of Escherichia coli heat-labile porcine enterotoxins

Veterinary Microbiology, 17 (1988) 83-90 Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

83

Short Communication

Double-sandwich Enzyme-linked Immunosorbent A s s a y for D e t e r m i n a t i o n of E s c h e r i c h i a coli Heatlabile P o r c i n e E n t e r o t o x i n s BRIGITTE PICARD 1, J.-M. ALESSANDRI 2° and YVONNE DUVAL-IFLAH1

II.N.R.A., Laboratoire d'Ecologie Microbienne, Centre de Recherches de Jouy-en-Josas, 78350 Jouy-en-Josas (France) 2LN.R.A., Station de Recherches de Nutrition, Centre de Recherches de Jouy-en-Josas, 78350 Jouy-en-Josas (France) (Accepted for publication 1 December 1987 )

ABSTRACT Picard, B., Alessandri, J.-M. and Duval-Iflah, Y., 1988. Double-sandwichenzyme-linkedimmunosorbent assay for determination of Escherichia coli heat-labile porcine enterotoxins. Vet. Microbiol., 17: 83-90. A "double-sandwich" ELISA for the detection and measurement of a heat-labile enterotoxin produced by porcine enterotoxigenic strains of Escherichia coli (LTp) is described. In contrast with other heat-labile toxins, LTp did not bind to agarose gels and exhibited a very low affinity for GM1 in the classical GM1-ELISA technique. The similarity of LTp with cholera toxin was confirmed by immunoblotting.This property allowed the binding of LTp to rabbit IgG anti-cholera toxin antibodies (covalently linked to polystyrene plates) and sheep anti-cholera toxin serum. The immunocomplex was revealed by anti-sheep immunoglobulin antibodies conjugated with peroxidase. Application of the "double-sandwich" ELISA to the quantitation of toxin production by two strains, which differ only in the presence or the absence of the K88ab antigen, showed that the Ent +, K88 + strain produced significantly less toxin than the Ent +, K88- derivative.

INTRODUCTION

The structure and mechanism of action of heat-labile toxins (LT) are similar to those of cholera toxin (CT) (reviewed by Eidels et al., 1983; Middlebrook and Dorland, 1984; Lai, 1986). The GM1-ELISA method is the most frequently used for determining CT, human LT (Svennerholm and Holmgren, 1978; Wolk et al., 1980; Holmgren et al., 1982, 1985) or porcine LT (Olsvik et al., 1983). The galactosyl-N-acetylgalactosaminyl- (N-acetylneuraminyl)-gal*Author to whom correspondence should be addressed.

0378-1135/88/$03.50

© 1988 Elsevier Science Publishers B.V.

84 actosylglucosylceramide, or GM1, (reviewed by Fishman and Brady, 1976) has been identified in plasma membranes as the main receptor for cholera toxin (Eidels et al., 1983 ). However, heat-labile toxins exhibit a variable affinity for GM1 depending on their origin (Olsvik et al., 1983; WadstrSm et al., 1984) and may recognize different receptors in vivo (Holmgren et al., 1982, 1985; Griffiths et al., 1986). The lower affinity for GM1 of LT produced by a porcine strain led us to develop an ELISA technique where GM1 was replaced by IgG anti-CT antibodies. MATERIALSAND METHODS Purification of IgG anti-CT antibodies by affinity chromatography Following the method of Guesdon and Avrameas (1976), 3 ml of serum from a rabbit immunized by 3 injections with 100 #g of pure cholera toxin (Sigma) in Freund's adjuvant, were submitted to filtration for 12 h at 4°C in closed circuit in a column containing AcA 34 gel (reagents IBF) coupled with CT according to a previously-described method (Ldger et al., 1982). The column was thereafter washed with 100 mM phosphate-buffered saline, NaC1 150 mM (PBS) and antibodies were desorbed with the glycine-HC10.2 N pH 2.8 buffer. Eluted fractions were pooled, neutralized by K2HPO4 1 M and dialyzed at 4 ° C against PBS. The total amount of gamma-globulins purified from 3 ml of serum determined according to the method of Lowry et al. ( 1951 ) was about 5 mg per 0.88 mg immobilized CT, i.e. a molar ratio I g G / C T of approximately 3. Production of L T crude extracts Porcine E. coli strains 5148 (K88ab +, LT +, ST + ), 5148 (K88ab-, LT +, ST + ) and JP1 (K88ab +, E n t - ) (Duval-Iflah et al., 1983; Duval-Iflah and Chappuis, 1984) were cultivated with vigorous shaking in 50 ml MOPS medium (morpholino propane potassium sulfonate, Serva) in 250-ml flasks at 37 ° C for 24 h, according to Gilligan and Robertson (1979). The pellet obtained by centrifugation was suspended in a 5-ml Tris buffer (10 mM), EDTA (10 mM), glucose (50 mM), lysosyme (50 mg m1-1, pH 8), and incubated for 15 minutes at 4°C. Spheroplasts were pelleted, lysed in EDTA (10 mM, pH 8) and cell fragments were washed once with PBS. The two supernatants (EDTAspnt and PBS-spnt, respectively) were dialyzed against PBS and constituted crude LT + (5148 K88ab + and 5148 K88ab- ) and L T - (JP1) extracts. Filtration of L T + crude extracts on agarose gels LT + crude extracts dialyzed against Tris buffer 10 mM, EDTA 1 mM, NaN3 0.01%, NaC1 150 m M (TEAN buffer) were submitted to filtration in a column

85 of agarose A-5m or A-1.5m (Biorad) equilibrated and eluted with the same buffer. The column was then washed with TEAN buffer containing 0.2 M galactose according to the method described by Clements and Finkelstein (1979).

SDS-PAGE Samples of purified IgG, CT and LT crude extracts were analyzed by slab electrophoresis on 10% polyacrylamide gels in the presence of SDS according to the technique of Laemmli {1970). Determination of molecular weights was made by comparison with two standard calibration kits {proteins were: lysosyme, soybean trypsic inhibitor, carbonic anhydrase, ovalbumin, bovine serum albumin, phosphorylase B, fl-galactosidase and myosin).

Immunoblotting Proteins fixed on polyacrylamide gel were transferred to a nitrocellulose sheet at a voltage of 60 V for 2 h. Anti-CT purified IgG was visualized by addition of rabbit anti-IgG antibodies conjugated with peroxidase (Institut Pasteur Production) diluted 500-fold in PBS, Triton X-100 1%, BSA (V) 10%, for 1 h at room temperature. Peroxidase was revealed by diaminobenzidic acid in the presence of nickel and copper salts. Toxin samples were revealed in the same conditions by sequential incubations with anti-CT IgG {diluted 200-fold) and rabbit anti-IgG antibodies coupled with peroxidase.

ELISA Polystyrene flat-bottomed microtiter plates (Limbro-Titertek, Flow) were used. Glutaraldehyde (grade II) was purchased from Sigma. Sheep anti-CT serum and sheep anti-IgG antibodies conjugated with peroxidase were obtained from Swiss and Institut Pasteur Production, respectively. The doublesandwich technique is given in Table I; the wells were filled with 100 pl of each reagent. After the saturation step plates can be used or stored at - 2 0 °C for several weeks. The development was stopped by the addition of 50 ~1 of H2SO4 5 N and reading was performed using a densitometer (Multiskan Titertek) at 492 nm. The titer of immunoreactive LT contained in the samples was determined with the formula: LT -- CT × 2y- where CT is the concentration of the cholera toxin parent solution (ng ml-1); and n and y, the respective numbers of CT and LT dilutions necessary to reach an absorbance at 492 n m equal to 0.5. The LT titer of samples was used to calculate the total amount of heat-labile toxin (in ng) produced per ml of culture. n

86

TABLE I ELISA procedure for determination of porcine LT Step

Reagents

Buffer

Procedure

Activation

Glutaraldehyde 1%

Distilled water

1 h at 4 ° C '(4 washes)

Coating

Rabbit anti-CT IgG (5 pg ml- 1)

PBS

12 h at room temperature (4 washes )

Saturation of sites

BSA 1%

PBS-Tween 0.5%

2 h at 37°C (3 washes)

Sample deposition

CT: range 4-612 ng mlLT: crude extracts diluted 2 by 2

PBS-Tween 0.5%

12 h at 4 ° C (4 washes)

Antiserum

Sheep anti-CT serum (1000-fold diluted)

PBS-Tw-BSA 1%

2 h at 37°C (4 washes)

Conjugated

Sheep anti-IgG antibodies conjugated with peroxidase (500-fold diluted)

PBS-Tw-BSA 1%

2 h at 37°C (8 washes)

Development

Tetracetate orthophenyldiamine 0.4 mg ml -~

Citrate 50 rnM pH 5, H202 0.4%

10 m i n in darkness

A

B

C

~ - 150 000

~-- 72 000

D 42 000 ---~ 51 000 --~ 29 0 0 0 ~ 26000~

( 14 0 0 0 - - ~

53000

p

Fig. 1. Illustrated i m m u n o b l o t t i n g s of pure cholera toxin (Lane A ), L T + -crude extract (Lane B ), a n d purified anti-cholera toxin IgG (Lane C). Samples were s u b m i t t e d to S D S - P A G E a n d t h e n transferred to nitrocellulose sheets. Toxin subunits were revealed by incubation of t h e slices with rabbit anti-cholera toxin IgG, followed by incubation with a n t i - r a b b i t IgG serum conjugated with peroxidase. B a n d s of purified r a b b i t immunoglobulins were revealed by incubation of the slices with a n t i - r a b b i t IgG serum conjugated with peroxidase.

87 RESULTS

Immunoblots of anti-CT purified IgG, of CT and LT Purity of anti-CT antibodies was checked by SDS-PAGE (not shown) and immunoblotting (Fig. 1 ). Cholera toxin is composed of 3 bands whose migration corresponds to molecular weights close to 31, 29 and 26 kDa, respectively. The first two bands correspond to the molecular weights of A (34.5 kDa) and A1 (29 kDa) subunits of CT. The recombination of two CT-B subunits (11.5 kDa) may account for the presence of a band close to 26 kDa. Monomeric B 1,5.

® E c:

I.

,,¢

0,5.

o

®

E c

o,~-

~2

~3

~

~5

~

:~7

~a Dilutions

Fig. 2. E L I S A of a crude extract from a culture of the E n t + K88ab ÷ strain (a) a n d from a culture of the E n t + K 8 8 a b - strain (b). E a c h p o i n t represents t h e m e a n value of a triplicate. The lowest and highest values are represented by horizontal bars. ( © ), E L I S A of LT contained in E D T A spnt (first extraction ); ( • ) , E L I S A of L T contained in P B S - s p n t (second extraction ).

88 subunits are located in the migration front (molecules with a molecular weight below 14 kDa). Similar results were obtained with L T p which is composed of a major immunoreactive band with an intermediate migration compared to that of CT-A and CT-A1 subunits, and a minor band with a molecular weight close to 42 kDa which may correspond to the dimeric association LT-A + LT-B. LT-B subunits migrate as CT-B subunits. The same results (not shown) were obtained using rabbit entire serum instead of purified IgG.

E L I S A for L T determination The range of sensitivity of the assay performed with a parent solution of CT at 20/~g m l - 1 is located between 160 and 5 ng m l - 1. Half-saturation of coating was obtained at a CT concentration of 30.7 ( + 1.0 ) ng ml-1. Results obtained with crude extracts were compared to the range of CT so as to determine the LT titer in each sample (see Fig. 2). The level of L T produced by each LT + strain was determined in the two supernatants, E D T A - s p n t and PBS-spnt, and the mean titer of LT per ml of culture (6 determinations) was obtained by adding the toxin concentrations of each of the two supernatants. It is equal to 44.5 ng m1-1 ( S . E . = 8 . 4 ) for strain 5148 K 8 8 a b - , E n t + a n d t o 25.4 ng m1-1 (S.E.--5.6) for strain 5148 K88ab +, E n t +. Seventy-four and 63% (mean values) of the total toxin was extracted in the EDTA-spnts of the K 8 8 a b - and K88ab + strains, respectively. Supernatants collected after lysosyme treatment of strains LT + and all supernatants of strain L T - were not significantly different from the base line.

Binding trials of L T to agarose gel

Detection of L T by means of the E L I S A procedure in the different fractions showed that all of the toxin was eluted with the T E A N buffer. The T E A N galactose buffer only eluted a residual fraction of proteins, negative with ELISA i.e., below 5 ng ml-1 toxin. DISCUSSION Preliminary experiments, not reported here, showed that the classical GM1E L I S A was not suitable for determination of crude extracts of LT produced by our strains. The detection of such toxins based on the affinity for a given type of receptor was therefore questionable, and the development of a more specific method was sought. A comparison using immunoblotting of heat-labile toxin and cholera toxin confirmed that the porcine L T exhibited remarkable antigenic similarities with CT. These interactions between LTp and anti-CT IgG

89 were used to develop a double-sandwich E L I S A which enhanced the immunological reactions. The sensitivity of this test was approximately 5 ng m l similar to that of the commonly used G M 1 - E L I S A for determination of heatlabile toxins of h u m a n origin (reviewed by Germani, 1986). The reproducibility of the trials performed on crude extracts of LT issued from two strains of porcine origin suggests that this test can be used for the systematic determination of LT in E. coli cultures isolated from diarrhea of piglets. The comprehensive analysis of the L T levels at each extraction step showed that in our culture conditions, the action of lysosyme did not release the toxin to the medium. The lysis of spheroplasts was necessary for the release of toxin, and further washing of m e m b r a n o u s fragments yielded more LT. Accordingly, it was necessary to add the yield from both steps to determine the total production of toxin per ml of culture. This was significantly different in the two strains studied, the variant which had lost the plasmid encoding for the K88ab adhesin synthetized more (about 2-fold) toxin in vitro than the parent strain. The lack of affinity for agarose of the porcine L T used, suggests that it does not bind to galactose. Accordingly, the receptors for this toxin in vivo may be different from the binding sites of galactose-binding toxins (WadstrSm et al., 1984). The poor recognition of GM1 and galactose by our porcine L T may be caused by only minor structural differences in the toxin molecule. Indeed, polyclonal antibodies recognize many epitopes, whereas GM1 only binds to one site of each B subunit. Thus, it is noteworthy that the replacement of the glycine residue at position 33 by aspartic acid induced by a mutagen on LT-B, is sufficent to inhibit completely the interactions between the modified LT and GM1 (Tsuji et al., 1985). ACKNOWLEDGEMENTS We wish to thank Dr. Boquet (Institut Pasteur, Paris ) for helpful advice and Annick Bouroche for the English translation of the manuscript. This work was partially supported by grants from the "Action thdmatique programm~e du C.N.R.S." entitled "Alimentation et Sant~ (P.I.R.E.N.)".

REFERENCES Clements, J.D. and Finkelstein, R.A., 1979. Isolation and characterization of homogeneous heatlabile enterotoxins with high specific activity from Escherichia coli cultures. Infect. Immun., 24(3): 760-769. Duval-Iflah, Y. and Chappuis, J.P., 1984. Influence of plasmids on the colonization of the intestine by strains ofEscherichia coli in gnotobiotic and conventional animals. In: M.J. Klug and C.A. Reddy (Editors), Current Perspectives in Microbial Ecology. Proceedings of the Third International Symposium on Microbial Ecology, Michigan State University, 7-12 August 1983. Duval-Iflah, Y., Chappuis, J.P., Ducluzeau, R. and Raibaud, P., 1983. Intraspecific interactions

90 between Escherichia coli strains in human newborns and in gnotobiotic mice and piglets. Prog. Food Nutr. Sci., 7: 107-116. Eidels, L., Proia, R.L. and Hart, D.A., 1983. Membrane receptors for bacterial toxins. Microbiol. Rev., 47(4): 596-620. Fishman, P.H. and Brady, R.O., 1976. Biosynthesis and function of gangliosides. Science, 194: 906-915. Germani, Y., 1986. Identification and assays methods for Escherichia coli enterotoxins. Bull. Inst. Pasteur, Paris, 84: 365-387. Gilligan, P.H. and Robertson, D.C., 1979. Nutritional requirements for synthesis of heat-labile enterotoxin by enterotoxigenic strains of Escherichia coll. Infect. Immun., 23 (1): 99-107. Griffiths, S.L., Finkelstein, R.A. and Critchley, D.R., 1986. Characterization of the receptor for cholera toxin and Escherichia coli heat-labile toxin in rabbit intestinal brush-borders. Biochem. J., 238: 313-322. Guesdon, J.L. and Avrameas, S., 1976. Polyacrylamide-agarose beads for the preparation of effective immunosorbents. J. Immunol. Methods, 11: 129-133. Holmgren, J., Fredman, P., Lindblad, M., Svennerholm, A.M. and Svennerholm, L., 1982. Rabbit intestinal glycoprotein receptor for Escherichia coli heat-labile enterotoxin lacking affinity for cholera toxin. Infect. Immun., 38 (2): 424-433. Holmgren, J., Lindblad, M., Fredman, P., Svennerholm, L. and Myrvold, H., 1985. Comparison of receptors for cholera and Escherichia coli enterotoxins in human intestine. Gastroenterology, 89: 27-35. Laemmli, U.K., 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (London), 227: 680-685. Lai, C.°Y., 1986. Bacterial protein toxins with latent ADP-ribosyl transferases activities. Adv. Enzymol., 58: 99-140. LSger, C., Alessandri, J.-M., Kann, G., Charles, M., Corring, T. and Flanzy, J., 1982. Binding between immobilized anti-colipase purified antibodies and colipase. Radioimmunoassay of colipase from pig plasma and pancreatic juice. Biochim. Biophys. Acta, 713: 208-221. Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J., 1951. Protein measurement with the Folin phenol reagent. J. Biol. Chem., 193: 265-275. Middlebrook, J.L. and Dorland, R.B., 1984. Bacterial toxins cellular mechanisms of action. Microbiol. Rev., 48: 199-221. Olsvik, O.D., Lund, A.D., Berdal, B.P. and Bergan, T., 1983. Differences in binding to the GM1 receptor by heat-labile enterotoxin of human and porcine Escherichia coli strains. Niph. Ann., 6(1 ): 5-15. Svennerholm, A-M. and Holmgren, J., 1978. Identification of Escherichia coli heat-labile enterotoxin by means of a ganglioside immunosorbent assay (GM 1-ELISA ). Curr. Microbiol., 1:1923. Tsuji, T., Honda, T., Miwatani, T., Wakabayashi, S. and Matsubara, H., 1985. Analysis of receptor-binding site in Escherichia coli enterotoxin. J. Biol. Chem., 260 {14): 8552-8558. WadstrSm, T., Baloda, S.B. and Ronnberg, B., 1984. New trends in purification of "old and new" bacterial protein toxins. In: Bacterial Protein Toxins, FEMS Symposium No. 24, Academic Press, p. 235-240. Wolk, M., Svennerholm, A-M. and Holmgren, J., 1980. Isolation of Escherichia coli heat-labile enterotoxin by affinity chromatography: characterization of subunits. Curr. Microbiol., 3: 339344.