Eur. J. Tmmunol. 1992. 22: 2277-2281
Nils Lycke, Takao Tsuji' and Jan Holmgren Department of Medical Microbiology and Immunology, University of Goteborg, Goteborg and Department of Microbiologyv, Fujita-Gakuen University School of Medicine, Toyoake
Adjuvant activity of CT and LT require ADP-ribosyltransferase activity
2277
The adjuvant effect of Vibrio cholerae and Escherichia coti heat-labile enterotoxins is linked to their ADP-ribosyltransferase activity This study addressed the question of whether the mucosal adjuvant property of cholera toxin (CT) and the structurally closely related Escherichiu coli heat-labile toxin (LT) requires the enterotoxic and adenylate cycIaseicAMP activating property of these molecules. Therefore, we investigated the cytotoxic and adjuvant abilities of the enterotoxins and compared the results with those obtained with the non-toxic CTand LTderivatives; recombinant CTB (rCTB) and a mutated LT (mLT), which had a single amino acid substitution in position 112 (Glu-+Lys) of the A subunit. Detailed functional studies revealed that, in contrast to the enterotoxins, both rCTB and mLT lacked ADP-ribosylating and CAMP-stimulating abilities. However, similar membrane ganglioside GM1receptor binding ability of all the putative adjuvants was demonstrated. When the probe antigen, keyhole limpet hemocyanin (KLH), was given perorally together with CTor LTstrong gut mucosal anti-KLH immune responses were stimulated, whereas no or very low anti-KLH responses were seen in the groups which received antigen admixed with rCTB or the mLT. Moreover, the specific serum antibody responses to the various immunization protocols closely paralleled the local anti-KLH response in the gut. From these results it appears that the adjuvant mechanism of LT, and probably also of CT, is linked to the ability to ADP-ribosylate and to stimulate cAMP formation. However, this study does not unequivocally rule out other possibilities such as interactions by the A1 fragment of CTor LTwith other G-proteins than Gsa or events that parallel or precede the effects on the adenylate cyclasekAMP system. Thus, the levels of ADPribosylation and CAMP-induction that are required and the key event or target cell that is responsible for the adjuvant effect of CT and LT remain to be elucidated. Studies are underway to address these issues.
1 Introduction There is a great need for substances that can act as adjuvants on local mucosal immune responses to perorally administered antigens and which could be included in future oral vaccines. Cholera toxin (CT). which is one of the best studied and most potent mucosal immunogens, has recently attracted much interest because of its strong adjuvant effects on mucosal IgA responses to unrelated antigens when given perorally or intranasally together with these antigens [l-31. We have reported that the adjuvant action of CT is dose dependent and requires simultaneous peroral administration of antigen admixed with CT [2]. CT adjuvant greatly enhances priming as well as booster IgA responses in the intestinal lamina propria and promotes the development of long-term, perhaps life-long, immunological memory to unrelated antigens in the gut [4, 51.
[I 105461
* The study was funded by grants from the Swedish Medical Research Council and the WHO Transdisease Vaccinology Program. Correspondence: Nils Lycke, Department of Medical Microbiology and Immunology, University of Goteborg, S-413 46 Giiteborg, Sweden Abbreviations: C T Cholera toxin rCTB: Recombinant B subunit of CT L T E. coli heat-labile toxin mLT Mutated E. coli heat-labile toxin SFC: Spot-forming cells
0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1992
In a series of in vitro and in vivo experiments we have sought to understand the mechanismis behind the adjuvant effect of CT. The conclusion to be made from these experiments is that CT is likely to have several immunomodulatory actions on the local immune system in the gut [4]. We have reported that CT enhances antigen presentation in macrophages as well as in epithelial cells and promotes differentiation and isotype switching in B cells [6-81. Others have provided evidence of selective inhibitory actions by CTon TI%lcells, while T H cells ~ seem to be functionally intact, i.e. are unaffected in lymphokine production upon activation, in the presence of CT [9]. In addition, the toxin increases gut permeability for luminal antigens and thereby perhaps provides increased access of certain antigens to the gut mucosal immune system [lo]. CT is presumed to cause diarrhea by stimulating intracellular cAMP in enterocytes [ll]. It is still unresolved whether activation of the adenylate cyclase/cAMP system is also required for the adjuvant effect of CT.Therefore, a key question for the possible use of CTas an adjuvant in humans is whether the adjuvant and enterotoxic activities can be separated. In our studies we have consistently been unable to detect any adjuvant function of the non-toxic, highly purified or recombinant B subunit of CT (CTB) in vivo [2, 5 , lO].This is in contrast to the adjuvant effect by CTB on mucosal immune responses to influenza virus reported by two other laboratories [12, 131. However, yet other laboratories, of which some also used the influenza virus system, have provided data which support the lack of adjuvant function of CTB [ 3 ] . 0014-2980/92/0909-2277$3.50+ ,2510
2278
N.Lycke. T. Tsuji and J. Holmgren
This study used E. coli heat-labile toxin (LT) enterotoxin, which is closely similar to CT both in structure and function and which also has been reported to exert adjuvant effects on mucosal immune responses [141.We investigated the role ot the toxic activity of this molecule for the adjuvant function by comparing the intact LTwith a mutant LTwith a single amino acid substitution in the A subunit resulting in an atoxic molecule [15]. We also used a completely nontoxic, genetically engineered, recombinant CTB (rCTB) and compared the adjuvant effect of this molecule with that of the whole toxin [16]. In contrast to commercially available CTB preparations, which are derived from the whole toxin, the rCTB is absolutely devoid of the toxic A subunit since the gene for this subunit is not included in the rCTB expression system [16].
2 Materials and methods 2.1 Mice, toxins and derivatives All experiments were carried out with C57BL/6 female mice, aged 8-12 weeks at the onset of the priming immunizations. In each experiment age-matched control female animals were used. CT was purchased from List Biological Laboratories Inc. (Campbell, CA) and rCTB was produced in Vibrio cholerae, strain CVD103, a derivative of the classical Inaba strain 569B, as described [16]. E. coli LT was purified by the method of Clements and Finkelstein [14]. The mutant LT (mLT) was generated from the same strain through chemical mutagenesis using hydroxylamine. A plasmid encoding the mLT was obtained which had a structure identical to that of the parent LT except for having lysine instead of glutamic acid in position 112 in the Al-region [15]. The chemical and biological properties of this mLT have reccntly been studied and the results show that the single amino acid substitution results in an LT molecule without any toxicity but with intact membrane GM1-receptor binding ability [15, 171. 2.2 Immunizations
Purified CT, rCTB, LT or mLT, 10 yg/dose, was administered perorally together with keyhole limpet hemocyanin (KLH; Sigma, St. Louis, MO), 2.5 mg/dose, in a volume of 0.5 ml phosphate-buffered saline (PBS) containing 3% (w/v) NaHC03 through a baby-feeding tube under light ether anesthesia [ 2 ] .Groups of six mice for each immunization protocol were immunized three times at intervals of 10 days between each administration of antigen plus putative adjuvant. 2.3 ELISPOT assay of antibody-producing cells in the lamina propria The procedures used for preparation of lamina propria cells and for determinations of anti-KLH and anti-CT spot (antibody)-forming cells (SFC) by the ELISPOT method have been described in detail previously [2]. The small intestines from two mice were pooled and analyzed in
Eur. J. Immunol. 1992. 22: 2277-2281
groups of six mice as indicated. Lamina propria cells were analyzed in duplicates for antigen-specific IgA SFC using a goat anti-mouse IgA (Cappel, Malven, CA) at 1/300 dilution followed by a rabbit anti-goat Ig-antibody conjugated to horseradish peroxidase (HRP; Dako, Glostrup, Denmark) used at 1/200 dilution. Antigen-specific antibody production in the lamina propria was expressed as IgA SFC f SD of each group. 2.4 ELISA The mice were bled immediately prior to sacrifice and the immune sera from pairs of mice were pooled, providing three sera per group, and stored at - 20 "C until assayed. Briefly, polystyrene microtiter plates (Nunc, Roskilde, Denmark) were coated with KLH (200 yg/ml) or GM1 ganglioside (0.5 nmol/ml; Sigma) followed by CT (0.5 pg/ml) as described [2]. After incubation with PBS/O.1% BSA for 30 min at 37"C, followed by thorough washing, the sera were added at a 1/50 dilution and serial threefold dilutions were performed. The plates were incubated overnight at 4 "C. Bound antibodies were demonstrated with HRP-conjugated rabbit anti-mouse Ig antibodies (Dako) at a 1/200 dilution and visualized using o-phenylenediamine (OPD) substrate (1 mg/ml)/0.04% H202in citrate buffer (pH 4.5). The reaction was read in a Titertek Multiscan spectrophotometer (Flow Labs, Irvine, Scotland) at 450 nm. The anti-KLH and antitoxin titers were defined as the interpolated absorbance reading giving rise to a value 0.4 above background. The mice were analyzed in pairs and each pooled serum sample assayed in duplicate. Serum loglo titers were expressed as means f SD of three pooled sera in each group. 2.5 Assay systems for determining toxicity and receptor binding ability 2.5.1 Rabbit ileal loop test The rabbit abdomen was opened under anesthesia and 5-cm loops were ligated in the mid part of the small intestine as described [4]. Purified CT, LT, rCTB or mLT in various doses in duplicates in 0.5 ml PBS was injected into the loops and the abdomen was closed. After 18 h the rabbit was killed and the ligated loop with its fluid content was weighed and its length determined. Values for fluid accumulation were expressed as weight per length ratio (mg/cm) and represents means of duplicates of two experiments SD < 5 % .
+
2.5.2 CAMP-stimulation test A commercial CAMP-test kit was used according to the manufacturers' instructions (Amersham Int., Amersham, GB). Briefly enriched spleen Bcells, i.e. depleted of Thy-1.2+ cells as described [MI, at 107/ml in Iscove's medium containing 5% fetal calf serum were incubated in the presence of 1 pg/ml of CT, rCTB, LTand mLT. After 60 min the cells were harvested and washed thoroughly in cold PBS before the determination of CAMP-concentrations. Values represent pg/ml of CAMP f SD < 5% of two independent experiments.
Adjuvant activity of CT and LT require ADP-ribosyltransferase activity
Eur. J. Immunol. 1992. 22: 2277-2281
2.5.3 ADP-ribosyltransferaseactivity The method previously described was used [17]. Briefly, NAD : agmatine ADP-ribosyltransferase activity was determined by assaying 10 pg/ml of each putative adjuvant and assessing the ADP-ribosylagmatine formation through incorporation of [U-14C]adenine. Each sample contained 50 mM potassium phosphate (pH 7.5), 100 pM GTP, 5 mM MgC12, 100 mM [U-I4C]adenine-labeled NAD, 10 mM agmatine, 0.1 mg/ml of ovalbumin, and the respective toxins and toxin derivatives as described (total volume 300 pl). After 60 min at 30"C, three 50 yl samples were transferred to AGl-X2 column which were washed four times with 1.25 ml water. Eluates containing [U-14C]adenine-labeled ADP-ribosyl-agmatine were collected for determinations of radioactivity. The values represent counts per minute (cpm). SD was < 5% in three independent experiments. 2.5.4 GM1-receptor binding ability The membrane GM1-receptor binding ability of CT, rCTB, LT and mLT was determined by incubating 1 yg/ml and serial fivefold dilutions of these reagents in the GM1ELISA as described [4]. Values represent means of duplicates and three separate experiments. SD was < 5 % . 2.6 Comparison of B cell epitopes between LT and mLT GM1-ELISA and a panel of mouse monoclonal antibodies (mAb) directed against human and porcine LT and CT epitopes were used to evaluate whether the single amino acid mutation in position 112 of the A subunit had caused steric changes in the LT molecule. The mAb used recognized the A or B subunits of CT and/or LT as indicated in Table 2 [19]. GM1-ELISA was performed in a microtiter
Table 1. Functional properties of CT, rCTB, LT and mLT AgentaJ CAMP-activationbj
/
CT
ADP-ribsylating activityf)
Loop
GM1-
testJ)
binding)
Nils Lycke, Takao Tsuji' and Jan Holmgren Department of Medical Microbiology and Immunology, University of Goteborg, Goteborg and Department of Microbiologyv, Fujita-Gakuen University School of Medicine, Toyoake
Adjuvant activity of CT and LT require ADP-ribosyltransferase activity
2277
The adjuvant effect of Vibrio cholerae and Escherichia coti heat-labile enterotoxins is linked to their ADP-ribosyltransferase activity This study addressed the question of whether the mucosal adjuvant property of cholera toxin (CT) and the structurally closely related Escherichiu coli heat-labile toxin (LT) requires the enterotoxic and adenylate cycIaseicAMP activating property of these molecules. Therefore, we investigated the cytotoxic and adjuvant abilities of the enterotoxins and compared the results with those obtained with the non-toxic CTand LTderivatives; recombinant CTB (rCTB) and a mutated LT (mLT), which had a single amino acid substitution in position 112 (Glu-+Lys) of the A subunit. Detailed functional studies revealed that, in contrast to the enterotoxins, both rCTB and mLT lacked ADP-ribosylating and CAMP-stimulating abilities. However, similar membrane ganglioside GM1receptor binding ability of all the putative adjuvants was demonstrated. When the probe antigen, keyhole limpet hemocyanin (KLH), was given perorally together with CTor LTstrong gut mucosal anti-KLH immune responses were stimulated, whereas no or very low anti-KLH responses were seen in the groups which received antigen admixed with rCTB or the mLT. Moreover, the specific serum antibody responses to the various immunization protocols closely paralleled the local anti-KLH response in the gut. From these results it appears that the adjuvant mechanism of LT, and probably also of CT, is linked to the ability to ADP-ribosylate and to stimulate cAMP formation. However, this study does not unequivocally rule out other possibilities such as interactions by the A1 fragment of CTor LTwith other G-proteins than Gsa or events that parallel or precede the effects on the adenylate cyclasekAMP system. Thus, the levels of ADPribosylation and CAMP-induction that are required and the key event or target cell that is responsible for the adjuvant effect of CT and LT remain to be elucidated. Studies are underway to address these issues.
1 Introduction There is a great need for substances that can act as adjuvants on local mucosal immune responses to perorally administered antigens and which could be included in future oral vaccines. Cholera toxin (CT). which is one of the best studied and most potent mucosal immunogens, has recently attracted much interest because of its strong adjuvant effects on mucosal IgA responses to unrelated antigens when given perorally or intranasally together with these antigens [l-31. We have reported that the adjuvant action of CT is dose dependent and requires simultaneous peroral administration of antigen admixed with CT [2]. CT adjuvant greatly enhances priming as well as booster IgA responses in the intestinal lamina propria and promotes the development of long-term, perhaps life-long, immunological memory to unrelated antigens in the gut [4, 51.
[I 105461
* The study was funded by grants from the Swedish Medical Research Council and the WHO Transdisease Vaccinology Program. Correspondence: Nils Lycke, Department of Medical Microbiology and Immunology, University of Goteborg, S-413 46 Giiteborg, Sweden Abbreviations: C T Cholera toxin rCTB: Recombinant B subunit of CT L T E. coli heat-labile toxin mLT Mutated E. coli heat-labile toxin SFC: Spot-forming cells
0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1992
In a series of in vitro and in vivo experiments we have sought to understand the mechanismis behind the adjuvant effect of CT. The conclusion to be made from these experiments is that CT is likely to have several immunomodulatory actions on the local immune system in the gut [4]. We have reported that CT enhances antigen presentation in macrophages as well as in epithelial cells and promotes differentiation and isotype switching in B cells [6-81. Others have provided evidence of selective inhibitory actions by CTon TI%lcells, while T H cells ~ seem to be functionally intact, i.e. are unaffected in lymphokine production upon activation, in the presence of CT [9]. In addition, the toxin increases gut permeability for luminal antigens and thereby perhaps provides increased access of certain antigens to the gut mucosal immune system [lo]. CT is presumed to cause diarrhea by stimulating intracellular cAMP in enterocytes [ll]. It is still unresolved whether activation of the adenylate cyclase/cAMP system is also required for the adjuvant effect of CT.Therefore, a key question for the possible use of CTas an adjuvant in humans is whether the adjuvant and enterotoxic activities can be separated. In our studies we have consistently been unable to detect any adjuvant function of the non-toxic, highly purified or recombinant B subunit of CT (CTB) in vivo [2, 5 , lO].This is in contrast to the adjuvant effect by CTB on mucosal immune responses to influenza virus reported by two other laboratories [12, 131. However, yet other laboratories, of which some also used the influenza virus system, have provided data which support the lack of adjuvant function of CTB [ 3 ] . 0014-2980/92/0909-2277$3.50+ ,2510
2278
N.Lycke. T. Tsuji and J. Holmgren
This study used E. coli heat-labile toxin (LT) enterotoxin, which is closely similar to CT both in structure and function and which also has been reported to exert adjuvant effects on mucosal immune responses [141.We investigated the role ot the toxic activity of this molecule for the adjuvant function by comparing the intact LTwith a mutant LTwith a single amino acid substitution in the A subunit resulting in an atoxic molecule [15]. We also used a completely nontoxic, genetically engineered, recombinant CTB (rCTB) and compared the adjuvant effect of this molecule with that of the whole toxin [16]. In contrast to commercially available CTB preparations, which are derived from the whole toxin, the rCTB is absolutely devoid of the toxic A subunit since the gene for this subunit is not included in the rCTB expression system [16].
2 Materials and methods 2.1 Mice, toxins and derivatives All experiments were carried out with C57BL/6 female mice, aged 8-12 weeks at the onset of the priming immunizations. In each experiment age-matched control female animals were used. CT was purchased from List Biological Laboratories Inc. (Campbell, CA) and rCTB was produced in Vibrio cholerae, strain CVD103, a derivative of the classical Inaba strain 569B, as described [16]. E. coli LT was purified by the method of Clements and Finkelstein [14]. The mutant LT (mLT) was generated from the same strain through chemical mutagenesis using hydroxylamine. A plasmid encoding the mLT was obtained which had a structure identical to that of the parent LT except for having lysine instead of glutamic acid in position 112 in the Al-region [15]. The chemical and biological properties of this mLT have reccntly been studied and the results show that the single amino acid substitution results in an LT molecule without any toxicity but with intact membrane GM1-receptor binding ability [15, 171. 2.2 Immunizations
Purified CT, rCTB, LT or mLT, 10 yg/dose, was administered perorally together with keyhole limpet hemocyanin (KLH; Sigma, St. Louis, MO), 2.5 mg/dose, in a volume of 0.5 ml phosphate-buffered saline (PBS) containing 3% (w/v) NaHC03 through a baby-feeding tube under light ether anesthesia [ 2 ] .Groups of six mice for each immunization protocol were immunized three times at intervals of 10 days between each administration of antigen plus putative adjuvant. 2.3 ELISPOT assay of antibody-producing cells in the lamina propria The procedures used for preparation of lamina propria cells and for determinations of anti-KLH and anti-CT spot (antibody)-forming cells (SFC) by the ELISPOT method have been described in detail previously [2]. The small intestines from two mice were pooled and analyzed in
Eur. J. Immunol. 1992. 22: 2277-2281
groups of six mice as indicated. Lamina propria cells were analyzed in duplicates for antigen-specific IgA SFC using a goat anti-mouse IgA (Cappel, Malven, CA) at 1/300 dilution followed by a rabbit anti-goat Ig-antibody conjugated to horseradish peroxidase (HRP; Dako, Glostrup, Denmark) used at 1/200 dilution. Antigen-specific antibody production in the lamina propria was expressed as IgA SFC f SD of each group. 2.4 ELISA The mice were bled immediately prior to sacrifice and the immune sera from pairs of mice were pooled, providing three sera per group, and stored at - 20 "C until assayed. Briefly, polystyrene microtiter plates (Nunc, Roskilde, Denmark) were coated with KLH (200 yg/ml) or GM1 ganglioside (0.5 nmol/ml; Sigma) followed by CT (0.5 pg/ml) as described [2]. After incubation with PBS/O.1% BSA for 30 min at 37"C, followed by thorough washing, the sera were added at a 1/50 dilution and serial threefold dilutions were performed. The plates were incubated overnight at 4 "C. Bound antibodies were demonstrated with HRP-conjugated rabbit anti-mouse Ig antibodies (Dako) at a 1/200 dilution and visualized using o-phenylenediamine (OPD) substrate (1 mg/ml)/0.04% H202in citrate buffer (pH 4.5). The reaction was read in a Titertek Multiscan spectrophotometer (Flow Labs, Irvine, Scotland) at 450 nm. The anti-KLH and antitoxin titers were defined as the interpolated absorbance reading giving rise to a value 0.4 above background. The mice were analyzed in pairs and each pooled serum sample assayed in duplicate. Serum loglo titers were expressed as means f SD of three pooled sera in each group. 2.5 Assay systems for determining toxicity and receptor binding ability 2.5.1 Rabbit ileal loop test The rabbit abdomen was opened under anesthesia and 5-cm loops were ligated in the mid part of the small intestine as described [4]. Purified CT, LT, rCTB or mLT in various doses in duplicates in 0.5 ml PBS was injected into the loops and the abdomen was closed. After 18 h the rabbit was killed and the ligated loop with its fluid content was weighed and its length determined. Values for fluid accumulation were expressed as weight per length ratio (mg/cm) and represents means of duplicates of two experiments SD < 5 % .
+
2.5.2 CAMP-stimulation test A commercial CAMP-test kit was used according to the manufacturers' instructions (Amersham Int., Amersham, GB). Briefly enriched spleen Bcells, i.e. depleted of Thy-1.2+ cells as described [MI, at 107/ml in Iscove's medium containing 5% fetal calf serum were incubated in the presence of 1 pg/ml of CT, rCTB, LTand mLT. After 60 min the cells were harvested and washed thoroughly in cold PBS before the determination of CAMP-concentrations. Values represent pg/ml of CAMP f SD < 5% of two independent experiments.
Adjuvant activity of CT and LT require ADP-ribosyltransferase activity
Eur. J. Immunol. 1992. 22: 2277-2281
2.5.3 ADP-ribosyltransferaseactivity The method previously described was used [17]. Briefly, NAD : agmatine ADP-ribosyltransferase activity was determined by assaying 10 pg/ml of each putative adjuvant and assessing the ADP-ribosylagmatine formation through incorporation of [U-14C]adenine. Each sample contained 50 mM potassium phosphate (pH 7.5), 100 pM GTP, 5 mM MgC12, 100 mM [U-I4C]adenine-labeled NAD, 10 mM agmatine, 0.1 mg/ml of ovalbumin, and the respective toxins and toxin derivatives as described (total volume 300 pl). After 60 min at 30"C, three 50 yl samples were transferred to AGl-X2 column which were washed four times with 1.25 ml water. Eluates containing [U-14C]adenine-labeled ADP-ribosyl-agmatine were collected for determinations of radioactivity. The values represent counts per minute (cpm). SD was < 5% in three independent experiments. 2.5.4 GM1-receptor binding ability The membrane GM1-receptor binding ability of CT, rCTB, LT and mLT was determined by incubating 1 yg/ml and serial fivefold dilutions of these reagents in the GM1ELISA as described [4]. Values represent means of duplicates and three separate experiments. SD was < 5 % . 2.6 Comparison of B cell epitopes between LT and mLT GM1-ELISA and a panel of mouse monoclonal antibodies (mAb) directed against human and porcine LT and CT epitopes were used to evaluate whether the single amino acid mutation in position 112 of the A subunit had caused steric changes in the LT molecule. The mAb used recognized the A or B subunits of CT and/or LT as indicated in Table 2 [19]. GM1-ELISA was performed in a microtiter
Table 1. Functional properties of CT, rCTB, LT and mLT AgentaJ CAMP-activationbj
/
CT
ADP-ribsylating activityf)
Loop
GM1-
testJ)
binding)
