The identification of plant lectins with mucosal adjuvant activity

Abstract

To date, the most potent mucosal vaccine adjuvants to be identified have been bacterial toxins. The present data demonstrate that the type 2 ribosome-inactivating protein (type 2 RIP), mistletoe lectin I (ML-I) is a strong mucosal adjuvant of plant origin. A number of plant lectins were investigated as intranasal (i.n.) coadjuvants for a bystander protein, ovalbumin (OVA). As a positive control, a potent mucosal adjuvant, cholera toxin (CT), was used. Co-administration of ML-I or CT with OVA stimulated high titres of OVA-specific serum immunoglobulin G (IgG) in addition to OVA-specific IgA in mucosal secretions. CT and ML-I were also strongly immunogenic, inducing high titres of specific serum IgG and specific IgA at mucosal sites. None of the other plant lectins investigated significantly boosted the response to co-administered OVA. Immunization with phytohaemagglutinin (PHA) plus OVA elicited a lectin-specific response but did not stimulate an enhanced response to OVA compared with the antigen alone. Intranasal delivery of tomato lectin (LEA) elicited a strong lectin-specific systemic and mucosal antibody response but only weakly potentiated the response to co-delivered OVA. In contrast, administration of wheatgerm agglutinin (WGA) or Ulex europaeus lectin 1 (UEA-I) with OVA stimulated a serum IgG response to OVA while the lectin-specific responses (particularly for WGA) were relatively low. Thus, there was not a direct correlation between immunogenicity and adjuvanticity although the strongest adjuvants (CT, ML-I) were also highly immunogenic.

Introduction

Since most pathogens colonize and invade the host at mucosal surfaces, the induction of immunity at these sites is a rational and attractive approach to prevent infection.¹ Mucosal routes for vaccine delivery are non-invasive so administration is relatively simple and inexpensive. Furthermore, the potential to induce mucosal and systemic immune responses after mucosal vaccine delivery allows the possibility of effective immunization against many diseases. Specific immunoglobulin A (IgA) alone can protect mice against intranasal (i.n.) infection with influenza virus,² and intestinal infection with Vibrio cholerae.³ However, mucosal delivery of non-replicating antigens generally does not stimulate strong immune responses, requires multiple high doses⁴ and may result in systemic unresponsiveness.¹ A number of strategies may be used to enhance responses to mucosally delivered vaccines including live bacterial vectors,⁵ biodegradable microparticles, or liposomes.^6,7 The most powerful mucosal adjuvants identified to date are cholera toxin produced by V. cholerae and heat-labile enterotoxin (LT) from enterotoxigenic strains of Escherichia coli.^8,9 Stimulation of toxin-specific local and systemic responses and responses to co-administered antigens after mucosal application distinguish these molecules from most soluble proteins.^10,11 Although toxicity prevents clinical application, molecules with retained adjuvanticity but of low toxicity have been generated by site-directed mutagenesis.¹²

Plant lectins are proteins containing at least one non-catalytic domain, which binds specifically and reversibly to a monosaccharide or oligosaccharide.¹³ Certain plant lectins have been investigated for specific targeting of molecules to the mucosal epithelium.¹⁴ For example Ulex europaeus agglutinin-1 (UEA-I), a fucose-specific lectin, selectively labels antigen-sampling M cells in the murine intestine.¹⁵ The principal inductive sites for i.n. antigens in humans and mice appear to be the palatine and nasopharyngeal tonsils, which form the nasopharyngeal-associated lymphoreticular tissues (NALT).¹⁶ M cells in hamster NALT are distinguished by the expression of glycoconjugates bearing terminal α(1-3)-linked galactose.¹⁴ Griffonia simplicifolia I isolectin-B₄ (GS I-B₄, galactose-specific) targeted follicle-associated epithelium (FAE) after i.n. delivery to hamsters. The lectin was endocytosed and i.n. immunization with a horseradish peroxidase (HRP) conjugate of either GS I-B₄ or wheatgerm agglutinin (WGA) stimulated a serum IgG response to HRP. Uptake of plant lectins from the gut has also been described in mice¹⁵ and humans,¹⁷ highlighting the potential of plant lectins as mucosal vaccine carriers. Receptor-mediated binding of lectins to mucosae was proposed as an important determinant of mucosal immunogenicity.¹⁸ We have previously demonstrated that some plant lectins, in particular mistletoe lectin I (ML-I) are highly immunogenic after mucosal delivery in mice.¹⁹ The present work is the first detailed investigation of plant lectins as mucosal adjuvants. Plant lectins were compared with cholera toxin (CT) as adjuvants for a bystander antigen, ovalbumin (OVA).

Materials and methods

Antigens

CT, OVA (type V, hen egg) and WGA were obtained from Sigma (Poole, UK). Phytohaemagglutinin (PHA) and ML-I were prepared as described previously.^20,21 UEA-I and tomato lectin (LEA) were obtained from Vector Laboratories (Peterborough, UK).

Animals

Eight-week-old female BALB/c mice (Harlan Olac, Bicester, UK) were given free access to commercial stock diet (Labsure, Manea, UK) and water.

Mucosal immunization

Groups of mice (n = 10) were bled 1 week prior to the first immunization. Mice were immunized on days 1, 14, 28 and 42, with phosphate-buffered saline (PBS), OVA (10 µg) alone, or OVA (10 µg) mixed with CT (1 µg), ML-I (1 µg), LEA (10 µg), PHA (10 µg), WGA (10 µg), or UEA-I (10 µg). While restrained, mice were dosed with 30 µl of each preparation (15 µl placed over each nostril) using fine tips attached to a pipette. Mice were held in place until the liquid was inhaled and were not anaesthetized for the procedure.

Collection of blood and mucosal secretions

Blood samples were collected 1 day before each immunization from the tail vein following a 10-min incubation at 37°. Two weeks after the final immunization, animals were terminally anaesthetized (hypnorm plus diazepam) to allow collection of salivary and vaginal secretions. Mice were then killed by anaesthetic overdose followed by exsanguination. Blood was immediately collected, centrifuged and the serum was stored at −20°.

Absorbent cellulose wicks (Whatman International, Maidstone, UK) were used for collection of saliva and vaginal fluid.²² Wash fluid [0·01 m PBS, 50 mm ethylenediaminetetraacetic acid (EDTA), 5 mm phenylmethylsulphonyl fluoride (PMSF), 5 µg/ml aprotinin (Sigma, Poole, UK)] (ice-cold) was used for elution of antibody from wicks and for nasal and intestinal washes. Saliva was collected by the insertion of a wick tip into the mouth for 2 min and antibody was extracted into 400 µl mucosal wash fluid. Vaginal fluid was collected by repeated flushing and aspiration of 50 µl wash fluid and insertion of a wick for 2 min and antibody was extracted into 400 µl wash fluid. Nasotracheal washes were collected from decapitated animals by back-flushing 0·5 ml of wash fluid from the trachea. Intestinal wash was obtained by flushing the small intestine with 10 ml of ice-cold wash fluid. All secretions were stored at −20° until required for analysis.

Enzyme-linked immunosorbent assays (ELISA)

Microtitre plates (Immunolon 4, Dynatech) were coated with 75 µl of antigen per well (1 µg/ml for CT/lectins, 50 µg/ml when measuring bystander responses to OVA) in carbonate–bicarbonate buffer, pH 9·6 and incubated at 4° overnight. After washing, plates were blocked with 2% gelatin/dilution buffer and incubated at 37° for 1 hr. Plates were washed; then samples were added, serially diluted and incubated at 37° for 1 hr. Biotinylated antiserum in dilution buffer was added and incubated at 37° for 1 hr. After further washes, ExtrAvidin® peroxidase (Sigma) at a dilution of 1 : 750, in dilution buffer was added and incubated at 37° for 30 min. Plates were washed and 50 µl/well of developing solution [TMB microwell peroxidase substrate (1-C) Kirkegaard and Perry Laboratories, Gaithersburg, MD] was added and incubated in the dark with shaking at 37° for 30 min. The reaction was stopped by addition of 1 m H₂SO₄ and the absorbance was read at 450 nm.

ELISA dilution buffers were as follows: CT [PBS + 0·1% Tween (PBST)], OVA (PBST), WGA (100 mm N-acetylglucosamine/PBST), PHA (0·1% Fetuin/PBST), UEA-I (30 mm l-fucose/PBST), LEA [Chitin hydrolysate (1 : 200) (Vector Laboratories)/PBST], ML-I (100 mm d-galactose/PBST). Sugars were included in dilution buffers to prevent sugar-mediated interactions between immunoglobulins and lectins.¹⁹ Working dilutions of anti-IgG (1 : 8000) and IgA (1 : 2600) biotinylated capture antisera (Sigma) were determined after preliminary assays with preimmune and pooled positive sera. Working dilutions of IgG subclass antisera (Serotec) were as recommended by the manufacturers (IgG1 (1 : 4000), IgG2a (1 : 4000), IgG2b (1 : 2000), IgG3 (1 : 2000)). End-point titres were determined as the dilution of a sample giving an OD value of 0·1 units greater than the mean of control samples at the same dilution.

Total IgA was quantified as specific IgA with the following modifications: plates were coated with goat anti-mouse IgA [1 : 8000 (α-chain specific, Sigma)], PBST was used as diluent and 2% gelatin in PBST was used as blocking solution. Total IgA levels were calculated from the linear region of the IgA (IgA kappa, Sigma) standard curve. Total IgA end-point titres were determined as the dilution of a sample giving an OD value of 0·1 units greater than buffer alone.

Statistics

Data are presented as arithmetic means, calculated by dividing the sum of the reciprocal titres by the number of samples in each group. anova was used to test for significance between groups. Where standard deviations were significantly different between groups, a non-paramentic test (Kruskal–Wallis test with Dunn's multiple comparison post test) was used to assess significance. Kruskal–Wallis non-parametric test with Dunn's multiple comparison post test was also used to assess the significance of the total IgA data.

Results

Adjuvant effect of plant lectins on OVA-specific serum antibody responses

Two weeks after a single immunization, OVA-specific serum IgG was detected in six of the ten mice immunized with CT + OVA and one of the ten mice immunized with ML-I + OVA (Fig. 1) but not in the other groups. After a second dose, specific IgG was detected in all mice immunized with CT + OVA (mean titre 40321) and in nine of the ten mice immunized with ML-I + OVA (mean titre 11090). Of the other groups, specific IgG was only detected in mice immunized with UEA-I + OVA (mean titre 91). After four doses, the highest mean IgG titres were in mice immunized with CT + OVA and ML-I + OVA, being approximately 286-fold and 118-fold higher, respectively, than in mice which received OVA alone. Titres in mice immunized with PHA + OVA were similar to those in mice administered with OVA alone. Administration of LEA + OVA resulted in approximately a fivefold increase in mean titre compared with OVA alone. Delivery of WGA or UEA-I with OVA, respectively, led to 41- and 51-fold increases in mean serum IgG anti-OVA titres compared with OVA alone. In contrast to the groups which received CT + OVA and ML-I + OVA, responses in these groups were highly variable. As a result, after the final dose only the CT + OVA and ML-I + OVA groups (difference not significant between groups) had mean OVA-specific IgG titres significantly higher than the OVA only group. Titres in these groups were also significantly higher than in the PHA + OVA group (P < 0·001).

Figure 1.

OVA-specific serum IgG antibody titres from mice immunized intranasally on days 1, 14, 28 and 42 with either OVA (10 µg) alone or OVA (10 µg) together with CT (1 µg), ML-I (1 µg), LEA (10 µg), PHA (10 µg), WGA (10 µg) or UEA-I (10 µg). Sera were collected 1 day before each immunization and at the termination of the study. (a) Serum IgG titres after one dose (day 13); (b) serum IgG titres after two doses (day 27); (c) serum IgG titres after three doses (day 41); (d) serum IgG titres after the final dose (day 56). Points refer to individual data and the symbol (–) represents the mean titre. P values in parentheses refer to significance of data compared with the OVA only group.

In contrast to the high levels of specific IgG, very low titres of OVA-specific serum IgA were detected (data not presented). After the final dose, significant levels of OVA-specific serum IgA were only detected in mice immunized with CT + OVA (mean titre, 220) and ML-I + OVA (mean titre, 80).

OVA-specific IgG subclass patterns

The IgG1 titres were similar to the titres of OVA-specific IgG in most groups (Fig. 2). In mice immunized with CT + OVA and ML-I + OVA, respectively, the mean titres were approximately 450-fold and 255-fold higher than in mice immunized with OVA alone. Titres in the CT + OVA group were significantly higher than in all groups except ML-I + OVA (P < 0·05) and titres in the ML-I + OVA group were significantly higher than in groups which received OVA alone or PHA + OVA (P < 0·001).

Figure 2.

OVA-specific serum IgG1 antibody titres measured in mice immunized intranasally on days 1, 14, 28 and 42 with either OVA (10 µg) alone or OVA (10 µg) together with CT (1 µg), ML-I (1 µg), LEA (10 µg), PHA (10 µg), WGA (10 µg), or UEA-I (10 µg). Samples were collected 2 weeks after the final immunization. Points refer to individual data and the symbol (–) represents the mean titre. P values in parentheses refer to significance of data compared with the OVA only group.

OVA-specific IgG2a was detected in eight of ten and two of ten mice immunized with CT + OVA (mean titre 561) and ML-I + OVA (mean titre 331), respectively, but not in the other groups. Specific IgG2b was only detected in two of the mice immunized with CT + OVA and in none of the other groups. Specific IgG3 was not detected. These data are strikingly different to the CT-specific IgG isotype responses in these mice where relatively high titres of specific IgG2a and significant levels of IgG2b and IgG3 were detected (Table 1).

Table 1.

CT- and plant lectin-specific antibody titres in sera and secretions of mice after four doses of CT/plant lectin+OVA.

	Serum IgG and IgG subclass titre					IgA titre
Antigen	IgG1	IgG2a	IgG2b	IgG3	Serum	Saliva	Vagina	Nasal	Gut
CT	532 480	339 840	83 200	3680	1200	4480	409·6	358·8	3200	211·2
ML-I	337 960	409 600	2880	900	–	1840	64	440·3	819·2	33·6
LEA	61 440	78 080	360	10	–	360	11	10·2	49·6	5·2
PHA	8560	7160	–	–	–	210	–	–	4·6	0·2
WGA	140	20	–	–	–	50	0·2	0·6	1·4	–
UEA-I	10 880	18 440	10	–	–	30	1·8	1	5	–
OVA	180	30	–	–	–	–	–	–	–	–

Mice were immunized by the i.n. route on days 1, 14, 28 and 42 with OVA (10 µg) either alone or together with CT (1 µg), ML-I (1 µg), LEA (10 µg), PHA (10 µg), WGA (10 µg), or UEA-I (10 µg). Samples were collected 2 weeks after the final immunization. The data presented are mean titres from 10 mice in each case. ELISAs were run using a coating concentration of 1 µg/ml antigen in all cases.

Adjuvant effect of plant lectins on OVA-specific mucosal IgA responses

Specific IgA was detected at all mucosal sites sampled in mice immunized with CT + OVA or ML-I + OVA with no significant difference between the two groups. OVA-specific salivary IgA was not detected in mice immunized with OVA alone, LEA + OVA, or PHA + OVA (Fig. 3) but was detected in two and four mice immunized with WGA + OVA and UEA-I + OVA, respectively. In contrast, specific salivary IgA was measured in nine of ten and ten of ten mice immunized with CT + OVA and ML-I + OVA, respectively, with a twofold higher mean titre in the CT + OVA group. Mean titres in the CT + OVA group were significantly higher (P < 0·05) than in all groups except ML-I + OVA and mean titres in the ML-I + OVA group were higher (P < 0·05) than in all groups except CT + OVA and UEA-I + OVA.

Figure 3.

OVA-specific IgA antibody titres measured in secretions of mice immunized intranasally on days 1, 14, 28 and 42 with OVA (10 µg) alone or OVA (10 µg) together with CT (1 µg), ML-I (1 µg), LEA (10 µg), PHA (10 µg), WGA (10 µg), or UEA-I (10 µg). Data are titres measured 2 weeks after the final immunization in (a) saliva, (b) vaginal wash, (c) nasotracheal wash and (d) intestinal wash. Points refer to individual data and the symbol (–) represents the mean titre. P values in parentheses refer to significance of data compared with the OVA only group.

In vaginal washes, OVA-specific IgA was detected in nine of ten and seven of ten mice immunized with CT + OVA and ML-I + OVA, respectively (difference not significant). The mean titre was fourfold higher in the CT + OVA group but this was largely the result of one high responder. OVA-specific vaginal wash IgA was not detected in mice immunized with PHA + OVA and was detected in only one of the ten mice immunized with either OVA alone, LEA + OVA or WGA + OVA and in three of the ten mice immunized with UEA-I + OVA.

High titres of OVA-specific IgA were detected in nasotracheal washes from all mice immunized with CT + OVA or ML-I + OVA with approximately a fivefold higher titre in the CT + OVA group. OVA-specific nasotracheal wash IgA titres were significantly higher in mice immunized with CT + OVA than in all groups except ML-I + OVA and WGA + OVA. Remarkably, the OVA-specific nasotracheal wash IgA titres in these groups were comparable to the serum IgA titres. Total IgA titres in sera from mice immunized with CT + OVA and ML-I + OVA, respectively, were 33-fold and 73-fold higher than in nasotracheal washes. Specific IgA was detected in the nasotracheal washes of one of the ten mice immunized with OVA alone but not in any mice immunized with PHA + OVA. OVA-specific nasotracheal wash IgA was measured in seven of the ten mice immunized with WGA + OVA and five of the ten mice immunized with UEA-I or LEA + OVA, respectively.

OVA-specific IgA was detected in gut washes from all mice immunized with CT + OVA or ML-I + OVA with approximately a fourfold higher titre in the CT + OVA group. Titres in these groups were not significantly different from each other but were significantly higher (P < 0·05) than in all other groups.

CT/plant lectin-specific responses

CT-specific serum IgG was detected in all animals after a single dose of CT + OVA and titres increased with each subsequent dose. Specific antibody of all four IgG subclasses was detected in sera after four doses (Table 1). The highest titres were of IgG1, although CT-specific IgG2a, IgG2b and IgG3 were also detected. After the final dose, CT-specific serum IgA was detected in all mice with a mean titre of 4481. Specific IgA was also detected in all animals in all secretions sampled. Salivary IgA titres were relatively consistent between animals (approximately 10-fold lower mean titre than in serum). Total IgA titres in saliva from these mice were 1340-fold lower than in serum. Vaginal IgA titres were highly variable with a single high responder increasing the mean titre. High titres of CT-specific IgA were measured in nasotracheal washes from all animals with a mean titre comparable to the serum IgA titre. Specific IgA was also detected in intestinal washes of all mice but at a lower mean titre than at the other mucosal sites sampled.

Intranasal delivery of a single dose of ML-I + OVA stimulated the production of ML-I-specific IgG in three of ten mice. After the second and subsequent doses, high titres of specific IgG were detected in all mice. Analysis of ML-I-specific serum IgG subclasses found high titres of ML-I-specific IgG1 (Table 1). ML-I-specific IgG2a and IgG2b were also detected. ML-I-specific IgA was detected in all mice in serum and at all mucosal sites sampled after four doses. Titres in the saliva were consistent for all animals while a high responder increased the mean titre in the vaginal washes. High ML-I-specific IgA titres were measured in the nasotracheal washes of all animals. As with CT, the mean ML-I-specific titre in nasotracheal washes was comparable with the serum IgA titre (approximately twofold lower) which was remarkable as the total IgA titres in nasotracheal washes were 73-fold lower than in sera. Specific IgA was also detected in gut washes from all animals.

In mice immunized with LEA + OVA, LEA-specific serum IgG was detected in nine of the ten mice after a single dose. The titre increased after each subsequent dose to a relatively high level after the final immunization. Analysis of IgG subclasses found high titres of LEA-specific IgG1 and a low mean IgG2a titre (Table 1). Specific serum IgA was detected in seven of ten mice after four doses but at a low level. Specific IgA was also detected in all four mucosal secretions tested although in comparison to the data in the CT + OVA and ML-I + OVA groups the titres were highly variable.

PHA-specific serum IgG was detected in eight of the ten animals after the final dose. Of the IgG subclasses, only specific IgG1 was detected (Table 1). Low titres of specific serum IgA were detected in all animals. PHA-specific IgA was not detected in saliva or vaginal washes but was detected in nasotracheal washes of five of ten mice and gut washes of one of ten mice.

The lowest titres of specific antibody were elicited to WGA, even after four doses of WGA + OVA. Specific IgG1 was detected in two of the ten mice but other IgG subclasses were not detected (Table 1). Specific IgA was detected in a number of mice after four doses but at a maximum titre of 1 : 100. Low titres of specific IgA were measured in a small number of mice in saliva, vaginal washes and nasotracheal washes. These data are in contrast to the OVA-specific data from this group where OVA-specific serum IgG was detected in a number of mice.

UEA-I-specific IgG was detected in sera from eight of ten mice after the final dose. Specific IgG1 was detected in nine of ten (Table 1), IgG2a in one of ten mice and IgG2b and IgG3 were not detected. Specific serum IgA was detected in three of ten mice after the final dose. Relatively low levels of IgA were detected in saliva, vaginal washes and nasotracheal washes. Administration of the plant lectins chosen did not result in a significant increase in total IgA levels in sera or secretions.

Discussion

These data demonstrate that ML-I is a strong mucosal adjuvant. The most potent mucosal adjuvants identified to date have been toxins from bacteria infecting mucosal surfaces including CT,⁸ LT,⁹ pertussis toxin (PT)²³ and zonula occludens toxin (zot).²⁴ CT and LT are well-characterized mucosal immunogens and adjuvants for bystander proteins. These toxins contain separate A and B subunits. The B subunits mediate binding to cell surface receptors.²⁵ After binding of the B subunit, the A subunit reaches the cytosol and activates adenyl cyclase leading to an increase in [cAMP]_i.^10,11 LT is structurally and functionally similar to CT and is a comparable adjuvant.^26,27 In mice, CT strongly stimulates humoral and cell-mediated immune responses.¹⁰ In line with previous data, the present work found that CT was a potent immunogen and adjuvant after intranasal delivery.

ML-I is a type 2 ribosome-inactivating protein (type 2 RIP), containing an N-glycosidase A subunit responsible for the ribosome-inactivating activity and a galactose-specific carbohydrate-binding B subunit.²⁸ ML-I is highly immunogenic when administered to mice by the oral and intranasal routes.¹⁹ The present study indicates that ML-I is a strong mucosal adjuvant, stimulating anti-bystander antigen (OVA) antibody titres in sera and secretions only two- to fivefold lower than those induced by CT. To address the possibility that the potent adjuvanticity of ML-I could be due to LPS contamination, the lectin preparation was tested by the limulus assay. The levels of LPS detected were less than 0·1 IU/ml so less than 1 pg LPS was delivered per mouse with these preparations. In addition the OVA preparation was tested for LPS contamination. Of the batches used the highest dose administered per mouse was less than 60 pg. These low levels of endotoxin contamination of lectins or antigen are highly unlikely to account for the immune responses measured. It was suggested that differences in the immune responses elicited to CT and LT may be due to binding to different receptors.²⁷ Differences in receptor-binding specificity may also partly explain the different responses elicited to CT and ML-I. Mistletoe extracts are used in cancer therapy and have been reported to have immunomodulatory activity.^28,29 The ML-I A chain depurinates an adenine residue in 28S ribosomal RNA while the B (lectin) chain interacts with galactose residues of glycoproteins and glycolipids on the target cell surface. ML-I has been crystallized as a complex with β-d-galactose.³⁰ The lectin has been cloned and recombinant lectin has been produced which had similar properties to the natural lectin.³¹ It was concluded that both lectin and N-glycosidase activity were necessary for cytotoxicity. The recombinant B chain did not exert cytotoxic effects at the doses tested. The present findings show that ML-I is a strong mucosal adjuvant but since the lectin is highly toxic it is not clinically useful at present. The availability of methods for producing functional recombinant ML-I will allow investigations to determine if non-toxic rational mutants with adjuvant activity can be produced.

Co-administration of the other plant lectins with OVA did not result in a significant increase in OVA-specific antibody levels. Furthermore, since these lectins were used at a 10-fold higher dose than ML-I and CT, the data do not indicate that these possess significant adjuvant activity. Co-delivery of PHA (isotype E2L2) did not induce an enhanced response to OVA. PHA is a potent mitogen in vitro but data on immunomodulation in vivo are less clear.^32,33 LEA is a strong immunogen when given orally and particularly by the i.n. route¹⁹ but the present data do not indicate that LEA has strong adjuvant activity. Administration of WGA or UEA-I with OVA stimulated an enhanced OVA-specific serum IgG response after a number of doses. However, these lectins induced poor OVA-specific mucosal IgA responses in comparison with CT and ML-I. Intranasal immunization of hamsters with lectin–HRP conjugates stimulated HRP-specific serum IgG while antigen delivered alone or mixed with lectin (GS I-B₄) did not.¹⁴ If adjuvanticity is linked to receptor binding, the effects in different species may vary due to differences in glycoconjugate expression. The genetic background of mice influenced the immune response to CT³⁴ so similar effects are likely with other antigens. Similarly, orally delivered CTB is highly immunogenic in humans but relatively ineffective in mice.³⁵

The factors that dictate mucosal immunogenicity, are not fully understood although the ability to bind to cells is regarded as important.^18,36 However, a number of proteins with affinity for molecules on the surface of eukaryotic cells were not strong oral immunogens in mice.³⁷ In contrast with CT and CTB, none of the lectins (Ricinus communis, concanavalin A, PHA, peanut agglutinin) stimulated intestinal secretory IgA levels above those of controls. Feeding of peanut agglutinin, PHA and concanavalin A induced oral tolerance while the toxic lectin from Ricinus communis did not. It was concluded that binding was not sufficient to confer mucosal immunogenicity. These data are similar to the present results where administration of some lectins (PHA, WGA and UEA-I in particular) poorly stimulated local specific antibody production. However, LEA did stimulate local responses and ML-I stimulated high levels of specific antibody in local secretions. Binding of antigens to the epithelium is unlikely to be sufficient to confer mucosal immunogenicity but may be an essential first step. Binding of LT and other ADP-ribosylating toxins such as pertussis toxin to target cells is necessary for their immunogenicity.³⁸ Fusions of the ADP ribosyltransferase-active CTA1 subunit of CT and a carrier peptide which targets the complex to antigen-presenting cells is a strong mucosal adjuvant, comparable to CT in mice.¹⁰ The promiscuous binding of CTB to GM1 on various cell types was regarded as a disadvantage and responsible for toxicity. The presence of an active A subunit dramatically enhanced adjuvant activity of LT mutants especially after a single dose of vaccine.³⁹ In the present work an adjuvant effect was found after one or two doses of the molecules with enzymatically active A subunits (CT, ML-I) while with WGA and UEA-I an effect was only seen after multiple immunizations. There have been conflicting reports regarding the adjuvant properties of CTB.^10,11 Recent work found that recombinant CTB bound to antigen-presenting cells could directly co-stimulate cytokine production from T cells which was suggested as a mechanism for adjuvant activity.²⁶ An enhancing effect on antigen presentation and co-stimulation has been claimed to be the most important immunomodulating effect of CT.¹⁰ Increased permeability of the intestinal mucosae exposed to CT (but not CTB) has been shown and it was suggested that this contributed to adjuvant activity.⁴⁰ It is not yet known if lectins can exert similar effects.

Intranasal immunization with CT and ML-I + OVA stimulated high titres of OVA-specific IgG1 with little IgG2a or IgG2b. Oral immunization with soluble proteins together with CT as adjuvant results in the induction of Th2-type responses.^41,42 This is reflected both in the cytokines produced by the cells and in the production of high titres of serum IgG1 and low levels of IgG2a. When OVA was co-administered with CT, high titres of specific IgG1 were detected with no IgG2a and little IgG2b.⁴³ Intranasal immunization with Tetanus Toxoid (TT) fragment C + CT induced responses suggesting induction of both T helper type 1 (Th1) and Th2 type cells while when PT or a PT mutant was used as adjuvant a more biased Th2 type response was induced.²³ Data from i.n. immunization of C57/B6 mice with CT or LT plus OVA indicated that in contrast to CT, LT elicited a dominant IgG2a response suggesting a skew toward a mixed or more Th1 type response.²⁷ T cells stimulated with LT released interleukin-2, interferon-γ and interleukin-5 and low, but detectable interleukin-4 indicating a mixed response.⁴⁴

The present data indicate that the type of response elicited to the adjuvant and to the bystander protein may differ. High titres of specific IgG1 were detected to both OVA and CT but while relatively high titres of CT-specific IgG2a were measured there was little OVA-specific IgG2a. Delivery of ML-I + OVA led to similar results although the ML-I-specific IgG2a titres were relatively low. Previous work found higher OVA-specific IgG1 than IgG2a titres after delivery of OVA + CT while higher titres of CT-specific IgG2a than IgG1 were found in the same mice.²⁷ Feeding mice with CT + keyhole limpet haemocyanin (KLH) stimulated a strong KLH-specific secretory IgA response in mice which were high responders to CT with a much smaller effect in poor responders.³⁵ It was concluded that the oral adjuvant effect of CT depended on a strong immune response to CT itself. The present data support this hypothesis, with the strongest adjuvants (CT, ML-I) being potent immunogens. However, WGA and UEA-I displayed a moderate adjuvant effect for serum IgG responses to OVA and were not highly immunogenic. A recent study found that several dietary lectins, including PHA, could trigger human basophils to release IL-4 and IL-13. It was suggested that lectins may enter the circulation and induce IL-4, which is required to switch towards a Th2-type response.⁴⁵

Despite the induction of high serum IgG titres to OVA in mice immunized with CT or ML-I + OVA, serum IgA was barely detectable. Previous work found similar results with i.n. delivered zot protein or LT + OVA²⁴ and oral delivery of LT + TT.²⁶ The high titres of specific IgG induced may block the coating antigens in ELISA assays and prevent binding of the less abundant IgA. Removal of IgG, for example using Staphylococcal protein A, may reveal higher specific IgA titres in these samples. While both CT and ML-I effectively stimulated anti-OVA IgA in all mucosal secretions the levels were highest in nasotracheal washes. Antibody titres in vaginal washes were highly variable, which may reflect hormonal influences.⁴³ Since the total serum IgA titres in mice immunized with CT + OVA and ML-I + OVA, respectively, were 33-fold and 73-fold higher than in nasotracheal washes and 1340-fold and 1176-fold higher than in saliva, the OVA-specific IgA titres at these sites indicate the induction of local responses. There is some debate about the fate of antigens delivered into the nasal cavity of mice. Depending on the volume of fluid delivered and on whether mice are anaesthetized, antigen may enter the lungs. In the present experiments, mice were not anaesthetized but the possibility of a small amount of antigen reaching the lungs cannot be excluded.