Linear PADRE T Helper Epitope and Carbohydrate B Cell Epitope Conjugates Induce Specific High Titer IgG Antibody Responses¹

Microbial or tumor polysaccharide Ags often display thymus-independent type 2 Ag characteristics. The Ab response generated against these polysaccharide Ags is typified by low titer IgM, which fail to class switch to IgG, little anamnestic response, and ineffective Ab affinity maturation (1, 2, 3). These characteristics represent significant drawbacks for their use as effective vaccines (4, 5, 6).

The use of large protein carriers such as tetanus toxoid or diphtheria toxoid conjugated to polysaccharides tends to overcome these difficulties (7, 8, 9, 10, 11, 12). However, the use of large protein carriers creates difficulties in terms of reproducibility of the conjugation reactions and of chemical characterization of the resulting complex immunogens. These difficulties can complicate large-scale production and may endanger vaccine effectiveness and/or practical feasibility (13, 14). Synthetic epitopes should offer distinct advantages in terms of manufacturing and chemical characterization because of their small size and defined chemical nature. The immunogenicity of synthetic peptides has indeed been demonstrated for both cell-mediated responses and Ab responses (15, 16, 17, 18). However, in the case of simple peptide-polysaccharide conjugates, lack of consistent T cell help can be a problem due to the extensive polymorphism of HLA molecules with which the peptide must bind (19, 20).

To address this drawback, we engineered a simple T cell helper epitope that is chemically defined, easily manufactured, and able to generate effective helper T cell responses in the general human population. This epitope is a synthetic, nonnatural pan HLA DR-binding Epitope (PADRE)³ peptide that binds with high or intermediate affinity to 15 of 16 of the most common HLA-DR types tested to date (20, 21). Because of its binding promiscuity, PADRE should overcome the problems posed by the extreme polymorphism of HLA-DR molecules in the human population. Furthermore, PADRE peptide was specifically engineered to be immunogenic in humans (20). When PADRE was evaluated for immunogenicity using human T cells in a proliferation assay, it was found to be approximately 100-fold more potent on a molar basis than a control tetanus toxoid-derived universal epitope (22). This property represents another significant feature of PADRE, suggesting its potential utility as a carrier to induce T cell help in vaccine constructs designed for human use.

Since PADRE is capable of binding to murine I-A^b molecules, it was possible to evaluate the in vivo immunogenicity of simple linear constructs using PADRE as a T helper epitope in conjunction with various malarial B cell peptide epitopes. Monovalent PADRE-B cell epitope constructs induced high titer, long lasting malaria-Ag-specific Ab responses. The isotype profile of the responses was composed predominantly of IgG, and the Abs reacted with intact sporozoites and inhibited in vitro infection of liver cells (23). Most importantly, immunization with the monomeric PADRE constructs induced responses capable of protecting mice from in vivo challenge with live Plasmodium yoelii sporozoites (24).

In this study, we investigate the immunogenicity of simple linear constructs containing PADRE and prototype carbohydrate B cell epitopes. Specifically, two model carbohydrate Ags were examined, lacto-N-fucopentose II and a dodecasaccharide from the O-specific polysaccharide of Salmonella typhimurium. The Ab responses were evaluated with regard to magnitude, isotype profiles, biological relevance, and adjuvant effects.

The structural formulae of the lacto-N-fucopentose II and the Salmonella-derived dodecasaccharide are provided in Fig. 1,A, panels 1 and 2, respectively. Lacto-N-fucopentose II, lacto-N-fucopentose II-PAII (LNF-PA), and lacto-N-fucopentose II-HSA (LNF-HSA) were purchased from Accurate Chemicals (Westbury, NY). The lacto-N-fucopentose II-BSA (LNF-BSA) conjugate was purchased from Dextra Laboratories (Reading, U.K.). The dodecasaccharides from the O-polysaccharide of S. typhimurium SH4809 and LPS (s) from S. typhimurium SH4305, S. typhimurium 253 Ty, and Salmonella newport (C2) (see Table III) were all purchased from Accurate Chemicals. The structures of synthetic O-polysaccharide-specific tri- (Fig. 1,B, lines 3–5) and tetrasaccharide-BSA glycoconjugates (Fig. 1,B, lines 6–10) listed in Table IV were described previously (25, 26).

The PADRE peptide aKXVAAWTLKAAaZC (X = l-cyclohexylalanine, Z = aminocaproic acid) was prepared according to standard solid phase F-moc peptide synthesis procedures. Glycopeptide conjugates were generated by first preparing the corresponding glycosylamine derivative (Fig. 1 C). A solution of lacto-N-fucopentose (10 mg) or dodecasaccharide (6.0 mg) in saturated NH₄HCO₃ (1–5 ml) was stirred at room temperature for 4 days with solid NH₄HCO₃ added in 250-mg fractions during this time to ensure saturation. The mixture was repeatedly lyophilized to constant weight. The resulting solid was resuspended in tetrahydrofuran (0.5–2.0 ml) and cooled to 4°C. Saturated NaHCO₃ solution (0.5–2.0 ml) was then added, and the glycosylamine was acylated by 6-bromocaproyl chloride (8–20 μl). After 10 min, an additional portion of 6-bromocaproyl chloride (8–20 μl) was added, and stirring was continued at 0°C for 2 h, and then at room temperature for 2 h. Water (5 ml) was added to the reaction mixture, and the solution was acidified to pH 5 by addition of 1.0 M HCl. This solution was extracted with diethyl ether, and the separated aqueous phase was concentrated to 2 ml and chromatographed (C₁₈ reverse silica gel, 0% methanol in H₂O to 20% methanol in H₂O). After concentration and lyophilization, a white solid was obtained (∼50% yield). To perform conjugation with peptide, a mixture of PADRE peptide (3–4 mg) and either bromocaproyl-lacto-N-fucopentose (2.3 mg) or bromocaproyl-dodecasaccharide (1.61 μmol) and cesium carbonate (10–14 mg) in anhydrous dimethylformamide (1.0 ml degassed with argon before use) was stirred at room temperature under an argon atmosphere for 14–24 h. The PADRE-lacto-N-fucopentose (LNF-PAII) was concentrated in vacuo, and the resulting solid was dissolved in water (0.5 ml). Purification of the PADRE-lacto-N-fucopentose II conjugate was done by HPLC (using a Vydac (Hespiria, CA) C₁₈ column, and 25% CH₃CN, 75% H₂O to 35% CH₃CN, 65% H₂O in 55 min). The yield was calculated to be ∼45%. The PADRE-dodecasaccharide was concentrated in vacuo, and the resulting solid was also dissolved in water (0.5 ml). Purification of the PADRE-dodecasaccharide conjugate (DODECA-PA) was done by HPLC (Vydac C₁₈ column, 25% CH₃CN, 75% H₂O to 42% CH₃CN, 58% H₂O in 55 min). The yield was calculated to be ∼54%.

Groups of at least three C57BL/6 (H-2^b) mice 8–16 wk old were injected at the base of the tail with 100 μg/mouse of immunogen in 100 μl of CFA (Difco, Detroit, MI) and bled 4 wk later. The mice were subsequently boosted with the same amount of immunogen in 100 μl of IFA (Difco) and bled 2 wk later. In the case of the alum formulations, the adsorptions were performed with alum hydroxide, (Rehsorptar from Intergen, Purchase, NY), which contained 1.8% Al₂O₃ or 2.8% Al(OH)₃. Briefly, one volume of the gel was washed once in 5 vol of PBS (pH 7.4) and then suspended to the original volume using PBS. To adsorb PADRE-fucopentose or HSA-fucopentose, 150 μl of the constructs at 4 mg/ml in PBS were mixed at room temperature with 150 μl of the gel at 10 mg/ml. After incubating with agitation for 18 h, the suspensions were centrifuged, and the percentage of adsorption was calculated by measuring 280 nM OD before and after adsorption (adsorption efficiency was between 70 and 90%). The mice were immunized, bled, and boosted at 1-mo intervals with 100 μg of conjugate adsorbed on 250 μg of Alum. When using QS-21 (Aquila Biopharmaceuticals, Framingham, MA) (27) as the adjuvant, the mice were immunized, as described above, with 100 μg of the constructs mixed with 20 μg QS-21 in PBS. The mice were subsequently bled and boosted at 2-wk intervals. When the combination of alum and QS-21 was used for immunization and boosting, the conjugates were prepared as described above for alum and QS-21 and mixed. Subsequently, a total of 100 μg/mouse of immunogen were used, and mice were bled and boosted at 20-wk intervals.

To measure fucopentose-specific Ab titers, 96-well, flat-bottom plates (Immunol II, Dynatech, Boston, MA) were first coated with 1 μg/ml of either LNF-BSA (Dextra Laboratories) in 100 μl of 0.1 M sodium bicarbonate (pH 8.2) or similar concentrations of synthetic glycoconjugates. Subsequently, the plates were blocked with 0.1% BSA, 0.05% Tween 20 in PBS, followed by addition of serial dilutions of pooled sera from the immunized mice. The plates were incubated 1 h at 37°C, washed with PBS 0.1% Tween 20, and then incubated for 2 h at room temperature with HRP-rat anti-mouse IgG or HRP-rat anti-mouse IgM (Caltag, Burlingame, CA). Plates were washed and then incubated with avidin DH-HRP (Vectastain ABC kit, Vector Laboratories, Burlingame, CA). Ab titers were defined as the reciprocal of the serum dilution yielding 0.3 OD units (450 nM). Isotype determination was also performed by ELISA. In brief, serial dilutions of pooled sera from immunized mice were incubated with plates coated with the appropriate Ag, followed by biotin anti-mouse IgG, IgG1, IgG2a, IgG2b, IgG3, IGA, and IgM (Caltag). Subsequently, plates were incubated with avidin DH-HRP, and titers were determined.

To measure dodecasaccharide-specific Ab titers, various coating Ags were used. These included tri- and tetra-BSA conjugates (Fig. 1 B, lines 3–10) (25, 26) and LPS from different Salmonella species (Accurate). The LPS was detoxified by partial delipidation before using as a coating Ag. Briefly, the polysaccharide (20 mg) was dissolved in 0.15 N NaOH (5.0 ml); solution was heated at 100°C for 2 h. The mixture was cooled to room temperature and acidified to pH 4 with 1 N HCl. The mixture was then extracted with methylene chloride, and the aqueous layer was loaded on a C₁₈ reverse phase column. Elution was with MeOH/H₂O, 0–30%. Delipidated LPS was collected in the 10% MeOH fractions, which were collected and lyophilized. Coating the plates and determining the Ab titers are essentially as described above.

The ATCC 14028 S. typhimurium strain that was used expresses O-factors 4 (abequose linked to mannose), 5 (2-O-acetylated abequose linked to mannose), and 12 (glucosylated galactose). This strain was chosen because the SH4809 strain carbohydrate used to generate the PADRE-dodecasaccharide immunogen could originally express O-factors 5 and 12, the work-up and phage endorhamosidase degradation of the LPS likely deplete these residues (28). These cells were fixed with 1% formaldehyde (Sigma, St. Louis, MO) for 18 h at 37°C. These bacteria were subsequently washed twice in PBS and 2.0% FCS and were incubated with normal or immune sera diluted 1:20 in PBS 2% FCS followed by detection of bound Ab using a goat IgG-FITC conjugate specific for mouse IgG (Jackson ImmunoResearch, West Grove, PA). The mean channel fluorescence (MCF) was assessed by FACS-scan analysis (Becton Dickinson, San Jose, CA). Specificity of the interaction was verified by preincubation of the immune sera with 500 μM Salmonella O-polysaccharide-specific octasaccharide (Gal-[Abe]-Man-Rha)₂, for 1 h at 37°C before addition to the inactivated S. typhimurium. After 1 h incubation at 4°C, the Salmonella were washed twice, and the detection Ab (goat anti-mouse IgG FITC, Jackson ImmunoResearch) was added and incubated for 30 min at 4°C. The MFC was determined by FACScan analysis.

The anti-dodecasaccharide sera (diluted to 10⁻⁵) or the mAb SE154.1 ascites (diluted to 10⁻⁷) were coincubated with 10-fold serial dilutions (500–0.005 μM) of the free O-specific octasaccharide from Salmonella for 1 h at 37°C, followed by addition to plates (Immulon II) coated with 1 μg/well of O-specific tetrasaccharide-BSA conjugate and then incubated for 2 h at 37°C. The plates were washed twice, and the Ab titers determined.

In the first series of experiments, we examined whether the PADRE epitope could provide help for carbohydrate-specific Ab responses. The fucopentose was selected as a model carbohydrate because of its availability from commercial sources and structural similarity to human glycolipid epitopes. Accordingly, mice were immunized with 100 μg of either fucopentose, human serum albumin (HSA)-lactofucopentose, or PADRE-lactofucopentose conjugates in CFA, followed 1 mo later by a boost of the same immunogens in IFA. To determine the primary and secondary Ab responses, mice were bled 1 mo following immunization and 2 wk following the boost. The results of standard ELISA assays of these sera are shown in Fig. 2 A.

Although the unconjugated fucopentose did not induce measurable Ab responses (data not shown), the PADRE-fucopentose conjugate induced vigorous IgG primary and secondary responses with titers of 1.1 × 10⁵ and 6.8 × 10⁵, respectively. These titers were comparable with values of 8.2 × 10⁴ and 8.8 × 10⁵ measured for HSA conjugates in primary and secondary responses, respectively. These data demonstrate that simple linear PADRE-carbohydrate conjugates can induce IgG responses, comparable to those induced by more complex protein carrier conjugates.

To further address the potential of PADRE conjugates, we evaluated the immunogenicity of PADRE-carbohydrate constructs in alum, an adjuvant approved for human use, or QS-21, an adjuvant currently used in human clinical trials. When the potency of PADRE-fucopentose and HSA-fucopentose adsorbed to alum was examined in C57BL/6 mice, it was found (Fig. 2,B) that both compounds were immunogenic, although, as expected, the absolute titers were lower than those observed when using CFA. Ab titers in the 2.5 × 10³ range were observed for both the PADRE-fucopentose and HSA-fucopentose constructs after one immunization, which increased to 2.3 × 10⁴ and 5.8 × 10⁴ for PADRE-fucopentose and HSA-fucopentose respectively, after the second boost (Fig. 2 B).

The responses induced by these immunogens mixed with the QS-21 adjuvant were examined next. On the basis of previous experiments, an immunization regimen of three injections 2 wk apart was utilized. As expected, little response was noted after only the first immunization. After the second boost, PADRE-fucopentose induced an Ab titer of 2.7 × 10⁵ whereas the HSA-fucopentose Ab titer was approximately 2-fold lower (1.4 × 10⁵, Fig. 2,C). The combination of alum and QS-21 adjuvants induced the highest Ab titers. Following the second boost, the Ab titers had increased to 5.0 × 10⁵ and 3.1 × 10⁵ for PADRE-fucopentose and HSA-fucopentose respectively, (Fig. 2 D).

To further characterize the quality of Ab responses induced, we examined the isotype profile obtained by immunization with PADRE constructs. The predominant isotype observed for the PADRE-fucopentose was greatly dependent on the adjuvant used. CFA/IFA and alum induced primarily IgG1 responses. In contrast, QS-21 induced primarily responses in which the IgG2a, 2b isotype, predominated (Fig. 3). Interestingly, when the two adjuvants alum and QS-21 were combined, a balanced response was obtained, with both IgG1 and IgG2b isotypes each contributing approximately half of the total IgG.

To examine Ab responses specific for the carrier induced by PADRE or HSA conjugates, standard ELISA assays, which used either the carriers themselves or BSA-fucopentose as coating Ags, were performed. It was found that fucopentose conjugated to either PADRE or HSA induced strong and comparable anti-fucopentose Ab responses, in the range of 4.1 × 10⁴ to 8.3 × 10⁵, following one or two booster injections (Table I). However, the responses directed to the carrier moiety of each immunogen were strikingly different, depending on the adjuvant used, with the anti-PADRE response being over 40- to 1500-fold lower than the anti-HSA response. Therefore, it appears that PADRE may allow a greater portion of the immune response to be directed to the therapeutically relevant portion of the vaccine.

To extend our analysis to a carbohydrate moiety derived from a potential human disease target, linear glycopeptide constructs composed of PADRE and the O Ag (dodecasaccharide) from S. typhimurium were used. Mice were primed with either detoxified LPS or the PADRE-dodecasaccharide construct emulsified in CFA and boosted 1 mo later with the same amount of immunogen in IFA. As expected, by itself, detoxified-LPS was incapable of inducing significant Ab responses (Table II). In contrast, the PADRE-dodecasaccharide construct, induced primary titers of 7.1 × 10³ and secondary titers of 8.7 × 10⁴. Although the primary response consisted of ∼30% IgM and 65% IgG, the secondary response was composed predominantly of IgG. These results demonstrate the utility of PADRE-carbohydrate constructs utilizing a B cell epitope derived from a microorganism of potential pathological significance.

To demonstrate the biological relevance of the Ab response induced by PADRE-dodecasaccharide, we assayed the antisera derived from mice immunized with this construct for its capacity to bind intact S. typhimurium.

The anti-dodecasaccharide serum was strongly reactive, giving a MFC of 400 (Fig. 4). The specificity of this interaction was demonstrated by the fact that the reactivity was almost completely inhibited by the addition of 500 μM octasaccharide (Gal-[Abe]-Man-Rha)₂. As expected, a normal mouse serum control did not appreciably react with intact Salmonella.

Salmonella species produce a series of O Ags that are structurally closely related. The fine specificity of the anti-dodecasaccharide Ab responses was therefore examined by coating ELISA plates with LPS from various Salmonella species possessing different O Ags, (Table III). Ab titers were normalized using S. typhimurium (SH 4809) LPS, which is the origin of the dodecasaccharide used in the PADRE-dodecasaccharide construct.

A drastic reduction of binding, down to 1.3% of the control, was demonstrated with LPS from S. typhi 253 Ty suggesting a detrimental effect when tyvelose replaces the abequose (dideoxyhexose, 3,6-dideoxy-d-galactose) sugar. Similarly, when LPS derived from another Salmonella species (S. newport: C2) was tested, almost no Ab binding was observed. In this case, the abequose structure is invariant, but the backbone from mannose-rhamnose-galactose is changed to rhamnose-mannose-mannose-galactose. The O Ag of the LPS isolated from S. typhimurium (SH 4305) is modified by addition of glucose linked to the main chain galactose residue. Binding was still appreciable although somewhat reduced to 54% of the control, possibly due to steric hindrance from the glucose. Taken together, these results suggest that the configuration of the abequose is critical to binding.

To further address fine specificity of anti-dodecasaccharide Ab reactivity, the capacity of the anti-sera to bind BSA conjugated tri- and tetrasaccharide conjugates of varied structure in ELISA assays was examined (Table IV). Ab titers were normalized to the BSA conjugate containing the tetrasaccharide, galactose-[abequose]-mannose-rhamnose. It was found that, when the abequose is replaced by paratose or tyvelose, binding is reduced, in the 3.4 to 11.2% range. The importance of the abequose side chain for Ab reactivity was further evaluated with additional conjugates. Altering the hydroxyl groups at positions 2 or 4 of abequose, to generate 2-deoxy-abequose and 4 deoxy-abequose reduced the binding to 8.4% and 65%, respectively. These results suggest that position 2 of abequose is significantly involved in binding, while position 4 appears to be less critical. Reducing the backbone structure from 4 to 3 carbohydrate units decreased the binding to 36%. As expected, when the abequose side chains on the trisaccharide conjugates were replaced by either paratose or tyvelose, all Ab reactivity was lost.

We next estimated the affinity of the anti-dodecasaccharide sera by measuring its sensitivity to inhibition by soluble ligands, as compared with the control mAb Se155-4 (26). Competition assays were performed using the soluble O-polysaccharide-specific octasaccharide as the competitor and an O-polysaccharide-specific tetrasaccharide-BSA glycoconjugate as the immobilized Ag. The IC₅₀ was found to be 20 μM in the case of the PADRE-dodecasaccharide polyclonal antisera, and 3 μM in the case of the Se155-4 Ab (Fig. 5). Thus, the affinity of the PADRE-dodecasaccharide-specific antisera is in the affinity range measured for Abs specific for the same epitope (26, 29).

Herein, we report the use of PADRE-carbohydrate conjugates to induce high titer, IgG Abs directed against oligosaccharide epitopes. In the fucopentose system, the responses obtained with PADRE conjugates were similar in magnitude to those observed with the more complex HSA-carrier conjugate. PADRE conjugates were also characterized in terms of potential use when formulated in different adjuvants, and in terms of the Ig class and subclass composition of the resulting antisera. The potency of PADRE-carbohydrate conjugates was also generalized to a different model epitope, a S. typhimurium-derived dodecasaccharide. The dodecasaccharide used in this study consists of three repeats of a tetrasaccharide unit (Fig. 1 A). A number of studies have examined the Ab responses directed against Salmonella-derived O-specific carbohydrate Ags, making this system a well-characterized model for carbohydrate-specific Ab responses (29, 30, 31, 32). In the case of Salmonella, the resulting antisera was characterized in terms of fine specificity, affinity and binding to intact S. typhimurium cells. Together the experiments presented herein demonstrate the potential of PADRE as a synthetic carrier for carbohydrate Ags.

The current study demonstrates that the responses obtained with monomeric PADRE-carbohydrate conjugates were equivalent to those obtained with a more complex HSA protein carrier. These data contrast with the view that simple linear monovalent constructs are ineffective in inducing responses against carbohydrate Ags, and that polyvalent constructs incorporating complex protein carriers are required for optimal immunogenicity. Moreover, it is possible that because of their optimal T cell activity, PADRE constructs may overcome the requirement for Ig cross-linking on the surface of B cells (23). These are important findings, particularly in view of the problems sometimes associated with large protein carriers available for use in vaccine conjugates, such as tetanus toxoid or diphtheria toxoid. When targeting weakly antigenic molecules such as carbohydrates, “booster” injections are important for conversion of the initial, transient IgM isotype-restricted Ab response to a strong, durable IgG isotype response. However, Ab responses directed against the vaccine carrier have been shown, in some cases, to negatively affect the booster response to the vaccine Ag (13, 14, 33). This phenomenon is known as the “carrier effect” and is likely related to rapid clearance of the vaccine mediated by anticarrier Abs. The use of a small synthetic carrier capable of inducing vigorous helper T cells, but potentially less readily recognized by Abs might be, in this respect, of significant interest.

Using the PADRE-fucopentose construct, we found that the carrier-specific Ab responses were 2- to 20-fold less than Ab responses to the carbohydrate. In contrast, HSA-fucopentose induced a 2- to 110-fold higher Ab response to the carrier. These data suggest that large protein carriers are likely to induce higher titer Ab responses to the carrier and this may interfere with subsequent vaccines using this same carrier. Further studies are required in the context of carrier-suppression to fully explore the potential advantages of PADRE relative to large protein carriers.

Thus, synthetic peptides, because of their small size and defined chemical nature offer an attractive alternative to large carrier proteins. Indeed, the use of synthetic peptides as immunogens has been described in several studies (34, 35, 36, 37, 38, 39, 40, 41, 42). However, simple linear B and T cell epitope constructs may suffer from lack of consistent T cell help arising from extensive polymorphism of HLA DR molecules, and decreased potency because of their simple monovalent nature. PADRE has been specifically engineered to address this limitation for human T cells, and is approximately 100-fold more potent in vitro than the previously described universal tetanus toxoid epitope (22).

Other important factors that influence effectiveness of a vaccine include potency in appropriate adjuvants, and induction of protective Ab isotypes. As demonstrated in this study, PADRE-fucopentose was highly immunogenic in both alum and QS-21 in terms of induction of high titer carbohydrate-specific IgG responses. In terms of isotype profiles resulting from immunization of PADRE conjugates, it was found that responses in which Th1- or Th2-associated Ig classes and subclasses are predominant could be obtained by the use of different adjuvants. These observations are consistent with previous reports that underline the influence of carrier type and particular adjuvant used in determining the profile of Ab isotypes generated in response to immunization (43, 44, 45).

Interestingly, when PADRE constructs were formulated in combination with Alum and QS-21, a balanced mix of IgG1 (Th2) and IgG2b (Th1) isotypes was observed. The capacity of PADRE constructs to induce a balanced response in terms of subclass composition is of interest. This capacity may allow simultaneous induction of Abs effective both in terms of complement fixation and opsonization (46, 47, 48, 49, 50). Importantly, no significant IgE responses were noted in response to PADRE conjugates, a fact consistent with the good safety profile of PADRE thus far in laboratory animals and humans (51).

Our observations were expanded to an O-polysaccharide-specific dodecasaccharide from S. typhimurium. Salmonella species produce a series of O Ags that are structurally similar. The difference may be as subtle as the stereochemistry of the immunodominant 3,6-dideoxyhexose or may involve posttranslational acetylation or glycosylation of the completed Ag. The net result is a modulation of Ab response so that sera against Salmonella species exhibit various degrees of cross-reactivity with the LPS from Salmonella of other serogroups. The Kaufmann-White serological classification scheme, which identifies O factors, has been correlated with structural features of the O Ags (52). Relevant aspects of such structural correlates are presented in Table III. The specificity and potential biological relevance of Ab responses were therefore analyzed. Our data suggest that the epitope recognized by the PADRE-carbohydrate antisera is composed of four carbohydrate units. Reactivity toward LPS from various Salmonella species depended on a mannose-rhamnose-galactose backbone with an absolute requirement for abequose as a branching residue attached to mannose as exhibited by two S. typhimurium species. As expected, the involvement of the backbone residues mannose-rhamnose-galactose are important for Ab binding. It is further evident that the galactose residue of the main polysaccharide chain plays a significant role in contacting the Ab site since its substitution by a glucose residue reduces Ab binding. From the biological standpoint, the results also underline the species-specificity of the antisera, in that the antisera bound with similar titers to different strains of S. typhimurium, whereas marginal or no binding was detected to different species of the same genus (S. typhi and S. newport). The immunodominance of abequose was confirmed by examining binding of the anti-dodecasaccharide Abs to tri- and tetrasaccharide-BSA conjugates. Similar findings were reported by Norberg and colleagues (29), which analyzed the specificity of antiserum obtained from rabbits immunized with heat-killed Salmonella essen (O Ags 4 and 12). These similar findings of specificity in both mice and rabbits emphasize the potent immunogenicity of this O Ag epitope.

Biological relevance of the anti-dodecasaccharide Ab response was suggested by demonstrating binding to intact S. typhimurium cells, thus suggesting that the antisera obtained as a result of PADRE-dodecasaccharide immunization should be capable of neutralizing activity in vivo. The affinity of the anti-dodecasaccharide was also measured and found to be in the range of affinities observed for other Abs directed against the same epitope induced by other approaches (25).

In conclusion, simple synthetic constructs encompassing the PADRE epitope are highly immunogenic in I-A^b mice, and effective for inducing Ab responses to oligosaccharide Ags. Murine strains other than C57BL/6 (b haplotype) mice were not evaluated due to lack of cross-reactivity of PADRE on class II molecules other than I-A^b (20). Our studies provide validation of PADRE as a helper epitope for inducing Ab responses in two carbohydrate Ag systems and with adjuvants compatible with human use. A large number of potential Ags both in the infectious disease and cancer arena have been described (53, 54) that could be targeted by the use of these types of immunogens.

We thank George Kallingal for performing mouse immunizations, bleedings and ELISA. We also thank Ms. Mara Capella for preparation of the manuscript. In addition, we thank Dr. Timothy J. Lithgow and Dr. Mark Newman for careful review of the manuscript.