Antigenicity and Immunogenicity of Modified Synthetic Peptides Containing D-Amino Acid Residues

The effect of introducing D-amino acid residues in an hexapeptide was examined both at the antigenic and immunogenic levels. A series of D-analogues of the model peptide of sequence IRGERA corresponding to the COOH-terminal residues 130-135 of histone H3 were produced. Four analogues contained a single change of an L-residue by the corresponding enantiomer, one peptide contained two D-residues and another one contained only D-residues @-enantiomer). A peptide analogue was also synthesized in which the 2 Arg residues were replaced by Lys residues. The parent peptide and peptide analogues were injected into mice after covalent coupling to small unilamellar liposomes containing monophosphoryl lipid A as adjuvant. The substitution of ~-Arg"l to Lys or D-Arg was found to change neither the antigenic nor immunogenic properties of the resulting peptides. In contrast, the substitution of Glul", Arg"', and Ala"' by the respective enantiomers drastically altered the antigenicity of the modified peptides. Each of the six analogues induced an immune response with an unusually high level of IgG3 antibodies. The D-enantiomer produced IgG3 antibodies which reacted with the homologous peptide as well as with the all L-peptide and the parent protein H3 in solution but not with analogues containing one or two D-residues only. IgG3 antibodies produced against the all L-peptide reacted with the free all D-peptide but not with the other analogues containing D-residues in position 133, 134, and 135.

The effect of introducing D-amino acid residues in an hexapeptide was examined both at the antigenic and immunogenic levels. A series of D-analogues of the model peptide of sequence IRGERA corresponding to the COOH-terminal residues 130-135 of histone H3 were produced. Four analogues contained a single change of an L-residue by the corresponding enantiomer, one peptide contained two D-residues and another one contained only D-residues @-enantiomer). A peptide analogue was also synthesized in which the 2 Arg residues were replaced by Lys residues. The parent peptide and peptide analogues were injected into mice after covalent coupling to small unilamellar liposomes containing monophosphoryl lipid A as adjuvant. The substitution of ~-Arg"l to Lys or D-Arg was found to change neither the antigenic nor immunogenic properties of the resulting peptides. In contrast, the substitution of Glul", Arg"', and Ala"' by the respective enantiomers drastically altered the antigenicity of the modified peptides. Each of the six analogues induced an immune response with an unusually high level of IgG3 antibodies. The D-enantiomer produced IgG3 antibodies which reacted with the homologous peptide as well as with the all L-peptide and the parent protein H3 in solution but not with analogues containing one or two D-residues only. IgG3 antibodies produced against the all L-peptide reacted with the free all D-peptide but not with the other analogues containing D-residues in position 133, 134, and 135.
Synthetic peptides have received considerable attention as potential vaccines because they are chemically defined, have very long shelf lives, and are likely to be safer and cheaper than conventional vaccines derived from infectious material. The design of a synthetic vaccine requires a thorough knowledge of the mechanisms of protective immunity involving B and T epitopes and of the parameters that affect peptide immunogenicity such as the mode of presentation to the immune system, the need for a carrier and adjuvant, and the influence of route and method of delivery. *This work was supported by Centre National de la Recherche Scientifique GCCB Project 28D4. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "uduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Over the last decade, several candidate synthetic vaccines have been tested up to the level of in vivo protection trials. There has been an increasing tendency to design peptides endowed with a particular conformation and several peptides have been assessed as potential vaccines (Brown, 1990;Milich, 1990;Arnon and Van Regenmortel, 1992). Of particular interest is the possibility of improving the antibody response, in terms of specificity, isotype, and duration, by altering the presentation or composition of a peptide. To this end, peptide analogues containing D-amino acid residues or modified peptide bonds which exhibit a much higher resistance to proteases may be rendered, by virtue of their increased stability, more immunogenic than the native peptide. Such modifications have found many applications in the field of peptide-based drugs where increased stability has resulted in significantly increased biological activities (Fauchgre and Thurieau, 1992). In contrast to studies dealing with pharmacological activities of peptides containing D-residues, very little has been published regarding the antigenic and immunogenic properties of such analogues (Geysen et al., 1986(Geysen et al., , 1987. In the present study, we have evaluated the influence of L-to D-amino acid substitutions on the antigenic and immunogenic properties of a model hexapeptide. A series of peptide D-analogues (several partially modified analogues and the mirror D-enantiomer; Wade et al., 1990) were tested in the hope of defining if changes in peptide structure lead to a longer duration of the immune response without impairing the capacity of modified peptides to induce antibodies cross-reacting with the cognate protein. In order to minimize any influence of the carrier, the parent peptide and analogues were injected into mice after coupling to small unilamellar liposomes containing monophosphoryl lipid A as adjuvant. The short peptide IRGERA which alone is not immunogenic (Frisch et al., 1991) was selected for this study. A short peptide was used since it seemed likely that with longer peptides, the presence of Damino acids may also affect processing and presentation to MHC molecules (Milton et al., 1992;Jung, 1992) in addition to altering peptide stability.

MATERIALS AND METHODS
Peptides-Assembly of the protected peptide chains was carried out on a multichannel peptide synthesizer (Neimark and Briand, 1993) using the procedure described previously (Friede et al., 1992). Briefly, all amino acids were protected at the a-amino position with the Fmoc group. The following side chain protecting groups were used tert-butyl ester for Glu and 2,2,5,7,8-pentamethyl-chroman-6sulfonyl for Arg. Twenty-five pmol Fmoc-Ala or Frnoc-D-Ala 4hydroxymethyl phenylmethyl polystyrene resin (Novabiochem, Switzerland) were placed in each reaction vessel. Fmoc amino acid deriv-26279 D-Analogues of Antigenic Peptides atives were coupled in 500 pl of dimethylformamide by using a BOP/ HOBt coupling procedure (Hudson, 1988). Double couplings were performed systematically. At the end of the synthesis and after the last deprotection step, the peptide resin was washed with dichloromethane and dried under nitrogen. The trifluoroacetic acid cleavage was performed on the machine in the same reaction vessels by using King's reagent (King et al., 1990). Four ml of this reagent were delivered into each reaction vessel. The total cleavage time was set to 2 h 30 min. At the end each peptide was collected in a polypropylene tube containing 25 ml of cold ether, centrifuged, washed again with ether, solubilized in 5 ml of H20/CH3CN (v/v), and lyophilized. Homogeneity of the crude peptides was assessed by analytical runs on a Novapak Cl8 column, 5 pm (3.9 X 150 mm) using a triethylammonium phosphate buffer system. Linear gradients detected at 210 nm were from 1-21% CH3CN in 12 mn, and the flow rate was 1.4 ml/min. Runs were performed with a Waters apparatus (Waters Corporation, Milford, MA). CD-Circular dichroism measurements were performed on a Jobin Yvon Dicrograph I11 under constant nitrogen flush at 25 'C. The spectra of peptides were recorded between 195 and 260 nm in trifluoroethanol (TFE)' using a 1-mm path length quartz cell. The results were expressed as mean residue ellipticity values (e) having the units deg . cm2. dmol".
Peptide Carrier Conjugation-Three methods were used to conjugate peptides to ovalbumin or bovine serum albumin (BSA). Coupling was performed using either glutaraldehyde, N-succinimidyl3-[2-pyridyl dithiolpropionate (SPDP), or photochemical activation involving the benzophenone species benzoylbenzoyl glycine which was added to the NH2-terminal residue of the peptide IRGERA at the end of the synthesis (Muller, 1988). The yield of coupling was obtained by determining the amino acid composition of the final conjugate (Briand et d., 1985) or by determining spectrophotometrically the release of 2-thiopyridone from SPDP derivatized BSA on interaction with cysteine-containing peptides (Muller, 1988).
Preparation of Liposome-associated Peptides-Peptides were covalently coupled to preformed small unilamellar vesicles (SUV) containing 4-(p-maleimidophenyl)butyrylphosphatidylethanolamine (MBP-PE) (Friede et al., 1993). For immunization, monophosphoryl lipid A (MPLA) was incorporated in SUV as adjuvant. Briefly, preparation of liposome-associatedpeptides was performed as follows: SUV were formed by drying a solution of lipids (egg-yolk phosphati-dyLcholine,~ cholesterol, and MBP-PE, (855015 mol %) containing MPLA (Ribi, Hamilton, MO) at 2 mg/mmol lipid, resuspension (12 pmol lipid/ml) in 100 mM NaCI, 50 mM HEPES, pH 6.5, and sonication with a probe sonicator for 1 h. After centrifugation at 10, 000 X g for 10 min to remove large vesicles and titanium particles, peptide was added at an equimolar ratio to the MBP-PE and incubated at room temperature for 1 h. Unbound peptide was removed by dialysis.
Antisera-The following rabbit antisera were used: antiserum Tri directed to complete histone H3, antiserum Me1 directed to peptide IRGERA conjugated to ovalbumin by means of glutaraldehyde, and antiserum Giz directed to peptide (benzoylbenzoyl glycine) IRGERA conjugated to ovalbumin by photochemical coupling. Their production and characterization have been described . Mouse antiserum was obtained from BALB/c mice immunized with liposome-associated peptides. Mice were injected intraperitonealy with the various SUV preparations containing 1 pmol of lipid, 2 pg of MPLA, and 60 pg peptide/injection/animal. Mice received three injections at intervals of 3 weeks. Blood was withdrawn from the retroorbital venous plexus 5 days after each injection and then regularly over 6 months after the last injection. Control serum was obtained from each mouse before the first injection.
ELISA-The ELISA procedure used to measure the binding of antibodies from rabbit or mouse sera was as follows: microtiter plates (Falcon) were coated overnight at 37 "C with 100 ng/ml histone H3 or with 2 p~ of the various peptides and peptide conjugates in solution in 0.05 M carbonate buffer, pH 9.6. After three washings of microtiter plates with phosphate-buffered saline containing 0.05% Tween (PBS-T), antiserum diluted in PBS-T containing 10 mg/ml BSA was added for 1 h at 37 "C. After repeated washings positive reactions were detected with AffiniPure goat anti-rabbit IgG (H+L) conjugated to horseradish peroxidase (Jackson Laboratories, Bar Harbor, ME; ' The abbreviations used are: TFE, trifluoroethanol; BSA, bovine serum albumin; SPDP, N-succinimidyl 3-[2-pyridyl dithiolpropionate; SUV, small unilamellar vesicles; MBP-PE, 4(p-maleimido-pheny1)butyrylphosphatidylethanolamine; ELISA, enzyme-linked immunosorbent assay; MPLA, monophosphoryl lipid A. working dilution 1:40,000) or with rabbit anti-mouse IgG (H+L) conjugated to peroxidase (Nordic, Tilburg, The Netherlands; working dilution 1:5,000). After 30 min of incubation and washing with PBS-T, the final reaction was visualized by incubation with 3,3',5,5'tetramethylbenzidine in the presence of HzOz . The resulting absorbance was measured at 450 nm.
To determine antibody isotypes, the unlabeled rabbit anti-mouse specific for IgG1, IgGZa, IgGZb, IgG3, IgA, and IgM from Nordic were used (working dilutions 1:5,000). After 30 min of incubation at 37 'C and washings, goat anti-rabbit IgG conjugate diluted 1:30,000 was added for 30 min at 37 "C. The other steps of the assay were as described above. A number of experiments were also performed using goat anti-mouse IgG3 conjugated to horseradish peroxidase (Nordic, working dilution 1:7,000).
tides or of H3 used as inhibitors were incubated for 1 h at 37 'C and For competition experiments, various concentrations of the pepthen overnight at 4 "C with antisera diluted in PBS-T. The mixture was then added to protein or peptide-coated wells, and the test was performed as described above. In order to provide a semiquantitative comparison of the relative reactivities of antibodies for the peptides and H3 in inhibition studies, we made the following assumptions. In the case of BSA-conjugated peptides, we assumed that the maximal binding of BSA coated into the wells of PVC microtiter plates was 400 ng (Sorensen and Brodbeck, 1986;Butler et al., 1992), corresponding to 0.062 nmol of peptide. In the case of individual histones, we showed previously using radioactive molecules that in the coating conditions used, nearly 90% of the antigen introduced in wells becomes attached to the plastic (Muller and Van Regenmortel, 1989). This corresponds to about 20 ng H3/well. In the case of both BSA conjugates and histone H3, we assumed that 100% of the bound antigen was available for binding to the antibodies present in the liquid phase. (Table I). Peptides 1-4 correspond to the natural sequence of the COOH-terminal end of histone H3 (IRGERA). Th' 1s region contains a major epitope of H3 which has been extensively studied in this laboratory (Muller et al., 1982;Frisch et al., 1991;Briand et al., 1992;Friede et al., 1993 ). In order to enhance the accessibility of the residues in the IRGERA and GERA peptides bound to a carrier, two additional residues of Gly were added to their NH2-terminal end (peptides 1 and 4-11). A Cys residue was added at the NHz terminus of these peptides to allow selective conjugation of the peptides to BSA and liposomes via reaction with maleimide groups present on the activated carriers. In peptide 5, Arg131 and Arg" were replaced by Lys residues and in peptides 6-11, D-amino acid residues replaced certain L-amino acid residues as indicated in Table I. The replacement by a D-optical isomer is not applicable to the symmetrical Gly13'. The degree of purity of peptides, as assessed by high performance liquid chromatography, was at least 95%. In CD measurements a longer peptide (residues 118-135) was introduced as control.

Synthetic Peptides-Eleven peptides were used in this study
Structure of Peptide IRGERA and Peptide Analogues-In absence of structural data, attempts to compare the structures  adopted by the different peptide analogues by molecular modeling using calculations of energy minimization were unsuccessful. No preferential conformation was observed and thus modeled structures were of very low reliability (data not shown)? The crystal structure of the core histone octamer has been determined at 3.1-A resolution (Arents et al., 1991). In the three-dimensional structure the residues in region 120-132 of H3 form an a-helix, while the 3 residues 133-135 are outside the helix? CD spectra show that in 100% TFE, the percent helix content in peptide 118-135 may be estimated as 30% (Fig. 1). Helical conformation is no longer observed in peptide 4 and a negative ellipticity at 198 nm is found indicating an unordered form. Although CD spectra of peptides in TFE need to be interpreted with great circumspection, it is interesting to observe that spectra obtained with peptides 4 and 11 are roughly symmetrical.

I K
Recognition in ELISA of IRGERA Analogues by Rabbit and Mouse Antibodies Induced against the Parent Peptide IR-GERA and H3"Three antisera raised against the peptide IRGERA (rabbits Me1 and Giz) and histone H3 (rabbit Tri) were first tested for their ability to react with the various peptides shown in Table I. These tests were performed in a competitive ELISA using H3 as antigen for coating plates and the different peptides as inhibitors. As shown in Fig. 2 and Table 11, the binding of anti-H3 antibodies to H3 was strongly inhibited by peptides 4, 5, and 6, less well by the shorter peptide 1 and not at all by peptides 7-11. Very little difference was seen when rabbit antibodies against peptide IRGERA instead of rabbit antibodies against the cognate protein H3 were used except for peptide 8 which inhibited slightly the interaction of anti-IRGERA antibodies with H3.
Antisera against peptide 4 were also raised in BALB/c mice by a series of intraperitoneal injections of the peptide coupled at the surface of SUV containing MPLA as adjuvant (Friede et al., 1993). When the antisera were tested in the competitive assay using H3 as antigen and the different peptides as inhibitors, very similar results to those described above with rabbit antisera were found (Table 11). Antibodies raised against peptide 4 reacted also strongly with H3 in solution (see below).
The cross-reaction of antibodies against peptide 4 with the various peptide analogues and with H3 was also studied in a direct ELISA format in which H3 and the different peptide for coating plates. The conditions of coupling were controlled in such a way that all BSA-peptide conjugates had the same peptide to carrier molar ratio of 10. As shown in Table 111, mouse antibodies against peptide 4 reacted well with homologous peptide 4 (see also Fig. 3A) as well as with the peptide analogues 5 and 6. They also reacted strongly with the parent protein H3 (Fig. 31). In contrast they did not bind to peptides From the results obtained with rabbit and mouse antisera, it can be concluded that the replacement of Arg13' and Arg'34 by Lys residues (peptide 5) had not effect on antibody recognition. Likewise, the replacement in peptide 6 of Argx3' by D-Arg had no detectable effect on the binding of anti-H3 and anti-IRGERA antibodies. In contrast, the replacement in peptide 4 of L-residues 133,134, and 135 by the respective Disomers (peptides 7-9) dramatically altered the recognition of the peptides by anti-H3 and anti-IRGERA antibodies. Likewise, peptides 10 and 11 containing 2 D-Arg residues in position 131 and 134 and 5 D-residues in position 130-135, respectively, were recognized neither by H3 nor by IRGERA antibodies.
Mouse Antibodies to IRGERA Analogues-Groups of four BALB/c mice were injected with the various peptide analogues conjugated to liposomes. An IgM response could be generally demonstrated by ELISA in bleeding 1 which slowly decreased from bleeding 3 onward and was no longer detectable after bleeding 5 (data not shown). IgG responses to the different peptides conjugated to BSA by means of SPDP are shown in Fig. 3 (open squares). IgG antibodies were generally detected from bleeding 2 onward and, depending on the peptide used as immunogen, their activity decreased from bleeding 5 (47 days after the last injection) to 7 (89 days after the last injection). A fairly low response was found in mice immunized with peptides 7 and 11 (Fig. 3, D and H). The antibody response measured by ELISA was very similar in the individual animals of each group.
Ability of Mouse Antibodies Induced against IRGERA Analogues to Cross-react with Histone H3 and Various Peptide Analogues-The capacity of mouse antibodies induced with IRGERA analogues coupled to liposomes to recognize histone H3, parent peptide 4, and other IRGERA analogues was studied in different ELISA formats. The reactivity of mouse antisera was first studied in a direct ELISA in which the different antigens were adsorbed on microtiter plates (Table  111, Fig. 3, open squares). The reactivity of antibodies against peptide analogues containing Lys residues at positions 131 and 134 (peptide 5) or a D-Arg residue at position 131 (peptide 6) was very similar to that of antibodies against peptide IRGERA (peptide 4). These antibodies reacted with peptides 1, 4, 5, and 6 and with the cognate protein H3 but not with peptides 7-11. In contrast, antibodies induced against peptides 7-11 recognized only the respective homologous peptides and did not react with heterologous peptides nor with H3.
The reactivity of the mouse antibodies was also tested in a competitive binding assay with free peptides in solution to determine if the free peptides were recognized as well as the peptide conjugates. The results presented in Table IV show the molar excesses of the various peptides required to inhibit 50% of the binding between anti-peptide antibodies and the respective homologous peptides. In agreement with the results obtained with immobilized conjugated peptides, peptides 4,5, and 6 strongly inhibited the binding of antibodies against these peptides, but not that of antibodies against peptides 7-11. As also observed in direct ELISA, only the homologous peptides inhibited the reaction between antibodies to peptides 7-11 and their respective peptides. H3 in solution inhibited 7-11.

FIG. 2. reaction
Inhibition of the ELISA between H3 (100 ndml) and three rabbit antisera, respectively, directed to H3 (rabbit Tri, A ) , peptide 2 conjugated to ovalbumin by means of glutaraldehyde (rabbit Mel, B ) , and peptide 3 conjugated to ovalbumin by photochemical coupling (rabbit Giz, C) by increasing concentrations of peptides 1 and 4-11. A control peptide corresponding to the sequence 149-158 of tobacco mosaic virus protein (TMV-p) was used as control. Rabbit antisera were diluted 1:3,000 and anti-rabbit IgG conjugate 1:40,000. Absorbance values were measured at 450 nm.

Recognition in competitive ELISA of IRGERA and IRGERA analogues by anti-peptide and anti-H3 antibodies
Microtiter plates were coated with 100 ng/ml H 3 and allowed to react with rabbit and mouse antisera diluted 1:3,000 and 1:500, respectively, and preincubated with the various peptides used as inhibitors. The reaction was revealed with anti-rabbit IgG (H+L) conjugate and anti-mouse IgG (H+L) conjugate diluted 1:40,000 and 1:5,000, respectively. Molar excesses were calculated as described under "Materials and Methods." the binding of antibodies induced against peptides 1,4,5, and 6, but not that of antibodies against peptides 7-11, to their respective homologous peptides coated as BSA-conjugates (Table V). Overall, when the results obtained in the direct and competitive ELISA tests were compared, the only difference which was observed concerned peptide 1. In solution this short peptide was only capable of inhibiting the binding between anti-peptide 1 antibodies and peptide 1 and was unable to compete with IRGERA analogues (Table IV). In a direct binding assay, this peptide coated as a BSA-conjugate was recognized by antibodies induced against peptides 4,5, and 6 while anti-peptide 1 antibodies recognized conjugated peptides 4, 5, and 6 ( Table 111).
The binding of H3 by antibodies induced against peptides

Reactivity in ELISA of mouse antibodies induced against IRGERA
and IRGERA analogues Microtiter plates were coated with 2 FM peptide conjugated to BSA (carrier to peptide molar ratio 1:lO) or with 100 ng/ml H 3 and allowed to react with mouse antisera diluted 1500 (bleeding of immunized mice after three injections except for mice immunized with peptides 1, 7, and 11 for which the respective bleedings 2 were used). Anti mouse IgG-peroxidase conjugate was diluted 1:5,000. 4,5, and 6 was strongly inhibited by the homologous peptides but not by peptide 1 and peptides 7-11 (data not shown). The binding of H3 by antibodies to peptide 1 was only inhibited by peptide 1.

Analysis of the Isotypes of Antibodies Raised to IRGERA and IRGERA Analogues Associated to SUV Containing MPLA-
The antibodies raised in BALB/c mice to the various peptides covalently attached to SUV were essentially of the IgG type. Antibodies of IgM isotype which were initially present gradually disappeared during the course of immunization, and no IgA antibodies were detected. Antibodies of all IgG subclasses were present in the various antipeptide sera except in antisera to peptide 11 which contained mainly IgG3 antibodies (Fig.  3H). The IgG3 subclass (Fig. 3, closed squares) was also prevalent in antisera to peptides 7 and 9 (Fig. 3, D and F).

TABLE IV
Recognition in competitive ELZSA of ZRGERA and ZRGERA analogues by mouse antibodies induced against homologous and analogue peptides Microtiter plates were coated with 2 p~ peptide conjugated to BSA by means of SPDP (carrier peptide molar ratio 1:lO) and allowed to react with mouse antisera raised against the homologous peptide (serum dilution 1:500) and preincubated with the various peptides used as inhibitors. Anti mouse IgG-peroxidase conjugate was diluted 1:5,000. By using in ELISA a conjugate specifically directed toward mouse IgG3, it was possible to detect a particular subpopulation of antibodies of the IgG3 isotype which was produced generally later during the course of immunization. This reactivity was not detected when anti-globulin conjugate was a mouse IgG (H+L) second antibody, presumably because the conjugate contained too few anti-IgG3 antibodies.

TABLE V Recognition in competitive ELZSA of H3 by mouse antibodies induced against peptide 4 and analogue peptides
Microtiter plates were coated with 2 pM peptide conjugated to BSA by means of SPDP (peptide-carrier ratio 1O:l) and allowed to react with mouse antisera raised against the homologous peptide (serum dilution 1:500) preincubated with various concentrations (0.25-125 pg/ml) of H3 used as inhibitor. The final reaction was revealed with rabbit anti-mouse IgG (H+L) conjugate (A) or with goat anti-mouse IgG3 (B) conjugated to horseradish peroxidase (working dilutions 1:5,000 and 1:7,000, respectively). Molar excesses were calculated as described under "Materials and Methods." Molar excess of H3 required to inhibit 50% of the Anti-mouse Ig reaction between antibodies against peptides 1 to 11 reveal reaction conjugate used to and the respective homologous peptides When the specificity of the IgG3 antibodies induced against the peptides 5-11 was tested in direct and competitive ELISA, it was found that these antibodies reacted in the same way with the various antigens as the other IgG subclasses except in the following two cases: 1) the IgG3 anti-peptide antibodies, especially those induced against peptides 7-11, reacted well and, in some cases, strongly with histone H3 (Fig. 3, GP). 2) free peptides 4 and 11 but not the other peptide analogues inhibited the binding between H3 and IgG3 antibodies induced against peptide 11 (Fig. 44). In reciprocal tests, free peptides 4 and 11 but not the other peptide analogues, inhib- Mouse antisera were diluted 1500 and anti-mouse IgG3 second antibody 15,000. Absorbance values were measured at 450 nm. The molar excesses of peptides 4 and 11 over H3 required to inhibit 50% of the antibody binding were about 300 in both cases.
ited the binding between H3 and IgG3 antibodies induced against peptide 4 (Fig. 4B). IgG3 antibodies raised against the all D-peptide 11 reacted strongly with histone H3 in solution. As shown in Table V, the parent protein H3 inhibited the binding of anti-peptide 11 antibodies to peptide 11. Similar results were not observed with IgG3 antibodies raised against other peptide analogues (Table VB) nor with IgG antibodies of other subclasses (Table VA). DISCUSSION Among the factors thought to affect the duration of the immune response, persistence of the antigen in an appropriate location is considered to play an important role (Gray and Skarvall, 1988). Thus one of the major problems in developing synthetic vaccines is to enhance the half-life of the peptides and increase the probability that they will interact with cells of the immune system, in particular with dendritic cells. In this study, we have used the liposome-peptide model developed previously (Friede et al., 1993) and explored the effect of replacing L-amino acid residues by the corresponding Damino acids on both the duration and specificity of the immune response. To the best of our knowledge this is the first detailed investigation of the influence of enantiomeric substitutions on the immunogenicity of peptides.
As measured by the antigenic properties of the seven analogues produced in this study, the data suggest that neither the substitution of Arg13' and Arg'% by Lys residues nor the replacement of Arg131 by a D-Arg changed the conformation of the resulting peptides sufficiently to alter the binding of rabbit and mouse antibodies induced against H3 or against peptides 4, 5, and 6. In contrast, the substitution of G l P , k g " , and Ala"' by the respective enantiomers (peptides 7, 8, and 9) drastically altered the antigenicity of the modified peptides. In none of the presentations (either coupled to BSA by means of SPDP or free in solution in inhibition assays) were the peptide analogues containing D-residues at positions 133, 134, and 135 recognized by antibodies against H3 or peptides 4, 5, and 6. The various partially D-analogues also differ significantly among themselves since they were not recognized by antibodies induced against the non-homologous peptide analogues either in the direct or in the competition ELISA format where peptides are in the free form.
Regarding the immunogenicity of the D-analogues, we found that each of the six D-analogues when coupled at the surface of SUV containing MPLA induced an immune response. The response, as revealed with anti-total IgG conjugate (which very likely detects IgG1, IgG2a, and IgG2b but not IgG3) was somewhat lower in the case of peptides 7 and 11 and, in general, of shorter duration in the case of all peptides containing D-residues. In contrast, IgG3 antibodies appeared later in the serum of immunized mice, reached a maximum in bleedings 4-5 and, in some cases (peptides 7, 9, and 111, remained at a significant level in bleeding 7, i.e. 70 days after the last injection. Thus the progression of IgG1, 2a, and 2b antibodies and that of IgG3 antibodies did not follow the same pattern. Furthermore, antibodies of the IgG1, 2a, and 2b subclasses were produced at a low level in response to injection of the D-enantiomer which initially led us to believe that this peptide was less immunogenic than the others. Overall, the results lead to some important conclusions. First, it seems that the presence of IgG3 antibodies in the immunized mice can be missed when an anti-mouse IgG (H+L) peroxidase conjugate is used. It is important, therefore, to determine which subclasses of IgG are actually detected by the indicator immunoconjugate and to use reagents that reveal all antibodies with the same efficacy.
Second, the high IgG3 response in all mice immunized with the various peptides coupled to the surface of SUV is particularly striking. Although we cannot directly compare the levels of reactivity of the antibodies of different subclasses because the anti-mouse conjugates used in our ELISA procedure were different and that one arbitrary working dilution of 15,000 was chosen, the IgG3 response detected in immunized mice was significantly higher than usual. This finding may result in part from the fact that we have used small liposomes with surface-bound peptide and not a carrier protein for presenting peptides, and that MPLA, and not Freund's adjuvant, was used as adjuvant. Interestingly it has been reported that mouse IgG3 subclass antibodies predominate in humoral responses to protein antigens linked to the surface of liposomes while they constitute a small component of the humoral response to encapsulated protein (ThBrien et al., 1991) which is suggestive of a T-independent B-cell activation (Mc Kearn et al., 1982;Snapper and Mond, 1993).
Third, when the specificity of antibody response is considered, the antibody population of the IgG1, 2a, and 2b subclasses appears very specific in the sense that the antibodies of these subclasses induced against peptides 7, 8, 9, and 10, for instance, recognized only the homologous analogue peptides and not the other partially modified D-analogues or the cognate protein H3. In contrast, the IgG3 population contains antibodies able to react both with the respective homologous peptides and with H3 on the plastic. Most interesting was the finding that the parent peptide as well as the protein H3, both in the free form, were also recognized by anti-all Dpeptide IgG3 antibodies. Reciprocally, IgG3 antibodies induced against the parent peptide did recognize the homologous peptide and H3 as well as the free all D-peptide. This cross-reactivity was only found between the parent peptide and the D-enantiomer (mirror image) but not when a single D-amino acid was introduced in the sequence of the hexapeptide. The common structure between a peptide and its mirror image is the peptide backbone (Guptasarma, 1992). Accordingly, the results suggest that the IgG3 antibodies recognize the backbone of the peptides 4 and 11 rather than the side chains. In this regard, it is noticeable that in this particular cross-reactivity with IgG3 antibodies to peptide 11, the parent peptide 4 and the peptide 5 containing 2 lysine residues in replacement of arginine residues were not antigenically equivalent.
It is well established that all L-amino acid polymers make right-handed a-helices whereas homopolymers of D-amino acid residues form left-handed ones. Therefore, the fact that IgG3 antibodies induced against the D-enantiomer reacted with the parent protein H3 and that IgG3 antibodies induced against the all L-peptide reacted with the free D-enantiOmer points to new possibilities of manipulating peptide antibody responses. Furthermore, the fact that it is only the IgG3 subclass that exhibits this cross-reactivity suggests that this class of antibody may bind to antigen in a unique manner possibly allowing a self-association which yields more effective binding, for example to the surface of bacteria (Greenspan and Cooper, 1992). NMR studies of peptide-antibody complexes involving these analogues and monoclonal antibodies (Cung et al., 1991) may unravel the structural basis of this particular type of interaction.