Recombinant proteins VP1 and VP3 of hepatitis A virus prime for neutralizing response

Journal of Medical Virology 31:277-283 (1990) Recombinant Proteins VP1 and VP3 of Hepatitis A - Virus Prime for Neutralizing Response Verena Gauss-Miiller, Zhou Mingquan, Klaus von der Helm, and Friedrich Deinhardt Institute for Medical Microbiology (V.G.-M.), Medical University of Liibeck, Lubeck, Federal Republic of Germany; Shanghai Hygiene and Anti-epidemic Center (Z.M.), Shanghai, P R China; and Max von Pettenkofer Institute for Hygiene and Medical Microbiology (K.V.D.H., F.D.), Ludwig Maximilian University, Munich, Federal Republic of Germany Six overlapping genomic regions of capsid pro- teins VP1 and VP3 of hepatitis A virus (HAV) in- serted into the expression vectors pBD or pUR respectively expressed p-galactosidase-HAV fu- sion proteins. The recombinant proteins were poorly soluble so they were difficult to detect by human anti-HAV sera in radioimmunoassay, but the fusion proteins dissolved in sodium dodecyl sulfate reacted with human and rabbit anti-HAV- positive sera in immunoblots. Antisera against VP1 and VP3 recombinant proteins reacted with the respective structural proteins of HAV in im- munoblots. Two recombinant proteins, one in- cluding the first 120 amino acids of the N-ter- minus of VP1 and the other containing all of VP1 except for the first 60 N-terminal amino acids, induced a transient neutralizing antibody re- sponse in rabbits. Antisera directed against other regions of VP1 and VP3 neither neutralized viral infectivity nor recognized native virus in a competitive radioimmunoassay. However, when immunized animals were challenged with a sub- immunogenic dose of HAV, all animals re- sponded with stable virus-neutralizing antibod- ies. KEY WORDS: P-galactosidase fusion proteins, anti-HAV rabbit sera, subunit HAV vaccine INTRODUCTION Both inactivated and attenuated viruses are used in conventional strategies to develop viral vaccines. Large amounts of virus are needed for inactivated vac- cines, and vaccination with attenuated virus carries the risk of reversion to virulence, a reason for continu- ing efforts to define the molecular basis of viral atten- uation. Economic and safety considerations are major stimuli for seeking alternatives to the classical ap- proaches in vaccine development, and recombinant DNA technology has made it possible to develop sub- 0 1990 WILEY-LISS, INC. unit vaccines derived from defined immunogenic re- gions of the infectious agents. Hepatitis A virus (HAV), a human picornavirus, is responsible worldwide for endemic and epidemic hepa- titis. Until now, prophylaxis against HAV infection can be provided only by passive immunization with human immunoglobulin. Inactivated and attenuated vaccines are currently under development from virus propagated in cell culture [Provost et al., 1986 a,b; Fleh- mig et al., 1989; Wiedermann et al., 19891. Since the entire viral genome has been cloned and sequenced, an experimental approach for characterizing antigenic sites has become available now [Paul et al., 19871. To identify more directly the molecular structure that contributes to antigenic properties of the viral structural proteins, substantial quantities of either pu- rified proteins or synthetic peptides representing indi- vidual epitopes have been isolated and tested for their ability to induce a neutralizing antibody response [Hughes and Stanton, 1985; Wheeler et al., 1986; Gauss-Muller and Deinhardt, 19881. Most results sug- gested that neutralizing epitopes are located on capsid proteins VP1 and VP3, and provided circumstantial evidence that such epitopes are of discontinuous na- ture. We report here on the production of well-defined antigenic regions of HAV in E. coli. We tested the an- tigenicity of various recombinant proteins represent- ing overlapping segments of VP1 and VP3 fused ami- noterminally to p-galactosidase, as well as their capacity to elicit neutralizing antibodies. Two con- structs induced transiently neutralizing antibodies; others primed the rabbit immune system for a neutral- izing antibody response which became evident by sub- sequent exposure to whole virus. Accepted for publication March 29, 1990. Address reprint requests to Dr. V. Gauss-Muller, Institute for Medical Microbiology, Medical University Lubeck, Ratzeburger Allee 160, D-2400 Lubeck, West Germany. 278 --- 5’ n n v I I 1 ” I 1 --- I , Gauss-Muller et al. plasmidA t d I €3 I C D 3 E H F 4 Fig. 1. Map of the genome of HAV strain MBB and relative position of expression plasmids within the P1 region (A-F). Vectors as in Table I. MATERIALS AND METHODS Construction of Plasmids Expressing VP1 and/or VP3 Genes Several overlapping subgenomic cDNA fragments (pBD 2A, pUR 290 5-2, pUR 290 17-7,2470,8101, and 3831) representing the coding region of VP3 and VP1 of HAV strain MBB were used for preparing constructs [Paul e t al., 1987; Ostermayr et al., 1987, 19881. The vectors used were pBD 2 and pUR 290 which are deri- vates of pUC and pUK and carry a multiple cloning site a t about nucleotides 1000 and 3000 of the P-galactosi- dase gene respectively [Broker, 1986; Ruther and Muller-Hill, 19831. Expression of VP3 in E. coli was achieved by inserting the Ssp I/Nco I fragment in frame into the vector pBD 2, thus coding for amino acids 215 to 486 of the polyprotein (plasmid A). Plasmid B was prepared by subcloning the HAV sequences of pUR 290 17-7 into pBD 2 coding for the C-terminal half of VP3 and for about one-third of VP1 [Ostermayr et al., 19871. Plasmid C (previously named pBD 2A) codes for ap- proximately 120 N-terminal amino acids of VP1 [Os- termayr et al., 19871. Plasmid D was constructed by extending plasmid C with the Bgl II/Pst I fragment (nucleotides 2622 to 3299) of plasmid pUR 290 5-2. It covers the genomic region of VP1 plus five amino acids of VP3 and about 22 amino acids of the N-terminus of P2. Plasmid E was constructed by inserting the Bam HI/Bgl I1 fragment (coding for amino acid 58 to 142 of VP1) into the Bam HI site of pBD 2. Plasmid F (named pUR 290 5-2 previously) spans the genomic region be- tween amino acid 58 of VP1 to about amino acid 22 of 2A [Ostermayr et al., 19871. The physical map and the relative position of all expression plasmids (A-F) are shown in Figure 1. Isolation of Bacterial Clones Containing HAV cDNA and Expression of Proteins Transformation of plasmid DNA into competent bac- teria (E. coli strains JM 109 and DH 5 a) was per- formed by the calcium chloride precipitation method LMandel et al., 19701. For identification and orientation of HAV-containing cDNA segments, a detailed restric- tion enzyme analysis was performed. For expression of fusion protein, 5 ml of an overnight culture was diluted into 100 ml of Luria broth supplemented with ampicil- lin (50 pg/ml) and grown a t 37°C to an optical density of about 0.4. Two hours following induction by isopro- pylthiogalactoside (IPTG), the cells were collected by HAV Recombinant Proteins 279 TABLE I. Characteristics of HAV Fusion Proteins Amino acid position Mol wt of expressed % HAV No. of amino Plasmid Vector in polyprotein protein (kd) of fusion protein acids of HAV A pBD2" 215-486 71 42 271 B pBD2 359-ca.610 69 41 252 C pBD2 487-ca.6 10 55 25 124 D pBD2 487-858 81 50 362 E pBD2 549-633 50 18 84 F ~ U R 2 9 0 ~ 549-858 144 31 309 "pBD2 codes for the first 375 N-terminal amino acids of P-galactosidase. bpUR290 codes for all of p-galactosidase. centrifugation, washed, and suspended in TE (50 mM Tris pH 7.5, 50 mM EDTA) a t a concentration of 20 OD/ml. Purification of Fusion Proteins IPTG-induced bacterial cultures (50 ml) were har- vested and suspended as described above. The insoluble fusion protein was extracted by repeated sonication and centrifugation in TE containing 0.5% Triton X 100. The final pellet was resuspended in TE. The proteins were separated on preparative 10% SDS-polyacryl- amide gels and visualized by staining with Coomassie brilliant blue. The immunoreactive band was excised and minced. The protein was eluted by 10 mM ammo- nium carbonate and 1% SDS and collected by precipi- tation with 9 volumes of acetone a t -20°C overnight. Analysis of Fusion Proteins Proteins from bacterial extracts were separated by SDS-PAGE and analyzed by Coomassie brilliant blue staining or immunoblotting [Gauss-Muller and Dein- hardt, 19881. In brief, following transfer to nitrocellu- lose membrane (Schleicher and Schull, BA 85), block- ing with 1% ovalbumin and incubation in a 8 M urea solution for 1 hour a t 60°C, the membrane was incu- bated overnight in human or rabbit antiserum directed either against whole virus or individual proteins [Gauss-Muller and Deinhardt, 19881. The immunore- active proteins were visualized with alkaline phos- phatase-conjugated goat anti-rabbit IgG or anti-human Ig (Dakopatts) and p-nitro blue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl phosphate as sub- strates. Immunization of Rabbits Rabbits were injected intradermally with approxi- mately 200 pg of protein emulsified in complete Freund adjuvant. Two to three booster injections were given subcutaneously in incomplete Freund adjuvant. In most instances, animals were challenged with a sub- immunogenic dose of HAV by injection of lo4 TCID,, intramuscularly. The animals were bled from the ear veins 2 to 3 weeks after the injections. Sera were ana- lyzed by radioimmunoassays (RIA, Abbott Laborato- ries), immunoblot, and neutralization tests (NT) in cell culture as described before [Gauss-Muller and Dein- hart, 19881. RESULTS Expression and Antigenicity of P-galactosidase-HAV Fusion Proteins Figure 1 depicts the overlapping HAV expression plasmids A to F and their mapping positions along the HAV genome. All plasmids contain one of two amino acids (Asp 70 in VP3 or Ser 102 in VP1) which have been described as immunodominant by analysis of neu- tralization escape mutants [Stapleton and Lemon, 1987; Ping et al., 19881. P-galactosidase-HAV fusion proteins were expressed after induction with IPTG, separated on SDS-PAGE, and detected by Coomassie blue staining (Table I and Fig. 2a). Plasmid B was not analyzed in this study since a similar construct had been described before [Ostermayr et al., 19871. Nonin- duced cells produced only minute amounts of recombi- nant proteins (not shown). Solubility of fusion proteins was in general poor in 0.5% Triton X 100 (less than lo%), as can be seen in Figure 2a. Subsequent extrac- tion with 1% SDS rendered almost all of the recombi- nant products soluble. HAV antigenicity of the major bands stained by Coomassie blue could be shown by reaction with rabbit serum directed against whole vi- rus following immunoblotting and incubation in 8 M urea at 60°C (Fig. 2b). In the Triton X 100-insoluble fraction of C a protein of approximately 65 kd immu- noreacted with the rabbit serum next to the expected fusion protein of 55 kd (Fig. 2b,C). When crude bacte- rial extracts were tested by RIA for their reactivity with human anti-HAV-positive sera, none gave posi- tive results, presumably because of the low solubility of HAV fusion proteins. However, once the recombinant products were solubilized by SDS, separated by PAGE, immunoblotted, and subsequently treated by 8 M urea, human anti-HAV-positive sera recognized all these proteins (not shown). This indicates that all our recom- binant HAV proteins are capable of refolding into or exposing naturally occurring epitopes provided that ex- perimental conditions are optimal. Immunogenicity to HAV Fusion Proteins In a second series of experiments, fusion proteins were used as immunogens in rabbits. By this approach, 280 -144kd Gauss-Muller et al. 69 / - 81 kd 71 kd a b a b a b a b a b a b a pBD2 C D F E A 92 kd - 68 kd - 46kd- a b a b a b a b a b a b pBD2 C D F E A Fig. 2. Analysis of bacterial lysates expressing HAV fusion pro- teins. a: Crude bacterial lysate was extracted with 0.5% Triton X 100 in TE, and the proteins of the soluble (a) and insoluble (b) fractions were separated on 106 SDS-PAGE and stained by Coomassie blue. Extracts were prepared from bacteria transformed with vector pBD 2, and plasmids C , D, F, E, A. b: The soluble (a) and insoluble (b) frac- tions of bacteria transformed by plasmids C , D, F, E, and A were separated by 10% SDS-PAGE, immunoblotted, treated with 8 M urea, and analyzed with rabbit sera directed against whole virus. we sought to identify immunodominant epitopes on HAV recombinant proteins. On the basis of their low solubility, HAV fusion proteins were partially purified from the bulk of bacterial proteins and subsequently separated by SDS-PAGE in preparative amounts. The major bands carrying VP1 or VP3 antigenicity were cut out of the gels, eluted, and concentrated by acetone precipitation. About 200 +g of purified fusion proteins were used for each injection into rabbits. After four to five injections, animals were challenged with a subim- munogenic dose of native HAV to see whether their immune systems were primed. All antisera were tested by immunoblot analysis, RIA, and NT in cell culture. The data are summarized in Table 11. All recombinant immunogens induced antibodies that reacted with the respective viral protein in immunoblot. However, most rabbits did not produce antibodies directed against na- tive virus when tested by RIA or NT. Only antisera of the rabbits immunized with the fusion proteins coded for by plasmids C and F reacted transiently in compet- itive RIA and in NT. These two fusion proteins were each injected into three rabbits, all of which produced HAV Recombinant Proteins 28 1 TABLE 11. Reactivity of Antisera to HAV Fusion Proteins Before challenge After challenge" Immunosen Immunoblot" RIA^ N T ~ RIA^ N T ~ + + n.t. n.t. + + + + - - VP3 A B VP3 VP1 (+I" + + VP1 (+IC C D VP1 VP1 - (+Y (+IC + + E F Whole virus VP1, VP2, VP3 + + n.t. n.t. None "The antisera were tested in a dilution of 1500 with extracts of HAV- and mock-infected cells or partially rurified virus. 'The reactivity of these rabbit sera was transient. Antisera obtained 10 days later was negative. dNeutralization test. 'The rabbit immune system was challenged by an intramuscular injection of a subimmunogenic dose of HAV. - - - - - - - - - - - Reaction with native virus was tested in a modified radioimmunoassay (RIA). neutralizing antibodies 62 days after the first injection. Antisera obtained a t day 84 no longer reacted with native virions in RIA or NT. Interestingly, when primed animals irrespective of their initial responses were challenged by injecting a subimmunogenic dose of native HAV, the resulting antiserum neutralized HAV infectivity in cell culture and competed with human anti-HAV-positive sera in RIA. Control animals which had received only a subimmunogenic dose showed no anti-HAV response (not shown). These results indicate that recombinant proteins representing VP3 and VP1 of HAV or only parts of each contain antigenic sites present on the native virion. DISCUSSION Compared to other picornaviruses, relatively little is known about the antigenic structure of HAV. Iodina- tion of highly purified virions indicated that VP1, VP2, and VP3 are exposed on the surface of the viral parti- cles [Gerlich and Frosner, 1983; Robertson et al., 19891, and a three-dimensional model has been proposed but no experimental data are available yet to verify the model [Luo et al., 19881. Epitope mapping using pep- tide antigens was little help in identifying neutralizing antigenic sites because only one of the peptides was reported to induce HAV-neutralizing antibodies [Em- ini e t al., 1985; Wheeler et al., 1986; Gauss-Muller and Deinhardt, 19881. Two immunodominant antigenic sites within VP1 and VP3 were identified by determi- nation of the amino acid sequence of neutralization mutants [Stapleton and Lemon, 1987; Ping et al., 19881. Both sites (residue 70 of VP3, Asp, and residue 102 of VP1, Ser) are parts of knob or loop structures predicted by the three-dimensional model and align with antigenic sites in polio and rhino virus [Luo et al., 19881. In our study, different parts of VP1 and VP3 were expressed and the antigenicity and immunogenic- ity of the resulting polypeptides were tested, to deter- mine the most probable antigenic sites on these viral proteins. Since all recombinant proteins were recog- nized by human anti-HAV-positive sera using our pro- tocol for immunoblots (which includes a renaturation step with 8 M urea) and all proteins tested primed for a neutralizing response, all carry relevant antigenic sites and can fold into the appropriate antigenic con- formation. For expression of recombinant HAV fusion proteins, vectors pBD 2 or pUR 290 which carry multiple cloning sites within or at the 3' end of the P-galactosidase gene were chosen [Broker 1986; Ruther and Muller-Hill, 19831. Since P-galactosidase is known to provide T helper cell specific antigenic sites, fusion proteins should elicit a potent immunological response. Anti- genic peptides of foot-and-mouth disease virus proteins fused to P-galactosidase have been expressed success- fully as soluble recombinant proteins [Winther et al., 19861. In contrast, the recombinant proteins described here were mainly insoluble although the relative pro- portion of HAV never exceeded 50% of the entire pro- tein. Due to their low solubility, HAV fusion proteins could be freed from the bulk of bacterial proteins, but were undetected in an RIA based on human anti-HAV- positive sera. Once the insoluble bacterial lysates were solubilized by SDS and separated under denaturing conditions by SDS-PAGE, transferred to nitrocellulose membrane, and treated with urea, the recombinant proteins were reactive with human sera in immuno- blot. Under these conditions, in general a stronger reac- tion was observed with rabbit sera directed against whole virus or individual proteins than with human sera. All recombinant proteins were recognized by anti- HAV rabbit sera, although a t different intensities. Fu- sion proteins including large epitopes of HAV (up to 362 amino acids for protein D) reacted much more strongly than those presenting small epitopes (83 amino acids of VP1 for protein El, indicating that in our expression system the tested regions of VP1 and VP3 seem to assume at least partially an antigenic structure similar to the native surface of the virion. In an earlier comparison of the reactivity of two of the six recombinant proteins (F and C) with rabbit an- 282 Gauss-Miiller et al. tiserum directed against whole virus, we found that the ACKNOWLEDGMENTS fusion protein covering the N-terminal part of VP1 (C) was reactive but not the protein containing the C-ter- minal portion of VP1 (F), leading us to speculate that a site of HAV antigenicity may be located in the region of the first 100 amino acids of the N-terminal part of VP1 [Ostermayr et al., 19871. In subsequent immunoblot experiments, we observed that by incubation of blotted viral proteins in a solution of 8 M urea a t 60°C more immunoreactivity could be achieved, probably because the proteins refolded [Gauss-Muller and Deinhardt, 19881. In the present study, to achieve the highest sen- sitivity possible, the blotted proteins were always treated with 8 M urea. Thus, all recombinant proteins were recognized by rabbit and human anti-HAV sera, indicating that they are a t least parts of regions ex- posed on the viral surface and represent immunodom- inant epitopes. Using different vectors, other investigators have ex- pressed HAV structural proteins and tested their abil- ity to elicit neutralizing antibodies [Johnston et al., 1987; Ross et al., 19881. A recombinant protein similar to our construct D containing parts of VP3, all of VP1, and 129 amino acids of P2 fused to Trp E coding se- quence also primed for an HAV-neutralizing response in rabbits [Johnston et al., 19871. Antisera to another, similar construct did not neutralize viral infectivity [Ross et al., 19881. Our observation that a transient neutralization re- sponse is induced by two of our recombinant constructs cannot be explained only by their HAV antigenicity. In contrast, our earlier observation that these fusion pro- teins seem to differ in their antigenicity (only protein C reacted with rabbit anti-HAV sera) indicated rather that only the N-terminal part of VP1 carries a n anti- genic site [Ostermayr et al., 19871. Since both recom- binant proteins (C and F) elicited the transient produc- tion of neutralizating antibodies in three animals each immunized by an identical immunization schedule in the present study, we must assume that they both carry an important antigenic site of VP1. Fusion pro- tein E which carries the antigenic region included in C and F, however, was not sufficient to elicit a neutral- izing response but primed for it. The lack of induction of neutralizing antibodies by protein E and the fusion protein described by Ross et al. as compared to proteins C and F may be due not only to a genuine difference in antigenicity or immunogenicity but also to varying amounts of immunogen used or different abilities of animals to respond. It cannot be exluded that a tran- sient immune response was missed in the case of B, D, and E. In addition, purification has a great impact on the immunogenicity of recombinant proteins [Winther et al., 19861. In conclusion, P-galactosidase-HAV fusion proteins function as antigens and immunogens priming for a neutralizing response. A combination of two or three of these proteins may be a n even better immunogen, and this will be tested. We thank U. Schwieger and S. Seelmair for excellent technical assistance, M. Broker for providing plasmid pBD 2 and the antiserum directed against plasmid B, and P. Muller for critical reading of the manuscript. This project was supported by the Deutsche Forschungs- gemeinschaft (Ga 304/1-2) and the Forderverein zur Bekampfung der Viruserkrankungen e.V. REFERENCES Broker M (1986): Vectors for regulated high-level expression of pro- teins fused to truncated forms of E.coli p-galactosidase. Gene Analysis and Technology 3:53-57. Emini EA, Hughes JV, Perlow DS, Boger I (1985): Induction of hep- atitis A virus-neutralizing antibody by a virus-specific synthetic peptide. Journal of Virology 55:836-839. Flehmig B, Heinricy U, Pfisterer M (1989): Immunogenicity of a killed hepatitis A vaccine in seronegative volunteers. Lancet: 1039-1041. Gauss-Muller V, Deinhardt F (1988): Immunoreactivity of human and rabbit antisera to hepatitis A virus. Journal of Medical Virology 24:219-228. Gerlich WH, Frosner GG (1983): Topology and immunoreactivity of capsid proteins of hepatitis A virus. Medical Microbiology and Immunology 172:lOl-106. Hughes JV, Stanton LW (1985): Isolation and immunization with hepatitis A viral structural proteins: Induction of antiprotein, an- tiviral and neutralizing responses. Journal of Virology 55:395- 401. Johnston JM, Harmon SA, Binn LN, Richards OC, Ehrenfeld E, Sum- mers DF (1987): Antigenic and immunogenic properties of a hep- atitis A virus capsid protein expressed in E.coli. Journal of Infec- tious Diseases 157:1203-1211. Luo M, Rossmann MG, Palmenberg AC (1988): Prediction of three- dimensional models for foot-and-mouth disease virus and hepatitis A virus. Virology 166503-514. Mandel M, Higa A (1970): Calcium dependent bacteriophage DNA infection. Journal of Molecular Biology 53:154. Ostermayr R, von der Helm K, Gauss-Muller V, Winnacker EL, Dein- hardt F (1987): Expression of hepatitis A virus cDNA in E. coli: Antigenic VP 1 recombinant protein. Journal of Virology 61: 3645-3647. Ostermayr R, von der Helm K, Seelmair S, Gauss-Muller V, Dein- hardt F (1988): Expression of HAV VP 1 antigen as recombinant protein in E. coli. In Zuckerman A (ed): “Viral Hepatitis and Liver Disease.” New York: Alan R. Liss, pp 59-61. Paul AV, Tada H, von der Helm K, Wissel T, Kiehn R, Wimmer E, Deinhardt F (1987): The entire nucleotide sequence of the genome of human hepatitis A virus (isolate MBB). Virus Research 8:153- 171. Ping LH, Jansen RW, Stapleton JT, Cohen JI, Lemon SM (1988): Identification of an immunodominant antigenic site involving the capsid protein VP3 of hepatitis A virus. Proceedings of the Na- tional Academy of Sciences USA 8523281-8285. Provost PJ, Hughes JV, Miller WJ, Giesa PA, Baker FS, Emini EA (1986a): An inactivated hepatitis A viral vaccine from infected cell culture origin. Journal of Medical Virology 19:23-31. Provost PJ, Bishop RP, Gerety RJ, Hilleman MR, McAleer WJ, Scolinck EM, Stevens CE (1986b): New findings in live, attenu- ated hepatitis A vaccine development. Journal of Medical Virology 20:165-175. Robertson BH, Brown VK, Holloway BP, Khanna B, Chan E (1989): Structure of hepatitis A virion: Identification of potential surface- exposed regions. Archives of Virology 104:117-128. Ross BC, Anderson BN, Gust ID (1988): Expression of the hepatitis A virus genome as P-galactosidase fusion proteins in Exoli. In Zuck- erman A (ed): “Viral Hepatitis and Liver Disease.” New York: Alan R. Liss, pp 62-64. Ruther U, Muller-Hill B (1983): Easy identification of cDNA clones. EMBO Journal 2:1791-1794. Stapleton JT, Lemon SM (1987): Neutralization escape mutants de- fine a dominant immunogenic neutralization site on hepatitis A virus. Journal of Virology 61:491-498. HAV Recombinant Proteins 283 Wheeler CW, Robertson BH, Van Nest G, Dina D, Bradley DW, Fields HA (1986): The structure of the hepatitis A virion: Peptide map- ping of the hepatitis A virus capsid region. Journal of ViroloD 58:307-313. Wiedermann G, Ambrosch F, Hofmann H, Kunz C, Safary A (1989): First results on safety and immunogenicity of a new inactivated hepatitis A vaccine. Abstract to the VIII. Congress of liver dis- eases, Basel. Winther MD, Allen G, Banford RH, Brown F (1986): Bacterially ex- pressed antigenic peptides from foot-and-mouth disease virus cap- sid elicit variable immunological responses in animals. Journal of Immunology 134:1835-1840.