- Email: [email protected]
Topographical analysis of epitope relationships on the envelope glycoprotein of yellow fever 17D vaccine and the wild type asibi parent virus
VIROLOGY
150,333-341 (1986)
Topographical Analysis of Epitope Relationships on the Envelope Glycoprotein of Yellow Fever 17D Vaccine and the Wild Type Asibi Parent Virus N. CAMMACK’
AND E. A. GOULD
Arbovims Research Unit, London School of Hygiene and Tropical Medicine, Winches Farm Field Station., 395 Hageld Road, St. A&MS, He& AL& OXQ, England Received October 7, 1985; accepted December 20, 1985 Monoclonal antibodies (MCA), with defined molecular specificity, were used in a competition binding enzyme-linked immunosorbent assay (ELISA) to locate the relative positions of the epitopes on the envelope glycoprotein of yellow fever 17D vaccine virus and its wild type parent virus, Asibi (AS). Five topographically distinct antigenic domains were defined on the E glycoprotein of the 1’i’D vaccine. Three of these (A, B, and C) were represented by one MCA each, a fourth (D) was represented by two MCA, and a fifth domain (E) comprised a major cluster of at least five overlapping epitopes. Asibi virus also possessed domain E which is proposed to be a conserved antigenic region within the envelope glycoprotein of all flaviviruses. Domains A and C were not represented on Asibi virus and one epitope, situated proximal to the E domain, showed structural alterations in physical overlap. Functional activities were assigned to physically mapped epitopes by haemagglutination inhibition (HAI), virus neutralisation (N), and passive protection in mice. The HA1 and N functions were not necessarily linked but only MCA with N activity were able to protect mice passively against lethal infection. All domains demonstrated a heterogeneous range of biological properties dependent upon the virus strain rather than the epitope. 8 1986Academic press, he. INTRODUCTION
Yellow fever (YF) is a mosquito-transmitted virus disease, It is classified as the type species of the family Flaviviridae (Westaway et al., 1985) and has considerable potential to act as a human pathogen despite the current availability of safe and effective vaccines. The yellow fever virus vaccines were developed empirically but it is now felt that a new and improved vaccine strain should be derived (see Report of the Pan American Health Organization, Washington D. C., 1984). The major viral surface component is the envelope glycoprotein which is responsible for the biological properties of haemagglutination (HA) (Della-Porta and Westaway, 19’77a) and neutralisation (N) (Qureshi and Trent, 1973). On the basis of complement fixation ‘To whom requests for reprints should be addressed. 333
haemagglutination inhibition (HAI), N tests, and the natural invertebrate host, a system of classification was derived which incorporated seven antigenic subgroups or complexes (Porterfield, 1980). Cross-protection virus-challenge experiments in animals have also demonstrated the antigenic relationships of the flaviviruses (Hammon and Sather, 1956; Tarr and Hammon, 1974). Moreover, monoclonal antibodies (MCA) specific for the envelope glycoprotein can confer immunity in mice to St. Louis encephalitis virus lethal infection (Mathews and Roehrig, 1984). In order to understand more completely the significance of flavivirus-host interactions and the changes that occur in vaccines as they are derived from parent viruses, it is necessary to know something of the three-dimensional structure of the envelope glycoprotein. Competition binding assays have previously been carried out to determine the relative distribution of ep0042-6822/86 $3.00 Copyright All rights
Q 1996 by Academic Press. Inc. of reproduction in any form reserved.
334
CAMMACK
itopes on the envelope glycoproteins of several flaviviruses (Heinz et aL, 1983, Kimura-Kuroda and Yasui, 1983; Roehrig et al, 1983; Henchal et al, 1985). We recently prepared and reported MCA which display a wide range of antigenic reactivities with all flavivirus subgroups (Gould et a& 1985b). Five antigenic domains were identified on the YF 17D vaccine virus envelope glycoprotein: 17D-specific, vaccine-specific, YF-specific, intermediate reactivity with flaviviruses, and broad reactivity with flaviviruses. We now report a comparison of the topographical maps of the envelope glycoprotein of YF 17D vaccine virus and the wild type Asibi (AS) parent YF virus. The relevant antigenic determinants have also been characterised using HAI, N, and mouse passive protection tests so that functional properties can be assigned to the physical maps.
AND
GOULD
Production of immune mouse asciticfluid Seven to ten days after priming 8-week-old male BALB/c mice with pristane (2,6,10,14tetramethyl pentadecane), 5 X lo6 of the appropriate hybridoma cells were inoculated intraperitoneally. Ascitic fluids were collected, clarified (2000 g for 10 min) and stored at -20”. Pur$cation and biotinylation of hybridomu antibodies. Monoclonal antibodies were clarified by centrifugation first at 2000 g for 10 min followed by 10,000 g for 30 min. Aliquots were dialysed using 0.02 M Tris-HCl, pH 8.0, and centrifuged at 2000 g for 5 min. The proteins were purified by affinity chromatography with DEAE affigel blue (Bio-Rad, England) to obtain the IgG immunoglobulin. Optimal fractions, identified by immunofluorescence titrations with YF-infected cell monolayers, were dialysed with 0.005 M PBS and concentrated by forced dialysis. Protein concentrations were estimated using the MATERIALS AND METHODS Bio-Rad Protein Assay (Bio-Rad) and adViruses and cells. The 1’7D-204 Arilvax justed to 1 mg/ml. The purified immunovaccine Batch BYF 1228 (Wellcome Labo- globulin was biotinylated using biotinylratories, England) and the parent Asibi N-hydroxy succinimide ester (BHSE, (AS) virus, kindly supplied by Dr. R. Fitz- Sigma, England) according to the method george (PI-IL& Porton, Salisbury, England) of Guesdon et aL (1979). The optimal conwere used throughout. Ten percent suck- centration giving biotinylated probes with ling mouse brains suspensions of these vi- strong self-competition in the competition ruses at an estimated input multiplicity of binding assay, varied between 75 and 200 0.1 PFU/cell were used to infect monolay- pg BHSE/mg protein. The reaction was alers of Vero cells maintained at 37’ in Lei- lowed to proceed for 4 hr at room temperbovitz L15 medium containing 2% fetal calf ature, before fractionation by Sephadex serum (FCS). Supernatant medium from chromatography (G-75). Peak fractions infected cells was clarified by centrifugawere dialysed extensively against 0.005 M tion at 2000 g for 10 min and virus was PBS and concentrated by forced dialysis. precipitated overnight at 4” with 7% polyCompetition antibody blocking assays usethylene glycol (PEG) and 0.4 M sodium ing the ELISA test. The standard ELISA chloride. Precipitates were collected by assay, using biotin and streptavidin, was centrifugation at 3000 g for 30 min. Virus described previously (Gould et al, 1985a). concentrates (loo-fold) were resuspended The titer of each antibody was estimated in borate-buffered saline at pH 9.0. as the reciprocal of the highest dilution Monoclonal antibodies. The preparation, showing an absorbance of 0.1 greater than the control. The blocking assay was based designation, isotype, and immunofluorescence characteristics of most of the mono- upon competition between biotinylated and clonal antibodies used in this report were unlabelled MCA for a previously deterantigen concentration. given previously (Gould et a& 1985b). Ad- mined limiting ditional antibodies (MCA 110,126, and 140) Nonspecific binding was reduced to a minare included in this report. These were imum by the addition of PBS-Tween 20 prepared against AS virus using the meth- containing 5% newborn calf serum for 1 hr. Competitor antibodies were added in ods described previously.
MAPPING
YF VIRUS
ENVELOPE
serial lo-fold dilutions and incubated overnight. The plates were then washed and biotinylated MCA was added for 1 hr. Binding was detected using the streptavidin-biotinylated peroxidase (S/P) complex and OPD substrate as described previously. The percentage binding was calculated by comparing the absorbance readings recorded in the presence of nonlabelled MCA with those recorded in the presence of control ascitic fluid (a monoclonal antibody against HBsAg). HAI tests. These assays followed the protocol of Clarke and Casals (1958) except that 24-hr chick erythrocytes were used at pH 6.2. Neutralisation assay. Vero cell monolayers in 5-cm petri dishes were infected with an estimated 50 PFU/dish of YF 1’7Dvirus in either the presence or absence of serial dilutions of heat inactivated (56” for 30 min) MCA. After absorption for 1 hr at 3’7” the monolayers were washed with 0.05 M PBS, pH 7.4, and Leibovitz-L15 medium containing 2% FCS and 1% agarose (Seakem, USA) was added. After incubation at 37” for ‘7days monolayers were fixed with 10% Formalin in saline and stained with naphthalene black. The titer of each antibody was the reciprocal of the highest dilution that reduced plaque numbers by 50%. Mouse passive protectiw (PP) tests. Groups of eight mice, 3-4 weeks old (strain TO, outbred; Tuck and Son, Bettlesbridge, Essex) were inoculated intraperitoneally with 0.1 ml of undiluted monoclonal ascitic fluid. After 24 hr the mice were challenged intracerebrally with 50 LDm per mouse of either YF 17D or AS virus. The antibody was considered positive if at least 50% of the mice survived. Negative control mice were inoculated with a monoclonal antibody against HBsAg. Immunofluorescence (IF) tests. These were performed as described previously (Gould et al, 1985a). RESULTS
Biological Characteristics of MCA A total of 11 MCA were used. Their designation, molecular specificity, isotype,
GLYCOPROTEIN
335
HA1 and N titers, reactivity with YF 17D virus in either indirect immunofluorescence or ELISA tests, and their ability to protect mice in passive antibody transfer experiments are all summarised in Table 1. All MCA reacted specifically with the 54K envelope glycoprotein of YF 17D virus (Gould et aJ., 198513).The results obtained by IF and ELISA tests were identical: MCA 864 and 411 reacted only with 17D virus but the other antibodies reacted equally well with both viruses as shown when antibody binding profiles were compared (see Cammack and Gould, 1986). However, whether or not HAI, N, or PP occurred was independent of antibody binding to the virus. Generally, assignment of biological properties to individual epitopes was dependent on the virus strain. The three tests showed that HAI, N, and PP were not linked; MCA 612 had HA1 but not N activity with AS virus, whereas MCA 427 had N but no HAI activity with YF 17D virus. MCA 411, the vaccine-specific antibody, had no biological activity but its titer by ELISA titration was comparable with other MCA (Table 1). For YF 17D virus, only MCA with N activity also showed PP activity. Moreover, only MCA 110 showed activity in N tests with AS virus and none did in PP tests with AS virus. Competition Blocking Tests Using YF 17D virus Each of the 11 MCA were used in competitive antibody blocking tests with YF 17D virus as described above. Preliminary studies indicated that a PEG-concentrated supernatant virus from virus-infected cells provided the most suitable antigen source. With this type of preparation, antibodies showed binding characteristics equivalent to those obtained with highly purified virions (unpublished results). The antibody binding specificities of all MCA, except 612 and 110, were retained after biotinylation. Thus, where possible, reciprocal competition experiments were performed to account for variation in antibody avidity, since lateral shifts of binding curves do indicate avidity differences. Each biotinylated antibody (designated BMCA, i.e.,
CAMMACK
336
AND GOULD
TABLE 1 CHARAC~ERISATION OFEPITOPESONTHE E GLYCOPROTEIN OFYELLOWFEVER (YF) VIRUS Passive protection Survivors/total
Endpoint (l/log,,) HA1
N
MCA
Subclass
Epitopeb
IF
ELISA
YFl’7D
YFAS
YFlPD
YFAS
YF17D
YFAS
864 825 411 126 140 110 813 843 868 612 427
IgG2A IgG2A IgG2A IgG2A IgG3 IgG3 IgG2A IgG2A IgG2A IgG3 IgG2A
gP54A gp54* gP54c gP54D gP54D gP54E g~54~ gp54’ gP54E gp54” gP54E
YF17D-204 YF-type YF-vaccine Flavi-inter Flavi-inter Flavi-broad Flavi-broad Flavi-broad Flavi-broad Flavi-broad Flavi-inter
6.0 6.0 5.7 5.8 5.0 5.7 5.7 5.7 6.0 5.4 5.7
4.3 1.9
6.6 4.2 cl.0 cl.0 2.2 4.7 3.0 2.8 4.4 1.3 2.6
16/16 12/16 O/16 2/16 4/16 l/16 9/16 l/16 12/16 2/16 4/16
O/16 O/16 O/16 O/16 O/16 O/16 O/16 O/16 O/16 O/16 O/16
‘Endpoint ELISA values represent relative avidities of MCA for purified YFlPD virus (2 gg/well). Haemagglutination inhibition (HAI) was performed using 4-8 HA units at pH 6.2.Neutralisation (N) was performed on Vero cells with 50 PFU/test. All results are expressed as l/log,,, endpoint antibody dilution. bThe letters A, B, C, D, and E represent the domains on the glycoprotein referred to in the text. ’ Immunofluorescence reactivities with flaviviruses as defined in the Introduction.
B825) was titrated to find an optimal concentration on the plateau of maximum binding. The conjugates were then challenged against a range of dilutions of unlabelled homologous or heterologous MCA. Binding of BMCA was inhibited by 70100% in the presence of excess unlabelled homologous MCA. The results are presented in Fig. 1. For YF 17D virus, MCA 411 showed no reciprocal or nonreciprocal competition with any other MCA (domain C); B864 was blocked nonreciprocally by MCA 825, 843, 813, and 427, whereas B825 was nonreciprocally blocked just by MCA 843 and 612. Only reciprocal and not nonreciprocal blocking was taken to indicate overlapping sites, therefore MCA 864 and 825 were each assigned to separate and distinct antigenic domains (A and B, respectively). On the other hand, MCA 813, 843, and 868 all showed reciprocal blocking with each other and, together with 612 and 110, which had shown one-way blocking, were placed together as one domain (E). Since MCA 427 was reciprocally blocked by 868 and 813 it was considered to be proximal to domain E. Finally, MCA 126 and 140 overlapped to
form another distinct domain (D) although both were nonreciprocally blocked by MCA 110 and were able to block (nonreciprocally) MCA 868. Competition Blocking Tests Using YF As&i Virus YF AS virus was used in a competition binding ELISA with each of the 11 MCA described above. There was remarkable similarity (Fig. 2) with each of the five overlapping antibodies represented by domain E for YF 17D virus. Furthermore, MCA 427 still retained its proximity to this major cluster of epitopes but overlapped with MCA 868 and 843 instead of 868 and 813. A further difference in the relative positions of the epitopes represented on the two viruses was seen with MCA 110 and 140 which did not nonreciprocally block MCA 868 as they had previously been shown to do with YF 17D virus. Assignment of Biological Functions to Maps Figures 3a and 3b show our attempts to construct maps of the epitopes based on their relative physical positions in the en-
MAPPING
YF VIRUS
ENVELOPE
80
60
40
20
loo
0
s E B 0
337
ruses, respectively. As can be seen, with YF 17D virus, different biological functions were represented in all domains. Thus, within an antigenic domain the biological properties of individual epitopes were not necessarily identical. MCA in domains A and B showed functional activity in all three biological tests, whereas MCA 411 in domain C did not react in any test. All five broadly cross-reactive MCA in domain E reacted in HA1 and N tests but only two (MCA 813 and 868) showed PP activity. Neither MCA 140 nor 126, in domain D, protected mice but MCA 140 had HA1 and N activity whereas MCA 126 had no N activity. In view of the differences in biological activity between YF 17D and AS virus it is not surprising that the functional maps appeared so different. Only domains B, D, and E were represented on the physical
1w
ii z 5
GLYCOPROTEIN
60
60
40
20
100
60
60
40
20
100
Ii
I I
-2
I -3
I -4
III -5
I -1 LOGlO
-2
1 -3 MC/\
I -4
-6
-*
-2
-3
-4
-5
60
DILUTION 60
FIG. 1. Competitive blocking ELISA with biotinlabelled MCA (BMCA) directed against the YF 17D E glycoprotein. Serial lo-fold dilutions of unlabelled MCA were allowed to react with YF antigen-coated microtitre plates. Unbound MCA were removed and the binding of a previously determined limiting concentration of biotin-labelled MCA was recorded. The extent of blocking by unlabelled MCA was calculated as described in the text and is expressed as percentage binding BMCA. In each case only the data for blocking MCA are presented, since all others showed 100% binding. 0,864; 0,825; A, 427; A, 612; Cl, 868; n , 843; 0,411; x, 813; 0,110; 0,126; v, 140.
40
20
loo
velope glycoproteins of YF 17D and AS viruses, respectively. Overlapping closed circles represent overlapping epitopes or epitopes thought to be in very close proximity. The open circles for 612 and 110 are proposed sites based on one-way blocking experiments. Figures 4a and b have been derived by analysis of the hypothetical physical maps and the data presented in Table 1. These figures show our attempts to assign functional a&Vi&g to the maps Of the envelope glycoproteins of both YF 17D and AS vi-
80
60
40
20
LOGlo
MCA
DILUTION
FIG. 2. Competitive blocking ELISA with biotinlabelled MCA (BMCA) directed against the YF AS E glycoprotein. Assays were performed as described in the legend for Fig. 1.
CAMMACK
333
FIG. 3. Physical model for the antigenic domains on the E glycoprotein of (a) YF 17D vaccine virus and (b) YF AS wild type parent virus. Overlapping closed circles represent reciprocal blocking between MCA. The open circles for MCA 110 and 612 are proposed sites based on one-way blocking experiments. Domains are labeled A, B, C, D, and E.
map and in contrast to the YF 17D, domain B showed no biological activity. Furthermore, none of the MCA in domains D and E had PP activity and only MCA 110 (domain E) neutralised AS virus infectivity. However, both MCA in domain D and three MCA (110, 813, and 612) in domain E did demonstrate HA1 activity. DISCUSSION
Knowledge of the topographical distribution of protective epitopes on the flavivirus envelope glycoprotein may have important implications for the future development of synthetic vaccines. Moreover, the arrangement of biologically active determinants may influence the outcome of infection, in other words, determine the pathogenesis of the virus.
AND GOULD
In this report we have attempted to combine the physical data from competition blocking assays with those from functional biological tests. In experiments with mice we have recently found that the average survival times after intracerebral challenge are longer (12.25 days) for the 1’7D virus than for AS virus (7.65 days) (Gould et aL, 1986). Thus, by analysing and comparing the envelope glycoprotein of the 17D vaccine derivative of YF virus with the parent wild type strain it was hoped that markers of virus virulence might be identified. Five topologically distinct domains were located on the 17D envelope glycoprotein. In many instances the overlap of epitopes was defined by complete reciprocal competition between antibodies with differing functional reactivity patterns. Until recently, this may not have seemed possible since neutralisation of virus infectivity, for example, was considered to be dependent upon antibody attachment to a critical area of the viral surface (Della-Porta and Westaway, 1977b). It is now known, however, that a critical site for a monoclonal antibody can determine whether or not neutralisation occurs (Buckley and Gould, 1985). Thus, the mere presence of an epitope in the glycoprotein does not necessarily lead to neutralisation. Only one of the five domains had no functional reactivity (represented by MCA 411) in the biological tests described; possibly other epitopes in this domain will prove to be biologically active, or alternatively it may represent an area within the envelope glycoprotein that serves a structural role rather than a direct biological role. MCA 864 was blocked nonreciprocally by several antibodies but did not itself block any antibodies. Affinity differences did not account for this but structural alterations induced by antibody binding at a distant site (LeFrancois and Lyles, 1982) could lead to the epitope for MCA 864 being hidden. Nevertheless, MCA 864 defines a very important functional site on YF 17D virus as exemplified by the very high titers found in all tests. Indeed, its absence on all viruses except 17D vaccine strains might be indirectly relevant to the virulence of the YF viruses. By analogy with the concept of Brown (1984) epitopes nec-
MAPPING
YF VIRUS ENVELOPE
GLYCOPROTEIN
339
(8)
A
0 0 664
B
825
(0) I
(b)
HAI
N 0
III
PP
FIG. 4. Assignment of functional activities to the physical maps of (a) YF 17D vaccine virus and (b) YF AS wild type parent virus based on HAI, N, and passive mouse protection tests. [Refer to key.] Domains are labeled A, B, C, D, and E.
essarily involved in pathogenesis are probably more stable throughout the group and perhaps less antigenic. Five epitopes defined a major antigenic domain of the YF 1’7DE glycoprotein which exist as a cluster of overlapping sites. This domain was identically conserved on the E glycoprotein of the wild type AS parent YF virus and we therefore suggest this might be the case for all flaviviruses. An important physical difference between the 1’7D vaccine virus and wild type parent AS virus was the location of the epitope for MCA 427. On the 17D virus E glycoprotein this epitope was situated proximal to the major flavivirus domain, overlapping epitopes for MCA 868 and 813. However, on the AS virus E glycoprotein the epitope for MCA 427 overlapped MCA 868 and 843 but not 813, suggesting that
alterations in the structural configuration of the E glycoprotein had occurred. Similar physical maps were obtained for both viruses and biological activities were assigned to five domains. Nevertheless, in general for both viruses, biological properties were associated only with epitopes and not with domains. In accordance with the findings of Kimura-Kuroda and Yasui (1983) for Japanese encephalitis virus, there was no evidence of linkage between HA1 and N activity. Varying extents of HAI activity for individual epitopes may reflect differences in optimal pH activity. Indeed, altering the pH for HA1 may simply reorientate the epitope (Skehel et aL, 1982). At a given pH, low relative levels of HA1 activity may result because epitopes are in a conformationally inappropriate position. Only minor steric hindrance ef-
CAMMACK
340
fects on the cell membrane binding site would then result (Kingsford, 1984). In this comparison of the antigenic structures of YF 17D and AS viruses we detected physical similarities but major functional differences. Both serological and structural differences have previously been found between these viruses by HAI, N, oligonucleotide RNA fingerprint analysis, and indirect immunofluorescence (Clarke, 1960; Monath et uL, 1983; Gould et al, 198533).Epitopes that determine the virulence of flaviviruses, equivalent to those identified by Fields and Greene (1982) for reoviruses, have yet to be discovered. Extension of the observations reported above with suitably selected variants from YF virus preparations may make this possible. ACKNOWLEDGMENTS This work was partly supported by a grant from the British Medical Research Council (G821 9412/T) and also a grant from the Wellcome Trust. REFERENCES BROWN,F. (1984).Synthetic viral vaccines. Anna. Rev. MtkTrdkd 38.221235. BUCKLEY,A., and GOULD,E. A. (1985). Neutralisation of yellow fever virus studied using monoclonal and polyclonal antibodies. .I. Gen. I%& 66,2523-2531. CAMMACK,N., and GOULD, E. A. (1986). Antigenic analysis of yellow fever virus glycoproteins: Use of monoclonal antibodies in enzyme-linked immunosorbent assays. J. Viral h&&ho&, in press. CLARKE,D. H. (1969). Antigenic analysis of certain group B arthropod-borne viruses by antibody absorption. J. Exp. Med 111,21-32. CLARKE,D. H., and CASALS,J. (1958). Technique for haemagglutination and haemagglutination inhibition with arthropod borne viruses. Amer. J. Trop. Med Hyg 7.561-573. DELLA-PORTA,A. J., and WESTAWAY,E. G. (1977a). Immune response in rabbits to virion and to nonvirion antigens of the flavivirus Kunjin. h&et Immm 15,874-882. DELLA-PORTA,A. J., and WESTAWAY,E. G. (1977b). A multi-hit model for the neutralisation of animal viruses. J. Gfm V+r& 38, l-19. FIJZLDS,B. N., and GREENE,M. I. (1982). Genetic and molecular mechanisms of viral pathogenesis: Implications for prevention and treatment. Nature (London)
366,19-23.
AND GOULD GOULD,E. A., BUCKLEY,A., BARRETT,A. D. T., and CAMMACK,N. (1986). Neutralising (54K) and nonneutralising (54K and 48K) monoclonal antibodies against structural and non-structural yellow fever virus proteins confer immunity in mice. J. Gen vim!
67,591-595.
GOULD,E. A., BUCKLEY,A., and CAMMACK,N. (1985a). Use of the biotin-streptavidin interaction to improve flavivirus detection by immunofluorescence and ELISA tests. J. Viral. Methods 11,41-48. GOULD,E. A., BUCKLEY,A., CAMMACK,N., BARRETT, A. D. T., CLEGG,J. C. S., ISHAK, R., and VARMA, M. G. R. (1985b). Examination of the immunological relationships between the flaviviruses using yellow fever virus monoclonal antibodies. J. Gen. Viral 66, 1369-1382. GUESDON,J. L., TERNYNCK,T., and AVRAMEAS, S. (1979). The use of avidin-biotin interaction in immunoenzymatic techniques. J. Histoch Q&&em. 27(8), 1131-1139. HAMMON,W. M., and SATHER,G. E. (1956). Immunity of hamsters to West Nile and Murray Valley viruses following immunisation with St. Louis and Japanese B encephalitis viruses. Proc Sot Exp. Bid Med 91, 521-524. HEINZ, F. X., BERGER,R., TUMA, W., and KUNZ, C. (1983). A topological and functional model of epitopes on the structural glycoprotein of tick-borne encephalitis virus defined by monoclonal antibodies. Virology 125.525-537. HENCHAL, E. A., MCCOWAN,J. M., BURKE, D. S., SEGUIN, M. C., and BRANDT, W. E. (1985). Epitope analysis of antigenic determinants on the surface of dengue 2 virions using monoclonal antibodies. Amer. J. Trap. Med Hyg. 34,162-169. KIMURA-KURODA, J., and YASUI, K. (1983). Topographical analysis of antigenic determinants on envelope glycoprotein V3(E) of Japanese encephalitis virus using monoclonal antibodies. J. V&l. 45, 124-132. KINGSFORD,L. (1984). Enhanced neutralisation of La Crosse virus by the binding of specific pairs of monoclonal antibodies to the Gl glycoprotein. viroZ.ogy136.265-273.
LEFRANCOIS,L., and LYLES, D. S. (1982). The interaction of antibody with the major surface glycoprotein of Vesicular Stomatitis virus. II. Monoclonal antibodies to nonneutralising and cross-reactive epitopes of Indiana and New Jersey serotypes. virobgg 121,168-174. MATHEWS,J. H., and ROFZIRIG,J. T. (1984).Elucidation of the topography and determination of the protective epitopes on the E glycoprotein of Saint Louis encephalitis virus by passive transfer with monoclonal antibodies. J. Immunol 132(3), 1533-1537. MONATH, T. P., KINNEY, R. M., SCHLESINGER,J. J., BRANDRISS,M. W., and BRES, P. (1983). Ontogeny
MAPPING
YF VIRUS
ENVELOPE
of yellow fever vaccine: RNA oligonucleotide fingerprint and monoclonal antibody analyses of vaccines produced world-wide. J. Gen Viral 64, 627637. PORTERFIELD,J. S. (1980). Antigenic characteristics and classification of togaviridae. In “The Togaviruses, Biology, Structure, Replication” (R. W. Schlesinger ed.), pp. 13-46. Academic Press, New York. QURESHI,A. A., and TRENT, D. W. (1973). Group B arbovirus structural and non-structual antigens: III. Serological specificity of solubilised intracellular viral proteins. Iqfect. Immun 8.993-999. Pan American Health Organisation. (1984). “Report of the Pan American Health Organisation: Meeting to Develop Guidelines and Protocols for the Production of Yellow Fever Vaccine in Cell Cultures, Washington, D.C. 21-23 February.” ROEHRIG,J. T., MATHEWS,J. H., and TRENT, D. W.
GLYCOPROTEIN
341
(1983). Identification of epitopes on the E glycoprotein of Saint Louis encephalitis virus using monoclonal antibodies. Virology 128,118126. SKEHEL, J. J., BALEY, P. M., BROWN,E. B., MARTIN, S. R., WATERFIELD,M. D., WHITE, J. R., WILSON, I. A., and WILEY, D. C. (1982). Changes in the conformation of the influenza virus haemagglutinin at the pH optimum of virus-mediated membrane fusion. Proc. Natl. Acad Sci USA 79,968-972. TARR, G. C., and HAMMON,W. M. (1974). Cross-protection between group B arboviruses: Resistance in mice to Japanese encephalitis and St. Louis encephalitis viruses induced by dengue virus immunisation. Iqfect. Immun. 9,909-915. WESTAWAY,E. G., BRINTON,M. A., GAIDAMOVICH,S. YA., HORZINEK,M. C., IGARASHI,A., K~~%RI&NEN, L., Lvov, D. K., PORTERFIELD,J. S., RUSSELL,P. K., and TREZT,D. W. (1985). Flaviviridae. InterviroEogy 24,183-192.
150,333-341 (1986)
Topographical Analysis of Epitope Relationships on the Envelope Glycoprotein of Yellow Fever 17D Vaccine and the Wild Type Asibi Parent Virus N. CAMMACK’
AND E. A. GOULD
Arbovims Research Unit, London School of Hygiene and Tropical Medicine, Winches Farm Field Station., 395 Hageld Road, St. A&MS, He& AL& OXQ, England Received October 7, 1985; accepted December 20, 1985 Monoclonal antibodies (MCA), with defined molecular specificity, were used in a competition binding enzyme-linked immunosorbent assay (ELISA) to locate the relative positions of the epitopes on the envelope glycoprotein of yellow fever 17D vaccine virus and its wild type parent virus, Asibi (AS). Five topographically distinct antigenic domains were defined on the E glycoprotein of the 1’i’D vaccine. Three of these (A, B, and C) were represented by one MCA each, a fourth (D) was represented by two MCA, and a fifth domain (E) comprised a major cluster of at least five overlapping epitopes. Asibi virus also possessed domain E which is proposed to be a conserved antigenic region within the envelope glycoprotein of all flaviviruses. Domains A and C were not represented on Asibi virus and one epitope, situated proximal to the E domain, showed structural alterations in physical overlap. Functional activities were assigned to physically mapped epitopes by haemagglutination inhibition (HAI), virus neutralisation (N), and passive protection in mice. The HA1 and N functions were not necessarily linked but only MCA with N activity were able to protect mice passively against lethal infection. All domains demonstrated a heterogeneous range of biological properties dependent upon the virus strain rather than the epitope. 8 1986Academic press, he. INTRODUCTION
Yellow fever (YF) is a mosquito-transmitted virus disease, It is classified as the type species of the family Flaviviridae (Westaway et al., 1985) and has considerable potential to act as a human pathogen despite the current availability of safe and effective vaccines. The yellow fever virus vaccines were developed empirically but it is now felt that a new and improved vaccine strain should be derived (see Report of the Pan American Health Organization, Washington D. C., 1984). The major viral surface component is the envelope glycoprotein which is responsible for the biological properties of haemagglutination (HA) (Della-Porta and Westaway, 19’77a) and neutralisation (N) (Qureshi and Trent, 1973). On the basis of complement fixation ‘To whom requests for reprints should be addressed. 333
haemagglutination inhibition (HAI), N tests, and the natural invertebrate host, a system of classification was derived which incorporated seven antigenic subgroups or complexes (Porterfield, 1980). Cross-protection virus-challenge experiments in animals have also demonstrated the antigenic relationships of the flaviviruses (Hammon and Sather, 1956; Tarr and Hammon, 1974). Moreover, monoclonal antibodies (MCA) specific for the envelope glycoprotein can confer immunity in mice to St. Louis encephalitis virus lethal infection (Mathews and Roehrig, 1984). In order to understand more completely the significance of flavivirus-host interactions and the changes that occur in vaccines as they are derived from parent viruses, it is necessary to know something of the three-dimensional structure of the envelope glycoprotein. Competition binding assays have previously been carried out to determine the relative distribution of ep0042-6822/86 $3.00 Copyright All rights
Q 1996 by Academic Press. Inc. of reproduction in any form reserved.
334
CAMMACK
itopes on the envelope glycoproteins of several flaviviruses (Heinz et aL, 1983, Kimura-Kuroda and Yasui, 1983; Roehrig et al, 1983; Henchal et al, 1985). We recently prepared and reported MCA which display a wide range of antigenic reactivities with all flavivirus subgroups (Gould et a& 1985b). Five antigenic domains were identified on the YF 17D vaccine virus envelope glycoprotein: 17D-specific, vaccine-specific, YF-specific, intermediate reactivity with flaviviruses, and broad reactivity with flaviviruses. We now report a comparison of the topographical maps of the envelope glycoprotein of YF 17D vaccine virus and the wild type Asibi (AS) parent YF virus. The relevant antigenic determinants have also been characterised using HAI, N, and mouse passive protection tests so that functional properties can be assigned to the physical maps.
AND
GOULD
Production of immune mouse asciticfluid Seven to ten days after priming 8-week-old male BALB/c mice with pristane (2,6,10,14tetramethyl pentadecane), 5 X lo6 of the appropriate hybridoma cells were inoculated intraperitoneally. Ascitic fluids were collected, clarified (2000 g for 10 min) and stored at -20”. Pur$cation and biotinylation of hybridomu antibodies. Monoclonal antibodies were clarified by centrifugation first at 2000 g for 10 min followed by 10,000 g for 30 min. Aliquots were dialysed using 0.02 M Tris-HCl, pH 8.0, and centrifuged at 2000 g for 5 min. The proteins were purified by affinity chromatography with DEAE affigel blue (Bio-Rad, England) to obtain the IgG immunoglobulin. Optimal fractions, identified by immunofluorescence titrations with YF-infected cell monolayers, were dialysed with 0.005 M PBS and concentrated by forced dialysis. Protein concentrations were estimated using the MATERIALS AND METHODS Bio-Rad Protein Assay (Bio-Rad) and adViruses and cells. The 1’7D-204 Arilvax justed to 1 mg/ml. The purified immunovaccine Batch BYF 1228 (Wellcome Labo- globulin was biotinylated using biotinylratories, England) and the parent Asibi N-hydroxy succinimide ester (BHSE, (AS) virus, kindly supplied by Dr. R. Fitz- Sigma, England) according to the method george (PI-IL& Porton, Salisbury, England) of Guesdon et aL (1979). The optimal conwere used throughout. Ten percent suck- centration giving biotinylated probes with ling mouse brains suspensions of these vi- strong self-competition in the competition ruses at an estimated input multiplicity of binding assay, varied between 75 and 200 0.1 PFU/cell were used to infect monolay- pg BHSE/mg protein. The reaction was alers of Vero cells maintained at 37’ in Lei- lowed to proceed for 4 hr at room temperbovitz L15 medium containing 2% fetal calf ature, before fractionation by Sephadex serum (FCS). Supernatant medium from chromatography (G-75). Peak fractions infected cells was clarified by centrifugawere dialysed extensively against 0.005 M tion at 2000 g for 10 min and virus was PBS and concentrated by forced dialysis. precipitated overnight at 4” with 7% polyCompetition antibody blocking assays usethylene glycol (PEG) and 0.4 M sodium ing the ELISA test. The standard ELISA chloride. Precipitates were collected by assay, using biotin and streptavidin, was centrifugation at 3000 g for 30 min. Virus described previously (Gould et al, 1985a). concentrates (loo-fold) were resuspended The titer of each antibody was estimated in borate-buffered saline at pH 9.0. as the reciprocal of the highest dilution Monoclonal antibodies. The preparation, showing an absorbance of 0.1 greater than the control. The blocking assay was based designation, isotype, and immunofluorescence characteristics of most of the mono- upon competition between biotinylated and clonal antibodies used in this report were unlabelled MCA for a previously deterantigen concentration. given previously (Gould et a& 1985b). Ad- mined limiting ditional antibodies (MCA 110,126, and 140) Nonspecific binding was reduced to a minare included in this report. These were imum by the addition of PBS-Tween 20 prepared against AS virus using the meth- containing 5% newborn calf serum for 1 hr. Competitor antibodies were added in ods described previously.
MAPPING
YF VIRUS
ENVELOPE
serial lo-fold dilutions and incubated overnight. The plates were then washed and biotinylated MCA was added for 1 hr. Binding was detected using the streptavidin-biotinylated peroxidase (S/P) complex and OPD substrate as described previously. The percentage binding was calculated by comparing the absorbance readings recorded in the presence of nonlabelled MCA with those recorded in the presence of control ascitic fluid (a monoclonal antibody against HBsAg). HAI tests. These assays followed the protocol of Clarke and Casals (1958) except that 24-hr chick erythrocytes were used at pH 6.2. Neutralisation assay. Vero cell monolayers in 5-cm petri dishes were infected with an estimated 50 PFU/dish of YF 1’7Dvirus in either the presence or absence of serial dilutions of heat inactivated (56” for 30 min) MCA. After absorption for 1 hr at 3’7” the monolayers were washed with 0.05 M PBS, pH 7.4, and Leibovitz-L15 medium containing 2% FCS and 1% agarose (Seakem, USA) was added. After incubation at 37” for ‘7days monolayers were fixed with 10% Formalin in saline and stained with naphthalene black. The titer of each antibody was the reciprocal of the highest dilution that reduced plaque numbers by 50%. Mouse passive protectiw (PP) tests. Groups of eight mice, 3-4 weeks old (strain TO, outbred; Tuck and Son, Bettlesbridge, Essex) were inoculated intraperitoneally with 0.1 ml of undiluted monoclonal ascitic fluid. After 24 hr the mice were challenged intracerebrally with 50 LDm per mouse of either YF 17D or AS virus. The antibody was considered positive if at least 50% of the mice survived. Negative control mice were inoculated with a monoclonal antibody against HBsAg. Immunofluorescence (IF) tests. These were performed as described previously (Gould et al, 1985a). RESULTS
Biological Characteristics of MCA A total of 11 MCA were used. Their designation, molecular specificity, isotype,
GLYCOPROTEIN
335
HA1 and N titers, reactivity with YF 17D virus in either indirect immunofluorescence or ELISA tests, and their ability to protect mice in passive antibody transfer experiments are all summarised in Table 1. All MCA reacted specifically with the 54K envelope glycoprotein of YF 17D virus (Gould et aJ., 198513).The results obtained by IF and ELISA tests were identical: MCA 864 and 411 reacted only with 17D virus but the other antibodies reacted equally well with both viruses as shown when antibody binding profiles were compared (see Cammack and Gould, 1986). However, whether or not HAI, N, or PP occurred was independent of antibody binding to the virus. Generally, assignment of biological properties to individual epitopes was dependent on the virus strain. The three tests showed that HAI, N, and PP were not linked; MCA 612 had HA1 but not N activity with AS virus, whereas MCA 427 had N but no HAI activity with YF 17D virus. MCA 411, the vaccine-specific antibody, had no biological activity but its titer by ELISA titration was comparable with other MCA (Table 1). For YF 17D virus, only MCA with N activity also showed PP activity. Moreover, only MCA 110 showed activity in N tests with AS virus and none did in PP tests with AS virus. Competition Blocking Tests Using YF 17D virus Each of the 11 MCA were used in competitive antibody blocking tests with YF 17D virus as described above. Preliminary studies indicated that a PEG-concentrated supernatant virus from virus-infected cells provided the most suitable antigen source. With this type of preparation, antibodies showed binding characteristics equivalent to those obtained with highly purified virions (unpublished results). The antibody binding specificities of all MCA, except 612 and 110, were retained after biotinylation. Thus, where possible, reciprocal competition experiments were performed to account for variation in antibody avidity, since lateral shifts of binding curves do indicate avidity differences. Each biotinylated antibody (designated BMCA, i.e.,
CAMMACK
336
AND GOULD
TABLE 1 CHARAC~ERISATION OFEPITOPESONTHE E GLYCOPROTEIN OFYELLOWFEVER (YF) VIRUS Passive protection Survivors/total
Endpoint (l/log,,) HA1
N
MCA
Subclass
Epitopeb
IF
ELISA
YFl’7D
YFAS
YFlPD
YFAS
YF17D
YFAS
864 825 411 126 140 110 813 843 868 612 427
IgG2A IgG2A IgG2A IgG2A IgG3 IgG3 IgG2A IgG2A IgG2A IgG3 IgG2A
gP54A gp54* gP54c gP54D gP54D gP54E g~54~ gp54’ gP54E gp54” gP54E
YF17D-204 YF-type YF-vaccine Flavi-inter Flavi-inter Flavi-broad Flavi-broad Flavi-broad Flavi-broad Flavi-broad Flavi-inter
6.0 6.0 5.7 5.8 5.0 5.7 5.7 5.7 6.0 5.4 5.7
4.3 1.9
6.6 4.2 cl.0 cl.0 2.2 4.7 3.0 2.8 4.4 1.3 2.6
16/16 12/16 O/16 2/16 4/16 l/16 9/16 l/16 12/16 2/16 4/16
O/16 O/16 O/16 O/16 O/16 O/16 O/16 O/16 O/16 O/16 O/16
‘Endpoint ELISA values represent relative avidities of MCA for purified YFlPD virus (2 gg/well). Haemagglutination inhibition (HAI) was performed using 4-8 HA units at pH 6.2.Neutralisation (N) was performed on Vero cells with 50 PFU/test. All results are expressed as l/log,,, endpoint antibody dilution. bThe letters A, B, C, D, and E represent the domains on the glycoprotein referred to in the text. ’ Immunofluorescence reactivities with flaviviruses as defined in the Introduction.
B825) was titrated to find an optimal concentration on the plateau of maximum binding. The conjugates were then challenged against a range of dilutions of unlabelled homologous or heterologous MCA. Binding of BMCA was inhibited by 70100% in the presence of excess unlabelled homologous MCA. The results are presented in Fig. 1. For YF 17D virus, MCA 411 showed no reciprocal or nonreciprocal competition with any other MCA (domain C); B864 was blocked nonreciprocally by MCA 825, 843, 813, and 427, whereas B825 was nonreciprocally blocked just by MCA 843 and 612. Only reciprocal and not nonreciprocal blocking was taken to indicate overlapping sites, therefore MCA 864 and 825 were each assigned to separate and distinct antigenic domains (A and B, respectively). On the other hand, MCA 813, 843, and 868 all showed reciprocal blocking with each other and, together with 612 and 110, which had shown one-way blocking, were placed together as one domain (E). Since MCA 427 was reciprocally blocked by 868 and 813 it was considered to be proximal to domain E. Finally, MCA 126 and 140 overlapped to
form another distinct domain (D) although both were nonreciprocally blocked by MCA 110 and were able to block (nonreciprocally) MCA 868. Competition Blocking Tests Using YF As&i Virus YF AS virus was used in a competition binding ELISA with each of the 11 MCA described above. There was remarkable similarity (Fig. 2) with each of the five overlapping antibodies represented by domain E for YF 17D virus. Furthermore, MCA 427 still retained its proximity to this major cluster of epitopes but overlapped with MCA 868 and 843 instead of 868 and 813. A further difference in the relative positions of the epitopes represented on the two viruses was seen with MCA 110 and 140 which did not nonreciprocally block MCA 868 as they had previously been shown to do with YF 17D virus. Assignment of Biological Functions to Maps Figures 3a and 3b show our attempts to construct maps of the epitopes based on their relative physical positions in the en-
MAPPING
YF VIRUS
ENVELOPE
80
60
40
20
loo
0
s E B 0
337
ruses, respectively. As can be seen, with YF 17D virus, different biological functions were represented in all domains. Thus, within an antigenic domain the biological properties of individual epitopes were not necessarily identical. MCA in domains A and B showed functional activity in all three biological tests, whereas MCA 411 in domain C did not react in any test. All five broadly cross-reactive MCA in domain E reacted in HA1 and N tests but only two (MCA 813 and 868) showed PP activity. Neither MCA 140 nor 126, in domain D, protected mice but MCA 140 had HA1 and N activity whereas MCA 126 had no N activity. In view of the differences in biological activity between YF 17D and AS virus it is not surprising that the functional maps appeared so different. Only domains B, D, and E were represented on the physical
1w
ii z 5
GLYCOPROTEIN
60
60
40
20
100
60
60
40
20
100
Ii
I I
-2
I -3
I -4
III -5
I -1 LOGlO
-2
1 -3 MC/\
I -4
-6
-*
-2
-3
-4
-5
60
DILUTION 60
FIG. 1. Competitive blocking ELISA with biotinlabelled MCA (BMCA) directed against the YF 17D E glycoprotein. Serial lo-fold dilutions of unlabelled MCA were allowed to react with YF antigen-coated microtitre plates. Unbound MCA were removed and the binding of a previously determined limiting concentration of biotin-labelled MCA was recorded. The extent of blocking by unlabelled MCA was calculated as described in the text and is expressed as percentage binding BMCA. In each case only the data for blocking MCA are presented, since all others showed 100% binding. 0,864; 0,825; A, 427; A, 612; Cl, 868; n , 843; 0,411; x, 813; 0,110; 0,126; v, 140.
40
20
loo
velope glycoproteins of YF 17D and AS viruses, respectively. Overlapping closed circles represent overlapping epitopes or epitopes thought to be in very close proximity. The open circles for 612 and 110 are proposed sites based on one-way blocking experiments. Figures 4a and b have been derived by analysis of the hypothetical physical maps and the data presented in Table 1. These figures show our attempts to assign functional a&Vi&g to the maps Of the envelope glycoproteins of both YF 17D and AS vi-
80
60
40
20
LOGlo
MCA
DILUTION
FIG. 2. Competitive blocking ELISA with biotinlabelled MCA (BMCA) directed against the YF AS E glycoprotein. Assays were performed as described in the legend for Fig. 1.
CAMMACK
333
FIG. 3. Physical model for the antigenic domains on the E glycoprotein of (a) YF 17D vaccine virus and (b) YF AS wild type parent virus. Overlapping closed circles represent reciprocal blocking between MCA. The open circles for MCA 110 and 612 are proposed sites based on one-way blocking experiments. Domains are labeled A, B, C, D, and E.
map and in contrast to the YF 17D, domain B showed no biological activity. Furthermore, none of the MCA in domains D and E had PP activity and only MCA 110 (domain E) neutralised AS virus infectivity. However, both MCA in domain D and three MCA (110, 813, and 612) in domain E did demonstrate HA1 activity. DISCUSSION
Knowledge of the topographical distribution of protective epitopes on the flavivirus envelope glycoprotein may have important implications for the future development of synthetic vaccines. Moreover, the arrangement of biologically active determinants may influence the outcome of infection, in other words, determine the pathogenesis of the virus.
AND GOULD
In this report we have attempted to combine the physical data from competition blocking assays with those from functional biological tests. In experiments with mice we have recently found that the average survival times after intracerebral challenge are longer (12.25 days) for the 1’7D virus than for AS virus (7.65 days) (Gould et aL, 1986). Thus, by analysing and comparing the envelope glycoprotein of the 17D vaccine derivative of YF virus with the parent wild type strain it was hoped that markers of virus virulence might be identified. Five topologically distinct domains were located on the 17D envelope glycoprotein. In many instances the overlap of epitopes was defined by complete reciprocal competition between antibodies with differing functional reactivity patterns. Until recently, this may not have seemed possible since neutralisation of virus infectivity, for example, was considered to be dependent upon antibody attachment to a critical area of the viral surface (Della-Porta and Westaway, 1977b). It is now known, however, that a critical site for a monoclonal antibody can determine whether or not neutralisation occurs (Buckley and Gould, 1985). Thus, the mere presence of an epitope in the glycoprotein does not necessarily lead to neutralisation. Only one of the five domains had no functional reactivity (represented by MCA 411) in the biological tests described; possibly other epitopes in this domain will prove to be biologically active, or alternatively it may represent an area within the envelope glycoprotein that serves a structural role rather than a direct biological role. MCA 864 was blocked nonreciprocally by several antibodies but did not itself block any antibodies. Affinity differences did not account for this but structural alterations induced by antibody binding at a distant site (LeFrancois and Lyles, 1982) could lead to the epitope for MCA 864 being hidden. Nevertheless, MCA 864 defines a very important functional site on YF 17D virus as exemplified by the very high titers found in all tests. Indeed, its absence on all viruses except 17D vaccine strains might be indirectly relevant to the virulence of the YF viruses. By analogy with the concept of Brown (1984) epitopes nec-
MAPPING
YF VIRUS ENVELOPE
GLYCOPROTEIN
339
(8)
A
0 0 664
B
825
(0) I
(b)
HAI
N 0
III
PP
FIG. 4. Assignment of functional activities to the physical maps of (a) YF 17D vaccine virus and (b) YF AS wild type parent virus based on HAI, N, and passive mouse protection tests. [Refer to key.] Domains are labeled A, B, C, D, and E.
essarily involved in pathogenesis are probably more stable throughout the group and perhaps less antigenic. Five epitopes defined a major antigenic domain of the YF 1’7DE glycoprotein which exist as a cluster of overlapping sites. This domain was identically conserved on the E glycoprotein of the wild type AS parent YF virus and we therefore suggest this might be the case for all flaviviruses. An important physical difference between the 1’7D vaccine virus and wild type parent AS virus was the location of the epitope for MCA 427. On the 17D virus E glycoprotein this epitope was situated proximal to the major flavivirus domain, overlapping epitopes for MCA 868 and 813. However, on the AS virus E glycoprotein the epitope for MCA 427 overlapped MCA 868 and 843 but not 813, suggesting that
alterations in the structural configuration of the E glycoprotein had occurred. Similar physical maps were obtained for both viruses and biological activities were assigned to five domains. Nevertheless, in general for both viruses, biological properties were associated only with epitopes and not with domains. In accordance with the findings of Kimura-Kuroda and Yasui (1983) for Japanese encephalitis virus, there was no evidence of linkage between HA1 and N activity. Varying extents of HAI activity for individual epitopes may reflect differences in optimal pH activity. Indeed, altering the pH for HA1 may simply reorientate the epitope (Skehel et aL, 1982). At a given pH, low relative levels of HA1 activity may result because epitopes are in a conformationally inappropriate position. Only minor steric hindrance ef-
CAMMACK
340
fects on the cell membrane binding site would then result (Kingsford, 1984). In this comparison of the antigenic structures of YF 17D and AS viruses we detected physical similarities but major functional differences. Both serological and structural differences have previously been found between these viruses by HAI, N, oligonucleotide RNA fingerprint analysis, and indirect immunofluorescence (Clarke, 1960; Monath et uL, 1983; Gould et al, 198533).Epitopes that determine the virulence of flaviviruses, equivalent to those identified by Fields and Greene (1982) for reoviruses, have yet to be discovered. Extension of the observations reported above with suitably selected variants from YF virus preparations may make this possible. ACKNOWLEDGMENTS This work was partly supported by a grant from the British Medical Research Council (G821 9412/T) and also a grant from the Wellcome Trust. REFERENCES BROWN,F. (1984).Synthetic viral vaccines. Anna. Rev. MtkTrdkd 38.221235. BUCKLEY,A., and GOULD,E. A. (1985). Neutralisation of yellow fever virus studied using monoclonal and polyclonal antibodies. .I. Gen. I%& 66,2523-2531. CAMMACK,N., and GOULD, E. A. (1986). Antigenic analysis of yellow fever virus glycoproteins: Use of monoclonal antibodies in enzyme-linked immunosorbent assays. J. Viral h&&ho&, in press. CLARKE,D. H. (1969). Antigenic analysis of certain group B arthropod-borne viruses by antibody absorption. J. Exp. Med 111,21-32. CLARKE,D. H., and CASALS,J. (1958). Technique for haemagglutination and haemagglutination inhibition with arthropod borne viruses. Amer. J. Trop. Med Hyg 7.561-573. DELLA-PORTA,A. J., and WESTAWAY,E. G. (1977a). Immune response in rabbits to virion and to nonvirion antigens of the flavivirus Kunjin. h&et Immm 15,874-882. DELLA-PORTA,A. J., and WESTAWAY,E. G. (1977b). A multi-hit model for the neutralisation of animal viruses. J. Gfm V+r& 38, l-19. FIJZLDS,B. N., and GREENE,M. I. (1982). Genetic and molecular mechanisms of viral pathogenesis: Implications for prevention and treatment. Nature (London)
366,19-23.
AND GOULD GOULD,E. A., BUCKLEY,A., BARRETT,A. D. T., and CAMMACK,N. (1986). Neutralising (54K) and nonneutralising (54K and 48K) monoclonal antibodies against structural and non-structural yellow fever virus proteins confer immunity in mice. J. Gen vim!
67,591-595.
GOULD,E. A., BUCKLEY,A., and CAMMACK,N. (1985a). Use of the biotin-streptavidin interaction to improve flavivirus detection by immunofluorescence and ELISA tests. J. Viral. Methods 11,41-48. GOULD,E. A., BUCKLEY,A., CAMMACK,N., BARRETT, A. D. T., CLEGG,J. C. S., ISHAK, R., and VARMA, M. G. R. (1985b). Examination of the immunological relationships between the flaviviruses using yellow fever virus monoclonal antibodies. J. Gen. Viral 66, 1369-1382. GUESDON,J. L., TERNYNCK,T., and AVRAMEAS, S. (1979). The use of avidin-biotin interaction in immunoenzymatic techniques. J. Histoch Q&&em. 27(8), 1131-1139. HAMMON,W. M., and SATHER,G. E. (1956). Immunity of hamsters to West Nile and Murray Valley viruses following immunisation with St. Louis and Japanese B encephalitis viruses. Proc Sot Exp. Bid Med 91, 521-524. HEINZ, F. X., BERGER,R., TUMA, W., and KUNZ, C. (1983). A topological and functional model of epitopes on the structural glycoprotein of tick-borne encephalitis virus defined by monoclonal antibodies. Virology 125.525-537. HENCHAL, E. A., MCCOWAN,J. M., BURKE, D. S., SEGUIN, M. C., and BRANDT, W. E. (1985). Epitope analysis of antigenic determinants on the surface of dengue 2 virions using monoclonal antibodies. Amer. J. Trap. Med Hyg. 34,162-169. KIMURA-KURODA, J., and YASUI, K. (1983). Topographical analysis of antigenic determinants on envelope glycoprotein V3(E) of Japanese encephalitis virus using monoclonal antibodies. J. V&l. 45, 124-132. KINGSFORD,L. (1984). Enhanced neutralisation of La Crosse virus by the binding of specific pairs of monoclonal antibodies to the Gl glycoprotein. viroZ.ogy136.265-273.
LEFRANCOIS,L., and LYLES, D. S. (1982). The interaction of antibody with the major surface glycoprotein of Vesicular Stomatitis virus. II. Monoclonal antibodies to nonneutralising and cross-reactive epitopes of Indiana and New Jersey serotypes. virobgg 121,168-174. MATHEWS,J. H., and ROFZIRIG,J. T. (1984).Elucidation of the topography and determination of the protective epitopes on the E glycoprotein of Saint Louis encephalitis virus by passive transfer with monoclonal antibodies. J. Immunol 132(3), 1533-1537. MONATH, T. P., KINNEY, R. M., SCHLESINGER,J. J., BRANDRISS,M. W., and BRES, P. (1983). Ontogeny
MAPPING
YF VIRUS
ENVELOPE
of yellow fever vaccine: RNA oligonucleotide fingerprint and monoclonal antibody analyses of vaccines produced world-wide. J. Gen Viral 64, 627637. PORTERFIELD,J. S. (1980). Antigenic characteristics and classification of togaviridae. In “The Togaviruses, Biology, Structure, Replication” (R. W. Schlesinger ed.), pp. 13-46. Academic Press, New York. QURESHI,A. A., and TRENT, D. W. (1973). Group B arbovirus structural and non-structual antigens: III. Serological specificity of solubilised intracellular viral proteins. Iqfect. Immun 8.993-999. Pan American Health Organisation. (1984). “Report of the Pan American Health Organisation: Meeting to Develop Guidelines and Protocols for the Production of Yellow Fever Vaccine in Cell Cultures, Washington, D.C. 21-23 February.” ROEHRIG,J. T., MATHEWS,J. H., and TRENT, D. W.
GLYCOPROTEIN
341
(1983). Identification of epitopes on the E glycoprotein of Saint Louis encephalitis virus using monoclonal antibodies. Virology 128,118126. SKEHEL, J. J., BALEY, P. M., BROWN,E. B., MARTIN, S. R., WATERFIELD,M. D., WHITE, J. R., WILSON, I. A., and WILEY, D. C. (1982). Changes in the conformation of the influenza virus haemagglutinin at the pH optimum of virus-mediated membrane fusion. Proc. Natl. Acad Sci USA 79,968-972. TARR, G. C., and HAMMON,W. M. (1974). Cross-protection between group B arboviruses: Resistance in mice to Japanese encephalitis and St. Louis encephalitis viruses induced by dengue virus immunisation. Iqfect. Immun. 9,909-915. WESTAWAY,E. G., BRINTON,M. A., GAIDAMOVICH,S. YA., HORZINEK,M. C., IGARASHI,A., K~~%RI&NEN, L., Lvov, D. K., PORTERFIELD,J. S., RUSSELL,P. K., and TREZT,D. W. (1985). Flaviviridae. InterviroEogy 24,183-192.