- Email: [email protected]
78 (H1N1) and its escape variants
Journal of Virological Methods, 42 (1993) 75-88 © 1993 Elsevier Science Publishers B.V. / All rights reserved / 0166-0934/93/$06.00
75
VIRMET 01461
Comparative investigation of the hemagglutinin epitopes of influenza virus A/Brazil/11/78 (H 1N 1) and its escape variants J. Drescher and W. Verhagen Institute of Virology, Medical School of Hannover, Hannover, (Germany) (Accepted 12 October 1992)
Summary A method described previously for determining the concentration of influenza virus antihemagglutinin antibody molecules, the number of epitopes per virus particle and the equilibrium constant of virus antibody interaction was adapted to the use with escape variants (EVs), produced by multiplication of influenza virus A/Brazil (H1N1) in presence of monoclonal antibody directed to each of the four hemagglutinin sites (Sa, Sb, Ca and Cb). The EVs were found to possess an altered antigenic site, which was both antigenic and immunogenic. By use of selected EVs and antibody preparations, the number of epitopes per antigenic site was determined and it was found that each of the four sites was represented by about 390 epitopes per virus particle, suggesting that each of the about 400 hemagglutinin spikes per virion possessed one epitope of the specificity Sa, Sb, Ca and Cb. Alteration of site Sa but not of site Ca increased the avidity of antibody to react with the unchanged sites.
Introduction The hemagglutinin (HA) of influenza virus possesses different antigenic sites (Webster and Laver, 1980; Breschkin et al., 1981; Jackson et al., 1982; Drescher et al., 1987). Two antigenic sites with strain specificity, designated Sa and Sb, and two sites common to antigenically related strains, referred to as Ca and Cb, have been defined operationally for H1 hemagglutinin (Gerhard et al., 1981; Correspondence to: J. Drescher, Institute of Virology, Medical School of Hannover, 3000 HannoverKleefeld, Konstanty-Gutschow Str. 8, Germany.
76
Caton et al., 1982), where Ca has been further divided into the subsites Ca 1 and Ca 2 (Caton et al., 1982). By clustering within the three-dimensional HA configuration of amino acid substitutions of variants selected by monoclonal antibody, 5 antigenic sites, designated A through E, have been proposed for H3 hemagglutinin (strain A / H o n g Kong/1/68)(Wiley et al., 1981, Wiley and Skehel, 1987). The H1 and H3 sites are at least partially equivalent (e.g., the H1 site Sb and the H3 site B) (Caton et al., 1982; Yewdell et al., 1983). Multiplication of influenza virus in presence of monoclonal antibody directed against one site can result in the selection of escape variants (EVs) which are present in the 'parent virus (PV) at a frequency of about 10-~ and do not combine with the antibody used for their selection (Yewdell et al., 1979; Webster and Laver, 1980; Webster et al., 1983). This is interpreted to reflect the loss of one antigenic site of EVs. Whether the lost site is regularly replaced by a new site not present on PV is unknown. For EVs of the H3N2 influenza virus strain A/Memphis/I/71, it was found that adsorption of sera raised against some EVs onto PV removed the HI activity to the EVs completely, while antibodies raised to other EVs were incompletely adsorbed on PV, suggesting that they possessed a new antigenic site (Webster et al., 1983). The experiments described in this paper were designed to examine comparatively the hemagglutinin epitopes of the H1N1 influenza virus strain A/Brazil/11/78 and its EVs. More specifically, the following points were examined: (i) By cross-adsorption of EV antibody onto PV and vice versa, it was established that EVs possessed new antigenic sites (subsequently referred to as altered antigenic site). (ii) The number of epitopes per antigenic site has been determined previously (Drescher et al., 1987) in terms of the number of epitopes per virus particle recognized on A/Brazil virus by monoclonal antibody, directed to each of the four hemagglutinin sites. However, the sum of the number of epitopes found exceeded by far the total number of epitopes present per virus particle. This result was interpreted to reflect cross-reaction of antibody with heterologous sites. In order to obtain more accurate values of the number of epitopes per antigenic site, we determined the s values recognized by anti-PV and anti-EV antibody on PV and vice versa. Thereby, the number of epitopes per antigenic site is obtained as difference of the respective s values found. Since this difference is independent from cross-reactions, more accurate estimates of the number of epitopes per antigenic site can be obtained by this approach. (iii) Furthermore, the alteration of one antigenic site on EVs was examined to determine whether this influences the avidity of reaction of antibody with the unaltered sites.
77 Material and Methods
Virus The egg-adaPted influenza virus HIN1 strain A/Brazil/11/78 was used. EVs were produced by means of three egg-passages in the presence of an excess of monoclonal antihemagglutinin antibody, directed to one site of A/Brazil virus, and two subsequent dilution limit passages (Webster and Laver, 1980). The EVs were earmarked by the symbol 'EV' followed by the antigenic site against which the antibody used for selection was directed. As example, EV-Sa refers to an EV selected by monoclonal antibody directed to site Sa. Furthermore, the reassortant A/Aichi/2/68 (H3) -A/Bel/42/41 (NI) was used. Virus was purified by means of sucrose gradient ultracentrifugation (Laver, 1960). The HA titers were determined by use of HA pattern test (Palmer et al., 1975) and by means of the photometric HCU (hemagglutinin concentration unit) method (Drescher et al., 1962). The concentration of virus particles per ml was enumerated by means of electron microscopy as described previously (Stachan and Drescher, 1987). The S.E. of the values found did not exceed 5.2%.
Antibody Antisera against A/Brazil virus (anti-PV) and the escape variants (anti-EV) were raised in chickens and goats as described previously (Drescher et al., 1987). Antibody preparations were heated to 56°C for 30 min. and were pretreated with M/90 KJO4 as described by Dowdle et al. (1979). The elimination of inhibitors was controlled by testing K104-treated prevaccination sera of animals against the virus strains employed. No HI activity was found. Antineuraminidase antibody was removed by adsorption onto the recombinant A/Aichi/2/68 (H3)-A/Bel/42 (N1), possessing irrelevant hemagglutinin and relevant neuraminidase. Antisera were tested for IgM and IgG antibody against homologous virus by use of sucrose gradient ultracentrifugation (Cremer and Riggs, 1979) and antibody of the IgG class was used for further experiments, only. Monoclonal A/Brazil antihemagglutinin antibody was produced and tested as described (Drescher et al., 1987). With our set of monoclonals, only 4 sites (Sa, Sb, Ca and Cb) were identified on A/Brazil virus and no division of site Ca into the subsites Cal and Ca2 as described for A/PR8/34 (H1N1) virus (Caton et al., 1982) was possible.
Isolation of antibody EV antisera were adsorbed onto an excess of PV and vice versa. An excess of virus was considered to be represented by the 10-fold virus dose, sufficient to adsorb all HI activity against the virus used for raising of antisera.
78 The resulting virus antibody complexes were separated from unadsorbed antibody by ultracentrifugation. They were subsequently split at pH 3.0, separated from virus by ultracentrifugation (Laver et al., 1974) and tested for HI activity against PV and EV.
Antibody titration Antibody titrations were carried out both by means of HI pattern test (Palmer et al., 1975) and by the use of a photometric HI test (Drescher, 1972). Photometrically determined titers were expressed as the reciprocal antibody dilution (ds0) yielding 50% hemagglutination inhibition.
Determ&ation of s and K The number of epitopes per virus particle (s) recognized by antibody and the equilibrium constant K were determined as described previously (Drescher et al., 1990). Briefly, antibody preparations were tested by equilibrium filtration (Fazekas and Webster, 1961), using graded dilutions (l/d) of a virus suspension containing V0 virus particles per ml. The slope (mv) and intercept (iF) of the straight lines obtained when plotting (l-a) versus (1-a)/(a. d) were determined, where a is the fraction of bound antibody. In addition, the antibody preparations where tested for the antibody dilution (1/ds0) yielding 50% hemagglutination inhibition in the photometric HI test (Drescher, 1972), using a virus concentration of Vo/dv, p. From these data, s was calculated as follows: s = (39.2. my-dso)l(dv,p" d v . as0)
(1)
dr is the reciprocal antibody dilution employed in equilibrium filtration and as0 the fraction of antibody bound at the dilution 1/d5o in the photometric HI test. as0 was calculated as follows: as0 = 0.5 (mp + ip + 1) -- x/0.5 2 (mp + ip + 1)2 -- mp where mp
=
(mF"
dso)/(dv,p
dr)
and ip
=
-
iv • dsoldv.
Then, the equilibrium constant K was obtained as follows:
(2)
79 K = (S • V o • i p ) / ( m p
•
dv,p)
(3)
In addition, the relative avidity of the reaction of a given antibody preparation with PV and EV was determined in terms of the HCU dose of virus yielding binding of 75% of HI activity (HAo.75), using the same starting dilution of antibody. Values of HAo.T5 were determined by equilibrium filtration.
Determination of the number of antibody molecules adsorbed per virus particle at the dilution (1/d5o) yielding 50% hemagglutination inhibition in the photometric HI t e s t ( A b , 5 o / V ) Values of A b , 5 0 / v were determined as described previously (Drescher et al., 1990). Briefly, antibody preparations were tested chemically for the concentration of antibody molecules per ml (= A) as described (Drescher et al., 1990) and were tested against EVs by equilibrium filtration (mv and iv) and by the photometric HI test (ds0) as described above. Values of mv, iF, ds0 and dv,p were used for calculating the fraction of antibody bound at the dilution 1/ds0 (= as0). Then, Ab,5o/V was obtained as follows: Ab,50/v
=
(A
• aso)/(dso
" v)
(4)
where v is the concentration of virus particles used in the photometric HI test.
Statistical analysis The standard errors (SE) of s and K values were determined as described previously (Drescher et al., 1990). The mean values of s, K and HAo.75 were tested for significant differences by means of a two-tailed t test (NCSS program), since the values met the normality test criteria.
Results
Patterns of adsorption of antibody onto virus Antisera raised against PV and EVs were adsorbed onto an excess of PV and of EVs. The completeness of antibody adsorption was tested by measuring the HI titers of supernatants against the viruses employed. Representative examples of the results obtained are listed in Table 1. Note that the anti-EV antibody were incompletely adsorbed onto PV and vice versa. These findings indicate that PV and EVs possessed a different antigenic site and that the altered site on EVs was both antigenic and immunogenic.
80 TABLE 1 Example for the patterns of adsorption of antibody onto parent virus and escape variants Antibody raised against virus
Antibody adsorbed onto excess of virus d
HI titers against virus PV
EV-Ca
EV-Sa
pv a
nil PV EV-Ca EV-Sa
2048 <4 128 256
1024 <4 54 128
1024 <4 64 <4
EV-Ca b
nil PV EV-Ca EV-Sa
1024 <4 <4 128
2048 128 <4 256
2048 <4 <4 <4
EV-Sa c
nil PV EV-Ca EV-Sa
2048 <4 128 <4
4096 <4 <4 <4
4096 128 256 <4
apv, parent virus A/Brazil. bEV-Ca, escape variant selected by anti-Ca antibody. CEV-Sa, escape variant selected by anti-Sa antibody. OThe virus dose employed was 100000 HCU/ml corresponding to 5 x l012 virus particles per ml. Since 1 virus particle has a weight of 5.25 x 10- 16 g (Reimer et al., 1966) and consists to 70% of protein (Ada and Perry, 1954), the virus dose employed contained 1.83 mg of viral protein.
Enumeration of the number of epitopes per antigenic site The method used for the determination of the number of epitopes per virus particle (s) recognized by antibody is based on the finding that there are on an average 39.2 antibody molecules adsorbed per virus particle at the antibody dilution (1/ds0), yielding 50% hemagglutination inhibition in the photometric HI test (Drescher et al., 1990). This was tested to establish whether it is also true when allowing antibody to react with EVs. Therefore, this value (Ab,50/V) was also determined using EVs. Representative examples of the results obtained are given in Table 2. Note that the values of Ab,5o/V obtained did not differ significantly from the value of 39.2 determined previously. These results indicate that the technique can be used also for measuring s of EVs. Antisera directed against PV and EVs were tested for the number of epitopes per virus particle (s) they did recognize on PV and EVs. Representative examples of the results obtained are shown in Table 3. In experimental group 1, antisera were allowed to react with the virus against which they were raised. Note that antisera in this group yielded s values not significantly differing one from another (average value 1561). This result suggested t h a t the altered site of EVs (e.g. site Sa) did not differ in s from the respective antigenic site of PV. Experimental groups 2 through 4 pertain to the reaction of anti-PV antibody with EV and vice versa. These reactions measure the number of epitopes of the
81
--
c5
-H
H-I
0
.
~
.
.
o
~
~
"a 0 . . . .
~
"d ¢,2 • .,-
0 0 0 o o
II
o 0 0 0 0 0 o
,.a
0
ok
~-~
o,
ol
O~
,,.o
"d 0 X
o
X ~o
0 ",~
::
o
~
o-n
I:= ,"a
UJ
¢'I
,~.~
~.~
' ~
¢~
.~.~
.~
0
~
82
0 ,.o
M •
A
0
U
3 e.~
++~+ +_~++ ..o ;.,, .o ..o +ca~, + +
++ ~ + +
++ ,.~ ..o + +
++ .,o~ + +
++ ,.o ..o ta~ ta~
"r. e~
E c~c~
0
<
i
e~
~
...(3
c~
~>.~
c)
83
r~l~
r~J~
r~ ~
÷
÷
~
It- ÷
~r~ l
~tI
Jr ÷
÷I~
7L÷
o
<
84
three antigen!c sites indicated. Note that an average value of s of 1181 was obtained. Group 5 comprised data obtained by allowing antibody directed to one EV to react with a different EV and relates to the number of epitopes of two antigenic sites (Sb and Cb), yielding an average value of 822. These data were used (see Table 4) to calculate the number of epitopes per antigenic site. As an example, the s value of Sa was calculated by subtracting the s value obtained for sites Sb, Ca and Cb from the respective value recorded for all four sites (Sa, Sb, Ca and Cb). The number of epitopes per antigenic site (average value 390) did not differ significantly one from each other.
Comparison of the avidity of epitope-paratope interaction for parent virus and the escape variants EV-Sa and EV-Ca Studies were carried out to determine whether the change of one antigenic site resulting in the appearance of an escape variant alters the avidity of reaction of antibody with the unaltered sites. For this purpose, monoclonal antibody directed to site Ca was allowed to react with parent virus and with the escape variant EV-Sa. The avidity of antigen-antibody interaction was recorded in terms of the HCU doses required to bind 75% of the initial HI activity (HA0.75) and in terms of the equilibrium constant K. Representative examples of the results obtained are listed in Table 5. Note that the anti-Ca antibody had significantly higher avidity (i.e. lower HA0.v5 values and K values) for the reaction with site Ca of the escape variant than with parent virus. This result indicates that the alteration of site Sa increased the avidity of reaction of antibody with site Ca. The same conclusion was reached when testing polyclonal antibody, recognizing the sites Sb, Ca and Cb, in like manner (lower part of Table 5). The antibodies were obtained by absorption onto and elution from the escape variant EV-Sa of anti-PV antisera and vice versa. Again, the antibody had higher avidity for their reaction with EV-SA than with PV. By contrast, significant differences of avidity were not found when comparing the reaction with EV-Ca and PV of antibody directed against the unaltered sites of PV-Ca (data not shown).
Discussion
The results presented in this paper indicate that the escape variants (EVs) examined each possessed an altered antigenic site which was antigenic and immunogenic. This conclusion is based on the finding that anti EV-antibody recognized all epitopes on the homologous EV, indicating that the altered site combined with antibody, and that adsorption of EV-antisera onto PV did not completely remove the HI activity against EV. The analogous result was obtained when adsorbing anti-PV antibody onto EV.
85 <
0 .,.0
I=
~14q
~.~.
,q.o.
.=_
.o
e4~
e4~
X
n~
:> ,,.., r',l
~
X
v
v
tm i.
I"---
0
~
I"xl
o
o~
.~. ~
o ° o°
t"q~
6e.,
Ob 0..~
.~. >~
0
>~,
~.
0 ",~ " ~
>~, II o "~
86
The number of epitopes per virus particle recognized by each of monoclonal antibodies directed to antigenic sites Sa through Cb has been previously determined (Drescher et al., 1987). The sum of the epitopes recognized by each antibody exceeded by far the total number of epitopes present per virus particle. This has been interpreted to reflect the potential of cross-reaction of antibody with antigenic sites or to result from the reaction of antibody with overlapping sites. As example, monoclonal anti-Ca antibody will not only react with the Ca epitopes but will also cross-react with epitopes of site Cb, resulting in an overestimation of the number of epitopes of site Ca. In order to obtain more accurate values of the number of epitopes per antigenic site, we determined the total number of epitopes per virus particle, recognized by anti-PV antibody and anti-EV-antibody on PV and vice versa. The reaction of anti-PV with PV measures the total number of epitopes per virus particle and the reaction of anti-EV with PV the number of epitopes per virus particle with the exception of the epitopes, belonging to the antigenic site not present on the EV used. The number of epitopes of the respective antigenic site is obtained by subtraction. Since the antibody directed to the altered antigenic site of EV does not cross-react with PV, these values are not influenced by cross-reaction of antibody. By use of this strategy, the number of epitopes per antigenic site was determined. The sum of the number of epitopes per antigenic site agreed well with the total number of epitopes present per virus particle, indicating that reliable estimates were obtained. On an average, each site comprised 390 epitopes. Since an influenza virion possesses about 400 hemagglutinin spikes (Laver, 1973) this finding indicates that each hemagglutinin spike possessed one epitope of each specificity (Sa, Sb, Ca and Cb). The avidity of epitope-paratope interaction was comparatively tested for EV-Sa, EV-Ca and PV. Alteration of site Sa (EV-Sa), but not of site Ca (EVCa), enhanced the avidity of reaction of antibody directed against unaltered sites, as indicated by the finding that anti-Ca antibody had a higher avidity against site Ca of EV-Sa than against site Ca of PV. This result indicates that alteration of one antigenic site can alter the avidity of reaction of unchanged sites with antibody. This could be interpreted to result from conformational changes. As an alternate explanation, it could be assumed that antibodies combine with lower avidity with heterologous sites than with the homologous site of PV. If the cross-reacting site is absent on EV, the measured avidity of the reaction of antibodies with E-V is higher than that recorded for their reaction with PV. At present, it is not known whether this interpretation is correct.
Acknowledgements
The authors acknowledge with much appreciation the excellent technical assistance of Mrs. R. Assmann, Mrs. D. Findeisen-Weidauer, Mrs. G. Harste,
87
Mrs. J. Milzer, Mrs. K. Stubbendorff, Mrs. A. Thiemer, and the competent help of Mrs. A. Meyer and Mrs. B. Thierkopf in preparing this manuscript. References Ada, G.L. and Perry, B.T. (1954) The nucleic acid content of influenza virus. Aust. J. Exp. Biol. Med. Soc. 32, 453-468. Breschkin, A.M., Ahern, J. and White, D.O. (1981) Antigenic determinants of influenza virus hemagglutinin.. III. Topography of the antigenic regions of influenza virus hemagglutinin determined by competitive radioimmunoassay with monoclonal antibodies. Virology 113, 130140. Caton, A.J., Brownlee, G.G., Yewdell, J.W. and Gerhard, H. (1982) The antigenic structure of the influenza virus A/PR/8/34 hemagglutinin (HI subtype). Cell 31,417-427. Cremer, N.E. and Riggs, J.L. (1979) In: D.H. Lennette and N.J. Schmidt (Eds.), Immunoglobulin Classes and Viral Diagnosis, 5th ed., American Public Health Association, Washington, D.C., pp. 191 208. Dowdle, W.A., Kendal, A.P. and Noble, G.R. (1979) Influenza viruses. In: E.H. Lennette and N.J. Schmidt (Eds.), Diagnostic Procedures for Viral, Rickettsial and Chlamydial Infections, 5th Ed., American Public Health Association, Washington, D.C. pp. 585 609. Drescher, J., Hennessy, A.V. and Davenport, F.M. (1962) Photometric methods for the measurement of hemagglutinating viruses and antibody. 1. Further experience with a novel photometric method for measuring hemagglutinin. J. lmmunol. 89, 794-804. Drescher, J. (1972) Patterns of cross-reaction of selected influenza virus A 2 strains isolated from 1957 to 1968 as determined by use of a photometric hemagglutination inhibition test. Am. J. Epidemiol. 95, 549-556. Drescher, J., Verhagen, W., Flik, J. and Stachan, R. (1987) Comparative investigation of monoclonal and polyclonal antibodies directed against strain-specific and common antigenic sites on influenza HIN1 virus hemagglutinin. Intervirology 27, 205-217. Drescher, J., Renger, D. and Flik, J. (1990) Method for determining the concentration of influenza virus antihemagglutinin antibody molecules, the number of epitopes per virus particle and the equilibrium constant of virus antibody interaction. J. Virol. Methods 27, 295 310. Fazekas de St. Groth, S. and Webster, R.G. (1961) Methods in immunochemistry of viruses I1. Evaluation of parameters from equilibrium measurements. Aust. J. Exp. Biol. Med. Sci. 39, 563582. Gerhard, W., Yewdell, J., Frankel, M.E. and Webster, R.G. (1981) Antigenic structure of influenza virus hemagglutinin defined by hybridoma antibodies. Nature (London) 290, 713 717. Jackson, D.C., Murray, J.M., White, D.O. and Gerhard, W.U. (1982) Enumeration of antigenic sites of influenza virus hemagglutinin. Infect. Immunity 37, 912 918. Laver, W.G. (1960) Purification of influenza virus. In: K. Habel and N.P. Salzmann (Eds.), Fundamental Techniques in Virology, Academic Press, New York. pp. 82 86. Laver, W.G. (1973) The polypeptides of influenza viruses. Adv. Virus Res. 18, 57 103. Laver, W.G., Downie, J.C. and Webster, R.G. (1974) Studies on antigenic variation in influenza virus. Evidence for multiple antigenic determinants on the hemagglutinating subunits of A/Hong Kong/68 (H3N2) virus and the A/England/72 strains. Virology 59, 23~244. Palmer, D.F., Coleman, I.T., Dowdle, W.R. and Schild, G.C. (1975) Advanced laboratory techniques for influenza diagnosis. Public Health Service, US Department of Health, Education and Welfare. Center for Disease Control, Atlanta, GA. Reimer, C.B., Baker, R.S., Newlin, T.E. and Havens, M.L. (1966) Influenza virus purification with the zonal ultracentrifuge. Science 152, 1379-1381. Stachan, R. and Drescher, J. (1987) Electron microscope method for determining the concentration of influenzavirus antihemagglutininantibodies of the IgG class. J. Virol. Methods 18, 179 192. Webster, R.G. and Laver, W.G. (1980) Determination of the number of non-overlapping antigenic
88 areas on Hong Kong (H3N2) influenza virus hemagglutinin with monoclonal antibodies and the selection of variants with potential epidem-ological significance. Virology 104, 139-148. Webster, R.G., Laver, W.G. and Air, G.M. (1983) In: P. Palese and D.W. Kingsbury (Eds.), Genetics of Influenza Virus, Springer-Verlag, Wien, New York, pp. 127-168. Wiley, D.C., Wilson, J.A. and Skehel, J.J. (1981) Structural identification of the antibody binding sites of Hong Kong influenza hemagglutinin and their involvement in antigenic variation. Nature (London) 289, 373-378. Wiley, D.C. and Skehel, J.J. (1987) The structure and function of the hemagglutinin membrane glycoprotein of influenza virus. Ann. Rev. Biochem. 56, 365-394. Yewdell, J.W., Webster, R.G. and Gerhard, W. (1979) Antigenic variation in three distinct determinants of an influenza type A hemagglutinin molecule. Nature 279, 246-248. Yewdell, J.W., Gerhard, W. and Bach, T. (1983) Monoclonal anti-hemagglutininantibodies detect irreversible antigenic alterations that coincide with the acid activation of influenza virus A/PR8/ 34-mediated hemolysis. J. Virol. 48, 239 248.
75
VIRMET 01461
Comparative investigation of the hemagglutinin epitopes of influenza virus A/Brazil/11/78 (H 1N 1) and its escape variants J. Drescher and W. Verhagen Institute of Virology, Medical School of Hannover, Hannover, (Germany) (Accepted 12 October 1992)
Summary A method described previously for determining the concentration of influenza virus antihemagglutinin antibody molecules, the number of epitopes per virus particle and the equilibrium constant of virus antibody interaction was adapted to the use with escape variants (EVs), produced by multiplication of influenza virus A/Brazil (H1N1) in presence of monoclonal antibody directed to each of the four hemagglutinin sites (Sa, Sb, Ca and Cb). The EVs were found to possess an altered antigenic site, which was both antigenic and immunogenic. By use of selected EVs and antibody preparations, the number of epitopes per antigenic site was determined and it was found that each of the four sites was represented by about 390 epitopes per virus particle, suggesting that each of the about 400 hemagglutinin spikes per virion possessed one epitope of the specificity Sa, Sb, Ca and Cb. Alteration of site Sa but not of site Ca increased the avidity of antibody to react with the unchanged sites.
Introduction The hemagglutinin (HA) of influenza virus possesses different antigenic sites (Webster and Laver, 1980; Breschkin et al., 1981; Jackson et al., 1982; Drescher et al., 1987). Two antigenic sites with strain specificity, designated Sa and Sb, and two sites common to antigenically related strains, referred to as Ca and Cb, have been defined operationally for H1 hemagglutinin (Gerhard et al., 1981; Correspondence to: J. Drescher, Institute of Virology, Medical School of Hannover, 3000 HannoverKleefeld, Konstanty-Gutschow Str. 8, Germany.
76
Caton et al., 1982), where Ca has been further divided into the subsites Ca 1 and Ca 2 (Caton et al., 1982). By clustering within the three-dimensional HA configuration of amino acid substitutions of variants selected by monoclonal antibody, 5 antigenic sites, designated A through E, have been proposed for H3 hemagglutinin (strain A / H o n g Kong/1/68)(Wiley et al., 1981, Wiley and Skehel, 1987). The H1 and H3 sites are at least partially equivalent (e.g., the H1 site Sb and the H3 site B) (Caton et al., 1982; Yewdell et al., 1983). Multiplication of influenza virus in presence of monoclonal antibody directed against one site can result in the selection of escape variants (EVs) which are present in the 'parent virus (PV) at a frequency of about 10-~ and do not combine with the antibody used for their selection (Yewdell et al., 1979; Webster and Laver, 1980; Webster et al., 1983). This is interpreted to reflect the loss of one antigenic site of EVs. Whether the lost site is regularly replaced by a new site not present on PV is unknown. For EVs of the H3N2 influenza virus strain A/Memphis/I/71, it was found that adsorption of sera raised against some EVs onto PV removed the HI activity to the EVs completely, while antibodies raised to other EVs were incompletely adsorbed on PV, suggesting that they possessed a new antigenic site (Webster et al., 1983). The experiments described in this paper were designed to examine comparatively the hemagglutinin epitopes of the H1N1 influenza virus strain A/Brazil/11/78 and its EVs. More specifically, the following points were examined: (i) By cross-adsorption of EV antibody onto PV and vice versa, it was established that EVs possessed new antigenic sites (subsequently referred to as altered antigenic site). (ii) The number of epitopes per antigenic site has been determined previously (Drescher et al., 1987) in terms of the number of epitopes per virus particle recognized on A/Brazil virus by monoclonal antibody, directed to each of the four hemagglutinin sites. However, the sum of the number of epitopes found exceeded by far the total number of epitopes present per virus particle. This result was interpreted to reflect cross-reaction of antibody with heterologous sites. In order to obtain more accurate values of the number of epitopes per antigenic site, we determined the s values recognized by anti-PV and anti-EV antibody on PV and vice versa. Thereby, the number of epitopes per antigenic site is obtained as difference of the respective s values found. Since this difference is independent from cross-reactions, more accurate estimates of the number of epitopes per antigenic site can be obtained by this approach. (iii) Furthermore, the alteration of one antigenic site on EVs was examined to determine whether this influences the avidity of reaction of antibody with the unaltered sites.
77 Material and Methods
Virus The egg-adaPted influenza virus HIN1 strain A/Brazil/11/78 was used. EVs were produced by means of three egg-passages in the presence of an excess of monoclonal antihemagglutinin antibody, directed to one site of A/Brazil virus, and two subsequent dilution limit passages (Webster and Laver, 1980). The EVs were earmarked by the symbol 'EV' followed by the antigenic site against which the antibody used for selection was directed. As example, EV-Sa refers to an EV selected by monoclonal antibody directed to site Sa. Furthermore, the reassortant A/Aichi/2/68 (H3) -A/Bel/42/41 (NI) was used. Virus was purified by means of sucrose gradient ultracentrifugation (Laver, 1960). The HA titers were determined by use of HA pattern test (Palmer et al., 1975) and by means of the photometric HCU (hemagglutinin concentration unit) method (Drescher et al., 1962). The concentration of virus particles per ml was enumerated by means of electron microscopy as described previously (Stachan and Drescher, 1987). The S.E. of the values found did not exceed 5.2%.
Antibody Antisera against A/Brazil virus (anti-PV) and the escape variants (anti-EV) were raised in chickens and goats as described previously (Drescher et al., 1987). Antibody preparations were heated to 56°C for 30 min. and were pretreated with M/90 KJO4 as described by Dowdle et al. (1979). The elimination of inhibitors was controlled by testing K104-treated prevaccination sera of animals against the virus strains employed. No HI activity was found. Antineuraminidase antibody was removed by adsorption onto the recombinant A/Aichi/2/68 (H3)-A/Bel/42 (N1), possessing irrelevant hemagglutinin and relevant neuraminidase. Antisera were tested for IgM and IgG antibody against homologous virus by use of sucrose gradient ultracentrifugation (Cremer and Riggs, 1979) and antibody of the IgG class was used for further experiments, only. Monoclonal A/Brazil antihemagglutinin antibody was produced and tested as described (Drescher et al., 1987). With our set of monoclonals, only 4 sites (Sa, Sb, Ca and Cb) were identified on A/Brazil virus and no division of site Ca into the subsites Cal and Ca2 as described for A/PR8/34 (H1N1) virus (Caton et al., 1982) was possible.
Isolation of antibody EV antisera were adsorbed onto an excess of PV and vice versa. An excess of virus was considered to be represented by the 10-fold virus dose, sufficient to adsorb all HI activity against the virus used for raising of antisera.
78 The resulting virus antibody complexes were separated from unadsorbed antibody by ultracentrifugation. They were subsequently split at pH 3.0, separated from virus by ultracentrifugation (Laver et al., 1974) and tested for HI activity against PV and EV.
Antibody titration Antibody titrations were carried out both by means of HI pattern test (Palmer et al., 1975) and by the use of a photometric HI test (Drescher, 1972). Photometrically determined titers were expressed as the reciprocal antibody dilution (ds0) yielding 50% hemagglutination inhibition.
Determ&ation of s and K The number of epitopes per virus particle (s) recognized by antibody and the equilibrium constant K were determined as described previously (Drescher et al., 1990). Briefly, antibody preparations were tested by equilibrium filtration (Fazekas and Webster, 1961), using graded dilutions (l/d) of a virus suspension containing V0 virus particles per ml. The slope (mv) and intercept (iF) of the straight lines obtained when plotting (l-a) versus (1-a)/(a. d) were determined, where a is the fraction of bound antibody. In addition, the antibody preparations where tested for the antibody dilution (1/ds0) yielding 50% hemagglutination inhibition in the photometric HI test (Drescher, 1972), using a virus concentration of Vo/dv, p. From these data, s was calculated as follows: s = (39.2. my-dso)l(dv,p" d v . as0)
(1)
dr is the reciprocal antibody dilution employed in equilibrium filtration and as0 the fraction of antibody bound at the dilution 1/d5o in the photometric HI test. as0 was calculated as follows: as0 = 0.5 (mp + ip + 1) -- x/0.5 2 (mp + ip + 1)2 -- mp where mp
=
(mF"
dso)/(dv,p
dr)
and ip
=
-
iv • dsoldv.
Then, the equilibrium constant K was obtained as follows:
(2)
79 K = (S • V o • i p ) / ( m p
•
dv,p)
(3)
In addition, the relative avidity of the reaction of a given antibody preparation with PV and EV was determined in terms of the HCU dose of virus yielding binding of 75% of HI activity (HAo.75), using the same starting dilution of antibody. Values of HAo.T5 were determined by equilibrium filtration.
Determination of the number of antibody molecules adsorbed per virus particle at the dilution (1/d5o) yielding 50% hemagglutination inhibition in the photometric HI t e s t ( A b , 5 o / V ) Values of A b , 5 0 / v were determined as described previously (Drescher et al., 1990). Briefly, antibody preparations were tested chemically for the concentration of antibody molecules per ml (= A) as described (Drescher et al., 1990) and were tested against EVs by equilibrium filtration (mv and iv) and by the photometric HI test (ds0) as described above. Values of mv, iF, ds0 and dv,p were used for calculating the fraction of antibody bound at the dilution 1/ds0 (= as0). Then, Ab,5o/V was obtained as follows: Ab,50/v
=
(A
• aso)/(dso
" v)
(4)
where v is the concentration of virus particles used in the photometric HI test.
Statistical analysis The standard errors (SE) of s and K values were determined as described previously (Drescher et al., 1990). The mean values of s, K and HAo.75 were tested for significant differences by means of a two-tailed t test (NCSS program), since the values met the normality test criteria.
Results
Patterns of adsorption of antibody onto virus Antisera raised against PV and EVs were adsorbed onto an excess of PV and of EVs. The completeness of antibody adsorption was tested by measuring the HI titers of supernatants against the viruses employed. Representative examples of the results obtained are listed in Table 1. Note that the anti-EV antibody were incompletely adsorbed onto PV and vice versa. These findings indicate that PV and EVs possessed a different antigenic site and that the altered site on EVs was both antigenic and immunogenic.
80 TABLE 1 Example for the patterns of adsorption of antibody onto parent virus and escape variants Antibody raised against virus
Antibody adsorbed onto excess of virus d
HI titers against virus PV
EV-Ca
EV-Sa
pv a
nil PV EV-Ca EV-Sa
2048 <4 128 256
1024 <4 54 128
1024 <4 64 <4
EV-Ca b
nil PV EV-Ca EV-Sa
1024 <4 <4 128
2048 128 <4 256
2048 <4 <4 <4
EV-Sa c
nil PV EV-Ca EV-Sa
2048 <4 128 <4
4096 <4 <4 <4
4096 128 256 <4
apv, parent virus A/Brazil. bEV-Ca, escape variant selected by anti-Ca antibody. CEV-Sa, escape variant selected by anti-Sa antibody. OThe virus dose employed was 100000 HCU/ml corresponding to 5 x l012 virus particles per ml. Since 1 virus particle has a weight of 5.25 x 10- 16 g (Reimer et al., 1966) and consists to 70% of protein (Ada and Perry, 1954), the virus dose employed contained 1.83 mg of viral protein.
Enumeration of the number of epitopes per antigenic site The method used for the determination of the number of epitopes per virus particle (s) recognized by antibody is based on the finding that there are on an average 39.2 antibody molecules adsorbed per virus particle at the antibody dilution (1/ds0), yielding 50% hemagglutination inhibition in the photometric HI test (Drescher et al., 1990). This was tested to establish whether it is also true when allowing antibody to react with EVs. Therefore, this value (Ab,50/V) was also determined using EVs. Representative examples of the results obtained are given in Table 2. Note that the values of Ab,5o/V obtained did not differ significantly from the value of 39.2 determined previously. These results indicate that the technique can be used also for measuring s of EVs. Antisera directed against PV and EVs were tested for the number of epitopes per virus particle (s) they did recognize on PV and EVs. Representative examples of the results obtained are shown in Table 3. In experimental group 1, antisera were allowed to react with the virus against which they were raised. Note that antisera in this group yielded s values not significantly differing one from another (average value 1561). This result suggested t h a t the altered site of EVs (e.g. site Sa) did not differ in s from the respective antigenic site of PV. Experimental groups 2 through 4 pertain to the reaction of anti-PV antibody with EV and vice versa. These reactions measure the number of epitopes of the
81
--
c5
-H
H-I
0
.
~
.
.
o
~
~
"a 0 . . . .
~
"d ¢,2 • .,-
0 0 0 o o
II
o 0 0 0 0 0 o
,.a
0
ok
~-~
o,
ol
O~
,,.o
"d 0 X
o
X ~o
0 ",~
::
o
~
o-n
I:= ,"a
UJ
¢'I
,~.~
~.~
' ~
¢~
.~.~
.~
0
~
82
0 ,.o
M •
A
0
U
3 e.~
++~+ +_~++ ..o ;.,, .o ..o +ca~, + +
++ ~ + +
++ ,.~ ..o + +
++ .,o~ + +
++ ,.o ..o ta~ ta~
"r. e~
E c~c~
0
<
i
e~
~
...(3
c~
~>.~
c)
83
r~l~
r~J~
r~ ~
÷
÷
~
It- ÷
~r~ l
~tI
Jr ÷
÷I~
7L÷
o
<
84
three antigen!c sites indicated. Note that an average value of s of 1181 was obtained. Group 5 comprised data obtained by allowing antibody directed to one EV to react with a different EV and relates to the number of epitopes of two antigenic sites (Sb and Cb), yielding an average value of 822. These data were used (see Table 4) to calculate the number of epitopes per antigenic site. As an example, the s value of Sa was calculated by subtracting the s value obtained for sites Sb, Ca and Cb from the respective value recorded for all four sites (Sa, Sb, Ca and Cb). The number of epitopes per antigenic site (average value 390) did not differ significantly one from each other.
Comparison of the avidity of epitope-paratope interaction for parent virus and the escape variants EV-Sa and EV-Ca Studies were carried out to determine whether the change of one antigenic site resulting in the appearance of an escape variant alters the avidity of reaction of antibody with the unaltered sites. For this purpose, monoclonal antibody directed to site Ca was allowed to react with parent virus and with the escape variant EV-Sa. The avidity of antigen-antibody interaction was recorded in terms of the HCU doses required to bind 75% of the initial HI activity (HA0.75) and in terms of the equilibrium constant K. Representative examples of the results obtained are listed in Table 5. Note that the anti-Ca antibody had significantly higher avidity (i.e. lower HA0.v5 values and K values) for the reaction with site Ca of the escape variant than with parent virus. This result indicates that the alteration of site Sa increased the avidity of reaction of antibody with site Ca. The same conclusion was reached when testing polyclonal antibody, recognizing the sites Sb, Ca and Cb, in like manner (lower part of Table 5). The antibodies were obtained by absorption onto and elution from the escape variant EV-Sa of anti-PV antisera and vice versa. Again, the antibody had higher avidity for their reaction with EV-SA than with PV. By contrast, significant differences of avidity were not found when comparing the reaction with EV-Ca and PV of antibody directed against the unaltered sites of PV-Ca (data not shown).
Discussion
The results presented in this paper indicate that the escape variants (EVs) examined each possessed an altered antigenic site which was antigenic and immunogenic. This conclusion is based on the finding that anti EV-antibody recognized all epitopes on the homologous EV, indicating that the altered site combined with antibody, and that adsorption of EV-antisera onto PV did not completely remove the HI activity against EV. The analogous result was obtained when adsorbing anti-PV antibody onto EV.
85 <
0 .,.0
I=
~14q
~.~.
,q.o.
.=_
.o
e4~
e4~
X
n~
:> ,,.., r',l
~
X
v
v
tm i.
I"---
0
~
I"xl
o
o~
.~. ~
o ° o°
t"q~
6e.,
Ob 0..~
.~. >~
0
>~,
~.
0 ",~ " ~
>~, II o "~
86
The number of epitopes per virus particle recognized by each of monoclonal antibodies directed to antigenic sites Sa through Cb has been previously determined (Drescher et al., 1987). The sum of the epitopes recognized by each antibody exceeded by far the total number of epitopes present per virus particle. This has been interpreted to reflect the potential of cross-reaction of antibody with antigenic sites or to result from the reaction of antibody with overlapping sites. As example, monoclonal anti-Ca antibody will not only react with the Ca epitopes but will also cross-react with epitopes of site Cb, resulting in an overestimation of the number of epitopes of site Ca. In order to obtain more accurate values of the number of epitopes per antigenic site, we determined the total number of epitopes per virus particle, recognized by anti-PV antibody and anti-EV-antibody on PV and vice versa. The reaction of anti-PV with PV measures the total number of epitopes per virus particle and the reaction of anti-EV with PV the number of epitopes per virus particle with the exception of the epitopes, belonging to the antigenic site not present on the EV used. The number of epitopes of the respective antigenic site is obtained by subtraction. Since the antibody directed to the altered antigenic site of EV does not cross-react with PV, these values are not influenced by cross-reaction of antibody. By use of this strategy, the number of epitopes per antigenic site was determined. The sum of the number of epitopes per antigenic site agreed well with the total number of epitopes present per virus particle, indicating that reliable estimates were obtained. On an average, each site comprised 390 epitopes. Since an influenza virion possesses about 400 hemagglutinin spikes (Laver, 1973) this finding indicates that each hemagglutinin spike possessed one epitope of each specificity (Sa, Sb, Ca and Cb). The avidity of epitope-paratope interaction was comparatively tested for EV-Sa, EV-Ca and PV. Alteration of site Sa (EV-Sa), but not of site Ca (EVCa), enhanced the avidity of reaction of antibody directed against unaltered sites, as indicated by the finding that anti-Ca antibody had a higher avidity against site Ca of EV-Sa than against site Ca of PV. This result indicates that alteration of one antigenic site can alter the avidity of reaction of unchanged sites with antibody. This could be interpreted to result from conformational changes. As an alternate explanation, it could be assumed that antibodies combine with lower avidity with heterologous sites than with the homologous site of PV. If the cross-reacting site is absent on EV, the measured avidity of the reaction of antibodies with E-V is higher than that recorded for their reaction with PV. At present, it is not known whether this interpretation is correct.
Acknowledgements
The authors acknowledge with much appreciation the excellent technical assistance of Mrs. R. Assmann, Mrs. D. Findeisen-Weidauer, Mrs. G. Harste,
87
Mrs. J. Milzer, Mrs. K. Stubbendorff, Mrs. A. Thiemer, and the competent help of Mrs. A. Meyer and Mrs. B. Thierkopf in preparing this manuscript. References Ada, G.L. and Perry, B.T. (1954) The nucleic acid content of influenza virus. Aust. J. Exp. Biol. Med. Soc. 32, 453-468. Breschkin, A.M., Ahern, J. and White, D.O. (1981) Antigenic determinants of influenza virus hemagglutinin.. III. Topography of the antigenic regions of influenza virus hemagglutinin determined by competitive radioimmunoassay with monoclonal antibodies. Virology 113, 130140. Caton, A.J., Brownlee, G.G., Yewdell, J.W. and Gerhard, H. (1982) The antigenic structure of the influenza virus A/PR/8/34 hemagglutinin (HI subtype). Cell 31,417-427. Cremer, N.E. and Riggs, J.L. (1979) In: D.H. Lennette and N.J. Schmidt (Eds.), Immunoglobulin Classes and Viral Diagnosis, 5th ed., American Public Health Association, Washington, D.C., pp. 191 208. Dowdle, W.A., Kendal, A.P. and Noble, G.R. (1979) Influenza viruses. In: E.H. Lennette and N.J. Schmidt (Eds.), Diagnostic Procedures for Viral, Rickettsial and Chlamydial Infections, 5th Ed., American Public Health Association, Washington, D.C. pp. 585 609. Drescher, J., Hennessy, A.V. and Davenport, F.M. (1962) Photometric methods for the measurement of hemagglutinating viruses and antibody. 1. Further experience with a novel photometric method for measuring hemagglutinin. J. lmmunol. 89, 794-804. Drescher, J. (1972) Patterns of cross-reaction of selected influenza virus A 2 strains isolated from 1957 to 1968 as determined by use of a photometric hemagglutination inhibition test. Am. J. Epidemiol. 95, 549-556. Drescher, J., Verhagen, W., Flik, J. and Stachan, R. (1987) Comparative investigation of monoclonal and polyclonal antibodies directed against strain-specific and common antigenic sites on influenza HIN1 virus hemagglutinin. Intervirology 27, 205-217. Drescher, J., Renger, D. and Flik, J. (1990) Method for determining the concentration of influenza virus antihemagglutinin antibody molecules, the number of epitopes per virus particle and the equilibrium constant of virus antibody interaction. J. Virol. Methods 27, 295 310. Fazekas de St. Groth, S. and Webster, R.G. (1961) Methods in immunochemistry of viruses I1. Evaluation of parameters from equilibrium measurements. Aust. J. Exp. Biol. Med. Sci. 39, 563582. Gerhard, W., Yewdell, J., Frankel, M.E. and Webster, R.G. (1981) Antigenic structure of influenza virus hemagglutinin defined by hybridoma antibodies. Nature (London) 290, 713 717. Jackson, D.C., Murray, J.M., White, D.O. and Gerhard, W.U. (1982) Enumeration of antigenic sites of influenza virus hemagglutinin. Infect. Immunity 37, 912 918. Laver, W.G. (1960) Purification of influenza virus. In: K. Habel and N.P. Salzmann (Eds.), Fundamental Techniques in Virology, Academic Press, New York. pp. 82 86. Laver, W.G. (1973) The polypeptides of influenza viruses. Adv. Virus Res. 18, 57 103. Laver, W.G., Downie, J.C. and Webster, R.G. (1974) Studies on antigenic variation in influenza virus. Evidence for multiple antigenic determinants on the hemagglutinating subunits of A/Hong Kong/68 (H3N2) virus and the A/England/72 strains. Virology 59, 23~244. Palmer, D.F., Coleman, I.T., Dowdle, W.R. and Schild, G.C. (1975) Advanced laboratory techniques for influenza diagnosis. Public Health Service, US Department of Health, Education and Welfare. Center for Disease Control, Atlanta, GA. Reimer, C.B., Baker, R.S., Newlin, T.E. and Havens, M.L. (1966) Influenza virus purification with the zonal ultracentrifuge. Science 152, 1379-1381. Stachan, R. and Drescher, J. (1987) Electron microscope method for determining the concentration of influenzavirus antihemagglutininantibodies of the IgG class. J. Virol. Methods 18, 179 192. Webster, R.G. and Laver, W.G. (1980) Determination of the number of non-overlapping antigenic
88 areas on Hong Kong (H3N2) influenza virus hemagglutinin with monoclonal antibodies and the selection of variants with potential epidem-ological significance. Virology 104, 139-148. Webster, R.G., Laver, W.G. and Air, G.M. (1983) In: P. Palese and D.W. Kingsbury (Eds.), Genetics of Influenza Virus, Springer-Verlag, Wien, New York, pp. 127-168. Wiley, D.C., Wilson, J.A. and Skehel, J.J. (1981) Structural identification of the antibody binding sites of Hong Kong influenza hemagglutinin and their involvement in antigenic variation. Nature (London) 289, 373-378. Wiley, D.C. and Skehel, J.J. (1987) The structure and function of the hemagglutinin membrane glycoprotein of influenza virus. Ann. Rev. Biochem. 56, 365-394. Yewdell, J.W., Webster, R.G. and Gerhard, W. (1979) Antigenic variation in three distinct determinants of an influenza type A hemagglutinin molecule. Nature 279, 246-248. Yewdell, J.W., Gerhard, W. and Bach, T. (1983) Monoclonal anti-hemagglutininantibodies detect irreversible antigenic alterations that coincide with the acid activation of influenza virus A/PR8/ 34-mediated hemolysis. J. Virol. 48, 239 248.