Antibodies as Measured by Means of the Photometric Antibody Concentration Unit Method

Transcription

1 JOURNAL OF VIROLOGY, Dec. 1974, p Copyright American Society for Microbiology Vol. 14, No. 6 Printed in U.SA. Kinetics of Inhibition of Influenza Virus Hemagglutination by Homologous Antihemagglutinin and Antineuraminidase Antibodies as Measured by Means of the Photometric Antibody Concentration Unit Method JOACHIM DRESCHER, ULRICH DESSELBERGER, AND DIETRICH LUBACH Institute of Virology, Medizinische Hochschule Hannover, D 3 Hannover-Kleefeld, Germany Received for publication 11 September 1974 The kinetics of the reactions of antihemagglutinin (AH) and antineuraminidase (AN) antibodies with homologous influenza virus were examined by use of a photometric hemagglutination inhibition test (antibody concentration unit [ACU] test). The following results were obtained. (i) The isotherms describing the reaction of AN antibodies with homologous virus were found to have much steeper slopes than did the corresponding AH antibody isotherm. This finding indicates that the photometric ACU method can discriminate objectively between AH and AN antibodies. (ii) The reaction of mixtures of AH and AN antibodies with homologous virus was examined and found that AH antibodies combine with virus independently from the presence of AN antibodies, whereas AH antibodies were found to reduce greatly the measurable degree of hemagglutination inhibition by AN antibodies. (iii) A formula was developed and experimentally confirmed by means of which the binding of virus by mixtures of AH and AN antibodies can be predicted. Thereby, the influence of the relative concentration of AH and AN antibodies on the establishing of isotherms and on their subsequent use for antibody titration by means of the photometric ACU method was investigated. It was found that the procedure routinely employed for establishing isotherms yielded data reflecting the reaction of AH antibodies only. As a consequence, the use of these isotherms will identify AH antibodies. In a preceding publication (7), a highly accurate photometric method for the determination of hemagglutination inhibiting influenza virus antibody has been described. This so called photometric ACU (antibody concentration unit) method is based on the finding that, when an excess of virus is reacted with graded dilutions of homologous antiserum, the ratio of bound and free virus to the concentration of antibody present can be described in terms of Freundlich's adsorption isotherm (10). As applied, log CN/C.b = log A + 1/N log C', where CN is the concentration of bound and C' of free virus, and Cab the concentration of antibody present. CN is determined in operational terms as difference between hemagglutinin concentration units (HCUs) added (C) and HCUs recovered (C'). A and 1/N are two constants which are experimentally determined for the strain of virus employed. Log A equals the intercept and 1/N the slope of the straight line obtained when plotting the values of log CN/Ca, against the values of log C'. With heterologous reactants, the reaction cannot be described in terms of Freundlich's isotherm, since a different type of proportionality pertains (2, 7). As a consequence, testing whether or not the kinetics of the reaction of an unknown antiserum with a given strain of virus can be described in terms of the isotherm established for the strain of virus employed enables the reactions of virus with homologous and heterologous antibody to be objectively discriminated (2, 7). This technique has been utilized to test the specificity of antibody patterns recorded by conventional hemagglutination inhibition (HI) tests in human sera (2, 3, 4). When these experiments were done, it was not known' that under certain circumstances antineuraminidase (AN) antibodies can cause hemagglutination inhibition (11, 16). Therefore, it was of interest to test whether or not AN antibody can influence recognition of specific antihemagglutinin (AH) antibody by means of the photometric ACU method. For this purpose, the reaction between the virus strains A/Bel/42 (HO N1), A/Sing/1/57 (H2 N2), and the recom- 1361

2 1362 DRESCHER, DESSELBERGER, AND LUBACH J. VIROL. binant A/Bel (HO)-A/Sing (N2) carrying the hemagglutinin of the strain A/Bel and the neuraminidase of the strain A/Sing/1/57 was examined. Since the strains A/Bel and A/Sing were found not to cross-react measurably, the reactions of A/Sing antibodies with A/Bel-A/ Sing virus and of A/Bel-A/Sing antibodies with A/Sing virus represent AN antibody-mediated hemagglutination inhibition. The experiments described in this paper were designed to investigate the mutual influence of AH and AN antibodies on the establishing of adsorption isotherms and on their use for antibody titration by means of the ACU method. For this purpose, the kinetics of reacting 'separately AH and AN antibodies with homologous virus was investigated. Since recognition of specific antibody by means of the ACU method is based on the kinetics of virus antibody interaction, the finding of different kinetics for AN and AH antibody-virus interaction would indicate that both types of antibodies can be discriminated by means of the ACU method. Furthermore, a formula was developed and experimentally confirmed by means of which the binding of virus by mixtures of AH and AN antibodies can be predicted. By use of this formula, the levels of AN antibodies influencing measurably the reaction of virus with AH antibodies can be determined. In addition, it was felt that the knowledge of the kinetics of the reaction of AH and AN antibodies with homologous virus as obtained by the experiments described in this paper could contribute to a better understanding of the role of each type of antibody in immunity to influenza virus infection. MATERIALS AND METHODS Virus. The strains of egg-adapted influenza virus employed were: A/Sing/1/57 (H2 N2), A/Bel/42 (HO N1), and the recombinant A/Bel/42 (HO)-A/Sing/I/ 57 (N2) carrying A/Bel/42 hemagglutinin and A/Sing/ 1/57 neuraminidase. (The virus suspensions used were inhibition as described by Seto and Rott (13), using found to have the following ratios of neuraminidase KIO4 pretreated sera. activity [mu determined using fetuin purchased from Neuraminidase inhibition titers were expressed in K and K Laboratories, Plainville, N.Y., as substrate] terms of the geometric means of the reciprocal highest to hemagglutinin (HA) activity: A/Bel mu/ha, serum dilutions yielding 50% inhibition, using neuraminidase starting concentrations yielding optical den- A/Bel-A/Sing mu/ha, and A/Sing mu/ha.) In addition, the strains A/England/1/61 sity readings of 0.05 at a wave length of 549 nm in a (H2 N2), A/AA/1/65 (H2 N2), and A/Aichi/2/68 (H3 model DB Beckman spectrophotometer (light path of N2) were used. Virus was purified by adsorption on 1 cm). and elution from BASO4 (8) and was suspended in buffered saline (0.15 M NaCi), buffered at ph 7.0 RESULTS with 0.01 M phosphate. Sera. Antisera were prepared by vaccination of Pattern of reactions of virus strains A/Sing chickens and rabbits with the virus strains listed (H2 N2), A/Bel (HO Ni), and the recombinant above. The inoculation schedule was 10 intraperito- A/Bel (HO)-A/Sing (N2). The reactions of the neal injections given at 24-h intervals. Each contained 100 HCUs of virus and was suspended in 20 ml of buffered saline (5). Blood samples were drawn prior to vaccination and 15, 30, 45, and 60 days thereafter. Sera were pretreated with M/90 KIO,. Prior to vaccination, none of the animals showed detectable antibody against the virus strains employed. Antibody titrations. (i) Hemagglutination inhibiting antibody were determined by means of the photometric ACU method (7). Titers were expressed in terms of ACU units (7). (One ACU unit is defined as that concentration of antibody which is just sufficient to give the value 1.0 for the ratio CN/AC'1/N when reacted with homologous virus.) In addition, the reaction of sera with virus was characterized by determining the reciprocal highest serum dilution (d5*) yielding binding of 50% of 90 to 110 HCUs of virus by means of a photometric hemagglutination inhibition test (5). (ii) For establishing adsorption isotherms, the technique used was that previously described (7). Each isotherm was based on 80 to 130 measurements. The "standard serum" (7) was assigned a convenient numerical value in order to have ACU titer values computed by use of the isotherm in the same numerical range of magnitude as HI titers determined by conventional HI tests. Hereby, the HI titers of A/Sing antiserum against A/Bel-A/Sing virus and of A/ Bel-A/Sing antiserum against A/Sing virus were used when establishing the corresponding isotherms. The isotherms were tested for linearity as described by Fazekas (9) as follows: for each isotherm, the values of 1/N and log A were calculated separately for values where C' (C' = concentration of free virus) ranged from 10 to 50 and for values where C' exceeded 50. The values of 1/N and log A were tested for significant differences by means of a t test (P < 0.05). When the 1/N and log A values of these partial isotherms failed to differ significantly, the isotherm was considered linear. (iii) HI pattern tests were carried out using four agglutinating doses of virus. Titers were expressed in terms of the reciprocal highest serum dilution yielding complete hemagglutination inhibition. (iv) Neuraminidase activity assays were carried out according to the method of Warren (15), using mucoprotein from human urine prepared according to Tamm and Horsfall (14) as substrate. Sera were tested repeatedly for neuraminidase

3 VOL. 14, 1974 INHIBITION OF INFLUENZA VIRUS BY ACU METHOD virus strains A/Bel, A/Sing, and the recombinant A/Bel-A/Sing with the corresponding antisera were examined by use of the photometric hemagglutination inhibition test (d,0), HI pattern test, and neuraminidase inhibition test. Representative examples of the results are given in Table 1. Note that the A/Bel (HO)-A/Sing (N2) antiserum yielded AN antibody-mediated hemagglutination inhibition of A/Sing virus and A/Sing antiserum of A/Bel (HO)-A/Sing (N2) virus. This conclusion is based on the finding that the A/Bel antiserum failed to react measurably with A/Sing virus and the A/Sing antiserum with A/Bel virus. Establishing of isotherms. The constants A and 1/N of the isotherms describing the reaction of the virus strains A/Sing, A/Bel, and A/Bel-A/Sing with the corresponding antisera were determined where possible. Table 2 gives a survey on the adsorption isotherm constants obtained. (The isotherms presented were found to be linear.) The values of the constants A and 1/N were found not to differ significantly (P < 0.5) for the reactions of A/Bel and A/Bel-A/Sing antisera with A/Bel and A/Bel-A/Sing virus, indicating that these data were not measurably influenced by the presence-or absence of homologous AN antibodies. TABLE 1. Not shown for convenience in presentation is that the ACU titers of A/Bel and of A/Bel-A/ Sing antisera obtained by using A/Bel virus were found not to differ significantly from the ACU titers recorded when reacting these antisera with A/Bel-A/Sing virus. In contrast, the isotherms pertaining to AN antibody-mediated hemagglutination inhibition (reaction of A/Sing antisera with A/Bel-A/Sing virus and of A/ Bel-A/Sing antisera with A/Sing virus) had significantly (P < 0.05) higher values of the constants 1/N and lower values of the constant A than the isotherms describing the reactions of these virus strains with AH antibodies (reaction of A/Bel antibody with A/Bel-A/Sing virus and of A/Sing antibody with A/Sing virus). This finding warrants the conclusion that AH and AN antibodies reacted separately with homologous virus will be identified correctly by use of the ACU method. Reaction of mixtures of AH and AN antibodies with homologous virus. To analyze the reaction of mixtures of AH and AN antibodies with homologous virus, it was necessary to obtain formulas by means of which the virus concentration bound by AH and AN antibodies separately can be precisely predicted. Binding of virus by AH antibodies. It was found that when reacting C.b,H ACUs of AH Pattern of cross-reactions of strains A/Bel (HO Nl), A/Sing (H2 N2), and the recombinant A/Bel (HO)-A/Sing (N2)a Antisera reacted with strain of virus Antisera obtained by vaccination with A/Bel (HO N1) A/Bel (HO)-A/Sing (N2) A/Sing (H2 N2) strain of virus dso HI NIT d5. HI NIT d5o HI NIT A/Bel (HO N1) 3,936 1, ,950 1,120 < 10 <62 <28 < 10 A/Bel (HO)-A/Sing 8,700 3,584 <10 13,987 3, ,527 A/Sing H2 N2) <62 < 28 < 10 3, ,090 j 8,960 2,740 a d0, Reciprocal highest serum dilution yielding binding of 50% of 90 to 110 HCUs; HI, reciprocal highest serum dilution yielding complete inhibition of four agglutinating doses of virus; NIT (neuraminidase inhibition test), reciprocal highest serum dilution inhibiting 50% of neuraminidase standard dose. TABLE 2. Constants A and 1/N of the isotherms describing the reaction of strains A/Bel (HO Nl), A/Bel (HO)-A/Sing (N2), and A/Sing (H2 N2) with antibodies Sera reacted with virus straina Antisera oriented to strain of virus A/Bel (HO N1) A/Bel (HO)-A/Sing (N2) A/Sing (H2 N2) A 1/N A 1/N A 1/N 1363 A/Bel (HO N 1) A/Bel (HO)-A/Sing (N2) A/Sing (H2 N2) a -, No isotherm obtained, since sera failed to react with virus.

4 1364 DRESCHER, DESSELBERGER, AND LUBACH J. VIROL. antibodies with C HCUs of homologous virus the concentration of bound (CNNH) and free virus (C 'H) can be precisely predicted by means of equation 1, the derivation of which has been previously (6) described: CN,H = C - C'H (1) where C'H equals ClyCab H, y equals JOAHC(1/N,H - 1)/2.303 and AN and 1/N,H are the constants of the isotherm describing the reaction of AH antibodies with homologous virus. Binding of virus by AN antibodies. It was empirically found that equation 1 may not be applied to data where 1/N exceeds 1.0. Therefore, it cannot be used for describing the reaction of AN antibody with homologous virus. However, it was empirically found that this reaction can be described in terms of equation 2: P = R log (TNQld) (2) where P is the fraction of virus bound, TN is the concentration of AN antibodies in the undiluted serum in terms of ACU units, and d is the reciprocal serum dilution tested. R and Q are two constants which are experimentally determined for the strain of virus employed (Table 3). The determination of these constants and the derivation of equation 2 are presented in the Appendix to this paper. Representative examples of the results obtained when testing the validity of equation 2 are given in Table 4. An antiserum oriented to A/Sing virus was reacted with A/Bel-A/Sing virus and the concentration of bound virus observed was compared with the binding of virus predicted by use of equation 2. Note that the values of the concentration of bound virus predicted agreed well with the experimentally observed values. Binding of virus by mixtures of AH and AN antibodies. The binding of homologous virus by graded dilutions of mixtures of AH antisera and AN antisera was photometrically determined, reacting mixtures of A/Sing and A/Bel antisera with A/Bel-A/Sing virus and mixtures of A/Sing and A/Bel-A/Sing antisera with A/Sing virus. It was found that the binding of virus by AH and AN antibody mixtures could be described means of an equation derived by making the following assumptions: it was assumed that AH antibody react independently from the presence of AN antibody. Then, the concentration of virus bound (CNNH) when reacting a concentration CG.,H of AH antibody with C HCUs of homologous virus is obtained by means of equation 1. by It was further assumed that the binding of virus by AN antibodies equals the binding of virus expected when the concentration of virus free after the reaction with AH antibodies (C'H) reacts with AN antibodies. Then, the virus starting concentration operative against the AN antibodies equals C H = A/yCab,H and the concentration of virus bound by AN antibody (CN N) is obtained by use of equation 3: CN.N = C R log (TN. Q/d)/yCab, H (3) As a consequence, the total concentration of bound virus (CN) equals the sum of the virus concentrations bound by AH and by AN antibodies: CN = C - C/yC ab. H + C.R log (TN. Q/d)/yCab. H (4) TABLE 3. Constants R and Q of the formulaa describing the reaction of antineuraminidase antibody with homologous virus Antiserum Reacted with R Q oriented to: virus strain A/Sing (H2 N2) A/Bel-A/Sing (HO H2) A/Bel-A/Sing A/Sing (H2 N2) (HO N2) apercent bound virus = R log (TN. Q/d), where TN equals ACU titer of AN antibody and d is the reciprocal serum dilution tested. TABLE 4. Examples of the reaction of AIBel-AlSing (HO N2) virus with A/Sing (H2 N2) antibodya Reciprocal serum Cab.N C CNObs. CN.exp. dilution , , , , , , , , , , , , , a CaO,N, Concentration of AN antibody reacted; C, virus starting concentration; CN.Oft., concentration of bound virus experimentally observed; CN.exp., concentration of bound virus calculated according to equation 2.

5 VOL. 14, 1974 INHIBITION OF INFLUENZA VIRUS BY ACU METHOD consequence, the conclusion is warranted that AH antibodies react independently from the presence of AN antibodies with homologous virus, whereas AH antibodies reduce the degree of measurable hemagglutination inhibition by AN antibodies. It should be noted that alternate models describing the reaction of mixtures of AH and AN antibodies with homologous virus were also tested. These models included the assumption Representative examples of the results obtained when testing experimentally the validity of equation 4 are charted in Table 5. Graded concentrations of A/Bel-A/Sing virus were reacted with mixtures of A/Bel and A/Sing antisera. The reciprocal serum dilutions and the corresponding antibody concentrations are charted in columns 2, 3, 6, and 7. Column 5 gives the concentration of virus bound by AH antibody calculated according to equation 1, and column 4 the corresponding concentration of free virus C'H (= C - CN.H). The fraction of virus bound by AN antibody was calculated according to equation 3, and the values found were listed in column 8. Then, the concentration of virus bound by AN antibody (CNN) was obtained by multiplication of C'H values with P and the values obtained were listed in column 9. The total binding of virus expected (sum of CN.H and CN.N) is charted in column 10 (CN expected). Note that the values obtained agreed well with the experimentally observed values (see column 11). The results indicate that equation 4 is valid for describing the reaction of homologous virus with mixtures of AH and AN antibodies. As a that AN antibodies combine with virus independently from the presence of AH antibodies and that AH antibodies react with the concentration of virus free after reaction with AN antibodies. However, the concentrations of bound virus predicted by use of an equation based on this assumption were found to differ significantly (P < 0.05) from the experimentally determined concentrations of bound virus, using virus starting concentrations from 130 to 310 HCUs, to ACUs of AH antibodies, and 1.72 to ACUs of AN antibodies. Therefore, this model was found not to be valid. The same conclusion was reached when assuming that binding of virus by mixtures of AH and AN antibodies would equal the sum of virus bound when each antibody is reacted with virus separately. Influence of AN antibodies on establishing of adsorption isotherms for the reaction of AH antibodies with homologous virus. It was examined whether or not the isotherms previously established for various strains of influenza virus represent AH antibodies only, or were measurably influenced by the levels of AN antibodies possibly present. For this purpose, isotherms for the reactions of A/Bel-A/Sing virus with graded mixtures of A/Sing (= AN) TABLE 5. Reaction of AIBeI-AlSing virus with a mixture of AH (= antiserum against A/Bel) and AN (= antiserum against A/Sing) antibodiesa Antiserum Antiserum c against A/Bel (192.1 ACU) against A/Sing (1000 ACUN) CN CN. expected observed ds C'H C-C'H ds P Cab,H Cab.N CN li 'CH C , , n- X-r _ -,I z_z A. A. _. A-; _:_ l - ALU f"ha,. a u, Virus starting concentration; ds, reclprocal serum dilutuon; -ab,h, concentration 01 Ati antioocy; L.? H, concentration of virus free after reaction with AH antibody; ACUN, ACU titer measured against A/Bel-A/Sing virus calculated with the isotherm constants A = and 1/N = (= AN antibodies); Cb.tN, concentration of AN antibody; P, fraction of virus bound by AN antibody; CN,N, concentration of virus bound by AN antibody; CN, total virus concentration bound. 1365

6 1366 DRESCHER, DESSELBERGER, AND LUBACH J. VIROL. and A/Bel (= AH) antisera were established. The results obtained are given in Table 6. The ratio of AH (= A/Bel) to AN (= A/Sing) antibodies expressed in terms of d,0 values is given in column 2. Note that A/Bel antiserum without addition of A/Sing antiserum was used in experiment no. 1. The number of data obtained in each experiment is given in column 3. Columns 4 and 6 list the isotherm constants found. By use of a t test (P < 0.05), it was examined whether the 1/N and log A values of experiments 2 to 7 differed significantly from the corresponding value of experiment 1 (see columns 5 and 7). The standard errors of the 1/N values were found not to exceed and those of the log A values not to exceed The results obtained indicate that AN antibodies failed to influence measurably the adsorption isotherms, when the serum mixtures contained at least three times higher levels of AH antibodies than AN antibodies in terms of d,0 values (see experiments 2 to 5). In contrast, the presence of higher relative concentrations of AN antibodies (see experiments 6 and 7) resulted in a significant increase of 1/N and log A values. When the relative concentration of AN antibodies was further increased (see experiments 8 to 9), no isotherms were obtained since no linear relationship between values of log CN/CGb and log C' was recorded (see column 8). In these experiments, the partial isotherms calculated for data where C' ranged from 10 to 50 showed TABLE 6. significantly lower 1/N and higher log A values than the partial isotherms calculated for data where C' exceeded 50. It should be noted that the influence of the relative concentration of AH and AN antibodies on the establishing of adsorption isotherms can be predicted by introducing into equation 4 assumed values of TH, TN, and d, and by calculating isotherms using the values of free and bound virus obtained. The isotherms calculated by this procedure for the reactions of A/Bel and A/Sing antibody mixtures with A/Bel-A/Sing virus agreed well with the isotherms experimentally determined. For the sake of brevity, no examples are given. In addition, antisera oriented to the virus strains A/Sing/1/57 (H2 N2), A/England/1/61 (H2 N2), A/AA/1/65 (H2 N2), and A/ Aichi/2/68 (H3 N2), which had been previously used for establishing adsorption isotherms, were tested in like manner with homologous virus and, in addition, with A/Bel-A/Sing virus. The ratios of AH to AN antibodies as measured in terms of d,0 values ranged from 9.7 to This finding suggests that the immunization procedure employed for preparing antisera for establishing adsorption isotherms yields relatively low levels of AN antibodies and, as a consequence, the isotherms established measure AH antibodies only. DISCUSSION When interpreting the results obtained, it should be emphasized that all conclusions Determination of the constants A and 1/N of the isotherms describing the reactions ofa/bel-a/sing virus with graded mixtures of A/Bel and A/Sing antisera Ratio of A/Bel Significance of Significance of Expt to A/Silng A/Sin antibodies iod/es NO. N. Of o difference between difference between 1/N 1/N and 1/N value of Log A log A and log A value Linearityb no. in terms of d,5 values values expt no. 1 ofexpt.no.1 (dso,h/daj O.N)a(P < 0.05) (P < 0.05) Yes No No Yes No No Yes No No Yes No No Yes Yes Yes Yes Yes Yes Yes No No a do,,h, Reciprocal highest dilution of A/Bel serum yielding binding of 50% of 90 to 110 HCUs of A/Bel-A/Sing virus; dio,n, reciprocal highest dilution of A/Sing serum yielding binding of 50% of 90 to 110 HCUs of A/Bel-A/Sing virus. b The linearity of each adsorption isotherm was tested by dividing the data into two groups, one comprising values where the concentration of free virus ranged from 10 to 50 and a second where the concentration of free virus ranged from 51 to 110. For each group, the values of log A and 1/N were calculated separately and tested for significant differences by means of a t test (P < 0.05). When no significant difference between log A and 1/N values was recorded, the isotherm was considered linear (9).

7 VOL. 14, 1974 INHIBITION OF INFLUENZA VIRUS BY ACU METHOD reached pertain to the strains of virus employed, and that at present, no statement can be made whether or not such conclusions are valid for further virus strains or their recombinants. The data presented in this paper warrant the conclusion that AH and AN antibodies have drastically different kinetics of their reaction with homologous virus as measured by means of the photometric ACU method. This finding indicates that the photometric ACU method can discriminate between AH and AN antibodies. When using an AH antibody isotherm, ACU titers will reflect AH antibodies, and vice versa. Using graded mixtures of AH and AN antibodies, the influence of the relative concentration of AH and AN antibodies on establishing adsorption isotherms was investigated, and it was found that AN antibodies failed to influence measurably the isotherm, if the ratio of AH to AN antibodies exceeded 3.0. Since it was found that the vaccination schedule employed for preparation of antisera used for establishing isotherms yields ratios of AH to AN antibodies exceeding 3.0, this finding suggests that previously established isotherms measure AH antibody only. This conclusion is consistent with a previous report (11), where AN antibody was also found not to influence measurement of ACU titers. A modification of the ACU method by means of which both AH and AN antibodies can be determined will be described in a subsequent publication. It might be questioned whether the found difference between AH and AN antibody isotherms is due to the fact that AN antibodies were measured by means of a hemagglutination inhibition test. Therefore, it should be mentioned that when isotherms were established for the reaction of AN antibodies with N2 neuraminidases of different strains by. use of enzyme inhibition test, an average value of the constant 1/N of 1.09 was obtained (1), which is in excellent agreement with the data presented in this paper. The finding that AN antibody isotherms have much steeper slopes than AH antibody isotherms can be interpreted as follows: under comparable conditions, the amount of virus bound per ACU unit of AN antibody will increase much more rapidly with increasing virus starting concentration than does the virus concentration bound per ACU unit of AH antibody. As an example, when reacting A/Bel-A/ Sing virus with 0.1 ACUs of A/Sing antibody, an increase of the virus starting concentration from 10 to 100 HCUs will result in 10-fold 1367 increase of the virus concentration bound per ACU unit, whereas with A/Bel antibody this ratio will increase 4.8-fold only. Vice versa, the binding of virus by AN antibodies decreases much more progressively with decreasing virus starting concentration than does binding of virus by AH antibodies. The relatively low binding capacity of AN antibody against low concentrations of virus could be at least partially responsible for the fact that AN antibody do not prevent virus infection where a low virus concentration is needed but interferes with the release of virus from infected cells where a much higher virus concentration is operative (16). When examining the reaction of mixtures of AH and AN antibodies with homologous virus, it was found that binding of virus by AH antibodies was not influenced by the presence of AN antibodies. In contrast, the presence of AH antibodies did greatly reduce the measurable degree of hemagglutination inhibition by AN antibodies, since the virus concentration bound by AN antibodies was found to equal that virus concentration bound if only the virus concentration free after the reaction with AH antibodies is reacted with AN antibodies. This finding could be explained by the following hypothetical model: consider that a virus concentration C is once reacted separately with AH antibodies, yielding binding of a- C HCUs, and once with AN antibodies, resulting in binding of b-c HCUs. Since AH and AN antibodies combine with different antigenic sites, it is reasonable to assume that both antibodies combine with virus independently from each other. Thus, when reacting a mixture of AH and AN antibodies with C HCUs of virus, the concentration of virus bound by AH antibodies will again equal a C, whereas the virus concentration a * b * C is bound both by AH and AN antibodies. As a consequence, the binding of virus by AN antibodies observed equals b (C - a- C), where C - ac is the virus concentration free after reaction with AH antibodies. Thus, the decrease in measurable binding of virus by AN antibodies in presence of AH antibodies is due to the fact that virus inhibited by AH antibodies combines with AN antibodies without additional inhibition of hemagglutinating activity. APPENDIX Determination of the constants R and Q for describing the reaction of AN antibodies with homologous virus. Graded dilutions (lid) of AN antisera containing TN ACUs of AN antibodies were reacted with graded concentrations of homologous

8 1368 DRESCHER, DESSELBERGER, AND LUBACH J. VIROL. virus, and the fraction of bound virus (P) was determined under the conditions of the photometric ACU method. The slope (R) and the intercept (E) of the regression line relating values of P to values of log (TN/d) were determined by means of the method of least squares yielding: P = R log (TN/d) + E (1) This can be rearranged to give 10 = (TNI/dR) * 10E (2) 10P = (TN,.10E/1R/d)R (3) Calling the value of 1OE/R Q gives: P = R log (TN. Q/d) (4) By means of equation 4, the fraction of virus bound when reacting a known concentration of AN antibodies (TN/d) with homologous virus can be calculated. (Values of P larger than 1.0 are scored as 1.0.) ACKNOWLEDGMENTS We acknowledge with much appreciation the help of G. C. Schild, World Influenza Centre, Mill-Hill, London, who generously supplied the virus strains A/Bel, A/Sing, and A/Bel-A/Sing. The mucoprotein samples employed were a kind gift of E. Hertzberger, N. V. Philips Duphar, Weesp, The Netherlands. The excellent technical assistance of G. Andrae, B. Brunler, D. Friedrich, M. Stanschus, K. Stubbendorff, and D. Weidauer is acknowledged. LITERATURE CITED 1. Baars, A. J., H. Frankena, and N. Masurel Antigenic relationship of influenza-virus neuraminidase from Asian, Hong Kong and Equi-2 strains. Antonie van Leeuwenhoek. J. Microbiol. Serol. 37: Davenport, F. M., A. V. Hennessy, J. Drescher, J. Mulder, and T. Francis, Jr Further observations on the relevance of serologic recapitulation of human infection with influenza viruses. J. Exp. Med. 120: Davenport, F. M., A. V. Hennessy, and E. Minuse Further observations on the significance of A/ equine-2/63 antibodies in man. J. Exp. Med. 126: Davenport, F. M., A. V. Hennessy, and E. Minuse The age distribution in humans of hemagglutinatinginhibiting antibodies reacting with avian strains of influenza A virus. J. Immunol. 100: Drescher, J Patterns of cross-reaction of selected influenza virus A2 strains isolated from 1957 to 1968 as determined by use of a photometric hemagglutination inhibition test. Amer. J. Epidemiol. 95: Drescher, J Photometric method for determination of doubly specific influenza antibody. Zentralbl. Bakteriol. Hyg. Abt. I. Orig. 225: Drescher, J., F. M. Davenport, and A. V. Hennessy Photometric methods for the measurement of hemagglutinating viruses and antibody. II. Further experience with antibody determinations and description of a technique for analysis of virus mixtures. J. Immunol. 89: Drescher, J., A. V. Hennessy, and F. M. Davenport Photometric methods for the measurement of hemagglutinating viruses and antibody. I. Further experience with a novel photometric method for measuring hemagglutinins. J. Immunol. 89: Frazekas de St. Groth, S Methods in immunochemistry of viruses. 2. Evaluation of parameters from equilibrium measurements. Aust. J. Exp. Biol. 39: Freundlich, H Colloid and capillary chemistry, p Dutton, New York. 11. Hennessy, A. V., and F. M. Davenport Effect of neuraminidase on antibody combining unit (ACU) titers of human sera determined by Drescher's photometric procedure. Proc. Soc. Exp. Biol. Med. 141: 6f Kilbourne, E. D Recombination of influenza A viruses of human and animal origin. Science 160: Seto, J. T., and R. Rott Functional significance of sialidase during influenza virus multiplication. Virology 30: Tamm, I., and F. L. Horsfall, Jr Mucoprotein derived from human urine which reacts with influenza, mumps and Newcastle disease virus. J. Exp. Med. 95: Warren, L The thiobarbituric acid assay of sialic acids. J. Biol. Chem. 234: Webster, R. G., and W. G. Laver Antigenic variation in influenza virus. Biology and chemistry. Progr. Med. Virol. 13: