Studies on Rubella Virus Strain Variation by Kinetic Haemagglutination-inhibition Tests

J. gen. Virol. (I97O), 9, 215-223 2I 5 Printed in Great Britain Studies on Rubella Virus Strain Variation by Kinetic Haemagglutination-inhibition Tests By JENNIFER M. BEST AND J. E. BANATVALA Department Of Clinical Virology, St Thomas's Hospital and Medical School, London, S.E. t (Accepted 7 September ~97o) SUMMARY Kinetic haemagglutination-inhibition tests were used to investigate possible antigenic variation between four strains of rubella virus which had been isolated in different years and in different geographical areas. No significant differences were detected using hyperimmune rabbit sera and human sera obtained following both naturally acquired infection and vaccination with an attenuated rubella vaccine. These results suggest that major antigenic variation is unlikely to constitute a problem with rubella. INTRODUCTION Laboratory tests have confirmed that reinfection may occasionally occur following both vaccination and natural infection by rubella virus (Horstmann, Pajot & Liebhaber, ~969; Wilkins et al. ~969; Strannegfird et al. I97O). Further, infants congenitally infected by recent epidemics of rubella have shown a wider range of anomalies (American Pediatric Society and Society for Pediatric Research, I965). It is therefore important to determine whether significant antigenic variation occurs amongst different strains of rubella virus. Attenuated rubella vaccines have recently been developed, but the detection of any marked degree of antigenic variation might affect their future composition. We previously reported that no significant antigenic variation could be detected employing standard cross haemagglutination-inhibition (HI) tests amongst eight different strains of rubella virus which were isolated in different geographical areas and at different times (Banatvala & Best, I969). Similar tests conducted by Casals 0964) with Eastern equine encephalitis (EEE) and by Young & Johnson (t969) with Venezuelan equine encephalitis (VEE) viruses also failed to show any significant degree of antigenic variation. However, when these workers employed kinetic haemagglutination-inhibition tests, significant degrees of antigenic variation closely related to the place of origin of these viruses were detected. In these tests, serial dilutions of both sera and antigens were incubated for varying intervals of time in order to determine whether titration of antigens in the presence of high dilutions of serum might reveal sharper differences between strains than those observed in standard HI tests. Since rubella virus has been shown to share certain morphological and physical characteristics with some arboviruses (Holmes & Warburton, I967; Holmes, Wark & Warburton, I969) we employed kinetic tests to determine whether any degree of antigenic variation could be detected amongst four strains of rubella virus which had different passage histories in cell culture and which were originally isolated in different geographical areas. I5-2

216 J. M. BEST AND J. E. BANATVALA METHODS Antigens. Haemagglutinating antigen (HA) for each of the four strains selected (Table I) was produced in BHK2I cell cultures according to the method described by Banatvala & Best 0969). HA titres varied from I/I6 to I/Z56. Since antigen titres of at least 1/32 were required for the kinetic test, cell culture fluid was subsequently treated with ether + Tween 80 (Norrby, ~ 962). This procedure increased antigen titres approximately fourfold. Antigens were stored at -7 o0 until required. Table I. Derivation and passage history of rubella virus strains Used for inoculation Strain Derivation of rabbits GIGUERE 1964; New Haven, U.S.A. ; rubella syndrome VMK (2), RKI3 (7) Hvv-77 I961; U.S.A.; adult with VMK (67), PMK (4), rubella RKI3 (2) Attenuated vaccine strain JUDITH I963; Liverpool, U.K.; child with rubella VMK (3), RKla (55) Ko-1 I967; Kochi, Japan; child with rubella VMK (2), BHK (3), RK 13 (4), Vero (1) Passage history of virus* Used as HA VMK (2), RKI3 (5), BHK (4) VMK (77), PMK (4), BHK (3) VMK (3), RKI3 (5o), BHK (9) VMK (2), RKI3 (4), BHK (5) * VMK = vervet-monkey kidney; PMK = patas-monkey kidney; RKI3 = continuous line of rabbit kidney; BHK2I = continuous line of baby hamster kidney; Vero = continuous line of vervet-monkey kidney; ( ) = passage number. Sera Rabbit antisera. The preparation of strain-specific hyperimmune rabbit sera has been described previously (Banatvala & Best, 1969) (table I). In most tests, the earliest available serum samples were used, the rabbits being bled some 2 to 3 weeks after the first intravenous inoculation. However, in some tests, antigens were tested against antisera which had been obtained from the same rabbits at intervals of up to ~o weeks after the first inoculation. Since a I/~6 dilution of most sera contained agglutinins for pigeon erythrocytes even after adsorption with these cells, any serum with an antibody titre of < I/~6 when tested using a virus-serum reaction time of 4 rain., was unsuitable for use in those kinetic HI tests employing this reaction time. Human sera. Serial serum samples were obtained from two patients following naturally acquired infection in London during I969. Serial serum samples were also obtained from one volunteer between 29 and 84 days after administration of an attenuated rubella virus vaccine 0qvv. DE5), but only the 84 day sample had a sufficiently high antibody titre for use in the kinetic HI test. The earliest rabbit and human sera used were tested for rubella-specific IgM by serum fractionation on sucrose density gradients (Best, Banatvala & Watson, ~ 969). Pre-treatment ofsera. Serum inhibitors and agglutinins were removed by pretreatment with heparin and manganous chloride and adsorption with pigeon erythrocytes as described by Plotkin, Bechtel & Sedwick (I968) and Cooper et al. (I969). However, in our studies their procedure was reversed, adsorption with pigeon erythrocytes preceding beparin-manganous chloride treatment, since this method was found to be more effective in reducing high levels

Rubella strain variation by kinetie HI 217 of pigeon cell agglutinins. No change in antibody titre was noted when pretreated sera were stored at -2o either before or after dilution with sterile diluent. Reagents. Tests were carried out using microtitre equipment and disposable 'V' plates (Linbro Chemical Co. Inc., New Haven, Connecticut, U.S.A.). HI diluent, ph 6-2, consisted of veronal-buffered saline (Oxoid Ltd, London, S.E. I), to which o.i ~ CaC12, ~.o8 ~o MgSO~. 7H20 and o.2 ~ bovine plasma albumin was added. As high antigen titres were required for kinetic HI tests, this diluent was used since it produced two- to fourfold higher antigen titres than other diluents recommended for the rubella HI test. The optimum ph for rubella haemagglutination (HA) has been shown to be 6.2 (Halonen, Ryan & Stewart, I967), but since Casals (~964) found that the ph optima for HA and HI were not necessarily the same for different strains of EEE, the four strains of rubella virus were tested at four different phs between 6.o and 6.6. However, no significant difference in HA titre or number of HA units inhibited was observed and all kinetic HI tests were carried out at ph 6.2. Pigeon erythrocytes (o.33 ~) were used since we found that they gave approximately fourfold higher antigen titres than day-old-chick erythrocytes. Kinetic HI tests. Serum antibody titres were determined by titration against 4 to 8 units of antigen, using a reaction time of 4 rain. The approximate HA titre of each antigen was determined on the day before the test. Antigens were then diluted to approximately twice the HA titre required, i.e. to 64 HA units/o.o25 ml. and held at 4 overnight, since overnight storage reduced fluctuations in HA titre which frequently occurred when antigens were retested on the day they were thawed. Antigens were titrated in duplicate on the day of the test and then diluted to 32 HA units/o.o25 ml. Duplicate antigen titrations were also set up in parallel with each test. A modification of the test described by Young & Johnson (~969) was employed for these studies. Four twofold dilutions of each antigen and at least six twofold dilutions of serum were prepared in tubes, taking the initial serum dilution as the dilution below the titre of the serum which had been determined in the preliminary test. A 'chessboard' was prepared by distributing o.o25 ml. of each serum dilution in the plates, followed by o'oz5 ml. of each antigen dilution. Each serum was tested simultaneously against all four antigens in triplicate. The following reaction times were employed; 4 rain., 3o rain., 2 hr and 17 hr, after which o.o25 ml. of o'33 ~ pigeon erythrocytes was added to each well. Plates were incubated at 4 for I hr and were then left for 3o min. at room temperature before the number of HA units inhibited by each serum dilution was recorded. The number of HA units inhibited was taken as the lowest dilution of antigen producing no agglutination with each serum dilution. Effect of different reaction times on HI titres RESULTS The first available serum from each rabbit was used in these tests. Table 2 shows the number of HA units inhibited at different reaction times by twofold dilutions of the four strain-specific rabbit antisera. Although the extent of inhibition increased with time, there was no significant difference (~ fourfold) in the number of HA units inhibited by any single dilution of antiserum at any of the reaction times, indicating that there was no difference in the rate of inhibition among the different virus strains. Since Casals (I964) showed that with short reaction times homologous reactions appeared first and were of such specificity that strains could be easily separated, whereas with longer reaction times cross-reactions developed, we selected the shortest possible reaction time (4 rain.) for subsequent tests.

218 J. M. BEST AND J. E. BANATVALA I I ~" v i i ~ '~ I I- w I I ~ ~ I [ ~oooo V I I "- v 1%~ ii ~ ~ o 11 ~' v i ~'~ I1~ ~ t t ~ -" '~ '~ II "'vv V V \ o bl V /A V V I- v If " vvv -~ o ~ II ~ vv i ~ V,~,IO [ I/A vv ~ " " vvv ~ ~ V [ ~ V o < ~ ~ VV V o ~o ~ i ~ v v v v "vvv vvvv I ~ vvvv 0 " " vvvv ~ v v v ~ vvvv [ ~ vvvv ~ VV ~4 VV ~ VV ~ V V V V e ~ "'V VV IVV VV VV ~ooo - ~ o - ~ o o,. ~ o o 0 [,.

(a) Using rabbit antisera Rubella strain variation by kinetic HI 219 Further tests using 4 rain. reaction time Because the earliest available rabbit antisera had relatively low antibody titres and sharp differences between strains are more likely to be detected if high serum dilutions are used, two additional strain-specific rabbit antisera with higher antibody titres obtained from later bleedings of each rabbit were also treated. The results are shown in Table 3 and are expressed as the ratio Ho/HT where H 0 denotes the number of homologous antigenic units inhibited and HT the number of heterologous antigenic units inhibited. The homologous antigenic reaction is defined as unity and increasing or decreasing ratios therefore indicate the degree of antigenic difference between strains. These results show that the four strains of rubella virus are antigenically similar, since no result falls outside the limits of o'25 and 4, which would indicate a significant fourfold difference between strains. The inhibition curves shown in Fig. I also illustrate the similarity between these strains. Table 3- Comparison of rubella virus strains using rabbit antisera obtained at different intervals after the first intravenous inoculation Antigens z. Antiserum Weeks* GIGUERE HPV JUDITH KO-I GIGUERE 2½ + 1'05 I"5--2"O I'5 I "5 5{- I-o o-7-i'o I-O o'7-i'3 10 1"0 0"5 I "0 0" 5 HPV 2 I "3 I "0 I "0--2'0 2"0 4 1.3 I.O I.O 2.0 10 0"5--2"0 I'0 0"5--2"0 0" 5 I'0 JUDITH 3 f O" 7 I" 5 I'O I' 5 8 o'7-i.o 2.0 I.o 2'0-3.0 Io o'7-i "3 o'5-i "o I.o 2.0 KO-I 3 I'5 2"O I'5 I'O 4 0'7--I "3 2"O 2"0--3"0 I "O 6 0"5 O"5--1 "O 2"0 I "0 (b) Using human sera * = Weeks after first intravenous inoculation. t = Sera in which rubella-specific IgM was detected. ~. = Ho/ HzHA units inhibited. Table 4 shows that there were no significant differences in the number of HA units inhibited when early and late convalescent sera from two patients (D.W. and V.C.) with naturally acquired rubella were tested. Both early convalescent sera contained rubellaspecific IgM. Similar results were obtained using the post-vaccination serum EA. Thus, no variation between the four strains of rubella virus could be detected when tests were conducted with sera containing antibodies resulting from both naturally acquired recent infection and following vaccination with the HeV. DE 5 attenuated rubella vaccine. General considerations These studies also illustrated several points of value for rubella HI tests: (1) Since no differences in the inhibiting activity of four different strains were detected, it is likely that most strains of rubella virus can be used for the preparation of antigen for HI tests. (2) Fluctuations in the titres of ether +Tween-treated antigens, which may perhaps be caused by

220 J. M. BEST AND J. E. BANATVALA 32 16 '~' ' ' (a)' -- I I I I (b) 2.o 7-0 \. 256 512 1024 2048 4096 I 64 128 256 512 ' 102' "~ 32 < "I" 16 (c) I I I I (d) I 64 128 256 512 1024 Reciprocal serum dilutions 64 128 256 512 I 1024 Fig. i. Comparison of four strains of rubella virus by the kinetic HI test, employing rabbit antisera obtained between 6 and Io weeks after the first inoculation of rubella virus and a reaction time of 4 min. (a) GI~UERE serum, (b) HVV serum, (e) JUDITH serum, (d) KO- I serum. GIGUERE antigen, 0--0; HPV antigen, ; JUDITH antigen, m; KO-I antigen,. Table 4. HA units inhibited by human sera t Sera A Reciprocal Patient Days* dilution Naturally acquired DW 8t 64 antibody 128 24 I28 256 42 64 I28 VC 7t 32 64 ]4 32 64 HPv.DE5 vaccine- EA 84 32 induced antibody 64 I28 * Days following onset of symptoms or after vaccination. t Sera in which rubella-specific IgM was detected. t Antigens GIGUERE HPV JUDITH KO- I 3z 24 24 16 8 6 6 4 32 ~ 24 20 16 8 I2 8 8 32 24 48 Z4 8 12 12 8 16 I6 I6 32 8 8 6 8 24 I6 I6 32 8 6 6 8 32 ~ 24 32 ~ 48 i6 i2 i2 24 6 6 4 6

Rubella strain variation by k&etic HI 221 virus aggregation, can be eliminated by overnight storage at 4. (3) Although inhibition of haemagglutination can be detected after as short a reaction time as 4 min., this procedure would be unsuitable for screening sera, since antibody titres are approximately fourfold lower than after a reaction time of I hr and therefore low levels of antibody might be missed. (4) The HI diluent used in these tests was easy to prepare and produced high HA titres. HA and HI titres were stable over a ph range of 6-0 to 6.6. (5) Pigeon erythrocytes produced higher HA titres than day-old-chick erythrocytes, but the high titres of pigeon agglutinins present in some sera may preclude their use for tests in which it is hoped to detect low levels of serum antibodies. (6) It was found that sera can be stored at -2o for at least 2 months after heparin and manganous chloride treatment without change in antibody titre. DISCUSSION Different strains of rubella virus have been shown to produce variable plaque morphology and cytopathic effect in RKI 3 cell cultures (Morgan, 1969; Fogel & Plotkin, 1969). Although Morgan (1969) and Lawrence & Gould (I969) employed strains of varying passage history, including recently isolated low passage strains, and detected some variation in plaque morphology, they concluded that this could not be correlated with passage history or degree of virus attenuation. Fogel & Plotkin (I969) showed that plaque morphology depended largely on the sensitivity of different virus strains to inhibitors in the agar overlay and to the ph of the medium during plaque formation. They were able to enhance differences in neutralization between homologous and heterologous viruses by adsorbing sera with cobalt chloride-precipitated antigens and in preliminary experiments were able to demonstrate some variation between the RA 27/3 vaccine strain of rubella virus and a high passage laboratory strain, WEST POINT. Whether these differences reflected variation amongst the isolates from which these strains were originally derived or were due to their subsequent passage histories could not be determined. Oxford (1969), employing microplaque-inhibition tests, detected slight variation between a low passage and a high passage strain of virus, but did not consider the degree of variation to be significant (J. S. Oxford, personal communication). The four strains we studied were isolated in different geographical areas in different years and also exhibited somewhat varying biological properties. Thus, JUDITH was originally isolated from a child with uncomplicated rubella in Liverpool, England, in I963 and has had over 50 passes in cell cultures, mostly in RKt3. GIGUERE was isolated from an infant in New Haven, Connecticut, with severe manifestations of congenitally acquired rubella following the extensive epidemic in the U.S.A. in I964. This epidemic was followed by many reports of unusually severe manifestations of congenitally acquired rubella, together with the frequent appearance of certain clinical features which had only rarely been observed previously (American Pediatric Society and Society for Pediatric Research, 1965). KO-1 was isolated in Japan in I967. Kono et al. (I969) have recently presented evidence to suggest that Japanese strains of rubella virus may be less tetratogenic than those isolated in the U.S.A. and Europe since, although infection by rubella virus occurs commonly in Japan and the proportion of women of child-bearing age who are susceptible differs little from that found in Western countries, the incidence of congenitally acquired rubella is considerably lower than in the U.S.A. Preliminary experiments conducted in rabbits using both Japanese and U.S. virus strains have shown that the Japanese virus strains did not infect or damage the foetus, whereas strains isolated in the U.S.A. caused both foetal infection and damage (Kono et al. 1969). The HPV-77 strain of virus was the first attenuated vaccine strain to be

222 J.M. BEST AND J. E. BANATVALA developed (Parkman et al. I966). Vaccination with this strain results in the development of serum antibodies in the absence of disease. Although these strains differed in their biological behaviour, we could detect no antigenic variation by kinetic HI tests using both hyperimmune rabbit and early convalescent human antisera. However, our tests were conducted with antigens which had been treated with ether and Tween since it was not possible to obtain adequate titres of haemagglutinin with untreated virus. Although it is possible that differences might have been detected if untreated virus had been used as antigen, Young & Johnson 0969) found that the sensitivity of their kinetic HI tests was not changed by use of ether+tween-treated VEE antigen. The sensitivity of tests to detect antigenic variation is likely to be greatest when sera are obtained as early as possible from animals which have received a minimum number of inoculations. Although multiple inoculations were necessary before satisfactory haemagglutination-inhibition titres were obtained in rabbit sera, two of the four early rabbit antisera and both early human antisera tested contained rubella-specific IgM, the presence of this class of immunoglobulin being characteristic of sera obtained shortly after a primary antigenic stimulus (Uhr & Finkelstein, ~963; Svehag & Mandel, 1964; Schluederberg, I965; Best et al. I969). It has also been shown that antisera to certain arboviruses prepared in different animals and using different vaccination schedules may vary in their specificity (Saturno & Henderson, I965; Young & Johnson, I969). Since we only tested rabbit and human antisera against four different strains we cannot be absolutely certain that there is no variation between strains of rubella virus. However, these results are encouraging in that they suggest that major antigenic variation is unlikely to constitute a problem with rubella, and that therefore antibody produced by any one of a number of vaccine strains is likely to confer adequate protection against naturally encountered strains of rubella virus. We wish to thank Dr K. Johnson (National Institutes of Health, Middle America Research Unit, Balboa Heights, Canal Zone, Panama) for his advice; Dr R. Kono (National Institute of Health, Tokyo) for providing the Japanese rubella virus strains Ko-i ; Dr J. A. Dudgeon (Institute of Child Health, London) for providing postvaccination serum samples; the Wellcome Research Laboratories, Beckenham, for providing pigeon erythrocytes, and Mrs J. Uren and Mrs W. Keating for technical assistance. This work was supported by a grant from the Medical Research Council. REFERENCES AMERICAN PEDIATRIC SOCIETY AND SOCIETY FOR PEDIATRIC RESEARCH (t965). Rubella Symposium. American Journal of Disease of Children xxo, 349- BANATVALA, J. E. & eest, J. M. (I969). Cross-serological testing of rubella virus strains. Lancet i, 695. BEST, J. M., BANATVALA, J. E. & WATSON, D. 0969). Serum IgM and IgG responses in postnatally acquired rubella. Lancet ii, 65. CASALS, J. (I964). Antigenic variants of eastern equine encephalitis virus. Journal of Experimental Medicine IX9, 547. COOPER, L. Z., MATTERS, B., ROSENBLUM, J. K. & KRtJGMAN, S. (I969). Experience with a modified rubella hemagglutination-inhibition antibody test. Journal of the American Medical Association zo7, 89. rogel, A. & PLOTKIN, S. (I969). Markers of rubella virus strains in RKI3 cell culture. Journal of Virology 3, I57. a~alonen, P. E., RVAN, J. M. & STEWART, J. A. (I967). Rubella haemagglutinin prepared with alkaline extraction of virus grown in suspension culture of BHK2I cells. Proceedings of the Society of Experimental Biology and Medicine IZS, I62. HOLMES, I. H. & WARBURTON, M. F. (I967). IS rubella an arbovirus? Lancet ii, I233. ~OLMES, I. H., WARI~, M. C. & WARBURTON, M. r. (I969). Is rubella an arbovirus? II. Ultrastructural morphology and development. Virology 37, I5.

Rubella strain variation by kinetic HI 223 HORSTMANN, D. M., PAJOT, T. G. & LIEBI-IABER, H. 0969). Epidemiology of rubella; subclinical infection and occurrence of reinfection. American Journal of Diseases of Children HS, 133. KONO, R., HIBI, M., HAYAKAWA, Y. & ISHII, K. (I969). Fxperimental vertical transmission of rubella virus in rabbits. Lancet i, 343. LAWRENCE, G. D. & GOULD, J. (I969). Morphology of rubella plaques in RKI3 cultures. International Symposium on Rubella Vaccines, London I968; Symposia Series" in Immunobiological Standardization xi, 177. Basel/New York: Karger. MORGAN, J. R. (1969). The use of plaque methods for strain comparisons. International Symposium on Rubella Vaccines, London 1968. Symposia Series in lmmunobiologieal Standardization H, 173. Basel New York: Karger. NORRBY, E. C. J. 0962). Hemagglutination by measles virus. IV. A simple procedure for production of high potency antigens for hemagglutination-inhibition tests. Proceedings of the Society of Experimental Biology and Medicine xxx, 814. OXFORD, J. S. (1969). A comparison of some 'wild' and 'attenuated' strains of rubella virus. International Symposium on Rubella Vaccines, London I968. Symposia Series in Immunobiological Standardization H, 181. Basel/New York: Karger. VARKMAN, V. D., MEYER, H. M., KIRSCHSTEIN, R. L. & HOPPS, H. E. (I966). Attenuated rubella virus. I. Development and laboratory characterization. New England Journal of Medicine 275, 569. VLOTKIN, S. A., BECnXEL, D. J. & SEDWICK, W. D. (1968). A simple method for removal of rubella haemagglutination inhibitors from serum adaptable to fingertip blood. American Journal of Epidemiology 88, 3oi. SATURNO, A. & ttenderson, J. R. (I965). Factors influencing determination of antigenic relationships between western equine encephalomyelitis virus strains. Journal of Immunology 94, 365 SCHLUEDERaER6, A. (I965). Immune globulins in human viral infections. Nature, London 205, I232. STRANNEG~,RD, O., HOLM, S. E., ttermodsson, S., NORRBY, R. & LYCKE, E. (1970). Case of apparent reinfection with rubella. Lancet i, 240. SVEHAG, S. E. & MANDEL, B. (1964). The formation and properties of poliovirus-neutralising antibodies. I. I9S and 7s formation. Differences in kinetics and antigen dose requirements for induction. Journal of Experimental Medicine xx9, 1. UHR, 1. W. & FINKELSTEIN, M. S. (1963). Antibody formation. IV. Formation of rapidly and slowly sedimenting antibodies and immunological memory to bacteriophage ~bx 174. Journal of Experimental Medicine x I7, 457. WmKINS, J., LEEDOM, S. M., PORTNOY, B. & SALVA'rORE, A. A. (1969). Reinfection with rubella virus despite live vaccine induced immunity. American Journal of Diseases of Children Ix8, 275. YOUNG, N.A. & JOHNSON, K. M. 0969). Antigenic variants of Venezuelan equine encephalitis virus: their geographic distribution and epidemiologic significance. American Journal of Epidemiology 89, 286. (Received 15 June 197o)