- Email: [email protected]
Single-radial-immunodiffusion as an in vitro potency assay for human inactivated viral vaccines
Veterinary Microbiology, 37 (1993) 253-262 Elsevier Science Publishers B.V., Amsterdam
253
Single-radial-immunodiffusion as an in vitro potency assay for human inactivated viral vaccines M.S. Williams Laboratory of Respiratory Viruses, Division of Virology, Centerfor BiologicsEvaluation and Research, Food and DrugAdministration, Bethesda, MD, USA (Accepted 14 July 1993 )
ABSTRACT Single-radial-immunodiffusion(SRID) assays have been used to determine the potency of all human inactivated influenza virus vaccines licensed by the Food and Drug Administration for use in the United States since 1978. SRID replaced less reliable tests which were based on the aggregation of erythrocytes by the hemagglutinins of influenza viruses. Similar SRID assays have been used experimentally to determine the potency of inactivated polio and rabies vaccines. In each case, the assays are based on the diffusion of viral antigen into an agarose gel containing specific antibodies to the antigen being measured. For influenza and rabies, disruption of the virions with a detergent is necessary to permit the diffusion of the appropriate antigens, where as with polio, intact virions are allowed to diffuse. The interaction between antigen and antibody produces a zone of precipitation whose size is directly proportional to the amount of antigen applied. A potency value for unknowns is obtained by comparing the sizes of zones produced by unknown preparations to the sizes of zones obtained with a calibrated reference of known antigen content. Once the specific reference antigens and antibodies are prepared and the test standardized, it is a remarkably simple technique which unlike agglutination assays is very reproducible, relatively unaffected by minor variations in test conditions and is far less time consuming and cumbersome than in vivo assays for potency such as those done by inoculatingmice or monkeys. More importantly, clinical studies demonstrate that standardization of influenza vaccines by SRID provides a better correlate of human immunogenicity than previous methods.
INTRODUCTION
The severity of disease caused by viruses such as rabies and polio is well known by the scientific and medical communities as well as the general public, but the severity of disease which can be caused by influenza viruses is less well understood. Influenza viruses cause considerable morbidity and mortalCorrespondence to: M.S. Williams, Laboratory of Respiratory Viruses, Division of Virology, Center for Biologics Evaluation and Research, Food and Drug Administration, 8800 Rockville Pike, Bethesda, MD 20892, USA.
2 54
M.S. WILLIAMS
ity in human populations, especially high-risk groups such as the elderly and persons with underlying health problems. During major epidemics, hospitalization rates for high-risk persons may increase 2- to 5-fold. Previously healthy children and young adults may also require hospitalization due to influenza or influenza-related complications, although the relative increase in their hospitalization rates is less than for persons in high-risk populations. Increased mortality results not only from influenza and pneumonia but also from cardiopulmonary and other chronic diseases that can be exacerbated by influenza infection. At least 10000 excess deaths have been documented in each of 19 different epidemics in the United States during the last 30 years, and more than 40000 excess deaths occurred in each of three of these epidemics. Approximately 80-90% of these excess deaths were among persons greater than 64 years of age. (ACIP, 1991 ) The proportion of elderly persons in the United States population is increasing. The numbers of younger persons at increased risk for influenza-related complications is also increasing due a number of factors, such as better survival rates for organ transplant recipients and diseases such as cystic fibrosis and acquired immunodeficiency syndrome, and the success of neonatal intensive care units. Therefore, the need for safe and effective vaccines for protection against influenza remains a very important public health issue for the present as well as the future (ACIP, 1991 ). There has been extensive experience in the last 15 years in testing human inactivated viral vaccines by single-radial-immunodiffusion (SRID). Tests for the potency of influenza, rabies and polio vaccines have been developed using similar methods. All three require the production of antibodies to a purified preparation of a target protein antigen from the virus, and the standardization of a reference antigen. The production and standardization of these reference reagents can be time consuming and costly. The test itself is, however, simple to perform and produces accurate and reproducible results. Optimal results may require modifications of the test for different antigens. For example, influenza viruses are disrupted with the detergent Zwittergent 3-14 and rabies uses the detergent Emulphogene BC-720. The SRID for polio viruses utilizes an autoradiographic technique causing zone enhancement (ZE), which is necessary to measure the low levels of antigen present in final vaccines. The SRID assays for these viruses have shown advantages in some ways over other types of tests, but are in different stages of acceptance for official tests for release of lots of vaccines. The SRID assay used to establish the potency of influenza virus vaccines has been used routinely by manufacturers and regulatory agencies for many years, and the description of this test serves as a model for the development of similar techniques. PREPARATION OF REAGENTS
The performance of SRID assays requires a specific antiserum to the purified antigen to be measured. For influenza viruses, the antigen of most im-
SINGLE-RADIAL-IMMUNODIFFUSION
255
portance is the hemagglutinin (HA), which is one of the two viral surface glycoproteins. Embryonated chicken eggs, 9 to 11 days old, are inoculated via the allantoic sac, incubated at 33 ° C to 35 ° C for 3 days, and the allantoic fluid harvested. A purified whole virus preparation is obtained by centrifugation through a sucrose gradient. One portion of the purified virus is used to prepare HA for the immunization of sheep. The HA is removed from the viral core by the method of Brand and Skehel (1972) using the enzyme bromelain. The method yields HA which is missing the hydrophobic membrane anchor and destroys the enzymatic and antigenic properties of the other surface antigen, neuraminidase (NA), in the process. A total of approximately 100 #g of purified HA is inoculated intramuscularly and intradermally into sheep (or other suitable animals) using Freund's Complete Adjuvant. Three weeks after the first injection, 2 to 4 booster doses of 20 to 30/tg in Freund's Incomplete Adjuvant are given at weekly intervals by the same routes of inoculation, and continued until antibody levels, which are tested by SRID before each subsequent immunization, have reached a desired working range. For influenza viruses, this range is about 1 to 10/~1 of antiserum per ml ofagarose. Antibody levels in this range should produce distinct and measurable zone diameters of about 8 to 10 mm for antigen preparations containing 30/tg of HA per ml. A reference standard of known antigen content is necessary to determine the potency of an unknown. The calibration of this reference may be accomplished in several ways, all of which rely on a protein determination. One method is to solubilize the HA from another portion of the whole virus preparation using the detergent Nonidet P-40 (NP-40). NP-40 solubilization leaves the membrane anchor region of the HA intact. After solubilization, the HA and NA antigens are separated by centrifugation through a sucrose gradient. Since NA and HA migrate similarly through the gradient, HA prepared in this manner contains some NA. The amount of NA present is about 10% of the total H A / N A (HANA) preparation, and it does not interfere with the standardization process. A protein value is determined for the HANA (minus 10% for the NA) and this material is used as a primary standard in SRID tests to assign a value to a lyophilized, formalin inactivated, whole influenza virus vaccine preparation. The latter lyophilized virus preparation is then used as the reference antigen for the actual potency testing of influenza virus vaccines which are antigenically identical to the reference standard. As a means of determining the accuracy of this standardization process, gg of HA protein by SRID and Ftg of total protein are determined on the whole virus preparation before lyophilization. Based on our experience over the years with many different strains of influenza viruses, the amount of HA protein should be about 30 to 50% of the total protein. Both of the purified HA preparations, due to their different properties, serve a unique role in the preparation of reagents. The bromelain material, because
2 56
M.S. WILLIAMS
it destroys NA, yields a highly purified HA preparation and therefore produces antibodies which recognize only the HA. However, since a substantial portion of the HA remains in the viral core, and standardization is based on total protein values, it does not serve well as a primary standard. The NP-40 solubilized HA remains whole and although it contains a small amount of NA making it unsuitable for producing antibodies, the NA can be deducted from calculations when used as a primary standard. Standardization of these reagents relies on protein values, and it is important to establish a protein assay which is reliable, reproducible, and free from interference by chemicals in the different preparations. For instance, the Lowery protein, commonly used in many laboratories, is subject to interference by detergents and is not suitable for measuring the protein content of the NP-40 material. Kjeldahl, amino acid analysis, and an automated o-phthalaldehyde fluorometric assay (Benson & Hare, 1975 ) have all been used successfully in our laboratory. The Kjeldahl measures nitrogen and care must be taken to avoid testing of samples with chemicals containing nitrogen such as tris buffers. The purity of the various preparations described above is checked by polypeptide analysis according to the method of Laemmli (1970). Samples are electrophoresed through 10% polyacrylamide gels. The gels are stained with Coomassie Brilliant Blue R-250, and analyzed to ensure that the preparations are free from contaminating proteins. P E R F O R M A N C E O F T H E TEST
The following procedure is a brief description of how SRID assays are performed for influenza vaccines and is a modification of the procedures described by Wood et al. ( 1977 ) and Williams et al. (1980). The first step in the performance of the test is the preparation of agarose gels. The amount of antiserum necessary to perform the test on samples at a given antigen concentration is determined in a set of preliminary experiments and the optimal amount is added to a 1% solution of agarose melted in phosphate buffered saline (PBS). Care should be taken to ensure that the agarose has been completely melted to allow for even diffusion and staining. The temperature of the solution must be lowered to 55 °C or below before addition of the antiserum or destruction of antibodies may occur. The agarose/antiserum mixture is injected slowly and steadily into a 240 × 80 X 2 m m casting plate, and placed at 4 °C for at least 5 rain to allow complete gelation. In order to prepare the gels for sample inoculation, wells, 4 m m in diameter, are punched in a 4 by 12 row configuration. This configuration allows for the inoculation of 1 dilution series of the reference and 5 dilution series of unknowns, each in duplicate. Samples of both reference and unknown antigens are pre-diluted to the de-
SINGLE-RADIAL-IMMUNODIFFUSION
257
sired testing range of approximately 30 ~g HA/ml in PBS and mixed at a 1:10 ratio with a 10% stock solution of the detergent Zwittergent 3-14 to give a final detergent concentration of 1%. The antigen/detergent mixture is incubated at room temperature for approximately 10 min. Samples are then diluted in PBS to obtain a dilution series of l: 1, 1:1.5, 1:2 and 1:4. Twenty/As of each dilution is inoculated into duplicate wells. The gel plates are placed in a moist chamber at room temperature for a minimum of 18 hours. Following the 18 h incubation, the gels are gently removed from the casting plates in a pan of tap water and floated onto a glass plate or a strip of Gel Bond. A moist piece of filter paper is placed on top of the gel and it is allowed to dry completely. The filter paper is removed and the dried gel is stained with Coomassie Brilliant Blue R-250 for l0 rain and destained as needed to visualize the zones of precipitation. The zone diameters are measured in two directions at right angles to each other to the nearest 0.1 m m using a microcomparator, a magnifying projector, or other suitable measuring device. Zones that are eliptical in shape should be measured across the long and the short axis. Potency determinations are made by comparing the zone sizes of the reference dilutions to those of the unknowns. For the potency of vaccines in the United States, calculations are performed by the parallel line bioassay method of Brownlee ( 1965 ) plotting log antigen dilution vs. log zone diameter. Another simpler calculation which is widely used in other parts of the world is the slope ratio method described by Wood et al. (1977). In this method, the slope of the vaccine curve is divided by the slope of the reference curve and multiplied by the potency value of the reference. The overall time required for performance of the test, including potency calculations, is approximately 24 hours. During the early development of SRID for influenza virus vaccines, the detergent used to disrupt the virus was sodium sarkosyl sulfate (SSS). We discovered that this detergent gave inconsistent results between different manufacturing processes and for some sub types of influenza actually yielded smaller precipitation zones when comparing SSS treated purified HA to purified HA with no detergent treatment. A large number of different detergents were studied for their ability to effectively disrupt the virion and produce precipitation zones of sufficient size and clarity (Table 1 ). Several non-ionic detergents, Triton X-100, Triton CF-54, Nonidet P-40, Octylglucoside, and Emulphogene BC-720 were found to give relatively large zone diameters as well as excellent quality of zone reaction as measured by intensity of staining and crispness of zone edges. The zwitterionic detergent Zwittergent 3-14 gave similar results to the non-ionic detergents listed above, but gave more consistent results when tested against all of the licensed influenza vaccine preparations.
2 58
M.S.WILLIAMS
TABLE 1 Comparison of detergents for the solubilization of influenza A H 1N 1 and influenza B antigens
Detergents
Source
Relative zone diameter size
Relative qualityof reaction
Supelco, Inc. Bellefonte, PA
Small Small
Poor Poor
Small
Good
Small
Good
Large Large Large Small Small Small Small Small Small Small Small Small Large
Good Good Good Good Good Good Good Good Good Poor Poor Poor Good Good
Large
Good
Anionic Triton Wr-1339 Sodium Taurocholate
Cationic Cetyltrimethylammonium Bromide Hyamine CL
Non-ionic Triton X- 100 ( TX- 100) Triton CF-54 Nonidet P-40 (NP-40) Emulphogene (BC-420) Emulphogene (BC-610)
Shell Oil Co., London GAF Corp., Linden N.J.
Emulphogene (BC-720) Emulphogene (BC-840) Emulphogene (BC-970)
Large
Lubrol WX Brij 35 Myrj 59 Tween 20 Tween 80
Supelco, Inc., Bellefonte, PA
Octylglucoside
Sigma Chemical Co., St. Louis, MO
Zwitterionic
Zwittergent 3-14
Calbiochem-Berhing Corp., La Jolla, CA
DISCUSSION
SRID has proven to be a successful replacement for other in vitro potency assays based on the agglutination of erythrocytes for influenza virus vaccines. The old potency assays were not reproducible within or between laboratories because they depended on the physical ability of the hemagglutinin to bind to red cells. The reproducibility of these tests was therefore susceptible to any condition which could affect that bond such as variation in pools of red cells, temperature, or vibrations. Different types of vaccines were also found to behave differently such that sub-unit or split virus preparations appeared to have a much higher antigen content than whole virus preparations. (Table 2 shows the manufacturers and the different types of influenza virus vaccines mar-
SINGLE-RADIAL-IMMUNODIFFUSION
259
TABLE 2 Influenza virus vaccine manufacturers, types and processes Manufacturer
Vaccine Type and Process
Connaught Laboratories, Inc.
1. whole virus vaccine, gradient purified 2. sub-unit vaccine, gradient purified, disrupted with Triton X-100 3. surface antigen vaccine, gradient purified, disrupted with Triton N- 101, and further purified to obtain a vaccine containing predominantly HA and NA 4. sub-unit vaccine, gradient purified, disrupted with Polysorbate 80 and extracted with ether 5. sub-unit vaccine, chromatographically purified, disrupted with tri (n) butylphosphate and Polysorbate 80
Lederle/Evans*
Parke-Davis, Inc. Wyeth-Ayerst, Inc.
*Marketed by Lederle Laboratories under a divided manufacturing agreement with Evans Medical Ltd.
keted in the United States at this time. ) Analysis of clinical studies clearly demonstrated that vaccines, regardless of their different manufacturing techniques, standardized by SRID correlated well with human immunogenicity and reactogenicity. In 1978, the year that SRID was first used as the potency test for influenza virus vaccines, large scale clinical studies were performed which provided information relating dose in #g of HA to immunologic responses and reaction rates in humans and formed the basis for the amount of antigen to be used in vaccines licensed for public use. That year also marked the return of the circulation of H 1N 1 viruses, so that clinical information was provided for populations of people who had been immunologically primed by exposure to H3N2 viruses, as well as a population of young people who were not immunologically primed to H 1N 1 viruses. The studies demonstrated that in adult, primed populations, whole virus and sub-unit vaccines standardized by SRID were quite comparable in human immunogenicity and reactogenicity. In younger, unprimed populations, sub-unit vaccines were found to be less reactive than whole virus vaccines and required two doses of vaccine to achieve protective levels of antibodies. The clinical information derived from these studies using vaccines whose potency was determined by SRID established a basis for how influenza virus vaccines are used and formulated today (Cate et al., 1983; La Montagne et al., 1983; Quinnan et al., 1983; Wright et al., 1983). The experience with SRID in potency testing of human rabies vaccines by
260
M.S. WILLIAMS
the methods of Ferguson and Schild ( 1982 ) and Fitzgerald and Needy (1986) is quite extensive, but still at an experimental level. In this vaccine, the protein which is measured by SRID is the envelope glycoprotein. Although SRID assays have several advantages over the currently used NIH mouse potency test as described by Seligman ( 1973 ) such as less timely and cumbersome to perform, more consistency from test to test and lot to lot, and they do not require challenge with live rabies virus, there is concern that the envelope glycoprotein may not be the only antigen which is important in conferring protection. Work by Dietzschold et al. (1987) has shown that the ribonucleoprotein also contributes to protection from lethal challenge. Some manufacturers are using SRID as an internal test in helping to formulate vaccine, but given the severe consequences of vaccine failure, more information will need to be obtained before SRID can replace the animal model for the purpose of releasing final vaccine to the public. SRID potency tests for inactivated poliovirus vaccines as described by Schild et al. (1980) have been used at various stages of production by some manufacturers. It has been shown to be a reliable method of determining the D antigen content, which is the component that produces neutralizing antibodies. However, because of the length of time involved in performing the ZE test (approximately 10 days) and because of a reluctance by some manufacturers to use radio isotopes in their facilities, SRID is not routinely used as a potency test for this vaccine. In process testing of vaccines during various manufacturing steps is usually performed utilizing an ELISA assay, and final vaccine testing still requires tests in animals. SRID tests can be highly strain specific, depending upon the properties of the antisera. This can be problematic as well as useful. For influenza virus vaccines, which currently contain three constantly changing subtypes, new antisera and reference antigens must be made and standardized each time a new strain is incorporated into the vaccine formula. There have been 21 different strains of influenza viruses in the vaccine formula in the last 15 years, requiring a large commitment of time and resources for the production of these reagents. This same strain specificity, however, also allows for the testing of individual strains in a multivalent vaccine, which was not possible with earlier assays. With poliovirus vaccines, this specificity can be used to distinguish between different strains within subtypes. CONCLUSION
SRID techniques have been successfully developed as potency tests for several types of human inactivated viral vaccines. It is a simple technique to perform and, in the case of influenza, has been shown to be more reliable and a better predictor of human immunogenicity than older in vitro assays and has therefore replaced those tests. For rabies vaccine, it is more reproducible,
SINGLE-RADIAL-IMMUNODIFFUSION
261
less time consuming and cumbersome, and safer than the in vivo N I H mouse potency test since personnel need not be exposed to infectious rabies virus. Studies are in progress to determine if SRID will replace the N I H test for rabies, or if the two can be used in conjunction to determine the immunogenicity o f vaccine lots. For inactivated poliovirus vaccine, practical considerations have kept SRID from becoming the test o f choice for vaccine potency testing. ACKNOWLEDGEMENTS I acknowledge the personal communications o f Dr. Ronald Lundquist from the Division o f Virology, Center for Biologics Evaluation and Research for information on poliovirus vaccines and Ms. Nancy Roscioli from the Division of Product Quality Control, Center for Biologics Evaluation and Research for information on rabies vaccine.
REFERENCES Benson, J.R. and Hare, P.E., 1975. o-Phthalaldehyde:fluorogenicdetection of primary amines in the picomolerange. Comparison with fluorescainineand ninhydrin. Proc. Nat. Acad. Sci., 72:619-662. Brady, M.I. and Furminger, I.G.S., 1976. A surface antigen influenzavaccine. I. Purification of haemagglutininand neuraminidase proteins. J. Hyg., 77:161-172. Brand, C.M. and Skehel, J.J., 1972. Crystallineantigen from the influenza virus envelope. Nature, New biology, 238: 145-147. Brownlee, K.A., 1965. In: Statistical Theory and methodology in Science and Engineering., Chapter I 1. New York: John WileyPublishers. Cate, T.R., Couch, R.B., Parker, D., and Baxter, B., 1983. Reactogenicity,immunogenicity,and antibody persistence in adults given inactivated influenza virus vaccines- 1978. Rev. Infect. Dis., 5, No. 4: 737-747. Dietzschold, B., Honghai, W., Rupprecht, C.E., Esteban, C., Tollis, M., Ertl, H., Heber-Katz, E. and Koprowski,H., 1987. Induction of protective immunity against rabies by immunization with rabies virus ribonucleoprotein. Proc. Nat. Acad. of Sci., 84:9165-9169 Ennis, F.A., Mayner, R.E., Barry, D.W., Manischewitz,H.E., Dunlap, R.C., Verbonitz, M.W., Bozeman, F.M. and Schild, G.C., 1977. Correlation of laboratory studies with clinical responses to A/New Jersey influenza vaccines. J. Infect. Dis., Supplement 136: $397-$406. Ferguson, M. and Schild, G.C., 1982. A single-radial-immunodiffusiontechnique for the assay of rabies glycoprotein antigen: Application for potency tests of vaccines against rabies. J. Gen. Virol., 59: 197-201. Fitzgerald, E.A. and Needy, C.F., 1986. Use of the single radial immunodiffusion test as a replacement for the NIH mouse potency test for rabies vaccine. In: IABS Symposiumon Reduction of Animal Usage in the Development and Control of Biological Products, London, United Kingdom, 1985. Dev. Biol. Stand., 64: 73-79. Basel: S. Karger. La Montagne, J.R., Noble, G.R., Quinnan, G.V., Curlin, G.T., Blackwelder,W.C., Smith, J.I., Ennis, F.A. and Bozeman, F.M., 1983. Summary of clinical trials of inactivated influenza vaccine- 1978. Rev. Infect. Dis., 5: 723-736.
262
M.S.WILLIAMS
Laemmli, U.K., 1970. Cleavage of structural proteins during the assembly of the lead bacteriophage T4. Nature (London), 227: 680-685. Mayner, R.E., Blackburn, R.J. and Barry, D.W., 1977. Quantitation of influenza vaccine hemagglutinin by immunoelectrophoresis. In: International symposium on Influenza Immunization (II), Geneva. Dev. Biol. Stand., 39:169-178. Basel: S. Karger. Prevention and control of influenza. Recommendations of the Immunization Practices Advisory Committee (ACIP). M M W R 1991, 40 RR-6: 1-15. Quinnan, G.V., Schooley, R., Dolin, R., Ennis, F.A., Gross, P. and Gwaltney, J.M., 1983. Serologic responses and systemic reactions in adults after vaccination with monovalent A/USSR/ 77 and trivalent A/USSR/77, A/Texas/77,B/Hong Kong/72 influenza vaccines. Rev. Infect. Dis., 5: 748-757. Schild, G.C., 1974. In: Proc., International Conference on Standardization of Diagnostic Materials, June 1973, in Atlanta, Georgia, U.S.A., pp. 243-248. Schild, G.C., 1977. The reference materials for the standardization of influenza vaccines. In: World Health Organization WHO/B5/77, 1147. Schild, G.C., Wood, J.M., Minor, P.D., Dandawate, C.N. and Magrath, D.I., 1980. Immunoassay of poliovirus antigens by single-radial-diffusion: Development and Characteristics of a sensitive autoradiographic zone size enhancement (ZE) technique. J. Gen. Virol., 51: 157170.
Seligman, E.B., Jr., 1973. The NIH test for potency. In: Laboratory Techniques in Rabies (WHO), 3rd edition, chapter 33, 279-285. Geneva: WHO. Williams, M.S., Mayner, R.E., Daniel, N.J., Phelan, M.A., Rastogi, S.C., Bozeman, F.M., and Ennis, F.A., 1980. New developments in the measurement of the hemagglutinin content of influenza virus vaccines by single-radial-immunodiffusion. J. Biol. Stand., 8: 289-296. Wood, J.M., Schild, G.C., Newman, R.W. and Seagroatt, V.A., 1977. An improved single-radial-immunodiffusion technique. Application for potency determinations of inactivated whole virus and subunit vaccines. J. Biol. Stand., 5: 237-247. Wright, P.F., Cherry, J.D., Hjordis, F.M., Glezen, W.P., Hall, C.B., McIntosh, K., Monto, A.S., Parrott, R.H., Portnoy, B. and Taber, L.H., 1983. Antigenicity and reactogenicity of influenza A/USSR/77 virus vaccine in children - a multicentered evaluation of dosage and safety. Rev. Infect. Dis., 5: 758-764.
253
Single-radial-immunodiffusion as an in vitro potency assay for human inactivated viral vaccines M.S. Williams Laboratory of Respiratory Viruses, Division of Virology, Centerfor BiologicsEvaluation and Research, Food and DrugAdministration, Bethesda, MD, USA (Accepted 14 July 1993 )
ABSTRACT Single-radial-immunodiffusion(SRID) assays have been used to determine the potency of all human inactivated influenza virus vaccines licensed by the Food and Drug Administration for use in the United States since 1978. SRID replaced less reliable tests which were based on the aggregation of erythrocytes by the hemagglutinins of influenza viruses. Similar SRID assays have been used experimentally to determine the potency of inactivated polio and rabies vaccines. In each case, the assays are based on the diffusion of viral antigen into an agarose gel containing specific antibodies to the antigen being measured. For influenza and rabies, disruption of the virions with a detergent is necessary to permit the diffusion of the appropriate antigens, where as with polio, intact virions are allowed to diffuse. The interaction between antigen and antibody produces a zone of precipitation whose size is directly proportional to the amount of antigen applied. A potency value for unknowns is obtained by comparing the sizes of zones produced by unknown preparations to the sizes of zones obtained with a calibrated reference of known antigen content. Once the specific reference antigens and antibodies are prepared and the test standardized, it is a remarkably simple technique which unlike agglutination assays is very reproducible, relatively unaffected by minor variations in test conditions and is far less time consuming and cumbersome than in vivo assays for potency such as those done by inoculatingmice or monkeys. More importantly, clinical studies demonstrate that standardization of influenza vaccines by SRID provides a better correlate of human immunogenicity than previous methods.
INTRODUCTION
The severity of disease caused by viruses such as rabies and polio is well known by the scientific and medical communities as well as the general public, but the severity of disease which can be caused by influenza viruses is less well understood. Influenza viruses cause considerable morbidity and mortalCorrespondence to: M.S. Williams, Laboratory of Respiratory Viruses, Division of Virology, Center for Biologics Evaluation and Research, Food and Drug Administration, 8800 Rockville Pike, Bethesda, MD 20892, USA.
2 54
M.S. WILLIAMS
ity in human populations, especially high-risk groups such as the elderly and persons with underlying health problems. During major epidemics, hospitalization rates for high-risk persons may increase 2- to 5-fold. Previously healthy children and young adults may also require hospitalization due to influenza or influenza-related complications, although the relative increase in their hospitalization rates is less than for persons in high-risk populations. Increased mortality results not only from influenza and pneumonia but also from cardiopulmonary and other chronic diseases that can be exacerbated by influenza infection. At least 10000 excess deaths have been documented in each of 19 different epidemics in the United States during the last 30 years, and more than 40000 excess deaths occurred in each of three of these epidemics. Approximately 80-90% of these excess deaths were among persons greater than 64 years of age. (ACIP, 1991 ) The proportion of elderly persons in the United States population is increasing. The numbers of younger persons at increased risk for influenza-related complications is also increasing due a number of factors, such as better survival rates for organ transplant recipients and diseases such as cystic fibrosis and acquired immunodeficiency syndrome, and the success of neonatal intensive care units. Therefore, the need for safe and effective vaccines for protection against influenza remains a very important public health issue for the present as well as the future (ACIP, 1991 ). There has been extensive experience in the last 15 years in testing human inactivated viral vaccines by single-radial-immunodiffusion (SRID). Tests for the potency of influenza, rabies and polio vaccines have been developed using similar methods. All three require the production of antibodies to a purified preparation of a target protein antigen from the virus, and the standardization of a reference antigen. The production and standardization of these reference reagents can be time consuming and costly. The test itself is, however, simple to perform and produces accurate and reproducible results. Optimal results may require modifications of the test for different antigens. For example, influenza viruses are disrupted with the detergent Zwittergent 3-14 and rabies uses the detergent Emulphogene BC-720. The SRID for polio viruses utilizes an autoradiographic technique causing zone enhancement (ZE), which is necessary to measure the low levels of antigen present in final vaccines. The SRID assays for these viruses have shown advantages in some ways over other types of tests, but are in different stages of acceptance for official tests for release of lots of vaccines. The SRID assay used to establish the potency of influenza virus vaccines has been used routinely by manufacturers and regulatory agencies for many years, and the description of this test serves as a model for the development of similar techniques. PREPARATION OF REAGENTS
The performance of SRID assays requires a specific antiserum to the purified antigen to be measured. For influenza viruses, the antigen of most im-
SINGLE-RADIAL-IMMUNODIFFUSION
255
portance is the hemagglutinin (HA), which is one of the two viral surface glycoproteins. Embryonated chicken eggs, 9 to 11 days old, are inoculated via the allantoic sac, incubated at 33 ° C to 35 ° C for 3 days, and the allantoic fluid harvested. A purified whole virus preparation is obtained by centrifugation through a sucrose gradient. One portion of the purified virus is used to prepare HA for the immunization of sheep. The HA is removed from the viral core by the method of Brand and Skehel (1972) using the enzyme bromelain. The method yields HA which is missing the hydrophobic membrane anchor and destroys the enzymatic and antigenic properties of the other surface antigen, neuraminidase (NA), in the process. A total of approximately 100 #g of purified HA is inoculated intramuscularly and intradermally into sheep (or other suitable animals) using Freund's Complete Adjuvant. Three weeks after the first injection, 2 to 4 booster doses of 20 to 30/tg in Freund's Incomplete Adjuvant are given at weekly intervals by the same routes of inoculation, and continued until antibody levels, which are tested by SRID before each subsequent immunization, have reached a desired working range. For influenza viruses, this range is about 1 to 10/~1 of antiserum per ml ofagarose. Antibody levels in this range should produce distinct and measurable zone diameters of about 8 to 10 mm for antigen preparations containing 30/tg of HA per ml. A reference standard of known antigen content is necessary to determine the potency of an unknown. The calibration of this reference may be accomplished in several ways, all of which rely on a protein determination. One method is to solubilize the HA from another portion of the whole virus preparation using the detergent Nonidet P-40 (NP-40). NP-40 solubilization leaves the membrane anchor region of the HA intact. After solubilization, the HA and NA antigens are separated by centrifugation through a sucrose gradient. Since NA and HA migrate similarly through the gradient, HA prepared in this manner contains some NA. The amount of NA present is about 10% of the total H A / N A (HANA) preparation, and it does not interfere with the standardization process. A protein value is determined for the HANA (minus 10% for the NA) and this material is used as a primary standard in SRID tests to assign a value to a lyophilized, formalin inactivated, whole influenza virus vaccine preparation. The latter lyophilized virus preparation is then used as the reference antigen for the actual potency testing of influenza virus vaccines which are antigenically identical to the reference standard. As a means of determining the accuracy of this standardization process, gg of HA protein by SRID and Ftg of total protein are determined on the whole virus preparation before lyophilization. Based on our experience over the years with many different strains of influenza viruses, the amount of HA protein should be about 30 to 50% of the total protein. Both of the purified HA preparations, due to their different properties, serve a unique role in the preparation of reagents. The bromelain material, because
2 56
M.S. WILLIAMS
it destroys NA, yields a highly purified HA preparation and therefore produces antibodies which recognize only the HA. However, since a substantial portion of the HA remains in the viral core, and standardization is based on total protein values, it does not serve well as a primary standard. The NP-40 solubilized HA remains whole and although it contains a small amount of NA making it unsuitable for producing antibodies, the NA can be deducted from calculations when used as a primary standard. Standardization of these reagents relies on protein values, and it is important to establish a protein assay which is reliable, reproducible, and free from interference by chemicals in the different preparations. For instance, the Lowery protein, commonly used in many laboratories, is subject to interference by detergents and is not suitable for measuring the protein content of the NP-40 material. Kjeldahl, amino acid analysis, and an automated o-phthalaldehyde fluorometric assay (Benson & Hare, 1975 ) have all been used successfully in our laboratory. The Kjeldahl measures nitrogen and care must be taken to avoid testing of samples with chemicals containing nitrogen such as tris buffers. The purity of the various preparations described above is checked by polypeptide analysis according to the method of Laemmli (1970). Samples are electrophoresed through 10% polyacrylamide gels. The gels are stained with Coomassie Brilliant Blue R-250, and analyzed to ensure that the preparations are free from contaminating proteins. P E R F O R M A N C E O F T H E TEST
The following procedure is a brief description of how SRID assays are performed for influenza vaccines and is a modification of the procedures described by Wood et al. ( 1977 ) and Williams et al. (1980). The first step in the performance of the test is the preparation of agarose gels. The amount of antiserum necessary to perform the test on samples at a given antigen concentration is determined in a set of preliminary experiments and the optimal amount is added to a 1% solution of agarose melted in phosphate buffered saline (PBS). Care should be taken to ensure that the agarose has been completely melted to allow for even diffusion and staining. The temperature of the solution must be lowered to 55 °C or below before addition of the antiserum or destruction of antibodies may occur. The agarose/antiserum mixture is injected slowly and steadily into a 240 × 80 X 2 m m casting plate, and placed at 4 °C for at least 5 rain to allow complete gelation. In order to prepare the gels for sample inoculation, wells, 4 m m in diameter, are punched in a 4 by 12 row configuration. This configuration allows for the inoculation of 1 dilution series of the reference and 5 dilution series of unknowns, each in duplicate. Samples of both reference and unknown antigens are pre-diluted to the de-
SINGLE-RADIAL-IMMUNODIFFUSION
257
sired testing range of approximately 30 ~g HA/ml in PBS and mixed at a 1:10 ratio with a 10% stock solution of the detergent Zwittergent 3-14 to give a final detergent concentration of 1%. The antigen/detergent mixture is incubated at room temperature for approximately 10 min. Samples are then diluted in PBS to obtain a dilution series of l: 1, 1:1.5, 1:2 and 1:4. Twenty/As of each dilution is inoculated into duplicate wells. The gel plates are placed in a moist chamber at room temperature for a minimum of 18 hours. Following the 18 h incubation, the gels are gently removed from the casting plates in a pan of tap water and floated onto a glass plate or a strip of Gel Bond. A moist piece of filter paper is placed on top of the gel and it is allowed to dry completely. The filter paper is removed and the dried gel is stained with Coomassie Brilliant Blue R-250 for l0 rain and destained as needed to visualize the zones of precipitation. The zone diameters are measured in two directions at right angles to each other to the nearest 0.1 m m using a microcomparator, a magnifying projector, or other suitable measuring device. Zones that are eliptical in shape should be measured across the long and the short axis. Potency determinations are made by comparing the zone sizes of the reference dilutions to those of the unknowns. For the potency of vaccines in the United States, calculations are performed by the parallel line bioassay method of Brownlee ( 1965 ) plotting log antigen dilution vs. log zone diameter. Another simpler calculation which is widely used in other parts of the world is the slope ratio method described by Wood et al. (1977). In this method, the slope of the vaccine curve is divided by the slope of the reference curve and multiplied by the potency value of the reference. The overall time required for performance of the test, including potency calculations, is approximately 24 hours. During the early development of SRID for influenza virus vaccines, the detergent used to disrupt the virus was sodium sarkosyl sulfate (SSS). We discovered that this detergent gave inconsistent results between different manufacturing processes and for some sub types of influenza actually yielded smaller precipitation zones when comparing SSS treated purified HA to purified HA with no detergent treatment. A large number of different detergents were studied for their ability to effectively disrupt the virion and produce precipitation zones of sufficient size and clarity (Table 1 ). Several non-ionic detergents, Triton X-100, Triton CF-54, Nonidet P-40, Octylglucoside, and Emulphogene BC-720 were found to give relatively large zone diameters as well as excellent quality of zone reaction as measured by intensity of staining and crispness of zone edges. The zwitterionic detergent Zwittergent 3-14 gave similar results to the non-ionic detergents listed above, but gave more consistent results when tested against all of the licensed influenza vaccine preparations.
2 58
M.S.WILLIAMS
TABLE 1 Comparison of detergents for the solubilization of influenza A H 1N 1 and influenza B antigens
Detergents
Source
Relative zone diameter size
Relative qualityof reaction
Supelco, Inc. Bellefonte, PA
Small Small
Poor Poor
Small
Good
Small
Good
Large Large Large Small Small Small Small Small Small Small Small Small Large
Good Good Good Good Good Good Good Good Good Poor Poor Poor Good Good
Large
Good
Anionic Triton Wr-1339 Sodium Taurocholate
Cationic Cetyltrimethylammonium Bromide Hyamine CL
Non-ionic Triton X- 100 ( TX- 100) Triton CF-54 Nonidet P-40 (NP-40) Emulphogene (BC-420) Emulphogene (BC-610)
Shell Oil Co., London GAF Corp., Linden N.J.
Emulphogene (BC-720) Emulphogene (BC-840) Emulphogene (BC-970)
Large
Lubrol WX Brij 35 Myrj 59 Tween 20 Tween 80
Supelco, Inc., Bellefonte, PA
Octylglucoside
Sigma Chemical Co., St. Louis, MO
Zwitterionic
Zwittergent 3-14
Calbiochem-Berhing Corp., La Jolla, CA
DISCUSSION
SRID has proven to be a successful replacement for other in vitro potency assays based on the agglutination of erythrocytes for influenza virus vaccines. The old potency assays were not reproducible within or between laboratories because they depended on the physical ability of the hemagglutinin to bind to red cells. The reproducibility of these tests was therefore susceptible to any condition which could affect that bond such as variation in pools of red cells, temperature, or vibrations. Different types of vaccines were also found to behave differently such that sub-unit or split virus preparations appeared to have a much higher antigen content than whole virus preparations. (Table 2 shows the manufacturers and the different types of influenza virus vaccines mar-
SINGLE-RADIAL-IMMUNODIFFUSION
259
TABLE 2 Influenza virus vaccine manufacturers, types and processes Manufacturer
Vaccine Type and Process
Connaught Laboratories, Inc.
1. whole virus vaccine, gradient purified 2. sub-unit vaccine, gradient purified, disrupted with Triton X-100 3. surface antigen vaccine, gradient purified, disrupted with Triton N- 101, and further purified to obtain a vaccine containing predominantly HA and NA 4. sub-unit vaccine, gradient purified, disrupted with Polysorbate 80 and extracted with ether 5. sub-unit vaccine, chromatographically purified, disrupted with tri (n) butylphosphate and Polysorbate 80
Lederle/Evans*
Parke-Davis, Inc. Wyeth-Ayerst, Inc.
*Marketed by Lederle Laboratories under a divided manufacturing agreement with Evans Medical Ltd.
keted in the United States at this time. ) Analysis of clinical studies clearly demonstrated that vaccines, regardless of their different manufacturing techniques, standardized by SRID correlated well with human immunogenicity and reactogenicity. In 1978, the year that SRID was first used as the potency test for influenza virus vaccines, large scale clinical studies were performed which provided information relating dose in #g of HA to immunologic responses and reaction rates in humans and formed the basis for the amount of antigen to be used in vaccines licensed for public use. That year also marked the return of the circulation of H 1N 1 viruses, so that clinical information was provided for populations of people who had been immunologically primed by exposure to H3N2 viruses, as well as a population of young people who were not immunologically primed to H 1N 1 viruses. The studies demonstrated that in adult, primed populations, whole virus and sub-unit vaccines standardized by SRID were quite comparable in human immunogenicity and reactogenicity. In younger, unprimed populations, sub-unit vaccines were found to be less reactive than whole virus vaccines and required two doses of vaccine to achieve protective levels of antibodies. The clinical information derived from these studies using vaccines whose potency was determined by SRID established a basis for how influenza virus vaccines are used and formulated today (Cate et al., 1983; La Montagne et al., 1983; Quinnan et al., 1983; Wright et al., 1983). The experience with SRID in potency testing of human rabies vaccines by
260
M.S. WILLIAMS
the methods of Ferguson and Schild ( 1982 ) and Fitzgerald and Needy (1986) is quite extensive, but still at an experimental level. In this vaccine, the protein which is measured by SRID is the envelope glycoprotein. Although SRID assays have several advantages over the currently used NIH mouse potency test as described by Seligman ( 1973 ) such as less timely and cumbersome to perform, more consistency from test to test and lot to lot, and they do not require challenge with live rabies virus, there is concern that the envelope glycoprotein may not be the only antigen which is important in conferring protection. Work by Dietzschold et al. (1987) has shown that the ribonucleoprotein also contributes to protection from lethal challenge. Some manufacturers are using SRID as an internal test in helping to formulate vaccine, but given the severe consequences of vaccine failure, more information will need to be obtained before SRID can replace the animal model for the purpose of releasing final vaccine to the public. SRID potency tests for inactivated poliovirus vaccines as described by Schild et al. (1980) have been used at various stages of production by some manufacturers. It has been shown to be a reliable method of determining the D antigen content, which is the component that produces neutralizing antibodies. However, because of the length of time involved in performing the ZE test (approximately 10 days) and because of a reluctance by some manufacturers to use radio isotopes in their facilities, SRID is not routinely used as a potency test for this vaccine. In process testing of vaccines during various manufacturing steps is usually performed utilizing an ELISA assay, and final vaccine testing still requires tests in animals. SRID tests can be highly strain specific, depending upon the properties of the antisera. This can be problematic as well as useful. For influenza virus vaccines, which currently contain three constantly changing subtypes, new antisera and reference antigens must be made and standardized each time a new strain is incorporated into the vaccine formula. There have been 21 different strains of influenza viruses in the vaccine formula in the last 15 years, requiring a large commitment of time and resources for the production of these reagents. This same strain specificity, however, also allows for the testing of individual strains in a multivalent vaccine, which was not possible with earlier assays. With poliovirus vaccines, this specificity can be used to distinguish between different strains within subtypes. CONCLUSION
SRID techniques have been successfully developed as potency tests for several types of human inactivated viral vaccines. It is a simple technique to perform and, in the case of influenza, has been shown to be more reliable and a better predictor of human immunogenicity than older in vitro assays and has therefore replaced those tests. For rabies vaccine, it is more reproducible,
SINGLE-RADIAL-IMMUNODIFFUSION
261
less time consuming and cumbersome, and safer than the in vivo N I H mouse potency test since personnel need not be exposed to infectious rabies virus. Studies are in progress to determine if SRID will replace the N I H test for rabies, or if the two can be used in conjunction to determine the immunogenicity o f vaccine lots. For inactivated poliovirus vaccine, practical considerations have kept SRID from becoming the test o f choice for vaccine potency testing. ACKNOWLEDGEMENTS I acknowledge the personal communications o f Dr. Ronald Lundquist from the Division o f Virology, Center for Biologics Evaluation and Research for information on poliovirus vaccines and Ms. Nancy Roscioli from the Division of Product Quality Control, Center for Biologics Evaluation and Research for information on rabies vaccine.
REFERENCES Benson, J.R. and Hare, P.E., 1975. o-Phthalaldehyde:fluorogenicdetection of primary amines in the picomolerange. Comparison with fluorescainineand ninhydrin. Proc. Nat. Acad. Sci., 72:619-662. Brady, M.I. and Furminger, I.G.S., 1976. A surface antigen influenzavaccine. I. Purification of haemagglutininand neuraminidase proteins. J. Hyg., 77:161-172. Brand, C.M. and Skehel, J.J., 1972. Crystallineantigen from the influenza virus envelope. Nature, New biology, 238: 145-147. Brownlee, K.A., 1965. In: Statistical Theory and methodology in Science and Engineering., Chapter I 1. New York: John WileyPublishers. Cate, T.R., Couch, R.B., Parker, D., and Baxter, B., 1983. Reactogenicity,immunogenicity,and antibody persistence in adults given inactivated influenza virus vaccines- 1978. Rev. Infect. Dis., 5, No. 4: 737-747. Dietzschold, B., Honghai, W., Rupprecht, C.E., Esteban, C., Tollis, M., Ertl, H., Heber-Katz, E. and Koprowski,H., 1987. Induction of protective immunity against rabies by immunization with rabies virus ribonucleoprotein. Proc. Nat. Acad. of Sci., 84:9165-9169 Ennis, F.A., Mayner, R.E., Barry, D.W., Manischewitz,H.E., Dunlap, R.C., Verbonitz, M.W., Bozeman, F.M. and Schild, G.C., 1977. Correlation of laboratory studies with clinical responses to A/New Jersey influenza vaccines. J. Infect. Dis., Supplement 136: $397-$406. Ferguson, M. and Schild, G.C., 1982. A single-radial-immunodiffusiontechnique for the assay of rabies glycoprotein antigen: Application for potency tests of vaccines against rabies. J. Gen. Virol., 59: 197-201. Fitzgerald, E.A. and Needy, C.F., 1986. Use of the single radial immunodiffusion test as a replacement for the NIH mouse potency test for rabies vaccine. In: IABS Symposiumon Reduction of Animal Usage in the Development and Control of Biological Products, London, United Kingdom, 1985. Dev. Biol. Stand., 64: 73-79. Basel: S. Karger. La Montagne, J.R., Noble, G.R., Quinnan, G.V., Curlin, G.T., Blackwelder,W.C., Smith, J.I., Ennis, F.A. and Bozeman, F.M., 1983. Summary of clinical trials of inactivated influenza vaccine- 1978. Rev. Infect. Dis., 5: 723-736.
262
M.S.WILLIAMS
Laemmli, U.K., 1970. Cleavage of structural proteins during the assembly of the lead bacteriophage T4. Nature (London), 227: 680-685. Mayner, R.E., Blackburn, R.J. and Barry, D.W., 1977. Quantitation of influenza vaccine hemagglutinin by immunoelectrophoresis. In: International symposium on Influenza Immunization (II), Geneva. Dev. Biol. Stand., 39:169-178. Basel: S. Karger. Prevention and control of influenza. Recommendations of the Immunization Practices Advisory Committee (ACIP). M M W R 1991, 40 RR-6: 1-15. Quinnan, G.V., Schooley, R., Dolin, R., Ennis, F.A., Gross, P. and Gwaltney, J.M., 1983. Serologic responses and systemic reactions in adults after vaccination with monovalent A/USSR/ 77 and trivalent A/USSR/77, A/Texas/77,B/Hong Kong/72 influenza vaccines. Rev. Infect. Dis., 5: 748-757. Schild, G.C., 1974. In: Proc., International Conference on Standardization of Diagnostic Materials, June 1973, in Atlanta, Georgia, U.S.A., pp. 243-248. Schild, G.C., 1977. The reference materials for the standardization of influenza vaccines. In: World Health Organization WHO/B5/77, 1147. Schild, G.C., Wood, J.M., Minor, P.D., Dandawate, C.N. and Magrath, D.I., 1980. Immunoassay of poliovirus antigens by single-radial-diffusion: Development and Characteristics of a sensitive autoradiographic zone size enhancement (ZE) technique. J. Gen. Virol., 51: 157170.
Seligman, E.B., Jr., 1973. The NIH test for potency. In: Laboratory Techniques in Rabies (WHO), 3rd edition, chapter 33, 279-285. Geneva: WHO. Williams, M.S., Mayner, R.E., Daniel, N.J., Phelan, M.A., Rastogi, S.C., Bozeman, F.M., and Ennis, F.A., 1980. New developments in the measurement of the hemagglutinin content of influenza virus vaccines by single-radial-immunodiffusion. J. Biol. Stand., 8: 289-296. Wood, J.M., Schild, G.C., Newman, R.W. and Seagroatt, V.A., 1977. An improved single-radial-immunodiffusion technique. Application for potency determinations of inactivated whole virus and subunit vaccines. J. Biol. Stand., 5: 237-247. Wright, P.F., Cherry, J.D., Hjordis, F.M., Glezen, W.P., Hall, C.B., McIntosh, K., Monto, A.S., Parrott, R.H., Portnoy, B. and Taber, L.H., 1983. Antigenicity and reactogenicity of influenza A/USSR/77 virus vaccine in children - a multicentered evaluation of dosage and safety. Rev. Infect. Dis., 5: 758-764.