AN EVALUATION OF METHODS FOR THE CONCENTRATION AND PURIFICATION OF INFLUENZA VIRUS

Assessing infectivity of emerging enveloped viruses in wastewater and sewage sludge: Relevance and procedures

The emergence of SARS-CoV-2 has heightened the need to evaluate the detection of enveloped viruses in the environment, particularly in wastewater, within the context of wastewater-based epidemiology. The studies published over the past 80 years focused primarily on non-enveloped viruses due to their ability to survive longer in environmental matrices such as wastewater or sludge compared to enveloped viruses. However, different enveloped viruses survive in the environment for different lengths of time. Therefore, it is crucial to be prepared to assess the potential infectious risk that may arise from future emerging enveloped viruses. This will require appropriate tools, notably suitable viral concentration methods that do not compromise virus infectivity. This review has a dual purpose: first, to gather all the available literature on the survival of infectious enveloped viruses, specifically at different pH and temperature conditions, and in contact with detergents; second, to select suitable concentration methods for evaluating the infectivity of these viruses in wastewater and sludge. The methodology used in this data collection review followed the systematic approach outlined in the PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analysis) guidelines. Concentration methods cited in the data gathered are more tailored towards detecting the enveloped viruses' genome. There is a lack of suitable methods for detecting infectious enveloped viruses in wastewater and sludge. Ultrafiltration, ultracentrifugation, and polyethylene glycol precipitation methods, under specific/defined conditions, appear to be relevant approaches. Further studies are necessary to validate reliable concentration methods for detecting infectious enveloped viruses. The choice of culture system is also crucial for detection sensitivity. The data also show that the survival of infectious enveloped viruses, though lower than that of non-enveloped ones, may enable environmental transmission. Experimental data on a wide range of enveloped viruses is required due to the variability in virus persistence in the environment.

Challenges of Making Effective Influenza Vaccines

Seasonal influenza vaccines prevent influenza-related illnesses, hospitalizations, and deaths. However, these vaccines are not as effective as other viral vaccines, and there is clearly room for improvement. Here, we review the history of seasonal influenza vaccines, describe challenges associated with producing influenza vaccine antigens, and discuss the inherent difficulties of updating influenza vaccine strains each influenza season. We argue that seasonal influenza vaccines can be dramatically improved by modernizing antigen production processes and developing models that are better at predicting viral evolution. Resources should be specifically dedicated to improving seasonal influenza vaccines while developing entirely new vaccine platforms.

The evolving history of influenza viruses and influenza vaccines

The isolation of influenza virus 80 years ago in 1933 very quickly led to the development of the first generation of live-attenuated vaccines. The first inactivated influenza vaccine was monovalent (influenza A). In 1942, a bivalent vaccine was produced after the discovery of influenza B. It was later discovered that influenza viruses mutated leading to antigenic changes. Since 1973, the WHO has issued annual recommendations for the composition of the influenza vaccine based on results from surveillance systems that identify currently circulating strains. In 1978, the first trivalent vaccine included two influenza A strains and one influenza B strain. Currently, there are two influenza B lineages circulating; in the latest WHO recommendations, it is suggested that a second B strain could be added to give a quadrivalent vaccine. The history of influenza vaccine and the associated technology shows how the vaccine has evolved to match the evolution of influenza viruses.

AGGLUTINATION OF STAPHYLOCOCCUS AUREUS IN THE PRESENCE OF CHORIOALLANTOIC FLUID FROM HEN'S EGGS

Certain strains of Staphylococcus aureus will agglutinate when grown in the presence of normal or influenza virus-infected chorio-allantoic fluid. Fluids are rendered inactive after heating at 56 degrees C. for one or two minutes. Surface tension and temperature are not important factors in the manifestation of the phenomenon so long as they are within the limits compatible with growth. The reaction is favored by an alkaline pH. Bacterial dissociation is not a feature of this phenomenon and the relationship of this reaction to bacteriophagy is probably not significant but is being investigated further. Similar reactions as described for the egg fluids uiay be seen when Berkefeld filtrates of influenza virus-infected and normal mouse lungs are used. A more detailed account of all the data will appear at a later date.

INFLUENCE OF pH AND OF CERTAIN OTHER CONDITIONS ON THE STABILITY OF THE INFECTIVITY AND RED CELL AGGLUTINATING ACTIVITY OF INFLUENZA VIRUS

A study has been made of the pH stability of centrifugally purified strains of influenza virus with respect to the biological properties of mouse infectivity and chicken red blood cell agglutinating activity. Observations also were made on the importance of composition of buffer, temperature of storage, and concentration of virus protein to the stability of the virus. When tested for stability at a protein concentration of 0.1 mg. per cc. in phosphate buffer, the infectivity of PR8 virus was found to be most stable at pH 6.5–7; the swine virus, at pH 7–7.9; and the Lee strain, at a pH of 7.9 or higher. The CCA activity of the PR8 virus in phosphate buffer was most stable at pH 7, that of the swine virus at pH 7–8, and that of the Lee virus at a pH greater than 9. Furthermore, the Lee virus was much less stable in dilute solution in phosphate buffer, even under optimum conditions of pH, than either the PR8 or swine strains. The different strains of influenza virus were found to possess certain characteristics in common. They lost infectivity and CCA activity on the acid side of optimum pH conditions much more rapidly than on the alkaline side. Under suitable conditions of buffer and pH, the infectivity decreased while the CCA activity remained unchanged. In general, the rate of loss in infectivity was greater than the rate of loss in CCA activity. When tests of stability were carried out at a protein concentration of 0.1 mg. per cc. in a composite phosphate-glycine-NaCl buffer, the virus strains showed less marked differences and possessed much higher stabilities of CCA activity and infectivity than when stored at the same concentration in phosphate buffer alone. Under the modified conditions, all three viruses possessed maximum stabilities of CCA activity and infectivity at pH 7–8 with the exception of the PR8 virus whose infectivity appeared more stable at pH 7 than at pH 8. In detailed experiments with the Lee virus, it was found that the infectivity and CCA activity of this strain at pH 7 and at a protein concentration of 0.1 mg. per cc. were maintained best in the composite phosphate-glycine-NaCl buffer, less well in a buffer containing glycine and NaCl, and least well in phosphate buffer alone. In tests with PR8 virus, the activity was found to be much more stable at 4° C. than at 23° C. When stored at a concentration of 2 mg. per cc. at 4° C. in phosphate buffer at pH 7, the PR8 and Lee strains were found to be much more stable than when stored at the concentration of 0.1 mg. per cc. At the higher concentration, no significant losses in either infectivity or CCA activity were observed over a period of 2 months.

THE STABILITY OF INFLUENZA VIRUS IN THE PRESENCE OF SALTS

The stability of centrifugally purified PR8 influenza virus at pH 7 in 0.001, 0.01, 0.1, and 1.0 M phosphate buffers, in veronal and borate buffers, and in adjusted solutions of saline and distilled water was investigated. The results demonstrate that the stability of this virus can vary considerably at pH 7 depending upon the nature and concentration of the salts present. Borate, veronal, and phosphate buffers at a concentration of about 0.1 M showed almost equal ability to maintain virus activity over several weeks of time at 4°C. In many cases, it may prove inconvenient to use veronal buffer, however, because of the difficulty in determining protein concentration in its presence. The 0.1 M phosphate buffer has proved in tests not described here to be slightly more consistent in preserving virus activity than the borate and may, therefore, be considered slightly superior to the latter. It is apparent that unbuffered saline is a poor solvent for preserving virus activity regardless of pH. The activity of partially inactivated virus in distilled water and in saline solutions was increased ten to 1000 times by diluting such solutions with 0.1 M phosphate buffer. Some reactivation was also effected with veronal but not with borate buffers.

ELECTROPHORETIC STUDIES ON PR8 INFLUENZA VIRUS

Crude preparations of PR8 influenza virus, obtained by high-speed centrifugation, contain two electrophoretically distinct components. One of these, present to the extent of 10 to 20 per cent, was identified by electrophoresis, centrifuge, and activity tests, as an impurity similar to or identical with a high molecular weight acidic substance shown by Knight to be elaborated by normal uninfected embryos. The other component, present to the extent of 80 to 90 per cent, appeared to represent the active virus. The virus fraction was separated from the impurity by repeated fractional centrifugation. It then appeared homogeneous in the analytical centrifuge and in the Tiselius apparatus, and possessed an isoelectric point at pH 5.3 as measured by the micro-electrophoresis method.

BIOPHYSICAL PROPERTIES OF PREPARATIONS OF PR8 INFLUENZA VIRUS

The biophysical properties of several preparations of PR8 influenza virus have been studied. Electron micrographs showed slightly irregular, circular particles with an average diameter of 115 mmicro and a standard deviation of the distribution of diameters of 15 per cent. The specific volume was determined with a pycnometer to be 0.79. The sedimentation rate was found to vary inversely with the concentration of virus. The extrapolated value for one preparation was 722 Svedberg units and the value for another preparation was 658 units. Sedimentation studies in sucrose solutions of varying densities showed a non-linear dependence of sedimentation rate upon solvent density, indicating that the density in solution increases with increasing sugar concentration. The virus particles floated in a sugar solution with a density of 1.18. The density in the absence of sucrose was estimated to be about 1.1. The virus was shown not to lose infectivity very rapidly in sucrose solutions. The spreading of the boundary during sedimentation was shown to be represented by a standard deviation of the sedimentation rate equal to 8 per cent of the mean. This corresponds to a distribution of diameters with a standard deviation of 4 per cent of the mean. A more slowly sedimenting component was observed with the ultracentrifuge in the highest concentrations of several of the preparations. The intrinsic viscosities, calculated on the basis of concentrations expressed as grams per cubic centimeter, of several highly purified preparations were determined to be between 11.3 and 16.5. The variation of sedimentation rate with concentration was shown to be due entirely to the variation of solution viscosity with concentration. The viscosity of the virus preparations can be explained as being due in great part to the presence of the slower sedimenting contaminating components which possess a very high intrinsic viscosity. The average particle diameter of the anhydrous PR8 virus particle was calculated from sedimentation and specific volume data to be about 80 mmicro. The discrepancy between this value and that obtained from the electron micrographs, the greater size distribution from the electron micrographs, the slight irregularities in some of the particles as observed in the electron micrographs, the behavior of the sedimentation process in sucrose solutions of different densities, and the inactivation of the virus by withdrawal of electrolytes can all be explained in a straightforward manner if it is assumed that the virus particles in solution contain, in addition to the constituents shown by chemical analysis, about 60 per cent by weight of water. It is estimated that such hydrated virus particles are about 100 mmicro in diameter.

TITRATION OF INFLUENZA VIRUS IN CHICK EMBRYOS

A study was made to establish the reproducibility of end points obtained in the titration of influenza viruses in chick embryos. Six tests were performed, each composed of five replicate titrations of a purified preparation of the PR8 strain of influenza virus. The data from these titrations were subjected to statistical analysis which revealed that the chances are 19 out of 20 that differences in end points of 0.37 and 0.62 logarithmic units are significant in titrations employing ten embryos and five embryos per dilution, respectively. Additional simultaneous titrations in embryos and in mice showed that chick embryos are sensitive to considerably smaller amounts of virus than are mice, and that the mouse titration is adversely affected by inactive virus under conditions which are without apparent influence on the embryo titration.

A SEDIMENTABLE COMPONENT OF ALLANTOIC FLUID AND ITS RELATIONSHIP TO INFLUENZA VIRUSES

Macromolecular material was isolated from normal allantoic fluid by a centrifugation procedure comparable to that currently employed for the concentration and purification of influenza viruses. The yield of material was found to vary with the age of the embryo, reaching a maximum average value after 14 days of incubation at 39°C. of about 0.02 mg. per ml. of allantoic fluid. The purified material was found to contain protein, carbohydrate, and lipid and to have a general composition similar to purified preparations of PR8 influenza virus. A typical preparation of normal material had an isoelectric point at pH 2.3. Sedimentation studies indicated that the normal material can give a variety of sedimentation constants depending upon the concentration and viscosity of the preparations. The sedimentation constant, corrected for viscosity, of the major component of a fresh preparation was 170 S. The diameters of the predominant particles shown in electron micrographs of the normal material and of preparations of PR8 influenza virus were about 40 and 100 mµ, respectively. Serological tests indicated that the normal material is a good antigen and that preparations of both A and B types of influenza virus obtained from allantoic fluids by centrifugation show a strong serological relationship to the normal material. Freezing and thawing of allantoic fluid, and repeated adsorption of virus on red cells, failed to provide a practical basis for the separation of normal protein from the virus entity in the case of PR8 virus. In the cases of similar preparations of F12 and of Lee viruses, a partial separation of a small component was accomplished by fractional centrifugation and this component and the normal protein were shown to be identical or very closely related. Antiserum to the purified normal material inhibited red cell agglutination by A and B types of influenza virus at serum dilutions of 600 to 700, but failed to show significant neutralizing capacity in chick embryo and in mouse tests at a serum dilution of 100. Rabbit antiserum to purified preparations of PR8 virus gave a 50 per cent red cell agglutination inhibition endpoint at a serum dilution of 112,000. Some of the implications of the findings are discussed.

THE NUCLEIC ACID AND CARBOHYDRATE OF INFLUENZA VIRUS

Both ribonucleic and desoxyribonucleic acids have been obtained from purified particles of PR8 influenza virus. These particles were also found by extraction with formamide to contain carbohydrate in addition to that of the nucleic acids. Carbohydrate-rich fractions, essentially devoid of nucleic acid, were obtained not only from the particles representing PR8 virus but from those of Lee influenza virus as well. The carbohydrate in each case appeared to be a polysaccharide composed of mannose, galactose, and glucosamine units.

THE PREPARATION AND PROPERTIES OF INFLUENZA VIRUS VACCINES CONCENTRATED AND PURIFIED BY DIFFERENTIAL CENTRIFUGATION

Influenza virus vaccines containing from 1 to 10 mg. of virus materials per cc. concentrated and purified from infectious allantoic fluids by means of one or two cycles of differential centrifugation and inactivated by different treatments have been prepared and subjected to laboratory tests. Suitable inactivation of the virus preparations with retention of full red cell agglutinating activity and immunizing potency in mice was achieved by treatment with minimal amounts of formaldehyde or ultraviolet light. Treatment with phenol or chloroform failed to cause adequate loss of virus activity. Excessive amounts of formaldehyde or of ultraviolet light were found to cause a loss in red cell agglutinating activity and in immunizing potency. Freezing resulted in the immediate loss of red cell agglutinating activity of the formalinized vaccine. Storage of the vaccines in the frozen state was accompanied by a gradual decrease in red cell agglutinating activity. Drying of the vaccines from the frozen state resulted in a loss of red cell agglutinating activity and, in the case of the formalinized vaccine, in a loss in immunizing potency. There appeared to be at least a rough correlation between red cell agglutinating activity and immunizing potency. The immunizing potency and red cell agglutinating activity of a purified formalinized vaccine containing 2 mg. of virus material per cc. were unchanged following 2 months' storage at 4 degrees but were measurably decreased following storage for 2 months at 18 to 25 degrees and at 37 degrees . At equivalent dosages of virus material the immunizing potency of formalinized centrifugally purified virus, of formalinized virus purified by the red cell elution method, and of infectious allantoic fluid was not measurably different. The immunizing potency of a formalinized polyvalent vaccine containing centrifugally purified Lee, PR8, and Weiss influenza virus materials at concentrations of 5, 2.5, and 2.5 mg. per cc., respectively, was found to be essentially the same as that of a similar vaccine prepared commercially. In both cases the protection afforded against the Weiss strain appeared to be better than that against the Lee and PR8 strains. The commercially prepared vaccine is being subjected to clinical tests in man at dosage levels ranging from 0.01 mg. to 10 mg. The latter corresponds to a level approximately 100 times that of infectious allantoic fluid. It was found that the bacterial contamination that frequently accompanies operation on a large scale can be controlled by the addition of one part per 10,000 of formalin plus one part per 100,000 of phenyl mercuric nitrate to the allantoic fluid immediately following harvesting, without affecting the quality of the vaccine. This procedure and the use of virus materials purified and concentrated by a single cycle of differential centrifugation by means of the Sharples centrifuge were found to be suitable for the production of influenza virus vaccines on a large scale. By means of this method influenza vaccines possessing 20 or more times the immunizing potency of infectious allantoic fluid and 10 or more times the immunizing potency of the usual commercial vaccine prepared by the red cell elution method can be manufactured rapidly on a very large scale with considerable ease and efficiency.

AMINO ACID COMPOSITION OF HIGHLY PURIFIED VIRAL PARTICLES OF INFLUENZA A AND B

Microbiological assays for amino acids were made on hydrolysates of four to five highly purified preparations each of influenza A virus (PR8 strain) and influenza B virus (Lee strain). The results of the assays indicated that these strains of influenza virus contain approximately the same amounts of alanine, aspartic acid, glycine, histidine, isoleucine, leucine, methionine, phenylalanine, proline, serine, threonine, and valine. However, significant differences were found in the values for arginine, glutamic acid, lysine, tryptophane, and tyrosine. It is believed that these differences may provide, at least in part, a chemical explanation for some of the differing properties of the PR8 and Lee strains of influenza viruses.

Vaccines for Pandemic Influenza. The History of our Current Vaccines, their Limitations and the Requirements to Deal with a Pandemic Threat

Fears of a potential pandemic due to A(H5N1) viruses have focussed new attention on our current vaccines, their shortcomings, and concerns regarding global vaccine supply in a pandemic. The bulk of current vaccines are inactivated split virus vaccines produced from egggrown virus and have only modest improvements compared with those first introduced over 60 years ago. Splitting, which was introduced some years ago to reduce reactogenicity, also reduces the immunogenicity of vaccines in immunologically naïve recipients. The A(H5N1) viruses have been found poorly immunogenic and present other challenges for vaccine producers which further exacerbate an already limited global production capacity. There have been some recent improvements in vaccine production methods and improvements to immunogenicity by the development of new adjuvants, however, these still fall short of providing timely supplies of vaccine for all in the face of a pandemic. New approaches to influenza vaccines which might fulfil the demands of a pandemic situation are under evaluation, however, these remain some distance from clinical reality and face significant regulatory hurdles. Key words: Adjuvant, Antigen, Cell-culture, Immune response, Immunogenicity, Influenza A(H5N1), Split vaccine

Downstream Processing: From Egg to Cell Culture-Derived Influenza Virus Particles

The establishment of cell culture-derived vaccine production requires the development of appropriate downstream processes. Until today, many of the downstream methods applied originate from egg-derived production processes. These methods have often been slightly modified in order to account for the new demands. However, efforts are currently underway to optimize these processes focusing, for example, on ion exchange or affinity based membrane adsorption chromatography. This review covers the main aspects relevant for the downstream processing of egg and mammalian cell culture-derived whole influenza viruses.

[Inactivated vaccine against trivalent influenza. Comparative study of the immune response by hemagglutination inhibition and simple radial hemolysis methods]

A vacina inativada contra gripe, trivalente, preparada no Instituto Butantan, contendo 200 unidades hemaglutinantes de cada uma das cepas de virus Influenza A/SP/1/80 (H3N2), A/SP/1/78 (H1N1) e B/England/847/73, foi administrada em 110 voluntários humanos adultos, sendo que 62 receberam uma dose de vacina e 48 duas doses, com intervalo de 21 dias. A resposta de anticorpos específicos para influenza foi analisada comparativamente pelos testes de Inibição da Hemaglutinação (IH) e Hemólise Radial Simples (HRS). Ocorreu aumento significativo do teor de anticorpos nos indivíduos vacinados, correspondente a um aumento de 4 vezes ou mais nos títulos obtidos pelo teste IH e a um aumento de 3,0 mm ou maior no diâmetro das zonas de hemólise pelo teste HRS. Os métodos demonstraram correlação satisfatória entre si.

Influenza virus purification with the zonal ultracentrifuge

Sufficient amounts of influenza virus (density, 1.185; size, 722S) can be highly purified (22,000 chicken cell agglutinating units per milligram of protein) with a zonal ultracentrifuge, used first in a rate process followed by isopycnic banding, to permit its detailed biological and physical-chemical evaluation.

Evaluation of some methods of concentration and purification of influenza virus

A number of techniques have been developed in the past 15 years for the concentration and purification of influenza virus, but there has been no comparative evaluation of their efficacy. In this paper the authors report on such an assessment, carried out with particular reference to the suitability of the various techniques for large-scale production of influenza vaccine, the aim being to recover a relatively pure antigen by simple means and without undue loss in the haemagglutinin and virus contents.The purest product was obtained with two cycles of aluminium phosphate treatment. Next in order came ultracentrifugation with and without methanol precipitation, red cell adsorption, single-cycle aluminium phosphate treatment, and finally the zinc hydroxide method.

Attempts to increase the yield of influenza virus from stored infected allantoic fluids

The deposit or precipitate formed after storage at -20 degrees C of allantoic fluids infected with influenza virus contains a considerable amount of the virus. It has been found that by treatment of this material, after thawing, with sodium chloride and subsequent purification of the treated fluid, all the virus originally present can be recovered, thus ensuring against loss of virus and irregular titration results. The authors describe in detail the method developed, which should prove useful in large-scale production of influenza vaccine when the material has to be stored in the deep-freeze for some time.

Serological and prophylactic trials of Asian influenza vaccines in Sweden, 1957

A serological trial of a formolized monovalent Asian influenza vaccine of conventional type, using a range of falling vaccine dilutions, showed a satisfactory antihaemagglutinin response in groups of volunteers vaccinated with relatively low numbers of haemagglutinating units (HU) of the vaccine.Booster injections, given 15-19 days after the primary vaccination, gave a further increase and more persistence of antibody response.In a field protection trial, performed in more than 1000 recruits who received a similar vaccine prepared from another line of the same virus strain, a good degree of protection (57%-78%) was obtained after a single injection of 2000 HU of aqueous vaccine. A lower but still statistically significant degree of protection (29%-45%) obtained with only 400 HU of the vaccine adsorbed on aluminium phosphate. Two injections or 2000 or 400 HU of vaccine (without AlPO(4)), given 9 days apart, protected, respectively, 69%-86% and 45%-51% of the vaccinated. The slightly better protective effect of the two-injection scheme appears to be due to the double amount of vaccine given rather than to a recall reaction.The numerical results of both trials were used for an evaluation of the quantitative relation between the antibody-stimulating and protective effects of influenza vaccines.Some pitfalls in the current method of expressing the potency of influenza vaccines in terms of their haemagglutinin content are discussed.

PRECIPITIN REACTIONS OF HIGHLY PURIFIED INFLUENZA VIRUSES AND RELATED MATERIALS

Antisera to purified PR8 virus, to purified protein from normal allantoic fluid, and to purified normal mouse lung particles were obtained from hyper-immunized rabbits and used in quantitative precipitin tests employing various purified preparations of influenza virus and related materials as antigens. The results of those tests indicated that the most highly purified preparations of PR8 or of Lee influenza virus obtained from infectious allantoic fluid contain an antigen characteristic of normal allantoic fluid and likewise that highly purified mouse lung PR8 virus contains an antigen characteristic of normal mouse lungs. Since the infectivity of virus preparations which were ultracentrifugally and electrochemically homogeneous was precipitated by the appropriate antisera to normal antigens, it was concluded that the normal antigens constitute a part of the 100 mmicro particles with which influenza virus activity is at present deemed to be associated. It was estimated from quantitative precipitin data that the most highly purified preparations of PR8 and of Lee influenza viruses obtained from infectious allantoic fluid contain at least about 20 and 30 per cent, respectively, of an antigenic structure characteristic of the sedimentable protein of normal allantoic fluid.

THE PREPARATION OF HIGHLY PURIFIED PR8 INFLUENZA VIRUS FROM INFECTED MOUSE LUNGS

Highly purified preparations of PR8 influenza virus were obtained from perfused, infected mouse lungs by a combination of methods involving adsorption of the virus on and elution from chicken red cells and differential centrifugation. Such preparations were found to possess 50 per cent infectivity end-points at 10(-11) to 10(-11.8), and 10(-13) to 10(-14.3) gm. of nitrogen in mice and in chick embryos, respectively. A sedimentation constant of 683 S was obtained for the material and electron micrographs revealed essentially spherical particles about 100 mmicro in diameter. The material was isoelectric at pH 5.4 and chemical analyses on several of the preparations indicated the presence of 10.1 per cent of nitrogen, 1.06 per cent of phosphorus, 6.2 per cent of carbohydrate and about 33 per cent of lipid. Positive tests were obtained for both ribonucleic and desoxyribonucleic acids. The virus was precipitated strongly by antisenim to purified PR8 virus obtained from the allantoic fluid of infected chick embryos and this serum inhibited the agglutination of red cells by the mouse virus to a dilution of about 40,000. In general, the properties of the virus isolated from infected mouse lungs were found to coincide with those of the virus obtained from the allantoic fluid of infected chick embryos.