WO2005113756A1

WO2005113756A1 - Method

Info

Publication number: WO2005113756A1
Application number: PCT/EP2005/005262
Authority: WO
Inventors: Emmanuel Hanon; Elisabeth Neumeier; Florence Nozay
Original assignee: Glaxosmithkline Biologicals S.A.; Saechsisches Serumwerk Dresden Branch Of Smithkline Beecham Pharma Gmbh & Co Kg
Priority date: 2004-05-14
Filing date: 2005-05-12
Publication date: 2005-12-01

Abstract

The invention relates to a method to produce influenza virus or influenza virus antigens in embryonated avian eggs. In particular, the invention relates to the introduction into the embryonated avian eggs of an influenza virus and any of the following components: a serine protease, such as trypsin, at least one amino acid or derivative thereof, insulin-like growth factor or type I interferon antagonist, to increase the yield of influenza virus or influenza virus antigens produced. The invention further relates to influenza virus obtainable by said process, to compositions comprising said virus, to their use in medicine and to prophylactic and/or therapeutic methods using said virus.

Description

METHOD

TECHNICAL FIELD

The present invention relates to a method to produce influenza virus or influenza virus antigens in embryonated avian eggs. In particular, the invention relates to the introduction into the embryonated avian eggs of an influenza virus and any of the following components: a serine protease, such as trypsin, at least one amino acid or derivative thereof, insulin-like growth factor or type I interferon antagonist, to increase the yield of influenza virus or influenza virus antigens produced.

TECHNICAL BACKGROUND

Influenza virus is one of the most ubiquitous viruses present in the world, affecting both humans and livestock. The economic impact of influenza is significant.

The influenza virus is an RNA enveloped virus with a particle size of about 125 nm in diameter. It consists basically of an internal nucleocapsid or core of ribonucleic acid (RNA) associated with nucleoprotein, surrounded by a viral envelope with a lipid bilayer structure and external glycoproteins. The inner layer of the viral envelope is composed predominantly of matrix proteins and the outer layer mostly of the host-derived lipid material. The surface glycoproteins neuraminidase (NA) and haemagglutinin (HA) appear as spikes, 10 to 12 nm long, at the surface of the particles. While antigenic differences between type A and B influenza strains is characterized by differences in their matrix or nucleoprotein, the antigenicity of influenza A subtypes is defined by differences in their surface proteins

Typical influenza epidemics cause increases in incidence of pneumonia and lower respiratory disease as witnessed by increased rates of hospitalization or mortality. The elderly or those with underlying chronic diseases are most likely to experience such complications, but young infants also may suffer severe disease. These groups in particular therefore need to be protected. Influenza vaccines are classically trivalent and can take the form of either live attenuated vaccines or inactivated formulations. They usually contain antigens derived from two influenza A virus strains and one influenza B virus strain. A standard injectable dose in most cases contain 15 μg of haemagglutinin (HA) antigen component from each strain, as measured by single radial immunodiffusion (SRD), usually in a 0.5 ml dose (J.M. Wood et al.: An improved single radial immunodiffusion technique for the assay of influenza haemagglutinin antigen: adaptation for potency determination of inactivated whole virus and subunit vaccines. J. Biol. Stand. 5 (1977) 237-247; J. M. Wood et al., International collaborative study of single radial diffusion and immunoelectrophoresis techniques for the assay of haemagglutinin antigen of influenza virus. J. Biol. Stand. 9 (1981) 317-330).

In certain circumstances, such as the occurrence of a pandemic influenza strain, it may be desirable to have a vaccine which contains only a single strain. This will help the speed of response to a pandemic situation. The influenza virus strains to be incorporated into influenza vaccine each season are determined by the World Health Organisation in collaboration with national health authorities and vaccine manufacturers.

Currently available influenza vaccines are either inactivated influenza vaccines or live attenuated influenza vaccines. Inactivated influenza vaccines comprise one of three types of antigen preparation: inactivated whole virus, sub-virions where the membrane of the virus particles is disrupted with detergents or other reagents to solubilise the lipid envelope (so-called 'split' vaccine) or subunits vaccines (produced either recombinantly or purified from disrupted viral particles) in particular HA and NA subunit vaccine. Virosomes are another type of inactivated influenza formulation, where the viral membrane is reconstituted following disruption. These inactivated vaccines are generally given parenterally, in particular intramuscularly (i.m.), but some virosome-based formulations and live attenuated vaccines have been administered intranasally.

As said above, the currently commercially available influenza vaccines remain the intramuscularly administered split vaccine, whole virus vaccine, subunit injectable or virosomes vaccines. In particular, the split and subunit vaccines are prepared by disrupting the virus particle, generally with an organic solvent or a detergent, and separating or purifying the viral proteins to varying extents. Split vaccines are prepared by fragmentation of whole influenza virus, either infectious or inactivated, with solubilizing concentrations of organic solvents or detergents and subsequent removal of the solubilizing agent and some or most of the viral lipid material. Split vaccines generally contain contaminating matrix (M) protein and nucleoprotein (NP) and sometimes lipid, as well as the membrane envelope proteins (such as HA and NA). Split vaccines will usually contain most or all of the virus structural proteins although not necessarily in the same proportions as they occur in the whole virus. Subunit vaccines on the other hand consist essentially of highly purified viral surface proteins, haemagglutinin (HA) and neuraminidase (NA), which are the surface proteins responsible for eliciting the desired virus neutralising antibodies upon vaccination. Matrix and nucleoproteins are either not detectable or barely detectable in subunit vaccines.

Standards are applied internationally to measure the efficacy of influenza vaccines. The European Union official criteria for an effective vaccine against influenza are set out in the table below. Theoretically, to meet the European Union requirements, an influenza vaccine has to meet only one of the criteria in the table, for all strains of influenza included in the vaccine. However in practice, at least two or more probably all three of the criteria will need to be met for all strains, particularly for a new vaccine such as a new intradermal vaccine. Under some circumstances two criteria may be sufficient. For example, it may be acceptable for two of the three criteria to be met by all strains while the third criterion is met by some but not all strains (e.g. two out of three strains). The requirements (Note for Guidance on harmonisation of requirements for influenza vaccines, CPMP/BWP/214/96, 12 March 1997) are different for adult populations (18-60 years) and elderly populations (>60 years): a) the following serological assessments should be considered for each strain in adult subjects, aged between 18 and 60, and at least one of the assessments should meet the indicated requirements: • number of seroconversions or significant increase in antihaemagglutinin antibody titre > 40%; • mean geometric increase > 2.5 • the proportion of subjects achieving an HI titre ≥40 or SRH titre >25 mm² should be > 70% b) the following serological assessments should be considered for each strain in adult subjects aged over 60, and at least one of the assessments should meet the indicated requirements: • number of seroconversions or significant increase in antihaemagglutinin antibody titre > 30%; • mean geometric increase > 2.0 • the proportion of subjects achieving an HI titre ≥40 or SRH titre >25 mm² should be > 60%

For an influenza vaccine to be commercially useful it will not only need to meet those standards, but also in practice it will need to be at least as immunologically efficacious as the currently available vaccines. It will need to be administered using a procedure that is reliable and straightforward for medical staff to carry out. Furthermore, it will also need to be produced by an acceptable process and will of course need to be commercially viable in terms of the amount of antigen and the number of administrations required.

The yield of influenza virus in eggs is critical to influenza vaccine production and availability, but the contribution of specific genes to the growth properties of influenza A and B viruses is not well understood. WHO aims to increase routine vaccination of all people at high risk - including at least 50% of the Elderly population - by 2006 (Clayton, et al. Nature Medicine, 2003, 9, 375). This implies that current production capacity of Influenza vaccine has to substantially increase. Among different possibilities, increasing yield of influenza virus production and/or influenza virus antigens in egg is a strategy of choice. For example, as the yield from one egg is commonly enough to produce influenza vaccine for about one to 1.5 doses, an increase of two-fold in the production of influenza virus, and in the production of haemagglutinin in particular, would already result in a decrease of 50% in the number of eggs required to manufacture the same amount of influenza vaccine. This may in particular represent a significant advantage in the case of pandemics preparedness when the output of vaccine doses may then be increased using the available manufacturing resources.

It has been reported that administration of a solution of synthetic amino acids into 7-days old eggs can substantially increase embryonic cellular growth and amino acid incorporation (Ohta et al. Poult. Sci., 2001 , 80, 1430-1436). It has also been reported that administration of insulin-like growth factor protects embryo from stress and can have a positive impact on growth of the embryo (Girbau et al., Endocrinology, 1987, 121 , 1477- 1482). It has further been reported that a functional haemagglutinin (HA) requires cleavage by a proteolytic enzyme such as trypsin. The latter is used for increasing the production of influenza virus in cell line in vitro (Tree et al., Vaccine, 2001 , 19, 3444- 3450). It is well known that the action of type I interferons (IFNs) on virus-infected cells and surrounding tissues elicits an antiviral state that is characterized by the expression and antiviral activity of IFN-stimulated genes (Katze MG, He Y, Gale M Jr. Nat Rev Immunol. 2002 Sep; 2(9) 675-87). In this regard, it has also been reported that the biological activity of type I interferons can be inhibited by using antagonist such as soluble receptor (B18R protein from Vaccinia virus) (Alcami A, Symons JA, Smith GL J Virol. 2000 Dec; 74 (23): 11230-9).

No reference however has ever suggested that introduction, into embryonated eggs, of any of the compounds described above and an influenza virus would result in an enhanced yield of influenza virus or to an enhanced yield of influenza viral antigens, in particular of haemagglutinin.

DESCRIPTION OF FIGURES

Figure 1 - Total content of HA units at three steps of the purification process

Figure 2 - Total protein content at three steps of the purification process

Figure 3 - Total content of HA units at three steps of the purification process

Figure 4 - Total protein content at three steps of the purification process

Figure 5 - HA Titer (Figure 5A) and protein concentration (Figure 5B)

Figure 6 - Flowchart for the selection of suitable amino acid combinations (choice of the amino acids, optimization of their concentrations, application to other strains)

Figure 7 - Best predicted combination (higher than 16 263, i.e. higher than the 95% confidence interval upper limit of control mean obtained in the experiment of Example V) for a given number of amino acids.

STATEMENT OF THE INVENTION

The present inventors have discovered that introducing certain additives in embryonated avian eggs had a positive impact not only on growth of the embryo, but also could lead to an increase of the yield of virus production and/or influenza viral antigen production, in particular to an increase in haemagglutinin production. In particular, the additives are selected from the list consisting of at least one amino acid or derivative thereof, a serine protease, insulin-like growth factor, type-1 interferon antagonist, and a combination of one of more of these compounds.

Accordingly, in a first aspect of the present invention, there is provided a method for producing influenza virus in embryonated avian eggs, comprising introducing into said eggs, an influenza virus and at least another component wherein said component is selected from the list consisting of: a serine protease, at least one amino acid or derivative thereof, insulin-like growth factor, type 1 interferon antagonist, and a combination of one or more thereof. Said 'at least one amino acid' may be selected from selected from the list in Table 1 , or may be any derivative of said amino acid.

In another aspect, there is provided a method for producing enhanced levels of haemagglutinin in embryonated avian eggs, comprising introducing into said eggs, an influenza virus and at least another component wherein said component is selected from the list consisting of: a serine protease, at least one amino acid or derivative thereof, insulin-like growth factor, type 1 interferon antagonist, and a combination of one or more thereof.

In a further aspect, the invention provides a method for producing an influenza virus vaccine comprising the steps of:

(a) producing influenza virus according to the method of the present invention,

(b) harvesting the virus from the eggs,

(c) purifying the virus, and

(d) formulating the virus of step (c) with a pharmaceutically acceptable carrier or excipient.

In a yet further aspect, the invention provides method for prophylaxis of influenza infection or disease in a subject which method comprises administering to the subject a vaccine produced according to the method claimed herein.

There is also provided by this invention the use of a vaccine as claimed herein or produced according to the method according to the invention, in the manufacture of a medicament to elicit an immune response to an influenza antigen in a patient susceptible to influenza infection.

Other aspects and advantages of the present invention are described further in the following detailed description of the preferred embodiments thereof. DETAILED DESCRIPTION

The present invention provides for a method for producing influenza virus or antigens thereof in embryonated avian eggs, comprising introducing into said eggs, an influenza virus and at least another component wherein said component is selected from the list consisting of: a serine protease, an amino acid or derivative thereof, insulin-like growth factor, type 1 interferon antagonist, and a combination of one or more thereof. In one embodiment, the claimed method produces enhanced levels of virus and/or influenza virus antigen yield obtained from embryonated eggs so treated as assessed by comparison with the virus and/or influenza virus antigen yield obtained from similar eggs which are handled in the same manner as the treated eggs but which are not subjected to the same procedure of exposure or contact with any of the additives or components listed above. For example, the virus or virus antigen yield obtained from eggs treated according to the process claimed (so-called the 'treated eggs') in the present invention is higher than the virus or virus antigen yield obtained from similar eggs (so-called the 'control eggs') which have been in contact with a standard solution, i.e. a solution similar to that used for the treated eggs, but which is devoid of any of the described components. The standard solution may be water or a buffered solution such as PBS for example.

Accordingly, in a specific embodiment, there is provided a method for producing influenza virus in embryonated avian eggs, comprising introducing into said eggs, an influenza virus and at least another component wherein said component is selected from the list consisting of: a serine protease, at least one amino acid or derivative thereof, insulin-like growth factor, type 1 interferon antagonist, and a combination of one or more thereof, wherein said eggs produce enhanced levels of influenza virus and/or antigen thereof compared to control eggs which have not been in contact with said additive(s).

In one specific embodiment, the influenza virus yield obtained from the treated eggs is higher than that obtained from the control eggs. It may be evaluated on the allantoic fluid after clarification (see example I, section 1.3) at different steps during the purification process, as assessed by the haemagglutinin titer (HA titer) and/or total protein content of purified virus at each process step.

In a specific embodiment, the influenza virus antigen(s) yield obtained from the treated eggs is higher than that obtained from the control eggs. In a specific embodiment, said influenza virus antigen is haemagglutinin (HA). Suitably then the influenza virus antigen yield is expressed as HA titer.

Another suitable mean to measure the effect of the additive may be performed by multiplying the number of harvested eggs by the protein content of harvested eggs. This product is referred to as 'response' in Example V of the document.

By higher yield or enhanced level of virus and/or virus antigen, both expression being used interchangeably, is meant a ratio of said virus yield or said virus antigen yield from the treated eggs relative to that from the control eggs higher than 1 as assessed by any suitable method as detailed herein above. The level of virus and/or virus antigen for the control may advantageously be a control mean when several controls have been run simultaneously, as it gives a higher precision to the ratio. Suitably a ratio of at least 1.2:1 (treated: control or control mean) is obtained. Particularly suitable is a minimum ratio of 1.5:1. Suitably a ratio of at least 2:1 , a ratio of at least 3:1 , a ratio of at least of 5: 1 is obtained.

The introduction of the component according to the invention may be done by any suitable way, such as inoculation with a needle or injection using a standardized apparatus.

When inoculated, the avian embryonated eggs are usually 9 to 12 days old, typically 10 or 1 1 days old. Suitably they are chicken eggs.

In one embodiment, the component is introduced in embryonated avian eggs, preferably concomitantly with the influenza virus, typically in the form of a solution, preferably as part of the same solution so as only one inoculation is needed. Alternatively the embryonated eggs may be injected with the component according to the invention, and then subsequently infected with influenza virus, or vice versa. Both the virus and the other component may be introduced separately or in the form of a mixture either in the allantoic liquid, or in the amniotic liquid, or in other egg compartment or directly in the embryo. Suitably the inoculation takes place in the allantoic fluid.

As used herein, the term influenza virus includes any viruses capable of causing a febrile disease state in animals (including human) characterised by respiratory symptoms, inflammation of mucous membranes and often systemic involvement. The method according to the present invention is especially useful in the production of a variety of influenza viruses, including equine, human, porcine, and avian strains. Influenza virus include any strains, subtypes and types, particularly those recommended by WHO, many of which are available from clinical specimens and from public depositories such as ATCC, ECCAC and BIBSC. Typically type A, type B and/or type C influenza viruses are contemplated in the present invention. Influenza strains which can be consistently propagated to high titers are especially contemplated. In particular, reassortant influenza viruses and influenza viruses exhibiting chime c viral surface molecules such as chimeric HA and/or NA surface molecules are contemplated (such as those described in WO 94/29439 for example). Recombinant influenza virus strains are also contemplated (such as those described in WO 91/03552 for example).

In one embodiment, said at least one amino acid is selected from the list of amino acids given in Table 1 or is a derivative of said amino acid. Suitably at least two, at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine, at least ten, at least eleven, at least twelve, at least thirteen amino acids are present. The minimum number of amino acids can be routinely determined experimentally. In particular, the number of amino acids will be determined so as to lead to a fixed minimum higher yield or enhanced level of virus and/or virus antigen compared to the yield obtained in the absence of said amino acids. What is meant by 'higher yield' is explained herein above. By way of illustration, non-limiting examples of suitable solutions of amino acids are found in Table 9 which exhibits results obtained with tested solutions, and in Table 12 which exhibits solutions predicted by the model detailed in example V. The solution(s) can be chosen by reference to the ratio treated: control or control mean which is desirable to be achieved (i.e. higher than 1) as illustrated in the last column of both tables, and taking into account experimental parameters such as the influenza virus strain, the amino acid concentration in the solution, etc. It may also be advantageous to select, between two solutions giving a similar virus or virus antigen yield, the solution(s) which have only a limited number of amino acids, so as to minimize the interaction between the amino acids to be included in said solution(s). For example, from the results illustrated in Table 9, 24 solutions are possible candidates giving a virus yield for treated: control higher than 1. Solutions 2 and 46 give similar yields (ratio treated: control of 1.44) but solution 46 might be preferred as it only contains 1 amino acid by comparison to solution 2 which contains 8 amino acids. If however a maximum yield (for example a ratio treated: control of above 4) is desirable then it may be desirable to select that or those solution(s) which allow(s) achieving said higher yield (such as solution 42, with a ratio of 4.6) even though more amino acids (10 for solution 42) may need to be included in the solution. It is to be understood that the same procedure can routinely be applied to different experimental conditions (change of virus strain, change in amino acid concentrations) and may results into different solution profiles.

In a specific embodiment, all amino acids from the list in Table 1 are present. Table 1 also shows the ranges of concentration for the amino acids, when present, which are appropriate. Optimization of said amino acid concentrations will be routinely performed.

In one embodiment, the serine protease is selected from the list consisting of: trypsin-like protease (such as rProtease, from Invitrogen, 3175 Staley Road, Grand Island, NY 14072, supplier catalogue number 02-106), porcine or bovine trypsin, or recombinant origin (such as Trypzean, a recombinant trypsin produced in corn, Prodigen, 101 Gateway Blvd, Suite 100 College Station, Texas 77845. Manufacturer code : TRY). Proteases from the trypsin-like protease family are commonly found in prokaryotes, animals and viruses. These enzymes participate in diverse physiological processes, the best known among them are digestions, fertilisation, blood clotting cascade and developmental processes. It is thought that they diverged from a common ancestral protein. These enzymes have been extensively described in the literature (A.J. Greer, "Comparative modelling methods - application to the family of mammalian serine proteases" Proteins, Vol. 7, p 317-334, 1990) and can be divided into different families bases on their structure (A. Sali & T. Blundell, "definition of general topological equivalence in protein structures" J. Mol. Biol., 212, p 403-428, 1990). A suitable protease is a serine protease such as recombinant trypsin or trypsin-like protease.

Suitably trypsin is used in the form of a solution, and used at an amount of 6.69 USP per egg. A typical range is from 0.067 to 670 USP, preferably 0.669 - 66.96 USP (standard unit for trypsin activity).

Other suitable serine proteases are from the family of subtilisin-like proteases, such as chymotrypsin, thermolysin, pronase or subtilisin A which are particularly suitable.

A suitable Insulin-like growth factor type is IGF-I used at a quantity of 0.4 - 40 μg per egg, typically at an amount of 4 - 20 μg per egg, preferably at an amount of 4 μg or 20 μg per egg. A suitable type I interferon antagonist is B18R protein from Vaccinia virus (Alcami A, Symons JA, Smith GL J Virol. 2000 Dec; 74 (23): 11230-9) used at a amount of 1 to 1000 μg per egg, suitably at an amount from 5 to 100, or 100 μg per egg. Functionally equivalent type I interferon antagonist are also contemplated.

It is possible to use a wide range of concentration or amount of the protease, the amino acid, the insulin-like growth factor or the type-1 interferon antagonist. The quantity can be optimised by routine experiments.

Typically the protease, the amino acid, the insulin-like growth factor or the type-1 interferon antagonist are used in the form of an aqueous solution. In particular, the amino acid solution for inoculation into the egg exhibits a pH from at least 6 to maximum 11. A pH range of 7 to 1 1 , in particular a pH range of 8 -10 is particularly suitable. An amino acid solution with a pH ranging from 9 to 10, Suitably with a pH at around 9.5 is especially contemplated. Typically, a pH at 'around of 9.5' is meant to include any pH ranging from 9.25 to 9.75 which will be understood to be contemplated.

Combinations of the various components described above are also contemplated in the present invention. In particular combinations of two or more of the components are sought. Typically such a combination will comprise at least one amino acid in admixture with a serine protease, in particular with trypsin. Another combination will comprise at least one amino acid in admixture with either IGF-1 , or type I interferon antagonist or in admixture with both IGF-1 and type I interferon antagonist.

The contact with the component according to the invention, whether given alone or in admixture with the virus may be carried out for periods of time varying from an hour to up to several days, suitably from one hour to 96 hours, in particular from 12 hours to 72 hours, typically for 24 hours or alternatively for 48 hours. A typical, although not limiting, concentration of the component is as depicted in Table 1 (third column) for the amino acids, when present, and/or 6.69 USP for the trypsin, and a typical period of incubation in the embryonated eggs is 48 hours, and for production, as a rule it is 72 hours.

Enhanced levels of influenza virus may be determined by measuring the increased yield of haemagglutinin (HA). By 'increased yield of HA' is meant an amount of HA, as determined by a quantitative assay, including SRD assay, which is higher than that obtained in the absence of injection in the embryonated egg of any of the component selected from the list comprising: a serine protease, at least one amino acid or derivative thereof, insulin-like growth factor, type 1 interferon and a combination of one or more thereof. The yield, and the increase in the yield, of influenza antigen may be determined by measuring the haemagglutinin titer per ml (or in HA units per ml) using suitable methods (Kendal, A.P., Pereira, M.S. and Skehel, J. 1982. Concepts and procedures for laboratory-based influenza surveillance, distributed by the viral diseases unit, WHO, Geneva, or the WHO Collaborating Center for the Surveillance, Epidemiology and Control of Influenza, Centers for Disease Control and Prevention, Atlanta, GA 30333, U.S.A.). Suitably, the haemagglutinin antigen component from each influenza strain, may be measured by single radial immunodiffusion (SRD) (J.M. Wood et al.: An improved single radial immunodiffusion technique for the assay of influenza haemagglutinin antigen: adaptation for potency determination of inactivated whole virus and subunit vaccines. J. Biol. Stand. 5 (1977) 237-247; J. M. Wood et al., International collaborative study of single radial diffusion and immunoelectrophoresis techniques for the assay of haemagglutinin antigen of influenza virus. J. Biol. Stand. 9 (1981 ) 317-330).

Also provided by the present invention is a method for producing an influenza virus or virus antigen vaccine, said method comprising the steps of:

(a) producing influenza virus or antigen thereof according to the method claimed herein, (b) harvesting the virus or antigen thereof from the eggs,

(c) purifying the virus or antigen thereof, and

(d) formulating the virus or antigen thereof of step (c) with a pharmaceutically acceptable carrier or excipient.

Alternatively the virus or antigen thereof may be harvested from the amniotic fluid of the injected embryonated eggs, from the embryo or from the allantoic fluid, which is particularly suitable.

The influenza vaccines according to the invention may be either whole, or split or subunit vaccines. Split and subunit vaccines are prepared by disrupting the virus particle, generally with an organic solvent or a detergent, and separating or purifying the viral proteins to varying extents. Split vaccines are prepared by fragmentation of whole influenza virus, either infectious or inactivated, with solubilizing concentrations of organic solvents or detergents and subsequent removal of the solubilizing agent and some or most of the viral lipid material. Split vaccines generally contain residual matrix protein and nucleoprotein and sometimes lipid, as well as the membrane envelope proteins. Split vaccines will usually contain most or all of the virus structural proteins although not necessarily in the same proportions as they occur in the whole virus. Subunit vaccines on the other hand consist essentially of highly purified viral surface proteins, haemagglutinin and neuraminidase, which are the surface proteins responsible for eliciting the desired virus neutralising antibodies upon vaccination.

The influenza virus antigen preparation may be produced by any of a number of commercially applicable processes, for example the split flu process described in patent no. DD 300 833 and DD 211 444, incorporated herein by reference. In particular the split flu vaccine is produced according to the methods disclosed in WO 02/097072, the disclosure of which is incorporated herein by reference. Traditionally split flu was produced using a solvent/detergent treatment, such as tri-n-butyl phosphate, or diethylether in combination with Tween™ (known as "Tween-ether" splitting) and this process is still used in some production facilities. Other splitting agents now employed include detergents or proteolytic enzymes or bile salts, for example sodium deoxycholate as described in patent no. DD 155 875, incorporated herein by reference.

Alternatively, the influenza virus may be in the form of a whole virus vaccine. This may prove to be an advantage over a split virus vaccine for a pandemic situation as it avoids the uncertainty over whether a split virus vaccine can be successfully produced for a new strain of influenza virus. For some strains the conventional detergents used for producing the split virus can damage the virus and render it unusable. Although there is always the possibility to use different detergents and/or to develop a different process for producing a split vaccine, this would take time, which may not be available in a pandemic situation. In addition to the greater degree of certainty with a whole virus approach, there is also a greater vaccine production capacity than for split virus since considerable amounts of antigen are lost during additional purification steps necessary for preparing a suitable split vaccine.

In another embodiment, the influenza virus preparation is in the form of a purified sub-unit influenza vaccine. Sub-unit influenza vaccines generally contain the two major envelope proteins, HA and NA, and may have an additional advantage over whole virion vaccines as they are generally less reactogenic, particularly in young vaccinees. Sub-unit vaccines can produced either recombinantly or purified from disrupted viral particles. In another embodiment, the influenza virus preparation is in the form of a virosome. Virosomes are spherical, unilamellar vesicles which retain the functional viral envelope glycoproteins HA and NA in authentic conformation, intercalated in the virosomes' phospholipids bilayer membrane.

Detergents that can be used as splitting agents include cationic detergents e.g. cetyl trimethyl ammonium bromide (CTAB), other ionic detergents e.g. laurylsulfate, taurodeoxycholate, or non-ionic detergents such as the ones described above including Triton X-100 (for example in a process described in Lina et al, 2000, Biologicals 28, 95- 103) and Triton N-101 , or combinations of any two or more detergents.

The preparation process for a split vaccine will include a number of different filtration and/or other separation steps such as ultracentrifugation, ultrafiltration, zonal centrifugation and chromatography (e.g. ion exchange) steps in a variety of combinations, and optionally an inactivation step eg with heat, formaldehyde or β-propiolactone or UN. which may be carried out before or after splitting. The splitting process may be carried out as a batch, continuous or semi-continuous process.

Suitable split flu vaccine antigen preparations according to the invention comprise a residual amount of Tween 80 and/or Triton X-100 remaining from the production process, although these may be added or their concentrations adjusted after preparation of the split antigen. Suitably both Tween 80 and Triton X-100 are present. Suitable ranges for the final concentrations of these non-ionic surfactants in the vaccine dose are: Tween 80: 0.01 to 1 %, in particular about 0.1 % (v/v) Triton X-100: 0.001 to 0.1 (% w/v), in particular 0.005 to 0.02% (w/v).

A micelle modifying excipient, in particular α-tocopherol or a derivative thereof as a haemagglutinin stabiliser may be used in the manufacture of an influenza vaccine. Suitably the α-tocopherol is in the form of an ester, such as the succinate or acetate and in particular the succinate. Suitable concentrations for the α-tocopherol or derivative are between 1 μg/ml - 10mg/ml, in particular between 10μg/ml - 500 μg/ml.

The vaccine according to the invention generally contains both A and B strain virus antigens, typically in a trivalent composition of two A strains and one B strain. However, tetravalent, divalent and monovalent vaccines are not excluded. Monovalent vaccines, containing antigens from a single type or subtype of influenza virus (such as e.g. H1 N1 , H2N2, H3N2 of type A, type B and type C) may be advantageous in a pandemic situation, for example, where it is important to get as much vaccine produced and administered as quickly as possible. Typically, for a divalent or a trivalent vaccine, monovalent bulks are produced and subsequently pooled together to produce the divalent or trivalent vaccine. Vaccines containing reassortant influenza viruses and influenza viruses exhibiting chimeric viral surface molecules such as chimeric HA and/or NA surface molecules, as well as vaccine containing recombinant influenza virus strains are also contemplated.

In a specific embodiment, the virus strain is associated to a pandemic outbreak or has the potential to be associated to a pandemic outbreak.

By way of background, during inter-pandemic periods, influenza viruses circulate that are related to those from the preceding epidemic. The viruses spread among people with varying levels of immunity from infections earlier in life. Such circulation, over a period of usually 2-3 years, promotes the selection of new strains that have changed enough to cause an epidemic again among the general population; this process is termed 'antigenic drift'. 'Drift variants' may have different impacts in different communities, regions, countries or continents in any one year, although over several years their overall impact is often similar. In other words, an influenza pandemics occurs when a new influenza virus appears against which the human population has no immunity. Typical influenza epidemics cause increases in incidence of pneumonia and lower respiratory disease as witnessed by increased rates of hospitalisation or mortality. The elderly or those with underlying chronic diseases are most likely to experience such complications, but young infants also may suffer severe disease.

At unpredictable intervals, novel influenza viruses emerge with a key surface antigen, the haemagglutinin, of a totally different subtype from strains circulating the season before. This phenomenon is called 'antigenic shift'. It is thought that at least in the past pandemics have occurred when an influenza virus from a different species, such as an avian or a porcine influenza virus, has crossed the species barrier. If such viruses have the potential to spread from person to person, they may spread worldwide within a few months to a year, resulting in a pandemic.

Accordingly, in an aspect of the present invention, the influenza virus strain is associated with a pandemic outbreak or has the potential to be associated with a pandemic outbreak. Such strain may also be referred to as 'pandemic strains' in the text below. Suitable strains are, but not limited to: H5N1 , H9N2, H7N7, H2N2 and H1 N1.

The vaccine may be administered by any suitable delivery route, such as intramuscular, subcutaneous, intradermal or mucosal e.g. intranasal or oral. Other delivery routes are well known in the art.

Any suitable device may be used for subcutaneous delivery, including devices with or without needle. Needleless syringes that are suitable for subcutaneous administration are for example described in the following patent families: WO 01/05454, WO 01/05453, WO 01/05452, WO 01/05451 , EP1090651 , WO 01/32243, WO 01/41840, WO 01/41839, WO 01/47585, WO 01/56637, WO 01/58512, WO 01/64269, WO 01/78810, WO 01/91835, WO 02/09796, WO 02/34317, WO 04/084977, WO 04/084975, WO 04/093944, WO 04/93947 all incorporated herein by reference. Functional equivalents of these needleless syringe are also contemplated by the present invention.

Any suitable device may be used for intradermal delivery, for example short needle devices such as those described in US 4,886,499, US 5,190,521 , US 5,328,483, US 5,527,288, US 4,270,537, US 5,015,235, US 5,141 ,496, US 5,417,662. Intradermal vaccines may also be administered by devices which limit the effective penetration length of a needle into the skin, such as those described in WO99/34850 and EP1092444, incorporated herein by reference, and functional equivalents thereof. Also suitable are jet injection devices which deliver liquid vaccines to the dermis via a liquid jet injector or via a needle which pierces the stratum corneum and produces a jet which reaches the dermis. Jet injection devices are described for example in US 5,480,381 , US 5,599,302, US

5,334,144, US 5,993,412, US 5,649,912, US 5,569,189, US 5,704,911 , US 5,383,851 , US 5,893,397, US 5,466,220, US 5,339,163, US 5,312,335, US 5,503,627, US 5,064,413, US 5,520, 639, US 4,596,556US 4,790,824, US 4,941 ,880, US 4,940,460, WO 97/37705 and WO 97/13537. Also suitable are ballistic powder/particle delivery devices which use compressed gas to accelerate vaccine in powder form through the outer layers of the skin to the dermis. Additionally, conventional syringes may be used in the classical mantoux method of intradermal administration. However, the use of conventional syringes requires highly skilled operators and thus devices which are capable of accurate delivery without a highly skilled user are preferred. The influenza antigen preparation for use in the invention may be selected from the group consisting of split virus antigen preparations, subunit antigens (either recombinantly expressed or prepared from whole virus), inactivated whole virus which may be chemically inactivated with e.g. formaldehyde, β-propiolactone or otherwise inactivated e.g. UN. or heat inactivated. Suitably the antigen preparation is either a split virus preparation, or a subunit antigen prepared from whole virus, particularly by a splitting process followed by purification of the surface antigen. Particularly suitable preparations are split virus preparations.

Suitably the concentration of haemagglutinin antigen for the or each strain of the influenza virus preparation is 1-1000 μg per ml, such as 3-300 μg per ml and in particular about 30 μg per ml (15 μg per 0.5 ml dose), as measured by a SRD assay. Lower doses are also contemplated in particular in the pandemics situation. In any case, an immunologically effective amount of viral antigen will be used, that is an amount of antigen which induces a protective immune response.

The vaccine according to the invention may further comprise an adjuvant or immunostimulant such as but not limited to detoxified lipid A from any source and non- toxic derivatives of lipid A, saponins and other reagents capable of stimulating a TH1 type response.

It has long been known that enterobacterial lipopolysaccharide (LPS) is a potent stimulator of the immune system, although its use in adjuvants has been curtailed by its toxic effects. A non-toxic derivative of LPS, monophosphoryl lipid A (MPL), produced by removal of the core carbohydrate group and the phosphate from the reducing-end glucosamine, has been described by Ribi et al (1986, Immunology and Immunopharmacology of bacterial endotoxins, Plenum Publ. Corp., ΝY, p407-419) and has the following structure:

A further detoxified version of MPL results from the removal of the acyl chain from the 3- position of the disaccharide backbone, and is called 3-O-Deacylated monophosphoryl lipid A (3D-MPL). It can be purified and prepared by the methods taught in GB 2122204B, which reference also discloses the preparation of diphosphoryl lipid A, and 3-O- deacylated variants thereof.

A suitable form of 3D-MPL is in the form of an emulsion having a small particle size less than 0.2μm in diameter, and its method of manufacture is disclosed in WO 94/21292. Aqueous formulations comprising monophosphoryl lipid A and a surfactant have been described in WO9843670A2.

The bacterial lipopolysaccharide derived adjuvants to be formulated in the compositions of the present invention may be purified and processed from bacterial sources, or alternatively they may be synthetic. For example, purified monophosphoryl lipid A is described in Ribi et al 1986 (supra), and 3-O-Deacylated monophosphoryl or diphosphoryl lipid A derived from Salmonella sp. is described in GB 2220211 and US 4912094. Other purified and synthetic lipopolysaccha des have been described (Hilgers et al., 1986, Int.Arch.Allergy.lmmunol., 79(4):392-6; Hilgers er a/., 1987, Immunology, 60(1 ):141-6; and EP 0 549 074 B1). A particularly suitable bacterial lipopolysaccharide adjuvant is 3D- MPL.

Accordingly, the LPS derivatives that may be used in the present invention are those immunostimulants that are similar in structure to that of LPS or MPL or 3D-MPL. In another aspect of the present invention the LPS derivatives may be an acylated monosaccharide, which is a sub-portion to the above structure of MPL.

Synthetic derivatives of lipid A are also known including, but not limited to:

OM174 (2-deoxy-6-o-[2-deoxy-2-[(R)-3-dodecanoyloxytetra-decanoylamino]-4-o- phosphono-β-D-glucopyranosyl]-2-[(R)-3-hydroxytetradecanoylamino]-α-D- glucopyranosyldihydrogenphosphate), (WO 95/14026)

OM 294 DP (3S, 9 R) -3--[(R)-dodecanoyloxytetradecanoylamino]-4-oxo-5-aza-9(R)-[(R)- 3-hydroxytetradecanoylamino]decan-1 ,10-diol,1 ,10-bis(dihydrogenophosphate) (WO99 /64301 and WO 00/0462 )

OM 197 MP-Ac DP ( 3S-, 9R) -3-[(R) -dodecanoyloxytetradecanoylamino]-4-oxo-5-aza-9- [(R)-3-hydroxytetradecanoylamino]decan-1 ,10-diol,1 -dihydrogenophosphate 10-(6- aminohexanoate) (WO 01/46127)

Another suitable immunostimulant for use in the present invention is Quil A saponin and its derivatives. Saponins are taught in: Lacaille-Dubois, M and Wagner H. (1996. A review of the biological and pharmacological activities of saponins. Phytomedicine vol 2 pp 363- 386). Saponins are steroid or triterpene glycosides widely distributed in the plant and marine animal kingdoms. Saponins are noted for forming colloidal solutions in water which foam on shaking, and for precipitating cholesterol. When saponins are near cell membranes they create pore-like structures in the membrane which cause the membrane to burst. Haemolysis of erythrocytes is an example of this phenomenon, which is a property of certain, but not all, saponins.

Saponins are known as adjuvants in vaccines for systemic administration. The adjuvant and haemolytic activity of individual saponins has been extensively studied in the art (Lacaille-Dubois and Wagner, supra). For example, Quil A (derived from the bark of the

South American tree Quillaja Saponaha Molina), and fractions thereof, are described in US 5,057,540 and "Saponins as vaccine adjuvants", Kensil, C. R., Crit Rev Ther Drug Carrier Syst, 1996, 12 (1-2):1-55; and EP 0 362 279 B1. Particulate structures, termed Immune Stimulating Complexes (ISCOMS), comprising fractions of Quil A are haemolytic and have been used in the manufacture of vaccines (Morein, B., EP 0 109 942 B1 ; WO 96/11711 ; WO 96/33739). The haemolytic saponins QS21 and QS17 (HPLC purified fractions of Quil A) have been described as potent systemic adjuvants, and the method of their production is disclosed in US Patent No.5,057,540 and EP 0 362 279 B1. QS-21 is a natural saponin derived from the bark of Quillaja saponaria Molina, which induces CD8+ cytotoxic T cells (CTLs), Th1 cells and a predominant lgG2a antibody response and is a particularly suitable saponin in the context of the present invention. Other saponins which have been used in systemic vaccination studies include those derived from other plant species such as Gypsophila and Saponaria (Bomford et al., Vaccine, 10(9):572-577, 1992).

An enhanced system involves the combination of a non-toxic lipid A derivative and a saponin derivative particularly the combination of QS21 and 3D-MPL as disclosed in WO 94/00153, or a less reactogenic composition where the QS21 is quenched with cholesterol as disclosed in WO 96/33739. The saponins forming part of the present invention may be separate in the form of micelles, or may be in the form of large ordered structures such as ISCOMs (EP 0 109 942 B1 ) or liposomes) when formulated with cholesterol and lipid, or in the form of an oil in water emulsion (WO 95/17210). The saponins may suitably be associated with a metallic salt, such as aluminium hydroxide or aluminium phosphate (WO 98/15287).

A particularly potent adjuvant formulation involving QS21 and 3D-MPL in an oil in water emulsion is described in WO 95/17210 and in WO 99/11241 and WO 99/12565, and are particularly suitable formulations.

Accordingly in one embodiment of the present invention there is provided a vaccine comprising an influenza antigen preparation of the present invention adjuvanted with detoxified lipid A or a non-toxic derivative of lipid A, such as a monophosphoryl lipid A or derivative thereof which are particularly suitable.

Suitably the vaccine additionally comprises a saponin, in particular QS21. Suitably the formulation additionally comprises an oil in water emulsion such as that described in WO 95/1720, WO 90/14837 and WO 93/14744. The present invention also provides a method for producing a vaccine formulation comprising mixing an antigen preparation of the present invention together with a pharmaceutically acceptable excipient, such as 3D-MPL.

The vaccines according to the invention may further comprise at least one surfactant which may be in particular a non-ionic surfactant. Suitable non-ionic surfactant are selected from the group consisting of the octyl- or nonylphenoxy polyoxyethanols (for example the commercially available Triton ™ series), polyoxyethylene sorbitan esters (Tween ™ series) and polyoxyethylene ethers or esters of general formula (I):

(I) HO(CH₂CH₂O)_n-A-R wherein n is 1-50, A is a bond or -C(O)-, R is C_1-50 alkyl or phenyl C_1-50 alkyl; and combinations of two or more of these.

Suitable surfactants falling within formula (I) are molecules in which n is 4-24, more particularly 6-12, and typically 9; the R component is C_1-50 , in particular C₄-C₂₀ alkyl and C₁₂ alkyl is particularly suitable.

Octylphenoxy polyoxyethanols and polyoxyethylene sorbitan esters are described in "Surfactant systems" Eds: Attwood and Florence (1983, Chapman and Hall). Octylphenoxy polyoxyethanols (the octoxynols), including t- octylphenoxypolyethoxyethanol (Triton X-100 ™) are also described in Merck Index Entry 6858 (Page 1162, 12^th Edition, Merck & Co. Inc., Whitehouse Station, N.J., USA; ISBN 0911910-12-3). The polyoxyethylene sorbitan esters, including polyoxyethylene sorbitan monooleate (Tween 80 ™) are described in Merck Index Entry 7742 (Page 1308, 12^th Edition, Merck & Co. Inc., Whitehouse Station, N.J., USA; ISBN 0911910-12-3). Both may be manufactured using methods described therein, or purchased from commercial sources such as Sigma Inc.

Particularly suitable non-ionic surfactants include Triton X-45, t-octylphenoxy polyethoxyethanol (Triton X-100), Triton X-102, Triton X-114, Triton X-165, Triton X-205, Triton X-305, Triton N-57, Triton N-101 , Triton N-128, Breij 35, polyoxyethylene-9-lauryl ether (laureth 9) and polyoxyethylene-9-stearyl ether (steareth 9). Triton X-100 and laureth 9 are particularly suitable. Also particularly suitable is the polyoxyethylene sorbitan ester, polyoxyethylene sorbitan monooleate (Tween 80™).

Further suitable polyoxyethylene ethers of general formula (I) are selected from the following group: polyoxyethylene-8-stearyl ether, polyoxyethylene-4-lauryl ether, polyoxyethylene-35-lauryl ether, and polyoxyethylene-23-lauryl ether.

Alternative terms or names for polyoxyethylene lauryl ether are disclosed in the CAS registry. The CAS registry number of polyoxyethylene-9 lauryl ether is: 9002-92-0. Polyoxyethylene ethers such as polyoxyethylene lauryl ether are described in the Merck index (12^th ed: entry 7717, Merck & Co. Inc., Whitehouse Station, N.J., USA; ISBN 0911910-12-3). Laureth 9 is formed by reacting ethylene oxide with dodecyl alcohol, and has an average of nine ethylene oxide units.

Two or more non-ionic surfactants from the different groups of surfactants described may be present in the vaccine formulation described herein. In particular, a combination of a polyoxyethylene sorbitan ester such as polyoxyethylene sorbitan monooleate (Tween 80 ™) and an octoxynol such as t-octylphenoxypolyethoxyethanol (Triton) X-100™ is particularly suitable. Another particularly suitable combination of non-ionic surfactants comprises laureth 9 plus a polyoxyethylene sorbitan ester or an octoxynol or both.

Non-ionic surfactants such as those discussed above have suitable concentrations in the final vaccine composition as follows: polyoxyethylene sorbitan esters such as Tween 80™: 0.01 to 1 %, in particular about 0.1% (w/v); octyl- or nonylphenoxy polyoxyethanols such as Triton X-100™ or other detergents in the Triton series: 0.001 to 0.1%, in particular 0.005 to 0.02 % (w/v); polyoxyethylene ethers of general formula (I) such as laureth 9: 0.1 to 20 %, preferably 0.1 to 10 % and in particular 0.1 to 1 % or about 0.5% (w/v).

For certain vaccine formulations, other vaccine components may be included in the formulation. As such the formulations of the present invention may also comprise a bile acid or a derivative thereof, in particular in the form of a salt. These include derivatives of cholic acid and salts thereof, in particular sodium salts of cholic acid or cholic acid derivatives. Examples of bile acids and derivatives thereof include cholic acid, deoxycholic acid, chenodeoxycholic acid, lithocholic acid, ursodeoxycholic acid, hyodeoxycholic acid and derivatives such as glyco-, tauro-, amidopropyl-1- propanesulfonic-, amidopropyl-2-hydroxy-1-propanesulfonic derivatives of the aforementioned bile acids, or N,N-bis (3Dgluconoamidopropyl) deoxycholamide. A particularly suitable example is sodium deoxycholate (NaDOC) which may be present in the final vaccine dose.

In another embodiment, the claimed method, use and vaccine composition relate to viruses other than influenza virus, that require or can be produced by propagation in embryonated eggs, in particular embryonated chicken eggs. Suitable viruses are selected from the list consisting of: orthomyxoviridae (including influenza virus), paramyxoviridae, flaviviridae, togaviridae, rhabdovi dae and coronaviridae families.

The teaching of all references in the present application, including patent applications and granted patents, are herein fully incorporated by reference.

For the avoidance of doubt the terms 'comprising', 'comprise' and 'comprises' herein are intended by the inventors to be optionally substitutable with the terms 'consisting of, 'consist of, and 'consists of, respectively, in every instance.

The invention will be further described by reference to the following, non-limiting, examples:

Example I describes the Improved yield of influenza virus and HA in embryonated chicken eggs in three distinct experiments using a solution of amino acids, or trypsin or a combination of both.

Example II describes the large scale preparation of a split influenza vaccine. Example III describes the Preparation of influenza virus antigen preparation for a thiomersal-free vaccine.

Example IV describes the SRD Method used to measure haemagglutinin content.

Example V describes and illustrates an experiment and a methodology for selecting suitable amino acid solutions.

Example I - Improved yield of influenza virus and HA in embryonated chicken eggs

Experiment A

1-1 • Inoculation of embryonated eggs 10-days old embryonated chicken eggs were inoculated with influenza strain

A Λ/yoming/3/2003 (H3N2) X-147 at a dilution of 10^'5 in PBS (pH 7.2), containing 0.025 mg/ml hydrocortisone. Additionally the inoculation solution contained (a) a solution of trypsin (720000 USP/L, porcine origin, BioWhitaker, catalog number BESP117E), (b) a solution of amino acids 1 a, (c) a solution of amino acids 1 b, (d) a solution of amino acids 2a, (e) a solution of amino acids 2b, (f) no additive. The eggs are inoculated with 0.5 ml of the inoculum solution by hand with a needle and a syringe, but an automatic egg inoculation apparatus is equally suitable. The composition of the amino acid solution is given in Table 2.

72 eggs were used for each experiment.

Solution of trypsin

0.5 ml solution containing 9.3 μl of concentrated trypsin solution (720000 USP/L, irradiated porcine origin, BioWhitaker)

NaCI 164 mM^* KCI 3.76 mM^* Na2HPO4 8.17 mM^* KH2PO4 1.91 mM^*

Trypsin at 720000 USP/L

^* The ionic concentration may slightly differ from one trypsin lot to another depending on the specific activity of trypsin, and will be adjusted routinely.

Amino acid solutions: are adapted from Ohta et al., 2001 , Poultry science 80: 1430-1436 (in particular table 1 on page 1431 ), with the following modifications Amino acid solution 1 has a pH of 9.7 and osmolahty of 1550; it does not contain valine. Amino acid solution 2 has a pH of 7.2 and osmolahty of 1033; it does not contain valine nor tyrosine. Solutions 1 b and 2b are a fifth of solutions 1 and 2, respectively.

The eggs were inoculated with each solution (0.5 ml) containing Influenza virus and the component according to the invention and prepared as described hereabove, and incubated at 33°C for 48 hours. Before harvest, the embryos were killed by chilling the eggs at 4°C for 24 hours. Typically the cooling temperature may range from 2 to 8°C and the storage time at this temperature may range from 12 to 60 hours. 1.2. Harvest

The allantoic fluid was harvested manually (it may be suitable to use an appropriate egg harvesting machine) and clarified by low-speed centrifugation. Usually, 8 to 10 ml of crude allantoic fluid can be collected per egg.

1.3. Concentration and purification of whole virus from allantoic fluid

Virus was pelleted by ultracent fugation at 17,500 rpm for 2 hours in a Beckman SW19 rotor through a cushion of 20% sucrose. The virus pellet was resuspended in 15 ml of PBS pH 7.4 (PELLET 1) and loaded on a 15 to 55% sucrose gradient. After centrifugation the fraction containing protein was collected and diluted to 35 ml with PBS pH 7.4. After centrifugation for 60 min at 25'200 in a Beckman SW28 rotor the virus pellet was resuspended in 3 ml of PBS, pH 7.4 (PELLET 2).

The virus yield was evaluated on the clarified allantoic fluid and at two steps during the purification process (PELLET 1 and PELLET 2). Hemagglutination titer (HA titer) and protein content of purified virus are indicators of the virus yield at each process step.

It was found that the addition of trypsin or amino acids had a positive effect on the virus yield as measured by the HA titer, and this increase was observed at each process step as compared to the control.

1.4. Assays

Haemaqqlutination assay

A 2-fold dilution series is prepared in reaction tubes by diluting 250 μl virus suspension with 250 μl PBS. 250 μl chicken red blood cells (RBC) (0.5% in PBS) are then added. The reaction is mixed and the RBCs are allowed to sediment for 1 hour at room temperature. The HA titre corresponds to the inverse of the last dilution that shows complete hemagglutination.

Protein assay Protein was measured by photometric determination of the total protein content by BIURET-BCA (Bicinchoninic acid) assay. Sample preparation was done using the Compat-Able(TM) Protein Assay Preparation Reagent Set (0023215, Pierce Chemical Company). Briefly, 50 μl of each sample was mixed with 500 μl of reagent 1 and incubated at room temperature for 5 to 10 minutes. 500 μl reagent 2 were added to each sample and mixed well. After centrifugation at 12,000 rpm for 5 min the supernatant was removed and the pellet was resuspended in 50 μl of BCA working solution. The working solution consisted of 20 μl BCA protein assay reagent B (Perbio Nr. 23224) mixed with 1 ml of BCA protein assay reagent A (Perbio Nr. 23222).

25 μl of the sample, prepared as described above, was dispensed into a microtiter plate (flat-bottom). 200 μl of the BCA working solution was added per well and mixed well with the sample. The plate was incubated at 30 min at 37°C, followed by 5 min +/- 0.5 min at room temperature. The absorption was measured at 560 nm. The protein content was determined using a BSA calibration curve.

Results are shown in Tables 3 and 4, and in Figures 1 and 2.

Experiment B

The following groups were tested. Exp.1 : trypsin solution

Exp.2: amino acid solution adapted from Ohta et al., 2001 , Poultry science 80: 1430-1436 (in particular table 1 on page 1431 ), with the following modifications: pH of 9.1 ; it does not contain valine.

Exp.3: Amino acid solution as above in admixture with trypsin.

Exp.4 and 5: solution of trypsin either three times concentrated (exp.4), or used at a third of solution of Exp.1 , respectively. Exp.6: control group

The eggs were inoculated with 0.5 ml of a solution containing influenza A/Wyoming/3/ 2003 (H3N2) IVR-147 at a dilution of 10^"5 in PBS, 0.025 mg/jml hydrocortisone and the component according to the invention and prepared as described hereabove, and incubated at 33°C for 72 hours. Before harvest, the embryos were killed by chilling the eggs at 4°C for 24 hours. Typically the cooling temperature may range from 2 to 8°C and the storage time at this temperature may range from 12 to 60 hours.

Results are shown in Tables 5 and 6, and in Figures 3 and 4.

Experiment C - Effect of pH on the yield of influenza virus and HA in embryonated chicken eggs

The following groups were tested.

1. Standard inoculum (strain A/New Caledonia/20/99 (H1 N1 ) IVR-116) (final volume 0.5ml)

2. Standard inoculum mixed with an equivalent volume of a solution at pH 8

3. Standard inoculum mixed with an equivalent volume of a solution at pH 9.5 4. Standard inoculum mixed with an equivalent volume of a solution at pH 11

5. Standard inoculum mixed with an equivalent volume of a solution at pH 11.5

The eggs were inoculated with each solution (0.5 ml) containing Influenza virus at a dilution 10^"5 and the component according to the invention and prepared as described hereabove, and incubated at 33°C for 72 hours. Before harvest, the embryos were killed by chilling the eggs at 4°C for 24 hours. Typically the cooling temperature may range from 2 to 8°C and the storage time at this temperature may range from 12 to 60 hours.

Results are shown in Figure 5A and B.

Example II - Large scale preparation of a split influenza vaccine

Monovalent split vaccine is prepared, at a large scale, according to the following procedure.

11.1. Preparation of virus inoculum

On the day of inoculation of embryonated eggs a fresh inoculum is prepared by mixing the working seed lot with a phosphate buffered saline containing gentamycin sulphate at 0.5 mg/ml and hydrocortisone at 25 μg/ml (virus strain-dependent). The component according to the invention is added to the solution. 11.2. Inoculation of embryonated eggs

Nine to eleven day-old embryonated eggs are used for virus replication. Shells are decontaminated. The eggs are inoculated with 0.2 ml of the virus inoculum. The inoculated eggs are incubated at the appropriate temperature (virus strain-dependent) for 48 to 96 hours. At the end of the incubation period, the embryos are killed by cooling and the eggs are stored for 12-60 hours at 2-8°C.

11.3. Harvest

The allantoic fluid from the chilled embryonated eggs is harvested. Usually, 8 to 10 ml of crude allantoic fluid is collected per egg. To the crude monovalent virus bulk 0.100 mg/ml thiomersal is optionally added.

11.4. Concentration and purification of whole virus from allantoic fluid

11.4.1. Clarification

The harvested allantoic fluid is clarified by moderate speed centrifugation (range: 4000 - 14000 g).

11.4.2. Adsorption step To obtain a CaHPO₄ gel in the clarified virus pool, 0.5 mol/L Na₂HPO₄ and 0.5mol/L CaCI₂ solutions are added to reach a final concentration of CaHPO₄ of 1.5 g to 3.5 g CaHPO₄/litre depending on the virus strain.

After sedimentation for at last 8 hours, the supernatant is removed and the sediment containing the influenza virus is resolubilised by addition of a 0.26 mol/L EDTA-Na₂ solution, dependent on the amount of CaHPO₄ used.

11.4.3. Filtration

The resuspended sediment is filtered on a 6μm filter membrane.

11.4.4. Sucrose gradient centrifugation

The influenza virus is concentrated by isopycnic centrifugation in a linear sucrose gradient (0 - 55 % (w/v)) containing 100 μg/ml Thiomersal. The flow rate is 8 - 15 litres/hour.

At the end of the centrifugation, the content of the rotor is recovered by four different fractions (the sucrose is measured in a refractometer): fraction 1 55-52% sucrose fraction 2 approximately 52-38% sucrose fraction 3 38-20% sucrose^* fraction 4 20- 0% sucrose ^* virus strain-dependent: fraction 3 can be reduced to 15% sucrose.

For further vaccine preparation, only fractions 2 and 3 are used. Fraction 3 is washed by diafiltration with phosphate buffer in order to reduce the sucrose content to approximately below 6%. The influenza virus present in this diluted fraction is pelleted to remove soluble contaminants.

The pellet is resuspended and thoroughly mixed to obtain a homogeneous suspension. Fraction 2 and the resuspended pellet of fraction 3 are pooled and phosphate buffer is added to obtain a volume of approximately 40 litres. This product is the monovalent whole virus concentrate.

11.4.5. Sucrose gradient centrifugation with sodium deoxycholate

The monovalent whole influenza virus concentrate is applied to a ENI-Mark II ultracenthfuge. The K3 rotor contains a linear sucrose gradient (0.2 - 55 % (w/v)) where a sodium deoxycholate gradient is additionally overlayed. Tween 80 is present during splitting up to 0.1 % (w/v). The maximal sodium deoxycholate concentration is 0.7-1.5 % (w/v) and is strain dependent. The flow rate is 8 - 15 litres/hour.

At the end of the centrifugation, the content of the rotor is recovered by three different fractions (the sucrose is measured in a refractometer) Fraction 2 is used for further processing. Sucrose content for fraction limits (47-18%) varies according to strains and is fixed after evaluation:

11.4.6. Sterile filtration The split virus fraction is filtered on filter membranes ending with a 0.2 μm membrane. Phosphate buffer containing 0.025 % (w/v) Tween 80 is used for dilution. The final volume of the filtered fraction 2 is 5 times the original fraction volume.

11.4.7. Inactivation The filtered monovalent material is incubated at 22 ± 2°C for at most 84 hours (dependent on the virus strains, this incubation can be shortened). Phosphate buffer containing 0.025% Tween 80 is then added in order to reduce the total protein content down to max. 250 μg/ml. Formaldehyde is added to a final concentration of 50 μg/ml and the inactivation takes place at 20°C ± 2°C for at least 72 hours.

11.4.8. Ultrafiltration

The inactivated split virus material is concentrated at least 2 fold in a ultrafiltration unit, equipped with cellulose acetate membranes with 20 kDa MWCO. The material is subsequently washed with phosphate buffer containing 0.025 % (w/v) Tween 80 and following with phosphate buffered saline containing 0.01 % (w/v) Tween.

11.4.9. Final sterile filtration

The material after ultrafiltration is filtered on filter membranes ending with a 0.2 μm membrane. The final concentration of haemagglutinin, measured by SRD (method recommended by WHO) should exceed 450 μg/ml.

11.4.10. Storage

The monovalent final bulk is stored at 2 - 8°C for a maximum of 18 months. To obtain a trivalent vaccine, monovalent bulk from different strains are pooled.

Example III - Preparation of influenza virus antigen preparation for a thiomersal- free vaccine

Monovalent split vaccine is prepared at a large scale according to the following procedure.

111.1. Preparation of virus inoculum

On the day of inoculation of embryonated eggs a fresh inoculum is prepared by mixing the working seed lot with a phosphate buffered saline containing gentamycin sulphate at 0.5 mg/ml and hydrocortisone at 25 μg/ml. (virus strain-dependent).

III.2. Inoculation of embryonated eggs

Nine to eleven day old embryonated eggs are used for virus replication. Shells are decontaminated. The eggs are inoculated with 0.2 ml of the virus inoculum. 60,000 inoculated eggs are incubated at the appropriate temperature (virus strain-dependent) for 48 to 96 hours. At the end of the incubation period, the embryos are killed by cooling and the eggs are stored for 12-60 hours at 2-8°C. 111.3. Harvest

The allantoic fluid from the chilled embryonated eggs is harvested. Usually, 8 to 10 ml of crude allantoic fluid is collected per egg.

111.4. Concentration and purification of whole virus from allantoic fluid

III. 4.1. Clarification

III. 4.2. Precipitation step

Saturated ammonium sulfate solution is added to the clarified virus pool to reach a final ammonium sulfate concentration of 0.5 mol/L. After sedimentation for at least 1 hour, the precipitate is removed by filtration on depth filters (0.5 μm nominal). Alternatively the precipitate may be removed by low speed centrifugation followed by a filtration step.

III. 4.3. Filtration

The clarified crude whole virus bulk is filtered on filter membranes ending with a validated sterile membrane (typically 0.2 μm).

III. 4.4. Ultrafiltration

The sterile filtered crude monovalent whole virus bulk is concentrated on a cassettes equipped with 1000 kDa MWCO BIOMAX™ membrane at least 6 fold. The concentrated retentate is washed with phosphate buffered saline at least 1.8 times.

III. 4.5. Sucrose gradient centrifugation

The influenza virus is concentrated by isopycnic centrifugation in a linear sucrose gradient (0-55 % (w/v). The flow rate is 8 - 15 litres/hour.

At the end of the centrifugation, the content of the rotor is recovered by four different fractions (the sucrose is measured in a refractometer): fraction 1 55-52% sucrose fraction 2 approximately 52-38% sucrose - fraction 3 38-20% sucrose* fraction 4 20- 0% sucrose ^* virus strain-dependent: fraction 3 can be reduced to 15% sucrose.

For further vaccine preparation, either only fractions 2 is used or fraction 2 together with a further purified fraction 3 are used.

Fraction 3 is washed by diafiltration with phosphate buffer in order to reduce the sucrose content to approximately below 6%. The influenza virus present in this diluted fraction is pelleted to remove soluble contaminants.

III. 4.6. Sucrose gradient centrifugation with sodium deoxycholate

The monovalent whole influenza virus concentrate is applied to a ENI-Mark II ultracentrifuge. The K3 rotor contains a linear sucrose gradient (0 - 60 % (w/v)) where a sodium deoxycholate gradient is additionally overlayed. Tween 80 is present during splitting up to 0.1 % (w/v) and Tocopherylsuccinate is added for B-strain viruses up to 0.5 mM. The maximal sodium deoxycholate concentration is 0.7-1.5 % (w/v) and is strain dependent. The flow rate is 8 - 15 litres/hour.

At the end of the centrifugation, the content of the rotor is recovered by three different fractions (the sucrose is measured in a refractometer) Fraction 2 is used for further processing. Sucrose content for fraction limits varies according to strains and is fixed after evaluation.

III. 4.7. Sterile filtration

The split virus fraction is filtered on filter membranes ending with a 0.2 μm membrane. Phosphate buffer containing 0.025 % (w/v) Tween 80 and (for B strains) 0.5 mM Tocopherylsuccinate is used for dilution. The final volume of the filtered fraction 2 is 5 times the original fraction volume.

III. 4.8. Inactivation The filtered monovalent material is incubated at 22 ± 2°C for at least 84 hours (dependent on the virus strains, this incubation can be shortened). Phosphate buffer containing 0.025% (w/v) Tween 80 is then added in order to reduce the total protein content down to maximum 500 μg/ml. For B-strains a phosphate buffered saline containing 0.025% (w/v) Tween 80 and 0.25 mM Tocopherylsuccinate is applied for dilution to reduce the total protein content down to 500 μg/ml. Formaldehyde is added to a final concentration of 100 μg/ml and the inactivation takes place at 20°C ± 2°C for at least 72 hours.

III. 4.9. Ultrafiltration

The inactivated split virus material is concentrated at least 2 fold in a ultrafiltration unit, equipped with cellulose acetate membranes with 20 kDa MWCO. The Material is subsequently washed with phosphate buffer containing 0.025 % (w/v) Tween 80 and following with phosphate buffered saline containing 0.01 % (w/v) Tween. For B-strain viruses a phosphate buffered saline containing 0.01% (w/v) Tween 80 and 0.1 mM Tocopherylsuccinate is used for washing.

III. 4.10. Final sterile filtration

The material after ultrafiltration is filtered on filter membranes ending with a 0.2 μm membrane. Filter membranes are rinsed and the material is diluted if necessary that the protein concentration does not exceed 500 μg/ml with phosphate buffered saline containing 0.01% (w/v) Tween 80 and, specific for B strains, 0.1 mM Tocopherylsuccinate.

III. 4.11. Storage

The monovalent final bulk is stored at 2 - 8°C for a maximum of 18 months.

Example IV - SRD Method used to measure haemagglutinin content

Glass plates (12.4 - 10.0 cm) are coated with an agarose gel containing a concentration of anti-influenza HA serum that is recommended by NIBSC. After the gel has set, 72 sample wells (3 mm 0) are punched into the agarose. 10 microliters of appropriate dilutions of the reference and the sample are loaded in the wells. The plates are incubated for 24 hours at room temperature (20 to 25°C) in a moist chamber. After that, the plates are soaked overnight with NaCI-solution and washed briefly in distilled water. The gel is then pressed and dried. When completely dry, the plates are stained on Coomassie Brillant Blue solution for 10 min and de-stained twice in a mixture of methanol and acetic acid until clearly defined stained zones become visible. After drying the plates, the diameter of the stained zones surrounding antigen wells is measured in two directions at right angles. Alternatively equipment to measure the surface can be used. Dose-response curves of antigen dilutions against the surface are constructed and the results are calculated according to standard slope-ratio assay methods (Finney, D.J. (1952): Statistical Methods in Biological Assay. London: Griffin, Quoted in: Wood, JM, et al (1977). J. Biol. Standard. 5, 237-247).

Example V - Selection of suitable amino acid combinations

V.1. Introduction

V.1.1. Outline of the methodology

The different amino acid combination runs carried out experimentally can be analysed by standard statistical or design of experiment software packages (such as the 'design expert 6' software that is used here) to correlate the response obtained to the various components in the combination. Such a software allows a model to be generated, linking the response obtained to the combination of components in each runs, and can then be used to predict how combinations that have not been experimentally tested might perform in respect of virus yield or antigen production, for example.

A brief description of the working procedure is herein below given. For detailed information regarding the definitions, it can be referred to the section V.1. and below.

The objective of the experiment detailed below was to study the effects of 18 amino acids at two levels (presence or absence of the amino acid in the combination) on the 'response' (defined as the total amount of viral protein that would be produced by 100 injected eggs as assessed by the product of i) the percentage of harvested eggs by ii) the protein content per harvested egg - see below). The entire possible solutions that one can prepare from 18 amino acids at two levels are 2^Λ18 solutions, that is to say 262144 solutions (from the solution without any amino acids to the solution containing all the amino acids).

Since it was not practically possible to actually test all these solutions, the following strategy, summarized in four steps, was undertaken: 1. Only a part of the entire possible solutions were performed and experimentally tested/injected in the eggs. These solutions have been, for the most part (40 different solutions, i.e. solutions used for runs 1 to 45, with the exceptionof solutions used for runs 8, 44 and 45 which are repeat of solutions used otherwise, and with the additional exception of solutions used for run 15 and 19), defined using the experimental design methodology (COCHRAN W. G., COX G. M. [1957]. Experimental Designs. New York, Wiley, 61 1 p.; LEWIS G. A., MATHIEU D., PHAN-TAN-LUU R. [1999] Pharmaceutical Experimental Design. New York, Marcel Dekker, 491 p). The other tested solutions have been selected based on the following criteria: solutions comprising individual amino acids, or solutions which had performed well in an earlier experiment.

2. These solutions then allowed fitting a statistical model, using well-known statistical or design of experiment software packages such as the 'design expert 6' software to predict the results which would have been obtained with anyone of the 2^Λ18 theoretically possible solutions, in other words, including also results which would have been obtained with solutions that have not been actually tested in the eggs.

The model that was selected is based on the experimentally tested solutions and represents a statistical equation (a mathematical equation to which experimental errors are added) that was defined to express the better the variations of the response by the caused variations of the amino acids. The model is selected so as to be appropriate to closely predict the observed responses of the actually tested solutions, and be used to predict also the response expected with anyone of the solutions that have not actually been tested, or alternatively to rank them by order of efficacy so as to be able to define categories in which several predicted solutions would be expected to lead to a similar range of response (for example a ratio treated: control of a given value, e.g. between 1.2 and 1.5).

This model was fitted using the Design-Expert 6 software, but this could be done in many standard statistical or design of experiment software packages, since Design-Expert 6 uses the classical multiple linear regression, coding the absence of an amino acid by a "- 1" and the presence by a "+1".

3. Then, using the equation of the model, it was possible to determine which solutions would give the highest responses. This calculation was done in SAS software, but could be done in any statistical software having an included programming language. 4. A confirmatory step is then ideally included to experimentally test the solutions selected to be the best solutions, since a model is always a simplified representation of the reality.

V.1.2. Definitions

As said above, the following example describes an experiment done in order to identify which amino acid, or combination of at least two amino acids, are suitable to lead to improved influenza virus yields, and/or in particular to the improved yield of HA, compared to a standard solution containing no amino acid.

To determine the suitable amino acids combinations, it is often desirable to take into consideration the possible interactions between the amino acids (synergies or inhibitions). Consequently the search for the best amino acids combinations will be done by testing amino acids combinations (containing from 1 to 18 amino acids), and not only amino acid individually tested.

The objective of the study was therefore to improve the total amount of viral proteins that would be produced by 100 injected eggs, and consequently the influenza virus yield, using either amino acids taken individually or amino acid combinations.

The outcome of the experiment was the total amount of viral protein that would be produced by 100 injected eggs as assessed by the product of i) the percentage of harvested eggs by ii) the protein content per harvested egg. This product was called 'response'. An increase in the total amount of protein produced may be due to a higher percentage of harvested eggs, or to a higher protein content per harvested egg or a combination of both.

The factors that were analysed were 18 different amino acids selected from the group consisting of: Phenylalanine, Lysine, Cystein, Asparagine, Glutamic acid, Histidine, Isoleucine, Leucine, Proline, Serine, Valine, Tryptophan, Tyrosine, Alanine, Methionine, Arginine, Threonine, Glycine. They were studied at two levels (i.e. the absence or the presence of the said amino acid in the combination). The concentration for each amino acid was adapted from Ohta et al. (2001 , Poultry science 80: 1430-1436) and used every time the amino acid is present in the combination (but for one combination).

Figure 6 presents a flowchart for the selection of suitable amino acid combinations (choice of the amino acids, optimization of their concentrations, implementation to other influenza virus strains).

V.2. Composition of the amino acid solutions and standard solution

The experiment illustrated in the Example section was performed in influenza virus A/New Caledonia/20/99 (H1N1) IVR-116 strain.

Table 7 describes the following experiment: the composition of the tested solutions and the day they were tested. Table 7 also indicates the letters that were used to code, for convenience, the amino acids names during the analysis.

The tested solutions consisted in:

- 40 different solutions (runs 1 to 45, but run 15 and 19), containing from 0 to 18 amino acids. Run 8 is a repeat of run 3, and runs 44 and 45 are repeats of run 24; 10 of the 18 amino acids (Cystein, Asparagine, Leucine, Proline, Serine, Valine, Tyrosine, Methionine, Threonine, Glycine) individually tested, at the same concentrations that were used for the 41 combinations (runs 46 to 55);

- two additional combinations (run 15 and 19). Run 15 consisted in all the amino acids at a concentration different from that used in the other solutions; a control (PBS, 0.025 mg/ml hydrocortisone, containing no exogenous amino acid) every day. As the experiment lasted 11 days, the total number of controls was 11 for this experiment.

All the solutions that did not contain tyrosine had a pH of 9.5. The solutions that contained tyrosine had a higher pH, i.e. a pH of between 9.7 and 10.17. Therefore, should there be an effect due to the pH difference between the tested solutions (e.g. with and without tyrosine), it will be added to the effect of tyrosine. In other words, that means that the estimate of the factor tyrosine effect will be due to the addition of i) the effect of tyrosine and ii) the correlated higher pH. V.3. Results and analysis

V.3.1. Determination of the reference value

The mean of the eleven controls can be used as a reference to which compare the amino acid combinations in order to define whether the total amount of viral protein that would be produced by 100 injected eggs has improved, and to which extent.

Table 8A presents the results of the control for the 11 days that the experiment lasted. The response (product of the percentage of harvested eggs with the protein content per harvested egg) was calculated as described above.

Before performing any statistics, the study of the distribution of the response has been done. The Shapiro-Wilk's normality test applied on the response did not reject the normality hypothesis. Should this test have rejected the normality hypothesis then the distribution of the response would have required a transformation. For example for the control values the same Shapiro-Wilk's normality test would have been applied to the square root, or the log transformation.

During the analysis of the runs 1 to 55 (but the run 15), the Box-Cox transformation method (classical method to search the best transformation to normalize the distribution of data. BOX G. E. P., COX D. R. [1964]. An analysis of transformations (with discussion). J. R. Stat. Soc, Ser. B, 26 (2), 211-252) confirmed no transformation was required. For example, should the Box-Cox have indicated another transformation then that latter transformation would have been used.

All the statistics were then performed directly on the response and are show in Table 8A.

Table 8B shows the mean, the standard deviation and the 95% confidence interval around the mean, as calculated from the response of Table 8A data.

The mean of the eleven control was 12572. This value has been used as a reference when calculating the ratio of the predicted response of each amino acids combinations relative to the mean of the control. A ratio greater than 1 indicated that the predicted response for this combination was higher than the estimated mean of the control. The 95% confidence interval upper limit of this mean was 16263. Therefore, a first step for selecting the most interesting amino acids combinations was to consider the ones with a predicted response higher than 16263.

V.3.2. Analysis of the runs data

The results of the runs are given in Table 9 (numbers are rounded).

Only the combinations of amino acids described in Table 7 were really performed. All other combinations of amino acids have been designed following predictions given by the model described below.

V.3.2.1. Fitting the model to put the response in relation with the amino acids

All 55 runs (but run 15) allowed fitting a statistical model to explain the changes of the response by the changes of the factors. Run 15, too much different from the others (different concentrations), was not included in this analysis.

The runs and the response were introduced in Design Expert 6 software (Design-Expert software for design of experiments, version 6 user's guide [2000]; Stat-Ease).

The software coded the levels of each factor: a "-1" when the amino acid is absent, and a "+1" when it is present.

It replaced also, for a better legibility of the output, the name of each amino acids by a letter, that can be found at the column heads of Table 7.

The software fitted a multiple linear regression to the data.

The 55 runs of the experiment have been shared between different days. Since there was a high probability that the variability between the days was higher than the variability within the day, a factor was introduced in the model to extract this additional variability due to the day from the residual error. This factor is called a block factor, and the days can be called blocks. The model can then be expressed in function of the amino acids, for an "averaged" block ("averaged" day).

The model equation obtained, in terms of coded factors ("absent" = "-1" and "present" =

"+1"), for the "averaged" block, is given below:

Response =

13589.16

-1.59812 ^*B

202.2241 ^*C

280.7663 ^*D

1232.878 ^*E

-95.467 ^*F

434.2554 ^*G

839.1349 ^*H

207.9162 ^*J

316.4423 ^*K

122.8089 ^*L

117.3943 ^*M

246.7557 ^*N

359.4403 ^*O

62.74559 ^*P

96.37456 ^*Q

-314.426 ^*R

759.6433 ^*S

492.925 *B^*F

-1209.72 ^*B*K

-534.994 ^*C*N

493.6713 ^*E^*L

-751.05 ^*J^*P

-836.504 ^*K^*M

1344.339 ^*L*N

-266.912 ^*L*P

2210.047 ^*M^*S

-2094.28 ^*Q^*R Only the terms that contributed to explain the response variations have been kept in the model (the hierarchy of the model was respected: if a factor interacted with another, its main effect was kept in the model, whatever its individual importance). The factor A (Phenylalanin) is not in the model. It can be fixed either at its level "present" or "absent" because it has no impact on the response value.

The analysis of variance table for the model used is shown in Table 10. It shows the importance of using a block factor (the mean square of the block is important).

The remaining seventeen amino acids are all included in the model, alone, and many are in one or more 2-factor interactions. They are present in the model because they have an effect on the response. But their effects are not of the same importance.

The diagnostic of the model has been done using the classical indicators (variance inflation factors, internally and externally residuals, Cook's distances, residual plots NETER J., KUTNER M. H., NACHTSHEIM C. J., WASSERMAN W. [1996]. Applied Linear Statistical Models. WCB/Mac Graw-Hill). None of them raised any concern.

Table 11 gives, for the response, the observed and predicted (by the model) values for the runs of the experiment (but the run 15).

V.3.2.2. Using the model to find the best combinations

When using a model to mimic the reality, it must be realized that however good the model is, it remains a model and is therefore "not absolutely (or strictly) correct", because it is most likely not the true model but an empirical approximation. However, should it be a good enough approximation, the model is very useful.

The model is now used to maximize the response and, having found the highest values of the response, to look at which amino acids combinations they correspond.

However, given the high number of possible workable solutions (2^Λ18 = 262144, including the eighteen individual amino acid solutions and the 'no amino acid' solution), the merit of this model was to avoid testing all combinations before determining which ones were predicted to be good solutions, and among these which ones were predicted to be the best. Since factor A (Phenylalanin) had no effect on the response, the search of the best combinations has been performed on the half possible solutions (the 131 072 solutions that did not contain the Phenylalanin). The remaining 131 072 solutions containing the Phenylalanin would have given the same predicted response.

Since the optimization algorithm of Design-Expert 6 was limited, the SAS software was used to predict the response of the 131 072 solutions, using the model equation.

Table 12 gives the list of the 249 best combinations (response higher than 25 000, combinations without Phenylalanin) predicted by the model, and the ratio of the predicted response relative to the mean of the twelve controls (12 572).

Table 13 and Figure 7 present the best predicted combination (i.e. which give a higher than 16263, 95% confidence interval upper limit of control mean) for a given number of amino acids. Compositions with 7, 8, 9, 10, 11 , 12, 13, 14 and 15 amino acids in Figure 7 correspond to compositions 227, 69, 21 , 10, 12, 2, 1 , 3, 134 in Table 13, respectively.

Composition 1-6 and 16 are below 16263 and have therefore not been exemplified in

Table 13.

Table 14 shows, for these same combinations, the 95% confidence and prediction intervals around the predicted response.

There is an uncertainty around each prediction of the model: - One that is not measurable because linked to the distance between the model used and the true model. This distance may be important depending on the number of solutions which have actually been tested (in this experiment, 41 of the 262144 possible solutions have actually been tested). This immeasurable uncertainty can be minimized by increasing the number of solutions which are actually tested before applying the model.

- One that is due to the residual error (used to calculate the 95% confidence and prediction intervals of Table 14).

- One that is due to the variance between the days. It should be taken into account if a confirmation step is performed, because it will be done at other days and then, the results could fall outside the intervals of the Table 14. V.4. Inoculation of embryonated eggs, harvest and concentration and purification of whole virus from allantoic fluid

10 to 1 1 days old embryonated chicken eggs were inoculated with influenza strain A/New Caledonia/20/99 (H1 N1 ) IVR-116. An inoculation solution was prepared by diluting the virus seed 10^"5 in PBS (pH 7.2), containing 0.025 mg/ml hydrocortisone. Additionally the inoculation solution contained the additives described above.

The eggs were inoculated manually with 0.5 ml of the inoculation solution. 72 eggs were used for each single experiment. After inoculation the eggs were incubated 72 hours at 33°C. Before harvest the embryos were killed by chilling the eggs at 2 to 8°C for 24 hours. The allantoic fluid was harvested manually and clarified by low-speed centrifugation. Usually 8 to 10 ml crude allantoic fluid can be collected per egg.

Virus particles were pelleted by ultracentrifugation at 17,500 rpm for 2 hours in a Beckman SW19 rotor through a 20% sucrose cushion. The virus pellet was resuspended in 15 ml PBS, pH 7.4 and loaded on a 15 to 55% sucrose gradient. After centrifugation the fraction containing the protein peak was collected and diluted to 35 ml with PBS, pH 7.4. Virus particles were pelleted by centrifugation at 25,200 rpm in a Beckman SW28 rotor and the virus pellet was resuspended in 3 ml of PBS, pH 7.4.

The virus yield was evaluated by determining the haemagglutination (HA) titer and protein content of purified virus.

V.5. Use of the procedure and model to design combinations using modified experimental conditions

This model can be further used when experimental conditions are modified such as the starting concentration of the amino acids within the tested solutions, the influenza virus strain, etc. Figure 6 represents a possible flowchart to follow for the optimization of the amino acids concentrations and the study of other strains for example.

V.5.1. Confirmatory step The robustness of the model can further be tested in a confirmatory step, where several amino acid combinations predicted by the model are tested in parallel with a control solution.

The solutions can be selected using Table 13, which lists the best solution for a given number of amino acids. Only the ones having a predicted response greater of 16000 have been included.

Figure 7 plots the predicted response of the Table 13 solutions in function of the number of amino acids contained, as said above.

Three combinations (or more) can be chosen from Figure 7 and Table 13 (based on the model), and two combinations can be added, chosen independently of the model, among the solutions really performed. A control must also be added. Repeat experiments are ideally made for at least four days.

For example, the best combination of seven, ten, and thirteen amino acids can be chosen.

V.5.2. Refining the concentrations

The concentrations of the amino acids that enter in the best combination selected from the confirmatory step can then be optimized using a response surface study (MYERS R. H., MONTGOMERY D. C. [2002]. Response Surface Methodology: Process and Product Optimization Using Designed Experiments. New York, Wiley, 798 p.).

The experimental design normally takes the 'day effect' into account, as a block effect, and will also be built in function of the resources and constraints.

A confirmation step will be done, where the optimum will be tested several days in parallel with the control.

V.5.3. Application to a different Influenza strain

The combination(s) that was (were) selected for the tested strain(s) will be compared with the new strain. Results may be comparable or not. If the results are not satisfactory enough, additional studies can be performed, and can be planned using the experimental design methodology (COCHRAN W. G., COX G. M. [1957]. Experimental Designs. New York, Wiley, 611 p.; LEWIS G. A., MATHIEU D., PHAN-TAN-LUU R. [1999]. Pharmaceutical Experimental Design. New York, Marcel Dekker, 491 p.), taking into account the resources, the constraints, and the day factor as explained above.

List of Tables referred to in this document

Table 1 - Preferred concentration range in the inoculum as adapted from Ohta et al., 2001 , Poultry science 80: 1430-1436). This reference is incorporated by reference.

* or a molecule derivated from specified amino acid at molar equivalence. Table 2 - Amino acid solution for Experiment A

Table 3. HA titer and protein concentration at three steps of the purification process

Table 4. Total content of HA units and protein at three steps of the purification process

Table 5. HA titer and protein concentration at three steps of the purification process

Total volume of clarified allantoic fluid: 400 ml Total volume of pellet 1 : 15 ml Total volume of pellet 1 : 3 ml Table 6. Total content of HA units and protein at three steps of the purification process

Table 8A: results for the controls (numbers are rounded).

Table 8B: mean, SD and 95% confidence interval around the mean of the response for the eleven controls (the numbers are rounded).

Table 9: results of the experiment (runs and control of each day), and ratio of the response of the runs on the response of the control of the block (numbers are rounded)

Table 10: Analysis of variance table [Partial sum of squares]

Table 12: 249 best combinations (response > 25 000) predicted by the model, and ratio of the predicted response on the mean of the twelve controls (12 572)

Table 13: best predicted combination (higher than 16 263, 95% confidence interval upper limit of control mean) for a given number of amino acids.

Table 14: predictions of the response, Cl and PI for each best combination of a given number of amino acids.

Cl low & Cl high = confidence limits of the predicted mean PI low & PI high = prediction limits of individual values

The variability between the blocks is not included in the intervals.

Claims

1. A method for producing influenza virus in embryonated avian eggs, comprising introducing into said eggs, an influenza virus and at least one amino acid or derivative thereof.

2. A method for increasing the production of haemagglutinin in embryonated avian eggs, comprising introducing into said eggs, an influenza virus and at least one amino acid or derivative thereof.

3. A method according to claim 1 or 2, wherein the virus and the other component are inoculated in the allantoid fluid of the embryonated eggs.

4. A method according to any of claim 1 to 3 wherein the virus and the other component are introduced in the form of a solution.

5. A method according to any of claim 1 to 4 wherein the other component comprises at least two amino acids or derivative thereof.

6. A method according to claim 5 wherein the other component comprises at least three amino acids or derivative thereof.

7. A method according to claim 6 wherein the other component comprises at least five amino acids or derivative thereof.

8. A method according to claim 7 wherein the other component comprises all amino acids or derivative thereof as listed in Table 1.

9. A method according to any of claim 4 to 8 wherein the pH of amino acid solution is from 6 to 11.

10. A method according to claim 9 wherein the pH of amino acid solution is from 7 to 10.

11. A method according to claim 10 wherein the pH of amino acid solution is around 9.5.

12. Influenza virus obtained by the method as claimed in any of claims 1 to 11.

13. A method for producing an influenza virus vaccine comprising the steps of:

(a) producing influenza virus according to the method claimed in any of claims 1 to 11 ,

(b) harvesting the virus from the egg, (c) purifying the virus, and

14. A method for producing an influenza virus vaccine as claimed in claim 12 additionally comprising the step of splitting the virus prior to step (d).

15. A method for producing an influenza virus vaccine as claimed in claim 13 additionally comprising the step of purifying the influenza structural protein HA or/and NA prior to step (d).

16. An influenza vaccine obtainable by the method as claimed in any of claims 13 to 15.

17. A method for prophylaxis of influenza infection or disease in a subject which method comprises administering to the subject a vaccine produced according to the method of any of claims 13 to 15.

18. A method according to claim 17, in which vaccine delivery is intradermal, intranasal, intramuscular, oral or subcutaneous.

19. Use of a vaccine according to claim 16 or produced according to the method claimed in any of claims 13 to 15 in the manufacture of a medicament to elicit an immune response to an influenza antigen in a patient susceptible to influenza infection.