WO2006117354A1 - Novel assay for the separation and quantification of hemagglutinin antigens - Google Patents

Abstract

The present invention relates to novel methods for separating hemagglutinin (HA) antigens, comprising the steps of applying a reduced and derivatized antigen preparation comprising solubilized HA antigens and a detergent in a pH controlled solution, on a Reversed-Phase High-Performance Liquid Chromatography (RP-HPLC) column; and eluting the HA antigens from the column with an ion pairing agent in an organic mobile phase. The invention further relates to quantifying methods using the methods for separating the antigens with the further step of measuring the peak area of the eluted antigen in a chromatogram resulting from the elution step.

Description

Title of the invention

Novel assay for the separation and quantification of hemagglutinin antigens.

Field of the invention

The invention relates to the field of vaccine manufacturing. More in particular, the invention relates to the production of influenza vaccines and the determination of antigen concentration in influenza virus preparations.

Background of the invention

Influenza viruses are generally divided into three types: A, B, and C, based on the antigenic differences between their nucleoprotein and matrix protein antigens. Influenza A viruses are further divided into subtypes depending on the antigenic nature of the two major viral surface proteins, the hemagglutinin (HA) and neuraminidase (NA) proteins. Currently, 15 subtypes of HA are known (Lamb and Krug. 2001). Both HA and NA carry antigenic epitopes. Antibodies that are raised against HA and NA are associated with resistance to infection and/or illness in humans and animals. The efficacy of a vaccination against influenza is largely determined by the amount of immunogenic HA in a vaccine (Wright and Webster. 2001) . For several decades the HA content of influenza whole- virus and split vaccines derived from this, has been assayed using Single Radial Immunodiffusion (SRID) . In this assay, influenza virions are disrupted by detergent and submitted to immunodiffusion for three days at room temperature in antibody-loaded agarose gels. Upon gel staining, the precipitation zone diameters of antigen-antibody complexes are measured, and the antigen content of virus preparations of a certain subtype is calculated by using a calibration curve obtained with a whole virus reference batch of this subtype (NIBSC, Hertfordshire, UK) with a known HA content (Wood et al. 1977) . However, this SRID assay has a number of disadvantages. Apart from being time consuming, laborious and not leaving room for very high throughput (Wood et al. 1977), the quantification of HA by SRID was shown to be inaccurate when analyzing split vaccines or subunit vaccines (Johannsen et al . 1985). In addition, the virus sample environmental background (its pH and ionic strength) and the choice of detergent for disintegrating the influenza virus and its HA were shown to affect the determination of the HA titer (Willkommen et al . 1983; Bizhanov et al. 1988). Despite all shortcomings of the SRID assay, and calls from experts in the field that in addition to the SRID assay a physico- chemical quantification method should be used for the quantification of HA (Pereira. 1973; Johannsen et al . 1985), immunodiffusion techniques are still the only methods approved by the regulatory authorities for the evaluation of influenza vaccines.

A Reversed-Phase High Performance Liquid Chromatography (RP-HPLC) method to separate influenza virus components has been described (Phelan and Cohen. 1983). Viral proteins were solubilized and denatured in guanidine-HCl, and reduced by incubation with dithiotreitol (DTT) for several hours at room temperature. It is a well-recognized fact in the art that, under denaturing conditions upon reduction, mature and activated HAO falls apart in the relatively hydrophilic subunit HAl and the hydrophobic subunit HA2, the latter still containing the trans membrane domain of the original HAO. Subsequently, analysis was performed by RP-HPLC at room temperature on an (C8) Aquapore column, applying a linear gradient of 0.05% TFA in water to 0.05% TFA in acetonitrile. However, the separation of the various virus components was far from optimal, whereas the recovery was low and not quantitative, presumably due to aggregation of the virus components and/or nonspecific adsorption to the HPLC system/column. In addition, in this HPLC assay HA2 could not be detected, presumably because it had been trapped on the column matrix due to its strong hydrophobic nature. Kemp et al . (1980) also discloses a method for separating influenza HA using RP-HPLC: radiolabeled tryptic glycopeptides (small parts) of HA are pre-isolated from SDS/PAGE gels and subsequently analyzed by HPLC. The method disclosed by Kemp et al . has the disadvantage of not being suitable for a high-throughput system, because the isolation from gel renders the method rather laborious. Moreover, the chromatographs clearly indicate the poor resolution of the peaks, overlapping with numerous other viral peaks, which makes that the method cannot be used for quantitative purposes. The isolation of numerous bands related to different peptides of different size from gel makes that the method is not suitable for very accurate quantification and repeatability. Moreover, the method of Kemp et al. is not suitable for real-life (non-radiolabeled) samples as the radiolabel is detected, and not suitable for crude sample analyses .

In yet another study (Van der Zee et al . 1983) a method has been disclosed for the purification of Sendai virus envelope proteins using RP-HPLC. Although Van der Zee et al . state that some proteins could be recovered in pure form, this was only assessed by SDS/PAGE, which method is not a very accurate means to show purity of a sample. The chromatograms show that resolution is poor: this indicates that any accurate quantification, based on the HPLC chromatograms is not possible using the purification method disclosed. Moreover, it seems that the detergent interferes with the peak of interest. Furthermore, carry-over of proteins from one analysis to the other is significant. In general, it is clear that the art does not disclose methods and means for an accurate determination of HA concentration in either crude or purified HA samples. Clearly, there is a strong need for a robust, accurate and fast method for reliable separation and quantification of HA in upstream- and downstream-process preparations, as well as for final vaccine formulations.

Description of the figures

Figure 1. Reversed-Phase HPLC of egg-derived, reduced and alkylated influenza A/Panama/2007/99 (Resvir-17; H3N2) 02/100. An amount corresponding to 10.0 μg HA (as determined by SRID) was injected. Numbers 1-4 correspond to the fractions applied on SDS-PAGE of figure 2.

Figure 2. SDS-PAGE silver staining of the four RP-HPLC fractions of Figure 1. A = antigen control. Fraction 1 is the flow through. M = kD size marker. Figure 3. Western blot analysis (anti-HA) of the four RP- HPLC fractions of Figure 1. A = antigen control. Fraction 1 is the flow through. The arrows indicate forms of HA antigen that was not cleaved before application on the column. Figure 4. Reversed-Phase HPLC of PER.C6^®-produced, reduced and alkylated influenza A/Panama/2007/99 (Resvir-17; H3N2) . An amount corresponding to 16.6 μg HA (as determined by

SRID) was injected. Numbers 1-8 correspond to the fractions applied on SDS-PAGE of figure 5. Figure 5. SDS-PAGE silver staining of the eight RP-HPLC fractions of Figure 4. A = antigen control. Fraction 1 is the flow through. M = kD size marker.

Figure 6. Western blot analysis of the eight RP-HPLC fractions of Figure 4. A = antigen control. Fraction 1 is the flow through. M = kD size marker. HAO = the mature antigen. HAl and HA2 = cleaved hemagglutinin antigens. Figure 7. Reversed-Phase HPLC of egg-derived, reduced and alkylated influenza A/Duck/Sing (H5N3) 00/522. An amount corresponding to 3.0 μg HA (as determined by SRID) was injected. The numbers 1, 2a, 2b, 3, 4, 5a, 5b, 5c, 6, 7 and 8 refer to fractions, with significant peaks, some of which were applied on SDS-PAGE as shown in figure 8. Figure 8. SDS-PAGE silver staining of the RP-HPLC fractions 1, 2b, 2a, 3, 4, 5a, 5b and 5c of Figure 7. 0.19 ug HA of the Duck/Sing strain was used as a positive control. Figure 9. Reversed-Phase HPLC of egg-derived reduced and alkylated influenza A/New Caledonia/20/99 (HlNl) 00/608. An amount corresponding to 15.0 μg HA (as determined by SRID) was injected. Numbers 1-7 correspond to the fractions applied on SDS-PAGE of figure 10.

Figure 10. SDS-PAGE silver staining (left panel) and Western blot analysis using an anti-HlNl antibody (right panel) of the seven RP-HPLC fractions of Figure 9. FT = Flow Through. A = antigen, positive control. M = kD size marker. HAO, HAl and HA2 are indicated by arrows.

Figure 11. Linearity study: (A) calibration curve by plotting the measured HAl peak area versus the injected amount of HA from formaldehyde-inactivated, egg-derived, reduced and alkylated Resvir 17 antigen. (B) idem, now for a PER.C6^®-derived BPL-inactivated A/Resvir-17 sample. Figure 12. (A) Reversed Phase-HPLC chromatograms of egg- derived, reduced and alkylated Resvir-17 antigen obtained with column temperatures of 25°C (upper panel) and 70⁰C (lower panel) . (B) Effect of column temperature on the recovery (peak area) of HAl from PER.C6^®-based influenza A/Resvir-17 (dark bars) and A/New Caledonia (light bars) . Figure 13. Reversed Phase-HPLC chromatograms of egg-derived, reduced and alkylated influenza A /Equine/Prague/56 (H7N7) 85/553 antigen. Numbers 1-5 correspond to the fractions applied on SDS-PAGE of figure 14.

Figure 14. SDS-PAGE silver staining of the RP-HPLC fractions 1-5 and loaded antigen of Figure 13. M = size marker. HAO, HAl and HA2 are indicated by arrows. Figure 15. Western blot analysis of BPL-inactivated PER.C6^®- based A/Resvir-17 HA protein upon treatment with trypsin in the presence of 1% Zwittergent (left, A) , or 1% SDS (right, B) in a time range from 15 min to 2 h. HAO, and its subunits HAl and HA2 are indicated by arrows. Figure 16. RP-HPLC of non-trypsinized, reduced/alkylated, BPL-inactivated PER.C6^®-based influenza A/Resvir-17. (A) Immediate injection after reduction/alkylation (approximately 13.6 μg HA). (B) Injection after 17 h storage at 4°C (approximately 5.8 μg HA). Figure 17. HAl peak shape monitoring of non-trypsinized influenza A/Resvir-17 after reduction and alkylation (A) , or after reduction only (B) .

Figure 18. HAl peak shape monitoring of non-trypsinized influenza A/Resvir-17 after reduction by DTT with different concentrations and subsequent storage at 4°C for 0 hours (A) or for 18 hours (B) .

Figure 19. Schematic flow-sheet of a preferred embodiment of the method of the invention indicating the preferred steps of trypsin incubation and the re-addition of the reducing agent after the alkylation step at a concentration of 25 mM, thereby reducing the undesired effects of the alkylating agent. Figure 20. (A) Effect of increasing amount of trypsin (mU) added to influenza preparations of A/New Caledonia, containing small amounts of HA, and (B) increasing times of trypsin treatment (using 120 mU) .

Figure 21. RP-HPLC of: (A) a culture supernatant of PER.C6^® cells grown in BMIV medium and infected with influenza

A/Resvir-17; (B) a trypsin-treated (192 mU) culture supernatant as in (A) ; and (C) a trypsin-treated (192 mU) and centrifuged pellet of culture supernatant as in (A) .

Figure 22. Flow diagram of the preferred steps involved in the preparation of crude culture supernatants containing influenza virus, for quantification of HA by RP-HPLC.

Figure 23. RP-HPLC of PER. C6-produced, reduced and alkylated influenza B/Jiangsu.

Figure 24. SDS-PAGE silver staining of the RP-HPLC fractions of the PER.Cβ-produced, reduced and alkylated influenza

B/Jiangsu sample of Figure 23.

Figure 25. RP-HPLC of egg-based, reduced and alkylated influenza B/Guangdong/120/2000 (01/546, NIBSC) .

Figure 26. SDS-PAGE silver staining of the RP-HPLC fractions of the egg-based, reduced and alkylated influenza

B/Guangdong/120/2000 (01/546, NIBSC) sample of Figure 25.

Summary of the invention

The present invention relates to methods for separating hemagglutinin (HA) antigens, comprising the steps of applying a reduced and derivatized antigen preparation comprising solubilized HA antigens and a detergent in a pH controlled solution, on a Reversed Phase High Performance Liquid Chromatography (RP-HPLC) column; and eluting the HA antigens from the column with an ion pairing agent in an organic mobile phase. One preferred embodiment of the present invention relates to a method according to the invention, wherein said elution is performed at a temperature between about 25°C and about 70⁰C, preferably between about 40⁰C and about 70⁰C, and more preferably between about 50⁰C and about 70⁰C, and most preferably between about 60⁰C and 70⁰C. In another preferred embodiment, the method comprises a step wherein the antigens are cleaved by a protease, such as trypsin.

The invention also relates to methods for quantifying the HA titer of an HA antigen preparation, said method comprising the method of separating hemagglutinin (HA) antigens according to the invention, with the further step of measuring the peak area of the eluted antigen in a chromatogram resulting from the elution step.

The invention in particular relates to the separation and quantification of HA of influenza B viruses.

Detailed description

Here, a novel separation and quantification assay for the determination of hemagglutinin (HA) concentration by Reversed-Phase High-Performance Liquid Chromatography (RP- HPLC) is disclosed. The problem with the RP-HPLC methods that have been described in the art for separation of the antigens from influenza virus was that the separation was not optimal, with poor resolution of the protein peaks of interest, and that the recovery was low and not quantitative. The inventors of the present invention have now solved many of these problems by using a certain RP-HPLC assay in a set-up in which the antigen is reduced in the presence of a detergent, after which an inert derivatization of the antigen preparation is performed thereby protecting the sulfhydryl groups on the antigen. Preferably, for this step, an alkylating agent is added to render a reduced and alkylated antigen preparation and wherein the antigen is present in a pH controlled solution. It was found that by increasing the temperature during elution of the antigens from the RP-HPLC column, the recovery and the reproducibility of the assay was increased. The assays known from the art were performed at room temperature.

The inventors of the present invention have also found that it is preferred to select a column material that is suitable to be used at higher temperatures of up to about 70⁰C. Preferred column material is therefore polymer-based, which generally can be used in these high temperature ranges. It is also preferred to keep the solution in which the antigen is dissolved pH controlled, preferably at neutral pH values. Preferably, values between about 5 and about 9 are used, more preferably values between about 6 and about 8 are used, while it is most preferred to use pH values between about 7 and about 8. Methods for buffering solutions are well known in the art and are herein not further elaborated on. The virus preparation can be brought on the column, eluted from the column and the quantities of the antigens can be calculated from the specific peak areas all in a single day. It is thus a fast and robust method. Moreover, the methods clearly show that the process is accurate (as found in comparison to the SRID assay) and reproducible. The invention relates thus to a fast and accurate means for determining the HA concentration in different kinds of samples within the manufacturing process of influenza vaccines, thereby overcoming most of the problems associated with the methods known in the art.

In the disclosed assay, the quantification of HA is based on the peak area of HAl, which is well separated from the other vaccine components. The applicability of the present invention is demonstrated for different influenza A subtypes, including HlNl, H3N2, H5N3, and H7N7, strongly suggesting that the assay can be broadly applied for different hemagglutinin antigens. The Neuraminidase (NA) component of the strains is not limiting the broad applicability of the invention, as it relates to the separation of the HA component. It is assumed that the invention will also be applicable for influenza B subtypes, as well as for other viruses comprising hemagglutinin antigens that behave in a similar manner on HPLC columns.

The present invention relates to a novel method for separating hemagglutinin (HA) antigens, said method comprising the steps of applying a reduced and derivatized antigen preparation comprising solubilized HA antigens and a detergent in a pH controlled solution, on a Reversed-Phase High-Performance Liquid Chromatography (RP-HPLC) column; and eluting the HA antigens from the column with an ion pairing agent in an organic mobile phase.

In one embodiment of the present invention, Influenza virus particles obtained from an upstream process of either egg-derived material or virus material from cell culture are first solubilized by the addition of a detergent, preferably a zwitterionic agent, more preferably Zwittergent, to a concentration of for example about 1% (w/v) , but this is not critical to the invention. Subsequently, samples are treated with a protease such as trypsin (typically present on beads) to cleave all HA molecules into the subunits HAl and HA2, which are only kept together by a single disulfide bridge upon this treatment. As trypsin is hindered by the presence of high concentrations of SDS, SDS is preferably not used when the trypsin step is added to the method. If SDS is used, the concentration should be low enough not to inhibit the protease. Thus, it is preferred to use a detergent that does not inhibit the activity of the (possible) additional protease.

The disulfide bridge is then broken by addition of a reducing agent, preferably dithiotreitol (DTT) to a concentration of for example 25 mM, although other concentrations may also be used, and reduction takes place for about 10 minutes at about 90⁰C. As in some cases the samples acidify to pH values of 4.0 and become slightly milky and turbid, it is preferred to perform the reduction under buffered (relatively neutral pH) conditions, such as in 150 mM Tris-HCl, pH 8.0. To prevent re-association and/or complex formation of HAl with HA2 and other proteins, it is highly preferred to have the sulfhydryl groups of all proteins protected, for instance by the addition of an alkylating agent, such as iodacetamide (IAA) or iodoacetic acid. Any inert derivatization of the -SH groups may be applied, such that no active groups remain. For example, any suitable sulfhydryl alkylating agent known in the art may be used. Examples of other suitable alkylating agents are N- ethylmaleimide; dithiobis (2-nitro) benzoic acid; nitrogen mustards_? such as chlorambucil and cyclophosphamide; cisplatin; nitrosoureas_? such as carmustine, lomustine and semustine; alkylsulfonates, such as busulfan; ethyleneiniines, such as thiotepa; and triazines_? such as dacarbazlne . The person skilled in the art Is aware of what compounds may be used to have the sulfhydryl groups protected and what compounds may be used to derivatize the antigen preparation, such that no active groups remain.

As disclosed herein, alkylation with IAA is usually performed at 37°C in the dark for about 45 minutes, but other conditions (T, time) will work as well. This step is then preferably followed by the step of adding an alkylation-inhibiting agent, for instance through re- addition of DTT to neutralize all remaining IAA molecules, upon which the samples are ready for RP-HPLC analysis. The HPLC analysis may be performed by using a POROS Rl/10 column (Applied Biosystems) , but other comparable columns would work as well. Usually, an acetonitrile gradient from 20 to 35% is applied at a column temperature that may be as high as 70⁰C, but preferably around 65°C, as the column cannot withstand temperatures that are much higher. In this high temperature range, the HAl peak is generally highest for this column. When the column material allows higher temperatures, the elution temperature may also be increased.

The column of choice is usually selected for its performance at high temperatures. Although silica-based columns such as C4 or C8 can be used, polymer-based columns, such as the POROS Rl/10 column, are preferred as they can still perform well at temperatures as high as 70⁰C. As outlined herein, higher temperatures ensure a better recovery from the column, and thus in a better quantitative method. In the course of the experiments, it was found that prolonged storage of the reduced/alkylated samples at 4°C may result in a deformation of the HAl peak in the RP-HPLC graphs, which would influence the accuracy of peak measurements and thus on the accuracy of the method. Although it does not count for crude samples, present in medium, this would limit the storage/shipment possibilities of the treated (reduced/alkylated) samples and thus on the overall usefulness of the method. This problem of deformation of the peak was solved by adding the reducing agent (exemplified by DTT) also after the alkylation step, thereby decreasing the harmful effects of the alkylating agent. So, the additional step of adding a reducing agent after alkylation, is highly preferred.

It was also found that certain concentrations of DTT increased the peak area of HAl when the samples were analyzed immediately. In general, it is preferred to use concentrations of the reducing agent that are higher than about 4.4 mM, more preferably at least about 11 mM, and most preferably about 22 to about 25 mM.

Generally, when the method of the present invention is carried out as a routine, using the same column for different runs, carry-over from HA from one run to another occurs (see for an example Van der Zee et al . 1983). To reduce the effect of carry-over, a wash step of the column with a detergent, such as 1% SDS or 1% Zwittergent is highly recommended between different runs on the same column, to remove all residual HA from the column material.

Hemagglutinin (HA) antigens are well known in the art. Although the method of the present invention has been demonstrated to work well for hemagglutinin antigens from influenza, it is likely that the method can also be applied for other hemagglutinin antigens derived from other viruses, such as measles virus. Thus, the present invention relates to a method for separating HA, wherein said HA is of an influenza virus or a measles virus. Preferably, said influenza virus is an influenza A virus or an influenza B virus. Also preferred are methods according to the invention wherein said HA is of an influenza A virus strain comprising an Hl, H3, H5, or H7 hemagglutinin. For an accurate calculation of the HA concentration, it is preferred to have the HAO mature form of influenza substantially separated into the subunits HAl and HA2, as HAl is generally the component which can be easily distinguished in chromatograms and of which the peak area can easily be assessed. The reduction of the antigen is preferably pH controlled, i.e., buffered to a suitable pH. Typically, as described herein, a pH of about 8.0 was applied. However, other suitable (relatively neutral) pH values may be used, such as pH 7. It is important to note that the antigen precipitates in solution at pH values that are too low, for instance at a pH value of 4, or even lower. pH values between about 5 and about 9 may typically be applied, while more preferably values between about 6 and about 8 are applied, since at pH 6 the hemagglutinin antigen unfolds during the infection process under natural conditions. It is most preferred to use a pH value between about 7 and about 8. The person skilled in the art will be capable in finding the correct pH value with which the antigens are still acceptably separated, while it is also readily visible when a pH value is too low as the antigen precipitates at such values. As stated intra, it is well within the skill of the skilled person to adjust pH values and to buffer solutions. Typically, as used herein, a solution is buffered with Tris/HCl, but this is not critical to the invention. Reduction of the antigen preparation is preferably performed using a reducing agent such as dithiotreitol (DTT) . To prevent the antigens from re-association or from complex formation, it is highly preferred to have the sulfhydryl groups protected. This is performed through inert derivatization. Derivatization is preferably performed using an alkylating agent. Alkylation is preferably performed by using IAA. Especially when methods are performed in high throughput setting in which the antigen preparations may be left a prolonged period of time before application on the column, it is preferred to keep the antigens dissociated from each other, and thus to have no active groups present on the proteins. Alkylation may also be used for regular settings, or short-term methods. The art discloses the use of an RP-HPLC method to separate influenza virus components (Phelan and Cohen. 1983) . In this, viral proteins were solubilized and denatured in guanidine-HCl, and reduced by incubation with DTT for several hours at room temperature, wherein the pH was not controlled. Subsequently, analysis was performed by RP-HPLC at room temperature on a silica-based (C8) Aquapore column, applying a linear gradient of 0.05% TFA in water to 0.05% TFA in acetonitrile . However, the separation of the various virus components was far from optimal, whereas the recovery was low and not quantitative, presumably due to aggregation of the virus components and/or nonspecific adsorption to the HPLC system/column. Alkylation, or any other inert derivatization of the preparation was not applied, which may have resulted in re-association of components and unwanted complexes may have been formed. In addition, in this HPLC assay HA2 could not be detected, presumably because it had been trapped on the column matrix due to its strong hydrophobic nature.

Although the RP-HPLC method is generally quick, the teaching of the art was that it was not suitable for accurate measurements of antigenic determinants, such as HA of influenza. Guanidine-HCl, as used in the art, is a chaotropic agent that seems too harsh for the antigens and may even destroy some of the antigens of interest. Thus, such chaotropic agents are preferably not used for the methods of the present invention. Suitable detergents that are typically applied for the present invention are for instance SDS and several zwitterionic detergents. Examples of zwitterionic detergents are Zwittergent^® 3-08, 3-10, 3-12 and 3-14 (the 3-14 compound of Calbiochem^® is synonym for n- Tetradecyl-N,N-dimethyl-3-ammonio-l-propanesulfonate) . The person of skill in the art is able to distinguish which detergents are suitable, as those would provide results with well-separated antigens that may easily be quantified. Unsuitable detergents that would disrupt or even destroy the antigens may just as easily be distinguished, since these would not give proper separated peaks in the chromatograms .

The problem to be solved was to provide an accurate, rapid and robust method, that would be applicable for high- throughput, and that would not have the disadvantages as found in the art. The inventors of the present invention have now found that RP-HPLC can nevertheless be used for this purpose and that such columns can be used if the temperature of the column was raised up to (but not including or above) the temperature with which the column material can no longer be used. The inventors have found that the hemagglutinin antigen, and especially the main determinant HAl is separated extremely well from the other proteins present in the preparation. This now enables the skilled person to measure the peak of the separated protein in an RP-HPLC chromatogram and to determine the amount present in the preparation, either by comparing it to other (known) values or to internal standards. Preferably, the elution is performed at a temperature above room temperature, typically above approximately 25°C. The inventors have found that values above room temperature are well suited for this purpose. It is also found that temperatures between about 50⁰C and about 70⁰C are even more suitable for said purpose. Thus, in a preferred embodiment, the invention relates to a method according to the invention, wherein said elution is performed at a temperature between about 25°C and about 70⁰C, more preferably between about 40⁰C and about 70⁰C, and even more preferably between about 50⁰C and about 70⁰C, most preferably between about 60⁰C and about 70⁰C. The best performance was detected at about 60⁰C and about 70⁰C, which latter value is close to the temperature with which the column material can no longer be applied for proper separation purposes. It is thus also part of the invention to perform the methods of the invention up to the highest temperature possible before the column material does no longer allow proper separation. Typically, most elutions were performed at a temperature of about 65°C. Suitable column materials that are typically used are polymer-based materials. Silica-based materials are less suitable, since they generally do not allow elution at high temperatures. The person skilled in the art of RP-HPLC can easily determine to what temperature certain column materials can be raised before they become useless for said purpose. So, in a highly preferred embodiment, the invention relates to a method according to the invention, wherein said elution is performed at a temperature of approximately 60⁰C, approximately 65°C or approximately 70⁰C.

It is to be understood that typical methods of RP-HPLC technology have been applied, and that a person skilled in the art of (RP-) HPLC is well aware of minor adjustments that would not alter the results to be obtained, such as different measurements at other suitable wavelengths or by the use of other column material that would not severely alter the results obtained by the present invention.

As mentioned infra, it is a well-known fact in the art that the mature influenza antigen HAO is processed to the sub-fragments HAl, and HA2, upon cleaving with for example trypsin. Since the methods according to the invention use the separation in RP-HPLC such that the HAl peak is measured for proper and accurate determination of the titer, it is preferred to have full cleavage of the mature protein. This can be achieved by a further step in which a protease compound is added that cleaves most if not all un-cleaved mature protein into the two desired sub-fragments. Typically, but not necessarily, the compound trypsin is used for this purpose. Thus, the invention also relates to a method according to the invention, comprising the further step of incubating the antigen preparation with a protease such as trypsin. This step is suitable for cleaving most if not all remaining un-cleaved mature forms of the HA antigen. Since the trypsin component is preferably removed from the solution before analysis, it is preferred to have the protease such as trypsin present on beads, preferably agarose beads. These beads can easily be removed by centrifugation, after the trypsin has cleaved most, if not all, HAO into its separate subunits. Clearly, in another setting, one could choose to add trypsin inhibitors after the trypsin has cleaved all HAO, in which case the use of beads is not necessary. Importantly, it was also noticed by the inventors that upon re-addition of DTT after alkylation, the HAl recovery seemed to be 6 to 10% higher than after reduction alone. Thus, in one preferred embodiment, a further step is included, wherein the reducing agent is added after alkylation of the reduced antigen in the sample preparation procedure.

The methods of the present invention now enable one of skill in the art to separate HAl from other proteins in a very robust, rapid and accurate way. The RP-HPLC chromatograms that are produced in machines applied for the methods of the present invention can also be used to determine the peak values of the separated proteins. Since these can be compared to known values of known antigens or to internal values used by the person carrying out the method, one is now able to accurately determine the amount of antigen present in the starting material. Thus, the present invention also relates to a method for quantifying the HA titer of an HA antigen preparation, said method comprising the method of separating the HA according to the invention, with the further step of measuring the peak area of the eluted antigen in a chromatogram resulting from the elution step. Preferably, said method of quantifying is applied for influenza antigens; a preferred embodiment relates to a quantification method according to the invention, wherein said HA antigen is of an influenza A virus . Examples

The following Influenza A antigens have been used herein (NIBSC-reference numbers underlined) :

A/New Caledonia/20/99 (HlNl) 00/608 A/Duck/Sing (H5N3) 00/522

A/Panama/2007/99 (Resvir-17; H3N2) 02/100 A/Equine/Prague/56 (H7N7) 85/553

All influenza antigens were obtained from the National Institute for Biological Standards and Control (NIBSC, Hertfordshire, United Kingdom) . The antigen A/Panama/2007/99 (D953-043F) was also produced using PER.C6^® cell-based technology.

Example 1. Determination of hemagglutinin in influenza preparations of A/Panama/2007/99 (Resvir-17; H3N2) using Reversed Phase HPLC. The egg-derived influenza antigen preparation

A/Panama/2007/99 (Resvir-17; H3N2) from NIBSC and the same antigen produced on PER.C6^®-based technology, were analyzed on Reversed Phase-HPLC (RP-HPLC) .

The production of antigen produced on PER.C6^® cells was performed as follows: PER.C6^® cells (as represented by the Human embryonic retina (HER) cells under ECACC no. 96022940 deposited with the European Collection of Cell Cultures (ECACC) at the Centre for Applied Microbiology and Research (CAMR) , Salisbury, Wiltshire, UK) were cultured in a bioreactor (37⁰C, DO = 50%, pH 7.3) until a viable cell density of IxIO⁶ cells/ml was accomplished. The cells were infected with influenza viruses of the strain Resvir-17 (H3N2) (35⁰C, with a multiplicity of infection of IxIO^"4) in the presence of 3 μg/ml trypsin/EDTA. The infection was continued for 5 days. The bioreactor content was then treated with 10 U/ml benzonase (Merck) for 30 min at 37⁰C. This was followed by clarification with a 3.0 μm filter (Clarigard, Millipore) and a 10-fold concentration step, using tangential flow filtration (Hollow-fiber module, Amersham) . Subsequently, the product was applied on sucrose gradient from 10 to 42% in PBS and centrifuged for 2 h at 22,000 rpm in an ultracentrifuge (Beckman) . The virus band was visible by the eye and was collected using a syringe. This material was used for development of the HPLC method.

Both batches of Resvir-17 antigen were disintegrated by addition of SDS (Gibco BRL) to a final concentration of 1% (w/v) , and reduced with 60 mM DTT in 0.15 M Tris, pH 8.0, for 30 min at 65°C. After cooling down, reduced proteins were alkylated by incubation with iodacetamide (IAA, final concentration of approximately 106 mM) at 37°C for 45 min in the dark. This alkylation step prevents the released proteins with free reactive sulfhydryl groups (e.g. HAl, HA2, and NA) from associating with each other.

Analysis was performed on an Agilent 1100 HPLC system with 900 μl loop injector, using a polystyrene dimethylbenzene POROS Rl/10 (2.1x100 mm) Reversed Phase column (Applied Biosystems) , and the gradient profile described in Table 1. Proteins were detected with a multiple wavelength detector at 215 nm.

Between 50-300 μl of sample was injected (approximately 10 μg HA as determined by SRID) , and RP-HPLC was performed with a flow of 0.8 ml/min and at a column temperature of 70⁰C.

The RP-HPLC assay according to the present invention for quantification of the HA titer in influenza virus preparations is based on measuring the peak area of its subunit HAl. The protein is solubilized upon addition by detergent, submitted to reduction/alkylation with DTT/IAA (respectively) , and subsequently analyzed utilizing the RP- HPLC procedure according to the schedule depicted in Table 1. As a consequence, a crucial parameter of the assay is the selectivity, i.e., the resolution between the HAl peak and other virus-derived material in a Reversed Phase chromatogram. The person skilled in the art is aware of the fact that the organic mobile phase may be performed with different agents. Typically, acetonitrile is used as solvent B (see Table 1) . Other solvents B that may be used are methanol, isopropanol and ethanol . As part of solution A and B (see Table 1) an anionic or cationic ion-pairing agent is typically used. Examples of anionic ion-pairing agents that may be used in the methods of the present invention are trifluoracetic acid (TFA) , pentafluoropropionic acid (PFPA) and heptafluorobutyric acid (HFBA) and the like. Examples of cationic ion-pairing agents that may be used in the methods of the present invention are tetramethylammonium chloride, tetrabutylammonium chloride and triethylamine.

The selectivity of the assay was explored first by analyzing formaldehyde-inactivated influenza A subtype Resvir-17 (H3N2) produced in chicken eggs at the NIBSC (Figure 1) . A total amount of Resvir-17 antigen corresponding to 10.0 μg HA was injected, and analyzed applying the acetonitrile gradient described in Table 1. The peak fractions as depicted in figure 1 were collected, and vacuum-evaporated for 45 min at 30⁰C to remove most acetonitrile from the samples. Subsequently, the fractions were concentrated on Microcon YM-IO filter devices (Amicon) according to the manufacturer's protocol, taken up in lithium dodecyl sulfate sample buffer (LDS, Invitrogen) , and analyzed by SDS-PAGE, silver staining and Western blot analysis to determine which fraction contained HAl. SDS-PAGE was carried out with NuPAGE 4-12% Bis-Tris gels (Invitrogen) at a constant voltage of 200 V for 55 min. Proteins were stained utilizing the SilverXpress^® silver staining kit (Invitrogen) according to the corresponding instruction manual. HA proteins and/or fragments were detected by Western blot analysis, using an antiserum from sheep raised against partially purified HA of A/Panama/2007/99 (H3N2) (NIBSC, catalogue no. 02/338) . For this purpose, the proteins analyzed on SDS-PAGE gels were blotted onto PVDF membranes (Millipore) for 1.5 h at 20 V. Next, the membranes were incubated for 1 h in blocking buffer (5% (w/v) non- fatty milk powder (BioRad) in TBST) , for 1 h in blocking buffer, containing the sheep anti-HA antiserum at a final dilution of 1:500, and finally in blocking buffer, containing rabbit anti-sheep horse radish peroxidase conjugate (Rockland, USA) at a final concentration of 1:6000. According to the instruction manual, ECL Western blotting reagents (Amersham) were used to detect the HA antigens .

The results of the silver stained SDS-PAGE gel are shown in figure 2. Apparently, the first peak with a retention time of about 8.9 min (fraction 2 in figure 1) contained all detectable HAl (molecular weight of approximately 55 kDa) , while being barely, if at all, contaminated with other proteins (Figure 2, lane 2). Western analysis confirmed that the 55 kDa band indeed contained HAl, as this band was clearly recognized by the anti-HA antiserum (Figure 3, lane 2) . Interestingly, in the starting material prior to the injection on the HPLC (Figure 3; lane A, which indicates the loaded antigen without purification over the column) a triplet of immunoreactive bands was visualized, most likely representing the intact monomeric, dimeric, and trimeric forms of HA, and therefore indicating that a substantial part of HA was resistant to cleavage into HAl and HA2. Complete cleavage is a prerequisite for an accurate quantification of HA samples. If it is unsure whether all HAO has been fully cleaved, it is thus preferred to have the HA fully cleaved by a protease before loading. This issue is further addressed below, in example 7. Arrows in figure 3 indicate the multimeric forms. This phenomenon has most likely been caused by the formaldehyde treatment of the antigen preparation, by which proteins together in a complex (like trimeric HA) are partly irreversibly cross- linked. As demonstrated in figure 3 (lane 4, corresponding to fraction 4 the in RP-HPLC of figure 1), these cross- linked HA forms eluted separately from the HAl form that eluted predominantly in fraction 2.

PER.C6^®-produced Resvir-17 antigen material was also analyzed by RP-HPLC (Figure 4) . This virus preparation was inactivated by beta-propiolactone (BPL) treatment, which in principle does not affect the characteristics of the viral proteins. In figure 4 the total amount of HA injected was approximately 16.6 μg. RP-HPLC analysis was performed utilizing the gradient profile as depicted in Table 1. Again, the peak fractions as denoted in figure 4 (eight in total) were collected, and prepared for SDS-PAGE, silver staining and Western blot analysis as already described in this section for egg-produced Resvir-17 antigen (figures 2 and 3, respectively) . It appeared that, in addition to the influenza virus encoded proteins, the PER. C6^®-produced batch of Resvir-17 antigen contained several other proteins (Figure 5, lane A, which indicates the antigen before application on the column) , most likely representing host cell proteins. This was also reflected by the RP-HPLC chromatogram of this batch, showing numerous peaks eluting between 10 and 15 min (Figure 4, peaks denoted as 3-6) . Nevertheless, the first peak with retention time of around 9 min (Figure 4) , contained HAl as demonstrated by the SDS- PAGE silver staining and Western blot analysis of the HPLC peak fractions (Figures 5 and 6, lanes 2), was well-resolved from other protein peaks, which shows that the methods are also very useful for methods in which the antigens are produced on tissue culture cells.

Consequently, these data indicate that the assay selectivity, i.e. the separation of HAl with the other viral components in both egg- and PER.C6^®-derived H3N2 Resvir-17 antigens, was excellent.

Example 2. Determination of hemagglutinin in influenza preparations of A/Duck/Sing (H5N3) and A/New Caledonia (HlNl) using Reversed Phase HPLC.

Further, it was investigated whether the RP-HPLC assay was also applicable for hemagglutinins from other influenza A subtypes. Hence, the selectivity of the assay with two other influenza A subtypes, A/Duck/Sing (H5N3) and A/New Caledonia (HlNl) was determined. First, an RP-HPLC was performed on egg-derived and formaldehyde-treated H5N3 from A/Duck/Sing. For this an amount corresponding to 3.0 μg HA was injected. Further procedures were as described in example 1, except that instead of SDS, Zwittergent 1% (w/v) was used as the detergent. In figure 7, a Reversed Phase chromatogram of the reduced/alkylated H5N3 antigen is shown. SDS-PAGE and subsequent silver staining (Figure 8) of the proteins demonstrated that fraction 2b contained most, if not all HAl (Lane 2b) . Notably, peak 1 (lane 1 in figure 8) , although eluting first after the flow through, did not contain HA proteins; hardly any proteins were discernible in this fraction in SDS-PAGE. An amount of 0.19 μg HA antigen that was not applied on the column was taken as a positive control (lane Duck/Sing) .

A graph of the RP-HPLC of egg-derived, reduced and derivatized influenza A subtype HlNl (A/New Caladonia) is shown in figure 9. An amount corresponding to 15 μg HA was reduced and alkylated under non-buffered conditions, injected on the HPLC, and subsequently analyzed running the acetonitrile gradient presented in Table 1. Further procedures were as described in example 1, except that Zwittergent 1% (w/v) was used as the detergent. The first peak (denoted as 1) with a retention time of about 11.4 min, contained predominantly HAl, as shown by silver staining (Figure 10, left panel) and Western blot analysis (Figure 10, right panel). The retention time of approximately 11.4 minutes differed significantly from the retention time of the HAl peak of A/Resvir-17 (Figure 1), which was about 8.9 minutes. HAl of A/Resvir-17 has a higher polarity (more hydrophilic) than its counterpart of influenza A/New Caledonia, probably due to the difference in amino acid content. Again, since this antigen batch had also been inactivated by formaldehyde treatment, not all HAO could be cleaved into its subunits, and, hence, a part of HAO apparently migrated as uncleaved and multimeric forms in the gel (Figure 10, right panel, lanes 5 and A, indicated by arrows) .

Taken together, these data demonstrate that the assay selectivity for quantification of HAl is excellent, and in addition, that the RP-HPLC assay is not specific for a particular influenza A subtype, but that it can be applied broadly for different types of influenza viruses.

Example 3. RP-HPLC assay linearity for quantification of HA of egg-derived influenza A subtype H3N2.

One of the key criteria of an analytical procedure is linearity, the ability (within a certain range) to obtain test results, which are directly proportional to the concentration (amount) of analyte in the sample. Assay linearity was studied with egg-derived, formaldehyde- inactivated, reduced and alkylated influenza antigen from A/Panama/2007/99 (A/Resvir-17; H3N2) . Increasing concentrations of HA were injected on the RP-HPLC system with a constant injection volume of 200 μl, and subsequently plotted versus the measured area of the HAl peak, resulting in a calibration curve as shown in figure HA. It is evident from the data in figure HA, that the assay linearity in the range between 0.3 and 10.6 μg HA injected was very good, as indicated by a correlation coefficient (R²) of more than 0.99. In principle, the real working range is also determined by the accuracy of the HA concentration of the calibration samples measured by utilizing the calibration curve. In a so-called residual analysis, in which the deviation of the actual data points from the regression line (calibration curve) was calculated, it was revealed that the percentage deviation (experimental from predicted HAl area) for most data points was smaller than 5%. Accepting ± 15% difference, in this particular experiment no data points of the curve had to be left out, and, hence, the actual operating range to determine the HA titer was limited between 0.3 and 10.6 μg HA injected. As an important conclusion, as discussed infra, the HA titer of formaldehyde-inactivated influenza samples like the one used above cannot be determined to the highest possible accuracy. Moreover, in the linearity study just discussed sample preparation had not been optimal. Taking these two points into account, linearity was, therefore, also studied with a PER.C6^®-derived but BPL-inactivated A/Resvir-17 sample. The data showed that for the BPL-inactivated influenza A/Resvir-17 sample good assay linearity could be achieved (Figure HB) .

Example 4. RP-HPLC assay precision .

Assay precision was studied by analyzing six injections of a reduced and alkylated sample of A/New Caledonia (HlNl) at a relatively low concentration (0.65 μg HA per injection) . In addition, precision was also explored by injecting four independently reduced and alkylated samples with a relatively high HA titer (about 3 μg HA per injection). Results are shown in Tables 2 and 3, and demonstrate that the precision was good for both sets of samples with CVs below 10%. Example 5. Effect of column temperature on RP-HPLC assay performance of egg-derived Resvir-17 antigen (H3N2) .

The effect of column temperature on the assay was also studied. In this respect, egg-derived Resvir-17 (H3N2) antigen was reduced and alkylated as described above, except that these reactions were conducted under non-buffered conditions. Subsequently, samples (approx. 4.3 μg HA per injection) were analyzed at the following column temperatures: 25°C, 40⁰C, 50⁰C, 60⁰C, and 70⁰C, using the acetonitrile gradient in an organic mobile phase as described in Table 4.

In figure 12A the chromatograms of column temperatures 25°C and 70⁰C are compared. It is evident that the recovery of all peaks was higher at a column temperature of 70⁰C. In Table 5, the peak areas of HAl and three other peaks, which were denoted in figure 12A (lower panel) as Peak 2, 3 and 4 and which were in general also obtained after RP-HPLC at the column temperatures mentioned above, are presented. Peak 3 could not be distinguished at a column temperature of about 25°C and about 40⁰C. It is herein demonstrated that for optimal recovery of HAl the column temperature range is preferably above about 25°C, more preferably above about 40⁰C and most preferably above about 50⁰C, whereas for recovery of the other three peaks it is preferred to use a column temperature of approximately 70⁰C. The results as shown in Table 5 show that the most preferred temperature range of the column is between about 50⁰C and about 70⁰C, while the best results were achieved with a temperature of approximately 60⁰C.

In a subsequent experiment, the effect of column temperature on RP-HPLC of influenza A/Resvir-17 was again investigated, and at each test temperature samples were analyzed in triplicate. In addition, influenza A/New Caledonia (HlNl) was taken along (single injection at each temperature) . As illustrated in Figure 12B, the recovery of HAl from influenza A/Resvir-17 (H3N2) was significantly enhanced when the column temperature was increased: between about 60⁰C and 70⁰C the HAl peak area was the largest. No temperatures higher than 70⁰C were explored (according to the manufacturer the maximum allowed temperature for this column is 80⁰C) . With regard to influenza A/New Caledonia

(HlNl) a similar tendency was seen, although less pronounced as for influenza A/Resvir-17. At about 50, 60 and 70⁰C more or less of an equilibrium with the same HAl recoveries was acquired for A/New Caledonia. Consequently, the data described in this section point out that utilizing this particular column a column temperature between about 60⁰C and 70⁰C was optimal for RP-HPLC quantification of HA.

Example 6. Determination of hemagglutinin in influenza preparation of A/Equine/Prague/56 (H7N7) using Reversed Phase HPLC.

It was further investigated whether the RP-HPLC assay was also applicable for hemagglutinin from influenza A subtype H7N7. Hence, the selectivity of the assay with formaldehyde-inactivated subtype A/Equine/Prague/56 (H7N7) was determined. Sample preparation and further procedures were generally as described in example 1 (1% SDS as detergent, reduction with 65 mM DTT, 65°C, 30 min, alkylation with IAA 116 mM, 37°C, 45 min in the dark) . In figure 13, a Reversed Phase-HPLC chromatogram of the reduced and alkylated H7N7 antigen is shown. SDS-PAGE and subsequent silver staining (Figure 14) of the proteins demonstrated that, based on the size of the protein as compared to the size marker, fraction 2 of figure 13, a relatively irregularly shaped peak, contained predominantly HAl (Figure 14, silver staining, Lanes depicted as '2'). Fraction 4 contained HA2 and the non-cleaved HAO forms.

Taken together, as mentioned under example 2, these data demonstrate that the assay selectivity for quantification of HAl is excellent, and in addition, that the RP-HPLC assay is not specific for a particular influenza A subtype, but that it can be applied broadly for different types of influenza viruses.

Example 7. Full cleavage of HAO in its subunits HAl and HA2 As noticed above, HA in both formaldehyde-inactivated egg-based and beta-propiolactone-inactivated PER.C6^®-derived influenza A/Resvir-17 (H3N2) did not turn out to be fully cleavable upon reduction (59 mM DTT, 30 min at 65°C) . This is a highly undesired situation, as the most accurate HA quantification by RP-HPLC requires a full cleavage of HA into HAl and HA2. It was explored whether this could be accomplished by applying more severe reduction conditions. However, the ratio between non-cleaved HAO and HAl was not affected by reduction for longer times and/or at higher temperatures (data not shown) . This indicated that most likely part of the HA in both preparations had not been cleaved enzymatically, and could therefore never be split upon reduction.

Then, it was investigated whether the cleavage of the residual un-cleaved HA was possible by additional trypsin treatment of both vaccins dissociated in 1% SDS or 1% Zwittergent. Under these conditions, HA is in its trimeric format, which is susceptible to the trypsin-induced specific cleavage into its subunits, but resistant to further proteolytic breakdown. For this, agarose beads conjugated with trypsin were utilized, which enables one to remove the beads conveniently by centrifugation after digestion, thereby also avoiding possible proteolytic degradation of trypsin-sensitive HAl during the further sample preparation (e.g. the reduction step). Buffer solution was 134 mM Tris- HCL, pH 8.0 and reduction was performed with DTT at 100⁰C for 10 min. Alkylation was performed by IAA treatment as in the previous examples. During trypsin treatment (15, 35, 60 and 120 min) , samples were rotated in an oven at 37⁰C to prevent the trypsin beads from precipitating in the sample tubes . Pre-treatment of BPL-inactivated PER.C6^®-based Resvir- 17 in 1% Zwittergent resulted in the disappearance of the residual amount of un-cleaved HAO, while further breakdown of HAl and/or HA2 was not detected (left panel, Figure 15) . In contrast, trypsinization in the presence of 1% SDS did not lead to cleavage of the residual HAO (right panel,

Figure 15) , most likely because the trypsin activity was abolished by this concentration of denaturing SDS. Thus, when trypsin is used to fully cleave the HAO protein, it is preferred not to use SDS, but rather to use a detergent such as Zwittergent. When the trypsinized sample in 1%

Zwittergent was analysed by RP-HPLC and compared with Resvir-17 HA that was not treated with trypsin, an increase of up to 10% in HAl peak area could be detected upon trypsin treatment prior to reduction and RP-HPLC analysis (Table 6) . These data demonstrate that trypsin pre-treatment (30 min at 37°C) is a highly preferred sample preparation step in the RP-HPLC method for HA quantification of influenza vaccines. In contrast, the situation for formaldehyde-inactivated egg-based A/Resvir-17 was quite different: in the presence of either SDS or Zwittergent cleavage of HAO by additional trypsin pre-treatment did not turn out to be possible (data not shown) . This was most likely due to the formaldehyde- inactivation treatment (different from the beta- propiolactone treatment discussed above) . Formaldehyde is known to cause irreversible cross-linking of proteins, which are in a complex, like the disulfide-linked HAl and HA2. Consequently, it is preferred to use trypsin to ensure a full cleavage of HAO, but if trypsin is used it should be used in a detergent such as Zwittergent, rather than SDS, while trypsin should preferably be used on samples that were previously inactivated through an inactivating agent such as BPL, rather than cross-linking inactivating agents such as formaldehyde .

Example 8. Sample stability: effect of reducing and alkylation conditions on RP-HPLC of HA of A/Resvir-17 (H3N2) Initially, influenza samples were reduced and alkylated under non-buffered conditions. When injected immediately after the non-buffered reduction and alkylation reactions (1% Zwittergent; reduction 59 mM DTT, 65°C 30 min; alkylation 106 mM IAA, 37°C 45 min in the dark), RP-HPLC of PER.C6^®-based A/Resvir-17 HA resulted in a sharp HAl peak, eluting at approximately 8.9 min in the chromatogram (Figure 16, panel A) . However, it was noted that this sample was not stable: overnight storage of this batch (approximately 17 h) at 4°C gave rise to a novel peak at 8.6 min, which was accompanied by a decrease of the original HAl peak eluting at 8.9 min (Figure 16, panel B). This suggested that part of the HAl became slightly more hydrophilic in time. Similar observations were made for egg-based Resvir-17. First, it was thought that this phenomenon was related to the non- buffered status of the samples. However, this did not appear to be the only explanation, because in another PER.C6^®-based influenza A/Resvir-17 batch (1% Zwittergent, reduction with 57 mM DTT 10 min at 100⁰C, buffered at pH 8.0, alkylation with 102 mM IAA 45 min at 37°C in the dark), stored at 21°C instead of 4⁰C, comparable changes in the HAl peak shape occurred (Figure 17A) , although to a lesser extent as shown in Figure 16B.

On the other hand, the observed HAl peak deformation in time might also be caused by the presence of residual amounts of IAA, as it is generally known that IAA may give rise to relatively strong adverse effects on the integrity of proteins. Then, the effect of omission of the alkylation step after reduction was studied. This however, did not have a significant effect on HAl recovery, but, interestingly, regarding the HAl peak shape, samples proved to be far more stable in time (Figure 17B) . Consequently, the data as depicted in Figure 17 suggest that HAl was relatively stable for 23 h at 21°C under strictly reducing conditions (DTT) , but not when most (if not all) DTT was neutralized by IAA.

To distinguish whether the apparent HAl instability was due to the absence of reducing circumstances or to possible disadvantageous side effects of the residual amount of IAA chemically modifying the protein, the HAl stability was also monitored after reduction (without subsequent alkylation) at various DTT concentrations before and after storage for 18 h at 4°C. As can be seen in Figure 18, at all tested DTT concentrations (1.1, 2.2, 4.4, 11 and 22 mM) , the previously observed additional peak that eluted just before the original HAl peak (see Figure 16B and 17A) was not observed anymore, when stored for at least 18 hours (panel B) , indicating that the HAl peak transformation must have been caused by the IAA-related chemical modifications of the protein. Unexpectedly, a different (putative) HAl-peak instability was observed: after 18 h at 4°C and at low DTT- concentrations (1-4 mM) a small, but significant peak was discernible in the tailing part of the original HAl peak (Figure 18B) . At higher DTT concentrations (11 and 22 mM) , this little peak did not evolve. So, these higher concentrations of DTT are preferred. Overall, it is preferred to use concentrations of DTT higher than about 4.4 mM, more preferably at least about 11 mM and most preferably about 22-25 mM. The stability of the HAl peak area was monitored in triplicate for both a reduced/alkylated/DTT treated sample and an only reduced PER.C6^®-based influenza A/Resvir-17 sample before and after storage for 20 h at 4°C. Notably, reduction was carried out at a DTT concentration of 25 mM, and after the alkylation reaction (as for half of the samples) DTT was re-added to a final concentration of 25 mM, to prevent any HAl peak deformation. It turned out that, unlike the experiment of Figure 18, the HAl peak area was barely, if at all, affected by storage for 20 h at 4°C (Table 7). Importantly, it was noticed that upon reduction/alkylation and re-addition of DTT to the samples the HAl recovery seemed to be at least 6 to 10% higher than after reduction alone. A possible explanation is that alkylated HAl exhibits less (non-specific) absorption to the column than its non-alkylated counterpart. An alternative explanation might be that the molar extinction coefficient of HAl was enhanced by the alkylation, leading to relatively higher signals at 215 nm. Whatever the reason, based on the data of Table 7, it is highly preferred to include the step of adding the reducing agent after alkylation in the sample preparation procedure. Similar results were obtained with the influenza A/New Caledonia (HlNl) strain derived from eggs. This embodiment of the method of the invention is depicted schematically in the flow diagram of Figure 19.

Example 9. Effect of trypsin concentration on recovery of HAl

Above, it has been indicated that trypsin treatment ensures a full cleavage of HA into its subunits HAl and HA2. It should be noted that if the cleavage, due to for instance cellular proteases may be complete and that an extra trypsin treatment may be omitted. Nevertheless, to ensure that all HAO is cleaved, it is preferred to add the additional trypsin step. An experiment was designed to explore the effect of increasing concentrations of trypsin (preferably present on beads) on the ultimate recovery of HAl from three samples of PER.C6^®-based influenza A/New Caledonia, which differed in the amount of virus and, hence, in HA content. It turned out that for each of the three samples addition of 120 mU trypsin beads, and subsequent incubation for 30 min at 37°C resulted in optimal HAl recoveries (Figure 20A) . It was also investigated whether these conditions were suitable for batches, containing higher amounts of influenza virus, having HA titers of approx. 3.4, 6.8, 13.6, and 17 μg/ml . As illustrated in Figure 2OB, this was indeed the case for the influenza samples containing up to 13.6 μg HA/ml : the maximal recovery of HAl from these samples was attained after 30 min incubation at 37°C with 120 mU trypsin beads and longer incubation times did not lead to higher HAl peak areas. As for preps with higher concentrations of influenza virus most likely longer incubation times may be required.

Example 10. Comparison RP-HPLC system versus SRID As disclosed herein, it was demonstrated that the assay selectivity, linearity and precision of the method according to the invention were good. To explore whether the RP-HPLC assay according to the invention would provide for a proper alternative for the cumbersome and slow SRID assay, results were compared between the two assays. In Table 8 a first comparison was made between both assays for a number of A/Resvir-17 samples. It must be noted that trypsin pretreatment and alkylation were not included as standard steps. Six different samples were compared, whereas the concentration of the samples A, C and D were determined in triplicate by HPLC (e.g. Al, A2, A3). The Table shows that the HA titers obtained by RP-HPLC closely resembled those acquired by SRID.

Interestingly, the data also demonstrated that formaldehyde treatment of A/Resvir-17 resulted in a greatly reduced HAl peak as compared to the same, but BPL- inactivated batch (Table 8; compare samples E and F) , supporting earlier conclusions that HA quantification in formaldehyde-inactivated influenza batches is far from accurate.

Subsequently, these comparative studies between RP-HPLC and SRID were repeated in more detail for a series of A/New Caledonia samples. As shown in Table 9, the HPLC data agreed well with the SRID-based titres. Consequently, it was thus established that the RP-HPLC assay is accurate, and represents a good alternative for the SRID assay to quantify the HA concentration in influenza virus containing batches. This is certainly the case when taking into account that the HPLC assay precision is better than the precision attained by SRID (Table 9, see RSD values for sample D) .

Example 11. HA quantification in crude samples

From a process development point of view, another important application of the assay would be measuring the HA concentration in crude culture supernatants of cells infected with influenza virus. To explore the feasibility of this potential application we studied the assay selectivity with regard to the sample matrix, i.e. conditioned growth medium of PER.C6^®. When analyzing a sample of a crude culture supernatant of PER.C6^® grown in AEM medium and infected with influenza A/Resvir-17 without any sample treatment other than reduction alone, a rather complex chromatogram was recorded (Figure 21A) . Although a peak was discernible having the same retention time as HAl (Figure 21A, indicated by an arrow) , it became immediately clear that quantification of HAl by measuring the HAl peak area was impossible due to the large amount of interfering material surrounding the putative HAl peak. It was reasoned that an additional trypsin treatment might solve the observed problem of lack of assay selectivity for crude samples by digesting the interfering proteins in the sample. This indeed turned out to be the case, as the RP- chromatogram of a comparable culture supernatant pre-treated with trypsin (coupled to agarose beads) exhibited a HAl peak almost free from other material and thus relatively easy to integrate (Figure 21B) . However, as it could not be excluded that part of the HAl might have been broken down during

(further) preparation of this sample due to the presence of soluble trypsin, which is already present in cell cultures when the influenza viral infection has to be stimulated (general procedure) , it was also investigated whether the HAl recovery could be augmented by first centrifuging the virus, and then treating the virus pellet with the trypsin beads (after removal of the supernatant containing the soluble trypsin) . It appeared that this approach indeed resulted in a significantly (about 30%) enhanced HAl peak area (Figure 21C) . Thus, for routine crude sample analysis for quantification of HA it is preferred to include at least one centrifugation step in the sample preparation to remove the harmful soluble trypsin and proteins, which interfere with the integration of the HAl peak in the chromatogram. Another advantage is that the virus is concentrated by this strategy. Centrifugation is typically performed for 30 min at 4°C with centrifugal force values of 4500 g or higher, preferably higher than 6000, more preferably higher than 9000, even more preferably higher than 12,000 g, whereas the it is most preferred to use at least 17,000 g, because the HA recovery values were up to 100% when 12,000 g to 17,000 g was used. One preferred embodiment of the method according to the invention in which HAl is quantified in crude (supernatant) samples of cells infected with influenza viruses is shown in Figure 22.

Example 12. Determination of hemagglutinin in influenza B viruses using Reversed Phase HPLC

It was investigated whether the RP-HPLC assay was also applicable for hemagglutinin from B-strains. Hence, the selectivity of the assay with PER.Cβ-produced influenza B/Jiangsu, and BPL-inactivated egg-based B/Guangdong was determined. Sample preparation and further procedures were as described for A-strains. The acetonitrile gradient was slightly changed from the method for A-strains (Table 10). In figure 23, a Reversed Phase chromatogram of the trypsin- treated, reduced and alkylated B/Jiangsu sample is shown. SDS-PAGE and subsequent silver staining of the proteins demonstrated that based on the size of the protein band fraction 2 of figure 23 contained predominantly HAl and a trace of putatively influenza Nucleoprotein (NP; Figure 24, Lane 3) . The HAl band was not seen in any of the other HPLC fractions. Together with the observation that non-cleaved HA formats were absent in the starting material (Figure 24, Lane A) and any of the HPLC fractions, this demonstrates good assay selectivity for HA quantification of B/Jiangsu samples .

In figure 25, a Reverse Phase chromatogram of the trypsin-treated, reduced and alkylated, egg-based

B/Guangdong sample is shown. SDS-PAGE and subsequent silver staining of the proteins demonstrated that based on the size of the protein band fraction 2 of figure 23 contained predominantly HAl (Figure 26, Lane 3) . It may be that not all HA was cleaved by the applied trypsin-treatment, as a band with the expected molecular weight for non-cleaved HA was still visible in the starting material and in the combined HPLC fractions 3, 4, and 5 (Figure 26) . Nevertheless, assay selectivity was acceptable, as all HAl eluted in a single peak, well resolved from the other vaccine components. These experiments indicate that the RP- HPLC assay for separation and/or quantification of HA antigens (in the form of HAl) from influenza viruses can be applied both for influenza A strains, as well as for influenza B strains. Table 1. Gradient profile of the RP-HPLC influenza quantification method. Solvent A is 0.1% trifluoracetic acid (TFA) in 5% acetonitrile, solvent B is 0.098% TFA in 100% acetonitrile.

Table 2. RP-HPLC assay precision (or repeatability) data of six injections of an egg-derived, reduced and alkylated sample of influenza A/New Caledonia HlNl. Amount HA per injection was 0.65 μg.

Table 3. RP-HPLC assay precision data of four independently reduced and alkylated samples of egg-derived influenza A/New Caledonia HlNl. Amount HA per injection was 3.0 μg.

Table 4. Gradient profile of the RP-HPLC influenza quantification method. Solvent A is 0.1% trifluoracetic acid (TFA) in 5% acetonitrile, solvent B is 0.098% TFA in 100% acetonitrile.

Table 5. Effect of column temperature on RP-HPLC of egg- derived, reduced and alkylated Resvir-17 antigen (H3N2) .

Temperature Peak Area ( 215 nm)

HA1 Peak 2 Peak 3 Peak 4

25 1084581 (8.51) 79310 (14.57) ? 236605 (27.41)

40 1092810 (8.32) 112307 (14.19) ? 264513 (27.51)

50 1150764 (8.08) 165212 (13.75) 18343 (22.73) 302645 (27.48)

60 1231606 (7.74) 220181 (13.15) 36399 (21.38) 337627 (27.37)

70 1200473 (7.35) 239249 (12.44) 63262 (19.60) 354590 (27.21)

Table 6. Effect of pre-treatment of influenza A/Resvir-17 sample with trypsin on the peak area of HAl in RP- chromatograms

Table 7. Effect of reduction/alkylation/DTT treatment versus reduction only on the recovery of HAl derived from a non- trypsinized PER.C6^®-based influenza A/Resvir-17 batch (H3N2) measured by RP-HPLC. Amounts injected: approx. 2.9 μg HA.

Table 8. Comparison of the HA titers of seven A/Resvir-17 samples determined by RP-HPLC and SRID. An A/Resvir-17 batch with a HA concentration of 1161 μg HA/ml was taken as reference (for calibration in HPLC) .

Table 9. Comparison of the HA titers of 5 influenza A/New Caledonia samples (A-D) determined by RP-HPLC and SRID. Different fractions were taken. An A/New Caledonia batch with a HA concentration of 90 μg HA/ml was taken as reference (for calibration in HPLC) .

Table 10. Gradient profile of the RP-HPLC influenza B virus quantification method. Solvent A is 0.1% trifluoracetic acid (TFA) in 5% acetonitrile, solvent B is 0.098% TFA in 100% acetonitrile.

References

Bizhanov, Kastrikina, Lonskaya, and Popov (1988) Influence of detergents on measurement of influenza haemagglutinin content in inactivated influenza vaccine by single radial immunodiffusion. Acta Virol 32:252-260

Johannsen, Moser, Hinz, Friesen, and Gruschkau (1985) Quantification of haemagglutinin of influenza Tween-ether split vaccines by immunodiffusion. Vaccine 3 (Suppl 1985) :235-240

Kemp MC, Holloway, Bennett and Compans (1980) Separation of Influenza hemagglutinin tryptic glycopeptides by ion-pair Reverse-Phase High-Performance Liquid Chromatography (HPLC) . J Biochem and Biophys Methods 3:61-63

Lamb and Krug. (2001) Orthomyxoviridae: the viruses and their replication. In: Fields Virology VoI 1 (4^th edition) , pp 1487-1531. Eds. Knipe, Howley, Griffin, Martin, Lamb, and Roizman. Lippincott, Williams & Wilkins, Philadelphia

Pereira (1973) Final discussion on standardization of influenza vaccines. Symp Ser Immunobiol Stand 20:378

Phelan and Cohen (1983) Gradient optimization principles in reversed-phase high performance liquid chromatography and the separation of influenza virus components. J Chromatography 266:55-66

Van der Zee R, Welling-Wester and Welling (1983) Purification of detergent-extracted Sendai virus proteins by Reversed-Phase High-Performance Liquid Chromatography. J Chromatography 266:577-584

Villkommen, Platen, and Staber (1983) The influence of pH and ionic strength on the single-radial-immunodiffusion test quantitative assay of influenza virus haemagglutinin. Acta Virol 27:407-411

Wood, Schild, Newman, and Seagroatt (1977) An improved single-radial-immunodiffusion technique for the assay of influenza haemagglutinin antigen: application for potency determinations of inactivated whole virus and subunit vaccines. J Biol Stand 5:237-247

Wright and Webster (2001) Orthomyxoviruses. In: Fields Virology Vol. 1 (4^th edition), pp 1533-1578. Eds. Knipe, Howley, Griffin, Martin, Lamb, and Roizman. Lippincott, Williams & Wilkins, Philadelphia

Claims

Claims

1. A method for separating hemagglutinin (HA) antigens from an influenza B virus, said method comprising the steps of: applying a reduced and derivatized antigen preparation comprising solubilized HA antigens from an influenza B virus, and a detergent in a pH controlled solution, on a Reversed-Phase High- Performance Liquid Chromatography (RP-HPLC) column; and eluting the HA antigens from the column with an ion pairing agent in an organic mobile phase.

2. A method according to claim 1, wherein said influenza HA is substantially separated into the subunits HAl and HA2.

3. A method according to claim 1, wherein derivatization of the antigen preparation is performed with an alkylating agent.

4. A method according to claim 3, wherein the alkylating agent is iodacetamide (IAA) .

5. A method according to claim 1, wherein the pH of the solution has a value between about 5 and about 9, preferably between about 6 and about 8, and more preferably between about 7 and about 8.

6. A method according to claim 1, wherein said detergent is SDS or a zwitterionic detergent.

7. A method according to claim 1, wherein said elution is performed with a column comprising polymer- based material.

8. A method according to claim 1, wherein said elution is performed using a column comprising a material that can withstand temperatures up to at least about 70⁰C.

9. A method according to claim 1, wherein said elution is performed at a temperature between about 25°C and about 70⁰C, preferably between about 40⁰C and about 70⁰C, more preferably between about 50⁰C and about 70⁰C, and most preferably between about 60⁰C and about 70⁰C.

10. A method according to claim 1, wherein said elution is performed at a temperature of approximately 60⁰C, approximately 65°C or approximately 70⁰C.

11. A method according to claim 1, comprising the further step of incubating the antigen preparation with trypsin for cleaving the mature form of the HA antigen into HAl and HA2.

12. A method according to claim 11, wherein said trypsin is present on beads.

13. A method according to claim 1, wherein the reduction of the antigen preparation is performed with dithiotreitol (DTT) .

14. A method according to claim 13, wherein the concentration of DTT is between about 4.4 and about 11 mM, preferably between about 11 and about 22 mM, and more preferably about 25 mM.

15. A method according to claim 3, further comprising the step of adding an alkylation-inhibiting agent after alkylation of the antigen preparation.

16. A method according to claim 15, wherein said alkylation-inhibiting agent is a reducing agent, preferably DTT.

17. A method for quantifying the HA titer of an HA antigen preparation, said method comprising the method of claim 1, with the further step of measuring the peak area of an eluted antigen in a chromatogram resulting from the elution step.

18. A method according to claim 17, wherein the peak that is measured represents the HAl subunit of influenza HA.