Application Note 20「A fast and high precision influenza vaccine potency assay」

APPLICATION NOTE 20 A fast and high precision influenza vaccine potency assay David Wheatley  1 Principal Scientist, Debi Saunders  1, Staff Scientist, Kyle Jones  1, R&D Scientist and David Apiyo  2, Marketing Applications Manager 1  Pall Life Sciences, 2  ForteBio Overview Vaccines are biological preparations that contain agents resembling disease causing microorganisms, and can improve immunity against a specific disease. They are typically prepared from inactivated or weakened forms of the microbe or its toxins, or surface proteins. Classical vaccines against the influen- za virus are developed in embryonated hen eggs and may include whole virus, split virus or a purified subunit with every component other than hemagglutinin (HA) or neuraminidase (NA) removed.1 The target molecule for the protective immune response triggered by vaccination is generally accepted to be the HA molecule; a glycoprotein found on the surface of the influenza virus. Measuring the vaccine potency or the biolog- Figure 1: Octet system with sample plate loaded. ically active components is critical to the determination of the vaccine’s effective dose. In addition, the stability of the vac- cine has major impact on its usage for immunization programs worldwide. Although real-time stability studies under different storage conditions is preferable, thermal stability testing using The relative standard deviation and dynamic range of a vaccine potency assays with samples subjected to heat or environ- titre assay was tested for the influenza virus using the Octet mental stress conditions can be used as predicators of vaccine platform and was found to be better than that encountered with stability over time.2 SRID. Unlike in the SRID technique where detergents are used A fast and accurate determination of vaccine titer during man- to expose the target HA molecule, with Octet systems, samples ufacturing is important in understanding vaccine development are analyzed in their natural state without the use of detergents. process performance, and for correctly scaling each process As a result, Octet systems are capable of analyzing whole virus, step. The Single Radial Immunodiffusion (SRID) technique has split virions and recombinant HA vaccine samples. been the most commonly used technique for vaccine titer determination. However, SRID is time consuming and generally Materials and reagents exhibits poor precision. An alternative assay that can speed up the analysis process and provide accurate and precise poten- • Samples of inactivated virus and antibody standards were cy data on different vaccine strains is therefore desirable. The purchased from NIBSC, South Mimms, UK. Split virion vaccine Octet® platform’s Bio-Layer Interferometry (BLI) technology samples were provided by Sanofi Pasteur, France, and recom- combines the high-throughput characteristics of a 96-well or binant hemagglutinin vaccine samples and antibodies were 384-well plate format with improvements in precision and repro- provided by Protein Sciences, USA. ducibility and is derived from a simpler and more direct vaccine/ • Sample Diluent (Part No 18-1048), Protein A and G biosen- antigen–antibody binding measurement method. They provide sors (Part No 18-5010 and 18-5082 respectively), were process development groups with a robust and easy to use provided by ForteBio. alternative to the SRID method. BLI reduces the assay time from • Black polypropylene 96-well sample plates from Greiner, days to just a few hours for a 96-well plate of samples. Part No 655209 (Sigma Aldrich Part No M9685) were used. 1

Equilibration Loading Baseline Association Figure 2: Assay workflow for the influenza vaccine titer assay. The assay is run using the Advanced quantitation setup in the Octet Data Acquisition software. Strain-specific Vaccine molecule antibody Method 3.5 The assay is based on the binding of the vaccine sample to 3.0 polyclonal antibodies that recognize the influenza epitopes presented by the vaccine. Protein A or Protein G derivatized bi- 2.5 osensors are first used to load the specific polyclonal antibody from the serum antibodies (Figure 2). The antibody-immobilized 2.0 biosensors are then dipped into the vaccine samples and a response signal that depends on binding epitope recognition 1.5 and vaccine concentration is registered. To determine the optimal antibody concentration to capture 1.0 onto the biosensor, a serial dilution of the serum antibody sam- ple was performed using ForteBio Sample Diluent. 0.5 Figure 3 shows the biosensor binding profiles of a range of 0 antibody dilutions starting from 1/10 to 1 /2000 of the neat 0 50 100 150 200 250 Time (sec) antibody samples obtained from NIBSC as California/7/2009 Figure 3: Antibody Load Scouting using Protein A biosensors – (H1N1); an A strain inactivated virus standard and antibody pair. A/California/7/2009(H1N1). Typical Octet assay antibody loading time is 300 seconds with a shake speed ranging from 400–1000 RPM. The antibody Loading Baseline loading time can vary significantly from virus strain to strain and should be evaluated. 1.0 While high precision Streptavidin (SAX) biosensors are rec- ommended for multi-step quantitation assays, Protein A or G 0.8 biosensors can be used especially when dealing with non-pu- rified IgG samples. The biosensors were first tested in five replicates for each lot in a biosensor lot to lot robustness study 0.6 for antibody loading (Figure 4, Table 1). The biosensors were found to be highly robust to loading variations. However, it is 0.4 critical to include a referencing biosensor (zero analyte) in each assay to subtract off potential post antibody baseline drift. The second stage of the assay optimization involves dipping the 0.2 antibody-loaded biosensors into a fixed concentration of the vaccine sample to monitor response (Figure 5). Both antibody loading and antigen binding during assay optimization are 0 100 200 300 400 500 600 Time (sec) Biosensors Average response % CV Figure 4: Antibody loading at 1/250 dilution using different lots of Protein A biosensors. Lot 1 0.489 2.41 Lot 2 0.463 2.63 Lot 3 0.455 1.53 Lot to Lot % CV 3.78 Table 1: Protein A biosensors lot to lot loading response monitored at T = 500 seconds 2 Binding (nm) Binding (nm)

performed in sequence on the Octet instrument using the Advanced Quantitation experimental setup in the Octet Data 1.6 Acquisition software. The lowest coating antibody concentration (1/250 dilution in this example) that gives maximal vaccine binding should be select- 1.2 ed for the studies. For influenza virus, the sample binding step is typically 300 seconds except for split virion analysis, where the step should take less than 30 seconds as virus heterogene- 0.8 ity can occur with time. To determine titer, a standard curve is required and is generated using a titration of standard samples whose concentration is known. The response signals obtained 0.4 from the standard samples are analyzed by calculating the bind- ing rate for the initial slope of the binding curve. The measured binding rate is then plotted as a function of the standard sam- 0 ples concentration and the data fit to a dose response equation 0 50 100 150 200 250 300 Time (sec) resulting in a standard curve from which unknown sample titre can be determined. Figure 5: Fixed concentration of antigen bound to the different antibody concen- trations from Figure 3 – A/California/7/2009 (H1N1). The antigen binding response for the three most concentrated antibody solutions overlap (Figure 5). The lack of distinction in response from these three antibody coating concentrations • Standard curves should have a minimum of eight data points, suggest biosensor binding saturation, which may imply that including a reference or zero concentration point for data sub- steric hindrance could play a role in antigen binding. As a result, traction, and should be run in replicates. The reference data the next lower antibody concentration should be selected for should be acquired using media similar to the vaccine analyte. the studies. • Check raw data for overlapping binding curves in both the antibody loading and antigen binding scouting steps. This will Tips for optimizing vaccine titer assays highlight data that cannot be differentiated at these levels. For • Select the appropriate biosensors for the study. For immo- antibody loading, choose the highest concentration of anti- bilization of non-purified samples such as serum antibodies, body that does not overlap (typically around 2 nm binding). Protein A or Protein G biosensors (ProA or ProG) are recom- For antigen binding, discount all standard levels that overlap. mended. When purified samples are available for capture, it Figure 5 shows examples of these overlaps in binding curves. is preferable to use High-Precision Streptavidin biosensors • In some cases, depending on size, particles such as viruses (SAX). In this case, the purified capture molecule should be will generate a negative signal, so the acquired data will need biotinylated prior to use. to be ‘flipped’ in the data analysis window prior to fitting. • Use the same batch/lot number of biosensors for both the • Read time should be set at 300 seconds for inactivated and standard curve and samples. recombinant hemagglutinin vaccine samples. Split Virion • Ensure that biosensors are hydrated in assay buffer for at sample read times should be ascertained from the raw data least 10 minutes. (typically < 30 seconds), and the appropriate window used to capture the linear region in the positive binding response. • Determine the shake speed for both the antibody loading and the antigen binding steps to obtain optimal conditions. Two • Antibody and vaccine samples give different results for each shake speeds, 400 RPM and 1000 RPM, are recommended strain, and so conditions such as shake speed, analysis time, for evaluation prior to the start of the assay. The same shaking antibody loading and linear range should be evaluated for speed should be used for both the standard curve and un- each strain. Typical conditions for the assay are shown in known concentration samples. Table 2. Antibody loading Vaccine binding Recombinant Recombinant Whole virus hemagglutinin Split virion Whole virus hemagglutinin Split virion Assay time (s) 300 300 300 300 300 30 Hold time (s) 600 600 600 N/A N/A N/A Shake speed (rpm) 400 Table 2: Typical conditions for vaccine titer assays. Hold time refers to temperature equilibration time with samples in the instrument prior to the start of the assay. 3 Binding (nm)

Results – inactivated virus The recombinant hemagglutinin H3/TX sample was diluted to 50 µg/mL using ForteBio Sample Diluent and was divided into The virus standard was tested from a range of 1 to 140 µg/mL two aliquots. One aliquot was treated at 95 °C for five minutes based on the HA concentration provided by NIBSC. The linear while the other was left at room temperature as a control. The range was established to be between 5 and 75 µg/mL with a samples were then analyzed using the using the method de- linear regression value of 0.9951, as shown in Figure 6. Once the scribed above with the H3/TX antibody bound to the biosen- linear range and antibody loading concentration are determined, sors. Figure 7 shows that the native sample gave an average the assay is ready to be used for sample titre determination. recovery concentration of 53.6 µg/mL, while the heat-treated sample showed a loss of response to an average recovery con- Stability indicating assay – heat denatured centration of 7.2 µg/mL. This proves the assay can successfully samples test heat-treated, stability-indicating samples and can distin- guish between native and deactivated antigen. A robust assay capable of distinguishing between native and deactivated antigen would also be suitable as a stability assess- ment assay. Such an assay can then be used to determine the A comparison between srid and Octet stability of the sample under accelerated degradation condi- analysis for split virion samples tions such as temperature, pH and oxidative conditions. SRID, the most often used technique for influenza virus titre, is a gel-based assay that is easy to use and relatively inexpensive to setup. It can, however, take as long as three days to run, and .045 can produce subjective data.3 Antigen diffusion into the aga- rose gel alone can take as long as 16 hours.4 R2=0.9951 Due to differences in sample processing, the two techniques .035 measure different molecular aspects of the sample. SRID measures the HA content after lysing with zwittergent 314, while the Octet system measures the diluted sample as is. The .025 Octet platform is a much quicker technique with easier sample processing, involving only a simple dilution in Sample Diluent. It also analyzes the vaccine sample itself and not a secondary .015 Standard sample that has been changed by the addition of a reagent. Fitting curve Precision and dynamic range produced by the Octet assay is also significantly better than what can be obtained from SRID. .005 0 20 40 60 80 100 Concentration (µg/mL) Platform Assay time for 96-samples Figure 6: Linear range of A/H1N1 California virus standard layout. SRID Several days Octet RED96 < 180 min .22 Octet RED384 < 90 min Octet HTX < 20 min .18 Table 3: SRID technique and Octet platforms assay time comparison .14 B/Massachusetts Octet assay (µg/mL) SRID (µg/mL) .10 Replicate Conc. High Low Conc. Conc. High Low Conc. Standard group avg %CV avg %CV .06 Fitting curve A 723.65 821 658 8.84 711 792 620 8.04 Unknown & control B 664.18 753 589 7.97 697 850 554 13.20 .02 C 737.58 796 686 5.21 697 802 538 10.04 0 0 20 40 60 80 100 120 Table 4: Comparison of SRID vs Octet assay data for split virion vaccine samples. Concentration (µg/ml) Figure 7: Heat denatured H3/TX recombinant hemaglutinin sample vs. control. Both samples are shown in red against the standard curve points shown in blue. 4 Binding (nm) Binding (nm)

Conclusion References Octet platform assays offer a number of advantages over the 1 Phillip D Minor, Assaying the potency of influenza vaccines, SRID assay for vaccine titre analysis: Vaccines 2015, 3, 90–104. • Higher quality data: The Octet assay has a wider dynamic 2 WHO Department of Immunization, Vaccines and Biologi- range (5–75 µg/mL), with greater precision and accuracy. cals. Temperature sensitivity of vaccines. World Health Orga- • Faster results with plate-based format: It uses a 96 or 384- nization, 2006 (http://www.who.int/vaccines-documents/ well plate format and can analyze a full plate in under three DocsPDF06/847. pdf). hours, including sample preparation time. 3 Aziza P. Manceur, Amine A. Kamen, Critical review of current • Simple sample prep: Samples only need to be diluted with no and emerging quantification methods for the development complex sample preparation. In-process and purified samples of influenza candidates, Vaccines, 2015, 33, 5913–5919. can be analyzed without encountering matrix effects. 4 Yingxia Wen et al., Conformationally selective biophysical • Direct vaccine measurement: the data collected is for the assay for influenza vaccine potency determination, Vaccines, vaccine itself and not a derived analyte produced from dena- 2015, 33, 5342-5349. turing the sample with reagents such as Zwittergent 314. • Use the same assay for all strains: A change in vaccine strains does not require a change in equipment or method. A limited requalification with the appropriate biosensor and strain specific antibody is sufficient. • Determine vaccine stability: The assay can analyze heat inac- tivated samples, hence can also be used as a vaccine stability indicating technique. ForteBio ForteBio Analytics (Shanghai) Co., Ltd. Molecular Devices (UK) Ltd. Molecular Devices (Germany) GmbH 47661 Fremont Boulevard No. 88 Shang Ke Road 660-665 Eskdale Bismarckring 39 Fremont, CA 94538 Zhangjiang Hi-tech Park Winnersh Triangle 88400 Biberach an der Riss 888.OCTET-75 or 650.322.1360 Shanghai, China 201210 Wokingham, Berkshire Germany www.fortebio.com fortebio.info@moldev.com salesops.china@moldev.com RG41 5TS, United Kingdom + 00800 665 32860 +44 118 944 8000 uk@moldev.com ©2019 Molecular Devices, LLC. All trademarks used herein are the property of Molecular Devices, LLC. Specifications subject to change without notice. Patents: www.moleculardevices.com/product patents. FOR RESEARCH USE ONLY. NOT FOR USE IN DIAGNOSTIC PROCEDURES. AN-4020 Rev C