Ethics

Animal experiments were carried out in compliance with the Public Health Service Policy on Humane Care and Use of Laboratory Animals47 and the Guide for the Care and Use of Laboratory Animals48, and were conducted with approved animal protocols from the Sanofi Institutional Animal Care and Use Committee. All animals were housed under specified pathogen-free conditions with food and water ad libitum, and acclimatized for 3 days (mice) or 7 days (ferrets) before entering the studies. Any ferret judged to be moribund, and where euthanasia was warranted as judged by a trained veterinarian (or undertaken as part of the study), were anesthetized and administered an overdose of euthanasia agent containing pentobarbital or other agents approved by American Veterinary Medical Association for euthanasia.

Influenza viruses

Reassortant H6 viruses used in the enzyme-linked lectin assay (ELLA) were generated by reverse genetics, with each reassortant expressing the targeted NA antigen, the HA from A/mallard/Sweden/81/2002 H6N1, and internal genes from A/Puerto Rico/8/1934 H1N1 [PR8]. Reassortant H1 viruses used in pre-infected ferret models were also generated by reverse genetics, with each reassortant expressing the targeted NA antigen, and the HA and internal genes from A/Puerto Rico/8/1934 H1N1 [PR8]. HA and NA segments including non-coding regions were generated by custom gene synthesis (Geneart AG), and PR8 segments were derived from a viral isolate as previously published49. All segments were cloned into a bi-directional transcription plasmid derived from pUC57 (Genscript) through the incorporation of polymerase (Pol) I and Pol II promoters, as described elsewhere49. Briefly, 293FT cells (Thermo Fisher Scientific) were transfected with a total of eight plasmids representing each influenza virus segment using Lipofectamine 2000 CD (Thermo Fisher Scientific). After 24 h, Madin-Darby canine kidney Cells (MDCK-) adult T-cell leukemia (ATL) cells (ATCC) were added to the transfected cells in the presence of tosyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin (Sigma) to allow influenza virus propagation. Cell culture supernatants containing influenza virus were harvested 7 days post-MDCK addition and passaged in 8–10-day-old embryonated chicken eggs (Charles River Laboratories, Inc.). Inoculated eggs were incubated at 37 °C for 48 h, then cooled to 4 °C for 12 h, harvested, and clarified by low-speed centrifugation (3000 rpm, 20 min). Virus titers were determined by plaque assay on MDCK cells.

Egg-grown stocks of A/Michigan/45/2015 (H1N1), A/Singapore/INFIMH-16-0019/2016 (H3N2), B/Colorado/06/2017 (B/Victoria/2/87-like lineage) and B/Phuket/3073/2013 (B/Yamagata/16/88-like lineage) included in HAI testing were provided by Sanofi Global Clinical Immunology (Swiftwater, PA). Wild-type influenza A/Perth/16/2009 (H3N2) used in the ferret challenge study was provided by IIT Research Institute (Chicago, IL). All viruses were stored at <–65 oC until use.

Vaccine antigens

Constructs were designed for the expression of recombinant, soluble influenza NA. Both tetrameric and monomeric NA construct design included an N-terminal CD5 secretion signal peptide, 6HIS tag and the globular neuraminidase head domain similar to other rNA constructs described elsewhere29. The tetrameric design also contains a tetrabrachion domain between the HIS tag and the globular head for multimerization. Using a defined amino acid sequence, a codon optimized synthetic gene was assembled from oligonucleotides and/or PCR products and the fragment was inserted into pcDNA3.4-TOPO (ThermoFisher). The plasmid DNA was purified from transformed bacteria and scaled to achieve appropriate concentration for transfection. Protein expression was performed in CHO-S cells using the ExpiCHO™ Expression System Max Titer Protocol (ThermoFisher). A clarification step was performed to separate secreted proteins from cells. NA protein was purified from host-cell proteins by affinity (HisTrap HP Column – GE Healthcare) followed by anion exchange chromatography (HiTrap Q HP – GE HealthCare), dialysis into 10 mM phosphate buffered saline (pH 7.2) and sterile filtration through a 0.2 μm filter. The NA vaccine preparations were produced in compliance with the current good research practices (cGRP).

Recombinant HA proteins were obtained from Protein Sciences. Briefly, purified HA proteins were produced by baculovirus expression in a continuous insect cell line (EXPRESSF + ®) derived from Sf9 cells and grown in serum-free medium. IIV was prepared from influenza virus propagated in embryonated chicken eggs, which was subsequently inactivated using formaldehyde, then concentrated, and purified by zonal centrifugation on a sucrose gradient, split with Triton® X-100, before repeat purification and resuspension in isotonic sodium phosphate-buffered sodium chloride solution. Preparations were sterile filtered using a 0.2 μm syringe filter. Live influenza virus-derived neuraminidase (LVNA) was isolated from influenza virus propagated in embryonated chicken eggs. Virus was purified by sucrose gradient ultracentrifugation and NA was extracted by detergent solubilization, further purified by column chromatography, and suspended in sodium phosphate-buffered isotonic sodium chloride solution. Preparations were sterile filtered using a 0.2 µm syringe filter.

Size exclusion chromatography with multiangle light scattering (SEC-MALS): assessment of molecular weight and purity of secreted/soluble NA

Size exclusion chromatography was carried out on a Waters ACQUITY Arc Bio HPLC system with a TSK-GEL G4000 PWXL (7.8 mm × 30 cm) column (Tosoh Bioscience) using phosphate buffer saline containing 0.02% sodium azide (pH 7.4) as mobile phase. A flow rate of 0.5 mL/min was used. The effluent was detected by a UV detector at 280 nm, followed by a Wyatt DAWN MALS detector (HELEOS II) and an Optilab TrEX differential refractive index (RI) detector. Empower® (Waters Corporation, Milford, MA) software was used for high performance liquid chromatography (HPLC) control and data analysis on the UV chromatogram. The purity of the sample was calculated based on the percentage of the specific peak area/the total peak areas. ASTRA software was used for MALS data collection and protein molecular weight determination using light scattering signals with a concentration detector (RI or UV).

2′-(4-methylumbelliferyl)-α-D-N-acetylneuraminic acid (MUNANA)-based assay: assessment of NA activity

The MUNANA-based assay was based on a previously described method with modifications50. Briefly, two-fold serial rTET-NA dilutions were prepared in 96-well plates using buffer (33.3 mM 2-[N-morpholino] ethanesulfonic acid [MES, pH 6.5], 4 mM CaCl2, 50 mM BSA) and mixed with MUNANA substrate (100 µM) and incubated for 1 h at 37 °C with shaking. The reaction was stopped by addition of alkaline solution (0.2 M Na2CO3). The fluorescence intensity (RFU, relative fluorescence unit) from the rTET-NA and MUNANA substrate mixture was measured using excitation and emission wavelengths of 355 nm and 460 nm, respectively. A standard curve was generated using 4-methylumbelliferone (4-MU) diluted in enzyme buffer at various concentrations; rTET-NA enzymatic activity was determined against a 4-MU reference with the results expressed in µM/60 min for total NA activity and nmol/min/µg for specific NA activity.

Oseltamivir-binding assay: detection of tetrameric NA

The oseltamivir phosphate (Tamiflu)-NA binding assay was performed on Octet-Red96 instrument (ForteBio, Sartorius), an Octet BLI Detection System using the bio-layer interferometry (BLI) technique that facilitates real-time label-free analysis for the determination of kinetics, affinity, and quantitation regarding the NA bound to the streptavidin biosensor tip coated with oseltamivir-phosphate–biotin conjugate. In brief, oseltamivir-phosphate–biotin conjugate (5–10 μg/ml) in 1xKB buffer (containing PBS pH 7.4; 0.02% Tween-20; 0.1% albumin; and 0.05% sodium azide) is first captured on a streptavidin-coated biosensor. The interaction between NA and oseltamivir-phosphate was then initiated by dipping oseltamivir-phosphate bound biosensors into sample wells containing a 2-fold dilution series of recombinant NA (0.16–40 μg/ml in 1xKB buffer). The binding between oseltamivir-phosphate and NA produces a measured shift in the interference pattern via the detector. The level of binding response is proportional to the concentration of NA. An Octet data acquisition software was used for instrument control and data collection (ForteBio, Sartorius), and an Octet Data Analysis software was used for data process (ForteBio, Sartorius).

Mouse studies

Female BALB/c mice (8 per group; sample size chosen to ensure 85% power of detecting a difference between groups) aged 6–8 weeks and with a body weight of 15–20 g (Charles River) were vaccinated twice 21 days apart with the same dose (0.2 or 1 μg) of various NA-containing preparations (50 µL/dose, intramuscularly) in the presence or absence of AF03 adjuvant containing 5% squalene (Sanofi). The mice were not anesthetized before any in-life study procedure. Terminal bleed (via cardiac puncture) was undertaken after isoflurane overdose exposure followed by cervical dislocation two weeks after booster vaccination; sera pools from two animals were created (stored at –20 °C until required), resulting in a total of 4 samples per group. The sera were tested by ELLA to assess NAI activity or via ELISA to derive NA-binding or tetrabrachion-binding antibodies.

Ferret studiesImmunization studies

Outbred male and female Fitch ferrets (Marshalls Farms North Rose, NY), aged 17–21 weeks old, with a bodyweight of at least 600 g, and seronegative (by HAI antibody assessment) to the four seasonal influenza vaccine strains, were used for vaccination studies. The ferrets were randomized into study groups (6 per group) using a body weight stratification procedure that produced groups with similar mean body weights. For standard immunogenicity assessments, naïve ferrets were vaccinated twice 21 days apart with the same dose (5 µg or 45 µg) of various NA-containing preparations (500 µL/dose, intramuscularly) with or without AF03 adjuvant. Animals used to assess impact of pre-existing influenza immunity on NA immunogenicity were initially primed by intranasal challenge with reassortant influenza virus A/Perth/16/2009 H1N2 or A/Kansas/14/2017 H1N2 on day 0 (1000 µL/dose, split evenly between nostrils). Three weeks after initial infection, animals received a single dose (1.8 to 45 µg) of A/Perth/16/2009 rTET-NA or A/Kansas/14/2017 rTET-NA (500 µL/dose, intramuscularly) with or without AF03 adjuvant.

The animals used to assess multivalent vaccine immunogenicity were vaccinated twice 21 days apart with a mixture of four rTET-NA antigens, each antigen comprising the NA head domain from one of the strains included in the quadrivalent 2018-19 seasonal influenza vaccine (A/Singapore/INFIMH-16-0019/2016 (N2), A/Michigan/45/2015 (N1), B/Colorado/06/2017 (B/Victoria/2/87-like lineage), and B/Phuket/3073/2013 (B/Yamagata/16/88-like lineage)). All antigens were administered at a 1:1 ratio, without adjuvant (45 µg/antigen dose) or adjuvanted with AF03 (5 µg/antigen and 45 µg/antigen doses).

All ferrets were bled under sedation at baseline and three weeks after primer (one day before or just before booster) and booster vaccination, and two weeks after challenge as required. For all in-life procedures, the ferrets were anesthetized via intramuscular (IM) injection with a ketamine HCl (25 mg/kg) and dexmedetomidine (0.001 mg/kg) solution. For euthanasia, ferrets were sedated with the same anesthetic cocktail as above, blood was collected, and then animals were administered an overdose of euthanasia agent containing pentobarbital (e.g., Beuthanasia-D) or other American Veterinary Medical Association approved method of euthanasia. Sera samples (stored at –20oC until required) were tested by ELLA to assess NAI activity. Additionally, the HAI assay and antibody forensics were undertaken to assess antibody responses to hemagglutinin antigens following multivalent vaccination.

Challenge studies

Outbred male Fitch ferrets aged 17–21 weeks old, with a bodyweight of at least 600 g, and seronegative (by HAI antibody assessment) to the four seasonal influenza vaccine strains (Triple F farms, Sayre, PA), were randomized into study groups (16 per group) using a body weight stratification procedure that produced groups with similar mean body weights. For all in-life procedures, the ferrets were anesthetized via IM injection with a ketamine HCl (25 mg/kg) and xylazine (2 mg/kg) mixture. Ferrets were initially vaccinated twice 21 days apart with the same dose (0.2 µg to 45 µg) of A/Perth/16/2009 N2 rTET-NA (500 µL/dose, intramuscularly) with or without AF03 adjuvant. Three weeks after booster vaccination, the ferrets received intranasal challenge with 107 PFU of A/Perth/16/2009 H3N2 wild-type influenza A virus (1000 µL/dose, split evenly between nostrils). Prior to intranasal administration, ferrets were anesthetized with a ketamine HCl (25 mg/kg) and xylazine (2 mg/kg) mixture. The ferrets were then held upright and 0.5 mL of the inoculum was administered dropwise into each nostril (for a total of 1.0 mL/ferret). The animals were monitored for 14 days post-challenge for clinical symptoms and changes in body weight once daily, and body temperature twice daily. Nasal washes were collected from all challenged animals on days 1, 3, 5 and 7 post-challenge and samples were stored at ≤ –65 °C for viral shedding assessment. Selected ferrets (1–2) from each group were euthanized with an intravenous dose of Beuthanasia-D (150 mg/kg) on days 1, 3, 6 and 14 post-challenge and necropsied. Lungs and nasal turbinates from necropsied ferrets were collected for viral titer analyses.

Viral titration of nasal wash and respiratory tissues

Nasal wash specimens were collected from experimentally infected ferrets on alternate days following intranasal challenge. Briefly, ferrets were initially anesthetized followed by 0.5 mL intranasal instillation into each nostril with sterile PBS containing gentamicin (50 μg/mL), penicillin (100 U/mL), and streptomycin (100 μg/mL), and collection of the nasal wash, which was then stored at ≤ –65 °C until assessment. Virus in the nasal wash specimens was titrated by standard 50% tissue culture infectious dose (TCID50) assay as follows. The nasal washes were thawed and then clarified by centrifugation. The resulting supernatant was 10-fold serially diluted and transferred to a 96-well plate for titration on a monolayer of MDCK cells. Sections of lungs (right and left cranial, and right left caudal lung lobes) and nasal turbinates harvested for viral titer assessment were weighed and flash frozen in an ethanol/dry ice bath or liquid nitrogen and stored at ≤ –65 °C until processing for virus titration by standard TCID50 assay as described.

Enzyme-linked lectin assay (ELLA): assessment of NAI responses

NAI antibody responses were measured against H6 reassortant viruses containing NA derived from strains of interest by ELLA as previously described51. Briefly, a H6 reassortant virus containing the NA derived from a strain of interest was titrated in fetuin-coated 96-well plates to determine the standard amount of virus that provides 70% of maximum NA enzymatic activity. Titration of NAI antibodies present in the sera was achieved by performing two-fold serial dilutions of heat inactivated sera. A total of 50 μL of diluted sera was then added to 50 μL of diluted virus corresponding to 70% of maximum NA enzymatic activity in a fetuin-coated plate. The serum-virus mixture was incubated at 37 oC overnight. The plate was washed four times, incubated with horseradish peroxidase- (HRP-) conjugated peanut agglutinin (Sigma) and washed again prior to developing by addition of o-phenylenediamine dihydrochlorid. Low or no signal relative to a virus control indicates inhibition of NA activity due to the presence of NA-specific antibodies. NAI titers were approximated with a non-linear four parameter logistic (4PL) curve using GraphPad Prism software and the 50% maximal inhibitory concentration (IC50) calculated.

Enzyme-linked immunosorbent assay (ELISA): assessment of NA-binding and tetrabrachion-binding titers

The rTET-NA derived from influenza virus strains A or B were immobilized on the surface of nickel coated 96-well microtiter plates (Pierce) by adding 50 μL of rTET-NA at working dilution of 0.5 μg/mL and coating overnight at 4oC. Coated plates were washed four times to remove any unbound protein and incubated with 5% w/v Blot-Quick Blocker (Biosciences) (300 μL per well) at room temperature for two hours to saturate the non-specific binding sites. After blocking, the plates were incubated with 50 μL of three-fold serially diluted individual serum samples at room temperature for two hours to allow binding of anti-NA antibody to immobilized rTET-NA protein. The plates were washed four times to remove any unbound antibody and incubated with 50 μL of HRP-conjugated anti-ferret or anti-mouse detection antibody (Abcam) and developed through incubation with 50 μL of the HRP enzyme substrate mix containing 3,3’,5,5’-tetramethylbenzidine and hydrogen peroxide. The reactions were stopped by addition of 50 μL of 0.16 M sulfuric acid (Thermofisher) and the absorbance at 450 nm read immediately. Titers were determined based on the serum dilution that achieves 50% binding (EC50) using a 4-parameter non-linear regression analysis from GraphPad Prism software.

For the assessment of antibodies that specifically bind to the tetrabrachion domain present in the r-TET-NA vaccines included in this study, recombinant protein containing a fragment of human serum albumin (domain III) fused to the tetrabrachion, and his-tagged at its N-terminus were immobilized on the surface of nickel coated 96-well microtiter plates and the assay undertaken in the same manner as described above.

Hemagglutination-inhibition (HAI) assay

Sera were treated with receptor-destroying enzyme (RDE; Denka Seiken, Co., Japan) to inactivate nonspecific inhibitors prior to the HAI assay. Briefly, three parts RDE were added to one part serum and incubated at 37 °C overnight. RDE was subsequently inactivated by incubation with 6 times excess serum volume at 56 °C for 30 min, followed by 2-fold serially dilution in 0.9% saline in 96-well v-bottom microtiter plates. Equal volumes of each virus from the panel, adjusted to 4 hemagglutinating units per 250 μL, were added to each well. For the current study, homologous virus panel included A/Michigan/45/2015 (H1N1), A/Singapore/INFIMH-16-0019/2016 (H3N2), B/Colorado/06/2017 (B/Victoria/2/87-like lineage) and B/Phuket/3073/2013 (B/Yamagata/16/88-like lineage) viruses grown in eggs. The plates were covered, incubated for 20 min at room temperature, and 1% turkey red blood cells (Lampire Biologicals) in PBS subsequently added. The plates were agitated to mix the red blood cells, covered and allowed to settle at room temperature for 1 h. The reciprocal dilution attained for the last well containing non-agglutinated red blood cells was taken as the HAI titer.

Antibody forensics assay

Antibody forensics methods were used to measure strain-specific rHA antibodies in ferret sera using magnetic bead array (MagPlex® Microspheres) with fluorescent dyes (as described in Ustyugova et al.)52. The strength of antibody binding to strain-specific rHA was presented in normalized mean fluorescent intensity units, calculated from raw fluorescent intensity signal multiplied by the serum dilution. The rHAs coupled to the magnetic beads were selected based on antigenicity data published in the annual and interim reports on the composition of influenza vaccines52. In addition to 2018–2019 Northern hemisphere recommended strains, the rH3 panel included strains for 2013 through 2016 seasons, while the H1 panel encompassed strains from 2009 through 2016 seasons. A complete list of rHAs included in this study is provided in the supplementary materials. Individual ferret sera were analyzed and the resultant antibody forensics data for 40 H3 and 18 H1 strains was evaluated.

Statistical analysis

To evaluate the immunogenicity of the rTET-NA vaccines, a multiple comparison analysis was performed using Tukey’s test to compare vaccine groups in terms of NAI titer production and to evaluate whether there was an adjuvant impact or a dose effect.

In the ferret challenge study, mean percent change in body weight graphs were created to show the pattern of the body weight loss from D44 to D57 per vaccine group in ferrets that were monitored up to D57 (n = 6 per group). The AUC of the body weight loss graph as well as the peak temperature rise were calculated, first individually, then per vaccine group. Bar charts were created to visualize these metrics per vaccine group. The AUC of the virus shedding was estimated per vaccine group using the trapezoidal rule.

To assess the association between anti-NA level and disease severity, a severity score was established based on a percentile approach. The individual severity score for each protection endpoint was defined by using the percentiles (25th, median and 75th) presented in Supplemental Table S1 (disease severity scoring). A combined severity score was then calculated per animal as the sum of the severity score of peak body weight loss, peak body temperature rise and the AUC of virus shedding. The combined severity score was calculated only for ferrets that had complete data for all three protection endpoints. Ferrets that had a combined symptom severity score above the 75th percentile of the distribution of the severity scores were considered as having severe disease. The comparison of NAI titers at D42 between groups was conducted using one-way ANOVA. The comparison was performed using log-transformed NAI titers to meet the normality assumption required for the use of the statistical model.

A binary outcome (severe or non-severe disease) logistic model was performed to evaluate the prediction of severity with NAI titer. In this model, the disease severity was defined as a dependent variable and the log-transformed NAI titer was included in the model as a predicted factor. The subsequent ROC curves generated were used to define cutoffs based on the maximum Youden’s index.

The AUC of the ROC curves were also estimated and used to evaluate the ability of the NAI titer to properly distinguish ferrets with severe disease from those with non-severe disease. An AUC of 0.5 means that the NAI titer is uninformative for the prediction of influenza severity. The closer the AUC is to 1, the better the discrimination power of the NAI titer.

All statistical tests were two-sided, and the nominal level of statistical significance was set to α = 0.05 for effect size estimates. Statistical analyses were conducted using SEG SAS v9.4® (WISE environment) and R Statistical Software (R version 3.5.1 on RStudio®).