Human Embryonic Kidney (HEK) 293T/17 cells were used for PV production, titration and as target cells. They were maintained in Dulbecco’s Modified Eagle Medium (DMEM) (PAN−Biotech P04-04510) supplemented with 15% (v/v) Fetal Bovine Serum (FBS) (PAN−Biotech P30-8500) and 1% (v/v) penicillin−streptomycin (Pen-Strep) (Sigma P4333).

Madin−Darby Canine Kidney (MDCK) and A549 cells were also used as target cells to study IBV PV transduction and tropism. Cells were cultured in DMEM supplemented with 5% FBS and 1% of Pen-Strep and in DMEM/F-12 media (Gibco 21041033) with 10% FBS and 1% Pen-Strep, respectively.

B/Bangladesh/3333/2007 and B/Brisbane/60/2008 HA genes were amplified from cDNA and subcloned into plasmid expression vector pI.18 (46, 47). B/Hong Kong/8/1973, B/Victoria/2/1987, B/Yamagata/16/1988 and B/Florida/4/2006 codon optimized HA genes ( Figure 1 ) were synthesized by GenScript and cloned into phCMV1 as previously described (32). B/Colorado/06/2017, B/Phuket/3073/2013, B/Brisbane/60/2008 and B/Washington/2/2019 plasmids were synthesized by GeneArt and cloned into pEVAC. HAT (human airway trypsin-like protease) and TMPRSS2 (type II transmembrane protease serine 2) (49) were cloned in pCAGGS; while TMPRSS4 (type II transmembrane protease serine 4) (50) was cloned in pCMV, and were used to activate influenza B HA during lentiviral vector production. The second-generation packaging construct pCMVΔR8.91 (51), and the self-inactivating lentiviral vectors pCSFLW expressing firefly luciferase and pCSemGW expressing emerald green fluorescence protein (emGFP) were used as reporters for PV production.

For IBV HA in pI.18, pseudotypes were produced in 6-well plates using branched polyethylenimine (PEI) (Sigma-Aldrich 408727) co-transfection of 500 ng pCMVΔR8.91, 750 ng pCSFLW or pCSemGW, and 500 ng of HA-encoding plasmid. Additionally, 125 ng or 250 ng of protease-encoding plasmid was also included in the DNA transfection mix to allow influenza B HA activation and cleavage. For IBV HA in pEVAC, cells were transfected using Opti-MEM™ (Thermo Fisher Scientific 31985062) and FuGENE^® HD Transfection Reagent (ProMega E2312) with 10 ng HA encoding plasmid, 375 ng pCSFLW, 250 ng pCMVΔR8.91, and 2.5-10 ng of HAT or T4 proteases. IBV PV were also produced in the absence of protease-encoding plasmid (Δ protease), and these served as controls. At least 8 hours post-transfection, 1 unit of exogenous Neuraminidase (Sigma N2876) was added. PV were collected 48 hours after transfection of HEK293T/17, passed through a 0.45 µm membrane filter, and stored at -80°C.

Titration experiments of firefly luciferase-expressing or emerald GFP PV were performed in Nunc™ F96 MicroWell™ white or transparent polystyrene plates (Thermo Fisher Scientific), respectively. The PV production titer was evaluated by transduction of HEK293T/17, MDCK, and A549 cells, with 2−fold serial dilutions of PV in a total volume of 100 µL. As a control for HA cleavage, Δ protease influenza B PV were also activated by addition of 1 mg/mL (TPCK)-trypsin (Sigma T1426) to a final concentration of 100 µg/mL per well. These wells were incubated at 37°C, 5% CO₂ for 30 minutes followed by addition of 50 µL trypsin neutralizing solution (Lonza CC-5002) to each well. For all samples and controls, 1.5x10⁴ HEK293T/17 cells were added to each well of the 96-well plate. Plates were incubated for 48 hours at 37°C, 5% CO₂. Afterwards, firefly luciferase gene expression was evaluated and quantified using the Bright−Glo™ assay system (Promega E2620) and GloMax Multi detection system luminometer (Promega) as shown previously (52). Influenza B PV produced with emerald GFP were also tested, using an equivalent protocol, for their ability to transduce HEK293T/17, MDCK, and A549 cells.

Antisera against B/Brisbane/60/2008 (NIBSC 11/136) was obtained from the National Institute for Biological Standards and Control (NIBSC) and used as the primary antibody in western blot experiments and positive control in neutralization assays. Post-vaccination mouse sera were obtained from the University of Cambridge as part of an ongoing influenza vaccination study and were employed to determine responses of different antigens used for vaccination against IBV PV. Human sera obtained during the clinical trial NCT00942071 (53) were used in PV neutralization assays (pMN). These sera were collected from adults aged 50 years and over pre− and post− administration of an influenza split virion vaccine (Sanofi Pasteur MSD, France) that contained HAs of A/California/7/2009 (H1N1), A/Perth/16/2009 (H3N2), and B/Brisbane/60/2008. Nine subjects received an injection of Modified Vaccinia Ankara (MVA) vectors expressing the A/Panama/2007/1999 (H3N2) nucleoprotein (NP) and matrix protein 1 (M1) (MVA−NP+M1) antigens as a single fusion protein (53, 54). The remaining subjects (n=8) received a trivalent inactivated vaccine (TIV) and saline placebo. These sera were previously analyzed (53) in an HI assay against B/Brisbane/60/2008. Seal serum samples (n=8) were obtained from the University of Veterinary Medicine, Vienna, Austria, as part of a wildlife and conservation study of seals caught in the Caspian Sea (CS) from 2006-2010 at the Europe/Asia border. We obtained previously characterized neutralizing monoclonal antibodies (55) from the Division of Viral Products, Food and Drug Administration, USA. Monoclonal antibodies designated 5A1 specific to the B/Victoria lineage, 3E8, specific to the B/Yamagata lineage, and 2F11 specific to both B/Victoria and B/Yamagata lineages are used herein to evaluate PV neutralization activity (55).

Briefly, 2 ml of PV was centrifuged at 3000xg at 4°C for 24 hours. Supernatant was removed and resuspended in 150 µL OptiMEM™. SDS−PAGE of reduced and denatured influenza B PV was carried out in 4−15% Mini−PROTEAN^® TGX™ precast polyacrylamide gels (Bio−Rad 4561024) following manufacturer’s instructions. The gel was transferred onto an Immuno−Blot^® low fluorescence polyvinylidene fluoride membrane (PVDF) (Bio−Rad 1620260) for 60 minutes at 100 V. The membrane was blocked overnight with blocking buffer (10% (w/v) dried milk, 0.01% (v/v) Tween20−PBS). The anti−B/Brisbane/60/2008 serum (NIBSC 11/136) was used as the primary antibody, diluted 1:500 in blocking buffer. The membrane was washed with 0.01% (v/v) Tween20−PBS prior to addition of secondary antibody, IgG Dylight^®800 antibody diluted 1:20000 in blocking buffer. After washing to remove unbound secondary antibody, the PVDF membrane was visualized using the Odyssey^® Sa Infrared Imaging System (LI−COR Bioscience) at 800 nm.

The neutralization activity of anti−B/Brisbane/60/2008 serum (NIBSC 11/136) (starting dilution 1:1000), post-vaccination mouse sera (University of Cambridge), sera from the NCT00942071 (53) human clinical trial, and mAbs from Food and Drug Administration (USA) were evaluated against different influenza B pseudotypes produced in this study by performing the pMN according to established protocol (52). Briefly, 2−fold serial dilutions of serum or mAbs were performed in a Nunc™ F96 MicroWell™ white polystyrene plate. The plate was centrifuged for 1 min at 500xg and 50 µL of influenza B PV solution with a concentration 1×10⁶ relative luminescence units (RLU) was added to each well. The plate was again centrifuged for 1 min at 500xg and incubated at 37°C, 5% CO₂. Afterwards 50 µL of 1.5×10⁴ HEK293T/17 cells were added to each well and after a final centrifugation for 1 minute at 500xg, the plate was incubated at 37°C, 5% CO₂ for 48 hours. Firefly luciferase gene expression was evaluated as previously described. Additionally, the neutralization activity of a panel of seal sera was also evaluated using a starting dilution of 1:100 ( Supplementary Table 1 ). Human positive controls were obtained from Prof. Emanuele Montomoli at the University of Siena (starting dilution 1:1000). The negative control used was FBS (starting dilution 1:100).

All analyses were performed with GraphPad Prism 8.12 for Windows (GraphPad Software). A non−parametric Analysis of variance (ANOVA) was carried out to determine whether differences exist among the means of three or more groups when compared to control. If differences were found, the post-hoc Tukey multiple comparison test was performed to identify if there is a significant difference between pairs of experimental groups. Wilcoxon matched−pairs signed rank test was used to assess statistical significance between day 0 and day 21 in the clinical trial as the data did not follow Gaussian distributions. Clinical trial data were also then stratified based on vaccine regimen: TIV + placebo or TIV + MVA−NP+M1, to evaluate the effect of MVA−NP+M1 in broadening the heterosubtypic antibody response. To avoid multiple comparisons, statistical analysis was performed using the IC₅₀ fold−increase between TIV + placebo and the TIV + MVA−NP+M1 groups in a Mann−Whitney U test.

Influenza B HA gene sequences that were used in PV production were used to calculate percentage identity between amino acid sequences, and for phylogenetic analysis. First, HA−encoding sequences were downloaded from the Influenza Virus Resource database (IVRD). Then, percentage identity between amino acid sequences was calculated by pair−wise alignment using Jalview (56).

For phylogenetic analysis, codon−based alignment was performed using the MUSCLE algorithm (57) in Molecular Evolutionary Genetics Analysis (MEGA) (58). The phylogenetic “Quick Tree” model used PhyML (48) and pre-defined settings by the Influenza Resource Database of the National Institute of Allergy and Infectious Diseases (NIH/DHHS) to infer phylogenies based on sequences (https://www.fludb.org/brc/tree.spg). The tree generated was then graphically elaborated with Archaeopteryx.js (https://sites.google.com/site/cmzmasek/home/software/archaeopteryx-js) in phyloXML and format.

Proteases were previously reported to be necessary for production of high titer influenza A pseudotypes (50, 59). Here we have employed three different proteases that are associated with influenza A HA cleavage: HAT, TMPRSS2, and TMPRSS4, to show how they will affect IBV PV production. We have used fixed amounts of HA-encoding plasmid, packaging construct, and lentiviral reporter plasmids, only varying the type and amount of protease-encoding plasmid in the transfection. For IBV HA in pI.18, the highest titers of around 10⁹ RLU/mL were achieved using HAT as protease, except for B/Brisbane/60/08 which achieved its highest titer of ~10⁸ RLU/mL with TMPRSS4 ( Figure 2A ). For all PV, co-transfection with TMPRSS2 achieved the lowest titers, ~10⁵-10⁸ RLU/mL, a log down from titers produced via HAT ( Figure 2A ). In the absence of protease (Δprotease), influenza PV are still produced, but titers only reached ~10⁵ RLU/mL at best, at least 3 logs lower than PV produced in the presence of protease. These PV (Δprotease) were added with TPCK-Trypsin post-production and prior to titration to activate HA cleavage. PV titers increased at least 3-logs compared to Δprotease titers ( Figure 2A ). We co-transfected with HAT and TMPRSS4 proteases to produce PV from IBV HA in pEVAC. For transfections with the IBV HA pEVAC plasmids, we used significantly less plasmid DNA, 10 ng of the IBV HA plasmid, 250 ng of the packaging plasmid, 375 ng of the luciferase reporter plasmid, and 2.5-10 ng of protease-encoding plasmid. We obtained titers of ~10¹⁰ RLU/mL for all PV produced via co-transfection with HAT, and 10⁹-10¹⁰ RLU/mL with TMPRSS4 co-transfection ( Figure 2B ). The difference in titers observed for B/Brisbane/60/2008 produced may be attributed to the employment of a different plasmid backbone (pI.18 vs pEVAC), and the transfection agents used. Results indicate that transduction titers of different influenza B PV are dependent on the type of protease used and on the quantity of protease-encoding plasmid employed in transfection, with amounts being adjusted to produce optimal titers.

The HA of both IAV and IBV are synthesized as an inactive precursor HA0, and its proteolytic cleavage into the mature HA1/HA2 form is a prerequisite for the low pH-induced, irreversible HA conformational change that is essential for virus infectivity (10, 14, 49, 60, 61). To confirm that the addition of an extracellular protease during transfection mediates HA cleavage, a western blot was performed on the B/Brisbane/60/2008 (from pI.18) PV produced. In all B/Brisbane/60/2008 PV produced with the addition of HAT, TMPRSS2 and TMPRSS4, bands corresponding to HA1 (~55kDa) can be observed ( Supplementary Figure 1 ). Darker and heavier HA1 bands are observed for PV produced with HAT and TMPRSS4 compared to TMPRSS2. These band intensities correlate with PV titers ( Figure 2A ). Interestingly, PV HA activated by HAT, TMPRSS2 and TMPRSS4 present heterogeneous glycosylation characteristics: more than one band between 50 kDa and 60 kDa, all corresponding to HA1, are present, whereas in the Δprotease PV before and after TPCK−trypsin treatment, only a single band at ~55 kDa is observed ( Supplementary Figure 1 ). This band at ~55 kDa shows limited HA0 cleavage and probable HA activation even without proteases ( Supplementary Figure 1 ). It is also observed that the HA cleavage mediated by proteases can be reproduced through TPCK−trypsin treatment of the Δ protease PV, as the HA0 band was replaced by bands corresponding to the HA cleavage products ( Supplementary Figure 1 ).

We tested whether IBV pseudotypes can transduce HEK293T/17, MDCK, and A549 cells at high titer. Employing transfection with pCSFLW (firefly luciferase) ( Figure 3A ) and emGFP ( Figure 3B ), it was found that all IBV pseudotypes are able to transduce all cell lines. It was shown that the IBV PV transduces HEK293T/17 cells more effectively than the classical influenza producing cell lines, MDCK and A549. Titers in HEK293T/17 cells are in the range of 5x10⁷-10¹⁰ RLU/mL ( Figure 3A ) whilst titers were in the range of 10⁴-10⁸ RLU/mL for pseudotypes using both MDCK and A549 as target cells ( Figure 3A ). Subsequently, we utilized HEK293T/17 cells for production, titration and neutralization assays involving IBV PV in this study.

We have investigated the utility of influenza B pseudotypes as substitutes for live viruses in influenza neutralization serological studies. We employed standard reference antisera, mouse sera from vaccination studies, human sera from an influenza clinical trial, and sera from seals, a known natural host of influenza B, to test the IBV PV we produced in this study. First, we tested if three of the six IBV PV from pI.18 we have generated ( Figure 2A ) and one IBV PV from pEVAC ( Figure 2B ) could be neutralized by anti-B/Brisbane/60/2008 HA serum (NIBSC 11/36). We chose to test B/Brisbane/60/2008 from the B/Victoria lineage, B/Florida/4/2006 from the B/Yamagata lineage, representing strains of co-circulating influenza B lineages previously or currently used in influenza seasonal vaccination, and B/Hong Kong/8/1973, as a representative strain circulating before the influenza B lineage division ( Figure 1 ). It was found that the reference serum employed completely neutralized all the IBV PV tested up to the dilution 1:1600 ( Figure 4A ). The anti-serum neutralized the homologous B/Brisbane/60/2008 most efficiently, followed by B/Hong Kong/8/1973, which shares 94.9% amino acid identity with B/Brisbane/60/2008, and B/Florida/4/2006 (93.5% amino acid identity) ( Figures 1 , 4A ). There was no difference in neutralization of B/Brisbane/60/2008 PV produced from either pI.18 or pEVAC when used in the same input RLU by anti-B/Brisbane/60/2008 HA serum (NIBSC 11/36) ( Figure 4A ). As such, the PV neutralization assay (pMN) with reference antisera was able to differentiate between the different PV, showing correlation between amino acid identity and neutralization efficacy.

Pre-clinical utility: We have obtained sera from Influenza B vaccination studies in mice at the University of Cambridge, and we have tested the ability of these mouse sera to neutralize the IBV pseudotypes, B/Phuket/3073/2013 from the B/Yamagata lineage and B/Washington/2/2019 from the B/Victoria lineage, that we produced using pEVAC-HA ( Figure 2B ). Naïve mice were vaccinated with the HA of B/Singapore/222/1972 (pre-split IBV), B/Colorado/06/2017 (B/Victoria lineage), B/Phuket/3073/2013 (B/Yamagata lineage), and PBS (negative control) to elicit an immune response against influenza B. Mouse terminal bleeds were taken, and sera were run on pMN assays to obtain IC₅₀ data and assess vaccine immunogenicity. Post-vaccination sera from groups given B/Colorado/06/2017 and B/Phuket/3073/2013 were able to neutralize both IBV pseudotypes tested regardless of lineage ( Figure 4B ) compared to the PBS control (*p<0.05). Vaccination with B/Colorado/06/2017 (B/Victoria) showed highest neutralization against the B/Victoria lineage PV, B/Washington/2/2019. Similarly, vaccination with B/Phuket/3073/2013 (B/Yamagata) neutralized the B/Phuket/3073/2013 PV better than the B/Victoria lineage PV ( Figure 4B ). Mice vaccinated with pre-split B/Singapore/222/1972 was unable to neutralize both IBV with IC₅₀ dilutions showing no significant difference (p>0.05) to the mice vaccinated with PBS ( Figure 4B ). IC₅₀ dilutions were in the range of 10² – 10⁵ for mice vaccinated with either B/Colorado/06/2017 or B/Phuket/3073/2013 against both IBV PV ( Figure 4B ). indicating development of an immune response post-vaccination that can be effectively assessed via the PV neutralization assay.

Clinical utility: To further investigate the suitability of influenza B pMN assays in vaccine immunogenicity studies and to detect cross−reactive responses to influenza B, we employed human sera from clinical trial NCT00942071 (53). The panel of human sera was provided by Professor Sarah Gilbert (University of Oxford). Using the IBV PV we produced in pI.18, we determined the levels of anti-IBV neutralizing antibodies pre- and post-vaccination with a Trivalent Influenza Vaccine (TIV) ( Figure 5A ). All participants (n=17) had pre-existing antibody response to Flu B – mean IC₅₀ dilution of ~61341. These responses were significantly boosted against 2/3 IBV PV tested; B/Hong Kong/8/1973 a pre-split PV ( Figure 5A (ii) (p=0.0046 Wilcoxon paired U test) and B/Florida/4/2006, of Yamagata lineage ( Figure 5A (iii) (p=0.0129 Wilcoxon paired U test). However, none of the participants had boosted responses against B/Brisbane/60/2008 of Victoria lineage (p=0.5791 Wilcoxon paired U test). Trial participants received an additional vaccination with either a placebo (saline solution) (n=8) or an MVA-NP-M1 fusion protein (n=9) (53). Antrobus et al. (53) have demonstrated that this additional boost with NP-M1 fusion MVA increased T cell responses to internal influenza proteins. They suggest that this may have helped HA-specific B cells leading to the increased seroprotection and seroconversion to H3N2 they reported (53). Therefore, post vaccination samples were compared between those who had received the NP-M1 fusion MVA and those who received the placebo to determine if this effect was responsible for the increase in post vaccination antibodies observed herein ( Figure 5A ). However, once post-vaccination IC₅₀ values were grouped based on the vaccine regimens administered, no significant difference was found in the IC₅₀ fold differences between pre-vaccination and post-vaccination values against any of the PV tested ( Figure 5B ). The range of responses was also greater in those participants who received TIV followed by MVA-NP-M1 fusion protein compared to those who received TIV followed by the placebo.

Currently the gold standard for determining vaccine efficacy is to test responses using hemagglutination inhibition (HI) and single radial hemolysis (SRH). We believe that these assays can be complemented by the use of pMN especially in the initial testing of novel vaccine candidate responses. Recently, it has been shown that influenza B pMN fold changes (IC₅₀) correlates strongly to SRH and HI assays (62). However, the HI assay is only able to detect antibodies directed against the HA head and we have not differentiated stem responses herein and so this may account for the weak correlation with the HI data presented in the clinical trial (r=0.1632, p=0.3563) (53) ( Supplementary Figure 2 ).

Serological surveillance studies monitor the virus and its reservoirs to aid in vaccine choice and treatment. To test whether the pseudotyped IBV could be used to test non-human sera, we obtained seal sera from an influenza surveillance study conducted by the University of Vienna. We evaluated these samples for neutralizing activity against influenza B strains, B/Hong Kong/8/1973, B/Yamagata/16/1988, and B/Victoria/2/1987 from pI.18 ( Figure 2A ) using pMN. In the limited number of sera tested, only a single positive signal was observed, against B/Victoria (IC₅₀ = 44) ( Supplementary Table 1 ), with none of the other serum samples showing neutralizing activity against any of the IBV PV tested. This correlates with previous data showing a prevalence of antibodies to IBV in only 2% of seals tested (n=971) after 1995 (5).

We tested the influenza B pseudotypes produced from pEVAC-HA ( Figure 2B ) against monoclonal antibodies obtained from the Food and Drug Administration, USA. Monoclonal antibody 3E8, specific to B/Yamagata viruses, was able to neutralize B/Phuket/3073/2013 from the B/Yamagata lineage with an IC₅₀ of 18.77 ng/mL ( Figure 6 ). However, 3E8 was unable to neutralize the B/Victoria lineage pseudotype. On the other hand, 5A1, specific to B/Victoria viruses, was able to neutralize B/Brisbane/60/2008 from the B/Victoria lineage with an IC₅₀ of 206 ng/mL ( Figure 6 ). Monoclonal antibody 5A1 was unable to neutralize the B/Yamagata lineage virus. Monoclonal antibody 2F11 which was previously reported to neutralize both B lineages was unable to neutralize both IBV PV at the concentrations tested ( Figure 6 ). Our results for 2F11 against B/Brisbane/60/2008 correlate with Carnell, 2021 (62) and we have employed a more recent B/Yamagata-like virus in this study, B/Phuket/3073/2013, compared to those tested previously, which may explain non-neutralization. Epitope mapping of the mAb binding site in these IBV may be done in future to determine if genetic change throughout the years may have influenced neutralization of IBV by this mAb. We have also shown here that neutralization of the IBV pseudotypes we tested was lineage-specific.