ELISA-based quantification of neuraminidase in commercial influenza vaccines using virus-derived reference antigens

ABSTRACT

Most commercial influenza vaccines are produced using inactivated viruses decorated with hemagglutinin (HA) and neuraminidase (NA) antigens from the recommended strains. However, only the HA antigen content is monitored, leaving the NA antigen content, stability, and potential contributions to efficacy unclear. We quantified functional NA amounts in commercial vaccines from two seasons by coupling NA enzyme-linked immunosorbent assays (ELISAs) with NA reference antigens that were isolated from egg-propagated vaccine strains (H1N1, H3N2, and type B) by Immobilized NA Active site Affinity Chromatography (INAAC). Our results show that nonexpired egg-based vaccines for the 2024–25 season contained the three expected NAs and that cell-based vaccines may require cell propagated NA reference antigens. NA amounts in egg-based vaccines varied by strain and manufacturer, ranging from 0.7 to 1.7 µg/dose or NA:HA ratios between 1:10 and 1:20. We confirmed the ELISA results with an NA activity analysis, indicating functional NA amounts were measured. In expired egg-based vaccines (2022–23 season), some NAs were at low or undetectable levels, indicative of structural integrity decay over time. These findings provide a foundation for quantifying NA content and stability in vaccines to evaluate NA’s potential efficacy contributions and optimize the design of future influenza vaccines that contain both NA and HA.

IMPORTANCE

The influenza virus surface antigens hemagglutinin (HA) and neuraminidase (NA) are both targeted by the immune system during infection. Although most licensed influenza vaccines contain both the HA and NA antigens from the recommended strains, only the HA content is monitored. Here, we measured functional NA amounts in commercial vaccines using a stability-indicating ELISA with NA reference antigens isolated from the vaccine viruses. Our results show that current egg-based vaccines contain the expected NAs at amounts that vary by strain and manufacturer. Expired vaccines had reduced or undetectable NA levels, suggesting NA may degrade over time. This methodology can be applied to measure NA content and stability in current vaccines and help to optimize the design of future vaccines that contain both NA and HA antigens.

INTRODUCTION

Although influenza virus infections induce antibodies against both the hemagglutinin (HA) and neuraminidase (NA) surface antigens (1), seasonal vaccines have historically focused on eliciting anti-HA antibodies (2). Consequently, all currently licensed seasonal influenza vaccines are multivalent and at least contain the HA antigens from recommended influenza A (H1N1 and H3N2) and B strains anticipated to circulate during the upcoming season. Most of these vaccines are produced using inactivated vaccine viruses grown in eggs or cells that also contain the NA antigens from the recommended strains (3). While the NA antigens could contribute to the vaccine efficacy, only the HA antigen amounts are monitored due to vaccine dosing recommendations, along with the lack of appropriate NA reference antigen standards and the lower NA content in viruses.

The primary role of HA is to facilitate viral entry by binding to sialic acid-containing receptors on host cells (4–6), whereas NA cleaves these sialic acid receptors to promote viral release and movement (7–10). Due to these distinct roles, antibodies that inhibit the function of HA or NA likely reduce viral replication through independent mechanisms that can potentially be exploited to improve vaccine efficacy. Supporting this possibility, antibodies against NA have been shown to be protective in humans and animal models (11–14). In addition, recent work suggests that antibodies against NA from seasonal H1N1 viruses may confer cross-protection against highly pathogenic H5N1 viruses (15–17) and that HA and NA antigenic drift occurs asynchronously (18, 19). This growing body of evidence supports quantifying NA in existing vaccines to assess contributions to efficacy and develop NA supplementation strategies.

Multiple studies have measured NA in vaccines using mass spectrometry (MS), enzymatic activity assays, immunoblotting, and enzyme-linked immunosorbent assays (ELISAs) (20–30). However, current MS-based approaches and immunoblotting do not distinguish native vs denatured protein, limiting the application for measuring NA stability and potency (e.g., properly folded NA amounts). Previously, functional NA amounts were measured using a common activity assay (23, 27, 29) and more recently a NA-specific titration reagent that eliminates the need of a NA standard (24), but both are currently challenging to implement for multivalent formulations. To analyze multivalent vaccine formulations, one study used an ELISA for quantifying NA in urea-denatured samples (25), whereas other reports used anti-NA antibodies in a microarray format that was shown to be specific, sensitive, and stability indicating (26, 28). Despite these advances, most current approaches are limited to relative or absolute values (e.g., NA polypeptide amounts) because appropriate NA reference standards are not available for quantifying functional NA content, which is likely to correlate with protective immune responses.

Due to the absence of NA reference antigens or standards, manufacturers typically confirm the presence of NA by measuring enzymatic activity in the monovalent (individual strain) bulks that are combined to formulate the final vaccine product. To address this deficiency, we recently developed an Immobilized NA Active-site Affinity Chromatography (INAAC) strategy for isolating enzymatically active NA directly from vaccine viruses (31). In this study, we measured NA content in seasonal vaccines by coupling a stability-indicating ELISA with INAAC-isolated NA reference antigens. Our results show that non-expired vaccines contain NA antigens from all three recommended vaccine strains at amounts that vary by strain and manufacturer. The implications of the results and future considerations are discussed.

RESULTS

Although more than 90% of currently administered influenza vaccines are manufactured using inactivated viruses containing the HA and NA antigens from the recommended strains (3), only the HA content is reported. The goal of this study was to measure the NA antigen amounts in commercial vaccines by coupling NA sandwich ELISAs with appropriate NA reference antigens or standards. To set up the ELISA, we initially produced several strep-tagged recombinant NA substrates. These included recombinant full-length NAs (rfNAs) and recombinant soluble NAs (rsNAs) from the recommended H1N1 (A/Victoria/2570/2019), H3N2 (A/Darwin/6/2021), and B-Victoria (B/Austria/1359417/2021) strains for the 2022–23 season (Fig. 1A). rsNAs were included because previous studies suggest some NAs are susceptible to cleavage in the stalk (31, 32). To produce rsNAs, we used the common approach of exchanging the N-terminal transmembrane region with a cleavable signal peptide and a tetrameric stabilization domain (33–37).

Fig 1.

Characterization of recombinant NA test antigens produced using insect cells. (A) Diagram of the recombinant full-length NA (rfNA) and soluble NA (rsNA) constructs expressed in Sf9 insect cells using recombinant baculovirus. The NA domains (e.g., transmembrane [TM], stalk, and head) and additional domains fused to the N-terminus of each construct are indicated. (B) Image of a Coomassie-stained gel containing the indicated rfNAs and rsNAs purified from insect cells. Proteins (~2 µg each) were reduced with DTT prior to resolution on SDS-PAGE (4%–12%) gels. (C) Activities of the indicated rfNA and rsNA serial dilutions were measured directly after preparation (left panel) and after storage at 4°C for 24 h. NA samples at the indicated storage concentrations were diluted 10-fold in reaction buffer containing MUNANA immediately before measuring the activity. Samples were analyzed in triplicate. All data points are displayed with the linear regression.

Each NA was expressed by infecting insect cells with recombinant baculovirus, isolated using Strep-Tactin XT and characterized. On reducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels, the isolated NAs resolved near the expected molecular weights (58–68 kDa) and the slower rsNA mobilities were in line with the presence of two additional N-linked glycan sites in the tetrabrachion tetramerization domain (Fig. 1B and Table S1). Size exclusion chromatography (SEC) showed that rfNAs in detergent form high molecular weight (>670 kDa) species (Fig. S1), indicative of oligomer formation. In contrast, rsNAs resolved as heterogeneous mixtures of tetramers and different order oligomers, in linesimilar to previous observations with rsNAs (35, 38). Consistent with the SEC analysis, rfNAs generally showed higher specific activity than the rsNAs with the reporter substrate 2′-(4-methylumbelliferyl)-α-D-N-acetylneuraminic acid (MUNANA), suggesting the rfNA preparations contain more functional protein (Fig. 1C). Interestingly, when we stored the diluted samples at 4°C for ~24 h, the rfNA-specific activities only showed slight variation, whereas the rsNA activity was largely lost, indicating rfNAs are likely more stable at lower concentrations.

We then used antibodies identified from literature to set up three different sandwich ELISAs for quantifying the subtype 1 (N1), subtype 2 (N2), and type B (NB) NA amounts in recent seasonal vaccines (Fig. 2A). N1 and NB were captured using the universal active-site binding antibody FNI-9 (12), whereas N2 was captured with the FNI-9/R27D mutant that binds recent N2s with higher affinity (39). For detection, we used horseradish peroxidase (HRP)-labeled NA-type-specific monoclonal antibodies (MAbs) that bound but did not inhibit the NA by either steric or active site-binding mechanisms identified by a Neuraminidase Active Site Proximity Assay or NASPA (39). These included the following MAbs, 5D11 raised against rfNA from the H1N1 strain A/Victoria/2570/2019, 48G11 raised against rsNA from the strain B/Austria/1359417/2021, and NDS.1 a N2 MAb isolated from humans (40).

Fig 2.

Specificity and topology bias of sandwich ELISAs for NAs from recent vaccine strains. (A) Diagram showing the sandwich ELISAs for detecting NAs from the H1N1, H3N2, and type B-Victoria lineage vaccine strains recommended for the 2022–23 season. MAbs for capturing and detecting each NA are indicated. (B) NA sandwich ELISA specificity was tested using the individual rfNAs and a mixture of all three rfNAs. rfNA samples were twofold serially diluted from the indicated concentrations and run in triplicate. All values are displayed with insets showing the area under the curve (AUC) for each run. P values are from one-way ANOVA multiple comparisons tests. (C) Topology bias for each NA sandwich ELISA was tested using the corresponding rsNA and rfNA antigens individually and combined in a mixture. NA test antigens at the indicated concentrations were twofold serially diluted and run in triplicate. All values are displayed with insets showing each run AUC. Indicated P values are from one-way ANOVA multiple comparisons tests. n.s., not significant (P > 0.05) and **** (P < 0.0001).

To confirm the specificity of each ELISA, we first analyzed serial dilutions of the individual rfNAs starting from 1.0 µg/mL and a mixed sample starting with 1.0 µg/mL of each rfNA. All three ELISAs detected the correct rfNA, and the signal largely overlapped with the mixed sample, especially at the lower dilutions (Fig. 2B). At the two highest concentrations, we noted slight cross-reactivity in the N1 ELISA, but it did not add to the mixed sample signal. To test topology bias, we used individual serial dilutions of matching rfNAs and rsNAs starting from 1.0 µg/mL and a mixed sample starting with 0.5 µg/mL of each. All the signals overlapped in the N1 ELISA, indicative of no bias (Fig. 2C). However, rsNB gave slightly higher signal in the NB ELISA and rfN2 gave higher signal in the N2 ELISA, suggesting the two may have some topology bias. Finally, we confirmed the ELISA can monitor the structural integrity of the NAs as the signal and enzymatic activity were both lost after incubating the rfNAs at 70°C for 10 min (Fig. S2).

Our results thus far suggested that the sandwich ELISAs could be coupled with appropriate reference antigens to measure NA content in current trivalent influenza vaccine formulations. To test this possibility, we used a recently developed Influenza NA Active-site Affinity Chromatography (INAAC) strategy (31) to isolate NA antigens directly from egg- propagated vaccine viruses (Fig. 3A). For the initial test, we used the H1N1 (A/Victoria/2570/2019), H3N2 (A/Darwin/6/2021), and B-Victoria (B/Austria/1359417/2021) strains that were recommended for egg-based vaccines during the 2022–23 season.

Fig 3.

ELISA analysis of expired commercial influenza vaccines with NAs isolated from the vaccine strains. (A) Schematic of the INAAC strategy for isolating unmodified viral NA antigens (vNAs) from egg-propagated viruses recommended for the 2022–23 season. (B) Image of a Coomassie-stained gel containing the indicated vNAs purified from egg-propagated vaccine viruses. Proteins (~1 µg each) were treated with and without DTT prior to resolution on SDS-PAGE (4%–12%) gels. Bands corresponding to NA monomers, dimers, SDS-resistant tetramers, and higher-order oligomers are indicated. Asterisks denote NB head domains that are potentially generated by proteolytic cleavage of the stalk. (C) Activity of the vNAs isolated by INAAC was measured in triplicate at the indicated protein concentrations using MUNANA. All data are displayed with the linear regression. (D) NA sandwich ELISA results comparing two expired commercial egg-based inactivated influenza virus vaccines with vNAs isolated from the recommended egg-based strains for the same season (2022–23). Vaccine samples and vNA reference antigens were twofold serially diluted and run in duplicate on the same plate. vNA reference antigen amounts per well are indicated. All data are displayed in a log-log scale with a 4-parameter logistic (4-PL) regression. Upper (ULOD) and lower limits of detection (LLOD) are indicated.

The viral NAs (vNAs) isolated from the egg-propagated vaccine strains predominantly migrated at the expected molecular weight on reducing SDS-PAGE gels (Fig. 3B and Table S1). The NA from the type B virus (vNB/Aus21) showed an additional faster migrating band (Fig. 3B, see asterisks), which we previously interpreted as head domains created by proteolytic cleavage of the stalk based on the purification scheme and reactivity with antibodies against the B head domain (31). Under non-reducing conditions, vNAs from the H1N1 virus (vN1/Vic19) and H3N2 virus (vN2/Dar21) both showed the expected disulfide-linked dimers, SDS-resistant tetramers, and higher-order oligomers (41). In contrast, vNB/Aus21 primarily migrated as monomers, disulfide-linked dimers, and a faster migrating band likely corresponding to oxidized head domains. As expected, all three vNAs showed enzymatic activity using MUNANA, and vN2/Dar21 displayed the highest specific activity followed by vNB/Aus21 and vN1/Vic19 (Fig. 3C).

After the characterization, we included the vNAs as reference antigens in an NA ELISA analysis of two expired egg-based vaccines for the 2022–23 season (Fig. 3D, and Table S2). One vaccine showed good signal in the N1 ELISA, and both displayed visible signal in the NB ELISA but poor signal in the N2 ELISA. Analysis of the NA ELISAs data for the vNAs showed reasonable log-log linearity (R² = 0.98–0.99) and accuracy (100% ± 15%) for the samples containing 1.6–50 ng of vN1/Vic19, 1.6–25 ng of vNB/Aus21, and 0.6–20 ng of vN2/Dar21 (Fig. S3 and Table S3). Within these limits, both vaccine samples in the NB ELISA and one in the N1 ELISA produced signal that showed parallelism with the vNA (e.g., similar slope), indicating the N1 and NB in these samples are similar to the vNAs. Finally, we used these data to calculate that one expired vaccine contained ~1.7 µg of N1 and ~1.4 µg of NB per dose and the other contained ~1.6 µg of NB per dose, which were all ~10% of the 15 µg formulated HA amount (Table S4).

The poor signal for both expired vaccines in the N2 ELISA and one in the N1 ELISA suggested that either the NA ELISAs coupled with the vNAs did not effectively measure the content of all three NAs in the vaccines or that the NAs have variable stability. To address these possibilities, we acquired two commercial egg-based vaccines (Fluzone and Flulaval) and one cell-based vaccine (Flucelvax) for the 2024–25 season (Table S2) and used INAAC to isolate NA antigens from the recommended egg-based vaccine viruses (Fig. 4A). The isolated vNAs migrated at the expected molecular weights on reducing SDS-PAGE gels and showed the expected disulfide-linked dimers and SDS-resistant tetramers (Fig. 4B and Table S1). The NA from the type B strain (vNB/Aus21) did not change resulting in a similar SDS-PAGE gel profile as before (Fig. 4B). We again verified that the isolated vNAs are functional, and vN2/Thai22 showed the highest specific activity followed by vNB/Aus21 and vN1/Vic22 (Fig. 4C).

Fig 4.

ELISA analysis of non-expired commercial influenza vaccines with NAs isolated from the vaccine strains. (A) Illustration of the INAAC capture antibodies used for isolating unmodified vNAs from egg-propagated viruses recommended for the 2024–25 season. (B) Representative Coomassie-stained gel image of the indicated vNAs purified from egg-propagated vaccine viruses. vNAs (~2 µg each) were treated with and without DTT prior to resolution on SDS-PAGE (4%–12%) gels. Bands corresponding to NA monomers, dimers, and SDS-resistant tetramers are indicated. Asterisks denote NB head domains that are potentially generated by proteolytic cleavage of the stalk. (C) Activity of the vNAs isolated by INAAC was measured at the indicated protein concentrations in triplicate using MUNANA. All data are displayed with the linear regression. (D) NA sandwich ELISA results comparing non-expired commercial egg-based (Fluzone and Flulaval) and cell-based (Flucelvax) inactivated influenza virus vaccines with vNAs isolated from the recommended egg-based strains for the same season (2024–25). Vaccine samples and vNA reference antigens were twofold serially diluted and run in triplicate on the same plate. vNA reference antigen amounts per well are indicated. All data are displayed in a log-log scale with a 4-PL regression. ULOD and LLOD are indicated. (E) Activities of the indicated vaccine samples and vNA reference antigen mixtures, formulated based on the ELISA-determined NA vaccine concentrations, were twofold serially diluted, and measured in triplicate using MUNANA. All data are displayed.

In contrast to the prior results, all three non-expired vaccines showed visible signal in each NA ELISA (Fig. 4D). Analysis of the vNA data showed log-log linearity (R² = 0.98–0.99) and accuracy (100% ± 20%) for the samples containing 6.3–100 ng of vN1/Vic22, 3.1–100 ng of vNB/Aus21, and 1.6–50 ng of vN2/Thai22 (Fig. S4 and Table S5). Within these ranges, results from the egg-based vaccines in all three NA ELISAs were largely parallel to the vNAs. With the exception of the NB ELISA, results from the cell-based vaccine were generally lower and showed suboptimal parallelism with the vNAs (Fig. 4D and Fig. S4). However, the cell-based vaccine results may have been influenced by our use of vNAs isolated from viruses grown in eggs rather than cells.

For calculating the NA amounts, we only analyzed the egg-based vaccine data because of the potential mismatch of the isolated vNAs with the cell-based vaccine (Table S6). All three NAs were present in both egg-based vaccines and the amounts of each ranged from 0.7 to 1.7 µg per dose (Table 1), suggesting the undetected NAs in the expired vaccines were attributed to instability rather than the ELISA. Interestingly, NA amounts per dose generally differed between the vaccines by a factor of 2 and the one with higher N1 amounts (~1.7 µg vs ~1 µg) had lower N2 (~0.7 µg vs ~1.7 µg) and NB (~0.9 µg vs ~1.7 µg) amounts. To test the ELISA results by an orthogonal method, we compared the NA-specific activity of each egg-based vaccine to a corresponding mixture of the vNAs at the measured concentrations. The specific activity of one vNA mixture was within 100% ± 15% of the corresponding vaccine, and the other was within 100% ± 33%, supporting that the ELISA measurement reflected functional NA amounts. Overall, our results demonstrate that vNAs isolated from vaccine viruses can be used to measure NA amounts in vaccines. In addition, we show that the NA amounts vary between vaccines likely because of differences in the manufacturing process, vaccine viruses, or NA stability in the formulation buffers.

TABLE 1.

NA ELISA quantification results for 2024–25 influenza vaccines

Vaccine	NA	Mean NA amount in 0.5 mL dose ± SD (µg)a
Fluzone	N1/Vic22	1.67 ± 0.07
	N2/Thai22	0.74 ± 0.07
	NB/Aus21	0.90 ± 0.05
Flulaval	N1/Vic22	0.97 ± 0.06
	N2/Thai22	1.72 ± 0.23
	NB/Aus21	1.67 ± 0.13

^a Raw data for the calculations are provided in Table S6.

DISCUSSION

Inactivated influenza virus vaccines comprise over 90% of administered doses (3). These vaccines are produced using egg- or cell- propagated viruses that contain the HA and NA antigens from the recommended strains. Despite the established importance of anti-NA immunity in protection against influenza (11–14, 42, 43), only the HA content in vaccines is monitored. Here, we measured functional NA amounts in commercial vaccines using ELISAs coupled with viral NA (vNA) reference antigens isolated from egg-propagated vaccine strains via INAAC (31). All three vaccine strain NAs were detected in non-expired egg-based vaccines, whereas some NAs were low or undetectable in expired egg-based vaccines, suggesting NA may decay over time. NA amounts in the non-expired vaccines ranged from 0.7 to 1.7 µg/dose (5%–10% of HA amounts), and the results aligned well with an NA activity analysis, indicating functional NA was measured. These findings establish a foundation for measuring NA amounts and stability in vaccines and suggest that existing NA antigens could contribute to antibody responses.

Currently, there is significant interest in improving influenza vaccines by increasing the suboptimal NA amounts through supplementation strategies that have shown varying levels of success (37, 44–46). However, these studies were performed without knowing the existing NA vaccine content that could negatively or positively influence the results between manufacturers and seasons. In this study, we quantified all three NAs (e.g., N1, N2, and NB) in multiple vaccines from two seasons by applying a recent approach for isolating functional NAs from viruses (31). Our results show that the functional NA content varies by strain and manufacturer, indicating each would likely require a unique NA supplementation strategy. Interestingly, when we set up the ELISAs using NA-specific MAbs for capture and the universal active-site MAb for detection, we observed a stronger signal from the mixed rfNA control vs the individual rfNAs, suggesting the different rfNAs can interact. While swapping the capture and detection antibodies addressed this property, it does not rule out that NA interactions in complex vaccine mixtures could impact the immunogenicity.

Our NA estimates of 0.7–1.7 µg/dose were calculated with Bicinchoninic Acid (BCA)-determined vNA concentrations, resulting in NA:HA ratios between 1:10 and 1:20 as each HA in the vaccines was formulated at ~15 µg/dose. These NA:HA ratios are lower than what was reported from isotope dilution MS or immunoblotting of other egg-based vaccines (20, 30), but consistent with estimates on virions (47, 48). We also did not adjust the vNA concentrations for purity on SDS-PAGE gels due to the vNB heterogeneity, which would have reduced N1 and N2 amounts by ~10%–20%. This raises questions about how vNA reference antigens should be characterized, especially since our results suggest cell-based vaccines may require cell-propagated NA reference antigens. Beyond specific activity and SDS-PAGE profiles, future vNA characterizations could include size in solution, additional purity assays, and functional vNA fractions using reagents such as titration reagent 1 or TR1 (24). The latter can provide significant insight into the vNA reference antigen quality as functional NA fractions have been shown to vary between NA preparations (36, 49) and are likely critical for protective antibody responses.

During the study, we made a few observations with potential implications. One was the inability to detect some NAs in expired vaccines that suggest stability issues. However, more thorough stability studies are needed during an influenza season to determine the relevance of this observation since these samples were analyzed ~2 years post-expiration. Another was that NA content varied ~2-fold between manufacturers, possibly reflecting differences in manufacturing, NA stability, or the choice of vaccine strains as NA content can vary between viruses (50). In addition, it is also possible that these values vary by lot, and this was not accounted for in our study as we only examined one lot of each vaccine. Lastly, the parallelism analysis showed that two vaccines had higher slopes than vNB and three had lower slopes than vN2. Based on our topology bias data, this suggests the vaccine NAs may undergo changes at 4°C such as stalk cleavage. Immunoblots could be used to more directly address this possibility, especially since changes from detergent-bound to soluble NAs could affect the immunogenicity and contribute to the development of antibodies that target the bottom of NA such as NDS.1 (40).

Overall, our results demonstrate that seasonal influenza vaccines contain functional NA amounts that vary by strain and manufacturer, which can potentially contribute to antibody responses and efficacy. To reach this conclusion, we also showed that vNA antigens isolated by INAAC can serve as reference antigens though factors including vNA characterization and source material (e.g., eggs vs cells) require further investigation. Although the antibodies we used in the ELISA were suitable for the analyzed vaccine NAs, we expect that the antibody selection can be improved and may require adaptation as strains evolve, which can additionally impact results. Nonetheless, the outlined approaches can support future studies of critical questions such as NA stability, lot-to-lot consistency, and relationships between NA quantity and immune responses to help to optimize functional NA content in vaccines.

MATERIALS AND METHODS

Generation of recombinant baculoviruses

Recombinant baculoviruses (BVs) for expressing the rNAs were generated using the Bac-to-Bac BV Expression system (Gibco). Briefly, NA sequences from the following vaccine strains A/Victoria/2570/2019 (EPI1741925), H3N2 strain A/Darwin/9/2021 (EPI1888081), and type B strain B/Austria/1359417/2021 (EPI1924335) were obtained from the Global Initiative on Sharing All Influenza Data (GISAID). For each rfNA, a Strep-Tag and linker sequence were fused to the N-terminus (residue 2), and for each rsNA, a sequence encoding a cleavable signal peptide, Strep-tag, and the tetrabrachion tetramerization domain were fused to the N-terminus (residue 35). All constructs were codon optimized for insect cells, synthesized, inserted into the pFastBac1 vector NotI and HindIII sites (GenScript), and used to transform DH10Bac E. coli cells. Positive clones, identified by blue-white screening, were grown in 5 mL culture, and plasmid DNA was isolated using a ZB BAC DNA miniprep kit (Zymo Research). P0, P1, and P2 working BV stocks were created using Spodoptera frugiperda 9 (Sf9) cells as previously described (51).

Recombinant NA expression, purification, and size exclusion chromatography analysis

Expression of rsNAs in Sf9 cells and purification from the culture medium by Strep-Tactin XT affinity chromatography (Cytiva) was performed as previously described (39). The expression of rfNAs in Sf9 cells was similar except the cells were harvested ~72 h post-infection by sedimentation (4,000 × g; 10 min) and stored at −80°C if not used directly. Cell pellets were solubilized at 4°C in Buffer A (20 mM HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) pH 7.0, 300 mM NaCl, and 1 mM CaCl₂ containing 2% Triton X-100 (TX100), 10 U/mL Benzonase (MilliporeSigma), and 1 × protease inhibitor (SIGMAFAST EDTA-free Protease Inhibitor Tablets), clarified by sedimentation (100,000 × g; 1 h), passed through a 0.22 μm filter, and loaded onto a Strep-Tactin XT column using an AKTA start. The column was washed with 0.025% TX100 Buffer A and eluted using Buffer B (20 mM HEPES pH 7.0, 150 mM NaCl, 1 mM CaCl₂, 0.025% TX100) containing 50 mM biotin. Fractions with NA activity were pooled, concentrated with a 100 kDa molecular weight cut-off (MWCO) centrifugal filter (Amicon), and dialyzed three times against Buffer B at 4°C in a 3.5 kDa Molecular weight cut-off slide-a-lyzer (Thermo Fisher Scientific). Protein concentrations were determined by a Micro BCA assay (Thermo Fisher Scientific) and adjusted to 0.5–1.0 mg/mL aliquoted and stored at −80°C. The size of the isolated rfNAs and rsNAs was examined using an Agilent 1260 prime HPLC equipped with a variable wavelength detector (VWD) and an AdvanceBio SEC 300A column (Agilent Technologies) as previously described (51). Briefly, samples (20 µL) were run at 1 mL/min either with a mobile phase buffer (20 mM HEPES pH 7.0, 150 mM NaCl, and 1 mM CaCl₂) containing 0.001% TX100 (rfNAs) or without TX100 (rsNAs), and the absorbance (Abs) at 280 nm was monitored with the VWD. AdvanceBio SEC 300A protein standard (Agilent) was used for molecular weight references.

NA enzymatic activity assay

NA activity assays were performed and analyzed as previously described (45). Briefly, samples containing indicated NAs were serially diluted in assay buffer (25 mM MES (2-(N-morpholino) ethanesulfonic acid) pH 7, 150 mM NaCl, 1 mM CaCl₂, and 0.075% TX100) and transferred (10 μL) to 96-well, black-wall, clear-bottom plates (Corning) prewarmed to 37°C. Reactions were initiated by adding 190 μL of 37°C substrate solution (185 μL assay buffer and 5 μL 2 mM MUNANA), and the fluorescence (λEx: 355 nm, λEm: 450 nm) at 37°C was immediately measured for 10 min at 30 s intervals using a Cytation 5 plate reader (BioTek). NA activity was determined based on the slope of the early linear region in the relative fluorescence unit (RFU)-vs-time graph.

NA sandwich ELISAs

Immulon 4HB flat-bottom microtiter plates (Thermo Fisher Scientific) were coated with a recombinant murine IgG1 FNI-9 MAb chimera (N1 and NB ELISAs) or the R27D point mutant (N2 ELISAs) as follows (12, 39). MAbs diluted to 5 μg/mL in phosphate-buffered saline (PBS) pH 7.4 were transferred (100 µL/well) to plates and incubated at 4°C overnight. Wells were blocked for 1 h at 37°C with 3.0% Bovine serum albumin (BSA) in dilution buffer (20 mM HEPES, pH 7.0, 300 mM NaCl, and 1 mM CaCl₂). Commercial vaccines, purchased by the LID Clinical Studies Unit, NIAID, NIH from McKesson, were stored at 4°C. Vaccine samples diluted 1:5 in dilution buffer containing 0.5% BSA were twofold serially diluted with the NA samples in Nunc 96-well natural plates (Thermo Fisher Scientific) using dilution buffer containing 0.5% BSA. Diluted samples were transferred (100 µL/well) to the ELISA plates and incubated 2 h at 37°C. Wells were washed (6 × 250 µL) with PBS pH 7.4 containing 0.025% TX100 and incubated 90 min at 37°C with HRP-conjugated NA subtype-specific MAbs diluted 1:1,000–1:5,000 in dilution buffer containing 0.5% BSA. Wells were washed (3 × 250 µL) with PBS pH 7.4 containing 0.025% TX100 and developed 10 min at 37°C with o-phenylenediamine dihydrochloride (Sigma), and reactions were stopped with 1 N H₂SO₄. Abs at 490 nm (A_490nm) was read with a Cytation 5 plate reader. The upper limit of detection was defined by the plate reader (A_490nm > 3.95), and the lower limit of detection was determined by the mean + 3 standard deviations of dilution buffer control samples on the same plate.

Purification of NAs from egg-propagated viruses

Vaccine strains obtained from the WHO were propagated in 10-day-old specific-pathogen-free (SPF) eggs from Charles River Labs. These included egg-based vaccine strains recommended for the 2022–23 season: A/Victoria/2570/2019 (H1N1), A/Darwin/9/2021 (H3N2), and B/Austria/1359417/2021; and 2024–25 season: A/Victoria/4897/2022 (H1N1), A/Thailand/8/2022 (H3N2), and the same B strain. Viruses were grown for 3 days at 33°C, and eggs were placed at 4°C for 2–18 h prior to harvesting the allantoic fluid. Allantoic fluid was clarified by centrifugation (2,000 × g; 5 min), and viruses were isolated by sedimentation (100,000 × g; 45 min) at 4°C through a sucrose cushion (25% [wt/vol] sucrose, PBS pH 7.2, and 1 mM CaCl₂). Viral pellets were stored at −80°C or directly solubilized with Buffer A containing 2% TX-100 and isolated by INAAC using TX100 as recently described (31). Protein concentrations were determined by Micro BCA Protein Assay Kit and stored at −80°C.

SDS-PAGE and Coomassie staining

NA samples were mixed with 2 × LDS sample buffer with or without 0.1 M dithiothreitol (DTT), heated at 50°C for 10 min, and resolved on a 4%–12% polyacrylamide Tris-Glycine SDS-PAGE wedge gel (Thermo Fisher Scientific). Gels were stained with simple blue and imaged with an Azure C600.

Data and statistical analysis

All assays were run in triplicate unless otherwise stated. GraphPad Prism 10 software (GraphPad Software, Inc) was used to perform linear regression analysis of NA-specific activity and the ELISA results for assessing parallelism, 4-parameter logistic (4-PL) regression analysis for calculating NA amounts, area under the curve (AUC) measurements, and P-value calculations. P-values were determined with a one-way ANOVA multiple comparisons test, and P < 0.05 was considered not significant (n.s.). Mean dose amounts, percent accuracy, percent coefficient of variation, and percent previous dilution concentration were all calculated with Excel software. Linearity of vNA reference antigens was determined by samples that showed 100% ± 15% accuracy. NA purity was calculated by densitometry analysis of reducing Coomassie- stained gels with ImageJ software by dividing the full-length NA band intensity by the sum of all band intensities in the lane.

DATA AVAILABILITY

All raw data sets used and analyzed in this study are provided in the Supplemental material.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/spectrum.01227-26.

Supplemental figures. spectrum.01227-26-s0001.pdf.

Fig. S1 to S4.

Supplemental material. spectrum.01227-26-s0002.xlsx.

Raw data associated with all main figures.

Supplemental tables. spectrum.01227-26-s0003.docx.

Tables S1 to S6.

OPEN PEER REVIEW. reviewer-comments.pdf.

An accounting of the reviewer comments and feedback.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.