NAI antibody measurements are substrate-dependent

NA activity is often measured using small monovalent reporter substrates like MUNANA or more biologically relevant multivalent substrates such as the glycoprotein fetuin. With MUNANA (Fig. 1a), NA activity is readily measured by using fluorescence to quantify the 4-methyl umbelliferone that is released following sialic acid cleavage. In contrast, NA activity measurements with glycoproteins require more complex assays like ELLA, which monitor sialic acid cleavage by using peanut agglutinin binding to the exposed galactose residues on the glycoprotein (Fig. 1b). Initially, we used both assays to measure NA activity from the H1N1 strain A/California/07/2009 (H1N1/CA09) in the presence of ferret or mouse antisera that were raised against a reassortant virus (H6N1/CA09) carrying H6 and the NA from H1N1/CA09. With the fetuin-based ELLA, ferret and mouse antisera both reduced the NA activity by more than 50% (Fig. 1c), whereas no significant activity changes were observed with the smaller substrate MUNANA (Fig. 1d). This observation is in line with the common practice of using ELLA to measure serum NAI antibodies and several reports showing NAI antibodies can sterically hinder the ability of the NA active site to bind sialic acid residues on large N-linked glycans, but not on smaller reporter substrates like MUNANA^{40,45,46,47,48}.

a, b Diagrams showing the detection of NA enzymatic activity with the (a) small monovalent reporter substrate MUNANA and (b) multivalent glycoproteins by ELLA. MUNANA cleavage is monitored by fluorescent detection of the released umbelliferone. Cleavage of the sialylated glycans on the glycoprotein is measured by HRP-conjugated peanut agglutinin (PNA-HRP) binding to the exposed terminal galactose residue. c, d NA activity inhibition by ferret and mouse antisera was measured with a H1N1/CA09 virus by (c) ELLA using the glycoprotein fetuin and (d) MUNANA. Ferret and mouse antisera were generated by intranasal infection with a reassortant virus (H6N1/CA09) carrying a mismatched HA (H6) and a matching NA (N1/CA09). e NA activity in the H1N1/CA09 virus was measured by ELLA in the presence of serially diluted H1 MAbs that inhibit the receptor binding function of the virus (Supplementary Table 1). Purified H1 MAbs were diluted as indicated from an initial concentration of ~1 mg/ml. f NA activity in the H6N1/CA09 virus was measured by ELLA in the presence of serially diluted ferret and mouse antisera raised against the H6N1/CA09 virus.

HA contributes to the viral NA activity measurements with multivalent glycoprotein substrates

Bovine fetuin contains up to three sialylated N-linked glycans and several sialylated O-linked glycans⁴⁹ that can act as substrates for NA or potential receptors for HA (Fig. 1b). Consequently, viruses can bind to fetuin via HA which localizes the neighboring NA proteins near the sialylated glycan substrates, likely increasing the apparent activity of NA in a virus^50,51. To test the contribution of HA binding when fetuin is localized to a surface, we measured the NA activity in the H1N1/CA09 virus by ELLA in the presence of several H1 monoclonal antibodies (MAbs) that inhibit H1N1/CA09 receptor binding (Supplementary Table 1). All the H1 MAbs significantly reduced the NA activity measured by ELLA (Fig. 1e), resulting in false positive NAI titers of more than 10,000 (e.g., reciprocal dilution of the half-maximal inhibitory concentration (IC₅₀) of the antibody). In line with this result, we also observed a significant NA activity decrease in the H6N1/CA09 virus when it was measured by ELLA in the presence of the H6N1/CA09 ferret and mouse antisera (Fig. 1f), which can also inhibit H6N1/CA09 receptor binding (Supplementary Table 1). With the H6N1/CA09 virus these antisera produced NAI titers of ~10,000 compared to ~160 with the HA subtype mismatched H1N1/CA09 virus. Together, these results illustrate how antibodies against HA, especially those that inhibit the receptor binding function, can produce false positive or inflated NAI antibody measurements by ELLA. Supporting this conclusion, several previous studies have also reported that ELLA is subject to interference by other HA antibodies^37,38,39, making it difficult to establish NAI correlations²⁷, or implement it for monitoring NA antigenic drift with the same reagents (viruses and ferret antisera) used for analyzing HA antigenic drift.

Development and pilot testing of an NA active site proximity assay (NASPA)

An ideal NAI assay should be compatible with the different NAs from circulating viruses, capable of differentiating steric versus active site inhibition, and insensitive to the presence of HA antibodies as this would eliminate the need for NA specific reagents (e.g., strains or antisera). We posited that this could be achieved by creating a NA active site proximity assay (NASPA) that uses a large chemical NA active-site-binding inhibitor (NAi), or monoclonal antibody, to replace the function of the sialylated glycans on fetuin that serve as NA substrates in ELLA (Fig. 2a). We initially set the NAi at a half-maximal concentration (IC₅₀) or higher to create NA populations with both bound and accessible active sites. Under these conditions, we hypothesized that sera containing steric NAI antibodies (Fig. 2a, step 1) would block the large NAi from binding the active site (Fig. 2a, step 2), resulting in NA activity increases that could be measured using the small substrate MUNANA (Fig. 2a, step 3). We speculated that these conditions would also enable detection of NAI antibodies that bind the active site as these could target the sites not occupied by the NAi, resulting in NA activity decreases that could be confirmed in a MUNANA only assay.

a Diagram of NASPA showing the expected outcomes of the three steps in the presence (blue region) or absence (orange region) of steric and active site binding NAI antibodies. Following (step 1) a sera titration with a fixed amount of intact or detergent treated virus, (step 2) a large ‘bulky’ NA inhibitor (NAi) is added at the half-maximal inhibitory concentration (IC₅₀), and (Step 3) the ability of steric and active site NAI antibodies to block (steric) the large NA inhibitor or contribute to inhibition (active site) is detected by MUNANA. b H1N1/CA09 virus in the absence or presence of TX100 (0.025% final concentration) was serially diluted and NA activity was measured for 10 min after MUNANA addition. c. Inhibition curves are displayed for the NAi-MAb (FNI-9) with untreated and TX100-treated H1N1/CA09 virus. The NAi-MAb was incubated 3 h at 37 °C prior to measuring NA activity by MUNANA. d, e NA activity data is displayed from NASPA that was performed using untreated and TX100-treated H1N1/CA09 virus with the indicated (d) ferret antisera or (e) H1 MAbs that inhibit the receptor binding function of the H1N1/CA09 virus. Assays were performed in duplicate using NAi-MAb amounts approximating the IC₈₀ and are displayed as the mean ± the standard deviation (SD). NA activities were read for 10 min after MUNANA addition. Gray regions indicate the mean (dashed line) ± 3 SDs (dotted line) that were determined using (d) negative control ferret sera or (e) multiple wells containing PBS.

We initially tested this concept with the H1N1/CA09 virus and several recently identified NA inhibiting monoclonal antibodies (NAi-MAbs) FNI-9, FNI-19, and 1G01, which inhibit a broad range of NAs from type A and B viruses by binding to the active site^52,53. The pilot test was also performed in the absence and presence of the detergent Triton X-100 (TX100) at a final concentration of 0.025% to compare results from intact virions to spatially separated HA and NA antigens (Fig. 2a). We first titrated the virus by measuring NA activity with MUNANA to identify amounts that cleaved ~0.25–0.5 pmol/sec of substrate (Fig. 2b). Next, we measured the IC₅₀ of the NAi-MAbs (FNI-9, FNI-19 and 1G01) with and without TX100 using the standardized virus amounts (Fig. 2c and Supplementary Fig. 1a). We then incubated the same virus amounts with serial dilutions of ferret antisera raised against H1N1/CA09 or a H6N1/CA09 reassortant virus to investigate if the sera possess detectable steric or active site NAI antibodies by NASPA and the influence of anti-HA antibodies against the test virus. Each well was then incubated with the NAi-MAb at an inhibitory concentration greater than 50% prior to measuring NA activity with MUNANA. Using this approach, both ferret antisera increased the NA activities above the negative control ferret sera with all three NAi-MAbs and the activities decreased to control levels as the sera were diluted (Fig. 2d, Supplementary Fig. 1b, c).

We then determined the NASPA titers for each serum by identifying the largest dilution factor that showed NA activities more than 3 standard deviations (SDs) above the mean NA activity from the negative control sera. The results with all three NAi-MAbs showed that TX100 reduced the NASPA titers from the ferret antisera against H1N1/CA09 that possessed antibodies against the HA in the H1N1/CA09 test virus (Table 1). Importantly, TX100 treatment also eliminated false positive NA activity increases that were observed when NASPA was performed with the three MAbs that recognize the receptor binding domain of the HA in H1N1/CA09 (Fig. 2e). Together, these results suggested that NASPA could be used to measure NAI antibodies and that NA and HA should be separated to avoid potential interference from antibodies against HA.

Compatibility and optimization of NASPA with the different NAs from influenza vaccines

Vaccines against circulating influenza viruses generally include two type A subtypes (H1N1 and H3N2) and a type B virus that is currently from the Victoria lineage. Therefore, we performed a more thorough analysis of NASPA using different vaccine strains and the NAi-MAb FNI-9 that possesses broad NA binding properties⁵². For the vaccine strains we used H1N1 (A/Brisbane/02/2018), H3N2 (A/Hong Kong/4801/2014), and a type B (B/Austria/1359417/2021) virus, as well as H6 reassortant viruses that carried NAs from the same vaccine strains (Fig. 3a).

a Schematic of the type A (H1N1 and H3N2) and B (HBNB) vaccine strains and the corresponding H6 reassortant viruses that were analyzed by NASPA. b The NAi-MAb (FNI-9) IC₅₀ was measured after different incubation times with the indicated vaccine viruses in the absence and presence of TX100. Results are displayed as the mean ± SD from two independent experiments. c NASPA results are displayed for the three vaccine strains in the absence (upper panels) or presence of TX100 (lower panels) and the indicated ferret antisera raised against H6 reassortant viruses carrying the vaccine strain NAs. Negative ferret serum was included as a control. d NASPA results are shown for the H6 reassortant viruses carrying the vaccine strain NAs with the ferret antisera that were raised against the same viruses. NASPA was performed in the absence (upper panels) or presence of TX100 (lower panels). Negative ferret serum was included as a control. All NASPA assays were performed in duplicate using NAi-MAb concentrations corresponding to the IC₈₀ and are shown as the mean ± SD of the NA activities measured for 10 min after MUNANA addition. Gray regions indicate the mean (dashed line) ± 3 SDs (dotted line) from the negative control sera.

After titrating the vaccine viruses, we measured the IC₅₀ of the NAi-MAb for each virus after different incubation times in the absence and presence of TX100 (Supplementary Fig. 2). A temporal plot of this data showed that the NAi-MAb binding required ~180 min at 37 °C to reach equilibrium for all three NAs (Fig. 3b). We did note some NA-dependent variation in the equilibration times and TX100 effects. Next, we tested different ferret sera incubation times and observed consistent results with each vaccine virus after 30 min (Supplementary Fig. 3). Finally, we used these parameters to analyze each vaccine virus in the absence and presence of TX100 with a panel of NA ferret antisera that were generated using H6 reassortant viruses to avoid recognition of HA in the vaccine strains (Fig. 3c). The H1N1 and the type B vaccine viruses were only recognized by matching NA ferret antiserum and the data did not change when TX100 was present (Fig. 3c, left and right panels). Unexpectedly, all three NA ferret antisera and the control sera caused NA activity increases with the H3N2 vaccine virus without TX100 (Fig. 3c, upper middle panel). TX100 addition minimized these nonspecific interactions (Fig. 3c, lower middle panel), suggesting that binding of the H3N2 virus to serum components, presumably via HA, may block the access of the NAi-MAb to the NA in the H3N2 virus.

Minimizing anti-HA antibody interference during NASPA

To examine how NASPA performs with different NAs in the presence of anti-HA antibodies, we tested the H6 reassortant viruses carrying the vaccine strain NAs (Fig. 3a) with the strain specific ferret antisera that were generated using these same viruses. In the absence of TX100, all three HA matched ferret antisera showed NA activity increases that varied with each virus (Fig. 3d, upper panels), likely due to the presence of antibodies against the HA (e.g., H6) in the reassortant viruses. However, when TX100 was present only the ferret antisera raised against the test viruses blocked the NAi-MAb and significantly increased NA activity (Fig. 3d, lower panels), indicating that anti-HA antibodies can likely interfere with the NAi-MAb binding to NA in an intact virus. Following this observation, we titrated the TX100 amount with the same viruses and found that final concentrations between 0.025–0.1% TX100, or 2–10 times the TX100 critical micelle concentration (CMC), prevented the anti-H6 antibody interference and produced similar results (Supplementary Fig. 4). We then applied these parameters and obtained similar NASPA results with viral and recombinant NAs (Supplementary Fig. 5), further supporting that these conditions, including detergent, limit potential interference from anti-HA antibodies.

Optimization of NAi-MAb concentrations, NA amounts, and substrate incubation times

NASPA titers are dependent on the ability of antibodies in sera to block the binding of the NAi-MAb. Therefore, we measured NASPA titers for the three vaccine NAs in the presence of increasing NAi-MAb concentrations to determine the functional range for the inhibitor. Although the minimal NAi-MAb concentrations needed to obtain a NASPA titer varied between the NAs, we did obtain consistent NASPA titers for all three NAs when the NAi-MAb amounts were above the IC₅₀ (Supplementary Fig. 6), indicating NASPA titers are relatively insensitive to changes in the NAi-MAb concentrations above the IC₅₀.

Lowering the NA amounts in NASPA can potentially increase the sensitivity for samples with low anti-NA antibody quantities. Therefore, we measured NASPA titers with decreasing amounts of the three vaccine NAs after different MUNANA incubation times. These included kinetic measurements for 10 min after MUNANA addition and endpoint measurements 10 min and 2 h after MUNANA addition (Supplementary Fig. 7). As expected, NASPA titers from the kinetic and 10 min endpoint readings were similar, and the ability to detect titers was lost with the lower NA amounts (Supplementary Fig. 7d–f). At the 2 h endpoint, NASPA titers were more consistent across a wide range of NA amounts (e.g., 4–16 fold differences), indicating longer substrate incubation times with lower NA amounts can increase the consistency and sensitivity of NASPA results.

NASPA and ELLA comparison using MAbs against N1, N2, and type B NA

Since NAI titers are commonly measured with a fetuin-based ELLA approach^40,41,54, we compared NASPA and ELLA titers with a panel of MAbs in hybridoma media that we validated by ELISA for binding against N1, N2, and type B NA from recent vaccine strains (Supplementary Figure 8 and Table 2). Contrary to ELLA, which only shows NA activity decreases (Supplementary Fig. 8d–f), NASPA displayed activity increases and decreases with the MAbs against the type B NA (Fig. 4a). We attributed the expected activity increases to steric inhibition of the NAi-MAb, and the activity decreases suggested NASPA may also recognize MAbs that bind the active site and inhibit NA because it uses NAi-MAb amounts in the IC₅₀-IC₈₀ range. Supporting this conclusion, each MAb that showed decreased activity in NASPA also reduced the NA activity measured by MUNANA alone (Supplementary Fig. 10). With the MAbs against N1, two out of the five gave positive NASPA titers (Fig. 4b). For the NA from the H3N2 vaccine strain A/Darwin/9/2021 (Dar21), we did not identify an IC₅₀ for the NAi-MAb (FNI-9) as it was above 4 mg/ml (Supplementary Fig. 9).

NASPA data obtained with TX100-treated (a) B/Austria21, (b) H1N1/Vic19, and (c) H6N2/Dar21 vaccine strains and the indicated MAb hybridoma medium are displayed. Medium from duplicate hybridoma clones was analyzed, and a single representative of each is shown. The type B and H1N1 virus analysis used the NAI MAb FNI-9 at an ~IC₈₀, whereas the H3N2 virus analysis used a modified NAi-MAb (FNI-9 R27D) at an ~IC₈₀. Endpoint NA activities (RFU) were measured after a 10 min incubation with MUNANA at 37 °C. Gray regions indicate the mean (dashed line) ± 3 SDs (dotted line) from control reactions containing serially diluted hybridoma medium. d Correlation plot of NASPA and ELLA titers obtained with the same viruses and MAb hybridoma media. Pearson’s correlation coefficient (R) for the linear regression (dotted line) of the 27 MAb hybridoma media is shown. Negative (active site binding) NASPA titers were treated as positive for the correlation, and values below the limit of detection were assigned 0.

We overcame this problem by introducing a structure-guided substitution (R27D in the FNI-9 light chain). This substitution increased the NAi-MAb binding affinity by more than 10-fold (Supplementary Fig. 9), likely by complementing the charge of residue 344 K in Dar21 that was commonly a negatively charged residue (E) in NAs from earlier H3N2 strains. With this modified NAi-MAb we observed positive NASPA titers with three of the five N2 MAbs (Fig. 4c). We then plotted the NASPA titers versus the ELLA titers and observed a good linear correlation (R = 0.9), although the NASPA titers were generally lower (Fig. 4d and Table 2). We also noted that 24 out of the 27 MAbs showed the same positive and negative results and that the three which differed gave low ELLA titers (Fig. 4a and Table 2). Together, these results show that NASPA provides similar data with individual antibodies as ELLA with the added advantage that it can also differentiate steric and active site-binding NAI antibodies in a shorter amount of time. While the need to modify the NAi-MAb upon rare evolutionary changes to the active site can introduce a reagent lag time, any need for an update is likely to reflect an antigenic change in NA.

NASPA can provide antigenic data for NAs from type A H1N1 and type B strains

Monitoring NA antigenic drift in circulating strains is an ideal application for NASPA as it is insensitive to the presence of antibodies against HA. We tested this possibility using a panel of recent H1N1 vaccine strains from 2007 to 2019 which contained several strains we previously analyzed by ELLA³⁵. After titrating the viruses with MUNANA and determining the NAi-MAb IC₅₀ for each strain (Supplementary Fig. 11a), we set up plates to individually analyze each strain with the five ferret antisera raised against H6 reassortant viruses carrying the same NAs. The NASPA graphs indicated that the NA from the H1N1 strain A/Brisbane/59/2007 (BR07) is antigenically distinct from the NAs of post-pandemic H1N1s with slight similarity to the NA from the original 2009 H1N1 pandemic strain CA09 (Fig. 5a). In contrast, NAs from the H1N1 strains A/Michigan/45/2015 (MI15), A/Brisbane/02/2018 (BR18) and A/Victoria/2570/2019 (Vic19) all showed high antigenic similarity that was slightly related to the NA from H1N1/CA09. This conclusion was supported by the NASPA titers which showed the same trends as the previously reported ELLA titers (Table 3 and ref. ³⁵).

a NASPA results for each TX100-treated H1N1 vaccine strain with ferret antisera raised against the indicated reassortant virus are shown. The H1N1 strains A/Brisbane/59/2007 (BR07), A/California/07/2009 (CA09), A/Michigan/45/2015 (MI15), A/Brisbane/02/2018 (BR18) and A/Victoria/2570/2019 (Vic19) were analyzed. b NASPA results for each TX100-treated type B vaccine strain with ferret antisera raised against the indicated wild type and reassortant virus are displayed. The B Yamagata lineage strain B/Phuket/3073/2013 (Phu13) and the B Victoria lineage strains B/Brisbane/60/2008 (BR08), B/Colorado/06/2017 (CO17), B/Washington/02/2019 (WA19) and B/Austria/1359417/2021 (Austria21) were tested. All analysis were performed in duplicate using NAi-MAb concentrations corresponding to the IC₈₀ and are shown as the mean ± SD of the NA activities measured for 10 min after MUNANA addition. Gray regions indicate the mean (dashed line) ± 3 SDs (dotted line) of the control wells.

NAs from type B vaccine strains have received little attention due to the difficulties in generating antisera that do not recognize HA in the test virus. Therefore, we analyzed several type B vaccine strains using ferret antisera raised against viruses that carried three of the type B NAs. After titrating and determining the NAi-MAb IC₅₀ for each virus (Supplementary Fig. 11b), we performed the analysis. The NASPA results showed that the NA in the Yamagata-lineage strain B/Phuket/3073/2013 (Phu13) is antigenically distinct from the four Victoria-lineage strains (Fig. 5b and Table 4). Within the Victoria lineage, results with B/Brisbane/60/2008 (BR08) antisera showed progressively lower reactivity by year of virus isolation, indicating that the NA antigenicity has incrementally changed from BR08-like to B/Austria/1359417/2021-like (Austria21). However, the incremental change was less apparent with the Austria21 antisera. These observations suggest that the cumulative changes near the active site of the NAs from the Victoria lineage led to the recognition of a new dominant epitope that is more cross reactive with earlier strains, and we speculate that the switching of dominant epitopes may contribute to NA evolution in type B viruses.

Adult human sera commonly possess steric and active site-binding NAI antibodies

NASPA can theoretically be coupled with a simple activity assay like MUNANA to effectively measure steric and active site-binding NAI antibody responses in human sera. As a pilot test, we obtained previously analyzed sera samples from an H1N1 human challenge study²⁸ and performed NASPA and a MUNANA analysis in a blinded fashion using the H1N1 challenge virus and a H6 reassortant carrying the same NA. With both test viruses steric and active site NAI antibodies were readily detected, and the latter was confirmed to be even more prevalent by the MUNANA analysis (Fig. 6a and Supplementary Fig. 12). Unblinding the results revealed that the combined steric and active site NAI titers observed with MUNANA aligned well with the previously obtained ELLA titers²⁸. Interestingly, the combined results also showed that some individuals possess either steric or active site NAI antibodies whereas others possessed both and these profiles showed some changes post-challenge (Fig. 6a, b).

a Endpoint titers determined with NASPA and a MUNANA activity assay are displayed for the indicated patients pre-challenge (upper panel) and 2 months post-challenge (lower panel) along with the ELLA titers previously reported for these samples²⁸. Data was obtained using a TX100-treated H6 reassortant virus grown in eggs that contained the NA from the H1N1/CA09 challenge strain b Pre-challenge (upper panel) and post-challenge (lower panel) correlation plots showing the steric and active site NAI endpoint titers that were obtained from each serum with NASPA and MUNANA, respectively. Limits of detections (LODs) are shown by dotted lines. Overlapping dots were offset, and values below the LOD were assigned 0. c, d Post-challenge sera that showed strong (c) steric or (d) active site NAI antibody responses were depleted of immunoglobulin with protein A and G agarose or control agarose beads prior to performing NASPA or MUNANA analyses. Results are from one of two independent experiments. Assays were run with TX100-treated H6 reassortant virus using NAi-MAb concentrations corresponding to the IC₈₀. NA endpoint activities were measured 2 h after MUNANA addition. Gray regions indicate the mean (dashed line) ± 3 SDs (dotted line) of the control wells.

The prevalence of active site NAI antibodies was quite unexpected. Therefore, we questioned if the activity of this labile and Ca²⁺ sensitive enzyme² was reduced by other factors than antibody binding. To examine this possibility, we depleted the immunoglobulin from the strongest post-challenge sera samples using a combination of protein A and G agarose and repeated the NASPA and MUNANA analyses (Fig. 6c, d). The activity changes measured by NASPA and MUNANA in all cases were reduced following protein A and G incubation, and in most they were completely lost, confirming the results were largely antibody-mediated. We also noted that some sera showed reduced signals after incubation with the control agarose beads likely due to nonspecific binding. These results indicate that NAI antibody responses in adults are complex and function by two different mechanisms of NA inhibition that can be differentiated by coupling NASPA with a simple MUNANA activity assay.

NASPA can utilize large “bulky” synthetic NA inhibitors

Although NASPA with the NAi-MAb provided reproducible results in a time efficient manner, we sought to improve it further by replacing the NAi-MAb with a large bulky synthetic inhibitor which can potentially bind different NAs with similar affinities. For this analysis, we took advantage of a recently synthesized biotinylated sialoside-based NA inhibitor (4-guanidino-Neu5Acα2–3 Galβ1–4GlcNAc–βProNH-PEG4-Biotin) that was shown to bind NAs with nanomolar affinity⁵⁵ and was also inhibited by ferret antisera in an ELISA format (Supplementary Fig. 13). Based on these results, we reasoned that the synthesized NAi size could be increased by mixing it with streptavidin (SA) and the mixture could be used for NASPA (Fig. 7a, b). After mixing with streptavidin (1:2 molar ratio, NAi + SA) the IC₅₀ of the synthesized NAi for N1 slightly increased and the equilibrium was reached within 10–30 min (Fig. 7c and Supplementary Fig. 14a), compared to ~3 h for the NAi-MAb. We then tested the compound in NASPA using the H1N1/BR18 virus and only observed NA activity increases with the ferret antisera raised against H6N1/BR18 and the NAi + SA mixture (Fig. 7d). Somewhat unexpected, the NASPA titer of this ferret antisera with the NAi + SA mixture (~160) was identical to the titer achieved with the NAi-MAb (Table 3), indicating the biotinylated sialoside-streptavidin complex is suitable for NASPA.

a Diagram showing the novel chemoenzymatically synthesized biotinylated NA inhibitor (NAi: 4-guanidino-Neu5Acα2–3 Galβ1–4GlcNAc–βProNH-PEG4-Biotin) that was combined with streptavidin (SA) for use in NASPA. b NASPA schematic showing how the synthetic NA inhibitor bound to streptavidin (NAi + SA) can detect steric (purple) or active site (red) NAI antibodies. c IC₅₀ values are displayed for the NAi, NAi + SA, NAi-MAb (FNI-9), and zanamivir after the indicated incubation times with TX100-treated H1N1/BR18 virus. d Streptavidin binding to the NAi is required for NASPA. NA activity data is displayed from NASPA performed using TX100-treated H1N1/BR18 virus with the NAi, NAi + SA, or SA and the indicated ferret antisera. Assays were performed in duplicate using amounts approximating the IC₆₅ and are displayed as the mean ± SD. Gray region indicates the mean (dashed line) ± 3 SDs (dotted line) of the mock control wells. e IC₅₀ values of the NAi, NAi + SA, and the NAi-MAbs (FNI-9 and FNI-9 R27D) for the indicated H3N2 vaccine strains are displayed. Each assay was run twice using either a 30 min incubation (NAi and NAi + SA) or 3 h incubation (NAi-MAbs) at 37 °C. f Representative NA activity data set is shown from NASPA that was performed using TX100-treated H3N2/Dar21 virus with the NAi + SA mixture and the indicated purified N2 MAbs. Purified N2 MAbs were diluted as indicated from a starting concentration of ~20 μg/ml. The assay was performed using NAi + SA amounts approximating the IC₈₀. Gray region indicates the mean (dashed line) ± 3 SDs (dotted line) of the mock control wells.

We previously modified the NAi-MAb (FNI-9) to improve the affinity for the NA from a recent H3N2 virus, so we measured the IC₅₀ of the synthesized NAi ± SA for several recent H3N2 viruses. In contrast to the NAi-MAb (FNI-9), both the NAi and NAi + SA mixture showed consistent IC₅₀ values suggesting it could be used at a set concentration for NASPA (Fig. 7e). Finally, we tested the synthetic NAi using purified N2 MAbs. With the NAi + SA mixture, steric inhibition was observed with three of the purified MAbs, and another showed an activity decrease indicative of active site binding. As these results were somewhat different than those from the hybridoma media with unknown antibody concentrations (Fig. 4c), we also analyzed the purified N2 MAbs using NASPA with modified NAi-MAb and observed a similar pattern (Supplementary Fig. 14b). We did note the activity increases from two of the MAbs were more pronounced, likely due to the different spatial footprints of the NAi + SA and the NAi-MAb. Interestingly, the active site binding N2 MAb (13F10), which was confirmed with a MUNANA assay (Supplementary Fig. 14c), could potentially serve as an effective NAi-MAb for NAs from recent H3N2 viruses. Altogether, these data demonstrate that NASPA is an efficient and cost-effective platform that can utilize different large ‘bulky’ NA active site inhibitors to profile NA antigenicity and NAI antibody responses, making it well suited to aid the development of new influenza vaccines containing NA.