Influenza viruses possess two surface glycoproteins, haemagglutinin and neuraminidase (NA). Although haemagglutinin plays a major role as a protective antigen, immunity to NA also contributes to protection. The NA protein consists of a stalk and a head portion, the latter of which possesses enzymatic NA (or sialidase) activity. Like haemagglutinin, NA is under immune pressure, which leads to amino acid alterations and antigenic drift. Amino acid changes accumulate around the enzymatic active site, which is located at the top of the NA head. However, amino acid alterations also accumulate at the lateral surface of the NA head. The reason for this accumulation remains unknown. Here, we isolated seven anti-NA monoclonal antibodies (mAbs) from individuals infected with A(H1N1)pdm09 virus. We found that amino acid mutations on the lateral surface of the NA head abolished the binding of all of these mAbs. All seven mAbs activated Fcγ receptor (FcγR)-mediated signalling pathways in effector cells and five mAbs possessed NA inhibition activity, but the other two did not; however, all seven protected mice from lethal challenge infection through their NA inhibition activity and/or FcγR-mediated antiviral activity. Serological analysis of individuals infected with A(H1N1)pdm09 virus revealed that some possessed or acquired the anti-NA-lateral-surface antibodies following infection. We also found antigenic drift on the lateral surface of the NA head of isolates from 2009 and 2015. Our results demonstrate that anti-lateral-surface mAbs without NA inhibition activity can provide protection by activating FcγR-mediated antiviral activity and can drive antigenic drift at the lateral surface of the NA head. These findings have implications for NA antigenic characterization in that they demonstrate that traditional NA inhibition assays are inadequate to fully characterize NA antigenicity.
Influenza A virus harbours two major glycoproteins on its envelope: haemagglutinin (HA) and neuraminidase (NA). HA and NA play pivotal roles in both the early and late stages of the virus replication cycle. HA binds to cellular receptors (sialic acid) on the cell surface to initiate entry1,2, whereas the enzymatic activity of NA cleaves off the sialic acid, allowing the release of progeny viruses from the cell surface3,4. The enzymatic activity of NA also contributes to virus entry by removing receptor decoys that present in the airways5.
Antibodies against HA and NA are elicited following influenza virus infection of individuals6,7,8,9. Antibodies against the head region of HA inhibit virus attachment to the receptor primarily by preventing receptor binding10 and those against the stem region of HA inhibit viral fusion or egress11,12,13. Although antibodies to HA are typically considered the mediators of protection from influenza virus infection, some studies have indicated the importance of anti-NA antibodies for protection9,14,15,16,17. Among the NA-specific monoclonal antibodies (mAbs), some restrict virus spread by interfering with the NA activity of the virus, resulting in the aggregation of progeny virions on the cell surface. Such NA-inhibiting (NI) antibodies are known to reduce the viral loads and symptoms in infected mice, ferrets and humans14,15,18,19,20,21. Although the major antiviral function of anti-NA antibodies seems to be their NI activity, one study reported that for in vivo protection, a broadly reactive NI mAb required interactions between an Fc region of IgG and Fcγ receptors (FcγRs; Fc–FcγR) to activate effector cells, such as natural killer cells, macrophages and neutrophils22. Natural killer cells express FcγRIIIa23, and macrophages and neutrophils express FcγRIIa24,25. Activated natural killer cells, macrophages and neutrophils suppress virus propagation through antibody-dependent cellular cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP) and antibody-dependent neutrophil-mediated phagocytosis, respectively26,27. This study suggests that anti-NA mAbs could trigger the FcγR-mediated activation of effector cells to promote in vivo protection. These functions have been studied by analysing anti-NA mAbs that possess NI activity and recognize epitopes around the enzymatic centre of the NA head. However, the contribution of anti-NA antibodies that recognize regions other than those near the enzymatic centre remains unknown.
To evade neutralizing antibodies, influenza viruses with amino acid changes in the epitopes of the major surface proteins (HA and NA) emerge (antigenic drift)28,29,30. Similar to the antigenic drift of HA, the antigenicity of NA changes as a result of amino acid mutations in the NA head that allow escape from anti-NA antibodies31,32. Several amino acid changes that allow viruses to escape from NI antibodies have been mapped around the enzymatic centre of the NA head33. However, amino acid substitutions also accumulate at the lateral surface of the NA head, which is far from the enzymatic centre33; the reason why mutations accumulate in this region remains unknown.
Here, we obtained seven mAbs that recognize the lateral surface of the NA head from two individuals naturally infected with A(H1N1)pdm09 virus during the 2015–2016 influenza season and characterized these mAbs to understand their properties, particularly their roles in in vivo protection and antigenic drift.
Isolation of human mAbs that recognize A(H1N1)pdm09 virus
Peripheral blood mononuclear cells (PBMCs) were obtained from two human volunteers who were infected with A(H1N1)pdm09 virus during the 2015–2016 influenza season. The PBMCs were fused with SPYMEG cells to generate hybridomas that express human antibodies. After several rounds of screening by enzyme-linked immunosorbent assays (ELISA) using purified A/California/04/2009 (CA/04/09) and A/Yokohama/94/2015 (YO/94/15) viruses, which were isolated during the first wave of the 2009 pandemic and in the 2015–2016 season, respectively, and classified into the genetic clades 1 and 6B.1 based on their respective HA sequences, hybridomas expressing antibodies that bound to the YO/94/15 but not to the CA/04/09 virus were biologically cloned. Of the 701 hybridomas screened, we obtained seven clones (DA05C23, HP02C70, DA05B02, HP02B69, HP02E74, HP02E63 and HP02B24) that expressed a mAb specific to the YO/94/15 virus. Analysis of the nucleotide sequences of their variable regions revealed that all seven mAbs used the same VH gene (IGHV3–15*01), although the CDR3 sequence of each heavy chain varied (Supplementary Table 1). Five mAbs used IGLV3–1*01, whereas HP02B24 and HP02E63 used IGLV3–21*01 and IGLV6–57*01, respectively.
The seven human mAbs mainly recognize the NA of A(H1N1)pdm09 virus isolated during the 2015–16 influenza season
To determine the breadth of recognition of the seven mAbs that we obtained, we performed an ELISA using A(H1N1)pre2009, A(H3N2) and YO/94/15 viruses. None of the mAbs recognized either A(H1N1)pre2009 or A(H3N2) virus (Fig. 1a). A broadly reactive human anti-N1-NA mAb clone 1000–3C05 (refs. 8,22) bound to the A(H1N1)pre2009 and YO/94/15 viruses, whereas the broadly reactive human anti-HA stem mAb clone CR9114 (ref. 34) recognized all three viruses tested (Fig. 1a). A negative control, anti-B HA mAb clone 1430E3/9, did not bind to any of these viruses. These results indicate that the seven mAbs isolated here are specific to A(H1N1)pdm09 virus. Next, we performed an ELISA using A(H1N1)pdm09 viruses isolated during different influenza seasons. HP02C70, DA05B02, HP02E74, HP02E63 and HP02B24 showed specific binding to the YO/94/15 virus, whereas HP02B69 and DA05C23 bound preferentially to the YO/94/15 virus and weakly to the A/Osaka/6/2014 virus (Fig. 1b). EM-3C02, which was reported to be a strain-specific human anti-N1-NA mAb8,22, bound to the CA/04/09 and A/Hiroshima/66/2011 viruses, whereas the broadly reactive mAb 1000–3C05 recognized all of the viruses tested. The mAb clone 1430E3/9 failed to recognize any of the viruses tested. These results indicate that the seven mAbs obtained are not broadly reactive but mainly recognize the YO/94/15 virus.
To determine which viral protein was recognized by these mAbs, we performed an ELISA using purified reassortant viruses that had HA and NA derived from either the YO/94/15 or the CA/04/09 virus. All seven mAbs bound to viruses possessing NA derived from YO/94/15 virus, whereas EM-3C02 recognized a virus possessing NA derived from CA/04/09, but not YO/94/15, virus (Fig. 1c). All of the viruses tested were detected by 1000–3C05. These results demonstrate that these seven human mAbs target the NA of A(H1N1)pdm09 virus.
We next determined the binding affinity of the seven mAbs to NA derived from the YO/94/15 virus by Scatchard analysis (Supplementary Fig. 1). DA05C23 and HP02C70 possessed high binding affinities (KD = 0.86 and 0.73 nM, respectively), which were comparable to that of 1000–3C05 (KD = 0.82 nM). However, DA05B02, HP02B69, HP02E74, HP02E63 and HP02B24 showed relatively lower binding affinities than that of 1000–3C05 (KD = 2.57, 3.24, 4.18, 3.22 and 3.62 nM, respectively).
The seven mAbs recognize the lateral surface of the NA head
To determine which amino acids are important for the binding of the seven anti-NA mAbs, we compared the amino acid sequences of the NA protein from the six viruses used in Fig. 1b and identified six unique amino acids in the NA of the YO/94/15 virus (Fig. 2a; 13I, 44I, 263I, 268K, 314M and 390K; N2 numbering). We then generated mutant YO/94/15 viruses possessing single amino acid substitutions (I13V, I44L, I263V, K268N, M314I or K390N) in their NA and examined the binding of the seven mAbs in an ELISA. HP02C70, DA05B02, HP02E74, HP02E63 and HP02B24 did not bind to virus possessing the K390N substitution in its NA but did bind to viruses with the I13V, I44L, I263V, K268N or M314I substitutions (Fig. 2b). DA05C23 and HP02B69 recognized all of the NA mutant viruses with single amino acid substitutions but showed decreased reactivity to the K390N mutant virus. As these two clones bound weakly to the A/Osaka/6/2014 virus (see Fig. 1b), we selected three amino acids that were shared by only the A/Osaka/6/2014 and YO/94/15 viruses (that is, 38V, 321V and 432E) for the generation of additional mutants. We generated mutant viruses possessing two substitutions (V38I and K390N, V321I and K390N, and E432K and K390N) in their NA and tested mAb binding. DA05C23 and HP02B69 did not bind to the virus that had both the K390N and E432K mutations in the NA but still bound to the viruses possessing V38I and K390N or V321I and K390N (Fig. 2b). These results suggest that the amino acid at position 390 and the amino acids at positions 390 and 432 are important for the recognition of five and two of the mAbs, respectively. We therefore mapped positions 390 and 432 together with the NA catalytic residues on the NA head molecule (Fig. 2c). Position 390 is located at the lateral face of the NA head, whereas position 432 is located at the edge of the enzymatic centre on the upper surface of NA (Fig. 2c). The distance between positions 390 and 432, based on the crystal structure of the NA head molecule, is approximately 3 nm. Given that an antibody footprint is typically 4‒10 nm2 (refs. 35,36), both amino acids are likely to be covered by one mAb. We therefore suggest that the five anti-NA mAbs that recognize an epitope around position 390 mainly recognize the lateral surface of NA and that the other two anti-NA mAbs that lose binding to NA with mutations at positions 390 and 432 recognize an epitope that spans the upper and lateral surfaces.
The seven anti-NA mAbs differ in their NI activity
To further characterize the seven anti-NA mAbs, we assessed their ability to inhibit viral sialidase activity with an enzyme-linked lectin assay (ELLA). DA05C23, HP02C70 and DA05B02 strongly inhibited NA activity with 50% inhibition concentration (IC50) values of 0.38, 0.60 and 0.73 nM, respectively (Fig. 3a). HP02B69 weakly inhibited NA activity with an IC50 value of 7.52 nM, which was similar to that of 1000–3C05 (IC50 = 12.05 nM). HP02E74 slightly inhibited NA activity at high concentrations (IC50 ≥ 26.67 nM), whereas HP02B24 and HP02E63 possessed no NI activity. To further confirm the NI activity in vitro, we performed a plaque-size reduction assay. In this assay, DA05C23, HP02C70 and DA05B02 decreased the viral-plaque size by at least 50% at their highest concentrations (Fig. 3b). The plaques that formed in the presence of HP02B69 and HP02E74 were about 30% smaller than those that formed in the absence of a mAb. In contrast, the plaque-size reduction induced by HP02B24, HP02E63 and 1000–3C05 was approximately 10% for all of the concentrations tested. Together, these results indicate that DA05C23, HP02C70 and DA05B02 possess relatively high NI activity, HP02B69 and HP02E74 possess relatively low NI activity, and HP02B24 and HP02E63 fail to inhibit NA activity.
The seven anti-NA mAbs activate the FcγR-mediated signalling pathway in effector cells
We next investigated whether these anti-NA mAbs elicit ADCC or ADCP with ADCC and ADCP reporter bioassays. Using these reporter bioassays, we found that the three mAbs that possess high NI activity (HP02C70, DA05C23 and DA05B02) efficiently activated the ADCC and ADCP signalling pathways in effector cells co-cultured with both virus-infected and NA-expressing cells, except for the ADCC reporter assay with DA05C23 and NA-expressing cells, which was moderately activated (Fig. 3c). HP02B69 and HP02E74, which possessed low NI activity, moderately activated the ADCC signalling pathway and efficiently activated the ADCP signalling pathway in the effector cells co-cultured with infected and NA-expressing cells. HP02B24 and HP02E63, which had no NI activity, induced low or similar ADCC and ADCP signalling activation to that induced by the other five mAbs. The mAb 1000–3C05, which protected mice from lethal infection by activating FcγR-mediated effector cells22, triggered ADCC and ADCP signal activation via both virus-infected and NA-expressing cells. All of the mAbs that possessed the N297Q mutation, which abolishes the interaction with FcγRs, and 1430E3/9 caused little-to-no activation in both reporter assays. These results indicate that our seven anti-NA mAbs have the potential to activate ADCC and ADCP.
The anti-NA mAbs protect mice from lethal infection via their NI and FcγR-mediated antiviral activities
As our mAbs possessed various levels of NI activity and activated the FcγRIIIa- and FcγRIIa-mediated signalling pathways to various extents, we examined their in vivo protective efficacy against lethal influenza virus infection. For these in vivo tests, we monitored the body weight loss and survival of mice infected with 10× the 50% mouse lethal dose (MLD50) of a challenge virus after the administration of the wild-type or mutant N297Q mAb at 30 mg kg−1 (the results of the administration of 5 and 1 mg kg−1 are shown in Supplementary Fig. 2). The positive and negative controls used were 1000–3C05, which requires FcγR-mediated effector cell activation to provide in vivo protection22, and 1430E3/9, respectively. All of the wild-type anti-NA mAbs, except DA05B02, protected all of the mice from lethal challenge infection, with the mice experiencing transient weight loss (Fig. 4). Among the groups administered with mutant N297Q mAbs, 75% of the mice that received either HP02C70 or DA05B02, which possessed high NI activity, survived (Fig. 4). In particular, the N297Q mutant mAb DA05B02 showed similar in vivo protective potency to that of its wild-type counterpart. The N297Q mutants DA05C23, HP02B69 and HP02E74, which possessed NI activity, protected 25% of the mice that experienced severe weight loss. The N297Q mutants HP02E63 and HP02B24, which showed no NI activity, failed to protect any mice. Wild-type 1000–3C05 showed protection with transient weight loss, whereas none of the mice that received the N297Q mutant of 1000–3C05 survived. The mice that received 1430E3/9 experienced similar weight loss to the mice that received PBS. These results suggest that both the NI activity and FcγR-mediated antiviral activity of the anti-NA mAbs contribute to protection against influenza virus infection. Moreover, FcγR-mediated antiviral activity played a central role in the in vivo protection provided by the anti-NA mAbs that lacked NI activity.
Antibodies that recognize the lateral surface of the NA head of A(H1N1)pdm09 virus in patient sera
To evaluate whether individuals infected with the A(H1N1)pdm09 virus had antibodies targeting the lateral surface of the NA head on the day they visited the hospital and one month after infection, we performed a competitive-binding ELISA using a protective mAb without NI activity (HP02E63 or HP02B24) or a cocktail of our seven anti-NA mAbs (Fig. 5a). The Fc region of these mAbs was replaced with that of mouse IgG2a to prevent recognition by the horseradish peroxidase (HRP)-conjugated antibody. We assessed antibody binding in the presence of the mAb cocktail or individual mAbs of the serum samples of patients infected with the A(H1N1)pdm09 virus during the 2015‒2016 influenza season to virus-like particles (VLPs) displaying the NA from YO/94/15 (YO/94/15-NA); the NA-displaying VLPs were produced by expressing NA and the VP40 protein from Ebola virus, as humans do not have antibodies to VP40. All of the patients were found to possess antibodies that competed with the anti-NA mAb cocktail 30 d after infection (Fig. 5b). The serum samples collected 30 d after infection from patients HP001, HP002, HP014 and HP021, and from HP002 and HP021 contained significant amounts of antibodies that competed with the antibodies HP02E63 and HP02B24, respectively (Fig. 5c,d). On the day of the visit to the hospital, patients HP001 and HP021 had antibodies that competed with the anti-NA mAbs cocktail, HP02E63 and/or HP02B24 (Fig. 5). These results demonstrate that substantial amounts of antibodies targeting epitopes on the lateral surface of the NA head were induced following virus infection.
NA antigenicity differs between the 2009 and 2015 isolates of the A(H1N1)pdm09 virus
Influenza viruses constantly evade suppressive antibodies by acquiring mutations in the epitopes of not only HA, but also NA. We therefore investigated the antigenicity of CA/04/09-NA and YO/94/15-NA in an ELISA using 22 serum samples that were collected during the 2009‒2010 influenza season from volunteers who had neutralization antibodies against the A(H1N1)pdm09 virus37. We found that the end-point titres to YO/94/15-NA were significantly lower than those to CA/04/09-NA (Fig. 6a). These results indicate that YO/94/15-NA differed antigenically from CA/04/09-NA.
To determine which amino acid mutations are responsible for the antigenic change in NA, we compared the amino acid sequences of CA/04/09-NA and YO/94/15-NA and found 6, 4 and 3 amino acid differences around the enzymatic centre of the NA head (positions 199, 240, 247, 321, 372 and 432), on the lateral surface of the NA head (positions 263, 268, 314 and 390) and in the stalk region (not shown in the structure; positions 38, 44 and 48), respectively (Figs. 2a and 6b). We then performed a cell-based ELISA using the 22 A(H1N1)pdm2009-positive sera and cells transiently expressing wild-type CA/04/09-NA or its mutants that possessed amino acid changes found in YO/94/15-NA. We concluded that a mutation affected the antigenicity of NA if the end-point titre of each serum against the mutant NA was at least 4× lower than that against the wild-type NA. Half of the sera (11/22; blue dots) showed reduced reactivity to NA with mutations around the enzymatic centre (Fig. 6c, left). Two of these 11 sera also showed decreased binding to the stalk region mutant (Fig. 6c, right). Two serum samples (2/22) mainly targeted the lateral surface of the NA head (red and orange dots in Fig. 6c, middle) and reacted with the NA mutants of the enzymatic centre and stalk region with similar titres. These results indicate that the antigenic drift of NA involves mutations on the lateral surface of the NA head and that the antibodies of some individuals mainly recognize the lateral surface of the NA head.
Furthermore, we investigated whether a single mutation at position 390, which is the key residue for the binding of our mAbs, contributes to the antigenic drift on the lateral surface of the NA head using a cell-based ELISA. Cells transiently expressing CA/04/09-NA with the N390K mutation as the antigen were incubated with the 22 A(H1N1)pdm2009-positive human sera used in Fig. 6c. We found that one sample (red dot) also showed decreased binding to the N390K mutant (Fig. 6d). These results suggest that the single N390K mutation is partially involved in the NA antigenic drift.
Antibodies that inhibit virus sialidase activity are produced in infected individuals and are important for the protection against seasonal influenza virus infection20,21,38. To evade such NI antibodies, amino acid mutations are accumulated around the enzymatic centre of NA33,39. Although, over time, many amino acid changes accumulate on the lateral surface of the NA head, which is far away from the enzymatic centre, the role of these mutations remained unknown33. Here, we isolated human mAbs that recognize the lateral surface of the NA head of A(H1N1)pdm09 virus isolated during the 2015‒2016 influenza season. These anti-NA-lateral-surface mAbs protected mice from lethal virus infection via their NI and/or FcγR-mediated antiviral activities. We also found that some individuals acquire anti-NA-lateral-surface antibodies after infection and that the antigenicity of the lateral surface of the NA head has changed during the replication of the A(H1N1)pdm09 virus in humans since 2009. Together, the findings of this study suggest that anti-NA antibodies similar to our anti-NA antibodies are likely to be involved in the antigenic drift on the lateral surface of the NA head. To determine whether amino acids change frequently at the lateral surface of the NA head, we compared the NA sequences derived from the A(H1N1)pdm2009, A(H1N1)pre2009 and A(H3N2) viruses, which were obtained from the Global Initiative on Sharing All Influenza Data database, and then calculated the Shannon entropy to determine the sequence variability. The calculated Shannon entropy was then mapped onto the NA structure (Supplementary Fig. 3). This analysis revealed that amino acid mutations at the lateral surface of the NA head occur in all subtypes of human virus NA at a frequency equal to or higher than that around the enzymatic centre. Although further epitope analysis is required to fully understand this phenomenon, the selective pressure of anti-NA-lateral-surface mAbs due to FcγR-mediated antiviral activity as well as NI activity probably contributes to the antigenic drift at the lateral surface.
Recent studies report that, in vaccinated humans, the levels of anti-NA antibodies correlate with a reduced disease severity score, decreased symptoms and even a reduction in the duration of virus shedding20,40, thus indicating that anti-NA antibodies are important mediators of protection against influenza virus infection. Attention has therefore been paid to the importance of NA in influenza vaccines9,16,17. In this context, the antigenic characterization of NA is essential to select appropriate vaccine strains, as has been done with HA. Traditionally, the antigenic properties of NA are evaluated using the NI assay, which mainly detects antigenic changes caused by amino acid changes around the enzymatic active centre and might not detect some changes in the lateral surface of NA. Our results demonstrate that anti-lateral-surface mAbs lacking NI activity are potentially involved in the antigenic drift of NA, which suggests that the traditional NI assays are not sufficient to fully characterize NA antigenicity. To this end, it is vital that we develop a new assay to properly characterize NA antigenicity.
In a study of human anti-NA antibodies, DiLillo et al.22 reported that the broadly reactive mAb clone 1000–3C05, which possesses low NI activity, requires FcγR-mediated effector cell activation to provide in vivo protection, whereas the strain-specific mAb EM-3C02, which possesses high NI activity, does not. Together with our results, these findings suggest that human anti-NA antibodies suppress virus pathogenicity by at least two distinct mechanisms: NI activity and FcγR-mediated effector cell activation. An anti-NA mAb with high NI activity could reduce virus spread in vivo mainly through its NI activity, whereas an anti-NA mAb without NI activity could suppress virus replication mainly through FcγR-mediated effector cell activation (for example, ADCC, ADCP and/or antibody-dependent neutrophil-mediated phagocytosis). Both NI activity and FcγR-mediated effector cell activation are required for anti-NA mAbs with low NI activity to reduce virus growth in vivo. Thus, the sum of the NI activity and the FcγR-mediated effector cell activation determines the potency of an anti-NA antibody in vivo.
Of the seven isolated anti-NA mAbs, binding to the NA protein of five clones (HP02C70, DA05B02, HP02E74, HP02E63 and HP02B24) was abolished by an amino acid substitution at position 390, which is located on the lateral surface, and that of two clones (DA05C23 and HP02B69) was eliminated by two amino acid mutations at positions 390 and 432, the latter of which is located on the upper surface. The amino acid at position 390 was essential for recognition. Of the former five mAbs, three (HP02C70, DA05B02 and HP02E74) showed NI activity, whereas the other two (HP02E63 and HP02B24) did not. The mouse mAb clone CD6, which recognizes an epitope located on the lateral surface of NA, possesses potent NI activity41,42. Co-crystallization revealed that CD6 could hinder NA activity by either preventing the sialic acid substrates from accessing the enzymatic centre (due to its size) or by cross-linking adjacent NA tetramers41. We hypothesize that our anti-NA mAbs (HP02C70, DA05B02 and HP02E74) may inhibit NA activity through a similar mechanism. However, we cannot explain why HP02E63 and HP02B24 do not possess NI activity despite that they share the key amino acid for binding. Co-crystallization of NA with these mAbs would help us to understand the lack of NI activity of such mAbs that recognize an epitope on the lateral surface of NA.
In summary, we found that anti-NA mAbs with low or no NI activity can contribute to protection from influenza virus infection. Furthermore, these mAbs are potentially involved in the antigenic drift of NA via FcγR-mediated antiviral activity. These findings suggest that the assessment of antigenic drift using the NI assay alone is insufficient and that methods that allow us to consider antigenic changes that cannot be detected using NI assays need to be developed for better selection of vaccine candidate strains.
All experiments were performed in compliance with the relevant ethical regulations. Human blood was collected from two volunteers by following a protocol approved by the Research Ethics Review Committee of the Institute of Medical Science of the University of Tokyo. Written informed consent was obtained from each participant. All experiments with mice were performed in accordance with the University of Tokyo’s Regulations for Animal Care and Use and were approved by the Animal Experiment Committee of the Institute of Medical Science of the University of Tokyo.
Madin–Darby canine kidney (MDCK) cells were maintained in Eagle’s minimal essential medium containing 5% newborn calf serum. Human lung carcinoma A549 cells were maintained in Ham’s F-12K medium containing 10% fetal calf serum (FCS). Human embryonic kidney 293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FCS. SPYMEG cells (MBL), which are a fusion partner cell line for human hybridoma, were generated by the fusion of the mouse myeloma cell line SP2/0 with the human leukaemia cell line MEG-01 and were maintained in DMEM containing 15% FCS43. These cells were incubated at 37 °C under 5% CO2. Expi293F cells (Thermo Fisher Scientific) were maintained in Expi293 expression medium (Thermo Fisher Scientific) at 37 °C under 8% CO2.
We used the following six A(H1N1)pdm09 viruses that have been reported previously44: CA/04/09 (clade 1), A/Hiroshima/66/2011 (clade 5), A/Osaka/83/2011 (clade 7), A/Osaka/33/2013 (clade 6C), A/Osaka/6/2014 (clade 6B) and YO/94/15 (clade 6B.1). A/Brisbane/59/2007 (H1N1pre2009) and A/Hong Kong/4801/2014 (H3N2) were also used. All viruses were propagated in MDCK cells.
Plasmid-based reverse genetics
The reassortant, NA mutant and challenge viruses were generated by plasmid-based reverse genetics as described previously45. For reassortant viruses, two pPol I plasmids encoding the HA and NA segments derived from either the YO/94/15 or CA/04/09 viruses and six pPol I plasmids encoding the remaining segments of the CA/04/09 virus were used for the plasmid transfection. For the preparation of NA mutant viruses, pPol I plasmids encoding HA and wild-type or mutated NA derived from YO/94/15 virus and six pPol I plasmids encoding the remaining segments of CA/04/09 virus were used. We used pPol I plasmids encoding the HA and NA segments of YO/94/15 virus and six pPol I plasmids encoding the remaining segments of mouse-adapted CA/04/09 for the preparation of the challenge virus46. These eight pPol I plasmids along with protein expression plasmids for PB2, PB1, PA and NP derived from A/Puerto Rico/8/34 were transfected into 293T cells using TransIT-293 (Mirus) according to the manufacturer’s instructions. At 48 h post-transfection, the supernatants were harvested and inoculated into MDCK cells. The rescued viruses were sequenced to ensure the absence of unwanted mutations. The primer sequences are available on request.
PBMCs were isolated from blood (30 ml) obtained from two volunteers who were infected with the A(H1N1)pdm09 virus in the 2015–2016 influenza season using Ficoll Paque Plus (GE Healthcare). The hybridomas that were generated by the fusion of PBMCs with SPYMEG cells were cultured and biologically cloned as described previously11. For the screening, we performed an ELISA using purified YO/94/15 and CA/04/09 viruses as antigens, as described below.
ELISA plates (96-well) were coated with each purified virus. After being blocked with Blocking One (1/5 dilution; Nakarai), the plates were incubated with either the culture medium of the hybridomas or 1 μg ml−1 purified mAb. A HRP-conjugated goat anti-human IgG, Fcγ-fragment-specific antibody (Jackson Immuno-Research) was used as the secondary antibody and the signal was developed using TMB (Thermo Fisher Scientific) as the substrate. The reaction was stopped with 2 N sulphuric acid and the A450 nm values were then measured. The A450 nm values of at least 0.1 were considered to indicate binding.
Expression and purification of monoclonal human IgG
The inhibition of NA activity was measured using an ELLA as described previously48. Briefly, two-fold serial dilutions of the test antibodies (4,000–7.8 ng ml−1) were mixed with a predetermined amount of purified YO/94/15 virus diluted in PBS containing 1% BSA and 0.05% Tween 20 (PBS–T). The mixture was transferred to 96-well plates coated with fetuin (Sigma) and then incubated for 16 h at 37 °C. The plates were washed with PBS–T and peroxidase-conjugated peanut agglutinin (Sigma) was then added to detect the galactose exposed by the removal of the sialic acids on fetuin. The plates were incubated at room temperature for 2 h in the dark and then washed with PBS–T before the addition of the substrate o-phenylenediamine dihydrochloride (Sigma). The reaction was stopped with the addition of 1 N sulphuric acid and the A490 nm values were read. The relative NI activity was calculated by dividing the A490 nm values of the test well by the A490 nm values of the ‘virus only’ well and multiplying by 100. The IC50 was determined nonlinear regression analysis (GraphPad Prism software) assuming that the molecular weight of the antibody was 150 kDa.
Plaque-size reduction assay
MDCK cells were infected with 100 plaque-forming units of YO/94/15 virus for 1 h at 37 °C. After the removal of the inoculum, the cells were washed with Eagle’s minimal essential medium containing 0.3% BSA and overlaid with agar containing the indicated amount of the test mAb and 1 μg ml−1 l-(tosylamido-2-phenyl) ethyl chloromethyl ketone-treated trypsin. The cells were incubated at 37 °C under 5% CO2 for 2 d and then stained with crystal violet solution to visualize the plaques. The plaque images were obtained by scanning the assay plates with a photo scanner (GT-X900, EPSON). The total plaque size of each well was measured using a script written in MATLAB (Mathworks) and the average plaque size was calculated by dividing by the total number of plaques. Infected cells overlaid with agar that lacked mAb served as a control. The reduction in plaque size was calculated by dividing the average plaque size in the presence of each mAb by the control plaque size.
ADCC and ADCP reporter assays
ADCC and ADCP reporter assays were performed using ADCC and ADCP Reporter Bioassay kits (Promega) according to the manufacturer’s instructions. The activation of ADCC and ADCP signalling in effector cells were measured using two kinds of target cells: A549 cells infected with the YO/94/15 virus and 293T cells expressing NA derived from the YO/94/15 virus. Briefly, A549 cells were plated at a density of 10,000 cells well−1 in a flat-bottom white 96-well plate. After 24 h, the cells were infected with YO/94/15 virus at a multiplicity of infection of 10. At 12 h post-infection, the cells were used as target cells. For the second target-cell type, 293T cells in a six-well plate were transfected with a plasmid encoding YO/94/15-NA using TransIT-293 2 d before the reporter assays were performed. The next day, the transfected cells were harvested and seeded at 20,000 cells well−1 into a white 96-well plate. The cells were used as target cells 24 h later. In the ADCC and ADCP reporter assays, the medium of the target cells was replaced with test mAb five-fold-diluted (10–0.016 μg ml−1) in the assay buffer (Promega). ADCC or ADCP bioassay effector cells (a Jurkat cell line stably expressing human FcγRIIIa or FcγRIIa, human CD3γ and an NFAT-response element driving expression of a firefly luciferase) were added to the antibody-treated target cells and incubated for 6 h at 37 °C. The firefly luciferase activity was then measured using luciferase assay regents (Promega).
In vivo protection test
Four six-week-old female BALB/c mice (Japan SLC) per group were intraperitoneally injected with PBS or the test antibody at 30 mg kg−1 in 250 μl PBS. After 24 h, the mice were anaesthetised with isoflurane and intranasally challenged with 10× MLD50 of the challenge virus in 50 μl PBS. The weight of the mice was monitored daily for 14 d and mice that lost 25% or more of their initial body weight were scored as dead and euthanized according to institutional guidelines.
Competitive binding assay
For the preparation of antigens, 293T cells were transfected with pCAGGS plasmids encoding the Ebola virus matrix protein VP40 and each indicated NA using TransIT-293. After 2 d, the supernatant of the transfected cells was harvested and centrifuged at 28,000 r.p.m. with a 20% sucrose cushion. The pellet containing the VP40-induced VLPs presenting NA was resuspended in PBS. ELISA plates were coated overnight at 4 °C with 5 μg ml−1 VLPs. After being blocked with Blocking One (1/5 dilution), these plates were incubated with a two-fold diluted cocktail of the seven anti-NA mAbs (DA05C23, HP02C70, DA05B02, HP02B69, HP02E74, HP02E63 and HP02B24), individual mAbs (HP02E63 and HP02B24) that possessed the Fc region of mouse IgG2a or PBS for 1 h. The plates were then washed with PBS–T followed by the addition of diluted serum samples. The serum samples used in this assay were collected from patients infected with A(H1N1)pdm09 virus during the 2015‒2016 influenza season. After 1 h, an HRP-conjugated goat anti-human IgG, Fcγ-fragment-specific antibody was added as the secondary antibody. The plates were incubated for 1 h and then washed with PBS–T before the addition of the TMB substrate. The reaction was stopped with 2 N sulphuric acid and the A450 nm values were then measured. The wells with added PBS (instead of mAb) served as the controls for each serum sample.
For each amino acid position of the aligned isolate sequences, an entropy value was computed using the formula –ΣPi × ln[Pi], as described by Chen and colleagues49. This formula follows the definition of Shannon entropy50 that has been used to evaluate diversity. In this study, entropy was used to measure the variability of amino acids at a given position, where i = 1–20 represents the 20 different amino acid residues and Pi represents the probability density of the respective residue. The entropy values range from 0 (only one residue present at that position) to 2.996 (all 20 residues are equally represented). The calculation was based on the NA sequences of 19,034 A(H1N1)pdm09 isolates, 4,626 A(H1N1)pre2009 isolates and 40,516 A(H3N2) isolates obtained from the Global Initiative on Sharing All Influenza Data database.
Cells (293T) were transfected with FLAG-tagged wild-type or mutant NA-expression plasmids, or an empty control plasmid using TransIT-293. At 48 h after the transfection, the cells were fixed with 4% paraformaldehyde in PBS for 30 min at room temperature. The cells were then washed with PBS–T and blocked with Blocking One (1/5 dilution) for 1 h. The cells were then incubated for 1 h with serial two-fold dilutions of serum samples, which were collected from volunteer schoolchildren and their parents who were seropositive for A(H1N1)pdm09 virus between November 2009 and March 2010 (ref. 37), followed by incubation for 1 h with an HRP-conjugated goat anti-human IgG, Fcγ-fragment-specific antibody. The signal was developed using TMB as the substrate and stopped with 2 N sulphuric acid. The A450 nm values were then measured. The expression of NA was normalized to an anti-FLAG tag antibody (Wako). The mean A450 nm values at each dilution was obtained by subtracting the A450 nm values for the control well (empty-plasmid-transfected cells). The end-point titres of the serum samples were determined as the reciprocal of the highest dilution providing an A450 nm > 0.1.
K D determination
KD values were determined from Scatchard plots using ELISA data51. In the ELISA, VP40-induced VLPs presenting the YO/94/15-NA were used as the antigen. After being blocked with Blocking One (1/5 dilution), antigen-coated plates were incubated with the anti-NA mAb diluted two-fold (10–0.005 μg ml−1). The HRP-conjugated goat anti-human IgG, Fcγ-fragment-specific antibody was used as the secondary antibody and the signal was developed using TMB as the substrate. The reaction was stopped with 2 N sulphuric acid and the A450 nm values were then measured. The concentrations of antibody bound to antigen and of free antibody were calculated on the basis of the A450 nm values (assuming that the molecular weight of the antibody was 150 kDa). When the ratio of bound to free antibody concentration was plotted against the bound antibody concentration, the negative inverse of the resulting slope was determined as the KD.
Amino acid positions were plotted on the crystal structure of the NA protein of the CA/04/09, A/Brevig Mission/1/1918 (H1N1pre2009) or A/Memphis/31/1998 (H3N2; PDB accession codes: 4B7R, 3B7E and 4H53, respectively) viruses using the PyMOL molecular graphics system.
The two-way and one-way ANOVAs followed by Dunnett’s tests, log-rank tests and Wilcoxon signed-rank tests were performed using the GraphPad Prism software. P <0.01 was considered significantly different. No samples were excluded from the analysis.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
All data analysed during this study are included in this article. The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Wilson, I. A., Skehel, J. J. & Wiley, D. C. Structure of the haemagglutinin membrane glycoprotein of influenza virus at 3 Å resolution. Nature 289, 366–373 (1981).
Weis, W. et al. Structure of the influenza virus haemagglutinin complexed with its receptor, sialic acid. Nature 333, 426–431 (1988).
Job, E. R. et al. Serum amyloid P is a sialylated glycoprotein inhibitor of influenza A viruses. PLoS ONE 8, e59623 (2013).
Palese, P., Tobita, K., Ueda, M. & Compans, R. W. Characterization of temperature sensitive influenza virus mutants defective in neuraminidase. Virology 61, 397–410 (1974).
Matrosovich, M. N., Matrosovich, T. Y., Gray, T., Roberts, N. A. & Klenk, H. D. Neuraminidase is important for the initiation of influenza virus infection in human airway epithelium. J. Virol. 78, 12665–12667 (2004).
Potter, C. W. & Oxford, J. S. Determinants of immunity to influenza infection in man. Br. Med. Bull. 35, 69–75 (1979).
Hobson, D., Curry, R. L., Beare, A. S. & Ward-Gardner, A. The role of serum haemagglutination-inhibiting antibody in protection against challenge infection with influenza A2 and B viruses. J. Hyg. 70, 767–777 (1972).
Wrammert, J. et al. Broadly cross-reactive antibodies dominate the human B cell response against 2009 pandemic H1N1 influenza virus infection. J. Exp. Med. 208, 181–193 (2011).
Chen, Y. Q. et al. Influenza infection in humans induces broadly cross-reactive and protective neuraminidase-reactive antibodies. Cell 173, 417–429 (2018).
Knossow, M. & Skehel, J. J. Variation and infectivity neutralization in influenza. Immunology 119, 1–7 (2006).
Yamayoshi, S. et al. A broadly reactive human anti-hemagglutinin stem monoclonal antibody that inhibits influenza A virus particle release. EBioMedicine 17, 182–191 (2017).
Brandenburg, B. et al. Mechanisms of hemagglutinin targeted influenza virus neutralization. PLoS ONE 8, e80034 (2013).
Skehel, J. J. & Wiley, D. C. Receptor binding and membrane fusion in virus entry: the influenza hemagglutinin. Annu. Rev. Biochem. 69, 531–569 (2000).
Schulman, J. L., Khakpour, M. & Kilbourne, E. D. Protective effects of specific immunity to viral neuraminidase on influenza virus infection of mice. J. Virol. 2, 778–786 (1968).
Webster, R. G. & Laver, W. G. Preparation and properties of antibody directed specifically against the neuraminidase of influenza virus. J. Immunol. 99, 49–55 (1967).
Wilson, J. R. et al. An influenza A virus (H7N9) anti-neuraminidase monoclonal antibody with prophylactic and therapeutic activity in vivo. Antiviral Res. 135, 48–55 (2016).
Wohlbold, T. J. et al. Broadly protective murine monoclonal antibodies against influenza B virus target highly conserved neuraminidase epitopes. Nat. Microbiol. 2, 1415–1424 (2017).
Bosch, B. J. et al. Recombinant soluble, multimeric HA and NA exhibit distinctive types of protection against pandemic swine-origin 2009 A(H1N1) influenza virus infection in ferrets. J. Virol. 84, 10366–10374 (2010).
Seto, J. T. & Chang, F. S. Functional significance of sialidase during influenza virus multiplication: an electron microscope study. J. Virol. 4, 58–66 (1969).
Monto, A. S. et al. Antibody to influenza virus neuraminidase: an independent correlate of protection. J. Infect. Dis. 212, 1191–1199 (2015).
Monto, A. S. & Kendal, A. P. Effect of neuraminidase antibody on Hong Kong influenza. Lancet 1, 623–625 (1973).
DiLillo, D. J., Palese, P., Wilson, P. C. & Ravetch, J. V. Broadly neutralizing anti-influenza antibodies require Fc receptor engagement for in vivo protection. J. Clin. Invest. 126, 605–610 (2016).
Perussia, B., Starr, S., Abraham, S., Fanning, V. & Trinchieri, G. Human natural killer cells analyzed by B73.1, a monoclonal antibody blocking Fc receptor functions. I. Characterization of the lymphocyte subset reactive with B73.1. J. Immunol. 130, 2133–2141 (1983).
Stuart, S. G. et al. Human IgG Fc receptor (hFcRII; CD32) exists as multiple isoforms in macrophages, lymphocytes and IgG-transporting placental epithelium. EMBO J. 8, 3657–3666 (1989).
Futosi, K., Fodor, S. & Mócsai, A. Neutrophil cell surface receptors and their intracellular signal transduction pathways. Int. Immunopharmacol. 17, 638–650 (2013).
Bournazos, S., DiLillo, D. J. & Ravetch, J. V. The role of Fc-FcγR interactions in IgG-mediated microbial neutralization. J. Exp. Med. 212, 1361–1369 (2015).
Pincetic, A. et al. Type I and type II Fc receptors regulate innate and adaptive immunity. Nat. Immunol. 15, 707–716 (2014).
Laver, W. G., Air, G. M., Webster, R. G. & Markoff, L. J. Amino acid sequence changes in antigenic variants of type A influenza virus N2 neuraminidase. Virology 122, 450–460 (1982).
Luther, P., Bergmann, K. C. & Oxford, J. S. An investigation of antigenic drift of neuraminidases of influenza A (H1N1) viruses. J. Hyg. 92, 223–229 (1984).
Webster, R. G., Laver, W. G., Air, G. M. & Schild, G. C. Molecular mechanisms of variation in influenza viruses. Nature 296, 115–121 (1982).
Air, G. M., Els, M. C., Brown, L. E., Laver, W. G. & Webster, R. G. Location of antigenic sites on the three-dimensional structure of the influenza N2 virus neuraminidase. Virology 145, 237–248 (1985).
Colman, P. M., Varghese, J. N. & Laver, W. G. Structure of the catalytic and antigenic sites in influenza virus neuraminidase. Nature 303, 41–44 (1983).
Fanning, T. G., Reid, A. H. & Taubenberger, J. K. Influenza A virus neuraminidase: regions of the protein potentially involved in virus-host interactions. Virology 276, 417–423 (2000).
Dreyfus, C. et al. Highly conserved protective epitopes on influenza B viruses. Science 337, 1343–1348 (2012).
Simonelli, L. et al. Rational engineering of a human anti-dengue antibody through experimentally validated computational docking. PLoS ONE 8, e55561 (2013).
Wright P., Neumann G. & Kawaoka Y. in Field Virology 5th edn, vol. 2 (eds Knipe, D. M. & Howley, P. M.) 1691–1740 (Lippincott Williams & Wilkins, Philadelphia, 2007).
Iwatsuki-Horimoto, K. et al. Seroprevalence of pandemic 2009 (H1N1) influenza A virus among schoolchildren and their parents in Tokyo, Japan. Clin. Vaccine Immunol. 18, 860–866 (2011).
Murphy, B. R., Kasel, J. A. & Chanock, R. M. Association of serum anti-neuraminidase antibody with resistance to influenza in man. N. Engl. J. Med. 286, 1329–1332 (1972).
Bragstad, K., Nielsen, L. P. & Fomsgaard, A. The evolution of human influenza A viruses from 1999 to 2006: a complete genome study. Virol. J. 5, 40 (2008).
Memoli, M. J. et al. Evaluation of antihemagglutinin and antineuraminidase antibodies as correlates of protection in an influenza A/H1N1 virus healthy human challenge model. mBio 7, e00417-16 (2016).
Wan, H. et al. Structural characterization of a protective epitope spanning A(H1N1)pdm09 influenza virus neuraminidase monomers. Nat. Commun. 6, 6114 (2015).
Jiang, L. et al. Comparative efficacy of monoclonal antibodies that bind to different epitopes of the 2009 pandemic H1N1 influenza virus neuraminidase. J. Virol. 90, 117–128 (2015).
Kubota-Koketsu, R. et al. Broad neutralizing human monoclonal antibodies against influenza virus from vaccinated healthy donors. Biochem. Biophys. Res. Commun. 387, 180–185 (2009).
Yasuhara, A. et al. Diversity of antigenic mutants of influenza A(H1N1)pdm09 virus escaped from human monoclonal antibodies. Sci. Rep. 7, 17735 (2017).
Neumann, G. et al. Generation of influenza A viruses entirely from cloned cDNAs. Proc. Natl Acad. Sci. USA 96, 9345–9350 (1999).
Sakabe, S., Ozawa, M., Takano, R., Iwastuki-Horimoto, K. & Kawaoka, Y. Mutations in PA, NP, and HA of a pandemic (H1N1) 2009 influenza virus contribute to its adaptation to mice. Virus Res. 158, 124–129 (2011).
Yamayoshi S. et al. Human protective monoclonal antibodies against the HA stem of group 2 HAs derived from an H3N2 virus-infected human. J. Infect. 76, 177–185 (2017).
Couzens, L. et al. An optimized enzyme-linked lectin assay to measure influenza A virus neuraminidase inhibition antibody titers in human sera. J. Virol. Methods 210, 7–14 (2014).
Chen, G. W. et al. Genomic signatures of human versus avian influenza A viruses. Emerg. Infect. Dis. 12, 1353–1360 (2006).
Shannon, C. E. A mathematical theory of communication. Bell Syst. Tech. J. 27, 379–423 (1948).
Ottaviano, Y. & Gerace, L. Phosphorylation of the nuclear lamins during interphase and mitosis. J. Biol. Chem. 260, 624–632 (1985).
We thank P. C. Wilson for providing us with anti-NA mAbs; K. Iwatsuki-Horimoto and T. Koibuchi for assistance with experiments; C. Kawakami, E. Takashita and S. Nakajima for providing us with influenza viruses and S. Watson for editing the manuscript. This work was supported by the Japan Initiative for Global Research Network on Infectious Diseases from the Japan Agency for Medical Research and Development (AMED; grant no. JP18fm0108006), Leading Advanced Projects for medical innovation from the AMED (grant no. JP18am001007), Grants-in-Aid for Scientific Research on Innovative Areas from the Ministry of Education, Culture, Science, Sports and Technology (MEXT) of Japan (grant nos. 16H06429, 16K21723 and 16H06434) and the Center for Research on Influenza Pathogenesis funded by the NIAID contract no. HHSN272201400008C.