Fcγ Receptors Contribute to the Antiviral Properties of Influenza Virus Neuraminidase-Specific Antibodies

There is a pressing need for next-generation influenza vaccine strategies that are better able to manage antigenic drift and the cocirculation of multiple drift variants and that consistently improve vaccine effectiveness. Influenza virus NA is a key target antigen as a component of a next-generation vaccine in the influenza field, with evidence for a role in protective immunity in humans. However, mechanisms of protection provided by antibodies directed to NA remain largely unexplored. Herein, we show that antibody Fc interaction with Fcγ receptors (FcγRs) expressed on effector cells contributes to viral control in a murine model of influenza. Importantly, a chimeric mouse-human IgG1 with no direct antiviral activity was demonstrated to solely rely on FcγRs to protect mice from disease. Therefore, antibodies without NA enzymatic inhibitory activity may also play a role in controlling influenza viruses and should be of consideration when designing NA-based vaccines and assessing immunogenicity.

serum was heat inactivated and characterized in vitro and in vivo (Fig. 1). Vaccination raised a mixture of IgG isotypes directed toward the rNA, including IgG1, IgG2a, and IgG2b, with IgG1 showing the strongest signal in an enzyme-linked immunosorbent assay (ELISA), whereas IgG3 and IgE were undetectable by ELISA ( Fig. 1A and data not Binding of mouse antisera to Bel/09 rNA. Wells of a 96-well flat-bottom ELISA plate were coated with 0.5 g/ml of Bel/09 rNA overnight in sodium carbonate buffer. Wells were then blocked with 1% BSA in PBS, and heatinactivated antiserum was subsequently applied in serial dilutions and incubated for 2 h at room temperature in 0.5 mg/ml BSA in PBST. Binding of antisera was detected with either anti-mouse IgG1, IgG2a, or IgG2b conjugated to HRP. Data represents the average from 2 independent experiments Ϯ standard deviation (SD). OD, optical density. (B) Ability of antisera to inhibit virus infectivity. One hundred TCID 50 of Bel/09 was preincubated with increasing dilutions of heat-inactivated antisera directed toward rNA or rHA of Bel/09 or preimmune serum in triplicate and subsequently added to prewashed confluent monolayers of MDCKs. Cells were incubated for 6 days in the presence of trypsin. Virus replication was detected by a hemagglutination assay with turkey red blood cells. Results are expressed as the percentage of wells that scored positive for agglutination. (C) Antisera directed toward Bel/09 rNA can inhibit the viral NA activity. Bel/09 was incubated with increasing dilutions of heat-inactivated anti-Bel/ 09 NA sera or PBS antisera, and NA activity was determined at 18 h postincubation on fetuin as described in Materials and Methods. Data represent at least 2 independent repeats and show the average Ϯ SD from one experiment performed in duplicate. (D) Anti-Bel/09 NA serum protects mice in a dosedependent manner. Mice were treated via the i.p. route with 100 l of heat-inactivated mock serum or increasing amounts of anti-Bel/09 rNA serum and subsequently infected with 1 LD 50 of Bel/09. Mice were monitored for 14 days for weight loss. Mice were sacrificed if they had lost Ն25% of their original body weight. Survival percentages are indicated on the bottom right-hand side of the graph. Text colors match the legend color for each group. Data are representative of 2 repeats and show the average weight loss over time Ϯ the standard error of the mean (SEM [n ϭ 3]).
Antineuraminidase IgGs Protect through Fc␥ Receptors ® shown). Next, the ability of the serum to neutralize infectivity of the mouse-adapted descendant of A/Belgium/1/2009 (named Bel/09) was tested. Serum raised to buffer with adjuvant (phosphate-buffered saline [PBS]) was unable to neutralize virus infectivity, whereas anti-recombinant soluble Bel/09 HA (rHA) immune serum neutralized the virus at all concentrations tested. For antiserum directed against rNA, there was a titratable effect on viral infectivity (Fig. 1B), in line with previous studies (21). Further, the anti-Bel/09 NA serum was also able to inhibit the viral NA activity in an NA inhibition (NI) assay, whereas the PBS serum could not (Fig. 1C). Finally, to determine an in vivo protective dose, BALB/c mice were treated via the intraperitoneal (i.p.) route with either PBS serum as a negative control (100 l) or increasing amounts of anti-Bel/09 rNA serum. One day later, the mice were infected with 1 50% lethal dose (LD 50 ) of Bel/09 and monitored for weight loss over 14 days. All mice treated with anti-NA serum were significantly protected from weight loss in comparison to mock serum-treated mice (P Ͻ 0.05, two-way analysis of variance [ANOVA], main column effects). Fifty microliters of rNA immune serum was selected for further studies, as this conserved the antiserum and showed almost 100% protection from weight loss (Fig. 1D).
We next examined the possible contribution of Fc␥Rs to the protection provided by anti-NA serum. Wild-type (WT), Fcer1g Ϫ/Ϫ or Fcgr1 Ϫ/Ϫ Fcgr3 Ϫ/Ϫ mice were pretreated i.p. with 50 l of heat-inactivated anti-NA or PBS antiserum, as a negative control. Fcer1g Ϫ/Ϫ mice lack the common ␥ chain that, in mice, is required for the functional expression of Fc␥RI, -III, and -IV, whereas Fcgr1 Ϫ/Ϫ Fcgr 3 Ϫ/Ϫ mice are unable to express both Fc␥RI and -III. One day following immune serum administration, the mice were challenged with 1 LD 50 of Bel/09 and monitored for weight loss over 14 days. WT, Fcer1g Ϫ/Ϫ , and Fcgr1 Ϫ/Ϫ Fcgr 3 Ϫ/Ϫ knockout (KO) mice treated with anti-NA sera were protected from weight loss and displayed no significant difference in weight loss over-time. WT and knockout mice treated with serum from the PBS-administered group displayed transient weight loss, with some mice succumbing to the infection ( Fig. 2A). Importantly there was also no significant difference between challenged mock-treated WT and KO mice, suggesting that there was no intrinsic susceptibility difference between the WT and KO mice for influenza virulence ( Fig. 2A). To examine the possible impact of Fc␥Rs on the ability to control virus replication, lung viral loads were examined. Mice were treated as in the weight loss experiments and challenged 1 day later with 0.1 LD 50 of Bel/09. On days 3 and 5 after infection, lung homogenates were assessed for the presence of infectious virus. Treatment of challenged mice with polyclonal anti-NA serum reduced viral titers in comparison to mock-treated mice in both WT and KO animals on days 3 ( Fig. 2B) and 5 ( Fig. 2C) postinfection. However, by day 5, at the dose tested in this passive serum transfer experiment, the strong reduction in viral titers observed in anti-NA-treated mice on day 3 was less pronounced. There was no significant difference between WT and KO mice in the control of virus replication on days 3 and 5. Taken together, these data show that polyclonal anti-NA responses with NI activity do not rely on Fc␥Rs for protection against a homologous virus challenge in the mouse model.
Fc␥Rs contribute to the control of viral replication by a monoclonal antibody with NI activity. Thus far, we have shown that a polyclonal anti-NA response does not rely on the engagement of Fc␥Rs to control viral infection. As polyclonal responses are complex involving a variety of antibodies with different affinities for NA and comprises antibodies with and without NI activity, it was important to address if a monoclonal antibody directed against NA would result in the same outcome. Previously, we isolated and defined an A(H1N1)pdm09 N1-specific IgG1 mouse monoclonal antibody, N1-C4, with NI activity, which could control infection of Bel/09 virus in mice (7). We sought to address if N1-C4 could protect against influenza virus infection in the absence of activating Fc␥Rs. WT, Fcer1g Ϫ/Ϫ , or Fcgr1 Ϫ/Ϫ Fcgr3 Ϫ/Ϫ BALB/c mice were treated intranasally with 1 mg/kg of body weight N1-C4 or an isotype control. Twenty-four hours later, the animals were challenged with 1 LD 50 of Bel/09 and monitored for weight change (Fig. 3A). Challenged isotype control-treated mice experienced transient severe weight loss with approximately half of the mice recovering from infection. There were no significant differences between KO and WT mice treated with the isotype control. N1-C4-treated WT, Fcer1g Ϫ/Ϫ , and Fcgr1 Ϫ/Ϫ Fcgr3 Ϫ/Ϫ mice displayed minor weight loss with all mice recovering from infection. Interestingly, on days 8 and 9 knockout mice displayed a slightly higher weight loss than WT counterparts, although this only reached a statistically significant level in the Fcgr1 Ϫ/Ϫ Fcgr 3 Ϫ/Ϫ mice. Next, viral loads were determined in the lungs of mice treated with N1-C4 or the isotype control followed by infection with Bel/09 at 0.1 LD 50 . On day 3 postinfection in the N1-C4-treated Fcer1g Ϫ/Ϫ mice, there was a clear trend for increased levels of virus compared to WT mice (Fig. 3B, left-hand panel). This was also evident in the double-KO FIG 2 Activating Fc␥Rs are not required to control influenza virus in mice treated with polyclonal anti-NA serum. Fifty microliters of heat-inactivated anti-NA sera (circles) or PBS antisera (squares) were passively transferred via the i.p. route into Fcer1g or Fcgr1/Fcgr3 knockout (KO) mice. BALB/c WT mice were included as controls. One day later, the mice were challenged with 1 LD 50 of Bel/09 and monitored daily, and any mice that had lost Ն25% of their original body weight were euthanized. Weight loss data represent the mean percentages Ϯ SEM of original body weight over time (n ϭ 8 to 10, pooled from 2 independent experiments). Survival percentages are indicated in the left-hand corner of the graphs where the text color matches colors indicated in the legend for each group (A). In a separate set of experiments, after passive transfer, KO or WT mice were challenged with 0.1 LD 50 of Bel/09 and viral titers assessed in lung homogenates on day 3 (B) or day 5 (C) by TCID 50 . The horizontal lines represent the mean (pooled from 2 independent experiments). The dotted line indicates the detection limit of the TCID 50 . ns, nonsignificant. *, P Ͻ 0.05, **, P Ͻ 0.01, and ***, P Ͻ 0.001, by one-way ANOVA. mouse strain, where the absence of FcR␥I and Fc␥RIII significantly impacted the ability of the mice to control viral loads (P Ͻ 0.05 compared to WT mice, one-way ANOVA) (Fig. 3B, right-hand panel). This trend remained on day 6 postinfection with higher levels of virus in the KO compared to the WT N1-C4-treated mice (Fig. 3C). Thus, we conclude that N1-C4 is capable of controlling infection in mice in the absence of Fc␥Rs; however, in the absence of the activating Fc␥RI and -III or in mice that lack the common ␥ chain, viral clearance within the lungs is less efficient than in WT mice.
Development and characterization of a human IgG1 chimeric anti-NA antibody. The data presented above have explored antibody responses with direct antiviral Fcer1g or Fcgr1/Fcgr3 KO mice (white symbols) were treated with 1 mg/kg of N1-C4 (circles) or a mouse IgG1 isotype control (squares) via the intranasal route under isoflurane sedation. BALB/c WT mice were included as controls (black symbols). One day later, the mice were challenged with 1 LD 50 of Bel/09 and monitored daily, and any mice that had lost Ն25% of their original body weight were euthanized. Survival percentages are indicated on the left-hand side of the graphs, where the text color matches colors indicated for each group in the legend. Weight loss data represent the mean percentages Ϯ SEM of original body weight over time (n ϭ 8 to 10 pooled from 2 independent experiments) (A). In a separate set of experiments, after passive transfer, KO or WT mice were challenged with 0.1 LD 50 of Bel/09, and viral titers were assessed in lung homogenates on day 3 (B) or day 6 (C) by TCID 50 . The horizontal lines represent the mean. The dotted line indicates the detection limit of the TCID 50 . ns, nonsignificant. *, P Ͻ 0.05, **, P Ͻ 0.01, and ***, P Ͻ 0.001, by one-way ANOVA. activity, and as such, it is not unreasonable that the antibodies do not rely solely on Fc␥Rs to control an influenza virus infection. Recent studies have indeed shown that antibodies without NI activity can protect mice in an Fc-dependent manner (20). We have previously isolated a mouse IgG1 monoclonal antibody, N1-7D3, that could bind to Bel/09 NA but was unable to inhibit the viral NA activity or protect mice in vivo. As mouse IgG1 is poor at engaging activating Fc␥Rs, interacting only with mouse Fc␥RIII (12), we next sought to enhance the affinity of N1-7D3 for Fc␥Rs before assessing its protective ability. A chimeric antibody was constructed with the variable domains of mouse N1-7D3 and the constant regions of human IgG1 (here called huN1-7D3). Human IgG1 can interact with all human activating Fc␥Rs. Likewise, N1-C4 was engineered to express human IgG1 constant domains (here named huN1-C4) and thus represents a chimeric human N1 NA-specific antibody that has direct antiviral activity.
As altering the constant region may impact the ability to bind to the antigen compared to the native antibody (22), we initially ascertained that huN1-7D3 and huN1-C4 behaved like the native mouse monoclonal antibodies. Similar to mouse N1-7D3 and N1-C4, huN1-7D3 and huN1-C4 were both able to bind to Bel/09 rNA in a dose-dependent manner (Fig. 4A). Furthermore, while human and mouse N1-C4 could block the NA activity of Bel/09, N1-7D3 (human and mouse) could not, consistent with our previous studies (Fig. 4B). Therefore, both huN1-C4 and huN1-7D3 behaved as expected.
It is known that core fucosylation of the conserved N-glycan at site Asn 297 in the Fc tail of human IgGs modulates interaction with the Fc␥RIIIa on NK cells and the loss of fucose at this site results in a substantial increase in ADCC activity (23). We next used glycan profiling to compare the extent of core fucosylation of the chimeric and the native mouse monoclonal antibodies. N-glycans from the monoclonal antibodies were isolated by PNGase F treatment, labeled with 8-amino-1,3,6-pyrenetrisulfonic acid (APTS), and subsequently treated or not with fucosidase. The labeled glycans were then separated by capillary electrophoresis. A labeled dextran ladder and N-glycans from bovine RNase B were taken along as standards. Four major glycan peaks were identified in mouse and the human chimeric antibodies. Treatment of both N1-7D3 and huN1-7D3 with fucosidase resulted in a mobility shift of the 4 main peaks, indicating that the antibody glycans were almost fully core fucosylated (Fig. 4C), consistent with typical glycosylation forms found on Asn 297 in wild type antibodies. Mouse and human chimeric N1-C4 were also both core fucosylated (data not shown). As such, it is expected that fucosylation at Asn 297 will not contribute to any additional biological activity of huN1-7D3 compared to N1-7D3 produced by the mouse hybridoma.
Fc␥R engagement is important for control of IAV infection by an NA-specific monoclonal antibody that lacks NI activity. To examine if huN1-7D3 could possibly protect against an influenza virus infection in vivo we utilized the transgenic humanized Fc␥R mice (huFc␥R mice), which are knocked out for mouse Fc␥Rs and transgenic for all human Fc␥-activating receptors (24). Mice were treated intranasally (i.n.) with 2.5 mg/kg of huN1-C4, huN1-7D3, or a human IgG1 isotype control. One day later, the mice were challenged with Bel/09 at 1 LD 50 and monitored for weight loss and survival. Both huN1-C4-and huN1-7D3-treated mice lost significantly less weight than the isotype-treated control mice, with 100% of the mice surviving the infection when treated with anti-NA monoclonal antibodies versus a 50% survival rate for the isotype control-treated mice. Hence, with an Fc that can engage multiple Fc␥-activating receptors, N1-7D3 now displayed the potential to protect against influenza virus infection in mice.
To explore a potential Fc-mediated mechanism of action for the protection provided by the human chimeric antibodies, an ADCC reporter assay was used. MDCK cells were infected with Bel/09 at a multiplicity of infection (MOI) of 1 and used as target cells for the antibodies to target NA expressed at the surface of the cells. Both huN1-C4 and huN1-7D3 were able to engage the Fc␥RIIIa eceptor expressed on Jurkat cells, resulting in a luciferase signal, whereas the isotype control could not (Fig. 5B).
Next, for more definitive evidence that huN1-7D3 controlled infection through engagement of Fc␥Rs, the Fc portion of huN1-7D3 and huN1-C4 was removed by treatment with IdeS enzyme to create F(ab=) 2 fragments (Fig. 5C). F(ab=) 2 fragments and intact antibodies were subsequently passively transferred in huF␥R mice at equimolar  (Fig. 5D). One day later, the mice were challenged with Bel/09 at 2 LD 50 s and monitored for weight loss and survival. As expected, all the mice treated with antibodies huN1-C4 and huN1-7D3 survived the challenge infection. However, huN1-C4-treated mice displayed significantly less weight loss than huN1-7D3 treated mice (P Ͻ 0.0001, two-way ANOVA, main column effects). The F(ab=) 2 fragment of huN1-C4 also protected from lethality, albeit mice lost significantly more weight than mice treated with intact huN1-C4 monoclonal antibody (P Ͻ 0.0001, two-way ANOVA, main column effects). In contrast, the huN1-7D3 F(ab=) 2 fragment was unable to protect the mice, where the majority of the mice succumbed to the infection by days 8 and 9. Therefore, we conclude that huN1-7D3 relies on Fc-Fc␥R interactions to provide protection in humanized mice.

DISCUSSION
Fc␥ effector functions have been shown to contribute to various aspects of antibody-mediated protection against influenza viruses, particularly for broadly binding antibodies. For example, the engagement of Fc␥R on alveolar macrophages has been shown to be crucial for M2e antibody-mediated protection (14). Furthermore, the majority of HA stem-specific antibodies require interaction with effector cells, including macrophages (18) and NK cells to control influenza virus infection in vivo. For NA antibodies, this area is ill-defined, with limited studies examining the contribution of Fc␥ effector functions.
Kim et al. (19) reported that in the absence of Fc␥Rs, anti-NA serum was still able to control infection with H1N1pdm09. Our results are in line with these studies where we observed no major difference in weight loss and survival over time in the presence of NI antiserum or monoclonal antibody N1-C4. Studies presented herein expanded on previous work by Kim et al. by examining the viral loads in the lungs at various days after infection. For polyclonal antiserum viral lung loads in Fc␥R-deficient mice were not significantly different from those in WT mice. However, Fc␥Rs enhanced the ability of monoclonal antibody N1-C4 to control virus loads in the lungs, particularly on day 6 postinfection. Additionally, treating huFc␥R mice with the F(ab=) 2 fragment of huN1-C4 could still protect, but mice displayed significantly more weight loss than those treated with intact monoclonal huN1-C4, although in this case, we cannot rule out that a decreased half-life of the F(ab=) 2 fragment, compared to the intact monoclonal antibody, contributed to the reduced protection. Besides ADCC and antibody-dependent cellular phagocytosis (ADCP) to aid in viral clearance, Fc␥Rs have been implicated in maturation and activation of antigen-presenting cells (APCs) and in turn activation of T cells, plus in B cell maturation and affinity selection (reviewed by Pincetic et al. [25]). We speculate that while NI monoclonal antibodies aid in the initial control of the virus, engagement of Fc␥Rs could play a role in enhancing viral clearance from the lungs. We observed that Fcer1g Ϫ/Ϫ mice treated with N1-C4 on day 6 postinfection displayed reduced viral clearance compared to Fcgr1 Ϫ/Ϫ Fcgr3 Ϫ/Ϫ mice (Fig. 3C). Fcer1g Ϫ/Ϫ mice lack the FcR␥ chain, which associates not only with FcRs I, III, and IV but also with FcRI and several other molecules. Of interest are members of the C-type lectin family-dendritic cell-associated C-type lectin-2 (Dectin-2), macrophageinducible C-type lectin (Mincle), dendritic cell immunoactivating receptor (DCAR) and blood dendritic cell antigen 2 (BDCA-2)-all of which have been shown to function as pattern recognition receptors for various human pathogens (26). There is a clear role described for the interaction of influenza virus and other reported C-type lectins (27). In addition, Mincle has been shown to be upregulated during influenza virus infection (28).
By altering the Fc tail of N1-7D3 from a mouse IgG1 to one of human IgG, which can engage multiple human Fc␥Rs, we were able to protect humanized Fc␥R mice against a potentially lethal challenge of H1N1pdm09. In line with this, by removing the Fc portion of huN1-7D3, protection was lost. As such it is clear that N1-7D3 relies on Fc␥R engagement to mediate protection, in contrast to N1-C4, which is able to control the virus by direct inhibition of viral NA activity. Glycan analysis revealed that both huN1-C4 and huN1-7D3 displayed core fucosylation at site Asn 297 . Engineering these monoclonal antibodies without a core fucose at this site would be of interest as studies have shown that this can significantly enhance ADCC activity (29), and thus, in turn could provide more potent protection against influenza virus. To our knowledge, core fucose mutants have never been investigated to increase protection against influenza virus; however, the GASD/ALIE Fc mutant of an anti-HA stalk antibody, 6F12, has been shown to provide significantly greater protection against PR8/34 infection in humanized Fc␥R mice compared to the wild-type antibody (30). This mutant shows enhanced binding to activating huFc␥Rs and increased activation of ADCC. The area of antibody engineering research holds great promise for the future development of potent anti-influenza antibodies.
DiLillo et al. (13) have previously shown that the broad anti-NA antibody 3C05 required the engagement of Fc␥Rs to protect against H1N1pdm09, whereas a strainspecific antibody did not. N1-7D3 also broadly binds to N1 viruses at a conserved linear peptide at the carboxy terminus of the NA (7). Results presented here confirm the studies by DiLillo et al.; however, it remains to be seen if huN1-7D3 can also protect in mice against a variety of N1 IAVs. Furthermore, while we demonstrated that huN1-7D3 could engage with Fc␥RIIIa on a reporter cell line, the contribution of this engagement to mediate protection in vivo is unknown and whether other effector functions are involved is also unknown. Future studies are needed to address this. Importantly, these future studies also need to examine the possible synergy between HA and NA antibodies and whether increased induction of NA immunity by influenza virus vaccines can provide better heterologous protection: e.g., against drifted influenza virus strains.

MATERIALS AND METHODS
Viruses, recombinant proteins and antiserum generation. The type A influenza virus (IAV) used in this study was the mouse-adapted H1N1 A/Belgium/1/2009 (Bel/09) strain (7). Bel/09 was propagated in Madin-Darby canine kidney (MDCK) cells in serum-free medium in the presence of TPCK (tosylsulfonyl phenylalanyl chloromethyl ketone)-treated trypsin (Sigma). The median tissue culture infective dose (TCID 50 ) and median lethal dose (LD 50 ) were calculated by the Reed and Muench method (31).
Recombinant soluble tetrameric Bel/09 NA (rNA) and trimeric Bel/09 HA (rHA) were produced as described previously (32). Briefly, the intracellular and transmembrane domains of NA as well as the stalk of NA were replaced by a type I secretion signal followed by a tetramerizing domain derived from Tetrabrachion (33). The HA trimer was stabilized with a trimerizing GCN4-derived leucine zipper (34). Secretion was facilitated by an N-terminal CD5-derived secretion signal, and a Strep-tag allowed purification from supernatants of transiently transfected HEK293T cells by affinity trap using a StrepTrap HP column followed by size exclusion chromatography in phosphate-buffered saline (PBS) using an AKTA explorer purification system (GE Healthcare Life Sciences).
One microgram of purified Bel/09 rNA or rHA was used per mouse per immunization step to produce mouse antiserum. Mice were primed and boosted 3 weeks apart in the presence of Sigma Adjuvant System (SAS), containing the immunostimulants monophosphoryl lipid A and synthetic trehalose dicorynmycolate. Mice were terminally bled via retro-orbital bleed, and serum was isolated by allowing clot formation.