Plasticity of Amino Acid Residue 145 Near the Receptor Binding Site of H3 Swine Influenza A Viruses and Its Impact on Receptor Binding and Antibody Recognition

The HA N 145 K substitution results in significant antigenic changes in both human and swine H3N2 IAVs (7, 16–18). Asparagine (Asn or N) was the most prevalent amino acid found at residue 145 across all analyzed hosts (Fig. 1A). In swine-origin IAVs, around 64% of isolates possessed Asn at residue 145, followed by lysine (Lys or K) at ∼33% and serine (Ser or S) at ∼2.5%. In human-origin IAVs, about 42% of isolates carried Asn, followed by Ser at ∼39% and Lys at ∼18%. In avian-origin IAVs, the frequency was ∼49% for Asn, ∼45% for Ser, and ∼4.6% for glycine (Gly or G). In canine-origin IAVs, around ∼73% of isolates possessed Asn, followed by aspartic acid (Asp or D) at ∼24%. In equine-origin IAVs, 99% of isolates carried Asn. Isolates possessing arginine (Arg or R), histidine (His or H), isoleucine (Ile or I), cysteine (Cys or C), methionine (Met or M), glutamine (Gln or Q), or threonine (Thr or T) at residue 145 were found at a frequency below 1%. Alanine (Ala or A), glutamic acid (Glu or E), leucine (Leu or L), phenylalanine (Phe or F), proline (Pro or P), tryptophan (Trp or W), tyrosine (Tyr or Y), and valine (Val or V) were not observed in any of the isolates from analyzed hosts.

To evaluate the impact of substitutions at residue 145, a panel of viruses carrying single substitutions representing each of the 20 amino acids was generated by reverse genetics (RG) (19, 20) (Fig. 1B). Since these viruses were generated in the context of a swine-origin H3N2 IAV (Fig. 1A), 145 N, 145 K, and 145 S are referred to as naturally occurring substitutions, whereas the remaining substitutions are termed alternative or nonnaturally occurring. Substitutions were introduced into the HA of A/turkey/Ohio/313053/2004 (H3N2), which here is referred to as OH/04 wild-type (wt) virus and naturally carries 145 N (Fig. 1C). Of the 19 OH/04 viruses with a single amino acid mutation at amino acid (aa) 145 (OH/04 HA aa 145 single mutant viruses) rescued, next-generation sequencing analysis revealed that 12 substitutions (145 A, 145 C, 145 G, 145 H, 145 K, 145 L, 145 M, 145 P, 145 Q, 145 R, 145 S, and 145 T) were well tolerated (98 to 100% of sequenced reads possessed the expected codon at residue 145) after 3 passages in Madin-Darby canine kidney (MDCK) cells. For substitutions 145 F, 145 I, 145 V, and 145 Y, the percentage of reads bearing the mutated codon was around 90% while the remaining 10% of reads showed a partial reversion to the wt codon (I 145 N, V 145 N, and Y 145 N) or a partial transition for a codon specifying Ser (F 145 S). None of these HA substitutions led to compensatory substitutions in the HA or neuraminidase (NA) segments. Three substitutions (145 D, 145 E, and 145 W) were not well tolerated, leading to partial reversion to the wt codon (D 145 N) or partial transition to a codon specifying either Gly (E 145 G) or Leu (W 145 L). Moreover, additional substitutions emerged on the HA for 145 E and 145 W (T 128 A) or on the NA for 145 E (T 148 I). For these reasons, 145 D, 145 E, and 145 W viruses were excluded from further analysis.

Viruses were assayed for their ability to agglutinate red blood cells (RBCs) from different species (turkey, chicken, and horse) by standard hemagglutination (HA) assay. Turkey and chicken RBCs are known to carry both SAα2-3Gal and SAα2-6Gal receptors on their cell surfaces, while horse RBCs display mainly SAα2-3Gal (21–23). Similar to the case for the 145 N (wt) virus, nearly all mutant viruses were found to agglutinate turkey RBCs efficiently, with the exception of 145 C virus, which displayed low HA titers (Table 1). In contrast, agglutination of chicken RBCs was less consistent, with HA titers of some mutant viruses comparable to that of the 145 N (wt) virus and other mutant viruses exhibiting an 8- to 16-fold decrease in HA titers (145 F, 145 G, 145 V, and 145 Y viruses) or no agglutination (145 C virus) (Table 1). Only the 145 F virus showed detectable, albeit low, HA titers using horse RBCs (Table 1). There was no correlation between the ability to agglutinate RBCs and virus titers. Due to the decreased ability to agglutinate RBCs, the 145 C virus was not tested in further assays.

Viruses were then compared in a multiple-step infection cycle in vitro. MDCK cells were infected at a low multiplicity of infection (MOI) (0.01). All viruses grew to high titers and displayed similar growth kinetics (Fig. 2A). There was no discernible difference in peak titers (∼6.5 to 7.0 log₁₀ 50% tissue culture infective dose [TCID₅₀]/ml equivalents) at 72 h postinfection (hpi), with the exception of the 145 P and 145 H viruses, which showed slightly lower peak titers (∼5.5 log₁₀ TCID₅₀/ml equivalents) (P < 0.001). Taken together, these results indicate that residue 145 shows high levels of plasticity in vitro despite of the limited detection of amino acid variation in nature. Furthermore, single amino acid substitutions did not have a major impact on growth kinetics in vitro but modulated the ability of mutant viruses to agglutinate RBCs from particular hosts.

To further expand on the receptor binding characterization of the aa 145 mutants, we analyzed whether these substitutions were involved in modulating receptor avidity or receptor specificity by measuring agglutination of turkey RBCs previously treated with different concentrations of bacterial neuraminidase (Fig. 2B and C). All of the mutant viruses bound to desialylated RBCs to various degrees. Naturally occurring substitutions showed the highest avidity to receptors. Binding of 145 N (wt) virus was indistinguishable from that of either the 145 K (RBC treated with both neuraminidases) or 145 S (RBC treated with Clostridium perfringens neuraminidase) virus. Nearly all mutant viruses carrying alternative substitutions at residue 145 displayed decreased receptor binding avidity, with the notable exception of the 145 M virus, which showed no discernible difference in binding compared to the 145 N (wt) virus. Overall, the results suggest differences in avidity regardless of the bacterial neuraminidases tested, although specificity cannot be completely ruled out.

Next, we performed a glycan-based enzyme-linked immunosorbent assay (ELISA) using monospecific preparations of horseradish peroxidase (HRP)-conjugated fetuin (Fet-HRP) as surrogates of binding to SAα2-3Gal (3-Fet-HRP) or SAα2-6Gal (6-Fet-HRP) (24) glycans (Fig. 3A to P). The assay reliably discriminated receptor binding specificity as evidenced by the viruses used as controls, with human pH1N1 showing a preference to SAα2-6Gal (Fig. 3Q) while avian ΔH5N1 displayed restricted binding to SAα2-3Gal (Fig. 3R). All of the mutant viruses retained binding to SAα2-6Gal with no residual binding to SAα2-3Gal. Naturally occurring substitutions (145 N [wt], 145 K, and 145 S viruses) showed the greatest binding to SAα2-6Gal (Fig. 3A, G, and M). Consistent with differences in avidity from previous results, mutant viruses carrying alternative substitutions at residue 145 displayed weaker binding to SAα2-6Gal (Fig. 3B, F, H to L, and N to P) than viruses possessing naturally occurring substitutions. Among mutant viruses carrying alternative substitutions, 145 M and 145 P viruses demonstrated the highest binding (Fig. 3I and J), while 145 F and 145 G viruses showed substantial decreases in binding to SAα2-6Gal (Fig. 3C and D). Overall, these results suggest that substitutions in HA at residue 145 do not affect receptor specificity but modulate receptor avidity. Nearly all the alternative substitutions led to decreased receptor binding avidity.

To further compare the impacts of substitutions at residue 145 on receptor binding specificity, glycan array analysis was performed for all viruses using a glycan microarray containing linear, O-linked, and N-linked glycans with extended poly-N-acetyl-lactosamine (poly-LacNAc) repeats found in human, swine, and ferret airway tissues (25). The array provides a broad semiquantitative view of the binding preference for specific viruses. All of the viruses retained specificity to SAα2-6Gal glycans (Fig. 4A to P), corroborating the results from the glycan-based ELISA (Fig. 3). Viruses carrying naturally occurring substitutions showed expanded binding to nearly all SAα2-6Gal glycan types in the array. OH/04 wt (145 N) virus had slightly preferred binding to O-glycans (Fig. 4A), while 145 K virus appeared to favor binding to N-glycans (Fig. 4G). 145 S virus showed a slight reduction in binding to linear and some of the smaller O-glycans (Fig. 4M) compared to OH/04 wt (145 N) virus (Fig. 4A). There were distinct patterns of binding for viruses carrying alternative substitutions. 145 F, 145 L, 145 M, 145 P, 145 Q, 145 T, and 145 V viruses demonstrated binding to a broader range of glycans that was comparable to that for OH/04 wt (145 N) virus (Fig. 4C, H, I, and K to O). The 145 A, 145 G, 145 H, 145 I, 145 R, and 145 Y mutant viruses displayed a more restricted binding to glycans in the array, although they appear to maintain a preference for O-linked glycans (Fig. 4B, D to F, L, and P). Taken together with the glycan-based ELISA results, these data indicated that while certain substitutions in HA at residue 145 lead to broad reductions in binding to SAα2-6Gal glycans, many nonnatural variants appear to be well tolerated, with only minor variance in receptor specificity detected.

For human H3N2 IAVs, it has been proposed that the emergence of the HA N 145 K substitution led to a change in antibody immunodominance (7). To assess whether a similar phenomenon occurs in swine-origin H3 IAVs and how alternative substitutions at residue 145 affect antigenicity, the serum reactivity of each mutant virus in the panel was tested by ELISA (Fig. 5A to C) and hemagglutination inhibition (HI) assays (Fig. 6A and B). For these experiments, swine antisera against NY/11 or IA/14 wild type viruses were tested. The HA1 domains of these viruses differ by only two amino acids: NY/11 virus possesses 145 N/289 P, while IA/14 carries 145 K/289 S. Based on its location at the base of the HA1 globular head and no prior evidence of an antigenic role, we assume that position 289 is antigenically irrelevant. Additionally, swine antisera against OH/04 wt virus were tested. To prevent confounding effects, the HA gene segments of the NY/11 and IA/14 viruses were rescued by reverse genetics in the background of 7 gene segments from the OH/04 wt virus. The 1 + 7 NY/11 and IA/14 reverse genetics viruses were used as controls in ELISA and HI assays.

The reactivities of sera generated against the OH/04 wt virus were similar across the entire panel of OH/04 aa 145 single mutant viruses (Fig. 5A), indicating that substitutions at residue 145 in the context of OH/04 HA do not change antigenicity against wt homologous sera. Consistent with amino acid divergence between OH/04 wt and control viruses, OH/04 wt virus antisera had reduced reactivity to both NY/11 and IA/14 control viruses. Surprisingly, sera raised against the NY/11 virus displayed no discernible difference in reactivity among all of the viruses evaluated regardless of the aa 145 substitution, including the NY/11 and IA/14 viruses (Fig. 5B). As observed for human H3N2 IAVs, the presence of 145 K in a swine-origin IAV results in an immunodominant epitope targeted by the antibody response (Fig. 5C). Sera generated against the IA/14 virus showed decreased reactivity to the NY/11 virus (possessing HA residue 145 N) compared to the IA/14 virus (carrying HA residue 145 K). Nearly all of the OH/04 aa 145 single mutant viruses, with the exception of 145 K virus, exhibited diminished reactivity to the IA/14 virus antisera. The 145 K virus was the only mutant virus to consistently show reactivity more similar to that of homologous IA/14 sera generated against the IA/14 virus, confirming the impact of HA residue 145 K in modulating antibody immunodominance.

Antigenic characterization by HI assay was mostly consistent with the ELISA data. The OH/04 wt virus antisera reacted to all OH/04 aa 145 single mutant viruses. Relative to the 145 N (wt) virus, changes in HI titers were almost indistinguishable (Fig. 6A and B). The OH/04 wt virus antisera had a large reduction in cross-HI reactivity against the IA/14 virus and only a modest reduction against the NY/11 virus. Similar to the ELISA data, most substitutions at residue 145 minimally affected NY/11 virus antiserum reactivity, with the 145 K virus causing the largest reduction in cross-HI titers. Reactivity of the IA/14 virus antisera was drastically reduced against nearly all of the OH/04 aa 145 single mutant viruses. In agreement with previous results, the 145 K virus was the only mutant virus to show minimal reduction in serum cross-reactivity compared to IA/14 virus. Collectively, these results indicate that residue 145 K has a profound impact on modulating antibody immunodominance and that serum raised against a 145 K bearing virus can recognize the epitope bearing such a substitution in the context of a distinct H3 HA.

Madin-Darby canine kidney (MDCK) and human embryonic kidney 293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Cells were propagated at 37˚C in a humidified incubator under a 5% CO₂ atmosphere.

The OH/04 wt strain is a prototypic swine-origin virus amenable to genetic manipulation by an established reverse genetics (RG) system (18, 47, 48). Single amino acid substitutions representing each of the 20 naturally occurring amino acids (Table 2) were inserted by site-directed mutagenesis at the codon corresponding to HA residue 145 (H3 numbering). To prevent the carryover of OH/04 wt HA plasmid DNA during PCR amplification, the cloned OH/04 wt HA segment was split into two overlapping plasmids, pDP-SD1 and pDP-SD2. The pDP-SD1 plasmid carries the mouse RNA polymerase I terminator (t1) (20) followed by nucleotides 1 to 522 of the OH/04 wt HA segment. The pDP-SD2 plasmid contains nucleotides 500 to 1762 of the OH/04 wt HA segment followed by the human RNA polymerase I promoter (polI) (20). To introduce the desired mutations into OH/04 wt HA, a PCR fragment containing t1 followed by the 5′ portion of the HA was generated with forward primer 5′-ACC GGA GTA CTG GTC GAC CTC CGA AGT TGG GGG GGA GCA AAA GCA GG-3′ and the respective reverse primer (Table 2) using pDP-SD1 as the template. Similarly, a PCR fragment comprising the 3′ portion of HA followed by polI was amplified from pDP-SD2 using reverse primer 5′-ATG CTG ACA ACG TCC CCG GCC CGG CGC TGC T-3′ and the respective forward primer (Table 2). All PCR products were purified by gel extraction using the QIAquick gel extraction kit (Qiagen, Valencia, CA) and combined to produce RG-ready PCR-based HA segments for individual OH/04 HA aa 145 single mutants by overlapping PCR as previously described (19). All PCRs were performed with Phusion high-fidelity DNA polymerase (New England Biolabs, Ipswich, MA), and products were confirmed to be free of unwanted mutations by sequencing. Viruses were rescued by PCR-based RG using a coculture of 293T and MDCK cells as previously described (19, 20). To generate OH/04 HA aa 145 single mutant viruses, the respective RG PCR-based HA segment was paired with the seven plasmids representing the remaining OH/04 wt gene segments. Following transfection, cells were incubated at 35˚C. After 24 h of incubation, medium was replaced with Opti-MEM I (Life Technologies, Carlsbad, CA) containing 1 μg/ml TPCK (tosylsulfonyl phenylalanyl chloromethyl ketone)-trypsin (Worthington Biochemicals, Lakewood, NJ) and 1% antibiotic-antimycotic solution (Sigma-Aldrich, St. Louis, MO). Following virus rescue, virus stocks were amplified in MDCK cells. Virus stocks were titrated by TCID₅₀, and virus titers were determined by the method of Reed and Muench (49).

Virus RNA from tissue culture supernatant virus stocks was purified using the RNeasy minikit (Qiagen, Valencia, CA) or the MagNA Pure LC RNA isolation kit (Roche Life Science, Mannheim, Germany). Isolated virus RNA served as the template in a one-step reverse transcriptase PCR (RT-PCR) for multisegment, whole-genome amplification (50). Amplicon sequence libraries were prepared as previously described (50) or using the Nextera XT DNA Library Prep kit (Illumina, San Diego, CA) according to the manufacturer’s protocol. Barcoded libraries were multiplexed and sequenced on the high-throughput Illumina MiSeq sequencing platform in a paired-end 150-nucleotide (nt) run format. De novo genome assembly was performed as described previously (50), and HA- and NA-specific reads were mapped to OH/04 reference sequences using Geneious 10.1.3 (51).

Confluent monolayers of MDCK cells were inoculated at a multiplicity of infection (MOI) of 0.01 for each virus. After 1 h of incubation, the virus inoculum was removed and cells washed twice with 1× phosphate-buffered saline (PBS). Opti-MEM I (Life Technologies, Carlsbad, CA) containing TPCK-trypsin (Worthington Biochemicals, Lakewood, NJ) and an antibiotic-antimycotic solution (Sigma-Aldrich, St. Louis, MO) was then added to the cells. At the indicated time points, tissue culture supernatant from inoculated cells was collected for virus titer quantification. Virus RNA from tissue culture supernatant was isolated using the MagMAX-96 AI/ND viral RNA isolation kit (Thermo Fisher Scientific, Waltham, MA). Virus titers were determined using a real-time reverse transcriptase PCR (rRT-PCR) assay based on the influenza A matrix gene (52). The rRT-PCR was performed in a LightCycler 480 real-time PCR instrument (Roche Diagnostics, Rotkreuz, Switzerland) using the LightCycler 480 RNA master hydrolysis probe kit (Roche Life Science, Mannheim, Germany). A standard curve was generated using 10-fold serial dilutions from an OH/04 wt virus stock of known titer to correlate quantitative PCR (qPCR) crossing-point (Cp) values with virus titers, as previously described (53). Virus titers were expressed as log₁₀ TCID₅₀/ml equivalents.

Chicken (Poultry Diagnostic and Research Center, Athens, GA), turkey (Poultry Diagnostic and Research Center, Athens, GA), and horse (Lampire Biologicals, Pipersville, PA) red blood cells (RBCs) were prepared from whole blood preparations using standard techniques (54). Virus HA assays were carried out using 0.5% (vol/vol in PBS) chicken RBCs, 0.5% (vol/vol in PBS) turkey RBCs, or 1% (vol/vol in PBS) horse RBCs. Briefly, 50 μl of RBC preparations was added to 50 μl of 2-fold serial dilutions of virus stocks and allowed to incubate at room temperature for 45 min. After incubation, agglutination was measured, and data were expressed as the inverse of the highest dilution that allowed full agglutination.

The receptor cell binding assay was performed as previously described (8, 18, 55). Briefly, 10% (vol/vol in PBS) turkey RBCs were pretreated for 1 h at 37°C with 2-fold serial dilutions of bacterial neuraminidase from Arthrobacter ureafaciens (New England BioLabs, Ipswich, MA) or Clostridium perfringens (New England BioLabs, Ipswich, MA). Following neuraminidase treatment, treated RBCs were washed twice with cold PBS and then resuspended to 1% (vol/vol in PBS), and then 50 μl of 1% RBCs treated with the different neuraminidase concentrations was added to 50 μl of each virus (8 HA units [HAU], as determined on untreated RBCs) and allowed to incubate at room temperature for 45 min. After incubation, agglutination was measured, and data were expressed as the maximal concentration of neuraminidase that allowed full agglutination.

The receptor-binding specificity was determined in a solid-phase direct binding assay using monospecific preparations of horseradish peroxidase (HRP)-conjugated fetuin (Fet-HRP). Monospecific preparations of Fet-HRP were synthesized using α-2,3-sialyltransferase from Pasteurella multocida (Sigma, St. Louis, MO) for 3-modified fetuin (3-Fet-HRP) or α-2,6-sialyltransferase from Photobacterium damsela (Sigma, St. Louis, MO) for 6-modified fetuin (6-Fet-HRP), essentially as described previously (24). Ninety-six-well native fetuin-coated flat-bottom plates (Greiner Bio-One, Monroe, NC) were incubated overnight at 4°C with 128 HAU of each virus in 0.02 M Tris-buffered saline (TBS), pH 7.2 to 7.4. Virus samples were run in duplicate. Plates were washed three times with PBS and blocked with blocking solution (BS) (PBS containing 0.1% neuraminidase-treated bovine serum albumin [BSA-NA]) for 2 h at room temperature. After blocking, plates were washed twice with ice-cold washing solution (WS) (PBS containing 0.02% Tween 80) and incubated with 2-fold serial dilution of 3-Fet-HRP (SAα2-3Gal) or 6-Fet-HRP (SAα2-6Gal) in reaction solution (RS) (PBS containing 0.02% Tween 80, 0.1% BSA-NA, and 2 μM oseltamivir carboxylate) for 1 h at 4°C. After incubation, plates were washed five times with ice-cold WS before adding freshly prepared substrate solution (SS) (0.01% 3,3′,5,5′-tetramethylbenzidine in 0.05 M sodium acetate with 0.03% H₂O₂). Reactions were allowed to proceed at room temperature for 30 min unless otherwise stated. Reactions were stopped with 3% (vol/vol in double-distilled water [ddH₂O]) H₂SO₄. Absorbance readings were obtained at 450 nm using a Victor ×3 multilabel plate reader (PerkinElmer, Waltham, MA).

Virus stocks were grown in MDCK cells, clarified by low-speed centrifugation, and inactivated by treatment with 0.1% β-propiolactone (BPL) for 1 day at 4°C. Inactivated virus stocks were concentrated as described previously (56). Concentrated virus stocks were resuspended in PBS with 5% glycerol to 1280 to 5120 HAU, aliquoted, and stored at −80°C. HA titers of concentrated virus stocks were determined by HA assay using 0.5% (vol/vol in PBS) turkey RBCs. Glycan array analysis was performed using an N-hydroxysuccinimide (NHS) ester-coated glass microarray slide containing six replicates of 128 diverse synthetic sialic acid-containing glycans, including terminal sequences as well as N-linked and O-linked glycans found on mammalian and avian glycoproteins and glycolipids (25). Whole influenza virus samples were diluted to 256 HAU in PBS containing 3% BSA (PBS-BSA) and incubated on the array surface for 1 h at room temperature in a humidity-controlled chamber. After incubation, slides were washed in PBS and incubated with OH/04 wt swine antiserum (18) diluted 1:200 in PBS-BSA for 1 h. Slides were washed in PBS and incubated for 1 h in goat anti-pig IgG conjugated to fluorescein isothiocyanate (FITC) (Thermo Fisher Scientific, Waltham, MA) diluted in PBS-BSA (20 μg/ml, final concentration). Slides were washed twice in PBS and in distilled water (dH₂O) and then dried prior to detection. Slide scanning to detect bound virus was conducted using an InnoScan 1100AL (Innopsys, Carbonne, France) fluorescent microarray scanner. Fluorescent signal intensity was measured using Mapix (Innopsys, Carbonne, France), and the mean intensity minus mean background of 4 replicate spots was calculated. A complete list of the glycans present in the array is presented in Table S1 in the supplemental material. The array is comprised of nonsialoside control (1 to 10; gray), SAα2-3Gal (11 to 76; yellow), and SAα2-6Gal (77 to 128; green) glycans. Glycans are grouped by structure type: L, linear; O, O linked; N, N linked; and L^x, sialyl Le^x.

Swine antisera against A/swine/New York/A01104005/2011 (H3N2) (NY/11), A/swine/Iowa/A01480656/2014 (H3N2) (IA/14), and A/turkey/Ohio/313053/2004 (H3N2) (OH/04) were generated in previous studies (18, 32). For each virus, two pigs were primed and boosted intramuscularly with UV-inactivated whole-virus vaccine prepared with commercial oil-in-water adjuvant (1:5 ratio) (Emulsigen D; MVP Laboratories, Inc., Ralston, NE). The pigs were humanely euthanized for blood collection.

Hemagglutination inhibition (HI) assays were performed as previously described (54). Prior to HI testing, swine antisera were treated overnight with receptor-destroying enzyme (Denka Seiken, Tokyo, Japan) and heat inactivated at 56°C for 30 min. Serial 2-fold dilutions starting at 1:10 were tested for the ability to inhibit the agglutination of 0.5% turkey RBCs with 4 HAU of each virus. HI titers were recorded as the inverse of the highest dilution that inhibited hemagglutination.

For enzyme-linked immunosorbent assay (ELISA), 96-well flat-bottom plates (Greiner Bio-One, Monroe, NC) were incubated overnight at 4°C with 16 HAU of each virus in 1× coating solution (Seracare, Milford, MA). Virus samples were run in duplicate. Plates were blocked with StartingBlock (PBS) blocking buffer (Thermo Fisher Scientific, Waltham, MA) for 1 h at room temperature. After blocking, plates were washed three times with PBS containing 0.05% Tween 20 (PBS-T). Twofold serial dilutions of swine antisera were added and allowed to incubate for 1 h at room temperature. After incubation, plates were washed three times with PBS-T before adding the secondary goat anti-swine IgG-HRP HRP-conjugated polyclonal antibody (Seracare, Milford, MA). Plates were incubated at room temperature for 1 h. After incubation, plates were washed three times with PBS-T before treatment with freshly prepared SS buffer or 2, 2'-azino-di(3-ethylbenzthiazoline-6-sulfonate) (ABTS) 1-component microwell peroxidase substrate (Seracare, Milford, MA) for 1 h at room temperature. The reaction was stopped by adding 3% (vol/vol in ddH₂O) H₂SO₄ or ABTS peroxidase stop solution (Seracare, Milford, MA), respectively. Absorbance readings were obtained at 405 nm using a Victor ×3 multilabel plate reader (PerkinElmer, Waltham, MA).

All statistical analyses were performed using the GraphPad Prism software, version 7 (GraphPad Software Inc., San Diego, CA). For multiple comparisons, one-way or two-way analysis of variance (ANOVA) followed by a post hoc Tukey test was performed. When indicated, a P value below 0.05 (P < 0.05) was considered significant.

A model of the structure of the HA of A/turkey/Ohio/313053/2004 (H3N2) was built by homology modeling using Modeller v9.16 (57) based upon the crystal structures of multiple H3 HA proteins (Protein Data Bank [PDB] codes 2YP7, 1HA0, 2YP2, 4WE8, and 4WE5). The generated model was subsequently rendered with PyMOL v2.1 (58).

The frequency distribution of amino acid identities at residue 145 (H3 numbering) was computed using the sequence variation analysis tool in the Influenza Research Database (IRD) (59). HA amino acid sequences from IAVs of the H3 subtype isolated from swine, human, avian, canine, and equine hosts and publicly available in the IRD as of 6 September 2017 were analyzed. For humans, the frequency was calculated from precomputed data in the IRD database. Only amino acids that reached a frequency of at least 1% are labeled in the plot legend.