HA stabilization promotes replication and transmission of swine H1N1 gamma influenza viruses in ferrets

Pandemic influenza A viruses can emerge from swine, an intermediate host that supports adaptation of human-preferred receptor-binding specificity by the hemagglutinin (HA) surface antigen. Other HA traits necessary for pandemic potential are poorly understood. For swine influenza viruses isolated in 2009–2016, gamma-clade viruses had less stable HA proteins (activation pH 5.5–5.9) than pandemic clade (pH 5.0–5.5). Gamma-clade viruses replicated to higher levels in mammalian cells than pandemic clade. In ferrets, a model for human adaptation, a relatively stable HA protein (pH 5.5–5.6) was necessary for efficient replication and airborne transmission. The overall airborne transmission frequency in ferrets for four isolates tested was 42%, and isolate G15 airborne transmitted 100% after selection of a variant with a stabilized HA. The results suggest swine influenza viruses containing both a stabilized HA and alpha-2,6 receptor binding in tandem pose greater pandemic risk. Increasing evidence supports adding HA stability to pre-pandemic risk assessment algorithms.

In addition to pandemics, swine IAVs cause sporadic spillovers from swine to humans. More than 400 cases of swine-IAV infections in humans have been documented between 2005 and 2018 (World Health Organization (WHO), 2018). H1N1-subtype IAVs are currently enzootic in swine, constituting a pool of genetically diverse and continuously evolving strains. Swine H1N1 strains are divided into three major lineages: classical swine (1A), human seasonal (1B), and Eurasian avian (1C) . The classical swine lineage branches into three clades based on early regional considerations: alpha (a; 1A.1), beta (b; 1A.2), and gamma (g; 1A.3) . In approximately 1999, gamma viruses split into two branches: swine gamma (1A.3.3.3) and swine viruses that later contributed the hemagglutinin (HA) gene to the 2009 human pandemic virus (1A.3.3.2). (Gao et al., 2017). Frequently after 2009, HA and other gene segments from H1N1pdm viruses transmitted from humans to swine, generating diverse reassortant viruses. Swine gamma viruses isolated after 2009 contain varying numbers of internal genes originating from human H1N1pdm viruses (Ducatez et al., 2011). Here, we studied H1N1pdm and gamma clade viruses because these two clades contain HA genes that are closely related to the 2009 pandemic and many gamma clade viruses have one or more H1N1pdm internal genes that are already human adapted.
During IAV entry, HA trimers bind to sialic-acid-terminating moieties, triggering endocytosis of virions. Avian IAVs have HA proteins evolved to bind glycoprotein and glycolipid receptors having terminal sialic acids with an aÀ2,3 linkage, and human-adapted IAVs have HA proteins containing receptor-binding pocket mutations that allow higher affinity binding to aÀ2,6-linked sialic acid Xiong et al., 2014;Gambaryan and Matrosovich, 2015). Within minutes after endocytosis, virions are exposed to increasing acidification (Mellman et al., 1986;Cain et al., 1989). Upon reaching a threshold pH that varies by virus and typically ranges from pH 4.8 to 6.2 (Galloway et al., 2013;Russell et al., 2018), spring-loaded HA nanomachines are triggered to irreversibly change their structures and fuse the viral envelope with the endosomal membrane (Skehel and Wiley, 2000). Thus, IAV entry is regulated by HA receptor binding specificity and HA stability.
Avian IAVs typically have HA proteins triggered at pH 5.5-6.2 while H1N1pdm isolates (2010)(2011) have been shown to trend lower at pH 5.3-5.5 (Russell et al., 2018). This suggests that avianto-human adaptation of the HA protein may include a decrease in activation pH. Multiple mechanisms may be operational. HA proteins with a lower activation pH have increased resistance to extracellular inactivation and have increased environmental stability (Poulson et al., 2016;Labadie et al., 2018;Russier et al., 2020). This may be important because the human upper respiratory tract is mildly acidic (England et al., 1999;Man et al., 2017). Intracellularly, opposing intracellular forces may select for intermediate HA stability (Singanayagam et al., 2019), which may help minimize interferon responses both in dendritic cells triggered by high-pH HA proteins (Russier et al., 2020) and in late endosomes triggered by low-pH HA proteins via interactions with IFITM2 and IFITM3 proteins (Gerlach et al., 2017). Consequently, an optimal HA stability is essential for virus replication.
A pandemic-capable IAV requires its HA protein has been unseen immunologically by the majority of the population and is able to support efficient replication and transmission between humans. Pandemic risk assessment tools developed by the Centers for Disease Control and Prevention (CDC) and World Health Organization (WHO) consider virologic, population, and ecological/epidemiological factors (Centers for Disease Control and Prevention (CDC), 2019 and World Health Organization (WHO), 2016). These are called IRAT and TIPRA, respectively. The algorithms explicitly consider HA receptor-binding specificity; however, other virologic properties that may be important are not included or are only considered indirectly by virtue of transmission experiments (reviewed in Russell et al., 2018;Lipsitch et al., 2016;Long et al., 2019b). Presently, HA stability is not considered in risk assessment tools but has been linked to IAV transmission and host adaptation. HA stabilization occurred after H1N1pdm circulation in humans (Maurer-Stroh et al., 2010;Russier et al., 2016) and has been shown to be necessary for airborne transmission of avian H5N1 and 2009 H1N1 pandemic (H1N1pdm) viruses in ferrets (Russier et al., 2016;Imai et al., 2012;Herfst et al., 2012). As swine provide an established pathway for the emergence of pandemic influenza and HA stability is an intrinsic property of the key gene involved in pandemics, we considered it important to determine the impact of HA stability on swine IAV replication and transmissibility.
In this study, we hypothesized that contemporary swine viruses vary in HA stability and strains having more stable HA proteins are more likely to transmit by the airborne route in ferrets. To test this hypothesis, we measured HA activation pH for a panel of H1N1 swine gamma viruses and infected ferrets with genetically related pairs of H1N1 swine gamma viruses that differed in HA stability.

Results
H1N1 gamma isolates from swine had higher HA activation pH values than pandemic isolates from swine and humans To survey HA stability of contemporary H1N1 influenza A viruses (IAVs), we collected 34 swine and 21 human H1N1 IAVs (2009-2016) (Supplementary file 1). Phylogenetic analyses of HA segments showed that 22 swine isolates were gamma clade while 21 human and 12 swine isolates were pandemic (pdm) clade ( Figure 1-figure supplement 1). All isolates were widely distributed within gamma and pdm clades and thus could be taken as genetically diverse representatives of the two targeted genetic clades (Figure 1-figure supplement 1).
We measured HA stability by syncytia assay, recording the highest pH at which virus-expressed HA proteins were induced to cause membrane fusion. A/Tennessee/1-560/2009 (H1N1) (A/TN/2009) had an HA activation pH of 5.5, a moderately stable value previously reported for human pandemic isolates at the start of the 2009 pandemic (Russier et al., 2016;Galloway et al., 2013). The pdm clade viruses isolated from humans and swine after 2009 had HA activation pH values ranging from 5.0 to 5.5 with an average of 5.3 ( Figure 1A, Supplementary file 1). In contrast, gamma clade swine viruses had significantly higher HA activation pH values that ranged from 5.5 to 5.9 and averaged 5.7 ( Figure 1A, p<0.0001 compared to the pdm clade). Overall, pdm-origin HA proteins were more stable than swine gamma, and HA stability for the two clades overlapped at pH 5.5.

Two similar gamma virus pairs differed in HA stability
For further work, we selected two pairs of gamma viruses that within each pair contained (a) substantial differences in HA stability, (b) only one amino-acid substitution in HA, and (c) limited amino-acid variations in other genes. Pair 1 consisted of A/swine/Illinois/2A-1213-G15/2013 (G15) and A/swine/ Illinois/2B-0314-P4/2014 (P4), which contain NP and M genes from the pdm clade and the six other genes (including HA) from swine gamma ( Figure 2A). Consensus sequences of these viruses differed at amino-acid positions HA1-210, PA-271, and PB2-648. HA1-210 is located at the HA receptorbinding domain near the trimer interface, where it could potentially modulate HA head dissociation during acid-induced activation ( Figure 2B). Next-generation sequencing (completed after ferret studies) revealed isolate P4 contained approximately 100% HA1-S210 while isolate G15 contained approximately 85% HA1-N210 and 15% HA1-S210 ( Figure 2C). Virus isolates P4 and G15 induced syncytia formation at pH 5.5 and 5.8, respectively, whether the HA and NA genes were expressed from the background of wild-type viruses or 6+2 reassortant viruses containing internal genes from A/TN/2009 ( Figure 3A). Thus, the PA and PB2 variations did not appear to contribute to differences in HA stability. Acid inactivation titrations showed G15 virus was susceptible to inactivation at 0.2 pH units higher than P4 ( Figure 3B). Viruses G15 and P4 grew to similar titers that were nearly 100-fold higher than A/TN/2009 in swine testicular (ST) and MDCK cells ( Figure 3C). The Pair 1 viruses bound to both a2,6and a2,3-linked receptors ( Figure 3D). P4 had higher binding to a2,6 and less binding to a2,3 than G15. In summary, Pair 1 gamma swine viruses were capable of efficient replication in cultured cells and binding to both a2,6and a2,3-linked receptors. However, P4 had an HA activation pH of 5.5 and receptor-binding skewed toward an a2,6 preference, whereas G15 had a substantially higher activation pH of 5.8 and receptor-binding skewed toward an a2,3 preference.
Pair 2 consisted of A/swine/Illinois/2E-0113-P19/2013 (P19) and A/swine/Illinois/2E-0113-P24/ 2013 (P24), which contain PB2, PB1, PA, and M genes from the pdm clade and the four other genes (including HA) from swine gamma ( Figure 2A). Consensus sequences of these viruses had only one major amino-acid variation in the genome, residue HA2-117. HA2-117 is located in helix D of the central triple-stranded coiled coil in the stalk, where it forms part of a pocket engaged by the fusion peptide of an adjacent protomer in the prefusion structure ( Figure 2B). Next-generation sequencing (completed after ferret studies) showed that isolate P24 contained approximately 100% HA2-N117 while isolate P19 contained approximately 83% HA2-T117 and 17% HA2-N117 ( Figure 2D). Virus isolates P24 and P19 caused syncytia formation at pH 5.6 and 5.9, respectively ( Figure 4A). In the background of 6 + 2 reassortment viruses containing internal genes from A/TN/2009, P24 caused membrane fusion at pH 5.6 while P19 was shifted up to 6.0 ( Figure 4A). P19 virus was susceptible to acid inactivation at 0.15 pH units higher than P24 ( Figure 4B). Viruses P24 and P19 grew to similar titers and approximately 10 to 100-fold higher than A/TN/2009 in ST and MDCK cells ( Figure 4C). The Pair 2 viruses bound to both a2,6and a2,3-linked receptors similarly ( Figure 4D). In summary, the Pair 2 viruses replicated efficiently in cell culture and bound similarly to a2,6and a2,3-linked  receptors. However, the Pair 2 viruses differed at stalk residue HA2-117, which resulted in P24 becoming activated at pH 5.6 for membrane fusion while P19 was activated at pH 5.9.

Pair 1 viruses showed consistently efficient transmissibility by contact but varying transmissibility through airborne exposures in ferrets
To investigate the impact of HA stability on virus pathogenicity and transmission, three donor ferrets were intranasally inoculated with 10 6 PFU of isolates G15 and P4, which had a low HA stability (activation pH 5.8) and a moderate HA stability (activation pH 5.5), respectively. Each group was in an isolated cubicle, and the G15 cubicle was processed before decontamination and handling the P4 group. Additionally, each group had three cages containing one donor ferret each. One day . Gene segments from pandemic and swine endemic are marked in black and white, respectively. (B) HA protein structure and locations of HA residue variations. HA1 residue 210 (yellow) is located in the HA receptor-binding domain head (green) in contact with an adjacent protomer (gray). HA2 residue 117 (yellow) is located in the stalk region (red) in contact with the fusion peptide (black) of an adjacent protomer. HA1 residues in the stalk are colored blue, and two protomers of the trimer are colored gray. The structure was generated by Mac PyMOL using A/California/04/2009 (H1N1) (PDB entry 3UBE). (C) Sequence variations between Pair 1 isolates G15 and P4. (D) Sequence variations between Pair 2 isolates P19 and P24. Whole-genome analyses were performed by next-generation sequencing. Amino-acid variations with frequencies greater than 5% are shown in colored pie charts. H3 numbering was used for the HA segment. following inoculation, three naive contact ferrets were cohoused with the donor ferrets, and three airborne ferrets were introduced individually into adjacent non-contact compartments separated by perforated dividers that allowed air flow. All ferrets were monitored for body weight and temperature daily, and nasal washes were collected every other day until day 14. Ferret sera were collected on day 21. The results showed that temperature and body weight changes of ferrets infected with G15 and P4 were limited and similar to each other (Figure 5-figure supplement 1), consistent with swine viruses typically causing only mild disease in ferrets (Russier et al., 2017;Pulit-Penaloza et al., 2018). On day 1, ferrets inoculated with G15 had nasal wash titer of 1.5 Â 10 5 TCID 50 /ml, whereas those from P4-infected ferrets were 10-fold higher (2.2 Â 10 6 TCID 50 /ml, p=0.23) ( Figure 5A). Peak days (day 2.3 and 3.0 for G15 and P4, respectively, p=0.21) and peak titers of infection (6.4 and 6.7 log 10 TCID 50 /ml for G15 and P4, respectively) were similar (Table 1). Moreover, both groups displayed similar contact transmission efficiency (3/3), contact virus growth, and contact peak nasal titers (6.95 and 7.90 log 10 TCID 50 /ml for G15 and P4, respectively, p=0.40) ( Figure 5B). With respect to airborne transmission, only 1/3 ferrets in the P4 group (activation pH 5.5) seroconverted and had detectable nasal wash virus, which peaked at 2.32 Â 10 5 TCID 50 /ml ( Figure 5C-D). The G15 (pH 5.8) airborne group had 3/3 seroconversion and 2/3 nasal virus isolation (peak titers of 2.32 Â 10 6 TCID 50 /ml and 50 TCID 50 /ml). In summary, both virus isolates replicated efficiently in donors, transmitted efficiently to contacts, and replicated to high titers in contacts. Ferrets inoculated with G15 supported efficient airborne transmission (3/3) while those inoculated with P4 had only partial (1/3).

Stabilizing variant HA1-S210 outgrew in the G15-infected and transmitted ferrets
Increased airborne transmission by the G15 group was unexpected, as G15's predominant HA1-N210 population resulted in a higher HA activation pH (5.8) and a decreased binding ratio of a2,6versus a2,3-linked receptors compared to P4. Using viruses collected from ferret nasal washes, we measured HA stability by syncytia assay. For the group inoculated with P4 isolate (pH 5.5), HA activation pH values in donor, contact, and airborne animals remained at an intermediate stability of pH 5.5-5.6 ( Figure 6A). The HA activation pH of the G15 inoculum was 5.8. G15-infected donor and exposed contact ferrets yielded nasal viruses with HA activation pH ranging from 5.6 to 5.9. While three G15-exposed airborne ferrets seroconverted and viruses were recovered from two, only one airborne ferret yielded sufficient virus for phenotypic analysis, which revealed a value of 5.6 on the first day collected (day 8, Figure 6A). Thus, P4 (starting with a moderately stable HA) retained its phenotype after transmission while G15 (starting with an unstable HA) had its HA protein become stabilized during airborne transmission.
We sequenced ferret nasal washes by next-generation sequencing. The percentages of HA variants (synonymous and nonsynonymous) occurring in G15 donor and contact ferrets were significantly higher than those of P4 ( Figure 6B), showing G15 was less genetically stable in ferrets. P4 had several nonsynonymous mutations arise in donor, contact, and airborne ferrets; however, none of these variants increased over time or became dominant (Figure 6-figure supplement 1A-C). G15 also had a variety of minor variants that did not become dominant or increase substantially from day-to-day ( Figure 6-figure supplement 1D-F). As described in Figure 2C, the inoculum of G15 contained 85% HA1-N210 and 15% HA1-S210. HA1-S210 outgrew HA1-N210 within one day in two inoculated ferrets and by 3 days in the remaining ferret ( Figure 6C, Figure 6-figure supplement 1D). In contact and airborne G15 ferrets,>99% variants consisted of HA1-S210. Contamination by P4 (which also contained HA1-S210) was ruled out, as HA1-S210-containing G15-group samples retained identifying PA-P271 and PB2-L648 residues that were not in P4. Collectively, airborne transmission for both Pair 1 groups occurred with a frequency of 4/6 seroconversion (3/6 virus isolation), Binding specificities toward a(2, 6)or a(2, 3)-linked sialic acid receptors. Viruses were inoculated onto fetuin-coated plates. Virus binding affinities toward a2,6or a2,3-linked sialylglycopolymers were measured by solid-phase receptor binding assay. Recombinant A/Puerto Rico/9/1934 (H1N1) with the HA segment from A/ Mallard/Alberta/383/2009 (H5N1) was used as a control. All data were means ± SD of at least two independent experiments. transmitted virus contained HA1-S210, and the HA activation pH of airborne-transmitted virus was 5.5-5.6 upon first isolation. In the G15 group, virus containing the unstable HA1-N210 variation (pH 5.8) was outcompeted and did not airborne transmit in ferrets; instead the minor HA1-S210 (pH 5.5) sub-population was airborne transmitted along with its unique PA-P271 and PB2-L648 variations.

HA stabilization was also favored during infection and transmission with Pair 2 viruses
Compared to HA1-N210, the HA1-S210 subpopulation that rapidly became dominant in G15-inoculated ferrets had increased HA stability and increased binding to a2,6versus a2,3-linked receptors ( Figure 3D). Both these phenotypic changes have been shown to promote IAV replication and transmission in ferrets (Russell et al., 2018). Therefore, we performed a ferret transmission study using Pair 2 isolates P19 and P24 that had only one nonsynonymous variation at HA2-117 that modulated HA stability but not receptor-binding specificity (Figure 4). The results for Pair 2 largely resembled those for Pair 1, albeit with a lower frequency of airborne transmission.
After one day of infection in inoculated donors, the average nasal wash titer of P24 was 250-times higher than that of P19 (3.0 Â 10 7 TCID 50 /ml and 1.2 Â 10 5 TCID 50 /ml, respectively) ( Figure 7A). This difference was not statistically significant (p=0.33), and both groups had similar titers on day 3. Peak titers were attained approximately 2.5 days quicker in the P24 group (p=0.047) but were similar in magnitude for both groups (6.6 and 6.9 log 10 TCID 50 for P19 and P24, respectively, Table 1). P24 donor ferrets had slightly higher body temperatures on several days and small increases in weight loss compared to P24 (Figure 7-figure supplement 1), but these differences were relatively small and both groups had only mild symptoms of infection. Both P19 and P24 transmitted with 100% efficiency by contact ( Figure 7B,D), and virus titers were comparable between groups ( Figure 7B). Airborne transmission was less efficient for these viruses than for Pair 1, as Pair 2 virus isolation and seroconversion were detected for only 1/3 ferrets in the P19 group and 0/3 in the P24 group ( Figure 7C,D).  HA activation pH of viruses recovered from donor and contact ferrets in the P24 group were maintained at pH 5.6-5.7 ( Figure 8A). In the P19 group inoculated with virus having an HA activation pH of 5.9, HA stability in one donor ferret was reduced to pH 5.7 at 3 dpi and was approximately 5.7 in all three contact ferrets upon first isolation ( Figure 8A). Just as with Pair 1 viruses, airbornetransmitted P19 virus had an HA activation pH of 5.6 ( Figure 8A), showing HA stabilization had occurred.
Next-generation sequencing showed the HA gene of P19 contained a higher frequency of variants than P24 ( Figure 8B). Most variations remained minor and did not increase over time for Pair 2 viruses. For example, minor variants P24 NP-I363 and PB2-V661 from the original inoculum ( Figure 2D) were enriched after one day of infection in donors but were not detected on day 3 and day 5 (Figure 8-figure supplement 1A). 2/3 of P24-infected contact ferrets acquired HA1-N16K mutations in the signal sequence on day 6, but this mutation was retained in only one ferret by day 8 with a decreased frequency (Figure 8-figure supplement 1B). All three P19-inoculated donors acquired a PB2-W98R mutation by day 3, but this mutation was less abundant on day 5 and did not  carry forward to the three contact and one airborne ferret (Figure 8-figure supplement 1C-E). As described above, the inoculum of P19 contained 83% HA2-T117 and 17% HA2-N117. HA2-N117 outgrew HA2-T117 in all three inoculated ferrets within 1 day ( Figure 8C, Figure 8-figure supplement 1C). HA activation pH values for 2/3 of the P19 donor ferrets remained at pH 5.9 despite containing a majority of the stabilizing HA2-N117 variant. High-pH syncytia formation for these samples was most likely induced by the remaining minority population of the HA2-T117 variant. Upon first detection in P19 contact ferrets, HA2-N117 was over 90% abundant, and this variant comprised over 99% of the population in the airborne ferret. Contamination by P24 (also contained HA2-N117) was ruled out, as HA2-N117-containing P19-group samples retained identifying nucleotide PA-1749T (P24 had nucleotide PA-1749C at this position). Similar to Pair 1, the dominant population of P19 in the inoculum with an unstable HA protein (pH 5.9, HA2-T117) was outcompeted and did not airborne transmit. Although airborne transmission in the Pair 2 experiment was infrequent (1/6 total), a moderately stable HA (pH 5.6, HA2-N117) was necessary for efficient replication and airborne transmission.

Discussion
This work showed that HA proteins from pandemic-clade human and swine IAVs are relatively stable (activation pH 5.0-5.5) while those from swine gamma viruses are less stable (pH 5.5-5.9). A moderately stable HA protein (pH 5.5-5.6) was necessary for efficient replication and partial airborne transmission of swine gamma viruses in ferrets. Despite replicating to higher levels in MDCK cells than human pandemic viruses, swine gamma viruses with unstable HA proteins (pH 5.8-5.9) were not airborne transmitted and were outgrown quickly by minor variants that contained stabilized HA proteins. Overall, these results are consistent with the hypothesis that HA stabilization is necessary for airborne transmission of swine IAVs in ferrets. The results also suggest that a stabilized HA (pH <5.6) and human-like receptor-binding in tandem are necessary for IAV airborne transmissibility in ferrets.
The HA protein has two primary roles during virus entry: receptor binding and acid-induced membrane fusion. Previous studies have indicated HA binding to aÀ2,6-linked sialic acid receptors and a relatively low HA activation pH may contribute to airborne transmissibility in ferrets (reviewed in de Cauldwell et al., 2014;Russell et al., 2018), a phenotype often considered a predictor of human adaptation. In the present study, ferrets were infected with natural isolates of swine H1N1 gamma viruses that varied in HA stability and had dual affinities toward aÀ2,6and aÀ2,3-linked sialic acid receptors. Viruses containing an unstable HA protein did not transmit by the airborne route, and all viruses isolated from airborne recipients contained stabilized HA proteins. For A/H1N1pdm, a destabilizing mutation (HA1-Y17H, pH 6.0) eliminated airborne transmissibility in ferrets despite the virus retaining aÀ2,6 receptor-binding specificity (Russier et al., 2016). This suggested that aÀ2,6 receptor-binding specificity was insufficient to promote airborne transmission in the absence of a stabilized HA protein. The mutant virus regained airborne transmissibility by acquiring two mutations that lowered its activation pH to 5.3. A/H5N1 viruses with relatively unstable HA proteins (pH >5.5) were engineered to have aÀ2,6 receptor-binding specificity but did not become airborne transmissible in ferrets until acquiring a mutation that decreased the HA activation pH to 5.5 or lower (Imai et al., 2012;Herfst et al., 2012;Linster et al., 2014). In other studies, HA stabilization of A/H5N1 viruses in the absence of aÀ2,6 receptor-binding specificity increased virus growth in the upper respiratory tract of ferrets but was insufficient to promote airborne transmissibility (Zaraket et al., 2013;Shelton et al., 2013). Altogether, these results suggest aÀ2,6 receptor-binding and a stabilized HA protein (pH <5.6) in tandem are necessary for IAV airborne transmission in ferrets; neither property appears sufficient without the other.
For the swine gamma viruses studied here, aÀ2,6 receptor binding and a relatively stable HA protein was necessary for airborne transmissibility in ferrets but was insufficient for robust transmission. The swine gamma viruses also bound aÀ2,3 receptors, which was associated with a loss of airborne transmissibility for A/H1N1/1918 in ferrets (Tumpey et al., 2007), perhaps due to increased binding to human bronchial mucins (Lamblin et al., 2001;Vahey and Fletcher, 2019). A/H7N9 viruses with dual receptor-binding affinities have also been shown to have limited or very poor airborne transmissibility (Richard et al., 2013;Belser et al., 2013), although these viruses may lack other properties necessary for airborne transmissibility such as a stabilized HA protein (Zaraket et al., 2015). Ferrets, humans, and other animals differentially express complex glycopolymers (Walther et al., 2013;Jia et al., 2014;Byrd-Leotis et al., 2014), and mammalian and avian influenza viruses bind differently to both sialylated and phosphorylated, nonsialylated glycans (Byrd-Leotis et al., 2019). Thus, refinements in binding to aÀ2,6-linked sialic acid (and perhaps other) receptors may be needed to enhance airborne transmissibility. For example, airborne transmission by the A/H1N1pdm09 virus was conferred by binding to long-chain aÀ2,6-linked sialic acids even when aÀ2,3 binding was retained (Lakdawala et al., 2015).
In addition to HA receptor binding and stability, other adaptive properties contributing to influenza virus transmission include polymerase activity, virus morphology, and resistance to host-based restrictions to replication (Mänz et al., 2012;Bogdanow et al., 2019;Long et al., 2019a). Such adaptations may also be needed for swine gamma viruses to adapt highly efficient airborne transmissibility in ferrets. In this study, only 1/3 of airborne-exposed ferrets in the P4 group seroconverted while G15-HA1-S210 airborne-transmitted 3/3. Both P4 and G15-HA1-S210 had identical HA sequences (including HA1-S210) and HA stabilities. As the two viruses differed only at amino-acid residues PA-271 and PB2-648, suboptimal polymerase activity is most likely responsible for the limited airborne transmissibility of P4.
Additional supporting data from follow-up studies could support the conclusion of this study that HA stabilization is necessary but not sufficient for airborne transmission of swine gamma viruses between ferrets. Samples obtained from nasal washes in this experiment could be plaque purified to isolate G15-HA1-N210, G15-HA1-S210, and P4-HA1-S210. G15-HA1-N210 and G15-HA1-S210 are identical except for the stability-altering residue at HA1-210; therefore, differences in replication and transmission in ferrets between the two viruses would map exclusively to the HA protein. Comparing infection with G15-HA1-S210 and P4-HA1-S210, which have identical HA genes but differ at polymerase residues PA-271 and PB2-648, may reveal the importance of the polymerase genes in supporting airborne transmission of the related swine gamma viruses.
Swine are susceptible to both avian and human IAVs, in part by supporting replication of viruses that bind to either aÀ2,3or aÀ2,6-linked receptors. We speculate swine may also serve as an avian-human intermediate host for HA stabilization. H1N1 viruses of the classical, Eurasian, and triple-reassortant swine lineages isolated before 2009 have HA activation pH values of 5.4-6.0 (Russier et al., 2016). The present study showed that gamma-clade IAVs isolated 2012-2016 also have moderately stable to unstable HA proteins (pH 5.5-6.0) and that a moderately stable HA protein (pH 5.5-5.6) is necessary for airborne transmission in ferrets. Recently, three H1N1 variant (H1N1v) viruses of the gamma clade that had been isolated from humans were found to have HA activation pH values of~5.7 (Pulit-Penaloza et al., 2018). Moreover, an alpha-lineage H1N2v virus with an HA activation pH of 5.5 promoted airborne transmission in ferrets while delta-clade H1N2 viruses with less stable HA proteins (pH 5.6-5.7) promoted only partial airborne transmission (1 or 2 out of 3) (Pulit-Penaloza et al., 2018). This supports the notion that increased HA stability promotes increased airborne transmissibility in ferrets. As the transmitted viruses collected from airborne recipients in the above study were not phenotyped and genotyped (Pulit-Penaloza et al., 2018), it is possible that the transmitted H1N1v and H1N2v viruses had activation pH values differing from input virus, similar to our results of gamma viruses G15 and P19.
In summary, contemporary swine H1N1 gamma viruses were found to have HA activation pH values higher than those of the currently circulating pandemic-clade, and HA stabilization enhanced gamma virus growth and was necessary but not sufficient for airborne transmission in ferrets. This work, combined with other recent studies (Imai et al., 2012;Herfst et al., 2012;Linster et al., 2014;Russier et al., 2016;Pulit-Penaloza et al., 2018), support the proposition that HA stability in tandem with receptor-binding specificity should be considered in pandemic risk assessment tools. Gamma viruses are one of the most prevalent lineages in swine (Diaz et al., 2017) and are widely distributed in many locations including USA, Mexico, China, Hong Kong, and South Korea Nelson et al., 2015). These viruses have caused sporadic human infections in recent years (World Health Organization (WHO), 2018). Prior studies suggest IAVs endemic in swine reassort readily with 2009 pandemic viruses and many have obtained pandemic internal genes (Rajão et al., 2017;Ducatez et al., 2011;Vijaykrishna et al., 2010). In this study, we found viruses G15 and P4 contained M and NP genes of 2009 human pandemic origin, while P19 and P24 contained M, PB2, PB1, and PA of 2009 human pandemic origin. The harboring of pandemic gene segments by swine gamma viruses increases their potential to evolve into a pandemic virus. For A/H3N2 swine viruses, reassortment with 2009 pandemic genes has been found to be associated with increased pathogenicity (Rajão et al., 2017). We conclude that the evolution of currently circulating swine IAVs, especially gamma clade, should be closely monitored for increased pandemic potential.

Cells and viruses
Madin-Darby canine kidney (MDCK) cells, African green monkey kidney (Vero) cells, swine testicular (ST) cells, and human embryonic kidney (HEK) 293 T cells were maintained in minimum essential medium (MEM, Thermo Fisher Scientific), Dulbecco's Modified Eagle Medium (DMEM, Life Technologies), DMEM, and Opti-MEM reduced Serum Medium (Life Technologies), respectively. All culturing media were supplemented with 10% HyClone standard fetal bovine serum (FBS, Life Technologies) and 1% penicillin/streptomycin (P/S, Thermo Fisher Scientific). All cells were grown at 37˚C with 5% CO 2 (Hu et al., 2016;Hu et al., 2017a). Cell lines were authenticated by short-tandem repeat profiling and tested free of mycoplasma contamination. Fifty-five H1N1 influenza A viruses (IAVs) (2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) were propagated in MDCK cells less than three passages with 1 mg/ml tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin, as described previously (Hu et al., 2016). The four viruses used in ferret experiments (G15, P4, P19, and P24) were recovered from swine and then passaged two times in MDCK cells before the experiments reported here. Recombinant viruses used in this study were generated using reverse genetics using the HA and NA genes from the indicated viruses along with internal genes from A/TN/2009, and were propagated in MDCK cells less than three passages (Hu et al., 2016;Hu et al., 2017b;Russier et al., 2016).

Virus growth assay
Tested viruses were inoculated into MDCK cells or ST cells in six-well plates at a multiplicity (MOI) of 0.01 PFU/cell. The inocula were removed after 1 hr incubation at 37˚C. Infected cells were washed by PBS (pH~7.4) twice and covered by 3 ml MEM culture media supplemented with 1 mg/ml TPCKtreated trypsin. Infected cells were then maintained at 37˚C. At indicated time points, 300 ml of cell culture supernatant was collected and titrated by TCID 50 on MDCK cells (Hu et al., 2017a;Russier et al., 2016).

HA stability
HA stability was measured using syncytia and/or virus inactivation assays. In the syncytia assay, Vero cells in 24-well plates were infected at an MOI of 3 PFU/cell. At 16 hr post-infection (hpi), infected cells were treated with DMEM supplemented with 5 mg/ml TPCK-treated trypsin for 15 min. Subsequently, infected cells were maintained with pH-adjusted PBS buffers ranging from 4.8 to 6.2 for 15 min. After aspiration of pH-adjusted PBS, infected cells were incubated in DMEM supplemented with 5% FBS for 3 hr at 37˚C. The cells were then fixed and stained using a Hema 3 Fixative and Solutions (Fisher Scientific). Photomicrographs of cells containing or lacking syncytia were recorded using a light microscope (Russier et al., 2016;Reed et al., 2010). A baseline of no virus-induced syncytia formation was obtained by exposing the cells to pH 6.2 media, a condition under which Vero cells formed no visible syncytia. Micrographs were scored positive for syncytia formation if a field contained at least two syncytia that had at least five nuclei. HA activation pH values for syncytia assays were reported as the highest pH that induced syncytia as judged by positive scoring. Differences in HA cell-surface expression could affect the number of syncytia observed per field but would not be expected to alter the reported threshold pH level at which syncytia are induced.
To perform the virus inactivation assay, tested viruses were mixed with pH-adjusted PBS buffers in a ratio of 1:100 and incubated at 37˚C for 1 hr. After that, the resulting infectious viruses were titrated by TCID 50 on MDCK cells (Russier et al., 2016;Zaraket et al., 2013).

Hemagglutinin (HA) assay
Serial two-fold dilutions of tested viruses were prepared in a U-shaped 96-well plate (50 ml/well). PBS was used as a negative control (50 ml/well). After that, 50 ml of 0.5% of turkey red blood cells (tRBCs) was added into each well and mixed gently. The plate was then incubated at room temperature for 30 min. The HA titers of the tested viruses were defined as the highest dilutions that tRBC clumping occurred in the bottom of the wells (Webster et al., 2002).

Solid-phase receptor binding assay
Virus receptor-binding specificity was evaluated using a solid-phase receptor binding assay (Zaraket et al., 2015;Russier et al., 2016). 50 ml of tested viruses (32 HA Units) were inoculated into 96-well plates coated with fetuin. After overnight incubation at 4˚C, inocula were removed. The plates were washed by PBS five times and blocked with 200 ml of PBS buffer with 0.1% desialylated bovine serum albumin (Sigma) for 2 hr. The plates were then washed by PBS three times and covered with 50 ml of indicated sialylglycopolymers. After incubation 1 hr at 4˚C, the sialylglycopolymers were replaced with 50 ml of horseradish peroxidase (HRP)-conjugated streptavidin. The plates were maintained at 4˚C for another 1 hr. After that, 50 ml of TMB substrate solution (Thermo Scientific) were added into each well, and the plates were incubated at room temperature for 30 min. Finally, 50 ml of 3% H 2 SO 4 were added and the plates were read at 450 nm to obtain absorbance values on BioTek microplate reader (BioTek Instruments, Inc).

Virus genome sequencing and analyses
Virus whole-genome sequences were obtained by next-generation sequencing (Russier et al., 2017;Zaraket et al., 2015). Viral RNAs were extracted from ferret nasal wash samples (unless otherwise indicated). Subsequently, virus cDNAs were obtained by reverse-transcription PCR using SuperScript III First-strand Synthesis System (Thermo Fisher Scientific). DNA libraries were prepared by PCR amplification using Phusion High-Fidelity PCR Master Mix with HF Buffer (New England BioLabs). The resulting PCR products were purified by QIAquick Gel Extraction Kit (Qiagen) and submitted to the St. Jude Harwell Center for virus whole-genome sequencing. Sequencing data were analyzed using CLC Genomics Workbench version 11.0.1. Variants were reported when the predefined quality scores were met, and they were presented in forward and reverse reads at an equal ratio. All variants presented in this study were supported by at least 10 reads with a minimum frequency of 5%. Heat maps and parts of whole were generated by using GraphPad Prism version 7 software (Graph-Pad Software, San Diego, CA).

Ferret transmission experiments
Thirty-six male Fitch ferrets (five-to six-month-old, purchased from Triple F Farms, Sayre, PA) were verified to be serologically negative for currently circulating IAVs, influenza B viruses, and viruses used in this study by HAI assay. Viruses A/swine/Illinois/2A-1213-G15/2013 (G15), A/swine/Illinois/ 2B-0314-P4/2014 (P4), A/swine/Illinois/2E-0113-P19/2013 (P19) and A/swine/Illinois/2E-0113-P24/ 2013 (P24) were used for ferret experiments. On day 0, 3 naive ferrets each (Donor) were intranasally inoculated with 10 6 PFU of tested viruses in 500 ml and were caged separately. On day 1, another 3 naive ferrets each (Contact) were co-housed with the Donor ferrets. At the same time, 3 naive ferrets each (Airborne) were introduced in adjacent cages. Aerosol exposure was permitted among ferrets within tested viruses but was not allowed between the tested viruses. Ferret body weight and temperature were monitored daily until day 14. Ferret nasal washes were collected every other day until day 14. Additionally, all ferrets were euthanized on day 21 to collect whole blood. Extracted ferret sera were subjected to HAI assay (Zaraket et al., 2015;Russier et al., 2016). During the experiments, temperature and humidity in each cubicle were set and controlled to 22˚C and 40-60%, respectively. Animal studies were performed in compliance with St. Jude Children's Research Hospital Animal Care and Use Committee guidelines.

Hemagglutinin inhibition (HAI) assay
Ferret sera were collected and treated with receptor-destroying enzymes (Denka Seiken Co., LTD). After incubation at 37˚C overnight and treatment at 56˚C for 30 min, the resulting samples were examined with 0.5% TRBCs. The HAI titers were defined as the highest dilutions that completely inhibited hemagglutinin (Zaraket et al., 2015).

Statistical analysis
All data analyses were performed using GraphPad Prism version 7. One-way ANOVA followed by a Tukey's multiple comparisons test, Mann-Whitney U test, or two-tailed Student's t-test were used to determine statistical significances. P values < 0.05 were considered significant.
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.