Human Immunity and Susceptibility to Influenza A(H3) Viruses of Avian, Equine, and Swine Origin

Influenza A viruses (IAVs) of subtype H3 that infect humans are antigenically divergent from those of birds, horses, and swine. Human immunity against these viruses might be limited, implying potential pandemic risk. To determine human risk, we selected 4 avian, 1 equine, and 3 swine IAVs representing major H3 lineages. We tested serum collected during 2017–2018 from 286 persons in Belgium for hemagglutination inhibiting antibodies and virus neutralizing antibodies against those animal-origin IAVs and tested replication in human airway epithelia. Seroprevalence rates for circulating IAVs from swine in North America were >51%, swine in Europe 7%–37%, and birds and equids ≤12%. Replication was efficient for cluster IV-A IAVs from swine in North America and IAVs from swine in Europe, intermediate for IAVs from horses and poultry, and absent for IAVs from wild birds and a novel human-like swine IAV in North America. Public health risk may be highest for swine H3 IAVs.

I nfluenza A viruses (IAVs) of the H3 subtype are endemic to humans, swine, and wild birds; they also cause outbreaks in horses and are often detected in domestic birds. An H3 IAV that crosses the species barrier from animals to humans can result in a pandemic if the virus carries a hemagglutinin (HA) against which humans lack protective antibodies and the virus readily replicates in and spreads among humans. For example, in 1968, transmission of an IAV with an avian-origin H3 HA to humans caused the influenza A(H3N2) pandemic (1).
The natural IAV reservoir is considered to be wild waterfowl, but transmission to domestic poultry is frequent. Avian H3 IAVs are classified as Eurasian and North American lineages, although the HA of these viruses is antigenically closely related (2,3). In contrast, after being introduced to humans in 1968, the HA of human H3 IAVs quickly drifted away from that of the avian precursor IAV. Consequently, contemporary human H3 IAVs are antigenically divergent from those in birds (2). Similarly, avian H3 IAVs were introduced into horses in the 1960s, after which their HA antigenically drifted. That evolution was, however, different and slower than for human H3 IAVs (4). Equine H3 IAVs of Florida clade 1 (FC1) are currently predominant (5). All swine H3 IAVs derived their HA from human IAVs.
H3 IAVs from swine in Europe originated from a human IAV that circulated in the late 1970s. Of the 2 major lineages cocirculating in North America, cluster IV-A was derived from human IAVs from the late 1990s and novel human-like swine H3 IAVs from human IAVs from the early 2010s (6). H3 IAVs undergo slower antigenic drift in swine than in humans. Consequently, persons born after the swine viruses' human ancestor IAV had circulated are unlikely to have cross-reactive antibodies against the swine H3 IAVs. Therefore, with time, human population immunity against swine H3 IAVs decreases, increasing the pandemic risk (7)(8)(9)(10).
The infectious potential of swine H3 IAVs for humans is evident from >400 recorded zoonotic infections in the United States caused by North American cluster IV-A or novel human-like H3 swine IAVs. Four zoonotic infections with H3 IAVs from swine in Europe have also been reported (6,(11)(12)(13). H3 IAVs from equids can infect humans under experimental conditions, but there are no confirmed cases of natural transmission (14). Animal H3 IAVs might, however, become more adapted to humans by accumulating mutations in their viral proteins, reassortment of gene segments with IAVs of different species, or both (6,15). Avian H3 IAVs can infect humans directly or via an intermediate host, such as poultry or swine (2,15). In 2019, an H3N1 IAV that Influenza A viruses (IAVs) of subtype H3 that infect humans are antigenically divergent from those of birds, horses, and swine. Human immunity against these viruses might be limited, implying potential pandemic risk.
To determine human risk, we selected 4 avian, 1 equine, and 3 swine IAVs representing major H3 lineages. We tested serum collected during 2017-2018 from 286 persons in Belgium for hemagglutination inhibiting antibodies and virus neutralizing antibodies against those animal-origin IAVs and tested replication in human airway epithelia. Seroprevalence rates for circulating IAVs from swine in North America were >51%, swine in Europe 7%-37%, and birds and equids ≤12%. Replication was efficient for cluster IAVs from swine in North America and IAVs from swine in Europe, intermediate for IAVs from horses and poultry, and absent for IAVs from wild birds and a novel human-like swine IAV in North America. Public health risk may be highest for swine H3 IAVs. originated from wild birds caused outbreaks at 82 poultry farms in Belgium and 3 in France without infecting humans but was unusually virulent for poultry (16). In 2022, two zoonotic infections with avian H3N8 IAVs were reported (17).
H3 IAVs continue to evolve in each host species. Therefore, frequent re-evaluation of human seroprevalence and replication potential in humans for circulating animal H3 IAVs is recommended. Serum hemagglutination inhibiting (HI) and virus neutralizing (VN) antibodies correlate with protection. Thus, prevalence of such antibodies against animal H3 IAVs in persons of different age groups can be used to estimate the public health risk (18). Recent seroprevalence studies are available for H3 IAVs from swine in North America, but studies in Europe were conducted with samples and IAVs collected before 2011 (7)(8)(9)(10)19). Studies for H3 IAVs from birds and equids are generally lacking, except for a few small-scale studies with historic IAV strains (20)(21)(22)(23)(24). The infectivity of animal H3 IAVs in humans was previously evaluated with mammalian models, outdated IAV strains, or both (15,(25)(26)(27)(28)(29). To help evaluate the public health risk posed by different animal H3 IAVs, we analyzed serum samples collected from persons of different age groups in Belgium for prevalence and titers of HI and VN antibodies against all major circulating swine, avian, and equine H3 IAV lineages. We also assessed the replicative capacity of selected IAVs in human airway epithelia. The Commission for Medical Ethics of the Ghent University Hospital approved the study (approval no. 2017/834).

Human Serum and Tissue Samples
During August 2017-January 2018, we selected 286 serum samples from immunocompetent persons in Belgium born during 1921-2017 who had unknown influenza infection or vaccination history. The male:female ratio was ≈1:1, and we used ≈3 samples per birth year.
From Epithelix Sàrl (https://www.epithelix. com), we purchased human airway epithelia (Mu-cilAir) reconstituted from primary cells of biopsy samples from 6 donors (Table 1). We maintained the tissues at the air-liquid interface with MucilAir culture medium (Epithelix) according to the manufacturer's instructions.
The major target of neutralizing antibodies is HA1. We downloaded the viruses' HA1 nucleotide sequences from GenBank and translated them to amino acids. We aligned sequences with the MUS-CLE algorithm (https://www.ebi.ac.uk/Tools/msa/ muscle) and constructed maximum-likelihood trees by using the Jones-Taylor-Thornton model and the nearest-neighbor-interchange heuristic method in MEGA7 (https://www.megasoftware.net) (33). We determined numbers of identical amino acids in presumed antigenic sites (34) and percentages of amino acid homology between test viruses by using MEGA7 and R version 3.5.3 (The R Foundation for Statistical Computing, https://www.r-project.org).
We received avian IAVs from the Flemish Institute for Biotechnology (Flanders, Belgium) and Ohio State University (Columbus, Ohio, USA), equine IAVs from St. Jude Children's Research Hospital (Memphis, Tennessee, USA), North American swine IAVs from the US Department of Agriculture-Agricultural Research Service (Bethesda, Maryland, USA), and the human IAV from Philipps University Marburg (Marburg, Germany). Viruses for serologic assays and inoculation of MucilAir tissues were grown in MDCK cells; only avian and equine viruses for HI assays were propagated in allantoic cavities of 10-day-old embryonated chicken eggs; and all underwent <4 passages.

Serologic Assays
HI and VN assays for antibodies against each test virus were performed according to standard procedures (35). We performed HI assays with 1% horse erythrocytes for avian and equine IAVs and 0.5% turkey erythrocytes for human and swine IAVs. Antibody titers represent the reciprocal of the highest serum dilution showing complete hemagglutination inhibition of 4 hemagglutinating units of virus (HI) or 50% neutralization of 100 tissue culture infective doses (TCID 50 ) of virus (VN). Starting dilutions were 1:20 in HI and 1:10 in VN. We considered a titer of >40 to be positive.

Virus Replication Kinetics
To standardize the amount of mucus, we washed the apical side of fully differentiated MucilAir tissues (1/ donor/condition) 1 time with culture medium. Three days later, we inoculated the tissues apically with 250 μL of medium (mock inoculation) or IAV at multiplicity of infection 0.01 TCID 50 . After incubating the samples for 1 hour at 34°C and 5% CO 2 , we removed the inoculum and washed the apical side of the tissues 1 time. At 0-4 days postinoculation (dpi), we measured transepithelial electrical resistance (TEER) with a Millicell-ERS2 Voltohmmeter (Merck KGaA, https://www.merckmillipore.com) and took samples for virus titration. For titration, we added 250 μL medium apically, allowed it to equilibrate for 30 min at 34°C, and collected it. We determined TCID 50 titers of inocula and samples by titration on MDCK monolayers, which 5 days later underwent immunocytochemical staining of IAV nucleoprotein, as previously described (36); the start dilution was 1:2.

Statistical Analyses
We used log 2 -transformed antibody titers to calculate geometric mean titers (GMTs) and 95% CIs against each virus for samples from persons each birth decade. We used log 10 -transformed virus titers to calculate the area under the curve (AUC) for each virus in each MucilAir tissue. Samples that tested negative were assigned a titer of half the detection limit (HI 10, VN 5, virus titration 0.65 TCID 50 /mL). We used Kruskal-Wallis and Mann-Whitney U tests to compare antibody titers between viruses for a certain age group or between age groups for a certain virus. We used the same tests to compare AUCs between viruses for a certain tissue or between tissues for a certain virus. We compared proportions of positive samples by using Fisher exact tests. We determined Spearman correlation coefficients (CCs) between HI titers or between VN titers against different viruses by using nonstratified data. For serologic data, we applied the Bonferroni adjustment of the p values and considered adjusted p<0.05 significant. For AUCs, we considered p<0.1 significant. We performed all analyses with R version 3.5.3.

Genetic Relatedness Between Test Viruses
For seroreactivity and replication studies in humans, we selected 9 H3 IAVs from humans, birds, horses, and swine. Their genetic relatedness was determined on the basis of HA1 amino acid sequence homology (    Table 2. Scale bar indicates amino acid substitutions per site.  (Figures 2, 3), and differences in seroprevalence rates between age groups were not significant. GMTs were below the detection limit for all age groups and no antibodies against chG19 were detected in persons born during 1987-2017 (Tables 4, 5).
Overall seroprevalence rates for swMO15 were 76% in HI and 72% in VN (Figures 2, 3). At least 50% of persons in each age group were positive in both HI and VN, with GMTs of ≥35 (Tables 4, 5). Seroreactivity was highest for persons born during 1987-1996, and significant differences in seroprevalence were found only between those in this group and those born during 2006-2017 in HI (97% vs. 59%; p = 0.02) and those born during 1947-1956 in VN (93% vs. 53%; p = 0.04). Seroreactivity was higher against IAVs of the swine H3 lineages that were more recently introduced to swine and peaked among persons born shortly before these introductions.
ChG19 isolated from poultry replicated to titers of up to 3.9 log 10 TCID 50 /mL in bronchial tissue of BD3 without affecting TEER (Figures 4, 5). ChG19 was also detectable in tissues of ND1 and BD1 at 4 dpi.
For the 3 IAVs of wild birds and North American novel human-like swine IAV swMO15, no virus was detectable in any of the tissues except dkUK63, mlOH18, and mlB18 had titers of <2.2 log 10 TCID 50 /mL at 4 dpi in tissues of ND1 and BD1, and an swMO15 titer of 3.0 log 10 TCID 50 /mL was detected at 2 dpi in the tissue of ND2 (Figures 4, 5). Because of large donor-to-donor variation, only a few differences in virus replication AUCs were significant (p<0.1). In nasal tissues, AUCs were significantly higher for swG17 and swIN16 than for all avian IAVs and swMO15. In bronchial tissues, AUCs were significantly higher for HK68 than for all avian IAVs, eqCH18, and swMO15.

Discussion
Antibody titers against animal H3 IAVs in serum samples from humans in Belgium depended on the virus 104 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 29, No. 1, January 2023   Table 2. Am., American; FC1, Florida clade 1; N. Am., North American. †Age at the end of 2017. ‡Human influenza A virus that no longer circulates.   Table 2. TCID 50 , 50% tissue culture infectious dose. . Black dashed lines represent the TEER below which tissue integrity is irreversibly lost (37). Complete isolate names are provided in Table 2. TEER, transepithelial electrical resistance.
horses and poultry, and minimal for H3 IAVs of wild birds and a North American novel human-like swine H3 IAV. Our results for cluster IV-A IAVs from swine in North America and IAVs from swine in Europe are consistent with previous findings in differentiated human (tracheo)bronchial epithelial cells (25,27). However, 1 study also reported efficient replication of a zoonotic novel human-like IAV from swine in North America (27). A previous study with historic strains detected substantial replication of avian H3 IAVs, whereas an equine H3 IAV did not replicate (25). Discrepancies between our findings and previous findings can result from the use of different cell systems and variation in the genetic background of human cell donors or IAV strains (39,40). Antibody titers against swine H3 IAVs reflect cross-reactive titers against human ancestor IAVs. Antibodies against human ancestor IAVs can be deduced from the theory of antigenic seniority: humans are likely to have antibodies against human IAVs that circulated after their birth, with peak titers against IAVs encountered early in life (41), confirmed by our results for HK68. Our findings, together with those of previous studies showing similar seroreactivity against older swine H3 IAVs and their human ancestor IAV (7,9,10), suggest slow antigenic drift of H3 HA in swine and indicate swine as a reservoir for historic human IAVs. We estimate that HA1 amino acid homology between our swine test viruses and their human ancestor is 87%-93%, with 29-32 identical amino acids in antigenic sites ( Figure 6). Although the HA1 sequences of the avian H3 IAVs were more closely related to that of human virus HK68, cross-reactive serum antibody titers were minimal. Accordingly, ferret serum against human H3 IAVs showed low cross-reactivity with avian H3 IAVs (42), which might be caused by a few key amino acid differences between HK68 and avian H3 IAVs. Compared with all avian IAVs, HK68 has 4 mutations in antigenic sites, of which N145S might be of particular relevance. This mutation mediated antigenic cluster transitions for swine and human H3 IAVs (43,44). Furthermore, higher HA glycosylation of human IAVs might mask certain epitopes shared with avian IAVs, preventing humans from raising antibodies against these epitopes. For example, glycosylation at positions 122, 133, and 144 masks epitopes in antigenic site A. In contrast, HA glycosylation patterns for swine IAVs and for the human ancestor IAV are similar (45). For equine H3 IAVs, the lack of cross-reactive antibodies can be explained by the closer relatedness to avian than to human IAVs and substantial antigenic drift in horses after the introduction from the avian reservoir (2,3).
Seroprevalence rates of <12% for avian and equine H3 IAVs suggest that these IAVs pose a high pandemic risk. Comparable seroprevalences of 2%-19% against the 2009 pandemic influenza A(H1N1) virus were detected right before the pandemic started (46). However, more efficient replication of H3 IAVs of swine in human respiratory tissues as opposed to those of birds or horses suggests that swine pose the highest risk for introduction of H3 IAVs to humans. Indeed, 434 human infections with cluster IV-A and novel human-like H3 IAVs from swine in North America and 4 infections with H3 IAVs from swine in Europe have been reported (11)(12)(13), whereas only 2 zoonotic infections with avian H3 IAVs and no zoonotic infections with equine H3 IAVs have been reported. Swine IAVs are derived from past human IAVs, which can explain their higher potential to infect humans. Swine IAVs prefer human-type α-2,6 sialic acid receptors, whereas avian and equine IAVs prefer avian-type α-2,3 receptors. Human cells also support polymerase activity of swine but not avian IAVs (47). In addition, humans frequently encounter dense swine populations and, unlike for horses and poultry, H3 IAVs are 106 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 29, No. 1, January 2023  Table 2.
endemic among swine. Because zoonotic infections generally result from close contact with infected animals, swine IAVs are also more likely than IAVs of horses or birds to be transmitted to humans (48). On the basis of the seroprevalence rates of >30% for persons >16 years of age, swine H3 IAVs are considered a lower pandemic risk (18). They do, however, pose a zoonotic risk to the youngest persons who lack cross-reactive antibodies, which can explain why most human infections with swine H3 IAVs occurred in persons <18 years of age (11)(12)(13). Our results suggest that population immunity will wane over time and that the human population will sooner become fully susceptible to H3 IAVs from swine in Europe than to H3 IAVs from swine in North America. We estimated the infection potential of animal H3 IAVs in humans on the basis of their replicative capacity in nasal and bronchial MucilAir tissues. However, adaptive and some innate immune responses that are not represented in this model might cause more restricted replication of swine, equine, or avian H3 IAVs in vivo. Also, in vitro experiments in differentiated human airway epithelia will, in the best case, reflect only replication efficiency in a single person and are in no way indicative of airborne transmission between humans (49).
In conclusion, our results stress the need to closely monitor circulating H3 IAVs in different animal species and to frequently evaluate humans for antibodies against these IAVs. This need applies especially to H3 IAVs of swine, which seem to pose the highest zoonotic risk.