Epitopes in the HA and NA of H5 and H7 avian influenza viruses that are important for antigenic drift

Abstract Avian influenza viruses evolve antigenically to evade host immunity. Two influenza A virus surface glycoproteins, the haemagglutinin and neuraminidase, are the major targets of host immunity and undergo antigenic drift in response to host pre-existing humoral and cellular immune responses. Specific sites have been identified as important epitopes in prominent subtypes such as H5 and H7, which are of animal and public health significance due to their panzootic and pandemic potential. The haemagglutinin is the immunodominant immunogen, it has been extensively studied, and the antigenic reactivity is closely monitored to ensure candidate vaccine viruses are protective. More recently, the neuraminidase has received increasing attention for its role as a protective immunogen. The neuraminidase is expressed at a lower abundance than the haemagglutinin on the virus surface but does elicit a robust antibody response. This review aims to compile the current information on haemagglutinin and neuraminidase epitopes and immune escape mutants of H5 and H7 highly pathogenic avian influenza viruses. Understanding the evolution of immune escape mutants and the location of epitopes is critical for identification of vaccine strains and development of broadly reactive vaccines that can be utilized in humans and animals.


Introduction
Influenza A virus (IAV) haemagglutinin (HA) and neuraminidase (NA) continue to e volv e antigenicall y to e v ade r ecognition by the host immune response .T he two major types of antigenic evolution of IAV are antigenic shift and antigenic drift (Webster et al. 1992 ), whic h ar e driv en by selection of esca pe m utants by antibody and cellular immunity.Antigenic shift occurs when the HA gene of a circulating virus is replaced by a novel HA gene segment (genetic shift), negating the ability of the host immune response to recognize the antigenically novel HA.Antigenic shift can arise from in toto infection of a novel influenza virus or from reassortment e v ents in co-infected cells.Antigenic drift describes the accumulation of mutations in the HA and/or NA genes that cause amino acid changes that enable IAV to e v ade host immunity.As part of pandemic preparedness frameworks, the genetic and antigenic evolution of highly pathogenic avian influenza viruses (HPAIVs) is closel y monitor ed to ensur e that HPAI candidate vaccine viruses (CVVs) are protective against circulating viruses, and to inform the selection of updated CVVs if necessary (World Health Organization 2022 ).
The HA gl ycopr otein structur e is composed of a membr ane distal globular head, the membr ane pr oximal stem domain, flexible linker, tr ansmembr ane domain, and cytoplasmic tail (Wilson et al. 1981, Benton et al. 2018 ).The receptor-binding domain (RBD) is located within the HA global head.Structur al featur es of the HA RBD include the and 130-loop , 150-loop , 190-helix, and 220-loop (Wu and Wilson 2020 ).Five antigenic regions on the globular head of H3 HA (antigenic sites A-E) were identified as primary targets of the host antibody response (Wiley et al. 1981 ).Analogous sites wer e subsequentl y identified on H5 (antigenic sites 1-5) (Philpott et al. 1989 ) and H1 HAs [antigenic sites Sb, Sa, Ca (which includes subsites Ca 1 and Ca 2 ), and Cb] (Gerhard et al. 1981, Caton et al. 1982 ) (Figs 1 and 2 ).Work with H3 and H1 seasonal influenza has informed our understanding of antigenic sites the H5 and H7 HAs and avian influenza NAs.
In silico studies have revealed numerous H5 and H7 HA residues that are under positive selection pressure in avian hosts (K osako vsky P ond et al. 2008, Duvvuri et al. 2009, Xiong et al. 2019 ) ( Table S1 ).Additional work examining positive selection pressure of human origin H5 HPAIVs identified similar amino acid subsites to those selected in avian systems (Duvvuri et al. 2009 ).Most of these residues cluster in antigenic sites A and B, and to a lesser extent, antigenic site D, suggesting that antigenic sites A and B ar e pr edominatel y tar geted by the host imm une r esponse and likely to play a crucial role in the antigenic evolution of avian influenza virus (AIV).Inter estingl y, some of the codons encoding these r esidues ar e under positiv e selection pr essur e in human H3 IAVs (Bush et al. 1999 ), suggesting that these sites ar e br oadl y integral to the ability of IAVs evade host immunity.Antigenic drift and associated antigenic cluster transitions of human H3 viruses Protein sequences were downloaded from Influenza Research Database (Zhang et al. 2017 ), and sequences with ambiguous base calls, labor atory-gener ated, and duplicate sequences were omitted ( n = 4971).Receptor column: (1) shaded rows indicate residues surrounding receptor binding site (Yang et al. 2016 ), (2) delta ( ) symbol indicates residues crucial to receptor specificity (Stevens et al. 2006 ), and (3) phi ( ) symbol indicates residues crucial to H3 antigenic cluster transitions (Koel et al. 2013 ).H1 (Ca, Cb, Sa, Sb) and H3 (A-E) HA1 antigenic r egions ar e indicated (Yang et al. 2016 ).Finally, H3, H5, and H7 numbering for each residue is indicated.from 1968 to 2003 were associated with se v en r esidues located adjacent to the RBD in antigenic site A, residue 145, and antigenic site B, residues 155, 156, 158, 159, 189, and 193 (H3 numbering) (Koel et al. 2013 ).Of these se v en r esidues, amino acids 145, 158, 159, and 193 (H3 numbering; H5: 141, 154, 155, and 189) have been identified as being under positive selection in H5 AIVs (Duvvuri et al. 2009 ).
Substitutions in antigenic site A likely play a major role in the antigenic drift of H7 AIVs.Computational analysis of Eurasian H7s antigenic epitopes r e v ealed higher substitution rates in epitopes A and B, whereas substitutions in North American H7s were higher in epitopes B and C (Liu et al. 2015 ).Following natural infection of humans with H7N9 AIV, isolated mAbs pr edominantl y bound antigenic site A or trimer interface site II, although mAbs that had haemagglutination inhibition (HI) activity predominately bound to antigenic sites A and B (Gilchuk et al. 2021 ).Characterization of a panel of murine mAbs r aised a gainst H7N9 AIV demonstr ated that all mAbs with neutralizing activity primaril y tar geted antigenic site A and to a lesser extent, to antigenic sites A and D (Ito et al. 2019 ).Substitutions in antigenic site A also play a role in antigenic evolution of H5 AIVs, although subsites are influential, they may not be the immunodominant antigenic epitope.An early study mapping the antigenic landscape of American-lineage H5 AIV described escape mutants with substitutions in antigenic site A (R122Q, S145P), although the frequency was lower than escape mutants containing substitutions in antigenic site B (Philpott et al. 1989(Philpott et al. , 1990 ) ).
Although the majority of H7 antigenic evolution studies have been performed using H7N9 AIVs and human or murine mAbs, one study has examined the antigenic evolution of Americanlineage H7N2 (A/turk e y/New York/4550-5/1994) using chicken pol yclonal antiserum.An esca pe m utant containing G129E substitution (plus others in antigenic sites D and E) (immature protein: G137E; H5: G124E; H7: G119E) a ppear ed following selection with c hic ken pol yclonal serum (Sitar as et al. 2020 ).Substitutions at subsite 129 have been fr equentl y r eported (Kav erin et al. 2002, He et al. 2013b, Henry Dunand et al. 2016, Timofee v a et al. 2020a ) (He, immatur e pr otein: G137R; Henry Dunand, immatur e pr otein: G137E; Timofee v a, H3 numbering: D129N)-this subsite is located in a r ecentl y identified epitope that is either partly or transiently exposed on the pre-fusion conformation of HA (Turner et al. 2019 ).
Alter ed r ece ptor binding d ynamics have been described for H5 esca pe m utants with substitutions at amino acid 129 (Ilyushina et al. 2004 ).Further studies selecting variants using avian polyclonal antisera would be of interest as this likely more closely reflects selection pr essur e in the natur al host.
Inter estingl y, phenotypic attributes conferr ed by one amino acid substitution may not be conferred if another amino acid was selected for at the subsite.Although an American-lineage R144G (H5: R140G; H7: R133G) esca pe m utant described abov e had increased affinity to numerous α-2,3 receptor analogues, subsequent phenotypic c har acterization of a gs/Gd-linea ge esca pe m utant containing the N144S substitution (H5: N140S; H7: N133S) had reduced binding to 3 SLN-PAA receptor analogue and to c hic ken erythr ocytes, and exhibited a r eduction in thermostability compared the parental strain (An et al. 2019 ).Another study c har acterizing Eur asian non-gs/Gd-linea ge esca pe m utants containing S145P/Y substitutions (H5: 141; H7: 134) demonstrated that HA thermostability of esca pe m utants was dependent on this amino acid (Timofee v a et al. 2020a ), and HI activity has been shown to be modulated by the amino acid present at subsite 57 (immatur e pr otein: 65) (Henry Dunand et al. 2016 ).These results support earlier work with H1 IAVs demonstrating that phenotypic attributes, such as antigenic escape, mediated by one amino acid may not be conferred if another amino acid is substituted at the antigenic subsite (Doud et al. 2017 ).
Collectiv el y, substitutions at subsites 144 and 145 ar e highl y influential on the antigenicity of H5 and H7 AIVs and are likely to be important subsites contributing to AIV antigenic evolution.

(Figs 1 and 2 ).
A seminal study mapping the antigenic landscape of the head domain of a North American-lineage H5N9 HA identified amino acid 62 (immature protein: 69; H5: 53; H7: 52) as a target of neutralizing antibodies (Philpott et al. 1989(Philpott et al. , 1990 ) ).There was no discernible effect on viral pathogenicity following challenge of chickens with this antigenic site E esca pe m utant, whic h is in contrast to esca pe m utants with substitutions in antigenic site B (Philpott et al. 1989 ). Immunoselection studies with H7N9 have identified Q78R/H substitutions (Ito et al. 2019 ).An H7N2 antigenic escape mutant selected for using polyclonal chicken antisera with a mutation at amino acid 79 (immatur e pr otein: 87; H5: 70; H7: 69) has been described (Sitaras et al. 2020 ).Epitope mapping of an anti-H5 mAb targeting the vestigial esterase domain revealed a role for amino acid 79 (in addition to 62 and 69) for recognition by the mAb, and mutation of this amino acid reduced mAb binding (Paul et al. 2017 ).Finally, gs/Gd H5N1 antigenic escape mutants with substitutions at N81D and P82 A Q (immatur e pr otein: N88D and P90Q) have been described (Höper et al. 2012 ).R81 is the most fr equentl y detected amino acid in curr entl y circulating subclade 2.3.4.4 H5 HPAIVs (Fig. 1 ), although lysine , serine , and aspar a gine are also detected (Fig. S1), and amino acid 81 has shown to be influential in the antigenic drift of 2.3.4.4 HPAIVs (Li et al. 2020a ).

Non-canonical haemagglutinin epitopes
Whilst most described esca pe m utants harbour substitutions at k e y antigenic r egions, imm une esca pe m utants with substitutions at non-canonical antigenic sites are described.These include an American-linea ge H5 esca pe m utant with E46K (associated with A186T Philpott et al. 1989A186T Philpott et al. , 1990 ) ). H7N9 imm une esca pe m utants with K173N and D348N (Henry Dunand et al. 2016 ), A149D (which flanks antigenic site A and it situated in a CD8 + T cell epitope) (Thornburg et al. 2016 ), V309I, R354K, I374N/T (Henry Dunand et al. 2015 ), andR256H (Vasude v an et al. 2018 ) substitutions have been described.Notably, amino acid 256 has been shown to be under positive selection in H5 viruses (Duvvuri et al. 2009 ).gs/Gdlineage H5 escape mutants with (S133P) + H244R + R326G (Nguyen et al. 2017 ) and E368K have been reported (Kalthoff et al. 2013 ).Se v er al esca pe m utants with substitutions in the HA cleav a ge site, the predominant marker of AIV virulence (Luczo et al. 2018 ), have been described including R326G (Nguyen et al. 2017 ), andK327Q andT328K (Lyashko et al. 2024 )-through ampliative reasoning, T328K may be r e v ersion of a mouse-ada pted AIV bac k to the wild type sequence upon pr opa gation in embryonated c hic ken eggs.

Molecular determinants of neuraminidase antigenic drift
Whilst the HA gl ycopr otein is the predominant protective antigen, NA is r eceiving incr easing attention as a pr otectiv e imm unogen.Vaccination of c hic kens with NA can elicit complete protection a gainst H5 HPAIV c hallenge (Webster et al. 1988 ), and mucosal NA immunity has been shown to pr e v ent IAV tr ansmission (McMahon et al. 2019 ).A deletion in the NA stalk, a known marker of poultry ada ptation (Matr osovic h et al. 1999 ) does not affect NA antigenicity (Els et al. 1985 ). NA antigenic epitopes and imm une esca pe m utants are not as well described compared to HA epitopes .Here , we expand our focus to escape mutants generated using avian isolates or isolates that harbour avian-origin NA due to the ability of H5 and H7 AIV to reassort with numerous NA subtypes.N2 numbering is used throughout.

Variable segment I
Variable segment I is located on the lateral solvent exposed surface of N A (Fig. 3 , y ello w residues) and includes residues 328-336.An early study examining pandemic N2 escape mutants (of which the NA is of avian origin) reported D329N, and N334S + K368E in v ariable r egion I following selection using mAbs (Air et al. 1985 ). Additionally, selection of escape mutant with N329D has been described for Austr alian-linea ge N9 isolate (Webster et al. 1987 ).

Variable segment II
Variable segment II is located adjacent to v ariable r egion I on the lateral plane of NA (Fig. 3 , or ange r esidues).Variable seg- ment II is composed of residues 339-347 and incor por ates the 340-loop (r esidues 342-347).Se v er al N2 esca pe m utants with substitutions at residue 344 are described.These include pandemic N2 with R344I (Laver et al. 1982 ), R344K/G/I/T/S (Lentz et al. 1984 ), and R344G/K (Air et al. 1985 ). N9 imm une esca pe m utants from the Yangtze River Delta-lineage ((H7)N9) with N345S and N347S (Xiong et al. 2020 ) and N8 esca pe m utants with N344K and G346R/E (Saito et al. 1994 ) substitutions have been mapped to this variable segment.An N2 escape mutant with R338S substitution is described (Wan et al. 2016 ), suggesting that amino acid 338, which flanks variable segment II, is also a part of variable segment II epitope.Functional consequence of substitutions in variable segment II include alterations to NA thermostability (Laver et al. 1982 ).

Variable segment III
Variable segment III is located on the top solvent exposed surface of NA and incor por ates amino acids 367-370 (Fig. 3 , light blue residues) and features loop 1 of the second sialic binding site, also called the hemabsorbing (HB) site (amino acids 367, 370, and 372) (Varghese et al. 1997 ).The NC-41 epitope partially maps to variable segment III (amino acids 368-372) (Colman et al. 1987 ).
A m ultitude of imm une esca pe m utants ma p to v ariable segment III.Immune escape mutants with substitutions at amino acid 367 include N8 escape mutants with S367N (Saito et al. 1994 ), N9 esca pe m utants with S367N (Webster et al. 1987 ), or S367N/G/R (Air et al. 1990 ), and an (H7)N9 immune esca pe m utant with S367P (Xiong et al. 2020 ).Notabl y, gl ycan shielding seems to be a mechanism of immune evasion at this subsite.
Esca pe m utants with substitutions at amino acid 368 ar e fr equentl y described.They include a N2 esca pe m utants with K368E alone (Laver et al. 1982 ) or in in combination with N334S (K368E + N334S) (Air et al. 1985 ). N9 escape mutants with substitutions at this position include I368R (Webster et al. 1987 ) and (H7)N9 immune escape mutant with T368 L (Xiong et al. 2020 ).
N9 esca pe m utants commonl y r eport substitutions at amino acid 369.These include an N9 esca pe m utant with A369D (Webster et al. 1987 ) and an (H7)N9 immune escape mutant with A369T (Xiong et al. 2020 ).N2 and N9/(H7)N9 escape mutants with S370 L ar e commonl y r eported (Air et al. 1985, Webster et al. 1987, Xiong et al. 2020 ).Amino acid 372 flanks variable segment III and numerous AIV NA escape mutants with substitutions at this position are reported (Webster et al. 1987, Air et al. 1990 ), suggesting that this amino acid is part of the variable segment III epitope.Esca pe m utants with substitutions at this position include N9 esca pe m utants with S372Y (Webster et al. 1987 ) or S372F (Air et al. 1990 ).

Variable segment IV
Variable segment IV is located on the top surface exposed surface of NA (Fig. 3 , gr een r esidues) and includes r esidues that form loop 2 of the HB site (Varghese et al. 1997 ).The NC-41 epitope partially maps to variable region IV (amino acids 400-403) (Colman et al. 1987 ).Ma pping of imm une esca pe m utants suggests that r esidue 399 also forms the variable segment IV epitope.

Variable segment V
Variable segment V is located on the top solvent exposed surface of NA (Fig. 3 , dark blue residues) and is comprised of amino acids 431-434.Adjacent to variable segment V are variable segments III and IV.The 430-loop is located within variable segment V, and it also features loop 3 of the second sialic acid binding site (HB site) (Varghese et al. 1997 ).The NC-41 epitope partiall y ma ps to v ariable r egion IV (amino acids 430-434) (Colman et al. 1987 ).
N9 escape mutants with K432N (Webster et al. 1987 ) or K432E + K435G (Air et al. 1990 ) have been described.N9/(H7)N9 esca pe m utants with substitutions at amino acids 435 and 436, both of which flank variable segment V, suggest that these residues also form part of variable segment V e pitope.In ad dition to the N9 esca pe m utant with a substitution at amino acid 435 described abo ve , (H7)N9 escape mutants with D435E and K436R have been reported (Xiong et al. 2020 ).Finally, a human N1 escape mutant with R430Q suggests that amino acid 430, which flanks variable segment V, also likely forms part of variable segment V epitope.

Variable segment VI
Variable segment VI is located on the lateral surface of NA, along the rim of the active site, and close to a pr otomer:pr otomer interface (Fig. 3 B, pink residues).Variable segment VI includes amino acids 197-199, and the Mem5 Fab footprint maps these amino acids (Venkatramani et al. 2006 ).Residue 198 forms part of the dimensional structure of the active site (reviewed in (Shtyrya et al. 2009 ), and numerous escape mutants with substitutions at these sites have been reported.These include (H5)N1 gs/Gd-lineage esca pe m utant with D198E (Nguyen et al. 2017 ), N2 imm une esca pe mutants with D198N and K199N (Wan et al. 2016 ), N8 immune esca pe m utant with S199P (Saito et al. 1994 ), and an (H7)N9 escape mutant with N198S (Xiong et al. 2020 ).
Amino acids 220 and 221 flank variable segment VI and are likely part of the same epitope, and the Mem5 epitope includes these amino acids (Venkatramani et al. 2006 ).Se v er al imm une esca pe m utants at these positions hav e been r eported, including N6 esca pe m utant with G220E that was associated with a str ong r eduction in NI activity (Strohmeier et al. 2022 ) and an N9 escape mutant with R220Q (Webster et al. 1987 ).A N2 immune escape mutant with D221H substitution has been described (Laver et al. 1982 ).

Variable segment VII
Variable segment VII is located on top of NA, along the rim of the active site, and close to a pr otomer:pr otomer interface (Fig. 3 , r ed residues).Colman et al. described only amino acid 153 as situated in this variable segment.To date, human N1 immune escape mutant with a substitution at this position (S153I) (Yasuhara et al. 2018 ), but not avian, have been described.Ho w ever, numerous avian immune escape mutants with substitution at amino acid 150, which flanks 153, are described.These include an N2 immune esca pe m utant with H150Q/N (Air et al. 1985 ), an N8 esca pe m utant with K150E (Saito et al. 1994 ), and an (H7)N9 imm une esca pe mutant with H150P (Xiong et al. 2020 ).It is likely that amino acids 150 and 153 belong to the same epitope.Notably, the Mem5 Fab footprint maps to amino acid 150 within variable segment VII, and amino acids 147 and 154 that flank variable segment VII (Venkatramani et al. 2006 ), suggesting that the epitope may be larger than initially described.The universally conserved Asn146 glycosite is located near variable segment VII, the absence of which is associated with H1N1 neurovirulence (Li et al. 1993 ) and altered N8 substrate preference (Saito and Kawano 1997 ).

Other neuraminidase epitopes
In addition to the se v en v ariable segments described by Colman et al. ( 1983 ), analysis of antigenic escape mutants suggests the presence of another epitope (Fig. 3 , light purple residues) situated on the NA lateral surface and surrounding the active site .T his additional epitope includes amino acids 245-249 (N1: 231-235).Recent work with N6 and (H7)N9 AIVs has ma pped numer ous substitutions in antigenic esca pe m utants to this r egion.Specificall y, N6 escape mutants contained P245Q, N248S, or R249G/K substitutions in this r egion.N248S r esulted in a significant reduction in neuraminidase inhibition (NI) activity and P246Q and R250G/K resulted in a complete loss of NI activity (Strohmeier et al. 2022 ).The Mem5 Fab footprint maps to amino acids 249 and 251 in this r egion (Venkatr amani et al. 2006 ).Ad ditionally, a stud y examining (H7)N9 escape mutants identified A246G and T247N substitutions in this region (Xiong et al. 2020 ).
Se v er al imm une esca pe m utants with substitutions located on the bottom surface of NA have been described (Fig. 3 B).An early study examining N2 antigenic drift described an escape mutant with R253S substitution (Lentz et al. 1984 ).Escape mutants with substitutions located on the underside of NA include an Eurasianlineage (H5)N3 escape mutant with I257M (Timofeeva et al. 2020a ), a gs/Gd-lineage (H5)N1 immune escape mutant with R130K and T187A (Höper et al. 2012 ), and an N8 escape mutant with N284G (Saito et al. 1994 ).Substitutions located underneath NA have also been described for human N2 (Kirkpatrick Roubidoux et al. 2021 ), suggesting that this region is targeted by both avian and mammalian hosts.

Conclusions
H5 and H7 AIVs continue to e volv e antigenicall y to e v ade the host imm une r esponse .Experimental studies that ha v e gener ated antigenicall y adv anced imm une esca pe m utants hav e pr ovided crucial insights to k e y antigenic epitopes on haemagglutinin and neur aminidase pr oteins.Within the HA gl ycopr otein, H7 esca pe m utants ar e fr equentl y detected with substitutions in antigenic site A, and in H5 esca pe m utants antigenic site B seems to play a major role in the antigenic drift of contemporary isolates.Moreover, the contribution of NA immunity is gaining increased recognition of its importance in anti-IAV immunity, although to date, remains understudied.Studies examining the functional fitness of antigenicall y adv anced esca pe m utants hav e pr ovided further insights as to why certain escape mutations may be selected and the br oad landsca pe of potential mutations among differ ent linea ges.Understanding the molecular determinants of antigenic drift of both HA and NA is crucial to the de v elopment of br oadl y pr otectiv e v accines to combat the threat of H5 and H7 AIVs.Finally, in addition to preserving food security and maintaining domestic and wild animal health, increasing research efforts to understand the emer gence, e volution, and fitness of antigenically novel strains is crucial to understanding risk and pandemic potential.