Hemagglutinin

Hemagglutinin (HA) is an integral Class I homotrimeric transmembrane glycoprotein densely arrayed across the lipid envelope of influenza viruses (family Orthomyxoviridae), where it constitutes approximately 80% of all surface spikes (~300 to 400 trimers per virion) alongside the receptor-destroying enzyme neuraminidase (NA) and the M2 proton channel. Named for its historic capacity to agglutinate vertebrate erythrocytes in vitro, hemagglutinin executes two sequential functions essential for viral infectivity: (1) specific attachment of the virion to terminal sialic acid (N-acetylneuraminic acid, Neu5Ac) glycoconjugates on host respiratory or enteric epithelial cells, and (2) acid-triggered conformational refolding within the maturing endosome after endocytosis, catalyzing the fusion of the viral lipid envelope with the host endosomal membrane to release viral ribonucleoprotein (vRNP) segments into the cytosol.

Among Influenza A viruses, encoded on negative-sense single-stranded RNA segment 4 (~1,778 nucleotides), 18 antigenically distinct hemagglutinin subtypes (H1 through H18) have been characterized and partitioned phylogenetically into Group 1 (H1, H2, H5, H6, H8, H9, H11, H12, H13, H16, H17, and H18) and Group 2 (H3, H4, H7, H10, H14, and H15). Subtypes H1 through H16 circulate predominantly in wild aquatic reservoir birds (orders Anseriformes and Charadriiformes), whereas H17 and H18 are evolutionarily divergent bat-associated subtypes (Artibeus spp.) that dispense with sialic acid binding entirely and instead utilize major histocompatibility complex class II (MHC-II) molecules for host entry. Because the solvent-exposed globular head of HA is the primary target of host neutralizing antibodies, rapid accumulation of point mutations in its antigenic loops and genetic reassortment of segment 4 drive antigenic drift, shift, and original antigenic sin across human seasonal and pandemic influenza waves.

Homotrimeric Organization of HA1 and HA2

Each hemagglutinin monomer is synthesized on rough endoplasmic reticulum ribosomes as an inactive single-chain polypeptide precursor, HA0 (~550 amino acids, ~75 kDa per monomer, ~225 kDa per trimer), which undergoes N-linked glycosylation and trimerization before transport through the Golgi apparatus to the plasma membrane. Crystallographic elucidation at 3.0 Å resolution by Ian Wilson, John Skehel, and Don Wiley (1981) revealed that each cylinder-like spike extends ~135 Å from the viral membrane and consists of two structurally distinct domains created upon post-translational cleavage of HA0:

• HA1 (~328 residues, ~50 kDa): Forms the membrane-distal globular head domain composed of an eight-stranded antiparallel β-sheet jelly-roll motif housing the shallow receptor-binding site (RBS) and the five hypervariable antigenic epitopes (designated sites A–E in H3 and Sa, Sb, Ca1, Ca2, and Cb in H1).
• HA2 (~221 residues, ~25 kDa): Forms the membrane-proximal fibrous stalk domain anchored by a C-terminal transmembrane α-helix and short cytoplasmic tail. The core of the trimeric stalk is a triple-stranded α-helical coiled-coil that cradles the hydrophobic N-terminal fusion peptide (the first ~20–23 conserved apolar residues of HA2) in a buried, metastable conformation at neutral pH.

Even after proteolytic cleavage, the HA1 and HA2 subunits remain covalently tethered by a single conserved interchain disulfide bond (e.g., Cys14 of HA1 to Cys137 of HA2 in H3 numbering).

Trypsin-Like Monobasic vs. Furin Polybasic Cleavage Sites

Cleavage of the precursor HA0 at a specific arginine residue between HA1 and HA2 is mandatory to liberate the free N-terminus of the HA2 fusion peptide; uncleaved HA0 can still bind sialic acid receptors but is fusion-incompetent and non-infectious. The amino acid sequence at the HA1/HA2 cleavage loop is a primary molecular determinant of tissue tropism and systemic virulence:

• Low-Pathogenicity Avian and Mammalian Influenza (Monobasic Site): Seasonal human strains (H1N1, H3N2) and low-pathogenicity avian influenza (LPAI) strains possess a single arginine residue (e.g., Q/E-X-R↓G) at the cleavage junction. This monobasic motif is recognized only by extracellular or membrane-anchored trypsin-like serine proteases—such as human airway trypsin-like protease (HAT) and transmembrane protease serine 2 (TMPRSS2)—whose expression is restricted to the respiratory and avian intestinal epithelia, confining viral replication to localized mucosal surfaces.
• High-Pathogenicity Avian Influenza (Polybasic Site): High-pathogenicity avian influenza (HPAI) strains within subtypes H5 and H7 (such as clade 2.3.4.4b H5N1) acquire insertions of multiple basic arginine and lysine residues (consensus motif R-X-R/K-R↓G) at the HA0 cleavage loop. This polybasic motif is cleaved intracellularly within the trans-Golgi network by furin and proprotein convertase subtilisin/kexin type 5 (PC5/6), calcium-dependent serine endoproteases expressed ubiquitously across all vertebrate tissues. Consequently, budding virions emerge pre-activated and can infect vascular endothelium, myocardium, and central nervous system neurons, driving fatal systemic disease in poultry and mammals.

Sialic Acid α-2,3 versus α-2,6 Linkage Geometry

The host range barrier separating avian influenza reservoirs from human-to-human airborne transmission is governed largely by stereochemical complementarity inside the HA1 receptor-binding site (RBS). The RBS is a shallow pocket at the apex of each HA1 monomer framed by three secondary structural elements—the 130-loop (residues 134–138), the 190-helix (residues 188–194), and the 220-loop (residues 221–228)—overlying a conserved floor of four aromatic and polar residues (Tyr98, Trp153, His183, and Tyr195 in H3 numbering). Although the monovalent dissociation constant (K_d) of a single HA1 site for sialic acid is weak (~1 to 5 mM), simultaneous multivalent engagement of multiple trimers across the virion surface generates high avidity (K_d,apparent ≈ 10⁻⁹ to 10⁻¹¹ M).

Terminal sialic acids on host cell glycoproteins and glycolipids are attached to the penultimate galactose sugar via two principal glycosidic linkages:

1. α-2,3-linked sialic acid (Neu5Ac-α2,3-Gal): Adopts an extended, linear trans conformation with a narrow cone-like topology. This linkage predominates in the intestinal tract epithelia of waterfowl and in the non-ciliated cuboidal bronchiolar and type II alveolar pneumocytes of the human lower respiratory tract. Avian HA subtypes accommodate this narrow cone via conserved Gln226 and Gly228 (in H3/H5/H7) or Glu190 and Gly225 (in H1).
2. α-2,6-linked sialic acid (Neu5Ac-α2,6-Gal): Adopts a bent, folded cis conformation with a wide "umbrella-like" topology. This linkage dominates the ciliated pseudostratified columnar epithelium and goblet mucins of the human upper respiratory tract (nasal mucosa, pharynx, and trachea). Adaptation of avian HA to efficient human respiratory droplet transmission requires amino acid substitutions that widen the binding pocket: Gln226Leu and Gly228Ser in H2 and H3 pandemics (1957 and 1968), or Glu190Asp and Gly225Asp in H1 pandemics (1918 and 2009). Because the tracheal epithelium of the domestic pig (Sus scrofa domesticus) co-expresses abundant α-2,3- and α-2,6-linked sialic acids on the same mucosal surface, swine populations frequently act as an epidemiological intermediate host ("mixing vessel") facilitating co-infection and genomic segment reassortment between avian and mammalian influenza viruses.

The Spring-Loaded Coiled-Coil Refolding Cascade

Following multivalent receptor attachment, the influenza virion is internalized via clathrin-mediated endocytosis or macropinocytosis. As the early endosome matures into a late endosome, vacuolar V-type H⁺-ATPase proton pumps acidify the endosomal lumen from physiological pH 7.4 down to pH 5.0–5.8. Protonation of conserved histidine and acidic salt-bridge residues at the HA1–HA1 and HA1–HA2 interfaces triggers an irreversible, kinetically trapped conformational transition from the metastable pre-fusion state to the thermodynamically minimal post-fusion state:

1. Head Dissociation and Loop-to-Helix Transition: The three globular HA1 heads detrimerize and swing outward while remaining tethered by their disulfide bonds. Simultaneously, an extended interhelical loop (the B-loop, residues 55–76 of HA2) connecting the short α-helix A to the central triple-stranded coiled-coil refolds into a continuous α-helix. This loop-to-helix transition propels the hydrophobic N-terminal fusion peptide ~100 Å upward like a molecular harpoon, inserting it directly into the target endosomal lipid bilayer.
2. Jackknife Hairpin Collapse and Hemifusion: Next, the lower portion of the long HA2 coiled-coil melts around residues 106–112 and folds back 180° in an antiparallel jackknife inversion, packing against the upper triple-stranded coiled-coil to form a stable six-helix bundle (hairpin trimer of hairpins). Because this hairpin places the N-terminal fusion peptide (embedded in the endosomal membrane) and the C-terminal transmembrane domain (embedded in the viral envelope) at the same end of a rigid 70 Å rod, the free energy released by coiled-coil zipping (~40–60 kcal/mol per trimer, concerted across 3 to 5 adjacent HA trimers) pulls the two membranes into direct apposition, overcoming hydration repulsion to form a lipid hemifusion stalk and subsequently an expanding fusion pore.

During this same endosomal transit, protons flow through the viral M2 tetrameric ion channel into the virion interior, dissociating the matrix protein M1 from the eight vRNP segments so they can exit through the HA-generated fusion pore and import into the host nucleus. Finally, during late viral assembly and budding, the functional balance between hemagglutinin receptor avidity and neuraminidase (NA) sialidase activity prevents progeny virions from self-aggregating on sialylated viral glycoproteins or remaining trapped on the infected cell surface.

Why do seasonal influenza vaccines require annual reformulation while stalk-directed antibodies offer broader protection?

The immunodominant globular head domain (HA1) tolerates rapid amino acid substitutions in its surface loops surrounding the receptor-binding site without losing function, enabling escape from preexisting neutralizing antibodies via antigenic drift, shift, and original antigenic sin. In contrast, the membrane-proximal HA2 stalk domain is highly conserved across Group 1 or Group 2 subtypes because its residues must execute the intricate loop-to-helix and jackknife refolding cascade. Broadly neutralizing antibodies (bnAbs) targeting the HA stalk (such as CR6261 for Group 1 or CR8020 for Group 2) lock the trimer in its pre-fusion conformation and prevent acid-induced membrane fusion, forming the structural basis for universal influenza vaccine design.

How does the pH of hemagglutinin activation affect airborne transmissibility in mammals?

Avian influenza viruses replicate primarily in the intestinal tract of waterfowl and typically possess less acid-stable HA trimers that trigger fusion at higher thresholds (pH 5.6–6.0). In the mammalian upper respiratory tract, however, respiratory droplets and nasal mucus are mildly acidic (pH 5.5–6.5); an HA that triggers prematurely above pH 5.5 undergoes irreversible conformational inactivation in extracellular airway mucus or airborne aerosols before reaching a host cell. Mammalian-adapted pandemic strains therefore require stabilizing stalk mutations (such as His110Tyr in H5 HA2) that lower the activation threshold to pH 5.0–5.4.

Why are avian H5N1 infections in humans severe yet rarely transmitted between people?

Human upper airway epithelia (nasal cavity and trachea) express predominantly α-2,6-linked sialic acids, whereas α-2,3-linked sialic acids are restricted to non-ciliated bronchioles and type II alveolar pneumocytes deep in the lower lungs. Wild-type H5N1 hemagglutinin binds α-2,3 linkages, so infection requires deep inhalation of high-titer aerosols into the alveoli, where viral replication and polybasic furin cleavage provoke diffuse alveolar damage and viral pneumonia, but sparse upper-airway replication limits expulsion via coughing or sneezing.

1. Wilson, I. A., Skehel, J. J., & Wiley, D. C. (1981). "Structure of the haemagglutinin membrane glycoprotein of influenza virus at 3 Å resolution." Nature, 289(5796), 366–373. https://doi.org/10.1038/289366a0
2. Bullough, P. A., Hughson, F. M., Skehel, J. J., & Wiley, D. C. (1994). "Structure of influenza haemagglutinin at the pH of membrane fusion." Nature, 371(6492), 37–43. https://doi.org/10.1038/371037a0
3. Skehel, J. J., & Wiley, D. C. (2000). "Receptor binding and membrane fusion in virus entry: the influenza hemagglutinin." Annual Review of Biochemistry, 69(1), 531–569. https://doi.org/10.1146/annurev.biochem.69.1.531
4. Gamblin, S. J., & Skehel, J. J. (2010). "Influenza hemagglutinin and neuraminidase membrane glycoproteins." Journal of Biological Chemistry, 285(37), 28403–28409. https://doi.org/10.1074/jbc.R110.129809