Epistasis facilitates the long-term antigenic evolution of the influenza B virus hemagglutinin

Technical Summary: Epistasis Facilitates the Long-Term Antigenic Evolution of the Influenza B Virus Hemagglutinin

Problem Statement
Current understanding of viral antigenic evolution is heavily reliant on studies of influenza A(H3N2), which evolves along a relatively linear trajectory in antigenic space. However, the generalizability of these mechanisms to other influenza viruses remains unclear. Influenza B virus (IBV) has circulated in humans for over 80 years, undergoing substantial genetic and antigenic diversification into two distinct lineages (B/Victoria and B/Yamagata), yet its evolutionary constraints and molecular drivers of antigenic drift are less characterized. Specifically, the molecular determinants of antigenic cluster transitions in IBV have largely remained experimentally unvalidated, and the role of epistasis (where the effect of a mutation depends on the genetic backbone) in balancing antigenic novelty with viral fitness in IBV is not well understood.

Methodology
The authors conducted a comprehensive phylogenetic and experimental characterization of the IBV hemagglutinin (HA) spanning 81 years of human circulation (1940–2021).

• Data Collection: The study analyzed 405 HA sequences and 287 IBV isolates paired with 66 ferret antisera.
• Computational Modeling: A Bayesian multidimensional scaling (BMDS) model was used to jointly reconstruct antigenic phenotypes and evolutionary relationships, identifying 12 antigenic clusters across three lineages (Ancestral, B/Victoria, B/Yamagata). Selection pressure analyses (aBSREL, BUSTED, etc.) and Bayesian Graphical Models (BGM) were employed to identify sites under positive selection and co-evolving pairs.
• Experimental Validation: Reverse genetics systems were used to generate recombinant viruses (and virus-like particles for B/Yamagata) with specific amino acid substitutions or deletions. These variants were tested in Hemagglutination Inhibition (HAI) assays using both ferret antisera and human sera to validate antigenic impacts.
• Fitness Assessment: The replicative fitness of antigenic variants was assessed in air-liquid interface (ALI) cultures of human airway epithelial cells (nasal, large airway, and small airway) and in the respiratory tracts of C57BL/6 mice.

Key Results

• Antigenic Trajectory: Unlike the linear drift observed in A(H3N2), IBV HA evolution followed a non-linear path. The study identified "punctuated" antigenic drift and, notably, an antigenic reversion in recent B/Victoria viruses, where they converged toward antigenic states resembling ancestral viruses from the 1960s.
• Molecular Drivers: The study experimentally validated that antigenic changes are primarily driven by a small set of residues: 136, 150, 162–165, 197, and 203. These residues are proximal to the receptor-binding site (RBS). While insertions and deletions (indels) in the 160-loop (specifically Δ162–164) were also critical, they were found to be less common in A(H3N2).
• Role of Epistasis: The impact of these mutations was highly context-dependent.
  • Mutation Recycling: The re-emergence of past antigenic phenotypes (reversion) was driven by the recycling of mutations at positions 150 and 203.
  • Epistatic Networks: The antigenic effect of a mutation often depended on the presence or absence of other mutations. For example, the deleterious fitness cost of the N150K/K203R combination was rescued by the prior presence of K136E and Δ162–164, which generated a "permissive backbone."
  • Fitness Costs: Mutations that conferred immune escape (e.g., N150K, K203R) often carried significant fitness costs (delayed replication) unless compensated by epistatic interactions.
• Human Serum Correlation: The antigenic effects observed in ferret antisera were largely consistent with reductions in HAI activity in human polyclonal sera, confirming the biological relevance of the identified residues.
• Egg Adaptation: The study noted that egg passage often introduces the N197D mutation (disrupting a glycosylation site), which can alter antigenicity, but confirmed that this change also occurs naturally in circulation.

Significance and Claims
The authors claim that their study provides the most comprehensive characterization of IBV antigenic space to date, moving beyond sequence correlation to experimental validation.

• Generalizability: The findings demonstrate that while IBV shares some evolutionary constraints with A(H3N2) (e.g., reliance on a small number of RBS-proximal residues), it exhibits distinct patterns, including non-linear antigenic paths and the use of indels.
• Epistasis as a Balancing Mechanism: The paper establishes that epistasis is critical for IBV evolution, serving two main functions: (1) increasing the number of accessible antigenic phenotypes through mutation recycling, and (2) rescuing deleterious fitness costs associated with immune escape mutations, thereby enabling long-term viral survival.
• Vaccine Implications: The identification of 10 key amino acid residues (136, 148–150, 162–165, 197, and 203) that underpin antigenic evolution over eight decades suggests these should be prioritized in surveillance and vaccine strain selection.
• Predictive Modeling: The authors conclude that while their genetic, antigenic, and phenotypic datasets support the development of genotype-to-phenotype prediction tools, such tools must accurately capture the complex, non-additive outcomes of epistasis to be effective. They do not claim to have solved the prediction problem but rather highlight the necessity of incorporating epistatic networks into future models.