Why One Antibody Rules Them All: Epistasis Blocks Broad Flu Protection

The influenza virus hemagglutinin stem has long been regarded as the most promising target for a universal flu vaccine. Unlike the rapidly mutating head domain that crowns the protein, the stem is structurally conserved across a wide range of viral strains, making it an attractive anchor for antibodies that could neutralize many influenza variants at once. Yet despite more than a decade of intensive searching, only a single human antibody, known as CR9114, has ever been shown to bind the hemagglutinin stems of both influenza A and influenza B viruses, the two branches of the virus family that circulate in humans. A new study published in PLOS Biology now explains why such cross-reactive antibodies are so extraordinarily rare, and the answer lies in the intricate genetic architecture of antibody evolution itself.

The research team, led by Katrine E. Dailey and Nicholas C. Wu, set out to map the evolutionary landscape that shaped CR9114. This antibody is encoded by the IGHV1-69 germline gene, a segment of the human antibody repertoire that is particularly prone to producing stem-directed influenza antibodies. CR9114 acquired its remarkable breadth not at birth but through somatic hypermutation, the process by which B cells randomly mutate their antibody genes after encountering an antigen, followed by selection for variants that bind more tightly. The investigators asked a deceptively simple question: if breadth is so valuable, why has the immune system so rarely stumbled upon it?

To answer this, the researchers performed four separate deep mutational scanning experiments, a high-throughput technique that allows scientists to test tens of thousands of antibody variants simultaneously. They compared the binding affinity landscapes of both the germline precursor and the mature, somatically mutated CR9114 against three distinct hemagglutinin targets: the H1 subtype of influenza A, the H3 subtype of influenza A, and the hemagglutinin of influenza B virus. By systematically mutating every position in the antibody’s binding regions and measuring how each change affected binding to each of the three targets, they generated an unprecedentedly detailed picture of the trade-offs that govern antibody evolution.

The central finding is stark: many mutations that have little effect on binding to H1 hemagglutinin, or that even improve it, are actively detrimental to binding the H3 hemagglutinin or the influenza B hemagglutinin. In other words, the three binding tasks are not compatible at the level of individual amino acid positions. An antibody mutation that helps the immune system track one viral subtype can silently sabotage its ability to recognize another. This creates a rugged fitness landscape in which the path toward broader recognition is littered with valleys of reduced function, valleys that natural selection within a single infected or vaccinated individual has little incentive to cross.

This incompatibility between binding targets is compounded by two well-known but often underappreciated genetic phenomena: epistasis and pleiotropy. Epistasis refers to situations in which the effect of one mutation depends on the presence of other mutations elsewhere in the protein. In the context of IGHV1-69-derived stem antibodies, the study revealed that epistasis is pervasive. A mutation that appears neutral in the germline background may become beneficial or harmful once other somatic mutations have accumulated, meaning that the evolutionary consequences of any single change cannot be predicted in isolation. Pleiotropy, meanwhile, describes the situation in which a single mutation affects multiple traits simultaneously, here the binding affinities to different hemagglutinin types, forcing unavoidable trade-offs.

The practical consequence of these constraints is that the evolutionary trajectory available to a developing B cell is far narrower than the raw diversity of mutations might suggest. Somatic hypermutation generates plenty of variants, but selection can only act on those that maintain or improve binding to the antigen currently driving the immune response. If improving binding to H1 hemagglutinin simultaneously degrades binding to H3 and influenza B hemagglutinins, then a B cell expanding during an H1 infection will be steered away from breadth even if breadth would be valuable in the long run. The immune system, in effect, optimizes locally rather than globally, and the local optima it reaches are frequently narrow.

CR9114, then, is best understood not as a representative example of what stem antibodies can achieve but as an evolutionary outlier, a rare combination of mutations that happened to satisfy the conflicting requirements of binding across both influenza A and B viruses. The deep mutational scanning data show that the mature antibody sits at a point in sequence space that is difficult to reach step by step, because many of the intermediate forms along plausible paths would have lost binding to one or more targets and been eliminated by selection. The rarity of CR9114-like antibodies is therefore not a failure of the immune repertoire to generate diversity but a predictable outcome of the geometry of the binding landscape.

These findings carry significant implications for vaccine design. Many current efforts to build a universal influenza vaccine aim to drive the immune system toward stem-directed antibodies by using engineered immunogens such as chimeric hemagglutinins, in which the head domain is repeatedly swapped to redirect the response toward the conserved stem. The new results suggest that such strategies must contend with the epistatic and pleiotropic constraints documented here. Sequential vaccination with different hemagglutinin subtypes may help, because it can apply selection pressure from multiple targets at once, but the study indicates that the order and combination of immunogens matter enormously, since mutations favored by one antigen may undermine recognition of the next.

The work also provides a mechanistic framework for evaluating antibody repertoires more broadly. If epistasis is prevalent in IGHV1-69-derived stem antibodies, then computational models that treat each mutation independently will systematically mispredict which antibody variants are viable. Deep mutational scanning of this kind, applied across multiple antibody families and viral targets, could help vaccine designers identify which evolutionary paths are accessible and which immunization regimens are most likely to shepherd B cells toward genuinely broad neutralization rather than toward narrow local optima.

Ultimately, the study reframes the quest for a universal influenza vaccine as a problem of navigating a constrained evolutionary landscape rather than simply eliciting more antibodies. The scarcity of antibodies like CR9114 is not an accident but a consequence of the fundamental genetics of antibody maturation, in which the same mutations that confer one binding capability often compromise another, and in which the value of any mutation depends on the company it keeps. Understanding these constraints in molecular detail brings vaccine developers one step closer to designing immunization strategies that can work with, rather than against, the evolutionary rules that govern the human immune response to influenza.

Subject of Research: Evolutionary constraints on broadly neutralizing influenza stem antibodies

Article Title: Epistasis and pleiotropy constrain the evolution of cross-reactive breadth in a broadly neutralizing influenza antibody

Article References: Dailey, K. E., Wang, Y., Teo, Q. W., Wang, C., Lv, H., Lei, R., Tong, M., Rodriguez, L. A., Wang, L., & Wu, N. C. (2026). Epistasis and pleiotropy constrain the evolution of cross-reactive breadth in a broadly neutralizing influenza antibody. PLOS Biology, 24(9), e3004021. https://doi.org/10.1371/journal.pbio.3004021

Image Credits: AI Generated

DOI: 10.1371/journal.pbio.3004021

Keywords: influenza, broadly neutralizing antibodies, CR9114, hemagglutinin stem, epistasis, pleiotropy, deep mutational scanning, IGHV1-69, somatic hypermutation, universal vaccine, PLOS Biology, antibody evolution