🧬 biology

Standing HA phenotypic breadth shapes H5N1 cross-host potential

This study reconstructs the evolution of hemagglutinin (HA) phenotypic traits across thousands of H5N1 viruses to demonstrate that standing phenotypic breadth, rather than simple stepwise mutations, enables the virus to rapidly diversify and adapt to new hosts like dairy cattle, offering a scalable framework for prioritizing cross-host variants.

Original authors: Justin Bahl, M H M Mubassir, Sachin Subedi, Tanin Rajamand, Mohamed Bakheet, Ludy Carmola, Sihua Peng, Rajan Kandel, Guppy Stott, Robert Woods, Stephen Tompkins, Geert-Jan Boons

Published 2026-07-21

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Justin Bahl, M H M Mubassir, Sachin Subedi, Tanin Rajamand, Mohamed Bakheet, Ludy Carmola, Sihua Peng, Rajan Kandel, Guppy Stott, Robert Woods, Stephen Tompkins, Geert-Jan Boons

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine the world of viruses as a massive, bustling city where every building is a different host—birds, cows, humans, and more. In this city, the virus's most important tool is a key called Hemagglutinin, or HA for short. Think of HA as a master key that must fit perfectly into a specific lock on a cell's door to let the virus inside. Usually, bird viruses have keys shaped for bird locks, and human viruses have keys shaped for human locks. For a virus to jump from birds to a new animal, like a cow, scientists used to think it needed to undergo a dramatic, step-by-step makeover, swapping out its old key for a completely new one. They believed the virus had to wait until it found the perfect "switch" mutation to unlock the new door. But what if the virus didn't need a brand new key? What if it already had a whole toolbox of slightly different keys in its pocket, ready to try out on new locks before it even left the bird world? This is the question researchers are asking: does a virus need to be perfectly optimized to jump species, or does it just need enough "flexibility" to figure it out as it goes?

This new study, led by Justin Bahl and his team at the University of Georgia, dives into the chaotic history of the H5N1 bird flu virus as it spread across North America. They looked at about 13,000 different versions of the virus's HA key to see how its shape and behavior changed over time. Instead of just looking for one specific "magic mutation," they mapped out the entire "personality" of these keys—how stable they are, how they wiggle, and how they try to grab onto different types of locks.

The researchers found that when the virus first arrived in North America, it was like a wild explorer, trying out thousands of different key shapes. This created a huge variety of "phenotypes," or physical traits, in the virus population. However, nature soon acted like a strict editor, pruning away most of these variations and leaving only a few survivors. Here is the surprising part: the keys that eventually allowed the virus to infect dairy cows weren't brand new inventions created after the virus jumped into the cows. Instead, these "cow-friendly" keys were already hiding in the bird population, waiting in the wings. The virus didn't need to wait for a single, perfect mutation to happen; it simply drew from the diverse pool of keys it already had.

Once the virus started spreading in cows, it didn't just stick to one shape. It continued to tinker, making its keys slightly more flexible and better at grabbing onto cow-cell locks, even though it never fully changed its basic "bird-like" design. The study suggests that the virus's ability to jump hosts didn't come from a single, dramatic switch, but from a distributed, subtle reshaping of its key's geometry. It's as if the virus didn't forge a new key; it just learned to wiggle its existing key in a way that fit a slightly different lock.

The team used powerful computer simulations to watch these keys in action, seeing how they moved and flexed when they tried to grab onto different types of sugar molecules (the "locks") found on bird cells versus cow cells. They found that the cow-associated viruses developed a bit more "wiggle room" in their binding sites, allowing them to hold onto cow locks a little tighter, even without the classic mutations scientists usually look for. While these findings are based on computer models and need real-world lab testing to confirm, they suggest that we might be missing the danger signs. If we only look for specific, known mutations, we might miss viruses that are already flexible enough to jump species, hiding in plain sight within the natural diversity of bird flu. The paper argues that the real risk might not be the "perfect" virus, but the one with the most adaptable, flexible toolkit.

Technical Summary: Standing HA Phenotypic Breadth Shapes H5N1 Cross-Host Potential

Problem Statement
The emergence of highly pathogenic avian influenza (HPAI) H5N1 clade 2.3.4.4b in North America, particularly its sustained transmission in dairy cattle, presents a critical challenge for pandemic preparedness. A major barrier to assessing cross-host potential is the difficulty in translating naturally occurring genetic variation into quantitative functional phenotypes. Traditional surveillance often focuses on identifying specific "switch" mutations (e.g., canonical receptor-binding substitutions) or relies on high-throughput experimental assays that are limited in scale. Consequently, current frameworks may overlook how standing genetic diversity within viral populations creates a reservoir of functional flexibility that enables host shifts without requiring immediate, stepwise adaptive optimization. There is a need for lineage-resolved approaches that connect naturally sampled sequence diversity to quantitative Hemagglutinin (HA) phenotypes to understand how viruses traverse multidimensional phenotype space over time.

Methodology
The authors constructed a sequence–structure–function landscape for approximately 13,000 U.S. H5N1 clade 2.3.4.4b HA sequences collected between late 2021 and May 2025. The study integrated multiple computational layers:

Phylogenetic Mapping: A time-scaled maximum-likelihood phylogeny was generated using IQ-TREE 2 and TreeTime. Phenotypic traits were mapped onto this tree to track temporal variation.
Structural Modeling: AlphaFold 3 was used to predict 3D structures for 2,195 unique HA amino acid variants. These structures were refined using Rosetta Cartesian relaxation protocols.
Thermodynamic and Stability Analysis:
- Folding Stability: Estimated using Rosetta (cartddg2020) and FoldX across a pH panel to assess pH-dependent stability.
- Receptor Binding: Binding free energies ( $\Delta\Delta G$ ) for $\alpha2,3$ - and $\alpha2,6$ -linked sialosides were calculated using Rosetta flex ddG and benchmarked against experimental data from Lin et al. (2024).
Molecular Dynamics (MD): All-atom MD simulations (500 ns, triplicate) were performed on representative variants using Amber24 with GLYCAM_06j-1 force fields to analyze receptor-binding site (RBS) flexibility and geometry.
Sequence-Derived Metrics: Biophysical properties (net charge, hydrophobicity, secondary structure) were calculated using Biopython.
Diversity Quantification: Phenotypic diversity was measured over time using a Shannon Hill number (effective bins) derived from quantile-based binning, controlling for sampling bias via rarefaction.

Key Results

Phenotypic Expansion and Pruning: Geographic expansion in avian hosts (2022–2023) generated extensive predicted HA phenotypic diversity, which peaked in early 2023. This diversity was subsequently pruned by a selective sweep in early 2024, collapsing the population into two distinct trajectories: an avian-associated lineage (Lineage A) and a cattle-associated lineage (Lineage B).
Pre-existing Cattle-Associated Backgrounds: The HA genetic background of early cattle-associated variants (e.g., V6) was detected circulating in avian and wild-mammal hosts as early as 2023, prior to the dairy outbreak. This indicates that cattle-enriched HA backgrounds were part of the standing avian diversity before spillover.
Distributed Structural Remodeling: Cattle-associated variants (Lineage B) exhibited distinct structural remodeling of the receptor-binding site (RBS) compared to avian variants (Lineage A). This included a shorter pocket tip and greater lateral opening. Crucially, these changes occurred without canonical receptor-switch mutations (e.g., Q226L, G228S).
Dynamic Flexibility and Binding Permissiveness: MD simulations revealed that cattle-associated variants (e.g., V10) displayed increased flexibility in the RBS loops (130-loop, 190-helix, 220-loop) when bound to avian-type ( $\alpha2,3$ ) receptors compared to avian lineage variants. While overall receptor preference remained largely avian-type, later cattle-associated variants (V8–V10) showed a progressive energetic shift toward increased permissiveness for $\alpha2,6$ -linked sialoside engagement.
Distributed Energetic Changes: The shift in receptor engagement was driven by distributed changes in the interaction network around the binding pocket rather than a single adaptive substitution. Energy decomposition showed that while conserved residues maintained core binding, surrounding residues contributed more strongly to $\alpha2,6$ binding in cattle variants.

Significance and Claims
The paper argues that cross-host potential may depend less on the accumulation of peak-fitness mutations in a single host and more on the "standing functional breadth" of circulating viral populations. The authors claim that:

Ecological Fitting: Host transitions can draw on pre-existing functional breadth within viral populations (ecological fitting) rather than arising solely through stepwise adaptive mutations immediately preceding emergence.
Limitations of Marker-Based Surveillance: Relying solely on canonical genetic markers (like specific receptor-switch mutations) may systematically overlook variants with latent host-range capacity that utilize distributed structural and energetic changes.
Scalable Framework: The integrated framework linking sequence variation to predicted protein properties provides a scalable strategy for prioritizing HA variants with cross-host-relevant features for targeted surveillance and experimental follow-up.
Reframing Risk Assessment: The study suggests shifting risk assessment from detecting signatures of prior adaptation to quantifying latent functional flexibility, allowing for the identification of variants with cross-host potential before they rise to dominance.

The authors explicitly state that these findings are based on predicted phenotypes and require experimental validation (e.g., glycan binding assays, crystallography). They do not claim to demonstrate ecological fitting experimentally but provide population-level evidence consistent with this interpretation, supported by recent independent experimental studies of H5N1 in cattle. The study focuses on HA and does not evaluate contributions from other viral genes (e.g., NA, PB2) or other host-range barriers.

More like this