The emergence and molecular evolution of H5N1 influenza viruses in United States dairy cattle

This study reveals that H5N1 influenza viruses (genotypes B3.13 and D1.1) spilled over from wild birds into US dairy cattle in late 2023 and 2024, respectively, followed by cryptic transmission and accelerated evolution driven by relaxed purifying selection and positive selection for adaptation, highlighting the urgent need for intensified genomic surveillance to mitigate human emergence risks.

The Gist

The Big Picture: A Viral "Heist" in the Dairy World

Imagine the flu virus as a master thief. For years, this thief (H5N1) has been robbing bird houses (wild birds and poultry). But in 2024, the thief decided to try a new target: dairy cows.

This paper is like a detective report written by a team of viral forensics experts. They looked at the genetic "fingerprints" of the virus to figure out exactly when the thief broke in, how they adapted to the new house, and whether they are getting better at the job.

Here are the four main chapters of their investigation:

---

1. The Break-In: Two Separate Heists

The researchers found that the virus didn't just jump into cows once; it happened twice with two different "versions" of the virus.

• The First Heist (B3.13): In late 2023, a specific version of the virus (called B3.13) jumped from birds to cows in Texas. The virus was so good at hiding that it started spreading silently among cows for 2 to 3 months before anyone noticed. It was like a burglar who moved into a house, started rearranging the furniture, and only got caught when the police finally checked the security cameras in March 2024.
• The Second Heist (D1.1): About a year later, a different version of the virus (D1.1) jumped into cows in Nevada and Arizona. This one was already famous for causing trouble in birds, but it managed to sneak into cows again, hiding for another 5 to 6 months before being detected.

The Takeaway: The virus isn't just a one-time accident. It seems to be finding ways to jump into cows repeatedly, and it's very good at hiding in plain sight before we catch it.

2. The "Speed Run": Evolution on Fast Forward

Once the virus got inside the cows, something interesting happened: it started evolving faster.

Think of the virus in birds as a car driving on a smooth, well-paved highway. It has to follow strict rules (natural selection), so it doesn't change much. But when the virus moved into cows, it was like the car suddenly drove off-road into a muddy, chaotic field.

• Relaxed Rules: In the cows, the virus faced fewer "rules" (scientists call this "relaxed purifying selection"). It was allowed to make more mistakes and try out more random changes.
• The Result: The virus started mutating and changing its genetic code much faster in cows than it does in birds. It was like the virus was hitting the "turbo button" to figure out how to survive in this new environment.

3. The "Gym Workout": Adapting to the New Host

Because the virus was evolving so fast, it started finding specific "workouts" that made it stronger in cows.

• The HA Protein: This is the virus's "key" that unlocks the cow's cells. The researchers found that the virus started changing the shape of this key specifically to fit cow locks better.
• The Polymerase (The Engine): The virus also tweaked its internal engine (polymerase) to run more efficiently in cow cells.
• The Analogy: Imagine the virus is a runner. In the bird world, it's a marathon runner. When it got to the cow world, it realized it needed to be a sprinter. So, it quickly changed its shoes and muscle structure to run faster on this new track.

4. The Future: Why We Should Watch Closely

The paper ends with a warning and a call to action.

• The "Practice Run" Danger: Because the virus is evolving so fast in cows, it is essentially "practicing" how to become a better mammal virus. Cows are mammals, just like humans. If the virus gets really good at infecting cows, it might accidentally learn how to infect humans more easily.
• The Immune Pressure: As more cows get infected (or vaccinated in the future), the virus will have to keep changing to survive. This is like a game of "Rock, Paper, Scissors" where the virus keeps changing its hand to beat the immune system.
• The Solution: We need to keep a very close eye on the genetic code of the virus in dairy cows. If we see it changing in specific ways, we can know early if it's becoming a bigger threat to humans.

Summary in One Sentence

This paper tells us that H5N1 flu viruses have successfully broken into US dairy herds twice, are evolving much faster there than in birds, and are quietly practicing how to become better adapted to mammals, which means we need to watch them like hawks to prevent a future human outbreak.

Technical Summary

1. Problem Statement

Prior to 2024, Highly Pathogenic Avian Influenza (HPAI) H5N1 clade 2.3.4.4b was primarily restricted to wild birds and poultry. In 2024 and 2025, the virus emerged in a novel host: United States dairy cattle. Two distinct reassortant genotypes, B3.13 and D1.1, were detected in cattle.

• B3.13 caused a massive outbreak starting in Texas in March 2024, spreading to 917 premises by the end of 2024.
• D1.1 was detected in Nevada and Arizona in early 2025.
• Knowledge Gap: The evolutionary dynamics of H5N1 in cattle were poorly understood. Key questions included: When did spillover occur? Was there cryptic transmission prior to detection? How do evolutionary rates and selective pressures differ between avian and bovine hosts? Did the virus adapt to the new host?

2. Methodology

The authors employed a comprehensive phylogenetic and phylodynamic framework to analyze whole-genome sequences of H5N1 viruses.

• Data Compilation:
  • Compiled 5,134 B3.13 and 3,177 D1.1 genomes from GISAID and NCBI (as of Sept 2025).
  • Constructed background datasets of avian viruses for comparison.
  • Created concatenated alignments for specific gene segments (PA, HA, NA, MP) to improve phylogenetic resolution, acknowledging that these segments share a common reassortment history (A1 genotype for B3.13).
• Phylogenetic Inference:
  • Used Maximum Likelihood (MAPLE) for tree topology.
  • Used Bayesian Phylodynamics (BEAST X) to estimate time-scaled trees, evolutionary rates, and divergence times. Crucially, the models allowed evolutionary rates to differ between avian and bovine hosts.
  • Analyzed both individual gene segments and concatenated alignments to account for reassortment.
• Selection Analysis:
  • dN/dS Ratios: Estimated the ratio of non-synonymous to synonymous substitutions to measure selective pressure.
  • RELAX Framework: Tested for shifts in selection intensity (relaxation vs. intensification) by comparing avian "background" branches to bovine "test" branches.
  • Site-Specific Selection: Used Robust Counting and FUBAR (Fast Unconstrained Bayesian AppRoximation) to identify specific amino acid sites under positive selection in cattle versus birds.
  • Structural Mapping: Mapped selected sites onto protein structures (HA, NA, Polymerase complex) to infer functional implications.

3. Key Contributions

• Refined Spillover Timing: Provided precise estimates for when H5N1 entered cattle, revealing significant periods of cryptic transmission before detection.
• Host-Specific Evolutionary Rates: Demonstrated that H5N1 evolves significantly faster in cattle than in birds, driven primarily by relaxed purifying selection.
• Genotype-Specific Dynamics: Showed that B3.13 and D1.1 had distinct emergence patterns (single vs. multiple introductions) and different ecological outcomes (B3.13 dominated in cattle but faded in birds; D1.1 dominates birds but has limited cattle spread).
• Adaptation Signatures: Identified specific genomic sites under positive selection in cattle, including those associated with antigenic escape and polymerase adaptation, suggesting active host adaptation.

4. Key Results

A. Emergence and Cryptic Transmission

• B3.13:
  • Spillover Time: The most recent common ancestor (tMRCA) of the cattle outbreak was estimated at mid-December 2023. The spillover event (tPMRCA) likely occurred in late October 2023.
  • Cryptic Period: This implies the virus circulated undetected in dairy cattle for 2–3 months before the first clinical cases in March 2024.
  • Single Introduction: Phylogenetic analysis confirmed a single introduction event from wild birds into cattle, followed by sustained transmission.
• D1.1:
  • Multiple Introductions: D1.1 was already widespread in wild birds. Analysis confirmed independent introductions into dairy cattle in Nevada and Arizona.
  • Cryptic Period: Spillover into Nevada cattle likely occurred between mid-October and early December 2024; Arizona spillover occurred between late August and late December 2024. Both outbreaks involved 1–6 months of cryptic transmission prior to detection.

B. Evolutionary Rates and Selection

• Elevated Evolutionary Rate: B3.13 evolved at a significantly higher rate in cattle compared to birds, particularly in the PA, NA, and MP segments. The HA segment showed similar rates but a distinct shift in selection pressure.
• Relaxed Purifying Selection:
  • The dN/dS ratio was significantly elevated in cattle compared to birds.
  • RELAX analysis confirmed a shift toward neutrality (K < 1) in cattle, indicating relaxed purifying selection. This suggests that constraints on the virus were reduced in the new host, allowing for the accumulation of mutations that might otherwise be deleterious in birds.
  • This pattern was observed in B3.13 (strongly supported) and D1.1 (trend observed, though limited by sample size).

C. Site-Specific Adaptation

• Positive Selection in HA: In cattle, positively selected sites were enriched in the HA1 globular head domain, particularly near the receptor-binding site.
  • Specific sites 104 and 147 (HA) showed repeated substitutions and are associated with antigenic escape.
• Polymerase Adaptation:
  • Site PB2-740 (D740N) was under positive selection in cattle but negative selection in birds. This mutation is known to boost polymerase activity in mammalian cells.
  • Other sites near the polymerase-ANP32 binding interface (e.g., PA-441, PB2-589) showed host-specific selection patterns.
• Convergent Evolution: While some canonical mammalian adaptations (e.g., PB2-E627K) appeared, the study suggests there are multiple evolutionary pathways to fitness in cattle, not just a fixed set of canonical mutations.

5. Significance

• Ecological Shift: The study confirms that dairy cattle are a competent host for sustaining H5N1 transmission, reshaping the understanding of avian influenza ecology.
• Surveillance Implications: The discovery of 2–6 months of cryptic transmission highlights the limitations of current surveillance and the need for intensified genomic monitoring in livestock to detect spillover events earlier.
• Zoonotic Risk: The relaxation of purifying selection allows the virus to explore a wider fitness landscape, potentially accelerating the acquisition of mutations that enhance mammalian transmission or antigenic escape.
• Vaccine and Control: The identification of antigenically relevant sites (HA 104, 147) and the potential for rapid evolution underscore the urgency of developing vaccines for cattle and monitoring for human adaptation.
• Methodological Advance: The study demonstrates the necessity of multi-segment concatenated analyses for accurately reconstructing the emergence of reassortant viruses, correcting previous overestimates of evolutionary rates derived from single-gene analyses.

In conclusion, the paper establishes that H5N1 spillover into US dairy cattle is not a rare anomaly but a recurring event with significant potential for cryptic spread and rapid host adaptation, driven by relaxed selective constraints in the bovine host.