Timing decides which influenza lineages win, cell-culture study finds

Influenza A virus lineages do not necessarily win out because their genomes are intrinsically fitter. According to a study published in EMBO Reports, a lineage gains disproportionate representation in the next generation of virus particles mainly when it shares a cell with other viral genomes at the same time, during a brief early window before infected cells become resistant to further entry. The finding, from a team led by Alvaro Fajardo and Benjamin tenOever, comes from single-cell transcriptomics combined with a neutral viral barcode system in A549 human lung alveolar cells infected with the A/Puerto Rico/8/1934 (PR8) H1N1 strain.

The starting puzzle was a long-standing tension in influenza biology. The virus evolves through replication errors that generate diversity, but those same errors can trigger cellular antiviral responses. Its segmented genome allows reassortment, yet segmentation also produces infection heterogeneity: many infected cells fail to express a complete set of viral genes because of incomplete genome packaging or defective gene expression. Superinfection exclusion, a time-dependent process by which already infected cells become refractory to secondary entry, further restricts the chance for incomplete infections to be complemented. Together these constraints should limit the consistent expansion of any individual variant, and indeed clear signatures of positive selection have been difficult to detect in real-world household transmission of influenza A virus. Yet the virus continues to adapt in nature, circulating as seasonal strains and repeatedly giving rise to pandemics.

To find the infection contexts that permit disproportionate lineage amplification, the researchers first mapped the infection landscape with single-cell RNA sequencing. A549 cells were infected at a multiplicity of infection of one and profiled at 8 and 24 hours post-infection. Because viral RNA can be carried over by the experimental workflow, the team distinguished bona fide infection from background signal by examining read distributions across the ten major viral open reading frames at single-cell resolution. The distributions were strongly bimodal, and comparisons with naturally occurring incomplete infections, particularly cells lacking segment 5, which encodes the viral polymerase component needed for replication, allowed the researchers to confirm that the higher-expression peaks reflected genuine viral transcription while the invariant lower peak represented carryover RNA.

The single-cell analysis revealed a landscape of marked heterogeneity. Although infection was initiated at a multiplicity predicted to infect roughly 60 percent of cells under a Poisson expectation, the fraction of infected cells remained stable at about 60 percent between 8 and 24 hours post-infection, consistent with limited expansion of the infected-cell population. Interferon engagement was rare: at 8 hours post-infection, interferon-expressing cells accounted for approximately 0.03 percent of infected cells, a frequency the authors attribute to viral antagonism and the stringent requirements of interferon promoter activation rather than a sequencing depth limitation. By 24 hours, the proportion of interferon-expressing cells had increased more than 200-fold, and broader antiviral-state activation, reflected in the interferon-induced gene MX1, was detectable across additional cell populations. The infection landscape, in other words, contained only a small number of interferon-initiating events early on, with antiviral signaling spreading later.

The key methodological advance was a lineage-tracking system built into the virus itself. Using transposon-mediated insertional mutagenesis of genome segments 7 and 8, the team identified a permissive site within the M2 coding region, just downstream of the 3-prime splice acceptor and between residues 24 and 25 of the M2 protein, where a randomized 9-nucleotide barcode encoding three amino acids could be inserted. The resulting library contained approximately 2,000 non-redundant barcodes. Two lines of evidence supported the neutrality of these tags: the encoded amino acids covered all five physicochemical classes in proportions consistent with codon-weighted expectation, and replication kinetics of the barcoded library in A549 cells were indistinguishable from the parental strain in both single-cycle and multi-cycle infection assays. Because the barcode sits in a transcript that accumulates late in infection, its recovery enriches for cells that have progressed beyond initial entry, and the researchers paired single-cell barcode detection with direct sequencing of released progeny to determine which lineages actually gained representation in the output.

In the central experiment, cells were infected with the barcoded library at a multiplicity of 1.5, chosen to generate roughly balanced populations of cells infected by one, two, or three or more viral genomes among infected cells, and analyzed at 10 hours post-infection after sorting for hemagglutinin surface expression. One third of infected cells carried a single barcode, one quarter carried two, and another third carried three or more. When barcode frequencies in the released progeny were compared with the inoculum, enrichment was operationally defined as a reproducible shift in barcode frequency in progeny virions relative to the input. The pattern across replicates was strikingly stochastic: 62 percent of enriched barcodes appeared in only a single biological replicate, 26 percent in two, 9 percent in three, and just 2 percent in all four. Barcodes enriched in one infection frequently appeared neutral or depleted in others, with no reproducible winners or losers, confirming that the apparent trajectories reflected the infection landscape rather than sequence-driven fitness effects.

Crucially, no host transcriptional program predicted success. Differential gene expression analysis comparing cells carrying enriched versus depleted single barcodes identified no genes with significant differential expression, and mapping barcode classifications onto the transcriptional map revealed no spatial enrichment within any cluster. What did matter was multiplicity. In all four biological replicates, a significantly larger fraction of enriched barcode detections was associated with cells harboring two or more distinct barcodes, and this overrepresentation of enriched barcodes among coinfected cells remained significant when compared with the background distribution of all detected barcode clusters, with Fisher’s exact test P values of 0.018, 0.031, 0.009, and 0.024 across the four replicates. Consistent with a complementation mechanism, cells containing multiple viral genomes produced markedly higher levels of viral transcripts than singly infected cells, with statistically significant increases between adjacent barcode-count categories up to three barcodes per cell.

The timing of this cooperative window was then measured directly. Using recombinant reporter viruses in which hemagglutinin was replaced by mCherry or eGFP, the team quantified how rapidly prior infection restricts secondary entry. When both viruses were introduced simultaneously, more than 40 percent of cells expressed both reporters, a level largely unchanged when infections were initiated two hours apart. But as the delay increased to 4, 6, 8, or 10 hours, coinfection rates fell sharply to 13, 6, 2, and 0.5 percent respectively. Viral production kinetics reinforced the picture: infectious particle output remained low across all tested multiplicities up to 6 hours post-infection, then rose sharply from 8 hours in proportion to the initial multiplicity, with the highest dose producing nearly an order of magnitude more infectious particles than the lowest at 8 hours. Normalizing these titers to the input multiplicity showed no progressive reduction in relative yield with increasing dose, meaning the amplification was not merely a dilution of per-particle productivity.

A direct competition experiment tested whether infection timing determines the opportunity for enrichment. Cells were first infected with a single, barcode-defined viral clone and then challenged with the complete barcoded library after delays of 0, 2, 4, 6, 8, 10, or 12 hours. With simultaneous inoculation, coinfection was extensive and the seeded barcode showed only modest enrichment. At a two-hour delay the seeded barcode gained further dominance while measurable diversity from the secondary inoculum persisted. By 4 to 6 hours the seeded virus accounted for nearly the entire output, and by 8 to 12 hours it dominated the recovered population with little detectable contribution from the late-arriving library. Once a single lineage established replication, even modest asynchrony restricted the contribution of later competitors, demonstrating that the window for cooperative infection is time-limited under these synchronized cell-culture conditions.

The authors emphasize the scope of their conclusions. The experiments were performed in cell culture with synchronized infections, and the team found no reproducible association between barcode enrichment and any specific host transcriptional program, while predicting, without directly testing, that interferon signaling would modify the permissive window. In vivo, cells enter exclusionary states asynchronously across anatomical sites, so cooperative coinfection may continue at different locations and times even though the opportunity for secondary entry into any individual cell remains restricted. The work is conceptually consistent with earlier in vivo barcoding studies in which transmission bottlenecks were largely independent of barcode-encoded fitness. The practical implication is that viral traits increasing burst output, or the likelihood of synchronous coinfection, such as those affecting virion aggregation, dispersal, and release, may disproportionately shape influenza population composition, independently of mutations that enhance intrinsic replication.

Subject of Research: How synchronous co-infection and infection timing determine which influenza A virus lineages gain representation in progeny populations

Article Title: Synchronous co-infection drives lineage representation in influenza A virus populations

Article References: Fajardo, A., Frere, J. J., Uhl, S., Cantergiani, R., Hsieh, L. L., Panis, M., Carrau, L., & tenOever, B. (2026). Synchronous co-infection drives lineage representation in influenza A virus populations. EMBO Reports. https://doi.org/10.1038/s44319-026-00968-1

Image Credits: AI Generated

DOI: 10.1038/s44319-026-00968-1

Keywords: influenza A virus, superinfection exclusion, co-infection, viral lineage tracking, single-cell RNA sequencing, barcoded virus library, A549 cells, viral burst output, complementation, interferon, infection timing, population composition