Influenza A virus reassortment in mammals gives rise to genetically distinct within-host subpopulations

Abstract

Influenza A virus (IAV) genetic exchange through reassortment has the potential to accelerate viral evolution and has played a critical role in the generation of multiple pandemic strains. For reassortment to occur, distinct viruses must co-infect the same cell. The spatio-temporal dynamics of viral dissemination within an infected host therefore define opportunity for reassortment. Here, we used wild type and synonymously barcoded variant viruses of a pandemic H1N1 strain to examine the within-host viral dynamics that govern reassortment in guinea pigs, ferrets and swine. The first two species are well-established models of human influenza, while swine are a natural host and a frequent conduit for cross-species transmission and reassortment. Our results show reassortment to be pervasive in all three hosts but less frequent in swine than in ferrets and guinea pigs. In ferrets, tissue-specific differences in the opportunity for reassortment are also evident, with more reassortants detected in the nasal tract than the lower respiratory tract. While temporal trends in viral diversity are limited, spatial patterns are clear, with heterogeneity in the viral genotypes detected at distinct anatomical sites revealing extensive compartmentalization of reassortment and replication. Our data indicate that the dynamics of viral replication in mammals allow diversification through reassortment but that the spatial compartmentalization of variants likely shapes their evolution and onward transmission.

Fig. 1. Viral genotypic diversity generated through reassortment in the mammalian nasal tract.

Results from guinea pigs are shown in A–D, ferrets in E–H, and swine in I–L. Stacked plots (A, E, I) show frequencies of unique genotypes detected in one representative animal over time. Blue and orange represent the WT and VAR parental genotypes, respectively. The frequency of both parental genotypes combined (B, F, J), richness (C, G, K), and Shannon–Weiner diversity (D, H, L) are each plotted as a function of time. Guinea pigs and ferrets inoculated at high dose (1 × 10⁵ ID₅₀) and low dose (1 × 10² ID₅₀) are indicated with dashed and solid lines, respectively. The distribution of parental genotype frequencies (M), richness (N), and diversity (O) across all time points in each species (guinea pigs n = 12; ferrets n = 12; swine n = 18) is shown with violin plots. Data are presented as violin plots featuring a box plot. The bounds of the box show the first and third quartiles. Whiskers indicate 1.5 times the interquartile range and contain ~99% of the data for a normal distribution. The bounds of the violin plots indicate the minima and maxima of the entire dataset. Differences in richness and diversity between swine and guinea pigs (richness p = 0.00065; diversity p = 0.00014) and between swine and ferrets (richness p = 0.0016; diversity p = 0.0028) were significant. Parental genotype frequencies differed significantly between swine and guinea pigs (p = 0.003) and were not significant between swine and ferrets (p > 0.05). All statistics were derived using one-way ANOVA. Animal silhouettes were generated using BioRender.com.

Fig. 2. Viral diversity generated through reassortment is greater in the ferret nasal turbinates than in the lungs.

Pie charts show frequencies of unique genotypes detected in the nasal turbinates (NT) and lung of ferrets inoculated with 1 × 10² FID₅₀ (A) or 1 × 10⁵ FID₅₀ (B). Ferrets shown were analyzed at the following times post-infection: Day 1 (F11, 12, 19, 20), Day 2 (F13, 14, 21, 22), Day 3 (F15, 16, 23), and Day 4 (F17, 18, 24). Blue and orange sections of the pie charts represent VAR and WT parental genotypes, respectively. Richness (C) and diversity (D) are plotted, with observed results (colored points) overlaid on the distribution of simulated data (gray violins) (n = 1000 simulations per ferret). Horizontal lines denote the 95th and 5th percentiles. The distribution of richness (E) and diversity (F) in NT and lungs of all individuals are shown with violin plots (Lung n = 14; Nasal turbinates n = 14). Data are presented as violin plots featuring a box plot. Data are presented as violin plots featuring a box plot. The bounds of the box show the interquartile range and the center of the box shows the 50th percentile. The median is shown by the white dot. Whiskers indicate 1.5 times the interquartile range and contain $~99\%$ of the data for a normal distribution. The bounds of the violin plots indicate the minima and maxima of the entire dataset. Differences between NT and lungs were significant by both measures (p < 0.01, two-sided paired t test).

Fig. 3. Reassortant viral populations in the ferret upper and lower respiratory tracts are distinct.

Heat map showing normalized beta diversity of viral populations in ferret lung and nasal turbinates (NT) (A). The inset shows a representative comparison between the tissues of ferrets F23 and F21, to indicate the position of NT and lung within the matrix. Normalized beta diversity is plotted in B with observed results (colored points) overlaid on the distribution of simulated data (gray violins) (n = 1000 simulations per ferret). Data are presented as violin plots featuring a box plot. Data are presented as violin plots featuring a box plot. The bounds of the box show the interquartile range and the center of the box shows the 50th percentile. Whiskers indicate 1.5 times the interquartile range and contain $~99\%$ of the data for a normal distribution. The bounds of the violin plots indicate the minima and maxima of the entire dataset. One star indicates that observed data is above the 95th percentile of the distribution; two stars indicates that observed data is above the 99th percentile. C Immunohistochemistry images of ferret F23 NT and lung sections stained for WT (green) and VAR (red) viruses at day 3 post inoculation. Gray staining marks epithelial cell borders. Yellow coloring in merged images indicates the presence of both WT and VAR HA antigens in the same cell. Zoomed insets of both NT and lung sections are shown with white arrows indicating co-infected cells and yellow arrows indicating singly infected cells. Scale bars are 20 µm. Four fields were analyzed for each tissue section and representative images are depicted here. D Immunohistochemistry images of ferret F24 lung sections stained for nucleoprotein (red) and counterstained with hematoxylin. Scale bars are 20 µm. Four fields were analyzed for each tissue section and representative images are depicted here.

Fig. 4. Local diversity arising through reassortment is modest within the swine respiratory tract.

Pie charts showing frequencies of unique genotypes present within each lung lobe of a given pig on day 3 (A–C) and the nasal tract and each lung lobe of a given pig on day 5 (D–F) with blue and orange sections representing WT and VAR parental genotypes, respectively. The tissue sites are abbreviated as NA-Nasal; LA-Left Apical; RA-Right Apical; LC-Left Cardiac; RC-Right Cardiac; LD-Left Diaphragmatic; RD-Right Diaphragmatic; IN-Intermediate (accessory). Simulated richness (G) and diversity (H) are plotted, with observed results (colored points) overlaid on the distribution of simulated data (gray violins) (n = 1000 simulations per pig). Data are presented as violin plots featuring a box plot. The bounds of the box show the interquartile range and the center of the box shows the 50th percentile. The median is shown by the white dot. Whiskers indicate 1.5 times the interquartile range and contain $~99\%$ of the data for a normal distribution. The bounds of the violin plots indicate the minima and maxima of the entire dataset. Horizontal lines denote the 95th and 5th percentiles.

Fig. 5. Reassortant viral populations show extensive compartmentalization within the swine respiratory tract.

Heat map showing normalized beta diversity of viral populations in swine lung and nasal tract (A). The inset shows a representative comparison between the tissues of pigs P3 and P6, to indicate the position of each tissue within the matrix. The tissue sites are abbreviated as NA-Nasal; LA- Left Apical; RA- Right Apical; LC-Left Cardiac; RC-Right Cardiac; LD-Left Diaphragmatic; RD-Right Diaphragmatic; IN-Intermediate (accessory). Normalized beta diversity between the left apical and left cardiac lobes is plotted (B) with observed results (colored points) overlaid on the distribution of simulated data (gray violins) (n = 1000 simulations per pig). Data are presented as violin plots featuring a box plot. Data are presented as violin plots featuring a box plot. The bounds of the box show the interquartile range and the center of the box shows the 50th percentile. Whiskers indicate 1.5 times the interquartile range and contain $~99\%$ of the data for a normal distribution. The bounds of the violin plots indicate the minima and maxima of the entire dataset. One star indicates that observed data is above the 95th percentile of the distribution; two stars indicates that observed data is above the 99th percentile. C Representative immunohistochemistry images of swine (P76) left apical lung section stained for WT (green) and VAR (red) viruses at day 3 post-inoculation. Gray staining marks epithelial cell borders. Yellow coloring in merged images indicates the presence of both WT and VAR HA antigens in the same cell. Zoomed insets of the lung are shown with white arrows indicating co-infected cells and yellow arrows indicating singly infected cells. Scale bars are 20 µm. Four fields were analyzed for each tissue section and representative images are depicted here. D Immunohistochemistry images of swine (P76) left apical lung sections stained for nucleoprotein (red) and counterstained with hematoxylin. Scale bars are 20 µm. Four fields were analyzed for each tissue section and representative images are depicted here.