Influenza A virus reassortment in mammals gives rise to genetically distinct within-host subpopulations – Nature.com

Ethical considerations

All the animal experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The studies were conducted under animal biosafety level 2 containment and approved by the IACUC of Emory University (DAR-2002738-ELMNTS-A) for guinea pig (Cavia porcellus), the IACUC of the University of Georgia (AUP A2015 06-026-Y3-A5) for ferret (Mustela putorius furo) and the IACUC of Kansas State University (protocol #4120) for swine (Sus scrofa). The animals were humanely euthanized following guidelines approved by the American Veterinary Medical Association.

MadinDarby canine kidney (MDCK) cells, a gift from Dr. Robert Webster, St Jude Childrens Research Hospital, Memphis, TN to D.R.P were used for all experiments. A seed stock of MDCK cells at passage 23 was subsequently amplified and maintained in Minimal Essential Medium (Gibco) supplemented with 10% fetal bovine serum (FBS; Atlanta Biologicals) and Normocin (Invivogen). 293T cells (ATCC, CRL-3216) were maintained in Dulbeccos Minimal Essential Medium (Gibco) supplemented with 10% FBS and PS. All cells were cultured at 37C and 5% CO2 in a humidified incubator. The cell lines were not authenticated. All cell lines were tested monthly for mycoplasma contamination while in use. The medium for the culture of IAV in MDCK cells (virus medium) was prepared by supplementing the basal medium for the relevant cell type with 4.3% BSA and Normocin.

Viruses used in this study were derived from influenza A/Netherlands/602/2009 (H1N1) virus (NL09) and were generated by reverse genetics51,52,53. In brief, 293T cells transfected with reverse genetics plasmids 1624h previously were co-cultured with MDCK cells at 37C for 4048h. Recovered virus was propagated in MDCK cells at a low multiplicity of infection to generate working stocks. Titration of stocks and experimental samples was carried out by plaque assay in MDCK cells. Silent mutations were introduced into each segment of the VAR virus by site-directed mutagenesis of reverse genetics plasmids. The specific changes introduced into the VAR virus were reported previously24,31. NL09 VAR virus was engineered to contain a 6XHIS epitope tag plus a GGGS linker at the amino (N) terminus of the HA protein following the signal peptide. NL09 WT virus carries an HA epitope tag plus a GGGS linker inserted at the N terminus of the HA protein31. For animal challenges, 1:1 mixture of NL09 WT and VAR viruses was prepared using methods described previously24. This mixture was validated in cell culture by quantifying cells positive for HIS and HA tags following infection of MDCK cells, revealing an empirically determined ratio of 0.95:1 (WT:VAR). The same mixture was used for all experiments reported herein.

Replication of NL09, NL09 WT, and NL09 VAR viruses was determined in triplicate culture wells. MDCK cells in 6 well dishes were inoculated at an MOI of 0.05 PFU/cell in PBS. After 1h incubation at 37C, inoculum was removed, cells were washed 3x with PBS, 2mL virus medium was added to cells, and dishes were returned to 37C. A 120 ul volume of culture medium was sampled at the indicated times points and stored at 80C. Viral titers were determined by plaque assay on MDCK cells.

Female, Hartley strain guinea pigs weighing 250350g were obtained from Charles River Laboratories and housed by Emory University Department of Animal Resources. Before intranasal inoculation and nasal washing, the guinea pigs were anaesthetized with 30mgkg1 ketamine and 4mgkg1 xylazine by intramuscular injection. The GPID50 of the NL09 virus was previously determined to be 1101 PFU32. To evaluate reassortment kinetics in guinea pigs, groups of six animals were infected with 1103 PFU (1102 ID50) or 1106 PFU (1105 ID50) of the NL09 WT/VAR virus mixtures. Virus inoculum was given intranasally in a 300l volume of PBS. Nasal washes were performed on days 16 post-inoculation and titered by plaque assay. Viral genotyping was performed on samples collected on days 1, 2, and 3 or 4 for each guinea pig. Day 3 was used for animals receiving the higher dose since the virus is cleared rapidly in this system and shedding has ceased by day 4.

Female ferrets, 20-weeks-old, from Triple F Farms (Gillett, PA) were used. All ferrets were seronegative by anti-nucleoprotein (anti-NP) influenza virus enzyme-linked immunosorbent assay, Swine Influenza Virus Ab Test, (IDEXX, Westbrook, ME) prior to infection. Five days prior to experimentation, ferrets were sedated, and a subcutaneous transponder (Bio Medic Data Systems, Seaford, Delaware) was implanted to identify each animal and provide temperature readings. Anesthetics were applied via intramuscular injection with ketamine (20mgkg1) and xylazine (1mgkg1). Infections were performed via intranasal inoculation of 1mL of virus diluted in PBS. Ferret nasal washes were carried out as follows. Ferrets were anesthetized and 1ml of PBS administered to the nose was used to induce sneezing. Expelled fluid was collected into Petri dishes and samples were collected in an additional volume of 1mL PBS. Infected ferrets were monitored daily for clinical signs, temperature, and weight loss. Ferrets were euthanized by intravenous injection of 1ml of Beuthanasia-D diluted 1:1 with DI water (Merck, Madison, NJ).

For determination of ferret ID50, six groups of four ferrets each were inoculated with increasing doses of the NL09 WT/VAR virus mixture (1100.1 PFU, 1100 PFU, 1101 PFU, 1102 PFU, 1103 PFU, and 1104 PFU). Nasal washes were collected daily for up to 6 days and titrated for viral shedding by plaque assay. The ferret ID50 was determined based on results obtained on day 2 and found to be equivalent to 3.2102 PFU.

For analysis of reassortment frequency and detection of viral antigen in tissues, ferrets were inoculated with 3.2104 PFU (1102 ID50) or 3.2107 PFU (1105 ID50). After infections, nasal washes were collected daily for up to 6 days and titrated by plaque assay. Viral genotyping was performed on samples collected on days 1, 3, and 5 for each ferret. Necropsies were performed on days 14 for the collection of nasal turbinate and lung tissues. A single lung lobe (the left caudal lobe) was sampled from each ferret. Tissue sections collected for virology were disrupted in 1mL of sterile PBS using the TissueLyser LT (Qiagen, Germantown, MD) at 30Hz for 5min twice, in microcentrifuge tubes with 3mm Tungsten Carbide Beads (Qiagen, St. Louis, MO). Supernatants were clarified by centrifugation and frozen at 80C until viral titration. For histology, tissues were submerged in 10% buffered formalin (Sigma Aldrich, St. Louis, MO) and stored at room temperature until evaluation.

The pig study was conducted at the Large Animal Research Center (a biosafety level 2+ facility) at Kansas State University in accordance with the Guide for the Care and Use of Agricultural Animals in Research and Teaching of the U.S. Department of Agriculture. To determine virus reassortment and viral antigen in tissues, 18 4-week-old influenza H1 and H3 subtype virus- and porcine reproductive and respiratory syndrome virus-seronegative gender-mixed crossbred pigs were randomly allocated into groups. Each pig was inoculated with 2106 PFU of NL09 WT/VAR mixture through both intranasal and intratracheal routes (106 PFU was administered in a 1ml volume by each of these two routes) under anesthesia as described previously54. Clinical signs for all experimental pigs were monitored daily throughout the experiment. Nasal swabs were collected at 1-, 3-, 5-, and 7-days post infection from each pig. Three infected pigs were euthanized at 3-, 5-, and 7-days post infection. During necropsy, nasal turbinate, trachea, and lung tissues from seven lobes collected from each pig were frozen at 80C for virus isolation and fixed in 10% buffered formalin for IHC examination.

Reassortment frequencies were evaluated by genotyping 21 clonal viral isolates per sample as described previously50. This analysis was applied to guinea pig nasal washes, ferret nasal washes, swine nasal swabs, ferret tissue homogenates, and swine tissue homogenates. Time points to be examined were chosen based on positivity in all animals in a treatment group. Thus, nasal wash samples from days 1, 2, 3, or 4 were evaluated from guinea pigs while samples from days 1, 3, and 5 were evaluated for swine and ferrets. Ferret tissues collected on days 1, 2, 3, and 4 and swine tissues collected on days 3 and 5 were analyzed.

Briefly, plaque assays were performed on MDCK cells in 10cm dishes to isolate virus clones. Serological pipettes (1ml) were used to collect agar plugs into 160l PBS. Using a ZR-96 viral RNA kit (Zymo), RNA was extracted from the agar plugs and eluted in 40l nuclease-free water (Invitrogen). Reverse transcription was performed using Maxima reverse transcriptase (RT; ThermoFisher) according to the manufacturers protocol. The resulting cDNA was diluted 1:4 in nuclease-free water and each cDNA was combined with segment-specific primers (Supplementary Data3)24,31 designed to amplify a region of approximately 100 base pairs. The amplicon for each segment contains the site of the single nucleotide change in the VAR virus. Quantitative PCR was performed with Precision Melt Supermix (Bio-Rad) using a CFX384 Touch Real-Time PCR Detection System (Bio-Rad). Quantitative PCR data was collected using CFX Manager Software v2.1 (Bio-Rad). Template amplification was followed by high-resolution melt analysis to differentiate the WT and VAR amplicons55. Precision Melt Analysis software v1.2 (Bio-Rad) was used to determine the parental origin of each gene segment based on the melting properties of the cDNA amplicons relative to WT and VAR controls. Each plaque was assigned a genotype based on the combination of WT and VAR genome segments, with two variants on each of eight segments allowing for 256 potential genotypes.

Tissue samples from nasal turbinates of ferrets, the right caudal lung lobe of ferrets, and all seven lung lobes of swine were fixed in 10% neutral buffered formalin for at least 24h before being embedded in paraffin. Nasal turbinates were decalcified prior to being embedded in paraffin. Sections from all the tissues were cut and slides were prepared. The tissues were deparaffinized by warming the slides at 60C on a slide warmer for 45min followed by immersion in xylenes (Sigma) for 25min. The slides were then immersed in 100% ethanol for 10min, 95% ethanol for 10min, and 70% ethanol for 5min. The slides were then washed by placing them in deionized water for 1h. Antigen retrieval was performed by steaming the slides in 10mM citric acid, pH 6.0 for 45min, followed by washing in tap water and 1 PBS (Corning) for 5min. The WT and VAR viruses were detected in the tissues using a mouse anti-HA Alexa Fluor 488 (Invitrogen catalog number A-21287; clone 16B12; 1:50 dilution) and mouse anti His Alexa Fluor 555 (Invitrogen catalog number MA1-135-A555; clone 4E3D10H2/E3; 1:50 dilution) while epithelial cell borders were stained using rabbit anti-Na+K+ ATPase Alexa Fluor 647 (Abcam catalog number 198367; clone EP1845Y; 1:100 dilution) at 4C overnight. Slides were washed three times in 1 PBS (Corning) and once in deionized water to remove excess antibody. The slides were mounted onto glass coverslips using ProLong Diamond Anti Fade mounting media (ThermoFisher). The images were acquired using an Olympus FV1000 Confocal Microscope at 60 magnification under an oil immersion objective. The specificity of the antibodies was confirmed by infecting MDCK cells with either the NL09 WT, NL09 VAR, or both viruses for 24h. The cells were fixed using 4% paraformaldehyde (Alfa Aesar) and stained for HA and His tags using the antibodies as described above (Supplementary Fig.9).

For morphological analysis via IHC, the slides were pre-treated in pH 9.0 buffer at 110C for 15min. Blocking was performed using hydrogen peroxide for 20min followed by PowerBlock (BioGenex) for 5min. Slides were washed with PBS thrice and NP antigen was detected using a goat anti-influenza NP polyclonal antibody (abcam catalog number ab155877; 1:1000 dilution) for 1h. Slides were washed thrice with PBS to remove excess antibody and incubated with a rabbit anti-goat biotinylated IgG (Vector laboratories catalog number BA-5000; 1:5000 dilution) for 10min. After washing, 4Plus Alkaline Phosphatase Label (BioCare Medical) was added for 10min. The antigen signal was detected by incubating the slides in Chromogen IP Warp Red stain (BioCare Medical) for 10min. Haematoxylin counterstaining was performed post-antigen staining.

Figures were generated using Python 3 v3.1056 and the packages matplotlib v3.6.057, NumPy v1.23.358, pandas v1.5.059, and seaborn v0.12.060. Simulations were conducted in Python 3 v3.10.

Here a viral genotype is defined as a unique combination of the eight IAV segments, where each segment is derived from either the variant or wild-type parental virus; therefore, there are 28 possible unique genotypes, with two parental genotypes and 254 reassortant genotypes. For any given sample, the frequency of each unique genotype can be calculated by dividing the number of appearances each unique genotype has in the sample by the total number of clonal isolates obtained for that sample.

Understanding the distribution of unique genotypes involves using both unweighted and weighted genotype frequency statistics. Genotype richness ((S)) does not incorporate genotype frequency and is given by the number of unique genotypes in a sample. Given our sample size of 21 plaque isolates, genotype richness, or the number of distinct genotypes detected in a sample, can range from a minimum of 1 (a single genotype is detected 21 times) to a maximum of 21 (21 unique genotypes detected).

Diversity was measured using the ShannonWeiner index (H), which considers both richness and evenness in the frequency with which genotypes are detected. In our dataset, diversity can range from 0 to 3.04. ShannonWiener diversity was calculated as:

$$H=-mathop{sum }limits_{i=1}^{S}({p}_{i} * {ln}{p}_{i})$$

(1)

where (S) is genotype richness and ({p}_{i}) is the frequency of unique genotype (i) in the sample (6).

To address whether evaluating 21 plaques per sample for this analysis was sufficient to yield robust results on genotype diversity, we used a computational simulation to test the sensitivity of the measured diversity values to the number of plaques sampled. In these simulations, we calculated the diversity present in samples generated by randomly picking n (out of the possible 21) plaques without replacement. At each sampling effort n, we simulated 1000 samples, with plaque replacement between samples. The results typically show that diversity values increase as n increases, with values asymptoting as n approaches 21, suggesting that further increases in n would not greatly change results and validating the use of 21 plaques (Supplementary Fig.10).

To evaluate the extent to which the spatial dynamics of viral reassortment and propagation shape the overall richness and diversity in a host, we sought to compare the observed richness and diversity at each anatomical site to that which would be expected if virus moves freely among anatomical locations. Thus, to simulate free mixing within the host, we randomly shuffled observed viral genotypes among all sites in a given animal. The average richness and ShannonWiener index of the simulated viral populations at each site were then calculated. The 5th and 95th percentiles for the simulated distribution of each animal were calculated and compared to the observed richness and diversity for each of the anatomical sites. If a sites observed richness and diversity fell below the 5th percentile or above the 95th percentile, then a barrier to the influx or efflux of reassortant genotypes from or to the other sites is suggested.

The dissimilarity between populations can be measured by beta diversity. For this study, we evaluated beta diversity from a richness perspective, focusing on dissimilarity in the unique genotypes detected and excluding consideration of their frequency. This approach was used to de-emphasize the effects of WT and VAR parental genotypes, which were likely seeded into all anatomical locations at the time of inoculation. We calculate the beta diversity by treating the viral genotypes in two lobes as two distinct populations:

$$beta=frac{{S}_{1+2}}{frac{1}{2}({S}_{1}+{S}_{2})}$$

(2)

where ({S}_{1+2}) is the richness of a hypothetical population composed of pooling the viral genotypes of the two lobes while (frac{1}{2}({S}_{1}+{S}_{2})) represents the mean richness of the lobes (7). The beta diversity of a single comparison can be normalized so that it ranges from zero to one:

$${{BD}}^{{prime} }=frac{{BD}-1}{{{BD}}_{{max }}-1}{beta }_{n}$$

(3)

where ({{BD}}_{{max }}) is the beta diversity calculated by assuming that there are no viral genotypes shared by both lobes (7). A ({{BD}}^{{prime} }{beta }_{n}) closer to one indicates that the lobes viral populations are more dissimilar while a ({{BD}}^{{prime} }{beta }_{n}) closer to zero suggests that the lobes have similar unique viral genotypes and overall viral richness. A ({{BD}}^{{prime} }) ({beta }_{n})of zero occurs when all unique genotypes present in one lobe are also present in the other.

To address whether evaluating 21 plaques per sample for this analysis was sufficient to yield robust results on beta diversity, we again used computational simulations. These simulations were designed to test the sensitivity of beta diversity values to the number of plaques sampled. In these simulations, we again generated plaque data subsets by randomly picking n (out of the possible 21) plaques without replacement. At each sampling effort n, we simulated 1000 samples, with plaque replacement between samples. Beta diversity values were then calculated based on these data subsets, at a given n. The results typically show that ({beta }_{n})values tend to stabilize as n approaches 21, suggesting that further increases in n would not greatly change results and validating the use of 21 plaques (Supplementary Fig.10). In a subset of cases that involve the nasal sample of Pig 5, however, the relationship between ({beta }_{n}) and n is less stable. In sharp contrast to most other samples from Pig 5, the nasal site showed 20WT parental isolates and one VAR parental isolate. The lung tissues had no WT parental genotypes detected. As a result, each successive plaque draw from the nasal sample increases the probability of detecting the VAR virus and therefore detecting a commonality between the nasal tract and any of the lung lobes. Thus, in situations where two tissue sites have a single, relatively rare genotype in common, the number of plaques sampled has a strong impact on ({beta }_{n}) outcomes.

To simulate free mixing between two lobes, we randomly shuffled the genotypes between each of the 28 pairwise combinations among pig tissues and the single ferret lung-NT combination and computed the ({{BD}}^{{prime} }{beta }_{n}) for each comparison. Free mixing for all combinations was simulated 1000 times. We reasoned that if compartmentalization was present in the observed dataset, then the dissimilarity values would fall at the high end of the simulated distribution (>95th percentile).

Percentiles were calculated using the percentileofscore method from the SciPy package v1.9.159. Paired t tests and ANOVA tests were performed using the ttest_rel method and the f_oneway method respectively from the SciPy Package v1.9.159.

Further information on research design is available in theNature Portfolio Reporting Summary linked to this article.