DNA mismatch repair controls the host innate response and cell fate after influenza virus infection

Differential cellular fate after influenza A virus infection can be modeled in vitro.

To observe cellular fate after IAV infection in vitro, we previously utilized LoxP-Stop-LoxP-reporter-containing NCI-H441 cells which activate a fluorescent reporter after infection with a Cre-expressing influenza virus (IAV-Cre)¹⁵. For this study, we generated an analogous reporter in an alveolar epithelial cell line (A549) to compare rates of survival after infection (Fig. 1a). Upon IAV-Cre infection, we found that while both cell types initially activate the ZsGreen reporter, the infected A549 cells ultimately succumbed while the H441-CR cells survived (Fig. 1b). A549-CR cells divide much more rapidly than H441-CR cells (Supplementary Fig. 1), discounting the possibility that higher numbers of survivors at later time points is simply due to rapid growth of surviving H441-CR cells.

To better understand the nature of infection in both A549 and H441 cells, we monitored infection and hemagglutinin (HA) surface trafficking via microscopy as well as the activity of the viral RNA-dependent RNA polymerase (RdRP); however, no significant differences were observed (Fig. 1c–d). We also conducted a multicycle growth curve and found that infectious progeny viruses are released from H441 cells, albeit at lower levels than A549 cells (Fig. 1e). These data suggested that H441 cells have the ability to clear actively replicating virus rather than being incompletely permissive. To monitor the kinetics of viral infection and clearance in H441 cells, we conducted an infection time course and visualized viral protein production (red) and induction of the ZsGreen reporter (green) over time on a per cell basis. Despite early viral protein expression, we observed rapid clearance of replicating virus from infected cells, which subsequently survive (Fig. 1f–g). H441 cells however, were killed after infection with an alphavirus (Sindbis virus), indicating that these cells are not just generally resistant to virus-dependent cytotoxicity (Fig 1h).

siRNA screening identifies a DNA mismatch repair protein as required for H441 cell survival after IAV infection.

In order to identify the genes that allow H441 cells to clear and survive IAV infection, we conducted an RNAi screen using a library targeting the druggable genome (23,349 siRNAs targeting 7,783 genes). We transfected H441 cells containing our ZsGreen reporter with siRNAs arrayed in 384-well plates, infected with IAV-Cre virus for 120 hours, and used automated imaging and analysis to determine genes that, when targeted by siRNA, altered H441 survival (Fig. 2a). We treated cells with type I interferon (IFN) as a negative control (which blocks infection and prevents ZsGreen signal) and transfected cells with an siRNA targeting the viral nucleoprotein (NP) as a positive control (which increases cell survival by suppressing viral replication). We performed the screen in duplicate, with two unique siRNAs per gene per well in two wells (for a total of four unique siRNAs per gene). A Z-score was calculated based on the gene’s ability to either significantly increase or decrease H441 cell survival (% ZsGreen⁺ cells) when knocked down, and a “hit” was defined as having a log₁₀ Z-score of greater than 1.0 or less than −1.0 in both replicates of the screen (Fig. 2b–c, Supplementary Table 1).

Using these criteria, we identified 87 candidate genes. In addition to those 87 genes, we also selected 9 additional genes that had large phenotypes (log₁₀ Z-score of greater than 1.3) in only one of the two replicate screens, for a total of 96 candidate genes for further testing. We obtained two additional siRNAs per gene and performed an RNAi validation screen (Fig. 2d–e, Supplementary Table 2). Because genes that directly affect viral infection could also indirectly impact cellular survival, we infected siRNA-transfected H441 cells with a luciferase-expressing influenza virus (IAV-Luc) and eliminated genes that modulated luciferase levels at 16 hpi (Fig. 2f–g, Supplementary Fig. 2). Finally, because some siRNAs can induce cellular toxicity irrespective of viral treatment, we focused on genes that did not decrease total cell numbers by more than 50% relative to control. Together, these analyses revealed 15 genes that significantly affected H441 survival in both the primary and validation screens, with only minor effects on virus replication levels and cell viability (Table 1).

Many of the validated genes play roles in cellular stress response pathways. Examples include genes involved in protein degradation (PSMA2, PSMA5), host transcription (POLR2C, POLR2E, CDK12), and protein folding (PPIAL4A, TCP1). One of the more unexpected hits was MSH6, a core member of the DNA mismatch repair (MMR) pathway. This pathway is responsible for excising and repairing mismatched nucleotides that arise during DNA replication or some types of DNA damage¹⁶. As an RNA virus, IAV infection is not generally thought to affect host DNA metabolic processes, and thus it was unclear why this gene would be required for cellular survival from IAV infection; we therefore selected this gene for further study.

DNA MMR is differentially regulated in A549 and H441 cells during IAV infection.

First, we determined whether MSH6 was required for H441 cell survival in the context of DNA MMR, or whether it had an alternative “moonlighting” activity. Knockdown of other key proteins involved in the canonical DNA MMR pathway (MSH2, MLH1, PMS2, EXO1) revealed that each member of the pathway was required for H441 cell survival (Fig. 3a–c, Supplementary Fig. 3). We next explored whether differences in the regulation of the DNA MMR pathway could explain the differential survival phenotypes of H441 and A549 cells. We therefore monitored expression of DNA MMR genes in cells mock-infected or infected with wild-type A/Puerto Rico/8/1934 (WT PR8), a laboratory-adapted H1N1 strain of IAV. At 9 hpi, PR8-infected A549 cells display a significant reduction in the gene expression of all core DNA MMR genes compared to uninfected cells, whereas PR8-infected H441 cells maintain relatively high levels of expression of each DNA MMR gene comparable to uninfected cells (Fig. 3d, Supplementary Table 3). Western blot analysis of two main DNA MMR proteins (MSH2 and MSH6) also displayed a reduction at 24 hpi in A549 cells, but not in H441 cells (Fig. 3e).

To determine whether differences in MMR protein levels impacted DNA MMR activity, we adapted a previously described¹⁷ DNA MMR assay by generating a NanoLuc plasmid with a single nucleotide mismatch (G-G) in the start codon (Fig. 3f). We compared NanoLuc expression from the ‘mismatch plasmid’ to the ‘WT plasmid’ to measure the level of DNA MMR activity. Cells transfected with non-targeting control siRNAs repaired and expressed the ‘mismatch plasmid’ in an MSH2/6 dependent manner (Fig. 3g). After infection of A549 and H441 cells, we observed that A549 cells lost >25% of their MMR activity while H441 cells maintained levels equal to mock-infected cells (Fig. 3h).

DNA MMR is required to repair ROS-induced DNA damage during IAV infection.

Since we identified a DNA repair pathway as critical for surviving viral infection, we next verified that DNA damage was induced during IAV infection. We infected H441 cells with WT PR8 and monitored phospho-H2AX levels, a histone modification associated with DNA repair (Fig. 4a–b). We observed strong induction of phospho-H2AX 8–12 hours after PR8 infection, on par with treatment with the control DNA damaging agents Etoposide and hydrogen peroxide (H₂O₂).

Apart from its major role in repairing spontaneous nucleotide mismatches following DNA replication, the DNA MMR pathway also plays a role in repairing reactive oxygen species (ROS)-induced DNA damage¹⁸. IAV infection is known to alter the redox state of infected cells^19,20 and increase intracellular ROS levels and oxidative damage in a variety of cell lines including A549 and MDCK cells^21–26. These data suggested a possible link between IAV infection and the DNA MMR pathway. Although total ROS levels have been reported to increase during IAV infection, it was unknown whether this leads to ROS-induced DNA damage. We therefore employed an assay to measure 8-hydroxy-2’-deoxyguanosine (8-OHdG), a common biomarker of ROS-induced DNA damage²⁷. We found increased levels of 8-OHdG upon PR8 infection, suggesting that IAV infection increases ROS-induced DNA damage (Fig. 4c–d).

To determine whether DNA MMR is required to repair the oxidative DNA damage induced by IAV infection, we employed a modified comet assay, wherein cells are treated with Fpg, a restriction enzyme that cleaves DNA only at sites of oxidative damage^28,29. H441 cells were transfected with non-targeting control siRNAs or depleted of DNA MMR genes, and subsequently infected with WT PR8. Comet tail lengths increased only 2.8-fold following PR8 infection in cells competent for DNA MMR, but tail lengths increased 9.4-fold in infected cells lacking MSH2+6 (Fig. 4e–f). Our data suggested that maintaining DNA MMR during infection controls ROS-induced DNA damage. Therefore, artificially decreasing ROS levels could potentially increase the rate of cellular survival. When we infected A549-CR cells (which normally succumb to IAV infection) with IAV-Cre in the presence of the ROS scavenger Trolox³⁰, cell survival roughly doubled (Fig. 4g). However, the same treatment of H441-CR cells had no effect on the percentage of survivors (Fig. 4h), likely due to their maintenance of DNA MMR activity.

DNA MMR impacts expression of innate antiviral genes.

We hypothesized that the avoidance of lethal oxidative genomic damage was part of a larger program required for epithelial cells to survive viral infection. ROS-induced DNA damage increases stalling of RNA polymerase II during transcription^31,32 and DNA MMR is associated with transcription-coupled repair^33–35. Although DNA MMR and antiviral gene expression have not been previously linked, we tested whether the reduction in DNA MMR activity and the subsequent accumulation of DNA damage in IAV-infected cells also attenuated transcription of host antiviral genes.

We first tested whether ROS-induced DNA damage affected cellular transcription in club cells. H₂O₂ treatment reduced expression of a cellular NanoLuc reporter by ~80% compared to PBS-treated control cells (Fig. 5a). The decrease in NanoLuc expression, however, was not due to cell death, as cell viability in both samples was comparable (Fig. 5b). We also monitored the effects of oxidative damage on expression of endogenous antiviral genes. Expression of the interferon-stimulated gene (ISG) Mx1 was significantly reduced upon treatment with H₂O₂ at both the RNA (Fig. 5c) and the protein level (Fig. 5d).

Previous reports have shown that DNA repair pathways in general are frequently downregulated during IAV infection^36–38. To our knowledge, the contribution of virally-induced oxidative DNA damage to host shutoff, however, is unexplored. Infection of our NanoLuc reporter club cells led to the expected reduction in luciferase levels (Fig. 5e–f). In order to test if the maintenance of DNA MMR activity was helping to limit host shutoff, we repeated this experiment after knockdown of DNA MMR genes and observed a further reduction of NanoLuc expression (Fig. 5e–f). To determine whether the DNA MMR mediated effects on host translation could extend to antiviral gene expression, we made use of an interferon reporter construct (ISRE-GFP). Reporter expression was significantly reduced upon knockdown of DNA MMR genes (Fig. 5g). Together, these data are consistent with a model in which DNA MMR activity repairs oxidative DNA damage to allow for transcriptional induction of antiviral genes (Fig. 5h).

In order to more broadly investigate the impact of DNA MMR on antiviral gene expression, we performed RNAseq on H441 cells transfected with control or DNA MMR siRNA and then infected with WT PR8 for 24 h (Fig. 5i, Supplementary Table 4, NCBI GEO Series GSE130189). We confirmed depletion of the MMR genes, as well as viral infection (Supplementary Fig. 4a–c). Out of the 282 genes induced >5-fold by PR8 infection, the expression of 115 genes was reduced (defined as a log₂(fold change)<−0.2) in cells transfected with MSH2+6 siRNA (Fig. 5i–j, Supplementary Table 5). More specifically, out of the 51 ISG and stress response genes induced >5-fold, the expression of 28 of those genes was reduced >10% by MSH2+6 depletion (Fig. 5k, Supplementary Table 5). We confirmed that induction of two of the ISGs identified in the RNA sequencing (IFI44L and IFIT1) was significantly attenuated in cells depleted of DNA MMR genes (Fig. 5l–o). We also verified that the reduction of IFIT1 mRNA led to reduced IFIT1 protein in infected H441 cells depleted of DNA MMR genes (Fig. 5p).

Enhanced viral suppression of the DNA MMR pathway decreases cellular survival and increases the severity of viral disease.

We wanted to generate a recombinant IAV strain that would suppress the DNA MMR pathway in all infected cells. In order to accomplish this goal, we inserted artificial microRNAs (amiRNAs) into an extension of the 3’ UTR of segment 6 (Fig. 6a). Processing of amiRNAs leads to the generation of small RNAs which are functionally equivalent to siRNAs³⁹. We generated two separate viruses: one with an amiRNA designed to target GFP as a control (GFP-amiRNA virus), and a second targeting mouse MSH6 in order to suppress DNA MMR activity (MSH6-amiRNA virus). We also inserted Cre recombinase into segment 8 of both viruses (Fig. 6a) to facilitate measurement of cellular survival. We verified functional Cre recombinase expression and monitored processing of the cleaved, mature amiRNAs via northern blot (Fig. 6b–c). We also confirmed that the mature amiRNAs reduced MSH6 RNA levels in mouse epithelial cells (Fig. 6d). In chicken cells where the amiRNA against mouse MSH6 is non-functional (Fig. 6e), we performed growth curves and observed that the growth of the GFP-amiRNA and MSH6-amiRNA viruses was identical (Fig. 6f).

In order to evaluate the effect of MSH6 targeting in primary cells, we generated air-liquid interface (ALI) epithelial cultures (Fig. 6g) from transgenic mice harboring a LoxP-Stop-LoxP-tdTomato cassette and infected them with our amiRNA viruses. An equal number of cells were initially infected with the GFP-amiRNA and MSH6-amiRNA viruses, but at later timepoints, we observed a significant reduction in cellar survival when MSH6 was targeted by the virally-encoded amiRNA (Fig. 6h–i). Next, we set out to determine the roles of DNA MMR and club cell survival in viral pathogenesis. We infected 8-week-old female C57BL/6J mice intranasally with a range of doses of the parental WT PR8, GFP-amiRNA, or MSH6-amiRNA viruses. Mice that were infected with the MSH6-amiRNA virus experienced increased morbidity and mortality relative to GFP-amiRNA-infected mice (Fig. 6j–l), although both modified viruses were attenuated relative to parental PR8. The MSH6-amiRNA virus also replicated to higher titers in the mouse lungs at 4 dpi (Supplementary Fig. 5), which is consistent with dampened antiviral gene expression.

Cells.

All cells were obtained from ATCC and grown at 37°C in 5% CO₂. H441 cells were grown in Roswell Park Memorial Institute (RPMI) medium supplemented with 10% fetal bovine serum, HEPES, glucose, sodium pyruvate, and penicillin-streptomycin. 293T and A549 cells (ATCC) were grown in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 5% fetal bovine serum, GlutaMAX, and penicillin-streptomycin. Madin-Darby canine kidney (MDCK) cells were grown in minimal essential medium (MEM) supplemented with 5% fetal bovine serum, HEPES, NaHCO₃, GlutaMAX, and penicillin-streptomycin. MLE15 cells were grown in DMEM supplemented with 10% fetal bovine serum, GlutaMAX, and penicillin-streptomycin. The Cre-reporter cassette was constructed and transduced into H441 and A549 cells as previously described¹⁵. The growth curve comparing division rates of A549-CR and H441-CR was conducted in 6-wells. Cells were seeded at 150,000 cells per well and were collected and counted at 24, 48, 72, and 96 h.

Viruses.

Recombinant viruses were generated using reverse genetics methods as previously described^62,63, using the bicistronic pDZ rescue plasmid system. All viruses were based off of the A/Puerto Rico/8/1934 (PR8) H1N1 influenza virus background. The WT PR8 virus contains all WT PR8 segments. The NEBuilder HiFi DNA assembly kit (NEB) was used to assemble DNA fragments to create the desired recombinant DNA gene segments. The IAV-Cre recombinant virus was engineered to express Cre recombinase within the PR8 PB2 segment (segment 1), as previously described⁷. The time-course measuring ZsGreen expression during IAV-Cre infection in A549-CR and H441-CR cells was conducted by infecting cells at MOI=1. Every 24 h, cells were fixed with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) and nuclei were stained with Hoescht dye before pictures were taken with the ZOE fluorescent cell imager (Bio-Rad). The IAV-Luc recombinant virus was also engineered to express luciferase in the PR8 PB2 segment (segment 1), as previously described⁶⁴. For the PR8-Cre+amiRNA viruses, the Cre recombinase was inserted into the PR8 NS segment (segment 8) following separation of the NS1 and NEP coding regions, and the amiRNA expression sequence was inserted into the PR8 NA segment (segment 6). After assembly and transformation of the constructs into high-efficiency competent cells (NEB), the plasmids were collected and used in separate rescue transfections in 293T/MDCK co-cultures. Rescued virus was expanded in day 10 fertilized chicken eggs and then plaque-purified on MDCK cells. Individual plaques were picked and expanded in day 10 fertilized chicken eggs before being validated via RT-PCR and sequencing of the recombinant segments. For the viruses containing Cre recombinase, Cre activity was confirmed by infecting H441-CR at MOI=2 and observing the expression of the Cre-controlled ZsGreen reporter in cell monolayers at 24–48 hpi.

siRNA screen.

Single siRNAs (2 sites/target, 25 nM final concentration, Qiagen) were spotted in collagen-coated 384-well plates complexed with HiPerFect (0.6 μL/well in 9.4 μL OptiMEM). Following complexing, H441-CR cells (1,625 cells/well) were added to each well and infected 48 h later with IAV-Cre at an MOI=5. After 120 h, cells were fixed with 4% PFA, washed, and stained with Hoechst stain. All solutions were added using automated liquid handling (Well-Mate, Thermo Fisher) to limit well-to-well variability. Images were captured (3 sites/well) at 10X using an ImageXpressMicro Microscope (Molecular Devices). Automated image analysis (MetaXpress) was used to calculate the total number of cells (Hoechst⁺) and the number of survivor cells (ZsGreen⁺). These values were used to calculate the median and interquartile range (on log transformed data), which were then used to calculate robust Z-scores. The median of each plate in the screen was used to calculate baseline. Hits were identified as those wells that exhibited a change in survivor cell percentage of total cells by the indicated standard deviations from calculated Z-scores within each plate.

siRNA survivor cell assays.

H441-CR cells were transfected with 0.2 μM siRNA using HiPerFect reagent (Qiagen) in OptiMEM. After 48 h, cells were infected with IAV-Cre at MOI=1 for 1 h before virus was removed and replaced with complete media. Cells were incubated at 37°C for 5 days before being fixed with 4% PFA and stained with Hoescht dye for nucleus labeling. Images were then collected using the ZOE fluorescent cell imager (Bio-Rad) to identify the level of ZsGreen-positive cells indicating survival of direct infection. Image analysis was done using ImageJ (NIH).

Viral replication counter-screen.

H441-CR cells were transfected with the siRNA mini-library and then infected after 48 h with IAV-Luc at MOI=0.5. At 16 hpi, cells were lysed and luciferase activity was measured using a luciferase assay system (Promega) as a readout of viral replication. All siRNAs were compared relative to the non-targeting control siRNA and those which altered viral replication greater than two standard deviations were eliminated from the final list of screen hits.

Viral growth comparisons.

The RdRP replication assay was conducted as previously described⁶⁴. For the multi-cycle growth assay comparing A549 and H441 cells, WT PR8 was added to cells at MOI=1 for 1 h before being maintained in OptiMEM with 0.3 μg/mL TPCK-treated trypsin protease at 37°C. For the experiment comparing IAV and Sindbis survival, H441-CR cells were infected with MOI=1 of WT PR8 influenza virus or Sindbis virus (gift from Benjamin tenOever). After 48 h, the total number of remaining cells was quantified compared to mock-infected controls using flow cytometry and counting beads. For the amiRNA virus growth curves, eggs were injected with 1000 PFU virus and incubated at 37°C for 24–72 h, or MLE15 cells were infected at MOI=0.1 for one hour before being maintained in OptiMEM with 0.4 μg/mL TPCK-treated trypsin protease for 24–72 h at 37°C. All samples were then titered using plaque assays.

Plaque assays.

Viral titers were determined using standard plaque assay procedures on MDCK cells. Serial virus dilutions were incubated on cells for 1 h before removing the virus and adding the agar overlay. Cells were then incubated at 37°C for 48 h before being fixed in 4% PFA for at least 12 h. The 4% PFA was then aspirated, and the agar layer was removed before washing cells with PBS. Serum from WT PR8-infected mice was diluted 1:2,000 in antibody dilution buffer (5% (w/v) nonfat dried milk and 0.05% Tween-20 in PBS) and incubated on cells at 4°C for 2 h. Cells were then washed three times with PBS and then incubated for 2 h in anti-mouse IgG horseradish peroxidase (HRP)-conjugated sheep antibody (GE Healthcare) diluted 1:4,000 in antibody dilution buffer. Assays were then washed three times with PBS and exposed to 0.5 mL of True Blue peroxidase substrate (KPL) for 15 min. Plates were then washed with water and dried before plaques were counted.

Cell infections.

For experiments measuring RNA and protein levels of MMR genes and ISGs, H441-CR cells were infected with WT PR8 at MOI=5 for one hour before virus was removed and complete media was added. Cells were incubated at 37°C for 3–24 h before total RNA was collected, or for 24–48 h before total protein was collected. For experiments using siRNA-treated cells, infection was done at 48 h after siRNA transfection to ensure maximum knockdown of target genes.

qRT-PCR assays.

For cell culture-based qRT-PCR experiments, total RNA was collected using Monarch Total RNA Miniprep Kits (NEB). For the mouse lung IAV NP qRT-PCR, lungs were collected at 4 dpi (with 1200 PFU virus) and homogenized directly in Trizol, before using chloroform and isopropyl alcohol to complete the RNA extraction. For the confirmation of siRNA knockdown, measurement of ISG RNA levels, the comparison of relative MSH6 RNA levels in amiRNA-transfected cells, and the quantification of mouse lung IAV NP levels, samples were analyzed using the EXPRESS One-Step Superscript Universal qRT-PCR Kit (Thermo) and gene-specific TaqMan Expression Assay Probes (Thermo) (Supplementary Table 6). For the MMR gene panel, the qRT-PCR was performed in two separate steps. First, single-stranded cDNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Thermo). Second, cDNA was loaded into Custom TaqMan Array 96-well Plates (Thermo) we designed to include all major genes directly involved in the DNA MMR pathway as well as two housekeeping gene controls for normalization.

Western blotting.

Equal amounts of protein were loaded into 4–20% acrylamide gels (Bio-Rad) and transferred to nitrocellulose membrane. PBS with 5% (w/v) nonfat dried milk and 0.1% Tween-20 was used to block for 2 h at 4°C. Primary antibodies were then incubated with the membrane overnight at 4°C. Antibodies used were rabbit anti-MSH2 (Cell-Signaling Technology D24B5), rabbit anti-MSH6 (Abcam EPR3945), mouse anti-HA (PY102, gift from Thomas Moran at Mt. Sinai), rabbit anti-Mx1 (Abcam 95926), and mouse anti-α-Tubulin (Sigma T5168). Membranes were washed five times in PBS with 0.1% Tween-20 and then anti-mouse-HRP or anti-rabbit-HRP secondary antibody was added for 1 h. The membrane was then washed five times and Clarity or Clarity Max ECL substrate (Bio-Rad) was added before being exposed to film and developed. Uncropped scans of all Western blots are displayed in Supplementary Figure 6.

DNA MMR activity assay.

The assay was designed as previously described¹⁷, with some adjustments. In short, we digested our synthesized pLCMV-NanoLuc plasmid with NheI and PflFI to remove the NanoLuc start codon region. We then ligated in annealed oligos containing a G-G mismatch in the third position of the NanoLuc start codon; this was referred to as our “Mismatch plasmid.” A set of annealed oligos with a WT start codon was also ligated in to be used as a normalization control. The annealed oligos also contained a disrupted HpaI restriction site, allowing us to specifically eliminate any uncut parental pLCMV-NanoLuc plasmid (which contains an intact HpaI site directly upstream of start codon) following plasmid ligation. Validation of the MMR activity assay was done using A549 cells transfected with MSH2 and MSH6 siRNA. For the comparison of A549 and H441 cells, the cells were infected at MOI=10, transfected with 0.5 μg of the WT or Mismatch plasmid at 24 hpi, and then analyzed at 16 h post-transfection. To measure NanoLuc expression, cells were lysed with Luciferase Cell Lysis Buffer (NEB) and then evaluated using the Nano-Glo Luciferase Assay Reagent Kit (Promega).

DNA damage assays.

For phospho-H2AX staining, H441 cells were plated on collagen-coated cover slides and infected with WT PR8 (MOI=5) for indicated lengths of time before fixation with 4% PFA. Positive controls were treated with 1 μM Etoposide (Abcam) or 5 mM H₂O₂ (EMD Millipore) for 30 min before fixation. Cells were stained using the Histone H2A.XS139ph antibody (Active Motif, #39117). DAPI was added in the mounting solution for nuclei staining. Images were collected using a Leica SP5 inverted confocal microscope. Images within the same panel were collected at the same time using the same settings and analyzed using ImageJ (NIH). Quantification of 8-OHdG was done using the OxyDNA Assay Kit (Calbiochem) following the manufacturer’s instructions with H441 cells infected with WT PR8 (MOI=10) for 48 h. The gating strategy used for the 8-OHdG flow cytometry analysis is shown in Supplementary Figure 7. The modified comet assay was performed as previously described^28,29, with some alterations. H441 cells were transfected with MSH2+6 or non-targeting control siRNA, and then infected with WT PR8 (MOI=10) 48 h after siRNA transfection. At 24 hpi, cells were collected and mixed with 1% TopVision Low Melting Point Agarose (ThermoScientific) before plating on agarose coated microscope slides. Agar-embedded cells were then lysed and treated with Fpg restriction enzyme. After alkaline treatment to neutralize the restriction enzyme, the cells were exposed to gel electrophoresis to induce migration of digested nuclei. Cells were washed with neutralization buffer and then stained with DAPI. Images were collected using a Leica SP5 inverted confocal microscope. Images within the same panel were collected at the same time using the same settings and analyzed using ImageJ (NIH). For the ROS scavenger assays, A549-CR and H441-CR cells were treated with 0.5 mM Trolox ((±)-6-Hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid, Sigma-Aldrich) for 1 h before infection with IAV-Cre (MOI=5). Fresh Trolox (0.5 mM) was then maintained in the cell media for 3 dpi before being removed and replaced with media containing no Trolox. At 5 dpi, cells were fixed with 4% PFA and stained with Hoescht dye before analyzing the percentage of ZsGreen⁺ cells using a Cellomics Arrayscan VTI HCS.

Expression of Nanoluc and ISRE-GFP reporters.

All experiments measuring expression of the pLCMV-NanoLuc plasmid were conducted in H441 cells and done using the uncut reporter plasmid we originally synthesized for the MMR activity assay described above. For the ROS experiments, 5 mM H₂O₂ was added to the media on the cells and incubated at 37°C for 30 min before being washed and replaced with fresh complete media. Immediately following the addition of fresh media, the cells were transfected with 0.5 μg of pLCMV-NanoLuc plasmid using Lipofectamine2000 (Thermo). After 6 h, the cells were lysed with Luciferase Cell Lysis Buffer (NEB) and luminescence was measured using the Nano-Glo Luciferase Assay Reagent Kit (Promega). For the IAV infection experiments, cells were transfected with siRNA and incubated for 48 h before being infected with WT PR8 at MOI=10. At 18 hpi, the infected cells were transfected with 0.5 μg of the pLCMV-NanoLuc plasmid. After 6 h, cells were lysed with Luciferase Cell Lysis Buffer (NEB) and luminescence was measured using the Nano-Glo Luciferase Assay Reagent Kit (Promega). Cell viability was confirmed in all experiments using cells treated/infected in parallel and the CellTiter-Glo Luminescent Cell Viability Assay (Promega). We designed the ISRE-GFP reporter in the pTRIP vector by inserting seven consensus ISRE sites directly upstream of a minimal CMV promoter. This promoter drives super-folder GFP fused to a mouse ornithine decarboxylase domain to decrease the protein half-life. For the ISRE-GFP reporter experiments, 293T cells were transfected with siRNA and incubated for 48 h before transfecting in the reporter. After 24 h, 1000 U/mL IFN-alpha was added to the cells and flow analysis was done to look at the level of GFP reporter expression.

RNAseq.

H441 cells were transfected with siRNA and incubated for 48 h before being infected with WT PR8 (MOI=10). Total RNA was collected at 24 hpi using Monarch Total RNA Miniprep Kits (NEB). RNA was then prepped for RNAseq submission using the NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB), NEBNext Ultra II RNA Library Prep Kit for Illumina (NEB), and NEBNext Multiplex Oligos for Illumina (NEB). Mapping of raw reads to the human hg19 reference genome was accomplished using a custom app on the Illumina BaseSpace Sequence Hub⁷. After data normalization, average read values were compared across samples. For comparisons in which some samples had zero reads detected for a specific gene, 1 read was added to all values in the sample to complete analyses that required non-zero values. Raw data is available at NCBI GEO (Series GSE130189).

ISG expression assays.

All experiments measuring ISG expression were conducted in H441 cells. For experiments including ROS treatment, cells were transfected with siRNA and after 48 h, 5 mM H₂O₂ was added to the media for 30 min. Media was then aspirated and replaced with media containing 500 U/mL IFN-alpha for activation of ISG expression. Cells were then incubated at 37°C for 3 h before total RNA was collected and for 24 h before total protein samples were collected. For IFI44L and IFIT1 experiments, cells were transfected with siRNA and incubated for 48 h before infecting with WT PR8 at MOI=10 and collecting total RNA at 48 hpi.

Northern blot.

Total RNA samples were collected from either MLE15 cells infected in a 10 cm dish with the PR8-Cre+amiRNA viruses for 24 h, or from MLE15 cells transfected in a 10 cm dish with 15 μg of the Seg6-amiRNA plasmids for 24 h. Equal amounts of RNA was loaded into a 15% TBE-urea gel (Bio-Rad) and transferred to a nitrocellulose membrane. The dried membrane was UV-crosslinked and then exposed to pre-hybridization solution for 1 h. Probes specific for the GFP amiRNA, MSH6 amiRNA, and U6 control snRNA were labeled with ³²P-ATP and then added to the membrane for 1 h. Membrane was washed two times with saline-sodium citrate (SSC) buffer plus 0.05% SDS and then exposed to film for 96 h before developing. Uncropped scans are displayed in Supplementary Figure 6.

amiRNA activity confirmation.

MLE15 cells were co-transfected in 24-well plates with 500 ng of the Seg6-amiRNA plasmids and 500 ng of a pLEX helper plasmid with puromycin resistance marker using Lipofectamine2000 reagent (Thermo). After 24h, co-transfected cells were selected for with 2.5 μg/mL puromycin. After 48 h of selection (72 h post-transfection), total RNA was collected using TRIzol reagent. MSH6 RNA levels were then quantified from each of the samples using one-step qRT-PCR.

ALI culture infections.

Tracheal epithelial cells were isolated from transgenic Cre-tdTomato-reporter mice and cultured at air-liquid interface for 14 days to allow formation of a pseudostratified epithelium. This was monitored using markers for mature differentiated cell types (KRT5 – basal stem cells, ACTUB – mature ciliated cells). These cultures were then infected with 500,000 PFU of WT PR8, GFP-amiRNA, or MSH6-amiRNA virus. At 1 and 8 dpi, cultures were collected, rinsed two times with PBS, and fixed with 2% PFA at room temperature for 15 min. Cultures at 1 dpi were stained with PY102 to quantify the number of infected cells and cultures at 8 dpi were imaged using tdTomato as a marker of survivor cells. All cultures were imaged on a Leica SP5 inverted confocal microscope using a 40X oil objective. Images within the same panel were collected at the same time using the same settings. Images were analyzed using Fiji (NIH) using macros that were written to automatically quantify the number of PY102+ and tdTomato+ cells per field of view.

Mouse infections.

Eight-week-old female C57BL/6J mice were used for all experiments, with a sample size of 4 mice per dose of virus. Prior to infection, mice were anesthetized with a 100-μL injection of a ketamine-xylazine mixture. After being tail-marked and weighed, mice were intranasally infected with 40 μL of virus diluted in pharmaceutical-grade PBS. Mice were weighed daily and euthanized once their body weight reached <80% of the starting weight as a humane endpoint. Euthanasia was performed via CO₂ as the primary method and a bilateral thoracotomy was performed as the secondary method. All procedures were approved by the Duke University IACUC.

Statistical analysis.

Comparison of data sets was performed using an unpaired, two-tailed Student t tests unless otherwise stated. Analysis was performed using Prism 7 and Prism 8 (GraphPad) software.