Dual effects of influenza A virus PA-X on suppression of cytokine signaling and MHC I pathway disruption in the human airway epithelium

Article activity feed

1. 

   Review Commons

   Sep 25, 2026

   Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Reply to the reviewers

   Revision Plan after Review Commons Reviews

   1. General Statements We are excited that the reviewers think that “The findings demonstrate that PA-X can modulate both the innate and adaptive immune responses to infection, providing important insights into the function of this protein, which is conserved across the majority of IAV strains.” and that they acknowledge that “the story is really strong and improves the current state of the art in this field.” We also thank the reviewers for recognizing the physiological relevance of our study, stating that “strength of the study is the use of ALI cultures, thus a very relevant primary cell model, closely mimicking …

   Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Reply to the reviewers

   Revision Plan after Review Commons Reviews

   1. General Statements We are excited that the reviewers think that “The findings demonstrate that PA-X can modulate both the innate and adaptive immune responses to infection, providing important insights into the function of this protein, which is conserved across the majority of IAV strains.” and that they acknowledge that “the story is really strong and improves the current state of the art in this field.” We also thank the reviewers for recognizing the physiological relevance of our study, stating that “strength of the study is the use of ALI cultures, thus a very relevant primary cell model, closely mimicking the in vivo situation” and that “use of primary human bronchial epithelial cultures and their analysis at the single cells level allowing to distinguish infected from bystander cells add important information” for immunomodulation by influenza A virus PA-X. We also thank the reviewers for noting that our study uses “thorough and comprehensive scRNA-seq analysis corroborated by orthogonal approaches to verify their claims (cytokine measurements and flow cytometry for MHCI)”

   One reviewer stated that our study “neither brings up a new concept, nor does it describe mechanisms which are not yet reported in the context of other viruses.” While influenza A virus is certainly not the first virus to antagonize MHC I antigen presentation, we disagree that this makes our study less significant for several reasons. First, while this is an accepted mechanism of immune evasion in many DNA viruses, it is not currently a process that is widely recognized in RNA viruses in general and influenza A virus in particular. There has only been one previous report to this effect in influenza A virus, and only in cell lines, and it is not a mechanism that is widely discussed in the field. However, reduced MHC I antigen presentation may have an important impact on effective responses to influenza A virus, in particular in individuals with existing immunity and memory T cells, i.e. vaccinated individuals. Therefore, our study is significant because it highlights this new aspect of influenza A virus immune evasion and suggests a revised model for influenza A viral adaptive immune evasion. In this model, antigen presentation inhibition is an active mechanism of evasion complementary to the well-established antigenic drift as “passive” evasion. Our study also highlights that antigen presentation inhibition by influenza A virus does occur in a relevant setting – a model of the bronchial epithelium. Lastly, the study is significant because it points to a new facet of PA-X activity and thus of influenza A virus-host interactions. So far studies of this protein have focused on effects on innate immunity and may thus be missing an important component of its function. We have revised the text to emphasize the importance of antigen presentation inhibition as an understudied mechanism of influenza A virus immune evasion and as a new function for the PA-X protein (lines 140-143, 505-509, 529-531).

   We also want to emphasize that our results provide the first description of a mechanism for MHC I downregulation in influenza A virus infection. Moreover, they suggest that PA-X decreases peptide-loaded surface MHC I levels in two ways. On the one hand, PA-X directly reduces the total amount of MHC I by downregulating MHC I mRNAs, thus presumably reducing new MHC I protein synthesis. On the one hand, PA-X decreases the expression of other genes involved in antigen presentation, which will indirectly reduce correct folding and loading of the MHC I complex. Unloaded and poorly folded MHC I does not carry out antigen presentation and is essentially “invisible” to the antibodies we used, which only detect correctly folded and loaded MHC I. Unloaded MHC I is also thought to be highly susceptible to protein degradation. We have clarified this model in the text (lines 521-527). Nonetheless, we plan to refine the PA-X regulation model by testing whether there is evidence for direct effects on trafficking (i.e. retention of the MHC I proteins in intracellular compartments in the presence of PA-X; see section 2).

   In this version of the manuscript, we have already substantially revised the text in response to the reviewers’ comments, as described in section 3 below. We have carried new analyses requested by the reviewers, adding several supplemental figure panels. We have edited the text to clarify the significance of the study and the model of PA-X regulation of MHC I in response to Reviewer 2’s concerns. We have clarified the number of replicates used, moved one of the supplemental figures to the main figures (Figure S7, now Figure 4C-G), and better described the 5’ RACE results in response to Reviewer 3’s comments. We provide a version of the manuscript with changes highlighted in yellow (including legends for new Figures or Figure panels).

   In addition, we plan to carry out additional experimental revisions as described in section 2 below. In particular, we will investigate several questions raised by Reviewer 2, where technically feasible, to further strengthen the mechanistic aspect and significance of our manuscript.

   We agree with the reviewers that this manuscript would be of interest to a broad immunology and virology audience as well as influenza biology researchers.

   1. Description of the planned revisions

   Reviewer 1

   The authors propose that PA-X directly targets for degradation several mRNA of genes involved in antigen processing and presentation, and they tested it in HEK 293T cells transfected with empty vector, wt or catalytically inactive PA-X. Could the authors confirm these data [5’ RACE] in hBEC, infected with WT or ∆X viruses?

   We agree that it would be ideal to confirm cleavage during viral infection in this more relevant model. We will attempt to perform 5’RACE experiments with infection in ALI cultures. We note that this may present a challenge for several technical reasons. First, because a low percent of cells are infected in ALI cultures, the abundance of the fragments may be too low to detect in the assay. In addition, the procedure requires a relatively large amount of starting RNA, which may not be easily obtainable from an ALI culture. Lastly, the 5’RACE experiments in the manuscript were performed in cells that had the cellular exonuclease Xrn1 knocked down, as we previously showed that this stabilizes the RNA fragments. Replicating Xrn1 knock-downs in the ALI culture may prove technically challenging.

   Reviewer 2

   • Conversely, one would expect also ISG-induction in bystander cells, due to higher IFN-lambda secretion in the delta-PA-X condition and this would hinder viral spread at later time points. An experiment I would recommend.

   To address this possibility, we will repeat ALI infections and measure titers and percentage infected cells up to day 7. However, we think it is unlikely that we will see a difference. In the current manuscript, we include viral titers up to day 4 (Figure 1C) and the percentage infected cells at day 3 (Figure S8A). In both cases we do not see a difference between cultures infected with WT and PA-X-deficient viruses. Moreover, although ISG responses in bystander cells are different at 1 day post infection (DPI), the differences are largely gone by 3 DPI (Figure 2D). The transitory nature of the ISG modulation likely prevents long-term effects on viral spread. The loss of ISG differences may be due to threshold effects of IFN signaling, such that maximal signaling is achieved in both conditions despite apparent differences in secreted IFN levels (Figures 2C-D).

   • Can PA-X alone, for instance expressed from a lentivirus, downmodulate MHCI and alter IFN signalling and cytokine secretion?

   To address this question, we will express PA-X in isolation and measure MHC I downregulation and cytokine secretion in the absence of other viral proteins. In the latter case, we will use synthetic agonists to induce IFN and/or other cytokines, as these signals will likely be otherwise absent in uninfected cells.

   • Does PA-X interact with MHCI, how does it alter trafficking? Is MHCI retarded in the ER, is it channeled to the lysosome etc?

   Because of PA-X’s function as an RNase and its predominantly nuclear localization (Khaperskyy, Schmaling et al. 2016, PMID: 26849127; Hayashi et al. 2016, PMID: 27226377; Daly et al. 2026, PMID: 41511082), it is not likely that PA-X directly interacts with MHC I. Our results suggest it downregulates MHC I at the RNA levels. However, PA-X may also indirectly alter trafficking, since our results also indicate that PA-X cleaves host mRNAs encoding proteins that are essential to MHC I antigen processing and presentation. Reduction in the levels of these proteins prevents the normal peptide loading trafficking of MHC I, and in some cases leads to increased MHC I protein degradation. For example, inhibition of HSP90 is sufficient to reduce surface levels of folded MHC I and increase surface levels of unfolded MHC I, which cannot present antigen (Callahan et al. PMID: 18216248). (Since the antibody we used only binds folded and peptide-loaded MHC I, poorly folded and unloaded complexes at the surface would be invisible in our experiments). To test the possibility that there is a change in MHC I trafficking independent of total levels, we will use confocal microscopy and/or flow cytometry to measure levels and localization of folded and unfolded MHC I protein. We will use an antibody specific to unfolded HLA-B and HLA-C (Ruggiero et al., 2025, PMID: 41067637) to uncover additional defects in the MHC I exposure pathway.

   Reviewer 3

   1. Additional statistical tests should be added to several figures including for viral titre comparisons (Figure 1C), heat map analysis (Figure 2D and Figure 4B) and ELISA analysis of IFN beta (Figure S5).

   We have already added statistical analysis for Figure 1C, 4B and S5 (see section 3 on revisions already incorporated in the manuscript). For Figure 2D, we plotted data from a custom set of genes, rather than relying on the results of a gene set analysis program. In the revision, we will add statistical testing for the general impact of PA-X on ISGs as a group, which may require us to develop some custom code.

   1. In line 387-391 the authors state that they also examined another viral strain to determine whether the reduction in MHC I expression is conserved, reporting a trend towards reduced MHC I expression in Figure S8C and D. However, this effect appears modest, with no statistically significant difference reported, and only a subset of donors was included in the analysis. Therefore, the subsequent description of the reduction as "striking" appears somewhat inconsistent with the data presented in Figure S8 and may benefit from rewording to more accurately reflect the findings.

   We apologize for the confusion - the “striking” comment was meant to refer to the MHC I phenotype overall, and mainly the results obtained with A/Perth/16/2009 H3N2, not the A/California/04/2009 H1N1 results. We have edited the text to clarify this (line 413). Nonetheless, we will repeat experiments measuring the effect of A/California/04/2009 H1N1 on MHC I levels in ALI cultures in additional donors.

   1. Description of the revisions that have already been incorporated in the transferred manuscript

   We have added analyses and new figure panels, as well as edited the text to address the following comments made by the reviewers. We have highlighted the changes in yellow in the text. We have also moved Figure S7 to the main figures (now Figure 4C-G), leading to changes in figure numbering. (Reviewer comments are in blue italics and our responses in black.)

   Reviewer 1

   Third, they show that PA-X could also limit the production of pro-inflammatory and lung injury associated cytokine. They should take advantage of the single cell RNA data and analyse the data as done for IFNs, to define if the observed reduction is a direct effect of PA-X in infected cells only or an indirect effect of PA-X presence/absence on all cells. Do bystander cells produce IL6, IL1b, G-CSF, FGF, etc? Is this production reduced in wt vs dX infections, in both infected and bystander cells?

   We have now added a heatmap (Figure S6A) that shows the expression of the differentially secreted cytokines and growth factors in infected vs. bystander cells. Interestingly some of the cytokines are expressed more highly in infected cells, while others in bystander cells. Moreover, in some cases the pattern changes at different timepoints. Surprisingly, this analysis shows that the relative mRNA levels in the different conditions do not correlate with secreted protein levels, i.e. PA-X does not clearly downregulate these cytokines and growth factors at the RNA level. Therefore, we cannot use the RNA profiles to determine whether secretion is changed in infected or bystander cells. Nonetheless, this analysis shows that PA-X is likely to have both cell-intrinsic and paracrine effects on the production of both inflammatory and repair-associated cytokines. Also, this analysis reinforces the importance of using programs like CytoSig to identify bystander cell responses to protein secretion rather than relying solely on the transcriptional regulation of the cytokine. We have added a discussion of this result in the text (lines 331-339).

   The data presented confirm that IAV infected cells downregulate expression of HLA-A, HLA-B, HLA-C, at both the protein and RNA level and that this reduction is partially dependent on PA-X. Is this happening in all infected cell types? In Figure 4c, I noticed a small proportion of dX infected cells significantly upregulating HLA-ABC expression. Are these a spillover of non infected cells?

   We have added heatmaps showing the levels of differentially expressed antigen presentation genes at 1 and 3 DPI in ciliated and club cells (Figure S7B). We find that PA-X reduces expression of these genes in the infected portion of both ciliated and club cells, indicating this is not a cell-type specific effect.

   We also noticed the small proportion of cells with significantly upregulated HLA-ABC expression. We do not know where this population comes from. It is possible that these cells are not infected but have bound virus on the surface. Alternatively, a small portion of cells may have been infected by defective viral particles that increase antiviral responses and do not properly express all viral proteins. Also, while the small population with high HLA-ABC is only presented in the ∆X-infected sample in the replicate we used as a representative example in the manuscript, in some of the other experimental replicates we see this population in both WT and ∆X-infected cells or only in some technical replicates. Overall, this population is not detected consistently. Because we used median fluorescence intensity instead of mean fluorescence intensity, we do not believe this small population makes a significance contribution to our measurements.

   In Fig 4b at 3dpi, HLA-G is indicated as "up with PA-X" but the color code indicates the opposite. Please clarify and discuss the possible implication in the context of NK cytotoxic activity.

   We thank Reviewer #1 for pointing out this error. We have amended the heat map (Figure 4B) to show that HLA-G is increased during infection with the PA-X deficient virus, specifically in infected cells. It is possible that the elevated HLA-G expression in ∆X infection results in greater surface protein expression, which could lead to a greater inhibitory effect on NK cell cytotoxicity vs. WT-infected or mock-infected cells. This is certainly an avenue for future exploration, as it would be interesting to explore the balance between PA-X MHC I antagonism and impacts on NK cell cytotoxicity. We now mention the possible effect of changes in levels of HLA-G (and of other non-classical MHC I complex) on NK cell activity in the Discussion (lines 556-559).

   Reviewer 2

   • If PA-X alters MHCI levels, this might activate NK-cell killing of infected cells, as they are triggered by low HLA-levels; so does PA-X also affect NK-cell ligands?

   We have analyzed the results for activating NK cell ligands in the scRNAseq dataset. Many of them are somewhat induced during infection (vs. mock conditions) and are also generally expressed at higher levels in WT vs. ∆X infections, particularly at 3 DPI. However, none of the changes reach statistical significance (padj > 0.05 with FindAllMarkers). Higher expression of NGK2D ligand genes in WT vs. ∆X infection is intriguing, and suggests that the virus does not prevent upregulation of these NK cell-activating signals through PA-X. Therefore, these results do not suggest that there is a compensatory change to protect cells from NK cell killing. We have now included these data as Figure S9 and discuss them in the Discussion section (lines 559-569).

   Reviewer 3

   1. The scRNAseq data is validated for several conclusions by additional assays such as Luminex to analyse cytokine levels and a 5' RACE workflow to look at cleavage of host genes by PA-X. However, for several of these additional assays it is unclear on how many replicate experiments were performed. This should be more clearly stated and ideally there should be three independent experiments conducted.

   We have edited the figure legends to more clearly indicate that all our experiments were carried out in ≥3 biological replicates, except for the scRNAseq, which was done in two biological replicates with different donors. This is true for experiments with ALI cultures, as well as Calu-3 and HEK293T cells. In addition, for each ALI experiment in the main figures of the manuscript, we carried out at least one biological replicate with each of the different donors we used, and we now explicitly state this in the legends.

   1. Many experiments in this paper involve comparisons between WT and PA-X-deficient viruses. To support the conclusions regarding the effects of PA-X, it is important to demonstrate that infection rates are comparable between the two viruses, including with appropriate statistical analysis.

   We agree that this is an important point. Indeed, we collected apical washes from most experiments to measure viral replication and for all flow cytometry experiments we assessed the percentage of NP+ positive cells. We also measured the percentage of infected cells based on influenza A virus reads in the scRNAseq. These data are already included as Figure 1C, Figure S8A, and Figure 1G. We have not seen any evidence of difference in infection rates between the two viruses. Therefore, our results indicate differences in infection rate are not a concern. We have added explicit statements to this effect in the text (lines 398-400).

   1. Additional statistical tests should be added to several figures including for viral titre comparisons (Figure 1C), heat map analysis (Figure 2D and Figure 4B) and ELISA analysis of IFN beta (Figure S5).

   We did not include p values for Fig 1 C and S5 as the differences were not statistically significant. We have now added statements to the figure legend to clarify that statistical testing was carried out and that differences were not significant. In Figure 4B, the heat map already only includes genes that were identified as differentially expressed genes (DEGs) in infections with WT vs. PA-X-deficient viruses by Seurat’s FindAllMarkers function. The cutoff for DEGs were |log2FC| > 0.25 and p-adjusted

   1. Figure 1 part C is listed to be a viral titre from a TCID50 assay but the units on the graph are PFU/ml. It should be TCID50/ml if it is a TCID50 assay but the materials and methods describe a plaque assay. Titres also seem quite high compared to what other literature appears to report. The figure legend mentions n=11 from 5 separate donors, is this from separate infections or were these technical replicates of the 5 donors?

   The confusion stems from the fact that we aggregated data from both TCID50 and plaque assays in the original figure. To do so, we converted the numbers to PFU/ml by multiplying by 0.7 according to the Poisson distribution. This conversion is commonly used (https://www.atcc.org/resources/culture-guides/virology-culture-guide). However, to avoid confusion and the potential confound of using different assays, we have now excluded the plaque assays from the analysis and changed the axis label to TCID50/ml. We also have recalculated the TCID50 numbers using the Reed-Muench method to ensure that they are correct. This figure now represents n = 8, which refers to 8 separate independent experiments with 1 experiment for donors 1 and 5, and two experiments each for donors 2-4. We now clearly state this in the legend. We collected apical washes to titer viruses for all replicate experiments shown in Figures 2C, 3C, and 5A (4D in the previous submission), hence the large number of replicates.

   We were also surprised at the robust viral titer that is produced in the primary ALI cultures and agree that it is several logs higher than we have observed in cell lines. However, anecdotally many people report that human seasonal viruses replicate much more efficiently in the ALI culture, likely because these cultures are closer to the natural site of viral replication. Additional sources of the difference may be the strain used and the inoculation and collection procedures.

   1. Figure S5-mislabelled graph- only has a part A in the figure but a part A and C are in the figure legend. Should IFN-X be IFN-beta? Perform statistical test to confirm no significant difference.

   We apologize for the mistake. We have corrected the figure legend and the figure, which was missing a panel label. Both now have panels A and B. As mentioned above, we have also now clarified in the figure legend that the differences were not statistically significant.

   1. Figure 3- could you clarify what the n= 5-11 independent experiments describes and how many independent experiments were performed for the data in Figure 3C.

   We thank the reviewer for pointing out this error, which was due to copy-pasting of text from another legend, and have corrected the figure legend to n = 5, with one biological replicate for each donor.

   1. Figure S6 B- These cytokines are either below or above the limit of detection of the standard curve of the assay. For cytokines above the upper limit of detection, could the samples be appropriately diluted and rerun to obtain values within the standard curve range? For cytokines below the lower limit of detection, it is unclear whether these data provide meaningful quantitative information and, therefore, if they should be included. Additionally, the presentation of the data in Figure S6B should ideally be consistent with the graphs in Figure 3C, including the labelling and formatting of the axes.

   We thank the reviewer for pointing this out. Considering that there does not appear to be any PA-X-dependent differences for cytokines above the limit of detection, we did not repeat this test at different dilutions. To avoid misrepresenting the data, we have removed the results for all cytokines that were outside of the standard curve range.

   1. For Figure S7 (5′ RACE workflow), this appears to be important data supporting a potential mechanism by which PA-X inhibits the expression of antigen-processing and presentation genes. Therefore, it may be more appropriate for this experiment to be presented in the main body of the manuscript. However, the DNA gel images are not particularly clear, and there appear to be several nonspecific bands. In particular, Figure S7E is not very convincing. Additionally, it appears that only one independent experiment was performed but ideally there should be at least three independent biological replicates.

   This experiment was indeed carried out in three independent biological replicates, which we have now clarified in the figure legend. We also have increased the contrast of the gels in ImageJ for better visualization of the bands and captured a clearer representative picture of the HSP90AA1 RACE experiments (formerly Figure S7E, now Figure 4G). The quality of the results is an unfortunate consequence of the 5’ RACE assay, especially when using endogenous mRNAs. As the RACE adaptors are ligated to all 5’ phosphate-bearing RNA fragments and the gene specific primers amplify any fragment coming from the target gene, many background fragments are also detected. To verify our results, we sequenced the PCR products we amplified. Importantly, the sequencing supported the idea that only the amplicons in the WT PA-X transfected sample represented the PA-X cleavage products identified in the 5’ RACE-seq. In these samples, the amplicons aligned to the gene reference only 3’ of the cleavage site and the adaptor sequence was detected 5’ of the cleavage site. In contrast, sequences from faint bands of similar sizes in cells transfected with empty vector and the PA-X D108A catalytic mutant aligned to the target gene (CALR or HSP90AA1) 5’ of the expected cut site and did not align to the RACE adapter sequence. We have emphasized this in the text (lines 377-379) and figure legend. We also agree with the reviewer’s point that this would be better suited as a main figure and have made this change (now Figure 4C-G).

   1. In line 339-341 a downregulation of genes relating to antigen processing and presentation is described for both ciliated and club cells and reference is made to table 1 but in table 1 this term is only listed for the club cells not the ciliated cells.

   We only listed the top three gene sets in this table, but antigen processing and presentation was within the top 10 gene sets for ciliated cell cluster 10 at 1 DPI. We have now clarified this in the text (lines 351-358). In addition, the “protein folding” gene set is in the top three gene sets from ciliated cells. This gene set contains several genes involved in antigen processing (for example, HSP90AA1, HSP90AB1, CALR, CANX). Additionally, the existing gene set analysis in all infected cells at 1 DPI in Figure 4A, the new one in Figure S7A for 3 DPI, and the new heatmaps analyzing antigen processing and presentation gene expression separately in ciliated and club cells (Figure S7B), confirm that the PA-X dependent downregulation of antigen processing and presentation genes is shared between cell types.

   1. In line 343 to 344 it is stated that antigen processing and presentation is one of the top downregulated gene sets for both 1 and 3 DPI with a reference to Figure 4A but in the Figure of 4A the figure legend and image only mention the 1 DPI data. Figure 4B has analysis from both timepoints.

   We have added the gene set enrichment analysis plot for 3 DPI in the supplemental data (Figure S7A) to confirm that antigen processing and presentation gene sets are downregulated by PA-X.

   1. Figure 4F/G appears to be the same data as Figure S8H- is this a duplication of the same data? Is this also only from one independent experiment?

   They are different results. Figure 4F/G represents the MHC I recovery after acid strip in the influenza A virus-infected (NP+) cells in the population. Figure S8H represents the MHC I recovery after acid strip in the uninfected (NP-) cells in the same sample. We have edited the panel titles to make this difference clearer. We separated the two populations because we found that MHC I trafficking is significantly reduced in cells that are infected, but in not bystander cells, indicating that the MHC I regulation is due to a cell intrinsic activity rather than paracrine signaling. We felt this is important to include because there is a PA-X-dependent transcriptional reduction in both infected and bystander populations at least at 1 DPI (Figure 4B). However, the results in Figure S8H show that in the bystander cells, this change in RNA levels does not translate to reduced trafficking. We have edited the text and figure panel to make this clearer. Like all other experiments in the manuscript, these results are averages of multiple biological replicates, in this case 3-4 (samples from one timepoint were accidentally lost during processing in one of the replicates). We have edited the legends to make this clear.

   1. Description of analyses that authors prefer not to carry out

   Reviewer 1

   An experiment in this direction would be a great addition to the paper. For example, the authors could infect mice with a primary wt virus, and then test CD8 activity 1 month later during a secondary heterotypic infection with either a wt or dX virus. The authors themselves recognize that functional effects of PA-X disruption of MHC I trafficking need to be tested in vivo, to understand how this disruption affects T cell clearance of infected cells during primary and, most importantly, secondary infections. An experiment in this direction would be a great addition to the paper.

   We thank the reviewer for this idea, and we agree that this experiment would be a great addition to this story. However, we feel this is beyond the scope of the current study because of the amount of optimization involved in changing species and potentially viral strain. Because the mRNAs of mice and humans have different sequences and PA-X cleaves target based on sequence (Gaucherand et al. 2023, PMID: 37349586), we would have to establish that PA-X targets the same mRNAs and proteins in mouse cells. In addition, to do a secondary infection and isolate T cell effects, we would need to use the PR8/X31 system. This system is based on the lab-adapted PR8 strain, as both strains carry the polymerase segments, and thus PA-X, from PR8. Therefore, we would have to first establish how well these strains recapitulate the effects of seasonal strains on MHC I. This is particularly a concern since a previous report of MHC I downregulation in influenza A virus observed strain-specific differences in A549 cells, and found that the common mouse-adapted strain PR8 had the least effect on MHC I levels among the tested strains (Koutsakos et al. 2019, PMID: 31191533). Nonetheless, we agree that it is very important to test in vivo effects of MHC I antagonism, and we plan to perform in vivo studies in the future.

   Reviewer 2

   • How are the effects on IFN-signalling when using a NS1, PA-X double KO as compared to a NS1-deficient IAV? Are both proteins acting complementary?

   We agree with the reviewer that the interplay of PA-X and NS1 is a very interesting question. However, in our opinion investigating this question is beyond the scope of this manuscript due to its complexity. First of all, loss of NS1 dramatically reduces influenza A virus replication in ALI cultures (for example, Haye et al. 2009, PMID: 19403682), making comparisons between NS1 mutant viruses and other strains tricky. In addition, a recent study suggested that NS1 is required for PA-X activity during infection (Bougon et al. 2024, PMID: 38629840), which further complicates the task of dissecting the specific effects of NS1 vs. PA-X. The molecular nature of this interaction is currently unknown, which means we cannot easily use mutants to separate the effects of NS1 on PA-X from those of NS1 on gene expression in general.

   • Can they feed IAV-specific T cells to the system and will WT infected cells perform much better under such conditions, i.e. evade killing as compared to PA-X infections?

   In principle this would be a great experiment. However, there is currently no system for modeling T cell cytotoxicity ex vivo. Anecdotally, according to other groups working on modeling T cells effects in ex vivo systems, introducing antigen-specific T cells into the ALI system for a cytotoxicity assay does not work. While we hope that in the future these techniques will be available, at present this experiment is not feasible.

   Read the original source
2. 

   Review Commons

   Sep 25, 2026

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #3

   Evidence, reproducibility and clarity

   The authors have employed an ex vivo model of the human airway and single cell RNA sequencing to identify effects of the IAV PA-X protein on infection. By comparing infection of a wild type IAV H3N2 strain and PA-X deficient virus, the authors identify that PA-X reduces secretion of several cytokines and decreases surface expression of MHC I on infected cells.

   Specific points:

   1. The scRNAseq data is validated for several conclusions by additional assays such as Luminex to analyse cytokine levels and a 5' RACE workflow to look at cleavage of host genes by PA-X. However, for several of these additional assays it is unclear on how many …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #3

   Evidence, reproducibility and clarity

   The authors have employed an ex vivo model of the human airway and single cell RNA sequencing to identify effects of the IAV PA-X protein on infection. By comparing infection of a wild type IAV H3N2 strain and PA-X deficient virus, the authors identify that PA-X reduces secretion of several cytokines and decreases surface expression of MHC I on infected cells.

   Specific points:

   1. The scRNAseq data is validated for several conclusions by additional assays such as Luminex to analyse cytokine levels and a 5' RACE workflow to look at cleavage of host genes by PA-X. However, for several of these additional assays it is unclear on how many replicate experiments were performed. This should be more clearly stated and ideally there should be three independent experiments conducted.
   2. Many experiments in this paper involve comparisons between WT and PA-X-deficient viruses. To support the conclusions regarding the effects of PA-X, it is important to demonstrate that infection rates are comparable between the two viruses, including with appropriate statistical analysis.
   3. Additional statistical tests should be added to several figures including for viral titre comparisons (Figure 1C), heat map analysis (Figure 2D and Figure 4B) and ELISA analysis of IFN beta (Figure S5).
   4. Figure 1 part C is listed to be a viral titre from a TCID50 assay but the units on the graph are PFU/ml. It should be TCID50/ml if it is a TCID50 assay but the materials and methods describe a plaque assay. Titres also seem quite high compared to what other literature appears to report. The figure legend mentions n=11 from 5 separate donors, is this from separate infections or were these technical replicates of the 5 donors?
   5. Figure S5-mislabelled graph- only has a part A in the figure but a part A and C are in the figure legend. Should IFN-X be IFN-beta? Perform statistical test to confirm no significant difference.
   6. Figure 3- could you clarify what the n= 5-11 independent experiments describes and how many independent experiments were performed for the data in Figure 3C.
   7. Figure S6 B- These cytokines are either below or above the limit of detection of the standard curve of the assay. For cytokines above the upper limit of detection, could the samples be appropriately diluted and rerun to obtain values within the standard curve range? For cytokines below the lower limit of detection, it is unclear whether these data provide meaningful quantitative information and, therefore, if they should be included. Additionally, the presentation of the data in Figure S6B should ideally be consistent with the graphs in Figure 3C, including the labelling and formatting of the axes.
   8. For Figure S7 (5′ RACE workflow), this appears to be important data supporting a potential mechanism by which PA-X inhibits the expression of antigen-processing and presentation genes. Therefore, it may be more appropriate for this experiment to be presented in the main body of the manuscript. However, the DNA gel images are not particularly clear, and there appear to be several nonspecific bands. In particular, Figure S7E is not very convincing. Additionally, it appears that only one independent experiment was performed but ideally there should be at least three independent biological replicates.
   9. In line 339-341 a downregulation of genes relating to antigen processing and presentation is described for both ciliated and club cells and reference is made to table 1 but in table 1 this term is only listed for the club cells not the ciliated cells.
   10. In line 343 to 344 it is stated that antigen processing and presentation is one of the top downregulated gene sets for both 1 and 3 DPI with a reference to Figure 4A but in the Figure of 4A the figure legend and image only mention the 1 DPI data. Figure 4B has analysis from both timepoints.
   11. In line 387-391 the authors state that they also examined another viral strain to determine whether the reduction in MHC I expression is conserved, reporting a trend towards reduced MHC I expression in Figure S8C and D. However, this effect appears modest, with no statistically significant difference reported, and only a subset of donors was included in the analysis. Therefore, the subsequent description of the reduction as "striking" appears somewhat inconsistent with the data presented in Figure S8 and may benefit from rewording to more accurately reflect the findings.
   12. Figure 4F/G appears to be the same data as Figure S8H- is this a duplication of the same data? Is this also only from one independent experiment?

   Significance

   This is an overall strong study that uses a relevant primary human model system and sophisticated technologies for analysis. The findings demonstrate that PA-X can modulate both the innate and adaptive immune responses to infection, providing important insights into the function of this protein, which is conserved across the majority of IAV strains.

   The overall limitations of the study are that some areas of the text require further clarification, and additional replicates and statistical analyses are needed to substantiate the conclusions. This study would be relevant to researchers of virology, innate immunity and the initiation of adaptive immunity.

   My field of expertise is in virology and innate immunity.

   Read the original source
3. 

   Review Commons

   Sep 25, 2026

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #2

   Evidence, reproducibility and clarity

   The authors analyze the function of the still more or less enigmatic influenza accessory protein PA-X in the context of ALI culture infection. They confirm its immune-modulatory role by performing single cell RNA seq experiments in these context relevant cultures. In detail, they found reduced IFN-lambda signalling and PA-X dependent decrease in certain infammatory cytokines when comparing WT to delta-PA-X influenza infections. More interestingly, they describe that PA-X transcrptionally dampens antigen-presentation pathways comprising MHCI, and furthermore it decreases cell surface MHCI levels.

   I have no major concerns to the data …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #2

   Evidence, reproducibility and clarity

   The authors analyze the function of the still more or less enigmatic influenza accessory protein PA-X in the context of ALI culture infection. They confirm its immune-modulatory role by performing single cell RNA seq experiments in these context relevant cultures. In detail, they found reduced IFN-lambda signalling and PA-X dependent decrease in certain infammatory cytokines when comparing WT to delta-PA-X influenza infections. More interestingly, they describe that PA-X transcrptionally dampens antigen-presentation pathways comprising MHCI, and furthermore it decreases cell surface MHCI levels.

   I have no major concerns to the data as it is. The experiments are performed, as far as I can judge, technically sound and the conclusions are supported by the data. I feel the conclusion related to disruption of MHCI-mediated antigen presentation is much stronger as compared to the relatively minor effects on Interferon-lambda and the other cytokines. It is understandable, that the authors do not see effects on virus replication and titers, when main effect of PA-X is on MHCI presentation and there are no IAV -specific CTLs in the system. Conversely, one would expect also ISG-induction in bystander cells, due to higher IFN-lambda secretion in the delta-PA-X condition and this would hinder viral spread at later time points. An experiment I would recommend.

   The strength of the study is the use of ALI cultures, thus a very relevant primary cell model, closely mimicking the in vivo situation. They use thorough and comprehensive scRNA-seq analysis corroborated by orthogonal approaches to verify their claims (cytokine measurements and flow cytometry for MHCI). Hence, from a descriptive point of view, the story is really strong and improves the current state of the art in this field. From physiological consequences and mechanisms it could be improved a lot and plenty additional experiments could be done. Some optional suggestions:

   • How are the effects on IFN-signalling when using a NS1, PA-X double KO as compared to a NS1-deficient IAV? Are both proteins acting complementary?
   • Can PA-X alone, for instance expressed from a lentivirus, downmodulate MHCI and alter IFN signalling and cytokine secretion?
   • Does PA-X interact with MHCI, how does it alter trafficking? Is MHCI retared in the ER, is it channeled to the lysosome etc?
   • If PA-X alters MHCI levels, this might activate NK-cell killing of infected cells, as they are triggered by low HLA-levels; so does PA-X also affecte NK-cell ligands?
   • Can they feed IAV-specific T cells to the system and will WT infected cells perform much better under such conditions, ie evade killing as compared to PA-X infections?

   All these suggestions, that likely will need some considerable amount of time, are not essential for the current conclusions of the manuscript, but would enhance its significance a lot. As it stands, I think it is more suited for specialist audience working in the context of Influenza biology and pathognenesis. The study neither brings up a new concept, nor does it describe mechanisms wich are not yet reported in the context of other viruses. Nevertheless, it is a sound descriptive study.

   No minor comments. The paper is clearly understandable and well written.

   Significance

   The strength of the study is the use of ALI cultures, thus a very relevant primary cell model, closely mimicking the in vivo situation. They use thorough and comprehensive scRNA-seq analysis corroborated by orthogonal approaches to verify their claims (cytokine measurements and flow cytometry for MHCI). Hence, from a descriptive point of view, the story is really strong and improves the current state of the art in this field.

   Read the original source
4. 

   Review Commons

   Sep 25, 2026

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #1

   Evidence, reproducibility and clarity

   In this paper by Setaro et al, the authors investigate the role of the IAV endoribonuclease PA-X in modulating immune responses to infection, in a model of primary human bronchial epithelial cultures. The data presented indicate that PA-X activity could influence both innate and adaptive immune responses by reducing both cytokine production and expression of MHCI molecules in infected cells.

   Previous papers (well referenced by the authors) have already described the effect of PA-X in curbing immune responses, in in vivo and in vitro models of infection. Nevertheless, the authors' use of primary human bronchial epithelial cultures …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #1

   Evidence, reproducibility and clarity

   In this paper by Setaro et al, the authors investigate the role of the IAV endoribonuclease PA-X in modulating immune responses to infection, in a model of primary human bronchial epithelial cultures. The data presented indicate that PA-X activity could influence both innate and adaptive immune responses by reducing both cytokine production and expression of MHCI molecules in infected cells.

   Previous papers (well referenced by the authors) have already described the effect of PA-X in curbing immune responses, in in vivo and in vitro models of infection. Nevertheless, the authors' use of primary human bronchial epithelial cultures and their analysis at the single cells level allowing to distinguish infected from bystander cells add important information to the matter.

   First, the authors showed that PA-X dampens IFN production (primarily IFN-λ) in all infected cells, independently of their subtype (ciliated or secretory).

   Second, they show that PA-X mediated reduction in IFN levels has a negative effect on ISG expression in uninfected bystander cells, at least at 1 dpi. This effect is not seen in infected cells, where the direct actions of other IAV proteins (e.g. NS1) could contribute to global suppression of gene and ISG expression, as discussed in the paper. Although this decrease in IFN signaling is not sufficient to affect influenza A viral replication in the closed experimental system of hBEC, it could still be relevant in vivo to influence inflammation, immune cell responses and lung injury.

   Third, they show that PA-X could also limit the production of pro-inflammatory and lung injury associated cytokine. They should take advantage of the single cell RNA data and analyse the data as done for IFNs, to define if the observed reduction is a direct effect of PA-X in infected cells only or an indirect effect of PA-X presence/absence on all cells. Do bystander cells produce IL6, IL1b, G-CSF, FGF, etc? Is this production reduced in wt vs dX infections, in both infected and bystander cells?

   The data presented confirm that IAV infected cells downregulate expression of HLA-A, HLA-B, HLA-C , at both the protein and RNA level and that this reduction is partially dependent on PA-X. Is this happening in all infected cell types? In Figure 4c, I noticed a small proportion of dX infected cells significanthly upregulating HLA-ABC expression. Are these a spillover of non infected cells? In Fig 4b at 3dpi, HLA-G is indicated as "up with PA-X" but the color code indicates the opposite. Please clarify and discuss the possible implication in the context of NK cytotoxic activity. The authors propose that PA-X directly targets for degradation several mRNA of genes involved in antigen processing and presentation, and they tested it in HEK 293T cells transfected with empty vector, wt or catalytically inactive PA-X. Could the authors confirms these data in hBEC, infected with WT or ∆X viruses?

   The authors recognize that functional effects of PA-X disruption of MHC I trafficking need to be tested in vivo, to understand how this disruption affects T cell clearance of infected cells during primary and ,most importantly, secondary infections. An experiment in this direction would be a great addition to the paper. For example, the authors could infect mice with a primary wt virus, and then test CD8 activity 1 month later during a secondary heterotypic infection with either a wt or dX virus.

   Overall, the data are well presented to allow reproducibility, well referenced and do not present obvious issues with number of replicates and statistical analysis. For the single cell data processing, I do not have enough expertise to properly evaluate the methods used.

   Significance

   Previous papers (well referenced by the authors) have already described the effect of PA-X in curbing immune responses, in in vivo and in vitro models of infection.

   Nevertheless, the authors' use of primary human bronchial epithelial cultures and their analysis at the single cells level allowing to distinguish infected from bystander cells add important information to the matter. The authors themselves recognize that functional effects of PA-X disruption of MHC I trafficking need to be tested in vivo, to understand how this disruption affects T cell clearance of infected cells during primary and ,most importantly, secondary infections. An experiment in this direction would be a great addition to the paper.

   The paper could be of interest for a broad immunologists/virologists audience of basic researchers.

   My field of expertise:

   Viral infection, innate immune responses to virus, airway epithelial biology and regeneration post injury

   Read the original source
5. 

   Version published to 10.64898/2026.01.30.702929 on bioRxiv

   Jan 31, 2026