SARS-CoV-2 inhibits induction of the MHC class I pathway by targeting the STAT1-IRF1-NLRC5 axis

The MHC class I-mediated antigen presentation pathway plays a critical role in antiviral immunity. Here we show that the MHC class I pathway is targeted by SARS-CoV-2. Analysis of the gene expression profile from COVID-19 patients as well as SARS-CoV-2 infected epithelial cell lines reveals that the induction of the MHC class I pathway is inhibited by SARS-CoV-2 infection. We show that NLRC5, an MHC class I transactivator, is suppressed both transcriptionally and functionally by the SARS-CoV-2 ORF6 protein, providing a mechanistic link. SARS-CoV-2 ORF6 hampers type II interferon-mediated STAT1 signaling, resulting in diminished upregulation of NLRC5 and IRF1 gene expression. Moreover, SARS-CoV-2 ORF6 inhibits NLRC5 function via blocking karyopherin complex-dependent nuclear import of NLRC5. Collectively, our study uncovers an immune evasion mechanism of SARS-CoV-2 that targets the function of key MHC class I transcriptional regulators, STAT1-IRF1-NLRC5.

3) It will be of interest that the authors were unable to replicate the pre-print observations regarding MHC-I downregulation by ORF8, however, the experiments have not been conducted or presented in such a way as to demonstrate that this is clearly the case. The authors only ever seem to test the expression of class I when ORF8 is co-transfected with NLRC5 which makes the results difficult to interpret. It is worth noting that the authors suggest the reason for the difference between this manuscript and the preprint by Zhang et al is that the effect may be cell type specific -but both groups use the same cell line -293T cells. 4)The authors repeatedly state without any reference that IFN-gamma (and not type I IFN) is the primary regulator of class I expression or that IFN-gamma is more potent. Please include the appropriate references for this statement in all cases.
Reviewer #2 (Remarks to the Author): The authors show that SARS-CoV-2 downregulates expression of MHC class I both in infected epithelial cells and COVID-19 patients. They report that the ORF6 protein suppresses IFNgammamediated STAT1 signaling, with a reduced expression of BLRC5 and IRF1, and also targets the nuclear import of NLRC5 mediated by the karyopherin complex.
The findings reported are of high relevance to understand immune evasion by SARS-CoV-2. The authors present a vast amount of experimental work to dissect how SARS-CoV-2 infection controls several pathways and to demonstrate the implication of ORF6. The experimental data support the conclusions.
Specific comments: Fig.2b. The authors indicate that 'induction of gene expression of the primary transcriptional regulators responsible for the MHC class I activation, such as NLRC5, IRF1 and STAT1 was also remarkably suppressed by SARS-CoV-2 infection'. But the results shown in Fig. 2b for STAT1 expression show no statistical significance in infected Calu-3 cells (Fig. 2b), nor a reduction of STAT1 is observed in Caco-2 or Huh7 cells (Suppl. Fig. 2). The authors do find downregulation of MHC class I protein expression by flow cytometry (Fig. 2c), but the transcriptional data on STAT1 does not correlate with a reduction in NLRC5 and IRF1 expression (Fig. 2b and Suppl. Fig. 2). To make these conclusions the authors compare the results in infected cells with the induction observed after Poly(I:C) treatment, a very strong activator of the IFN response. It is not clear whether this is the correct comparison, the control demonstrates that the cells can activate the IFN response, but IFN activation in the context of infection may not be that potent (even in the absence of viral inhibitors of these pathways). The ideal control would be the infection with a SARS-CoV-2 mutant lacking the ORF6 gene. The authors should clarify this.  . The results show that expression from the IRF1 promoter is also reduced, to a lesser extend compared to ORF6, in the presence of Nsp1 and ORF8 … the authors should comment on this in the Results section. A similar effect is seen with Nsp1 when testing the NLRC5 promoter activity. The Discussion should include alternative roles of Nsp1 and ORF8.
Discussion, page 15. It is interesting that impaired expression/function of NLRC5 may influence the levels of MHC class I and affect the CD8 response in cancer patients. This means that the immune response to SARS-CoV-2 and the development of COVID19 disease may also depend on the genetic variability of different patients leading to differential expression of NLRC5. The authors may comment on this interesting possibility in the Discussion.
There are some errors with the format of the manuscript. Pages and lines are not numbered and this makes difficult the writing of the report. Some parts of Figures 3 and 6 are missing.
Reviewer #3 (Remarks to the Author): This study by Yoo et al sets out to define the mechanism by which SARS-CoV-2 can inhibit the MHC I pathway to provide a possible explanation of why anti-viral CD8+ T cell responses in COVID-19 patients are impaired. RNA-Seq analysis of samples from COVID-19 patients as well as infected primary lung epithelial cells were performed to establish that MHC I gene expression is reduced and associated with decreased NLRC5 expression. This result was also observed using the epithelial cell line, Calu-3, where upon infection led to reduction of STAT-1 and IRF-1 expression. The authors then turn to the use of transfection system using HEK293T and HeLa cells to comprehensively define the mechanisms associated with these outcomes. Through this, they identified the ORF6 viral protein as the key player in inhibiting STAT-1 activity, IFN-g induced outcomes resulting in reduced MHC I expression at the gene and protein level, and confirmed the role of NLRC5. Additionally, experiments also defined how ORF6 inhibits NLRC5 function by blocking karyopherin complex-mediated nuclear import. Altogether, these findings provide an insight into a possible mechanism by which SARS-CoV-2 can evade host detection. This is a comprehensive, well written and presented study.
The role of NLRPC5 as a key regulator of the MHC I pathway in cancer and other viral infections is well studied and its mechanisms of action in this arena have been well defined. Thus, its role during SARS-CoV-2 infection is perhaps not completely unexpected. Nonetheless, this work serves to provide first evidence of its possible involvement in this setting, particularly in infected human samples and notably epithelial cells. Key components of the pathway that are affected are unravelled and the role of ORF6 is clearly implicated, albeit this is studied in non-respiratory continuous cell lines. While the reviewer understands that these systems lend themselves well for such mechanistic studies and are also more accessible, confirming some of the later major findings, especially those relating to ORF6 mediated NLRPC5 function and karyopherin-mediated nuclear import, in respiratory epithelial cells are encouraged. These data would strengthen the conclusions of the manuscript.
Other comments: Fig. 1a Data on MHC-I downregulation in SARS-CoV-2 patients in comparison to healthy controls should be also verified at the protein level by flow cytometry.
Supplementary Figure 1b. UMAP analysis of cells from normal lung tissue showed MHC I and CD45 gene expression associated with different cell populations. Given that correlation data for NLRC5 expression is available (in Supp Fig 1a), please also show the distribution of NLRC5 expression within the epithelial cell population. . The authors posit that this also occurs in Calu-3 cells, however, it appears that upregulation by immunostimulants is very minimal thus in comparison with infection, is not as convincing. Data should be plotted to show statistically significant results. Although these expression levels may be intrinsic to the cell type, another form of measure to demonstrate clearer differences, including replicates, would be useful to prove that the effects observed at the gene expression level are also observed at the protein level (i.e. by Western blot).
Supplementary Figure 2 title: States that it is "epithelial cells" but the results are using Caco-2 and Huh7 which are colon and liver cells.  In HEK293T cells, IFN-g-mediated STAT gene expression is inhibited by ORF expression. This is only shown for ORF6 and results of how the other viral proteins affect STAT-1 gene expression would add weight to the conclusion that that they do not have an effect on STAT signalling (in Fig 3c and d). The x-axis labels in my version are also missing and the bottom of Fig 3f is cut-off .   Fig 4a and b. In HEK293T cells, IFN-g induced IRF1 GAS and NLRC5 promoter activity is reduced by overexpression of ORF6. Graphs for depicting promoter activity (%) do not appear to show error bars, suggesting this is a single replicate. If the results are from 3 independently conducted experiments, individual replicates should be included to present the range of variation for each group.  In HeLa cells, NLRC5 stably expressed cells were treated with Leptomycin B where ORF6 expression showed reduced NLRC5 in the nucleus. My understanding is that Leptomycin B inhibits the export of proteins from the nucleus into the cytoplasm and yet in this experiment, what is measured in the translocation of NLRC5 into the nucleus. Can the authors please clarify the mechanisms/process of how Leptomycin B is affecting the presence of NLRC5 in the nucleus? 1

Point-by-point responses
Reviewer #1 (Remarks to the Author): Q: While for the most part, the work appears to have been competently conducted, I find some issues which I would very much hope the authors will consider before their next submission. These mainly relate to the attempts by the authors to imply that SARS-CoV-2 is specifically targeting the MHC-I pathway rather than IFN signalling overall, especially in the first figure.
A: We thank this reviewer for his/her appreciation that the study has been competently conducted. While it may have appeared so in the original version of the manuscript, we are not attempting to imply that SARS-CoV-2 specifically targets the MHC-I pathway rather than IFN signaling. Indeed, our data in this manuscript and data from other groups show that SARS-CoV-2 has successfully evolved to target BOTH IFN signaling as well as the MHC-I pathway. In order to avoid misunderstanding, we repeatedly emphasize this notion in the revised manuscript. (page 5 line 104, page 8 line 184 (section for Fig. 3), and page 9 line 216 (section for Fig. 4)).
Q: The analysis in figure one of existing datasets is extremely selective and could be described as misleading. I will outline my problems with these datasets but overall my advice would be that the authors remove this figure, and the associated claims in future submissions.
A: We thank this reviewer for pointing out the potential risk of misleading the reader of Figure 1. We have reanalyzed the data based on this reviewer's suggestion as illustrated below. The newly created Figure 1 is more convincing with much less risk of misleading the reader. While we think the new Figure 1 is appropriate to present in this manuscript, if this and other reviewers and/or the editors think it is preferable to remove the revised Figure 1, we will comply with this request. A: We agree that the level of CD45 between the healthy donors and the patient group should be comparable in order to analyze the differences in the gene expression between the two groups. To minimize the potential influence of high immune cell influx infection. With additional panels, the data have become more solid. We appreciate this reviewer's suggestion.
Q: It will be of interest that the authors were unable to replicate the pre-print observations regarding MHC-I downregulation by ORF8, however, the experiments have not been conducted or presented in such a way as to demonstrate that this is clearly the case.
The authors only ever seem to test the expression of class I when ORF8 is co-transfected with NLRC5 which makes the results difficult to interpret. It is worth noting that the authors suggest the reason for the difference between this manuscript and the preprint by Zhang et al is that the effect may be cell type specific -but both groups use the same cell line -293T cells.
A: We appreciate this reviewer's pointing out that our experiments regarding to the MHC class I downregulation by ORF8 was not sufficient in order to replicate the observation in the pre-print paper. We performed the experiments in which effects of

Please include the appropriate references for this statement in all cases.
A: We thank this reviewer for pointing out this deficiency. The references have The experimental data support the conclusions.
A: We thank this reviewer for his/her appreciation of the high relevance of the study, the vast amount of experimental work, and the experimental data to support the conclusion.

Q: Fig.2b. The authors indicate that 'induction of gene expression of the primary transcriptional regulators responsible for the MHC class I activation, such as NLRC5, IRF1 and STAT1 was also remarkably suppressed by SARS-CoV-2 infection'. But the results shown in Fig. 2b for STAT1 expression show no statistical significance in infected
Calu-3 cells (Fig. 2b), nor a reduction of STAT1 is observed in Caco-2 or Huh7 cells (Suppl. Fig. 2).
A: We apologize that our description of the data was not accurate. These transcriptional regulators were not induced upon SARS-CoV-2 infection in Calu-3 and Huh-7 cells. In the Caco-2 cell, the expression of STAT1 and NLRC5 was induced, although at a very modest level. We modified the text in the Results section (page 7, line 166-168) to reflect these corrections. (Fig. 2c), but the transcriptional data on STAT1 does not correlate with a reduction in NLRC5 and IRF1 expression ( Fig. 2b and Suppl. Fig. 2).

Q: The authors do find downregulation of MHC class I protein expression by flow cytometry
A: This reviewer correctly pointed out that the transcriptional data on STAT1 does not correlate with a reduction in NLRC5 and IRF1 expression. Although the exact mechanism of the discrepancy of the expression of STAT1 and NLRC5/IRF1 is not clear, it might be possible that inhibition of the function of STAT1 by ORF6 might play a role.
Although the STAT1 expression level was not down-regulated by SARS-CoV-2 infection, the function of STAT1 (judging from STAT1 nuclear translocation) is blocked by SARS-7 CoV-2 ORF6 (Fig. 3d). This may result in the transcriptional suppression of IRF1 and NLRC5 gene expression, both of which play a role in the transcription of components in the MHC class I pathway. infection. It has been reported that 'SARS-CoV-2 induces substantial but delayed IFN-β production' (Lei et al., Nature Communications, 2020, 11, 3810, doi:10.1038/s41467-020-17665-9 and Yin et al., Cell Reports, 2021 A: We agree that we should comment on the effect of Nsp1 and ORF8 on the MHC class I pathway. We observed marginal reduction of IFNg-mediated IRF1 and NLRC5 promoter activity by Nsp1 and ORF8. However, only ORF8 showed an inhibitory effect on NLRC5 expression upon IFNg stimulation. This effect was much milder compared to ORF6 (Fig. 4abc) A: We apologize for any inconvenience caused by this deficiency. It seems that formatting errors were created during conversion of the files to PDF format. We have corrected these errors.
Reviewer #3 (Remarks to the Author): Q: Altogether, these findings provide an insight into a possible mechanism by which SARS-CoV-2 can evade host detection. This is a comprehensive, well written and presented study.
A: We thank this reviewer for his/her appreciation of the study for its comprehensiveness.
Q: Key components of the pathway that are affected are unravelled and the role of ORF6 is clearly implicated, albeit this is studied in non-respiratory continuous cell lines. While the reviewer understands that these systems lend themselves well for such mechanistic studies and are also more accessible, confirming some of the later major findings, especially those relating to ORF6 mediated NLRPC5 function and karyopherin-mediated nuclear import, in respiratory epithelial cells are encouraged. These data would strengthen the conclusions of the manuscript.
A: We thank this reviewer for the suggestion to use more relevant cells for the study. In the assay for karyopherin-mediated nuclear localization of NLRC5, we used additional cell lines including A549 (lung epithelial cell line), Calu-3 (lung epithelial cell line) and Caco-2 (intestinal epithelial cell line) (Supplementary Fig. 9). Also, we confirmed ORF6 inhibition of NLRC5 CITA function by reporter assay using the Calu-3 cell line (Supplementary Fig. 7). The Results section has been modified accordingly (page 11 line 261-262 and page 12 line 291-292). New data revealed that ORF6 inhibits NLRC5 nuclear localization also in lung epithelial cells. We again thank the reviewer for the excellent suggestion.  Supp Fig 1a), please also show the distribution of NLRC5 expression within the epithelial cell population.
A: We thank for this reviewer's suggestion. First, just in case there was a misunderstanding, the data in Supplementary Fig. 1a and 1b

Q: Fig 3e. In HEK293T cells, IFN-g-mediated STAT gene expression is inhibited by ORF expression. This is only shown for ORF6 and results of how the other viral proteins affect
STAT-1 gene expression would add weight to the conclusion that that they do not have an effect on STAT signalling (in Fig 3c and d).
A: This is an excellent point. We performed the experiments for STAT-1 gene expression in HEK293T cells expressing other SARS-CoV-2 genes. Please find Supplementary Fig. 6. Other SARS-CoV-2 genes, including Nsp1, Nsp15, ORF8 or N did not suppress the expression of STAT-1 at both steady-state and after IFNg stimulation. These serve as excellent controls for the function of ORF6 and we would like to thank this reviewer for suggesting this revision.

Q: The x-axis labels in my version are also missing and the bottom of Fig 3f is cut-off.
A: We apologize for any inconvenience caused by this deficiency. It seems that formatting errors were created during file conversion to a PDF format. We have corrected these errors.
13 Q : Fig 4a and  A: We appreciate this suggestion and pointing out our shortcomings. In the revised Fig. 4a and b, we have analyzed data from multiple repeats and the error bar and P-values have been appended.

Can the authors please clarify the mechanisms/process of how Leptomycin B is affecting the presence of NLRC5 in the nucleus?
A: Leptomycin B, as this reviewer correctly notes, inhibits CRM1-dependnet exportation of proteins from the nucleus to the cytoplasm.
Once translated, NLRC5 protein is dynamically shuttling between the nucleus and cytoplasm, and under the steady state conditions, the majority of NLRC5 protein is observed in the cytoplasm. In our experimental conditions, we monitored whether SARS-CoV-2 ORF6 inhibited nuclear translocation of NLRC5, and we used Leptomycin B to capture NLRC5 proteins that were translocated to the nucleus. Once NLRC5 translocated to the nucleus, Leptomycin B blocked the nuclear export of NLRC5.

REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): While the authors agree with the points raised in the previous round of review, the paper itself is not substantially different and retains many of the previous problems without bringing substantial novelty and I cannot recommend it for publication.
1) As a general point, the authors seem to have decided that when comparing different datasets or their own results, "significantly down-regulated' and 'not changing' can be taken to be equivalentboth being 'inhibition of up-regulation'.
This has a certain logic -i.e. assume that all virus infections will induce class-I expression unless inhibited -but this also glosses over some significant issues with consistency between the data and analysis presented here. MHC-I not changing at all is repeatedly used to corroborate data where MHC-I is strongly down-regulated -with no attempt to explain why the data should be discordant.
It also leads to some fairly odd statements like: " Supplementary Fig. 4. SARS-CoV-2 inhibits surface expression of the MHC class I proteins" -to title a figure where SARS-CoV-2 does not, in fact, change MHC-I protein expression on the cell surface.
2) The first example of this is in figure 1, which retains the most significant problems of the first round of review (despite the authors claims to the contrary). In the RNA-seq analysis, the authors claim "up to 66% reduction in mean expression of MHC class I genes". The proposed mechanism for this reduction is cell-autonomous -i.e. SARS-CoV-2 infected cells reduce class-I RNA because of ORF6 blocking NLRC5.
If the cells were a uniform population this would mean that if SARS-CoV-2 infection completely reduced MHC-I RNA levels to 0, on average, 66% of cells would have to be infected in the samples. But the swab contains a mix of cells -probably 66% aren't even infectable with SARS-CoV-2, let alone infected. And as the authors will know, any the lymphocytes present contribute a disproportionately large amount of class I.
It's simply inconceivable that sufficient numbers of cells could be infected to cause this downregulation. If they wanted to confirm this they would have to look at MHC-I expression in virally infected cells -not by correlation.
I don't know what the source of this apparent down-regulationis -it cannot be related to ORF6 blocking NLRC5 in infected cells -nor can it ever provide useful evidence for this issue.
This down-regulation in the the authors' re-analysis of the patient RNA-seq data is 'supported' by the RNA-seq and scRNA data from in vitro experiments with air liquid interface cells, which shows that there is *no change* in class I expression. This is correct -but this is not the same as a significant down-regulation! Again, this is as stated e.g. by Blanco-Melo et al that there is not a substantial IFN response in those cells, not a novel observation.
3) I assume that the reason the authors are keen to find a precedent in figure 1 for class I downregulation is that their own qPCR data shows a down-regulation of these genes. I am not aware of this being found in any of the many available RNA-seq and scRNA datasets and as the authors show in figure 1-other groups carrying out infections in the best possible model (air-liquid interface HBECS) find that there is no significant change, or even a slight up-regulation.
Furthermore, this down-regulation is not corroborated by their own flow cytometry data. -this is all somewhat strange. There is a strong down-regulation of class I genes in Calu3 but no effect at the cell surface -the authors claim that Calu3 express so little Class I that they have to measure on a linear scale (which doesn't seem correct and is somewhat concerning). It certainly does not match our experience (where MHC I is readily measurable on Calu3 with w6/32) or the relative expression levels of HLA-A or HLA-B at the RNA level between e.g Calu3 and Caco2 on public databases e.g. at https://portals.broadinstitute.org/ccle The down-regulation in Calu3 at the RNA level appears far more potent than in the other cell lines tested, where changes are more modest. In fact, as statistics used are generally not appropriate, I suspect many changes would not be significant if the correct test was used. See point 4.
None of the cells show a down-regulation of MHC-I at the cell surface. In fact, this might be consistent -e.g. very little change in Caco2 at the RNA level and no change at the protein level. But instead figure 2 presents a dramatic change at the RNA level in Calu cells and no change in Caco cells by flow cytometry -as if this was the same outcome. It is not and results in less confidence in both assays, not more. Again, the authors do not attempt to explain these discrepancies, only call it all 'inhibition of upregulation'.
In the end, ORF6, the supposed driver for this phenotype, has no effect on MHC-I expression at steady state levels by qPCR or flow cytometry, so the downregulation at the RNA level in infection is not explained.

4)
Student T-tests are repeatedly used for analysis with multiple groups. This is not appropriate.
Reviewer #2 (Remarks to the Author): The authors have addressed my concerns. I have no further comments.
Reviewer #3 (Remarks to the Author): the authors addressed my concern and greatly improved the manuscript.

Point-by-point responses
Reviewer #1 (Remarks to the Author): Q: While the authors agree with the points raised in the previous round of review, the paper itself is not substantially different and retains many of the previous problems without bringing substantial novelty and I cannot recommend it for publication. distinct. In addition, this manuscript provides molecular mechanisms of impaired upregulation of MHC class I during SARS-CoV-2 infection. Therefore, our manuscript describes novel findings, which are not predictable from the previous literature.
As this reviewer pointed out correctly below, the published RNAseq data have not illustrated impaired upregulation of MHC class I, perhaps because the focus of those studies was on other subjects. Although Figure 1 in our manuscript is our starting point for finding the failed upregulation of MHC class I in SARS-CoV-2 infected airway cells, we would like to point out that a detailed analysis of MHC class I regulation has not been documented.
Q: 1) As a general point, the authors seem to have decided that when comparing different datasets or their own results, "significantly down-regulated' and 'not changing' can be taken to be equivalent -both being 'inhibition of up-regulation'.
This has a certain logic -i.e. assume that all virus infections will induce class-I expression unless inhibited -but this also glosses over some significant issues with consistency between the data and analysis presented here. MHC-I not changing at all is repeatedly used to corroborate data where MHC-I is strongly down-regulated -with no attempt to explain why the data should be discordant.
A: We agree that "significantly down-regulated" and "not changing" are not completely equivalent. As this reviewer pointed out, if the virus infection leads to MHC class I upregulation, which has been documented in many virus strains, both of these two cases suggest that there is 'inhibition of upregulation' of the MHC class I pathway by the virus. We apologize that there were shortcomings in describing the data which appeared not to indicate the same outcomes. We have modified the text to describe the following figures (  differently regulated, and their functions are distinct, it is hard to conclude that the study in this manuscript is predictable from Blanco-Melo's paper. Furthermore, we have clarified the molecular mechanism of the ORF6 inhibition on CITA function of NLRC5, which has not been shown before. Thus, we believe that the study in this manuscript combines novelty with a comprehensive analysis. Q: 3) I assume that the reason the authors are keen to find a precedent in figure 1 for class I down-regulation is that their own qPCR data shows a down-regulation of these genes. Q: Furthermore, this down-regulation is not corroborated by their own flow cytometry data.this is all somewhat strange. There is a strong down-regulation of class I genes in Calu3 but no effect at the cell surface -the authors claim that Calu3 express so little Class I that they have to measure on a linear scale (which doesn't seem correct and is somewhat concerning). (equivalent to 'no infection') similar to our result, thereby corroborating our observation.
Therefore, these observations indicate that regardless of the quality of the antibody or the inconsistent cell line status or conditions, the MHC class I is not upregulated in SARS-Cov-2 infected cells.
We cannot comment on the expression level of MHC class I in Calu3 cells observed by this reviewer or in the website that this reviewer mentioned (Cancer Cell Line Encyclopedia at the Broad Institute, https://portals.broadinstitute.org/ccle), both of which are not currently accessible. Since there might be differences in sub-cell lines or culture conditions, we need to compare the data side by side. However, again, it was observed unanimously that MHC class I expression was NOT induced upon SARS-CoV-2 infection in all cell lines that we analyzed, regardless of the level of MHC l class I expression. We thank this reviewer for his/her careful examination of this point.
Q: The down-regulation in Calu3 at the RNA level appears far more potent than in the other cell lines tested, where changes are more modest. In fact, as statistics used are generally not appropriate, I suspect many changes would not be significant if the correct test was used.

See point 4.
A: We thank this reviewer for pointing out if the correct test was used in order to analyze for the statistical significance. We agree that this is important. As we summarized in the answer for point 4 below, we confirmed that correct methods were used for all the statistical testing.
Q: None of the cells show a down-regulation of MHC-I at the cell surface. In fact, this might be consistent -e.g. very little change in Caco2 at the RNA level and no change at the protein level. But instead figure 2 presents a dramatic change at the RNA level in Calu cells and no change in Caco cells by flow cytometry -as if this was the same outcome. It is not and results in less confidence in both assays, not more. Again, the authors do not attempt to explain these discrepancies, only call it all 'inhibition of upregulation'.
A: We agree that although all three cell lines showed a reduced mRNA expression by SARS-CoV-2 infection, downregulation of MHC class I on the cell surface was not observed in all cell lines. As we mentioned above, cell surface expression level and mRNA level may not always correlate because cell surface expression is regulated not only by the transcription level. Since the turnover rate of HLA proteins is relatively long, it is more likely that the regulation of the HLA protein half-life may be involved in the discrepancy between mRNA level and protein level. Again, it is important to note that upregulation of surface expression of MHC class I was NOT observed in any cell lines that we used for the entire study. Therefore, these data unanimously suggest that there is a mechanism to suppress the upregulation of the MHC class I pathway at the mRNA and cell surface level. As requested by this reviewer, we modified the text to describe the difference among cell types (page 8, lines 179-187). We thank this reviewer for the careful data interpretation.
Q: In the end, ORF6, the supposed driver for this phenotype, has no effect on MHC-I expression at steady state levels by qPCR or flow cytometry, so the downregulation at the RNA level in infection is not explained.
A: We agree that ORF6 did not reduce the expression of MHC class I under the steady-state condition. As we explained above, ORF6 suppressed the function of three key molecules, STAT1, IRF1, and NLRC5 all of which are critical for the upregulation of MHC class I upon IFN treatment and under the inflammatory conditions. It is likely that besides ORF6, there might be alternative mechanisms to suppress the MHC class I expression at steady state. We incorporated a consideration of this possibility in the revised Discussion section (page 19, lines 478-485).
Q: 4) Student T-tests are repeatedly used for analysis with multiple groups. This is not appropriate.
A: We thank this reviewer to point out the methods of statistical testing. We are well aware of this issue that is crucial for making a scientifically sound conclusion.
As this reviewer correctly pointed out, Student T-test is NOT appropriate IF values are analyzed comparing among multiple groups. It is appropriate IF an 'unpaired t-test' is used in order to compare the mean value between two independent groups with equal variance.
We used an unpaired t-test for the statistical analysis for the comparison between 'two independent groups' (e.g. 'mock' vs 'infected/stimulated' or 'empty control' vs 'viral gene expressed') as shown in Figures 3,4,5,6,and 7,and Supplementary Figures 3,5,6,7,8,and 9. Therefore, we believe that the student t-test is an appropriate method to measure the statistics for these figures. Please note that the student T-test was not used for comparing the multiple groups.
However, in the case of Figure 1, since the samples comprise randomly selected values, we used the Mann-Whitney U test or Wald-test adjusted by Benjamini-Hochberg to reduce the false discovery rate.

Reviewer #2 (Remarks to the Author):
Q: The authors have addressed my concerns. I have no further comments.
A: We thank this reviewer again for acknowledging the novelty and significance of the study and providing constructive criticisms which helped to significantly improve our manuscript.