SARS-CoV-2-Induced Type I Interferon Signaling Dysregulation in Olfactory Networks Implications for Alzheimer’s Disease

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

General comments:

The interferon signaling (IFN-1), COVID-19, and Alzheimer’s are interesting and proposed hypothesis and/or the model is well though, but for being an Insilco work, the works has a separate and inherent gap in the presentation and are recognized as in below.

Title: The title needs a thorough change as it not reflecting the work it presented.

In Introduction Page -2 : “ The attractiveness of type I interferon” is not a good phrase. Change it.

Section 2.2 : For differentially expressed genes.: It is hard to gauge the data with data sets with having a differential data set. Most of the data in the GEO have the expansion raw files which can well be used for the actual differential expression studies and then compare. That will be a  real comparison and differentiate. Why it was not done, and explanations must be given, as a shortcoming. They should also consider the GEO data sets to search first and the annotated files and the linked references.

 

Results: 3.1 This reviewer needs real pathways build with a most updated software.

 

Figure -1: Only 14 are in common, looks something else. Did they try to build the pathways with these 4 only and compare with those independent ones? Need more depreciative notes besides its technical part that the figures can stand alone with the explanation of the data.

 

Figure -2: The networks of  the 14 gene connection doesn't make any sense unless the Pathways are depicted. also need a line legend clearly, A low quality explanations.

 

3.3: What is the significance of the target risk score. Where is the validation data for the analyzed as well as the captured referred detests?

 

4.2: No doubt a new concept emerging. But intricately the Inflammatory responses to Tau and Amyloid beta pathogenesis is required along the validation, which is not here. Which is more to come.

 

In Conclusion: Last two lines: Too optimistic and should have low keyed because of any experimental data by the authors and the clinical testing.

 

Technical Notes: At some point Supplementary files are mentioned but at the end, it said Supplementary file none. Not also witnessed the same.

 

Comments on the Quality of English Language

Jut need a few editing and notational and expressional changes.

Author Response

The interferon signaling (IFN-1), COVID-19, and Alzheimer’s are interesting and proposed hypothesis and/or the model is well though, but for being an Insilco work, the works has a separate and inherent gap in the presentation and are recognized as in below.

1. Title: The title needs a thorough change as it not reflecting the work it presented.
2. In Introduction Page -2 : “ The attractiveness of type I interferon” is not a good phrase. Change it.
3. Section 2.2 : For differentially expressed genes.: It is hard to gauge the data with data sets with having a differential data set. Most of the data in the GEO have the expansion raw files which can well be used for the actual differential expression studies and then compare. That will be a  real comparison and differentiate. Why it was not done, and explanations must be given, as a shortcoming. They should also consider the GEO data sets to search first and the annotated files and the linked references.

 

4. Results: 3.1 This reviewer needs real pathways build with a most updated software.

 

5. Figure -1: Only 14 are in common, looks something else. Did they try to build the pathways with these 4 only and compare with those independent ones? Need more depreciative notes besides its technical part that the figures can stand alone with the explanation of the data.

 

6. Figure -2: The networks of  the 14 gene connection doesn't make any sense unless the Pathways are depicted. also need a line legend clearly, A low quality explanations.

 

7. 3: What is the significance of the target risk score. Where is the validation data for the analyzed as well as the captured referred detests?

 

8. 2: No doubt a new concept emerging. But intricately the Inflammatory responses to Tau and Amyloid beta pathogenesis is required along the validation, which is not here. Which is more to come.

 

9. In Conclusion: Last two lines: Too optimistic and should have low keyed because of any experimental data by the authors and the clinical testing.

 

10. Technical Notes: At some point Supplementary files are mentioned but at the end, it said Supplementary file none. Not also witnessed the same

 

Thank you for your comments. A point-by-point answer follows.

1. The title has been amended accordingly, to reflect the paper’s workflow. Thank you.
2. Amended, thank you.
3. We thank the reviewer for their comment. We agree with the principles outlined in the reviewer’s comment. For the comparisons and concept considered in our hypothesis however, there is currently one appropriate published study in PubMed, whose data on amygdala and OB gene expression of COVID-19 patients are available by its supplementary files. Therefore, while we acknowledge and agree with these principles on data usage, we are similarly restricted by the current paucity of reposited data on the subject of our study. GEO does not contain data appropriate for our study design, and please not that not every GEO dataset is necessarily curated. Thank you.

It should be noted that supplementary material provided by peer-reviewed publications is a valid source of data, and its intended use that of secondary analysis (e.g., see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6157527/ ). As a case in point, the Ma’ayan laboratory, developing Enrichr, is also developing tools to leverage supplementary gene expression data from PubMed Central literature (see https://www.biorxiv.org/content/10.1101/2023.10.03.560783v1 ). Thank you.

4. We are unsure of the reviewer’s meaning in general, and the specific meaning of “real” pathways. The supplementary data provided in this paper can fully replicate our findings and the analyses by any software currently available. Furthermore, the specific usage of IFN-I pathways is not an object of scrutiny in our study; we consider it the expected biological response to infection and immune stimulation in general, based on the literature on COVID-19. Thank you.
5. Thank you for your comment. The 14 genes at the central intersection are those that are common between all datasets. Those in other intersections represent other combinations that are potentially interesting, e.g., all amygdala IFN-I genes fully overlap with the AD gene set. Note that the Venn diagram has been redrawn to show comparisons between genes from the “Interferon a/b signalling” pathway between all datasets.

Please also note that among our findings, the AD gene signature overlaps with at least one other COVID-19 CNS tissue. We have focused on the 14 genes common between every IFN-I signature however, as per our original aim.

6. Thank you for your comment. The Figure is directly reported from STRING. Following the reviewer’s suggestion, we have provided the node scores in Supplementary File 5, and we had specified that this putative PPI is statistically significant in-text. Pathway enrichment is provided in Table 1 and raw files in Supplementary File 6.
7. Thank you for your comment. We have added the significance of the Target Risk Score as an aggregate of scores, and define its significance as provided by Agora, i.e., a liner relationship with risk. We furthermore specify that the genes we report on belong to increased risk as reported by Agora, provide the raw files in Supplementary File 7, and clarify that we report on genes differentially expressed in at least one CNS region in the Agora datasets. The validation data for Agora are available via previous publications and directly on site; we do not provide them.
8. Thank you for your comment. We are unsure of the meaning conferred. The AD signature we chose is a IFN-I signature that is significantly enriched in a comparison between Aβ and Tau pathology. Furthermore, we refer to published data and studies on the convergence of Αβ and tau in cGAS-STING, a pathway directly involved with interferon signalling and nucleic acid surveillance – two processes that are also identified in our analyses. We have expanded the discussion to outline them.
9. Thank you for your comment. We toned them down significantly.
10. Thank you for your comment. Supplementary files have been restructured and reuploaded.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have performed an Agora search to find shared genes between SARS-Cov-2 and Alzheimer's Disease risk. This is an interesting research question of broad interest.

While the title has been generalized, it does not reflect the specificity of the current study, which focused on Alzheimer's associations. The title should be amended.

The data set selection was not comprehensively worded. The authors should follow Cochrane or Prospero protocols for listing selection of studies and include a PRISMA diagram that shows where studies were excluded at each step and why.  If possible, a forest plot with potential for data set bias should also be included.

A description of the data attributes was not given.  A careful summary of the input data along with key attribute descriptions and sample sizes is needed within the main paper. The authors state raw data is given in the Supplement but no Supplement file was provided. This will need to be checked/updated.

The quantitative results are not presented in a reproducible manner. Figure 2 uses line thickness to illustrate relationships. However, these should also be quantified in tabular format and post-hoc statistical comparison performed where possible. Some of this is given in Table 1 but needs much better explaining and integration into the paper. 

The description and literature review of the nodes is not substantive enough, lacking in both breadth and depth.  Interconnecting hypotheses by the authors are also not described to provide more context.

 

 

Comments on the Quality of English Language

Minor editing

Author Response

The authors have performed an Agora search to find shared genes between SARS-Cov-2 and Alzheimer's Disease risk. This is an interesting research question of broad interest.

While the title has been generalized, it does not reflect the specificity of the current study, which focused on Alzheimer's associations. The title should be amended.

Thank you for your comment. It has been appropriately changed to reflect the focus on Alzheimer’s disease.

 

The data set selection was not comprehensively worded. The authors should follow Cochrane or Prospero protocols for listing selection of studies and include a PRISMA diagram that shows where studies were excluded at each step and why.  If possible, a forest plot with potential for data set bias should also be included.

Thank you for your comment. Please note that both Cochrane and Prospero protocols refer to systematic reviews and meta-analyses, which do not apply to our study design as it attempts neither, both due to the paucity of available data on amygdala (a singled dataset is currently available) and due to our elected study design which aims to (a) approximate a neuroanatomical pathway (b) compare two distinct diseases on the basis of shared perturbations in innate immunity. What our study attempts is a validation of a previously developed proof of concept, with all the restrictions and caveats that come with attempting this validation in silico.

Specifically, the relative scarcity of data on the hypothesis we explore, renders a data-driven meta-analysis not feasible. The amygdala and olfactory bulb dataset is currently unique in the literature. Similarly, the laser-captured dataset from Alzheimer’s disease represents a unique approach that has similarly generated unique data. Finally, the olfactory epithelia study utilizes a single cell approach and provides data on ciliated cells, contrary to any other available study.

Beyond that however, we designed a hypothesis driven, gene-overlap analysis between enriched interferon signatures from previously published datasets.

Following your suggestion we have added this rationale to the methods section, and as a limitation. We have furthermore added:

1. A concept design section to describe how this concept was formulated and the rationale behind dataset selection.
2. Search results for each query, as well as any potential reason to exclude studies in supplementary materials 1, as well as the gene lists retrieved from each study.

A description of the data attributes was not given.  A careful summary of the input data along with key attribute descriptions and sample sizes is needed within the main paper. The authors state raw data is given in the Supplement but no Supplement file was provided. This will need to be checked/updated.

Thank you for your comment. We acknowledge the suggestion, and have restructured the raw data and inputs.

The quantitative results are not presented in a reproducible manner. Figure 2 uses line thickness to illustrate relationships. However, these should also be quantified in tabular format and post-hoc statistical comparison performed where possible. Some of this is given in Table 1 but needs much better explaining and integration into the paper. 

Thank you for your comment. Following the reviewer’s suggestion we have (1) restructured supplementary files to provide data on a. studies screened for supplementary data, specific datasets used and specific gene signatures used. (2) Expanded the methods section to include rationale and the processes employed by Enrichr (3) For figure 2, we provided the quantitative data as generated by STRING in supplementary file 5.

The description and literature review of the nodes is not substantive enough, lacking in both breadth and depth.  Interconnecting hypotheses by the authors are also not described to provide more context.

Thank you for your comment. We did not attempt an explicit review of each gene due to space constraints and the potential redundancy with previously published works from our group.

Following the reviewer’s suggestion, we have added relevant sections in the discussion where appropriate, focusing on mechanisms however, rather than singlular genes.

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript by Vavougios et al. explores the relationship between SARS-CoV-2 infection and Alzheimer's disease or dementia/cognitive dysfunction, focusing on the IFN signaling pathway. Despite achieving interesting results, addressing some important issues is essential to ensure both the proper form and content for publication.

Major issues
Beyond PubMed, other gene-expression databases offer better avenues for obtaining these data.
It is unclear why the author chose these datasets from among all those retrieved in the search. The selection should be motivated and systematic.

The tools and conditions for gene enrichment, such as ORA and GSEA, need to be described. Additionally, a thorough analysis of the results is required. For instance, some sectors of the Venn diagram share the same or nearly identical items as those chosen for the results but are not explained. Furthermore, neither the relationship between the obtained genes nor their functions is described, beyond IFITM or OAS.

Additionally, the statistical analyses need to be described thoroughly and in detail.

Only one dataset is referenced in the manuscript, specifically in Section 3.1. Table 1 is not mentioned in the text. It is unclear which database these references pertain to. Additionally, consider removing Section 4.4 'Tables' since there is only one table.

Minor issues

The title should include the terms AD or dementia.

The keywords should also encompass dementia or cognitive decline in addition to AD.

Abbreviate Alzheimer's Disease as AD for the initial mention, and consistently use this acronym thereafter.

Are there any other works, aside from those by the authors, that explore this olfactory pathway? Please cite and discuss any similarities and differences found in the literature.

The supporting description for Agora is provided in Section 2.4 (inappropriate, remove it) and in the acknowledgments.
Figure 1 could be enhanced by indicating the datasets, perhaps using colors as illustrated in the figure caption.

Comments on the Quality of English Language

It is recommended that an English grammar and style expert review the manuscript to enhance its overall coherence, refine certain structures, and address repetitive expressions.

Author Response

Reviewer 3

The manuscript by Vavougios et al. explores the relationship between SARS-CoV-2 infection and Alzheimer's disease or dementia/cognitive dysfunction, focusing on the IFN signaling pathway. Despite achieving interesting results, addressing some important issues is essential to ensure both the proper form and content for publication.

Major issues

Beyond PubMed, other gene-expression databases offer better avenues for obtaining these data.It is unclear why the author chose these datasets from among all those retrieved in the search. The selection should be motivated and systematic.

We thank the reviewer for their comment. We agree with the principles outlined in the reviewer’s comment. For the comparisons and concept considered in our hypothesis however, there is currently one appropriate published study in PubMed, whose data on amygdala and OB gene expression of COVID-19 patients are available by its supplementary files. Therefore, while we acknowledge and agree with these principles on data usage, we are similarly restricted by the current paucity of reposited data on the subject of our study.

It should be noted however that supplementary material provided by peer-reviewed publications is a valid source of data, and its intended use that of secondary analysis (e.g., see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6157527/ ). As a case in point, the Ma’ayan laboratory, developing Enrichr, is also developing tools to leverage supplementary gene expression data from PubMed Central literature (see https://www.biorxiv.org/content/10.1101/2023.10.03.560783v1 ).

Following the reviewer’s suggestion, we have incorporated an extended description of our search for data appropriate for the comparisons, and present them in Supplementary Material 1. This includes, search data, studies included/excluded and reasons for exclusion, the supplementary file designation in the original study, and the gene signatures extracted for further analysis.

 

The tools and conditions for gene enrichment, such as ORA and GSEA, need to be described. Additionally, a thorough analysis of the results is required. For instance, some sectors of the Venn diagram share the same or nearly identical items as those chosen for the results but are not explained. Furthermore, neither the relationship between the obtained genes nor their functions is described, beyond IFITM or OAS. Additionally, the statistical analyses need to be described thoroughly and in detail.

Thank you for your suggestions. The Venn diagram has been redrawn to show comparisons between genes from the “Interferon a/b signalling” pathway between all datasets. The 14 genes at the central intersection are those that are common between all datasets. Those in other intersections represent other combinations that are potentially interesting, e.g., all amygdala IFN-I genes fully overlap with the AD gene set.

Following the reviewer’s suggestion, we have expanded the discussion to include information on HLAs, PSMB8 and XAF1. Please note however that due to space constraints and a specific hypothesis, we cannot expand more in the current paper.

We have expanded the methods sections to better describe the tools and the derived statistics from Enrichr.

Only one dataset is referenced in the manuscript, specifically in Section 3.1. Table 1 is not mentioned in the text. It is unclear which database these references pertain to. Additionally, consider removing Section 4.4 'Tables' since there is only one table.

Thank you for your suggestions. We have referenced the studies (Ziegler et al, Serrano et al – 2 datasets, Das et al) and included them in the reworked supplementary files. Table 1 has been linked to the text and section 4.4. Removed as suggested.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thanks to the authors for making substantial changes as request. This has now been much improved.  

Author Response

We thank the reviewer for their appraisal of our work.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have made substantial improvements to the clarity and framing of their study.  Most major issues have been rectified. Only two minor concerns remain.

While it is clear this is not a meta-analysis, the study does rely on identification of data from literature to be used in secondary analysis. As such, rigor in how studies were evaluated for inclusion and exclusion is critical. While the authors were pretty clear why Das study was chosen, it was less clear why other identified studies returned from the search(es) were explicitly excluded.  The addition to the Supplement is noted and certainly helpful. However, a few additional explanatory sentences on the handful of identified potential studies that were ultimately excluded would improve rigor in the main manuscript.  Perhaps a table or bulleted list with a reason for each study?

The present study's results are interesting and important. Overall, the authors do a good job in supporting their findings.  However, one minor edit that remains is insuring the Limitations mentions the number of original study datasets and corresponding sample sizes included in the present study's analysis. The lack of data source diversity is a limitation that must be considered when assessing how variance and/or other unknown biases in data selection may impact reproducibility and the framing of future work.

Author Response

We thank the reviewer for their comments and acumen. 

Following the reviewers suggestion, we have added a numbered list in text describing exclusion criterions on study design and gene expression data. In the supplement, beside each study, a comment describing exclusion reason is available.

We concur on the limitations and the selection bias introduced by the limited number of studies available. We have added the following statement: 

A major limitation to be considered when interpreting our analysis is the current data source diversity. Specifically, there is a single study on gene expression data from the amygdala and olfactory bulb combined (Supplementary Materials 1; n=36; 18 controls vs. 18 COVID-19 patients donating amygdala tissue, specifically[37]). Likewise, detailed data on ciliated cells infected by COVID-19 fitting the criteria of our study were available from one out of 40 studies (Supplementary Materials 1; n=58[29]).The lack of other studies providing data on the extended olfactory network enforces a modicum of selection bias that is currently unavoidable; therefore, its impact on the reproducibility and the context of our work should be tested by subsequent studies. Indirect evidence (i.e., from studies of human brain interactomes albeit different loci however indicate that IFN-I dysregulation is a plausible and as elsewhere noted, dysregulated mechanism in the CNS following COVID-19 exposure and in the absence of fulminant neuroinfection[31]). Our analyses, despite their limitations on that aspect, would therefore reinforce this concept.

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript could be accepted in its present form, as all previous issues have been fixed.

Author Response

We thank the reviewer for their comments and appraisal of our work.