Network analyses identify liver‐specific targets for treating liver diseases

Abstract We performed integrative network analyses to identify targets that can be used for effectively treating liver diseases with minimal side effects. We first generated co‐expression networks (CNs) for 46 human tissues and liver cancer to explore the functional relationships between genes and examined the overlap between functional and physical interactions. Since increased de novo lipogenesis is a characteristic of nonalcoholic fatty liver disease (NAFLD) and hepatocellular carcinoma (HCC), we investigated the liver‐specific genes co‐expressed with fatty acid synthase (FASN). CN analyses predicted that inhibition of these liver‐specific genes decreases FASN expression. Experiments in human cancer cell lines, mouse liver samples, and primary human hepatocytes validated our predictions by demonstrating functional relationships between these liver genes, and showing that their inhibition decreases cell growth and liver fat content. In conclusion, we identified liver‐specific genes linked to NAFLD pathogenesis, such as pyruvate kinase liver and red blood cell (PKLR), or to HCC pathogenesis, such as PKLR, patatin‐like phospholipase domain containing 3 (PNPLA3), and proprotein convertase subtilisin/kexin type 9 (PCSK9), all of which are potential targets for drug development.

The reviewers' recommendations are rather clear so I think that there is no need to repeat the points listed below. Of course, feel free to contact me in case you would like to discuss any particular point in further detail.
On a more editorial level, we would like to ask you to address the following issues: -Datasets should be provided as individual .xls files labeled and cited as Dataset EV1, Dataset EV2 etc. The description of each dataset should be included in a separate tab in the corresponding .xls file.
-The link http://sysmedicine.com/cns providing access to the networks generated in the study is not working. For long term archival, we would also ask you to provide these files as EV Datasets or to upload them in a public repository. For more information please refer to our author guidelines for Data Deposition.
-The related unpublished study by Mardinoglu et al, should be cited in the text as "Mardinoglu et al, in preparation" (unless it has in the meantime been accepted, and is currently in press). For more information regarding citation of unpublished work you can refer to our Author Guidelines.
-Please provide a "standfirst text" summarizing the study in one or two sentences (approximately 250 characters), three to four "bullet points" highlighting the main findings and a "synopsis image" (211x157 pixels, jpeg format) to highlight the paper on our homepage.
-When you resubmit your manuscript, please download our CHECKLIST and include the completed form in your submission. *Please note* that the Author Checklist will be published alongside the paper as part of the transparent process. The authors present an elegant study where co-expression networks (CNs) were constructed for 46 human tissues and liver cancer, followed up by detailed analysis of liver CNs and subsequently by validation studies in vitro and in vivo.
The manuscript is well written and the presentation is clear.
There are two issues the authors may consider elaborating further.
There is little information what samples were used to contruct CNs. Specifically for liver which has been studied in more detail than other tissues in the study, one would expect that the type of samples included in construction of CNs would affect the co-expression patterns. How were the covariates such as gender, age, amount of liver fat been taken into account?
The method for construction of CNs is rather straightforward and based on Person correlation. No particular statistic was used to evaluate significance of associations and instead a cutoff value was used based on the ranking of associations. The motivation for this choice needs to be explained. Furthermore, Person correlation would lead to large number of spurious associations. Ideally, one would rather apply alternative methods for construction of CNs such as those based on partial correlation. The authors should explain better the reasoning for their choice and perhaps discuss the limitation in the Discussion.
Reviewer #2: In this study, lee et al have generated tissue-specific co-expression networks for 46 major human tissues and human hepatocellular carcinoma (HCC), and deciphered their tissue-specific functions.
The results indicate that hepatic fatty acid synthase (FASN) is co-expressed with a few liver-specific genes associated with de novo lipogenesis that usually occurs during non-alcoholic fatty liver disease (NAFLD) and HCC. In fact, their reduced expression was found to be associated with reduced FASN expression. In addition, the authors potentially identify the cannabinoid type-1 receptor (CB1R) as the upstream modulator of these genes. This work complements their previously documented findings that describe other genomic analysis methodologies (such as transcriptional regulatory and protein-to-protein interaction networks) to identify tissue-specific genes that are correlate with functionality during health and disease. Overall, these are interesting findings that provide compelling evidence for using such strategy to identify novel genes that are dysregulated during diseases. The authors provide a wealth of novel data supporting the fact that FASN expression is associated with PKLR, PNPLA3, and PCSK9, inhibition of which might be used as a potential treatment for NAFLD and HCC. Having said that, there are numerous issues that need to be addressed to improve the quality of the manuscript: 1. To identify the functional relationship between FASN and the hepatic-specific genes identified during NAFLD progression, the authors oddly decided to use a two-week high sucrose diet (HSD) mouse model. Since NAFLD is a well-known chronic condition that affects the liver, it would be important to show that a similar pattern of expression exists following a long-term HSD feeding. Moreover, beside evaluating hepatic triglyceride content (Fig. 5A), the authors should fully characterize their model by providing histological evaluation of the liver, assessing NASH score, determining serum ALT and AST levels, etc. 2. In accordance with the reduced FASN and its associated targets in PCSK9-null mice, are they also resistant to the development of hepatic steatosis on regular chow and HSD feeding? 3. The authors mentioned that the attenuation of FASN and its related targets in CB1R-null mice treated with DEN is associated with decreased tumor growth. However, no data representing tumor size are provided. I understand that they used samples that have been already described (Mukhopadhyay et al., 2015), but at least correlations between tumor size and gene expression should be added. 4. Since peripheral CB1R blockade has been shown to ameliorate HCC (Mukhopadhyay et al., 2015), it would be important to determine whether this effect is associated with reduced FASN expression and its related genes. 5. Identifying PKLR, PCSK9, and PNPLA3 genes to be associated with FASN expression as well as NAFLD and HCC progression is a major novelty of the current paper. Indeed, the authors conclude that targeting these genes/proteins by using a chemical compounds or monoclonal antibodies could be considered as an effective treatment for such diseases. Therefore, to further validate their conclusion, they should consider testing PCSK9 inhibitor for reversing NAFLD or attenuating HCC in their mouse models. Minor comments: 1. It isn't clear whether the data in all figures are presented as mean {plus minus} sem. 2. In figure 5D & E the gene expression is normalized to % of HCC, why? Isn't it better to describe it against the expression in the noncancerous tissues?

RESPONSE TO REVIEWER COMMENTS:
Reviewer #1: The authors present an elegant study where co-expression networks (CNs) were constructed for 46 human tissues and liver cancer, followed up by detailed analysis of liver CNs and subsequently by validation studies in vitro and in vivo. The manuscript is well written and the presentation is clear. There are two issues the authors may consider elaborating further.
We would like to thank the reviewer for careful reading of our manuscript and providing constructive comments.
There is little information what samples were used to construct CNs. Specifically for liver which has been studied in more detail than other tissues in the study, one would expect that the type of samples included in construction of CNs would affect the co-expression patterns. How were the covariates such as gender, age, amount of liver fat been taken into account? From GTEx database, we obtained RNA-seq data of tissues of all deceased donors and we used data as it is, considering that they are from normal population without selection bias. Therefore, coexpression networks based on these cohorts are expected to identify more general functional relationships rather than associations with a specific clinical indication.
Considering the comment of the reviewer, we discussed the limitation of cohorts used in this study in the Discussion section. We have not considered the effect of covariates while constructing CNs because their effects would be distinct by tissues, especially reproductive tissues, and may increase biases while comparing different tissue CNs.
The method for construction of CNs is rather straightforward and based on Person correlation. No particular statistic was used to evaluate significance of associations and instead a cutoff value was used based on the ranking of associations. The motivation for this choice needs to be explained.
Furthermore, Person correlation would lead to large number of spurious associations. Ideally, one would rather apply alternative methods for construction of CNs such as those based on partial correlation. The authors should explain better the reasoning for their choice and perhaps discuss the limitation in the Discussion. Reviewer #2:

Considering unbiased investigations across diverse tissues, we preferred to choose straightforward
In this study, Lee et al have generated tissue-specific co-expression networks for 46 major human tissues and human hepatocellular carcinoma (HCC), and deciphered their tissue-specific functions.
The results indicate that hepatic fatty acid synthase (FASN) is co-expressed with a few liver-specific genes associated with de novo lipogenesis that usually occurs during non-alcoholic fatty liver disease (NAFLD) and HCC. In fact, their reduced expression was found to be associated with reduced FASN expression. In addition, the authors potentially identify the cannabinoid type-1 receptor (CB1R) as the upstream modulator of these genes. This work complements their previously documented findings that describe other genomic analysis methodologies (such as transcriptional regulatory and protein-to-protein interaction networks) to identify tissue-specific genes that are correlate with functionality during health and disease.
Overall, these are interesting findings that provide compelling evidence for using such strategy to identify novel genes that are dysregulated during diseases. The authors provide a wealth of novel data supporting the fact that FASN expression is associated with PKLR, PNPLA3, and PCSK9, inhibition of which might be used as a potential treatment for NAFLD and HCC. Having said that, there are numerous issues that need to be addressed to improve the quality of the manuscript: We would like to thank the reviewer for careful reading of our manuscript and providing constructive comments.
1. To identify the functional relationship between FASN and the hepatic-specific genes identified during NAFLD progression, the authors oddly decided to use a two-week high sucrose diet (HSD) mouse model. Since NAFLD is a well-known chronic condition that affects the liver, it would be important to show that a similar pattern of expression exists following a long-term HSD feeding. Moreover, beside evaluating hepatic triglyceride content (Fig. 5A), the authors should fully characterize their model by providing histological evaluation of the liver, assessing NASH score, determining serum ALT and AST levels, etc.
In our study, we investigated the relationship between hepatic steatosis and the expression of FASN and our target genes. In order to show this association, we fed the mice with HSD for two weeks and measured the Hepatic TG levels as well as the expression of the genes. We also found a reference reporting that it is reasonable for generation of a NAFLD mouse model after 2-3 weeks of HSD ("Koteish A, Diehl A M. Animal models of steatosis, Seminars in liver disease (2001).") In another independent mouse experiment, we fed the mouse with HSD for longer period and measured the hepatic steatosis and the expression of the genes using RT-PCR. We found very similar results. These liver tissue samples will be sent out for RNA sequencing and the global metabolic alterations will be studied in a fallow up study.
2. In accordance with the reduced FASN and its associated targets in PCSK9-null mice, are they also resistant to the development of hepatic steatosis on regular chow and HSD feeding? , it would be important to determine whether this effect is associated with reduced FASN expression and its related genes.

We observed that in the absence of CB1R, there is very low level FASN mRNA in the control and tumor of CB1R-/-mice. It further strengthens our hypothesis and FASN is detectable in higher level in HCC mice when tumor is present.
5. Identifying PKLR, PCSK9, and PNPLA3 genes to be associated with FASN expression as well as NAFLD and HCC progression is a major novelty of the current paper. Indeed, the authors conclude that targeting these genes/proteins by using a chemical compounds or monoclonal antibodies could be considered as an effective treatment for such diseases. Therefore, to further validate their conclusion, they should consider testing PCSK9 inhibitor for reversing NAFLD or attenuating HCC in their mouse models. Dear Adil, Thank you for sending us your revised manuscript. We have now heard back from reviewer #2 who was asked to evaluate your study. As you will see below, the reviewer is satisfied with some of the modifications made but has some remaining concerns, which we would ask you to address in a revision of the manuscript.
While we think that the experiments in mice fed a long-term HSD (point #1 of the reviewer) are not mandatory for acceptance of the paper, we think that points #4 and #5 need to be addressed, since they would indeed enhance the conclusiveness of the study.
On a more editorial level, I would like to draw your attention to the following points: -I have made some changes in the standfirst text, bullet points and abstract (see attached file). Could you please let me know whether you agree with these changes or if you would prefer to further edit the text?
-I would also suggest the slightly modified title: "Network analyses identify liver-specific targets for treating liver diseases".
Please resubmit your revised manuscript online, with a covering letter listing amendments and responses to each point raised by the referees. Please resubmit the paper **within one month** and ideally as soon as possible. If we do not receive the revised manuscript within this time period, the file might be closed and any subsequent resubmission would be treated as a new manuscript. Please use the Manuscript Number (above) in all correspondence.
As a matter of course, please make sure that you have correctly followed the instructions for authors as given on the submission website.
Thank you for submitting this paper to Molecular Systems Biology. 1. The authors have decided not to provide data, which evaluate their HSD-induced hepatic steatosis mouse model. Instead, they have chosen to cite a paper describing the model. As I mentioned before, NAFLD is a chronic condition, and I still believe that validating their data in a new set of animals would greatly add to the impact of the current paper. 2. Thank you for clarifying this point. 3. I am satisfied with the explanation provided. 4. The authors did not determine the expression levels of FASN in mice treated with a peripheral CB1 receptor blocker, which has been shown to reduce HCC occurrence. 5. There are currently two PCSK9 inhibitors, Evolocumab (Repatha) and Alirocumab (Praluent) that have been developed and tested, so stating that there is no inhibitor available is basically not too accurate.

RESPONSE TO REVIEWER COMMENT:
We would like to thank the reviewer for careful reading of our manuscript and providing constructive comments.
1. The authors have decided not to provide data, which evaluate their HSD-induced hepatic steatosis mouse model. Instead, they have chosen to cite a paper describing the model. As I mentioned before, NAFLD is a chronic condition, and I still believe that validating their data in a new set of animals would greatly add to the impact of the current paper.
We performed additional independent mouse experiment, generated RNAseq data and confirmed our results. Data has been presented in the thesis of Mattias Bergentall and will be published in a follow up study. Please let me know if you want me to send the thesis or the draft of the paper. Considering that we had similar results, we decided not to include in this paper.
2. Thank you for clarifying this point.
3. I am satisfied with the explanation provided.
4. The authors did not determine the expression levels of FASN in mice treated with a peripheral CB1 receptor blocker, which has been shown to reduce HCC occurrence. Figure 5. Moreover, we had tumor size values from individual mice and the FASN gene expression was measured in these set of mice.

We have already included the expression of FASN in the noncancerous and tumor samples obtained from wild type and CBR1 KO mice in
5. There are currently two PCSK9 inhibitors, Evolocumab (Repatha) and Alirocumab (Praluent) that have been developed and tested, so stating that there is no inhibitor available is basically not too accurate.  Do the data meet the assumptions of the tests (e.g., normal distribution)? Describe any methods used to assess it.
Is there an estimate of variation within each group of data?
Is the variance similar between the groups that are being statistically compared?
Yes. We used appropriate statistical method.
Yes. The data meet the assumptions of the tests.

Yes.
Yes. We used appropriate statistical method.

YOU MUST COMPLETE ALL CELLS WITH A PINK BACKGROUND 
In each mouse and cell line experiment we included more than 3 samples and adjusted p--values for multiple testing.
We included 10 animals in each group of mice fed with CD and HSD and have also verified the results in an independent experiment.
No samples were excluded from the analysis.
We have separeted animals in two two group and suplemented one group with the HSD to see its effect on liver fat. The lipid and gene expression measurements were performed by a staff scientist with blinded samples.
We have separeted animals in two two group and suplemented one group with the HSD to see its effect on liver fat. The person who handled the mice did not perform the lipidomics analyses of the samples.
The lipidomics analyses were performed with blinded samples. The clinical chemistry measurements were performed by a professional accredited clinical chemistry laboratory with blinded sampes. The measurement of liver fat were performed by specialized clinical staff without knowing anything about the experiment.
We have not consider blinding of the investigator. The reason is that we wanted the same person fed the mice, in order to minmize variation. In addition, this person did not perform any subsequent analysis of the samples.

Captions
The data shown in figures should satisfy the following conditions: Source Data should be included to report the data underlying graphs. Please follow the guidelines set out in the author ship guidelines on Data Presentation. a statement of how many times the experiment shown was independently replicated in the laboratory.
Any descriptions too long for the figure legend should be included in the methods section and/or with the source data.
Please ensure that the answers to the following questions are reported in the manuscript itself. We encourage you to include a specific subsection in the methods section for statistics, reagents, animal models and human subjects.
In the pink boxes below, provide the page number(s) of the manuscript draft or figure legend(s) where the information can be located. Every question should be answered. If the question is not relevant to your research, please write NA (non applicable).

B--Statistics and general methods
the assay(s) and method(s) used to carry out the reported observations and measurements an explicit mention of the biological and chemical entity(ies) that are being measured. an explicit mention of the biological and chemical entity(ies) that are altered/varied/perturbed in a controlled manner. the exact sample size (n) for each experimental group/condition, given as a number, not a range; a description of the sample collection allowing the reader to understand whether the samples represent technical or biological replicates (including how many animals, litters, cultures, etc.).

Reporting Checklist For Life Sciences Articles (Rev. July 2015)
This checklist is used to ensure good reporting standards and to improve the reproducibility of published results. These guidelines are consistent with the Principles and Guidelines for Reporting Preclinical Research issued by the NIH in 2014. Please follow the journal's authorship guidelines in preparing your manuscript.