Inhibition of NF-κB Signaling Alters Acute Myelogenous Leukemia Cell Transcriptomics

Round 1

Reviewer 1 Report

The author in this study try to address the effect of NF-kB inhibition on global gene expression changes. 16 leukemia patient samples were used in this study and the cells isolated from the leukemia patients were either treated with BMS-345541 or DMSO as control followed by micro-analysis. The author found that inhibitor treatment resulted in misregulation of about 280 genes. Following this, the author tried to analyze the micro-array data in different ways and unfortunately, the data is unconvincing in its totality. The study provides descriptive explanation of outcome of the gene expression analysis. In this period when RNAseq is usually carried out, performing only microanalysis and trying to wrap a story around the dataset is probably not enough to get a paper published in a journal like cells. Moreover, the data presented in the paper looks inconclusive and lacked proper explanation of the experiment or data presented. For e.g volcano plot showed on the genes which are downregulated. The meta analysis was difficult to understand. It seems gene expression also changed due to DMSO treatment only and therefore the data needs to be analyzed by normalizing the gene expression changes without any treatment. What does a protein interaction map drawn based on misregulated genes even demonstrate? Unfortunately, I do not agree with the author and I recommend this paper not be published in the current form.

Author Response

Reviewer 1:

The author in this study try to address the effect of NF-kB inhibition on global gene expression changes. 16 leukemia patient samples were used in this study and the cells isolated from the leukemia patients were either treated with BMS-345541 or DMSO as control followed by micro-analysis. The author found that inhibitor treatment resulted in misregulation of about 280 genes. Following this, the author tried to analyze the micro-array data in different ways and unfortunately, the data is unconvincing in its totality. The study provides descriptive explanation of outcome of the gene expression analysis. In this period when RNAseq is usually carried out, performing only microanalysis and trying to wrap a story around the dataset is probably not enough to get a paper published in a journal like cells. Moreover, the data presented in the paper looks inconclusive and lacked proper explanation of the experiment or data presented. For e.g volcano plot showed on the genes which are downregulated. The meta analysis was difficult to understand. It seems gene expression also changed due to DMSO treatment only and therefore the data needs to be analyzed by normalizing the gene expression changes without any treatment. What does a protein interaction map drawn based on misregulated genes even demonstrate? Unfortunately, I do not agree with the author and I recommend this paper not be published in the current form.

The comments made by the reviewer are mainly general and differ especially from the comments given by reviewers 2 and 3. We have tried to carefully address those examples given by the reviewer.

The Volcano plot has been left out, this was also commented by reviewer 4.

We are aware of the increasing use of RNA sequencing for molecular genetic profiling. However, microarray methodology and transcriptomics in our opinion also represent an established and highly standardized and validated methodological strategy that has been widely used for two decades in cancer research, including AML. These previous studies have demonstrated considerable consistence with regard to characterization of gene expression profiles, e.g. for AML subclassification, prognostication and as a basis for targeted therapy. This is briefly commented in a new chapter in the Discussion section (lines 347-356).

Regarding the presentation of our data we believe the methodological approaches and visualization of the data are based on robust and established methodology. However, we agree as also comment by reviewer 4 that that the information given by Volcano plot is limited. Accordingly, we have left the volcano plot out in the Revised Version of our manuscript.  We also agree that the description of the meta-analysis was incomplete, and accordingly we have improved the explanation of these approaches in our revised version (as also comment by reviewer 4).

Regarding the use of DMSO in control cultures, in our opinion the control cultures have to include DMSO as long as the inhibitor BMS-345541 is dissolved in DMSO. The control cultures contained the same final DMSO concentration as the BMS-345541 containing cultures. This is now clearly stated in Material and methods (lines 86-88 ). It should also be emphasized that the final DMSO concentration in the control and the BMS-345541 cultures is very low, and these concentrations do not alter the viability and proliferation of AML cells compared with cells incubated in medium alone. This is also stated in the Material and method section.

We agree that the results from our network analyses have to be interpreted with care because we analyzed gene expression profiles whereas the interaction networks are based on protein interactions. However, based on the important role of NF-κB in the regulation of gene transcription we think it is important to try to indicate the possible functional importance of our present results. For this reason, we hope it can be acceptable to include these data in our Revised Version, but we have added a new chapter to the Discussion section where this is commented (lines 357-369).

Reviewer 2 Report

In the present manuscript, authors have conducted a study to investigate the global gene expression profile with NF-kB inhibitor (BMS-345541) in primary leukemic blasts. Authors have performed various analysis tests and with the use of databases such as PANTHER and STRING, they were able to discriminate the genes affected by NF-kB inhibition. Their bioinformatics analysis indicated that the genes which seem to be affected by the NF-kB inhibition are mostly involved in the three major biological systems: cytokine and interleukin signaling, metabolic systems and immune system. Moreover, they found some key genes involved in leukemogenesis as well as leukemic stem cell profiles which are affected by NF-kB inhibition. This article is quite comprehensive and is useful for oncologists and cancer researchers. There are a few minor points that should be addressed in the manuscript.

1. The current study is based on the data obtained with the treatment of an NF-kB inhibitor- BMS-345541 in AML patients. Thus, while one appreciates the volume of work undertaken in this study, it is important to validate the findings with other NF-kB inhibitors.

2. The study requires validation of NF-kB inhibition in the cells used in this study. Author should provide western blot data and/or inhibition assay which shows NF-kB inhibition in the primary blasts.

1. After performing Feature subset analysis, the authors identified 277 genes that were differentially expressed after treatment with NF-κB inhibitor. This includes 183 genes that were downregulated and 94 genes which were upregulated. The authors further performed gene ontology enrichment analysis among the downregulated genes only. Why authors have only chosen downregulated genes for further analysis? What happens when the enrichment analysis is conducted among upregulated genes?

1. There are several published studies suggesting the role of NF-kB activation in AML, thereby suggesting NF-kB inhibitors as a treatment option. Therefore, it is important for authors to describe the novelty and significance of the current study in detail in the conclusion section. 

5. Are there any in vitro studies or preclinical data available that supports the author’s findings of altered cytokine and interleukin signaling after NF-κB inhibition?  

6. Based on the data obtained by the authors with BMS-345541 treatment in leukemic blasts, can authors suggest BMS-345541 as a treatment option for AML patients? 

7. It is recommended to check the grammatical and spelling errors throughout the manuscript. For instance ‘signaling instead of signalling’. 

Author Response

In the present manuscript, authors have conducted a study to investigate the global gene expression profile with NF-kB inhibitor (BMS-345541) in primary leukemic blasts. Authors have performed various analysis tests and with the use of databases such as PANTHER and STRING, they were able to discriminate the genes affected by NF-kB inhibition. Their bioinformatics analysis indicated that the genes which seem to be affected by the NF-kB inhibition are mostly involved in the three major biological systems: cytokine and interleukin signaling, metabolic systems and immune system. Moreover, they found some key genes involved in leukemogenesis as well as leukemic stem cell profiles which are affected by NF-kB inhibition. This article is quite comprehensive and is useful for oncologists and cancer researchers. There are a few minor points that should be addressed in the manuscript.

 We are grateful for this in general positive comments regarding our present manuscript.

1. The current study is based on the data obtained with the treatment of an NF-kB inhibitor- BMS-345541 in AML patients. Thus, while one appreciates the volume of work undertaken in this study, it is important to validate the findings with other NF-kB inhibitors.

We agree in this comment. However, we would emphasize that we used a well characterized inhibitors that has been used in several previous studies. These previous results are summarized in a new chapter in the Discussion section. Relevant references have been included (lines 362-374 ).

1. The study requires validation of NF-kB inhibition in the cells used in this study. Author should provide western blot data and/or inhibition assay which shows NF-kB inhibition in the primary blasts.

Molecular studies including Western blots have been included in the previous studies of this inhibitor; this is explained in the new chapter of the Discussion section (lines 436-448).

3. After performing Feature subset analysis, the authors identified 277 genes that were differentially expressed after treatment with NF-κB inhibitor. This includes 183 genes that were downregulated and 94 genes which were upregulated. The authors further performed gene ontology enrichment analysis among the downregulated genes only. Why authors have only chosen downregulated genes for further analysis? What happens when the enrichment analysis is conducted among upregulated genes?

 We also performed a such analysis on upregulated genes, although this analysis is not so profound and conscience as downregulated genes. We have comment this our revised version of the manuscript.

4. There are several published studies suggesting the role of NF-kB activation in AML, thereby suggesting NF-kB inhibitors as a treatment option. Therefore, it is important for authors to describe the novelty and significance of the current study in detail in the conclusion section. 

 We agree in this comment. An additional comment is now included in the Conclusion section.

1. Are there any in vitro studies or preclinical data available that supports the author’s findings of altered cytokine and interleukin signaling after NF-κB inhibition?  

An additional comment is now included in the Discussion section (lines  413-415). This is an important point, and we have highlighted studies incorporating these findings to includes studies regarding cytokine metabolism in leukemic cells as results of BMS-345541 inhibition.

6. Based on the data obtained by the authors with BMS-345541 treatment in leukemic blasts, can authors suggest BMS-345541 as a treatment option for AML patients? 

We appreciate this comment, emphasizing the importance of translational aspect, purchasing and “bench to bedside” approach. Although NF-κB inhibition seems both as a rational and suitable approach, our data are not sufficient to support entering of BMS-345541 into clinical trials. However, we believe this data is support as backbone for further entering of BMS-345541 and other BMS-345541 inhibitors in clinical trials. We have highlighted these features in the conclusion of our revised manuscript.

7. It is recommended to check the grammatical and spelling errors throughout the manuscript. For instance ‘signaling instead of signalling’. 

We have carefully controlled the grammar and spelling.

Reviewer 3 Report

The paper by H. Reikvam is mostly descriptive and reports gene regulation upon BMS-345541 (NFkB pathway inhibitor) treatment of 16 primary AML patient samples in culture. No statistics are provided. The number of samples is by far too low to conclude anything. In addition, they are many typos in the text such as: principle instead of principal, finale instead of final, ..... Finally, none of the figure brings reliable information as detailed below.

Major points:

Fig 1: Volcano plot is not a volcano plot!

Fig 2: X values are not defined nor p values!

Fig 3 is not informative

Fig 4: The different networks mentioned in the text must be highlighted

Fig 5 is not informative

Fig 6: the LSC signature, to which the author refers, is an old LSC signature with a large number of genes described in 2011 (Eppert, Nat Med, 2011). More recently the same tream published  a new LSC signature with only 17 LSC genes (Ng, Nature, 2016). Fig 6 should be interpreted in the light of these most recent results.

Author Response

The author has performed a study aiming to evaluate the role of NF-kB as a therapeutic target and the association between the alterations in GEP and the treatment with NF-kB inhibitor BMS-345541 in primary leukemic blasts from 16 consecutive AML patients.

The design of study, statistical analysis and endpoints are good, and the study can be really interesting for readers. The paper is written in a good English, and it can be clear for readers.

The design of the study is original and really useful. The data are well presented and results are well discussed.

In particular, it is interesting that key genes involved in leukemogenesis result affected by NF-kB inhibition.

Into discussion, it could be useful to add a table with other studies in the same topic.

However, the idea is good and well presented and the paper could be really interesting for readers, and it can be accepted for publication after minor revision.

We are very grateful for these general comments.

As suggested by the reviewer we have added a short new Table 2 where important observations from previous studies of NF-κB inhibition in AML are summarized.

Reviewer 4 Report

The author has performed a study aiming to evaluate the role of NF-kB as a therapeutic target and the association between the alterations in GEP and the treatment with NF-kB inhibitor BMS-345541 in primary leukemic blasts from 16 consecutive AML patients.

The design of study, statistical analysis and endpoints are good, and the study can be really interesting for readers. The paper is written in a good English, and it can be clear for readers.

The disegn of the study is original and really useful. The data are well presented and results are well discussed.

In particular, it is interesting that key genes involved in leukemogenesis result affected by NF-kB inhibition.

Into discussion, it could be useful to add a table with other studies in the same topic.

However, the idea is good and well presented and the paper could be really interesting for readers, and it can be accepted for publication after minor revision.

Author Response

The paper by H. Reikvam is mostly descriptive and reports gene regulation upon BMS-345541 (NFkB pathway inhibitor) treatment of 16 primary AML patient samples in culture. No statistics are provided. The number of samples is by far too low to conclude anything. In addition, they are many typos in the text such as: principle instead of principal, finale instead of final, ..... Finally, none of the figure brings reliable information as detailed below.

We agree that AML is a heterogeneous disease, and for this reason the number of patients is far to low to address the question of whether AML patients are also heterogeneous with regard to the effect of NFkB inhibition. However, despite the heterogeneity of AML patients with regard to the biological characteristics of their leukemic cells, we could identify a relatively large number of genes with altered expression after NFkB inhibition. These genes represent an effect of NFkB inhibition that seems to be common for all or most patients. We would emphasize that dditional effects may be important for single patients or subsets of patients. This is now clearly stated in a new hapter of the Discussion section.

Grammar and spelling has been carefully controlled. If the article is accepted we will then use the journals service for additional control of the English language.

We will emphasize that all figure legends have been carefully rewritten to better describe the figures. Detailed comments for each individual figure are given below.

Major points:

Fig 1: Volcano plot is not a volcano plot!

We agree that this plot could be misleading and not an optimal way of presenting our data (this was also commented by reviewer 1). Accordingly, we have left out this figure in our Revised Version.

Fig 2: X values are not defined nor p values!

We agree in this comment, and accordingly Figure 2 has been reviewed and we belive that the new version gives a better version of our data. Only significant GO-terms are presented. The legend has also been rewritten

Fig 3 is not informative

In our opinion Figure 3 should be included in the study. One should not only present GO terms; in our opinion a list of the most important individual genes should also be available to the reader. For this reason we hope that the figure can be accepted. However, we agree that the figure legend was too brief, and accordingly we have carefully revised the legend.

Fig 4: The different networks mentioned in the text must be highlighted

We appreciate this comment and agree that highlighting of the mention networks would make the interpretation easier for the readers. The four main networks have therefore been highlighted. The figure legend has also been rewritten; the main functional mplications of the four networks are briefly stated in the legend and are not only mentioned in the text.

Fig 5 is not informative

Although we agree that the network analysis presented in Figure 5 is less important than the analysis presented in Figure 4 presenting the downregulated genes, we believe it is important to also describe the upregulated genes and this was also requested by reviewer 2. For these reasons we hope it is acceptable to include this figure also in the Revised Version of our article.

Fig 6: the LSC signature, to which the author refers, is an old LSC signature with a large number of genes described in 2011 (Eppert, Nat Med, 2011). More recently the same tream published  a new LSC signature with only 17 LSC genes (Ng, Nature, 2016). Fig 6 should be interpreted in the light of these most recent results.

We are grateful for this comment. We are aware of the fact that different leukemic stem cell signatures/profiles have been obtained, and that the signature of NG is more validated and robust based on fewer genes. Accordingly, we have reanalyzed our data based on this 17-gene stem cell profile. The results are presented and highlighted in a new and revised version of Figure 6. The more recent reference is included

Round 2

Reviewer 1 Report

The author has significantly amended the manuscript that currently sounds ok. However, I still do not agree with the author on the textual revision of the protein interaction map. According to the author those pieces of data has to be "interpreted with great care". According to me, this statement would suggest the data is inconclusive and therefore should not be in the manuscript. Alternatively, can the author carry out experiment and show that atleast one major node each from downregulated and upregulated gene list, actually has inter-regulation or inter-connection? 

Author Response

We are grateful for the general comments to the First Revised Version of the manuscript. We also agree that we have to present and discuss the interaction map more in detail to justify its inclusion in the manuscript. The problem by leaving it out is that one of the other reviewers required an additional analysis, i.e. we should not only include analysis of the downregulated genes but an additional analysis of the upregulated genes. The two last reviewers did not have any critical comments to this analysis. We have tried to make a compromise between the reviewers in our Second Revised Version. We hope this can be accepted, if the reviewer cannot accept our present solution we will of course leave these parts out. We have now done the following alterations in our manuscript:

• The figure showing the results for the upregulated genes is now left out from the article and is presented as Figure S1.
• The interaction analysis of the downregulated genes is still in the article.
• We have now included a more detailed description of the downregulated gene analysis in our new Section 3.4; we now give a brief description of each of the four gene subsets identified in Figure 4.
• Each individual gene from the four subsets identified in Figure 4 are now briefly described in a new Table S2. Several new references are also included to that these genes are relevant for AML both with regard to leukemogenesis and chemosensitivity.
• The upregulated genes are presented in the same way, except that the figure is now included as Figure S1. A detailed description of the identified genes is given in the new Table S3. Additional references are given.
• The discussion and conclusions based on these analyses has been extended in the new chapter in the Discussion section. We also state why we mean that these observations should be interpreted with care; although their importance is supported by previous studies (i.e. these genes and proteins are indeed relevant in AML) one has to be careful because protein expression is regulated at different levels and not only at the mRNA level.

We hope our compromise/solution can be accepted by the reviewer, otherwise we will leave these parts out as we stated above.

Reviewer 3 Report

Although the manuscript has been improved top some extent, the study still remains mostly descriptive. In addition, and contrary to what the authors claim, the Fig 6 has not been changed. Only the legend has been changed !!!! 

Author Response

I am very sorry for this mistake that was due to technical problems with the submission. I can only apologize. I have now uploaded the correct new Revised Version of this figure, i.e. Figure 5 in the Revised Version of the manuscript. In our opinion this new figure should be more suitable for presentation of our data.