Insulin and 5-Aminoimidazole-4-Carboxamide Ribonucleotide (AICAR) Differentially Regulate the Skeletal Muscle Cell Secretome

Round 1

Reviewer 1 Report

In this manuscript, the authors used a label-free high-resolution MS-based proteomic analysis to evaluate the effects of 5-aminoimidazole-4-carbox- 21 amide ribonucleotide (AICAR – an AMPK activator) and insulin-treated on the secretome of culture media and the proteome from differentiated C2C12 myotubes. This article is well written and highlighted several interesting point including a deep analysis of the secretome adaptations upon the insulin and AICAR effects. To consider this manuscript for publication, this authors should add some data as mentioned in the comments below.

GENERAL COMMENTS

ABSTRACT

No comment

INTRODUCTION

No comment

EXPERIMENTAL SECTION

• The authors should give more details about the mouse UniProt sequence database they used. Swissprot database? Which release?
• The authors should specify how many technical replicates and biological replicates they performed for secretomic and proteomic study
• The information on the criteria used to define the down- and up-regulated proteins should be clearly indicated in the experimental section or results section
• The authors should indicate which software they used to performed the Gene Ontology analysis and the parameter used.
• The authors should indicate if they used at least 2 or 1 unique peptides to consider the protein as identified.
• Line 154: The authors wrote “The Pfam database 154 (pfam.xfam.org) was used for domain predictions and enrichment analyses.” It would be interesting to clarify what are the domain prediction and what for as well as what kind of enrichment analysis. If it is for Gene Ontology, why the authors did use it while it is mostly dedicated to alignment/prediction sequence rather than gene ontology enrichment analysis. And what is the advantage of it compared to other softwares such as David bioinformatics resources or ingenuity pathway, or pathway studio

RESULTS

• Line 252: when the authors wrote “owever, a weak correlation (Pearson=0.445; R2=0.198) was observed when comparing the median LFQ intensities of proteins from both datasets, again suggesting that few, if any, of the identified secreted proteins were released due to cell death (Figure 2E).”, it would have been important to perform apoptotic assays to support this statement.
• Do the authors control the concentration of AICAR or insulin that were internalized into the cells?
• Do the authors perform LC-MS analysis on control quality samples (QC) to enseure the stability of the signal during the LC-MS analysis?
• It may be interesting to mention other proteomic studies performed on secretome of media cultures with particular interest in metabolic adaptations (such as Goron A., Breuillard C., Cunin V., Bourgoin-Voillard S., Seve M., Moinard C., Modulation of muscle protein synthesis by amino acids What consequences for the secretome? - A preliminary in vitro study, Amino Acids, 2019)

Author Response

Reviewer 1

Comments and Suggestions for Authors

In this manuscript, the authors used a label-free high-resolution MS-based proteomic analysis to evaluate the effects of 5-aminoimidazole-4-carbox- 21 amide ribonucleotide (AICAR – an AMPK activator) and insulin-treated on the secretome of culture media and the proteome from differentiated C2C12 myotubes. This article is well written and highlighted several interesting point including a deep analysis of the secretome adaptations upon the insulin and AICAR effects. To consider this manuscript for publication, this authors should add some data as mentioned in the comments below.

We thank Reviewer 1 for their detailed feedback and critical assessment of the manuscript. We have carefully considered each point and addressed them below. We have clarified several aspects of the experimental design and analyses as described below.

GENERAL COMMENTS

ABSTRACT

No comment

INTRODUCTION

No comment

EXPERIMENTAL SECTION

The authors should give more details about the mouse UniProt sequence database they used. Swissprot database? Which release?

The MS/MS spectra were searched against the mouse UniProt FASTA database (version November 2017). This information has been added to the manuscript at lines 141-142.

The authors should specify how many technical replicates and biological replicates they performed for secretomic and proteomic study.

The number of replicates are indicated in the figure legends and on line 150. However, this information has also been added to the manuscript text on line 108.

The information on the criteria used to define the down- and up-regulated proteins should be clearly indicated in the experimental section or results section

We acknowledge that this is an important detail to provide. Therefore, we have added a more detailed statistics section within the Methods (which clarifies the paired comparison test.

The authors should indicate which software they used to performed the Gene Ontology analysis and the parameter used.

 This has been clarified in the new Statistics section. All statistical and enrichment analyses were performed using Perseus software [PMID: 27348712]. Enrichment analyses were performed using a Fisher’s exact test on secreted proteins that were differentially regulated within each comparison (i.e., Ctrl vs. AICAR and Ctrl vs. insulin). The 2D enrichment was assessed using the same software package, and has been thoroughly described previously [PMID: 23176165]. These detailed descriptions have again been added to the Methods.

The authors should indicate if they used at least 2 or 1 unique peptides to consider the protein as identified.

We thank the reviewer for raising this important question. For the protein quantification, we used the default settings in MaxQuant which takes into account both unique and razor peptide-based quantification. In the secretome, it was observed that only 21/1192 (~2%) of proteins were identified with fewer than two unique peptides. This information is included in Table S3 and has been clarified in the Methods at lines 147-148.

 Line 154: The authors wrote “The Pfam database 154 (pfam.xfam.org) was used for domain predictions and enrichment analyses.” It would be interesting to clarify what are the domain prediction and what for as well as what kind of enrichment analysis. If it is for Gene Ontology, why the authors did use it while it is mostly dedicated to alignment/prediction sequence rather than gene ontology enrichment analysis. And what is the advantage of it compared to other softwares such as David bioinformatics resources or ingenuity pathway, or pathway studio

For enrichment analyses, the Pfam annotations from EMBL-EBI were used to identify groups of proteins sharing a similar domain that were overrepresented in our dataset [PMID: 33125078]. We have used a similar approach earlier where we described the secretome of insulin resistant C2C12 cells (PMID: 26457550). Similar to Gene Ontology enrichment analysis, we believe that this approach extracts additional information on protein domains overrepresented in specific populations (in this case significantly regulated secreted proteins). Software solutions such as David, Ingenuity and Perseus perform enrichment analysis in slightly different manners. Compared to the enrichment analysis in other software solutions, Perseus uses total quantified proteins in the given experiment (instead of the complete species-specific database). This approach is more stringent and widely accepted, providing a more reliable enrichment analysis. Nevertheless, it is unlikely that our observations would have changed if we had used David or Ingenuity pathway analysis.

RESULTS

Line 252: when the authors wrote “owever, a weak correlation (Pearson=0.445; R2=0.198) was observed when comparing the median LFQ intensities of proteins from both datasets, again suggesting that few, if any, of the identified secreted proteins were released due to cell death (Figure 2E).”, it would have been important to perform apoptotic assays to support this statement.

While these assays would provide a proxy of cell apoptotic pathway activation, they do not estimate membrane damage or changes to cytosolic protein leakage. To that end, we feel our statement is adequately supported by the observed lack of difference in cytosolic lactate dehydrogenase (LDH) content across experimental groups, which is mentioned on lines 234-238. Any relative differences in secreted protein responses across the groups are unlikely to be due to protein leakage from membrane damage or cell death, particularly that which is unlikely to arise following acute insulin/AICAR treatment. LDH release has been long regarded as an accurate measure to estimate cell death [PMID: 22294120; 8872918].

Do the authors control the concentration of AICAR or insulin that were internalized into the cells?

There was no direct measurement of compound internalization into cells since we were predominantly interested in the downstream changes to secreted proteins following the activation of these distinct cellular pathways. As such, AICAR and insulin signaling were confirmed by western blot of proteins known to be activated (phosphorylated) by each treatment.

Do the authors perform LC-MS analysis on control quality samples (QC) to ensure the stability of the signal during the LC-MS analysis?

We thank the reviewer for highlighting a very important consideration. As a standard practice, QC runs involving BSA and HeLa peptides are performed before and after measuring the samples. These runs were used to monitor the MS signals and HPLC/column performance. These QC runs ensured that LC and MS performance were not affected due to our samples.

It may be interesting to mention other proteomic studies performed on secretome of media cultures with particular interest in metabolic adaptations (such as Goron A., Breuillard C., Cunin V., Bourgoin-Voillard S., Seve M., Moinard C., Modulation of muscle protein synthesis by amino acids What consequences for the secretome? - A preliminary in vitro study, Amino Acids, 2019)

Thank you for this excellent and very relevant suggestion. This recent study further supports the use of myotubes as an effective model to characterize the secretome in response to nutritional/pharmacological stressors. In this case, the independent effects of hypo/hyperaminoacidemia as well as leucine and citrulline (two amino acids known to regulate muscle protein synthesis) were assessed. We have added this reference and briefly highlighted the paper in text at lines 284-286.

Reviewer 2 Report

The paper entitled " Insulin and 5-aminoimidazole-4-carboxamide Ribonucleotide (AICAR) Differentially Regulate the Skeletal Muscle Cell Secretome” is very interesting, deeply and provide novel data. The paper is well written and has merit of publication.

• Raw data should be submitted via PRIDE or similar.

Author Response

Reviewer 2

Comments and Suggestions for Authors

The paper entitled " Insulin and 5-aminoimidazole-4-carboxamide Ribonucleotide (AICAR) Differentially Regulate the Skeletal Muscle Cell Secretome” is very interesting, deeply and provide novel data. The paper is well written and has merit of publication.

Raw data should be submitted via PRIDE or similar.

We thank Reviewer 2 for their positive comments regarding the data and the manuscript. The raw MS data has been submitted via the PRIDE repository under the identifier PXD025687, Username: [email protected], Password: WKMh4wZ4. This is indicated in the manuscript under “Data Availability”. We will ensure that all data is publicly available.

Reviewer 3 Report

General comments

• Line 43 Although authors noted "While the understanding of the skeletal muscle secretome remains incomplete, myokines (e.g. interleukin-6; IL-6) are proven regulators of inflammation, immune function and energy metabolism (7, 8).", is it true? Recent study shows that macrophages invaded after exercise produce IL-6 massively, and your logic might be denied in the future.  
• Not all experimental methods performed are described in the methods parts.
• The authors should describe the details of the experimental methods without omission (e.g., the sample numbers of biological replicate, statistical analysis for ELISA, immunoblotting, PCA analysis and ELISA for IGFBP7)
• Please standardize the description of the number of biological replicates in the figure caption.
• It is unclear which measurement parameters use the same sample or whether independent experiments were performed, and please provide details in methods part.

Specific comments

• Line 25  The capitalization is necessary for "Insulin"?
• Line 93 The description “7000g, 10 min” is different from other description (8000g x 10 min).
• In Figure 1B, the authors should describe which bar is which group in the graph.
• In Figure 1B, I cannot understand which asterisk indicates a significant difference between the groups, please show it as Figure 1C.
• For the pAkt/totalAkt data in Figure 1B, please note that the bars in the Insulin-stimulated group omit the middle value.
• Line 104 Please rewrite to cellular proteome as it is easily confused.
• Line 128 27%; can be 27%.  
• Line 238 It may be Figure 2B, not Figure 2A.
• Figure 2C is not explained in the results section.
• Line 241-246 If serum-deprivation influences apoptosis, apoptosis markers such as cleaved caspase-3 should be measured in cell lysis. Generally apoptosis is triggered by muscle differentiation (https://doi.org/10.1007/s10495-013-0922-7); therefore, it may be reasonable to include intracellular protein in the condition medium. The LDHA data only indicate that there was no difference in the effect of cell death between the groups. If the authors want to demonstrate the effect of serum-deprivation, the authors must compare between the presence and absence of serum in the medium.
• Line 249-255 The correlation between secretome and cellular proteome does not indicate whether intracellular protein was released by cell death or not. Also, since the database is used to filter out secreted proteins, it may not be a problem if the secretome contains intracellular proteins.
• Line 255-257 Please describe in the Methods part how you calculated the correlation coefficient. Also, please describe in the Methods part how you determined that there are more media than cells, and explain this in the Results. In addition, since the sentences are difficult to understand, please rewrite them in concise sentences.
• Line 262 Please replace "This study" with "That previous study/above study" to avoid confusion.
• Line 275　The sentences may be mistaken.
• Line 303-304 Please describe all details of the analysis methods used in this experiment, including the method of PCA analysis, in the methods section.
• Line 366-368 There is n = 4-5, can't you do a statistical analysis?
• Line 368-378 In the data of Figure 4E-F, please describe which analysis method was used (e.g., GO or KEGG). What does Figure 4E-F show? The p-value on the horizontal axis is positive on both sides. In pathway enrichment analysis, it usually analyzes how many GO terms are tagged to variable proteins and does not compare the increase or decrease of GO terms.
• Line 400-402 The details of the experiment with the addition of culture medium are not described in the Methods part.
• For the experiments in Figure 5D-G, why is the number of biological replicates large?

Author Response

Reviewer 3

Comments and Suggestions for Authors

We thank Reviewer 3 for their very thorough and critical assessment of the manuscript and data analysis. We have carefully considered each suggestion, updated the appropriate figures and text, and feel that these changes have strengthened the overall quality of our submission.

General comments

Line 43 Although authors noted "While the understanding of the skeletal muscle secretome remains incomplete, myokines (e.g. interleukin-6; IL-6) are proven regulators of inflammation, immune function and energy metabolism (7, 8).", is it true? Recent study shows that macrophages invaded after exercise produce IL-6 massively, and your logic might be denied in the future.

This statement is not specific to exercise, although macrophage-derived IL-6 has yet to be shown as exercise intensity-dependent, despite evidence to suggest that circulating IL-6 levels increase with increases in exercise intensity [PMID: 31262006]. Beyond myocytes and macrophages, there are also several other IL-6-producing cell types and organs. The relative contribution of each to plasma IL-6 concentrations are highly dependent on a variety of factors, including hypoxia, energy status and nutrition. Skeletal muscle undoubtedly remains an important source of circulating IL-6 which can regulate several biological processes [PMID: 8379461, 11600669, 15059966] and post-exercise circulating IL-6 concentrations are reduced in skeletal muscle-specific IL-6 knockout animals [PMID: 29253016, 33178046].

Not all experimental methods performed are described in the methods parts.

The authors should describe the details of the experimental methods without omission (e.g., the sample numbers of biological replicate, statistical analysis for ELISA, immunoblotting, PCA analysis and ELISA for IGFBP7)

An additional “Statistics'' section has been added to the methods for clarity. The number of replicates for proteomic analysis have already been mentioned in the methods on line 150, however, we have also added the number of replicates for each experiment in the appropriate sections (e.g. ELISA, immunoblots, and mitochondrial respiration).

Please standardize the description of the number of biological replicates in the figure caption.

We apologize for the inconsistency here, the figure letters and sample sizes have now been standardized across all figure captions.

It is unclear which measurement parameters use the same sample or whether independent experiments were performed, and please provide details in methods part.

We apologize if we do not fully grasp this comment, but we assume that the reviewer is asking whether LCMS analysis was performed on the same samples. To that end, the LCMS analysis of media (secretome) and cell lysate (proteome) was performed on 5-6 replicates from the same experiments.

Specific comments

Line 25  The capitalization is necessary for "Insulin"?

Thank you for pointing this out. The capitalization has been removed.

Line 93 The description “7000g, 10 min” is different from other description (8000g x 10 min).

This typo has been corrected in the manuscript, as both steps were performed at 8000g.

In Figure 1B, the authors should describe which bar is which group in the graph.

We have more clearly shown in colour which bar represents each group.

In Figure 1B, I cannot understand which asterisk indicates a significant difference between the groups, please show it as Figure 1C.

All statistics are now shown in the same manner as Figure 1C. We apologize for the inconsistency.

For the pAkt/totalAkt data in Figure 1B, please note that the bars in the Insulin-stimulated group omit the middle value.

This has been corrected.

Line 104 Please rewrite to cellular proteome as it is easily confused.

This has been added.

Line 128 27%; can be 27%. ???

This has been corrected.

Line 238 It may be Figure 2B, not Figure 2A.

We thank the reviewer for noticing this detail, which has now been corrected.

Figure 2C is not explained in the results section.

We apologize for the oversight. This has been added to the manuscript at lines 230-231.

Line 241-246 If serum-deprivation influences apoptosis, apoptosis markers such as cleaved caspase-3 should be measured in cell lysis. Generally apoptosis is triggered by muscle differentiation (https://doi.org/10.1007/s10495-013-0922-7); therefore, it may be reasonable to include intracellular protein in the condition medium. The LDHA data only indicate that there was no difference in the effect of cell death between the groups. If the authors want to demonstrate the effect of serum-deprivation, the authors must compare between the presence and absence of serum in the medium.

Ideally, serum would be present for the entirety of the experiment, however even a small amount of serum contaminants during MS runs severely compromises peptide identification. This is an inherent limitation to label-free MS-based secretomics. Ultimately, we are not in disagreement that some intracellular proteins would be expected in the media, but we are more concerned with whether differences in the relative abundance of secreted proteins arise from differences in cell death/membrane damage across treatments.  While we cannot directly test this, the LDH data suggests that protein leakage is not different between groups.

Line 249-255 The correlation between secretome and cellular proteome does not indicate whether intracellular protein was released by cell death or not. Also, since the database is used to filter out secreted proteins, it may not be a problem if the secretome contains intracellular proteins.

We thank the reviewer for raising an important point, and we are in agreement that these data cannot be used to interpret changes to cell death or protein release consequent to cell death. These claims have been removed, and this paragraph has been reworked to avoid any potential misinterpretation. We have kept the figure to demonstrate that secreted proteins displayed higher intensities in the secretome vs. proteome as a proof-of-concept that these filters likely identify bona fide secreted proteins.

Line 255-257 Please describe in the Methods part how you calculated the correlation coefficient. Also, please describe in the Methods part how you determined that there are more media than cells, and explain this in the Results. In addition, since the sentences are difficult to understand, please rewrite them in concise sentences.

The correlation coefficient was calculated using the Pearson correlation tool built-in to the Perseus software package. This statement is not intended to convey differences in the absolute abundance or number of secreted proteins, but rather the LFQ intensity of peptides common to both the secretome (media) and proteome (lysate) samples. We feel that the figure effectively shows that these intensities are qualitatively higher in the secretome, however nothing further may be concluded from this.

Line 262 Please replace "This study" with "That previous study/above study" to avoid confusion.

This has been clarified.

Line 275 The sentences may be mistaken.

This has been clarified.

Line 303-304 Please describe all details of the analysis methods used in this experiment, including the method of PCA analysis, in the methods section.

The theory and detailed description for the PCA algorithm included in the Perseus software package can be found in the published Nature Methods article PMID: 27348712, which we have cited on line 181 in Methods. As described in the Nature Methods paper, Perseus performs PCA analysis based on singular value decomposition (PMID: 10963673), a form that computationally performs well on high-dimensional data. We have added a sentence to the manuscript (line 178) to clarify that PCA analysis was also performed using this software.

Line 366-368 There is n = 4-5, can't you do a statistical analysis?

We apologize if this was unclear, however this figure is indeed a representation of the paired statistical comparison (i.e. Ctrl/AICAR, Ctrl/insulin) that was performed using a student’s t-test, as outlined in the Methods. This has been emphasized in the Figure 4 legend, at line 457.

Line 368-378 In the data of Figure 4E-F, please describe which analysis method was used (e.g., GO or KEGG). What does Figure 4E-F show? The p-value on the horizontal axis is positive on both sides. In pathway enrichment analysis, it usually analyzes how many GO terms are tagged to variable proteins and does not compare the increase or decrease of GO terms.

Indeed, the pathway analysis in Figure 4E-F describes proteins that significantly increased or decreased in the secretome, and were enriched in various annotated pathways (including GO and KEGG terms, as well as UniProt keywords). These bars do not denote positive or negative p-values, and this has been clarified in the Methods under Statistics. Additional information regarding the annotations used in Figure 4E-F as well as specific p-values and enrichment factors can be found in Table S2. For clarity, we have also added headers to indicate that the bars represent proteins that “increased” or “decreased” within these pathways.

Line 400-402 The details of the experiment with the addition of culture medium are not described in the Methods part.

This experiment is described under “Mitochondrial Respiration” in the methods, however further details have been added for clarity at lines 169-171.

For the experiments in Figure 5D-G, why is the number of biological replicates large?

We apologize for the lack of clarity, as this is meant to indicate the number of wells on the Seahorse plate per treatment, however this has been modified to “Data shown are means ± SEM from 2 independent experiments with 14-16 replicates/group in each experiment”. We feel this is less confusing for the reader. In addition, we have added the number of replicates for this experiment under the Mitochondrial Respiration section of the Methods.

Round 2

Reviewer 3 Report

Thanks for the revisions. However, some further revisions are necessary.

Line255-260 This sentence can be interpreted to mean that there was no effect of cell death in this study. The data can only show that there was no effect of cell death between treatments. Whether serum-deprivation influences apoptosis/cellar damage is the limitation of this study.

Line 167-168 Does this mean that you measured n=14-16 on one plate, and then n = 14-16 on another plate, for a total of n=27-30? Or does it mean that you measured n =14-16 with a technical duplicate? If the former is true, then the notation of n=27-30 is acceptable.

Author Response

Thanks for the revisions. However, some further revisions are necessary.

Comment: Line255-260 This sentence can be interpreted to mean that there was no effect of cell death in this study. The data can only show that there was no effect of cell death between treatments. Whether serum-deprivation influences apoptosis/cellar damage is the limitation of this study.

Response : Thank you, we agree and this has been re-worded.

Comment: Line 167-168 Does this mean that you measured n=14-16 on one plate, and then n = 14-16 on another plate, for a total of n=27-30? Or does it mean that you measured n =14-16 with a technical duplicate? If the former is true, then the notation of n=27-30 is acceptable.

Response: We apologize for not being clearer. The first is correct, in that we performed the mitochondrial respiration experiment with n=14-16, plated new cells and then performed an additional independent experiment with n=14-16. We have again adjusted the notation to n=27-30 in the Methods and Figure legend, but left the clarification that two independent experiments were performed.

New text is highlighted with the track changes.