ETNK1 mutations induce a mutator phenotype that can be reverted with phosphoethanolamine

Recurrent somatic mutations in ETNK1 (Ethanolamine-Kinase-1) were identified in several myeloid malignancies and are responsible for a reduced enzymatic activity. Here, we demonstrate in primary leukemic cells and in cell lines that mutated ETNK1 causes a significant increase in mitochondrial activity, ROS production, and Histone H2AX phosphorylation, ultimately driving the increased accumulation of new mutations. We also show that phosphoethanolamine, the metabolic product of ETNK1, negatively controls mitochondrial activity through a direct competition with succinate at mitochondrial complex II. Hence, reduced intracellular phosphoethanolamine causes mitochondria hyperactivation, ROS production, and DNA damage. Treatment with phosphoethanolamine is able to counteract complex II hyperactivation and to restore a normal phenotype.

These studies investigate the mechanisms of cellular dysregulation caused by catalytically inactivating mutations in ETNK1 in leukemias. Although these mutations are recurrent in aCML and CMML, very little is known about the mechanisms by which they contribute to oncogenesis, making these studies of biological interest. Experiments showing altered mitochondrial activity, ROS, and H2AX levels with ETNK mutations and restoration of appropriate levels of activity upon P-Et supplementation are compelling and mechanistically interesting. The finding of increased H2AX levels in ETNK1 mutant aCML samples adds nice clinical relevance to this study. It is very interesting the ETNK1 mutation are early events in disease evolution, however it is hard to evaluate the validity of these claims based on the data presented. There are some improvements that need to be made with respect to statistical analysis and data interpretation, but otherwise it is an interesting and novel story.
Major comments-1. Lines 132-135: I would be reluctant to point out changes in gene expression that are not significant. Furthermore, I'm not sure what biological relevance such a small difference in expression of a biosynthetic enzyme would have. In general, it takes fairly large changes to enzyme abundance to alter function. This portion of the manuscript could be omitted without significantly altering the conclusions.
2. Statistical analyses in many figures with multiple groups use a t-test. This may not be an appropriate test in the setting of multiple comparisons. The authors should consult a statistician, but an Anova with a post-test may be a more appropriate analysis.
3. For figure 4B the changes in oxoG seem very subtle with the exception of chr18…honestly, I'm a bit surprised that the differences are statistically significant and seem subtle relative to the change in level of dsDNA damage. Can the authors provide some additional information about the global magnitude of this change? Are there potential oxoguanine-independent mechanisms by which dsDNA damage could be occurring downstream of mutant ETNK1? 4. Line 316-322. I see the data for the bulk based sequencing to assign the mutations with oncogenic potential in supplemental tables 12-16, but don't see a reference to where the data from the sequencing of individual clones is presented used to assign the hierarchies depicted in 6A-D. It seems like this data should be in the supplement, but I am unable to find it. How does the reconstruction by mutational analysis of individual colonies compare to the assignment that would have been made based on VAF?
Minor-5. Line 53-synthesize (spelling) 6. Lines 71-75-run on sentence 7. Figures 1C-E. While it is easy to appreciate that there are areas of low electron density in the mutants it is not easy to appreciate that the WT mitochondria have a rounder shape. Maybe just omit that statement, or alternatively present a measurement of circularity. I believe that the level of magnification is consistent between the images, but the scale bar is a bit challenging to see (especially in E), perhaps make the scale bar more prominent.
8. Line 124 and 138. I don't understand what is meant by vicariate in this particular context. 9. Figure S3. Can the font size be increased on the legends? What do the green bars represent? 10. Figure S4. A brief description of the method used in the figure legend would be helpful.
11. Line 203. It would be more accurate to say that it restored to near normal levels 12. Line 218. I would soften the statement that it "almost completely restored the normal ROS levels" perhaps to "reduced ROS levels to an intermediary level" or something similar. The levels of ROS are still fairly elevated relative to baseline.
13. Line 243. A sentence or two explaining the 6-TG assay would be helpful here.
14. Line 250. I may just be misreading this section, but if the 6-TG resistance assays were performed with P-ET and tigecycline as new assays from those described in line 245 they should be shown separately and not as a combined graph ie untreated controls should be from the same experiment. It looks like the control data is identical between supplemental figure 4 and 5…so I can't tell why they are presented as two different tables, unless the untreated is being shown for reference, but was not part of the second experiment.
15. Line 266. What is the evidence for aCML being characterized by chromosomal instability? A reference would be useful here.
16. 366-374 The discussion of the time involved to do functional studies on newly identified mutations does not seem necessary to me.
17. 396-397. I believe that it would be more correct to say "dominant" than "dominant negative". Dominant negative implies that the mutant protein interferes with the activity of WT ETNK1, which I don't think has been demonstrated here.
Reviewer #3 (Remarks to the Author): Summary of the work: The authors have evaluated the effect of mutations in the kinase ETNK1 on mitochondrial activity in this paper. Using CRISPR KO and knock-in 293 HEK cell lines, they show that loss of ETNK1 has no significant effect on phospholipid levels in the mitochondria or in the cell membrane, suggestive of mechanisms compensating for loss of ETNK1 enzymatic activity. However, they note that ENTK1 loss leads to increased mitochondrial mass, ROS production and Oxo-DG and gH2AX, which is hypothesized to be due to a regulatory mechanism of mitochondrial complex II by phosphoethanolamine. Importantly, they demonstrate that the increased ROS and Oxo-DG in ETNK1 mutant cells is reversible by p-ET addition. Overall I found the discoveries to be interesting, and relevant/ topical. However, the findings need to be defined further in a suitable cellular model, the link between ENTK1 loss and DNA damage needs to be clarified, and the presentation of data needs to be significantly improved in its rigour/ attention to methodological detail.
Major comments: 1. The majority of the experiments are based on ETNK1 knockout/knockin in 293 HEK cells. Given the significant differences in phenotype of mutations in distinct cellular contexts, I think it is required to carefully validate all critical findings in a leukemic or haematopoietic cell line, to justify the conclusion of ETNK1 regulating leukemogenesis through metabolism. The overexpression experiments in TF1 cells suggest that mutant ETNK1 can act as a dominant negative, but there is no data presented on the extent of overexpression that leads to this effect (the mutations in aCML are noted to be heterozygous). 2. The cell line models are not adequately validated. Do the authors see a change in p-ET in their 293 KO/ knockin cells, similar to their previous observations? These data should be presented along with actual FISH analysis, sequencing, and western blots to show validation of the knockout and knockin in the 293 cell line model (along with validation of the overexpression system in the haematopoietic line). 3. For all analysis represented in the paper, the authors are requested to specify the imaging/ quantification technique used, and the number of cells analyzed per experiment. Eg. Mitochondrial mass estimation has been performed with mitotracker green, but the methods and figure legend do not elaborate how the analysis was performed. The authors should also provide a table with all the antibodies used (clone/ product numbers), and concentration of primary antibody for each application. 4. ETNK1 mutations are hypothesized to reduce p-ET levels, leading to reduced "brakes" on SDH activity, and therefore increased ROS production from complex II, and subsequent DNA damage. Is there evidence of increased complex II activity in lysates from the ETNK1 KO/ knockin cells? Do these cells show evidence of altered respiration in a seahorse assay? 5. If ROS generation is key to ETNK1 mutations in leading to genomic instability, does reduction of ROS reverse it? Would quenching of ROS (eg. by N-acetyl cysteine) reduce the amount of Oxo-DG and/or gH2AX? This would be pertinent given recently described links between SDH mutations and impaired homologous recombination, as alternate means of leading to elevated gH2AX/ genomic instability. 6. The section on ETNK1 being an early/ truncal event in leukemogenesis does not fit into the current flow of the paper. It could serve as an introductory section if required, or could be removed from this paper.
Specific comments: 1. The Figure 1 title reads "Mitochondria morphology and activity". The authors use mitotracker green/red ratio for mitochondrial potential. As this is a key message of the paper, it would be good to cross-validate these findings with an orthogonal assay such as TMRE or JC1 fluorescence, in both the CRISPR KO and overexpression system. Positive and negative controls are important to include for these experiments, for example FCCP/CCCP treatment to demonstrate depolarization ( Figure 1 and 2); ROS quenchers and inducers ( Figure 3B). 2. In Figure1 (panel C, D, E) authors show TEM images of mitochondria suggesting an abnormal mitochondrial morphology in ETNK1 mutant and KO cell lines. Are these accompanied by/ due to changes in cellular morphology? 3. In figure 2 (and throughout the paper), please report actual p-values when t-tests are done. How many cells were analyzed for each experiment and using what method? 4. How does the in-vitro dose of Tigecycline of 2.5 microM compare to the pharmacokinetics of Tigecylcine dosage in humans? 5. In figure 3, why were only WT and KO cells (and not the mutant) chosen for the O2 consumption experiments? How many cells were analyzed per experiment for the CellRox experiments? How were the imaging data quantified? 6. Does ETNK1 loss/ mutation lead to increased overall OxoDG staining (as assessed by immunofluorescence)? For the 6TG assay ( Figure 4D); please show representative images in the main or supplementary data 7. gH2AX is smaller than Actin, and ideally should be presented below it in a western blot figure. The quality of the blots is not satisfactory (especially for in-vitro samples). 8. Figure 5B is not representative at present. Please highlight an area from figure 5A and enlarge for clearer view. 9. How were three replicates performed for figure S10 and 5I (single patient sample)? Were they performed at three different occasions or processed simultaneously? How was the p-value calculated here? 10. "side pathways" could be phrased better Anand D Jeyasekharan National University of Singapore We are glad that the three Reviewers found merits in our work and we did our best to address their points. The point-by-point answers to the Reviewers are reported here below.

Answers to Reviewer #1
Fontana et al., have evaluated the functional effects of ETNK1 mutations. ETNK1 regulates the production of the phospholipid P-Et. Knockout and mutations of ETNK1 lead to increased mitochondrial mass, increased oxygen consumption and ROS production. ETNK1 inhibition should lead to decreased P-Et and supplementing respiratory complexes with P-Et impaired respiratory chain complex II activity. The authors report the novel finding that P-Et impairs complex II activity likely be competing with succinate for binding to SDH. This work sheds new light on the regulation of complex II. I have the following comments about the manuscript:

1) The vast majority of experiments were conducted in 293 cells and only a few experiments were validated in a leukemia cell line. It is unclear how the findings from these cell lines would translate to patient samples. Specifically, it is unclear whether primary samples with ETNK1 mutations or deletions have metabolic abnormalities and alterations in P-Et levels. It is also unknown whether they have abnormalities in mitochondrial and respiratory chain complex function. As such, the relevance of their findings to human disease is unclear.
To address the appropriate Reviewer's concerns, we extended the entire analysis to a second cell line (TF-1), which is hematological and myeloid and, as such, much closer to the cell type involved in the disease than 293 (Fig 2, Suppl Fig S11, S12, S13, S19, S22 and corresponding text). Also, whenever possible taking into account the rarity of the primary samples, we extended the study to primary ETNK1+/-aCML samples (Fig 5, Suppl Fig S8, S20). Specifically, we replicated on the myeloid TF-1 line the following experiments: -mitochondria activity with MitoTracker Red and Green in absence/presence of P-Et; -mitochondria activity with JC1 plus/minus CCCP; -ROS production in absence/presence of P-Et; -H2AX Double-strand DNA damage in the absence/presence of P-Et; -Mitochondria Complex I, II, III and IV activity in the absence/presence of increasing concentration of P-Et; -succinate competition assay for Mitochondria Complex II.

On primary samples, we performed the following new analyses:
-membrane Lipid class composition analysis; All the new experiments went in the same direction as the previous ones, therefore, strengthening the conclusions of the work.
2) The authors did not measure levels of P-Et in their cell lines, Does knockout or mutation of ETNK1 alter P-Et levels? Of note, mutations and knockout of ETNK1 did not alter other lipid levels.
We thank the Reviewer for pointing out this very important question. To address this point, we extended our previous analysis in order to measure: 1) intracellular P-Et levels (Suppl Fig S10); 2) total lipid levels (Suppl Fig S5); and 3) detailed lipid composition, with specific but not exclusive emphasis on phosphatidylethanolamines (Suppl Fig S6).

3) What are the lipid levels in patient samples with this mutation?
In line with the points discussed in (2), we extended the same analysis to patient samples. We found that, in line with the cell models, lipid levels don't change in ETNK1+ vs ETNK1patients (Suppl. Fig. S8).

4) A potential implication of their work is that patients with ETNK1 should be treated with P-
Et to restore normal mitochondrial function. However, without mouse models and primary samples, this question has not been tested.
We totally agree with the Reviewer. This work, however, is intended as a first study focused aimed at inferring the molecular mechanisms responsible for the ETNK1-mediated prooncogenic process occurring in aCML patients. The potential of P-Et as a therapeutic agent, which emerged here, will be the subject of future investigations using in vivo models. Remarkably, these models are quite complex, given that the role we propose for mutated ETNK1 is to induce a very peculiar 'mutator phenotype' in the target cells and the definition of a valid readout for such a model is not trivial. The translational potentialities of the study will be addressed in the near future.

5) The authors treat cells with mM concentrations of P-Et to see cellular effects. Are these concentrations physiologically relevant?
It is very difficult to say, as the data available in the literature are very limited. Recent data generated in mouse (Cell Rep. 2019 Oct 1;29(1):89-103.e7. doi: 10.1016/j.celrep.2019 suggest that those concentrations may be physiologically relevant.  Fig. S2; p>0.05), while it is four-fold less expressed in the knock-out line compared to the WT one (Suppl. Fig. S2; p=0.0033). This finding is compatible with an expected nonsense-mediated RNA decay occurring in the knock-out.

6) The authors treat complexes with uM concentrations to see changes in activity. But, changes in cells require mM concentrations. Can the authors be certain that the effects on mitochondrial function are only due to alterations in respiratory chain complex II
8) As a minor comment, the manuscript results and figure legends should specify which cells were used. One had to refer to the methods to determine the experiments were performed in 293 cells.
Likewise, times of incubation should be noted in the text.
In order to improve the manuscript readability, we modified the text and legends by specifying both the cell line used and the time of incubation of all the experiments.

Answers to Reviewer #2 (Remarks to the Author):
These studies investigate the mechanisms of cellular dysregulation caused by catalytically inactivating mutations in ETNK1 in leukemias. Although these mutations are recurrent in aCML and CMML, very little is known about the mechanisms by which they contribute to oncogenesis, making these studies of biological interest. Experiments showing altered mitochondrial activity, ROS, and H2AX levels with ETNK mutations and restoration of appropriate levels of activity upon P-Et supplementation are compelling and mechanistically interesting. The finding of increased H2AX levels in ETNK1 mutant aCML samples adds nice clinical relevance to this study. It is very interesting the ETNK1 mutation are early events in disease evolution, however it is hard to evaluate the validity of these claims based on the data presented. There are some improvements that need to be made with respect to statistical analysis and data interpretation, but otherwise it is an interesting and novel story. We thank the Reviewer for providing this valuable suggestion. We decided to completely omit this section of our work, accordingly.
2. Statistical analyses in many figures with multiple groups use a t-test. This may not be an appropriate test in the setting of multiple comparisons. The authors should consult a statistician, but an Anova with a post-test may be a more appropriate analysis.

We agree that multiple group tests should be addressed using Anova rather than individual t-test. We modified the analyses, figures and text accordingly.
3. For figure 4B the changes in oxoG seem very subtle with the exception of chr18…honestly, I'm a bit surprised that the differences are statistically significant and seem subtle relative to the change in level of dsDNA damage. Can the authors provide some additional information about the global magnitude of this change? Are there potential oxoguanine-independent mechanisms by which dsDNA damage could be occurring downstream of mutant ETNK1?
The limited difference is, in the opinion of the authors, in large part explained by the extremely volatile nature of the oxoG signal. When oxoG is generated at gDNA level, the error must be corrected very quickly, before the onset of the next cell cycle, in order to avoid the irreversible incorporation of the error in one of the daughter cells. 4. Line 316-322. I see the data for the bulk based sequencing to assign the mutations with oncogenic potential in supplemental tables 12-16, but don't see a reference to where the data from the sequencing of individual clones is presented used to assign the hierarchies depicted in 6A-D. It seems like this data should be in the supplement, but I am unable to find it. How does the reconstruction by mutational analysis of individual colonies compare to the assignment that would have been made based on VAF?
We thank the Reviewer for highlighting this problem, however, following the specific requests of Reviewer #3, we completely removed this part of our study.

Figures 1C-E. While it is easy to appreciate that there are areas of low electron density in the mutants it is not easy to appreciate that the WT mitochondria have a rounder shape. Maybe just omit that statement, or alternatively present a measurement of circularity. I believe that the level of magnification is consistent between the images, but the scale bar is a bit challenging to see (especially in E), perhaps make the scale bar more prominent.
We thank the Reviewer for this suggestion. We added a measure of circularity, which confirms that WT mitochondria have a rounder shape. We also modified the scale bar of the figure in order to improve readability.

Line 124 and 13I don't understand what is meant by vicariate in this particular context.
We rephrased to clarify our statement.
9. Figure S3. Can the font size be increased on the legends? What do the green bars represent?
Bars are now more visible and font size increased.
10. Figure S4. A brief description of the method used in the figure legend would be helpful.
A brief description was added to the previous legend.

Line 203. It would be more accurate to say that it restored to near normal levels
The text was modified as suggested.
12. Line 218. I would soften the statement that it "almost completely restored the normal ROS levels" perhaps to "reduced ROS levels to an intermediary level" or something similar. The levels of ROS are still fairly elevated relative to baseline.
We agree. Text was modified as suggested.
13. Line 243. A sentence or two explaining the 6-TG assay would be helpful here.
We thank the Reviewer for this suggestion. A description of the 6-TG assay was added.

Line 250. I may just be misreading this section, but if the 6-TG resistance assays were performed with P-ET and tigecycline as new assays from those described in line 245 they should be shown separately and not as a combined graph ie untreated controls should be from the same experiment. It looks like the control data is identical between supplemental figure 4 and 5…so I can't tell why they are presented as two different tables, unless the untreated is being shown for reference, but was not part of the second experiment.
This is correct: we performed a single assay with all the lines and treatments in a single experiment. To avoid any possible confusion we rephrased the entire paragraph.

Line 266. What is the evidence for aCML being characterized by chromosomal instability? A reference would be useful here.
A reference was added (Blood. 2014 Apr 24;123(17):2645-51).

366-374 The discussion of the time involved to do functional studies on newly identified mutations does not seem necessary to me.
The paragraph focused on functional studies was deleted.

396-397. I believe that it would be more correct to say "dominant" than "dominant negative". Dominant negative implies that the mutant protein interferes with the activity of WT ETNK1, which I don't think has been demonstrated here.
In agreement with the Reviewer, "dominant negative" was changed to "negative"..

Answers to Reviewer #3
Summary of the work: The authors have evaluated the effect of mutations in the kinase ETNK1 on mitochondrial activity in this paper. Using CRISPR KO and knock-in 293 HEK cell lines, they show that loss of ETNK1 has no significant effect on phospholipid levels in the mitochondria or in the cell membrane, suggestive of mechanisms compensating for loss of ETNK1 enzymatic activity. However, they note that ENTK1 loss leads to increased mitochondrial mass, ROS production and Oxo-DG and gH2AX, which is hypothesized to be due to a regulatory mechanism of mitochondrial complex II by phosphoethanolamine. Importantly, they demonstrate that the increased ROS and Oxo-DG in ETNK1 mutant cells is reversible by p-ET addition. Overall I found the discoveries to be interesting, and relevant/ topical. However, the findings need to be defined further in a suitable cellular model, the link between ENTK1 loss and DNA damage needs to be clarified, and the presentation of data needs to be significantly improved in its rigour/ attention to methodological detail.
Major comments: 1. The majority of the experiments are based on ETNK1 knockout/knockin in 293 HEK cells. Given the significant differences in phenotype of mutations in distinct cellular contexts, I think it is required to carefully validate all critical findings in a leukemic or haematopoietic cell line, to justify the conclusion of ETNK1 regulating leukemogenesis through metabolism. The overexpression experiments in TF1 cells suggest that mutant ETNK1 can act as a dominant negative, but there is no data presented on the extent of overexpression that leads to this effect (the mutations in aCML are noted to be heterozygous).
We thank the Reviewer for this important observation, which was made also by the Reviewer #1. Indeed, the use of a single, non hematological cell line may lead to unreliable results. To overcome this limitation, in the revised version of our work we extended the use of the myeloid TF-1 cell line to all the experiments, confirming our previous findings (Fig 2, Suppl Fig S11, S12, S13, S19, S22 and corresponding text). Also, whenever possible taking into account the rarity of the primary samples, we extended the study to primary ETNK1+/-aCML samples (Fig 5, Suppl Fig S8, S20). Specifically, we replicated on the myeloid TF-1 line the following experiments: -mitochondria activity with MitoTracker Red and Green in absence/presence of P-Et; -mitochondria activity with JC1 plus/minus CCCP; -ROS production in absence/presence of P-Et; -H2AX Double-strand DNA damage in absence/presence of P-Et; -mitochondria Complex I, II, III and IV activity in absence/presence of increasing concentration of P-Et; -succinate competition assay for Mitochondria Complex II.

On primary samples we performed the following new analyses:
-membrane Lipid class composition analysis; -ROS production in absence/presence of P-Et; -H2AX Double-strand DNA damage in absence/presence of P-Et.   Fig. S2; p>0.05), while it is four-fold less expressed in the knock-out line compared to the WT one (Suppl. Fig. S2; p=0.0033). This finding is compatible with an expected nonsense-mediated RNA decay occurring in the knock-out.
3. For all analysis represented in the paper, the authors are requested to specify the imaging/ quantification technique used, and the number of cells analyzed per experiment. Eg. Mitochondrial mass estimation has been performed with mitotracker green, but the methods and figure legend do not elaborate how the analysis was performed. The authors should also provide a table with all the antibodies used (clone/ product numbers), and concentration of primary antibody for each application.
We thank the Reviewer for this observation. We modified manuscript and figure legends accordingly. A table summarizing name, clone, brand, code, and dilution of used antibodies for each assay was added in the revised version (Suppl. Tab. S15).
4. ETNK1 mutations are hypothesized to reduce p-ET levels, leading to reduced "brakes" on SDH activity, and therefore increased ROS production from complex II, and subsequent DNA damage. Is there evidence of increased complex II activity in lysates from the ETNK1 KO/ knockin cells? Do these cells show evidence of altered respiration in a seahorse assay?
The altered respiration, as mitochondrial respiratory capacity and oxygen consumption (OCR), was measured in ETNK1-WT and ETNK1-KO intact cells using high-resolution respirometry Oroboros O2K ( Figure 3A). While the argument is definitely complex and difficult to address, the complementary alteration of glucose uptake and lactate production (ECAR) Fig. S18). Globally, we think that this new data significantly strengthen the proposed model.
6. The section on ETNK1 being an early/ truncal event in leukemogenesis does not fit into the current flow of the paper. It could serve as an introductory section if required, or could be removed from this paper.
We thank the Reviewer for pinpointing this problem. We removed this part of the manuscript, as suggested.
Specific comments: 1. The Figure 1 title reads "Mitochondria morphology and activity". The authors use mitotracker green/red ratio for mitochondrial potential. As this is a key message of the paper, it would be good to cross-validate these findings with an orthogonal assay such as TMRE or JC1 fluorescence, in both the CRISPR KO and overexpression system. Positive and negative controls are important to include for these experiments, for example FCCP/CCCP treatment to demonstrate depolarization (Figure 1 and 2); ROS quenchers and inducers ( Figure 3B).
This is an important point, as the alteration of mitochondrial potential in ETNK1-mutated samples is one of the key elements of this work. Following the Reviewer's suggestions, we cross-validated our findings by adding an orthogonal assay based on JC-1, which we tested in both 293 and TF-1 lines (Suppl. Fig. S3, S12).. Also, we added the missing positive and negative controls (Suppl. Fig. S14, S15), thanks for highlighting. All the new experiments confirmed our previous findings, thereby strengthening our original hypothesis.

In Figure1 (panel C, D, E) authors show TEM images of mitochondria suggesting an abnormal mitochondrial morphology in ETNK1 mutant and KO cell lines. Are these accompanied by/ due to changes in cellular morphology?
We didn't notice any relevant change in cell morphology in ETNK1-WT, mutated, or KO lines. This likely suggests that the effect is restricted to mitochondria.
3. In figure 2 (and throughout the paper), please report actual p-values when t-tests are done. How many cells were analyzed for each experiment and using what method?
P-values and statistical tests are now explicitly reported in the manuscript, as well as the number of cells used for each experiment.

In figure 3, why were only WT and KO cells (and not the mutant) chosen for the O2 consumption experiments? How many cells were analyzed per experiment for the
6. Does ETNK1 loss/ mutation lead to increased overall OxoDG staining (as assessed by immunofluorescence)? For the 6TG assay ( Figure 4D); please show representative images in the main or supplementary data We tried to perform OxoDG staining in immunofluorescence with antibody Anti-8-Oxoguanine Antibody, clone 483.15 (MAB3560; Sigma-Aldrich) but unfortunately we failed to produce good quality results. We now show the images of the 6TG experiments in the supplementary material (Suppl. Fig. S16) as suggested.
7. gH2AX is smaller than Actin, and ideally should be presented below it in a western blot figure. The quality of the blots is not satisfactory (especially for in-vitro samples).
We shifted the position of H2AX/Actin in both panels of figure 5 (panels D and H) and we improved the quality of the image, as requested.
8. Figure 5B is not representative at present. Please highlight an area from figure 5A and enlarge for clearer view.
To improve the readability of figure 5B, we generated new, high resolution single cell images of both WT and KO lines.
9. How were three replicates performed for figure S10 and 5I (single patient sample)? Were they performed at three different occasions or processed simultaneously? How was the pvalue calculated here?
They were three independent replicates processed at the same time point. The p-value was calculated using GraphPad Prism6 software, comparing primary bone marrow cells in the absence and presence of phosphoethanolamine 1mM.
10. "side pathways" could be phrased better Changed to "alternative pathways".
We sincerely hope we addressed all Reviewer' points and the revised manuscript is now suitable for publication in Nature Communications.