MYCN-driven fatty acid uptake is a metabolic vulnerability in neuroblastoma

Neuroblastoma (NB) is a childhood cancer arising from sympatho-adrenal neural crest cells. MYCN amplification is found in half of high-risk NB patients; however, no available therapies directly target MYCN. Using multi-dimensional metabolic profiling in MYCN expression systems and primary patient tumors, we comprehensively characterized the metabolic landscape driven by MYCN in NB. MYCN amplification leads to glycerolipid accumulation by promoting fatty acid (FA) uptake and biosynthesis. We found that cells expressing amplified MYCN depend highly on FA uptake for survival. Mechanistically, MYCN directly upregulates FA transport protein 2 (FATP2), encoded by SLC27A2. Genetic depletion of SLC27A2 impairs NB survival, and pharmacological SLC27A2 inhibition selectively suppresses tumor growth, prolongs animal survival, and exerts synergistic anti-tumor effects when combined with conventional chemotherapies in multiple preclinical NB models. This study identifies FA uptake as a critical metabolic dependency for MYCN-amplified tumors. Inhibiting FA uptake is an effective approach for improving current treatment regimens.


Reviewer #1 (Remarks to the Author):
The paper by Ling Tao et al. is a comprehensive metabolomic study involving both cell and mouse models with overexpressed or knocked-down MYCN, as well as patient samples, to identify a mode of action of MYCN in neuroblastoma. Myc has previously been linked to lipid and fatty acid metabolism in other oncologies but here a highly specific FA transporter is identified and shown to be a pharmacological target, with an inhibitor demonstrating efficacy in orthotopic patient-derived tumour models.
It is generally well written and the data clearly presented.
There are minor modificatons required prior to publication.
Line 127 "in two MYCN-induced NB cell line models and in NB primary tumors" One model is Tet-ON Dox-induced, while the other here is conditionally knocked down by Dox-inducible shRNA expression. Perhaps a better way to describe them therefore is "conditionally-expressed MYCN", or other way to reflect that they are not MYCN-induced lines per se.
Line 194 "These findings were validated in a second MYCN system, SK-N-AS MycN-ER." Since this is a new assay (lipid droplets) being introduced to extend the FA phenotype to an additional cell type, why isn't this assay performed in parallel with with MYCN3-Tet-ON cells to show that they respond similarly?

Fig 2A
This appearance of lipid droplets after MYCN activation contrasts with the report from Marie Arsenian Henriksson where such droplets were observed after Myc inhibition in neuroblastoma (Zirath et al. 2013). Please briefly discuss this discrepancy.

Fig 2D
While the effects of FA supplementation on the MNA cells vs. SHEP are clear, for the SK-N cells with or without activated MycN-ER it is more subtle (although significant). It is a little confusing to represent the same data in a different format here (Fig 2D right panel vs. the other panels), and perhaps drawing the data in the same way may show how the MycN-ER cells are more similar to the MNA cell lines, whereas without Myc-ER they are more similar to SHEP cells (therefore, plot the right panel of Fig 2D in 2 separate line graphs, not as single bar graph).
Line 286 "supporting that NB cells highly depends" The grammar requires correction here.

Fig 4J
There is no legend for the blue and black bars (in fact, they are coloured incorrectly and should be red, presumably).
Line 425-6 "These effects can be rescued by FA supplementation." From Fig 2D, it seems that there is only a partial rescue, and in fact in IMR32 and MYCN-ER cells, it is a relatively small effect. This should be reflected in the text here (since as written it implies a complete rescue) plus one could speculate as to why there is not a greater rescue with FA supplementation. This extends to Line 230 too ("Collectively, our data indicate that MNA-induced cell growth critically depends on exogenous FAs."). Again, if it is critically dependent, why isn't the rescue from FA supplementation more notable? It is a modest -although significant -effect in Fig 2E ( This is a very interesting and ambitious study proposing a new approach for neuroblastoma therapy based on the identification of a novel metabolic target . However, there are several concerns that should be addressed before acceptance.
Major concerns: 1. Please show WB on MYCN, proliferation, and apoptosis markers after treatment with the inhibitors in the cell lines. In addition, authors only show changes in MYCN expression in LAN5 and MYCN3 in SFigure 1A, but not in any other of their cell lines after induction/ repression of MYCN. 2. The authors have analyzed lipid accumulation in some of their experiments. It has previously been shown that inhibition of MYCN results in lipid accumulation see for instance Zirath PNAS, 2013. How does the lipid accumulation occurring after CB5 treatment correspond to the lipid accumulation caused by MYCN inhibition? 3. Figure 4 and legend. Photos of LAN5 and SHEP cells are shown after BOIDPY. It is obvious by eye that the fluorescence is decreasing more in the LAN5 cells but is also clearly decreasing in the SHEP cells. However, in the graph is looks more or less as if there is no decrease in the SHEP cells. In addition, it is indicated that * means significant changes after CB5 treatment and that # indicates significant differences between LAN5 and SHEP cells at same CB5 dose. Please clarify. The use of # is unusual.
4. Lines 407-410 "Our metabolomics data show reduced glutamine and glutamate in MYCNoverexpressing cells and MNA tumors, suggesting enhanced consumption of glutamine and glutamate. Future research should include in vivo imaging and flux studies to better understand MYCN reprogramming of glutamine metabolism." Other studies have demonstrated that MYCN de novo glutamine synthesis. Please mention and give reference. 5. The authors confirm that SLC27A2 is required for NB survival performing shRNA in LAN5 cells. These results should be validated in a second MNA NB cell line. Moreover, to prove that it is a target of MYCN and not c-MYC, authors should also prove that shSLC27A2 does not affect viability in non-MYCN NB cells. 6. Author demonstrated lipid droplet accumulation as a result of increased fatty acid uptake and synthesis was demonstrated in SK-N-AS MYCN-ER cells. Authors should validate that this is a result due to MYCN, by repeating this experiment in another cell line, for instance the MYCN3, in which they performed the tracing experiments. This will prove that lipid accumulation is not due to treatment with 4-OHT which could induce stress. Are there any E-boxes in that region? It would also be useful to check available ChIP-seq data in the ENCODE database and see if c-MYC also binds to the promoter of SLC27A2. 9. The authors show a non-canonical E-box in the promoter of SLC27A2 and show MYCN binding to the promoter. To show that it is a target gene authors need to mutate the E-Box and analyze binding by ChIP. Furthermore, both wt and mutant E-box-reporters, for instance luciferase, and assays should be performed to validate transcriptional activation of SLC27A2 by MYCN. 10. The major concern is the lack of biological replicates in many of the in vitro experiments shown in this manuscript. In the materials and methods section, the authors mention that at least six mice are necessary for in vivo experiments to reach an 80% statistical power at p<0.05. But for several of the in vitro experiments, the authors only present two replicates. Statistical analysis cannot be performed with a n<3. Authors should discuss how the statistical analysis was performed and represent experiments with at least three times for significance. It is important to present at least three replicates per experiment with the corresponding statistical analysis. This concerns the following Figures: Fig 2B ( 11. The authors present data from seven orthotopic mouse models: I. LAN5 shCTRL and shSLC27A2 cells were implanted in the renal capsule of NCr nude mice. MRI. Tumor weight. ORO-staining. Concerns re animal experiments: 11A. Please show the actual size of the tumors from the different mouse models so it is possible to compare them. 11B. Figure 5H. Authors have used mice with normal immune system and write that it is important that tumor growth can be reduced by inhibiting SLC27A2. Is it not the growth of the TH-MYCN spheres which is the major achievement in this experiment, ie. that tumors can form even in the setting of a normal immune system rather than that the tumor burden can be reduced? Comparing the graphs in Figure 5F with Figure 5H, respectively, show that the tumors in fact get larger in the syngenic model compared to in the nude mice during the same time frame. Do the authors have any idea why the tumors from the TH-MYCN+/+ cells become larger in the 129x1/svj compared to the NRc nude mice? Please comment on this in the discussion. 11C. The authors have analyzed some of the tumors generated for lipid accumulation but not stained for any other markers ( Figure 4G andFigure 5G). Please show stainings of the tumors at endpoint for at least MYCN, proliferation, apoptosis and maybe also for hypoxia, and angiogenesis markers to understand more about the mechanism for reduced tumor growth. In addition, how did the authors choose the mice to analyze for lipid accumulation? Please explain why you chose to stain for lipids only in the LAN5 shCTRL and shSLC27A2 and in the TH-MYCN +/+ derived allograft model in NCr nude mice. 11D. Why did authors wait 2 weeks in the first and then 5,5 weeks in the experiments with the MNA patient tumors?

II. LAN 5-luciferase cells in
11E. The representation of tumor volume is unclear in Figure 5E. The authors show a significant difference between control and CB5-treated mice, however the graph show tumor volumes similar for both groups. Please, provide the average of each group.
11F. SFigure 5C and SFigure 5D. Why do the mice increase weight during the experiments with treatment with CB5 in TH-MYCN+/+ derived syngenic and in patient derived MNA xenograft models, respectively? Furthermore, no statistics is given in these experiments. The same is true for the experiments shown in SFigure 5A and SFigure 5B.
11G. Authors show that CB5 has minimal effect in non-MNA cells in vitro but they have not proved this in vivo. Treatment of mice implanted with non-MNA NB cells should be performed to further validate the in vitro results. Moreover, NCr nude mice were treated with 25 mg/kg b.i.d, while in the immunocompletent 129x1/svj mice the concentration of CB5 was 30 mg/kg b.i.d. Please, explain why the concentration is higher in the latter. 11H. The In vivo experiments show that the mice do not lose weight as a sign of toxicity derived from inhibition of SLC27A2. In the discussion it is mentioned that normal tissue toxicity is missing while inhibiting FA synthesis. Does inhibition of SLC27A2 affecting organs as the liver or kidneys?
11I. There are important differences between the mice treated with CB5 in Figure 5E but still the differences with the control group are significant at a p-value of 0.006. Please revise and provide a graph with the average and SD of all the mice. The present Figure 5E can be shown in SI.
Minor comments: 1.To avoid any confusions, authors should mention that SLC27A encodes FATP2 earlier in the text, also in the abstract. Along the same line, authors use abbreviations in the results part that are not spelled out. For easier reading of the manuscript is it preferable to spell out the abbreviations in the text rather than in the Figure  3. The authors do not specify the time point in which the IC50 values in Figure 2C were calculated. 4. In line 246, Figure 3B is mentioned while it should be Figure 3C. 5. In Lines 352 and 353, Figure 5J is mentioned but there is no Figure 5J in Figure 5. Please revise. 6. The authors write "Our study provides the first direct evidence that MYCN-driven tumors rely on FA uptake and lipid synthesis for survival, making targeting of FA uptake a promising therapeutic approach for high-risk MNA patients." Again, there are several articles also demonstrating this statement, so the words "first evidence" should be avoided. Authors use similar expressions elsewhere in the manuscript, please do not use "for the first time", "first report" and similar expressions. 7. Lines 178-180. How do the authors explain the mechanism of MYCN promoting FA desaturation? Does MYCN regulate any enzyme involved in that process? 8. Figure 6A-B and Supplementary Figure 6A-B. Please show the cell viability graphs for each cell line and treatment. SHEP cells do not show Caspase 3/7 activity, but we cannot know whether these compounds still affect cell viability. There is the possibility that the selected compounds induce cell death in SHEP by a mechanism other than apoptosis.
9. Figure 4J. The different concentrations are presented as white (0), grey (5 uM) and red (10 uM) CB5, but there are bars in the Figure with blue and black colors. Nothing is neither mentioned about the blue and black bars in the legend. Please advice and revise.
10. The authors are using several different inducible neuroblastoma cells for different assays. Please clarify the reasoning behind using these specific cells for the specific assays and if they have tested also some of the other cell lines in the experiments presented than the ones shown, with similar or with different results.
11. Please use italics when referring to the gene and for in vitro, in vivo, i.p, b.i.d., etc 12. Line 381 "All treatments did not cause significant body weight loss or clinical signs of toxicity." Revise to -> No treatment caused any significant body weight loss or clinical signs of toxicity.
13. Figure Figure 3C, SFigure 3B and SFigure 3C. Please specify the regions and the primers used for the PCRs.
Reviewer #3 (Remarks to the Author): The manuscript by Tao and colleagues describes a multi-omics analysis of MYCN-amplified and nonamplified neuroblastoma (NB) systems, including patient-derived tumour samples. This analysis points towards NMY-dependent changes in glycerolipids, mostly triglycerides (TG) and diacylglycerides (DG). The authors then go on to show that MYCN-amplified NB cells are dependent on lipid uptake as they are more effected by lipid depleted serum or inhibition of fatty acid uptake than non-amplified counterparts. This is then linked to enhanced expression of the fatty acid transporter FATP2 (SLC27A2), which the authors identify as an MYCN-target gene by chromatin-IP and promoter studies. They also analyse public datasets derived from NB patients to show that SLC27A2 expression correlates with MYCN expression/activity and poor patient survival. They then use RNAi-mediated gene silencing to show that FATP2 is required for in vitro and in vivo growth of MYCN-amplified NB cells. They also perform lipidomics analysis of tumour material after SLC27A2 silencing to confirm that TG levels are reduced. In addition, they block the activity of FATP2 using the small molecule compound CB5 (also termed grassofermata in previous studies) to show reduced tumour growth and extended survival in several MYCN-dependent in vivo models. Finally, they show that CB5 synergises with two chemotherapeutic agents (VP16 and temozolomide) both in vitro and in vivo. They conclude that enhanced fatty acid uptake through induction of FATP2 promotes NB growth and therapy resistance.
Overall, the manuscript shows a plethora of data generated across multiple in vitro and in vivo systems. It makes a strong case for enhance fatty acid uptake in MYCN-amplified NB cells and tumours and provides evidence for treatment resistance, albeit without deeper mechanistic insight. There are a number of points that should be addressed before this manuscript can be considered for publication. 1) Several of the bar graphs shown in the Figures do not show error bars in the controls (i.e. Fig. 2E, 3B  etc). Variability of the controls should also be displayed.
2) The different sensitivity of SK-N-AS NMYC-ER cells to medium lipid depletion is quite small. The cells without MYC activation are already quite sensitive to lipid depletion (similar to LAN5 cells). Is this explained by a leakiness in the induction system? How would parental SK-N-AS cells respond? Given that these cells are non-MYC amplified, a high sensitivity towards lipid depletion would be contradictory to the authors' conclusions.
3) The specificity of the inhibitors used in this study is purely based on literature evidence. As a large proportion of the in vivo findings rely on the CB5 compound, it would be good to establish at least some level of specificity of this drug in the systems used here. For example, the authors could monitor restoration of fatty acid uptake after overexpression of another fatty acid transporter that should not be affected by the compound.
4) The lipidome analysis of tumour tissue is somewhat confusing. The majority of lipids displayed in the heatmap in Fig. 4H, including phosphatidylcholine species, actually show a strong upregulation in SLC27A2-silenced tumours. Only DG and TG species seem to be selectively downregulated. Does this mean that fatty acids taken up by the tumour cells are funnelled specifically into the TG synthesis pathway? How does this compare to the lipid species identified to be altered in the initial metabolomics analysis performed on cell lines and tumours (Fig. 1)? This experiment needs deeper exploration and discussion.

5)
In the orthotopic allograft model (TH-MYCN+/+ in NCr nude mice, Fig. 5D-G), there are two clear outliers in the CB5 treatment group. Where these tumours also analysed by Oil-Red-O staining? This would be important to judge whether failure to inhibit tumour growth could be due to poor availability of the drug in the target tissue or due to potential resistance mechanisms (i.e. activation of fatty acid synthesis). Would it be possible to confirm drug availability in the tumour tissue?
6) The manuscript is quite difficult to read. This is most likely due to that large amount of data but some editing for clarity would help making the study more accessible. In particular, the first paragraph of the discussion section should be reconsidered. The authors mention several isolated findings and speculate about potential mechanisms that were not investigated in the manuscript. This distracts from the discussion of the main conclusions from the study. In particular, the discussion of potential roles of polyamine and glutamine metabolism in MYCN-amplified NB does not add further insight. Moreover, the discussion refers to the downregulation of glycerophosphoglycerols by MYCN but this is not mentioned in the results section and cannot easy be delineated from the figures. The section on metabolism and immune function is also quite unclear as the authors report reduced tumour growth in both immunocompetent and deficient in vivo systems.

Specific points:
Line 101: Something seems to be missing from this sentence. Should this be in a new paragraph? Line 128: The description of the experimental systems used in this study is a bit confusing. It should be made clear that MYCN overexpression (MYCN3 model) was performed in non-MYCN amplified SHEP cells. Figure 1C: It would be useful to assign the indicators of significance to the respective comparison (i.e. tumours, overexpression or KD). Asterisks could be organised in a grid. It would be particularly interesting to see pathways that are only significantly regulated in tumours. Remarkably, only one pathway (mapping to DG metabolism) is significantly altered in all three experimental systems. This should be discussed. Figure 1E: Why were these TG species (TG 42:0 and TG 48:2) chosen for display as box plots? Line 154: GSEA on the same data cannot be used to "validate" findings. This should be rephrased. Line 205: The statement that both A939572 nor CB5 have no cytotoxic effect but that CB5 is less toxic does not make much sense. This should be rephrased. Line 246: Reference to figure should be 3C. Line 317: The MNA LAN5 orthotopic implantation model has already been introduced on the previous page. Line 453: The meaning of this sentence is unclear. Do the authors mean immune cell activity targeted against MYCN positive NB tumours or MYCN function in immune cells. Line 455: Insert new paragraph after "therapeutic targets"? Line 461: Typo in temozolomide.

Dear Reviewers,
We thank the Reviewers for expressing a high level of enthusiasm for our manuscript entitled "MYCN-driven Fatty Acid Uptake is a Novel Metabolic Vulnerability in Neuroblastoma" (NCOMMS-21-29324), and providing insightful comments and suggestions. We have revised the manuscript following editorial requests and addressed the reviewers' concerns point by point below. The changes are highlighted in gray throughout the text.
Reviewer #1: Comment 1: Line 127: "in two MYCN-induced NB cell line models and in NB primary tumors". One model is Tet-ON Dox-induced, while the other here is conditionally knocked down by Dox-inducible shRNA expression. Perhaps a better way to describe them therefore is "conditionally-expressed MYCN", or other way to reflect that they are not MYCN-induced lines per se. Response: Thank you for the comment. We revised the text and better described the MYCN systems (line #114).  Fig. R1a-b). In addition, no significant changes in LDs were observed in SHEP cells treated with DOX and in SK-N-AS cells treated with 4-OHT ( Fig.  R1a-b), suggesting that DOX and 4-OHT do not play a role in LDs accumulation. These data are in accordance with our previous LDs observations. Notably, inhibiting FA synthesis (via A939572) and FA transport (via CB5) attenuated MYCN-driven LD accumulation (Fig. R1c), suggesting that FA synthesis and transport contribute to glycerolipids, particularly TGs (FDR < 0.25, new Fig. 1e). Because glycerolipids are the major components of LDs, this supports our LDs finding. Lipidomics profiling provides more specific characterization of the lipid categories altered by MYCN and these data were included in the new Fig.1e. LDs are dynamic organelles involved in energy regulation, signaling, and lipid metabolism. MYC(N) dynamically alters lipid homeostasis by promoting FA uptake (findings presented in this manuscript), de novo lipogenesis, LDs, and β-oxidation to generate ATP. Here we show that activation of MYCN promotes FA uptake and lipogenesis ( Fig. 2a and  supplementary Fig. 2a), thus resulting in LDs accumulation. Importantly, blocking FA uptake in the context of MYCN-amplification reduces LDs and exerts anti-tumor activity (Fig. R1c, Fig. 4e-h and Fig. 5f-g). However, inhibition of MYCN in NB also leads to LDs accumulation due to inhibition of FA β-oxidation and redirection of FAs towards de novo lipogenesis (Zirath et al. 2013). The relative contributions of FA uptake, synthesis, storage and oxidation to MYCN reprogrammed lipid homeostasis will need to be dynamically assessed in future investigations.
Comment 3: Fig 2D: While the effects of FA supplementation on the MNA cells vs. SHEP are clear, for the SK-N cells with or without activated MycN-ER it is more subtle (although significant). It is a little confusing to represent the same data in a different format here (Fig 2D right panel vs. the other panels), and perhaps drawing the data in the same way may show how the MycN-ER cells are more similar to the MNA cell lines, whereas without Myc-ER they are more similar to SHEP cells (therefore, plot the right panel of Fig 2D in 2 separate line graphs, not as single bar graph). Response: Thank you for the suggestion. To address also Reviewer #3 comment #2, we repeated cell viability study in SK-N-AS MYCN-ER ™ cells over time (0-6 days) and plotted the results as two separate line graphs as suggested (new Supplementary Fig. 2e, left). In addition, we examined the effects of lipid deprivation on cell viability and apoptosis in an extended panel of MNA and non-MNA lines (new Fig. 2c). Our results demonstrate that cells harboring MYCN-amplification or high MYCN activity are more dependent on exogenous lipids.
Comment 4: Line 286: "Supporting that NB cells highly depends" The grammar requires correction here. Response: The grammar has been corrected.
Comment 5: Fig 4J: There is no legend for the blue and black bars (in fact, they are colored incorrectly and should be red, presumably). Response: Thank you for the comment. We revised the graph, which is now correctly colored (new Fig. 4j).
Comment 6: Line 425-6: "These effects can be rescued by FA supplementation." From Fig 2D, it seems that there is only a partial rescue, and in fact in IMR32 and MYCN-ER cells, it is a relatively small effect. This should be reflected in the text here (since as written it implies a complete rescue) plus one could speculate as to why there is not a greater rescue with FA supplementation. This extends to Line 230 too ("Collectively, our data indicate that MNA-induced cell growth critically depends on exogenous FAs."). Again, if it is critically dependent, why isn't the rescue from FA supplementation more notable? It is a modest -although significant -effect in Fig  2E (right panel) for MYCN-ER active cells vs. inactive. Response: Thank you for the comment. We revised the text to emphasize that it is a partial rescue. We repeated cell viability study in SK-N-AS MYCN-ER ™ cells over time (0-6 days). MYCN-ER activated cells were more susceptible to lipid deprivation. Moreover, FA supplementation almost completely rescued the viability of MYCN-ER activated cells (p=0.0001; new Supplementary Fig. 2e). In addition, we examined the effects of lipid deprivation on cell viability and apoptosis in additional MNA (SK-N-BE(2c)) and non MNA (SK-N-AS parental) cell lines. MNA cells were consistently more dependent on exogenous lipids compared with non MNA cells. Moreover, FA supplementation partially rescued cell viability in all the lines tested (p<0.001) and completely rescued the deprivation-induced apoptosis in LAN5 and SK-N-BE(2c) cells (new Fig. 2c). Overall, our new data suggest that MYCN-driven cells depend on exogenous FAs.
Reviewer #2: Comment 1: Please show WB on MYCN, proliferation, and apoptosis markers after treatment with the inhibitors in the cell lines. In addition, authors only show changes in MYCN expression in LAN5 and MYCN3 in S. Figure  1A, but not in any other of their cell lines after induction/repression of MYCN. Response: As suggested, we assessed the protein expression of MYCN, c-MYC, p53, p21, total and cleaved Caspase-3, as well as total and cleaved PARP in MNA (LAN5 and IMR32) and non-MNA (SHEP) cells upon CB5 treatment (new Fig. 4k). Western blots were performed in triplicate and quantifications are shown in the new Supplementary Fig. 5a. CB5 preferentially induced cleaved PARP and cleaved Caspase-3 expression in MNA cells (new Fig. 4k and new Supplementary Fig. 5a). These data are in agreement with the Caspase 3/7 activity analysis shown in Fig 4j. CB5 also inhibited MYCN but not c-MYC protein expression, supporting the selective targeting of MNA cells. p53 and its downstream target p21 play critical roles in cell cycle, proliferation, and apoptosis in the context of MYCN amplification. We found that CB5 increased both p53 and p21 expression in MNA cells. Collectively, our data suggest that CB5 inhibits MYCN and activates p53 signaling to suppress cell growth and promote apoptosis. As requested, we added MYCN WB in Tet21/N (Tet-Off) system (new Fig. 3b).
In the MYCN-ER ™ system, 4-OHT induces MYCN transcriptional activity. In Fig. 3a we included mRNA expression of ODC1, a known MYCN target gene, as positive control for MYCN activation. A more extensive validation of the MYCN-ER ™ system can be found in our recent manuscript (Moreno-Smith et al., Nat. Commun. 2021, doi: 10.1038Fig. 3c).  . 1c and response to reviewer #1 comment #2). Importantly, blocking FA uptake via CB5 in the context of MYCN-amplification reduces the expression of FA synthesis (ACC and SCD1) and transport (FATP2) proteins (new Supplementary Fig. 5d), and thus glycerolipids and LDs accumulation ( Fig. 4g-h). It has been previously shown that inhibition of MYCN in NB also leads to LDs accumulation due to inhibition of FA β-oxidation and redirection of FAs towards de novo lipogenesis (Zirath et al. 2013; included in revised discussion). The relative contributions of FA uptake, synthesis, storage and oxidation to MYCN reprogrammed lipid homeostasis will need to be dynamically assessed in future investigations.
Comment 3: Figure 4 and legend. Photos of LAN5 and SHEP cells are shown after BOIDPY. It is obvious by eye that the fluorescence is decreasing more in the LAN5 cells but is also clearly decreasing in the SHEP cells. However, in the graph is looks more or less as if there is no decrease in the SHEP cells. In addition, it is indicated that * means significant changes after CB5 treatment and that # indicates significant differences between LAN5 and SHEP cells at same CB5 dose. Please clarify. The use of # is unusual. Response: Thank you for the comment. To quantify the fluorescence intensity, we took five random fields per treatment group and determined the CTCF as described in the methods section. In the new Fig 4i, each white circle indicates the average CTCF from one experiment. The study was repeated three times. We have now included more representative images. We have also separated LAN5 from SHEP graphs for individual statistical analysis, and individual p-values are now presented. Response: Based on Reviewer #3 comment #6, we revised the discussion section focusing on lipid metabolism. Thus, the above paragraph has been removed from the discussion. However, this reference is now cited in the introduction section (line #64).

Comment 5:
The authors confirm that SLC27A2 is required for NB survival performing shRNA in LAN5 cells. These results should be validated in a second MNA NB cell line. Moreover, to prove that it is a target of MYCN and not c-MYC, authors should also prove that shSLC27A2 does not affect viability in non-MYCN NB cells. Response: Thank you for the suggestion. We genetically depleted SLC27A2 in a second MNA cell line (IMR32) and in a non MNA cell line (SK-N-AS). We found that silencing SLC27A2 reduced FA uptake and cell growth in IMR32 but not SK-N-AS cells (new Supplementary Fig. 4c), confirming that these phenotypes are selective to MNA cells.

Comment 6:
Author demonstrated lipid droplet accumulation as a result of increased fatty acid uptake and synthesis was demonstrated in SK-N-AS MYCN-ER cells. Authors should validate that this is a result due to Response: We appreciate the reviewer's observation, and looked into the metabolomics analysis of lipid species. Indeed, MYCN appears to reduce phospholipid synthesis and upregulate glycerolipid synthesis and DG levels (Fig. 1c, Supplementary Fig. 1b). This likely explains why silencing SLC27A2 in the context of MYCNamplification effectively inhibits MYCN-induced glycerolipid accumulation (mostly DG and TG species;Figs. 5gh). When looking at the phospholipid and glycerolipid biosynthesis pathways, DGs are derived from phosphatic acid (PA) by phosphatidic acid phosphatases and converted to TGs by diacylglycerol-acyltransferase1 or 2 (DGAT1 and DGAT2). On the other end, PA can be converted to CDP-DGs by CDS1 and CDS2, leading to other types of phospholipids (PC, PE, PG, PI, and PS). We speculate that MYCN may directly regulate glycerolipid and TG synthesis by upregulating DGAT2 to store excess FAs in TGs and LDs. DGAT2 is highly expressed in MNA patients across two large datasets (GSE45547 and GSE85047), and high expression predicts NB poor clinical outcomes (p < 0.01, new Supplementary Fig. 7a). In addition, MYCN upregulates DGAT2 expression in multiple in vitro NB models (p < 0.05), and blocking FA transport via CB5 suppresses its expression (p < 0.05, new Supplementary Fig. 7b). Because glycerolipid and phospholipid biosynthesis share the same precursor PA, a shift towards glycerolipid synthesis may inversely affect phospholipid synthesis. Future investigations remain necessary to elucidate how MYCN dynamically maintains lipid homeostasis. Future studies will aim at tracing FA incorporation into lipids and performing a comprehensive evaluation of the enzymatic activities involved in glycerolipid and phospholipid metabolism. This comment was addressed in the revised discussion.

Comment 5:
In the orthotopic allograft model (TH-MYCN+/+ in NCr nude mice, Fig. 5D-G), there are two clear outliers in the CB5 treatment group. Where these tumours also analyzed by Oil-Red-O staining? This would be important to judge whether failure to inhibit tumour growth could be due to poor availability of the drug in the target tissue or due to potential resistance mechanisms (i.e. activation of fatty acid synthesis). Would it be possible to confirm drug availability in the tumour tissue? Response: Thank you for the comment. Indeed, two tumors escaped CB5 treatment (Fig. 5f). Supporting our findings, these tumors did not exhibit lipid inhibition by Oil Red O staining (their neutral lipid level is comparable to CTRL tumors, new Supplementary Fig. 5d, bottom). Unlike CB5-responsive tumors, these tumors also showed increased MYCN activation and FA synthesis/transport activity (SCD1 and FATP2, new Supplementary  Fig. 5d, middle). CB5 likely fails to inhibit MYCN and MYCN-mediated FA transport and synthesis in these two tumors, resulting in LDs accumulation.

Comment 6:
The manuscript is quite difficult to read. This is most likely due to that large amount of data but some editing for clarity would help making the study more accessible. In particular, the first paragraph of the discussion section should be reconsidered. The authors mention several isolated findings and speculate about potential mechanisms that were not investigated in the manuscript. This distracts from the discussion of the main conclusions from the study. In particular, the discussion of potential roles of polyamine and glutamine metabolism in MYCN-amplified NB does not add further insight. Moreover, the discussion refers to the downregulation of glycerophosphoglycerols by MYCN but this is not mentioned in the results section and cannot easy be delineated from the figures. The section on metabolism and immune function is also quite unclear as the authors report reduced tumour growth in both immunocompetent and deficient in vivo systems. Response: We appreciate the suggestion. We substantially revised our manuscript language to improve clarity. The first paragraph of the discussion was revised and is now focused on lipid/DG pathway as suggested, and the section of metabolism and immune function was clarified. We believe our revised manuscript is much improved in terms of clarity and focus.
Comment 7: Line 101: Something seems to be missing from this sentence. Should this be in a new paragraph? Response: Thank you, this sentence is now revised (line #89).
Comment 8: Line 128: The description of the experimental systems used in this study is a bit confusing. It should be made clear that MYCN overexpression (MYCN3 model) was performed in non-MYCN amplified SHEP cells.

Response:
We appreciate the comment. The MYCN models, including MYCN3, are now described in more detail (line #120 of the results section; see also Reviewer #1 comment #1).
Comment 9: Figure 1C: It would be useful to assign the indicators of significance to the respective comparison (i.e. tumours, overexpression or KD). Asterisks could be organised in a grid. It would be particularly interesting to see pathways that are only significantly regulated in tumours. Remarkably, only one pathway (mapping to DG metabolism) is significantly altered in all three experimental systems. This should be discussed.