ALK ligand ALKAL2 potentiates MYCN‐driven neuroblastoma in the absence of ALK mutation

Abstract High‐risk neuroblastoma (NB) is responsible for a disproportionate number of childhood deaths due to cancer. One indicator of high‐risk NB is amplification of the neural MYC (MYCN) oncogene, which is currently therapeutically intractable. Identification of anaplastic lymphoma kinase (ALK) as an NB oncogene raised the possibility of using ALK tyrosine kinase inhibitors (TKIs) in treatment of patients with activating ALK mutations. 8–10% of primary NB patients are ALK‐positive, a figure that increases in the relapsed population. ALK is activated by the ALKAL2 ligand located on chromosome 2p, along with ALK and MYCN, in the “2p‐gain” region associated with NB. Dysregulation of ALK ligand in NB has not been addressed, although one of the first oncogenes described was v‐sis that shares > 90% homology with PDGF. Therefore, we tested whether ALKAL2 ligand could potentiate NB progression in the absence of ALK mutation. We show that ALKAL2 overexpression in mice drives ALK TKI‐sensitive NB in the absence of ALK mutation, suggesting that additional NB patients, such as those exhibiting 2p‐gain, may benefit from ALK TKI‐based therapeutic intervention.


30th Jun 2020 1st Editorial Decision
Dear Dr Palmer, Thank you for the submission of your manuscript (EMBOJ-2020-105784) to The EMBO Journal. Please accept my apologies for the delay with the peer-review of your work due to protracted referee input and detailed discussions in the team. Your manuscript has been initially sent to three reviewers, however one referee indicated a (positive) conflict of interest only during peer-review. We have thus considered input from the other two experts, which I enclose below.
As you will see, the referees acknowledge the potential interest and novelty of your results, although they also express a number of issues that will have to be conclusively addressed before they can be supportive of publication of your manuscript in The EMBO Journal. In particular, referee #2 raises concerns regarding the human pathophysiological relevance of your findings and requires complementary experiments (ref#2, pts 1,2,6). Referee #1 agrees in that histopathological analyses should be expanded (ref#1, pts 9,11). In addition, the reviewers raise a number of points related to statistics and replicates, data presentation as well as discussion of the findings and context, which would need to be conclusively addressed to achieve the level of robustness and clarity needed for The EMBO Journal.
I judge the comments of the referees to be generally reasonable and given their overall interest, we are in principle happy to invite you to revise your manuscript experimentally to address the referees' comments.
Further information is available in our Guide For Authors: https://www.embopress.org/page/journal/14602075/authorguide The revision must be submitted online within 90 days; please click on the link below to submit the revision online before 28t h Sep 2020. This is a nicely written study that outlines the role of the ALK ligand ALKAL2 in driving neuroblastoma development independent of activating mutations in ALK. While it has been known for some time that mutant forms of ALK which are ligand-independent, are expressed in neuroblastoma, the effects of full-length ALK towards the pathogenesis of neuroblastoma and whether this is ligand-dependent has not been investigated. Indeed, for many years, the natural ligands for ALK were unknown and had been incorrectly attributed to other proteins. Hence, the previous discovery of ALKAL2 by this research group was an impressive step forward in our understanding of the normal physiological role of ALK. This group now follow-up on this important work to show that ALKAL2 also plays a role in the development of neuroblastoma and furthermore that ALKAL2-driven ALK activity is targetable by ALK tyrosine kinase inhibitors inducing disease remission in cell line and murine models. These data therefore represent a huge step forward in our understanding of the pathogenesis of neuroblastoma.

Link Not Available
I have no major concerns with regards to the data that are presented. The experiments have been conducted with the most appropriate model systems and these data, as presented are sound and conclusive.
Minor concerns: 1. In some places it is not stated how many repeats and whether these are biological or technical replicates have been conducted for each experiment, e.g. Western blots. 2. Supplementary Figure S1 Western blots -I cannot see protein bands for ALK nor pALK for the IMR-32 cell lines -perhaps the contrast can be adjusted and a second image shown of the higher contrast so that these are visible. 3. spell out the meaning of the acronym SRF at first use 4. Figure 1 -Western blots could be added for ARC, EGR2 and 3 if antibodies are available. A western blot for IMR32 could also be shown in the supplementary data. 5. Figure 2 -were any of the proteins encoded for by the transcripts detected in figure 1 also seen in the proteomics analysis? 6. Could the authors please mark on the volcano plots with lines where the cut-off lies for significance and also for the logFC 7. Figure 2e -It looks to me like the post-translational modification of FOXO3 detected does not change following treatment with lorlatinib in the volcano plot, but does in the Western blot -do the authors have an explanation for this? Can the authors also show the data for the 24 hr timepoints with lorlatinib. 8. As for the proteomics in figure 2 -do the corresponding transcript levels change in the data presented in Figure 1? 9. Figure 3E -could the authors expand on the differences in the histopathology of the MYCN versus ALK mutant/MycN tumours, particularly the description of the H&E images 10. Figure 3 legend p-values -can the authors clarify which comparisons the stated p-values relate too. The figure has *, ** or *** and the legend states an absolute value, but the value for which comparison? 11. Figure 4B -can the authors expand on the sites of tumour development -they say abdominal but attached to which organ? 12. Figure 4D -it looks like ALK levels increase in the ALKAL/Myc mouse-is ALK expressed at a higher level in these mice or is it better stabilised? A pALK blot would also be beneficial here to see if phosphorylation levels achieved with ALKAL and WT ALK are higher than those in the mutant ALK mice compared to the ThmycN (ALK WT mouse) tumours -maybe higher levels of phosphorylation might account for higher ALK expression levels due to stabilisation of the protein?? 13. Fig 5G -are there any significant differences in overall survival attributable to VGF levels? 14. Figure 5F, is there a negative control? What does VGF expression look like in Th-MycN mouse tumours? 15. It is interesting that the mutant form of ALK activates more genes that ALKAL activated WT ALK ( Figures 5C and D). Can the authors expand on this finding? 16. FIGURE 6D -the authors state that smaller spheroids are more sensitive to brigatinib -are these data significant? No p-values are presented. 17. Figure 7e -the legend says 5 and 15 days whilst the figure says 4 and 14. 18. Materials and methods -a lot of antibodies are described but were they all used in the manuscript?
Referee #2: Borenäs et al. make a case for the ALK ligand, Alkal2, as a potential driver of tumorigenesis and tumor progression in neuroblastoma (NB). They provide strong experimental data, including transgenic in vivo models, to support an oncogenic role of Alkal2 overexpression. This has potential relevance in the clinic, as the authors show that Alkal2-driven NB are sensitive to ALK tyrosine kinase inhibitors. This is a compelling and very interesting story that beautifully combines tumor biology with therapeutic intervention. Generally, the data is very strong, but in the latter parts of the manuscript the level of scientific rigor is somewhat decreasing.
Strengths: -multi-omics approach for unbiased identification of Alkal2 upregulation in NB cell lines -novel transgenic mouse model supporting Alkal2 as driver of tumorigenesis and progression in NB Weaknesses: -lack of survival data for ALK TKI application in vivo -no validation of Alkal2 deregulation is driving NB in human patients Major Points: 1) The authors try to emphasize the point that aberrant ligand-receptor signaling promotes tumor formation and progression in NB, particularly in the absence of oncogenic driver mutations in ALK. This point could be substantially stronger if there is evidence for ALK phosphorylation in ALK nonmutant patient tumors, or if the authors could verify that the identified cluster of 6 downstream transcription factors are upregulated in human samples (and/or linked to patient outcome). 2) Fig. 5G: the survival analysis should be based on the 'median', not a 'scan' method which plots the best fit. This can force p values for data that are not necessarily meaningful.
3) Fig. 7B lacks a description of the statistical test that has been used. 4) Fig. S1: As the authors note, IMR-32 cells show lower expression of ALK. The ALK, p-ALK and p-AKT bands are hardly visible for IMR-32 cells. The authors should consider probing IMR-32 cells on a separate gel to show ALK expression and phosphorylation in IMR-32 cells, or perhaps showing the same membranes with longer exposure to demonstrate the same. 5) p.21 l.10 -change 'CPG island' to 'CpG island' Manuscript: EMBOJ-2020-105784 Title: Overexpression of the ALK ligand ALKAL2 drives neuroblastoma in the absence of ALK mutation.
General response to both reviewers: We would like to thank the reviewers for their constructive comments and suggestions, and the editor for the opportunity to re-submit our manuscript. We have made a number of changes, corrections and additions that we feel have significantly strengthened the work. Despite the current difficult working circumstances we have attempted to address all of the reviewers comments, and as detailed below, we outline the changes we have made.
We have not been able to complete all of the mouse experiments asked of us by the reviewers, in part due to ethical restrictions on our treatment protocols and in part due to breeding problems in our mouse colony. Our breeding problems are a direct result of our university building a new faculty building next to the mouse house with bedrock blasting every Tuesday and Thursday, which has severely limited our mouse breeding's due to stress (they are on a sensitive 129 background). Having said this, we feel that our revised manuscript has been tightened up on many fronts and that the main message is built on a very solid foundation. We very much hope that the reviewers will also appreciate our revised manuscript.
Below is a detailed point-by-point response to the individual reviewers' comments.

Referee #1:
This is a nicely written study that outlines the role of the ALK ligand ALKAL2 in driving neuroblastoma development independent of activating mutations in ALK. While it has been known for some time that mutant forms of ALK which are ligand-independent, are expressed in neuroblastoma, the effects of full-length ALK towards the pathogenesis of neuroblastoma and whether this is liganddependent has not been investigated. Indeed, for many years, the natural ligands for ALK were unknown and had been incorrectly attributed to other proteins. Hence, the previous discovery of ALKAL2 by this research group was an impressive step forward in our understanding of the normal physiological role of ALK. This group now follow-up on this important work to show that ALKAL2 also plays a role in the development of neuroblastoma and furthermore that ALKAL2-driven ALK activity is targetable by ALK tyrosine kinase inhibitors inducing disease remission in cell line and murine models. These data therefore represent a huge step forward in our understanding of the pathogenesis of neuroblastoma.
I have no major concerns with regards to the data that are presented. The experiments have been conducted with the most appropriate model systems and these data, as presented are sound and conclusive.
Minor concerns: 1. In some places it is not stated how many repeats and whether these are biological or technical replicates have been conducted for each experiment, e.g. Western blots.

Authors reply:
The reviewer has a good point and we have now stated the number of repeats and whether they are biological or technical throughout. Figure S1 Western blots -I cannot see protein bands for ALK nor pALK for the IMR-32 cell lines -perhaps the contrast can be adjusted and a second image shown of the higher contrast so that these are visible.

Authors reply:
We have done our best to improve these blots, including a second image as suggested by the reviewer. However, ALK and pALK are always difficult when running IMR-32 together with NB1 samples due to the low levels of ALK.
3. spell out the meaning of the acronym SRF at first use Authors reply: We thank the reviewer for spotting this; we have now fixed this in the revised version. 4. Figure 1 -Western blots could be added for ARC, EGR2 and 3 if antibodies are available. A western blot for IMR32 could also be shown in the supplementary data.

Authors reply:
We have tried very hard to look at ARC and EGR2/3, however, we have not been able to find antibodies that work for these proteins -at least not at the endogenous level in neuroblastoma cells. During the revision process, we purchased a number of antibodies that should detect these proteins, but none detected endogenous proteins of the correct size and we were unable to generate any trustable data despite intensive effort. Therefore, we are not able to provide any solid immnuoblotting data to address this point. Authors reply: The reviewer raises a good point here. We have checked our RNAseq data with the proteomics and phoshoproteomics data, but it is clear that we do not detect scarce proteins. In summary, for EGR1 -3, ARC, FOS, FOSB and SRF there is no data in our proteomics or phosphoproteomics and we only see them by RNASeq. However, e.g. EGR1 and FOS are clearly detected with antibodies. Since we do not detect the proteins, either by mass spectrometry or immunoblotting, we are currently unable to add any insightful data regarding this question. While we have a rich proteomics and phosphoproteomics dataset, we have been following current discussion regarding overlap and integration of proteomic (total and phosphor) and RNASeq data and realise that this is not trivial. We are currently investigating this and working to produce deeper phosphor datasets with which to tackle these challenges. 6. Could the authors please mark on the volcano plots with lines where the cut-off lies for significance and also for the logFC Authors reply: We opted for a hyperbolic threshold in the proteomics data analysis, analogous to other proteomics tools and studies (e.g. Singh S, proteomics 2016; Tyanova S, Nature Biotech 2016; Emdal, Science signaling 2018). This approach implies a smoother transition between significance (log P values) and logFC cut-offs (i.e. a protein which is on the edge of being significant will be required to have a higher logFC value and vise versa). To clarify this for the reader we have added the following sentence to the figure caption: "Dashed lines indicate differential expression thresholds (hyperbolic cut-off with asymptotes P=0.05 and log2(FC)=+/-2; see Methods for details)." 7. Figure 2e -It looks to me like the post-translational modification of FOXO3 detected does not change following treatment with lorlatinib in the volcano plot, but does in the Western blot -do the authors have an explanation for this? Can the authors also show the data for the 24 hr timepoints with lorlatinib.

Authors reply:
We agree with the reviewer's interpretation. While the blots indeed suggest lorlatinib sensitivity, in contrast to the other genes we labelled in fig. 2E, FOXO3 phosphorylation at S253 does not substantially change after lorlatinib inhibition of ALK, suggesting regulation that is more complex. This complexity is further supported by the ALKdependent transient nature of the response (disappears after 24hr in control conditions, but not after lorlatinib treatment). To illustrate this, we have now added the 24hr lorlatinib volcano plot to panel E as well as an additional panel G focusing on FOXO3 (phosphor)protein dynamics. We note that there are many examples of complex regulation in our dataset, which can be compared with e.g. STAT3 dynamics, which 'behave' rather as expected (Fig 2I). Some of these, such as FOXO3 are rather interesting, and shed light on additional layers of regulation other than phosphorylation. Despite continuing experimental analysis we do not yet fully understand what is going on in many of these cases, and therefore in the revised manuscript we have tried to highlight these observations more clearly and hope that other investigators as well as ourselves in the future will dig into more detail to provide mechanistic insight. 8. As for the proteomics in figure 2 -do the corresponding transcript levels change in the data presented in Figure 1?

Authors reply:
We have looked at a quite a few proteins, including VGF, TNC, IGFBP5 and FOSL2, and their corresponding transcript levels, and find that this becomes very complex in each individual case. Right now, we think that it is difficult to add anything useful regarding this to the manuscript without getting into a lot of detail that can become rather confusing. 9. Figure 3E -could the authors expand on the differences in the histopathology of the MYCN versus ALK mutant/MycN tumours, particularly the description of the H&E images Authors reply: We have revised our text and included more description of the images, this can be found on page 11. 10. Figure 3 legend p-values -can the authors clarify which comparisons the stated p-values relate too. The figure has *, ** or *** and the legend states an absolute value, but the value for which comparison?
Authors reply: We have revised the legend for Figure 3 to clarify this for the reader.
11. Figure 4B -can the authors expand on the sites of tumour development -they say abdominal but attached to which organ?

Authors reply:
We have now included data regarding the sites of tumor development. Interestingly, in almost all cases there is no involvement of the adrenal gland, rather the tumour arises in close proximity, most likely in the nearby peripheral nervous system. We observe this in all genotypes, and have now included this. This is rather interesting and something that we would like to follow up in future work in more detail. We have added this data as Fig 4C and have included text addressing this on pages 12/13 in the revised manuscript.
12. Figure 4D -it looks like ALK levels increase in the ALKAL/Myc mouse-is ALK expressed at a higher level in these mice or is it better stabilised? A pALK blot would also be beneficial here to see if phosphorylation levels achieved with ALKAL and WT ALK are higher than those in the mutant ALK mice compared to the ThmycN (ALK WT mouse) tumours -maybe higher levels of phosphorylation might account for higher ALK expression levels due to stabilisation of the protein??
Authors reply: We agree with the reviewer that the blot might suggest higher ALK expression in Rosa26_Alkal2;Th-MYCN mice. We verified this at the transcriptome level in our RNA-Seq data (Table S4) and we could not confirm any significant differential expression between Rosa26_Alkal2;Th-MYCN and Th-MYCN (logFC -0.19, P=0.52) nor between Alk-F1178S;Th-MYCN and Th-MYCN (logFC=0.44, P=0.14). Based on this, we have rerun these samples and have included another ALK panel here. Despite several attempts, we have been unable to see a clear pALK signal in these tumour samples and have therefore not included pALK data. 13. Fig 5G -are there any significant differences in overall survival attributable to VGF levels?
Authors reply: Based on the comments from both reviewer #1 and 2 we have now updated the survival analyses to use median values as cut-offs between low/high expression. We also checked overall survival and found that this was also significant (log-rank test P=3.3e-16 for the Kocak cohort and log-rank test P=5.2e-05 for the Versteeg cohort). While overall survival is not shown in the revised Figure, we have included this information in the text (on page 15).
14. Figure 5F, is there a negative control? What does VGF expression look like in Th-MycN mouse tumours?
Authors reply: We agree with the reviewer that this would be good to include. We find that there is VGF protein in the MYCN tumors, however, the staining is not as strong as in the Rosa26_Alkal2;Th-MYCN and the Alk-F1178S;Th-MYCN tumours, in agreement with our RNAseq data. These data have now been included in Fig 5G of the revised version. 15. It is interesting that the mutant form of ALK activates more genes that ALKAL activated WT ALK ( Figures 5C and D). Can the authors expand on this finding?

Authors reply:
The reviewer is quick to notice this point, and we have also been considering what this means, and have discussed this at length in the lab. We have some ideas here about what is going on here. In our experiments in both neuroblastoma cell lines and in the model organism Drosophila melanogaster, we have noted that overexpression of ligand appears to downregulate the ALK protein. Moreover, when we mutate ALK, the protein appears to have a slightly different localisation, and to be more persistent. There is some previous data to support this, such as Mazot et al, Oncogene 2011, in which activation of ALK was shown to affect receptor trafficking. We suspect that differences in the trafficking of the wild-type versus the mutant ALK receptor may at least be part of the explanation for our observations. We have now added text in the discussion addressing this. 16. FIGURE 6D -the authors state that smaller spheroids are more sensitive to brigatinib -are these data significant? No p-values are presented.

Authors reply:
We thank the reviewer for spotting this; somehow, we missed these and have now included them in the revised version. The reviewer is indeed correct that the smaller Rosa26_Alkal2;Th-MYCN spheroids (2000 cells or 4000 cells) are more sensitive to brigatinib treatment than the bigger spheroids (8000 cells), and this is statistically significant (p-values of 0.01 and 0.014). However, this trend was not detected in Alk-F1178S;Th-MYCN spheroids. While interesting, we currently have no explanation for this observation. However, we have now commented on this in the text and added the p-values in the revised manuscript. 17. Figure  Authors reply: We thank the reviewer for spotting this error, we have now corrected this in the revised version.

Authors reply:
We thank the reviewer for pointing this out. We have now gone through the materials and methods and checked the antibodies used carefully. The antibodies and dilutions used are now clearly detailed in a table (S5) in the revised version.

Referee #2:
Borenäs et al. make a case for the ALK ligand, Alkal2, as a potential driver of tumorigenesis and tumor progression in neuroblastoma (NB). They provide strong experimental data, including transgenic in vivo models, to support an oncogenic role of Alkal2 overexpression. This has potential relevance in the clinic, as the authors show that Alkal2-driven NB are sensitive to ALK tyrosine kinase inhibitors. This is a compelling and very interesting story that beautifully combines tumor biology with therapeutic intervention. Generally, the data is very strong, but in the latter parts of the manuscript the level of scientific rigor is somewhat decreasing.
Strengths: -multi-omics approach for unbiased identification of Alkal2 upregulation in NB cell lines -novel transgenic mouse model supporting Alkal2 as driver of tumorigenesis and progression in NB Weaknesses: -lack of survival data for ALK TKI application in vivo -no validation of Alkal2 deregulation is driving NB in human patients Major Points: 1) The authors try to emphasize the point that aberrant ligand-receptor signaling promotes tumor formation and progression in NB, particularly in the absence of oncogenic driver mutations in ALK. This point could be substantially stronger if there is evidence for ALK phosphorylation in ALK nonmutant patient tumors, or if the authors could verify that the identified cluster of 6 downstream transcription factors are upregulated in human samples (and/or linked to patient outcome).

Authors reply:
We appreciate the reviewers comment here. On looking back at the initial submitted manuscript, we realise that we have not provided much background on the evidence regarding ALK levels and activity in NB, outside of the ALK mutant status. To clarify this we now discuss results from others that provide evidence for correlation of ALK expression/activity and prognosis at the transcript and protein level, including a recently published study. These findings fit with our own unpublished data from a limited number of patient samples. However, we are not a clinical laboratory, and have only very limited access to patient samples for analysis by either immunohistochemistry or proteomics methods. We hope to be able to look at ALKAL2 expression in patient samples in a future collaboration. As suggested by the reviewer we have looked into the highlighted transcription factors in the context of patient outcome and for EGR1, EGR2, ARC, FOS, FOSB and FOSL2 the data is very mixed depending on which dataset one looks at. We also looked at SRF, and here we saw a consistent association in all datasets looked at, where high SRF is always associated with poor outcome. We thank the reviewer for this suggestion and have now included this data as Extended View Figure 3, where we show by both median and scan outcome related to SRF levels.
2) Better evaluation of histopathological hallmarks of NB in both Alk-F1178S/Th-MYCN and Alkal2/Th-MYCN tumors would strengthen the case that the tumors in the respective animal model closely resemble human disease pathology. This applies to Figs. 3 and 4. On a related note, do R26-Alkal2 mice show hyperplasia in celiac ganglia like Alk-F1178S mice?
Authors reply: Regarding the potential hyperplasia in R26-Alkal2 ganglia, our initial experiments suggest they ae enlarged, however we have not repeated this with sufficient numbers of mice and controls to include this data. Given our current breeding speed I estimate this would take us quite a few more months to achieve. We have tried to improve the text describing the tumours arising in  Figure 3D, this should also be used in Figure 4A. For the R26:Alkal2/Th-MYCN the authors mention in the discussion (p.18 l.10) that 3/24 mice have not developed tumors at 200 days. I think in this light Fig. 4A is somewhat misleading, and the authors should include the full cohort of 24 R26:Alkal2/Th-MYCN animals in the Kaplan-Meier curve.

Authors reply:
We have now included all data from our mouse colony (Fig 4F). We note that we also have seen one Alk-F1178S/Th-MYCN animal that did not present with tumor before 200 days, which is also included in this data. This data includes all mice in our colony that were monitored for 200 days as well as those mice that were used in treatment upon tumor detection. For animals that were used in treatment schedules we estimated survival as 2 weeks after tumor detection, as estimate of survival in the absence of treatment. This data provides a more accurate overview of the overall penetrance. In our initial experiment ( Fig  4A), all mice were carefully monitored and controlled and we present a highly controlled experiment. We have not tried to mislead the reader, in fact we have striven to be very honest. We also cannot say whether mice that were tumour free at 200 days would have presented tumors later, however. We now have 45 R26:Alkal2/Th-MYCN mice, of which 5 mice have reached 200 days without presenting with a tumour, therefore we are starting to feel confident that somewhere around 10% of mice will not present with a tumour by 200 days. This makes our model highly penetrant, although not quite as much as our Alk-F1178S/Th-MYCN, which is almost fully penetrant at 200 days (97.8%). Fig. 4D it is unclear to me why the Alkal2 band in R26:Alkal2/Th-MYCN samples is of the same intensity as the other samples. Based on the overexpression of Alkal2 in this mouse line I would have expected increased protein levels of Alkal2.

Authors reply:
We realise that this may appear confusing in the original version of the manuscript. We have been able to detect ALKAL2 protein in both human neuroblastoma cell lines (Javanmardi, 2019) and patient tumors (unpublished). This is in agreement with our RNASeq data, which detects Alkal2 mRNA in tumor samples of all three mouse genotypes that we have used here (Table S4). As the Alkal2 transgene is codon optimized, we can clearly state that all tumors transcribe Alkal2 from the endogenous locus. We are unable to comment on whether the Alkal2 protein observed in the mouse tumors originates from the tumor cells themselves or from stromal tissue that contributes to the tumour, or indeed even more remotely, and this is something we are interested in looking at in future experiments. However, when we isolate cell lines from the mouse tumors, we observe a clear increase in Alkal2 protein in the cells derived from R26:Alkal2/Th-MYCN mouse tumors as compared with cells isolated from Alk-F1178S/Th-MYCN tumors. This data is now included in the revised version of Figure 6 ( Figure 6B) and we have also added text in the results that hopefully clarifies this for the reviewer and for the reader. 5) Fig. 6B/C: 'spheroid area' is a very unusual readout for sphere formation. Importantly, this does not reveal whether changes in sphere area are due to sphere formation, cell adhesion, or cell death. A more direct quantification of sphere numbers, preferably in limiting dilution format, would be better to support this point. I would further encourage the authors to include cell viability assays in Figs. 6/7, as the quantification of cell confluence may also be skewed by cell contractility or other processes not related to changes in viability.

Authors reply:
We appreciate the reviewer's comments here and as suggested, we have reanalyzed the data as sphere numbers instead of spheroid area. For spheroid number quantification, spheroids with a diameter larger than 100um were counted. Brigatinib treatment significantly decreased the spheroid number in both #3540 and #3546 cell lines. We have also performed cell viability assays with resazurin, and these new data are included in the revised Figure 6 (as Figure 6F). 6) Ideally, a survival analysis after lorlatinib treatment should be included in Fig. 7.

Authors reply:
We appreciate the reviewer's comments here; however we have been unable to provide this. We thank the reviewer however; as this comment has meant that we have included this possibility as an option in our updated Ethical permission, which is currently under evaluation. 7) In several instances, statistical analysis is missing. This includes Fig. 6 panels A, C, D and Fig. 7 panel A Authors reply: We thank the reviewer for spotting this; somehow, we missed these and have now included them in the revised version.

19th Oct 2020 1st Revision -Editorial Decision
Dear Dr Palmer, Thank you for submitting your amended manuscript (EMBOJ-2020-105784R) to The EMBO Journal. Your revised study was sent back to the referees #1 and #2, and we have received comments from both of them, which I enclose below. As you will see the referees find that their concerns have been sufficiently addressed and they are now broadly in favour of publication.
Thus, we are pleased to inform you that your manuscript has been accepted in principle for publication in The EMBO Journal, pending the a number of minor aspects related to formatting and data representation as detailed below are addressed at resubmission.
Further, I will share additional edits and comments from our production team during the next days to be considered.  Do the data meet the assumptions of the tests (e.g., normal distribution)? Describe any methods used to assess it.
Is there an estimate of variation within each group of data?
Animal RNA-Seq was performed with a n=4 or more. Detailed histological analyses of mice was performed with n=4 or more. Kaplan survival curves were performed with total cohort of 126 (3D);total cohort of 42 (4A); total cohort of 234 (4F). RNA-Seq from mouse tumours was performed with at least n=4 biological samples. Mouse tumour treatment was carried out with at least n=6 Animals suffering from unrelated illnesses were removed in accordance with ethical regulations. These animals were excluded from analyses.
No specific steps were takem.

Manuscript Number: EMBOJ-2020-105784
Yes, statistical tests are detailed in the Figure legends and materials and methods.
Data was checked for normality using the appropriate test in GraphPad Prism 6/7.

NA
Animals were followed over time in a non-biased manner. Treated animals were selected on tumour detection. Treatment was predetermined prior to tumour detection, meaning that the order of tumour development determined the animals treatment with either vehicle or drug, irrespective of age and sex. No.
All animals were followed for tumour occurance and ethical protocols strictly followed. All personnel would have had access to genotyping data for each animal as well as being aware of whether the animal was treated with either vehicle or drug.

Data
the data were obtained and processed according to the field's best practice and are presented to reflect the results of the experiments in an accurate and unbiased manner. figure panels include only data points, measurements or observations that can be compared to each other in a scientifically meaningful way.
The data shown in figures should satisfy the following conditions: Source Data should be included to report the data underlying graphs. Please follow the guidelines set out in the author ship guidelines on Data Presentation.
Please fill out these boxes ê (Do not worry if you cannot see all your text once you press return) a specification of the experimental system investigated (eg cell line, species name).
RNA-Seq, Proteomics and Phospohr samples were performed with at least an n=3 (biological repeats). Detailed histological analyses of mice was performed with n=4 or more. Kaplan survival curves were performed with total cohort of 126 (3D);total cohort of 42 (4A); total cohort of 234 (4F). RNA-Seq from mouse tumours was performed with at least n=4 biological samples. Mouse graphs include clearly labeled error bars for independent experiments and sample sizes. Unless justified, error bars should not be shown for technical replicates. if n< 5, the individual data points from each experiment should be plotted and any statistical test employed should be justified the exact sample size (n) for each experimental group/condition, given as a number, not a range; Each figure caption should contain the following information, for each panel where they are relevant:

B-Statistics and general methods
the assay(s) and method(s) used to carry out the reported observations and measurements an explicit mention of the biological and chemical entity(ies) that are being measured. an explicit mention of the biological and chemical entity(ies) that are altered/varied/perturbed in a controlled manner.
a statement of how many times the experiment shown was independently replicated in the laboratory.
Any descriptions too long for the figure legend should be included in the methods section and/or with the source data.
In the pink boxes below, please ensure that the answers to the following questions are reported in the manuscript itself. Every question should be answered. If the question is not relevant to your research, please write NA (non applicable). We encourage you to include a specific subsection in the methods section for statistics, reagents, animal models and human subjects.
definitions of statistical methods and measures: a description of the sample collection allowing the reader to understand whether the samples represent technical or biological replicates (including how many animals, litters, cultures, etc.).

Reporting Checklist For Life Sciences Articles (Rev. June 2017)
This checklist is used to ensure good reporting standards and to improve the reproducibility of published results. These guidelines are consistent with the Principles and Guidelines for Reporting Preclinical Research issued by the NIH in 2014. Please follow the journal's authorship guidelines in preparing your manuscript.