The tumor-suppressive long noncoding RNA DRAIC inhibits protein translation and induces autophagy by activating AMPK

  1. Shekhar Saha
  2. Ying Zhang
  3. Briana Wilson
  4. Roger Abounader
  5. Anindya Dutta

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Long noncoding RNAs (lncRNAs) are long RNA transcripts that do not code for proteins and have been shown to play a major role in cellular processes through diverse mechanisms. DRAIC, a lncRNA that is downregulated in castration-resistant advanced prostate cancer, inhibits the NF-κB pathway by inhibiting the IκBα kinase. Decreased DRAIC expression predicted poor patient outcome in gliomas and seven other cancers. We now report that DRAIC suppresses invasion, migration, colony formation and xenograft growth of glioblastoma-derived cell lines. DRAIC activates AMP-activated protein kinase (AMPK) by downregulating the NF-κB target gene GLUT1, and thus represses mTOR, leading to downstream effects, such as a decrease in protein translation and increase in autophagy. DRAIC, therefore, has an effect on multiple signal transduction pathways that are important for oncogenesis, namely, the NF-κB pathway and AMPK–mTOR–S6K/ULK1 pathway. The regulation of NF-κB, protein translation and autophagy by the same lncRNA explains the tumor-suppressive role of DRAIC in different cancers and reinforces the importance of lncRNAs as emerging regulators of signal transduction pathways.

This article has an associated First Person interview with the first author of the paper.

Article activity feed

  1. Version published to 10.1242/jcs.259306
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Reviewer #1 (Evidence, reproducibility and clarity (Required)):

    Saha and colleagues investigated the functions of the long non-coding RNA (lncRNA) DRAIC in malignant glioma. They find that DRAIC expression decreases cell migration/invasion and tumorsphere/colony formation in vitro, and tumor growth in vivo using established cell lines. Mechanistically, DRAIC is known to inhibit NF-kB signaling and the authors demonstrate that DRAIC activates AMPK leading to repression of mTOR, which decreases protein synthesis and increases autophagy. This is a solid study highlighting a potentially interesting pathway of tumor growth and invasion in brain tumors.

    __Answer: __We appreciate Reviewer 1 for the positive feedback of our study

    __Major Comments: __

    1) It is unclear whether the presented values (mean +/- SD) in the histograms refer to repeat measurements (in which case n = 1) or independent experiments (n>1). The number of replicate experiments is not stated in the methods or figure legends. This must be included.

    __Answer: __We want to thank Reviewer 1 for pointing out this omission. We have now included this information in the Materials and methods and in figure legends section.

    2) I don't think the immunoblot for p62 in Fig. 5C shows a convincing increase following DRAIC knockout, so the statement on p.8 should be revised.

    __Answer: __We have revised the statement to say: Consistent with DRAIC decrease being associated with a decrease in autophagic flux, and despite a decrease in p62 mRNA, the level of P62 protein is increased in three of the DRAIC KO prostate cancer cells (Fig. 5C, KO1, KO2, KO4 compared to WT) and unchanged in the other two.

    3) On p.8/Fig.5 the authors make a case that increased DRAIC levels increase lysosomal degradation of autophagosome core proteins LC3 II / p62 (resulting in decreased protein levels of both), while simultaneously increasing gene expression of LC3B and p62 (causing increased mRNA levels). The data for DRAIC overexpression fit this logic fairly well (even though I think more work is needed to fully support this claim), but I am finding it difficult to reconcile the DRAIC knockout data with this scenario - here, loss of DRAIC results in increased protein levels to decreased autophagy, but also decreased gene expression. To fully support this argument, rescue experiments would be needed using FoxO3a knockout/overexpression.

    __Answer: __ Note that the mRNA level is not always correlated with protein expression. This is particularly true for LC3 and p62, whose protein levels are significantly affected by the extent of fusion of autophagosomes and lysosomes and subsequent degradation in autophagolysosomes. Thus, although the mRNA of these genes is decreased in DRAIC KO cells (Fig. 5E), the proteins are increased (Fig. 5C) because of decrease of autophagic flux (and decrease of degradation in the autophagolysosomes).

    The overexpression of FoxO3a in the DRAIC KO cells will not restore mRNA levels of LC3 or p62, because we show in Fig. 4H that FoxO3 phosphorylation by AMPK is suppressed by DRAIC KO. This phosphorylation is important for the induction of LC3 or p62 mRNA by FoxO3.

    FoxO3 knockout or knockdown in DRAIC OE cells should decrease LC3B or p62 mRNA in Fig. 5D, but it is already known from the Literature that FoxO3a is necessary for inducing LC3B or p62 mRNA. Cell Metab. 2007 Dec;6(6):458-71. doi: 10.1016/j.cmet.2007.11.001.PMID: 18054315.

    4) Similarly, the data supporting increased autophagy following DRAIC overexpression (Fig. 5F/G) are a bit weak and lack controls (is the LC3B-GFP overlapping with endogenous LC3B and autophagosomes? Was the transfection efficiency comparable? Is there fusion with lysosomes?). In the absence of stronger data, the authors should temper their claims that DRAIC increases autophagy.

    __Answer: __ LC3B of fusion protein LC3B-GFP is known to overlap with the p62 puncta (similar to endogenous LC3B). This result is in Fig. 4A of the citation that we have now added (Proc Natl Acad Sci U S A. 2016 Nov 22;113(47): E7490-E7499. doi: 10.1073/pnas.1615455113. Epub 2016 Oct 17)

    To support our hypothesis that DRAIC OE induces more autophagy compared to empty vector, we used Bafilomycin A1 in Figure 5B to inhibit the autophagosome and lysosome fusion. We see the accumulation of more LC3B upon treatment with Bafilomycin A1 in the DRAIC OE cells (compared to EV containing U251 cells), consistent with the idea that autophagosome-lysosome fusion is increased by DRAIC OE.

    5) No information is provided on animal numbers used in this study. How many mice were used per cohort? Were male and female mice used? Authors should follow ARRIVE guidelines in reporting animal experiments. The method for calculating tumor volume needs to be specified.

    __Answer: __ We have included the details about the animal study in methods section of our modified manuscript .

    6) Student's T-test is inappropriate for comparisons of more than two groups (i.e. all experiments using DRAIC knockout cells) - for these experiments a Kruskal Wallis test or ANOVA should be used. Did the authors test for normal distribution of their data? This may affect statistical testing and should be taken into consideration.

    __Answer: __ We have now modified our statistical calculation and included in the statistical analysis section in our modified manuscript.

    Minor Comments:

    7) Authors mention that DRAIC expression is undetectable in immortalized astrocytes and GBM cancer stem cells (Fig. S1). What is the source of these cells and how were they cultured?

    __Answer: __The immortalized astrocytes and GBM stem cells and their culture conditions is now described.

    8) The immunoblot in Fig. 3D could be replaced with a slightly lower exposure to make the difference between WT and DRAIC KO more obvious.

    __Answer: __We have now replaced the immunoblot with lower exposure.

    9) Some immunoblots in Fig. 3 (panel E, p-S6K and S6K; panel H, actin) are not of the best quality and an effort should be made to replace them.

    __Answer: __We have now replaced the immunoblot p-S6K as reviewer mentioned.

    10) Why are different loading controls used in Fig. 3 (a-Tubulin v actin)?

    Answer: We use multiple loading control to make sure that we are not underestimating or overestimating changes in the experimental protein because of unexpected changes in the loading controls.

    11) Compared to other blot images in the same figure (e.g. Fig. 3E), the bands for p-mTOR and mTOR in Fig. 3F look compressed and should be shown appropriately sized.

    __Answer: __ We have modified the Figure as reviewer suggested.

    12) The layout of Fig. 4 is somewhat confusing. I would suggest organizing this according to DRAIC overexpression in A172 and U373 cells versus DRAIC knockout in LNCaP cells. Each immunoblot should be clearly labelled with the corresponding cell line, and it should be clearly explained why p-FoxO3a was tested in U251 cells, rather than A172/U373 as in the rest of the figure.

    __Answer: __We thank the reviewer for the constructive criticism. We have labeled all the cell lines in the Figure as reviewer suggested. We have now systematically alternated the prostate cancer cells (for KO) and the GBM cells (for OE), as we looked at each relevant marker. We have now included the western blot for p-FoxO3a from another glioblastoma cell line U373. Please find the modified Figure 4K for p-FoxO3a.

    13) Labelling of immunoblot in Fig. 5B is confusing and should be improved.

    __Answer: __We have modified the Fig. 5B to make the label clearer.

    14) Changes in GLUT1 expression (Fig. 7A) should be validated on the protein level.

    __Answer: __We have included the immunoblot for GLUT1 from DRAIC KO cells in Figure 7B. GLUT1 protein is increased upon DRAIC KO.

    Reviewer #1 (Significance (Required)):

    The authors describe a novel link between the lncRNA DRAIC and AMPK activation through inhibition of NF-kB-mediated regulation of GLUT1. This study extends their previous work on DRAIC inhibition of NF-kB in prostate cancer (Saha et al. Cancer Res 2020). There is one study describing DRAIC effects on growth and invasion in glioma cell lines (Li et al. Eur Rev Med Pharmacol Sci 2020), but the work presented by Saha and colleagues contains stronger experimental data and a more detailed and previously undescribed mechanism.

    The current study presents a mechanistic advance that increases the understanding of tumor growth and protein synthesis in cancer cells. The data presented in the study are not supported by in vivo experiments (other than suppression of tumor growth by DRAIC overexpression), validation in human tissue and/or primary patient-derived human glioblastoma cells, or even substantial rescue experiments. This limits the influence of the work on the field. I'm also not sure how transferable findings from DRAIC knockout in prostate cancer cell lines are to glioma, although the results are mostly complementary to the data from glioma cell lines. This is particularly relevant to the proposed mechanism of GLUT1 regulation by NF-kB, as the bulk of experimental data in Figures 6 and 7 was generated in prostate cancer cell lines and is only poorly validated in glioma cells. The study results will be most relevant for researchers investigating cell signaling pathways and autophagy in cancer.

    __Answer: We like to thank reviewer for the positive comments on our study. The DRAIC KO experiments of Fig. 6 and 7 cannot be done in glioma cells, because as we show if Supp. Fig. S1, there are no glioma cells or GSC that express DRAIC to levels comparable to LnCaP. We have shown that GLUT1 mRNA decreases in the glioma cells when DRAIC is overexpressed (Supp. Fig. S4. We also show in Fig. 7G that AMP levels increase when DRAIC is overexpressed in glioma cells.

    __

    Reviewer #2 (Evidence, reproducibility and clarity (Required)):

    In this manuscript, the authors describe DRAIC as a lncRNA downregulated in prostate cancer. They postulate that DRAIC expression surpasses invasion, migration and growth. Mechanistically, the authors show that DRAIC activates AMPK by suppressing NFkB target gene TOR and indirectly impacting translation and autophagy. Collectively the observation is interesting and robust. However, I have several technical requests, particularly regarding the mechanistic part of the paper.

    __Answer: __ We appreciate the positive feedback. We have addressed the reviewer’s concerns in our modified manuscript.

    Major Comments:

    1) The authors should rescue Ko phenotypes by over expressing DRAIC to consider potential off target effects.

    __Answer: __DRAIC OE alone is sufficient to have exactly the opposite effect as DRAIC KO in protein translation (Fig. 3C-F), so DRAIC OE will rescue the effect of DRAIC KO. We make a similar argument for all the phenotypes, including mTOR, S6K and ULK1(S757) phosphorylation (Fig. 3G-J), AMPK and FoxO3a phosphorylation (Fig. 4B-C; J-L), autophagic flux (Fig. 5B, C) and effects on LC3B and p62 mRNAs (Fig. 5D, E). The same is true for our published phenotypes of DRAIC KO on invasion, migration and NF-kB activity (Saha, Cancer Research, 2020)

    2) The blots showing TOR and ULK1 phosphorylation need to be repeated. This is an important part of the paper and I feel that these blots are hard to interpret. p-S6K typically run a bit higher in gels. there may be a technical problem.

    __Answer: __We are not sure which specific blots the reviewer is referring to, and it is possible that the blots the other reviewers pointed to are the ones under question. We have changed those blots so that the results are clear.

    3) GLUT1-related results are interesting, but the authors should provide genetic evidence that the effects are mediated by GLUT1. How do we know that glucose uptake is indeed upregulated upon knockout?

    __Answer: __In Fig. 7 C-F we show that the effects of DRAIC KO on invasion, protein translation, AMP levels and AMPK activity are reversed by the GLUT1 inhibitor Bay-876. This is a cleaner result than using siRNA to knockdown GLUT1. siRNAs can have off-target activity and sometimes cannot decrease a protein sufficiently below the threshold necessary to see reversal of action.

    Minor Comments:

    4) The figures need to be updated. FOnts are all different, lots of unaligned graphs, quality of the blots are poor.

    __Answer: __ We have updated the Figures and changed fonts as reviewer mentioned.

    Reviewer #2 (Significance (Required)):

    The observation is interesting, but the mechanism is incompletely understood. This is a nice addition to the literature, even without the mechanism.

    __Answer: __ We want to thank the reviewer for the constructive criticism.

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):

    **Summary:**

    Shaha and colleagues present a study demonstrating the tumor suppressive role of DRAIC, a long non-coding RNA transcript, through transmission of the signal from IKK/NF-kB to the AMPK/mTOR pathway via regulation of GLUT1 expression. The inhibition of mTOR by this pathway results in the reduction of protein translation, cellular invasion and activation of autophagy. Several diseases and models as well as multiple genetic and pharmacological manipulations were used to investigate the mechanisms at play. The manuscript is well written and the experiments are well designed. The conclusions are supported by the results. The following major and minor comments should be addressed:

    __Answer: __We appreciate the reviewer for the positive comments on our study.

    Major Comments:

    1) In addition to reporting the effect of DRAIC overexpression on tumor volume, the authors should present survival studies with one or more models.__

    Answer: __We thought of doing the survival study in our glioblastoma model but unfortunately, the tumor growth is very rapid (exceeding the size permitted by our IACUC in 2-3 weeks). The animal ethics welfare committee did not allow us to keep the mice for a longer time to perform the survival study.

    2) Since the authors study metabolic energy sensor pathways, related to glycolysis, it would be important to perform some of the key experiments in physiological level of glucose: e.g., pmTOR, pAMPK, LC3-II expression level in DRAIC overexpressing and deficient cells.

    Answer: The concentration of glucose in plasma is 1G/L, while that of the RPMI medium is 2G/L. We do not think we are too far from the physiological levels of glucose.

    3) In addition to RT-PCR data, GLUT1 protein levels should be investigated in the different DRAIC expressing cells.

    Answer: We have incorporated the GLUT1 protein expression data from DRAIC KO cells in Figure 7B and DRAIC overexpressing cells in supplementary Figure 4G-H. The blots from the same gels were split into different panels, the loading control GAPDH remain same in Figure 4K and Supplementary Figure S4H.

    4) The effect of DRAIC on GLUT1 expression is also measured in condition of glucose saturation, which does not reflect disease state. The decrease of GLUT1 in response to DRAIC overexpression and the increased GLUT1 level in DRAIC deficient cells should be investigated in physiological levels of glucose.

    Answer Same as above. We are near physiological levels of glucose.

    __Minor Comments:

    __

    5) All the data are generated with established cell lines (e.g., U87) but more clinically relevant models, such as patient-derived primary cells like the ones used in Fig. S1, could be used to replicate some of the key findings.

    __Answer: __As we showed in Fig. S1 that DRAIC is not expressed in glioma stem cells, and so knockout experiments are not possible. We believe that the knockout experiments are the most relevant to this paper because they do not run the risk of artefacts from overexpression of an RNA far beyond physiological levels.

    6) Also please provide further details about the patient-derived cells from Fig. S1.

    __Answer: __ We have mentioned the details of the cell lines in our modified manuscript.

    7) The statistical analysis section states that the number of measurements is indicated however I don't see the sample size of the experiments.

    __Answer: __We have now incorporated the number the experiments in our modified text.

    __Reviewer #3 (Significance (Required)): __ The study reports a new model of regulation of tumor via long non-coding RNA. This article adds to the growing literature The topic and content of the article is relevant and significant to the field of tumor research but the significance and impact could be enhanced with the use of more physiologically relevant models and conditions as pointed in the major comments.

    __Answer: __ We want to thank the reviewer for the positive feedback on our study.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    Summary:

    Shaha and colleagues present a study demonstrating the tumor suppressive role of DRAIC, a long non-coding RNA transcript, through transmission of the signal from IKK/NF-kB to the AMPK/mTOR pathway via regulation of GLUT1 expression. The inhibition of mTOR by this pathway results in the reduction of protein translation, cellular invasion and activation of autophagy. Several diseases and models as well as multiple genetic and pharmacological manipulations were used to investigate the mechanisms at play. The manuscript is well written and the experiments are well designed. The conclusions are supported by the results. The following major and minor comments should be addressed:

    Major comments:

    1. In addition to reporting the effect of DRAIC overexpression on tumor volume, the authors should present survival studies with one or more models.
    2. Since the authors study metabolic energy sensor pathways, related to glycolysis, it would be important to perform some of the key experiments in physiological level of glucose: e.g., pmTOR, pAMPK, LC3-II expression level in DRAIC overexpressing and deficient cells.
    3. In addition to RT-PCR data, GLUT1 protein levels should be investigated in the different DRAIC expressing cells.
    4. The effect of DRAIC on GLUT1 expression is also measured in condition of glucose saturation, which does not reflect disease state. The decrease of GLUT1 in response to DRAIC overexpression and the increased GLUT1 level in DRAIC deficient cells should be investigated in physiological levels of glucose.

    Minor comments:

    1. All the data are generated with established cell lines (e.g., U87) but more clinically relevant models, such as patient-derived primary cells like the ones used in Fig. S1, could be used to replicate some of the key findings.
    2. Also please provide further details about the patient-derived cells from Fig. S1.
    3. The statistical analysis section states that the number of measurements is indicated however I don't see the sample size of the experiments.

    Significance

    The study reports a new model of regulation of tumor via long non-coding RNA. This article adds to the growing literature The topic and content of the article is relevant and significant to the field of tumor research but the significance and impact could be enhanced with the use of more physiologically relevant models and conditions as pointed in the major comments.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    In this manuscript, the authors describe DRAIC as a lncRNA downregulated in prostate cancer. They postulate that DRAIC expression surpasses invasion, migration and growth. Mechanistically, the authors show that DRAIC activates AMPK by suprressing NFkB target gene TOR and indirectly impacting translation and autophagy. Collectively the observation is interesting and robust. However, I have several technical requests, particularly regarding the mechanistic part of the paper.

    • The authors should rescue Ko phenotypes by over expressing DRAIC to consider potential off target effects.
    • The blots showing TOR and ULK1 phosphorylation need to be repeated. This is an important part of the paper and I feel that these blots are hard to interpret. p-S6K typically run a bit higher in gels. there may be a technical problem.
    • GLUT1-related results are interesting but the authors should provide genetic evidence that the effects are mediated by GLUT1. How do we know that glucose uptake is indeed upregulated upon knockout?

    Minor:

    The figures need to be updated. FOnts are all different, lots of unaligned graphs, quality of the blots are poor.

    Significance

    The observation is interesting, but the mechanism is incompletely understood. This is a nice addition to the literature, even without the mechanism.

  5. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Saha and colleagues investigated the functions of the long non-coding RNA (lncRNA) DRAIC in malignant glioma. They find that DRAIC expression decreases cell migration/invasion and tumorsphere/colony formation in vitro, and tumor growth in vivo using established cell lines. Mechanistically, DRAIC is known to inhibit NF-kB signaling and the authors demonstrate that DRAIC activates AMPK leading to repression of mTOR, which decreases protein synthesis and increases autophagy. This is a solid study highlighting a potentially interesting pathway of tumor growth and invasion in brain tumors.

    Major comments:

    • It is unclear whether the presented values (mean +/- SD) in the histograms refer to repeat measurements (in which case n = 1) or independent experiments (n>1). The number of replicate experiments is not stated in the methods or figure legends. This must be included.
    • I don't think the immunoblot for p62 in Fig. 5C shows a convincing increase following DRAIC knockout, so the statement on p.8 should be revised.
    • On p.8/Fig.5 the authors make a case that increased DRAIC levels increase lysosomal degradation of autophagosome core proteins LC3 II / p62 (resulting in decreased protein levels of both), while simultaneously increasing gene expression of LC3B and p62 (causing increased mRNA levels). The data for DRAIC overexpression fit this logic fairly well (even though I think more work is needed to fully support this claim), but I am finding it difficult to reconcile the DRAIC knockout data with this scenario - here, loss of DRAIC results in increased protein levels to decreased autophagy, but also decreased gene expression. To fully support this argument, rescue experiments would be needed using FoxO3a knockout/overexpression.
    • Similarly, the data supporting increased autophagy following DRAIC overexpression (Fig. 5F/G) are a bit weak and lack controls (is the LC3B-GFP overlapping with endogenous LC3B and autophagosomes? Was the transfection efficiency comparable? Is there fusion with lysosomes?). In the absence of stronger data, the authors should temper their claims that DRAIC increases autophagy.
    • No information is provided on animal numbers used in this study. How many mice were used per cohort? Were male and female mice used? Authors should follow ARRIVE guidelines in reporting animal experiments. The method for calculating tumor volume needs to be specified.
    • Student's T-test is inappropriate for comparisons of more than two groups (i.e. all experiments using DRAIC knockout cells) - for these experiments a Kruskal Wallis test or ANOVA should be used. Did the authors test for normal distribution of their data? This may affect statistical testing and should be taken into consideration.

    Minor comments:

    • Authors mention that DRAIC expression is undetectable in immortalized astrocytes and GBM cancer stem cells (Fig. S1). What is the source of these cells and how were they cultured?
    • The immunoblot in Fig. 3D could be replaced with a slightly lower exposure to make the difference between WT and DRAIC KO more obvious.
    • Some immunoblots in Fig. 3 (panel E, p-S6K and S6K; panel H, actin) are not of the best quality and an effort should be made to replace them.
    • Why are different loading controls used in Fig. 3 (a-Tubulin v actin)?
    • Compared to other blot images in the same figure (e.g. Fig. 3E), the bands for p-mTOR and mTOR in Fig. 3F look compressed and should be shown appropriately sized.
    • The layout of Fig. 4 is somewhat confusing. I would suggest organizing this according to DRAIC overexpression in A172 and U373 cells versus DRAIC knockout in LNCaP cells. Each immunoblot should be clearly labelled with the corresponding cell line, and it should be clearly explained why p-FoxO3a was tested in U251 cells, rather than A172/U373 as in the rest of the figure.
    • Labelling of immunoblot in Fig. 5B is confusing and should be improved.
    • Changes in GLUT1 expression (Fig. 7A) should be validated on the protein level.

    Significance

    The authors describe a novel link between the lncRNA DRAIC and AMPK activation through inhibition of NF-kB-mediated regulation of GLUT1. This study extends their previous work on DRAIC inhibition of NF-kB in prostate cancer (Saha et al. Cancer Res 2020). There is one study describing DRAIC effects on growth and invasion in glioma cell lines (Li et al. Eur Rev Med Pharmacol Sci 2020), but the work presented by Saha and colleagues contains stronger experimental data and a more detailed and previously undescribed mechanism.

    The current study presents a mechanistic advance that increases the understanding of tumor growth and protein synthesis in cancer cells. The data presented in the study are not supported by in vivo experiments (other than suppression of tumor growth by DRAIC overexpression), validation in human tissue and/or primary patient-derived human glioblastoma cells, or even substantial rescue experiments. This limits the influence of the work on the field. I'm also not sure how transferable findings from DRAIC knockout in prostate cancer cell lines are to glioma, although the results are mostly complementary to the data from glioma cell lines. This is particularly relevant to the proposed mechanism of GLUT1 regulation by NF-kB, as the bulk of experimental data in Figures 6 and 7 was generated in prostate cancer cell lines and is only poorly validated in glioma cells. The study results will be most relevant for researchers investigating cell signaling pathways and autophagy in cancer.

    Reviewer keywords:

    neurooncology, cancer stem cells, signaling pathways in cancer

  6. Version published to 10.1101/2021.06.04.447165v1 on bioRxiv