Large pan-cancer cell screen coupled to (phospho-)proteomics underscores high-dose vitamin C as a potent anti-cancer agent
Curation statements for this article:-
Curated by eLife
eLife assessment
This useful study utilizes proteomics analysis across a large panel of 51 cancer cell lines to elucidate mechanisms underlying the sensitivity of cancer cells to high-dose vitamin C (Ascorbate). While the associations between specific molecular pathways and sensitivity to ascorbate are interesting, a major limitation is that the study is largely descriptive and incomplete, lacking evidence on the molecular underpinnings of cancer cells' sensitivity to high-dose vitamin C.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Cancer Biology (eLife)
Abstract
Increasing preclinical and clinical evidence has positioned high-dose vitamin C as a promising anti-cancer treatment that merits more clinical attention. Multiple cytotoxicity mechanisms have been described, including pro-oxidant effects. To contribute to the preclinical understanding of the broad pan-cancer effects of high-dose vitamin C in a global manner, we determined the IC50 of a large panel of cancer cell lines (n=51) representing 7 solid tumour types and generated proteome data. The majority of cell lines were highly sensitive (IC50 range 0.036-10mM, mean 1.7 ± 0.4 mM), well below a clinically achievable dose. The proteome data (>5000 proteins per sample), showed that high sensitivity is associated with proliferation, as indicated by functional enrichment of cell cycle, RNA splicing and chromatin organization, while lower sensitivity is linked to extracellular vesicles, glycolysis, fatty acid metabolism and mitochondria. Moreover, (phospho-)proteome analysis of on-treatment vitamin C effects on four pancreatic ductal adenocarcinoma (PDAC) cells dosed at a range of IC50 values (Hs766 T, 2 mM; Capan-2, 0.6 mM; PANC-1, 0.14 mM and Suit-2, 0.1 mM) revealed, next to cell line specific effects, down-modulation of AKT-MTOR signalling and immune suppressive signalling, while IFN-α response was enhanced upon vitamin C. Altogether, our comprehensive pharmacological and (phospho-)proteome analysis is the first to assess cancer vulnerabilities and effects of vitamin C on a large cancer cell line panel and underscores the potential of high-dose vitamin C as an anti-cancer agent.
Article activity feed
-
Author Response
We thank eLife Senior Editor and reviewers for the comprehensive evaluation and constructive comment on our manuscript. We are grateful that all 3 reviewers recognize the value of the large pharmacological and proteomics screen of 51 cancer cell lines in relation to vitamin C IC50 values. As reviewer 1 points out, our findings are of interest as high dose vitamin C is in clinical trials. Most importantly, we show that all 51 cell lines tested can be killed at a dose range that is achievable by intravenous administration in the clinic. These pharmacological findings underscore high-dose vitamin C as a potent anti-cancer agent. Moreover, we provide an elaborate description of functional terms associated with the vitamin C IC50 values in the different cell panels (Figs 1-5) and the common denominators across panels (Figs 6, …
Author Response
We thank eLife Senior Editor and reviewers for the comprehensive evaluation and constructive comment on our manuscript. We are grateful that all 3 reviewers recognize the value of the large pharmacological and proteomics screen of 51 cancer cell lines in relation to vitamin C IC50 values. As reviewer 1 points out, our findings are of interest as high dose vitamin C is in clinical trials. Most importantly, we show that all 51 cell lines tested can be killed at a dose range that is achievable by intravenous administration in the clinic. These pharmacological findings underscore high-dose vitamin C as a potent anti-cancer agent. Moreover, we provide an elaborate description of functional terms associated with the vitamin C IC50 values in the different cell panels (Figs 1-5) and the common denominators across panels (Figs 6, 7 and 8), thereby enhancing our biological insights of sensitivity to vitamin C treatment. This study indeed is of descriptive nature and our large scale pharmacological and proteomics scale dataset should be seen as a resource for further research. The raw and processed data will be available in the ProteomeXchange repository (accession number and reviewer password were provided before) and the resubmission will include all processed proteome and phosphoproteome data as a supplementary file.
It is beyond the scope of our study to do mechanistic studies with knock-downs to see if we can further sensitize cancer cell lines that are less sensitive. We do not call these cell lines resistant as cell growth can be inhibited at a clinically achievable dose.
In our detailed rebuttal we will follow up on the suggestion of reviewer 1 to put our data also in the context of NCI-60 growth inhibition data for other cytotoxic agents. This will expand our comparative analysis to cisplatin in the lung cancer panel (Fig 5A) where we show that vitamin C IC50 values and cisplatin IC50 values are not one-on-one correlated as one of the most cisplatin resistant NSCLC cell lines in our panel was very sensitive to high dose vitamin C. Furthermore, we will clarify method details and annotate mutational status in our panels and explore potential genomic associations to high-dose vitamin C sensitivity as presented in previous studies (e.g. mutant BRAF and/or KRAS tumors, https://doi.org/10.1126/science.aaa5004).
Finally, we will critically read the manuscript and add references where needed.
-
-
-
eLife assessment
This useful study utilizes proteomics analysis across a large panel of 51 cancer cell lines to elucidate mechanisms underlying the sensitivity of cancer cells to high-dose vitamin C (Ascorbate). While the associations between specific molecular pathways and sensitivity to ascorbate are interesting, a major limitation is that the study is largely descriptive and incomplete, lacking evidence on the molecular underpinnings of cancer cells' sensitivity to high-dose vitamin C.
-
Reviewer #1 (Public Review):
Summary:
The study by Valles-Marti et al. was aimed at elucidating mechanisms of high-dose vitamin C (Ascorbate) sensitivity using proteomics of a large panel of cancer cell lines. The study is primarily based on correlating protein expression to vitamin C sensitivity based on IC50 from cell viability studies. As expected, cancer type-specific proteome patterns emerge and the authors conclude that some pan-cancer pathways, such as proliferation correlate with high sensitivity to VitC. In a subset of PDAC cells proteomics and phospho proteomics were also carried out following vitamin C treatment, albeit those studies did not identify significant changes in response to treatment.Strengths:
The premise for the work is of interest as high dose vitamin C is in clinical trials and thus studies investigating …Reviewer #1 (Public Review):
Summary:
The study by Valles-Marti et al. was aimed at elucidating mechanisms of high-dose vitamin C (Ascorbate) sensitivity using proteomics of a large panel of cancer cell lines. The study is primarily based on correlating protein expression to vitamin C sensitivity based on IC50 from cell viability studies. As expected, cancer type-specific proteome patterns emerge and the authors conclude that some pan-cancer pathways, such as proliferation correlate with high sensitivity to VitC. In a subset of PDAC cells proteomics and phospho proteomics were also carried out following vitamin C treatment, albeit those studies did not identify significant changes in response to treatment.Strengths:
The premise for the work is of interest as high dose vitamin C is in clinical trials and thus studies investigating mechanisms of sensitivity and potential resistance mechanisms to this therapy are of interest to the field. The authors have collected large proteomic datasets on some of the most common cancer cells used and these data may be a useful resource for others when made publicly available. Although this is not necessarily novel, since proteomics data sets for some of the included cell lines are already available.Weaknesses:
The title suggests that the proteomics data presented "underscores high-dose vitamin C as a potent anti-cancer agent" However, while the proteomic data are extensive, it is my assessment that without further validation there are no clear pathways identified by the presented proteomics data that conclusively determine vitamin C sensitivity.A major question arising from this work is how specific the proteomics data reflect sensitivity to vitamin C over general sensitivity to other cytotoxic agents. It would be of interest to compare the correlation of proteomic data and ascorbate sensitivity to the sensitivity of cell lines to other cytotoxic agents. (e.g. comparison to NCI-60 growth inhibition data). In other words, do the proteomic data that correlate with ascorbate sensitivity simply reflect susceptibility to other cytotoxic agents? The comments that vitamin C toxicity is not dependent on underlying histological or genetic subtypes of cancers ("one size fits all") suggest this.
The genetic backgrounds of tumor cells have not been taken into consideration in the analysis and how this may influence VitC susceptibility. An example that comes to mind is KEAP1/Nrf2 aberrations in lung cancer.
The study would be significantly strengthened if some of the proteins identified were further validated in eliciting low or high sensitivity to Vitamin C. Of particular interest are proteins that have functions related to known mechanisms of action of Vitamin C toxicity, such as iron homeostasis. Some of the metabolic-related protein changes are also of interest. For example, HCCS expression is mentioned several times as being associated with lower sensitivity to ascorbate. Providing experimental evidence that this protein is of significance to Vitamin C sensitivity and if this is due to its effects on iron and subsequent generation of ROS in response to VitC would be of significance.
Similarly, an interesting aspect of the findings is the authors' conclusion that proliferation is associated with Vitamin C sensitivity. The authors propose in their discussion that Vitamin C may be an attractive alternative to treat heavily pretreated and chemoresistant cancers. Thus it would be important to know which of the highly proliferative cell lines tested have a chemoresistance phenotype and are also more susceptible to Vitamin C toxicity. Perhaps partitioning the cells further into chemoresistant and sensitive cell lines to standard chemotherapy and then assessing which protein signatures are associated with Vitamin C sensitivity will allow for better elucidation of sensitivity mechanisms that are more relevant to using Vitamin C as an alternate therapy for chemoresistant tumors.
Following on from this, there is an interesting mechanistic question as to why more proliferative cells are more sensitive to vitamin C, and whether this is related to changes in metabolism and underlying changes in their steady-state levels of ROS. Further investigating this mechanistically based on the identified proteomic signatures could make the findings more significant.
Vitamin C can also generate H2O2 extracellularly in the presence of iron. Thus, Vitamin C toxicity could be affected by different abilities of the tumor cells to scavenge extracellular H2O2, such as different expression levels of extracellular antioxidant enzymes. Judging from the methods section, it does not appear that proteomic data include secreted proteins. Can the authors comment on how this may be a potential caveat?
In light of this, the strong effects of exogenous catalase addition on cell viability suggest that H2O2 may be produced by ascorbate in the media.
Similarly, can the authors comment on the cell culture conditions used to compare IC50s between cell lines, specifically if different media and FBS batches were used, as these have the potential to vary in metal/iron concentrations that might influence the pro-oxidant generation by high dose ascorbate in media. Specifically, have the authors looked into the iron content and how these different conditions may be contributing to intracellular H2O2 and extracellular H2O2 (AmplexRed) production in response to Vitamin C.
Other comments relate to methods:
How was ascorbate prepared? There is no mention of degassing of H2O and ensuring that H2O does not have mental impurities, which can lead to auto-oxidation.
The OxiSelect probe is based on DCFDA, which is an oxidant-sensitive probe that has been described to be fraught with artifacts. Thus it is advised to mention the caveats associated with the use of this probe (as outlined in PMCID: PMC3911769) and consider backing up these experiments with additional Oxidant probes.
-
Reviewer #2 (Public Review):
Summary:
The authors generated proteome profiles of 51 cancer cell lines treated with pharmacologic ascorbate. The idea was to identify players responsible for the sensitivity or relative resistance to ascorbate to delineate mechanisms of action of this potentially transformative new treatment.Strengths:
The proteomic profiles themselves. The identification of MAPK and mTOR as overrepresented proteomic elements and close correlations between proliferation, cell cycle mediators, and sensitivity to ascorbate indicate that rapidly proliferating cancer may be more sensitive to ascorbate. Also, the finding that sensitivity to ascorbate is correlated to different pathways in different types of cancer is interesting. For instance, in some pancreatic and lung cancers sensitivity seemed to be related to iron …Reviewer #2 (Public Review):
Summary:
The authors generated proteome profiles of 51 cancer cell lines treated with pharmacologic ascorbate. The idea was to identify players responsible for the sensitivity or relative resistance to ascorbate to delineate mechanisms of action of this potentially transformative new treatment.Strengths:
The proteomic profiles themselves. The identification of MAPK and mTOR as overrepresented proteomic elements and close correlations between proliferation, cell cycle mediators, and sensitivity to ascorbate indicate that rapidly proliferating cancer may be more sensitive to ascorbate. Also, the finding that sensitivity to ascorbate is correlated to different pathways in different types of cancer is interesting. For instance, in some pancreatic and lung cancers sensitivity seemed to be related to iron handling while in breast DNA damage/repair seemed to be most involved.Weaknesses:
The study is quite descriptive. Although the proteomes indicate what pathways are more or less represented after ascorbate challenge there is little mechanistic information about their relevance to the sensitivity to ascorbate. Since activity is not assessed, proteins may be present in higher or lower abundance but not necessarily at the peak of their activity. Also, many statements are made as "known facts" but no references are provided. -
Reviewer #3 (Public Review):
Summary:
In the manuscript titled "Large pan-cancer cell screen coupled to (phospho-)proteomics underscores high-dose vitamin C (VitC) as a potent anti-cancer agent," the authors use a combination of proteomics and cell viability assays to understand the effect of Vitamin C on different solid tumor models in 51 different cancer cell lines. They found that many cancer cell lines are sensitive to high-dose Vitamin C, with IC50 values in the micromolar to millimolar range. Given that Vitamin C, when administered intravenously, can reach 20mM, this suggests that Vitamin C could provide some benefits to patients. The authors also generate and analyze bulk proteomic data for all 51 cell lines. They perform statistical analysis of these data to identify proteins that are up or downregulated in sensitive vs …Reviewer #3 (Public Review):
Summary:
In the manuscript titled "Large pan-cancer cell screen coupled to (phospho-)proteomics underscores high-dose vitamin C (VitC) as a potent anti-cancer agent," the authors use a combination of proteomics and cell viability assays to understand the effect of Vitamin C on different solid tumor models in 51 different cancer cell lines. They found that many cancer cell lines are sensitive to high-dose Vitamin C, with IC50 values in the micromolar to millimolar range. Given that Vitamin C, when administered intravenously, can reach 20mM, this suggests that Vitamin C could provide some benefits to patients. The authors also generate and analyze bulk proteomic data for all 51 cell lines. They perform statistical analysis of these data to identify proteins that are up or downregulated in sensitive vs resistant cell lines in the same tumor and commonly across tumors. They then focus on PDAC cell lines and measure bulk and phosphoproteomics of PDAC cell lines 2, 4 and 24 hours after Vitamin C treatment.Strengths: The strengths of the study are the rather large datasets accumulated on bulk proteomics of 51 different cancer cell lines. The IC50 values of these cell lines in response to Vitamin C is also useful.
Weaknesses:
Though identifying targets to sensitize cancer cells to Vitamin C treatment is interesting, I felt the manuscript delved too much into listing off genes they found, with speculation on why the particular protein would be enriched in sensitive or resistant cell lines without testing any key claims experimentally.Major Issues
(1) The overall premise of the study is that proteins that are enriched in Vitamin C-sensitive cell lines point to mechanisms of sensitivity and those enriched in Vitamin C resistant lines underlie mechanisms of resistance. Yet this is never directly tested. To show that the authors would need to knockdown/knockout a gene enriched in resistant lines and show this sensitizes cells to Vitamin C treatment or overexpress a protein associated with resistance and show that this leads to resistance in an otherwise sensitive cell line.
(2) One of the key strengths of this study is the large datasets generated, namely the proteomics data for 51 different cell lines. Yet the data is not included as a supplement or uploaded to a public repository.
-
