Pan-cancer proteomic analysis of vitamin C sensitivity.

(A) Study design of the pan-cancer proteome profiling of 51 human cell lines divided by cancer (sub)type panel: PDAC, LUAD, LUSC, SCLC, CRC, BC and PC, with the determination of vitamin C sensitivities. The candidate selection of sensitivity-associated proteins per panel was supported by two complementary statistical analyses: a Spearman correlation of protein expression with IC50 values (Corr), and a two-group comparison between protein expression in more and less sensitive cell lines (2g). (B) Number of identified proteins per sample. (C) (D) VitC sensitivities plotted as IC50s, measured by cell viability assays and represented (C) in millimolar (n=51 cell lines) and (D) in pmol/cell range (n=40 cell lines). Mean ± SEM plotted, average n ≥ 2. (E) Targeted analysis of potential VitC-associated processes(1). The significance by two group comparison and Spearman correlation is shown per panel. Proteins with at least one significant pvalue in at least one panel were included.

Pancreatic ductal adenocarcinoma VitC sensitivity and protein biomarker discovery analysis.

(A) Vitamin C sensitivities (IC50 pmol/cell) of n=12 PDAC cell lines. Epithelial and mesenchymal classification is annotated based on morphology and ssGSEA EMT signature. (B) Hierarchical clustering heatmap of proteins correlated with high and low VitC sensitivity (Spearman Correlation Rho ≥ or ≤ 0.7, pvalue ≤ 0.05, n=127 identified proteins). (C) Hierarchical clustering heatmap of proteins enriched in highly sensitive and lowly sensitive cell lines (2-group comparison, FC ≥ or ≤ 2, pvalue ≤ 0.05, n=299 identified proteins). (D), (E) Venn diagram describing the overlap between significant proteins found in both statistical analysis, 2-group comparison and Spearman Correlation for (D) high and (E) low VitC sensitivity, as input for the top 10 enriched biology terms. Terms for (D) high and (E) low VitC sensitivity are ranked on –log10 pvalue. (F), (G) Top 20 candidate markers associated with (F) high and (G) low VitC sensitivity ranked on log2FC. Gene ontology terms of known VitC related processes were further annotated. Significant proteins found in these VitC-related processes were also added.

Colorectal cancer VitC sensitivity and protein biomarker discovery analysis.

(A) Vitamin C sensitivities (IC50 mM) of n=13 CRC cell lines. CMS subtype information was retrieved from two recent studies(46,47). (B) Hierarchical clustering heatmap of proteins correlated with high and low VitC sensitivity (Spearman Correlation Rho ≥ or ≤ 0.6, pvalue ≤ 0.05, n=191 identified proteins). (C) Hierarchical clustering heatmap of proteins enriched in highly sensitive and lowly sensitive cell lines (2-group comparison, FC ≥ or ≤ 2, pvalue ≤ 0.05, n=275 identified proteins). (D), (E) Venn diagram describing the overlap between significant proteins found in both statistical analysis, 2-group comparison and Spearman Correlation for (D) high and (E) low VitC sensitivity, as input for the top 10 enriched biology terms. Terms for (D) high and (E) low VitC sensitivity are ranked on –log10 pvalue. (F), (G) Top 20 candidate markers associated with (F) high and (G) low VitC sensitivity ranked on log2FC. Gene ontology terms of known VitC related processes were further annotated. Significant proteins found in these VitC-related processes were also added.

Breast cancer VitC sensitivity and protein biomarker discovery analysis.

(A) Vitamin C sensitivities (IC50 pmol/cell) of n=5 BC cell lines. Molecular subtype was annotated based on a reference study(51). (B) Hierarchical clustering heatmap of proteins correlated with high and low VitC sensitivity (Spearman Correlation Rho ≥ or ≤ 0.7, pvalue ≤ 0.05, n=851 identified proteins). (C) Hierarchical clustering heatmap of proteins enriched in highly sensitive and lowly sensitive cell lines (2-group comparison, FC ≥ or ≤ 2, pvalue ≤ 0.05, n=573 identified proteins). (D), (E) Venn diagram describing the overlap between significant proteins found in both statistical analysis, 2-group comparison and Spearman Correlation for (D) high and (E) low VitC sensitivity, as input for the top 10 enriched biology terms. Terms for (D) high and (E) low VitC sensitivity are ranked on –log10 pvalue. (F), (G) Top 20 candidate markers associated with (F) high and (G) low VitC sensitivity ranked on log2FC. Gene ontology terms of known VitC related processes were further annotated. Significant proteins found in these VitC-related processes were also added.

Lung adenocarcinoma VitC sensitivity and protein biomarker discovery analysis.

(A) Vitamin C sensitivities (IC50 mM) of n=7 LUAD cell lines. Cisplatin sensitivities (IC50s μM) were further annotated. (B) Hierarchical clustering heatmap of proteins correlated with high and low VitC sensitivity (Spearman Correlation Rho ≥ or ≤ 0.7, pvalue ≤ 0.05, n=890 identified proteins). (C) Hierarchical clustering heatmap of proteins enriched in highly sensitive and lowly sensitive cell lines (2-group comparison, FC ≥ or ≤ 2, pvalue ≤ 0.05, n=891 identified proteins). (D), (E) Venn diagram describing the overlap between significant proteins found in both statistical analysis, 2-group comparison and Spearman Correlation for (D) high and (E) low VitC sensitivity, as input for the top 10 enriched biology terms. Terms for (D) high and (E) low VitC sensitivity are ranked on –log10 pvalue. (F), (G) Top 20 candidate markers associated with (F) high and (G) low VitC sensitivity ranked on log2FC. Gene ontology terms of known VitC related processes were further annotated. Significant proteins found in these VitC-related processes were also added.

Enriched common biology associated with VitC sensitivity among cancer cell line panels.

Overlapping significant (pvalue adjusted) gProfiler terms (GO:BP,CC, MF and REACTOME) in (A) high sensitive cells and (B) less sensitive cells from shared proteins between spearman correlation and 2-group comparison analyses per panel. Maximum of top forty terms ordered by adjusted pvalue were used to assess term overlap between panels. (C) Cancer hallmarks (ssGSEA) associated with VitC sensitivity across cancer panels. Heatmap built with spearman correlation coefficients (Rho) between IC50s and ssGSEA hallmark enrichment scores per panel. Significant pvalue is annotated with an asterisk. LUSC panel was excluded from this analysis due to a low sample size (non-reliable results).

Pan-cancer biology associated with high VitC sensitivity.

(A) Upset plot highlighting the overlap of proteins associated with high VitC sensitivity identified in each cancer panel. Overlap between panels is indicated with joined black circles. (B) Top 15 protein-protein interaction clusters and enriched biology. All proteins in panel A were used as input for the networks. Node colours refer to cell line panel, see Fig. 7A. Bold nodes refer to shared proteins between at least two cancer cell panels. See the expanded network in Fig. S5.

Pan-cancer biology associated with low VitC sensitivity.

(A) Upset plot highlighting the overlap of proteins associated with low VitC sensitivity identified in each cancer panel. Overlap between panels is indicated with joined black circles. (B) Top 15 protein-protein interaction clusters and enriched biology. All proteins in panel A were used as input for the networks. Node colours refer to cell line panel, see Fig. 8A. Bold nodes refer to shared proteins between at least two cancer cell panels. See the expanded network in Fig. S6.

Effects of VitC on the proteome and phospho-proteome of a subset of PDAC cells.

(A) Cell confluence images of PANC-1 and Capan-2 cell lines for treated (5mM) and untreated conditions. (B) Intracellular ROS levels upon multiple VitC concentrations in PANC-1 and Capan-2 cell lines after 10 minutes, 2h and 24h. For every time point, intracellular ROS levels are normalized to the levels of the untreated condition. Data from n=2 biological replicates. Statistical significance between timepoints was calculated by a paired t-test. (C) Cell growth of PANC-1 and Capan-2 upon VitC, and upon VitC followed by the exposure to catalase. (D) Representative images of PANC-1 cells VitC treated (2.5 and 20mM) with and without catalase exposure. (E) Workflow scheme of phospho-proteomic profiling of VitC treated cell lines at 2h, 4h and 24h. See Methods. (F) Unsupervised clustering based on ssGSEA results using hallmarks of cancer. Enrichment scores shown as z-scores. (G) Interesting ssGSEA enrichments between control and VitC samples after 24h. Colours depict cell lines in order of VitC sensitivity and VitC dose (0.1mM Suit-2, 0.14mM PANC-1, 0.6mM Capan-2, 2mM Hs 766T). (H) Phosphosite changes upon VitC at 2h, 4h and 24h, calculated by FC between control and VitC, of downstream regulators MAPK1/3 and AKT2 in all cell lines. Colours depict cell lines in order of VitC sensitivity and VitC dose (0.1mM Suit-2, 0.14mM PANC-1, 0.6mM Capan-2, 2mM Hs 766T). Rectangle indicates double phosphorylation required for MAPK1/3 activation. (I) Phosphosite changes upon VitC at 2h, 4h and 24h, calculated by FC between control and VitC in Hs 766T cell line. (J) Kinase activity changes upon VitC treatment at 2h, 4h and 24h timepoints. Log2 fold changes were calculated for the top 20 ranked kinases based on the cumulative INKA score. A fold change cut-off of (−) 1.5 was used to determine which kinases are altered upon VitC treatment. In grey, FC values cannot be determined due to the absence of detected INKA kinases in at least one of the replicates, either in the VitC or control group. Arrows indicate members of MAPK and PI3K/AKT signalling.