Fig. 2

Pan-cancer bioinformatics data mining and prognostic analysis. A, Bioinformatics data mining from Oncomine databases on differential gene expressions of INHBA, NEK2, TOP2A and DNMT1 across 20 different human cancer types as indicated. The number within each coloured box indicates the number of significant unique studies. Red and blue colours indicate gene expression upregulation and downregulation, respectively. Cell colour scale is determined by the best gene rank percentile for the analyses (dark red/blue = top 1%; red/blue = top 5%; pale red/blue = top 10%). B, Kaplan-Meier RNA-seq transcriptome prognostic analysis for INHBA, NEK2, TOP2A and DNMT1 on 21 different human cancer types. Hazard ratios (with logrank P < 0.05) extracted from KM-plotter database were plotted here as beeswarm dot-plot with box-and-whisker overlays (minimum, box: median, and 25–75%, percentiles and maximum) to demonstrate individual marker prognostic value for each cancer type. Dark red indicates marker associated with poor prognosis and green for markers associated with good prognosis (abbreviations listed in panel E). *Note: outlier (Thymoma, Log₁₀ HR = 8.66) was plotted outside the chart for reference. C, Table listing corresponding number (and %) of cancer type analysed for each marker with poor or good prognosis. D, Individual Kaplan-Meier plots for INHBA, NEK2, TOP2A and DNMT1 in HNSCC tumour samples (n = 500) with hazard ratio (HR) and logrank P values as shown within each panel. E, All four markers (alone and in combinations) were examined for their synergistic prognostic values for each cancer type, the most significant prognostic marker (or markers in combinations) are tabulated here. Hazard ratio (HR) values were shown with colour scales applied to indicate poor (dark red) or good (green) prognosis with their corresponding logrank P values (colour scales indicate their relative levels of significance) for each cancer type (n = the number of samples in each cancer type). Marker abbreviations: I, INHBA; N, NEK2; T, TOP2A and D, DNMT1). F-I, Drug library screen to identify drug-gene interactions for counteracting chemoresistance in HNSCC cells. Nine compounds (D1-D9; Additional File 2: Fig. 17) were selected. Shown here are two compounds (D4, and D7) with dose-dependent inhibition on both INHBA (F) and NEK2 (G) gene expression in WT and CR CaLH2 cells. Each datapoint represents relative gene expression (mean ± SEM) of quadruplicates quantified using RT-qPCR. Drug potencies (IC₅₀) on respective gene inhibition are displayed within each panel. H, Chemical structure and identity of compounds D4 and D7. I, Re-sensitisation of CR CalH2 cells by addition of D4 or D7 (1 µM) to cisplatin dose-response measured using AlamarBlue cell viability assay. Each datapoint represents a mean ± SEM of n = 6 replicates. Cisplatin potency values (mean IC₅₀ ± SEM of n = 6) in the absence or presence of either D4 or D7 single concentration (1 µM) are shown within the figure. Statistical t-test was performed between cisplatin alone vs. cisplatin + D4 or cisplatin + D7 and their corresponding P-values are indicated within the figure as ***<0.001.