Fig. 1

Identification and validation of candidate genes responsible for conferring multidrug resistance in HNSCC. A, Transcriptome data mining from eight independent gene expression microarray studies (comparing HNSCC tumour and normal oral tissues samples) identified 28 short-listed genes involved in the regulation of matrix remodelling, immune modulation, cell proliferation & differentiation, stem cell renewal, epigenetic programming and genomic instability. B, Pharmacological dose-response screening for multidrug-resistant differentially expressed genes in three cell lines (SVpgC2a, SVFN8 and CaLH2) each between parental WT cells and their corresponding drug-resistant strains for four chemotherapeutic drugs (R1: Cisplatin, R2: 5FU, R3: PTX and R4: DTX). For each cell/drug resistant combination strain, differential expression of the 28 genes were measured by RT-qPCR to identify drug dose-dependent response genes (see Additional File 2: Fig. S1-S12). Table shows the top differentially expressed genes in corresponding drug-resistant cell strains are shown in coloured (statistically significant) gene symbols (non-significant genes in white text). Genes with underlines indicate downregulation in drug-resistant cells, otherwise, upregulation. The list of 28 genes were ranked in descending order according to their frequency of occurrence as top significant genes across the whole panel of 12 cell/drug combinations. C, Validation of gene expression levels of TOP2A, DNMT1, INHBA and NEK2 in two different HNSCC patient cohorts: Top panel, a UK cohort with adjacent margin (n = 98) and HNSCC tumour core tissues (n = 123). The relative mRNA expression levels of each of the four genes were measured using RT-qPCR against two reference genes (YAP1 and POLR2A) measured in duplicate wells. Data were plotted as beeswarm dot-plot with box-and-whisker overlays (minimum, box: median, and 25–75%, percentiles and maximum). Statistical t-test were performed between the margin and tumour samples and all four genes showed P < 1 × 10^− 5. Bottom panel: Differential expression of the four genes in HNSC cohort from TCGA/GTEx transcriptomic data comparing margin (n = 44) and HNSCC (n = 519) were all significantly upregulated in tumour (P < 0.01, one-way ANOVA). D, Effects of siRNA gene silencing of TOP2A, DNMT1, INHBA and NEK2 on reversal of chemoresistance (or re-sensitisation). Summary of relative chemoresistance following gene-specific siRNA knockdown in SVpgC2a, SVFN8 and CaLH2 cells, each resistant to either cisplatin, 5FU, PTX or DTX. Relative chemoresistance was calculated as fold-change between IC₅₀ of drug-resistant cells and IC₅₀ of corresponding WT cells. IC₅₀ drug potency values of each of the four chemotherapeutic drugs on WT and drug-resistant cells were measured using crystal violet cell viability assay (Additional File 2: Fig. S14-S16). Statistical t-test was performed between controls (mock transfection/+H₂O and siCTRL were combined as one group) vs. each of the gene-specific siRNA and their corresponding P-values are indicated (*<0.05; **<0.01; ***<0.001; ns, not significant) within the charts