Cancers acquire resistance to systemic treatment with platinum-based chemotherapy (eg, cisplatin [CDDP]) as a result of a dynamic intratumoral heterogeneity (ITH) and clonal repopulation. However, little is known about the influence of chemotherapy on ITH at the single-cell level. Here, mapping the transcriptome of cancers treated with CDDP by scRNA-seq, Karolinska Institutet researchers uncovered a novel gene, COX7B, associated with platinum-resistance, and surrogate marker, CD63. Knockdown of COX7B in cancer cells decreased the sensitivity of CDDP whereas overexpression recovered the sensitivity of CDDP. Low COX7B levels correlated with higher mortality rates in patients with various types of cancer and were significantly associated with poor response to chemotherapy in urinary bladder cancer. Tumor samples from patients, who underwent CDDP therapy, showed decreased COX7B protein levels after the treatment. Analyzing scRNA-seq data from platinum-naïve cancer cells demonstrated a low-COX7B subclone that could be sorted out from bulk cancer cells by assaying CD63. This low-COX7B subclone behaved as cells with acquired platinum-resistance when challenged to CDDP. These results offer a new transcriptome landscape of platinum-resistance that provides valuable insights into chemosensitivity and drug resistance in cancers, and identify a novel platinum resistance gene, COX7B, and a surrogate marker, CD63.
scRNA‐seq identifies innate platinum‐resistant cells in human urinary bladder cancer
A, Principal component analysis (PC1 vs PC2) of the COX7B levels listed in Figure 1B (n = 249). The high‐magnification images are of the boxed region with and without pseudo‐coloring. B, Workflow for determining a cell surface marker associated with the COX7B level in human urinary bladder cancers. C, Principal component analysis (PC1 vs PC2) of the CD63 levels listed in Figure 1B (n = 249). Receiver operating characteristic analysis (upper panels) of CD63 for detecting the low COX7B level (<25th percentile) group in all 249 single cells (D) or TCGA provisional for 408 human urinary bladder cancer patients (E). AUC values are indicated. Scatter plots in lower panels show COX7B vs CD63 levels in each dataset. r, Spearman’s correlation coefficient. F, Scatter plots depicting the FACS analysis of COX7B and CD63 from 5637 cells. Fluorescence‐minus 5637 cells (gray) were used as control. G, The COX7B protein levels of sorted 5637 cells for CD63 were determined by FACS re‐analysis. H, Histogram (blue) plots depicting the FACS analysis of CD63 from 5637 cells. Data are the mean percentages of three independent experiments for sorted 5637 CD63high and CD63low cells. Fluorescence‐minus 5637 cells (gray) are used as a control. Right panel shows effect of CDDP on the viability of FACS‐sorted 5637 CD63high and CD63low cells measured as relative fold change in IC50 (the half maximal inhibitory concentration). Un‐sorted 5637 cells were used as a control. *P < 0.05, compared to sorted 5637 CD63low cells. The P value from the two‐tailed Student’s t test. I, Histogram (blue) depicting the FACS analysis of CD63 from 5637PR cells. Data are the mean percentages of three independent experiments for sorted 5637PR CD63high and CD63low cells. Fluorescence‐minus 5637PR cells (gray) were used as a control. Right panel shows effect of CDDP on the viability of FACS‐sorted 5637PR CD63high and CD63low cells measured as relative fold change in IC50. Un‐sorted 5637PR cells were used as a control. **P < 0.05, compared to sorted 5637PR CD63low cells