Signatures of malignant cells and novel therapeutic targets revealed by single-cell sequencing in lung adenocarcinoma

Abstract

Background: Single-cell transcriptomics has been used to investigate various tumors to elucidate the molecular distinction of all cell type compositions of a complex mix.

Aims: This study aimed to investigate malignant-cell-specific genes to explore diagnostic and therapeutic biomarkers using single-cell transcriptomic data of lung adenocarcinoma.

Materials & methods: 10X single-cell RNA-seq data of fourteen patients with lung adenocarcinoma were analyzed. Genes that expressed differentially and those with higher confidence to distinguish tumor cells from normal cells were picked out using the ROC curves. The LASSO regression method was used to select most markedly correlated genes to predict the malignancy of every single cell within a model. We also conducted further experiments to determine their roles in lung cancer in vitro.

Results: Twenty two thousand four hundred and ninety one tumor and 181 666 normal single cells were analyzed where 369 genes were found to be specifically expressed in single malignant cells. Seventy of them, encoding secreted or membrane-bound proteins, showed involvement in cell-to-cell communications in tumor biology. KRT18 and the other six genes were identified as predictors to distinguish single malignant cells and were integrated to construct an accurate (96.1%) predicting model. Notably, IRX2, SPINK13, and CAPN8 outperformed the other four genes. Further experiments confirmed the upregulation of them in lung adenocarcinoma at both tissue and cell levels. Proliferative capacities of lung adenocarcinoma cells were attenuated by knocking-down of either of them. However, targeting CAPN8, IRX2, or SPINK13 hardly exerted a cytotoxic effect on these cells.

Discussion: Apart from the current model, similar tools were still warranted using single-cell RNA-seq data of more types of tumors. The three genes identified as potential therapeutic targets in the present study still need to be validated in more in lung cancer.

Conclusion: Our model can aid the analyses of single-cell sequencing data. CAPN8, IRX2, and SPINK13 may serve as novel targets of targeted and immune-based therapies in lung adenocarcinoma.

Keywords: lung adenocarcinoma; single-cell transcriptomics; therapeutic targets.