Identification of downstream signaling cascades of ACK1 and prognostic classifiers in non-small cell lung cancer

Abstract

Activated Cdc42-associated kinase 1 (ACK1) is an oncogene in multiple cancers, but the underlying mechanisms of its oncogenic role remain unclear in non-small cell lung cancer (NSCLC). Herein, we comprehensively investigated the ACK1-regulated cell processes and downstream signaling pathways, as well as its prognostic value in NSCLC. We found that ACK1 gene amplification was associated with mRNA levels in The Cancer Genome Atlas (TCGA) lung cancer cohort. The Oncomine databases showed significantly elevated ACK1 levels in lung cancer. In vitro, an ACK1 inhibitor (dasatinib) increased the sensitivity of NSCLC cell lines to AKT or MEK inhibitors. RNA-sequencing results demonstrated that an ACK1 deficiency in A549 cells affected the MAPK, PI3K/AKT, and Wnt pathways. These results were validated by gene set enrichment analysis (GSEA) of data from 188 lung cancer cell lines. Using Cytoscape, we dissected 14 critical ACK1-regulated genes. The signature with the 14 genes and ACK1 could significantly dichotomize the TCGA lung cohort regarding overall survival. The prognostic accuracy of this signature was confirmed in five independent lung cancer cohorts and was further validated by a prognostic nomogram. Our study unveiled several downstream signaling pathways for ACK1, and the proposed signature may be a promising prognostic predictor for NSCLC.

Keywords: ACK1; NSCLC; TCGA; dasatinib; prognosis.

Figure 1

The implication of ACK1 in NSCLC. Genetic alterations of the ACK1 gene in the TCGA-LUAD and TCGA-LUSC cohorts (A) (https://www.cbioportal.org). The association between ACK1 gene copy number and mRNA expression levels (B). Significantly elevated mRNA expression levels of the ACK1 gene in lung cancer in comparison with normal tissues in the independent cohorts from the Oncomine database (C). Immunohistochemistry of ACK1 in lung cancer (D, Human Protein Atlas). Abbreviation: FC, fold change.

Figure 2

Inhibitory efficiency of ACK1 inhibitor alone or in combination with MK-2206/selumetinib on NSCLC cell lines. Proliferation assay of the A549 cell line (A), H358 cell line (B), and H23 cell line (C) treated with drugs as indicated. Combined therapy performed significantly better than single agents in suppressing cell survival. * and ** denoted P<0.05 and P<0.01, respectively. Data are represented as the mean ± SD.

Figure 3

Knockdown (KD) of ACK1/TNK2 in A549 cells, followed by RNA-seq. ACK1 was silenced using three lentivirus-mediated shRNAs (A). The shRNA showing the highest efficiency of the ACK1 gene knockdown was used for subsequent experiments. Overlapping genes were identified in the negative control (NC) and KD groups (B). The volcano plot (C) indicated the significantly up- and downregulated genes after the silencing of ACK1 [absolute value of log2 (fold change) ≥1, P<0.001]. Based on differentially expressed genes (DEGs), three NC and three KD samples (shACK1/TNK2) were well clustered (D). Gene Ontology enrichment analysis of DEGs (E). KEGG pathway annotation of DEGs (F).

Figure 4

Analysis of 219 hub genes with degree ≥10. GO biological process analysis using the BinGo plug-in of Cytoscape (A). KEGG pathway analysis using the ClueGo plug-in of Cytoscape (B). Volcano plot of the enriched pathways with P values and enrichment scores (C). Altered genes in the Wnt signaling pathway (D). Altered genes in the MAPK signaling pathway (E). Orange and cyan rectangles indicate the upregulation and downregulation of genes, respectively, after the ACK1 gene knockdown. Only DEGs (fold change ≥2 and adjusted P value <0.001) are colored.

Figure 5

Analysis of the ACK1 signaling pathways. RNA-seq data of 188 lung cancer cell lines were retrieved from the CCLE database. GSEA was performed after dividing cell lines into ACK1^high and ACK1^low groups by the average ACK1 expression level (A). Principal component analysis (PCA) of the 57 hub genes (degree≥20) was carried out in the TCGA-LUAD (B) and TCGA-LUSC (C) versus normal tissues. The most significant 14-gene module was derived from DEGs, which captured 50 coexpressed genes in the TCGA-LUAD cohort (D). Most of the 64 genes were enriched in the ubiquitin-mediated proteolysis and the proteasome (E), which was validated by GSEA using CCLE lung cancer cell data (F).

Figure 6

Prognostic values of the gene signature comprising ACK1 and the 14 genes of the most significant module in the TCGA lung cancer patients. Patients were classified into low (green) and high (red) risk groups according to risk scores. Heatmaps of 15 gene expression profiles in the low and high risk LUAD (A) and LUSC (B) patients. Comparison of the 15 gene expression levels between high (blue) and low (red) risk groups in LUAD (C) and LUAD (D). Univariate (upper panel) and multivariate (lower panel) Cox regression analyses in LUAD (E) and LUSC (F).

Figure 7

Prognostic values of the 15-gene signature in the TCGA lung cancer patients. The distribution of risk scores and survival statuses of patients in the LUAD (A) and LUSC cohorts (B). Kaplan-Meier survival curves of patients defined by low and high risk scores in LUAD (C) and LUSC (D). ROC curves with different characteristics of patients, as indicated in LUAD (E) and LUSC (F).

Figure 8

Validation of the prognostic power of risk scores in the independent lung cancer cohorts. Kaplan-Meier survival curves of 6 independent lung cancer cohorts (A). Performance of risk scores in the NCI (B), KOHNO (C), HOU (D), BILD (E), and ZHU (F) cohorts.

Figure 9

Prognostic nomogram for TCGA lung cancer cohorts. Nomogram for evaluating the survival probability of TCGA-LUAD patients (A). The calibration curves for predicting patient survival in TCGA-LUAD (B, D) and TCGA-LUSC (C, E). Overall survival (OS) derived from the nomogram is plotted on the x-axis, and actual OS is displayed on the y-axis. A plot approaching the 45° dashed line would show an ideal calibration model indicating the perfect concordance between the predicted probabilities and the actual survival.