HIF-2α-dependent TGFBI promotes ovarian cancer chemoresistance by activating PI3K/Akt pathway to inhibit apoptosis and facilitate DNA repair process

Abstract

Hypoxia-mediated chemoresistance plays a crucial role in the development of ovarian cancer (OC). However, the roles of hypoxia-related genes (HRGs) in chemoresistance and prognosis prediction and theirs underlying mechanisms remain to be further elucidated. We intended to identify and validate classifiers of hub HRGs for chemoresistance, diagnosis, prognosis as well as immune microenvironment of OC, and to explore the function of the most crucial HRG in the development of the malignant phenotypes. The RNA expression and clinical data of HRGs were systematically evaluated in OC training group. Univariate and multivariate Cox regression analysis were applied to construct hub HRGs classifiers for prognosis and diagnosis assessment. The relationship between classifiers and chemotherapy response and underlying pathways were detected by GSEA, CellMiner and CIBERSORT algorithm, respectively. OC cells were cultured under hypoxia or transfected with HIF-1α or HIF-2α plasmids, and the transcription levels of TGFBI were assessed by quantitative PCR. TGFBI was knocked down by siRNAs in OC cells, CCK8 and in vitro migration and invasion assays were performed to examine the changes in cell proliferation, motility and metastasis. The difference in TGFBI expression was examined between cisplatin-sensitive and -resistant cells, and the effects of TGFBI interference on cell apoptosis, DNA repair and key signaling molecules of cisplatin-resistant OC cells were explored. A total of 179 candidate HRGs were extracted and enrolled into univariate and multivariate Cox regression analysis. Six hub genes (TGFBI, CDKN1B, AKAP12, GPC1, TGM2 and ANGPTL4) were selected to create a HRGs prognosis classifier and four genes (TGFBI, AKAP12, GPC1 and TGM2) were selected to construct diagnosis classifiers. The HRGs prognosis classifier could precisely distinguish OC patients into high-risk and low-risk groups and estimate their clinical outcomes. Furthermore, the high-risk group had higher percentage of Macrophages M2 and exhibited higher expression of immunecheckpoints such as PD-L2. Additionally, the diagnosis classifiers could accurately distinguish OC from normal samples. TGFBI was further verified as a specific key target and demonstrated that its high expression was closely correlated with poor prognosis and chemoresistance of OC. Hypoxia upregulated the expression level of TGFBI. The hypoxia-induced factor HIF-2α but not HIF-1α could directly bind to the promoter region of TGFBI, and facilitate its transcription level. TGFBI was upregulated in cisplatin-sensitive and resistant ovarian cancer cells in a cisplatin time-dependent manner. TGFBI interference downregulated DNA repair-related markers (p-p95/NBS1, RAD51, p-DNA-PKcs, DNA Ligase IV and Artemis), apoptosis-related marker (BCL2) and PI3K/Akt pathway-related markers (PI3K-p110 and p-Akt) in cisplatin-resistant OC cells. In summary, the HRGs prognosis risk classifier could be served as a predictor for OC prognosis and efficacy evaluation. TGFBI, upregulated by HIF-2α as an HRG, promoted OC chemoresistance through activating PI3K/Akt pathway to reduce apoptosis and enhance DNA damage repair pathway.

Keywords: Akt; Chemoresistance; DNA repair; Hypoxia; Ovarian cancer.

Figure 1

Prognostic value of the hypoxia risk signature in OC. (a) The univariate Cox regression analysis of 12 hypoxia-related genes. (b) The univariate Cox regression analysis of 6 hypoxia-related genes. (c) The survival probability of prognosis model based on the risk score in training cohort. (d) The area under the curve of prognosis model based on the risk score in training cohort. (e) Distribution of risk score, OS, and OS status of the 6 prognostic hypoxia risk gene signatures in the training cohort. (f) The heatmap of the 6 prognostic hypoxia risk gene signatures in the training cohort.

Figure 2

The role of hypoxia-related genes in hypoxic condition. (a) The area under the curve for four diagnostic models. (b) The mRNA expression level of TGFBI, AKAP12, GPC1 and TGM2 in normoxia and hypoxia condition. (c) The mRNA expression level of TGFBI, AKAP12, GPC1 and TGM2 after transfected with HIF1a-plasmid. (d) The mRNA expression level of TGFBI, AKAP12, GPC1 and TGM2 after transfected with HIF2a-lentivirus. (e) The protein expression level of TGFBI after transfected with siHIF2a. (f) The Pearson and Spearman correlation analysis of TGFBI and EPAS1. (g) The promotor region of TGFBI and binding motif of HIF2a. (h) Expression level of TGFBI in ovarian cancer and normal tissues. (i) K–M survival curves of TGFBI in ovarian cancer. (j–m) Validation of TGFBI at the translational level using the Human Protein Atlas (HPA) database (IHC).

Figure 3

The role of TGFBI in ovarian cancer proliferation, migration and invasion. (a) The protein expression level of TGFBI in normal ovarian cell line and 5 ovarian cancer cell lines. (b) The mRNA and protein expression level of TGFBI after knocking down by siRNA. The cell proliferation (c), migration (d) and invasion (e) after knocking down TGFBI in 3AO cell line. The blots have been cropped to improve the conciseness and clarity of the display.

Figure 4

The role of TGFBI in ovarian cancer chemoresistance. (a) Evaluation of TGFBI sensitivity to chemotherapy drugs on tumor. (b) Cell survival between A2780 and A2780/CDDP followed by the concentration gradient stimulation of cisplatin. (c) The mRNA and protein expression level of TGFBI in cisplatin-sensitive ovarian cancer cell line-A2780 and cisplatin-resistant ovarian cancer cell line-A2780/CDDP. (d) The protein expression level of TGFBI after time gradient of cisplatin in A2780 (IC₅₀ as 119.2 μM). (e) The protein expression level of TGFBI after time gradient of cisplatin in A2780/CDDP (IC₅₀ as 258.8 μM). (f) The protein expression level of TGFBI after knocking down by siRNA in A2780/CDDP. (g) Cell survival after A2780/CDDP were transfected with siRNA, followed by the concentration gradient stimulation of cisplatin (0, 2.5, 5, 10, 20, 40, 80, 160, 320 and 640 μM).

Figure 5

TGFBI promoted chemoresistance through inhibiting apoptosis and facilitating DNA damage repair via activating PI3K/Akt signaling pathway. (a) The protein expression level of MRP1 and MDR1 in A2780/CDDP after TGFBI knockdown. (b) The protein expression level of Homologous Recommendation DNA repair related markers in A2780/CDDP after TGFBI knockdown. (c) The protein expression level of Non-Homologous End Joining (NHEJ) DNA repair related markers in A2780/CDDP after TGFBI knockdown. (d) The KEGG enrichment pathway analysis of TGFBI-related differentially expressed genes in OC. (e) The protein expression level of PTEN, PI3K/p110, PI3K/p85, p-Akt, Akt, BCL2, Bax and Beclin1 in A2780/CDDP after TGFBI knockdown.

Figure 6

A model of the role of TGFBI in chemoresistance.