Genome-wide transcriptome analysis and drug target discovery reveal key genes and pathways in thyroid cancer metastasis

Abstract

Introduction: Metastasis is the major cause of thyroid cancer morbidity and mortality. However, the mechanisms are still poorly understood.

Methods: We performed genome-wide transcriptome analysis comparing gene expression profile of metastatic thyroid cancer cells (Met) with primary tumor cells established from transgenic mouse models of papillary thyroid cancer (PTC), follicular thyroid cancer (FTC), poorly differentiated thyroid cancer (PDTC), and anaplastic thyroid cancer (ATC).

Results: Genes involved in tumor microenvironment (TME), inflammation, and immune escape were significantly overexpressed in Met cells. Notably, IL-6-mediated inflammatory and PD-L1 pathways were highly active in Met cells with increased secretion of pro-inflammatory and pro-metastatic cytokines such as CCL2, CCL11, IL5, IL6, and CXCL5. Furthermore, Met cells showed robust overexpression of Tbxas1, a thromboxane A synthase 1 gene that catalyzes the conversion of prostaglandin H2 to thromboxane A2 (TXA2), a potent inducer of platelet aggregation. Application of aspirin, a TXA2 inhibitor, significantly reduced lung metastases. Mertk, a member of the TAM (Tyro, Axl, Mertk) family of RTKs, was also overexpressed in Met cells, which led to increased MAPK activation, epithelial-mesenchymal transition (EMT), and enrichment of cancer stem cells. Braf-mutant Met cells developed resistance to BRAFV600E inhibitor PLX4720, but remained sensitive to β-catenin inhibitor PKF118-310.

Conclusion: We have identified several overexpressed genes/pathways in thyroid cancer metastasis, making them attractive therapeutic targets. Given the complexity of metastasis involving multiple pathways (PD-L1, Mertk, IL6, COX-1/Tbxas1-TXA2), simultaneously targeting more than one of these pathways may be warranted to achieve better therapeutic effect for metastatic thyroid cancer.

Keywords: CD274 (PD-L1); IL6; MERTK; TBXAS1; thyroid cancer metastasis.

Figure 1

Experimental lung metastasis. (A) Schematic diagram summarizing experimental lung metastatic procedure. BVE, BVE^Trp53null, KGD, or KGD^Cdkn2anull cells (1 x 10⁶ cells) were injected to tail vain of 5 nude mice for each group. Six weeks after injection, lung metastatic tumors were collected to establish Met1 cell lines. The established Met1 cells were re-injected (1 x 10⁶ cells) to a new group of nude mice (n=5 for each group) via tail vain for enrichment of cells with high metastatic potential. Lung metastatic tumors were harvested three weeks following injection to establish Met2 cell lines. Met1, Met2 and primary tumor cells were subject to RNA sequencing to identify differentially expressed genes (DEGs). (B) Lung metastatic foci after tail vain injection of BVE thyroid cancer cells (H&E staining). Metastatic foci are indicated by black arrows and tumor infiltrating monocytes and macrophages are indicated by green arrows.

Figure 2

Genome-wide transcriptome analysis of DEGs in Met2 cells. (A) Heatmap of DEGs present in all 4 Met2 cells (BVE-Met2, BVE^Trp53null-Met2, KGD-Met2, and KGD^Cdkn2a_null-Met2). Several key genes involved in inflammation, immune checkpoint regulation and cancer stem cells are indicated by an arrow. (B) Top 20 Enriched ontology clusters of DEGs by Metascape analysis. Regulation of cytokine production, inflammatory response, phagocytosis pathway, negative regulation of immune system, and regulation of IL6 production are the top 5 enriched pathways. (C) PPI network and hub genes identification by NetworkAnalyst. STRING interactome with high confidence score of 900 (of maxium1000) was chosen for hub gene identification. Eighteen genes with the degree of a node >10 (bigger red nodes) were selected as hub genes (D) PPI network based on InnateDB. Six genes with the degree of a node >10 (bigger red nodes) were selected as hub genes. (E) Kaplan-Meier analyses of hub gene expression on disease-specific survival of PTC patients. Overexpression of 4 hub genes (Tnf, Nckap1, Nlrp3, and Card11) was associated with poor disease-specific survival. TCGA-THCA mRNA dataset (n=498) was used for Kaplan-Meier analysis.

Figure 3

Cytokine/Chemokine production by Met2 cells. (A) Cytokine/Chemokine production was detected in conditioned media of both primary and Met2 cells using MILLIPLEX mouse ELISA assay. Increased production of inflammatory cytokines and chemokines were observed in Met2 cells. (B) Western blot analysis of IL-6 and IL-6 receptor expression in primary and Met2 cells. Increased IL-6 and IL-6 receptor expression were observed in Met2 cells. (C) FACS analysis of Ep-CAM and CD11b staining of both primary and Met2 cells. Ep-CAM is an epithelial cell marker, while CD11b is a myeloid cell marker highly expressed on the surface of innate immune cells, including macrophages and neutrophils. CD11b-positive cells were present in less than 5% of Met2 cell population *P<0.05, **P<0.01, ***P<0.001.

Figure 4

Contribution of Cd274 and Tbxas1 in pulmonary pre-metastatic niche formation. (A) Overexpression of Cd274 (PD-L1) and Tbxas1 mRNA in Met2 cells as compared to primary cells detected by qRT-PCR analysis.*p<0.01. (B) Western blot analysis of CD274 expression. Increased CD274 expression were observed in Met2 cells. (C) Microscopic metastatic foci of KGD-Met2 cells. Significant infiltration of RBCs, lymphocytes, and macrophages are noted at the interface between metastatic tumor foci and lung. Tumor cells are indicated by green arrows and marked as TC. Increased infiltration of monocytes and macrophages is indicated by black arrows. Increased RBC aggregation is noted and platelets aggregation on tumor cells is indicated by a blue arrow. (D) Lung metastasis of BVE^Trp53null-Met2 upon low dose aspirin treatment. The metastatic foci on the lung surface were counted and plotted as a bar graph. Significant reduction of lung metastatic foci was observed when mice were given aspirin (25 mg/kg) by oral gavage for 4 weeks. Data are presented as mean ± SEM. Representative metastatic foci are indicated by an arrow. (E) Histology (H&E staining) of lung metastases of thyroid tumor cells. The size and number of metastatic foci were reduced after aspirin treatment.

Figure 5

Up-regulation of MerTK-mediated signaling. (A) Mertk mRNA overexpression in Met2 cells detected by qRT-PCR analysis. *p<0.01. (B) Western blot analysis of E-cadherin, vimentin, p-Erk, and p-AKT expression in Met2 cell lines. Increased expression of vimentin and p-Erk and decreased expression of E-cadherin were demonstrated in all Met2 cells. Increased p-AKT expression was present only in the BVE^trp53null-Met2 cells. (C) FACS analysis of CD44/CD24 expression in Met2 cells. Increased CD24^lowCD44^high staining was observed in Met2 cells. (D) Tumorspheres formation in ultralow attachment stem cells culture conditions. Representative tumorspheres are shown. (E) The number of spheres was counted and plotted as a bar graph. The number of tumorspheres was increased in Met2 cells, *p<0.01. (F) Inhibition of tumorspheres by PKF118-310. Significant reduction of tumorspheres was noted upon treatment with 1µM PKF118-310, *p<0.01. (G) Sensitivity of BVE-Met2 cells to BRAF^V600E and β-catenin inhibitors. BVE and BVE-Met2 cells were cultured in different low cell seeding numbers. A representative colony formation assay was shown. (H) The number of colonies was counted and plotted as a bar graph. Increased colony formation was observed in low seeding number of BVE-Met2 cells, which were resistant to BRAF^V600E inhibitor PLX4720 but remained sensitive to β-catenin inhibitor PKF118-310. *P<0.05, ****P<0.0001, NS, not statistically significant. (I) Flow cytometric analysis of apoptotic BVE and BVE-Met2 cells after treatment with BRAF^V600E or β-catenin inhibitors. (J) Flow cytometric analysis of apoptotic BVE and BVE-Met2 cells after combined treatment of BRAF^V600E and β-catenin inhibitors.

Figure 6

Schematic diagram of metastatic cascade of thyroid cancer cells and potential drug targets. Genetic changes and epigenetic modifications drive tumor progression and metastatic reprogramming. To evade immune elimination and survive in the blood circulation, metastatic cells express high levels of ‘Don’t eat me’ signals such as Cd274 (PD-L1) and Cd52 as well as Tbxas1 and Mertk. At the metastatic site, the metastatic cells induce a cytokine storm by secreting a large amount of inflammatory cytokines and chemokines to help form a pre-metastatic niche and eventual colonization at the distant site.