TWIST1/miR-584/TUSC2 pathway induces resistance to apoptosis in thyroid cancer cells

Abstract

TWIST1, a transcription factor, plays a pivotal role in cancer initiation and progression. Anaplastic thyroid carcinoma (ATC) is one of the deadliest human malignancies; TWIST1 is overexpressed in ATC and increases thyroid cancer cell survival, migration and invasion. The molecular mechanisms underlying the effects of TWIST1 are partially known. Here, using miRNome profiling of papillary thyroid cancer cells (TPC-1) ectopically expressing TWIST1, we identified miR-584. We showed that TWIST1 directly binds miR-584 using chromatin immunoprecipitation. Importantly, miR-584 was up-regulated in human ATC compared to papillary thyroid carcinoma (PTC) and normal thyroid samples. Overexpression of miR-584 in TPC cells induced resistance to apoptosis, whereas stable transfection of anti-miR-584 in TPC-TWIST1 and 8505C cells increased the sensitivity to apoptosis. Using bioinformatics programs, we identified TUSC2 (tumor suppressor candidate 2) as a novel target of miR-584. TUSC2 mRNA and protein levels were decreased in TPC miR-584 and increased in TPC-TWIST1 anti-miR-584 cells. Luciferase assays demonstrated direct targeting. Restored expression of TUSC2 rescued the inhibition of apoptosis induced by miR-584. Finally, qRT-PCR and immunohistochemical analysis showed that TUSC2 was down-regulated in ATC and PTC samples compared to normal thyroids. In conclusion, our study identified a novel TWIST1/miR-584/TUSC2 pathway that plays a role in resistance to apoptosis of thyroid cancer cells.

Keywords: TUSC2; TWIST1; anaplastic thyroid carcinoma; apoptosis; miR-584.

Figure 1. Expression levels of the indicated miRNAs in TPC-TWIST1 cells

qRT-PCR analysis of the indicated miRNAs in TPC-TWIST1 cells compared to vector control cells. The expression level of each miRNA was normalized to the level of U6 snRNA, and fold changes were evaluated using the 2^{− Δ Ct} method, assuming that the value of the vector control cells is equal to 1. The average results of three independent experiments are reported with error bars indicating standard deviation.

Figure 2. TWIST1 directly binds a region upstream of miR-584

a. Schematic representation of all E-box binding sites (indicated as —) in the genomic locus of miR-584. Black ▼ indicate the E-boxes amplified in ChIP assays. b. ChIP assay, followed by qRT-PCR, was performed with TPC-TWIST1 or control (pcDNA) cells. Columns represent the average of four independent experiments □± SD. * p< 0.05. c. The amplicon of the GAPDH promoter was used as a negative control. The figure represents four independent experiments.

Figure 3. Expression of miR-584 in thyroid tissue samples

qRT-PCR analysis of miR-584 in normal thyroid (NT) (n=10), PTC (n=11), and ATC (n=6) snap-frozen tissue samples. The expression level of miR-584 was normalized to the U6 snRNA levels. The level of miR-584 in each sample was measured by comparing its fluorescence threshold with the average fluorescence threshold of the NT samples. The average results of quadruplicate experiments are plotted. ***, p< 0.001.

Figure 4. Effects of ectopic expression of miR-584 in TPC cells

a. TPC miR-584 and TPC miR-null cell lines weretreated with different doses of doxorubicin and staurosporine for 48 and 24 hours, respectively, and counted in triplicate. The IC₅₀ values were calculated with GraphPad software. The average of three independent experiments ± SD is shown. b. Dual staining of apoptotic cells with 7-AAD and Annexin V after treatment with doxorubicin and staurosporine. The percentages of apoptotic cells (P3) or of viable cells (P2) are indicated. c. TPC miR-584 and TPC miR-null cells were treated with 300 nM of doxorubicin or with 500 nM of staurosporine, lysed after 48 and 24 hours, respectively, and blotted with the antibodies for Cleaved CASPASE 3 or α-TUBULIN.

Figure 5. Effects of silencing of miR-584 in TPC-TWIST1 and 8505C cells

The indicatedcells were treated with increasing doses of doxorubicin or staurosporine and counted after 48 and 24 hours, respectively. Values represent the average of four independent experiments ± SD.

Figure 6. TUSC2 is a direct target of miR-584

a. Immunoblotting of TUSC2 and α-TUBULIN in TPC miR-584 or control cells. b. qRT-PCR of TUSC2 in TPC miR-584 or control cells; data were normalized to the level of POLR2A assuming that the value of TPC miR-null is equal to 1. c. Immunoblotting of TUSC2 and α-TUBULIN in TPC-TWIST1 anti-miR-584 or control cells. d. qRT-PCR of TUSC2 in TPC-TWIST1 anti-miR-584 or control cells. e. Predicted duplex formation between human 3′UTR-TUSC2 and miR-584;the strikethrough sequence of 3′UTR of TUSC2 corresponds to the predicted binding site of miR-584 deleted in the mutant 3′UTR-TUSC2 plasmid (3′UTR del). f. Luciferase activity of TUSC2 wild-type (3′UTR WT) or deleted (3′UTR del) reporter genes in TPC cells stably transfected with miR-584 or control. Luciferase activity was normalized with a co-transfected β-actin plasmid. **, p< 0.01.

Figure 7. TUSC2 partially rescues the phenotype induced by miR-584

a. TPC miR-584 cells were transiently transfected with TUSC2 plasmid or with the vector control (CMV) and analyzed for TUSC2 protein expression by Western blotting. An antiα-TUBULIN antibody was used for normalization. b. TPC miR-584 cells were transfected with TUSC2 or vector controlandtreated with 50 nM of doxorubicin and 300 nM of staurosporine; after 48 and 24 hours, respectively, cells were stained with trypan blue and counted in triplicate. The percentage viability± SD is shown, **, p < 0.01.

Figure 8. Expression of TUSC2 in thyroid tissue samples

a. Immunohistochemical analysis of TUSC2 protein expression in normal and malignant thyroid tissues. Representative sections (20X magnification) of PTC, ATC and contralateral NT stained with an anti-TUSC2 antibody are shown. b. qRT-PCR of TUSC2 in NT (n=10), PTC (n=11) and ATC (n=6) snap-frozen tissue samples. The expression level of TUSC2 in each sample was measured by comparing its fluorescence threshold with the average fluorescence threshold of the NT samples. The average results of quadruplicate samples are plotted. The expression level in each sample was normalized with the β-Actin measurement. * p< 0.05.