Thyroid Hormone Receptor β Suppression of RUNX2 Is Mediated by Brahma-Related Gene 1-Dependent Chromatin Remodeling

Abstract

Thyroid hormone receptor β (TRβ) suppresses tumor growth through regulation of gene expression, yet the associated TRβ-mediated changes in chromatin assembly are not known. The chromatin ATPase brahma-related gene 1 (BRG1; SMARCA4), a key component of chromatin-remodeling complexes, is altered in many cancers, but its role in thyroid tumorigenesis and TRβ-mediated gene expression is unknown. We previously identified the oncogene runt-related transcription factor 2 (RUNX2) as a repressive target of TRβ. Here, we report differential expression of BRG1 in nonmalignant and malignant thyroid cells concordant with TRβ. BRG1 and TRβ have similar nuclear distribution patterns and significant colocalization. BRG1 interacts with TRβ, and together, they are part of the regulatory complex at the RUNX2 promoter. Loss of BRG1 increases RUNX2 levels, whereas reintroduction of TRβ and BRG1 synergistically decreases RUNX2 expression. RUNX2 promoter accessibility corresponded to RUNX2 expression levels. Inhibition of BRG1 activity increased accessibility of the RUNX2 promoter and corresponding expression. Our results reveal a mechanism of TRβ repression of oncogenic gene expression: TRβ recruitment of BRG1 induces chromatin compaction and diminishes RUNX2 expression. Therefore, BRG1-mediated chromatin remodeling may be obligatory for TRβ transcriptional repression and tumor suppressor function in thyroid tumorigenesis.

Figure 1.

BRG1 expression is reduced in thyroid cancer cell lines and thyroid cancer tissue samples. (A) Representative immunoblot illustrates BRG1, Baf57, and Baf60 protein expression in thyroid cell lines derived from a spectrum of thyroid cancer subtypes. (B) The histogram illustrates quantitation of BRG1, Baf57, and Baf60 protein levels, standardized to β-actin, averaged from three independent experiments performed in triplicate. Error bars are SD; significance compared with nonmalignant cells (Nthy-ORI) is indicated. (C) The graph illustrates BRG1, Baf57, and Baf60 mRNA levels, standardized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH), averaged from three independent experiments performed in triplicate. Error bars are SD; significance compared with nonmalignant cells (Nthy-ORI) is indicated. (D) Publically available microarray expression data (GSE76039, GSE3467), analyzed using GEOR2, reveals differential expression of BRG1 in thyroid cancer patient tumor samples. Error bars are SD; significance compared with normal thyroid tissue is indicated. *P < 0.05; **P < 0.01; ****P < 0.0001. ADU, arbitrary densitometric units; ATC, anaplastic thyroid cancer; PDTC, poorly differentiated thyroid carcinoma; PTC, papillary thyroid cancer.

Figure 2.

BRG1 and TRβ are colocalized and associate with the RUNX2 promoter. (A) Immunofluorescence illustrates detection of TRβ and BRG1 proteins in whole-thyroid cells. (B) Quantitation of the colocalization of TRβ and BRG1 in each thyroid cell line demonstrates a high concordance in localization. (C) TRβ is detected by immunoblot after immunoprecipitation from nonmalignant thyroid nuclear lysate using an anti-BRG1 antibody. (D) BRG1 is detected by immunoblot after immunoprecipitation from nonmalignant thyroid nuclear lysate using an anti-TRβ antibody. (E) BRG1 and TRβ are both detected by immunoblot after DNA pulldown using oligos containing the native sequence of the RUNX2 promoter. Binding is lost when an oligo containing a mutant thyroid hormone response element (TRE) is used, and when a 10× unlabeled competitor probe is used. (F) ChIP–qRT-PCR shows enrichment of BRG1 and TRβ at the RUNX2 promoter in nonmalignant cells (Nthy-ORI) but not in anaplastic thyroid cancer cells (SW 1736). Values were calculated as fold enrichment compared with the matched IgG antibody control. Error bars are SD; significance compared with IgG is indicated. *P < 0.05; ***P < 0.001. DAPI, 4′,6-diamidino-2-phenylindole; IB, immunoblot; IP, immunoprecipitation.

Figure 3.

BRG1 silencing leads to increased RUNX2 and RUNX2 target gene expression. (A) Representative immunoblot illustrates that transfection of nonmalignant thyroid cells (Nthy-ORI) with siBRG1 results in a concentration-dependent decrease in BRG1 protein. (B) Knockdown of BRG1 also results in corresponding decreases in BRG1 mRNA. Loss of BRG1 results in increased RUNX2 mRNA levels. Data are the averaged fold change in transfected cells normalized to mock-transfected cells. Graph summarizes quantitation of three separate experiments performed in triplicate; values are standardized to GAPDH. Error bars are SD; significance compared with nontargeting control is indicated. (C) Knockdown of BRG1 increases Runx2 target gene expression: MMP9, cyclin D1, and VEGF. Data are the averaged fold change in transfected cells normalized to mock-transfected cells. Graph summarizes quantitation of three separate experiments performed in triplicate; values are standardized to GAPDH. Error bars are SD; significance compared with nontargeting control is indicated. *P < 0.05; ***P < 0.001; ****P < 0.0001.

Figure 4.

BRG1 mediates RUNX2 suppression. (A) Representative immunoblot shows increased TRβ expression after lentiviral transduction of SW 1736 (SW) cells compared with cells transduced with an empty vector (EV). (B) qRT-PCR results show that transient transfection with a BRG1 expression vector increases BRG1 mRNA levels in both the SW-EV transduced cell line and the SW-TRβ transduced cell line. Error bars are SD; significance compared with empty vector control is indicated. (C) qRT-PCR results show changes in RUNX2 mRNA levels after stably transduced cells were transiently transfected with a BRG1 expression vector. Overexpression of BRG1 in the absence of TRβ does not repress RUNX2. BRG1 transfection in the presence of TRβ represses RUNX2 further than TRβ alone. Error bars are SD; significance compared with empty vector control is indicated. *P < 0.05; ***P < 0.001. WT, wild-type.

Figure 5.

BRG1 mediates RUNX2 promoter accessibility. (A) The active chromatin modification, H3K27ac, is enriched at the RUNX2 promoter in anaplastic thyroid cancer cells (SW 1736) but not in nonmalignant thyroid cells (Nthy-ORI). Values were calculated as fold enrichment compared with the matched IgG antibody control. Error bars are SD; significance compared with IgG is indicated. (B) The RUNX2 promoter is more accessible in anaplastic thyroid cancer cells (SW 1736) than nonmalignant (Nthy-ORI) by the DNase I hypersensitivity assay. Values were calculated as relative sensitivity to DNase digestion compared with a known heterochromatinized site. Error bars are SD; significance compared with nonmalignant cells is indicated. (C) Treatment with a BRG1 inhibitor (PF1-3, 30 µM) for 6 hours results in an increase in sensitivity to DNase I digestion of the RUNX2 promoter in nonmalignant cells (Nthy-ORI). Error bars are SD; significance compared with vehicle control is indicated. (D) Treatment with PFI-3 results in a loss of H1.2 enrichment at the RUNX2 promoter detected by ChIP. Values were calculated as fold enrichment compared with the matched IgG antibody control. Error bars are SD; significance compared with vehicle control is indicated. *P < 0.05; **P < 0.01; ***P < 0.001.