Repression of Runx2 function by TGF-beta through recruitment of class II histone deacetylases by Smad3

Abstract

Transforming growth factor-beta (TGF-beta) inhibits osteoblast differentiation through inhibition of the function of Runx2 (Cbfa1) by Smad3. The mechanism through which TGF-beta/Smad3 inhibits Runx2 function has not been characterized. We show that TGF-beta induces histone deacetylation, primarily of histone H4, at the osteocalcin promoter, which is repressed by TGF-beta, and that histone deacetylation is required for repression of Runx2 by TGF-beta. This repression occurs through the action of the class IIa histone deacetylases (HDAC)4 and 5, which are recruited through interaction with Smad3 to the Smad3/Runx2 complex at the Runx2-binding DNA sequence. Accordingly, HDAC4 or 5 is required for efficient TGF-beta-mediated inhibition of Runx2 function and is involved in osteoblast differentiation. Our results indicate that class IIa HDACs act as corepressors for TGF-beta/Smad3-mediated transcriptional repression of Runx2 function in differentiating osteoblasts and are cell-intrinsic regulators of osteoblast differentiation.

Figure 1

Effect of TSA on Runx2-mediated transcription from OSE2 sequences. (A) TSA reverses TGF-β-induced repression of Runx2 activity at 6OSE2-luc in NIH3T3 cells. Cells were transfected with a Runx2 plasmid and the 6OSE2-luc reporter. After 24 h, cells were treated with TSA for 16 h in the presence or absence of TGF-β, and the luciferase activity was measured. (B, C) TSA does not affect activation of transcription by TGF-β at the 3TP or Smad7 promoters. (D) TSA inhibits TGF-β-induced repression of endogenous osteocalcin mRNA expression in the ROS17/2.8 cells, as assessed by real-time PCR.

Figure 2

TGF-β-induced chromatin changes at the osteocalcin and Smad7 promoters. (A) TGF-β induces deacetylation of histone H4 at the osteocalcin promoter in ROS17/2.8 cells. Chromatin from cells treated with or without TGF-β for 4 h was immunoprecipitated with anti-acetylated histone H3 (anti-AcH3) or H4 (anti-AcH4), or control beads, and analyzed with PCR using primers spanning the −235 to +26 sequence in the osteocalcin promoter, including the OSE2 sequence. (B) Rapid deacetylation of histone H4 upon TGF-β treatment. ROS17/2.8 cells were treated with TGF-β, and ChIP assays were performed as in panel A. (C) TGF-β stimulates histone H3 and H4 acetylation at the Smad7 promoter, as assessed by ChIP assays using primers that span the −462 to −277 sequence that includes the SBE sequence. (D) The levels of histone H3 and H4 acetylation from panels B and C were quantified.

Figure 3

Class IIa HDACs are corepressors in TGF-β-mediated repression of Runx2-dependent transcription. (A) Effects of individual HDACs on Runx2-mediated transcription in the absence or presence of TGF-β. ROS17/2.8 cells were transfected with −147OC-luc reporter and individual HDAC plasmids, with or without TGF-β. (B) Wild-type or Smad3−/− MEFs were transfected with 6OSE2-luc and the indicated plasmids, with or without TGF-β. (C) HDAC4 and 5 repress Gal–Smad3-mediated transcription. NIH3T3 cells were transfected with a plasmid encoding Gal4–DNA-binding domain (Gal–DBD), Gal–Smad3 or Gal–VP16, without or with HDAC4, 5 or 1 plasmid, as well as the Gal4-luc reporter with five tandem Gal4-binding sites. An activated TβRI receptor was coexpressed to activate TGF-β signaling.

Figure 4

Localization of HDAC4 and 5 mRNA by in situ hybridization. Hybridization of ³⁵S-labeled riboprobes in E17.5 mouse bone sections was visualized using dark-field microscopy to localize mRNAs of HDAC4, 5, Runx2, or osteocalcin. Counterstaining with DAPI visualized the nuclei. HDAC4 mRNA was localized in the digits of the front paw (A), and HDAC5 mRNA was localized in the humerus (B), and frontonasal bone of the head (C). Sense control probes show the specificity for HDAC4 or 5. Colocalization of Runx2 and HDAC5 mRNA is shown by superimposing the staining, taking into account the scattering of the activated silver grains (B, C). A higher magnification of a framed field is shown in the lower panel. (D) Detection by RT–PCR of HDAC4 and 5 mRNA in ROS17/2.8, MC3T3-E1, and caIB 2T3 cells. (E) Detection of mRNA or protein of HDAC4 and 5 by RT–PCR or Western blotting, respectively, in primary calvarial osteoblasts.

Figure 5

Interaction of Smad3 with HDAC4 and 5. (A, B) COS cells were transfected with the indicated plasmids and processed for immunoprecipitation with anti-Flag antibodies. Smad3NL contains the MH1 domain and linker segment, while Smad3C contains the MH2 domain. (C) Interaction of endogenous Smad3 and HDAC4 or 5. ROS17/2.8 cells were treated with or without TGF-β for 3 h before lysis and immunoprecipitation with anti-Smad3 or IgG, and immunoblotting for HDAC4, 5, or Smad3. (D) Interaction of Smad3 with HDAC4 or 5 in vitro. ³⁵S-labeled, in vitro translated HDAC4 or 5 was adsorbed to GST or GST-Smads (or derivatives) bound to glutathione-Sepharose, and analyzed by SDS–PAGE and autoradiography. The Coomassie Blue-stained gel shows the purified GST proteins. (E) HDAC5 decreases the binding of Smad3 to CBP. 293T cells expressing the indicated proteins were subject to immunoprecipitation with anti-Flag antibodies. (F) Mapping of the HDAC5 segment that interacts with Smad3. COS cells expressing Flag-tagged Smad3 and HA-tagged HDAC5 or mutants were processed for immunoprecipitation assays.

Figure 6

Effects of TGF-β/Smad3 on repressor complex formation. (A) TGF-β induces interaction of endogenous Runx2 and HDAC5. ROS 17/2.8 cells were incubated for 4 h with or without added TGF-β. Immunoprecipitation assays were performed with anti-HDAC5 antibodies. (B, C) TGF-β and Smad3/4 enhance the interaction of Runx2 with HDAC5. COS cells expressing the indicated proteins with or without an active TGF-β receptor chimera were processed for immunoprecipitations and Western blotting with indicated antibodies. (D) Runx2 enhances the interaction of Smad3 with HDAC5. Increasing levels of Runx2 were coexpressed with Smad3 and HDAC5, and lysates were subject to immunoprecipitation and Western blotting. (E) Runx2, Smad3, and HDAC5 form a ternary complex. 293T cells expressing the indicated plasmids were subject to immunoprecipitation with anti-Flag. Protein complexes were eluted from the antibody and subjected to immunoprecipitation with anti-Myc and Western blotting. (F) Effects of Smad3 on recruitment of HDAC5 to the OSE2 sequence. COS cells expressing the indicated proteins were subjected to DNA precipitation assay using biotinylated 2xOSE2 oligonucleotide. Protein–DNA complexes were analyzed by Western blotting. (G) Recruitment of HDAC4 or 5 to the endogenous osteocalcin promoter in response to TGF-β. ROS17/2.8 cells or ROS17/2.8 cells expressing Flag-tagged HDAC5 were treated with or without TGF-β for 4 h and ChIP assays were performed using anti-HDAC4, anti-Flag for HDAC5 or control beads, and analyzed with PCR using primers spanning the −235 to +26 sequence in the osteocalcin promoter, including the OSE2 sequence (Figure 2A).

Figure 7

HDAC4 and 5 are required for repression of Runx2 function by TGF-β. (A) HDAC5(931–1121) interferes with the interaction of Smad3 with HDAC5. COS cells expressing Smad3, HDAC5, and increasing amounts of HDAC5(931–1121) were processed for immunoprecipitations and Western blotting. (B) HDAC5(931–1121) inhibits TGF-β-induced deacetylation of histone H4 at the osteocalcin promoter. ROS17/2.8 cells expressing HDAC5(931–1121) (ROS-H14 cells) or harboring empty vector (ROS-C2 cells) were treated with or without TGF-β, and subjected to ChIP assays with anti-acetylated histone H4 antibodies. The PCR primers spanned the −235 to +26 segment of the osteocalcin promoter (Figure 2A). (C) HDAC5(931–1121) decreases TGF-β repression of the osteocalcin promoter, but does not significantly affect transcription from the Smad7 promoter. ROS17/2.8 cells expressing HDAC5(931–1121) (ROS-H14 cells) or control ROS-C2 cells were transfected with −147OC-luc or Smad7pro-luc, with or without expression of a dominant-negative TGF-β receptor II (TβRII-DN), with or without added TGF-β, and luciferase activities were measured. The numbers above the bar graphs show the fold repression. (D) Effectiveness of siRNA-mediated depletion of HDAC4 and 5. Endogenous HDAC4 or 5 in the presence or absence of siRNAs was visualized by Western blotting. (E) Effects of HDAC4 or 5 siRNA on repression of osteocalcin gene expression by TGF-β. ROS17/2.8 cells were transfected with HDAC4 or 5 siRNA, and osteocalcin mRNA expression was measured by real-time PCR.

Figure 8

Class IIa HDACs contribute to the inhibition of osteoblast differentiation by TGF-β. (A) caIB 2T3 cells were infected with LPCX or LPCX-HA-HDAC5(931–1121) (upper panel) or transfected with HDAC4 or HDAC5 siRNAs (lower panel). Western analysis validated expression of HA-HDAC5(931–1121) (upper panel) or reduction of HDAC4 or 5 levels (lower panel). (B) HDAC5(931–1121) expression inhibits the reduction of osteocalcin mRNA expression by TGF-β in caIB 2T3 cells at day 6 in differentiation medium. Ethidium bromide staining shows similar RNA loading for Northern hybridization. Osteocalcin mRNA was quantified using real-time PCR and normalized to RPL19 mRNA expression, which is not affected by TGF-β. (C) HDAC4- or 5-specific siRNAs inhibit the repression of osteocalcin mRNA expression by TGF-β in caIB 2T3 cells at day 6 in differentiation medium. Real-time PCR values were normalized against RPL19 mRNA. (D) HDAC5 (931–1121) reduces the inhibition of matrix mineralization by TGF-β, as assessed by Alizarin Red staining on day 11 of differentiation.