Hypoxia inhibits osteogenesis through direct regulation of RUNX2 by TWIST

INTRODUCTION: Bone loss induced by hypoxia is associated with various pathophysiological conditions such as ischemia, vascular diseases, and osteolytic bone metastases, however, little is known about the mechanism of hypoxia-regulated osteogenesis and bone formation. RUNX2 (also known as CBFA1) is a master regulator of skeletogenesis and its expression is required for osteoblast differentiation and maturation. Using human bone marrow mesenchymal stem cells (MSCs), which survived under hypoxia, we discovered TWIST, a downstream target of HIF1-α, acts as a transcription repressor of RUNX2 through binding to the E-box located on the promoter of Type 1 RUNX2. Suppression of RUNX2 inhibited the expression of BMP2, Type 2 RUNX2 and downstream targets of RUNX2 such as alkaline phosphatase and osteocalcin in MSCs. Further, knockdown of twist1a and twist1b in zebrafish induced an increase in runx2b (one orthologue of mammalian Runx2) expression, which was associated with a ventralized embryo and an increased in bone formation. Our finding pointed out an important role of hypoxia-mediated signalling in osteogenic differentiation and bone formation through direct regulation of RUNX2 by TWIST. METHODS: Primary human mesenchymal stem cells (MSCs) induced in osteogenic induction medium (OIM) under hypoxia or DFX or zebrafish were exposed to hypoxia or DFX. We plan to investigate makers of osteogenesis under hypoxia which induced HIF-1α-TWIST pathway to inhibit osteogenesis by real-time PCR analysis and western blot analysis. To examine whether overexpression of TWIST can suppress T1, T2 RUNX2 and RUNX2 in osteogenesis, we used pFLAG-TWIST or pFLAG-CMV1 transfected MSCs in normaxia or pSuper-TWIST-si or pSuper-Scramble in hypoxia. To determinate an E-box, a novel role of TWIST binding domain of T1 RUNX2 in bone formation through direct interaction under hypoxia, we used RUNX2 promoter assay, truncation bHLH of TWIST (pFLAG-tbTWIST) promoter assay, site-directed mutagenesis E-boxs of T1 RUNX2 promoter assay, EMSA and ChIP to prove TWIST binding T1 RUNX2 through E1-box. In vivo, zebrafish used to confirm bone loss of hypoxia or DFX by Alizarin Red S & Alcian blue stain. Further, runx2b expression was analyzed by runx2b whole-mount in situ hybridization and real-time PCR and HE stain under embryos were microinjected twist1a or twist1b atgMOs. RESULTS: Because MSCs isolated from bone marrow, which is hypoxic in nature (1-7% O2), survive under hypoxia, we induced bone marrow MSCs from three individual donors in osteogenic induction medium (OIM) under normoxia (21% O2) and hypoxia (1% O2) to understand the effects of oxygen tension on osteogenic differentiation. The expression of RUNX2 was detected at 3 days of differentiation and the expression level was greater under normoxia than hypoxia or hypoxia mimic condition, deferoxamine (DFX) both as mRNA and protein in all three MSCs. Further both hypoxia and DFX induced a decrease in the expression of RUNX2 downstream target genes, such as OC, ALK-P, BSP, COL1A1, and OP. Induction under hypoxia also had an inhibitory effect on the functional mineralization of MSCs both at 14 and 21 days of osteogenic differentiation. To demonstrate the inhibitory effect of hypoxia and DFX on bone formation in vivo, zebrafish were also confirmed. Collectively, these data suggest hypoxia or DFX inhibited osteogenic differentiation both in vitro and in vivo. MSCs induced in OIM that contains dexamethasone, increased in T1 RUNX2 and BMP2 (bone morphogenetic protein2) expression as early as 12 h after induction, followed by a delayed increase in the expression of T2 RUNX2 at 24 h. To explore the key molecule that hypoxia or DFX targeted to regulate osteogenesis, we first found both the expression of T1 and T2 RUNX2 were downregulated at 3 days under hypoxia. Further, the presence of BMP2 abrogated the DFXmediated downregulation of T2 RUNX2, OC and OP but not T1 RUNX2. To examine the molecular mechanism that hypoxia mediated to inhibit osteogenesis, we first examined the expression of HIF-1α, the main transcription factor induced by hypoxia and TWIST, which is upregulated by HIF-1α in tumor, under hypoxia. Both HIF-1α and TWIST were upregulated under hypoxia and DFX treatment in MSCs from three individuals. Interestingly, overexpression of TWIST suppressed the expression of T1, T2 RUNX2 and OC under normoxia, while knockdown of TWIST stimulated the expression of these genes under hypoxia. These results demonstrated the HIF-1α-TWIST pathway regulates osteogenesis by MSCs. To demonstrate whether RUNX2 was directly regulated by TWIST, analysis of the human RUNX2 P2 and P1 promoter activity using luciferase reporter constructs in the immortalized MSC cell line revealed the upregulation of P2 but not P1 promoter activity upon induction in OIM and suppressed by TWIST or DFX treatment in a dose-dependent manner. To identify the minimum promoter region required for inhibition by TWIST, a series of 5' P2 promoter deletion constructs were generated. A further finding where TWIST bind to pGL3-RUNX2 P2 (WT) promoter, site-directed mutagenesis of the putative E-boxs or truncation bHLH of TWIST in the P2 promoter prevented suppression under TWIST was examined. Electrophoretic mobility shift assays (EMSAs) using an oligonucleotide containing the E1-box sequence from RUNX2 P2 promoter incubated with nuclear extracts of TWISToverexpressing 293T cells demonstrated the direct binding of TWIST to the wild-type probe. Finally, Chromatin immunoprecipitation (ChIP) assays confirmed the direct binding of TWIST to the E1-box of the T1 RUNX2 promoter. PCR amplification of the antiTWIST antibody immunoprecipitants showed that the fragment of the T1 RUNX2 promoter containing the E1-box (180 bp) existed in the MSC-TWIST transient or 293T-TWIST stable line. These data taken together suggest TWIST downregulated P2 promoter activity by direct binding through the bHLH domain to the E1-box in P2 promoter. To test whether twist downregulates runx2/runx2b and inhibits bone formation in vivo, we used the zebrafish developmental model, in which the distribution of both runx2b and twist1a,1b,2 and 3 are well clarified. Interestingly, microinjection of twist1a and twist1b atgMOs but not MO-SC (scramble), twist2 and twist3 atgMOs dose-dependently induced an increase in class 3 and 4 ventrailzed embryos, suggesting knockdown of twist1a and twist1b increased the expression of runx2b. Consistently, quantitative RT-PCR revealed twist1a and twist1b atgMOs increased in T1 and T2 runx2b expressions compared to wild type/ MOSC embryos at 8, 14 and 48hpf. Whole-mount in situ hybridization at 8, 14 and 48 hpf further demonstrated the expression of runx2b was induced by twist1a and 1b atgMO. Finally, we determined whether twist1a and twist1b atgMOs injection promoted functional mineralization both under normoxia and hypoxia. The embryos survived and had normal morphology with up to an 8 dpf increase in bone formation as shown by an apparent increase of ARS staining at the Ot in whole embryos studies, and as well as an increase in cell condensation and mineralization in the cranial and pharyngeal region in thin tissue sections stained with H&E Staining with or without DFX treatment. These data taken together suggest knockdown of twist1a or twist1b in zebrafish enhanced runx2b transcription, induced ventralized patterning, and promoted bone formation both under normoxic and hypoxic conditions (DFX treatment). DISCUSSION: This is the first time T1 RUNX2 has been demonstrated as an important target for controlling osteogenesis and bone formation by hypoxia or HIF-1α-TWIST, an important environment encountered by a lot of pathophysiological conditions associated with normal development and regeneration, or acquired and genetic diseases. Therefore, signaling pathways or molecules that control the transcription of T1 RUNX2 may be applied to control in vitro osteogenesis and in vivo bone formation. We further demonstrated the direct downregulation of RUNX2 by TWIST through binding to E1-box of P2 promoter. In addition, we proved knockdown of twist1a and twist1b in zebrafish increased runx2b expression, which controlled dorsoventral patterning and bone development. Therefore, the roles of twists controlled dorsoventral pattern, skeleton development and bone formation through a direct role in controlling the expression and an indirect role in regulating the function of runx2b. The phenotypes of knockdown of twist1a and twist1b include the abnormalities in eyes, fusion of fore/midbrain and hindbrain, notochord, trunk, and other skeleton deformity. The similar phenotype abnormalities are seen in human disorders with TWIST mutation, such as Saethre–Chotzen syndrome (SCS), an autosomal dominant disorder with characteristic abnormalities of eyes (exophthalmos and ptosis), craniosynostoses resulted from premature closure of cranial sutures, short stature, and developmental limb deformities (syndactyly and polydactyly).Future exploration of the twist signaling pathways may help in developing strategies in controlling dorsoventral patterning and skeleton development through runx2b suppressions. In conclusion, these data provide convincible evidences on the important roles of TWIST in controlling dorsoventral patterning, skeleton development and bone formation. Further exploration of the mechanism that TWIST mediates to regulate skeleton development and regeneration may provide new strategies for treating these diseases. Paper No. 287 • 56th Annual Meeting of the Orthopaedic Research Society