Histone demethylase JMJD3 is required for osteoblast differentiation in mice

JMJD3 (KDM6B) is an H3K27me3 demethylases and emerges as an important player in developmental processes. Although some evidence indicated the involvement of JMJD3 in osteoblast differentiation in vitro, its role as a whole in osteoblast differentiation and bone formation in vivo remains unknown. Here we showed that homozygous deletion of Jmjd3 resulted in severe delay of osteoblast differentiation and bone ossification in mice. By biochemical and genetical methods, we demonstrated that JMJD3 mediated RUNX2 transcriptional activity and cooperated with RUNX2 to promote osteoblast differentiation and bone formation in vivo. These results strongly demonstrated that JMJD3 is required for osteoblast differentiation and bone formation in mice.

whether JMJD3 is also required for intramembranous ossification in vivo, we examined the membranous bones of E14.5-E18.5 littermates by alcian blue and alizarin S red staining. No obvious skeletal difference between WT and Jmjd3 +/− mice was detected (data not shown). However, unlike wild-type embryos, which exhibited well calcified skeletons, homozygous mutant embryos showed continually less calcification in parietal bone from E14.5 to E18.5 (Fig. 1A). Detailed investigation revealed that Jmjd3 −/− mice exhibited open fontanelles, less mineralized cranial bones and hypoplastic clavicles compared to WT littermates at E18.5 (Fig. 1B,C). These features suggested that JMJD3 is required for intramembraneous ossification in vivo. JMJD3 is required for osteoblast differentiation during both intramembraneous and endochondral bone formation. Since osteoblast maturation is essential for intramembranous ossification and endochondral bone formation, we next examined whether JMJD3 is required for osteoblast differentiation in vivo. Firstly, we examined the expressed level of JMJD3 in osteoblasts of E14.5, E16.5 and E18.5 parietal bones by western blot assays. The results showed that JMJD3 expression increased from E14.5 to E18.5 ( Fig. 2A) and was primarily in the nuclei of osteoblasts (Fig. 2B). Subsequently, we performed histological investigations by von Kossa (Von) and alkaline phosphatase (AP) staining and observed much less mineralization and lower intensity of AP staining in the sections of parietal bones of Jmjd3 −/− embryos compared to WT littermates (Fig. 2C). Furthermore, we also observed much less mineralization and lower intensity of AP staining in Jmjd3 −/− long bones, such as in humerus (Fig. 2D). These results suggested JMJD3 is required for osteoblast maturation during both intramembraneous and endochondral bone formation.
To further characterize the defects of osteoblast differentiation in vivo, we performed in situ hybridization with RNA probes of molecular marker for osteoblast differentiation at different stages. COL1A1, the most abundant protein in the bone matrix, is an early marker of osteoblast differentiation. In E13.5 and E15.5 embryos, in situ hybridization indicated that Col1a1 mRNA level was much less in the perichondrium of Jmjd3 −/− than that of WT humerui (Fig. 2Ea-d). At E18.5, the level of osteocalcin (Bglap), a late marker of osteoblast differentiation, markedly decreased in perichondral and periosteal osteoblasts of Jmjd3 −/− embryos compared to WT littermates (Fig. 2Ee,f). These results revealed that defects of osteoblast differentiation occurred at both early and late developmental stages of Jmjd3 −/− embryos. Consistently, the mRNA level of Fgf18, a factor crucial for both early and late osteoblast differentiation, also obviously reduced in mutants compared with WT embryos (Fig. 2Eg,h).
In addition, we examined the differentiation potential of primary cultured preosteoblasts from cranial bones of E14.5 WT and Jmjd3 −/− embryos in osteogenic medium. Alizarin S red and von Kossa staining  JMJD3 regulates RUNX2 transcriptional activity during osteoblast differentiation. To explore the molecular mechanism of JMJD3 in osteoblast differentiation, we first analyzed expression of osteoblast-specific genes in WT and Jmjd3 −/− embryos. RT-qPCR showed that the mRNA levels of Runx2, Sp7, Spp1, Bglap, Col1a1 and Fgf18 in parietal osteoblasts were reduced in Jmjd3 −/− embyros compared with WT mice (Fig. 3A). In contrast, the mRNA level of atf4 was not affected. To test whether JMJD3 directly regulates these genes transcription, we examined JMJD3 level at the promoters of Runx2 and Bglap by chromatin immunoprecipitation (ChIP) followed by qPCR analysis in primary osteoblasts. Indeed, JMJD3 was showed to strongly recruit to the promoters of the two genes ( Fig. 3Ba,b). Consistently, the H3K27me3 level at the promoters of Runx2 and Bglap genes markedly increased in Jmjd3 −/− compared with WT primary osteoblasts (Fig. 3Bc,d). These results clearly demonstrated that JMJD3 directly promoted Runx2 and Bglap transcription by erasing H3K27me3 in osteoblasts. This was consistent with a recent report that knockdown of Jmjd3 in MC3T3-E1 osteoblasts resulted in increased H3K27me3 level at the promoter regions of Runx2 18 . In addition, western blot assay showed that knockout of Jmjd3 increased the total level of H3K27me3 in primary osteoblasts (Fig. 3C). This indicated that JMJD3 might regulate more genes transcriptions for osteoblast differentiation.
Since RUNX2 has been shown to bind and activate its own promoter via a positive feedback loop 20 , and Sp7, Spp1, Bglap and Col1a1 are target genes of RUNX2 21 , we speculated that JMJD3 may act as a coactivator of RUNX2 in osteoblast differentiation. To test this possibility, we firstly test whether JMJD3 biochemically interacts with RUNX2. Co-immunoprecipitation assays showed that RUNX2 really pull down JMJD3, or vice versa, in the primary culture of osteoblasts from E18.5 mice (Fig. 3D). In addition, immunofluorescence under confocal microscope indicated that JMJD3 co-localized with RUNX2 in the nuclei of primary osteoblasts (Fig. 3E). Next, to test whether JMJD3 can functionally cooperate with RUNX2 to regulate their target genes transcription, we performed luciferase reporter assays with Bglap gene promoter. The results showed that JMJD3 quite obviously synergized with RUNX2 to potentiate the expression of Bglap. This synergistic effect was reduced by shRNA against Jmjd3 in a dose dependent manner (Fig. 3F). These results strongly supported that JMJD3 is a coactivator of RUNX2 in osteoblast differentiation in vivo and suggested that RUNX2 may recruit JMJD3 to target genes during bone formation. In order to test whether RUNX2 can affect JMJD3 binding to target genes during osteoblast differentiation, we examined JMJD3 occupancies at promoters of Runx2 and Bglap genes by ChIP-qPCR in WT and Runx2 −/− primary osteoblasts from parietal bones. The results showed that the level of JMJD3 was reduced at the promoters of Runx2 and Bglap genes in Runx2 −/− compared with WT osteoblasts (Fig. 3Ga,b). Consistently, the levels of H3K27me3 at the promoters of Runx2 and Bglap increased in Runx2 −/− versus WT primary osteoblasts (Fig. 3Gc,d). These results supported that RUNX2 can recruit JMJD3 to target genes during osteoblast differentiation.

Discussion
Osteoblast differentiation is essential for both intramembranous and endochondral ossification. Abnormality of this process will result in various bone diseases, such as osteoporosis or osteosclerosis 22,23 . It has been shown previously that H3K27 methyltransferase EZH2, a core subunit of PRC2 group, suppresses mesenchymal stem cells (MSCs) from differentiating into osteoblasts 24 , while WDR5, a core component of trithorax group (TrxG), promote osteoblasts 25 . These results implicate other PRC2 and trithorax proteins may also be involved in skeleton development. As an H3k27me3 demethylases, JMJD3 counteracts the transcription repression by Polycomb group (PcG) and biochemically interacts with the core subunits of TrxG, such as WDR5, RbBP5, ASH2L and BRG1 9,26 . Thus, these biochemical features begged the question of the possible role of JMJD3 during osteoblast differentiation. Recently, Ye et al. 19 showed that osteogenic differentiation of human bone marrow MSCs is positively regulated by JMJD3, while Yang et al. 18 demonstrated that osteoblast differentiation is inhibited upon knocking-down Jmjd3 in vitro. In addition, both papers also showed by in vivo models that JMJD3 positively regulates osteogenic differentiation, either using a model injecting MSCs in which JMJD3 was silenced into mice 19 , or by finding impaired local bone formation in adult mice upon local application of siRNA against Jmjd3 in the calvaria of living mice 18 . However, the detail roles of JMJD3 for osteoblast differentiation during both intramembranous and endochondral ossification as a whole in vivo is not clear. Here, using Jmjd3 knockout mice, we demonstrated that JMJD3 is required for osteoblast differentiation during both intramembranous and endochondral ossification. This conclusion was supported by the following reasons: (1)  JMJD3 was expressed in osteoblasts, (2) Jmjd3 mutant mice displayed severe retardation of osteoblast differentiation by skeletal preparation and histological analysis, and (3) JMJD3 genetically and biochemically cooperated with RUNX2 to promote osteoblast differentiation and bone ossification in vivo.
The existence of a fairly typical cleidocranial dysplasia phenotype in Jmdj3 −/− embryos suggested that JMJD3 genetically associates with RUNX2 in vivo. Indeed, we detected that JMJD3 directly promotes Runx2 transcription in osteoblasts by CHIP-qPCR assay. This was consistent with a recent report that knockdown of Jmjd3 in MC3T3-E1 osteoblasts resulted in increased H3K27me3 level at the promoter regions of Runx2 18 . The result was also consistent with previous report that EZH2 directly bound to the promoter of Runx2 to inhibit differentiation of MSCs to osteoblasts 24 . Moreover, since both Runx2 is its own target genes 21 , we speculated that JMJD3 may act as a coactivator of RUNX2. This was verified through Co-IP and luciferase reporter assays by showing that JMJD3 favors RUNX2 transcriptional activity. More importantly, we demonstrated that JMJD3 genetically cooperated with RUNX2 to promote osteoblast differentiation and bone ossification in vivo. Therefore, our results not only confirmed the role of JMJD3 in osteoblast differentiation in vivo but also expanded the molecular mechanism of JMJD3 in bone formation. In addition, these results implicated that fine tuned counteraction of TrxG and PcG in gene transcription may play critical roles in modulating osteoblast maturation and bone development. Further studies need to explore the physiological or pathological roles of JMJD3 and other subunits of TrxG or PcG in bone biology.
Whole-mount staining of skeletal preparation, histological analyses, alkaline phosphatase staining and in situ hybridization investigations were performed as described previously 17 . Briefly, the embryos were skinned, eviscerated, and fixed in 95% ethanol. After 4 days fixation in 95% ethanol, embryos were stained in Alcian blue solution overnight. After washing with 70% ethanol, the embryos were stained by Alizarin S red solution overnight and transferred into 1% KOH for 1week. Lastly, embryos were transferred into 1% KOH/20% glycerol for 2 days and stored in 50% ethanol/50% glycerol. For histological analyses, embryos fixed in 10% formalin, processed, and embedded in paraffin or OCT (Tissue-Tek; Thermo Fisher Scientific). Serial sections were taken at 4 μ m thickness. To detect mineral deposition, Von Kossa staining was performed and counterstained by alcian blue and nuclear fast red. Alkaline phosphatase staining was performed on 7 μ m cryostat sections using alkaline phosphatase kit (Millpore) following the manufacturer's instructions. For in situ hybridization, digoxigenin-11-UTP-labed single-stranded RNA probes were prepared with a DIG RNA labeling kit (Roche) according to the manufacturer's instructions. Fgf18 probe was from David Onnitz (Washington University School of Medicine) and Col1a1 and Bglap probes sequences were described before 27,28 . All animal procedures were conducted in accordance with the Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Shanghai Institutes for Biological Sciences.