CHD7 regulates bone-fat balance by suppressing PPAR-γ signaling

Chromodomain helicase DNA-binding protein 7 (CHD7), an ATP-dependent eukaryotic chromatin remodeling enzyme, is essential for the development of organs. The mutation of CHD7 is the main cause of CHARGE syndrome, but its function and mechanism in skeletal system remain unclear. Here, we show conditional knockout of Chd7 in bone marrow mesenchymal stem cells (MSCs) and preosteoblasts leads to a pathological phenotype manifested as low bone mass and severely high marrow adiposity. Mechanistically, we identify enhancement of the peroxisome proliferator-activated receptor (PPAR) signaling in Chd7-deficient MSCs. Loss of Chd7 reduces the restriction of PPAR-γ and then PPAR-γ associates with trimethylated histone H3 at lysine 4 (H3K4me3), which subsequently activates the transcription of downstream adipogenic genes and disrupts the balance between osteogenic and adipogenic differentiation. Our data illustrate the pathological manifestations of Chd7 mutation in MSCs and reveal an epigenetic mechanism in skeletal health and diseases.

A s the common progenitors for osteoblasts and adipocytes, bone marrow mesenchymal stem cells (MSCs) show strict control of cell fate decisions to maintain the balance between osteogenic and adipogenic differentiation to ensure the health of the skeletal system 1,2 . Characterized by low bone mass and extensive marrow adipose tissue (MAT) accumulation, senile osteoporosis in elderly patients leads to a series of complications, e.g., increasing bone fragility and susceptibility to fracture [3][4][5] . Bone marrow MSCs preferentially differentiate into adipocytes under senility or pathological stimuli, e.g., hormone disorders, leading to increased MAT and progressive bone loss 1,3,6 . Unfortunately, the precise mechanism by which MSC lineage allocation favors adipogenic differentiation toward osteogenic differentiation remains to be explored.
As typical ATP-dependent eukaryotic chromatin remodeling enzymes, the members of the chromodomain helicase DNAbinding protein (CHD) superfamily interrupt the interaction between DNA and chromatin by translocating nucleosomes along the DNA strand 7,8 . The CHD family is essential for the expression of normal genes and the maintenance of dynamic chromatin structure via their ability to activate or suppress specific histone markers 9,10 . Hence, the CHD superfamily is critical for the maintenance and proliferation of stem cells, in addition to cell differentiation and cell fate decisions 11,12 .
As one of the most researched members of the CHD family, CHD7 is widely expressed in various tissues and plays a critical role in the development of organs 10,12 . Mutation of CHD7 is the main cause of CHARGE syndrome, a kind of developmental disorder involving multiple organ system defects, characterized as coloboma of the eye, heart defects, atresia of the choanae, retarded growth, genital anomalies, ear malformations and deafness [13][14][15] . The acronym of these six major manifestations was defined as CHARGE syndrome in 1981 16 .
As previously indicated, CHD7 is an ATP-dependent chromatin remodeling enzyme that regulates the location of nucleosomes and alters DNA accessibility 17,18 . According to previous studies, one of the main functions of the CHD family is binding to methylated histone residues and promoting methylated histone H3 at lysine 4 (H3K4me) interactions 8,19 . Previously found by chromatin immunoprecipitation (ChIP), CHD7 recruitment was strongly linked to histone modification, e.g., colocalization with H3K4me1 at the enhancer subregion and with H3K4me3 at the promoters of the transcription initiation site 20 .
We previously showed that CHD7 could promote the osteogenic differentiation of human bone MSCs by interacting with SMAD1 and binding to the enhancer region of SP7 21 . In this work, we generated Chd7 conditional knockout mice and unveiled the essential role of CHD7 in MSC fate decisions by suppressing the peroxisome proliferator-activated receptor gamma (PPAR-γ) signaling pathway. Such findings might provide insight into the pivotal regulatory role of CHD7 in bone development and diseases, as well as a probable clinical therapeutic strategy.

Results
Deletion of Chd7 in MSCs leads to skeletal development disorder. To investigate the potential role of CHD7 in skeletal development, we examined the expression of CHD7 in mouse femurs. Immunohistochemical (IHC) staining showed that CHD7 was prevalently expressed in the nuclei of bone cells and bone marrow in both primary and secondary ossification centers but marginally expressed in chondrocytes of the growth plate (Supplementary Fig. 1a). Immunofluorescence (IF) staining images showed the same trend ( Supplementary Fig. 1b).
We conducted whole-mount skeletal staining on pups 2 days after birth and observed that the skeletal development of Prx1cre;Chd7 fl/fl pups is significantly delayed compared with that of Chd7 fl/fl littermates, with shorter forelimbs and hindlimbs and a lower degree of mineralization in the cranial and maxillofacial bone (Fig. 1b, c). The skulls of 2-week-old and 4-week-old mice were harvested and scanned by microcomputed tomography (microCT). The reconstructed images showed nonclosure of fontanels in the Prx1-cre;Chd7 fl/fl mice, indicating that knockout of Chd7 in MSCs leads to cranial bone hypoplasia (Fig. 1d, e).
Deletion of Chd7 in MSCs leads to low bone mass and high marrow adiposity. MicroCT analysis of distal femur metaphysis indicated that bone mineral density (BMD) and bone volume/ tissue volume ratio (BV/TV) were significantly reduced in both male and female Prx1-cre;Chd7 fl/fl mice compared to their Chd7 fl/fl littermates at 4 weeks of age. In addition, Chd7 deletion impaired the trabecular number (Tb.N), the trabecular thickness (Tb.Th) and the midshaft cortical thickness (Ct.Th), while elevating the trabecular separation (Tb.Sp) (Fig. 2a, b).
The undecalcified sections after Von Kossa staining further validated the low bone mass phenotype of the Prx1-cre;Chd7 fl/fl mice (Fig. 2c). The Prx1-cre;Chd7 fl/fl mice also showed a slower mineral apposition rate (MAR) and bone formation rate (BFR) (Fig. 2e). In addition, the Prx1-cre;Chd7 fl/fl mice had fewer osteoblasts (N.Ob/B.Pm), indicating a diminished bone formation. The osteoclast number (N.Oc/B.Pm) showed no obvious differences between groups (Fig. 2d, e). Chd7 knockout also led to a slightly shortened growth plate ( Supplementary Fig. 1a).
Notably, in both male and female Prx1-cre;Chd7 fl/fl mice, the loss of bone mass was associated with obviously increased bone marrow adipose tissue accumulation in both the primary and secondary ossification centers (Fig. 2f). The number and density of bone marrow adipocytes were apparently elevated in the Prx1-cre;Chd7 fl/fl mice (Fig. 2g). For unknown reasons, the phenotype of increased MAT was more striking in female Prx1-cre;Chd7 fl/fl mice.
Loss of Chd7 in preosteoblasts leads to low bone mass and high marrow adiposity. To further investigate whether depletion of Chd7 from committed osteoblast progenitors could induce a similar bone loss phenotype, we mated the Sp7-cre transgenic mice with the Chd7 fl/fl mice 28 . The Sp7-cre;Chd7 fl/fl mice also exhibited a smaller size, lighter weight, lower survival rate and delayed growth features ( Supplementary Fig. 2a, b). MicroCT analysis of the femur metaphysis revealed a diminishment in trabecular and cortical bone in the Sp7-cre;Chd7 fl/fl mice, as shown by impaired skeletal parameters, e.g., BMD, BV/TV, Tb.N, Tb.Th and Ct.Th, and elevated Tb.Sp (Fig. 3a, b).
The Von Kossa staining validated the low bone mass phenotype of the Sp7-cre;Chd7 fl/fl mice (Fig. 3c). The Sp7cre;Chd7 fl/fl mice also showed declined dynamic skeletal histomorphometric features, manifested as slower MAR and BFR (Fig. 3e). In addition, the decrease in osteoblast numbers further confirmed the osteopenic manifestations of the Sp7-cre;Chd7 fl/fl mice (Fig. 3d, e).
Similarly, increased MAT accumulation was also observed in both the primary and secondary ossification centers of femurs from the Sp7-cre;Chd7 fl/fl mice (Fig. 3f, g). Moreover, the MAT accumulation phenomena were more remarkable in the tibia of the Sp7-cre;Chd7 fl/fl mice at the age of 4 weeks. Considerable MAT was observed at the mesial region, while the distal region was mostly occupied by MAT, with a very small amount of bone tissue left at the distal region of tibiae in the female Sp7-cre;Chd7 fl/fl mice ( Supplementary Fig. 2c).
Chd7-deficient MSCs exhibit reduced osteogenic and increased adipogenic potential. Next, we sought to isolate MSCs from the Chd7 fl/fl and Prx1-cre;Chd7 fl/fl mice and compared their osteogenic and adipogenic potential in vitro ( Supplementary Fig. 3a). Flow cytometry was conducted to identify the primary mouse bone marrow mesenchymal stem cells with anti-CD29 and anti-CD90.2 antibodies.
CHD7 suppresses the transactivation of PPAR-γ. Next, we performed RNA-seq of MSCs to identify the critical mechanism. Gene set enrichment analysis (GSEA) confirmed the restricted osteogenesis in Prx1-cre;Chd7 fl/fl MSCs after osteogenic induction (Fig. 5a, b). Interestingly, significant enhancement of the PPAR signaling pathway was found in the Prx1-cre;Chd7 fl/fl group, indicating that the PPAR family might play a crucial role in the formation of the abovementioned phenotype (Fig. 5a).
Meanwhile, among MSCs after adipogenic induction, significant enhancement of the PPAR signaling pathway was also found in the Prx1-cre;Chd7 fl/fl group, along with enhancement of the adipogenesis pathway and increased expression of downstream genes of PPAR-γ (Fig. 5c, d).
Next, we sought to verify whether CHD7 directly interacts with PPAR-γ by conducting a coimmunoprecipitation (co-IP) assay. In MSCs isolated from Chd7 fl/fl mice, CHD7 was successfully detected in PPAR-γ immunoprecipitates (Fig. 5e). However, H3K4me3 was barely observed in PPAR-γ immunoprecipitates, ARTICLE indicating that PPAR-γ might not be activated in wild-type MSCs. In contrast, H3K4me3 was detected in PPAR-γ immunoprecipitates in the Chd7-deficient MSCs (Fig. 5e). To intuitively observe the altered binding pattern of PPAR-γ and H3K4me3, we conducted IF staining to detect their colocalization within the nucleus. In the Prx1-cre;Chd7 fl/fl MSCs, the merged rates of PPAR-γ and H3K4me3 were significantly higher than those in wild-type MSCs, in accordance with the co-IP assay ( Fig. 5f, g).
Next, we aimed to determine how this binding alteration regulates cell fate decisions. By analyzing ChIP-sequencing (ChIP-seq) data of Mus musculus (GSM2104246 and GSM1571714) 29,30 , we confirmed that H3K4me3 and PPAR-γ could coaggregate at the promotor region of Adipoq (Fig. 5h). On this basis, we performed a ChIP-PCR assay on the MSCs. The enrichment of H3K4me3 and PPAR-γ at the promoter regions of Adipoq was significantly increased in MSCs isolated from Prx1cre;Chd7 fl/fl mice (Fig. 5i, j).
PPAR-γ inhibition partially rescues the phenotype of Chd7deficient MSCs. To verify that activated PPAR-γ contributed to pathogenic phenotype upon Chd7 depletion, we treated the Prx1cre;Chd7 fl/fl MSCs with the PPAR-γ-specific antagonist GW9662. Although the protein level of CHD7 in the Prx1-cre;Chd7 fl/fl MSCs was unchanged, the expression of PPAR-γ was significantly restricted. Meanwhile, the PPAR-γ antagonist partially recovered the expression of RUNX2 and alkaline phosphatase (Fig. 6a).
Administration of a PPAR-γ inhibitor partially rescues the skeletal phenotype of Prx1-cre;Chd7 fl/fl mice. Intraperitoneal injection of the PPAR-γ inhibitor GW9662 partially recovered the body size and weight of the Prx1-cre;Chd7 fl/fl mice (Fig. 7a-c). MicroCT analysis of the femur metaphysis showed partially recovered trabecular and cortical bone in the treated Prx1cre;Chd7 fl/fl mice, as shown by recovered skeletal parameters, e.g., BMD, BV/TV, Tb.N, Tb.Th, Ct.Th, and Tb.Sp (Fig. 7d, e).
The Von Kossa staining validated the restored bone mass of the treated Prx1-cre;Chd7 fl/fl mice (Fig. 7f). The treated Prx1cre;Chd7 fl/fl mice also showed recovered skeletal histomorphometric features, manifested as elevated BFR and increased osteoblast numbers (Fig. 7g).
Moreover, decreased MAT accumulation was also observed in femurs from the treated Prx1-cre;Chd7 fl/fl mice (Fig. 7h). The number and density of bone marrow adipocytes were evidently reduced in the treated Prx1-cre;Chd7 fl/fl mice (Fig. 7i). Consequently, the three major manifestations of Chd7 depletion in MSCs-whole-body skeletal system development disorder, low bone mass, and high marrow adiposity-could be partially rescued by administration of a PPAR-γ inhibitor.

Discussion
As a dynamic process, stem cell differentiation is regulated by activating or suppressing a specific locus and dividing the genome into transcriptionally active or inactive chromadomains 31,32 . Epigenetic regulation plays an essential role in the growth and development of multiple organ systems in mammals, e.g., DNA methylation, histone modification, chromosome remodeling, and RNA interference 33,34 . As a hotspot in current research, the methylation and acetylation of histone lysines is a crucial modification at the tails of histones, playing an important role in stem cell differentiation and even cell fate decisions 35,36 . Generally, H3K4me1 is considered as a marker of enhancers, while H3K4me2 and H3K4me3 modifications are mostly enriched at promoters near transcription initiation sites to activate specific Accumulating evidence has uncovered the regulatory role of CHD7 in epigenetics. As reported previously, in mammals, CHD7 could be recruited at the promoters or superenhancers of the transcriptionally active region of specific genes to increase or decrease transcription 20,45 . Existing studies of CHD7 preferentially shed light on nervous system development and the differentiation of neural stem cells and neural crest stem cells, and CHD7 has been proven to be crucial for neurogenesis and adult neural stem cell maintenance 9,10,13,46,47 . Widely observed in clinical work, the alteration of CHD7 expression levels in specific cell lineages leads to severe malfunction or diseases, e.g., CHARGE syndrome, Kallmann syndrome, autism spectrum disorder, developmental disorders of multiple organ systems, and tumorigenesis 13,15,[45][46][47][48][49] . Some of these conditions were caused by the mutation of CHD7, while others resulted from changes in CHD7 histone modification or chromatin remodeling functions 45,[48][49][50] .
Here, we discovered that CHD7 regulates osteogenesis and cell fate decisions in bone marrow MSCs by suppressing the PPAR-γ signaling pathway. By using different temporal markers of osteogenic stem cells, Prx1 and Sp7, we established two lines of transgenic mice to conditionally knockout Chd7 from the limb bud mesenchyme at a very early time or from preosteoblasts at the initial stage of differentiation 27,28 . We found that loss of Chd7 in MSCs disrupted the cell fate decision in mice, leading to the pathological manifestations of osteoporosis, e.g., low bone mass with an impaired osteogenic capacity, as well as high marrow adiposity with an enhanced adipogenic ability, revealing the effective and specific regulation of CHD7 on MSCs.
Previously, PPAR-γ was reported to act as a pivotal nuclear receptor and regulate stem cell differentiation and cell fate decisions 51 . This process relies on the association with coactivators or corepressors cooperating with the methylation and acetylation of histone lysine 52 . Admittedly, osteogenesis, adipogenesis, and chondrogenesis are three major directions of the multidirectional differentiation potential of MSCs. The activation of PPAR-γ primarily drives the cell fate of MSCs to adipogenesis 52,53 . However, via the inhibiting function of PPAR-γ of specific progenitor cytokines or other corepressors, the cell fate of MSCs favoring adipogenic differentiation could also be reversed toward osteogenic differentiation 53,54 . Consistent with our findings, the enhanced PPAR signaling pathway was consistent with the phenotype of the Prx1-cre;Chd7 fl/fl mice in vivo and in vitro. Further, the administration of a PPAR-γ antagonist largely converted the biased MSC adipogenic differentiation and allowed osteogenesis to dominate, which also further confirmed the PPAR signaling pathway as an innovative mechanism whereby CHD7-mediated histone modification manipulated the cell lineage allocation of bone marrow MSCs.
Our findings indicated that when CHD7 bound PPAR-γ, the interaction between PPAR-γ and H3K4me3 was reduced, and consequently, the expression of the downstream adipogenic genes were downregulated, while that of osteogenic-related genes was upregulated, leading to the osteogenic differentiation of MSCs. In contrast, when CHD7 was depleted in MSCs, PPAR-γ colocalized with and contacted H3K4me3 and then activated the transcription of downstream adipogenic genes, leading to the adipogenic differentiation of MSCs ( Supplementary Fig. 4a).
Notably, no significant difference (p > 0.05) in the osteoclast number (N.Oc/B.Pm) was found between the Prx1-cre;Chd7 fl/fl mice and their Chd7 fl/fl control littermates. We speculated that such minor alterations might not be the major factor of the low bone mass in the Prx1-cre;Chd7 fl/fl mice. According to a previous study, bone marrow adipocytes might be the source of RANKL 23 . Therefore, the minor alteration of the osteoclast number in the Prx1-cre;Chd7 fl/fl mice might be a side effect of MAT accumulation, which deserves further research.
Certain limitations in our work should be listed to provide ideas for further research. The MAT accumulation in bone marrow suggested that there might be specific influences of CHD7 on adipocytes. Conditional knockout of Chd7 in adipose tissue in vivo would be a prospective research direction. Moreover, the lack of specimens from patients was a limitation of this research. As previously reported, CHARGE syndrome is mainly caused by the CHD7 mutation in neural crest stem cells, while patients with CHD7 mutations in MSCs have not been studied.
In conclusion, we demonstrate that conditional knockout of Chd7 in bone marrow MSCs leads to the activation of the PPAR signaling pathway. Such alteration would promote the MSC differentiation balance toward adipogenesis, resulting in whole body developmental disorders of the skeletal system, low bone mass, and high marrow adipose tissue. Our research provided evidence that CHD7 regulates stem cell differentiation and cell fate decisions and revealed the functional association between histone modification and osteoporosis, which might shed light on therapeutic strategies for bone pathological disorders.

Methods
Generation of conditional Chd7 knockout mice. Based on the CRISPR-Cas9 approach, Chd7 fl/+ mice were generated with a C57BL6/J background [55][56][57] . Briefly, two sgRNAs were designed to target regions either upstream or downstream of exon 2 of Chd7 with the aid of a CRISPR design tool (http://crispr.mit.edu). Then, these molecules were screened for on-target activity via a Universal CRISPR Activity Assay (UCATM, Biocytogen, Inc., Beijing). Donor vectors containing LoxP flanking exon 2 mixed with sgRNAs and Cas9 mRNA were coinjected into the cytoplasm of single-celled C57BL6/J zygotes. Such injected zygotes were transferred into ovarian ducts of Kunming pseudopregnant female mice to generate F0 mice. F0 mice with the expected genotype identified by tail genomic DNA PCR and sequencing were then mated with C57BL6/J mice to obtain germline-transmitted F1 founders. Both PCR genotyping and Southern blot examination were applied to further confirm the correct genotype of the generated Chd7 fl/+ mice.
Prx1-cre and Sp7-cre mice were purchased from the Jackson Laboratory (Bar Harbor, ME) and were crossed with Chd7 fl/+ mice for more than three generations to obtain Prx1-cre;Chd7 fl/fl and Sp7-cre;Chd7 fl/fl conditional knockout mice 27,28 . The genotypes were identified by PCR amplification of tail genomic DNA. Primers for the Chd7 knockout allele and Cre transgene genotyping are listed in Supplementary Table 1. All mice were propagated and raised in specific pathogenfree facilities. Whole-mount skeletal staining. Mouse pups were harvested at postnatal day 2. After removal of the skin and viscera, the whole mount of the pups was fixed with 95% ethanol. After being processed with acetone, the specimens were stained with Alcian blue (Solarbio) solution and decolorized with 70% and 95% ethanol. After being processed with potassium hydroxide solution, the specimens were stained with Alizarin Red S (ARS) (Solarbio) solution. After being processed with potassium hydroxide solution and glycerol, the specimens were imaged with an Epson Perfection V370 Photo Scanner. Length of humeri, femurs, and tibiae was measured with ImageJ (National Institutes of Health) 58 .
MicroCT. Mice were sacrificed at 4 weeks old and intraperitoneally injected with calcein (10 mg/kg of body weight; Sigma) at 7 days and 2 days before sacrifice. The collected bone tissues were fixed in 4% paraformaldehyde for 24 h and then transferred to phosphate-buffered saline (PBS) at 4°C before scanning. According to the guidelines, microCT analysis was performed via a μCT50 microCT system (Scanco Medical, Bassersdorf, Switzerland) with a spatial resolution of 8 μm (55 kV, 114 mA, 500 ms integration time). The volume of interest (VOI) of femurs was defined as a cylindrical area at the distal region of the growth plate, 1200 µm thickness for spongiosa and 200 µm thickness for cortical bone. Bone volume was calculated within the delimited VOI. Data analysis and reconstructed images were processed on IMARIS software (version 9.1.2; Oxford Instruments). Quantification of cranial defect was calculated with ImageJ (National Institutes of Health), by measuring the hole area in the cranium in the reconstructed images of skulls 3,44,59,60 .
Undecalcified histomorphological analysis. After microCT scanning, femurs were successively dehydrated in 15% and 30% sucrose solutions and then embedded in NEG-50 (Thermo Scientific) at −80°C. Specimens were cut into 5µm-thick sections with a freezing microtome (CM3050 S; Leica) with a tungsten steel blade at -35°C and were stuck to adhesive film (Cryofilm; Section-Lab Co., Ltd.).
Sections of undecalcified bone specimens were mainly processed for Von Kossa staining and analysis of double labeling. The VOI was defined as the distal end of femurs. Calcium deposits were visualized according to the Von Kossa/nuclear fast red staining method and were photographed by microscopy (BX53; Olympus). The dynamic osteogenic index was defined as the distance between 2 calcein bands by microscopy with fluorescence excitation (BX53; Olympus) and was analyzed with OsteoMeasure software (OsteoMetrics; Decatur, GA) 26,59,61 .
Decalcified histomorphological analysis. After being fixed in 4% polyoxymethylene for 24 h and then decalcified in 10% EDTA for 2 weeks at 4°C, femurs and tibiae were dehydrated in ethanol and xylene by Excelsior AS (Thermo Scientific). Embedded in paraffin, bone specimens were cut into 5-µm-thick sections with a paraffin microtome (RM2255; Leica) 3,5 .
Sections of decalcified bone specimens were mainly processed for hematoxylin and eosin (HE) staining, tartrate-resistant acid phosphatase (TRAP) staining, IHC staining, and IF staining. The VOI was defined as the distal end of femurs. According to the IHC and IF staining protocol, slides were incubated in sodium citrate antigen retrieval solution at 100°C for 10 min and then incubated with rabbit anti-CHD7 antibody (Abcam, ab117522, 1:200). IF staining slides were imaged by laser scanning confocal microscopy (FV3000; Olympus), and other  Cell culture and flow cytometry. Primary MSCs were isolated by flushing the bone marrow of femurs, tibiae, and humeri. The bone marrow suspension was cultured in 21 cm 2 petri dishes in alpha minimum Eagle's medium (α-MEM, HyClone) with 10% fetal bovine serum (FBS, Gibco), 100 units/mL penicillin, and 100 μg/mL streptomycin. After incubation at 37°C with 5% CO 2 , the culture medium was changed every 48 h. MSCs at passage 3 were used for subsequent research 3,62,63 .
ALP, ARS, and Nile red staining. MSCs fixed with 4% paraformaldehyde in 24-pore plates were then stained with a BCIP/NBT Alkaline Phosphatase Color Development Kit (Beyotime) after 7 days of osteogenic induction. The reaction was terminated after 15 min of light-free incubation at room temperature, and the pore plates were imaged with an Epson Perfection V370 Photo Scanner. Quantitative analysis of ALP activity was processed with a BCA Protein Assay Kit (Beyotime) and Alkaline Phosphatase Assay Kit (Beyotime). The BCA curve was obtained from the absorbance of the BCA protein concentration gradient. The corresponding ALP activity was calculated from the standard curve of ALP absorbance by reaction with 0.5 mM p-nitrophenyl phosphate 59,62 .
MSCs in 24-pore plates were fixed with 4% paraformaldehyde and then stained with Alizarin Red S solution (Cayagen) at room temperature after 21 days of osteogenic induction. The pore plates were imaged with an Epson Perfection V370 Photo Scanner. The calcium nodules were detained by cetylpyridinium chloride for 15 minutes to quantify the calcium concentration. The absorbance at 562 nm was measured with a Multiskan Sky Microplate Spectrophotometer (Thermo Scientific), in contrast with the standard calcium absorbance curve 5,26 .
For Nile red staining, MSCs on glass coverslips were fixed with 4% paraformaldehyde and then stained with Nile red solution (Solarbio) at 37°C for 10 min after 21 days of adipogenic induction. After DAPI staining and mounting, images were obtained with laser scanning confocal microscopy (FV3000; Olympus) and were analyzed with ImageJ (National Institutes of Health) 64 .
Cell immunofluorescence staining. MSCs on glass coverslips were fixed with 4% paraformaldehyde and then permeabilized with 0.5% Triton X-100 in PBS at room temperature for 10 minutes. MSCs were incubated with 1% BSA, 5% goat serum, and 0.1% Triton X-100 in PBS at room temperature for a 30-minute block. MSCs were incubated with primary antibodies from different species in blocking buffer at 4°C overnight. After removal of the primary antibodies and rinsing with PBS the next day, MSCs were successively incubated with Alexa Fluor 488 goat anti-mouse antibody (Abcam, ab150113, 1:100) and Alexa Fluor 647 goat anti-rabbit antibody (Abcam, ab150083, 1:100) in the dark at room temperature for 1 h. After DAPI staining and mounting, images were obtained with laser scanning confocal microscopy (FV3000; Olympus) and were analyzed with ImageJ (National Institutes of Health) 5 . The primary antibodies and concentrations are as follows: mouse anti-PPAR-γ antibody (Santa Cruz, sc-271392, 1:100) and rabbit anti-trimethylhistone H3 (Lys4) antibody (CST, #9751, 1:100).
Quantitative RT-PCR and RNA-seq. Total RNA was isolated with TRIzol Reagent (Invitrogen) 5 days after osteogenic or adipogenic induction. RNA was reverse transcribed via a PrimeScript RT reagent Kit (TaKaRa). Quantitative RT-PCR was processed with SYBR Premix Ex Taq (TaKaRa) and LightCycler 96 (Roche). The housekeeping gene GAPDH was used as the baseline to quantitatively analyze the gene expression of the osetogenic group, and 36B4 was used for the adipogenic group 59,62 .
For RNA sequencing, 12 RNA samples of MSCs in 4 groups were prepared according to the manufacturer's protocol of a NEBNext Ultra RNA Library Prep Kit for Illumina (USA). RNA samples were subjected to HiSeq 2500 (Illumina). FastQC (v0.11.5) and the FASTX toolkit (0.0.13) were applied for quality control. Further analysis, e.g., KEGG enrichment, GO enrichment, heatmaps, and GSEA, was performed to explore the downstream pathways on this basis 3,26 .
For the co-IP assay, MSCs were lysed in mild lysis buffer (MLB) with protease/ phosphatase inhibitor cocktail (CST, #5872, 1:100) for 10 min with gentle shaking at 4°C. After centrifugation at 14,000 × g for 10 min, 10% of the supernatant was heated with SDS-PAGE sample loading buffer (Beyotime) at 100°C for 5 min as input. The remaining 90% of supernatant was collected and incubated with IgG or selected antibody overnight at 4°C with mixing. After washes with MLB, 5 μg Pierce Protein A/G Magnetic Beads (Thermo Scientific) was added to each sample and incubated for 1 h at 4°C with mixing. After collection and washing, the beads were heated with SDS-PAGE Sample Loading Buffer (Beyotime) at 100°C for 5 min for subsequent western blot analysis. The co-IP antibodies were rabbit (DA1E) mAb IgG XP® Isotype Control (CST, #3900, 1:100) and rabbit anti-PPARγ antibody (CST, #2443, 1:100) 65 .
ChIP assay. ChIP-seq data were obtained from GSM2104246 and GSM1571714 and were analyzed with WashU EpiGenome Browser (Washington University) to ascertain the binding site and the experimental design of ChIP-PCR 29,30 . According to the manufacturer's protocol of the Magna ChIP HiSens Chromatin Immunoprecipitation kit (Millipore), 2 × 10 6 cells were harvested in each ChIP reaction. Protein and DNA of each sample were crosslinked by applying 37% formaldehyde solution. SDS lysis buffer with protease/phosphatase inhibitor cocktail (CST, 1:100) was added after cell harvesting. After restriction enzyme digestion, centrifugation, and precipitation with beads, the precipitated DNA samples were quantified with specific primers via real-time PCR 5 .
Rescue strategy in vitro and in vivo. Based on the verification of PPAR-γ signaling pathway, MSCs were treated with a PPAR-γ specific antagonist GW9662 (MedChemExpress) to rescue the phenotype in vitro. The application concentration was 10 μM, dissolved in DMSO 66 .
To partially rescue the general body phenotype in vivo, we intraperitoneally injected 3 mg/kg/day GW9662 into the female Prx1-cre;Chd7 fl/fl and control littermates from postnatal day 10 until 4 weeks old before sacrifice. Hereafter, microCT and decalcified and undecalcified histomorphological analyses were conducted on the specimens to clarify the rescue effect 67,68 .
Statistical analyses. All data were calculated as the mean ± standard deviation (SD). Statistical differences were analyzed via Student's t test for independent sample tests or one-way ANOVA with Tukey's post hoc tests for multiple comparisons. A p value <0.05 was considered statistically significant 3,62 .
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.