Disruption of Dhcr7 and Insig1/2 in cholesterol metabolism causes defects in bone formation and homeostasis through primary cilium formation

Human linkage studies suggest that craniofacial deformities result from either genetic mutations related to cholesterol metabolism or high-cholesterol maternal diets. However, little is known about the precise roles of intracellular cholesterol metabolism in the development of craniofacial bones, the majority of which are formed through intramembranous ossification. Here, we show that an altered cholesterol metabolic status results in abnormal osteogenesis through dysregulation of primary cilium formation during bone formation. We found that cholesterol metabolic aberrations, induced through disruption of either Dhcr7 (which encodes an enzyme involved in cholesterol synthesis) or Insig1 and Insig2 (which provide a negative feedback mechanism for cholesterol biosynthesis), result in osteoblast differentiation abnormalities. Notably, the primary cilia responsible for sensing extracellular cues were altered in number and length through dysregulated ciliary vesicle fusion in Dhcr7 and Insig1/2 mutant osteoblasts. As a consequence, WNT/β-catenin and hedgehog signaling activities were altered through dysregulated primary cilium formation. Strikingly, the normalization of defective cholesterol metabolism by simvastatin, a drug used in the treatment of cholesterol metabolic aberrations, rescued the abnormalities in both ciliogenesis and osteogenesis in vitro and in vivo. Thus, our results indicate that proper intracellular cholesterol status is crucial for primary cilium formation during skull formation and homeostasis.


INTRODUCTION
Cellular cholesterol amounts are regulated by multiple steps of biosynthesis and feedback mechanisms. 1 Infants with cholesterol metabolism aberrations have malformations in the craniofacial region. [2][3][4][5] For example, mutations in genes involved in cholesterol synthesis (DHCR7, SC5D, and DHCR24) have been found in patients with Smith-Lemli-Opitz Syndrome (SLOS) lathosterolosis, and desmosterolosis, who display craniofacial bone abnormalities. 1 In addition, high-cholesterol diets during pregnancy are known to be a risk factor for birth defects, including craniofacial bone abnormalities. 6,7 Despite these findings, it remains unclear how cholesterol metabolism contributes to craniofacial bone formation, particularly intramembranous ossification.
The 7-dehydrocholesterol reductase (DHCR7) catalyzes the final step of cholesterol biosynthesis; 8 mutations in DHCR7 cause cholesterol deficiency and an excess of cholesterol precursors, resulting in craniofacial deformities (e.g., microcephaly, cleft palate, craniosynostosis, and micrognathia), intellectual disability, and behavioral problems in humans. 9,10 Dhcr7 −/− mice show a suckling defect, weight less, immature lungs, distended bladders, and variable craniofacial abnormalities. 11 The molecular mechanism of craniofacial anomalies in these conditions is still elusive. The insulin-induced genes 1 and 2 (INSIG1 and INSIG2) are endoplasmic reticulum (ER) retention proteins that play roles in both the regulation of the activity of the 3-hydroxy-3-methylglutarylcoenzyme A (HMG-CoA) reductase and the translocation of the sterol regulatory element-binding protein (SREBP) to the nucleus for gene regulation. 12 Mice deficient for Insig1 and Insig2 (Insig1 −/ − ;Insig2 −/− mice), which are negative regulators of cholesterol biosynthesis, 13 show high-cholesterol levels in craniofacial tissues and display craniofacial deformities such as midfacial cleft, cleft palate, calvarial deformities and micrognathia, while mice deficient for either Insig1 or Insig2 are normal. 3,12 These craniofacial deformities are rescued by the normalization of cholesterol levels in Insig1/2 null mice; 3 however, it remains elusive how high-cholesterol levels cause craniofacial deformities and which cells are responsible for the craniofacial anomalies seen in Insig1/2 null mice.
Primary cilia, microtubule-based organelles that function in sensory and signaling pathways, are enriched with cholesterol-rich microdomains (known as lipid rafts) that recruit or retain receptors and ciliary membrane proteins. 14 An association between lipid rafts and ciliary membrane proteins has been suggested in other organisms, including vertebrate photoreceptors, 15 Chlamydomonas reinhardtii, 16 mammalian spermatozoa, 17 and Leishmania major. 18 Defects of primary cilia cause various deformities, including craniofacial abnormalities (altogether known as ciliopathies, a group of genetic syndromes associated with defects in primary cilia). 19,20 The broad-spectrum phenotypes in Dhcr7 −/− mice and individuals with SLOS 11 (e.g., presenting with craniofacial anomalies such as craniosynostosis, hypertelorism, and cleft palate, as well as immature lungs and enlarged bladders) are similar to those seen in ciliopathies. The phenotypic similarity between ciliopathies and cholesterol synthesis defects suggests that cholesterol metabolism (level and function of mature cholesterol and cholesterol intermediates) can regulate bone development through modulation of primary cilium formation and function. While over the past decade the underlying mechanism of ciliopathies has focused on the inner structures of primary cilia such as intraflagellar transport (IFT) and kinesin (KIF) proteins, 21 little is known about the role of the surface membrane characteristics of primary cilia in ciliogenesis.
In this study, we investigated the link between cholesterol metabolic aberrations and craniofacial bone abnormalities by employing both loss-of-function and gain-of-function mouse models: mice with a deletion of Dhcr7 and mice with a deletion of Insig1/2, respectively. To identify the bone abnormalities with either low or high-cholesterol levels, we carried out microCT, skeletal staining, and histological analyses in these mice. Our study aims to elucidate how Dhcr7 and Insig1/2 regulate bone formation.

RESULTS
Dhcr7 deficiency increases osteogenesis Dhcr7 −/− knockout (KO) mice presented microcephaly, accelerated bone formation, and thicker calvaria bones at birth with complete penetrance, and died within 1 day after birth (Fig. 1a-c and Supplementary Fig. S1). The accelerated bone formation resulted in immature suture fusion after culturing calvaria explants for 3 days (Supplementary Fig. S1d). To examine the cellular mechanism of how cholesterol metabolic aberrations cause accelerated bone formation in Dhcr7 −/− mice, we carried out biological analyses, namely BrdU incorporation assays and Ki67 immunohistochemistry for cell proliferation, TUNEL assays for apoptosis, and von Kossa staining for mineralization, and immunoblotting for type I collagen for osteogenic differentiation. We found that osteogenic differentiation, but not cell proliferation and apoptosis, was increased in Dhcr7 −/− frontal bones (Fig. 1d, e and Supplementary Fig. S2). Next, to determine the regulatory mechanism of osteogenic differentiation, we performed quantitative RT-PCR (qRT-PCR) analyses for osteogenic factors (Runx2, Alp, Col1a1, Col1a2, Bglap, Sparc, Sp7, and Spp1) using frontal bones from Dhcr7 −/− mice at embryonic day (E) E14.5, E15.5, E16.5, and postnatal day (P) P0 ( Fig. 1f and Supplementary Fig. S3). Col1a1 gene expression was significantly and consistently upregulated in Dhcr7 −/− frontal bones compared to controls at E14.5-P0. Expression of RUNX2, COL1A1 and SP7 (aka Osterix) was increased, and the area with these positive signals was expanded compared to controls and correlated with increased expression of these genes in Dhcr7 −/− frontal bones (Fig. 1f, g). Next, we evaluated the effect of loss of Dhcr7 on osteogenic differentiation using cultured osteoblasts from P0 frontal bones. The Dhcr7 −/− osteoblast characteristics (no proliferation defect but accelerated osteogenic differentiation through upregulated Col1a1 expression) were well conserved in cultured primary osteoblasts ( Fig. 1h and Supplementary Fig. S3). Taken together, our results indicate that a failure in cholesterol synthesis causes accelerated osteogenesis through upregulated Col1a1 expression.

Insig1/2 deficiency suppresses osteogenesis
To determine the tissue-specific contribution of high-cholesterol status to craniofacial deformities, we employed Insig1/2 conditional knockout (cKO) mice 12 in cranial neural crest (CNC) cells 22 , which give origin to the majority of craniofacial bones. 23 Insig1/2 cKO mice were viable, but they exhibited very thin frontal bones at birth and later in life (Fig. 2a, b), while the other craniofacial structures derived from CNC cells in these mice were intact. To examine the cellular mechanism of how high-cholesterol amount results in decreased bone formation in Insig1/2 cKO mice, we carried out biological analyses, as performed in Dhcr7 −/− mice. We found that osteogenic differentiation, but not cell proliferation and apoptosis, was decreased in Insig1/2 cKO frontal bones ( Supplementary Fig. S4). We also found that Col1a1 gene expression was significantly and consistently downregulated in frontal bones of Insig1/2 cKO mice at E14.5, E16.5 and P0 (Fig. 2c), with protein expression correlating with gene expression (Fig. 2d). Expression of COL1A1 and SP7, but not RUNX2, was decreased, which was correlated with the expressions of these genes in Insig1/2 cKO frontal bones (Fig. 2e). We further evaluated the effect of loss of Insig1/2 on osteogenic differentiation using cultured osteoblasts from Insig1/2 cKO mice ( Fig. 2f and Supplementary Fig. S4). The Insig1/2 cKO osteoblast characteristics (no proliferation defect but reduced osteogenic differentiation through decreased Col1a1 expression) were well conserved in cultured primary osteoblasts. Since COL1 expression was decreased in Insig1/2 cKO mice, the abnormalities may resemble osteogenesis imperfecta, which is a congenital bone disorder characterized by thinner and fragile bones that affects 6-7 in 100 000 individuals worldwide. 24 Altogether, our results indicate that either too much or too little cholesterol causes calvarial bone abnormalities through dysregulation of Col1a1 expression.
Ciliogenesis is altered in either Dhcr7-or Insig1/2-deficient mice As cholesterol is abundant in cellular membranes, we carefully investigated cellular membranous structures and found that there were fewer and shorter primary cilia in Dhcr7 −/− osteoblasts compared to controls (Fig. 3a-c). Ciliogenesis starts with the interaction of the basal body (mother centriole) with primary ciliary vesicles (CVs), which can be labeled with RAB11, 25 and then the axoneme grows within the ciliary membrane while fusing with secondary CVs, which can be labeled with RAB8. 26 The elongated primary cilium eventually fuses with the plasma membrane, allowing the distal part of the cilium to interact with the extracellular milieu. 19,27 To track ciliogenesis in Dhcr7 −/− osteoblasts, CVs were immunostained with anti-RAB11 and anti-RAB8 antibodies. We found that RAB11-positive primary CVs accumulated in the cells and that RAB8-positive secondary CVs failed to gather at sites of cilium formation in Dhcr7 −/− osteoblasts (Fig. 3d, e).
osteoblasts compared to controls (Fig. 4g), while the expression of other PCM-related genes was not altered ( Supplementary Fig.  S6b). We further confirmed that PLK4 protein expression was increased in the mutant osteoblasts compared to controls (Fig.  4h). Taken together, our results indicate that Dhcr7 and Insig1/2 play an important role in primary cilium formation.
WNT/β-catenin and hedgehog signaling pathways are downstream signaling cascades responsible for the bone phenotype of Dhcr7-and Insig1/2-deficient mice We examined how the ciliary phenotypes resulted in osteogenic abnormalities. The primary cilia coordinate with multiple signaling pathways such as hedgehog (HH) and WNT during  osteogenesis. [36][37][38] To investigate HH signaling in Dhcr7 −/− calvaria, we analyzed Gli1 and Ptch1 levels by qRT-PCR, which are readout gene expressions for HH signaling activity. [39][40][41][42][43] We found that expression of these genes was significantly downregulated in Dhcr7 −/− calvaria (Fig. 5a). Furthermore, we employed Gli1-LacZ reporter mice to evaluate the HH signaling activity in vivo. We found that HH signaling was compromised in the calvaria of Dhcr7 −/− mice compared to controls (Fig. 5b). To further confirm the reduced HH signaling activity in the calvaria of Dhcr7 −/− mice, we performed cell fractionation and the consequent immunoblotting analyses for GLI1. Previous studies indicate that in the absence of HH ligands, GLI1 is truncated and degraded by the proteasome in the cytoplasm, while GLI1 translocates into the nucleus in the presence of HH ligands by escaping from degradation. 44,45 As expected, full-length GLI1 was decreased in the nuclear fraction from Dhcr7 −/− osteoblasts (Fig. 5c).
Next, we investigated the HH signaling activity in Insig1/2 cKO mice. We carried out qRT-PCR for Gli1 and Ptch1 in Insig1/2 cKO calvaria and found that expression of these genes was significantly upregulated in Insig1/2 cKO mice (Fig. 5d). Furthermore, we performed LacZ staining for Gli1 expression in Insig1/2 cKO;Gli1-LacZ and control mice. As expected, Gli1-LacZ expression was increased in the calvaria of Insig1/2 cKO;Gli1-LacZ mice compared to that of control mice (Fig. 5e). Next, to examine whether Col1a1 expression was regulated through HH signaling, we conducted promoter analyses for GLI binding on the Col1a1 promoter and found that the Col1a1 promoter contained a putative GLI-binding site (GGCCACGCA; −68 bp to -60 bp) ( Supplementary Fig. S7a). ChIP assays validated that GLI binding to the Col1a1 promoter region was correlated with the activity of HH signaling in osteoblasts from Dhcr7 −/− and Insig1/2 cKO mice ( Fig. 5f, g).
Thus, HH signaling activity was correlated with the primary cilium phenotype of Dhcr7 −/− and Insig1/2 cKO mice. However,  c Quantitative RT-PCR for the indicated osteogenic genes in WT (blue bars) and Insig1/2 cKO (yellow bars) mice at E14.5, E16.5, and P0. n = 6 per genotype per stage. ***P < 0.001. d Immunoblotting for COL1 in newborn WT and Insig1/2 cKO mice. GADPH was used as loading control. e Immunohistochemical analysis for RUNX2, COL1A1 and SP7 (Osterix) in newborn WT and Insig1/2 cKO mice. Nuclei were counterstained with 0.04% methylene blue. Scale bar, 50 µm. f Alkaline phosphatase (top) and Alizarin Red (bottom) staining of osteoblasts isolated from the frontal bones of newborn WT and Insig1/2 cKO after induction of osteogenic differentiation at Day 28. Osteogenesis and primary cilium formation A Suzuki et al.
previous studies show that upregulated and downregulated HH signaling cause defective and normal craniofacial bone formation, respectively. [46][47][48] This suggests that altered HH signaling was not responsible for the bone phenotypes in Dhcr7 −/− and Insig1/2 cKO mice. We, therefore, investigated another candidate cilium-mediated signaling that regulates intramembranous ossification, WNT/ β-catenin signaling, 49 in Dhcr7 −/− and Insig1/2 cKO mice. Recent studies indicate that primary cilia negatively regulate WNT/β-catenin signaling. 37,50,51 To investigate the activity of WNT/β-catenin signaling pathway, we conducted quantitative RT-PCR for Axin2, a readout of WNT/β-catenin signaling activity, and found that its expression was upregulated and downregulated in Dhcr7 −/− and Insig1/2 cKO osteoblasts, respectively (Fig. 6a, b and Supplementary  Fig. S8). To validate the relationship between WNT/β-catenin signaling and cholesterol metabolism, we treated primary osteoblasts from Dhcr7 −/− and Insig1/2 cKO mice with either WNT3A or lithium chloride (LiCl), a known WNT/β-catenin signaling activator. While both WNT3A and LiCl induced Axin2 expression five-fold in wild-type controls, Axin2 expression was upregulated and downregulated in Dhcr7 −/− and Insig1/2 cKO osteoblasts, respectively (Fig.  6c, d). We confirmed that in mice with Topgal reporter for the WNT/ β-catenin signaling activity, its activity was increased at osteogenic front of the calvaria in Dhcr7 −/− mice compared to control mice (Fig.  6e). Next, to examine whether Col1a1 expression was regulated through WNT/β-catenin signaling, we conducted promoter analyses for β-catenin binding on the Col1a1 promoter and found four putative β-catenin binding sites ( Supplementary Fig. S7b). To evaluate these predicted binding sites, we conducted ChIP assays for β-catenin binding to the Col1a1 promoter region, and found that β-catenin bound at binding sites 1 and 4, and the binding was correlated with the activity of WNT/β-catenin signaling in osteoblasts from Dhcr7 −/− and Insig1/2 cKO mice ( Supplementary Fig. S9). Lastly, to test the functional significance of WNT/β-catenin signaling, we generated and investigated compound mutant mice with a haploinsufficiency of Axin2 in the Dhcr7 mutant background and found that normalized WNT/β-catenin signaling restored the accelerated bone formation in Dhcr7 −/− mice ( Fig. 6f-h). Altogether, these findings are well supported by the fact that WNT/β-catenin signaling positively regulates osteogenesis. 52 The bone phenotype in Dhcr7-and Insig1/2-deficient mice is restored by the normalization of aberrant cholesterol metabolism with simvastatin To test whether abnormalities in primary cilium and bone formation can be restored by the normalization of aberrant cholesterol metabolism, we treated Dhcr7 −/− and Insig1/2 cKO osteoblasts with simvastatin, which can normalize cholesterol metabolic aberrations by inhibiting the activity of the HMG-CoA reductase. Simvastatin treatment restored primary cilium formation in cultured osteoblasts from Dhcr7 −/− mice (Fig. 7a-c). In addition, the in vivo administration of simvastatin to Dhcr7 −/− mice (10 mg·kg −1 body weight per day, intraperitoneal injection to a pregnant mouse, E12.5-E18.5) could normalize the accelerated bone formation in newborn mice (Fig. 7d, e). We confirmed that expression of genes related to bone formation and WNT signaling was restored in Dhcr7 −/− mice treated with simvastatin (Fig. 7f). These results indicate that cholesterol intermediates caused the bone defects in Dhcr7 −/− mice. In Insig1/2 cKO mice, simvastatin treatment restored both the increased number and length of primary cilia (Fig. 8a, b) and the reduced bone formation in Insig1/2 cKO osteoblasts (Fig. 8c). Importantly, simvastatin treatment (10 mg·kg −1 body weight per day from P7 to P42)   improved the decreased bone formation in these mice (Fig. 8d). As expected, the expression of genes related to osteogenesis, cilia and WNT signaling was normalized in Insig1/2 cKO mice treated with simvastatin (Fig. 8e). Taken together, our results indicate that proper cholesterol metabolic status is crucial for normal primary cilium formation, which is responsible for osteogenesis in osteoblasts (Fig. 9).

DISCUSSION
Previous studies show that cholesterol metabolism is important for the formation and homeostasis of bone and cartilage during endochondral ossification. 53 54 Thus, either too much and too low of cholesterol results in short limbs through defective endochondral ossification. In this study, we aimed to study the role of cholesterol in intramembranous ossification using both gain-offunction (Insig1/2 cKO mice) and loss-of-function (Dhcr7 −/− mice) mouse models, and found that cholesterol metabolism plays a role in intramembranous ossification in a dose-dependent manner. These findings suggest that effects of abnormal  55 Previous studies indicate that suppression of DHCR7 activity results in upregulated vitamin D 3 levels in vivo after birth. [56][57][58] While patients with SLOS show a high level of Primary cilia transduce extracellular cues as a mechanical sensor, as well as a chemical sensor, for morphogens and growth factors. 65,66 Therefore, both endochondral and intramembranous ossification are affected in mice with defective primary cilia. 38 For example, mice deficient for Kif3a, a ciliary protein, in osteoblasts display decreased bone formation and osteopenia. 67 A deficiency for cilia-related protein SPEF2 results in osteoblast differentiation defects. 68 In this study, we found that HH and WNT/β-catenin signaling is inverted in conditions of the primary cilia in Dhcr7 −/− and Insig1/2 cKO mice during intramembranous ossification. While both WNT/β-catenin and HH signaling pathways are essential for endochondral ossification as well as for the differentiation and maturation of cultured osteoblasts, 36 previous mouse genetic studies suggest that WNT/β-catenin signaling may be more dominant than HH signaling for osteogenesis in intramembranous ossification during skull formation.
In regard to HH signaling, previous studies show that HH signaling is compromised in Dhcr7 mutant mammalian models; for example, HH signaling is inhibited in Dhcr7 −/− mouse embryonic fibroblasts (MEFs) 69 , as well as in mice deficient for Dhcr7 (Dhcr7 ΔEx8/ΔEx8 mice). 70,71 Our results show that the number of ciliated cells, as well as the length of cilia, is decreased in Dhcr7 −/− osteoblasts, which is responsible for diminished HH signaling.
Previous studies indicate that HH signaling is not dominant in regulating intramembranous ossification; for example, mice with loss of Ihh (Ihh −/− mice), which is expressed at the osteogenic front in cranial bones, 72,73 develop small but normal calvaria, 47,48 and osteogenesis is not affected during intramembranous ossification. 74 Mice deficient for Smo (Smo n/c ;Wnt1-Cre mice) display small but normal skulls in contrast to other severe craniofacial defects. 75 In addition, mice deficient for Shh (Shh n/c ; K14-Cre mice and Shh F/F ;K14-Cre mice), which is specifically expressed in the epithelium in craniofacial regions, show normal skull formation while they exhibit cleft palate. 2,76,77 Thus, loss-offunction of HH signaling does not affect calvaria formation. By contrast, gain-of-function of HH signaling results in defective bone formation; for example, mice with an ectopic HH signaling activation in CNC cells (Smo +/M2 ;Osr2-IrsCre) show cleft palate and osteogenesis defects such as the absence of the palatine processes of the premaxilla and maxilla and a short mandible with ossification defects. 46 In addition, overexpression of Shh in the epithelium (K14-Shh transgenic mice) results in craniofacial bone defects, cleft palate, short limbs, and polysyndactyly through the increased and expanded SHH signaling activation in the mesenchyme. 78 Taken together, although HH signaling may contributes to intramembranous ossification at some degree, upregulation and downregulation of HH signaling in Dhcr7 −/− and Insig1/2 cKO mice is not in agreement with the hypothesis that HH is a factor responsible for cilia-mediated osteogenesis during intramembranous ossification. As to WNT/β-catenin signaling, previous studies show that it positively regulates osteogenesis; for example, constitutive active WNT/β-catenin signaling in bones results in increased osteogenesis in mice, [79][80][81] and WNT/β-catenin signaling inhibition results in a failure of osteogenesis; for example, mice with inactivated β-catenin in mesoderm-derived osteoblast and chondrocyte progenitor cells (Catnby c/c ;Dermo1-Cre mice) show drastically diminished osteogenesis and ectopic cartilage formation in both intramembranous and endochondral ossification. 82 Kif3a −/− and Kif3a F/F ;Wnt1-Cre mice as well as MEFs or embryonic stem (ES) cells from Kif3a −/− , Ofd1 −/− and Ift88 orpk/orpk mice with no functional primary cilia show hyper-activated WNT/β-catenin signaling, compared to ciliated control cells, 83,84 as seen in Dhcr7 −/− osteoblasts. Thus, WNT/β-catenin signaling is well correlated with the osteogenic phenotype in Dhcr7 −/− and Insig1/2 cKO mice during intramembranous ossification.
In summary, we demonstrated the biological significance of intracellular cholesterol metabolism in the regulation of osteogenesis and ciliogenesis using Dhcr7 −/− and Insig1/2 cKO mice and derived cultured osteoblasts. Dhcr7 −/− osteoblasts displayed fewer and shorter primary cilia compared to controls, while Insig1/2 cKO osteoblasts exhibited supernumerary and longer primary cilia compared to controls. Our work places a new focus on primary cilium formation regulated by cholesterol metabolism in the bone. The principles learned from this study promise to be fertile ground for future molecular genetic studies of craniofacial bone development, and may lead to the development of innovative preventive and therapeutic approaches for bone diseases and ciliopathies related to cholesterol metabolic aberrations. Dhcr7 −/− ;Topgal and Dhcr7 +/+ ;Topgal, Dhcr7 −/− ;Axin2 LacZ/+ , and Dhcr7 +/+ ;Axin2 LacZ/+ mice. Genotyping was performed using PCR primers, as previously described. 12,22,85 Pregnant females were treated with simvastatin (S6196; Sigma-Aldrich) at a dose of 10 mg/kg −1 body weight (BW) from E12.5 to E18.5, or from day 7 to day 42, administered by intraperitoneal injection.

Skeletal staining
The three-dimensional architecture of the skeleton was examined by modified whole-mount Alcian blue-Alizarin Red S staining, as previously described. 88 MicroCT MicroCT analysis was performed using a Scanco µCT40 unit in the microCT core facility at Baylor College of Medicine. The data were collected at a resolution of 20 µm. Three-dimensional reconstruction was performed using the BoneJ software.
Histology H&E, BrdU staining, von Kossa staining, LacZ staining, immunohistochemistry, and the TUNEL assay were performed as previously described. 88,89 Antibodies used for immunohistochemistry were the anti-COL1A1 rabbit polyclonal (Abcam), anti-RUNX2 rabbit monoclonal (Cell Signaling Technology), anti-SP7 rabbit polyclonal (Abcam), anti-BrdU rat monoclonal (Abcam), and anti-Ki67 rabbit monoclonal (Abcam) antibodies (Supplementary Table  S1). Click-iT® Plus TUNEL Assay with Alexa 594 (C10618, Molecular probes) was used to detect apoptotic cells, according to the manufacturer's instructions. Fluorescence images were obtained using a confocal microscope (Ti-C2, Nikon).  Fig. 9 Model of primary cilium formation altered in Dhcr7 −/− and Insig1/2 mutant osteoblasts. Ciliogenesis starts with the interaction of the basal body (aka mother centriole) with primary ciliary vesicles (CVs), which can be labeled with RAB11, and then the axoneme grows within the ciliary membrane while fusing with secondary CVs, which can be labeled with RAB8. Eventually the elongated primary cilium fuses with the plasma membrane, allowing the distal part of the cilium to interact with the extracellular milieu. In Dhcr7 knockout (KO) osteoblasts, primary cilia were fewer and shorter than in controls. By contrast, Insig1/2 conditional KO (cKO) osteoblasts showed supernumerary and longer primary cilia compared to controls. Primary CVs labeled with RAB11 accumulated in the cells, and secondary CVs labeled with RAB8 failed to gather at cilium formation sites in Dhcr7 KO osteoblasts. Secondary CVs stained with RAB8 accumulated in Insig1/2 cKO osteoblasts, and the number of basal bodies was abnormally increased in Insig1/2 cKO osteoblasts. During osteogenesis, WT osteoblasts have a single primary cilium that suppresses WNT/β-catenin signaling and activates HH signaling, which induces Col1a1 expression. In Dhcr7 KO osteoblasts, WNT/ β-catenin signaling is hyper-activated (WNT: + + ) and HH signaling is compromised (HH: -), while in Insig1/2 cKO osteoblasts WNT/β-catenin signaling is inhibited (WNT: −) and HH signaling is hyper-activated (HH: + + ).
Osteogenesis and primary cilium formation A Suzuki et al.
Cell culture Primary osteoblasts were obtained from newborn frontal bones. Briefly, the frontal bones were dissected at birth and incubated with 2 mg·mL −1 collagenase II solution (Gibco) while shaking at 350 r·min −1 at 37°C for 1 h. Cells were resuspended in minimum essential medium alpha (MEM-α) supplemented with 10% fetal bovine serum (FBS), penicillin/streptomycin, and L-glutamine. For osteogenic differentiation, osteoblasts were cultured in 12-well plates, and osteogenic differentiation was induced with osteogenic induction medium (MEM-α supplemented with 100 µg·mL −1 ascorbic acid, 5 mmol·L −1 beta-glycerol phosphate, 10% FBS, penicillin/ streptomycin, and L-glutamine) for 28 days. To evaluate osteogenic differentiation, alkaline phosphatase staining, Alizarin Red staining, and von Kossa staining were performed. To induce ciliogenesis, osteoblasts were starved of serum for 24 h, followed by immunofluorescent staining or collection of RNA and protein. To evaluate WNT signaling activity, osteoblasts were starved of serum for 24 h and then treated with either 20 mmol·L −1 lithium chloride (LiCl) or 20 mmol·L −1 NaCl (a negative control) in a serum-free medium for 24 h. For WNT3A treatment, WNT3A-conditioned medium and control medium were prepared from murine WNT3Aoverexpressing cells (L-Wnt3a cells: the American Type Culture Collection [ATCC]) and from control L-cells (ATCC), according to manufacturer's instructions, and used at 10% for each assay for 24 h, respectively. For the cell proliferation assay, osteoblasts were plated onto 96-well plates at a density of 5 000 cells per well and then counted by CCK8 (Dojindo Molecular Technologies) at 24, 48, and 72 h. For the BrdU incorporation assay, osteoblasts were plated onto a 35-mm culture dish at a density of 10 000 cells per dish for 2 days, and then 100 µg·mL −1 BrdU was added for 1 h. To generate Insig1/2 cKO osteoblasts from Insig1 F/F ;Insig2 −/− osteoblasts, the Ad5-CMV-LacZ-Cre virus (Vector Development Laboratory) was transduced at 300 multiplicity of infection (MOI) for 2 days. The Cre recombinase efficiency was evaluated with LacZ staining.

Statistics
The two-tailed student's t-test was applied for statistical analysis. A P-value < 0.05 was considered statistically significant. For all graphs, data are represented as mean ± standard deviation.

Study approval
All animal experiments were reviewed and approved by the Animal Welfare Committee and the Institutional Animal Care and Use Committee of UTHealth.