Conditional Disruption of Pkd1 in Osteoblasts Results in Osteopenia Due to Direct Impairment of Bone Formation*

PKD1 (polycystin-1), the disease-causing gene for ADPKD, is widely expressed in various cell types, including osteoblasts, where its function is unknown. Although global inactivation of Pkd1 in mice results in abnormal skeletal development, the presence of polycystic kidneys and perinatal lethality confound ascertaining the direct osteoblastic functions of PKD1 in adult bone. To determine the role of PKD1 in osteoblasts, we conditionally inactivated Pkd1 in postnatal mature osteoblasts by crossing Oc (osteocalcin)-Cre mice with floxed Pkd1 (Pkd1flox/m1Bei) mice to generate conditional heterozygous (Oc-Cre;Pkd1flox/+) and homozygous (Oc-Cre;Pkd1flox/m1Bei) Pkd1-deficient mice. Cre-mediated recombination (Pkd1Δflox) occurred exclusively in bone. Compared with control mice, the conditional deletion of Pkd1 from osteoblasts resulted in a gene dose-dependent reduction in bone mineral density, trabecular bone volume, and cortical thickness. In addition, mineral apposition rates and osteoblast-related gene expression, including Runx2-II (Runt-related transcription factor 2), osteocalcin, osteopontin, and bone sialoprotein, were reduced proportionate to the reduction of Pkd1 gene dose in bone of Oc-Cre;Pkd1flox/+ and Oc-Cre;Pkd1flox/m1Bei mice. Primary osteoblasts derived from Oc-Cre;Pkd1flox/m1Bei displayed impaired differentiation and suppressed activity of the phosphatidylinositdol 3-kinase-Akt-GSK3β-β-catenin signaling pathways. The conditional deletion of Pkd1 also resulted in increased adipogenesis in bone marrow and in osteoblast cultures. Thus, PKD1 directly functions in osteoblasts to regulate bone formation.

PC1 (polycystin-1) is a highly conserved, receptor-like multidomain membrane protein widely expressed in various cell types and tissues (1,2). Mutations of human PKD1 (polycystic kidney disease gene 1) cause autosomal dominant polycystic kidney disease (ADPKD) 2 (3,4). The genetics of ADPKD is complex, because it is widely held that inactivation of the normal copy of the PKD1 gene by a second somatic mutation in conjunction with the inherited mutation of the other allele is required for renal cyst formation, which occurs in only a subset of the dually affected tubules (5). Although primarily affecting the kidney, ADPKD is also a multisystem disorder (6,7). Extrarenal manifestations include intracranial and aortic aneurysms and cystic disease of liver and pancreas (8 -11). The biological functions of PC1 are poorly defined in some tissues that express PKD1 transcripts, such as bone. Indeed, the absence of clinically demonstrable skeletal abnormalities in patients with ADPKD initially delayed the investigation of PKD1 function in bone. The apparent lack of abnormalities in other tissues expressing PC1 may arise because of differences in the frequency of a second hit somatic mutation, the presence of other modifying factors that may compensate for lack of PC1 function in other organs (12), or failure to detect more subtle phenotypes. For example, lung was not thought to be affected by PKD1 mutations until computed tomography scans of lungs of ADPKD patients showed a 3-fold increase in the prevalence of bronchiectasis compared with controls (13).
Pkd1 is highly expressed in bone, and several mouse models with inactivating mutations of Pkd1 have skeletal abnormalities in the setting of polycystic kidney disease and embryonic lethality (6, 7, 14 -16). Most recently, however, the heterozygous Pkd1 m1Bei mouse, which has an inactivating mutation of Pkd1 and survives to adulthood without polycystic kidney disease, has been shown to develop osteopenia and impaired osteoblastic differentiation (17,18), suggesting that Pkd1 may function in bone. Because homozygous PKD1/Pkd1 mutations in humans and mice are lethal, and most of the existing models are globally Pkd1-deficient, the significance of inactivation of Pkd1 in osteoblasts remains uncertain, and the bone changes might reflect an indirect effect due to loss of PKD1, in the kidney or other tissues.
In the current study, to determine if PKD1 in osteoblasts has a direct function in regulating postnatal skeletal functions, we used mouse genetic approaches to conditionally delete Pkd1 in osteoblasts. We demonstrate that conditional deletion of Pkd1 from osteoblasts using Oc-Cre results defective osteoblast function in vivo and in vitro, and osteopenia, indicating that PKD1 has a direct role to regulate osteoblast function and skeletal homeostasis.

EXPERIMENTAL PROCEDURES
Mice-We obtained the floxed Pkd1 mice from Dr. Gregory Germino at Johns Hopkins University (19) and Oc (osteocalcin)-Cre mice from Dr. Thomas Clemens at the University of Alabama (20). The Pkd1 m1Bei heterozygous mice were available in our laboratory as described previously (18). These mice were bred and maintained on a C57BL/6J background. At first, we created double heterozygous Oc-Cre;Pkd1 m1Bei/ϩ mice and homozygous Pkd1 flox/flox mice. Then double heterozygous Oc-Cre;Pkd1 m1Bei/ϩ mice were mated with homozygous Pkd1 flox/flox mice to generate excised floxed Pkd1 heterozygous (Oc-Cre;Pkd1 flox/ϩ ) and null mice (Oc-Cre;Pkd1 flox/m1Bei or Pkd1 Oc-cko ) as well as Beier Pkd1 heterozygous mice (Pkd1 m1Bei/flox ) and Oc-Cre negative control mice (Pkd1 flox/ϩ , equivalent to wild type). These mice were used for phenotypic analysis. Animal experiments were performed, following review and approval by the University of Kansas Medical Center's Animal Care and Use Committee.
Bone Densitometric, Histomorphometric, and Microcomputed Tomography (CT) Analysis-Bone mineral density (BMD) of femurs was assessed at 16 weeks of age using a LUNAR PIXIMUS bone densitometer (Lunar Corp., Madison, WI). Calcein (Sigma) double labeling of bone and histomorphometric analyses of periosteal mineral apposition rate (MAR), mineralized surface per bone surface (MS/BS), and bone formation rate (BFR) in tibias were performed using the osteomeasure analysis system (Osteometrics). Goldner and Von Kossa staining were performed according to standard protocols (21,22). The distal femoral metaphyses were also scanned using a Scanco CT 40 (Scanco Medical AG, Brüttisellen, Switzerland). A three-dimensional image analysis was done to determine bone volume/total volume and cortical thickness as described previously (18,21).
Detection of Bone Marrow Adipocytes in Long Bones by Oil Red O Lipid Staining-Whole intact femurs with encapsulated marrow were dissected from 18-week-old mice, fixed for 48 h in phosphate-buffered paraformaldehyde, decalcified in 14% EDTA, and then embedded in tissue freezing medium. Cryosectioning was performed on a Leica CM1900 cryostat (Leica, Nussloch, Germany) equipped with a CryoJane frozen sectioning kit (Instrumedics, Hackensack, NJ). 10-m thick sections were then stained with Oil Red O for bone marrow adipocytes as described before (23). Briefly, the sections were rinsed in 60% isopropyl alcohol; stained for 20 min in 0.5% Oil Red O, isopropyl alcohol solution; differentiated in 60% isopropyl alcohol; rinsed in tap water; and mounted in glycerin jelly. Sections were examined with a Leica DM LB microscope equipped with an Optronics digital camera.
Detection of Bone Marrow Fat in Long Bones by CT-Whole intact tibiae with encapsulated marrow were dissected from 18-week-old mice, fixed for 48 h in phosphate-buffered paraformaldehyde, decalcified in 14% EDTA, and stained for 2 h in 2% aqueous osmium tetroxide (OsO 4 ). Bones were rinsed in water for 48 h and then scanned at 6 m resolution using a Scanco CT 40, 45 keV, and 177 A. Quantification of fat volume, density, and distribution throughout the marrow was registered to low contrast decalcified bone.
Real-time RT-PCR-For quantitative real-time RT-PCR, 2.0 g of total RNA isolated from either the long bone of 16-weekold mice or 10-day cultured primary osteoblasts in differentiation medium was reverse transcribed as described previously (24). PCRs contained 100 ng of template (cDNA or RNA), 300 nM each forward and reverse primers, and 1ϫ iQ TM SYBR Green Supermix (Bio-Rad) in 50 l. The threshold cycle of tested gene product from the indicated genotype was normalized to the threshold cycle for cyclophilin A. Expression of total Pkd1 transcripts was performed using the following Pkd1 allelespecific primers: in exon 26, forward primer of normal Pkd1 ϩ transcript (5Ј-CTG GTG ACC TAT GTG GTC AT-3Ј), forward primer of mutant Pkd1 m1Bei transcript (5Ј-CTG GTG ACC TAT GTG GTC AG-3Ј), and common reverse primer (5Ј-AGC CGG TCT TAA CAA GTA TTT C-3Ј); in exons 2-4, forward primer of normal Pkd1 ϩ transcript (5Ј-ATA GGG CTC CTG GTG AAC CT-3Ј) and reverse primer (5Ј-CCA CAG TTG CAC TCA AAT GG-3Ј). The normal Pkd1 ϩ versus cyclophilin A is normalized to the mean ratio of five control mice, which has been set to 1. The percentage of conditional deleted and mutant transcripts was calculated from the relative levels of the normal Pkd1 ϩ transcripts in different Pkd1 exons (25). Serum Biochemistry-Serum osteocalcin levels were measured using a mouse osteocalcin enzyme immunoassay kit (Biomedical Technologies Inc., Stoughton, MA). Serum urea nitrogen was determined using a serum urea nitrogen diagnostic kit from Pointe Scientific, Inc. Serum calcium was measured by the colorimetric cresolphthalein binding method, and phosphorus was measured by the phosphomolybdate-ascorbic acid method (Stanbio Laboratory, TX). Serum TRAP (tartrate-resistant acid phosphatase) was assayed with the enzyme-linked immunosorbent assay-based SBA Sciences mouseTRAP TM assay (Immunodiagnostic Systems, Fountain Hills, AZ).
Primary Osteoblast Culture for Proliferation, Differentiation, and Western Blot Analysis-Primary osteoblasts from newborn mouse calvarias were cultured in ␣-minimum essential medium containing 10% fetal bovine serum and 1% penicillin/ streptomycin as described previously (24). Cell proliferation was detected by bromodeoxyuridine incorporation assays as the manufacturer describes (QIA58, Calbiochem). To induce differentiation, primary osteoblasts were plated at a density of 1 ϫ 10 5 cells/well in a 6-well plate and grown for period of up to 21 days in ␣-minimum essential medium containing 10% fetal bovine serum supplemented with 5 mM ␤-glycerophosphate and 25 g/ml ascorbic acid. Alkaline phosphatase activity and alizarin red-S histochemical staining for mineralization were performed as described previously (24). Total DNA content was measured with a PicoGreen double-stranded DNA quantitation reagent and kit (Molecular Probes, Inc., Eugene, OR).
To examine the amounts of cytoplasmic Akt, GSK, and ␤-catenin, the cells were prepared using 1ϫ passive lysis buffer for 30 min at 4°C (Promega, Madison, WI) and centrifuged at 100,000 ϫ g for 45 min at 4°C. Protein concentrations of the supernatant were determined with a Bio-Rad protein assay kit (Bio-Rad). Equal quantities of protein were subjected to NuPAGE TM 4 -12% BisTris gel (Invitrogen) and were analyzed with standard Western blot protocols (horseradish peroxidaseconjugated secondary antibodies from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) and ECL from Amersham Biosciences). Antibodies against phospho-Akt (Ser-473), Akt, phospho-GSK (Ser-9), and GSK were from Cell Signaling Technology (Beverly, MA). Anti-␤-catenin (sc-7199) and anti-␤-actin (sc-47778) antibodies were from Santa Cruz Biotechnology, Inc.
Transient Transfection-Both MC3T3-E1 and primary osteoblasts were cultured in ␣-minimum essential medium containing 10% fetal bovine serum and 1% penicillin/streptomycin. To examine if PC1 regulates Runx2-P1 promoter activity by coupling with PI3K-Akt signaling, 1 ϫ 10 6 MC3T3-E1 cells were transfected with either control expression vector (sIgဧ) or gain-of-function PC1 C-tail construct (PC1-AT) along with the Runx2-P1 luciferase reporter (p0.42Runx2-P1-Luc) construct by electroporation using Cell Line Nucleofector Kit R according to the manufacturer's protocol (Amaxa Inc., Gaithersburg, MD). A total of 10.2 g of plasmid DNA was used for each electroporation, with 3.6 g of PC1 C-tail construct, 2.4 g of p0.42Runx2-P1-Luc reporter, the indicated amounts of a dominant negative Akt (dn-Akt) construct in combination with empty vector (3.6 g), and 0.6 g of Renilla luciferase-null (RLnull) as internal control plasmid. Promoter activity was assessed by measuring luciferase activity 48 h after transfection in the presence or absence of a PI3K inhibitor (0.1-10 M; LY294002) and a dn-Akt construct (1.2-3.6 g) as described previously (17,18).
To explore potential abnormalities of the WNT pathway in excised floxed Pkd1 null mice, control (Pkd1 flox/ϩ ) and excised floxed Pkd1 null (Pkd1 Oc-cko ) osteoblasts were transiently cotransfected with either pTOPFLASH or pFOPFLASH along with Renilla luciferase-null (RL-null; Promega, Madison, WI) as an internal control as described above. Promoter activity will be assessed by measuring luciferase activity 48 h after transfection in the presence or absence of 100 ng/ml recombinant Wnt3a treatment for the last 8 h.
Statistics-We evaluated differences between groups by oneway analysis of variance. All values are expressed as means Ϯ S.D. All computations were performed using GraphPad Prism5 (GraphPad Software, Inc., La Jolla, CA).
Additive Effects of Global Mutant (Pkd1 m1Bei ) and Conditional Deleted (Pkd1 ⌬flox ) Pkd1 Alleles Suggest a Direct Role for PKD1 in Bone-At 16 weeks of age, the gross appearance and body weight of single global and conditional heterozygous (Pkd1 flox/m1Bei or Oc-Cre;Pkd1 flox/ϩ ) and control (Pkd1 flox/ϩ ) mice were not significantly different. The global Pkd1 heterozygous mice (Pkd1 flox/m1Bei ) and heterozygous conditional deleted Pkd1 mice (Oc-Cre;Pkd1 flox/ϩ ), however, were osteopenic, as evidenced by respective 9 and 7% reduction BMD in both male and female adult mice ( Fig. 2A). The phenotype of the Pkd1 Oc-cKO mice was more severe. The body weight of both male and female Pkd1 Oc-cKO mice was reduced by ϳ16 and 12% (data not shown) compared with the control mice (Pkd1 flox/ϩ ). In addition, Pkd1 Oc-cKO mice had greater loss in BMD, with respective reductions in BMD of 14 and 13% reduction in male and female adult mice ( Fig. 2A). The abnormalities in BMD in the various groups, although present at 6 weeks of age, segregated by gene dose by 16 weeks of age (Fig. 2B). CT analysis revealed that the reduction in bone mass in heterozygous Pkd1-deficient mice (either Pkd1 flox/m1Bei or Oc-Cre;Pkd1 flox/ϩ ) was caused by a reduction in trabecular bone volume (24.3 and 25.5%, respectively) and cortical bone thickness (9.7 and 10.8%, respectively) (Fig. 2C). Pkd1 cKO/m1Bei had greater loss in both trabecular (44.5%) and cortical bone (21.0%) than did single heterozygous mice (Fig. 2C).
Consistent with a low bone mass phenotype by BMD and CT analysis, both Goldner staining in distal femur and Von Kossa staining in vertebrae confirmed marked reductions in bone volume and cortical thickness in Pkd1 Oc-cKO null mice (Fig. 3A). In addition, we found that bone loss was associated with a significant Pkd1 gene dose-dependent decrease in periosteal MAR, MS/BS, and BFR. In this regard, MAR, MS/BS, and BFR were reduced by ϳ36%, ϳ33%, and ϳ57% in heterozygous Pkd1 flox/m1Bei and Oc-Cre;Pkd1 flox/ϩ mice and ϳ65, ϳ60, and ϳ87% in Pkd1 Oc-cKO null mice compared with age-matched controls, respectively (Fig. 3B).
To investigate the effects of Pkd1 deficiency on gene expression profiles in bone, we examined by real-time RT-PCR the expression levels of a panel of osteoblast lineage-, osteoclast-, and chondrocyte-related mRNAs from the femurs of 16-weekold control, heterozygous Pkd1-deficient (Oc-Cre;Pkd1 flox/ϩ and Pkd1 m1Bei/ϩ ), and Oc-Cre;Pkd1 flox/m1Bei mice (Table 1). Bone derived from heterozygous Oc-Cre;Pkd1 flox/ϩ and Pkd1 m1Bei/ϩ mice had measurable reductions in the osteoblast lineage gene transcripts, including Runx2-II, total Runx2, osteocalcin, osteopontin, Bsp (bone sialoprotein), Opg (osteoprotegerin), RankL (Rank ligand), Dmp1 (dentin matrix protein 1), and Phex (phosphate-regulating gene with homologies to endopeptidases on the X chromosome) mRNA levels compared with control mice. Significantly greater reductions of Runx2-II, total Runx2, osteocalcin, osteopontin, Bsp, RankL, and Dmp1 were observed in Pkd1 Oc-cKO null mice. In this regard, the Opg/ RankL expression ratio was increased in a gene dose-dependent manner (Table 1). Consistent with a ratio of Opg/RankL that favors the reduced osteoclastogenesis, bone expression of Trap Osteocalcin-Cre-mediated recombination of excised floxed Pkd1 (Pkd1 ⌬flox ) allele occurred exclusively in bone, whereas non-skeletal tissues retained the floxed Pkd1 allele (Pkd1 flox ). C, real-time RT-PCR analysis of total Pkd1 transcripts. Data are expressed as the percentage expression of wild-type (Pkd1 ϩ and Pkd1 flox ), mutant (Pkd1 m1Bei ), and conditional deleted (Pkd1 ⌬flox ) Pkd1 alleles for each genotype from 5-6 tibias of 16-week-old mice. Expression of total Pkd1 transcripts was performed using Pkd1-allele-specific primers as described under "Experimental Procedures." The normal Pkd1 ϩ versus cyclophilin A is normalized to the mean ratio of five control mice, which has been set to 1. The percentage of conditional deleted and mutant transcripts was calculated from the relative levels of the normal Pkd1 ϩ transcripts in different Pkd1 exons. D, histology of adult kidney. Hematoxylin-eosin-stained sections from 16-week-old mice failed to identify any cystic tubules in either cortical or medullary regions of kidney from Oc-Cre;Pkd1 flox/ϩ or Pkd1 Oc-cko mice, consistent with the absence of Oc-Cre expression in the kidney. In contrast, ablation of Pkd1 in the kidney caused massive cyst formation, which served as a positive control. Cy, cyst. Scale bars, 100 m.
and Mmp9 (matrix metalloproteinase 9), markers of bone resorption, were also reduced in heterozygous Pkd1-deficient mice and to a greater extent in Pkd1 Oc-cKO null mice ( Table 1). Transcripts of chondrocyte-related genes did not differ between heterozygous Pkd1-deficient and Pkd1 Oc-cKO null mice (Table 1).
Changes in gene expression in bone correlated with alterations in serum biomarkers. In this regard, further evidence for osteoblast dysfunction includes a reduction in osteocalcin in serum from 16-week-old heterozygous Oc-Cre;Pkd1 flox/ϩ and Pkd1 m1Bei/ϩ mice (Table 2). Serum levels of TRAP, a marker of bone resorption, were also reduced in heterozygous Pkd1-deficient mice compared with control littermates ( Table 2). As with other parameters, Pkd1 Oc-cKO null mice had greater reductions in bone formation and resorption markers, indicating additive effects of inactivation of both Pkd1 alleles in bone homeostasis. Collectively, these findings suggest that Pkd1-mediated bone loss results from low bone formation rates rather than increased bone resorption (Fig. 2, A-C, and Table 2).
PPAR␥ (peroxisome proliferator-activated receptor ␥), an adipocyte transcription factor, and adipocyte markers, including Lpl (lipoprotein lipase) and aP2 (adipocyte fatty acid-binding protein 2) were increased femurs of Pkd1-deficient mice in a Pkd1 gene dosage-related manner (Table 1). Consistent with increased adipogenic markers, bone marrow exhibited an increased percentage of fat cells in Pkd1 Oc-cKO mice, as evidenced by a higher number of adipocytes and volume of fat droplets in decalcified femurs and tibias stained with Oil Red O and OsO 4 (Fig. 3C). In addition, the inflammatory cytokine IFN␥ (interferon-␥) level, but not TNF␣ (tumor necrosis factor-␣) and TRAIL (TNF-related apoptosis-inducing ligand) expression levels, was significantly elevated in Pkd1-deficient mice (Table 1). Moreover, Alox15 (arachidonate 15-lipoxygenase) was also markedly increased in conditional Pkd1-deficient mice.
Effect of Conditional Deletion of Pkd1 on Osteoblastic Function ex Vivo-To determine the impact of conditional deleted Pkd1 on osteoblast function ex vivo, we examined cell proliferation and osteoblastic differentiation and gene expression profiles in primary osteoblast cultures derived from control and Pkd1 Oc-cKO null mice. Consistent with defects in bone formation, we found that Pkd1 Oc-cKO null osteoblasts had a higher bromodeoxyuridine incorporation than control osteoblasts, indicating a greater proliferation rate in Pkd1 Oc-cKO null osteoblasts (Fig. 4A). In addition, conditional null osteoblasts displayed impaired osteoblastic differentiation and maturation, as evidenced by lower alkaline phosphatase activity, diminished calcium deposition in extracellular matrix, and reduced osteoblastic differentiation markers compared with controls (Fig. 4, B-D). In agreement with increased adipogenic activity in vivo, the cultured primary calvarial cells under osteogenic conditions exhibited a marked increase of adipocyte markers, including PPAR␥, aP2, and Lpl (Fig. 4E), suggesting impairment of osteogenesis and enhancement of adipogenesis in Pkd1 Oc-cKO null osteoblast cultures.

Effect of Conditional Deletion of Pkd1 on PI3K-Akt-GSK-␤-Catenin Signaling Pathway in Osteoblasts-PC1 encoded by
Pkd1 is coupled to multiple signal transduction pathways, including activation of canonical WNT/␤-catenin and Akt-dependent pathways. There is evidence that Akt-dependent serine 9 phosphorylation of GSK3␤ prevents phosphorylation and degradation of ␤-catenin (26 -29). Because the PI3K/Akt path- A, effects of Pkd1 ⌬flox allele on BMD at 16 weeks of age. Similar to Beier Pkd1 heterozygous mice (Pkd1 m1Bei/ϩ ), there was ϳ7-9% reduction in both male and female of BMD in single excised floxed Pkd1 heterozygous mice (Oc-Cre; Pkd1 flox/ϩ ) compared with age-matched control mice (Pkd1 flox/ϩ ), and an even greater reduction (13-14%) in double heterozygous Oc-Cre;Pkd1 flox/m1Bei (Pkd1 Oc-cko ) mice, indicating an additive effect of global mutant and conditional deleted Pkd1 alleles on loss of bone mass. B, age-dependent effects of Pkd1 ⌬flox allele on BMD. Double heterozygous Pkd1 Oc-cko mice displayed a significant decrease in femur BMD compared with Pkd1 m1Bei/ϩ mice until 16 weeks of age but not at 6 weeks of age, indicating an age-dependent effect of Pkd1 ⌬flox allele on bone mass. C, effects of Pkd1 ⌬flox allele on bone structure of femurs and midshaft diaphyses. CT analysis of the distal femoral metaphyses and midshaft diaphyses revealed that double heterozygous Pkd1 Oc-cko mice had greater loss in both trabecular and cortical bone than single Oc-Cre; Pkd1 flox/ϩ and Pkd1 m1Bei/ϩ heterozygous mice, consistent with additive effects of global mutant and conditional deleted Pkd1 alleles on bone structure and a direct role of Pkd1 in bone. Data represent the mean Ϯ S.D. from 8 -10 individual mice. *, significant difference from control (Pkd1 flox/ϩ ); @, significant difference from single heterozygous Oc-Cre;Pkd1 flox/ϩ ; #, significant difference from single heterozygous Oc-Cre;Pkd1 flox/ϩ and Pkd1 flox/m1Bei mice at p Ͻ 0.05, respectively. way and WNT signaling play a role in osteoblastic development (30,31), we determined the level of Akt phosphorylation in Pkd1 Oc-cKO derived osteoblasts. Basal phospho-Akt relative to total Akt expression and phosphorylation of serine 9 of GSK3␤ were reduced in Pkd1 Oc-cKO null osteoblasts compared with controls. To determine if potential cross-talk between the PC1/PI3K/Akt pathway and the WNT/GSK3␤/␤-catenin pathways have functional consequences (28), we next examined the response of osteoblasts derived from Pkd1 Oc-cKO mice to Wnt3a. Pkd1 Oc-cKO -derived osteoblasts had a reduced response to Wnt3a-mediated phosphorylation of Akt when compared with control cells (Fig.  5A). In contrast, the addition of Wnt3a resulted in an increase in phosphorylation of GSK3␤ in control cells, leading to the inactivation of GSK3␤ (Fig. 5A). To determine if inhibition of GSK3␤ activated ␤-catenin (32), we assessed the accumulation of free ␤-catenin in the cytoplasm (33). Consistent with reduction of GSK3␤ phosphorylation, we found that the basal level of cytosolic ␤-catenin was lower in Pkd1 Oc-cKO null osteoblasts and exhibited no increase following Wnt3a treatment, whereas control cells increased cytoplasmic ␤-catenin levels following Wnt3a stimulation (Fig. 5A).
To examine the effect of Pkd1 inactivation on Wnt/␤-catenin transcriptional activity, we examined TOPFlash activity in Pkd1 Oc-cKO null osteoblasts. We observed a significant reduction of basal TOPFLASH activity in primary osteoblasts derived from Pkd1 Oc-cko mice compared with the controls. In addition, Wnt3a-induced TOPFLASH activity was more than 2-fold above basal level in the control osteoblasts, whereas Wnt3a-induced TOPFLASH activity was only 1.4-fold above basal level in the Pkd1 Oc-cKO null mice (Fig. 5B), indicating that loss of PC1 significantly attenuates responsiveness of the Wnt/␤-catenin pathway in osteoblasts.
To examine the role of PC1-dependent Akt activation on transcriptional control of osteoblast development, we examined Runx2-II promoter activity. In previous studies, we have shown that PC1 is coupled to Runx2-II expression, a master regulator

PC1 Regulates Osteoblast Function via PI3K/Akt/GSK3␤
in osteoblast function, and that transfection of the C-terminal region of PC1 (PC1-AT) was sufficient to activate the Runx2-II P1 promoter/reporter construct (17). As shown in Fig. 5, C and D, we found that either the PI3K inhibitor LY294002 or cotransfection with a dn-Akt construct resulted in a dose-dependent inhibition of PC1-AT-mediated increase in Runx2-II P1 promoter activity.
Examination of Wnt-related gene expression in Pkd1 Oc-cKO null osteoblasts further supports impairment of WNT/␤-catenin signaling. In this regard, we found evidence for down-regulation of Wnt10b, Axin2, Cox2, and Runx2-II and up-regulation of the negative regulators sFrp1 and sFrp4 (Fig. 5E).

DISCUSSION
PC1 is expressed in cells within the osteoblast lineage (18), and skeletal abnormalities have been reported in Pkd1 mutant mouse models (14,15,17,18), but from these generalized lossof-function observations it was not clear if the observed skeletal abnormalities were an indirect consequence of loss of PC1 in multiple tissues or a direct effect of loss of PC1 in osteoblasts. In the present studies, we have addressed this question by using Oc-Cre to conditionally inactivate Pkd1 in mature osteoblasts postnatally. First, we have shown that the heterozygous conditional reduction of Pkd1 in osteoblasts in Oc-Cre;Pkd1 flox/ϩ results in an osteopenic bone phenotype indistinguishable from the global Pkd1 flox/m1Bei mice. The fact that Oc-Cre;Pkd1 flox/ϩ mice had a ϳ25% reduction in Pkd1 expression, whereas Pkd1 flox/m1Bei mice had a 50% reduction but had identical effects on bone, suggests that loss of Pkd1 function in osteoblasts is responsible for the observed reduction in bone mass in both models. Moreover, the additive effects on the severity of osteopenia in combined conditional deletion of Pkd1 in osteoblasts superimposed on the inactivated Pkd1 m1Bei allele in Oc-Cre;Pkd1 flox/m1Bei (or Pkd1 Oc-cKO ) mice, which resulted in ϳ2-fold greater (ϳ14%) reduction in BMD associated with a 75% overall reduction in Pkd1 expression in bone, compared with a 50% reduction in other tissues, indicates a dose-dependent function of Pkd1 in mature osteoblasts.
We purposely did not create Oc-Cre;Pkd1 flox/flox mice, due to the relative inefficiency of the Oc-Cre, which, based on the 25% reduction in Pkd1 transcripts observed in Oc-Cre;Pkd1 flox/ϩ mice, may have only reduced expression in Oc-Cre;Pkd1 flox/flox

TABLE 1 Gene expression profiles in 16-week-old mice
Data are mean Ϯ S.D. from 5-6 tibias of 6-week-old individual mice and expressed as the -fold changes relative to the housekeeping gene cyclophilin A subsequently normalized to control (Pkd1 flox/ϩ ) mice.

Gene
Accession no.

TABLE 2 Biochemistry analysis of serum in 16-week-old mice
Data are mean Ϯ S.D. from nine individual mice. Osteocalcin is produced by osteoblasts, and TRAP is produced by osteoclasts. Ex vivo assessment of primary osteoblasts isolated confirmed an intrinsic impairment of osteoblast maturation as well as identified increased proliferation, which are typically inversely related (36). A feature of ADPKD renal cells is increased proliferation rate as well as impaired differentiation of epithelial cells (37); our findings suggest that this phenotypic switch may also occur in osteoblasts. Indeed, the Pkd1 Oc-cKO null osteoblasts showed a greater proliferation rate as well as impaired osteoblastic differentiation and maturation. Because others have shown that the response of the renal epithelium to acquired loss of Pkd1 is determined by the developmental state of the organ (38), our data showing a defect in osteoblast maturation in remodeling bone raises the possibility that Pkd1 may also play a role in the embryogenesis of bone. Osteoblasts and adipocytes undergo renewal from bone marrow-derived precursors, and the ratio of osteoblasts and adipocytes appears to be reciprocally controlled in response to physiological stimuli and aging (39). We also found an inverse dose relationship between Pkd1 expression and adipogenesis marker expression in the Pkd1 Oc-cKO osteoblast cultures compared with the control. In agreement with this in vitro data, our in vivo study also showed strong evidence for lipid droplet accumulation in the bone marrow of Pkd1 Oc-cKO mice compared with the control mice via Oil Red O and osmium staining. This phenomenon was also supported by an enhanced expression of inflammatory mediators, such as IFN␥ and Alox15 (40,41), which, produced by adipose tissues, strongly suppress osteoblastogenesis. The findings of increased fat cells in bone marrow of Pkd1 Oc-cKO mice and impaired development of osteoblast cultures derived from Pkd1 Oc-cKO mice suggest that Pkd1 deficiency may lead to impaired differentiation of mesenchymal stem Primary cultured Pkd1 Oc-cKO osteoblasts exhibited a higher bromodeoxyuridine incorporation than control Pkd1 flox/ϩ osteoblasts for 6 h, indicating increased proliferation in the Pkd1 Oc-cKO osteoblasts. B, alkaline phosphatase (ALP) activity. Primary cultured Pkd1 Oc-cKO osteoblasts displayed time-dependent increments in alkaline phosphatase activities for 14 days of culture, but the alkaline phosphatase activity was significantly lower at different time points compared with control Pkd1 flox/ϩ osteoblasts. C, quantification of mineralization. Alizarin Red-S was extracted with 10% cetylpyridinium chloride and quantified as described under "Experimental Procedures." Primary cultured Pkd1 Oc-cKO osteoblasts had time-dependent increments in Alizarin Red-S accumulation for 21 days of culture, but the accumulation was significantly lower at different time points compared with control Pkd1 flox/ϩ osteoblasts. D and E, gene expression profiles by real-time RT-PCR. 10-day cultured Pkd1 Oc-cKO osteoblasts in osteogenic differentiation medium showed a significant attenuation in osteogenesis compared with control osteoblasts, as evidenced by a significant reduction in osteoblastic markers, including osteocalcin (Oc), bone sialoprotein (Bsp), osteopontin (Opn), and osterix (Osx). However, a marked increase of adipocyte markers, such as PPAR␥, aP2, and Lpl, was observed from the Pkd1 Oc-cKO osteoblasts under the same differentiation medium when compared with control osteoblasts. Data are mean Ϯ S.D. from triple three independent experiments. *, significant difference from control (Pkd1 flox/ϩ ) mice at p Ͻ 0.05. Values sharing the same superscript in B and C are not significantly different at p Ͻ 0.05.

cells into osteoblasts and enhanced differentiation into adipocytes.
Because Oc-Cre-mediated deletion of Pkd1 in osteoblasts would carry forward to the terminal differentiated osteocytes, we cannot exclude a possible role of the osteocyte in the bone phenotype in Pkd1 Oc-cko null mice. In addition to being expressed throughout the osteoblastic lineage during embryogenesis and in postnatal bone, Pkd1 is also expressed in chondrocytes (6). Consequently, the more severe skeletal abnormalities in global Pkd1 null compared with Pkd1 Oc-cKO mice could be due to PC1 regulation of osteoblast differentiation or chondrocyte function during development or to systemic effects caused by the presence of polycystic kidneys (6, 7). Additional studies that ablate Pkd1 earlier in the osteoblast lineage during embryogenesis or later in osteocytes, as well in mature and immature chondrocytes, will be needed to establish a direct role of PC1 in skeletal development and to establish the respective contributions of PKD1 function in preosteoblasts, osteoblasts, osteocytes, and chondrocytes.
We previously demonstrated that PC1 selectively regulates Runx2-II P1 promoter activity in osteoblasts through an intracellular calcium pathway linked to nuclear factor I family and AP-1 transcription factors (17). The PC1 regulates a variety of signal transduction pathways in renal epithelial cells, including Akt1 and GSK3␤ (26 -28), which are also important regulators of bone mass (42,43). The present studies indicate that Pkd1 also regulates PI3K-Akt-GSK3␤ signaling in osteoblasts and that this pathway is upstream of Runx2. The decline in Akt-GSK3␤-␤-catenin signaling in osteoblasts from Pkd1 Oc-cko null mice and the finding that PC1-PI3K-Akt signaling regulates Runx2-II P1 promoter activity suggest that this may represent another PKD1-dependent pathway regulating osteoblast function. The interactions between WNT-and PKD1-dependent signaling pathways, however, are complex, and further studies will be needed to define the mechanism whereby loss of polycystin 1 results in decreased ␤-catenin activity.
The broader role of PC1 in regulating bone physiology is not revealed by our studies. There is no known ligand for PC. Polycystin potentially functions as a mechanosensor, a chemosensor, or as a sensor of cell-cell or cell-matrix interactions (44,45). PC1 colocalizes to the primary cilium (46), which is also present in osteoblasts, osteocytes, and chondrocytes (47,48). Also, cleavage of PC1 at the G protein-coupled receptor proteolytic site upstream of the first TM segment may be required for bioactivity (49), suggesting that PC1 may be constitutively active. Although the precise activator in bone remains to be defined, PC1 presence or its ability to sense environmental cues during bone remodeling may allow it to function as a "hub" or a common connection point that permits cells in the osteoblast lineage to sense diverse environmental signals during skeletal development and translate FIGURE 5. Signaling pathways in Pkd1 Oc-cKO osteoblasts. A, Western blot analysis. Comparison of Akt, GSK3␤, and ␤-catenin expressions in control and Pkd1 Oc-cKO osteoblasts treated with or without Wnt3a (100 ng/ml) for the indicated times. Phosphorylated Akt at Ser-473 (panel 1) coincides with phosphorylation of GSK3␤ at Ser-9 (panel 3), reflecting its inactivation by Akt and followed by accumulation of cytoplasmic ␤-catenin, as detected by Western blot. Total Akt (panel 4), GSK3␤ (panel 4), and ␤-actin (panel 6) were used as loading controls for phospho-Akt, phospho-GSK3␤, and ␤-catenin in the cytoplasm, respectively. Pkd1 Oc-cKO osteoblasts exhibited suppressed activity of the PI3K-Akt-GSK3␤-␤-catenin signaling pathway under basal and Wnt3a-stimulated conditions. B, TCF/LEF-dependent transcriptional activation as assessed by pTOPFLASH activity. Pkd1 Oc-cKO osteoblasts display lower levels of basal TCF/LEF-dependent reporter activity (TOPFLASH) when compared with control Pkd1 flox/ϩ osteoblasts. Wnt3a (100 ng/ml) induced reporter activity by more than 2.0-fold above control in Pkd1 flox/ϩ osteoblasts, whereas Wnt3a induced reporter activity by only 1.4-fold above control in Pkd1 Oc-cKO osteoblasts. C and D, PC1-mediated regulation of Runx2-II-P1 prompter activity via the PI3K/Akt pathway. Wild-type MC3T3-E1 cells were transiently transfected either with control expression vector (sIgဧ) or gain-of-function PC1 C-tail construct (PC1-AT) along with the Runx2-II-P1 luciferase reporter (p0.42Runx2-P1-Luc) construct in the presence or absence of PI3K inhibitor LY294002 or dn-Akt construct. PC1-mediated activation of Runx2-II-P1 promoter activity was dose-dependently diminished by either PI3K inhibitor LY294002 or dn-Akt construct. E, gene expression profiles by real-time RT-PCR. 10-day cultured Pkd1 Oc-cKO osteoblasts in differentiation medium showed a significant attenuation in WNT/␤-catenin signaling compared with control osteoblasts, as evidenced by a significant down-regulation of WNT/␤-catenin-targeting genes, including Wnt10b, Axin2, Cox2, and Runx2-II, and up-regulation of its antagonist genes, such as sFrp1 and sFrp4. Data are mean Ϯ S.D. from triple independent experiments. *, significant difference from control (Pkd1 flox/ϩ ) mice at p Ͻ 0.05. Values sharing the same superscript in C and D are not significantly different at p Ͻ 0.05. these into multiple signaling pathways required for maintenance of bone mass.
In contrast to the role of PKD1 in mouse bone, a PC1-dependent bone phenotype has not been reported in humans with ADPKD, who have superimposed renal osteodystrophy.
It may yet be possible to detect small, clinically unapparent reductions of bone density in humans heterozygous for PKD1 mutations prior to developing kidney failure. In addition, there are some very interesting but rare families with early onset of polycystic kidney disease in newborns that are associated with severe skeletal malformations, similar to those observed in mice with homozygous Pkd1 mutations, suggesting that broadly disrupting both alleles in osteoblasts/osteocytes (analogous to the Pkd1 null mouse as opposed to random second hits in ADPKD) might cause clinically apparent bone disease in humans (12,50).
In conclusion, the finding that selective reduction of Pkd1 in osteoblasts results in osteopenia defines a new signaling paradigm in bone that has the potential to expand our understanding of how bone senses environmental signals maintaining bone mass as well as regulating osteoblast growth and development. The further study of polycystin function in bone could provide a variety of new insights, ranging from new mechanosensing mechanisms potentially involving primary cilium to identifying a molecular target for the development of pharmacological approaches to increased bone mass in osteopenic disorders.