Notch Signaling in Osteocytes Differentially Regulates Cancellous and Cortical Bone Remodeling*

Background: Notch signaling regulates skeletal development and remodeling. Results: Activation of Notch preferentially in osteocytes induces osteoprotegerin and Wnt signaling, decreases cancellous bone remodeling, and increases cortical bone formation; as a result, Notch prevents immobilization-induced osteopenia. Conclusion: In osteocytes, Notch decreases cancellous bone remodeling and enhances cortical bone formation. Significance: Notch has distinct actions in osteocytes leading to a marked increase in bone mass. Notch receptors play a role in skeletal development and homeostasis, and Notch activation in undifferentiated and mature osteoblasts causes osteopenia. In contrast, Notch activation in osteocytes increases bone mass, but the mechanisms involved and exact functions of Notch are not known. In this study, Notch1 and -2 were inactivated preferentially in osteocytes by mating Notch1/2 conditional mice, where Notch alleles are flanked by loxP sequences, with transgenics expressing Cre directed by the Dmp1 (dentin matrix protein 1) promoter. Notch1/2 conditional null male and female mice exhibited an increase in trabecular bone volume due to an increase in osteoblasts and decrease in osteoclasts. In male null mice, this was followed by an increase in osteoclast number and normalization of bone volume. To activate Notch preferentially in osteocytes, Dmp1-Cre transgenics were crossed with RosaNotch mice, where a loxP-flanked STOP cassette is placed between the Rosa26 promoter and Notch1 intracellular domain sequences. Dmp1-Cre+/−;RosaNotch mice exhibited an increase in trabecular bone volume due to decreased bone resorption and an increase in cortical bone due to increased bone formation. Biomechanical and chemical properties were not affected. Osteoprotegerin mRNA was increased, sclerostin and dickkopf1 mRNA were decreased, and Wnt signaling was enhanced in Dmp1-Cre+/−;RosaNotch femurs. Botulinum toxin A-induced muscle paralysis caused pronounced osteopenia in control mice, but bone mass was preserved in mice harboring the Notch activation in osteocytes. In conclusion, Notch plays a unique role in osteocytes, up-regulates osteoprotegerin and Wnt signaling, and differentially regulates trabecular and cortical bone homeostasis.

Notch receptors play a role in skeletal development and homeostasis, and Notch activation in undifferentiated and mature osteoblasts causes osteopenia. In contrast, Notch activation in osteocytes increases bone mass, but the mechanisms involved and exact functions of Notch are not known. In this study, Notch1 and -2 were inactivated preferentially in osteocytes by mating Notch1/2 conditional mice, where Notch alleles are flanked by loxP sequences, with transgenics expressing Cre directed by the Dmp1 (dentin matrix protein 1) promoter. Notch1/2 conditional null male and female mice exhibited an increase in trabecular bone volume due to an increase in osteoblasts and decrease in osteoclasts. In male null mice, this was followed by an increase in osteoclast number and normalization of bone volume. To activate Notch preferentially in osteocytes, Dmp1-Cre transgenics were crossed with Rosa Notch mice, where a loxP-flanked STOP cassette is placed between the Rosa26 promoter and Notch1 intracellular domain sequences. Dmp1-Cre ؉/؊ ;Rosa Notch mice exhibited an increase in trabecular bone volume due to decreased bone resorption and an increase in cortical bone due to increased bone formation. Biomechanical and chemical properties were not affected. Osteoprotegerin mRNA was increased, sclerostin and dickkopf1 mRNA were decreased, and Wnt signaling was enhanced in Dmp1-Cre ؉/؊ ; Rosa Notch femurs. Botulinum toxin A-induced muscle paralysis caused pronounced osteopenia in control mice, but bone mass was preserved in mice harboring the Notch activation in osteocytes. In conclusion, Notch plays a unique role in osteocytes, up-regulates osteoprotegerin and Wnt signaling, and differentially regulates trabecular and cortical bone homeostasis.
Bone remodeling is a process that results in the coordinated resorption and formation of skeletal tissue carried out in basic multicellular units (1)(2)(3). There, osteoclasts resorb bone, and the cavity formed is filled by osteoblasts with newly synthesized matrix. Osteoblasts are derived from mesenchymal cells, and following their maturation, they may die of apoptosis, become quiescent lining cells, or become entombed in the mineralized matrix of bone as osteocytes (4). These cells with a dendritic morphology rest in lacunae and communicate with osteoblasts, lining cells and other osteocytes, through a canalicular network. Osteocytes play a fundamental role in bone remodeling, and their ablation results in bone loss and microstructural deterioration (5).
Bone is a dynamic tissue that responds to mechanical stress with changes in mass and structure to achieve a balance between stress and load-bearing capacity. Mechanical loading can stimulate new bone formation and prevent bone loss (6,7). Conversely, unloading of the skeleton or immobilization can enhance bone resorption, reduce bone formation, and increase bone loss and the risk for osteoporotic fractures (8 -10). Osteocytes play a central role in mechanotransduction and respond to mechanical loading by converting extracellular forces into intracellular signals that regulate specific pathways and by communicating the signals to other skeletal cells to regulate skeletal homeostasis (11)(12)(13)(14).
Notch (Notch1 to -4) are single-pass transmembrane receptors that play a critical role in cell fate decisions (15)(16)(17)(18). Notch regulates skeletal development and homeostasis and osteoblast and osteoclast differentiation (15, 19 -22). Notch is activated by Notch-ligand interactions. There are five classic Notch canonical ligands, which are Jagged1 and -2 and Delta-like 1, 3, and 4 (15). Notch-ligand interactions result in the release of the Notch intracellular domain (NICD) 2 and in its translocation to the nucleus, where it displaces transcriptional repressors and interacts with CSL (for C promoter-binding factor 1, Suppressor of hairless, and Lag-1), termed Rbpj in mice, and Mastermind-like (Maml) proteins to regulate transcription (23)(24)(25)(26). This leads to the induction of Hes1 (hairy enhancer of split 1), Hes5, and Hes7 and Hey1 (Hes-related with YRPW motif 1), Hey2, and HeyL. Skeletal cells express Notch1, Notch2, and low levels of Notch3 transcripts; Notch1 and -2 are considered responsible for the effects of Notch in the skeleton (19,27). Transgenic expression of the NICD of Notch1 under the control of a 3.6-kilobase (kb) fragment of the Col1a1 (collagen type I ␣1) promoter causes osteopenia, whereas NICD expression under the control of a 2.3-kb fragment of the same promoter causes the formation of abundant woven bone (20,21). In accordance with these observations, the inactivation of Notch1 and Notch2 in the developing skeleton causes a transient increase in trabecular bone volume due to enhanced osteoblastic differentiation followed by osteopenia, due to an increase in osteoclast differentiation and bone resorption (19). These and related findings suggest that Notch inhibits both osteoblast and osteoclast differentiation (28).
Recently, we explored the effects of Notch1 in cells of the osteoblastic lineage at various stages of cell differentiation. Activation of Notch in undifferentiated and differentiated osteoblasts caused marked osteopenia. In contrast, activation in osteocytes caused a pronounced increase in bone mass due to a suppression of bone resorption (29). Although the observations indicate a distinct function of Notch in osteocytes, the mechanisms involved and the consequences of the inactivation of Notch1 and Notch2 in osteocytes were not explored. Because of the distinct effect of Notch activation in osteocytes, we postulated that it may protect from the deleterious consequences of immobilization.
The intent of the present study was to define the function of Notch1 and -2 in osteocytes in vivo. To avoid any possible genetic compensation, we studied dual conditional Notch1 and -2 null mice by crossing mice where Notch1 and 2 alleles were targeted with loxP sequences with transgenics expressing the Cre recombinase under the control of the Dmp1 (Dentin matrix protein 1) promoter (Dmp1-Cre) (19,20). We also extended the characterization of mice expressing the NICD of Notch1 preferentially in osteocytes and explored the protective effect of Notch in a model of paralysis-induced osteopenia. For this purpose, Dmp1-Cre transgenics were crossed with Rosa Notch mice, where a loxP-flanked STOP cassette was cloned into the Rosa26 locus upstream of the coding sequence of the Notch1 NICD.

EXPERIMENTAL PROCEDURES
Notch1 and -2 Conditional Mice-Conditional Notch1 and Notch2 mice were provided by F. Radtke (Lausanne, Switzerland) and T. Gridley (Scarborough, ME), respectively (30,31). For the creation of the Notch1 conditional allele, a region of the Notch1 locus containing the 3.5 kb upstream of the putative transcriptional start site and the first exon was flanked by loxP sequences, so that Cre recombination results in the removal of the Notch1 signal peptide (31). For the creation of the Notch2 conditional allele, exon 3 was flanked by loxP sequences so that Cre recombination leads to a frameshift and consequent truncation of the Notch2 protein (30). Notch1 and Notch2 conditional mice were mated to create dual Notch1 loxP/loxP and Notch2 loxP/loxP mice in a 129SvJ/C57BL/6 background. To inactivate Notch1 and -2 in osteocytes, transgenic mice expressing the Cre recombinase under the control of an ϳ10-kb fragment of the Dmp1 promoter (Dmp1-Cre) in a C57BL/6 genetic background were obtained from J. Feng (Dallas, TX) (32). Notch1 loxP/loxP ;Notch2 loxP/loxP mice were mated with Dmp1-Cre transgenics to create Dmp1-Cre ϩ/Ϫ ;Notch1 loxP/loxP ; Notch2 loxP/loxP to cross with Notch1 loxP/loxP ;Notch2 loxP/loxP mice so that 50% of the progeny would be inactive for both Notch genes (Dmp1-Cre ϩ/Ϫ ;Notch1 ⌬/⌬ ;Notch2 ⌬/⌬ ), and 50% would serve as controls (Notch1 loxP/loxP ;Notch2 loxP/loxP ). Male and female experimental and control littermate mice were compared at 1, 3, and 6 months of age. Genotyping was carried out in tail DNA by polymerase chain reaction (PCR) using specific primers ( Table 1). Deletion of loxP-flanked sequences by the Cre recombinase was documented by PCR in DNA extracted from femurs of 1-month-old mice using specific primers (Table 1).
Rosa Notch Mice-Rosa Notch mice were created by D. A. Melton (Harvard University, Cambridge, MA) and obtained from Jackson Laboratory in a 129SvJ/C57BL/6 genetic background (Bar Harbor, ME) (33,34). In these mice, the Rosa26 locus is targeted with a DNA construct encoding Notch1 NICD, preceded by a STOP cassette flanked by loxP sites, cloned downstream of the Rosa26 promoter. Expression of NICD from the targeted Rosa26 locus occurs following the excision of the STOP cassette by Cre recombination of loxP sequences (35,36). To study the activation of Notch1 in osteocytes, homozygous Rosa Notch mice were mated with Dmp1-Cre ϩ/Ϫ transgenics to create Dmp1-Cre ϩ/Ϫ ;Rosa Notch experimental and Rosa Notch littermate controls. Male experimental and control mice were compared at 4 and 7 weeks of age. Genotyping of Dmp1-Cre transgenics and Rosa Notch mice was carried out by PCR in tail DNA extracts ( Table 1). Deletion of the loxP-flanked STOP cassette by the Cre recombinase was documented by PCR in DNA from tibiae using specific primers, and the induction of NICD and Notch target gene expression in femurs was documented by quantitative RT-PCR.
Muscle Immobilization Protocol-The effect of muscle immobilization on the skeleton was tested in 1-month-old Dmp1-Cre ϩ/Ϫ ;Rosa Notch and littermate control male mice. For this purpose, control and experimental mice were injected with botulinum toxin A (BtxA) (Allergan Inc., Irvine, CA) at a dose of 2 units/100 g of body weight administered in equal amounts in the quadriceps and the triceps surae muscle groups or injected with saline under anesthesia using a ketamine/xylazine mixture intraperitoneally (37)(38)(39)(40). The contralateral limb was used as an internal control. Experimental and control mice were sacrificed 3 weeks post-BtxA injection, when bone loss following muscle paralysis reaches its nadir (37). The degree of muscle paralysis was assessed 3 days postinjection using the digital abduction scoring test, where the degree of abduction of the digits in the paralyzed limb is assessed and scored using a scale of 0 -4, with 0 reflecting normal muscle function and 4 indicating total paralysis (41). All animal experiments were approved by the Animal Care and Use Committee of Saint Francis Hospital and Medical Center.
Microcomputed Tomography (CT)-Bone microarchitecture of femurs from experimental and control mice was determined using a microcomputed tomography instrument (CT 40; Scanco Medical AG, Bassersdorf, Switzerland), which was calibrated weekly using a phantom provided by the manufacturer (42,43). Femurs were scanned in 70% ethanol at high resolution, energy level of 55 peak kV, intensity of 145 A, and integration time of 200 ms. A total of 100 slices at midshaft and 160 slices at the distal metaphysis were acquired at an isotropic voxel dimension of 6 m or 216 m 3 and a slice thickness of 6 m and chosen for analysis. Trabecular bone volume fraction and microarchitecture were evaluated starting ϳ1.0 mm proximal from the femoral condyles. Contours were manually drawn every 10 slices a few voxels away from the endocortical boundary to define the region of interest for analysis. The remaining slice contours were iterated automatically. Trabecular regions were assessed for total volume, bone volume, bone volume fraction (bone volume/total volume), trabecular thickness, trabecular number, trabecular separation, connectivity density, and structure model index, using a Gaussian filter ( ϭ 0.8) and user-defined thresholds (42,43). For analysis of femoral cortical bone, contours were iterated across 100 slices along the cortical shell of the femoral midshaft, excluding the marrow cavity. Analyses of bone volume/total volume, porosity, cortical thickness, total cross-sectional and cortical bone area, periosteal perimeter, endosteal perimeter, and material density were performed using a Gaussian filter ( ϭ 0.8) and user-defined thresholds.
Bone Histomorphometric Analysis-Static and dynamic histomorphometry was carried out on experimental and control mice after they were injected with calcein (20 mg/kg) and demeclocycline (50 mg/kg) at an interval of 2 days in 1-monthold or of 5 days in 7-week-old animals. Five-m longitudinal sections of femurs and cross-sections at the mid-femoral diaphysis were cut on a microtome (Microm, Richards-Allan Scientific, Kalamazoo, MI) and stained with 0.1% toluidine blue or von Kossa. Static parameters of bone formation and resorption were measured in a defined area between 360 and 2160 m from the growth plate, using an OsteoMeasure morphometry system (Osteometrics, Atlanta, GA). For dynamic histomorphometry, mineralizing surface per bone surface and mineral apposition rate were measured on unstained sections under ultraviolet light, using a triple diamidino-2-phenylindole/fluorescein/Texas red set long pass filter, and the bone formation rate was calculated. The terminology and units used are those recommended by the Histomorphometry Nomenclature Committee of the American Society of Bone and Mineral Research (44,45).
Mechanical Integrity Testing-Fresh frozen femurs from Dmp1-Cre ϩ/Ϫ ;Notch1/2 ⌬/⌬ conditional null and Notch1/ 2 loxP/loxP controls and from Dmp1-Cre ϩ/Ϫ ;Rosa Notch and control mice were thawed and tested in torsion to measure whole bone mechanical properties. The proximal and distal ends were potted in methacrylate, aligning the longitudinal bone axis with the central axis of applied torque. Tissue hydration was maintained with saline-soaked gauze. The non-embedded central 6-mm span of femur diaphysis was tested by securing the potted distal femur in a mated fixture, twisting the proximal potted end in external rotation at 1°/s (Bose Endura Tec 3200, Minnetonka, MN). Applied torque and twist angle were acquired at 10 Hz. Strength and maximum deformation (twist angle) were defined at maximal applied torque. Torsional stiffness was measured in the linear portion of the torque-deformation curve up to 20°of rotational deformation.
Fourier Transform Infrared (FTIR) Imaging-The compositional properties of bones from Dmp1-Cre ϩ/Ϫ ;Notch1/2 ⌬/⌬ conditional null and Notch1/2 loxP/loxP controls and from Dmp1-Cre ϩ/Ϫ ;Rosa Notch and control littermate mice were mapped in 1-2-m femoral sections of bones embedded in polymethylmethacrylate by FTIR imaging using a PerkinElmer Spotlight 300 infrared imaging System (PerkinElmer Life Sciences) with a pixel size of 6.25 m, at a spectral resolution of 4 cm Ϫ1 , in 256 repeated scans. Three random areas containing either trabecular bone or cortical bone were selected for analysis. Spectra were processed using ISYS software (Spectral Dimensions, Olney, MD), as described (46). In brief, the base-line spectra parameters were normalized to the area of the polymethylmethacrylate peak, the polymethylmethacrylate contribution was subtracted, and the following parameters were calculated for each cortical and trabecular image independently: mineral/matrix ratio, carbonate/phosphate ratio, collagen maturity, crystallinity, and acid phosphate content. Quantitative Reverse Transcription-Polymerase Chain Reaction-Total RNA was extracted from femurs following the removal of the bone marrow by centrifugation, and mRNA levels were determined by quantitative RT-PCR (47,48). For this purpose, equal amounts of RNA were reverse-transcribed using the iScript RT-PCR kit (Bio-Rad), according to the manufacturer's instructions, and amplified in the presence of specific primers (Table 2) and iQ SYBR Green Supermix (Bio-Rad) at 60°C for 45 cycles. Transcript copy number was estimated by comparison with a serial dilution of cDNA for Notch1, Sclerostin (Sost) (both from Thermo Scientific, Waltham, MA), Hey1, Hey2 (both from T. Iso, Los Angeles, CA), HeyL (from D. Srivastava, Dallas, TX), Opg (osteoprotegerin), Wisp1 (Wnt1inducible signaling pathway protein) (both from American Type Tissue Culture Collection (ATCC), Manassas, VA), Tnfsf11 (encoding for Rankl (receptor activator of nuclear factor-B ligand) (from Source BioScience, Nottingham, UK), and Dkk1 (Dickkopf-related protein 1) (from C. Niehrs, Heidelberg, Germany) (49 -51). Reactions were conducted in a CFX96 quantitative RT-PCR detection system (Bio-Rad), and fluorescence was monitored during every PCR cycle at the annealing step. Data were obtained from male mice and are expressed as copy number corrected for Gapdh (glyceraldehyde-3-phosphate dehydrogenase) copy number, estimated by comparison with a serial dilution of cDNA for Gapdh (R. Wu, Ithaca, NY) (52).
Immunofluorescence-Frozen sections of femurs from 1month-old Dmp1-Cre ϩ/Ϫ ;Rosa Notch and Rosa Notch controls were exposed to rabbit polyclonal anti-Nicd antibody (Abcam, Cambridge, MA) or to normal rabbit IgG (Santa Cruz Biotechnology, Inc.) at a 1:500 dilution and to goat anti-rabbit IgG conjugated to fluorophore Alexa 488. Slides were counterstained with 4Ј,6-diamidino-2-phenylindole (DAPI) to label nuclear DNA. The Alexa 488 fluorophore was excited with an argon laser emitting at 488 nm, and the signal was detected with a 505-530-nm band pass emission filter. Images were acquired with a photomultiplier tube using Zeiss ZEN software.
Statistical Analysis-Data are expressed as means Ϯ S.E. Statistical differences were determined by unpaired Student's t test.

RESULTS
Femoral Microarchitecture and Histomorphometry of Dmp1-Cre;Notch1 and -2 Conditional Null Mice-To study the conditional inactivation of Notch preferentially in osteocytes, heterozygous Dmp1-Cre ϩ/Ϫ transgenics were mated with Notch1 loxP/loxP ;Notch2 loxP/loxP mice to create Dmp1-Cre ϩ/Ϫ ; Notch1 loxP/loxP ;Notch2 loxP/loxP mice to be mated with Notch1 loxP/loxP ;Notch2 loxP/loxP and generate Dmp1-Cre ϩ/Ϫ ; Notch1 ⌬/⌬ ;Notch2 ⌬/⌬ experimental and Notch1 loxP/loxP ; Notch2 loxP/loxP littermate controls. To ensure the validity of the controls, in preliminary experiments, we documented that Dmp1-Cre transgenics and Notch1 loxP/loxP ;Notch2 loxP/loxP mice were not appreciably different from wild type mice by femoral CT at 1 month of age. Cre-mediated recombination of loxP sequences was documented in femoral extracts from Dmp1-Cre ϩ/Ϫ ;Notch1 ⌬/⌬ ;Notch2 ⌬/⌬ mice (Fig. 1). The appearance of Notch1 and -2 conditional null mice and their weight were not different from controls. CT of the distal femur revealed that male and female Notch1 and -2 conditional null mice had increased trabecular bone volume at 3 months of age due to an increased number of trabeculae (Table 3 and Fig. 1). The increased bone volume was associated with increased connectivity, and structure model index revealed a tendency toward platelike trabeculae in female mice. The trabecular microarchitectural changes were observed in male and female Notch1 and -2 conditional The GenBank TM accession number identifies the transcript recognized by primer pairs.

Gene Strand Sequence 5-3 GenBank TM accession number
Forward TCCAGGAGTTAAGTGATTTGCTCA NM_018865 Reverse CATGTTACATGACACTGGGCTTC null mice, although they were more pronounced and sustained in female mice. Accordingly, an increase in trabecular bone volume was observed in 3-and 6-month-old female Notch1 and -2 conditional null mice, whereas male mice displayed an increase in trabecular bone only at 3 months of age. Total and cortical bone area at the femoral midshaft were increased in 6-month-old female mice, whereas male mice displayed a modest increase in cortical area at 3 months of age. Other cortical bone microarchitecture parameters determined at the femoral midshaft were not changed in 1-6-month-old Notch1 and -2 conditional null mice of either sex except for a small and transient decline in cortical bone volume and higher porosity noted in 3-month-old male mice (Table 3).
Bone histomorphometric analysis of femurs from Notch1 and -2 conditional null mice at 3 months of age confirmed the microarchitectural findings (Table 4) and demonstrated increased bone volume/tissue volume secondary to an increase in trabecular number in male and female Notch1 and -2 conditional null mice. At 1 month of age, there was an increase in osteoblast number/perimeter and surface and a decrease in osteoclast surface/bone surface in male and female Notch1 and -2 conditional null mice. At 3 months of age, the cellular phenotype evolved, and the number of osteoblasts was not different between Notch1 and -2 conditional null mice and littermate controls. In contrast, there was a 1.5-fold significant increase in osteoclast number and eroded surface in male but not in female mice, possibly explaining the sustained increase in trabecular bone volume observed in 6-month-old female but not male Notch1 and -2 conditional null mice ( Table 4). Parameters of bone formation were not changed in 1-or 3-month-old male or female Notch1 and -2 conditional null mice. Osteoid surface/ bone surface was increased 2-3-fold, although the change was statistically significant only in 1-month-old female and 3-month-old male Notch1 and -2 conditional null mice. Although Notch1 and -2 conditional null female mice had an ϳ2-fold increase in trabecular bone volume, cortical morphometry was similar to controls (Table 3). Biomechanical testing of 5-6-month-old female Notch1 and -2 conditional null mice revealed only a modest reduction in strength (Table 5). Accordingly, compositional analysis of bones from 3-monthold Notch1 and -2 null mice by FTIR imaging revealed modest changes, including an 11% decrease in carbonate/phosphate ratio in trabecular bone and a 5% decrease in collagen maturity in cortical bone (Table 5).
Femoral Microarchitecture and Histomorphometry of Dmp1-Cre;Rosa Notch Mice-The general skeletal phenotype of Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice expressing Notch preferentially in osteocytes was described in a previous report from this laboratory and demonstrated no sexual dimorphism (29). Therefore, the experiments we describe in the Rosa Notch model were conducted in male mice. In agreement with the former study, femoral CT of newly created mice revealed that 1-month-old male Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice had a 3-fold increase in trabecular bone volume/tissue volume due to an increase in trabecular number and had a 9-fold increase in connectivity and a 1.5-fold increase in total cross-sectional and 3.5-fold increase in cortical bone area when compared with control mice (Table 6). Histomorphometric analysis of these newly created mice confirmed the increase in trabecular bone volume and revealed decreased osteoclast number and eroded surface and no changes in osteoblast number (Table 6). Calcein and demeclocycline labels were diffuse, and only a few areas of cancellous bone from Dmp1-Cre;Rosa Notch mice contained well defined condensed labels; these were used to estimate bone formation, which was suppressed by 75% (p Ͻ 0.05). A detailed analysis of the cortical bone revealed that cortical thickness was increased by 3-fold and the periosteal perimeter was increased, whereas the endocortical perimeter was decreased in Dmp1-Cre ϩ/Ϫ ; Rosa Notch mice when compared with Rosa Notch littermates. Histomorphometric analysis of the cortical bone revealed a 3-fold increase in the number of osteoblasts on the endocortical sur-   by CT of 1-, 3-, and 6-month-old Dmp1-Cre ؉/؊ ;Notch1/2 ⌬/⌬ conditional null mice and Notch1/2 loxP/loxP littermate controls CT was performed in distal femurs for trabecular bone and midshaft for cortical bone from 1-, 3-, and 6-month-old male and female Dmp1-Cre ϩ/Ϫ ;Notch1/2 ⌬/⌬ conditional null mice (Notch1/2 ⌬/⌬ ) and control Notch1/2 loxP/loxP littermates (Control). Values are means Ϯ S.E.; n ϭ 4 -7. *, significantly different from controls, p Ͻ 0.05 by unpaired t test.   (Fig. 2), whereas the number of osteoclasts was not changed (not shown). Due to technical reasons, it was difficult to identify and quantify with certainty the cells present on the periosteal surface following the analysis of either methacrylate or paraffin-embedded sections. Mineral apposition rate was increased in both the endocortical and periosteal surfaces of Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice when compared with Rosa Notch controls. The decrease in endocortical and increase in periosteal perimeter, in conjunction with an increase in mineral apposition rate in Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice, is consistent with an increase in bone formation at cortical sites. This effect contrasts with the decrease in bone resorption observed in cancel-lous bone. Although trabecular bone volume and cortical thickness were increased in Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice, their biomechanical properties were not different from those of Rosa Notch control mice possibly because the cortical bone had increased porosity and a trabecular appearance on CT. Similarly, the chemical properties were not different from those of control littermates (Table 6).
Mechanisms Operational in Dmp1-Cre;Rosa Notch Mice-To explore mechanisms that may explain the phenotype of Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice, femurs were analyzed for changes in gene expression. Transcript levels for the Notch1 NICD and its target genes Hey1, Hey2, and HeyL were increased in femurs of
Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice, confirming activation of Notch signaling in skeletal cells (Fig. 3). Immunofluorescence demonstrated Notch1 NICD expression in the bone marrow and femurs of control mice and an increased expression in femurs of Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice, although the cells overexpressing Notch1 NICD were difficult to identify with certainty and included osteocytes as well as other cells present in the bone structure. Notch activation caused a 3-fold increase in Opg and a lesser increase (2-2.5-fold) in Rankl expression. Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice exhibited a marked suppression of Sost, the gene encoding for the Wnt antagonist sclerostin, a 60% decrease in Dkk1 mRNA, and an increase in the Wnt target gene Wisp1. In accordance with the suppression of the Wnt antagonists and induction of Wisp1, immunohistochemistry of femoral sections from 1-month-old Dmp1-Cre ϩ/Ϫ ;Rosa Notch male mice revealed increased ␤-catenin expression in comparison with controls, confirming activation of Wnt/␤-catenin canonical signaling. Most of the increased ␤-catenin staining was observed in osteocytes, although some staining was noted in other skeletal cells and in the bone marrow space.
Effects of Immobilization on the Skeleton of Dmp1-Cre; Rosa Notch Mice-To test whether the preferential Notch activation in osteocytes and consequent induction of Opg and upregulation of Wnt signaling protected the skeleton from the bone loss that occurs during immobilization, the quadriceps and triceps surae muscle groups of 1-month-old Dmp1-Cre ϩ/Ϫ ;Rosa Notch male mice and sex-matched littermate controls were injected with BtxA or saline. Mice were sacrificed 3 weeks later, the known nadir of the effect of BtxA-induced immobilization on bone mass (37). Muscle paralysis was confirmed by the digital abduction test, and saline-injected mice scored 0, whereas Dmp1-Cre ϩ/Ϫ ;Rosa Notch and Rosa Notch controls injected with BtxA scored (means Ϯ S.E.; n ϭ 5-6) 3.6 Ϯ 0.4 and 3.5 Ϯ 0.5, respectively, confirming adequate paralysis (41). CT and histomorphometric analysis revealed a 65-70% decrease in femoral trabecular bone volume in control Rosa Notch mice secondary to a decrease in trabecular number and thickness and a 20% decrease in cortical bone area (Table 7 and Fig. 4). Connectivity density was markedly decreased, and structure model index revealed a tendency toward rod-like trabeculae. There were no changes in osteoblast number or bone formation, but there was an increase in osteoclast number and a not statistically significant 15% increase in eroded surface. Confirming the phenotype observed in 1-month-old mice, saline-injected Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice exhibited a marked increase in trabecular bone volume compared with Rosa Notch controls (Table 7). Dmp1-Cre;Rosa Notch mice exhibited trabecularization of the cortex at mid-diaphysis so that it was not possible to discern cortical from cancellous bone, and the tissue filled the entire section so that there was little if any bone marrow cavity space (Fig. 4A, middle). In contrast to the results observed in control mice, BtxA-induced paralysis did not cause a loss of trabecular bone in Dmp1-Cre ϩ/Ϫ ;Rosa Notch

TABLE 7 Effect of botulinum toxin A-induced immobilization on femoral microarchitecture assessed by CT and histomorphometry in 7-week-old male conditional Dmp1-Cre ؉/؊ ;Rosa Notch mice and controls
CT and bone histomorphometry were performed on femurs from 7-week-old male Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice (Notch) and control littermates, injected with saline or BtxA. Values are means Ϯ S.E.; n ϭ 5-6. *, significantly different between saline and BtxA injection, p Ͻ 0.05 by unpaired t test.
mice, so that their trabecular bone volume, microarchitecture, and histomorphometric parameters were not different from those of limbs of the Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice injected with saline. These observations confirm that BtxA-induced immobilization causes bone loss and indicate that activation of Notch in osteocytes can prevent the osteopenia induced by paralysis.

DISCUSSION
Our findings indicate that the inactivation of Notch1 and -2 preferentially in osteocytes causes an increase in cancellous bone volume, but despite the fact that cortical bone is rich in osteocytes, Notch inactivation had modest consequences in this skeletal compartment. There is no apparent explanation for this finding, particularly in light of the marked skeletal phenotype observed when Notch is activated in osteocytes (29). The phenotype of mice harboring a Notch1 and -2 inactivation could be attributed to an initial increase in the number of osteoblasts and a decrease in osteoclast number. It is of interest that there was a late increase in osteoclast number and eroded surface observed only in male Notch1 and -2 conditional null mice, explaining the transient nature of the increase in bone volume in male and the sustained increase in cancellous bone in female mice. The phenotype observed in conditional Notch1 and -2 null male mice and the late increase in bone resorption is similar to the one reported following the deletion of Notch1 and -2 in the limb bud (19). It is possible that Notch1 and -2 conditional null female mice did not exhibit the resorptive phenotype because of the inhibitory effect of estrogens on bone resorption (53)(54)(55). The findings add credence to the necessity of an independent phenotypic analysis of mice of different sexes (43,56). The in vivo phenotype described indicates that Notch1 and -2 are not only required for the previously reported role of Notch signaling in skeletal development but also required for postnatal bone remodeling, because the inactivation of Notch1 and -2 in osteocytes leads to an increase in cancellous bone (Fig. 5) (19).
In this study, we confirm earlier observations demonstrating an increase in bone mass following the activation of Notch1 preferentially in osteocytes (29). It is important to note that Notch operated by distinct mechanisms in different bone compartments (Fig. 5). In cancellous bone, resorption was suppressed, whereas in cortical bone, the number of osteoblasts and bone formation were increased. Despite a pronounced increase in bone mass, neither the mechanical nor the chemical properties of bone from Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice were affected. This is consistent with findings in other genetically modified mice displaying suppressed bone resorption. For example, Opg transgenic mice exhibit increased bone volume, but the mechanical properties of bone are not enhanced when compared with controls (57). The osteopetrotic phenotype observed following the preferential activation of Notch1 in osteocytes can be explained by an induction in Opg mRNA with a more modest change in Rankl expression, leading to an inhibition of bone resorption and decreased remodeling of cancellous bone. In fact, an osteopetrotic phenotype is also observed following the inactivation of Rank or Tnfsf11, encoding for Rankl, and the transgenic overexpression of Opg (13,(57)(58)(59). An additional mechanism involved in the phenotype of Dmp1-Cre ϩ/Ϫ ;Rosa Notch mice is the activation of Wnt signaling secondary to the suppression of the Wnt antagonists sclerostin and Dkk1 (60). This was documented by the demonstration of increased ␤-catenin in cells present in cancellous bone and the enhanced skeletal expression of the Wnt target gene Wisp1.
Wnt signaling induces osteoblastogenesis, and alterations in this signaling pathway result in profound changes in bone mass, and its activation could explain the increased osteoblast number and bone formation observed in cortical bone (61)(62)(63). In addition, Wnt has a less recognized but important inhibitory effect on osteoclastogenesis and bone resorption, and the constitutive activation of ␤-catenin in the osteoblast and osteoclast lineages causes osteopetrosis (61,62,64). The inhibitory effect of Wnt signaling on bone resorption has been explained by an increase in Opg expression by cells of the osteoblastic lineage and by direct effects on osteoclast precursors independent of Opg (65)(66)(67). Notch can induce Opg directly in cells of the osteoblastic lineage, possibly by regulating Opg transcription because Opg promoter regions contain multiple CSL (Rbpj) consensus sequences, and CSL is the component of the Notch transcription complex that binds to DNA (68). Therefore, the induction of Opg observed following the activation of Notch in osteocytes either could be a direct effect of Notch on Opg expression or could be secondary to enhanced Wnt signaling.
A limitation of the present work is that the expression of Dmp1, used to direct Cre and activate or inactivate Notch, is preferentially but not absolutely selective to osteocytes, and Dmp1 is also expressed by other cells, including terminally mature osteoblasts (69). However, activation of Notch in osteocytes resulted in a unique phenotype that is in contrast to the osteopenic phenotype observed following the activation of Notch in osteoblasts. Because the unloading of the skeleton is characterized by decreased bone formation and increased resorption, leading to bone loss, we postulated that the Notch activation in osteocytes would preserve the bone mass in a model of disuse osteoporosis (37,39,40). Microarchitectural and histomorphometric analyses revealed a decrease in trabecular bone volume and an increase in osteoclast surface in BtxAparalyzed limbs of control mice but not of Notch-expressing mice. These results are congruent with those observed following the conditional inactivation of Tnfsf11 (encoding for Rankl) in osteocytes demonstrating a protective effect from bone loss in a murine tail suspension model (14). They also suggest that FIGURE 5. Notch can suppress osteoblastogenesis, thereby inhibiting bone formation, and Notch inactivation results in an increase in bone mass due to an enhanced number of osteoblasts. Notch inhibits osteoclastogenesis; its inactivation results in an increase in osteoclast number and increased bone resorption. Overexpression of Notch in osteocytes has a prevalent impact in cells of the osteoclast lineage; Notch induces osteoprotegerin directly and through inhibition of sclerostin expression and enhancement of Wnt signaling. As a result, Notch decreases osteoclastogenesis and bone resorption and increases cancellous bone volume. In cortical bone, Notch causes increased bone formation possibly by inducing Wnt signaling, but the bone structure has a cancellous bone appearance.
by inducing Opg and decreasing bone resorption, Notch can prevent bone loss in experimental models of disuse osteoporosis and possibly following skeletal unloading. In addition, the suppression of Wnt antagonist expression and enhanced cortical bone formation may contribute to the protective effect of Notch. Disuse osteoporosis carries a significant risk of fractures, and to prevent or correct the bone loss would prove highly beneficial to individuals afflicted by this disease. Exploring mechanisms responsible for the disease and ways to correct them should offer important information for the development of future therapies to prevent or ameliorate this type of osteoporosis. This report reveals a unique function of the Notch activation in osteocytes in controlling both cancellous and trabecular bone homeostasis, suggesting that Notch may play a role in mechanotransduction and skeletal adaptation. In support of this hypothesis, recently we found that fluid flow shear stress activates Notch signaling in the osteocytic cell line MLO-Y4. 3 In conclusion, Notch activation preferentially in osteocytes leads to an induction of Opg expression and of Wnt/␤-catenin signaling, with a consequent suppression of bone resorption in cancellous bone, and enhancement of bone formation in cortical bone. Notch may prevent the bone loss induced by immobilization.