Antisense oligonucleotides targeting Notch2 ameliorate the osteopenic phenotype in a mouse model of Hajdu-Cheney syndrome

Notch receptors play critical roles in cell-fate decisions and in the regulation of skeletal development and bone remodeling. Gain–of–function NOTCH2 mutations can cause Hajdu-Cheney syndrome, an untreatable disease characterized by osteoporosis and fractures, craniofacial developmental abnormalities, and acro-osteolysis. We have previously created a mouse model harboring a point 6955C→T mutation in the Notch2 locus upstream of the PEST domain, and we termed this model Notch2tm1.1Ecan. Heterozygous Notch2tm1.1Ecan mutant mice exhibit severe cancellous and cortical bone osteopenia due to increased bone resorption. In this work, we demonstrate that the subcutaneous administration of Notch2 antisense oligonucleotides (ASO) down-regulates Notch2 and the Notch target genes Hes-related family basic helix–loop–helix transcription factor with YRPW motif 1 (Hey1), Hey2, and HeyL in skeletal tissue from Notch2tm1.1Ecan mice. Results of microcomputed tomography experiments indicated that the administration of Notch2 ASOs ameliorates the cancellous osteopenia of Notch2tm1.1Ecan mice, and bone histomorphometry analysis revealed decreased osteoclast numbers in Notch2 ASO-treated Notch2tm1.1Ecan mice. Notch2 ASOs decreased the induction of mRNA levels of TNF superfamily member 11 (Tnfsf11, encoding the osteoclastogenic protein RANKL) in cultured osteoblasts and osteocytes from Notch2tm1.1Ecan mice. Bone marrow-derived macrophage cultures from the Notch2tm1.1Ecan mice displayed enhanced osteoclastogenesis, which was suppressed by Notch2 ASOs. In conclusion, Notch2tm1.1Ecan mice exhibit cancellous bone osteopenia that can be ameliorated by systemic administration of Notch2 ASOs.

(6955C3 T) in the mouse genome to reproduce a mutation (6949C3 T) found in a subject with HCS (10,20,22). The mutation creates a stop codon in exon 34 leading to the translation of a truncated NOTCH2 protein of 2318 amino acids (10). The mouse line, termed Notch2 tm1.1Ecan , exhibits NOTCH2 gain-of-function, and homozygous mice display craniofacial developmental abnormalities and newborn lethality. Heterozygous Notch2 tm1.1Ecan mice have cancellous and cortical bone osteopenia due to enhanced bone resorption. This is secondary to an increase in the number of osteoclasts due to enhanced expression of RANKL by cells of the osteoblast lineage as well as due to direct effects of NOTCH2 on osteoclastogenesis (10,15). The discovery of the mechanisms responsible for the bone loss provided clues to offer improved treatments to individuals with HCS, such as the use of the RANKL antibody denosumab (27). However, none of the available interventions offers the opportunity to correct the mechanisms responsible for the disease.
Approaches to down-regulate Notch signaling include the use of biochemical inhibitors of Notch activation, thapsigargin, antibodies to nicastrin, which forms part of the ␥-secretase complex, or to Notch receptors or their ligands, and stapled peptides that prevent the assembly of a NICD/RBPJ/MAML ternary complex (28 -32). A limitation of these approaches is that either they are not specific inhibitors of Notch signaling or they prevent the indiscriminate activation of all Notch receptors, leading to a generalized Notch activation knockdown and side effects. Anti-Notch NRR antibodies have been effective at preventing the activation of specific Notch receptors (33)(34)(35). However, the pronounced down-regulation of Notch activation may result in gastrointestinal toxicity.
Antisense oligonucleotides (ASOs) are single-stranded synthetic nucleic acids that bind target mRNA by Watson-Crick pairing resulting in mRNA degradation by RNase H (36,37). The administration of ASOs has emerged as a novel therapeutic approach to down-regulate WT and mutant transcripts, and it has been successful in the silencing of mutant genes in the central and peripheral nervous system, retina, and liver (38 -45). ASOs have been used to down-regulate specific genes in the skeleton, although information about their possible use as a therapeutic intervention in genetic disorders of the skeleton is limited (46,47).
The purpose of this work was to answer the question whether the phenotype of the Notch2 tm1.1Ecan mouse model could be ameliorated or reversed by down-regulating Notch2 expression with Notch2-specific ASOs. To this end, heterozygous Notch2 tm1.1Ecan and control littermate mice were treated with second generation phosphorothioate-modified ASOs targeting Notch2 and characterized by bone microarchitectural analysis. The direct effects of the Notch2 ASO on osteoblast, osteocyte, and osteoclast cultures from control and experimental mice also were tested.

Effect of Notch2 ASOs on Notch2 expression and signaling in vivo
In initial experiments, we tested whether mouse Notch2 ASOs down-regulated Notch2 mRNA in vivo in tissues where Notch2 is expressed and is known to have a function (10, 48 -52). The subcutaneous administration of ASOs targeting murine Notch2 to C57BL/6 WT mice at a dose of 50 mg/kg caused an ϳ40 -50% down-regulation of Notch2 mRNA 40 h later in the spleen, kidney, and femur and an 80% reduction of Notch2 transcripts in the liver (Fig. 1). In a subsequent experiment, Notch2 ASOs, administered subcutaneously to WT C57BL/6 mice at 50 mg/kg, down-regulated Notch2 mRNA in femur by ϳ40% 48 -96 h after the administration of the ASO.
There was evidence of enhanced Notch signaling in skeletal tissue from Notch2 tm1.1Ecan mice, and the Notch target genes Hey1, Hey2, and HeyL were induced in bone extracts from mutant mice in relationship to control littermates (Fig. 2). The subcutaneous administration of mouse Notch2 ASOs decreased the expression of Notch2 and Notch2 6955C3 T mutant mRNA. Notch2 ASOs also decreased the Notch target genes Hey1 and Hey2 in bone extracts from WT mice and Hey1, Hey2, and HeyL in extracts from Notch2 tm1.1Ecan mice demonstrating a suppressive effect of Notch2 ASOs on Notch signaling in the skeleton. As a result, the mRNA levels of Hey1, Hey2, and HeyL in tibiae from Notch2 tm1.1Ecan mice treated with Notch2 ASOs approached the levels found in tibiae from WT mice treated with control ASOs. A modest induction of Tnfsf11 (encoding RANKL, p Ͼ 0.05) was observed in tibiae from Notch2 tm1.1Ecan mice, and this was reduced by Notch2 ASOs.

Effect of Notch2 ASOs on general characteristics, femoral microarchitecture, and histomorphometry of Notch2 tm1.1Ecan mice
Heterozygous Notch2 tm1.1Ecan mutant male mice were compared with WT sex-matched littermate mice in a C57BL/6 genetic background because the skeletal phenotype was similar in both sexes and the homozygous mutation of Notch2 tm1.1Ecan results in perinatal lethality. Confirming prior results, Notch2 tm1.1Ecan heterozygous mice had ϳ10% less weight than littermate controls, and their femoral length was slightly shorter than that of controls ( Fig. 3) (10). Following the administration of mouse Notch2 ASOs, control and Notch2 tm1.1Ecan experimental mice appeared healthy, although a 6% decrease in weight was noted in WT mice treated with Notch2 ASOs when compared with control ASOs. Femoral length was not affected
Validating previous observations, CT of the distal femur revealed that 2-month-old Notch2 tm1.1Ecan mutant male mice had a significant decrease in trabecular bone volume/total volume (BV/TV) associated with reduced connectivity and a higher structure model index (SMI) (10). Trabecular number and thickness were both reduced in Notch2 tm1.1Ecan mice, contributing to the decrease in BV/TV (Fig. 4). The subcutaneous administration of mouse Notch2 ASOs once a week at 50 mg/kg for 4 weeks did not change microarchitectural parameters of femoral bone in WT mice. In contrast, Notch2 tm1.1Ecan mice receiving Notch2 ASOs had a BV/TV that was 30% greater than in mutant mice receiving control ASOs. As a consequence, BV/TV in Notch2 tm1.1Ecan mice was reduced by 28% when compared with control WT mice, whereas Notch2 tm1.1Ecan treated with control ASOs exhibited a 45% reduction in BV/TV compared with WT littermate controls (Fig. 4). The partial restoration of BV/TV by Notch2 ASOs was associated with a significant increase in trabecular number. Notch2 tm1.1Ecan mice presented with cortical osteopenia and cortical bone was thin, and bone area and cortical thickness were reduced ( Table 1). The cortical osteopenia was not affected by Notch2 ASOs, so the cortical bone area and thickness in Notch2 tm1.1Ecan mice treated with Notch2 ASOs were not different from values obtained in mutant mice treated with control ASOs.
Cancellous bone histomorphometric analysis revealed that osteoclast number was increased in Notch2 tm1.1Ecan mice; Notch2 ASOs did not change osteoclast number in WT mice, but significantly reduced the osteoclast number in Notch2 tm1.1Ecan mice so that the osteoclast number was not different between Notch2 tm1.1Ecan mice treated with Notch2 ASOs and control littermate WT mice ( Table 2). Confirming prior observations, osteoblast number was not different between control and Notch2 tm1.1Ecan mice. Accordingly, dynamic parameters of bone formation were not different between WT and mutant mice and were not affected by Notch2 ASOs. In accordance with the cellular phenotype of Notch2 tm1.1Ecan mice, fasting serum levels of carboxyl-terminal collagen cross-links (CTX) were increased from (means Ϯ S.D.; n ϭ 5-6) control 34.6 Ϯ 2.4 to 49.2 Ϯ 8.9 ng/ml (p Ͻ 0.05) in Notch2 tm1.1Ecan mice treated with control ASOs. Notch2 ASOs reduced the serum levels of CTX in both WT mice to 24.1 Ϯ 9.7 ng/ml (p Ͻ 0.052) and Notch2 tm1.1Ecan mice to 23.2 Ϯ 3.9 ng/ml (p Ͻ 0.05) demonstrating a normalization of bone resorption in experimental mice.

Effect of Notch2 ASOs on Notch2 expression and signaling in osteoblast and osteocyte cell cultures
Mouse Notch2 ASOs added to the culture medium of osteoblast-enriched cells from WT C57BL/6 mice at 1-20 M decreased Notch2 mRNA by ϳ40 to ϳ80% 72 h after ASO addition without evidence of cellular toxicity or changes in cell replication (Fig. 5). The effect of the Notch2 ASO was specific for Notch2 mRNA because, at a concentration as high as 20 M, it did not decrease the expression of Notch1, -3, or -4 mRNA. The NOTCH2 intracellular domain (N2ICD), representative of NOTCH2 cleavage and signal activation, was increased in Notch2 tm1.1Ecan osteoblasts, and the truncated form of NOTCH2, lacking the PEST domain (N2ICD ⌬PEST ), was detected only in Notch2 tm1.1Ecan cells. Therefore, the total levels of N2ICDs, intact and truncated, were ϳ2-fold greater in Notch2 tm1.1Ecan cells than in control cells (Fig. 5). Notch2 ASOs decreased the total levels of N2ICD in WT and Notch2 tm1.1Ecan cells demonstrating a suppression of NOTCH2 activation. Notch2 6955C3 T transcripts were present in cells from Notch2 tm1.1Ecan mutant mice but not in control cultures, and Hey1 and Hey2 transcripts were increased in Notch2 tm1.1Ecan  Numerical values express data normalized to a control value of 100%. *, significantly different between Notch2 tm1.1Ecan mutant and control mice, p Ͻ 0.05; #, significantly different between Notch2 ASO and control ASO, p Ͻ 0.05.
Notch2 6955C3 T mRNA was present in osteocyte-enriched cultures from Notch2 tm1.1Ecan mice and not in control cultures, and Hey2 and Tnfsf11 were significantly increased in Notch2 tm1.1Ecan cells (Fig. 7). Notch2 ASOs suppressed Notch2 mRNA in WT and mutant cells and Notch2 6955C3 T mRNA levels in cells from Notch2 tm1.1Ecan mice and suppressed Hey2 and Tnfsf11 in Notch2 tm1.1Ecan cells to levels that were similar to those found in WT cells treated with control ASOs.

Effect of Notch2 ASOs on Notch2 expression and activity in BMM cultures and osteoclast formation
Notch2 ASOs were added to either BMM cultures at the initiation of the culture period or following the addition of RANKL for 2 days to determine their effect in cells of the myeloid lineage and in osteoclast precursors. Mouse Notch2 ASOs at 1 and 5 M suppressed Notch2 mRNA levels in BMMs from WT C57BL/6 mice by 85-95% and in osteoclast precursors by 70 -85% without evidence of cellular toxicity and without altering cell proliferation (Fig. 8). Confirming results in osteoblast cultures, the N2ICD was slightly increased in Notch2 tm1.1Ecan osteoclasts, and the truncated form of NOTCH2 lacking the PEST domain (N2ICD ⌬PEST ) was detected only in Notch2 tm1.1Ecan cells. Consequently, the total levels of N2ICD, intact and truncated, were ϳ2-fold greater in Notch2 tm1.1Ecan cells than in control cells (Fig. 8). Notch2 ASOs decreased the total levels of

ASOs and Hajdu-Cheney
N2ICD in WT and Notch2 tm1.1Ecan cells demonstrating a suppression of NOTCH2 activation. There was a significant increase in osteoclast formation in BMMs from Notch2 tm1.1Ecan mice cultured in the presence of M-CSF and RANKL (Fig. 9). The increased osteoclastogenesis was prevented by the addition of Notch2 ASOs to BMM cultures at 1 M so that the osteoclastogenic potential of Notch2 tm1.1Ecan cells cultured with Notch2 ASOs was no longer different from that of control cells. The decrease in osteoclastogenesis by Notch2 ASOs in Notch2 tm1.1Ecan cells was associated with a concomitant decrease in Notch2 WT and Notch2 6955C3 T mutant transcripts.

Discussion
Findings from this work confirm that a mouse model replicating a mutation found in HCS displays femoral cancellous and cortical bone osteopenia. The osteopenic phenotype is manifested early in life in mice of both sexes; and in this study, we elected to treat 1-month-old male mice with Notch2 ASOs in an attempt to ameliorate the osteopenic femoral phenotype of Notch2 tm1.1Ecan mice (10). Because only male mice were treated, one needs to be cautious and not to extrapolate the results to female mice. Phenotypic alterations of experimental and control mice were assessed by CT, and analyses required the ex vivo exam of bone following the sacrifice of mice. Consequently, the same animal could not be analyzed before and after the administration of Notch2 ASOs. Another limitation of the work is the fact that all the analyses were performed in femoral bone because the osteopenia of Notch2 tm1.1Ecan mice was established at this skeletal site (10). Although Notch2 ASOs down-regulated Notch2 WT and mutant transcripts in femoral bone, it was not determined whether the same effect occurs at other skeletal, possibly less vascularized, sites. The Notch2 ASO utilized is specific to Notch2 so that the results obtained should not be attributed to the down-regulation of other Notch receptors.

ASOs and Hajdu-Cheney
acro-osteolysis (53). In this work, we confirm that Notch2 has unique actions on trabecular bone physiology and induces osteoclastogenesis by increasing the expression of RANKL by cells of the osteoblast lineage and by inducing the differentiation of cells of the myeloid lineage toward mature osteoclasts (15). Notch2 ASOs decreased both effects in vitro and decreased serum levels of CTX, a marker of bone resorption, so that CTX levels in Notch2 tm1.1Ecan ASO-treated mice were not different from those of WT mice. These effects would explain the amelioration of the osteopenia observed in Notch2 tm1.1Ecan mice.
Notch2 ASOs down-regulated Notch2 and Notch2 6955C3 T transcripts and decreased the enhanced Notch signaling found in Notch2 tm1.1Ecan cells as well as in bone extracts without an effect on basal levels of Notch activation. Only Notch2 tm1.1Ecan mutant cells synthesized the truncated form of the N2ICD (N2ICD ⌬PEST ) and the intact N2ICD. The summation of the intact and truncated forms of N2ICD resulted in an ϳ2-fold greater expression of N2ICD in Notch2 tm1.1Ecan mutants than in control cells, and this was suppressed by Notch2 ASOs confirming the down-regulation of Notch2 signaling. The N2ICD ⌬PEST is more stable than WT N2ICD because it is resistant to ubiquitin-mediated degradation, explaining the gain-of-NOTCH2 function and the induction of Notch target genes in Notch2 tm1.1Ecan cells. Notch2 tm1.1Ecan mice do not exhibit an increase in osteoblast number or a bone-forming response to the increase in bone resorption, indicating a possible negative regulation of osteoblastogenesis or osteoblast function by the Notch2 mutation. However, in this study, we confirm that osteoblast gene markers, such as Bglap (osteocalcin), are not affected in cells from Notch2 tm1.1Ecan mice. The inactivation of Notch2 in cells of the osteoblastic lineage causes an increase in the osteogenic potential of these cells suggesting an inhibitory role of Notch signaling in osteoblastogenesis (54 -56).
Although approaches to down-regulate Notch signaling are various, they are often not specific to this signaling pathway or to a specific Notch receptor. A recent alternative has been the use of antibodies to the negative regulatory region (NRR) of specific Notch receptors that prevent the exposure of the NRR to the ␥-secretase complex and thus the activation of Notch (33)(34)(35). Recently, we demonstrated that anti-Notch2 NRR antibodies reverse the skeletal phenotype of Notch2 tm1.1Ecan mice, and anti-Notch3 NRR antibodies reverse the skeletal phenotype of Notch3 tm1.1Ecan mice, a model of lateral meningocele syndrome (34,35). Although anti-Notch NRR antibodies are

ASOs and Hajdu-Cheney
specific, the significant down-regulation of the Notch receptor throughout the organism may lead to potential side effects, such as gastrointestinal toxicity. In this study, we demonstrate that down-regulation of Notch expression by specific Notch ASOs is a suitable alternative to decrease Notch activation in conditions of Notch gain-of-function. Although the effect of Notch2 ASOs was less pronounced than the one reported with anti-Notch2 NRR antibodies, Notch2 ASOs were effective at ameliorating the skeletal phenotype of Notch2 tm1.1Ecan mice and appeared to be well-tolerated by this experimental model of HCS.
Although attempts have been made to transport ASOs to bone, complex delivery systems were necessary, and the technology has not been applied to the correction of gene mutations in the skeleton (57). In this study, we used a practical systemic approach to down-regulate Notch2 in skeletal and nonskeletal tissue. We demonstrate that a second generation phosphorothioate-modified murine Notch2 ASO down-regulated Notch2 in tissues where the gene is expressed and has a function, including bone. The decrease in Notch2 in a mouse model of Notch2 gain-of-function was associated with a concomitant decrease in Notch target gene expression in skeletal cells documenting a tempering effect on Notch activation. As a consequence, a recovery of bone mass was observed. Although this was not complete, a significant effect on BV/TV was achieved with amelioration of the Notch2 tm1.1Ecan skeletal phenotype.
In conclusion, Notch2 ASOs down-regulate Notch2 expression and signal activation, decrease RANKL and osteoclastogenesis in a model of HCS, and consequently ameliorate its osteopenic phenotype. The down-regulation of NOTCH2 may offer a potential therapeutic opportunity for subjects with HCS in the future.

Notch2 antisense oligonucleotides
ASOs targeting Notch2 mRNA were designed in silico by scanning through the sequence of murine Notch2 pre-mRNA. The entire Notch2 pre-mRNA sequence was covered for potential 16-mer oligonucleotides complementary to the pre-mRNA. Sequence motifs that were intrinsically problematic because of unfavorable hybridization properties, such as polyG stretches, or potential toxicity due to immunogenic responses were avoided. Notch2 ASOs were tested for activity in vitro for down-regulation of Notch2 mRNA in HEPA 1-6 cells at Ionis Pharmaceuticals (Carlsbad, CA), and 14 ASOs targeting Notch2 mRNA were screened for activity and toxicity in vivo at the Korea Institute of Toxicology (Daejeon, Korea). To this end, 7-week-old BALB/c male mice were administered ASOs at a dose of 50 mg/kg once a week by subcutaneous injection for a total of 3.5 weeks (four doses). Body weights were measured weekly, and mice were euthanized 48 h after the last dose of ASO. Liver, kidney, and spleen were weighed, normalized to body weight, and compared with organs from control mice. Blood was obtained by cardiac puncture, and plasma was collected for the measurement of alanine aminotransferase, aspartate aminotransferase, total bilirubin, albumin, and blood urea nitrogen. Total RNA was extracted from liver samples to deter-mine Notch2 mRNA levels corrected for cyclophilin A expression. Based on the information obtained, ASOs found to downregulate Notch2 liver mRNA by more than 75% compared with a control mismatched ASO without toxicity in vivo were selected. Procedures were performed at and approved by the Animal Care and Use Committee of the Korea Institute of Toxicology. For this study, mouse Notch2 ASO Ionis 977472 of sequence GTTATATAATCTTCCA and control mismatched ASO Ionis 549144 of sequence GGCCAATACGCCGTCA were selected.

Notch2 tm1.1Ecan mutant mice
A mouse model of HCS, termed Notch2 tm1.1Ecan , harboring a 6955C3 T substitution in exon 34 of Notch2 was previously reported and validated (10). Notch2 tm1.1Ecan mice were backcrossed into a C57BL/6J background for eight or more generations, and genotyping was conducted in tail DNA extracts by PCR using forward primer Nch2Lox gtF 5Ј-CCCTTCTCTCT-GTGCGGTAG-3Ј and reverse primer Nch2Lox gtR 5Ј-CTCA-GAGCCAAAGCCTCACTG-3Ј. In this study, 1-month-old mice heterozygous for the Notch2 6955C3 T allele and control mice were obtained by crossing heterozygous mutants with WT mice to assess the impact of Notch2 ASOs on the Notch2 tm1.1Ecan skeletal phenotype. One-month-old male Notch2 tm1.1Ecan heterozygous mutant and control sex-matched littermate mice were treated with Notch2 ASO (Ionis 977472) or control ASO (Ionis 549144) that was suspended in PBS and administered subcutaneously at a dose of 50 mg/kg once a week for 4 consecutive weeks. Mice were euthanized at 2 months of age. Studies were approved by the Institutional Animal Care and Use Committee of UConn Health.

CT
Bone microarchitecture of femurs from experimental and control mice was determined using a CT (CT 40; Scanco Medical AG, Bassersdorf, Switzerland), which was calibrated periodically using a phantom provided by the manufacturer (58, 59). Femurs were scanned in 70% ethanol at high resolution, energy level of 55 kilovoltage peaks, 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 size of 216 m 3 , with a slice thickness of 6 m, and then chosen for analysis. Trabecular bone volume fraction and microarchitecture were evaluated starting ϳ1.0-mm proximal from the femoral condyles. Contours were manually drawn a few voxels away from the endocortical boundary every 10 slices 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 SMI, using a Gaussian filter ( ϭ 0.8), and a threshold of 240 per mil eq to 355.5 mg/cm 3 hydroxyapatite (58, 59). For analysis of femoral cortical bone, contours were iterated across 100 slices along the cortical shell of the femoral midshaft, excluding the marrow cavity. Analysis of bone volume/total volume, porosity, cortical thickness, total cross-sectional and cortical bone area, periosteal perimeter, endosteal perimeter, and material density

ASOs and Hajdu-Cheney
were performed using a Gaussian filter ( ϭ 0.8, support ϭ 1), and a threshold of 400 per mil eq to 704.7 mg/cm 3 hydroxyapatite.

Bone histomorphometric analysis
Static cancellous bone histomorphometry was carried out on experimental and control mice. The 5-m longitudinal sections of undecalcified femurs embedded in methyl methacrylate were cut on a microtome (Microm, Richards-Allan Scientific, Kalamazoo, MI) and stained with 0.1% toluidine blue. Static and dynamic 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). Stained sections were used to measure osteoblast and osteoclast number and eroded surface. Mineralizing surface per bone surface and the mineral apposition rate were measured on unstained sections visualized under UV light and a triple diamidino-2-phenylindole/fluorescein/Texas Red set long-pass filter, and bone formation rate was calculated. The terminology and units used are those recommended by the Histomorphometry Nomenclature Committee of the American Society for Bone and Mineral Research (60,61).

Osteoblast-enriched cell cultures
The parietal bones of 3-5-day-old control and Notch2 tm1.1Ecan mutant mice were exposed to Liberase TL 1.2 units/ml (Sigma) for 20 min at 37°C, and cells were extracted in five consecutive reactions (62). Cells from the last three digestions were pooled and seeded at a density of 10 ϫ 10 4 cells/cm 2 , as described previously (63). Osteoblast-enriched cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with nonessential amino acids (both from Life Technologies, Inc.) and 10% heat-inactivated fetal bovine serum (FBS) (Atlanta Biologicals, Norcross, GA) in a humidified 5% CO 2 incubator at 37°C. Confluent osteoblast-enriched cells were exposed to DMEM supplemented with 10% heatinactivated FBS, 100 g/ml ascorbic acid, and 5 mM ␤-glycerophosphate (both from Sigma) in the presence of Notch2 ASO or control ASO at various doses as indicated in figure legends.

Osteocyte-enriched cultures
Femurs from 6-to 7-week-old WT or Notch2 tm1.1Ecan mice were collected after sacrifice; the surrounding tissues were dissected; the proximal epiphysis was excised; and the bone marrow was removed by centrifugation. The distal epiphysis was removed, and to release the endosteal and periosteal cellular layers, the femoral fragments were sequentially exposed for 20-min periods to type II collagenase pretreated with 17 g/ml N ␣ -tosyl-L-lysine chloromethyl ketone hydrochloride and 5 mM EDTA (Life Technologies, Inc.) at 37°C, as described previously (4,64). Osteocyte-enriched bone fragments were obtained and cultured individually in DMEM supplemented with nonessential amino acids, 100 g/ml ascorbic acid, and heat-inactivated 10% FBS for 72 h in a humidified 5% CO 2 incubator at 37°C in the presence of control or Notch2 ASOs, as indicated in figure legends (4, 65).

Bone marrow-derived macrophage (BMM) cultures and osteoclast formation
To obtain BMMs, bone marrow cells were isolated from long bones by flushing the marrow with a 26-gauge needle. Red blood cells were lysed in lysis buffer containing 150 mM NH 4 Cl, 10 mM KHCO 3 , and 0.1 mM EDTA (pH 7.4). The cell suspension was centrifuged, and the pellet was suspended in ␣-minimum essential medium (Life Technologies, Inc.) containing 10% heat-inactivated FBS and recombinant human macrophage colony-stimulating factor (M-CSF) at 30 ng/ml. M-CSF complementary DNA (cDNA) and expression vector were obtained from D. Fremont (St. Louis, MO), and M-CSF was purified as reported previously (66). Cells were seeded at a density of 3 ϫ 10 5 cells/cm 2 on uncoated Petri dishes and cultured for 3 days.
For osteoclast formation, cells were collected following treatment with 0.05% trypsin/EDTA and seeded at a density of 4.7 ϫ 10 4 cells/cm 2 on tissue culture plates in the presence of M-CSF at 30 ng/ml and murine RANKL at 10 ng/ml until the formation of multinucleated tartrate-resistant acid phosphatase (TRAP)positive cells. RANKL cDNA and expression vector were obtained from M. Glogauer (Toronto, Ontario, Canada), and GSH S-transferase-tagged RANKL was expressed and purified as described (67). TRAP enzyme histochemistry was conducted using a commercial kit (Sigma), in accordance with manufacturer's instructions. TRAP-positive cells containing more than three nuclei were considered osteoclasts. Cultures were carried out in the presence of Notch2 or control ASO at various doses as indicated in the figure legends.

Quantitative reverse transcription (qRT)-PCR
Total RNA was extracted from either cultured cells or tibiae following the removal of the bone marrow by centrifugation, and mRNA levels were determined by qRT-PCR (68,69). 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 were amplified in the presence of specific primers (Table 3, all primers from Integrated DNA Technologies (IDT), Coralville, IA), and iQ SYBR Green Supermix (Bio-Rad), at 60°C for 35 cycles. Transcript copy number was estimated by comparison with a serial dilution of cDNA for Bglap (from J. Lian, Burlington, VT), Hey1 and Hey2 (both from T. Iso, Gunma, Japan), HeyL (from D. Srivastava, San Francisco, CA), Notch2 (from Thermo Fisher Scientific), Notch1 (from J. S. Nye, Cambridge, MA), Notch4 (from Y. Shirayoshi, Tottori, Japan), or Tnfsf11 (from Source BioScience, Nottingham, UK) (70 -75). Notch3 copy number was estimated by comparison with a serial dilution of an ϳ100-base pair synthetic DNA template (IDT) cloned into pcDNA3.1 (Thermo Fischer Scientific)
To measure levels of the Notch2 6955C3 T mutant transcript, total RNA was reverse-transcribed with Moloney murine leukemia virus reverse transcriptase in accordance with the manufacturer's instructions (Life Technologies, Inc.) in the presence of reverse primers for Notch2 and of reverse primers for ribosomal protein L38 (Rpl38) ( Table 3). Notch2 cDNA was amplified by PCR in the presence of specific primers, a 6-carboxyfluorescein-labeled DNA probe of sequence 5Ј-CATTGCCTAGGCAGC-3Ј covalently bound to a 3Ј-minor groove binder quencher (Life Technologies, Inc.), and SsoAdvanced Universal Probes Supermix (Bio-Rad) at 60°C for 45 cycles (10,77). Notch2 6955C3 T transcript copy number was estimated by comparison with a serial dilution of a synthetic DNA fragment (IDT) containing ϳ200 bp surrounding the 6955C3 T mutation in the Notch2 locus, and it was cloned into pcDNA3.1(Ϫ) (Life Technologies, Inc.) by isothermal single-reaction assembly using commercially-available reagents (New England Biolabs, Ipswich, MA) (76).
Amplification reactions were conducted in a CFX96 qRT-PCR detection system (Bio-Rad), and fluorescence was monitored during every PCR cycle at the annealing step. Data are expressed as copy number corrected for Rpl38 copy number, as estimated by comparison with a serial dilution of Rpl38 cDNA (from ATCC) (78). In selected experiments, control data were normalized to one following correction for Rpl38 expression.

Serum carboxyl-terminal collagen cross-links assay
Serum samples from control and experimental mice were obtained after an overnight fast. CTX levels were measured using an ELISA kit according to manufacturer's instructions (Immunodiagnostic Systems, Gaithersburg, MD).

Statistics
Data are expressed as means Ϯ S.D. Statistical differences were determined by analysis of variance with Holm-Sidak's post hoc analysis for pairwise or multiple comparisons.