Article

Nature Medicine 14, 299 - 305 (2008)
Published online: 24 February 2008 | doi:10.1038/nm1712

Dimorphic effects of Notch signaling in bone homeostasis

Feyza Engin1, Zhenqiang Yao2,6, Tao Yang1,6, Guang Zhou1, Terry Bertin1, Ming Ming Jiang1,3, Yuqing Chen1,3, Lisa Wang4, Hui Zheng1, Richard E Sutton5, Brendan F Boyce2 & Brendan Lee1,3

Notch signaling is a key mechanism in the control of embryogenesis. However, its in vivo function during mesenchymal cell differentiation, and, specifically, in bone homeostasis, remains largely unknown. Here, we show that osteoblast-specific gain of Notch function causes severe osteosclerosis owing to increased proliferation of immature osteoblasts. Under these pathological conditions, Notch stimulates early osteoblastic proliferation by upregulating the genes encoding cyclin D, cyclin E and Sp7 (osterix). The intracellular domain of Notch1 also regulates terminal osteoblastic differentiation by directly binding Runx2 and repressing its transactivation function. In contrast, loss of all Notch signaling in osteoblasts, generated by deletion of the genes encoding presenilin-1 and presenilin-2 in bone, is associated with late-onset, age-related osteoporosis, which in turn results from increased osteoblast-dependent osteoclastic activity due to decreased osteoprotegerin mRNA expression in these cells. Together, these findings highlight the potential dimorphic effects of Notch signaling in bone homeostasis and may provide direction for novel therapeutic applications.

Evolutionarily conserved Notch signaling has a crucial role in cell fate determination and various developmental processes, as it translates cell-cell interactions into specific transcriptional programs1, 2. Temporal and spatial modulation of this pathway can markedly affect proliferation, differentiation and apoptotic events3. Moreover, the timing of Notch signaling can lead to diverse effects within the same cell lineage4, 5. In mammals, activation of up to four Notch receptors by membrane-bound ligands initiates a process leading to presenilin-mediated cleavage and release from the membrane of the Notch intracellular domain (NICD), which then traffics to the nucleus. The NICD subsequently regulates the expression of genes, in cooperation with the transcription factor RBP-Jkappa and Mastermind-like proteins.

The observation that mutations in the Notch ligand Delta-like–3 (Dll-3) and in the gamma-secretase presenilin-1 cause axial skeletal phenotypes initially caused researchers to link Notch signaling with axial skeletal development6, 7. Recently, several in vitro studies yielded conflicting results that implicated the Notch pathway in the regulation of osteoblast differentiation; however, the in vivo role of Notch signaling in bone homeostasis still remains unknown8, 9, 10, 11, 12.

In this study, we investigate the tissue, cellular and molecular consequences of both gain and loss of function of Notch signaling in committed osteoblasts.

Top

Results

Gain of Notch function results in severe osteosclerosis

To determine the pathological consequences of in vivo gain of Notch function during bone formation and homeostasis, we generated transgenic mice expressing the Notch1 intracellular domain (N1ICD) under the control of the type I collagen (Col1a1) promoter (Supplementary Fig. 1a,b online). Here, gain of Notch function would occur in committed osteoblastic cells, as Col1a1 is both an early and a late marker of the osteoblastic lineage. Notably, founder mice expressing high levels of the transgene were small at birth and showed progressive growth retardation. Analysis of three established lines at 4 weeks of age showed increased bone mass on radiographs and a thickened, osteosclerotic appearance after skeletal preparations (Fig. 1a and Supplementary Fig. 1c). As determined by histology, marrow spaces in 4-week-old transgenic mice were filled with trabecular bone composed predominantly of immature woven, rather than lamellar, bone and surrounded by fibrotic marrow containing cells with morphologic features of early osteoblastic precursors, suggesting the increased proliferation of these cells (Fig. 1b). The cortices of the bones were also composed of woven bone, and this phenotype was present in 11-week-old mice as well. Toluidine blue staining of 11-week-old transgenic mice indicated an increased number of osteoblasts in vertebrae (Fig. 1c). Quantitative histomorphometry of an established mouse line confirmed the significant increase in trabecular bone volume and osteoblast surface area, which would be consistent with the high bone mass being due to increased osteoblastic activity (Fig. 1d). This increased osteoblastic activity led to increased production of osteoid (Fig. 1d) and bone formation (Supplementary Fig. 1d).

Figure 1: Gain of Notch function in transgenic mice cause osteosclerosis.

Figure 1 : Gain of Notch function in transgenic mice cause osteosclerosis.

(a) X-ray and skeletal preparations of 4-week-old transgenic mice show severe osteosclerosis (white arrows) in the skull, ribs and long bones. WT, wild-type mice; Tg, transgenic mice. (b) H&E staining of 4-week-old WT (top) and Tg (bottom) mouse hind limbs show immature woven bone formation with little distinction between cortex and marrow (black arrow). B, bone; BM, bone marrow. Scale bars, 500 mum (left) and 50 mum (right). (c) Toluidine blue staining of lumbar sections of 11-week-old WT (top) and Tg (bottom) mice reveals an increased number of osteoblasts (black arrow). Ob, osteoblasts lining bone. Scale bars, 500 mum (left) and 50mum (right). (d) Von Kossa staining of lumbar sections of 11-week-old mice (top left) and trabecular bone volume density (BV/TV) analyses of 4-week- and 11-week-old mice (n = 4; top right). Scale bar, 500 mum. Histomorphometric analyses of 4-week-old mice (bottom left) (n = 5) and Goldner's staining of spinal trabecular bone (bottom right). Boxed area is enlarged in the right panel. Scale bars, 500 mum (left) and 50 mum (right). Ob.S/BS, osteoblast surface per bone surface; OS/BS, osteoid per bone surface. (e) BrdU staining of osteoblastic cells from P6 calvaria (n = 5). Scale bars, 200 mum (left) and 20 mum (right). (f) qRT-PCR of total RNA from 4-week-old mouse calvaria (n = 4). Osx, osterix; Alp, alkaline phosphatase; Bsp: bone sialoprotein; Oc, osteocalcin. (g) qRT-PCR of total RNA from 4-week-old mouse calvaria and forelimbs (n = 4). Opg, osteoprotegerin. *P < 0.05 between WT and Tg.

Full size image (97 KB)

Because bone formation and resorption are coupled in vivo, we analyzed the status of osteoclasts by staining for tartrate-resistant acid phosphatase (TRAP) activity in osteoclasts from bone sections of 4-week-old mice. Although total TRAP staining was qualitatively increased in the limb sections of transgenic mice (Supplementary Fig. 1e), consistent with increased bone mass and remodeling, the osteoclastic parameters normalized to bone surface (that is, osteoclast number per mm of bone surface and osteoclast surface) were significantly decreased in trabecular bone of transgenic mice (Supplementary Fig. 1e). Together, these data support the notion that gain of Notch function in committed osteoblastic lineage cells stimulates the proliferation of early osteoblastic precursors that differentiate into immature osteoblasts producing increased amounts of immature woven bone. Although osteoclastic activity was secondarily stimulated by this massive osteoblastic proliferation, bone formation much greatly outweighed bone resorption, leading to an osteosclerotic phenotype.

To determine the underlying cellular and molecular mechanism for the increase in early osteoblastic precursors in transgenic bone, we cultured postnatal day 6 (P6) calvarial osteoblasts and found significantly increased numbers of BrdU-positive cells, consistent with increased cellular proliferation (Fig. 1e). Quantitative real time RT-PCR (qRT-PCR) of 1-month-old calvarial total RNA showed an increased abundance of early osteoblastic differentiation markers, including osterix (encoded by Sp7), alkaline phosphatase and bone sialoprotein. In contrast, later markers of osteoblast differentiation, including osteocalcin, were downregulated (Fig. 1f). To exclude the possibility that the increased bone mass was due to decreased osteoclastic activity, we assessed the expression of markers that regulate macrophage differentiation along the osteoclastic lineage in the forelimbs of 4-week-old transgenic mice. RANK ligand (RANKL, encoded by Tnfrsf11a), osteoprotegerin, TRAP and macrophage colony–stimulating factor (M-CSF) were all highly expressed, suggesting that the hyperproliferation of the early osteoblastic pool was associated with increased production of both pro- (RANKL and M-CSF) and anti- (osteoprotegerin) osteoclastic differentiation factors by osteoblasts (Fig. 1g).

Notch regulates osteoblastic transcription factors

How Notch signaling regulates these processes on a biochemical level is unknown. Osteoblast differentiation from mesenchymal stem cells and subsequent maturation steps require the function of the runt-domain transcription factor Runx2 and the zinc-finger transcription factor osterix. Runx2 is required for commitment of mesenchymal osteochondroprogenitors to the osteoblastic lineage, differentiation into mature osteoblasts and terminal differentiation into osteocytes. In contrast, osterix is important in expansion of the early osteoblastic pool13. Whereas bone sialoprotein and alkaline phosphatase are markers of early osteoblasts, osteocalcin is a marker of later, mature osteoblasts. To determine the mechanistic basis of Notch action in this context, we tested the effects of Notch expression on these key transcriptional regulators of osteoblast differentiation and maturation. N1ICD alone was able to directly bind Runx2 and repress its transactivation of a reporter osteocalcin enhancer in vitro (in Cos7 and rat osteosarcoma Ros17/2.8 cells; Fig. 2a–c and Supplementary Fig. 1f). Electrophoretic mobility shift assays (EMSAs) showed that N1ICD could inhibit Runx2 binding to a target cis element in the type X collagen promoter (Supplementary Fig. 1g). Notably, there was marked downregulation of Runx2 protein in P2 calvaria of transgenic mice (Fig. 2d). Thus, the downregulation of osteocalcin and the delay in late osteoblast differentiation in vivo is probably due, in part, to direct repression of Runx2 by Notch at the protein level. At the same time, we observed upregulation of osterix mRNA expression in the P2 calvaria of transgenic mice (Fig. 1f). Moreover, N1ICD activated the Sp7 promoter in transient transfection studies in C2C12 cells that were induced to differentiate into osteoblasts with bone morphogenic protein-2 treatment (Fig. 2e). These data suggest that Notch can induce proliferation of committed osteoblast precursors by directly upregulating transcription of Sp7, while inhibiting their maturation by repressing the function of Runx2.

Figure 2: Notch regulates key osteoblast transcription factors and cell cycle proteins.

Figure 2 : Notch regulates key osteoblast transcription factors and cell cycle proteins.

(a) Relative luciferase activity in Cos7 cells transfected with Runx2 and Runx2-dependent osteocalcin enhancer luciferase reporter with increasing dosage of N1ICD. Western blot analyses of cell lysates using the indicated antibodies are shown. (b) Transfection of N1ICD inhibits endogenous Runx2 activity in ROS17/2.8 osteosarcoma cells as shown by relative luciferase activity of the Runx2-responsive osteocalcin enhancer. Western blot analyses of cell lysates using the indicated antibodies are shown. (c) GST pull-down assay with amino-terminal–truncated (GST-NICDRA) and carboxy-terminal–truncated (GST-NICDTAD) Notch fusion proteins and in vitro transcribed-translated, 35S-methionine–labeled Runx2 (IVT-Runx2). Strongest binding is noted with the carboxy-terminal portion of Notch. (d) Decreased Runx2 protein abundance was detected by western blot analyses and quantified with densitometry on P2 calvarial protein extracts in Tg and WT mice. (e) Relative luciferase activity in C2C12 cells transfected with Osterix promoter luciferase reporter gene and N1ICD. (f) qRT-PCR of cell cycle markers on RNA obtained from 4-week-old mouse calvaria (n = 4). (g) Western blot analyses of P2 calvarial protein extracts to detect cell cycle markers and quantification by densitometry. *P < 0.05 between WT and Tg.

Full size image (88 KB)

To further understand the biochemical basis of the effects of Notch on osteoblastic proliferation, we analyzed the expression of cell cycle markers and detected increased mRNA expression of cyclin D and cyclin A by qRT-PCR in osteoblasts overexpressing N1ICD (Fig. 2f). This correlated with increased cyclin D and cyclin E expression at the protein level (Fig. 2g). We did not, however, observe significant differences in the abundance of two other important cell cycle regulators implicated in bone homeostasis, p53 and Rb. Of note, it has been shown with in vitro and ex vivo studies that Runx2 can suppress osteoblast proliferation and promote osteoblast maturation by supporting exit from the cell cycle14, 15. Moreover, cyclin D1–cyclin-dependent kinase-4 (cdk4) can induce Runx2 ubiquitination and degradation, and thus Runx2 activity can be regulated by the cell cycle machinery16. Hence, gain of Notch function can inhibit osteoblast maturation by direct repression of Runx2 activity, as well as by repression of Runx2's antiproliferative effects via cyclin D1 upregulation.

Loss of Notch signaling leads to age-related osteoporosis

To determine whether the pathological effects of gain of Notch signaling reflect a physiological function during bone homeostasis, we generated a tissue-specific model of loss of Notch signaling in osteoblasts. Because all Notch receptors are expressed in osteoblasts (data not shown), we abolished Notch signaling by generating null mice for both presenilin-1 (Psen1) and presenilin-2 (Psen2). Because Psen2-null mice are viable and fertile, we generated double homozygotes for the Psen2-null allele and the Psen1 floxed allele, but heterozygous for the type 1 collagen Cre recombinase transgene (Psen1f/fPsen2-/-Col1a1Cre/+, or DKO). DKO mice were compared to their Psen1f/fPsen2-/- littermates as controls, and efficient deletion of the Psen1f/f allele (approximately 92%) was confirmed by RT-PCR for presenilin mRNA expression and genomic PCR for DNA recombination in both calvarial osteoblasts and tail DNA, respectively (Supplementary Fig. 2a,b online). Moreover, we confirmed that this led to decreased NICD processing by western blot analysis (Supplementary Fig. 2c). Histomorphometric analyses of DKO mice showed that 6-month-old, but not 3-month-old, mice were osteoporotic, a decreased tissue bone mass phenotype that is the opposite of the osteosclerotic tissue phenotype in gain-of-Notch-function transgenic mice (Fig. 3a–c). Bone formation rates, osteoblast surfaces and mineralized surfaces in vertebrae and long bones in the DKO mice were similar to those in control mice (Supplementary Fig. 3 online). However, osteoclast numbers, osteoclast surfaces and eroded surfaces were increased in DKO vertebrae and long bones at 6 months of age, but not at 3 months of age (Fig. 3d,e and Supplementary Fig. 3). These findings suggest that loss of Psen1 and Psen2, and thus all Notch signaling in osteoblasts, led to osteoporosis through activation of osteoclastogenesis and the subsequent increased bone resorption over bone formation with age-related penetrance.

Figure 3: Loss of Notch signaling via presenilin deletion causes osteoporosis.

Figure 3 : Loss of Notch signaling via presenilin deletion causes osteoporosis.

(a) Von Kossa staining of 6-month-old lumbar vertebrae of Psen1f/fPs2-/- (Control) and DKO mice showed osteoporosis. Scale bar, 500 mum. (b) Micro-CT reconstruction of distal femur from 6-month-old DKO mice showing decreased trabecular bone. Scale bar, 1 mm. (c) Histomorphometry of 3-month-old and 6-month-old DKO L3–L4 spine showing age-related penetrance of low bone mass phenotype. (d) TRAP staining (red) of 6-month-old lumbar vertebral sections from DKO and control mice indicated increased osteoclast staining. Scale bar, 200 mum. (e) Histomorphometry of 6-month-old DKO and control (n = 5 per group) mouse tibiae revealed decreased BV/TV (top left) and increased osteoclast number per bone volume (N.Oc/BV; top right), osteoclast number per bone surface (N.Oc/BS), osteoclast surface per bone surface (Oc.S/BS) and eroded surface per bone surface (ES/BS) in DKO mice. *P < 0.05.

Full size image (98 KB)

Activated osteoblasts support osteoclast formation and differentiation from osteoclast precursors by expressing M-CSF and RANKL, but they also inhibit this process through osteoprotegerin, which binds to and inactivates RANKL. To further examine the effects of loss of Psen1 and Psen2 in osteoblasts on osteoclastogenesis, we performed osteoblast-osteoclastic precursor (OCP) coculture studies. In this ex vivo assay, P7 DKO calvarial osteoblasts stimulated the formation of more osteoclasts from wild-type, spleen-derived OCPs than did wild-type osteoblasts, suggesting that Psen1 and Psen2 deletions can affect osteoclastogenesis in a non–cell-autonomous fashion (Fig. 4a). To determine whether this effect was specific for Notch signaling, we tested whether heterologous expression of N1ICD after lentiviral transduction of DKO osteoblasts could suppress osteoclastogenesis in coculture studies (Fig. 4b and Supplementary Fig. 4a online). Compared to control vector expressing EGFP, lentiviral transduction of N1ICD into Psen1-Psen2 mutant osteoblasts was able to suppress osteoclastogenesis, suggesting that the DKO phenotype was primarily due to presenilin effects on Notch signaling.

Figure 4: Loss of Notch signaling through presenilin deletion increases the osteoclastogenic pool.

Figure 4 : Loss of Notch signaling through presenilin deletion increases the osteoclastogenic pool.

(a) Cocultures of WT or DKO P7 (n = 2, as indicated by 'Pair 1' and 'Pair 2') spleen cells (Spl) and osteoblasts (Ob) stained (images) and quantified (graph) for TRAP+ osteoclasts. Scale bar, 250 mum. (b) Cocultures of P7 (n = 2, as indicated by 'Pair 1' and 'Pair 2') N1ICD lentivirus–transduced DKO osteoblasts with WT spleen cells quantified for TRAP+ osteoclasts. (c) Bone marrow cells obtained from 3-month-old (n = 3) WT and DKO mice were stained and subjected to FACS to detect the CD11b+Gr-1-/lo cells and the c-Fms+ population specific for osteoclast precursors in both total gated and CD11b+Gr-1-/lo-gated populations. Representative histograms show total c-Fms+ cells and c-Fms+ cells in the CD11b+Gr-1-/lo population. Percentages indicate the proportion of total cells in a given gate (indicated by boxes). (d) qRT-PCR for osteoblast markers in total RNA obtained from P4 mouse (DKO versus control calvaria (n = 5 per group)). *P < 0.05.

Full size image (94 KB)

The in vivo relevance of this was confirmed by flow cytometric analysis of markers on bone marrow cells from 3-month-old DKO mice. This showed increased staining of early OCPs in total cells (cFMS+) and in more differentiated OCPs (CD11b+Gr-1-/lo) compared to controls, indicating an expansion of the OCP pool in DKO mice (Fig. 4c). To determine whether this increase in osteoclast differentiation was due to an imbalance of osteoblastic inductive (RANKL and M-CSF) versus suppressive (osteoprotegerin) signals, we analyzed their mRNA expression in DKO versus control bone at P4. We found comparable expression of RANKL mRNA in DKO and control mice, but expression of osteoprotegerin mRNA was markedly decreased (Fig. 4d) in DKO mice. Similarly, we found decreased osteoprotegerin production in cultured DKO calvarial osteoblasts as compared to wild-type calvarial osteoblasts (Supplementary Fig. 4b). Hence, under physiological conditions, Notch signaling enabled by Psen1 and Psen2 function in osteoblasts represses osteoclast differentiation by regulating osteoprotegerin expression.

Together, these in vivo gain-of-function and loss-of-function studies lend support for a central role of Notch and presenilin signaling in regulating both osteoclastogenesis and immature osteoblastic proliferation during bone homeostasis (Fig. 5).

Figure 5: Model for Notch's dimorphic effects in bone homeostasis.

Figure 5 : Model for Notch's dimorphic effects in bone homeostasis.

In established osteoblastic lineages, pathological gain of Notch function activates expansion of the immature osteoblastic pool by increasing transcription of the genes encoding osterix, cyclin D and cyclin E and by repressing the function of Runx2 by direct interaction and inhibition of its binding. Physiologically, it inhibits osteoclastogenesis by increasing osteoprotegerin production over RANKL production.

Full size image (28 KB)

Top

Discussion

Until now, few primary signaling mechanisms regulating osteoblast differentiation and function during bone homeostasis have been identified in vivo by genetic and biochemical studies. Wnt signaling via LRP5/6 coreceptors and canonical beta-catenin activity are required for osteoblast lineage commitment and function17, 18, 19. Activation of this pathway leads to high bone mass20, 21. Activating mutations in transforming growth factor-beta in humans is associated with increased bone formation and inhibition of bone resorption22. However, and not unexpectedly, apparently discrepant results in vivo have been observed, depending on the timing of the gain versus the loss of transforming growth factor-beta function. Similarly, Notch signaling probably has temporal and spatial dependence, as well.

In bone, our data suggest that Notch and presenilin signaling may be important in the physiological regulation of osteoclastogenesis by osteoblasts. Moreover, it raises the question of whether loss of Notch signaling contributes to age-related osteoporosis, as this type of osteoporosis is associated with increased resorption over bone formation23, as is seen in our DKO model. We discovered that one function of Notch in committed osteoblasts is to regulate osteoclastogenesis via regulation of osteoprotegerin production. The magnitude of osteoprotegerin dysregulation and the age-related penetrance of the osteoporosis in the loss-of-function mouse phenotype correlate well with epidemiological data in humans, in which age-related osteoporosis has been associated with changes in osteoprotegerin production24, 25, 26, 27. Furthermore, the report that heterozygote osteoprotegerin-mutant mice have an age-related osteoporotic phenotype suggests that this mechanism is sufficient for disease pathogenesis28. What is unclear is whether osteoprotegerin dysregulation is due to direct regulation by N1ICD or by its target transcription factors, given the still poorly characterized osteoprotegerin genetic regulatory region. Further studies showing chromatin immunoprecipitation analysis of a well-defined functional osteoprotegerin promoter with N1ICD or with its target genes would help to address this issue. Likewise, our studies do not address the potential role of Notch signaling before osteoblastic commitment in the mesenchymal stem cell (Fig. 5). Here, Runx2 has the central role in osteoblastic commitment. Our data on the Notch-Runx2 interaction suggest that early loss of function of Notch would actually lead to increased commitment to the osteoblastic lineage and perhaps depletion of the mesenchymal stem cell compartment.

In a pathological disease context, our findings show that activation of Notch signaling in the committed osteoblastic lineage leads to expansion of an immature osteoblast pool. The primary mode of action is transcriptional upregulation of the early osteoblast transcription factor osterix and increase of cyclin D and cyclin E proteins. These data raise the question of the potential contribution of activation of Notch signaling in human diseases related to osteoblastic proliferation, such as in bone pathologies including human osteosarcomas. The substantial upregulation of cyclin D1 in the transgenic mice correlates with the observation in humans that 10% of osteosarcomas show amplification of the chromosomal region encoding cyclin D1(ref. 29). Although our data suggest that Notch can directly interact with Runx2 to inhibit its binding to target cis elements and its prodifferentiation function, this is probably not the main determinant of the gain-of-function phenotype in mice.

Finally, our data have key therapeutic implications. There are few anabolic bone agents for the treatment of osteoporosis, with most therapies targeted at inhibition of bone resorption. Upregulation of Notch signaling may represent a potential approach for increasing bone formation over bone resorption, as well as for inhibiting osteoclastogenesis. However, it is clear that temporal effects of Notch on other cellular compartments, such as the mesenchymal stem cell pool, would have to be considered; that is, Notch inhibition of Runx2 function could inhibit mesenchymal stem cell commitment to the osteoblastic lineage. In opposing fashion, inhibition of Notch signaling may be a therapeutic option to investigate for the treatment of proliferative disorders of the osteoblast, such as osteosclerotic diseases or bone cancers.

From a mechanistic perspective, the function of Notch signaling in bone constitutes an example of a signaling pathway capable of regulating both osteoblastic and osteoclastic lineages. Gain of Notch function in osteoblasts affects osteoblastic differentiation in a cell-autonomous fashion, whereas loss of Notch function in osteoblasts affects osteoclastogenesis in a non–cell-autonomous manner. A remaining question is how Notch-Notch ligand interactions with neighboring cells, such as stromal and osteoclastic cells, may further modify biological function in the respective lineages. For example, ephrin B2 signaling in bone is bidirectional, with consequences for both the cells expressing the ligand and the cells expressing the receptor. In this case, reverse signaling through ephrin B2 ligand expressed by osteoclasts suppresses osteoclast precursors, whereas forward signaling through EphB4 receptor expressed by osteoblasts enhances osteoblast formation30, 31. Together, our data point to a dimorphic role for Notch signaling in osteoblast biology; that is, the stimulation of osteoblastic precursors in a pathological context and the inhibition of osteoclastogenesis in the physiological regulation of bone mass and homeostasis.

Top

Methods

Mice.

We cloned Myc-His–tagged N1ICD, which includes amino acids 1760–2556 of Notch (gift of T. Kadesch), under the control of the 2.3–kilobase (kb) osteoblast-specific Col1a1 promoter in a coat-color vector containing the tyrosinase minigene and the woodchuck posttranscriptional regulatory element (WPRE) sequences32. We generated transgenic founders by pronuclear injections using standard techniques. We maintained all transgenic lines on a FVB/N background. We identified the transgenic mice at birth by eye pigmentation and confirmed by PCR using primers specific for the WPRE. We crossed previously described Psen1f/f and Psen2-/- mice with Col1a1-Cre mice (gift of G. Karsenty) to generate osteoblast-specific Psen1-Psen2 DKO mice. These studies were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee.

Skeletal analyses, histology and bone histomorphometry.

We cleared and stained skeletons from 1-month-old mice with Alcian blue for cartilage and Alizarin red for bone as described33. We killed the mice and fixed the whole skeleton in 10% neutral-buffered formalin for 18 h. For radiographic analyses, we analyzed the skeletons by contact radiography with a Faxitron X-ray cabinet (Faxitron X-ray). We sectioned paraffin-embedded tissues to a 4–7–mum thickness and stained the section with H&E. We performed toluidine blue, Von Kossa and Goldner's stains on 5–7–mum undecalcified lumbar vertebral plastic sections by using standard protocols. We performed all static and dynamic histomorphometry analyses according to standard protocols using the OsteoMeasure histomorphometry system (Osteometrics). We performed histomorphometric analyses on 4-week-old transgenic mice (n = 3) and 6-month-old knockout mice (n = 5–7).

We analyzed mu–computed tomography (muCT) scanning of the trabecular bone of the distal femur by the muCT system (muCT-40, Scanco Medical).

Plasmids.

Osterix-luciferase was a gift of M.S. Nanes. For lentivirus vector production, we constructed plasmid pHIV-N1-IRES-eYFP by inserting a FLAG-tagged version of intracellular activated form of Notch1 (N1), just upstream of the 1.4-kb IRES-eYFP cassette (an internal ribosomal entry site fused to enhanced yellow fluorescence protein) of pHIV-IRES-eYFP34. For the primary osteoblasts, the lentiviral vectors used were self-inactivating and had the 0.5-kb mouse phosphoglycerate kinase promoter inserted upstream of either N1-IRES-eYFP or IRES-eYFP35. We produced vesicular stomatitis virus G protein–pseudotyped vector supernatants as previously described36. After 72 h, we harvested cell culture supernatants and clarified them. Typical titers after concentration by ultracentrifugation were in excess of 1 times 108 international units (IU)/ml for the two self-inactivating vectors as assessed on HOS cells by epifluorescence microscopy. Titers of unconcentrated non-SIN vectors were in excess of 1 times 107 IU/ml.

Coculture studies.

We cocultured 5 times 103 calvarial osteoblasts from 1-week-old mice (n = 3) with 5 times 104 spleen cells per well in 96-well plates for 7 d in the presence 10-8 M vitamin D3. We then stained the cells for TRAP activity and counted as described previously37. For the lentiviral rescue experiment, we cultured 5 times 103 osteoblasts isolated from calvaria of 10-d-old Psen1-Psen2 DKO mice in a 96-well plate for 2 d. We then infected the cells with either 5 mul N1ICD lentivirus or the YFP lentiviral vector supernatant for 24 h in 100 mul alpha-MEM containing 10% FBS and 8 mug polybrene/ml. We then cocultured the infected cells with 5 times 104 spleen cells from 10-d-old WT mice for 7 d in the presence of 10-8 M vitamin D3.

Fluorescence-activated cell sorting and cell sorting analyses.

After lysis of erythrocytes with ammonium chloride solution, we incubated 2 times 106 cells from bone marrow or spleen for 5 min with antibody to murine CD16/32 to block Fc receptor–mediated antibody binding, and then followed with triple staining with allophycocyanin-conjugated antibody to mouse CD11b, FITC-conjugated antibody to Gr-1 and phycoerythrin-conjugated antibody to c-Fms for 30 min. We then subjected the cells to FACS to analyze the CD11b+Gr-1-/lo cells that contain osteoclast precursors and c-Fms+ cells in both the total gated and the CD11b+Gr-1-/lo populations.

Bromodeoxyuridine incorporation.

We isolated osteoblasts from calvaria of P6 transgenic mice and wild-type littermates (n = 5 per group) as previously described33. We replated cells 48 h after the initial culture and expanded them for an additional day. We treated cells with BrdU labeling reagent (Zymed) according to the manufacturer's instructions for 6 h, washed them with PBS and fixed them with 70% ethanol for 25 min at 4 °C. Three to five areas for each genotype (n = 3 slides) were counted by two independent observers blinded to genotype. We scored BrdU-positive cells over total cells visually and with Automeasure software (Zeiss Axiovision).

Western blot analysis.

We extracted proteins from P2 mice by homogenizing the calvaria (n = 3 per group) in a buffer containing 5% SDS and 0.0625 M Tris HCl. We performed western blot analyses with antibody to p53 (gift of L. Donehower), polyclonal antibody to Runx2 (M70; Santa Cruz Biotechnology), antibody to cyclin D1 (H-2953; Santa Cruz Biotechnology) and antibody to cyclin E (ab-7959; Abcam). We normalized protein content using mouse monoclonal antibody to-gamma-tubulin (Sigma).

Glutathione-S-transferase pulldown.

We expressed GST, GST-NICDDeltaTAD, and GST-NICDDeltaRA (gift of T. Kadesch) in the BL21 strain of Escherichia coli (Stratagene). We induced production of the GST proteins were induced with 0.2 mM isopropyl beta-D-thiogalactopyranoside (IPTG; Promega) and allowed the to grow an additional 4–5 h. After induction, we lysed the cells by sonication. We bound GST proteins to glutathione resin (Amersham Biosciences). We generated 625-muM Met-labeled, FLAG-tagged Runx2 proteins by a T7 in vitro transcription-translation kit (Novagen) and incubated them with GST or GST-NICDTAD or GST-NICDRA immobilized on glutathione-Sepharose beads at 4 °C for 2 h. We then washed the beads five times with TNN buffer (1 M TrisHCl, 5 M NaCl, 14.2 M 2-mercaptoethanol) containing 1% Nonidet P-40, boiled the proteins in 2 times SDS sample loading buffer and separated them by SDS-PAGE. We performed western blotting to detect the FLAG-tagged Runx2 protein with M2 monoclonal antibody to FLAG (Sigma).

RNA extraction and quantitative reverse transcription PCR analysis.

We extracted total RNA from calvaria and forelimbs of P4 and 4-week-old mice (n = 5 and n = 3, respectively) with TRIzol reagent (Invitrogen). We synthesized cDNAs from extracted RNA with the Superscript III First Strand RT-PCR kit (Invitrogen). We performed real-time quantitative PCR amplifications in a LightCycler (Roche) using a TaqMan assay (Applied Biosystems probe HS00172878-M1). We used the genes encoding beta-actin and beta2-microglobulin as internal controls for the quantity and quality of the cDNAs in real-time PCR assays.

DNA transfection.

We transfected Cos7 and Ros17/2.8 cells with the 6XOSE2-luc reporter gene by using Lipofectamine Plus according to manufacturer's recommendations (Invitrogen). We assayed luciferase and beta-galactosidase activities 48 h after transfection. We transfected C2C12 cells with -1269/91 Osx-p-luc (gift of M.S. Nanes) by using Fugene6 according to the manufacturer's instructions (Roche). We induced the cells 24 h after the transfection with 300ng/ml recombinant human bone morphogenic protein-2 (R&D Systems), and we harvested and assayed the cells the next day. We performed all transfections in triplicate with pSV2betagal as an internal control for transfection efficiency.

Statistical analyses.

Data are expressed as mean values plusminus s.d. We computed statistical significance with Student's paired t-test. A P value of <0.05 was considered statistically significant.

Note: Supplementary information is available on the Nature Medicine website.



Top

Acknowledgments

We thank T. Kadesch (University of Pennsylvania) for Myc-His–tagged N1ICD, GST-NICDDeltaTAD and GST-NICDDeltaRA, G. Karsenty (Columbia University) for Col1a1-Cre mice, M.S. Nanes (Emory University) for osterix-luciferase and -1269/91 Osx-p-luc, and L. Donehower (Baylor College of Medicine) for antibody to p53. We thank M. Acar and O. Sirin for technical assistance. This work was supported by US National Institutes of Health grants ES11253 (B.L.), HD22657 (B.L.), DE016990 (B.L.) and AR43510 (B.F.B.).

Received 14 September 2007; Accepted 19 December 2007; Published online 24 February 2008.

Top

References

  1. Artavanis-Tsakonas, S., Rand, M.D. & Lake, R.J. Notch signaling: cell fate control and signal integration in development. Science 284, 770–776 (1999). | Article | PubMed | ISI | ChemPort |
  2. Bray, S.J. Notch signalling: a simple pathway becomes complex. Nat. Rev. Mol. Cell Biol. 7, 678–689 (2006). | Article | PubMed | ISI | ChemPort |
  3. Weinmaster, G. The ins and outs of Notch signaling. Mol. Cell. Neurosci. 9, 91–102 (1997). | Article | PubMed | ISI | ChemPort |
  4. Daudet, N. & Lewis, J. Two contrasting roles for Notch activity in chick inner ear development: specification of prosensory patches and lateral inhibition of hair-cell differentiation. Development 132, 541–551 (2005). | Article | PubMed | ChemPort |
  5. Brennan, C.A. & Moses, K. Determination of Drosophila photoreceptors: timing is everything. Cell. Mol. Life Sci. 57, 195–214 (2000). | Article | PubMed | ChemPort |
  6. Bulman, M.P. et al. Mutations in the human delta homologue, DLL3, cause axial skeletal defects in spondylocostal dysostosis. Nat. Genet. 24, 438–441 (2000). | Article | PubMed | ISI | ChemPort |
  7. Shen, J. et al. Skeletal and CNS defects in presenilin-1–deficient mice. Cell 89, 629–639 (1997). | Article | PubMed | ISI | ChemPort |
  8. Deregowski, V., Gazzerro, E., Priest, L., Rydziel, S. & Canalis, E. Notch 1 overexpression inhibits osteoblastogenesis by suppressing Wnt/beta-catenin but not bone morphogenetic protein signaling. J. Biol. Chem. 281, 6203–6210 (2006). | Article | PubMed | ChemPort |
  9. Sakamoto, K., Chao, W.S., Katsube, K. & Yamaguchi, A. Distinct roles of EGF repeats for the Notch signaling system. Exp. Cell Res. 302, 281–291 (2005). | Article | PubMed | ChemPort |
  10. Sciaudone, M., Gazzerro, E., Priest, L., Delany, A.M. & Canalis, E. Notch 1 impairs osteoblastic cell differentiation. Endocrinology 144, 5631–5639 (2003). | Article | PubMed | ChemPort |
  11. Tezuka, K. et al. Stimulation of osteoblastic cell differentiation by Notch. J. Bone Miner. Res. 17, 231–239 (2002). | Article | PubMed | ISI | ChemPort |
  12. Zamurovic, N., Cappellen, D., Rohner, D. & Susa, M. Coordinated activation of Notch, Wnt, and transforming growth factor-beta signaling pathways in bone morphogenic protein 2–induced osteogenesis. Notch target gene Hey1 inhibits mineralization and Runx2 transcriptional activity. J. Biol. Chem. 279, 37704–37715 (2004). | Article | PubMed | ChemPort |
  13. Nakashima, K. et al. The novel zinc finger–containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell 108, 17–29 (2002). | Article | PubMed | ISI | ChemPort |
  14. Galindo, M. et al. The bone-specific expression of Runx2 oscillates during the cell cycle to support a G1-related antiproliferative function in osteoblasts. J. Biol. Chem. 280, 20274–20285 (2005). | Article | PubMed | ISI | ChemPort |
  15. Pratap, J. et al. Cell growth regulatory role of Runx2 during proliferative expansion of preosteoblasts. Cancer Res. 63, 5357–5362 (2003). | PubMed | ISI | ChemPort |
  16. Shen, R. et al. Cyclin D1–cdk4 induce Runx2 ubiquitination and degradation. J. Biol. Chem. 281, 16347–16353 (2006). | Article | PubMed | ISI | ChemPort |
  17. Hill, T.P., Spater, D., Taketo, M.M., Birchmeier, W. & Hartmann, C. Canonical Wnt/beta-catenin signaling prevents osteoblasts from differentiating into chondrocytes. Dev. Cell 8, 727–738 (2005). | Article | PubMed | ISI | ChemPort |
  18. Day, T.F., Guo, X., Garrett-Beal, L. & Yang, Y. Wnt/beta-catenin signaling in mesenchymal progenitors controls osteoblast and chondrocyte differentiation during vertebrate skeletogenesis. Dev. Cell 8, 739–750 (2005). | Article | PubMed | ISI | ChemPort |
  19. Glass, D.A., II et al. Canonical Wnt signaling in differentiated osteoblasts controls osteoclast differentiation. Dev. Cell 8, 751–764 (2005). | Article | PubMed | ISI | ChemPort |
  20. Krishnan, V., Bryant, H.U. & Macdougald, O.A. Regulation of bone mass by Wnt signaling. J. Clin. Invest. 116, 1202–1209 (2006). | Article | PubMed | ChemPort |
  21. Tu, X. et al. Noncanonical Wnt signaling through G protein–linked PKC-delta activation promotes bone formation. Dev. Cell 12, 113–127 (2007). | Article | PubMed | ChemPort |
  22. Janssens, K., ten Dijke, P., Janssens, S. & Van Hul, W. Transforming growth factor-beta1 to the bone. Endocr. Rev. 26, 743–774 (2005). | Article | PubMed | ChemPort |
  23. Mezquita-Raya, P. et al. The contribution of serum osteoprotegerin to bone mass and vertebral fractures in postmenopausal women. Osteoporos. Int. 16, 1368–1374 (2005). | Article | PubMed |
  24. Fahrleitner-Pammer, A. et al. Osteoprotegerin serum levels in women: correlation with age, bone mass, bone turnover and fracture status. Wien. Klin. Wochenschr. 115, 291–297 (2003). | PubMed | ChemPort |
  25. Arko, B., Prezelj, J., Kocijancic, A., Komel, R. & Marc, J. Association of the osteoprotegerin gene polymorphisms with bone mineral density in postmenopausal women. Maturitas 51, 270–279 (2005). | Article | PubMed | ChemPort |
  26. Choi, J.Y. et al. Genetic polymorphisms of OPG, RANK, and ESR1 and bone mineral density in Korean postmenopausal women. Calcif. Tissue Int. 77, 152–159 (2005). | Article | PubMed | ChemPort |
  27. Hofbauer, L.C. et al. Estrogen stimulates gene expression and protein production of osteoprotegerin in human osteoblastic cells. Endocrinology 140, 4367–4370 (1999). | Article | PubMed | ChemPort |
  28. Bucay, N. et al. Osteoprotegerin-deficient mice develop early onset osteoporosis and arterial calcification. Genes Dev. 12, 1260–1268 (1998). | Article | PubMed | ISI | ChemPort |
  29. Wang, L.L. Biology of osteogenic sarcoma. Cancer J. 11, 294–305 (2005). | PubMed | ChemPort |
  30. Boyce, B.F. & Xing, L. Osteoclasts, no longer osteoblast slaves. Nat. Med. 12, 1356–1358 (2006). | Article | PubMed | ChemPort |
  31. Zhao, C. et al. Bidirectional ephrinB2-EphB4 signaling controls bone homeostasis. Cell Metab. 4, 111–121 (2006). | Article | PubMed | ChemPort |
  32. Zhou, G. et al. Dominance of SOX9 function over RUNX2 during skeletogenesis. Proc. Natl. Acad. Sci. USA 103, 19004–19009 (2006). | Article | PubMed | ChemPort |
  33. Ducy, P. et al. A Cbfa1-dependent genetic pathway controls bone formation beyond embryonic development. Genes Dev. 13, 1025–1036 (1999). | Article | PubMed | ISI | ChemPort |
  34. Segall, H.I., Yoo, E. & Sutton, R.E. Characterization and detection of artificial replication-competent lentivirus of altered host range. Mol. Ther. 8, 118–129 (2003). | Article | PubMed | ISI | ChemPort |
  35. Dai, C., McAninch, R.E. & Sutton, R.E. Identification of synthetic endothelial cell–specific promoters by use of a high-throughput screen. J. Virol. 78, 6209–6221 (2004). | Article | PubMed | ChemPort |
  36. Sutton, R.E., Wu, H.T., Rigg, R., Bohnlein, E. & Brown, P.O. Human immunodeficiency virus type 1 vectors efficiently transduce human hematopoietic stem cells. J. Virol. 72, 5781–5788 (1998). | PubMed | ISI | ChemPort |
  37. Xing, L. et al. NF-kappaB p50 and p52 expression is not required for RANK-expressing osteoclast progenitor formation but is essential for RANK- and cytokine-mediated osteoclastogenesis. J. Bone Miner. Res. 17, 1200–1210 (2002). | Article | PubMed | ISI | ChemPort |
  1. Department of Molecular and Human Genetics, Baylor College of Medicine, One Baylor Plaza, Houston, Texas 7703, USA
  2. Department of Pathology and Laboratory Medicine, University of Rochester Medical Center, 601 Elmwood Avenue, Rochester, New York 14642, USA
  3. Howard Hughes Medical Institute, One Baylor Plaza, Houston, Texas 77030, USA
  4. Department of Pediatrics, Baylor College of Medicine, One Baylor Plaza, Houston, Texas 77030, USA
  5. Department of Molecular Virology and Microbiology, Baylor College of Medicine, One Baylor Plaza, Houston, Texas 77030, USA
  6. These authors contributed equally to this work.

Correspondence to: Brendan Lee1,3 e-mail: blee@bcm.tmc.edu