Muons penetrate much further than X-rays, they do essentially zero damage, and they are provided for free by the cosmos.
Image credit: Science Source/Digital Globe.
Contributed by Maria I. New, February 29, 2016 (sent for review December 10, 2015; reviewed by Di Chen, Christopher Huang, and Roberto Pacifici)
Significance
Notch is a critical regulator of skeletal development, but its role in remodeling of the adult skeleton is unclear. Using genetically modified mice, we show that Notch stimulates skeletal mineralization by bone-building osteoblasts. Thus, overexpression of the Notch intracellular domain in mice results in an increase in bone mass, prevents bone loss following ovariectomy and during aging, and promotes fracture healing. Notch is therefore a potential therapeutic target for conditions of bone loss, including osteoporosis.
Abstract
Notch controls skeletogenesis, but its role in the remodeling of adult bone remains conflicting. In mature mice, the skeleton can become osteopenic or osteosclerotic depending on the time point at which Notch is activated or inactivated. Using adult EGFP reporter mice, we find that Notch expression is localized to osteocytes embedded within bone matrix. Conditional activation of Notch signaling in osteocytes triggers profound bone formation, mainly due to increased mineralization, which rescues both age-associated and ovariectomy-induced bone loss and promotes bone healing following osteotomy. In parallel, mice rendered haploinsufficient in γ-secretase presenilin-1 (Psen1), which inhibits downstream Notch activation, display almost-absent terminal osteoblast differentiation. Consistent with this finding, pharmacologic or genetic disruption of Notch or its ligand Jagged1 inhibits mineralization. We suggest that stimulation of Notch signaling in osteocytes initiates a profound, therapeutically relevant, anabolic response.
Notch is a transmembrane receptor protein that upon ligand binding is cleaved by γ-secretases, including presenilin-1/2 (Psen1/2), to yield an active Notch intracellular domain (Nicd) that undergoes nuclear translocation (1). Within the nucleus, Nicd interacts with the DNA binding protein, recombination signal-binding protein for Ig κ J region (Rbpj), to regulate the expression of target genes. This canonical Notch signaling pathway includes the receptors Notch1–4, their ligands Jagged1/2 and Dll1-3, the primary nuclear binding protein Rbpj, and downstream genes, such as Hes1/5 (2). Global deletion of any one or more of these genes renders mice embryonically lethal, likely due to deficiencies in somite maturation (3⇓⇓⇓⇓⇓–9). Analysis of these embryos reveals defects in axial skeletal development, which is broadly recapitulated in descriptions of people harboring loss-of-function mutations in Notch1/2 or Dll3 (8, 10⇓–12).
Though studies on the global deletion of Notch components have been difficult to interpret, there are seemingly conflicting reports on the skeletal phenotype of mice in which Notch1/2 or Psen1/2 is deleted, even in a cell-specific manner. Early deletion of Notch1, for example, using Prx1-Cre mice, causes postnatal lethality and radiodense bones (13). In contrast, late-stage Notch1 deletion using a Col2.3-Cre line yields an osteoporotic phenotype (14). The effects of Nicd overexpression also yield opposing phenotypes depending upon the promoter used. Osteoporosis results when Nicd overexpression is driven by the Col3.6 promoter (15), whereas mice are rendered severely osteosclerotic when either Col2.3 or Dmp1 promoters are used (16, 17).
Thus, it has been difficult to separate known early effects of Notch on skeletogenesis and postnatal modeling from potential therapeutically relevant effects on the mature skeleton, even in mutants that survive. In addition, gain-of-function studies have been fraught with interpretational dilemmas due to the apparent cell- and stage-specific function of Notch. Here, we show that cell-specific Notch induction in adult mice triggers a skeletal anabolic response. This bone-forming action, which we find is driven predominantly by increased mineralization, overcomes both age-related and ovariectomy-induced bone loss, and promotes bone healing in an osteotomy model; this prompts the potential for exploiting the Notch pathway to a therapeutic advantage.
Results
Notch Expression in Osteocytes and New Bone Formation.
It is well recognized that Notch is critical for skeletogenesis (18⇓–20), but its expression profile and physiological function in adult bone is not established. We therefore first sought to investigate the expression of activated Notch in trabecular and cortical bone of adult mice using a Notch reporter mouse, cp-EGFP [transgenic Notch reporter (TNR)]. TNR mice, initially developed for studying Notch expression in neural (21) and hematopoietic stem cells (22), respond to the intranuclear accumulation of Nicd upon activation. Thus, cellular fluorescence is noted only when Notch is activated.
Frozen sections of trabecular (femur metaphysis and spine), membranous (calvaria), or cortical (femur diaphysis) bone showed that cp-EGFP is localized to osteoblasts and osteocytes (Fig. 1 A–D). Notably, whereas cp-EGFP+ osteocytes were embedded within the bone matrix (red arrow), surface cells were spindle-shaped, reminiscent of an osteoblast-to-osteocyte transition phenotype (white arrow). Single labeling with Xylenol orange of nondecalcified femur showed that cp-EGFP+ cells lay in the vicinity of and, in fact, at points overlapped with sites of new bone formation (Fig. 1A). cp-EGFP+ cells were not detected in a range of mouse tissues, including liver, spleen, lung, kidney, heart, vessel, muscle, and skin, with minimal expression noted in adult brain (Fig. S1). This finding suggests physiologically specific localization of Notch signaling to bone and brain.
Fig. 1.
Osteocytic expression of active Notch. Representative fluorescence micrographs showing EGFP expression (green) in trabecular [femur metaphysis (A) (40× magnification) and spine (B) (20× magnification)], membranous [calvarial sutures (C) (20× magnification)], and cortical [femur diaphysis (D) (40× magnification)] bone in adult, 2-mo-old cp-EGFP TNR mice. Red and blue staining, respectively, show Xylenol orange (XO) labeling, indicating areas of new bone formation, and DAPI staining of cell nuclei. Note the close proximity of EGFP expression and XO staining (A). n = 6 (male and female, aged 2–4 mo). (E) BMSC cultures (day 14–16; 20× magnification) from TNR mice (n = 4 per culture), showing EGFP+ cells (Left, green) localized to mineralizing nodules (Center, black), with overlapping images shown (Right, triplicates/time point, five repeats). Expression (quantitative PCR) of Notch1-4 (F), the ligands Jagged1/2 (Jag1/2) and Dll1/3/4 (G), and downstream target genes, Hes1/5 (H) in BMSC cultures from C57BL/6J mice, as a function of days in differentiating medium. n = 5, triplicates per time points. ***P < 0.001, **P < 0.01, *P < 0.05, compared with day 8 (one-way ANOVA with Bonferroni post hoc test). All experiments were performed at least two times.
Fig. S1.
High Notch expression mainly in bone tissue. Representative fluorescence and corresponding bright light micrographs showing strong EGFP expression (green) in bone, weak expression in the brain, and virtually no expression in the liver, spleen, lung, kidney, heart, blood vessels, muscle, and skin of adult (2-mo-old) cp-EGFP TNR mice. Blue, DAPI staining of cell nuclei.
We sought further to explore the temporal expression of activated Notch as a function of osteoblast differentiation in bone marrow stromal cell cultures (BMSCs). cp-EGFP+ cells were not detected during early osteoblast differentiation (days 4 and 7) by fluorescent microscopy. However, strongly EGFP+ cells resembling osteocytes were found embedded in mineralized areas at ∼day 10 of culture (Fig. S2). After day 16, upon the completion of mineralization, active cp-EGFP+ cells were lost. This narrow window of Notch activity in adult bone specifically points to its putative role in regulating mineralization.
Fig. S2.
Notch expression as a function of mineralization. Representative fluorescence micrographs showing EGFP expression (green) in differentiating cultures (day 10–22) of BMSCs isolated from adult (2-mo-old) cp-EGFP TNR mice. Red and blue staining, respectively, show XO labeling, indicating mineralization, and DAPI staining of cell nuclei.
The presence of active Notch in vivo would require the presence not only of the Notch receptor, but also of ligand and target gene(s). We therefore examined the expression (quantitative PCR) in bone marrow cultures of the Notch1–4, the Notch ligands Jag1/2 and Dll1/3/4, and the downstream target genes Hes1/5. Notch1 and 4 showed a progressive increase in expression over time (Fig. 1F). Increasing expression was also noted for Jag1 and Hes1, but not for Jag2, Dll1/3/4, or Hes5 (Fig. 1 G and H). These results are consistent with a function of Notch in the mineralization of adult bone.
Notch Signaling Is Required for Mineralization.
To determine whether Notch signaling is required for mineralization, we used four complementary approaches. We first examined mineralization by isolated BMSCs in the presence of a γ-secretase inhibitor DAPT (N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester); the latter is known to block Notch endoproteolysis. Cells were cultured in the presence of β-glycerolphosphate and stained for cfu-fibroblastoid (cfu-f; alkaline phosphatase labeled) at day 10, or cfu-osteoblastoid (cfu-ob; von Kossa labeled) at days 14, 17, and 21. DAPT strongly inhibited cfu-ob, but not cfu-f formation, consistent with a late effect of Notch signaling inhibition on mineralization (Fig. 2 A and B); this was associated with markedly reduced expression of mineralization-associated genes—namely, matrix extracellular phosphoglycoprotein (Mepe), dentin matrix protein-1 (Dmp1), sclerostin (Sost), and phosphate regulating endopeptidase homolog X-linked (Phex). Importantly, gene expression at earlier time points remained unaffected (Fig. 2C).
Fig. 2.
Pharmacologic and genetic evidence for a role for Notch in mineralization. Representative images (A) and colony area (B) showing the effect of the γ-secretase inhibitor DAPT (or DMSO, vehicle) on alkaline phosphatase-labeled cfu-f or von Kossa-labeled cfu-ob in BMSC cultures undergoing differentiation. RNA extracted from BMSC cultures treated with DAPT or DMSO underwent quantitative PCR for markers of osteocytes and mineralization, including Mepe, Dmp1, Sost, and Phex. Statistics: unpaired Student’s t test, comparisons between DAPT and DMSO; P values shown (B), n = 3 wells per group; in quadruplicate (C). (D) Dramatic reduction of mineralization [phase contrast images, Upper; Xylenol orange (XO) labeling, Lower] when BMSC cultures from Jag1fl/fl mice were infected with Ad-Cre to delete exon 4 of the Jag1 gene under differentiating conditions. n = 3 wells/group in quadruplicate. (E) Similarly dramatic reduction in von Kossa-labeled cfu-ob colonies in cultures from Mx1-Cre+:Jag1fl/fl mice, shown both as representative plates and as staining area. n = 3 wells per group, in quadruplicate. (F) Profound reduction in cfu-ob with modest decrements in early cfu-f colonies in BMSCs from Psen1 haploinsufficient (Psen1+/−) mice. n = 3 wells per group. ***P < 0.0001, **P < 0.001, compared with day 8 (one-way ANOVA with Sidak’s correction for multiple comparisons). All experiments were performed at least two times.
Second, we studied the mineralizing activity of BMSCs in which the Notch ligand, Jag1, was conditionally deleted using Ad-Cre. A profound reduction in mineralization was visualized using phase contrast microscopy and by Xylenol orange staining (Fig. 2D). In a third set of experiments, we cultured BMSCs isolated from mice in which Jag1 was deleted conditionally in cells of the hematopoietic and mesenchymal lineages using an Mx1-Cre promoter and polyI:polyC. Again, a gross reduction in mineralization was noted on von Kossa staining (Fig. 2E). Finally, we cultured bone marrow stromal cells from mice haploinsufficient for Psen1 (23). Expectedly, there was a profound reduction in cfu-ob formation; however, there was also an unexplained decrease in cfu-f, likely arising from alternative actions of Psen1 (Fig. 2F). These data together confirm that the activation of Notch is a requirement for mineral deposition.
To explain the cellular basis for the reduced mineralization in Psen1+/− mice, we tracked cells of the osteoblast lineage using reporter mice that expressed GFP during early and late osteoblast progression. We crossed Psen1+/− mice with Col3.6-GFPtpz or Col2.3-EGFP mice (24), respectively, and lineage tracked GFP+ cells in adult mice in vivo, as well as ex vivo in BMSC cultures at 15 and 21 d. Previous studies with Col3.6-GFPtpz mice have shown that the 3.6 kb Col1a1 promoter is extensively expressed, in addition to skin, in osteoblasts lining the periosteal and endosteal trabecular surfaces, with weak expression in periosteal spindle-shaped preosteoblasts (24). In contrast, the 2.3-kb Col1a1 promoter marks mature osteoblasts and osteocytes in Col2.3EGFP mice (24). In differentiating bone marrow stromal cell cultures, GFP expression is seen as early as days 7 and 14, respectively, for Col3.6-GFPtpz and Col2.3EGFP mice (24) (Fig. 3A).
Fig. 3.
Lineage tracing osteoblasts in Psen1 haploinsufficient mice reveals a block in differentiation. A schematic representation of expression patterns for Col3.6-GFPtpz and Col2.3-EGFP (A). Tracking EGFP+ osteoblasts in Col3.6-GFPtpz (early expression) or Col2.3-EGFP (late expression) mice on a wild-type or Psen1+/− background, both in vivo in the cortical and trabecular compartments of the femur (B) (female, Col3.6-GFPtpz aged 3 mo and Col2.3-EGFP aged 6 mo), as well as in vitro in BMSC cultures (C). Representative frozen sections and wells are shown: (B) 20× magnification; box showing magnified area 40×; (C) 20× magnification (n = 3–5 per group, aged 2–4 mo). All experiments were performed at least three times.
We found a notable abundance of GFP+ cells in frozen sections (Fig. 3B) and BMSC cultures (Fig. 3C) from Col3.6-GFPtpz:Psen1+/− mice compared with Col3.6-GFPtpz:Psen1+/+ littermates. However, there was a marked reduction in GFP+ cells in Col2.3-EGFP:Psen1+/− mice compared with Col2.3-EGFP:Psen1+/+ mice (Fig. 3 B and C). This suggests that haploinsufficiency of Psen1, a surrogate for Notch signaling, blocked lineage progression to mature mineralizing osteoblasts.
Controlled Notch Overexpression in Osteoblast Lineage Cells Stimulates Bone Formation.
To test the direct effects of Notch activation on bone formation and bone mass, we characterized the skeletons of mice overexpressing Nicd specifically in cells of the osteoblast lineage. Because Notch is expressed in osteocytes (Fig. 1), we initially crossed Dmp1-Cre mice with Rosa-NicdTg/Tg mice (25). Nicd overexpression in Dmp1-Cre+:Rosa-NicdTg/Tg mice led to postnatal death at ∼20 d. H&E staining of femur epiphyseal sections showed significantly increased bone volume, with infiltration and ablation of the bone marrow cavities (Fig. S3). To circumvent osteosclerosis resulting from the noninducible, persistent expression of Nicd by Dmp1-Cre, we selected Col3.2-CreERT2 to achieve a spatiotemporal control of Nicd activity in mature osteoblasts and osteocytes in bone, as previously reported (26⇓⇓–29). Of note is that the expression of the 3.2-kb Col1a1 promoter is restricted to mature osteoblasts and osteocytes similarly to the 2.3-kb Col1a1 promoter, but is distinct from that of the Col3.6 promoter. Notably, the 3.2-kb promoter is not active in spindle-shaped periosteal preosteoblasts, where the 3.6-kb promoter is predominantly expressed (24). As such, our premise is that expression of Nicd using the inducible 3.2-kb promoter will closely mimic any effects driven by the 2.3-kb promoter (26, 29).
Fig. S3.
Severe osteosclerosis in Dmp1-driven Notch overexpression. Representative micrographs displaying severe osteosclerosis and bone marrow infiltration in the femur diaphysis (also showing marrow cavity) (Upper) and trabecular bone of L4 vertebrae (Lower) from 1-mo-old Dmp1-Cre+:Rosa-NicdTg/Tg mice. Dmp1-Cre-:Rosa-NicdTg/Tg littermates served as controls.
In separate experiments, we found that Ad-Cre-mediated Nicd overexpression in BMSCs caused a profound increase in mineralization (Fig. S4). The direct effect of Nicd activity on mineralization was confirmed in tamoxifen-induced BMSCs from Col3.2-CreERT2:Rosa-NicdTg/Tg mice (Fig. 4A). Using Col3.2-CreERT2:Rosa-red/EGFP reporter mice, we confirmed tamoxifen-induced EGFP expression selectively and efficiently in osteocytes and osteoblasts (Fig. 4B); its expression pattern is known to be similar to that of Col2.3-EGFP and is different from Col3.6-GFPtpz, notably without detectable EGFP expression in periosteal preosteoblasts (24). However, low levels of EGFP expression were noted in noninduced, corn oil-treated mice (Fig. S5A). Microstructural bone and osteoid parameters on µCT or von Kossa staining remained unaffected despite the leaky EGFP expression (Fig. S5 B and C), thus validating noninduced, corn oil-treated Col3.2-CreERT2:Rosa-NicdTg/Tg mice as controls.
Fig. 4.
Conditional activation of Notch in osteoblasts of adult mice induces new bone formation. (A) Representative plates showing increased von Kossa-labeled mineralization in response to Tmx treatment of BMSC cultures from Col3.2-CreERT2:Rosa-NicdTg/Tg mice (n = 4 triplicate wells per group). (B) Fluorescence micrographs of frozen section of femur from reporter Col3.2-CreERT2:Rosa-red/EGFP mice to demonstrate EGFP expression in response to Cre activation by Tmx (Left, 5× magnification; Right, 20×). Note that mice treated with vehicle (oil) show minimal EGFP leakage (white arrow, also see Fig. S5A) (n = 3–6 per group, male, aged 2–3 mo). Effects of Notch activation by Tmx in 16-wk-old Col3.2-CreERT2:Rosa-NicdTg/Tg mice. The femur (trabecular and cortical bone) was evaluated by µCT (C), H&E staining (D, 5× and 20× magnification), or von Kossa labeling (E, 5× magnification). Note the substantial increase in bone mass. Microstructural parameters include bone volume/total volume (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), osteoid surface/bone surface (OS/BS), osteoid volume/bone volume (OV/BV), and osteoid thickness (O.Th). Shown also are plasma osteocalcin levels (G). Dynamic histomorphometry in calcein/Xylenol orange dual-labeled mice displaying the dramatic effect of Notch induction on mineralizing surface (MS), MAR, and BFR (F) (units as shown). Representative images are also shown. Measures of bone strength, including elastic modulus and maximum bending load, on a three-point bending test, are shown (H). (I) Toluidine blue and H&E staining of femoral metaphysis and cortex, respectively, showing osteoblasts (arrows, 40× magnification). Unpaired Student’s t test, comparisons between oil- and Tmx-treated mice; P values shown, n = 5 mice per group, male. Results are expressed as mean ± SEM.
Fig. S4.
Induction of Notch in BMSC cultures dramatically enhances mineralization. Schematic representation of the strategy used to activate transcription of the Nicd domain in vitro by removing the stop codon using Ad-Cre; this leads to a profound increase in mineralization in differentiating BMSC cultures, displayed as phase contrast images. Ad-EGFP was used as control.
Fig. S5.
Leaky Notch activation due to Col3.2-CreERT expression does not significantly affect phenotype. (A) EGFP expression (green) reflecting leakiness in the femur epiphysis of Col3.2-CreERT2:Rosa-red/EGFP mice treated with vehicle (corn oil). Red staining, ubiquitin expression. µCT (B) or von Kossa staining (C) and corresponding microstructural parameters (as in Fig. 4), including BV/TV, Tb.N, Tb.Th, BS/BV, OS/BS, OV/BV, and O.Th (B and C). No significant changes on unpaired Student’s t testing; n = 5 mice per group.
µCT of trabecular and cortical bone from adult homozygotic Col3.2-CreERT2:Rosa-NicdTg/Tg showed a dramatic increase in bone mass. There was a significant increase in bone volume/total volume (BV/TV) and trabecular number (Tb.N) in tamoxifen-induced mice compared with corn oil-treated controls (Fig. 4C). H&E staining showed marked increases in trabecular bone with infiltration of bone marrow (Fig. 4D). Von Kossa staining further showed not only increases in BV/TV, but also reduced osteoid parameters, including osteoid surface/bone surface (OS/BS), osteoid volume/bone volume (OV/BV), and osteoid thickness (O.Th), consistent with increased mineralization (Fig. 4E). Dynamic histomorphometry showed profound increases in mineral apposition rate (MAR) and bone formation rate (BFR; Fig. 4F), indicative of overall increases in bone formation. The enhanced bone formation, confirmed by increased serum osteocalcin levels (Fig. 4G), together with disproportionately high hypermineralization, likely resulted in markedly improved bone strength—namely, elastic modulus and maximum bending on a three-point bending test (Fig. 4H). Overall, therefore, selective activation of Notch signaling in mature osteoblasts and osteocytes stimulates new bone synthesis and increases bone strength. An almost-identical phenotype was evident in heterozygotic Col3.2-CreERT2:Rosa-NicdTg/+ mice (Fig. S6).
Fig. S6.
Active Notch in heterozygote transgenics increases bone mass. µCT analysis of bone from the metaphyseal and diaphyseal compartments of femur in heterozygotic Col3.2-CreERT2:Rosa-NicdTg/+ mice, showing increased bone mass. Microstructural parameters (as in Fig. 4) include BV/TV, Tb.N, Tb.Th, and Tb.Sp. Statistics by unpaired Student’s t test; P values as shown; n = 6 mice per group.
At the cellular level, we examined osteoblast and osteoclast numbers through toluidine blue and tartrate-resistant acid phosphatase (Trap) staining, respectively. Osteoblast numbers were expectedly increased in Col3.2-CreERT2:Rosa-NicdTg/Tg mice (Fig. 4I). There was also an increase in parameters of bone resorption, including Oc.N/BS and Oc.S/BS, as well as serum Trap5b levels, likely arising from retained osteoblast–osteoclast coupling (Fig. S7 A and B). This increase in bone resorption is consistent with an increase in osteoclast formation in response to low subnanomolar concentrations of Jag1 added to BMSCs. Osteoclast formation was inhibited at higher nanomolar concentrations (Fig. S7C).
Fig. S7.
Notch signaling activates osteoclastic bone resorption. Histomorphometric parameters of bone resorption, quantitated as osteoclast surface/bone surface (Oc.S/BS), osteoclast number/bone surface (Oc.N/BS), and osteoclast number/total volume (Oc.N/TV), in bones of Col3.2-CreERT2:Rosa-NicdTg/Tg mice with oil or Tmx (A). (B) Serum Trap5b levels. Statistics by unpaired Student’s t test; P values as shown; n = 5 mice per group. (C) Effect of the Notch ligand Jag1 on Trap-positive osteoclast formation in bone marrow hematopoietic cell cultures treated with macrophage colony-stimulating factor (Mcsf) and receptor activator of nuclear factor kappa-B ligand (Rankl) for 5 d. Statistics by unpaired Student’s t test corrected for multiple comparisons by Bonferroni; *P < 0.05, **P < 0.01.
Biologically Relevant Effects of Notch Signaling in Osteocytes.
To establish Notch as a viable target for the modulation of adult bone remodeling, we injected recombinant Jag1 directly onto calvaria of 3-mo-old adult C57BL/6J mice. There was a profound anabolic effect, with a significant increase in new bone formation assessed in H&E stain (Fig. 5A).
Fig. 5.
Conditional Notch activation in adult mice rescues bone loss and promotes bone healing. (A) Anabolic action of recombinant Jag1 injected onto calvarial bone of adult mice (40 µg/kg, 20× magnification), n = 5 per group, male B6, aged 2 mo. Evaluation of spine by µCT (B) and femur metaphysis by von Kossa staining (C, 5× magnification) following the induction of Notch expression in 8-mo-old Col3.2-CreERT2:Rosa-NicdTg/+ mice by Tmx injections. Microstructural parameters as in Fig. 4. (D) Complementary dynamic histomorphometry of calcein/XO dual-labeled mice displaying the effect of Notch induction on MS, MAR, and BFR (units as shown). Representative images are also shown (40× magnification). (E) Von Kossa staining (5× magnification) and microstructural parameters, including osteoid parameters, of femur metaphysis following Tmx injection in ovariectomized Col3.2-CreERT2:Rosa-NicdTg/Tg mice. Unpaired Student’s t test, comparisons between oil- and Tmx-treated mice; P values shown (B), n = 4–5 mice/group. (F) Plain radiograph (In-Vivo Xtreme) showing the effect Notch activation by Tmx on the healing of an osteotomy defect induced in femur diaphysis, n = 5 per group.
To complement a skeletal anabolic response genetically in relevant biological models, we examined the effects of Nicd overexpression in aging and ovariectomized mice, as well as in an osteotomy model. Induction of Nicd by tamoxifen in Col3.2-CreERT2:Rosa-NicdTg/+ mice at 8 mo of age, at which time bone mass is known to be reduced, resulted in osteoprotection. Notably, µCT of spinal trabecular bone showed marked increases in BV/TV and Tb.N, with a corresponding decrease in Tb.Sp (Fig. 5B). Von Kossa staining of femur metaphyseal (trabecular) bone not only confirmed an increase in BV/TV, but also showed diminished OV/BV and O.Th, consistent with increased mineralization (Fig. 5C); this was accompanied by an overall increase in bone formation parameters—namely, mineralizing surface (MS) and BFR (Fig. 5D). Additionally, there was dramatic thickening of the cortical component of the femur (Fig. 5C).
In separate experiments, 3-mo-old Col3.2-CreERT2:Rosa-NicdTg/Tg mice were ovariectomized and immediately injected with either corn oil or tamoxifen. Expectedly, there was a dramatic loss of bone in corn oil-treated control mice 8 wk following surgery (Fig. 5E). However, there was an obvious and significant increase in BV/TV and Tb.Th in tamoxifen-induced mice (Fig. 5E), which was accompanied by a dramatic reduction in osteoid parameters, confirming that Notch signaling protected against bone loss following ovariectomy.
Finally, we assessed the effect of Notch activation in adult mice on fracture healing using a femur osteotomy model. For this, a 3-mm defect was created surgically, 3 d after which the mice were given corn oil or tamoxifen for 4 wk. Plain radiography [In-Vivo Xtreme (Bruker)] showed that there was complete restoration of the surgical defect in tamoxifen-induced Col3.2-CreERT2:Rosa-NicdTg/Tg mice, whereas mice receiving oil continued to display nonunion (Fig. 5F).
Discussion
We report the effects of active Notch signaling on adult bone. Hitherto, Notch has been considered a key developmental gene that coordinates a variety of transcriptional events to regulate somitogenesis and stem cell fate (30⇓⇓⇓⇓⇓–36). In the skeletal context, studies have focused primarily on bone development in both mice and people. Mutations of Notch components are known to cause defects in axial skeletal development. For example, spondylocostal dysostosis results from mutations in human DLL3 (10, 11, 37, 38), a Notch ligand, whereas NOTCH2 mutations alone cause Hagdu–Cheney syndrome, a rare disease characterized by facial abnormalities, acro-osteolysis and, interestingly, osteoporosis (39, 40). Mutations of JAG1 and NOTCH2 in concert cause Alagille syndrome, marked by impaired craniofacial development due to somite segmentation defects.
However, mice lacking individual Notch receptors are invariably lethal in utero, likely due to an early deficiency in somite segmentation (7, 9⇓–11, 41⇓⇓⇓⇓⇓–47). Nonetheless, double-mutant mice, wherein Prx1-Cre-mediated deletion of Psen1 is restricted to the mesenchyme on a Psen2−/− background, survive for up to 10 wk and display radiodense bones (13). This phenotype is recapitulated in mice in which both Notch1 and Notch2 are deleted from the early mesenchyme, again using Prx1-Cre (13). However, deletion of Psen1 solely in osteoblast-lineage cells using a Col2.3-Cre line yields an osteoporotic phenotype (14). The opposing skeletal phenotypes of mice in which Notch components are deleted using Prx1- or Col2.3-Cre lines suggests that Notch effects are cell specific and dependent on the stage of development.
Gain-of-function studies also yield opposing phenotypes depending on promoter used to drive Notch overexpression. Mice transgenically expressing Nicd under Col2.3 or Dmp1 promoters, which allow restricted expression in osteoblast-lineage cells and osteocytes, respectively, display osteosclerosis with bone marrow infiltration at 4 wk followed by early postnatal death (17). These findings support the possibility that Notch signaling has an anabolic action on adult bone. In contrast, early osteoblast lineage overexpression of Nicd using Col3.6-Cre mice suppresses progression of the osteoblast to a more mature, mineralizing phenotype, yielding an osteopenic phenotype (15, 48). These conflicting phenotypes seem to arise from both cell- and stage-dependence of Notch effects, and call for a definitive phenotypic characterization of mice in which Nicd is overexpressed solely during adulthood.
Because Col3.6-Cre is noninducible, and expressed in bone, teeth, and other nonbone tissues, such as skin, brain, kidney, liver, and lung (24, 49), we used Col3.2-CreER in which Cre is inducible and expressed in a similar pattern to Col2.3-Cre (24)—namely, its expression is limited in bone and teeth with a lower expression in tendon, but without detection of GFP expression in other tissues, such as skin, brain, kidney, liver, and lung) (26, 29). Within bone, Col3.6 is expressed in precursor cells in the suture mesenchyme, as well as in preosteoblasts and the fibrous layers of the inner and outer periosteum and their downstream progeny (24). In contrast, Col2.3 and Col3.2 are identified specifically in osteoblasts and osteocytes (26⇓⇓–29). We surmise that the reported osteoporotic phenotype of Col3.6-Cre:Rosa-NicdTg/Tg may be caused by early and constant Nicd activation that likely interrupts physiological lineage transition from the preosteoblast into mature osteoblasts and osteocytes (15). In addition, Col3.6-GFP has been reported to be expressed in osteoclasts (50), making it further difficult to interpret outcomes. In contrast, we report that Col3.2-CreERT:Rosa-NicdTg/Tg mice that are induced to overexpress Nicd in osteoblast lineage cells during adulthood display a dramatic anabolic bone phenotype. Arising from increased osteoblastic bone formation, prominently as a result of advanced mineralization, this anabolic phenotype overcomes age-associated osteopenia and prevents postovariectomy-induced bone loss. These latter studies suggest that Notch, if targeted during adult life, could offer potential opportunities for new therapies. To this end, we further show that induction of Nicd in osteoblast lineage cells can result in the rapid healing of surgical defects in an osteotomy model.
Another issue unresolved from previous studies relates to the cell type that Notch primarily regulates in adult bone. We find that the actions of Notch in osteoblast lineage cells are primarily osteocyte driven. Several findings underscore this premise. First, our analysis of Notch-EGFP reporter mice documents EGFP+ cells within bone matrix; osteocytes are the only cells known to be embedded in bone. Second, mice in which Nicd is overexpressed using the Dmp1 promoter, known to be relatively specific for osteocytes, demonstrate profound osteosclerosis and postnatal lethality. Third, the osteocyte-mediated function of Notch is consistent with the predominant mineralization defect noted as reduced von Kossa staining and osteocytic gene expression in DAPT-treated BMSC cultures and absent mineralization in cultures in which Jag1 is deleted in vitro using Ad-Cre or ex vivo in Mx1-Cre+:Jag1fl/fl mice. Fourth, by lineage tracing using two GFP reporter mice, we provide direct evidence that Psen1 haploinsufficiency almost completely blocks the transition of osteoblast precursors to a more mature, mineralizing phenotype, with the consequent accumulation of precursor cells. This latter finding also explains osteoporotic phenotype of Psen1/2-deficient mice noted by others (13, 51). With that said, we are by no means excluding other actions of Notch signaling during earlier stages of osteoblast development, although we believe that the predominant effects occur late during mineralization.
Our complementary approaches that collectively demonstrate direct anabolic effects of Notch on adult bone raise the question of whether Notch or its components are drug-able targets for osteoporosis. More than 100 million men and women suffer from osteoporosis worldwide, and many of these patients have low-turnover bone loss that could be best treated by an anabolic agent (52). Nonetheless, the armamentarium of anabolic therapies is restricted to just recombinant human parathyroid hormone. Targeting the osteocyte toward an anabolic response may provide a viable alternative, which could be achieved either by stimulating the Wnt pathway, an approach that is currently underway (53⇓⇓–56), or, as we suggest, by stimulating Notch signaling.
However, the extent to which any bone-forming therapy would have a net positive effect on bone mass in people would depend on whether it also stimulates bone resorption to counter its anabolic action. We find that, despite the stimulation of osteoclastic bone resorption, Notch overexpression, at least in mice, yields a profound net anabolic phenotype, which is in contrast to the stimulation of Wnt signaling, wherein the anabolic action of an anti-sclerostin antibody is paralleled by an inhibition, rather than stimulation, of bone resorption (20, 56⇓–58). Though this finding may suggest that Wnt stimulation may be efficacious with a greater net effect on bone mass, a key advantage of targeting Notch over the Wnt pathway resides in the relative selectivity of its expression in bone, with limited expression in the adult brain. Wnt, however, is significantly more ubiquitously distributed and has a role in oncogenesis (59). Therefore, expression of Notch in the brain may necessitate a fuller evaluation, because mutations in presenilins and γ-secretases are known to predispose to Alzheimer’s disease.
Methods
Mice.
All mouse experiments were performed per approved protocols by the Institutional Animal Care and Use Committees at Sichuan University and Icahn Medical School at Mount Sinai. Three mouse lines were purchased from JAX, including the TNR line [Tg(Cp-EGFP)25Gaia/ReyaJ; 018322]; conditional Nicd transgenic line [Gt(ROSA)26Sortm1(Notch1)Dam/J; 008159]; Rosa-ACTB-tdTomato,-EGFP [Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J; 007576]; and Mx1-Cre transgenic lines [Tg(Mx1-Cre)1Cgn/J; 003556]. The Col3.6-GFPtpz and Col2.3-EGFP lines were obtained from D.W.R.’s laboratory (24). Col3.2-CreERT2 mice were kindly provided by Jerry Feng, Texas A&M University, Dallas and Henry M. Kronenberg, Harvard University, Boston (27). The 3.6-kb collagen 1a1 proximal promoter (Col3.6) is active during early osteoblast differentiation, whereas the shorter 2.3-kb and 3.2-kb promoters (Col2.3 and Col3.2) are active during late osteoblast differentiation. Jag1flox/flox [B6;129S-Jag1tm2Grid/J; 010618] and Notch transgenic mice [C57BL/6J-Tg(ACTB-NOTCH1)1Shn/J; 006481] mice were kindly supplied by Thomas Gridley (JAX). Psen1 mutant mice were provided by Jorge Busciglio (60). For conditional Notch activation, male Col3.2-CreERT2 mice were crossed with female transgenic mice (either Rosa-Nicd or Rosa-ACTB-tdTomato,-EGFP). Tamoxifen (Tmx; Sigma-Aldrich; catalog no. 10540-29-1) was injected at a dose of 75 mg/kg, once a day for 6 or 10 d, respectively, into mature (8- to 10-wk-old) or aged (8-mo-old) mice, as described previously (61, 62). Littermate controls were injected with corn oil (vehicle). The mice were killed 2 wk after injection. Each group consisted of five or six mice.
Mouse Bone Marrow Stromal Cell Culture.
Primary BMSCs were performed using bone marrow from mouse femurs and tibia as previously described (63). Briefly, bone marrow was flushed from femora and tibia and cultured in α-modified Eagle's MEM (Life Technologies) supplemented with 10% (vol/vol) FBS (HyClone Laboratories) and 1% penicillin/streptomycin (Life Technologies) at a cell density of ∼20 × 106/mL. For osteogenic differentiation, cells were maintained in basal medium for 7 d, and then changed to osteogenic medium (supplemented with 10−8 M dexamethasone, 8 mM β-glycerophosphate, and 50 µg/mL ascorbic acid) for 14 d. Alkaline phosphatase (Alp)-positive colonies (or cfu-f) and von Kossa-labeled mineralizing colonies (or cfu-ob), respectively, were counted 7 and 21 d after culture. ImageJ software v1.41a (NIH) was used to determine areas of Alp and von Kossa staining. Ad-Cre and Ad-EGFP were obtained from Yingzi Yang (NIH).
Osteoclast cultures were performed as described previously (64). Briefly, bone marrow hematopoietic cells were plated in the presence of macrophage colony-stimulating factor (MCSF) (30 ng/ml) for 2 d, followed with both MCSF and receptor activator of nuclear factor kappa-B ligand (Rankl) (25 ng/ml) for 3–5 d. Osteoclast cultures were fixed in 3.7% (vol/vol) formaldehyde and 0.1% Triton X-100 for 5 min, and stained for Trap. Trap+ cells with more than two nuclei were counted as osteoclasts.
Histology and Bone Histomorphometry.
Decalcified and nondecalcified sections of bone were obtained, as described previously (65). Briefly, mice were injected s.c. with calcein (10 mg/kg; Sigma) 8 d (adult mice) or 12 d (aging mice), after which they received a Xylenol orange (90 mg/kg; Sigma) injection. Lumbar vertebrae (L1–4) were fixed with 4% (vol/vol) paraformaldehyde for 18 h at 4 °C. Nondecalcified bones were embedded in optimum cutting temperature compound (O.C.T. compound, Tissue-Tek), and 5- to 7-µm-thick frozen sections were prepared using a transparent film. Static and dynamic histomorphometric analyses were performed according to standard protocols using software kindly provided by Robert J. van’t Hof.
ELISA Assay.
Mouse plasma was stored at −80 °C. ELISA kits were purchased for tartrate-resistant acid phosphatase 5b (Trap5b) and osteocalcin assays and used per manufacturer's protocol (catalog nos. SEA902Mu and SEA471Mu; Cloud Clone Corp.).
Microtomography.
High-resolution µCT scanning (μCT50; Scanco) was performed to measure morphological indices of metaphyseal regions of femur or lumbar vertebrae (L5–6) as previously described (66). The bones were dissected, cleaned, fixed in 10% formalin, transferred to 75% (vol/vol) ethanol, loaded into 10-mm diameter scanning tubes, and imaged. Imaging analysis of metaphyseal regions was performed using 100 slices (10 µm/slice). The most proximal slice was defined as the plane where the growth plate had just disappeared. A Gaussian filter (σ, 0.8; support, 1) was applied to all analyzed scans. Key parameters were as follows: X-ray tube potential, 55 kVp; X-ray intensity, 145 μA; integration time, 200 ms; and threshold, 220 mg/cm3.
Gene Expression Analysis.
Total RNA was extracted with TRIzol reagent (Invitrogen) per manufacturer’s protocol (67). RNA was treated with DNase and first-strand cDNA was synthesized using oligo(dT) primer and SuperScript II reverse transcriptase (Invitrogen). Quantitative real-time PCR was performed in triplicate by using iQ SYBR Green Supermix on a iCycler Real-Time Detection System (BioRad). The relative amount of mRNA was normalized by cyclophilin A expression.
Local Calvarial Injection of Recombinant Jagged1.
Recombinant Jag1 protein was purchased from Sino Biological Inc. (catalog no. 11648-H02H) and injected onto the right side of the calvaria of adult mice using a technique described previously (68⇓–70). Briefly, 8-wk-old C57BL/6J mice were injected with vehicle alone or Jag1 (40 µg/kg, once daily, for 6 d, 50 μL). On day 16, calvariae were harvested, fixed, and subjected to H&E staining, followed by measurement of new bone width.
Mechanical Testing.
Three-point bending and compression/traction of long bones (femurs and tibias) were performed as described previously (71) using a mechanical testing device (Twin Column Table Mounted Testing System; Instron 5565).
Femur Osteotomy.
Mouse femur osteotomy was performed in Col3.2-CreERT2:Rosa-NicdTg/Tg male mice previously reported (72) with modification. Briefly, general anesthesia was induced by i.p. injections of ketamine hydrochloride (750 mg/kg body weight) and xylazine (25 mg/kg body weight). The intramuscular septum between the vastus lateralis and the hamstring muscles was divided by blunt dissection to localize the femur. An external fixation device was attached to the right femur. Then, a 3-mm defect was created in the femur at the midshaft by means of a transverse osteotomy with a dental saw. Absorbable sutures were used to close the intramuscular septum and skin incision. Mice were randomized into each group. At 3 d postoperation, Tmx was injected i.p. every other day for 10 d (n = 5), and mice were killed 3 wk after the last injection. Corn oil was used as control (n = 5). Radiographs were obtained using an Optical In-Vivo Xtreme Imaging System (Bruker).
Acknowledgments
We thank Drs. Jian Q. (Jerry) Feng and Henry M. Kronenberg for providing Col3.2-CreERT2 and Dmp1-Cre mice; Drs. Mary L. Bouxsein and Brian C. Dawson for help with µCT protocols; Dr. Rob van't Hof for providing a bone histomorphometry software; Dr. Jorge Busciglio for Psen1+/− mice; Dr. Thomas Gridley for Jag1fl/fl and Actb-Notch1Tg mice; Dr. Xi Jiang and Ms. Li Chen for assistance with frozen sections and fluorescence imaging; Dr. Liping Wang for assistance with the mouse osteotomy model; and Dr. Yingzi Yang for providing Ad-Cre. Support for this work was provided by National Science Foundation of China Grant 81070689 (to P. Liu); NIH Grants AG40132, AG23176, AR65932, and AR67066 (to M.Z.); and start-up funding from the National Key Lab of Oral Diseases, West China School of Stomatology, Sichuan University, China (P. Liu).
Footnotes
↵1P. Liu, Y.P., M.M., D.Z., L.S., and M.Z. contributed equally to this work.
- ↵2To whom correspondence may be addressed. Email: mone.zaidi{at}mssm.edu, liupossible{at}gmail.com, or maria.new{at}mssm.edu.
Author contributions: P. Liu, D.W.R., and M.Z. designed research; P. Liu, Y.P., M.M., D.Z., C.L., S.Z., S.G., Y.J., F.L., F.Y., P. Lu, A.S., M.B., C.W., L.Z., and K.W. performed research; P. Liu, Y.P., M.M., D.Z., C.L., S.Z., S.G., R.C., T.Y., L.S., and M.Z. analyzed data; and P. Liu, M.I.N., T.Y., L.S., and M.Z. wrote the paper.
Reviewers: D.C., Rush University; C.H., University of Cambridge; and R.P., Emory University School of Medicine.
Conflict of interest statement: M.Z. consults for Merck, Roche, Novartis, and a number of other companies, including Jeffries, Gerson Lehrman Group, and Guidepoint.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1603399113/-/DCSupplemental.
References
Anabolic actions of Notch on mature bone
Article Classifications
- Biological Sciences
- Medical Sciences
Sign up for Article Alerts
Jump to section
You May Also be Interested in
Researchers are exploring whether these ubiquitous fluorinated molecules might worsen infections or hamper vaccine effectiveness.
Image credit: Shutterstock/Dmitry Naumov.
One protein seems to play a key role in touch sensitivity for flytraps and other meat-eating plants.
Image credit: Shutterstock/Kuttelvaserova Stuchelova.
Adam Mastroianni and Daniel Gilbert explore why conversations almost never end when people want them to.
A study finds that giant pandas roll in horse manure to increase their cold tolerance.
Image credit: Fuwen Wei.