Notch Signaling in Osteoblast Progenitor Cells is Required for BMP-Induced Bone Formation

Abstract

Notch signaling is active during bone formation and prior studies have shown that it is required for both robust intramembranous and endochondral bone regeneration. Particularly, the systemic blockade of Notch signaling has been shown to inhibit BMP-induced bone formation in a murine calvarial defect model. In this study, we genetically disrupted the expression of both the dominant Notch receptor, Jagged-1, and the essential Notch signaling transcription factor Rbpj in osteoblast progenitors during calvarial bone healing. We found that Jagged-1 (and Jagged-2) expression by alpha Smooth Muscle Actin (αSMA) expressing progenitors is required for bone formation. Similarly, we found that Notch transcriptional activity within the αSMA lineage is required for BMP-induced bone regeneration. Inhibition of Notch signaling in the αSMA lineage resulted in decreased osteoblast progenitors, reduced vascularization, and sustained inflammation 10 days post-injury, with enhanced inflammation still present 42 days post-injury. We conclude that Jagged ligand induced Notch signaling within the osteoblast progenitor lineage is therefore required for bone morphogenetic proteins (BMP) induced bone regeneration. Modulation of Notch signaling may represent a new approach to promote bone repair.

INTRODUCTION

The Notch signaling pathway has emerged as a key regulator in biological processes across species, including in organ formation, tissue function, and tissue repair. Such signaling is facilitated by a highly conserved ligand-receptor mechanism which plays a vital role in regulation of cellular functions such as proliferation, differentiation, migration, and stem cell renewal. [1][2]

Activation of the cell-to-cell signaling mechanism occurs when one of the Notch ligands (Jagged 1-2 or Delta like 1,3-4) interacts with one of the Notch receptors (Notch 1-4) on a nearby cell. Subsequent enzymatic proteolysis by the gamma-secretase complex results in the cytoplasmic release of the Notch intracellular domain (NICD), which translocates to the nucleus where it binds to central regulator CSL/Rbpj protein and mastermind-like-1 (MAML) and activates transcription of downstream target genes, including the Hes and Hey family of transcriptional repressors. [3]

Notch signaling is an important regulator of skeletal cell proliferation and differentiation during bone healing. [4]–[8] Expression of Notch receptors and ligands have been detected during all phases of endochondral and intramembranous bone formation with the most highly expressed ligand and receptor being Jagged-1 (Jag1) and Notch-2. [4] Disruption of Jag1 in mesenchymal progenitor/stem cells (MSCs) during intramembranous bone formation in a marrow ablation model demonstrated decreased bone formation. [5] Additionally, localized delivery of recombinant Jag1 promoted osteoblast differentiation without diffuse bone hypertrophy. [5] Overexpression of NICD1 has also been shown to result in a more mineralized fracture callus with increased mechanical properties. [6] Prx1Cre mediated deletion of notch signaling modulator Rbpj resulted in tibial fracture nonunions due to depletion of progenitor population. [7] Further, in a transverse tibial fracture model, we previously found that whole-body, inducible disruption of Notch signaling by activated expression of dominant negative MAML in an Mx1-Cre model impairs bone healing while increasing inflammation in the fracture callus. [8] The effects of pharmacological Notch inhibition have also been examined during bone healing. Published work for our laboratory showed treatment with systemic small molecule inhibitor diabenzazepine (DBZ, a gamma secretase inhibitor) reduced bone volume during Bone morphogenetic protein-2 (BMP2) mediated calvarial defect healing in mice, demonstrating that Notch signaling is an important determinant of BMP-induced bone formation. [9]

BMPs are multi-functional growth factors that belong to the transforming growth factor beta (TGFbeta) superfamily. [10] Members of the BMP superfamily affect almost all aspects of bone, cartilage and joint biology. [11] BMP2 was the first BMP to be identified in bone and has been well characterized and extensively investigated as a candidate for the enhancement of skeletal repair. Because of its osteogenic potential, the Food and Drug Administration (FDA) approved usage of recombinant human BMP2 (rhBMP-2) during spinal fusion surgery, tibial shaft repair, and maxillary sinus reconstructive surgery. [12] However, the wider use of BMP has been precluded due to multiple factors including heterogeneity in responses, significant side effects and necessity of supraphysiological doses in treatment. [12–14] Therefore, a need exists to better understand the mechanism controlling BMP2-induced osteogenesis to improve treatment options.

The objective of this study was to investigate the intersection of Notch signaling and BMP-induced osteogenesis and to understand the cellular basis for the deleterious effects seen during reduced Notch signaling. We utilized conditional genetic deletion models in progenitor cells at the time of BMP2-mediated healing. Specially, we disrupted essential Notch signaling transcription factor Rbpj or Notch signaling ligands Jag1 and Jagged-2 (Jag2) in tandem. Our results demonstrate that inhibition of canonical Notch signaling in progenitors decreases bone regeneration, progenitor proliferation and cell number, vascularization, and results in altered inflammation.

MATERIALS AND METHODS

Mice

The study was developed under the guidance of the university-wide AAALAC-accredited laboratory animal medicine program and under the direction of veterinarians with specialization in laboratory animal medicine. All animals in this study were used in compliance with the University of Michigan guidelines and approved IACUC protocol. All experiments were conducted on both male and female mice and sexes were analyzed together. The following transgenic mouse strains on a C57BL/6 background were also used: αSMACreERT2, [15] B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J (Ai9, TdTomato) [16], Jagged-1 and Jagged-2 double floxed [5], and Rbpj flox. [17] The mice were group-housed in ventilated cages within a vivarium and kept at 20 to 23 °C, 30 to 50% relative humidity, a 12-hour light/dark cycle, and given ad libitum access to water and food (Laboratory Rodent Diet 5LOD, PMI Lab Diet, St Louis, MO). Genotyping for all mouse strains was performed by Transnetyx (Cordova, TN) using real-time PCR.

Calvarial defect model

αSMACreERT2 Jag 1/2 mice (Cre− n=12 (8 females, 4 males); Cre+ n=14 (10 females, 4 males)), αSMACreERT2 Rbpj (Cre− n=30 (12 females, 18 males); Cre+ n=28 (13 females, 15 males), αSMACreERT2+ TdTomato (n=4 (2 females, 2 males) aSMACreERT2 Rbpj TdTomato (n=6 (1 female, 5 male) underwent calvarial defect surgery. All mice were aged 14-20 weeks. Animals were sedated through anesthetic inhalation of 4-5% isoflurane for induction and 1-2% isoflurane for maintenance. Sedation was confirmed by lack of responsiveness to a hind-limb toe pinch. The 72-hour extended-release analgesia, Buprenorphine XR (3.25 mg/kg) was administered at the time of surgery, and ocular lubricant was applied at the time of anesthesia to prevent corneal drying. Parietal bones were exposed by single midline sagittal incisions and the periosteum was scraped off to facilitate drilling. Bilateral 3 mm critical-size defects were then created using a piezoelectric drill. A collagen sponge graft (Advanced Biomatrix SpongeCol 5135) was loaded with a solution containing BMP2 (0.25 μg) or saline solution (vehicle) and placed over the bone defect. Mice were treated with 75 mg/kg tamoxifen via intraperitoneal injection on 0, 2 and 4 days post-injury (DPI). On 10 or 42 DPI, whole calvariae were removed and fixed in 10% neutral buffered formalin (NBF) for 48 hours and processed for histology and immunohistochemistry. Samples from 42 DPI were then scanned on a microCT and analyzed using DragonFly software. Data was analyzed using GraphPad Prism. On days 4 through 6 DPI, αSMACreERT2 TdTomato animals were treated with 10 μg/kg DBZ, a Notch inhibitor, or a vehicle control via intraperitoneal injection as described previously for in vivo Notch inhibition. [18] At 6.5 DPI, mice were euthanized using CO₂ and euthanasia was confirmed through secondary methods of bilateral pneumothorax or cervical dislocation. Whole calvariae were removed and processed for immunohistochemistry as described below.

Quantitative RT-PCR analysis

For mouse calvarial defects, a 4-mm biopsy punch (Fisher Scientific) was used to collect tissue centered over each defect, which was subsequently flash-frozen in Trizol reagent (Invitrogen) and stored at −80°C until ready for further processing. Tissue was than homogenized using a tissue homogenizer (Bertin Technologies). Total RNA was isolated and quantified using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). 450 ng of total RNA was reverse transcribed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) in a 10 μL reaction. The resulting cDNA was diluted to a final concentration of 5 ng/μL and 1 μL was amplified using Power SYBR^® Green PCR Master Mix and gene-specific primers in a QuantStudio 6 Pro (Thermo Fisher Scientific) following manufacturer recommendations. Relative expression for each gene was normalized against GAPDH, calculated using the formula 2^−ΔC(t) and reported as mean ± standard deviation. For each unique sample, the qPCR was run in triplicate. Proper amplicon formulation was confirmed by melt curve analysis.

Histology and immunohistochemistry

Tissue was fixed in 10% NBF for 48 hours, then decalcified in a formic acid-based solution (Immunocal Decalcifier,StatLab, McKinney, TX) for 4 hours before being embedded in paraffin or Optimal Cutting Temperature compound (Fisher Healthcare Tissue-Plus, Thermo Fisher Scientific, Waltham, MA). Tissue was sectioned in a coronal plane at 5-7 μm thickness and stained with H&E and Russell-Movat’s modified pentachrome (ab245884, Abcam, Waltham, MA) to assess bone morphology using conventional, qualitative bright-field light microscopy. Immunostaining was performed for PDGFRβ (R&D Systems AF1042, 1:250), F4/80 (Abcam ab300421, 1:100), CD31 (Abcam ab222783, 1:100), and Ly6G (Abcam ab25377, 1:100). Sections for immunostaining were air-dried, post-fixed in 10% NBF for 10 minutes and rehydrated in phosphate-buffered saline (PBS). Sections were permeabilized for 20 minutes with 0.3% Triton X-100 and treated with blocking serum (5% donkey or 5% goat serum, 1% BSA, 2% glycine) for 60 minutes at room temperature. Blocked sections were incubated with primary antibodies diluted in blocking serum overnight at 4°C in a humidified chamber. After washing, secondary antibodies diluted to 1:500 were applied for 1 hour at room temperature (goat anti-rabbit Alexa Fluor 647, Invitrogen A32733; for PDGFRβ staining, donkey anti-goat Alexa Fluor 647, Thermo Fisher Scientific A48265; for Ly6G staining, goat anti-rat Alexa Fluor 647, Invitrogen A48365). Nuclei were counterstained with DAPI (Invitrogen H3570) after secondary antibody incubation. Sections were washed four times with PBS before mounting with Prolong Gold (Thermo Fisher Scientific) Finally, the slides were air-dried and sealed with nail polish. A modified protocol was used for CD31: sections were incubated in UNI-TRIEVE Universal Retrieval Solution (Innovex Biosciences, Richmond, CA) overnight at 40°C prior to blocking and the permeabilization step was omitted. High resolution images were acquired on a Nikon TI2 widefield microscope equipped with NIS-Elements software using a 20x objective at laser wavelengths of 640 and 390. All image analysis was performed with Fiji software. For all quantification, identically sized regions of interest (ROIs) were drawn to encompass each defect edge and surrounding mesenchymal tissue. Fluorescence intensity value represents the average intensity of fluorescence within the ROI. Standardized thresholds were manually set for each fluorescence channel. Measurements were taken on each bone edge of the defect and the average value was used as the final value for the sample. Vascular density represents the ratio of vasculature area to total selection area and was quantified using the Vessel Analysis plugin for Fiji (http://imagej.net/Vessel_Analysis). Cell counts were obtained utilizing the identical images for fluorescent intensity and CellProfiler (v. 4.2.7). Optimal signal diameter was established based on locality of the fluorescence. Nuclear signal was assigned a signal diameter threshold of 9-25 pixel units and cytosolic signal was assigned a signal diameter threshold of 12-30 pixel units. Positive/negative αSMA signal was segmented based on the built-in robust background method. Positive/negative F4/80 and Ly6G signal was segmented based on the built-in Otsu method. Positive/negative PDGFRβ and EdU fluorescent signal was segmented based on the built-in minimum-cross entropy method. Minimum fluorescence threshold was established by analyzing negative controls and determining the highest threshold allowed before CellProfiler recognized false positives. Consistent signal diameter and threshold parameters were then applied to all sections in a single batch.

Histomorphometric Analysis

Light microscopic evaluation of H&E and Russell-Movat’s pentachrome was performed at magnifications ranging from 20x to 600x by a board-certified veterinary pathologist (ILB). Histopathology evaluation was performed with knowledge of experimental groups. Each sample contained two defects which were differentiated by tissue dye (left side = red, right side = green). Descriptive changes of bone repair followed utilized terminology recommended as in the most recent (2012) update to the American Society of Bone and Mineral Research (ASBMR) Histomorphometry Nomenclature Committee. [19] Specifically, the following morphologically based terms were utilized: (1) ossification centers: aggregates of osteoblasts with osteoid deposition but without cancellous bone formation (no defined trabeculae or marrow spaces); (2) woven bone: immature bone characterized by increased osteoblast cellularity and irregular matricellular organization; (3) lamellar bone: mature bone characterized by regular, parallel arrangement of bone matrix (lamellae) with entrapped osteocytes. Each bone defect was assessed individually as to the presence or absence of each of these three morphologic features.

Inflammation was graded semi-quantitatively by an American College of Veterinary Pathology Board-certified pathologist (IB) based on whether neutrophils and/or macrophages were present. Each defect was given a binary score of 1 if inflammation was present, and a score of 0 if inflammation was not present. Descriptive findings indicating type of inflammation (neutrophil, macrophage or mixed) were given for each sample.

Cell Proliferation

To analyze cell proliferation after injury, mice were injected intraperitoneally with 1 mg of 5-ethynyl-2′-deoxyuridine (EdU) (A10044, Thermo Fisher Scientific) dissolved in PBS at 5 DPI, and harvested 6 (Rbpj mice) or 6.5 (GSI treated) DPI. Whole calvariae were removed and processed histologically as described above. Following sectioning, Click-iT EdU Alexa Fluor 647 Imaging Kit (ThermoFisher C10340) was used to stain the sections for EdU and 4′,6-diamidino-2- phenylindole (DAPI; #D9542; 1:1000 dilution; Sigma-Aldrich, St. Louis, MO, USA). EdU in the cells was detected using a fluorescence microscope (BioTek Lionheart FX Automated Microscope). EdU-positive cell numbers were counted and calculated in NIS-Elements using the object count function to identify EdU-positive cells.

Micro-Computed Tomography (microCT)

Calvaria were scanned using microCT (Scanco microCT 100) at the Michigan School of Dentistry’s microCT facility and analyzed in Comet Dragonfly. For calvarial defect analysis, a cylindrical ROI centered over each defect side was defined as 1.45mm in radius and 1.5mm in height and was thresholded using Otsu’s method [20]. The average of the vehicle samples was taken as the Otsu threshold and applied to all the samples to normalize bone density. The defect was analyzed at every 10 slices and a spline interpolation was performed between individual slices to generate the defect ROI. Next, volumetric analysis was preformed to determine bone volume, bone mineral content, bone mineral density, tissue mineral content, tissue mineral density, and bone volume fraction. Bone volume fraction was calculated by dividing the bone volume by the total volume. Bone mineral density was determined by dividing bone mineral content by total volume. Tissue mineral density was determined by dividing the tissue mineral content by the bone volume.

Statistical Analysis

Data are presented as the means ± standard deviation. Statistical analyses were performed using GraphPad Prism (version 10.0). Comparisons between two groups were analyzed using a 2-tailed, unpaired Student’s t test. P ≤0.05 was set as a statistically significant difference. The number of samples assessed, and replicates used for each experimental condition are indicated in each figure legend. Data was tested for normal distribution by QQ plot and passed the D’Agostino & Pearson, Anderson-Darling, Shapiro-Wilk, and Kolmogorov-Smirnov tests. While the experimental design was not focused on determining sex-differences, given that we utilized both male and female mice, two-way ANOVA was utilized to verify that there was not an effect of Sex on the primary outcome of Bone Volume, and that there was not an interaction between Sex and Genotype.

RESULTS

Jagged-1 disruption in mesenchymal cells decreases bone regeneration in cranial defects

We have previously demonstrated that pharmacological Notch inhibition using DBZ reduces bone volume during BMP2 mediated calvarial defect healing of mice. [9] To further explore the intersection between Notch and BMP signaling pathways during bone regeneration, we investigated the impact of Notch inhibition through conditional genetic deletion models in progenitor cells at the time of healing using the Cre-LoxP system. First, we disrupted Notch ligands Jag1 and Jag2 by utilizing an inducible model in αSMA osteoprogenitor cells with the αSMACreERT2 mice. Jag1 is expressed 100-fold more than Jag2 in the mesenchymal lineage, but Jag2 deletion was included to abrogate any potential effect of compensatory Jag2 upregulation. Tamoxifen was administered on the day of surgery and 2 and 4 DPI (Figure 1A). We evaluated effects at 10 and 42 DPI using histologic analysis, microCT, and gene expression (Figure 1A). There was significantly decreased expression of Jag1 (Figure 1B), as well as downstream Notch target gene Hey1 (Figure 1C) in αSMACreERT2+ Jag1/2 calvarial defect tissue compared to Cre-controls at 10 DPI. As our Cre driver targeted aSMA+ cells, the 40% decrease in Jag expression in the total defect is representative of the heterogenous cell population within the defect (Figure 1B). Bone volume via microCT analysis was significantly reduced in αSMACreERT2+ Jag1/2 defects at 42 DPI (Figure 1D, 1E).

Figure 1. Jagged-1 disruption in mesenchymal cells (αSMA-CreERT2) at the time of injury and BMP treatment decreases bone regeneration in cranial defects.

(A) Experimental design for calvarial defect studies. Quantitative gene expression of both (B) Jag1 and (C) Hey1 at 10 DPI. (D) Bone volume via microCT analysis is significantly reduced in Jag 1/2 αSMACreERT2+ defects at 42 DPI. (E) Representative microCT reconstructions of Cre− and Cre+ defect sites. Data are presented as mean +/−1 SD., **P< 0.01, **** P<0.0001 as assessed using a two-tailed Student’s t test. (B-C) n=5 Cre− and Cre+ and (D) n=13 Cre− mice and Cre+ n=17 . Open circles = Cre− females, closed circles = Cre− males; open squares = Cre+ females, closed squares = Cre+ male.

Rbpj deletion in αSMA+ cells perturb BMP-Notch signaling leading to diminished bone formation

To further elucidate the impact of Notch signaling perturbation on BMP-induced bone formation, a tamoxifen-inducible αSMACreERT2 mouse model was used to disrupt the Rbpj gene, a key transcription factor involved in the Notch pathway, in progenitor cells responsible for calvarial bone regeneration. BMP2-mediated bone defect healing in mice was evaluated at 10 and 42 DPI with gene expression, histological analysis, and microCT. Tamoxifen was administered on days 0, 2 and 4 DPI (Figure 2A). Cre recombination was confirmed by the Rbpj gene deletion at the LoxP site containing Exon 6 (Figure 2B). The relatively low level of Rbpj deletion noted in the RNA analysis is presumably because of the heterogenous nature of the injured tissue and Cre was only active aSMA expressing cells. There was significantly decreased expression of Runx2 (Figure 2C), a key transcriptional regulator of osteoblastogenesis, as well as down-stream Notch target gene HeyL (Figure 2D) in αSMACreERT2+ Rbpj calvarial defect tissue compared to Cre-controls at 10 DPI. In micro-CT images, ablation of Rbpj in αSMA expressing cells during the early healing phase resulted in a 36% decrease (p<0.01) in bone volume at 42 DPI (Figure 2E, 2F).

Figure 2. Deletion of Rbpj in αSMA+ cells at the time of injury impairs BMP-induced calvarial bone defect repair.

(A) Experimental design for calvarial defect studies. Quantitative gene expression of (B) Rbpj Exon6, (C) Runx2 (C) and (D) HeyL at 10 DPI. (E) Bone volume via microCT analysis is significantly reduced in αSMACreERT2+ Rbpj defects at 42 DPI. (F) MicroCT reconstruction of the calvaria and the defect site. Left panel, Cre−. Right panel, Cre+. (G) Representative histological staining of calvarial bone repair at 42 DPI. 8x (upper panels) and 20x (lower panels) images of Movat’s pentachrome-staining of bone regeneration in defects. Note the more robust bone formation in controls (αSMACreERT2- Rbpj). Bone edges of defect (red dotted line area), upper panels. Left hand panels, Cre−. Right hand panels, Cre+. In the lower panels, black arrow shows mineralized compact lamellar bone (red staining), WB: woven bone (yellow staining), M: mesenchyme (within sponge matrix), BM: bone marrow space, black scale bar: 300 μm, red scale bar: 70 μm. In graphs, each dot represents a single animal. Data are presented as mean ±1 SD. *P<0.05, **P< 0.01, *** P <0.001 as assessed using a two-tailed Student’s t test. (B-D) n=6 Cre− and Cre+ (E) n=10 Cre− and Cre+. Open circles = Cre− females, closed circles = Cre− males; open squares = Cre+ females, closed squares = Cre+ male.

In addition to the quantitative evaluation performed on the microCT reconstructions, 42 DPI defects were also descriptively evaluated by a board-certified veterinary pathologist (Table 1). These results suggest an overall reduction in ossification centers, woven and lamellar bone, and overall repair score in αSMACreERT2+ Rbpj mice (Figure 2G and Table 1). Pentachrome staining revealed woven immature (yellow) to early mineralized mature bone (red), appearing as red stippling within the defects of αSMACreERT2+ Rbpj mice (Figure 2G, upper and lower right panels) but most of the defect contained mesenchymal tissue and sponge matrix. In comparison, Cre- control defects contained large areas of woven bone and early cancellous bone within the defect and multiple organized areas of mineralized compact lamellar bone (Figure 2G, upper and lower left panels). Overall, these findings indicate inferior bone formation when Rbpj expression is reduced in the αSMA+ stromal cells at 42 DPI.

Table 1. Semiquantitative histology score for bone regeneration.

Day 42 semiquantitative histomorphometry results, shown as percentage of analyzed defects per group that were positive. Cre− n=10 and for Cre+ n=15 defects per group.

Rbpj Flox Mice	Ossification Centers	Woven (Immature) Bone	Lamellar (Mature) Bone
Cre−	100%	50%	40%
Cre+	60%	20%	7%

Notch signaling inhibition in αSMA+ cells result in reduced mesenchymal derived cells in early bone regeneration

To determine potential mechanisms driving impaired bone healing, earlier time points were investigated. Previous studies have shown Notch signaling is required for proliferation and/or migration of MSCs. [7] To visualize the cell population, the αSMACreERT2+ Rbpj floxed mice that we used were crossed with Ai9 reporter mice (TdTomato), which allowed for fluorescent labeling of cells with Cre mediated recombination (Figure 3A). To interrogate MSC recruitment to the injury site, we quantified percentage of αSMA+ cells at the defect site in aSMACreERT2+ TdTomato+ and aSMACreERT2+ RBPJ floxed/floxed TdTomato+ (Figure 3B–D) and assessed total fluorescence intensity of TdTomato (Figure 3E). We found a reduction in TdTomato cell recruitment to the injury site in the Rbpj floxed mice relative to the control mice. The percent of TdTomato+ cell numbers were significantly reduced more than 50% in RBPJ floxed mice (Figure 3D) and Fluorescence Intensity (Figure 3E) was also significantly decreased. Relatedly, we observed a corresponding decrease in the PDGFRβ expression, a progenitor cell marker, both by immunofluorescence (IF) (Figure 3F–J) and gene expression levels (Figure 3K). The number of PDGFRβ positive cells (Figure 3H) and the percent of positive cells (3I) were both statistically significant different between WT and RBPJ floxed animals with decreases of more than 100%. Similarly, fluorescence intensity (Figure 3J) and expression of PDGFRβ (Figure 3K) were statistically significant. This suggests that Notch signaling within the activated mesenchymal cells is necessary for proper cell proliferation and/or recruitment in response to injury.

Figure 3. Notch signaling inhibition results in diminished progenitor cell populations at 10 DPI.

(A) Fluorescence of αSMACreERT2+ TdTomato (A) without (control) or (A’) with Rbpj floxed. Left hand panels are low magnification and right-hand panels are high magnification of the yellow box regions in the left side panels. Red represents αSMA+ cells. White dashed lines indicate bone edges. (B-E) Quantification of TdTomato fluorescence. αSMACreERT2+ TdTom, n=4; Rbpj αSMACreERT2+ TdTom, n=6 (B-D), n=4 (E). Immunofluorescence staining was used to assess PDGFRβ within the calvarial defect at 10 DPI (F-J). Representative fluorescent images of (F) Cre− and of (F’) Cre+ stained with anti PDGFRβ antibody. Left hand panels are low magnification images, and right-hand panels are high magnifications of the yellow boxed regions. Magenta represents PDGFRβ staining. White dashed lines indicate bone edges. (G-J) Quantification of PDGFRβ immunofluorescence in control or Rbpj floxed mice. Cre−, n=4 (J); Cre+, n=5. In graphs, each dot represents a single animal. Scale bars, 200 μM (white), 100 μM (red). Data are presented as mean ±1 SD. *P<0.05, ****P<0.0001. Open circles = Cre− females, closed circles = Cre− males; open squares =Cre+ females, closed squares = Cre+ male.

Decrease in proliferation and impaired angiogenesis by Notch pathway disruption

Next, we assessed potential differences in proliferation at 6 DPI and angiogenesis at 10 DPI (Figure 4A). Proliferation was assessed by EdU staining on 6 DPI calvarial defects (Figure 4B). αSMACreERT2+ Rbpj mice had a significant 50% decrease in the percent of Edu+ cells relative to Cre- controls (Figure 4C–E). Similar findings were seen using the small molecule inhibitor DBZ which effectively disrupts the Notch pathway systemically (Figure S1A), we found that there was a significant decrease in cell proliferation, as assessed by EdU staining (Figure S1B, S1C) as well as Ki67 gene expression (Figure S1D). Overall, these findings support that Notch is active in MSCs, and the disruption of the Notch signaling pathway negatively impacts their proliferation in response to injury.

Figure 4. Decrease in proliferation and impaired angiogenesis by Notch pathway disruption.

(A) Experimental design for calvarial defect studies. Cellular proliferation was analyzed in control or Rbpj floxed mice by (B-E) EdU immunofluorescent staining. (B)Representative EdU staining. Left panel is αSMACreERT2− Rbpj flox (control). Right panel is αSMACreERT2+ Rbpj flox. Magenta represents EdU+ cells. Scale bars, 100 μM (red) (C-E) EdU quantification. Cre−, n=4; Cre+, n=5. Angiogenesis was assessed by (F-H) CD31 immunofluorescence staining and (I) Ang1 quantitative gene expression at 10 DPI. (F) αSMACreERT2− Rbpj flox (control). (G) αSMACreERT2+ Rbpj flox. Left hand panels are low magnification and right-hand panels are high magnification of the yellow boxed regions. Magenta staining is CD31. White dashed lines indicate bone edges. Scale bars, 200 μM (white) and 100 μM (red). (H) Quantification of CD31 staining, Cre−, n=4; Cre+, n=4. (I) Ang1 gene expression. Cre−, n=7; Cre+, n=6. Data in graphs are presented as mean ±1 SD. *P<0.05, ****P<0.0001. Open circles = Cre− females, closed circles = Cre− males; open squares = Cre+ females, closed squares = Cre+ male.

Angiogenesis is a key component of bone repair in the early period of fracture healing, and Notch signaling is known to influence blood vessel formation in bone and during bone regeneration. [21] To evaluate if impaired vasculature contributed to bone healing deficiencies with reduced Notch signaling, we performed CD31 immunostaining at 10 DPI (Figure 4F–G). There was more than a 50% statistically significant decrease in overall vascular density at the defect site in αSMACreERT2+ Rbpj mice relative to Cre- controls (Figure 4F–H). There was also decrease in gene expression of proangiogenic factor, angiopoietin 1 (Figure 4I), which has been shown to enhance BMP2-induced cranial bone regeneration with increased pericyte recruitment. [22]

Rbpj deletion results in sustained neutrophilic-based inflammation during bone regeneration

Surprisingly, histopathological assessment of healing defects at day 10 revealed a dense to moderate neutrophilic inflammation within the mesenchyme of αSMACreERT2+, Rbpj mice (Figure 5A, right panel) but neutrophilic inflammation was absent in Cre-control mice (Figure 5A, left panel). This was semi-quantitively graded, 11% of control specimens (1 out of 9) had some macrophagic inflammation, but 40% of the Cre+ mice (6 of 15) had mixed neutrophil and macrophagic inflammation. Consistent with neutrophil inflammation, Tumor Necrosis Factor-α (TNFα, Figure 5B) and Interleukin-1β (IL-1β, Figure 5C) were upregulated in αSMACreERT2+ Rbpj defects at Day 10 DPI. Macrophage number was quantitatively assessed by F4/80 immunohistochemical staining. There were no differences in the percentage of F4/80+ cells between αSMACreERT2+ Rbpj and Cre-control defects (Figure 5D), suggesting that the increased inflammatory response at this timepoint seen in αSMACreERT2+ Rbpj mice was indeed primarily neutrophilic, as assessed by histopathology (Table 2).

Figure 5. Rbpj deletion results in sustained neutrophilic inflammation during bone regeneration.

(A) Descriptive histolopathologic evaluation of H&E-stained samples at 10 DPI showed macrophage (black arrows) and neutrophil (red arrows) inflammatory infiltrates. Left hand panel is Cre− and right-hand panel is Cre+, 40x. The extent of inflammation was analyzed via quantitative gene expression of inflammatory markers (B) TNFα and (C) IL-1B. Cre−, n=5; Cre+, n=5 and (D) immunofluorescent quantification of F4/80 positive cells at 10 DPI. Cre−, n=5, Cre+, n=7. (E) H&E specimens of Cre+ at 42 DPI continue to show inflammation (red arrows, neutrophils). Top panels are Cre− and bottom panels are Cre+. Left panels are low magnification (8x) and right panels are higher magnification (40x). (F-J) Ly6G immunofluorescence staining was used to assess neutrophilic inflammation at 42 DPI. (F) representative fluorescent images of Cre− and (F’) of Cre+. Left panels are low magnification and right panels are magnification of the area shown in the yellow box. Magenta staining is Ly6G. White dashed lines indicate bone edges. (G-J) Quantitative assessment of fluorescence of Ly6G staining at 42 DPI. Cre−, n=4; Cre+, n=6. (K-P) F480 immunofluorescence staining was used to assess macrophage inflammation at 42 DPI. (K) representative fluorescent images of Cre− and (K’) of Cre+, Left panels are low magnification and right-hand panels are magnification of the area shown in the yellow box. Green staining is F480. White dashed lines indicate bone edges. (L-P) Quantification of F480 immunofluorescence. Cre−, n=4; Cre +, n=6. Scale bars: 300 μM (blue), 200 μM (white), 100 μM (red), 60 μM (green). Data are presented as mean ±1 SD. *P<0.05, ***P<0.001. Open circles = Cre− females, closed circles = Cre− males; open squares = Cre+ females, closed squares = Cre+ male.

Table 2. Semiquantitative histology score for inflammation.

Day 42 semiquantitative histomorphometry results, shown as percentage of defects affected per group total. Cre− Day 10 n=9 and Day 42 n=10 defects per group. Cre+ n=15 for both time points. Mixed = macrophagic and neutrophilic.

Rbpj Flox Mouse		n	% of Analyzed Defects with Inflammation	Type of Inflammation
αSMACreERT2−	Day 10	9	11.1	Macrophage
	Day 42	10	0.0	None
αSMACreERT2+	Day 10	15	40.0	Mixed
	Day 42	15	46.7	Neutrophil

Surprisingly, areas of inflammation were still present in αSMACreERT2+ Rbpj mice at 42 DPI (Figure 5E, lower panels). In total, 46.7% (7 out of 15) of defects were determined to have neutrophilic inflammation present (Table 2). Cre- control defects had little/no inflammation present (Figure 5E, upper panels). To quantitatively characterize this altered inflammatory process, both neutrophil (Ly6G+) and macrophages (F480+) were evaluated by immunofluorescence (F-P). There are many fewer F480 cells (Figure 5M) relative to Ly6g (Figure 5H). There was increased neutrophilic marker Ly6G staining in αSMACreERT2+ Rbpj defects at 42 DPI compared to control counterpart by assessing fluorescence intensity (Figure 5J), although no apparent differences in percent of total cells that were neutrophils (Figure 5G–I). There was no difference in F4/80+ macrophages between groups at this later time point (Figure 5K–P). Our IF results did indicate macrophage presence in both groups at 42 DPI, but this is likely endogenous F480+ macrophages in the bone and adjacent connective tissue. Collectively, the data in Figure 5 demonstrate that Rbpj deletion increased neutrophilic based inflammation that appears to be sustained until day 42.

DISCUSSION

Our results show that Notch signaling within the αSMA lineage – the activated progenitor cells for calvarial bone regeneration – is required for robust BMP-induced bone formation. Both disruption of the Notch signaling ligands Jag1 and Jag2 or Notch signaling canonical transcription factor, Rbpj, resulted in negative effects on BMP-induced healing.

Jag1 is the most highly expressed Notch ligand during both long bone and calvaria bone regeneration [4] and Jag1 is required for osteoblast differentiation. [9] Due to the abundance of Jag1 and its known role in Notch canonical signaling, we sought to interrogate the role of Jag1 in intramembranous, BMP-induced healing in a calvarial defect. To eliminate the potential for compensation by Jag2, experiments were conducted using Jag1/Jag2 double floxed mice. These results are consistent with our previous results of Jag1/Jag2 disruption in a marrow ablation model that did not use BMP-induction for new bone. [5]

Similarly, we have shown that intact Notch downstream signaling through the Rbpj transcription factor in αSMA progeny cells is required for BMP-induced bone formation in calvarial defects. Rbpj has been disrupted in a previous bone healing study in Prx1-Cre cells (non-inducible), resulting in non-unions [7] but to our knowledge has not been disrupted during BMP-induced bone formation. This result, coupled with prior systemic inhibition of Notch signaling [6][9] and Jag1/2 genetic disruption in the marrow ablation model, [5] suggests that Notch signaling within the osteoblast progenitor population is required for proper bone formation. This is also consistent with recent studies showing that genetic activation of Notch signaling in the αSMA progenitor population through overexpression of Notch 1 intracellular domain, can increase bone healing. [6]

Undoubtedly, some effect of disruption of Notch signaling in the progenitor population on BMP-induced bone healing is related to a failure of cells to proliferate. We show reduced EdU staining of calvaria and reduced αSMA TdTomato+ progeny using lineage tracing when Rbpj is disrupted and reduced PDGFRβ positive cells. Both PDGFRα [23][24] and PDGFRβ [25] positive cells expand during bone regeneration and are required for proper healing. Particularly, Xing et al. [24] has shown a requirement for PDGFRα positive cells to proliferate for calvarial healing. Our findings of reduced osteoblast progenitor number are consistent with our previous finding that blocking Notch signaling in long bone fracture can reduce progenitor cell number and proliferation. [6]

The healing calvarial tissue is a heterogenous population and, as we also show here, the effect of Notch signaling disruption in the mesenchymal cell population is complex, and there are apparent cell non-autonomous effects. Angiogenesis is decreased and inflammation, particularly neutrophils, is sustained when Notch signaling within the αSMA population is disrupted through genetic deletion of Rbpj. Notch signaling disruption within endothelial cells been shown to be an impactful regulator of new blood vessel formation. [26][27] Interestingly, excess Notch signaling within the endothelial cells have been shown to increase blood vessel density but has little effect on bone regeneration. [28]

In addition, Notch signaling can have a pro-inflammatory effect. [29] Rbpj is implicated in rheumatoid arthritis with aberrant T and B cell activity. [30] A previous study has shown that overexpression of a dominant negative inhibitor of Notch signaling using Mx1-Cre (dnMAML) results in sustained inflammation and altered bone healing. [8] There is a strong link between sustained inflammation and reduced tissue regeneration, particularly in bone healing. [31–33] The current literature suggests that increased or prolonged influx of neutrophils into the fracture hematoma may mediate impairment of bone regeneration after hyper-inflammatory conditions. [34]

The mechanism(s) by which reduced Notch signaling within the bone forming cell lineage (αSMA progeny) decreases angiogenesis and increases neutrophilic inflammation are not yet understood. These effects could be due to an alteration in direct cell to cell communication. For instance, perhaps inhibition of Notch signaling within progenitor cells, or differentiated osteoblasts, alters the secretome of the cells such that endothelial progenitor proliferation is reduced. Osteoblastogenesis and vascularization are generally thought to be tightly linked. [35] These cells produce both positive and negative modulators of angiogenesis, including VEGF and angiopoietin and thrombospondins, respectively. We show a reduction in angiopoietin expression likely by the osteogenic progenitors.

It is unclear how progenitor cells and their progeny are involved with clearing inflammation and why there is increased neutrophilic inflammation in the absence of Notch signaling in osteogenic progenitors. Stromal progenitor cell recruitment to the injury site is necessary for the resolution of inflammation. [36] The increase in inflammation we observe with reduced Notch signaling is likely because the typical signals osteogenic progenitors and differentiating osteoblasts release to clear inflammation [32] are disrupted when Notch signaling is inhibited. Similarly, the effect of chronic inflammation could be an indirect effect due to the failure of bone to properly form. While pathologic increases in blood vessels can contribute to inflammation, vascularization plays a key role in regulating inflammation. [37] Decreased vascularization may result in increased retention of inflammatory cells at the injury site. [38]

Future studies will continue to probe the mechanistic underpinnings of these complex cellular observations. While our inducible deletion models have utility, it becomes increasingly difficult to differentiate primary, cell autonomous effects from downstream cell non-autonomous effects. It is likely that progenitor proliferation, vascularization, and inflammation are inextricably linked and that it will be difficult to dissect direct effects of Notch from downstream effects. Importantly, this work further supports that clinically, activating Notch signaling could become an effective mechanism to enhance bone healing, and may do this through multiple pro-regenerative effects.

Highlights.

• Jagged ligand expression by osteoblast progenitors is required for BMP-induced calvarial bone regeneration

• Notch signaling within the osteoblast progenitor cells is required for BMP-induced induced calvarial bone regeneration

• Notch signaling is required for BMP-induced progenitor cell proliferation and angiogenesis within calvarial defects

• Decreases in Notch signaling result in sustained neutrophil inflammation