Department of Clinical Studies-New Bolton Center, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
Department of Animal Biology, School of Veterinary Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
Department of Orthopaedic Surgery, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
Dean W. Richardson Chair in Equine Disease Research, Associate Professor of Musculoskeletal Research and Orthopaedic Surgery, 311 Hill Pavilion, 380 S. University Ave., Philadelphia, Pennsylvania 19104, USA
Author contributions: F.Z.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; M.S.: collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; K.H.: conception and design, financial support, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
first published online in STEM CELLS EXPRESS December 29, 2012.
JAG1, the gene for the Jagged-1 ligand (Jag1) in the Notch signaling pathway, is variably mutated in Alagille Syndrome (ALGS). ALGS patients have skeletal defects, and additionally JAG1 has been shown to be associated with low bone mass through genome-wide association studies. Plating human osteoblast precursors (human mesenchymal stem cells—hMSCs) on Jag1 is sufficient to induce osteoblast differentiation; however, exposure of mouse MSC (mMSC) to Jag1 actually inhibits osteoblastogenesis. Overexpression of the notch-2 intracellular domain (NICD2) is sufficient to mimic the effect of Jag1 on hMSC osteoblastogenesis, while blocking Notch signaling with a γ-secretase inhibitor or with dominant-negative mastermind inhibits Jag1-induced hMSC osteoblastogenesis. In pursuit of interacting signaling pathways, we discovered that treatment with a protein kinase C δ (PKCδ) inhibitor abrogates Jag1-induced hMSC osteoblastogenesis. Jag1 results in rapid PKCδ nuclear translocation and kinase activation. Furthermore, Jag1 stimulates the physical interaction of PKCδ with NICD. Collectively, these results suggest that Jag1 induces hMSC osteoblast differentiation through canonical Notch signaling and requires concomitant PKCδ signaling. This research also demonstrates potential deficiencies in using mouse models to study ALGS bone abnormalities. STEM Cells2013;31:1181–1192
Jagged-1 (Jag1), a single transmembrane glycoprotein encoded by JAG1, is one of the main ligands for Notch receptors on mammalian cell membranes and is expressed in a wide-variety of tissues. Mutations in JAG1 are associated with Alagille syndrome (ALGS), a variably expressed autosomal dominant condition [1, 2]. While the dominant feature of ALGS is liver choleostasis, there are also bone manifestations including bone demineralization and many patients experience significant morbidity due to pathologic fractures that heal poorly and are prone to recurrence [3–5]. Additional research linking Jag1 to the regulation of human bone includes an association of Jag1 with low bone mass through genome-wide association studies (GWAS) .
In mammals, four Notch receptors (Notch 1–4) and five ligands (Jagged-1, Jagged-2, Delta-like-1, Delta-like-3, and Delta-like-4) have been identified [7, 8]. Jag1 binding to Notch receptors results in a series of enzymatic cleavages that release the Notch intracellular domain (NICD). The released NICD translocates from the cytoplasm to the nucleus and initiates transcription of Notch target genes by interacting with the DNA binding CSL/RBP-J protein, MAML (the transcriptional coactivator, mastermind-like-1), and other transcription factors. Formation of a ternary complex involving NICD, CSL (CBF1/RBPJ-kappa in vertebrates), and MAML is required for initiating transcription associated with canonical Notch signaling, and drives the expression of Hey and Hes family genes .
The role of Notch signaling in mesenchymal stem cell (MSC) differentiation and bone formation has yielded conflicting results. Some studies report that Notch either directly or indirectly inhibits nuclear factor of activated T-cell activity or as a consequence can inhibit osteoblastogenesis through its interactions with Foxo1 . In murine preosteoblast cells, Notch also represses osteoblast maturation through the binding of NICD or the Notch target genes Hes1 and Hey1 to Runx2 [11–13]. In addition, another group reported that Notch signaling in mesenchymal progenitor cells causes osteopenia and impairs osteoblastogenesis by inhibiting the Wnt/β-catenin pathway . In contrast, other studies suggest that Notch signaling enhances osteoblastogenesis in murine osteoblast cell MC3T3E1 and human vascular smooth muscle cells, although the mechanisms involved in this pathway are poorly understood [15, 16].
Notch signaling is a key mechanism in the control of stem cell differentiation and embryogenesis. However, the function during mesenchymal cell differentiation, and specifically, in bone homeostasis, remains largely unknown. In our previous studies of murine fracture healing, we have found that Notch signaling components, especially Jag1 and Notch-2, were upregulated during both endochondral and intramembranous bone regeneration . In this study, both primary human and mouse MSCs were used to evaluate the effects of Jag1-activated Notch signaling on MSC osteoblast differentiation. Pharmacological analyses of Jag1-induced osteoblastogenesis were conducted and show that Jag1 is sufficient to induce osteoblast differentiation in hMSC, while conversely inhibiting osteoblastogenesis in mMSCs. Importantly, we also discovered that protein kinase C δ (PKCδ) activity is required for Jag1-induced hMSC osteoblastogenesis. NICD directly interacts with PKCδ and Jag1 directly increases PKCδ kinase activity and nuclear translocation. These findings demonstrate the potential importance of Jag1 in regulating human osteoblastogenesis and provide a possible explanation for why human patients lacking Jag1 may have low bone mass and a poor ability to heal fractures.
MATERIALS AND METHODS
Isolation and Culture of MSC
Primary hMSC were isolated from vertebral bodies obtained through the National Disease Research Interchange (Philadelphia, PA) or extracted from mononuclear cells from Lonza (Walkersville, MD, http://www.lonza.com/research), as previously reported . hMSC were cultured in assay medium with 20% defined fetal bovine serum (FBS) (Hyclone, Logan, UT, http://www.thermoscientific.com). Assay medium is McCoy's 5a medium (Invitrogen, Grand Island, NY, http://www.invitrogen.com) supplemented to additionally contains 20 mg/ml asparagine, 10 mg/ml serine, 0.75× Minimum essential media (MEM) vitamins, 0.38× MEM amino acids, 75 μm nonessential amino acids, 100 U/ml penicillin/streptomycin, 2.3 mM L-glutamine, 1.3 mM Sodium pyruvate, 0.06% sodium bicarbonate, and 50 mM Beta-mercaptoethanol (Invitrogen). Primary mMSCs were harvested from femora and tibiae of 3–4-month-old BLK6 mice as previously reported . Single-cell suspensions from three mice were pooled and then cultured in Mesenchymal progenitor cells (MPC) culture media (α-minimum Eagle's medium supplemented with 10% FBS, 2 mM L-glutamine, 25 μg/ml sodium ascorbate, and 100 U/ml/streptomycin). MSCs were induced in osteogenic medium (MPC medium plus 25 mg/ml ascorbic acid and 5 mM β-glycerol phosphate).
Immobilized and Transient Jagged-1 Treatment
Immobilized Jagged-1 (R&D, Minneapolis, MN,http://www.rndsystems.com) was used in long-term (over 8 hours) treatment of MSCs . Briefly, tissue culture plates were precoated with 10 μg/ml of antibody against the Fc portion of human IgG (Jackson ImmonoResearch, West Grove, PA, http://www.jacksonimmuno.com) for 1 hour. Excess antibody was removed and then wells were incubated with the indicated concentration of recombinant rat Jagged-1/human Fc IgG chimeric protein for 2 hours. Control plates were incubated with human Fc IgG only or untreated as indicated. MSCs were then plated to the precoated plates.
For transient Jag1 treatment, agarose protein G beads were incubated with the indicated concentration of recombinant Jagged-1/human Fc IgG chimeric protein for 2 hours at 4°C. Protein G beads (21 μl) and Jag1 solution (158 μl) were incubated per cm2 of plate area. Control agarose protein G beads were treated similarly but with phosphate buffered saline (PBS) alone. Jagged-1 bound or control beads were then used to treat MSCs. Beads were removed upon cell harvest from 0 to 24 hours as indicated.
Primary antibodies against PKCδ (Santa Cruz Biotechnology, http://www.scbt.com/), NICD for Western blotting (Cleaved Notch1 Val1744, Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com), NICD for immunostaining (Activated Notch1, Abcam, Cambridge, U.K., http://www.abcam.com), and β-actin (Santa Cruz Biotechnology) were used. Secondary antibodies conjugated to horseradish peroxidase were purchased from Chemicon (Chemicon International, Billerica, MA, http://www.chemicon.com). The Alexa Fluor 488 or 594 donkey anti-rabbit IgG was used for immunofluorescent staining (Invitrogen).
Chemicals and Recombinant Protein
Fc portion of human IgG was from Jackson ImmonoResearch. Recombinant rat Jagged-1/human Fc IgG chimeric protein was from R&D systems (http://www.rndsystems.comMinneapolis, MN, http://www.rndsystems.com). Gamma-secretase inhibitor (GSI) was from Sigma-Aldrich (Sigma, St. Louis, MO, http://www.sigmaaldrich.com). SB203580, U0126, PD168393, and Rottlerin were from Calbiochem (http://www.millipore.com). H89, bisindolylmaleimide (Bis-I), Gö6976, SP600125, LY294002, and D4476 were kind gifts of Dr. Serge Fuchs.
Plasmid Construction and Retroviral Supernatants
Retrovirus plasmids MigR1, MigR1-NICD2, and MigR1-dnMAML were kind gifts of Dr. Warren Pear. Full-length human cDNA sequence for Hey1 (MHS4771-99611155, OpenBiosystems, Huntsville, AL, http://www.thermoscientificbio.com/openbiosystems) was subcloned into the murine myeloproliferative sarcoma virus-based retroviral vector (PRLP2). Retroviral supernatants were harvested as per our previous study .
Cell Viability Assay
Cell viability assay was performed with CellTiter 96 AQueous Non-Radioactive Cell Proliferation Assay kit (Promega, G5421, Madison, WI, http://www.promega.com) according to the manufacturer's instructions. Briefly, 4 × 103 cells were plated into each well of Jag1 precoated 24-well plates. Cells in 200 μl medium were incubated with 40 μl (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS)/phenazine methosulfate)(PMS)) solution for 2 hours at 37°C in a humidified, 5% CO2 atm at the indicated culturing time course, and then 100 μl reacted MTS/PMS solution was used to measure the absorbance at 490 nm in 96-well plate using Varioskan Flash plate reader (Thermo Electron Corporation).
Alkaline phosphatase (ALP) staining was carried out using an alkaline phosphatase kit (Sigma-Aldrich, 86R-1KT) according to the manufacturer's instructions. Briefly, cells were fixed by citrate-acetone-formaldehyde fixative solution for 30 seconds at 6th day after treated with Jag1, then incubated with alkaline-dye mixture for 15 minutes, and rinsed with deionized water twice.
For Alizarin red S staining, cells were harvested between days 10 and 14 (unless otherwise indicated). The cells were washed in PBS and fixed in 50% ethanol for 5 minutes. The ethanol solution was removed and an Alizarin red S solution (1%) was added for 5 minutes. Each well was rinsed three times with PBS.
Immunocytochemistry and Quantification
MSCs were plated on sterile glass coverslips precoated for 2 hours with culture medium containing 10% FBS. Wells were treated with either Jag1 bound or plain agarose beads for the length of time indicated, rinsed once in PBS containing calcium and magnesium, and then fixed in 4% formalin. Fixed cells were washed gently and repeatedly to remove as many beads as possible. Cells were then permeabilized with 0.05% Tween 20 and then blocked in 2.5% goat serum + 4% bovine serum albumin for 20 minutes at room temperature. Cells were incubated with rabbit antitype PKCδ (1:50) or rabbit antitype NICD (1:75) for 1 hour at room temperature or overnight at 4°C, and with the indicated secondary antibody (1:200) for 30 minutes at room temperature. Coverslips were mounted with Vectashield containing 4′,6-diamidino-2-phenylindole (DAPI) to stain all nuclei (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). Cells were imaged using an Olympus BX51 fluorescent microscope. Images were obtained using SPOT RT3 camera and acquisition software. Contrast was adjusted postimaging equally for all images.
The ratio of nuclear to cytoplasmic PKCδ staining intensity was obtained using MetaMorph software. Briefly, the region corresponding to the nucleus in each image was selected based on DAPI staining. To avoid inclusion of the cytoplasm, the outer five pixels of each nucleus was excluded. A ring 12 pixels away from the outer perimeter of the nucleus, and 13 pixels thick, was automatically selected for the cytoplasm. Rings that did not fit entirely into a cell body were automatically excluded. The average intensity of these two regions for each cell was used to calculate the nuclear/cytoplasmic ratio. Experiments were repeated three to four separate times. For each treatment, 9–10 unique ×20 fields from two to four coverslips were used for quantification.
Total RNA was isolated at various time points using the RNeasy kit (Qiagen, Valencia, CA, http://www.qiagen.com) according to the manufacturer's instructions. RNA was dissolved in DEPC ddH2O and stored at −80°C. All primers were designed using the Primer3 program (Whitehead Institute, Cambridge, MA, http://frodo.wi.mit.edu). cDNA was synthesized using Superscript III reverse transcriptase (Invitrogen). Quantitative RT-PCR was performed using an ABI 7600 Fast Thermocycler with SYBR Green (BioWhittaker Molecular Application, Rockland, ME, http://www.biowhittaker.com). All reaction products were normalized to the expression level of mRNA. The primer sets used are as follows: primers for human: ALP forward, 5′-GTA AGG ACA TCG CCT ACC AG-3′; reverse, 5′-GGC TTT CTC GTC ACT CTC AT-3′; Osterix forward, 5′-TGC TTG AGG AGG AAG TTC AC-3′; reverse, 5′-AGG TCA CTG CCC ACA GAG TA-3′; IBSP forward, 5′-ACA ACA CTG GGC TAT GGA GA-3′; reverse, 5′-CCT TGT TCG TTT TCA TCC AC-3′; Id1 forward, 5′-CTC TAC GAC ATG AAC GGC TGT-3′; reverse, 5′-TGC TCA CCT TGC GGT TCT G-3′; PKCδ forward, 5′-GCC AAC CTC TGC GGC ATC A-3′; reverse, 5′-CGT AGG TCC CAC TGT TGT C-3′; Hey1 forward, 5′-ACG AGA CCG GAT CAA TAA CA-3′; reverse, 5′-ATC CCA AAC TCC GAT AGT CC-3′; Hes1 forward, 5′-TTT TGG ATG CTC TGA AGA AAG-3′; reverse, 5′-GTA CTT CCC CAG CAC ACT TG-3′; β-actin forward, 5′-AGA CCT GTA CGC CAA CAC AG-3′; reverse, 5′-CGA TCC ACA CGG AGT ACT TG-3′. Primers for murine: ALP forward, 5′-GTC ATC ATG TTC CTG GGA GA-3′; reverse, 5′-GGC CCA GCG CAG GAT-3′; Osterix forward, 5′-GGT CCC CAG CTC GAG GAT-3′; reverse, 5′-CTA GAG CCG CCA AAT TTG CT-3′; Integrin-binding sialoprotein (IBSP) forward, 5′-TTC CAT CGA AGA ATC AAA GC-3′; reverse, 5′-TCG CCG TCT CCA TTT TCT TC-3′; Hey1 forward, 5′-GGT ACC CAG TGC CTT TGA GA-3′; reverse, 5′-ACC CCA AAC TCC GAT AGT CC-3′; Hes1 forward, 5′-CCA AGC TAG AGA AGG CAG ACA-3′; reverse, 5′-GTC ACC TCG TTC ATG CAC TC-3′; β-actin forward, 5′-AAG AGC TAT GAG CTG CCT GA-3′; reverse, 5′-TGG CAT AGA GGT CTT TAC GG-3′. To ensure primer specificity, melt curves were performed after 45 cycles of PCR. Fold differences in gene expression were calculated using the ΔΔCt method with normalization to β-actin.
Immunoprecipitation and Western Blotting
The experimental procedures for immunoprecipitation and immunoblotting were performed as described previously . Briefly, cleared lysates were transferred to a clean Eppendorf tube and incubated with primary antibody on ice for 30 minutes, then protein G beads were added and the mixture was incubated at 4°C for 2 hours with rotation. The beads were washed with lysis buffer three times. The immunoprecipitation complexes or lysates were directly dissolved in SDS-PAGE sample buffer and separated by SDS-PAGE. The immune detection was performed as instructed by the enhanced chemiluminescence kit (Thermo Scientific).
PKCδ Activity Assay
PKCδ activity assay was carried out using Pep Tag Assay Kit for Nonradioactive Detection of PKC (Promega V5330) as described in the manufacturer's instructions. Briefly, 5 × 105 hMSC cells treated with Jag-1 bonded Protein G for indicated time were homogenized to extract the crude PKC in 0.5 ml extraction buffer. PKCδ was pull down from the crude PKC extracts by incubating with primary anti-PKCδ and protein G beads. The immunoprecipitation complexes were used for the assay, which was performed according to the manufacturer's instructions. Phosphorylated peptide migrated toward the anode (+), while nonphosphorylated peptide migrated toward the cathode (−). The intensity of the phosphorylated bands was quantified using ImageJ software (National Institutes of Health, Bethesda, MD, http://rsbweb.nih.gov/ij).
Student's t tests were used to determine whether treated groups were significantly different from controls at p < .01.
Jag1 Is Sufficient to Induce hMSC Osteoblastogenesis and Inhibits mMSCs Osteoblastogenesis
Notch signaling is activated through direct binding of Notch ligand and the receptor via cell–cell contact [14, 23]. Experimentally, soluble Jag1 lacks the ability to stimulate the Notch receptor (data not shown). Therefore, to determine the role of Jag1 in osteoblastogenesis, we plated hMSC or mMSC on Jag1 (0, 2, and 10 μg/ml) immobilized to precoated tissue culture dishes for 48 hours. By quantitative RT-PCR, Notch target genes Hey1 and Hes1 are significantly increased in both hMSC and mMSC after immobilized Jag1 treatment for 48 hours (Fig. 1A). In hMSC, the osteoblast marker genes hALP (tissue nonspecific ALP, liver/bone/kidney), BSP (bone sialoprotein), and Osterix are all increased between 1.2- and 1.5-fold with 2 μg/ml of Jag1 and between two- and fivefold with 10 μg/ml of Jag1, indicating a positive dose-response to Jag1 treatment (Fig. 1A). In contrast to hMSC, ALP in mMSC decreased 20% with 2 μg/ml Jag1 treatment and 50% with 10 μg/ml Jag1 treatment; BSP in mMSC decreased 50% with 2 μg/ml Jag1 treatment and 65% with 10 μg/ml Jag1 treatment; and Osterix in mMSCs decreased 20% with 2 μg/ml Jag1 treatment and 75% with 10 μg/ml Jag1 treatment (Fig. 1A).
Consistent with the qPCR results, ALP activity (Fig. 1B) and Alizarin red S (Fig. 1C) staining for mineralized matrix results show that immobilized Jag1 induces hMSC terminal osteoblast differentiation, while inhibiting mMSCs osteoblast differentiation. In order to account for potential hMSC donor-to-donor variability, three individual donors were used to determine the effect of Jag1 on hMSC osteogenesis. Results from all three donors show that although there is donor-to-donor variability, Jag1 has similar effects on osteoblast induction (supporting information Fig. S1A–S1C).
To better understand the temporal dynamics of Jag1-induced osteoblastogenesis, we treated hMSC for 12, 24, or 48 hours with immobilized Jag1. Upregulation of ALP and BSP RNA required 24 hours of treatment, whereas expression of Hey1 and Hes1 was increased as early as 12 hours (Fig. 2A). Conversely, Id1, a target gene of Bone morphogenetic protein (BMP)-Smad signaling, does not increase with Jag1 treatment (Fig. 2A) indicating that BMP-Smad signaling is not involved in Jag1-induced hMSC osteogenesis.
MTS assay were used to evaluate the effect of Jag1 on the growth of hMSC and mMSC at days 0.5, 4, 7, and 14 (Fig. 2B). There is no effect of 2 μg/ml Jag1 treatment on hMSC proliferation. Intriguingly, Jag1 significantly inhibited hMSC proliferation at 10 μg/ml. mMSCs display 18% less growth at day 4 and 8% less growth at day 7, but have 8% more cells at day 14 than untreated cells at the same time points.
CSL-NICD-MAML Complex Is Required for Jag1-Induced hMSC Osteoblast Differentiation
To determine whether Jag1 induction of hMSC osteoblastogenesis occurs through canonical Notch signaling, retroviral transduction was used to enforce expression of either NICD2 or dnMAML (dominant-negative MAML1). Following 24 hours of viral transduction, cells were replated to Jag1 (10 μg/ml) precoated dishes for 48 hours before harvest of RNA for qRT-PCR analysis. In the NICD2 overexpressing cells, Hey1 expression is increased by 80-fold, Hes1 by 2.5-fold, ALP by over 34-fold, and BSP is increased by approximately 3.5-fold relative to cells transduced with the control vector. Furthermore, treatment of NICD2 overexpressing cells with 10 μg/ml of Jag1 further induced the expression of Hes1, Hey1, ALP, and BSP (Fig. 3A). dnMAML overexpression conversely decreased Hey1 expression by 60%, Hes1 by 37%, ALP by 58%, and BSP by 80% when compared with control cells. Overexpression of dnMAML also inhibited Jag1 induction of Hey1, Hes1, ALP, and BSP mRNA expression in transduced cells. The gene expression results are consistent with both staining for ALP activity (Fig. 3B) and staining for mineralized matrix (Fig. 3C). These histochemical stains demonstrate that NICD2 can induce both ALP activity and mineralization independent of exogenous Jag1, and that dnMAML can partially block the osteoinductive effects of Jag1.
To investigate whether Notch target genes are involved in the process of Jag1-induced osteogenesis, Hey1 was overexpressed in hMSC. Hey1 gene expression in the overexpressing cells was increased over 7,000-fold relative to control cells (supporting information Fig. S2A); however, there were only minor changes in the expression of ALP (supporting information Fig. S3A, S3B). Interestingly, Hey1 overexpression slightly attenuated Jag1-induced Osterix expression, this finding is consistent with the ALP staining and mineralization staining results (supporting information Fig. S2B, S2C). Collectively, these studies show that Jag1 induces hMSC osteogenesis via CSL-NICD-MAML, but that the effect is not Hey1 dependent.
To further investigate the mechanism of Jag1 induction of hMSC osteogenesis, hMSC from two donors were subjected to various pharmacologic kinase inhibitors in conjunction with Jag1 treatment. We evaluated pathways that have either been previously associated with osteoblastogenesis or Notch signaling. The inhibitors evaluated included Notch inhibitor GSI (2 μM), CK1 inhibitor D4476 (5 μM), PKC inhibitors Gö6976 (1 μM) and Bis-I (4 μM), PKA inhibitor H89 (2 μM), Erk1/2 inhibitor U0126 (10 μM), p38 inhibitor SB203580 (10 μM), EGFR inhibitor PD1683963 (10 μM), Akt inhibitor LY294002 (10 ng/ml), and JNK inhibitor sp600125 (40 nM). hMSCs were plated on Jag1, treated with inhibitors, and ALP gene expression was evaluated (supporting information Fig. S3A--S3D). Treatment with GSI abolishes Jag1-induced ALP expression which is consistent with the NICD2 overexpression results. Interestingly, PKC inhibitors Gö6976 and Bis-I show strongly significant effects on reducing Jag1-induced ALP expression and the results were consistent between the two donors, compared to variable results with other kinase inhibitors. Gö6976 treatment reduced ALP expression approximately 55% (donor 1) or 81% (donor 2) in Jag1 (10 μg/ml)-treated hMSC, while having no effect on ALP expression in Jag1-untreated cells. Furthermore, Bis-I treatment reduced ALP expression approximately 80% (donor 1) or 94% (donor 2) in Jag1-treated hMSC, although Bis-I shows some limited repression of ALP expression in Jag1-untreated cells.
To further determine the role that PKC plays in regulating Jag1-induced osteogenesis, multiple doses of PKC inhibitors were used to treat hMSC, including the PKCδ-specific inhibitor Rottlerin. All the three inhibitors reduced Jag1 induction of both ALP expression and mineralization in hMSC in dose-dependent manner (Fig. 4A--4D). Particularly, Rottlerin treatment shows a more effective inhibition of mineralization than Gö6976 and Bis-I. The addition of Rottlerin reduced ALP in a dose-dependent manner (Fig. 4C). Taken together, PKC inhibitors Gö6976, Bis-I, and the PKCδ-specific inhibitor, Rottlerin, reduce Jag1-induced hMSC osteogenesis and mineralization.
PKCδ Regulates Jag1-Induced hMSC Osteoblastogenesis by Interacting with NICD
Since Jagged-1 binding to Notch results in release of NICD rapidly (Fig. 5A, supporting information Fig. S5A), we developed a new approach to treat hMSC with molecularly oriented Jag1 for shorter periods of time than is required in plating cells on immobilized Jag1. Recombinant Jagged-1/human Fc chimeric protein was bound to Protein G Agarose beads (or Protein A) through the binding of the Fc fragment fused to Jag1 to the Protein G. Western blot analysis with HRP-conjugated (HRP, Horseradish peroxidase) anti-Fc antibody was performed to evaluate the binding efficiency of Protein G Agarose beads and recombinant Jagged-1/human Fc chimeric protein, and results showed that Protein G Agarose beads could bind the recombinant Jag1/Fc efficiently (supporting information Fig. S4A). Comparing empty Protein G Agarose beads treated control cells, the expression of ALP and Hes1 was significantly increased in all three donor hMSC treated with Jag1 bound Protein G Agarose beads in a dose- and duration-dependent manner (supporting information Fig. S4B, S4C) that was consistent with our results using tissue culture plate immobilized Jag1.
To investigate the mechanism by which PKCδ regulates NICD-induced osteogenesis, we examined the association between PKCδ and NICD using coimmunoprecipitation. In retroviral NICD2-transduced hMSC, PKCδ directly associates with NICD2 as demonstrated with coimmunoprecipitation (Fig. 5B). Furthermore, an interaction between endogenous NICD and PKCδ was stimulated upon Jag1 treatment (Fig. 5A, 5C). No changes in PKCδ levels were detected with Jag1 treatment, but as anticipated levels of NICD increase rapidly following Jag1 exposure (Fig. 5A). Collectively, the data demonstrate that Jag1 treatment stimulates the physical interaction of PKCδ and NICD and the amount of NICD bound to PKCδ increase over time consistent with enhanced NICD present following Jag1 signaling.
Jag1 Treatment Results in PKCδ Translocation and Kinase Activation
Considering that PKCδ physically associates with NICD and blocking PKCδ inhibits Jag1-induced osteogenesis, the localization of PKCδ and NICD was evaluated. Immunostaining revealed that in untreated hMSC cells, a small amount of NICD was mainly located in nucleus, whereas PKCδ was present diffusely in the cytosol, but largely excluded from the nucleus (supporting information Fig. S5A, S5B). Within 4 hours of Jag1 stimulation, NICD was increased, released to cytosol, and then translocated to nucleus. In association, PKCδ was redistributed to the nucleus without apparent changes in protein expression level (Fig. 5A, supporting information Fig. S5B). The ratio of PKCδ located in nucleus versus the cytosol shows a 1.4-fold increase with 10 minutes of Jag1 treatment and a 2.8-fold increase with 4 hours Jag1 treatment (Fig. 6A). Next, the kinase activity of immunoprecipitated PKCδ from Jag1-treated hMSC was measured. The PKCδ kinase activity was enhanced 1.9-fold with 10 minutes and 2.6-fold with 4 hours of PGB-Jag1 treatment (Fig. 6B). Taken together, Jag1 treatment stimulates the conuclear translocation of PKCδ and NICD, and this coincides with an increase in PKCδ kinase activity.
JAG1 is variably mutated in ALGS patients with skeletal defects and poor bone healing. As well, JAG1 has been identified as a gene associated with osteoporosis in GWAS and a recently identified Jag1 null mutation is responsible for the development of osteogenesis imperfecta. Despite the clinical and genetic evidence that suggests that Jag1 may positively regulate bone mass in humans, there is a paucity of data that has explored the role of Jag1 in human osteoblastogenesis.
Recent studies in mice demonstrate a suppressive effect of Notch signaling in osteoblast differentiation through the repression of Runx2 activity by physical interaction with NICD or Hey1 [11–14]. This finding is consistent with our results for Jag1-treated murine MSC, which show reduced expression of osteogenic marker genes, ALP, BSP, and Osterix. However, in marked contrast, we found that Jag1-activated Notch signaling was sufficient to induce human MSC osteoblastogenesis, increasing osteogenic marker genes, ALP and BSP, and enhancing ALP activity and mineralization. Intriguingly, the expression level of Osterix and ALP, which are putative down-stream targets of Runx2, was enhanced by Jag1 treatment within 48 hours, which suggests that Runx2 activity is not suppressed by Notch in hMSC, as reported in mMSC [11–14].
This clear difference between murine and human MSC responses to Jag1 treatment is extremely intriguing. One potential explanation for this difference is the greater heterogeneity of murine MSC cultures. While the murine and human purification and culture protocols are essentially the same and both result in multipotential MSC with expression of common cell surface markers, it is well-documented that the murine cultures display significant heterogeneity and contamination with hematopoietic origin cells, particularly CD11b-positive cells [24, 25]. It is conceivable that the presence of these contaminating hematopoietic cells could be directly or indirectly altering the osteoblastic response of the MSC to Jag1.
However, it is also worth noting that despite the similarities between hMSC and mMSC in well-described phenotypic markers, and similar profiles of differentiation, hMSC and mMSC have been shown to be functionally different in a wide-variety of studies. As one example, hMSC and mMSC show a spectrum of cytokine receptor expression that is remarkably different . Similarly, while hMSC highly express high levels of the immunosuppressive molecule indoleamine 2,3-dioxygenase, mMSCs express very little of it, and instead express high levels of inducible nitric oxide synthase, which is not expressed by hMSC . In a similar manner, it is possible that the repertoire of Notch receptors, the extracellular modifications of Notch receptors, or intracellular Notch signaling molecules are different between hMSC and mMSC and thus the cells of the two species respond differently. Additionally, there is strong evidence that murine and human cells respond differently to some osteogenic inducers. While vitamin D3 promotes osteocalcin expression and osteoblast differentiation in human osteoblast progenitors, in murine cells, vitamin D3 inhibits oseteoblastogenesis [28, 29].
Our data show that BMP signaling is likely not involved in Jag1-induced human osteoblastogenesis since Id1, a down-stream target gene of BMP signaling, is not increased after Jag1 treatment. Our findings also show that the presence of 10 μg/ml Jag1 significantly inhibits hMSC proliferation and that hMSCs plated on Jag1 begin to mineralize even in normal growth media (data not shown). This suggests that the cells plated on Jag1 are potently osteoblastic, and this is a possible reason for decreased hMSC proliferation in higher Jag1 doses [23, 30].
Our results show that the Notch inhibitor GSI abrogates Jag1-induced enhancement of ALP and Hes1 expression. Relatedly, dnMAML abolishes Jag1-induced hMSC osteogenesis, while overexpression of the NICD2 is sufficient to induce osteoblastogenesis. However, Hey1 does not seem to be involved in Jag1-induced hMSC osteogenesis, as Hey1 overexpression has minimal impact on osteoblast differentiation. Collectively, these results suggest that Jag1 induces hMSC directly via the canonical CSL-NICD-MAML complex. Consensus CSL-binding sequence CGTGGGAA or CCTGGGAA was analyzed in silica in the ALP promoter region −4,000 to +3,000 . Intriguingly, two potential CSL binding sites are found located around −900 and +870 in the human ALP gene, but these sites are not present in the mouse ALP gene (data not shown). It is possible that differences in expression of osteoblast genes, and subsequent osteoblast differentiation, between human and mouse MSC reflect differences in osteoblast gene promoters.
We used pharmacologic inhibitors to screen for intersecting signaling pathways that could be involved in Jag1-induced osteogenesis. These pathways included Wnt/TGFβ-related CK1 (inhibited with D4476), PKA (inhibited with H89), PKC (inhibited with Gö6976, Bis-I, and Rottlerin), MAPK (inhibited with U0126 for ERK1/2, SB203580 for P38, and PD1683963 for JNK), EGFR (inhibited with SP600125), and PI3K-Akt (inhibited with LY294002). Inhibition of PKC signaling showed a pronounced abrogation of Jag1-induced osteoblastogenesis. In particular, our results show that the PKCδ-specific inhibitor Rottlerin had a pronounced dose-dependent inhibitory effect on osteoblast gene expression and osteoblastogenesis. In contrast to other PKC isoforms, PKCδ has been previously reported to upregulate osteoblastogenesis and bone formation [32, 33] Genetic deletion of PKCδ results in an early bone formation deficit in the mouse embryo, furthermore, PKCδ null MSCs exhibit significantly lower ALP and Osterix expression level and fewer bone nodules than wild-type cells . Correlatively, we observed that Jag1 increased expression of genes such as Osterix and ALP, while Rottlerin treatment reverses this stimulation (data not shown).
PKC is a serine/threonine protein kinase that is known to be involved in multiple cell signal transduction pathways that mediates cellular functions such as proliferation, differentiation, and apoptosis [34, 35] Following stimulation, PKC isozymes translocate from inactive pools to active cell loci. Recent work indicates that active PKC isozymes translocate to a variety of subcellular structures including: membrane vesicles, perinuclear/nuclear structure, and the cytoskeleton .
Intriguingly, we observed that Notch signaling by Jag1 in hMSC rapidly redistributes PKCδ to the nucleus and activates PKCδ kinase activity, but does not change protein levels. To our knowledge, the direct activation and relocalization of PKCδ by Jag1-induced Notch signaling have not been reported previously. We also found that NICD and PKCδ directly interact. This is consistent with the results reported by Kim et al. who demonstrate that overexpression of PKCδ together with NICD results in binding which inhibits proteasomal degradation of NICD . Phosphorylation of intracellular domain of Notch has been shown to be necessary for optimal cleavage and release of NICD, and phosphorylated NICD shows enhanced nuclear translocation [38, 39]. Intriguingly, two serine residues of Ser2152 and Ser2173 in the NICD are putative phosphorylation sites of PKCδ [37, 38]. We hypothesize that activation of PKCδ by Jag1 enhances NICD stability and nuclear translocation through phosphorylation of NICD (Fig. 6C). NICD then acts to promote the direct transcriptional regulation of osteoblast-associated genes.
PKCδ has also been shown to increase Runx2 transcription activity through a physical interaction with Runx2 in FGF2-stimulated MC3T3-E1 preosteoblastic cells . In our study, Jag1 did not directly affect Runx2 gene expression (results not shown), but we did observe that Jag-1 increased expression of genes such as Osterix and ALP, which are reported to be regulated by Runx2, while Rottlerin treatment reverses this stimulation (data did not show). Thus, it is possible that PKCδ also augments Jag1-induced hMSC osteogenesis through PKCδ-activated Runx2 to enhance the expression level of Osterix, which is the direct downstream gene of Runx2 (Fig. 6C). Thus, the enhancement of PKCδ kinase activity associated with Jag1 treatment could also regulate hMSC osteogenesis by increasing Runx2 transcriptional activity .
In a conclusion, treatment of hMSC with Jag1 potently induces osteoblastogenesis. Jag1 results in the release of NICD; NICD interacts directly with PKCδ which is activated by Jag1. Our studies suggest that the processes of Notch regulation of osteoblast differentiation are not fully conserved in mice and humans, but the work supports a novel platform for development of new treatments for ALGS or disease conditions associated with low bone mass, such as age-associated osteoporosis. The generation and characterization of specific small molecules that are capable of regulating Jag1-mediated Notch and/or PKCδ could provide a potent means for activating hMSC osteoblastogenesis.
The authors gratefully acknowledge Dr. Gordon Ruthel for excellent assistance with quantifying the translocation of PKCδ. We thank Hailu Shitaye, Yanjian Wang, Michael Dishowitz, Hui Zhen, Patricia Mutyaba, and Derek Dopkin for assistance with experiments. This work was supported by Department of Defense Award W81XWH-10-1-0825.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.