Laboratory for Molecular Endocrinology (KMEB), Department of Endocrinology and Metabolism, University Hospital of Odense, Odense C, Denmark
Address reprint requests to: Moustapha Kassem, MD, PhD, DSc Laboratory for Molecular Endocrinology (KMEB), Department of Endocrinology and Metabolism, University Hospital of Odense, Winsløwparken 25, 1, DK-5000 Odense C, Denmark
The authors state that they have no conflicts of interest.
Genetic mutations in the LRP5 gene affect Wnt signaling and lead to changes in bone mass in humans. Our in vivo and in vitro results show that activated mutation T253I of LRP5 enhances osteogenesis and inhibits adipogenesis. Inactivating mutation T244M of LRP5 exerts opposite effects.
Introduction: Mutations in the Wnt co-receptor, LRP5, leading to decreased or increased canonical Wnt signaling, result in osteoporosis or a high bone mass (HBM) phenotype, respectively. However, the mechanisms whereby mutated LRP5 causes changes in bone mass are not known.
Materials and Methods: We studied bone marrow composition in iliac crest bone biopsies from patients with the HBM phenotype and controls. We also used retrovirus-mediated gene transduction to establish three different human mesenchymal stem cell (hMSC) strains stably expressing wildtype LRP5 (hMSC-LRP5WT), LRP5T244 (hMSC-LRP5T244, inactivation mutation leading to osteoporosis), or LRP5T253 (hMSC-LRP5T253, activation mutation leading to high bone mass). We characterized Wnt signaling activation using a dual luciferase assay, cell proliferation, lineage biomarkers using real-time PCR, and in vivo bone formation.
Results: In bone biopsies, we found increased trabecular bone volume and decreased bone marrow fat volume in patients with the HBM phenotype (n = 9) compared with controls (n = 5). The hMSC-LRP5WT and hMSC-LRP5T253 but not hMSC-LRP5T244 transduced high level of Wnt signaling. Wnt3a inhibited cell proliferation in hMSC-LRP5WT and hMSC-LRP5T253, and this effect was associated with downregulation of DKK1. Both hMSC-LRP5WT and hMSC-LRP5T253 showed enhanced osteoblast differentiation and inhibited adipogenesis in vitro, and the opposite effect was observed in hMSC-LRP5T244. Similarly, hMSC-LRP5WT and hMSC-LRP5T253 but not hMSC-LRP5T244 formed ectopic mineralized bone when implanted subcutaneously with hydroxyapatite/tricalcium phosphate in SCID/NOD mice.
Conclusions:LRP5 mutations and the level of Wnt signaling determine differentiation fate of hMSCs into osteoblasts or adipocytes. Activation of Wnt signaling can thus provide a novel approach to increase bone mass by preventing the age-related reciprocal decrease in osteogenesis and increase in adipogenesis.
Wnt signaling pathways play an important role in a variety of cellular activities, including cell fate determination, proliferation, migration, polarity, and gene expression.(1) The Wnt/β-catenin pathway, commonly referred to as the canonical Wnt pathway, is initiated through binding of a Wnt ligand to the seven-transmembrane domain-spanning Frizzled receptor and the low-density lipoprotein receptor-related protein 5 and 6 (LRP5/6) co-receptors. On this interaction, cytosolic β-catenin will escape phosphorylation by glycogen synthase kinase (GSK)-3β and subsequent degradation by the ubiquitin/proteasome pathway. The stabilized β-catenin translocates to the nucleus, binds to T-cell factor/lymphoid enhancer binding factor (TCF/LEF) transcription factors, and regulates downstream gene expression.(2)
LRP5 was cloned in 1998 by three independent groups.(3–5) It encodes a single transmembrane protein belonging to the low-density lipoprotein receptor (LDLR) superfamily known for their roles in cholesterol homeostasis and mediating endocytosis of macromolecules.(6) Similar to other LDLR proteins, the extracellular domains of LRP5 contain four EGF-like motifs separated by YWTD spacers and three LDLR ligand binding motifs followed by a single transmembrane domain. The intracellular domain of LRP5 lacks the NPXY internalization motif found in other LDLRs, and instead, possesses five SH3-like domains (PPPSP) characteristic of cytokine and growth factor receptors and essential for triggering the Wnt/β-catenin signal pathway.(7) LRP5 was identified as a potential Wnt receptor when a loss-of-function mutation in the Drosophila ortholog, Arrow, was found to impair signaling through Wingless, the fly ortholog of Wnt-1.(8,9)
In humans, different mutations in LRP5 cause either inactivation or activation of Wnt signaling and lead to osteoporosis (a disease known as osteoporosis-pseudoglioma syndrome [OPPG]) or high bone mass (HBM), respectively.(10–14) In 2001, Gong et al.(10) reported 14 mutations in LRP5 in 28 OPPG families and for the first time showed a relationship between Wnt signaling through LRP5 and bone development. Another large scale LRP5 mutation screening was performed in 37 OPPG probands and revealed new OPPG mutations that reduced Wnt signaling transduction ability.(14) On the other hand, activation mutations of LRP5 have been detected in different patients characterized by HBM, including patients with the HBM phenotype and patients with endosteal hyperstosis, autosomal dominant osteosclerosis, and van Buchem disease.(11–13).
The cellular target for changes in Wnt signaling in bone and the mechanisms leading to changes in bone mass are not clear. Bone marrow-derived human mesenchymal stem cells (MSCs; also known as skeletal stem cells and bone marrow stromal cells) are multipotent cells capable of differentiation into mesoderm-type cells (e.g., osteoblasts and adipocytes).(15) Alterations in the balance between osteoblast and adipocyte differentiation of MSCs can lead to bone loss associated with aging and other clinical conditions.(16) We hypothesized that mutations in LRP5 lead to changes in Wnt signaling affecting the differentiation potential of MSCs into osteoblasts and adipocytes. We studied bone marrow composition in iliac crest bone biopsies obtained from patients with the HBM phenotype and normal controls. In addition, we infected our well-characterized human MSC line (hMSC-TERT)(17,18) with wildtype human LRP5 or two contrasting mutant forms and studied their in vitro and in vivo differentiation potential. Our in vitro and in vivo data showed that mutations in the LRP5 gene and the corresponding levels of Wnt signaling determined the differentiation fate of MSCs into osteoblasts or adipocytes, and consequently, the development of either HBM or low bone mass.
MATERIALS AND METHODS
Bone biopsies obtained from patient with HBM phenotype and controls
The patient population and methods for obtaining and preparation of bone biopsies have been described in details in previous publications.(19,20) We examined bone biopsies from nine patients with a genetically verified HBM phenotype (six men and three women), 26–63 yr of age, and five normal controls (men) without any medications or disorders known to interfere with bone metabolism.(19,20) Tissue sections were viewed at ×40 magnification using an Olympus BX50 microscope with a mounted camera and a motorized stage system (Prior ProScan). Each section was scanned stepwise using the CAST software from Olympus (version 2,3,2,1). Subsequently, the micrographs were imported into Adobe Photoshop (Adobe Systems, Mountain View, CA, USA) and aligned to make a large single micrograph of each tissue section. Quantification of adipose tissue volume (ATV) and trabecular bone tissue volume (TBV) were determined by counting the pixels of each respective tissue based on color difference and expressed as a percentage of the total tissue volume (TV).
pcDND3.1 constructs contained full-length wildtype human LRP5, LRP5 with HBM mutation T253I, or OPPG mutation T244M, kindly provided by Dr Matthew L Warman (Department of Genetics and Center for Human Genetics, Case School of Medicine and University Hospital of Cleveland, Cleveland, OH, USA), as well as Flag-tagged human DKK1 constructs.(14,21) TCF-firefly luciferase reporter plasmid (Topflash) and Renilla luciferase vector pRL-SV40 were purchased from Upstate and Promega, respectively. Retrovirus vector pBABEpuro was obtained from Dr Thomas Jensen (Institute of Human Genetics, University of Aarhus, Aarhus, Denmark). To subclone the wildtype and mutant type of human LRP5 into retrovirus vector pBABEpuro, the pcDNA3.1-LRP5 constructs were cut with EcoRI, XbaI, and PvuI (Promega), and the 5-kb LRP5 fragment was recovered using Promega SV gel and the PCR clean-up system and was blunt-end ligated into the SnaBI site of a pBABEpuro vector. The orientation of the inserts was analyzed by PvuI and SalI (Promega) digestion, and the mutation was confirmed by sequencing.
Cell culture and preparation of conditioned medium
hMSC-TERT were cultured in phenol red-free MEM (Gibco Invitrogen) supplemented with 10% FBS (PAA laboratories) and 1% penicillin/streptomycin (Gibco Invitrogen) as described.(17,18) The phoenix amphotropic packaging cells and HEK293T cells were cultured in DMEM with 10% FBS and 1% penicillin/streptomycin. Wnt3a conditioned medium (Wnt3a-CM) was obtained from an overexpressing Wnt3a stable murine L-Wnt3a cell line (ATCC; CRL-2647) and control conditioned medium (Co-CM) from control L cell line (ATCC; CRL-2648). Cell culture and preparation of the conditioned medium were performed as described by ATCC. To prepare DKK1 conditioned medium (DKK1-CM), HEK293T was transfected by empty pcDNA or pcDNA-DKK1, and the conditioned medium was collected as described by Ai et al.(21)
Retrovirus infection and establishing of stable cell strains
Phoenix cells (70–80% confluent) cultured in 6-cm2 dishes were transfected with either pBABE empty vector or pBABE-LRP5 constructs (10 μg) by the calcium phosphate method. The supernatants containing virus particles were collected 24 and 48 h after transfection, filtered with a 0.45-μm filter, diluted 1:1 with the culture medium, and added to hMSC-TERT cells in 10-cm2 dishes supplemented with 8 μg/ml polybrene for infection. Twenty-four hours after a second round of infection, 1 μg/ml puromycin was added for selection until all control cells were killed. The puromycin-resistant cells were expanded and maintained in MEM medium plus 0.2 μg/ml puromycin.
DNA analysis of transduced cells
The identity of the stably transduced cell strains were verified using PCR-based assays. Genomic DNA was isolated by the DNeasy tissue kit (Qiagen) and treated with DNase-free RNaseA (Sigma). The LRP5 primer pair (forward: 5′-CATCAAGCAGACCTACCTG-3′; reverse: 5′-GTAGAGCTTCCCCTCCTG-3′) and hTERT primer pair (forward: 5′-AGCTGACGTGGAAGATGAGC-3′; reverse: 5′-GCTGAACAGTGCCTTCACC-3′) were used to amplify the exogenous intronless LRP5 and hTERT genes.
Detection of LRP5 proteins and biotinylation of cell surface proteins
LRP5 protein was detected by Western blot analysis. Cells were lysed by RIPA buffer (Sigma), and protein concentration was determined by Coomassie Plus Bradford assay kit (Pierce). For each sample, 30 μg total protein was dissolved by NuPAGE 10% Bis-Tris gel in reducing 3-(N-morpholino) propane sulfonic acid (MOPS) SDS running buffer (Invitrogen) and transferred to polyvinylidene difluoride (PVDF) membrane. The membrane was probed by polyclonal rabbit anti-LRP5 antibody (Zymed Laboratories), incubated with HRP-labeled anti-rabbit IgG (Cell Signaling), and detected by ECLplus Western blotting detection reagents (Amersham Biosciences). To identify the membrane-associated LRP5, we performed biotinylation of cell surface proteins. Cells cultured in 10-cm2 dishes were labeled with 0.5 mg/ml EZ-link Sulfo-NHS-Biotin (Pierce) at room temperature for 30 min and stopped by rinsing with PBS plus 100 mM glycine three times. Cells were lysed, and protein concentration was quantitated as above. A mixture of 200 μg protein and 50 μl immobilized streptavidin (Pierce) in 500 μl RIPA buffer was incubated at 4°C for 1 h. After washing with PBS, the proteins were eluted in boiled SDS-PAGE loading buffer.
Detection of telomerase activity by telomerase repeat amplification protocol
Telomerase activity was quantitated by Telo TAGGG Telomerase PCR ELISA kit (Roche) in triplicate according to the manufacturer's instructions. Absorbance values were recorded as the A450nm reading against reference A690nm, and the mean of the negative control readings was subtracted from those of test samples.
FACS analysis of MSC surface markers
One million cells were stained with PE-conjugated CD44 or CD166 or FITC-conjugated CD29, CD34, CD63, and CD73 antibodies for 60 min at 4°C and then washed twice before FACS analysis (Becton-Dickinson).
Wnt signaling analysis using dual-luciferase reporter assay
Transfection was performed in 24-well plates when cells reached ∼90% confluence by Fugene 6 (Roche). Briefly, 1 h before transfection, cells were incubated with MEM medium without serum and antibiotics. Three hundred nanograms Topflash and 10 ng pRL-SV40 were mixed with 1 μl Fugene 6 in 25 μl OptiMEM and incubated at room temperature for 30 min before being added to the cells. Four hours after transfection, the medium was changed to one of the following media: medium with 50% Co-CM, 50% Wnt3a-CM, or 50% Wnt3a-CM plus either 50% pcDNA-CM or 50% DKK1-CM. Twenty hours after transfection, the dual luciferase reporter assay (Promega) was performed in a luminometer (BMG Labtechnology) according to the manufacturer's instructions.
Cell proliferation assay
Cells were seeded into 96-well plates (n = 6; 4000 cells/cm2). After overnight incubation, the cells were incubated with either normal culture medium or medium containing one of the following: 50% Co-CM, 50% Wnt3a-CM, or culture medium containing 50 ng/ml mouse recombinant Wnt3a (day 0). Cell number was determined using an MTT assay performed at days 0, 2, and 4. Briefly, cells were incubated with medium containing 50 μg thiazole blue tetrazolium bromide (MTT, Sigma) for 4 h, and MTT-formazan crystals were dissolved in 20% SDS and 50% dimethylformamide (pH 7.4). The absorbance was recorded at 570 and 690 nm as reference. The validity of the MTT assay was determined by comparing the results of the MTT assay with direct cell counting using a hemocytometer. In addition, cell number was also determined by the NucleoCounter System (Chemometec). Briefly, one part of single-cell suspension was mixed with one part of lysis buffer and one part of stabilizing buffer. The mixed sample was added into the nucleocassette containing propidium iodide. Cell counting and calculations were performed according to the manufacturer's guidelines.
In vitro hMSC cell differentiation
For osteoblast differentiation, cells were seeded at a density of 104 cells/cm2. After 24 h, the medium was changed to one of the following media: normal culture medium, medium containing 50% Co-CM, 50% Wnt3a-CM, or osteogenic medium (OS; 10 mM β-glycerophosphate, 50 μg/ml 2-phoshate ascorbate, 10 nM dexamethasone, and 10 nM 1,25-dihyrdroxyvitamin D3). The media were renewed every 3 days throughout the study period. For adipocyte differentiation, confluent cells were incubated with adipocyte induction medium (AIM) containing 10% FBS, 10% horse serum, 10 nM dexamethasone, 450 μM 1-methyl-3-isobutylxanthine (IBMX; Sigma-Aldrich), 1 μM rosiglitazone (BRL49653; kindly provided by Novo Nordisk, Bagsvaerd, Denmark), 3 μg/ml human recombinant insulin (Sigma-Aldrich) alone or AIM plus either 10% Co-CM or 10% Wnt3a-CM. Cells were induced for 7 days, and the media were renewed every 3 days.
RT-PCR and real-time PCR
Total RNA was isolated by the GenElute Mammalian Total RNA Kit (Sigma), treated with RNase free DNase (Sigma) to eliminate genomic DNA contamination, and reverse transcribed by the iScript cDNA synthesis kit (Bio-Rad). Real-time PCR analysis of Wnt signaling genes, Dkk1, osteoblastic markers, CbfaI/Runx2 and alkaline phosphatase (ALP), and adipocyte markers, peroxisome proliferator-activated receptor γ (PPARγ2), lipoprotein lipase (LPL), adipocyte-specific fatty acid binding protein (aP2), and adiponectin (APM1), was performed as described previously.(18,22) Primers for RT-PCR analysis of Wnt signaling genes were as described previously.(23,24)
Cytochemical staining and ALP activity assay
ALP staining, Oil red O staining, and ALP activity assay were performed as described previously.(18,22)
In vivo bone formation assay
One million cells were mixed with hydroxyapatite/tricalcium phosphate (HA/TCP; Zimmer Scandinavia, Broendby, Denmark) and implanted subcutaneously into NOD/LtSz-Prkdcscid mice for 8 wk. The transplants were embedded undecalcified in methyl methacrylate, and tissue sections were stained by Goldner's trichrome staining as described previously.(18) Also, other implants were embedded in paraffin after decalcification. Percentage bone volume (mineralized bone and unmineralized bone [osteoid]) per HA volume was determined as described above.
Statistical testing was determined by Student's t-test, and p < 0.05 was considered significant.
Patients with the HBM phenotype exhibit increased bone volume but decreased fat tissue volume in bone marrow
The patients included in this study exhibited typical clinical characteristics of the HBM phenotype,(19) as well as a T253I mutation in the LRP5 gene.(13) Several histomorphometric measurements performed on iliac crest bone biopsies obtained from these patients have been reported previously.(19,25) Compared with controls, these patients exhibit increased cortical width, and consequently, reduction in the fractional width of cancellous bone.(19,25) In addition, the fractional trabecular bone volume and trabecular thickness were also increased.(19,25) Using the same biopsies, we found a significant difference in bone marrow composition between HBM phenotype patients and normal controls. TBV/TV was significantly increased in HBM patients compared with controls (44.9 ± 10.1% versus 17.3 ± 3.0%, p < 0.005). Interestingly, ATV/TV was significantly decreased in HBM compared with controls (3.3 ± 3.1% versus 26.5 ± 8.0%, p < 0.005), suggesting enhanced osteoblastogenesis and impaired adipogenesis of MSCs in vivo (Fig. 1).
Characterization of LRP5-transduced hMSC-TERT cells
To further examine the biological mechanisms mediated by LRP5 and its mutated forms in regulation of hMSC differentiation, human wildtype LRP5, LRP5 with OPPG mutation T244M, or LRP5 with HBM mutation T253I was subcloned into a retroviral vector, pBABEpuro, and introduced into early passage hMSC-TERT (population doubling level [PDL], 106). Infected cells surviving puromycin selection were expanded in medium containing 0.5 μg/ml puromycin until confluent. The cells were cultured and passaged in the presence of standard medium containing 0.2 μg/ml puromycin. The cell strains used in this study were named as hMSC-TERT (parental hMSC cell line), hMSC-pBABE (vector control), hMSC-LRP5WT (wildtype LRP5), hMSC-LRP5T244 (T244M OPPG mutation), or hMSC-LRP5T253 (T253I HBM mutation).
The exogenous intronless LRP5 gene was detected by PCR. As shown in Fig. 2A, the expected band of ectopic LRP5 was detected in hMSC-LRP5WT, hMSC-LRP5T244, and hMSC-LRP5T253 but not in hMSC-TERT and hMSC-pBABE, indicating a stable chromosomal integration of LRP5. Western blot analysis further confirmed that the ectopic LRP5 was expressed highly and equally in hMSC-LRP5WT, hMSC-LRP5T244, and hMSC-LRP5T253 (Fig. 2B). We could not detect endogenous LRP5 in hMSC-TERT and hMSC-pBABE, suggesting that the basal level of LRP5 is very low in our hMSC-TERT cells. To ascertain the level of membrane LRP5, biotin was used to label membrane proteins. Western blot analysis revealed that the membrane trafficking ability of LRP5T244 is reduced dramatically compared with LRP5WT or LRP5T253 (Fig. 2C). Furthermore, because of post-translational glycosylation, membrane LRP5 exhibited a higher molecular weight compared with LRP5 from total cell lysates(Fig. 2C).(21) We did not find abnormal post-translational glycosylation in the ectopically expressed LRP5 or its mutated forms (data not shown).
To determine whether ectopic expression of LRP5 or its mutated forms affect the phenotype of hMSC-TERT cells, we examined the cells for variations in their telomerase activity and in the expression of surface CD markers associated with hMSC phenotype. All five cell strains expressed high and comparable levels of hTERT, confirming that ectopic LRP5 or its mutated forms did not interfere with the activity of the hTERT gene (Figs. 2A and 2D). FACS analysis showed that all of five cell strains expressed the same level of CD29, CD44, CD63, CD73, and CD166 and were negative for hematopoietic marker CD34 (data not shown). Based on these data, hMSC-LRP5WT, hMSC-LRP5T244, and hMSC-LRP5T253 were used in further analysis.
Expression of Wnt signaling components in hMSC-TERT and LRP5-transduced cell strains
RT-PCR was used to determine the expression profile of Wnt ligands, receptors, antagonists, and some of the cytoplasmic and nuclear components of the Wnt signaling pathway in hMSC-TERT and LRP5-transduced hMSC cell strains cultured in control media. Among 19 human Wnts, only Wnt2b, Wnt5a,Wnt5b, and Wnt11 were detected in our cell strains. Fz receptors, Fz1, 2, 4, 5, and 7, and Wnt co-receptors LRP5 and LRP6 were also expressed. We also detected Wnt antagonists such as DKK1 and DKK2, secreted frizzled-related proteins (sFRPs), and some of the cytoplasmic components including adenomatosis polyposis coli (APC), Axin2, Dvl-2, β-catenin, GSK3β, and transcription factor 4 (TCF4) (Table 1). Among these components, the steady-state mRNA levels of DKK1 and DKK2 in hMSC-LRP5WT and hMSC-LRP5T253 were much higher than in hMSC-LRP5T244.
Table Table 1.. Expression Profile of Wnt Components*
LRP5-transduced cell strains exhibited different ability to transduce Wnt signaling
Dual luciferase assay showed that the level of Wnt signaling in the cell strains treated with 50% Co-CM was very low (Fig. 3A), and these levels were similar to those detected in cells cultured in normal culture medium (data not shown). Thus, it can be considered as the basal level of Wnt signaling in these cell strains. Using 50% Wnt3a-CM, Wnt signaling increased 24- and 20-fold in hMSC-LRP5WT and hMSC-LRP5T253, respectively, but only 9-fold in hMSC-LRP5T244 (Fig. 3A) compared with the basal level of Wnt signaling. To further confirm the normal biological responses of the cells, we found that Wnt signaling activation was inhibited by adding DKK1-CM in all of cell strains (Fig. 3B).
LRP5T253 mutation inhibited cell proliferation and was associated with decreased DKK1 expression
To determine the effect of the presence of wildtype LRP5 or its mutated forms on hMSC proliferation, we studied the effects of adding Wnt3a on the proliferation of hMSC-LRP5WT, hMSC-LRP5T244, and hMSC-LRP5T253 cell strains by MTT assay and by determining cell number using the NucleoCounter system. As shown by the MTT method, treatment with 50% Co-CM or Wnt3a-CM for 2 days reduced proliferation rate of hMSC-LRP5WT and hMSC-LRP5T253 (p < 0.01 and 0.05, respectively). Proliferation rate was unchanged in hMSC-LRP5T244 (Fig. 4A). The inhibitory effects persisted in hMSC-LRP5WT and hMSC-LRP5T253 (p < 0.01 for both) but still was not apparent in hMSC-LRP5T244 after 4 days of treatment (Fig. 4B). To confirm these results and to avoid the effects of the interaction of Wnt with “unknown factors” present in the condition medium, cells were treated with 50 ng/ml mouse recombinant Wnt3a for 2 or 4 days. Similar effects on cell proliferation were observed after 4 days (data not shown). Cell number determined by the NucleoCounter method confirmed that cell number in hMSC-LRP5WT and hMSC-LRP5T253 was significantly reduced after Wnt3a treatment for 2 days (p < 0.01; Fig. 4C). FACS analysis of propidium iodide-stained cells revealed absence of sub-G1 peaks, suggesting that the decreased cell number was not caused by enhanced apoptosis (data not shown).
It has been previously reported that the expression of Wnt antagonist DKK1(26,27) is required for maintaining hMSCs in the cell cycle.(28) Thus, we analyzed its expression profile in hMSC-LRP5WT, hMSC-LRP5T244, and hMSC-LRP5T253. As shown in Fig. 4D, basal expression level of DKK1 in proliferating hMSC-LRP5WT and hMSC-LRP5T253 (cells cultured in normal culture medium) was significantly higher than in hMSC-LRP5T244 (p < 0.01). Furthermore, the expression level of DKK1 was decreased by 60% in Wnt3a-CM-treated hMSC-LRP5WT and hMSC-LRP5T253 compared with Co-CM-treated cells (Fig. 4D). The DKK1 expression level in hMSC-LRP5T244 did not decrease with Wnt3a-CM compared with Co-CM treatment (Fig. 4D).
LRP5T253 mutation enhances in vitro osteoblast differentiation of hMSCs
We examined steady-state gene expression of osteoblast-specific markers in Wnt3a-treated hMSC-LRP5WT, hMSC-LRP5T244, and hMSC-LRP5T253. The expression of CbfaI/Runx2 in cells treated with Co-CM was low and similar to that of cells treated with normal culture medium. CbfaI/Runx2 expression in hMSC-LRP5WT and hMSC-LRP5T253 was significantly increased by treating the cells with Wnt3a-CM for 3 days (p < 0.01 and 0.05, respectively; Fig. 5A). In contrast, culturing hMSC-LRP5T244 in the presence of Wnt3a-CM for 5 days led to only a slight insignificant increase in CbfaI/Runx2 expression (Fig. 5A). When normalized to expression levels in Co-CM treatment, CbfaI/Runx2 expression was increased by Wnt3a: 1.6-fold in hMSC-LRP5T244, 3.3-fold in hMSC-LRP5WT, and 2.3-fold in hMSC-LRP5T253 (Fig. 5A). The expression of ALP was analyzed by a cytochemical staining method. As shown in Fig. 5B, ALP was increased by Wnt3a treatment in hMSC-LRP5WT and hMSC-LRP5T253 but not in hMSC-LRP5T244 compared with Co-CM treatment, and both the number of ALP+ cells and their staining intensity in hMSC-LRP5WT and hMSC-LRP5T253 was increased compared with hMSC-LRP5T244 (Fig. 5B). To confirm these results, mRNA expression level of ALP after treatment with Co-CM or Wnt3a-CM for 7 days was quantitated by real-time PCR. Steady-state ALP gene expression was increased by 3- or 3.6-fold after Wnt3a treatment in hMSC-LRP5WT and hMSC-LRP5T253, respectively, but only 1.6-fold in hMSC-LRP5T244 (Fig. 5C). This ALP induction pattern was detectable as early as 3 days after treatment (data not shown). Similarly, ALP enzyme activity was induced by Wnt3a in hMSC-LRP5WT and hMSC-LRP5T253: 2.5- and 4.1-fold, respectively, but only 1.5-fold in hMSC-LRP5T244 (Fig. 5C).
LRP5T253 mutation inhibited in vitro adipocyte differentiation of hMSCs
Treating hMSCs with AIM for 7 days resulted in adipocyte differentiation, shown by the presence of high levels of mRNA expression of both early and late adipocyte differentiation markers PPARγ2,LPL, aP2, and APM1. Notably, expression levels of these markers in hMSC-LRP5T244 were significantly higher than those observed in hMSC-LRP5WT and hMSC-LRP5T253 (p < 0.05). These results suggest that Wnt signaling inhibited adipocyte differentiation of hMSCs (Fig. 6B). To confirm these results, Co-CM or Wnt3a-CM (10%) were mixed with AIM, and the effects on adipocyte differentiation were examined. Oil red O staining for mature adipocytes containing lipid droplets was reduced in hMSC-LRP5WT and hMSC-LRP5T253 cultured in AIM plus 10% Co-CM compared with hMSC-LRP5T244 (Fig. 6A). The adipocyte differentiation of hMSC-LRP5WT and hMSC-LRP5T253 could be efficiently blocked by adding Wnt3a-CM (10%); however, hMSC-LRP5T244 was resistant to these effects (Fig. 6A, insets). Real-time PCR analysis confirmed that the expression of adipocyte-specific differentiation markers on Wnt3a-CM treatment was significantly decreased in hMSC-LRP5WT and hMSC-LRP5T253 compared with the levels observed in hMSC-LRP5T244 (Fig. 6B).
LRP5T253 mutation enhanced in vivo bone formation
To confirm the observed in vitro effects on hMSC differentiation, we examined the in vivo bone formation capacity of hMSC-LRP5WT, hMSC-LRP5T244, and hMSC-LRP5T253 after subcutaneous implantation of cells mixed with HA/TCP scaffold in immune-deficient mice. hMSC-LRP5WT and hMSC-LRP5T253 formed high amounts of heterotopic bone compared with hMSC-LRP5T244 (Figs. 7A–7C). The bone formed was of human origin as shown by staining with human specific antibodies (data not shown).(17) Furthermore, quantitative analysis of the amount of bone formed revealed that similar amounts were formed by parental cell line (hMSC-TERT), and hMSC-LRP5T244 and bone formation was enhanced by hMSC-LRP5WT and hMSC-LRP5T253. Interestingly, Goldner trichrome staining of undecalcified implants revealed that most of the bone formed in parental cell line (hMSC-TERT) and hMSC-LRP5T244 was osteoid (unmineralized bone), whereas hMSC-LRP5WT and hMSC-LRP5T253 formed both osteoid and mineralized lamellar bone (Fig. 7D).
During the last few years, the Wnt/β-catenin signaling pathway has been increasingly recognized as a main regulatory pathway controlling bone mass accrual and turnover. This is based on the identification of mutations in LRP5 as the cause of rare inherited bone diseases involving decreased or increased bone mass. Inactivation mutation in LRP5 leading to decreased Wnt signaling results in low bone mass and osteoporosis, and activation mutation in LRP5 leading to increased Wnt signaling results in high bone mass and osteopetrosis phenotype. The role of Wnt signaling in control of bone mass has been further strengthened by establishing murine models mimicking these two human diseases.(29,30) Also, several murine models that either alter Wnt ligands,(31) Wnt antagonists,(32) or β-catenin turnover(33,34) exhibit a bone phenotype and thus confirm the importance of Wnt signaling pathway for bone mass control. Despite the clear effects of changes of Wnt signaling on bone mass in vivo, the mechanism whereby Wnt signaling regulates bone mass is still unclear. In this study, we showed that patients with the HBM phenotype exhibit high trabecular bone mass and reduced fat mass in the bone marrow. Also, hMSCs carrying the HBM phenotype mutation T253I in LRP5 exhibited increased Wnt signaling activity, enhanced in vitro osteoblast but not adipocyte differentiation, and enhanced in vivo bone formation. hMSCs carrying the OPPG mutation T244M in LRP5 exhibited the opposite phenotype.
We found that both overexpression of wildtype and HBM LRP5 mutation T253I enhanced Wnt signaling, whereas OPPG LRP5 mutation T244M impaired Wnt signaling, thus confirming previous findings in a variety of cell types.(14,21) The molecular basis for changes in Wnt signaling caused by the LRP5 mutation is under study. A proposed mechanism for HBM mutations is that it disrupts the interaction between the mutated LRP5 and Mesd (mesoderm development in mouse), a chaperone protein required for transport of LRP5/6 co-receptors to cell surfaces.(35) Alternatively, the HBM mutation may decrease functional activity of Wnt antagonists (e.g., resistant to DKK1(21) or SOST inhibition.(36)) In hMSC-LRP5WT and hMSC-LRP5T253, DKK1 inhibited Wnt signaling, suggesting that hMSC-LRP5T253 cells are not resistant to the effects of DKK1 and resistance to DKK1 may not be the only mechanism for enhanced Wnt signaling. Actually, in the study of Ai et al.,(21) HBM LRP5 mutation T253I was shown to be resistant to DKK1 inhibition, but this inhibition was only apparent in the presence of Wnt10b or Wnt1. When Wnt3a (as in the case of our study) was used as a ligand, the inhibitory effects were not clear, suggesting that DKK1 resistance may be ligand-dependent.
We found that activation of Wnt signaling by Wnt3a led to a decreased proliferation rate of hMSCs, and this was not indirectly caused by enhanced apoptosis. Generally, canonical Wnt signaling is linked to increased cell proliferation through activating of cell cycle genes (e.g., c-Myc or cyclin D1).(37) Also, Wnt signaling is known to regulate stem cell self-renewal in some stem cell types, including embryonic stem cells, intestinal stem cells, skin stem cells, and hematopoietic stem cells.(38) In hMSCs, canonical Wnt signaling has also been reported to stimulate cell proliferation.(39,40) Thus, our results seem to be inconsistent with previous observations. These discrepancies may be related to cell types, culture conditions, differentiation state of cells, or the make up of Wnts and antagonists of Wnt signaling in the pericellular microenvironment. For example, we observed significant differences in the levels of gene expression between Wnts and antagonists of Wnt signaling in our cell lines compared with that reported in other pre-osteoblastic and hMSC cultures.(39) In the cell lines used in our experiments, most of Wnt ligands were not expressed, and low basal Wnt signaling was detectable. Also, DKK1 was highly expressed in proliferating hMSC-LRP5WT and hMSC-LRP5T253, but lower in hMSC-LRP5T244, suggesting an autocrine mechanism aiming at maintaining Wnt signaling at low basal levels. On the other hand, activation of canonical Wnt signaling by Wnt3a led to downregulation of DKK1 in hMSC-LRP5WT and hMSC-LRP5T253, but not in hMSC-LRP5T244, suggesting that Wnt signaling controls cell proliferation in an autocrine or paracrine fashion based on antagonism of Wnt signaling by DKK1. Our results corroborate previous findings showing that DKK1 is required for hMSC entry into the cell cycle, and consequently, cell proliferation, through its inhibitory effects on Wnt signaling(28) and consistent with a previously proposed model for Wnt signaling during endochondral bone formation, where canonical Wnt signaling inhibits proliferation and accelerates chondrocyte differentiation.(41)
We observed significant changes in the bone marrow composition in patients with HBM, with increased amount of trabecular bone and reduced fat tissue volume compared with controls. This suggests that LRP5 status and/or Wnt signaling may function as a molecular switch determining the cell lineage fate of hMSCs. By using in vitro hMSC cell models, we confirmed that activation of Wnt signaling enhanced osteoblast and inhibited adipocyte differentiation. The mechanism whereby Wnt signaling regulated bone mass is not clear and most probably involves several levels of control.(31–34) In mice deficient in LRP5, low bone mass phenotype was observed because of reduced proliferation of precursor cells and decreased bone matrix deposition.(29) In mice overexpressing HBM LRP5 mutation G171V, high bone mass was associated with enhanced osteoblast activity and reduced osteoblast apoptosis.(30) Some recent studies have also suggested that enhanced Wnt signaling impaired osteoclast activity caused by enhanced production of osteoprotegerin(42) and in vitro cultures of osteoclastic cells obtained from patients with HBM mutation T253I exhibited impaired activity.(43) Also, Wnt signaling has been suggested to be a target for normal physiological response of mechanical loading in bone,(44) and thus enhanced Wnt signaling, as observed in HBM phenotype patients and HBM mouse models, may be caused by an exaggerated physiological bone response to mechanical strain.(44–46)
Our findings suggest a novel mechanism for Wnt signaling control of bone mass. Activation of Wnt signaling because of genetic changes in LRP5 leads to changes in bone mass caused by switching of the differentiation of hMSCs into osteoblastic cells instead of adipocytic cells. These results were further confirmed by the findings in hMSC-LRP5T244 cells, where impaired Wnt signaling was associated with enhanced adipogenesis and impaired bone formation. Aging and osteoporosis are associated with decreased bone mass and increased adipose tissue volume in the bone marrow,(47) and enhanced adipogenesis and decreased bone formation of MSCs are possible underlying mechanisms.(16) Our findings support the validity of this concept and suggest that changes in bone mass during normal aging and in age-related osteoporosis may be caused by alterations in Wnt signaling. However, further studies are needed to confirm this hypothesis.
Several possible molecular mechanisms can explain the basis for Wnt signaling in determining the differentiation fate of MSCs. For example, Runx2 had been reported as a Wnt target gene.(48) Activation of Wnt signaling in ST2 bipotential cell line led to suppression of gene expression of C/EBPα and PPARγ and increased expression of Runx2 and osterix.(49) Also, inhibition of Wnt signaling induced expression of adipogenic transcription factors C/EBPα and PPARγ and stimulated adipogenesis.(50,51) In addition, it has been shown that the balance between PPARγ and β-catenin is regulated by GSK3β, and this in turn determines adipogenesis.(52) Recently, upregulation of DKK1 and downregulation of LRP5/6 co-receptors have been reported to be coordinately regulated to promote early adipogenesis.(53)
In addition to the effects of Wnt signaling on enhancing osteoblast differentiation of hMSCs, we observed that Wnt signaling enhanced late stages of maturation of osteoblastic cells as shown by increased mineralized bone formation in vivo by hMSC-LRP5WT and hMSC-LRP5T253. Currently, knowledge regarding the function of Wnt signaling in late stages of osteoblast differentiation and matrix mineralization is limited. DKK1 (a Wnt antagonist) has been reported to induce a dose-dependent suppression of osteoblast matrix mineralization.(54) In DKK2-deficient mice, an osteopenic phenotype has been observed caused by decreased osteoblastic ability for matrix mineralization.(55) Also, in this mouse model, canonical Wnt signaling has been shown to enhance expression of DKK2 in differentiated osteoblastic cells.(55) We observed that the parental hMSC-TERT and hMSC-LRP5T244 exhibited reduced ability to form mineralized bone in vivo, and this was associated with low expression levels of DKK2, which is in contrast with high DKK2 levels in hMSC-LRP5WT and hMSC-LRP5T253 (data not shown). Thus, Wnt signaling may affect both osteoblast commitment and osteoblast functions. These effects could be regulated independently in autocrine or paracrine fashions through changes in a number of Wnt antagonists and agonists acting on cell surface (e.g., DKK1 and DKK2). This hypothesis may have clinical implications in selective targeting of DKK1 or DKK2 and other Wnt signaling modulators to elicit specific effects on bone mass.
In conclusion, LRP5 mutations and the level of Wnt signaling determine the differentiation fate of hMSCs into osteoblasts or adipocytes. Further studies, exploring the mechanism underlying the biological effects of LRP5 mutations and how Wnt signaling regulates hMSC biology, including possible cross-talk with other signaling pathways, will provide important information needed for possible targeting of hMSC to increase bone formation with an anabolic therapy for bone loss states.
This work was supported by grants from the Danish Medical Research Council, The Danish Stem Cell Center (DASC), the Novo Nordisk foundation, and a grant from the County of Funen, Denmark, as well as a PhD fellowship from Novo Nordisk foundation, the Nordic Network of Endocrinology (TEA). The authors thank Hongbing Zhang and Lars Grøntved for helping in retrovirus infection experiments and Martin Rasmussen and Lone Christiansen for excellent technical assistance. We also thank Jorge S Burns for grammatical control.