Disclosure of potential conflicts of interest is found at the end of this article.
Author contributions: G.D.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; D.C.H.: collection and assembly of data, data analysis; G.D. and D.C.H. contributed equally to the manuscript; L.G.S.M.: data analysis and interpretation, manuscript writing; R.K.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; M.M.: collection and assembly of data, data analysis and interpretation; D.R.: collection and assembly of data, data analysis; X.F.Y.: collection and assembly of data, data analysis.
First published online in STEM CELLSExpress December 18, 2008.
Interferon (IFN)γ is a strong inhibitor of osteoclast differentiation and activity. However, its role in osteoblastogenesis has not been carefully examined. Using microarray expression analysis, we found that several IFNγ-inducible genes were upregulated during early phases of osteoblast differentiation of human mesenchymal stem cells (hMSCs). We therefore hypothesized that IFNγ may play a role in this process. We first observed a strong and transient increase in IFNγ production following hMSC induction to differentiate into osteoblasts. We next blocked this endogenous production using a knockdown approach with small interfering RNA and observed a strong inhibition of hMSC differentiation into osteoblasts with a concomitant decrease in Runx2, a factor indispensable for osteoblast development. Additionally, exogenous addition of IFNγ accelerated hMSC differentiation into osteoblasts in a dose-dependent manner and induced higher levels of Runx2 expression during the early phase of differentiation. We next examined IFNγ signaling in vivo in IFNγ receptor 1 knockout (IFNγR1−/−) mice. Compared with their wild-type littermates, IFNγR1−/− mice exhibited a reduction in bone mineral density. As in the in vitro experiments, MSCs obtained from IFNγR1−/− mice showed a lower capacity to differentiate into osteoblasts. In summary, we demonstrate that the presence of IFNγ plays an important role during the commitment of MSCs into the osteoblastic lineage both in vitro and in vivo, and that this process can be accelerated by exogenous addition of IFNγ. These data therefore support a new role for IFNγ as an autocrine regulator of hMSC differentiation and as a potential new target of bone-forming cells in vivo. STEM CELLS2009;27:550–558
Maintenance of skeletal integrity is a complex phenomenon characterized by timely regulation between bone formation and bone resorption . Osteoblast formation occurs within the bone marrow by differentiation of pluripotent mesenchymal stem cells (MSCs) [2–4]. Regulation of osteoblastogenesis is critical in the acquisition and maintenance of bone mass throughout life [1, 2, 4]. MSCs obtained from the bone marrow and exposed to ascorbic acid in vitro produce a collagenous extracellular matrix and express specific genes associated with osteoblastic phenotypes, such as alkaline phosphatase (ALP), osteopontin (OPN), and osteocalcin (OCN) . The transcription factors involved in the process of differentiation of MSCs into osteoblasts are a subject of intense research since their identification could provide us with potential targets for bone formation. Microarray studies of osteoblast differentiation in mouse and human models in vitro have identified novel transcription factors that may be important in the establishment and maintenance of differentiation . Most models are based on temporal changes postinduction of MSCs that include a proliferative phase (week 1), a collagen matrix deposition phase (week 2), and a mineralization phase (week 3) .
In this study, we initially used human (h)MSCs that display a stable phenotype, remain monolayer in vitro, and can be induced to differentiate into adipocytes, chondrocytes, or osteoblasts . We performed a functional analysis of gene expression changes when hMSCs are induced to differentiate into osteoblasts. Our data indicated that several interferon (IFN)γ-inducible genes were expressed transiently during the proliferative phase by hMSCs undergoing osteoblastic differentiation in vitro. Considering that there is a known interplay between interferon and bone cells, mostly with the osteoclast lineage [1, 7], and that the role of IFNγ in osteoblast differentiation and function remains unknown, we assessed the role of IFNγ in osteoblastogenesis both in vitro and in vivo.
In this study, we looked at IFNγ production when hMSCs were induced to differentiate into osteoblasts and at the consequences of disrupting IFNγ production by small interfering (si)RNA in hMSCs prior to their commitment to osteoblastogenesis. We also examined the effect of exogenous addition of IFNγ to cultured cells as well as the expression of transcription factors required in osteoblast differentiation [8, 9]. Finally, we tested whether, as in our in vitro experiments, the absence of IFNγ signaling affects osteoblastogenesis in vivo. Our data indicate that IFNγ plays an important role in hMSC commitment to osteoblastogenesis both in vitro and in vivo.
MATERIALS AND METHODS
Osteogenic Differentiation of hMSCs In Vitro
The induction of osteogenic differentiation of hMSCs was described previously . Briefly, hMSCs (BioWhittaker, Walkersville, MD, http://www.cambrex.com) were plated at a density of 5 × 105 cells per well in 150-cm2 dishes containing MSC growth medium (GM) (BioWhittaker) with 10% fetal calf serum (FCS) and incubated at 37°C. After the cells reached 60% confluence, the medium was replaced with either GM or osteoblastogenesis induction medium (OM) (prepared with MSCGM, 10% FCS, 0.2 mM dexamethasone, 10 mmol/l β glycerol phosphate, and 50 μg/ml ascorbic acid) for 21 days. The medium was changed every 3 days. Media obtained at the beginning of the first, second, and third week of differentiation as well as conditioned media were collected for measurement of IFNγ. Concentrations of IFNγ in the conditioned media were measured from both GM- and OM-treated cells using the Human NT-4 DuoSet ELISA Development Kit (R&D Systems, Minneapolis, http://www.rndsystems.com).
cDNA Microarray Analysis of hMSCs
Total RNA was extracted from hMSCs treated with either GM or OM after the first and third week of differentiation using an Easy-Kit miniprep (Qiagen, Valencia, CA, http://www1.qiagen.com). Generation of cDNA, fluorescent labeling, hybridization to the gene chip, and data analysis were performed by the Genomics Laboratory at McGill University as previously described . We examined 12,000 human genes and expressed sequences tags on the array Human Genome U95A (Affymetrix, Inc., Santa Clara, CA, http://www.affymetrix.com) and analyzed the results using the MicroDB Software (Affymetrix). Expression values of the differentiated and nondifferentiated hMSCs were compared using Student's t-test. Genes with significant changes were then grouped depending on their known function. The biological function of each gene product was obtained from literature searches in medical databases. This experiment was repeated twice and significant changes in gene expression were determined by the method of biological duplicates as previously described [11, 12].
Treatment of hMSCs with IFNγ Under Conditions of Osteoblastic Differentiation
hMSCs were plated in 4-cm2 dishes containing GM at a density of 4 × 104 cells per dish. After 48 hours, the medium was replaced with GM or OM containing either IFNγ (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) at different concentrations (1, 10, and 100 ng/ml) or vehicle. The medium was changed every 3 days. After 1 week of treatment, the medium was aspirated and cells were stained for both ALP, using TT-blue+ (Sigma-Aldrich), and mineralization, using Alizarin Red (Sigma-Aldrich). The number of ALP+ cells per field was quantified in 10 different fields per well. Matrix mineralization was quantified by extracting the Alizarin Red staining with 100 mM cetypyridinium chloride (Sigma) at room temperature for 3 hours. The absorbance of the extracted Alizarin Red S staining was measured at 570 nm. Data represented are expressed as units of Alizarin Red per mg of protein in each culture normalized to the number of cells per well.
IFNγ Knockdown by siRNA
Knockdown of the IFNγ was obtained by gene siRNA in differentiating hMSCs as previously described . Briefly, hMSCs were grown in MSCGM containing 10% FCS until 60% confluence in six-well plates. The medium was then removed and replaced with serum-free MSCGM and transfected with 300 nmol of siRNA using the siRNA oligo transfection kit (sc-29528; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com) according to the manufacturer's directions. We used a double-stranded siRNA oligonucleotide against human IFNγ (sc-39606) and a negative control siRNA (sc-37007) from Santa Cruz Biotechnology Inc. Cells were incubated for 5-8 hours in serum-free MSCGM and the medium was then replaced with OM. siRNA transfection was repeated every 3 days and cells were treated for up to 14 days. Specific siRNAs directed against human IFNγ were a pool of the following three separate strands: sense strand (a): 5′-CGAAGAGCAUCCCAGUAAUtt-3′ (position 616-634), sense strand (b): 5′-CUGUGACUGUCUCACUUAAtt-3′ (position 807-826), and sense strand (c): 5′-GCAAGGCUAUGUGAUUACAtt-3′ (position 849-867) (Fig. 2B). The positions of the siRNA strands were obtained through GenBank mRNA accession number NM-000619.2. Their specificity was verified in the nonredundant human DNA database using a Basic Local Assignment Search Tool algorithm (accessed through the National Center for Biotechnology Information). Control siRNA did not lead to any specific degradation of known cellular mRNA and was selected because it exhibited no cellular toxicity.
At the time points indicated, cells were collected and washed twice in phosphate-buffered saline (PBS). Poly(A)+ mRNA was isolated with the QuickPrep Micro mRNA purification kit according to the manufacturer's specifications (GE Healthcare Bio-Sciences Inc., Baie d'Urf<1272>, QC, Canada, http://www.gehealthcare.com), dissolved in diethylpyrocarbonate-treated water, and subjected to DNAse I treatment. Two μg of poly(A)+ mRNA was reverse transcribed using Qiagen One Step reverse transcription-polymerase chain reaction (RT-PCR) Enzyme Mix (Qiagen). The resulting cDNA was amplified by 35 PCR cycles with an annealing temperature of 58°C. Oligonucleotide primers for IFNγ amplification were obtained from Santa Cruz Biotechnology Inc.. The predicted size of the IFNγ product was 545 bp (Fig. 2B). Genes were randomly selected from the list of genes with a significant change after the first and third week of differentiation (see supporting information). Primers were synthesized by Alpha DNA Inc. (Montreal, QC, Canada, http://www.alphadna.com). Selected primers for the first week included IFNγ-inducible protein 35 (IFI 35) (5′-CTCTGCTCTGATCACCTTTGATCAC-3′ upstream and 5′-GCTTCTGGAAGTGGATCTCCAGGA-3′ downstream), interleukin 10 receptor (5′-GAACCTGACTTTCACAGCTCAGTAC-3′ upstream and 5′-TCAGGTGCTGTGGAAGAGAATTC-3′ downstream), growth-related oncogene 1 (5′-GAACATCCAAAGTGTGAACGTGAAG-3′ upstream and 5′-ATTTGCTTGGATCCGCCAGCCTCTA-3′ downstream), and importin (5′-GAGAATTGCAGAATTGGCCTGACCT-3′ upstream and 5′-CTATGTCTGAGTACTTCATGCCA-3′ downstream). For the third week, selected primers included: transforming growth factor β receptor 2 (5′-TACATCGAAGGAGAGCCATTCGC-3′ upstream and 5′-TGCAGCACACTCGATATGGACCAG-3′ downstream), LDL receptor related protein five (5′-GTCGTAGTCGATGGCAATGGCGT-3′ upstream and 5′-ACGGACTCAGAGACCAACCGCATC-3′ downstream), OPN (5′-ACTCTGGTCATCCAGCTGACTCGT-3′ upstream and 5′-CTCCTAGGCATCACCTGTGCCATA-3′ downstream), OCN (5′-TGGCCGCACTTTGCATCGCTGG-3′ upstream and 5′-CGATAGGCCTCCTGAAAGCCGATG-3′ downstream), and runt-related transcription factor 2 (Runx2) (5′-TGGCCGCACTTTGCATCGCTGG-3′ upstream and 5′-CGATAGGCCTCCTGAAAGCCGATG-3′ downstream).
Amplification of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or β-actin was used as a control. The expected sizes of the amplification products were between 400 and 800 bp. Amplified products were analyzed by either 1.5% or 2% agarose gel electrophoresis. The signals were quantified by densitometry (Bio-Rad, Hercules, CA, http://www.bio-rad.com) and normalized according to GAPDH or β-actin density.
Western Blot Analysis
Cells were collected at the times indicated, lysed, separated by gel electrophoresis, and then transferred to a nitrocellulose membrane. After blocking, the membrane was incubated overnight at 4°C using an antibody directed against Runx2, IFNγ, OCN, IFNγ receptor 1 (IFNγR1) (Santa Cruz Antibodies), or tubulin (Sigma), and the bound antibodies were detected with the corresponding secondary antibodies conjugated with horseradish peroxidase. Blots were developed by enhanced chemiluminescence using Lumi-GLO reagents (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD, http://www.kpl.com). The signals were quantified by densitometry and expression ratios were normalized according to tubulin density.
Cell Viability and Proliferation Assays
To assess whether treatment of differentiating hMSCs with either IFNγ or IFNγ knockdown had an effect on their proliferation and survival, cells were plated onto 24-well cultures and treated with either with IFNγ or IFNγ siRNA as previously described. After treatment, cells were released by 0.25% vol/vol trypsin-EDTA treatment and resuspended in 10% FCS-containing medium. After centrifugation (1,000g for 5 minutes) cells were resuspended in 10% vol/vol FCS-containing medium mixed with 0.4% w/v trypan blue in 1:1 ratio, and a viable cell count was performed using a hemacytometer (nonviable cells were stained with trypan blue).
For the proliferation analysis, MSCs were seeded at a density of 4 × 102 cells/well in 96-well cluster plates (Falcon; Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com). Cells were induced to differentiate and treated with either IFNγ or IFNγ siRNA as previously described. Cell proliferation was quantified by 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS)-formazan analysis (CellTiter 96 AQueous Cell Proliferation Assay, Promega, Madison, WI, http://www.promega.com). Briefly, a stock solution of MTS was dissolved in PBS at a concentration of 5 mg/ml and was added in a 1:10 ratio (MTS/Dulbecco's modified Eagle's medium [DMEM]) to each well and incubated at 37°C for 2 hours, and the optical density was determined at a wavelength of 490 nm on a microplate reader model 3550 (Bio-Rad). In preliminary experiments, the absorbance was found to be directly proportional to the number of cells over a wide range (2 × 102 to 5 × 104 cells/well). The percent proliferation was defined as [(experimental absorbance − blank absorbance)/control absorbance − blank absorbance)] × 100, where the control absorbance is the optical density obtained for 1 × 104 cells/well (number of cells plated at the start of the experiment), and blank absorbance is the optical density determined in wells containing medium and MTS alone.
We purchased IFNγR1−/− mice (strain name 129-Ifngrtm1 on a C57BL/6 background) from Jackson Laboratory (Bar Harbor, ME, http://www.jax.org). As a control strain, we purchased the strain 129S1/SvImJ on a C57BL/6 background from the same source. Mice were housed in cages in a limited access room. Animal husbandry adhered to Canadian Council on Animal Care Standards, and all protocols were approved by the McGill University Health Center Animal Care Utilization Committee.
Bone Mass Measurements by Dual Energy X-Ray Absorptiometry
Hip and spine bone mineral density (BMD) were measured at 4, 8, and 12 weeks using a PIXIMUS bone densitometer (GE Medical Systems). A quality control phantom was used to calibrate the densitometer prior to each experiment.
Ex Vivo Cultures of Bone Marrow Cells
Bone marrow cells were prepared and induced to differentiate into osteoblasts as previously described . Briefly, tibiae from one side of 4- and 8-week-old IFNγR−/− and IFNγR+/+ mice (n = 12 per group per time point) were flushed using a 21-gauge needle attached to a 10-ml syringe filled with DMEM (Gibco-BRL, Gaithersburg, MD, http://www.gibcobrl.com). The bone marrow cells were filtered through a cell strainer with a 70-μm nylon mesh (BD Biosciences, Bedford, MA, http://www.bdbiosciences.com) and plated in 10-cm2 tissue culture dishes. The cells were incubated in GM at 37°C with 5% humidified CO2 and isolated by their adherence to tissue culture plastic. The medium was aspirated and replaced with fresh medium every 2-3 days to remove nonadherent cells. Adherent MSCs were grown to confluency for about 7 days and defined as MSCs at passage 0, harvested with 0.25% trypsin and 1 mM EDTA for 5 minutes at 37°C, diluted 1:3 in GM, plated, and grown to confluency for further expansion. After the second and third passages, MSCs were used for subsequent experiments.
To induce differentiation, a total of 104 cells were diluted in OM and plated in 24 dishes per group, each 4 cm2. The medium was aspirated and replaced with fresh OM every 3 days. At 21 days, the medium was removed and cultures were fixed in 10% vol/vol formol/saline solution for 5 minutes. Colony-forming units-osteoblasts (CFU-OB) were detected by Alizarin Red (pH 7.4) staining. The total number of CFU-OB per dish was counted macroscopically with a flatbed scanner fitted with a transparency adapter.
All data are expressed as mean ± standard deviations of three replicate determinations. Unless otherwise stated, all experiments were repeated three times. Statistical analysis was performed by one-way analysis of variance or Student's t-test. A probability value of p < .05 was considered statistically significant.
IFNγ-Inducible Genes Are Upregulated During hMSC Differentiation into Osteoblasts In Vitro
We exposed hMSCs to either GM or OM for a total of 3 weeks. Subsequently, we examined 12,000 human genes in the array Human Genome U95A (Affymetrix, Inc.) to assess gene expression during the proliferative (week 1) and mineralization (week 3) phases of osteoblastic differentiation. The scatter plots of gene expression levels were determined for each time point at week 1 and week 3 by averaging mRNA expression levels from GM- and OM-exposed cells plotted in a double-logarithmic scale (see supporting information). Genes identified as differentially expressed between GM- and OM-treated cells had a minimum of a 2.5-fold increase or decrease in gene expression. At weeks 1 and 3, a subset of markers of early osteoblast differentiation [5, 15–20] as well as markers of mature osteoblasts [20, 21] were differentially expressed in response to OM (Fig. 1). In a set of 103 genes during the proliferative phase (week 1), a predominant expression of markers of early osteoblast differentiation was found (Fig. 1A). Furthermore, at week 3 of differentiation, in a set of 104 genes, strong expression of the markers of mature osteoblasts [20, 21], including ALP, OPN, OCN, and bone sialoprotein 2, was found (Fig. 1B). In addition, a set of genes known as IFNγ-inducible genes [22–26] displayed striking and transient changes in gene expression at week 1 (Fig. 1A), and returned to normal at week 3 (Fig. 1B). Finally, to estimate the reliability of our microarray results, we compared our results with those obtained by RT-PCR normalized for GAPDH on randomly selected genes at each time point (Fig. 1C, 1D).
IFNγ Is Produced Following Induction of hMSCs to the Osteoblastic Lineage and Regulates Their Differentiation in an Autocrine Manner
After identification that IFNγ-inducible genes were upregulated during early osteoblastogenesis, we then assessed the capacity of hMSCs to produce IFNγ prior to, and following, exposure to OM. Concentrations of IFNγ were almost undetectable in the conditioned medium of uncommitted hMSCs at weeks 1 and 2 (Fig. 2A), but increased significantly in the third week, confirming previous reports of IFNγ production by confluent MSCs [27, 28]. In contrast, conditioned hMSCs exposed to OM showed a dramatic increase in IFNγ production at week 1, and returned to baseline at week 2 (Fig. 2A) (p < .01).
We then assessed the effect of IFNγ blockade on the ability of hMSCs to differentiate into osteoblasts. IFNγ blockade did not affect cell survival or proliferation (data not shown). We compared hMSC differentiation into osteoblasts in cells treated with control siRNA and IFNγ siRNA. IFNγ mRNA levels remained stable throughout the experiment when treated with control siRNA (Fig. 2C, left panel). In contrast, the addition of IFNγ siRNA induced a rapid and sustained inhibition of IFNγ mRNA for the duration of the experiment (Fig. 2C, right panel). This sustained inhibition was also seen with IFNγ protein levels (Fig. 2D). IFNγ blockade in the presence of OM resulted in strong inhibition of differentiation, as shown prior to (Fig. 2E) and after (Fig. 2F) staining with Alizarin Red.
IFNγ Induces Osteoblastogenesis in a Dose-Dependent Manner
The expression of IFNγR1 was significantly increased by exposure of hMSCs to OM and by treatment with IFNγ as compared with GM-treated hMSCs (p < .01) (Fig. 3A, 3B). This increase occurred early (at week 1) and was sustained for the duration of the experiment (Fig. 3B). In addition, when hMSCs were induced to differentiate with OM, no significant difference in cell proliferation between IFNγ- and vehicle-treated cells was found (Fig. 3C). We then examined the capacity of IFNγ to alter the differentiation of hMSCs into osteoblasts. IFNγ was added simultaneously with OM. At week 1 after the addition of IFNγ to hMSCs, a significant dose-dependent and early increase in ALP-expressing cells was observed as compared with untreated differentiating cells (76% ± 4% versus 8% ± 2%; p < .01), (Fig. 3D, 3E). Higher doses of IFNγ did not cause a further increase in the percentage of ALP-expressing cells (data not shown). Similar results were obtained using Alizarin Red staining, a measure of the ability of osteoblasts to mineralize (Fig. 3E, 3G). Quantification of Alizarin Red staining showed that hMSCs treated with IFNγ mineralized earlier in a dose-dependent manner (p < .01) (Fig. 3G). Additionally, when hMSCs were treated with increasing concentrations of IFNγ in GM alone, no effect on either ALP expression or mineralization was observed (Fig. 3D, 3F).
Expression of Osteogenic Transcription Factors Following IFNγ Knockdown
We first assessed whether the high levels of early osteoblast differentiation and mineralization induced by IFNγ in OM-treated hMSCs were concomitant with changes in the expression of Runx2. hMSCs treated with GM either alone or with IFNγ (100 ng/ml) expressed low levels of Runx2 at weeks 1, 2, and 3 (Fig. 4A, 4B). As expected , levels of Runx2 expression progressively increased after exposure of hMSCs to OM (Fig. 4A, 4B). In contrast, the addition of IFNγ to OM induced a significant increase in expression of Runx2 at week 1, compared with cells treated with GM alone. IFNγ-treated cells continued to display a strong induction of Runx2 at week 2 as compared with OM alone, which decreased at week 3 to the same levels observed with OM (Fig. 4A, 4B).
We next assessed the effect of IFNγ blockade on the expression of Runx2 and OCN. Concomitant with lower levels of osteoblast differentiation and mineralization, a dramatic decrease in the expression of Runx2 and OCN was observed in cells treated with IFNγ siRNA as compared with control siRNA (Fig. 4C, 4D).
IFNγ Signaling Regulates Osteoblast Differentiation In Vivo
We then determined the effect of IFNγ signaling on osteoblast differentiation in vivo in a mouse model in which the IFNγ receptor had been disrupted by homologous recombination, the IFNγR1−/− model. These animals are normal at birth, are fertile, have a normal growth rate and body weight, and cannot be differentiated from control IFNγR1+/+ mice, except for subtle differences in the immune system [29, 30]. Initially, we examined BMD changes in IFNγR1−/− mice and in control IFNγR1+/+ mice. BMD measured at the spine and femora indicated that IFNγR1−/− mice had a significantly lower bone mass in the spine (41% lower) than IFNγR1+/+ mice (0.038 ± 0.004 g/cm2 versus 0.064 ± 0.005 g/cm2; p < .01) and in the femur (31% lower) (0.032 ± 0.005 g/cm2 versus 0.046 ± 0.002 g/cm2; p < .01) (Fig. 5A). Subsequently, to assess whether, as in our in vitro experiments, MSCs lose their capacity to differentiate into osteoblasts in the absence of IFNγ signaling in vivo, we cultured bone marrow cells derived from 4- and 8-week-old IFNγR1−/− and IFNγR1+/+ mice in OM. After 3 weeks of treatment, the number of CFU-OB derived from IFNγR1−/− mice was significantly lower than the number derived from 4-week-old IFNγR1+/+ mice (6 ± 2 versus 21 ± 5; p < .01) and 8-week-old IFNγR1+/+ mice (12 ± 5 versus 46 ± 2; p < .01) (Fig. 5B, 5C). Similarly, bone marrow cells derived from 4- and 8-week-old IFNγR1+/+ mice and treated for 14 days with IFNγ siRNA had significantly fewer CFU-OB than bone marrow cells treated with control siRNA (data not shown).
In this study, we identified a new role for IFNγ in MSCs committed to differentiate into the osteoblastic lineage both in vitro and in vivo. We first determined the changes in gene expression that take place during the differentiation of hMSCs into osteoblasts. We demonstrated that, in addition to the expression of previously described signaling pathways of osteoblast differentiation , a strong and transient induction of a new signaling pathway, IFNγ-inducible genes, occurred during the proliferative phase of hMSC differentiation.
Among these IFNγ-inducible genes, IFI-16 and absent in melanoma 2 belong to the IFNγ-induced nuclear protein family NIH-200, which is known to play an important role in cell differentiation and embryonic development [22, 23]. Earlier studies had shown that, after exposure to IFNγ, IFI-16 and absent in melanoma 2 interact with Stat heterodimer and are translocated to the nucleus by karyopherin, where they associate with casein kinase 2 [23, 26], a protein involved in osteogenic differentiation of hMSCs [21, 26]. Our study also demonstrates that karyopherin, also known as importin β2, is upregulated at the end of the proliferative phase.
The pattern of gene expression observed therefore supports the coordinated and early activation of these IFNγ-inducible genes during the commitment of hMSCs to the osteoblastic lineage. The transient nature of IFNγ-inducible gene expression was striking during the proliferative phase (week 1) and returned to baseline during the mineralization phase. These changes were suggestive of a potential role of IFNγ signaling at the early stage of the differentiation process of hMSCs into osteoblasts and triggered us to investigate the regulation of IFNγ during this process. In fact, IFNγ production by proliferating and osteoblastic stromal cells has been reported earlier [27, 28]. However, the production of IFNγ by MSCs undergoing osteogenic differentiation was considered a response to the presence of factors released by immune cells into the medium .
An important finding from our study was the novel observation of a sharp and transient increase in IFNγ production by hMSCs induced to differentiate into osteoblasts in the absence of either immune cells or their released factors. This was concomitant with the transient expression of IFNγ-inducible genes, which showed a significantly higher expression level in hMSCs committed to differentiate into osteoblasts than in noncommitted hMSCs at week 1. It is important to note that this difference in gene expression does not imply that IFNγ-responsive genes have not been activated in undifferentiated (noncommitted) hMSCs at week 3, but rather indicates no significant difference in IFNγ-responsive gene relative expression between OM- and GM-treated cells at week 3. Consistent with previous reports [27, 28], we found that hMSCs express receptors for IFNγ (IFNγR1) and that induction of osteogenic differentiation is accompanied by a rapid and sustained increase in their expression, suggesting a synergistic interaction between IFNγR1 expression and the autocrine production of its ligand. Addition of IFNγ further enhanced this rapid and sustained increase in IFNγR1 expression only in the presence of OM, further supporting this synergistic interaction. Interestingly, IFNγR1 expression was initially low in nonosteogenic conditions (GM) and increased rapidly, reaching a maximum at week 2, suggesting that the late (delayed) autocrine production of IFNγ by hMSCs may be responsible for this delayed increase in IFNγR1 expression. Taken together, these findings suggest that modulation of IFNγR1 plays an important role for hMSC differentiation and acts synergistically with IFNγ in an IFNγ autocrine loop to modulate osteoblast differentiation. The critical role of this autocrine loop was then demonstrated by blocking endogenous IFNγ production by siRNA in differentiating hMSCs and also by inducing the differentiation of MSCs, obtained from the osteopenic IFNγR1−/− mice, into the osteoblastic lineage. We expected that both approaches used here would block endogenous IFNγ production prior to induction with OM. We found that the absence of IFNγ not only inhibited the early peak in IFNγ production by hMSCs observed at week 1 in vitro but also significantly inhibited MSC differentiation into osteoblasts at week 3, both in vitro and in vivo. This evidence indicates that expression of IFNγ by MSCs is an important early step during their commitment to the osteoblastic lineage.
Furthermore, to assess the potential effect of exogenous treatment with IFNγ on osteoblastogenesis, we next added IFNγ to OM-treated hMSCs committed to differentiate into osteoblasts. At week 1, IFNγ accelerated hMSC differentiation into osteoblasts in a dose-dependent manner only in the presence of OM, indicating that an optimal osteogenic milieu is necessary for the IFNγ effect to occur. This effect was not dependent on cell proliferation, indicating that IFNγ stimulates differentiation without affecting proliferation in the presence of OM.
Another important finding in this study was the observed changes in Runx2 expression in early osteoblastogenesis prior to and following IFNγ knockdown. Previous studies have indirectly linked IFNγ and Runx2 through crosstalk between Runx2 and Stat1, a transcription factor regulated by IFNγ [5, 31–33]. Unphosphorylated Stat1 physically interacts with Runx2 to inhibit its function [5, 32]. However, a direct link between IFNγ and Runx2 in the absence of immune cells or immune cell-released factors has not been reported. In our study, Runx2 expression was strongly inhibited following knockdown of endogenous IFNγ production, supporting a mechanistic link between IFNγ signaling and Runx2. Additionally OCN, an osteogenic protein downstream of the Runx2-activated pathway , was also inhibited after IFNγ knockdown.
Addition of IFNγ to MSCGM did not show an effect on Runx2 expression over the levels seen with GM alone. In contrast, the strongest induction of Runx2 occurred at week 1 following the addition of IFNγ to OM. This strong induction was sustained at week 2 but decreased abruptly at week 3. The strong initial increase is consistent with Runx2-mediated osteoblastogenesis with IFNγ. The mechanism underlying the progressive decrease in Runx2 expression at weeks 2 and 3 is unclear but could be related to downregulation following completion of the differentiation process. Taken together, these results strongly suggest that IFNγ signaling is a key regulator of hMSC differentiation into osteoblasts, acting at least in part through the Runx2-related pathway. Further studies looking at the link between Runx2 and IFNγ signaling are required.
The significance of our findings resides in the importance that osteoimmunology has acquired in recent years. Although IFNγ is considered to play an important role in bone turnover, it has been proposed that its role is predominantly in the regulation of osteoclastic activity . IFNγ is a strong inhibitor of osteoclastogenesis in vitro [7, 34], but it stimulates osteoclastogenesis in vivo . A significant lack of knowledge exists regarding the role of IFNγ in osteoblastogenesis and bone formation both in vitro and in vivo.
Overall, our results suggest that, in addition to its reported role in osteoclastogenesis and bone resorption [31, 34–36], IFNγ may also play an important role in osteoblastogenesis and bone formation both in vitro and in vivo. In addition, our findings that hMSCs secrete IFNγ in the absence of immune cells and that its inhibition by siRNA blocks hMSC differentiation into osteoblasts suggest that the autocrine secretion of IFNγ by hMSCs is an essential step in the early commitment of hMSCs towards osteoblastogenesis in vitro. Considering the importance of crosstalk between the immune system and skeletal homeostasis [1, 7], our data showing an effect of IFNγ on hMSC commitment to the osteoblastic lineage add yet another level to the complexity of the IFNγ role in bone biology.
In conclusion, our data support the concept of an IFNγ autocrine loop as an important early step for hMSC commitment to the osteoblastic lineage. Furthermore, the observation that activation of IFNγ signaling further enhances hMSC differentiation into osteoblasts indicates that additional studies are necessary to determine its overall contribution to skeletal homeostasis.
Dr. Duque holds a Fellowship from the University of Sydney Medical Research Foundation. Dr. Kremer holds a Chercheur National Award from the Fond de la Recherche en Santé du Québec. This study was supported by the Canadian Institutes for Health Research (CIHR MOP 10839), the Dairy Farmers of Canada, and NSERC to R. Kremer. We thank Pat Hales and Leigh Bambury for preparation of the manuscript.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.