Protein Kinase Inhibitor γ Reciprocally Regulates Osteoblast and Adipocyte Differentiation by Downregulating Leukemia Inhibitory Factor

Authors

  • Xin Chen,

    1. Department of Orthopaedics, University of North Carolina, North Carolina, USA
    Search for more papers by this author
  • Bryan S. Hausman,

    1. Department of Orthopaedics, University of North Carolina, North Carolina, USA
    Search for more papers by this author
  • Guangbin Luo,

    1. Department of Genetics and Genome Sciences, University of North Carolina, North Carolina, USA
    Search for more papers by this author
  • Guang Zhou,

    1. Department of Orthopaedics, University of North Carolina, North Carolina, USA
    2. Department of Genetics and Genome Sciences, University of North Carolina, North Carolina, USA
    Search for more papers by this author
  • Shunichi Murakami,

    1. Department of Orthopaedics, University of North Carolina, North Carolina, USA
    2. Department of Genetics and Genome Sciences, University of North Carolina, North Carolina, USA
    3. Center for Regenerative Medicine, School of Medicine, Case Western Reserve University, Cleveland, Ohio, USA
    Search for more papers by this author
  • Janet Rubin,

    1. Department of Medicine, University of North Carolina, North Carolina, USA
    Search for more papers by this author
  • Edward M. Greenfield

    Corresponding author
    1. Department of Orthopaedics, University of North Carolina, North Carolina, USA
    2. Center for Regenerative Medicine, School of Medicine, Case Western Reserve University, Cleveland, Ohio, USA
    3. Department of Pathology, School of Medicine, Case Western Reserve University, Cleveland, Ohio, USA
    • Correspondence: Edward M. Greenfield, Ph.D., Harry E. Figgie III M.D. Professor of Orthopaedics, Director of Orthopaedic Research, Department of Orthopaedics, Case Medical Center, Case Western Reserve University, Biomedical Research Building, Room 331, 2109 Adelbert Road, Cleveland, Ohio 44106, USA. Telephone: 216-368-1331; Fax: 216-368-1332; e-mail: emg3@cwru.edu

    Search for more papers by this author

  • Author contributions: X.C.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; B.S.H.: conception and design, provision of study material or patients, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; G.L., G.Z., E.M.G., and S.M.: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript; J.R.: provision of study material or patients, data analysis and interpretation, manuscript writing, and final approval of manuscript.

Abstract

The protein kinase inhibitor (Pki) gene family inactivates nuclear protein kinase A (PKA) and terminates PKA-induced gene expression. We previously showed that Pkig is the primary family member expressed in osteoblasts and that Pkig knockdown increases the effects of parathyroid hormone and isoproterenol on PKA activation, gene expression, and inhibition of apoptosis. Here, we determined whether endogenous levels of Pkig regulate osteoblast differentiation. Pkig is the primary family member in murine embryonic fibroblasts (MEFs), murine marrow-derived mesenchymal stem cells, and human mesenchymal stem cells. Pkig deletion increased forskolin-dependent nuclear PKA activation and gene expression and Pkig deletion or knockdown increased osteoblast differentiation. PKA signaling is known to stimulate adipogenesis; however, adipogenesis and osteogenesis are often reciprocally regulated. We found that the reciprocal regulation predominates over the direct effects of PKA since adipogenesis was decreased by Pkig deletion or knockdown. Pkig deletion or knockdown also simultaneously increased osteogenesis and decreased adipogenesis in mixed osteogenic/adipogenic medium. Pkig deletion increased PKA-induced expression of leukemia inhibitory factor (Lif) mRNA and LIF protein. LIF neutralizing antibodies inhibited the effects on osteogenesis and adipogenesis of either Pkig deletion in MEFs or PKIγ knockdown in both murine and human mesenchymal stem cells. Collectively, our results show that endogenous levels of Pkig reciprocally regulate osteoblast and adipocyte differentiation and that this reciprocal regulation is mediated in part by LIF. Stem Cells 2013;31:2789–2799

Introduction

The cAMP/protein kinase A (PKA) signaling pathway is activated by Gαs-coupled receptors and potently regulates multiple cellular processes. Because of the importance of this pathway, its activity is precisely controlled by the interaction of multiple mechanisms [1-4], including inactivation of PKA by the protein kinase inhibitor (Pki) gene family [5-8]. We previously showed that inactivation of nuclear PKA by PKIγ is a primary mechanism that reduces PKA signaling induced by Gαs-coupled receptors in mesenchymal cells [9, 10]. Consequently, inactivation of nuclear PKA by Pkig reduces PKA-dependent activation of transcription factors and the resultant expression of rapidly and transiently induced mRNAs [9, 10]. There have been extensive and elegant investigations of the biochemistry and structure of the PKI family and of the inactivation of PKA by PKI overexpression [5-8], but prior to our studies [9-11], there were only two publications that showed effects of endogenous PKI expressed at physiological levels [12, 13]. In this regard, genetic deletion of the other two members of the Pki family, either separately or in combination, caused little detectable phenotype in the resultant mice because of compensatory regulation by other components of the PKA pathway [14, 15].

PKA signaling stimulates bone formation due to effects on the osteoblast lineage, including increased proliferation and differentiation of osteoblast precursors, activation of quiescent lining cells, and prevention of apoptosis [16, 17]. We previously showed that inactivation of nuclear PKA by PKIγ is a primary mechanism that reduces the antiapoptotic effects of Gαs-coupled receptors in osteoblasts [11]. Moreover, overexpression of PKIγ was reported to reduce BMP2-induced osteoblast differentiation [18]. One of the goals of this study was therefore to determine whether endogenous PKIγ, expressed at physiological levels, reduces PKA-induced osteoblast differentiation.

Adipocyte differentiation is also stimulated by PKA signaling [19-22] and therefore might be expected to be reduced by PKIγ. However, despite the preponderance of evidence that PKA signaling stimulates adipocyte differentiation [19-22], there are also reports that PKA signaling can have the opposite effect [23-26]. As such, timing and magnitude of the PKA signal may be important and the effects of PKI may be complex. It is also unknown whether the direct effects of PKA on adipocyte differentiation will predominate over the reciprocal regulation of adipocyte and osteoblast differentiation that frequently occurs because both lineages differentiate from the same mesenchymal precursor cells [27, 28]. Examples of reciprocal regulation that are particularly relevant to this study are the reduction of adipocyte differentiation and stimulation of osteoblast differentiation in response to PKA activation by forskolin [29], parathyroid hormone (PTH) [30, 31], or transgenic expression of constitutively active PTH receptors [32]. The second goal of this study was therefore to determine whether endogenous PKIγ, expressed at physiological levels, stimulates or reduces adipocyte differentiation and to resolve whether the direct effects of PKA predominate over effects due to the reciprocal regulation of adipocyte and osteoblast differentiation.

Leukemia inhibitory factor (LIF), a member of the IL6 family of gp130-dependent cytokines, can stimulate osteoblast differentiation [33-38] and reduce adipocyte differentiation [34, 37, 39, 40]. We previously found that Lif is one of the mRNAs that are rapidly and transiently induced by PKA signaling in osteoblasts in cell culture and in vivo [41-43]. The third goal of this study was therefore to determine whether downregulation of LIF expression by PKIγ mediates regulation of osteoblast and adipocyte differentiation.

Materials and Methods

Murine Embryonic Fibroblasts

Pkig/ mice generated in our lab were backcrossed with C57BL/6J mice for four generations. Wild-type and Pkig/ mice were obtained by breeding of the resultant Pkig+/ mice and murine embryonic fibroblasts (MEFs) were prepared from embryos 12.5 to 13.5 days postcoitum using standard techniques [44]. MEFs were maintained in growth medium consisting of α-MEM medium (SH30265.01, HyClone, Logan, UT, www.thermoscientific.com/hyclone) supplemented with 10% fetal bovine serum (SH30071.03, HyClone), 2 mM l-glutamine (25-005-Cl, Cellgro, Manassas, VA, www.cellgro.com), nonessential amino acids (25-025-Cl, Cellgro), 100 units/ml penicillin, and 100 µg/ml streptomycin (30-002-Cl, Cellgro). For experiments, MEFs that had been passaged less than five times were plated at 1 × 104/cm2 in growth medium.

Osteoblast differentiation by MEFs was induced as previously described [45]. Briefly, confluent MEFs were incubated in osteogenic medium consisting of the growth medium supplemented with 50 µg/ml 2-phospho-l-ascorbic acid (49752, Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) and 10 mM β-glycerophosphate (G-9891, Sigma-Aldrich) plus 1 µM forskolin (344270, Calbiochem, Gibbstown, NJ, http://www.emdmillipore.com) or 0.1% dimethyl sulphoxide (DMSO) as a vehicle control with medium changes every 3 or 4 days. Specificity of the effects of forskolin was confirmed by showing that the inactive analog, dideoxy-forskolin [46], does not induce osteoblast differentiation (Supporting Information Fig. S1). Indicated cultures also received 20% conditioned medium from CHO cells that stably express BMP4, a kind gift from Dr. Tatsuya Koike, Osaka City University Medical School [47, 48]. The ability of this conditioned medium to stimulate osteoblast differentiation is shown in Supporting Information Figure S2.

Adipocyte differentiation was induced by incubating 2-day postconfluent MEFs with multiple cycles of 3 days in induction medium and 1 day in maintenance medium as described [49]. The maintenance medium was the growth medium supplemented with 10 µg/ml insulin (I5500, Sigma-Aldrich). The induction medium was the maintenance medium further supplemented with 1 µM dexamethasone (D8893, Sigma-Aldrich), either 200 µM indomethacin (I7378, Sigma-Aldrich) or 0.4% ethanol, and either 0.5 mM 3-isobutyl-1-methylxanthine (IBMX, I701, Sigma-Aldrich) or 0.1% DMSO.

To simultaneously induce both osteoblast and adipocyte differentiation, confluent MEFs were incubated with a 1:1 mixture of the osteogenic and adipogenic induction media as recommended [49, 50]. For this purpose, the adipogenic induction medium described above was supplemented with both indomethacin and IBMX, but neither forskolin nor BMP4 was added to the osteogenic medium. Media were changed every 3 or 4 days.

Murine Marrow-Derived Mesenchymal Stem Cells

The murine marrow-derived mesenchymal stem cell (mdMSC) line was obtained from bone marrow of adult male C57BL/6 mice and was previously described [51]. mdMSCs were maintained in the same growth medium described above for MEFs. For reverse transfection, 5 × 104 mdMSCs in 1 ml of growth medium were added to individual wells of a 12-well plate containing 200 pmoles of murine Pkig siRNA (Dharmacon On-TARGET plus J-049150-10, Thermo Scientific, Waltham, MA, www.thermoscientific.com) or nontargeting siRNA (Dharmacon D-001810-10-20, Thermo Scientific) that had been preincubated for 30 minutes at room temperature with DharmaFECT 3 transfection reagent (T-2003-03, Thermo Scientific) and DharmaFECT cell culture reagent (B-004500-100, Thermo Scientific). Transfection reagents were replaced with growth medium after 4 hours of incubation at 37°C in 5% CO2. Twenty hours later, mdMSCs were incubated with a 1:1 mixture of the osteogenic and adipogenic media previously described for these cells [51] further supplemented with 0.25 mM IBMX. Media were changed every 3 days.

Human Mesenchymal Stem Cells

The human MSCs (hMSCs) were obtained from Lonza (PT-2501, Walkersville, MD, www.lonza.com) and maintained in growth medium consisting of Dulbecco's modified Eagle's medium/high glucose medium (SH30243.01, HyClone) supplemented with 10% fetal bovine serum, 2 mM l-glutamine, nonessential amino acids, 100 units/ml penicillin, and 100 µg/ml streptomycin. For reverse transfection, 4 × 104 hMSCs in 400 µl of growth medium were added to individual wells of a 24-well plate containing 100 pmoles of human PKIG siRNA (Dharmacon On-TARGET plus J-017218-06, Thermo Scientific) or nontargeting siRNA that had been preincubated as described in the previous paragraph. Transfection reagents were replaced with growth medium after 4 hours of incubation at 37°C in 5% CO2. Twenty hours later, the hMSCs were then used in osteogenic or adipogenic experiments. Osteogenic medium consisted of the growth medium supplemented with 50 µg/ml 2-phospho-l-ascorbic acid and 10 mM β-glycerophosphate and was changed every 3 days. Adipocyte differentiation was induced by three cycles of 3 days of adipogenic induction medium (growth medium supplemented with 5 µg/ml insulin, 100 nM dexamethasone, 50 µM indomethacin, and 0.5 mM IBMX) and 2 days of adipogenic maintenance medium (growth medium supplemented with 5 µg/ml insulin) as recommended by the manufacturer.

Quantitative Real-Time RT-PCR

Gene expression was measured by quantitative real-time RT-PCR using the linear portion of standard curves as we previously described [52]. All PCR primers were designed to overlap exon-exon junctions at their 3′-termini using primer3 software [53] except for Pkig and those mRNAs that are encoded by a single exon (Gapdh and C/ebpa). All primers were validated by sequencing of PCR amplicons. Murine and human primers are listed in Supporting Information Tables S1 and S2, respectively. All assays included negative controls without RNA and both analysis of melting curves and agarose gel electrophoresis to confirm the presence of single PCR amplicons.

PKA Activity Assay

Nuclear fractions were prepared as we previously described [10]. PKA activities were assayed using fluorescently labeled-kemptide as substrate (Nonradioactive PepTag Assay kit, V5340, Promega, Madison, WI, www.promega.com). The fraction of phosphorylated kemptide was quantified by measuring fluorescence (excitation at 535 nm and emission at 600 nm) on a 4000MM In-Vivo F Image Station (Kodak, Rochester, NY, www.kodak.com) after separation from nonphosphorylated kemptide by electrophoresis on 0.8% agarose gels in Tris-Cl (pH 8.0) buffer. More than 90% of kemptide phosphorylation by nuclear fractions from forskolin-treated MEFs is inhibited by PKIα (Supporting Information Fig. S3) and is therefore due to PKA activity. PKA activity was expressed as µmol/minute/µg protein. Total protein was measured using the bicinchoninic acid (BCA) protein assay kit (23227, Pierce, Rockford, IL, www.piercenet.com).

Western Blotting

MEFs were lysed in SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, and 10% glycerol) plus protease inhibitors (cocktail tablet 04693159001, Roche, Mannheim, Germany, www.roche.com) and phosphatase inhibitors (1 mM sodium pyrophosphate, 1 mM sodium orthovanadate, and 1 mM β-glycerophosphate). Equivalent amounts of protein (BCA protein assay) per lane were electrophoresed on 12% Tris-Glycine gels (59503, Lonza, Rockland, ME) and transferred to polyvinylidene difluoride membranes. Membranes were blocked in Tris-buffered saline (pH 8.0) containing 0.05% Tween 20 and 5% non-fat dry milk and incubated with 1:1,000 dilutions of primary antibody (pSer133-cAMP response element-binding protein (CREB), 9191; CREB, 9197; ATF4, 11815; β-actin, 4970; or GAPDH, 2118; all from Cell Signaling, Danvers, MA, www.cellsignal.com) or with 1:500 dilutions of primary antibody (Runx2, 5356-1, Epitomics, Burlingame, CA, www.epitomics.com; or Osterix, ab22552, Abcam, Cambridge, MA, www.abcam.com). The membranes were then incubated with horseradish peroxidase-conjugated anti-rabbit IgG (Cell Signaling, 7074). Bands were detected by chemiluminescence (Amersham ECL Plus, RPN2132, GE Health Care, Piscataway, NJ, www.gelifesciences.com).

Alkaline Phosphatase, Alizarin Red S, and Oil Red O Assays

Alkaline phosphatase activity was measured biochemically as previously described [54]. Alkaline phosphatase staining was performed using a kit (86C-1KT, Sigma-Aldrich) following the recommendations of the manufacturer. Staining of mineralization with Alizarin Red S (A5533, Sigma-Aldrich) was performed and quantified following extraction with 10% cetylpyridinium chloride as previously described [55]. Staining of lipid accumulation with oil red O (O0625, Sigma-Aldrich) was performed and quantified following extraction with isopropanol as previously described [56]. For costaining, alkaline phosphatase staining was performed first and the cells were then rinsed with water twice before the oil red O staining.

LIF ELISA and Neutralizing Antibodies

LIF protein in cell culture supernatants was measured using a kit (MLF00, R & D Systems, Minneapolis, MN, www.rndsystems.com) following the recommendations of the manufacturer. To determine whether LIF mediates the effects of Pkig deletion or knockdown, indicated concentrations of either a neutralizing affinity-purified goat anti-murine LIF antibody (AF449, R & D Systems) or a normal goat control immunoglobulin (AB-108-C, R & D Systems) were incubated with MEFs or mdMSCs experiments. To determine whether LIF mediates the effects of PKIG knockdown, 10 µg/ml of either a neutralizing mouse anti-human LIF monoclonal antibody (MAB250, R & D Systems) or an isotype control mouse IgG2B (MAB004, R & D Systems) was incubated with hMSCs.

Statistical Analysis

Unless otherwise indicated, all quantitative data are expressed as mean ± SD of three independent experiments, each containing triplicate culture wells assayed in triplicate. Statistical analysis was by Student's t test for experiments with two groups and by one-way ANOVA with Bonferroni post hoc tests of the indicated comparisons for experiments with three or more groups. All statistical tests were performed with SigmaStat 3.0 software (Systat Software, Inc., Point Richmond, CA). p values less than .05 were considered statistically significant. All nonquantitative results were also derived from three independent experiments. Images from representative experiments were selected based on the results from the quantitative portions of the same experiments. For example, the images in Figure 1C are from the experiment that was closest to the means in Figure 1B.

Figure 1.

Pkig deletion substantially extends PKA signaling following stimulation with FSK. Murine embryonic fibroblasts from wild-type (WT) and Pkig−/− (KO) mice were incubated for 12 hours in osteogenic medium (A) or for the indicated times (B--F) in osteogenic medium containing 1 µM FSK or 0.1% DMSO as a vehicle control (C). mRNAs encoding members of the Pki family (A) and PKA-dependent genes (D–F) were measured by quantitative real-time RT-PCR, nuclear PKA activity was assessed biochemically (B), phosphorylation of CREB and ATF1 and total CREB were assessed by Western blotting (C). * denotes p < .05 as compared to the indicated groups (A) or the WT groups at the same time points (B, D--F). In (A), the horizontal bars represent the means of the three independent experiments indicated by the individual symbols that represent means of three culture wells/group, each assayed in triplicate. In (B, D--F), the symbols represent the mean ± SD of three independent experiments, each consisting of three culture wells/group, each assayed in triplicate. Abbreviations: FSK, forskolin; LIF, leukemia inhibitory factor; PKA, protein kinase A.

Results

Pkig Deletion Increases PKA Signaling

Pkig mRNA is the dominant Pki family member expressed in osteoblasts and other mesenchymal cells [6, 10]. Consistent with those findings, Pkig mRNA expression is 3.8-fold higher than Pkia mRNA expression in wild-type MEFs and Pkib mRNA is barely detectable (first three groups in Fig. 1A). We previously showed that Pkig knockdown by either siRNA or antisense approaches substantially increases PKA signaling [10, 11]. To determine whether off-target effects might be responsible for the effects of Pkig knockdown, we compared wild-type and Pkig/ MEFs. Pkig deletion does not induce detectable compensatory changes in mRNAs encoding either of the other PKI family members (last three groups in Fig. 1A). Pkig deletion substantially increases the magnitude and extends the time course of both nuclear PKA activation and phosphorylation of CREB and ATF1 transcription factors induced by forskolin in MEFs (Fig. 1B, 1C). Similarly, Pkig deletion substantially increases stimulation by forskolin of c-fos, Lif, and IL6 mRNAs (Fig. 1D--1F). Taken together, these results demonstrate that PKA signaling is increased by Pkig deletion in MEFs. They therefore confirm our previous studies of Pkig knockdown [10, 11] and validate MEFs as a cellular model system for investigating the effects of Pkig deletion.

Pkig Deletion Increases Osteoblast Differentiation

PKA signaling induced by the adenylate cyclase activator forskolin stimulates osteoblast differentiation in vitro [29, 48, 57]. To examine whether the enhanced PKA signaling induced by Pkig deletion also increases osteoblast differentiation, we took advantage of the ability of MEFs to model MSCs by undergoing osteoblastic differentiation [45, 58-61] as well as adipocytic and chondrocytic differentiation [62, 63]. In osteogenic medium, Pkig remains the dominant Pki family member expressed in MEFs and Pkig deletion does not induce detectable compensatory changes in the other Pki family members (Fig. 2A--2C). The transcription factors, Runx2 and osterix, were expressed at high levels by both wild-type and Pkig/ MEFs at the beginning of the culture period at both the mRNA (first group in Fig. 2D, 2E) and protein levels (Supporting Information Fig. S4). Runx2 and osterix mRNA levels decreased in the wild-type MEFs during culture (red symbols in Fig. 2D, 2E) but remained elevated in the Pkig/ MEFs (blue symbols in Fig. 2D, 2E). As expected, ATF4 was induced during osteoblast differentiation and was modestly increased at the mRNA level by Pkig deletion at day 7 (Fig. 2F).

Figure 2.

Pkig deletion substantially increases osteoblast differentiation. Murine embryonic fibroblasts from wild-type (WT) and Pkig−/− (KO) mice were incubated for the indicated times in osteogenic medium supplemented with 1 µM FSK or 0.1% DMSO as a vehicle control in the presence or absence of BMP-4. mRNAs encoding members of the Pki family (A--C) and osteoblastic marker genes (D--H) were measured by quantitative real-time RT-PCR. Osteoblast differentiation was also assessed by examining ALP biochemically and cytochemically (I), and by Alizarin Red S staining of mineral (J), after 7 or 14 days, respectively, in osteogenic medium. * denotes p < .05 and NS denotes not significant. The horizontal bars represent the means of the three (A--H, J) or four (I) independent experiments indicated by the individual symbols that represent means of three culture wells/group, each assayed in triplicate. Abbreviations: ALP, alkaline phosphatase; FSK, forskolin.

Alkaline phosphatase was used as a marker of the middle stage of osteoblast differentiation. Culture in osteogenic medium containing forskolin increases alkaline phosphatase mRNA levels (red symbols in Fig. 2G). Pkig deletion further increases forskolin-induced alkaline phosphatase mRNA levels by 3.7- and 5.3-fold, respectively, after 3 and 7 days (blue symbols in Fig. 2G). As expected, for a late stage marker of osteoblast differentiation, osteocalcin mRNA levels are only modestly affected after 3 days of incubation in osteogenic medium containing forskolin (middle pair of groups in Fig. 2H). However, Pkig deletion increases osteocalcin mRNA levels by 16.5-fold after 7 days (last pair of groups in Fig. 2H).

To confirm the effects observed on the mRNA markers, alkaline phosphatase activity and mineralization were assessed. These experiments were conducted in the absence and presence of BMP4 because osteoblast differentiation induced by BMPs is enhanced by forskolin [48, 57] and blocked by PKIγ overexpression [18]. Forskolin increases alkaline phosphatase activity and mineralization both in the absence and presence of BMP4 (red symbols in Fig. 2I, 2J). Pkig deletion further increases forskolin-induced alkaline phosphatase activity assessed biochemically both in the absence and presence of BMP4 by 2.3- and 3.9-fold, respectively, and increases forskolin-independent alkaline phosphatase activity both in the absence and presence of BMP4 by 4.3- and 4.9-fold, respectively (blue symbols in Fig. 2I). Staining of alkaline phosphatase activity confirms these results (lower panel of Fig. 2I). Pkig deletion also increases mineralization in cultures continuously stimulated with BMP4 by 6.4- and 1.6-fold, respectively, in the absence and presence of forskolin (third and fourth pairs of groups in Fig. 2J) and there was a trend toward increased mineralization by the Pkig/ MEFs in the absence of BMP4 (first and second pairs of groups in Fig. 2J). These results collectively demonstrate that Pkig deletion substantially increases osteoblast differentiation by MEFs.

Pkig Deletion Decreases Adipocyte Differentiation

Osteoblast and adipocyte differentiation are often regulated reciprocally but both lineages are stimulated by PKA signaling (see Introduction for details). We therefore next asked which of these mechanisms would predominate during adipogenesis when PKA signaling is enhanced by Pkig deletion. In adipogenic medium, Pkig remains the dominant Pki family member expressed in MEFs and Pkig deletion does not induce detectable compensatory changes in the other Pki family members (Fig. 3A--3C). As expected [19-22], induction of PKA signaling with the phosphodiesterase inhibitor IBMX increases adipocyte differentiation as assessed by measuring expression of both early and late adipogenic marker genes (red symbols in Fig. 3D, 3E for early markers and in Fig. 3F, 3G for late markers) or by staining for lipid accumulation with oil red O (red symbols in Fig. 3H). In contrast, further enhancement of PKA signaling by Pkig deletion decreases IBMX-induced expression of both the early and late adipogenic marker genes by 20%–52% (blue symbols in Fig. 3D--3G). Moreover, Pkig deletion inhibits oil red O staining by 33% in the presence of IBMX and indomethacin (last pair of groups in Fig. 3H) and there was a trend toward inhibition by Pkig deletion in all other tested conditions (first three pairs of groups in Fig. 3H). Taken together, these results demonstrate that Pkig deletion decreases adipocyte differentiation.

Figure 3.

Pkig deletion decreases adipocyte differentiation. Murine embryonic fibroblasts from wild-type (WT) and Pkig−/− (KO) mice were incubated for the indicated times in repeated cycles of adipogenic induction medium and maintenance medium as described in Materials and Methods. mRNAs encoding members of the Pki family (A--C) and adipocyte marker genes (D--G) were measured by quantitative real-time RT-PCR. Adipocyte differentiation was also assessed by staining lipid accumulation with oil red O after 14 days (H). * denotes p < .05 and NS denotes not significant. The horizontal bars represent the means of the three independent experiments indicated by the individual symbols that represent means of three culture wells/group, each assayed in triplicate.

Pkig Deletion Simultaneously Increases Osteoblast Differentiation and Decreases Adipocyte Differentiation

We next determined the effect of Pkig deletion on the simultaneous differentiation of osteoblasts and adipocytes in mixed osteogenic/adipogenic medium. Pkig deletion increases osteoblast differentiation as alkaline phosphatase activity, alkaline phosphatase staining, and mineralization are increased (Fig. 4A, 4B). Pkig deletion decreases adipocyte differentiation as assessed by oil red O staining by 30% on both day 4 and day 14 (Fig. 4C, 4D). Costaining for alkaline phosphatase and oil red O confirmed that the Pkig/ MEFs are more osteogenic (blue staining in Fig. 4E) and substantially less adipogenic (red staining in Fig. 4E) than the wild-type MEFs. These images also confirm that the two lineages are distinct from each other since almost no individual cells exhibited positive staining for both lineages (Fig. 4E). In summary, the results with the mixed osteogenic/adipogenic medium are consistent with the lineage-specific media reported above as Pkig deletion simultaneously increases osteoblast differentiation and inhibits adipocyte differentiation and the magnitude of the effect on osteoblast differentiation is larger than the effect on adipocyte differentiation.

Figure 4.

Pkig deletion simultaneously increases osteoblast differentiation and decreases adipocyte differentiation in mixed osteogenic/adipogenic medium. Murine embryonic fibroblasts from wild-type (WT) and Pkig−/− (KO) mice were incubated for the indicated times in the mixed osteogenic/adipogenic medium. Osteoblast differentiation was assessed by examining ALP biochemically and cytochemically (A) and by Alizarin Red S staining of mineral (B). Adipocyte differentiation was assessed by examining oil red O staining (C, D). In (A--D), the horizontal bars represent the means of the three independent experiments indicated by the individual symbols that represent means of three culture wells/group, each assayed in triplicate. In (E), cultures were costained for ALP activity (blue) and oil red O (red) after 4 days in the mixed osteogenic/adipogenic medium. Scale bars = 100 um. * denotes p < .05. Abbreviation: ALP, alkaline phosphatase.

The Effects of Pkig Knockdown in Murine and Human MSCs Are Similar to the Effects of Pkig Deletion in MEFs

To determine whether the results with MEFs are generalizable to other mesenchymal precursor cells, we first measured the effects of Pkig knockdown on mdMSCs cultured in mixed osteogenic/adipogenic medium. Pkig is also the dominant Pki family member expressed in the mdMSCs and Pkig knockdown does not induce detectable compensatory changes in the other Pki family members (first pair of groups in Fig. 5A--5C). Pkig knockdown increases expression of alkaline phosphatase and osteocalcin mRNAs by 2.2- and 1.6-fold compared to control groups transfected with nontargeting siRNA duplexes (first pair of groups in Fig. 5D, 5E). Simultaneously, Pkig knockdown decreases expression of both the early and late adipogenic marker genes by 30%–37% (first pair of groups in Fig. 5F--5H).

Figure 5.

LIF mediates the effects of Pkig knockdown on osteogenesis and adipogenesis in murine marrow-derived mesenchymal stem cells (mdMSCs). mdMSCs reverse transfected with nontargeting siRNA duplexes as a control (C) or siRNA targeting Pkig (KD) were incubated for 6 days in mixed osteogenic/adipogenic medium. Culture medium was supplemented with 0.3 µg/ml of either control immunoglobulin (IgG) or the anti-murine LIF Antibody (LIF Ab). mRNAs encoding members of the Pki family (A--C), osteoblastic marker genes (D, E), and adipogenic marker genes (F--H) were measured by quantitative real-time RT-PCR. Leptin mRNA was undetectable in all groups (not shown). * denotes p < .05 and NS denotes not significant. In (A--H), the horizontal bars represent the means of the three independent experiments indicated by the individual symbols that represent means of three culture wells/group, each assayed in triplicate. Abbreviation: LIF, leukemia inhibitory factor.

To further determine whether the results are generalizable to human mesenchymal precursor cells, we measured the effects of PKIG knockdown on hMSCs cultured in either osteogenic or adipogenic medium. PKIG is also the dominant PKI family member expressed in the hMSCs and PKIG knockdown does not induce detectable compensatory changes in other PKI family members (first pair of groups in Fig. 6A--6C and Supporting Information Fig. S5). PKIG knockdown increases expression of both alkaline phosphatase and osteocalcin mRNAs by 1.8-fold compared to control groups transfected with nontargeting siRNA duplexes (first pair of groups in Fig. 6D, 6E). In contrast, PKIG knockdown decreases mRNA expression of PPARG by 38% and adiponectin by 46% (first pair of groups in Fig. 6F, 6G). The effects of PKIγ knockdown with mdMSCs and hMSCs therefore confirm the results with MEFs (Figs. 2--4) that Pkig deletion reciprocally increases osteoblast differentiation and decreases adipocyte differentiation.

Figure 6.

LIF mediates the effects of PKIG knockdown on osteogenesis and adipogenesis in human mesenchymal stem cells (hMSCs). hMSCs reverse transfected with nontargeting siRNA duplexes as a control (C) or siRNA targeting PKIG (KD) were incubated for 6 days in osteogenic medium (A--E) or three cycles of adipogenic induction medium (3 days) and adipogenic maintenance medium (2 days) (F, G). Culture medium was supplemented with 10 µg/ml of either control immunoglobulin (IgG2B) or the anti-human LIF monoclonal antibody (LIF MAb). mRNAs encoding members of the PKI family (A--C), osteoblastic marker genes (D, E), and adipogenic marker genes (F, G) were measured by quantitative real-time RT-PCR. C/EBPA and leptin mRNA were undetectable in all groups with adipogenic media (not shown). * denotes p < .05 and NS denotes not significant. Horizontal bars represent the means of the three independent experiments indicated by the individual symbols that represent means of three culture wells/group, each assayed in triplicate. Abbreviation: LIF, leukemia inhibitory factor.

Reciprocal Regulation of Osteoblast and Adipocyte Differentiation by Pkig Is Mediated by LIF

Lif expression is induced by PKA activation (Fig. 1E and [41-43]) and LIF increases osteoblast differentiation [33-38] and decreases adipocyte differentiation [34, 37, 39, 40]. We therefore asked whether LIF mediates the effects of Pkig deletion. Consistent with the results that Pkig deletion increases basal and forskolin-induced expression of Lif mRNA (Fig. 1E), Pkig deletion also increases basal (2.7-fold), IBMX-induced (9.6-fold), and forskolin-induced (7.6-fold) secretion of LIF protein (Fig. 7A). Concomitantly, Pkig deletion increases basal (3.5-fold), IBMX-induced (3.9-fold), and forskolin-induced (5.9-fold) alkaline phosphatase activity (Fig. 7B) as expected based on the results in Figures 2 and 4.

Figure 7.

LIF mediates the effects of Pkig deletion on osteogenesis and adipogenesis in murine embryonic fibroblasts (MEFs) (G). MEFs from wild-type (WT) and Pkig−/− (KO) mice were incubated for 1 day (A), 4 days (B--D), or 14 days (E, F) in mixed osteogenic/adipogenic medium. Culture medium was supplemented with 0.25 mM 3-isobutyl-1-methylxanthine (IBMX), 1 µM forskolin (FSK), or 0.1% DMSO (C) as a vehicle control (A, B). Culture medium was supplemented with IBMX plus the indicated concentrations of the neutralizing anti-murine LIF antibody (LIF Ab) in (C) and 0.3 µg/ml of either control immunoglobulin (IgG) or the LIF antibody (LIF Ab) in (D--F). LIF protein in the cell culture supernatants was measured by ELISA (A). Osteoblast differentiation was assessed by examining alkaline phosphatase (ALP) biochemically (B--D) and cytochemically (lower panel in D) and by Alizarin Red S staining of mineral (E). Adipocyte differentiation was assessed by examining oil red O staining (F). * denotes p < .05 and NS denotes not significant. In (A, B, D--F), the horizontal bars represent the means of the three independent experiments indicated by the individual symbols that represent means of three culture wells/group, each assayed in triplicate. In (C), the symbols represent the mean ± SD of three independent experiments, each consisting of three culture wells/group, each assayed in triplicate. Abbreviations: ALP, alkaline phosphatase; FSK, forskolin; LIF, leukemia inhibitory factor; PKA, protein kinase A.

An antibody that neutralizes murine LIF was used to determine whether the effects of Pkig deletion are mediated by LIF. Initial experiments showed that 0.3 µg/ml of the antibody maximally inhibits osteoblast differentiation in Pkig/ MEFs (Fig. 7C) and that concentration was therefore used for all subsequent experiments. That concentration of control immunoglobulin does not alter the effects of Pkig deletion on osteogenesis or adipogenesis (compare first pair of groups in Fig. 7D--7F with Figs. 2--4). In contrast, the LIF antibody inhibits alkaline phosphatase activity, alkaline phosphatase staining, and mineralization in the Pkig/ MEFs (blue symbols and lower panels in Fig. 7D, 7E). The LIF antibody also enhances lipid accumulation in Pkig/ MEFs (blue symbols in Fig. 7F). As a result, in the presence of the LIF antibody, there are no significant differences between the wild-type and the Pkig−/− MEFs in mineralization or lipid accumulation (last pair of groups in Fig. 7E, 7F). These results demonstrate that the LIF antibody inhibits the effects of Pkig deletion on both osteoblast and adipocyte differentiation by MEFs.

The anti-murine LIF antibody also inhibits the effects of Pkig knockdown on the mRNA expression of the osteoblast marker genes by the mdMSCs (blue symbols in Fig. 5D, 5E). As a result, in the presence of the LIF antibody, there are no significant differences between the control and the Pkig knockdown groups in osteocalcin mRNA levels (second pair of group in Fig. 5E). There are also no significant differences between the control and the Pkig knockdown groups in levels of the adipocyte marker genes in the presence of the LIF antibody (second pair of groups in Fig. 5F--5H).

To determine whether LIF also mediates the effects of PKIG knockdown in human cells, we used a monoclonal antibody that neutralizes human LIF. Ten micrograms per milliliter of that monoclonal antibody was incubated with hMSCs based on the results in Figure 7C and the manufacturer's measurements of the relative potencies of the two antibodies. The anti-LIF monoclonal antibody inhibits the effects of PKIG knockdown on the mRNA expression of osteoblast and adipocyte marker genes (blue symbols in Fig. 6D--6G). As a result, in the presence of the LIF monoclonal antibody, there are no significant differences between the control and the PKI knockdown groups in ALP and osteocalcin mRNA levels (second pair of groups in Fig. 6D, 6E). There are also no significant differences between the control and the PKI knockdown hMSCs in levels of the adipocyte marker genes in the presence of the LIF monoclonal antibody (second pair of groups in Fig. 6F--6G). The mRNA expression results with both mdMSCs and hMSCs therefore confirm and extend the biochemical and cytochemical results with MEFs (Fig. 7) that the effects of Pkig deletion on the reciprocal regulation of osteoblast and adipocyte differentiation are mediated by LIF.

Discussion

This study showed that activation of nuclear PKA by Pkig deletion or knockdown increases osteoblast differentiation (right side of Fig. 7G). This occurred both in the presence of exogenous activators of the PKA pathway (forskolin and IBMX) and in their absence. The results in the absence of exogenous PKA activation likely reflect inhibition of basal PKA activity by PKIγ. These results, together with our previous demonstration that PKIγ is a primary mechanism that reduces the antiapoptotic effects of PKA signaling in osteoblasts [11], indicate that PKIγ is likely to be a major constraint on bone formation induced by Gαs-coupled receptors. Downregulation of PKIγ may therefore provide a useful cotherapy in combination with PKA agonists, such as intermittent PTH, for bone loss in conditions such as osteoporosis. Downregulation of PKIγ may be particularly useful for the approximately 20% patients who are unresponsive to therapy with intermittent PTH [64, 65]. Downregulation of PKIγ may also be useful to enhance repair of bone defects by mesenchymal precursor cells by programming their fate toward the osteoblastic lineage [66-68]. For example, our demonstration of relatively long-term knockdown of Pkig may provide a feasible approach to downregulate PKIγ ex vivo in mesenchymal precursor cells prior to their implantation in bone defects. The primary limitation of our study with regard to these potential clinical applications is that all our experiments were conducted in vitro and future studies will therefore be needed to assess the in vivo outcomes of Pkig deletion or knockdown.

This study also showed that activation of nuclear PKA by Pkig deletion or knockdown decreases adipocyte differentiation simultaneously with the increase of osteoblast differentiation (left side of Fig. 7G). Thus, the reciprocal regulation between differentiation of mesenchymal precursor cells into osteoblasts and adipocytes [27, 28] predominates over the direct effects of PKA signaling to stimulate adipocyte differentiation [19-22]. However, the direct effects of PKA signaling (gray arrow in Fig. 7G) likely partially counterbalance the reciprocal regulation, which may explain why Pkig deletion or knockdown has generally more modest effects on adipocyte differentiation than on osteoblast differentiation. Nonetheless, the decrease in adipocyte differentiation gives further credence to the possibility that downregulation of PKIγ may be useful to enhance repair of bone defects, especially in situations where adipogenesis impairs defect repair [67, 68].

This study further showed that the reciprocal regulation of osteoblast and adipocyte differentiation by PKIγ is mediated, at least in part, by LIF (middle of Fig. 7G). Thus, Pkig deletion increased PKA-induced Lif mRNA and LIF protein. Moreover, the effects of Pkig deletion or knockdown on osteogenesis and adipogenesis were inhibited by two different antibodies that, respectively, neutralize either murine or human LIF. Our results are consistent with prior studies that LIF can increase osteoblast differentiation [33-38] and inhibit adipogenesis [34, 37, 39, 40]. Interestingly, the effects of LIF in those studies are primarily on early events in differentiation [34, 36-38], perhaps reflecting regulation of mesenchymal lineage specification. It is likely that LIF acts together with other factors to mediate the reciprocal regulation of osteogenesis and adipogenesis by PKIγ. Future studies will be needed to identify those other factors and elucidate the details of how they interact with LIF to reciprocally regulate osteoblast and adipocyte differentiation.

This study also confirmed our previous results [10, 11] that PKIγ reduces PKA signaling as well as PKA-dependent phosphorylation of CREB and the resultant PKA-dependent expression of rapidly induced mRNAs. We previously characterized c-fos, Lif, and IL6 as immediate-early genes induced by PKA signaling [11, 42]. However, the quantitative real-time RT-PCR measurements of gene expression in this study distinguished between the mRNAs. c-fos showed a typical immediate-early pattern of gene expression both in the wild-type and Pkig/ MEFs. In contrast, Lif and IL6 mRNAs were induced and declined more slowly and this was especially pronounced in the Pkig/ MEFs.

Conclusion

In conclusion, our results show that Pkig reciprocally favors adipogenesis over osteogenesis and, together with our previous studies [9-11], demonstrate that PKIγ is a major regulator of PKA signaling in mesenchymal cells. PKA activity was recently shown to be regulated by miRNA-27a or miRNA-155 that targets Pkia in hematopoietic cells [69-71]. Intriguingly, miRNA-155 can induce osteoblast differentiation [72]. A transient increase in PKIβ level was also shown to be required for neuronal differentiation in cell culture [73]. Those results, together with our studies and the earlier publications on PKIα [12, 13], represent accumulating evidence that endogenous levels of the PKI family impact multiple cellular processes by regulating PKA activity.

Acknowledgments

CHO cells that stably express BMP4 were a kind gift from Dr. Tatsuya Koike, Osaka City University Medical School. This work was supported by NIH R21 AR055230 to EMG and the Harry E. Figgie III M.D. Professorship to EMG.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.

Ancillary