|Home | About | Journals | Submit | Contact Us | Français|
Mesenchymal stem cells are essential for repair of bone and other supporting tissues. Bone morphogenetic proteins (BMPs) promote commitment of these progenitors toward an osteoblast fate via functional interactions with osteogenic transcription factors, including Dlx3, Dlx5, and Runx2, and also can direct their differentiation into bone-forming cells. BMP-2-stimulated osteoblast differentiation additionally requires continual signaling from insulin-like growth factor (IGF)-activated pathways. Here we identify Akt2 as a critical mediator of IGF-regulated osteogenesis. Targeted knockdown of Akt2 in mouse primary bone marrow stromal cells or in a mesenchymal stem cell line, or genetic knockout of Akt2, did not interfere with BMP-2-mediated signaling but resulted in inhibition of osteoblast differentiation at an early step that preceded production of Runx2. In contrast, Akt1-deficient cells differentiated normally. Complete biochemical and morphological osteoblast differentiation was restored in cells lacking Akt2 by adenoviral delivery of Runx2 or by a recombinant lentivirus encoding wild-type Akt2. In contrast, lentiviral Akt1 was ineffective. Taken together, these observations define a specific role for Akt2 as a gatekeeper of osteogenic differentiation through regulation of Runx2 gene expression and indicate that the closely related Akt1 and Akt2 exert distinct effects on the differentiation of mesenchymal precursors.
Continual bone remodeling is necessary to maintain bone integrity and function throughout life. In the adult skeleton, optimal bone growth, remodeling, and repair require a balance between bone-forming osteoblasts and bone-resorbing osteoclasts (22, 43, 58). Osteoclasts are of hematopoietic origin, while osteoblasts are derived from pluripotent mesenchymal stem cells (43, 48, 50). With aging and under conditions such as osteoporosis, bone formation is diminished relative to the rate of resorption, leading to net bone loss and an increased risk of fractures (43, 48). Under both normal and pathological conditions, multiple local and systemic signals derived from hormones, growth factors, and other agents control different aspects of bone remodeling (43, 58). Among key factors that promote bone formation are bone morphogenetic proteins (BMPs) and insulin-like growth factors (IGFs) (30, 43, 58).
BMPs, members of the transforming growth factor β (TGF-β) superfamily, are potent inducers of osteoblast differentiation from mesenchymal progenitors (47) and have been used clinically to promote fracture repair (7, 25). BMPs bind to specific transmembrane type I and type II receptors and, by sequentially inducing receptor serine kinase activity, stimulate the intracellular mediators, Smad1, -5, and -8 (16), which transmit the BMP signal into the nucleus to regulate target gene transcription (31). Among osteogenic genes induced by BMP-activated Smads are those encoding the osteoblast determination and differentiation factors Dlx3, Dlx5, and Runx2 (14, 15, 28, 29). Runx2 in particular is a master regulator of osteoblast fate and function and is essential for bone formation (8, 17, 19). Its absence in mice results in a complete lack of bone (8, 17, 19), while heterozygous defects in humans lead to the bone deficiency disease cliedocranial dysplasia (62).
IGF actions also are important for normal bone growth and exert positive effects on bone size, bone formation, and bone density in mammals (20). Studies using genetically modified mice have shown that targeted overexpression of IGF1 in osteoblasts increased bone formation rates and enhanced trabecular and cortical bone volumes (61). In contrast, deleting the IGF1 receptor in osteoblasts caused impaired bone formation and reduced mineralization (60). The two IGFs, IGF1 and IGF2, bind to and signal through the IGF1 receptor, a ligand-activated tyrosine protein kinase, which activates multiple intracellular signaling pathways in osteoblasts and other cell types via adaptor molecules, including IRS1 and IRS2 (39). Previous studies from our laboratory have demonstrated that IGF binding protein 5 (IGFBP-5) functions to inhibit IGF actions in bone cells and blocks both osteoblast differentiation and bone growth (36), further illustrating the key positive role of intact IGF signaling in bone development and function.
Despite the multiplicity of signaling cascades activated by the IGF1 receptor, growing evidence points to the phosphatidylinositol 3-kinase (PI3-kinase)-Akt pathway as being critical for IGF actions in bone (11-13, 27, 37, 44). For example, we have shown that this pathway is required for BMP-2-mediated osteoblast differentiation in culture and for ex vivo bone growth and development (37). Similarly, osteoblast-specific genetic deletion of Pten, an endogenous PI3-kinase pathway inhibitor, resulted in both enhanced bone formation and osteoblast differentiation (32).
The three Akts found in mammals are structurally and functionally similar serine-threonine protein kinases that are the products of related genes (3, 33). Akt1 and Akt2 are ubiquitously expressed, while Akt3 has a more restricted tissue distribution (3, 33). Combined genetic knockout of Akt1 and Akt2 in mice resulted in defective osteogenesis along with severe deficiencies in other organs and tissues, including skeletal muscle, that led to early postnatal death (41). Global deficiency of Akt1 in mice was associated with low-turnover osteopenia with mild reductions in bone mass that were attributable primarily to defective coupling between osteoblasts and osteoclasts (21), while systemic loss of Akt2 or Akt3 in mice has not been linked to a bone phenotype (5, 10).
Based on our recent observations that a dominant negative Akt inhibited osteoblast differentiation and bone growth (37), we sought to define the specific roles of individual Akts in osteogenesis. We now find that Akt2 is required for BMP-2-mediated osteoblast differentiation from mesenchymal precursors, including bone marrow-derived stem cells, while Akt1 appears to be dispensable. Loss of Akt2 prevents induction of Runx2 gene expression in these cells, and forced expression of Runx2 restores normal differentiation. The actions of Akt2 to promote osteogenic development contrast with its limited role in myogenesis from the same mesenchymal progenitors, in which Akt1 is a critical signaling molecule for differentiation (46, 55). Our studies thus provide evidence for distinct biological roles for Akt1 and Akt2 in mesenchymal cell differentiation.
Fetal calf serum, newborn calf serum, horse serum, Dulbecco's modified Eagle's medium (DMEM), and phosphate-buffered saline (PBS) were purchased from Mediatech-Cellgrow (Herndon, VA). Okadaic acid was from Alexis Biochemicals (San Diego, CA); protease inhibitor tablets and nitroblue tetrazolium (NBT)-5-bromo-4-chloro-3-indolylphosphate (BCIP) tablets were from Roche Applied Sciences (Indianapolis, IN). Restriction enzymes, buffers, ligases, and polymerases were purchased from Roche Applied Sciences, BD Biosciences-Clontech (Palo Alto, CA), and Fermentas (Hanover, MD). Alizarin red, sodium orthovanadate, ascorbic acid, and β-glycerol phosphate were from Sigma-Aldrich (St. Louis, MO), and trypsin-EDTA solution and the Superscript III first-strand synthesis kit were from Invitrogen (Carlsbad, CA). AquaBlock EIA/WIB solution was from East Coast Biologicals (North Berwick, ME). The bicinchoninic acid (BCA) protein assay kit was purchased from Pierce Biotechnologies (Rockford, IL). Immobilon-FL was from Millipore Corporation (Billerico, MA). The production and use of recombinant BMP-2 have been described previously (36). Other chemicals and reagents were purchased from commercial suppliers.
The following polyclonal antibodies were purchased from commercial vendors: anti-Smad1, anti-Akt, anti-phospho-Akt (Ser473), and anti-Akt2, Cell Signaling Technology (Beverly, MA); anti-α-tubulin, Sigma Aldrich (St. Louis, MO); anti-Dlx5 and anti-phospho-Smad1, -5, and -8, Santa Cruz Biotechnology (Santa Cruz, CA); and anti-Runx2, R&D Systems (Minneapolis, MN). The monoclonal anti-Akt1 antibody was from Abcam (Cambridge, United Kingdom). Goat anti-rabbit IgG-IR800 and goat anti-mouse IgG-IR680 were from Rockland Immunochemical (Gilbertsville, PA).
The following adenoviruses (Ad) have been described previously: Ad-βgal (53) and Ad-shAkt1 (short hairpin RNA [shRNA] against mouse Akt1) and Ad-shAkt2 (55). A recombinant adenovirus that expressed the coding region of mouse Runx2 (type II isoform) was prepared by standard procedures, as described previously (36). The Runx2 cDNA was generated by reverse transcription-PCR (RT-PCR) from RNA isolated from mouse calvarial osteoblasts (PCR primers are as follows: sense strand, 5′-AAGTCGACATGCTTCATTCGCCTCACAAACAACCAC-3′ [ATG codon underlined]; antisense strand: 5′-AAGATATCTCATATGGCCGCCAAACAGACTCATCC-3′). Viruses were purified by centrifugation through CsCl density gradients, and titers were determined by optical density (54). Prior to use, all adenoviruses were diluted in DMEM plus 2% fetal calf serum and filtered through a Gelman syringe filter (0.45 μM).
Recombinant lentiviruses expressing the coding regions of enhanced green fluorescent protein (EGFP), mouse Akt1, or mouse Akt2 were generated. The Akt cDNAs were cloned into the PmeI and SmaI sites of the lentiviral plasmid pWXPL (Addgene Inc., Cambridge, MA). Recombinant lentiviruses were prepared by transfecting the pWXPL plasmid along with packaging plasmids (Addgene Inc., Cambridge, MA) into Hek293FT cells. Lentiviruses were concentrated by centrifugation of the lentiviral supernatant at 20,000 × g at 4°C for 2 h. The pellet was resuspended in PBS with 0.1% bovine serum albumin (BSA), aliquoted, and stored at −80°C.
All cells were incubated at 37°C in humidified air with 5% CO2. Mouse embryo fibroblasts (MEFs) from wild-type mice and mice lacking either Akt1 or Akt2 have been described previously (46). Primary bone marrow stromal cells (MSCs) were isolated and propagated from 8-week old male mice as described previously (49). Confluent MEFs, MSCs, and C3H10T1/2 cells (ATCC catalog no. CCL-226) were incubated in osteogenic medium (OM) (DMEM, 10% fetal calf serum, 50 μg/ml ascorbic acid, 10 mM β-glycerol phosphate, and 200 ng/ml BMP-2). OM was replaced every 48 h for up to 12 days. For infections of MSCs and C3H10T1/2 cells with Ad-βgal, Ad-shAkt1, or Ad-shAkt2, viruses were added to cells at ~50% of confluent density at a multiplicity of infection (MOI) of 750. One day later, cells were washed with phosphate-buffered saline and OM was added. For infections of C3H10T1/2 cells with Ad-shAkt2 plus Ad-Runx2 or with Ad-shAkt2 plus Ad-βgal, each virus was added to cells at ~50% of confluent density at an MOI of 500 to 750, and 1 day later, cells were washed and OM added. MEFs were infected with Ad-Runx2 or Ad-βgal at ~50% of confluent density, and cells were washed and OM added 1 day later. Cells were infected with lentiviruses at ~10 to 15% of confluent density by adding the lentiviral solution to DMEM supplemented with Polybrene (6 μg/ml) and 10% fetal calf serum. One day later, cells were washed and growth medium (DMEM with 10% fetal calf serum) was added. Adenoviral infections of C3H10T1/2 cells were performed the following day, and cells were incubated in OM 1 day later, as described above. For MEFs, OM was added at 3 days after lentiviral infection.
For measurement of alkaline phosphatase activity, cells were washed with PBS, fixed with 70% ethanol for 10 min, and incubated with 500 μl of NBT-BCIP solution (one tablet in 10 ml distilled water) for 20 min at 20°C (36). After three washes with distilled water, images were captured and analyzed with the LiCor Odyssey infrared imaging system, using software version 1.2 (LiCor, Lincoln, NE). For detection of mineralization, cells were fixed in 70% ethanol for 10 min and stained with 2% Alizarin red solution (pH 4.1 to 4.5) for 1 min at 20°C (36). Images were obtained by scanning plates with the LiCor Odyssey system.
Whole-cell RNA (2 μg), isolated as described previously (36), was reverse transcribed with the Superscript III first-strand synthesis kit using oligo(dT) primers in a final volume of 20 μl. PCR was performed with 1 μl of cDNA per reaction mixture and previously published primer pairs for mouse Sox9, Dlx5, Runx2, osterix, osteocalcin, and S17 (36) and Akt1 and Akt2 (55). Mouse Id1 primers are as follows: sense strand, 5′-GAGTCTGAAGTCGGGACCAC-3′; antisense strand, 5′-ATGCGCCTGAAAAGTAAGGA-3′. Cycle numbers for PCR varied from 20 to 30 and were within the linear amplification range for each primer pair. Results were visualized after agarose gel electrophoresis.
Whole-cell protein lysates were prepared as described previously (36, 38), and aliquots were stored at −80°C until use. Protein samples (20 to 30 μg/lane) were resolved by SDS-PAGE and transferred to Immobilon-FL membranes. After blocking with 25 to 50% AquaBlock solution for 1 h at 20°C, membranes were incubated sequentially with primary and secondary antibodies (36). The following primary antibodies were used at a dilution of 1:1,000: anti-Akt; anti-Akt2; anti-phospho-Akt (Ser473); anti-Runx2; anti-phospho-Smad1, -5, and -8; and anti-Smad. Anti-Akt1 was used at 1:2,000, anti-α-tubulin at 1:15,000, and anti-Dlx5 at 1:2,000. Secondary antibodies were used at 1:5,000. Results were visualized and images captured using the LiCor Odyssey system and version 1.2 analysis software.
We recently showed that BMP-2-initiated osteoblast differentiation of mouse mesenchymal stem cells was blocked by the PI3-kinase inhibitor LY294002 but could be restored to normal by infection with a recombinant adenovirus encoding an inducible-active Akt (37). We also found that both osteoblast differentiation and the growth and mineralization of mouse metatarsal bones in short-term primary culture were inhibited by an adenovirus encoding a dominant negative Akt (37). The goal of current experiments was to establish how Akt-mediated signaling promoted osteogenic differentiation.
Of the three mammalian Akts, Akt1 and Akt2 are found in most tissues outside of the central nervous system, while Akt3 is more restricted in its distribution (3, 33). We first assessed Akt expression in the pluripotent C3H10T1/2 mesenchymal stem cell line under conditions in which BMP-2 stimulated osteogenesis. We found that the levels of Akt1 and Akt2 mRNA and protein remained relatively constant during a 7-day period in which BMP-2 caused a rapid and sustained increase in phosphorylation of Smad1, -5, and -8, along with expression of bone-specific transcription factors Dlx5 and Runx2. In contrast, minimal Akt3 mRNA was detected (data not shown). BMP-2 rapidly and potently induced mRNA expression of the BMP- and Smad-dependent genes for Id1 and Sox9, as well as for Dlx5 and Runx2 (Fig. 1A and B), osteoblast-specific transcription factors whose genes are known targets of BMP-2 (47). Other observations confirmed that treatment with BMP-2 induced osteogenesis in this cellular model system, as transcripts for the osteoblast transcription factor osterix (Osx) and the secreted osteoblast protein osteocalcin (Ocn) increased markedly (Fig. (Fig.1B),1B), as did activity of bone-specific alkaline phosphatase (AP) and mineralization of extracellular matrix, as measured by Alizarin red staining (Fig. (Fig.1C).1C). Thus, despite the apparently important role of Akt in osteogenic differentiation in these cells (37), neither Akt1 nor Akt2 changed in abundance.
We next asked if loss of either Akt1 or Akt1 could modify osteogenic differentiation. C3H10T1/2 cells were infected with a control adenovirus (for β-galactosidase, Ad-βgal) or with adenoviruses encoding short hairpin interfering RNAs for either mouse Akt1 or Akt2 (Fig. (Fig.2A).2A). As pictured in Fig. Fig.2B,2B, each individual Akt knockdown was effective and specific and did not lead to compensatory upregulation of the other Akt isoform. In addition, Akt phosphorylation was reduced by ~50%. Akt deficiency did not reduce BMP-2-stimulated Smad phosphorylation or BMP-2-induced expression of mRNAs for Id1, Sox9, or the osteoblast transcription factor Dlx5 (Fig. 2B and C). Dlx5 protein also accumulated in the absence of Akt1 or Akt2 (Fig. (Fig.2B).2B). Cells infected with Ad-βgal or with Ad-shAkt1 underwent rapid and extensive osteoblast differentiation, as indicated by sequential induction of transcripts for Runx2, Osx, and Ocn and by progressively increasing AP activity and mineralization (Fig. 2C to E). In contrast, cells infected with Ad-shAkt2 did not express Runx2, Osx, or Ocn mRNAs or Runx2 protein; had minimal AP activity; and failed to undergo mineralization (Fig. 2B to E), even when incubated for up to 10 days in OM (data not shown).
To confirm and extend these results, analogous experiments were performed with mouse bone marrow stromal cells (MSCs) in primary culture (Fig. (Fig.3A).3A). Each Ad-shAkt effectively reduced the abundance of the respective protein without any compensatory changes in the other Akt and led to an ~50% decline in Akt phosphorylation but did not cause any alterations in BMP-2-mediated Smad phosphorylation or gene activation of Id1, Sox9, or Dlx5 (Fig. 3B and C). MSCs infected with either Ad-βgal or Ad-shAkt1 differentiated normally over a 12-day time course, as measured by progressive expression of Runx2, Osx, and Ocn mRNAs; AP activity; and mineralization (Fig. 3C and D). In contrast, knockdown of Akt2 resulted in markedly impaired biochemical and morphological differentiation (Fig. 3C and D), as indicated by reduced Runx2 gene expression and by minimal AP activity and mineralization. Thus, in primary MSCs, as in the C3H10T1/2 mesenchymal stem cell line, Akt2 appeared to be required for osteogenic differentiation.
To further extend our observations, we evaluated the effects of a total absence of Akt1 or Akt2 on BMP-2-mediated osteogenic differentiation by studying mouse embryo fibroblasts (MEFs) derived from mice genetically deficient in either Akt1 or Akt2 (5, 6). Loss of Akt1 or Akt2 did not lead to upregulation of the other Akt, and individual Akt protein expression was constant during a 7-day incubation in OM with BMP-2 in both wild-type control and knockout cells. In addition, in each knockout cell line Akt phosphorylation was reduced by ~50% compared with that in wild-type cells (Fig. (Fig.4B).4B). As was observed in C3H10T1/2 cells and in MSCs, both BMP-2-stimulated Smad phosphorylation and upregulation of Dlx5 protein and Id1, Sox9, and Dlx5 gene expression were unaltered by Akt1 or Akt2 deficiency (Fig. 4B and C). As seen in Akt2 knockdown cells, genetic loss of Akt2 completely prevented osteogenic differentiation, as demonstrated by minimal Runx2, Osx, or Ocn gene expression; little Runx2 protein production; and absent AP activity and mineralization (Fig. 4B to D). In contrast, osteogenesis of Akt1-deficient MEFs was similar to that of wild-type cells (Fig. 4C and D). Thus, based on the results pictured in Fig. Fig.22 to to4,4, we conclude that Akt2 but not Akt1 is required for BMP-2-mediated osteogenic differentiation.
To further test the hypothesis that Akt2 plays a critical role in osteogenesis, we next asked if a lentivirus encoding a wild-type mouse Akt2 that was resistant to siRNA-mediated knockdown could promote differentiation in cells deficient in Akt2 (Fig. (Fig.5A5A and and6A).6A). In Akt2-deficient cells, infection with either lentiviral Akt1 or Akt2 increased total Akt concentrations to equivalent degrees and led to similarly enhanced Akt phosphorylation. In contrast, lentiviral EGFP was ineffective (Fig. (Fig.5B5B and and6B).6B). Addition of Akt2 restored normal osteoblast gene induction (Fig. (Fig.5C5C and and6C),6C), simulated Runx2 production (Fig. (Fig.5B5B and and6B),6B), and promoted both AP activity and mineralization to levels nearly identical to those of control osteoblasts after incubation for 7 days in BMP-2 plus OM (Fig. (Fig.5D5D and and6D).6D). In contrast, addition of Akt1 was unable to restore any aspect of osteogenesis (Fig. (Fig.55 and and6).6). Thus, based on these observations, Akt2 signaling is necessary for osteoblast differentiation.
Runx2 is a key transcription factor for osteoblast differentiation and bone formation (8). BMP-2 stimulates Runx2 gene and protein expression by several mechanisms, including indirectly via the transcription factors Dlx3 and Dlx5 (14, 29) and directly through active Smad1, -5, and -8, which have been shown to bind to the osteoblast-specific Runx2 promoter and activate its gene transcription (42). In our experiments, Akt2 deficiency did not perturb BMP-2-stimulated Smad activation (Fig. (Fig.2B,2B, ,3B,3B, and and4B);4B); did not impair induction of the Smad target genes Id1, Sox9, and Dlx5 (Fig. (Fig.2C,2C, ,3C,3C, and and4C);4C); and did not block production of Dlx5 protein (Fig. (Fig.2B2B and and4B).4B). However, Runx2 gene and protein expression were inhibited (Fig. (Fig.2C,2C, ,3C,3C, and and4C)4C) but could be restored to normal by Akt2 (Fig. (Fig.5B5B and and6B).6B). Based on these observations, we tested the hypothesis that the critical step in osteogenic differentiation blocked by lack of Akt2 was production of Runx2. We introduced recombinant adenoviruses encoding mouse Runx2 or a control protein (βgal) into Akt2-deficient cells and assessed osteoblast differentiation over a 7-day interval (Fig. (Fig.7A7A and and8A).8A). Under the conditions of these experiments, Ad-Runx2 restored Runx2 mRNA and protein to levels similar to those of control Akt2-replete cells (Fig. (Fig.7B7B and and8B).8B). Neither Ad-Runx2 nor Ad-βgal enhanced Akt1 expression (Fig. (Fig.7B).7B). Ad-Runx2 but not Ad-βgal stimulated production of Osx and Ocn mRNAs (Fig. (Fig.7C7C and and8B),8B), induced AP activity, and promoted mineralization to levels similar to what was seen in control cells (Fig. (Fig.7D7D and and8C).8C). Based on these results, we conclude that at least in two cell models, Akt2 acts upstream of Runx2 in an osteogenic signaling cascade.
BMPs play a key role in osteogenic differentiation and bone development and can drive uncommitted mesenchymal precursor cells toward the osteoblast lineage (30, 47). In previous studies we have defined cross talk between signaling steps regulated by BMP-2 and the IGF1-activated PI3-kinase-Akt pathway that are required to initiate osteogenic differentiation, to promote maturation of committed osteoblasts, and to regulate longitudinal bone growth by enhancing both chondrocyte and osteoblast development and function (37). The current experiments were designed to determine if individual Akt isoforms played specific roles in osteogenesis. We find that Akt2 is required for BMP-2-stimulated osteoblast differentiation but that Akt1 is dispensable. The critical step regulated by Akt2 appears to be induction of gene expression of the osteoblast-specific transcription factor Runx2, and osteoblast differentiation could be restored to normal in Akt2-deficient cells by a recombinant adenovirus encoding Runx2. Taken together, our results demonstrate that signaling by IGF-activated Akt2 but not Akt1 is selectively required to promote osteoblast differentiation.
The three Akts found in mammals are closely related serine-threonine protein kinases that are activated by a variety of growth factors and hormones via stimulation of production of phosphatidylinositol 3,4,5-trisphosphate (PIP3) by the class Ia PI3-kinases (3, 33). Despite similarities in overall structure and kinase activity of the three Akts, a growing literature supports their selective functions in cells and in whole-organism physiology, with initial insights coming from individual gene knockout experiments with mice. For example, mice lacking Akt1 exhibit a postnatal growth deficiency (6), while mice deficient in Akt2 develop impaired glucose tolerance (5) and mice without Akt3 have defective brain development (10). More complicated phenotypes have resulted from combinatorial knockouts and demonstrate that neither Akt2 nor Akt3 by itself is able support postnatal life (9). In contrast, mice expressing only Akt1 were viable but had significant deficits in somatic growth and in carbohydrate metabolism (9).
Additional insights toward elucidating distinct functions for each Akt have been made through the study of individual cell types and tissues derived from Akt-deficient mice or generated by targeted gene knockdowns. For example, vascular smooth muscle cell differentiation in culture appears to require activation of Akt2 by IGF1-stimulated PI3-kinase, with Akt1 being dispensable (34), while in contrast, adipocyte differentiation of the 3T3-L1 cell line just needs sustained activation of Akt1 (56), as does differentiation of primary keratinocytes (52). As a further variation, activity of both Akt1 and Akt2 appear to be necessary for complete differentiation of primary bone marrow-derived macrophages into osteoclasts (51). However, in none of these examples have the precise signaling steps that uniquely distinguish the mechanisms of action of one Akt from another been characterized, nor have cell-type-specific modifiers that might explain distinct biological effects of a single Akt within a background of otherwise shared signaling pathways been identified.
The anabolic effects of IGF1 on bone growth and skeletal mineral accretion have been well documented (20). In our previous studies, we demonstrated that IGFBP-5, the PI3-kinase inhibitor LY294002, and adenovirus-delivered dominant negative Akt were each able to equivalently block osteogenic differentiation of cultured cells and prevent longitudinal growth and mineralization of neonatal mouse metatarsal bones incubated in serum-free medium (36, 37), confirming that locally produced IGFs are responsible for activating the PI3-kinase-Akt pathway in bone, and showing that this signaling pathway is essential for both skeletal development and growth. The skeletal phenotype in our metatarsal model resembled the delayed ossification, osteopenia, and reduced long bone growth observed in mice with engineered IGF1 receptor deficiency (60) and also seen with a combined systemic knockout of Akt1 and Akt2 (41). An opposite phenotype resulted in osteoblasts from mice lacking Pten, the endogenous PI3-kinase inhibitor, in which bone mineral density and volume increased progressively throughout life (32), thus further underscoring the importance of IGF-mediated PI3-kinase-Akt signaling in bone growth and anabolism.
Very few studies have examined the actions of individual Akts in bone, and our nearly identical results with three different cell models provide strong support for Akt2 being a central signaling factor in BMP-stimulated osteogenic differentiation. Despite the congruency of our observations, our results are seemingly at odds with studies by Kawamura et al., who found that mice lacking Akt1 exhibited mild osteopenia secondary to defects in both osteoblast and osteoclast function (21). However, their in vitro observations, showing that differentiation of Akt1-deficient calvarial osteoblasts in response to BMP-2 was normal, are consistent with our data, although in the absence of serum the osteoblasts lacking Akt1 were more susceptible than controls to apoptotic death (21). In our experiments, osteoblast differentiation was studied in osteogenic medium containing 10% fetal calf serum, and minimal cell death occurred, with no increases seen in the absence or Akt1 or Akt2 (data not shown) or in the presence of a dominant negative Akt (37). Direct analyses of mice with chondrocyte-, osteoblast-, or osteoclast-specific loss of Akt1 or Akt2 will help establish the specific roles of each of these protein kinases in different aspects of bone development and function throughout the life span.
Loss of Akt2 or Akt1 in our experimental models did not appear to inhibit the actions of BMP-2. Phosphorylation of BMP receptor-activated Smads was normal, as was the induction of BMP-2- and Smad-dependent target genes, including Id1 (expressed in many cell types ), Sox9 (specific to chondrocytes and osteoblasts ), and most importantly Dlx5 (expressed early in committed osteoblast precursors (23). However, despite accumulation of Dlx5 protein and functional Smads, which collaboratively can stimulate Runx2 gene transcription (18, 28, 29, 42), little Runx2 mRNA or protein was produced in the absence of Akt2. Since lentiviral Akt2 could promote Runx2 expression and restore normal osteoblast differentiation in cells lacking Akt2, as could adenovirus-delivered Runx2, our results in aggregate indicate that signaling pathways downstream of Akt2 are required for induction of Runx2 gene expression, although they do not identify the relevant molecular mechanisms.
The activity of Runx2 also may be regulated by Akt-mediated signaling pathways. Fujita et al. found that a chemical inhibitor of PI3-kinase or a dominant negative Akt reduced the ability of Runx2 to bind directly to DNA in vitro or to DNA in chromatin in cells (11). Other studies have shown that the intracellular enzyme glycogen synthase kinase 3β (GSK-3β), can negatively regulate both the DNA binding and transcriptional actions of Runx2 through inhibitory phosphorylation, which could be the mechanism whereby active GSK-3β impairs osteoblast differentiation (24). As GSK-3β itself is a direct target of inhibitory phosphorylation by Akt (40), it is possible that Akt2 could promote osteoblast differentiation in several ways: first, by stimulating Runx2 gene expression, and second, by enhancing the activity of Runx2 by reducing the effects of inhibitors. Future experiments will be designed to test these and other possibilities.
Mesenchymal progenitors can be directed toward the adipoblast, chondroblast, myoblast, and osteoblast lineages through combinatorial interactions of cell-type-specific transcription factors with environmental cues directed by locally derived growth factors (4). In previous studies using MEFs from mice lacking Akt1 or Akt2, we found that Akt1 activity was required for initiation of muscle differentiation but that Akt2 was dispensable (46). In contrast, as shown here, Akt2 but not Akt1 was needed for osteoblast differentiation. Thus, the same precursor cell became either responsive or refractory to differentiation-inducing signals depending on which Akt was expressed and activated by an apparently identical IGF-stimulated pathway.
In summary, our studies define a key role for Akt2 in osteogenesis and provide examples of how IGF-mediated signaling pathways differentially control mesenchymal cell fate and function. Signaling through the IGF1 receptor also has been associated with enhanced aging in experimental animals (45, 57) and with increased cancer risk in humans and animals (26), yet treatment with IGF1 has been proposed for bone and muscle disorders in humans (1, 2). Further understanding of the biochemical and molecular mechanisms of IGF action will be needed to separate the therapeutic and pathogenic effects of this potent growth factor.
We thank Morris Birnbaum of the University of Pennsylvania for wild-type and Akt-deficient MEFs.
These studies were supported in part by National Institutes of Health grant R01 DK42748 (to P.R.).
Published ahead of print on 7 December 2009.