Adult Human Mesenchymal Stem Cell Differentiation to the Osteogenic or Adipogenic Lineage Is Regulated by Mitogen-activated Protein Kinase*

Adult human mesenchymal stem cells are primary, multipotent cells capable of differentiating to osteo-cytic, chondrocytic, and adipocytic lineages when stimulated under appropriate conditions. To characterize the molecular mechanisms that regulate osteogenic differentiation, we examined the contribution of mitogen-activated protein kinase family members, ERK, JNK, and p38. Treatment of these stem cells with osteogenic supplements resulted in a sustained phase of ERK activation from day 7 to day 11 that coincided with differentiation, before decreasing to basal levels. Activation of JNK occurred much later (day 13 to day 17) in the osteogenic differentiation process. This JNK activation was associated with extracellular matrix synthesis and increased calcium deposition, the two hallmarks of bone formation. Inhibition of ERK activation by PD98059, a specific inhibitor of the ERK signaling pathway, blocked the osteogenic differentiation in a dose-dependent manner, as did transfection with a dominant negative form of MAP kinase kinase (MEK-1). Significantly, the block-age of osteogenic differentiation resulted in the adipogenic differentiation of the stem cells and the expression of adipose-specific mRNAs peroxisome pro-liferator-activated receptor g 2, aP2, and lipoprotein lipase. These observations provide a potential mechanism involving MAP kinase activation in osteogenic differentiation of adult stem cells and suggest that commitment of hMSCs into osteogenic or adipogenic lineages is governed by activation or inhibition of ERK, polymerization. For JNK assays, GST-c-Jun (amino 1–79) fusion protein was purified by glutathione-agarose chroma- tography of Escherichia coli extracts containing the expression plasmid pGEX2T-c-Jun, and 40 m g/ml was co-polymerized with the SDS-PAGE solution. After the denaturation/renaturation procedure, kinase reac-tions were performed by incubating the gel with kinase buffer (20 -glycerophosphate, 1.0 dithiothreitol, 0.1 EGTA, containing 200 of [ g - 32 P]ATP. The gel was incubated fo r and autoradiograms were developed after washing several trichloroacetic acid and sodium pyrophosphate to remove free isotope. Phosphorylated bands were excised from the gel, and incorpo- rated radioactivity was measured by Cerenkov counting.

cultured under defined in vitro conditions (1)(2)(3). The hMSCs do not differentiate spontaneously, and their in vitro and in vivo osteogenic potential has been very well characterized by us and others (4 -6). When cultured in the presence of the synthetic glucocorticoid dexamethasone, ascorbic acid, and ␤-glycerophosphate (osteogenic supplements, OS), hMSCs differentiate to the osteogenic lineage, producing bone-like nodules with a mineralized extracellular matrix containing hydroxyapatite (4). The similar developmental phenomenon has also been described by others (7,8) using bone marrow-derived cells. Other than the osteoinductive effect that OS has on MSCs, OS also acts as a mitogen (9). Presumably, the osteoinductive and mitogenic effects are due to dexamethasone present in OS because glucocorticoids are potent regulators of cellular growth and differentiation (10). However, the underlying molecular mechanisms of OS-induced mitogenic and osteogenic differentiation are presently unknown.
In order to acquire a new cell phenotype, uncommitted hM-SCs must undergo proliferative and differentiative changes, the two most fundamental biological processes in the life cycle of cells. One of the potential signal transduction pathways that might regulate the proliferation and differentiation of hMSCs is the MAP kinase pathway. Activation of the MAP kinase pathway in other cell types such as neuronal cells, adipocytes, T-cells, and muscle cells promotes cell differentiation (11,12). Most recently, a role for MAP kinase in chondrogenesis of limb bud mesenchyme has also been demonstrated (13). The first discovered members of the MAP kinase family, ERK1 and ERK2 (extracellular signal-regulated kinase), which were first identified by Ray and Sturgill (14), are predominantly activated by growth factors via a Ras-dependent signal transduction pathway (15). ERKs are activated by dual phosphorylation on tyrosine and threonine residues separated by a glutamate residue (TEY) by a single upstream kinase known as MEK (MAP kinase or ERK kinase). Other than growth factors, hormones such as estrogen and parathyroid hormone, which have a profound effect on bone development and remodeling, will also transiently activate ERK1 in MCF-7 cells and ERK2 in the osteosarcoma cell line UMR 106-01 (16,17). Activation of ERK1/ERK2 has also been demonstrated as an important signaling mechanism in the differentiation of the preadipocyte cell line 3T3-L1 into mature adipocytes (18). The ERK phosphorylation of peroxisome proliferator-activated receptor ␥ (PPAR␥), which is a key component of transcription machinery of adipogenic differentiation and selectively expressed in adipocytes, in vitro as well as in vivo, results in reduction of PPAR␥ transcriptional activity (19,20). Growth factors and cytokines, such as tumor necrosis factor-␣, which activate MAP kinases (21,22), are potent inhibitors of adipocyte differentiation (23)(24)(25). The relationship between MAP kinases and adipogenesis is further strengthened by the recent demonstration that leptin, product of the ob gene, induced activation of MAP kinase in mouse embryonic cell line C3H10T1/2 (26). Leptin-induced MAP kinase activation may cause a reduction in the adipogenic differentiation by phosphorylating PPAR␥ and thus might play an important role in controlling adipogenesis.
Recently, two new members of the MAP kinase family have been described as follows: the c-Jun N-terminal kinase (JNK), also known as stress-activated protein kinase, and p38-reactivating kinase (p38 RK or simply, p38). These kinases can be activated by a variety of cytokines, environmental stress, as well as ultraviolet and ionizing radiation (27). Transforming growth factor-␤ stimulates ERK (28) as well as p38 (29) and acts as a physiological regulator of osteoblast differentiation and bone remodeling by regulating and coordinating the activities of osteoblasts and osteoclasts (30). Cytokines such as tumor necrosis factor-␣, interleukin-1, and interleukin-6, which are important regulators of bone resorption, activate JNK and p38 in human and mouse osteoblastic cells (31,32).
Despite the rapid progress in the elucidation of the signaling pathways involving MAP kinases, the most crucial question of how the specificity and integration of multiple signals are processed in parallel remains unanswered. In the present study, we demonstrate that activation of MAP kinases in hM-SCs, cultured in OS medium, resulted in their differentiation into osteocytes and that inhibition of the MAP kinase pathway by PD98059, a specific MEK inhibitor (33), caused these cells to develop into fully mature adipocytes. These results provide a potential mechanism of action for ERK in regulating the lineage commitment of multipotential adult stem cells. These results may also have important clinical implications during the aging process or diseases, such as osteoporosis, where bone marrow stroma is progressively replaced by fat (34).

EXPERIMENTAL PROCEDURES
Materials-The monoclonal MAP kinase antibody raised against a C-terminal synthetic peptide (amino acids 324 -345) was purchased from Zymed Laboratories Inc. Laboratory, San Francisco, CA. Phosphospecific MAP kinase and phosphospecific p38 MAP kinase antibodies were from New England Biolabs, Beverly, MA. MEK1 inhibitor PD98059 was purchased from Calbiochem. [␥-32 P]ATP, ECL detection kit, horseradish peroxidase-conjugated mouse and rabbit IgG were from Amersham Pharmacia Biotech. Alkaline Phosphatase Diagnostic Kit 85 and Calcium Diagnostic Kit 587 were purchased from Sigma. All other chemicals were obtained commercially and were of the highest purity available.
Cell Culture and Preparation of Lysates-Bone marrows were obtained from normal human volunteer donors (age 23-40 years) after informed consent. hMSCs were purified from the marrow by Percoll density gradient centrifugation method (2,3,6). The hMSCs were cultured in 100-mm Petri dishes and treated with osteogenic supplements (OS) as described earlier (4). To make lysates, cells were washed two times with cold PBS and were suspended in 0.5 ml of lysis buffer (20 mM Tris-HCl, pH 7.5, 1.0 mM EDTA, 1.0 mM EGTA, 1.0 mM dithiothreitol, 150 mM NaCl, 1% Triton X-100, 1.0 mM sodium orthovanadate, 10 mM NaF, 25 mM p-nitrophenyl phosphate, 0.1% SDS, 0.5 mM phenylmethylsulfonyl fluoride, and 2.0 g/ml aprotinin and leupeptin). The SDS was omitted from lysis buffer in samples prepared for kinase assays. The lysates were sonicated briefly (2 ϫ 20 s) on ice and centrifuged at 100,000 ϫ g for 35 min. Protein concentrations were determined using the bicinchoninic acid assay (Pierce). In each experiment, control and OS-treated cells were processed in parallel. In experiments using the MEK1 inhibitor PD98059, the 50 mM stock solution was prepared in Me 2 SO. The final concentration of Me 2 SO never exceeded 0.1%, and the same amount of Me 2 SO vehicle was added to control wells.
Immunoblot Analysis-Cell lysates containing 20 g of protein were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore Corp., Bedford, MA). The membranes were incubated for 2 h at room temperature in blocking buffer (5% non-fat dry milk in Tris-buffered saline ϩ 0.2% Triton X-100) and then incubated overnight at 4°C with various antibodies (1:1000). Antigen-antibody complexes were visualized by incubation of the blots in a 1:5000 dilution of horseradish peroxidase-conjugated goat anti-mouse or goat antirabbit immunoglobulin G and the ECL detection system (Amersham Pharmacia Biotech).
Alkaline Phosphatase and Calcium Assays-Alkaline phosphatase activity was assayed by measuring the formation of p-nitrophenol from p-nitrophenyl phosphate as described previously (4). Alkaline phosphatase histochemistry was performed using Sigma Diagnostic Kit 85. Selected specimens were also stained for mineral deposition by the von Kossa method (4). Accumulated calcium was measured in 0.5 N HCl extracts according to the manufacturer's instructions contained in Sigma Diagnostic Kit 587. Total calcium was calculated from standard solutions prepared in parallel and expressed as g/mg cellular protein. Cell numbers were determined by using the nuclear dye, crystal violet (35).
Kinase Assay-In-gel kinase assays were performed using whole cell lysates (20 g of protein) on SDS-PAGE separated proteins that were renatured in the gel according to the procedure of Kamehita and Fugisawa (36). Polyacrylamide (10%) gels were cast with 0.5 mg of myelin basic protein (MBP) per ml added as substrate to the SDS-PAGE gel solution prior to polymerization. For JNK assays, GST-c-Jun (amino acids 1-79) fusion protein was purified by glutathione-agarose chromatography of Escherichia coli extracts containing the expression plasmid pGEX2T-c-Jun, and 40 g/ml was co-polymerized with the SDS-PAGE solution. After the denaturation/renaturation procedure, kinase reactions were performed by incubating the gel with kinase buffer (20 mM Tris-HCl, pH 7.2, 20 mM MgCl 2 , 15 mM ␤-glycerophosphate, 1.0 mM dithiothreitol, 0.1 mM EGTA, and 0.5 mM sodium vanadate) containing 200 Ci of [␥-32 P]ATP. The gel was incubated for 3 h at room temperature, and autoradiograms were developed after washing several times in 5% trichloroacetic acid and 1% sodium pyrophosphate to remove free isotope. Phosphorylated bands were excised from the gel, and incorporated radioactivity was measured by Cerenkov counting.
Plasmid Constructions and Co-transfections-The epitope-tagged (EEEEYMPME, termed "EE") wild type (MEK-WT), dominant negative (S218A, S222A; MEK-2A), and constitutively active (S218E, S218E; MEK-2E) rat MEK-1 vectors driven by cytomegalovirus promoters (62) were obtained from Dennis Templeton (Case Western Reserve University, Cleveland, OH). XL1-Blue supercompetent cells (Stratagene) were transformed with 20 ng of plasmid DNA following the manufacturer's protocol, and large scale plasmid DNA was prepared from overnight cultures (250 ml) using Qiagen Endofree Midi Prep Kit. For transfection, 20 g of DNA and 2 g of pCI neo (Promega) was prepared in 150 mM NaCl to a final volume of 50 l and then added to 3 ϫ 10 6 hMSCs in 200 l of Opti-MEM (Life Technologies, Inc.). The cells were electroporated in 0.4-cm gap size cuvettes at 970 microfarads, 200 V with the time constant between 40 and 42 using a Gene Pulser II (Bio-Rad). The cells were then poured into culture dishes containing prewarmed Dulbecco's modified Eagle's medium ϩ 10% fetal bovine serum. The medium was changed after 24 h to remove the dead cells and debris, and the cells were allowed to grow for 48 h. The cells were then treated with OS, and alkaline phosphatase activity and calcium deposition were measured as described above. The expression of epitope-tagged MEK-1 protein was confirmed by Western blotting using anti EE antibody (Berkeley Antibody Co., Berkeley, CA).
Determination of Triglyceride in Differentiated Adipocytes-The accumulation of intracellular triglyceride droplets was visualized by staining with Oil Red O as described previously (37). A fluorescent quantitative method for determining Nile Red-stained fat droplets was employed to determine the extent of adipogenesis (3). Briefly, cells growing in 6-well plates were fixed for 30 min in 10% neutral buffered formalin at room temperature. The cells were washed twice with PBS, and background fluorescence was measured in the presence of 1.0 ml PBS/well with a Molecular Devices fMAX Fluorescent Microplate Reader using 355/460 and 485/538 nm filter sets. Cells were then incubated for 15 min with 0.2% saponin, 8 g/ml DAPI (4Ј,6-diamidino-2-phenylindole), and 1 g/ml Nile Red in PBS at room temperature. Cells were washed three times with PBS, and fluorescence was measured as above, and background values were subtracted. Cell numbers were normalized by the use of DAPI staining.

Activation of ERK2, JNK, and p38 by OS Treatment-As
previously demonstrated (4), hMSCs differentiate into osteoblasts in response to OS over 14 -16 days. Therefore, we examined ERK activation over the entire period of 3 weeks during treatment with OS. ERK activity in OS-treated cultures was determined by immunoblot analysis using phospho-specific MAP kinase antibody and by an in-gel kinase assay. As shown in Fig. 1A, OS did not have an effect on ERK activation up to day 5 of treatment. However, a robust and sustained activation of ERK was observed from day 7 to day 11, and the vast increase in phosphorylation was on ERK2. The maximal activation was observed at day 7, and ERK activity declined to basal levels after day 11. We also tested ERK activation by an in-gel kinase assay using MBP as a substrate (Fig. 1, C and D). The immunoblot and in-gel kinase assays gave consistent results. A densitometric analysis of the autoradiogram from ingel kinase assays revealed a 6 -7-fold increase in kinase activity at day 7 in OS-treated cells as compared with non-treated controls. This increase declined to a 3-fold increase at day 11 and subsequently returned to basal level by day 13.
Next, we measured JNK activity by an in-gel kinase assay using a GST-c-Jun (amino acids 1-79) fusion protein as a substrate. JNK2 binds c-Jun and catalyzes the phosphorylation of serine 63 and serine 73 (38). The continuous presence of OS led to a dramatic increase in JNK activity starting at day 13, peaking at day 15, and tapering off after day 17 (Fig. 2B). A 5-fold increase in JNK2 activation over non-treated controls was observed at day 15 of OS treatment as assessed by radioactivity incorporated into the JNK band after in-gel kinase assay (Fig. 2C).
The third member of the MAP kinase family, p38, appears to have a distinct function in cells due to its substrate specificity that differs from those of ERK and JNK. Therefore, we tested the activation of p38 in OS-treated hMSCs. The p38 activation was measured by an immunoblotting procedure using a phospho-specific p38 antibody. Prolonged treatment with OS stimulates the p38 activation and essentially paralleled ERK activation, starting at day 9 and declining to basal levels after day 13 ( Fig. 2A). Since p38 is known to be activated by environmental stresses such as hyperosmolarity (39), it is important to note that the time window when p38 is activated (9 -13 days) is associated with Gla protein and extracellular matrix synthesis, increased alkaline phosphatase activity, and initiation of mineral deposition (40) which may collectively cause an increase in intra-or extracellular osmolarity.
Inhibition of the ERK Signaling Pathway Blocks Osteogenic Differentiation-To ask whether ERK activation is necessary for osteogenic differentiation, PD98059, a selective inhibitor of MEK1, was used to prevent the phosphorylation and activation of the ERKs (41). To quantify osteogenic differentiation, alkaline phosphatase (AP) activity was measured at day 7 of OStreated hMSCs by both histochemical and biochemical methods. Continuous incubation of hMSCs with OS for 7 days resulted in an 8-fold increase in AP activity over the nontreated control (3.15 nmol of p-nitrophenyl phosphate/10 6 control versus 25 nmol of p-nitrophenyl phosphate/10 6 OS-treated cells, p Ͻ 0.01) (Fig. 4A). Treatment of hMSCs with OS-containing PD98059 produced a concentration-dependent inhibition of AP activity as determined by both histochemical and biochemical methods (Fig. 3A and 4A). A 50 M dose of PD98059 inhibited AP activity by 83%, whereas 10 and 25 M

FIG. 1. Osteogenic medium induces activation of ERKs in hM-
SCs. Cells were treated (ϩ) or not treated (Ϫ) with osteogenic differentiation medium (OS), and lysates were prepared at the indicated times. Lysates were subjected to immunoblot analysis using phosphospecific (A) and non-activated MAP kinase (B) antibody. An aliquot of each lysate was subjected to kinase detection assay within a polyacrylamide gel containing MBP as a substrate. Phosphorylation was assessed by incorporation of [␥-32 P]ATP into the ERK proteins, followed by SDS-PAGE and autoradiography. The autoradiogram-revealed activation of ERK between 7 and 11 days was shown (C). Quantification of ERK phosphorylation in non-treated (squares) and treated (circles) lysates as determined by radioactivity incorporated in bands cut out of dried gels is shown (D) .   FIG. 2. Activation of p38 and JNK in response to osteogenic differentiation medium. Cultures of hMSCs were incubated in OS, and cell lysates were prepared at the indicated times. Lysates were subjected to immunoblot analysis using an anti-phospho-specific p38 antibody (A) and to an in-gel kinase assay within a polyacrylamide gel cross-linked with GST-c-Jun fusion protein as substrate (B). Phosphorylation was assessed by incorporation of [␥-32 P]ATP in the GST-c-Jun fusion protein followed by SDS-PAGE and autoradiography. C, quantification of GST-Jun phosphorylation in lysates of non-treated (squares) and OS-treated hMSCs (circles) was determined by measuring radioactivity incorporated into proteins cut out of dried gels.
PD98059 produced 71 and 62% inhibition, respectively. To confirm that inhibition of AP activity by PD98059 was due to abolition of MEK1 activity, we measured the activation of ERKs in parallel experiments by Western blotting using an anti-phospho-MAP kinase antibody. The dose-dependent inhibition of AP by PD98059 paralleled the decrease in ERK2 tyrosine phosphorylation (Fig. 3B).
We also tested the effect of PD98059 on the ability of hMSCs to mineralize the extracellular matrix that they produce when cultured in the presence of OS. hMSCs grown in OS deposited a calcium-rich mineralized matrix by day 15 as measured by a sensitive colorimetric calcium assay (Fig. 4B). Importantly, the mineralization pattern as judged by the von Kossa staining was distributed throughout the culture rather than localized to a few discrete foci (data not shown). However, when added with OS, PD98059 significantly inhibited calcium deposition in a dose-dependent manner at all concentrations tested. Cells grown in control or OS medium containing 50 M PD98059 failed to deposit any detectable calcium throughout the culture period. These results demonstrate that OS-induced APase activity and mineral deposition, the hallmarks of bone formation, are associated directly with ERK2 stimulation and, therefore, ERK2 plays an important role in osteogenic differentiation of hMSCs. We attempted to test inhibitors of p38 and JNK for effects on the osteogenic differentiation of the hMSCs. We were unable to acquire an inhibitor of JNK for testing. The inhibitor of p38, compound SB203580, did not inhibit AP activity significantly nor did it inhibit calcium accumulation (data not shown), and therefore its role in osteogenic differentiation is unclear. However, elevated expression of activity for this kinase is likely associated with other cellular processes, and JNK activity occurred at a much later time than ERK2 activation.
Attenuation of ERK-stimulated Osteogenesis Promotes the Adipogenic Differentiation of hMSCs-A striking observation of the present study was that large numbers of adipocytes appeared in cultures that were treated with the MEK1 inhibitor PD98059 together with osteogenic differentiation medium. As early as 6 -7 days after the onset of treatment, adipocytes began to appear morphologically, and their numbers increased with time. At day 15, approximately 15-20% of cells appeared to be adipocytes as determined by Oil Red O staining (Fig. 5D). This differentiation appears to be dose-dependent, as fewer adipocytes were seen in cultures treated with 10 and 25 M PD98059 as compared with 50 M (Fig. 6A). Non-treated control cultures and cultures that were treated with PD98059 alone or with OS alone did not show any adipocyte formation. To rule out the possibility that dexamethasone alone induced adipogenic differentiation in hMSCs, cells were treated with 10 Ϫ5 , 10 Ϫ6 , and 10 Ϫ7 M dexamethasone. Over a 16-day culture period, no adipocyte formation was observed by phase contrast microscopy or after Oil Red O staining (data not shown). These results confirm that hMSCs are not predestined to become adipocytes but rather are truly multipotential cells that can differentiate into several cell lineages depending upon the stimuli. We further quantified the number of adipocytes by staining the cultures with the fluorescent lipophilic dye Nile Red. As is evident from Fig. 6B, appearance of adipocytes in hMSC cultures treated with PD98059 in the presence of OS is directly proportional to the dose of PD98059. PD98059 alone did not result in any adipogenic differentiation of hMSCs (Fig. 6A).
To characterize further the molecular response, we tested the induction of adipocyte-specific genes such as PPAR␥2 (42), lipoprotein lipase (43), and fatty acid binding protein, aP 2 (44), at the expense of osteogenic specific genes such as osteopontin (45) in hMSCs cultured in OS containing PD98059. These adipogenic and osteogenic markers are undetectable or expressed at very low levels in undifferentiated hMSCs. The expression of osteogenic markers is induced to high levels in OS-treated hMSCs (Fig. 7). However, interruption of the osteogenic program by treatment with PD98059 caused these cells to enter the adipocytic lineage with high expression of PPAR␥2, aP 2 , and lipoprotein lipase genes that are involved in lipid metabolism and trigger terminal differentiation of preadipocytes into adipocytes (46). This expression was directly correlated with the disappearance of bone marker gene osteopontin. The transcription factor C/EBP␣ is also induced during adipogenic differentiation of hMSCs (data not shown). It is unlikely that there are endogenous PPAR ligands in the culture system that cause the adipogenic differentiation because control samples do not show the presence of adipocytes even after extended culture periods, and addition of 15-deoxy-⌬ 12,14 -prostaglandin J 2 to hMSCs will result in adipocytes in the culture. 2 Similarly, high density lipoproteins from human plasma in serum-free medium were not effective in inducing adipogenic differentiation (data not shown).
Expression of Dominant Negative MEK-1 Inhibits Osteogenic Differentiation-To confirm the role of ERKs in osteogenic differentiation, we transfected hMSCs with MEK-2A, MEK-2E, as well as with wild type MEK plasmids. The expression of various glutamic acid-tagged MEK constructs in hMSCs was determined by immunoprecipitation and Western blotting using EE-specific antibody (62). As shown in Fig. 8A, various EEtagged MEK mutants are expressed in hMSCs and were detectable by Western blot for at least 15 days in these transiently transfected cells. The mutation of both serine 218 and 222 to alanine resulted in a dominant negative isoform of MEK that is incapable of activating its downstream target, ERK (62). As is evident from Fig. 8B, expression of MEK-2A in hMSCs significantly (p Ͻ 0.05) inhibits the OS-induced alkaline phosphatase activity. A 41% reduction in alkaline phosphatase ac-tivity was observed at day 7 of OS treatment. Similarly, a more significant reduction in mineralization was observed in these cells as reflected by a 56% decrease in calcium deposition as compared with cells transfected with wild type MEK (MEK-WT). Mutation of serine 218 and 222 with negatively charged glutamic acid has been shown to cause a constitutively active form of MEK (62). We therefore transfected hMSCs with a plasmid bearing this constitutively active form, MEK-2E, to see if it would potentiate the effects of OS on osteogenesis. In hMSCs transfected with MEK-2E, OS treatment indeed resulted in increased alkaline phosphatase activity and calcium deposition (increases of 31 and 17% respectively) compared with control plasmid-transfected cells (Fig. 8, B and C). DISCUSSION Although activation of the MAP kinase signaling cascade has been associated with many different cellular signaling events, there is little previous evidence implicating this pathway in differentiation of hMSCs into osteoblasts. In the present study, we investigated the differentiation of multipotent mesenchymal stem cells into osteoblasts through regulation of MAP 2 M. F. Pittenger and S. C. Beck, unpublished observations. kinases. We found that differentiation in osteoinductive culture medium strongly activated ERK2, JNK2, and p38 in a time-dependent manner. Although ERK1 and ERK2 were both expressed at similar protein levels, ERK2 was strongly activated by phosphorylation during differentiation. The ERK pathway regulates two mutually antagonistic processes, cellular proliferation and cellular differentiation. Stimulation of MAP kinase by fluoride has been shown to lead to the potentiation of bone cell proliferation (63). There is considerable evidence that differentiation of cells requires sustained activation of ERK, whereas transient activation of ERK leads to proliferation (15). The division of hMSCs does not appear to be stimulated by exposure to dexamethasone alone. In fact, the addition of OS to human marrow stromal cells has been shown to inhibit proliferation while promoting osteogenic differentiation (7). This observation, combined with other evidence of differentiation occurring as early as 2 days after OS treatment (47), implies that the requirement of sustained ERK2 activation fosters continued lineage progression and furthers osteogenic differentiation. Furthermore, sustained activation of ERK2 between days 7 and 10 is correlated with up-regulation of osteopontin expression, matrix deposition, and initiation of mineralization (48). Based on the requirement for sustained ERK2 activation, we suggest the possibility that the ERK signaling pathway is involved in the establishment of an autocrine signaling loop, contributing to the accumulation of a factor responsible for hMSC differentiation. Further efforts to characterize the precise cell cycle dynamics in relation to addition of OS, the nature of the factors involved, and acquisition of the osteoblastic phenotype will help to clarify the role of MAP kinases in this osteogenesis system.
The requirement for the MAP kinase pathway for osteogenic differentiation of hMSCs was tested by the use of a specific inhibitor of MEK1, PD98059, and the introduction of a dominant negative MEK1 isoform into hMSCs. The PD98059 compound selectively blocks the activity of MEK1 and the subsequent activation of ERKs in a variety of intact cells (33,41). Additionally, the use of a dominant negative MEK1 expression plasmid, which avoids any drug-related pleiotropic effects, confirmed a specific role for the MEK1 signaling pathway. Our results demonstrated that inhibition of ERK activation paralleled the reduction in alkaline phosphatase activity and calcium deposition, suggesting a regulatory role of ERK in osteogenic differentiation of hMSCs and osteogenesis. Moreover, the introduction of a constitutively active form of MEK1 into the hMSCs resulted in increased AP activity and calcium deposition (Fig. 8, B and C). In addition to ERK2, OS-induced differentiation also stimulated JNK, and p38 as well, albeit at later time points. Although the ERK, JNK, and p38 MAP kinases are structurally related and have similar kinase cascade patterns, they are activated by different extracellular stimuli (49) and have distinct substrate specificities (21,51). These differences could explain the temporal relationship of kinase activation seen in the results, which reveal that these kinases are activated at different times and likely regulate distinct biological functions required in osteogenesis, such as commitment, differentiation, and maturation. Growth factor-stimulated activation of the ERK signaling pathway is required for proliferation and differentiation, whereas stimulation of p38 and JNK signaling pathways can contribute to other cellular processes, including cell death (52). Different types of stress such as physical strain (stretch) and hyperosmolarity generated by progressive accumulation and calcification of soft extracellular matrix around the differentiated osteoblasts and during nodule formation could lead to stimulation of p38 and JNK. Physiological calcification of soft extracellular matrix occurs in bone as it is converted into a rigid material capable of supporting mechanical force (53). It has been demonstrated that apoptosis occurs in heavily mineralized cultures in and around mature nodule areas during osteogenesis (40).
The osteogenic and adipogenic differentiation of bone marrow-derived stromal cells, and cells isolated from human trabecular bone have been characterized extensively; however, the regulatory mechanism(s) responsible remain unidentified. The lineage potentiality of these cells depends upon factors as diverse as 1, 25-dihydroxyvitamin D 3 (54) and BMPs (55), as well as cell passage number (59). Based upon our finding of the inverse regulation of osteogenesis and adipogenesis by MAP kinase in hMSCs, we suggest a role for ERK as a regulatory switch for these differentiation pathways.
Many growth factors and cytokines that activate MAP kinases also inhibit adipose differentiation in culture or in vivo (56,57). At the transcriptional level, two families of factors control adipogenesis as follows: PPARs and C/EBPs, members of the nuclear hormone receptor family (46). Among the three PPAR isoforms, ␣, ␦, and ␥, PPAR␥ is responsible for the initiation and maintenance of the adipocyte phenotype in vivo (58). The specific ligands that bind to PPAR␥, including antidiabetic thiazolidinediones, and prostaglandin J2, induce a strong adipogenic response in hMSCs. 2 PPAR␥ possesses a MAP kinase phosphorylation consensus sequence at the N terminus which is phosphorylated when growth factors known to activate MAP kinases interact with surface receptors (19,20,59). Phosphorylation of PPAR␥ decreases its transcriptional activities and thus contributes to the anti-adipogenic effects of these agents. Most of these studies were performed using preadipocytic cell lines and demonstrated that PPAR␥ acts as a master conductor of the entire adipogenic program. However, the potent activators of PPARs such as thiazolidinedione or 2-bromopalmitate that induce a potent adipogenic differentiation in preadipocytic cells could not commit multipotential embryonic stem cells into the adipogenic lineage (60). These observations suggest that PPAR␥ does not play its critical role alone in the initial steps of adipose cell commitment and development from a multipotent stem cell. There are likely other factors that play an important role in the early stages of initiating adipogenic programming. To decipher the molecular mechanism will require further work, preferably in a serumfree system.
To our knowledge, this is the first report that the switch from osteogenesis to adipogenesis is regulated by ERK in hMSCs. However, the relevant downstream targets of ERK still remain unknown at present. Whereas activation of ERK correlates with proliferation and initial differentiation, the activation of p38 and JNK likely plays a role in later cellular differentiation or, in the end stage, apoptosis. The presence of multiple MAP kinase signal transduction pathways and the extent to which each is activated during differentiation may reflect the integration of multiple signals and the balance among ERK, p38, and JNK MAP kinases.
Human bone marrow contains a variety of cell types including hMSCs, osteoblasts, osteoclasts, blood progenitor cells, stromal cells, and adipocytes. Under normal conditions there is a balance between osteoblasts and adipocytes in bone marrow. However, this balance is lost in diseases such as osteoporosis and osteopenia where a decrease in bone volume is accompanied by an increase in adipose tissue. One mechanism for the decrease could be imbalance in differentiation of hMSCs into osteoblastic and adipocytic lineages at the expense of the osteoblasts. The precise molecular mechanism for this imbalance is presently unknown, and the interrelationship between adipocytes and bone formation is an area of active research. The present studies suggest a vital role of ERK1/2 in balancing the differentiation of osteoblastic and adipogenic lineages from a common progenitor (hMSCs) and are directly relevant to the understanding of the mechanism of osteoporosis.