Oxysterols Regulate Differentiation of Mesenchymal Stem Cells:Pro-Bone and Anti-Fat

Hoa Ton Kha,1 Benjamin Basseri,1 Daniel Shouhed,1 Jennifer Richardson,1 Sotirios Tetradis,2 Theodore J Hahn,1,3

and Farhad Parhami1

ABSTRACT: Pluripotent mesenchymal stem cells can undergo lineage-specific differentiation in adult organ-isms. However, understanding of the factors and mechanisms that drive this differentiation is limited. Weshow the novel ability of specific oxysterols to regulate lineage-specific differentiation of mesenchymal stemcells into osteogenic cells while inhibiting their adipogenic differentiation. Such effects may have importantimplications for intervention with osteoporosis.

Introduction: Oxysterols are products of cholesterol oxidation and are formed in vivo by a variety of cells includingosteoblasts. Novel pro-osteogenic and anti-adipogenic effects of specific oxysterols on pluripotent mesenchymal cellsare demonstrated in this report. Aging and osteoporosis are associated with a decrease in the number and activity ofosteoblastic cells and a parallel increase in the number of adipocytic cells.Materials and Methods: The M2–10B4 pluripotent marrow stromal cell line, as well as several other mesenchymalcell lines and primary marrow stromal cells, was used to assess the effects of oxysterols. All results were analyzedfor statistical significance using ANOVA.Results and Conclusion: Pro-osteogenic and anti-adipogenic effects of specific oxysterols were assessed by theincrease in early and late markers of osteogenic differentiation, including alkaline phosphatase activity, osteocalcinmRNA expression and mineralization, and the decrease in markers of adipogenic differentiation including lipoproteinlipase and adipocyte P2 mRNA expression and adipocyte formation. Complete osteogenic differentiation of M2 cellsinto cells expressing early and late markers of differentiation was achieved only when using combinations of specificoxysterols, whereas inhibition of adipogenesis could be achieved with individual oxysterols. Oxysterol effects werein part mediated by extracellular signal-regulated kinase and enzymes in the arachidonic acid metabolic pathway, i.e.,cyclo-oxygenase and phospholipase A2. Furthermore, we show that these specific oxysterols act in synergy with bonemorphogenetic protein 2 in inducing osteogenic differentiation. These findings suggest that oxysterols may play animportant role in the differentiation of mesenchymal stem cells and may have significant, previously unrecognized,importance in stem cell biology and potential therapeutic interventions.J Bone Miner Res 2004;19:830–840. Published online on January 12, 2004; doi: 10.1359/JBMR.040115

Key words: oxysterols, stem cells, osteogenesis, adipogenesis, bone morphogenetic protein

INTRODUCTION

MESENCHYMAL TISSUE REGENERATION during adult life ishighly dependent on a population of pluripotent stem

cells commonly known as mesenchymal stem cells or marrowstromal cells (MSCs).(1,2) These cells are present in a variety oftissues and are prevalent in bone marrow stroma.(3,4) They canbe readily isolated and expanded in culture, and on treatmentwith a variety of agonists, they can differentiate into lineage-specific cells including osteoblasts, chondrocytes, myocytes,adipocytes, and fibroblasts.(1–4) MSCs are thought to be anexcellent potential tool for interventions in many diseases thatresult from reduced or impaired functioning of these cells.(5,6)

Such interventions may include regeneration of defective ex-tracellular matrix by systemic infusion of MSCs in individualswith generalized defects such as in osteogenesis imper-fecta,(7,8) repair of damaged tissues such as bone and cartilageby localized injection into the affected areas,(9,10) and thetherapeutic delivery of specific genes to diseased tissues.(11)

All this, however, requires an understanding of the factors andmolecular mechanisms involved in determining lineage-specific differentiation of MSCs and the identification of newstrategies for driving lineage-specific differentiation in vitroand in vivo.

One situation in which MSCs might be used in humandisease is in age-related and postmenopausal osteoporosis,where the decreased number and osteogenic activity ofosteoprogenitor MSCs, in part, leads to decreased boneThe authors have no conflict of interest.

1Department of Medicine, David Geffen School of Medicine at UCLA, Los Angeles, California, USA; 2University of California at LosAngeles Dental School, Los Angeles, California, USA; 3West Los Angeles VA Medical Center, Los Angeles, California, USA.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 19, Number 5, 2004Published online on January 12, 2004; doi: 10.1359/JBMR.040115© 2004 American Society for Bone and Mineral Research

830

formation.(6,12–14) A reduction in bone formation by osteo-blasts in the face of increased bone resorption by osteoclastsin these disorders results in decreased bone mass, increasedsusceptibility to fractures, and impaired fracture healing.(14)

Therefore, the systemic and/or local application of osteo-progenitor MSCs, or factors that enhance their osteogenicdifferentiation, could be of great potential benefit in boost-ing bone forming capacity in osteoporosis. Future improvedtreatment of osteoporosis will likely require the use of boneanabolic agents that can enhance the osteogenic differenti-ation and bone-forming capacity of osteoblastic precursorcells, and hence increase bone mass and reduce fracturerisk.(15–17) Several growth factors and hormones have beentested in this regard, including bone morphogenetic protein(BMP), growth hormone, and parathyroid hormone(PTH).(15,16) BMPs play critical roles in the differentiationof MSCs into osteoblasts both in vitro and in vivo.(18,19)

BMP2 is the most potent known inducer of bone formationin vivo, and it enhances the differentiation of osteoprogeni-tor and non-osteoprogenitor precursors into cells with os-teoblastic phenotype.(20) The use of BMPs in fracture heal-ing is currently hampered by the large concentrations of therecombinant protein necessary to induce adequate boneformation.(20) Currently, only PTH, which on intermittentinjection increases bone mass and reduces bone fractureincidence, has been approved by the Food and Drug Ad-ministration (FDA). However, because of its potential ad-verse side effects and cost, it is currently only used forseverely osteoporotic patients. Hence identification of novelfactors and strategies for enhancing bone formation is cur-rently an area of intense investigation.

Because MSCs are the common progenitors for bothosteoblasts and adipocytes,(1–4) reductions in the number ofosteoblastic cells in aging and osteoporosis have been at-tributed, in part, to an increased differentiation of thesecommon progenitor cells into adipocytes rather than osteo-blasts.(14) It has been observed that the numbers of adipo-cytes in the bone marrow increases in parallel with a de-crease in the number of osteoblasts in a variety of types ofosteoporosis.(21) Moreover, the volume of adipose tissue inbone increases with age in normal subjects and is substan-tially elevated in age-related osteoporosis,(22) with the num-ber of adipocytes adjacent to bone trabeculae increasing inparallel to the degree of trabecular bone loss.(23) Based onthis and similar observations, it has been suggested thatbone loss in age-related osteoporosis is at least in partcaused by a shift in MSC differentiation from the osteoblas-tic to the adipocytic pathway.(14,21) If this is true, it ispossible that intervening in this apparent shift in lineage-specific differentiation of MSCs could increase the numberof cells capable of undergoing differentiation into function-ing osteoblasts.(21)

Oxysterols form a large family of oxygenated derivativesof cholesterol that are present in the circulation and inhuman and animal tissues.(24–26) These compounds may beformed either by auto-oxidation, as a secondary byproductof lipid peroxidation, or by the action of specific mono-oxygenases, most of which are members of the cytochromeP450 family of enzymes.(27) In addition, oxysterols may bederived from dietary intake.(28) Oxysterols have potent ef-

fects in physiologic and pathological processes includingcholesterol metabolism, inflammation, apoptosis, steroidproduction, and atherosclerosis.(24–26,29) Moreover, becauseof the abundance of cholesterol in living organisms, thepro-oxidant nature of our cellular metabolic processes, andthe multitude of enzymatic and non-enzymatic pathways foroxysterol production, we speculate that oxysterols may playadditional, as yet unidentified, roles in biological systems.Recently, several reports have noted the possible role ofoxysterols in cellular differentiation.(30–32) For example, theoxysterols 22(R)- and 25-hydroxycholesterol induce the dif-ferentiation of human keratinocytes in vitro,(30,31) whereasmonocyte differentiation is induced by the oxysterol7-ketocholesterol.(32)

We previously reported that products of the cholesterolbiosynthetic pathway are important for the proper osteo-genic differentiation and activity of MSCs.(33) Because ox-ysterols may be a derivative of the endogenous cellularcholesterol biosynthetic pathway,(25–27) we hypothesizedthat the oxysterols generated by osteoprogenitor cells, aswell as those derived from exogenous sources, may beinvolved in their osteogenic differentiation. In this report,we present the first line of evidence for osteogenic activityof specific oxysterols and their ability to inhibit adipogenicdifferentiation. In addition, we show the synergistic inter-action of these oxysterols with BMP2.

MATERIALS AND METHODS

Materials

Oxysterols, �-glycerophosphate (�GP), silver nitrate, andOil red O were obtained from Sigma (St Louis, MO, USA);RPMI 1640, �-MEM, and DMEM were from Irvine Scien-tific (Santa Ana, CA, USA); and FBS was from Hyclone(Logan, UT, USA). PD98059 was purchased from BI-OMOL Research Labs (Plymouth Meeting, PA, USA); TO-901317, SC-560, NS-398, Ibuprofen, and Flurbiprofen werefrom Cayman Chemical (Ann Arbor, MI, USA); ACA andAACOCF3 were from Calbiochem (La Jolla, CA, USA);and recombinant human BMP2 was from R&D Systems(Minneapolis, MN, USA). Antibodies to phosphorylatedand native extracellular signal-regulated kinases (ERKs)were obtained from New England Biolabs (Beverly, MA,USA) and troglitazone was from Sankyo (Tokyo, Japan).

Cells

The M2–10B4 mouse marrow stromal cell line obtainedfrom American Type Culture Collection (Rockville, MD,USA) was derived from bone marrow stromal cells of a(C57BL/6J � C3H/HeJ) F1 mouse and supports human andmurine myelopoiesis in long-term cultures (as per ATCC).These cells were cultured in RPMI 1640 containing 10%heat-inactivated FBS and supplemented with 1 mM sodiumpyruvate, 100 U/ml penicillin, and 100 U/ml streptomycin(all from Irvine Scientific). The MC3T3-E1 mouse preos-teoblastic cell line was purchased from ATCC and culturedin �-MEM containing 10% heat-inactivated FBS and sup-plements as described above. C3H-10T1/2 mouse pluripo-tent embryonic fibroblast cells were kindly provided by DrKristina Bostrom (UCLA) and were cultured in DMEM

831OXYSTEROLS INDUCE OSTEOGENIC DIFFERENTIATION

containing 10% heat-inactivated FBS and supplements asdescribed above. Primary mouse marrow stromal cells wereisolated from male 4- to 6-month-old C57BL/6J mice andcultured and propagated as previously reported.(34)

Alkaline phosphatase activity assay

Colorimetric alkaline phosphatase (ALP) activity assayon whole cell extracts was performed as previously de-scribed.(34)

von Kossa and Oil red O staining

Matrix mineralization in cell monolayers was detected bysilver nitrate staining as previously described.(34) Oil red Ostaining for detection of adipocytes was performed as pre-viously described.(34)

45Ca incorporation assay

Matrix mineralization in cell monolayers was quantifiedusing the 45Ca incorporation assay as previously de-scribed.(35)

Western blot analysis

After treatments, cells were lysed in lysis buffer, proteinconcentrations were determined using the Bio-Rad proteinassay (Hercules, CA, USA), and SDS-PAGE was performedas previously described.(34) Probing for native and phos-phorylated ERKs was performed as previously reported.(34)

RNA isolation and Northern blot analysis

After treatment of cells under appropriate experimentalconditions, total RNA was isolated using the RNA isolationkit from Stratagene (La Jolla, CA, USA). Total RNA (10mg) was run on a 1% agarose/formaldehyde gel and trans-ferred to Duralon-UV membranes (Strategene) and cross-linked with ultraviolet light. The membranes were hybrid-ized overnight at 60°C with a 32P-labeled mouse osteocalcincDNA probe,(36) mouse lipoprotein lipase (LPL), mouseadipocyte protein 2 (aP2) PCR-generated probes, and hu-man 28S and 18S rRNA probes obtained from GenekaBiotechnology (Montreal, Quebec, Canada) and MaximBiotech (San Francisco, CA, USA), respectively. The PCRproducts were generated using primer sets synthesized byInvitrogen (Carlsbad, CA, USA) with the following speci-fications: mouse aP2 gene (accession no. M13261); sense(75–95) 5�-CCAGGGAGAACCAAAGTTGA-3�, antisense(362–383) 5�-CAGCACTCACCCACTTCTTTC-3�, gener-ating a PCR product of 309 bp. Mouse LPL (accession no.XM_134193); sense (1038–1058) 5�-GAATGAAGAAAA-CCCCAGCA-3�, antisense (1816–1836) 5�-TGGGCCA-TTAGATTCCTCAC-3�, generating a PCR product of 799bp. The PCR products were gel-purified and sequenced bythe UCLA sequencing core, showing the highest similarityto their respective GenBank entries. After hybridization, theblots were washed twice at room temperature with 2� SSCand 0.1%SDS and twice at 60°C with 0.5� SSC and 0.1%SDS and exposed to X-ray film. The extent of gene induc-tion was determined by densitometry.

BMP2 ELISA

Conditioned medium from M2 cells cultured in 24-wellplates was prepared after treatment with oxysterols or con-trol buffer for 24 or 48 h, followed by rinsing the cells toremove the treatment media and replacing it with 0.5 mlRPMI containing 0.1% bovine serum albumin (BSA)/well.After 24 h, the conditioned medium was collected, centri-fuged to remove debris, and used in the BMP2 ELISAimmunoassay according to the manufacturer’s instructions(R&D Systems).

Statistical analyses

Computer-assisted statistical analyses were performed us-ing the StatView 4.5 program. All p values were calculatedusing ANOVA and Fisher’s projected least significant dif-ference (PLSD) significance test. A value of p � 0.05 wasconsidered significant.

RESULTS

Novel osteogenic activities of oxysterols

To begin testing our hypothesis, we examined the effectsof oxysterols on indices of osteoblastic differentiation in invitro cultures of MSCs. In cultures of MSCs in vitro, stim-ulation of ALP activity, osteocalcin gene expression, andmineralization of cell colonies are indices of increaseddifferentiation into osteoblast phenotype.(37,38) We foundthat specific oxysterols, namely 22(R)-hydroxycholesterol(22R), 20(S)-hydroxycholesterol (20S), and 22(S)-hydroxycholesterol (22S), induced ALP activity, an earlymarker of osteogenic differentiation, in pluripotent M2–10B4 murine MSCs (M2; Fig. 1A). This effect was specific,because other oxysterols, including 7-ketocholesterol (7K),did not induce ALP activity in these cells (Fig. 1A). Theinduction of ALP activity was both dose- and time-dependent at concentrations between 0.5 and 10 �M, andshowed a relative potency of 20S � 22S � 22R. A 4-hexposure to these oxysterols followed by replacement withosteogenic medium without oxysterols was sufficient toinduce ALP activity in M2 cells, measured after 4 days inculture.

Although induction of ALP activity is an early markerof osteogenic differentiation and plays an important rolein both the differentiation and eventual mineralization pro-cesses,(39) this response was not sufficient for the inductionof mineralization in M2 cells by individual oxysterols. In-dividual oxysterols (22R, 20S, and 22S) at concentrationsbetween 0.5 and 10 �M were unable to induce mineraliza-tion after as many as 14 days of exposure, despite theirability to cause large increases in ALP activity measured 4days after treatment (data not shown). The individual oxys-terols also had minimal to no effect on osteocalcin geneexpression after as many as 14 days of treatment (data notshown). However, ALP activity (Fig. 1B), robust mineral-ization (Figs. 1C and 1D), and osteocalcin gene expression(Figs. 1E and 1F) were all induced in M2 cultures by acombination of the 22R � 20S or 22S � 20S oxysterols.Other combinations of oxysterols including 22R � 22S, orcombinations of 22R or 22S with 7K, did not induce min-

832 KHA ET AL.

eralization in M2 cell cultures (data not shown). The com-bination of 20S with either 22R or 22S also producedosteogenic effects in the mouse pluripotent embryonic fi-broblast C3H10T1/2 cells (Fig. 1G), in murine calvarial

pre-osteoblastic MC3T3-E1 cells, and in primary mouseMSCs (Fig. 1H), as assessed by stimulation of ALP activityand mineralization. Thus, we concluded that the combina-tion of 20S with either S or R stereoisomers of 22-

FIG. 1. Osteogenic effects of oxysterols inMSCs. (A) M2 cells at confluence were treatedwith control vehicle (C) or 10 �M individualoxysterols as indicated, in an osteogenic mediumconsisting of RPMI 1640, to which 10% FBS, 50�g/ml ascorbate, and 3 mM �GP were added.After 3 days of incubation, ALP activity wasdetermined in cell homogenates by a colorimet-ric assay as previously described. Results from arepresentative of five experiments are shown,reported as the mean � SD of quadruplicatedeterminations, normalized to protein concentra-tion (*p � 0.01 for C vs. oxysterol-treated cells).(B) M2 cells at confluence were treated in os-teogenic medium with control vehicle (C) or acombination of 22R and 20S oxysterols, at theindicated concentrations. ALP activity was mea-sured after 3 days as described above. Resultsfrom a representative of four experiments areshown, reported as the mean � SD of quadru-plicate determinations, normalized to proteinconcentration (*p � 0.01 for C vs. oxysterols).(C) M2 cells at confluence were treated in os-teogenic medium with control vehicle or 5 �Moxysterols, alone or in combination as indicated.After 14 days, mineralization was identified byvon Kossa staining, which appears black, as pre-viously described. (D) M2 cells were treatedwith control vehicle (C) or a combination of 22Rand 20S oxysterols at increasing concentrationsas indicated. After 14 days, matrix mineraliza-tion in cultures was quantified using a 45Ca in-corporation assay as previously described. Re-sults from a representative of four experimentsare shown, reported as the mean � SD of qua-druplicate determinations, normalized to proteinconcentration (*p � 0.01 for C vs. oxysterol-treated cultures). (E) M2 cells at confluence weretreated with control vehicle (C) or a combinationof 22R and 20S oxysterols (Ox; 5 �M each) inosteogenic medium. After 4 and 8 days, totalRNA from duplicate samples was isolated andanalyzed for osteocalcin (Osc) and 28S rRNAexpression by Northern blotting as described.Data from densitometric analysis of the Northernblot is shown in F as the average of duplicatesamples, normalized to 28S rRNA. (G)C3H10T1/2 cells at confluence were treated inosteogenic medium with control vehicle (C) or acombination of 22R and 20S oxysterols, at theindicated concentrations. ALP activity was mea-sured after 3 days as described above. Resultsfrom a representative of three experiments areshown, reported as the mean � SD of quadru-plicate determinations, normalized to proteinconcentration (*p � 0.01 for C vs. oxysterols).(H) Primary MSCs at confluence were treated inosteogenic medium with control vehicle (C) or acombination of 22R and 20S oxysterols, at theindicated concentrations. After 10 days, matrixmineralization in cultures was quantified using a45Ca incorporation assay as described earlier.Results from a representative of three experi-ments are reported as the mean � SD of qua-druplicate determinations, normalized to proteinconcentration (*p � 0.05 for C vs. oxysterols).

833OXYSTEROLS INDUCE OSTEOGENIC DIFFERENTIATION

hydroxycholesterol has osteogenic effects on osteoblast pre-cursor cells. Although stimulation of MSCs by BMP2 canenhance their osteogenic differentiation,(37) the osteogeniceffects of the oxysterols were not caused by the induction ofBMP2 expression in M2 cells. RT-PCR analysis of BMP2mRNA expression in M2 cells treated for 4 or 8 days with22R � 20S oxysterols (5 �M) showed no induction byoxysterol treatment (data not shown). In addition, ELISAassay using conditioned media from M2 cells treated with22R � 20S (5 �M) for 24 and 48 h also did not show anyinduction of BMP2 protein expression (data not shown).Interestingly, the BMP inhibitor noggin (Ng) at 200 ng/mlcaused a 40% inhibition in 22R � 20S (RS, 2.5 �M)-induced ALP activity and a 90% inhibition of that inducedby rhBMP2 (100 ng/ml; Control � 4 � 2; RS � 75 � 7; RS� Ng � 42 � 5; BMP2 � 120 � 11; BMP2 � Ng � 10 �2 activity units/mg protein; p � 0.05 for Control versus RSand BMP2 and for RS and BMP2 in the presence andabsence of Ng). We speculate that in light of our RT-PCRand ELISA data that did not show any induction of BMP2by oxysterols, the inhibition of response to oxysterols bynoggin might be caused by inhibition of synergism betweenoxysterols and BMP2 present in FBS. Alternatively, oxys-terols may induce the expression of other members of theBMP family that can interact with noggin, such as BMP4and BMP7, the expression of which would not be detectedby our BMP2-specific ELISA assay.

Synergistic osteogenic effects of oxysterols with BMP2

As with other osteoprogenitor cells, BMP2 is able toinduce osteoblastic differentiation of M2 cells in vitro.(40)

Interestingly, we found that osteogenic combination of22R � 20S oxysterols acted synergistically with BMP2in inducing ALP activity (Fig. 2A), osteocalcin mRNAexpression (Figs. 2C and 2D), and mineralization by M2cells (Fig. 2B). Although synergism in stimulating ALPactivity was found when individual oxysterols 20S, 22S,and 22R were added with BMP2, synergy in induction ofmineralization was only produced when oxysterols wereadded in combinations of 22R � 20S or 22S � 20Soxysterols with BMP2.

Novel anti-adipogenic activities of oxysterols

Adipogenesis of adipocyte progenitors including MSC isregulated by the transcription factor peroxisome proliferatoractivated receptor � (PPAR�), which on activation byligand-binding, regulates transcription of adipocyte-specificgenes.(41) We previously reported that, as expected withMSCs, M2 cells have the ability to undergo adipogenicdifferentiation in response to the PPAR� activator, Trogli-tazone (Tro).(34) In M2 cells treated with Tro to induceadipogenesis, 20S, 22S, and 22R, alone or in combination,inhibited adipogenesis (Figs. 3A and 3B). The relative anti-adipogenic potency of these oxysterols was similar to theirrelative potency in stimulating ALP activity in M2 cells,with 20S � 22S � 22R. Similar to its lack of osteogeniceffect, 7K was also unable to inhibit adipogenesis in M2cells (data not shown). Inhibition of adipogenesis was alsoassessed by an inhibition of the expression of the adipogenicgenes LPL and aP2 by 20S (Figs. 3C and 3D). However,

addition of the oxysterols to already formed adipocytes inM2 cell cultures did not reduce the number of adipocytesafter 8 days of treatment (data not shown), suggesting thatthe oxysterols were active only at the early stages of adi-pogenesis. Inhibitory effects of these three oxysterols onadipogenesis were also demonstrated using C3H10T1/2 andprimary mouse MSC, in which adipogenesis was inducedeither by Tro or a standard adipogenic cocktail consisting ofdexamethasone and isobutylmethylxanthine (data notshown).

Mechanism of osteogenic activity of oxysterols

Mesenchymal cell differentiation into osteoblasts is reg-ulated by cyclo-oxygenase (COX) activity.(42–44) We exam-ined the possible role of COX in mediating the osteogeniceffects of oxysterols on M2 cells. In the presence of FBS,which corresponds to our experimental conditions, M2 cellsin culture express both COX-1 and COX-2 mRNA at allstages of osteogenic differentiation (data not shown). Con-sistent with the role of COX in osteogenesis, our studiesshowed that the COX-1 selective inhibitor SC-560, at 1–20

FIG. 2. Synergistic osteogenic effects of oxysterols and BMP2 inMSCs. (A) M2 cells at confluence were treated with control vehicle(C), 50 ng/ml recombinant human BMP2, or a combination of 22R and20S oxysterols (RS, 2.5 �M each), alone or in combination in osteo-genic medium as described in Fig. 1. ALP activity was measured after2 days as previously described. Results from a representative of fourexperiments are shown, reported as the mean � SD of quadruplicatedeterminations, normalized to protein concentration (*p � 0.001 forBMP � RS vs. BMP and RS alone). (B) M2 cells were treated asdescribed in A. After 10 days, matrix mineralization in cultures wasquantified using a 45Ca incorporation assay as previously described.Results from a representative of four experiments are shown, reportedas the mean � SD of quadruplicate determinations, normalized toprotein concentration (p � 0.01 for BMP � RS vs. BMP and RSalone). (C) M2 cells were treated under similar conditions as thosedescribed above. After 8 days, total RNA was isolated and analyzed forosteocalcin (Osc) and 18S rRNA expression by Northern blotting aspreviously described. Data from densitometric analysis of the Northernblot is shown in D as the average of duplicate samples, normalized to18S rRNA.

834 KHA ET AL.

�M, significantly inhibited the osteogenic effects of the 22R� 20S and 22S � 20S oxysterol combinations. SC-560inhibited oxysterol-induced ALP activity (Fig. 4A), miner-alization (Fig. 4B), and osteocalcin gene expression (Figs.4C and 4D). Although less effective than SC-560, the non-selective COX inhibitors, ibuprofen and flurbiprofen, atnontoxic doses of 1–10 �M, also significantly inhibited theosteogenic effects of 22R � 20S oxysterol combination by25–30%. In contrast, the selective COX-2 inhibitor, NS-398, at the highest nontoxic dose of 20 �M, had onlynegligible inhibitory effects. SC-560 (10 �M) also inhibitedthe synergistic induction of ALP activity by oxysterols andBMP2 (Fig. 4G). Furthermore, the osteogenic effects of theoxysterol combination on ALP activity (Fig. 4E) and min-eralization (Fig. 4F) were also inhibited by the generalphospholipase A2 (PLA2) inhibitor ACA and by the selec-tive cytosolic PLA2 inhibitor, AACOCF3 (AAC). More-over, rescue experiments showed that the effects of theCOX-1 and PLA2 inhibitors on oxysterol-induced ALP ac-tivity were reversed by the addition of 1 �M PGE2 (Fig. 4H)and 25 �M arachidonic acid (Fig. 4I), respectively.

The ERK pathway is another major signal transductionpathway previously associated with osteoblastic differenti-ation of osteoprogenitor cells.(45,46) Interestingly, the 20Soxysterol used alone or in combination with 22R oxysterolcaused a sustained activation of ERK1 and ERK2 in M2cells (Figs. 5A and 5B). Inhibition of the ERK pathway bythe inhibitor PD98059 inhibited oxysterol-induced mineral-ization (Fig. 5C) but not ALP activity or osteocalcin mRNA

expression in M2 cell cultures (data not shown). Theseresults suggest that sustained activation of ERK is importantin regulating certain specific, but not all, osteogenic effectsof oxysterols.

Liver X receptors (LXR) are nuclear hormone recep-tors that, in part, mediate certain cellular responses tooxysterols, including 22R and 20S, but not 22S.(47,48)

LXR� is expressed in a tissue-specific manner, whereasLXR� is ubiquitously expressed.(47,48) By Northern blotanalysis, we demonstrated the expression of LXR�, butnot LXR�, in confluent cultures of M2 cells (data notshown). To assess the possible role of LXR in mediatingthe effects of osteogenic oxysterols, we examinedwhether activation of LXR� by the pharmacologic LXRligand TO-901317 (TO) had effects similar to those ex-erted by 22R and 20S in M2 cells. Interestingly, incontrast to 22R and 20S, TO at 1–10 �M caused adose-dependent inhibition of ALP activity in M2 cells(control [C]: 18 � 2; ligands used at 10 �M: 22R � 45 �5; 20S � 140 � 12; and TO � 3 � 0.5 activity units/mgprotein; p � 0.01 for C versus all ligands). Furthermore,TO treatment did not induce osteocalcin gene expressionor mineralization after 10 days (data not shown). There-fore, the osteogenic effects of the oxysterols on M2 cellsseem to be independent of the LXR receptor, as sug-gested by the potent osteogenic activity of the non-LXRoxysterol ligand 22S and the lack of osteogenic effects inresponse to the LXR ligand TO.

FIG. 3. Inhibition of adipogenesis in MSCs byoxysterols. (A and B) M2 cells at confluencewere treated in RPMI containing 10% FBS withcontrol vehicle or 10 �M Tro in the absence orpresence of 10 �M 20S or 22S oxysterols. After10 days, adipocytes were visualized by Oil red Ostaining and quantified by light microscopy,shown in B, and as previously described. Datafrom a representative of four experiments areshown, reported as the mean � SD of quadru-plicate determinations (*p � 0.001 for Tro vs.Tro � 20S and Tro � 22S). (C and D) M2 cellswere treated at confluence with 10 �M Tro aloneor in combination with 10 �M 20S oxysterol.After 10 days, total RNA was isolated and ana-lyzed for LPL or aP2 gene expression or 18SrRNA expression by Northern blotting as de-scribed. Data from densitometric analysis of theNorthern blot is shown in D as the average ofduplicate samples, normalized to 18S rRNA. (E)M2 cells were treated with control vehicle (C) orosteogenic medium containing ascorbate and�-glycerophosphate (Ost) alone or in combina-tion with troglitazone (Tro, 20 �M) and 20Soxysterol (5 �M). ALP activity was measuredafter 4 days. Results from a representative ofthree experiments are reported as the mean �SD, normalized to protein concentration (*p �0.01 for C vs. Ost, Ost vs. Ost � Tro, and Ost �Tro vs. Ost � Tro � 20S).

835OXYSTEROLS INDUCE OSTEOGENIC DIFFERENTIATION

FIG. 4. Mechanism of osteogenic activity of oxysterols in MSC. (A) M2 cells at confluence were pretreated for 4 h with control vehicle (C)or 10 �M COX-1 inhibitor SC-560 (SC) in osteogenic medium as described earlier. Next, a combination of 22R and 20S oxysterols (RS, 2.5 �Meach) was added in the presence or absence of SC as indicated. After 3 days, ALP activity was measured as described earlier. Data from arepresentative of three experiments are shown, reported as the mean � SD of quadruplicate determinations, normalized to protein concentration(*p � 0.001 for RS vs. RS � SC). (B) M2 cells were treated as described in A, and after 10 days, matrix mineralization in cultures was quantifiedby a 45Ca incorporation assay as described earlier. Results from a representative of three experiments are shown, reported as the mean � SD ofquadruplicate determinations, normalized to protein concentration (*p � 0.01 for RS vs. RS � SC). (C) M2 cells were pretreated with 20 �MSC for 4 h, followed by the addition of RS in the presence or absence of SC as described above. After 8 days, total RNA was isolated and analyzedfor osteocalcin (Osc) and 18S rRNA expression by Northern blotting as previously described. Data from densitometric analysis of the Northernblot is shown in D as the average of duplicate samples, normalized to 18S rRNA. (E) M2 cells at confluence were pretreated for 2 h with controlvehicle (C) or PLA2 inhibitors ACA (25 �M) and AACOCF3 (AAC, 20 �M) in osteogenic medium. Next, a combination of 22R and 20Soxysterols (RS, 2.5 �M) was added in the presence or absence of the inhibitors as indicated. After 3 days, ALP activity was measured as previouslydescribed. Data from a representative of three experiments are shown, reported as the mean � SD of quadruplicate determinations, normalizedto protein concentration (*p � 0.01 for RS vs. RS � ACA and RS � AAC). (F) M2 cells were treated as described in E. After 10 days, matrixmineralization in cultures was quantified using a 45Ca incorporation assay as previously described. Results from a representative of threeexperiments are shown, reported as the mean � SD of quadruplicate determinations, normalized to protein concentration (*p � 0.01 for RS vs.RS � ACA and RS � AAC). (G) M2 cells at confluence were treated in osteogenic medium with control vehicle (C), rhBMP2 (BMP, 50 ng/ml),and oxysterol combination 22R � 20S (RS, 2.5 �M) alone or in combination. Pretreatment of some cells with SC-560 (SC, 10 �M) was donefor 2 h before the addition of BMP and RS. ALP activity was measured after 4 days. Results from a representative of three experiments are reportedas mean � SD of quadruplicate determinations, normalized to protein concentrations (*p � 0.005 for BMP � RS vs. BMP � RS � SC). (H)M2 cells were treated in osteogenic medium with RS oxysterol combination (2.5 �M) alone or in combination with COX-1 inhibitor SC-560 (SC,10 �M) and PGE2 (1 �M). ALP activity was measured after 4 days and reported as described above. Results from a representative of threeexperiments are reported as the mean � SD of quadruplicate determinations, normalized to protein concentrations (*p � 0.01 for RS vs. RS �SC and RS � SC and RS � SC � PGE2). (I) M2 cells were treated in osteogenic medium with RS oxysterol combination (2.5 �M) alone or incombination with PLA2 inhibitor ACA (25 �M) and arachidonic acid (AA, 25 �M). ALP activity was measured as described above. Results froma representative of three separate experiments are reported as the mean � SD of quadruplicate determinations, normalized to protein concentration(*p � 0.01 for RS vs. RS � ACA and for RS � ACA and RS � ACA � AA).

836 KHA ET AL.

Mechanism of anti-adipogenic activity of oxysterols

As noted above, ERKs are important in the proliferationand osteogenic differentiation of MSCs. In addition to pos-itively regulating the osteogenic activity of osteoblast pro-genitor cells, ERK activation also negatively regulatesthe adipogenic differentiation of adipocyte progenitor cells,and inhibition of ERK enhances adipogenic different-iation.(49–51) Consistent with this effect of ERK, inhibitionof oxysterol-induced ERK activation by ERK pathway in-hibitor PD98059 completely abolished the anti-adipogeniceffects of 20S and 22S on Tro-induced adipogenesis in M2cells (Fig. 5D). Interestingly, PD98059 potentiated the adi-pogenic effects of Tro (Fig. 5D), suggesting that, similar tothe situation in pre-adipocytes,(51) spontaneous activity ofERK is inhibitory to the adipogenic differentiation of MSC.In contrast, the selective inhibitors for COX-1 and COX-2,SC-560, and NS-398, respectively, were not able to abolishthe anti-adipogenic effects of the oxysterols (data notshown). These results suggested that the anti-adipogenicactivity of oxysterols is mediated through activation of theERK pathway.

DISCUSSION

This is the first demonstration of the ability of oxysterolsto regulate lineage-specific differentiation of MSCs in favorof osteoblastic and against adipogenic differentiation. This

effect of the oxysterols is in part mediated through COX/PLA2- and ERK-dependent mechanisms (Fig. 6). COX-1and COX-2 are both present in osteoblastic cells and seemto be primarily involved in bone homeostasis and repair,respectively.(42,52) Metabolism of arachidonic acid intoprostaglandins, including prostaglandin E2 (PGE2), by theCOXs mediates the osteogenic effects of these enzymes.(53)

COX products and BMP2 have complementary and additiveosteogenic effects.(44) Activation of PLA2 releases arachi-donic acid from cellular phospholipids and makes it avail-able for further metabolism by COX enzymes into prosta-glandins.(54,55) Consistent with previous reports ofoxysterol-stimulated metabolism of arachidonic acid,(56,57)

the present results suggest that the osteogenic activity of theoxysterols in MSC are in part mediated by the activation ofPLA2-induced arachidonic acid release and its metabolisminto osteogenic prostanoids by the COX pathway. Further-more, COX activity also seems to be important in regulatingthe synergism between the osteogenic oxysterols andBMP2, but not in the anti-adipogenic effects of the oxys-terols, suggesting the specific role of COX/PLA2 pathway inmediating the oxysterol-induced lineage specific differenti-ation of M2 cells into osteogenic cells.

These results also show another signaling pathway medi-ating M2 responses to oxysterols is the ERK pathway.Sustained activation of ERKs mediates the osteogenic dif-ferentiation of human MSCs,(45) and activation of ERKs in

FIG. 5. Role of ERK in mediating the responses of MSCs to oxysterols. (A) M2 cells at confluence were pretreated for 4 h with RPMI containing1% FBS, followed by treatment with control vehicle or 5 �M 20S oxysterol for 1, 4, or 8 h. Next, total cell extracts were prepared and analyzedfor levels of native or phosphorylated ERK (pERK) using specific antibodies as previously described. Data from a representative of fourexperiments are shown; each treatment is shown in duplicate samples. (B) M2 cells were treated as described in A, with 5 �M 20S and 22R aloneor in combination. After 4 h, cell extracts were prepared and analyzed for levels of native and pERK. Data from a representative of twoexperiments are shown; each treatment is shown in duplicate samples. (C) M2 cells at confluence were pretreated for 2 h with control vehicle (C)or 20 �M PD98059 (PD) in osteogenic medium as previously described. Next, a combination of 22R and 20S oxysterols (RS, 5 �M each) wereadded to appropriate wells as indicated. After 10 days of incubation, matrix mineralization was quantified by the 45Ca incorporation assay aspreviously described. Data from a representative of three experiments are reported as the mean � SD of quadruplicate determinations, normalizedto protein concentration (*p � 0.01 for RS vs. RS � PD). (D) M2 cells at confluence were pretreated for 2 h with 20 �M PD98059 (PD) in RPMIcontaining 5% FBS. Next, the cells were treated with control vehicle (C), 10 �M Tro, or 10 �M of 20S or 22S oxysterols alone or in combinationas indicated. After 10 days, adipocytes were visualized by Oil red O staining and quantified by light microscopy as previously described. Datafrom a representative of three experiments are reported as the mean � SD of quadruplicate determinations.

837OXYSTEROLS INDUCE OSTEOGENIC DIFFERENTIATION

human osteoblastic cells results in upregulation of expres-sion and DNA binding activity of Cbfa1, the master regu-lator of osteogenic differentiation.(46) Furthermore, ERKactivation seems to be essential for growth, differentiation,and proper functioning of human osteoblastic cells.(58) Inaddition, the ERK pathway is a negative regulator of adi-pogenic differentiation of adipocyte progenitor cells.(49–51)

Consistent with the role of the ERK pathway in both osteo-genic and adipogenic differentiation, oxysterol-inducedmineralization and inhibition of adipogenic differentiationwere blocked by the inhibitor, PD98059, and the oxysterolsinduced a sustained activation of ERK. It must be notedthat, because both individual and combinations of oxys-terols activated ERK, whereas only the combinations ofoxysterols were able to induce full osteogenic differentia-tion of M2 cells, ERK activation seems to be an essentialbut not sufficient step in mediating the osteogenic effects ofthe oxysterols.

We also show for the first time that the osteogenic oxys-terols synergistically stimulate BMP2-induced osteogenicdifferentiation of MSCs. It is of interest that when BMPswere originally extracted from bone matrix, they were foundto be associated with lipids.(59) These crudely characterizedlipids potentiated the osteogenic effects of BMP, as evi-denced by a great reduction in the bone forming activity ofBMP on their removal.(59) This line of evidence and theprevious demonstration of the presence of lipids in healingfracture callus(60) suggest that lipids such as oxysterols mayplay an important role in regulating osteogenesis and theeffects of BMP on this process. These studies suggest thatspecific oxysterols may potentiate the osteogenic differen-tiation of osteoblast precursor cells and that this effect ofoxysterols is in part mediated through a potentiation of theosteogenic activity of BMPs. Furthermore, the synergisticeffects of osteogenic oxysterols with BMP2 may provide an

exciting new strategy for using BMP2 in local stimulationof fracture healing. However, because bone turnover isregulated by a combination of osteoblastic bone formationand osteoclastic bone resorption, it is important to note thatthe putative in vivo osteogenic capacity of oxysterol com-binations reported here will depend also on their effect onosteoclastic differentiation/activity and bone resorption.

The possibility of inducing lineage-specific differentia-tion of MSCs into designated cells and tissues has potentialimplications for the prevention and treatment of a number ofdisorders that result from age-related tissue deterioration orfrom genetic defects.(6) For example, as noted earlier, age-related osteoporosis seems to be, at least in part, caused bya decreased osteogenic differentiation of osteoprogenitorcells.(61–63) Accordingly, the focus of developing new ther-apeutic approaches that would positively impact osteoporo-sis has largely shifted from finding additional antiresorptionagents to finding anabolic agents that can enhance boneformation.(15,16) The possible use of lipid-based treatmentsinvolving oxysterols, rather than protein- or peptide-basedinterventions as currently used, introduces a whole newstrategy for creating therapeutics that would potentiallyintervene in osteoporosis. Further identification of thedownstream targets of these molecules in MSCs holds greatpotential for discovering new and previously unsuspectedways to regulate the lineage-specific differentiation of keyMSC populations. This could improve the potential for theuse of autologous MSCs in treatment of a variety of con-nective tissue diseases, as well as in tissue engineering.(6) Inaddition, fat cells increase significantly in mesenchymaltissues with age, in parallel with a decline in osteoblasts,muscle cells, and other key derivatives of MSCs.(64) Thebasis of this apparent shift in MSC differentiation is un-known, but the ability to reverse this process could signif-icantly affect the management of age-related disorders.

ACKNOWLEDGMENTS

This work was supported by National Institute on AgingPepper Center Grant IP60-AG10415 and National Institutesof Health Grant HL30568.

REFERENCES

1. Prockop DJ 1997 Marrow stromal cells as stem cells for nonhe-matopoietic tissues. Science 276:71–74.

2. Caplan AI 1994 The mesengenic process. Clin Plast Surg 21:429–435.

3. Chen JL, Hunt P, McElvain M, Black T, Kaufman S, Choi ES 1997Osteoblast precursor cells are found in CD34� cells from humanbone marrow. Stem Cells 15:368–377.

4. Majors AK, Boehm CA, Nitto H, Midura RJ, Muschler GF 1997Characterization of human bone marrow stromal cells with respectto osteoblastic differentiation. J Orthop Res 15:546–557.

5. Gerson SL 1999 Mesenchymal stem cells: No longer second classmarrow citizens. Nat Med 5:262–264.

6. Caplan AI, Bruder SP 2001 Mesenchymal stem cells: Buildingblocks for molecular medicine in the 21st century. Trends MolMed 7:259–264.

7. Horwitz EM, Prockop DJ, Fitzpatrick LA, Koo WW, Gordon PL,Neel M, Sussman M, Orchard P, Marx JC, Pyeritz RE, BrennerMK 1999 Transplantability and therapeutic effects of bonemarrow-derived mesenchymal cells in children with osteogenesisimperfecta. Nat Med 5:309–313.

8. Pereira R, O’Hara MD, Laptev AV, Halford KW, Pollard MD,Class R, Simon D, Livezey K, Prockop, DJ 1998 Marrow stromal

FIG. 6. Pathways regulating the effects of oxysterols in MSCs.Osteogenic oxysterols act on MSC by activating at least two signalingpathways: (1) the ERK pathway and (2) the COX/PLA2 pathway.Activation of both pathways seem to be involved in the osteogeniceffects of the oxysterols in MSCs and are inhibited by the inhibitor ofERK pathway PD98059, the COX-1 inhibitor SC-560, and the PLA2

inhibitors ACA and AACOCF3. The anti-adipogenic effects of theoxysterols seem to be mainly mediated by the activation of the ERKpathway and are inhibited by PD98059.

838 KHA ET AL.

cells as a source of progenitor cells for nonhematopoietic tissues intransgenic mice with a phenotype of osteogenesis imperfecta. ProcNatl Acad Sci USA 95:1142–1147.

9. Mendes SC, Tibbe JM, Veenhof M, Bakker K, Both S, PlantenburgPP, Oner FC, De Bruijn JD, Van Blintterswijk CA 2002 Bonetissue-engineered implants using human bone marrow stromalcells: Effect of culture conditions and donor age. Tissue Eng8:911–920.

10. Murphy M 2001 Injected mesenchymal stem cells stimulate me-niscal repair and protection of articular cartilage. Trans Orthop ResSoc 26:193.

11. Lieberman JR, Le LQ, Wu L, Finerman GA, Berk A, Witte ON,Stevenson S 1998 Regional gene therapy with a BMP-2 producingmurine stromal cell line induced heterotopic and orthotopic boneformation in rodents. J Orthop Res 16:330–339.

12. Quarto R, Thomas D, Liang CT 1995 Bone progenitor cell deficitsand the age-associated decline in bone repair capacity. CalcifTissue Int 56:123–129.

13. Mullender MG, van der Meer DD, Huiskes R, Lips P 1996 Os-teocyte density changes in aging and osteoporosis. Bone 18:109–113.

14. Chan GK, Duque G 2002 Age-related bone loss: Old bone, newfacts. Gerontology 48:62–71.

15. Mundy GR 2002 Directions of drug discovery in osteoporosis.Annu Rev Med 53:337–354.

16. Rodan GA, Martin TJ 2002 Therapeutic approaches to bone dis-eases. Science 289:1508–1514.

17. Goltzman D 2002 Discoveries, drugs and skeletal disorders. NatRev Drug Discov 1:784–796.

18. Reddi AH 1995 Bone morphogenetic proteins, bone marrow stro-mal cells, and mesenchymal stem cells. Clin Orthop Relat Res313:115–119.

19. Katagiri T, Takahashi N 2002 Regulatory mechanisms of osteo-blast and osteoclast differentiation. Oral Dis 8:147–159.

20. Lieberman JR, Daluiski A, Einhorn TA 2002 The role of growthfactors in the repair of bone. J Bone Joint Surg Am 84:1032–1044.

21. Nuttall ME, Gimble JM 2000 Is there a therapeutic opportunity toeither prevent or treat osteopenic disorders by inhibiting marrowadipogenesis? Bone 27:177–184.

22. Meunier P, Aaron J, Edouard C, Vignon G 1971 Osteoporosis andthe replacement of cell populations of the marrow by adiposetissue: A quantitative study of 84 iliac bone biopsies. Clin OrthopRelat Res 80:147–154.

23. Burkhardt R, Kettner G, Bohm W, Schmidmeier M, Schlag R,Frisch B, Mallmann B, Eisenmenger W, Gilg TH 1987 Changes intrabecular bone, hematopoiesis and bone marrow vessels in aplas-tic anemia, primary osteoporosis, and old age: A comparativehistomorphometric study. Bone 8:157–164.

24. Bjorkhem I, Diczfalusy U 2002 Oxysterols: Friends, foes, or justfellow passengers? Arterioscler Thromb Vasc Biol 22:734–742.

25. Edwards PA, Ericsson J 1999 Sterols and isoprenoids: Signalingmolecules derived from the cholesterol biosynthetic pathway.Annu Rev Biochem 68:157–185.

26. Schroepfer GJ 2000 Oxysterols: Modulators of cholesterol metab-olism and other processes. Physiol Rev 80:361–554.

27. Russell DW 2000 Oxysterol biosynthetic enzymes. Biochim Bio-phys Acta 1529:126–135.

28. Lyons MA, Samman S, Gatto L, Brown AJ 1999 Rapid hepaticmetabolism of 7-ketocholesterol in vivo: Implications for dietaryoxysterols. J Lipid Res 40:1846–1857.

29. Bjorkhem I 2002 Do oxysterols control cholesterol homeostasis?J Clin Invest 110:725–730.

30. Hanley K, Ng DC, He SS, Lau P, Min K, Elias PM, Bikle DD,Mangelsdorf DJ, Williams ML, Feingold KR 2000 Oxysterolsinduce differentiation in human keratinocytes and increase Ap-1dependent involucrin transcription. J Invest Dermatol 114:545–553.

31. Komuves GL, Schmuth M, Fowler AJ, Elias PM, Hanley K, ManM, Moser AH, Lobaccaro JA, Williams ML, Mangelsdorf DJ,Feingold KR 2002 Oxysterol stimulation of epidermal differenti-ation is mediated by liver X receptor-� in murine epidermis.J Invest Dermatol 118:25–34.

32. Hayden JM, Brachova L, Higgins K, Obermiller L, Sevanian A,Khandrika S, Reaven PD 2002 Induction of monocyte differenti-ation and foam cell formation in vitro by 7-ketocholesterol. J LipidRes 43:26–35.

33. Parhami F, Mody N, Gharavi N, Ballard AJ, Tintut Y, Demer LL2002 Role of the cholesterol biosynthetic pathway in osteoblasticdifferentiation of marrow stromal cells. J Bone Miner Res 17:1997–2003.

34. Parhami F, Jackson SM, Tintut Y, Le V, Balucan JP, Territo M,Demer LL 1999 Atherogenic diet and minimally oxidized lowdensity lipoprotein inhibit osteogenic and promote adipogenic dif-ferentiation of marrow stromal cells. J Bone Miner Res 14:2067–2078.

35. Parhami F, Morrow AD, Balucan J, Leitinger N, Watson AD,Tintut Y, Berliner JA, Demer LL 1997 Lipid oxidation productshave opposite effects on calcifying vascular cell and bone celldifferentiation. Arterioscler Thromb Vasc Biol 17:680–687.

36. Celeste AJ, Rosen V, Buecker JL, Kriz R, Wang EA, Wozney JM1986 Isolation of the human gene for bone gla protein utilizingmouse and rat cDNA clones. EMBO J 5:1885–1890.

37. Rickard DJ, Sullivan TA, Shenker BJ, LeBoy PS, Kazhdan I 1994Induction of rapid osteoblast differentiation in rat bone marrowstromal cell cultures by dexamethasone and BMP-2. Dev Biol161:218–228.

38. Hicok KC, Thomas T, Gori F, Rickard DJ, Spelsberg TC, RiggsBL 1998 Development and characterization of conditionally im-mortalized osteoblast precursor cell lines from human bone mar-row stroma. J Bone Miner Res 13:205–217.

39. Stein G, Lian J 1993 Molecular mechanisms mediatingproliferation/differentiation interrelationships during progressivedevelopment of the osteoblast phenotype. Endocr Rev 14:424–442.

40. Zebboudj AF, Imura M, Bostrom K 2002 Matrix GLA protein, aregulatory protein for bone morphogenetic protein-2. J Biol Chem277:4388–4394.

41. Tontonoz P, Hu E, Spiegelman BM 1994 Stimulation of adipo-genesis in fibroblasts by PPAR gamma 2, a lipid-activated tran-scription factor. Cell 79:1147–1156.

42. Zhang X, Schwartz EM, Young DA, Puzas JE, Rosier RN,O’Keefe RJ 2002 Cyclooxygenase-2 regulates mesenchymal celldifferentiation into the osteoblast lineage and is critically involvedin bone repair. J Clin Invest 109:1405–1415.

43. Chikazu D, Li X, Kawaguchi H, Sakuma Y, Voznesensky OS,Adams DJ, Xu M, Hoshi K, Katavic V, Herschman HR, Raisz LG,Pilbeam CC 2002 Bone morphogenetic protein 2 induces cyclo-oxygenase 2 in osteoblasts via a Cbfa1 binding site: Role in effectsof bone morphogenetic protein 2 in vitro and in vivo. J Bone MinerRes 17:1430–1440.

44. Simon AM, Manigrasso MB, O’Connor JP 2002 Cyclo-oxygenase2 function is essential for bone fracture healing. J Bone Miner Res17:963–976.

45. Jaiswal RK, Jaiswal N, Bruder SP, Mbalaviele G, Marshak DR,Pittenger MF 2000 Adult human mesenchymal cell differentiationto the osteogenic or adipogenic lineage is regulated by mitogen-activated protein kinase. J Biol Chem 275:9645–9652.

46. Pziros PG, Gil AR, Georgakopoulos T, Habeos I, Kletsas D,Basdra EK, Papavassiliou AG 2002 The bone-specific transcrip-tional regulator Cbfa1 is a target of mechanical signals in osteo-blastic cells. J Biol Chem 277:23934–23941.

47. Pete DJ, Janowski BA, Mangelsdorf DJ 1998 The LXRs: A newclass of oxysterol receptors. Curr Opin Genet Dev 8:571–575.

48. Edwards PA, Kast HR, Anisfeld AM 2002 BAREing it all: Theadoption of LXR and FXR and their roles in lipid metabolism. JLipid Res 43:2–12.

49. De Mora JF, Porras A, Ahn N, Santos E 1997 Mitogen-activatedprotein kinase activation is not necessary for, but antagonizes,3T3–L1 adiopocytic differentiation. Mol Cell Biol 17:6068–6075.

50. Kim S, Muise AM, Lyons PJ, Ro H 2001 Regulation of adipogen-esis by a transcriptional repressor that modulates MAPK activa-tion. J Biol Chem 276:10199–10206.

51. Hansen JB, Petersen RK, Jørgensen C, Kristiansen K 2002 Dereg-ulated MAPK activity prevents adipocyte differentiation of fibro-blasts lacking the retinoblastoma protein. J Biol Chem 277:26335–26339.

52. Raisz LG, Pilbeam CC, Fall PM 1993 Prostaglandins: Mechanismsof action and regulation of production in bone. Osteoporos Int3(Suppl 1):136–140.

53. Balsinde J, Winstead MV, Dennis EA 2002 Phospholipase A2

regulation of arachidonic acid metabolism. FEBS Lett 531:2–6.

839OXYSTEROLS INDUCE OSTEOGENIC DIFFERENTIATION

54. Capper EA, Marshall LA 2001 Mammalian phospholipases A2:Mediators of inflammation, proliferation and apoptosis. Prog LipidRes 40:167–197.

55. Lahoua Z, Astruc ME, Barjon JN, Michel F, Crastes de Paulet A1989 Mechanism of the activation of arachidonic acid release byoxysterols in NRK 49F cells: Role of calcium. Cell Signal 1:569–576.

56. Lahoua Z, Vial H, Michel F, Crastes de Paulet A, Astruc ME 1991Oxysterol activation of arachidonic acid release and prostaglandinE2 biosynthesis in NRK 49F cells is partially dependent on proteinkinase C activity. Cell Signal 3:559–567.

57. Wohlfeil ER, Campbell WB 1997 25-Hydroxycholesterol en-hances eicosanoid production in cultured bovine coronary arteryendothelial cells by increasing prostaglandin G/H synthase-2. Bio-chim Biophys Acta 1345:109–120.

58. Lai C, Chaudhary L, Fausto A, Halstead LR, Ory DS, Avioli LV,Cheng S 2001 Erk is essential for growth, differentiation, integrinexpression, and cell function in human osteoblastic cells. J BiolChem 276:14443–14450.

59. Urist MR, Behnam K, Kerendi F, Raskin K, Nuygen TD, ShamieAN, Malinin TI 1997 Lipids closely associated with bone morpho-genetic protein (BMP) and induced heterotopic bone formation.Connect Tissue Res 36:9–20.

60. Lane JM, Boskey AL, Li WKP, Eaton B, Posner AS 1979 Atemporal study of collagen, proteoglycan, lipid and mineral con-stituents in a model of endochondral osseous repair. Metab BoneDis Relat Res 1:319–324.

61. Bonyadi M, Waldman SD, Liu D, Aubin JE, Grynpas MD, Stan-ford WL 2003 Mesenchymal progenitor self-renewal deficiency

leads to age-dependent osteoporosis in Sca-1/Ly-6A null mice.Proc Nat Acad Sci USA 13:5840–5845.

62. Ichioka N, Inaba M, Kushida T, Esumi T, Takahara K, Inaba K,Ogawa R, Iida H, Ikehara S 2002 Prevention of senile osteoporosisin SAMP6 mice by intrabone marrow injection of allogeneic bonemarrow cells. Stem Cells 20:542–551.

63. Chen XD, Shi S, Xu T, Robey PG, Young MF 2002 Age-relatedosteoporosis in biglycan-deficient mice is related to defects in bonemarrow stromal cells. J Bone Miner Res 17:331–340.

64. Kirkland JL, Tchkonia T, Pirtskhalava T, Han J, Karagiannides I2002 Adipogenesis and aging: Does aging make fat go MAD? ExpGerontol 37:757–767.

Address reprint requests to:Farhad Parhami, PhD

University of California Los Angeles Division ofCardiology

Center for the Health SciencesRoom 47-123

10833 Le Conte AvenueLos Angeles, CA 90095, USA

E-mail: fparhami@mednet.ucla.edu

Received in original form September 15, 2003; in revised formNovember 26, 2003; accepted Janaury 9, 2004.

840 KHA ET AL.