Original Contribution

OXIDATIVE STRESS MODULATES OSTEOBLASTIC DIFFERENTIATION OFVASCULAR AND BONE CELLS

NILAM MODY,* FARHAD PARHAMI ,† THEODORE A. SARAFIAN,† and LINDA L. DEMER*†

*Department of Physiology and†Department of Medicine, UCLA School of Medicine, Los Angeles, CA, USA

(Received18 October2000;Accepted15 May 2001)

Abstract—Oxidative stress may regulate cellular function in multiple pathological conditions, including atherosclero-sis. One feature of the atherosclerotic plaque is calcium mineral deposition, which appears to result from thedifferentiation of vascular osteoblastic cells, calcifying vascular cells (CVC). To determine the role of oxidative stressin regulating the activity of CVC, we treated these cells with hydrogen peroxide (H2O2) or xanthine/xanthine oxidase(XXO) and assessed their effects on intracellular oxidative stress, differentiation, and mineralization. These agentsincreased intracellular oxidative stress as determined by 2,7 dichlorofluorescein fluorescence, and enhanced osteoblasticdifferentiation of vascular cells, based on alkaline phosphatase activity and mineralization. In contrast, H2O2 and XXOresulted in inhibition of differentiation markers in bone osteoblastic cells, MC3T3-E1, and marrow stromal cells,M2-10B4, while increasing oxidative stress. In addition, minimally oxidized low-density lipoprotein (MM-LDL),previously shown to enhance vascular cell and inhibit bone cell differentiation, also increased intracellular oxidativestress in the three cell types. These effects of XXO and MM-LDL were counteracted by the antioxidants Trolox andpyrrolidinedithiocarbamate. These results suggest that oxidative stress modulates differentiation of vascular and bonecells oppositely, which may explain the parallel buildup and loss of calcification, seen in vascular calcification andosteoporosis, respectively. © 2001 Elsevier Science Inc.

Keywords—Osteoblastic differentiation, Minimally oxidized low-density lipoprotein, Xanthine/xanthine oxidase, Hy-drogen peroxide, Reactive oxygen species, Antioxidants, Free radicals

INTRODUCTION

The development of atherosclerosis involves an increasein oxidative stress [1,2]. Macrophages within the athero-sclerotic lesion produce reactive oxygen species (ROS)such as hydrogen peroxide and superoxide anion [3].Endothelial cells also release ROS into the artery wall inresponse to various agents and conditions, such as bra-dykinin, hypoxia, and hypercholesterolemia [4–6]. Fur-thermore, they respond to oxidized LDL and leptin byincreasing intracellular ROS [7,8]. Similarly, smoothmuscle cells increase intracellular ROS generation,mostly by NADPH oxidase, in response to PDGF,TNF-a, and angiotensin II [9–11].

One feature of atherosclerotic lesions that has important

clinical consequences is vascular calcification. As an invitro model of vascular calcification, we have previouslyisolated and cloned a subpopulation of aortic medial smoothmuscle cells, which undergo osteoblastic differentiation[12]. These cells, called calcifying vascular cells (CVC),express several markers of osteoblastic differentiation, suchas collagen I, alkaline phosphatase, and osteocalcin, andproduce hydroxyapatite mineral [12]. Furthermore, severaloxidized lipids with atherogenic properties, including min-imally oxidized low-density lipoprotein (MM-LDL), oxi-dized palmitoyl arachidonoyl phosphatidylcholine, and iso-prostaglandin E2, were previously shown to inducedifferentiation of CVC [13].

Previous studies have shown that increased arterialcalcification correlates with an increase in osteoporosis[14,15]. In osteoporosis, bone loss involves both in-creased osteoclastic bone resorption and decreased os-teoblastic bone formation [16,17]. ROS enhance oste-oclastic activity [18,19], but the effect of ROS onosteoblastic function is unknown.

Address correspondence to: Linda L. Demer, M.D., Ph.D., Divisionof Cardiology, Department of Medicine, Center for the Health Sci-ences, Room 47-123, UCLA Medical Center, Los Angeles, CA 90095-1679, USA; Tel: (310) 206-2677; Fax: (310) 825-4963; E-Mail:ldemer@mednet.ucla.edu.

Free Radical Biology & Medicine, Vol. 31, No. 4, pp. 509–519, 2001Copyright © 2001 Elsevier Science Inc.Printed in the USA. All rights reserved

0891-5849/01/$–see front matter

PII S0891-5849(01)00610-4

509

In this report we tested the hypothesis that oxidativestress modulates osteoblastic differentiation of vascularand bone cells. To determine the effect of oxidativestress on CVC, we treated the cells with hydrogen per-oxide (H2O2) or xanthine/xanthine oxidase (XXO), areaction in which xanthine oxidase converts xanthine touric acid and generates superoxide anion. Both agentsinduced osteoblastic differentiation of CVC. We alsotreated MC3T3-E1, a preosteoblastic cell line, and M2-10B4, a marrow stromal cell line that undergoes osteo-blastic differentiation [20], with XXO and H2O2. Wefound that, in contrast to CVC, both inducers of oxida-tive stress inhibited differentiation of both osteoblasticcell lines. Previously, we found that atherogenic oxidizedlipids, such as MM-LDL, may contribute to vascularcalcification and osteoporotic bone loss by differentiallyregulating the osteoblastic differentiation of vascular andbone cells. In this report, we demonstrate that MM-LDLalso increased oxidative stress in all three cell types,while inducing a reciprocal response in vascular andbone cells. In addition, antioxidants inhibited the effectsof XXO and MM-LDL, which suggests that productionof ROS mediates the enhanced differentiation of vascularcells and inhibits the differentiation of osteoblastic cellsin response to these agents.

MATERIALS AND METHODS

Materials

[3H]Thymidine and 45CaCl2 were from AmershamCorp. (Piscataway, NJ, USA). Benzoic acid, DMSO, glu-cose oxidase, hydrogen peroxide (30% v/v), 3-[4,5-dimeth-ylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT,thiazoyl blue), xanthine, and xanthine oxidase, pyrro-lidinedithiocarbamate (PDTC), and 2,7 dichlorofluoresceindiacetate were from Sigma (St. Louis, MO, USA). Defer-oxamine and 6-hydroxy-2,5,7,8-tetramethylchroman-2-car-boxylic acid (Trolox) were from Calbiochem (San Diego,CA, USA).

Cell culture

CVC, the clonal subpopulation of primary bovineaortic smooth muscle cells, were harvested and cloned asdescribed previously [12]. CVC were grown in Dulbec-co’s modified Eagle’s medium (Irvine Scientific, SantaAna, CA, USA) containing 15% heat-inactivated fetalbovine serum (Hyclone Labs, Logan, UT, USA) andsupplemented with sodium pyruvate (1 mM), penicillin(100 units/ml), and streptomycin (100 units/ml), all fromIrvine Scientific. MC3T3-E1 mouse preosteoblast cellline (Riken Cell Bank, Tsukuba, Japan) was grown inaMEM containing 10% FBS, supplemented as indicated

for DMEM above. M2-10B4 mouse marrow stromal cellline (ATCC, Rockville, MD, USA) was grown in RPMIcontaining 10% FBS, supplemented as indicated forDMEM above. The medium was changed every 3–4 dwith agents, if applicable.

Timing of treatments

For H2O2 treatment, CVC and MC3T3-E1 weretreated daily for 4 d and then assayed for alkaline phos-phatase activity; M2-10B4 were treated every other dayfor 6 d and then assayed. For XXO and MM-LDL, allcell types were treated with agent or control buffer every3 d and then assayed for alkaline phosphatase activity or45Ca incorporation.

Alkaline phosphatase activity assay

CVC, MC3T3-E1, and M2-10B4, seeded in a 24 or 48well plates, were treated at 70–90% confluence withvehicle alone or agents in media (DMEM,aMEM, andRPMI for CVC, MC3T3-E1, and M2-10B4, respec-tively) containing 2.5% FBS as described above. H2O2

concentration was determined by induction of alkalinephosphatase activity in CVC and cytotoxicity as assessedby the MTT assay. XXO concentration was determinedby induction of alkaline phosphatase activity in CVC andcytotoxicity as assessed by the MTT assay. MC3T3-E1and M2-10B4 were treated with similar concentrations ofH2O2 and XXO, and cells remained adherent to cultureplates for the duration of the experiment. For antioxidanttreatments, cells were pretreated with the antioxidant inmedia containing 2.5% FBS for either 30 min or 2 h, asindicated in the text, at 37°C and then treated with eithervehicle, XXO, or MM-LDL, with antioxidant. Alkalinephosphatase activity was assayed as previously described[13]. Alkaline phosphatase activity was normalized tototal protein concentration as assessed by the Bradford(Bio-Rad, Hercules, CA, USA) assay. The data werefrom a representative of at least three experiments shownas the mean6 SD of quadruplicate wells.

MTT assay

The MTT assay was used to measure cell viability[21]. Cells were seeded in 24 well plates and treated inthe same manner as the alkaline phosphatase activityassay. At the end of the incubation period, MTT wasdiluted to a final concentration of 0.5 mg/ml in theculture media. Cells were incubated for 2 h, and themedia was carefully removed. The formazan crystalswere dissolved in acidified isopropanol (40 mM HCl inisopropanol) and absorbance was determined at 590 nm.

510 N. MODY et al.

The data were from a representative of at least threeexperiments shown as a percentage of control6 SD andwere measured in quadruplicate wells.

45Ca incorporation assay

45Ca incorporation to detect mineralization was per-formed as previously described [13]. CVC were plated in24 well plates and treated with 2.5% FBS, supplementedwith 4 mM CaCl2, 5 mM b-glycerophosphate, and agentsto be tested. MC3T3-E1 were treated with 2.5% FBS,supplemented with 2 mM CaCl2, 5 mM b-glycerophos-phate, and 25mg/ml ascorbic acid. M2-10B4 weretreated with 2.5% FBS, supplemented with 0.5 mMCaCl2, 3 mM b-glycerophosphate, and 50mg/ml ascor-bic acid. The data for45Ca incorporation were from arepresentative of three experiments shown as the mean6SD of five wells.

Lipoproteins

Human LDL was isolated by density-gradient centrif-ugation of serum and stored in phosphate-buffered 0.15mol/l NaCl containing 0.01% EDTA. MM-LDL wasprepared by iron oxidation of human LDL as previouslydescribed [22]. Minimal oxidation of LDL resulted in a2- to 3-fold increase in conjugated dienes and 2 to 3 nmolof thiobarbituric acid-reactive substances per milligramof cholesterol after dialysis. The concentrations of li-poprotein used are reported in micrograms of protein.

Oxidative stress measurements

2,7 Dichlorofluorescein diacetate (2,7-DCFH-DA) isa cell permeable dye, which becomes fluorescent uponreaction with hydroxyl radical, hydrogen peroxide, orperoxynitrite. Cells were plated on a 48 well plate, and,at 80–90% confluence, they were loaded with 20mg/mlof 2,7-DCFH-DA in modified Kreb’s Ringer buffer (125mM NaCl, 5 mM KCl, 1.2 mM KH2PO4, 25 mMHEPES, 6 mM Glucose, 1.2 mM MgSO4, 1.0 mM CaCl2,pH 7.4). The cells were assayed as described previously[23]. Cellular oxidative stress was measured fluorometri-cally by monitoring the oxidation of intracellular DCFHusing a Cytofluor 2300 plate reader (PerSeptive Biosys-tems, Framingham, MA, USA). For H2O2 and XXO,DCF fluorescence was measured over the course of 1 h.For MM-LDL, cells were treated with MM-LDL for 30min and DCF fluorescence was measured after the incu-bation buffer was removed. DCF fluorescence measure-ments were normalized to cell number as determined bypropidium iodide and digitonin, as previously described[23]. The data were from a representative of at least three

experiments shown as the mean6 SD of quadruplicatewells.

Statistical analysis

Computer-assisted statistical analyses were performedusing the Student’s two-tailedt-test. Ap value of, .05was considered to be statistically significant.

RESULTS

Reactive oxygen species production in response tooxidative stress in vascular cells and bone cells

Intracellular ROS production by 25mM xanthine and25 mU/ml xanthine oxidase (XXO) and by 1 mM hydro-gen peroxide (H2O2) was measured by DCF fluorescencein CVC and MC3T3-E1 over the course of 1 h. Inresponse to XXO, DCF fluorescence increased in bothcell types (2.26 0.4-fold in CVC; 2.66 0.4-fold inMC3T3-E1) compared to control at 1 h (Fig. 1A, C). Inresponse to H2O2, the increase was 5.16 1.0-fold and7.1 6 1.7-fold for CVC and MC3T3-E1 cells, respec-tively, compared to control at 1 h (Fig. 1B, D). We alsomeasured a 2-fold increase in DCF fluorescence in M2-10B4 cells in response to 0.25 mM H2O2 over the courseof 30 min (data not shown, two experiments performedin quadruplicate,p , .01 for each experiment). In CVC,the increases in control samples over the course of 1 hwere 1.5- to 2-fold, and, in MC3T3-E1 cells, the in-creases in control samples over the course of 1 h were notsignificant.

Osteoblastic differentiation of vascular cells and bonecells in response to oxidative stress

Because XXO and H2O2 increased oxidative stress inthese cell types, we used these agents to determine theeffect of oxidative stress on cellular differentiation. XXOinduced alkaline phosphatase activity in CVC (Fig. 2A).A similar effect was produced by 4 d daily treatment with1 mM H2O2 in 2.5% serum (Fig. 2B, 2.56 0.5-foldincrease). Although hydrogen peroxide caused a greaterincrease in DCF fluorescence than XXO, the increase inalkaline phosphatase activity in response to H2O2 ismuch smaller than the increase in alkaline phosphataseactivity for XXO. One possibility is that the DCF signalis falsely high because of the production of O2

•2, whichresults in higher H2O2 levels because of the dispropor-tionation of O2

•2 [24]. Furthermore, addition of 7.5mU/ml glucose oxidase, which, with glucose from themedia, releases H2O2, resulted in a 1.5-fold increase inalkaline phosphatase activity (data not shown, two ex-periments performed in quadruplicate,p , .005 for each

511ROS and osteoblast differentiation

experiment). To determine the effect of XXO on miner-alization of CVC, radiolabeled calcium incorporationwas assayed. Results showed a significant increase(2.3 6 0.2-fold) in 45Ca incorporation, indicating en-hanced mineralization (Fig. 2C) [25].

To determine whether these agents caused cytotoxicityto the CVC, viability was measured using the MTT (thia-zoyl blue) assay [21,26]. With either 1 mM H2O2 or XXO,there was no significant decrease in cell viability (86616% and 896 14%, respectively, data not shown,n $ 3).

In contrast, XXO and H2O2 inhibited alkaline phospha-tase activity of the osteoblastic cell line, MC3T3-E1 (Fig.3A, B), and bone marrow stromal cell line, M2-10B4 (Fig.4A, B). The decrease in alkaline phosphatase activity withXXO was 2.46 0.7-fold in MC3T3-E1 and 4.06 0.1-foldin M2-10B4 (Figs. 3A, 4A). With 1 mM H2O2 (the con-centration used with the CVC), a significant cytotoxic effectwas found by the MTT assay in MC3T3-E1 cells (3069%, data not shown,n 5 4, p , .01); therefore, wemeasured alkaline phosphatase activity at 0.5 mM H2O2

and found a significant decrease in alkaline phosphataseactivity (Fig. 3B, 2.26 0.7-fold). At this concentration ofH2O2, the cytotoxic effect was less than 15% (data not

shown,n 5 4). For these reasons, we chose to use XXO asthe agent for generating oxidant stress in subsequent exper-iments. Furthermore, XXO decreased45Ca incorporation inMC3T3-E1 cells (Fig. 3C) and M2-10B4 cells (Fig. 4C) by11-fold and 2-fold, respectively.

To exclude an effect of the different media used forthe different cell types, we tested whether the inductionof alkaline phosphatase activity by XXO in CVC wasaffected by the choice of media. The cells were culturedas usual and then treated in eithera-MEM or RPMI.There was a similar fold induction of alkaline phospha-tase activity ina-MEM as compared to DMEM (data notshown,n 5 3, p , .005 for each experiment). XXO-treatment of CVC in RPMI also resulted in a significantinduction of alkaline phosphatase activity, though not ashigh as with DMEM (data not shown,n 5 3, p , .001for each experiment).

Minimally oxidized low-density lipoprotein as anoxidative stress

Previously, we reported that minimally oxidized-LDL(MM-LDL) enhanced alkaline phosphatase activity and

Fig. 1. Increased accumulation of ROS in calcifying vascular cells (CVC) and osteoblastic cell line, MC3T3-E1, stimulated by XXOand H2O2. CVC (A–B) or MC3T3-E1 (C–D) were cultured in 48 well plates and were loaded with 20mg/ml DCHF-DA for 20 min.The cells were then treated with 25mM xanthine and 25 mU/ml xanthine oxidase (XXO) or with 1 mM hydrogen peroxide (H2O2).The figures are as follows: (A) CVC and XXO, (B) CVC and H2O2, (C) MC3T3-E1 and XXO, (D) MC3T3-E1 and H2O2. DCFfluorescence was monitored over the course of 1 h and normalized to cell number. Results from a representative of four experimentsare shown as the mean6 SD of four or five determinations (*p , .005, compared with controls at 1 h). The combined analysis of themultiple experiments for each cell type and agent also yielded significant results (p # .01).

512 N. MODY et al.

Fig. 2. Induction of osteoblastic differentiation and mineralization byoxidative stress in CVC. (A) Alkaline phosphatase activity in responseto XXO. CVC were incubated for 3 d with control buffer or with 25mM xanthine and 25 mU/ml xanthine oxidase. (B) Alkaline phospha-tase activity in response to H2O2. CVC were treated daily for 4 d withmedia or 1 mM H2O2. For (A) and (B), alkaline phosphatase activitywas measured from whole cell lysates. Results from a representative offour experiments are shown as the mean6 SD of quadruplicatedeterminations (*p , .001 for H2O2 and XXO, compared with con-trols). With XXO, each experiment yielded significant results (p ,.005). With H2O2, the combined analysis of multiple experiments wasstatistically significant (p , .01). (C) Ca45 incorporation in response toXXO. CVC were treated with control buffer or with 25mM xanthineand 25 mU/ml xanthine oxidase and were incubated for 7–9 d. Resultsfrom a representative of three experiments are shown as the mean6SD of quintuplicate determinations (*p , .001 for XXO compared withcontrol). Analysis of the fold change of the three experiments alsoyielded significant results (p , .001).

Fig. 3. Inhibition of osteoblastic differentiation by oxidative stress inMC3T3-E1. (A) Alkaline phosphatase activity in response to XXO.MC3T3-E1 cells were incubated with control buffer or with 25mMxanthine and 25 mU/ml xanthine oxidase for 3 d. (B) Alkaline phos-phatase activity in response to H2O2. MC3T3-E1 were treated daily for4 d with media or 0.5 mM H2O2. Alkaline phosphatase activity wasmeasured from whole cell lysates. Results presented as in Fig. 2 (*p ,.001 for H2O2 and XXO, compared with controls). The combinedanalysis of four experiments for each agent yielded significant results(p , .05) (C) Ca45 incorporation in response to XXO. MC3T3-E1 weretreated with control buffer or with 25mM xanthine and 25 mU/mlxanthine oxidase were incubated for 27–30 d (*p , .001 for XOcompared with control). The combined analysis of three experimentswas statistically significant (p , .01).

513ROS and osteoblast differentiation

mineralization in CVC and inhibited alkaline phospha-tase activity and mineralization in MC3T3-E1 and M2-10B4 cells [13,20]. To determine whether MM-LDLgenerates ROS, cells were treated for 30 min with 200mg/ml MM-LDL, native LDL, or control buffer. DCFfluorescence was measured after the buffer containingexcess DCF was replaced to reduce any potential directinteraction between DCF and the LDL particle. Theresults showed that MM-LDL increased ROS production2–4-fold in CVC, MC3T3-E1, and M2-10B4 (Fig. 5).

Antioxidants

To determine whether the MM-LDL effect is due tooxidative stress, we examined the effects of two antioxi-dants on osteoblastic differentiation. We used pyrro-lidinedithiocarbamate (PDTC), a thiol-containing antiox-idant [27,28], and Trolox, a hydrophilic vitamin Eanalogue [29,30]. CVC were pretreated at 70–80% con-fluence with antioxidant for 2 h, followed by treatmentwith 200 mg/ml MM-LDL for 4–5 d or XXO for 3 d inthe presence of these agents. PDTC inhibited XXO-induced alkaline phosphatase activity in a dose-depen-dent manner with maximal inhibition (636 7%) at 10mM PDTC, while 100mM Trolox inhibited activity byonly 316 5% (Fig. 6A, B). This inhibition by PDTC wasnot due to toxicity as assessed by the MTT assay (datanot shown,n 5 3). In contrast, Trolox inhibited MM-LDL-induced alkaline phosphatase activity in a dose-dependent manner with complete inhibition at 100mMTrolox (Fig. 6C,p , .005 for each experiment). PDTC(10 mM) failed to block alkaline phosphatase activityinduced by MM-LDL (Fig. 6D).

To determine the role of hydroxyl radicals and Fentonproducts, CVC were pretreated with DMSO, benzoicacid, or deferoxamine for 30 min and then treated withXXO for 3 d in the presence of these agents. Resultsshowed that 1% DMSO caused a partial inhibition (5469%) of XXO-induced alkaline phosphatase activity (datanot shown,n 5 3, p 5 .01); however, benzoic acid hadno effect on XXO-induced alkaline phosphatase activity(data not shown,n 5 3). Previously reported concentra-tions of deferoxamine (1 mM) were toxic to CVC overthe course of the 3 d treatment. Deferoxamine, up to 10mM, did not inhibit XXO-induced alkaline phosphataseactivity (data not shown,n 5 3).

Because PDTC inhibited XXO effects and Troloxinhibited MM-LDL effects in CVC, we determined theeffect of PDTC with XXO and Trolox with MM-LDL inMC3T3-E1. As expected, in MC3T3-E1 cells, Troloxblocked the inhibition of alkaline phosphatase activity byMM-LDL, and PDTC counteracted the XXO-inducedinhibition of alkaline phosphatase activity (Table 1).Similar experiments with PDTC and Trolox were at-

Fig. 4. Inhibition of osteoblastic differentiation by oxidative stress inthe bone marrow stromal cell line, M2-10B4. (A) Alkaline phosphataseactivity in response to XXO. M2-10B4 were incubated with controlbuffer or with 25mM xanthine (2/1) and the indicated concentrationsof xanthine oxidase for 6 d. (B) Alkaline phosphatase activity inresponse to H2O2. M2-10B4 were treated on alternate days for 4 d withthe indicated concentrations of H2O2. Alkaline phosphatase activitywas measured from whole cell lysates. Results presented as in Fig. 2(*p , .005 for 0.25 and 0.5 mM H2O2 and p , .001 for 5 and 25mU/ml of XO). The combined analysis of three experiments for eachagent yielded significant results (p , .005) (C) Ca45 incorporation inresponse to XXO. M2-10B4 were treated with control buffer or with 25mM xanthine and the indicated concentrations of xanthine oxidase andwere incubated for 18–20 d (*p , .001 for 1, 5, 25 mU/ml of XOcompared with control). The combined analysis of four experimentswas statistically significant (p , .05).

514 N. MODY et al.

tempted on M2-10B4 cells. TenmM PDTC was toxic toM2-10B4 cells, and 5mM PDTC did not affect XXO-induced inhibition of alkaline phosphatase activity.

Trolox (100mM) did not inhibit the effects of 200mg/mlMM-LDL. These results suggest that marrow stromalcells are especially sensitive to oxidative stress.

DISCUSSION

The present study suggests that oxidative stress hasopposite effects on the differentiation of calcifying vas-cular cells and bone cells. Xanthine/xanthine oxidase andH2O2 increased differentiation in CVC while inhibitingdifferentiation in bone cells, as assessed by their effecton an early differentiation marker, alkaline phosphataseactivity. Furthermore, XXO increased mineral formation,a late marker of differentiation, in CVC while inhibitingmineralization in M2-10B4 and MC3T3-E1 cells. XXOand H2O2 increased ROS generation in all three celltypes. We had previously found that MM-LDL alsoaffected differentiation oppositely in bone cells andCVC. In this report, we further demonstrated that MM-LDL increased intracellular oxidative stress in all celltypes tested. Furthermore, the effects of XXO and MM-LDL on alkaline phosphatase activity were inhibited bytreating cells with antioxidants.

In 1990 Sohal and Allen presented the “oxidativestress hypothesis of aging,” in which they state that agingis one stage of development, and that changes in geneexpression during different stages of development ordifferentiation can be influenced by oxidative stress [31].Previous evidence has suggested that oxidative stressmay mediate cellular differentiation in various cell types,such as hepatoma cells, peripheral blood dendritic cells,and osteoclastic cells [18,32,33]. The present resultssupport this role of oxidative stress in differentiation.Induction of ROS generation strongly influencedosteoblastic differentiation by stimulating vascular celldifferentiation and inhibiting bone cell differentiation.Previous reports used a lower concentration of H2O2 inserum-free media [34,35]; however, in order to avoidcompounding effects of oxidative stress and serum star-vation, we used a higher concentration of H2O2 in thepresence of 2.5% serum.

Furthermore, H2O2 has been shown to enhance oste-oclastic differentiation [19], while our results showed aninhibition of osteoblastic differentiation. Other studies,using TGF-b and nitric oxide, have also found opposingeffects on osteoclastic and osteoblastic differentiation,suggesting that H2O2 may not be the only agent that hasa reciprocal effect on these two cell types [36–38]. ROS,released by osteoclasts [39], may be involved in degrad-ing bone matrix and inhibiting differentiation of osteo-blasts, resulting in promotion of bone resorption andinhibition of bone formation in close proximity [40,41].

Antioxidants, such as PDTC and Trolox, inhibit in-creases in oxidative stress [27–30]. The present results

Fig. 5. Increased intracellular ROS production in CVC, MC3T3-E1,M2-10B4 stimulated by minimally oxidized-LDL (MM-LDL). DCF flu-orescence in response to MM-LDL (200mg/ml) in CVC (A), MC3T3-E1(B), and M2-10B4 (C). All cell types were loaded with DCFH-DA as inFig. 1 and were treated with MM-LDL, native LDL (N-LDL), or controlbuffer for 30 min. DCF fluorescence was measured after removal of themedia and normalized to cell number. Results are the mean6 SD of atleast four wells and one representative experiment for each cell type isshown (*p , .001 for MM-LDL compared with either buffer or N-LDL).For each cell type, at least three experiments were performed. The com-bined analysis of the multiple experiments for each cell type also yieldedsignificant results (p , .05).

515ROS and osteoblast differentiation

showed that PDTC inhibited and Trolox partially inhib-ited XXO-induced differentiation of CVC, while Troloxcompletely blocked MM-LDL-induced differentiation ofCVC. Similarly, PDTC and Trolox blocked the inhibi-tory effects of XXO and MM-LDL, respectively, ondifferentiation in MC3T3-E1 cells. Interestingly, Troloxalone enhanced alkaline phosphatase activity inMC3T3-E1 cells, suggesting that antioxidants may en-hance bone cell differentiation by reducing basal ROSgeneration. The different effects of the two antioxidantsmay relate to their mechanisms. Trolox inhibits lipidoxidation by oxidized LDL in cultured endothelial cells,as assessed by TBARS [42], and scavenges peroxyl

radical and peroxynitrite [29,30]. PDTC reduces proteinand low molecular weight thiols and inhibits NFkB bind-ing to DNA [43,44]; increased levels of reduced gluta-thione would help in scavenging H2O2 [27]. This sug-gests that XXO and MM-LDL may be functioning byproducing different ROS. However, the failure of PDTCto inhibit completely MM-LDL-induced alkaline phos-phatase activity in CVC may be due to pro-oxidanteffects of PDTC after prolonged incubation [45,46]. Al-ternatively, the concentration of MM-LDL may havebeen too high to observe PDTC protection becausePDTC alone displayed cytotoxicity at higher concentra-tions (. 10 mM). These results suggest that oxidative

Fig. 6. Inhibition of XXO- and MM-LDL-induced alkaline phosphatase activity by antioxidants in CVC. CVC were pretreated for 2 hwith the indicated concentration of Trolox or PDTC, and followed by addition of XXO (X5 25 mM and XO 5 25 mU/ml) orMM-LDL (200 mg/ml) with the antioxidant. The figures are as follows: Trolox and XXO (A), PDTC and XXO (B), Trolox andMM-LDL (C), PDTC and MM-LDL (D). Alkaline phosphatase activity was measured 2 d after XXO addition or 4 d after MM-LDLaddition. Results from a representative of three experiments are shown as the mean6 SD of quadruplicate determinations (*p # .001).The combined analysis of three experiments for panel (A) and (B) also yielded significant results (p , .01). Each experiment yieldedsignificant results for panel (C) (p , .005).

516 N. MODY et al.

events induce differentiation of CVC, but that MM-LDLand XXO may generate different ROS intracellularly, asevidenced by the differing effects of the antioxidants. Tofurther identify the ROS involved in inducing differen-tiation of CVC, these cells were treated with deferox-amine, benzoic acid, or DMSO in the presence of XXO.Our results with deferoxamine suggest that Fenton prod-ucts do not play a role in XXO-induced alkaline phos-phatase. Inhibition of alkaline phosphatase activity byDMSO, but not by benzoate, suggest that hydroxyl rad-icals may have little or no role in CVC differentiation.

DCFH oxidation occurs in the presence of peroxyni-trite and H2O2 in the presence of peroxidase, cytochromec, or Fe12, but not in the presence of O2

•2 [47,48]. DCFfluorescence may also be directly affected by peroxidaseenzymes [49]. Thus, differences in peroxidase activity inCVC, MC3T3-E1, and M2-10B4 cells may have contrib-uted to observed differences in DCF data. Furthermore,oxidation of DCFH to DCF leads to superoxide anionproduction [24]. H2O2 forms as a result of the dismuta-tion of superoxide and can result in an amplification ofthe DCF fluorescence signal. Thus, the increase in DCFfluorescence with H2O2 may exaggerate real changes inoxidative stress.

Altogether, the present study demonstrates that oxi-dative stress enhances differentiation of calcifying vas-cular cells and inhibits differentiation of bone cells. Wehave hypothesized that accumulation of atherogenic ox-idized lipids, such as MM-LDL, in the vessel wall andbone may be the common underlying mechanism behindthe parallel development of vascular calcification andosteoporosis. The present study further suggests thatoxidative stress is the mechanism of MM-LDL action invascular and bone cells. Because lipids also appear toinduce vascular calcification and bone loss in vivo [50,

51], and antioxidants inhibit the in vitro effects of MM-LDL, interventions that target oxidative stress may serveas future strategies against both diseases. In fact, prelim-inary in vivo studies suggest that antioxidants have ben-eficial effects on bone [52].

Acknowledgements— This research was supported by NIH grantHL30568 and the Laubisch Fund. N. Mody was also supported by theJennifer Buchwald Endowment, and F. Parhami is a recipient of NIHgrant AG10415 through the Claude D. Pepper Older American Inde-pendence Center at UCLA. We thank J. Berliner and B. Premack fortheir guidance, V. Le for technical assistance, and J. Bishop and A.Gasparyan for manuscript preparation.

REFERENCES

[1] Carew, T. E.; Schwenke, D. C.; Steinberg, D. Antiatherogeniceffect of probucol unrelated to its hypocholesterolemic effect:evidence that antioxidantin vivo can selectively inhibit low den-sity lipoprotein degradation in macrophage-rich fatty streaks andslow the progression of atherosclerosis in the Watanabe heritablehyperlipidemic rabbit.Proc. Natl. Acad. Sci. USA84:7725–7729;1987.

[2] Keaney, J. F. Jr.; Xu, A.; Cunningham, D.; Jackson, T.; Frei, B.;Vita, J. A. Dietary probucol preserves endothelial function incholesterol-fed rabbits by limiting vascular oxidative stress andsuperoxide generation.J. Clin. Invest.95:2520–2529; 1995.

[3] Badwey, J. A.; Karnovsky, M. L. Active oxygen species and thefunctions of phagocytic leukocytes.Annu. Rev. Biochem.49:695–726; 1980.

[4] Shimizu, S.; Ishii, M.; Yamamoto, T.; Kawanishi, T.; Momose,K.; Kuroiwa, Y. Bradykinin induces generation of reactive oxy-gen species in bovine aortic endothelial cells.Res. Commun.Chem. Pathol. Pharmacol.84:301–314; 1994.

[5] Yang, W.; Block, E. R. Effect of hypoxia and reoxygenation onthe formation and release of reactive oxygen species by porcinepulmonary artery endothelial cells.J. Cell. Physiol.164:414–423; 1995.

[6] Ohara, Y.; Peterson, T. E.; Harrison, D. G. Hypercholesterolemiaincreases endothelial superoxide anion production.J. Clin. Invest.91:2546–2551; 1993.

[7] Cominacini, L.; Garbin, U.; Pasini, A. F.; Davoli, A.; Campag-nola, M.; Pastorino, A. M.; Gaviraghi, G.; Lo Cascio, V. Oxidizedlow-density lipoprotein increases the production of intracellularreactive oxygen species in endothelial cells; inhibitory effect oflacidipine.J. Hypertens.16:1913–1919; 1998.

[8] Bouloumie, A.; Marumo, T.; Lafontan, M.; Busse, R. Leptininduces oxidative stress in human endothelial cells.FASEB J.13:1231–1238; 1999.

[9] Griendling, K. K.; Ushio-Fukai, M. Redox control of vascularsmooth muscle proliferation.J. Lab. Clin. Med.132:9–15; 1998.

[10] Sundaresan, M.; Yu, Z.-X.; Ferrans, V. J.; Irani, K.; Finkel, T.Requirement for generation of H2O2 for platelet-derived growthfactor signal transduction.Science270:296–299; 1995.

[11] Meyer, M.; Schreck, R.; Mu¨ller, J. M.; Baeuerle, P. A. Redoxcontrol of gene expression by eukaryotic transcription factorsNF-kB, AP-1 and SRF/TCF. In: Pasquier, C.; Olivier, R.Y.;Auclair, C.; Packer, L., eds.Oxidative stress, cell activation andviral infection. Basel, Boston, Berlin: Birkhauser Verlag; 1994:217–234.

[12] Watson, K. E.; Bostro¨m, K.; Ravindranath, R.; Lam, T.; Norton,B.; Demer, L.L. TGF-b1 and 25-hydroxycholesterol stimulateosteoblast-like vascular cells to calcify.J. Clin. Invest.93:2106–2113; 1994.

[13] Parhami, F.; Morrow, A.D.; Balucan, J.; Leitinger, N.; Watson,A. D.; Tintut, Y.; Berliner, J. A.; Demer, L. L. Lipid oxidationproducts have opposite effects on calcifying vascular cell andbone cell differentiation: a possible explanation for the paradox of

Table 1. Effect of Antioxidants on MC3T3-E1

TreatmentAlkaline phosphatase activity

(percent of control)

Control 100XXO 41 6 4.3PDTC 1 XXO 93 6 17PDTC (10mM) 73 6 5.4Control 100MM-LDL (200 mg/ml) 446 6.3Trolox 1 MM-LDL 161 6 4.5Trolox (100mM) 1486 17.4

Inhibition of XXO- and MM-LDL-inhibited alkaline phosphataseactivity by antioxidants in MC3T3-E1. MC3T3-E1 were pretreated for2 h with 100 mM Trolox, followed by addition of MM-LDL (200mg/ml) with the antioxidant; or 10mM PDTC, followed by addition ofXXO (X 5 25 mM and XO 5 25 mU/ml) with the antioxidant.Alkaline phosphatase activity was measured 4 d after XXO addition or6 d after MM-LDL addition. Results from a representative of threeexperiments are shown as the mean6 SD of quadruplicate determi-nations (p , .01 for each experiment).

517ROS and osteoblast differentiation

arterial calcification in osteoporotic patients.Arterioscler.Thromb. Vasc. Biol.17:680–687; 1997.

[14] Banks, L. M.; Macsweeney, J. E.; Stevenson, J. C. Effect ofdegenerative spinal and aortic calcification on bone density mea-surements in post-menopausal women: links between osteoporo-sis and cardiovascular disease?Eur. J. Clin. Invest.24:813–817;1994.

[15] Ouchi, Y.; Akashita, M.; De Souza, A. C.; Nakamura, T.;Orimo, H. Age-related loss of bone mass and aortic/aorticvalve calcification—re-evaluation of recommended dietary al-lowance of calcium in the elderly.Ann. N. Y. Acad. Sci.676:297–307; 1993.

[16] Roholl, P. J. M.; Blauw, E.; Zurcher, C.; Dormans, J. A. M. A.;Theuns, H. M. Evidence for a diminished maturation of preos-teoblasts into osteoblasts during aging in rats: an ultrastructuralanalysis.J. Bone Miner. Res.9:355–366; 1994.

[17] Shih, M.-S.; Cook, M. A.; Spence, C. A.; Palnitkar, S.; McElroy,H.; Parfitt, A. M. Relationship between bone formation rate andosteoblast surface on different subdivisions of the endosteal en-velope in aging and osteoporosis.Bone14:519–521; 1993.

[18] Garrett, I. R.; Boyce, B. F.; Oreffo, R. O. C.; Bonewald, L.; Poser,J.; Mundy, G. R. Oxygen-derived free radicals stimulate oste-oclastic bone resorption in rodent bone in vitro and in vivo.J. Clin. Invest.85:632–639; 1990.

[19] Fraser, J. H. E.; Helfrich, M. H.; Wallace, H. M.; Ralston, S. H.Hydrogen peroxide, but not superoxide, stimulates bone resorp-tion in mouse calvariae.Bone19:223–226; 1996.

[20] Parhami, F.; Jackson, S. M.; Tintut, Y.; Le, V.; Balucan, J. P.;Territo, M.; Demer, L. L. Atherogenic diet and minimally oxi-dized low density lipoprotein inhibit osteogenic and promoteadipogenic differentiation of marrow stromal cells.J. Bone Miner.Res.14:2067–2078; 1999.

[21] Mossmann, T. Rapid colorimetric assay for cellular growth andsurvival: application to proliferation and cytotoxicity assays.J. Immunol. Methods65:55–63; 1983.

[22] Parhami, F.; Fang Z. T.; Fogelman, A. M.; Andalibi, A.; Territo,M. C.; Berliner, J. A. Minimally modified low density lipoprotein-induced inflammatory responses in endothelial cells are mediatedby cyclic adenosine monophosphate.J. Clin. Invest.92:471–478;1993.

[23] Sarafian, T. A.; Vartavarian, L.; Kane, D. J.; Bredesen, D. E.;Verity, M. A. bcl-2 expression decreases methyl mercury-inducedfree-radical generation and cell killing in a neural cell line.Toxi-col. Lett.74:149–155; 1994.

[24] Rota, C.; Chignell, C. F.; Mason, R. P. Evidence for freeradical formation during the oxidation of 29-79-dichlorofluo-rescin to the fluorescent dye 29-79-dichlorofluorescin by horse-radish peroxidase: possible implications for oxidative stressmeasurements.Free Radic. Biol. Med.27:873– 881; 1999.

[25] Stein, G. S.; Lian, J. B. Molecular mechanisms mediating prolif-eration/differentiation interrelationships during progressive devel-opment of the osteoblast phenotype.Endocr. Rev.14:424–442;1993.

[26] Santanam, N.; Auge´, N.; Zhou, M.; Keshava, C.; Parthasarathy, S.Overexpression of human catalase gene decreases oxidized lipid-induced cytotoxicity in vascular smooth muscle cells.Arterio-scler. Thromb. Vasc. Biol.19:1912–1917; 1999.

[27] Moellering, D.; McAndrew, J.; Jo, H.; Darley-Usmar, V. M.Effects of pyrrolidine dithiocarbamate on endothelial cells: pro-tection against oxidative stress.Free Radic. Biol. Med.26:1138–1145; 1999.

[28] Nathens, A. B.; Bitar, R.; Davreux, C.; Bujard, M.; Marshall,J. C.; Dackiw, A. P. B.; Watson, R. W. G.; Rotstein, O. D.Pyrrolidine dithiocarbamate attenuates endotoxin-induced acutelung injury. Am. J. Respir. Cell Mol. Biol.17:608–616; 1997.

[29] Salgo, M. G.; Pryor, W. A. Trolox inhibits peroxynitrite-mediatedoxidative stress and apoptosis in rat thymocytes.Arch. Biochem.Biophys.332:482–488; 1996.

[30] Regoli, F.; Winston, G. W. Quantification of total oxidant scav-enging capacity of antioxidants for peroxynitrite, peroxyl radicals,

and hydroxyl radicals.Toxic. Appl. Pharmacol.156:96–105;1999.

[31] Sohal, R. S.; Allen, R. G. Oxidative stress as a causal factor indifferentiation and aging: a unifying hypothesis.Exp. Gerontol.25:499–522; 1990.

[32] Ren, J.-G.; Zheng, R.-L.; Shi, Y.-M.; Gong, B.; Li, J.-F. Apopto-sis, redifferentiation and arresting proliferation simultaneouslytriggered by oxidative stress in human hepatoma cells.Cell Biol.Int. 22:41–49; 1998.

[33] Rutault, K.; Alderman, C.; Chain, B. M.; Katz, D. R. Reactiveoxygen species activate human peripheral blood dendritic cells.Free Radic. Biol. Med.26:232–238; 1999.

[34] Tournier, C.; Thomas, G.; Pierre, J.; Jacquemin, C.; Pierre, M.;Saunier, B. Mediation by arachidonic acid metabolites of theH2O2-induced stimulation of mitogen-activated protein kinases(extracellular-signal-regulated kinase and c-Jun NH2-terminal ki-nase).Eur. J. Biochem.244:587–595; 1997.

[35] Shibanuma, M.; Kuroki, T.; and Nose, K. Release of H2O2 andphosphorylation of 30 kilodalton proteins as early responses ofcell cycle-dependent inhibition of DNA synthesis by transforminggrowth factor beta 1.Cell Growth Differ.2:583–591; 1991.

[36] Filvaroff, E.; Erlebacher, A.; Ye, J.; Gitelman, S. E.; Lotz, J.;Heillman, M.; Derynck, R. Inhibition of TGF-beta receptor sig-naling in osteoblasts leads to decreased bone remodeling andincreased trabecular bone mass.Development126:4267–4279;1999.

[37] Erlebacher, A.; Filvaroff, E.H.; Ye, J.; Derynck, R. Osteoblasticresponses to TGF-b during bone remodeling.Mol. Biol. Cell9:1903–1918; 1998.

[38] Chae, H. J.; Park, R. K.; Chung, H. T.; Kang, J. S.; Kim, M. S.;Choi, D. Y.; Bang, B. G.; Kim, H. R. Nitric oxide is a regulatorof bone remodeling.J. Pharm. Pharmacol.49:897–902; 1997.

[39] Steinbeck, M. J.; Appel, W. H. Jr.; Verhoeven, A. J.; Karnovsky,M. J. NADPH-oxidase expression and in situ production of su-peroxide by osteoclasts actively resorbing bone.J. Cell Biol.126:765–772; 1994.

[40] Ries, W. L.; Key, L. L. Jr.; Rodriguiz, R. M. Nitroblue tetrazo-lium reduction and bone resorption by osteoclasts in vitro inhib-ited by a manganese-based superoxide dismutase mimic.J. BoneMiner. Res.7:931–939; 1992.

[41] Suzuki, H.; Hayakawa, M.; Kobayashi, K.; Takiguchi, H.; Abiko,Y. H2O2-derived free radicals treated fibronectin substratum re-duces the bone nodule formation of rat calvarial osteoblast.Mech.Ageing Dev.98:113–125; 1997.

[42] Mabile, L.; Fitoussi, G.; Periquet, B.; Schmitt, A.; Salvayre, R.;Negre-Salvayre, A.a-Tocopherol and trolox block the early in-tracellular events (TBARS and calcium rises) elicited by oxidizedlow density lipoproteins in cultured endothelial cells.Free Radic.Biol. Med.19:177–187; 1995.

[43] Brennan, P.; O’Neill, L. A. 2-Mercaptoethanol restores the abilityof nuclear factor kappa B (NFkB) to bind DNA in nuclear extractsfrom interleukin 1-treated cells incubated with pyrollidine dithio-carbamate (PDTC). Evidence for oxidation of glutathione in themechanism of inhibition of NFkB by PDTC. Biochem. J.320:975–981; 1996.

[44] Schreck, R.; Meier, B.; Me`annel, D. N.; Dre`oge, W.; Baeuerle,P. A. Dithiocarbamates as potent inhibitors of nuclear factorkappa B activation in intact cells.J. Exp. Med.175:1181–1194;1992.

[45] Nobel, C. S. I.; Kimland, M.; Lind, B.; Orrenius, S.; Slater,A. F. G. Dithiocarbamates induce apoptosis in thymocytes byraising the intracellular level of redox-active copper.J. Biol.Chem.270:26202–26208; 1995.

[46] Pinkus, R.; Weiner, L. M.; Daniel, V. Role of oxidants andantioxidants in the induction of AP-1, NF-kB, and glutathioneS-transferase gene expression.J. Biol. Chem.271:13422–13429;1996.

[47] LeBel, C. P.; Ischiropoulos, H.; Bondy, S. C. Evaluation of theprobe 29,79-dichlorofluorescin as an indicator of reactive oxygenspecies formation and oxidative stress.Chem. Res. Toxicol.5:227–231; 1992.

518 N. MODY et al.

[48] Possel, H.; Noack, H.; Augustin, W.; Keilhoff, G.; Wolf, G.2,7-Dihydrodichlorofluorescein diacetate as a fluorescent markerfor peroxynitrite formation.FEBS Lett.416:175–178; 1997.

[49] Rota, C.; Fann, Y. C.; Mason, R. P. Phenoxyl free radical forma-tion during the oxidation of the fluorescent dye 29,79-dichlorofluo-rescein by horseradish peroxidase.J. Biol. Chem.274:28161–28168; 1999.

[50] Roselaar, S. E.; Kakkanathu, P. X.; Daugherty, A. Lymphocytepopulations in atherosclerotic lesions of apoE2/2 and LDL

receptor2/2 mice. Decreasing density with disease progres-sion. Arterioscler. Thromb. Vasc. Biol.16:1013–1018;1996.

[51] Wang, G. J.; Chung, K. C.; Shen, W. J. Lipid clearing agents insteroid-induced osteoporosis.J. Formos. Med. Assoc.94:589–592; 1995.

[52] Xu, H.; Watkins, B. A.; Seifert, M. F. Vitamin E stimulatestrabecular bone formation and alters epiphyseal cartilage mor-phometry.Calcif. Tissue Int.57:293–300; 1995.

519ROS and osteoblast differentiation