Regulation of interleukin-6 expression in osteoblasts by oxidized phospholipids.

Epidemiological evidence suggests that cardiovascular disease is associated with osteoporosis, independent of age. Bone resorptive surface is increased in mice on a high-fat diet, and osteoclastic differentiation of bone marrow preosteoclasts is promoted by oxidized phospholipids. Because osteoclastic differentiation requires cytokines produced by osteoblasts, we hypothesized that the stimulatory mechanism of oxidized phospholipids is via induction of osteoclast-regulating cytokines in osteoblasts. To investigate the effects of oxidized phospholipids on expression of such cytokines, murine calvarial preosteoblasts, MC3T3-E1, were treated with oxidized 1-palmitoyl-2-arachidonyl-sn-glycero-3-phosphocholine (ox-PAPC), an active component of oxidized lipoproteins. Results showed that ox-PAPC increased expression of interleukin-6 (IL-6) and tumor necrosis factor-alpha. IL-6 expression was also elevated in calvarial tissues from hyperlipidemic but not in wild-type mice. Ox-PAPC also induced IL-6 protein levels in both MC3T3-E1 and primary calvarial cells. Promoter-reporter assay analysis showed that ox-PAPC, but not PAPC, induced murine IL-6 promoter activity. Effects of ox-PAPC on IL-6 expression and the promoter activity were attenuated by H89, a PKA inhibitor. Analysis of deletion and mutant IL-6 promoter constructs suggested that CAAT/enhancer binding protein (C/EBP) partly mediates the ox-PAPC effects. Taken together, the data suggest that oxidized phospholipids induce IL-6 expression in osteoblasts in part via C/EBP.


Animal tissue
Calvarial tissues were harvested from 15-week-old wild-type (C57BL/6) and hyperlipidemic ( Ldlr Ϫ / Ϫ ) mice (Jackson Laboratory, ME), and total RNA was isolated. The animals were given a standard Purina Chow diet. The experimental protocol was approved by the Institutional Animal Care and Use Committee of the University of California, Los Angeles.

RNA isolation and real-time RT-quantitative PCR
Total RNA was isolated using TRIzol reagent (Invitrogen). Real-time PCR was performed using the One-Step qRT-PCR SuperMix Kit (BioChain, Inc.) and Mx3005P (Stratagene).

DNA constructs
Generation of murine IL-6 promoter luciferase reporter constructs was previously described ( 22 ). The constructs contain either 1277, 231, or 84 bp upstream of the transcriptional start site or a 1277 bp fragment containing mutant C/EBP response elements (-161 to -147) and are designated here as -1277 Luc, -231 Luc, -84 Luc, and mC/EBP Luc, respectively. The plasmids were verifi ed by sequencing.

Transient transfections and luciferase reporter assay
Plasmid transfection was performed with Effectene transfection reagent (Qiagen), according to the manufacturer's instructions. Twenty-four h after transfection, cells were serum starved for 7 h using media containing 1% FBS and treated in the same media for 19 h. Luciferase activity (expressed as relative light units, RLU) was assayed using the Dual-Luciferase Reporter Assay System (Promega). Firefl y luciferase activity was normalized for transfection effi ciency to Renilla luciferase.

IL-6 protein levels
Confl uent cells were treated for 24 h in serum-free, phenol red-free DMEM (Sigma Aldrich). Secreted IL-6 protein levels in cell media were measured using Quantikine Mouse IL-6 Immunoassay Kit (R&D Systems), following the manufacturer's protocol. Assays were performed in triplicate.

Data analysis
Experiments ( у triplicate wells) were performed at least three times and data from one representative experiment are expressed as mean ± SEM. Results were compared using a two-tailed paired Student's t -test. In comparisons across more than two groups, two-way ANOVA, followed by Fisher's Protected Least Signifi cant Difference, was performed using StatView ( v 4.5, Abacus). P < 0.05 is considered statistically signifi cant.

Effects of ox-PAPC on osteoclast-regulating cytokines
To determine whether ox-PAPC regulates expression of osteoclast-regulatory cytokines, MC3T3-E1 were treated with ox-PAPC for 6 and 72 h, and mRNA expression level was assessed by real-time RT-qPCR for IL-6, TNF-␣ , RANKL, and osteoprotegerin (OPG), a decoy receptor of RANKL that inhibits osteoclastic differentiation ( 19 ). As shown in Fig. 1A , ox-PAPC signifi cantly induced IL-6 at both 6 and 72 h, whereas it only increased TNF-␣ at 72 h. OPG and RANKL component of biologically active LDL that is known to cause infl ammatory responses both in vitro and in vivo is ox-PAPC ( 11,12 ).
Oxidized lipids are also likely to accumulate within the skeletal bone. We previously found lipid deposits within the perivascular space of Haversian canals in osteoporotic but not in normal cortical bone ( 6 ). In addition, we have detected oxidized lipids in the bone marrow of hyperlipidemic mice ( 6 ). Recent studies showed that osteoblasts are also capable of oxidizing LDL particles ( 13 ), suggesting a possible increased local concentration of oxidized phospholipids in bones of hyperlipidemic patients.
Elevated levels of osteoclast-promoting cytokines, including receptor activator of nuclear factor kappa B ligand (RANKL), IL-6, and TNF-␣ have been linked to postmenopausal bone loss ( 14 ). IL-6 stimulates osteoclastic differentiation and activity in bone marrow cells ( 15 ). IL-6 also mediates parathyroid hormone-induced bone resorption in vivo ( 16 ), and has been implicated in contributing to the increased sensitivity to bone loss that is often observed in peri-menopausal or estrogen-defi cient women who concurrently suffer from primary hyperthyroidism ( 17,18 ). We previously found that oxidized phospholipids induced osteoclast differentiation when cocultured with osteoblasts ( 7 ). Because osteoclastic differentiation is regulated by cytokines produced by osteoblasts ( 19 ), we hypothesize that oxidized phospholipids promote production of osteoclastregulating cytokines by osteoblasts. To test the hypothesis, we investigated the effects of oxidized phospholipids on expression of osteoclastogenic cytokines in the murine calvarial cell line, MC3T3-E1. Our results showed that oxidized lipids induced mRNA expression of IL-6, and that CAAT/enhancer binding protein (C/EBP) response elements partly mediated the transcriptional regulation of IL-6 expression by oxidized phospholipids.

Cell culture
Calvarial preosteblasts ( 21 ) were a kind gift from Dr. S. Tetradis (UCLA). Both calvarial preosteoblasts and MC3T3-E1 cells were cultured in ␣ -MEM (Invitrogen) supplemented with 10% FBS, penicillin, streptomycin, and sodium pyruvate. Media was replenished every 3 to 4 days, when necessary. Short-term (6-24 h) treatment media consisted of ␣ -MEM without ascorbic acid (Gibco BRL) supplemented with 1% FBS, whereas longer-term was at the level of transcriptional activation, we assessed IL-6 promoter activity using -1277 Luc, which contains luciferase reporter gene under the transcriptional control of the IL-6 promoter (1277 bp upstream of the transcriptional start site). As shown in Fig. 2A , ox-PAPC induced IL-6 promoter activity in a dose-dependent manner. The IL-6 promoter activity was also induced by ox-AA but not unoxidized PAPC ( Fig. 2B ), confi rming that the oxidized arachidonic side chain is the active component. As shown previously ( 23 ), AA (40 µg/ml) also induced IL-6 promoter activity (control: 0.63 ± 0.32 vs. AA: 31.96 ± 6.83, P < 0.05). Interestingly, IL-6 induction was not induced by POVPC, PGPC, or PEIPC, some of the active components of ox-PAPC ( Fig. 2C ). In primary calvarial osteoblasts, ox-PAPC also induced both IL-6 protein levels and promoter activity ( Fig. 3A , B ). expressions were not signifi cantly affected by ox-PAPC ( Fig.  1A ). RANKL expression was not detectable at 72 h in these cells. Because hyperlipidemic mice have elevated levels of oxidized phospholipids, we compared the in vivo cytokine profi le of calvarial tissue from hyperlipidemic and wild-type mice with our in vitro results. We found that IL-6 expression, but not RANKL, OPG, or TNF-␣ , was elevated in the hyperlipidemic Ldlr Ϫ / Ϫ mice compared with wild-type mice ( Fig.  1B ). Results from IL-6 ELISA showed that treatment of MC3T3-E1 cells with ox-PAPC also induced IL-6 protein levels in a dose-dependent manner ( Fig. 1C ).

Intracellular mechanism mediating ox-PAPC effects
Because IL-6 expression is induced both in vitro and in vivo, we further investigated the mechanism of IL-6 induction by ox-PAPC. To investigate whether IL-6 induction activate two other kinases that are also inhibited by H89, mitogen-and stress-activated protein kinase 2 or p70 ribosomal protein s6 kinase 1 (data not shown). Multiple transcription factors, including activator protein-1 (AP-1) and C/EBP, have been reported to regulate IL-6 promoter activity in several cell types ( 22,27,31 ). Therefore, we tested whether AP-1 and C/EBP mediate ox-PAPC effects. To delineate the involvement of AP-1 sites, we used two deletion constructs. As shown in Fig. 5A , ox-PAPC retained induction of IL-6 in the -231 Luc construct where upstream AP-1 response element (-277 to -271) was deleted. In contrast, ox-PAPC effect was attenuated in the -84 Luc construct containing downstream AP-1 response elements (-61 to -55), indicating that AP-1 response elements do not mediate ox-PAPC induction of IL-6 promoter activity. However, the -84 Luc construct lacks the C/EBP response element (-161 to -147), suggesting that C/EBP partly mediates the effects of ox-PAPC. Consistent with this hypothesis, ox-PAPC effects were attenuated partially in the construct containing mutations in the C/EBP response element in the full-length IL-6 promoter (mC/EBP Luc) ( Fig. 5B ). The incomplete attenua-Because the protein kinase A pathway mediates effects of oxidized lipoproteins and ox-PAPC ( 7, 24-26 ) as well as IL-6 gene expression in epithelial cells ( 27 ), we tested the effects of a PKA inhibitor, H89, on ox-PAPC induction of IL-6. As shown in Fig. 4A , H89 attenuated basal IL-6 expression, as reported previously ( 28 ). In addition, H89 attenuated ox-PAPC-induced IL-6 mRNA expression. Consistent with this fi nding, H89 also inhibited IL-6 promoter activity ( Fig. 4B ). Because rho-associated kinase II (ROCK-II) has been shown to be inhibited by H89 ( 29 ) and also shown to mediate parathyroid hormone-induced IL-6 promoter activity ( 30 ), we tested the effect of Y-27632, which inhibits 87% of ROCK-II activity but only 9% of PKA activity ( 29 ). As shown in Fig. 4C, Y -27632 did not attenuate ox-PAPC induction of IL-6 promoter activity, suggesting that ROCK-II does not mediate ox-PAPC effects. Western blot analyses also showed that ox-PAPC did not  induction through osteoblasts. The data suggest that oxidized phospholipids promote osteoblastic production of the osteoclastogenic cytokines IL-6 and TNF-␣ . The induction of IL-6 expression is also evident in the calvarial tissues of hyperlipidemic mice, suggesting that local cytokine expression may be increased. The present results also showed that oxidized lipids did not alter expression of RANKL and OPG, the main regulators of osteoclast differentiation ( 19 ). Lack of RANKL induction and/or OPG inhibition by oxidized phospholipids both in vitro and in tissue suggests that hyperlipidemia/oxidized phospholipid regulation of osteoclastogenesis via osteoblasts may be secondary to their direct effects on preosteoclasts, as we showed previously ( 7 ).
We previously found that oxidized phospholipids have differential effects on vascular and bone cells ( 32 ). They tion suggests the involvement of other transcription factors in mediating ox-PAPC effects.

DISCUSSION
We previously found that bone marrow cells from hyperlipidemic ( Ldlr Ϫ / Ϫ ) mice have increased osteoclastic potential ( 6 ), and that oxidized phospholipids promote osteoclastic differentiation both directly and also indirectly through osteoblasts ( 7 ). Because osteoblasts are principal physiological regulators of osteoclastic differentiation, we investigated the mechanism of lipid-mediated osteoclast  induce osteogenic differentiation of vascular cells whereas they inhibit osteogenic differentiation of bone cells. In the vasculature, certain components of ox-PAPC, namely POVPC, PGPC, and PEIPC, have been shown to mediate activation of monocytes and/or endothelial cells in vitro ( 25 ). Our present data showed that although ox-PAPC effects are due to AA side chain, the bioactive components that are shown to be active in the vasculature may not be involved in the induction of IL-6 in osteoblasts, suggesting that either other components or a combination of two or more components of ox-PAPC are required for IL-6 induction in osteoblasts.
The results also showed that effects of oxidized phospholipids on IL-6 expression are at the transcriptional level. Several transcription factors have been shown to control IL-6 promoter activity upon activation by various stimuli ( 22,27 ). Our results show that deletion or mutation of C/EBP response elements results in partial attenuation, suggesting that C/EBP may be a part of the transcriptional complexes mediating ox-PAPC effects. The incomplete attenuation suggests the involvement of other transcription factors in mediating ox-PAPC effects. Alternatively, the partial inhibition results from the residual activity of the mutated C/EBP site. Analysis of the IL-6 promoter between -231 and -84 sequences with the Transcription Element Search System program revealed several other transcription factor response elements, including cAMP response element binding protein, serum response factor, and the Ets family of transcription factors. Interestingly, a response element closely matching the extended antioxidant/electrophile responsive element ( 33 ) is also found within this region. More studies are required to delineate the factors or response element(s) activated downstream of ox-PAPC stimuli.
In the larger context, the present fi ndings elucidate the regulatory mechanism of oxidized phospholipids on bone cells. Our previous studies showed that oxidized lipids directly inhibit osteoblastic differentiation. The present study suggests that they also induce expression of osteoclastogenic cytokines IL-6 and possibly TNF-␣ , creating a pro-osteoporotic condition. Our result may partly elucidate the fi ndings of Hirasawa et al. ( 4 ) that show decreased bone formation rate and increased bone resorption surface when mice were fed a high-fat diet. Therefore, atherogenic oxidized lipids, which trigger pathogenesis of atherosclerosis, may also lead to osteoporotic conditions. Better understanding of the molecular actions of oxidized phospholipids could lead to novel or more effective therapies to cooperatively combat the commonly associated cardiovascular disease and osteoporosis.
We thank M. Huang and M. Mazzotta for their excellent technical assistance.