Activating transcription factor 4 regulates stearate-induced vascular calcification.

Previously, we reported that stearate, a saturated fatty acid, promotes osteoblastic differentiation and mineralization of vascular smooth muscle cells (VSMC). In this study, we examined the molecular mechanisms by which stearate promotes vascular calcification. ATF4 is a pivotal transcription factor in osteoblastogenesis and endoplasmic reticulum (ER) stress. Increased stearate by either supplementation of exogenous stearic acid or inhibition of stearoyl-CoA desaturase (SCD) by CAY10566 induced ATF4 mRNA, phosphorylated ATF4 protein, and total ATF4 protein. Induction occurred through activation of the PERK-eIF2α pathway, along with increased osteoblastic differentiation and mineralization of VSMCs. Either stearate or the SCD inhibitor but not oleate or other fatty acid treatments also increased ER stress as determined by the expression of p-eIF2α, CHOP, and the spliced form of XBP-1, which were directly correlated with ER stearate levels. ATF4 knockdown by lentiviral ATF4 shRNA blocked osteoblastic differentiation and mineralization induced by stearate and SCD inhibition. Conversely, treatment of VSMCs with an adenovirus containing ATF4 induced vascular calcification. Our results demonstrated that activation of ATF4 mediates vascular calcification induced by stearate.

Cardiovascular disease, such as vascular calcifi cation, is the leading cause of death in patients with chronic kidney disease, accounting for over 50% of deaths ( 1,2 ). Vascular calcifi cation is frequently observed in advanced atherosclerotic lesions and is a highly regulated process that recapitulates osteogenesis in bone formation. Recent in vivo and in vitro studies have implicated the involvement of numerous positive and negative regulators, including serum phosphate, several lipid-derived molecules (saturated fatty acids, oxys-MOVAS-1 cells were treated with triacsin C (Enzo Life Sciences) and free fatty acid-BSA complexes. The fatty acid-BSA complexes were generated as previously described ( 12,26 ). The medium was changed every 2-3 days. Seven days after reaching confl uence, the cells were stained with Alizarin red to identify calcium deposits.

Calcium content in cultured cells
Calcium deposition in the plates was quantifi ed as previously described ( 11 ). Cells were decalcifi ed using a 0.6 M HCl solution. After collecting the supernatant, the cells were washed with PBS and solubilized with a 0.1 N NaOH/0.1% SDS solution for protein quantifi cation. Calcium content was quantifi ed calorimetrically using the o-cresolphthalein method. Protein content was measured using a BCA protein assay kit.

RNA analysis
Total RNA was isolated using Tri reagent in conjunction with an RNAeasy kit. Real-time quantitative PCR assays were performed by using an Applied Biosystems StepOne qPCR instrument. In brief, 1 g of total RNA was reverse transcribed with random hexamers by using the High Quality Reverse Transcription Reagents Kit (Applied Biosystems). Each amplifi cation mixture (10 l) contained 25 ng cDNA, 900 nM forward primer, 900 nM reverse primer, and 5 l of Universal Fast PCR Master Mix. The quantifi cation of given genes was expressed as the mRNA level normalized to a ribosomal housekeeping gene (18S or 36B4) using the ⌬ ⌬ Ct method. Primer sequences are available upon request. The spliced form of X-box binding protein-1 (XBP-1) was analyzed by RT-PCR coupled with PstI digestion, as described previously ( 27 ).

Adenoviral transduction for MOVAS-1 cells
MOVAS-1 cells were infected with recombinant adenoviruses at a multiplicity of infection ( MOI ) of 40. An adenovirus expressing ATF4 was generated using the ViraPower Adenovirus Expression System (Invitrogen). MOVAS-1 cells were infected with the adenovirus in DMEM with 10% FBS. After 6 h, the infected cells were treated with fresh media for 7 days.

Western blotting
Cell and tissue lysates were prepared using RIPA buffer (Cell Signaling). The samples were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and immunoblotted with an ATF4 antibody (Santa Cruz Biotechnology), CHOP antibody and GAPDH (Santa Cruz Biotechnology), and phospho-serine antibody (Millipore). Samples were visualized using horseradish peroxidase coupled to an anti-mouse secondary antibody, with enhancement by an ECL detection kit. For phosphorylated (p-)ATF4 detection, the promoted mineralization of VSMCs compared with other fatty acids. Inhibition of acetyl-CoA carboxylase or acyl-CoA synthetase reduced mineralization of VSMCs, whereas inhibition of stearoyl-CoA desaturase (which converts stearate to oleate) promoted mineralization and osteoblastic differentiation ( 12 ). We also found that increased cholesterol synthesis and uptake by LXR and SREBP-1c activations contributed to vascular calcifi cation ( 13 ). Therefore, we concluded that a stearate metabolite derived from lipogenesis promotes vascular calcifi cation. However, the molecular mechanism by which stearate metabolites promote osteoblastic differentiation and mineralization of VSMCs remains elusive.
The endoplasmic reticulum (ER) is a major site for the regulation of calcium and lipid homeostasis. ER stress (also known as the unfolded protein response, UPR) is an integrated signal transduction pathway involved in the localization and folding of secreted and transmembrane proteins. A number of cellular stress conditions lead to the accumulation of unfolded or misfolded proteins in the ER lumen. The UPR is initiated by activation of three molecules: PKR-like endoplasmic reticulum kinase (PERK), inositol-requiring enzyme 1 (IRE1), and activating transcription factor 6 (ATF6) ( 14,15 ). Activation of PERK leads to the phosphorylation of the ␣ -subunit of eukaryotic initiation factor 2 (eIF2 ␣ ), which inhibits the assembly of the 80S ribosome and inhibits protein synthesis ( 16,17 ). In contrast to most proteins, activating transcription factor 4 (ATF4) is not affected by the translational attenuation of eIF2 ␣ phosphorylation because ATF4 has two small upstream open reading frames in its 5 ′ -untranslated region. These upstream open reading frames, which prevent translation of the true ATF4 under normal conditions, are bypassed only when eIF2 ␣ is phosphorylated, and thereby permit ATF4 translation ( 16,18 ). ATF4 is a pivotal transcription factor that mediates not only ER stress but also osteoblastic differentiation ( 18,19 ). The transcriptional activity of ATF4 is regulated through posttranslational phosphorylation by ribosomal S6 kinase 2 (RSK2) and PKA (18)(19)(20). In addition to ER stress markers, such as ATF3 and C/EBP homologous protein (CHOP), transcriptional targets of ATF4 include osteocalcin, an osteoblast-specifi c marker for the late stage of osteoblast differentiation ( 18 ), and osterix, another essential transcription factor in osteoblast differentiation ( 18,(20)(21)(22). ATF4 is also required for preserving mature osteoblast functions, including the synthesis of collagen, the most abundant extracellular protein found in bones and calcifi ed vasculatures ( 18 ). It has recently been reported that ER stress mediated via the PERK-eIF2 ␣ -ATF4 pathway is involved in osteoblast differentiation induced by bone morphogenetic protein-2 ( 23 ). However, the role of ATF4 in the regulation of vascular calcifi cation and vascular osteogenesis has yet to be determined.

Cell culture studies
MOVAS-1 cells were kindly provided by Dr. Husain at the University of Toronto and cultured in DMEM containing 10% FBS with either 3.0 mM phosphate or 5.0 mM glycerophosphate ( 24,25 ). compared with no treatment ( Fig. 1A ) ( 12 ). Other fatty acids, such as palmitoleate, oleate, and vaccenate, did not affect mineralization of MOVAS-1 cells ( Fig. 1A ). The procalcifi c effect of stearate showed dose dependency ( Fig. 1B ). In addition, stearate treatment but not other fatty acid treatments signifi cantly induced osteoblastic differentiation as assayed with ALP activity ( Fig. 1C ) and osteocalcin (OCN) gene expression ( Fig. 2B ).
We next examined whether treatment of MOVAS-1 cells with stearate or other fatty acids increases the expression of ATF4, a pivotal transcription factor not only in osteogenesis but also in ER stress. MOVAS-1 cells were treated with either 200 M of stearate or another fatty acid, such as palmitate, palmitoleate, oleate, or vaccenate, for 6 h. Stearate treatment increased ATF4 protein and mRNA levels by 23.9-fold and 7.0-fold, respectively, compared with no treatment ( Fig. 2A, B ). Palmitate treatment also increased ATF4 protein levels, but the effect was weaker than it was with stearate treatment ( Fig. 2C ). Treatment of MOVAS-1 cells with unsaturated fatty acids, including lysates were immunoprecipitated with an ATF4 antibody and immunoblotted using a phospho-serine antibody.

Stearate levels in ER
VSMC homogenate was centrifuged at 10,000 g twice to remove the large debris. The resulting supernatant was layered on a discontinuous sucrose gradient (2 ml of 30% sucrose and 4 ml of 38% sucrose prepared in 10 mM HEPES, pH 7.4) and subjected to centrifugation at 100,000 g for 2 h to obtain the ER, which precipitated. Total lipids from the ER fraction were isolated by Bligh and Dyer's method. Fatty acids were quantifi ed using gas chromatography as previously described ( 28 ). Protein content was measured using a BCA protein assay kit.

Stearoyl-CoA desaturase activity
MOVAS-1 cells treated with CAY10566 were incubated with 200 M stearate-BSA complex containing 1 Ci 14 C-stearate. Total lipids were saponifi ed with 3 M sodium hydroxide/ethanol. The saponifi ed fatty acids were separated by 10% silver nitratecoated thin-layer chromatography. The ratio of the cpm in the band corresponding to oleic acid to the cpm in the band corresponding to stearate was used to calculate stearoyl-CoA desaturase (SCD) activity as previously described ( 28 ).

Alkaline phosphatase activity
Alkaline phosphatase (ALP) activity was measured using p -nitrophenyl phosphate as the substrate ( 29 ).

Statistical analysis
Data were collected from more than two independent experiments and were reported as the means ± SE. Statistical analysis for two-group comparison was performed using the Student t -test or one-way ANOVA with a Student-Newman posthoc test for multigroup comparison. Signifi cance was accepted at P < 0.05.

Stearate treatment increases CHOP, phosphorylated ATF4 protein, and total ATF4 protein, associated with increased mineralization and osteoblastic differentiation of VSMCs
Stearate and palmitate treatment induced mineralization of mouse vascular smooth muscle cell line MOVAS-1, an immortalized cell line recently described as an in vitro model of vascular calcifi cation ( 24 ). Similar to our previous observations, both stearate and palmitate increased calcium content by 4.3-fold and 2.4-fold, respectively,  MOVAS-1 cells were treated with 200 M stearate for up to 16 h. Phosphorylated ATF4 levels continuously increased compared with no treatment (time 0), whereas total ATF4 protein levels transiently increased up to 2.9-fold after 6 h of stearate treatment ( Fig. 3A ). We also examined whether PERK-eIF2 ␣ signaling contributed to the induction of ATF4 expression. Phosphorylated eIF2 ␣ levels were quickly and transiently induced between 0.5 and 2 h of treatment prior to the induction of ATF4 expression. CHOP protein was induced by 35.6-fold after 6 h of treatment ( Fig. 3A ). The effect of stearate on p-ATF4, total ATF4, CHOP, and p-eIF2 ␣ protein expression exhibited dose dependency ( Fig. 3B ). At a 200 M concentration, 12 h stearate treatment induced ATF4, CHOP, and sXBP-1 mRNA expression by 3.8-fold, 7.1-fold, and 2.7-fold, respectively ( Fig. 3C, E, G ). Consistent with protein expression, the mRNA levels of ATF4, CHOP, and sXBP-1 continuously increased with 6 h of stearate treatment in a dose-dependent manner ( Fig. 3D, F, H ). The expressions of ATF4, CHOP, and sXBP-1 correlated with increases in calcium content of MOVAS-1 cells by stearate treatment ( Figs . 1 and 3). Similar to the expression of genes involved in ER stress, stearate treatment induced several osteogenic markers, including ALP, OCN, osteoprotegerin (OPG), sodiumdependent phosphate transporter 1 (Pit1), Runx2, and Osterix with maximum induction at 6 h, 12 h, 3 h, 12 h, 12 h, and 48 h, respectively ( Fig. 4 ). oleate, palmitoleate, and vaccenate, did not affect ATF4 protein and mRNA levels ( Fig. 2A-C ). Consistently, mRNA and protein expression of CHOP, a major ATF4 target, were highly induced by stearate treatment but not oleate treatment ( Fig. 2A, B ). Levels of GAPDH protein used as a loading control did not vary between stearate and oleate treatment. In addition, stearate treatment caused a 3.3-fold increase in mRNA levels of the spliced form of X-box binding protein-1 (sXBP-1), another common marker of ER stress ( Fig. 2D ). The unspliced form of XBP-1 (uXBP-1) remained unchanged in MOVAS-1 cells treated with stearate ( Fig. 2D ). We also examined time-and dose-dependent effects of stearate on the expression of p-ATF4, which is an active form of ATF4 in osteoblastic differentiation.  Even in concentrations as low as 10 nM, CAY10566 induced CHOP mRNA expression, indicating that CAY10566 is as potent as other common inducers of ER stress, such as thapsigargin and tunicamycin, which also induce vascular calcifi cation and ATF4 expression ( Fig. 7C and data not shown). We also examined the time-dependent effect of CAY10566 on the expression of ATF4 and other ER stress markers. CAY10566 treatment transiently induced p-ATF4 expression. After 2 h of treatment, p-ATF4 levels were increased by 2.9-fold. Total ATF4 and CHOP protein levels were time-dependently increased by 8.2-fold and 8.1-fold, respectively, at 16 h of treatment ( Fig. 6F ). Phosphorylated PERK levels were increased by 15.7-fold at 2 h of CAY10566 treatment, whereas p-eIF2 ␣ levels were transiently increased by 1.91-fold at 6 h of CAY10566 treatment ( Fig. 6F ). The expressions of ATF4, CHOP, and sXBP mRNA were induced up to 96 h of 300 nM CAY10566 treatment ( Fig. 7B, D To confi rm whether SCD inhibition induces ATF4 and CHOP expression through the PERK-eIF2 ␣ pathway, we treated PERK-knockdown MOVAS-1 cells with 300 nM CAY10566 for 16 h. PERK knockdown completely inhibited the induction of ATF4 and CHOP protein induced by stearate ( Fig. 6G ).
In contrast to inhibition of SCD by CAY10566, SCD1 overexpression by an adenovirus system completely blocked mineralization of MOVAS-1 cells induced by stearate treatment ( Fig. 6H ). Treatment with adenoviruses containing SCD1 increased the expression of SCD1 by 5.3-fold ( Fig. 6I ), which was comparable to the induction of SCD1 by T0901317 treatment. The induction of total ATF4 and CHOP protein expression by stearate treatment was completely blocked by SCD overexpression ( Fig. 6I ).
Not only stearate but also palmitate induced ATF4 expression and mineralization ( Figs. 1 and 2 ). We therefore hypothesized that the elongation of palmitate to stearate is required for palmitate-induced ATF4 induction and mineralization. To examine our hypothesis, we treated Elovl6-knockdown MOVAS-1 cells with palmitate ( Fig. 6J ). Elovl6 knockdown signifi cantly reduced palmitate-induced ATF4 expression and mineralization ( Fig. 6K, L ).

ATF4 knockdown blocks mineralization and osteoblastic differentiation of VSMCs, whereas ATF4 overexpression induces vascular calcifi cation
ATF4-knockdown MOVAS-1 cells were generated by treating MOVAS-1 cells with lentiviral shRNAs of ATF4. MOVAS-1 cells were infected with lentiviruses containing fi ve different shRNAs for ATF4 or a control shRNA. The cells were treated with 5 g/ml puromycin for 7 days to isolate colonies. Three shRNAs out of fi ve effectively reduced ATF4 protein and mRNA expression (data not shown). In this study, we also used MOVAS-1 cells treated with a lentivirus generated by the clone

Alteration of stearate metabolism affects expression of ATF4 through the PERK-eIF2 ␣ pathway
To examine whether alteration of stearate metabolism affects ATF4 expression and ER stress, we used specifi c inhibitors triacsin C and CAY10566 to inhibit acyl-CoA synthetase and stearoyl-CoA desaturase, two major enzymes in stearate metabolism. MOVAS-1 cells were cotreated with stearate (200 M) and triacsin C (5 M). Corresponding with the effect of triacsin C on mineralization and osteoblastic differentiation ( Fig. 5A , B ), triacsin C treatment drastically reduced the expression of p-ATF4 and total ATF4 protein ( Fig. 5C ), CHOP protein and mRNA ( Fig. 5C, D ), and sXBP-1 mRNA ( Fig. 5E ).

DISCUSSION
ATF4 is a critical transcription factor that mediates not only UPR/ER stress but also osteoblastic differentiation during bone formation ( 16,18,(30)(31)(32)(33). However, the role of ATF4 in vascular calcifi cation and osteogenesis in the vasculature has not previously been determined. The results of our present study demonstrate, for the fi rst time to our knowledge, that the activation and induction of ATF4 are key events in the pathogenesis of vascular calcifi cation induced by stearate.
There are four reasons for using MOVAS-1 cells in this study, instead of the primary bovine calcifying vascular cells used in our previous studies ( 11,12 ). First, MOVAS-1 cells have been established as a cell culture model of vascular calcifi cation. Second, all of our previous observations in bovine calcifying vascular cells were perfectly replicated in MOVAS-1 cells. Third, mouse but not bovine lentiviral shRNAs are readily available from several commercial sources. Fourth, puromycin selection for the shRNA experiment signifi cantly affects the growth, mineralization, osteoblastic differentiation, and morphology of primary vascular cells but not of MOVAS-1 cells.
We previously reported that stearate derived from SREBP-1-dependent de novo lipogenesis promotes vascular calcifi cation of bovine aortic calcifying vascular cells ( 12 ). Stearate is one of the major saturated fatty acids in mammals and is acquired through two pathways: i ) dietary fat absorption and ii ) de novo lipogenesis. Although mammals obtain a large amount of stearate from their diets, stearate can also accumulate through de novo fatty acid synthesis. Consistent with our previous report using bovine cells ( 12 ), stearate most potently induced vascular osteoblastic differentiation and calcifi cation in mouse aortic VSMCs. In a number of recent studies using cell culture systems, saturated fatty acids, including stearate, have been shown to exert lipotoxic effects via ER stress (34)(35)(36)(37). We therefore hypothesized that the induction of ATF4 through an ER stress signaling pathway mediates stearate-induced vascular calcifi cation and osteogenesis. Consistent with our hypothesis was our fi nding that stearate treatment drastically and dose-dependently induced p-eIF2 ␣ , ATF4, CHOP, and sXBP-1 expression. Phosphorylated eIF2 ␣ levels increased prior to the induction of ATF4 protein expression in response to stearate and CAY10566 treatment, suggesting that the translation of ATF4 is induced by the activation of eIF2 ␣ .
The effect of stearate on vascular calcifi cation and ATF4 is specifi c compared with other fatty acids. Neither mineralization nor induction of proteins involved in ER stress (ATF4, p-eIF2 ␣ , and CHOP) was detected in MOVAS-1 cells treated with oleate and other unsaturated fatty acids ( Figs. 1 and 2 ). Although palmitate induced mineralization and ER stress in MOVAS-1 cells to a lesser extent than stearate, the induction of ATF4 expression by palmitate of MOVAS-1 cells induced by CAY10566 ( Fig. 8C ). As expected, ATF4 knockdown attenuated the induction of p-ATF4 and total ATF4 protein expression induced by CAY10566 ( Fig. 8D ) or stearate treatment (data not shown), resulting in the reduction of p-ATF4 and CHOP protein expression. Consistent with the immunoblot analysis, qPCR analysis showed that ATF4 defi ciency alleviated the expression of ER stress markers [ATF3, asparagine synthetase (ASNS), CHOP, GADD34, and GRP78] and osteogenic markers (ALP, OCN, and Pit1) induced by CAY10566 treatment ( Fig. 8E ). Consistent results were obtained in ATF4-knockdown MOVAS-1 cells by a lentivirus derived from a different clone TRCN0000071724 (data not shown). In contrast to shRNA-mediated knockdown of ATF4, ATF4 overexpression mediated by the adenovirus (Ad) expression system induced mineralization and osteoblastic differentiation of MOVAS-1 cells. Although infection with Ad-ATF4 increased ATF4 expression by 4.3-fold ( Fig. 8F ), it was lower than ATF4 induction in MOVAS-1 cells treated with stearate (7.0fold, shown in Fig. 2B ). Calcium content, ALP activity, and OCN mRNA levels were increased by 3.9-, 4.5-, and protein expressions, which directly correlated with the reduction of SCD activity in MOVAS-1 cells. CAY10566 treatment also dose-dependently increased stearate levels in the ER of MOVAS-1 cells. In contrast to SCD inhibition, the overexpression of SCD1 completely attenuated ATF4 protein expression and mineralization induced by stearate treatment. In addition, PERK knockdown completely inhibited the induction of ATF4 expression by SCD inhibition and stearate ( Fig. 6G and data not shown), suggesting that PERK senses stearate levels in the ER and activated the eIF2 ␣ -ATF4 signaling pathway. Because sXBP-1 mRNA levels were also increased in MOVAS-1 cells treated with stearate and CAY10566, other UPRs, such as the ATF6 and IRE-1 pathways, may also sense stearate levels in the ER. These pathways may contribute to stearate-induced vascular calcifi cation, as it was recently reported that the IRE-1-XBP-1 pathway contributes to BMP-2-dependent vascular calcifi cation by increasing Runx2 expression ( 40 ). treatment was attenuated by the shRNA-mediated knockdown of Elovl6, which generates stearate from palmitate ( Fig. 6J, K ). These data suggest that the effects of palmitate on ATF4 expression are due to an increase in stearate. We also found that inhibition of acyl-CoA synthetase by triacsin C completely blocked the induction of ATF4 protein and CHOP mRNA expression induced by stearate treatment, accompanied with a reduction of mineralization and osteoblastic differentiation. This inhibition suggests that stearoyl-CoA or its downstream metabolite mediates the induction of ATF4 and ER stress. Stearoyl-CoA desaturase is an ER transmembrane enzyme that regulates the level of stearate in the ER by converting stearate to oleate ( 38,39 ). CAY10566 is able to inhibit the activity of both SCD1 and SCD2, as SCD activity was undetectable in MOVAS-1 cells treated with 300 nM of CAY10566 treatment. Similar to stearate treatment, the treatment of MOVAS-1 cells with CAY10566 dose-dependently and potently induced p-PERK, p-eIF2 ␣ , ATF4, and CHOP   9. Proposed mechanism by which stearate promotes vascular calcifi cation. Palmitate (16:0), derived from lipoproteins and de novo lipogenesis, is elongated to stearate (18:0) in the ER. To induce ER stress, stearate has to be converted to the CoA-conjugated form. Stearoyl-CoA is incorporated into an ER membrane lipid, leading to ER stress that induces the expression of ATF4 through the activation of the PERK-eIF2 ␣ pathway. Activated ATF4 induces osteoblastic differentiation and mineralization in VSMCs.