Protein kinase R-like endoplasmic reticulum kinase and glycogen synthase kinase-3α/β regulate foam cell formation.

Evidence suggests a causative role for endoplasmic reticulum (ER) stress in the development of atherosclerosis. This study investigated the potential role of glycogen synthase kinase (GSK)-3α/β in proatherogenic ER stress signaling. Thp1-derived macrophages were treated with the ER stress-inducing agents, glucosamine, thapsigargin, or palmitate. Using small-molecule inhibitors of specific unfolded protein response (UPR) signaling pathways, we found that protein kinase R-like ER kinase (PERK), but not inositol requiring enzyme 1 or activating transcription factor 6, is required for the activation of GSK3α/β by ER stress. GSK3α/β inhibition or siRNA-directed knockdown attenuated ER stress-induced expression of distal components of the PERK pathway. Macrophage foam cells within atherosclerotic plaques and isolated macrophages from ApoE(-/-) mice fed a diet supplemented with the GSK3α/β inhibitor valproate had reduced levels of C/EBP homologous protein (CHOP). GSK3α/β inhibition blocked ER stress-induced lipid accumulation and the upregulation of genes associated with lipid metabolism. In primary mouse macrophages, PERK inhibition blocked ER stress-induced lipid accumulation, whereas constitutively active S9A-GSK3β promoted foam cell formation and CHOP expression, even in cells treated with a PERK inhibitor. These findings suggest that ER stress-PERK-GSK3α/β signaling promotes proatherogenic macrophage lipid accumulation.

modulator of distal, proapoptotic elements of the PERK signaling pathway. Moreover, we demonstrate that inhibition of PERK-GSK3 ␣ / ␤ signaling attenuates macrophage lipid biosynthesis and uptake, lipid accumulation, and foam cell formation induced by ER stress.

Cell culture and treatments
Thp-1 human monocytes were cultured in Roswell Park Memorial Institute 1640 medium (Invitrogen) containing 10% fetal bovine serum at 37°C and 5% CO 2 . Monocytes were differentiated into macrophages by exposing the cells to 100 nM PMA for 72 h. Thioglycolate-elicited peritoneal macrophages were isolated from 8-week-old female C57BL6 mice or ApoE Ϫ / Ϫ mice and cultured in DMEM (Life Technologies) containing 10% fetal bovine serum.

Real-time PCR
Total RNA was isolated from cultured cells using an RNeasy mini kit (Qiagen). RNA was quantifi ed by measuring the absorbance at 260 nm, and RNA purity was verifi ed by calculating the ratio of the absorbance at 260 and 280 nm (A 260 /A 280 ). cDNA was reverse transcribed from 1 g of RNA using Superscript II Reverse Transcriptase (Invitrogen). Real-time PCR was performed on the StepOne Plus (Applied Biosystems) using iQ SYBR Green Supermix (Bio-Rad), 1 g cDNA, and 500 nM forward and reverse primers. See supplementary Table I for primer sequences and amplifi ed product size.
ER stress and UPR activation have been associated with the progression and development of atherosclerotic plaques. Multiple cardiovascular risk factors including hyperglycemia ( 6,7 ), hyperhomocysteinemia ( 7,8 ), obesity ( 9 ), cigarette smoke ( 10 ), and elevated concentrations of unesterifi ed cholesterol ( 11 ) or palmitic acid (PA) ( 12 ) have each been shown to induce ER stress. ER stress and the activation of the UPR have been observed in atherosclerosis-prone areas of the vessel wall prior to lesion development ( 13 ) as well as at all stages of plaque progression ( 14 ). Furthermore, the alleviation of ER stress with a chemical chaperone reduces atherosclerotic plaque size in apoE-defi cient (ApoE Ϫ / Ϫ ) mice ( 15 ). ER stress has also been associated with the dysregulation of lipid metabolism by disruption of sterol-regulatory element binding proteins (SREBPs) ( 8,16 ), the induction of infl ammation by nuclear factor B (NF-B) upregulation ( 17,18 ), and activation of the proapoptotic process by induction of CCAAT/enhancer binding protein homologous protein (CHOP) expression ( 19,20 ). However, the molecular mechanisms by which ER stress and the UPR activate these and other proatherogenic pathways remain unresolved.
Glycogen synthase kinase (GSK) 3 is a serine/threonine kinase involved in several different cell signaling pathways [reviewed by Doble and Woodgett ( 21 )]. There are two forms of GSK3 encoded on separate genes, a 51 kDa ␣ form and a 47 kDa ␤ form. Regulation of GSK3 ␣ / ␤ activity is predominantly, but not entirely, through phosphorylation. Phosphorylation at residue serine 21 of ␣ and serine 9 of ␤ is indicative of inhibition, while phosphorylation at tyrosine 279 of ␣ and tyrosine 216 of ␤ is associated with kinase activation. GSK3 ␣ / ␤ activity can also be regulated by altering its subcellular localization ( 22,23 ). Although GSK3 ␣ and GSK3 ␤ share 90% homology within the kinase domain, the enzymes have been shown to have both distinct as well as common substrates (24)(25)(26). Recent evidence suggests that the role of GSK3 ␣ / ␤ in cell metabolism extends to ER stress and the activation of proatherogenic pathways. In cultured cells, conditions of ER stress activate GSK3 ␤ ( 27,28 ). In vivo studies have also demonstrated a role for GSK3 ␣ / ␤ in the regulation of NF-B ( 29 ). Furthermore, our group and others have shown that inhi bition of GSK3 ␣ / ␤ is associated with attenuated atherosclerotic development and reduced hepatic steatosis in different mouse models ( 7,30,31 ). However, the mechanisms by which ER stress modulates GSK3 ␣ and/or ␤ , and how GSK3 ␣ / ␤ regulates proatherogenic processes, remain unresolved. In this study, we present evidence showing that ER stress-induced GSK3 ␣ / ␤ does not regulate the adaptive components of UPR signaling but instead acts as a Liver and heart, including the aortic root, were removed and embedded in paraffi n. Serial sections (4 m thick) of aortic root were collected on precoated glass slides. Sections were stained with primary antibodies against KDEL, CHOP (cat#sc-575, Santa Cruz), or Mac3 (cat#553322, Becton Dickson Co.). Serial sections were stained with preimmune IgG, in place of primary antibodies, to control for nonspecifi c staining. Images were captured with an Olympus microscope and a 12.5 megapixel DP71 digital camera. Immunofl uorescence was quantifi ed using ImageJ 1.43 software. Briefl y, 12 aortic sections from each animal (n = 6 to 7 mice per treatment group), representing the entire length of the lesion, were stained and imaged. Staining intensity above background was determined over a fi xed threshold. The staining intensity of the 12 aortic sections from each animal was averaged to provide a staining intensity for each animal. Data shown are average staining intensities for each animal within the group.

Kinase activity assay
Total GSK3 ␣ / ␤ activity was determined from 250 g total cell protein (supplementary Fig. I). For isoform-specifi c analysis, GSK3 ␣ or GSK3 ␤ were immunoprecipitated from 600 g total cell protein in kinase buffer using a monoclonal antibody specifi c for GSK3 ␤ (cat#610202, BD Transductions) or GSK3 ␣ Technologies), membranes were developed using the Immobilon Western chemiluminescent HRP substrate (Millipore). Protein band intensities were quantifi ed and normalized to ␤ -actin.

Five-week-old female ApoE
Ϫ / Ϫ (B6.129P2-ApoE tm1Unc ) mice (n = 24), purchased from Jackson Labs (stock #002050), were placed on a high-fat diet (HFD) containing 21% milk fat and 0.2% cholesterol (TD97363, Harlan Teklad). After 1 week, half of the mice (n = 12) were switched to an HFD supplemented with 625 mg/kg sodium valproate (VPA), while the other half of the mice (n = 12) remained on unsupplemented HFD. All mice had unrestricted access to both food and water throughout the study. Mice were euthanized at 24 weeks of age for atherosclerotic plaque analysis or 8 weeks of age for analysis of peritoneal macropahges, and blood and tissues were collected for further analysis. The McMaster University Animal Research Ethics Board approved all procedures.

Statistical analysis
All experiments are representative of at least three independent biological experiments. All data are expressed as mean ± SD. An unpaired Student's t -test or one-way ANOVA test was used, as appropriate, to determine statistical signifi cance. A value of P < 0.05 was considered statistically signifi cant.

GSK3 ␣ / ␤ inhibition does not affect the adaptive UPR
Thp-1 human monocytic cells were differentiated into macrophages by exposure to 100 nM PMA for 72 h. The small-molecule GSK3 ␣ / ␤ inhibitor CT99021 was used to directly inhibit GSK3 ␣ / ␤ activity ( 32 ). To confi rm inhibition, GSK3 ␣ and GSK3 ␤ were immunoprecipitated from Thp-1 macrophage lysates, and kinase activity was determined in the presence or absence of 0. To determine the impact of GSK3 ␣ / ␤ inhibition on ER stress-induced chaperone expression, macrophages were pretreated for 2 h in the presence or absence of 4 M CT99021 and then challenged with ER stress-inducing agents, including 1 M Thaps, 5 mM GLN, or 600 M PA, for 18 h. Neither ER stress nor GSK3 ␣ / ␤ inhibition reduced Thp-1 macrophage cell viability below 80% (supplementary Fig. III). Total RNA was isolated, and quantitative real-time PCR was performed. The expression levels of the cellular chaperones and foldases, glucose-related protein (GRP) 78, GRP94, calreticulin, and PDI, were determined ( Fig. 1 ). These components of the adaptive ER stress response were signifi cantly upregulated by Thaps, GLN, and PA ( Fig. 1 ). GSK3 ␣ / ␤ inhibition did not alter GRP78, GRP94, calreticulin, or PDI expression ( Fig. 1 ). Consistent with these fi ndings, siRNA-directed knockdown of GSK3 ␣ / ␤ did not alter the ability of Thaps, GLN, or PA to increase GRP78 protein levels (supplementary Fig. IVA-C). These results suggest that GSK3 ␣ / ␤ activity is not required for early, adaptive UPR signaling.

GSK3 ␣ / ␤ is a target of the PERK signaling pathway
We next investigated the three branches of UPR and the potential role of GSK3 ␣ / ␤ in each of these signaling pathways. Initially, the effect of ER stress on GSK3 ␣ / ␤ activation was determined. ER stress induced by Thaps, GLN, and PA signifi cantly increased GSK3 ␣ / ␤ activity in Thp-1 macrophages ( Fig. 2A ). Macrophages were then exposed to inhibitors of each of the three UPR signaling pathways. Inhibition of the PERK, but not IRE or ATF6, signifi cantly (cat#07-389, Cell Signaling) and Ultra Link immobilized Protein A Plus (Pierce). Kinase activity was measured by monitoring the incorporation of 32 P onto phospho-glycogen synthase peptide-2 (pGS-2; Upstate Biotech). Briefl y, either cell lysate or immunoprecipitated GSK3 ␣ or ␤ was combined with 15 M pGS-2 and 0.5 Ci/ l [ ␥ 32 P]ATP in a reaction mixture containing 20 mM MOPS, 50 M EDTA, 0.25 mM Mg acetate, 5 mM MgCl 2 , 5 mM ␤ -glycerol phosphate, 1 mM EGTA, 0.25 mM Na 3 VO 4 , 0.2 mM DTT, and 35 M ATP in a total volume of 40 l. As background controls, a subset of samples were incubated with 0.5 M CT99021. After 60 min at room temperature, samples were placed on ice, then spotted onto Whatman P81 phosphocellulose paper (GE Healthcare) and washed 3× with 0.75% o -phosphoric acid and once with acetone. 32 P incorporation onto the substrate was determined by scintillation counting, and total counts minus background are reported.

Lipid analysis
Esterifi ed and unesterifi ed cholesterol levels were determined in macrophages using a Cholesterol Quantitation Kit (cat#MAK043, Sigma-Aldrich) according to the manufacturer's instructions. Briefl y, lipids were extracted from 1 × 10 6 cells with chloroform/ isopropanol/IGEPAL CA-630 (7:11:0.1). Lipids were incubated with a cholesterol probe and either with (total cholesterol) or without (free cholesterol) cholesterol esterase for 60 min at 37°C. Esterifi ed cholesterol levels were determined by the difference between total and free cholesterol levels. The absorbance of the sample was determined at 570 nm (A 570 ) and compared with a known standard (supplementary Fig. I). Protein concentrations were determined in the organic phase using a Bradford assay (Bio-Rad). Lipid uptake was determined by treating cells as indicated and then supplementing media with Alexa fl uor 488-AcLDL (7 g/ml) (Life Technologies) for 2 h at 37°C and 5% CO 2 . To observe lipid droplets, cells were grown or differentiated onto glass coverslips and stained with Oil Red O (0.5% w/v) dissolved in isopropanol/PBS (4:3) followed by 4',6-diamidino-2-phenylindole (DAPI). Coverslips were mounted onto slides in Crystalmount media. All images were captured using an Olympus microscope and a 12.5 megapixel DP71 digital camera. Oil Red O and Alexa fl uor 488-AcLDL was quantifi ed using ImageJ 1.43 software. Briefl y, each biological experiment and treatment was conducted a minimum of four times. From each of these biological replicates, a minimum of fi ve images, each containing ‫ف‬ 200 cells, were captured. The stained area over background as well as cell number were quantifi ed. Data from each image of a biological replicate were combined providing an average stained area per cell with a minimum of 1,000 cells. Data shown are average by densitometry (C, D). Total RNA was isolated from similarly treated cells and the expression level of ATF4 (E) and CHOP (F) were quantifi ed by real-time PCR. n = 4, * P < 0.05 relative to untreated cells, # P < 0.05 relative to cells of the same treatment without CT99021.  attenuated ER stress-induced GSK3 ␣ / ␤ activity ( Fig. 2A  and supplementary Fig. V). Activated PERK phosphorylates the eukaryotic initiation factor (eIF) 2 ␣ at serine 51. This phosphorylation event results in the attenuation of general protein translation and the specifi c upregulation of ATF4 and CHOP. Immunoblot analysis of protein lysates from macrophages challenged with Thaps, GLN, or PA shows the expected ER stress-induced phosphorylation of to KDEL staining pattern or intensity in the VPA-supplemented mice. Next, serial aortic sections were costained with anti-Mac3 and anti-CHOP antibodies. We observed a signifi cant reduction in CHOP staining within macrophages of the atherosclerotic plaque in VPA-supplemented mice ( Fig. 3B, D ). Peritoneal macrophages were isolated from a separate group of ApoE Ϫ / Ϫ mice fed an HFD or an HFD supplemented with VPA for 3 weeks. Macrophages isolated from mice fed an HFD supplemented with VPA displayed signifi cantly reduced CHOP protein and mRNA expression, while GRP94 and GRP78 mRNA and protein levels were unchanged relative to HFD-fed mice ( Fig. 3E-H ). These observations are consistent with our in vitro data showing that GSK3 ␣ / ␤ does not regulate upstream, adaptive components of the UPR but does modulate distal components of the PERK signaling pathway.

GSK3 ␣ / ␤ inhibition attenuates ER stress-induced lipid accumulation in macrophages
We hypothesize that ER stress may play a role in the dysregulated accumulation of lipids in atherosclerotic foam cells. Our data suggest that inhibition of GSK3 ␣ / ␤ will modulate this response. To test this, cultured Thp-1 macrophages were challenged with ER stress-inducing agents in the presence or absence of CT99021. After 18 h, the expression levels of selected transcripts involved in lipid biosynthesis and uptake were quantifi ed by real-time PCR. Results show that ER stress was associated with signifi cantly enhanced expression of genes regulating lipid and cholesterol metabolism including FAS, SREBP-1c, SREBP-2, HMG-CoA, and LDL receptor (LDLR) ( Fig. 4 ). Inhibition eIF2 ␣ , indicative of the activation of the PERK signaling pathway ( Fig. 2B, C ). P-eIF2 ␣ levels were unaffected by GSK3 ␣ / ␤ inhibition suggesting that GSK3 ␣ / ␤ does not affect PERK activity directly. However, ER stress-induced CHOP and ATF4 expression were blocked by GSK3 ␣ / ␤ inhibition and siRNA knockdown ( Fig. 2B, D-F , and supplementary Fig. IVA-D). These results indicate that GSK3 ␣ / ␤ plays a role in the regulation of downstream components of the PERK branch of the UPR.

GSK3 ␣ / ␤ inhibition is associated with attenuated CHOP expression in atherosclerotic macrophages
Having identifi ed a role for GSK3 ␣ / ␤ in PERK signaling in vitro, we then asked if CHOP expression in macrophages within the atherosclerotic plaque could be attenuated by GSK3 ␣ / ␤ inhibition. Five-week-old female ApoE Ϫ / Ϫ mice were placed on an HFD containing 21% milk fat and 0.2% cholesterol for 20 weeks. A subset of mice were given an HFD supplemented with VPA (625 mg VPA/kg body weight), a small molecule shown to inhibit GSK3 ␣ / ␤ both in vitro and in vivo ( 7,30,(34)(35)(36). We have previously reported that ApoE Ϫ / Ϫ mice fed an HFD supplemented with VPA present with attenuated GSK3 ␤ activity in hepatic tissue and within the aortic wall and have signifi cantly reduced atherosclerotic plaque volume (by ‫ف‬ 10%) and necrotic core area (by ‫ف‬ 27%) ( Table 1 ) ( 7 ). Aortic sections were costained with a macrophage-specifi c anti-Mac3 (CD107) antibody and an anti-KDEL antibody specifi c for GRP78 and 94. Serial sections were stained with appropriate preimmune IgG antibodies to control for nonspecifi c staining (supplementary Fig. VI). Representative images ( Fig. 3A ) and quantitation ( Fig. 3C ) show no detectible alterations  To determine if these changes in gene expression affected lipid and cholesterol accumulation within macrophages, of GSK3 ␣ / ␤ activity by CT99021 signifi cantly attenuated ER stress-induced FAS, SREBP-1c, SREBP-2, HMG-CoA, and LDLR expression ( Fig. 4 ). Transcript levels of genes involved in other metabolic pathways were not signifi cantly  ed (A, B). Thp-1 macrophages were treated with Thaps, GLN, or PA, in the presence or absence of CT99021, and then exposed to Alexa fl uor 488-AcLDL (7 g/ml). Representative images of AcLDL uptake are shown (C) and quantifi ed (D). n = 4, * P < 0.05 relative to untreated cells, # P < 0.05 relative to cells of the same treatment without CT99021. Fig. 6. PERK-GSK3 ␣ / ␤ signaling regulates foam cell formation in primary mouse macrophages. Thioglycolate-elicited mouse peritoneal macrophages were isolated from female C57BL6 mice. Macrophages were exposed to 1 M Thaps, 5 mM GLN, or 600 M PA in the presence or absence of PERK inhibitor (3 M) or CT99021 (4 M). After treatment, cells were stained with Oil Red O and DAPI. Representative images are shown (A) and quantifi ed (B, C). Primary mouse macrophages were infected with adenovirus encoding constitutively active GSK3 ␤ (Ad-S9A-GSK3 ␤ ) or an empty vector control (Ad-CMV-Null). ER stress was induced in the cells expressing GSK3 ␤ -S9A as above in the presence or absence of CT99021 or PERK inhibitor. Cells were stained with Oil Red O and DAPI. Representative images are shown (D) and quantifi ed (E, F). Protein lysates from primary macrophages expressing GSK3 ␤ -S9A with or without being exposed to PERK inhibitor were resolved by SDS-PAGE and probed for CHOP and ␤ -actin. Representative blots are shown (G) and quantifi ed (H). CHOP mRNA expression was determined by quantitative RT-PCR in primary macrophages expressing S9A-GSK3 ␤ in the presence or absence of PERK inhibitor (I). n = 4-5, * P < 0.05 relative to untreated cells, # P < 0.05 relative to cells of the same treatment without inhibitor. esterifi ed and unesterifi ed cholesterol levels were quantifi ed. ER stress induction by Thaps, GLN, and PA resulted in signifi cant accumulation of both esterifi ed and unesterifi ed cholesterol ( Fig. 5A, B ). GSK3 ␣ / ␤ inhibition by CT99021, as well as two other GSK3 ␣ / ␤ inhibitors, signifi cantly attenuated the accumulation of free and esterifi ed cholesterol in the ma crophages ( Fig. 5A, B  and supplementary Fig. VIII). Consistent with these fi ndings, Oil Red O staining showed increased lipid droplet formation in the macrophages exposed to ER stress-inducing agents, and siRNA-directed GSK3 ␣ / ␤ knockdown attenuated this effect (supplementary Fig. IV).
The mechanism by which ER stress signaling through GSK3 ␣ / ␤ promotes macrophage lipid accumulation and foam cell formation could involve altered lipoprotein uptake/effl ux and/or altered intracellular lipid biosynthesis. To determine if ER stress signaling through GSK3 ␣ / ␤ plays a role in uptake of modifi ed LDL, Thp-1 macrophages were incubated with acetylated LDL (AcLDL) labeled with Alexa 488. Pretreatment with ER stressinducing agents enhanced AcLDL uptake, and this effect was blocked by GSK3 ␣ / ␤ inhibition ( Fig. 5C, D ). To examine the effect on lipid biosynthesis, macrophages were treated with ER stress-inducing agents and then cultured in the absence of lipoproteins. ER stress enhanced cellular unesterifi ed cholesterol levels, and this effect was attenuated by GSK3 ␣ / ␤ inhibition (supplementary Fig. IX). Neither ER stress nor GSK3 ␣ / ␤ inhibition signifi cantly altered macrophage cholesteryl ester levels. Taken together, these data suggest that ER stress signaling through GSK3 ␣ / ␤ plays a role in lipid uptake and biosynthesis.

PERK-GSK3 ␣ / ␤ signaling regulates ER stress-induced foam cell formation in primary macrophages
To investigate the relevance of PERK-GSK3 ␣ / ␤ signaling in primary macrophages, peritoneal macrophages were isolated from C57Bl6 mice. Mouse macrophages were cultured in the presence or absence of ER stress-inducing agents, Thaps, GLN, or PA. Consistent with our fi ndings in Thp-1 cells, ER stress increased intracellular lipid concentrations, as assessed by Oil Red O staining, relative to untreated controls ( Fig. 6A-C ). Inhibition of PERK or GSK3 ␣ / ␤ was suffi cient to block ER stress-induced lipid accumulation, and cells treated with either of these inhibitors had lipid concentrations similar to those observed in untreated cells ( Fig. 6A-C ).
A constitutively active form of GSK3 ␤ , S9A-GSK3 ␤ , was overexpressed in primary macrophages using an adenovirus vector (Ad-S9A-GSK3 ␤ , MOI of 10) (supplementary Fig. X). As a control, cells were infected with an empty adenovirus (Ad-Null, MOI of 10). S9A-GSK3 ␤ overexpression resulted in signifi cant lipid accumulation, which was not detectibly altered by the presence or absence of ER stress ( Fig. 6D-F ). PERK inhibition, which blocked ER stress-induced lipid accumulation ( Fig. 6B ), did not alter the ability of SA9-GSK3 ␤ to increase macrophage lipid content ( Fig. 6E ). Exposing the cells to CT99021 attenuated S9A-GSK3 ␤ -induced lipid accumulation ( Fig. 6F ). Constitutive GSK3 ␤ activation also resulted in the induction of CHOP protein levels and signifi cantly elevated mRNA expression in primary macrophages, even in the presence of the PERK inhibitor ( Fig. 6G-I ). Together, these data are consistent with our hypothesis that ER stress signals through PERK-GSK3 ␣ / ␤ to induce macrophage lipid accumulation and foam cell formation.

DISCUSSION
Recent evidence suggests a causative role for ER stress in the initiation, development, and progression of atherosclerosis ( 7,13,14,30 ). The mechanistic details of how ER stress induces the proatherogenic process are poorly understood. We present data showing that the PERK branch of the UPR signals through GSK3 ␣ / ␤ to promote macrophage foam cell formation. Our results suggest that GSK3 ␣ / ␤ does not modulate the adaptive components of ER stress signaling including chaperone expression and translation attenuation in human macrophages. Rather, ER stress-induced GSK3 ␣ / ␤ activation plays a role in regulating the distal components of the PERK pathway and is involved in the upregulation of transcription factors including ATF4 and CHOP. Inhibition or knockdown of PERK or GSK3 ␣ / ␤ attenuates ER stress-induced macrophage lipid accumulation. Our results suggest that both lipid uptake and biosynthesis pathways may be affected; however, it is also possible that lipid effl ux from macrophages is also regulated by GSK3 ␣ / ␤ ( 37 ). Constitutive GSK3 ␤ activation is able to overcome the effect of PERK inhibition and promotes lipid accumulation and foam cell formation. Constitutive GSK3 ␤ activation also induces CHOP protein and mRNA expression even when PERK is inhibited. Taken together, these results illuminate a novel signaling mechanism by which ER stress may promote lipid accumulation in macrophage foam cells by activation of the PERK-GSK3 ␣ / ␤ pathway.
Multiple cardiovascular risk factors are capable of inducing ER stress in cell culture and animal models; however, the mechanisms by which individual risk factors promote ER stress are not fully understood. Elevated levels of ER stress have been observed in animal models of hyperglycemia, obesity, and dyslipidemia ( 6,7,30,37,38 ). In a hyperglycemic state, GLN, a metabolite of glucose, accumulates within cells and acts as a potent inducer of ER stress (39)(40)(41). In addition, lipids such as PA and unesterifi ed cholesterol are thought to disturb ER function by disrupting the composition of the ER membrane ( 11,42 ). ER stress/UPR activation may be a central mechanism by which multiple cardiovascular risk factors promote atherosclerosis development.
GSK3 ␣ / ␤ is involved in a number of signaling networks and regulates many aspects of cell metabolism and physiology. The role of GSK3 ␣ / ␤ in the regulation of infl ammatory cytokines is relatively well established. For example, in monocytes and macrophages, induction of NF-B is mediated by GSK3 ␣ / ␤ ( 43 ). Interestingly, GSK3 ␣ / ␤ appears to play a role in the regulation of interleukin (IL)-10 expression, an anti-infl ammatory cytokine. In monocytes and macrophages, stimulation of the phosphatidylinositol 3-OH kinase (PI3K)/AKT pathway results in GSK3 ␤ inhibition, through phosphorylation of serine 9, and results in a signifi cant increase in IL-10 expression ( 44,45 ). Moreover, GSK3 ␣ / ␤ inhibition may also inhibit the expression of the proinfl ammatory cytokines IL-1 ␤ and IL-6 ( 44 ). To the best of our knowledge, we present the fi rst evidence of a role for GSK3 ␣ / ␤ in the regulation of lipid metabolism in macrophages. Together, these fi ndings may indicate that GSK3 ␣ / ␤ has a role in regulating the differentiation between proinfl ammatory, lipid-engorged foam cells and anti-infl ammatory macrophages.
The involvement of ER stress signaling in metabolic disease has been well established, and the PERK signaling branch of the UPR has been the focus of many studies. In both the ApoE and LDLR knockout mice, CHOP defi ciency decreases atherosclerotic plaque size and decreases plaque complexity ( 46 ). ApoE Ϫ / Ϫ CHOP Ϫ / Ϫ and LDLR mice fed a Western diet develop atherosclerotic plaques that are less necrotic and display decreased caspase-3 activation ( 46 ). Similarly, Oyadomari et al. ( 47 ) examined a mouse model in which eIF2 ␣ phosphorylation was impaired by enhanced growth arrest and DNA damage-inducible protein (GADD) 34 expression. These mice had decreased signaling through the downstream components of the PERK pathway and displayed signifi cantly reduced hepatic fat droplet and triglyceride deposition as well as dramatically improved glucose and insulin tolerance ( 46 ). The authors characterized these mice further by observing reduced expression of many genes involved in lipid metabolism including stearoyl-CoA desaturase (SCD)-1 , FAS, and Lpl ( 47 ). Moreover, ATF4 knockout mice have decreased circulating serum triglycerides and FFAs along with decreased white adipose tissue size and reduced SCD1, FAS, and SREBP-1c expression ( 48 ). While our results show that the IRE1 and ATF6 pathways are not directly involved in ER stress signaling through GSK3 ␣ / ␤ , other studies have suggested that the IRE1 and ATF6 do play a role in the regulation of lipid metabolism (49)(50)(51). Although poorly defi ned, cross talk between UPR pathways appears to coordinate the appropriate response to an external stimulus, and thus these pathways may collectively regulate lipid homeostasis ( 52 ). These studies, along with the results presented here, suggest that activation of the PERK branch of the UPR leads to increased lipid synthesis and uptake while attenuation of PERK signaling impairs lipid deposition.
Our investigations defi ne an important role for GSK3 ␣ / ␤ in the regulation of downstream components of the PERK signaling branch of the UPR. Moreover, PERK-GSK3 ␣ / ␤ signaling may play a critical role in the regulation of macrophage lipid accumulation and foam cell formation in vivo. There are a number of important aspects of this pathway that remain to be investigated. For example, it will be important to understand the mechanism by which PERK regulates GSK3 ␣ / ␤ and whether other signaling networks such as the PI3K/AKT, Wnt , or MAPK pathways are involved. Second, we need to identify the direct targets of GSK3 ␣ / ␤ and understand their roles in the regulation of ATF4 and CHOP expression, as well as lipid metabolism. This will be a challenging task because of the number of putative GSK3 ␣ / ␤ substrates that have already been identifi ed ( 53 ). It will also be interesting to determine the specifi c roles of GSK3 ␣ and GSK3 ␤ in this process. The delineation of these pathways may lead to the identifi cation of novel targets for the development of future antiatherogenic therapeutics. Our work presented here provides an initial step toward a clearer understanding of the mechanisms linking ER stress and proatherogenic processes.