Deficiency of TDAG51 Protects Against Atherosclerosis by Modulating Apoptosis, Cholesterol Efflux, and Peroxiredoxin‐1 Expression

Background Apoptosis caused by endoplasmic reticulum (ER) stress contributes to atherothrombosis, the underlying cause of cardiovascular disease (CVD). T‐cell death‐associated gene 51 (TDAG51), a member of the pleckstrin homology‐like domain gene family, is induced by ER stress, causes apoptosis when overexpressed, and is present in lesion‐resident macrophages and endothelial cells. Methods and Results To study the role of TDAG51 in atherosclerosis, male mice deficient in TDAG51 and apolipoprotein E (TDAG51−/−/ApoE−/−) were generated and showed reduced atherosclerotic lesion growth (56±5% reduction at 40 weeks, relative to ApoE−/− controls, P<0.005) and necrosis (41±4% versus 63±8% lesion area in TDAG51−/−/ApoE−/− and ApoE−/−, respectively; P<0.05) without changes in plasma levels of lipids, glucose, and inflammatory cytokines. TDAG51 deficiency caused several phenotypic changes in macrophages and endothelial cells that increase cytoprotection against oxidative and ER stress, enhance PPARγ‐dependent reverse cholesterol transport, and upregulate peroxiredoxin‐1 (Prdx‐1), an antioxidant enzyme with antiatherogenic properties (1.8±0.1‐fold increase in Prdx‐1 protein expression, relative to control macrophages; P<0.005). Two independent case–control studies found that a genetic variant in the human TDAG51 gene region (rs2367446) is associated with CVD (OR, 1.15; 95% CI, 1.07 to 1.24; P=0.0003). Conclusions These findings provide evidence that TDAG51 affects specific cellular pathways known to reduce atherogenesis, suggesting that modulation of TDAG51 expression or its activity may have therapeutic benefit for the treatment of CVD.

C ardiovascular disease (CVD) is an acute clinical manifestation of atherothrombosis that accounts for the majority of deaths in North America. 1 A number of risk factors are known to accelerate CVD, including hypercholesterolemia, smoking, diabetes, hypertension, hyperhomocysteinemia, and obesity. Despite the diversity of these risk factors, the development and progression of atherosclerotic lesions is remarkably similar. Endothelial cell dysfunction and the accumulation of cholesterol-rich lipoproteins in the vessel wall are early events in atherogenesis, resulting in the recruitment of circulating monocytes, their adhesion to the endothelium, their subsequent differentiation into macrophages, and the accumulation of lipid to form foam cells. 2 In humans, these fatty streaks can progress to more advanced lesions characterized by a lipid-rich necrotic core and a fibrous cap consisting of smooth muscle cells and collagen.
The acute clinical manifestations of atherosclerosis result from plaque rupture, thrombus formation, and vessel occlusion. 3 Apoptotic cell death is a key feature of unstable plaques 4 and is induced by a number of cellular stress pathways, including oxidative and endoplasmic reticulum (ER) stress. [5][6][7] The distribution of cell death is heterogeneous within advanced lesions, but is most prominent in the lipid-rich necrotic core that contains a high density of macrophages. Apoptotic cell death increases the risk of plaque rupture by decreasing the number of viable smooth muscle cells necessary for collagen production and compromising the structural integrity of the fibrous cap following release of matrix metalloproteinases from dead macrophages. 8 Furthermore, plaque thrombogenicity is enhanced because lesion-resident cells undergoing apoptosis express active cell surface tissue factor (TF), 9 the major physiological initiator of the coagulation cascade. 10 Previous studies have demonstrated that the absence of specific proapoptotic factors such as Bax 11 or Rb 12 decreases macrophage apoptosis as well as necrotic core size in hyperlipidemic mice. Consistent with these findings, a reduction in apoptosis and plaque necrosis was observed in advanced atherosclerotic lesions from ApoE À/À mice deficient in the ER stress effector CHOP. 5 TDAG51 is a member of the pleckstrin homology-like domain family having proapoptotic characteristics. 13 Furthermore, TDAG51 is induced by ER stress, [14][15][16] and its overexpression in human vascular endothelial cells induces apoptotic cell death by disrupting cytoskeletal structure and impairing cell adhesion. 15 Conversely, deficiency of TDAG51 contributes to apoptosis resistance and growth dysregulation in metastatic melanomas in vivo. 17 TDAG51 can also regulate energy metabolism by modulating adipogenesis and hepatic lipogenesis, which correlates with mature-onset metabolic disease. 18 Several lines of evidence implicate TDAG51 in atherosclerotic lesion development. TDAG51 expression is increased in lesionresident macrophages and endothelial cells during all stages of atherogenesis. 6,15 Furthermore, TDAG51 mRNA is significantly increased in cultured human vascular endothelial cells following exposure to athero-prone waveform stimulation. 19 Although these findings suggest that TDAG51 contributes to the atherosclerotic process, it is currently unknown if TDAG51 is causally related to atherogenesis or if its mechanism of action stems from its previously described role as a proapoptotic factor.
In this report, we investigated whether loss of TDAG51 alters the development and progression of atherosclerosis by crossing TDAG51 À/À mice 20 with ApoE À/À mice, an established hyperlipidemic mouse model of accelerated atherosclerosis. 21 Our findings provide the first in vivo evidence that deficiency of TDAG51 reduces atherosclerotic lesion growth. Furthermore, such inhibition of atherogenesis because of TDAG51 deficiency likely involves the action of PPARc on specific cellular targets and pathways that are known to affect atherosclerotic lesion development and progression.

Mice and Dietary Conditions
ApoE À/À mice were obtained from the Jackson Laboratory (Bar Harbor, ME). TDAG51-deficient (TDAG51 À/À ) mice have been previously described. 20 TDAG51 À/À (knockout [KO]) mice were backcrossed >9 generations onto a C57BL/6 background. TDAG51-deficient mice were crossbred with TDAG51 +/+ /ApoE À/À mice (also on a C57BL/6 background) to generate TDAG51 À/À /ApoE À/À double-knockout (dKO) mice as well as ApoE À/À littermate controls. Given previous studies showing that the PPARc ligand effect on atherosclerosis is sex specific toward male mice, 22 only male mice were used in this study. Mice were housed with free access to regular chow diet. All experimental procedures using mice were approved by the McMaster University Animal Research Ethics Board.

Measurement of Total Cholesterol, Triglyceride, and Glucose
Plasma total cholesterol was measured using an Infinity cholesterol measurement kit (Thermo Electron Corporation, Melbourne, Australia). Triglyceride and glucose levels in mice plasma were determined similarly to the above described cholesterol assay (Thermo Electron Corporation).
Macrophages were pretreated in the presence or absence of 10 lmol/L GW9662 in FBS-deficient media for 4 hour before incubation with acetyl-LDL (50 lg/mL) or acetyl-LDL+GW9662 (10 lmol/L) for 24 to 48 hours. Cellular lipids were isolated via Bligh and Dyer chloroform:methanol extraction 25 with subsequent assessment of total and free cholesterol levels using Cholesterol E and Free Cholesterol E kits (Wako Pure Chemical Industries, Ltd, Osaka, Japan). Lipid content was normalized against cellular protein and the data expressed as fold-change.

Measurement of Cholesterol Efflux in Peritoneal Macrophages
HDL/APOA1-dependent cholesterol efflux to the medium was determined as described previously. 26 Peritoneal macrophages isolated from TDAG51 À/À and wild-type C57BL/6 mice were plated at a density of 5910 5 cells/well and loaded with 1 lCi/mL [ 3 H]cholesterol (PerkinElmer Life Sciences) in RPMI-1640 media containing 5% LPDS for 48 hours. To equilibrate cholesterol pools, cells were washed twice in media containing 2% fatty acid-free BSA and cultured overnight in the same media. Media were removed, and cells were incubated for 1 to 5 hours in media containing 0.2% BSA in the absence or presence of 40 lg/mL HDL. Following incubation, radioactivity of culture supernatants and cell lysates was measured by liquid scintillation. Results were normalized to total cellular protein content and expressed as the percentage of radioactivity in the medium divided by the total radioactivity in the cells and medium.

Detection of Superoxide
Superoxide levels in cells were measured using fluorescentdye dihydroethidium (Invitrogen, Carlsbad, CA) as previously described. 16 Fluorescence intensity is reported as relative fluorescence units (RFU).

Measurement of Mouse Plasma Lipid Profiles Using FPLC
Mouse plasma was fractionated into lipid components using gel filtration-fast protein liquid chromatography (FPLC), as described previously. 27

Tissue Sample Preparation
Following perfusion-fixation with 10% neutral buffered formalin, 6,28 hearts (including the aortic roots) were cut transversely and embedded in paraffin. Serial sections, 4 lm thick, were cut starting from the aortic root origin and collected for measurement of lesion size (hematoxylin/eosin staining) and immunohistochemical analyses. 29 In each mouse, the atherosclerotic lesion area was measured in 5 sections separated by 80 lm (ie, within 320 lm from the aortic valve). 29 The lesion was traced manually and measured using computer-assisted image analysis equipment (Olympus BX41 microscope, Olympus DP70 CCD camera, and ImagePro Plus software). Lesion size is expressed as the mean of 5 sections. Therefore, this number is directly proportional to the volume of the lesions in the first 320 lm of the ascending aorta.

Immunohistochemical Analysis
Immunohistochemical staining of atherosclerotic lesions was performed as described previously. 6,29 Sections were counterstained with hematoxylin. Human carotid arteries were obtained at the time of endarterectomy from consenting patients. The protocol was approved by the institutional ethics review boards of Hamilton Health Sciences and St. Joseph's Healthcare. Tissue was fixed with formalin and embedded in paraffin. Double immunofluorescence was performed on the sections as described previously. 29 Antibodies used for immunostaining were anti-GRP78 (sc-1050; Santa Cruz, CA), anti- Immunostaining was quantified by extracting the blue component of the RGB image, thresholding for the staining, and measuring the stained area as well as the total area of the lesion (by manual tracing) using ImagePro 6.3 software. Results are expressed as a percentage of the total lesion area.

Indirect Immunofluorescence of Endogenous TDAG51 Expression in Lesion-Resident Macrophages
Six-week-old TDAG51 +/+ /ApoE À/À mice were placed on control chow diet for 25 weeks. Mice were euthanized, and hearts containing the aortic roots were removed, embedded in paraffin, sectioned, and immunostained. Primary antibodies were detected using either Alexa 488 or Alexa 594 donkey anti-goat IgG (Molecular Probes). A Carl Zeiss LSM510 laser-scanning confocal microscope was used to examine endogenous TDAG51 localization.

Collagen Content in Atherosclerotic Lesions From Aortic Roots
Six-week-old TDAG51 +/+ /ApoE À/À or TDAG51 À/À /ApoE À/À mice were placed on control chow diet for 25 weeks. Hearts containing aortic roots were removed, embedded in paraffin, serially sectioned, and stained with Masson's trichrome. For quantification of collagen content, 1 section per mouse, close to the aortic root origin, was assessed. The amount of collagen was quantified by thresholding for the blue color using Image Pro software, and collagen area is expressed as a percentage of the total lesion area in that section.

En Face Oil Red O Staining of Mouse Aortas
Unopened aortas were stained with Oil Red O. Aortas were then opened longitudinally, and the stained area was measured as percentage of the total aorta.

Case-Control Genetic Association Studies
The association between CVD and single-nucleotide polymorphisms (SNPs) in a 56-kbp region (chromosome 12, positions 74 680 495 to 74 736 823) that included the entire TDAG51 gene (chromosome 12, positions 74 705 495 to 74 711 823) and 25 kbp upstream and downstream of the gene was examined in 2 case-control studies. We evaluated results from the CVD case-control study published by the Wellcome Trust Case-Control Consortium (WTCCC). 30 The WTCCC study included 2000 cases with documented CVD and 3000 controls; all cases and controls were white. The University of California, San Francisco (UCSF) case-control study included 731 myocardial infarction patients and 797 healthy controls, collected by investigators at the UCSF Genomic Resource in Arteriosclerosis. 31 The number of cases and controls who were successfully genotyped is reported for each genotype group in Tables 3 and 4.

Statistical Analysis
Experimental values are presented as meanAEstandard error (SE). Unless otherwise noted, statistical comparisons for all experiments were performed using Mann-Whitney (for 2 groups) and Kruskal-Wallis (for >2 groups) tests. Statistical comparisons for en face Oil Red O staining of mouse aortas and Oil Red O staining of peritoneal macrophages were performed using the unpaired Student t test and ANOVA. P<0.05 was considered statistically significant for all tests.
The association between SNPs and myocardial infarction was performed by logistic regression analysis that adjusted for age and sex. A combined analysis of the results from the UCSF study and the WTCCC study was carried out using the fixed-effects Mantel-Haenszel method that combined odds ratios across these studies; homogeneity of the odds ratios was assessed by the Breslow-Day test.

Loss of TDAG51 Reduces Atherosclerotic Lesion Development and Progression
To investigate the role of TDAG51 in atherosclerotic lesion development, we generated TDAG51 À/À /ApoE À/À doubleknockout male mice (dKO), as well as TDAG51 +/+ /ApoE À/À littermate/sex controls ( Figure 1). To focus on the effect of TDAG51 deficiency, male mice were maintained on a control chow diet, as opposed to a high-fat chow diet, to minimize the potential contribution of obesity and insulin resistance 32 to atherosclerotic lesion development and necrosis.
Total plasma lipids, glucose, and inflammatory cytokines, as well as aortic atherosclerotic lesion size and composition, were analyzed at 25 and 40 weeks of age. No significant differences in total plasma cholesterol or triglycerides were observed between dKO and ApoE À/À mice (Figure 2A and 2B). Consistent with these findings, plasma lipid profiles were indistinguishable between these groups at 25 weeks of age ( Figure 2C). No significant changes in body weight (data not shown), plasma glucose level ( Figure 2D), or morphology of pancreatic b cells from the islets of Langerhans ( Figure 3) were observed in dKO and ApoE À/À mice.
Immunohistochemical analysis revealed that dKO and ApoE À/À mice developed fatty streaks and mature atherosclerotic lesions, consisting of macrophages and smooth  Figure 1. Identification of TDAG51 À/À /ApoE À/À mice. Mouse genomic DNA was amplified by PCR using specific primer sets, and the PCR products were separated on agarose gels to identify mice containing the wild-type TDAG51 allele (wt-TDAG51) (A) or the disrupted TDAG51 allele (ko-TDAG51) (B). PCR amplification and gel electrophoresis was also used to identify mice that were deleted for the ApoE allele (C). Neg, negative controls for PCR reactions (A through C). Liver tissue obtained from PCR identified TDAG51 À/À /ApoE À/À dKO or control TDAG51 +/+ /ApoE À/À mice was homogenized, and total protein lysates were examined by immunoblotting using an anti-TDAG51 antibody (D). b-actin was used as a loading control. TDAG51 indicates T-cell death-associated gene 51; PCR, polymerase chain reaction; ApoE, apolipoprotein E; ko, knockout; dKO, double knockout. muscle cells, in the atherosclerotic cap region ( Figure 6). However, at 25 and 40 weeks of age, there were no significant differences in the content of either macrophages or smooth muscle cells in the lesions of dKO and ApoE À/À mice ( Figure 6 and Table 1). In support of the en face lipid staining ( Figure 5C) and increased necrotic core size ( Figure 5D and 5E), cholesterol crystals were prevalent in the necrotic core regions of the 25-week-old ApoE À/À mice ( Figure 5D). Collagen content in the lesions, as measured by Masson's trichrome staining, showed high intralesion variability and did not differ significantly between the ApoE À/À and dKO groups (23.4% versus 27.9%, P=0.48; Figure 7). However, at 40 weeks dKO mice exhibited less vascular calcification, as assessed by von Kossa staining, compared with ApoE À/À mice ( Figure 8). Inflammatory cytokines are known to influence several cellular processes that accelerate atherosclerotic lesion growth and stability. 33 However, loss of TDAG51 had no significant effect on plasma levels of proinflammatory cytokines, including IL-12, TNF-a, or MCP-1, up to 40 weeks (Table 2). These data suggest the decrease islets of Langerhans. Representative sections from 5 mice per group of pancreatic tissue from 25-week-old TDAG51 +/+ /ApoE À/À (ApoE À/À ) or TDAG51 À/À /ApoE À/À (dKO) mice immunostained for insulin (red) or glucagon (green). Scale bars, 50 lm. TDAG51 indicates T-cell deathassociated gene 51; ApoE, apolipoprotein E; dKO, double knockout. in lesion area was primarily a result of reduction in necrotic core size.

TDAG51 Deficiency Is Cytoprotective Against ER Stress and Oxidative Stress
TDAG51 expression is increased by ER stress and causes detachment-induced apoptotic cell death when overexpressed. 15,34 Recent studies have reported that reducing ER stress by treatment with the small chemical chaperone 4-PBA 35 or by genetic ablation of CHOP 5 attenuates apoptosis, plaque necrosis, and atherogenesis in ApoE À/À mice. To investigate whether the absence of TDAG51 decreases apoptosis in vivo, atherosclerotic lesions were stained for TUNEL and activated caspase-3. 6 TUNEL staining as well as activated caspase-3 staining was reduced in the necrotic core of advanced lesions from dKO mice, compared with ApoE À/À mice ( Figure 9A). In support of these findings, we observed that TDAG51 À/À peritoneal macrophages ( Figure 9B) were resistant to cell death induced by ER stress and oxidative stress.

TDAG51 Negatively Regulates the Expression of PPARc and Its Downstream Targets
To provide further mechanistic insight into the antiatherogenic effects of TDAG51 deficiency, we investigated the expression of PPARc because of its involvement in modulating various pathways contributing to atherogenesis, such as inflammation, lipid metabolism, and oxidative stress. Furthermore, we have previously observed an inverse correlation between TDAG51 and PPARc expression in 3T3-L1 cells. 18 Prior studies have demonstrated the expression of PPARc in lesion-resident macrophages from LDLR À/À mice. 36 To determine whether deficiency of TDAG51 alters the expression of PPARc in lesional macrophages, paraffin sections from the aortic roots of chow-fed dKO and ApoE À/À mice after 15 or 25 weeks were immunostained for PPARc ( Figure 10A). Intense nuclear staining for PPARc was observed in lesionresident macrophages from dKO mice compared with ApoE À/À mice. As a positive control, nuclear staining for PPARc was observed in adipocytes from dKO and ApoE À/À mice ( Figure 10B). Again, the intensity of nuclear staining for PPARc was increased in adipocytes from dKO mice as well as in cultured TDAG51 À/À macrophages ( Figure 10C). Furthermore, TDAG51 À/À peritoneal macrophages exhibited increased mRNA expression of PPARc (2.8AE0.5-fold, P<0.05) and its target gene, LXRa (1.9AE0.3-fold, P<0.05), as well as PPARc-inducible gene ABCG1 (1.7AE0.1-fold, P<0.05), but not ABCA1, compared with wild-type macrophages ( Figure 11).
Consistent with these findings, reintroduction of TDAG51 using a retrovirus expression construct caused a significant reduction in PPARc mRNA expression (P<0.001) in TDAG51 À/À mouse embryonic fibroblasts (data not shown). Furthermore, siRNA knockdown of TDAG51 in HeLa cells also showed enhanced PPARc expression (data not shown).

TDAG51 Deficiency Enhances PPARc-Dependent Inhibition of Inflammatory Cytokine Expression in Macrophages
Inflammation has been shown to modulate several processes that contribute to atherogenesis and lesion stability. 33 TDAG51 À/À peritoneal macrophages exhibited no basal differences in the expression of either MCP-1 or TNFa relative to controls, as assessed by RT-PCR ( Figure 12A). However, in the presence of the PPARc agonist rosiglitazone, TDAG51 À/À macrophages expressed significantly lower levels of both MCP-1 and TNFa, relative to wild-type macrophages ( Figure  12A). Furthermore, TDAG51 deficiency caused a significant reduction in LPS-induced expression of MCP-1 (22.4AE0.8-fold versus 31.6AE1.5-fold, P<0.05), but not TNF-a, compared with wild-type macrophages ( Figure 12B). Human THP-1-derived A B C Figure 4. Expression of TDAG51 in atherosclerotic lesions. Six-week-old TDAG51 +/+ /ApoE À/À and TDAG51 À/À /ApoE À/À mice were fed control chow diet for 25 weeks. Mice were euthanized, and hearts containing aortic roots were removed, embedded in paraffin, sectioned, and immunostained for TDAG51. Indirect immunofluorescence detection of TDAG51 in lesions (arrows) from aortic roots of TDAG51 +/+ /ApoE À/À mice (A and B). B, Magnified region from A. Representative images from 5 mice are shown. To assess nonspecific immunofluorescence, aortic root sections from TDAG51 À/À /ApoE À/À mice were immunostained using an anti-TDAG51 antibody (C). Scale bars: 100 lm in A and C; 50 lm in B.  Figure 5. Effect of TDAG51 deficiency on atherosclerotic lesions in ApoE À/À mice. TDAG51 À/À /ApoE À/À mice (dKO) and ApoE À/À control mice were fed standard chow diets for 25 or 40 weeks (n=8 to 9 per group). A and B, Aortic root sections were stained with hematoxylin/eosin, and mean atherosclerotic lesion size was determined. Significant reduction (*P<0.05) in lesion size was observed at both 25 and 40 weeks in dKO compared with ApoE À/À control groups. Black line demarcates lesion area. Scale bar=500 lm. C, Representative en face Oil Red O (ORO)stained aortas and quantitative assessment showed significant reduction (*P=0.0016) in lipid deposition in aortas from 40-week-old dKO mice compared with the control group. For quantitative data, results from 3 independent experiments are shown. D, Representative images are shown of necrotic core sizes in aortic root lesions of 25-and 40-week-old dKO and ApoE À/À mice (n=5). Red line demarcates necrotic core area. Scale bar=100 lm. E, Necrotic core sizes of 25-and 40-week-old dKO mice were smaller in aortic root lesions compared with their respective ApoE À/À control groups (n=5). Data are shown as mean necrotic core areaAESE (*P<0.005). TDAG51 indicates T-cell death-associated gene 51; ApoE, apolipoprotein E; dKO, double knockout. macrophages, but not THP-1 monocytes, as well as aortic endothelial cells (HAECs) and aortic smooth muscle cells (HASMCs), expressed TDAG51 protein ( Figure 12C). Consistent with these findings, TDAG51 was also expressed in macrophages (CD68), smooth muscle cells (SMA), and endothelial cells (vWF) in lesions from human carotid arteries, as determined by dual immunohistochemistry ( Figure 12D).

Loss of TDAG51 Increases Cholesterol Efflux in Cultured Peritoneal Macrophages
Given that PPARc increases reverse cholesterol transport in lesion-resident macrophages and is considered atheroprotective, 37,38 we examined the effects of TDAG51 deficiency on macrophage foam cell formation and cholesterol efflux. To assess macrophage foam cell formation, peritoneal macrophages were cultured in the absence or presence of acetylated low-density lipoprotein (LDL), and the accumulation of intracellular lipids was examined by ORO staining ( Figure 13A). TDAG51 À/À macrophages showed a significant reduction in ORO staining at 48 hours compared with wildtype macrophages ( Figure 13B). TDAG51 À/À macrophages also accumulated significantly less cellular total cholesterol (1.8AE0.1-fold versus 3.1AE0.1-fold, P<0.05; Figure 13C) and free cholesterol (0.9AE0.1-fold versus 1.5AE0.1-fold, P<0.05; Figure 13D) after incubation with acetyl-LDL (50 lg/mL) for 48 hours, relative to wild-type macrophages. Consistent with the hypothesis that increased PPARc results in reduced lipid accumulation, the PPARc antagonist GW9662 (10 lmol/L) increased cellular total cholesterol in both wild-type and KO macrophages after 48 hours incubation with 50 lg/mL acetyl-LDL, compared with acetyl-LDL treatment alone (Figure 13E). To determine whether the reduced lipid accumulation with TDAG51 deficiency was the result of increased cholesterol efflux, macrophages were loaded with [ 3 H]cholesterol, and the percent change in intracellular radiolabeled cholesterol in the presence or absence of high-density lipoprotein (HDL) was measured ( Figure 13F). Cholesterol efflux was significantly increased in TDAG51 À/À peritoneal macrophages 2, 3, and 5 hours after the addition of HDL, compared with WT macrophages ( Figure 13F). PPARc  Lesions from 25-or 40-week-old TDAG51 À/À /ApoE À/À (dKO) and ApoE À/À mice were sectioned and probed for Mac-3 (macrophages) and SMA (smooth muscle actin). Macrophage or smooth muscle cell area was measured, and data are expressed as percentage of lesion area. TDAG51 indicates T-cell death-associated gene 51; ApoE, apolipoprotein E; dKO, double knockout; n, mouse number; NS, not significant.
agonists induce the expression of ABCG1, a protein transporter mediating cholesterol efflux from macrophages. 39 Consistent with these results, ABCG1 mRNA expression, but not ABCA1 expression, was significantly increased in TDAG51 À/À peritoneal macrophages compared with wild-type macrophages ( Figure 11). Taken together, these results suggest the cumulative effect of increased cholesterol efflux over time leads to significantly lower intracellular total and free cholesterol levels in TDAG51 À/À macrophages.

Single-Nucleotide Polymorphisms in the Human TDAG51 Gene Are Associated With CVD
The WTCCC case-control study of CVD published data for 5 single-nucleotide polymorphisms (SNPs) in the 56-kbp region that included the TDAG51 gene. 30 Two of these SNPs (rs10880022 and rs2367446) were associated with CVD in this study (Table 3); both remained significant after Bonferroni correction for testing 5 SNPs. We genotyped rs10880022 and rs2367446 in the UCSF study 31

Discussion
Several lines of evidence suggest that TDAG51 modulates lesion progression and plaque stability. First, TDAG51 is expressed in athero-prone vascular endothelial cells 19 as well as lesion-resident macrophages and endothelial cells from ApoE À/À mice with diet-induced hyperhomocysteinemia. 6,15 Second, TDAG51 is expressed in apoptotic cells within the  Figure 9. TDAG51 deficiency reduces cell death in atherosclerotic lesions and peritoneal macrophages. TDAG51 À/À /ApoE À/À (dKO) or ApoE À/À mice were placed on control chow diet for 25 weeks. A, Representative images of atherosclerotic lesions from 5 mice per group were stained for TUNEL and cleaved caspase-3. Microscope magnification 920. B, Peritoneal macrophages isolated from wildtype C57BL/6 (C57) mice or TDAG51 À/À mice were treated with 2.5 lg/mL tunicamycin (Tm), 100 nmol/L thapsigargin (Tg), or 10 lmol/L 7-ketocholesterol ( lipid-rich necrotic core. 6,15 Third, overexpression of TDAG51 in cultured human vascular endothelial cells leads to detachment-induced apoptosis. 15,34 Additional observations of increased TDAG51 expression during all stages of atherogenesis in the absence of HHcy 6 suggest that other pathophysiological conditions that induce ER stress and/or TDAG51 expression modulate lesion growth and stability. In support of this hypothesis, we have reported that peroxynitrite, a proatherogenic agent generated from nitric oxide and superoxide, induces ER stress, TDAG51 expression, and apoptosis in cultured vascular endothelial cells. 16 Given that ER stress plays a major role in lesion progression and plaque stability [5][6][7]42,43 and that TDAG51 is an ER stress-inducible gene 6,14-16 expressed in lesion resident macrophages and endothelial cells, we sought to investigate the causal role of TDAG51 in atherogenesis. In this study, TDAG51 À/À /ApoE À/À dKO male mice showed significant reductions in lesion and necrotic lipid core sizes in aortic roots at 25 and 40 weeks compared with age-matched ApoE À/À controls. Thus, in the setting of a normal chow diet, ApoE À/À mice lacking TDAG51 exhibited reduced growth of atherosclerotic lesions. As TDAG51 is expressed in lesion-resident cells undergoing apoptotic cell death, 6,15 a potential mechanism through which TDAG51 deficiency contributes to the reduced atherosclerotic lesion size and necrosis observed in this study is decreased lesional apoptosis. Decreased lesional necrosis from dKO mice was associated with reduced apoptosis, as determined by immunohistochemical staining for TUNEL and cleaved caspase-3. Our data are consistent with the hypothesis that retardation of the rate of necrotic core and lesion growth is in  Figure 10. TDAG51 deficiency increases PPARc expression and nuclear localization in lesion-resident macrophages. A, TDAG51 À/À /ApoE À/À (dKO) or ApoE À/À mice were placed on control chow diet for 15 or 25 weeks. Atherosclerotic lesions from the aortic roots were sectioned and immunostained for PPARc. Arrows indicate PPARc-positive staining macrophages. Representative images from 5 mice per group are shown. Scale bar=50 lm. B, Identification of PPARc in adipose tissue from TDAG51 +/+ /ApoE À/À (ApoE À/À ) and TDAG51 À/À /ApoE À/À (dKO) mice fed chow diet for 40 weeks. Fat pads were removed, embedded in paraffin, sectioned, and immunostained for PPARc. Arrows indicate positive nuclear immunostaining for PPARc. Consistent with lesion-resident TDAG51 À/À macrophages, intensity of nuclear PPARc staining was increased in TDAG51 À/À adipocytes. Representative images from 5 mice per group are shown. Scale bar=100 lm. C, Optical sections 0.8 lm in thickness were obtained through TDAG51 À/À and C57BL/6 (C57) peritoneal macrophages at the plane of the nuclei (arrows). In C57 macrophages, little PPARc (green) was visualized within the nucleus, whereas in TDAG51 À/À macrophages, PPARc (green) was found to colocalize (merged green and red producing yellow) with nucleic acids (red) as shown by ethidium bromide (EtBr) staining in the nuclei (arrows). Representative images from 3 independent experiments are shown. Scale bar=10 lm. TDAG51 indicates T-cell death-associated gene 51; ApoE, apolipoprotein E; dKO, double knockout; PPARc, peroxisome proliferator-activated receptor c.  Figure 11. Detection of PPARc and its target genes in peritoneal macrophages. Total RNA was isolated from male TDAG51 À/À or C57BL/6 (C57) peritoneal macrophages and mRNA expression assessed by qRT-PCR. Data were normalized to 18s, and fold-change in expression relative to C57 was determined. Results from 5 independent experiments are shown as meanAESE. *P<0.05, relative to C57 controls. PPARc indicates peroxisome proliferator-activated receptor c; TDAG51, T-cell death-associated gene 51; qRT-PCR, quantitative real-time polymerase chain reaction.
part a consequence of increased resistance to macrophage cell death resulting from TDAG51 deficiency. Consistent with this hypothesis, peritoneal macrophages isolated from TDAG51 À/À mice were resistant to ER and oxidative stress-induced cell death. Cytoprotection was associated with increased levels of Prdx-1 in lesions of dKO mice as well as in TDAG51 À/À macrophages and endothelial cells. This was accompanied by reduced superoxide levels in TDAG51 À/À macrophages at both baseline and when exposed to oxidative stress, a finding consistent with previous studies. 44,45 These data suggest that the elevation of Prdx-1 levels associated with TDAG51 deficiency contributes to increased resistance to oxidative stress-induced apoptosis, thereby reducing atherosclerotic lesion progression and rupture. This is consistent with the finding that Prdx-1 deficiency in ApoE À/À mice causes endothelial activation (increased leukocyte rolling, endothelial P-selectin, and vWF expression) and accelerates atherosclerosis. 41 Further studies should clarify whether TDAG51-mediated endothelial dysfunction contributes to atherogenesis. It has been previously reported that inhibiting macrophage apoptosis increases atherosclerotic lesion size. Macrophagespecific deletion of the proapoptotic gene p53 resulted in significantly increased lesion area after 15 or 20 weeks on a high-fat diet in LDLR À/À mice. 46  attributed the larger lesion size to increased cell proliferation rather than decreased apoptosis, as no significant differences in apoptosis were observed. Liu et al 11 demonstrated that macrophage-specific deletion of Bax, a proapoptotic protein, decreased macrophage apoptosis and was associated with increased lesion area in LDLR À/À mice after 10 weeks on a Western diet. Alternatively, deletion of the macrophage apoptosis inhibitory factor (AIM) was associated with increased apoptosis and smaller lesion area in LDLR À/À mice after 5 or 12 weeks on a Western diet. 47 Given the variability in genetic mouse models, diets used, and stage of atherosclerosis examined, a direct comparison to the current study is not possible. However, in the current study, in which lesions were assessed in dKO mice on a chow diet at later times (25 and 40 weeks) than the above studies (5 to 20 weeks on high-fat or Western diets), data are consistent with the hypothesis that elevated macrophage apoptosis leads to reduced atherosclerosis at early stages and increased atherosclerosis at later stages of lesion growth. Furthermore, given that the model used in our study was global ablation of the TDAG51 gene, the observed reduction in lesion size may not be solely attributable to decreased macrophage apoptosis; other cell types and mechanisms may be involved. Previous studies have indicated that PPARc acts as a positive regulator of antioxidant defenses. 48 Consistent with the inverse correlation between TDAG51 deficiency and PPARc expression, rosiglitazone treatment induced Prdx-1 protein expression by 23% (P<0.05) in TDAG51 À/À macrophages, compared with 12% (P<0.05) in wild-type macrophages ( Figure 14F). These findings are consistent with previous studies showing that 15d-PGJ 2 , a natural PPARc agonist, induces expression of antioxidant proteins including Prdx-1 49 and supports the hypothesis that the increased Prdx-1 expression associated with TDAG51 deficiency is a consequence of elevated expression and/or activity of PPARc, leading to the reduced atherosclerotic lesion growth observed in the dKO mice. Given the ability of TDAG51 to modulate transcriptional activity, 13 we are currently investigating the possibility that TDAG51 modulates PPARc transcriptional activity and/or nuclear localization.
We recently reported that TDAG51 deficiency induced ageassociated adipogenesis and hepatic steatosis in TDAG51 À/À mice, 18 further implicating PPARc in modulating the effects of TDAG51 deficiency. Taken together, our results suggest that TDAG51 deficiency, despite its effects on adipogenesis and hepatic lipogenesis, reduces atherosclerotic lesion growth via activation of multiple cellular pathways that regulate apoptosis, antioxidant status, and lipid storage/export.
Although studies reporting the beneficial role of PPARc agonists in mitigating atherogenesis have been fairly consistent in mouse models, some controversy exists over the cardiovascular risk associated with the use of thiazolidinediones (TZDs) in patients. Several reports have indicated rosiglitazone is associated with significantly increased risk of myocardial infarction. 53,54 In contrast, others have found the evidence for an association between rosiglitazone and myocardial infarction and CVD mortality to be inconclusive. 55 Furthermore, pioglitazone, another PPARc agonist, was reported to have significantly reduced cardiovascular risk in the PROactive study in diabetic patients with preexisting CVD. 56 Given the current debate over the potential beneficial and adverse events associated with TZDs in regard to CVD risk, it remains to be determined whether increased PPARc associated with TDAG51 deficiency and decreased atherosclerotic lesion growth in a mouse model can be extrapolated to a clinical setting. It should also be noted that this mechanism is likely not the only one that drives the   Figure 16. TDAG51 deficiency protects vascular smooth muscle cells (VSMCs) against endoplasmic reticulum (ER) and oxidative stress. VSMCs isolated from C57BL/6 (C57), TDAG51 +/À , and TDAG51 À/À mice were incubated in the presence or absence of 1 lg/mL tunicamycin (Tm), 200 nmol/L thapsigargin (Tg), or 1 lmol/L staurosporine for 48 hours. Cytotoxicity was determined by measuring LDH release. Results from 4 independent experiments are shown as meanAESE. *P<0.05 relative to respective genotype controls;^P<0.05 compared with respective C57 group; # P<0.05 compared with respective TDAG51 +/À group. TDAG51 indicates T-cell death-associated gene 51; LDH, lactate dehydrogenase.
antiatherogenic effect of TDAG51 deficiency. This could reflect TDAG51's role in apoptosis or signaling pathways that modulate macrophage viability and lipid metabolism. The genetic association results from 2 independent casecontrol studies 30,31 suggest that genetic variants in the TDAG51 region are associated with CVD. This finding is intriguing given the biological evidence for this gene in atherosclerosis. However, meta-analyses of genomewide association studies 57,58 did not find an association between SNPs in the TDAG51 region and CVD at the genomewide significance level (P<5910 À8 ). Therefore, the genetic association results we report here may have overestimated the risk associated with the SNPs we examined. Additional studies are thus required to further evaluate whether genetic variations in the TDAG51 gene are associated with altered expression of this gene and CVD. Functional characterization of these SNPs will also be helpful to understand the potential role of these SNPs in CVD. Because rs2367446 is in an intergenic region 20 kbp from TDAG51, it is possible that this SNP affects transcription of the TDAG51 gene.
In summary, we have characterized a previously unknown role for TDAG51 in modulating atherosclerotic lesion development and progression while addressing the underlying cellular mechanisms by which TDAG51 deficiency modulates these processes. Because the effects of TDAG51 are mediated by multiple pathways that affect lesion development and stability, our findings provide a unique opportunity to develop novel therapeutic approaches that decrease the risk of CVD by targeting TDAG51 expression and/or activity.