Epigenetic upregulation of p66shc mediates low-density lipoprotein cholesterol-induced endothelial cell dysfunction

Abstract

Hypercholesterolemia characterized by elevation of low-density lipoprotein (LDL) cholesterol is a major risk factor for atherosclerotic vascular disease. p66shc mediates hypercholesterolemia-induced endothelial dysfunction and atheromatous plaque formation. We asked if LDL upregulates endothelial p66shc via changes in the epigenome and examined the role of p66shc in LDL-stimulated endothelial cell dysfunction. Human LDL stimulates human p66shc promoter activity and p66shc expression in human endothelial cells. LDL leads to hypomethylation of two CpG dinucleotides and acetylation of histone 3 in the human p66shc promoter. These two CpG dinucleotides mediate LDL-stimulated p66shc promoter activity. Inhibition or knock down of DNA methyltransferases negates LDL-induced endothelial p66shc expression. p66shc mediates LDL-stimulated increase in expression of endothelial intercellular adhesion molecule-1 (ICAM1) and decrease in expression of thrombomodulin (TM). Mirroring these changes in ICAM1 and TM expression, p66shc mediates LDL-stimulated adhesion of monocytes to endothelial cells and plasma coagulation on endothelial cells. These findings indicate that LDL cholesterol upregulates human endothelial p66shc expression via hypomethylation of CpG dinucleotides in the p66shc promoter. Moreover, they show that LDL-stimulated p66shc expression mediates a dysfunctional endothelial cell surface, with proadhesive and procoagulant features.

hypercholesterolemia leads to vascular endothelial dysfunction (27). Lowering of low-density lipoprotein (LDL) cholesterol in hypercholesterolemic subjects improves endothelial function (33) and reduces the risk of myocardial infarction (34). LDL can be oxidized (Ox-LDL) in vivo (23), and Ox-LDL induces endothelial cell death and endothelial dysfunction (30). In addition, uptake of Ox-LDL by macrophages leads to the formation of foam cells, major culprits in the pathogenesis of an atherosclerotic plaque. Independent of oxidation, native (nonoxidized) LDL (n-LDL) also affects the endothelium, resulting in a proinflammatory and proadhesive phenotype. For example, n-LDL promotes endothelial cell adhesiveness by inducing expression of intercellular adhesion molecule-1 (ICAM1) (29).

Cardiovascular risk associated with hypercholesterolemia may not all be attributable to the direct toxic effects of Ox-LDL or n-LDL on the endothelium and vasculature as a whole. In this regard, modification of chromatin has gained traction as a mechanism for the long-term and heritable cardiovascular risk associated with hypercholesterolemia. This is underscored by experimental findings showing epigenetic modifications in animal models of hypercholesterolemia (20). In human pathological samples as well, genomic hypomethylation has been observed in lesions of vascular atherosclerosis (10).

p66shc belongs to the shcA family of adaptor proteins. It governs the cellular redox state via several mechanisms, including regulation of mitochondrial electron transport (6) and activity of the rac1 GTPase (13). Under pathophysiological conditions, p66shc promotes endothelial dysfunction (4) and high-fat diet-induced atheroma formation (22). In addition, downregulation of p66shc under physiological conditions increases endothelium-dependent vasorelaxation by stimulating endothelial nitric oxide synthase activity (37). Moreover, p66shc expression in peripheral blood monocytes correlates well with the extent of atherosclerotic coronary disease, suggesting a role for p66shc in such disease (5). p66shc expression is regulated at several levels, including phosphorylation-mediated increase in protein stability (14) and methylation of CpG dinucleotides in its promoter (26, 31). Whether and how LDL cholesterol stimulates p66shc expression has not been reported previously.

Empowered by prior findings showing a change in the global epigenome associated with hypercholesterolemia, we asked if LDL cholesterol affects endothelial p66shc expression via changes in methylation of the p66shc promoter. In addition, we questioned whether p66shc plays a causal role in the dysfunction of endothelial cells triggered by LDL cholesterol.

MATERIALS AND METHODS

Cell culture and transfection.

Human umbilical vein endothelial cells (HUVECs) were purchased from Clonetics (San Diego, CA), cultured in endothelial growth medium-2 (Lonza, Walkersville, MD), and used until passage 10. Human embryonic kidney 293 (HEK 293) cells and U937 monocytes were purchased from American Type Culture Collection. Bovine aortic endothelial cells (BAEC) were purchased from Cell Applications (San Diego, CA). HEK 293 cells and BAECs were cultured in DMEM (Mediatech, Manassas, VA) supplemented with 10% FBS, 100 μg/ml streptomycin, and 100 μg/ml penicillin. U937 monocytes were cultured in RPMI 1640 (Mediatech). HEK 293 cells were transfected with plasmid, and HUVECs were transfected with validated small-interfering RNA (siRNA)-p66Shc, siRNA-DNA methyltransferase (DNMT) 1, siRNA-DNMT3b, or negative control siRNA using Lipofactamine 2000 (InVitrogen, Carlsbad, CA) per the recommendations of the manufacturer.

Low-density lipoprotein.

n-LDL isolated by density ultracentrifugation form fresh blood obtained from normal volunteers was purchased from Lee Biosolutions (St. Louis, MO). Purity of LDL was verified by Helena Lipoprotein cellulose acetate electrophoresis as a single band. Oxidation of LDL was measured by thiobarbituric acid reactive substance assay and was consistently below 2 nmol MDA/mg apolipoprotein B.

Antibodies and immunoblotting.

Anti-shcA antibody was purchased from BD Biosciences Pharmingen (San Jose, CA) and anti-ICAM1, anti-thrombomodulin (TM), and anti-β-actin antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Fifty micrograms of whole cell lysate were subjected to SDS-PAGE, transferred to nitrocellulose membrane, and probed with each antibody, and the appropriate horseradish peroxidase-labeled secondary antibody. Chemiluminescent signal was developed using Super Signal West Pico or Femto substrate (Pierce, Rockford, IL), and blots were imaged and quantified with a Gel Doc 2000 Chemi Doc system with Quantity One software (Bio-Rad).

Monocyte-endothelial cell adhesion assay.

Methodology to quantify adhesion of U937 monocytes to HUVEC has been described previously (16).

Plasma cholesterol.

Total cholesterol in mouse plasma was measured with Cayman's Cholesterol Assay Kit (Cayman Chemical, Ann Arbor, MI) per the recommendation of the manufacturer.

Promoter-reporter assay.

The human p66Shc promoter-reporter construct has been described previously (15). Promoter-reporter constructs were cotransfected with a constitutive Renilla luciferase plasmid using Lipofectamine 2000. Firefly and Renilla luciferase luminescence were measured using the Dual luciferase reporter kit (Promega) according to the recommendations of the manufacturer. The firefly-to-Renilla ratio was calculated to normalize for variations in transfection efficiencies.

Site-directed mutagenesis.

The CpG(6,7) mutant in the human p66Shc promoter-reporter construct has been described previously (15).

Quantitative real-time PCR.

Total RNA from cells or mice aorta were isolated with TRIzol Reagent (InVitrogen) according to the recommendations of the manufacturer. Quantitative real-time PCR was performed using the Prism 7000 Sequence Detection System (Applied Biosystems) with the SuperScript III Platinum SYBR Green One-Step qRT-PCR Kit (InVitrogen). Primer sequences for human p66Shc are as follows: forward, 5′-AAG TAC AAT CCA CTC CGG AAT GA-3′ and reverse, 5′-GG GCC CCA GGG ATG AAG-3′. The amplicon generated using these primers is within the coding region of the unique CH2 domain of p66Shc. The primer sequences for mouse p66Shc are forward, 5-GAC GAT AGT CCG ACT ACC CTG TGT-3′ and reverse, 5-CAG CAG GAT TGG CCA GCT T-3. The primer sequences for human ICAM1 are forward, 5′-GCC CTT TCT CGT CAT TTA GAT CTC-3′ and reverse, 5′-CT ATG TTA ATA TGA GCT TTG ACA AAA-3′. Human and mouse GAPDH was used as internal controls. The primer sequences for human GAPDH are forward, 5′-ATG GCA TCA AGA AGG TGG TG-3′ and reverse, 5′-CAT ACC AGG AAA ATG AGC TTG-3′ and for mouse GAPDH are forward, 5′-GGC AAA TTC AAC GGC ACA GT-3′ and reverse, 5′-CGC TCC TGG AAG ATG GTG AT-3′.

Adenoviral constructs.

Recombinant adenoviruses encoding a short-hairpin RNA (shRNA) targeting p66Shc (Adp66ShcRNAi) and the inert Escherichia coli LacZ gene (AdLacZ) have been described previously (37). Cells were infected with viruses for 4–6 h. Medium was changed, and cells were incubated for another 24 h.

Plasma clotting assay.

Clotting of human platelet-rich plasma on HUVEC monolayers has been described previously (18).

Chromatin immunoprecipitation assay.

Chromatin immunoprecipitation assays were performed in HUVEC using a ChIP Assay Kit (Upstate) and a rabbit polyclonal antibody for anti-acetyl-lysine 9,14 for histone 3 (H3) (Millipore). Primers (forward 5′-GCC TGT GAG TCA GCA CTG TCC TCA CG-3′, reverse 5′-GTG GCT TCC GCT CCC CAG CTC AG-3′) were used to PCR amplify a 169-bp fragment of human p66hc promoter. Input chromatin was used for PCR amplification of the same region of the promoter.

Quantification of CpG methylation in HUVEC.

CpG(6,7) methylation was quantified by methylation-specific real-time PCR as previously described (9) with 100 ng of bisulfite-converted genomic DNA as the template and the following methylation-specific primers: forward 5′-TGT GAG TTA GTA TTG TTT TTA- 3′, reverse 5′-CAA ACC CCA TCC CCG CCC AAC G- 3′.

Mice.

Wild-type C57Bl/6 mice were purchased from Jackson Laboratories (Bar Harbor, ME). Six- to eight-week-old mice were fed a normal chow or a high-fat diet (TD.88137; Harlan Teklad) for 12 wk. Mice were killed by cervical dislocation, and aortas were harvested. All studies on animals were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh and conformed with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health.

Statistical analysis.

All experiments were performed at least three times. Data are expressed as means ± SD. Statistical analysis was performed with SigmaStat. Data in which two conditions were compared were tested using the Student's t-test. Data in which more than two conditions were compared in a single experiment were tested using ANOVA or repeated measures of ANOVA as appropriate. Correlation between variables was evaluated using the Pearson product method. A P value of <0.05 was considered statistically significant.

RESULTS

LDL stimulates endothelial p66shc expression.

We first determined if n-LDL changes p66shc expression in cultured human endothelial cells. p66shc expression in HUVEC, both at the protein (Fig. 1A) and mRNA (Fig. 1B) level, was stimulated in a dose-dependent fashion by LDL. Moreover, LDL stimulated activity of a p66shc promoter-driven luciferase construct (Fig. 1C). To determine if the same holds true in vivo, we measured p66shc expression in a high-fat diet model of hypercholesterolemia. Aortic p66shc expression was significantly higher in mice fed a high-fat diet compared with wild-type mice on a normal diet (Fig. 1C). Thus LDL cholesterol in vitro and hypercholesterolemia in vivo stimulate endothelial cell and vascular p66shc transcription, respectively.

Fig. 1.

Low-density lipoprotein (LDL) increases endothelial p66Shc expression. A: representative immunoblot for p66Shc in lysates of human umbilical vein endothelial cells (HUVEC) incubated with LDL (24 h). Data are expressed as relative fold change in p66shc normalized to β-actin. *P < 0.05, **P < 0.01, and ***P < 0.001 (n = 3). B: quantitative real-time PCR for p66shc in HUVEC incubated with LDL (200 μg/ml for 8 h). Data are expressed as fold change in p66shc RNA normalized to GAPDH RNA. ***P < 0.001 (n = 3). C: p66Shc promoter activity in bovine aortic endothelial cells (BAEC) challenged with LDL (200 μg/ml for 24 h). Normalized promoter activity is expressed relative to untreated cells. ***P < 0.001 (n = 3). D: quantitative real-time PCR for p66shc in whole aortas of wild-type C57Bl/6 mice on normal diet [wild type (WT), ND] and high-fat diet (WT, HFD) for 12 wk. Plasma cholesterol was measured in the same mice. Data are expressed as fold change in p66shc RNA normalized to GAPDH RNA. *P < 0.05 (n = 4).

LDL induces p66shc promoter activity via CpG hypomethylation.

Next, we asked if LDL-induced p66shc transcription is mediated by hypomethylation of CpG dinucleotides in the p66shc promoter. Our prior work has shown that two CpG dinucleotides [CpG(6,7)] in the human p66shc are methylated in human endothelial cells, and hypomethylation of these dinucleotides by homocysteine is responsible for homocysteine-stimulated p66shc transcription (16). Hypothesizing that these same CpG dinucleotides play a role in LDL-stimulated p66shc expression, we first measured the effect of LDL on methylation of these dinucleotides in HUVEC. Challenge with LDL led to hypomethylation of CpG(6,7) (Fig. 2A). Next, we determined if CpG(6,7) mediate the effect of LDL on p66shc transcription. To do this, we compared the effect of LDL on two human p66shc promoter-reporter constructs: one that consists of 1,141 bp of the wild-type sequence encompassing Cpg(6,7) and the other in which cytosines of both CpG(6,7) are mutated to nonmethylatable adenosines [CpG(6,7) mutant promoter]. Under resting conditions, activity of the CpG(6,7) promoter was higher compared with the wild-type promoter (Fig. 2B), underscoring the importance of these CpG dinucleotides to repression of basal promoter activity. Moreover, compared with the wild-type promoter, response of the CpG(6,7) mutant promoter to LDL was blunted significantly (Fig. 2B). These findings show that exposure of endothelial cells to LDL leads to hypomethylation of CpG(6,7), and hypomethylation of these dinucleotides mediates LDL-stimulated p66shc promoter activity.

Fig. 2.

LDL increases p66Shc transcription via hypomethylation of CpG dinucleotides in the promoter. A: quantification of CpG(6,7) methylation in the human p66shc promoter by real-time methylation-specific PCR. PCR was performed on genomic DNA from untreated HUVEC and those challenged with LDL (200 μg/ml for 24 h). Data are expressed relative to untreated HUVEC. **P < 0.01 (n = 6). B: CpG(6,7) mediates LDL-stimulated p66shc promoter activity. Human embryonic kidney (HEK) 293 cells transfected with luciferase reporter constructs encoding WT or mutant [nonmethylatable on CpG(6,7)] promoters were untreated or challenged with LDL (200 μg/ml for 24 h). Normalized promoter activities are expressed relative to the untreated p66shc WT promoter. **P < 0.01 and ***P < 0.001. NS, nonsignificant (n = 3). C: knock down of DNA methyltransferases (DNMT) 1 and 3B prevents LDL-stimulated p66shc expression. Basal and LDL-stimulated (200 μg/ml for 24 h) p66shc protein in HUVECs in which DNMT1 and DNMT3b were knocked down. p66shc expression was quantified densitometrically normalized to β-actin, and expressed relative to untreated cell transfected with control small-interfering (si) RNA (siRNA control). Bottom, quantitative real-time PCR for DNMT1 and DNMT3b, normalized to GAPDH, and expressed relative to siRNA control. ***P < 0.001 (n = 3). D: inhibition of DNMT activity suppresses LDL-stimulated p66shc expression. Quantitative real-time PCR for basal and LDL-stimulated (200 μg/ml for 24 h) p66shc expression in HUVEC treated with the DNMT inhibitor 5′-azacytidine (5′-AZA, 10 μM for 24 h). p66shc RNA, normalized to GAPDH RNA, is expressed relative to untreated cells. ***P < 0.001 (n = 3).

To establish if changes in DNA methylation are responsible for LDL-stimulated p66shc expression, we asked if knock down of DNMT negates the effect of LDL on p66shc transcription. The two principal DNMTs expressed postnatally, DNMT1 and -3b, were simultaneously knocked down in HUVEC with targeted siRNAs (Fig. 2C). Knock down of DNMTs increased basal p66shc expression (Fig. 2C). In addition, LDL failed to increase p66shc in HUVEC in which DNMT1 and -3b expression was knocked down (Fig. 2C). In complementary experiments, we determined if pharmacological inhibition of DNMT activity achieved the same results as DNMT knock down. DNMT was inhibited in HUVEC with the pan-DNMT inhibitor 5′-azacytidine (5′-AZA). Under basal conditions, similar to that observed with DNMT knock down, 5′-AZA increased p66shc expression (Fig. 2D). Moreover, in HUVEC pretreated with 5′-AZA, LDL did not further increase p66shc expression (Fig. 2D). These findings show that LDL acts via DNMTs to exert its effect on endothelial p66shc expression.

LDL stimulates H3 acetylation on the p66shc promoter.

Changes in histone modifications often accompany those in DNA methylation. CpG hypomethylation partners with histone acetylation to result in an open chromatin accessible to transcription factors. Therefore, we asked whether LDL leads to signature changes in histone acetylation in the p66shc promoter. In HUVEC, LDL stimulated acetylation of H3 on lysines 9 and 14 (Fig. 3A), an epigenetic mark associated with a transcriptionally active chromatin. This effect of LDL was similar to that observed with the histone deacetylase inhibitor trichostatin A (TSA) (Fig. 3A). Thus, hypomethylation of CpG(6,7) in the p66shc promoter by LDL is accompanied by hyperacetylation of H3 at this locus.

Fig. 3.

LDL modifies chromatin on the p66shc promoter. A: LDL hypoacetylates H3 on the p66shc promoter. Chromatin immunoprecipitation (IP) assay for H3 (H3-Ac-K) in untreated HUVEC, and HUVEC challenged with LDL (200 μg/ml for 48 h) or treated with trichostatin A (TSA, 300 nM for 48 h). Amplified band represents a 169-bp region in the human p66shc promoter encompassing CpG(6,7). Nonimmunoprecipitated chromatin was used as input. B: p66shc promoter activity in HEK 293 cells challenged with 200 μg/ml LDL or 300 nM TSA or both for 24 h. Normalized activities are expressed relative to untreated cells. **P < 0.01 and ***P < 0.001 (n = 3).

We next assessed the role of histone acetylation on p66shc transcription. Similar to LDL, TSA increased p66shc promoter activity (Fig. 3B). Moreover, LDL in the presence of TSA did not lead to any further increase in promoter activity (Fig. 3B). These findings illustrate the importance of histone acetylation in stimulating p66shc transcription. In addition, they are in accordance with the effect of LDL on H3 acetylation (Fig. 3A), suggesting that the effect of LDL on p66shc promoter methylation leads to a transcriptionally active chromatin via H3 acetylation, and that this is necessary and sufficient for p66shc transcription.

p66shc mediates LDL-stimulated ICAM1 expression.

Vascular endothelial dysfunction is associated with upregulation of inflammatory genes, including expression of ICAM1 on the endothelium (7). Moreover, LDL upregulates endothelial adhesion molecules, including ICAM1 (28, 36). Therefore, we asked if p66shc mediates LDL-stimulated expression of endothelial ICAM1. LDL increased endothelial expression of ICAM1 both at the protein (Fig. 4A) and mRNA (Fig. 4B) levels. We then examined the role of p66shc in this upregulation. LDL-induced ICAM1 expression was blunted significantly in HUVEC in which p66shc expression was knocked down with siRNA (Fig. 4C). These findings underscore the role of p66shc in LDL-stimulated endothelial cell inflammatory gene expression.

Fig. 4.

p66shc mediates LDL-induced intercellular adhesion molecule-1 (ICAM1) expression and monocyte adhesion. LDL induces endothelial ICAM1. A: quantitative real-time PCR for ICAM1 in HUVEC challenged with LDL (200 μg/ml for 6 h). ICAM1 mRNA, normalized to GAPDH mRNA, is expressed relative to untreated cells. ***P < 0.001 (n = 5). B: representative immunoblot and quantification of ICAM1 in untreated HUVEC and HUVEC challenged with LDL for 24 h. ICAM1, normalized to β-actin, is expressed relative to untreated cells. **P < 0.01 and ***P < 0.001 (n = 3). C: p66shc mediates LDL-induced endothelial ICAM1. HUVEC transfected with negative control siRNA (siRNA control) or p66shc siRNA (siRNA-p66Shc) were untreated or challenged with LDL (200 μg/ml for 24 h). Representative immunoblots for ICAM1 and p66shc (top), quantification of ICAM1 protein, normalized to β-actin, and expressed relative to untreated cells (middle), and ICAM1 mRNA, normalized to GAPDH mRNA, and expressed relative to untreated cells (bottom) are shown. ***P < 0.001 (n = 3). D: adhesion of U937 monocytic cells to HUVEC infected with AdLacZ or Adp66shcRNAi, and challenged with LDL (200 μg/ml for 6 h). Data are expressed as the no. of adherent U937 cells/mm². **P < 0.01 (n = 3). Representative photomicrographs are shown (bottom).

p66shc mediates LDL-stimulated adhesion of monocytes to endothelial cells.

Upregulation of adhesion molecules on endothelial cells leads to adhesion and infiltration of inflammatory leukocytes into the vessel wall and is one of the hallmarks of a dysfunctional endothelium associated with hypercholesterolemia (25). Therefore, we asked if p66shc is responsible for LDL-stimulated adhesion of leukocytes to endothelial cells. To do this, we examined the effect of LDL on adhesion of the monocytic cell line U937 to HUVEC. HUVEC were infected with a control adenovirus encoding the E. coli LacZ gene (AdLacZ) or a virus encoding and shRNA targeting p66shc (Adp66shcRNAi). In control AdLacZ-infected HUVEC, challenge with LDL resulted in marked adhesion of monocytes to endothelial cells (Fig. 4D). Importantly, LDL-induced increase in monocyte adhesion was significantly blunted in HUVEC infected with Adp66shcRNAi (Fig. 4D). Thus, p66shc, in part, mediates the adhesion of LDL-induced monocytes to endothelial cells.

p66shc mediates LDL-stimulated downregulation of endothelial cells.

Hypercholesterolemia is also associated with a prothrombotic phenotype (32). TM is an anti-inflammatory and anti-thrombotic protein expressed on the endothelial cell surface and is downregulated on a dysfunctional endothelium (35). We asked if downregulation of TM is one feature of a prothrombotic phenotype induced by LDL in endothelial cells, and whether p66shc mediates this phenotype. We first examined the effect of LDL on TM expression in HUVEC. TM expression in AdLacZ-infected HUVEC was significantly downregulated by LDL (Fig. 5A). Moreover, the LDL-induced decrease in TM expression was negated in HUVEC infected with Adp66shcRNAi (Fig. 5A). Thus, p66shc mediates LDL-stimulated inhibition of the anti-thrombotic molecule TM in endothelial cells.

Fig. 5.

LDL decreases thrombomodulin (TM) expression and stimulates coagulation on endothelial cells via p66Shc. A: p66shc mediates LDL-induced downregulation of endothelial TM. Representative immunoblot and quantification of TM in HUVEC infected with AdLacZ or Adp66shcRNAi and challenged with LDL (200 μg/ml, 24 h). Data are expressed as relative fold change in TM normalized to β-actin. ***P < 0.001 (n = 3). B: time to clotting of normal human platelet-rich plasma on HUVEC infected with AdLacZ or Adp66ShcRNAi, and challenged with LDL (200 μg/ml for 18 h). Values are expressed as half-time to complete clot formation. ***P < 0.001 (n = 3).

p66shc mediates LDL-induced platelet thrombus formation on endothelial cells.

Next, we explored the role of p66shc in promoting a procoagulant endothelial cell phenotype. We used an assay that measures the time to platelet-rich clot formation on a monolayer of HUVEC and asked whether knock down of p66shc prolongs this time. Under basal conditions, time to clot formation was significantly higher in Adp66shcRNAi-infected cells compared with AdLacZ-infected cells (Fig. 5B). In addition, LDL-induced time to clot formation was prolonged significantly in Adp66shcRNAi-infected cells compared with AdLacZ-infected HUVEC (Fig. 5B). Thus, p66shc is, in part, responsible for shifting endothelial cells to a prothrombotic phenotype, both under basal conditions and also when challenged with LDL.

DISCUSSION

The promoter region of p66shc does not have a CpG island. However, there is evidence that functional CpG methylation occurs in non-CpG islands (11, 12) and CpG methylation silences p66shc in several cell types (31). In HUVEC, only two CpG dinucleotides in the proximal p66shc promoter, which lie after the transcription start site, are methylated (16), and therefore it was these dinucleotides on which we focused our attention. The relative paucity of methylated CpG dinucleotides in the p66shc promoter in HUVEC compared with cells in which p66shc expression is completely silenced explains why p66shc is expressed in HUVEC. However, even though the p66shc promoter is not heavily methylated in HUVEC, demethylation of only a few CpGs can affect gene expression, as has been observed for the estrogen receptor-β gene in endothelial cells (17), and for the tumor suppressor p53 (19).

Hypercholesterolemic apolipoprotein E-deficient (ApoE^−/−) mice have aberrant DNA methylation patterns (20). These mice have decreased global methylation in aorta and peripheral blood mononuclear cells but no change in DNA methylation in liver and skeletal muscle at 6 mo of age, suggesting tissue-specific hypomethylation in the setting of hypercholesterolemia. DNA hypomethylation has also been observed in atherosclerotic lesions that ApoE^−/− mice develop when fed a high-fat diet (10). Although these reports indicate that hypercholesterolemia leads to global CpG hypomethylation, at least in vascular tissue, it would be naïve to conclude that hypercholesterolemia results in hypomethylation of all genes expressed in the vasculature which are regulated by CpG methylation. There are many examples of LDL or modified LDL hypermethylating genes in vascular cells. For example, Ox-LDL results in downregulation of its own receptor (LOX-1) in endothelial cells via hypermethylation of the LOX-1 promoter (21). Thus the hypercholesterolemic state may lead to change in CpG methylation in both a tissue- and gene-specific manner.

The role of DNMT in regulating endothelial p66shc expression is likely to be complex. The fact that 5′-AZA resulted in an increase in basal p66shc expression indicates that the net result of inhibiting all DNMT is an increase in p66shc transcription. This was also corroborated in cells in which both DNMT1 and -3b were knocked down. In addition, that 5′-AZA or knock down of DNMT1 and -3B abrogates LDL-induced upregulation of p66shc suggests a role for these DNMTs in hypercholesterolemia-stimulated p66shc expression. However, the magnitude of increase in p66shc with 5′-AZA was less than that seen with LDL, hinting at the possibility that some DNMT, not involved in LDL-induced upregulation of p66shc, may be acting to increase basal p66shc transcription in HUVEC.

It is important to note that our data show that p66shc is an important, but not the sole, mediator of a dysfunctional endothelium triggered by LDL. LDL-induced ICAM1 expression, adhesion of monocytes, and platelet-rich clot formation were inhibited, but not abrogated, by knock down of p66shc. These findings illustrate that LDL, and other triggers of endothelial dysfunction, activate a host of signaling pathways, only some of which may involve p66shc. It is also important to appreciate that p66shc expressed in cell types, other than endothelial cells, may be important in promoting endothelial dysfunction and contributing to the pathogenesis of atherosclerosis. One such cell type may be circulating monocytes themselves, which play a central part in the inflammatory component of atherogenesis. Supporting this hypothesis are observations that p66shc expression in peripheral blood monocytes correlates with extent of atherosclerotic disease (5).

The effect of lipid particles and hypercholesterolemia on CpG methylation begs the question as to how precisely this occurs. Although research on this subject is still in its infancy, emerging data show that the protein moieties of lipid particles can be detected in the nuclei of cells and can associate with the chromatin (8), raising the possibility that lipoproteins may modify chromatin directly. An alternative to direct effects of lipoproteins could be an indirect mechanism whereby lipoproteins activate cellular signaling cascades such as kinases, which in turn affect DNA and histone modification enzymes. Yet another mechanism could involve targeting of specific DNMTs by lipoproteins as has been shown for microRNA-mediated downregulation of DNMT3b by Ox-LDL (3).

Hypercholesterolemia not only affects DNA methylation patterns in the adult organism but, when present during pregnancy, may also have long-lasting impact on the newborn. Maternal hypercholesterolemia results in changes in aortic gene expression of offspring that persist long after the fetal exposure to hypercholesterolemia. This phenomenon of in utero or “developmental” programming has gained growing acceptance as a mechanism to explain epidemiological data showing increased risk of cardiovascular disease in adults who were exposed to perturbed metabolic conditions such as hyperglycemia and hypercholesterolemia in utero (24). Notably, heritable in utero epigenetic modifications have been implicated in the pathogenesis of postnatal cardiovascular disorders (1, 2). Thus, LDL-induced epigenetic changes in the p66shc gene may have relevance not only in the postnatal vasculature but also in the development of endothelial dysfunction and its sequelae in adults who have been prenatally exposed to a hypercholesterolemic environment.

GRANTS

This work was supported by National Institutes of Health Grants HL-065608, HL-070929, HL-094959, and HL-098892 to K. Irani.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: Y.-R.K. and K.I. conception and design of research; Y.-R.K., C.-S.K., A.N., A.K., and S.K. performed experiments; Y.-R.K., A.K., T.A.H., and K.I. analyzed data; Y.-R.K. and K.I. interpreted results of experiments; Y.-R.K. and K.I. prepared figures; Y.-R.K. and K.I. drafted manuscript; Y.-R.K. and K.I. edited and revised manuscript; Y.-R.K. and K.I. approved final version of manuscript.