Skip main navigation

Hemodynamic Disturbed Flow Induces Differential DNA Methylation of Endothelial Kruppel-Like Factor 4 Promoter In Vitro and In Vivo

Originally published Research. 2014;115:32–43



    Hemodynamic disturbed flow (DF) is associated with susceptibility to atherosclerosis. Endothelial Kruppel-Like Factor 4 (KLF4) is an important anti-inflammatory atheroprotective transcription factor that is suppressed in regions of DF.


    The plasticity of epigenomic KLF4 transcriptional regulation by flow-mediated DNA methylation was investigated in vitro and in arterial tissue.

    Methods and Results:

    To recapitulate dominant flow characteristics of atheroprotected and atherosusceptible arteries, human aortic endothelial cells were subjected to pulsatile undisturbed flow or oscillatory DF containing a flow-reversing phase. Differential CpG site methylation was measured by methylation-specific polymerase chain reaction, bisulfite pyrosequencing, and restriction enzyme-polymerase chain reaction. The methylation profiles of endothelium from disturbed and undisturbed flow sites of adult swine aortas were also investigated. In vitro, DF increased DNA methylation of CpG islands within the KLF4 promoter that significantly contributed to suppression of KLF4 transcription; the effects were mitigated by DNA methyltransferase (DNMT) inhibitors and knockdown of DNMT3A. Contributory mechanisms included DF-induced increase of DNMT3A protein (1.7-fold), DNMT3A enrichment (11-fold) on the KLF4 promoter, and competitive blocking of a myocyte enhancer factor-2 binding site in the KLF4 promoter near the transcription start site. DF also induced DNMT-sensitive propathological expression of downstream KLF4 transcription targets nitric oxide synthase 3, thrombomodulin, and monocyte chemoattractant protein-1. In support of the in vitro findings, swine aortic endothelium isolated from DF regions expressed significantly lower KLF4 and nitric oxide synthase 3, and bisulfite sequencing of KLF4 promoter identified a hypermethylated myocyte enhancer factor-2 binding site.


    Hemodynamics influence endothelial KLF4 expression through DNMT enrichment/myocyte enhancer factor-2 inhibition mechanisms of KLF4 promoter CpG methylation with regional consequences for atherosusceptibility.


    Although associated with systemic environmental risk factors, genetic defects and aging,1 atherosclerotic lesions typically originate locally, developing preferentially at predictable sites of curvature, branching, and bifurcation in large arteries where disturbed blood flow (DF) is a prominent characteristic.2 In contrast, regions dominated by pulsatile unidirectional laminar undisturbed flow (UF) are relatively atheroresistant. Hemodynamic DF, although usually also laminar, is characterized by flow separation, transient flow reversals, and average low shear forces that define the atherosusceptible regional environment.3

    The endothelium plays a central role in the initiation and development of inflammatory atherosclerosis. Endothelial phenotypes in prelesional atherosusceptible regions are subtly different from those located at nearby atheroresistant sites where UF is prevalent. In prelesional DF regions of mouse and swine, differential transcriptional profiling has identified endothelia that are sensitized for proinflammatory pathways, coagulation, and redox,36 and the chronic low level activation of endoplasmic reticulum stress and the unfolded protein response.7 When the principal characteristics of UF and DF are recapitulated in vitro, important protective pathways are suppressed by DF including the expression and activity of Kruppel-like factor 4 (KLF4) and nitric oxide synthase 3 (NOS3).5,8 The expression of NOS3, essential for regulation of vascular tone and maintenance of the quiescent state of endothelium,8,9 is regulated by hemodynamic shear stress linked to a series of upstream transcription factors that includes KLF4, KLF2, and RelA/p65.1012 KLF4 and KLF2 are zinc-finger regulatory transcription factors for gene networks that confer atheroresistant anti-inflammatory and antithrombotic properties to the endothelium; localized dysfunction or suppression of KLF4 is, therefore, propathological.10,13,14 Dysregulation of KLF4 and NOS3 by genetic manipulation of the endothelium of mice significantly contributes to the development and progression of atherosclerosis.9,10,15

    The mechanisms linking hemodynamics characteristics such as UF and DF to endothelial phenotype, function, and pathosusceptibility are under intensive study at multiple levels of regulation, most recently epigenetic. Flow-induced histone modification and miRNAs have been shown to shape endothelial phenotype identities,1620 but differential DNA methylation responses to different flow profiles encountered in vivo and their recapitulation in vitro have not been addressed.

    DNA methylation is one of the critical epigenetic mechanisms controlling gene expression.21 In vertebrates, DNA methylation occurs at carbon 5 of cytosine in CpG dinucleotides (5-methylcytosine). When occurring within the promoter regions of genes, it dramatically suppresses transcription by direct inhibition of transcription factor binding and recruitment of methyl-CpG-binding proteins, which further hinder access to the recognition site of transcription factors or modulate chromatin structure by the recruitment of histone-modifying proteins.2224 The DNA methylation landscape of the genome is established by methylation and demethylation enzymes. DNA methyltransferase 1 (DNMT1) maintains tissue-specific DNA methylation patterns via methylation of a hemimethylated nascent DNA strand during cell proliferation.25 DNMT3A and 3B are required for genome-wide de novo methylation and play crucial roles in the establishment of DNA methylation patterns during development.25 Methylation by DNMTs is counterbalanced by passive and active DNA demethylation in which the Tet methylcytosine dioxygenase (TET) genes pathway has been suggested to play a central role in oxidizing 5-methylcytosine to 5-hydroxymethylcytosine.23

    An appreciation of DNA methylation dynamics in physiological and pathological gene regulation is emerging.21 Although the postdevelopment DNA methylation status associated with many genes tends to remain stable and is often linked to the maintenance of cell identity, epigenetic plasticity including DNA methylation/demethylation dynamics may be important for cellular adaptation responses including endothelial phenotype identity in different arterial hemodynamic environments. Here, we demonstrate the plasticity of endothelial DNA methylation within the promoter of the important atheroprotective transcription factor KLF4. We show that DF-induced hypermethylation significantly suppresses KLF4 transcription and regulates its downstream targets NOS3, thrombomodulin, and monocyte chemoattractant protein-1 (MCP-1). As far as we are aware these data are the first demonstrated changes in DNA methylation induced by physiological characteristics of flow and are supported by steady-state measurements in endothelial cells isolated from in vivo regions of hemodynamic DF and UF in swine aorta.


    Reagents and detailed molecular biology procedures are described in detail in Online Data Supplement.

    Cell Culture and Flow Experiments

    Human aortic endothelial cells (HAECs; passage 4–6; Lonza, Allendale, NJ) were cultured in complete endothelial cell growth medium-2 (EGM2) medium to confluence on 0.1% gelatin-coated glass slides (75×38 mm). The flow experiments were conducted as previously described.26 Postconfluent HAECs were subjected to pulsatile UF or DF in a parallel plate flow chamber for 2 days. UF waveform is characterized by a higher mean wall shear stress and fully antegrade flow (Figure 1A). In contrast, the DF waveform exposes cells to lower mean wall shear stress and a retrograde flow for one third of each cycle. The flow waveforms capture the dominant characteristics of human arterial hemodynamics flow behavior in UF and DF arterial sites. All flow in large arteries is unsteady (pulsatile). The defining feature of DF regions is that there is flow reversal during the cardiac cycle, whereas in UF, the flow is always unidirectional. Waveforms were generated digitally and converted to analog signals by a data acquisition card (USB-6229, National Instruments, Austin, TX) that controlled a 520U Watson-Marlow peristaltic pump (Cornwall, England). Flow was measured with an ultrasonic flowmeter (Transonic Systems Inc, Ithaca, NY) to ensure experimental repeatability. Both waveforms were sinusoidal while differing in amplitude, mean wall shear stress, and oscillatory shear index values. Wall shear stress values for the UF waveform ranged from 9.6 to 1.5 dyne/cm2 (mean 5.1 dyne/cm2) and for DF from 2 to −1.2 dyne/cm2 (mean of 0.4 dyne/cm2). The oscillatory shear index for UF equaled 0, whereas for DF, it was 0.32. An oscillatory shear index value of 0 corresponds to fully antegrade flow and 1 to fully retrograde flow.

    Figure 1.

    Figure 1. Undisturbed flow (UF) and disturbed flow (DF) regulation of Kruppel-like factor 4 (KLF4) and nitric oxide synthase 3 (NOS3) promoter methylation and gene transcription. A, Schematic illustration of the parallel plate flow apparatus and the arterial flow waveforms. Confluent human aortic endothelial cells were subjected to UF or DF for 2 days. Flow characteristics were confirmed by ultrasonic flowmeter. B, DF-induced suppression of KLF4 transcription. mRNA and premature mRNA (pre-mRNA) of KLF4 and NOS3 were determined by quantitative polymerase chain reaction (qPCR). The relative expression of DF to UF is computed after normalization of UF or DF to no-flow cells. All qPCR are normalized to ubiquitin B (UBB) and expressed as mean±SEM fold of UF. C, Schematic illustration of CpG island and CpG sites in human KLF4 and NOS3 promoters. D, DF differentially regulates KLF4 and NOS3 promoter methylation. DNA methylation was determined by methylation-specific PCR using primers targeting KLF4 (−152/+9) and NOS3 (−369/−196) promoter region. Data are normalized to UBB promoter and expressed as mean±SEM fold of UF. E, Methylation and hydroxymethylation of KLF4 and NOS3 promoters. After glucosylation by T4 β-glucosyltransferase (BGT) and restriction enzyme MspI and HpaII digestion, the digested genomic DNA were used for qPCR with gene-specific primers targeting CCGG-9 of the KLF4 promoter and CCGG-137 of the NOS3 promoter. Data are normalized to uncut DNA and expressed as mean±SEM. *P<0.05. n=4.

    Animal Studies

    Endothelia were obtained from adult pigs (6-month-old; ≈250 lb) immediately after euthanasia at a local slaughterhouse (Clemens Foods, Hatfield, PA). Aortas were harvested, and the vessel lumen was rinsed with ice-cold PBS. Endothelial cells were freshly harvested by gentle scraping of 1 cm2 regions located at the inner curvature of the aortic arch and nearby descending thoracic aorta representing DF and UF, respectively. Cells were transferred directly to lysis buffer for DNA or RNA extraction. Endothelial purity was routinely assessed with antibodies against platelet endothelial cell adhesion molecule-1 and alpha smooth muscle actin (ACTA2). Endothelial purity was also monitored by examining ACTA2 promoter hypermethylation (Online Figure VIII).

    Statistical Analysis

    Results are expressed as mean±SEM. Statistical analysis was performed by using an independent Student t test for 2 groups of data and ANOVA. If a normality test failed, data were compared by Mann–Whitney rank-sum test. P value <0.05 was considered significant.


    Induction of Endothelial KLF4 Promoter Methylation by DF In Vitro

    UF and DF characteristics were monitored in real time. The pulsatile flow was always antegrade in UF, whereas a brief reverse flow phase (negative shear stress, retrograde flow) lasting one third of the cycle provided an oscillatory shear index in DF (Figure 1A).

    Gene Expression

    Figure 1B (upper) demonstrates that, when referenced to UF, DF significantly inhibited the expression of both KLF4 premature mRNA (pre-mRNA) and mature mRNA by 65% and 75%, respectively. The introns are spliced out in mature mRNA, which is composed of exons only. DF also inhibited NOS3 pre-mRNA and mRNA by 41% and 61%, respectively (Figure 1B, lower). These data agree with previous reports of DF suppression of endothelial KLF4 and NOS3.5,8

    DNA Methylation

    A CpG island (CGI) exists at the transcription start site (TSS) of human KLF4 but not at the TSS of NOS3 (Figure 1C). The KLF4 promoter CGI is 2000 bp in length and the CpG observed/expected (CpGo/e) = 0.74. Promoter methylation is usually associated with gene regulation21; therefore, its association with the suppression of transcription was interrogated. Methylation status was determined by methylation-specific polymerase chain reaction (PCR) using specific primers targeting the KLF4 promoter. With reference to UF, DF significantly enhanced the KLF4 promoter methylation to 2.0-fold in CpG-rich regions after 2 days (Figure 1D, left). NOS3, an atheroprotective and KLF4 downstream target gene, was also evaluated. In contrast to KLF4, the human NOS3 promoter has poor CpG content (CpGo/e < 0.4). Importantly, methylation of NOS3 promoter was unchanged by DF (Figure 1D, right) despite suppression of NOS3 transcript expression. Thus, DF-suppressed KLF4 gene transcription was directly associated with hypermethylation of the KLF4 promoter, a status lacking in the NOS3 promoter.

    Minimal Contribution by Hydroxymethylation

    Hydroxymethylation at CpG sites is suggested to be associated with a DNA demethylation pathway that influences gene transcription regulation.23,27 Methylation-specific PCR does not discriminate between methylation and hydroxymethylation of CpG.28 Therefore, we quantified hydroxymethylation and methylation by using restriction enzyme-PCR. Two CCGG sites in the KLF4 promoter (804 and 9 bp upstream of TSS) were interrogated for DNA methylation and hydroxymethylation after exposure to UF or DF for 2 days (Figure 1E and Online Figure IA). At both sites, T4 β-glucosyltransferase and MspI enzyme treatment followed by quantitative PCR (qPCR) demonstrated that hydroxymethylation at the KLF4 promoter was low and remained unchanged by either flow treatment. In contrast, HpaII treatment followed by qPCR confirmed that DF significantly enhanced KLF4 promoter methylation in the same experiments (Figure 1E and Online Figure IA). Three CCGG sites (−745, −194, and −137) in the NOS3 promoter were also tested (Figure 1E and Online Figure I). T4 β-glucosyltransferase and MspI enzyme treatment followed by qPCR demonstrated that hydroxymethylation of NOS3 promoter was almost undetectable, whereas methylation remained low and not significantly different between UF and DF. We conclude that DNA methylation of the KLF4, but not NOS3, promoter region is influenced by flow characteristics and that the contribution by hydroxymethylation is not significant.

    DNA Methylation Inhibits the Transcriptional Activity of Myocyte Enhancer Factor-2

    It has been suggested that tumor necrosis factor α (TNFα) and resveratrol induction of the KLF4 gene can be regulated by myocyte enhancer factor-2 (MEF2) transcription factors.13,14,18,19 In silico analysis suggested only 1 MEF2 binding site TATTTAAAGTA (−64/−55) in the human KLF4 promoter. To test if MEF2 can bind to KLF4 promoter in cells, the MEF2 enrichment of chromatin was tested by chromatin immunoprecipitation PCR (ChIP-PCR) assay using 4 primers targeting the promoter region of KLF4. MEF2 was dramatically enriched in the region from −161 to −25 (MEF2 binding site), but not in other regions of the KLF4 promoter (Online Figure II).

    To confirm the ability of MEF2 to bind to the KLF4 promoter, nuclear protein extract from flow-acclimated HAEC was incubated with fluorescent-labeled oligonucleotide containing MEF2 binding site. Gel mobility assay showed 1 shifted band, which was abolished by anti-MEF2 antibody directed to the C terminus of MEF2, suggesting that C terminus of MEF2 is required in the formation of MEF2 complex and its association with the KLF4 promoter (Online Figure III). This was not a test of differential flow; protein extracts from DF and UF cells induced similar binding capacity of MEF2 to the unmethylated oligonucleotide.

    To test if hemodynamic forces could regulate the methylation status of CpG near the MEF2 binding sequence, bisulfite pyrosequencing was used to quantify the methylation levels at individual CpG sites (−135, −133, −66, −31, and −19 bp from KLF4 TSS). Consistent with methylation-specific PCR analysis (Figure 1D), DF further enhanced CpG methylation close to the MEF2 binding sequence. Methylation of CpG site −66 was enhanced from 34% to 73% and methylation of CpG site −31 was enhanced from 16% to 26% (Figure 2A). Consistent with robust DF-enhanced methylation near the MEF2 binding site, DF reduced the chromatin loading of MEF2 protein to KLF4 promoter by 80% (Figure 2B), confirming that MEF2-enhanced KLF4 gene transcription is impeded by methylation of KLF4 promoter.

    Figure 2.

    Figure 2. Kruppel-like factor 4 (KLF4) promoter methylation blocks myocyte enhancer factor-2 (MEF2)-mediated transcriptional activation of KLF4. A, Genomic DNA were isolated from confluent human aortic endothelial cells (HAECs), which were subjected to undisturbed flow (UF) or disturbed flow (DF) for 2 days. Bisulfite pyrosequencing quantified the methylation levels of individual CpG sites near the MEF2 binding sequence in the KLF4 promoter. B, Two-day UF and DF induction of MEF2 loading at the KLF4 promoter (−161/−25) was analyzed by ChIP–quantitative polymerase chain reaction. Data are normalized to chromatin enrichment of the ACTA2 promoter and are expressed as mean±SEM fold of UF. C, Methylated and unmethylated KLF4 promoter construct containing MEF2 binding sequence was transfected into HAEC. KLF4 promoter transcription activity was measured by luciferase enzyme activity. Data are expressed as mean±SEM fold of unmethylation. D, Tumor necrosis factor α (TNFα) induction of MEF2 binding to KLF4 promoter sequence. Confluent HAECs were treated with TNFα for 6 hours. Nuclear protein (10 μg) was incubated with carboxyfluorescein-labeled oligonucleotides with MEF2 binding sequence (TATTTAAAGTA) of the KLF4 promoter for 30 minutes. Anti-MEF2 antibody was preincubated with protein before adding the oligonucleotides (left). Competition experiments were performed by preincubations of the protein extract with anti-MEF2 antibody, competitor, or mutant oligonucleotides (right). Two CpG sites (−66 and −31 bp from KLF4 transcription start site) and the corresponding mutants flanking the core MEF2 binding sequence are underlined. *P<0.05. n=4 to 6.

    To further test if methylation of KLF4 promoter can impede KLF4 transcription, an in vitro luciferase reporter assay was performed. After successfully methylating a KLF4 promoter construct that contains MEF2 binding sequence (Online Figure IV), HAECs were transiently transfected with methylated and mock-methylated constructs. Methylation of KLF4 promoter construct significantly decreased its promoter activity by 64%, compared with the unmethylated construct (Figure 2C), further demonstrating inhibition of MEF2-enhanced KLF4 gene transcription by methylation of KLF4 promoter.

    Endothelia in prelesional DF regions of artery are sensitized for proinflammatory pathways.37 To test if proinflammatory cytokine TNFα can induce similar interactions of MEF2 with KLF4 promoter, oligonucleotide containing MEF2 binding sequence was incubated with nuclear protein extract from HAEC treated with or without TNFα. Gel mobility assay showed a prominent shifted band after TNFα treatment that was completely abolished by MEF2 antibody (Figure 2D, left), whereas no shifted band was observed in cells without TNFα treatment. The specificity of MEF2 site in the KLF4 promoter to bind endogenous MEF2 factors was then confirmed by progressive abolition of MEF2 protein binding by wild-type competitor (5- to 50-fold molar excess) as the molar concentration increased (Figure 2D, right). The importance of CpG in mediating MEF2 binding to KLF4 promoter was tested by mutation of 2 CpG dinucleotides (−66 and −33) flanking the MEF2 binding sequence. These mutations resulted in a mild effect on competing MEF2 binding, suggesting that the CpG sites flanking the core binding sequence are critical in mediating MEF2 binding. Taken together, these data suggest that DF-induced hypermethylation and mutation of CpG sites flanking MEF2 binding sequence (Figure 2A) can suppress the chromatin loading of MEF2 to KLF4 promoter (Figure 2B and 2D) and inhibit transcriptional activity (Figure 2C).

    DF-Induced DNMT3A Enrichment in the KLF4 Promoter

    DNA methylation at CpG sites is the net balance of methylation and demethylation dynamics. To test if DF can change the net methylation equilibrium, mRNA of the enzymes involved in methylation and demethylation processes were examined by qPCR (Figure 3A). The relative transcript expressions of methylation and demethylation enzymes were not significantly different between UF and DF; the mRNA of DNMT3L was not detectable by qPCR.

    Figure 3.

    Figure 3. Undisturbed flow (UF) and disturbed flow (DF) regulation of methylation and demethylation enzymes. Confluent human aortic endothelial cells (HAECs) were subjected to UF or DF for 2 days. A, mRNA of methylation enzymes (DNMT1, 3A, 3B, 3 L) and enzymes involved in demethylation pathways23 (TET1, 2, 3; TDG1; GADD45B; MBD4; SMUG1) were determined by quantitative polymerase chain reaction (qPCR). Data are normalized to ubiquitin B and expressed as mean±SEM fold of UF. B, Proteins of DNMT1, DNMT3A, and TET1 were analyzed by Western blot. Optical density of DNMT3A normalized to β-actin is expressed as mean±SEM fold of UF. C, UF- and DF-induced DNMT3A and TET1 enrichment in Kruppel-like factor 4 (KLF4) (−161/−25) and nitric oxide synthase 3 (NOS3) (−167/−15) promoter were analyzed by ChIP–qPCR. Data are normalized to the chromatin loading to ACTA2 promoter and are expressed as mean±SEM. D, Confluent HAECs were subjected to DF with or without RG108 for 2 days. Effect of RG108 on DF-induced DNMT3A loading to KLF4 promoter (−161/−25) was analyzed by ChIP–qPCR. Data are normalized to chromatin loading of DNMT3A to ACTA2 promoter and expressed as mean±SEM fold of UF. E, Confluent HAECs on glass slide post-transfected with DNMT3A-specific shRNA or scramble short hairpin RNA (shRNA) were subjected to DF for 2 days. Bisulfite pyrosequencing was used to quantify the methylation level of individual CpG sites at the KLF4 promoter. *P<0.05. n=4. DMSO indicates dimethyl sulfoxide; and TSS, transcription start site.

    DNMT3A protein was 1.7-fold enhanced in DF (P<0.05, Figure 3B). An increase in DNMT3A protein without changing the mRNA levels may suggest a potential post-transcriptional and post-translational mechanism (eg, sumoylation)29 in regulating DNMT3A by flow. Although DNMT1 and TET1 protein levels were comparable between UF and DF, DNMT3B protein was not detectable in HAEC by Western blot using 2 different antibodies from 2 vendors.

    Chromatin loading of DNMT3A enzyme at the KLF4 promoter and NOS3 promoter was examined by ChIP–PCR assay (Figure 3C, left). DF significantly enhanced (11-fold; P<0.00002) the enrichment of DNMT3A protein at the KLF4 promoter (−161/−25) but not at the NOS3 promoter (−167/−15) where DNMT3A was undetectable. The chromatin loading of TET1 enzyme at the KLF4 promoter remained unchanged between UF and DF, whereas TET1 at the NOS3 promoter was undetectable (Figure 3C, right). Pretreatment with RG108, a specific DNMT inhibitor, significantly suppressed (by 55%; P<0.05) DF-induced chromatin loading of DNMT3A to the KLF4 promoter (Figure 3D). Specific knockdown of DNMT3A significantly inhibited the DF-induced methylation of CpG sites near the MEF2 binding sequence (Online Figure V and Figure 3E). These data demonstrate that DF induced the chromatin enrichment of DNMT3A, leading to the hypermethylation of KLF4 promoter. They also suggest that methylation/demethylation enzymes DNMT3A and TET1 do not regulate NOS3 promoter methylation and hydroxymethylation in HAEC (Figure 1D, right and Online Figure I) under hemodynamic forces.

    DF-Enhanced, DNMT-Mediated KLF4 Promoter Methylation and Gene Silencing

    To determine if DF-enhanced methylation of KLF4 promoter is mediated by the activation of DNMTs, 5-azacytidine (5-Aza) and RG108 were used to block DNMT activity. 5-Aza can incorporate into DNA and covalently trap and inhibit DNMTs. RG108 has been shown to bind specifically to DNMTs and inhibit the enzyme activity with long half-life (20 days) and without significantly inducing apoptosis, cytotoxicity, and genotoxicity.30 In HAEC cultured under static conditions (no flow), dose–response curves showed that RG108 ≤100 μmol/L and 5-Aza ≤1 μmol/L did not significantly inhibit the methylation of the KLF4 promoter (Online Figure VIA).

    DF-mediated methylation was then measured after treatment with DNMT inhibitors. RG108 (20 μmol/L) and 5-Aza (1 μmol/L) completely prevented DF-specific (versus UF) methylation of the KLF4 promoter (−152/+9, Figure 4A), consistent with DF-induced KLF4 promoter hypermethylation mediated by DNMT. We also tested if DF-induced hypermethylation in other regions of the KLF4 promoter could be blocked by DNMT inhibitors. Consistent with the findings in −152/+9, DF-induced hypermethylation in regions −1520/−1328, 1339/−1141, and −808/−648 was completely blocked by RG108 (Online Figure VIB) presumably through (non-MEF2) DNMT global inhibition.

    Figure 4.

    Figure 4. Effects of RG108 and 5-azacytidine (5-Aza) on undisturbed flow (UF)- and disturbed flow (DF)-induced Kruppel-like factor 4 (KLF4) promoter methylation and gene expression. Confluent human aortic endothelial cells were subjected to UF and DF with RG108 (20 μmol/L), 5-Aza (1 μmol/L), or vehicle (dimethyl sulfoxide [DMSO]) for 2 days. A, KLF4 promoter methylation was analyzed by methylation-specific PCR. Data are normalized to ubiquitin B (UBB) promoter and expressed as mean±SEM fold of UF. B, Premature mRNA (pre-mRNA) and (C) mRNA of KLF4 were determined by quantitative polymerase chain reaction. Data are normalized to UBB and expressed as mean±SEM fold of UF. *,#P<0.05. n=4 to 6. *Different from UF. #Different from vehicle control.

    The effects of RG108 and 5-Aza on gene transcription were then examined. KLF4 pre-mRNA was completely restored to UF levels by RG108 and to 90% UF levels by 5-Aza (Figure 4B), consistent with the release of DF-suppressed KLF4 transcription by inhibition of DNMT. Mature KLF4 mRNA was rescued from 10% to 50% and 30% of UF levels by RG108 and 5-Aza, respectively (Figure 4C), indicating a post-transcriptional partial inhibition of KLF4 mRNA by DF. Our recent finding20 that intronic miRNA-92a could decrease KLF4 (and KLF2) mRNA stability may be related to this finding.

    Implications for the Regulation of KLF4 Target Genes

    KLF4 protein expression was inhibited by 38% (P<0.05) in DF (Figure 5A), consistent with in vivo data that endothelial KLF4 protein is downregulated in the DF region of aortic arch.20 DF-suppressed KLF4 protein levels in HAEC were rescued by the DNMT inhibitor RG108, which selectively increased KLF4 protein levels in DF without affecting UF levels (Figure 5A). As an atheroprotective transcription factor, KLF4 upregulates anti-inflammatory and antithrombotic factors such as NOS3 and thrombomodulin, whereas it inhibits the expression of proinflammatory factor, monocyte chemoattractant protein-1 (MCP-1).10,20 To show the effects of DF on the transcript expressions of these molecules and to test if blocking DNA methylation pathway could potentially restore the atheroprotective phenotypes of these KLF4 gene targets, the expression of NOS3, thrombomodulin, and MCP-1 were evaluated by qPCR in the absence and presence of DNMT inhibitor RG108 (Figure 5B). In controls (dimethyl sulfoxide vehicle), DF significantly inhibited NOS3 and thrombomodulin gene expression and enhanced MCP-1 gene expression (Figure 5B). This atherosusceptible profile was partially rescued by RG108. Suppression of thrombomodulin by DF was completely reversed by RG108, whereas inhibition of NOS3 and enhancement of MCP-1 by DF were both two thirds reversed by RG108. The suppression and restoration of these KLF4 target genes by RG108 was not associated with changes in DNA methylation of low CpG promoter of NOS3 or high CpG promoter of thrombomodulin (Online Figure VII). These results demonstrate that DF-induced proinflammatory and prothrombotic profiles in HAEC can be rescued by blocking upstream DNA methylation pathways in KLF4.

    Figure 5.

    Figure 5. Effect of RG108 on Kruppel-like factor 4 (KLF4) protein and KLF4-downstream genes transcription. Confluent human aortic endothelial cells were subjected to undisturbed flow (UF) and disturbed flow (DF) with RG108 (20 μmol/L) or vehicle control (dimethyl sulfoxide [DMSO]) for 2 days. A, KLF4 proteins in cytoplasmic extract (CE) and nuclear extract (NE) were determined by Western blot. Data normalized to β-actin are expressed as fold of UF. B, Thrombomodulin, monocyte chemoattractant protein-1 (MCP-1), and nitric oxide synthase 3 (NOS3) mRNA were determined by quantitative polymerase chain reaction. Data are normalized to ubiquitin B and expressed as mean±SEM fold of UF. *,#P<0.05. n=4. *Different from UF. #Different from vehicle control.

    KLF4 and NOS3 in the Swine Genome

    The hemodynamic environment, transcriptome, and atherosclerosis susceptibility in the arterial tree are similar between humans and swine.4 To explore parallel in vivo/in vitro epigenomic regulatory mechanisms, endothelial cells were isolated from distinct hemodynamic sites of DF (aortic arch; atherosusceptible) and UF (descending thoracic aorta; atheroresistant) in adult swine (Figure 6A). Consistent with the in vitro data, KLF4 and NOS3 mRNA and pre-mRNA in vivo were elevated in the UF region and suppressed in the DF region (Figure 6B).

    Figure 6.

    Figure 6. Differential gene expression and promoter methylation of swine arterial endothelial Kruppel-like factor 4 (KLF4) and nitric oxide synthase 3 (NOS3) in vivo. A, Schematic illustration of targeted undisturbed flow (UF) and disturbed flow (DF) regions in swine aorta. Endothelial cells were scraped gently from the descending thoracic aorta (DT) where UF is dominant, and from the inner curvature of aortic arch (AA) where DF is dominant. B, mRNA and premature mRNA (pre-mRNA) of swine KLF4 and NOS3 were determined by quantitative polymerase chain reaction. Data are normalized to the geometric mean of GAPDH and platelet endothelial cell adhesion molecule-1 and are expressed as mean±SEM fold of UF. C, Schematic illustration of the CpG island and CpG sites in swine KLF4 and NOS3 promoters. D, Methylation of KLF4 promoter and NOS3 promoter in UF and DF region of swine aorta was determined by using methylation-specific PCR targeting CpG-rich region. Data are normalized to ubiquitin B promoter without CpG sites and expressed as mean±SEM fold of UF. E, Alignment of KLF4 promoter region in multiple mammalian species showing a highly conserved region of myocyte enhancer factor-2 binding sequence. F, Methylation level of individual CpG sites in swine KLF4 promoter determined by bisulfite pyrosequencing. *Different from UF. P<0.05. n=6. TSS indicates transcription start site.

    Alignment of human and swine protein sequences showed homologies of 94.9% for KLF4 protein and 96.0% for NOS3 protein, and 89.3% similarity for KLF4 mRNA and 88.5% for NOS3 mRNA. Analysis of swine KLF4 promoter indicated 2 CGIs with CpGo/e 0.80 and 1.0 (Figure 6C), which are higher than that in human (0.74). In contrast to human NOS3 promoter that does not contain CGIs (Figure 1C, −500/+101 CpGo/e = 0.23; −200/+101 CpGo/e = 0.36), swine NOS3 promoter contains a CGI at the TSS (400 bp, −226/+174 CpGo/e = 0.61).

    DNA methylation in swine endothelium isolated from UF and DF aortic regions was, therefore, measured in CpG-rich regions of KLF4 and in NOS3. As shown in Figure 6D (left), although DNA methylation remained unchanged in NOS3 promoter, in the KLF4 promoter, there was significantly increased methylation in DF (3-fold; P<0.05). Multiple sequence alignment of human, mouse, cow, and pig revealed 1 highly conserved region of MEF2 binding sequence in the KLF4 gene promoter (Figure 6E); for swine, this site was −741/−686. Bisulfite pyrosequencing demonstrated hypermethylation of CpG sites inside and flanking the MEF2 binding sequence (−710, −666, −653, −621) in swine arterial endothelium isolated from DF sites (Figure 6F). These 4 CpG sites were hypomethylated (≈20%) in UF, whereas the methylation level was increased 1.5- to 2.7-fold in DF, consistent with the methylation-specific PCR analysis (Figure 6D). Thus, the swine KLF4 promoter methylation findings in vivo were in agreement with the in vitro HAEC profile of CGIs. Furthermore, the DNA methylation of NOS3 promoter remained unchanged between UF and DF (Figure 6D, right) in agreement with our in vitro HAEC data, despite the presence of a CGI near the TSS of swine NOS3.

    A suggested mechanism of enhanced DNA methylation that is part of the dynamic regulation of KLF4 transcription is outlined in Figure 7. DF, a recapitulation of the in vivo hemodynamic regions susceptible to atherosclerosis, induced KLF4 promoter hypermethylation of cytosine. The broad-acting DNMT inhibitors RG108 and 5-Aza suggested that DNMT balance is changed by DF. Our data show that the promoter becomes enriched in methylation enzyme DNMT3A, whereas the demethylation side of the equilibrium, reflected by TET1, remains unchanged. The resulting hypermethylation of KLF4 promoter induced gene silencing by preventing the chromatin binding of MEF2 to the KLF4 promoter.

    Figure 7.

    Figure 7. Summary schematic of Kruppel-like factor 4 (KLF4) promoter methylation mechanisms contributing to suppression of transcription. Disturbed flow–induced DNMT3A enrichment of endothelial KLF4 promoter near the transcription start site increased CpG methylation. Hypermethylation prevented myocyte enhancer factor-2 (MEF2) complex binding resulting in inhibition of KLF4 transcription. Decreased KLF4 expression can lower the interaction of KLF4 with its transcription targets independently of their methylation status leading to a proinflammatory, proatherosclerosis phenotype. Intervention by DNMT inhibitors (RG108; 5-Azacytidine) can rescue this pathway. MCP-1 indicates monocyte chemoattractant protein-1; and NOS3, nitric oxide synthase 3.


    As noted in a recent review,21 DNA methylation can be evaluated in different genomic contexts that result in functional purposes that range from cell identity to splicing to dynamic, as well as fixed, transcriptional regulation. Here, we have demonstrated that KLF4 promoter CpG methylation is responsive to different physiological flow profiles of pathological importance to human arterial endothelial function and we present a plausible DNMT3A/MEF2 mechanism for CpG methylation of promoter sequence near the TSS. In vivo patterns of steady-state KLF4 promoter methylation in aortic endothelium from UF and DF regions in swine supported the in vitro interpretation. The characteristics of DF may, therefore, contribute to the atherosusceptibility of regions associated with branches, curvatures, and bifurcations via inducible methylation of the KLF4 promoter that results in its transcriptional suppression with downstream effects on expression of its targets.

    KLF4 has been characterized as an essential transcription factor in the regulation of inflammation and maintenance of a quiescent endothelium. The consequences of enhanced DNA methylation by hemodynamic DF include inhibition of KLF4 expression that removes a degree of protection against the proinflammatory pathways that lead to atherogenesis. Among its targets is NOS3, which is also suppressed in regions of DF.

    Although histone modification has been reported to be involved in shear stress–induced NOS3 transcription,31 DF did not affect methylation, hydroxymethylation, and chromatin-enrichment of DNMT3A and TET1 in promoter regions of human NOS3. However, we noted that DF also failed to methylate a CGI proximal to the TSS of swine NOS3 promoter suggesting resistance of NOS3 to hypermethylation. This is broadly consistent with the report by Chan et al32 who noted the resilience of the human NOS3 gene to hypermethylation, attributing it to a critical determinant of endothelial cell identity established during development. The same study32 established the principle of a role for DNA methylation in NOS3 transcription by demonstrating reduced NOS3 promoter activity after in vitro transfection of methylated promoter/reporter constructs. However, as demonstrated in this study, neither human nor swine NOS3 promoter methylation was directly affected by DF and it seems that at least part of the NOS3 downregulation may be secondary to DNMT-sensitive DNA methylation effects on upstream transcription factors including, but not necessarily exclusive to, KLF4.

    DF-induced KLF4 hypermethylation in HAEC also inhibited the expression of thrombomodulin and upregulated MCP-1 (which is normally inhibited by KLF4).33 All 3 genes are KLF4 targets that contain or lack CGIs in their promoters; their responses reflect a proinflammmatory pattern we previously observed in HAEC when siRNA was used to knockdown KLF4.20

    The observation that hemodynamic forces are capable of changing the methylation of KLF4, an important transcription factor for flow-sensitive genes,13 sheds new light on how physical factors such as blood flow influence gene expression and disease susceptibility. By translating and integrating the effects of biochemical and biomechanical stimuli, KLF4 coordinates atheroprotective and atheroprone gene expression.13 Together with other epigenetic mechanisms that relate to endothelial flow responses, eg, chromatin and miRNA regulation, we show that DNA methylation is a potent contributor to the mechanistic link between the genome and environment that is important in the spatial distribution of atherogenesis. Indeed, our earlier study on the regulation of KLF4 gene expression by miRNA-92a suggests that KLF4 is tightly regulated by multiple flow-related mechanisms.20 The 50% rescue of KLF4 spliced mature mRNA by DNMT inhibitors, whereas the recovery of pre-mRNA expression was complete (Figure 4), is consistent with KLF4 regulation by intronic miRNA mechanisms,20 as well as by promoter methylation contributions.

    Many experimental studies have reported the effect of undisturbed laminar flow on endothelial responses when referenced to no-flow. However, extrapolation of no-flow comparisons to physiological arterial flow is problematic because of the constantly changing blood velocity/shear stress throughout each cardiac cycle. In the present study, pulsatile UF and DF, recapitulating dominant dynamic characteristics of in vivo arterial flow, were directly compared. DF and UF results were normalized to no-flow controls in each set of experiments only to control for unknown criteria unrelated to flow. The epigenomic measurements were conducted after 48 hours of in vitro flow to establish a degree of cell adaptation to the flow to better match the steady-state in vivo environment. This time window avoided the first 24 hours of flow (UF or DF) when transient changes in gene expression occur caused by the shift from no-flow. At 48 hours, much of this activity has subsided. In support of this, choice is the correlative evidence for a similar pathway in vivo in site-specific swine aortic endothelial cells (Figure 6); these cells are in an adapted steady state at the time of harvest.

    The association of hypermethylation of KLF4 promoter and downregulation of KLF4 have been reported in lymphoma and epithelial tumors where KLF4 typically functions as a tumor suppressor.34,35 In the cardiovascular system, the association of genomic DNA methylation with cardiovascular diseases has been noted in peripheral blood mononuclear cells of hyperhomocysteinemia patients and in the whole aorta tissue of apoE-null mice.36,37 In mammals, DNMTs use S-adenosyl methionine as a methyl group donor for DNA methylation. Hyperhomocysteinemia and the subsequent decreased production or bioavailability of S-adenosyl methionine is associated with an increased risk of cardiovascular disease.36 Furthermore, atheroprone apoE-null mice show changes in DNA methylation patterns in the whole aorta before the appearance of histologically detectable vascular lesion.37 In contrast, our work reveals DNA methylation plasticity in response to hemodynamics representative of atherosusceptible and protected locations in arteries. It demonstrates the first in vitro hypermethylation by spatially differential hemodynamic force characteristics in endothelial cells with strong corroborative evidence in vivo.

    In endothelium exposed to DF (atherosusceptible), the kinetics of signaling pathways (KLFs, proinflammatory molecules; endoplasmic reticulum stress/unfolded protein response, etc) are set differently resulting in differential phenotype expression.17,20,38 Indeed, we consider DF itself as a risk factor in sensitizing the cell to pathological change and the basis of atherosusceptibility. Yet, although these may be biomarkers of susceptibility, they are also adaptive responses and no overt pathology is evident. We suggest that atherogenesis is a 2-hit initiation process where the addition of a second risk factor such as hypercholesterolemia stress may be required to initiate inflammatory pathological change.

    The dynamic nature of DNA methylation and demethylation may offer opportunities for therapeutic intervention. Unlike DNA mutations, DNA methylation abnormalities are reversible by drugs in a laboratory setting and this reversal allows cancer cells to reactivate the silenced genes and produce normal proteins. In the present study, the specific DNMT inhibitors 5-Aza and RG108 could both rescue DF-suppressed atheroprotective gene expression and negatively regulate DF-induced atherosusceptible genes. 5-Aza has been approved (as Vidaza) by the US Food and Drug Administration for the treatment of myelodysplastic syndrome, a preleukemic bone marrow disorder, by inhibiting DNA methylation and cell proliferation.

    The physicochemical mechanism(s) by which the endothelium distinguishes between UF and DF is unclear and may reside in subcellular spatiotemporal mechanotransduction criteria.3,39 Transmission of flow-related deformation forces throughout the cytoskeleton is a plausible mechanical link to the nuclear membrane, the mechanics of which may influence gene regulation.40 Furthermore, the local redox environment near the cell surface may influence DNA methylation via the presence of reactive oxygen species that are significantly elevated in endothelium at atherosusceptible (DF) sites in normal swine.38 In response to reactive oxygen species, the CpG-rich KLF4 promoter may recruit the silencing complex (DNMTs, SIRT1, and polycomb members).41 Because nuclear factor-kB pathway is also more active in DF17 and Rel/p65 could recruit DNMTs to specific genome loci,42 these pathways may influence DNMT3A enrichment. As reflected in Figure 7, mechanotransduction experiments aimed at the induction of DNMT3A may be a fruitful avenue of investigation.

    The present studies, in which flow characteristics are the principal experimental variable leading to phenotype adaptation and increased susceptibility to atherosclerosis, were conducted in normal human cells and normocholesterolemic swine. The added introduction of hypercholesterolemia to initiate atherogenesis and associated cytokine-stimulated inflammatory responses will facilitate further evaluation of the epigenetic and epigenomic regulation of endothelial phenotype adaptation during early pathological change.

    Nonstandard Abbreviations and Acronyms




    CpG island


    disturbed flow


    DNA methyltransferase


    human aortic endothelial cell


    Kruppel-like factor 4


    monocyte chemoattractant protein-1


    myocyte enhancer factor-2


    nitric oxide synthase 3


    Tet methylcytosine dioxygenase


    tumor necrosis factor α


    transcription start site


    undisturbed flow


    We thank Drs Diamond, Manduchi, and Stoeckert of the University of Pennsylvania for discussions.


    In March 2014, the average time from submission to first decision for all original research papers submitted to Circulation Research was 12.63 days.

    The online-only Data Supplement is available with this article at

    Correspondence to Peter F. Davies, PhD, Robinette Foundation Professor of Cardiovascular Medicine, Perelman School of Medicine, University of Pennsylvania, 1010 Vagelos Bldg, 3340 Smith Walk, Philadelphia, PA 19104. E-mail


    • 1. Lusis AJ. Genetics of atherosclerosis.Trends Genet. 2012; 28:267–275.CrossrefMedlineGoogle Scholar
    • 2. Davies PF. Endothelial transcriptome profiles in vivo in complex arterial flow fields.Ann Biomed Eng. 2008; 36:563–570.CrossrefMedlineGoogle Scholar
    • 3. Davies PF. Hemodynamic shear stress and the endothelium in cardiovascular pathophysiology.Nat Clin Pract Cardiovasc Med. 2009; 6:16–26.CrossrefMedlineGoogle Scholar
    • 4. Passerini AG, Polacek DC, Shi C, Francesco NM, Manduchi E, Grant GR, Pritchard WF, Powell S, Chang GY, Stoeckert CJ, Davies PF. Coexisting proinflammatory and antioxidative endothelial transcription profiles in a disturbed flow region of the adult porcine aorta.Proc Natl Acad Sci U S A. 2004; 101:2482–2487.CrossrefMedlineGoogle Scholar
    • 5. Ni CW, Qiu H, Rezvan A, Kwon K, Nam D, Son DJ, Visvader JE, Jo H. Discovery of novel mechanosensitive genes in vivo using mouse carotid artery endothelium exposed to disturbed flow.Blood. 2010; 116:e66–e73.CrossrefMedlineGoogle Scholar
    • 6. Cheng C, van Haperen R, de Waard M, van Damme LC, Tempel D, Hanemaaijer L, van Cappellen GW, Bos J, Slager CJ, Duncker DJ, van der Steen AF, de Crom R, Krams R. Shear stress affects the intracellular distribution of eNOS: direct demonstration by a novel in vivo technique.Blood. 2005; 106:3691–3698.CrossrefMedlineGoogle Scholar
    • 7. Davies PF, Civelek M, Fang Y, Fleming I. The atherosusceptible endothelium: endothelial phenotypes in complex haemodynamic shear stress regions in vivo.Cardiovasc Res. 2013; 99:315–327.CrossrefMedlineGoogle Scholar
    • 8. Won D, Zhu SN, Chen M, Teichert AM, Fish JE, Matouk CC, Bonert M, Ojha M, Marsden PA, Cybulsky MI. Relative reduction of endothelial nitric-oxide synthase expression and transcription in atherosclerosis-prone regions of the mouse aorta and in an in vitro model of disturbed flow.Am J Pathol. 2007; 171:1691–1704.CrossrefMedlineGoogle Scholar
    • 9. Kawashima S, Yokoyama M. Dysfunction of endothelial nitric oxide synthase and atherosclerosis.Arterioscler Thromb Vasc Biol. 2004; 24:998–1005.LinkGoogle Scholar
    • 10. Zhou G, Hamik A, Nayak L, et al. Endothelial Kruppel-like factor 4 protects against atherothrombosis in mice.J Clin Invest. 2012; 122:4727–4731.CrossrefMedlineGoogle Scholar
    • 11. Davis ME, Grumbach IM, Fukai T, Cutchins A, Harrison DG. Shear stress regulates endothelial nitric-oxide synthase promoter activity through nuclear factor kappaB binding.J Biol Chem. 2004; 279:163–168.CrossrefMedlineGoogle Scholar
    • 12. Lin Z, Kumar A, SenBanerjee S, Staniszewski K, Parmar K, Vaughan DE, Gimbrone MA, Balasubramanian V, García-Cardeña G, Jain MK. Kruppel-like factor 2 (KLF2) regulates endothelial thrombotic function.Circ Res. 2005; 96:e48–e57.LinkGoogle Scholar
    • 13. Hamik A, Lin Z, Kumar A, Balcells M, Sinha S, Katz J, Feinberg MW, Gerzsten RE, Edelman ER, Jain MK. Kruppel-like factor 4 regulates endothelial inflammation.J Biol Chem. 2007; 282:13769–13779.CrossrefMedlineGoogle Scholar
    • 14. Villarreal G, Zhang Y, Larman HB, Gracia-Sancho J, Koo A, García-Cardeña G. Defining the regulation of KLF4 expression and its downstream transcriptional targets in vascular endothelial cells.Biochem Biophys Res Commun. 2010; 391:984–989.CrossrefMedlineGoogle Scholar
    • 15. Kuhlencordt PJ, Gyurko R, Han F, Scherrer-Crosbie M, Aretz TH, Hajjar R, Picard MH, Huang PL. Accelerated atherosclerosis, aortic aneurysm formation, and ischemic heart disease in apolipoprotein E/endothelial nitric oxide synthase double-knockout mice.Circulation. 2001; 104:448–454.LinkGoogle Scholar
    • 16. Illi B, Nanni S, Scopece A, Farsetti A, Biglioli P, Capogrossi MC, Gaetano C. Shear stress-mediated chromatin remodeling provides molecular basis for flow-dependent regulation of gene expression.Circ Res. 2003; 93:155–161.LinkGoogle Scholar
    • 17. Fang Y, Shi C, Manduchi E, Civelek M, Davies PF. MicroRNA-10a regulation of proinflammatory phenotype in athero-susceptible endothelium in vivo and in vitro.Proc Natl Acad Sci U S A. 2010; 107:13450–13455.CrossrefMedlineGoogle Scholar
    • 18. Kumar A, Lin Z, SenBanerjee S, Jain MK. Tumor necrosis factor alpha-mediated reduction of KLF2 is due to inhibition of MEF2 by NF-kappaB and histone deacetylases.Mol Cell Biol. 2005; 25:5893–5903.CrossrefMedlineGoogle Scholar
    • 19. Wang W, Ha CH, Jhun BS, Wong C, Jain MK, Jin ZG. Fluid shear stress stimulates phosphorylation-dependent nuclear export of HDAC5 and mediates expression of KLF2 and eNOS.Blood. 2010; 115:2971–2979.CrossrefMedlineGoogle Scholar
    • 20. Fang Y, Davies PF. Site-specific microRNA-92a regulation of Kruppel-like factors 4 and 2 in atherosusceptible endothelium.Arterioscler Thromb Vasc Biol. 2012; 32:979–987.LinkGoogle Scholar
    • 21. Jones PA. Functions of DNA methylation: islands, start sites, gene bodies and beyond.Nat Rev Genet. 2012; 13:484–492.CrossrefMedlineGoogle Scholar
    • 22. Ordovás JM, Smith CE. Epigenetics and cardiovascular disease.Nat Rev Cardiol. 2010; 7:510–519.CrossrefMedlineGoogle Scholar
    • 23. Bhutani N, Burns DM, Blau HM. DNA demethylation dynamics.Cell. 2011; 146:866–872.CrossrefMedlineGoogle Scholar
    • 24. Shirodkar AV, Marsden PA. Epigenetics in cardiovascular disease.Curr Opin Cardiol. 2011; 26:209–215.CrossrefMedlineGoogle Scholar
    • 25. Jurkowska RZ, Jurkowski TP, Jeltsch A. Structure and function of mammalian DNA methyltransferases.Chembiochem. 2011; 12:206–222.CrossrefMedlineGoogle Scholar
    • 26. Jiménez JM, Prasad V, Yu MD, Kampmeyer CP, Kaakour AH, Wang PJ, Maloney SF, Wright N, Johnston I, Jiang YZ, Davies PF. Macro- and microscale variables regulate stent haemodynamics, fibrin deposition and thrombomodulin expression.J R Soc Interface. 2014; 11:20131079.CrossrefMedlineGoogle Scholar
    • 27. Sérandour AA, Avner S, Oger F, et al. Dynamic hydroxymethylation of deoxyribonucleic acid marks differentiation-associated enhancers.Nucleic Acids Res. 2012; 40:8255–8265.CrossrefMedlineGoogle Scholar
    • 28. Huang Y, Pastor WA, Shen Y, Tahiliani M, Liu DR, Rao A. The behaviour of 5-hydroxymethylcytosine in bisulfite sequencing.PLoS One. 2010; 5:e8888.CrossrefMedlineGoogle Scholar
    • 29. Ling Y, Sankpal UT, Robertson AK, McNally JG, Karpova T, Robertson KD. Modification of de novo DNA methyltransferase 3a (Dnmt3a) by SUMO-1 modulates its interaction with histone deacetylases (HDACs) and its capacity to repress transcription.Nucleic Acids Res. 2004; 32:598–610.CrossrefMedlineGoogle Scholar
    • 30. Brueckner B, Garcia Boy R, Siedlecki P, Musch T, Kliem HC, Zielenkiewicz P, Suhai S, Wiessler M, Lyko F. Epigenetic reactivation of tumor suppressor genes by a novel small-molecule inhibitor of human DNA methyltransferases.Cancer Res. 2005; 65:6305–6311.CrossrefMedlineGoogle Scholar
    • 31. Chen W, Bacanamwo M, Harrison DG. Activation of p300 histone acetyltransferase activity is an early endothelial response to laminar shear stress and is essential for stimulation of endothelial nitric-oxide synthase mRNA transcription.J Biol Chem. 2008; 283:16293–16298.CrossrefMedlineGoogle Scholar
    • 32. Chan Y, Fish JE, D’Abreo C, Lin S, Robb GB, Teichert AM, Karantzoulis-Fegaras F, Keightley A, Steer BM, Marsden PA. The cell-specific expression of endothelial nitric-oxide synthase: a role for DNA methylation.J Biol Chem. 2004; 279:35087–35100.CrossrefMedlineGoogle Scholar
    • 33. Shen B, Smith RS, Hsu YT, Chao L, Chao J. Kruppel-like factor 4 is a novel mediator of Kallistatin in inhibiting endothelial inflammation via increased nitric-oxide synthase expression.J Biol Chem. 2009; 284:35471–35478.CrossrefMedlineGoogle Scholar
    • 34. Yasunaga J, Taniguchi Y, Nosaka K, Yoshida M, Satou Y, Sakai T, Mitsuya H, Matsuoka M. Identification of aberrantly methylated genes in association with adult T-cell leukemia.Cancer Res. 2004; 64:6002–6009.CrossrefMedlineGoogle Scholar
    • 35. Guan H, Xie L, Leithäuser F, Flossbach L, Möller P, Wirth T, Ushmorov A. KLF4 is a tumor suppressor in B-cell non-Hodgkin lymphoma and in classic Hodgkin lymphoma.Blood. 2010; 116:1469–1478.CrossrefMedlineGoogle Scholar
    • 36. Ulrey CL, Liu L, Andrews LG, Tollefsbol TO. The impact of metabolism on DNA methylation.Hum Mol Genet. 2005; 14:R139–R147.CrossrefMedlineGoogle Scholar
    • 37. Lund G, Andersson L, Lauria M, Lindholm M, Fraga MF, Villar-Garea A, Ballestar E, Esteller M, Zaina S. DNA methylation polymorphisms precede any histological sign of atherosclerosis in mice lacking apolipoprotein E.J Biol Chem. 2004; 279:29147–29154.CrossrefMedlineGoogle Scholar
    • 38. Civelek M, Manduchi E, Riley RJ, Stoeckert CJ, Davies PF. Coronary artery endothelial transcriptome in vivo: identification of endoplasmic reticulum stress and enhanced reactive oxygen species by gene connectivity network analysis.Circ Cardiovasc Genet. 2011; 4:243–252.LinkGoogle Scholar
    • 39. Davies PF, Helmke BP. Endothelial Mechaotransduction.In: , Mofrad RK, Kamm RD, eds. Cellular Mechanotransduction: Diverse Perspectives from Molecules to Tissue. Chapter 2. Cambridge University Press: Cambridge, UK; 2010:20–60.Google Scholar
    • 40. Gieni RS, Hendzel MJ. Mechanotransduction from the ECM to the genome: are the pieces now in place?J Cell Biochem. 2008; 104:1964–1987.CrossrefMedlineGoogle Scholar
    • 41. O’Hagan HM, Wang W, Sen S, Destefano Shields C, Lee SS, Zhang YW, Clements EG, Cai Y, Van Neste L, Easwaran H, Casero RA, Sears CL, Baylin SB. Oxidative damage targets complexes containing DNA methyltransferases, SIRT1, and polycomb members to promoter CpG Islands.Cancer Cell. 2011; 20:606–619.CrossrefMedlineGoogle Scholar
    • 42. Liu Y, Mayo MW, Nagji AS, Smith PW, Ramsey CS, Li D, Jones DR. Phosphorylation of RelA/p65 promotes DNMT-1 recruitment to chromatin and represses transcription of the tumor metastasis suppressor gene BRMS1.Oncogene. 2012; 31:1143–1154.CrossrefMedlineGoogle Scholar

    Novelty and Significance

    What Is Known?

    • Atherosclerotic plaques develop preferentially in regions of disturbed flow (DF).

    • Irreversible epigenetic DNA methylation patterns are established during development and cell differentiation.

    • Suppression of endothelial expression of the key atheroprotective transcription factor Kruppel-like factor 4 (KLF4) predisposes to atherogenesis. DNA hypermethylation is proposed as a potent inhibitor of KLF4 transcription.

    What New Information Does This Article Contribute?

    • DF induces DNA methylation of CpG islands in the KLF4 promoter.

    • Hypermethylation of KLF4 promoter is due to an imbalance in methylation/demethylation activities involving competition for myocyte enhancer factor-2 binding sites.

    • Change in the biophysical environment can influence pretranscriptional endothelial gene expression, a mechanism that may represent epigenomic adaptive physiological regulation.

    The distribution of atherosclerotic lesions has been linked to arterial branches, bifurcations, and curvatures where the blood vessel geometry causes the flow to create vortices and eddies, referred to as DF. In the absence of disease at these sites, the endothelium nevertheless expresses proinflammatory genes that collectively sensitize it to the initiation of atherosclerosis. These cells are, therefore, considered to be atherosusceptible, a prepathological state of stress adaptation to the local hemodynamic environment. Regulatory mechanisms that link DF to the susceptible endothelial phenotype include inhibition of KLF4 transcription and translation by DF. KLF4 is suppressed in endothelium of regions with DF in vivo where we found DNA hypermethylation of its promoter, the fingerprint of a potent epigenetic suppression mechanism. Using human aortic endothelial cells and controlled flow environments in vitro, we show that DF-induced KLF4 hypermethylation to be a dynamic epigenomic response with proatherogenic consequences. This analysis of flow-related epigenomic plasticity of an atheroprotective gene may be representative of a broader adaptive epigenomic response to environmental conditions, some of which may be truly epigenetic (inheritable).


    eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

    Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.