Biochimica et Biophysica Acta 1851 (2015) 1317–1326
Contents lists available at ScienceDirect
Biochimica et Biophysica Acta
journal homepage: www.elsevier.com/locate/bbalip
Metabolomic analysis revealed the role of DNA methylation in the
balance of arachidonic acid metabolism and endothelial activation
Shan-Shan Xue a,1, Jin-Long He a,1, Xu Zhang a, Ya-Jin Liu a, Feng-Xia Xue b, Chun-Jiong Wang a,
Ding Ai a,⁎, Yi Zhu a,⁎
a
b
Department of Physiology and Pathophysiology, Collaborative Innovation Center of Tianjin for Medical Epigenetics, Tianjin Medical University, Tianjin 300070, China
Department of Gynecology and Obstetrics, Tianjin Medical University General Hospital, Tianjin 300052, China
a r t i c l e
i n f o
Article history:
Received 17 March 2015
Received in revised form 19 June 2015
Accepted 6 July 2015
Available online 11 July 2015
Keywords:
DNA methylation
Arachidonic acid metabolism
Metabolomics
Endothelial cell
a b s t r a c t
Arachidonic acid (AA) metabolism plays an important role in vascular homeostasis. We reported that DNA hypomethylation of EPHX2 induced a pro-inflammatory response in vascular endothelial cells (ECs). However, the
change in the whole AA metabolism by DNA methylation is still unknown. Using a metabolomic approach, we
investigated the effect of DNA methylation on the balance of AA metabolism and the underlying mechanism.
ECs were treated with a DNA methyltransferase inhibitor 5-aza-2′-deoxycytidine (5-AZA), and AA metabolic profiles were analyzed. Levels of prostaglandin D2 (PGD2) and thromboxane B2 (TXB2), metabolites in the cyclooxygenase (COX) pathway, were significantly increased by 5-AZA treatment in ECs resulting from the induction of
PGD2 synthase (PTGDS) and thromboxane A synthase 1 (TBXAS1) expression by DNA hypomethylation.
This phenomenon was also observed in liver and kidney cell lines, indicating a universal mechanism.
Pathophysiologically, homocysteine, known to cause DNA demethylation, induced a similar pattern of the change
of AA metabolism. Furthermore, 5-AZA activated ECs, as evidenced by the upregulation of adhesion molecules.
Indomethacin, a COX inhibitor, reversed the effects of 5-AZA on the levels of PGD2 and TXB2, EC activation and
monocyte adhesion. In vivo, the plasma levels of PGD2 and TXB2 and the expression of In vivo PTGDS and
TBXAS1 as well as adhesion molecules were increased in the aorta of the mice injected with 5-AZA. In conclusion,
using a metabolomic approach, our study uncovered that DNA demethylation increased AA metabolites PGD2
and TXB2 by upregulating the expression of the corresponding enzymes, which might contribute to the DNA
hypomethylation-induced endothelial activation.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
Accumulating evidence has strongly suggested that arachidonic
acid (AA) metabolism plays an important role in cardiovascular
diseases. AA, catalyzed by phospholipase A2 from membrane phospholipids, can be metabolized to hundreds of metabolites by three
pathways: cyclooxygenase (COX), lipoxygenase (LOX), and cytochrome P450 (CYP). Hundreds of biological active metabolites of
AA are related to the progression of diseases. Prostaglandins (PGs)
Abbreviations: AA, arachidonic acid; COX, cyclooxygenases; LOX, lipoxygenases; CYP,
cytochrome P450; EETs, epoxyeicosatrienoic acids; HETEs, hydroxyeicosatetraenoic
acids; PGs, prostaglandins; PGD2, prostaglandin D2; PTGDS, prostaglandin D2 synthase;
TXB2, thromboxane B2; TBXAS1, thromboxane A synthase 1; sEH, soluble epoxide hydrolase; VCAM-1, vascular cell adhesion molecule 1; ICAM-1, intercellular adhesion molecule
1; ECs, endothelial cells; HUVECs, human umbilical vein endothelial cells; Hcy, homocysteine; VSMCs, vascular smooth muscle cells; 5-AZA, 5-aza-2′-deoxycytidine; IND, indomethacin; MSP, methylation-specific PCR; BSP, bisulfite sequencing PCR.
⁎ Corresponding authors.
E-mail addresses: [email protected] (D. Ai), [email protected] (Y. Zhu).
1
Contributed equally to this study.
http://dx.doi.org/10.1016/j.bbalip.2015.07.001
1388-1981/© 2015 Elsevier B.V. All rights reserved.
and thromboxanes (TXs) are crucial bioactive molecules derived
from the COX pathway and subsequent PG synthase that are implicated in inflammation, fever and pain. Hydroxyeicosatetraenoic
acids (HETEs), metabolites in the LOX pathway, have been implicated
in numerous biological processes, such as angiogenesis, platelet activation and asthma. CYP enzymes metabolize AA to multiple products including epoxyeicosatrienoic acids (EETs) and ω-hydrolased 20-HETE.
EETs act as anti-hypertensive and anti-atherosclerotic mediators for
vasculature and are degraded by soluble epoxide hydrolase (sEH),
whereas 20-HETE is a pro-inflammatory and pro-fibrotic eicosanoid.
Considering that various metabolites of AA play different roles in cardiovascular diseases [1], the metabolic profiling of AA during important
physical and pathophysiological processes is needed.
Gene expression is regulated epigenetically by histone modification,
DNA methylation, and microRNAs. The mechanism of DNA methylation
is well known: the cytosine of CpG dinucleotide adds a methyl to form
5-methylcytosine, which results in changed expression. DNA methylation is an essential component of normal development and transcriptional regulation, while aberrant patterns of DNA methylation are
associated with a number of pathological conditions and diseases,
1318
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
such as tumor formation, inflammation and cardiovascular diseases
[2–7]. Since hypomethylated extracellular superoxide dismutase was
found associated with atherosclerosis in 1999 [8], DNA methylation
has been implicated as a novel risk factor of cardiovascular disease [9].
In view of the recent findings of global or promoter-aberrant DNA
methylation in atherosclerosis [10] and the use of DNA methyltransferase inhibitor 5-aza-2′-deoxycytidine (5-AZA), an FDA-approved drug,
reducing atherosclerotic lesion formation and endothelial activation
[11], DNA methylation has been closely associated with the development of atherosclerosis. A genome-wide analysis revealed the hypomethylation of chromosomal DNA predominates in atherosclerotic
plaques, and two-thirds of genes were up-regulated in comparison to
healthy mammary arteries [3]. It was also reported that the atherosclerotic aortas had a tendency for decreased 5-methylcytosine content in
atherosclerotic aortas [8]. Further, DNA methylation polymorphisms
preceded any histological sign of atherosclerosis in ApoE knockout
mice [12]. Our previous work showed that homocysteine (Hcy) upregulated platelet-derived growth factor level via DNA demethylation in
endothelial cells (ECs), which affected cross-talk between ECs and vascular smooth muscle cells (VSMCs) and led to VSMC dysfunction [13].
Moreover, DNA demethylation may coordinately contribute to Hcyinduced sEH upregulation and EC activation [14], which implies that
DNA methylation associated with AA metabolism. However, whether
other enzymes involved in AA metabolism can be methylated, which
may result in changed AA metabolic profile and in turn, endothelial dysfunction is unknown. A highly specific approach to determine the AA
metabolic profile could be valuable.
As a core analytical technique of metabolomics and lipidomics, mass
spectrometry (MS) combined with liquid chromatography (LC–MS) is
used to detect biological metabolites at the parts-per-billion (ppb)
level in a “global” or “targeted” manner. LC–MS/MS-based targeted
metabolomics is a powerful and accurate tool to explore metabolic profiling under aberrant DNA methylation conditions in certain diseases
[15]. Aebersold and colleagues revealed the metabolism of AA with a
genomics–proteomics–metabolomics approach but did not focus on
the balance in the three pathways nor examine the metabolic change
related to the epigenetic regulation of genes [16].
In the present study, we used an LC–MS/MS-based metabolomic
approach and data analysis to investigate the AA metabolic profile
with 5-AZA treatment to elucidate the effect of DNA methylation
on the balance of AA metabolism in vitro and in vivo. Certain AA
metabolism-related enzymes, such as prostaglandin D2 synthase
(PTGDS) and thromboxane A synthase 1 (TBXAS1), were epigenetically regulated by DNA demethylation, which suppressed gene silencing. Accordingly, levels of their corresponding products, PGD2
and TXB2, were increased in parallel with the upregulation of vascular adhesion molecule 1 (VCAM-1) and intercellular adhesion molecule 1 (ICAM-1) in ECs. This work shows the crosstalk between DNA
methylation and AA metabolism in vitro and in vivo and reveals aberrant AA metabolism induced by DNA demethylation involved in EC
dysfunction.
from Waters Co. (Milford, MA). Centrifuge tube filters were from
Corning Co. (Corning, NY). All other chemical reagents were from
Sigma.
2.2. Cell culture
Human umbilical vein ECs (HUVECs) were isolated and maintained as described [17]. Experiments were performed with HUVECs
between 4 and 6 passages. Human liver carcinoma HepG2 and
human embryonic kidney 293T (HEK293T) cell lines were cultured
as described [18]. All cells were cultured at 37 °C in a humidified atmosphere containing 5% CO 2 . For experiments, subconfluent ECs
were treated with the DNA methyltransferase inhibitor 5-AZA or
COX inhibitor IND as indicated.
2.3. Metabolomic analysis
The method of metabolite detection in cell lysates, cultured medium or plasma and tissues from mice was described in our previous
work [19]. The tissues of the aorta, liver or kidney were homogenized
before lipid extraction. HUVECs were lysed by repeated freeze–
thawing cycles and then pre-processed in the solvent of methanol.
After centrifugation, the supernatant was extracted by ethyl acetate
twice, and then the upper organic phase was evaporated. The residue
was dissolved in 100 μL 30% acetonitrile. The subsequent procedure
involved ultra-high-performance liquid chromatography (Waters,
Milford, MA) with a 5500 QTRAP hybrid triple-quadruple linear ion
trap mass spectrometer (AB Sciex, Foster City, CA) equipped with a
Turbo Ion Spray electrospray ionization source (LC–MS/MS) as
described [19].
2.4. Quantitative RT-PCR (qPCR)
Total RNA was collected from cells or tissues in QIAzol and purified
by the use of the QIAGEN miRNeasy Mini Kit. Total RNA was reverse
transcribed into cDNA with SuperScript III and random primers
(Invitrogen), as described previously [18]. qPCR for specific genes involved the Brilliant II SYBR Green qPCR Master Mix (Stratagene)
with custom-designed primers and the ABI 7900HT Real-Time PCR
System (Life Technologies, CA). Results were normalized to β-actin.
Primer sequences are in Supplemental Table S1.
2.5. Western blot analysis
The western blot was performed as described previously [18]. Cell
lysates were resolved by 12% SDS-PAGE and then transferred to a
PVDF membrane. The membrane was incubated with primary antibodies then horseradish peroxidase-conjugated secondary antibody. The
level of β-actin was also measured as an internal control. The densities
of the bands were quantified and normalized to that of β-actin by the
use of the Scion Image software (Scion Corp., Frederick, MD).
2. Materials and methods
2.6. Animal experiments
2.1. Chemicals and reagents
C57BL/6 male mice, 8 weeks old, fed standard laboratory chow
and tap water ad libitum and bred in a 12-h light/dark cycle, were divided into 2 groups for treatment (control group, n = 13; 5-AZA
group, n = 15). The mice were injected with either phosphate buffered saline (PBS) or 5-AZA (1 mg/kg/d) intraperitoneally for 5 days
and then anesthetized with pentobarbital sodium and sacrificed.
The blood plasma, liver, kidney and aorta were collected. The investigation conformed to the Guide for the Care and Use of Laboratory
Animals by the US National Institutes of Health (NIH Publication
No. 85-23, updated 2011). The animal experimental protocol was approved by the Tianjin Medical University Institutional Animal Care
and Use Committee.
Butylated hydroxytoluene (BHT), 5-AZA, acetic acid and indomethacin (IND) were purchased from Sigma Aldrich Inc. (St.
Louis, MO); pGEM-T easy vector was purchased from Promega
(Madison, WI); 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl (BCECF-AM) was from Invitrogen (Carlsbad,
CA). Antibodies against PTGDS and TBXAS1 were purchased from
Abcam (Cambridge, UK). Mobile-phase acetonitrile (LC–MS grade),
methanol and n-hexane (HPLC grade) were from Merck (Darmstadt,
Germany). Ethyl acetate, formic acid, and glacial acetic acid were from
Fisher Scientific (Pittsburgh, PA). Oasis HLB 10 mg SPE cartridges were
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
2.7. Methylation-specific PCR (MSP) and bisulfite sequencing PCR (BSP)
Bisulfite modification and MSP were performed as described [18].
The relative DNA abundance was quantified by densitometry with use
of NIH ImageJ. PCR products of BSP were gel-purified and cloned into
the pGEM-T easy vector and confirmed by sequencing. The MSP primers
are in Supplemental Table S2.
1319
ECs, we detected the mRNA levels of these genes in HepG2 and
HEK293T cells treated with 5-AZA. The affected genes were also upregulated in both cell lines with 5-AZA treatment, which agreed with the
results in HUVECs (Fig. 2B). These results suggest that DNA demethylation plays an important role in the balance of AA metabolism universally
rather than tissue-specifically in ECs.
3.3. TBXAS1 and PTGDS were demethylated by 5-AZA
2.8. Monocyte adhesion assay
Subconfluent ECs in 6-well plates were treated with the DNA methyltransferase inhibitor 5-AZA or COX inhibitor IND as indicated. THP1
cells were labeled with BCECF-AM and then plated on the 6-well plates
with confluent HUVECs at 2 × 106 cells/well. After incubation with
HUVECs for 30 min at 37 °C, non-adherent cells were removed by washing 3 times with PBS. The numbers of stained adhering cells in 5 random
fields were counted for each group under a fluorescence microscopy.
2.9. Statistical analysis
Data were presented as mean ± SEM and analyzed by Student's t
test between 2 groups or by ANOVA with Bonferroni correction for
multiple group comparisons (GraphPad software, San Diego, CA). All experiments were performed with least 3 independent experiments.
p b 0.05 was considered statistically significant.
3. Results
3.1. AA metabolic profile affected by DNA demethylation in ECs
To elucidate the change of AA metabolite profile, LC–MS/MS was
performed in both cell lysates and cultured media from HUVECs
treated with 5-AZA or vehicle. Long interspersed element-1 sequences (LINE-1) are broadly distributed throughout the human genome and hypermethylated in normal cells. Methylation of LINE-1 is
a good indicator of global DNA methylation. We first used LINE-1
methylation status to validate the effect of 5-AZA. MSP revealed increased unmethylation level of LINE-1 (data not shown), which
reflected that 5-AZA caused genomic DNA hypomethylation in ECs.
Given the notion that balance of AA metabolism reflects enzyme activity
change, we detected the metabolic profiles in cell lysates by AA metabolomics. Orthogonal partial least squares discriminant analysis (OPLSDA) was used to produce a 2D visual summary of the observed variation
in eicosanoid profiles between different samples and revealed different
metabolism patterns in cell lysates from treated and control cells
(Fig. 1A). In 5-AZA-treated ECs, the metabolites in COX pathway such
as PGs and TXB2 were increased compared with those in LOX and CYP
pathways (Fig. 1B and Supplementary Table S3). PGs and TXB2 were especially enriched (Fig. 1C). This pattern suggests that DNA methylation
is important in regulating PG and TXB2 pathways in ECs.
3.2. DNA demethylation regulated the gene expression of AA metabolic
enzymes
To validate the metabolomic data and explore the underlying mechanism, we evaluated the methylation possibility of all genes involved in
AA metabolism by analyzing the content of CpG islands with the bioinformatics software Methyl Primer Express to reveal the DNA methylation site. The genes were analyzed and integrated with the KEGG
pathway (Fig. 2A). The predicted genes with enriched CpG islands are
highlighted by a red box. Furthermore, RT-PCR was used to test the
mRNA level of metabolic enzymes related to the changed metabolites.
As predicted, most genes of metabolic enzymes with CpG islands in
the COX pathway were upregulated, and most genes in LOX and CYP
pathways, except sEH, as we reported previously [18], were mainly suppressed (Fig. 2B). To explore whether the phenomenon was specific to
From the above results, we hypothesized that TBXAS1 and PTGDS
were regulated by DNA methylation and played a vital role in the balance
of AA metabolism. Because TBXAS1 and PTGDS mRNA levels were increased by 5-AZA, we analyzed the CpG islands of these two genes by
Methyl Primer Express. TBXAS1 and PTGDS had rich CpG islands located
on the promoter and the first exon, which suggested their possible
regulation by methylation (Fig. 3A, C). Further analysis by BSP showed
that the status of some CpG islands changed from methylation to
unmethylation with 5-AZA treatment (Fig. 3B, D). Thus, both bioinformatics and methylation analysis supported that TBXAS1 and PTGDS
were upregulated by DNA demethylation, which increased the level of
their products PGD2 and TXB2 and in turn activated ECs in vitro.
3.4. The effect of homocysteine on AA metabolites similar with 5-AZA
High levels of Hcy resulted in DNA hypomethylation of CpG islands
in the promoter region of the genes, causing its upregulation. We previously reported that Hcy-induced DNA hypomethylation induced sEH
and other genes expression and caused EC dysfunctions [14]. To address
the pathological significance of our finding, we investigated the effect of
Hcy on the balance of AA metabolism. As shown in Fig. 4A and B, the AA
metabolic profile in ECs treated with Hcy was similar to that of 5-AZA
treatment: levels of both TXB2 and PGD2 were increased (Fig. 4C).
Consistently, the mRNA levels of TBXAS1 and PTGDS were also upregulated by Hcy treatment (Fig. 4D).
3.5. Metabolites in the COX pathway mediated 5-AZA-activated ECs
We previously reported that Hcy induced DNA hypomethylation
resulting in the upregulation of VCAM-1 and ICAM-1 in ECs [14]. To investigate whether DNA demethylation-activated ECs was mediated by
metabolites of the COX pathway, we treated ECs with the general COX
inhibitor IND for 6 h to block the activity of both COX-1 and COX-2 before treatment with 5-AZA, and then measured the mRNA levels of
VCAM-1 and ICAM-1. The 5-AZA-induced upregulation of VCAM-1 and
ICAM-1 was largely attenuated by pretreatment with IND (Fig. 5A–B).
The monocyte-endothelial cell interaction plays a crucial role in atherogenesis and is mediated by complex interactions between multiple
adhesion molecules, such as VCAM-1 and ICAM-1. Resulting from
increased adhesion molecules, the monocyte adhesion induced by
5-AZA was increased, which was attenuated by IND pretreatment
(Fig. 5C–D). Consistently, the levels of metabolites PGD2 and TXB2
(Fig. 5E–F) and the expression of their corresponding genes PTGDS
and TBXAS1 were increased by 5-AZA, which was reversed by IND
(Fig. 5G–H). Moreover, the protein level of PTGDS and TBXAS1 were
up-regulated by 5-AZA, which was decreased by IND (Fig. 5I–K). Therefore, changes in levels of PGD2 and TXB2 resulted from DNA demethylation of their corresponding enzymes, might be involved in DNA
demethylation induced EC activation.
3.6. DNA demethylation switched AA metabolism toward the COX pathway
in vivo
To investigate the effect of DNA methylation on the AA metabolic
profile in vivo, we treated C57BL/6 mice by intraperitoneal injection
with 5-AZA or vehicle for 5 days. Metabolomics study was used to detect
AA metabolites in the plasma, liver and kidney. The plasma of the same
1320
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
Fig. 1. The level of arachidonic acid (AA) metabolites of the cyclooxygenase (COX) pathway were increased in human umbilical vein endothelial cells (HUVECs) treated with 5-aza-2′deoxycytidine (5-AZA). LC–MS/MS revealed metabolites in HUVECs with 4 μM 5-AZA treatment for 72 h. (A) Orthogonal partial least squares discriminant analysis (OPLS-DA); the red
plot represents the 5-AZA group and the black plot represents the control group. (B–C) Changes in the levels of arachidonic acid (AA) metabolites in cell lysates of HUVECs. (B) Red
and green represent increased and decreased levels, respectively, by Log2 analysis (fold of change, treated vs. control). (C) The concentration of metabolites in the COX pathway in cell
lysates. Data are mean ± SEM (n = 3). *, p b 0.05 compared with control; N.D.: not detectable.
treated group was pooled for LC–MS/MS detection. Levels of some COX
pathway metabolites were higher in the 5-AZA-treated group
compared to those in the control group (Fig. 6A). Consistent with the
profiles in the plasma, liver (Fig. 6B) and kidney (data not shown)
showed increased levels of some COX pathway metabolites, including
PGF2α, PGD2 and TXB2.
Furthermore, the mRNA levels of key metabolic enzymes, such as
TBXAS1 and PTGDS, were increased both in the aorta and liver of 5AZA-treated mice (Fig. 6C–D), which agrees with in vitro results. To further investigate the pathological significance of DNA demethylation of
AA metabolism-related genes, we tested whether ECs were activated
by 5-AZA treatment. The mRNA levels of both VCAM-1 and ICAM-1,
markers of EC activation, were upregulated in the mouse aorta with 5AZA treatment (Fig. 6C).
4. Discussion
Imbalanced AA metabolism in vasculature may lead to impaired
vascular homeostasis. Our previous studies showed that Hcy
transcriptionally upregulated sEH expression and then reduced
the level of EETs in human ECs via ER stress and DNA demethylation,
which led to endothelial activation [14]. We also found that Hcy upregulated platelet-derived growth factor and telomerase reverse transcriptase
via DNA demethylation in ECs [13,20]. Here, we used a metabolomic approach to explore the role of DNA methylation in AA metabolism and to
search for affected genes in hypomethylation-induced endothelial dysfunction and the underlying mechanism. We found that 1) both 5-AZA
and Hcy treatments increased the levels of COX-pathway metabolites,
specifically PGD2 and TXB2, and the expression of the associated enzymes PTGDS and TBXAS1 in ECs; 2) the DNA methylation of PTGDS
and TBXAS1 was regulated by 5-AZA in ECs; 3) plasma levels of PGD2
and TXB2 and the expression of PTGDS and TBXAS1 in the aorta and
liver in C57BL/6 mice injected with 5-AZA was increased; and 4) 5-AZA
activated ECs, and a COX inhibitor attenuated the effect of 5-AZA on the
increased levels of PGD2 and TXB2, EC activation and monocyte adhesion
on ECs. Thus, we utilized metabolomics to define the role of DNA demethylation in AA metabolites which suggested that DNA demethylationincreased PGD2 and TXB2 might contribute to EC activation.
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
1321
Fig. 2. The metabolic enzymes of COX pathway in ECs were upregulated by 5-AZA. (A) Bioinformatic prediction of AA metabolism pathways with the KEGG metabolic pathway. The red
rectangle represents enzymes with enriched CpG islands, and the blue rectangle represents enzymes without CpG islands. Oval box represents metabolites. (B) RT-PCR analysis of the
mRNA level of AA metabolic enzymes in HUVECs, HepG2 and HEK293T cells. The Y axis shows the Log2 value (relative fold of control). The value above the X axis indicates the changed
levels of gene expression. Data are mean ± SEM (n = 3). *, p b 0.05; **, p b 0.01 compared with control.
Previous study advocated that AA and associated metabolites regulate diverse biological processes such as development, metabolism,
and reproduction by binding and activating retinoid X receptor [21].
Another study demonstrated that altered CYP levels contributed to
vascular and tubular abnormalities in renal disease [22]. Accumulating
evidence suggests that the expression of enzymes involved in AA
metabolism is related to the progression of many diseases, such as
pulmonary artery hypertension and inflammation [23,24]. Given that
balanced levels of AA metabolites regulate cell function and disease
condition, changes in the level of specific AA metabolites may reflect
systematic responses to stimulation.
AA is metabolized in a network manner, with the level of one metabolite usually related with that of others. Thus, the study of metabolites is a powerful tool for characterizing complex phenotypes and
1322
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
Fig. 3. Methylation analysis of demethylated CpG islands of thromboxane A synthase 1 (TBXAS1) and prostaglandin D2 synthase (PTGDS) in HUVECs. (A) Methyl primer express-predicted
CpG islands of TBXAS1. The genomic DNA obtained from treated HUVECs was modified by bisulphate. (B) Bisulfite sequencing PCR (BSP) of the promoter of TBXAS1 in HUVECs treated
with 4 μM 5-AZA for 72 h. (C) Methyl primer express-predicted CpG islands of PTGDS. (D) BSP of the promoter of PTGDS in HUVECs treated with 4 μM 5-AZA for 72 h. All experiments
involved at least 3 independent experiments.
for developing biomarkers for specific physiological responses.
Sabidó et al. reported the quantification of changes in protein levels
by integrated genomic, proteomic, and metabolomic analysis of AA
biosynthetic pathways using RAW264.7 macrophages [16]. Metabolomics is a convenient and systemic approach to study AA metabolism
and has been used to investigate the function of key metabolites
under different conditions and treatments. Using an LC–MS/MS-based
AA metabolomic approach, Hammock and colleagues discovered an unexpected biomarker, 20-HETE, with increased level in murine plasma
for rofecoxib-mediated cardiovascular events [25]. Here, we established
a similar LC–MS/MS method and detected 32 AA metabolites for exploring the effect of DNA methylation on AA metabolism with an integrated
perspective. This is the first use of a systematic approach to establish the
relationship between DNA methylation and AA metabolic balance
in vitro and in vivo.
Patients with atherosclerosis have disturbed global DNA methylation status [26], and we reported that Hcy-activated ECs with the hypomethylation of sEH and, in turn, reduced levels of EETs [14]. Here,
combined with our metabolomic study of 5-AZA-treated human ECs,
we screened the expression of related enzymes predicted by bioinformatic analysis and confirmed the upregulation of sEH and decreased
levels of EETs. As well, we revealed several significant changes in
the metabolite profile with 5-AZA-induced DNA demethylation,
which indicated the switch of AA metabolism toward a COX pathway
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
1323
Fig. 4. The effect of Hcy on the AA metabolic in HUVECs was similar with 5-AZA. HUVECs were starved for 12 h in a serum-free medium, then treated with 200 μM Hcy for 48 h. (A) OPLS-DA
has shown an AA metabolic profile in cell lysates of HUVECs treated with Hcy detected by LC–MS/MS metabolomics;the red plot represents the 5-AZA group and the black plot represents
the control group. (B) Heatmap has shown changes in the levels of arachidonic acid (AA) metabolites in cell lysates of HUVECs. Red and green represent increased or decreased levels,
respectively, by Log2 analysis (fold of change, treated vs. control,). (C) The concentration of metabolites in the COX pathway in cell lysates. (D) RT-PCR analysis of the mRNA levels of
the metabolic enzymes in COX, LOX, and CYP pathways. The Y axis shows the Log2 value (relative fold of control). (*, p b 0.05; **, p b 0.01; N.D.: not detectable). Data are representative
of at least 3 independent experiments.
by eliminating the DNA methylation-mediated suppression of target
enzymes. The DNA hypomethylation induced ECs activation might
be due to elevated levels of PGD2 and TXB2 and the change in levels
of AA metabolites. The results from Hcy-treated ECs further supported the effect of hypomethylation on AA metabolism. Adhesion of
circulating monocytes onto the intimal endothelium is considered
as one of the earliest events of atherosclerosis. Our results demonstrated that 5-AZA upregulated adhesion molecules and increased
monocyte adhesion on ECs, suggesting the involvement of aberrant
DNA demethylation in the atherogenesis.
TBXAS1 is responsible for isomerization of PGH2 into thromboxane
[27]. PGD2 is synthesized by hematopoietic PGD synthase or PTGDS.
Both PTGDS and TBXAS1 are regulated by DNA methylation [28,29].
We found the hypomethylation and upregulation of both enzymes by
5-AZA in different cells and mice, which agrees with previous reports.
Since the whole COX pathway was activated, the COX inhibitor IND
could largely attenuated the effect of 5-AZA on EC activation. IND decreased the level of PGD2 and TXB2 and inhibited the upregulated
expression of adhesion molecules, markers of endothelial activation.
This observation was confirmed in liver and kidney cell lines and in
plasma of C57BL/6 mice injected with 5-AZA. The gene expression pattern was also similar, which suggests that DNA methylation regulated
the balance of AA metabolism universally. However, we did not investigate the pathophysiological significance of the AA profile change in
other organs.
PGs may be heavily involved in atherosclerosis. The stable metabolites thromboxane and prostacyclin are imbalanced in arteriosclerotic
lesions in patients and animal models. An abnormally high level of
TXB2 in plasma was reported in patients with atherosclerosis obliterans
and was positively correlated with serum triglyceride content and vascular injury index [30]. TBXAS1 was found increased in level in blood
vessels and induced vasoconstriction and platelet activation [29].
Antagonist or deletion of the receptor for thromboxane retarded atherogenesis in apolipoprotein E and low-density lipoprotein receptor double
knockout mice [31]. PGD2 has pro- or anti-inflammatory effects in
clinically important pathological conditions [32,33]. PTGDS is highly
1324
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
Fig. 5. Metabolites mediated 5-AZA-induced endothelial activation. HUVECs were starved for 12 h in a serum-free medium, then treated with 4 μM 5-AZA for 72 h with or without pretreatment with the COX inhibitor indomethacin (IND, 5 μM) for 6 h. RT-PCR analysis of the mRNA levels of VCAM-1 (A) and ICAM-1 (B). The adhering THP-1 under a fluorescence microscopy with different treatments (C), the statistics data shown on the right panel (D). LC–MS/MS of the levels of PGD2 (E) and TXB2 (F). RT-PCR analysis of the mRNA levels of PTGDS (G) and
TBXAS1 (H). (I) Western blot analysis of the protein level of human PTGDS and TBXAS1. The statistics result of the protein level of PTGDS in (J) and TBXAS1 in (K). β-Actin was a loading
control. Data are representative of 3 independent experiments. *, p b 0.05; **, p b 0.01; N.D.: not detectable.
expressed in atherosclerotic intima [34,35], and induced by shear stress
via activator protein 1 in ECs [36]. Urade et al. reported an increased
serum pocalin-type PTGDS level associated with the progression of atherosclerosis in patients and suggested that serum PTGDS is a powerful
biomarker for severity of stable coronary artery disease [37]. In this
study, as an independent risk factor of cardiovascular diseases, Hcy
was found to upregulated TBXAS1 and PTGDS and increased levels of
their metabolites in ECs. Thus, imbalanced AA metabolism, with high
levels of TXB2 and PGD2, might play a critical role in the progression
of cardiovascular disease. A limitation of this study is that we did not
verify the effect of Hcy and 5-AZA in a disease model in vivo.
Taken together, DNA demethylation upregulated the expression of
key enzymes in AA metabolism and their corresponding products,
which affected endothelial cell function. We revealed that PTGDS and
TBXAS1 had essential effect on balancing AA metabolism which was
attributed to DNA methylation with 5-AZA treatment. The new study
approach combining metabolomics and bioinformatics to study the
gene regulation function in vitro and in vivo can help in the investigation
of the mechanism of complicated diseases. Furthermore, this model of
study provides a powerful tool for clinical studies to uncover the molecular mechanism of disease and drug development.
Transparency document
The Transparency document associated with this article can be
found, in the online version.
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
1325
Fig. 6. Levels of metabolites and metabolic enzymes of the COX pathway were increased in vivo. C57BL/6 mice were treated with 5-AZA by intraperitoneal injection for 5 days. The plasma
for metabolic analysis was a mixture with the same amount of 13 mice in the control group, and 15 mice in the 5-AZA group. Metabolomic study to detect the AA metabolites in the plasma
(A) and liver (B). Quantitative RT-PCR analysis of the mRNA level of related enzymes and VCAM-1 and ICAM-1 in the aorta (C) and related enzymes in the liver (D). Data are mean ± SEM.
*, p b 0.05; **, p b 0.01 compared with control.
Acknowledgments
This work was supported in part by grants from the Major National
Basic Research Grant of China [No. 2012CB517500] and the National Natural Science Foundation of China [81130002; 81322006;
81370396; 81400320]. Zhu Y is an investigator in the Collaborative
Innovation Center of Cardiovascular Research in Beijing.
[3]
[4]
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.bbalip.2015.07.001.
[5]
[6]
References
[1] G.J. Dusting, S. Moncada, J.R. Vane, Prostaglandins, their intermediates and precursors: cardiovascular actions and regulatory roles in normal and abnormal circulatory
systems, Prog. Cardiovasc. Dis. 21 (1979) 405–430.
[2] G. Amabile, A. Di Ruscio, F. Muller, R.S. Welner, H. Yang, A.K. Ebralidze, H. Zhang, E.
Levantini, L. Qi, G. Martinelli, T. Brummelkamp, M.M. Le Beau, M.E. Figueroa, C. Bock,
D.G. Tenen, Dissecting the role of aberrant DNA methylation in human leukaemia,
Nat. Commun. 6 (2015) 7091.
E. Aavik, H. Lumivuori, O. Leppanen, T. Wirth, S.K. Hakkinen, J.H. Brasen, U.
Beschorner, T. Zeller, M. Braspenning, W. van Criekinge, K. Makinen, S. YlaHerttuala, Global DNA methylation analysis of human atherosclerotic plaques reveals extensive genomic hypomethylation and reactivation at imprinted locus
14q32 involving induction of a miRNA cluster, Eur. Heart J. 36 (2015) 993–1000.
P.L. De Jager, G. Srivastava, K. Lunnon, J. Burgess, L.C. Schalkwyk, L. Yu, M.L. Eaton,
B.T. Keenan, J. Ernst, C. McCabe, A. Tang, T. Raj, J. Replogle, W. Brodeur, S. Gabriel,
H.S. Chai, C. Younkin, S.G. Younkin, F. Zou, M. Szyf, C.B. Epstein, J.A. Schneider, B.E.
Bernstein, A. Meissner, N. Ertekin-Taner, L.B. Chibnik, M. Kellis, J. Mill, D.A. Bennett,
Alzheimer's disease: early alterations in brain DNA methylation at ANK1, BIN1,
RHBDF2 and other loci, Nat. Neurosci. 17 (2014) 1156–1163.
P. Sharma, G. Garg, A. Kumar, F. Mohammad, S.R. Kumar, V.S. Tanwar, S. Sati, A.
Sharma, G. Karthikeyan, V. Brahmachari, S. Sengupta, Genome wide DNA methylation profiling for epigenetic alteration in coronary artery disease patients, Gene
541 (2014) 31–40.
K. Nones, N. Waddell, S. Song, A.M. Patch, D. Miller, A. Johns, J. Wu, K.S. Kassahn, D.
Wood, P. Bailey, L. Fink, S. Manning, A.N. Christ, C. Nourse, S. Kazakoff, D. Taylor, C.
Leonard, D.K. Chang, M.D. Jones, M. Thomas, C. Watson, M. Pinese, M. Cowley, I.
Rooman, M. Pajic, G. Butturini, A. Malpaga, V. Corbo, S. Crippa, M. Falconi, G.
Zamboni, P. Castelli, R.T. Lawlor, A.J. Gill, A. Scarpa, J.V. Pearson, A.V. Biankin, S.M.
Grimmond, Genome-wide DNA methylation patterns in pancreatic ductal adenocarcinoma reveal epigenetic deregulation of SLIT-ROBO, ITGA2 and MET signaling, Int. J.
Cancer 135 (2014) 1110–1118.
1326
S.-S. Xue et al. / Biochimica et Biophysica Acta 1851 (2015) 1317–1326
[7] J. Martinez-Galan, B. Torres-Torres, M.I. Nunez, J. Lopez-Penalver, R. Del Moral, J.M.
Ruiz De Almodovar, S. Menjon, A. Concha, C. Chamorro, S. Rios, J.R. Delgado, ESR1
gene promoter region methylation in free circulating DNA and its correlation with
estrogen receptor protein expression in tumor tissue in breast cancer patients,
BMC Cancer 14 (2014) 59.
[8] M.O. Laukkanen, S. Mannermaa, M.O. Hiltunen, S. Aittomaki, K. Airenne, J. Janne, S.
Yla-Herttuala, Local hypomethylation in atherosclerosis found in rabbit ec-sod
gene, Arterioscler. Thromb. Vasc. Biol. 19 (1999) 2171–2178.
[9] C. Schiano, M.T. Vietri, V. Grimaldi, A. Picascia, M.R. Pascale, C. Napoli, Epigeneticrelated therapeutic challenges in cardiovascular disease, Trends Pharmacol. Sci. 36
(2015) 226–235.
[10] W. Perng, E. Villamor, M.R. Shroff, J.A. Nettleton, J.R. Pilsner, Y. Liu, A.V. Diez-Roux,
Dietary intake, plasma homocysteine, and repetitive element DNA methylation in
the Multi-Ethnic Study of Atherosclerosis (MESA), Nutr. Metab. Cardiovasc. Dis. 24
(2014) 614–622.
[11] J. Dunn, H. Qiu, S. Kim, D. Jjingo, R. Hoffman, C.W. Kim, I. Jang, D.J. Son, D. Kim, C. Pan,
Y. Fan, I.K. Jordan, H. Jo, Flow-dependent epigenetic DNA methylation regulates endothelial gene expression and atherosclerosis, J. Clin. Invest. 124 (2014) 3187–3199.
[12] G. Lund, L. Andersson, M. Lauria, M. Lindholm, M.F. Fraga, A. Villar-Garea, E.
Ballestar, M. Esteller, S. Zaina, DNA methylation polymorphisms precede any histological sign of atherosclerosis in mice lacking apolipoprotein E, J. Biol. Chem. 279
(2004) 29147–29154.
[13] D. Zhang, Y. Chen, X. Xie, J. Liu, Q. Wang, W. Kong, Y. Zhu, Homocysteine activates
vascular smooth muscle cells by DNA demethylation of platelet-derived growth factor in endothelial cells, J. Mol. Cell. Cardiol. 53 (2012) 487–496.
[14] D. Zhang, X. Xie, Y. Chen, B.D. Hammock, W. Kong, Y. Zhu, Homocysteine
upregulates soluble epoxide hydrolase in vascular endothelium in vitro and
in vivo, Circ. Res. 110 (2012) 808–817.
[15] A.K. Petersen, S. Zeilinger, G. Kastenmuller, W. Romisch-Margl, M. Brugger, A. Peters,
C. Meisinger, K. Strauch, C. Hengstenberg, P. Pagel, F. Huber, R.P. Mohney, H. Grallert,
T. Illig, J. Adamski, M. Waldenberger, C. Gieger, K. Suhre, Epigenetics meets metabolomics: an epigenome-wide association study with blood serum metabolic traits,
Hum. Mol. Genet. 23 (2014) 534–545.
[16] E. Sabido, O. Quehenberger, Q. Shen, C.Y. Chang, I. Shah, A.M. Armando, A. Andreyev,
O. Vitek, E.A. Dennis, R. Aebersold, Targeted proteomics of the eicosanoid biosynthetic pathway completes an integrated genomics–proteomics–metabolomics picture of cellular metabolism, Mol. Cell. Proteomics 11 (2012) (M111 014746).
[17] Y. Zhu, J.H. Lin, H.L. Liao, O. Friedli Jr., L. Verna, N.W. Marten, D.S. Straus, M.B.
Stemerman, LDL induces transcription factor activator protein-1 in human endothelial cells, Arterioscler. Thromb. Vasc. Biol. 18 (1998) 473–480.
[18] D. Zhang, D. Ai, H. Tanaka, B.D. Hammock, Y. Zhu, DNA methylation of the promoter
of soluble epoxide hydrolase silences its expression by an SP-1-dependent mechanism, Biochim. Biophys. Acta 1799 (2010) 659–667.
[19] L. Li, N. Li, W. Pang, X. Zhang, B.D. Hammock, D. Ai, Y. Zhu, Opposite effects of gene
deficiency and pharmacological inhibition of soluble epoxide hydrolase on cardiac
fibrosis, PLoS One 9 (2014) e94092.
[20] D. Zhang, X. Sun, J. Liu, X. Xie, W. Cui, Y. Zhu, Homocysteine accelerates senescence
of endothelial cells via DNA hypomethylation of human telomerase reverse transcriptase, Arterioscler. Thromb. Vasc. Biol. 35 (2015) 71–78.
[21] J. Lengqvist, A. Mata De Urquiza, A.C. Bergman, T.M. Willson, J. Sjovall, T. Perlmann,
W.J. Griffiths, Polyunsaturated fatty acids including docosahexaenoic and arachidonic acid bind to the retinoid X receptor alpha ligand-binding domain, Mol. Cell.
Proteomics 3 (2004) 692–703.
[22] F. Fan, Y. Muroya, R.J. Roman, Cytochrome P450 eicosanoids in hypertension and
renal disease, Curr. Opin. Nephrol. Hypertens. 24 (2015) 37–46.
[23] G. Fredman, L. Ozcan, S. Spolitu, J. Hellmann, M. Spite, J. Backs, I. Tabas, Resolvin D1
limits 5-lipoxygenase nuclear localization and leukotriene B4 synthesis by inhibiting
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
a calcium-activated kinase pathway, Proc. Natl. Acad. Sci. U. S. A. 111 (2014)
14530–14535.
L. Zhou, Z. Chen, P. Vanderslice, S.P. So, K.H. Ruan, J.T. Willerson, R.A. Dixon,
Endothelial-like progenitor cells engineered to produce prostacyclin rescue
monocrotaline-induced pulmonary arterial hypertension and provide right ventricle benefits, Circulation 128 (2013) 982–994.
J.Y. Liu, N. Li, J. Yang, H. Qiu, D. Ai, N. Chiamvimonvat, Y. Zhu, B.D. Hammock, Metabolic profiling of murine plasma reveals an unexpected biomarker in rofecoxibmediated cardiovascular events, Proc. Natl. Acad. Sci. U. S. A. 107 (2010)
17017–17022.
E. Aavik, H. Lumivuori, O. Leppanen, T. Wirth, S.K. Hakkinen, J.H. Brasen, U.
Beschorner, T. Zeller, M. Braspenning, W. van Criekinge, K. Makinen, S. YlaHerttuala, Global DNA methylation analysis of human atherosclerotic plaques
reveals extensive genomic hypomethylation and reactivation at imprinted
locus 14q32 involving induction of a miRNA cluster, Eur. Heart J. 36 (2014)
993–1000.
S. Kuwamoto, H. Inoue, Y. Tone, Y. Izumi, T. Tanabe, Inverse gene expression of prostacyclin and thromboxane synthases in resident and activated peritoneal macrophages, FEBS Lett. 409 (1997) 242–246.
C.A. Payne, S. Maleki, M. Messina, M.G. O'Sullivan, G. Stone, N.R. Hall, J.F. Parkinson,
H.R. Wheeler, R.J. Cook, M.T. Biggs, N.S. Little, C. Teo, B.G. Robinson, K.L. McDonald,
Loss of prostaglandin D2 synthase: a key molecular event in the transition of a
low-grade astrocytoma to an anaplastic astrocytoma, Mol. Cancer Ther. 7 (2008)
3420–3428.
A.A. Mousa, J.F. Strauss III, S.W. Walsh, Reduced methylation of the thromboxane
synthase gene is correlated with its increased vascular expression in preeclampsia,
Hypertension 59 (2012) 1249–1255.
R. Jouve, P.H. Rolland, C. Delboy, C. Mercier, Thromboxane B2, 6-keto-PGF1 alpha,
PGE2, PGF2 alpha, and PGA1 plasma levels in arteriosclerosis obliterans: relationship to clinical manifestations, risk factors, and arterial pathoanatomy, Am. Heart
J. 107 (1984) 45–52.
K.M. Egan, M. Wang, S. Fries, M.B. Lucitt, A.M. Zukas, E. Pure, J.A. Lawson, G.A.
FitzGerald, Cyclooxygenases, thromboxane, and atherosclerosis: plaque destabilization by cyclooxygenase-2 inhibition combined with thromboxane receptor antagonism, Circulation 111 (2005) 334–342.
V. Angeli, D. Staumont, A.S. Charbonnier, H. Hammad, P. Gosset, M. Pichavant, B.N.
Lambrecht, M. Capron, D. Dombrowicz, F. Trottein, Activation of the D prostanoid receptor 1 regulates immune and skin allergic responses, J. Immunol. 172 (2004)
3822–3829.
H. Hammad, H.J. de Heer, T. Soullie, H.C. Hoogsteden, F. Trottein, B.N. Lambrecht,
Prostaglandin D2 inhibits airway dendritic cell migration and function in steady
state conditions by selective activation of the D prostanoid receptor 1, J. Immunol.
171 (2003) 3936–3940.
Y. Eguchi, N. Eguchi, H. Oda, K. Seiki, Y. Kijima, Y. Matsu-ura, Y. Urade, O. Hayaishi,
Expression of lipocalin-type prostaglandin D synthase (beta-trace) in human heart
and its accumulation in the coronary circulation of angina patients, Proc. Natl.
Acad. Sci. U. S. A. 94 (1997) 14689–14694.
Y. Taba, T. Sasaguri, M. Miyagi, T. Abumiya, Y. Miwa, T. Ikeda, M. Mitsumata, Fluid
shear stress induces lipocalin-type prostaglandin D(2) synthase expression in vascular endothelial cells, Circ. Res. 86 (2000) 967–973.
M. Miyagi, Y. Miwa, F. Takahashi-Yanaga, S. Morimoto, T. Sasaguri, Activator
protein-1 mediates shear stress-induced prostaglandin d synthase gene expression
in vascular endothelial cells, Arterioscler. Thromb. Vasc. Biol. 25 (2005) 970–975.
T. Inoue, Y. Eguchi, T. Matsumoto, Y. Kijima, Y. Kato, Y. Ozaki, K. Waseda, H. Oda, K.
Seiki, K. Node, Y. Urade, Lipocalin-type prostaglandin D synthase is a powerful biomarker for severity of stable coronary artery disease, Atherosclerosis 201 (2008)
385–391.