Mechanistic Role of Cytochrome P450 Monooxygenases in Oxidized Low-Density Lipoprotein–Induced Vascular Injury
Therapy Through LOX-1 Receptor Antagonism?
Oxidized low-density lipoprotein (oxLDL) is an important risk factor for vascular injury. Its role on coronary vasoconstriction remains speculative. Endothelial monooxygenases (cytochrome P450s [CYPs]) are regulators of vascular tonus through production of epoxy fatty acids. We investigated the effects of oxLDL on CYP monooxygenases in human arterial coronary endothelial cells and explanted healthy and atherosclerotic aortae. We found oxLDL to induce radical oxygen species production via the action of NADPH oxidase NOX4. Intracellular radical oxygen species production prompted reduced protein expression of the transcriptional regulator nuclear factor 1 (NF-1). We identified novel DNA binding sites for NF-1 in promoter regions of CYPs. DNA binding of NF-1 was confirmed by electromobility shift assays. OxLDL repressed DNA binding of NF-1 and diminished transcript level of CYP genes targeted by this factor. The production of endothelial-derived hyperpolarization factor, a key regulator of vascular tonus, was also reduced. Repression of CYP monooxygenases was reversed, and production of endothelial-derived hyperpolarization factor was normalized after treatment of endothelium with the lectin-like oxLDL receptor antagonist κ-carrageenan or blocking of LOX-1 with a specific antibody. This suggests a mechanistic role of CYP monooxygenases in oxLDL-induced vascular injury. Therapy of endothelial dysfunction through LOX-1 receptor antagonism will be an interesting avenue to explore. The full text of this article is available online at http://www.circresaha.org.
- endothelial dysfunction
- endothelial-derived hyperpolarization factor
- arachidonic acid
- oxidized low-density lipoprotein
- cytochrome P450 monooxygenases
Enhanced oxidation of low-density lipoprotein particles (oxLDL) leads to vascular injury and atherosclerotic plaque formation.1,2 As of today, the molecular basis of disease and specifically the mechanism of endothelial dysfunction remains uncertain. Notably, oxLDL is a ligand for the lectin-like oxLDL receptor (LOX-1).3 On intracellular availability, oxLDL triggers an array of events, including expression of various adhesion molecules, exaggerated production of radical oxygen species (ROS), scavenging for NO, thereby reducing its availability, and synthesis of toxic peroxynitrite.4 Overall, these events impair endothelial function and regulation of vascular tonus.
There is evidence that cytochrome P450 (CYP) monooxygenases produce vasoactive molecules, some of which are termed endothelial-derived hyperpolarization factors (EDHF).5 For instance, the CYP monooxygenases isoforms CYP1A, CYP2B6, CYP2C, and CYP2J catalyze oxidation of arachidonic acid6,7 to yield epoxy fatty acids, including 11,12-epoxyeicosatrienoic acid (11,12-EET). This particular epoxy fatty acid has an established role in the regulation of vasodilation of coronary and mesenterial arteries.8–10
Furthermore, several lines of evidence suggest that oxLDL interferes with vasodilatory mechanisms, which in turn lead to coronary vasoconstriction by as-yet unknown mechanisms.11 In view of the role of CYP in the production of signaling molecules for vascular tonus,5 it was highly interesting to investigate the effects of oxLDL on CYP monooxygenase regulation in endothelial cells. Recently, we showed significant repression of CYP monooxygenases in dysfunctional endothelium and demonstrated activation of OCT-1, a transcriptional repressor, to coincide with loss of the endothelial phenotype.12 In this study, we focus on the role of oxLDL on CYP regulation and investigated, among others, production of intracellular NO and CYP-catalyzed production of the epoxy fatty acid 11,12-EET. We investigated the effects of competitive oxLDL receptor antagonism to delineate the role of oxLDL in CYP monooxygenase regulation as well as NADPH oxidase activation to foster ROS production. Additionally, we compared findings from in vitro cell culture experiments with CYP gene expression profiles of explanted human healthy and atherosclerotic aorta ascendens biopsy materials to enable translation of laboratory findings to human disease. Because of its redox sensitivity and its established role in CYP gene regulation, DNA binding of the transcription factor nuclear factor 1 (NF-1) to target genes was also studied. We additionally used an in silico approach to predict novel binding sites within CYP promoters.13–15 Evidence for oxLDL to interfere with NF-1 regulation stems from Western immunoblotting studies. We additionally investigated activation of the P65 subunit of NF-κB to explore the consequences of oxLDL-induced ROS production in human arterial coronary endothelial cells (HCAECs). Finally, we correlated electromobility gel shift assay (EMSA) findings with novel consensus binding sites of NF-1 and NF-κB to provide a better understanding of deregulated promoter activation of CYP genes and its consequences on the production of vasoactive molecules. Overall, we aimed to explore the role of oxLDL in CYP regulation to provide molecular insight into vascular injury.
Materials and Methods
Primary human coronary arterial endothelial cells were obtained from Clonetics and cultured in 75-cm2 plastic flasks in EGM-2MV medium (Clonetics). Confluent cultures were detached by trypsin/EDTA and plated on six wells until 90% confluence was reached. Cultured endothelial cells (forth passage) were checked by inverse-phase contrast microscopy before and after treatment with oxLDL (magnification ×20). We additionally examined expression of the endothelial-specific surface protein platelet-endothelial cell adhesion molecule-1 as a quality control using fluorescence-activated flow cytometry, as described previously.12 Endothelial dysfunction was triggered by incubation of cells with oxidized LDL and measured using endothelial NO synthase (eNOS) gene expression and enzyme activity, as well as gene expression of intracellular adhesion molecule-1 (ICAM-1), LOX-1, and monocyte chemoattractant protein 1 (MCP-1), which were previously shown to be inducible by oxLDL and therefore served as positive controls.16–18
Explanted Human Material
Approval for the use of tissue material was obtained from the ethical committee of the Medical School of Hannover, Germany. Immediately after explantation and during heart transplantation (in patients with ischemic cardiomyopathy), biopsy material was removed from the aorta ascendens (male donors, n=3) and excised tissue was shock-frozen in liquid nitrogen and stored at −80°C until analyzed. Explanted aortic tissue macroscopically suffered severely from arteriosclerosis. Total aortic RNA from healthy human donors (male donors, n=3) was obtained from Biocat GmbH.
Enzyme Markers for Membrane Integrity
Lactate dehydrogenase activity (endothelial cells, 220.127.116.11 U/L) was measured in endothelial cell cultures treated with ascending doses of oxLDL (0, 10, 20, 100, and 300 μg/mL) for 24 hours according to the manufacturer’s recommendation using an automated device for a kinetic UV test at 37°C (Cobas Fara, Roche).
Preparation of LDL and oxLDL
LDL was isolated from human plasma by sequential gradient ultracentrifugation, as described previously.19 Oxidized LDL was prepared by incubation of LDL with 5 μmol/L CuSO4 for 24 hours at 37°C. Oxidation was monitored using the thiobarbituric acid–reactive substances (TBARS) assay (see below) with tetraethoxypropane as an internal standard.
RNA and cDNA
RNA was isolated from endothelial cells using a total RNA Isolation System (Macherey-Nagel) according to the manufacturer’s recommendation. Quality and quantity of isolated RNA were checked using capillary electrophoresis (Bioanalyzer 2100, Agilent Technologies) following the manufacturer’s instructions. Total RNA 2 μg from each sample was used for reverse transcription (RT), as described previously.20 The resulting cDNA was frozen at −20°C until additional experimentation.
Real-Time Semiquantitative Polymerase Chain Reaction
Real-time reverse transcription–polymerase chain reaction (RT-PCR) measurements were done with the Lightcycler (Roche Diagnostics), as described previously.21 After an initial denaturation step at 95°C for 30 seconds, the PCR reaction was initiated with an annealing temperature of 55°C for 8 seconds followed by an extension phase for 14 seconds at 72°C and a denaturation cycle at 95°C for 1 second. The PCR reaction was stopped after a total of 30 to 40 cycles, and at the end of each extension phase, fluorescence was observed and used for quantitative measurements within the linear range of amplification. Exact quantification was achieved by a serial dilution with cDNA produced from endothelial total RNA extracts using 1:5 dilution steps. Gene expression levels were then given as the ratio of the gene of interest (nominator) versus a stable expressed housekeeping gene (cyclophilin A, denominator22).
PCR reactions were carried out in a thermal cycler (T3, Biometra). Detailed oligonucleotide sequence information and amplification settings can be obtained on request from the authors or from a recently published study.20 DNA contamination was checked for by direct amplification of RNA extracts before conversion to cDNA. Contamination of RNA extracts with genomic DNA could be excluded. PCR reactions were done within the linear range of amplification, and amplification products were separated using a 1.5% agarose gel and stained with ethidium bromide. Gels were photographed on a transilluminator (Kodak Image Station 440), and amplicons were quantified using the Kodak 1D 3.5 network software.
Nitric Oxide Analysis
Intracellular eNOS activity was determined using flow cytometry and 4,5-diaminofluorescein diacetate as reagent to quantify NO. After treatment with different doses of oxLDL or native LDL (nLDL), cells were incubated for 40 minutes with 10 μmol/L 4,5-diaminofluorescein diacetate at 37°C and then washed with PBS buffer and incubated for an additional 30 minutes to allow complete deesterification of the intracellular diacetates. Fluorescence emission was then measured at 515 nm using flow cytometry (FACScan, Becton Dickinson). The NOS inhibitor Nω-nitro-l-arginine methyl ester (L-NAME, Sigma, 500 μmol/L for 2 hours) was used as a positive control.
Intracellular ROS Measurements
After treatment with oxLDL (10, 20, and 100 μg/mL; 24 hours) or nLDL (100 μg/mL; 24 hours), cells were incubated with 10 μmol/L 2′-7′-dichlorofluorescin diacetate for 30 minutes at 37°C. We also cotreated the 100-μg/mL oxLDL group with inhibitors of CYP monooxygenases (clotrimazole, 5 μmol/L, 24 hours; SKF525, 10 μmol/L, 24 hours), xanthine oxidase (allopurinol, 1 mmol/L, 24 hours), cyclooxygenases (indomethacin, 100 μmol/L, 24 hours or acetylsalicylic acid, 100 μmol/L, 24 hours), and NAPDH oxygenase (diphenyliodonium, 10 μmol/L, 24 hours). Concomitantly, the 100-μg/mL oxLDL group was treated with the competitive oxLDL receptor antagonist κ-carrageenan (250 μmol/L, 24 hours) or with a specific LOX-1 antibody (oxLDL receptor 1 antibody, Santa Cruz Biotech, 10 μg/mL; pretreatment of cells for 30 minutes). We used H2O2 (50 μmol/L, 30 minutes) as a positive control for ROS production, as described previously.23 In addition, cells were pretreated with PEG catalase (1 mg/mL, 4 hours, Sigma) and PEG-superoxide dismutase (1 mg/mL, 4 hours, Sigma) before adding 100 μg/mL oxLDL for 24 hours. Endothelial cells were harvested, centrifuged for 5 minutes and 1200 rpm at 4°C, and washed in PBS, and the resulting cell pellet was resuspended with 800 μL PBS buffer. After addition of propidium iodide (1 μg/mL), fluorescence emission was detected at 530±30 nm (fluorescin) and 585±42 nm (propidium iodide) after excitation of cells at 488 nm using flow cytometric methods (FACScan, Becton Dickinson).
Endothelial cell cultures were treated with 10, 20, and 100 μg/mL oxLDL and 100 μg/mL nLDL for 24 hours. Then cells were collected and sonified (20 times for 0.5 seconds) in 2% sodium dodecyl sulfate (SDS, 400 μL per cell pellet) to prepare total protein extracts. Protein concentrations were determined according to Smith et al24 and were adjusted to 1 mg protein/mL. One milligram of total protein was used for determination of TBARS, as described previously.10 TBARS levels of nLDL and oxLDL preparations were additionally measured. The amounts of TBARS were calculated using tetraethoxypropane in 1% sulfuric acid at a concentration range of up to 500 nmol/L as a standard.
Ethoxy- and Pentoxy- Resorufin-O-Deethylase (EROD and PROD) Assay
This assay was done essentially as described previously.25 Control cells and cells treated with 10, 20, and 100 μg/mL oxLDL (24 hours) were incubated with 2 μmol/L 7-ethoxyresorufin (or 7-pentoxyresorufin) and 10 μmol/L dicumarol (Sigma) for 4 hours. To assess the effect of LDL oxidation, we also incubated cells with nonoxidized nLDL (100 μg/mL, 24 hours). The impact of the lectin-like oxLDL receptor was investigated by cotreatment of the 100-μg/mL oxLDL group with the LOX-1 inhibitor κ-carrageenan (250 μmol/L) or pretreatment of cells with an inhibiting LOX-1 antibody (10 μg/mL).26 We used Aroclor 1254 as a positive control for CYP1A1 activation, as described previously.12,27 In addition, we incubated cells with H2O2 (50 μmol/L, 24 hours) alone as well as with H2O2 and κ-carrageenan (250 μmol/L, 24 hours). Two hundred fifty microliters of the samples was treated with 250 μL ammonium acetate (pH 4.5) and with or without 100 U/mL of β-glucuronidase (Sigma) overnight at 37°C to assess the product release of β-glucuronide conjugates. After addition of 500 μL glycine buffer (pH 10.3), fluorometric analysis was carried out on a spectro-fluoro-photometer (BioRad). Calibration of the system was done with resorufin as an appropriate standard at a concentration range of up to 100 nmol/L.
Control cells and cells treated with oxLDL (10, 20, and 100 μg/mL for 24 hours) were incubated with 100 μmol/L testosterone for 4 hours. Testosterone and its metabolites were analyzed by high-performance liquid chromatography (HPLC) with slight modifications, as described previously.20,28 11-α-Hydroxyprogesterone was used as an internal standard for the quantitative determination of testosterone and its metabolites.
Metabolism of Arachidonic Acid and Production of EDHF
11,12-EET was extracted by solid-phase extraction (SPE) followed by high-performance liquid chromatography mass spectrometry. Cell cultures of HCAECs were treated with 10 to 100 μg/mL oxLDL or 100 μg/mL nLDL for 24 hours. In addition, cells were treated with the CYP2C9 inhibitor sulfaphenazole (Sigma, 10 μmol/L) for 24 hours. Thereafter, 30 μmol/L of arachidonic acid was added and incubated for 5 minutes. Subsequently, the supernatant was harvested, shock-frozen, and stored at −80°C to await analysis. Before SPE, 10 ng tridecanoic acid (Sigma) was added as an internal standard and protein was precipitated by adding 100 μL of 1 mol/L acetic acid (Sigma) to 2 mL of supernatant. The samples were then applied to a SPE column (Oasis HLB, 60 mg, Waters) that had been preconditioned with 1 mL methanol (with 0.0005% butylated hydroxytoluene) and equilibrated with 1 mL of 5% acetic acid. The column was washed with 1 mL of 5% acetic acid and dried under the stream of nitrogen gas. Epoxy fatty acids were eluted with 3 mL methanol (and 0.0005% butylated hydroxytoluene). The eluate was filtered through sodium sulfate and dried under a gentle stream of nitrogen gas. The extract was resuspended in 75 μL acetonitrile/acetic acid (0.5%; 1:1).
The 11,12-EET standard (Sigma) was added to blank culture media (0.25 to 5.0 ng/mL) and extracted as detailed above. Epoxy fatty acids were separated on a Securigard ODS pre-column (4×2 mm, Phenomenex) and a C18-discovery column (150×2.1 mm, 5 μmol/L, Supelco). The injection volume was 10 μL, and the HPLC consisted of an Agilent 1100 HPLC coupled to an Esquire 3000 plus mass spectrometer (Bruker Daltonik). A acetonitrile/acetic acid (A/B) was used as a mobile phase, and the gradient started from 55% A and 45% B (5 minutes) to 100% B for 38 minutes followed by 55% A and 45% B for 17 minutes. The flow rate was 200 μL/min, and the retention time for 11,12-EET was 27.3 minutes.
The ion trap mass spectrometer (Esquire 3000plus, Bruker Daltonik) was operated in negative ion electrospray conditions and full scan or mass spectrometry/mass spectrometry (MS2) mode. The nebulizer pressure was set to 15 psi, and the dry gas temperature was 320°C, whereas +4.2 kV was applied to the nebulizing capillary. Full mass spectra were acquired by scanning the mass range of m/z 150 to 340. Identification of metabolites was done by collision-induced dissociation experiments in multiple liquid chromatography mass spectrometry/mass spectrometry experiments. For quantification, the following mass fragments of 11,12-EET were used: 301.3 m/z, 275.3 m/z, 257.2 m/z, 221.2 m/z, and 179.1 m/z.
Western Blotting Experiments
Western immunoblotting was done as follows: total protein (100 μg) or nuclear protein (100 μg) extracts from cultured endothelial cells were denaturated at 95°C for 5 minutes, followed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) on 12% polyacrylamide gels, and blotted onto a polyvinylidene difluoride membrane (NEN) at 350 mA for 2 hours in a buffer containing 400 mmol/L glycine and 50 mmol/L Tris (pH 8.3). Nonspecific binding sites were blocked with Rotiblock (Roth) in 1× TBS buffer. After electroblotting of proteins, membranes were incubated with polyclonal antibodies for NF-κB, P65 (Santa Cruz Biotechnology, dilution 1:200), NF-1 (Santa Cruz Biotechnology, dilution 1:400), or CYP2C (Chemicon, dilution 1:1000) for 1 hour and washed three times with 1× TBS buffer containing 0.1% Tween-20 (Roth). Subsequently, the membranes were incubated with a 1:5000 diluted anti-α-rabbit (NF-κB, NF-1) or anti-α-sheep (CYP2C) antibody (Chemicon) for 1 hour at room temperature followed by three successive washes with 1× TBS buffer containing 0.1% Tween-20 (Roth). Immunoreactive proteins were visualized with a chemiluminescence reagent kit (NEN) according to the manufacturer’s instructions, and bands were scanned with the Kodak Image Station CF 440 and analyzed using the Kodak 1D 3.5 imaging software (Eastman Kodak Company).
Preparation of Nuclear Extracts
HCAEC nuclear extracts were prepared by the modified Dignam C method.29 Twenty-four hours after treatment with oxLDL (100 μg/mL), cells were washed twice with ice-cold PBS, scraped into microcentrifuge tubes, and centrifuged for 5 minutes at 1780g and 4°C. Cell pellets were resuspended in hypotonic buffer (10 mmol/L Tris, pH 7.4, 2 mmol/L MgCl2, 140 mmol/L NaCl, 1 mmol/L dithiothreitol (DTT), 4 mmol/L Pefabloc, 40 mmol/L β-glycerophosphate, 1 mmol/L Na3VO4, 10 μL aprotinin/mL buffer, and 0.5% Triton X-100) for 10 minutes at 4°C, transferred onto one volume of 50% sucrose in hypotonic buffer (see above), and centrifuged at 14 000g and 4°C for 10 minutes. Nuclei were resuspended in Dignam C buffer (20 mmol/L HEPES, pH 7.9, 25% glycerol, 420 mmol/L NaCl, 1.5 mmol/L MgCl2, 0.2 mmol/L EDTA, 1 mmol/L DTT, 4 mmol/L Pefabloc, 40 mmol/L β-glycerophosphate, 1 mmol/L Na3VO4, and 10 μL aprotinin/mL buffer) and gently rocked for 30 minutes at 4°C. Nuclear debris was removed by centrifugation at 14 000g and 4°C for 10 minutes, and extracts were aliquoted and stored at −80°C. Protein concentrations were determined, as described previously.24
Annealing of Synthetic Oligonucleotides and [32P] Labeling
Oligonucleotides representing a high-affinity consensus NF-1 or NF-κB binding site were chosen, as described.30 Oligonucleotide sequences were 5′-ATTTTGGCTTGAAGCCAATATG and complement 5′-CATATTGGCTTCAAGCCAAAAT for NF-1 and 5′-GCCATGGGGGGATCCCCGAAGTCC and complement 5′-GGACTTCGGGGATCCCCCCATGGC for NF-κB. Oligonucleotides were annealed at a concentration of 19.2 pmol/L per μL in 200 mmol/L Tris (pH 7.6), 100 mmol/L MgCl2, and 500 mmol/L NaCl at 80°C for 10 minutes and then cooled slowly to room temperature overnight and stored at 4°C. Annealed oligonucleotides were diluted to 1:10 in Tris-EDTA (pH 8.0) buffer and labeled using [32P]ATP (Amersham, 250 μCi, 3000 Ci/mmol) and T4 polynucleotide kinase (New England Biolabs). End-labeled probes were purified from unincorporated [32P]ATP by a Microspin G-25 Column (Amersham) and eluted in a 100-μL volume.
The procedure for EMSA was adapted from a previously described method.29 Briefly, 10 μg of endothelial nuclear extract was incubated with the binding buffer consisting of 25 mmol/L HEPES (pH 7.6), 5 mmol/L MgCl2, 34 mmol/L KCl, 2 mmol/L DTT, 2 mmol/L Pefablock (Boehringer Mannheim), 0.5 μL aprotinin (2.2 mg/mL, Sigma), 50 ng poly (dl-dC), and 80 ng BSA (PAA). The binding reaction was carried out for 20 minutes on ice, and free DNA and DNA/protein complexes were resolved on a 6% polyacrylamide gel. Competition studies were done by adding a specified amount (50-fold) of unlabeled oligonucleotides or a specific NF-1 or NF-κB P65 antibody (Santa Cruz Biotechnology) to the reaction mix 10 minutes before addition of the labeled oligonucleotides. Gels were blotted to Whatman 3-mm paper, dried under vacuum, exposed to imaging screens (Imaging Screen-K, BioRad) for autoradiography for 24 hours at room temperature, and analyzed using a phosphor imaging system (Molecular Imager FX pro plus; BioRad) and the Quantity One Version 4.2.2 software (BioRad).
Transcription Factor Binding Sites in CYP Promoters
We searched for NF-1 and NF-κB binding sites in the promoters of CYP1A1, CYP2B6/7, CYP2C8, CYP2E1, and CYP2J2 using the transcription factor database TRANSFAC Professional 6.2 (http://transfac.gbf.de/TRANSFAC/). Core and matrix similarities for all binding sites were set to 1.0 and 0.85 to obtain specific results.
Statistical analysis was done using the Wilcoxon signed-rank test and was considered to be significant at P<0.05.
Cellular morphology was studied by phase-contrast microscopy. No signs of altered morphology or cellular toxicity were observed after treatment of cultures with ascending doses of oxLDL (10 to 100 μg/mL). Expression of endothelial-specific surface antigen platelet-endothelial cell adhesion molecule-1 served as differentiation marker and was expressed >95%, as determined by flow cytometry measurement12 (data not shown). LDH activity was used as a marker for membrane integrity and was assayed after treatment of cell cultures with incremental doses of oxLDL (24 hours). Ascending doses of oxLDL caused LDH leakage into culture media with >40 U/L in the high-dose group (data not shown). Based on LDH activity, we used 100 μg/mL oxLDL as the highest tolerable noncytotoxic dose.
Gene Expression of CYP Monooxygenases and Epoxide Hydrolase
We investigated gene expression of a total of 18 CYP isoforms in RNA extracts of HCAECs. We observed expression of CYP1A1, CYP2A6/7, CYP2B6/7, CYP2C8, CYP2C9, CYP2E1, and CYP2J2, whereas transcript level of CYP2C18, CYP2D6, CYP3A4, CYP3A5, CYP3A7, and CYP4B1 was below the limit of detection (Figure 1A). Expression of CYP1A2, CYP2A13, CYP2C19, and CYP2S1 was close to the limit of detection.
We treated cultures of HCAECs with ascending doses of oxLDL and observed dose-dependently repression of CYP1A1, CYP2A6/7, CYP2B6/7, CYP2C8, CYP2C9, CYP2E1, and CYP2J2 transcript levels to 25%, 60%, 10%, 0%, 15%, 20%, and 40% of controls, respectively (Figure 1B). In contrast, no change in the gene expression of epoxide hydrolase or cyclophilin was detected after treatment, and nonoxidized LDL (100 μg/mL, 24 hours) had no effect (see Figure 1B). Furthermore, treatment of HCAECs with the oxLDL receptor inhibitor κ-carrageenan or with a specific LOX-1 antibody restored gene expression levels of CYP1A1, CYP2C8, CYP2C9, CYP2E1, and CYP2J2 to normal (see Figure 1B).
Additionally, we studied CYP gene expression in healthy (n=3) and diseased (n=3) aortic tissue and did observe repressed CYP1A1, CYP2B6/7, CYP2E1, CYP2J2, CYP2C8, and CYP2C9 transcript levels to 0%, 9%, 21%, 32%, 27%, and 0% in diseased human aortic tissue, respectively (Figure 2A). In contrast, no change in the expression of the housekeeping genes cyclophilin or 18SrRNA was detected (Figure 2A), and total RNA extracts of healthy and diseased aortae were of high quality (Figure 2B). The repressed CYP gene transcription in diseased aortae is therefore specific.
Markers of Endothelial Dysfunction
Treatment of HCAECs with 10 μg/mL oxLDL resulted in a nonsignificant 30% increase in eNOS gene expression, whereas dose escalation to 20 and 100 μg/mL repressed significantly eNOS gene expression to 70% and 30% of controls, respectively. Furthermore, oxLDL produced dose-dependently up to 4-fold increases in MCP-1, ICAM-1, and LOX-1 gene expression (see Figure 2C), whereas expression of NOX4 was evident in HCAECs treated with 100 μg/mL oxLDL (24 hours) only (Figure 2D). Notably, gene expression of NOX1 and inducible NOS was below the limit of detection (data not shown), and treatment with nonoxidized LDL had no effects. In contrast, concomitant treatment of cell cultures with oxLDL (100 μg/mL) and κ-carrageenan or inhibition of LOX-1 with an antibody led, in part, to control eNOS, ICAM-1, MCP-1, NOX4, and LOX1 expression levels (sees Figures 2C and 2D).
Enzyme Activity of CYP Monooxygenases (EROD and PROD Assay)
We used 7-ethoxyresorufin and 7-pentoxyresorufin as marker substrates to distinguish enzyme activity of CYP monooxygenase isoforms. We observed dose-dependent but not dose-linear repression of CYP1A1-catalyzed EROD activity to 85%, 65%, and 65% (n=6, P<0.001) of controls (see Figure 3A). We also determined secondary metabolism of this substrate. On average, 5% of oxidized product was glucuronidated, which suggests minimal glucuronidase activity in cultures of HCAECs (data not shown). Likewise, studies with 7-pentoxyresorufin evidenced a dose-dependent repression of monooxygenase activity to 60%, 40%, and 7% (n=3, P<0.05) of controls, respectively (see Figure 3B). No glucuronide of the oxidized product was detected. Importantly, treatment of HCAECs with nLDL (none oxidized) alone did not change monooxygenase activity with EROD and PROD as substrate (see Figure 3).
Using the protocol of Moriwaki et al,26 we treated HCAECs with the lectin-like oxLDL receptor (LOX1) antagonist κ-carrageenan or with an inhibitory antibody of this receptor. Treatment of HCAECs with κ-carrageenan alone did not change EROD or PROD activities (see Figure 3). Strikingly, repressed EROD activity was recovered, in part, in cultures treated concomitantly with κ-carrageenan or an inhibitory LOX-1 antibody. No effect on PROD activity was obvious (see Figures 3A and 3B). For comparison, treatment of endothelial cells with H2O2 (50 μmol/L, 24 hours) resulted in a 70% repression in EROD and PROD activity. Addition of κ-carrageenan alone (250 μmol/L, 24 hours) to H2O2 (50 μmol/L, 24 hours) did not prevent repressed EROD and PROD activities. We used Aroclor 1254 (20 μmol/L, 24 hours) as a positive control12 and demonstrate an approximate 4-fold increase in EROD activity (Figures 3A and 3B). This evidences responsiveness and reliability of the cell culture assay.
Production of 11,12-EET
We used an HPLC electrospray MS2 method to detect 11,12-EET production in cultures of HCAECs. After treatment of cells with 10 and 100 μg/mL of oxLDL, a dramatic decrease in the production of 11,12-EET to 41% and 20% of controls was observed. Treatment with nLDL (100 μg/mL) did not change 11,12-EET production observed in controls. Strikingly, treatment with the oxLDL receptor antagonist κ-carrageenan restored production levels of 11,12-EET to normal (see Figure 3C). Importantly, treatment with the specific CYP2C9 inhibitor sulfaphenazole (10 μmol/L, 24 hours) repressed 11,12-EET production to 35% of controls (see Figure 3C).
Enzyme Activity of CYP Monooxygenases (Testosterone Assay)
Metabolism of testosterone in oxLDL-treated endothelial cells was decreased. Production of androstenedione, 2a-HT, 6a-HT, 6b-HT, and 7a-HT was reduced to 20%, 40%, 55%, 65%, and 0% of controls in cell cultures treated with 100 μg/mL oxLDL (24 hours) (Figure 3D).
Malondialdehyde and ROS Measurement
TBARS level of an oxLDL solution (100 μg/mL) was 33-fold above nLDL preparations, eg, 207.7±25.8 nmol/L per mg compared with 6.3±0.9 nmol/mg. We observed a dose-dependent increase in TBARS formation in cultures of HCAECs treated with 10, 20, and 100 μg/mL oxLDL, the level being 1, 10, and 45 nmol/L per mg protein (Figure 4A). Cells treated with nLDL (100 μg/mL) alone produced TBARS levels comparable to controls. Treatment of cultures of HCAECs with 100 μg/mL oxLDL increased intracellular ROS production by 3-fold, whereas nLDL had no effect. Notably, ROS production was completely abolished in the presence of the NADPH inhibitor diphenyliodonium (10 μmol/L, 24 hours) or before treatment of cells with catalase (1 mg/mL, 4 hours) and superoxide dismutase (1 mg/mL, 4 hours). Treatment of cultures of HCAECs with inhibitors of CYP monooxygenases (clotrimazole, 5 μmol/L; SKF525, 10 μmol/L), xanthine oxidase (allopurinol, 1 mmol/L), or cyclooxygenases (indomethacin, 100 μmol/L, or acetylsalicylic acid, 100 μmol/L) had no effect on ROS production (Figures 4B and 4C). Importantly, treatment of HCAECs with oxLDL (100 μg/mL, 24 hours) and κ-carrageenan (250 μmol/L, 24 hours) or with an inhibitory LOX-1 antibody (10 μg/mL, 24 hours) reduced significantly (P<0.05) enhanced intracellular ROS production (Figure 4B).
Production of Intracellular NO
When compared with controls, intracellular NO production decreased to ≈60% of controls with the 100 μg/mL oxLDL dose (Figures 4D and 4E). This effect was reversed with the oxLDL receptor antagonist κ-carrageenan or an inhibitory LOX-1 antibody. Treatment with nLDL (100 μg/mL) did not change intracellular NO production. We used L-NAME as an inhibitor for NO production (positive control) and observed reduced NO levels to 30% of controls.
Western Blotting Studies
Protein expression of the inducible P65 subunit of NF-κB was unchanged in total or nuclear protein extracts of control or oxLDL (100 μg/mL)-treated cell cultures (see Figure 5A). Protein expression of NF-1 was strongly repressed when cells were treated with 100 μg/mL oxLDL, but nLDL (100 μg/mL) had no effect. Protein expression of NF-1 returned to normal in the presence of the oxLDL receptor antagonist κ-carrageenan or prior treatment of cultures with an inhibitory LOX-1 antibody (see Figure 5A).
CYP2C protein was expressed in extracts of control cell cultures and human liver. The latter served as positive control for its abundant expression of CYP proteins. Treatment of cultures of HCAECs with oxLDL (100 μg/mL) abolished almost completely expression levels of CYP2C protein, whereas nLDL (100 μg/mL) had no effect. Treatment of cell cultures with the LOX-1 antagonist κ-carrageenan or with an inhibitory LOX-1 antibody restored CYP2C protein expression to normal (see Figure 5A). CYP1A1, CYP2A, and CYP2B protein expression was below the limit of detection (data not shown).
Nuclear extracts were prepared from control cell cultures or cultures treated with nLDL (100 μg/mL), oxLDL (100 μg/mL, 24 hours), oxLDL (100 μg/mL, 24 hours) plus κ-carrageenan (250 μmol/L, 24 hours), oxLDL (100 μg/mL, 24 hours) plus a specific LOX-1 antibody (10 μg/mL, 24 hours), as well as oxLDL (100 μg/mL, 24 hours) plus catalase (1 mg/mL) and superoxide dismutase (1 mg/mL). DNA binding of NF-1 as well as NF-κB was assayed with established consensus binding sites.30 We observed reduced NF-1 protein/DNA binding in nuclear extracts of oxLDL-treated HCAECs, whereas NF-κB protein/DNA binding was unchanged. Strikingly, concomitant treatment of oxLDL-treated HCAECs with κ-carrageenan or a specific LOX-1 antibody restored NF-1/DNA binding. Nonoxidized LDL did not change NF-1/DNA binding. Pretreatment with catalase and SD recovered repressed NF-1 DNA binding of oxLDL (100 μg/mL)-treated HCAECs. Competition studies with unlabeled NF-1 or NF-κB (50-fold) as well as supershifts with a NF-1 or NF-κB antibody demonstrate specificity of the assay (Figure 5B).
In Silico Promoter Analysis
Promoter sequences of CYP monooxygenases were interrogated with Matrix Search for Transcription Factor Binding Sites (MATCH Version 1.8, Biobase, Germany, http://www. biobase.de/cgi-bin/biobase/transfac/start.cgi). Several binding sites of NF-1 and NF-κB were identified, as detailed in the Table.
This study focused on the effects of oxidized lipoproteins on CYP regulation in cultures of human coronary arterial endothelial cells. We found oxLDL to be a powerful repressor of CYP monooxygenases and were able to reverse the detrimental effects of oxLDL with the lectin-like oxLDL receptor antagonist κ-carrageenan or with an inhibitory LOX-1 antibody. OxLDL initiated exaggerated ROS production through activation of the NADPH oxidase NOX4, whereas nonoxidized LDL was without effect. NF-1 is ROS-sensitive and an important regulator of CYP monooxygenases.31 Consequently, high levels of ROS prompted reduced availability of NF-1 protein and resulted in repressed NF-1/DNA binding. Despite high intracellular oxidative stress, NF-κB was not activated, as evidenced by Western immunoblotting of P65 and EMSA. Repressed expression of endothelial CYP monooxygenases led to diminished production of 11,12-EET, an endothelial-derived hyperpolarization factor8 (see Figures 3C and 6⇓). This suggests a mechanism by which oxLDL represses CYP monooxygenase-catalyzed EDHF production through diminished NF-1 availability.
In the present study, dose selection was based on published and clinically confirmed oxLDL plasma levels.2 Oxidized LDL was previously shown to repress eNOS activity.32 We confirm repressed intracellular NO levels on treatment of HCAECs with ascending doses of oxLDL. There is clear evidence that NO can be scavenged by ROS to produce peroxynitrite, the latter being a powerful toxicant for endothelium.33 Concomitantly, NO is less available to result in metabolic deregulation.
Next to NO, epoxy fatty acids are important signaling molecules in the regulation of vasodilation. Arachidonic acid serves as a substrate for the production of epoxy fatty acids and is a substrate for lipoxygenases and CYP monooxygenases.7 There is speculation about the relative contribution of individual CYP isoforms in the production of epoxy fatty acids, and, indeed, next to CYP2C8 and CYP2C9, other isoforms have been implicated, and this includes CYP2J2, CYP2B6, as well as CYP1A1.7,9 All of the above-mentioned CYP isoforms are capable of arachidonic acid oxidation, resulting in the production of signaling molecules of vascular tonus.
We observed repression of major CYP isoforms involved in the production of 11,12-EET. Indeed, 11,12-EET is considered to be an endothelial-derived hyperpolarization factor (EDHF) and was shown to produce vasodilation in isolated aortic rings.34 CYP1A and CYP2B6 are additional enzymes involved in arachidonic acid oxidation to yield EDHF products.9 We link repressed CYP monooxygenase expression levels to reduced availability of EDHFs, which are NO- and prostacyclin-independent signaling molecules of vasodilation.5 Moreover, CYP2C9 is a key player in EDHF production, and inhibition of CYP2C9 with sulfaphenazole repressed EDHF production (Figure 3C). Additional evidence for a major role of CYP monooxygenases in vascular injury stems from gene expression studies with diseased human atherosclerotic aortae, where transcript levels of EDHF-catalyzing CYP isoforms were ≈30% of controls (Figure 2A). Unfortunately, healthy aortic tissue is scarcely available and, correspondingly, we could not perform incubation assays with microsomes isolated from healthy and diseased aortic tissue. Notably, repressed CYP monooxygenases in explanted human atherosclerotic aortic tissue may be the result of complex disease processes, including inflammation. Thus, the correlation between repressed CYP monooxygenases and oxLDL may be circumstantial. Nonetheless, our findings are intriguing and should stimulate in-depth research on the role of CYP monooxygenases in blood vessel disease.
Next to porcine coronary vessels,35 CYP induction resulted in restored EDHF-mediated relaxation in small mesenteric arteries and normalization of mean arterial pressure in insulin-resistant rats.36 The therapeutic benefit of CYP induction in vascular tissue should therefore be explored.
An important finding of our study was rescue of CYP1A1 and CYP2C on treatment of cell cultures with the oxLDL receptor antagonist κ-carrageenan or with an inhibitory LOX-1 antibody. This suggests cross talk of LOX-1 with transcriptional regulators of CYP1A1.
Based on EMSA assays, we demonstrate reduced DNA binding of NF-1 to a validated consensus binding site30 on oxLDL-induced ROS production. This transcription factor is an important regulator for several P450 monooxygenases and was shown to be redox-sensitive, as evidenced in rat hepatoma cells treated with H2O2.13 In this particular study, reduced availability of NF-1 led to repressed CYP1A and CYP2E1 transcript levels. In the present study, we demonstrate reduced DNA binding of NF-1 in nuclear extracts of oxLDL-treated HCAECs and link reduced endothelial CYP monooxygenase transcript levels to reduced availability of NF-1. Moreover, we used catalase and superoxide dismutase to probe for ROS-specific effects on NF-1 expression and its DNA binding. Addition of catalase and superoxide dismutase resulted in ROS levels below controls (Figure 4B). Consequently, NF-1 expression and DNA binding returned to normal (Figure 5B). We thus demonstrate ROS to be specifically responsible for diminished NF-1 expression and reduced DNA binding. The oxLDL inhibitor κ-carrageenan and addition of an inhibitory LOX-1 antibody reversed this effect (see Figure 5B). We also performed an in silico promoter analysis to confirm NF-1 binding sites in promoters of regulated CYP monooxygenases (see the Table). Furthermore, we extend the findings of Heinloth et al,37 who investigated the effects of oxLDL on NADPH oxidase activation in endothelial cells of the human umbilical vein (HUVECs). We observed oxLDL to trigger dose-dependently NOX4 activation to result in significant ROS production, whereas nLDL had no effect (see Figure 4B). There is evidence for NOX1 and NOX4 to be involved in vascular oxidative stress.37,38 Its expression was reported for vascular smooth muscle cells of rats.38 In addition, Rueckschloss et al39 reported expression of multimeric neutrophil NAD(P)H oxidase complex isoforms Gp91phox, Gp67phox, Gp47phox, and Gp22phox in HUVECs and demonstrated increased superoxide anion formation on oxLDL treatment. Based on gene expression studies, we suggest ROS production to be driven by NOX4, whereas NOX1 was below the limit of detection.
In the study by Graier et al,40 enhanced CYP monooxygenase activity was observed in assays with xanthine oxidase/hypoxanthine. This study therefore suggests exogenous ROS to stimulate CYP monooxygenase activity. However, the findings of Graier et al cannot easily be compared with our study for the following reasons.
Porcine aortic endothelial cells were used in the study of Graier et al, which may respond differently to the human coronary arterial endothelial cells used in our study. Furthermore, Graier et al used thapsigargin for depletion of intracellular calcium stores. This again will impact the biology of endothelial cells, and thus straightforward comparisons cannot be made. Specifically, the authors did not study gene and protein expression of CYP monooxygenases and DNA binding of NF-1. Finally, Graier et al studied the effects of ROS 1 hour after treatment; we investigated ROS 24 hours later. Whether superoxide has a dual role in the regulation of CYP monooxygenases requires additional investigations, but our findings clearly demonstrate a negative role of ROS in human coronary endothelial cells.
CYP monooxygenase expression may be repressed through activation of the NF-κB pathway,41,42 but there is conflicting information about its role in oxLDL-mediated endothelial dysfunction. Indeed, Li et al43 demonstrated activation of NF-κB in endothelial cells after treatment with oxLDL, whereas Heermeier et al44 reported suppression. We did not observe changes in the protein expression of the inducible NF-κB subunit P65 or NF-κB/DNA binding after treatment with oxLDL (see Figures 5A and 5B).
It was also shown that oxLDL enhances coronary vasoconstriction by increasing the activity of protein kinase C,11 and Imig et al45 suggest a unique role of epoxide hydrolase (EH) in arterial blood pressure, because inhibition of EH was shown to be antihypertensive. We thus studied gene expression of EH after treatment of HCAECs with oxLDL but did not find changes in transcript levels (Figure 1B). We thus assume no specific role of oxLDL on EH regulation, even though Imig et al45 demonstrated increased EH expression in angiotensin II hypertensive kidney.
In conclusion, we demonstrate oxLDL to be a powerful repressor of vascular CYP monooxygenases. We propose enhanced ROS production to diminish NF-1 availability for transcriptional activation of arachidonic acid–catalyzing monooxygenases (see Figure 6). This results in reduced production of EDHF. Expression of CYP monooxygenases was rescued with the LOX-1 inhibitor κ-carrageenan, as well as by blocking LOX-1 with a specific antibody. We provide a rationale for reduced vascular EDHF tissue production in patients with high circulating oxLDL plasma levels to result in endothelial dysfunction. CYP epoxygenases may therefore be valuable drug targets for the treatment of endothelial metabolic disease.
The financial support of the Lower Saxony Ministry of Science and Culture to J.B. is gratefully acknowledged. We kindly acknowledge the expert technical assistance of Chan Rong Lai.
Original received September 29, 2003; resubmission received November 7, 2003; accepted November 20, 2003.
Hein TW, Liao JC, Kuo L. oxLDL specifically impairs endothelium-dependent, NO-mediated dilation of coronary arterioles. Am J Physiol Heart Circ Physiol. 2000; 278: H175–H183.
Holvoet P, Vanhaecke J, Janssens S, Van de WF, Collen D. Oxidized LDL and malondialdehyde-modified LDL in patients with acute coronary syndromes and stable coronary artery disease. Circulation. 1998; 98: 1487–1494.
Cominacini L, Garbin U, Pasini AF, Davoli A, Campagnola M, Pastorino AM, Gaviraghi G, Lo CV. Oxidized low-density lipoprotein increases the production of intracellular reactive oxygen species in endothelial cells: inhibitory effect of lacidipine. J Hypertens. 1998; 16: 1913–1919.
Fleming I. Cytochrome p450 and vascular homeostasis. Circ Res. 2001; 89: 753–762.
Borlakoglu JT, Edwards-Webb JD, Dils RR. Evidence for the induction of fatty acid desaturation in proliferating hepatic endoplasmic reticulum in response to treatment with polychlorinated biphenyls: are fatty acid desaturases cytochrome P-450-dependent monooxygenases? Int J Biochem. 1991; 23: 925–931.
Rifkind AB, Lee C, Chang TK, Waxman DJ. Arachidonic acid metabolism by human cytochrome P450s 2C8, 2C9, 2E1, and 1A2: regioselective oxygenation and evidence for a role for CYP2C enzymes in arachidonic acid epoxygenation in human liver microsomes. Arch Biochem Biophys. 1995; 320: 380–389.
Yu Z, Huse LM, Adler P, Graham L, Ma J, Zeldin DC, Kroetz DL. Increased CYP2J expression and epoxyeicosatrienoic acid formation in spontaneously hypertensive rat kidney. Mol Pharmacol. 2000; 57: 1011–1020.
Giardina JB, Tanner DJ, Khalil RA. Oxidized-LDL enhances coronary vasoconstriction by increasing the activity of protein kinase C isoforms α and ε. Hypertension. 2001; 37: 561–568.
Thum T, Haverich A, Borlak J. Cellular dedifferentiation of endothelium is linked to activation and silencing of certain nuclear transcription factors: implications for endothelial dysfunction and vascular biology. FASEB J. 2000; 14: 740–751.
Morel Y, Barouki R. Down-regulation of cytochrome P450 1A1 gene promoter by oxidative stress: critical contribution of nuclear factor 1. J Biol Chem. 1998; 273: 26969–26976.
Ricci MS, Toscano DG, Mattingly CJ, Toscano WA Jr. Estrogen receptor reduces CYP1A1 induction in cultured human endometrial cells. J Biol Chem. 1999; 274: 3430–3438.
Li D, Mehta JL. Antisense to LOX-1 inhibits oxidized LDL-mediated upregulation of monocyte chemoattractant protein-1 and monocyte adhesion to human coronary artery endothelial cells. Circulation. 2000; 101: 2889–2895.
Li D, Mehta JL. Upregulation of endothelial receptor for oxidized LDL (LOX-1) by oxidized LDL and implications in apoptosis of human coronary artery endothelial cells: evidence from use of antisense LOX-1 mRNA and chemical inhibitors. Arterioscler Thromb Vasc Biol. 2000; 20: 1116–1122.
Pritchard KA Jr, Groszek L, Smalley DM, Sessa WC, Wu M, Villalon P, Wolin MS, Stemerman MB. Native low-density lipoprotein increases endothelial cell nitric oxide synthase generation of superoxide anion. Circ Res. 1995; 77: 510–518.
Thum T, Borlak J. Testosterone, cytochrome P450, and cardiac hypertrophy. FASEB J. 2002; 16: 1537–1549.
Telemaque-Potts S, Kuc RE, Maguire JJ, Ohlstein E, Yanagisawa M, Davenport AP. Elevated systemic levels of endothelin-1 and blood pressure correlate with blunted constrictor responses and downregulation of endothelin(A), but not endothelin(B), receptors in an animal model of hypertension. Clin Sci (Lond). 2002; 103 (suppl 48): 357S–362S.
Moriwaki H, Kume N, Sawamura T, Aoyama T, Hoshikawa H, Ochi H, Nishi E, Masaki T, Kita T. Ligand specificity of LOX-1, a novel endothelial receptor for oxidized low density lipoprotein. Arterioscler Thromb Vasc Biol. 1998; 18: 1541–1547.
Niehof M, Streetz K, Rakemann T, Bischoff SC, Manns MP, Horn F, Trautwein C. Interleukin-6-induced tethering of STAT3 to the LAP/C/EBPβ promoter suggests a new mechanism of transcriptional regulation by STAT3. J Biol Chem. 2001; 276: 9016–9027.
Morel Y, Barouki R. Down-regulation of cytochrome P450 1A1 gene promoter by oxidative stress: critical contribution of nuclear factor 1. J Biol Chem. 1998; 273: 26969–26976.
Liao JK, Shin WS, Lee WY, Clark SL. Oxidized low-density lipoprotein decreases the expression of endothelial nitric oxide synthase. J Biol Chem. 1995; 270: 319–324.
Fang X, Kaduce TL, Weintraub NL, VanRollins M, Spector AA. Functional implications of a newly characterized pathway of 11,12-epoxyeicosatrienoic acid metabolism in arterial smooth muscle. Circ Res. 1996; 79: 784–793.
Bauersachs J, Christ M, Ertl G, Michaelis UR, Fisslthaler B, Busse R, Fleming I. Cytochrome P450 2C expression and EDHF-mediated relaxation in porcine coronary arteries is increased by cortisol. Cardiovasc Res. 2002; 54: 669–675.
Heinloth A, Heermeier K, Raff U, Wanner C, Galle J. Stimulation of NADPH oxidase by oxidized low-density lipoprotein induces proliferation of human vascular endothelial cells. J Am Soc Nephrol. 2000; 11: 1819–1825.
Rueckschloss U, Galle J, Holtz J, Zerkowski HR, Morawietz H. Induction of NAD(P)H oxidase by oxidized low-density lipoprotein in human endothelial cells: antioxidative potential of hydroxymethylglutaryl coenzyme A reductase inhibitor therapy. Circulation. 2001; 104: 1767–1772.
Graier WF, Hoebel BG, Paltauf-Doburzynska J, Kostner GM. Effects of superoxide anions on endothelial Ca2+ signaling pathways. Arterioscler Thromb Vasc Biol. 1998; 18: 1470–1479.
Ke S, Rabson AB, Germino JF, Gallo MA, Tian Y. Mechanism of suppression of cytochrome P-450 1A1 expression by tumor necrosis factor-α and lipopolysaccharide. J Biol Chem. 2001; 276: 39638–39644.
Li D, Saldeen T, Romeo F, Mehta JL. Oxidized LDL upregulates angiotensin II type 1 receptor expression in cultured human coronary artery endothelial cells: the potential role of transcription factor NF-κB. Circulation. 2000; 102: 1970–1976.
Heermeier K, Leicht W, Palmetshofer A, Ullrich M, Wanner C, Galle J. Oxidized LDL suppresses NF-κB and overcomes protection from apoptosis in activated endothelial cells. J Am Soc Nephrol. 2001; 12: 456–463.
Imig JD, Zhao X, Capdevila JH, Morisseau C, Hammock BD. Soluble epoxide hydrolase inhibition lowers arterial blood pressure in angiotensin II hypertension. Hypertension. 2002; 39: 690–694.