INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

CYTOCHROME P-450 (CYP450) enzymes belong to one of the largest and most complicated family of isozymes with over 500 known members (47). Even though products of these proteins have been characterized in detail by xenobiologists and steroid chemists, unraveling a functional role for these enzymes at the molecular level has posed a challenge to investigators in cardiovascular physiology. Pharmacological inhibitors for these enzymes have clearly demonstrated that they play a crucial role in maintaining vascular tone in humans (16, 33, 34) as well as other species (25, 29, 38, 41) in addition to mediating cell proliferation (4, 6, 7, 18, 22, 43) and resisting inflammation (37). Using a specific inhibitor for CYP450 isoforms that metabolize arachidonic acid (AA) and an in vitro assay made up of rat cerebral endothelial cells and astrocytes in coculture (35, 53), we indirectly demonstrated that the epoxygenase enzymes may mediate angiogenesis. This result has important implications for the treatment of a number of brain injuries and may be applicable to other organs and tissue.

Molecular definition of CYP450 epoxygenases in endothelial and surrounding cells has made characterization of specific CYP450s possible in the vascular system. Our laboratory demonstrated that the rat epoxygenase 2C11 mediates astrocyte-driven functional hyperemia in the brain by RT-PCR, DNA sequencing, and antisense oligonucleotides in conjunction with functional analysis (2, 17). Also, with the use of a combination of RT-PCR and antisense oligonucleotides, the endothelium-derived hyperpolarizing factor in porcine coronary arteries was identified as epoxyeicosatrienoic acids (EETs) synthesized by the enzyme CYP2C34 (porcine)/2C8 (human) (10). Human umbilical vein endothelial cells (HUVEC) express a number of CYP450 epoxygenases of which CYP2E1, CYP3A, and CYP1A2 were not detected by Northern blotting with radiolabeled probes but were found to be present when examined by the sensitive RT-PCR method (8). Other CYP450 families present in microvascular endothelial cells are 1A, 2C, and 2J (10, 20). Overexpression of CYP2J2, another epoxygenase in human endothelial cells, resulted in decreased cytokine-induced expression of vascular cell adhesion molecule-1 (37). Knockout mice with deleted CYP4A (an AA -hydroxylase) demonstrated a hypertensive phenotype that was more severe in males (21). However, molecular characterization of an enzyme that promotes angiogenesis remains to be defined.

Many CYP450 isoforms are readily induced by pharmacological agents, including CYP2C8, which is consistently enhanced by -naphthoflavone (10). In addition, 2C8 and 2C9 are inhibited by the drug sulfaphenazole, and these properties make 2C8 and 2C9 favored targets for molecular analysis. Both isoforms have been cloned and characterized (10, 11). Although epoxygenase isoforms such as members of the 1A family as well as 2J2 might be present in pulmonary endothelial cells, we chose the epoxygenase 2C9, which is found in human endothelial cells (9, 11) and has been described to metabolize AA to EETs (51), with the most abundant product being 14,15-EET (51), as a candidate to test the angiogenic properties of epoxygenases. Because this EET profile of 2C9 matches that of the lung where 14,15-EET is also the most abundant endogenous EET (52), we used it in conjunction with primary cells in culture from human lung microvascular endothelium (HMVEC-L) in a model of in vitro tube formation. Pharmacological regulation of the 2Cs will enhance the ability to follow up on results obtained with these isoforms. To manipulate 2C9 levels in HMVEC-L, we constructed a recombinant adenovirus carrying the coding sequence of the enzyme in sense and antisense orientations and used it along with a control virus expressing the reporter green fluorescent protein (GFP). We have previously been successful in using adenoviral vectors to deliver functional CYP450 epoxygenases to cells in culture as well as in whole animals (29, 31). This study supports a role for epoxygenase 2C9 in mediating proliferation and tube formation of cultured HMVEC-L in vitro. It also demonstrates that 14,15-EET, a product of this enzyme, promotes angiogenesis in vivo.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Growth and culture of HMVEC-L. The HMVEC-Ls were purchased from Clonetics (Walkersville, MD) and maintained exactly as recommended by the manufacturer. Cells were cultured with the EGM-2-MV bullet kit containing growth factors that maintain the phenotype of the cells as judged by acetylated LDL uptake, von Willebrand factor staining, and surface platelet endothelial cell adhesion molecule (PECAM) expression. The cells are primary cultures and will remain differentiated for up to 15 passages if maintained as described above. We did not use cells beyond passage 13 and substituted the basal media EBM (Clonetics) as the serum-free medium (SFM) described in some experiments only in the final stages of the experiment. The cells were grown in a tissue culture water-jacketed incubator under 95% air-5% CO₂ following Biosafety 2 levels for human samples as well as recombinant adenovirus.

Construction of functionally expressing recombinant Ad5-2C9. The full-length coding sequence of 2C9 was restricted with EcoRI, cloned into the EcoRI site of pAdTrack-CMV (kindly provided by Bert Vogelstein, Ref. 19), and checked for orientation by restriction mapping. This construct was recombined into the adenovirus genome by homologous recombination in Escherichia coli as described (19) at the Adenoviral Core Facility at the Medical College of Wisconsin, with a few modifications. Briefly, the shuttle vectors carrying CYP2C9 (sense orientation) was digested with PmeI and dephosphorylated by incubation at 37°C for 30 min with 0.5 units of alkaline phosphatase (calf intestinal mucosa) obtained from Amersham Pharmacia Biotech (Piscataway, NJ). The enzyme was inactivated by heating to 85°C for 15 min. The linear DNA was purified by phenol-chloroform as described (19) and then gel purified using Quantum Prep Freeze'N Squeeze DNA Gel Extraction Column (Bio-Rad; Hercules, CA). This fragment was transformed along with pAdEasy-1 into E. coli BJ5183 and kanamycin (70 µg/ml)-resistant colonies analyzed by restriction digestion with PacI. Plasmids having inserts were transformed into E. coli DH10B for large-scale isolation.

Construction of Ad5-2C9 antisense (Ad5-2C9AS). The 2C9 cDNA (11) was subcloned in an adenovirus shuttle plasmid pCA3 (obtained from Microbix Biosystems; Toronto, Canada) by restriction to completion with EcoRI. The orientation of recombinant shuttle plasmid pCA3-2C9AS (antisense) was determined by restriction analysis. The shuttle plasmid with the insert in antisense orientation was restricted with ClaI, and the 2C9-containing fragment along with adjacent DNA was purified from the gel. This fragment was ligated into the ClaI-digested adenovirus helper dl 327 (12) to generate an intact virus by ligation in vitro. Subsequent transfection in the permissive host cell line HEK 293 generated plaques containing recombinant virus that were tested for 2C9 sequence by PCR with 2C9-specific primers. The plaque was purified by two rounds of subplaquing (13, 14, 27). This resultant virus along with Ad5-2C9 (sense orientation) were tested by Western and Northern blotting. Constructs that expressed 2C9 sequence (detected by Northern blotting) were also tested by Western blot analysis for CYP2C9 protein. Only one orientation (sense) demonstrated increased 2C protein expression. One virus for each orientation of 2C cDNA was selected for amplification to a large-scale preparation with titer >10¹⁰ plaque-forming units (pfu)/ml by replicating the recombinant virus in HEK 293 cells and purifying the virions through two cesium-chloride gradients (13, 14).

Growth of adenovirus. The Ad5-GFP, Ad5-2C9, and Ad5-2C9AS were grown, amplified (13, 14, 27), and purified at the Adenoviral Core Facility at the Medical College of Wisconsin. Each batch of virus was assayed for toxicity in a preliminary experiment using a multiplicity of infection (MOI) of 50 viruses/cell (29).

Western blot. Endothelial cells were cultured in 35-mm dishes to 80% confluency, washed, incubated with SFM + 0.1% FBS for 6 h. Virions (MOI 1:50) were added, and after 24 h the cells were washed three times with PBS and the proteins solubilized and extracted with 50 µl RIPA buffer [50 mM Tris (pH 8.0), 150 mM NaCl, 0.5% SDS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM EDTA, and 1× protease inhibitor cocktail (Pharmingen; San Diego, CA)]. The lysates were used to estimate their protein content with the Bio-Rad DC Protein Assay Reagent. Equal amounts of protein (10 µg) from each sample were electrophoresed on a 7.5% SDS polyacrylamide gel with running buffer as previously described (2, 28). Microsomes from the rat liver (2.5 µg) were included as a positive control for the antibody. The proteins in the gel were transferred to nitrocellulose membranes as described (2). The membrane was treated with a primary antibody (1:5,000 dilution of goat anti-rat 2C13 antibody from Gentest manufactured by Daichi Pure Chemicals) for 18 h and washed three times before incubating with a secondary antibody (1:2,500 dilution of anti-goat horseradish peroxidase) for 45 min. The protein bands were developed with chemiluminescence reagents, and the film was inspected for bands corresponding to the positive control (~55 kDa molecular mass). The membrane was stripped, rinsed, and redeveloped with anti--actin monoclonal antibody Ac-74 (A5316, Sigma; St. Louis, MO).

Northern analysis. HEK 293 cells were infected with Ad5-GFP and Ad5-2C9AS for 48 h at a MOI of 10. Northern blotting was carried out as described before (28, 32), and total RNA was extracted from the cells using the TRIZol reagent (GIBCO-BRL; Gaithersburg, MD) as specified by the manufacturer. Equal amounts of denatured RNA from each sample (20 µg) were loaded on a 1% formaldehyde agarose gel as described (28, 32) and electrophoresed in MOPS buffer (20 mM MOPS, pH 6.8, 5 mM sodium acetate, and 1 mM EDTA) at 10 V/cm, visualized under UV light, documented in a Vistra Fluorimager, and blotted onto Nytran Plus Membrane (Schleicher and Schuell, Keene, NH) using a TurboBlotter (Schleicher and Schuell). The blots were prehybridized at 42°C for 3 h with 5 ml of buffer A [formamide 50%, 5× SSPE (150 mM NaCl, 10 mM NaH₂PO₄, and 1 mM EDTA), 1× Denhardt's solution, 10% dextran sulfate, 0.1% SDS, and 100 µg/ml denatured salmon sperm DNA] and probed overnight at the same conditions with denatured labeled cDNA, which was prepared as follows: the cloned EcoRI fragment of 2C9 cDNA was isolated after appropriate restriction and electrophoreses on a 1% agarose gel, and a single band of ~2 kb was purified using Quantum Prep Freeze'N Squeeze DNA Gel Extraction Column (Bio-Rad). This fragment was used as a cDNA probe by labeling 25 ng of it with [³²P]dCTP (3,000 Ci/mmol, NEN Life Science Products; Boston, MA) using Ready-To-Go -dCTP DNA-labeling beads (Amersham Pharmacia Biotech). The labeled product was separated from unincorporated nucleotides by Microspin S-200 HR columns (Amersham, Pharmacia Biotech) and over 10⁶ counts/min put in the prehybridization solution after the probe was boiled for 5 min to denature it. After overnight hybridization, the filters were washed three times at room temperature with 2× SSC (0.3 M NaCl and 0.03 M sodium citrate, pH 7.0)-0.1% SDS followed by two washes with 0.1× SSC-0.5% SDS at 65°C. They were exposed to X-ray film (Kodak X-Omat AR) for autoradiography and developed in an automatic developer.

Thymidine incorporation. Primary cultures of HMVEC-Ls were transferred into 24-well tissue culture clusters at a density of 20,000 cells/well. After 24 h, the wells were infected with a control virus (Ad5-GFP) or Ad5-2C9 at a MOI of 50 in SFM + 2% FBS. After 18 h, [³H]thymidine (2 µCi/well) was added and allowed to incorporate for 4 h. The wells were washed three times with PBS (Sigma) and treated with ice-cold 15% trichloroacetic acid for 30 min at 0°C to precipitate the macromolecules. The trichloroacetic acid was removed, and the cells were washed twice with distilled water and dried in a hood. Precipitates were solubilized with 0.5 ml/well of 1 N NaOH for 20 min at 37°C. The NaOH was neutralized with an equal volume of HCl (1 M), and the contents were quantitatively removed to scintillation tubes. Scintillation fluid (2.5 ml/ml solution of Ecoscint from National Diagnostics; Atlanta, GA) was added to the samples, which were counted for 5 min each. The means ± SE were computed using SigmaStat 2.0 software and analyzed statistically in a t-test. There was a significant difference (P < 0.05) between cells expressing recombinant GFP alone versus those overexpressing GFP + 2C9.

Cell proliferation. The HMVEC-L cells were cultured in 24-well clusters (1 × 10⁴ cells/well) until they were 80% confluent (2 days). The medium was changed to SFM + 0.1% FBS to starve the cells for 6 h before they were infected with Ad5-GFP or Ad5-2C9 (5 × 10⁵ pfu/well) (29). The cells were growth arrested for 20 h in SFM + 0.1% FBS after which each well was treated with 25 µl of assay solution (CellTiter Aqueous One Solution from Promega; Madison, WI) for 2 h at 37°C. The absorbance in each well was measured at 490 nm in a plate reader to determine the formazan produced by this assay, which is proportional to the number of cells.

Morphogenesis assay (tube formation) in Matrigel. HMVEC-L cells were infected with Ad5-GFP, Ad5-2C9, and Ad5-2C9AS separately at MOI of 1:50 in SFM + 2% FBS for 24 h. The cells were lifted and seeded at 4 × 10⁴ cells/well into four-well Lab-Tek II chamber slides (Nalge Nunc; Naperville, IL) coated with Matrigel (Bectin Dickinson Labware; Bedford, MA). The coating of Matrigel was applied after diluting the stock (1:1) with SFM on ice to a final protein concentration of ~5.5 mg/ml. Matrigel (250 µl) was applied per squared centimeter of each well, and the matrix was allowed to gel for 30 min at 37°C before the addition of the cells, which were pretreated for some experiments as mentioned, with vehicle or epoxygenase inhibitors miconazole (10 µM) or N-methylsulfonyl-6-(2-propargyloxyphenyl) hexanamide (MS-PPOH, 20 µM, 45) for 30 min. The inhibitors remained in the samples during the experiment. All the wells were examined after an 18-h incubation period in the tissue culture incubator after which cells were carefully removed and mounted for microscopy. Each well was then scanned under low power (×10), and equal numbers (3-5/experiment) of fields with maximal tube formation were focused and captured using an Eclips 600 (Nikon) microscope with attached digital camera and SPOT software. The images were printed at a constant magnification (×300), and the length of the tubes were manually measured and summated to give the total length of tubes formed per image. Tube formation was measured after a total of three independent infections each containing multiple wells of the same viral construct.

Angiogenesis in vivo. The protocols (24, 40) were modified as follows: Sprague-Dawley rats (6 wk old) were lightly anesthetized with pentobarbital sodium (Nembutal, 50 mg/kg ip) and were injected subcutaneously with 0.7 ml Matrigel premixed with vehicle (ethanol) or 14,15-EET (150 µM). The injection was made rapidly with a 22-gauge needle to ensure the entire content was delivered in one plug. Each rat carried both a control and 14,15-EET-treated plug along the dorsal midline. The rats were allowed to recover and 7 days later the animals were anesthetized (pentobarbital sodium, 60 mg/kg ip), and the Matrigel plugs were harvested from under the skin. The plugs were homogenized in a hypotonic lysis buffer (~250 µl of 0.1% Brij-35/plug) and centrifuged for 5 min at 5,000 g. The supernatant was used in duplicate to measure hemoglobin with Drabkin's Reagent (Sigma) along with a hemoglobin standard, as directed by the manufacturer. Protein was assayed with Bio-Rad Protein Assay Reagent and compared with a known standard assayed at the same time. A total of 10 plugs carrying vehicle and control were assessed from 5 animals.

Statistical analysis. Pooled data obtained in each experiment were used to calculate the means ± SE for control (vehicle treated) or experimental groups (treated with adenovirus or pharmacological agents). The data were tested for significance by a nonpaired Student's t-test using Jandel SigmaStat softwares. All experiments with P < 0.05 and power of performed test above 0.8 were considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Expression of CYP2C9 by recombinant adenovirus, Ad5-2C9, and Ad5-2C9AS. The cloned 2C9 cDNA sequence that has been tested for activity in a number of vascular endothelial cells (9, 11) was used to generate recombinant virus. Protein products expressed by the recombinants were analyzed by Western blotting to ensure that the sense construct expressed full-length epoxygenase protein, whereas the antisense did not. The results are shown in Fig. 1A. Uninfected cells (HMVEC-L) and those infected with Ad5-GFP (control), Ad5-2C9, and Ad5-2C9AS were lysed, and equal amounts of protein were examined with goat anti-rat 2C13 antibody, which also recognizes human 2C enzymes, including 2C8 and 2C9. Rat liver microsomes were loaded as a positive control (Fig. 1A). A very strong band that migrated with the positive control was seen in the lane with proteins from the 2C9-infected HMVEC-L cells, indicating a high level of expression by Ad5-2C9. Much longer (15 times) exposures of the membrane revealed a band in uninfected cells (Fig. 1A, lane 6) implicating endogenous expression of 2C proteins in HMVEC-L cells, which was attenuated by overexpression of 2C9AS. To ensure that the Ad5-2C9AS expressed the 2C9 antisense RNA, we carried out a Northern blot with cultured human embryonic kidney cells that have been passaged extensively to downregulate endogenous epoxygenses. These cells were infected with Ad5-GFP and Ad5-2C9AS, and the total RNA from each was hybridized with ³²P-labeled 2C9 cDNA probe. A strong band at the expected position (with respect to 18S rRNA, Ref. 51) was seen in the RNA of cells infected with Ad5-2C9AS (Fig. 1B), which was absent in the control Ad5-GFP.

View larger version (25K): [in this window] [in a new window]   Fig. 1.   A: Western blot for 2C proteins derived from human microvascular endothelial cells from lungs (HMVEC-Ls). Lane 1, uninfected cells (control); lane 2, cells infected with Ad5-GFP; lane 3, cell infected with Ad5-2C9; lane 4, cells infected with Ad5-2C9 antisense (Ad5-2C9AS); lane 5, male rat liver microsomes (positive control); lanes 6 and 7, overexposed images of lanes 1 and 4, respectively. Note dark band in cells infected with 2C9, which is of comparable size to the 2C13 band in rat liver microsomes. Last two lanes depict overexposure of uninfected cells (lane 6) and cells infected with Ad5-2C9AS (lane 7) to show expression of endogenous 2C protein before and after expression of antisense 2C9. B: Northern blot analysis demonstrating expression of 2C9 mRNA sequence in HEK 293 cells infected with Ad5-2C9AS. Note a strong band that is absent in Ad5-GFP-infected cells, which does not get translated into protein as seen in A, lanes 4 and 7.

2C9-mediated increase in DNA synthesis in HMVEC-L. Epoxygenases have been reported to induce mitogenesis and proliferation in a number of cells, including rat cerebral microvascular endothelial cells (35). Because endothelial cell growth is an important requisite for angiogenesis, we were interested to document increased proliferation of HMVEC-L cells after intracellular delivery of 2C9 cDNA. Incorporation of [³H]thymidine was measured 18-24 h after introduction of recombinant 2C9 using the Ad5-2C9 construct with Ad5-GFP as a control in HMVEC-L cells (Fig. 2B). The infections routinely gave over 90% delivery of transgenes as determined by fluorescence (Fig. 2A), in which HMVEC-L cells were infected with Ad5-2C9 and viewed under phase contrast (Fig. 2A,i) as well as fluorescence (Fig. 2A,ii). As seen in Fig. 2, a majority of the cells expressed the GFP marker on the vector pAdEasy-Track (19) and were fluorescent. Figure 2B demonstrates the results in which DNA synthesis was stimulated significantly in the cells expressing 2C9.

View larger version (50K): [in this window] [in a new window]   Fig. 2.   A: phase-contrast (i) and fluorescent (ii) images of HMVEC-Ls after infection with Ad5-2C9. Note high frequency of gene transfer shown by presence of green fluorescent cells, which were not visible in uninfected cells or cells infected with Ad5 that does not code for GFP (not shown). B: graphical representation of increase in [3H]thymidine incorporation (expressed as counts per minute, CPM) in HMVEC-Ls infected with Ad5-2C9 versus Ad5-GFP. C: CYP2C9-mediated cell growth of HMVEC-L containing Ad5-GFP and Ad5-2C9. The optical density (OD; represented on the y-axis) is proportionate to cell number, which was determined independently in a standard curve for HMVEC-Ls by titering known amounts of cells and assaying with CellTiter Aqeous One Solution.

2C9-mediated proliferation of HMVEC-L. To measure whether cell number was also increased by overexpression of 2C9, the cells were infected with Ad5-GFP and Ad5-2C9 for 18 h before cell density was assayed with the use of the proliferation assay (CellTiter Aqeous One Solution from Promega). There was a significant increase in cell proliferation (>25%) as shown in Fig. 2C. Preliminary results (not shown) with Ad5-2C9AS did not show much decrease of cell number below the Ad5-GFP control, indicating that even in the absence of expression of endogenous CYP2C proteins, cell numbers were maintained.

Effect of CYP2C9 on HMVEC-L-tube formation in vitro. One of the characteristics of microvascular cells is to form tubes when embedded in matrices containing extracellular proteins like fibrinogen, gelatin, or Matrigel. In our hands, HMVEC-Ls were potent participants in tube formation in Matrigel and began to align and make cell-cell contact within an hour after being seeded on a matrix at a density of 4 × 10⁴ cells/cm². We, therefore, tested the effect of CYP2C9 on tube formation by first infecting the cells with Ad5-expressing recombinant products, then plating them on Matrigel for up to 18 h, and summating the length of the tubes formed in three to five defined fields of each well. Our results (see Fig. 3A) show that Ad5-2C9 was able to increase the already robust tube formation of these cells compared with Ad5-GFP-infected cells (Fig. 3A) as measured in 30 fields. The antisense orientation, however, decreased tube formation by one-half compared with GFP-expressing cells and by one-third compared with cells infected with Ad5-2C9. This result indicated that endogenous epoxygenase activity was important for tube formation, unlike the effect we observed on maintenance of cell number (see previous section) and prompted us to treat uninfected cells with an epoxygenase inhibitor before tube formation. As seen in Fig. 3B, the epoxygenase inhibitor miconazole (10 µM) significantly decreased tube formation. We repeated this assay with similar result using a more specific epoxygenase inhibitor, MS-PPOH (45) (Fig. 3C). Results are summarized in Fig. 3D,i-iii, showing all the cells participated in forming slender, contiguous tubes after infection with Ad5-2C9 (see red arrow in Fig. 3C,i), but in cells infected with Ad5-2C9AS (ii) or treated with miconazole (iii), there were shorter tubes surrounded by many cells that did not participate in the reaction (marked with black arrows). In fact, Fig. 3C,ii-iii, was captured from fields that show the best tubes in the well, which are atypical and difficult to find after infection with Ad5-2C9AS or treatment with inhibitors. Although most known angiogenic factors like VEGF and basic fibroblast growth factor are generally protein in nature, EETs is an example of a lipid that induces morphogenesis of endothelial cells. Sphingosine-1-phosphate is another example and exogenous application of 300 nM induced a five- to sixfold increase in tube formation of HUVECs (26).

View larger version (92K): [in this window] [in a new window]   Fig. 3.   A: representation of relative lengths of tubes formed by cells infected with recombinant adenovirus as marked. Note inhibition of tube formation after infection with Ad5-2C9AS compared with Ad5-GFP and Ad5-2C9. B: ~2.5-fold decrease in tube formation by uninfected HMVEC-Ls after treatment with the epoxygenase inhibitor miconazole (10 µM). C: significant inhibition of tube formation in HMVEC-L after treatment with epoxygenase-specific inhibitor N-methylsulfonyl-6-(2-propargyloxyphenyl) hexanamide (MS-PPOH, 20 µM). D: morphogenesis in Matrigel is demonstrated by fields viewed by phase microscopy showing tubes formed after treatment with Ad5-2C9 (i), Ad5-2C9AS (ii), and miconazole (iii). Note formation of long tubes marked by red arrow in A compared with the picture of tubes and nonparticipating cells (black arrows) in ii and iii. Magnification: ×100.

14,15-EET supports angiogenesis in vivo. In vitro tube formation in Matrigel is not conclusive evidence of angiogenesis, and angiogenic agents should be tested in vivo. We therefore embedded 14,15-EET (150 µM) in Matrigel and injected this mixture along with a similar plug-carrying vehicle. Both plugs were infused subcutaneously on the dorsal midline of a rat, and angiogenesis was measured a week later by retrieving the plugs and assaying hemoglobin content in each. Paired comparisions of four experiments showed that plugs carrying 14,15-EETs contained 1.6 times more hemoglobin than the vehicle-treated controls (Fig. 4A), despite the known short half-life of EETs, confirming that EETs are potent angiogens. Frozen sections of plugs were stained with anti-PECAM antibody and examined microscopically to verify the presence of endothelial-related vascularization (Fig. 4B).

View larger version (28K): [in this window] [in a new window]   Fig. 4.   A: hemoglobin measured in vehicle-containing Matrigel plugs versus those carrying 14,15-epoxyeicosatrienoic acid (EET) (150 µM). B: immunohistochemical staining of a 10-µm frozen section from a Matrigel plug treated with EET and developed with antibody against rat platelet endothelial cell adhesion molecule-1. Note invasion of the Matrigel by endothelial cells participating to form a network. Magnification: approximately ×1,000.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Endothelial epoxygenases have earned increasing recognition as vasoactive agents in resistance arterioles (4a, 12a, 36, 38, 44), including in human coronary vessels (33, 34). The most common isoforms in microvascular endothelium, as detected by RT-PCR, belong to the 2B, 2C, and 2J (10, 20) families, of which 2C8 has been identified as the EDHF. Other members of the 2C subfamily that are classified on the basis of sharing significant homology include 2C9, 2C10, and 2C19. The 2C gene family is recognized as one of the most complex human cytochrome subfamilies, and its members were among the first to be purified and characterized (50). The translated proteins of 2C9 and 2C10 differ by only two amino acids and may even represent the same gene or allelic variant (51). Their catalytic abilities are identical, and they are the most prevalent 2C isoforms in the kidney. Molecular cloning and subsequent characterization of the product of 2C10 showed 14,15-EET was the most abundant AA metobolite of this enzyme representing 52% of the total EETs [the other products being 11,12-EET (30%), 8,9-EET (17%) and 5,6-EET <1%] (51). Interestingly, the endogenous 14,15-EET concentration is also the highest of all four EET isomers in the rabbit lung (52). We therefore decided to use the cloned CYP2C9 enzyme (11) to test our hypothesis of epoxygenase-mediated angiogenesis in human microvascular endothelium derived from the lung.

In addition to EETs, cytochrome epoxygenases also generate superoxide and related free radicals (3, 11), which affect a number of cellular signaling cascades as well as normal and pathological endothelial phenotypes. Thus a study of angiogenesis by application of exogenous EETs alone or by using pharmacological inhibitors with nonspecific actions would not be as specific as using gene delivery of a single epoxygenase coding sequence and its complimentary antisense. We chose an adenoviral vector for gene delivery in preference to liposome-mediated transfer because primary cultures of human endothelial cells are notoriously resistant to transfection. The viral vectors in our hands routinely give >90% efficiency in gene transfer. Another advantage for use of viral vectors is that they can be delivered to the lung for in vivo testing in future studies. There are, however, potential disadvantages of these vectors, the most obvious being the viral proteins they carry in cis, which are not favorable to the host cell. We always included Ad5-GFP as a control in our experiments to negate this effect but realize this potential pitfall of viral vectors.

Our results showed a statistically significant stimulation in growth of HMVEC-L after 24 h of infection with CYP2C9. It is possible that a larger effect may have resulted if the viral genome was not delivered along with the CYP2C9. For example, recombinant human VEGF (2.5 ng/ml) has been reported to more than triple the cell number in bovine adrenal cortex-derived endothelial cells after 5-7 days (39). Although growth of HMVEC-L by overexpressing 2C9 for 18 h was much less, the cells continued to show growth after 18 h as measured by increase in thymidine incorporation. Epoxygenase products may not be as potent mitogens as other vascular growth factors like VEGF or they may regulate multiple biochemical pathways after introduction in a viral vector, and the growth we observed was a net result of the summation of these. Although endothelial cell proliferation and tube formation are separate events, they are sequential in angiogenesis, and our aim was to verify whether CYP2C9 induced both events. The growth-promoting ability of EETs depend on activation of a number of pathways, including mitogen-activated protein kinases and protein kinase B (Akt) (5, 7). Variations in the mechanisms in different cell types may represent cross-talk between EETs and other growth pathways. For example, epidermal growth factor increases EETs in proximal tubule cells which in turn modulates intracellular Ca²⁺ concentration to trigger mitogenesis (4, 7). The mechanisms of EET-induced proliferation of microvascular endothelial cells including possible interaction with VEGF and/or its receptor remain to be explored.

Our next goal was to assess the role of CYP2C9 in the differentiation of HMVEC-L. The results pointed to a significant attenuation by Ad5-2C9AS, which expressed a full-length antisense RNA, with very close homology to members of the 2C subfamily (2C8 and 2C9 have 85% homology at the nucleotide level). The data point out the importance of low levels of endogenous 2C epoxygenases in tube formation. It was therefore not surprising that one AA metabolite of CYP2C9, 14,15-EET, enhanced angiogenenesis in vivo. The single dose of EET was sufficient to increase functional vasculature in a Matrigel plug. We observed invasion of the Matrigel by endothelial cells (staining positive for PECAM-1) upon sectioning harvested plugs and examining the sections under the microscope (Fig. 4B). Our results show the hemoglobin content in the plugs, which is a more measurable and accepted quantitation of these data (26, 40), were increased by 14,15-EET. The concentration of 14,15-EET we used in the Matrigel was three times higher than that previously reported in the lung (147 ng/g, see Ref. 52) because this treatment was a bolus and because EETs are labile compounds. This experiment is a simplistic approach to replacing CYP2C9 with EET, because epoxygenases catalyze multiple substrates to form numerous products. However, EETs can also be released independent of de novo synthesis by epoxygenase. This was observed when EETs potentiated endothelial-dependent relaxation by being released from preformed stores (and not by action of epoxygenase, Ref. 46) that were esterified into the membrane phospholipids. In fact, the majority (>85%) of EETs synthesized by epoxygenases that are present endogenously in mammalian tissues are incorporated in cellular glycerophospholipids (50) and released on stimulation with varying agonists.

Results demonstrating that overexpression of epoxygenase enhance growth and differentiation of endothelial cells have important implications in normal as well as pathological microvascular physiology. Vasoactive agents such as bradykinin stimulate release of EETs from endothelial cells (23, 30), and this function may, in addition to mediating vascular tone, also promote a functional endothelium especially after damage in vascular disease. EETs have been shown to be neuroprotective after experimental transient ischemic attack, and this could be via the growth and angiogenic effects following ischemia. Similarly, EETs also mediate protection to ischemia-reperfusion injury in the heart (48). In addition, CYP450 enzymes in the vasculature may be regulated by constant ingestion of pharmacological drugs or exposure to toxins making study of their cardiovascular effects very significant.

Besides CYP2C other epoxygenases also contribute to EET biosynthesis in vascular endothelium (reviewed recently, Ref. 50), although the relative contributions by CYP2C versus molecules like CYP2J remains unknown. Overexpression of CYP2J2 prevented leukocyte adhesion to the vascular wall by inhibiting NF-B (37) and IB kinase. It also increased cAMP-mediated tissue plasminogen activator promoter activity (33). More recently, introduction of CYP2J2 in cultured bovine aortic endothelial cells protected against hypoxia-reoxygenation injury (49), increasing the list of epoxygenase-driven vascular homeostatic functions.

In conclusion, we have evidence showing angiogenic properties of a human epoxygenase, CYP2C9. This is the first report of a specific CYP450 isoform to induce both growth and differentiation of microvascular endothelium and promote angiogenesis. Because epoxygenases are inducible (10), they can now be targeted experimentally to initiate angiogenesis in pathological conditions. Also because of the multiple functions of EETs, molecular characterization of different epoxygenases will help to choose feasible candidates for gene therapy.