Selective Augmentation of Prostacyclin Production by Combined Prostacyclin Synthase and Cyclooxygenase-1 Gene Transfer
Background—We tested the hypothesis that combined cyclooxygenase-1 (COX-1) and prostacyclin synthase (PGIS) gene transfer selectively augments prostacyclin production without a concurrent overproduction of other prostanoids.
Methods and Results—ECV304 cells were transfected with bicistronic pCOX-1/PGIS versus pCOX-1 or pPGIS, and prostanoids were analyzed. Contrary to the high prostaglandin E2 synthesis in pCOX-1 transfected cells, selective prostacyclin formation was noted with bicistronic plasmid transfection. Next, we determined the optimal ratio of Ad-COX-1 to Ad-PGIS by transfecting human umbilical vein endothelial cells with various titers of these 2 adenoviral constructs and determined the level of protein expression and prostanoid synthesis. Our results show that optimal ratios of adenoviral titers to achieve a large prostacyclin augmentation without overproduction of prostaglandin E2 or F2α were 50 to 100 plaque forming units (pfu) of Ad-COX-1 to 50 pfu of Ad-PGIS per cell. A higher Ad-PGIS to Ad-COX-1 ratio caused a paradoxical decline in prostacyclin synthesis.
Conclusions—Prostacyclin synthesis can be selectively augmented by cotransfecting endothelial cells with an optimal ratio of COX-1 to PGIS. Combined COX-1 and PGIS gene transfer has the potential for therapeutic augmentation of prostacyclin.
Prostacyclin (PGI2) is a potent inhibitor of platelet aggregation and a vasodilator, and it acts in concert with nitric oxide, ectonucleotidase, and other endothelial molecules to maintain vascular homeostasis and vasoprotection.1 It is synthesized primarily in vascular endothelial and smooth muscle cells after appropriate stimulation by specific agents. Its biosynthesis is catalyzed by a series of enzymes: cytosolic phospholipase A2 cleaves arachidonic acid (AA) from the sn-2 position of phospholipids, cyclooxygenase (COX) converts AA to prostaglandin (PG) H2, and PGI2 synthase (PGIS) converts PGH2 to PGI2.2 PGH2 is a precursor of several biologically active prostanoids, including PGE2, PGD2, PGF2α, and thromboxane A2.
Two COX isoforms (COX-1 and COX-2) have been identified in endothelial cells (EC): COX-1 is expressed constitutively, whereas COX-2, which is undetectable in resting cells, is induced by proinflammatory and mitogenic factors.3 4 COX expression is considered a key step in determining the capacity for the synthesis of PGI2 and other PGs.4 We previously showed by retrovirus-mediated COX-1 gene transfer that the overexpression of COX-1 in ECs is accompanied by a marked increase in prostanoid synthesis, notably PGI2 and PGE2 in response to stimulation by AA, ionophore A23187, or thrombin.5 Similarly, adenovirus-mediated COX-1 gene transfer in ECs enhanced the production of PGI2, and the direct administration of Ad-COX-1 into injured porcine carotid arteries abrogated thrombus formation, as determined by histological examinations and flow measurements, as a result of increased PGI2 production by the injured artery.6 The antithrombotic effect depended on the titer of Ad-COX-1.6
Overexpression of PGIS by gene transfer reportedly increased PGI2 production and inhibited smooth muscle cell proliferation in a rat carotid artery injury model.7 These experimental results suggest that overexpression of PGI2 synthetic enzymes may potentially be useful in treating pulmonary hypertension, peripheral vascular disease, and other vascular disorders. However, there are drawbacks to overexpressing a single synthetic enzyme such as COX-1 or PGIS for enhancing PGI2 synthesis. In COX-1 gene-transferred cells, besides augmented PGI2 synthesis, a large quantity of PGE2 is also produced. PGE2 is a proinflammatory mediator, and its overproduction may have undesirable effects.8 In PGIS-overexpressed cells, augmentation of PGI2 synthesis is limited because of low cellular levels of COX-1, which is further compromised by autoinactivation during catalysis.9 10 11 12 We postulate that the cotransfection of ECs with COX-1 and PGIS genes at appropriate ratios will shunt PGH2 through the PGIS pathway, with a selective augmentation of PGI2 synthesis.
In this report, we tested this hypothesis by cotransfecting ECs with bicistronic COX-1/PGIS plasmids and by co-transfecting ECs with different titers of replicating defective adenoviruses containing COX-1 (Ad-COX-1) or PGIS (Ad-PGIS) cDNA insert and analyzing the metabolite profile in the transfected cells. Our results indicate that cotransfection of human umbilical vein endothelial cells (HUVECs) with 50 to 100 plaque-forming units (pfu) of Ad-COX-1 and 50 pfu of Ad-PGIS per cell yielded a large increase in PGI2 without a concurrent overproduction of PGE2 or other eicosanoids.
Cell Culture and Materials
ECV304 cells were initially used in our transfection experiments before the report that they exhibit endothelial, epithelial, and bladder cancer cell characteristics.13 We noted that the ECV304 cells used in our experiments stained positively for von Willebrand factor, were capable of expressing adenovirus-transferred COX-1 and PGIS transgenes that were colocalized to the endoplasmic reticulum (as were the natively expressed enzymes), and had a large PGI2 synthetic capacity.14 Thus, ECV304 cells retain EC properties, and they were suitable for the initial feasibility test. ECV304 and human embryonic kidney (HEK)-293 cells were obtained from the American Type Culture Collection (Manassas, Va). They were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 100 U/mL penicillin, and 100 μg/mL streptomycin at 37°C in a humidified 5% CO2 atmosphere.
HUVECs were prepared from freshly obtained umbilical veins and cultured as previously described.5 We routinely used HUVECs in passages 3 through 6. Although COX-1 expressions decline with increasing HUVEC passages, we found that HUVECs from passages 3 through 6 are capable of synthesizing prostanoids that are quantitatively reduced but qualitatively similar to cells at earlier passages. Furthermore, HUVECs from passages 3 through 6 expressed adenovirus-mediated COX-1 and PGIS transgenes competently, as demonstrated by confocal immunofluorescent microscopy.14 There was no significant difference in transfection results among cells at passages 3 to 6. Cell culture media and antibiotics were obtained from BRL Life Technologies. [1-14C]-AA (55 mCi/mmol) was obtained from Amersham.
Recombinant Plasmid Construction
pCOX-1 and pPGIS were constructed in pSG5, as previously described.6 15 The bicistronic pCOX-1/PGIS was constructed in pSG5 by SalI digestion to remove the COX-1 expression cassette, and it was then further subcloned into the NdeI site of pPGIS. The final construct contained 2 expression cassettes that were driven by independent SV40 promoters (Figure 1A⇓).
Recombinant Adenovirus Production
Replication-defective adenoviruses were produced as described previously.6 16 17 The adenovirus shuttle plasmid vector pAd-cytomegalovirus (CMV), which was kindly supplied by Dr S.-H. Chen at Mount Sinai School of Medicine (New York, NY), contains a CMV promoter and a polyadenylation signal of bovine growth hormone. The recombinant adenovirus (rAd) was prepared by cotransfecting HEK-293 cells with pAd-CMV containing the candidate cDNAs in expression cassettes and pJM17, which was kindly provided by Dr. L. Chan at Baylor College of Medicine (Houston, Tex), using an Effectene (Qiagen) transfection system. Two to 3 weeks after transfection, rAd plaques were picked, propagated, and screened for specific cDNA sequence using polymerase chain reaction and protein expression by Western blot analysis. A large-scale production of high titer rAd was performed as described previosly,6 with minor modifications.
HEK-293 cells grown to ≈95% confluence were infected with rAd for 36 to 44 hours, harvested by centrifugation, and resuspended in fresh culture medium. After freezing and thawing 5 times, cells were centrifuged to remove cell debris. The supernatant was collected, and rAd was harvested by CsCl gradient ultracentrifugation. The opalescent band containing viral particles was collected, loaded onto the top of 1.33 g/mL CsCl, and centrifuged again at the same condition for 18 hours. The opalescent band recovered was dialyzed 3 times against 1 liter of buffer containing 10 mmol/L Tris (pH 7.4), 1 mmol/L MgCl2, and 10% (v/v) glycerol at 4°C for 18 hours. Virus stocks were separated into aliquots and stored at −80°C. Viral titers were determined by a plaque-assay method. The HEK-293 cells were infected with serially diluted viral preparations and then overlaid with low-melting-point agarose after infection. Numbers of plaques formed were counted within 2 weeks. Plaque-forming units per cell (pfu/cell) are referred to as multiplicity of infection (moi).
Extraction and Analysis of AA Metabolites
ECV304 cells were transfected with recombinant plasmids by lipofectamine (Gibco). Forty-eight hours after plasmid transfection and 24 hours after rAd infection, cells were washed and incubated in serum-free Dulbecco’s modified Eagle’s medium containing 10 μmol/L [1-14C] AA at 37°C for 10 minutes. The media were collected, and eicosanoids in the media were extracted by Sep-Pak Cartridge (Waters Associates), as previously described.18 The extracted eicosanoids were analyzed by reverse-phase high-pressure liquid chromatography (HPLC), as previously described.19 The eicosanoid peaks were identified by the retention time of the authentic radiolabeled standards. The quantity of each eicosanoid peak was determined by relating the integrated area (mV · s) of the peak to the standard obtained from authentic radiolabeled eicosanoids and AA. A 1000-mV · s integrated area was equivalent to 6.18 ng of AA, 7.44 ng of 6-keto-PGF1α, 7.12 ng of PGE2, 7.16 ng of PGF2α, and 5.66 ng of 12-hydroxy-5,8,10-heptadecatrienoic acid (HHT), respectively.
Western Blot Analysis
A total of 15 μg of cell lysate proteins were applied to each lane and analyzed by Western blots, as previously described.20 PGIS antibodies21 and COX-1 antibodies (Santa Cruz) were each diluted to 1:2000. Peroxidase-conjugated anti-rabbit or anti-mouse IgG (1:2000 dilution) was used as the second antibody to detect PGIS and COX-1 bands, respectively, by enhanced chemiluminescence (Amersham).
To evaluate the feasibility of our proposal, we initially transfected ECV304 cells with pCOX-1, pPGIS, or the bicistronic pCOX-1/PGIS and determined the protein expression and the arachidonate metabolites in the transfected cells. Basal COX-1 and PGIS protein levels were low in untransfected or pUC18-transfected ECV304 (Figure 1B⇑). There was a concentration-dependent increase in COX-1 and PGIS when cells were transfected with 1, 2, or 4 μg of bicistronic pCOX-1/PGIS (Figure 1B⇑). The COX-1 protein level expressed in cells transfected with 4 μg of pCOX-1/PGIS was comparable to that with 4 μg of pCOX-1, whereas the PGIS protein level in pCOX-1/PGIS–transfected cells was only ≈50% of that in pPGIS-transfected cells (Figure 1B⇑).
There was a marked difference in the HPLC profile of cells transfected with pCOX-1/PGIS versus pCOX-1 or pPGIS. Transfection of cells with pCOX-1 (4 μg) increased PGE2 synthesis by ≈100-fold over control levels (Figure 2⇓ and Table 1⇓). In contrast, pCOX-1/PGIS transfection increased 6-keto-PGF1α levels in a concentration-dependent manner, without a concurrent PGE2 increase (Figure 2⇓ and Table 1⇓). The 6-keto-PGF1α level produced by cells transfected with 4 μg of pCOX-1/PGIS was ≈12-fold higher than that produced by cells transfected with 4 μg of pCOX-1 (Figure 2⇓ and Table 1⇓). HHT production was increased in cells transfected with pCOX-1 and pCOX-1/PGIS. These transient transfection results support the notion that co-overexpression of COX-1 and PGIS redirects PGH2 through the PGIS pathway.
Because ECV304 cells possess epithelial cell and bladder cancer cell characteristics, we performed subsequent experiments in cultured HUVECs. We transfected HUVECs with a mixture of Ad-COX-1 and Ad-PGIS at different pfu ratios to determine whether the relative quantities of prostanoids produced are influenced by different ratios of COX-1 to PGIS overexpression. HUVECs were transfected with 50 moi (pfu/cell) of Ad-COX-1 or Ad-PGIS alone, a fixed 50 moi of Ad-COX-1 with 10 to 100 moi of Ad-PGIS, or a fixed 50 moi of Ad-PGIS with 10 to 100 moi of Ad-COX-1. The transfected cells were lysed, and COX-1 and PGIS protein levels were determined by Western blot analysis. The transgenic COX-1 and PGIS protein levels were markedly elevated by individual Ad-COX-1 and Ad-PGIS transfection, respectively, compared with untransfected or Ad-CMV controls (Figure 3⇓). Combined Ad-COX-1 and Ad-PGIS transfections produced COX-1 and PGIS protein levels comparable to those of individual Ad-COX-1 or Ad-PGIS transfections (Figure 3⇓).
Eicosanoids generated by transfected cells treated with [1-14C]AA (10 μmol/L) are shown in Figure 4⇓. The left panel shows the eicosanoid profile of cells transfected with a fixed 50 moi of Ad-COX-1 combined with 0 to 100 moi of Ad-PGIS, and the right panel shows the control vector profile and the profile of cells transfected with a fixed 50 moi of Ad-PGIS and 0 to 100 moi of Ad-COX-1. Quantitative data on the major prostanoid peaks shown in this figure are displayed in Table 2⇓. PGE2 was the major product (Figure 4⇓); it constituted 78% of the total prostanoids produced, whereas PGF2α and 6-keto-PGF1α made up 13% and <1%, respectively, of the total prostanoids produced by HUVECs transfected with Ad-COX-1 (50 moi) alone (Table 2⇓). By contrast, 6-keto-PGF1α was the predominant peak in cells transfected with combined Ad-COX-1/Ad-PGIS (Figure 4⇓).
Cotransfection of HUVECs with 100 moi of Ad-COX-1 and 50 moi of Ad-PGIS yielded the highest 6-keto-PGF1α level, which constituted 85% of total prostanoids (Table 2⇑). When Ad-PGIS moi was in excess of Ad-COX-1 moi, the 6-keto-PGF1α and the total prostanoids produced were markedly reduced (Figure 4⇑ and Table 2⇑). For example, the 6-keto-PGF1α and the total prostanoid levels produced by cells transfected with 50 moi of Ad-COX-1 plus 100 moi of Ad-PGIS were only 44% and 40%, respectively, of those produced by cells transfected with 50 moi each of Ad-COX-1 and Ad-PGIS (Table 2⇑). HHT levels were increased by Ad-COX-1 transfection and by Ad-COX-1/Ad-PGIS cotransfection but not by Ad-PGIS transfection (Table 2⇑). The highest value of HHT was produced by cells transfected with Ad-COX-1/Ad-PGIS in a 50/10 to 50/20 ratio (Table 2⇑). Its level was reduced when the Ad-PGIS titer was in excess of the Ad-COX-1 titer. Only trace amounts of hydroxyeicosatetraenoic acid (HETE)-like eicosanoids were detected, and their values did not vary by combined transfections. Therefore, we did not characterize these HETE-like peaks further.
Results from this study demonstrate that PGI2 synthesis can be augmented without overproduction of other prostanoids by cotransfecting ECs with Ad-COX-1/Ad-PGIS. The extent of PGI2 augmentation depends on the ratio of Ad-COX-1 to Ad-PGIS titers. Our data with HUVECs indicate that transfection with 100 moi of Ad-COX-1 and 50 moi of Ad-PGIS produced the highest PGI2 augmentation when PGE2 and PGF2α levels remained very low (only slightly above the basal level). This is in contrast to the large increase in PGE2 synthesis seen in HUVECs transfected with Ad-COX-1 alone. Because the total prostanoids produced by Ad-COX-1 versus Ad-COX-1/Ad-PGIS–transfected cells are equivalent, the data are consistent with the concept that an equivalent amount of PGH2 is produced by the overexpressed COX-1, but a majority of PGH2 is metabolized via the PGIS pathway when there is an abundant PGIS. Our results provide direct evidence to support the hypothesis that the formation of diverse prostanoids from PGH2 in cells is governed by the level of the final synthetic enzymes. Further, they suggest that combined gene transfer has a general application for engineering selective augmentation of prostanoid synthesis.
It is intriguing that PGI2 synthesis declines significantly when the Ad-PGIS titer is in excess of the Ad-COX-1 titer. At Ad-PGIS excess, the level of HHT, a breakdown product of PGH2, was proportionally reduced, which is consistent with reduced PGH2 synthesis. Because no major peaks at the HETE fractions were detected in individual or combined transfections, it is unlikely that the reduced PGH2 is due to a shunt to the lipoxygenase pathways. The mechanism by which an excess of PGIS expression reduces PGH2 generation is unclear. It may be speculated that PGI2 synthesis requires a certain spatial and quantitative relationship between COX-1 and PGIS located on the endoplasmic reticulum membrane and an excessive PGIS expression may perturb this relationship, thereby hindering PGI2 synthesis. We recently showed that overexpressed COX-1 and PGIS,14 like constitutively expressed enzymes,22 are colocalized to the endoplasmic reticulum. COX-1 binds to the luminal endoplasmic reticulum membrane by hydrophobic interactions, whereas PGIS anchors to the cytosolic side of the endoplasmic reticulum membrane through a single transmembrane domain.21 It has been presumed that COX-1 and PGIS are located close to each other to facilitate metabolite transfer, but the quantitative and spatial relationship of these 2 enzymes are unclear and require further investigation.
The quantity of HHT produced by cotransfected cells was higher than that of control cells and proportionally higher than that of PGE2 or PGF2α (Table 2⇑). HHT is generated from PGH2 by several potential pathways. Thromboxane synthase catalyzes the conversion of PGH2 to HHT, malondialdehyde, and thromboxane A2 in a 1:1:1 ratio.23 Cytochrome P450 was recently reported to convert PGH2 to HHT and malondialdehyde.24 HHT can also be converted from PGH2 by a nonenzymic pathway. Cultured HUVECs possess very low levels, if any, of thromboxane synthase activity. It is unlikely that HHT is derived from this pathway. However, we cannot exclude the possibility that HHT was derived from cytochrome P450 or from nonenzymic conversion.
Several clinical trials have shown that PGI2 and its stable analog iloprost are efficacious in treating primary pulmonary hypertension.25 26 27 A recent randomized, multicenter, controlled trial provides objective evidence for the beneficial effect of beraprost, a stable analog of PGI2, on peripheral vascular disease.28 PGI2 is likely to have broad effects on diverse vascular disorders. However, systemic administration of these drugs is associated with undesirable side effects. PGI2 acts locally in an autocrine and paracrine manner. It will be desirable to administer PGI2 locally at a targeted site. Local administration of PGI2 and its more stable analogs to the targeted vascular region remains a challenge because of the relatively short half-life of these drugs. Gene transfer to overexpress PGI2 synthetic enzymes, thereby augmenting PGI2 productions at a targeted site of the vascular system, has been shown to be feasible in several animal models.6 7 In addition to local PGI2 productions, gene transfer approaches have the following other advantages: (1) PGI2 is produced only when stimuli are present, and (2) PGI2 synthetic enzyme(s) are expressed for a longer period of time than drugs.
Previous experimental work used a single gene approach to augment PGI2 production. Results from the present study indicate that overexpression of COX-1 alone is accompanied by an overproduction of PGE2, which is a key proinflammatory mediator and may potentially contribute to vascular inflammation, whereas overexpression of PGIS alone has a minimal effect on increasing PGI2 synthesis. In fact, PGIS overexpression may suppress PGI2 synthesis. In contrast, combined COX-1/PGIS gene transfer at appropriate ratios selectively augments PGI2 productions and should be a better approach than the single gene transfer approach. Work is in progress in our laboratory to evaluate the effects of combined Ad-COX-1/Ad-PGIS transfer on vascular lesions in appropriate experimental animal models.
In summary, it has become possible for the first time to engineer selective augmentation of PGI2 synthesis by combined transfer of Ad-COX-1 and Ad-PGIS. By selecting an optimal ratio of COX-1 to PGIS overexpression, it is possible to shunt a majority of PGH2 through the PGIS pathway. Combined COX-1/PGIS gene transfer is potentially useful in gene therapy for diverse vascular diseases.
This work was supported by grants from the Department of Health and the National Science Council of Taiwan under “The Frontier Program of Medical Genetics Research” (DOH89-TD-1132) and grants from the National Institutes of Health (P50-NS 23327 and RO1-HL 50675). We thank Angela Wang and Susan Mitterling for secretarial assistance.
The first 2 authors contributed equally to this work.
Guest Editor for this article was Joseph Loscalzo, MD, PhD, Boston University School of Medicine, Boston, Mass.
- Received June 16, 2000.
- Revision received November 8, 2000.
- Accepted November 20, 2000.
- Copyright © 2001 by American Heart Association
Wu KK, Kulmacz RJ, Wang L-H, et al. Molecular biology of prostacyclin biosynthesis. In: Rubanyi G, Vane JR, eds. Prostacyclin: New Perspectives for Basic Research and Novel Therapeutic Indications. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1992:11–23.
Wu KK. Inducible cyclooxygenase and nitric oxide synthase. Adv Pharmacol.. 1995;33:179–207.
Smith WL, Marnett LJ. Prostaglandin endoperoxide synthase: structure and catalysis. Biochem Biophys Acta.. 1990;1083:1–14.
Xu X-M, Ohashi K, Sanduja SK, et al. Enhanced prostacyclin synthesis in endothelial cells by retrovirus-mediated transfer of prostaglandin H synthase cDNA. J Clin Invest.. 1993;31:1843–1849.
Zoldhelyi P, McNatt J, Xu X-M, et al. Prevention of arterial thrombosis by adenovirus-mediated transfer of cyclooxygenase gene. Circulation.. 1996;93:10–17.
Tanaka T, Yokoyama C, Yamamoto H, et al. Gene transfer of human prostacyclin synthase prevents neointimal formation after carotid balloon injury in rats. Stroke.. 1999;30:419–426.
Hinson RM, Williams JA, Schacter E. Elevated interleukin 6 by prostaglandin E2 in a murine model of inflammation: possible role of cyclooxygenase-2. Proc Natl Acad Sci U S A.. 1996;93:4885–4889.
Egan RW, Paxton J, Kuehl FA Jr. Mechanism of irreversible self-deactivation of prostaglandin synthase. J Biol Chem.. 1976;251:7525–7535.
Kent RS, Diedrich SL, Whorton R. Regulation of vascular prostaglandin synthesis by metabolites of arachidonic acid in perfused rabbit aorta. J Clin Invest.. 1983;72:455–465.
McIntire TM, Zimmerman GA, Satoh K, et al. Cultured endothelial cells synthesize both PAF and PGI2 in response to histamine, bradykinin and ATP. J Clin Invest.. 1985;76:271–280.
Sanduja SK, Tsai A-L, Matijevic-Aleksic N, et al. Kinetics of prostacyclin synthesis in a PGHS-1 overexpressed endothelial cell. Am J Physiol.. 1994;267:C1459–C1466.
Liou J-Y, Shyue S-K, Tsai M-J, et al. Colocalization of prostacyclin synthase with prostaglandin H synthase-1 but not phorbol ester-induced PGHS-2 in cultured endothelial cells. J Biol Chem.. 2000;275:15314–15320.
Shyue S-K, Ruan K-H, Wang L-H, et al. Prostacyclin synthase active sites: identification by molecular modeling guided site directed mutagenesis. J Biol Chem.. 1997;272:3657–3662.
Gomez-Foix AM, Coats WS, Baque S, et al. Adenovirus-mediated transfer of the muscle glycogen phosphorylase gene into hepatocytes confers altered regulation of glycogen metabolism. J Biol Chem.. 1992;267:25129–25134.
Herz J, Gerard RD. Adenovirus-mediated transfer of low-density lipoprotein receptor gene acutely accelerates cholesterol clearance in normal mice. Proc Natl Acad Sci U S A.. 1993;980:2812–2816.
Eling T, Tainer B, Ally A, et al. Separation of arachidonic acid metabolites by high-pressure liquid chromatography. In: Lands WEM, Smith WL, eds. Methods in Enzymology. San Diego, Calif: Academic Press; 1983:511–517.
Sanduja SK, Mehta K, Xu X-M, et al. Differentiation-associated expression of prostaglandin H and thromboxane A synthase in monocytoid leukemia cell lines. Blood. 1991;78:3178–3185.
Wu KK, Sanduja R, Tsai A-L, et al. Aspirin inhibits interleukin 1-induced prostaglandin H synthase expression in cultured endothelial cells. Proc Natl Acad Sci U S A.. 1991;88:2384–2387.
Morita I, Schindler M, Regier MK, et al. Different intracellular locations for prostaglandin H synthase-1 and–2. J Biol Chem.. 1995;270:10902–10908.
Ohashi K, Ruan K-H, Kulmacz RJ, et al. Primary structure of human thromboxane synthase determined from the cDNA sequence. J Biol Chem.. 1992;267:789–793.
Plastaras JP, Guengerich FP, Nebert DW, et al. Xenobiotic-metabolizing cytochrome P450 convert prostaglandin endoperoxide to hydroxyheptadecatrienoic acid and malondialdehyde. J Biol Chem.. 2000;275:11784–11790.
Wensel R, Opitz CF, Ewert R, et al. Effects of iloprost inhalation on exercise capacity and ventilatory efficiency in patients with primary pulmonary hypertension. Circulation.. 2000;101:2388–2392.
Lievre M, Morand S, Besse B, et al. Oral beraprost, a prostaglandin I2 analogue, for intermittent claudication: a double-blind randomized multicenter controlled trial. Circulation. 2000;102:426–431.