INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

OXIDANT STRESS leads to the formation of reactive oxygen species, which have been implicated in numerous diseases (49). One of the main targets of oxygen free radicals is unsaturated fatty acids from cellular membranes, leading to peroxidation and cellular injury (49). In the retina, peroxidation appears to play an important role in the genesis of various disorders such as ischemic retinopathies, most notably retinopathy of prematurity and diabetes (42, 52, 55). Oxidant stress can alter retinal hemodynamics (9) by causing marked vasoconstriction, which is sustained in the newborn (1, 3, 9). We and other investigators have shown that reactive oxygen species stimulate the cyclooxygenase pathway to produce thromboxane, which appears to be involved in this constriction (3, 9, 29, 51). However, the cascade of events leading to this production of thromboxane is not known.

A series of prostaglandin-like compounds, termed isoprostanes, have recently been shown to be produced during oxidant stresses in vivo and in vitro in animals and humans (40, 41). Isoprostanes are produced independently of the cyclooxygenase pathway, and their formation results from oxidation of the ubiquitous arachidonic acid by free radicals (40, 41). In contrast to prostaglandins formed by cyclooxygenase, the isoprostanes are formed in situ on esterified phospholipids and are released in free form presumably by phospholipases (36). Isoprostanes are stable products, and their formation increases markedly in animal models subjected to free radical injury (32, 36, 38, 46).

8-Isoprostaglandin F₂ (8-iso-PGF₂), which is an abundantly produced isoprostane in vivo, is a highly potent renal vasoconstrictor with an EC₅₀ in the low nanomolar range (38, 50). It has also been shown that 8-iso-PGF₂ constricts bronchioles, as well as coronary, pulmonary, and cerebral vessels, which can be inhibited by thromboxane receptor antagonists (7, 21, 26, 27). However, binding studies suggest that 8-iso-PGF₂ does not directly interact with the thromboxane receptor (13, 43, 56). 8-Iso-PGF₂ has also been shown to stimulate endothelin-1 release from aortic endothelial cells, but the mechanism of endothelin release is not known (14). Altogether, the mechanism of isoprostane action remains unclear, such that the potency and efficacy of 8-iso-PGF₂ in tissues studied so far vary markedly (7, 21, 26, 27, 38, 50) and cannot be extrapolated to other tissues such as the retina in the present case.

Unlike other tissues, including the brain, the retina contains a specific profile of membrane phospholipids (6). The retina and its vasculature predominantly generate prostaglandin I₂ (PGI₂) (45, 47), rather than prostaglandin E₂ (PGE₂), which is generated by most other neural and nonneural tissues (23, 30). Although in general the responses of the retinal vasculature resemble those of other surface neural vessels, namely, pial ones, the retinal vasculature exhibits distinct vasomotor responses to a variety of agents, including prostanoids and peroxides (1, 3, 4, 9, 31). Hence, because of the susceptibility of the retina to peroxidation, we proceeded to investigate the effects of 8-iso-PGF₂ on the retinal vasculature of piglets and propose a new mechanism of action for 8-iso-PGF₂.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Tissue preparation. Animals were used according to a protocol of the Animal Care Committee of Hôpital Sainte Justine in accordance with the principles of the Guide for the Care and Use of Experimental Animals of the Canadian Council on Animal Care. Piglets (1-3 days old) were obtained from Fermes Ménard (L'Ange-Gardien, PQ, Canada). Animals anesthetized with halothane (2.5%) and injected with pentobarbital sodium (90 mg/kg) were subjected to thoracotomy. Blood was removed from the circulation by perfusion with systemically heparinized (1 U/ml) saline (~200 ml; corresponding to ~2 blood volumes) through a beating left ventricle through which a 16-gauge needle was inserted; blood was concomitantly disposed from the circulation via a right atriotomy. At the end of the procedure, additional pentobarbital (30 mg/kg) was injected into the heart to ensure that the animals were killed. The eyes were removed and placed immediately in ice-cold Krebs buffer (pH 7.4) of the following composition (mM): 120 NaCl, 4.5 KCl, 2.5 CaCl₂, 1.0 MgSO₄, 27 NaHCO₃, 1.0 KH₂PO₄, and 10 glucose; 1.5 U/ml heparin was added to the buffer. Previous histological studies confirmed the absence of platelets and other blood elements in the ocular vasculature (9). For biochemical measurements, tissues were frozen in liquid N₂ and stored at 80°C.

Vasomotor response to 8-iso-PGF₂ and other agents. Eyecup preparations were used to study the response in situ of the relatively undisturbed retinal vasculature, as previously described (1, 3, 9, 18). Briefly, a circular incision was made 3-4 mm posterior to the ora serrata to remove the anterior segment and vitreous body with minimal handling of the retina. The remaining eyecup was fixed with pins to a wax base in a 20-ml tissue bath containing Krebs buffer (pH 7.35-7.45) equilibrated with 21% O₂ and 5% CO₂ and maintained at 37°C (3). The preparations were allowed to stabilize for 30-45 min, during which they were rinsed two or three times with fresh buffer.

2

2

Retinal microvascular endothelial and smooth muscle cell cultures. For primary cultures of retinovascular endothelial and smooth muscle cells, retinas were collected in Hanks' balanced salt solution (HBSS) buffer (pH 7.4) of the following composition (mM): 2.8 KCl, 0.2 KH₂PO₄, 68 NaCl, 0.16 Na₂HPO₄, 2.8 glucose, 100 HEPES, and 0.01 phenol red. Retinal microvessels were prepared as previously described (2). Briefly, retinas were gently homogenized with a Wheaton pestle in 5 mM Tris · HCl buffer (pH 7.4) containing (mM) 1.1 acetylsalicylic acid, 0.5 EGTA, 1 benzamidine, and 0.1 phenylmethylsulfonyl fluoride; 100 µg/ml soybean trypsin inhibitor was added to the buffer. The homogenate was mixed with Ficoll 400 (40%) at a 1:1 (vol/vol) ratio and centrifuged at 20,000 g for 20 min at 4°C. The pellet, which contains the microvessels, was washed in HBSS three times. Purity of the microvessel preparation was confirmed by high-power microscopy and by -glutamyl transpeptidase activity, which was higher in vasculature (5.6-6.1 mU/mg protein) than in neural parenchyma (0.3-0.35 mU/mg protein) (2).

Thromboxane, prostaglandin, and endothelin assays. The effects of 8-iso-PGF₂ on thromboxane, 6-ketoprostaglandin F₁ (6-keto-PGF₁), PGE₂, and endothelin production were also studied. Retinas were unstimulated or stimulated with 8-iso-PGF₂ (1 µM) for 15 min after pretreatment for 20 min with saline (equivalent volume of 100 µl in 15 ml bath), indomethacin (10 µM), CGS-12970 (1 µM), BQ-123 (1 µM), phosphoramidon (10 µM), SKF-96365 (20 µM), nifedipine (5 µM), or EGTA (5 mM), and the reaction was stopped with liquid N₂. Thromboxane B₂ (TxB₂, a stable TxA₂ metabolite), 6-keto-PGF₁ (a stable PGI₂ metabolite), and PGE₂ were determined as previously described (4, 18). Briefly, retinas were suspended in a cold buffer (pH 7.4) of the following composition (mM): 5 Tris · HCl, 1.1 acetylsalicylic acid, 1 EDTA, and 0.045 butyl hydroxytoluene. The tissue was homogenized with a tissue grinder (30,000 rpm, twice for 30 s); proteins were measured in aliquots by the dye-binding method (8). The homogenate was centrifuged at 1,000 g for 10 min at 4°C to remove undisrupted cells and nuclei. The supernatant was rehomogenized and then centrifuged at 28,000 g for 45 min at 4°C to remove membranes and enhance extraction of prostanoids on octadecylsilyl silica columns. The supernatant was dissolved in 100% ethanol and acidified to pH 3 with glacial acetic acid. The samples were applied to the octadecylsilyl silica columns preactivated with methanol and distilled water and subsequently washed with 15% aqueous ethanol and then with petroleum ether. Prostanoids were subsequently eluted with methyl formate and evaporated under vacuum to dryness. TxB₂, 6-keto-PGF₁, and PGE₂ were measured by radioimmunoassay as previously described (4, 18). The recovery of prostanoids was 95%, and the interassay variability was <5%. A similar procedure was used to measure prostanoid levels in the culture media of retinovascular endothelial cells and smooth muscle cells stimulated with 8-iso-PGF₂ (1 µM) for 15 min in the absence or presence of indomethacin (10 µM), CGS-12970 (1 µM), SKF-96365 (20 µM), or nifedipine (5 µM).

Intracellular Ca²⁺ measurements. Intracellular Ca²⁺ concentration ([Ca²⁺]_i) was measured using the fluorescent indicator fura 2-AM. Confluent endothelial cells and smooth muscle cells were trypsinized in a solution containing 0.05% trypsin and 0.02% EDTA for 2 min, then 5 ml of HBSS were added. Cells were centrifuged at 250 g for 10 min and resuspended in a buffer containing (in mM) 20 HEPES, 10 D-glucose, 4.6 KCl, 118 NaCl, and 0.5 CaCl₂, as well as 1% fetal bovine serum. Cell viability was determined by trypan blue exclusion and was >90%. Fura 2-AM (2 µM) and 0.2% Pluronic F-127 were added to cell suspensions, which were incubated at 37°C for 30 min. The loaded cells were then washed twice and resuspended in HBSS with Ca²⁺ (2.5 mM) and 1% fetal bovine serum with or without pretreatment for 15 min with SKF-96365 (20 µM), nifedipine (5 µM), or EGTA (5 mM), followed by stimulation with 8-iso-PGF₂ (1 µM), ATP (1 µM), or U-46619 (1 µM). The [Ca²⁺]_i was determined in 2 ml of fura 2-loaded cell suspension (~2 × 10⁶ cells/ml) continuously stirred and measured by a spectrofluorometer (model LS 50, Perkin-Elmer, Beaconsfield, UK) by using excitation wavelengths of 340 and 380 nm and emission at 510 nm. Calibration of the fluorescent signal was determined on 2 ml of cell suspension by sequential addition of 0.2% Triton X-100 to obtain the maximal fluorescence ratio (R_max) and to 5 mM EGTA plus 10 µM ionomycin to obtain the minimal fluorescence ratio (R_min). Autofluorescence was determined by measuring fluorescence from nonloaded cells and subtracting it from the fluorescence produced by fura 2-loaded cells to calculate the fluorescence ratio R corresponding to the values produced at 340 and 380 nm. The [Ca²⁺]_i was calculated from the equation of Grynkiewicz et al. (16): [Ca²⁺]_i = K_d [(R  R_min)/(R_max  R)](S_f2/S_b2), where K_d (224 nM) is the effective dissociation constant of the fura 2-Ca²⁺ complex and S_f2/S_b2 is the ratio of fluorescence intensity at 380-nm wavelength in the presence of EGTA to that in the presence of Triton X-100.

Measurement of 8-iso-PGF₂ in isolated retina subjected to oxidation. Retinas were exposed in vitro to hydroxyl-generating conditions with H₂O₂ (0.1 mM) and FeCl₂ (20 µM) for 30 min in the presence or absence of indomethacin (10 µM) or the free radical scavenger dimethylthiourea (1 mM) (17). Extraction of isoprostanes was performed as described above for thromboxane, reflecting the free active unesterified isoprostanes (36). 8-Iso-PGF₂ was measured by enzyme immunoassay technique with a commercial kit (Cayman Chemical, Ann Arbor, MI). Briefly, dried samples containing extracted 8-iso-PGF₂ were reconstituted in 250 µl of 100 mM phosphate buffer (pH 7.4) with 1.5 mM NaN₃, 0.4 M NaCl, 1 M EDTA, and 1 g/l BSA. Fifty-microliter samples were placed in microtiter wells precoated with mouse monoclonal anti-rabbit IgG; then 50 µl of acetylcholinesterase linked to 8-isoprostane (tracer) were added, and the samples were incubated for 18 h at room temperature. Unbound reagents were washed five times with 10 mM phosphate buffer (pH 7.4) containing 0.05% Tween 20, and the reaction was developed in 60-90 min with the acetylcholinesterase substrate acetylthiocholine as well as DTNB (Ellman's reagent). Plates were read spectrophotometrically at 405-420 nm to assay formation of 5-thio-2-nitrobenzoic acid generated by the reaction of enzymatically formed thiocholine with DTNB. The specificity for the assay was 100% for 8-isoprostanes, with 85 ± 3% specificity for 8-iso-PGF₂ and <1% for 8-iso-PGE₂ tested in our laboratory (n = 4); cross-reactivity of antibody with TxB₂, PGE₂, and PGF₂ was 0.1%. The intra- and interassay variability was 5%.

Measurement of 8-iso-PGF₂ in retina of anesthetized piglets subjected to oxidant stress. Oxidant was produced in vivo by subjecting animals to asphyxia followed by reoxygenation, as previously described (9). Piglets were anesthetized with 2% halothane for tracheostomy and catheterization of a femoral vein for drug injection. Animals were ventilated with air by means of a Harvard small animal respirator. Halothane was discontinued, and immediately thereafter piglets were sedated with -chloralose (50 mg/kg bolus followed by 10 mg · kg¹ · h¹) and paralyzed with pancuronium (0.1 mg/kg twice). Animals were kept under a radiant warmer to keep body temperature at 38°C. After 1.5 h of recovery from surgery, piglets were asphyxiated by interruption of ventilation for 5 min. Ventilation was resumed normally for 45 min, and the animals were then killed (120 mg/kg iv pentobarbital sodium) and the eyes were removed. Retinas were collected to measure 8-iso-PGF₂, as described above.

Chemicals. L-670596 and CGS-12970 were generous gifts from Merck-Frosst (Pointe-Claire, PQ, Canada) and Ciba-Geigy (Summit, NJ), respectively. The following products were purchased: 8-iso-PGF₂ (>99% pure), 8-iso-PGF₂ enzyme immunoassay kit, and U-46619 (Cayman Chemicals); acetylsalicylic acid, ATP, aprotinin, benzamidine, butylated hydroxytoluene, econazole, EDTA, EGTA, FeCl₂, H₂O₂, dimethylthiourea, OPPC, indomethacin, ionomycin, nifedipine, N-(-rhamnopyranosyloxyhydroxyphosphinyl)-Leu-Trp (phosphoramidon), PMSF, soybean trypsin inhibitor (type II-S), Triton X-100, and Tris · HCl (Sigma Chemical, St. Louis, MO); 1-[-[3-(4-methoxyphenyl)propoxy]-4-methoxyphenethyl]-1H-imidazole hydrochloride (SKF-96365; BioMol, Plymouth Meeting, PA); endothelin-1 and cyclo(-D-Trp-D-Asp(ONa)Pro-D-Val-Leu) (BQ-123; Research Biochemicals, Natick, MA); N-cis-2,6-dimethylpiperidinocarbonyl-L-MeLeu-D-Trp(COOMe)-D-Nle-ONa (BQ-788), fura 2-AM, and Pluronic F-127 (Calbiochem, La Jolla, CA); TxB₂, 6-keto-PGF₁, and PGE₂ radioimmunoassay kits (Amersham, Oakville, ON, Canada); endothelin enzyme immunoassay kit (Peninsula); endothelium and smooth muscle growth medium (Clonetics); factor VIII antibody, smooth muscle-specific actin antibody, and glial fibrillary acidic protein antibody (Dako, Carpinteria, CA); FITC-conjugated goat anti-rabbit antibody, FCS, goat serum (Jackson Immunoresearch Laboratories, West Grove, PA); all other chemicals (Fisher Scientific, Montreal, PQ, Canada).

Data analysis. Results were analyzed using Student's t-test and a two-way ANOVA, with factoring for concentrations and different treatments. Post-ANOVA comparisons among means were performed using the Tukey-Kramer method (48). Statistical significance was set at P < 0.05. Values are means ± SE.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Vasoconstrictor effects of 8-iso-PGF₂, U-46619, and endothelin-1 on retinal arterioles. 8-Iso-PGF₂, U-46619, and endothelin-1 caused concentration-dependent constriction of retinal arterioles (Fig. 1A). Maximal efficacy of 8-iso-PGF₂ was nearly one-half that of U-46619 and endothelin-1. The EC₅₀ of 8-iso-PGF₂, U-46619, and endothelin-1 were 5.9 ± 0.5, 159 ± 26, and 0.9 ± 0.7 nM, respectively.

View larger version (29K): [in this window] [in a new window]   Fig. 1.   Constrictor concentration-response curves to 8-isoprostaglandin F2 (8-iso-PGF2), thromboxane A2 (TxA2) receptor mimetic U-46619, and endothelin-1 on retinal arterioles of piglets in absence (A) and presence of cyclooxygenase inhibitor indomethacin (10 µM), phospholipase A2 blocker oleoyloxyethyl phosphocholine (OPPC, 50 µM), TxA2 synthase inhibitor CGS-12970 (1 µM), and TxA2 receptor antagonist L-670596 (100 nM) (B); ETA receptor antagonist BQ-123 (1 µM), ETB receptor antagonist BQ-788 (25 µM), and endothelin-converting enzyme inhibitor phosphoramidon (10 µM) (C); and L-type voltage-gated Ca2+ channel blocker nifedipine (1 µM) and putative receptor-mediated Ca2+ channel blocker SKF-96365 (20 µM) (D). Values are means ± SE of 4-7 separate experiments. * P < 0.05 compared with effects of 8-iso-PGF2 in absence of antagonists (by 2-way ANOVA).

Roles of thromboxane and endothelin in 8-iso-PGF₂-induced constriction of retinal arterioles. Retinal vasoconstriction to 8-iso-PGF₂ was markedly inhibited by the cyclooxygenase blocker indomethacin and the phospholipase A₂ inhibitor OPPC, as well as the thromboxane synthase inhibitor CGS-12970 and the thromboxane receptor antagonist L-670596 (Fig. 1B).

2

2

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Effects of 8-iso-PGF2 (1 µM) on endothelin-1 (ET-1) formation by retina (A) and thromboxane B2 (TxB2) and 6-ketoprostaglandin F1 (6-keto-PGF1) production by retina (B and C) and cultured retinal endothelial cells (D and E). Tissues and cells were pretreated for 20 min with saline or inhibitors as follows: indomethacin (10 µM), CGS-12970 (1 µM), BQ-123 (1 µM), phosphoramidon (10 µM), SKF-96365 (20 µM), nifedipine (5 µM), and EGTA (5 mM). Values are means ± SE of 4-6 experiments. * P < 0.05 compared with basal values;  P < 0.05 compared with saline (+8-iso-PGF2).

Effects of Ca²⁺ channel blockers on vasoconstriction produced by 8-iso-PGF₂. Because formation of thromboxane and release of endothelin were suspected to contribute to effects of 8-iso-PGF₂ (on the basis of inhibition of constriction by CGS-12970 and phosphoramidon), we speculated that Ca²⁺ entry, which is required for the generation and release of these autacoids, may be involved in the action of 8-iso-PGF₂. Because L-type voltage-gated Ca²⁺ channels play a major role in the entry of Ca²⁺ in smooth muscle and other cell types as well as in vascular contraction (20), their contribution to the 8-iso-PGF₂ effect was tested using the selective blocker nifedipine. Nifedipine (1 µM) did not modify retinal vasoconstriction to 8-iso-PGF₂ (Fig. 1D); fivefold higher concentrations of nifedipine also did not alter effects of 8-iso-PGF₂ but virtually abolished constriction to KCl (20 mM). In contrast, the Ca²⁺ entry blocker SKF-96365 fully inhibited the vasoconstrictor response to 8-iso-PGF₂. Econazole (10 µM), another non-voltage-dependent Ca²⁺ channel blocker (19), and SKF-96365 inhibited the constrictor effects of 8-iso-PGF₂ to an equal degree (data not shown).

Effects of 8-iso-PGF₂ on thromboxane, prostaglandin, and endothelin release. 8-Iso-PGF₂ stimulated endothelin release (Fig. 2A) from the retina, which was inhibited by pretreatment with phosphoramidon and significantly reduced by SKF-96365. 8-Iso-PGF₂ also increased production of TxB₂ (~7-fold) and, to a lesser extent (by ~2-fold), 6-keto-PGF₁ and PGE₂ (latter not shown) in the retina (Fig. 2, B and C). These effects were markedly diminished by indomethacin, by the Ca²⁺ chelator EGTA, and by the Ca²⁺ entry blocker SKF-96365; as expected (5), the thromboxane synthase inhibitor CGS-12970 reduced formation of TxB₂, but not 6-keto-PGF₁ and PGE₂. Basal (unstimulated) TxB₂ and 6-keto-PGF₁ formation was not affected by SKF-96365, as previously shown (28), but was reduced by ~80% by indomethacin; CGS-12970 reduced basal TxB₂ by ~70%. 8-Iso-PGF₂-induced prostanoid production was slightly attenuated by BQ-123 (1 µM) and unaltered by nifedipine (5 µM; Fig. 2, B and C).

1

2

1

1

1

2

1

2

Effects of 8-iso-PGF₂ on intracellular Ca²⁺ transients in endothelial and smooth muscle cells from retinal vasculature. 8-Iso-PGF₂ elicited an increase in intracellular Ca²⁺ in endothelial cells (Fig. 3, A and B). This effect of 8-iso-PGF₂ on endothelial cells was not significantly affected by nifedipine but was virtually abolished by SKF-96365 and annulled by EGTA and absence of extracellular Ca²⁺. However, 8-iso-PGF₂ did not cause an increase in intracellular Ca²⁺ in smooth muscle cells, whereas ATP was effective (Fig. 3, C and D). Conversely, U-46619 stimulated an increase in intracellular Ca²⁺ in smooth muscle cells but not in endothelial cells (Fig. 3, B-D).

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Effects of 8-iso-PGF2 and U-46619 on intracellular Ca2+ transients in endothelial and smooth muscle cells from retinal vasculature. Typical tracings and peak increases in intracellular Ca2+ concentrations ([Ca2+]i) are presented in endothelial (A and B) and smooth muscle cells (C and D). Cells were pretreated for 15 min with SKF-96365 (20 µM), nifedipine (5 µM), EGTA (5 mM), or saline with or without extracellular Ca2+ (2.5 mM) before addition of 8-iso-PGF2 (1 µM), ATP (1 µM), or U-46619 (1 µM). Values are means ± SE of 4-6 experiments.  P < 0.01 compared with saline (+8-iso-PGF2).

8-Iso-PGF₂ formation in retina subjected to oxidation. Finally, generation of 8-iso-PGF₂ was verified in the retina after in vitro and in vivo oxidant stress. Incubation of retinas with H₂O₂ and FeCl₂ caused a marked increase in 8-iso-PGF₂ generation, which was prevented by the free radical scavenger dimethylthiourea and unaffected by indomethacin (Fig. 4A). An increase in 8-iso-PGF₂ concentrations was also detected in the retina of animals subjected to an asphyxic episode followed by reoxygenation (Fig. 4B).

View larger version (12K): [in this window] [in a new window]   Fig. 4.   8-Iso-PGF2 synthesis in retina of piglets subjected to in vitro and in vivo oxidant stress. A: retinas incubated for 30 min without (control) or with H2O2 (100 µM) and FeCl2 (20 µM) with or without 15 min of pretreatment with indomethacin (10 µM) or dimethylthiourea (DMTU, 1 mM). B: 8-iso-PGF2 in retinas of piglets subjected to 5-min asphyxic episode followed by 45 min of reoxygenation. Values are means ± SE of 4 experiments. * P < 0.01 compared with control or preasphyxia.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Isoprostanes are stable oxidation products of arachidonic acid produced by a free radical mechanism in numerous tissues (32, 36, 37, 40, 41). One of the F₂-isoprostanes that has been shown to be produced in vivo is 8-iso-PGF₂ (39); in this study we showed that 8-iso-PGF₂ is produced by the retina when subjected to an oxidant stress in vitro and in vivo (Fig. 4). 8-Iso-PGF₂ causes constriction of various vascular beds, albeit at different efficacy and potency (7, 21, 26, 27). Although the effects of 8-iso-PGF₂ have been found to be markedly inhibited by thromboxane receptor blockers (7, 21, 26, 27, 50), binding studies suggest that 8-iso-PGF₂ does not directly interact with the thromboxane receptor (13, 43, 56) but possibly with distinct binding sites (13, 52). Briefly, the mechanisms of action of 8-iso-PGF₂ are not clearly defined. The primary purpose of this study was to investigate the effects and potential mechanisms of action of 8-iso-PGF₂ on retinal vessels. The data suggest that 8-iso-PGF₂ elicits retinal vasoconstriction by releasing endothelin and more importantly the prostanoid thromboxane from retinal parenchymal and endothelial cells after Ca²⁺ entry into cells possibly through non-voltage-dependent Ca²⁺ channels. In this context, it is of relevance that reactive oxygen species can also increase capacitative Ca²⁺ influx (11), as well as activate cyclooxygenase, and cause vasoconstriction by increasing the synthesis of thromboxane (3, 51). Thus isoprostanes may serve as mediators in peroxidation-induced vasoconstriction.

8-Iso-PGF₂ caused retinal vasoconstriction with an EC₅₀ in the low nanomolar range and with an efficacy somewhat less than that of the thromboxane mimetic U-46619 and endothelin-1. These vasoconstrictor effects of 8-iso-PGF₂ were almost completely suppressed by inhibition of synthesis and action of thromboxane and, to a significantly lesser extent, by blockers of endothelin synthesis and of ET_A receptors. In addition, 8-iso-PGF₂ stimulated the formation of endothelin and thromboxane in the retina and in retinal endothelial cells but not in retinovascular smooth muscle cells. These data suggest that thromboxane and endothelin are involved in the vasoconstrictor effect of 8-iso-PGF₂. However, the fact that thromboxane synthesis and receptor blockers were more effective than those of endothelin in antagonizing the effects 8-iso-PGF₂ implies a relatively more important role for thromboxane in the retinal vasoconstriction to 8-iso-PGF₂. Along the same lines, although PGI₂ (measured by 6-keto-PGF₁) and PGE₂ slightly increased in response to 8-iso-PGF₂ (Fig. 2, C and E), these respective retinal vasodilators and constrictors (4) contributed negligibly to vasomotor effects of 8-iso-PGF₂ compared with thromboxane, since specific inhibitors of thromboxane action (like those of cyclooxygenase) caused near abolition of 8-iso-PGF₂-induced constriction (Fig. 1B).

Equivalent inhibition of the constrictor effect of 8-iso-PGF₂ by the phospholipase A₂ blocker OPPC, the cyclooxygenase inhibitor indomethacin, and the thromboxane synthase inhibitor CGS-12970 (5) (Fig. 1B) suggests that 8-iso-PGF₂ acts on the synthesis of thromboxane, rather than on its receptors, by stimulating the release of arachidonic acid, which is metabolized by cyclooxygenase into prostanoids of which thromboxane dominates in importance. This suggestion implies that the inhibition of the vasoconstrictor effects of 8-iso-PGF₂ by the thromboxane receptor antagonist L-670596 is exerted by a blockade of the effects of thromboxane released in response to 8-iso-PGF₂. This inference is further supported by the marked increase in the synthesis of thromboxane, compared with other prostanoids, in retina and retinal endothelial cells after stimulation with 8-iso-PGF₂ (Fig. 2), consistent with previously reported effects of peroxides in the retina (1, 3, 9). Additional evidence that 8-iso-PGF₂ does not seem to act on the thromboxane receptor is provided by divergent actions of 8-iso-PGF₂ and the thromboxane mimetic U-46619 on Ca²⁺ influx in endothelial and smooth muscle cells (Fig. 3, B-D). Hence, it can be inferred that in the retina 8-iso-PGF₂ seems to interact on a binding site distinct from the thromboxane receptor, concordant with suggestions of others (13, 43, 56).

The relative contribution of the retinal parenchyma per se in generating endothelin and cyclooxygenase products is difficult to clarify because of problems in separating ex vivo tissue parenchyma from its vasculature. Nonetheless, in separate experiments we found that retinas of 1-day-old rats, which are developmentally avascular, are capable of generating thromboxane in response to 8-iso-PGF₂; these observations suggest that vasculature and parenchyma participate in 8-iso-PGF₂-induced prostanoid production in the retina. In addition, it should be noted that potentially trapped platelets are an unlikely source of TxB₂, because they are not detected histologically in ocular vasculature of perfused animals (9) and, more importantly, platelets do not generate TxB₂ in response to 8-iso-PGF₂ (43).

It is of interest that, in contrast to retinal parenchymal and endothelial cells, smooth muscle cells did not generate prostanoids or exhibit an increase in Ca²⁺ transients in response to 8-iso-PGF₂. These observations would suggest that 8-iso-PGF₂ might exert little, if any, direct action on retinovascular smooth muscle and that its vasoconstrictor effects are mediated indirectly by release of thromboxane (and endothelin) from retinal parenchymal and endothelial cells (Fig. 2). In contrast to our findings, 8-iso-PGF₂ has been shown to exert diverse actions on rat aortic smooth muscle cells, including stimulation of DNA, inositol trisphosphate synthesis, and increase in cytosolic Ca²⁺ (13, 43). The other distinct finding in our study compared with others applies to the predominance of cyclooxygenase dependence of 8-iso-PGF₂ action. Effects of 8-iso-PGF₂ have been reported to be unrelated to cyclooxygenase products in kidney and lung (7, 50) and slightly dependent on metabolites of this enzyme in aorta (54). Reasons for disparities between our study in porcine retina and those of others regarding effects of 8-iso-PGF₂ on smooth muscle cells and cyclooxygenase dependence are not clear but may be due to differences in tissues as well as species; for instance, 8-iso-PGF₂ constricts bovine, but not ovine, coronary arteries (27).

In addition to thromboxane, endothelin also appears to play a role, albeit to a lesser extent, in the vasoconstrictor response to 8-iso-PGF₂. Endothelins are potent vasoconstrictors that can be produced by oxidant stress (10). 8-Iso-PGF₂ has been shown to stimulate endothelin-1 release from aortic endothelial cells (14). In the present study the vasoconstrictor effects of 8-iso-PGF₂ were reduced to approximately the same degree by the endothelin-converting enzyme inhibitor phosphoramidon and by the ET_A-receptor blocker BQ-123. Furthermore, 8-iso-PGF₂ induced the release of endothelin from the retina. These findings suggest that vasoconstrictor effects of 8-iso-PGF₂ are partially mediated by endothelin via the ET_A receptor. Because endothelin-1 can release thromboxane (44), it is possible that a part of the endothelin-mediated effects of 8-iso-PGF₂ is exerted indirectly via thromboxane. Indeed 8-iso-PGF₂-induced thromboxane formation was reduced by BQ-123 (Fig. 2B).

Because the release of endothelin and formation of thromboxane are Ca²⁺-dependent processes, we tested the role of Ca²⁺ in the action of 8-iso-PGF₂. 8-Iso-PGF₂ elicited an increase in intracellular Ca²⁺ that was dependent on extracellular Ca²⁺ (prevented by EGTA as well as by the absence of Ca²⁺) (Fig. 3). The L-type voltage-gated Ca²⁺ channels do not seem to be implicated, since nifedipine did not inhibit the effects of 8-iso-PGF₂ on retinal vessel tone, endothelin and thromboxane release, and intracellular Ca²⁺ transients. In contrast, the vasomotor effects, the generation of endothelin and thromboxane, and the increase in intracellular Ca²⁺ by 8-iso-PGF₂ were significantly inhibited by SKF-96365, a blocker of Ca²⁺ entry, including that by non-voltage-dependent Ca²⁺ channels (33, 34).

Non-voltage-gated Ca²⁺ channels, which comprise receptor-operated channels, are for the most part not well characterized (22). In addition, their physiological role is difficult to elucidate in the absence of selective blockers. SKF-96365 has been reported to inhibit receptor-mediated Ca²⁺ entry at 30 µM; however, at >100 µM, SKF-96365 also blocks voltage-gated Ca²⁺ channels. In the present study the inhibitory effects of SKF-96365 on 8-iso-PGF₂ action were observed at <30 µM. Further evidence that non-voltage-gated Ca²⁺ channels were involved in the vasoconstrictor action of 8-iso-PGF₂ was obtained in our study with econazole, which can also block Ca²⁺ influx from non-voltage-dependent Ca²⁺ channels (19). Altogether, these data suggest that 8-iso-PGF₂ increases influx of Ca²⁺ possibly via receptor-operated channels [present in excitable and nonexcitable cells (22, 23)], which in turn leads to stimulation of endothelin release and, more importantly, activation of phospholipase A₂ and metabolism of arachidonic acid into prostanoids, among which thromboxane predominates in mediating 8-iso-PGF₂-induced retinal vasoconstriction. The involvement of a receptor-operated Ca²⁺ channel would be consistent with the existence of a distinct 8-iso-PGF₂ receptor site (13), which remains to be characterized.

In conclusion, this study reveals a retinal vasoconstrictor effect of 8-iso-PGF₂ by a previously undescribed mechanism. Our results suggest that the effect of 8-iso-PGF₂ on retinal vasculature is mediated mostly by cyclooxygenase-generated formation of thromboxane and, to a lesser extent, by endothelin, probably through non-voltage-gated cation channels. Because isoprostanes are produced in the retina during oxidant stress (Fig. 4), it is possible that 8-iso-PGF₂ may contribute to the pathogenesis of ischemia-reperfusion retinal injury such as in retinopathy of prematurity and of diabetes. Along these lines, a role for thromboxane has been proposed in ischemic retinopathies (3, 35).

Perspectives

2

2