Angiotensin II-Induced Constrictions Are Masked by Bovine Retinal Vessels

Transcription

1 Angiotensin II-Induced Constrictions Are Masked by Bovine Retinal Vessels Prasad S. Kulkarni, Humera Hamid, Michelle Barati, and Djenita Butulija PURPOSE. TO unmask the vasoconstricting effect of angiotensin II (Ang 10 on retinal smooth muscle by studying its interaction with endothelium-derived paraine substances. This study focused specifically on determining the changes in vascular diameter and the release of endothelial-derived vasodilators, nitric oxide (NO) and prostaglandin (PG) I 2, from isolated retinal miovessels. METHODS. Bovine retinal central artery and vein were cannulated, and arterioles and venules were perfused with oxygenated/heparinized physiological salt solution at 37 C. This ex vivo perfused retinal miocirculation model was used to observe the contractile effects of Ang II on arterioles and venules of different diameters. The NO and PGI 2 synthase inhibitors, 1-NOARG and flurbiprofen, respectively, were used to unmask Ang II vasoconstriction; the changes in vascular diameters were then measured. Enzyme immunoassays were used to measure the release of cgmp (an index of NO release) and 6-keto-PG-F la (a stable metabolite of PGI 2 ) from isolated bovine retinal vessels. RESULTS. Topically applied Ang II (1O~ 10 M to 10~ 4 M) caused significant (P < 0.05) arteriolar and venular constrictions in a dose-dependent manner, with the smallest retinal arterioles (7 ± 0.2 jam luminal diameter) and venules (12 ± 2 /am luminal diameter) significantly more sensitive than larger vessels. After the inhibition of endogenous NO and PGI 2 synthesis by 1-NOARG and flurbiprofen, respectively, the vasoconstriction effects of Ang II became more pronounced. Again, the smallest vessels tested were significantly more sensitive, and synthesis of endothelial-derived relaxing factor (EDRF), therefore, may be most important in these vessels. Vasoactive doses of Ang II (10~' M to 10~ 4 M) caused a dose-dependent inease in the release of NO and PGI 2 from isolated bovine retinal vessels, indicating that the inease in EDRF may nullify direct Ang II-induced vasoconstriction. Interestingly, intraluminal administration of Ang II caused only vasodilation. NCSIONS. This study demonstrates that the retinal vascular endothelium acts as a buffer against the vasoconstricting agent Ang II via release of vasodilators NO and PGI 2, and the vasoconstriction effects due to Ang II are most prominent in the smallest diameter vessels. (Invest Ophthalmol Vis Sci. 1999:40: ) The aim of present study was to enhance our knowledge regarding the mechanisms that regulate blood flow in the retina. The vascular endothelium plays a major role in the regulation of vascular reactivity and tone. 1 " 3 Endothelial cells regulate the contractile state of smooth muscle cells, and thus the arterial tone, by the release of paraine substances, such as nitric oxide (NO), prostaglandin (PG) I 2, and endothelin. 1 One of the most important circulating peptides, which induces the release of paraine substances via its interaction with the vascular endothelium and thus plays a ucial role in the regulation of vascular tone, is angiotensin II (Ang II). 2 " 9 From the Department of Ophthalmology and Visual Sciences, School of Medicine, University of Louisville, Kentucky. Supported by a Grant-in-Aid from the Kentucky Heart Association, Louisville, Kentucky, an affiliate of the American Heart Association, Dallas, Texas; the Kentucky Lions Eye Research Foundation, Louisville, Kentucky; and an unrestricted grant from Research to Prevent Blindness, New York, New York. Submitted for publication June 22, 1998; revised October 7, 1998; accepted November 4, Proprietary interest category: N. Reprint requests: Prasad S. Kulkarni, Department of Ophthalmology and Visual Sciences, School of Medicine, University of Louisville, Louisville, KY Angiotensin II is a potent vasoconstrictor; however, there appears to be considerable heterogeneity in response of large vessels and miovessels to Ang II, even within the same vascular bed. For example, in 1987 it was observed that die topical application of Ang II failed to constrict first-order (Al) and second-order (A2) arterioles (with luminal diameters of 126 ± 6 jam and 79 8 /am, respectively) of the rat emaster muscle. 3 However, in 1982 it was observed that although A2 arterioles (86 ± 2 /am luminal diameter) of the rat emaster muscle did not constrict in response to locally applied Ang II, third-order arterioles (A3; 17 ± 2 jam luminal diameter) did constrict significantly. 6 In retinal vascular beds, disepancies about vascular contractile effects of Ang II still exist. In the past it has been shown that Ang II fails to have any effect even at high doses in human, cat, and bovine retinal blood vessels. 10 " 12 This is an unexpected finding, because synthesis of Ang II and the presence of Ang II-binding sites in these vessels have been demonstrated. 13 ' 14 Although Ang II is a potent vasoconstrictor, vascular endothelium has been shown to contribute to its differential vasoactive properties in blood vessels. For instance, Ang II is known to stimulate the release of PGI 2, PGE 2, and NO (all potent vasodilators) from vascular endothelium. 115 " 17 These vasodilators may nullify direct vasoconstricting effects of Ang Investigative Ophthalmology & Visual Science, March 1999, Vol. 40, No. 3 Copyright Association for Research in Vision and Ophthalmology 721

2 722 Kulkarni et al. IOVS, March 1999, Vol. 40, No. 3 II. In fact, the Al and A2 rat emaster arterioles, which did not show a contractile response to Ang II alone, constricted in response to the topical application of Ang II after PG synthesis was inhibited by cyclooxygenase inhibitors such as indomethacin or mefenamic acid. 3 In most miocirculatory beds, blood flow is regulated via vasoactive substances (such as NO, PGs, and endothelin) and adrenergic and cholinergic autonomic innervations. However, an interesting point is that retinal miocirculation is not innervated by adrenergic nerves, despite the presence of a- and /3-adrenergic receptors. Thus, the endothelial-derived relaxing factors play a ucial role in retinal miocirculation regulation. M The primary goal of this study was to further investigate the endothelium-dependent response of the retinal miovasculature to various concentrations of Ang II. Based on our current knowledge of Ang II effects on other vascular beds, discussed previously, we formulated the hypothesis that Ang II probably stimulates the release of NO and PGs into the retinal vascular bed and counterbalances the vasoconstriction effect. Second, with the blockage of the release of these vasodilators, the vasoconstrictor effect of Ang II should be more pronounced in smaller arterioles than in larger ones, because the smaller the diameter, the greater the percentage of constriction will be. And, finally, we wanted to study the correlation between the concentration of Ang II applied and the amount of NO and PGs released. We searched for the answers to these questions via two routes. An ex vivo perfused bovine retinal miocirculation model 1819 was used to observe the contractile effects of Ang II on retinal arterioles and venules of different diameters. Second, effects of Ang II on release of NO and PGs (PGI 2 in this case) from isolated bovine retinal vessels were determined via enzyme immunoassays. MATERIALS AND METHODS Perfused Bovine Retinal Miocirculation Model Bovine eyes were obtained through a local slaughterhouse and on removal were immediately placed in aflaskcontaining 150 ml of ice-cold heparinized physiological salt solution (PSS) and transferred to the laboratory. The PSS contained heparin (1%; U/l) and the following chemical compounds (in millimoles): NaCl, 130; NaH 3, 14.9; KC1, 4.7; KH 2 PO 4, 1.17; CaCl 2, 1.6; MgSO 4, 0.7; dextrose, 5.5; and EDTA, The globe was dissected free of any muscle and placed in a Petri dish containing 50 ml ice-cold PSS that was heparinized and oxygenated with O 2 (95%) and 2 (5%). An incision on the bovine eye was made in the sclera just posterior to the limbus, and the scleral portion of the eye was then completely excised. The eye was then lifted up by the optic nerve, and the anterior portion was pulled off the eye with slight pressure. The vitreous body was carefully removed using forceps. After the vitreous body was removed, oxygenated PSS was added to the resultant cup to prevent the retina from drying out. The retina was then carefully teased free of the choroid with a cottontipped applicator. The anterior edge of the retina was first completely separated from the choroid before working the rest of the way down to the optic nerve. The detached retina was then placed in the muscle chamber containing PSS bubbled with 95% O 2 and 5% 2. The retina was then stretched aoss the muscle chamber, and small pieces of clay were used to delicately tack the retina down, exposing the central artery and vein. Once the retina had been completely tacked down, an incision on the central retinal artery and one on the retinal vein were made using mioscissors. Close to the optic nerve, the retinal artery and retinal vein were cannulated using toothless forceps. The free end of the arterial-side cannula was connected to a minipulse pump, and the free end of the venular cannula was placed in a measuring cylinder. Then, whole retinal circulation was perfused at 20 jal/min with oxygenated/ heparinized PSS. At 20 /xl/min perfusion rate, the perfusion pressure was 80 mm Hg (Kulkarni et al. 18 and Kulkarni and Payne 19 ). The perfused retinal vasculature preparation in heparinized and oxygenated PSS was observed using closed-circuit video mioscopy as follows. The preparation was positioned on the portable stage of a triocular mioscope, and light was passed through the optical port in the bath through the tissue and into the mioscope. The image of the mio vessels was transferred via a video camera to a video monitor and simultaneously stored on videotape for analysis at a later time. Images (magnification, X0) of Al and VI (just above the first branch), A2 and V2 (just above the second branch), and A3 and V3 (subsequent branches) were recorded on the television seen and used for analysis. Before and after drug administration, the vascular diameters (at the same site for each vessel) were measured by a caliper of the selected vessels at the same site to determine changes in diameters in each branch. Bovine retinal arterioles were identified on the basis of their branching pattern. The primary arteriole supplying the retina (i.e., central retinal artery) was designated the first-order arteriole (Al). Arterioles branching from Al were defined as A2, and subsequent branches were termed A3 arterioles. Venules branching from the primary retinal vein (VI) were designated as V2 and subsequent branches as V3. In observations from 24 bovine retinal preparations, luminal diameters of Al, A2, and A3 arterioles ranged from 45 ± 2 /am, 32 ± 2 /urn, and 7 ± 0.2 jam, respectively. Diameters of VI, V2, and V3 were 65 ± 2 jam, 45 ± 1 jam, and 12 ± 1 /xm, respectively (n = 24). Retinal Vascular Responses to Vasoactive Compounds in the Presence and Absence of Metabolic Inhibitors Angiotensin II, saralasin (Ang receptor antagonist), flurbiprofen (a cyclooxygenase inhibitor), and 1-NOARG (a NO-synthase inhibitor; Sigma Chemical; St. Louis, MO) were used. Retinal arteriolar preparations are equilibrated in heparinized and oxygenated PSS at 37 C for 1 hour before cumulative dose responses to extraluminal or intraluminal applications of Ang II. After each dose, a minimum 10-minute recording period was allowed to attain the maximum response before the next dose was applied. After completion of each dose response, the preparation was washed with fresh heparinized and oxygenated PSS at 15-minute intervals for 1 hour to allow arteriolar diameters to return to baseline. Then, either saralasin (10~ 7 M), or flurbiprofen (10~ 6 M), or 1-NOARG (10~ 7 M), or both flurbiprofen and 1-NOARG were added to the bathing medium at least 15 minutes before obtaining a second dose-response curve to Ang II.

3 IOVS, March 1999, Vol. 40, No. 3 Ang n Effects on Bovine Retinal Miovasculature 723 TABLE 1. Effects of Ang II on Bovine Retinal Arterioles (Al, A2, A3) and Venules (VI, V2, V3) Ang 11 Concentration OM) saralasin Al ± 0 ± 0 97 ± 0.8* 96 ± 0.03* 95 ± 0.2* 93 ± 0.3* ± 0* A2 ±0 ± 0 96 ± 0.2* 93 ± 0.2* 90 ±0.4* 90 ± 0.5* ±0* % Change in Basal Luminal Diameter A3 ± 0 ± 0 95 ± 0.5* 90 ± 1* 84 ± 1* 84 ± 1* ± 0* VI ± 0 ± 0 99 ± 0.11* 98 ± 0.9* 97 ± 0.2* 96 ± 0.2* ± 0* V2 ±0 ±0 98 ± 0.2* 96 ± 0.4* 95 ±0.5* 95 ± 0.5* ±0* V3 ±0 ±0 98 ± 0.4* 95 ± 0.4* 95 ±0.2* 93 ± 0.2* ± 0* Numbers in column 1 are miomolar concentrations of Ang II and saralasin, a specific Ang II receptor antagonist, added to tissue media in a cumulative fashion; numbers in other columns represent percent change in basal luminal diameters ± SEM of six vessels. Luminal diameters of Al, A2, and A3 arterioles were 52 ± 1 jam, 27 ± 1 /xm, and 7 ± 0.2 p,m, respectively, whereas basal luminal diameters of VI, V2, and V3 were 65 1 /xm, 44 ± 1 jam, and 14 ± 0.5 jam, respectively. Percent inside arteriolar and venular basal diameters ± SEM were calculated as percent of luminal diameters prior to cumulative additions of Ang II to muscle bath. P< Effects of Ang n on Release of NO and PGI 2 from Isolated Bovine Retinal Vessels Bovine retinal vessels were isolated according to the procedure desibed previously by Kulkarni and Payne. 19 cgmp and 6-keto-PG-F lq! were used as an index for NO and PGI 2 release, respectively. Concentrations of Ang II used in this study were concentrations that induced vasoconstrictions of bovine retinal vessels in ex vivo preparations. Isolated bovine retinal vessels were incubated with different concentrations of Ang II for 30 minutes in 1 ml cell culture medium (Cellgro: Dulbecco's modified of Eagle's medium IX; Fischer Scientific; Pittsburgh, PA) at 37 C, and the reactions for PG and NO synthesis were stopped by acidifying samples. Measurements of cgmp (and index for NO) and 6-keto- PG-F, a (a stable metabolite of PGI 2 ) in tissues and media were performed as desibed in their specific enzyme immunoassay kits (Amersham Life Science, Arlington Heights, IL). Protein content of the isolated vessels was determined by Bio-Rad assay kits (Richmond, CA). Statistical Analysis Changes of vascular luminal diameters are expressed as a percentage of basal luminal diameter for each branch, whereas cgmp and 6-keto-PG-F, a content are expressed as per milligram tissue protein. Results are given as mean ± SEM with the number of preparations (control and experimental after inhibition of PG, NO, or both). Vascular diameters were compared in the same preparations. For the statistical analysis the level of significance (P < 0.05) is accepted using paired Student's t-test. RESULTS Effects of Topical Ang H on Bovine Retinal Mioarterioles and Venules Angiotensin II (0.001 JLLM to 10 /xm) applied topically has a small contractile effect on Al, A2, and A3 arterioles, with A3 vessels being the most responsive (P < 0.05). Also in these preparations, VI, V2, and V3 responded (Table 1) in smaller but significant (P < 0.05) vasoconstrictions to Ang II (0.001 /xm to 10 jixm). Angiotensin II (10 /xm)-induced vasoconstrictions in arteriolar and venular preparations were completely antagonized by the Ang II receptor antagonist saralasin (0.1 JLLM; Table 1). Vasoconstricting Effects of Topical Ang n on Bovine Retinal Arterioles and Venules after Cyclooxygenase Inhibition Figure 1 shows bovine retinal arteriolar (Al, A2, A3) and venular (VI, V2, V3) dose-response curves of topical Ang II (10 M to 10~ 4 M) in the presence and absence of flurbiprofen (10~ 6 M; a cyclooxygenase inhibitor). Angiotensin II-induced vasoconstriction curves were obtained before the addition of flurbiprofen. This figure shows that there was a significant (P < 0.05) and dose-dependent vasoconstriction of all arterioles and venules to Ang II (10~ 10 M to 10~ 4 M). When PG synthesis was inhibited byflurbiprofen,the vasoconstriction to topical Ang II was slightly enhanced in Al, A2, VI, and V2 vessels. However, in the smaller vessels (A3, V3), Ang II-induced constrictions were significantly ineased after flurbiprofen treatment (n = 6, P < 0.05) and were dose dependent in the 10" 10 M to 10~ 7 M range. Vasoconstrictor Effects of Topical Ang n on Bovine Retinal Arterioles and Venules after NO Synthase Inhibition Figure 2 demonstrates the dose-response curves of bovine retinal arterioles (Al, A2, A3) and venules (VI, V2, V3) to Ang II. All arterioles and venules (n = 6) responded with significant dose-dependent vasoconstrictions to topical Ang II (10~ 10 M to 10~ 4 M). Inhibition of NO synthase by 1- NOARG significantly enhanced Ang II-induced vasoconstrictions in all vessels (n = 6, P < 0.05). When 1-NOARC was added to the bathing medium before Ang II, it caused significant vasoconstriction of all vessels. A3 and V3 vessels were significantly more responsive to this than were the larger vessels. Angiotensin II-induced constrictions were dose dependent over the range of 1O~ 10 M to 10~ 7 M in these vessels.

4 724 KuUkarni et al. IOVS, March 1999, Vol. 40, No A1 ( ^ n r V1 70 rr 65 H I I I I I I I < < A2 rr s. CD V2 rr rr < g z A3 ANG II + FRBI -ANG II H a z H 1 I H ANG II + FRBI -ANG II V3 H 1 \- -I 1 1 -logangiotensin II [M] FIGURE 1. Dose-response curves of bovine retinal arterioles (Al, A2, A3) and venules (VI, V2, V3) to topical Ang II (10" 1O M to 10~ 4 M) in the absence and presence of (1CT 6 M) flurbiprofen (Flurbi). The basal inside diameters of Al, A2, A3 and VI, V2, and V3 were 53 ± 2 /xm, 34 ± 2 /xm, 8 ± 2 /am and 74 ± 2 /am, 53 ± 3 /urn, 15 ± 1 /xm, respectively. Percent inside vascular diameters were calculated as percent of basal inside diameters before cumulative Ang II addition to the muscle bath. Points represent mean of six vascular preparations in the absence and presence of flurbiprofen (Flurbi). Error bars, ±SEM. Vasoconstricting Effects of Topical Ang n on Bovine Retinal Arterioles and Venules after Cyclooxygenase and NO Synthase Inhibition Figure 3 shows that topical Ang II (1O~ 10 M to 10~ 4 M) caused significant vasoconstriction of arterioles (Al, A2, A3) and venules (VI, V2, V3) in a dose-dependent manner (P < 0.05, n = 6). Flurbiprofen and 1-NOARG combined significantly enhanced the vasoconstriction of all vessels to Ang II doses (n = 6,P < 0.05). Flurbiprofen and 1-NOARG added together caused marked vasoconstriction of all vessels before Ang II treatment. Interestingly, from Figures 2 and 3 it appears that these two combined drug treatments resulted in significantly greater vasoconstriction of only A3-level arterioles, whereas Ang II-induced responses were dose dependent over the concentration range of 1O~ 10 M to 10~ 7 M in A3 and V3 vessels. Intraluminal Administration of Ang H Figure 4 illustrates contractile effects of Ang II on bovine retinal miocirculation when perfused through retinal central artery. Ang II (10~ 10 M to 10~ 4 M) perfusion resulted in a

5 IOVS, March 1999, Vol. 40, No. 3 Ang n Effects on Bovine Retinal Miovasculature V1 H CD 3 m Q Z D Z a A-ANG II + I-NOARG -O-ANG II H A-ANG II + I-NOARG -O-ANG II H log ANGIOTENSIN II [M] FIGURE 2. Dose-response curves of bovine retinal arterioles (Al, A2, A3) and venules (VI, V2, V3) to topical Ang n (10~ 10 M to 10~ 4 M) in the absence and presence of 10~ 7 M 1-NOARG. The basal inside diameters of Al, A2, A3 and VI, V2, and V3 were 45 ± 1 /xm, 30 ± 5 jam, and 8 ± 2 /xm and 68 ± 1 /xm, 43 ± 2 /xm, and 11 ± 1 /xm, respectively. Percent inside vascular diameters were calculated as percent of basal inside diameters before cumulative Ang II additions to the muscle bath. Points represent mean of six vascular preparations in the absence and presence of 1-NOARG. Error bars, ±SEM. It should be noted that 1-NOARG applied topically before Ang II treatment caused vasoconstriction of all arterioles and venules. significant vasodilation of Al, A2, and A3 arterioles and V3 venules without affecting VI or V2 venules (n = 5, data not shown). It should be noted that these vessels are not precontracted and that the maximum vasodilation was observed at the lowest dose (1O~ 10 M) tested. Angiotensin II did not cause vasoconstriction even at the highest dose (10~ 4 M) administered intraluminally. Inhibition of Ang DT-Induced Vasoconstriction by Saralasin In six preparations, topical Ang II (10~' M to 10~ 4 M) in the presence of flurbiprofen caused significant vasoconstriction responses, especially of A3 and V3 retinal vessels, which were completely antagonized by topical administration of the Ang II antagonist saralasin (10~ 7 M; data not shown).

6 726 Kulkarni et al. IOVS, March 1999, Vol. 40, No V H h 5 5 CQ OLAR tr Q Z> Cj a a-^ A A2 a _J < CD JLAR Z > U z t V2 \ N_ r\ ci a a \ 95- \ -*-ANG II + FRBI + I-NOARG ' \ -a-angii K \ A3, X X -O O a q a^ 95- W * \ V \ ^b o_o a a ^ ^ V3 III 75- " -A-ANG II + FRBI +I-NOARG O-ANG II 80 - i i i i I I I log ANGIOTENSIN [M] FIGURE 3. Dose-response curves of bovine retinal arterioles (Al, A2, A3) and venules (VI, V2, V3) to topical Ang II (1O~ 10 M to 10~ 4 M) in the absence and presence of 1-NOARG (10~ 7 M) plus flurbiprofen (Flurbi; 10~ 6 M). The basal inside diameters of Al, A2, A3 and VI, V2, V3 were 50 ± 1 /un, 30 ± 1 ^m, and 9 ± 1 ju-m and 68 ± 3 /xm, 44 ± 2 ixm, and 13 ± 1 jxm, respectively. Percent inside vascular diameters were calculated as percent of basal inside diameters before cumulative Ang II addition to the muscle bath. Points represent mean of six vessel preparations in the absence and presence of inhibitors. Error bars, ±SEM. It should be noted that 1-NOARG and Flurbi applied topically before Ang II treatment caused vasoconstriction of all arterioles and venules. Effects of Ang n on Release of NO and PGI 2 Angiotensin II (10~ 8 M to 10~ 4 M) ineased the release of cgmp, used as a marker for NO, and 6-keto-PG-F la, a stable metabolite of PGI 2, from isolated bovine retinal vessels in a dose- dependent manner (Table 2). DISCUSSION This is the first study to demonstrate that Ang II causes a small but significant vasoconstriction of bovine retinal arterioles and venules via Ang II receptors in an ex vivo perfused bovine retinal miocirculation preparation. Interestingly, it has been

7 IOVS, March 1999, Vol. 40, No T log ANGIOTENSIN II [ M ] FIGURE 4. Dose-response curves of Al ( ), A2 (A), A3 ( ), and V3 (D) bovine retinal vessels to intraluminally administered Ang II (10~' M to 10"' 1 M) through the retinal central artery. The basal inside diameters of Al, A2, A3, and V3 were 48 ± 3, 30 ± 2, 7 ± 0.2, and 12 ± 2 jam, respectively. Percent inside vascular diameters were calculated as percent of inside basal diameters before retinal central intra-arterial perfusion of Ang II. Points represent mean of five vessels. Error bars, ±SEM. reported in the past that topically applied Ang II does not constrict bovine retinal arteriolar rings; however, it constricts posterior ciliary arteriolar ring in vitro. 12 Also, it has been previously demonstrated that intra-arterial injection of Ang II in cats does not affect retinal blood flow.'' In both of these studies" 12 vascular endothelial cells were intact. It has been reported that Ang II stimulates the release of potent vasodilators NO, 1 ' 2 ' 5 PGI 2, and PGE from vascular endothelium. Bovine retinal vascular endothelium also synthesizes NO, 19 ' 20 PGI 2, 21 andpge In 1987, Ang II was shown to stimulate the release of PGs from rat mesenteric vascular beds. 3 This study indicated that Ang n Effects on Bovine Retinal Miovasculature 727 Ang II does not constrict mesenteric arterioles; however, after inhibition of PG synthesis by indomethacin, constriction of these arterioles does occur due to Ang II. The study, thus, suggested that vasodilating PGs (L 2 and E 2 ) mask direct vasoconstrictor effects of Ang II in the nit mesenteric vascular bed. Now it is well known that PGE 2, PGI 2, and NO are potent vasodilators. 1 Therefore, we examined our hypothesis that little or no effect of Ang II on bovine retinal vessels is due to the stimulation of vasodilating NO and PGs release from retinal vessels. Although use of saralasin (a specific Ang II receptor antagonist) indicated no change in basal luminal diameter, the present study demonstrated that Ang II alone caused a small vasoconstriction of bovine retinal arterioles in a dose- dependent manner. After inhibition of the release of PGs by flurbiprofen, there was enhancement in Ang II-induced vasoconstrictions, especially in smaller vessels. After the inhibition of NO release, an even greater potentiation of these responses was observed in all retinal vessels. Interestingly, a significant and marked potentiation of Ang II was observed when NO and PG release was inhibited in smaller rather than larger vessels, suggesting that PGs and NO as vasodilators may play a vital role in precapillary arterioles and postcapillary venules. The cyclooxygenase inhibitor flurbiprofen alone did not cause vasoconstriction of any retinal vessels. However, 1-NOARG, a NO synthase inhibitor, significantly caused vasoconstriction of all vessels, whereas combined treatment with both inhibitors resulted in significantly greater vasoconstriction of only small arterioles at the A3 level. From these results it appears that PGI 2 is probably of no significance in larger vessels and that vasodilator)' NO is probably the only, and most important, substance in this respect. Thus, bovine retinal vessels mask Ang II-induced vasoconstriction by releasing both PG and NO. Furthermore, direct evidence to support that Ang II stimulates the release of PG (PGI 2 in this case) and NO was provided from isolated bovine retinal vessels. This showed that Ang II ineased the release of cgmp and 6-keto-PG-F- lcv (markers for NO and PGI 2, respectively) in a dose-dependent manner. In an earlier study, it was reported that vasoactive compounds like norepinephrine, histamine, and Ang II (which are water soluble) did not cause vasoconstriction; whereas fatsoluble papaverine caused vasodilation (i.e., ineased blood flow) in cat retinal circulation when injected intra-arterially.'' It was suggested that the water-soluble compounds do not oss the blood-retinal barriers and, therefore, do not affect retinal blood flow. However, it is now known that Ang 11 causes the release of vasodilators such as PGs and NO from TABLE 2. Effects of Ang II in Release of NO and PGI 2 from Isolated Bovine Retinal Vessels Ang n Dose, M Control 10" 10' " cgmp fm/mg protein 6-keto-PG-F la nivl/mg protein 3.75 ± ± ± 0.5* 5.4 ± 0.3* 7.7 ± 0.8* 2.5 ± ± ± 0.4* 7.3 ± 0.5* 4.4 ± 0.6* 10" ~ 6 M 1-NOARG 3 ± 0.8* N/A Numbers in columns represent cgmp or 6-keto-PG-F la content ± SEM, which were used as an index to determine the release of NO and PGI 2 respectively. The number of preparations used was seven, one for a control and one for each dose of Ang II. N/A, not available. * P < Significance was determined by comparing values of experimental (Ang II-treated) and control vessels.

8 728 Kulkarni et al. IOVS, March 1999, Vol. 40, No. 3 vascular endothelial cells that are able to oss the bloodretinal barrier. We tested the effects of Ang II in an ex vivo perfused bovine miocirculation preparation in which the blood-retinal barrier is intact. I8 Interestingly, in this study Ang II administered intraluminally caused only vasodilation, even at the highest dose (10~ 4 M) tested, suggesting that Ang II does not oss the retinal barrier and that it probably stimulates the release of endothelial-derived vasodilators like NO and PGs. This study may be further explored using inhibitors of PGs and NO. Although earlier studies have shown that Ang II does not cause vasoconstriction of the cat retinal vessels, those studies did not report whether intra-arterial injections of this polypeptide ineased the retinal blood flow. It is possible that the difference in the observations made in cat versus bovine studies could be due to in vitro versus in vivo studies; difference of species; or differences in the technique used. In 1982, the differential effects of Ang II on larger and smaller rat emaster muscle arterioles were reported. 6 Angiotensin II did not cause vasoconstriction of the larger arterioles (>130 /nm in diameter), whereas the smaller arterioles (<17 jam in diameter) showed vasoconstriction. In the perfused retinal miocirculation preparation we observed the effects of Ang II on the larger and smaller vessels simultaneously. The smallest arterioles and venules were significantly more sensitive to Ang II than the larger vessels, especially after PG and NO inhibition. This observation may be significant because in vascular diseases and diabetic retinopathy the blood-retinal barrier is disrupted, 22 and hyperglycemic conditions affect vascular NO/PG synthesis. 23 At such states, sensitivity of the smallest vessels to Ang II is extremely important because resultant vasoconstriction of the vessels will disrupt blood flow and thus the nutritional supply to the retinal neural tissue. NCSIONS As of today, the effects of Ang II on retinal vessels, in vivo or in vitro, are controversial. The reason is that in addition to vasoconstriction, Ang II also releases vasodilators such as NO and PGI 2 from the vascular endothelium. Thus, the net result is little or no response. This study shows that NO and PGI 2 synthase inhibition unmask Ang II-induced vasoconstrictions of the bovine retina arterioles and venules. However, it appears that NO is perhaps the only substance to play a regulatory role in the entire retinal miocirculation, whereas PGI 2 probably only contributes to smaller arterioles. The smallest bovine retinal arterioles (i.e., precapillary vessels) are more sensitive than the larger ones. This is important because these precapillary vessels are the ones more responsible for tissue blood supply. Angiotensin II causes vasoconstriction of retinal venules (this is the first study to show this; most veins and venules do not respond to this polypeptide). 3 ' 6 Retinal vascular endothelial cells play a primary role in local regulation by releasing vasodilator)' NO and PGs, which is important because retinal vessels are not innervated by autonomic nerves. These findings are important to not only understand the physiological effects of Ang II on vascular smooth muscle and endothelium but also pathophysiological conditions such as hypertension (high levels of circulating Ang II) and diabetes (dysfunctional vascular endothelium) in which Ang II may affect retinal vessels. References 1. Moncada S, Palmer RMJ, Higgs EA. Nitric oxide: physiology, pathophysiology and pharmacology. Pharmacol Rev. 1992;43:109-l De Nicola L, Blantz RC, Gabbai FB. Nitric oxide and angiotensin II: glomerular and tubular interaction in the rats./ Gin Invest. 1992; 89: Fleming JT, Harris PD, Joshua IG. Endogenous prostaglandins selectively mask large arteriole constriction to angiotensin II. Am] Physiol. 1987;253:H1573-H Brown SM, Janpol LM. New concepts of regulation of retinal vessel tone: mechanisms of ophthalmic disease. Arch Ophthalmol. 1996; 114: Ito AS, Ren Y, Juncos LA, Carretero OA. Endothelium-derived relaxing factor/nitric oxide modulates angiotensin II action in the isolated mio-perfused rabbit afferent but not efferent arteriole. JOin Invest. 1993;91: Bishop DA. Angiotensin reactivity at different arteriolar levels in skeletal muscle of hypertensive and decerebrate rats. Columbia, MO: University of Missouri; 1982: PhD dissertation. 7. Dollery CT, Hill DW, Hodge JV. The response of normal retinal blood vessels to angiotensin and noradrenaline. J Physiol. 1963; 165: Vacek L, Braveny P. Effect of angiotensin II on blood pressure and on miovascular beds in mesentery, skin and skeletal muscle of the rat. Miovasc Res. 1978; 16: Desjardins-Giasson S, Gutkowska J, Garcia R, Genest J. Effect of angiotensin II and norepinephrine on release of prostaglandins E 2 and I 2 by the perfused rat mesenteric artery. Prostaglandins. 1982;24:105-ll Rockwood EJ, Fantes F, Davis EB, Anderson DR. The response of retinal vasculature to angiotensin. Invest Ophthalmol Vis Sci. 1987;28: Aim A. Effects of norepinephrine, angiotensin, dihydroergotamine, papaverine, isoproterenol, histamine, nicotinic acid, and xanthinol nicotinate on retinal oxygen tension in cats. Acta Ophthalmol. 1972;50: Nyborg NCB, Nielsen J, Prieto D, Benedito S. Angiotensin II does not contract bovine retinal resistance arteries in vitro. Exp Eye Res. 1990;50: Ferrari-Dileo G, Davis EB, Anderson DR. Angiotensin II binding receptors in retinal and optic nerve head blood vessels. Invest Ophthalmol Vis Sci. 1991;32: Ferrari-Dileo G, Davis EB, Anderson DR. Angiotensin binding sites in bovine and human retinal blood vessels. Invest Ophthalmol Vis Sci. 1987;28: Mullane KM, Moncada S, Vane JR. Prostacyclin release induced by bradykinin may contribute to the anti-hypertensive action of angiotensin-converting enzyme inhibitors. Adv Prostaglandin Thromboxane Res. 1980;7: Gryglewski RJ, Splawinski J, Korbut R. Endogenous mechanisms that regulate prostacyclin release. Adv Prostaglandin Thromboxane Res. 1980;7: Faber JE, Harris PD, Miller PN. Miovascular sensitivity of PGE 2 and PGI 2 in skeletal muscle and decerebrate rat. Am J Physiol. 1982;243:H844-H Kulkarni PS, Roberts AM, Joshua JG, Barnes G. A novel method to assess reactivities of retinal miocirculation. Miocircul Res. 1994;48: Kulkarni PS, Payne S. Eicosanoids in bovine retinal miocirculatioa / Ocul Pharmacol Ther. 1997;13:139-l Benedito S, Prieto D, Nielsen PJ, Nyborg NCB. Role of the endothelium in acetylcholine-induced relaxation and spontaneous tone of bovine isolated retinal small arteries. Exp Eye Res. 1991;52: Robertson PL, Diane AR, Goldstein GW. Phospholipid metabolism and prostacyclin formation in retinal miovascular endothelium: stimulation by adenine nucleotide. Exp Eye Res. 1990;50: Small KW, Stefansson E, Hatchell DL. Retinal bloodflowin normal and diabetic dogs. Invest Ophthalmol Vis Sci. 1987;28: Tesfamariam B, Brown ML, Deykm D, Cohen RH. Elevated glucose promotes generation of endothelial-derived vasoconstrictor prostanoids in rabbit aorta./ Clin Invest. 1990;85: