-3 Regulation of Retinal Hemodynamics in Nondiabetic and Diabetic Rats

Fumihiko Mori, George L. King, Allen C. Clermont, Dahlia K. Bursell, and Sven-Erik Bursell

PURPOSE. To investigate the mechanisms of action of endothelin (ET)-3 on the regulation of retinal hemodynamics in diabetic and nondiabetic rats.

METHODS. Retinal blood flow changes were measured using video fluorescein angiography. Mea- surements were made before and after intravitreal injections of different ET-3 concentrations in nondiabetic rats and rats with streptozotocin (STZ)-induced diabetes. The effect of ET-3 on retinal G blood flow was also investigated in nondiabetic rats after pretreatment with N -monomethyl-L- arginine (L-NMMA), a nitric oxide synthase (NOS) inhibitor; BQ-788, an ET receptor B (ETB) antagonist; and BQ-123, an ET receptor A (ETA) antagonist. Control animals were injected intrav- itreally with vehicle alone.

RESULTS. In nondiabetic rats, ET-3 induced a dose-dependent rapid increase in retinal blood flow 2 minutes after intravitreal injection (maximal at 10Ϫ8 M, P Ͻ 0.01) followed 15 and 30 minutes after ET-3 injection by dose-dependent decreases in retinal blood flow (maximal effect at 10Ϫ6 M, P Ͻ 0.05). The ET-3–stimulated retinal blood flow increase was inhibited by 10Ϫ4 M BQ-788 (P Ͻ 0.01) Ϫ3 and 10 M L-NMMA (P Ͻ 0.05). The ET-3–stimulated decrease in retinal blood flow at later times (15 minutes) was inhibited (P Ͻ 0.03) by 10Ϫ4 M BQ-123. In diabetic rats, baseline retinal blood flows were decreased compared with nondiabetic rats (P Ͻ 0.01), showed dose-dependent increases 2 minutes after ET-3 injection (P Ͻ 0.03), and at later times remained significantly increased (P Ͻ 0.05) in contrast to flows in nondiabetic rats.

CONCLUSIONS. The ET-3–induced initial rapid retinal blood flow increase in nondiabetic rats is mediated by the ET-3/ETB and NOS action. The subsequent retinal blood flow decrease is mediated by ET-3/ETA action. Diabetic rats showed comparable ET-3–induced retinal blood flow increases indicating normal ET-3/ETB action. However, at later times, retinal blood flow remained increased, suggesting an abnormal ET-3/ETA action. (Invest Ophthalmol Vis Sci. 2000;41:3955–3962)

n understanding of the physiological regulation of ret- a potent retinal vasoconstrictor9–13 binding to the high-affinity inal hemodynamics and the maintenance of vascular ET receptor A (ETA)24 in retinal vascular smooth muscle cells A tone by endogenous vasoactive hormones and cyto- and pericytes.25,26 ET-3 also binds to the ETA but with lower kines in association with the development of retinal hemody- affinity and less vasoconstrictor action than in the ET-1/ETA namic changes in diabetes is important, especially because action. ET-1 has a role in maintaining normal vascular tone, abnormal retinal hemodynamics in diabetes can manifest in the and, in the diabetic rat retina, increased ET-1 production con- early stages of the disease, both in animals1–5 and in patients tributes to measured retinal blood flow reduction.9–11 In con- with no diabetic retinopathy,6–8 and may be associated with trast, ET-3 interacts primarily with the endothelial cell ET an increased risk for development of diabetic retinopathy. receptors type B (ETB), which have an equal affinity for both 27,28 Factors responsible for regulating vascular tone in the retina ET-1 and ET-3. ETB action initiates vasodilation through 29 include the (ETs),9–15 nitric oxide (NO),16–18 pros- NO and/or prostacyclin. Changes in tissue ET-3/ETB interac- 30,31 taglandins,19,20 and angiotensin.21,22 However, the role of tions in diabetes have been reported ; however, the ET-3/ these factors in the development of abnormal retinal hemody- ETA/ETB interactions in the retinal hemodynamics in diabetes namics in diabetes is not well defined. are not well characterized, prompting the current investiga- ETs are potent vasoactive agents,23 and in the retina, ET-1 tion. and ET-3 appear to play a role in vascular homeostasis. ET-1 is METHODS From the Research Division and Beetham Eye Institute, Joslin Instrumentation Diabetes Center, Harvard Medical School, Boston, Massachusetts. Supported by National Institutes of Health Grant EY09178 and by The video fluorescein angiography (VFA) system used for these a grant from the Massachusetts Lions Eye Research Fund, Inc. studies has been described previously1–4 and consists of a Submitted for publication January 20, 2000; revised July 6, 2000; fundus camera (NFC-50; Nikon, Tokyo, Japan) interfaced to a accepted August 14, 2000. Commercial relationships policy: N. video camera (SIT; Dage-MTI, Michigan City, IN). The video Corresponding author: Sven-Erik Bursell, Joslin Diabetes Center, camera output was directly digitized (512 ϫ 512 pixels by 8 One Joslin Place, Boston, MA 02215. [email protected] bits) at 30 frames/sec and stored in a data storage system

Investigative Ophthalmology & Visual Science, November 2000, Vol. 41, No. 12 Copyright © Association for Research in Vision and Ophthalmology 3955

Downloaded from iovs.arvojournals.org on 09/29/2021 3956 Mori et al. IOVS, November 2000, Vol. 41, No. 12

(Trapix Plus/DataStore; Recognition Concepts, Carson City, in vehicle of 2.5% Emulphor EL-620 (GAF Chemical, Wayne, NV). The digitized angiogram images were analyzed on a frame- NJ) in phosphate-buffered saline (PBS). Rats injected intravit- by-frame basis to determine the retinal circulatory parameters really with vehicle alone served as control subjects. of interest. VFA recordings were obtained before and at 2, 5, 15, and 30 minutes after intravitreal injection. Blood pressures and Animals heart rates were monitored using a noninvasive tail-cuff sensor One hundred three male Sprague–Dawley rats (Taconic Farms, and monitoring system (Ueda Electronics, Tokyo, Japan). Ani- Germantown, NY) with initial weights between 200 and 250 g mals were maintained on a heated pad during the course of the were used. All experiments were performed in accordance measurements. with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Animal Care NOS Inhibitor and ETB Antagonist Action and Use Committee of the Joslin Diabetes Center. Diabetes was Thirty-two nondiabetic rats were used. Pretreatment was per- induced in 34 rats with an intraperitoneal injection of 65 formed with intravitreal injections of either 10Ϫ3 M NG-mono- mg/kg of streptozocin (STZ; Sigma, St. Louis, MO) in 10 mM methyl-L-arginine (L-NMMA; Sigma) a nitric oxide synthase citrate buffer (pH 4.5) after a 12-hour fast. Diabetes was con- (NOS) inhibitor, or 10Ϫ4 M BQ-788 (Sigma), a specific ETB firmed with blood glucose measurements (Ͼ250 mg/dl) 24 antagonist. For both agents, the vehicle was 2.5% Emulphor Ϫ3 Ϫ5 hours after STZ injection. The rats were housed under standard EL-620 in PBS. Pretreatments with 10 M L-NMMA (8 ϫ 10 conditions with free access to water and standard food. All M effective vitreous concentration) or 10Ϫ4 M BQ-788 (8 ϫ animals were maintained for 2 weeks before retinal blood flow 10Ϫ6 M effective vitreous concentration) were used to ensure measurements. Blood glucose levels and body weights were maximal NOS33 and ETB inhibition.34 VFAs were recorded at monitored every other day. baseline and 15 minutes after intravitreal pretreatment with Twenty-four hours before retinal blood flow measure- L-NMMA, BQ-788, or vehicle. Each rat then underwent an ments, all animals (under anesthesia, 0.1 mg/kg amobarbital intravitreal injection of 10Ϫ8 M ET-3 (maximal retinal hemody- sodium; Eli Lily, Indianapolis, IN) underwent catheterization namic response), and subsequent VFA recordings were ob- with a polyvinyl catheter inserted into the right jugular vein.5 tained at 2 and 15 minutes after ET-3 injection. The catheter was flushed with 0.1 ml of 1000 U sodium hep- arin before and after implantation. It was positioned subcuta- ETA Antagonist Action neously along the shoulder, and the distal end was externalized Nine nondiabetic rats were used. Pretreatment was performed to the back of the neck. with intravitreal injections of 10Ϫ4 M BQ-123 (American Pep- tide, Sunnyvale, CA), a specific ETA antagonist with vehicle VFA Procedure consisting of 2.5% Emulphor EL-620 in PBS. At this concentra- Immediately before VFA measurements, each rat was anesthe- tion, prior results10 have shown a maximal retinal blood flow tized, the left eye was dilated (1% tropicamide, Mydriacyl: increase at 5 minutes after intravitreal injection with a return to Alcon, Fort Worth, TX), and a syringe (Hamilton, Reno, NV) baseline values by 15 minutes after injection. In five animals, containing 10% sodium fluorescein was connected to the ex- baseline VFA recordings were obtained followed by an intrav- ternalized jugular vein catheter. The rats were positioned on a itreal injection of BQ-123. VFA recordings were repeated 2 platform attached to the retinal fundus camera. The optic disc minutes after the BQ-123 injections. Immediately after these was centered and focused in the field of view, the VFA record- recordings, an intravitreal injection of 10Ϫ7 M ET-3 was per- ing sequence was initiated, and a 5-␮l bolus of fluorescein dye formed, and VFA recordings were repeated at 2, 5, 15, and 30 was rapidly injected into the jugular vein catheter.4,5 The minutes after the ET-3 injection. In four animals, only BQ-123 injection time was marked on the video recording. was injected at baseline and VFA recordings were performed at Baseline angiograms were recorded from each rat before 2, 5, 15, and 30 minutes after injection. intravitreal injection with the different agents under investiga- tion. A further series of angiograms were then recorded at Data Analysis selected time points after the intravitreal injection. The recorded fluorescein angiograms were digitized on a Intravitreal injections were performed by inserting a 27- frame-by-frame basis and analyzed densitometrically to deter- gauge needle, attached to a 10-␮l syringe (Hamilton), into the mine retinal vessel diameters and retinal mean circulation vitreous from a site 1 mm posterior to the limbus. Infusion was times (MCTs).4,5 performed directly over the optic disc region under direct Vessel diameters in units of pixels were determined from visualization, and a timer was started. VFA recordings were images recorded before fluorescein dye injection at defined obtained at selected times after injection. The effective vitreal vessel sample sites using a boundary-crossing algorithm. The concentrations of the injected agents were estimated knowing average vessel diameters for each eye represent the average of 32 that the rat vitreous volume is approximately 120 ␮l. Thus the individual vessel diameters for that eye. the retina would be exposed to a 12-fold lower concentration At the fixed vessel sites, the average vessel fluorescence than the injected concentration. within a sample area defined by the vessel width was measured on a frame-by-frame basis to generate temporal fluorescence Time Course and Dose Response of ET-3 in intensity or dye dilution curves. The resultant artery and vein Nondiabetic and Diabetic Rats fluorescence data were fit to a log normal distribution func- Intravitreal injections in 34 STZ-induced diabetic rats and 37 tion5 from which average arterial and venous circulation times nondiabetic rats were performed using different concentra- were calculated. The arterial appearance time (AT) of the dye tions (10Ϫ9 to 10Ϫ6 M) of ET-3 (Sigma, St. Louis, MO) dissolved bolus, defined as the time between dye injection and the first

Downloaded from iovs.arvojournals.org on 09/29/2021 IOVS, November 2000, Vol. 41, No. 12 Endothelin-3 Regulation of Retinal Blood Flow 3957

TABLE 1. Baseline Characteristics for Nondiabetic and Diabetic Rats cantly prolonged MCTs (P Ͻ 0.01), significantly reduced reti- nal blood flow (P Ͻ 0.01), but no significant differences in Nondiabetic Diabetic Rats Rats primary retinal artery or vein diameters compared with nondi- abetic rats. n 37 34 Body weight (g) 338.3 Ϯ 18.3 293.3 Ϯ 10.3* The Effect of ET-3 on Retinal Hemodynamics in Blood glucose (mg/dl) 102.5 Ϯ 14.3 386.0 Ϯ 88.5* Nondiabetic Rats Hematocrit (%) 47.7 Ϯ 3.6 46.8 Ϯ 6.8 Mean blood pressure (mm Hg) 77.2 Ϯ 15.5 77.7 Ϯ 20.2 After intravitreal injections of ET-3 in the nondiabetic rats, Heart rate (beats/min) 358.8 Ϯ 41.7 334.5 Ϯ 38.3 there were no significant changes in heart rate (403.5 Ϯ 67.5 Appearance time (sec) 1.97 Ϯ 0.35 2.16 Ϯ 0.75 Artery diameter (pixels) 6.6 Ϯ 0.6 6.2 Ϯ 0.7 vs. 377.8 Ϯ 74.4 beats/min), mean blood pressure (78.2 Ϯ 17.2 Vein diameter (pixels) 7.3 Ϯ 0.7 7.1 Ϯ 0.7 vs. 77.2 Ϯ 15.5 mm Hg) or AT (2.2 Ϯ 0.7 vs. 2.2 Ϯ 0.3 seconds) Mean circulation time (sec) 1.04 Ϯ 0.30 1.63 Ϯ 0.50* compared with preinjection measurements, indicating that in- 2 Retinal blood flow (pixel /sec) 100.3 Ϯ 24.7 58.2 Ϯ 15.7* travitreal injection of ET-3 had no significant effect on the systemic circulation. * P Ͻ 0.01 compared with nondiabetic rats. Retinal MCT and blood flow responses to intravitreally injected ET-3 or vehicle alone are summarized in Figures 1A and 1B, respectively. The MCT response to ET-3 was charac- detectable appearance of dye in the retinal artery, represents teristically biphasic in time with an initial rapid dose-depen- an assessment of systemic circulation times. The MCT was dent decrease (maximum 2 minutes after injection) followed at calculated as the difference between the retinal arterial and later times (15 and 30 minutes) by dose-dependent increases in venous circulation times for corresponding artery and vein MCTs compared with baseline or vehicle values. The maxi- pairs, and the average retinal MCT for each rat represents the mum decrease in MCT at 2 minutes occurred at 10Ϫ8 M ET-3 average of the individual artery–vein MCTs. Segmental retinal (0.61 Ϯ 0.17 seconds) and was significantly (P Ͻ 0.01) de- blood flows (in square pixels per second) were calculated from creased compared with vehicle (0.95 Ϯ 0.22 seconds). The the individual MCTs and the corresponding vessel diameter primary retinal artery and vein diameters tended to dilate at determinations, assuming that blood flow was proportional to this time and concentration but were not significantly different the sum of the squares of the arterial and venous diameters from baseline (arteries, 6.7 Ϯ 0.7 vs. 6.9 Ϯ 0.9 pixels; veins, 35 divided by the MCT. The average segmental retinal blood 7.6 Ϯ 0.5 vs. 8.1 Ϯ 0.8 pixels) or vehicle (arteries, 6.6 Ϯ 0.6 vs. flow represented the average of the individual segmental flows. 6.4 Ϯ 0.8 pixels; veins, 7.7 Ϯ 0.6 vs. 7.6 Ϯ 0.8 pixels). At later times after ET-3 intravitreal injection, there was a Statistical Analysis dose-dependent prolongation of the MCT at concentrations of All values are reported as the mean Ϯ SD. Statistical analysis 10Ϫ8 to 10Ϫ6 M ET-3; however, at 10Ϫ9 M ET-3, there was no software (SigmaStat; Jandel Scientific, San Rafael, CA) was used prolongation of the MCT, with values reverting to baseline at for statistical comparisons. One-way repeated-measures analy- 15 and 30 minutes. The vasoconstrictive response 30 minutes sis of variance (ANOVA) was used to compare values for the after injection of 10Ϫ6 M ET-3 was characterized by significant same rats at the different measurement times. Group compar- MCT prolongation (2.73 Ϯ 0.34 seconds) compared with ve- isons were performed using one-way ANOVA. Population nor- hicle (1.03 Ϯ 0.05 seconds; P Ͻ 0.02) and significant primary mality and equality of variances were tested using the Kolmog- retinal vessel diameter constriction (arteries, 5.8 Ϯ 0.5; veins, orov–Smirnov test and the Levene median test, respectively. If 6.6 Ϯ 0.3 pixels) compared with vehicle (arteries, 6.9 Ϯ 0.5; either test failed, then the Kruskal–Wallis ANOVA on ranks was veins, 7.9 Ϯ 0.6 pixels; P Ͻ 0.01). performed. All pair-wise multiple comparisons were per- In parallel with the MCT decrease, the retinal blood flow formed using the Student–Newman–Keuls test. Values of P Ͻ (Fig. 1B) was significantly increased 2 minutes after intravitreal 0.05 were considered to be statistically significant. Power anal- injection of 10Ϫ9 to 10Ϫ7 M ET-3 (193.6 Ϯ 33.2 pixel2/sec at yses for retinal blood flow measurements, based on the mea- 10Ϫ8 M ET-3;) compared with vehicle (105.6 Ϯ 7.6 pixel2/sec; sured variances in retinal blood flow, showed that a difference P Ͻ 0.03). At later times, retinal blood flow decreased, and 30 of 25 pixel2/sec in retinal blood flow could be detected at a minutes after injections of 10Ϫ8 to 10Ϫ6 M ET-3 was signifi- significance of 0.05 with a power of 0.8, using six rats per cantly reduced (28.4 Ϯ 3.3 pixel2/sec at 10Ϫ6 M) compared group. with vehicle (109.6 Ϯ 4.3 pixel2/sec; P Ͻ 0.01). At 10Ϫ9 M ET-3 the retinal blood flow showed an initial rapid increase followed at later times with a reversion to baseline but no RESULTS further decrease. The animal characteristics at the time of retinal blood flow The Effect of ET-3 on Retinal Hemodynamics in measurements are summarized in Table 1 for the 37 nondia- Diabetic Rats betic and the 34 STZ-induced diabetic rats used in the ET-3 dose–response and time course experiments. Group compari- After intravitreal injections of ET-3 in diabetic rats, there were sons showed that although diabetic rats all gained weight, they no significant changes compared with baseline in heart rate gained significantly (P Ͻ 0.01) less weight than the nondiabetic (370.0 Ϯ 43.9 vs. 358.8 Ϯ 41.7 beats/min), mean blood pres- rats and had significantly (P Ͻ 0.01) higher blood glucose sure (83.2 Ϯ 22.7 vs. 77.7 Ϯ 20.2 mm Hg), or retinal AT (2.3 Ϯ levels. There were no significant differences in hematocrit, 0.3 vs. 2.4 Ϯ 0.5 seconds). mean blood pressure, heart rate, and retinal ATs between The diabetic rat retinal MCT responses to ET-3 are sum- diabetic and nondiabetic rats. Diabetic rats showed signifi- marized in Figure 1C. The MCTs at 2 minutes after ET-3 injec-

Downloaded from iovs.arvojournals.org on 09/29/2021 3958 Mori et al. IOVS, November 2000, Vol. 41, No. 12

9 8 7 6 FIGURE 1. (A) Effect of intravitreal injection of ET-3 (10Ϫ M, E, n ϭ 7; 10Ϫ M, , n ϭ 8; 10Ϫ M, ƒ, n ϭ 6; 10Ϫ M, f, n ϭ 6) and vehicle alone (F, n ϭ 10) on retinal MCT in nondiabetic rats at different times after ET-3 injection. (B) Corresponding retinal blood flow responses to ET-3 in nondiabetic rats. (C) Effect of intravitreal injection of ET-3 (10Ϫ9 M, E, n ϭ 6; 10Ϫ8 M, , n ϭ 9; 10Ϫ7 M, ƒ, n ϭ 6; 10Ϫ6 M, f, n ϭ 5) and vehicle alone (F, n ϭ 8) on retinal MCT in diabetic rats at different times after ET-3 injection. (D) Corresponding retinal blood flow responses to ET-3 in diabetic rats. *P Ͻ 0.05 compared with vehicle injection.

tion showed an initial rapid decrease; however, baseline MCTs sponse with an initial rapid decrease followed at later times by were prolonged compared with nondiabetic rats (P Ͻ 0.01), a prolongation that at 30 minutes after injection (2.81 Ϯ 0.20 and at 2 minutes after ET-3 injection remained prolonged seconds) was significant compared with vehicle (1.55 Ϯ 0.21 compared with corresponding MCTs in nondiabetic rats (P Ͻ seconds; P Ͻ 0.01). There were no significant changes in the 0.05; Fig. 1A). In contrast to nondiabetic rats, the retinal MCT major retinal vessel diameters after ET-3 injections. decrease in diabetic rats was sustained for a longer period, with The baseline retinal blood flow in diabetic rats was de- the maximal response occurring 15 minutes after injection at creased compared with that in nondiabetic rats and in re- 10Ϫ9 to 10Ϫ7 M ET-3. At 15 minutes after injection, MCTs were sponse to 10Ϫ9 to 10Ϫ7 M ET-3 increased at 2, 5, and 15 significantly decreased at 10Ϫ9 M (0.80 Ϯ 0.33 seconds) and minutes (Fig. 1D) with a significant maximal response at 15 10Ϫ8 M ET-3 (0.71 Ϯ 0.41 seconds) compared with vehicle minutes after injection of 10Ϫ9 M (127.1 Ϯ 23.2 pixel2/sec) (1.52 Ϯ 0.59 seconds; P Ͻ 0.01), and the decrease was sus- and 10Ϫ8 M (151.7 Ϯ 52.0 pixel2/sec) compared with vehicle tained at 30 minutes after ET-3 injection (P Ͻ 0.05). At 10Ϫ6 M (62.4 Ϯ 10.8 pixel2/sec; P Ͻ 0.01). By 30 minutes after injec- ET-3 the retinal MCT showed a time-attenuated biphasic re- tion, retinal blood flows reverted to baseline. At 10Ϫ6 M ET-3

Downloaded from iovs.arvojournals.org on 09/29/2021 IOVS, November 2000, Vol. 41, No. 12 Endothelin-3 Regulation of Retinal Blood Flow 3959

TABLE 2. The Effects of L-NMMA and BQ-788 on Retinal Hemodynamics in Nondiabetic Rats

Time after Intravitreal Injection

15 30 Baseline minutes minutes

10Ϫ3 M L-NMMA (n ϭ 7) MCT (sec) 1.09 Ϯ 0.3 1.05 Ϯ 0.24 1.11 Ϯ 0.18 Da (pixels) 6.8 Ϯ 0.4 6.4 Ϯ 0.4 6.9 Ϯ 0.3 Dv (pixels) 7.1 Ϯ 0.6 6.9 Ϯ 0.4 7.2 Ϯ 0.2 RBF (pixel2/sec) 93.0 Ϯ 20.9 90.1 Ϯ 25.0 91.8 Ϯ 12.9 10Ϫ4 M BQ-788 (n ϭ 7) MCT (sec) 0.88 Ϯ 0.34 1.03 Ϯ 0.36 1.07 Ϯ 0.29 Da (pixels) 6.3 Ϯ 0.6 5.7 Ϯ 0.7 6.3 Ϯ 0.5 Dv (pixels) 6.8 Ϯ 0.6 6.2 Ϯ 0.7 6.9 Ϯ 0.6 RBF (pixel2/sec) 108.9 Ϯ 38.4 76.3 Ϯ 19.7* 88.8 Ϯ 28.0

* P Ͻ 0.05 compared to baseline. Da, retinal artery diameter; Dv, FIGURE 2. The dose response characteristics of the percentage retinal vein diameter; RBF, retinal blood flow. change from baseline of retinal blood flow at 2 and 15 minutes after ET-3 injections in diabetic and nondiabetic rats. *P Ͻ 0.05 and **P Ͻ 0.01 compared with vehicle injection. (P Ͻ 0.05) and at 30 minutes reverted to flows comparable to baseline. The retinal response to 10Ϫ8 M ET-3 in nondiabetic rats Ϫ3 after a 15-minute pretreatment with 10 M L-NMMA (n ϭ 6), the initial retinal blood flow increase was less pronounced than Ϫ4 at lower ET-3 concentrations, reached a maximum effect at 5 10 M BQ-788 (n ϭ 6), or vehicle (n ϭ 6) is summarized in minutes after injection, and was significantly decreased at 30 Figure 3. In the eyes pretreated with vehicle alone, the ex- 2 pected biphasic retinal blood flow response to ET-3 was mea- minutes (27.7 Ϯ 2.3 pixel /sec) compared with vehicle 2 sured with a rapid transient increase in retinal blood flow (58.4 Ϯ 12.2 pixel /sec; P Ͻ 0.01). (173.6 Ϯ 71.0 pixel2/sec compared with baseline 101.6 Ϯ 29.4 2 Retinal Hemodynamic Dose Responses to ET-3 pixel /sec; P Ͻ 0.05) 2 minutes after ET-3 injection, followed by a reduction in retinal blood flow 15 minutes after ET-3 Figure 2 summarizes the diabetic and nondiabetic rat retinal injection (86.1 Ϯ 49.8 pixel2/sec). In the eyes pretreated with blood flow dose responses to ET-3 calculated as the percentage the NOS inhibitor L-NMMA, the characteristic blood flow re- change from baseline at 2 and 15 minutes after injection. There sponse to ET-3 was completely inhibited. Blood flow at 2 was a dose-dependent increase in the percentage of retinal minutes (98.3 Ϯ 31.7 pixel2/sec) was significantly decreased blood flow change 2 minutes after ET-3 injection, with the compared with that in vehicle-pretreated eyes (173.6 Ϯ 71.0 Ϫ8 maximum at 10 M ET-3. There were no significant differ- pixel2/sec; P Ͻ 0.05). Similarly, in the eyes pretreated the ETB ences, however, in the magnitude of the percentage of retinal blood flow increase between diabetic (104.9% Ϯ 85.4%, at 10Ϫ8 M) and nondiabetic rats (106.6% Ϯ 61.7% at 10Ϫ8 M). Fifteen minutes after ET-3 injection, there was a dose- dependent decrease in the percentage of retinal blood flow change in nondiabetic rats that reached significance at 10Ϫ6 M compared with vehicle (P Ͻ 0.05). In diabetic rats, by contrast, the percentage of retinal blood flow change at 15 minutes dose dependently increased to a maximum at 10Ϫ8 M ET-3 (161.3% Ϯ 84.7%; P Ͻ 0.01 compared with vehicle) and decreased at higher concentrations. Additionally, the magnitude of the per- centage of retinal blood flow change in diabetic rats was significantly different compared with nondiabetic rats for the 10Ϫ9 to 10Ϫ7 M ET-3 (P Ͻ 0.01).

NOS Inhibitor and ETB Antagonist Action The retinal hemodynamic responses to intravitreal injection of Ϫ3 Ϫ4 10 M L-NMMA or 10 M BQ-788 alone in nondiabetic rats

are presented in Table 2. After NOS inhibition, there were no 8 FIGURE 3. The effect of 10Ϫ M ET-3 on retinal blood flow after a significant retinal hemodynamic changes at 15 minutes or 30 15-minute pretreatment with L-NMMA (10Ϫ3 M, n ϭ 6), BQ-788 (10Ϫ4 minutes; however, 40 minutes after injection, there was a M, n ϭ 6), or vehicle alone (n ϭ 6). Time 0 represents initiation of significant decrease in retinal blood flow compared with base- pretreatment with L-NMMA, BQ-788, or vehicle. ET-3 was injected after 2 line measurements (77.5 Ϯ 18.2 pixel /sec; P Ͻ 0.05; data not 15 minutes (arrow), and subsequent retinal blood flow measurements shown). After injection of the ETB antagonist, retinal blood were made at 2 and 15 minutes after ET-3 injection. *P Ͻ 0.05 com- flow was significantly decreased 15 minutes after injection pared with vehicle pretreatment.

Downloaded from iovs.arvojournals.org on 09/29/2021 3960 Mori et al. IOVS, November 2000, Vol. 41, No. 12

with diabetes of 2 weeks’ duration, there was an ET-3–stimu- lated dose-dependent initial retinal blood flow increase. How- ever, compared with nondiabetic rats, this increase was sustained for longer times except at the highest ET-3 concentration at which a temporally attenuated biphasic response was ob- served. In addition, because baseline diabetic rat retinal blood flows were lower, the magnitude of the ET-3–induced blood flow increase was reduced compared with that in nondiabetic rats. However, the dose-dependent percentage change in reti- nal blood flow increases (Fig. 3) was comparable to that in nondiabetic rats, which indicates that the initial vasodilatory response to ET-3 was not impaired in diabetic rats.

The median effective concentration (EC50) for the initial retinal blood flow increase 2 minutes after ET-3 intravitreal injection was 8 ϫ 10Ϫ11 M (effective vitreous concentration), consistent with results in other studies.36 This phenomenon appeared to be primarily associated with microcirculatory va- sorelaxation, rather than with dilation of the primary retinal

Ϫ7 vessels, because the diameter changes in these vessels were FIGURE 4. The effect of 10 M ET-3 on percentage decrease of retinal MCT after a 2-minute pretreatment with 10Ϫ4 M BQ-123, an ETA not statistically significant. The ET-3–induced vasorelaxation antagonist (n ϭ 5), BQ-123 alone (10Ϫ4 M, n ϭ 4), and ET-3 alone. Data appeared to be mediated through NO action, because the taken from Figure 1A for 10Ϫ7 M ET-3 (n ϭ 6) in nondiabetic rats. effect was abolished with NOS inhibitor pretreatment. Other Arrow: Time of ET-3 injection. *P Ͻ 0.03 compared with BQ-123 and preliminary results have shown that endothelium-independent ET-3 alone. NO action also induces a rapid (2-minute) transient retinal blood flow increase comparable in magnitude to the ET-3 responses.37 These results indicate that the initial ET-3–associ- antagonist (BQ-788) the characteristic retinal blood flow re- ated retinal vasorelaxation is mediated by ET-3 binding to the sponse to ET-3 was significantly inhibited (P Ͻ 0.02, compared G-–coupled ETB, subsequent activation of NOS, and 34,38–42 with vehicle pretreated eyes). BQ-788 pretreatment also signif- NO production, which causes vascular smooth muscle icantly reduced retinal blood flow (74.9 Ϯ 20.3 pixel2/sec) and pericyte cell relaxation through cyclic guanosine mono- 43,44 compared with baseline (108.9 Ϯ 38.4 pixel2/sec; P Ͻ 0.05) phosphate increases. before intravitreal ET-3 injection. There was a characteristic dose-dependent ET-3–induced reduction in retinal blood flow in nondiabetic rats at the later Ϫ9 ETA Antagonist Action measurement times, with an EC50 of 8 ϫ 10 M effective Ϫ7 vitreous concentration 15 minutes after injection, consistent The retinal response to 10 M ET-3 after a 2-minute pretreat- 45 ment with 10Ϫ4 M BQ-123, an ETA antagonist, is summarized in with a prior study. The EC50 for the ET-3–mediated vasocon- Figure 4. The results are plotted with respect to the percentage striction was 100 times greater than that for the initial ET-3– mediated vasodilation. This difference was reflected in the decrease in the retinal MCT from baseline. For comparison, the Ϫ9 retinal MCT data shown in Figure 1A for the 10Ϫ7 M ET-3 temporally augmented vasodilatory response to 10 M ET-3, concentration are presented also as the percentage of decrease which was sustained for a longer period (15 minutes) than the in MCT from baseline. The retinal response to BQ-123 alone responses at the higher ET-3 concentrations. Additional data was comparable to that measured previously,10 with a maximal showed that pretreatment with an ETA antagonist resulted in a decrease in MCT occurring 5 minutes after injection and a significant attenuation of this later retinal vasoconstrictive re- return to baseline by 15 minutes after injection. In contrast, the sponse to ET-3. These data indicate that the later retinal blood retinal response to ET-3 after pretreatment with BQ-123 was flow reductions were related to ET-3/ETA interaction with a decreased vasoconstrictive action compared with ET-19,46 characterized by a similar rapid decrease in MCT. However, the 24,25 following increase in MCT at later times (15 minutes) was (ETA affinity for ET-3 is 1000 times less than for ET-1 ). significantly attenuated compared with the response to BQ-123 Thus, the measured biphasic retinal hemodynamic response to alone and compared with the vasoconstrictive retinal response ET-3 depends on a balance between the vasodilatory actions of to ET-3 alone at this measurement time and concentration ETB and the vasoconstrictive actions of ETA. (P Ͻ 0.03; Fig. 1A) In diabetic rats at baseline, the MCT was prolonged, pri- mary vessel diameters were not different, and retinal blood flow was decreased compared with nondiabetic rats consistent 1–5 DISCUSSION with prior studies. The absence of any changes in primary vessel diameters would indicate that the hemodynamic In nondiabetic rats, there was a dose-dependent, biphasic ret- changes were associated with increased flow resistance in the inal blood flow response to intravitreal injection of ET-3 with microcirculation. Prior studies showed that the decreased ret- an initial transient rapid increase and subsequent decreased inal blood flow in diabetic rats was related to increased protein retinal blood flow. The initial retinal blood flow increase was kinase C-␤ activation3,4 and increased ET-1 expression.10,47 completely inhibited by L-NMMA, a NOS inhibitor and by BQ- Diabetic rats also responded to ET-3 with a dose-depen- 788, an ETB antagonist. The retinal blood flow decrease at later dent initial retinal blood flow increase. However, the maximal times was inhibited by BQ-123, an ETA antagonist. In the rats response occurred 15 minutes after injection, and the increase

Downloaded from iovs.arvojournals.org on 09/29/2021 IOVS, November 2000, Vol. 41, No. 12 Endothelin-3 Regulation of Retinal Blood Flow 3961

was sustained at 30 minutes after injection, which was char- clearance in patients with type 1 Diabetes. Diabetes Care. 1999; acteristically different from the effect in nondiabetic rats. It 22:1245–1251. was only at 10Ϫ6 M ET-3 that a temporally attenuated biphasic 8. Konno S, Feke GT, Yoshida A, Fujio N, Goger DG, Buzney SM. Retinal blood flow changes in type I diabetes: a long-term fol- blood flow response was noted with the maximal vasodilatory low-up study. Invest Ophthalmol Vis Sci. 1996;37:1140–1148. effect at 5 minutes and a retinal blood flow decrease at 30 9. Bursell SE, Clermont AC, Oren B, King GL. The in vivo effect of minutes after injection. The sustained ET-3–mediated retinal endothelins on retinal circulation in nondiabetic and diabetic rats. blood flow increase in the diabetic rats could be associated Invest Ophthalmol Vis Sci. 1995;36:596–607. with a decrease in ET-3/ETA interaction associated with com- 10. Takagi C, Bursell SE, Lin YW, et al. Regulation of retinal hemody- petition from increased ET-1 in the diabetic rat retina.10 Data namics in diabetic rats by increased expression and action of 48 endothelin-1. Invest Ophthalmol Vis Sci. 1996;37:2504–2518. from other studies suggest that chronically increased levels 11. Takagi C, King GL, Takagi H, Lin YW, Clermont AC, Bursell SE. of ET-1 in diabetes may lead to decreased affinity or downregu- Endothelin-1 action by endothelin receptors is primary mechanism lation of ETAs. Thus a decrease in the ET-3/ETA action in modulating retinal circulatory response to hyperoxia. Invest Oph- diabetic rats would lead to an attenuation of the expected thalmol Vis Sci. 1996;37:2099–2109. vasoconstriction and the resultant temporal augmentation of 12. Meyer P, Flammer J, Luscher TF. Endothelium-dependent regula- tion of the ophthalmic microcirculation in the perfused porcine the ET-3 vasodilatory effect. eye: role of nitric oxide and endothelins. Invest Ophthalmol Vis The sustained elevated retinal blood flow in diabetic rats Sci. 1993;34:3614–3621. may be associated with other vasodilatory actions. For exam- 13. Chakravarthy U, Gardiner TA, Anderson P, Archer DB, Trimble ER. ple, in cultured endothelial cells vasodilatory prostacyclins are The effect of on the retinal microvascular pericyte. produced in a time-dependent manner after ET-3 stimulation,38 Microvasc Res. 1992;43:241–254. in human hepatic stellate cells ETBs are upregulated by ET-1/ 14. Hirata Y, Hayakawa H, Suzuki E, et al. Direct measurements of endothelium-derived nitric oxide release by stimulation of endo- ETB–stimulated production of cyclic adenosine monophos- thelin receptors in rat kidney and its alteration in salt-induced 49 phate and prostacyclin, and in diabetes NOS activation is hypertension. Circulation. 1995;91:1229–1235. increased by ETB action through calcium-calmodulin and pro- 15. Takahashi K, Brooks RA, Kanse SM, Ghatei MA, Kohner EM, Bloom tein tyrosine kinase–dependent pathways.50 Thus, upregula- SR. Production of endothelin-1 by cultured bovine retinal endo- tion of the ETB and/or activation of postreceptor intracellular thelial cells and presence of endothelin receptors on associated pericytes. Diabetes. 1989;38:1200–1202. signal transduction cascades could result in increased produc- 16. Haefliger IO, Zschauer A, Anderson DR. Relaxation of retinal peri- tion and action of NO and/or prostacyclin, which could con- cyte contractile tone through the nitric oxide-cyclic guanosine tribute to sustained ET-3–stimulated retinal blood flow increase monophosphate pathway. Invest Ophthalmol Vis Sci. 1994;35: in the diabetic rats. 991–997. Intravitreally introduced ET-3 caused an initial rapid in- 17. Bohlen HG, Lash JM. Topical hyperglycemia rapidly suppresses crease in retinal blood flow followed at later times by reduced EDRF-mediated vasodilation of normal rat arterioles. Am J Physiol. 1993;265:H219–H225. retinal blood flow in nondiabetic rats. The initial retinal blood 18. Kakoki M, Hirata Y, Hayakawa H, et al. Effects of hypertension, flow increase was mediated through ET-3/ETB action and NO diabetes mellitus, and hypercholesterolemia on endothelin type B release. The subsequent retinal blood flow decrease was asso- receptor-mediated nitric oxide release from rat kidney. Circula- ciated with ET-3/ETA action. Diabetic rats showed comparable tion. 1999;99:1242–1248. early retinal blood flow increases, indicating that ET-3 action 19. Nielsen PJ, Nyborg NC. Contractile and relaxing effects of arachi- donic acid derivatives on isolated bovine retinal resistance arteries. may not be impaired. However, the prolongation of this blood Exp Eye Res. 1990;50:305–311. flow increase suggests a diminution in ET-3/ETA action. 20. Iwasaki T, Notoya M, Hayasaki–Kajiwara Y, et al. Endothelium- independent vascular relaxation mediating ETB receptor in rabbit References mesenteric arteries. Am J Physiol. 1999;276:H383–H390. 21. Qiu C, Baylis C. Endothelin and angiotensin mediate most glomer- 1. Clermont AC, Brittis M, Shiba T, McGovern T, King GL, Bursell SE. ular responses to nitric oxide inhibition. Kidney Int. 1999;55: Normalization of retinal blood flow in diabetic rats with primary 2390–2396. intervention using pumps. Invest Ophthalmol Vis Sci. 22. Rockwood EJ, Fantes F, Davis EB, Anderson DR. The response of 1994;35:981–990. retinal vasculature to angiotensin. Invest Ophthalmol Vis Sci. 2. Shiba T, Inoguchi T, Sportsman JR, Heath WF, Bursell SE, King GL. 1987;28:676–682. Correlation of diacylglycerol level and protein kinase C activity in 23. Yanagisawa M, Kurihara H, Kimura S, et al. A novel potent vaso- rat retina to retinal circulation. Am J Physiol. 1993;265:E783– constrictor produced by vascular endothelial cells. Na- E793. ture. 1988;332:411–415. 3. Ishii H, Jirousek MR, Koya D, et al. Amelioration of vascular 24. Sakurai T, Yanagisawa M, Masaki T. Molecular characterization of dysfunctions in diabetic rats by an oral PKC ␤ inhibitor. Science. endothelin receptors. Trends Pharmacol Sci. 1992;13:103–108. 1996;272:728–731. 25. Ihara M, Noguchi K, Saeki T, et al. Biological profiles of highly 4. Bursell SE, Takagi C, Clermont AC, et al. Specific retinal diacylglyc- potent novel endothelin antagonists selective for the ETA recep- erol and protein kinase C beta isoform modulation mimics abnor- tor. Life Science. 1992;50:247–250. mal retinal hemodynamics in diabetic rats. Invest Ophthalmol Vis 26. Eguchi S, Hirata Y, Ihara M, Yano M, Marumo F. A novel ETA Sci. 1997;38:2711–2720. antagonist (BQ-123) inhibits endothelin-1-induced phospho-inosit- 5. Takagi C, King GL, Clermont AC, Cummins DR, Takagi H, Bursell ide breakdown and DNA synthesis in rats vascular smooth muscle SE. Reversal of abnormal retinal hemodynamics in diabetic rats by cells. FEBS Lett. 1992;302:243–246. acarbose, an alpha-glucosidase inhibitor. Curr Eye Res. 1995;14: 27. Lin HY, Kaji EH, Winkel GK, Ives HE, Lodish HF. Cloning and 741–749. functional expression of human endothelin-1 receptor cDNA. Proc 6. Bursell SE, Clermont AC, Kinsley BT, Simonson DC, Aiello LM, Natl Acad Sci USA. 1991;88:3185–3189. Wolpert HA. Retinal blood flow changes in patients with insulin- 28. Ogawa Y, Nakao K, Arai H, et al. Molecular cloning of a non- dependent diabetes mellitus and no diabetic retinopathy. Invest isopeptide-selective human . Biochem Bio- Ophthalmol Vis Sci. 1996;37:886–897. phys Res Commun. 1991;178:248–255. 7. Bursell SE, Clermont AC, Aiello LP, et al. High-dose vitamin E 29. Inoue A, Yanagisawa M, Kimura S, et al. The human endothelin supplementation normalizes retinal blood flow and creatinine family: Three structurally and pharmacologically distinct isopep-

Downloaded from iovs.arvojournals.org on 09/29/2021 3962 Mori et al. IOVS, November 2000, Vol. 41, No. 12

tides predicted by three separate . Proc Natl Acad Sci USA. 41. Emori T, Hirata Y, Ohta K, et al. Endothelin-3 stimulates produc 1989;86:2863–2867. tion of endothelium-derived nitric oxide by phosphoinositide 30. Fukui M, Nakamura T, Ebihara I, et al. expression for endo- breakdown. Biochem Biophys Res Commun. 1991;174:228–235. thelins and their receptors in glomeruli of diabetic rats. J Lab Clin 42. Hirata Y, Emori T, Eguchi S, et al. Endothelin receptor subtype B Med. 1993;122:149–156. mediates synthesis of nitric oxide by cultured bovine endothelial 31. Lawrence E, Brain SD. Altered microvascular reactivity to endothe- cells. J Clin Invest. 1993;91:1367–1373. G lin-1, endothelin-3 and N -nitro-L-arginine methyl ester in strepto- 43. Ignarro LJ, Burke TM, Wood KS, Wolin MS, Kadowitz PJ. Associa- zocin-induced diabetes mellitus. Br J Pharmacol. 1992;106:1035– tion between cyclic GMP accumulation and acetylcholine-elicited 1040. relaxation of bovine intrapulmonary artery. J Pharmacol Exp Ther. 32. Hughes A. A schematic eye for rat. Vision Res. 1979;19:569–588. 1984;228:682–690. 33. Wheeler MA, Smith SD, Garcia–Carena G, Nathan CF, Weiss RM, 44. Haefliger IO, Zschaer A, Anderson DR. Relaxation of retinal peri- Sessa WC. Bacterial infection induces nitric oxide synthase in cyte contractile tone through the nitric oxide-cyclic guanosine human neutrophils. J Clin Invest. 1997;99:110–116. monophosphate pathway. Invest Ophthalmol Vis Sci. 1994;35: 34. Karaki H, Sudjarwo AS, Hori M. Novel antagonist of endothelin 991–997. ETB1 and ETB2 receptors, BQ-788: effect on blood vessel and small 45. Yanagisawa M, Inoue A, Ishikawa T, et al. Primary structure, intestine. Biochem Biophys Res Commun. 1994;205:168–173. synthesis, and biological activity of rat endothelin, an endothelium- 35. Bulpitt CJ, Dollery CT. Estimation of retinal blood flow by mea- derived vasoconstrictor peptide. Proc Natl Acad Sci USA. 1988; surement of mean circulation time. Cardiovasc Res. 1971;5:406– 85:6964–6967. 412. 46. Cardell LO, Uddman R, Edvinsson L. Evidence for multiple endo- 36. Yokokawa K, Johnson J, Kohno M, et al. Phosphoinositide turn- thelin receptors in the guinea pig pulmonary artery and trachea. over signaling stimulated by ET-3 in endothelial cells from sponta- Br J Pharmacol. 1992;105:376–380. neously hypertensive rats. Am J Physiol. 1994;267:R635–R644. 47. Yamauchi T, Ohnaka K, Takayanagi R, Umeda F, Nawata H. En- 37. Bursell SE, Mori F, Clermont A, Koya D, King GL. Nitric oxide (NO) mediation of retinal circulation in diabetic and nondiabetic rats hanced secretion of endothelin-1 by elevated glucose levels from [ARVO Abstract]. Invest Ophthalmol Vis Sci. 1997;38(4):S1135. cultured bovine aortic endothelial cells. FEBS Lett. 1990;267:16– Abstract nr 5324. 18. 38. Owada A, Tomita K, Terada Y, Sakamoto H, Nonoguchi H, Marumo 48. DelaRubia G, Oliver J, Inoguchi T, King GL. Induction of resistance F. Endothelin (ET)-3 stimulates cyclic guanosine 3,5-monophos- to endothelin-1’s biochemical actions by elevated glucose levels in phate production by ET-B receptor by producing nitric oxide in retinal pericytes. Diabetes. 1992;41:1533–1539. isolated rat glomerulus, and in rat mesangial cells. J Clin Invest. 49. Mallat A, Preaux AM, Gal CSL, et al. Growth inhibitory properties 1994;93:556–563. of endothelin-1 in activated human hepatic stellate cells: a cyclic 39. Fukuda N, Izumi Y, Soma M, et al. L-NG-monoethyl arginine inhibits adenosine monophosphate-mediated pathway. J Clin Invest. 1996; the vasodilating effects of low dose of endothelin-3 on rat mesen- 98:2771–2778. teric arteries. Biochem Biophys Res Commun. 1990;167:739–745. 50. Tsukahara H, Ende H, Magazine H, Bahous W, Goligorsky MS. 40. Emori T, Hirata Y, Marumo F. Specific receptors for endothelin-3 in Molecular and functional characterization of the non-isopeptide- cultured bovine endothelial cells and its cellular mechanism of selective ET-B receptor in endothelial cells. J Biol Chem. 1994; action. FEBS Lett. 1990;263:261–264. 269:21778–21785.

Downloaded from iovs.arvojournals.org on 09/29/2021