ATP-Induced Relaxation of Porcine Retinal Arterioles In Vitro Depends on Prostaglandin E Synthesized in the Perivascular Retinal Tissue

Abstract

Purpose.: It has been shown that inhibition of prostaglandin synthesis in the perivascular retinal tissue can prevent the relaxation of retinal arterioles induced by N-methyl-d-aspartic acid (NMDA) and adenosine triphosphate (ATP). The purpose of the present study was to identify the prostaglandins involved in this retina-dependent relaxation.

Methods.: Porcine retinal arterioles were mounted in a myograph for isometric tone measurements. The effect of the prostaglandins (PGs) PGE₂, PGF_2α, PGD₂, and PGI₂ and of thromboxane A₂ (TXA₂) on vascular tone was recorded before and after removal of the perivascular retina, and the specificity of the responses were confirmed by blocking with specific antagonists. Finally, the coupling between prostaglandins found to have a specific vasoactive effect, dependent on the perivascular retina, and the individual vasorelaxing effects of NMDA, ATP, and adenosine were studied.

Results.: All prostaglandins tested showed a significant relaxation of precontracted arterioles at the highest concentrations, whereas PGF_2α induced a significant constriction of isolated noncontracted arterioles. In the presence of perivascular retinal tissue, the dilating effect of PGE₂ increased significantly, an effect that was blocked by a prostaglandin E prostanoid (EP₁) receptor blocker, whereas PGD₂ induced a dual response, with a significant contraction at low concentrations and a significant dilation at high concentrations. Inhibition of the cyclo-oxygenase (COX) enzyme with ibuprofen, as well as the EP₁ receptor, blocked the vasodilating effect of ATP, but not that of NMDA and adenosine, in the presence of perivascular retinal tissue.

Conclusions.: ATP-induced vasodilation depends on the production of PGE in the perivascular retina. However, the regulation of retinal arteriolar tone involves COX products other than PGE.

An elucidation of the mechanisms underlying tone regulation in retinal arterioles is a precondition for improving the treatment of diseases in which disturbances in retinal perfusion are involved in pathogenesis. Results in studies have shown that agonists to the glutamate N-methyl-d-aspartic acid (NMDA) receptor in the tissue surrounding porcine retinal arterioles cause a vasorelaxation, which is mediated through ATP hydrolysis in the perivascular retina and adenosine receptor (A1/A2) activation in the arteriolar wall. ^1,2 This suggests that the unknown “retinal relaxing factor” described in in vitro studies ^3,4 is related to the glutamate released from the retinal tissue. NMDA-induced vasorelaxation has been shown to be blocked by cyclo-oxygenase (COX) inhibition, suggesting that a cyclo-oxygenase product in the perivascular retina is involved in the vasodilating effect. ¹ The exact nature of this COX product remains to be elucidated, but the presence of specific prostaglandin receptors in the retina suggest a physiological role of ligands to these receptors in retinal physiology including the regulation of vascular tone. ^5–7  

The purpose of the present study was to investigate how ATP-induced vasodilation is influenced by the prostaglandins (PG) PGE₂, PGF_2α, PGD₂, and PGI₂ and thromboxane A₂ (TXA₂), which have affinity for the prostaglandin receptors that have been identified in retinal tissue. Porcine retinal arterioles were mounted in a wire myograph, the effect of prostaglandins on relaxed and precontracted arterioles was studied, and the specificity of the observed effects was tested with the application of specific antagonists. Finally, the coupling between vasodilation induced by NMDA, ATP, and adenosine and the prostaglandins that had a specific vasodilating effect was studied. 

Materials and Methods

Solutions

Physiologic saline solution (PSS) containing (in mM): NaCl 119, KCl 4.7, MgSO₄ 1.17, NaHCO₃ 25, KH₂PO₄ 1.18, CaCl₂ 1.6, EDTA 0.026, glucose 5.5, and HEPES 5.0 (pH = 7.4) was used for transportation and storage and for the pharmacologic experiments. During dissection and normalization of the vessel diameter, a similar solution was used, in which CaCl₂ was omitted (Ca²⁺-free PSS). 

Tissue

The details of the procedure are described elsewhere. ¹ In brief, porcine eyes were collected from a local abattoir and were transported to the laboratory in 4°C PSS within 1 hour. The eyes were bisected by a frontal section through the equator, the vitreous was removed, and the retina was detached from the underlying pigment epithelium by injection of PSS between the two structures. Subsequently, an arteriolar segment with a length of <2 mm with approximately 2 mm retinal tissue attached on each side of the vessel was dissected from the retina. 

Mounting Procedure

The vascular segment was placed in the chamber of a small-vessel myograph (610M Multi-myograph; Danish Myo Technology, Århus, Denmark) and mounted on 25-μm diameter tungsten wires. After it was mounted, the preparation was suspended freely in PSS between the myograph jaws, and bubbling of the bath commenced, with a mixture of 95% atmospheric air and 5% CO₂, to give a pH of 7.4. 

Normalization

Normalization was performed according to published procedures. ⁸ In short, the arteriolar diameter was increased in four steps in Ca²⁺-free PSS, and the passive tensions (corresponding to transmural pressures between 0 and approximately 70 mm Hg) were measured. This diameter–tension relationship was exponential, and the intercept between this curve and a straight line based on the Laplace equation (wall tension = transmural pressure × radius) with the transmural pressure set to 70 mm Hg was calculated. Using the built-in micrometer screw, the jaws of the myograph were adjusted to 93.5% of the intercept length, at which tension the arteriole develop the maximal tone (i.e., the optimal length for contraction of the vascular smooth muscle cells). 

Compounds

Ibuprofen, NMDA, and ATP were purchased from Sigma-Aldrich (Vallensbaek, Denmark). All prostaglandin agonists and antagonists were purchased from Cayman Chemicals, Europe, Tallinn, Estonia). 

Tone Modulators.

Preconstriction was performed with the thromboxane analogue U46619 (9,11-dideoxy-11α,9α-epoxymethano-prostaglandin F_2α) and vasodilatation with NMDA, ATP, and adenosine. 

Prostaglandin Synthesis Inhibitor.

Ibuprofen, a nonselective inhibitor of COX, was used to block the COX-1 and -2 enzymes simultaneously. 

Prostaglandin Receptor Agonists.

The agonists to the prostaglandin receptors included PGE₂ (14010) an agonist to the EP receptors 1 to 4; PGF_2α (16,010), an agonist to the FP receptor; PGD₂ (12010), an agonist to the DP receptor; PGI₂ (18220), an agonist to the IP receptor, and carbocyclic thromboxane A₂ (CTA2; 19010), a stable agonist to the TP receptor (catalog numbers are in parentheses). 

Prostaglandin Antagonists.

The selective EP₁ receptor antagonist SC-19220 (14060), the selective TP receptor antagonist SQ-29548 (19025), and the selective FP receptor antagonist AL-8810 (16735) were used. 

U46619, NMDA, and ATP were dissolved in distilled water as stock solutions 10³ times the highest concentration used in the experiments and stored frozen for later use. All other compounds were prepared on the day of the experiment and dissolved in PSS. 

Experiments

Only one arteriole was used from each animal. For each experimental condition, at least six observations were obtained. Three different experimental series were conducted as shown in Figure 1. 

Figure 1.

 

The experimental protocols.

Series 1.

Equilibration.

Arterioles were allowed to equilibrate for a period of 10 minutes in PSS, for the tone to stabilize (Fig. 1A). 

Preconstriction.

U46619 was added in a concentration of 10⁻⁶ M to preconstrict the vessel, and the vascular tone was allowed to stabilize for at least 10 minutes or until no further increase in vascular tone was observed. 

Concentration–Response Experiment.

Subsequently, a prostaglandin agonist was added to the myograph chamber in increasing concentrations, with an increment of 0.5 log units between each new concentration. PGE₂ was added in 7 concentrations ranging between 10⁻⁹ and 10⁻⁶; PGF_2α in 8 concentrations between 10⁻⁸ and 3 × 10⁻⁵ M; PGD₂ in 8 concentrations between 10⁻⁸ and 3 × 10⁻⁵ M; and PGI₂ in 10 concentrations between 10⁻⁹ and 3 × 10⁻⁵ M, and CTA2 was added in 11 concentrations between 10⁻¹⁰ and 10⁻⁵ M. Each change in concentration was followed by a period of at least 3 minutes during which the vascular tone as allowed to stabilize. When the tone had been recorded after addition of the highest concentration of the relaxing compound, the myograph chamber was washed four times with PSS. 

Removal of the Perivascular Tissue.

The myograph was moved to a stereo microscope and the perivascular retina was gently removed within 2 minutes, using two pairs of fine forceps without touching the arteriole. 

Repetition of the Concentration–Response Experiment.

The myograph was replaced in the recording unit and the procedure for equilibration were repeated on the isolated arteriole. Subsequently, all experiments in series 1 were repeated without preconstriction with U46619, to study whether a per se constrictive response of the studied prostaglandins had been masked by the preconstriction. 

Series 2.

The concentration–response experiments in series 1 with PGE₂ (with preconstriction) and with CTA2 and PGF_2α (without preconstriction) were repeated, but in the presence of their specific antagonists from the beginning of the equilibration period and throughout the experiment: (1) PGE₂ with the selective EP₁ receptor antagonist SC-19220 (5 × 10⁻⁵ M); and (2) CTA2 with the selective TP receptor antagonist SQ-29548 (10⁻⁵ M) or (3) PGF_2α with the selective FP receptor antagonist AL-8810 (10⁻⁶ M). Because of the lack of commercially available specific antagonists to the DP receptor, experiments could not be performed to antagonize the effect of PGD₂ (Fig. 1B). 

Series 3.

The experiments in series 1 were repeated with the following modifications: The selective EP₁ receptor antagonist SC-19220 (5 × 10⁻⁵ M) was added at the beginning of the equilibrium period, so that it was present throughout the experiments. The concentration–response experiments were performed with the following compounds: NMDA added in seven concentrations between 10⁻⁶ and 10⁻³ M, adenosine triphosphate (ATP) in seven concentrations between 10⁻⁸ and 3 × 10⁻⁶ M, and adenosine in eight concentrations between 10⁻⁷ and 3 × 10⁻⁴ M. 

Test for Viability

Arterioles were considered to be viable and were included if the vascular tone increased >0.2 N/m compared with the condition in Ca²⁺-free PSS, either during precontraction with U46619 or during the final addition of PSS containing 120 mM K⁺ (vessels without precontraction). 

Control Experiments

COX Inhibition.

The experiment in series 1 was repeated with the following modifications: The nonselective COX inhibitor ibuprofen (10 μM) was added so that it was present throughout the experiment, and concentration–response experiments were made with the following compounds: NMDA in seven concentrations between 10⁻⁶ and 10⁻³ M and ATP in seven concentrations between 10⁻⁸ and 3 × 10⁻⁶ M. 

Interaction from the Vascular Wall.

To test whether the serial experimental design was biased by signal desensitization in the arterioles, we repeated the experiments in series 1 with the following modifications: (1) Concentration–response experiments with PGE were performed without removing the perivascular retinal tissue before repeating the experiments; (2) the perivascular retinal tissue was removed before preconstriction, and consequently concentration–response experiments with PGE, NMDA, and ATP were repeated on the isolated arterioles. There was no significant difference between the first and the second concentration–response curve. 

Time Controls.

To ensure that no time-dependent vasorelaxation occurred, we conducted a series of control experiments in which only the solvent (H₂O) was added to the preconstricted arterioles. These experiments revealed no time-dependent vasorelaxation in arterioles with perivascular retina or in isolated arterioles (not shown). 

Data Analysis

The tensions produced by the mounted arterioles were sampled at 1 Hz and displayed on a computer monitor as a function of time during the experiments. The data were stored in a data sheet (Excel; Microsoft, Redmond, WA) file for subsequent analysis. The tone recorded after the addition of each concentration of agonist was normalized to the tone produced after addition of 10⁻⁶ M U46619. In the experiments without preconstriction, vascular tone was normalized to the initial basal tone before addition of the agonists. The normalized tone was plotted as a function of the agonist concentration. The statistical analysis (Prism, ver. 4.02; GraphPad, San Diego, CA) showed that saturation kinetics at high agonist concentrations could be fitted to the Michaelis-Menten equation to obtain the EC₅₀. 

Statistical Analysis

In each concentration–response experiment, repeated-measures ANOVA was used to test whether the tone changed significantly after addition of increasing concentrations of agonist. Student's paired t-test was used to test for differences in the tone response of arterioles with and without perivascular retinal tissue for each concentration of the prostaglandins used in series 1. The unpaired t-test was used to determine whether the EC₅₀ and the tone response at each concentration of prostaglandins in series 1 were significantly changed by the specific antagonists used in series 2. The paired t-test was also used to test whether there was a significant difference between experiments with and without perivascular retinal tissue in series 3. The test was applied to EC₅₀ values when this parameter could be calculated for both experimental conditions and otherwise to the tone produced at the maximum prostaglandin concentration. 

The differences between isolated vessels and vessels with preserved perivascular retinal tissue were confirmed by repeated-measures ANOVA on the differences between tone induced in isolated vessels and in vessels with preserved perivascular retinal tissue in cases in which the concentration–response curves were monotonous. 

Results

All the tested prostaglandins induced a significant concentration-dependent vasorelaxation in precontracted arterioles, with and without perivascular retinal tissue at high concentrations (P < 0.01 for all comparisons, Student's paired t-test). 

The results from the experiments in series 1 are shown in Figure 2. PGE₂ induced significantly increased relaxation in the presence of perivascular retinal tissue over that in isolated vessels at the five highest concentrations and, accordingly, a significantly lower EC₅₀ in the presence of perivascular retinal tissue (mean: 6.27 × 10⁻⁹ M; CI₉₅: 3.31 × 10⁻⁶–1.20 × 10⁻¹¹ M) than in the isolated retinal arterioles (mean: 2.69 × 10⁻⁷ M; CI₉₅: 5.89 × 10⁻⁸–1.23 × 10⁻⁶ M) (P < 0.05, paired t-test; Fig. 2A). PGD₂ showed a dual response in the presence of the perivascular retinal tissue, with a significant constriction at lower (10⁻⁸–10⁻⁷ M) concentrations (P < 0.05, paired t-test), but a significantly increased relaxation at higher (3 × 10⁻⁵ M) concentrations than that in the isolated vessel (P < 0.05, paired t-test; Fig. 2C). There was no significant difference between the vasorelaxing effect of CTA2, PGI₂ and PGF_2α between isolated retinal arterioles and in arterioles with preserved perivascular tissue (Figs. 2B, 2D, 2E). 

Figure 2.

 

The effect of (A) PGE₂, (B) PGI₂, (C) PGD₂, (D) PGF_2α, and (E) CTA2 on the tone of precontracted porcine retinal arterioles in the presence of the COX inhibitor ibuprofen. *Significant difference between arterioles with preserved perivascular tissue (black) and isolated arterioles (gray).

Results of the experiments in series 2 are shown in Table 1 and Figure 3. The EP₁ receptor antagonist SC-19220 significantly reduced PGE₂-induced vasorelaxation of precontracted arterioles with perivascular retina (P < 0.05 at concentrations of 10⁻⁸–3 × 10⁻⁷ M, nonpaired t-test; Fig. 3A), whereas the FP receptor antagonist AL-8810 (10⁻⁶ M) blocked PGF_2α-induced vasoconstriction in nonprecontracted vessels (P < 0.05 at concentrations of 3 × 10⁻⁶–3 × 10⁻⁵ M, nonpaired t-test; Fig. 3B). 

Table 1.

 

Results of Experiments in Series 2

	+ Retina		− Retina
	EC50 × 10−7 M	Maximum Response	EC50 × 10−7 M	Maximum Response
PGE2	20.4 (16.2–25.1)	0.95 (0.92–0.98)	631 (204–195)	0.74 (0.65–0.83)
PGE2 and SC-19220	0.74 (0.22–2.45)	0.82 (0.64–0.98)	2.69 (2.04–3.55)	0.26 (0.23–0.29)
	NS	NS	P < 0.001	P < 0.001
PGF2α	—	0.98 (0.89–1.07)	—	1.35 (1.27–1.43)
PGF2α and AL-8810	—	0.7 (0.58–0.82)	—	0.97 (0.84–1.10)
	—	NS	—	P < 0.05

 Data are expressed as the mean (range). + Retina, vessels with peripheral vascular tissue attached; − Retina, isolated vessels.

Figure 3.

 

The effect of specific inhibition of (A) PGE₂ and (B) PGF_2α on porcine retinal arterioles in vitro. Arterioles with preserved perivascular tissue (black), and isolated arterioles (gray). #Significant differences in the unpaired test between experiments with PGE₂ on precontracted arterioles, with and without the PGE₂ antagonist, and experiments with PGF_2α on uncontracted arterioles, with and without the PGF_2α antagonist.

The results of the experiments in series 3 are shown in Figure 4. The specific EP₁ receptor blocker SC-19220 significantly reduced ATP-induced vasorelaxation of arterioles in the presence of perivascular retinal tissue at the three highest ATP concentrations (P < 0.01, nonpaired t-test; Figs. 4C, 4D), whereas no significant effect was observed on NMDA- or adenosine-induced vasodilatation (P ≥ 0.31 and P ≥ 0.09, unpaired t-test; Figs. 4A, 4B and 4E, 4F, respectively. 

Figure 4.

 

The vasorelaxing effect of (A, B) NMDA, (C, D) ATP, and (E, F) adenosine in the presence of and without the PGE₂ antagonist SC-19220. #Significant difference between arterioles with preserved perivascular tissue in the absence and presence of the antagonist.

Figure 5 confirms that blocking of the COX enzyme with ibuprofen inhibited vasodilation induced by both NMDA and ATP. 

Figure 5.

 

The effect of (A) NMDA and (B) ATP on the tone of retinal arterioles in vitro in the presence of the COX inhibitor ibuprofen. No significant difference occurred between arterioles with preserved perivascular tissue (black), and isolated arterioles (gray).

Discussion

The present findings contribute to evidence suggesting that prostaglandins may participate in the normal physiological regulation of retinal vascular tone. Thus, prostaglandins have been shown to have vasoactive effects generally, ^9–11 and both prostaglandins and receptors to these compounds have been found in retinal tissue. ^6,12,13 In addition, previous studies have shown that inhibition of prostaglandin synthesis potentiates constriction of retinal arterioles in the presence of perivascular retinal tissue. ¹  

The in vitro model used is limited to no-flow conditions, and consequently the involvement of prostaglandins in responses elicited by shear stress and other consequences of the flowing blood could not be studied. Furthermore, studies of the vasoactive effect of prostaglandins are hampered by the possible existence of still unidentified prostaglandins, by the fact that specific antagonists to receptors of some already-known prostaglandins are missing, and by the fact that known agonists and antagonists may show cross-reactivity to several receptors. ⁷ These facts may provide the background for the finding that, after preconstriction, all the studied prostaglandins showed a vasorelaxing effect at the highest concentrations used, that PGF_2α showed the opposite effect by decreasing the vascular tone after the vessels had been precontracted with U46619, and that CTA2 and U46619 had different effects, although both have been characterized as specific agonists to the TP receptor. ¹⁴ It has also been shown that the involvement of prostaglandins in the tone regulation of retinal arterioles displays considerable species variation, ^4,15 but despite this, no functional pharmacologic differences have hitherto been shown in the tone regulation in porcine and human retinal arterioles. ¹⁶  

The control experiments confirmed results in previous studies that showed that the vasorelaxing effect of NMDA and ATP in retinal arterioles depends on a COX product synthesized in the perivascular retina. ¹ The present results suggest that this COX product may be PGE. Thus, a similar relaxing effect of PGE on retinal arterioles was found to depend on the perivascular retina, and this effect was inhibited by a specific prostaglandin E receptor (PGE_rec) antagonist. Furthermore, the effect of the PGE_rec antagonist on ATP-induced vasodilation was not found in adenosine-induced vasodilation, which does not depend on the perivascular retina. ² It has been shown that PGE receptors exist in ocular tissue, ^6,7 and that PGE may induce vasodilation after intravitreal injection in minipigs, ⁵ an effect that can be reversed by indomethacin. ¹⁷ Furthermore, intravenously administered PGE₁ and PGE₂ increase retinal perfusion in rats ¹⁸ although this effect cannot be reproduced with PGE₁ in humans. ¹⁹ Finally, recent studies of brain arterioles from the rat suggest that PGE₂ uptake from the extracellular space can be inhibited by lactate secondary to hypoxia and thereby induce vasodilation. ⁹ The fact that blocking of the (EP₁) receptor did not affect NMDA-induced vasorelaxation suggests that NMDA also induces vasorelaxation through a COX product that does not involve ATP. Altogether, the evidence suggests that the observed effect of PGE occurs downstream from the synthesis of ATP, but before adenosine synthesized from ATP exerts an effect on the retinal vascular smooth muscle cells. We suggest that one or more compounds related to this pathway may be the previously published unknown retinal relaxing factor that is released from the perivascular retinal tissue. 

The existence of DP receptors in human retinal vessels makes it likely that ligands to these receptors have a physiological role for regulating retinal vascular tone. ⁷ However, the dual effect of PGD₂ found on arterioles with the perivascular retina with vasoconstriction at low PGD₂ concentrations and vasorelaxation at high concentrations may be due to a different effect on one EP receptor ⁷ or a mixed effect on several EP receptors with different receptor specificity at low and high concentrations of the ligand. ^20–23  

In previous studies PGF_2α has been found to induce vasoconstriction which can be antagonized by a calcium channel blocker. ¹⁵ In the present study PGF_2α induced a modest vasoconstriction in isolated retinal arterioles that was abolished in the presence of the perivascular retina, which confirms findings in previous studies of bovine and porcine retinal tissue. ⁴ The background for this effect of the perivascular retinal tissue remains obscure but may be related to the finding that PGF is involved in the physiological regulation of vascular tone during hypercapnia. ^5,24 Alternatively, PGF_2α may have nonspecific effects on other prostaglandin receptors in the retina. 

In previous studies, PGI₂ has been found to induce vasorelaxation after release from retinal vascular endothelial cells. ²⁵ This effect may have been overlooked in the present study because of the lack of penetration of the vascular endothelial cells of the PGI₂ applied to the extravascular retinal tissue. However, PGI₂ is the primary prostaglandin synthesized in pericytes, ²⁶ and PGI₂ has been shown to stimulate BkCa channels through PKC. ²⁷ Since both the activity of PKC and the function of retinal pericytes have been found to be disturbed in diabetes mellitus, ^28,29 changes in the metabolism of PGI₂ may be involved in the reduced tone regulation in retinal arterioles in diabetic patients. ^30,31  

Several studies have shown an involvement of prostaglandins in the regulation of arteriolar tone ^9,15,18 and have shown that inhibition of prostaglandin synthesis can significantly reduce hypercapnia-induced vasodilatation. ^5,24 In streptozotocin-induced diabetic rat retinas, prostaglandin synthesis has been shown to be increased by 40%, ³² and PGE₂ has been found to be significantly increased in diabetic versus normal rats, an effect that could be blocked by selective inhibition of the COX2 enzyme. ³³ This finding suggests that the metabolism of prostaglandin in the retinal may become a target for pharmacologic intervention in disturbances in retinal perfusion. The present study points to PGE, but also PGD and PGF_2α as possible candidates for such interventional studies. 

Footnotes

 Disclosure: K. Holmgaard, None; T. Bek, None