Pioglitazone, a Peroxisome Proliferator–Activated Receptor-γ Agonist, Induces Dilation of Isolated Porcine Retinal Arterioles: Role of Nitric Oxide and Potassium Channels

Abstract

Purpose.: Pioglitazone, a peroxisome proliferator–activated receptor (PPAR)-γ agonist, has anti-inflammatory and atheroprotective effects on vascular tissue and may reduce cardiovascular risk in patients with diabetes. The effect of pioglitazone on the retinal microvascular diameter was examined, and it was determined whether the effect depends on the endothelium and/or potassium channels in smooth muscle to reveal the signaling mechanisms involved in this vasomotor activity.

Methods.: Porcine retinal arterioles were isolated, cannulated, and pressurized without flow in vitro. Video microscopic techniques recorded diametric responses to pioglitazone.

Results.: The retinal arterioles dilated in a concentration-dependent (10 nM–10 μM) manner in response to pioglitazone and decreased by 60% after endothelium removal. The nitric oxide (NO) synthase inhibitor N ^G-nitro-l-arginine methyl ester (l-NAME) inhibited pioglitazone-induced vasodilation comparable to denudation. Inhibition of soluble guanylyl cyclase (1H-1,2,4-oxadiazolo[4,3-a]quinoxalin-1-one), blockade of phosphatidylinositol (PI) 3-kinase (wortmannin), and pretreatment with compound C, an AMP-activated protein kinase (AMPK) inhibitor, were comparable to l-NAME. Pioglitazone-induced vasodilation also was inhibited by a nonselective K⁺ channel blocker, tetraethylammonium, and a voltage-gated K⁺ (Kv) inhibitor, 4-aminopyridine (4-AP). Treatment with intraluminal and extraluminal GW9662, a PPAR-γ antagonist, similarly inhibited pioglitazone-induced vasodilation. Co-administration of l-NAME and 4-AP almost eliminated pioglitazone-induced vasodilation.

Conclusions.: Pioglitazone elicits endothelium-dependent and -independent dilation of retinal arterioles mediated by NO release and Kv channel activation, respectively. The NO-mediated dilation pathway probably occurs via activation of guanylyl cyclase, PI3-kinase/Akt, and AMPK signaling. Understanding the effect of pioglitazone on retinal vasculature may provide new insights into therapeutic advances for treating diabetic retinopathy.

Type 2 diabetes mellitus, a multifactorial condition characterized by hyperglycemia, is attributable mainly to insulin resistance, leading to long-term microvascular and macrovascular complications such as neuropathy, nephropathy, retinopathy, and atherosclerosis. ¹ Although the results of the abnormalities in retinal blood flow (RBF) in patients with diabetes mellitus remain controversial, ^{2 –4} we recently showed that the RBF decreases in patients with type 2 diabetes mellitus without diabetic retinopathy (DR) and in those with mild DR, ⁵ suggesting that impaired retinal microcirculation may be involved in the development and progression of DR in patients with type 2 diabetes. Therefore, improvement of the impaired retinal circulation could lead to new therapeutic modalities for DR, especially in patients with type 2 diabetes. 

It has been reported that the thiazolidinediones (TZDs), commonly used as an insulin sensitizer in patients with type 2 diabetes, ⁶ may exert anti-inflammatory, ⁷ antiatherogenic, ⁸ neuroprotective, ⁹ and antioxidative ¹⁰ effects via stimulation of peroxisome proliferator-activated receptor (PPAR)-γ. Therefore, TZDs are potential therapeutic agents for diabetic microvascular complications such as DR. However, few studies have examined the effect of TZDs, including pioglitazone and rosiglitazone, on DR. Although TZDs have the potential to prevent proliferative DR in humans ¹¹ and inhibit laser photocoagulation-induced choroidal neovascularization ¹² and ischemic retinopathies ¹³ in animal models, the effect of TZDs on the retinal microvasculature is not fully understood. Therefore, we examined the effect of pioglitazone on the retinal microvascular diameter, using a vessel isolation technique to exclude the confounding effects of metabolic, hemodynamic, humoral, and glial/neuronal factors associated with in vivo experiments. We also investigated whether the effect of pioglitazone on retinal arterioles depends on the endothelium and/or potassium channels in smooth muscle, to reveal the signaling mechanisms involved in this vasomotor activity. 

Materials and Methods

Animal Preparation

The Animal Care Committee of Asahikawa Medical University approved all animal procedures, which were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Porcine eyes were enucleated immediately from pigs of either sex (age, 16–24 weeks; weight, 10–15 kg), after they were killed in a local abattoir and transported to the laboratory in a moist chamber on ice. The techniques for identification and isolation of retinal microvessels have been described. ^{14 –16} Briefly, the anterior segment and vitreous body were removed, guided by a dissection microscope. The posterior segment was placed in a cooled dissection chamber (∼8°C) containing a physiologic salt solution (PSS [in mM]: NaCl 145.0, KCl 4.7, CaCl₂ 2.0, MgSO₄ 1.17, NaH₂PO₄ 1.2, glucose 5.0, pyruvate 2.0, EDTA 0.02, and MOPS 3.0) with 1% albumin. Single second-order retinal arterioles (internal diameter in situ, 90–130 μm; length, 0.6–1.0 mm) were dissected with forceps under a stereomicroscope. After the residual neural connective tissues were removed, the arterioles were transferred for cannulation to a Lucite vessel chamber containing PSS-albumin solution equilibrated with room air at ambient temperature. One end of the arteriole was cannulated with a glass micropipette filled with a PSS-albumin solution, and the outside of the arteriole was secured to the pipette with an 11-0 ophthalmic suture (MANI, Tochigi, Japan). The other end of the vessel was then cannulated with a second micropipette and tied with a suture. After cannulation, the vessel and pipettes were transferred to the stage of an inverted microscope coupled to a video camera, a video micrometer (V-94; Living Systems Instrumentation, Burlington, VT), and a data-acquisition system (PowerLab; ADInstruments, Colorado Springs, CO) for continuous measurement and recording of the internal diameter during the experiment. ¹⁴ The micropipettes were connected to independent pressure reservoirs. By adjusting the reservoir height, we pressurized the vessel to 55 cmH₂O (∼40 mm Hg) intraluminal pressure without flow and kept it constant throughout the experiment. This pressure level was used based on documented pressure ranges in retinal arterioles in vivo. ¹⁷ To examine the role of the endothelium in a specific intervention, we inserted a PE-10 tube filled with a pharmacologic inhibitor into the sidearm of the micropipette holder and advanced to the tip of the cannulation-micropipette. The fluid in the cannulation-micropipette was replaced with a drug, and a small amount of flow was introduced into the vessel from the drug-filled micropipette by creating a pressure gradient (∼5 cmH₂O) across two pressure reservoirs. After perfusing the vessels with drug-containing solution for 10 minutes, the flow was stopped, and the vessels were incubated with an inhibitor for at least 30 minutes before agonist stimulation. 

Experimental Protocols

Control Experiment.

Cannulated arterioles were bathed in PSS at 36°C to 37°C to allow vessels to develop stable basal tone for approximately 30 to 40 minutes. The concentration-dependent vasodilatory response to pioglitazone, 10 nM to 10 μM, then was constructed based on evidence that the pioglitazone plasma level can reach 10 nM to 10 μM in patients treated with therapeutic doses. ¹⁸ The vessels were exposed to each concentration of agonists for 3 to 5 minutes until a stable diameter was established. As a result of the responses of each agonist concentration, the concentration-response curve was obtained. After measurement of the control concentration response of pioglitazone without any drugs, the vessels were washed with PSS to allow redevelopment of the basal tone. The vasodilation elicited by pioglitazone was re-examined after 30 minutes to confirm the reproducibility of the response (n = 6). 

Endothelial Denudation.

To examine whether the effect of pioglitazone on diametric regulation of retinal arterioles depends on the endothelium and/or potassium channels in smooth muscle, the following series of experiments was performed. The role of the endothelium in pioglitazone-induced dilation was evaluated by comparing the responses before and after endothelium removal. The vessel was perfused with a nonionic detergent, 3-[(3-cholamidopropyl) dimethylammonio]-1 propane sulfonate (CHAPS, 0.4%), for 1 to 2 minutes, to remove the endothelial cells. ^16,19 To ensure that the vascular smooth muscle function was uncompromised by CHAPS, we examined the concentration-dependent dilation of the vessel in response to the endothelium-independent vasodilator sodium nitroprusside (SNP; 0.1–100 μM) before and after denudation. Only vessels that exhibited the same basal tone as before CHAPS showed no vasodilatory reaction to the maximum dose of the endothelium-dependent vasodilator bradykinin (10 nM) ¹⁹ and that showed unaltered vasodilation in response to SNP after CHAPS were accepted for further study with pioglitazone, to determine the effect of removal of the endothelium from isolated retinal arterioles. 

To confirm the removal of the endothelium by CHAPS immunohistochemically, retinal arterioles were embedded and frozen in OCT compound (Tissue-Tek; Sakura Finetek, Torrance, CA) after perfusion of CHAPS or PSS (as a control). Frozen sections (10-μm thick) were fixed in 4% paraformaldehyde and immunolabeled with either an anti-α-smooth muscle actin antibody (Sigma-Aldrich) or an anti-eNOS (Santa Cruz Biotechnology, Santa Cruz, CA) antibody. Afterward, the slides were incubated with Cy3-labeled or fluorescein-labeled (GE Healthcare Life Sciences, Piscataway, NJ) secondary antibodies. The slides were observed for red (Cy3) and green (fluorescein) images and analyzed with a confocal microscope (Fluoview FV 1000; Olympus, Tokyo, Japan). Merged images were created with Image J software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). 

Mechanistic Study of Pioglitazone-Induced Dilation

Role of Endothelium-Derived Factors.

The involvement of endothelium-derived vasodilators (i.e., prostaglandins, nitric oxide [NO], and cytochrome P450 metabolites) in mediating the vascular response was assessed in the presence of known effective concentrations of the specific enzyme inhibitors indomethacin (10 μM), ^20,21 N ^G-nitro-l-arginine methyl ester (l-NAME, 10 μM), ^14,20 and sulfaphenazole (10 μM), ²⁰ respectively. The role of guanylyl cyclase/cyclic guanosine monophosphate (cGMP) signaling was assessed by treating the vessels with the soluble guanylyl cyclase inhibitor 1H-1,2,4-oxadiazolo[4,3-a]quinoxalin-1-one (ODQ, 0.1 μM). ^15,16 To study the involvement of phosphatidylinositol (PI) 3-kinase and AMP-activated protein kinase (AMPK) in the endothelium, we studied the pioglitazone-induced response after incubating intraluminally the vessels with the PI3-kinase inhibitor wortmannin (0.1 μM) ²¹ and the AMPK inhibitor compound C (10 μM). ²² Moreover, because AMPK is expressed in endothelial and smooth muscle cells, we added extraluminal compound C (10 μM). Because PPAR-γ is also expressed in endothelial ²³ and smooth muscle cells, ²⁴ we examined the role of PPAR-γ on the effect of pioglitazone using the isolated vessels pretreated intraluminally and extraluminally with the PPAR-γ antagonist GW9662 (10 μM) ^25,26 1 hour before measuring the pioglitazone-induced response. 

Role of the Potassium Channels.

Since activation of potassium channels is thought to be a major mechanism in vasodilation of the retinal arterioles, ^14,15,27,28 we examined this pathway by treating the vessels with various potassium channel inhibitors: the nonselective potassium channel blocker tetraethylammonium (TEA, 10 mM), ²⁹ the large-conductance Ca²⁺-activated potassium channel blocker iberiotoxin (0.1 μM), ^29,30 the small-conductance Ca²⁺-activated potassium channel blocker apamin (0.1 μM), ³¹ the voltage-gated K⁺ (K_V) channel inhibitor 4-aminopyridine (4-AP) (0.1 mM), ³² the adenosine triphosphate-sensitive potassium channel blocker glibenclamide (5 μM), ²⁰ and the inward rectifier K⁺ (K_ir) channel blocker BaCl₂ (30 μM). ³³  

Response to SNP.

SNP (1 nM-10 μM) was used to probe endothelium-independent NO-mediated vasodilation. The vascular response to SNP was examined in the presence of various interventions, as mentioned previously. 

All drugs were administered extraluminally, unless otherwise stated. In addition the extraluminal administration, which may involve both the endothelium and smooth muscle, we added the AMPK inhibitor compound C and the PPAR-γ inhibitor GW9662 intraluminally to confirm that drugs can reach the endothelium without permeating the smooth muscle extraluminally, because AMPK and PPAR-γ are reported to be expressed in both the endothelium and smooth muscle, as mentioned previously. Each pharmacologic inhibitor was incubated with the vessels for at least 30 minutes. 

Measurement of Nitrite/Nitrate

The stable NO end products nitrite and nitrate, collectively NOx, were measured by high-performance liquid chromatography (ENO-20; Eicom, Kyoto, Japan). We collected samples from chambers 5 minutes after administration of pioglitazone 10 μM. After injection of 10 μL of the pretreated sample into the system, NOx production was measured by the Griess method. ³⁴  

Chemicals

Pioglitazone was obtained from LKT Laboratories, Inc. (St. Paul, MN). Compound C was obtained from Calbiochem (San Diego, CA). Other drugs were obtained from Sigma-Aldrich (St. Louis, MO). l-NAME, TEA, iberiotoxin, 4-AP, apamin, BaCl₂, and SNP were dissolved in PSS. Indomethacin, sulfaphenazole, and ODQ were dissolved in ethanol. Pioglitazone, wortmannin, GW9662, glibenclamide, and compound C were dissolved in dimethyl sulfoxide (DMSO). Subsequent dilutions of these drugs were prepared in PSS. The final concentrations of ethanol and DMSO in the vessel bath were less than 0.1%. ¹⁵ Vehicle control studies indicated that these final concentrations of solvents did not affect the arteriolar function. 

Data Analysis

At the end of each experiment, the vessel was relaxed in EDTA (1 mM) calcium-free PSS to obtain the maximum diameter at 55 cmH₂O intraluminal pressure. ^14,16 The diametric changes in response to pioglitazone and SNP were normalized to the resting diameters and expressed as percentage change in diameter. ^14,16 Data are reported as the mean ± SEM, and n represents the number of vessels studied. Statistical comparisons of the change in resting tone by antagonists were performed with the Student's t-test. Differences between pioglitazone and vehicle treatment in NOx production were examined using the Mann-Whitney U test. The two-way analysis of variance, followed by the Bonferroni multiple-range test, was used to determine the significance of the difference between control and experimental interventions. P < 0.05 was considered significant. 

Results

Dilation of Retinal Arterioles Induced by Pioglitazone

The basal tone in all vessels (n = 105) ranged from 50% to 70% (average, ∼63% ± 1%) of their maximum diameter. The average resting and maximum vessel diameters were 64 ± 2 and 103 ± 3 μm, respectively. Pioglitazone induced consistent concentration-dependent maximum dilation of the retinal arterioles within 2 to 3 minutes. The threshold concentration for vasodilation was 3 nM, and the highest concentration (10 μM) elicited approximately 50% of the maximum dilation (Fig. 1). Further study showed that pioglitazone-induced dilation was reproducible and did not deteriorate after repeated application (Fig. 1). 

Figure 1.

 

The concentration-dependent vasodilatory effect of pioglitazone was examined (first trial, resting diameter, 65 ± 3 μm; maximum diameter, 103 ± 3 μm), and then the trial was repeated after a 30-minute washout period (second trial, resting diameter, 63 ± 3 μm; maximum diameter, 103 ± 3 μm).

Role of the Endothelium

In another series of experiments, 12 vessels were subjected to the denudation protocol. After perfusion with CHAPS, 3 of the 12 vessels lost basal tone and three showed partial inhibition by the endothelium-dependent vasodilator bradykinin. These apparently damaged or partially denuded vessels were excluded from further study. The remaining six vessels maintained basal tone (control 65% ± 3% vs. denudation 62% ± 3%; P = 0.15) and the vasodilation induced by bradykinin (10 nM) was abolished (control 91% ± 2% vs. denudation: 1% ± 1%). In addition, these vessels exhibited normal vasodilation in response to SNP (Table 1). In these six denuded vessels, the pioglitazone-induced dilation decreased partly (P = 0.001; Fig. 2). 

Table 1.

 

Diametric Responses of Retinal Arterioles to SNP

	n	SNP (μM)
		0.1	1	10	100
Control	10	5.2 ± 1.0	21.2 ± 2.3	48.7 ± 4.0	86.2 ± 1.7
Denudation	4	6.6 ± 3.8	30.5 ± 7.2	59.2 ± 8.6	88.3 ± 4.0
L-NAME	4	11.5 ± 4.6	32.8 ± 4.0	50.7 ± 6.0	87.3 ± 3.4
Wortmannin	6	3.7 ± 4.1	22.0 ± 2.4	45.9 ± 5.9	82.2 ± 4.1
Compound C intraluminally	4	5.9 ± 2.2	17.8 ± 4.4	39.8 ± 7.4	83.8 ± 5.5
Compound C extraluminally	4	7.0 ± 2.6	22.1 ± 4.8	48.0 ± 8.1	84.7 ± 5.7
GW9662 extraluminally	4	9.9 ± 2.0	28.6 ± 6.3	57.8 ± 7.2	82.4 ± 1.3
GW9662 intraluminally	4	8.8 ± 1.7	22.0 ± 1.2	47.4 ± 4.7	80.7 ± 4.1
GW9662 + TEA	4	7.4 ± 0.6	19.9 ± 0.6	43.5 ± 1.4	92.7 ± 2.0
TEA	6	8.7 ± 3.7	33.1 ± 4.7	63.1 ± 3.7	90.9 ± 3.1
TEA + denudation	4	5.0 ± 1.4	17.8 ± 0.7	59.3 ± 4.6	88.3 ± 2.7
TEA + l-NAME	4	6.6 ± 1.9	28.6 ± 9.2	55.3 ± 8.5	86.7 ± 1.2
4-AP	5	9.1 ± 1.9	26.8 ± 5.1	57.7 ± 5.9	83.1 ± 1.1
4-AP + l-NAME	4	8.6 ± 2.4	27.3 ± 6.5	58.9 ± 7.5	82.4 ± 1.1

 Data are expressed as the mean percentage of maximum dilation ± SEM.

 Based on two-way ANOVA, the responses to SNP were unaffected by any perturbations, compared with the control. n, number of vessels.

Figure 2.

 

Effect of the removal of the endothelium by perfusion with 0.4% CHAPS. *P < 0.05 versus control.

In addition, we confirmed histologically that the endothelial cells were almost completely removed after application of CHAPS in other vessels (n = 4) that maintained basal tone and had no responses to bradykinin and normal responses to SNP after CHAPS (Fig. 3). 

Figure 3.

 

Immunohistochemical imaging of denudation in isolated retinal arterioles. Isolated and pressurized retinal arterioles were incubated intraluminally with PSS (control) or PSS containing CHAPS (denudation) for 1 to 2 minutes, followed by addition of anti-eNOS (green) or anti-α-smooth muscle actin (SMA, red) antibody. (A, Control) High levels of immunostaining were detected in endothelial (arrow) and smooth muscle (arrowhead) layers, respectively. The merged image shows the lack of overlap staining. (B, Denudation) Immunostaining was detected in smooth muscle for SMA but was only slightly detected in endothelium for eNOS. The merged image shows the lack of overlap staining. Data shown are representative of three separate experiments. Bar, 50 μm.

Role of Endothelium-Derived Factors

Inhibition of cytochrome P450 epoxygenase and prostaglandins by sulfaphenazole and indomethacin did not affect the vasodilatory response to pioglitazone (Fig. 4A). The NOS inhibitor l-NAME inhibited pioglitazone-induced vasodilation (P < 0.001, Fig. 4A) comparable to that produced by denudation (l-NAME versus denudation; P > 0.05; Figs. 2, 4A). The NOx level in the chamber significantly increased 5 minutes after administration of pioglitazone compared to the vehicle (Fig. 4B). ODQ did not change the basal tone and significantly reduced the vasodilatory response to pioglitazone in a manner similar to l-NAME (Fig. 5). The intraluminal PI3-kinase blocker, wortmannin, and the AMPK blocker, compound C, significantly reduced pioglitazone-induced vasodilation in the same manner as l-NAME (Fig. 6). Further, extraluminal and intraluminal compound C significantly decreased pioglitazone-induced dilation to the same extent (Fig. 6). Intraluminal and extraluminal GW9662 significantly inhibited vasodilation in response to pioglitazone to a similar extent (Fig. 7), indicating that pioglitazone-induced dilation of the retinal arterioles was involved with PPAR-γ in the endothelial cells. A higher concentration of and longer pretreatment time with GW9662 did not inhibit additional pioglitazone-induced vasodilation (data not shown). Co-administration of intraluminal GW9662 and TEA almost eliminated pioglitazone-induced vasodilation. These agents did not affect the basal tone. 

Figure 4.

 

(A) Effect of incubation with the NOS inhibitor l-NAME (10 μM), the cyclooxygenase inhibitor, indomethacin (10 μM), and the cytochrome P450 epoxygenase inhibitor, sulfaphenazole (10 μM). *P < 0.05 versus control. (B) The NOx production response to pioglitazone (10 μM) or vehicle was examined 5 minutes after injection of pioglitazone or vehicle into the chamber. *P < 0.05 versus vehicle.

Figure 5.

 

Effect of incubation with the soluble guanylyl cyclase inhibitor ODQ (0.1 μM). *P < 0.05 versus control.

Figure 6.

 

Effect of intraluminal incubation with the PI3-kinase inhibitor wortmannin and the AMPK inhibitor compound C. To study the involvement of AMPK in retinal smooth muscle arterioles, the concentration-dependent vasodilatory response was examined before and after extraluminal pretreatment with compound C. *P < 0.05 versus control.

Figure 7.

 

Effect of extraluminal and intraluminal incubation with the PPAR-γ inhibitor GW9662 (10 μM) for 1 hour. Residual vasodilation in the presence of GW9662 (10 μM) also was examined after co-incubation with TEA (10 mM). *P < 0.05 versus control.

Role of Potassium Channels

TEA significantly inhibited pioglitazone-induced vasodilation of the retinal arterioles (Fig. 8A). To determine whether the K channel related to vasodilation by pioglitazone is present in endothelial cells or smooth muscle cells, we incubated TEA in denuded vessels. The residual vasodilation after denudation decreased further with subsequent TEA treatment (Fig. 8A). In addition, 4-AP attenuated pioglitazone-induced dilation of the retinal arterioles similar to that attenuated by TEA (Figs. 8A, 8B), but glibenclamide, BaCl₂, iberiotoxin, and apamin were ineffective (Fig. 8B). The residual vasodilation in the presence of extraluminal TEA and 4-AP decreased further with subsequent l-NAME treatment (Figs. 8A, 8B). Any pretreatment did not significantly alter the basal tone. 

Figure 8.

 

(A) Effect of extraluminal incubation with TEA (10 mM). Residual vasodilation in the presence of TEA also was examined after denudation and co-incubation with l-NAME (10 μM). (B) Effect of the incubation with specific potassium channel blockers (i.e., 4-AP, glibenclamide, BaCl₂, apamin, and iberiotoxin). Residual vasodilation in the presence of 4-AP (0.1 mM) was examined after co-incubation with l-NAME (10 μM). *P < 0.05 versus control.

Response to SNP

Various interventions did not affect the SNP-induced dilation of the retinal arterioles (Table 1), suggesting that vascular smooth muscle function was unaltered by these interventions. 

Discussion

The present study showed for the first time that pioglitazone, a PPAR-γ agonist, induces concentration-dependent vasodilation of the retinal arterioles with approximately 50% dilation at high concentrations (3–10 μM; Fig. 1). Previous studies also have reported that pioglitazone 10 μM elicits the same degree of dilation in rat aorta ^35,36 and rat mesenteric artery. ³⁷ Those results were comparable to our findings, suggesting that the vasodilatory effect of pioglitazone is favorable in large arteries as well as the microvasculature, including the retinal arterioles. Because the plasma pioglitazone concentration reaches 0.83 or 1.82 μM within 2 to 3 hours after oral administration of 15 or 30 mg, respectively, of pioglitazone in patients with type 2 diabetes mellitus, ¹⁸ the current data clearly showed that pioglitazone may have clinical potential to elicit 30% to 40% vasodilation of the retinal arterioles at these concentrations (Fig. 1) in patients with type 2 diabetes. Because we found that the RBF decreases in patients with type 2 diabetes and no or minimal retinopathy, ⁵ the vasodilatory potential of pioglitazone on the retinal arterioles may shed light on the novel treatment of DR. Further clinical study is needed to examine the potential of pioglitazone as a novel treatment of retinal vascular disorders. 

Animal studies to examine the effects of pioglitazone on vascular responses in the aorta have reported conflicting results as to whether pioglitazone causes endothelium-dependent ³⁸ or -independent vasodilation, ³⁹ or both. ³⁵ In the present study, we first observed that the removal of the vascular endothelium by CHAPS, which we confirmed with immunohistochemical staining (Fig. 3), significantly decreased but did not eliminate the pioglitazone-induced vasodilation, suggesting that there are both endothelium-dependent and -independent pathways in pioglitazone-induced vasodilation of the retinal arterioles. 

We also observed that NOS blockade and denudation to a similar extent inhibited pioglitazone-induced vasodilation (Fig. 4A) and the levels of NO metabolites (nitrite and nitrate) were elevated in the chamber after pioglitazone administration (Fig. 4B), suggesting that the retinal arterioles dilated via NO production from the endothelium in the retinal arterioles. NO involvement in pioglitazone-induced vasodilation was also reported in isolated rat ³⁵ and mouse aortas. ³⁸ Collectively, it is reasonable that NO contributes greatly to the endothelium-dependent component of pioglitazone-induced vasodilation in the retinal microcirculation. In contrast to l-NAME, the vasodilatory response to pioglitazone was unaffected by inhibition of synthesis of prostacyclin or cytochrome P450 metabolites, the other two vasodilators secreted from the endothelium (Fig. 4A). Taken together, NO primarily contributes the endothelium-dependent component of pioglitazone-induced vasodilation of the retinal arterioles, independent of prostaglandins and cytochrome P450 metabolites. 

We found that pioglitazone caused dilation of the retinal arterioles within a few minutes. Although no study has reported that pioglitazone stimulates eNOS phosphorylation in human aortic endothelial cells in less than 1 hour, ²⁶ we speculated that pioglitazone may phosphorylate eNOS within a few minutes in the endothelial cells of the retinal arterioles. This rapid action was supported by the observation that NOx significantly increased in the vessel chamber 5 minutes after pioglitazone injection (Fig. 4B). Further study is needed to elucidate whether pioglitazone induces rapid eNOS phosphorylation in retinal arterioles. 

Among pathways that are involved in NO vasodilation, the NO/cGMP pathway is considered a major vasodilatory mechanism, ⁴⁰ but previous reports have not confirmed that pioglitazone increases the cGMP level. Inhibition of pioglitazone-induced vasodilation by ODQ incubation (Fig. 5) was comparable to l-NAME (Fig. 4A) in the present study, suggesting that vasodilation of the retinal arterioles induced by pioglitazone occurs via the NO/cGMP pathway. 

Inhibition of the PI3-kinase/Akt pathway reduced pioglitazone-induced vasodilation of the retinal arterioles in the same manner as l-NAME and denudation (Fig. 6). It was previously reported that pioglitazone protected against ischemia/reperfusion injury in isolated rat heart through activation of the PI3-kinase/Akt pathway. ⁴¹ It is likely that activation of the PI3-kinase-Akt-eNOS pathway in the vascular endothelium plays an important role in pioglitazone-induced dilation of the retinal arterioles. 

Although AMPK was expressed in endothelial ⁴² and smooth muscle cells, ⁴³ incubation with pioglitazone stimulated AMPK activity and phosphorylation in human aortic endothelial cells. ²⁶ Therefore, the current finding that extraluminal and intraluminal compound C inhibited to a similar extent the vasodilatory response to pioglitazone (Fig. 6) indicates that pioglitazone-induced dilation of the retinal arterioles may be involved in AMPK activation in endothelial cells. In addition, the inhibitory effects of compound C were comparable to those of wortmannin and l-NAME (Fig. 6), suggesting that both PI3-kinase-Akt and AMPK are involved in eNOS phosphorylation and consequent vasodilation of the retinal arterioles elicited by pioglitazone. It has been reported that AMPK activation may be upstream from that of Akt in the endothelial cells ⁴⁴ and that PI3-kinase may be the upstream kinase for the Akt and AMPK ⁴⁵ pathways. Taking these findings together, we speculate that activation of the PI3-kinase-AMPK-Akt-eNOS pathway is involved in pioglitazone-induced vasodilation of the retinal arterioles. 

Although PPAR-γ is expressed in the endothelial cells ²³ and smooth muscle cells ²⁴ of the vessel wall, there have been conflicting findings that pioglitazone induces PPAR-γ-dependent vasodilation in mouse aorta ³⁸ or PPAR-γ-independent vasodilation in isolated rat aorta. ³⁵ We also found that extraluminal and intraluminal administration of the PPAR-γ antagonist GW9662 inhibited vasodilation in response to pioglitazone to the same extent, indicating that pioglitazone-induced vasodilation may be involved in activation of PPAR-γ in the endothelium of retinal arterioles, probably resulting in increased NO production in the endothelium. Moreover, the current finding that inhibitors of PPAR-γ also reduced pioglitazone-induced vasodilation in a manner similar to PI3-kinase/Akt and AMPK suggested that NO production by pioglitazone is PPAR-γ-dependently generated from the endothelium of the retinal arterioles, probably via activation of the PI3-kinase-Akt pathways and AMPK leading to phosphorylation of eNOS. 

Previous studies have reported the involvement of various potassium channels in the vasodilation of the retinal arterioles. In porcine retinal arterioles without endothelium, the small- and intermediate-conductance Ca²⁺-activated potassium channel opener NS309 caused dilation. ²⁷ In addition, it has been reported that specific potassium channel blockers (i.e., apamin, ²⁷ iberiotoxin, ²⁸ and glibenclamide ^14,15 ) attenuated the drug/agonist-induced vasodilation of the porcine retinal arterioles. The current data show that the Kv channel blocker 4-AP inhibited pioglitazone-induced vasodilation (Fig. 8A) in the same manner as the nonselective K channel blocker TEA, whereas this vasodilation was unaffected by other specific K channel blockers, suggesting that mainly the Kv channel is involved in the pioglitazone-induced vasodilation of the retinal arterioles. Pioglitazone inhibited voltage-dependent Ca²⁺ currents in vascular smooth muscle cells of rat aorta ³⁹ and guinea pig mesenteric artery. ⁴⁶ Furthermore, pioglitazone-induced vasorelaxation of vascular smooth muscle in rat aorta is mediated by activation of K_V channels and K_ir channels. ³⁵ Their findings were comparable to ours obtained from the retinal arterioles. Moreover, pioglitazone-induced vasodilation was almost eliminated by 4-AP and l-NAME coadministration (Fig. 8A) and TEA incubation after denudation (Fig. 8B). Taken together, we speculate that pioglitazone-induced dilation of retinal arterioles may be induced primarily through the activation of potassium channels, mainly Kv channels, in smooth muscle and NO production in endothelial cells. 

In summary, we showed that pioglitazone, a PPAR-γ agonist, elicits potent dilation of the retinal arterioles, which has two components of endothelium-dependent and -independent pathways. Endothelium-dependent dilation is mediated via activation of the AMPK and PI3-kinase-Akt-eNOS signaling pathways for NO release and consequent activation of the soluble guanylyl cyclase/cGMP pathway. Endothelium-independent dilation is related mainly to activation of the Kv channel in smooth muscle. Taking all evidence together, pioglitazone induces vasodilation of retinal arterioles through PPAR-γ activation in endothelial cells. Because RBF is impaired in early-stage DR in patients with type 2 diabetes mellitus, ⁵ pioglitazone-induced vasodilation may be a novel potential drug for treating DR. Further clinical study is needed to determine whether pioglitazone can improve impaired RBF in patients with type 2 diabetes and retinal vascular disorders. 

Footnotes

 Supported by a Grant-in-Aid for Scientific Research (C) 18591904 from the Ministry of Education, Science, and Culture, Tokyo, Japan; the Uehara Memorial Foundation, the Takeda Foundation, and the Akiyama Life Science Foundation (TN).

Footnotes

 Disclosure: T. Omae, None; T. Nagaoka, None; I. Tanano, None; A. Yoshida, None

The authors thank Lih Kuo and Travis W. Hein for technical support and Lynda Charters for reviewing the manuscript.