Effect of Glucose on Endothelin-1-induced Calcium Transients in Cultured Bovine Retinal Pericytes*

Published work has shown that endothelin-1-induced contractility of bovine retinal pericytes is reduced after culture in high concentrations of glucose. The purpose of the present study was to establish the profile of endothelin-1-induced calcium transients in pericytes and to identify changes occurring after culture in high concentrations of glucose. Glucose had no effect on basal levels of cytosolic calcium or on endothelin-1-induced calcium release from intracellular stores. However, influx of calcium from the extracellular medium after endothelin-1 stimulation was reduced in pericytes that had been cultured in 25 mm d-glucose. L-type Ca2+ currents were identified by patch clamping. The L-type Ca2+ channel agonist, (−)-Bay K8644, caused less influx of calcium from the extracellular medium in pericytes that had been cultured in 25 mm d-glucose than in those cultured with 5 mm d-glucose. However, 3-O-methylglucose, a nonmetabolizable analogue of glucose which can cause glycation, had similar effects to those of high concentrations of glucose. The results suggest that reduced function of the L-type Ca2+ channel that occurs in bovine retinal pericytes after culture in high concentrations of d-glucose is probably due to glycation of a channel protein.

Published work has shown that endothelin-1-induced contractility of bovine retinal pericytes is reduced after culture in high concentrations of glucose. The purpose of the present study was to establish the profile of endothelin-1-induced calcium transients in pericytes and to identify changes occurring after culture in high concentrations of glucose. Glucose had no effect on basal levels of cytosolic calcium or on endothelin-1-induced calcium release from intracellular stores. However, influx of calcium from the extracellular medium after endothelin-1 stimulation was reduced in pericytes that had been cultured in 25 mM D-glucose. L-type Ca 2؉ currents were identified by patch clamping. The L-type Ca 2؉ channel agonist, (؊)-Bay K8644, caused less influx of calcium from the extracellular medium in pericytes that had been cultured in 25 mM D-glucose than in those cultured with 5 mM D-glucose. However, 3-O-methylglucose, a nonmetabolizable analogue of glucose which can cause glycation, had similar effects to those of high concentrations of glucose. The results suggest that reduced function of the L-type Ca 2؉ channel that occurs in bovine retinal pericytes after culture in high concentrations of D-glucose is probably due to glycation of a channel protein.
In epidemiological studies hyperglycemia is the single factor most consistently associated with diabetic microangiopathy, including retinal microangiopathy (1). Longitudinal studies in diabetes have shown that alterations in capillary function and blood flow occur in advance of the structural changes seen in diabetic retinopathy such as pericyte loss, microaneurysm formation, and neovascularization (2). Pericyte tone normally regulates retinal capillary blood flow (3). Because retinal capillaries are devoid of extrinsic innervation (4), pericyte tone is controlled to a large extent by changes in the partial pressures of blood gases (5) and endothelium-derived agonists such as endothelin-1 (ET-1) 1 (6).
We have previously shown (7) that exposure to high ambient glucose concentrations in vitro caused a reduction in the contractile response of the bovine retinal pericyte (BRP) to ET-1 without any reduction in either ET-1 binding or generation of inositol trisphosphate (Ins(1,4,5)P 3 ). The aim of the present study was 1) to establish the profile of ET-1-induced changes in cytosolic calcium concentrations ([Ca 2ϩ ] i ) in the cultured pericyte and 2) to determine whether prior exposure to raised concentrations of glucose would alter ET-1-induced calcium transients in these cells.

EXPERIMENTAL PROCEDURES
Isolation and Culture of Bovine Retinal Pericytes-BRP were isolated as described by Gitlin and D'Amore (8). Briefly, bovine retinal homogenates were subjected to microscopically controlled enzyme digestion, and the resultant microvessel fragments were trapped on a 53-m mesh. The microvessels were suspended in Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Inc.) containing 5 mM glucose, antibiotics (200 units/ml penicillin and 200 g/ml streptomycin, Sigma), and 20% fetal calf serum (FCS, Life Technologies, Inc.), a growth medium which promotes pericyte proliferation and discourages the growth of endothelial cells. The suspension was then seeded into 25-cm 2 flasks (Falcon, Becton Dickinson Ltd., Cowley, UK) to be maintained at 37°C in a mixture of 5% CO 2 and air until confluent. The pericytes were identified by their stellate appearance and immunocytochemically by the presence of smooth muscle actin (Amersham Pharmacia Biotech); the lack of Factor VIII staining confirmed the absence of endothelial cells. All experiments were performed on BRP derived from passages 2 and 3.
Determination of [Ca 2ϩ ] i by Fura-2 Fluorescence Spectroscopy-BRP from a common cell pool were seeded onto glass slips and maintained in DMEM and 20% FCS containing 5 mM glucose, 25 mM glucose, or 5 mM glucose ϩ 20 mM 3-O-methylglucose (3-OMG) for 3, 7, or 10 days; experiments were also carried out using 5 mM glucose ϩ 20 mM mannitol to assess the effect of osmolality. Cells were serum-starved for 24 h prior to experiments by placing them in basal DMEM containing the appropriate hexose concentration. BRP were loaded with the intracellular calcium probe, fura-2, by incubating them in basal DMEM (5 mM glucose, 25 mM glucose, or 5 mM glucose ϩ 20 mM 3-OMG) containing 3 M fura-2-acetoxymethyl ester and 0.02% Pluronic F127 (Molecular Probes, Eugene, OR) for 80 min at 37°C. Cells were washed for 10 min in Hepes-buffered Krebs solution, pH 7.4, containing 0.1% bovine serum albumin, placed into the cuvette holder of a Perkin-Elmer LS-50B spectrofluorimeter, and stirred constantly throughout the procedure while being maintained at 37°C. 1 nM ET-1 (Bachem UK, Saffron Walden, UK) was added to the same medium that contained either 1 mM Ca 2ϩ or was essentially Ca 2ϩ -free. Cells in Ca 2ϩ -free medium were previously washed in this solution and had 0.1 mM EGTA added to the cuvette 15 s prior to the addition of ET-1. In parallel experiments cells were preincubated with the calcium channel inhibitor, 0.2 mM diltiazem (Sigma), for 30 min at 37°C prior to stimulation with ET-1 in the presence of 1 mM calcium. In a separate series of experiments cells were stimulated alone with 1 M dihydropyridine L-type calcium channel agonist (Ϫ)-Bay K8644 (Research Biochemicals Ltd., Natick, MA) (9). Fluorescence ratios were measured at 510 nM with excitation at 340 and 380 nM, and following determinations of R max and R min (the fluorescence ratios under saturating and Ca 2ϩ -free conditions, respectively), [Ca 2ϩ ] i was determined by the method of Grynkiewicz et al. (10). To determine R min , 4 M ionomycin was added to the cuvette at the end of the experiment, followed by 5 mM EGTA 10 s later. To determine R max , in alternate runs the [Ca 2ϩ ] i of the medium was raised to 4 mM (3 mM in the case of Ca 2ϩ -free buffer) 10 s after the addition of ionomycin. All plateau phase [Ca 2ϩ ] i calculations were made at a standard time of 210 s after application of the stimulus. Thus, the final result was the mean of 30 consecutive values recorded at 0.5-s intervals equally spaced before and after 210 s. DMEM contains iron. Free/redox-sensitive iron was measured in DMEM in the absence and presence of 10% FCS by the bleomycin method (11). 8-epi-PGF 2␣ levels in pericytes were measured by a specific enzyme immunoassay kit (Cayman Chemicals, Ann Arbor, MI).
Patch Clamping-Cultured pericytes were resuspended and allowed to settle out on 0.12-mm-thick glass slips. These slips were placed on a glass-bottomed recording chamber on an inverted microscope through which bathing solution flowed. Drug solutions were delivered through a separate flow tube, 0.3 mm in diameter, directed at the cell. Conventional whole cell patch recordings were made at 34°C from cells that had begun to flatten out. Recordings were made in voltage clamp mode, and the cells were held at Ϫ40 mV. The bathing medium contained the following (in mM, pH 7.3): sodium, 145; potassium, 6; magnesium, 1.3; calcium, 2; chlorine, 150; HEPES, 10; and glucose, 5. Theaflavin (20 mM) was added to block potassium currents. The solution in the patch electrode contained the following (in mM): cesium gluconate, 145; HEPES, 10; MgCl 2 , 1; EGTA, 5; Na 2 ATP, 2; phosphocreatine, 2; GTP, 0.1; and Ca 2ϩ , 50 nM (calculated).
Statistical Analysis-The Mann-Whitney U, Wilcoxon, and Student's t tests were used as appropriate. Within each experiment a given condition was represented by 2-4 replicates; p values less than 0.05 were taken to be significant.

ET-1-induced Changes in [Ca 2ϩ
] i -In 1 mM extracellular calcium the ET-1-evoked calcium transient was characterized by an initial peak in [Ca 2ϩ ] i followed by a sustained plateau phase during which the [Ca 2ϩ ] i remained elevated (Fig. 1A). In calcium-free medium the initial peak induced by ET-1 was attenuated, and the plateau phase was abolished (Fig. 1B). Under these conditions rises in [Ca 2ϩ ] i were due to ET-1/ Ins(1,4,5)P 3 -induced calcium release from the endoplasmic reticulum. If BRP were preincubated in medium containing 1 mM Ca 2ϩ and the calcium channel inhibitor, 0.2 mM diltiazem, for 30 min ET-1 produced an effect similar to that in Ca 2ϩ -free medium (Fig. 2). These observations indicate that a large portion of the ET-1-induced rise in [Ca 2ϩ ] i in BRP was due to influx from the extracellular medium.
Characterization of Ca 2ϩ Channels in the Plasma Membrane of BRP-(Ϫ)-Bay K8644 (10 -20 M) generated an enhanced inward current that activated at Ϫ30 mV and peaked at ϩ10 mV (107 Ϯ 70 pA) (Fig. 3). 30% of those cells showing stable patches had transient inward currents of 10 -200 pA that activated at Ϫ30 mV and peaked at 0 to ϩ10 mV (Fig. 4). The inward current was increased by raising the Ca 2ϩ concentration to 8 mM (Fig. 4). This was reduced by 80% with 0.2 M nifedipine (not shown) and completely blocked with 1 M nifedipine (Fig. 4) or 100 M Cd 2ϩ (not shown). All of these effects are consistent with a L-type Ca 2ϩ current.
Effect of Glucose on ET-1-induced Calcium Transients-Basal levels of [Ca 2ϩ ] i were similar in cells maintained in 5 and 25 mM glucose (Table I). Cells grown in 25 mM glucose for 10 days produced a smaller rise in [Ca 2ϩ ] i with ET-1 (1 nM) than those maintained in 5 mM glucose (Fig. 1A, Table II). No significant differences were seen at 3 or 7 days (results not shown). All subsequent experiments utilized cells grown for 10 days in 25 mM glucose. In Ca 2ϩ -free medium there was no difference in the [Ca 2ϩ ] i response to ET-1 between the two groups of cells.   (Table III).
To clarify whether the observed attenuation of the ET-1induced calcium transient in BRP was due to glucose metabo-lism or was in fact due to non-enzymic glycation of a functional calcium channel protein cells were also grown in the presence of the non-metabolizable glucose analogue 3-OMG. With ET-1, BRP cultured for 10 days in 5 mM glucose ϩ 20 mM 3-OMG showed lower plateau phase levels of [Ca 2ϩ ] i than those grown in 5 mM glucose (Table II, top). This suggests that the effect was not dependent on glucose metabolism by the pericyte. Cells grown in 5 mM glucose ϩ 20 mM mannitol did not display any differences in basal or ET-1-evoked [Ca 2ϩ ] i changes from those grown in 5 mM glucose alone (data not shown), indicating that the effects seen with 25 mM glucose were not due to an osmotic effect. When stimulated with (Ϫ)-Bay K8644, BRP grown in 5 mM glucose ϩ 20 mM 3-OMG also had a smaller increase in [Ca 2ϩ ] i than those grown in 5 mM glucose alone (Table III,  bottom).
We have previously shown that, in contrast to vascular smooth muscle cells, pericytes show no increase in malondialdehyde levels after 10 days of exposure to 25 mM glucose (12) suggesting the absence of free radical mediated damage. In this series of experiments the F 2 -isoprostane, 8-epi-PGF 2␣ , was measured. The levels were 100.6 pg/mg of protein and 94.4 pg/mg of protein for 5 and 25 mM glucose, respectively (mean, n ϭ 2). Bleomycin-detectable iron content of DMEM was 0.221 Ϯ 0.022 M (mean Ϯ S.D., n ϭ 3) in the absence of FCS. No bleomycin-detectable iron was found in DMEM containing 10% FCS. DISCUSSION ET-1 produced a biphasic rise in [Ca 2ϩ ] i that consisted of an initial peak followed by a sustained plateau phase. The plateau phase was abolished in Ca 2ϩ -free medium, showing that this phase arose from influx through Ca 2ϩ channels. The probability that these were L-type Ca 2ϩ channels was suggested by identification of a nifedipine-sensitive inward current and by (Ϫ)-Bay K8644-induced increases in [Ca 2ϩ ] i . The plateau phase was depressed in high concentrations of glucose; the implication of this is that culture in high concentrations of D-glucose attenuated Ca 2ϩ channel function. This idea was supported by the effects of (Ϫ)-Bay K8644, an L-type channel agonist that had a lesser effect on Ca 2ϩ influx in cells grown in 25 mM glucose compared with those grown in 5 mM glucose. These differences were unlikely to result from differences in resting membrane potential, which is unchanged by high concentrations of glucose (13). The present study shows that sustained exposure to high concentrations of glucose alters the properties of L-type Ca 2ϩ channels in BRP.
The initial phase of the ET-1-induced Ca 2ϩ transient is composed of Ins(1,4,5)P 3 -induced Ca 2ϩ release from the endoplasmic reticulum, and this is superimposed on the earliest part of influx from the extracellular medium. In the absence of extracellular Ca 2ϩ , the Ins(1,4,5)P 3 -induced increase in [Ca 2ϩ ] i was   similar in BRP grown in 5 or 25 mM glucose. This is consistent with previous results that showed that ET-1 binding and Ins(1,4,5)P 3 production by BRP were unaffected by glucose under similar culture conditions (7). Thus, the present results show that continued exposure to high concentrations of glucose attenuates the [Ca 2ϩ ] i rise produced by ET-1 by an action on L-type Ca 2ϩ channels but has no effect on that mediated by Ins(1,4,5)P 3 . The results from studies with 3-OMG strongly suggest that glucose affects the Ca 2ϩ channel by a mechanism involving glycation. We have not been able to detect any evidence of free radical attack, at least on lipids, either in this study where F 2 -isoprostanes were measured or in a previous study (12) where malondialdehyde was measured after 10 days of exposure to 25 mM glucose; we have detected increased malondialdehyde after 18 days only. The attenuated increase in [Ca 2ϩ ] i may explain, at least in part, the previously observed reduced contractile response to ET-1 of BRP grown in high concentrations of glucose despite the fact that ET-1 binding and Ins(1,4,5)P 3 production were unchanged (7).
In conclusion, culture of BRP in high concentrations of glucose for extended periods leads to reduced function of L-type Ca 2ϩ channels; the observed effects may be due to glycation of a channel protein. These results suggest that glycation of Ca 2ϩ channel proteins may affect the quality of retinal pericyte response to ET-1 and, hence, control of retinal capillary blood flow in the presence of sustained hyperglycemia.