Purinergic regulation of cation conductances and intracellular Ca²⁺ in cultured rat retinal pigment epithelial cells

Abstract

1. We used whole-cell patch clamp and fluorescent calcium imaging techniques to investigate the effects of adenosine 5′-triphosphate (ATP) on membrane currents and intracellular calcium concentration ([Ca²⁺]_i)in rat retinal pigment epithelial (RPE) cells. In 62 % of RPE cells, application of 100 μM ATP elicited a fast inward current at negative membrane potentials. In 38 % of RPE cells recorded, a biphasic response to ATP was observed in which activation of the fast inward current was followed by activation of a delayed outward current.

2. The ATP-activated inward current was a non-selective cation (NSC) current that showed inward rectification, reversed at −1.5 ± 1 mV and was permeable to monovalent cations. The NSC current was insensitive to the P2 purinoceptor antagonists, suramin or PPADS but was activated by the purinoceptor agonists UTP, ADP and 2MeSATP.

3. The outward current activated by ATP reversed at −68 ± 3 mV (equilibrium potential for potassium (E_K) = −84 mV) and was blocked by Ba²⁺ ions, consistent with the activation of a K⁺ conductance. The outward K⁺ conductance was also reduced by the maxi-K_Ca channel blocker iberiotoxin (IbTX; 10 nM), suggesting that ATP activated an outward Ca²⁺-activated K⁺ channel in rat RPE cells. The Ca²⁺-activated K⁺ current (I_K(Ca)) was also activated by the purinoceptor agonists UTP, ADP and 2MeSATP.

4. In fluo-3 or fluo-4 loaded RPE cells, ATP and the pyrimidine agonist UTP elevated [Ca²⁺]_i. The increase in Ca²⁺ was not dependent on extracellular Ca²⁺ influx, but was sensitive to the Ca²⁺-ATPase inhibitor thapsigargin, confirming the involvement of intracellular Ca²⁺ stores release.

5. These results suggest that rat RPE cells express both P2X purinoceptors that gate activation of a non-selective cation conductance and G protein-coupled P2Y purinoceptors that mediate Ca²⁺ release from intracellular stores and activation of a calcium-activated K⁺ current.

The retinal pigment epithelium (RPE) carries out a number of roles that are essential for the maintenance and viability of the neurosensory retina. These roles include phagocytosis of shed rod and cone outer segments, melanin synthesis and recycling and regulation of subretinal volume via ion-coupled fluid absorption (Steinberg & Miller, 1979; Zinn & Benjamin-Henkind, 1979; Clark, 1986). In order to carry out these diverse functions, the RPE must be able to detect and respond to paracrine signals coming from the adjacent choroidal and/or neural retinal tissue and/or via systemic sources.

A number of metabotropic receptors have been identified on the RPE including those for dopamine, acetylcholine, adrenaline (epinephrine) and adenosine (Friedman et al. 1988; Dearry et al. 1990; Frambach et al. 1990). Activation of these receptors by their respective signalling molecules has been linked to changes in light-evoked responses (Dearry et al. 1990; Gallemore & Steinberg, 1990), phagocytic ability (Gregory et al. 1994) and ion and fluid transport across the RPE (Edelman & Miller, 1991; Joseph & Miller, 1992). Recently, in monolayers of bovine and rat RPE, extracellular adenosine 5′-triphosphate (ATP) and uridine triphosphate (UTP) were demonstrated to induce changes in intracellular Ca²⁺ and transepithelial ion and fluid movement (Stalmans & Himpens, 1997; Peterson et al. 1997). A role for intracellular ATP has also recently been demonstrated for the activation of a delayed inwardly rectifying K⁺ current (I_K(IR)) in isolated bovine RPE cells (Hughes & Takahira, 1998). These findings support the presence of metabotropic purinoceptors and suggest that ATP may act as an important paracrine signal in the RPE.

Purinoceptors are divided into two main classes, P1 and P2, based on their selectivity for adenosine and ATP, respectively (Burnstock & Kennedy, 1985). Adenosine or P1 receptors are G protein-coupled receptors that respond only to purine nucleosides and nucleoside monophosphates. ATP or P2 receptors are activated primarily by purine or pyrimidine nucleotide triphosphates, and are organized into two major subclasses: P2X purinoceptors and the P2Y purinoceptors (Barnard et al. 1997). P2X receptor subtypes comprise a novel family of ligand-gated cation channels (Barnard, 1992). Seven P2X receptor subtypes have been identified and classified on the basis of sequence homology, agonist/antagonist sensitivity and kinetics of receptor desensitization. The seven subtypes of P2Y receptors thus far identified belong to the larger superfamily of seven transmembrane proteins that are G protein-coupled (metabotropic) receptors. Although signalling molecules downstream from P2Y receptors have not been fully classified, all subtypes to date have been linked by heterotrimeric G proteins to phospholipase C (PLC) activation, generation of inositol trisphosphate (IP₃) and diacylglycerol (DAG), and subsequent release of Ca²⁺ from internal stores (Barnard et al. 1994).

Pharmacological and functional data support the presence of P2Y purinoceptors in rat and bovine RPE (Stalmans & Himpens, 1997; Peterson et al. 1997). More recently, P2Y₂ purinoceptors were found to be expressed in cultured human RPE cells (Sullivan et al. 1997). In the present study we report that extracellular ATP, UTP and other nucleotides, but not adenosine, activate two distinct cation conductances. The rapid activation of an inward cation current suggests for the first time, the presence of functional P2X purinoceptors in the rat RPE. We also provide evidence for the presence of G protein-coupled P2Y purinoceptors linked to increases in [Ca²⁺]_i and activation of a Ca²⁺-sensitive outward K⁺ conductance.

Our findings that extracellular ATP (or UTP) mediates cation channel activation and [Ca²⁺]_i mobilization in the RPE support a role for ATP and/or UTP as paracrine signals in the RPE and suggest that they may act to regulate RPE function under physiological and/or pathological conditions.

METHODS

Cell dissociation and culture

Long-Evans rats (age 8–14 days) were anaesthetized using halothane, killed by decapitation and the eyes enucleated. All procedures were carried out in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The RPE cells were isolated using a modification of the method of Wang et al. (1993). Enucleated eyes were placed in calcium-free Hanks' solution containing EDTA (CFHE; Gibco BRL, Burlington, Ontario). The connective tissue was removed and the globes were bisected along the equator just posterior to the ora serrata, leaving the neural retina attached to the posterior eye cup. The anterior portion of the eye containing the cornea, lens and vitreous was discarded. The posterior eye cup was then placed in calcium-free Hanks' EDTA solution containing 220 U ml⁻¹ hyaluronidase type III (Sigma Chemical Company, St Louis, MO, USA) and 65 U ml⁻¹ collagenase A (Boehringer Mannheim, Laval, Quebec) for 10 min at 37°C. Under a dissecting microscope, in fresh CFHE without enzyme, the neural retina was gently peeled from the eye cup using forceps. The remaining posterior segment was incubated at 37°C for 5 min in CFHE enzyme solution to facilitate the dissociation of the RPE from Bruch's membrane. Following a second transfer to fresh CFHE, sheets of RPE cells were gently removed from the remaining eye cup using forceps so as not to disturb the underlying choroid. Retinal pigment epithelial tissue was triturated through a narrow-bore fire-polished Pasteur pipette to yield a suspension of single cells and small clumps of RPE tissue. The cell suspension was washed and resuspended in 500 μl of Dulbecco's modified Eagle's medium (DMEM) plus 20 % fetal calf serum, 0.5 % penicillin-streptomycin (Gibco BRL, Burlington, Ontario), and gentamicin (50 μg ml⁻¹; Sigma). Retinal pigment epithelial cells were seeded onto glass coverslips (12 mm diameter) in 4-well culture dishes and placed in a 37°C incubator with an atmosphere of 5 % CO₂-95 % O₂. The medium was changed to DMEM containing 10 % fetal calf serum and antibiotics 24 h following the initial plating.

Solutions and drugs

Retinal pigment epithelial cells, attached to glass coverslips, were placed in a shallow recording chamber (volume ∼1 ml) and positioned on the stage of a Nikon inverted microscope (Nikon Instruments, Canada Inc.). The recording chamber was superfused (1–2 ml min⁻¹) with a variety of solutions via gravity inflow from elevated reservoirs. The flow rate was regulated by a series of valves. Solutions containing NaHCO₃ were continuously bubbled with 5 % CO₂-95 % O₂. All extracellular and intracellular solutions were adjusted to pH 7.3–7.4, and the osmolarity was measured by freezing point depression (Osmette A, Fischer Scientific, Nepean, Ontario). Extracellular solutions had a final osmolarity of 330–340 mosmol l⁻¹ and the osmolarity of intracellular solutions was between 310 and 320 mosmol l⁻¹. The use of a slightly hyperosmotic external solution was found to be effective in eliminating transient changes in ionic conductances, which occurred as a result of osmotic changes during the initial period of whole-cell recording. The standard extracellular solution (5 K⁺) contained (mM): NaCl, 130; KCl, 5; Na-Hepes, 10; NaHCO₃, 10; MgCl₂, 1; CaCl₂, 1; and glucose, 10. Replacement of extracellular Na⁺ was accomplished using equimolar substitution with either choline chloride or N-methyl-D-glucamine (NMDG) chloride. In some experiments, the extracellular solution was made nominally Ca²⁺ free by including 1.5 mM EGTA with 0.05 mM CaCl₂. The extracellular calcium concentration with this substitution was estimated to be < 10 nM. Calcium concentration was calculated using a software program based on the algorithm of Goldstein (1979; provided by J. Kleinschmidt). The standard intracellular solution (140 K⁺) contained (mM): KCl, 140; Hepes acid, 20; MgCl₂, 1; CaCl₂, 0.4; EGTA, 1; and 0.1 GTP. Low Cl⁻ intracellular solution contained (mM): potassium aspartate, 100; KCl, 10; Hepes acid, 20; MgCl₂, 1; CaCl₂, 0.4; EGTA, 1; ATP, 1; and GTP, 0.1.

Stock solutions of adenosine, ATP, UTP, 2MeSATP, ADP and α,β-methylATP were all diluted in the extracellular recording solution and applied via pressure injection from a micropipette. The micropipette, ≥ 2 mM in diameter, was positioned 50–100 μM from the cell and 14–34 kPa (2–5 lb in²) pressure was applied to the back of the micropipette using a Picospritzer II (General Valve Corp., Fairfield, NJ, USA). Purinergic antagonists (suramin and PPADS), potassium channel blockers (Ba²⁺ and iberiotoxin (IbTX)) and the chloride channel blocker (DIDS), were all included in the extracellular solution and bath perfused for a minimum of five, and usually for ten, complete (1 ml) bath exchanges. All drugs were used at the concentration cited in Results.

Electrophysiological recording techniques

We used the whole-cell patch clamp technique to measure currents in isolated RPE cells (Hamill et al. 1981). Patch electrodes were pulled from borosilicate glass with diameters of 1.5 mm outside and 1.1 mm inside (Sutter Instruments, Novato, CA, USA) using a two-stage vertical microelectrode puller (Narishige model PP83, Tokyo, Japan). Electrodes were coated with beeswax in order to reduce input capacitance and had resistances of 2–3 MΩ when filled with intracellular solution and placed in the extracellular bathing solution. The reference electrode used was a sealed electrode/salt bridge combination (Dri-Ref-2; World Precision Instruments, Sarasota, FL, USA). Offset potentials were nulled using the amplifier circuitry before seals were made on cells. All the data and current-voltage relationships shown have been corrected for liquid junction potentials (LJPs) arising between the bath and the electrode. LJPs were measured experimentally and were also calculated using a software program (JPCalc, version 2.00; P. H. Barry, Sydney, Australia). Liquid junction potentials were defined as the potential of the bath solution with respect to the pipette solution (Barry & Lynch, 1991). For whole-cell recording, the membrane potential of the cell, V_m, was calculated as V_m = V_p− LJP, (where V_p is pipette potential). All data and current-voltage relationships shown have been corrected for LJPs, which were 2 mV for standard internal and standard external Ringer solutions and 9 mV for low chloride/potassium aspartate internal solutions and standard extracellular solutions or cation-substituted extracellular solutions.

Membrane potential and ionic currents were recorded with an Axopatch 1D amplifier (Axon Instruments, Foster City, CA, USA). Voltage step commands were generated using Clampex version 5.5.1 (pCLAMP software, Axon Instruments). Currents were filtered with a 4-pole low-pass Bessel filter (−3 dB at 1 kHz) and digitized at a sampling frequency of 5 kHz using pCLAMP software (Axon Instruments). Current and voltage were displayed on a Gould TA240 chart recorder and were stored on computer disk. Values for cell capacitance were obtained from the capacitance compensation circuitry on the amplifier and averaged 34 ± 2 pF (n = 100). Measures of series resistance were obtained directly from the amplifier and were always less than 15 MΩ. Capacitance subtraction and series resistance compensation (80 %) were used in all recordings. The mean resting membrane potential of cultured RPE cells was −47 ± 2 mV (mean ±s.e.m.n = 33). These values for mean resting membrane potential and cell capacitance are comparable with those reported in other studies for isolated and cultured mammalian RPE cells (Hughes & Steinberg, 1990; Straußet al. 1994; Tao et al. 1994). All experiments were conducted at room temperature (20–24°C).

Calcium imaging

RPE cells were isolated as previously described and were cultured on glass coverslips for 2–5 days before loading with the membrane-permeant (acetomethoxy ester) form of the Ca²⁺ indicator fluo-3 (fluo-3 AM) or fluo-4 (fluo-4 AM; Molecular Probes, Eugene, OR, USA). When bound to Ca²⁺ the fluorescence emitted by these dyes increases (Kao et al. 1989). We chose to use fluo-3/fluo-4 as the calcium indicator dye(s) because of their large optical signals, which allow for a very good signal-to-noise ratio. Stock solutions of dye were prepared by dissolving 50 μg of the dye in 50 ml of DMSO. The dye stock solution was then diluted in standard extracellular 5 K⁺ solution containing 0.01 % pluronic f-127 to give a final concentration of 10 μM fluo-3 AM/fluo-4 AM. To load RPE cells with fluo-3/fluo-4, cells were incubated with the dye solution for 1 h at 37°C, and then rinsed with 5 K⁺ solution.

Retinal pigmental epithelium cells, attached to glass coverslips, were placed in a shallow bath chamber (∼1–2 ml) on the stage of Nikon UM-2 fluorescence microscope. The cells were continuously superfused with standard extracellular or Ca²⁺-free extracellular solution at a rate of 1–2 ml min⁻¹ (see Methods: ‘Solutions and drugs’). ATP, UTP or GTP were dissolved directly into the extracellular solutions and bath superfused at the concentrations cited in Results. Thapsigargin (5 mM stock in DMSO) was diluted into low Ca²⁺ extracellular solution to a final concentration of 5 μM and then bath superfused. All experiments were conducted at room temperature (20–24°C).

Isolated RPE cells were viewed by a Zeiss × 40 water immersion objective on a Nikon UM-2 fluorescence microscope. RPE cells loaded with fluo-3/fluo-4 were imaged using a cooled Photometrics CH250 CCD camera (Photometrics, Tuscon, AZ, USA) with a 200 ms exposure time. Stimulation for the excitation and fluorescence was provided by a 100 W mercury-vapour lamp filtered by a −1.6log neutral density filter, to minimize damage of the RPE cells with excessive light energy, and an appropriate filter set (Nikon B-2A, excitation wavelength of 450–490 nm; emission 520–560; dichroic 510 nm). An electronic shutter (Uniblitz, Rochester, NY, USA) was used to limit light exposure and images (100 pixels²) were recorded and saved every 5.5 s by the IPLab Spectrum program (Signal Analytics Co., Fairfax, VA, USA).

The fluorescence intensity over the RPE cell body was measured before, during and after drug application. A baseline intensity level was established in standard extracellular solution before addition of any drugs. Changes in fluorescence intensity were then quantified by determining the increase in intensity (F) during drug application and then normalizing this increase by dividing by the mean baseline intensity calculated for 10 images before exposure of the cell to drug (F_rest). These relative changes in fluorescence intensity were expressed as:

The ΔF/F values obtained for RPE cells in response to drug application are intended only as qualitative indicators of changes in [Ca²⁺]_i, as these values are not linearly related to absolute values of free [Ca²⁺]_i. Nonetheless, from the reported K_d for fluo-3 (390 nM; Haugland, 1996), the changes of fluorescence observed are likely to be within physiologically relevant ranges. Using IPLab software, the ΔF/F values obtained for RPE cells in response to drug application were converted into pseudocolour image representations.

Statistical analysis

Data are presented as means ±s.e.m. and analysed by use of Student's unpaired t test unless otherwise noted. A significance level of P≤ 0.05 was accepted.

Chemicals

α,β-MethylATP, 2-methylthioATP (2MeSATP), suramin, pyridoxalphosphate-6-azophenyl-2′,4′-disulphonic acid (PPADS), thapsigargin and pluronic F-127 were obtained from Calbiochem (La Jolla, CA, USA). Fluo-3 AM and fluo-4 AM were obtained from Molecular Probes. Iberotoxin (IbTX) was from Peninsula Laboratories Inc. (Belmont, CA, USA). All other chemicals were from Sigma.

RESULTS

Effects of extracellular ATP on whole-cell current

Recent evidence indicates the presence of P2 receptors on mammalian RPE cells (Stalmens & Himpens, 1997; Peterson et al. 1997). We used ATP to examine the effect of P2 receptor activation on whole-cell currents in cultured rat RPE cells in standard 140 K⁺ intracellular and standard 5 K⁺ extracellular recording solution. We initially chose ATP as our agonist as this nucleotide has been demonstrated to be an effective agonist at all P2X and P2Y subtypes described, with the exception of the P2Y₆ subtype (Barnard et al. 1997). Figure 1 shows the effects of ATP on whole-cell currents in rat RPE cells. Figure 1A shows the response of a representative RPE cell to three sucessive 4 s puffer applications of 100 μM ATP. Each application of ATP was separated by a 4 min recovery period. When the cell was held at −62 mV or −22 mV, 100 μM ATP evoked an inward current that activated rapidly and showed some decay towards the end of the ATP application. In this cell, at positive potentials (+18 mV) ATP had little effect on whole-cell current. In 57/ 92 rat RPE cells tested, ATP activated only an inward current. The mean peak amplitude of the ATP-evoked inward current measured at −62 mV was −12.5 ± 1 pA pF⁻¹ (n = 57). The time constant (t) for deactivation of the cation current, measured from the peak of the inward current back to baseline current measured at the holding potential, varied widely from 4.6 to 14 s (8.3 ± 0.7 s). Figure 1B shows the current-voltage plot for a representative RPE cell before and after puffer application of 100 μM ATP. The I–V plots were obtained from voltage-ramp commands applied between −122 and +48 mV. Very little whole-cell current was present in this cell before drug application. The I–V curve shows a small inwardly rectifying current that reverses around −52 mV. Subsequently, application of ATP to this cell evoked a large inward current at all potentials negative to 0 mV. The ATP-activated current, shown in Fig. 1B, was obtained by subtracting currents recorded under control conditions from currents recorded during ATP application. The ATP-activated current was inwardly rectifying and reversed direction close to 0 mV, suggesting the activation of a non-selective cation or an anion conductance.

Figure 1. ATP activates two conductances in rat RPE cells.

ATP activates an inward cation current. A, shows the current activated by a 4 s puff application of 100 μM ATP in a representative RPE cell held at −62, −22 and +18 mV. Fast-activating inward current is apparent at −62 and −22 mV, with little current activated at +18 mV. Cell capacitance was 35 pF. B shows I–V curves for another RPE cell constructed from two sequential 2 s voltage ramps between −122 and +38 mV in the absence or presence of 100 μM ATP. The activation of an inwardly rectifying current is apparent between −122 and 0 mV. C and D, ATP activates a biphasic conductance change. C shows the current activated by a 4 s puff application of 100 μM ATP at −62, +22 and +38 mV in a representative RPE cell. Cell capacitance was 22 pF. Inward current is apparent at −62 mV and outward current is evident at depolarized potentials. D, the I–V curves for three sequential 2 s voltage ramps obtained from a representative RPE cell with a biphasic response to ATP is shown. In the presence of ATP, activation of fast inward current (ATP 1) is followed by activation of a large outwardly rectifying current (ATP 2).

In 35/92 RPE cells tested we also noted that ATP induced a biphasic conductance change. Figure 1C shows a representative RPE cell that had a biphasic response to 100 μM ATP, suggesting activation of two conductances. At a holding potential of −62 mV the fast inward current described above was apparent. However, in contrast to Fig. 1A, at positive potentials (+18 and +38 mV) the activation of a large outward current is apparent. The I–V relationships shown in Fig. 1D were obtained from a representative RPE cell that had a biphasic response to ATP. Three sequential 2 s voltage ramps between −122 and +48 mV are shown. Control current, in the absence of drug, had both a small inward current and an outward current that activated around −40 mV. As above, puffer application of ATP initially activated the fast inward current which shifted the reversal potential of the whole-cell current 25 mV positive (ATP 1). The peak of the outward current (ATP 2) was reached within 8 s of ATP application at which time the I–V relationship reveals an outwardly rectifying current that reverses at −56 mV. Subsequently, between 12 and 24 s after ATP application, inward current decreased at potentials between −122 and −62 mV, but outward current remained increased at more positive potentials. When the inward current had returned to control values the outward current was still increased over control. In most RPE cells showing this biphasic response, there was prominent delay between activation of the inward current and peak of the outward current. The slow shift in the reversal potential from −15 to −65 mV, towards the calculated reversal potential for K⁺ (−84 mV) under our experimental conditions, suggested that, in addition to a cation current, ATP was also activating a K⁺-selective current in rat RPE cells.

The ATP-activated inward cation conductance

We characterized the ionic selectivity of the ATP-activated voltage-dependent inward current using low chloride/potassium aspartate intracellular solution and standard 5 K⁺ extracellular solution, designed to separate Cl⁻ currents (reversing around −40 mV) from cation currents (reversing at 0 mV) and decrease the Cl⁻ current contribution. Under these recording conditions the ATP-activated current reversed closer to the potential expected for a cation-selective channel (−1.5 ± 1 mV; n = 27), confirming that the current activated by ATP was selective for cations and was not carried by chloride ions. To verify this further, we also tested the sensitivity of the ATP-activated inward current to the disulphonic stilbene DIDS, which has been demonstrated to block Cl⁻ channels in a variety of cell types, including the RPE (Hughes & Segawa, 1993; Staußet al. 1998). Neither the reversal potential nor the amplitude of the ATP-activated inward current measured in four RPE cells at −69 mV (−225 ± 92 pA) was significantly affected after 5 min superfusion of the cells with 500 μM DIDS (−170 ± 57 pA). These results rule out the involvement of Cl⁻ current and identify the conductance activated by ATP as a cation current.

We investigated the cation selectivity of the ATP-inward current by substituting extracellular Na⁺ with other monovalent cations. Figure 2A shows the current-voltage plot for a representative RPE cell which displayed ATP-activated inward current but no outward current. The ATP-induced inward current was recorded in standard recording solution, and in a solution in which extracellular Na⁺ was replaced with NMDG. Substitution of NMDG for Na⁺ shifted the reversal potential of the ATP-activated current from −5 to −45 mV and reduced the current amplitude measured at −62 mV from −15 to −4 pA pF⁻¹. In cells with a biphasic response to ATP neither the delayed outward current amplitude, nor the reversal potential were affected by cation substitution (data not shown). Figure 2B shows that the sequential substitution of extracellular Na⁺ by choline or NMDG shifted the mean (±s.e.m.) reversal potential of the ATP-activated cation current in three rat RPE cells from −4 ± 7 mV (NaCl) to −21 ± 8 mV (choline choride) to −33 ± 8 mV (NMDG-Cl). The mean (±s.e.m.) current amplitude for the ATP-activated inward current measured in 5–7 cells was also significantly reduced when standard Na⁺ solution (−18 ± 2 pA pF⁻¹) was replaced with choline (−9 ± 2 pA pF⁻¹) or NMDG-substituted (−5 ± 0.6 pA pF⁻¹) solutions (Fig. 2C).

Figure 2. Inward current activated by ATP is a non-selective cation current.

A, representative current-voltage curves for the ATP-activated inward current in a representative rat RPE cell showing the effect of extracellular Na⁺ replacement. When Na⁺ was substituted for by NMDG, current magnitude was significantly reduced and current reversal potential shifted from −5 to −40 mV. B, histogram shows means ±s.e.m. reversal potential of the ATP-activated cation current measured in three cells in regular Na⁺ (□), choline-substituted (▪) or NMDG-substituted () extracellular solution. C, histogram shows means ±s.e.m. amplitude of the ATP-activated cation current measured at −69 mV in 5–7 cells in regular Na⁺ (□), choline-substituted (▪) or NMDG-substituted () extracellular solution. All data have been normalized for cell capacitance. * Significance at P < 0.05 in this and subsequent figures.

The cation permeability, anion impermeability and positive reversal potential of this ATP-activated inward current in rat RPE cells identifies it as a non-selective cation (NSC) current. The electrophysiological characteristics of this NSC current are similar to those described for cloned and native P2X purinoceptors in a variety of cell types (Humphrey et al. 1995; Soto et al. 1996). Non-selective cation channels gated by P2X receptors are not only permeant to Na⁺ and K⁺ but also have varying degress of permeability to larger divalent cations such as Ca²⁺ (Humphrey et al. 1995; Soto et al. 1996). We therefore next examined the effects of removal of extracellular Ca²⁺ on the ATP-induced cation current by activating the current in nominally Ca²⁺-free external solution ([Ca²⁺]_o < 10 nM). Removal of extracellular Ca²⁺ had no significant effect on that current activation or amplitude measured at −69 mV in three out of four cells tested (−9 ± 1 vs. −7 ± 0.8 pA pF⁻¹; P > 0.05). These results suggest that extracellular Ca²⁺ is not required for the activation of this non-selective cation channel.

In attempts to identify the subtype of P2X purinoceptor involved in the electrophysiological responses to ATP in rat RPE cells, the sensitivity of the ATP-activated inward cation conductance to pharmacological blockade by PPADS and suramin was investigated. PPADS and suramin have been purported to act as selective antagonists at P2X receptors, and have been used as pharmacological tools to discriminate between P2X and P2Y receptors (Surprenant et al. 1995; Buell et al. 1996). More recent evidence, however, indicates that these antagonists lack selectivity for P2X receptors and are weak or ineffective blockers of certain P2X receptor subtypes (Collo et al. 1996; Soto et al. 1996; Barnard et al. 1997). Figure 3A shows the mean (±s.e.m.) data for the effect of 100 μM suramin or PPADS on the inward current activated by 100 μM ATP. Peak current amplitude for the ATP-induced inward current was measured at −69 mV in the same three cells in the absence and presence of antagonist. Cells were allowed a 10 min recovery period between successive ATP applications. In our experiments a minimum of 4–5 min has been previously shown to allow for complete recovery from receptor desensitization (J. S. Ryan & M. E. M. Kelly, unpublished observation). The mean amplitude of the cation current activated by ATP in the absence of either antagonist (−9 ± 1 and −17 ± 9 pA pF⁻¹) was not significantly affected by either suramin (−13 ± 7 pA pF⁻¹) or PPADS (−15 ± 10 pA pF⁻¹), respectively.

Figure 3. ATP-activated NSC current is insensitive to P2 antagonists.

A, histogram shows the means ±s.e.m. amplitude of the ATP-activated inward current measured at −69 mV in the same three cells before (□) and after (▪) 15 min bath superfusion with 100 μM of the P2 antagonist suramin (n = 3) or PPADS (n = 3). Neither P2 antagonist had any effect on the amplitude of the ATP-activated NSC current. Data have been normalized for cell capacitance. B, histogram shows the means ±s.e.m. amplitude of inward NSC current measured at −69 mV activated by ATP (□; n = 57), UTP (; n = 7), 2MeSATP (▪; n = 5) or ADP (; n = 6). Agonists were applied by pressure application for durations of 4–10 s. All data have been normalized for cell capacitance.

We chose ATP as our initial agonist to explore its possible effects on whole-cell currents in rat RPE as this nucleotide has been demonstrated to be an effective agonist at all P2X and P2Y subtypes described with the exception of the P2Y₆ subtype (Barnard et al. 1997). To characterize further the subtypes of purinergic receptors involved in the activation of the cation current in RPE cells, we tested the effect of different purinoceptor agonists. The potency of these agonists at purinoceptors can be used as a means to discriminate between different P2 purinoceptor families (Barnard et al. 1997). Figure 3B summarizes results of experiments in which RPE cells were exposed to 100 μM concentrations of ATP, UTP, 2MeSATP, ADP or α,β-methylATP. All agonists, with the exception of α,β-methylATP, were equally effective in activating the inward cation current. Inward current was never observed in rat RPE cells exposed to α,β-methylATP (n = 12). The mean current amplitudes of the inward cation currents measured at −69 mV evoked by the purinergic agonists: UTP (−12 ± 6 pA pF⁻¹; n = 7), ADP (−10 ± 2.4; n = 6) or 2MeSATP (−15 ± 3 pA pF⁻¹; n = 5) were not significantly different from the inward cation current activated by ATP (−12.5 ± 1 pA pF⁻¹; n = 57). Thus, all P2 agonists tested in this study were equal in their ability to activate the P2X receptor subtype gating the non-selective cation current.

The ATP-activated K⁺ outward current

To confirm that the delayed outward current activated by ATP was K⁺ selective, we tested its susceptibility to blockade by a number of known K⁺-channel blockers. Figure 4A shows I–V curves for the ATP-activated currents measured in a representative RPE cell that exhibited a biphasic response to ATP. Puffer application of 100 μM ATP initially only elicited the fast inward non-selective cation (NSC) current that reverses close to 0 mV. Subsequently, 10 s after ATP application, the fast inward current decayed and a delayed outwardly rectifying current activated. Activation of this current shifted the reversal potential for the whole-cell current to −66 mV, towards the reversal potential for K⁺ under our recording conditions (E_K = −84 mV). Figure 4B shows the I–V curve for the same RPE cell 5 min after superfusion with 5 mM Ba²⁺. This concentration of external Ba²⁺ has been shown previously to block K⁺-selective outward currents in cultured rat RPE cells (Poyer et al. 1996). Puffer application of 100 μM ATP, as before, activated the fast inward cation current in the presence of Ba²⁺ (E_rev = 0 mV). However, the second I–V plot recorded 10 s after ATP application had very little whole-cell current, indicating that Ba²⁺ had blocked the delayed activation of the ATP-activated outward current. The amplitude of the ATP-activated outward current under control conditions (703 ± 200 pA) was significantly reduced in the presence of Ba²⁺ (50 ± 14 pA) (n = 6; P≤ 0.05). Similarly, the mean (±s.e.m.) reversal potential of the ATP-activated K⁺ current (−68 ± 3 mV) in the same six RPE cells was shifted significantly more positive with 5 mM Ba²⁺ (−9 ± 8 mV; P≤ 0.05). The negative reversal potential of the ATP-activated outward current (close to the calculated K⁺ equilibrium potential; E_K = −84 mV) and the sensitivity of the outward current to Ba²⁺ ions, confirms that ATP is activating a K⁺-selective current in addition to a non-selective cation current in rat RPE cells.

Figure 4. ATP activates a Ca²⁺-activated K⁺ current in rat RPE cell.

A, current-voltage (I–V) plots for a representative RPE cell that has a biphasic response to puffer application of 100 μM ATP. ATP initially activates a NSC current that reverses around 0 mV (•). Ten seconds after ATP application the inward cation current is inactivated and an outwardly rectifying current is activated that reverses at −70 mV (▵). B, the I–V plot shows ATP-activated current measured in the same RPE cell after a 10 min bath superfusion of 5 mM Ba²⁺. ATP still activates the initial fast inward cation current (•), but the activation of the outward current is abolished (▵). C, histogram shows means ±s.e.m. percentage increase in outward current measured at +31 mV following ATP application in the presence or absence of 10 nM IbTX. ATP (100 μM; □) significantly increased the outward current. In the presence of 10 nM IbTX (▪) the ATP-induced increase in outward current is reduced. D, histogram shows means ±s.e.m. amplitude of the outward K⁺ current (I_K(Ca)) activated by ATP (n = 23), UTP (n = 6), 2MeSATP (n = 3) or ADP (n = 6). Nucleotide-activated I_K(Ca) was measured at + 31 mV. All agonists were applied by pressure application for a duration of 4–10 s. All data have been normalized for cell capacitance.

At least two major types of outward K⁺-selective currents, a delayed rectifying K⁺ current (I_K(V)) and a calcium-activated K⁺ current (I_K(Ca)), have been described in mammalian RPE cells (Straußet al. 1994; Tao & Kelly, 1996). In a variety of cell types, including the RPE (Sullivan et al. 1997; Peterson et al. 1997), activation of metabotropic P2Y receptors have been shown to be coupled to elevations in [Ca²⁺]_i. We wanted to test the hypothesis that the delayed K⁺ current activated by ATP was a Ca²⁺-activated K⁺ current (I_K(Ca)) secondary to ATP-evoked elevations in [Ca²⁺]_i. To do this we looked at the sensitivity of the ATP-activated outward current to the selective ‘maxi’ K_Ca channel blocker iberiotoxin (Giangiacomo et al. 1992). Iberiotoxin (IbTX) has been shown to block potently agonist-stimulated activation of ‘maxi’ K_Ca channels and I_K(Ca) in ocular epithelial cells (Tao & Kelly, 1996; Ryan et al. 1998). Figure 4C shows the mean increase in current amplitude of the ATP-activated outward K⁺ current measured at +31 mV in RPE cells in the presence or absence of 10 nM IbTX. Puffer application of 100 μM ATP in the absence of IbTX increased outward K⁺ current (18 ± 4 pA pF⁻¹) in four RPE cells tested. Superfusion of the same four cells with 10 nM IbTX for 5 min completely blocked I_K activation in two of the cells tested and in two other cells reduced the mean increase in outward current to 2 ± 1 pA pF⁻¹ (P≤ 0.05). The ability of IbTX to block ATP-activated outward current suggests that purinoceptors can activate a Ca²⁺-activated K⁺ current (I_K(Ca)) in rat RPE cells.

Figure 4D shows the effectiveness of ATP and its analogues to activate I_K(Ca) in rat RPE cells. Similar to the findings with the inward cation current, α,β-methylATP was ineffective in all cells (n = 12) tested, but outward K⁺ current was activated in RPE cells treated with ATP, UTP, 2MeSATP, or ADP. The mean amplitude of the nucleotide-activated outward current measured at +31 mV by UTP (12 ± 3 pA pF⁻¹; n = 6), ATP (23 ± 6 pA pF⁻¹; n = 28), 2MeSATP (25 ± 6 pA pF⁻¹; n = 3) or ADP (27 ± 8 pA pF⁻¹; n = 6) showed no significant differences.

Effects of extracellular adenosine on whole-cell current

The presence of functional adenosine receptors on mammalian RPE cells (Gregory et al. 1994), suggested that these purinoceptors may play a role in modulating ionic currents in these cells. We examined the effect of adenosine on whole-cell currents in isolated rat RPE cells using standard intracellular and extracellular solutions. Figure 5A shows whole-cell currents recorded from a representative RPE cell before and after puffer application of 100 μM adenosine. In this, and eight other cells tested, the mean amplitude of the control current measured at +58 mV (106 ± 25 pA pF⁻¹) and − 62 mV (−115 ± 12 pA pF⁻¹) was unaffected by 100 μM adenosine (84 ± 21 pA pF⁻¹ at +58 mV and −84 ± 12 pA pF⁻¹ at −62 mV) (Fig. 5B). The results show that activation of adenosine receptors in cultured rat RPE cells does not result in alterations in whole-cell current.

Figure 5. Effect of adenosine on whole-cell current in rat RPE.

A, whole-cell current recorded from a representative RPE cell in the absence (left panel) or presence (right panel) of 100 μM adenosine. Currents were recorded in standard 140 K⁺ intracellular and standard 5 K⁺ extracellular solutions. Cell capacitance was 35 pF. B, means ±s.e.m. current amplitude (pA) measured at +58 and −62 mV in the absence (□) and presence (▪) of 100 μM adenosine (ADO). Adenosine had no effect on whole-cell current in isolated rat RPE cells (n = 9).

Intracellular calcium responses

The activation of an IbTX-sensitive I_K(Ca) by ATP suggested the involvement of G protein-coupled P2Y purinoceptors in modulating ion channels in the rat RPE. P2Y purinoceptors typically couple to signalling pathways linked to the activation of a phospholipase C/inositol 1,4,5-trisphosphate (PLC/IP₃) pathway and subsequent elevations in cytosolic Ca²⁺ (Salter & Hicks, 1995; Heilbronn et al. 1997). We used the calcium indicator dye fluo-3 or fluo-4 as a qualitative measure of alterations in [Ca²⁺]_i following ATP or UTP agonist stimulation (Kao et al. 1989). The top of Fig. 6 shows a bright field (A) and pseudocolour image (B) of the [Ca²⁺]_i response in a representative RPE cells in response to bath superfusion of 100 μM ATP. Figure 6C shows the time course for the ATP-mediated increase in fluo-3 fluorescence intensity for the same cell as in panels A and B. ATP increased [Ca²⁺]_i within 60 s of bath application. The rapid peak increase in [Ca²⁺]_i was followed by a phase of slow decrease in which [Ca²⁺]_i remained elevated above baseline for ∼3 min. The ΔF/F for this cell was 0.27. Bath superfusion of 100 μM UTP also caused large transient increases in [Ca²⁺]_i in rat RPE cells. Figure 6D shows the time course for the increase in [Ca²⁺]_i in a representative rat RPE cell in response to 100 μM UTP. The ΔF/F for this cell was 0.15. The mean response of three RPE cells to 100 μM UTP (ΔF/F = 0.09 ± 0.02) was not significantly different from that obtained for ATP (ΔF/F = 0.1 ± 0.03; n = 7) suggesting that both agonists were equipotent in mediating elevations in [Ca²⁺]_i. The mean fluorescence baseline signal or the stimulated increase in [Ca²⁺]_i was not significantly altered by repeated applications of either ATP or UTP. This indicates that the calcium response was not desensitizing and that the cells were not significantly affected by photobleaching.

Figure 6. ATP/UTP-evoked increases in [Ca²⁺]_i.

A, bright-field image of a representative RPE. B, pseudocolour image of the same RPE cell illustrating the increase in [Ca²⁺]_i in response to bath application of 100 μM ATP. C, time course of the change in [Ca²⁺]_i in response to 100 μM ATP for the same cell. D, time course of the change in [Ca²⁺]_i in response to 100 μM UTP in a representative rat RPE cell.

We also examined the effect of another purine nucleotide, guanosine triphosphate (GTP) and the P1 agonist adenosine on [Ca²⁺]_i rat RPE cells. GTP has little, if any, effect on P2X or P2Y purinoceptor subtypes. In three RPE cells, application of 100 μM ATP resulted in a mean increase ΔF/F of 0.1. Bath application of 100 μM GTP in the same three cells failed to produce any detectable change in fluorescence. A second application of ATP, however, produced a response that was undiminished from the initial change in [Ca²⁺]_i (mean ΔF/F = 0.13). The lack of effect of GTP suggests that the Ca²⁺ response is selective for adenine nucleotides. Similarly, adenosine treatment in four cells did not significantly alter baseline ΔF/F, consistent with the lack of effect of this agonist on whole-cell current (P > 0.05).

To identify the source of the Ca²⁺ mobilized in response to ATP, we measured the change in fluorescence in nominally Ca²⁺-free extracellular solution. Figure 7A shows the response of a representative RPE cell to stimulation with 100 μM ATP in regular external Ca²⁺ (2 mM), and in nominally Ca²⁺-free external solution with 1.5 mM EGTA added ([Ca²⁺]_o < 10 nM). The [Ca²⁺]_i response was undiminished in nominally Ca²⁺-free solution (ΔF/F = 0.26), compared with the first ATP application in 2 mM [Ca²⁺]_o (ΔF/F = 0.24). The mean ΔF/F value in five rat RPE cells exposed to ATP in nominally Ca²⁺-free solution (0.29 ± 0.07) was not significantly different from ATP application in regular Ca²⁺ solution (0.28 ± 0.04), confirming that under the experimental conditions used extracellular Ca²⁺ influx does not contribute to ATP-induced elevations in [Ca²⁺]_i (Student's paired t test; P > 0.05). We next examined whether ATP-induced increases in intracellular Ca²⁺ was mediated by release of Ca²⁺ from intracellular stores by exposing RPE cells to thapsigargin in nominally Ca²⁺-free Ringer solution. Thapsigargin mobilizes Ca²⁺ from intracellular stores by inhibiting Ca²⁺ sequestration from IP₃-sensitive pools (Thastrup et al. 1990). Figure 7B shows the [Ca²⁺]_i response to 100 μM ATP in a representative RPE cell prior to and after 5 mM thapsigargin. In regular (2 mM) extracellular Ca²⁺, prior to thapsigargin application, the ΔF/F in response to ATP was 0.2. A second application of 100 μM ATP in the same cell following 10 min superfusion with 5 μM thapsigargin failed to produce a significant increase in [Ca²⁺]_i (ΔF/F = 0.01). In 28 other cells tested the mean [Ca²⁺]_i response to ATP in the presence of thapsigargin was essentially abolished (0.01 ± 0.002) compared with control (0.48 ± 0.03). The ability of ATP to evoke a [Ca²⁺]_i response in the absence of extracellular Ca²⁺ and the ability of thapsigargin to inhibit ATP-induced [Ca²⁺]_i response, indicates that intracellular Ca²⁺ stores are involved in mediating changes in cytosolic- free Ca²⁺. These results provide further evidence for the presence of metabotropic P2Y receptors coupled to intracellular Ca²⁺ release and subsequent K⁺ channel modulation in rat RPE cells. The demonstration that ATP and UTP were equipotent in mediating increases in [Ca²⁺]_i is most consistent with the pharmacological profile of cloned and native P2Y₃ or P2Y₄ purinoceptors (Barnard et al. 1997).

Figure 7. ATP-evoked increase in [Ca²⁺]_i is dependent on intracellular Ca²⁺-stores release.

A, optical recording from a representative RPE cell illustrating the effect of Ca²⁺-free extracellular solution on the ATP-evoked increase in [Ca²⁺]_i. Removal of extracellular Ca²⁺ failed to abolish the ATP effect. Removal of extracellular Ca²⁺ failed to significantly affect the amplitude of the ATP-evoked increase in [Ca²⁺]_i. B, optical recording from a representative RPE cell illustrating the effects of 5 μM thapsigargin on ATP-evoked increase in [Ca²⁺]_i.

DISCUSSION

This study demonstrates that extracellular ATP activates two separate cation conductances in the rat RPE. Actions of adenine nucleotides on cation currents have not previously been described in rat RPE cells. In the majority of cells examined, ATP activated an inwardly rectifying non-selective cation conductance providing evidence for the existence of P2X purinoceptors in the RPE. In cells with a biphasic response to ATP, the activation of the inward current was followed by the activation of a calcium-activated K⁺ current (I_K(Ca)). The ability of the Ca-ATPase inhibitor thapsigargin essentially to abolish ATP-evoked elevations in [Ca²⁺]_i suggests that P2Y purinoceptor-mediated intracellular Ca²⁺ release may be involved in I_K(Ca) activation.

The initial transient inward current activated by ATP was identified as a non-selective cation current based on several observations. The ATP-activated inward current reversed close to 0 mV in standard 140 Cl⁻ solutions and in solutions at which E_Cl was set at about −40 mV suggesting that the current was selective for cations. Non-selectivity of the current was confirmed by experiments demonstrating that large cations, such as choline and NMDG, were also able to permeate the ATP-activated channel but with a reduced permeability compared with Na⁺. Removal of extracellular Ca²⁺ had no significant effect on the activation of the ATP-evoked current, suggesting that activation of this NSC current under physiological conditions is not dependent on extracellular Ca²⁺. The I–V plot for the ATP-activated non-selective cation current showed strong inward rectification, a common feature of ligand-gated P2X channels in virtually all cell types studied (Humphrey et al. 1995). Thus, the electrophysiological characteristics of the ATP-activated ion channel in rat RPE cells are consistent with the presence of P2X purinoceptors.

The identification of P2X purinoceptors is hindered by the lack of subtype-specific blockers. Suramin and PPADS were once considered to be selective antagonists for the P2X receptor family. These antagonists, however, are no longer considered selective for a given P2 purinoceptor family or subtype. They are weak or ineffective at some P2X subtypes and antagonize responses to virtually all P2Y subtypes as well nicotinic, GABA- and glutamate-gated ion channels (Humphrey et al. 1995). Of the seven P2X subtypes thus far identified, P2X₁, P2X₂, P2X₃ and P2X₅ are blocked by suramin and PPADS at the concentration used in this study while P2X₄ and P2X₆ have been found to be resistant, or partially resistant, to antagonist blockade by suramin and PPADS (Soto et al. 1996; Buell et al. 1996; Collo et al. 1996). The antagonist insensitivity of our ATP-evoked response may suggest the involvement of a P2X₄ and/or P2X₆ purinoceptor subtype in rat RPE cells.

The agonist profile obtained for the ATP-activated inward cation current in rat RPE cells (ATP = UTP = 2MeSATP = ADP), however, correlates poorly with the potency profiles reported for cloned P2X receptors. Our findings that the NSC current was not activated by α,β-methyl ATP suggests the involvement of an α,β-methyl ATP-insensitive P2X subtype (P2X₂, P2X₄, P2X₅ or P2X₆) (Barnard et al. 1997). The demonstration that UTP was equally as effective as ATP, suggests that this uridine nucleotide acts as a full agonist at the P2X receptor in rat RPE cells. Very few reports have documented effects of UTP at P2X purinoceptor suptypes. Recently, however, in dorsal root ganglion neurons, it has been demonstrated that UTP mediates transient inward currents via stimulation of P2X₃ purinoceptors (Rae et al. 1998). Similarly, UTP has been reported to activate an inward cation current gated by P2X₁ receptors in rat vascular smooth muscle cells (McLaren et al. 1998). These findings support our observation that UTP gates a NSC current via P2X receptor activation in rat RPE cells.

The varied pharmacological profile obtained for the ATP-activated cation current in rat RPE cells makes it plausible that our results are indicative of a combination of two or more homomeric purinoceptors, giving a novel pharmacological profile. A number of P2X receptors have been shown to associate in multimers resulting in pharmacological and kinetic profiles differing from homoligomeric channels. For example, in sensory neurons it has been proposed that the response to ATP is mediated via a P2X₂-P2X₃ heteromultimer (Lewis et al. 1995). Alternatively, the pharmacological profile obtained in the rat RPE may also be indicative of P2X receptor heterogeneity. In many cell types P2X receptor expression is not limited to only one subtype (Humphrey et al. 1995; Nori et al. 1998). The inward cation current in rat RPE cells could therefore be gated by multiple P2X subtypes, each with subtle differences in their sensitivity to purinoceptor agonists.

In a subpopulation of RPE cells, ATP activated an outward current with a slightly delayed onset compared with the NSC channel. The I–V relationship indicated that the delayed current activated at negative potentials, was outwardly rectifying at depolarized potentials and was blocked by Ba²⁺, indicating the activation of a K⁺ current. ATP-evoked activation of this current was sensitive to the selective ‘maxi’ K_Ca channel blocker IbTX, identifying it as a Ca²⁺-activated potassium current (I_K(Ca)). Maxi-K_{Ca or} BK_Ca channels are large conductance K⁺-selective channels that have been described in a variety of cell types (reviewed by McManus, 1991). Despite their ubiquitous nature, BK_Ca channels share common characteristics including activation by both Ca²⁺and membrane depolarization and sensitivity to IbTX. Intracellular Ca²⁺ ions activate these channels by interacting directly with the cytosolic surface of the channel protein. Calcium-activated potassium channels are targets for modulation by a number of different agonists linked to signalling pathways that stimulate elevations in [Ca²⁺]_i (Marty, 1989; Ryan et al. 1998). An IbTX-sensitive I_K(Ca) in rabbit RPE cells, similar to the one activated by ATP in rat, has been shown to be activated by α₁-adrenergic-induced elevations in intracellular Ca²⁺ (Q.-P. Tao & M. E. M. Kelly, unpublished observations).

The demonstration in this study that ATP and other purine/pyrimidine analogues could activate I_K(Ca) and elevate [Ca²⁺]_i in rat RPE cells suggested that ATP-evoked elevations in [Ca²⁺]_i are involved in mediating K_Ca channel activation. In a number of cell types purinergic agonists have been shown to regulate ion channels that are sensitive to [Ca²⁺]_i. In vascular smooth muscle an ATP-mediated [Ca²⁺]_i increase has been shown to result in activation of both a Ca²⁺-sensitive Cl⁻ and K⁺ channel (Strobaek et al. 1996). Ca²⁺-activated K⁺ conductances have also been reported to be activated by ATP in colonic smooth muscle (Koh et al. 1997), rat osteoclasts (Weidema et al. 1997) and mouse microglia (Waltz et al. 1993). More recently, in monolayers of bovine RPE, ATP-evoked elevations in [Ca²⁺]_i were reported to be involved in mediating changes in basolateral Cl⁻ and apical membrane K⁺ fluxes (Peterson et al. 1997). In all these studies, ATP-mediated activation of Ca²⁺-sensitive ion channels was linked to P2Y purinoceptor activation coupled to intracellular Ca²⁺ stores release. Based on these studies, and our experimental findings, we propose the presence of functional P2Y purinoceptors in rat RPE linked to increases in intracellular [Ca²⁺]_i and I_K(Ca) activation. Our results are consistent with previous functional and molecular studies which reported the existence of G protein-coupled P2Y purinoceptors in the mammalian RPE (Stalmans & Himpens, 1997; Peterson et al. 1997). However, as with P2X purinoceptor-derived responses and NSC channel activation in RPE cells, conclusive identification of the P2Y subtype(s) involved in I_K(Ca) activation awaits further molecular characterization of the purinoceptors in rat RPE cells.

In the RPE, adenosine receptor pharmacology is much better characterized than that for P2 purinoceptors. Messenger RNA for A_2A purinoceptors has been demonstrated in the RPE and pharmacological data support the presence of A_2B receptors that are sensitive to micromolar amounts of adenosine (Blazynski, 1993; Kvanta et al. 1997). A functional role for adenosine A₂ receptors in modulating phagocytosis of rod outer segments has also been proposed in the rat RPE (Gregory et al. 1994). Under the conditions of this study, 100 μM adenosine had no affect on whole-cell current in the rat RPE cells examined. Thus, despite a functional role in phagoctyosis, activation of adenosine receptors does not appear to play a role in modulating membrane currents in RPE cells. Our electrophysiological findings, coupled with the lack of an adenosine effect in rat RPE cells, are consistent with previous studies in which adenosine had no demonstrable affect on intracellular [Ca²⁺]_i responses in human RPE cells (Sullivan et al. 1997), or on fluid or ion transport in monolayers of bovine RPE (Peterson et al. 1997). Our findings that adenosine had no effect on whole-cell currents also supports our suggestion that ecto-ATPase activity played no role in our electrophysiological findings with ATP.

In a variety of other cell types ATP, acting on cell surface P2 purinoceptors, is involved in a plethora of biological processes including neurotransmission, vascular smooth muscle contraction, platelet aggregation, proliferation, secretion, cellular fluid and ion transport and phagocytosis (Mason et al. 1991; Ichinose, 1995; Heilbronn et al. 1997; Boarder & Hourani, 1998). To date, there is little evidence to support a role for ATP as a neurotransmitter in the retina. However, several studies have shown that in response to various stimuli ATP can be released from cells in the neural retina. Light flicker and/or depolarizing conditions has been demonstrated to evoke the co-release of ATP and acetylcholine from retinal amacrine cells (Neal & Cunningham, 1994) and nucleotides have been shown to be released from rabbit retina via both Ca²⁺-independent and Ca²⁺-dependent mechanisms (Blazynski & Perez, 1991). Similarly, a recent study has reported a release mechanism for stored ATP in ciliary epithelial cells (Mitchell et al. 1998). These studies suggest that under certain conditions ATP may be released in significantly high amounts to cross the short diffusion distance between the neural retina and RPE and activate P2 purinoceptors. Significant concentrations of purines may also be present in the choroidal blood supply as a result of platelet activation (ATP) and/or vascular cell lysis (ATP/UTP) in response to inflammation and/or injury. Given the close approximation of the choroid to the RPE, it is plausible that under pathophysiological conditions in which the blood retinal barrier is compromised, ATP/UTP from the choroidal vasculature could permeate in sufficiently high amounts to activate P2 receptors on the RPE. Additionally, ATP/UTP could also be released from damaged retinal cells or RPE cells in response to ocular trauma. Under these conditions, extracellular ATP would act as an autocrine or paracrine factor to alert the surrounding cells to a pathological event.

This current study, and others, demonstrating purinergic modulation of ion channels and transport properties (Peterson et al. 1997) in the RPE suggests that extracellular ATP may play an important functional role in vivo. For example, activation of the P2X-gated NSC current in rat RPE cells would cause sufficient depolarization to activate or inactivate a variety of ion channels, pumps and transporters. The net effect on the epithelium would be an altered ion/fluid transport accompanied by changes in the composition of the subretinal space. P2Y purinoceptor-stimulated Ca²⁺-activated K⁺ currents may play a similar role to the NSC current in modulating RPE ion/fluid transport. In support of this hypothesis, Ca²⁺-activated K⁺ channels have been shown to regulate transepithelial transport in a variety of epithelial cell types (Petersen & Maruyama, 1984; Klaerke, 1997). ATP-activated K⁺ current in the RPE may also play a role in the [K⁺]_o rebound that is observed in the subretinal space following light-induced decrease in [K⁺]_o (Joseph & Miller, 1991). ATP, possibly released from adjacent neural retina cells, may act as a paracrine signal for this rebound by stimulating apical BK_Ca channels, leading to K⁺ efflux and thus buffering the drop in subretinal K⁺ to maintain a homeostatic environment for the photoreceptors. Alternatively, activation of I_K(Ca) could result in hyperpolarization of the transepithelial (TEP = V_b−V_a) potential, where V_b is the basal membrane potential and V_athe apical membrane potential. This would stimulate net ion absorption by increasing the driving force for Cl⁻ secretion through passive ion channels in the basolateral membrane. The TEP is also known to be the driving force for ion movement through the paracellular pathway in the RPE (Edelman & Miller, 1991), and thus, BK_Ca channels may also stimulate net ion/fluid movement via this route.

In conclusion, we have provided evidence for the presence of P2X and P2Y purinoceptors on cultured rat RPE cells that are functionally linked to activation of a NSC current, a calcium-activated K⁺ current and elevations in [Ca²⁺]_i. The presence of different purinoceptor subtypes on RPE cells suggests that multiple signalling pathways may be activated in response to extracelllar ATP or UTP, which serve to modulate RPE functions under physiological and/or pathophysiological conditions.