Effects of KCNQ channel modulators on the M-type potassium current in primate retinal pigment epithelium

Abstract

Recently, we demonstrated the expression of KCNQ1, KCNQ4, and KCNQ5 transcripts in monkey retinal pigment epithelium (RPE) and showed that the M-type current in RPE cells is blocked by the specific KCNQ channel blocker XE991. Using patch-clamp electrophysiology, we investigated the pharmacological sensitivity of the M-type current in isolated monkey RPE cells to elucidate the subunit composition of the channel. Most RPE cells exhibited an M-type current with a voltage for half-maximal activation of approximately −35 mV. The M-type current activation followed a double-exponential time course and was essentially complete within 1 s. The M-type current was inhibited by micromolar concentrations of the nonselective KCNQ channel blockers linopirdine and XE991 but was relatively insensitive to block by 10 μM chromanol 293B or 135 mM tetraethylammonium (TEA), two KCNQ1 channel blockers. The M-type current was activated by 1) 10 μM retigabine, an opener of all KCNQ channels except KCNQ1, 2) 10 μM zinc pyrithione, which augments all KCNQ channels except KCNQ3, and 3) 50 μM N-ethylmaleimide, which activates KCNQ2, KCNQ4, and KCNQ5, but not KCNQ1 or KCNQ3, channels. Application of cAMP, which activates KCNQ1 and KCNQ4 channels, had no significant effect on the M-type current. Finally, diclofenac, which activates KCNQ2/3 and KCNQ4 channels but inhibits KCNQ5 channels, inhibited the M-type current in the majority of RPE cells but activated it in others. The results indicate that the M-type current in monkey RPE is likely mediated by channels encoded by KCNQ4 and KCNQ5 subunits.

the retinal pigment epithelium (RPE) carries out a host of functions that are critical to the integrity of the adjacent photoreceptors, including the phagocytosis of outer segments, the regeneration of photopigment, and the supply of nutrients and removal of wastes (31). In addition, the RPE helps control the volume and ionic composition of fluid in the subretinal space, the extracellular compartment that is bounded by the photoreceptor outer segments and the apical aspects of the RPE and Müller (radial glial) cells (9). Photoreceptor function critically depends on the maintenance of subretinal K⁺ concentration within narrow limits, which is achieved by K⁺ transport mechanisms in the RPE and Müller cells.

K⁺ channels in the RPE apical and basolateral membranes play pivotal roles in K⁺ secretion and absorption and strongly influence anion and fluid absorption, as well as mechanisms governing RPE intracellular Ca²⁺ and pH homeostasis (9). Although various classes of K⁺ current have been described in cultured RPE cells (37), in native RPE cells there appear to be two principal types that are active at physiological voltages: an inwardly rectifying K⁺ current (11) and a sustained, outwardly rectifying K⁺ current that has electrophysiological properties resembling those of M-type currents in excitable cells (12, 33). It is well established that the inwardly rectifying K⁺ current is mediated by Kir7.1 channels located in the RPE apical membrane (40), where they recycle K⁺ that enters the cell via the Na⁺/K⁺ pump and the Na⁺-K⁺-2Cl⁻ cotransporter. The identity and location of the channels that generate the M-type current in the RPE are poorly understood, as is the composition of the basolateral K⁺ conductance that functions in K⁺ absorption.

In other cell types, the M-type current is mediated by channels composed of KCNQ (K_V7) and KCNE β-subunits. In addition to regulating excitability in neurons and cardiac myocytes, KCNQ/KCNE channels also help set the membrane potential in certain epithelia (22). Recently, we demonstrated that the M-type current in monkey RPE is blocked by the KCNQ channel blocker XE991 and that this tissue expresses transcripts for KCNQ1, KCNQ4, and KCNQ5 (41), establishing these subunits as candidates for the M-type channel. We also demonstrated the expression of KCNQ5 protein in Western blots of monkey RPE plasma membrane proteins and localized KCNQ5 to the basolateral membrane by immunohistochemistry. KCNQ1 and KCNQ4 proteins were not detected in the RPE by either approach, but these negative findings could have been due to low protein abundance or poor antibody reactivity. The results indicate that KCNQ5 subunits may contribute to the basolateral membrane K⁺ conductance and suggest a potential role for KCNQ1 and KCNQ4 subunits as well.

In the present study, we investigated the kinetics and pharmacology of the M-type current in freshly isolated monkey RPE cells to define the relationship between the underlying channel(s) and KCNQ1, KCNQ4, and KCNQ5 subunits. We found that the M-type current in isolated monkey RPE cells is blocked by the nonselective KCNQ channel blockers linopirdine and XE991 but is relatively insensitive to the KCNQ1 blockers tetraethylammonium (TEA) and chromanol 293B. In addition, we found that the M-type current is activated by the KCNQ channel openers retigabine, zinc pyrithione (ZnPy), and N-ethylmaleimide (NEM), but not by cAMP. Finally, we found that diclofenac, which activates KCNQ4 channels but inhibits KCNQ5 channels, had variable effects on the RPE M-type current. The results are consistent with the idea that KCNQ5 and, perhaps, KCNQ4 subunits contribute to the channel(s) underlying the M-type current in primate RPE.

METHODS

Cell isolation.

Monkey (Macaca fascicularis or Macaca mulatta) eyes were obtained from the National Primate Centers or the Lonza Group (Walkersville, MD). Animals were euthanized for reasons unrelated to this study. The globes were shipped in HCO₃⁻-free culture medium on ice and received on the following day. RPE cells were isolated essentially as described previously (10). Briefly, pieces of RPE-choroid were incubated in a 0 Na⁺, Ca²⁺- and Mg²⁺-free solution (see below) containing activated papain (25 U/ml; Worthington Biochemical, Lakewood, NJ) for 30 min at 37°C, washed in Ringer solution containing 1% bovine serum albumin, and then gently triturated. Isolated RPE cells were stored in Ringer solution containing 0.5 mM taurine and 1 mM reduced glutathione at 4°C and used within 32 h. There were no obvious differences between the properties of RPE cells recorded on the day of isolation and those of day-old cells, and data were pooled.

Solutions.

The standard bath solution (HEPES-Ringer) consisted of (in mM) 135 NaCl, 5 KCl, 10 HEPES, 10 glucose, 1.8 CaCl₂, and 1 MgCl₂ and was titrated to pH 7.4 with NaOH (290 mosM). In some experiments, NaCl was replaced with TEA-Cl. The standard pipette solution was (in mM) 30 KCl, 83 K-gluconate, 10 K₂HEPES, 5.5 K₂EGTA, 0.5 CaCl₂, and 4 MgCl₂, pH 7.2. In whole cell recording experiments, 4 mM K₂ATP and 0.1 mM Na₂GTP were added to the pipette solution. Retigabine was a gift from Valeant Pharmaceuticals International (Aliso Viejo, CA). IBMX, 8-(4-chlorophenylthio)-cAMP (CPT-cAMP), 8-bromo-cAMP, dibutyryl-cAMP, and all other chemicals and drugs were obtained from Sigma-Aldrich (St. Louis, MO).

Electrophysiology.

Enzymatically dissociated monkey RPE cells were plated on the bottom of a recording chamber and constantly superfused with bath solutions delivered by gravity feed. Whole cell currents were recorded at room temperature using the perforated-patch or standard whole cell configuration of the patch-clamp technique. For perforated-patch recording, patch pipettes were front-filled with standard pipette solution and then back-filled with pipette solution containing 120 μg/ml amphotericin B. Series resistance was normally <10 MΩ and was uncompensated. Patch electrodes were pulled from 1.65-mm-OD glass capillary tubing (Warner Instruments, Hamden, CT) using a multistage programmable microelectrode puller (model P-97, Sutter Instruments, San Rafael, CA) and had an impedance of 2–5 MΩ after fire polishing. Signals were amplified with an Axopatch 200 amplifier (Molecular Devices, Sunnyvale, CA).

Data acquisition and analysis were carried out using pCLAMP 9.0 (Molecular Devices). Statistical analysis and curve fitting were performed using Excel (Microsoft) and SigmaPlot 10 (Systat Software, San Jose, CA). The M-type current was identified from its characteristic activation and deactivation kinetics. Tail current analysis was used to determine the M-type conductance (g) as follows. The membrane potential was stepped from a holding potential of −10 mV to various test potentials ranging from −160 to +40 mV for 1 or 2 s. These voltage steps produced relaxations in currents resulting from the deactivation or activation of M-type channels and were of sufficient duration for the conductance to reach >95% of its steady-state value. On return of the voltage to −10 mV, tail currents developed, reflecting a change in conductance [Δg(−10 mV)]. The amplitude of Δg(−10 mV) was calculated using the following relationship

$$\Delta g(-10\text{mV})=I(-10\text{mV})/(73\text{mV})$$ (1)

where I(−10 mV) is the amplitude of the tail current at −10 mV and 73 mV is the driving force on K⁺ at −10 mV. The conductance that was open at each prepulse potential was calculated as the difference between the maximal conductance and the conductance activated by stepping the voltage from the prepulse potential to −10 mV or

$$g(V)=g_{\text{max}}-\Delta g(-10\text{mV})$$ (2)

where V is the prepulse voltage and g_max is the maximal M-type conductance. The activation curve was determined by performing a nonlinear least-squares fit of the data to a Boltzmann equation

$$g/g_{\text{max}}=1/\{1+\text{exp}[(V_{1/2}-V_{\text{m}})/\text{S}]\}$$ (3)

where V_1/2 is the half-activation potential (i.e., the potential at which g = ½g_max), V_m is the membrane potential, and S is the slope factor.

All voltages were corrected for an offset potential resulting from the liquid junction potential between the pipette tip and bath solution, which was measured to be −10 mV using flowing saturated KCl bridges.

Statistics.

Values are means ± SE. Comparisons between two experimental conditions were evaluated by Student's t-test.

RESULTS

Freshly isolated monkey RPE cells typically had a “figure-eight” shape, similar to that observed previously in RPE cells acutely dissociated from bovine (33), human (12), and monkey retinas (36), with the apical membrane domain recognizable by the presence of short microvillus processes extending from its surface.

The zero-current potential (V₀) of isolated RPE cells was measured in the zero-current clamp mode after the series resistance had decreased to ∼30 MΩ or, in the case of whole cell recording, immediately after rupture of the membrane patch. For cells recorded in the perforated-patch configuration, the mean V₀ was −34 ± 2 mV (n = 70); for cells in the whole cell configuration, V₀ was −33.3 ± 10.6 mV (n = 127). For all cells, membrane capacitance averaged 30.7 ± 0.9 pF (n = 191).

We observed three types of K⁺ current in isolated monkey RPE cells: an M-type K⁺ current (33), an inwardly rectifying K⁺ current consistent with Kir7.1 channels (29), and a delayed rectifier K⁺ current (33). The inward rectifier and delayed rectifier currents in most monkey RPE cells were relatively small, resulting in a predominant M-type current. The M-type current could be isolated by holding the membrane potential at a depolarized potential to inactivate the delayed rectifier and analyzing the time-dependent component of whole cell currents to separate it from the inward rectifier. M-type current was observed in most (99 of 117) of the cells successfully recorded in this study.

Figure 1A shows a representative family of whole cell currents in an isolated monkey RPE cell recorded in the perforated-patch configuration. Depolarizing voltage steps positive to −50 mV from a holding potential of −70 mV evoked an M-type current that showed no signs of inactivation during the 1-s depolarization. Figure 1B shows a family of currents recorded in the same cell evoked from a holding potential of −10 mV. Depolarizing voltage steps evoked small time-dependent outward currents, whereas hyperpolarizing voltage pulses produced current relaxations resulting from channel deactivation that reversed near the K⁺ equilibrium potential (approximately −83 mV). The steady-state current-voltage relationships generated from the two holding potentials were nearly identical (Fig. 1C), confirming that the outward current underwent little inactivation.

Fig. 1.

M-type current. A: families of currents evoked from a holding potential of −70 mV by voltage steps ranging from +40 to −60 mV. Currents were recorded in the perforated-patch configuration with Ringer solution in the bath. Horizontal line indicates zero current level (V₀). B: currents evoked in the cell shown in A from a holding potential of −10 mV by voltage steps ranging from +40 to −120 mV. Hyperpolarizing voltage steps caused M-type channel deactivation, resulting in current relaxations that reversed near the K⁺ equilibrium potential (−83 mV). C: steady-state current-voltage relationships obtained with holding potentials (HPs) of −10 and −70 mV. D: normalized conductance (g/g_max)-voltage curves calculated from tail currents at −10 mV following prepulses to various potentials. Values are means ± SE of 32 cells in the perforated-patch configuration and 58 cells in the whole cell configuration; smooth curves are least-squares fits of data to the Boltzmann equation, with half-maximal voltage (V_1/2) = −30.0 and −35.6 mV for perforated patch and whole cell, respectively.

To determine the voltage dependence of M-type channel gating, we determined its activation curve by analyzing the tail currents that developed at a holding potential of −10 mV following prepulses to various potentials (see methods) in cells that lacked delayed rectifier currents and had well-resolved M-type currents. Figure 1D summarizes the results of experiments in 32 cells recorded in the perforated-patch mode and plots the normalized M-type conductance as a function of prepulse potential. The least-squares fit of the data yielded average values for V_1/2 and S of −35.6 ± 1.6 mV and 11.0 ± 0.4, respectively.

When cells were recorded in the whole cell recording mode, the M-type current often grew within the first several minutes after rupture of the membrane patch (not shown), resulting in an M-type conductance that, on average, was larger than that observed in cells recorded in the perforated-patch mode [4.85 ± 0.54 nS (n = 67) vs. 2.30 ± 0.30 nS (n = 32)]. The reason for the increase in M-type current amplitude during whole cell recording is unclear, but it could be related to the washout of an inhibitory factor from the cytoplasm or the diffusion into the cell of a stimulatory factor. In addition, V_1/2 was somewhat more depolarized in cells recorded in the whole cell mode (−30.3 ± 1.1 mV, n = 58; Fig. 1D). The M-type current ran down within minutes when ATP was omitted from the pipette solution (n = 4, not shown), presumably due to the depletion of membrane phosphatidylinositol 4,5-bisphosphate (32).

In the majority of cells, M-type current activation evoked by a depolarizing step to +40 mV was essentially complete within 500 ms. Current activation at voltages positive to −20 mV could be well described by a second-order exponential function. The slow activation time constant (τ_act,slow) was voltage-independent (Fig. 2A), whereas the fast activation time constant (τ_act,fast) decreased with increasing depolarization (Fig. 2B). The slow activation component contributed to about half of the M-type current at positive voltages (Fig. 2C). Deactivation kinetics generally could be well described by a single-exponential function. Figure 2D shows that the deactivation time constant (τ_deact) increased as the membrane was depolarized. The τ_act,slow was somewhat larger in cells recorded in the whole cell mode than in those recorded in the perforated-patch mode (P < 0.005); otherwise, the kinetics of activation and deactivation were not significantly different in the two recording configurations.

Fig. 2.

Activation and deactivation kinetics of M-type current. A: slow time constant of activation (τ_act,slow) as a function of membrane voltage. Values are means ± SE of 6–12 cells in the perforated-patch configuration and 10–28 cells in the whole cell configuration. B: fast time constant of activation (τ_act,fast) as a function of membrane voltage. Values are means ± SE of 11–12 cells in the perforated-patch configuration and 22–28 cells in the whole cell configuration. C: relative contribution of the slow component of activation to M-type current amplitude. Values are means ± SE of 5–12 cells in the perforated-patch configuration and 13–28 cells in the whole cell configuration. D: deactivation time constant (τ_deact) as a function of membrane voltage. Values are means ± SE of 9–11 cells in the perforated-patch configuration and 19–27 cells in the whole cell configuration. Time courses of current activation and deactivation were fitted by single exponentials, except for current activation at voltages positive to −20 mV, which was fitted by double exponentials.

Blocker sensitivity.

We tested the effect of several K⁺ channel blockers on the M-type current. Linopirdine, a relatively specific blocker of recombinant and native KCNQ channels (22), was an effective inhibitor of the M-type current in the micromolar range. Figure 3A shows families of whole cell currents recorded in the absence and presence of 300 nM linopirdine. The effect of linopirdine on the M-type conductance is depicted in Fig. 3B, which shows that M-type conductance in this cell was blocked by ∼64%. Figure 3C summarizes the results of five experiments in which various concentrations of linopirdine were tested. The smooth curve is the best fit of the data to a first-order equation describing 1:1 binding of linopirdine to the channel with an apparent IC₅₀ of 0.5 μM. XE991, a structurally related and more potent inhibitor of KCNQ channels (22), blocked the M-type conductance with an apparent IC₅₀ of 0.3 ± 0.1 μM (n = 3; not shown); at 10 μM, XE991 completely blocked the M-type current (2.5 ± 1.5% of control, n = 9). Blockade of the M-type current by 10 μM linopirdine and 10 μM XE991 depolarized V₀ from −56.0 ± 3.8 to −36.7 ± 7.8 mV (P < 0.002, n = 8) and from −54.4 ± 4.7 to −40.5 ± 8.1 mV (P < 0.005, n = 8), respectively. The depolarizations produced by these blockers indicate that the M-type conductance contributes substantially to the RPE membrane potential.

Fig. 3.

Effect of linopirdine on M-type current. A: whole cell currents recorded in a cell in the absence and presence of 300 nM extracellular linopirdine. B: effect of 300 nM linopirdine on M-type conductance-voltage relationship. Data are from the cell depicted in A. C: dose-response curve for linopirdine-induced block. Values are means ± SE (n = 5). Least-squares fit (solid line) of data yields IC₅₀ = 500 nM.

As shown previously in studies on bovine (33) and human (12) RPE cells, the M-type current was significantly inhibited by 5 mM barium (Fig. 4C) but was remarkably insensitive to block by TEA. Figure 4A compares whole cell currents of a monkey RPE cell recorded in the absence and presence of 135 mM TEA (Na⁺ replacement). TEA caused a decrease in the amplitude of a time-independent inward Na⁺ current but had little effect on the size of inward current relaxations and tail currents corresponding to M-type channel deactivation and activation, respectively. Similar results were obtained in nine other cells, with 135 mM TEA having no significant effect on M-type current amplitude (Fig. 4C). Among homomeric KCNQ channels, KCNQ1 and KCNQ2 exhibit high sensitivity to block by extracellular TEA, whereas KCNQ3, KCNQ4, and KCNQ5 are relatively TEA-insensitive (2, 22). The M-type current in the RPE was moderately sensitive to block by chromanol 293B, a KCNQ1 channel blocker (22). Exposure to chromanol 293B at 100 μM, a concentration that completely blocks KCNQ1 channels, inhibited the M-type current by only 43.5 ± 9.8% (Fig. 4, B and C). Together, the results obtained with TEA and chromanol 293B argue against a major contribution of KCNQ1 or KCNQ2 channels to the M-type current in the RPE.

Fig. 4.

Effects of 135 mM tetraethylammonium (TEA, A) and 100 μM chromanol 293B (B) on whole cell current. C: percent control M-type current in the presence of 5 mM barium (n = 6), 135 mM TEA (n = 9), and 100 μM chromanol 293B (n = 5). Values are means ± SE. *P < 0.05; **P < 0.001. n.s, Not significant.

KCNQ1 subunits can co-assemble with KCNE2 (7) or KCNE3 (28) subunits to form constitutively open channels that exhibit linear current-voltage relationships. If either of these heteromeric channels were functional in the RPE, then the application of TEA or chromanol 293B would be expected to inhibit a time-independent K⁺ current. As mentioned above, the substitution of TEA for Na⁺ reduced an inward Na⁺ current (Fig. 4A), which could have obscured the inhibition of inward K⁺ current. Exposure of RPE cells to 100 μM chromanol 293B, however, had no significant effect on the amplitude of the inward current at −100 mV (94.0 ± 6.0% of control, n = 4, P > 0.2). This result argues against a contribution of KCNQ1 channels to the background K⁺ current.

Effect of channel openers.

Retigabine, an antiepileptic drug that was recently approved for clinical use, has been shown to activate native M-type currents (34), as well as KCNQ2, KCNQ3, KCNQ4, and KCNQ5 currents (25). A tryptophan residue in the S5 segment of KCNQ subunits is crucial for the effects of retigabine (25), although additional residues have also been implicated (16). Figure 5A shows the families of currents in a monkey RPE cell in the absence and presence of 10 μM retigabine. Retigabine increased the amplitude of the M-type conductance at +40 mV from 6.00 to 9.46 nS (Fig. 5B) and hyperpolarized V₀ from −49 to −60 mV (Fig. 5C). In 10 cells, 10 μM retigabine increased the M-type conductance from 3.71 ± 0.87 to 6.93 ± 0.81 nS (Fig. 7) and hyperpolarized V₀ from −47.2 ± 2.2 to −61.3 ± 1.8 mV. As shown in Fig. 6A and Table 1, retigabine also caused a ∼20-mV negative shift in the V_1/2 of the M-type conductance-activation curve but had essentially no effect on S (not shown). Retigabine caused a pronounced slowing of deactivation kinetics (Fig. 6B) but had no consistent effect on activation kinetics (Fig. 6, C and D; Table 1). The effects of retigabine were reversible (not shown).

Fig. 5.

Effect of retigabine on M-type current. A: whole cell currents in the absence and presence of 10 μM retigabine. B: effect of 10 μM retigabine on M-type conductance-voltage relationship. Retigabine increased maximum conductance and shifted the threshold for activation to a more hyperpolarized potential. C: effect of 10 μM retigabine on whole cell current-voltage relationship. Data in A–C are from the same cell.

Fig. 7.

Effects of various KCNQ channel openers [10 μM retigabine (n = 10), 10 μM zinc pyrithione (ZnPy, n = 6), 300 μM zinc (n = 6), 100 μM N-ethylmaleimide (NEM, n = 6), and cAMP (1 mM cAMP or cAMP cocktail, n = 7)] on M-type conductance. Values are means ± SE. *P < 0.05; **P < 0.01.

Fig. 6.

Effect of retigabine on M-type conductance-voltage dependence and kinetics. A: normalized M-type conductance-voltage curves in the absence (control, n = 10) and presence (n = 10) of 10 μM retigabine. Smooth curves are least-squares fits of data to Eq. 3, with V_1/2 = −30.7 and −53.0 mV for control and retigabine, respectively. Retigabine (10 μM) shifted the voltage-activation curve to more hyperpolarized voltages. Values are means ± SE. B: effect of retigabine (10 μM) on M-type current deactivation time constant. Values are means ± SE of 5–8 untreated (control) cells and 4–8 retigabine-treated cells. C: effect of retigabine on M-type current τ_act,slow. Values are means ± SE of 5 control and 6–7 retigabine-treated cells. D: effect of retigabine on M-type current τ_act,fast. Values are means ± SE of 7 cells in each group. Retigabine had no significant effect on τ_act,slow or τ_act,fast.

Table 1.

Effect of channel openers on voltage dependence and kinetics

	V1/2, mV	τdeact (−60 mV), ms	τact, fast (−10 mV), ms	τact, slow (−10 mV), ms
Retigabine
Control	−31 ± 3 (10)	81 ± 13 (13)	72 ± 9 (11)	382 ± 65 (10)
Treated	−52 ± 4* (10)	212 ± 23* (13)	74 ± 9† (11)	342 ± 42† (10)
ZnPy
Control	−33 ± 5 (4)	114 ± 23 (4)	55 ± 16 (3)	400 ± 213 (4)
Treated	−52 ± 7* (4)	504 ± 205† (3)	171 ± 99† (3)	486 ± 152† (4)
NEM
Control	−37 ± 4 (5)	68 ± 12 (7)	56 ± 13 (7)	288 ± 42 (7)
Treated	−34 ± 5† (4)	77 ± 14† (7)	46 ± 9† (7)	294 ± 39† (7)
Diclofenac
Control	−33 ± 6 (13)	55 ± 19 (7)	57 ± 13 (8)	252 ± 91 (4)
Treated	−55 ± 7* (13)	86 ± 28* (7)	43 ± 14* (8)	140 ± 41† (4)

Values are means ± SE of number of cells in parentheses. V_1/2, half-maximal potential; τ_deact, deactivation time constant; τ_{act, fast}, fast time constant of activation; τ_{act, slow}, slow time constant of activation; ZnPy, zinc pyrithione; NEM, N-ethylmaleimide.

^*P < 0.05;

^†P > 0.05 (2-tailed t-test).

ZnPy, a reversible and potent opener of all KCNQ channels except KCNQ3 (39), also augmented the M-type current in monkey RPE cells. The results are summarized in Fig. 7, which shows that 10 μM ZnPy increased the M-type conductance from 1.88 ± 1.13 to 7.86 ± 3.06 nS (n = 6, P < 0.05). Consistent with reported effects of ZnPy on heterologously expressed KCNQ channels (39), ZnPy produced a ∼20-mV negative shift in the V_1/2 of the M-type current activation curve (Table 1) and hyperpolarized V₀ from −51.1 ± 5.2 to −69.7 ± 2.8 mV (P < 0.01, n = 6). In addition, ZnPy slowed M-type current activation and deactivation rates, increasing τ_act,fast at −10 mV about threefold and increasing τ_deact at −60 mV nearly fivefold (Table 1). It has been reported that mouse KCNQ5 is activated by ionic zinc in the hundred micromolar concentration range (13). In six cells tested, the M-type conductance was 3.57 ± 1.17 and 2.81 ± 0.83 nS in the absence and presence of 300 μM ZnSO₄, respectively (Fig. 7), a difference that is not significant (P > 0.05). Hence, the effects of ZnPy on the M-type current in RPE cells cannot be attributed to ionic zinc.

NEM, a cysteine-modifying agent, augments native M-type channels and KCNQ2, KCNQ4, and KCNQ5 channels (19), but not KCNQ1 or KCNQ3 channels (23). Bath application of 50 μM NEM caused augmentation of the M-type current in six of eight RPE cells tested, and in four of these cells, M-type current amplitude declined after reaching a peak to a steady-state level that was still greater than the initial baseline (results not shown). For the six cells that were responsive, NEM increased the steady-state M-type conductance from 2.68 ± 0.98 to 4.30 ± 1.30 nS (Fig. 7; P < 0.01). NEM had no obvious effects on the kinetics or the voltage dependence of activation (Table 1), nor did it affect V₀ (−52.5 ± 2.9 and −54.0 ± 2.3 mV for control and NEM, respectively, n = 6, P > 0.05).

cAMP has been reported to activate heterologously expressed KCNQ1/KCNE1 (15), KCNQ1/KCNE2 (7), and KCNQ1/KCNE3 (28), as well as homomeric KCNQ4 (4), channels. The addition to the bathing solution of exogenous 1 mM dibutyryl-cAMP or 8-bromo-cAMP (n = 4) or a cocktail of cAMP-elevating agents consisting of 10 μM forskolin, 0.5 mM IBMX, and 0.1 mM CPT-cAMP (n = 3) had no significant effect on the amplitude of the M-type current in RPE cells (Fig. 7), nor did it affect the background K⁺ conductance (data not shown).

It was recently shown that diclofenac, a nonsteroidal anti-inflammatory drug that enhances KCNQ2/3 current (21), has different effects on KCNQ4 and KCNQ5 channels heterologously expressed in A7r5 rat aortic smooth muscle cells. Diclofenac at 100 μM increased KCNQ4 conductance, produced a rapid voltage-dependent block of KCNQ5 conductance, and induced a slower and less extensive inhibition of KCNQ4/5 channels (3). In addition, diclofenac had different effects on the deactivation rates and the voltage dependence of activation of these channels.

In the present study, we found that the effects of diclofenac on the M-type current amplitude in RPE cells were variable. In a total of 13 cells tested, the application of 100 μM diclofenac caused augmentation of M-type current amplitude in 3 cells, inhibition in 9 cells, and no significant change in 1 cell. Figure 8A shows the results of an experiment in which exposure of an RPE cell to 100 μM diclofenac resulted in a twofold increase in M-type current amplitude; for the three cells that responded similarly, diclofenac increased the M-type current by 65.4 ± 23.1%. Figure 8C summarizes the diclofenac-induced changes in M-type conductance produced by 10 and 100 μM diclofenac and shows that in two cells the M-type current augmentation by diclofenac was dose-dependent but in the third cell the current was augmented by 10 μM diclofenac but partially inhibited by 100 μM diclofenac. The diclofenac-induced activation of M-type current developed slowly, taking several minutes for completion (results not shown).

Fig. 8.

A and B: variable effects of diclofenac on M-type current. Whole cell currents were recorded in a cell in the absence (control) and presence of 100 μM diclofenac. In A, diclofenac induced augmentation of M-type current; in B, diclofenac induced inhibition of M-type current. Diclofenac did not affect the shape of deactivating (arrow at bottom left) and activating (arrow at top right) M-type currents. C: summary of results obtained in 13 cells showing variability of changes in M-type conductance (g) induced by 10 and 100 μM diclofenac. Each symbol represents data from an individual cell. D: time course of changes in whole cell current measured at +40 mV caused by superfusion of a retinal pigment epithelium cell with 100 μM diclofenac. E: normalized M-type conductance-voltage curves in the absence (n = 9) and presence (n = 9) of 100 μM diclofenac. Values are means ± SE. Smooth curves are least-squares fits of data to Eq. 3, with V_1/2 = −33.2 and −52.0 mV for control and diclofenac, respectively.

Figure 8B shows the results obtained in another cell in which 100 μM diclofenac caused M-type current inhibition; for the nine cells exhibiting this type of response, 100 μM diclofenac reduced the M-type current by 66.2 ± 8.4%. The diclofenac-induced inhibition of M-type current was dose-dependent in all cells expect one in which diclofenac caused inhibition at 10 μM but partial recovery at 100 μM (Fig. 8C). As with diclofenac-induced activation, diclofenac inhibition of M-type current developed over the course of several minutes, and in five cells, inhibition was preceded by a transient current increase. An example of this biphasic response is shown in Fig. 8D, which plots the whole cell current at a holding potential of +40 mV as a function of time. In four cells, using periodic voltage ramps, we monitored the time course of diclofenac-induced current changes and determined that the transient increase in current reflected changes in outwardly rectifying current, presumably the M-type current (results not shown).

It was reported previously that the voltage-dependent block of homomeric KCNQ5 channels by diclofenac is manifested by the appearance of time-dependent changes in current at the beginning of voltage steps, specifically, inactivating currents evoked by membrane depolarization due to channel block and activating currents evoked by membrane hyperpolarization due to channel unblock (3). However, as shown in Fig. 8B, right, diclofenac did not alter the overall kinetics of RPE M-type current deactivation or activation. Thus some mechanism other than voltage-dependent block must be responsible for the diclofenac-induced inhibition of the M-type current.

In every cell tested, 100 μM diclofenac hyperpolarized V_1/2 (range −17 to −28 mV, mean change −21 ± 4 mV; Table 1). There was no significant difference in the size of this shift between cells exhibiting conductance inhibition and cells exhibiting conductance augmentation [−23.0 ± 1.2 mV (n = 9) and −18.4 ± 1.4 mV (n = 3), respectively, P > 0.05], and the combined results for all cells are summarized in Fig. 8E, which plots the normalized M-type conductance as a function of voltage. Diclofenac also slowed M-type current deactivation and accelerated its fast component of activation (Table 1), and again the results were indistinguishable between cells exhibiting conductance inhibition and cells exhibiting conductance augmentation.

Taken together, the results obtained with diclofenac suggest that RPE cells may be heterogeneous in their expression of KCNQ4 and KCNQ5 subunits and that although some cells may express homomeric KCNQ4 channels, the majority of RPE cells express heteromeric KCNQ4/5 channels.

DISCUSSION

Previous work in our laboratory determined that native human (12) and bovine (33) RPE cells exhibit a sustained outwardly rectifying K⁺ current that kinetically resembles the M-type current in excitable cells. This current was blocked by 2 mM extracellular barium but was relatively insensitive to extracellular TEA (12). In the intervening years, it has been well established that channels mediating M-type currents are composed of KCNQ (Kv7) subunits. Recently, we demonstrated the expression of KCNQ1, KCNQ4, and KCNQ5 transcripts in monkey RPE (41). In the present study, we characterized the biophysical properties of the M-type current in freshly dissociated monkey RPE cells and used a variety of KCNQ channel blockers and openers to obtain evidence for contributions by KCNQ subunits to the generation of this current. The results suggest that channels comprising KCNQ4 and KCNQ5 subunits contribute to the channel mediating the M-type current.

We found that the M-type current's voltage dependence of activation could be well described by a Boltzmann function with a V_1/2 of −35.6 mV and S of 11.0 mV/e-fold change in conductance. The activation time course was best fit by the sum of two exponentials, with the slow component being voltage-independent and the τ_act,fast decreasing with greater depolarization. The fast and slow components of activation contributed equally to the M-type current amplitude over a wide voltage range, and current activation at +40 mV was essentially complete within 500 ms. These results are similar to those reported in a previous study on fresh human RPE cells (12) and are compatible with the properties of heterologously expressed and native KCNQ channels.

To define which KCNQ channel subtypes contribute to the M-type current, we took advantage of the fact that KCNQ channels differ in their sensitivities to block by various pharmacological agents. Linopirdine and its more potent analog XE991 have been useful in the study of heterologously expressed and native KCNQ channels and have been shown to block all homomeric KCNQ channels with an IC₅₀ of 0.5–75 μM. Although linopirdine and XE991 also block other voltage-dependent currents (5, 26, 38), they do so at higher concentrations (IC₅₀ >100 μM). We found that the M-type current in RPE cells was quite sensitive to block by linopirdine, with an apparent IC₅₀ of 0.5 μM. Consistent with our previous finding (41), XE991 was also effective, with an apparent IC₅₀ of 0.3 μM. The IC₅₀ values for the block of the M-type current by linopirdine and XE991 are similar to those of KCNQ1, with IC₅₀ for block of 9 and 0.8 μM, respectively (35), but are about an order of magnitude lower than those of homomeric KCNQ4 and KCNQ5 channels, with IC₅₀ of 6–14 and 16–75 μM, respectively (18, 27, 30). This would seem to argue against the involvement of KCNQ4 or KCNQ5 channels in the M-type current and in favor of KCNQ1, but it is possible that the M-type channel sensitivity to these blockers is influenced by interactions with KCNE subunits or other factors.

The results obtained with other KCNQ channel blockers argue against a contribution of KCNQ1 channels to the M-type current. Consistent with previous results obtained on M-type currents in human (12) and bovine (33) RPE cells, we found that the M-type current in monkey RPE was insensitive to block by TEA (Fig. 4). KCNQ1 channels are sensitive to block by extracellular TEA in the millimolar range (6), whereas KCNQ5 channels are relatively TEA-insensitive, with IC₅₀ of 43 mM (2). It has been stated that the KCNQ4 channel has an IC₅₀ for block by TEA of 3 mM (6), but a more recent study reported an IC₅₀ of ∼50 mM (2); the latter value is in agreement with our results testing the effect of TEA on hKCNQ4_v1 and hKCNQ4_v2 channels expressed in Chinese hamster ovary cells (unpublished data). In addition, chromanol 293B, which blocks homomeric and heteromeric KCNQ1 channels with an IC₅₀ of <30 μM (17), blocked only 44% of the M-type current at a concentration of 100 μM. The low sensitivity of the M-type current to this drug is consistent with KCNQ4 and KCNQ5 channels, both of which are blocked moderately by 100 μM chromanol 293B (17).

To further define the relationship between the M-type current and KCNQ subunits, we tested the effects of several KCNQ channel openers. Retigabine, which activates KCNQ2, KCNQ3, KCNQ4, and KCNQ5, but not KCNQ1, channels (25), augmented the M-type current, as did ZnPy, a potent opener of all KCNQ channels except KCNQ3 (39). In addition to increasing M-type current amplitude at saturating voltages, both openers also produced hyperpolarizing shifts in V_1/2 and caused marked slowing of deactivation kinetics, effects that have been reported for KCNQ channels. The M-type current amplitude in the RPE was also increased by exposure to the alkylating agent NEM, which activates KCNQ2, KCNQ4, and KCNQ5, but not KCNQ1 or KCNQ3, channels (23), although it did so without producing the negative shift in voltage dependence and acceleration in the rates of activation and deactivation that have been reported for KCNQ channels. Thus the responses of M-type current to these KCNQ blockers and openers argue against the involvement of KCNQ1 channels but are compatible with contributions by KCNQ4 and KCNQ5 channels.

Recently, it was shown that the nonsteroidal anti-inflammatory drug diclofenac affects KCNQ4 and KCNQ5 channels overexpressed in A7r5 rat aortic smooth muscle cells differently: 100 μM diclofenac increased the maximal conductance of homomeric KCNQ4 channels, inhibited homomeric KCNQ5 channels, and partially inhibited heteromeric KCNQ4/5 channels (3). The diclofenac-induced inhibition of KCNQ5 channels was rapid and associated with changes in current kinetics that were consistent with voltage-dependent block, whereas the effects on KCNQ4 and KCNQ4/5 channels were slow to develop. In addition, diclofenac produced larger changes in the voltage-activation curve and deactivation rates of KCNQ5 than KCNQ4 or KCNQ4/5 channels.

In the present study, we found that the effects of diclofenac on the M-type current amplitude in the RPE varied from cell to cell: in 13 cells tested, 100 μM diclofenac augmented the current in 3 cells, inhibited it 9 cells, and had no effect in the remaining cell. Superficially, the results suggest that RPE cells may be heterogeneous with respect to their expression of KCNQ4 and KCNQ5 subunits. Two aspects of the diclofenac-induced inhibition of the RPE M-type current argue against the idea that it is mediated by homomeric KCNQ5 channels: 1) M-type current inhibition developed slowly, rather than abruptly, and 2) inhibition was not associated with current relaxations at the beginning of voltage steps that would be expected for voltage-dependent block and unblock of the channel. These features are more compatible with M-type currents in the RPE being mediated by KCNQ4/5 channels. On the other hand, the response characteristics of the RPE M-type current and those of overexpressed KNCQ4/5 channels differ in other respects. KCNQ4/5 channels reportedly are unaffected by 10 μM diclofenac, but we observed significant inhibition of M-type current in RPE cells at this concentration. Furthermore, compared with KCNQ4/5 channels, inhibition of M-type channels by 100 μM diclofenac was more extensive and was associated with a larger negative shift in the voltage-activation curve than has been reported for KCNQ4/5 channels (−22 vs. −15 mV). The reason for these discrepancies is unknown, but it is possible that the response of KCNQ4/5 channels in the RPE to diclofenac is modified by their association with auxiliary KCNE subunits or other proteins, their phosphorylation state, or other factors.

In a study of KCNQ4 expressed in Chinese hamster ovary cells, phosphorylation by PKA increased KCNQ4 currents and shifted the activation curve to more negative potentials (4). The present data, however, show that maneuvers expected to elevate intracellular cAMP in RPE cells had no significant effect on the M-type current. If we assume that KCNQ4 channels contribute to the M-type current in the RPE, the lack of an effect of cAMP suggests a low level of PKA activity in these cells. Additional experiments involving the dialysis of RPE cells with the catalytic subunit of PKA might help answer this question.

In some epithelia, KCNQ1 subunits co-assemble with KCNE2 (7) or KCNE3 (28) β-subunits to form constitutively active channels that exhibit little rectification. Similar to KCNQ1/KCNE1 channels, KCNQ1/KCNE2 (7) and KCNQ1/KCNE3 (28) heteromeric channels are inhibited by chromanol 293 and enhanced by cAMP. In the present study, neither chromanol 293 nor cAMP affected the time-independent K⁺ current in monkey RPE cells, indicating that if KCNQ1/KCNE2 or KCNQ1/KCNE3 channels are present in these cells, they must make up a small fraction of the K⁺ conductance.

We recently localized KCNQ5 protein to the basal membrane of monkey RPE (41), suggesting a possible contribution of KCNQ5 channels to the basolateral membrane K⁺ conductance. As a major route for K⁺ efflux across the basolateral membrane, K⁺ channels play a key role in active K⁺ absorption (20), which, together with active Cl⁻ absorption, promotes retinal adhesion by providing an important driving force for the absorption of subretinal fluid that lies between the rod and cone photoreceptors and the RPE (24). In native human RPE, the basolateral membrane potential is in the range −45 to −55 mV. At this voltage, the M-type conductance would be partially activated and exert a hyperpolarizing influence, which would help promote Cl⁻ absorption by maintaining the electrochemical gradient for Cl⁻ efflux across the basolateral membrane.

Another potential function of the M-type channels in the RPE is in cell volume regulation. Studies on cultured human (14) and native frog (1) RPE have implicated the basolateral membrane K⁺ conductance in regulatory volume decrease. Because KCNQ4 and KCNQ5 channels are sensitive to small changes in cell volume (8, 13), it seems likely that these channels could be involved in regulatory volume decrease in the RPE. Additional experiments on intact RPE with KCNQ blockers and openers could help confirm the role of KCNQ channels in K⁺ transport and cell volume regulation. The specific contributions of KCNQ4 and KCNQ5 subunits to these processes require experiments involving gene knockdown or knockout approaches.

GRANTS

This work was supported by National Eye Institute Research Grant EY-08850 and Core Grant EY-07703, the Foundation Fighting Blindness, and a Research to Prevent Blindness Lew R. Wasserman Merit Award to B. A. Hughes.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.R.P. and B.A.H. are responsible for conception and design of the research; B.R.P. and B.A.H. performed the experiments; B.R.P. and B.A.H. analyzed the data; B.R.P. and B.A.H. interpreted the results of the experiments; B.R.P. and B.A.H. drafted the manuscript; B.R.P. and B.A.H. edited and revised the manuscript; B.R.P. and B.A.H. approved the final version of the manuscript; B.A.H. prepared the figures.