Abstract
Various neurotransmitters excite neurons by suppressing a ubiquitous, voltage-dependent, noninactivating K+ conductance called the M-conductance (gM). In bullfrog sympathetic ganglion neurons the suppression of gM by the P2Y agonist ATP involves phospholipase C (PLC). The present results are consistent with the involvement of the lipid and inositol phosphate cycles in the effects of both P2Y and muscarinic cholinergic agonists on gM. Impairment of resynthesis of phosphatidylinositol 4,5-bisphosphate (PIP2) with the phosphatidylinositol 4-kinase inhibitor wortmannin (10 μm) slowed or blocked the recovery of agonist-induced gM suppression. This effect could not be attributed to an action of wortmannin on myosin light chain kinase or on phosphatidylinositol 3-kinase. Inhibition of PIP2 synthesis at an earlier point in the lipid cycle by the use of R59022 (40 μm) to inhibit diacylglycerol kinase also slowed the rate of recovery of successive ATP responses. This effect required several applications of agonist to deplete levels of various phospholipid intermediates in the lipid cycle. PIP2 antibodies attenuated the suppression of gM by agonists. Intracellular application of 20 μm PIP2 slowed the rundown of KCNQ2/3 currents expressed in COS-1 or tsA-201 cells, and 100 μm PIP2 produced a small potentiation of native M-current bullfrog sympathetic neurons. These are the results that might be expected if agonist-induced activation of PLC and the concomitant depletion of PIP2 contribute to the excitatory action of neurotransmitters that suppress gM.
- KCNQ
- Gq/11
- phospholipase C
- lipid kinase
- M-current
- P2Y
Introduction
The M-conductance (gM) is a noninactivating, voltage- and time-dependent K+ conductance (Brown and Adams, 1980) that represents the product of KCNQ2/3 genes (Wang et al., 1998). Suppression of this conductance by neurotransmitters that act via Gq/G11 produces excitatory effects in a variety of neuron types (Caulfield et al., 1994; Marrion, 1997; Cruzblanca et al., 1998; Haley et al., 1998). Although bradykinin inhibits gM in rat sympathetic neurons via phospholipase C (PLC) and Ca2+ release from inositol trisphosphate (InsP3)-sensitive stores (Cruzblanca et al., 1998), this transduction mechanism does not explain the actions of other neurotransmitters hitherto studied (Pfaffinger et al., 1988; Brown et al., 1989; Selyanko et al., 1990; Marrion, 1997; Stemkowski et al., 2002; Suh and Hille, 2002).
The P2Y agonists ATP and UTP suppress gM in both amphibian (Groul et al., 1981; Adams et al., 1982a) and mammalian sympathetic neurons (Bofill-Cardona et al., 2000). In both cell types the inhibition is reduced by the PLC inhibitor U73122, implying a role for phosphatidylinositol 4,5-bisphosphate (PIP2) hydrolysis. In rat neurons available evidence suggests that inhibition results from the consequential formation of InsP3 and the subsequent release of Ca2+ (Bofill-Cardona et al., 2000). However, in frog neurons neither InsP3 or Ca2+ nor the other consequential pathway for PIP2 hydrolysis, activation of protein kinase C (PKC), appears to be involved (Stemkowski et al., 2002). This raises the possibility that ATP-induced gM suppression in bullfrog sympathetic ganglion (BFSG) neurons involves signaling via the PLC-induced depletion of PIP2. This type of “upstream” signaling from PLC underlies agonist modulation of Ca2+-permeant TRPM7 channels in cardiac myocytes (Runnels et al., 2002), KATP channels expressed in COS7 cells (Xie et al., 1999), and voltage-gated Ca2+ channels (Wu et al., 2002). Indeed, recent data suggest that a reduction in PIP2 concentration, as would occur after PLC activation (Willars et al., 1998), also could be responsible for agonist-induced closure of KCNQ2/3 channels expressed in Xenopus oocytes or Chinese hamster ovary (CHO) cells (Zhang et al., 2003). Because PIP2 increases putative M-channel activity in rat sympathetic neurons, this mechanism also may control gM in intact neurons (Zhang et al., 2003). Moreover, ATP-dependent resynthesis of PIP2 is required for the recovery of muscarinic suppression of gM in the same cell type. This implies that an agonist-induced reduction in PIP2 may have been responsible for M-channel closure in the first place (Suh and Hille, 2002). This emerging concept has been referred to as the “lipid kinase and PI-polyphosphate hypothesis” (Suh and Hille, 2002).
The results of the present experiments with enzyme inhibitors, PIP2 antibodies, and PIP2 itself are mainly consistent with the general lipid kinase and PI-polyphosphate hypothesis (Suh and Hille, 2002; Zhang et al., 2003). The specific findings are explicable in terms of the hypothesis that PLC-mediated depletion of PIP2 mediates gM suppression by P2Y agonists in BFSG neurons (Stemkowski et al., 2002). Moreover, this hypothesis also may explain the classical defining effect of muscarinic agonists on gM in BFSG (Brown and Adams, 1980).
A preliminary report of this work has appeared (Ford et al., 2002).
Materials and Methods
Animals were cared for in accordance with the principles and guidelines of the Canadian Council on Animal Care, and experimental protocols were approved by the Health Sciences Animal Welfare Committee of the University of Alberta.
BFSG experiments. Neurons in the seventh to tenth paravertebral sympathetic ganglia of male or female Rana catesbeiana were dissociated with trypsin and collagenase as previously described (Selyanko et al., 1990; Kurenny et al., 1994). Experiments were done on neurons that were maintained for 1–2 d in a culture medium that consisted of diluted L-15 medium (73%) supplemented with 10 mm glucose, 1 mm CaCl2, 100 U/ml penicillin, 100 μg/ml streptomycin, and 10 μm cytosine arabinoside. Electrophysiological recordings were done at ∼20°C via whole-cell or nystatin-perforated patch recording (Stemkowski et al., 2002) (Axoclamp 1B amplifier, pClamp 5.5. software, Axon Instruments, Foster City, CA). Resting membrane potential was -50 to -55 mV, and cells were held at -30 mV. Experiments with agonists (ATP, UTP, and muscarine) were performed on B-cells (Cin > 30 pF), whereas PIP2 was introduced into the smaller C-cells (Cin < 25 pF) of BFSG (Dodd and Horn, 1983; Kurenny et al., 1994). Because the currents to be recorded were usually <0.5 nA, no corrections were made for the voltage drop across the series resistance (Selyanko et al., 1990). Current—voltage relationships were obtained by using a 4.5 sec ramp command from the holding potential of -30 to -110 mV (∼18 mV/sec). Leak current (IL) at -30 mV was estimated by extrapolation of the I—V plot obtained at voltages between -75 and -90 mV. The percentage of agonist-induced gM suppression was calculated as in our previous work (Stemkowski et al., 2002). Extracellular solution contained (in mm): 113 NaCl, 6 KCl, 2 MgCl2, 2 CaCl2, 5 HEPES/NaOH, pH 7.2, and 10 d-glucose. Patch pipettes had direct current resistances of 3–10 MΩ. The pipette solution contained (in mm): 110 KCl, 10 NaCl, 2 MgCl2, 0.4 CaCl2, 4.4 EGTA, 5 HEPES/KOH, pH 6.7, 10 d-glucose, and 2 Na2ATP (pCa = 7) (Selyanko et al., 1990). Intracellular ATP was omitted in experiments in which the effect of PIP2 was examined.
Experiments on tsA-201 cells. tsA-201 or COS-1 cells were maintained in DMEM supplemented with 2 mm l-glutamine, 10% fetal calf serum, and 0.1% penicillin/streptomycin at 37°C with 10% CO2. KCNQ2 and KCNQ3 channel subunit clones were kindly provided by Dr. D. McKinnon (State University of New York, Stony Brook, NY). Cells were plated at ∼40–70% confluency on 35 mm dishes 4–6 hr before transfection. Clones were inserted into the mammalian expression vector pCDNA3 and transfected into COS-1 cells by using Lipofectamine reagent per the manufacturer's instructions (Invitrogen, San Diego, CA) or into tsA-201 cells by using the calcium phosphate precipitation technique. For expression of heteromultimers a 1:1 molar ratio of KCNQ2 and KCNQ3 cDNA was transfected. Successfully transfected cells were identified by coexpression of the green fluorescent protein plasmid (pGreen Lantern, Invitrogen), visualized via fluorescence optics. Recordings were made at 48–72 or 72–96 hr after transfecting for COS-1 or tsA-201 cells, respectively, in whole-cell mode with 2–6 MΩ electrodes. Data were acquired with pClamp software (version 8.0), using an Axopatch 200B amplifier (Axon Instruments), and were filtered at 2 kHz. Capacitance transients were canceled by using circuitry within the amplifier, and the series resistance compensation was set at 40–60%. Whole-cell recordings from transfected cells were made at ∼20°C, using an extracellular solution consisting of (in mm): 144 NaCl, 2.5 KCl, 2 CaCl2, 0.5 MgCl2, 5 HEPES, and 10 d-glucose, pH 7.4, and an intracellular solution consisting of (in mm): 80 K-acetate, 30 KCl, 40 HEPES, 3 MgCl2, 3 EGTA, and 1 CaCl2, pH 7.4. For recording from excised inside-out patches containing KCNQ2/3 channels, the extracellular solution (with 2.5 mm K +) was placed in the pipette, and the intracellular solution (with 110 mm K +) was placed in the recording dish. Both faces of the excised patches thus were exposed to “physiological” solutions. Holding potential was 0 mV; because EK was -95 mV, the driving force was 95 mV. Movement of K + ions through KCNQ2/3 channels in these patches results in current flow toward the pipette, and this is displayed as downward deflections on recordings.
The KATP channel Kir6.2 subunit clone from mouse was generously provided by Dr. S. Seino (Graduate School of Medicine, Chiba University, Japan), and the SUR1 subunit clone from hamster was generously provided by Drs. L. Aguilar Bryan and J. Bryan (Baylor College of Medicine, Houston, TX). Clones were inserted into the mammalian expression vector pCNDA3. KATP channels were transfected as 1:1 molar ratios of Kir6.2 and SUR1 into tsA-201 cells with the green fluorescent protein plasmid in an analogous manner to KCNQ2/3 channels. KATP channels were recorded under symmetrical K + conditions with a pipette solution and superfusion system containing (in mm): 110 KCl, 30 KOH, 5 HEPES, 1 MgCl2, and 10 EGTA, pH 7.4, in an inside-out patch configuration. Membrane patches were held at -60 mV, and their cytoplasmic faces were exposed directly to test Mg-ATP solutions via a multi-input perfusion pipette (time to change solution at the tip of the recording pipette was <2 sec). All experiments were performed at room temperature (20–22°C).
Drugs and chemicals. Drugs were applied to BFSG neurons by using a rapid superfusion system that was constructed from 0.8-mm-diameter polyimide tubes (for details, see Stemkowski et al., 2002). Aliquots of U73122, U73343, or R59022 were dissolved first in chloroform, which then was allowed to evaporate under a stream of nitrogen to yield a filmy residue. This was stored at -20°C until the day of the experiment when it was dissolved in fresh DMSO. LY294002, ML-7, and wortmannin also were dissolved in fresh DMSO. Serial dilutions of drugs in external solution were arranged so that the final concentration of DMSO in solutions applied to cells was <0.1%. Acute application of 0.1% DMSO had no noticeable effect on their electrophysiological properties (Table 1). PIP2 was dissolved in chloroform, and aliquots were stored at -20°C under N2. On the day of the experiment the chloroform was evaporated with a stream of N2 to leave a filmy residue of PIP2. Recording solution was mixed with this residue for ∼10 min before sonication on ice until the solution was clear (∼30 min). PIP2 antibody was provided as a 1:25 dilution of immune serum. This was dissolved 1:4 in internal solution to achieve a final dilution of 1:100. For control experiments the horse serum was dissolved 1:4 in internal solution. Solutions, especially those containing AlCl3, were made in HPLC water to avoid the formation of AlF4-.
U73122, U73343, ML-7, R59022, and wortmannin were obtained from Biomol (Plymouth Meeting, PA), PIP2 (α-phosphatidyl-d-myo- inositol-4,5-bisphosphate, triammonium salt from bovine brain) was from Calbiochem (San Diego, CA), LY294002 was from Alomone (Jerusalem, Israel), and PIP2 antiserum was from Assay Designs (Ann Arbor, MI). All other chemicals were obtained from Sigma (Oakville, Ontario, Canada). Data are expressed as the mean ± SEM; significance of difference was estimated by using Student's two-tailed t test. Paired tests were used for pairs of observations on the same cell, and unpaired tests were used for a comparison of data from groups of cells. Data were considered significantly different when p < 0.05.
Results
Inhibition of PIP2 resynthesis slows recovery from gM suppression
Wortmannin is a nonselective kinase inhibitor that inhibits myosin light chain kinase (Nakanishi et al., 1992) and phosphatidylinositol 3-kinase (PtdIns 3-kinase) (Yano et al., 1993). The latter is involved in the synthesis of phosphatidylinositol 3,4,5-trisphosphate (PIP3) from PIP2 (Akasu et al., 1993). At relatively high concentrations wortmannin also inhibits phosphatidylinositol 4-kinase (PtdIns 4-kinase) (Willars et al., 1998; Xie et al., 1999). This enzyme plays a pivotal role in the lipid and inositol phosphate cycles because it is involved in the synthesis of phosphatidylinositol 4-phosphate (PIP) from phosphatidyl inositol (PI). PIP is phosphorylated to PIP2 by phosphatidylinositol phosphate kinase (PtdInsP kinase; Fig. 1). If PLC-induced PIP2 depletion is required for agonist-induced gM suppression, the recovery of the response would require PIP2 resynthesis. This recovery should be slowed in the presence of wortmannin because it would impair PIP2 resynthesis by its action on PtdIns 4-kinase (Runnels et al., 2002; Suh and Hille, 2002) and reduction in the steady-state level of PIP. The data illustrated in Figure 2a, which are whole-cell recordings from a BFSG B-neuron, are consistent with this prediction. The traces in Figure 2a, which are chart recordings of steady-state IM at -30 mV, are interrupted by responses to voltage ramps that were used to monitor agonist-induced changes in membrane conductance. These ramps produce I—V plots such as those shown in Figure 2e (Selyanko et al., 1990). Before the application of 10 μm wortmannin, an extracellular application of 250 μm ATP produces pronounced gM suppression (Fig. 2a). This is reflected as a reduction in steady-state outward current in the gM activation range (positive to -70 mV; Fig. 2e). Steady-state IM/gM returns to its control value within 1–2 min of initiating ATP washout (Fig. 2a, white arrow). After 5 min in 10 μm wortmannin the recovery phase of the ATP response is slowed so that the response is almost irreversible (Fig. 2a, black arrow), and persistent suppression of gM is seen. The time for 50% recovery of ATP responses increased by 905%, from 19.7 ± 2.5 to 198.1 ± 33.5 sec (n = 11; p < 0.0005) in neurons studied with whole-cell recording (Table 1). Similar effects were seen with cells studied with nystatin-perforated patches (Table 1). In these, wortmannin slowed the recovery of ATP responses by 549%, from 21.6 ± 2.4 to 118.0 ± 36.0 sec (n = 9; p < 0.04). Wortmannin also slowed the recovery of responses evoked with 250 μm UTP (data not shown). No significant slowing of rate of recovery of successive ATP responses was seen in neurons studied in the absence of wortmannin with the whole-cell or perforated patch methods (Table 1). Figure 2b shows that wortmannin had little effect on steady-state gM/IM when it was applied alone for the 10 min time period of the experiment. Wortmannin thus produces an “agonist-dependent” block of gM.
If the effect of wortmannin reflected an action of PtdIns 3-kinase rather than PtdIns 4-kinase, a PtdIns 3-kinase inhibitor, such as LY294002, should affect ATP responses in a similar manner to wortmannin. This was not the case. Although LY294002 (10 μm) slightly slowed the recovery of ATP responses that were studied with whole-cell recording, this effect was not significant when compared with the time-matched control group (Table 1). Moreover, no slowing of recovery of ATP responses was seen when the effects of LY294002 were studied by the use of perforated patches (Table 1). LY294002 thus did not produce an agonist-dependent block of steady-state gM (Fig. 2c). LY294002 did not affect the time course of recovery of responses evoked with UTP (250 μm; data not shown).
Wortmannin was still capable of slowing the recovery of ATP responses in the presence of LY294002. The time for 50% recovery of ATP responses recorded after 10 min in 10 μm LY294002 was 23.25 ± 4.3 sec (n = 4). After a further 5 min in 10 μm wortmannin plus 10 μm LY294002 this increased to 201.3 ± 62.0 (n = 4; p < 0.03). This effect of wortmannin cannot be attributed to the inhibition of PtdIns 3-kinase because it persisted after this enzyme was inhibited with LY294002.
To show that the effect of wortmannin did not reflect its action on myosin light chain kinase, we tested whether the inhibitor ML-7 would slow the recovery of ATP responses in a similar manner to wortmannin. ML-7 (10 μm) produced a rapidly developing and rapidly reversible reduction in gM that was associated with an increase in membrane noise (Fig. 2d, gray arrow). Although this is suggestive of a channel block mechanism, the effect of ML-7 on gM did not exhibit any obvious voltage or use dependence (our unpublished observations). It therefore may reflect a bona fide action on myosin light chain kinase, an enzyme that has been reported to affect gM (Akasu et al., 1993; Tokimasa et al., 1995). Despite this, ATP still suppressed the gM that persisted in the presence of ML-7; more importantly, the time course of recovery of the response was changed little (Fig. 2d, Table 1). The pronounced slowing of recovery seen with wortmannin (Fig. 2a) thus cannot be attributed to an action on myosin light chain kinase.
The I—V plot shown in Figure 2e is derived from current responses to the voltage ramps shown in Figure 2a. This shows that the effect of the extracellularly applied P2Y agonist ATP is confined to the gM activation range. The absence of any conductance change at potentials more negative than -70 mV excludes the participation of P2X receptors in the response to ATP.
Inhibition of the lipid cycle slows recovery from gM suppression
The above observations on ATP responses are similar to those of Suh and Hille (2002), who suggested that resynthesis of PIP2 is required for the recovery of muscarine-induced gM suppression in rat superior cervical ganglion neurons and for the recovery of muscarinic suppression of KCNQ2/3 channel currents expressed in tsA-201 cells. If this is so, depletion of PI, which is the substrate for PtdIns 4-kinase (Fig. 1), also would be expected to slow gM recovery. This is because, by classical enzyme kinetics, the rate of synthesis of phosphatidylinositol phosphates by PtdIns 4-kinase will depend on the concentration of the substrate, PI. We therefore interrupted the lipid cycle after diacylglycerol (DAG) by using the DAG kinase inhibitor R59022 (40 μm). A typical experiment is illustrated in Figure 3a. The initial 5 min exposure to R59022 had little effect on the rate of recovery of the ATP response. However, a second test of ATP in the continued presence of R59022 produced a response with slowed recovery. This trend (progressive slowing of recovery of successive ATP responses in the continued presence of R59022) is apparent in the summarized data for all of the cells that were tested (Fig. 3b). Very little slowing of the recovery of successive ATP responses was seen in control neurons. Thus the time for 50% recovery of the ATP response in untreated neurons was 32.6 ± 5.1 sec (n = 8), and this was significantly less than the 56.7 ± 8.1 sec (n = 6) seen in neurons after 10 min in 40 μm R59022 (p < 0.025). We suggest that, after one test of ATP in the presence of R59022, the membrane concentrations of PI and PIP are still sufficient to sustain the normal rate of resynthesis of PIP2 via PtdIns 4-kinase and PtdInsP kinase (Fig. 1). M-channels thus reopen at the control rate. With the second application of ATP, however, PI and PIP may be depleted irreversibly because their resynthesis has been interrupted by the R59022 inhibition of DAG kinase (Fig. 1). The corresponding rate of gM recovery thus is reduced. These findings are consistent with a role for PIP2 resynthesis in gM recovery (Suh and Hille, 2002) and further implicate earlier intermediates of the lipid cycle as the source of phosphatidic acid for this effect (Fig. 1).
PIP2 antibodies attenuate ATP-induced gM suppression
Studies of PIP2 modulation of other types of K+ channels (Huang et al., 1998; Xie et al., 1999; Leung et al., 2000; Bian et al., 2001) have led to the development of well characterized PIP2-neutralizing antibodies that reduce PIP2-dependent channel activity in a variety of cell types (Huang et al., 1998; Liou et al., 1999; Bian et al., 2001; Chuang et al., 2001). Under the conditions of our experiments, however, inclusion of PIP2-neutralizing antiserum (100:1) in the patch pipette failed to affect resting gM in BFSG neurons. Despite this, the modulation of gM by ATP was disrupted consistently. Thus after 5 min of recording with antibody in the pipette, 250 μm ATP suppressed gM by 85.6 ± 3.1%, but after 25 min of recording the agonist effectiveness was reduced so that only 42.9 ± 15.2% suppression was seen (n = 4; p = 0.05, paired t test). A typical experiment is illustrated in Figure 4a. The steady state-gM was little changed by the antibody (Fig. 4a, gray arrows), whereas the amount of ATP-induced gM suppression was far greater for the first response (Fig. 4a, white arrow) than for the second response (Fig. 4a, black arrow). The effect of antiserum was not seen in control experiments in which an appropriate amount of horse serum was included in the recording pipette (Fig. 4b). Thus ATP suppressed gM by 83.4 ± 8.0% after 5 min with the intracellular horse serum and by 76.1 ± 7.9% after 25 min (n = 5; p > 0.5). One interpretation of these findings is that the antibody preferentially inhibits the PIP2—PLC interaction rather than a PIP2—M-channel interaction.
Direct effects of PIP2 on KCNQ2/3 channels
If M-channel closure is promoted by the PLC-mediated hydrolysis of PIP2 (Suh and Hille, 2002), steady-state gM should be increased in the presence of exogenously applied PIP2 (Ikeda and Kammermeier, 2002). One recent report suggests that exogenously applied PIP2 increases putative M-current activity in cell-attached excised patches from rat sympathetic neurons (Zhang et al., 2003). This seemingly simple experiment is remarkably difficult to perform. One issue may be the simple physical problem of taking a hydrophobic lipid, such as PIP2, and delivering to the cell membrane via the aqueous environment of the recording solutions. We therefore first sought to verify the effectiveness of our aqueous solutions of PIP2.
Because PIP2 is known to decrease the sensitivity of KATP channels to blockade by intracellular ATP (Fan and Makielski, 1997; Baukrowitz et al., 1998), we used this effect to confirm the activity of our PIP2 solutions. KATP channels were expressed in tsA-201 cells, and PIP2 was included in the recording pipette on the extracellular side of excised inside-out patches (Fig. 5a, inset). PIP2 is able to reach the cytosolic face to modulate KATP channels when it is applied to the extracellular face (Wada et al., 2002). After 5 min the control channels displayed similar ATP sensitivity to that initially recorded (n = 2; Fig. 5a). Channels exposed to PIP2 (20 μm) exhibited a marked reduction in sensitivity to intracellular ATP, because substantially greater current was available at 0.1 mm ATP and even at 1 mm (a concentration that fully inhibited control channels; n = 2; Fig. 5b).
Because a majority of the studies demonstrating the effects of PIP2 on K+ channels have been performed on expressed (Huang et al., 1998; Zhang et al., 1999; Leung et al., 2000; Bian et al., 2001) rather than on native channels, we next used whole-cell recording to examine the effect of PIP2 on KCNQ2/3 channels expressed in tsA-201 or COS-1 cells (Wang et al., 1998). In cells in which the channels were transfected successfully, voltage commands from -20 to -50 mV produced a classical gM relaxation (Brown and Adams, 1980) in response to the start of the voltage pulse and reactivation of the conductance when the membrane voltage was restored to -20 mV (Fig. 5c, inset). Under the conditions of our experiments KCNQ2/3 channel currents in tsA-201 or COS-1 cells displayed significant rundown during the initial 5 min of recording with ATP-free internal solution in the pipette. The presence of ATP in the intracellular pipette or metabolic substrates such as pyruvate or glucose in the extracellular solution are known to slow the rate of rundown of M-channels during whole-cell recording (Pfaffinger et al., 1988; Simmons and Schneider, 1998; Suh and Hille, 2002). This suggests that ATP may be required for lipid kinase activity (i.e., PtdIns 4-kinase, PtdInsP kinase) and thus for the maintenance of PIP2 levels in the membrane (Suh and Hille, 2002). If PIP2 is required to maintain M-channel function, application of this lipid via the patch pipette therefore would be expected to slow rundown. Figure 5c compares the time course of change of KCNQ2/3 channel current, measured from tail currents at -50 mV, in 16 control cells with that in 11 cells recorded with 20 μm PIP2 in the recording pipette. The current amplitude decreased to 45.2 ± 8.8% of the initial (n = 16) after 3 min of recording. In contrast, KCNQ2/3 channel current recorded with 20 μm PIP2 in the recording pipette decreased significantly less, to 83.3 ± 16.6% of the initial (n = 11; p < 0.05). In other experiments the I—V characteristics of expressed KCNQ2/3 channels were studied over a wide range by using 4.75 sec commands from -80 to +50 mV. In these experiments the current recorded at +50 mV decreased to 58.6 ± 6.2% of control (n = 27) after 5 min with normal intracellular solution. Inclusion of 20 μm PIP2 in the pipette significantly reduced rundown of these currents such that those recorded after 5 min were still 85.1 ± 10.8% of control (n = 12; p < 0.05). Typical experiments are illustrated in Figure 5, d and e. Thus 20 μm PIP2 is able to modulate KCNQ2/3 channels expressed in tsA-201 or COS-1 cells. This concentration was chosen because it is within the 8–50 μm range of concentrations used in other investigations of phospholipid modulation of ion channels in other systems (Zhang et al., 1999; Bian et al., 2001; Runnels et al., 2002; Rohacs et al., 2003).
Effect of PIP2 on M-currents in BFSG cells
To study the effect of PIP2 on native neurons, we examined its effect on the smaller C-cells of BFSG that express an agonist-sensitive gM (Jones, 1987) similar to that of the larger B-cells that were used above. We postulated that the smaller volume of the C-cell cytoplasm (Dodd and Horn, 1983; Kurenny et al., 1994) would favor the partition of PIP2 into their membranes. As with the experiments on KCNQ2/3 channels in COS-1 or tsA-201 cells, ATP was excluded from the intracellular solution when PIP2 was tested on BFSG neurons. Although 20 μm PIP2 slowed the rundown of KCNQ2/3 currents, this concentration failed to affect gM in BFSG C-cells. Table 2 shows that there was no significant change in gM in these cells between 30 sec and 4 min of recording either with or without 20 μm PIP2 in the recording pipette (Student's paired t test). Moreover, there was no significant difference in the amount of gM in the two groups of cells after 4 min of recording (Student's unpaired t test).
In contrast, the amplitude of gM recorded after the intracellular application of a high concentration of PIP2 (100 μm) was significantly larger after 4 min of recording than after 30 sec of recording (p < 0.02, using Student's paired t tests on data from each cell to compare current values at 30 sec and 4 min; Table 2). This effect of PIP2 is rather modest because there was no significant difference in the amount of gM in control cell population as compared with the 100 μm PIP2-treated population after 4 min of recording (Student's unpaired t test was used to compare all control current values, with all values measured in the presence of PIP2 at 4 min; Table 2). Figure 5f compares the recorded values of gM over the first 5 min of recording in cells with and without 100 μm PIP2. Although significant differences are seen at early time intervals, the effect of PIP2 on gM in BFSG is very small when compared with its effect on KCNQ2/3 channels (Fig. 5c).
Intracellular application of 100 μm PIP2 also did not interfere with the ability of P2Y agonists to suppress gM. Thus extracellular application of 250 μm ATP suppressed gM by 66.9 ± 6.2% (n = 11) in PIP2-treated cells, and this was no different from the amount of suppression seen in control cells 61.5 ± 9.2% (n = 9; p > 0.6). This lack of effect on agonist-induced responses may have reflected our inability to alter PIP2 concentration in the vicinity of the channels; had this occurred, the amplitude of the steady-state current may have been increased more convincingly (Table 2).
Investigation of limited effectiveness of PIP2 on BFSG neurons
The limited effectiveness of PIP2 in BFSG neurons could be attributable to degradation of exogenously applied PIP2 by endogenous lipid phosphatases (Zhang et al., 1999). To test this, we inhibited lipid phosphatases by using an internal solution containing (in mm): 5 sodium fluoride, 0.1 sodium orthovanadate, and 10 sodium pyrophosphate (FVPP solution) (Huang et al., 1998). Inclusion of a FVPP solution in the recording pipette, however, did not increase the effectiveness of PIP2 (Table 3).
Alternatively, the limited effectiveness of exogenously applied PIP2 may reflect saturation of binding sites on M-channels by endogenous PIP2. In an attempt to deplete endogenous PIP2, we applied (extracellular) ATP to activate P2Y receptors and to promote PLC-induced PIP2 hydrolysis. This was done in the presence of wortmannin to prevent PIP2 resynthesis before gaining whole-cell access to the cell. Cells were pretreated thus with 10 μm wortmannin for 5 min after which ATP was applied for 20 sec. At this time whole-cell access was obtained, and the IM recorded after 30 sec was compared with that recorded after 4 min. When cells were treated in this manner, the inclusion of PIP2 (20 μm) in the recording pipette did not alter gM (Table 3).
PIP2 (20 μm) also was included in the patch pipette after PLC had been inhibited with U73122 (5 min; 10 μm) to determine whether tonically active PLC was hydrolyzing the applied PIP2. Under these conditions, however, PIP2 still failed to increase gM (Table 3).
Thus the poor efficacy of exogenously applied PIP2 in modulating gM under the conditions of our experiments is unlikely to reflect the presence of endogenous phosphatases (Huang et al., 1998), saturation of possible binding sites on M-channels by endogenous PIP2, or ongoing catabolism of PIP2 by the action of PLC.
Effects of Al3+
Because negative charges on the phosphate moieties of phosphatidylinositides are involved in their binding to positively charged residues on K+ channels (Zhang et al., 1999), trivalent cations such as Al3+ strongly disrupt PIP2 channel interactions (Hilgemann and Ball, 1996). To test whether Al3+ affected M-channels, we studied KCNQ2/3 channels in inside-out macropatches excised from tsA-201 cells. Apart from leak, almost all current in such patches should be attributable to KCNQ2/3 current. We found that the total current at 0 mV was suppressed by 47.0 ± 6.7% by 50 μm Al3+ (n = 5). In the experiment illustrated in Figure 6, the addition of 1 mm ATP to the intracellular surface of the membrane so as to promote PIP2 synthesis (Suh and Hille, 2002) partly reversed the effect of Al3+.
PIP2 and suppression of gM by muscarine
The M-current was defined first as a muscarine-sensitive current in BFSG neurons (Brown and Adams, 1980). Suppression of gM by muscarinic agonists in rat superior cervical ganglion neurons is attenuated by 3 μm U73122, and this had led to the conclusion that PLC likely participates in the transduction mechanism for this agonist in mammalian sympathetic ganglia (Suh and Hille, 2002). Although the PLC inhibitor U73122 (10 μm) antagonized the effect of 5–10 μm muscarine on gM in BFSG B-neurons (Fig. 7a), this effect also was seen with the inactive isomer U73343 (Fig. 7b). The amount of muscarine-induced gM suppression seen in the presence of U73122 thus did not differ from the amount of suppression seen in the presence of the inactive isomer (p = 0.5; Table 4). Although at 3 μm a slightly more selective effect was seen with the U73122 as compared with the inactive isomer (p < 0.05; Table 4), it was not possible on the basis of these data to implicate PLC unequivocally in the action of muscarine on gM in BFSG neurons.
Despite this, the muscarine response was affected in much the same way as the ATP response by both wortmannin (Fig. 7c) and the PIP2 antiserum (Fig. 7d,e). Thus wortmannin slowed recovery of muscarine responses. Before the application of wortmannin the time for 50% recovery of responses to 2 μm muscarine was 23.0 ± 3.9 sec. This increased to 174 ± 72.9 sec after 10 min in wortmannin. After 5 min recording with the PIP2 antibody in the pipette, 2 μm muscarine suppressed gM by 87.0 ± 6.3% (n = 5). After 25 min with the antibody the muscarine became significantly less effective because it suppressed gM by only 64.0 ± 8.7% (n = 5; p < 0.002). In control experiments the muscarine suppressed gM by 82.0 ± 4.5% after 5 min with the intracellular horse serum and by 76.6 ± 3.0% after 25 min (n = 6; p > 0.3).
The effect of muscarine and P2Y agonists on gM therefore may proceed via similar mechanisms.
Discussion
The lipid kinase and PI-polyphosphate hypothesis
These experiments address the hypothesis that PLC-induced depletion of PIP2 is the signal for agonist-induced M-channel closure in BFSG neurons. This mechanism, termed the lipid kinase and PI-polyphosphate hypothesis, was suggested to explain gM inhibition by muscarinic agonists in rat sympathetic neurons (Suh and Hille, 2002; Zhang et al., 2003). Although there is now good evidence for its applicability to receptor-mediated inhibition of expressed KCNQ2/3 channels (Zhang et al., 2003), unequivocal verification of the lipid kinase and PI-polyphosphate hypothesis for agonist-induced gM suppression in an intact neuron is not yet available. The present observations therefore are discussed in relation to four testable predictions of the hypothesis. These are that (1) PLC is involved in agonist-induced gM suppression, (2) gM is increased by PIP2, (3) PIP2 removal promotes M-channel closure, and (4) PIP2 resynthesis is involved in the recovery of agonist-induced gM suppression.
Role of PLC
Although ATP responses are antagonized by the PLC inhibitor U73122 and not by the inactive isomer U73343 (Stemkowski et al., 2002), a preferential effect of U73122 on muscarine-induced gM inhibition was more difficult to demonstrate (Table 4). Certainly, at 3 μm the active isomer seemed more effective. This is consistent with a role for PLC in muscarine-induced gM suppression in BFSG neurons as has been suggested in rat neurons by Suh and Hille (2002). Because 10 μm U73343 failed to affect ATP responses (Stemkowski et al., 2002) yet completely blocked muscarine responses (Fig. 7b), this inactive isomer may interact with muscarinic receptors.
Effects of PIP2
Although 20 μm PIP2 increased KCNQ2/3 currents in tsA-201 or COS-1 cells (Fig. 5c—e), a somewhat higher concentration (100 μm) produced, at best, a weak effect on gM in BFSG neurons (Fig. 5f, Table 2). One of many possible explanations for this difference is the presence of more membranous organelles in native neurons as compared with tsA-201 or COS cells. These could serve as sinks for exogenously applied PIP2. It also should be noted that a majority of studies of phospholipid modulation of ion channels have been performed on expressed channels (Huang et al., 1998; Xie et al., 1999; Zhang et al., 1999; Leung et al., 2000; Bian et al., 2001; Runnels et al., 2002; Rohacs et al., 2003), and few reports (Fan and Makielski, 1997; Zhang et al., 2003) document modulation of native channels by PIP2. The experiments of Zhang et al. (2003), which studied the effects of PIP2 on putative M-channels in rat sympathetic neurons, were done with inside-out patches held at +10 mV. Similar experiments would be difficult in frog neurons because recordings at this voltage would be dominated by openings of maxi gK,Ca channels (Adams et al., 1982b). We therefore used whole-cell recordings to study the effects of PIP2. The poor efficacy of exogenously applied PIP2 under these conditions did not reflect the presence of endogenous phosphatases (Huang et al., 1998), ongoing catabolism of PIP2 by the action of PLC, or saturation of possible binding sites on M-channels by endogenous PIP2 (Table 3). We have noted, however, that the PLC inhibitor U73122 transiently enhances gM in BFSG cells (Stemkowski et al., 2002) (Fig. 7a). Because this effect is not seen with the inactive isomer U73343 (Fig. 7b), it may reflect PIP2 accumulation and increased M-channel activity after the inhibition of basal PLC activity.
PIP2 removal and M-channel closure
Zhang et al. (2003) reported that PIP2 antibodies inhibit KCNQ2/3 channels in inside-out oocyte macropatches. This observation is consistent with the hypothesis that PIP2 removal invokes channel closure. Our observation that PIP2 antibodies failed to affect steady-state gM in BFSG neurons was, therefore, unexpected. Despite this, the antibody consistently attenuated the effect of both ATP and muscarine. One explanation for this may be that the antibodies protected PIP2 from the action of PLC while preserving its interaction with M-channels. If this were so, our result would support the hypothesis that PIP2 hydrolysis is required for agonist action. It does not, however, show that PIP2 removal invokes channel closure.
PIP2 resynthesis and recovery of agonist-induced gM suppression
Slowing of recovery of both P2Y and muscarinic responses by wortmannin can be attributed to impairment of PIP2 resynthesis after the inhibition of PtdIns 4-kinase (Suh and Hille, 2002; Zhang et al., 2003). It does not reflect actions on PtdIns 3-kinase or myosin light chain kinase because inhibitors of these enzymes, LY294002 and ML7, do not mimic the effect of wortmannin. Similar mechanisms appear to operate in intact and perfused BFSG neurons because the effects of wortmannin were seen with both whole-cell and perforated-patch recording (Table 1).
Our findings differ from those of a previous study in which wortmannin had a direct inhibitory effect on gM in BFSG neurons that was independent of the presence of agonist (Tokimasa et al., 1995). Differences in the nucleotide content of the internal solutions used in the two studies may provide an explanation for this disparity. Whereas our internal solution contained 2 mm ATP and no GTP, that used by Tokimasa and colleagues contained less ATP (1.15 mm) and 1.5 mm GTP. Our solution with the relatively high ATP content would favor PIP2 synthesis (Suh and Hille, 2002), whereas the presence of GTP in the solution used by Tokimasa et al. (1995) may favor constitutive G-protein turnover, tonic activation of PLC, and hence a tendency toward PIP2 hydrolysis. The addition of wortmannin under these conditions would prevent PIP2 resynthesis, and this may explain the direct, agonist-independent action of wortmannin on gM observed by (Tokimasa et al., 1995). Wortmannin (10 μm) also affected steady-state gM in rat sympathetic neurons (Suh and Hille, 2002) to a greater extent than in BFSG. This may reflect species differences. For example, there may be greater pools of PIP2 available in frog neurons as compared with rat neurons. Alternatively, the presence of a low concentration of GTP (0.1 mm) in the pipette solutions used by Suh and Hille (2002) may have favored ongoing PLC activity.
Inhibition of the lipid cycle at the level of DAG kinase by the use of R59022 also slowed recovery of ATP responses in BFSG neurons (Fig. 3), but this effect developed more slowly than that of wortmannin (Fig. 2a). This may reflect depletion of PIP at a relatively remote site as compared with wortmannin such that more agonist-induced turns of the cycle were required to deplete the membrane concentrations of the intermediates phosphatidic acid, CDP-DAG, PI, and PIP (Fig. 1). A further observation that is consistent with the lipid kinase and PI polyphosphate hypothesis is the slowing of the recovery of agonist responses by nonhydrolyzable ATP analogs (Suh and Hille, 2002), an effect that also occurs in BFSG neurons (Chen and Smith, 1992). These substances may reduce the activity of PtdIns 4-kinase and PtdInsP kinase by impairment of ATP hydrolysis that is required for the formation of PIP2.
Transduction mechanism(s) for agonist-induced gM suppression
Experiments reported here and in the work of Suh and Hille (2002) implicate PLC in the onset of agonist-induced suppression of gM in neurons and PIP2 resynthesis in the termination of this effect. Although the observation that PIP2 can increase neuronal gM (Fig. 5f) (Zhang et al., 2003) is supportive, there is, as yet, little direct evidence that the PLC-induced removal of PIP2 from M-channels is the mechanism for agonist-induced inhibition. This situation in neurons contrasts with findings in expression systems for which Zhang et al. (2003) have presented several lines of evidence to support a role for PLC-induced removal of PIP2 from KCNQ2/3 channels as a mechanism for agonist-induced inhibition. Perhaps their most compelling observation is that mutated KCNQ channels, which display reduced sensitivity to PIP2, exhibit increased susceptibility to receptor-induced inhibition (Zhang et al., 2003). Unfortunately, manipulations of this type are not generally feasible in native neurons. Thus although there is good evidence to support the lipid kinase and PI polyphosphate hypothesis for the inhibition of expressed KCNQ2/3 channels, the physiological significance of the observations can be questioned. In contrast, the evidence from native neurons, where the physiological significance is clear, is indirect and less compelling.
In view of the clear demonstration of PLC-mediated PIP2 depletion in the mediation of muscarinic suppression of calcium-permeant TRPM7 channels in both cardiac cells and in an expression system (Runnels et al., 2002) and numerous reports of regulation of cation channel function by PIP2 (Fan and Makielski, 1997; Huang et al., 1998; Xie et al., 1999; Zhang et al., 1999; Leung et al., 2000; Bian et al., 2001; Wu et al., 2002; Rohacs et al., 2003), the hitherto elusive transduction mechanism for agonist-induced gM suppression, even in native neurons, may reflect a commonplace mechanism for channel modulation. Despite this, it is unclear whether PIP2 depletion alone is sufficient to explain the robust suppression of neuronal gM that is seen with all Gq-coupled agonists (Ikeda and Kammermeier, 2002). It is possible, for example, that the release of DAG, as a result of agonist-activation of PLC, may contribute to M-channel closure via a PKC-independent process (Chen et al., 1994). Moreover, an obligatory (Kirkwood et al., 1991; Selyanko and Brown, 1996) or permissive role of Ca2+ (Beech et al., 1991) has been discussed frequently. If a Ca2+-dependent isoform of PLC such as PLC-β (Haley et al., 2000) is involved in the PIP2 depletion mechanism, Ca2+ released via the InsP3 pathway may serve as a positive feedback mechanism during agonist-induced gM suppression. The possible importance of Ca2+ and DAG in this process thus requires further assessment.
Footnotes
- Received February 11, 2003.
- Revision received April 2, 2003.
- Accepted April 4, 2003.
This work was supported by the Canadian Institutes for Health Research (MOP 57798). C.P.F. received student-ship support from the Alberta Heritage Foundation for Medical Research and Neuroscience Canada. We thank Dr. David McKinnon for gifts of KCNQ2 and KCNQ3 cDNA; Dr. S. Seino for the KATP channel Kir6.2 subunit clone; Drs. L. Aguilar Bryan and J. Bryan for the SUR1 subunit clone; Diana Steckley, Lynn Eisner, and Kenneth Wong for technical assistance; Dr. Jocelyn Manning Fox for useful discussions and advice on experimental procedures; and Drs. Bill Colmers and Fred Tse for their comments on a previous version of this manuscript.
Correspondence should be addressed to Peter A. Smith, Department of Pharmacology, 9.75 Medical Sciences Building, University of Alberta, Edmonton, Alberta, Canada T6G 2H7. E-mail peter.a.smith{at}ualberta.ca.
Copyright © 2003 Society for Neuroscience 0270-6474/03/234931-11$15.00/0