Hydrolyzable ATP and PIP₂ Modulate the Small-conductance K⁺ Channel in Apical Membranes of Rat Cortical-Collecting Duct (CCD)

The cortical-collecting duct (CCD)^* is an important site for potassium secretion (Wang et al., 1997; Giebisch, 1999). The secretory process is confined to principal cells and initiated by active uptake of potassium across the basolateral membrane through the activity of Na-K-ATPase. This is followed by passive diffusion along a favorable electrochemical gradient across the apical membrane through potassium channels. A small-conductance potassium channel (SK) in the apical membrane of CCD is the major pathway for this potassium secretion (Wang et al., 1997). SK channels in principal cells of the mammalian nephron are regulated by pH, phosphorylation and dephosphorylation processes, cell messengers, and the cytoskeleton (Wang et al., 1992, 1997; Kubokawa et al., 1995b; Palmer, 1999; Schulte and Fakler, 2000).

A distinguishing feature of the SK channel is the modulation of activity by adenosine triphosphate (ATP) (Wang and Giebisch, 1991a), typical of other ATP-sensitive (K_ATP or Kir6.x) K⁺ channels (Ashcroft and Ashcroft, 1990; Yokoshiki et al., 1998; Ashcroft, 2000). Low concentrations of ATP in the presence of protein kinase A (PKA) often restore the activity of run-down SK channels in CCD, whereas higher concentrations of ATP inhibit channel activity (Wang and Giebisch, 1991a). The former is thought to involve several distinct phosphorylation processes, whereas the latter may be due to direct binding of ATP to the channel as in other K_ATP channels (Tanabe et al., 1999, 2000; Ashcroft, 2000; MacGregor et al., 2001). ADP also affects SK channel activity by antagonizing the inhibitory effect of ATP (Wang and Giebisch, 1991b).

Phosphatidylinositol phosphates (PIPs) like PIP₂ provide membrane signals (Martin, 1998; Corvera et al., 1999) that stimulate both K_ATP (Kir6.x) channels (Fan and Makielski, 1997; Baukrowitz et al., 1998; Huang et al., 1998; Shyng and Nichols, 1998; Shyng et al., 2000b) and other nonnucleotide-gated inward rectifier K⁺ channels (Huang et al., 1998; Zhang et al., 1999). The concentration of these PIPs in the membrane is regulated by phospholipases (like phospholipase C, PLC), phospholipid phosphatases, and phosphoinositide kinase (PI kinase). Hydrolyzable ATP and/or PIP₂ have been shown to reactivate run-down Kir6.2 channels and the effect of Mg-ATP, but not PIP₂, is abolished by wortmannin, a PI kinase inhibitor (Xie et al., 1999b). Since protein kinase inhibitors did not block the ability of Mg-ATP to reactivate these Kir6.2 channels, this suggests that the Mg-ATP–dependent channel recovery is mediated by membrane lipid kinases rather than protein kinases. However, another study has suggested that protein kinase–mediated phosphorylation of Kir6.2 maintains responsiveness to PIP₂ (Ribalet et al., 2000). A role for PI kinases in regulating K_ATP channel activity is also supported by the modulation of ATP sensitivity of K_ATP channels in COSm6 cells transfected with PI kinase (Shyng et al., 2000a). In addition, the hydrolysis of PIP₂ by phospholipases such as PLC leads to the formation of inositol triphosphate and diacyl glycerol and loss of the stimulatory effect of PIP₂ on Kir channels (Xie et al., 1999a; Kobrinsky et al., 2000).

ROMK, cloned from the outer medulla of rat kidney (Ho et al., 1993; Zhou et al., 1994), shares many properties with the native SK channel (Palmer et al., 1997; Wang et al., 1997). ROMK exists in three isoforms, ROMK1, 2, and 3 (Boim et al., 1995; Lee and Hebert, 1995). In situ hybridization has shown expression of ROMK1 and ROMK2 mRNA in principal cells (Lee and Hebert, 1995), and immunohistochemistry has identified ROMK in apical membranes of principal cells in CCD (Lee and Hebert, 1995; Kohda et al., 1998). Thus, ROMK is believed to encode the SK channel in apical membranes of principal cells. ROMK1 channels expressed in oocytes can be activated by PIP₂ (Huang et al., 1998) and this effect is modulated by PKA-dependent phosphorylation of ROMK (Liou et al., 1999), as well as by changes in cytosolic-side pH (Leung et al., 2000).

Although the studies of PIP₂ effects on ROMK1 channels suggest a potential role for PIPs in regulating the native SK channel, there is currently no information about the effects of these phospholipids or PI kinases on SK channel activity in the mammalian CCD. In the present study we investigated the possible roles of ATP, PI kinase, PIP₂, and PKA in modulating SK channel activity in principal cells from rat kidney. We found that hydrolyzable ATP can restore the low SK channel activity that follows excision of patches through a mechanism that involves membrane-bound kinases. Full channel activation requires both PI kinases increasing PIPs (like PIP₂) and PKA phosphorylation of the SK channels.

Sprague-Dawley rats (80–100 g) were purchased from Taconic Farms, Inc. and kept on normal K-containing diet for 7–10 d before being killed. Left kidneys were removed and cut into two coronary slices for manual dissection. CCD were isolated using forceps under microscopic view in ice-cold bath solution (∼4°C) as previously described (Wang and Giebisch, 1991b). One or two single CCD segments were placed on Cell-Tak (Becton Dickinson) –coated cover glass (2.5 × 2.5 mm) to immobilize them, and transferred to the experimental chamber mounted on the stage of an inverted microscope (IMT-2; Olympus). CCD tubules were opened with a glass pipette to expose the lumen surface for patch-clamping of the apical membrane. Only principal cells, identified by their large flat surface and hexagonal shape, were chosen and all experiments were conducted at room temperature (22–24°C). The time between sacrificing the animal to opening the tubules was 5–10 min.

Control bath solutions contained (mM): 140 NaCl, 5 KCl, 1.8 MgCl₂, 1.8 CaCl₂, and 10 HEPES. Ca²⁺-free bath solution was the same except that it also contained (mM): 0 CaCl₂, 0.8 MgCl₂, and 1 EGTA. Ca²⁺/Mg²⁺-free bath solution contained (mM): 140 NaCl, 5 KCl, 1 EGTA, 1 EDTA, and 10 HEPES. All solutions (pH adjusted to 7.4 with NaOH) were kept in the refrigerator and used within 3 d. ATP-containing bath solutions were freshly prepared every day. Pipette solution contained (in mM): 140 KCl, 1.8 MgCl₂, and 10 HEPES (pH adjusted to 7.4 with KOH). Adenosine 5′-triphosphate (magnesium salt, Mg-ATP), adenosine 5′-[γ-thio]triphosphate (tetralithium salt, ATPγS), 5′-adenylylimido-diphosphate (AMP-PNP), wortmannin, 2-deoxy-D-glucose, antimycin A, and DL-dithiothreitol (DTT) were purchased from Sigma-Aldrich. Alkaline phosphatase from calf intestine (cAP) was purchased from New England Biolabs, Inc. LY294002, protein kinase A inhibitor (PKI), and quercetin were purchased from Calbiochem-Novabiochem. L-α-phosphatidylinositol-4, 5-bisphosphate (PIP₂) was purchased from Roche Diagnostics Corp. Phosphatidylinositol 4,5-bisphosphate antiserum (anti-PIP₂) was purchased from Assay Designs, Inc. Stock solutions of PIP₂ were sonicated in a cold bath for 30 min and kept in the freezer (−70°C). After sonicating for 1 min, PIP₂ was added directly to the perfusion chamber by hand using a small pipette. All chemical solutions were freshly made before each experiment. Agents were dissolved in either DMSO or ethanol with final concentrations in bath solutions of <0.1%. These solvents had no effects on SK channel activity.

Microelectrode pipettes were pulled from borosilicate glass capillaries (Dagan Corp.) using a two-step Narishige PP83 puller (Narishige Scientific Instrument Laboratory) and polished using a Narishige microforge (MF-83). The pipette resistances were 6–8 MΩ after filling with pipette solution. Single channel recordings were obtained using an EPC-7 amplifier (List Electronics) and filtered at 1 kHz by a Bessel filter (902 LPF; Frequency Devices, Inc.). Currents were digitized at a sampling rate of 4 kHz using a Digidata 1200 interface (Axon Instruments, Inc.) and stored in a PC (E-3100, Gateway 2000) for further statistical analysis and printing. NP_o represents the channel activity, in which N is the number of channels in the patch and P_o is the single channel open probability, calculated at a modified filter frequency of 500 Hz by using pCLAMP software (version 6.0.4 of Fetchan and pSTAT; Axon Instruments, Inc.). The changes of channel activity were determined when a steady-state was reached and calculated over a 30s interval before and after changing bath solutions or additions of agents. Statistical data are shown as mean ± SEM and P <0.05 was considered significant. Mean paired differences were calculated by Student's t test.

Fig. 1 shows a representative single channel recording upon changing from a cell-attached to an inside-out patch configuration on the apical membrane of a principal cell. With the cell-attached configuration and Ringer's solution in the bath, channel conductance was 38.3 ± 1.2 pS (calculated at pipette holding potential [−V_pipette] between 0 and −80 mV). The I-V curve showed slight inward rectification (35.4 ± 0.8 pS, −V_pipette between 20 and −20 mV; 31.2 ± 1.5 pS, −V_pipette between 40 and 0 mV, n = 13). In the inside-out patch configuration with symmetrical 140 mM KCl solutions, the channel conductance was 31.3 ± 0.6 pS (−V_pipette between 20 and −80 mV, n = 4). With bath solution containing 140 mM NaCl and 5 mM KCl, the channel showed inward rectification with a conductance of 26.0 ± 0.3 pS, −V_pipette between 0 and −80 mV (n = 6). The latter conductance is typical of the SK conductance under these conditions (Frindt and Palmer, 1989; Wang et al., 1990). In all remaining experiments, inside-out patches were exposed to 140 mM NaCl and 5 mM KCl bath solutions. Channel kinetics calculated at a pipette holding potential (−V_pipette) of −40 mV were typical of the SK channels in mouse (Lu et al., 2000) and rat (Wang and Giebisch, 1991a,b): one open time (T_o = 37.6 ± 0.2 ms) and one closed time (T_c = 1.4 ± 0.3 ms); open probability (P_o) was 0.89 ± 0.17. Consistent with previous observations (Wang and Giebisch, 1991b; Kubokawa et al., 1995a), channel activity progressively decreased in inside-out patches and channel activity disappeared completely within 1.4 ± 0.4 min (n = 21; Fig. 1).

The concentration of ATP on the cytosolic face of principal cells plays an important role in modulating SK channel activity: in excised patches sub-mM concentrations of ATP can reactivate run-down SK channels whereas mM concentrations of Mg-ATP reduce channel activity (Wang and Giebisch, 1991a). Antimycin A, an inhibitor of mitochondrial electron transfer, and 2-deoxy-D-glucose, a glycolysis inhibitor, deplete cellular ATP stores in cultured renal epithelial cells (Mandel et al., 1994). We used these metabolic inhibitors to assess effects of depletion of intracellular ATP on SK channel activity. Fig. 2 is a representative recording from a cell-attached patch showing that SK channel activity decreased following exposure to 20 μM antimycin A and 20 mM 2-deoxy-D-glucose. NP_o decreased from a control value of 4.7 ± 0.5 to 0.9 ± 0.3 (n = 3) within 7 min. The inhibitory effect of antimycin A + 2-deoxy-D-glucose was partially reversible. Thus, ATP depletion in principal cells reduces SK channel activity.

To investigate further the effects of ATP on SK channel activity, we assessed the ability of Mg-ATP or ATP analogs (ATPγS or ANP-PNP) on reactivation of run-down SK channels. ATPγS has been reported to be a substrate for some protein kinases (Eckstein, 1985; Jeong and Nikiforov, 1999) but not for lipid kinases in oocytes (Liou et al., 1999) while AMP-PNP is a nonhydrolyzable ATP analogue. Fig. 3 shows the effects of Mg-ATP on SK channel run-down. Fig. 3 A shows the decline of channel activity following exposure to Ca²⁺/Mg²⁺-free and Mg-ATP-free bath solutions. Additions of 0.5 mM Mg-ATP (n = 6) reactivated SK channels following run-down. Fig. 3 B shows that ATPγS (n = 11) and AMP-PNP (unpublished data; n = 3) failed to maintain channel activity, whereas addition of 0.5 mM Mg-ATP reactivated channels after run-down in each condition. To provide additional evidence that the activating effect of Mg-ATP was due to phosphorylation processes, we exposed excised patches to calf intestinal alkaline phosphatase (cAP, a broad-spectrum phosphatase) in the presence of 0.5 mM Mg-ATP. Fig. 4 shows a representative recording demonstrating that cAP abolished SK channel activity even in the presence of Mg-ATP (from an NP_o = 3.3 ± 0.3 to an NP_o = 0.2 ± 0.1; n = 16). This effect of cAP was reversible (NP_o = 2.9 ± 0.6; 88 ± 11% of control after 5 min of cAP wash-out), likely due to both reestablishment of PKA-dependent phosphorylation of proteins (including the SK channel) and lipid kinase production of PIP₂.

In rat principal cells, phosphorylation processes have been associated with both channel activation (PKA) and inhibition (PKC or Ca²⁺-dependent calmodulin-activated kinase II [CaMK II]). Increasing cytosolic Ca²⁺ reduces SK channel activity in cell-attached patches by activating PKC (Wang and Giebisch, 1991b; Wang et al., 1993) and/or CaMK II (Kubokawa et al., 1995a,b). However, addition of bath Ca²⁺ in excised patches did not reduce channel activity in the absence of exogenous PKC (Wang et al., 1990). Moreover, Mg²⁺ can activate membrane-bound phosphatases and reduce SK channel activity in excised patches (Kubokawa et al., 1995a). To test whether Ca²⁺ and/or Mg²⁺ affect channel run-down after patch excision, experiments were performed in bath solutions lacking Ca²⁺ or both Ca²⁺ and Mg²⁺. There was no difference in time to run-down between experiments using Ca²⁺-free (Mg²⁺-containing) solutions or Ca²⁺/Mg²⁺-free solutions (1.8 ± 0.3 min [n = 6] vs. 1.8 ± 0.6 min [n = 5], respectively). These times to run-down were also not different from the time of run-down in control Ca²⁺/Mg²⁺-containing solutions (Figs. 1 and 3 A). Thus, these kinase inhibitors appear to play no significant role in our present experiments with rats fed a normal K diet. Although in a previous study in rats fed a high K diet, the time to run-down was significantly prolonged in the absence of Ca²⁺ and Mg²⁺ (Kubokawa et al., 1995a), it is not clear whether this difference was due to K in the diet or to other undetermined factors.

In further experiments, we examined the role of PI kinases in mediating the Mg-ATP–dependent phosphorylation effects on maintaining SK channel activity (Figs. 5–7). Two experimental approaches were used to test whether PIP₂ affects SK channel activity. First, we investigated the effects of PIP₂ applied to the cytosolic face of excised membrane patches in which activity had declined after exposure to solutions free of Mg-ATP and Ca²⁺. As shown in Fig. 5 A, a significant increase of NP_o was produced by addition of 50 μM PIP₂ (NP_o from 0.9 ± 0.3 to 3.7 ± 0.7; n = 8). This result is similar to the increase in the activity of ROMK1 produced by PIP₂ in oocytes (Huang et al., 1998).

Second, we also investigated the effect of PIP₂ antibodies (anti-PIP₂). As shown in Fig. 5 B, the addition of anti-PIP₂ to the cytosolic face of excised membrane patches led to a prompt decline of SK channel activity in the presence of Mg-ATP–containing bath solution. NP_o decreased from 5.6 ± 1.2 to 0.1 ± 0.1 (n = 10). Fig. 5 C shows that the inhibition of channels could not be reversed by repeated wash-out with anti-PIP₂–free and Mg-ATP–containing/Ca²⁺-free bath solutions. However, adding DL-dithiothreitol (DTT, 100 μM) to the bath restored channel activity to 88 ± 9% of control values (NP_o = 4.9 ± 0.9, n = 5). These results are consistent with those of Huang and coworkers in ROMK1 expressed in oocytes showing that the reversal of the PIP₂ antibody effect also required exposure to reducing agents like DTT (Liou et al., 1999). This effect of DTT may be due reduction of bonds in the anti-PIP₂ immunoglobulin molecule (Liou et al., 1999). In another study, the low pH–inhibited ROMK1 channels expressed in oocytes required reducing agents for reactivation after returning pH to control (Schulte et al., 1998).

In further experiments we investigated the relationship between hydrolyzable ATP and PI kinases. The effect of LY294002 on SK channel activity is shown in Fig. 6. Although Fig. 6 A shows that LY294002 abolished SK channel activity in this patch, this was not always the case (Fig. 6 B). On average, SK channel activity was reversibly diminished to 17 ± 3% of control by 20 μM LY294002, a selective inhibitor of PI kinase (NP_o decreased from 4.3 ± 0.6 to 0.7 ± 0.3 and recovered to 3.2 ± 0.6; n = 14; Figs. 6 A and 7). This inhibition was also rapidly reversed in the continued presence of LY294002 by addition of 50 μM PIP₂ (Fig. 6 B; NP_o decreased from 3.1 ± 0.1 to 0.4 ± 0.1 and then increased to 2.6 ± 0.5 after addition of PIP₂; n = 7).

Experiments were also performed using two additional PI kinases inhibitors: quercetin (60 μM), a nonselective inhibitor of PI kinase, and wortmannin (20 μM), a specific covalent inhibitor of PI kinase. Results of experiments using these two PI kinase inhibitors are summarized in Fig. 7. NP_o was reduced to 37 ± 4% of control by quercetin (n = 5) or 25 ± 7% of control by wortmannin (n = 6). The average reduction by all PI kinase inhibitors was to 26 ± 6% of control, a value significantly different from zero. Thus, none of the PI kinase inhibitors completely abolished channel activity. PIP₂ reversed channel activity in the presence of quercetin and wortmannin (Fig. 7). Similar reactivation was also observed by addition of PIP₃ (20μM, n = 3; unpublished data).

In Xenopus oocytes, PKA phosphorylation of ROMK has several effects that enhance channel activity. Phosphorylation of specific sites on ROMK modulates either P_o or channel density, NP_o (MacGregor et al., 1998). In addition, the stimulation of channel activity by PIP₂ is enhanced by PKA phosphorylation of ROMK1 (Liou et al., 1999). To investigate the interaction of PKA and PIP₂ in the native SK channel, two approaches were used.

Fig. 8 A summarizes the effect of blocking PKA activity by protein kinase A inhibitor (PKI) in the presence of 0.5 mM Mg-ATP. In control experiments NP_o was maintained over 10 min after patch excision in the absence of PKI (Fig. 8 A, solid circles). SK channel activity declined to 64 ± 13% of control (Fig. 8 A, open circles) with 20 μM PKI in the presence of Mg-ATP. This 36% reduction in SK channel activity is consistent with our previous observations of the activating effect of PKA phosphorylation on SK channel activity (Wang and Giebisch, 1991a,b). In the presence of PKI (Fig. 8 A), Mg-ATP would be expected to maintain the concentration of PIPs in the membrane. Thus, PKA phosphorylation appears to be necessary for full activation of SK channels even in the presence of PIPs.

To further investigate the interplay of PKA phosphorylation processes and PIPs, we took advantage of the ability of cAP to abolish channel activity in the presence of Mg-ATP, as shown in Fig. 4. Fig. 8 B shows the run-down of channel activity by cAP in the presence of both 0.5 mM Mg-ATP and 20 μM PKI. Following removal of cAP, but in the continued presence of Mg-ATP and PKI, channel activity was only partially restored by endogenous production of PIP₂ even after prolonged observation (∼10 min; 68 ± 7% recovery; n = 7). These observations support that PKA phosphorylation is necessary for full PIP₂ responsiveness, and are consistent with observations in oocytes that PKA phosphorylation reduced the concentration of PIP₂ necessary for activation of ROMK1 channels (Liou et al., 1999). A similar conclusion regarding the necessity for protein phosphorylation of Kir6.2 in maintaining responsiveness to PIP₂ had also been reached (Ribalet et al., 2000).

To begin to understand the relevance of ATP–PIP₂ interactions in regulating SK channel activity, we examined the effects of PIP₂ in modulating both extracellular and intracellular ATP-mediated channel inhibition. We showed previously that extracellular ATP could reduce SK channel activity in mouse principal cells by stimulated purinergic (P₂Y₂) receptors (Lu et al., 2000). This effect of extracellular ATP on SK channel activity was due to enhanced phosphatase activity as it could be prevented by inhibiting phosphatases by okadaic acid, activating PKA with cyclic AMP, or blocking protein kinase G with KT5823.

In the representative experiments shown in Fig. 9, cell-attached apical membrane patches were obtained, extracellular 0.2 mM Mg-ATP added to the bath, and then the patch was excised in the inside-out configuration. Extracellular Mg-ATP inhibited SK channel activity in 1–3 min in cell-attached patches, confirming our previous observations in the mouse (Lu et al., 2000). As expected, no SK channel activity was restored after the patch was excised into Ca²⁺-free and Mg-ATP–free bath solution (Fig. 9 A). However, patch excision in a bath containing 0.5 mM Mg-ATP (Fig. 9 B) led to 48 ± 7% recovery of SK channel activity in 10 of 12 patches: NP_o changed from 6.68 ± 0.92 (cell-attached) to 3.17 ± 0.68 (inside-out). If PIP₂ (50 μM) alone was present in the bath upon patch excision, SK channel activity could be restored to 56 ± 9% of control: NP_o changed from 9.79 ± 1.25 (cell-attached) to 5.45 ± 0.82 (inside-out); n = 3 (unpublished data). Further addition of 50 U of PKA to the 0.5 mM Mg-ATP-containing bath restored channel activity to 81 ± 10% of that in the cell-attached configuration: NP_o changed from 6.53 ± 2.14 (cell-attached) to 5.28 ± 1.30 (inside-out); n = 4; Fig. 9 B. Thus, channel recovery requires both PIP₂ and PKA-dependent phosphorylation processes.

Since the intracellular ATP concentration is in the range of 2–4 mM in rat principal cells and 1 mM Mg-ATP in the bath inhibits SK channel activity in inside-out patches (Wang and Giebisch, 1991a), mechanisms must exist in intact cells to maintain K⁺ secretion through SK channels. One possibility is PIP₂ competition of Mg-ATP binding to the SK channel. PIP₂ has been shown to modulate the ATP sensitivity of the pancreatic β-cell K_ATP channel, Kir6.2 (Shyng and Nichols, 1998; Enkvetchakul et al., 2000; Loussouarn et al., 2001) and to directly compete with ATP for binding to the COOH-termini of Kir1.1 (ROMK) and Kir6.1 channels (MacGregor et al., 2001).

To examine the interactions of Mg-ATP and PIP₂ in rat principal cells, we assessed the Mg-ATP concentration dependence of SK channel activity in the presence of 50 μM PIP₂ in excised patches (Fig. 10). SK channel activity was completely inhibited by 1 mM Mg-ATP in the absence of added PIP₂ (Fig. 10 B, solid circle), confirming our previous observation in rat SK channels in principal cells (Wang and Giebisch, 1991a). Mg-ATP sensitivity is considerably lower in ROMK channels expressed in X. laevis oocytes (EC₅₀ ∼2–3 mM; McNicholas et al., 1996), but similar to native SK channels when ROMK is coexpressed with CFTR (Ruknudin et al., 1998). A representative experiment showing the concentration dependence of Mg-ATP inhibition in the presence of PIP₂ is shown in Fig. 10 A. The patch was excised into a Ca²⁺-free and Mg-ATP–free bath solution containing 50 μM PIP₂ that maintained SK channel activity for 2 min. Addition of 1 mM Mg-ATP had almost no effect on SK channel activity: control NP_o 3.12 ± 0.66 versus 2.90 ± 0.62 (n = 9). Increasing Mg-ATP led to a concentration-dependent inhibition of SK channel activity with 3 mM Mg-ATP abolishing activity (Fig. 10 B). PIP₂ reduced the sensitivity of SK channels to Mg-ATP: EC₅₀ for Mg-ATP was 1.80 ± 0.02 mM. Thus, PIP₂ reduced the Mg-ATP affinity of SK channels to a value close to that of the normal cell ATP concentration.

The present study provides evidence that both PI kinase and PKA play important roles in the regulation of the native SK channels in principal cells from rat kidney. Several lines of evidence support this conclusion. First, hydrolyzable ATP (Mg-ATP) is required to maintain channel activity or to reactivate channels after run-down through the endogenous PKA and PIP₂ (Figs. 3 and 4). This effect of Mg-ATP depends on phosphorylation reactions since exposure of SK channels to alkaline phosphatase in the presence of Mg-ATP nearly abolished channel activity. Second, stimulation of channel activity by Mg-ATP is mediated, at least in part, by generation of PIP₂ through endogenous PI kinases. This is supported by our observations that declining channel activity after excision of membrane patches is mitigated by directly applied PIP₂, that PIP₂ antibodies inhibit channel activity in the presence of Mg-ATP (Fig. 5, A and B), and that Mg-ATP or PIP₂ partially reactivated SK channels inhibited by extracellular Mg-ATP (Fig. 9). It should be noted in this regard that ATPγS does not interact with PI kinases to generate PIP₂ (Liou et al., 1999). This explains our observation that ATPγS does not sustain channel activity (Fig. 3 B). Moreover, inhibition of PI kinases reduced SK channel activity and attenuated Mg-ATP effects on maintenance of channel activity (Figs. 6 and 7). Third, processes involving PKA phosphorylation provide an additional mechanism for modulating SK channel activity by Mg-ATP. Maximal PIP₂ effects require PKA phosphorylation (Figs. 8 and 9 B).

PI kinases are involved in the synthesis of PIP₂ and several isoforms have been identified (Anderson et al., 1999). Together with membrane phosphatases and phospholipases, PI kinases are regulated by several hormones and growth factors (Martin, 1998; Leevers et al., 1999). PIP₂ has been shown to be an important regulator of a number of ion channels and transporters (Hilgemann et al., 2001). Our present results add the SK channel in principal cells to this growing list of PIP₂-regulated ion channels. However, the cellular distribution of PI kinases between membranes and the cytosol, and their regulation in principal cells has not been explored.

PIPs have thus emerged as important regulators of inward rectifier K⁺ channels (Hilgemann, 1997; Huang et al., 1998; Hilgemann et al., 2001), including ROMK. The generation of PIP₂ by ATP-dependent PI kinases has been shown to activate ROMK1, and this effect is enhanced by PKA phosphorylation of the channel protein (Liou et al., 1999). Our observations in principal cells support this view by demonstrating that PIP₂ activates the SK channel and that PKA phosphorylation processes are required to give full SK channel activation by PIPs (Figs. 7–9).

Given that ATPγS is a substrate for PKA phosphorylation of ROMK1 channels (Jeong and Nikiforov, 1999; Liou et al., 1999), the inability of ATPγS to maintain channel activity in the absence of Mg-ATP indicates that PKA phosphorylation of SK is insufficient to sustain channel activity in the absence of PIP₂. In addition, PKI reduced channel activity in the presence of Mg-ATP (Fig. 8 A) where PI kinases would be active and generating PIP₂. This latter observation is consistent with the modulation of PIP₂ sensitivity of ROMK1 by PKA in oocytes (Liou et al., 1999). Moreover, PIP₂ antibodies abolished channel activity in the presence of a concentration of Mg-ATP sufficient to sustain PKA phosphorylation processes. Furthermore, after dephosphorylation by cAP, Mg-ATP did not fully reactivate channel activity in the presence of PKI. Together, our observations on SK channels strongly support the notions that PIP₂ can reactivate channels in the absence of PKA, but the sensitivity of PIP₂-mediated reactivation is modulated by PKA phosphorylation. Thus, our observations in SK channels are fully consistent with the interactions of PIP₂ and PKA in the regulation of ROMK1 expressed in oocytes. However, we were unable to consistently fully inhibit channel activity by any one of several PI kinase inhibitors. It is possible that the inhibitor concentrations were insufficient to completely abolish PI kinase activity or that insufficient phosphatase or phospholipase activity was present so that a low level of PIPs remained in the membrane even in the presence of the PI kinase inhibitors.

Although the physiological actions of PIPs in the regulation of SK channel activity are incompletely understood, we have shown two circumstances where PIPs can modulate Mg-ATP inhibition. First, extracellular Mg-ATP can inhibit SK channels in intact principal cells by activation of P₂Y₂ receptors in mouse (Lu et al., 2000) and rat (Fig. 9). This appears to be due to P₂Y₂-mediated activation of phosphatases (Lu et al., 2000). The partial reactivation of the purinergic receptor-inhibited SK channels by bath Mg-ATP in excised patches (Fig. 9 B) is likely due to the action of PI kinase, since a similar degree of reactivation was observed with PIP₂ in the absence of Mg-ATP. These latter observations would be consistent with P₂Y₂-activated phosphatases reducing the concentration of PIPs in the plasma membrane. The further reactivation of these channels by addition of PKA suggests that the P₂Y₂-activated phosphatases also dephosphorylate ROMK, which is then rephosphorylated by addition of PKA. Furthermore, increase in SK channel activity by PKA is consistent with the necessity of PKA phosphorylation process for full activity of PIP₂.

Second, while low concentrations of ATP (<1 mM; K_d ≈ 0.5 mM) are important for channel activation by phosphorylation processes (via PI kinase, PKA, and potentially other kinases), SK channels are inhibited by higher concentrations of ATP (>1 mM; Wang and Giebisch, 1991a). The activity of SK channels in vivo and in intact cells in vitro is clearly maintained under conditions in which the cytosolic ATP is 2–4 mM, a concentration higher than that known to block SK channels in excised membrane patches. The data presented in Fig. 10 demonstrate that PIP₂ shifts the ATP sensitivity in SK channels to the right and thereby help to maintain native channel activity at the relatively high cell ATP concentrations. PIPs also modulate the ATP sensitivity of the pancreatic β-cell K_ATP channel, Kir6.2 (Shyng and Nichols, 1998; Enkvetchakul et al., 2000; Loussouarn et al., 2001). Recently, we have shown that PIPs compete with ATP for binding to the carboxy terminus of ROMK and Kir6.1 channels (MacGregor et al., 2001) and this could provide a mechanism for PIPs maintaining SK channels in the presence of high cytosolic concentrations of ATP. Finally, given the recent observations by Marx and coworkers (Marx et al., 2002) of a macromolecular signaling complex for the sympathetic regulation of outward K⁺ current in heart, it is possible that all of the proteins regulating SK channel activity (PKA, lipid kinases, phosphatases) are similarly compartmentalized into a macromolecular complex.

The authors thank Dr. Chou-Long Huang for constructive suggestions.

This work was supported by National Institutes of Health grants DK54999 (S.C. Herbert) and DK54998 (G. Giebisch).