Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

K_ATP channels are heteromultimers formed by association of an inwardly rectifying K⁺ channel, Kir6.2/Kir6.1, and an ATP binding cassette protein, the sulfonylurea receptor SUR1/SUR2 (Inagaki et al., 1995; Clement et al., 1997; Babenko et al., 1998). K_ATP channels show complex regulation by intracellular factors such as ATP and ADP and pharmacological compounds such as sulfonylureas and K⁺ channel openers. These agents interact with the channel at different subunits and alter channel activity by different mechanisms (Ashcroft, 1988; Babenko et al., 1998; Seino, 1999; Baukrowitz and Fakler, 2000b).

ATP and ADP have opposing effects on K_ATP channel activity. ATP interacts with the Kir6.2 subunit and inhibits channel activity. The exact mechanism for ATP inhibition and the location of the ATP binding site are presently unknown. However, several studies indicate that ATP may reduce the open probability of K_ATP channels by stabilizing a closed channel state (Alekseev et al., 1998; Trapp et al., 1998; Enkvetchakul et al., 2000). Mutagenesis work has uncovered three regions in the cytoplasmic N and C termini of Kir6.2 that are possibly involved in the binding of ATP (Tucker et al., 1997; Drain et al., 1998; Trapp et al., 1998). MgADP antagonizes ATP inhibition and thereby effectively activates K_ATP channels under physiological concentrations of ATP. The action of MgADP is mediated by the nucleotide binding domains of the SUR subunit (Nichols et al., 1996; Gribble et al., 1997; Shyng et al., 1997). The opposing effects of ATP and ADP on channel activity make the K_ATP channel a metabolic sensor that couples membrane excitability to the cellular metabolic state reflected by the ATP/ADP ratio. For this reason K_ATP channels are involved in many physiological functions such as insulin secretion (Ashcroft, 1988) and protection of cardiac myocytes during periods of metabolic impairment (e.g., ischemia) (Nichols and Lederer, 1991). Sulfonylureas (e.g., glibenclamide and tolbutamide) inhibit K_ATP channels and are used in treatment of diabetes because they promote secretion of insulin in pancreatic -cells (Ashcroft, 1988; Edwards and Weston, 1993). These drugs bind to the SUR (Ashfield et al., 1999) and reduce channel open probability by a poorly understood mechanism. K⁺ channel openers (KCOs) are a chemical divers group of drugs that bind to the SUR subunits (Uhde et al., 1999) and activate K_ATP channels. The KCOs cromakalim and P1075 act on cardiac K_ATP channels, whereas diazoxide is specific for pancreatic K_ATP channels (Edwards and Weston, 1993). The activating effect of these drugs results from their ability to antagonize inhibition of intracellular ATP by a mechanism that involves ATP binding, and, probably, hydrolysis at the nucleotide binding domains of the SUR (Gribble et al., 1997; Shyng et al., 1997; Schwanstecher et al., 1998).

Recent work has uncovered a new class of regulatory molecules for K_ATP channel gating (Hilgemann, 1997; Baukrowitz and Fakler, 2000b). Membrane phosphatidylinositol phosphates (PIPs) such as PIP₂ and PIP were found to interact with K_ATP channels, resulting in increased open probability (Furukawa et al., 1996; Hilgemann and Ball, 1996; Fan and Makielski, 1997) and profoundly reduced ATP sensitivity (Baukrowitz et al., 1998; Shyng and Nichols, 1998; Fan and Makielski, 1999). The effect of PIPs on ATP inhibition seems to involve electrostatic interactions because the negatively charged phosphate groups at the inositol ring seem to be critical for the phospholipid effect on ATP inhibition. Highly negatively charged phosphatidylinositol-3,4,5-trisphosphate and PIP₂ are very effective in reducing ATP inhibition, whereas PIP is less potent and PI has no effect (Baukrowitz et al., 1998; Shyng and Nichols, 1998). Moreover, polycations such as polylysine and neomycin, which are known to bind to phospholipids and neutralize their negative charges, abolish the effect of PIPs on ATP inhibition (Deutsch et al., 1994; Shyng and Nichols, 1998). Furthermore, PIP₂ diminishes the potency of the KCO diazoxide to antagonize ATP inhibition (Baukrowitz et al., 1998; Koster et al., 1999) and reduces inhibition of K_ATP channels by tolbutamide (Koster et al., 1999).

Most studies on the modulation of K_ATP by PIPs have focused on open probability and ATP sensitivity, whereas little is known about the effects of PIPs on the channel's pharmacological properties. To gain further insight into the interference of PIPs with the pharmacological properties of K_ATP channels we address in this study the following questions: Do PIPs contribute to the variability reported for K_ATP channels in respect to inhibition by sulfonylureas? What are the structural features critical for PIPs to modulate the pharmacological properties of K_ATP channels? Do PIPs modulate the pharmacological properties of K_ATP channels and ATP-sensitivity by the same mechanisms?

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Mutagenesis and Expression of K_ATP Channels. Murine K_ir6.2 and rat SUR2A (Inagaki et al., 1995; Inagaki et al., 1996) were used in this study. Site-directed mutagenesis of Kir6.2 (K185Q, R176A) was carried out as described previously (Baukrowitz et al., 1999). For oocytes expression, constructs were subcloned into the pBF expression vector, which provides the 5'- and 3'-untranslated regions of the Xenopus laevis globin gene (Baukrowitz et al., 1999). Capped cRNAs specific for K_ir6.2 and SUR2A were synthesized in vitro using SP6 polymerase (Promega, Heidelberg, Germany) and stored in stock solutions at 70°C.

Electrophysiology. Giant patch recordings (Baukrowitz et al., 1999) in inside-out configuration under voltage-clamp conditions were made at room temperature (approximately 23°C) 3 to 7 days after injection. Polylysine [poly(L-lysine), hydrobromide, M_r 30,000-70,000], neomycin, heparin, ATP, glibenclamide, and cromakalim were purchased from Sigma and P1075 was kindly provided by Dr. U. Quast (Department of Pharmacology, University of Tübingen, Tübingen, Germany). Pipettes used were made from thick-walled borosilicate glass, had resistances of 0.2 to 0.4 M (tip diameter of 20-30 µm) and were filled with 120 mM KCl, 10 mM HEPES, and 1.8 mM CaCl₂ (pH adjusted to 7.3 with KOH). Currents were recorded with an EPC9 amplifier (HEKA Electronics, Lamprecht, Germany) and sampled at 1 kHz with analog filter set to 3 kHz (3 dB). Solutions were applied to the cytoplasmic side of excised patches via a multibarrel pipette and had the following composition (K_int): 100 mM KCl, 10 mM HEPES, 2 mM K₂EGTA (total K⁺ concentration was 120 mM, pH adjusted to 7.3 with KOH). Solutions with P1075, cromakalim, and glibenclamide had 1.4 mM MgCl₂. Computational work was done on Macintosh PowerPC 7600/132 Mhz using commercial software (IGOR; WaveMetrics, Lake Oswego, OR) and Excel 98 for the Macintosh (Microsoft, Redmond, WA).

Preparation of Lipid Solutions. L--Phosphatidyl-D-myo-inositol-4,5-bisphosphate [PI(4,5)P₂ from bovine brain] and L--phosphatidyl-D-myo-inositol-4-phosphate [PI(4)P, from bovine brain] was purchased from Roche Diagnostics GmbH (Mannheim, Germany); 1,2-dihexadecanoyl-rac-glycero-3-phosphocholine, synthetic (PC), L--phosphatidylinositol (PI, from bovine liver), and 1,2-di[cis-9-octadecenoyl]-sn-glycerol (DOG) from Sigma; L--phosphatidylinositol-3,4,5-triphosphate, dipalmitoyl-, heptaammonium salt [PI(3,4,5)P₃] and L--phosphatidylinositol-3,4-bisphosphate, dipalmitoyl-, pentaammonium salt [PI(3,4)P₂] from Calbiochem-Novabiochem GmbH (Bad Soden, Germany); and DOGS-NTA (1,2-dioleoyl-sn-glycero-3-succinyl[N(5-amino-1-carboxypentyl)iminodiacetic acid], ammonium salt) from Avanti Polar Lipids (Alabaster, AL). All lipids were stored as stocks at 20°C; PI(4)P, PI(4,5)P₂, PI(3,4)P₂, PI(3,4,5)P₃, DOG (1 mM), and DOGS-NTA (0.5 mM) in K_int, PI at concentration of 1 mM in dimethyl sulfoxide, and PC at concentration of 1 mM in ethanol (99%); lipid stocks were sonicated for 15 min before storage. For experiments lipids were diluted in K_int solution to final concentrations and sonicated for 30 min and used within 6 h.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Variability for Inhibition by ATP and Glibenclamide Is Linked to Membrane Phospholipids. For cardiac K_ATP channels large variability in sensitivity to ATP and sulfonylureas has been reported (Findlay and Faivre, 1991). We here investigated this variability for heterologously expressed cardiac-type (Kir6.2/SUR2A) K_ATP channels in inside-out patches from X. laevis oocytes. As shown in Fig. 1A, in some patches K_ATP channels showed rather weak sensitivity to inhibition by ATP and glibenclamide, whereas in others high sensitivity was observed. Current inhibition by 100 µM ATP and 10 µM glibenclamide ranged from 98 to 40% measured in 54 patches excised from five different batches of oocytes. For each patch ATP inhibition closely matched glibenclamide inhibition. This observation was quantified in Fig. 1B, where relative glibenclamide inhibition is plotted against ATP inhibition. The linear dependence indicates a direct link between ATP and glibenclamide sensitivity for each patch. The variability observed for ATP inhibition has been related to differences in membrane concentrations of phosphatidylinositol phosphates such as PIP₂ and PIP (Baukrowitz et al., 1998; Shyng and Nichols, 1998). To test for an effect on glibenclamide inhibition, PIP₂ was applied to the intracellular side of patches and inhibition by glibenclamide and ATP was monitored. As shown in Fig. 1C glibenclamide inhibition was reduced by successive application of 10 µM PIP₂ and this reduction occurred in parallel to the reduction of ATP inhibition (Fig. 1D). Thus, the close correlation between inhibition by ATP and glibenclamide, which characterizes the endogenous variability, was preserved for the effect of exogenously applied PIP₂. Furthermore, we tested whether the effects of PIP₂ on K_ATP channels could be reversed when the patch was brought back into contact with the cytoplasm of the oocyte. For this purpose an inside-out patch was first exposed to PIP₂, resulting in large reduction of ATP and glibenclamide sensitivity, and then crammed back into the oocyte and subsequently excised again to assay sensitivity to glibenclamide and ATP (Fig. 1E). Every round of cramming successively restored sensitivity to glibenclamide and ATP (Fig. 1, E and F). We presume that contact of the patch with the cytoplasm enables either membrane-bound or cytoplasmic enzymes (e.g., lipases or PIPs-phosphatases) to break down PIP₂ and thereby reduce the PIP₂ content of the membrane toward the original level before PIP₂ application.

View larger version (28K): [in this window] [in a new window]   Fig. 1.   Variability of sensitivity of KATP channels to ATP and glibenclamide is linked to PIPs. A, variable inhibition of currents mediated by Kir6.2/SUR2A channels in response to application of ATP and glibenclamide shown for two inside-out patches. Currents were recorded at membrane potentials of 80 mV stepped to 60 mV for 200 ms every second; only currents at 80 mV are shown as upward deflection. Patches were exposed to ATP and glibenclamide (Glib.) as indicated. B, relative (rel.) current at 10 µM glibenclamide (I(10 µM Glib.)/Imax) plotted against the rel. current at 100 µM ATP (I(100 µMATP)/Imax) from experiments as in A, plot represents summary of n = 54 experiments. C, change in inhibition by ATP and glibenclamide upon application of 10 µM PIP2 as indicated. D, rel. current with glibenclamide plotted against the rel. current with ATP from experiments as in C;  (rel. currents before PIP2) and  (rel. currents after PIP2), plot represents summary of n = 10 experiments. E, reversal of the PIP2 effect on ATP and glibenclamide inhibition upon insertion of the patch into the oocyte (cramming), application of 10 µM PIP2 and duration of cramming as indicated (cramming caused reversible current inhibition due to high concentrations of ATP in the oocyte). F, rel. current with glibenclamide plotted against rel. current with ATP from experiments as in E;  (current before PIP2),  (current after PIP2), and  (current after cramming), plot represents summary of n = 22 experiments.

Structural Requirements of Lipids for Inference with Glibenclamide and ATP Inhibition. To inquire for the structural determinants of the lipids to modulate glibenclamide inhibition and ATP inhibition, lipids of different charge and structure were tested. Application of 50 µM phosphatidylcholine (PC), 50 µM phosphatidylinositol (PI), and 50 µM dioleoylglycerol (DOG) for 45 s had no effect on ATP or glibenclamide inhibition, whereas application of 10 µM PI(4)P for 45 s virtually abolished inhibition by 100 µM ATP and 10 µM glibenclamide inhibition (Fig. 2A). The negatively charged PIPs PI(4)P, PI(4,5)P₂, PI(3,4)P₂, and PI(3,4,5)P₃ were tested in an assay as described in Fig. 1, C and D, and found to reduce glibenclamide inhibition with similar potency than ATP inhibition (Fig. 2C). Furthermore, the synthetic anionic lipid dioleoylglycerol-succinyl-nitriloacetic acid (DOGS-NTA) was tested, which lacks the inositol head group of PIPs but carries a negatively charged NTA head group instead (structure shown in Fig. 2B). DOGS-NTA reduced sensitivity of K_ATP channels to inhibition by 100 µM ATP and 10 µM glibenclamide in parallel similar to PIPs (Fig. 2, C and D). In summary, neutral or weakly negatively charged lipids (PC, DOG, and PI) had no effect on neither ATP nor glibenclamide inhibition, whereas the highly negatively charged PIPs and DOGS-NTA reduced both properties with similar potency.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Lipid specificity for modulation of ATP and glibenclamide inhibition. A, effect of 50 µM PC (n = 3), 50 µM PI (n = 6), 50 µM DOG (n = 3), and 10 µM PI(4)P (n = 5) on ATP and glibenclamide sensitivity. Lipids were applied for 45 s and subsequently inhibition by 100 µM ATP and 10 µM glibenclamide was plotted as columns; mean ± S.E.M.; control (contr.) represents inhibition before lipid application (n = 16). B, structure of PI(4,5)P2 and DOGS-NTA. C, change in inhibition by ATP and glibenclamide upon application of PI(4)P (10 µM), PI(3,4)P2 (10 µM), PI(3,4,5,)P3 (10 µM), and DOGS-NTA (10 µM) obtained from experiments as in D plotted as rel. currents as described in Fig. 1D. D, currents from a patch exposed to 100 µM ATP, 10 µM glibenclamide, 10 µM DOGS-NTA, and 100 µg/ml polylysine (poly-K) as indicated.

PIP₂ Modulates Sensitivity to Glibenclamide and ATP by Distinctive Mechanisms. To learn about the mechanism underlying the modulation of glibenclamide inhibition by PIP₂, glibenclamide concentration-response (CR) curves were obtained before and after application of PIP₂. Before PIP₂ application glibenclamide inhibited about 80% of the K_ATP current with an inhibitory constant (K_i) of 15 nM and a Hill coefficient of 1, whereas 20% of the K_ATP channels were insensitive to glibenclamide (n = 5). Application of PIP₂ successively increased the fraction of glibenclamide insensitive channels, but had little effect on the K_i value and the Hill coefficient of the glibenclamide-sensitive fraction (Fig. 3A; n = 7). The effect of PIP₂ on the glibenclamide CR curve was strikingly different from the effect on the ATP CR curve from Baukrowitz et al. (1998), which is shown in Fig. 3B for comparison. PIP₂ reduced ATP inhibition by gradually shifting the ATP CR curve toward higher concentrations without a change in steepness or indication for multiple components.

View larger version (34K): [in this window] [in a new window]   Fig. 3.   Distinctive mechanisms for the effects of PIP2 on glibenclamide and ATP sensitivity. A, CR curves for glibenclamide inhibition of KATP channels before and subsequent to application of PIP2. To obtain mean and S.D. of each data point CR curves were pooled by the respective ATP sensitivity measured shortly before the CR curves were taken. CR curve without PIP2 corresponds to patches with KATP channels inhibited to 80 to 90% by 100 µM ATP; CR curves succeeding application of PIP2 correspond to KATP channels inhibited to 50 to 60% (short PIP2 application) and 25 to 35% (prolonged PIP2 application). Continuous lines represent fit to the Hill equation I / Imax = 1 / [1 + ([ATP] / Ki(Glib.))nH], where Imax is the current amplitude without glibenclamide, Ki(Glib.) is the concentration for half-maximal inhibition, and nH is the Hill coefficient. Ki(Glib.)and nH were 16 nM/0.9 (before PIP2), 29 nM/1 (after PIP2) and 33 nM/1.2 (after prolonged PIP2). B, change of the CR curve for ATP inhibition induced by PIP2, reproduced from Baukrowitz et al. (1998). C, polylysine abolishes the effect of PIP2 on ATP but not on glibenclamide inhibition. The patch was preincubated with PIP2 (data not shown) and exposed to ATP, glibenclamide, 100 µg/ml polylysine (poly-K) and 100 µg/ml heparin as indicated. D, relative (rel.) currents with 100 µM ATP and 10 µM glibenclamide from experiments as in C before and subsequent to PIP2, polylysine and heparin treatment plotted as columns, mean ± S.E.M. of n = 6 experiments. E, effect of neomycin on ATP and glibenclamide inhibition subsequent to PIP2 treatment; rel. currents with 100 µM ATP and 10 µM glibenclamide before and subsequent to PIP2 and in the presence of 500 µM neomycin plotted as columns, mean ± S.E.M. of n = 6 experiments. F, rel. currents with 100 µM ATP and 10 µM glibenclamide before and subsequent to application of 10 µM DOGS-NTA and subsequent to treatment with 100 µg/ml poly-K plotted as columns, mean ± S.E.M. of n = 4 experiments.

R176A Attenuates the Effect of PIP₂ on Glibenclamide and ATP Inhibition. Recently a conserved C-terminal arginine (R181 in Kir1.1 and R176 in Kir6.2) has been identified that contributes to PIP₂ binding (Fan and Makielski, 1997; Huang et al., 1998). Mutating arginine 176 to alanine (R176A) in Kir6.2 reduced the ability of PIP₂ to affect ATP inhibition (Baukrowitz et al., 1998; Shyng and Nichols, 1998). Therefore, we tested whether R176A also reduced the potency of PIP₂ to attenuate glibenclamide inhibition. Figure 4 shows the time course for attenuation of inhibition produced by 100 µM ATP (Fig. 4A) and 10 µM glibenclamide (Fig. 4B) upon application of 10 µM PIP₂ for wild-type (wt) channels (n = 5) and R176A mutant channels (n = 14). The experiments for wt and mutant channels were performed on the same day with the same batch of oocytes and the same solutions to minimize differences in the experimental conditions. We found that inhibition by ATP and glibenclamide is reduced more slowly by PIP₂ for R176A channels compared with wt channels. The differences are not large but are significant (p < 0.002, unpaired t test) for every single time point. Thus, the mutation R176A attenuated the effect of PIP₂ on ATP and glibenclamide inhibition consistent with R176 contributing to a PIP₂ binding site that modulates both properties.

View larger version (10K): [in this window] [in a new window]   Fig. 4.   R176A attenuates the effects of PIP2 on glibenclamide and ATP inhibition. A, time course for attenuation of inhibition by 100 µM ATP for wt channels (n = 6) and R176A mutant channels (n = 17) upon application of 10 µM PI(4,5)P2. B, time course for attenuation of inhibition by 10 µM glibenclamide for wt channels (n = 5) and R176A mutant channels (n = 14) upon application of 10 µM PI(4,5)P2. Asterisks indicate a significant difference between wt and R176A mutant channels (*p < 0.05, **p < 0.002, unpaired t test). Prolonged application of PIP2 (>2 min) completely abolished inhibition by glibenclamide and ATP for wt and mutant channels (data not shown).

Negatively Charged Lipids Render K_ATP Channels Insensitive to KCOs by a Polycation-Insensitive Mechanism. For the pancreatic type of K_ATP channels (Kir6.2/SUR1) we and others have recently shown an attenuating effect of PIP₂ on the potency of the KCO diazoxide (Baukrowitz et al., 1998; Koster et al., 1999). The interference of PIP₂ with cardiac-type K_ATP channels (Kir6.2/SUR2A) was investigated using the KCOs cromakalim and P1075. As shown in Fig. 5A, application of 10 µM P1075 activated 77% of the current inhibited by 100 µM ATP (n = 7). To test for the effect of PIP₂, patches were treated with PIP₂ until ATP sensitivity was reduced about 10-fold so that 1 mM ATP was necessary to produce inhibition comparable with that of 100 µM ATP before PIP₂ treatment. Application of 10 µM P1075 to those patches failed to reverse ATP inhibition (Fig. 5B; n = 8). Very similar results were obtained for the KCO cromakalim (before PIP₂ activation was 85 ± 1%, subsequent to PIP₂ activation was 12 ± 5%, n = 9, data not shown). In control experiments (no PIP₂) P1075 only weakly activated K_ATP channels in the presence of 1 mM ATP (Fig. 5C). This outcome is not surprising because KCOs are thought to antagonize ATP inhibition by an apparent competitive mechanism. Thus, the apparent ability of KCOs to overcome ATP inhibition ceases with increasing inhibition by ATP (Thuringer and Escande, 1989). Consistently, when ATP sensitivity was reduced (about 10-fold, comparable with that of PIP₂-treated patches in Fig. 5B) via a mutation in Kir6.2 (K185Q, K_i = 137 ± 20, n = 8; Tucker et al., 1997) 10 µM P1075 potently activated K_ATP channels inhibited by 1 mM ATP (Fig. 5D; n = 6). Thus, the experimental conditions in Fig. 5B are appropriate to assess the impact of PIP₂ on the activation of K_ATP channels by KCOs.

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Negatively charged lipids abolish sensitivity to KCOs by a polycation-insensitive mechanism. A, activation of KATP channels inhibited by ATP upon application of 10 µM P1075 as indicated. Activation was quantified as percentage of the current inhibited by 100 µM ATP that was reopened by P1075, activation (%) = [(I(100 µM ATP + 10 µM P1075)  I(100 µM ATP))/Imax]/(1  I(100 µM ATP)/Imax] * 100 and plotted as columns, mean ± S.E.M. (n = 7). B, KATP current from a patch exposed to 10 µM PIP2, ATP, and 10 µM P1075 as indicated. Columns represent percentage of activation of the current inhibited by 1 mM ATP that was reopened by P1075, mean ± S.E.M. (n = 7). C, KATP current from a patch exposed to 1 mM ATP and 10 µM P1075 (no PIP2). Columns represent percentage of activation of the current inhibited by 1 mM ATP that was reopened by P1075, mean ± S.E.M. (n = 9). D, activation of K185Q mutant channels by 10 µM P1075. Columns represent percentage of activation of the current inhibited by 1 mM ATP that was reopened by P1075, mean ± S.E.M. (n = 5). E, polylysine (poly-K) restored high ATP sensitivity but not activation by P1075. KATP current from a patch exposed to 10 µM PIP2, 100 µg/ml polylysine, ATP, and 10 µM P1075 as indicated. Columns represent mean ± S.E.M. percentage of activation of the current inhibited by 1 mM ATP that was reopened by P1075 subsequent to treatment with PIP2 and poly-K (n = 9). A similar outcome was obtained for 10 µM DOGS-NTA and subsequent poly-K application (n = 5). F, patches were exposed to 50 µM PC (n = 9), 50 µM PI (n = 7) and 100 µM DOG (n = 4) for 45 s and subsequently activation by P1075 was determined. Columns represent percentage of activation of the current inhibited by 100 µM ATP that was reopened by 10 µM P1075.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

PIPs Modulate Sensitivity to ATP, Sulfonylureas, and KCOs by Distinctive Mechanisms. We report here that PIPs profoundly alter the sensitivity of cardiac type K_ATP channels to ATP, glibenclamide, and the KCOs P1075 and cromakalim. Although PIPs simultaneously desensitize K_ATP channels to ATP, sulfonylureas, and KCOs several findings suggest different modulatory mechanisms. First, the impact of PIP₂ on the CR curves for inhibition by ATP and glibenclamide are very distinct. The effect of PIP₂ on the ATP CR curve is characterized by a shift toward higher concentrations without a change in steepness or indication for multiple components. In contrast, PIP₂ abolishes glibenclamide inhibition by increasing the fraction of glibenclamide-insensitive channels without altering the CR curve of the glibenclamide-sensitive fraction. Thus, PIP₂ seems to alter glibenclamide sensitivity in an all-or-nothing manner, whereas ATP sensitivity is shifted gradually. This gradual shift suggests the existence of multiple PIP₂ binding sites, which contribute, to the overall shift in ATP sensitivity in an additive manner. The "all-or-nothing" effect of PIP₂ on glibenclamide inhibition suggests that the channel exists in two states with respect to glibenclamide sensitivity and binding of PIP₂ promotes the channel from a sensitive into an insensitive state.

Possible Mechanisms for the Action on PIP₂ on K_ATP Channel Properties. Two mechanisms have been proposed to account for the regulation of ATP sensitivity by PIPs: a competitive mechanism where PIP₂ binds in proximity to the ATP binding site on Kir6.2, reducing binding of ATP by an electrostatic mechanism (Deutsch et al., 1994; Fan and Makielski, 1999) and an allosteric mechanism where PIP₂ increases the open probability of K_ATP channels and thereby decreases the frequency of a closed state to which ATP binds to inhibit channel activity (Enkvetchakul et al., 2000). Furthermore, a recent article provides evidence that both mechanisms coexist and reduction in ATP sensitivity is brought about by a direct effect on ATP binding and an indirect effect involving modulation of the open probability (Ribalet et al., 2000). The effect of PIP₂ on glibenclamide inhibition and activation by KCOs is presumably allosteric because the PIP₂ interaction sites are likely on the Kir6.2 subunit, whereas sulfonylureas and KCOs bind to the SUR. In line with this concept Koster et al. (1999) recently reported correlation between the open probability of K_ATP channels and their sensitivity to inhibition by tolbutamide and suggested allosteric modulation of tolbutamide inhibition by PIP₂. How to explain in this context the differential sensitivity to polycations for the effect of PIP₂ on ATP, glibenclamide, and KCO sensitivity? Polycations might by neutralizing the negative charge of PIP₂ reduce electrostatic interactions necessary for a direct effect on ATP binding that, however, are less important for the allosteric effect on glibenclamide inhibition and activation by KCOs.

Structural Requirements of Lipids to Modulate K_ATP Channels. To elucidate the structural requirements for modulation of K_ATP channels lipids with different structure and charge were tested for their potency to affect ATP inhibition, glibenclamide inhibition, and activation by KCOs. The zwitterionic PC, the neutral DOG, and the weakly negatively charged PI had no effect on the sensitivity of K_ATP channels for ATP, glibenclamide, or P1075. In contrast, the highly negatively charged phosphatidylinositol phosphates [PI(4)P, PI(4,5)P2, PI(3,4)P2, and PI(3,4,5)P₃] were potent modulators of all three properties. Thus, a negatively charged head group attached to a hydrophobic tail are necessary structural requirements for the lipids to reduce glibenclamide inhibition and activation by openers, as has been previously demonstrated for ATP inhibition (Fan and Makielski, 1997; Baukrowitz et al., 1998; Shyng and Nichols, 1998). Furthermore, we show that the structure of the negatively charged head group is less critical because the artificial lipid DOGS-NTA acts as a full substitute for PIPs but possesses a negatively charged NTA group instead of the phosphorylated inositol ring of PIPs (Fig. 2B). DOGS-NTA might serve as a useful agent for further functional and structural studies because it is considerably less expensive and more stable than PIP₂.

Implications for Physiology and Pharmacology of K_ATP Channels. For cardiac K_ATP channels large variability with respect to sulfonylurea and ATP inhibition has been reported. ATP sensitivity measured in excised patches from cardiac myocytes can vary 60-fold (Findlay and Faivre, 1991). In addition, ATP sensitivity may change during different metabolic states of cardiac myocytes; e.g., metabolic stress has been demonstrated to profoundly reduce ATP sensitivity of K_ATP channels (Deutsch and Weiss, 1993). This mechanism might be important for the cardioprotective function of K_ATP channels. Intriguingly, also glibenclamide sensitivity is dramatically reduced under metabolic stress situations (Findlay, 1993a,b). Thus, it seems that sulfonylurea and ATP sensitivity of K_ATP channel are regulated by a shared pathway. We report here that also heterologously expressed cardiac type K_ATP channels show considerable variability for ATP and glibenclamide inhibition in excised patches (Fig. 1). This variability is characterized by a close correlation between ATP and glibenclamide sensitivity. Application of PIPs reduced ATP and glibenclamide sensitivity in a way closely resembling the native variability because ATP and glibenclamide sensitivity is changed in parallel. The effects of PIP₂ on K_ATP channels are stable (even in presence of high Mg²⁺ and Ca²⁺, data not shown) in excised patches but can be readily reversed upon cramming of the patch back into the cytoplasm of the oocyte. Thus, exogenously applied PIP₂ can serve as substrate for endogenous PIPs-degrading enzymes. These results suggest that PIPs modulate K_ATP channels in a reversible manner as expected for a regulatory mechanism occurring in vivo.