Nucleotides and phospholipids compete for binding to the C terminus of K_ATP channels

Abstract

Inwardly rectifying, ATP-sensitive K⁺ channels (K_ATP) couple metabolism to either cell excitability (Kir6.x) or potassium secretion (Kir1.1). Phosphatidylinositol phospholipids, like PI(4,5)P₂, antagonize nucleotide inhibition of K_ATP channels enhancing the coupling of metabolic events to cell electrical or transport activity. The mechanism by which phospholipids relieve ATP block is unclear. We have shown that maltose-binding fusion proteins (MBP) containing the COOH termini of K_ATP channels (Kir1.1, Kir6.1, and Kir6.2) form functional tetramers that directly bind at least two ATP molecules with negative cooperativity. Here we show that purified phosphatidylinositol phospholipids compete for 2,4,6,-trinitrophenyl (TNP)-ATP binding to the COOH termini of K_ATP channels with EC₅₀ values for PIP₂ between 6–8 μM. The phospholipid potency profile was PIP₃ > PIP₂ = PIP > PI, suggesting that net phospholipid charge was important. A role for head group charge was supported by polycations (neomycin, spermine, and polylysine) reversing the effect of PIP₂ on TNP-ATP binding to the Kir1.1 channel COOH terminal fusion protein. In contrast, the water-soluble charged hydrolytic product of PIP₂, inositol(1,4,5)P₃ (IP₃), had no effect on TNP-ATP binding, suggesting that the acyl chain of PIP₂ was also necessary for its effect on TNP-ATP binding. Indeed, neutral and charged lipids had weak, but significant, effects on TNP-ATP binding. Whereas μM concentrations of PIP₂ could compete with TNP-ATP, we found that mM concentrations of MgATP were required to compete with PIP₂ for binding to these K_ATP channel COOH termini. Thus the COOH termini of K_ATP channels form a nucleotide- and phospholipid-modulated channel gate on which ATP and phospholipids compete for binding.

ATP-sensitive or ATP-regulated potassium (K_ATP) channels couple metabolism to either cell excitability (Kir6.x) or potassium secretion (e.g., Kir1.1 in kidney; refs. 1–7). K_ATP channels are formed by an octameric complex of four pore-forming subunits (Kir6.x or Kir1.1) and four sulfonylurea receptors, SUR for Kir6.x (2, 8) or the cystic fibrosis transmembrane conductance regulator, CFTR, or SUR for Kir1.1 (9–11).

Phosphatidylinositol phosphates provide spatially restricted membrane signals (12, 13) that compete with ATP for gating of K_ATP channels (14–18) and also modulate the activity of other non-nucleotide gated Kir channels (14, 19). The effect of PIP₂ to antagonize the inhibitory action of ATP provides a mechanism for opening K_ATP channels in intact cells in the presence of millimolar concentrations of cytosolic ATP (15, 18, 20, 21). Mutation of basic residues in the COOH termini of both K_ATP and other Kir channels have been shown to alter PIP₂ binding and reduce channel gating by phosphatidylinositol phosphates (14–19, 22).

ATP inhibition of K_ATP channel activity is thought to involve nucleotide binding to the Kir pore-forming subunit (16, 23–29). We have recently demonstrated that the fluorescent ATP analogue, TNP-ATP (2′,3′-O-(2,4,6-trinitrophenylcyclo-hexadienylidene) adenosine 5′-triphosphate) specifically binds to MBP-fusion proteins of the C termini of all of the known K_ATP channels (Kir1.1, Kir6.1, and Kir6.2) but not to the C terminus of an ATP-insensitive inward rectifier K⁺ channel, Kir2.1 (unpublished observations). Although the mechanism underlying PIP₂ antagonism of ATP inhibition is unclear, the observations that basic residues on the COOH terminus of Kir are also critical for the effects of phosphatidylinositol phospholipids (14–18) suggests that PIP₂ and ATP may directly compete for binding to the COOH termini of K_ATP channels. To examine this hypothesis, we assessed the interactions of phosphatidylinositol phospholipids and ATP binding to the COOH termini of Kir1.1, Kir6.1, and Kir6.2.

Methods

Construction of MBP-Fusion Protein Expression Vectors and Protein Purification.

The cDNA encoding the C termini of Kir6.1 (encoding amino acids 178–424; 247 aa) and Kir6.2CΔ36 (encoding amino acids 169–354; 186 aa) were obtained by reverse transcription (RT)-PCR from rat kidney and brain, respectively. The C terminus of rat Kir1.1 (encoding amino acids 183–391; 209 aa) was obtained by PCR from the previously cloned Kir1.1 in the pSport vector (30). Sequences of all recombinant cDNA constructs were determined by fluorescently labeled dideoxynucleotide dye termination (Keck Foundation Biotechnology Resource Laboratory, Yale University, New Haven, CT). Recombinant proteins were expressed using the pMALT vector [kindly provided by G. A. Altenberg, University of Texas Medical Branch, Galveston, TX (32)], which was derived from the pMAL-c2 vector (maltose binding protein, MBP fusion vector, New England Biolabs). Epicurian coli BL21-CodonPlus-RIL competent cells (Stratagene) were transformed with pMALT vectors containing MBP-Kir C terminus fusion-protein constructs and a fresh colony was selected from LB plates supplemented with 0.1 mg/ml ampicillin.

Recombinant proteins were expressed as per the manufacturer's instructions (New England Biolabs). Briefly, 1 l of LB medium supplemented with 0.1 mg/ml ampicillin and 0.5% glucose was inoculated with 10 ml of an overnight culture grown in the same medium. The culture was grown at 37°C until an optical density of A₆₀₀ ≈ 0.5 and then induction was performed with addition of 0.3 mM isopropyl β-D-thiogalactoside (IPTG, American Bioanalytical, Natick, MA) and incubation was continued for 2 h at 37°C. Cells were harvested by centrifugation at 4,000 × g for 20 min at 4°C. The cell pellet was resuspended in 50 ml of column buffer (in mM: 20 Tris⋅HCl/200 NaCl/1 EDTA, pH 7.4) and frozen overnight at −20°C. The sample was thawed in water at 37°C and soluble protein was released from the bacteria by membrane rupture during sonication on ice (9 s on, 9 s off for 4 min, using a Vibracell probe sonicator, Sonics and Materials, Newtown, CT). The suspension was then centrifuged at 10,000 × g for 30 min at 4°C and the supernatant was diluted 1:5 with column buffer and loaded onto preequilibrated amylose columns (8–10 ml of amylose resin). Unbound proteins were washed off the column by 12 column volume washes with column buffer and the bound fusion protein was eluted with column buffer containing 10 mM maltose and collected in 1.5-ml fractions. Fractions with A₂₈₀ > 0.5 were pooled and buffer exchanged with 50 mM Tris⋅HCl, pH 7.5 by using a Sephadex G-25 superfine resin based column and a Pharmacia ÄKTA FPLC protein purification system (Amersham Pharmacia Biotech). Fusion protein was quantified using the BCA protein assay method (Pierce) and yields of purified soluble fusion protein were ≈20 mg/l.

TNP-ATP Binding.

Binding of TNP-ATP (2′,3′-O-(2,4,6-trinitrophenylcyclo-hexadienylidene) adenosine 5′-triphosphate) to recombinant proteins was performed as described by Faller (31). Briefly, 5 μM recombinant protein was dissolved in 50 mM Tris⋅HCl at pH 7.5 and TNP-ATP binding was detected by the concentration-dependent increase in TNP-ATP fluorescence after equilibration for 30 s following addition (FluoroMax-3 spectrofluorometer, Jobin Yvon, Edison, NJ: excitation, 403 nm; emission, 546 nm at 22°C). Fluorescence enhancement, stoichiometry, and EC₅₀ values were determined by fitting to a modified version of the binding equation derived by Faller (31), using GraphPad PRISM 3.0 software (San Diego). Total fluorescence is given by Eq. 1:

1

Q and Q₂ are derived independently from the concentration dependence of TNP-ATP fluorescence in buffer alone and account for the inner filter effect. R and R₂ are derived independently from the concentration dependence of light scattering of individual proteins. P_o is protein concentration; γ is the enhancement factor; N is the TNP-ATP to protein subunit stoichiometry; and K_d is the binding affinity.

Lipid Preparation.

L-α-Phosphatidyl-D-myo-inositol-4,5-bisphosphate [PI(4,5)P₂] was purchased from Roche Diagnostics, Calbiochem, and Avanti Polar Lipids. L-α-Phosphatidylinositol-3,4,5-trisphosphate [PI(3,4,5)P₃] was from Calbiochem. l-α-Phosphatidylinositol-4-phosphate [PI(4)P], phosphatidylinositol (PI), phosphatidylserine (PS), and l-α-phosphatidylcholine (PC) were from Avanti Polar Lipids. No difference was observed in the properties of PIP₂ from the different sources. Most of the data were obtained using PI(4,5)P₂ from Avanti Polar Lipids (denoted as PIP₂), which had the stearyl (18:0 or 18:1) and arachidonyl (20:4) chains in its two acyl positions. The PIP(4,5)P₂ from Roche Diagnostics and Calbiochem had similar acyl chain configurations. The PI(3,4,5)P₃ from Calbiochem was synthesized with two palmitoyl chains, whereas all other lipids [PI(4)P, PI, PS, PC] from Avanti Polar Lipids were derived from natural sources and linked to relatively long (≈16–20) hydrocarbon chains. Lipids were dissolved in chloroform, dried under a stream of N₂, hydrated (50 mM Tris⋅HCl, pH 7.5, experimental buffer), and sonicated (Bransonics 1510 bath sonicator, Branson, Danbury, CT) at a temperature above Tc (the gel-liquid crystal transition temperature) for 30 min (this ranged from room temperature to ≈40°C, depending on specific lipid), or until an optically clear suspension was formed. Once prepared, the lipid was kept at room temperature for the duration of the experiment and used only on day of preparation. Poly-D-lysine (P0296, Sigma) with a molecular mass of ≈3–4 kDa, neomycin, and spermine were purchased from Sigma.

Liposome Preparation and Binding Assay.

The protocol was essentially similar to Chockalingam et al. (37). Lipids were dissolved at 10 mg/ml egg PC or 1 mg/ml PI(4,5)P₂ in chloroform under nitrogen and mixed to give solutions of 100% PC or 1% PI(4,5)P₂ and 99% PC (wt/wt). Mixtures were dried under a stream of nitrogen, rehydrated in PBS to a final lipid concentration of 3.4 mg/ml, vortexed, and sonicated on ice for 1 h to form liposomes. Liposomes were stored at room temperature under nitrogen and used on the day of preparation. COOH-terminal fusion protein was dialyzed against PBS overnight at 4°C and centrifuged (Sorvall discovery M150 MicroUltracentrifuge with S-100-AT3 rotor) at 100,000 × g for 15 min at 4°C. Liposomes (20 μl) were mixed with 2 μg of recombinant protein and diluted to a final volume of 40 μl with PBS (final lipid concentration 1.7 mg/ml) and incubated at room temperature for 10 min and on ice for 5 min. Lipid/protein mixtures were then centrifuged at 100,000 × g for 15 min at 4°C, the supernatant was saved, and the pellet was rinsed with PBS. MgATP competition of recombinant protein binding to PIP₂ liposomes (1% PIP₂ with 99% egg PC) was determined by incubating with varying concentrations of MgATP in PBS at pH 7.4. Pellets and supernatants were resuspended in Laemmli buffer and proteins were resolved by SDS/PAGE and visualized with Coomassie brilliant blue.

Results

Phosphatidylinositol Phospholipid and TNP-ATP Interactions with the COOH Terminus of Kir1.1 (ROMK) Channel.

Phosphatidylinositol phosphates are required for maintenance of Kir1.1 channel activity (14, 17). To determine whether there is interaction between ATP and phosphatidylinositol phospholipids in Kir1.1 channels, we assessed the ability of variously phosphorylated phosphatidylinositol phospholipids to compete for TNP-ATP binding on the MBP-Kir1.1 C-terminal fusion protein (MBP_1.1C; Fig. 1). Total TNP-ATP fluorescence (F_T; Fig. 1A Left) increased in a concentration-dependent manner (10 μM maximal concentration) and could be fit by Eq. 1 to yield a K_d, enhancement factor (γ), and stoichiometry (N; ratio of TNP-ATP to protein) for TNP-ATP binding of 0.89 ± 0.31 μM, 4.66 ± 1.07, and 0.36 ± 0.09, respectively (n = 5). To determine whether phosphatidylinositol phospholipids can compete with ATP binding to MBP_1.1C, we added increasing concentrations of PIP₂ in the presence of 10 μM TNP-ATP (Fig. 1A Right). PIP₂ decreased F_T in a concentration-dependent fashion with an EC₅₀ of 5.62 ± 0.49 μM (n = 8). In addition, preexposure of MBP_1.1C to 10 μM PIP₂ abolished TNP-ATP binding (Fig. 1B). Thus, PIP₂ competes with TNP-ATP binding to the Kir1.1 COOH terminus.

Figure 1.

PIP₂ reduces TNP-ATP binding to the COOH-terminal MBP-fusion protein of Kir1.1 (MBP_1.1C). (A Left) Concentration-dependent increases in TNP-ATP fluorescence in the presence of 5 μM MBP_1.1C (■; n = 5), indicating binding of this nucleotide to the protein. The solid line was fit according to Eq. 1. (A Right) PIP₂ decreases TNP-ATP fluorescence in a dose-dependent manner to the levels seen with buffer alone, (□; n = 8), indicating that this phosphatidylinositol phospholipid competes off TNP-ATP binding to MBP_1.1C. The solid line was fit to a downhill dose-dependent model (GraphPad PRISM V. 3.02) giving an IC₅₀ of 5.62 ± 0.49 μM and a Hill coefficient of 1.28 ± 0.11. (B) Concentration-dependent increases in TNP-ATP fluorescence with MBP_1.1C (5 μM) in the presence (▴; dashed line) or absence (■; solid line) of 10 μM PIP₂. For comparison, TNP-ATP fluorescence in Tris⋅Cl buffer only is shown (⋄; dotted line). Thus, preexposure of MBP_1.1C to PIP₂ abolishes TNP-ATP binding to MBP_1.1C.

Increasing the phosphorylation state of phosphatidylinositol phospholipids enhanced their potency to compete with TNP-ATP binding to MBP_1.1C (Fig. 2). The EC₅₀ values for the phosphatidylinositol phospholipids were: PIP₃ (4.5 ± 0.4 μM; n = 6) > PIP₂ (5.6 ± 0.5 μM; n = 6) = PIP (5.6 ± 0.2 μM; n = 5) > PI (20.4 ± 0.9 μM; n = 5). The anionic lipid phosphatidylserine (PS; EC₅₀ = 54.9 ± 13.1 μM; n = 5) and the neutral lipid PC (EC₅₀ = 265 ± 17 μM; n = 4) also reduced TNP-ATP binding although requiring much higher concentrations than the phosphatidylinositol phospholipids. Receptor-mediated activation of phospholipase C cleaves PIP₂ to generate the water-soluble products inositol(1,4,5)P₃ (IP₃) and diacylglycerol, but these compounds do not antagonize ATP gating of K_ATP channels (17). Consistent with their lack of functional antagonism, but in marked in contrast to the effects of phosphatidylinositol phospholipids, IP₃ and DOG had no effect on TNP-ATP binding to MBP_1.1C (Fig. 2). Thus, the acyl chain of the phosphatidylinositol phospholipids appears to be critical for the effect of phospholipids to antagonize TNP-ATP binding.

Figure 2.

Phosphatidylinositol (PI) phospholipids compete with TNP-ATP binding to the COOH terminus of the Kir1.1 channel (MBP_1.1C). The fractional change in 10 μM TNP-ATP fluorescence is plotted against the logarithm of test solute (S) concentration. PI phospholipids compete for binding to MBP_1.1C with a potency profile of PIP₃ (■; n = 6) > PIP₂ (●; n = 6) = PIP (▴; n = 5) ≫ PI (▾; n = 5). Neutral (phosphatidylcholine; PC; ○; n = 4) and charged (phosphatidylserine; PS; □; n = 5) lipids compete weakly, but significantly, with TNP-ATP binding, suggesting that the lipid acyl chains may interact with MBP_1.1C to alter TNP-ATP interactions. In contrast, the hydrolytic products of PIP₂ [IP₃ (▵) and dioctanoylglycerol (DOG; ▿)] had no effect on TNP-ATP fluorescence or binding.

Polycations like neomycin, polylysine, and spermine can reverse the PIP₂-mediated antagonism of ATP inhibition of K_ATP channels. The polycations presumably bind to the negatively charged phosphatidylinositol phospholipids in the cell membrane and prevent them from interacting with the K_ATP channel (17). Fig. 3 shows time-dependent changes in total fluorescence (F_T) with 5 μM MBP_1.1C protein. Additions of 10 μM TNP increased F_T significantly to a plateau values indicating steady-sate binding of this nucleotide to the COOH-terminal fusion protein. Additions of 10 μM PIP₂ rapidly decreased F_T to baseline (TNP-ATP fluorescence in buffer value), presumably by displacing TNP-ATP from its binding sites. Individually the polycations neomycin (Fig. 3A), polylysine (Fig. 3B), and spermine (Fig. 3C) reversed the PIP₂ competition of TNP-ATP binding. The order of potency in reversing the PIP₂ inhibition of TNP-ATP binding was polylysine > neomycin > spermine. The increase in TNP-ATP fluorescence with polycation additions was due to rebinding of TNP-ATP to MBP_1.1C protein because final addition of 10 mM MgATP reduced the enhanced TNP-ATP F_T due to binding by 50–75%. This reduction in F_T is due to MgATP competing with TNP-ATP for binding. The remaining signal is due to intrinsic free TNP-ATP fluorescence in buffer and protein light scatter. Thus neomycin, polylysine, and spermine reversed the antagonism of TNP-ATP binding to MBP_1.1C by PIP₂.

Figure 3.

Polycations reverse the effect of PIP₂ competition of TNP-ATP binding to the COOH termini of Kir1.1 (MBP_1.1C). Representative time-dependent changes in TNP-ATP fluorescence in the presence of 5 μM Kir1.1 COOH termini fusion protein are shown for neomycin (A), poly-D-lysine (B), and spermine (C). Addition of 10 μM TNP-ATP (arrow) rapidly increases fluorescence to a steady-state maximal value. Ten micromolar PIP₂ (second arrow) rapidly decreases TNP-ATP fluorescence indicating unbinding of the nucleotide from MBP_1.1C. Incremental additions of the polycations (arrow; numbers indicate total μM concentrations of polycations) reverse the effect of PIP₂ on TNP-ATP fluorescence or binding. The rebinding of TNP-ATP was confirmed by the final decrease in TNP-ATP fluorescence by 10 mM MgATP.

A direct interaction of PIP₂ and ATP with Kir1.1 COOH terminus also predicts that ATP would compete with PIP₂ for binding to MBP_1.1C. We tested this possibility by assessing the ability of MgATP to reverse the binding of MBP_1.1C to PIP₂-containing liposomes (Fig. 4). Fig. 4A shows that by using 100% PC liposomes, about 10–20% of MBP_1.1C is found in the pellet vs. supernatant, consistent with weak interactions between neutral lipids and the relatively hydrophobic MBP_1.1C. This suggests an important role of the acyl chain in PIP₂ for competing with TNP-ATP binding, as shown in Fig. 2. In contrast, the hydrophilic MBP did not associate with the pellet. Liposomes containing 99% PC/1% PIP₂ bind >90% of MBP_1.1C protein (associated with the pellet), but little MBP (Fig. 4B). Fig. 4C demonstrates that MBP_1.1C protein binding to 99% PC/1% PIP₂ liposomes was decreased by increasing concentrations of MgATP, which competes for binding with the liposomes. Approximately one-half of the protein is in each fraction at 20 mM MgATP (Fig. 4C)

Figure 4.

MgATP competes PIP₂ binding to the COOH terminus of Kir1.1 (MBP_1.1C). Representative Coomassie stained 10% PAGE gels are shown. Fusion proteins were mixed with liposomes and the segregation of protein to the supernatant (S; buffer) or pellet (P; liposome) following centrifugation was examined. 10–20% of MBP_1.1C associates with 100% PC liposomes (A Left), indicating a weak binding to MBP_1.1C to this neutral lipid and consistent with the amphipathic nature of this protein. No significant amounts of MBP (without fusion protein) associate with 100%PC (A Right) or 1%PIP₂/99%PC (B Right). In contrast, over 90% of MBP_1.1C associates with the 1%PIP₂/99%PC liposomes (B Left). (C) The binding of MBP_1.1C to 1%PIP₂/99%PC liposomes is competed by MgATP. Over 90% of MBP_1.1C initially associates with the pellet (P) at 0 mM MgATP with a gradual shift to the supernatant (S) as MgATP is added in increasing concentrations up to 40 mM. An approximate 50–50% optical density distribution between S and P was observed at 20 mM MgATP.

Phosphatidylinositol Phospholipid and TNP-ATP Interactions with the COOH Termini of Kir6.x Channels.

Phosphatidylinositol phosphates antagonize ATP gating of the classic K_ATP channel, Kir6.2 found in excitable cells (15, 18). In Fig. 5 we show TNP-ATP binding and PIP₂ interactions to MBP_6.1C (Fig. 5A) and MBP_6.2Δ36 (Fig. 5B). We used the MBP_6.2CΔ36 protein because deletion of the last 36 aa from the end of the COOH terminus of Kir6.2 gives rise to functional and ATP-sensitive channel activity in cells in the absence of SUR1 (24). TNP-ATP F_T increased in a concentration-dependent manner in the presence of either MBP_6.1C (Fig. 5A Left) or MBP_6.2CΔ36 (Fig. 5B Left). Total fluorescence saturated at 10 μM TNP-ATP with K_d values of 0.51 ± 0.09 μM (n = 5) for MBP_6.1C and 0.91 ± 0.11 μM (n = 6) for Kir6.2Δ36. The N and γ values were 0.26 ± 0.03 and 23.10 ± 3.24 for MBP_6.1C and 0.44 ± 0.04 and 2.83 ± 0.18 for MBP_6.2CΔ36, respectively. PIP₂ competed with TNP-ATP for binding with EC₅₀ values of 8.29 ± 0.43 μM for MBP_6.1C (Fig. 5A Right) and 6.33 ± 0.16 μM for MBP_6.2CΔ36 (Fig. 5B Right).

Figure 5.

PIP₂ competition of TNP-ATP binding to the COOH termini classic K_ATP channels (Kir6.1; MBP_6.1C) and (Kir6.2; MBP_6.2CΔ36). Concentration- dependent increases in TNP-ATP fluorescence with MBP_6.1C (A Left; ■; n = 5) or MBP_6.2CΔ36 (B Left; ■; n = 6). The solid lines are fit according to Eq. 1. Incremental additions of PIP₂ reduced TNP-ATP fluorescence to MBP_6.1C (A Right) and MBP_6.2CΔ36 (B Right) in a concentration-dependent fashion. The solid lines were fit to a downhill dose-dependent model (GraphPad prism v. 3.02) giving an IC₅₀ of 8.29 ± 0.43 μM and Hill coefficient of 1.73 ± 0.05 for MBP_6.1C and an IC₅₀ of 6.33 ± 0.16 μM and Hill coefficient of 1.55 ± 0.01 for MBP_6.2CΔ36.

We also assessed the ability of MgATP to reverse the binding of MBP_6.1C to PIP₂-containing liposomes (Fig. 6). With 100% PC liposomes, about 10–20% of MBP_6.1C bound to the liposomes, consistent with the amphipathic nature of the COOH terminus of Kir6.1 and with the necessity of the acyl chain of PIP₂ in the effects of this phosphatidylinositol phosphate on ATP interactions with the Kir6.x COOH termini (see Fig. 5). As with MBP_1.1C (Fig. 4), PC liposomes containing 1% PIP₂ (Fig. 6B) bind ≈90% of MBP_6.1C protein. Moreover, increasing millimolar concentrations of MgATP competed with PC/PIP₂ liposomes for binding to the MBP_6.1 fusion protein (Fig. 6C). Approximately 5 mM MgATP competed with MBP_6.1C for binding to about 50% of the PC/PIP₂ liposomes, a much lower concentration of MgATP than was required to reverse the MBP_1.1C binding (Fig. 4C).

Figure 6.

MgATP competes PIP₂ binding to the COOH terminus of Kir6.1 (MBP_6.1C). Representative Coomassie stained 10% PAGE gels are shown. Fusion proteins were mixed with liposomes and the segregation of protein to the supernatant (S; buffer) or pellet (P; liposome) following centrifugation was examined. 10–20% of MBP_6.1C associates with 100% PC liposomes (A Left), indicating a weak binding to MBP_6.1C to this neutral lipid and consistent with the amphipathic nature of this protein. More than 90% of MBP_6.1C associates with the 1%PIP₂/99%PC liposomes (B). (C) The binding of MBP_6.1C to 1%PIP₂/99%PC liposomes is competed by MgATP. Over 90% of MBP_1.1C initially associates with the pellet (P) at 0 mM MgATP with a gradual shift to the supernatant (S) as MgATP is added in increasing concentrations up to 40 mM. An approximate 50–50% optical density distribution between S and P was observed at 5 mM MgATP.

Discussion

To our knowledge this is the first study to show direct competition between ATP and PIP₂ for binding to the COOH termini of K_ATP channels. We demonstrate not only the ability of PIP₂ to reduce TNP-ATP binding to the COOH termini of the three known K_ATP channels (Figs. 1–3 and 5) but also competition by MgATP of MBP_1.1C (Fig. 4) and MBP_6.1C (Fig. 6) binding to PIP₂-containing liposomes. The affinity for PIP₂-mediated competition of TNP-ATP binding to Kir1.1, Kir6.1, and Kir6.2 COOH-terminal fusion proteins ranges between 6–8 μM, consistent with strong interactions with the COOH termini of K_ATP channels. The high ATP concentrations necessary (20–40 mM) to reverse completely PIP₂ binding to MBP_1.1C and MBP_6.1C fusion protein (Figs. 4 and 6) is consistent with the several-log-fold lowering (μM to mM) of ATP sensitivity of Kir6.2 channel gating by this phosphatidylinositol phosphate (15, 18). In addition, our finding that PIP₂ binds to the COOH termini of K_ATP channels is consistent with mutations of amino acids in the COOH termini of Kir1.1 (14) and Kir6.2 (15–18) dramatically reducing PIP₂ effects on channel gating.

Phosphatidylinositol phospholipids modulate the activity of both ATP-sensitive (15, 18, 21) and ATP-insensitive (14, 19) inwardly rectifying K⁺ channels by increasing open probability. In addition, phosphoinositides alter the activity of Na⁺–Ca²⁺ (20) and Na⁺–H⁺ (33) exchangers, indicating a wide range of effects on transport systems. K_ATP channels do not contain known pleckstrin homology domains (34) that form classic recognition sites for phosphoinositides. Instead, K_ATP channels have short sequences rich in basic amino acids in their COOH termini that may provide the binding site(s) for phosphatidylinositol phospholipids (13). Thus, the positive charge in these short regions on K_ATP channel COOH termini may be necessary for interaction with phosphatidylinositol phospholipids. The potency profile for phosphatidylinositol phospholipids shown in Fig. 2 indicates that the net negative charge of the phosphatidylinositol phosphates is important in the effects of these phospholipids on competing with TNP-ATP binding. Thus, charge effects are important in phospholipid–ATP binding interactions. In addition, polycations have been shown to reduce the EC₅₀ for ATP inhibition of K_ATP channels (35) and the effects of phosphatidylinositol phospholipids on channel gating (17, 36), suggesting that charge properties are also important in channel gating. Our observations that neomycin, spermine, and polylysine reverse the competition of PIP₂ on TNP-ATP binding (Fig. 3) are consistent with a charge effect and suggests that these polycations may bind to the PIP₂ liposomes and shield them from interacting at the nucleotide binding sites on the COOH terminus.

Interestingly, and in contrast to PIP₂, IP₃ does not activate K_ATP channels (17) but does bind to pleckstrin homology domains (34), consistent with the absence of a pleckstrin homology domain in the COOH termini of K_ATP channels. Our finding that IP₃ does not compete with TNP-ATP binding (Fig. 2) is also consistent with the lack of effect of IP₃ on channel gating. These observations suggest that charge interactions may be required, but are not sufficient, to permit interaction of phospholipids with the nucleotide binding site. The ability of neutral and charged lipids to compete, albeit weakly, with TNP-ATP binding (Fig. 2) suggests that the acyl chain of the phosphatidylinositol phospholipids may also be necessary for the effects of these phospholipids on K_ATP channels (and possibly other transporters) and ATP binding. This notion is supported by our finding that both MBP_1.1C and MBP_6.1C proteins weakly bind to 100% PC liposomes (Figs. 4A and 6A). In this regard, the COOH termini of K_ATP channels (as well as other Kir channel) are quite hydrophobic. Thus, the acyl chains of phosphatidylinositol phosphates may bind to hydrophobic regions on the COOH termini of K_ATP channels and help position the phosphate head groups in the nucleotide-binding pocket.

In conclusion, the direct competition of ATP binding by phosphatidylinositol phospholipid provides a model to explain the antagonism of ATP inhibition of K_ATP channels by PIP₂. This effect depends both on the charged phosphate head group and the neutral acyl chain of these phospholipids. Thus, the COOH termini of K_ATP channels form a nucleotide- and phospholipid-modulated channel gate on which ATP and phospholipids compete for binding.

Abbreviations

K_ATP

ATP-sensitive K⁺ channels

TNP

2,4,6,-trinitrophenyl

PI

phosphatidylinositol

PC

phosphatidylcholine