The mechanism of adenosine triphosphate (ATP)-sensitive potassium (KATP) channel activation by Mg-nucleotides was studied using a mutation (G334D) in the Kir6.2 subunit of the channel that renders KATP channels insensitive to nucleotide inhibition and has no apparent effect on their gating. KATP channels carrying this mutation (Kir6.2-G334D/SUR1 channels) were activated by MgATP and MgADP with an EC50 of 112 and 8 µM, respectively. This activation was largely suppressed by mutation of the Walker A lysines in the nucleotide-binding domains of SUR1: the remaining small (∼10%), slowly developing component of MgATP activation was fully inhibited by the lipid kinase inhibitor LY294002. The EC50 for activation of Kir6.2-G334D/SUR1 currents by MgADP was lower than that for MgATP, and the time course of activation was faster. The poorly hydrolyzable analogue MgATPγS also activated Kir6.2-G334D/SUR1. AMPPCP both failed to activate Kir6.2-G334D/SUR1 and to prevent its activation by MgATP. Maximal stimulatory concentrations of MgATP (10 mM) and MgADP (1 mM) exerted identical effects on the single-channel kinetics: they dramatically elevated the open probability (PO > 0.8), increased the mean open time and the mean burst duration, reduced the frequency and number of interburst closed states, and eliminated the short burst states. By comparing our results with those obtained for wild-type KATP channels, we conclude that the MgADP sensitivity of the wild-type KATP channel can be described quantitatively by a combination of inhibition at Kir6.2 (measured for wild-type channels in the absence of Mg2+) and activation via SUR1 (determined for Kir6.2-G334D/SUR1 channels). However, this is not the case for the effects of MgATP.
ATP-sensitive potassium (KATP) channels serve as metabolic sensors, coupling cell metabolism to plasmalemmal potassium fluxes and electrical activity in a variety of cell types (Miki and Seino, 2005; Ashcroft, 2007). Their activity is primarily determined by intracellular adenosine nucleotides, with ATP having an inhibitory effect and Mg-nucleotides (MgATP or MgADP) having a stimulatory effect on channel activity.
The β cell KATP channel is an octameric (4:4) complex of two different types of protein subunit: Kir6.2 and SUR1 (Clement et al., 1997; Mikhailov et al., 2005). Four Kir6.2 subunits form a potassium-selective, weakly inwardly rectifying pore. Each Kir6.2 subunit also bears a binding site for adenine nucleotides, and the binding of ATP or ADP to one or more of these sites leads to closure of the pore. Each Kir6.2 subunit is complexed with an SUR1 subunit, and both subunits contain ER retention signals that prevent traffic to the plasma membrane in the absence of their partner subunit. Truncation of the last 26 (or 36) residues of Kir6.2 (Kir6.2ΔC), however, deletes the ER retention signal and enables surface expression of Kir6.2 channels in the absence of SUR1 (Tucker et al., 1997).
SUR1 is a member of the ATP-binding cassette protein family (Aittoniemi et al., 2009). SUR1 endows the Kir6.2 pore with sensitivity to the stimulatory effects of nucleotides (Nichols et al., 1996; Gribble et al., 1997b, 1998) and to activation or inhibition by therapeutic drugs. In common with other ABC proteins, SUR1 contains two nucleotide-binding domains (NBDs) that are involved in the binding and hydrolysis of Mg-nucleotides. These NBDs associate in a head-to-tail fashion with the Walker A and Walker B motifs of one NBD interacting with the linker domain of the other NBD to form two composite ATP-binding sites. Occupancy of site 2 by MgADP, either by direct binding of the nucleotide or by hydrolysis from MgATP, is thought to activate the channel (Zingman et al., 2001). What happens at site 1 is less clear, but mutagenesis experiments indicate that nucleotide binding at this site is also required for channel activation and ATP hydrolysis by SUR (Nichols et al., 1996; Gribble et al., 1997b, 1998).
Metabolic regulation of the KATP channel involves both inhibition at Kir6.2 and activation via SUR1. To analyze these processes in detail, it is necessary to study them in isolation from one another. In the case of inhibition, this can be achieved by experiments in Mg-free solutions, as nucleotides do not interact with SUR1 in the absence of the cation. Consequently, there have been numerous studies of how ATP blocks the channel by binding to Kir6.2 (for review see Proks and Ashcroft, 2009). In contrast, much less attention has been paid to how Mg-nucleotides stimulate the KATP channel because it has been difficult to isolate activation from the confounding effects of simultaneous inhibition at Kir6.2. A further complication is that nucleotide activation and inhibition may not be independent but might interact.
The first step in elucidating the mechanism of Mg-nucleotide activation is to measure its effect in the absence of inhibition at Kir6.2. This can be achieved by using a mutant Kir6.2 that is insensitive to ATP block but retains normal gating properties. To date, many mutations in Kir6.2 that strongly impair ATP inhibition have been identified, but in most cases, some block still remains at ATP concentrations >1 mM. The Kir6.2-G334D mutation stands out, as it appears to be insensitive to ATP even at very high concentrations (Drain et al., 1998; Masia et al., 2007). Residue G334 lies within the putative ATP-binding site of Kir6.2, and its mutation to aspartate is believed to prevent ATP binding electrostatically (Antcliff et al., 2005). Previous studies have shown no obvious effect of the G334D mutation on the intrinsic (i.e., unliganded) single-channel kinetics of Kir6.2ΔC26 channels (Drain et al., 1998) or the open probability of Kir6.2-G334D/SUR1 channels (Masia et al., 2007). These results make KATP channels carrying this mutation potentially suitable for studying the mechanism of nucleotide activation of the wild-type KATP channel. Here, we have used Kir6.2-G334D as a tool to quantify channel activation by Mg-nucleotides at both the macroscopic and single-channel level.
MATERIALS AND METHODS
Human Kir6.2 (available from GenBank/EMBL/DDBJ under accession no. NM_000525 with E23 and I377) and rat SUR1 (available from GenBank/EMBL/DDBJ under accession no. L4_0624) were used in this study. We used rat SUR1 rather than human SUR1 because in our hands it expressed larger currents in excised patches exposed to nucleotide-free solution: we make the assumption that rat SUR1 behaves like human SUR1. Site-directed mutagenesis of SUR1 or Kir6.2, preparation of mRNA, and isolation of Xenopus laevis oocytes were performed as described previously (Proks et al., 2005b). Oocytes were injected with 0.8 ng of wild-type or mutant Kir6.2 mRNA and ∼4 ng of SUR1 mRNA. Currents were recorded 1–3 d after injection.
Patch electrodes were pulled from thick-walled borosilicate glass (Harvard Electronics). PIP2 and PIP3 antibodies were obtained from Invitrogen, and all other chemicals were from Sigma-Aldrich. Nucleotide-containing solutions were applied using a rapid solution changer (model RSC-200; Biological Science Instruments). Currents were recorded using an Axopatch 200B (Axon Instruments) and pClamp software (Axon Instruments).
All experiments were performed at room temperature (20–22°C). The pipette solution contained (in mM): 140 KCl, 1.2 MgCl2, 2.6 CaCl2, and 10 HEPES, pH 7.4 with KOH. In most experiments, the internal (bath) solution contained (in mM): 107 KCl, 1 CaCl2, 2 MgCl2, 10 EGTA, and 10 HEPES, pH 7.2 with KOH. Adenosine nucleotides were added to intracellular solution as the Mg2+ salt, and the pH was then readjusted with KOH. The Mg-free solution contained (in mM): 107 KCl, 1 K2SO4, 10 EGTA, and 10 HEPES, pH 7.2 with KOH. ATP was added as the potassium salt. Macroscopic currents were recorded from giant inside-out patches at a constant holding potential of −60 mV. Currents were filtered at 0.15 kHz and digitized at 0.5 kHz. Single-channel activity of 3–20-min duration was recorded at −60 mV, filtered at 5 kHz, and digitized at 50 kHz.
To quantify the potency of different Mg-nucleotides at activating KATP channels carrying the G334D mutation and to examine the effect of the PKA inhibitory peptide (PKIP), activation was expressed as (IN−IC)/(I0−IC)*100 (%), where IN is the steady-state KATP current in the presence of the nucleotide, I0 is the current in nucleotide-free solution immediately after patch excision (i.e., before channel rundown), and IC is the current in nucleotide-free solution before nucleotide application.
When measuring the relationship between nucleotide concentration and channel activation, the effect of rundown was compensated for using the protocol shown in Fig. 5 B. In brief, each test nucleotide concentration [X] was interrupted by a maximal stimulatory nucleotide concentration. The latter was determined in preliminary experiments and fixed at 10 mM for MgATP and 1 mM for MgADP. The increase in current due to the application of nucleotide concentration [X] was calculated as (IX−IC)/(IMAX−IC), where IX is the steady-state KATP current in the presence of test nucleotide concentration [X] (measured after switching from nucleotide-free solution), IMAX is the current in the presence of the maximal stimulatory nucleotide concentration, and IC is the current in nucleotide-free solution before switching to the test solution. The relationship between nucleotide activation and nucleotide concentration (Fig. 5 C) was then fit with Eq. 1:
where EC50 is the nucleotide concentration at which activation is half-maximal, and h is the slope factor (Hill coefficient).
The relationship between nucleotide concentration and inhibition of the wild-type KATP current in Mg-free solutions (Fig. 10) was fit with Eq. 2:
where IX is the steady-state KATP current in the presence of test nucleotide concentration [X], IC is the current in nucleotide-free solution obtained by averaging the current before and after application of the nucleotide, IC50 is the nucleotide concentration at which the inhibition is half-maximal, and h is the Hill coefficient.
The concentration–response curve for Mg-nucleotide action of the wild-type KATP channel shown in Fig. 10 was fit with the product of two Hill equations, one for activation and one for inhibition (Eq. 3):
where IX is the steady-state KATP current in the presence of the test nucleotide concentration [X], IC is the current in nucleotide-free solution (obtained by averaging the current before and after application of the nucleotide), a is the maximal increase in KATP current (produced by Mg-nucleotide activation) expressed as a fraction of IC, IC50 is the nucleotide concentration at which the inhibitory effect is half-maximal, EC50 is the nucleotide concentration at which the stimulatory effect is half-maximal, and h1 and h2 are the Hill coefficients for inhibition and activation, respectively.
For noise analysis, macroscopic currents were recorded at −60 mV, sampled at 20 kHz, and filtered at 5 kHz with an eight-pole Bessel filter. Unless blocked by ATP, KATP channels have mean open times >1 ms (see Figs. 3, 8, and 9). Thus, our conditions for data acquisition should not introduce substantial errors into our estimates of PO and N (Silberberg and Magleby, 1993). The macroscopic mean current (I) and variance (σ) were determined from data segments of 1-s duration (see Fig. 1). Control data were taken immediately before nucleotide application and test data once a steady-state condition was reached. PO and N were calculated according to the following equations:
Single-channel amplitude at −60 mV (i) was taken as 4 pA (Proks et al., 2001).
To examine the effect of the duration of the measurement period on PO and N, we simulated KATP currents using the program QuB (Qin et al., 1989) for n = 100 and n = 1,000 using the recording conditions given above and the kinetic scheme below (Gillis et al., 1989; Furukawa et al., 1993; Alekseev et al., 1998):
where O is the open state, CF is the short intraburst closed state, and CS1 and CS2 are long interburst closed states: a = 2,500 s−1, b = 500 s−1, d = 100 s−1, and f = 10 s−1. Values for c and e were calculated from PO and the mean burst duration (τB), where
PO and τB were varied between 0.4–0.86 and 10-1,000 ms, respectively. A good agreement between the simulated and the calculated values of PO and N was observed for time periods ≥1 s. For intervals ≥1 s, the values of PO calculated using Eq. 4 became significantly larger than those used in simulations.
Single-channel data were analyzed using a combination of Clampfit (OriginLab) and in-house software. Dwell-time histograms were constructed and analyzed at a resolution of 0.15 ms, as described previously (Davies et al., 1992). Open time distributions were fit with a single exponential; up to three exponentials were fitted to distributions of burst durations, and up to four exponentials were fitted to closed dwell-time distributions. The mean open time (τO) was corrected for missed events (τO,COR), as described previously (Proks et al., 2001). Burst durations were determined using the criterion for the critical time described by Magleby and Pallotta (1983):
where af and τf are the area and mean lifetime of the intraburst state, and asi and τsi are the area and mean lifetime of the interburst state i (i = 1 to n).
Single-channel kinetics were analyzed in patches that showed stationary kinetics in nucleotide-free solution over a 10-min period. It is important to point out that not all channels showed this degree of stability. In ∼50% of recordings, the activatory effect of MgADP (and MgATP) was substantially impaired. Furthermore, the extent of activation was sometimes not constant but decreased during the application of the nucleotide (as found for macropatches).
All values are given as mean ± SEM. Statistical significance was determined using Student’s t test.
The goal of this study was to obtain a quantitative description of KATP channel activation by Mg-nucleotides. Two criteria are needed to achieve this aim. First, it is essential to remove nucleotide inhibition at Kir6.2 so that activation can be studied in isolation. Second, it is necessary to determine a reliable way of measuring activation.
Nucleotide inhibition of the KATP channel is mediated by binding to Kir6.2 (Tucker et al., 1997). Thus, we exploited the fact that the Kir6.2-G334D mutation abolishes channel inhibition completely, without affecting the single-channel kinetics in the absence of nucleotide (Drain et al., 1998; Masia et al., 2007), to quantify the properties of activation. We make the assumption that this mutation does not affect channel activation by SUR1.
Two phenomena complicate measurements of channel activation: (1) the rundown in channel activity (Findlay and Dunne, 1986) and (2) the decline in nucleotide activation observed after excision of inside-out patches and exposure to nucleotide-free solution (Bokvist et al., 1991). We next explore whether these processes are altered by the G334D mutation and consider how their effects may be minimized.
Rundown of Kir6.2/SUR1 and Kir6.2-G334D/SUR1 channel activity in excised patches
A common feature of KATP channels is that they show a time-dependent decline of channel activity after patch excision into nucleotide-free solution. This rundown imposes two problems on quantification of channel activation. First, it is necessary to ensure that channel activity is as stable as possible throughout the period during which channel activation is measured, and that corrections are made for any time-dependent rundown that is present. Second, rundown will influence the capacity for channel activation. As the channel open probability can never be greater than unity, the apparent magnitude of maximal activation may be less for channels with high PO than low PO, even when the actual activation capacity is identical.
Fig. 2 compares the behavior of Kir6.2/SUR1 and Kir6.2-G334D/SUR1 channels expressed in Xenopus oocytes before and after excision into nucleotide-free solution. Before excision there is little detectable Kir6.2/SUR1 current as the channels are almost completely closed by the high intracellular ATP concentration ([ATP]i) found in the oocyte (Gribble et al., 1997a). Formation of an inside-out patch is followed by a rapid increase in current as ATP block is relieved by washout of adenosine nucleotides from the cytosolic side of the membrane. However, the current almost immediately starts to decline (Fig. 2 A).
In contrast to wild-type channels, Kir6.2-G334D/SUR1 channels are highly active in the cell-attached configuration, and there is usually only a small increase in current after patch excision (Fig. 2 B). This is because the G334D mutation almost completely abolishes the inhibitory effect of ATP (Drain et al., 1998; Markworth et al., 2000). In two thirds of patches (24/36), Kir6.2-G334D/SUR1 currents remained relatively stable after patch excision into nucleotide-free solution (declining <5%), provided that the intracellular solution was not perfused; however, perfusion induced an immediate and dramatic rundown of current (see Fig. 2 B for an example). In the other 30% of patches (12/36), Kir6.2-G334D/SUR1 currents declined gradually after excision, even when the solution was not perfused, but rundown was accelerated when perfusion was initiated (e.g., see Fig. 5 B).
For both types of KATP channel, once perfusion was initiated, the rate and extent of rundown varied dramatically between patches. To determine whether rundown is influenced by the G334D mutation, we compared its time course for Kir6.2/SUR1 and Kir6.2-G334D/SUR1 currents during the first 60 s of current decline (Fig. 2 C). In both cases, the decline in current was well fit with the sum of two exponentials and an offset value of 0.17, which represents the steady-state value of the rundown current at 60 s. There was no significant difference in the time constants and amplitudes of the two exponentials between Kir6.2/SUR1 and Kir6.2-G334D/SUR1 channels. The time constant of the fast component was 2.4 ± 1.1 s (amplitude 13 ± 3%; n = 10) for wild-type and 4.4 ± 1.2 s (amplitude 19 ± 4%; n = 10) for Kir6.2-G334D/SUR1 channels. The time constant of the slow component was 20 ± 1 s (amplitude 70 ± 2%; n = 10) for wild-type and 22 ± 1 s (amplitude 64 ± 4%; n = 10) for Kir6.2-G334D/SUR1 channels.
The similarity in the time course and extent of rundown found for Kir6.2/SUR1 and Kir6.2-G334D/SUR1 channels suggest that the mutation does not influence rundown. The data also suggest that activation is best examined at least 30 s after patch excision, when the fast component of rundown is largely complete.
We next determined if the intrinsic single-channel kinetics (i.e., those in the absence of added Mg-nucleotides) were similar in wild-type and mutant channels. Rundown of channel activity was also observed at the single-channel level. However, in some recordings, single-channel activity reached a quasi–steady state within ∼30 s and remained stable for up to ∼10 min, which enabled analysis of the channel kinetics. There was no difference between the mean open probability (PO(0)) of Kir6.2/SUR1 (0.43 ± 0.05; n = 6) and Kir6.2-G334D/SUR1 (0.41 ± 0.06; n = 6) channels when measured in quasi–steady-state conditions in nucleotide-free solutions. A similar finding was reported previously (Masia et al., 2007). In addition, there were no obvious qualitative differences in channel kinetics (Fig. 3). Both wild-type and mutant channels exhibited one open state, one short intraburst state, three to four long interburst states, and three burst components (Fig. 3, B and D, and Tables I and II).
Collectively, we conclude that the G334D mutation has no effect on the rundown or intrinsic gating of KATP channels; thus, channels with this mutation constitute a reasonable model for studying the mechanism by which Mg-nucleotides activate the KATP channel.
There are several possible explanations for the marked difference in the behavior of Kir6.2/SUR1 and Kir6.2-G334D/SUR1 channels during the period between patch excision and the start of the perfusion. For example, the high activity of Kir6.2-G334D/SUR1 channels after patch excision might be maintained by agents (such as PIP2) that bind more tightly to the mutant channel and require more vigorous perfusion to remove. It is known that PIP2 binding to wild-type KATP channels is antagonized by ATP (Fan and Makielski, 1997); thus, if ATP binding is reduced, PIP2 binding may be facilitated.
Decline of the activatory effect of Mg-nucleotides in excised patches
As shown in Fig. 4 A, the maximal extent of Kir6.2-G334D/SUR1 channel activation by MgATP (at 10 mM) gradually declined with repetitive applications: a similar decline was also observed for MgADP (not depicted). It has also been reported for wild-type KATP channels (Bokvist et al., 1991). To distinguish the decrease of channel activity in nucleotide-free solutions (rundown) from the decline of the activatory effect of Mg-nucleotides in excised patches, we term the latter “decline of activation by Mg-nucleotides” (DAMN).
To determine if rundown and DAMN affect the number of functional channels (N), the open probability (PO), or both, we used stationary noise analysis. We analyzed the effects of a maximal stimulatory concentration of 10 mM MgATP or 1 mM MgADP for six successive nucleotide applications applied at 30-s intervals. A representative experiment is shown in Fig. 4 A. Currents in the presence of Mg-nucleotide were measured at steady state over a 1-s interval, and control currents were measured for the same time interval immediately before nucleotide application (see Fig. 1). Channel activity (NPO; Fig. 4 B) and N (Fig. 4 D) were expressed as a fraction of that measured immediately after patch excision (for further details refer to Materials and methods).
As previously shown for wild-type KATP channels (Proks and Ashcroft, 1993), rundown of Kir6.2-G334D/SUR1 channels in nucleotide-free solution was accompanied by a decrease in both PO (to ∼0.4; Fig. 4 C) and N (to ∼30% of its initial value; Fig. 4 D). Although PO appeared to stabilize within ∼30 s, N took longer to reach a steady state (Fig. 4, C and D), suggesting that they may be influenced by separate processes. Data obtained in the absence of nucleotide provide an approximation of how N and PO change during rundown, but it should not be forgotten that rundown may be affected by the applied pulses of nucleotide.
The application of 1 mM MgADP or 10 mM MgATP increased both PO and N, but the potency of this effect declined with successive applications (Fig. 4, B–D). Interestingly, the number of functional channels that could be reactivated appeared to decrease faster (to ∼65% after six trials) than the decline in PO (to ∼75%). These data show that Mg-nucleotide activation declines with time for Kir6.2-G334D channels, as it does for wild-type channels (Bokvist et al., 1991); thus, this must be taken into account when constructing concentration–response curves for channel activation by Mg-nucleotides.
Activation by MgADP and MgATP
The results presented above show that to reliably quantify Kir6.2-G334D/SUR1 channel activation by Mg-nucleotides, it is necessary to correct for both rundown and for the decline in the stimulatory effect of Mg-nucleotide that occurs after patch excision. This was achieved as shown in Fig. 5 B. Each test nucleotide concentration was interrupted by the maximal stimulatory concentration of the nucleotide. Channel activation was then defined as the ratio (IX−IC)/(IMAX−IC), where IMAX is the KATP current at the maximal stimulatory nucleotide concentration, IX is the KATP current in the presence of the test nucleotide concentration ([X]), and IC is the KATP current in the absence of the nucleotide (refer to Materials and methods for further details).
Mean concentration–activation relationships are shown in Fig. 5 C. The potency of MgADP (EC50 = 8 ± 3 µM; n = 8) appears to be >10-fold that of MgATP (EC50 = 112 ± 11 µM; n = 8). The Hill coefficients for both curves are slightly higher than unity (∼1.3), indicating that the binding and cooperative interaction of more than one nucleotide molecule per KATP channel is necessary for channel activation. The EC50 for MgATP is similar to that previously obtained for Kir6.2-G334D/SUR1 channels (183 µM) (Masia et al., 2007), using a different expression system and no correction for DAMN.
Kinetics of Mg-nucleotide activation and deactivation
At a concentration of 1mM, MgADP and MgATP were equally potent at activating Kir6.2-G334D/SUR1 (Fig. 6 A). However, two exponential components were always required to fit the time course of MgATP activation, whereas in most experiments (9 out of 13) the time course of MgADP activation was well fit with a single exponential (Fig. 6, B and C). Furthermore, the fast on-rate and the off-rate were significantly faster for MgADP than for MgATP (Fig. 6, B and C). Although the off-rate for MgADP might be limited by the rapidity of the solution change, the slower off-rate of MgATP must reflect the time required for ATP hydrolysis and/or Pi unbinding. In all cases, the off-rates were slower than the fast on-rate.
Effects of nonhydrolyzable adenosine nucleotides
According to the current consensus view, MgATP has to be hydrolyzed to MgADP by the NBDs of SUR to stimulate channel activity (Zingman et al., 2001). Interestingly, the poorly hydrolyzable ATP analogue MgATPγS (1 mM) was as effective as MgATP at stimulating the activity of Kir6.2-G334D/SUR1 channels (Fig. 6 A). In both cases, the time course of channel activation was well fit with the sum of two exponentials, whereas a single exponential was sufficient to fit the time course of deactivation on return to nucleotide-free solution. The mean on-rates and off-rates for MgATPγS activation were only slightly slower than those of MgATP (Fig. 6 B), suggesting that any differences in the rate of hydrolysis are small or that other steps in the activation process are rate-limiting.
In contrast to MgATPγS, the application of the nonhydrolyzable analogue MgAMP-PNP (1 mM) to the intracellular membrane surface produced only a tiny (∼5%) increase in Kir6.2-G334D/SUR1 current, which did not reach significance (Fig. 6 A). Furthermore, MgAMP-PNP did not impair the activatory effect of 1 mM ATP, suggesting that it does not compete with ATP for the nucleotide-binding sites on SUR1. This argues that MgAMP-PNP binds poorly to SUR1.
Mg-nucleotide activation in the absence of functional NBDs of SUR1
Previous studies have shown that the activatory effect of Mg-nucleotides is mediated by the NBDs of SUR1 and is abolished by mutation of residues in the Walker A, Walker B, or linker motifs of the NBDs (Gribble et al., 1997b; Shyng et al., 1997). To confirm that this is also the case for Kir6.2-G334D/SUR1 channels, we mutated the Walker A lysine in both NBD1 and NBD2 to alanine (SUR1-KAKA) and examined the effects of a maximally effective concentration of Mg-nucleotide.
As shown in Fig. 7 A, the application of 1 mM MgATP elicited a small, slowly developing increase in Kir6.2-G334D/SUR1-KAKA current. The mean steady-state value of this increase was 8.8 ± 0.2% (n = 10). To determine if this increase was mediated via Kir6.2 or SUR1, we tested the effect of 1 mM MgATP on a truncated Kir6.2 subunit containing the G334D mutation (Kir6.2-G334DΔC), expressed in the absence of SUR1 (Fig. 7 B). The mean current increase was 7.8 ± 0.9% (n = 8), not significantly different from that observed for Kir6.2-G334D/SUR1-KAKA channels. This suggests that the slow increase seen for Kir6.2-G334D/SUR1-KAKA channels does not involve SUR1 but is intrinsic to Kir6.2.
A possible explanation for the residual activation by MgATP is that it is due to MgATP-generated PIP2, which is well established to stimulate the KATP channel (Baukrowitz et al., 1998; Shyng and Nichols, 1998). A similar idea has been advanced for wild-type KATP channels (Xie et al., 1999). To test this hypothesis, we used 0.1 mM LY294002 to inhibit both PI3 kinase (IC50 = 1.5 µM) and PI4 kinase (IC50 = 34 µM), which generate PIP2 (Rosado and Sage, 2000). In the absence of nucleotide, LY294002 had no immediate effect on Kir6.2-G334D/SUR1-KAKA channels but completely abolished the stimulatory effect of 1 mM MgATP (mean current increase 0.1 ± 0.1% vs. 5.4 ± 0.1% for control, measured in the same patch; n = 5). The application of a PIP2 antibody (15 µg/ml), however, was without effect on Kir6.2-G334D/SUR1-KAKA activation (increase 8.6 ± 0.3% vs. 8.2 ± 0.3% for control; n = 5). A similar result was obtained with a PIP3 antibody (n = 5; not depicted). Thus, it is not clear whether PIP2 and PIP3 are involved.
In contrast to its effect on Kir6.2-G334D/SUR1-KAKA channels, LY294002 caused a rapid reversible inhibition of Kir6.2-G334DΔC currents in nucleotide-free solution (52 ± 2%; n = 5), making it unsuitable for studying the role of PIP2 on this channel.
Collectively, these experiments suggest that MgATP activation of KATP channels is largely mediated by Mg-nucleotide interaction with the NBDs of SUR1, but that there is a small (5–10%) component that is mediated by a different mechanism. We have not attempted to correct for this component, as it varies in magnitude and time course between patches, and is dependent on ATP concentration.
PKA is not involved in MgATP activation of Kir6.2-G334D currents
It has been proposed that MgATP enhances KATP channel activity via PKA-dependent phosphorylation (Ribalet et al., 1989; Béguin et al., 1999; Lin et al., 2000). The results described above argue that MgATP activation of Kir6.2-G334D is largely mediated via nucleotide interaction with the NBDs of SUR1. However, to conclusively exclude the possible involvement of PKA-dependent phosphorylation, we used the protein kinase A inhibitory peptide (PKIP) (Lin et al., 2000).
There was no significant difference in the extent of activation of Kir6.2-G334D/SUR1 currents in the presence of 1 mM MgATP (79 ± 9%; n = 6) and 1 mM MgATP plus 200 µg/ml PKIP (80 ± 9%; n = 6). We also examined the effect of PKIP on the small, slowly developing stimulatory effect of MgATP on Kir6.2ΔC-G334D currents. Again, the increase in current produced by 1 mM MgATP was not different in the absence (7.8 ± 0.9%; n = 6) or presence (7.2 ± 0.7%; n = 6) of 200 µg/ml PKIP.
Lin et al. (2000) have shown that the stimulatory effect of PKA on KATP channel activity is mediated by threonine 224 in the Kir6.2 subunit, and that the effect of PKA is abolished by mutation of T224 to alanine. As a further test for the possible involvement of PKA-dependent phosphorylation, we made the T224A mutation on the Kir6.2-G334D background and examined MgATP activation. We found that 1 mM MgATP activated Kir6.2-G334D-T224A/SUR1 channels to a similar extent as Kir6.2-G334D/SUR1 channels: 77 ± 9% (n = 6) versus 79 ± 9% (n = 6). Finally, to investigate the possibility that KATP channel activity is modulated by endogenous PKA in intact cells (before patch excision), we measured PO in cell-attached patches containing a low number of channels (one to five). There was no obvious difference in the mean PO between Kir6.2-G334D-T224A/SUR1 channels (PO = 0.76 ± 0.02; n = 6) and Kir6.2-G334D/SUR1 channels (PO = 0.77 ± 0.01; n = 6).
Collectively, these experiments do not support the idea that activation by MgATP of either Kir6.2-G334D channels or Kir6.2ΔC-G334D channels involves PKA-dependent phosphorylation.
Effect of Mg-nucleotides on the single-channel kinetics
We next analyzed the effect of a maximal stimulatory concentration of 10 mM MgATP and 1 mM MgADP on the single-channel kinetics of Kir6.2-G334D/SUR1 channels (Figs. 8 and 9, and Tables III–VI). We selected only channels that showed stationary kinetics in nucleotide-free solution over a 10-min period. A similar increase in PO was produced by MgATP and MgADP: from 0.46 ± 0.18 to 0.83 ± 0.01 (n = 3) for MgATP, and from 0.41 ± 0.16 to 0.83 ± 0.01 (n = 3) for MgADP. In both cases, the increase in PO was caused by an increase in the mean open (τO,COR) and mean burst (τB) duration, a decrease in the frequency (aC2 and aC3) and mean duration (τC2 and τC3) of the two shorter interburst components, and loss of the longest interburst component (aC4 = 0). In addition, both nucleotides eliminated the short burst components (aB2 = aB3 = 0) (Tables III and V). It is noteworthy that the properties of maximally activated channels are similar to those of mutant KATP channels with strongly impaired gating (PO(0) ∼0.8; e.g., Trapp et al., 1998; Enkvetchakul et al., 2001; Proks et al., 2005a, 2007).
Our results show that MgATP and MgADP activate Kir6.2-G334D/SUR1 channels to similar extents, but that they do so with very different efficacies. The potency of MgADP (EC50, 8 µM) was much greater than that of MgATP (EC50, 112 µM), consistent with the idea that occupancy of the NBDs of SUR1 by MgADP drives channel opening, and that ATP must be hydrolyzed before it is effective. In agreement with this assumption, the EC50 for MgATP activation was similar to the Ki for MgATP displacement of 8-azido ATP binding to NBD2 of SUR1 (60 ± 26 µM) (Matsuo et al., 2000). Likewise, it is close to the Km for MgATP hydrolysis obtained for purified SUR1 (0.1 mM) (de Wet et al., 2007), and not much lower than that found for the purified KATP channel complex (0.4 ± 0.2 mM) (Mikhailov et al., 2005).
The EC50 for MgADP activation (8 µM) is much smaller than the Ki for MgADP displacement of ATP binding to either NBD of SUR1, measured when SUR1 was expressed in the absence of Kir6.2 (26–100 µM) (Matsuo et al., 2000). The Kd for MgADP binding to the KATP channel complex has not been measured, but it clearly must be less than the EC50 for channel activation (i.e., <8 µM). Thus, our data suggest that that the affinity of SUR1 for MgADP may be modulated by the presence of Kir6.2, or that MgADP binding to SUR1 may be altered by the presence of MgATP. The latter idea is not unreasonable because MgADP binding to NBD2 has been shown to stabilize MgATP binding to NBD1 (Ueda et al., 1997), raising the possibility that nucleotide binding at the two NBDs interacts. Similarly, the affinity of Kir6.2 for ATP is enhanced by SUR1 (Tucker et al., 1997), indicating that interactions between the two subunits can influence nucleotide binding at the partner subunit.
Effects of other nucleotides
Our results demonstrate that the extent of activation by MgATP and MgATPγS (1 mM) was similar, indicating that MgATPγS is hydrolyzed by SUR1. The fast time constant of activation by ATPγS was only slightly (30%) larger than that of MgATP, suggesting that any differences in the rate of hydrolysis are either small or that other steps in the activation process are rate-limiting.
In contrast, MgAMP-PNP had little or no stimulatory effect and was unable to prevent activation by MgATP. This suggests that MgAMP-PNP binds poorly to the NBDs of SUR1. Because AMP-PNP blocks wild-type channels (albeit not as effectively as ATP) (Schwanstecher et al., 1994), it may be useful as a tool when it is desired to block the channel at Kir6.2 without causing channel activation at SUR1.
On-rates and off-rates of activation
The time course of activation and deactivation were studied at near-saturating nucleotide-binding concentrations. The off-rate must represent the sum of nucleotide unbinding and the rate of channel closure. The latter can be expected to be independent of nucleotide. Because it is believed that channel activation is produced by the MgADP-bound state, one might expect the off-rate would be similar for both MgADP and MgATP. There are three possible explanations for the fact that this is not the case: (1) some channels are in the MgATP-bound state when nucleotide is removed and must undergo MgATP hydrolysis before they close; (2) some channels are in the MgATP-bound state when nucleotide is removed and undergo MgATP unbinding, which occurs more slowly than that of MgADP; and (3) if both the MgADP and the MgADP.Pi states of the hydrolytic cycle support channel activation, this might explain the slower off-rate of MgATP compared with MgADP.
The time constant of channel activation by MgADP was much smaller than that of deactivation. This might be attributable to a faster rate of MgADP binding than unbinding: a similar conclusion was reached for cardiac KATP channels (Bienengraeber et al., 2004). Alternatively, the rate of channel opening might be faster than that of channel closing. This could happen if the binding of MgADP to a single SUR1 subunit is needed to open the channel but additional molecules subsequently bind and MgADP must unbind from all subunits before the channel can close.
The on-rate for MgATP activation must include the additional time required for nucleotide hydrolysis. This may explain why activation is faster for MgADP than for MgATP. The very similar rates of activation and deactivation for MgATP and MgATPγS indicate that there is little difference in any of the reaction cycle rates for these nucleotides.
Properties of NBD-independent activation
Consistent with previous results (Gribble et al., 1997b, 1998; Shyng et al., 1997), channel activation by Mg-nucleotides requires functionally intact NBDs. A small increase in current in response to MgATP is observed when the Walker A lysines are mutated, or for Kir6.2-G334DΔC alone. The simplest explanation is that this represents current activation by PIP2 generated by endogenous kinases when MgATP is present. However, the lack of effect of a PIP2 antibody makes this less likely. Whatever the mechanism, these results serve to show that all of the activation produced by MgADP, and ∼90% of that produced by MgATP, is mediated via the NBDs of SUR1.
At saturating binding concentrations, ATP and ADP appear to have virtually identical effects on the single-channel kinetics of Kir6.2-G334D/SUR1 channels. Thus, prehydrolytic states appear to have no influence on the gating of KATP channels, at least when Mg-nucleotide inhibition at Kir6.2 is abolished by the G334D mutation. It remains possible, however, that these states may affect the gating of KATP channels when ATP is bound to Kir6.2.
As previously shown for the effects of MgATP on Kir6.2/SUR1 (Li et al., 2002) and MgUDP on Kir6.2/SUR2A channels (Alekseev et al., 1998), adenosine nucleotides affect both the open and the interburst states of the channel, producing an increase in the duration of openings and bursts, and a decrease in the frequency and duration of interburst closures. The lack of an effect of MgADP on the brief intraburst closings indicates that the nucleotide has little or no effect on the “fast” gate that governs the intraburst closures, as is also the case for the inhibitory effect of ATP at Kir6.2 (Li et al., 2002). Rather, MgADP operates upon the “slow” gate that governs the long interburst closures. The results further indicate that even under conditions of maximal activation, the slow gate can still close.
Implications for modulation of Kir6.2/SUR1 channel activity by Mg-nucleotides in excised patches
Can we now predict the behavior of the wild-type channel in the presence of Mg-nucleotides? Fig. 10 A shows the concentration-inhibition curve for ADP for wild-type channels, measured in the absence of added Mg2+ to exclude any interaction with SUR1, and the MgADP concentration–activation curve measured for Kir6.2-G334D/SUR1 channels to exclude any interaction with Kir6.2. The third curve shows the concentration–response relation for wild-type channels in MgADP, where both activation and inhibition are present. This exhibits a bell-shaped dependence on MgADP concentration with a peak at ∼50 µM MgADP, in agreement with a previous report (Hopkins et al., 1992). It is evident that the increase in the wild-type KATP current at low MgADP concentrations (<40 µM) is paralleled by a similar increase in Kir6.2-G334D/SUR1 current, and that the peak current is close to the IC50 for ADP inhibition in Mg-free solution (62 µM, this paper; 64 µM, Dabrowski et al., 2003).
It has been previously argued that the stimulatory effect of MgADP is mediated by a reduction in MgATP binding to Kir6.2 (Nichols et al., 1996; Shyng et al., 1997; Matsuo et al., 2000; John et al., 2001; Abraham et al., 2002). The fact that MgADP is able to stimulate channel activity in the complete absence of MgATP indicates that this cannot be the only mechanism of channel activation, and that MgADP (and MgATP) must have a direct effect on channel gating. Nevertheless, it remains possible that Mg-nucleotide interaction with SUR1 may also impair nucleotide binding to Kir6.2.
If we assume that the activatory and inhibitory effects of MgADP are independent, then the concentration–response curve for wild-type channels exposed to MgADP may be described by the product of two Hill equations (Eq. 3). When we used the EC50 determined for MgADP activation of Kir6.2-G334D/SUR1 (8 µM) and the IC50 measured for ADP inhibition of Kir6.2/SUR1 channels in the absence of Mg2+ (62 µM), we obtained a rather poor fit to the data for wild-type channels in the presence of Mg2+ (Fig. 10 A, dotted line). A much better fit was obtained if we assumed that MgADP activation impairs ADP binding to Kir6.2, and allowed IC50 to be a free parameter. In this case, the relationship between MgADP and the wild-type KATP current was well fit with the product of two Hill equations and an IC50 of 280 µM for ADP block (Fig. 10 A, solid line). This value is ∼4.5-fold greater than that measured in Mg-free solution, and it is consistent with nucleotide interactions at SUR1 reducing the affinity of Kir6.2 for ADP. However, further studies are required to substantiate this idea and to determine the underlying mechanism.
In contrast to MgADP, comparison of the MgATP response curve for wild-type channels with those for ATP inhibition and MgATP activation (Fig. 10 B) reveals a puzzling discrepancy at low ATP concentrations (≤10 µM). Namely, the inhibitory effect of MgATP is shifted to higher ATP concentrations with respect to that in the absence of Mg2+, yet there is no substantial activatory effect of MgATP at these concentrations (at least as measured for Kir6.2-G334D/SUR1 channels). There are several possible explanations for this effect. First, the G334D mutation may somehow impair binding of MgATP (but not MgADP) to the NBDs of SUR1. Second, ATP binding to wild-type Kir6.2 may enhance the affinity of the NBDs for Mg-nucleotide. Third, this effect may involve an unknown mechanism with a high affinity for MgATP.
The two first hypotheses assume that the shift in the ATP dose–response curve in the presence of Mg2+ at [ATP] ≤10 µM is mediated by nucleotide interaction with the NBDs. If this is the case, no shift should occur when the NBDs are not operational, as when the WA lysines are mutated. However, as shown in Fig. 10 C, Kir6.2/SUR1-KAKA channels exhibited a similar shift in ATP sensitivity when Mg2+ is present to that of the wild-type channel. This result suggests that the Mg2+-induced shift in the ATP dose–response relation, as least at low ATP concentrations, is produced via a mechanism that is independent of the NBDs but requires Mg2+. This cannot be a PKA-dependent phosphorylation because we found no difference in the effect of 10 µM MgATP on Kir6.2/SUR1 currents in the presence of 200 µg/ml PKIP (the fractional current remaining in the presence of the nucleotide was 0.58 ± 0.07, n = 6, in both the presence and absence of the inhibitor).
Collectively, these results indicate that NBD-dependent activation contributes little to the increase in the value of IC50 for channel inhibition by ATP found in the presence of Mg2+ (from 7 to 12 µM; Fig. 10 B). Instead, it is expected to alter the magnitude of the ATP-resistant current at higher ATP concentrations and to account for the small ATP-resistant current seen at [ATP]i >1 mM.
As shown in Fig. 10 B, Mg2+ has little apparent effect on the KATP channel sensitivity to ATP inhibition at [ATP] >1 mM when plotted on linear scale. However, if we display the current amplitude on log scale, it becomes obvious that there is actually a very marked (>100-fold) difference in the current at high [ATP] (1–60 mM) in the presence and absence of Mg2+ (Fig. 10 D). This difference persists over the physiological range of nucleotide concentrations found in pancreatic β cells (1–5 mM; denoted by the bar in Fig. 10 D). Because the resting membrane potential of the β cell is very sensitive to small changes in the KATP current (Tarasov et al., 2006; Proks and Ashcroft, 2009), the ATP-resistant current produced by NBD-mediated activation is expected to exert a strong influence on β cell electrical excitability.
This work was supported by grants from the European Union (BioSim LHSB-CT-2004-005137, EuroDia SHM-CT-2006-518513, and EDICT HEALTH-F4-2007-201924) and the Wellcome Trust.
Lawrence G. Palmer served as editor.
decline of activation by Mg-nucleotides
PKA inhibitory peptide