Direct Inhibition of the Pancreatic β-Cell ATP-regulated Potassium Channel by α-Ketoisocaproate*

The ATP-regulated potassium (K ATP) channel plays an essential role in the control of insulin release from the pancreatic β-cell. In the present study we have used the patch-clamp technique to study the direct effects of α-ketoisocaproate on the K ATPchannel in isolated patches and intact pancreatic β-cells. In excised inside-out patches, the activity of the K ATPchannel was dose-dependently inhibited by α-ketoisocaproate, half-maximal concentration being approximately 8 mm. The blocking effect of α-ketoisocaproate was fully reversible. Stimulation of channel activity by the addition of ATP/ADP (ratio 1) did not counteract the inhibitory effect of α-ketoisocaproate. In the presence of the metabolic inhibitor sodium azide, α-ketoisocaproate was still able to inhibit single channel activity in excised patches and to block whole cell K ATP currents in intact cells. No effect of α-ketoisocaproate could be obtained on either the large or the small conductance Ca2+-regulated K+ channel. Enzymatic treatment of the patches with trypsin prevented the inhibitory effect of α-ketoisocaproate. Based on these observations, it is unlikely that the blocking effect of α-ketoisocaproate is due to an unspecific effect on K+ channel pores. Leucine, the precursor of α-ketoisocaproate, did not affectK ATP channel activity in excised patches. Our findings are compatible with the view that α-ketoisocaproate not only affects the β-cell stimulus secretion coupling by generation of ATP but also by direct inhibition of the K ATPchannel.


EXPERIMENTAL PROCEDURES
Preparation-Adult obese mice (gene ob/ob) of both sexes were obtained from a local colony (9). The mice were starved for 24 h and then killed by decapitation. Pancreatic islets were isolated by a collagenase technique (10), and a cell suspension was prepared and washed essentially as described previously (11). The cells were resuspended in RPMI 1640 culture medium (Flow Laboratories, Scotland, UK) containing 11 mM glucose supplemented with 10% fetal bovine serum, 100 IU/ml penicillin, 100 g/ml streptomycin, and 60 g/ml gentamycin. Collagenase was obtained from Boehringer Mannheim. The cell suspension was seeded into Petri dishes (Corning Glass, Corning, NY) and incubated at 37°C in 5% CO 2 for 1-3 days.
Solutions-The bath solution (i.e. the "intracellular" solution) consisted of 125 mM KCl, 1 mM MgCl 2 , 10 mM EGTA, 30 mM KOH, and 5 mM HEPES-KOH (pH 7.15) unless otherwise indicated. In the experiments using the inside-out configuration, the pipettes were filled with standard extracellular solution containing 138 mM NaCl, 5.6 mM KCl, 1.2 mM MgCl 2 , 2.6 mM CaCl 2 , and 5 mM HEPES-NaOH (pH 7.40). In the perforated patch experiments, the pipette solution contained 10 mM KCl, 76 mM K 2 SO 4 , 10 mM NaCl, 1 mM MgCl 2 , and 10 mM NaOH (pH 7.15), and 200 g of amphotericin B/ml (dissolved in Me 2 SO). The final concentration of Me 2 SO was less than 0.1%. ATP and ADP (both supplied by Sigma) were added to the intracellular solution as indicated. When nucleotides were added as their Na ϩ salt (ADP), Mg 2ϩ was added to maintain an excess of Mg 2ϩ . ␣-Ketoisocaproate-Na ϩ salt was obtained from two different suppliers (Sigma and Fluka Chemie AG, Neu-Ulm, Switzerland), and 2-oxopentanoate was obtained from Aldrich. Trypsin-EDTA was purchased from Life Technologies, Inc.
Electrophysiology-K ATP channel activity and membrane potential were recorded using the patch-clamp technique (12). Pipettes were coated with Sylgard resin (Dow Corning, Kanagawa, Japan) near their tips to reduce capacitance transients and, finally, were fire-polished. Currents were recorded using an Axopatch 200 patch-clamp amplifier (Axon Instruments, Inc., Foster City, CA). Experiments were stored on magnetic tape using a video cassette recorder (JVC, Tokyo, Japan) and a digital data recorder (VR-10B, Instrutech Corp., Elmont, NY). The recorded signal was stored with an upper cut-off frequency of 2 kHz. Patches were excised into a nucleotide-free solution, and 0.1 mM ATP was first added to test for channel inhibition. With the solutions used, ion currents were outward, and channel records were displayed according to the convention; upward deflections denote outward currents. All experiments were performed at room temperature (20 -24°C), and channel activity was measured at 0 mV unless otherwise indicated. Whole cell K ATP currents and ␤-cell membrane potential were recorded using the perforated patch configuration of the patch- For analysis of single channel kinetics, records were low-pass filtered at 0.2 kHz using an 8-pole Bessel filter (Frequency Devices, Haverhill, MA) digitized at 0.8 kHz using a TL-1 DMA interface (Axon Instruments) and stored in a computer. Open time kinetics were determined using in-house software by digitizing segments of the current records (ϳ60-s long) and forming histograms of base-line and open-level data points. Analysis of the distribution of K ATP channel open times was restricted to segments containing no more than three active channels. Events were identified using a 50% amplitude criterion. The kinetic constants were derived by approximation of the data to exponential functions by the method of maximum likelihood (13). Channel activity (NP O ) was calculated as the mean current (I X ) divided by the single channel current amplitude. The mean current (I X ) was calculated according to Equation 1.
where N is the number of samples, i j is the current observed in sample j, and i B is the value of a user-defined base line. Results are expressed as mean Ϯ S.E., and statistical significance was analyzed using paired or unpaired Student's t tests.

RESULTS AND DISCUSSION
K ATP Channel Activity Is Inhibited by ␣-Ketoisocaproate-In the intact ␤-cell, the activity of the K ATP channel is the main regulator of resting membrane potential, and the intracellular ATP/ADP ratio is considered to constitute the primary determinator of channel activity (14). It is obvious that substances that are capable of modulating channel activity may also influence resting membrane potential. Fig. 1A shows a typical trace exposing an excised patch to 20 mM ␣-ketoisocaproate. When exposing patches to 20 mM ␣-ketoisocaproate, channel activity (NP O ) decreased to 31 Ϯ 5% (n ϭ 6; p Ͻ 0.001) compared with what was found before the addition of the substance. In Fig. 1B, we exposed isolated patches to an ATP/ADP ratio of 1, which efficiently stimulates K ATP channel activity (15). Upon the addition of the nucleotides, mean current increased with 247 Ϯ 74% (n ϭ 3) and induced a typical kinetic pattern with openings of long duration (16). The addition of ␣-ketoisocaproate (20 mM) to the patch in the continued presence of nucleotides reduced channel activity to 26 Ϯ 7% (n ϭ 4; p Ͻ 0.05). These data clearly show that the K ATP channel can be directly modulated by ␣-ketoisocaproate in inside-out patches.
Since ␣-ketoisocaproate is a Na ϩ salt, the decrease in single channel unitary conductance from 19.1 Ϯ 0. 3 picosiemens to 14.3 Ϯ 0.4 picosiemens (n ϭ 7; p Ͻ 0.01) after the addition of this fuel secretagogue could be explained by the presence of Na ϩ alone. In the millimolar range, Na ϩ is known to interact with channel conductance (17,18). Thus, the obtained decreases in channel conductance by the various concentrations of the ␣-ketoisocaproate-Na ϩ salt were identical to those reported earlier by Na ϩ (17). When correcting for the Na ϩ concentration during and before exposure of the patches to ␣-ketoisocaproate, no alteration in unitary conductance was observed (Fig. 1C).
In a series of experiments we exposed patches to four different concentrations of ␣-ketoisocaproate ranging from 2 to 20 mM. It is clear that ␣-ketoisocaproate dose-dependently decreased K ATP channel activity ( Fig. 2A). Compiled data of the concentration-inhibition relation is shown in Fig. 2B. Values are expressed as the ratio of channel activity obtained in the presence of ␣-ketoisocaproate (NP O ), and the activity assessed in the standard intracellular solution before the addition of ␣-ketoisocaproate (NP O Control). The mean value points were fitted to the Hill equation. Estimates of the concentration causing a 50% reduction in channel activity was found to be 8.1 mM, with a Hill coefficient of 2.3, which suggests that the block involves the binding of more than one molecule to the K ATP channel complex.
We further examined the effects of ␣-ketoisocaproate on kinetic properties of the K ATP channel. In Fig. 2C we analyzed the open-time distribution in the presence of ␣-ketoisocaproate using patches containing no more than three simultaneously active channels. The Inhibitory Effects ␣-Ketoisocaproate on the K ATP Channel Is Not Dependent on Metabolism-To further verify that ␣-ketoisocaproate has the ability to directly inhibit the K ATP channel without involving metabolism of the substance, we investigated K ATP channel activity in intact cells in the presence of the metabolic inhibitor sodium azide (NaN 3 ) using the perforated patch configuration of the patch-clamp technique. It is well established that the input conductance in the unstimulated ␤-cell is dominated by the K ATP channel and this conductance is virtually completely inhibited by the sulfonylurea com- pound tolbutamide (14). To measure input conductance, we used a voltage-clamp protocol with Ϯ5-mV excursions from a holding potential of Ϫ70 mV. As shown in Fig. 3A, the addition of 1 mM NaN 3 in the presence of 3 mM glucose increased input conductance 7-fold (698 Ϯ 78%; p Ͻ 0.05; n ϭ 6). However, the addition of ␣-ketoisocaproate in the continued presence of NaN 3 decreased input conductance to 50 Ϯ 3% (n ϭ 5; p Ͻ 0.01), whereas no significant change was observed when increasing the glucose concentration to 15 mM (106 Ϯ 6%; n ϭ 3; n.s.). Fig. 3B shows compiled data of a series of measurements of input conductance after the addition of glucose and ␣-ketoisocaproate, respectively, in the presence of NaN 3 .
We have previously shown that glucose-induced membrane depolarization and increase in cytosolic Ca 2ϩ concentration are slower at room temperature compared with 37°C (19), most likely reflecting a lower metabolic rate in the ␤-cell. We therefore estimated the time from the addition of ␣-ketoisocaproate or 15 mM glucose to the appearance of action potentials at high (34°C) and low (26°C) temperature. Top traces in Fig. 3C show recordings of membrane potential after the addition of 20 mM ␣-ketoisocaproate at low and high temperature. No significant change in the latency between the addition and appearance of action potential was observed using ␣-ketoisocaproate, 11.2 Ϯ 0.9 s versus 9.8 Ϯ 1.6 s at low and high temperature, respectively (n ϭ 4; n.s.). In contrast, when adding 15 mM glucose (bottom traces), a significant delay was observed when lowering the temperature to 26°C, 37.7 Ϯ 6.7 s at 34°C 86.2 Ϯ 20.5 s at 26°C (n ϭ 6; p Ͻ 0.05). It should be pointed out that it has been reported that the onset of insulin secretion is more rapid after stimulation with ␣-ketoisocaproate compared with stimulation with glucose (4). Finally, we studied the effect of NaN 3 in inside-out patches. As seen in Fig. 3D, channel activity remained unchanged after the addition of NaN 3 to the bath. In the presence of NaN 3 , ␣-ketoisocaproate (bottom trace) still potently inhibited channel activity. Taken together, these data strongly suggest that ␣-ketoisocaproate can block K ATP channel activity independent of metabolism of the keto acid.
Effects on K ATP Channel Activity by L-Leucine and 2-Oxopentanoate-Several studies have described insulinotrophic effects of L-leucine (3,20,21). Because of the blocking effect of ␣-ketoisocaproate on the K ATP channel, we also investigated whether the precursor amino acid, L-leucine, had direct effects on the channel in excised patches. As shown in Fig.  4A, the addition of 20 mM L-leucine to an isolated patch neither effected single K ATP channel activity nor channel mean open time (30.4 Ϯ 2.8 ms; n ϭ 4; n.s.). Subsequent inclusion of 20 mM ␣-ketoisocaproate in the same patch significantly inhibited channel activity. L-Arginine, which has a close structural resemblance to L-leucine, was also without effect on K ATP channel activity (data not shown). Oxopentanoate, which is derived from pentanoic acid, differs from ␣-ketoisocaproate in lacking a methyl group. Like ␣-ketoisocaproate, 2-oxopentanoate potently inhibited K ATP channel currents (Fig. 4B).
Specific Effect on the K ATP Channel-In contrast to the potent inhibition of the K ATP channel by ␣-ketoisocaproate, no effect could be seen on the large conductance Ca 2ϩ -regulated K ϩ (K BK ) channel (14). As seen in Fig. 5A, exposing inside-out patches to 20 mM ␣-ketoisocaproate, no effect on the mean current of K BK channel could be monitored. In a series of experiments, K BK channel activity (NP O ) was assessed to

FIG. 3. Effects of NaN 3 on ␣-ketoisocaproate-induced inhibition of K ATP channel activity in intact cells (A-C) using the perforatedpatch configuration and in excised inside-out patches (D).
A, cells were voltage-clamped (V-C) at a holding potential of Ϫ70 mV, and voltage excursions of ϩ/Ϫ5mV (200 ms) were performed (top left). In the presence of 1 mM NaN 3 , input conductance increased more than 5-fold, from 0.9 nS to 5.2 nS (bottom left). The addition of 15 mM glucose in the presence of NaN 3 did not affect the K ATP currents. In the presence of NaN 3 , input conductance was estimated to 3.0 nS and 3.1 nS before and after the addition of glucose, respectively (top right). Bottom right shows currents in the presence and absence of ␣-ketoisocaproate in the continued presence of sodium azide. The addition of the keto acid decreased the conductance from 5.3 nS to 2.5 nS. B, compiled data on the effects on input conductance after the addition of 15 mM glucose and 20 mM ␣-ketoisocaproate in the presence of 1 mM NaN 3 . The recordings were made at 34°C, and the effect of glucose was estimated after Ն120 s, which should be sufficient time for the fuel to act. C, effects of temperature on membrane potential when stimulated with 20 mM ␣-ketoisocaproate (top panels). At low temperature (26°C; left), the addition of the keto acid caused the cell to depolarize and display continues action potentials. The time from the addition of ␣-ketoisocaproate to the appearance of action potentials was estimated to 14.5 s. The time was not significantly affected by increasing the temperature to 34°C (13.2 s; right). Lower panels, at 26°C, the time span from the addition of 15 mM glucose to the appearance of action potentials was 55.1 s (left). Elevating the temperature to 34°C resulted in a shortening of the time to 31.5 s (right). D, 1 mM NaN 3 did not affect K ATP channel currents in inside-out patches (top trace); NP O before azide was estimated to 3.72 and after addition of azide to 3.45. The small decrease in channel activity could well be accounted for by spontaneous channel run down. In the presence of NaN 3 , channel activity was calculated to 6.17, which decreased to 1.99 after the addition of 20 mM ␣-ketoisocaproate (bottom trace). Glucose and ␣-ketoisocaproate were added using a puffer system in A, B, and C, allowing a momentaneous application and precise estimation of time from the addition of a substance to membrane depolarization. ** p Ͻ 0.01, compared with NaN 3 alone. 0.61 Ϯ 0.09 during control solution, compared with 0.60 Ϯ 0.10 in the presence of ␣-ketoisocaproate (n ϭ 3; n.s.). The activity of the small Ca 2ϩ -regulated K ϩ conductance channel (K SK ) (14) was also unaffected after inclusion of 20 mM ␣-ketoisocaproate in the bath solution (data not shown).
To further investigate whether the effect of ␣-ketoisocaproate is specific for the K ATP channel, we studied the effects of ␣-ketoisocaproate on channel activity in trypsin-modified patches. Exposure of isolated membrane patches to 20 g/ml trypsin for Ϸ5 min has been reported to alter K ATP channel activity with a specific pattern. Thus, trypsin treatment takes away the activating effect of MgADP and the inhibitory effect of sulfonylureas on K ATP channel activity, whereas the inhibitory effect of ATP remains, although with slightly decreased efficiency (22,23). Evidently this action of trypsin treatment results from its specific proteolytic effects on the K ATP channel complex (22), trypsin having a primary affinity for arginine and lysine residues (24). After trypsin modification, inclusion of ␣-ketoisocaproate (20 mM) in the bath solution was unable to affect K ATP channel activity (Fig. 5C). This experimental protocol was repeated four times, and in all four patches the inhibitory effect of ␣-ketoisocaproate on K ATP channel activity was lost subsequent to trypsin treatment (Fig. 5D). Channel activity (NP O ) was 3.1 Ϯ 0.5 and 3.0 Ϯ 0.5 (n ϭ 4; n.s.) before and after the addition of ␣-ketoisocaproate, respectively. Based on these observations, it is highly unlikely that the effect of ␣-ketoisocaproate is due to a nonspecific block of any K ϩ channel pore but rather involves a specific interaction with the K ATP channel protein complex.
Concluding Remarks-It has been known for the last decade that the deamination product of L-leucine, ␣-ketoisocaproate, is a potent stimulator of insulin secretion (21). A number of studies have demonstrated that this compound, like glucose, causes inhibition of the K ATP channel in the pancreatic ␤-cell (6 -8, 21). In addition, blocking of the respiratory cycle reduces the insulinotrophic effect of ␣-ketoisocaproate (4). These observations have suggested a role for oxidative phosphorylation-generated messengers to promote closure of the K ATP channel. In contrast to these findings, the results presented in this study clearly show that ␣-ketoisocaproate directly and reversibly inhibits the K ATP channel in the pancreatic ␤-cell without involving metabolism of the substance. Since the resting potential in the intact ␤-cell is mainly determined by the activity of the K ATP channel (14), it is clear that ␣-ketoisocaproate will also influence membrane potential and thereby insulin secretion by direct inhibition of the channel. Thus, in view of our findings, the use of ␣-ketoisocaproate as a tool to study the role of mitchondrial metabolism in intact ␤-cells should be interpreted cautiously. To what extent direct effects of ␣-keto acids on K ATP channel activity are involved in the ␤-cell stimulus-secretion coupling is at present difficult to assess. In our view, the concentrations required to affect the activity of the K ATP channel, as found in the present study, are not likely to be reached during physiological conditions. The present design does not allow us to evaluate relative contribution of the direct effect of ␣-ketoisocaproate on the K ATP channel versus effects resulting from metabolism of the substance. In this context it is interesting that ␣-ketoisocaproate stimulates insulin release more rapidly than glucose (4). The direct effect of ␣-ketoisocaproate on the K ATP channel may explain this difference in kinetics in the stimulation of insulin release.