The aminoguanidine carboxylate BVT.12777 activates ATP-sensitive K+ channels in the rat insulinoma cell line, CRI-G1

Background 3-guanidinopropionic acid derivatives reduce body weight in obese, diabetic mice. We have assessed whether one of these analogues, the aminoguanidine carboxylate BVT.12777, opens KATP channels in rat insulinoma cells, by the same mechanism as leptin. Results BVT.12777 hyperpolarized CRI-G1 rat insulinoma cells by activation of KATP channels. In contrast, BVT.12777 did not activate heterologously expressed pancreatic β-cell KATP subunits directly. Although BVT.12777 stimulated phosphorylation of MAPK and STAT3, there was no effect on enzymes downstream of PI3K. Activation of KATP in CRI-G1 cells by BVT.12777 was not dependent on MAPK or PI3K activity. Confocal imaging showed that BVT.12777 induced a re-organization of cellular actin. Furthermore, the activation of KATP by BVT.12777 in CRI-G1 cells was demonstrated to be dependent on actin cytoskeletal dynamics, similar to that observed for leptin. Conclusions This study shows that BVT.12777, like leptin, activates KATP channels in insulinoma cells. Unlike leptin, BVT.12777 activates KATP channels in a PI3K-independent manner, but, like leptin, channel activation is dependent on actin cytoskeleton remodelling. Thus, BVT.12777 appears to act as a leptin mimetic, at least with respect to KATP channel activation, and may bypass up-stream signalling components of the leptin pathway.


Background
ATP-sensitive K + (K ATP ) channels are important regulators of cell function, coupling energy metabolism with electrical activity. K ATP channels are comprised of two proteins, derived from the sulphonylurea receptor (SUR) family and an inwardly rectifying K + channel (Kir6.x family), the exact composition of these being dependent upon tissue [1,2]. For example, pancreatic β-cells and insulin-secreting clonal cell lines express K ATP channels consisting of Kir6.2 and SUR1 subunits [3]. K ATP channels are present in numerous tissues and are the target for drugs that inhibit or increase channel activity [4,5]. The archetypal inhibitors of these channels are the sulphonylurea class of drugs, which bind to the SUR subunit of the channel. Modulation of K ATP channel activity in pancreatic β-cells has profound effects on insulin secretion and glucose homeostasis [6]. Sulphonylureas such as tolbutamide and glibenclamide inhibit channel activity, resulting in β-cell depolarization, increased electrical activity, enhanced calcium entry and consequently increased insulin secretion [7]. In contrast, pancreatic β-cell K ATP channel activation induces hyperglycaemia in animals and man [8]. This latter action is caused by membrane hyperpolarization, reduction in cell excitability and decreased intracellular calcium resulting in reduced secretion of insulin. Such effects have been reported following application of the benzothiadiazine, diazoxide, which has been used on occasion to treat persistent hyperinsulinemic hypoglycaemia of infancy [8]. It has been demonstrated that diazoxide interacts with the sulphonylurea receptor subunit, SUR1, encompassing transmembrane domains 6-11 and the first nucleotide binding fold [9]. A similar conclusion has also been reached using a novel diazoxide analogue [10]. The presence of K ATP channels in many other tissues, notably muscle and central neurons, has stimulated interest in the development of novel, selective K ATP channel openers for the treatment of various diseases [10,11].
The ob gene product leptin has been demonstrated to activate K ATP channels in pancreatic β-cells [12] and insulinsecreting cell lines [13], consistent with a potential role in modifying insulin secretion [14]. One of the primary functions for this hormone is its role in the regulation of food intake and body weight [15]. Interestingly, leptin also activates K ATP channels of hypothalamic glucoseresponsive neurones [16,17] indicating a possible role for this channel in the control of energy homeostasis and body weight. In addition, Kir6.2 knock-out mice have deficits in central glucose sensing leading to loss of glucose mediated feeding response and a defective hypoglycaemic compensatory response [18]. These latter findings suggest that hypothalamic K ATP channels may also be an important target for drug manipulation with respect to centrally driven control of glucose and energy homeostasis. The aminoguanidine carboxylate, BVT.12777 (Figure 1), is one of a series of structurally related molecules based on the anti-diabetic/anti-obesity agent 3-guanidinopropionic acid [19], which, like leptin, have been demonstrated to reduce body weight in obese diabetic (ob/ob) mice [20]. Here we demonstrate that BVT.12777 opens K ATP channels in the CRI-G1 insulin secreting cell line, a useful model for pancreatic β-cells [21], and for analysing the mechanism by which leptin opens K ATP channels [13,22,23].

BVT.12777 activates K ATP channels
Under current clamp conditions with 5 mM ATP in the pipette solution to maintain K ATP channels in the closed state, the mean resting potential was -38.7 ± 1.7 mV (n = 10), similar to values reported in previous studies [13,22] under these recording conditions. Application of BVT.12777 (100 µM) hyperpolarized CRI-G1 cells ( Figure  2A) to -66.3 ± 2.7 mV (n = 10). Examination of the volt-age-clamped macroscopic currents indicates that prior to the addition of BVT.12777 the slope conductance of the cells was 0.43 ± 0.03 nS (n = 10), and following exposure to BVT.12777 (100 µM), this increased to 3.45 ± 1.17 nS (n = 10). The reversal potential (obtained from the point of intersection of the current-voltage relationship) associated with the BVT.12777-induced conductance increase (Figure 2A) was -78.5 ± 0.8 mV (n = 10), close to the calculated value for E k of -84 mV in this system, indicating increased K + conductance. CRI-G1 cells responded to BVT.12777 in an all or none manner, with cells undergoing full hyperpolarization and increase in conductance, at all concentrations (100 -300 µM) examined. Such an effect has also been reported for leptin on CRI-G1 cells [13]. Removal of BVT.12777 from the bath solution did not fully recover the membrane potential and conductance to control values over the next 15-30 minutes (not shown). Application of the K ATP channel inhibitor, tolbutamide (100 µM) during BVT.12777 exposure ( Figure  2A) completely reversed the BVT.12777-induced hyperpolarization and decreased conductance, to -41.0 ± 4.8 mV (n = 5) and 0.58 ± 0.07 nS (n = 5) respectively, values indistinguishable from control (P > 0.05). These data indicate that BVT.12777 increases K ATP current in this cell line. This is demonstrated more clearly in cell-attached recordings from CRI-G1 cells, where bath application of BVT.12777 (100 µM) resulted in activation of single K ATP channel currents ( Figure 2B; n = 7). The increase in channel activity was evident within 5 minutes of drug application, was sustained over the time course of exposure (~30 minutes) and was not immediately reversed following removal of the drug. Figure 2C shows mean channel activity (N f .P o ), normalised to the control for each recording, plotted against time of exposure to BVT.12777. BVT.12777 activation of K ATP channels was demonstrated to be reversibly inhibited by 100 µM tolbutamide (n = 4; Structure of BVT.12777 The effect of BVT.12777 on K ATP channel activity in excised membrane patches was also examined. Recordings were made from inside-out patches in symmetrical (140 mM KCl in pipette and bath solutions) K + at a membrane potential of -40 mV. K ATP channels were identified by inhibition of channel activity following application of 100 µM MgATP to the inner membrane aspect of the patch, which reduced normalised N f P o from 1.0 to 0.23 ± 0.05 (n = 4; P < 0.05). Subsequent application of 100 µM BVT.12777, in the continued presence of MgATP, induced a gradual increase in K ATP channel activity (Figure 3), to levels similar to that of control (in the absence of MgATP). For example 15 minutes after 100 µM BVT.12777 application normalised mean channel activity had recovered to 1.18 ± 0.46 (n = 4). In experiments where no drug was added, K ATP channel currents, in the presence of 100 µM MgATP, did not activate spontaneously (n = 4).

Heterologously expressed K ATP currents are not activated by BVT.12777
Oocytes injected with Kir6.2 and SUR1 cRNAs were challenged with sodium azide (3 mM) to elicit a reversible increase in current, which was completely blocked by 1 µM glibenclamide or 0.5 mM tolbutamide, indicating that the current was due to K ATP activation, as described previously [24,25]. In oocytes, previously exposed to sodium azide in order to verify Kir6.2-SUR1 expression, application of BVT.12777 (10 µM -1 mM) did not produce any consistent increase in K ATP current (n = 16; data not shown). Consequently, we utilized an alternative expression system, the HEK 293 cell line [25]. Application of BVT.12777 (100 µM) to the bathing solution using the cell-attached recording configuration resulted in no significant increase in mean channel activity above control levels over a 30-minute period, although subsequent addition of sodium azide (3 mM) did cause a rapid increase in channel activity, which was reversed by the addition of 100 µM tolbutamide (n = 4, data not shown  shown). These data are in agreement with the lack of BVT.12777 sensitivity to PI3K inhibitors on activation of K ATP channels. However, activation of MAPK has been implicated as a significant intermediate for both insulin and leptin signalling pathways in various cell types [26][27][28][29]. Thus, we examined the effect of UO126, a potent and specific inhibitor of the activation of the classical MAPK cascade [ BVT.12777 activates K ATP channels in inside-out patches

min
In addition, increasing UO126 to 10 µM had no effect on BVT.12777 induced K ATP channel activation.

BVT.12777 activation of K ATP channels is dependent on actin cytoskeleton dynamics
Leptin activation of K ATP channels in the CRI-G1 cell line is dependent upon reorganisation of the cytoskeleton, a process downstream from PI3K activation [31]. Therefore, we examined whether BVT.12777 opening of CRI-G1 K ATP channels occurs through alteration of actin filament dynamics. For this series of experiments the heptapeptide mushroom toxin phalloidin [32] was used to stabilise the polymerised form of actin (F-actin). As phalloidin is membrane-impermeant, it was directly applied to the internal aspect of the cell membrane. In whole-cell experiments, 10 µM phalloidin was added to the electrode solution and allowed to dialyse into the cell. The mean resting potential and slope conductance were -38.0 ± 0.6 mV and 0.66 ± 0.04 nS (n = 4) respectively, and following addition of 200 µM BVT.12777 no significant change in these parameters was observed ( Figure 5A), with a mean membrane potential of -41.7 ± 1.1 mV and slope conductance of 0.60 ± 0.08 nS (n = 4; P > 0.05). The presence of phalloidin (10 µM) in the bath solution also prevented K ATP channel activation by BVT.12777 in the inside-out isolated patch configuration ( Figure 5B). Application of 0.1 mM MgATP to the cytoplasmic aspect of inside-out patches caused 97.5 ± 2.1% inhibition of K ATP channel activity (n = 3; P < 0.05) and subsequent addition of 10 µM phalloidin had no further effect, as reported previously [29]. Subsequent addition of BVT.12777 (100 µM) failed to increase K ATP channel activity, with mean Nf.Po values of 0.06 ± 0.05 and 0.03 ± 0.01 in the absence and presence of BVT.12777 respectively (n = 3; P > 0.05). In contrast, the direct K ATP channel opener, diazoxide activates K ATP channels in the presence of phalloidin. In whole-cell experiments ( Figure 5C), diazoxide (200 µM) hyperpolarized CRI-G1 cells from a mean membrane potential of -42.6 ± 0.1 mV to -70.1 ± 0.8 mV (n = 4; P < 0.05), and increased slope conductance from 0.87 ± 0.23 to 7.39 ± 0.72, actions reversed by tolbutamide (100 µM).

F-actin is disrupted by BVT.12777
The prevention of BVT.12777-induced K ATP activation by phalloidin mirrors the effect of this toxin on leptin activation of K ATP [31]. Thus, we visualised F-actin by staining with rhodamine-conjugated phalloidin. In untreated CRI-G1 cells there was pronounced phalloidin-positive labelling of the cell membrane, with more diffuse, granular staining within the cytoplasm ( Figure 6A). In contrast, cells treated with BVT.12777 (100 µM) or leptin (10 nM) for 40 min showed a marked reduction in phalloidin fluorescence intensity, with disjointed labelling at the cell membrane ( Figure 6A). The actin filament disrupter cytochalasin B [33] also reduced the intensity of phalloidin labelling but in a more punctate manner on visualisation of treated cells compared with controls (data not shown). Analysis of the mean fluorescence intensity at the cell membrane following the actions of BVT.12777 and leptin demonstrated that both treatments caused a significant reduction of the intensity of rhodamine-phalloidin labelling, by 43.0 ± 4.2% (n = 6; P < 0.05) and 62.2 ± 6.0% (n = 6; P < 0.05), respectively, compared to untreated cells ( Figure 6B). However, the directly acting K ATP channel opener, diazoxide did not cause disruption of the actin cytoskeleton ( Figure 6A,6B), with a relative intensity of rhodamine-phalloidin staining of 0.98 ± 0.16 (P > 0.05).

Discussion
BVT.12777 induced hyperpolarization of CRI-G1 cells, with an associated increase in K + conductance, an action likely caused by the activation of K ATP channels, as the sulphonylurea tolbutamide completely reversed its effects. Cell-attached and inside-out single channel current recordings demonstrate directly that BVT.12777 activates K ATP channels. The increased K ATP current generated in isolated membrane patches resembles the effects of K ATP activators such as diazoxide [34] and sodium azide [35], which have also been shown to activate insulinoma or pancreatic β-cell K ATP channels in isolated patches in the presence of Mg-ATP. Thus, although not tested here, BVT.12777 as an activator of K ATP would be expected, as observed for diazoxide, to inhibit insulin release from CRI-G1 cells stimulated by metabolizable substrates or tolbutamide [36], although this would clearly be dependent on its action on other β-cell conductances, notably calcium channels. BVT.12777 activation of K ATP channels was only slowly reversed on withdrawal of the drug, unlike the actions of diazoxide or sodium azide, which are rapidly reversed on washout [35,36]. Indeed, following removal of BVT.12777 in the absence or presence of tolbutamide, enhanced K ATP channel activity was apparent for a considerable time. The slow reversibility on washout of BVT.12777 resembles the effects of the hormone leptin on CRI-G1 cell membrane potential and K ATP channel activation [13].
Leptin, via activation of the main signalling form of the leptin receptor (ObRb), has been shown to increase the phosphorylation of STAT3, MAPK and to stimulate PI3K pathways in various peripheral tissues, cell lines [37], and in hypothalamic neurones [38]. BVT.12777 although stimulating phosphorylation of STAT3 and MAPK did not stimulate PI3K dependent pathways as demonstrated by the lack of effect on the phosphorylation status of the PI3K output indicators, PKB and GSK3. It is unclear at present how this molecule induces STAT3 and MAPK phosphorylation. As K ATP activation by BVT.12777 is rapid and occurs in isolated membrane patches it is unlikely that any JAK-STAT pathway (which drives changes in tran- scription) contributes to this action. Leptin activation of K ATP channel currents in CRI-G1 cells has previously been shown to be independent of MAPK, but prevented by the inhibitors of PI3K [22]. However, BVT.12777 activation was not only insensitive to inhibition by the MAPKK inhibitor, UO126, it was also insensitive to the presence of the PI3-kinase inhibitors, wortmannin and LY294002, at concentrations sufficient to prevent leptin activation of K ATP in this cell line. These data led us to suspect that BVT.12777, irrespective of its ability to initiate various signalling cascades in this cell line, increased K ATP channel activity by a more direct effect on the channel subunits in a manner analogous to diazoxide, which is purported to interact directly with the SUR1 subunit [9,10]. This possibility was tested by heterologous expression of the β-cell subunits of K ATP channels, Kir6.2 and SUR1, in Xenopus oocytes, a commonly utilised expression system for electrophysiological studies of these recombinant channels [24,25]. However, BVT.12777 did not activate Kir6.2-SUR1 currents in oocytes, demonstrated to express functional K ATP channel currents. Thus we explored this question further by utilising a second heterologous expression system for Kir6.2-SUR1, HEK293 cells. Recordings from inside-out patches demonstrated that BVT.12777 did not activate Kir6.2-SUR1 currents in the presence of Mg-ATP, in contrast to diazoxide [39] or BVT.12777 disrupts the actin cytoskeleton

Normalised Intensity
Diazoxide * * B sodium azide [35]. Overall these data strongly suggest that expression of the K ATP channel subunits, Kir6.2 and SUR1 are insufficient per se to bring about sensitivity to BVT.12777, and indicate that this opener may activate this channel type by an indirect mechanism (which is not available in oocytes or HEK cells).
Although the activation of K ATP channels by leptin in CRI-G1 cells is PI3-kinase dependent the lipid products of this enzyme system, such as PtdIns(3,4,5)P 3 also do not interact directly with K ATP channels [22]. Recent studies demonstrate that both leptin and PtdIns(3,4,5)P 3 increase K ATP channel activity indirectly, through changes in cytoskeletal dynamics [31]. It is well established that many ion channels and transporters are anchored in the membrane by either direct or indirect association with the cytoskeleton. In addition, there is growing evidence that altering the integrity of cytoskeletal elements, in particular actin filaments, can modulate the activity of a variety of ion channels [40] and receptors [41]. For example, disruption of actin filaments with cytochalasin is shown to increase K ATP channel activity in cardiac myocytes [42] and CRI-G1 cells [31]. Indeed, a number of lipid kinases, including PI 3-kinase, are also localised to the cytoskeleton and their activities are modulated by a variety of cytoskeletal proteins, especially those associated with actin [40]. Actin filament structure is controlled by reversible polymerisation of G-actin, which forms F-actin, and this process is under the dynamic control of various actinbinding proteins [43]. The heptapeptide mushroom toxin phalloidin [32] binds to filamentous F-actin with high affinity and stabilises the actin in this form. The addition of phalloidin to the intracellular aspect of CRI-G1 cells prevented BVT.12777, but not diazoxide, from activation of K ATP channel currents in whole cell and inside out recording configurations indicating that this molecule likely causes the opening of K ATP channels by a membrane delimited alteration of cytoskeletal dynamics. This mechanism of action is identical to that proposed for leptin and PtdIns(3,4,5)P 3 activation of K ATP in this cell line [31]. Fluorescence staining of CRI-G1 cells with rhodamine-conjugated phalloidin revealed disassembly of actin filaments by both BVT.12777 and leptin, but not diazoxide. These data provide direct support for an important role for cytoskeletal dynamics in the control of K ATP channel activity by both leptin and BVT.12777. The lack of effect of diazoxide on the actin filament structure is also supportive of this opener acting directly on the K ATP channel subunits.

Conclusions
BVT.12777 activation of K ATP channels in CRI-G1 cells was evident regardless of whether it was applied to the external or internal surface of the cell. BVT.12777 signalling to K ATP channels is not mediated by PI 3-kinase or MAPK, but does appear to depend on actin filament re-modelling. As leptin hyperpolarizes a sub-population of hypothalamic neurones by opening K ATP channels [16], it is feasible that at least part of the anti-obesity action of BVT.12777 may be through the activation of this potassium channel. Furthermore, as BVT.12777 acts downstream of PI3K, such an agent may act to overcome the putative central leptin resistance associated with the obese state [37]. Thus, although BVT.12777 and its close structural analogues are unlikely per se to be useful anti-obesity agents as they display hepatotoxicity [44], understanding the general principles underlying their mechanism of action may reveal clues for future anti-obesity drug development.

Cell culture and transfection
Cells from the insulin secreting cell line, CRI-G1, and the human embryonic kidney cell line, HEK 293, were grown as described previously [25,35]. The preparation of mouse Kir6.2 (provided by Professor F. Ashcroft, University of Oxford), rat SUR 1 (provided by Dr G. Bell, University of Chicago) and CD4 cDNAs and transfection procedures were as described by [25]. Transfected cells were selected by visible binding of anti-CD4 coated beads (Dynal, Oslo) following incubation with the beads for 20 min.

Oocyte collection and preparation
Ovarian lobes were removed from mature female Xenopus laevis frogs (Blades Biological, UK) following killing of the animal by destruction of the brain. The use of animals was in accordance with the Home Office Animals (Scientific Procedures) Act (1986) and approved by the local ethics committee. Separation and selection of oocytes and the preparation and injection of cRNAs were performed as described by [25].

Cytoskeletal fluorescence imaging and analysis
CRI-G1 cells were gently washed in normal saline (containing in mM): NaCl 135, KCl 5, MgCl 2 1, CaCl 2 1, HEPES 10, pH 7.4, and incubated for 40 min with either 100 µM BVT.12777, 10 nM leptin, 200 µM diazoxide or 3 mM sodium azide for 30 min with the cytoskeletal disrupter, cytochalasin B (10 µM). Cells were then fixed, permeabilised, stained with rhodamine-conjugated phalloidin (2.66 U ml -1 ) and visualised using a BioRad Microradiance, confocal imaging system as described by [31]. The intensity of rhodamine-conjugated phalloidin staining in the plasma membrane was determined using BioRad Lasersharp processing software (Bio-Rad, CA, USA). Analysis lines were drawn along randomly selected regions of the plasma membrane and the fluorescence intensity determined. A histogram giving the mean fluorescence intensity was constructed for a minimum of 5 cells on each stimulated or control dish on at least 3 separate occasions. Within a given experimental series all conditions for capturing images were constant. In order to allow for quantification of experimental data obtained on separate days, the results were normalised relative to the mean plasma membrane fluorescence measured in the control cells for each day and presented as mean ± S.E.M. Statistical analyses were performed using Student's unpaired t test. p < 0.05 was considered significant.

Electrophysiological recording and analysis
Whole cell currents from Xenopus oocytes were measured using a two-electrode voltage clamp technique as described by [25]. Recordings were made in a high-potassium bath solution, KD96 containing (mM): KCl 96, NaCl 2, CaCl 2 1.8, HEPES 5 (pH 7.4 with KOH). Working concentrations of drugs were prepared in KD96 and superfused into the bath. Whole-cell current-clamp recordings with excursions to voltage clamp mode were used to monitor membrane potential and macroscopic currents from CRI-G1 cells. Cell-attached and excised inside-out recordings were made from CRI-G1 cells and HEK cells expressing Kir6.2 and SUR1 to examine single channel responses as described previously [25,35]. Single channel data were analysed for current amplitude and channel activity (N f .P o ; where N f is the number of functional channels in the patch and P o is the open probability) as described previously [45]. All data were normalised to control and are expressed as mean ± S.E.M. Statistical analyses were performed using Student's unpaired t test. P < 0.05 was considered significant. Recording electrodes were pulled from borosilicate glass and had resistances of 2-5 MΩ for whole cell recordings and 7-10 MΩ for cell-attached and inside-out experiments when filled with electrolyte solution. The pipette solution for whole-cell recordings comprised (in mM): KCl 140, MgCl 2 0.6, CaCl 2 2.73, Mg-ATP 5.0, EGTA 10, HEPES 10, pH 7.2 (free [Ca 2+ ] of 100 nM), whereas for single channel recordings the pipette solution contained (in mM): KCl 140, CaCl 2 1, MgCl 2 1, HEPES 10, pH 7.2. The bath solution for wholecell and cell-attached recordings was normal saline whereas for inside-out patches the bath solution contained (in mM): KCl 140, MgCl 2 1, CaCl 2 2, EGTA 10, HEPES 10, pH 7.2 (free [Ca 2+ ] of 30 nM). All solution changes were achieved by superfusing the bath with a gravity feed system at a rate of 10 ml min -1 , which allowed complete exchange within 2 min. All experiments were performed at room temperature (22-25°C).

Antibodies & drugs
Anti-PKB, which recognises all three isoforms of PKB, and the phospho-specific PKB (Thr308), GSK3α/β (Ser21/9), STAT3 (Tyr705) and p44/42 MAPK (Thr202/Tyr204) antibodies were obtained from Cell Signalling Technology Inc. Recombinant human leptin, wortmannin and LY 294002 were obtained from Novachem-Calbiochem and BVT.12777 ([2-(hydrazinoiminomethyl) hydrazino] acetic acid) was a gift from Biovitrum (Stockholm, Sweden). Tolbutamide, Mg-ATP, diazoxide, sodium azide, phalloidin and cytochalasin B were obtained from Sigma. Rhodamine-conjugated phalloidin was obtained from Molecular Probes and UO126 from Promega. BVT.12777 was prepared as a 100 mM stock solution in normal saline and stored at -70°C prior to use. Leptin was prepared as a 10 µM stock solution in normal saline containing 0.2 % bovine serum albumin as carrier. Rhodamine-conjugated phalloidin (200 U ml -1 ) and LY 294002 (10 mM) were stored as stock solutions in 1% methanol at -20°C. Cytochalasin B was stored as a 10 mM stock solution, and diazoxide and tolbutamide as 100 mM solutions, all in DMSO at 2-4°C. Mg-ATP was stored at -20°C as a 100 mM solution in 10 mM HEPES (pH 7.2). Wortmannin and UO126 were stored as 10 mM stock solutions in Me 2 SO at -20°C.