The 2H-1,4-benzoxazine derivatives are novel drugs structurally similar to nucleotides; however, their actions on the pancreatic β cell ATP-sensitive K+ (KATP) channel and on glucose disposal are unknown. Therefore, the effects of the linear/branched alkyl substituents and the aliphatic/aromatic rings at position 2 of the 2H-1,4-benzoxazine nucleus on the activity of these molecules against the pancreatic β cell KATP channel and the Kir6.2ΔC36 subunit were investigated using a patch-clamp technique. The effects of these compounds on glucose disposal that followed glucose loading by intraperitoneal glucose tolerance test and on fasting glycemia were investigated in normal mice. The 2-n-hexyl analog blocked the KATP (IC50 = 10.1 × 10−9 M) and Kir6.2ΔC36 (IC50 = 9.6 × 10−9 M) channels, which induced depolarization. In contrast, the 2-phenyl analog was a potent opener (drug concentration needed to enhance the current by 50% = 0.04 × 10−9 M), which induced hyperpolarization. The ranked order of the potency/efficacy of the analog openers was 2-phenyl > 2-benzyl > 2-cyclohexylmethyl. The 2-phenylethyl and 2-isopropyl analogs were not effective as blockers/openers. The 2-n-hexyl (2–10 mg/kg) and 2-phenyl analogs (2–30 mg/kg) reduced and enhanced the glucose areas under the curves, respectively, after glucose loading in mice. These compounds did not affect the fasting glycemia as is observed with glibenclamide. The linear alkyl chain and the aromatic ring at position 2 of the 1,4-benzoxazine nucleus are the determinants, which confer the KATP channel blocking action with glucose-lowering effects and the opening action with increased glucose levels, respectively. The opening/blocking actions of these compounds mimic those that were observed with ATP and ADP. The results support the use of these compounds as novel antidiabetic drugs.
Drugs that target the pancreatic β cell ATP-sensitive K+ (KATP) channels are used in glucose/insulin dysmetabolisms (Gribble and Reimann, 2003; Jahangir and Terzic, 2005; Arnoux et al., 2010; Ashcroft, 2010). This channel is a complex that is composed of the sulfonylureas receptor type-1 (SUR1) and the inwardly rectifying K+ channel (Kir6.2) subunits (Seino and Miki, 2003). The SUR1 subunit carries the binding sites for the KATP channel openers and blockers (Babenko et al., 2000; Moreau et al., 2005). KATP channel blockers are prescribed to individuals who are in the diabetes-aged population and who display impaired glucose-induced insulin release (Nourparvar et al., 2004). These drugs belong to the following distinct classes: the first-generation sulfonylureas (chlorpropamide, tolbutamide, and tolazamide), which are low-affinity ligands of the SUR1 subunit that cause the release of insulin at micromolar concentrations; the second- (glipizide and glyburide) and third-generation sulfonylureas (glimiperide and acetohexamide), which are high-affinity and slowly reversible ligands of the SUR1 subunit that act at nanomolar concentrations; and the glinides, which lack the sulfonylurea moiety and include repaglinide, nateglinide, and meglitinide, which are high-affinity SUR1 ligands, exhibit a rapid onset/offset action compared with that of the sulfonylureas (Gribble and Reimann, 2003).
The sulfonylureas and the glinides are being investigated for use in the treatment of hypotension that results from septic shock, ischemic trauma, and neonatal diabetes (Koster et al., 2005; Flechtner et al., 2006; Pearson et al., 2006; Flanagan et al., 2007; Mlynarski et al., 2007; Simard et al., 2008; Ashcroft, 2010). Severe hypoglycemia, weight gain, and cardiovascular side effects limit their use in special populations (Zünkler, 2006). Long-term exposure of β cells to sulfonylureas induces a reduction in the insulin content and the number of KATP channels (Takahashi et al., 2007). Therapeutic concentrations of sulfonylureas and high doses of glinides induce the apoptosis of β cells, β cell lines, or cell lines that express the recombinant KATP channel subunits, and these effects are mediated by SUR1 (Maedler et al., 2005; Hambrock et al., 2006). Glibenclamide also causes atrophy of the skeletal muscles (Tricarico et al., 2010). It has been proposed that sulfonylurea-induced atrophy contributes to the loss of β cell mass that characterizes the progression of diabetes (Takahashi et al., 2007).
The pancreatic KATP channel openers (KCOs) belong to the benzothiadiazine class, which includes the diazoxide and thiadiazine-1,1-dioxide derivatives, such as N-(3,3-dimethylbutan-2-yl)-1,1-dioxo-4H-pyrido[4,3-e][1,2,4]thiadiazin-3-amine (BPDZ-62), 7-chloro-3-isopropylamino-4H-1,2,4-benzothiadiazine-1,1-dioxide (BPDZ-73), 6-chloro-3-(1-methylcyclobutyl) amino-4H-thieno[3,2-e]-1,2,4thiadiazine 1,1-dioxide (NNC-55-0462), 6-chloro-1,1-dioxo-N-propan-2-yl-4H-thieno[3,2-e] [1,2,4]thiadiazin-3-amine (NNC-55-0118), and 6-chloro-N-(1-methylcyclopropyl)-1,1-dioxo-4H-thieno[3,2-e][1,2,4]thiadiazin-3-amine (NN414); cyanoguanidines; nitropyrazoles; and 4-sulfamoylphenylbenzamides (Hansen, 2006; Carosati et al., 2007). Diazoxide is prescribed for the treatment of acquired and tumor-related hypoglycemia, familial hyperinsulinemia hypoglycemia in infancy, and polycystic ovary syndrome (Jahangir and Terzic, 2005; Hussain, 2007; Arnoux et al., 2010). Diazoxide protects the β cells from sulfonylurea-induced apoptosis, high glucose, and cytokine-induced toxicity, which preserves the insulin stores (Teshima et al., 2003). In skeletal muscle, diazoxide prevents glibenclamide-induced atrophy (Tricarico et al., 2010). The SUR1-selective openers have been proposed to treat epilepsy and neurodegenerative disorders, in which the SUR1/Kir6.2-Kir6.1 subunits play a role (Carosati et al., 2007). However, diazoxide is unselective, which can lead to hypotension and tachycardia, severe fluid retention, sedation, and weakness.
The 2H-1,4-benzoxazine derivatives have emerged as novel KATP channel modulators. In the presence of ATP, these molecules display KCO activity, whereas in the absence of the nucleotide they display a blocking action (Tricarico et al., 2003, 2008). These compounds have structural similarities with ATP and ADP molecules. A conformational analysis showed that the planar area of the 2H-1,4-benzoxazines overlaps that of the adenine nucleotide tri- and diphosphates. The electronic distribution profile of this area is also similar to that of the adenine nucleotides, which were reached by electrons; this result suggests that these compounds share a common ATP/ADP interaction interface on the receptor sites (Tricarico et al., 2008). No data are available regarding the effects or possible dual actions of these nucleotide analogs on the pancreatic β cell KATP channel or glucose disposal in normal mice.
The molecular determinants that were responsible for the activity of the 2H-1,4-benzoxazine derivatives toward the β cell KATP channels and the Kir6.2ΔC36 subunit that was expressed in HEK293 cells therefore were investigated using patch-clamp techniques. The influence of the linear/branched alkyl chain substituents at position 2 of the 2H-1,4-benzoxazine nucleus on the biological activity of the molecules against the KATP channel was evaluated by testing the effects of the 2-n-hexyl- and 2-isopropyl-1,4-benzoxazine derivatives. The influence of an aromatic/aliphatic ring on the biological activity of the molecules against the KATP channels was evaluated by testing the effects of 2-cyclohexylmethyl-, 2-phenyl-, 2-benzyl-, and 2-phenylethyl-1,4-benzoxazines (Fig. 1). The drug effects on glucose disposal and fasting glycemia in normal mice also were investigated.
Materials and Methods
Pancreatic β Cells and tsA201 Cells Expressing the Kir6.2ΔC36 Channel.
All of the experiments were conducted in accordance with the Italian Guidelines for the Use of Laboratory Animals, which conform with the European Community Directive that was published in 1986 (86-609-EEC). The method of pancreatic β cell preparation has been described previously (Rolland et al., 2002, 2006). Sixty to 100 islets per culture from two mice were prepared. The islets were identified by optical morphological inspection and insulin secretion. The β cells were identified by their spherical form/size, capacitance of 8 ± 3 pF (number of cells = 45) (α or δ cells have a lower capacitance), current/membrane potential response to high or low glucose concentrations (see step 1 below), current responses to diazoxide or tolbutamide (see step 5 below), and insulin secretion in response to increasing concentrations of external glucose (Fig. 2). Studies on the Kir6.2ΔC36 channel in HEK293 cells used Kir6.2ΔC36 (mouse) that was inserted in the mammalian expression vector pCDNA3, which was provided generously by Professor F. M. Ashcroft (Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK). The HEK293 cells were cotransfected with 4 μg of DNA that encoded the channel and a lower amount of plasmid DNA that encoded the CD8 receptors using Lipofectamine 2000 and Opti-MEM (Invitrogen, Carlsbad, CA).
Triplicates of the eight islets were cultured in 10% fetal calf serum and 1 ml of Krebs-Ringer-bicarbonate buffer with HEPES medium with 2 mg/ml bovine serum albumin and 5 mM glucose overnight in air (5% CO2) at 37°C before the experiments. An equilibrium period of 120 min with 0 mM glucose of the islets was performed; this was followed by a period of incubation with different glucose concentrations (0.5–10 mM) for 60 min in 5% CO2 at 37°C to evaluate insulin secretion in the medium. The islets were sonicated to facilitate insulin extraction. The insulin concentrations in the sonicate and the medium were determined with the use of rat/mouse insulin enzyme-linked immunosorbent assays (Millipore Corporation, Billerica, MA) (Fig. 2C).
Solutions and Drugs.
For the perforated or conventional whole-cell recordings, the extracellular bath solution contained 140 mM NaCl, 4.8 mM KCl, 2.5 mM CaCl2, 1.2 mM MgCl2, 10 mM HEPES (pH 7.4), and various concentrations of glucose. The electrical contact was established by adding amphotericin B to the pipette solution (stock, 60 mg/ml in dimethyl sulfoxide; final concentration, 300 μg/ml). For the perforated patch mode, the pipette solution contained 70 mM K2SO4, 10 mM NaCl, 10 mM KCl, 3.7 mM MgCl2, and 5 mM HEPES (pH 7.1); 5 mM K2ATP and 1 mM Na3GTP were added to the pipette solution for the conventional whole-cell mode. K2SO4 was replaced by an equimolar concentration of KCl using BaCl2. For the inside-out recordings, the bath solution contained 140 mM KCl, 5 mM EGTA, and 20 mM HEPES (pH 7.2). The intrapipette solution contained 140 mM KCl, 2.6 mM CaCl2, 1 mM MgCl2, 10 mM EGTA, and 20 mM HEPES (pH 7.4). The synthesis of the 2H-1,4-benzoxazine derivatives has been described previously (Fig. 1) (Tricarico et al., 2003). All of the other chemicals were purchased from Sigma-Aldrich (St. Louis, MO).
The drug experiments were performed using the patch-clamp methodology in the voltage-clamp mode (perforated whole-cell, conventional whole-cell, and inside-out configurations), and the current-clamp mode was performed using Clampex software (Molecular Devices, Sunnyvale, CA) (Rolland et al., 2006; Tricarico et al., 2008).
The pancreatic β cells were perfused continuously with different solutions at a flow rate of 2 ml/min, and the changes in the currents were monitored. The patch perforation and whole-cell configurations were evaluated by monitoring the changes in the current capacitance and access resistance, which were 22 ± 6 and 30 ± 3 MΩ in the whole-cell and perforated patch experiments, respectively. The protocol that was applied for the identification of β cells and the drug test was as follows:
external glucose 10, 6, or 3 mM and NaN3 (5 mM)
external glucose 10, 6, or 3 mM and NaN3 (5 mM) (control)
external glucose 10, 6, or 3 mM, NaN3 (5 mM), and drugs
external glucose 10, 6, or 3 mM, NaN3 (5 mM), and diazoxide (2.5 × 10−4 M) or tolbutamide (5 × 10−4 M).
No more than two drug concentrations per cell were tested. The drug solutions were applied to the patches using a fast perfusion system (AutoMate Scientific, Berkeley, CA).
It should be noted that the low patch-clamp performance in 10 mM glucose can mask the blocking actions of the investigated drugs. In this experimental condition, we indeed measured a very low residual current that did not allow us to quantify the blocking action of the drugs on this parameter.
Intraperitoneal Glucose Tolerance Test.
This test was performed on six groups of 8-week-old nondiabetic mice after an overnight fast as described previously (Rolland et al., 2006). The noncyclic 2H-1,4-benzoxazine derivatives (0.2, 2, 10, or 30 mg/kg), the 2-phenyl analog (2, 10, and 30 mg/kg), glibenclamide (0.2 and 2 mg/kg), and diazoxide (2 mg/kg) were dissolved in a solution that contained polyoxyethylene sorbitan monooleate (10%) and 0.9% NaCl (90%).
This treatment was followed with evaluations at 24 h and 14 days postdose to identify any possible intolerability of the single intraperitoneal injection of the compounds. The primary end points that were evaluated were mild symptoms, such as piloerection, sedation, and weakness. The secondary end points were severe symptoms, such as abdominal cramps, convulsion, and moribundity.
Analysis of the Results and Statistics.
The off-line analysis was performed using Clampfit, Fetchan, pSTAT (Molecular Devices), and Excel (Microsoft, Redmond, WA). The mean current (single-channel current × open probability × number of channels) was evaluated in an excised patch as the digital average of the sampled points, whereas the single-channel conductance (pS) was evaluated as the slope of the current-voltage relationship. The experiments are illustrated by the traces, means, or representative results that were obtained from at least three different cultures. The concentration-response data were fitted by an equation that described the interaction of the ligand with one inhibitory site, as in the case of the 2-hexyl analog data, or by the sum of two equations that described the interaction of the ligand with two sites that exerted opposite actions, as in the case of the 2-cyclic analog data (Tricarico et al., 2008). The stimulatory component can be described by the following term: (Idrug − 1) × 100 = Amax/(1 + (DE50/[Drug])n); and the inhibitory component can be described by the following term: (Idrug − 1) × 100 = Imax/(1 + ([Drug]/IC50)n. Imax indicates the percentage maximal inhibition of the pancreatic KATP channel and Kir6.2ΔC36 channel currents that were caused by the compounds. Amax indicates the percentage maximal activation of the KATP currents that was produced by the compound under investigation. DE50 indicates the concentration of the drug that was needed to enhance the current by 50%, which was calculated with respect to the current levels that were recorded in the absence of drugs. IC50 is the concentration of the drug that is needed to reduce the current by 50%, which was calculated with respect to the current levels that were recorded in the absence of the drugs. N indicates the number of sampled cells per drug, whereas n indicates the slope factor of the curves. The algorithms of the fitting procedures that were used were based on a Marquardt least-squares fitting routine. The data analysis and the plot were performed using SigmaPlot software (Systat Software, Inc., San Jose, CA). The data are expressed as the mean ± S.E. The statistical significance of the differences between the means was assessed by the Student's t test. The differences were considered significant for P < 0.05.
Effects of 2H-1,4-Benzoxazine Derivatives on the Native Pancreatic β Cell KATP Channels of Mice and on the Kir6.2ΔC36 Channel That Was Expressed in HEK293 Cells.
We first evaluated the effects of 2-isopropyl and 2-n-hexyl analogs on the β cell KATP channel in the presence of 3, 6, or 10 mM glucose in the bath using the perforated and/or conventional whole-cell patch clamp mode. The exposure of the patches to a solution that was enriched with the 2-linear alkyl chain analog (10−10–10−4 M) inhibited the KATP channel currents of β cells in the perforated and conventional whole-cell modes in a concentration-dependent manner (Fig. 3, A, B, D, and F). In the perforated whole-cell patch mode, at −50 mV (Vm) the current was reduced from 42 ± 3 pA in the control to 18 ± 3 pA after the application of a 10−7 M concentration of the drug (Fig. 3A). In the conventional whole-cell mode, at −50 mV (Vm), the current was reduced from 38 ± 2 pA in the control to 25 ± 2 and 19 ± 3 pA after the application of 10−8 and 10−7 M concentrations of the drug, respectively (Fig. 3B). In the intact cell, the inhibitory effect of this compound was fully reversible after a 5-s washout of the drug solution. The 2-n-hexyl analog (from 10−10 to 10−4 M) reduced the Kir6.2ΔC36 channel currents that were recorded in the HEK293 cells pulsed to a test potential of −100 mV (Vm) using the conventional whole-cell patch mode. This drug reduced the Kir6.2ΔC36 current from −980 ± 22 pA in the control to 750 ± 36 and 200 ± 21 pA in the presence of 10−8 and 10−6 M concentrations of the drug, respectively (Fig. 3C). In the presence of 6 mM glucose, in the perforated patch mode, the current was reduced from 21 ± 3 pA in the control to 10 ± 2 pA after the application of a 10−7 M concentration of the blocker (Fig. 3D). In the perforated patch mode, the 2-branched alkyl chain analog failed to affect the KATP currents in the presence of either 3 mM glucose and 5 mM NaN3 or 6 mM glucose (Fig. 3, A and D), and it did not affect the Kir6.2ΔC36 channel. In the presence of 10 mM glucose, the 2-n-hexyl and 2-isopropyl analogs (10−8–10−6 M) did not significantly affect the currents (Fig. 3E). A 5 × 10−4 M concentration of tolbutamide almost fully reduced the currents in 3 and 6 mM glucose, and it showed a mild effect in 10 mM glucose (Fig. 3, A, D, and E). The mild blocking effect of tolbutamide that was observed in 10 mM glucose may be related to the low patch-clamp performance that was described under Materials and Methods.
The concentration-response curve analysis showed that at −50 mV (Vm), in 3 mM glucose, and using the perforated patch method, the IC50 of the 2-hexyl analog that was present in the range of concentrations tested was 10.1 ± 3 × 10−9 M; in the conventional whole-cell patch mode, the IC50 was 9.8 ± 2 × 10−9 M in the β cells and 9.6 ± 1 × 10−9 M in the cell line that was expressing the Kir6.2ΔC36 channel (Fig. 3F; Table 1). These values were not statistically different.
The coapplication of 10−9 M glibenclamide and the 2-n-hexyl analog (10−8 M) to the β cells caused a 53 ± 4% reduction in the currents (number of cells = 3) in the conventional whole-cell patch mode. Glibenclamide (10−9 M) and the 2-n-hexyl analog (10−8 M) caused a 25 ± 3% (number of cells = 2) and 27 ± 4% (number of cells = 3) reduction in the currents, respectively. The KATP current indeed was reduced from 39 ± 2 pA in the control to 29 ± 2 pA in the presence of 10−9 M glibenclamide and 16 ± 2 pA after the coapplication of 10−9 M glibenclamide and 10−8 M of the 2-n-hexyl analog to the cell (Fig. 3B).
Conversely, in the perforated patch-clamp experiment and in the presence of 3 or 6 mM external glucose, the 2-phenyl-, 2-benzyl-, and 2-cyclohexylmethyl-1,4-benzoxazine derivatives enhanced the KATP currents at sub-nanomolar concentrations. In 3 mM glucose and 5 mM NaN3, the percentage activations of the currents were 91.2 ± 4% (number of cells = 12) and 58.2 ± 6% (number of cells = 7) for 10−10 M of the 2-phenyl and 2-benzyl analogs, respectively. The currents that were measured at −50 mV (Vm) in 3 mM glucose and 5 mM NaN3 were 40 ± 4 pA in the control and 65 ± 4 and 75 ± 5 pA with 0.7 × 10−10 and 10−10 M concentrations of the 2-phenyl analog, respectively; they were 61 ± 5 pA after the application of a 10−10 M concentration of the 2-benzyl analog (Fig. 4A). For the 2-phenyl and 2-benzyl analogs in 6 mM glucose, the activations were 96.1 ± 8% (number of cells = 4) and 61.2 ± 6% (number of cells = 4), respectively. The currents that were measured at −50 mV (Vm) in 6 mM glucose were 19.5 ± 3 pA in the control and 38 ± 4 and 32 ± 4 pA after the application of a 10−10 M concentration of the 2-phenyl and 2-benzyl analogs, respectively (Fig. 4B). In 3 mM glucose and 5 mM NaN3 or 6 mM glucose, 10−7 M of the 2-phenyl- analog produced 17 ± 5% (number of cells = 12) and 20 ± 6% (number of cells = 3) reductions of the respective currents with respect to those of the controls, whereas the 2-benzyl analog produced 21 ± 4% (number of cells = 7) and 16 ± 5% (number of cells = 4) increases in the currents, respectively (Fig. 4, A and B). The 2-cyclic analogs appeared to be more effective in 6 mM glucose than 3 mM glucose, but the observed differences were not statistically significant. However, in 10 mM glucose, the 2-cyclic derivatives (10−9–10−7 M) were less effective (Fig. 4C).
Bell-shaped concentration-response curves were observed with the effective 2-cyclic-1,4-benzoxazine derivatives in the native pancreatic β cells (Fig. 4A). All of the effective 2-cyclic-1,4-benzoxazine derivatives showed a reduced efficacy at concentrations of >10−9 M. The 2-phenyl analog (10−11–10−6 M) was the most potent and effective compound in enhancing the KATP currents; its DE50 was 0.04 ± 0.01 × 10−9 M (Fig. 3A; Table 1). The 2-cyclohexylmethyl analog (10−10–10−7 M) was less effective, whereas the 2-phenylethyl analog (10−10–10−7 M) did not affect the KATP current (Fig. 4, A and C).
The percentage activations of the currents by 2.5 × 10−4 M diazoxide were 220 ± 12, 400 ± 20, and 150 ± 11% (number of cells = 6) with respect to the controls in 3 mM glucose and 5 mM NaN3, 6 mM glucose, and 10 mM glucose, respectively. A 2.5 × 10−4 M concentration of diazoxide in 3 mM glucose enhanced the whole-cell currents from 43 ± 5 pA in the control to 140 ± 7 pA and in 6 mM glucose from 27 ± 6 pA and 137 ± 9 pA in the control and after the application of the drug to the cells, respectively (Fig. 4, A and B). Significant channel activation also was observed in 10 mM glucose (Fig. 4C). Partial inhibitory responses of the Kir6.2ΔC36 channel were observed with the 2-cyclic analogs (Table 1).
The application of the 2-n-hexyl-1,4-benzoxazine derivative in 3 mM glucose and 5 mM NaN3 induced a dose-dependent cell depolarization that was recorded in the current-clamp mode (Fig. 5B). In 6 mM glucose, this compound also depolarized the cells, but the 2-isopropyl analog was inactive (Fig. 5A). The 2-phenyl analog hyperpolarized the cells at sub-nanomolar concentrations, but it did not cause a significant cell depolarization at higher concentrations (Fig. 5, A and B). The 2-benzyl analog showed approximately 20 mV cell hyperpolarization at sub-nanomolar concentrations (Fig. 5A). The 2-n-hexyl analog induced insulin release in vitro, whereas the 2-phenyl analog reduced this insulin release (Fig. 6).
In the excised patch experiments, 10−6 M of the 2-n-hexyl analog caused a channel block of 64 ± 5% (number of patches = 3) in the presence of ATP and 59 ± 4% (number of patches = 3) in the absence of the nucleotide. This blocking action was reversed after a 4-s washout. The mean currents were −30 ± 1 pA in the control, −13.2 ± 0.4 pA in the presence of the 2-n-hexyl analog (10−6 M) and 10−4 Μ ATP, −12.1 ± 1 pA in the presence of the 2-n-hexyl analog alone (10−6 M), −28 ± 0.9 pA after the washout of the drug solution, and −4.5 ± 0.3 pA with 10−4 Μ ATP (Fig. 7A). No effects were observed on the single-channel conductances, which were 68 ± 2 and 67 ± 3 pS (number of patches = 3) in the control and in the presence of a 10−6 M concentration of the drug, respectively. A 10−9 M concentration of the 2-phenyl analog led to a channel activation of 59 ± 4% (number of patches = 3) in the presence of 10−4 M internal ATP, but it produced a reduction of channel activity at a concentration of 10−7 M. The mean currents were −31.5 ± 1 pA in the control, −20.3 ± 0.4 pA in the presence of the 2-phenyl analog (10−9 M) and 10−4 Μ ATP, −7.1 ± 0.3 pA with the 2-phenyl analog (10−7 M) and 10−4 Μ ATP, −24.4 ± 3 pA after the washout of the drug solution, and −4.3 ± 0.4 pA with 10−4 Μ ATP (Fig. 7B).
In the presence of 2 × 10−4 M ADP, 10−9 M of the 2-phenyl analog led to a channel activation of 81 ± 6% (number of patches = 3) with respect to that of the control, whereas a mild reduction of channel activity was observed at a 10−7 M concentration (number of patches = 3) (Fig. 7C). The mean currents were −29.1 ± 1 pA in the control, −36.2 ± 2 pA with 2 × 10−4 M ADP, −51 ± 4 pA with the 2-phenyl analog (10−9 M) and 2 × 10−4 M ADP, −23 ± 1 pA with the 2-phenyl analog (10−7 M) and 2 × 10−4 M ADP, −25.1 ± 2 pA after the washout of the drug solution, and −3.5 ± 0.3 pA with 10−4 Μ ATP (Fig. 7C). The 2-isopropyl analog failed to affect the KATP channel in the excised patch experiments (data not shown).
In Vivo Effects of the 2H-1,4-Benzoxazine Derivatives on the Areas under the Curves and Fasting Glucose Levels in Normal Mice.
No macroscopic signs of intolerability were observed in the benzoxazine derivative-treated mice. The mice did not show any type of local reaction after the acute intraperitoneal administration of a single dose of these compounds. No adverse effects were observed at 24 h or 14 days postdose or thereafter.
The in vivo effects of the 2-isopropyl, 2-n-hexyl, and 2-phenyl analogs were evaluated by intraperitoneal glucose tolerance test and compared with the effects of glibenclamide or diazoxide. In the control mice that were injected with the vehicle, the administration of 2 g/kg glucose 30 min before the first blood glucose measurement resulted in a typical increase in blood glucose levels from a basal value of 0.86 ± 0.1 mg/ml to a peak of 2.72 ± 0.7 mg/ml as shown in Fig. 8A. The administration of the 2-n-hexyl analog led to a bimodal response in glucose disposal; at a 0.2 mg/kg dose, the blood glucose levels changed from a basal value of 0.6 ± 0.07 mg/ml to a peak of 3.31 ± 0.9 mg/ml. However, the group injected with 10 mg/kg of the 2-n-hexyl analog showed an improvement in the glucose tolerance curve; their blood glucose levels increased from a basal value of 0.66 ± 0.06 mg/ml to a peak of 1.64 ± 0.8 mg/ml. No significant differences were observed in the peak glucose levels of the mice that were treated with the 2-isopropyl analog with respect to those of the vehicle-treated mice. The administration of 10 mg/kg of a 2-phenyl analog caused an increase of the glucose level from a basal value of 0.67 ± 0.06 mg/ml to a peak of 3.3 ± 0.9 mg/ml. A dose of 25 mg/kg diazoxide caused an enhancement of the glucose levels from a basal value of 0.68 ± 0.08 mg/ml to a peak of 4.14 ± 0.4 mg/ml (Fig. 8A). The animals that were treated with 2 mg/kg glibenclamide showed the most significant improvement in the glucose tolerance curve, with a peak glucose level of 0.89 ± 0.07 mg/ml (Fig. 8A).
The administration of increasing doses (0.2, 2, and 10 mg/kg) of the 2-n-hexyl analog led to a bimodal response in the area under the curve (AUC) of blood glucose. In fact, at the lower dose of 0.2 mg/kg, the compound significantly increased the AUCs with respect to those of the vehicle; however, at a higher dose of 10 mg/kg, the 2-n-hexyl analog significantly decreased the AUC (Table 2). The 2-isopropyl analog increased the AUCs at the higher dose. The glibenclamide treatment (0.2 and 2 mg/kg) significantly decreased the AUCs with respect to those of the vehicle treated-group. The 2-phenyl analog (2–30 mg/kg) increased the AUC in a dose-dependent manner, and diazoxide (25 mg/kg) caused the maximal observed AUC increase (Table 2).
The 2-n-hexyl, 2-isopropyl, and 2-phenyl analogs did not significantly affect the fasting blood glucose levels of the mice compared with the 2 mg/kg dose of glibenclamide (Fig. 8B).
In the present work, we showed that the 2-n-hexyl-1,4-benzoxazine derivative is a reversible, potent blocker of the KATP channels of native pancreatic β cells, which leads to cell depolarization at nanomolar concentrations. Other Kir6.2 blockers that showed glucose-lowering effects exerted their action at micromolar concentrations, and they include the polyamine analogs; the vasodilating iptakalim; the synthetic imidazolines; the antipsychotics haloperidol and olanzapine; the antiarrhythmics cibenzoline, disopyramide, and mexiletine; the antimalarials mefloquine and quinine; and the antibiotics lomefloxacin and gatifloxacin (Bleck et al., 2005; Loussouarn et al., 2005; Zünkler, 2006; Misaki et al., 2007). The 2-n-hexyl analog equally blocked the KATP channels of the native β cells in the perforated and conventional whole-cell patch clamp methodologies and in the cells that expressed the truncated Kir6.2 subunit in the absence of SURs; these results suggest that the Kir6.2 subunit carries the inhibitory site(s) for this molecule.
The 2-n-hexyl analog showed antihyperglycemic activity in the intraperitoneal glucose tolerance test, but it was less effective than glibenclamide. A biphasic behavior was observed in the glucose disposal curves that were determined in vivo after the administration of increasing doses of the 2-n-hexyl analog; these data can be explained by accounting for the opposite in vitro effects of this compound on pancreatic β cells and skeletal muscle KATP channels (Fig. 9) (Tricarico et al., 2008).
The 2-phenyl analog activated the pancreatic KATP channels at sub-nanomolar concentrations in the perforated patch-clamp mode; the activation of these channels induces cell hyperpolarization. This activation was more potent but less effective than that of diazoxide. The rank order of potency and efficacy of the channel openers was 2-phenyl > 2-benzyl > 2-cyclohexylmethyl analogs, whereas the 2-phenylethyl analog was not effective. Other pancreatic KCOs that activate the channel at nanomolar concentrations include the thiadiazine-dioxide analogs, such as the NN414, BPDZ-73, 6-chloro-3-(1-methyl-1-phenylethyl)amino-4H-thieno[3,2-e]-1,2,4-thiadiazine-1,1-dioxide, NNC-55-0462, and 6-chloro-3-cyclobutylamino-4H-1,2,4-benzothiadiazine-1,1-dioxide (Nielsen et al., 2006; Carosati et al., 2007; Judge and Smith, 2009; Fischer et al., 2010; Pirotte et al., 2010).
The actions of the 2-cyclic aromatic analogs were observed mostly in the presence of medium and low glucose concentrations, although they were less effective in the presence of high external glucose concentrations, which suggests that they require a proper ATP/ADP ratio. This hypothesis is confirmed by the fact that in excised patch experiments the 2-phenyl analog is capable of activating the KATP channel in the presence of ATP and ADP. However, the 2-cyclic analogs showed a reduced efficacy in activating the KATP channel currents at a concentration of >10−9 M; this concentration can be related to the low-affinity interaction of these molecules with the Kir6.2 subunit. This is supported by the fact that these molecules inhibited the truncated Kir6.2 subunit in the nanomolar concentration range, but they showed a lower efficacy than the 2-linear alkyl chain analog.
As expected for a KATP channel opener, the 2-phenyl analog is the most effective 2H-1,4-benzoxazine derivative in increasing the AUC in normal mice, and this effect is similar to that of diazoxide; however, diazoxide caused the maximal observed increase of the AUC.
The introduction of a branched alkyl chain at position 2 of the 2H-1,4-benzoxazione nucleus leads to the formation of an inactive compound on the pancreatic KATP channel, as in the case of the 2-isopropyl analog; however, this analog is effective on the skeletal muscle subtype (Tricarico et al., 2008).
The 2-isopropyl analog induced a dose-dependent enhancement of the AUC of blood glucose; this result can be related to the activating action of the skeletal muscle KATP channels (Fig. 9).
Although glibenclamide affected the fasting blood glucose levels of the normal mice, the 2-n-hexyl-, 2-phenyl-, and 2-isopropyl-1,4-benzoxazine derivatives did not affect the fasting blood glucose levels of these mice. These compounds did not affect the fasting glucose levels with respect to those of glibenclamide because of their different modes of action on the target tissues. For example, the decrease of blood glucose due to the insulinotropic effect of the 2-n-hexyl analog is somehow counterbalanced by the KCO action on the skeletal muscle subtype. The actions of these drugs on neuronal KATP channels also should be considered. This specific channel controls the electrical activity of many types of neurons that are sensitive to the glucose level, thereby influencing glucose homeostasis. For example, a reduction in extracellular glucose causes the opening of the KATP channels in glucose-sensitive neurons of the ventromedial hypothalamus, which triggers adrenaline-dependent glucagon secretion and the counterregulatory response to hypoglycemia. The KATP channel activation, which is sufficient to abolish the electrical activity of the pro-opiomelanocortin-expressing neurons in the arcuate nucleus, leads to hyperphagia and increased body weight. Their partial activation, which reduces but does not abolish electrical activity, prevents glucose sensing and leads to impaired glucose tolerance (McTaggart et al., 2010). The differences in the kinetics of the actions of these compounds, which are rapidly reversible with the linear alkyl chain derivative compared with the sulfonylurea compound, also may play a role.
In conclusion, the 2-linear alkyl chain is the molecular determinant that confers the blocking action to the 2H-1,4-benzoxazines in pancreatic β cells, whereas the branched substituent is the molecular determinant that confers the activating action in skeletal muscles. The aromatic substituent confers the opening action to the 2H-1,4-benzoxazines in both the pancreatic β cells and the skeletal muscle cells. SUR1 shows an overall identity of 79% with SUR2A; larger differences exist in the amino acid sequences of the first nucleotide binding fold-2 (NBF2) and second transmembrane domain (TMD2) regions, and additional minor differences were found in the NBF2 region (Moreau et al., 2005). The NBF2 region confers sensitivity to the SUR2 openers with a benzopyran-like structure (cromakalim) and the pyridine (pinacidil) (Moreau et al., 2005). We propose that 2-isopropyl-1,4-benzoxazine, which is a structurally related analog of nucleotides, can bind to a site that is located on the NBF2 of SUR2A, although its interaction with TMD2 cannot be excluded. The 2-n-hexyl analog blocks the pancreatic and skeletal muscle KATP channels by interacting with the Kir6.2 subunit, but it also activates the muscle KATP channel by potentially interacting with the TMD2-NBF2 region. 2-Phenyl-1,4-benzoxazine activates both the pancreatic β cells and the skeletal muscle channels with a comparable potency and efficacy. These actions can be mediated by the binding of a drug to the TMD2-NBF2 region and to an additional site located on TMD1-NBF1, which is the main region that is responsible for the binding and activation of the pancreatic β cell KATP channel by diazoxide (Babenko et al., 2000; Moreau et al., 2005) (Fig. 9).
The 2-n-hexyl analog action that is mediated by the Kir6.2 subunit is an alternative mechanism of action with respect to that of the hypoglycemic sulfonylureas that target SUR1. This can be related to the structural similarity of this compound with the ATP molecule (Tricarico et al., 2008). The 2-phenyl analog may represent a potential drug treatment in patients who are intolerant to diazoxide. The 2H-1,4-benzoxazines have a differential pharmacological profile with respect to those of the SUR1 blockers or openers; in vitro and in vivo, they show a combined activating/blocking action toward KATP channels that confers significant glycemic control after the administration of a glucose bolus, but they do not affect fasting glycemia. In addition, their opening action of the skeletal muscle KATP channels may be cytoprotective, although pure SUR2A openers may lead to arrhythmia (Jahangir and Terzic, 2005). A similar mechanism has been proposed for iptakalim, which is a potent sub-nanomolar vascular KATP channel opener. At a high micromolar concentration, iptakalim shows a blocking action of the pancreatic channel, and it is in clinical development as an antihypertensive drug (Misaki et al., 2007; Pan et al., 2010).
Participated in research design: Tricarico and Rolland.
Conducted experiments: Rolland, Cannone, Mele, and Cippone.
Contributed new reagents or analytic tools: Laghezza, Carbonara, Fracchiolla, Tortorella, and Loiodice.
Performed data analysis: Cannone and Mele.
Wrote or contributed to the writing of the manuscript: Tricarico, Mele, and Conte Camerino.
Manuscript editing was performed with the contribution of the American Journal Experts.
This work was supported by the Italian Telethon [Grant GGP10101].
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- ATP-sensitive K+
- area under the curve
- drug concentration needed to enhance the current by 50%
- human embryonic kidney
- KATP channel openers
- inwardly rectifying K+ channel
- nucleotide binding fold-2
- second transmembrane domain
- sulfonylureas receptor type-1, BPDZ-62, N-(3,3-dimethylbutan-2-yl)-1,1-dioxo-4H-pyrido[4,3-e][1,2,4]thiadiazin-3-amine
- 6-chloro-3-(1-methylcyclobutyl)amino-4H-thieno[3,2-e]-1,2,4thiadiazine 1,1-dioxide.
- Received July 7, 2011.
- Accepted October 24, 2011.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics