Selective Enhancement of Nutrient-Induced Insulin Secretion by ATP-Sensitive Kϩ Channel-Blocking Imidazolines

0022-3565/09/3313-1033–1041$20.00 THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS Vol. 331, No. 3 Copyright © 2009 by The American Society for Pharmacology and Experimental Therapeutics 152751/3533565 JPET 331:1033–1041, 2009 Printed in U.S.A.

Kathrin Hatlapatka, Antje Wienbergen, Claudia Ku¨hne, Anne Jo¨rns, Michael Willenborg, and Ingo Rustenbeck Institute of Pharmacology and Toxicology, Technical University of Braunschweig, Braunschweig, Germany (K.H., A.W., C.K., M.W., I.R.); and Institute of Anatomy, Hannover Medical School, Hannover, Germany (A.J.) Received February 27, 2009; accepted September 9, 2009 Downloaded from

ABSTRACT ϩ ␮ ␮ The contribution of ATP-sensitive K channel (KATP channel)- M RX871024 or 100 M efaroxan) KATP channel activity was dependent and -independent signaling to the insulinotropic strongly reduced, the membrane was depolarized, and the characteristics of imidazolines was explored using perifused cytosolic Ca2ϩ concentration was elevated in the presence of mouse islets and ␤-cells. Up to a concentration of 100 ␮M basal glucose. Insulin secretion by sulfonylurea receptor

efaroxan had no insulinotropic effect in the presence of a basal (SUR)1 knockout (KO) islets, which have no functional KATP jpet.aspetjournals.org glucose concentration, but enhanced the effect of a stimulatory channels, was not increased by efaroxan (100 or 500 ␮M) or by concentration of glucose or nonglucidic nutrients (ketoisocap- 10 ␮M RX871024 but was increased by 100 ␮M RX871024. The roate plus glutamine). The secretion by a non-nutrient (40 mM imidazolines phentolamine and alinidine (100 ␮M) were also KCl) was not enhanced. At 500 ␮M, efaroxan stimulated insulin ineffective on SUR1 KO islets. It is concluded that a significant ␮ secretion when glucose was basal. Likewise, at 0.1 to 10 M KATP channel block is compatible with a purely enhancing RX871024 [2-(imidazolin-2-yl)-1-phenylindole] showed a purely effect of the imidazolines on nutrient-induced insulin secretion. enhancing effect, but at 100 ␮M it elicited a strong KCl-like Only RX871024 has an additional, nondepolarizing effect, at ASPET Journals on March 10, 2015 secretory response in the presence of basal glucose. At 0.1 and which at a high drug concentration is able to elicit a KATP 1 ␮M RX871024 did not significantly depolarize the ␤-cell mem- channel-independent secretion. brane. However, at a purely enhancing drug concentration (10

The insulinotropic effect of the prototypical imidazoline ing occurs with nanomolar affinity, binding to I3 sites re- phentolamine was originally believed to be due to an antag- quires micromolar concentrations of the imidazoline (Rust- ␣ ␤ onism at 2-adrenoceptors of the -cell (Robertson and Porte, enbeck et al., 1997). 1973; Efendic et al., 1975). The observation that the insuli- The demonstration that phentolamine and other imidazo- notropic effect of this compound was not shared by other lines block ATP-sensitive Kϩ channels (K channels) in ␣ ATP antagonists at 2-adrenoceptors but was shared by other pancreatic ␤-cells offered an explanation for their insulino- compounds with imidazoline moieties (Ostenson et al., 1988; tropic property (Plant and Henquin, 1990; Chan and Morgan, Schulz and Hasselblatt, 1989) led to the hypothesis that the 1990). However, it remained unclear how the effect on the insulinotropic effect was mediated by a ␤-cell-specific sub- KATP channel was related to the nonadrenergic imidazoline type of the imidazoline receptor (Chan et al., 1994). In other binding sites on ␤-cells (Rustenbeck et al., 1997). The concept tissues, at least two nonadrenergic imidazoline binding sites of a ␤-cell imidazoline receptor mediating multiple effects by have been identified, and consequently the hypothetical signal transduction cascades was rendered less likely by the ␤-cell subtype was named I receptor (Eglen et al., 1998; 3 observation that the K channel block by imidazolines was Morgan, 1999). In contrast to the I and I sites where bind- ATP 1 2 exerted directly at Kir6.2, the pore-forming subunit of the channel (Proks and Ashcroft, 1997; Grosse-Lackmann et al.,

This work was supported by grants from the Deutsche Forschungsgemein- 2003). Kir6.2 was then suggested to be the correlate of the I3 schaft [Ru 368/4-1, Ru 368/5-1]; and by the German Diabetes Society. receptor (Monks et al., 1999; Morgan, 1999). Article, publication date, and citation information can be found at http://jpet.aspetjournals.org. The demonstration that the imidazoline RX871024 in- doi:10.1124/jpet.109.152751. creased insulin secretion not only by blocking KATP channels

ϩ 2ϩ ABBREVIATIONS: KATP channel, ATP-sensitive K channel; RX871024, 2-(imidazolin-2-yl)-1-phenylindole; [Ca ]c, cytosolic free calcium concentration; BL11282, 5-chlor-3-(imidazolin-2-yl)-2-methylindole; NNC77-0074, (ϩ)-2-(2-(4,5-dihydro-1H-imidazol-2yl)-thiophene-2-yl-ethyl)- pyridine; SUR, sulfonylurea receptor; KO, knockout; KIC, ␣-ketoisocaproic acid; KU14R, 2-(2-ethyl-2,3-dihydrobenzo[b]furan-2-yl)-1H-imidazole. 1033 1034 Hatlapatka et al.

ϩ but also by acting at a site distal to Ca2 influx led to the mixing for 1 min. Islets and single islet cells were cultured in cell alternative hypothesis that imidazolines exert effects at mul- culture medium RPMI 1640 with 10% fetal calf serum (5 mM glu- tiple independent sites (Zaitsev et al., 1996). The experimen- cose) in a humidified atmosphere of 95% air and 5% CO2 at 37°C. Electrophysiological Recordings. K channel activity was tal model to demonstrate the KATP channel-independent ef- ATP fect was the permeabilized cell where the cytosolic free measured by a standard patch-clamp technique using the cell-at- 2ϩ 2ϩ tached configuration (Hamill et al., 1981). Pipettes were pulled from calcium concentration ([Ca ]c) was clamped by use of Ca - buffered incubation media. RX871024 (Zaitsev et al., 1996) borosilicate glass (2 mm o.d. and 1.4 mm i.d.; Hilgenberg, Malsfeld, and later efaroxan (Chan et al., 2001) were shown to increase Germany) by a two-stage vertical puller (List Electronics, Darm- stadt, Germany) and had resistances between 3 and 6 M⍀ when insulin release from such permeabilized cell preparations. It filled with solution. Currents were recorded by an EPC 7 patch- was hypothesized that these imidazolines sensitize the exo- 2ϩ clamp amplifier (List Electronics) and low pass-filtered by a four-pole cytotic machinery to [Ca ]c, which in turn would be in- Bessel filter at 2 kHz and stored on a videotape. The pipette holding creased by the KATP channel-blocking effect of these com- potential was 0 mV in cell-attached recordings. The membrane po- pounds. A similar combination of effects, comprising a tential of ␤-cells was determined using the whole-cell and perforated protein kinase C-mediated sensitization of the exocytotic ma- patch modes under current clamp condition (Smith et al., 1990). chinery in addition to a KATP channel block, has been sug- Exposure to the test compounds and washout was done by changing gested to explain the insulinotropic characteristics of sulfo- the bath medium with a slow bath perfusion system. The composi- nylureas (Eliasson et al., 1996). The hypothesis that the tions of the bath and pipette media were as given by Zu¨nkler et al. dependence on stimulatory glucose concentrations is due to (1988). All experiments were performed at room temperature (20– Downloaded from an effect at a site distal to Ca2ϩ influx seemed to be con- 23°C). Data were analyzed off-line using pClamp 6.03 software (Axon firmed by the development of imidazolines (e.g., BL11282 or Instruments, Foster City, CA). Microfluorimetric Measurements of the [Ca2؉] . Islets were NNC77-0074), which do not block KATP channels, but stim- c ulate insulin secretion nevertheless (Efanov et al., 2001b; cultured on collagen-coated glass cover slips in Petri dishes and were Hoy et al., 2003). These imidazolines were then termed “sec- used from days 2 to 4 after isolation. Fura-PE3/AM was loaded at a concentration of 2 ␮M (dissolved in Krebs-Ringer medium with 5 mM ond-generation imidazolines” to distinguish them from the jpet.aspetjournals.org glucose) for 45 min at 37°C. The coverslip with the attached cells or K channel-blocking imidazolines, which were termed ATP islets was inserted in a purpose-made perifusion chamber on the “first-generation imidazolines” (Efendic et al., 2002). stage of an epifluorescence microscope fitted with a Zeiss Fluar (40ϫ) However, a sensitization of the exocytotic machinery to 2ϩ objective. The fluorescence (excitation at 340 or 380 nm; emission [Ca ]c should also amplify subthreshold signals of non-nu- Ͼ470 nm) was recorded by a slow-scan charge-coupled device camera trient secretagogues and thus seems an unlikely explanation (TILL Photonics, Gra¨felfing, Germany). All perifusions were per- for the selective enhancement by efaroxan and RX871024 of formed at 35°C using a HEPES-buffered Krebs-Ringer-bicarbonate the effect of suprathreshold glucose concentrations (Zaitsev medium. Image pairs were taken at intervals as indicated in the at ASPET Journals on March 10, 2015 et al., 1996; Efendic et al., 2002; Bleck et al., 2004). In this figures; illumination time for each image was 800 ms. RX871024, but study, we present evidence that the selective enhancement not efaroxan, proved to be fluorescent with an excitation maximum 2ϩ does not consist of a sensitization of exocytosis to [Ca ]c at 331 nm and a broad emission peak with a maximum at 460 nm. increases but of an enhanced efficacy of amplifying signals This necessitated performing the experiments in duplicate using derived from the metabolism of suprathreshold concentra- sham-loaded islets to subtract the RX871024 fluorescence from the tions of nutrient secretagogues (Henquin, 2000). Despite this Fura fluorescence emission. common property, only RX871024 but not efaroxan, phentol- Electron Microscopy. Collagenase-isolated islets from SUR1 amine, or alinidine had a K channel-independent insuli- KO mice were pooled and cultured in batches of 50 islets. After ATP ␮ notropic effect, arguing against an indispensable role of the culturing for 20 h in RPMI 1640 medium with 100 M efaroxan or RPMI 1640 medium alone (5 mM glucose concentration), the islets distal site for the characteristics of imidazoline-induced in- were fixed for electron microscopy by immersing in a solution of sulin secretion. para-formaldehyde (2%) and diglutaraldehyde (2%) in cacodylate buffer (0.1 M, pH 7.3), postfixing in osmium tetroxide (1%) for 1 h, and embedding in epoxy resin. Fifty-nanometer sections of the islets Materials and Methods were cut by an ultramicrotome (Ultracut; Leica-Reichert-Jung, Wet- zlar, Germany), placed on nickel grids, and contrast-stained with Chemicals. Efaroxan was purchased from Tocris-Cookson (Bris- uranyl acetate and lead citrate. Transmission electron microscopy tol, UK), quinine from Sigma-Aldrich (Taufkirchen, Germany), and was performed using a Zeiss EM 9 electron microscope. Medium- tolbutamide from Serva (Heidelberg, Germany). Phentolamine was sized islets (diameters approximately 150 ␮m) were chosen for ul- kindly donated by Novartis/Ciba-Geigy (Lo¨rrach, Germany), alini- trastructural examination of the granulation state. The islets ana- dine by Boehringer Ingelheim GmbH (Ingelheim, Germany), and ␤ RX871024 by Reckitt and Colman (Kingston, UK). Fura-PE3/AM lyzed contained 50 to 90 -cells per section. Two independent was purchased from TEFLabs (Austin, TX), collagenase P from incubations were performed for each condition. Boehringer Mannheim/Roche Diagnostics (Mannheim, Germany), Measurement of Insulin Secretion. Batches of 50 freshly iso- cell culture medium RPMI 1640 from Gibco-Invitrogen (Karlsruhe, lated NMRI or SUR1 KO mouse islets were perifused (1 ml/min) at Germany), and fetal calf serum from PAA (Co¨lbe, Germany). All 37°C with a HEPES-buffered Krebs-Ringer medium containing the other reagents of analytical grade were from E. Merck (Darmstadt, respective secretagogue. The insulin content in the fractionated ef- Germany). fluate was determined by an enzyme-linked immunosorbent assay Tissue Culture. Islets were isolated from the pancreas of NMRI (Mercodia, Uppsala, Sweden). or sulfonylurea receptor (SUR)1 knockout (KO) mice (10–14 weeks Data Handling and Statistics. Statistical calculations were per- old) by a collagenase digestion technique and hand-picked under a formed with Prism and Instat software (GraphPad Software Inc., stereomicroscope. Single cells were obtained by incubation of the San Diego, CA). If not specified otherwise, differences were consid- islets for 10 min in a Ca2ϩ-free medium and subsequent vortex- ered significant if p Ͻ 0.05. Insulinotropic Effect of Imidazolines 1035 Results was only slightly affected by 10 ␮M efaroxan, whereas 30 ␮M efaroxan had a clear enhancing effect (Fig. 1A). After a first- Concentration-Dependent Changes in the Insulino- phase-like increase a plateau was established with a secre- tropic Characteristics of Imidazolines. The glucose de- tory rate of approximately 350% of the prestimulatory level. pendence of the insulinotropic effect was characterized in In the presence of 100 ␮M efaroxan the stimulatory effect of dependence of the drug concentration. The experimental pro- 10 mM glucose was strongly enhanced, resulting in a first- tocol was designed to assess the effect of the secretagogue in phase-like increase of secretion, which peaked after 6 min at the presence of a substimulatory glucose concentration and 710% of the prestimulatory rate. After 10 min a nadir was then to raise the glucose concentration to a moderately stim- reached; thereafter, the secretion rate increased again until ulatory value to assess the enhancing effect of the secreta- the end of the efaroxan perifusion (Fig. 1A). At 500 ␮M gogue. Efaroxan at 10, 30, or 100 ␮M did not stimulate efaroxan, in contrast, was markedly effective in the presence insulin secretion when the glucose concentration was 5 mM of 5 mM glucose, and within 8 min a plateau of 520% of the (Fig. 1A). When glucose was raised to 10 mM, the secretion prestimulatory level was established. Raising the glucose concentration to 10 mM led to a further increase to 930%, which remained stable until the end of the efaroxan exposure. Very low concentrations of efaroxan (0.1 and 1.0 ␮M) were entirely ineffective (Fig. 1A).

The insulinotropic effect of 10 ␮M RX871024 resembled Downloaded from that of 30 or 100 ␮M efaroxan in that no stimulation of secretion was seen in the presence of 5 mM glucose, but a marked enhancement of secretion with a prominent first phase occurred when the glucose concentration was raised to 10 mM (Fig. 1B). However, at 100 ␮M RX871024, a com-

pletely different pattern emerged. After 4 min of exposure in jpet.aspetjournals.org the presence of 5 mM glucose a massive increase of secretion started, which peaked after 10 min at 1080%. This increase was followed by a steady decline of secretion, which was only transiently reversed when glucose was raised to 10 mM (Fig. 1B). This response pattern resembled that of a strongly depolarizing concentration of KCl (40 mM). Again, a very high

peak value was quickly reached in the presence of 5 mM at ASPET Journals on March 10, 2015 glucose and thereafter the secretion rate declined steadily and was only transiently increased when glucose concentration was raised to 10 mM (Fig. 1B). In contrast with efaroxan, very low concentrations of RX871024 (0.1 and 1.0 ␮M) clearly enhanced the stimulatory effect of 10 mM glucose (Fig. 1B).

KATP Channel Block and Plasma Membrane Depo- larization by Efaroxan and RX871024. The concentration-dependent change in the insulinotropic characteristics was compared with the concentration-dependent block of

KATP channel activity. Efaroxan reduced the KATP channel activity in intact ␤-cells to 45.4 Ϯ 4.9% at 10 ␮M, 34.0 Ϯ 5.8% at 30 ␮M, 19.1 Ϯ 5.3% at 100 ␮M, and 8.2 Ϯ 2.4% at 300 ␮M. Under the same condition, RX871024 reduced channel activity to 17.5 Ϯ 6.7% at 10 ␮M and 4.3 Ϯ 1.9% at 100 ␮M. Thus, ␮ the KATP channel-blocking effect of 10 M RX871024 was ␮ Fig. 1. Concentration-dependent change in the glucose dependence of the equivalent to with that of 100 M efaroxan and the effect of insulinotropic effect of imidazolines. A, freshly isolated islets were peri- 100 ␮M RX871024 was not significantly different (p ϭ 0.23, fused with Krebs-Ringer medium containing 5 mM glucose for 90 min. t test) from that of 300 ␮M efaroxan. From 60 to 120 min the medium contained either 1 ␮M(E), 30 ␮M(F), ␮ 100 ␮M(Ⅺ), or 500 ␮M(f) efaroxan. From 90 to 130 min the glucose Consistent with these data, it was found that 10 M concentration was increased to 10 mM. The effect of 1 ␮M efaroxan was RX871024 was significantly more effective than 10 ␮M efar- not significantly different from that of glucose alone (control, ——) nor oxan in depolarizing the ␤-cell membrane. RX871024 depo- ␮ ␮ was that of 10 M (data not shown). Only at 500 M was efaroxan larized from a resting potential of Ϫ69.3 Ϯ 2.3 to Ϫ40.3 Ϯ 2.8 effective in the presence of 5 mM glucose. Values are means Ϯ S.E.M. of three to five experiments. B, freshly isolated islets were perifused with mV (n ϭ 7), whereas efaroxan depolarized from Ϫ72.4 Ϯ 2.6 Krebs-Ringer medium containing 5 mM glucose for 90 min. From 60 to mV to Ϫ54.6 Ϯ 5.1 mV (n ϭ 5). However, at 100 ␮M, the 120 min the medium contained 0.1 ␮M(E), 1.0 ␮M(F), 10 ␮M(Ⅺ), or 100 Ϫ Ϯ ␮ f membrane depolarization by RX871024 ( 26.3 2.5 mV, M( ) of RX871024. From 90 to 130 min the glucose concentration was ϭ increased to 10 mM. Note that the strong stimulatory effect of 100 ␮M n 5) was not significantly different from that by efaroxan RX871024 is different from the effect of 500 ␮M efaroxan in that the (Ϫ28.6 Ϯ 3.4 mV, n ϭ 7). At a very low concentration (0.1 ␮M) massive increase was followed by a decline that could be overcome only RX871024 was unable to significantly depolarize the ␤-cell transiently by raising the glucose concentration to 10 mM. This characteristic resembles the effect of 40 mM KCl (– – –) under the same condi- membrane. In conclusion, RX871024 is clearly more potent Ϯ tion. Values are means S.E.M. of four or five experiments. asaKATP channel blocker than efaroxan, but at concentra- 1036 Hatlapatka et al. tions greater than 100 ␮M, both compounds were practically equieffective. 2؉ Role of [Ca ]c Signaling in the Imidazoline En- hancement of Glucose-Induced Insulin Secretion. The 2ϩ relevance of [Ca ]c signaling for the glucose dependence of the insulinotropic effect was measured using the same experimental protocol as that for the secretion measurements. A preceding perifusion with a high Kϩ concentration (40 mM) for 10 min served as an internal standard. In the presence of 5 mM glucose 10 ␮M efaroxan had a negligible effect on 2ϩ [Ca ]c (Fig. 2A). As under control conditions, raising the glucose concentration to 10 mM evoked an immediate in- 2ϩ crease in [Ca ]c, which developed into an oscillatory pat- 2ϩ tern. Similar slow oscillations of [Ca ]c were induced by 100 ␮M efaroxan in the presence of 5 mM glucose and continued with a higher frequency (0.31 minϪ versus 0.16 minϪ1; n ϭ 6 each) when the glucose concentration was raised to 10 mM 2ϩ

(Fig. 2B). A transition from an oscillatory [Ca ]c increase to Downloaded from 2ϩ a sustained [Ca ]c increase occurred reproducibly by in- creasing the efaroxan concentration from 100 to 500 ␮Min the presence of 5 mM glucose (Fig. 2C). At 100 ␮M, RX871024 led to a massive sustained increase in the Fura fluorescence ratio (Fig. 3A). However, the ex- traordinary magnitude of this increase was largely due to the jpet.aspetjournals.org fluorescence of RX871024 itself, as could be shown by performing the same experiment with islets that had not been loaded with Fura (Fig. 3B). Calculating a net Fura fluores- 2ϩ cence ratio revealed that RX871024 increased the [Ca ]c to approximately the same values as the preceding Kϩ depolarization (Fig. 3C). Because of the autofluorescence of RX871024 and the necessary correction procedure, the occur- 2ϩ at ASPET Journals on March 10, 2015 rence of [Ca ]c oscillations could not be observed. Efaroxan Enhancement of Insulin Secretion In- duced by Nonglucidic Nutrients and Non-Nutrient Stimuli. To test whether the enhancement of secretion by efaroxan was specifically dependent on a stimulatory glucose concentration or more generally dependent on a stimulatory concentration of any fuel secretagogue, the glucose content in the perifusion medium was replaced by the combination of ketoisocaproic acid (KIC) and glutamine. At a concentration of 1 mM each, insulin secretion was basal, but raising the KIC concentration to 3 mM produced a moderate monophasic 2ϩ increase (Fig. 4A). At 100 ␮M efaroxan had no effect in the Fig. 2. Concentration-dependent effects of efaroxan on [Ca ]c of peri- presence of 1 mM KIC plus 1 mM glutamine, but the increase fused pancreatic islets. Fura-PE3/AM-loaded islets were perifused with Krebs-Ringer medium containing 5 mM glucose. After a depolarization of secretion elicited by 3 mM KIC was markedly enhanced in with 40 mM Kϩ for 10 min the islets were perifused with 10 ␮M (A) or 100 the presence of 100 ␮M efaroxan (Fig. 4A). ␮M (B) efaroxan. In the continued presence of the drug, the glucose At a nonstimulatory concentration of glutamine plus KIC concentration was raised from 5 to 10 mM. This perifusion protocol 2ϩ corresponds to that of the secretion measurements shown in Fig. 1. The ␮ ϩ (1 mM each) repeated [Ca ]c waves were elicited by 100 M ␮ 2 response pattern obtained with 10 M efaroxan (no [Ca ]c increase in 2ϩ efaroxan, and this pattern continued when KIC was raised to the presence of 5 mM glucose and an oscillatory [Ca ]c increase after 3 mM (Fig. 4B). When mean values were calculated, the raising glucose to 10 mM) corresponded to that of a control experiment. C, 2ϩ raising the efaroxan concentration from 100 to 500 ␮M in the continuous transient increase of [Ca ]c elicited by 3 mM KIC was presence of 5 mM glucose led to a transition from an oscillatory to a clearly discernible above the plateau resulting from the 2ϩ sustained [Ca ]c increase. Shown are traces from six subregions of a 2ϩ phase-shifted [Ca ]c waves elicited by efaroxan (Fig. 4C). single islet to demonstrate the synchronous oscillatory character of the [Ca2ϩ] increase. Representative recordings of four to five experiments Washout of efaroxan led to a return to basal levels (Fig. 4C). c each. To measure the effect of efaroxan on the insulin secretion elicited by a non-nutrient stimulus the glucose concentration was kept at 5 mM throughout. Exposure to 40 mM KCl for 10 initial response (Fig. 5A). When the second exposure to high min gave a marked secretory response, which returned to Kϩ was performed in the presence of 100 ␮M efaroxan, the baseline values within 15 min after the Kϩ concentration was percentage was not significantly increased (85% of the initial reduced to the physiological level. Repeating the 10-min ex- secretory response to KCl). In these experiments, the expo- posure to high Kϩ after an interval of 40 min gave a secretory sure to efaroxan preceded the second Kϩ depolarization for response (area under the curve) that amounted to 64% of the 20 min; during this time no increase in secretion was ob- Insulinotropic Effect of Imidazolines 1037 Downloaded from jpet.aspetjournals.org at ASPET Journals on March 10, 2015

Fig. 4. Nutrient dependence of the insulinotropic effect of efaroxan. A, freshly isolated islets were perifused with Krebs-Ringer medium containing nonglucidic nutrients (1 mM glutamine and 1 mM KIC) for 110 min. From 80 to 140 min the medium contained 100 ␮M efaroxan (F). From 110 to 140 min the KIC concentration was increased to 3 mM. In the 2ϩ Fig. 3. Effect of RX871024 on [Ca ]c of perifused pancreatic islets. control experiment (E) efaroxan was absent. Efaroxan had a clear en- Fura-PE3/AM-loaded islets were perifused with Krebs-Ringer medium ϩ hancing effect when the KIC concentration was raised. B, effect of efar- containing 5 mM glucose. Top, after a depolarization with 40 mM K for oxan on [Ca2ϩ] of a Fura 2-loaded islet using the same perifusion protocol ␮ c 10 min the islets were perifused with 100 M RX871024. In the continued as for the secretion measurements shown in A. Shown are traces from six presence of the drug, the glucose concentration was raised from 5 to 10 subregions of a single islet to demonstrate the synchronous oscillatory mM. This perifusion protocol corresponds to that of the secretion mea- 2ϩ Ϯ character of the [Ca ]c increase. C, mean values S.E.M. of five exper- surements shown in Fig. 1. Note the massive increase in the Fura fluo- 2ϩ iments as shown in B. Note the increase in [Ca ]c in the presence of rescence ratio by addition of RX871024. Middle, the same experiment as basal concentrations of the nutrients and the reversibility after efaroxan in A performed without Fura loading of the islet. Note that there is no washout. fluorescence ratio increase during Kϩ depolarization, but there is an increase after addition of 100 ␮M RX871024, suggestive of the autofluo- 40-min interval (data not shown). When 100 ␮M efaroxan rescence of this compound. Bottom, net Fura fluorescence ratio by sub- was added to the perifusion medium before the second Kϩ tracting the recordings obtained without Fura loading from the record- ϩ 2ϩ 2 ings of the regular experiments. Note that the amplitude of the [Ca ]c depolarization, an oscillatory increase of [Ca ]c resulted. increase is now in the range of that of the preceding Kϩ depolarization. The amplitude of the mean value corresponded to the plateau Representative traces of four to five experiments each. 2ϩ ϩ value of the [Ca ]c increase by K depolarization (Fig. 5B). The second Kϩ depolarization, which was performed in the ϩ 2ϩ served (Fig. 5A). The 10-min K depolarization led to a presence of efaroxan, resulted in a transient [Ca ]c peak 2ϩ 2ϩ square wave-like [Ca ]c increase that was essentially un- discernible above the mean [Ca ]c value established by efar- changed when the Kϩ depolarization was repeated after a oxan (Fig. 5B). 1038 Hatlapatka et al. Downloaded from jpet.aspetjournals.org at ASPET Journals on March 10, 2015

Fig. 5. Inability of efaroxan to enhance non-nutrient-induced insulin secretion. A, freshly isolated islets were perifused with Krebs-Ringer Fig. 6. Comparison of the effect of efaroxan and RX871024 on insulin medium containing 5 mM glucose throughout. From 60 to 70 min and secretion by SUR1 KO islets. Top, freshly isolated SUR1 KO islets were again from 110 to 120 min the Kϩ concentration was raised to 40 mM. A perifused with Krebs-Ringer medium containing 5 mM glucose for 90 100 ␮M concentration of efaroxan (F) was present from 90 to 130 min; i.e., min. From 60 to 120 min the medium contained 100 ␮M(F)or500␮M(E) the second Kϩ depolarization took place in the presence of efaroxan. In efaroxan. From 90 to 130 min the glucose concentration was increased to the control experiment (E) efaroxan was absent throughout. Values are 10 mM. Values are means Ϯ S.E.M. of four experiments. Bottom, freshly means Ϯ S.E.M. of four experiments each. B, effect of efaroxan and KCl isolated SUR1 KO islets were perifused with Krebs-Ringer medium con- 2ϩ taining 5 mM glucose for 90 min. From 60 to 120 min the medium on [Ca ]c of perifused islets. A Fura-PE3/AM-loaded islet was perifused with Krebs-Ringer medium containing 5 mM glucose throughout. KCl contained 100 ␮M(f), 10 ␮M(Ⅺ), 0.1 ␮M(F), or no RX871024 (control, was raised to 40 mM first in the absence and then in the presence of E), From 90 to 130 min the glucose concentration was increased to 10 2ϩ mM. The effect size (picograms per minute ϫ islet) of glucose stimulation efaroxan. Note the efaroxan-induced increase of [Ca ]c in the presence of 5 mM glucose. Values are means Ϯ S.E.M. of five experiments. alone can be estimated from Fig. 7C. Values are means Ϯ S.E.M. of four experiments.

Contribution of KATP Channel-Independent Signal- ing of the Imidazolines to the Glucose Dependence. To the response patterns can be regarded as typical for the class assess the role of KATP channel-independent effects in the of KATP channel-blocking imidazolines the effect of phentol- secretory characteristics of efaroxan and RX871024, isolated amine and alinidine (100 ␮M each) on insulin secretion by islets from SUR1 KO mice were perifused using the same SUR1 KO islets was characterized (Fig. 7). The secretory protocol as that for the initial secretion measurements (Fig. response of SUR1 KO islets to both compounds was negligible 1). Neither at 5 mM glucose nor at 10 mM glucose did 100 ␮M compared with that of NMRI islets (Fig. 7, A and B). Quinine, efaroxan increase secretion by SUR1 KO islets. Even at 500 which, like imidazolines, blocks KATP channels at the pore- ␮M efaroxan was completely ineffective (Fig. 6A). Likewise, forming subunit, had a moderate insulinotropic effect on SUR1 RX871024 at 0.1 and 10 ␮M was unable to significantly KO islets (Fig. 7C), but, similar to RX871024, this effect was increase insulin secretion when the glucose concentration already visible in the presence of 5 mM glucose. The inefficiency was raised to 10 mM. In contrast, a clear increase in secretion of efaroxan to stimulate insulin secretion by SUR1 KO islets (2.5 fold) was produced by 100 ␮M RX871024. However, this was confirmed by ultrastructural examination of the granula- increase had already occurred when the islets were perifused tion state. A 20-h exposure to 100 ␮M efaroxan, which strongly with 5 mM glucose. When the glucose concentration was degranulated ␤-cells in NMRI mouse islets, did not affect the raised to 10 mM, a marked transient decrease down to the content of insulin granules compared with that of control cul- prestimulatory level occurred (Fig. 6B). To decide which of tured SUR1 KO islets (Fig. 8). Insulinotropic Effect of Imidazolines 1039

Fig. 8. Effect of a long-term exposure to efaroxan on the ultrastructure of pancreatic ␤-cells in SUR1 KO islets. Isolated pancreatic islets were cultured for 20 h in RPMI 1640 medium with 5 mM glucose in the presence (A) or absence (B) of 100 ␮M efaroxan. The ␤-cells in control cultured SUR1 KO islets had a content of insulin granules comparable with that of normal NMRI mouse ␤-cells. There was no degranulation by

the exposure to efaroxan. Original magnification, 4000ϫ. Downloaded from

secretion at stimulatory but not at subthreshold glucose concentrations and that this selectivity is lost at a higher drug concentration (Zaitsev et al., 1996; Efendic et al., 2002). However, our other observations led us to a reinterpretation of the underlying mechanism of action. It has been proposed jpet.aspetjournals.org that the loss of selectivity at high concentrations of Ͼ ␮ RX871024 ( 50 M) was due to the KATP channel-blocking effect of this imidazoline (Efendic et al., 2002), because it is

widely accepted that a block of KATP channels is sufficient to stimulate insulin secretion. Our recent observation that the imidazoline efaroxan did not stimulate insulin secretion of normal mouse islets at a basal glucose concentration, even at ASPET Journals on March 10, 2015 though it reduced KATP channel activity, depolarized the plasma membrane, and induced oscillatory increases in 2ϩ [Ca ]c under the same condition (Bleck et al., 2004), was not compatible with this explanation and prompted us to reas-

sess the mechanism of action of KATP channel-blocking imidazolines. The secretion measurements in this study showed that similar to the insulinotropic effect of RX871024, that of efaroxan lost its selectivity for stimulatory glucose concentrations at a high drug concentration. The potency of the insu-

linotropic effect was paralleled by the potency of the KATP channel block; in both respects RX871024 was approximately 5- to 10-fold more potent than efaroxan. Both 10 ␮M Fig. 7. Lack of insulinotropic effect of imidazolines on islets from SUR1 KO mice. A, freshly isolated SUR1 KO islets (F) and NMRI islets (E) were RX871024 and 100 ␮M efaroxan were similarly effective in perifused with Krebs-Ringer medium containing 5 mM glucose for 90 inhibiting KATP channel activity (by approximately 80% in min. From 60 to 120 min the medium contained 100 ␮M phentolamine. intact ␤-cells) and produced an insulinotropic effect of similar From 90 to 130 min the glucose concentration was increased to 10 mM. Values are means Ϯ S.E.M. of four experiments. B, freshly isolated SUR1 strength, which was selective for stimulatory glucose. With KO islets (F) and NMRI islets (E) were perifused with Krebs-Ringer both compounds a reduction of KATP channel activity by medium containing 5 mM glucose for 90 min. From 60 to 120 min the approximately 95% was paralleled by a stimulatory effect at medium contained 100 ␮M alinidine. From 90 to 130 min the glucose concentration was increased to 10 mM. Values are means Ϯ S.E.M. of a basal glucose concentration. four experiments. C, freshly isolated SUR1 KO islets were perifused with However, there was a remarkable qualitative difference Krebs-Ringer medium containing 5 mM glucose for 90 min. From 60 to between the stimulatory effects of high concentrations of 120 min the medium contained 100 ␮M quinine (F) or no further addition E efaroxan and of RX871024. Although the secretion plateau (control, ; same as in Fig. 6B). From 90 to 130 min the glucose concen- ␮ tration was increased to 10 mM. In the presence of quinine the secretion induced by 500 M efaroxan in the presence of 5 mM glucose (area under the curve from 60 to 120 min) was significantly increased. could be further increased by raising glucose to 10 mM, 100 Values are means Ϯ S.E.M. of four to five experiments. ␮M RX871024 produced a very high secretion peak in the presence of 5 mM glucose, which spontaneously turned into a Discussion steady decline that could be reversed only transiently by raising glucose to 10 mM. This pattern resembled that pro- In the present investigation, we confirmed the earlier ob- duced by a strongly depolarizing KCl concentration and is servation that the imidazoline RX871024 enhances insulin reminiscent of the earlier observations that the insulino- 1040 Hatlapatka et al.

tropic action of non-nutrient secretagogues leads to de- ric Fura measurements. This problem could be overcome by creased responsiveness upon renewed stimulation, whereas calculating a net Fura ratio from the mean values. Some the opposite, a potentiation, results from nutrient stimula- earlier conclusions concerning the effects of RX871024 on 2ϩ tion (Nesher and Cerasi, 1987, 2002). [Ca ]c may need reevaluation in light of this observation. As a nonglucidic nutrient we used KIC in combination with Nevertheless, effects on intracellular Ca2ϩ handling by glutamine. KIC requires a transamination partner to contin- RX871024 could contribute to the different insulinotropic

uously produce both triggering and amplifying signals in the characteristics compared with those of the other KATP chan- Krebs cycle (Gao et al., 2003; Panten and Rustenbeck, 2008). nel-blocking imidazolines (Efanov et al., 1998; Efanova et al., This way, high concentrations of KIC, which have a direct 1998). ␤ blocking effect on KATP channels of the -cells, could be The role of KATP channel-independent signaling was tested avoided (Heissig et al., 2005). As with glucose, efaroxan by using islets of SUR1 KO mice, which have no functional 2ϩ raised [Ca ]c but not insulin secretion when the KIC con- KATP channels in the endocrine pancreas (Seghers et al., centration was nonstimulatory. Raising the nutrient concen- 2000). In an earlier investigation, 50 ␮M RX871024 in- tration to a stimulatory level revealed the enhancing effect of creased insulin secretion from such islets 3- to 4-fold in the efaroxan on secretion. Apparently, efaroxan enhances not presence of either 3 or 17 mM glucose (Efanov et al., 2001a). only glucose-induced secretion, but also, more generally, nu- The increase in secretion by SUR1 KO mice in the present trient-induced insulin secretion, whereas it did not enhance investigation was of a comparable magnitude, but it occurred ␮ ␮ the secretion elicited by a non-nutrient (40 mM KCl). Thus, only at the high (100 M) but not at the low (10 M) or very Downloaded from the earlier hypothesis that the characteristics of the imida- low (0.1 ␮M) concentration, which selectively enhanced glu- zoline effect are due to a sensitization of the exocytotic ma- cose-stimulated insulin secretion. In marked contrast to 2ϩ chinery to [Ca ]c (Zaitsev et al., 1996) may need revision. RX871024, efaroxan had practically no insulinotropic effect Concurrent with their KATP channel-blocking activity, both on SUR1 KO islets. This result was seen with both the efaroxan and RX871024 produced substantial increases of nutrient-selective concentration (100 ␮M) and the nonselec- [Ca2ϩ] in the presence of 5 mM glucose, which raises the tive concentration (500 ␮M). The lack of effect was confirmed

c jpet.aspetjournals.org 2ϩ ␮ ␤ question as to why this [Ca ]c increase was not transduced by the inability of 100 M efaroxan to affect the -cell gran- into increased secretion. The characteristic feature of the ule content of SUR1 KO islets during a 20-h tissue culture, 2ϩ ␤ [Ca ]c increase by efaroxan is its oscillatory nature. It has a whereas -cells in NMRI islets were markedly degranulated Ϫ1 2ϩ frequency (0.3 min ) like the slow [Ca ]c oscillation typi- by this procedure (Rustenbeck et al., 2004). cally induced by moderate stimulatory glucose concentra- Phentolamine and alinidine, two other KATP channel- 2ϩ tions (Liu et al., 1998). Most likely, the [Ca ]c oscillation is blocking imidazolines that were used in early studies on this the direct consequence of the oscillatory ␤-cell membrane group of compounds (Schulz and Hasselblatt, 1989; Proks at ASPET Journals on March 10, 2015 depolarization elicited by efaroxan in the presence of 5 mM and Ashcroft, 1997; Rustenbeck et al., 1997), were also un- glucose (Bleck et al., 2005). The repolarization despite the able to stimulate insulin secretion from SUR1 KO mice. On continuous presence of efaroxan may result from the marked the other hand, quinine, which, similar to imidazolines,

susceptibility of the KATP channel block by efaroxan to the blocks KATP channels at the pore-forming subunit and stim- opening effect of nucleoside diphosphates (Wienbergen et al., ulates insulin secretion (Grosse-Lackmann et al., 2003) but 2ϩ 2007). The [Ca ]c oscillation caused by efaroxan may thus does not have an imidazoline moiety, had a moderate insuli- reflect an underlying oscillation of the ATP/ADP ratio at notropic effect on SUR1 KO islets in the presence of both 5 basal glucose levels (Dryselius et al., 1994). and 10 mM glucose, similar to RX871024. Taken together,

The observation that raising the efaroxan concentration the present results suggest that the ability to elicit KATP from a nutrient-selective (100 ␮M) to a nonselective level channel-independent insulin secretion is neither a general ␮ 2ϩ (500 M) transformed the [Ca ]c oscillation into a sustained property of imidazoline compounds nor is it essential for the 2ϩ increase suggested a critical role of the oscillatory [Ca ]c marked glucose dependence of their insulinotropic effect on 2ϩ pattern. On the other hand, a short-term [Ca ]c increase by normal mouse islets. 40 mM KCl elicited a strong secretory response. Our earlier In view of this heterogeneity, is the imidazoline moiety a assumption that imidazolines may exert an inhibitory action pharmacophore with regard to the insulinotropic effect? The distal to Ca2ϩ influx, which is overcome by increased energy existence of the second-generation imidazolines (Efanov et metabolism (Bleck et al., 2004, 2005) cannot explain this al., 2001b; Hoy et al., 2003) and the inability of the imidazole discrepancy, because the insulinotropic effect of the short analog of efaroxan, KU14R, to enhance insulin secretion

KCl pulse persisted practically unchanged in the presence of despite a KATP channel-blocking and depolarizing effect efaroxan. The difference may rather be due to the fact that 40 (Bleck et al., 2005) suggested that this is the case. In the

mM KCl has a depolarizing strength approximately double latter report it was hypothesized that in addition to the KATP that of a KATP channel closure (approximately 40 versus 20 channel two imidazoline-specific sites of action would exist: 2ϩ mV). It was recently noted that a prolonged 20 mV depolar- one site that mediates the larger [Ca ]c increase of efaroxan ization by 15 mM KCl had only a modest transient effect on compared with that of KU14R and another site that mediates secretion (Hatlapatka et al., 2009). Thus, the oscillatory pat- the enhanced response to nutrients. The present observa- 2ϩ tern may keep the [Ca ]c signal at a subthreshold level for tions are best explained by assuming that this may be one secretion when the nutrient concentration is insufficient to and the same effect: the imidazoline-induced increase in 2ϩ activate metabolic amplifying pathways. [Ca ]c is practically insufficient to stimulate secretion on its 2ϩ RX871024-induced [Ca ]c oscillations could not be ob- own, but enhances the effect of nutrient stimuli, which acti- served because of the fluorescence of aqueous solutions of vate the metabolic amplifying pathway(s) (Henquin, 2000; this compound, which was not cancelled out by the ratiomet- Panten and Rustenbeck, 2008). Such a model would also Insulinotropic Effect of Imidazolines 1041 explain why the insulin secretion by 40 mM KCl, which has Eliasson L, Renstro¨m E, Amma¨la¨ C, Berggren PO, Bertorello AM, Bokvist K, Chiba- lin A, Deeney JT, Flatt PR, Ga¨bel J, et al. (1996) PKC-dependent stimulation of no metabolic amplifying effect, was not enhanced. exocytosis by sulfonylureas in pancreatic ␤ cells. Science 271:813–815. So is the selective enhancing effect of KATP channel-block- Gao Z, Young RA, Li G, Najafi H, Buettger C, Sukumvanich SS, Wong RK, Wolf BA, and Matschinsky FM (2003) Distinguishing features of leucine and ␣a- ing imidazolines on nutrient-induced insulin secretion only ketoisocaproate sensing in pancreatic ␤-cells. Endocrinology 144:1949–1957. due to their limited depolarizing effect? This finding seems to Grosse-Lackmann T, Zunkler BJ, and Rustenbeck I (2003) Specificity of nonadrenergic imidazoline binding sites in insulin-secreting cells and relation to the block be true for efaroxan, phentolamine, and alinidine but not for ϩ of ATP-sensitive K channels. Ann N Y Acad Sci 1009:371–377. RX871024. At 0.1 ␮M, RX871024 enhanced glucose-induced Hamill OP, Marty A, Neher E, Sakmann B, and Sigworth FJ (1981) Improved insulin secretion but did not yet significantly depolarize patch-clamp techniques for high-resolution current recordings from cells and cell- free membrane patches. Pflugers Arch 391:85–100. ␤-cells. It is apparent that RX871024 has two mechanisms of Hatlapatka K, Willenborg M, and Rustenbeck I (2009) Plasma membrane depolar- action with different concentration dependencies (one depo- ization as a determinant of the first phase of insulin secretion. Am J Physiol Endocrinol Metab 297:E315–E322. larizing and one nondepolarizing), both of which are compat- Heissig H, Urban KA, Hastedt K, Zu¨nkler BJ, and Panten U (2005) Mechanism of ible with a purely enhancing characteristic at the lower con- the insulin-releasing action of ␣-ketoisocaproate and related ␣-keto acid anions. Mol Pharmacol 68:1097–1105. centration range. The accumulation of phospholipase A2 Henquin JC (2000) Triggering and amplifying pathways of regulation of insulin reaction products such as arachidonic acid may be compatible secretion by glucose. Diabetes 49:1751–1760. Hoy M, Olsen HL, Andersen HS, Bokvist K, Buschard K, Hansen J, Jacobsen P, with such a nondepolarizing effect (Jones et al., 2004; Sha- Petersen JS, Rorsman P, and Gromada J (2003) Imidazoline NNC77-0074 stimu- royko et al., 2007). At a high concentration of RX871024, the lates insulin secretion and inhibits glucagon release by control of Ca2ϩ-dependent combination of both effects may be responsible for the mas- exocytosis in pancreatic ␣- and ␤-cells. Eur J Pharmacol 446:213–221. Jones PM, Burns CJ, Belin VD, Roderigo-Milne HM, and Persaud SJ (2004) The role sive stimulation of secretion, which exceeded the maximal of cytosolic phospholipase A2 in insulin secretion. Diabetes 53:S172–S178. Downloaded from effect of efaroxan. The nutrient-selective insulinotropic effect Liu YJ, Tengholm A, Grapengiesser E, Hellman B, and Gylfe E (1998) Origin of slow and fast oscillations of Ca2ϩ in mouse pancreatic islets. J Physiol 508:471–481. of the second-generation imidazolines, the molecular mecha- Monks LK, Cosgrove KE, Dunne MJ, Ramsden CA, Morgan NG, and Chan SL (1999) nism of which has eluded clarification up to now, would then Affinity isolation of imidazoline binding proteins from rat brain using 5-amino- efaroxan as a ligand. FEBS Lett 447:61–64. be phenomenologically similar, but mechanistically unre- Morgan NG (1999) Imidazoline receptors: targets for novel antihyperglycemic lated to that of the first-generation imidazolines. agents. Exp Opin Invest Drugs 8:575–584. Nesher R and Cerasi E (1987) Biphasic insulin release as the expression of combined

inhibitory and potentiating effects of glucose. Endocrinology 121:1017–1024. jpet.aspetjournals.org Acknowledgments Nesher R and Cerasi E (2002) Modeling phasic insulin secretion: immediate and time-dependent effects of glucose. Diabetes 51:S53–S59. Joseph Bryan and Lydia Aguilar-Bryan kindly provided the SUR1 ␣ Ostenson CG, Pigon J, Doxey JC, and Efendic S (1988) 2-Adrenoceptor blockade KO mice. Skillful technical assistance by Petra Weber, Verena Lier- does not enhance glucose-induced insulin release in normal subjects or patients Glaubitz, and Sabine Warmbold is gratefully acknowledged. with noninsulin-dependent diabetes. J Clin Endocrinol Metab 67:1054–1059. Panten U and Rustenbeck I (2008) Fuel-induced amplification of insulin secretion in References mouse pancreatic islets exposed to a high sulfonylurea concentration: role of the NADPH/NADPϩ ratio. Diabetologia 51:101–109. Bleck C, Wienbergen A, and Rustenbeck I (2004) Glucose-dependence of imidazoline- Plant TD and Henquin JC (1990): Phentolamine and yohimbine inhibit ATP- induced insulin secretion: different characteristics of two ATP-sensitive Kϩ chan- ϩ ␤

sensitive K -channels in mouse pancreatic -cells. Br J Pharmacol 101:115–120. at ASPET Journals on March 10, 2015 nel-blocking compounds. Diabetes 53 (Suppl 3):S135–S139. Proks P and Ashcroft FM (1997) Phentolamine block of KATP channels is mediated by Bleck C, Wienbergen A, and Rustenbeck I (2005) Essential role of the imidazoline Kir6.2. Proc Natl Acad Sci U S A 94:11716–11720. moiety for the insulinotropic effect but not the K channel-blocking effect of ATP Robertson RP and Porte D Jr (1973) Adrenergic modulation of basal insulin secretion imidazolines; a comparison of the effects of efaroxan and its imidazole analogue, in man. Diabetes 22:1–8. KU14R. Diabetologia 48:2567–2575. Rustenbeck I, Herrmann C, Ratzka P, and Hasselblatt A (1997) Imidazoline/ Chan SL, Brown CA, Scarpello KE, and Morgan NG (1994) The imidazoline site guanidinium binding sites and their relation to inhibition of K channels in involved in insulin secretion: characteristics that distinguish it from I and I sites. ATP 1 2 pancreatic B-cells. Naunyn Schmiedebergs Arch Pharmacol 356:410–417. Br J Pharmacol 112:1065–1070. Rustenbeck I, Wienbergen A, Bleck C, and Jo¨rns A (2004) Desensitization of insulin Chan SL and Morgan NG (1990) Stimulation of insulin secretion by efaroxan may secretion by depolarizing insulin secretagogues. Diabetes 53:S140–S150. involve interaction with potassium channels. Eur J Pharmacol 176:97–101. Schulz A and Hasselblatt A (1989) An insulin-releasing property of imidazoline Chan SL, Mourtada M, and Morgan NG (2001) Characterization of a K channel- ATP derivatives is not limited to compounds that block ␣-receptors. Naunyn Schmiede- independent pathway involved in potentiation of insulin secretion by efaroxan. bergs Arch Pharmacol 340:312–317. Diabetes 50:340–347. Seghers V, Nakazaki M, DeMayo F, Aguilar-Bryan L, and Bryan J (2000) SUR1 Dryselius S, Lund PE, Gylfe E, and Hellman B (1994) Variations in ATP-sensitive ϩ knockout mice. A model for K(ATP) channel-independent regulation of insulin K channel activity provide evidence for inherent metabolic oscillations in pancreatic ␤-cells. Biochem Biophys Res Commun 205:880–885. secretion. J Biol Chen 275:9270–9277. Efanov AM, Høy M, Bra¨nstro¨m R, Zaitsev SV, Magnuson MA, Efendic S, Gromada Sharoyko VV, Zaitseva II, Leibiger B, Efendic´S, Berggren PO, and Zaitsev SV (2007) J, and Berggren PO (2001a) The imidazoline RX871024 stimulates insulin secre- Arachidonic acid signalling is involved in the mechanism of imidazoline-induced ␤ KATP channel-independent stimulation of insulin secretion. Cell Mol Life Sci tion in pancreatic -cells from mice deficient in KATP channel function. Biochem Biophys Res Commun 284:918–922. 64:2985–2993. Smith PA, Ashcroft FM, and Rorsman P (1990) Simultaneous recordings of glucose- Efanov AM, Zaitsev SV, Efanova IB, Zhu S, Ostenson CG, Berggren PO, and Efendic´ ϩ S (1998) Signaling and sites of interaction for RX-871024 and sulfonylurea in the dependent electrical activity and ATP-regulated K -currents in isolated mouse ␤ stimulation of insulin release. Am J Physiol 274:E751–E757. pancreatic -cells. FEBS Lett 261:187–190. Efanov AM, Zaitsev SV, Mest HJ, Raap A, Appelskog IB, Larsson O, Berggren PO, Wienbergen A, Bleck C, Lackmann TG, and Rustenbeck I (2007) Antagonism of the and Efendic S (2001b) The novel imidazoline compound BL11282 potentiates insulinotropic action of first-generation imidazolines by openers of KATP channels. glucose-induced insulin secretion in pancreatic ␤-cells in the absence of modula- Biochem Pharmacol 73:94–102. Zaitsev SV, Efanov AM, Efanova IB, Larsson O, Ostenson CG, Gold G, Berggren PO, tion of KATP channel activity. Diabetes 50:797–802. Efanova IB, Zaitsev SV, Brown G, Berggren PO, and Efendic´ S (1998) RX871024 and Efendic´S (1996) Imidazoline compounds stimulate insulin release by inhibi- ϩ induces Ca2 mobilization from thapsigargin-sensitive stores in mouse pancreatic tion of KATP channels and interaction with the exocytotic machinery. Diabetes ␤-cells. Diabetes 47:211–218. 45:1610–1618. Zu¨nkler BJ, Lins S, Ohno-Shosaku T, Trube G, and Panten U (1988) Cytosolic ADP Efendic´S, Cerasi E, and Luft R (1975) Effect of phentolamine and preperfusion with ϩ glucose on insulin release from the isolated perfused pancreas from fasted and fed enhances the sensitivity to tolbutamide of ATP-dependent K channels from rats. Diabetologia 11:407–410. pancreatic B-cells. FEBS Lett 239:241–244. Efendic S, Efanov AM, Berggren PO, and Zaitsev SV (2002) Two generations of insulinotropic imidazoline compounds. Diabetes 51:S448–S454. Address correspondence to: Dr. Ingo Rustenbeck, Institute of Pharmacol- Eglen RM, Hudson AL, Kendall DA, Nutt DJ, Morgan NG, Wilson VG, and Dillon ogy and Toxicology, University of Braunschweig, Mendelssohnstr. 1, D-38106 MP (1998) ‘Seeing through a glass darkly’: casting light on imidazoline ‘I’ sites. Braunschweig, Germany. E-mail: [email protected] Trends Pharmacol Sci 19:381–390.