The sulfonylurea receptor

45

Biochimica et Biophysica Acta, 1175 (1992) 45-59 © 1992 Elsevier Science Publishers B.V. All rights reserved 0167-4889/92/$05.00

Minireview

BBAMCR 13306

The sulfonylurea receptor Stephen J.H. Ashcroft and Frances M. Ashcroft Nuffield Department of Clinical Biochemistry and Laboratory of Physiology, University of Oxford, Oxford (UK) (Received 21 July 1992)

Key words: ATP-sensitive potassium channel; Sulfonylurea; Diazoxide; Glibenclamide

Contents Sulfonylureas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Discovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Mechanism of action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

45 45 46

The sulfonylurea receptor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Pharmacology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Changes with development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Specificity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Effects of nucleotides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Involvement of phosphorylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Effects of fluorescein derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H. Endosulfine, an endogenous modulator of the sulfonylurea receptor . . . . . . . . . . . . . . . . . . . I. Effects of other modifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. Location of binding sites for glibenclamide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . K. Mechanism of interaction of sulfonylureas with the receptor . . . . . . . . . . . . . . . . . . . . . . . . . . L. Receptor characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

47 47 47 48 50 50 51 51 51 52 52 53 53

III. Relationship between the sulfonylurea receptor and the K-ATP channel . . . . . . . . . . . . . . . . . . .

55

IV. Intracellular actions of sulfonylureas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

56

V.

56

II.

Model of the sulfonylurea receptor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

VI. Future goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

I. Sulfonylureas I-A. Discovery W i t h t h e r ap i d d e v e l o p m e n t o f s u l f o n a m i d e t h e r a p y d u r i n g t h e S e c o n d W o r l d W a r , t h e b a c t e r i o s t a t i c action

Correspondence to: S.J.H. Ashcroft, Nuffield Department of Clinical Biochemistry, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, UK.

o f t h e s e drugs was a p p l i e d to m a n y infectious diseases. M a r c e l J a n b o n in M o n t p e l l i e r t r e a t e d a r o u n d 30 typ h o i d p a t i e n t s with p - a m i n o b e n z e n e s u l f a m i d o i s o p r o p y l t h i o d i a z o l e (2254 R P ) an d r e p o r t e d , in 1942, a high i n c i d e n c e of s y m p t o m s correctly i n t e r p r e t e d as d u e to h y p o g l y c a e m i a [1]. T h e r e a s o n for this r e s p o n s e was e l u c i d a t e d by A u g u s t e L o u b a t i ~ r e s w h o showed, in dogs, that 2254 R P c a u s e d h y p o g l y c a e m i a via stimulation o f insulin secretion. T h e s e p i o n e e r i n g studies (rev i e w e d by L o u b a t i ~ r e s [2]) led to clinical trials a nd t he successful use o f this class o f s u l f o n a m i d e s (called

46 sulfonylureas) to treat human diabetes [3-5]. The first sulfonamide lacking bacteriostatic action but retaining the hypoglycaemic and antidiabetic effects of 2254 RP was tolbutamide, in which the para-amino group of 2254 RP was replaced by a methyl group [6], and which subsequently came into general use in diabetes therapy. More than 12000 analogues have since been synthesized [7] and second-generation sulfonylureas such as glibenclamide were produced that have over 1000fold greater potency than tolbutamide. Interestingly, it was also Loubati~res who noted that certain sulfonamides, of which the most active was 3-methyl-7chloro- 1,2,4-benzothiadiazine- 1,1-dioxide (diazoxide) had a hyperglycaemic effect [8], later shown to result from inhibition of insulin secretion [9].

tion, shown in vitro to involve direct stimulation of the /j-cell [10], was the demonstration that tolbutamide depolarized pancreatic /j-cells and induced electrical activity [11]. Studies of ion fluxes suggested that the depolarisation induced by sulfonylureas resulted from a decrease in the K+-permeability of the/J-cell plasma membrane [12]. Diazoxide suppressed the effects of tolbutamide on both electrical activity and ion fluxes [13]. Application of the patch clamp technique then demonstrated that sulfonylureas were specific blockers of ATP-sensitive K-channels (K-ATP channels) [14] whilst diazoxide elicited opening of these channels [15,62]. K-ATP channels were first identified in cardiac muscle by Noma [16], and subsequently shown to occur in many tissues and play a variety of physiological roles [17]. These channels are inhibited by an increase in the concentration of ATP at the intracellular membrane surface and are modulated in a complex way by other adenine nucleotides (for review see Ashcroft [18]). In the /J-cell, the main physiological stimulus to

LB. Mechanism of action The first important landmark in understanding the mechanism of action of sulfonamides on insulin secre-

¢

Cl o

.

II

C-N(CH2) 2

S-N-C-N ii l I

H'

II

~,--/oH

H~--/

OCHs Glibenclamide (glyburide)

Linosi~de --N

0

~

II

H~c

0

m

c-Ncc-~

0

m

/---'%

Glipizide O

Meslitinide 0

II II ~-$ -N-C-N-(CH,2)3CH It I I

H3C-( ¢

°

S-N-C-N,

~/-- O H

¢o

3

H

Tolbutamide 0 0 0 H CO A l I 2r-]k l! II 3 ,~I"" " N - (CH2h--~ ")--S-N.C,N-(

~ O H

H ~''/

)

AZ-DF 26S

Me Me NH O

Gliquidone

N s=L~s'

%~o C1~ S ' - N H

RP-49356 Diazoxide

Nil 2 Minoxidil sulfate

Pinacidil

O

N N

C

~

OH

'LXoC Cmmakalim

Nlcorandil

Fig. 1. D r u g s i n t e r a c t i n g w i t h t h e s u l f o n y l u r e a r e c e p t o r .

47 insulin secretion is a rise in blood glucose concentration. This rise leads to an increased rate of glucose metabolism by the/3-cell and a consequent increase in the intracellular ratio of [ATP]/[ADP] which inhibits K-ATP channels. The resulting depolarisation opens voltage-sensitive CaZ+-channels and the ensuing influx of Ca 2÷ triggers insulin release (for review see Refs. 19,20). Hence, K-ATP channels play a central role in the response of the /3-cell to the main physiological regulator of insulin secretion and also to the main pharmacological agents used clinically to stimulate (sulfonylureas) or inhibit (diazoxide) insulin release. A key question, considered further below, is whether sulfonylureas and/or diazoxide interact directly with the K-ATP channel, or whether their effects are mediated by binding to some other (channel-related) protein. The first evidence for high affinity specific binding sites for sulfonylureas was obtained in studies on rat brain and /3-cell tumour [21,22]. Since then, as appreciation of the importance of K-ATP channels in a variety of tissues has grown, there has been a considerable amount of effort devoted to characterizing the sulfonylurea receptor. The present review attempts to summarize our current knowledge. We confine the review to effects of sulfonylureas in animals and do not consider their action as herbicides [23,24].

TABLE I

Brain rat whole rat forebrain rat cerebral cortex rat cerebral cortex rat cerebral cortex rat cerebellum rat cerebral cortex pig cerebral cortex pig cerebral cortex rat hypothalamus

II. The suifonylurea receptor

Heart muscle rat ventricle rat ventricle dog ventricle

Both high and low affinity binding sites for sulfonylureas have been identified. In this review we focus primarily on the high-affinity sulfonylurea-binding site which we refer to as the sulfonylurea receptor. Only limited data are available on the low-affinity binding sites. II-A. Pharmacology Fig. 1 gives the formulae of a number of drugs which have been shown to interact with the sulfonylurea receptor. These include the hypoglycaemic sulfonylureas tolbutamide and glibenclamide; non-sulfonylurea hypoglycaemic drugs such as meglitinide and linogliride; and K-channel openers such as diazoxide, nicorandil, cromakalim and minoxidil. II-B. Distribution Binding sites for sulfonylureas have been studied by autoradiography of tissue sections exposed to 3H- or 125I-labelled sulfonylureas, or by binding assays. The latter have employed intact ceils, isolated membrane and solubilized membrane preparations. High-affinity binding sites for [3H]glibenclamide ( K i < 10 nM) have been described in cardiac and smooth muscle, in pancreatic /3-cells and in brain microsomes. Table I lists

Glibenclamide binding - high-affinity sites Tissue

Endocrine pancreas HIT/3-cell HIT/3-cell HIT/3-cell (early passage) HIT/3-cell (late passage) RINm5F/3-cell ob / o b mouse islet ob / o b mouse islet rat insulinoma Smooth muscle Guinea-pig myenteric Guinea-pig illeal longitudinal Canine aorta

a

Kj (nM)

0.8 1.2 6.9 3.8 0.3 0.4 0.3 7 0.42 0.06 10

Bmax (fmol/mg protein) 1090 546 1640 160 930 1400 930

Reference

99 42 81 81 100 79 36 105 a

156 23 420

101 38 102

0.195 0.14 0.06 0.07 1.5 0.44 2.13 0.05 0.8 2.7

340 92 63 209 110 156 85 141 400 50

103 33 40 38 34 a 27 104 40 75 104

0.05 0.077 0.2

36 65 6

38 33 48

Used [3H]glipizide.

the affinity and density of high affinity glibenclamidebinding sites in the tissues so far studied. The reported K d for glibenclamide ranges from 0.05 to 10 nM and the/3-cell appears to have the highest density of sites. Low-affinity binding sites (K~ > 10 nM) have been reported for all of these tissues and also for adipocytes. Their properties are summarized in Table II. Quantitative autoradiography has been used to determine the localization of [3H]glibenclamide binding within the brain. In the adult animal, [3H]glibenclamide binding sites are concentrated in the substantia nigra (zona reticulata), the septohippocampal nucleus, the globus paflidus, the caudate putamen, the sensory motor cortex, the molecular layer of the cerebellum and regions of the hippocampus [25,27,32]. Similar results have been found using [125I]iodoglibenclamide which has the advantage of a rapid exposure time [28,29]. Within the hippocampus the highest levels of binding are observed in the dentate gyrus and the stratum lucidum of the CA3, where the mossy fibres synapse with the dendrites of the CA3 neurones. Evidence that sulfonylurea binding is localized to the

48 TABLE II

eas of the brain, such as the ventromedial hypothalamus which is known to contain K-ATP channels, were also poorly labelled [27]. A role for K-ATP channels in the response to cerebral ischaemia is provided by the finding that the density of glibenclamide-binding sites is reduced in the CA3 region, hilus and dentate gyrus of the hippocampus following forebrain ischemia [26]. It has also been reported that the density of glibenclamide binding sites is reduced in the neonatal rat and turtle brain, which are far more resistant to anoxia than the adult rat brain [106]. Also of interest is the finding that there is an 80% decrease in the density of binding sites in the molecular layer of the cerebellum of the weaver mouse, in which a single gene mutation gives rise to defective motor activity [32].

Glibenclamide binding - low-affinity sites Tissue

Kd (nM)

Endocrine pancreas HIT fl-cell HIT fl-cell HIT/3-cell (early passage) HIT/3-cell (late passage)

Bmax (fmol/mg protein)

136 277 16100 14000

800000 960000

42 74 81 81

395

4956

38

83 2

3100 53

101 38

1800

220 000

96

136 40

336 295

38 48

Brain rat cerebral cortex Smooth muscle guinea-pig myenteric guinea-pig illeal longitudinal Adipocytes rat

575

Reference

H-C. Changes with development

Heart rat ventricle dog ventricle

A detailed study of the ontogenesis of sulfonylurea binding sites in the rat brain used quantitative autoradiography to localize changes in the density and distribution of binding sites during development [32]. This study show that the density of binding sites was very low in newborn animals, with only the hindbrain showing some very small degree of labelling. By the third postnatal day some binding activity was detectable which increased with age, reaching about 50% of maximum around 18 days after birth. The develop-

mossy fibre terminals comes from experiments in which sulfonylurea binding activity disappeared following selective destruction of the mossy fibres [30,31]. Regions of the hindbrain such as the pons and the medulla, and the spinal cord, show little binding activity. In adult rat, binding sites were observed in the spinal cord with a rostral to cordal gradient [33]. Glucose-responsive ar-

TABLE III

Displacement of glibenclamide from high-affinity binding sites (K i) brain

HIT-/3cell

RIN 0-cell

o b / o b islet

smooth muscle

Species

hamster

rat

mouse

guinea pig

rat

dog

chicken

guinea pig

rat

[101] b

[38]

[48]

[86l

[86]

[38]

0.09 10

0.4 5 200

Reference Glibenclamide (nM) Gliquidone (nM) Glibornuride (nM) Glipizide (nM) Meglitinide (/z M) Chlorpropramide (/zM) Tolbutamide (p,M) Tolazamide (/zM) RP49356 (/zM) Diazoxide (p,M)

[99] 0.7

[41]a 7

45 60 25 7

55

[100] 0.3 1 90 0.8 0.26 9 9

[79] 0.4

10 4

0.16 0.27 136 2.8 0.27

15

0.09

2.5 2500

1 22 5 0.8

0.07

0.8

10

15

9 2.8 0.6

[34] c 0.25 1.6 120 4.4 32 10

4.8 > 140

Dexamethasone (/xM) Prazosine (/z M) Diazepam (p,M) Verapamil (#M) Nifedipine (/~ M) Quinine (mM) Data are IC50: divide by 2 to obtain K i. b Data are IC50: divide by 8 to obtain K i. c Used [3H]glipizide. Data are IC50: divide by 2 to obtain K i. a

cardiac ventricle

Tissue

500

150

14 44 190

14 44 190

200 600 300 75

49 opment. Miller et al. [33] compared the density of high affinity [3H]glibenclamide binding sites in rat neuronal and cardiac tissue during development. The density of

ment of binding activity appeared to correlate with the maturation of neuronal connectivity. The Kd, however, remained constant at around 0.5 nM throughout devel-

TABLE IV Effects o f nucleotides on the sulfonylurea receptor

Nucleotide

Glibenclamide receptor

Effect on binding

Source

Preparation

- Mg 2 +

+ Mg 2 +

Rat cortex Rat cortex Mouse islet HIT T15/3-cells

Membranes Solubilized Membranes Membranes

0 0 0 0

~, $

Rat cortex

Membranes

~ 1"

$

Rat cortex Pig brain HIT T15/3-cells HIT T15/3-cells Mouse islet

Membranes Purified Intact cells Membranes Membranes

0 ~ nt J, 0

$

AMP

Pig bain Rat cortex

Purified Membranes

0 0

0 0

[39] [38]

ATPyS

Rat cortex Mouse islets HIT T15/3-cells

Membranes Membranes Membranes

~ 0 nt

$ $ 0

[381 [36] [35]

ADP/3 S

Rat cortex

Membranes

~

AMP-PNP

Mouse islets HIT T15 fl-cells Rat cortex Pig brain

Membranes Membranes Membranes Purified

0

AMPS

Rat cortex

Membranes

~

AMP-PCP

HIT T15/3-cells

Membranes

nt

0

GTP

Rat cortex Rat cortex Mouse islets

Membranes Membranes Membranes

0 nt 0

J,

Rat cortex Rat cortex Rat cortex Mouse islets

Membranes Membranes Solubilized Membranes

,L 0 nt 0

$ $

GMP

Rat cortex Pig brain

Membranes Purified

0 0

0 0

[381 [39]

GTPyS

Rat cortex Mouse islets

Membranes Membranes

0 0

$ $

[38] [36]

GDP/3S

Rat cortex Mouse islets

Membranes Membranes

~ 0

,L ],

[381 [36]

GMP-PNP

Mouse islets Rat cortex

Membranes Membranes

0 0

0 0

[36]

UTP

Rat cortex

Membranes

0

$

UDP

Rat cortex

Membranes

0

ITP

Rat cortex

Membranes

0

IDP

Rat cortex

Membranes

J,

ATP

ADP

GDP

nt = not tested.

0 J,

Comment

References

J, Bmax I'Kd Required DTT SBmax I"Kd No effect of DI"T

[38,40] [40] [36,37] [35]

1"K~ ADP ~,Bmax below 3 mM but increased binding above 3 mM Attributed to ATP formation MgADP ],Bmax K0.s 100/~M

$ Attributed to ATP formation

[381 [40] [39] [43] [35,42,43] [36]

[38] 0 0 nt J,

Max. ~,25% K0.5 200/zM

[36] [35] [40] [39] [38] [35]

As potent as MgATP

Attributed to ATP formation

[38] [401 [36] [38] [401 [40] [36]

[401 As potent as MgATP

[38] [38]

As potent as MgATP ],

[381 [38]

50 binding sites increased 250% in forebrain membranes during postnatal development but was constant in ventricular membranes.

II-D. Specificity The radiolabelled sulfonylurea most frequently utilized for binding studies is [3H]glibenclamide although [3H]glipizide [34,105] and [3H]gliquidone [21,72] have also been used. The specificity of the sulfonylurea receptor has been assessed by comparing the ability of unlabeled sulfonylureas and other drugs to displace [3H]glibenclamide binding. The reported K i values are given in Table III. In general the K~ values of sulfonylureas correlate well with their ability to stimulate insulin secretion and close the /3-cell K-ATP channel, In other tissues, as discussed below, the correlation appears less good. Diazoxide displaces [3H]glibenclamide binding from intact HIT/3-cells at concentrations similar to those required to inhibit insulin release and open K-ATP channels [35], but, as considered below, its effect on /3-cell membranes is influenced by the concentration of MgATP.

II-E. Effects of nucleotides Since electrophysiological studies have indicated that nucleotides modulate K-ATP channels in several different ways, the effects of nucleotides on [3H]glibenclamide binding have been studied in some detail. Table IV summarizes the effects of different nucleotides tested on sulfonylurea receptors from various sources. Inhibition of [3H]glibenclamide binding by MgATP has been observed for membranes from HIT T15/3-cells [35], mouse islets [36,37], rat cerebral cortex [38-40] and pig brain [39]. In all of these studies there was no effect of ATP in the absence of Mg 2÷. Inhibition by MgATP was potentiated by dithiothreitol in brain membranes [38,40] but there are conflicting reports as to whether the effect of dithiothreitol persists after membrane solubilization [39,40]. In both mouse islet membranes [36] and solubilized brain membranes [40], MgATP increased the apparent K d for [3H]glibenclamide, which could indicate a competitive effect. However, the apparent K d was not the linear function of MgATP concentration expected for competitive inhibition [37]. This indicates that MgATP does not compete for the sulfonylurea binding site. As is the case for ATP, the ability of GTP to displace [3H]glibenclamide binding from rat cortex and mouse islet membranes is dependent on the presence of Mg 2÷ [36,38,40] The data for nucleoside diphosphates are conflicting. MgADP caused marked inhibition of [3H]glibenclamide binding to membranes from HIT T15 /3-cells [35], rat cortex [38,40] and mouse islets [36]. This

inhibitory effect of MgADP persisted after solubilization of the receptor from pig brain [39] but not from HIT B-cells [41] and was found to be non-competitive (K~ = 0.1 mM) in solubilized pig brain membranes [39]. However, in the absence of Mg 2+, inhibition of [3H]glibenclamide binding by ADP was observed in intact HIT/3-cells and HIT/3-cell membranes [42,43] but not in mouse islets [36]. No effect of ADP in the absence of Mg 2÷ was found in rat cortex in one study [40] but in another study [38] a dual effect of ADP was observed; at concentrations below 3 mM, ADP (in the absence of Mg 2+) inhibited [3H]glibenclamide binding non-competitively (IC5o = 0.6 mM) but at higher concentrations actually increased [3H]glibenclamide binding. Small but significant inhibition by ADP in the absence of Mg 2÷ was also found for the solubilized pig brain receptor [39]. GDP was effective in displacing [3H]glibenclamide binding from rat cortex both in the presence and absence of Mg z+ in one study [38] but required the presence of Mg 2+ in another [40] and also was Mg2+-dependent in mouse islets [36]. Neither AMP nor GMP, whether in the presence or absence of Mg 2÷, has any significant effect on [3H]glibenclamide binding to the brain sulfonylurea receptor [38,39]. ATPyS displaced [3H]glibenclamide binding to rat cortex membranes both in the absence and the presence of MgZ+: at 1 mM it was even more effective than MgATP [38]. However, ATPyS was effective only in the presence of Mg 2+ in mouse islet membranes [36] and was without effect at all on HIT cell membranes [35]. GTPyS is effective only in the presence of Mg z÷ both in rat cortex and in mouse islet membranes [36,38]. ADP/3S and GDP/3S reduced [3H]glibenclamide binding to rat cortex membranes equally well in the absence or presence of Mg 2+ [38] but GDP/3S was only effective on mouse islet membranes in the presence of Mg 2+ [36]. AMP-PCP, AMP-PNP and GMP-PNP are in general reported to be ineffective in inhibiting [3H]glibenclamide binding to brain or/3-cell membranes [35-37], although a small inhibition by AMP-PNP of binding to the purified pig brain receptor was observed [39]. UTP and ITP in the presence, but not the absence, of Mg 2+ were as effective as MgATP in inhibiting [3H]glibenclamide binding to rat cortex membranes; IDP resembled ADP in that inhibition of binding was independent of the presence of Mg 2+ but UDP was only effective in the presence of Mg z+ [38]. These data lead to the following conclusions. Since in rat cortex membranes the inhibition of [3H]glibenclamide binding by ADP in the absence of Mg 2+ was prevented by ATP or other nucleoside triphosphates, two interacting nucleotide-binding sites may be present in the sulfonylurea receptor, one resulting in inhibition, the other in stimulation of sulfonylurea binding

51 [38]. The former binds MgADP, Mg-nucleoside triphosphates and nucleoside diphosphates, including ADP (for which the affinity constant is less than 1 raM). Occupation of the second site by ADP at higher concentrations stimulates binding relative to the lowered value obtained when the first site is occupied. Electrophysiological studies have also led to models for the K-ATP channel in which nucleotides are envisaged to bind to two separate sites [44-47]. The situation is further complicated in HIT T15 /3-cells where ADP appears able to compete with glibenclamide at an external binding site [43]. The data also raise the possibility that phosphorylation of the sulfonylurea receptor may modulate its binding properties which is considered in the next section. There is evidence that the low-affinity glibenclamide-binding site in cardiac ventricular membranes is modulated by a G-protein since Gpp[NH]p altered the binding characteristics of the low- (but not the high-) affinity site [48].

II-F. Involvement of phosphorylation In intact HIT T15/3-cells, ATP-depletion with 2-deoxyglucose and oligomycin, which decreases intracellular ATP to submillimolar levels, decreased the specific binding of [3H]glibenclamide, suggesting that protein phosphorylation may play a positive role in the binding between glibenclamide and its receptor [35]. However, as Table IV makes clear, MgATP has been invariably found to decrease the binding of [3H]glibenclamide to its receptor. That this effect of MgATP may be mediated by phosphorylation of the receptor is suggested by the requirement for Mg 2+, and by the ineffectiveness of most non-hydrolysable ATP analogues. Consistent with this idea, we have found that /3-cell membrane preparations contain both endogenous protein kinase and phosphatase activities. Additionally it has been shown that the effect of MgATP on the apparent dissociation constant for binding of [3H]glibenclamide to mouse islet membranes is not in accordance with a competitive interaction of glibenclamide and MgATP [37]. Further evidence that protein phosphorylation may be involved in sulfonylurea binding is the reversal of MgATP-induced inhibition of [3H]glibenclamide binding to mouse islet membranes by exogenous alkaline phosphatase [36] On the other hand, the inhibitory effects of nucleoside diphosphates are not readily explained in terms of protein phosphorylation. Schwanstecher et al. [36,37] suggest that MgADP and MgGDP may be converted to MgATP and MgGTP by transphosphorylation, since in their studies MgADP and MgGDP were not effective when the concentrations of MgATP and MgGTP were kept low by the addition of hexokinase and glucose; how-

ever there is no direct evidence for this possibility. Moreover both Niki & Ashcroft [35] and Gopalakrishnan et al. [38] have raised the opposite possibility i.e. that ADP might mediate the effect of ATP, since rapid conversion of ATP to ADP occurs with membrane preparations from both rat cortex and HIT /3-cell membranes. Studies of the effects of K-channel openers have revealed a further way in which protein phosphorylation may modulate the sulfonylurea receptor. Patch clamp studies suggested that protein phosphorylation was required for the opening effect of diazoxide on K-ATP channels [49,50]. There is also evidence that other K-ATP channel openers require MgATP to open /3-cell K-ATP channels but become inhibitory in the absence of ATP [51,52]. Diazoxide was shown to decrease [3H]glibenclamide binding to intact HIT T15 /3-cells, an effect which was less pronounced after ATP depletion [35], but with isolated membrane preparations diazoxide alone was unable to inhibit [3H]glibenclamide binding [36,42]. However, when MgATP is also present diazoxide is able to displace [3H]glibenclamide from HIT /3-cell [35] or islet [36] membranes. This effect of MgATP is reproduced by ATPyS but not by non-hydrolysable ATP analogues (AMP-PNP or AMPPCP) or by ATP in the absence of Mg 2+ [35,36]. The presence of MgATP was also necessary for displacement of [3H]glibenclamide from HIT cell and rat cortex membranes by pinacidil, an effect which was sustained after solubilization of the membranes [53]. Both diazoxide [35] and pinacidil [53] decreased the number of binding sites but not the affinity of [3H]glibenclamide for the receptor. This suggests that channel openers and sulfonylureas do not compete for the same site.

II-G. Effects of fluorescein derivatives Fluorescein and its derivatives have been used to label the nucleotide-binding sites in various ATPases. De Weille et al. [54] have shown that fluorescein derivatives have both activatory and inhibitory effects on K-ATP channel activity in HIT T15/3-cells. Binding studies showed that specific binding of [3H]glibenclamide to HIT cell membranes was inhibited by phloxine B and Bengal rose with K0.5 values of 2 and 0.2 nM, respectively. These data provide further support for the notion that there are interactions between the sulfonylurea-binding site and one or more nucleotidebinding sites on the sulfonylurea receptor.

II-H. Endosulfine, an endogenous modulator of the sulfonylurea receptor It has been suggested that the high affinity of the sulfonylurea receptor for glibenclamide might indicate the presence of a natural ligand for the receptor [34]

52 which would modulate biological processes by the same mechanism(s) as that involved in the pharmacological response to sulfonylureas. A peptide capable of displacing [3H]glibenclamide from rat cerebral cortex membranes was detected in extracts from rat cortex [55]. This endogenous ligand has been extracted from ovine brain, purified using cation exchange chromatography and HPLC and partially characterized; two peptides, which differed in their isoelectric point, were obtained and given the names a- and fl-endosulfine [56]. Each form of endosulfine had a similar ability to displace [3H]glibenclamide from rat brain membranes and the B-form was shown to stimulate insulin release from B-TC cells. This latter effect suggests that, like sulfonylureas, endosulfine may inhibit K-ATP channel activity. These findings have important implications for B-cell function where endosulfine may be envisaged as playing a role in regulating both B-cell secretion and proliferation. Furthermore, lack of endosulfine may be of significance in the processes leading to non-insulindependent diabetes. Endosulfine may also be of importance in regulating K-ATP channel activity in the CNS. For example, it might stimulate ventromedial hypothalamic neurones and thereby influence appetite control. It has been argued that K-ATP channels may play a role in the tissue regulation of cerebral blood flow, since activation of K-ATP channels during metabolic stress may be expected to lead to enhanced K-efflux and thus to vasodilation [17]. Inhibition of this response by endosulfine may thus play a role in the response to cerebral ischaemia. Cloning of endosulfine will permit investigation of these possibilities.

II-I Effects of other modifiers A requirement for dithiothreitol has been noted for inhibition by MgATP of [3H]glibenclamide binding to rat cortex membranes [38,39]. This may suggest the involvement of thiol groups at or near the sulfonylurea binding site, consistent with electrophysiological studies of the effects of sulfhydryl reagents on K-ATP channels [57]. However, after solubilization of the rat cortex receptor, the requirement for dithiothreitol was no longer evident [40]. Modification of the membrane by enzymatic treatment can give information about receptor structure. Exposure of rat brain membranes to proteolytic (trypsin or chymotrypsin) or lipolytic (phospholipases A 2 or C) reduced subsequent binding of [3H]glipizide. Approx. 50% of binding activity was lost after treatment with the proteolytic enzymes and up to 93% after the phospholipases [34] These data indicate that the brain sulfonylurea receptor contains a protein component associated with phospholipid. Interestingly, trypsin also markedly affects K-ATP channel activity [58]. When applied to the intracellular membrane surface in mouse

B-cells, trypsin increased channel activity in the absence of ATP and also decreased the inhibitory effect of ATP.

II-J. Location of binding sites for glibenclamide Early studies reviewed by Gylfe et al. [59] suggested that tolbutamide reached its site of action in the B-cell membrane directly from the extracellular space. Thus the distribution volume in pancreatic islets for tolbutamide and several other sulfonylureas was shown to only slightly exceed that for extracellular space markers [60,61]. However, Trube et al. [62] showed that K-ATP channels were blocked by tolbutamide regardless of whether the drug was applied to the inside or the outside of the membrane. Tolbutamide was also effective in cell-attached patches when added to the bath solution, which suggested that tolbutamide could dissolve in the membrane and diffuse laterally to the blocking site. The location of the sulfonylurea receptor is not defined by these observations, however, since tolbutamide could be reaching an extracellular receptor via diffusion through the membrane. Experiments to test how sulfonylureas approach their receptor were reported by Ziinkler et al. [63]. Tolbutamide and related compounds are weak acids and the concentration of the undissociated, lipid-soluble form increases 100fold on decreasing pH from 8.4 to 6.4. When the concentration of the undissociated form of the drug was kept constant as the pH of the bath solution was increased it was found that neither the rate nor the extent of block of K-ATP channels by tolbutamide or meglitinide was affected. Raising the pH of the bath solution at a constant concentration of the sulfonylurea, however, decreased both the rate and extent of block. These data show that it is the undissociated form of the sulfonylurea that interacts with the K-ATP channel. Since these studies used the whole-cell configuration, the receptor site could be accessible to the undissociated sulfonylurea either directly from the extracellular space or from the lipid phase of the membrane. However, calculation of the forward rate constants for block of K-ATP channels from analysis of the kinetics of the current records suggested that the drugs might have to be distributed into the membrane before reaching their receptor. The faster onset of response to glibenclamide is consistent with the greater lipid solubility of this sulfonylurea. The effects of pH on inhibition by sulfonylureas of K-ATP channels in cardiac muscle also support the view that the unionised form of the drug forms the actual inhibitory ligand by virtue of its ability to enter the membrane lipid [64]. On the other hand, two pieces of evidence argue that the sulfonylurea-binding site is accessible from the extracellular milieu. Firstly both ADP and ADPagarose competively displaced [3H]glibenclamide from

53 intact HIT T15 O-cells [43]. The effect of ADP was accompanied by increased intraceUular Ca 2÷ and insulin secretion. Secondly the existence of an endogenous peptide ligand for the sulfonylurea receptor, endosulfine, which displaces [3H]glibenclamide and stimulates insulin secretion [56] also suggests that the [3H]glibenclamide-binding site is extracellular.

II-K. Mechanism of interaction of sulfonylureas with the receptor Perhaps the most striking result from studies of structure/function relationships of hypoglycaemic sulfonylureas is that the SO2NHCONH group itself is not essential for functional activity. Geisen et al. [65] showed that meglitinide, a benzoic acid derivative similar to the non-sulfonylurea moiety of glibenclamide, also had hypoglycaemic properties. It was subsequently demonstrated that the mode of action of meglitinide on the O-cell was indeed identical to that of tolbutamide or glibenclamide [66]; similar findings were reported for the non-sulfonylurea moiety of gliquidone [67]. Brown & Foubister [68] studied the blood glucose lowering activity of analogues of meglitinide. It was found that potent hypoglycaemic activity required a carboxyl group or a group that can be oxidized to carboxyl in vivo, such as methyl, to be present in the aromatic ring of the phenethyl group. The role of the benzamide group in meglitinide was ascribed to protein binding. However, a benzamide group is also nonessential. The benzoic acid derivative AZ-DF-265, in which a substituted carbamoyl group replaces the benzamide group of meglitinide, also behaves similarly to the sulfonylureas in stimulating insulin release and decreasing 86Rb-efflux from islets of Langerhans [69] with a potency similar to that of glibenclamide. AZDF-265 has also been shown to close K-ATP channels and to displace [3H]glibenclamide from intact 0-cells [70]. Henquin et al. [71] tested the stimulatory effects on mouse islets of Langerhans of derivatives of meglitinide. Of the two halves of the parent molecule, p-ethylbenzoic acid was weakly effective but 5-C1-2methoxybenzamide was inactive. Replacement of the COOH group decreased but did not entirely abolish activity. Both positive and negative effects of modification of the substituents of the benzamide ring were noted. There is thus evidence that the sulfonylurea receptor recognises two groups, the sulfonylurea or other acidic group and a benzamide or related group. Neither moiety is essential either for binding or for closure of K-ATP channels. Glibenclamide which possesses both sulfonylurea and benzamide moieties may be a bifunctional agent, each molecule binding to two closely located sites on the receptor. This would be consistent with the positive cooperativity (Hill coefficient = 2.1)

observed for binding to RIN O-cells [70]. It is possible that the NHCOCH group of AZ-DF-265 permits this analogue to bind simultaneously at both sites; steric hindrance could then account for the negative cooperativity of AZ-DF-265 (Hill coefficient = 0.5) in displacing [3H]glibenclamide [70]. It is not clear whether the two sites are separate binding domains on the same protein or different parts of the same domain. Verspohl et al [72] have provided evidence that the binding site for gliquidone in RIN 0-cell membranes is distinct from that for glibenclamide.

II-L. Receptor characterization Photoaffinity labelling of the sulfonylurea receptor was first attempted by Kramer et al. [73] Membranes prepared from a transplantable rat O-cell tumour were incubated with [3H]glibenclamide in the dark and then illuminated for 2 min with UV light at 254 nm. Following SDS-PAGE and autoradiography, radioactive bands of 140 and 33 kDa were obtained. Since the extent of labelling of the former but not the latter was decreased specifically by sulfonylureas it was concluded that the 140 kDa polypeptide was a component of the sulfonylurea receptor. This study indicated two potential problems in the use of glibenclamide as a photoaffinity probe. Firstly, the sulfonylurea is able to label nonspecific proteins, in particular albumin. Secondly, the efficiency of covalently labelling the putative membrane receptor is low; the peak area of the 140 kDa species was approx. 300 cpm corresponding to around 5 fmol [3H]glibenclamide-binding sites per 60/zg protein. Since 0-cell membranes contain around 1000 fmol glibenclamide-binding sites per mg, the efficiency of labelling was thus only about 0.75%. A similar efficiency was found by Aguilar-Bryan et al. [74] using an iodinated glibenclamide derivative as a photoaffinity probe. Using pig brain microsomes, Bernardi et al. [75] also observed photoaffinity labelling by [3H]glibenclamide of a 150 kDa polypeptide. A similar sized species was labelled in both intact and solubilized receptors and in active fractions eluted from WGA-AffiGel and ADPagarose affinity columns. However, after SDS-PAGE of HIT cell proteins prepared from intact cells covalently labelled by UV illumination after incubation with an iodinated glibenclamide derivative, Niki et al. [41] found that the main species labelled was 65 kDa. When isolated rat islets were photolabeled by the 125I-glibenclamide derivative, a band of identical M r was detected. It was suggested that the 140-150 kDa species could be a dimer formed by cross-linking on photoillumination. A problem with these studies is the unknown nature of the photo-reaction. Nelson et al. [76] have carried out a detailed examination of the effects of varying the

54 conditions for photo-labelling /3-cell microsomes with [5-125I]-2-hydroxyglibenclamide. Results were found to depend markedly on the wavelength and energy of the UV light. Under certain conditions at 254 nm a 65 kDa species was the major band whereas at 312 nm a 140 kDa species predominated. It was also found necessary to use a Laemmli sample buffer at pH 9 to avoid aggregation. The main finding from this study was that covalent labelling of the 140 kDa peptide was reduced by nanomolar concentrations of unlabeled glibenclamide, whereas the three lower M r species (65, 50, 33 kDa) required micromolar concentrations. It was concluded that the 140 kDa species contains the /3-cell high-affinity glibenclamide binding site. Further evidence that the 140-150 kDa species may represent the sulfonylurea receptor has come from studies using covalent labelling of nucleotide binding sites [39]. 5-[a32p]ATP was oxidized with metaperiodate and then incubated with a purified sulfonylurea-binding protein from pig brain. Sodium cyanoborohydride was present to reduce the Schiff base formed between aldehyde groups of the oxidized [a-32p]ATP and the receptor. SDS-gel electrophoresis and autoradiography showed that the major incorporation of 32p was into a 145 kDa species. Excess non-radioactive ATP prevented labelling, indicating that incorporation was occurring at an ATP-binding site. Labelling of three additional peptides of M r 80, 64 and 50 kDa was also prevented by ATP. The 145 kDa species labelled by oxidized [a32P]ATP migrated identically to the protein labelled by photoillumination with [3H]glibenclamide. Thus the same protein appears to possess both sulfonylurea- and ATP-binding sites. Moreover diazoxide inhibited covalent labelling of the 145 kDa species by oxidized [a32p]ATP consistent with this peptide containing a binding site for the channel opener. ADP, AMP-PNP and ATP-yS protected against labelling with oxidized [o~32p]ATP but AMP and adenosine were ineffective; sulfonylureas also did not protect. Using probes based on known K-channels we and others have so far been unable to isolate clones from /~-cell cDNA libraries that may correspond to the KATP channel. Therefore expression screening or protein purification are likely to be the necessary routes to successful characterization of the K-ATP channel/ sulfonylurea receptor. An important prerequisite for purification is the ability to solubilize the sulfonylurea receptor in active form. Using HIT cell membranes photolabeled with [5-~25I]-2-hydroxyglibenclamide, Aguilar-Bryan et al. [74] were able to solubilize labelled proteins with a variety of detergents. The solubilized membrane retained sulfonylurea binding activity [77]. Niki et al. [41] solubilized active sulfonylurea receptor from HIT-T15 /3-cells using CHAPS. The solubilized receptor showed a single class of non-interacting sites with a K d of 3.3 nM, similar to that

observed for binding to intact HIT cells. The relative potencies of sulfonylureas in displacing [3H]glibenclamide from the solubilized receptor were similar to that observed in microsome preparations. The sulfonylurea receptor solubilized from HIT cells also displayed inhibition of [3H]glibenclamide binding by MgATP and an enhancing effect of MgATP on displacement of [3H]glibenclamide by pinacidil [53]. Active sulfonylurea receptor has also been solubilized from rat and pig cerebral cortex [40,53,75]. Bernardi et al. [75] used digitonin to solubilize active sulfonylurea binding activity from pig brain microsomes; a yield of 40% of active sulfonylurea receptor was reported. However, the properties of the solubilized receptor were not defined. In detergent extracts of rat or pig cerebral cortex, 300/zM-MgATP reduced the number of high-affinity [3H]glibenclamide binding sites by 52% and increased the K d for [3H]glibenclamide eightfold; MgATP was half maximally effective at 41/zM [40]. Purification of the sulfonylurea receptor has been reported for pig brain [75] and for HIT/3-cells [74]. Pig brain membranes were first solubilized with digitonin. Sulfonylurea binding activity was purified 2500-fold by a four-step procedure involving (i) hydroxyapatite chromatography; (ii) ADP-agarose affinity chromatography; (iii) wheat germ agglutinin chromatography; (iv) chromatography on a mixture of AMP-agarose/GMPagarose/hydroxyapatite. The best preparation, which gave a single species of 150 kDa on SDS-PAGE, had a binding capacity of 1 nmol/mg protein. This is considerably less than the value of 6.7 nmol/mg expected for a 150 kDa protein with one binding site. The discrepancy was attributed to underestimation of the final binding capacity because of the short half-life of the purified receptor (12 h at 4°). Partial purification (60-fold) of HIT cell sulfonylurea receptor was reported by Aguilar-Bryan et al. [74]. These workers did not follow binding activity through their procedure. Instead HIT cell membrane were first photo-labelled with [5-125I]-2-hydroxyglibenclamide. Photolabeled membranes were then solubilized with digitonin and subjected to (i) DEAE-Sephacel ion exchange chromatography; (ii) agarose exclusion chromatography; (iii) dodecyl-agarose hydrophobic interaction chromatography. At each stage the receptor was equated with the presence of a radiolabelled species of M r 140000 on SDS-PAGE. Although this procedure could only have yielded a final preparation that was about 0.3% pure receptor, the 140 kDa species was well resolved from other proteins on SDS-PAGE and could be excised from the gel in a form suitable for microsequencing. Disappointingly, neither of the above studies has so far led to publication of any sequence data for the sulfonylurea receptor. Our own studies have investigated the use of affinity

55 chromatography on immobilized sulfonylurea derivatives. Such matrices are highly effective in binding sulfonylurea receptor and we have successfully eluted active receptor with glibenclamide. Around 500-fold purification can be achieved [78]. Thus up to the present time molecular characterization of the sulfonylurea receptor remains to be achieved. As yet there are no primary sequence data nor any reports on the M r of the native protein. Definite answers to a number of key questions, in particular the relationship of the receptor to the K-ATP channel, are currently awaiting success in cloning the sulfonylurea receptor. III. Relationship between the sulfonylurea receptor and the K-ATP channel In this section we shall be concerned only with the effects of sulfonylureas on ion channel activity. We shall not discuss the intracellular actions of sulfonylureas (described below) because the high lipid solubility of the drugs suggests that there may be separate intracellular receptors which mediate these events. In pancreatic/3-cells, there is a very good agreement between the concentration of glibenclamide or tolbutamide required to half-maximally inhibit channel activity and the binding affinity of these drugs to /3-cell membranes [79]. Moreover sulfonylureas are without effect on other types of ion channels in the/3-cell [80]. Estimates of channel density invariably yield lower values than the measured number of [3H]glibenclamide-binding sites, suggesting they may be different proteins. However, estimates of channel density are notoriously unreliable as they rest upon the value that is taken for the channel open probability in the intact cell, a value which is very difficult to quantify. Another argument that the sulfonylurea receptor and the K-ATP channel must be distinct entities is that the response to both tolbutamide and diazoxide disappear with time in excised patches, although channel activity remains. It is contended that this arises as a result of the loss of some soluble cytosolic factor which links the sulfonylurea receptor to the channel and which is gradually lost after patch excision. An alternative interpretation of these findings, however, is that some cytosolic constituent is simply required for the binding of these drugs to the channel/receptor complex (or for its activation). Similar opposing arguments can be applied to the finding that limited proteolysis by trypsin removes the ability of the channel to respond to tolbutamide [114]. Perhaps the strongest argument in favour of the idea that the sulfonylurea receptor and the K-ATP channel are closely linked, however, is provided by the finding that there is a parallel loss in K-ATP channel activity and high-affinity binding sites for [3H]glibenclamide with passage number in the /3-

cell line HIT T15 [81]. In the/3-cell therefore, the data are consistent with the view that the sulfonylurea receptor and the K-ATP channel are very closely associated. It is clear, however, that in other tissues the sulfonylurea receptor is not as closely associated with the K-ATP channel as in the/3-cell. There is an accumulating body of evidence that sulfonylureas act on ion channels other than the K-ATP channel. Thus glibenclamide has been reported to inhibit other types of K-channel, including the delayed outward current of neuroblastoma cells [82] and the slowly inactivating D-type K-current of hippocampal neurones (30% block by 3-10/zM) [83,84]. Effects on chloride channels have also been described. The cystic fibrosis transmembrane conductance regulator (CFTR) is a chloride channel which is blocked by tolbutamide and glibenclamide at half-maximal concentrations of 150 ~M and 20 /zM, respectively [85]. In contrast to the K-ATP channel, the K channel openers diazoxide, minoxidil and BRL38227 also inhibit whole-cell chloride currents. It is of interest that glibenclamide inhibition became voltage-dependent in mutants in which the R-domain of the channel was deleted: this domain is intracellular and contains a large number of phosphorylation sites [85]. These findings, and the fact that the CFTR channel is activated by direct binding of ATP, may indicate some structural similarities between CFTR and the K-ATP channel. They also suggest that the sulfonylurea receptor may associate with other types of ion channel. In some tissues the correlation between the affinity of [3H]glibenclamide binding and channel inhibition is less good than in the /3-cell. For example, in cardiac myocytes, the K i for glibenclamide binding is 2.5 nM [86] whereas that for channel inhibition was reported to be much higher (0.5/xM) [87]. A similar discrepancy is found for tolbutamide [86,88]. Although this may indicate that the receptor and the channel constitute separate proteins, it remains possible that the difference is simply the result of dialysis of the cell interior with the pipette solution. For example, ADP is known to decrease the channel sensitivity to tolbutamide in cardiac cells [87] and this nucleotide is likely to increase in dialysed ceils as a consequence of reduced metabolism. Perforated patch recordings are now needed to distinguish between these possibilities. It is worth noting that, in contrast to cardiac muscle, cytosolic ADP enhances the sensitivity to tolbutamide of K-ATP channels in /3-cells [89]. Moreover, in isolated ventricular myocytes from guinea pig hearts, glibenclamide inhibited the whole-cell KAT e currents with an ECs0 of 6 nM, close to the K i for glibenclamide binding [113], calling into question the apparent discrepancy previously observed. The effects of sulfonylureas on K-ATP channel activity in ventromedial hypothalamic neurones also ar-

56 gue that the receptor and the channel may be distinct. These neurones possess a K-ATP channel which differs from that found in most other tissues in being voltageactivated, of larger conductance (140 pS) and requiring approx. 3 mM ATP to produce half maximal inhibition [90]. Tolbutamide blocks channel activity in cell-attached patches but is ineffective when applied in the inside-out configuration [91]. A cytosolic constituent which facilitates tolbutamide block has not yet been identified. More surprising, perhaps, is the finding that glibenclamide does not inhibit channel activity itself but is able to block the effects of tolbutamide [91]. One important question is whether the effects of sulfonylureas on the activity of K-ATP channels in tissues other than the B-cell, and on other types of ion channel are mediated by different sulfonylurea receptors. The inhibitory effects of sulfonylureas have invariably been tested on other channel types at a concentration several times higher than the K i for sulfonylurea binding and in many cases dose-response curves have not been published. It is therefore possible that the K-ATP channel is linked to a class of high-affinity sulfonylurea receptors whereas other types of ion channel are linked to a different class of receptor of lower affinity. The fact that the relative potency of sulfonylureas varies between tissues may also suggest a family of sulfonylurea receptors. For example, in the substantia nigra, gliquidone > glipizide > glibenclamide > tolbutamide [92], a different sequence from that found in fl-cells [93]. In conclusion, the question of whether the sulfonylurea receptor forms (part of) the K-ATP channel remains open. The recent cloning of a K-ATP channel from kidney [115] should permit resolution of this question.

IV. Intracellular actions of sulfonylureas While there is no doubt that the major hypoglycaemic action of sulfonylureas is mediated by stimulation of insulin secretion, many studies have suggested additional extra-pancreatic effects including inhibition of gluconeogenesis [94], stimulation of glucose uptake by muscle [95] and adipose tissue [96], and activation of glycogen synthesis [97]. It is unclear to what extent any of these effects may in fact be mediated by K-ATP channels. We shall restrict discussion therefore to the one case in which binding studies have been simultaneously performed. Martz et al. [96] correlated the stimulatory action of sulfonylureas on glucose transport in adipocytes with their binding to adipocyte membranes. Two distinct classes of glibenclamide-binding sites were observed. A non-saturable site was shown to represent lipid binding. A saturable site bound glibenclamide with a K d of around 2 ~M, similar to the activation constant for potentiation of glucose transport by

Op-Dz ~

Op-MgADP-Dz

kinase

C

C-Sulph

=

~

Cp ~

Op

~

Op-MgADP

Bp-ATP

Fig. 2. Model for the sulfonylurea receptor/K-ATP channel..

glibenclamide. Displacement of [3H]glibenclamide was elicited by tolbutamide and tolazamide, which also increased glucose transport, but also by meglitinide and ciglitazone which did not. It was suggested that two subpopulations of sulfonylurea-binding sites were present, only one of which is relevant to effects on glucose transport. The exact role of this low-affinity receptor in glucose transport potentiation is not known. Evidence for an intracellular action of sulfonamides on insulin secretion has been reported in abstract form [98]. Using permeabilized islets or RINm5F /3-cells, insulin release was enhanced by tolbutamide, glipizide and glibenclamide and inhibited by diazoxide. Assessment of the significance of these findings requires demonstration of specific intracellular sulfonylurea-binding sites, which have not so far been reported.

V. Model of sulfonylurea receptor In this section we attempt to construct a model of the sulfonylurea receptor which accommodates all the present experimental findings (Fig. 2). The following makes the assumption that the K-ATP channel and the sulfonylurea receptor form a single functional unit. This does not imply that they are the same protein. From kinetic analysis of K-ATP channel openings it appears that there is a least one open state and two closed states [107,108]. The channel openings occur in bursts, with the channel flickering between the open and a short closed state during the burst and entering a much longer closed state(s) between bursts of openings. Although there may be further (long) closed states, we have grouped all the long closed states together together as a single state, C. The short closed state we refer to as Cp. It is well documented that K-ATP channel activity declines upon formation of an inside-out membrane patch (rundown) but can be restored by brief exposure of the intracellular membrane surface to MgATP [109]. Since Mg-free solutions and non-hydrolysable ATP analogues are ineffective, it is suggested that restoration of channel activity may involve protein phosphorylation. As rundown is associated with an increase in the long closed state, we suggest that in this state (C) the channel is dephosphorylated and that phosphoryla-

57 tion results in the channel spending an increased time in the shorter closed state, Cp. We also assume that channel remains phosphorylated in the open state (Op). ATP clearly binds to the open state of the channel and inhibits it: we define this ATP-blocked state as Bp-ATP. Other nucleotides may also bind to this state and inhibit channel activity but as they are less effective they must do so with lower affinity. In pancreatic /3-cells [110], but not cardiac muscle [44], the free ion A T P 4- is more effective than MgATP. In some tissues, such as cardiac muscle [44], more than one ATP molecule appears to bind to the channel since the Hill coefficient is greater than one. In this model, we have lumped all ATP-blocked states together as Bp-ATP. Although we have shown the channel entering the ATP-blocked state from the open state, the data are equally consistent with the channel entering the ATPblocked state from the phosphorylated closed state Cp. Electrophysiological studies also indicate that both diazoxide [15,50,62] and MgADP [111,112] can increase channel activity. We therefore assume that these agents bind to and stabilise the open state and refer to them as Op-Dz and OpMgADP, respectively. By contrast, sulfonylureas inhibit channel activity by producing an increase in the long closed time. We therefore suggest that these drugs bind preferentially to the dephosphorylated closed state C. Although the model is based on electrophysiological data, it can account for most of the data on sulfonylurea binding. In the absence of MgATP it may be assumed that most of the channels reside in the dephosphorylated closed state (C.) Thus because sulfonylureas bind to this state (C), ATP would not be expected to influence binding, as is indeed found experimentally. In the presence of MgATP, some channels will reside in the open state (Op) and may therefore enter the blocked state Bp-ATP. This accounts for the inhibitory effect of MgATP on sulfonylurea binding. The fact that diazoxide reduces sulfonylurea binding in the presence of MgATP (but not ATP alone) is explained by the fact that diazoxide preferentially binds to the open phosphorylated state (Op). The model is open to test. For example, it suggests that in the presence of diazoxide (or MgADP) there must be two open states. If the time spent in these open states differs then it may be possible to resolve this in kinetic studies of K-ATP channel activity. There are a number of findings which this model does not easily account for. First, Mg-ADP can decrease sulfonylurea binding. Our model can only explain this finding if we assume that MgADP is converted to MgATP by endogenous transphosphorylating enzymes, as suggested by Schwanstecher and colleagues [36]. Secondly, the model cannot account for the ability of fluorescein derivatives (a) to displace glibenclamide binding; and (b) to reactivate rundown

K-ATP channels in the absence of ATP [54]. However, there is no direct evidence that these agents bind to the ATP-binding site as is commonly assumed: it is also possible that these fluorescein derivatives modulate the activity of a kinase a n d / o r phosphatase, which could explain their effects on both sulfonylurea binding and channel activity. Finally, we have not attempted to account for the finding that in pancreatic/3-cells intracellular MgADP increases the efficacy of tolbutamide inhibition of K-ATP channel currents [89] whereas the opposite is true in cardiac muscle [87].

VI. Future goals A major target for future studies must be the cloning of the sulfonylurea receptor and determination of its primary structure. This should answer the question of whether the sulfonylurea receptor and the K-ATP channel constitute separate proteins or whether they are a single entity. If the former is the case, it will then be necessary to explore whether the sulfonylurea receptor may associate with ion channels other than the K-ATP channel. It will be important to determine whether there exists a family of sulfonylurea binding proteins with different tissue distribution and phamacological profiles and to identify the relationship between the low and high affinity binding sites. Cloning of the sulfonylurea receptor may also be expected to elucidate our understanding of the complex influences of nucleotides on sulfonylurea binding. The interaction between diazoxide and other K-channel openers with the sulfonylurea receptor also requires further analysis and the reason for the marked tissue specificity of the K-channel openers needs to be identified.

Acknowledgements Work from our own laboratories has been supported by the Medical Research Council, the British Diabetic Association, the Wellcome Trust, the Parkinson's Disease Society, the Royal Society, the E.P. Abraham Fund, the University of Oxford Medical Research Fund, Nordisk U.K. & Glaxo Inc. (U.S.A.)

References 1 Janbon, M., Chapal, J., Vedel, A. and Schaap, J. (1942) Montpellier m6d. 21-22, 441. 2 Loubati~res, A. (1969) Acta Diabet. lat 6 (Suppl. 1), 20-56. 3 Loubati~res, A. (1955) Montpellier m6d. 48, 618. 4 Bertram, F., Bendfeldt, E. and Otto, H. (1955) Deutsch Med. Woschenschr. 1455. 5 Franke, H. and Fuchs, J. (1955) Deutsch Med. Woschenschr. 80, 1449. 6 B~inder, A., Creutzfeldt, W., Dorfmiiller, T., Erhart, H., Marx, R., Maske, H., Meier, W., Mohnike, G., Pfeiffer, E.F., Schlagin-

58 weit, S., Sch6ffling, K., Scholz, J., Seidler, I., Steigerwald, H., Stich, W., St6tter, G. and Ulrich, H. (1956) Deutsch Med. Woschenschr. 81,823. 7 Biinder, A. (1969) Acta Diabet. lat. 6 (suppl.l), 429-440. 8 Loubati~res, A. (1944) C.R. Soc. Biol. (Paris) 138, 830. 9 Loubati~res, A., Alric, A., Mariani, M.M., De Malbosc, H. and Ribes, G. (1966) C.R. Soc. Biol. (Paris) 160, 168. 10 Bouman, P.R. and Goorenstroom, J.H. (1961) Metabolism 10, 1095 - 1099 11 Dean, P.M. and Matthews, E.K. (1968) Nature 219, 389-390. 12 Henquin, J.C. (1980) Diabetologia 18, 151-160. 13 Henquin, J.-C and Meissner, H.P. (1982) Biochem. Pharmacol. 31, 1043-1052. 14 Sturgess, N.C., Ashford, M.L.I., Cook, D.L. and Hales, C.N (1985) Lancet ii: 474-475. 15 Sturgess, N.C., Kozlowski, R.Z., Carrington, C.A., Hales, C.N. and Ashford, M.L.J. (1988) Br. J. Pharmacol. 95, 83-94 16 Noma, A. (1983) Nature 305, 147-148. 17 Ashcroft, S.J.H. and Ashcroft, F.M. (1990) Cell. Signal. 2, 197214. 18 Ashcrofl, F.M. (1988) Annu. Rev. Neurosci. 11, 97-118 19 Ashcroft, S.J.H. and Ashcroft, F.M. (1989) in Hormones and Cell Regulation No. 14 (Nunez, J. and J.E. Dumont, J.E., eds.), Colloque INSERM/J. Libbey Eurotext, 198, 99-103. 20 Ashcroft, F.M. and Rorsman, P. (1989) Prog. Biophys. Mol. Biol. 54, 87-143. 21 Kaubisch, N. Hammer, R., Wollheim, C.B., Renold, A.E. and Offord, R.E. (1982)Biochem. Pharmacol. 31, 1171-1174. 22 Geisen, K., Hitzel, V., 0komonopoulos, R., Printer, J., Weyer, R. and Summ, H.D. (1985) Arzn. Forsch 35, 707-712. 23 Blair, A.M. and Martin, T.D. (1988) Pestic. Sci. 22, 195-219. 24 Brown, H.M. (1990) Pestic. Sci. 29, 263-281. 25 Mourre, C., Ben Ari, Y., Bernardi, H., Fosset, M. and Lazdunski, M. (1989) Brain Res. 486, 159-164. 26 Mourre, C., Smith, M.-L, Siesj6, B.K. and Lazdunski, M. (1990) Brain Res. 526, 147-152. 27 Treherne, J.M. and Ashford, M.L.J. (1991) Neuroscience 40, 523-531. 28 Gehlert, D.R., Gackenheimer, S.L., Mais, D.E. and Robertson, D.W. (1991) J. Pharmacol. Exp. Ther. 257:901-907 29 Gehlert, D.R., Mais, D.E., Gackenheimer, S.L., Krushinski, J.H. and Robertson, D.W. (1990) Eur. J. Pharmacol. 186, 373-375 30 Mourre, C., Widmann, C. and Lazdunski, M. (1991) Brain Res. 540, 340-344. 3l Tremblay, E., Zini, S. and Ben-Ari, Y. (1991) Neurosci. Lett. 127, 21-24. 32 Mourre, C., Widmann, C. and Lazdunski, M. (1990) Brain Res 519, 9-43. 33 Miller, J.A., Velayo, N.L., Dage, R.C. and Rampe, D. (1991) J. Pharmacol. Exp. Ther. 256, 358-364. 34 Lupo, B. and Bataille, D. (1987) Eur. J. Pharmacol. 140, 157-169. 35 Niki, I. and Ashcroft, S.J.H. (1991) Biochim. Biophys. Acta Mol. Cell Res. 1133, 95-101. 36 Schwanstecher, M., L6ser, S., Rietze, I. and Panten, U. (1991) Naunyn-Schmiedeberg's Arch. Pharmacol. 343, 83-89. 37 Schwanstecher, M., l~ser, S., Brandt, C., Scheffer, K., Rosenberger, F. and Panten, U. (1992) Br. J. Pharmacol. 105, 531-534. 38 Gopalakrishnan, M., Johnson, D.E., Janis, R.A. and Triggle, D.J. (1991) J. Pharmacol. Exp. Ther. 257, 1162-1171. 39 Bernardi, H., Fosset, M. and Lazdunski, M. (1992) Biochemistry 31, 6328-6332. 40 Schwanstecher, M., Schaupp, U., L6ser, S. and Panten, U. (1992) J. Neurochem. 59, 1325-1335. 41 Niki, I., Welsh, M., Berggren, P.-O., Hubbard, P. and Ashcroft, S.J.H. (1991) Biochem. J. 277, 619-624. 42 Niki, I., Kelly, R.P., Ashcroft, S.J.H. and Ashcrofl, F.M. (1989) Pfliigers Arch. 415, 47-55.

43 Niki, I., Nicks, J.L. and Ashcrofl, S.J.H. (1990) Biochem. J. 268, 713-718. 44 Lederer, W.J. and Nichols, C.G. (1989) J. Physiol. (Lond.) 419, 193-211. 45 Findlay, I. (1987) J. Physiol. (Lond.) 391,611-629 46 Schwanstecher, C., Dickel, C. and Panten, U. (1992) Mol. Pharmacol. 41, 480-486. 47 Hopkins, W.F., Fatherazi, S., Peter-Riesch, B., Corkey, B.E. and Cool, D.L. (1992) J. Membr. Biol., in press. 48 French, J.F., Riera, L.C., and Sarmiento, J.G. (1990) Biochem. Biophys. Res. Commun. 167, 1400-1405. 49 Dunne, M.J. (1989) FEBS Lett. 250, 262-266. 50 Kozlowski, R.Z., Hales, C.N. and Ashford, M.L.J. (1989) Br. J. Pharmacol. 97, 1039-1050. 51 Dunne, M.J. (1990) Br. J. Pharmacol. 99, 487-492 52 Dunne, M..J, Aspinall, R.J. and Petersen, O.H. (1990) Br. J. Pharmacol. 99, 169-175 53 Schwanstecher, M., Brandt, C., Behrends, S., Schaupp, U. and Panten, U. (1992) Br. J. Pharmacol. 106, 295-301. 54 De Weille, J.R., Miiller, M., Lazdunski, M. (1992) J. Biol. Chem. 267, 4557-4563. 55 Virsolvy-Vergine, A., Briick, M., Dufour, M., Cauvin, A., Lupo, B. and Bataille, D. (1988) FEBS Lett. 24, 265-69. 56 Virsolvy-Vergine, A., Leray, H., Kuroki, S., Lupo, B., Duf0ur, N. and Bataille, D. (1992) Proc. Natl. Acad. Sci. 89, 6629-6623. 57 Weik, R., Neumcke, B. (1989) J. Membr. Biol. 110, 217-226 58 Trube, G., Hescheler, J. and Schr6ter, K. (1989) In Secretion and its control, Society of General Physiologists Series, Vol. 44, pp. 84-95 (Oxford, G.S. and Armstrong, C.M., eds.). 59 Gylfe, E., Hellman, B., Sehlin, J. and T~iljedal, I.-B. (1984) Experientia 40, 1126-1134 60 Hellman, B., Sehlin, J. and Tiiljedal, I.-B. (1971) Biochem. Biophys. Res. Commun. 45, 1384-88 61 Sehlin, J. (1973) Acta Diabetol. Eat. 10, 1052-1060 62 Trube, G. Rorsman, P. and Ohno-Shosaku, T. (1986) Pfliigers Arch. 407, 493-499. 63 Ziinkler, B.J., Trube, G. and Panten, U (1989) NaunynSchmiedeberg's Arch. Pharmacol. 340, 328-332 64 Findlay, I. (1992) J. Pharmacol. Exp. Ther. 262, 71-79. 65 Geisen, K., Hiibner, M., Hitzel, V., Hrstka, V.E., Pfaff, W., Bosies, E., Regitz, G., Kiihnle, H.F., Schmidt, F.H. and Weyer, R. (1978) Arzn. Forsch. 28, 1080-1083 66 Garrino, M.-G., Schmeer, W., Nenquin, M., Meissner, H.P. and Henquin, J.C. (1985) Diabetologia 28, 697-703 67 Garrino, M.G., Meissner, H.P. and Henquin, J.C. (1986) Eur. J. Pharmacol. 124, 309-. 68 Brown, G.R. and Foubister, A.J. (1984) J. Med. Chem. 27, 79-81. 69 Garrino, M.G. and Henquin, J.C. (1988) Br. J. Pharmacol. 93, 61-68. 70 Ronner, P., Hang, T.L., Kraebber, M.J. and Higgins, T.J. (1992) Br. J. Pharmacol. 106, 250-255 71 Henquin, J.-C, Garrino, M.G. and Nenquin, M. (1987) Eur. J. Pharmacol. 14, 243. 72 Verspohl, E.J., Ammon, H.P.T. and Mark, M. (1990) J. Pharm. Pharmacol. 42, 230-235 73 Kramer, W., Oekonomopulos, R., Punter, J. and Summ, H.-D. (1988) FEBS Lett 229, 355-359. 74 Aguilar-Bryan, L., Nelson, D.A., Vu, Q.A., Humphrey, M.B. and Boyd, A.E. lII (1990) J. Biol. Chem. 265, 8218-8224 75 Bernardi, H., Fosset, M. and Lazdunski, M. (1988) Proc. Natl. Acad. Sci. USA 85, 9816-9820. 76 Nelson, D.A., Aguilar-Bryan, L. and Bryan, J. (1992) J. Biol. Chem. 267, 14928-14933. 77 Boyd, A.E. III, Aguilar-Bryan, L., Bryan, J., Kunze, D.L., Moss, L., Nelson, D.A., Rajah, A.S., Raef, H., Xiang, H. and Yaney, G.C. (1991) Rec. Prog. Horm. Res. 47, 299-317

59 78 Niki, I., Hubbard, P.M. and Ashcroft, S.J.H. (1991) Diabetologia 34 (Suppl 2), A63 79 Panten, U., Burgfeld, J., Goerke, F., K~m-~ick~, M., Schwanstecher, M., Wallasch, A., Ziinkler, B.J. and Lenzen,, S. (1989) Biochem Pharmacol 38, 1217-1229. 80 Trube, G., Rorsman, P., Ohno-Shosaku, T. (1986) Pfliigers Arch. 407, 493-499 81 Aguilar-Bryan, L., Nichols, C.G., Rajan, A.S., Parker, C. and Bryan, J. (1992) J. Biol. Chem. 267, 14934-14940. 82 Reeve, H.L., Vaughan, P.F.T. and Peers, C. (1992) Neurosci. Lett. 135, 27-40 83 Cr6pel, V., Krnjevic, K., Ben-Ari, Y. (1992). J. Physiol., in press. 84 Cr6pel, V., Krnjevic, K. and Ben-Ari, Y. (1992) Can. J. Physiol. Pharmacol. 70, 306-307. 85 Sheppard, D.N. and Welsh, M.J. (1992) J. Gen. Physiol. 100, 573-593. 86 Fosset, M., de Weille, J.R., Green, R.D., Schmid-Antomarchi, H. and Lazdunski, M. (1988) J. Biol. Chem. 263, 7933-7936 87 Venkatesh, N., Lamp, S.T. and Weiss, J.N. (1991) Circ. Res. 69, 623-637. 88 Belles, B., Hescheler, J. and Trube, G. (1987) Pfliigers Arch. 409, 582-588 89 Ziinkler, B.J., Lins, S., Ohno-Shosaku, T., Trube, G. and Panten, U. (1988) FEBS Lett. 239, 241-244 90 Ashford, M.L.J., Boden, P.R. and Treherne, J.M. (1990) Pfliigers Arch. 415, 479-483. 91 Ashford, M.L.J., Boden, P.R. and Treherne, J.M. (1990) Br. J. Pharmacol. 101,531-540. 92 Amoroso, S., Schmid-Antomarchi, H., Fosset, M. and Lazdunski, M. (1990) Eur. J. Pharmacol. 183, 459. 93 Zunckler, B.J., Lenzen, S., Manner, K., Panten, U. and Trube, G. (1988) Naunyn-Schmeidebergs Arch. Pharmacol. 337, 225-30. 94 Cabello, M.A., Rodr~guez-Tarduchy, G., Ortega, J.L., Samper, B. and Felcju, J.E. (1991) Metabolism 40, 934-940. 95 Davidson, M.B., Molnar, I.G., Furman, A. and Yamaguchi, D. (1991) Diabetes 40, 1531-1538. 96 Martz, A., Jo, I., Jung, C.Y. (1989) J. Biol. Chem. 264, 1367213678.

97 Mojena, M., Marcos, M.L., Monge, L., Fel~u, J.E. (1989) Metabolism 38, 466-470. 98 Shibier, O., Flatt, P.R., Efendlc, S. and Berggren, P.-O. (1991) Diabetologia 34 (Suppl. 2), A91 99 Gaines, K.L., Hamilton, S. and Boyd, A.E. III (1988) J. Biol. Chem. 263, 2589-92. 100 Schmid-Antomarchi, H., De Weille, J., Fosset, M., Lazdunski, M. (1987) J. Biol. Chem. 262, 15840-4. 101 Zini, S., Yehezkel, B.-A. and Ashford, M.J.L. (1991) J. Pharmacol. Exp. Therap. 259, 566-573 102 Kovacs, R.J. and Nelson, M.T. (1991) Am. J. Physiol. 261, H604-H609 103 Robertson, D.W., Schober, D.A., Krushinski, J.H., Mais, D.E., Thompson, D.C. and Gehlert, D.R. (1990) J. Med. Chem. 33, 3124-3126. 104 Ashford, M.L.J., Boden, P.R. and Treherne, J.M. (1990) J. Physiol. 420, 93P 105 Siconolfi-Baez, L., Banerji, M.A. and Lebovitz, H.E. (1990) Diabetes-Care 13 (Suppl 3), 2-8 106 Jiang, C., Ying, X. and Haddad, G.G. (1992) J. Physiol. 448, 599-612 107 Ashcroft, F.M., Ashcroft, S.J.H., Harrison, D.E. (1988) J. Physiol. 400, 501-527 108 Davies, N.W., Standon, N.B. and Stanfield, P.R. (1991) J. Bioenerg. Membr. 23, 509-535 109 Ohno-Shosaku, T. Ziinkler, B.J. and Trube, G (1987) Eur. J. Physiol. 408, 133-138. 110 Ashcroft, F.M. and Kakei, M. (1989) J. Physiol. (Lond.) 416, 349-367 111 Kakei, M. Kelly, R.P., Ashcroft, S.J.H. and Ashcroft, F.M. (1986) FEBS Lett. 208, 63-66. 112 Bokvist, K., ,~mm~l~i, C., Ashcroft F.M., Berggren, P.-O., Larsson, O. and Rorsman, P. (1991). Proc. R. Soc. B: 243, 139-144.6 113 Findlay, I. (1992) J. Pharmacol. Exp. Ther. 261, 540-545. 114 Proks, P. and Ashcroft, F.M. (1992) J. Physiol., in press. 115 Ho, K., Vassilev, P.M., Kanazirska, M.V., Lytton, J. and Hebert, S.C. (1992) Nature, in press.