Mechanisms by which glucose can control insulin release independently from its action on adenosine triphosphate sensitive K+ channels in mouse B cells

Mechanisms by which glucose can control

insulin release independently from its action on

adenosine triphosphate-sensitive K+ channels

in mouse B cells.

M Gembal, … , Z Y Gao, J C Henquin

J Clin Invest.

1993;

91(3)

:871-880.

https://doi.org/10.1172/JCI116308

.

Glucose stimulation of insulin release involves closure of ATP-sensitive K+ channels

(K(+)-ATP channels), depolarization, and Ca2+ influx in B cells. However, by using diazoxide to

open K(+)-ATP channels, and 30 mM K to depolarize the membrane, we could demonstrate

that another mechanism exists, by which glucose can control insulin release independently

from changes in K(+)-ATP channel activity and in membrane potential (Gembal et al. 1992.

J. Clin. Invest. 89:1288-1295). A similar approach was followed here to investigate, with

mouse islets, the nature of this newly identified mechanism. The membrane

potential-independent increase in insulin release produced by glucose required metabolism of the

sugar and was mimicked by other metabolized secretagogues. It also required elevated

levels of cytoplasmic Cai2+, but was not due to further changes in Cai2+. It could not be

ascribed to acceleration of phosphoinositide metabolism, or to activation of protein kinases

A or C. Thus, glucose did not increase inositol phosphate levels and hardly affected cAMP

levels. Moreover, increasing inositol phosphates by vasopressin or cAMP by forskolin, and

activating protein kinase C by phorbol esters did not mimic the action of glucose on release,

and down-regulation of protein kinase C did not prevent these effects. On the other hand, it

correlated with an increase in the ATP/ADP ratio in islet cells. We suggest that the

membrane potential-independent control of […]

Research Article

Find the latest version:

http://jci.me/116308/pdf

Mechanisms by

Which Glucose Can Control Insulin

Release

_{Independently}

_from

Its

Action

on

Adenosine

Triphosphate-sensitive

K+

Channels in

Mouse B

Cells

MarekGembal, Philippe Detimary, Patrick Gilon,Zhi-YongGao, and Jean-ClaudeHenquin

Unite'deDiabtologieetNutrition, Faculty of Medicine, University ofLouvain, B-1200Brussels, Belgium

Abstract

Glucose stimulation of insulin release involves closure of ATP-sensitiveK+channels(K+-ATP channels),

depolarization,

and

Ca"

influx in B cells. However, by usingdiazoxide to open K+-ATP channels, and 30 mM K todepolarizethemembrane,

wecoulddemonstrate that anothermechanism

exists,

bywhich glucose cancontrol insulinrelease

independently

from changes

inK+-ATP channelactivityand in membranepotential (Gem-bal et al. 1992. J. Clin. Invest.

89:1288-1295).

A similar ap-proach wasfollowedhere to

investigate,

withmouse

islets,

the natureof this newly identified mechanism. Themembrane

po-tential-independent

increase in insulin release

produced by

glu-cose required metabolism ofthe sugar and was mimickedby

other metabolizedsecretagogues.Italso

required

elevated lev-elsofcytoplasmic

CO',

butwasnotduetofurther

changes

in

CO'.

It could not be ascribed to acceleration of phosphoinosi-tide

metabolism,

or to activation of

protein

kinases A or C.

Thus, glucose did not increase inositol phosphate levels and

hardly affected cAMP levels. Moreover,

increasing

inositol

phosphates byvasopressinorcAMP by

forskolin,

and

activat-ingprotein kinase C by phorbolestersdidnotmimic theaction

of glucoseon

release,

and

down-regulation

of

protein

kinaseC

didnot preventtheseeffects. Onthe other

hand,

itcorrelated

withanincreasein the

ATP/ADP

ratioinisletcells. We sug-gest that the membrane

potential-independent

controlof

insu-lin release exerted by glucose involves changes inthe energy

stateofB cells.(J. Clin. Invest.

1993.91:871-880.)

Key

words:

adenine nucleotides *calcium * insulin secretion-

pancreatic

islets*stimulus-secretion coupling

Introduction

The mechanisms by which glucose controls pancreatic Bcell functionarecomplex(

1-4).

Anumberof

studies

have estab-lished thata

metabolic

controlof ionicevents inBcells is a

critical event in stimulus-secretion

coupling

(4-9), and

as-Partsof this workwerepresented atthe28th Annual Meetingof the EuropeanAssociation for the Study of Diabetes, Prague, 8-11 Sep-tember1992,andhave appeared in abstract form ( 1992.Diabetologia. 35[Suppl. l]:A122).

AddressreprintrequeststoDr. J. C.Henquin,Unite d'Endocrino-logieet Metabolisme, UCL 55.30, Avenue Hippocrate, 55, B-1200 Brussels, Belgium.

Receivedforpublication20 July 1992 and in revisedform 16 No-vember 1992.

signed

toATP-sensitiveK+ channels(K+-ATP channels)' in the

plasma

membranea

pivotal

rolein thiscontrol

( 10-14).

It

isnowunanimously acceptedthat theentryofglucoseinB cells isfollowed

by

anaccelerationofmetabolismthatgeneratesone or several

signals

which close these K+-ATP channels. The

resulting

decreaseinK+ conductanceleadstodepolarizationof the membranewith

subsequent

openingof

voltage-dependent

Ca"

channels.

Ca"

influx through these channels then

in-creases,

leading

toarise incytoplasmic

Ca2+,

which

eventually

activatesaneffectorsystemresponsibleforexocytosisof insu-lingranules. The essential role ofK+-ATP channelsis

strik-ingly

illustrated

by

the inhibition of insulin release

brought

about

by

diazoxide.This

drug selectively

opens

K+-ATP

chan-nels( 15, 16) and, thereby,causesrepolarization of the

plasma

membrane

(

17) andarrest of

Ca2+

influx (17-19),

although

themetabolismof

glucose

isunabated

( 19, 20).

Recently, however,

we

presented

evidence that

glucose

can control insulinreleaseindependently from its actionon

K+-ATPchannels

(21 ).

Theexistenceofthismechanismcouldbe

established

by opening

K+-ATP channels withdiazoxide,but

restoring

Ca2+

influx by depolarizingthe membranewith

high

extracellular K .Underthese conditionsglucosestill causeda

concentration-dependent increase of insulin release (21 ). In the present studywe have investigated the possible mecha-nismsunderlying thismembranepotential-independentwayof

regulating insulin release.

Methods

Solutions. The medium used was a bicarbonate-buffered solution whichcontained 120 mM NaCl, 4.8mMKCI, 2.5mMCaCl2, 1.2mM MgCl2,and 24 mM NaHCO3. It wasgassed with02/C02 (94:6)to maintain pH 7.4 andwassupplemented with bovine serum albumin ( 1 mg/ml). Ca2+-free solutionswerepreparedby replacingCaCI2 with MgCl2 and addition of 100,uMEGTA. When the concentration of KC1 wasincreasedto30 mM, that of NaCl wasdecreased accordingly to maintain isoosmolarity.

Measurements of insulin release. Allexperimentswere performed with islets isolated by collagenase digestionof the pancreas of fed fe-maleNMRI mice (25-30 g), killed by decapitation.

Inafirsttypeof experiments, the isletswerefirst preincubated for 60minin acontrolmedium containing 15 mM glucose. The islets were thenplaced in batches of 25 in parallelperifusionchambers and peri-fusedat aflowrateof1.25ml/min(22). Effluent fractionswere col-lectedat2-minintervals,andinsulinwasmeasuredby a double-anti-bodyradioimmunoassaywith ratinsulinasthestandard (Novo Re-search Institute, Bagsvaerd, Denmark).

In anothertypeofexperiments,after the initial preincubation of 60

1. Abbreviationsused in thispaper:AVP, argininevasopressin; K+-ATPchannels, ATP-sensitive K+ channels; KIC, a-ketoisocaproate; aPDD, 4a-phorbol 12,13-didecanoate; PKC, protein kinase C; PMA, phorbol 12-myristate 13-acetate.

Glucose Control ofInsulin Release Independent from K+Channels 871 J.Clin. Invest.

© The American Society for Clinical Investigation,Inc.

min the islets were distributed in batches of three. Each batch of islets was then incubated for 60 min in 1 ml of medium containing appro-priate concentrations of glucose and test substances. A portion of the medium was withdrawn at the end of the incubation and appropriately

diluted for insulinassay.

Measurements ofcAMP. After theinitial preincubationof 60min

in a controlmedium,batches of 12 islets were incubated for 60 min in 1 mlof medium containing appropriate concentrations of glucose and test substances. At the endofthe incubation the tubes were briefly

centrifugedand 0.95 ml of medium withdrawn and saved for insulin assay. The medium was immediately replaced by 0.95 ml of ice-cold acetate buffer containing 0.25 mM isobutyl-methylxanthine. Islet cAMP content was thendeterminedwith a commercially available kit (Du Pont-New England Nuclear, Boston, MA) as described previ-ously(23).

Measurements ofATPandADP.Aftertheinitial preincubationof 60 min inacontrolmedium,batches of five islets were incubated for 60 min in 0.4 ml of medium containing the appropriate concentrations of glucose and test substances. Theincubationswerestopped by addition of 0.6 ml of ice-coldtrichloraceticacidtoafinal concentration of 5%. The tubes were then vortexed, left at room temperature for 15min,and

centrifuged for5 min inamicrofuge(Beckman Instruments, Inc., Palo Alto, CA).Afraction (400 ,A) of the supernatant was then thoroughly mixed with 1.5 mlof diethylether, and the ether phase containing

trichloraceticacidwasdiscarded.Thisstep wasrepeated three times to ensurecomplete elimination of trichloracetic acid.The extracts were thendiluted with 400

_,d

ofabuffer containing40 mM Hepes, 3 mM

MgCl2, and KOH to adjustpH at 7.75. They were then frozen at -20'Cuntilthedayofthe assay. ATP was assayed byaluminometric method(24).Analiquot (50

,gl)

ofeach samplewasmixed with 190,u ofthe abovebufferand60Mlofacommerciallyavailable ATP moni-toringreagentcontaining firefly luciferaseandluciferin (LKB-Wallac,

Turku, Finland). Theemitted lightwasmeasured inaluminometer (BiocounterM2000, Lumac,Landgraaf,TheNetherlands). Another

SO-,ul aliquot ofeachsamplewasfirst incubatedfor30 minwith 190M1 of buffercontaining 1.5 mM phosphoenol pyruvate and 2.3 U/ml pyruvatekinaseto convertADPinto ATP. ATPwasthenassayedas

describedaboveand ADPcontent wascalculatedbydifference. Appro-priate blanks, ATP, and ADP standardswere runthrough the entire procedureincludingtheextractionstep. Allmeasurementsweremade induplicates.

Measurementsofinositolphosphates.Afterisolation,theisletswere loaded with [2-3H]myo-inositol(The Radiochemical Centre, Amer-sham,UK) duringapreincubation of2 hin the presence of 15 mM glucose.After washing,batchesof20-25isletswerethenincubatedfor

60 min in 1 ml ofmediumcontaining1 mMinositol,5mMLiCl,and

appropriateconcentrationsofglucoseandtestsubstances.Attheend of theincubation,0.1 mlofthe mediumwastaken for insulin measure-mentbeforeadditiontoeach tube of 3 ml of CHCl3/CH3OH/concen-trated HCl(200:100:1, vol/vol/vol)and 100MlofEDTA( 100 mM).

Freeinositolandinositolphosphatescontained in theisletswerethen

separated by anion exchangechromatography (25),asdescribed

previ-ously(26).Inthisstudy, however,all inositolphosphateswereeluted together, byadding16 mlofamixture of 1 M ammonium formate and

0.1Mformic acidtothe columns.

Measurements of cytosolicCa2". After isolation, the isletswere culturedfor 1-2 d in RPMI 1640 mediumcontaining10% heat-inacti-vated fetal calf serum,100IU/ml ofpenicillinand100 ugg/mlof

strep-tomycin.Theconcentration ofglucosewas10mM.CytosolicCa2+was

measuredbymicrospectrofluorimetryin wholeislets loaded with fura-2(MolecularProbes, Inc., Eugene,OR).Thetissuewasexcited succes-sivelyat340 and380nmandthefluorescence emittedat510nmwas capturedbyaCCDcamera(PhotonicScienceLtd., Tunbridge Wells,

UK). The images were analyzedwith the system Magical (Applied

Imaging,Sunderland,UK). Thetechniquewasaspreviouslydescribed (21 ) except that the time interval between ratioedimageswas9 instead of 12s.

Down-regulation of protein kinase C (PKC). After isolation, the islets were cultured for 20-22 h in RPMI medium containing 5.5 mM glucose and supplemented with either 200 nM phorbol 12-myristate 13-acetate (PMA) or 200 nM of the inactive phorbol ester4a-phorbol

12,13-didecanoate (a-PDD). After washing and preincubation for 30 min in a control medium, the islets were incubated, in batches of 3, in 1 ml of medium containing appropriate concentrations of glucose and testsubstances. At the endof theincubation,analiquotof the medium wastaken for insulin assay. Insulin content of the islets was also deter-mined, for each batch, after extraction in 0.5 ml of acid ethanol.

Materials. Diazoxide was obtained from Schering-Plough Avon-dale (Rathdrum, Ireland); deoxyglucose, D-glyceraldehyde, manno-heptulose, cycloheximide, PMA, and a-PDD were obtained from Sigma Chemical Co. (St. Louis, MO); dibutyryl (db)-cAMP and a-ke-toisocaproate (KIC) were obtained from Aldrich Chemie (Steinhein, FRG); arginine vasopressin (AVP) was obtained from Peninsula Labo-ratories Inc. (Belmont, CA); forskolin was obtained from Calbiochem-Behring Corp. (San Diego, CA); ATP, ADP, phosphoenolpyruvate, andpyruvate kinase were from Boehringer (Mannheim, FRG), and all other reagents were obtained from Merck A.G. (Darmstadt, FRG).

Presentation ofresults.Allresults are presented asmeans±SEMfor theindicated number of experiments or of batches of islets (obtained in theindicated number of experiments). The statistical significance of differences between means was assessed by analysis of variance fol-lowed by a test of Dunnett when several experimental groups are com-pared with a control group, or by a test of Newman-Keuls when multi-plecomparisons are made. In a few instances, when only two groups were involved, a test oft was used. Differences were considered signifi-cant at P < 0.05.

Results

Characteristics of the effects

of

glucose on K-induced insulin

release. Perifused islets were used tocharacterizethe effects of various glucose concentrations on insulin release induced by 30 mM K in the presence of 250MM diazoxide.Asshown in

Fig. 1, control release in normal K was low in all glucose con-centrations, which attests that the normal effect of glucose is fully blocked by diazoxide. Increasing theconcentrationof K to 30 mM rapidly triggered insulin release. Both the pattern and the amplitudeof this response to high K were influenced bytheprevailingglucoseconcentration. In the absence of glu-cose, the rate of insulin release increased rapidly and markedly (by about 13-fold) beforeprogressivelyreturning to basal val-ues. In the presenceof3 or 6 mM glucose, K-induced insulin release was biphasic. The initial response was smaller than that seen in aglucose-freemedium, but was followed by a second-aryincrease,which was moresustainedin 6 mM than in 3 mM glucose (Fig. 1). When the concentration of glucose was 10 mM or higher (up to 30 mM), the pattern of insulin release was no longer clearly biphasic, but the rate ofincrease was still higher during the initial minutes than later.

When the concentration dependencyofthe effects of glu-cose wascomputedby integrating total insulin release over the 60-min period of perifusion with high K, the dose-response

curve displayed twocomponents:a first increase between 0 and 6mM, significantat 3 mM (P < 0.05 by analysis of variance and Dunnett's test), and a second increase > 6 mM (Fig. 2 A). This curve contrasts with the control, sigmoidalrelationship

between insulin release and the concentration of glucose in a perifusion medium containing4.8 mM K and nodiazoxide. Half-maximal stimulation ofinsulin releasewasproducedby 10

mM

glucose in high K and diazoxide, and by 15.5 mM

400 K 30mM K 30mM K 30mM

+

_'

200 -1i _

0 20 40 60

Time (min)

K 30mM

G 20

I I

0 20 40 60

Time (min)

K 30mM

KIC 2

I I

0 20 40

Time (min)

60

Figure 1. Effects of glucose and of KIConthe

stimulation of insulin release by 30 mM K in

mouseislets perifused withamedium contain-ing diazoxide. The experiments lasted 110

min,but only the last 10 min of the initial equilibrationperiodareshown. The

concen-tration of glucose (G 0-G 20 mM) and of KIC (2 mM) remainedconstantthroughout. Dia-zoxide (250ttM)wasalsopresentfromthe

startof the experiment. The concentration of Kwasincreased from4.8to30 mM for 60 minasindicated. Valuesarethe mean±SEM for sixor sevenexperiments.

Whentotal insulin releasewasintegratedoverthe first 10

min,

aclearinhibitory effect of 3 and 6 mM glucose (P<0.01) wasevidenced, whereas the response wasnotsignificantly af-fectedby 1 mMglucose orby concentrations of 10 mM and

higher (Fig. 2 B). Similar resultswereobtainedby computing the rate of release at4 and 6 min (not shown). The curve

.!n

a)

c

3

2

o

0 3 6 10 15 20

Glucose (mM) 30

0-10min B

characterizing the effect of glucose during the last 50 minof stimulation still showedtwocomponents(Fig. 2 C).It,

how-ever,tendedtoanhyperbolicshape for thelast10min ofstimu-lation (Fig. 2 D). During this last period, half-maximal stimula-tionwasachievedby 9.5 mMglucose, in contrastto 15 mM glucose under control conditions (low K andnodiazoxide).

Role of cytoplasmic

Ca".

The concentration of cytoplas-mic

Ca2"

in Bcellswasestimated in whole islets loaded with the Ca indicator, fura-2. At thestartoftheexperiments, when the concentration of Kwas4.8mM,

Ca3'

waslow andsimilar in 0, 6, and 20 mM glucose (55±4,65±5, and 63±5 nM,

respec-tively;n =9), because of thepresenceof diazoxide.When the concentration of Kwasraisedto30mM,abiphasic increasein

Ca?'

occurred, withaninitialpeak thatwashigher (P<0.01 ) in 0 glucose than in 6 mM glucose (Fig. 3). During steady-state stimulation with 30 mM K,

Ca3'

remained stable in 6 mM glucose, but slowly and regularly increased with time in the absence of glucose. The changes in

Ca?'

measuredinthe

pres-K30mM 400r

Glucose (mM)

Figure2. Concentration dependency oftheeffects ofglucoseon

in-sulin releaseby perifusedmouseislets stimulatedwith 30 mM K in thepresenceof250 ,uM diazoxide(e). Theexperimentsweresimilar tothoseillustratedinFig. 1.Insulinreleasewasintegratedoverthe

indicatedperiods of stimulation with30 mM K. Control experiments (o)wereperformed withamedium containing4.8 mM K,no diaz-oxideand thesameglucose concentration throughout.Insulin release

wasintegratedoverthesametime periodsasfortheexperiments with

highK.Valuesarethe mean±SEM (whenlarger than the symbol) for sixor sevenexperiments.

200

.E

IDp 0

-I. >, 100

0

I

1,I

I1

-^-

~~--GO

)r

lk

G 6mM

Smin

Figure3. Effect ofglucose (G)ontheincreaseincytoplasmicCa2"

broughtabout inislet cellsby30 mM K.The concentrationofglucose

wasconstantanddiazoxide (250 MM)waspresentthroughoutthe experiments,whereas the concentration of Kwasincreased from4.8 to30 mMasindicatedbythearrow.Thetracescorrespondtothe

meanresponse(±SEM)obtained in nine islets.

Glucose ControlofInsulinReleaseIndependent from K+ Channels 873

C

E

4-a,

C2 40

cn 2,601

GJ ix401

20(

0

0

K 30mM

f14

G 10

0

:

oL

. p-

I

ence of 20 mM glucose were similar to thosein6mMglucose

(not shown). At the peak, Ca2+ averaged (n = 9) 314±17,

236+8,

and 244±7 nM in 0, 6, and 20 mM glucose, respec-tively. After 30 min of stimulation with high K, it averaged 346±13, 199±7, and212±7 nM in 0, 6,and 20 mMglucose, respectively. When theexperiments wereperformed in a Ca-free medium, high K did not increase

Ca?'

inthe absence or presence of glucose (not shown).

Wealsoverifiedthat theeffect ofglucose oninsulinrelease

was not alteredin cultured islets loadedwith fura-2. Loading

with the Ca indicator had no effect on the insulin response to high K in the presence of diazoxide and absence ofglucose (4.0±0.1 vs. 4.2±0.2ng/isletper 60

min,

n =3).Thisalsodid notprevent6mMglucose from reducing theinitial responseby

30%andincreasing the late one by 150%.

Effects

of

various agents. Fig. 1 shows that2 mMKIC

af-fected insulin release induced by 30mM K in thepresence of

TableI.EffectsofVariousAgentsonInsulinRelease byMouseIsletsIncubatedinthePresence

of

30mM K

and250

,gM

Diazoxide

Experimental conditions

Glucose Testsubstance Insulin release mM mM percentageofcontrols

SeriesA

0 100±3(60)

20 281±11*(60)

0 Deoxyglucose20 101±8(16)

0 Galactose20 91±6(16)

0 Glyceraldehyde 10 203±12*(20)

0 Pyruvate20 101±5(14)

0 Lactate 20 102±7(12)

0 Glutamine 20 149±12*(12)

0 Leucine20 253±20*

(12)

0 Glutamine 10+leucine 10 293±22*

(16)

0 Inositol 10 97±6

(12)

0 Azide 2 51±4*

(16)

0 NH4C1 1 65±7*

(14)

0 NH4Cl5 32±3*

(14)

0 NoCaC12 14±2*(12)

Series B

20 100±3(32)

20 Mannoheptulose20 50±3*

(21)

20 Cycloheximide0.05 99±5

(12)

20 Azide 2 54+3*

(21)

20 NH4Cl1 77+4*

(12)

20 NH4C15 50±3*(17)

20 NoCaC12 4±1*(12)

Batches of 3 isletswereincubated for 1 hin 1 mlof medium con-taining30mM Kand250,gMdiazoxide.Themedium also contained the indicated concentration ofglucoseand oftestsubstance. Results areexpressedas apercentage of the average release of insulinby

con-trol islets within thesameexperiments. InseriesA,control release in 0mMglucose amountedto7.2±0.3ng/isletper h.Inseries B, control release in 20mMglucoseamountedto 19.8± 1.0ng/isletper h. Valuesaremeans±SEM for the indicated number of batches of islets, obtained inatleast three separateexperiments. * P<0.01 vs.

controls withouttestsubstanceatthesameglucoseconcentration

(Analysisof variance followedbytestofDunnett).

diazoxidein asimilarway as 6 mM glucose. Compared to the response measured in a glucose-free medium without KIC, the

initial release (0-10min) was decreased by 60% (P < 0.001 byt

test), whereas the late response ( 10-60 min) was increased morethanthreefold.

Theeffects ofother agents were tested with islets incubated

for 1 h inthe presence of 30 mM K and 250 ,M diazoxide (TableI).Deoxyglucose, galactose, pyruvate, lactate, and

inosi-tol were ineffective, whereas glyceraldehyde, and glutamine

andleucine, aloneorin combination,increasedinsulin release.

The stimulatory effect ofglucose was unaffected by

cyclo-heximide, which suggeststhatsynthesis of newproteinsis not

required. Onthe other hand, it was decreased by mannoheptu-lose andazide, whichrespectively inhibit glucose

phosphoryla-tion by glucokinase and interferewithmitochondrial

respira-tion. Azide also inhibited insulinrelease in the absence of glu-cose(Table I).

Theeffects of NH4Clwere tested because glucose raises

cy-toplasmicpHin Bcells(27), and because alkalinization of B

cells withNH4C1 increases insulinrelease undercertain

condi-tions

(28). NH4C1was ratherfoundto cause adose-dependent inhibition of K-inducedinsulinrelease in the absence and pres-enceof 20 mMglucose. Finally, omission of extracellular Ca

abolished insulin release in 0 and 20 mM glucose, as already

reported(21 ).

Role

of

cAMP. ThecAMPconcentration of islets incubated

for 1 hinamedium containing0glucose, 30 mM K, and 250

MMdiazoxideamounted to 12.0±1.0

fmol/islet

(n = 18). It was notsignificantly affectedby 3 or 6 mM glucose, and

mar-ginally increased (P<0.05)by 10and20 mMglucose(Fig. 4).

A

80-.-- ₇₀

U) _ 30_

E

-20

10F

nB

I-sa)

LA

a)

v3O

B 20

15

10

0

+Forskolin

[ z

irliri~~~~~~~~~~-- +

-Forskolin

0 3 6 10 20

Glucose (mM)

Figure4.Effectsof glucose and of forskolin (I

uMM)

oncAMP levels in andinsulin release bymouseislets stimulated by 30mM Kin the presence of 250

MM

diazoxide. Batches of 12 isletswereincubated foronehour inamediumcontainingtheindicatedconcentration of glucose withorwithout forskolin.Atthe endof theincubation, insu-linwasmeasured inthemediumand cAMPin thetissue.Valuesare themean±SEM for 18 batches of islets from six separateexperiments.

Thisweak effect ofglucosecontrasted with the strongincrease brought about by 1 ,tM forskolin: twofold in the absence of

glucose _{and fourfold in 6 mM glucose. There was no}

correla-tion between islet cAMP concentrationsand insulin release. Insulin release wassimilarly increased by6 mMglucoseandby

forskolin in the absence ofglucose,but cAMP levelswereabout 65% higher in the latter situation. Moreover, insulin release

was25% higher(P <0.01)in 20mMglucose than in 6 mM

glucose plus forskolin although cAMP levels were 4.4-fold

lower.

Giventhe opposite effects of certain glucoseconcentrations

on the earlyand late phases of insulin release inducedby 30

mM K in thepresence ofdiazoxide, the effects offorskolin werealso tested in perifused islets. Fig. 5 shows thatforskolin did not alter the time course ofrelease,but

simply

shiftedthe rateof releasetohigher values.Inparticular, it didnot decrease the initial response and increase the lateoneasdoesglucose.

Themembranepermeantdb-cAMP hadno

significant effect

in theabsence ofglucose, but mimickedthose offorskolin in 6 mMglucose (Fig. 5).

K 30mM

4001 0-* + Forskolin

+db-cAMP

2001

40 Time (min)

Figure5.Effects of forskolin(1,tM)and db-cAMP (0.5 mM)_onthe

stimulation of insulin releaseby30 mM K inmouseislets perifused withamedium containing diazoxide.Theexperimentslasted 110

min,butonlythelast 10 min of the initial equilibration periodare

shown. Theconcentration ofglucose(G 0-G 6 mM) remained con-stantthroughout. Diazoxide (250kLM)andforskolinordb-cAMP

werealsopresentfromthestartoftheexperiment. Valuesarethe mean±SEM for five experiments, exceptwith db-cAMPwherethe

valuesarethemeanfortwoexperiments.

0

E ' _E+AVP

7p

B 15 II

n 10

vl

C

L0

J +AVP

0 3 6

Glucose (mM)

II

10 20

Figure

6.Effects ofglucoseand of AVP(100nM)oninositol phos-phatelevels in and insulin releasebymouseislets stimulatedby30 mM K in thepresenceof 250AMdiazoxide. Batches of 20-25 islets wereincubated for one hour in a mediumcontainingthe indicated concentration ofglucosewithorwithoutAVP,and5 mM LiCl. At the end of theincubation,insulinwasmeasured in the medium and inositolphosphatelevels in the tissue. Valuesarethe mean±SEM for

12batches of islets from sixseparate experiments, exceptfor the

groupswith AVP where valuesarethemeanforeightbatches of islets.

Role

of phosphoinositides

and PKC.

To test

whether the

effects of

glucose

on_{insulin release involve changes} in

phos-phoinositide

turnover,theisletswerelabeledwith

[2-3H

H]myo-inositol before

being

incubatedin the presence of

high K,

diaz-oxide,

5 mM

lithium,

and variousconcentrations of

glucose.

Their content in total inositol phosphates was then deter-mined.Asshownin

Fig.

6,

inositol

phosphate

levelswerenot

affected

by

any concentration of

glucose

under the present

ex-perimental conditions, although

the sugarproducedits usual

dose-dependent

stimulation ofinsulin release. This

sharply

contrastswith the

large

increase in inositol

phosphate

levels

brought

about

by

100 nM

AVP,

which, however,didnot

mod-ify

insulin release.

Interestingly,

the effect ofAVPoninositol

phosphates

was

larger (P

<0.01)in 6 mMglucosethan inthe

glucose-free

medium.

Stimulationof PKC

by

25 nM PMA increasedtotal K-in-duced insulin release

by

50%

(P

<

0.05)

in the absence of

glucose, by

65%

(P

<

0.01)

in 3 mM

glucose,

and

by

about 300%in 6 mMglucose.Thispotentiationdid not

substantially

alter the timecourseofrelease (Fig.

7).

Toevaluate furtherapossibleroleof PKC in the stimula-tion

by glucose,

the effectsofthe sugar were tested in islets

which had beenculturedfor20-22hin the presenceofa

high

concentrationof

PMA,

aprocess

widely

usedto

down-regulate

PKC. Isletsculturedinparallelwith theinactive phorbolester

a-PDDservedascontrols. Because thepresenceof PMA in the

culture medium markedly stimulates insulinrelease theislet insulincontent wasdeterminedin each batch ofislets.Table II thus presents resultsasabsoluteamountsofinsulin releaseand

Glucose Control ofInsulin Release Independent from K+Channels 875

I-cr

0.

4,)

cn

K 30mM

IA 0.

01)

.7

CL

Ix C

-.* +PMA

600 G 6

400-2001

oL

Time (min)

Figure 7.Effects ofPMA(25 nM)onthestimulation of insulin re-leaseby30 mM K inmouseisletsperifused withamedium contain-ingdiazoxide. Theexperimentslasted 110min, butonlythelast 10 minoftheinitialequilibrationperiodareshown. Theconcentration of glucose (G0-G 6mM) remainedconstantthroughout. Diazoxide (250,uM)wasalso presentfromthestartoftheexperiment.Values arethemean±SEMfor fiveexperiments.The controlswerethesame asthose inFig.4.

as apercentageoftheinsulincontent,atthestartofthe

incuba-tion.

Experimental seriesAservedtoestablishthat

down-regula-tion of PKCwasachieved. After culture of the islets with

a-PDD, 15 mM glucose stimulated insulin release 1 -fold,

com-pared

to0glucose. This effectwasstrongly potentiated by 25 nMPMAor

by

100 nM mezerein,anactivatorofPKC

struc-turally

different from phorbol esters. In islets cultured with PMA, glucose stimulated insulin release almost to the same extent(ninefold) but this representedamuch larger

fractional

release of the hormone. On the other hand, the stimulatory effect of glucosewas nolonger potentiated byPMA or

meze-rein (Table II).

Whenislets cultured for20hwitha-PDDwere incubated

in thepresence of 30mM Kand250

,M

diazoxide,glucose

produced its normal, concentration-dependent increase in in-sulin release, and 25 nM- PMA strongly amplified the effect of glucose. Down-regulation of PKC by culture withPMAdidnot prevent glucose from increasing insulin release under these

conditions, but abolished the acute amplification of this effect by 25 nM PMA. The absolute changesin insulin release in-duced by glucose weresmaller than in islets treated with a-PDDbut the fractional changes were larger (TableII).

Role ofadenine nucleotides. IsletATPandADPlevelswere

measuredafter 60min of incubation inamedium containing 30 mM K, 250MM diazoxide, and various concentrations of glucose. ATP levels increased when the concentration of

glu-coseincreased from 0to 10-20 mM, whereasADPlevelsdid notsignificantly change (Fig. 8). This resulted in a concentra-tion-dependentincrease inthe ATP/ADP ratio, that was

dou-bledby20-30 mMglucose. Half-maximal increase occurred at

11 mMglucose. An excellentcorrelation was found between the ATP/ ADP ratiomeasuredinislets attheend ofthe 60min

ofincubation, and the amount of insulinreleasedduring the last 10min of the one-hour perifusion with high K, diazoxide and various concentrations of glucose (Fig. 9 A).

The influenceofother agents on ATP and ADP levels was

Table11.

Effects

ofDown-Regulation

of

PKConInsulinRelease

by

IncubatedMouseIslets

Islets cultured witha-PDD Islets cultured with PMA

Insulin Insulin

Incubation conditions content Insulinrelease content Insulinrelease

ng/islet ng/isletper h percentageofcontent ng/islet ng/isletper h percentageofcontent

SeriesA:4.8mM K

Glucose 0 235±12 0.39±0.05 0.17±0.02 60±5 0.76±0.11 1.31±0.21 Glucose 15mM 248±8

4.40±0.46*

1.79±0.20* 59±3 6.93±0.53*

11.9±1.10*

G 15mM+PMA25nM 227±8 27.9±2.4t 12.5±1.26$ 70±7 7.29±0.46* 11.0±0.92*

G 15 mM+Mezerein 100 nM 227±8 16.2±0.8* 7.17±0.32* 58±4 6.74±0.72* 12.0±1.35*

Series B: 30 mM K+diazoxide

Glucose 0 243±7 15.7±0.9 6.53±0.35 73±4 5.11±0.30 7.17±0.43 Glucose10 mM 249±14 19.0±0.9* 7.81±0.36* 73±4 7.06±0.53* 10.0±0.76* G 10mM+PMA25 nM 236±10 35.0±1.5* 14.9±0.42* 67±3 6.68±0.55* 9.94±0.64* Glucose 20 mM 243±7 24.4±1.1* 10.1±0.44* 66±3 8.54±0.62* 13.2±1.00*

also studied (Table III). In the absence ofglucose, deoxyglu-coseslightly decreased ATPlevels;galactose, pyruvate,and lac-tate were withouteffect;azidedecreased theATP/ADPratio;

glyceraldehyde and the combination glutamine plus leucine

increased ATP levels and the ATP/ADP ratio. On the other hand, theeffects of 20mMglucosewereinhibited by

manno-heptulose, azide, and NH4Cl. A good correlation was found

between the effectsof these various agentsontheATP/ADP ratio intheislets and oninsulinrelease(Fig. 9

B).

A notable

exceptionwastheomission ofextracellular Ca which abolished insulin release in 0 and 20 mM glucose (Table II) without

affectingtheATP/ADPratio in islet cells(TableIII).

Discussion

Thepresentstudy extendsourrecentdemonstration (21) that

glucoseisstill abletoregulate insulin release when itnolonger

can control K+-ATP channel activity and,hence, the mem-brane potentialof B cells. It alsosuggeststhat thisproperty of glucosemaybe linkedtochanges in theenergystateof Bcells,

while excluding several other possible mechanisms.

Inourprevious experiments performedwithincubated

is-letsweobserved that theconcentrationdependencyof the

ef-fect of glucoseoninsulin releasein thepresenceofhighKand diazoxidedisplayedapeculiarbiphasicshape,withanincrease alreadyat2 mM glucose.Asimilarcurve wasagainobtained

here, when perifused islets werestimulated for 60 min with

high K in different glucose concentrations. These dynamic ex-periments alsorevealedan unsuspected complexity of the

ef-fect of glucose on the time-course of release. Whereas the

steady-state responseprogressively increased with theglucose concentration, therapidresponsewasparadoxicallydecreased

by low glucoseconcentrations. Thedose-responsecurveisthus

thesumofaninhibitorycomponentdetected at 3 and 6 mM

glucose,andastimulatorycomponentobservedatallglucose concentrations. Itispossible,however,that theinhibitory

com-1.

4-0

E M

o

r-M

01-10 20

Glucose (mM)

Figure 8. Effects of various concentrations of glucoseonATPand ADPlevelsandontheATP/ADPratio inmouseislets stimulated by30 mM K in thepresenceof 250AMdiazoxide. Batchesof five isletswereincubated for I h in thepresenceofthe indicated

concen-tration ofglucose. Valuesarethe mean±SEM for 25 batchesof islets from fiveseparateexperiments.

20

15 a-10 c0 ,w 10

.c 75

B

e

k

+g

id

r2=0.92

4f

2

ATP /ADP

3

Figure9.Correlations between insulin release and the ATP/ADP

ra-tio inmouseislets stimulated by 30 mM K in thepresenceof250

AM

diazoxide. (A) Experiments performed withvarious

concentra-tions ofglucose. Valuesaretaken from Fig. 2 D for insulin release

andFig. 8 for the ATP/ADP ratio. (B) Experiments performed in thepresenceof variousagents.Valuesaretakenfrom Table I for in-sulin release and Table III for the ATP/ADP ratio. The conditions

are:a,G 0; b, G 0 and 20 mM deoxyglucose;c,G0and20 mM

ga-lactose; d, G 0 and 10 mM glyceraldehyde;e,G0and 10 mM

gluta-mine and leucine;f, G 0 and 2 mM azide;g,_{G 0 and 20 mM lactate;}

h,G 0 and 20 mMpyruvate;i, G 20mM;j,G 20 mM and 20 mM mannoheptulose; k, G 20 mMand2 mMazide; 1, G 20 mM and 5 mMNH4Cl. Valuesareshownasthemean±SEM.

ponentdoesnotdisappearat 10mMglucoseandabove, butis

masked by the fastdevelopment ofthe stimulatory

compo-nent. Anacute inhibitoryeffect of glucoseon insulin release

has been observedpreviouslywith islets perifused with a me-dium containing diazoxide and a normal K concentration

( 19).It has been ascribedtothe ability of thesugarto promote

Ca2+ sequestrationin the absence of

Ca2"

influx.Our

observa-tion that 6 mM glucose attenuated the rapidincreasein

cyto-plasmic

Ca?'

broughtabout by high K isconsonantwith this hypothesis which,admittedly,requiresmoredirect testing.

Elimination of theinhibitorycomponentmadethe

dose-response curve morethoughnotcompletely hyperbolic, thus

accentuating thedifferenceswith the sigmoidal relationship be-tweenthe glucose concentration and insulin release under

con-trol conditions. A significant effect ofglucosewasobservedat3 mM and thehalf-maximalresponseat9-10mM,ascompared

withmorethan 6 and 15mMincontrolislets.

Whenisletsareincubated for 60_min,only thestimulatory

componentof the glucose actioncanbereliably studied. Itwas

inhibited by mannoheptulose, which inhibits glucose

phos-Glucose ControlofInsulin ReleaseIndependent from K+Channels 877

6

C

.X-

..-EU

C>

4

-LA

c

-_do

A

+

r2- 0.96

Table III. Effects ofVarious Agents on A TPandADPLevelsin MouseIslets IncubatedinthePresenceof3OmM Kand 250

_,4M

Diazoxide

Experimentalconditions

Glucose Test substance ATP ADP ATP/ADP

mM mM pmol/islet pmol/islet

SeriesA

0 7.1±0.2 4.4±0.1 1.61±0.03

0 Deoxyglucose 20 6.3±0.2* 4.2±0.2 1.52±0.05

0 Galactose 20 7.3±0.3 4.3±0.1 1.70±0.04

O Pyruvate 20 7.5±0.2 4.4±0.2 1.76±0.10

O Lactate20 7.2±0.3 4.5±0.2 1.64±0.09

0 Glyceraldehyde 10 9.5±0.5$ 4.2±0.1 2.23±0.08*

0 Glutamine 10+leucine 10 13.6±0.5t 4.0±0.2 3.47±0.19t

0 Azide2 6.3±0.3* 4.6±0.2 1.39±0.04*

0 NoCaCI2 8.3±0.4* 5.1±0.3* 1.65±0.03

SeriesB

20 12.6±0.3 4.0±0.1 3.24±0.10

20 Mannoheptulose 20 7.7±0.3* 4.0±0.2 1.95±0.08t

20 Azide 2 7.5±0.4t 5.0±0.2t 1.47±0.03t

20 NH4CI 5 9.6±0.6t 4.5±0.3 2.14±0.09*

20 NoCaC12 13.0±0.4 4.4±0.1 2.96±0.13

Batchesof 5 isletswereincubated for 1 h in 0.4 ml of mediumcontaining30 mM K and 250,gMdiazoxide. The medium also contained the indicatedconcentrationof glucose andtestsubstance. Valuesaremeans±SEMfor 55(seriesA)or40(seriesB) batchesofcontrolislets without testsubstance andof15batchesof islets witheachtestsubstance. * P<0.05, tP<0.01 vscontrols withoutsubstance butatthesameglucose concentration(Analysisof variancefollowed bytestofDunnett).

phorylation (29, 30) and by azide, whichpoisons

mitochon-dria (31);it wasreproduced by glyceraldehyde, leucine with

glutamine,

and KIC, whichare all well metabolized

(32); it

was notmimicked by 2-deoxyglucose, galactose,pyruvate, or

lactate,

which

arenot or

only

poorly

metabolized in

islet

cells (29, 30).It

is

thusclear that the

stimulatory

effect

of glucose

is linkedto anincrease inBcell

metabolism.

Omission of extracellular Ca

prevented

the rise in

Ca?'

and the

stimulation

of insulin release

produced by high

Kboth in theabsence orpresence

of glucose.

The

stimulatory

effect of glucoseon

insulin

release is thus

Ca2' dependent. It, however,

doesnotresult froma

further

rise in

Ca?'.

Steady-state

Ca?'

(in

thepresence

of

high K)

was

similar

in20 and6 mM

glu-cose,whereastherate

of insulin

releasewastwice

higher

in the

former situation. Moreover,

Ca?'

was

higher

in0

glucose

than in 6or20mM

glucose,

but

insulin

releasewasmuch smaller. Therewasalsono

correlation

between the timecourse

of

the

changes

in

Ca?'

and

insulin

release.Intheabsence of

glucose,

Ca?'

slowly

increased with

time,

whereas the rate ofinsulin release decreased. It is thus

obvious

that

glucose, somehow,

modulates the

effectiveness of

Ca?'

on the

secretory

ma-chinery.

ThemodestincreaseinisletcAMPconcentrationthat

glu-cosecausesisnotthe

triggering signal

of insulinrelease under

physiological conditions (4, 33,

34),

but cAMP isknownto

sensitizethe

releasing

systemtoCa2 .

Participation

ofPKAin

thephenomenon studiedherewasthus

plausible. However,

no

correlation wasfound between islet cAMPlevelsand insulin

release in the presence of various concentrations of glucose, with or without forskolin. We, therefore, conclude that the membranepotential-independent effect of glucose on insulin

release is notsecondaryto anincreasein cAMPlevels,butwe cannotexcludethe

possibility

thatactivation ofPKA

by

basal cAMPlevels

participates

inthe effects of

glucose.

Nutrient insulin secretagogues slightly stimulate inositol

phosphate

formation

ininsulin-secreting cells (35, 36). This

effect is,

atleastpartly, duetoactivation of phospholipaseC by

theinflux

of

Ca2"

that thesecretagogues cause. Under our

ex-perimental conditions,

glucose

did

not affect inositol

phos-phate levels while increasing insulinrelease. Conversely, strong

stimulation of phosphoinositidebreakdown by AVP (26) did not augment insulin release. We, therefore, suggest that the membranepotential-independent effectof glucose is not due to anacceleration of

phosphoinositide

turnover.

The roleof PKC in glucose-induced insulin releaseisstill

controversial,

but the balancenegates(37-40) rather than sup-ports(41

) its

importance.

However, because PKC is believed

to

sensitize

the

releasing machinery

to

Ca2

,its involvement in the

effects of

glucose

studied

herewaspossible. This doesnot seemtobe thecase,forPMA wasonly poorly activeon K-in-duced insulin release unless glucose was present, and never gave toinsulin release thepatternproduced byanincrease in glucose concentration. This clearly showsthat aneffectof glu-coseother thananactivationof PKC playsapreeminent role, but doesnottotally excludes the

participation of

PKC. Totest

furtherthis

possibility,

islets were pretreated overnight with PMA, atechnique widelyused todown-regulate PKC(37-40).

Theinability ofPMA andof mezereintoincrease insulin re-leaseacutely aftersuch treatment attests thatdown-regulation of PKCwasachieved.

This,

however, didnotpreventglucose fromstimulating insulin releaseunder controlconditions,or in the presenceofhighKanddiazoxide.We,therefore, conclude

that neither the effect ofglucose under physiological

condi-tions,

nor its membrane

potential-independent effect,

more

specifically

studied here, is critically dependenton PKC. GlucoseincreasesATPlevels (42-45) and the ATP/ADP

ratio(44, 45) inisolated isletsstudiedunderphysiological

modulating the activity ofK+-ATPchannels, the changes in

cytoplasmicATP/ADP ratio constitute a major link between glucose metabolism and the control of B cell membrane poten-tial. They, however, neither prove this hypothesis nor exclude theintervention of other coupling factors. Thus, a stimulation of insulin release by the perfused rat pancreas, without

detect-able changesinthe ATP/ADP ratio in islet cells, was recently

observedupon acute stimulation withglucosein thepresence

ofamino acids(46). Inthepresent experiments (made under

nonphysiological conditions) glucose was found to cause a

dose-dependent increase in theATP/ADP ratio in islet cells,

and thisincrease correlatedvery well with the changes in

insu-linreleasebrought about bythe sugar. Moreover, anexcellent

correlation was also found between insulin release in the pres-enceof variousagents(activeones, alone and with inhibitors, or inactive ones)and theATP/ADP ratio in the tissue. The

only exceptionwasthe absence of extracellularCa,which

to-tally blocked release without affectingtheATP/ADPratio.

Underourexperimental conditions, K+-ATPchannels are nolongeroperativeastheyaremaintainedwidelyopenbythe

high concentration of diazoxide. We, therefore, suggestthat thereexists anotherstepof stimulus-secretion couplingwhere

adenine nucleotidesplayanimportant role.The fact that

acti-vation ofPKA by cAMP orof PKCby PMA waspoorly effec-tivein theabsenceof glucose in face of the

high

concentration of

cytosolic

Ca?+

indicates that thisstepisrather distal in the

chainofeventsleadingtoexocytosisof insulingranules. Two

possibilities

arethatATPissimply required tosustainan

en-ergy-dependent phenomenon (e.g.,the movementofgranules)

orthat aminimum concentrationof ATP is necessary for ade-quatephosphorylation of critical targets.Bothare

compatible

withtheobservationthatforskolinand PMA becomeeffective in thepresenceofglucose,butdo notreadily explain why the correlation between insulin release and the ATP/ADP ratio

extends overthe whole range of glucose concentrations. An

alternative is that another,yet

unidentified,

substance, the

con-centration of which varies inparallel with the ATP/ADP ratio, is thetruecoupling factor.

Inconclusion, the mechanism by which glucosecancontrol insulin releaseindependently from changes inK+-ATP

chan-nelactivity and changes inBcell membranepotential requires stimulation ofBcellmetabolism, butdoes notinvolve changes incytoplasmic

Ca'+,

cAMPlevels, inositol phosphate levels,or

PKCactivity. On the other handchangesin theenergy stateof

Bcellsmaybeinvolved.Thepossibility is raisedthat the ATP/ ADPratio(or arelatedmessenger)controls insulinrelease by

regulating boththe activity

of

K+-ATP channelsand a much

laterstepofstimulus-secretion coupling. Atthe latter level, it

appears to amplify the effectiveness of the rise in

Ca?'

that

followstheprimary action ofglucoseonK+-ATPchannels and membranepotential. Animportantchallenge will be to

estab-lishtherelative

contribution

of defectsat both levels of control

in the abnormalities of glucose-induced insulin release in B cellsofnon-insulin-dependent diabetic patients.

Acknowledgments

Wearegratefulto M.Gerardfor skillful assistanceand to M. Nenquin for editorial help. We thank Professor G.R.Rousseaufor allowingus tousetheLuminometer.

Thiswork wassupported bygrant3.4607.90from the Fonds de la RechercheScientifiqueMWdicale,Brussels, andby grantSPPS-AC 89/ 95-135fromtheMinistryofScientific Policy,Brussels. MarekGembal

is a postdoctoral researchfellow on leavefromTheMedicalAcademy

ofGdansk; Philippe Detimary is Aspirant andJean-ClaudeHenquin is

Directeur de Recherches of the FondsNationaldelaRecherche

Scien-tifique, Brussels; Patrick Gilon is Charge de Recherchesofthe

Univer-siteCatholique de Louvain.

References

1. Matschinsky, F. M., and F. J. Bedoya. 1989. Metabolism of pancreatic islets andregulation of insulin and glucagon secretion. In Endocrinology, Volume 2. L. J. Degroot, editor. W. B. Saunders Co., Philadelphia. 1290-1303.

2. Rasmussen, H., K. C. Zawalich, S. Ganesan, R. Calle, and W. S. Zawalich. 1990. Physiology and pathophysiology of insulin secretion. Diabetes Care. 13:655-666.

3. Cook, D. L., and G. J. Taborsky. 1990. B-cell function and insulin secre-tion. In Diabetes Mellitus: Theory and Practice, 4th edisecre-tion. H. Rifkin and D. Porte, editors. Elsevier, Amsterdam. 89-103.

4. Prentki, M., and F. M. Matschinsky. 1987. Ca2 , cAMP, and phospholipid-derived messengers in coupling mechanisms of insulin secretion. Physiol. Rev. 67:1185-1248.

5. Henquin,J. C., and H. P. Meissner. 1984. Significance of ionic fluxes and changes in membrane potential for stimulus-secretion coupling in pancreatic B-cells. Experientia(Basel).40:1043-1052.

6. Wollheim, C. B., and T. J. Biden. 1986. Signal transduction in insulin secretion: comparison between fuel stimuli and receptor agonists. Ann. N. Y. Acad. Sci. 488:317-333.

7. Wolf, B. A., J. R. Colca, J. Turk, J.Florholmen,and M. L. McDaniel. 1988. Regulation of Ca homeostasis by islet endoplasmic reticulum and its role in insulin secretion. Am. J.Physiol. 254:E121-E136.

8. Draznin, B. 1988. Intracellular calcium, insulin secretion, and action. Am. J. Med. 85:44-58.

9. Atwater,I., P. Carroll, and M. X. Li. 1989. Electrophysiology of the pancre-atic B-cell. In Insulin Secretion. B. Draznin, S. Melmed, D. Leroith, editors. Alan R.Liss, Inc., New York. 49-68.

10. Cook, D. L., L. S. Satin, M. L. J. Ashford, and C. N. Hales. 1988. ATP-sensitive K' channels in pancreaticfl-cells:spare-channel hypothesis. Diabetes. 37:495-498.

11. Ashcroft, F. M., and P. Rorsman. 1989. Electrophysiology of the pancre-atic 3-cell.Prog. Biophys.Mol. Biol. 54:87-143.

12. Ashford, M. L. J. 1990. Potassium channels and modulation of insulin secretion. In Potassium Channels: Structure, Classification, Function and Thera-peutic Potential. N. S. Cook, editor. Ellis Harwood Ltd., Chichester, UK. 300-325.

13. Rajan, A. S., L. Aguilar-Bryan, D. A. Nelson, G. C. Yaney, W. H. Hsu, D. L. Kunze, and A. E. Boyd. 1990. Ion channels and insulin secretion. Diabetes Care. 13:340-363.

14. Henquin, J. C., A. Debuyser, G. Drews, and T. D. Plant. 1992. Regulation of K' permeability and membrane potential in insulin-secreting cells. In Nu-trient Regulation of Insulin Secretion. P. R. Flatt, editor. Portland Press, London.

173-191.

15. Trube, G., P. Rorsman, and T. Ohno-Shosaku. 1986. Opposite effects of tolbutamide and diazoxide on the ATP-dependent K' channel in mouse pancre-aticj3-cells.

_Pfllugers

Arch. Eur. J. Pharmacol. 407:493-499.

16. Gillis, K. D., W. M. Gee, A. Hammoud, M. L. McDaniel, L. C. Falke, and S. Misler. 1989. Effects of sulfonamides on a metabolite-regulatedATP,-sensitive K+ channel in rat pancreatic B-cells. Am.J. Physiol.257:C 119-Cl127.

17. Henquin, J. C., and H. P. Meissner. 1982. Opposite effects of tolbutamide and diazoxide on 86Rb+fluxes and membrane potential in pancreatic B-cells. Biochem.Pharmacol. 31:1407-1415.

18.Arkhammar, P., T. Nilsson, P. Rorsman, and P.O.Berggren. 1987. Inhibi-tion of ATP-regulated K' channels precedes depolarizaInhibi-tion-induced increase in cytoplasmic free Ca2" concentration in pancreatic (3-cells. J. Biol. Chem. 262:5448-5454.

19. Bergsten, P., E. Gylfe,N.Wesslen, and B. Hellman. 1988. Diazoxide unmasks glucose inhibition of insulin release by counteracting entry ofCa"2. Am. J.Physiol. 255:E422-E427.

20. Ashcroft, S. J. H., C. J. Hedeskov, and P. J. Randle. 1970. Glucose metabo-lism in mouse pancreatic islets.Biochem.J. 118:143-154.

21. Gembal, M., P. Gilon, and J. C. Henquin. 1992. Evidence that glucose can control insulin release independently from its action on ATP-sensitive K' chan-nels in mouse B cells.J. Clin. Invest. 89:1288-1295.

22.Henquin, J. C. 1978. D-Glucose inhibits potassium efflux from pancreatic islet cells. Nature(Lond.).271:271-273.

23.Henquin, J. C. 198 1. Effects of trifluoperazine andpimozideon stimulus-secretion coupling in pancreatic B-cells: suggestion for a role of calmodulin? Biochem. J. 196:771-780.

24. Hampp, R. 1986. Luminometric Method. In Methods of Enzymatic Anal-ysis.H. U.Bergmeyer,editor.VCHVerlagsgeselschaft, Weinheim,FRG. 370-379.

25.Berridge,M.J.,R. M.C. Dawson, C.P.Downes,J. P.Heslop,andR. F.

Irvine. 1983. Changes in the levelsofinositolphosphatesafter agonist-dependent hydrolysis of membrane phosphoinositides. Biochem. J. 212:473-482.

26. Gao, Z. Y.,G. Drews, M. Nenquin, T. D. Plant, and J. C. Henquin. 1990. Mechanismsofthestimulation of insulinrelease byarginine-vasopressinin nor-mal mouseislets. J.Biol.Chem.265:15724-15730.

27.Best,L. 1992.IntracellularpH and B-cell function. In Nutrient Regula-tion of insulin secreRegula-tion.P. R.Flatt,editor. PortlandPress, London. 157-171.

28.Lindstrom,P., and J.Sehlin.1986.Effectofintracellularalkalinizationon pancreaticislet calcium uptake and insulin secretion.Biochem. J.239:199-204. 29.Zawalich, W.S. 1979. Intermediary metabolismandinsulin secretion from isolated ratisletsofLangerhans.Diabetes.28:252-260.

30.Ashcroft, S.J. H. 1980. Glucoreceptormechanisms and the controlof insulin release andbiosynthesis. Diabetologia. 18:5-15.

31. Stryer, L.1975. Biochemistry.Freeman &Co.,SanFrancisco. 1-877. 32.Malaisse, W. J. 1986.Branched-chain aminoand ketoacidmetabolism in pancreaticislets.Adv.EnzymeRegul. 25:203-217.

33.Sharp, G. W.G. 1979.Theadenylatecyclase-cyclicAMP system in islets ofLangerhans anditsrole in the controlofinsulin release.Diabetologia. 16:287-296.

34.Henquin,J. C. 1985. TheinterplaybetweencyclicAMPandions inthe stimulus-secretion couplinginpancreaticB-cells.Arch. Int.Physiol. Biochim. 93:37-48.

35. Biden, T.J., andC.B.Wollheim. 1989.Generation, metabolismand function of inositol phosphates during nutrient-andneurotransmitter-induced insulinsecretion.InInositolLipids in CellSignalling.R. H.Michell,A.H. Drum-mond, C. P. Downes,editors. Academic Press,Ltd.,London.405-425.

36. Morgan, N.G.,and W. Montague. 1992.Phospholipidsandinsulin secre-tion. In NutrientRegulationof insulinsecretion. P. R.Flatt,editor.Portland Press, London.125-155.

37. Metz,S.A. 1988.Isprotein kinaseCrequired forphysiologic insulin release?Diabetes.37:3-7.

38.Wolf,B. A., R. A. Easom, M. L.McDaniel,and J. Turk. 1990. Diacylgly-cerolsynthesisde novofromglucose bypancreaticisletsisolatedfrom rats and humans.J. Clin.Invest.85:482490.

39.Wollheim, C. B., and R. Regazzi. 1990. Protein kinase C in insulin releas-ing cells:putativerole in stimulus secretion coupling. FEBS (Fed. Eur. Biochem.

Soc)Lett.268:376-380.

40.Jones, P.M., S. J. Persaud, and S. L. Howell. 1991.Proteinkinase C and theregulation of insulinsecretionfrompancreatic B cells. J. Mol.Endocrinol. 6: 121-127.

41.Ganesan, S., R. Calle, K. Zawalich, J.I.Smallwood, W. S. Zawalich, and H.Rasmussen. 1990.Glucose-induced translocation of proteinkinase C in rat pancreatic islets.Proc.Natl.Acad. Sci.USA.87:9893-9897.

42. Hellman, B., L. A.Idahl, and A. Danielsson. 1969. Adenosine triphos-phate levelsofmammalianpancreaticBcellsafterstimulation with glucose and hypoglycemic sulfonylureas. Diabetes. 18:509-516.

43.Ashcroft,S. J. H., L. C. C. Weerasinghe, and P. J. Randle. 1973. Interrela-tionship of isletmetabolism, adenosine triphosphatecontentand insulin release. Biochem. J. 132:223-231.

44.Malaisse,W.J., and A. Sener. 1987.Glucose-inducedchangesin cytosolic ATP contentinpancreatic islets. Biochim. Biophys.Acta.927:190-195.

45.Meglasson, M. D., J. Nelson, D. Nelson, and M.Erecinska.1989. Bioener-geticresponseof pancreaticisletstostimulation by fuelmolecules.Metab. Clin. Exp.38:1188-1195.

46.Ghosh,A., P.Ronner,E.Cheong,P. Khalid,and F. M.Matschinsky. 1991. The role of ATP and free ADP in metabolic coupling during fuel-stimu-latedinsulinreleasefromislet13-cellsin theisolatedperfusedratpancreas. J.Biol. Chem. 266:22887-22892.