Rapid and reversible secretion changes during uncoupling of rat insulin producing cells

Rapid and reversible secretion changes during

uncoupling of rat insulin-producing cells.

P Meda, … , C Wollheim, L Orci

J Clin Invest.

1990;

86(3)

:759-768.

https://doi.org/10.1172/JCI114772

.

To determine whether insulin secretion is affected by a blockage of gap junctions between

B cells, we have studied the secretion of rat pancreatic islets of Langerhans, primary

dispersed islet cells, and cells of the RINm5F line, during short-term exposure to heptanol.

Within minutes, this alkanol blocked gap junctions between the B cells of intact islets and

abolished their normal secretory response to glucose. These two changes were rapidly and

fully reversible after return of the islets to control medium. We further found that heptanol

had no significant effect on the glucose-stimulated secretion of single B cells but inhibited

that of B cell pairs. In the clone of RINm5F cells, whose junctional coupling and

D-glyceraldehyde-induced stimulation of insulin release by aggregated cells were also

inhibited by heptanol, this alkanol did not perturb intracellular pH and Ca2+ and the most

distal steps of the secretion pathway. In summary, a gap junction blocker affected the

secretion of insulin-producing cells by a mechanism which is dependent on cell contact and

is not associated with detectable pleiotropic perturbations of the cell secretory machinery.

The data provide evidence for the involvement of junctional coupling in the control of insulin

secretion.

Research Article

Rapid and Reversible Secretion Changes

during Uncoupling of

Rat Insulin-producing Cells

P. Meda, D. Bosco, M. Chanson, E. Giordano, L.Vallar,* C.Wollheim,*andL. Orci

DepartmentofMorphologyand *Institute ofClinical Biochemistry, UniversityofGeneva MedicalSchool, CH-1211 Geneva,Switzerland

Abstract

Todetermine whether insulin secretion is affected by a block-age of gap junctions between B cells,wehave studied the se-cretion of rat pancreatic islets of Langerhans, primary dis-persed islet cells, and cells of the RINm5F line, during short-term exposure to heptanol. Within minutes, this alkanol blocked

gap junctions

between the B cells of intact islets and abolished their normal secretory response to glucose. These two changes were rapidly and fully reversible after return of

the

islets

to

control medium.

We further

found

that heptanol

had no

significant

effect on the glucose-stimulated secretion of

single

B cellsbut inhibited that of B cell pairs. In the clone of

RINm5F cells, whose

junctional

coupling and D-glyceralde-hyde-induced stimulation of insulin release by aggregated cells

werealso

inhibited

by heptanol, this alkanol did not perturb

intracellular pH and Ca2+ and the most distal steps of the secretion pathway. In summary, a gap junction blocker af-fected the

secretion of insulin-producing

cellsby a

mechanism

which is dependent on cell contact and is not associated with

detectable pleiotropic perturbations

of the cell secretory

ma-chinery.

The data

provide evidence for

the involvement of

junctional coupling

in the

control of

insulinsecretion.

(J. Clin.

Invest.

1990. 86:759-768.)

Key words: B cells*calcium*gap

junctions,, pancreatic islets

* reverse

hemolytic

plaque assay

Introduction

The B

cells of the endocrine

pancreasare

connected

by

gap

junctions (1),

the membrane

specializations

thought to

pro-vide

channels for direct cell-to-cell communications, i.e., for

the

cytoplasm-to-cytoplasm

exchange of

ions and molecules

of

< -900 daltons

(2, 3).

Two lines ofevidence indicatethat

these

communications,

aneventalso referredtoas

junctional

coupling,

may

participate

in the control of insulinsecretion.

First,

gap

junctions

and

coupling

increase

between B cells

dur-ing sustained stimulation of insulin

secretion

(4, 5).

Secondly,

isolated

(uncoupled)

B

cells show

impaired

secretion

that

is

corrected,

atleast

partially,

after

reaggregation

and

reestablish-ment

of

gap

junctions

(4, 6).

As yet,

however,

the critical evidence that

changes

in

insu-lin

secretion

occur

during

a

rapid

and reversible modulation of

gap

junction permeability

hasnotbeen

provided.

Toaddress

Address reprintrequeststo Dr. Meda, Department ofMorphology, Universityof Geneva, C.M.U., 1,rueMichelServet,CH- 1211Geneva 4, Switzerland.

Receivedfor publication22February1988andin

revisedform

20

February1990.

this

question,

wehave studied now the secretion of single

(un-coupled)

and clustered

insulin-producing

cells during

short-term

incubations in the

presence

of

a

long-chain

alcohol that

blocks junctional communications

(7).

Methods

Preparation of islets ofLangerhans andinsulin-producingcells. Intact islets of Langerhans were isolated by enzymatic digestion and Ficoll purification fromthesplenic pancreas ofadult(250-350 g body wt)

maleSprague-Dawley rats (8).

Toprepare primary cells, isolated islets were first incubated for 15 min atroom temperature in a

Ca2'-free

Krebs-Ringer buffer bicar-bonate(KRB) and then dispersed by repeatedaspirationthrougha

series ofsyringeneedles,asdescribed in Salomon andMeda(8). Aftera

3-min incubation inaspinner flask containinga

Ca2+-free

medium supplemented with0.1% trypsin (1:250;DifcoLaboratories,Inc., De-troit, MI), the cellswere centrifuged in sterile RPMI1640medium (Gibco Laboratories, Grand Island, NY) supplemented with 10% fetal calf serum,platedonbacteriological culture dishes,andkept overnight

at

37'C

inaair/CO2incubator(8).

RINcells (9) were obtained by mild trypsinization ofcultures ofthe

m5Fclone(10)thatwerepassagedonceevery5d and also subcultured

on RPMI1640 mediumsupplemented with 10% fetal calfserum.

Dyecouplingstudies. JunctionalcouplingofprimaryBcells was determined byinjectingLucifer Yellow into individual cells located deeply within the isolated islets (I 1, 12).Injectionswereperformedon

theheated

(37'C)

stage ofaUEMmicroscope (Carl Zeiss, Inc., Ober-kochen, FRG) using glass electrodes (- 150 MQ) thatwerefilled witha 4%solution of LuciferyellowinI Mlithiumchloride,bufferedtopH 7.2with 10 mMHepes. Thetracerwasinjected iontophoretically and undercontinuouselectrophysiologicalcontrol,asreported(13,14).At

theendof eachinjection,theisletswerephotographedandthenfixed in4%paraformaldehydeinphosphate-bufferedsaline for subsequent embedding in Epon and serial sectioning

(1

1, 12). Evaluation of

cou-plingwasrestrictedto Bcells thatwereidentified by immunostaining withanantiinsulinserumand/or by their location in the deep regions of the islets(I1, 12). Couplingextent wasevaluatedby measuring the

areaofLucifer yellow diffusiononphotographsof the intact islets

(1

1, 13) and by scoring the number ofB cellslabeled by thetracer on

sections that were cut serially throughout the microinjected islets (11, 12).

JunctionalcouplingofRINm5Fcells (passages 107-110)was de-termined in a similar mannerby injecting individual cellswithin monolayer clusters of 3-4-d-old cultures.Fortheseexperiments,2.5

X

104

cells per mlwereplatedinitially.Thenumberof cells labeledby

Lucifer yellowwasdeterminedonphotographsof the intactclusters,

whichweretakenimmediately after eachinjection.

Insulin secretion studies. For measurement of insulin secretion,

batches of 10freshlyisolated isletswerepreincubated 15minat37'C

inKRBcontaining12.5mMHepes, 0.5% bovineserumalbumin,and 2.8mMglucose(control KRB) and, then,incubatedin1mlof control KRB,supplementedornotwith the different secretagogues

(16.7

mM glucose and 1 mM3-isobutyl-1-methyl-xanthine

[IBMXJ1)

and

un-coupling agent (0.5-3.5 mM heptanol) to be tested. Heptanol was

1. Abbreviations usedin this paper: IBMX, 3-isobutyl-

1-methyl-xanthine J.Clin.Invest.

©TheAmerican SocietyforClinical Investigation,Inc.

0021-9738/90/09/0759/10

$2.00

added directly to the medium in which it was mixed by a 3-min sonication. At the end of the 15-min incubation at 370C, aliquots of the medium were removed and rapidly frozen for measurement of secreted insulin. The islets were then thoroughly rinsed in control KRB,extracted for 24 h at 40C in 1 ml of acid-ethanol and the extract wasstored frozen to assess insulin content. To assess the reversibility of theheptanol effects, the islets were first incubated in the presence of the alcohol as just described, then rinsed repeatedly, and finally incubated again in a test medium devoid of heptanol for 15 min. Immunoreac-tiveinsulin of media and extracts was measured using a radioimmuno-assay with a charcoal separation step. Values of secreted insulin were expressed bothasabsolute insulin concentrations and as percentage of theinitialinsulin content of the B cells (6, 1 1).

The insulin secretion of primary dispersed islet cells was studied using areverse hemolytic plaque assay. To this end, the cells were mixed with 4% (vol/vol) packed sheep red blood cells coated with proteinAandintroduced with control KRB into poly-L-lysine-coated glasschambers, as described (8). After cell attachment, each chamber wasthoroughly rinsed and then filled with control KRB supplemented with aheat-inactivated polyclonal antibodyagainstinsulin (Dr. P. H. Wright, Indiana University School of Medicine, Indianapolis, IN) di-luted 1:50, and containing either 2.8 or 16.7 mM glucose with or without 0.5 mMheptanol. After a 30-min incubation at 370C, the chambers wererinsed, filled with control KRB supplemented with 1:10 guinea pig complement (Behring Institut, Marburg, FRG), and incubated 60 min at370C.Thechambers were then filled with a 0.2% solutionof trypan blue and incubated for 10 minat370C.Atthe endof this lastincubation,theywererinsedagainandfinally fixed in Bouin's solution (8). Insulin-containing B cells wereidentified by

immuno-staining with the same antibodies used for the secretion test, now

diluted1:200,asdescribed (8, 15). Six experiments (100-200 insulin-immunostained and trypan blue-negative singleBcells andBcellpairs were scored per condition and per experiment) were performed to evaluatethe percentage of B cells that formedahemolytic plaque and theindividualareasof all theseplaques,twoparameters that indicate theproportionofsecretingcells andgivearelative estimate of their individual output,respectively(8, 15-17).Fromthese data,we calcu-latedthe total plaque development perBcell aggregate and the average totalplaque development perBcell,asdescribed in(8, 15).Values and mean±SEMestimateswerecomparedbyanalysisof variance and bya

Student'sunpairedttest,respectively.

To testthesecretion of single RINm5F cells, freshly passaged cells werediluted at a density of 2X105/ml and kept at370Cunder

contin-uousrotationwithinaspinner flask,whichwasfilled withRPMI1640 medium supplemented with 25 mM Hepes and 1% newborn calf

serum.Under theseconditions, virtuallyallcellswerestillsingleand excluded trypan blue 3 h after cell passage. Atthistime,aliquots of 500,000single RINm5Fcellswereprepared into siliconized tubes for thetestincubation. Totestthesecretion of aggregated RINm5F cells, 150,000cellswereplated per 35-mm culture dish (2mlof medium) and cultured for 5 d.Atthistime,mostcells had formed extensive aggregates andwereusedfor the secretiontest.Tothis end, bothsingle

andaggregated cell aliquotswereincubated 30 minat37°C into2 ml

of control KRB,supplementedornotwith 10mMD-glyceraldehyde

and3.5mMheptanol. Media and cellswerecollected forinsulin mea-surement asdescribed above. Radioimmunoassayfor insulin and

ex-pression of datawerealsoasdescribed for intact islets of Langerhans.

Tostudy the distal steps of thesecretorypathway,singleRINcells

werepermeabilized by 10 pulses of 30

Ms

ofa3 kV/cm electric field,as

TableI.ExtentofCoupling between Insulin-producingBCells

Extentof dye Dye-labeled Uncoupled Group diffusion* Bcells$ Bcells

n

"m2 n %

Control 527±23 3.0±0.4 20

(64) (20)

Heptanol 3.5mM 330±26§ 1.5±0.2§ 68

(56)

(25)

Heptanolreversibility 480±21" 2.9±0.3" 30 (88) (30)

Values are mean±SEM and werecompared using the unpaired Stu-dent'sttest.The number of *islets thatweremicroinjectedandof

tinjected

isletsthat weresectionedis indicated in parentheses.§_P <0.001 ascompared withthecorrespondingcontrolgroup;"lP

<0.001 ascompared with the heptanolgroup.

described in (10). After permeabilization,106cells were diluted in 1 ml

ofa 140 mMpotassium glutamate buffer, supplemented with the

different

Ca2"

concentrations specifiedin the text, and tested for

insu-linsecretionasdescribed (10).

Measurementofcytosolic

[Ca2"]

andpH.Suspensionsof RIN cells

were loadedfor30min at

370C

with 1 mMfura-2/acetoxymethylester

or 10MAM 2',7'-bis (carboxyethyl)-5,6-carboxyfluorescein/acetoxy-methylester (BCECF), respectively. Afterwashing, 106cellswere

sus-pended in 1 mlof controlKRB and placedinthe glass cuvette of an

LS3 spectrofluorometer(Perkin-Elmer Corp., Norwalk, CT) under

continuousstirringat370C.Measurements and calibrations were

per-formedaspreviously described (13, 18).

Results

Bcell coupling. Fig. 1 shows the appearance of pancreatic rat

islets

microinjected with Lucifer yellow

in their core, a

region

formed almost

exclusively by insulin-producing

B cells. In control

islets,

most

microinjections led

totheintercellular

ex-change of Lucifer yellow, indicating

the

frequent coupling

of B

cells

(Fig.

1, A

and

B). Characteristically, coupling

was

spa-tially restricted,

the average number of

communicating

Bcells

being

3.4±0.2

for

the 46

injections

(all

experimental

groups

pooled)

that

revealed

coupling

in this study. By

contrast,

in

islets

incubated

in the presence

of heptanol,

most

injections

led to the

retention of Lucifer yellow within

the injected

cell,

indicating uncoupling (Fig. 1,

C and

D)

of the

insulin-con-taining

B

cells

(Fig.

1

E). This uncoupling

wasconfirmed

by

a

quantitative analysis

in control and

heptanol-exposed

islets.

As

shown

in Table

I,

heptanol

significantly

(P

<

0.001)

de-creased

the extent

of dye

diffusionand the number ofBcells

labeled

by

the

dye after

each

injection.

The alcohol also

mark-edly increased

the

proportion

of B cells that

appeared

uncou-pled, as

judged

by

their

inability

to

exchange

Lucifer

yellow

Figure1. IsletsofLangerhans

microinjected

withLucifer

yellow.

(A)

Two successive

injections

inacontrol islet reveal the intercellular diffu-sion of the

injected

tracerwithin limitedareasof the islet

profile. (B)

Thecorrespondingsection,throughthese two smallareas,shows that Lu-ciferyellowhaddiffused into three

coupled

Bcells inonecommunication

territory

(arrow)and intotwoB cells in the other(arrowhead).In the

two

territories,

thearrowand the arrowhead

point

tothe

injected

cell.(C)Afterexposure to3.5 mMheptanol, microinjectionfailed to reveal

intercellular diffusion of Lucifer

yellow. (D)

Accordingly, only

the

injected

cell

(arrow)

wasfound labeledbythetracerin thecorresponding

Table II. Insulin Secretion from Isolated Islets ofLangerhans

Secretedimmunoreactive insulin

Heptanol Heptanol Incubation Group Control 3.5 mM 0.5mM

ng/JOisletsper15min

1st Glucose 2.8 mM 13.1±1.3 30.3±3.6§ 15.2±2.1

(29)

(25)

(15)

1st Glucose 16.7 mM 40.2±2.9* 28.7±4.6§ 29.5±3.0

(28)

(24)

(15)

1st Glucose 16.7 mM 90.6±8.9*t 39.7±6.0 39.9±2.9

+ IBMX 1 mM (30) (23) (15)

2nd Glucose2.8 mM 5.4±0.5** 6.3±0.7tt ND

reversibility (10) (10)

2nd Glucose 16.7 mM 26.6±2.5' 36.6±4.9 ND

reversibility (10) (10)

2nd Glucose 16.7 mM 53.8±6.611 56.7±5.311 ND

+IBMX I mM (10) (10)

reversibility

During thefirst 15-minincubation,islets incubated in the presence

of2.8 mMglucose, 16.7 mMglucose,or16.7 mM glucose plus 1 mMIBMX(control groups), were compared withisletsincubated in thecorrespondingmedium,supplemented nowwitheither3.5 or 0.5 mMheptanol(heptanol groups). When themediawere collectedfor

measurementoftheimmunoreactive insulinsecretedduringthefirst incubation,heptanol was removedfromalltubes, theisletswere

rap-idlywashed in controlmediumand testedagain duringasecond 15-minincubationin the absence (glucose2.8mM group) or presence

ofsecretagogues(16.7mMglucoseand 16.7 mMglucoseplus 1 mM IBMXgroups).Thus,during thissecondsecretiontest,theisletsthat hadbeen exposed toheptanolduringthefirst15-minincubation (heptanol3.5 mMgroups)werecomparedtoisletsthat had been processed inparallel withouteverbeing exposedtothealkanol

(con-trolgroups).Valuesaremean±SEM of the numbersof isletbatches

indicated within parenthesis. ND,notdetermined.

*P<0.001ascompared withthevalue observed in theglucose2.8 mM group; $ P<0.001 ascompared withthevalue observed in the glucose 16.7 mM group; §P<0.001ascompared with the

corre-spondingcontrol group; P<0.05, ¶P<0.02,** P<0.005,$$ P

<0.001 ascompared withthecorrespondinggroupofthefirst

incu-bation.

with their neighbors.

Thesechanges were all reversible within

15 min

after

the removal

of

heptanol. At this

time,

B cell

coupling

was similar in islets that had been exposedto

hep-tanol and then returnedto acontrolmedium

(heptanol

revers-ibility

group) and in control

islets

that hadnotbeen exposedto

the

uncoupling

agent

(Table

I).

Insulin secretion

of

primaryBcells. The insulin secretion of control and

heptanol-exposed

intact islets is shown bothas

absolute insulin output

(Table II)

and as percentage

of

the

initial

content

of

the B cells

(Fig. 2).

Both parameters show

thatin

controls, insulin

secretion was

markedly (P

<

0.001)

stimulatedoverthe basal value observed in the presence of 2.8 mM glucose, by both glucose (16.7

mM)

and glucose

(16.7

mM) plusIBMX(1 mM).Two

changes

of this normal pattern

were seen

during

a 15-min

incubation

in the presence

of

a

concentration

of heptanol (3.5 mM)

causing

cell

uncoupling.

First,

basal secretion of insulin

increased significantly (P

<

0.001). Secondly,

both

glucose

and

glucose plus

IBMXdid

not anymorestimulatesecretion over basallevel(Table IIand

FIRST INCUBATION

z

w 6-z

0

°

4-

U-o

2-H

0

0~

,,,10

Cr

w

z6

-J 8 (1)4

z

O

2-0 Glucose mM 2.8

IBMX1mM

-I

SECOND INCUBATION

off,

16.7 16.7

- +

2.8 16.7 16.7

- +

Figure2.Insulin secretion of isolatedratislets duringtwosuccessive 15-min incubations at 370C. (A) Duringthefirst incubation, the

in-sulin secretion of control islets (opencolumns)wasstimulated

two-fold by 16.7 mMglucose and fivefoldby addition of1 mM IBMX.

Additionof 3.5 mMheptanol(solidcolumns) increased significantly (P<0.001) the basal secretion observed in the presence of2.8 mM

glucoseandabolished thestimulatory effect of both 16.7 mM

glu-cose and 16.7mMglucose plus 1 mM IBMX.(B) Duringthe second

15-min incubation, which was performed after removal of heptanol,

thesecretion ofisletsthat had beenexposedto theuncoupler(dark shaded columns)wasalmostidenticaltothatofcontrols that had

been processedinparallel withoutbeing exposedtoheptanol (light shadedcolumns). Insulin release wasexpressedaspercentage ofthe

initial hormone content of the islets. Values are mean±SEM ofthe

number ofisletbatchesindicated within thecolumns.

Fig. 2 A). These two changes were fully reversible within 15

min after

removal

of

the alcohol and return of the isletsto a

controlmedium (TableII andFig. 2B). Separateexperiments showed a similar inhibition of secretagogue-induced release from isolated islets incubated for 15 min in thepresence ofa

much lower concentration (0.5 mM) ofheptanol (Table II). Under the latterconditions, the isletoutput expressedas per-centage of insulincontent,represented 46.7% and 54.4% of the corresponding control value in thepresence of 16.7 mM

glu-coseand 16.7 mMglucoseplus1 mM IBMX, respectively (not shown).

To assesswhetherthe loss ofBcell responsiveness to

glu-cose could be duetothe heptanol-induced B cell uncoupling,

wefurther compared theeffects of heptanolonthe secretion of

single (uncoupled) B cells and of B cell pairs, the smallest

assembliesin which B cells cansharegapjunctions.2 Usinga

hemolytic plaque assay (Figs. 3 and4), wefound that, after a 2. Dual patch-clampwhole-cellrecording and dye microinjection have

--

_9.A:

_Rui-

_j

><

_*.

_?t

;

4 4~~~~~~~~~~~'. 0

46

*,

4

o .

lk

*. '

'-

4b''

*+au*. * t *-##w~~~~~~~~~~~~~~'"" _4i

X_

Figure 3. Phase-contrast and

corresponding

fluorescence views of

dispersed

islet cells studied in the

hemolytic

plaqueassay for insulin. Aftera

30-min incubation in the presence of 16.7mMglucoseand0.5mMheptanol,singleBcells(A)andBcell

pairs

(B)that secreted insulinare

identified byasurroundinghemolytic plaquethatcomprisesthe

nonrefringent

ghostsof thecomplement-lysed erythrocytes.

Immunostaining

for insulin (C and D) revealedthat,undertheconditions ofourtest, all

hemolytic

plaquesdevelopedarounddifferentiated

_{insulin-containing}

B

cells. Bar, 15gminallpanels.

30-min

stimulation by

16.7 mM

glucose, 48.8±3.9% of

single

B cells had

formed

a

hemolytic plaque3 with

an area

of

3. This percentagewas9.2timehigher (P<0.001)than thatofsingleB cellsreleasingdetectableamountsof insulininthe presence of 2.8mM

glucose(Table III),indicatingthatdispersedB cells hadretained

re-sponsivenesstotheirphysiologicalsecretagogue.

1,368±117

Am2

(Table

III).

These valueswere

significantly (P

<

0.001)

lower

than

those evaluated

for

B

cell

pairs

(Table

III).

Asa

result,

total

plaque

development

(the

product

of the

pro-portion of

plaque-forming

cells

by

thearea

of individual

he-molytic

plaques) of

B

cell

pairs

wasthree-to

fourfold

higher (P

<

0.001)

than that

of

single

Bcells

(Fig.

4

A).

Correcting

for the

different

numbers

of

B

cells

thataccountfor

this

higher figure,

z W 2500 a.

-J 2000

w

o 1500

CY ₁₀₀₀ -J

-J 500

H 01

A

6

D z 2500

.1J

a

-2000 9 "E

uJ

-J1500-a-J

UJ

00

=)

1000-500;

0 S

H< 0

0

H- SINGLE B-CELLS

6

T

-B3-CELL PAIRS

Figure4.Totaldevelopmentofhemolytic plaquesreflectingthe

in-sulin secretion ofsingleBcellsand B cellpairsaftera30-min incu-bation in the presenceof 16.7mMglucose (opencolumns)and 16.7 mMglucose plus0.5mMheptanol(solid columns).(A) Under con-trolconditions (opencolumns),totalplaquedevelopmentwas3.1 times higher(P<0.001)forBcellpairs (2059±197 AMm2,n=6) than forsingleBcells(661±72 AMm2,n = 6). Addition ofheptanol(solid columns)didnotmodifysignificantlythe totalplaque development ofsingleBcells(488±43 AMm2,n=6) butmarkedly decreased (P

<0.02)thatofBcellpairs(1213±251 AMm2,n=6).(B) Expression of

these data on a percellbasis resulted inatotal plaque development which, under controlconditions (opencolumns), was higher (P

<0.01) foraBcellwithinapair(1,029±99

_Am2,

n=6) than fora

singleBcell(661±72 AMm2,n=6).Asthesizeofahemolyticplaque isafunction of theamountofhormonesecreted,these dataindicate

thattheaverage outputof control individualBcellswasabouttwice

ashigh whenaBcellwas partofapairthan when it occurredas a

single cell. This differencewas no moreevident inthe presenceof heptanol(solidcolumns),owingto asignificantreduction (P<0.02)

of plaque development estimated for the individual Bcellsthat formedpairs(606±125

AMm2,

n=_6).

we

found

that total plaque development per cell (the ratio of

total

plaque development by

the number

of

B

cells)

was, in average, about

twice

as

large (P

<

0.01) for

Bcell

pairs

than

for

single

Bcells

(Fig.

4

B). Addition of

0.5 mM

heptanol did

not

affect

significantly

the

proportion of

single

B cells thatwere

glucose-responsive

andthearea

of

the

hemolytic

plaquesthey

formed (Table III). By

contrast,

these

parameterswere

signifi-cantly (P

<

0.02-0.05)

reduced in B cell

pairs (Table III),

leading

to amarked

reduction

(P<

0.02) of

total

plaque

devel-opment

(Fig.

4

A).

As a

result,

even

though this

parameterwas

still

significantly

(P

<

0.02)

higher for

B cell

pairs

than

for

single

Bcells

(Fig.

4

A),

total

plaque development

per cellwas

now

similar

whether B cellswerepart

of pairs

oroccurredas

single

units

(Fig.

4

B).

Comparable

resultswere

obtained

in the presence

of

3.5 mM heptanol, even though in the presence

of

this

alkanol

concentration,

the

proportion of single

B cells

stained by

trypan bluewas

increased (not shown).

Table III. Parameters Characterizing Primary B Cells in the Plaque Assayfor Insulin

Cell Plaque-forming Individual arrangement Group aggregates plaque area

% Am2

SingleBcells Glucose2.8 mM 5.3±1.0 937±213

(32/6) (32/6)

Glucose 16.7mM 48.8±3.9* 1,368±117

(1331/6) (655/6)

Glucose16.7 mM 42.2±2.8* 1,162±80 +heptanol 0.5 mM (989/6) (427/6) Bcellpairs Glucose 2.8 mM 5.7±1.8 1,061±190

(11/6) (11/5)

Glucose16.7 mM 83.0±2.9*tt 2,467±208*tt (370/6) (305/6)

Glucose 16.7mM 66.1±5.4*t** 1,761±249§

+heptanol 0.5mM (241/6) (166/6) Numbers in parentheses are number of single cells and cell pairs/

numberof experimentsscored(forthepercentage ofplaque-forming

aggregates parameter) and number of hemolytic plaques/numberof experimentsscored(fortheindividual plaque area parameter).

*P <0.01 ascompared with the values found in the glucose 2.8 mM

group;tP <0.02ascomparedwiththe valuesfoundin the glucose

16.7 mM group;§P <0.05,1P<0.02,1P<0.01, ** P < 0.003, tP <0.001 ascomparedwith the correspondingvaluesofsingle B cells.

Coupling

and secretion

of

RINm5F cells. Toassessfurther

whether heptanol had pleiotropic, nonspecific effectson

var-ious steps of the secretory machinery, we used the RINm5F line whichisaconveniencesourceofwell-characterized

insu-lin-secreting cells.

Under control conditions, microinjection experiments

re-vealed that 54% of the RIN cells exchanged Lucifer yellow with alimited number (1.7±0.1) of companion cells. As with primary Bcells, the percentage was

significantly

(P<0.015) reduced after a few minutes of exposure to 3.5 mM heptanol, a condition under which about 61% of RIN cellswere

dye

un-coupled (Fig. 5 and Table IV). RIN cells recovered a control level of dye

coupling

shortly after heptanol was removed and

thecultureswerereturnedtocontrolmedium (Table IV). TableIV. Extent

of Coupling

between RINm5F Cells

Dye-labeled Uncoupled Group RINcells* RIN cells

n %

Control 1.8±0.1 46

(41)

Heptanol 3.5mM 1.4±0.1 61 (49)

Heptanolreversibility 1.8±0.1§ 47

(32)

Values aremean±SEMand werecomparedusingtheunpaired

Stu-dent'sttest.* Thenumberofmicroinjectedclusterswhichwere

scored isindicated inparentheses.t P<0.015ascomparedwith the

correspondingcontrolgroup;§P<0.017ascomparedwith the

hep-tanol group.

Figure5. Clustersof RINm5Fcellsmicroinjectedwith Luciferyellow. (A)In acontrol cluster, the tracer was exchanged between three adjacent cells, revealing junctional coupling.(B) In acluster exposedfor 5minto3.5mMheptanol, Lucifer yellow remained within the injected cell, re-vealinguncouplingofthis cell from itsneighbors. Bar, 20

*Am

inboth panels.

Interms

of insulin secretion,

RIN cells, as primary B cells, responded more to secretagogue stimulation when they were

aggregated

than when they were single (uncoupled). In the

presence

of D-glyceraldehyde,4

the increment above basal

se-cretion

was4.2- and 1.9-fold in thesetwogroups,

respectively

(Table V and

Fig.

6). Furthermore, again as with primary B

cells, heptanol markedly inhibited (P< 0.01) the

secreta-4.This secretagogue, that mimics the effect ofglucose onnormalB cell

secretion,waschosen becauseRINm5F cellsare unresponsiveto

D-glucose (21).

gogue-stimulated secretion of

aggregated RIN cells, but had no

significant effect

onthe

corresponding

secretion of single RIN

cells (TableVand Fig. 6).

To assess whether heptanol perturbed the most distal steps

of secretion, single

RINm5F cells were

electropermeabilized

and then assessed for

insulin

release in the presence of different

Ca2+

concentrations.

Inthis system, an increase in

Ca2+

con-centration from

10-7 to 10-5 M,

elicited

athree- to sevenfold

increase (P < 0.01)

in insulin

release (Table V and Fig. 7). Addition of 3.5 mM heptanol did not modify the basal insulin release observed in the presence of

10-7

M

Ca2+

nordid it Table V.Insulin Secretion ofRINm5F Cells

Immunoreactive insulin

Cellarrangement Group Control Heptanol3.5mM

ng/sampleper 30 min

Intactsinglecells Glucose 2.8 mM 20.2±3.9 19.7±1.0

(15)

(15)

Glucose 2.8 mM + D-glyceraldehyde10 mM 38.0±3.5* 35.1±1.6t

(15)

(15)

Intactaggregatedcells Glucose 2.8 mM 49.4±4.6 37.7±2.2§

(15) (15)

Glucose 2.8 mM+D-glyceraldehyde 10mM 208.7±8.3*

70.8±5.2t'

(15)

(15)

Permeabilized singlecells 10-7MCa2+ 3.4±0.2 4.2±0.2t

(9)

(9)

0-5M

Ca2+

12.8±2.0** 14.4±2.3** (9) (9)

Intheexperimentsonintactsingle cells, each tube contained500,000cells; intheexperimentsonintactaggregated cells,each dish contained about2,500,000 cells;intheexperimentson

electropermeabilized

cells,eachtubecontained 1,000,000cells. Thecorresponding insulin

con-tents(mean±SEM)were362±13.4ng/tube (n=60), 1979±59ng/dish (n =60)and 215±14ng/tube (n= 36), respectively. * P<0.005,* P

<0.001ascompared with the value found intheglucose2.8 mMgroup; §P<0.005,11P<0.02,t P<0.001ascompared with

corresponding

control group;**P<0.001 ascomparedwith the valuefoundin the 10-7Ca2+group.

I~-z

|

8

ILi

06

w

-i

Cl)

w-J z 2

z o

T

[Ca2+]

mM

1O7

Glucose 2.8mM + D-Glyceraldehyde

10mM

AGGREGATED RIN CELLS

-I-15

+

Figure 6. Insulin secretion of singleand aggregated RINm5F cells.

(A) Under control conditions (open columns),theinsulinsecretion of

single cells increased 1.9-fold (P<0.005) duringa30-min

incuba-tioninthepresenceof 10 mMD-glyceraldehyde. Addition of 3.5

mMheptanol (solid columns) didnotmodify this

glyceraldehyde-in-ducedincrease.(B) Aggregated RIN cellswerestimulated 4.3-fold(P

<0.001) by D-glyceraldehyde. Addition of heptanol markedly de-creased(P<0.001) this normal glyceraldehyde-inducedresponse. In-sulin releasewasexpressedasthepercentageof theinitial hormone

contentof thecells. Valuesaremean±SEM ofthe number of cell samples indicated within the columns.

affect the increased insulin output induced by

l0-`

M Ca2+ (TableVandFig. 7).

Cytosolic

Ca2`

andpH. As shown inFig. 7, 3.5mM

hep-tanol did not alter significantly the basal concentration (145±16 nM, n = 7) of free

Ca2`

([Ca2+]i)

in RINm5Fcells

loaded with fura-2 and didnotmodifytheextent(98±9,n=7)

and time course of the

[Ca2+]i

increment caused by 10 mM

D-glyceraldehyde (Fig. 8,upperpanel).Inthepresenceof hep-tanol, these two parameters were 146±5 nM (n = 5)and

99±30 nM(n= 5), respectively.

Heptanol also didnotalter basalintracellular pH (pHi)in RINm5Fcells loaded with BCECF(7.11±0.03,n =4) and did

notmodifythedecreasein

pHi

(0.09±0.01,n =5) induced by D-glyceraldehyde (Fig. 8, lowerpanel).Inthepresenceof hep-tanol, these two parameters were 7.12±0.02 (n = 4) and

0.10±0.001 (n = 5), respectively. Similarly, heptanoldidnot

alter the increase in

_pHi

normallyinducedby NH4Cl (Fig. 8).

T

10-5

10-7

lo-5

Figure 7.Insulinsecretion ofsingleelectropermeabilizedRINm5F cells. Under control conditions(opencolumns), raisingthe

Ca2"

con-centration of the incubation medium from

IO-'

to IO-' M eliciteda

3.7-fold increase(P<0.001)in the insulin secretion of RIN cells that had beenelectropermeabilizedtopermitthefreeaccessof the cationtocytoplasm. Addition of 3.5mMheptanol

(solid columns),

didnotchange thispatternsignificantly.Insulin releasewas

ex-pressedaspercentageoftheinitialhormonecontentof the cells.

Values aremean±SEM ofthenumber ofsamples indicated within the columns.

Discussion

We have shown here that a long-chain alkanol that blocks

gap

junctions

in several systems (7, 13, 14, 22-24), also

un-couples

B cells

within intact

pancreatic islets

and affects in

parallel

their insulin secretion. Most

strikingly, uncoupling

concentrations

of

heptanol abolished

within

minutes

the

nor-mal

increase

in

insulin secretion

causedby glucose. Both the

coupling

and thesecretion

changes

were

fully

reversible within

minutes after removal of

heptanol and, thus, cannot be easily

accounted

for by

a

major toxic effect of

the alcoholon Bcells. Rather, the

magnitude of

thesechanges and

their

reversibility

suggest that

heptanol

actswithsome

selectivity

on one or more

step(s)

of

the B cell secretory

pathway that is (are) critical for

a

proper

insulin

secretion.

The

obvious

importance of unraveling

the

mechanism(s)

whereby B cells may be, or become unresponsive to glucose under certain conditions

(6, 8, 15, 25, 26),

led us to further

investigate

the

rapid,

marked, and

fully

reversible

glucose

unresponsiveness induced by heptanol.

We

first

found

that the

heptanol effect

was

dependent

on cell contact.

Thus,

in the

presence

of

the

alcohol,

glucose-stimulated insulin secretion

was not

significantly modified

in

single

Bcells, butwas

mark-edly

inhibited in B cell aggregates. The latter inhibition was

already evident

in

pairs,

the smallest

assemblies

in

which

two B

cellscan

interact

directly

via

gap

junctions

(19,

20). Whereas,

under control

conditions,

the average

secretory

contribution of

individual

B cellswas

twice

as

high

within

pairs

than when

these cellswere

single,

suchadifferencewas no moreobserved in the presence

of heptanol.

Under the latter

condition,

Bcells

forming pairs secreted,

individually,

asif

they

were

single

cells.

The

contact-dependent mechanism

responsible

for

the

heptanol-induced glucose

unresponsiveness

remains

tobe

elu-cidated.

The

observation

that

heptanol

blocks

almost

com-pletely

B cell gap

junctions

without

disrupting

the

physical

z

w

I-z

0 C.) 0 IL.

LL 0

e-z

w

w

a.

w

Cl)

.4

w

-J

w

z

-J C/)

z

121B

9.

[Ca2+]j

(10-7M)

2.75 1.50

-_

D-GLYCERALDEHYDE 10 mM

2.75 1

1.50--to

I

2 min

I

HEPTANOL D-GLYCERALDEHYDE 3.5 mM 10 mM

pHi

7.30 - pH 7.41

7.20 -7.10 7.00

7.30

-7.20

-7.10

-7.00 _

D-GLYCERALDEHYDE 10 mM

,j

NH4CI

10mM

pHo=7.43

I

I

HEPTANOL D-GLYCERALDEHYDE NH4CI

3.5 mM 10 mM 10 mM 2 min

Figure8.Cytosolic [Ca

2']

andpHofRINm5Fcells.(Upperpanel)

The upper trace showsthe normal increase ofcytosolic free

Ca2+

([Ca2+],)

causedbyD-glyceraldehyde.The lowertraceshowsthat3.5

mMheptanol didnotaffect

[Ca2+],

significantlyanddidnotmodify the normalchange of thisparameterduringasubsequent stimulation by D-glyceraldehyde. (Lowerpanel)Theuppertraceillustratesthe

normalchanges ofcytosolic pH (pHi)in cellsexposedto D-glyceral-dehydeandtoNH4Cl.The lowertraceshows thatheptanol didnot

modify significantlythesechangesandbasalpHi.Inbothtracings,

pH.

isthepH oftheextracellular mediumthatwasmonitoredwitha

pHelectrode.

integrity

of intercellular contacts, favors the view that the

se-cretory

changes

observed are somehow related to the

inhibi-tion

of

the

direct

cytoplasmic

exchange

of ions and small

mol-ecules that gap

junction permit

between

adjacent

Bcells.

Ob-viously,

the

alternative

possibility

that

heptanol

has

pleiotropic

effects which

could

perturb

in

parallel

but

independently

the

coupling

and the secretion ofBcellswasconsidered. Totest

this

possibility,

we

first

studied

whether

heptanol

affected the

cytoplasmic free

Ca2'

and

pH of

insulin-producing

cells,

since

thesetwo

factors

are

thought

to

be

implicated

in the

physiolog-ical blockade of gap

junctions (2, 3, 27)

andareknowntoalso

influence

several steps

of

insulin secretion

(28, 29).

Since

in-tracellular

Ca2"

and pH measurements are not readily carried

out on

intact islets

and require large numbers of dispersed

cells, we used, as a model, cells of the RIN line (9). When compared to

primary

B cells under control (basal or stimu-lated)

conditions,

the RIN cells we used showed a less frequent andextensive coupling and secreted much less insulin. In spite

of

these

quantitative

differences, the patterns of RIN cell cou-pling and

secretion

showed the major features observed with primaryBcells. Thus coupling, as revealed by the exchange of Lucifer yellow, was not a consistent feature of every RIN cell and, when

it

occurred, it never involved the entire cell

popula-tion

comprising

aculture cluster. Furthermore,insulin

secre-tion

was two to

five times

larger

from

aggregated than from

single

RINcells, a pattern also observed with primary B cells.

After

exposure to heptanol, the behavior of RIN cells was also

reminiscent

of that

of

Bcells. Thus, RIN cells also uncoupled

in arapid and fully reversible way and became unresponsive to D-glyceraldehyde, a secretagogue that mimics the effects of glucose on islet function, in a manner which, as for primary B cells, depended oncell contact. Thus, in

spite of

some

differ-ences,

for

example as far as the effect of heptanol on basal insulin secretion was concerned, RIN cells represented an ac-ceptable model system in the context

of

our

experiments.

Using

these

cells,

we

found

that concentrations

of

heptanol

causing

uncoupling

and

inhibiting

the

glyceraldehyde-induced

increase in

insulin

release,

did

not causedetectable

alterations

of

either

cytosolic

[Ca2"]

or pH

of

RIN cells,

suggesting

that thesetwo

factors

most

likely did

not

mediate

the

changes

in

both

coupling

and

secretion.

Even

though

RINcellsdiffer in

several respects from

primary

B cells,

previous

reports have shown that thesetwocell types handle similarly their intracel-lular

Ca2"

and pH under a variety of other experimental

con-ditions (21),

thus

favoring

the

view

thatthey will not behave

differently

in the presence

of

heptanol. As a second approach

tofurther

investigate possible,

nonspecific effects of heptanol, weassessed whether the alcohol perturbed the most distal steps

of

the

secretory machinery

thatlead to the

fusion

of

secretory

granules with

the plasma membrane. The

observation

that

uncoupling concentrations of heptanol did

not

modify

the

Ca2+-induced

secretion

of

single electropermeabilized

RIN

cells

did

not

provide

evidence for

sucha

perturbation.

In

sum-mary, the unchanged

intracellular

Ca2 , intracellular pH and

Ca2+-induced

insulin

release, together

with the lack of effect of

heptanol

onthesecretion of

single cells,

strongly indicate

that,

even if the alcohol alters Bcell

organelles (30)

andchannels5 other than gap

junctions,

suchan

effect

alone couldmost

un-likely

explain

the

secretory

changesweobserved inaggregated cells.

Insummary, the present data provide several lines of evi-dence that the

rapid

loss of responsiveness to secretagogues observedafter exposure of aggregated insulin-producing cells

to heptanol was most

likely

related to the alcohol-induced blockadeofB cell gap

junctions.

Control of secretion by gap

junction-mediated

cell-to-cell

communication

has been

sug-5.Usingthewhole-cellconfigurationofthepatch-clamptechnique,we

havefound that,occasionally, heptanolinhibitspartiallythe outward currents recorded insingleBcells(M.Chanson and P.Meda,

unpub-lishedobservations).

InsulinSecretionandJunctionalCoupling 767

.--gestedas a

possible mechanism coordinating

the

functioning

of the numerous secretory cells that form most vertebrate glands(3, 4, 13, 14,

24, 32).

Theexperimentswe report here providethenovelevidence that short-termandfully reversible changes injunctional coupling are paralleled by drastic and rapidchangesin insulin secretion from intact pancreatic islets andinsulin-producing cellsincontact. Theabolished

secreta-gogue-induced

stimulation

of insulin releaseseenduring B cell uncoupling

is in

marked contrast with the secretion change which is associated with the uncoupling of the acinarcells of theexocrinepancreas(13, 14,24, 31).Inthis gland, that

differs

markedly from the islets of Langerhans in several respects,

includingthe composition and amountofgapjunctions, the restingorganization of intercellular communicationsandtheir

modulation during secretagogue challenge (32), heptanol causes arapid, marked andfullyreversibleincrease inamylase

secretion (13, 14, 24). Although the reason(s) for this opposite behavior remain(s) to be elucidated, the different secretory

changeof endocrineandexocrinecellsuncoupledbyheptanol

arguesfurther againstageneralized, nonspecific cell perturba-tion induced by the alkanol. Rather, thedata imply a more

subtle, tissue-specific effect of this agent, likely relatedtothe

inhibition of the gapjunction-mediated cell-to-cell communi-cationsitcauses.

Acknowledaments

We thank L. Burkhardt, A. Charollais, S. De Mitri, P. Ruga, J.-P.

Gerber, and C. Bartley for technical assistance.

This work was supported by grant 31-26625.89 from the Swiss National Science Foundation and grant 187384 from the Juvenile Diabetes Foundation International.

References

1.Orci,L., R. H.Unger, andA.E.Renold. 1973.Structural

cou-plingbetweenpancreaticislet cell. Experientia(Basel).29:1015-1018.

2.Loewenstein,W. R. 1981.Junctionalintercellular

communica-tiorV

the cell-to-cell membrane channel.Physiol.Rev.61:829-913. 3. Hertzberg, E.L., andR.G. Johnson, editors. 1988. Gap Junc-tions. AlanR.Liss, Inc., New York. 548pp.

4. Meda, P., A. Perrelet, and L. Orci. 1984. Gap junctionsand

cell-to-cellcoupling in endocrine glands. Mod. Cell Biol. 3:131-196.

5. Meda, P., A.Perrelet,and L. Orci. 1979.Increaseofgap junc-tions between pancreatic B-cells duringstimulation of insulin

secre-tion. J. Cell Biol. 82:441-448.

6. Halban, P. A., C. B. Wollheim, B. Blondel, P. Meda, E. N.

Niesor, and D. N.Mintz. 1982. The possible importance ofcontact

betweenpancreatic islet cells for thecontrolof insulin release. Endocri-nology. 111:86-94.

7.Johnston,M.F., S.A.Simon,andF.Ramon. 1980. Interaction of anaesthetics with electricalsynapses.Nature(Lond.). 286:498-500. 8. Salomon, D., andP. Meda. 1986.Heterogeneity and

contact-dependentregulationof hormone secretion by individual B-cells. Exp. Cell Res. 162:507-520.

9. Gazdar,A.F., W. L.Chick,H.K.Oie,H.L.Sims,D. L.King, G. C. Weir, and V. Lauris. 1980. Continuous, clonal, insulin- and

somatostatin-secretingcelllinesestablished fromatransplantablerat

isletcelltumor.Proc.Natl. Acad.Sci. USA.77:3519-3523.

10. Vallar, L., T. J. Biden, and C. B. Wollheim. 1987. Guanine

nucleotides induce

Ca2"-independent

insulin secretion from

perme-abilizedRINmSFcells. J. Biol. Chem. 262:5049-5056.

11. Meda, P., R. L. Michaels, P. A. Halban, L. Orci, and J. D.

Sheridan. 1983.Invivo modulation of gap junctionsanddye coupling between B-cellsofthe intact pancreatic islet. Diabetes. 32:858-868.

12.Meda, P.,R. M.Santos, andI. Atwater. 1986. Direct identifi-cation ofelectrophysiologically monitored cellswithin intact mouse

isletsof Langerhans. Diabetes. 35:232-235.

13.Meda,P.,R.Bruzzone,S.Knodel, andL.Orci. 1986. Blockage ofcell-to-cell communication within pancreaticacini is associated with

increasedbasalreleaseof amylase. J. Cell Biol. 103:475-483.

14.Meda, P., R. Bruzzone, M.Chanson, D.Bosco, and L. Orci.

1987. Gapjunctional coupling modulates secretion ofexocrine

pan-creas.Proc. Natl.Acad. Sci. USA. 84:4901-4904.

15. Bosco, D., L. Orci, and P. Meda. 1989. Homologous but not heterologous contact increases the insulin secretion of individual pan-creatic B-cells. Exp. CellRes. 184:72-80.

16. Jerne, N. K., C. Henry,A. A.Nordin,H.Fuji, A. M. C. Koros,

andI.Lefkovits. 1974. Plaque formingcells:methodologyandtheory. Transplant. Rev. 18:130-191.

17.Alberts, W.,A.Wouters,D. Van derMassen,A.Persoons,and

C.Denef. 1988. Adiffusion-adsorption model for thecomputationof

theamountof hormone aroundasecreting cell, detected bythe reverse

hemolytic plaque assay. J. Theor. Bio. 131:441-459.

18. Wollheim, C. B., and T. J. Biden. 1986. Second messenger function of inositol 1,4,5-trisphosphate. J. Bio. Chem. 261:8314-8319.

19. Perez-Armendariz, E. M., D. C. Spray, and M. V. L. Bennett.

1988.Properties of gap junctions between pairsof pancreaticbeta cells

of mice.Biophys.J. 53:53a. (Abstr.)

20. Meda, P., D.Bosco, E.Giordano, andM.Chanson.In Biophys-ics of Gap Junction Channels. C. Peracchia, editor.CRC Press,Boca Raton, FL.Inpress.

21.Wollheim, C. B., and T. J. Biden. 1986. Signal transductionin

insulin secretion: comparisonbetweenfuelstimuli andreceptor

ago-nists.Ann.NYAcad. Sci.488:317-333.

22. Bernardini, G., C. Peracchia, and L. L. Peracchia. 1984.

Re-versible effects of heptanol on gapjunction morphology in

mamma-lian heart. J. Membr. BiW. 34:307-312.

23.DNleze,J.,and J. C.Herve. 1983.Effect of severaluncouplersof cell-to-cellcommunication on gapjunction morphology in

mamma-lian heart. J. Membr. Biol. 74:203-215.

24. Chanson, M., R. Bruzzone, D. Bosco, and P. Meda. 1989.

Effectsof n-alcohols onjunctional couplingandamylasesecretion of pancreaticacinar cells. J. Cell. Physiol. 139:147-156.

25. Cerasi, E., R. Luft, and S.Efendic. 1971. Decreased sensitivity of pancreatic betacells toglucoseinprediabetic and diabetic subjects:a

glucose dose-responsestudy.Diabetes21:224-234.

26. Bonner-Weir, S., D. F. Trent, R. N. Honey, andG. C. Weir.

1981.Responsesof neonatal ratisletstostreptozotocin.Limited B-cell regeneration and hyperglycemia. Diabetes. 30:64-69.

27. Spray, D.C.,andL. V.Bennet. 1985.Physiologyand pharma-cology of gap junctions.Ann.Rev. Physiol. 47:281-303.

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

29.Pace, C. S., J. T.Tarvin,and J.S.Smith. 1983. Stimulus-secre-tion couplingin B-cells: modulation by pH.Am. J. Physiol.

244:E3-E18.

30. Knodel,S., P.Meda, andL.Orci. 1987.Rapid in vitro

forma-tionofsmoothendoplasmic reticulumaggregateswithin

peptide-pro-ducingisletcells. J. Cell. Physiol. 133:111-118.

31. Bruzzone, R., and P. Meda. 1988. The gap junction: a channel formultiplefunctions? Eur. J. Clin.Invest. 18:444-453.

32. Meda, P. Gapjunctionalcoupling andsecretion inendocrine