Preservation of insulin mRNA levels and insulin secretion in HIT cells by avoidance of chronic exposure to high glucose concentrations

Preservation of insulin mRNA levels and insulin

secretion in HIT cells by avoidance of chronic

exposure to high glucose concentrations.

R P Robertson, … , K L Pyzdrowski, T F Walseth

J Clin Invest.

1992;

90(2)

:320-325.

https://doi.org/10.1172/JCI115865

.

Glucose toxicity of the pancreatic beta cell is considered to play a secondary role in the

pathogenesis of type II diabetes mellitus. To gain insights into possible mechanisms of

action of glucose toxicity, we designed studies to assess whether the loss of insulin

secretion associated with serial passages of HIT-T15 cells might be caused by chronic

exposure to high glucose levels since these cells are routinely cultured in media containing

supramaximal stimulatory concentrations of glucose. We found that late passages of HIT

cells serially cultured in media containing 11.1 mM glucose lost insulin responsivity and had

greatly diminished levels of insulin content and insulin mRNA. In marked contrast, late

passages of HIT cells cultured serially in media containing 0.8 mM glucose retained insulin

mRNA, insulin content, and insulin responsivity to glucose in static incubations and during

perifusion with glucose. No insulin gene mutation or alteration of levels of GLUT-2 were

found in late passages of HIT cells cultured with media containing 11.1 mM glucose. These

data uniquely indicate that loss of beta cell function in HIT cells passed serially under high

glucose conditions is caused by loss of insulin mRNA, insulin content, and insulin secretion

and is preventable by culturing HIT cells under low glucose conditions. This strongly

suggests potential genetic mechanisms of action for glucose toxicity of beta cells.

[…]

Research Article

Find the latest version:

Preservation of Insulin

mRNA

Levels and Insulin

Secretion in

HIT

Cells

by Avoidance of Chronic Exposure

to

High

Glucose

Concentrations

R. PaulRobertson,* Hui-Jian Zhang,* Kathryn L.Pyzdrowski,*andTimothyF.Walseth*

*DiabetesCenter and theDivision ofEndocrinologyandMetabolism, Department ofMedicine; and$Department of Pharmacology, UniversityofMinnesota MedicalSchool, Minneapolis,Minnesota55455

Abstract

Glucose

toxicity

of the

pancreatic

betacellis consideredto

play

asecondary role in the

pathogenesis

oftypeIIdiabetes

melli-tus.To

gain

insights

into

possible mechanisms

of action of

glu-cose

toxicity,

we

designed

studiesto assesswhether the loss of insulin secretionassociated with serial passagesof HIT-T15 cellsmightbecaused by chronicexposure to

high

glucose levels since these cellsareroutinely cultured in media containing

su-pramaximal stimulatory concentrations ofglucose. Wefound

thatlatepassagesofHITcells

serially

culturedinmedia

con-taining 11.1 mM glucose lost insulin

responsivity

and had

greatlydiminishedlevelsof insulincontentandinsulin mRNA.

Inmarkedcontrast,latepassagesof HIT cells cultured serially inmedia

containing

0.8 mM glucose retained insulin mRNA, insulincontent, andinsulin

responsivity

toglucose in static in-cubationsand

during perifusion

with glucose. Noinsulingene

mutationoralteration of levels ofGLUT-2werefound in late

passagesofHITcells culturedwith media

containing

11.1mM glucose. These datauniquely indicatethat lossof beta cell func-tionin HITcellspassed

serially

under

high

glucose conditions is causedby loss of insulinmRNA, insulincontent,and insulin secretion and is preventable by

culturing

HITcells underlow glucose conditions. This strongly suggests potential genetic mechanisms of action for glucose

tokicity

ofbetacells.(J. Clin. Invest.

1992.90:320-325.)

Key words:glucosetoxicity*insulin

gene *diabetes mellitus

Introduction

Thethesis that glucose toxicity plays a secondary role in the

pathogenesis

oftype II

diabetes

mellitus maintainsthatafter hyperglycemia is established,glucoseitself inhigh

concentra-tions hasdeleterious

effects

on pancreatic isletbeta cell

func-tion (1-18). TypeIIdiabetes is characterizedby aspecific lack

of normal

recognition

of glucose signals, sinceallother insulin secretagogues (except glucose) elicit first phase insulin re-sponseswheninjected intravenously(19-22). Type I diabetes

ismarkedlydifferentinthis regard because it is caused by beta cell death and complete absence ofinsulin secretion. In support of theglucosetoxicityconcept, defective glucose-induced

insu-Address correspondence to R. Paul Robertson, M.D., The Diabetes

Center,Box 101 UMHC, University of Minnesota, Minneapolis, MN 55455.

Receivedfor publication 22May1991and in revisedform 13 Febru-ary 1992.

lin secretion in type II diabetic patients can be partially

re-versed byaninfusion ofexogenous insulin to lowercirculating

glucose levels into the normal range (1, 5).This suggests that hyperglycemia is notonlyan effect oftypeIIdiabetes,but may be acontributing cause of progressive deterioration of insulin secretion.However, as yet no biochemical mechanism for glu-cosetoxicity has been identified ( 15).

The HIT-T1 5 cell is a clonal cell line ofpancreaticislet beta cells derived from SV-40 transfection ofSyrian hamster pancre-atic islets(23). The HIT cell is a valuable research tool because itis one of only a few cell lines that secrete insulin in response toglucoseand because itrespondsto

physiologic

inhibitors of

insulin secretion(23-31). However,insulincontentand insu-linsecretionbythis cell lineprogressivelydiminish with serial passage(23, 26, 30).Theexplanationfor this loss of secretion

hasnotbeenforthcoming, althoughfurther transformation of the cell is often mentionedas apossibility byinvestigatorswho study this cell line. However, because loss ofinsulin secretion is

highly predictable bypassage number andnotsporadicin na-ture, further cellular transformation as a sole explanation

seemsquestionable.

HIT cells areroutinely seriallyculturedbymost investiga-torsusingmedia

containing

concentrations of 11.1 mM

glu-cose(23-31),aconcentration that issupramaximalfor

stimu-lating

insulin secretion from these cells

(30).

Consequently,

we

designedstudies to address(a)whether the loss of insulin

secre-tion associated with serial passages of HIT cells

might

be

caused by adverse effectsof chronic

culturing

the cells under

high glucoseconditions; and, if so,(b) whether this

phenome-nonis associated withaninsulingenemutation,a

deficiency

in thepancreaticisletglucosetransporter(GLUT-2),ordecreased

levelsof insulin mRNA; and, ifso, (c) whether such changes

might

beprevented by

culturing

thecells inmedia

containing

a

lowerglucose concentrationthatismorephysiologic

for

HIT

cells.

Methods

Cell culture. Stock cultures of HIT-T1 5 cells wereroutinelygrown in 5%C0J95%humidified airat370C,maintained inRPM! 1640

me-dium supplemented with 10% fetal calf serum, passaged once weekly

following detachment using trypsin-EDTA, and fed every 48 h by

changingmediumaspreviouslydescribed (30). Thefinal concentra-tion of insulin(fromfetal calf serum) in theworkingmedia averaged

5-10

guU/ml.

Noantibiotics were used. Beginning with passage 70, cells weresplit weekly and continuously passed in media containing either 0.8mMglucose,aslightly stimulatory concentration, or 11.1mM glu-cose,asupramaximalstimulatory concentration (30). Cell population

doubling levels anddoubling times were determined by plating cells into 12-well plates (4cm2/well)at0.1million or 0.5 million cells/well. Theseconditionsreflect the densities at which the cells were routinely cultured. Dailythereafter for 6 d, cells were detached with

trypsin-320 R.P.Robertson, H.-J. Zhang,K L.Pyzdrowski, andT. F. Walseth

J.Clin. Invest.

© The AmericanSocietyfor ClinicalInvestigation,Inc.

0021-9738/92/08/0320/06 $2.00

EDTA and counted.Viablecellsweredetermined bytrypan blue

ex-clusion,andbothviableand nonviable cellswerecounted. The

cumu-lative doubling levels and doubling timeswere calculated usingthe

following equations:

Numberofdoublingonday i =

ni

-

log

[TCJ/AC(i

1)]

log2 where

TCj=total number cells on day i(aliveanddead).

AC(iI)

=numberofalive cells on the day previous to day i.

6

Cumulative

doubling

level=

Ini

doublingtime(h)= 24h/dX 6 d

cumulativedoublinglevel

Insulinsecretion. Static insulinsecretion in responseto

progres-sively increasingconcentrations ofglucose(0-1 1.1mM) wasexamined

aspreviously described (30). Phasicinsulin secretion in response to 1.1 mM glucose perifusion in the presence of 0.1 mM 3-isobutyl

1-methylxanthine (IBMX)'wasexamined as previously described (32). Insulin mRNA. Cells were rinsed twice with ice-cold Dulbecco's PBS and scraped on ice with denaturing solution (4 M guanidine thio-cynate; 25 mM sodiumcitrate,pH 7;0.5% sarcosyl;0.1M 2-mercap-toethanol). The lysed cells were passed several times through a 22-gaugeneedletoshear chromosomal DNA. RNA was isolated by the procedure of Chomczynski et al. (33). Total RNA (5-15

_Ag)

was frac-tionatedon1.5%agarose-formaldehydegels and transferred to a nylon

hybridizationmembrane,(0.22gmmicron;Micron Seps. Inc.,

West-boro, MA) by electroblotting. The membrane was prehybridized in 50%formamide;5X standard saline citrate(SSC); lOxDenhardt's; 50

mMsodium phosphate, pH 6.5; 1 mg/ml salmon sperm DNA, and

0.1%SDSat30'Covernight.The membranewasthenhybridized

over-night with

32P-labeled

human insulin genomic DNA probe in 50%

formamide; 5xSSC;2x Denhardt's; 50 mM sodium phosphate, pH

6.5;0.1mg/mlsalmon sperm DNA;0.1mg/ml yeast tRNA; and0.1%

SDSat30°C.The membranewaswashedtwicefor 45 minat50°Cin 0.2X SSC and0.1% SDS,andexposedtox-ray film(KodakX-Omat AR)for16-24 h. The probe was humaninsulin genomicDNA(phins

214;Amer.Type CultureCollection, Rockville,MD)oligolabeledwith

[32P]dCTP

(34). The probehybridizedwith a 0.5-kb band on agarose gel fractionation oftotal HIT cell RNA consistent withHITcell insulin mRNA(35).Under thehybridizationconditions employed, the probe also labeled HIT cell 18S and 28SrRNA, and thedensity oflabeled18S rRNAwasproportionalto total RNAoverthe range of total HIT cell

RNAappliedtogels.InsulinmRNAlevelswerequantitated by

scan-ning densitometry of autoradiographs; advantage was taken of 18S rRNAlabelingbynormalizing insulinmRNAblotsforthedensityof thecorresponding18S rRNA.Therefore,dataareexpressedastheratio ofdensityofinsulinmRNAtodensity of18S rRNA.Thevalidityof

this approachhas beenverified byde Leeuw(36). Also,weperiodically

usedaprobeforj3-actinmRNAtoverifyuseof18S rRNAfor

normal-ization.

Restriction enzymeanalysisofinsulingene DNA. DNA was iso-lated from HIT cellsbyManiatis'procedure(37).Genomic DNAwas divided intofoursamples,whichwereeachdigestedwithoneof four

restriction endonucleases(EcoRI,PstI,HindIII,orBamHI).The restric-tionfragmentswereseparatedbyelectrophoresisina0.8% agarosegel.

BstEII digested lambda phage DNAwasusedas a size marker. The separatedDNAfragmentsweretransferred from the agarosegeltoa

nylon membraneasdescribed by Southern(38).DNAwasretainedon thenylonmembranebyultravioletcrosslinkingand thenhybridized

1. Abbreviation usedin thispaper: IBMX, 3-isobutyl

1-methylxan-thine.

with 32P-labeled human insulin genomic DNA probe as described above. The restrictionfragments containingsequencescomplementary

to theinsulinDNAprobeweredetected byexposingx-rayfilmto the nylon membrane.

Glucosetransporter. The amount ofglucose transporter (GLUT-2) was assessed by immunoblot. Antisera (W-19) against GLUT-2 was a generousgiftfrom Dr. Mike Muecklerof WashingtonUniversity, St. Louis, MO. HIT cell extracts obtained from various passages cultured undervariousconditions were prepared asdescribedpreviously and wereelectrophoresed in 1% polyacrylamide gels (39). After

electro-transfertoImmobilon-P,theimmunoblot procedure of Harris et al. (40) was employed usingW-19antisera at a dilution of 1/250.

Statistics. Intergroup comparisonsfordensitometric data and insu-linsecretion data were performed by Student's t tests. Data are pre-sented asmean±SE.

Results

Insulin secreted by HIT cells during cell growth. After splitting

passage 71 into flasks with media containing either0.8 mM

glucoseor 11.1mMglucose,cells wereseriallypassedfor40 wk

byweeklysplitting and continued culturing in media contain-ingtheinitialglucose concentration of either 0.8 or 11.1 mM.

Tocompensatefor theempirical observation oflesser numbers ofcells attachedtothewalls of flasks when 0.8mMglucosewas

used in

media, approximately

fourtofive timesasmanycells

wereinitially plated atthebeginning of each week for earlier

passages

serially

cultured inmedia

containing

0.8mMglucose.

In this manner, approximately equal numbers of cells were presentby the end of each week when comparing low and high glucose conditions. However, itwasascertained by

daily

total cellcountsin

duplicate

wells

performed

in

duplicate

(thesum

of attached cells plus cells

floating

in

media)

forseven

consecu-tive days forpassages75,

85,

and95 that thepopulation

dou-bling

times and cumulativedoubling levels were not

signifi-cantly

different (Fig.

1). Rather, a greater

proportion

of cells reproducibly detached into the

media

whentheywerecultured underconditions of low glucose, which accounted for the lesser numbers of cellsattachedtothe walls

of

the

flasks

atthetime of

subculturing. Initially

and

through

passage

79,

HITcells

ex-posedtothe

higher

glucose concentration secreted insulin into the media that reached levels of -

3,500-4,500

MU/million

cells per48 h

(Fig. 2). Thereafter,

insulin levels in the media

diminished

progressively through

passage 90 to - _100-250

AuU/million

cells. Media insulin levels secretedbycellscultured

in 0.8mMglucosewere

initially 1,000-2,500 MU/million

upto

so 11.1mMGLUCOSE

00.8mMGLUCOSE

10

P75 P85 P95

PASSAGE

Figure

1.

Comparison

ofHITcell

population

doubling

times and

cu-mulative

doubling

levels

over6 datpassages75,

85,and95,when

seri-ally

cultured in media

containing

either 11.1

-i 0 ₅₀₀₀

r-E 4000

-Z

z

3000-1000

-70

-*- 11.1 mMGLUCOSE - 0.8mMGLUCOSE

I

.z

F_i

75 80 85 90 95 100 PASSAGE

Figure 2.Media insulinconcentrationsdeterminedafter 48 h of

cul-turein RPMI-1640containing eitherlowglucose(0.8 mM)orhigh

glucose ( 1.1mM)concentrationsinserial passagesofHIT cells. Data

arei±SEfortriplicate determinations.

passage

79,

and then declinedgraduallyovertime, remaining stableat

400-1,000

I.U/million

cells frompassage 81to pas-sage95.

Insulin secretion

during

staticincubation and

during

perifu-sion.Glucose concentration-insulin responsecurves were

de-termined

using

selected passages

serially

cultured in media containing either 0.8or 11.1 mMglucose.Cells cultured in 0.8 mMglucoseweresubculturedin0.8mM

glucose

for2dorin

11.1 mMglucose for 2orfor 9d

before

secretion

experiments

wereperformedtostimulate insulingene

transcription

and in-sulin

synthesis (35).

HIT cells from passages 96-97

serially

passed in media

containing

11.1 mMglucose failedtosecrete

insulin

inresponse to

glucose during

1-hstatic incubations.In

striking

contrast, latepassages ofHIT cells

serially

passed

in

media containing

0.8mM

glucose

demonstratedtwo- to three-fold

insulin

responses to

glucose

(Fig. 3),

a

magnitude usually

found in passage 70 cultured in mediacontaining 11.1 mM

glucose (30).

Phasic

insulin

secretion in responseto

glucose

was

exam-ined during

perifusion

with 11.1 mM

glucose

and 0.1 mM IBMX. Passages 95-98 ofHITcells

serially

passed in

media

containing

11.1mMglucose hadno

phasic

insulinresponses to

glucose. However, thesamelatepassagesof HIT cells that had been

serially

passed in media

containing

0.8 mMglucose and then subcultured in 11.1 mM

glucose

demonstratedfirst and second phase insulin

responsivity during

glucose

perifusion

(Fig.

4) thatwassimilar in

magnitude

tothat observed inearly

passagescultured inmedia containing 11.1 mMglucose (32).

Comparable

insulin

responsiveness

wasobserved wheneither 2

or9 dwereusedfor

subculturing

in11.1mMglucose after cells

had been serially passed under low glucose conditions.

Insulin

content

and

insulin mRNA. HIT cell passages 70,

80, 90, 100, 1

10,

and 130 cultured inmediacontaining 1

1.1

mMglucoseshowed a progressive diminution of insulin con-tent(insulincontent= 2468, 1730, 629, 522, 379, and 68

jU/

mgprotein, respectively). Insulin content was greater in pas-sages87, 95,and96 when HIT cells were serially cultured in

mediacontaining0.8 mM rather than 1 1.1 mM glucose (Fig.

5). Correspondingly,

insulinmRNAin laterpassagesofHIT cellsculturedinmediacontaining11.1 mMglucose was

mark-edly diminished(Fig. 6), whereas insulin mRNA of HIT cells

serially

cultured in media containing 0.8 mM glucose was greater compared to cells cultured in media containing 1 1.1

0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0

[GLUCOSE], mM

Figure 3.Insulinresponsestoincreasing glucose concentrationsin 1 h

static incubationsof late passagesofHITcellsserially passedinmedia containingeither 0.8 mMglucoseor11.1mMglucose.Before these

studies,cellsserially passedinmediacontaining0.8 mMglucosewere culturedinmediacontaining 1.1mMglucosefor2or9 d. Dataare

x±SE forduplicateexperiments foreach passageusingeach condi-tion.Inset compares basal andglucose-stimulatedinsulinlevels in P-97 cells cultured in 11.1 mMglucose (A),in0.8 mMglucose(B),

and in 0.8 mMglucose plus9 dsubculture in11.1 mMglucose (C).

Both(B) and (C)conditionswereassociatedwithapproximately

two-tothreefoldinsulin releasebyglucoseduringthestaticstimulations,

whereas(A) was not.

mMglucose

(Figs.

5 and7). The

experiments

inwhich insulin

contentandmRNA wereobserveddidnotinvolve subcultur-ing under high glucose conditions but ratherwerecarriedout

entirely

under0.8 mM

glucose

conditions.

Southern

analysis of

insulin gene and

Western

analysis of

GLUT-2. Examination of

HIT cells

from early

and late pas-sagescultured in

media

containing

either 11.1 mMor0.8mM

0

0

0

E

Z,-z -Z

z

60

50

40

30

20

10

0 1 0 20 30 40

MINUTES

Figure 4. Phasic insulin secretion during perifusion with glucose and

IBMXfrom late passages of HIT cells serially passed in media

con-tainingeither 0.8mMglucoseor

11.

1 mMglucose. Before these

studies,cellsserially passed in media containing 0.8 mM glucose were cultured in mediacontaining 1.1 mMglucose for 2 or 9 d. Data are

x±SEfor duplicate experiments for each passage using each condition.

18S

r-.2 2

CL co E

M=L

z

LU

z 0 0

-j

:3 W z

z E Un z C: z

87 95 96 87 95 96 PASSAGE

Figure 5.Comparison of insulincontentandinsulinmRNAlevelsof

HITcellsserially passedin RPMImediacontainingeither11.1 mM glucose or0.8 mM glucose. mRNAdensitometricdataareexpressed aninsulinmRNA/ I 8S rRNA ratio. Dataare meansof triplicate

ex-periments for insulincontentandduplicate experiments for insulin

mRNAusingeachcondition.

glucose

failed

to reveal

differences

in

insulin

gene

fragments

(Fig. 8) or levels of GLUT-2

(Fig.

9).

Discussion

The HIT cell has proven to beavaluable

experimental

model

ofbeta cellfunction. Insulin secretion from HIT cells is glucose concentration-related and insulin release

during perifusion

is

biphasic

(23-31). However, disappointing

loss ofinsulin secre-tionfromHITcells with serial passage has beena

major

short-coming

(23,

26, 30).

We

considered

the

possibility

thatan 11.1

mM

concentration

of

glucose,

which has been

routinely

usedin

media for

culturing

these cells

(23-31)

and

is supramaximal

for HIT cell insulin

secretion (30), might

be

responsible

for this

effect.

We

found

that late passages

of

HIT cells cultured in

media containing

0.8mM

glucose

demonstrated normal insu-lin

responsivity

toglucose

in static incubations

and normal first and second phase

glucose-induced

insulin secretion

during

per-ifusion. Incontrast, late passages cultured

with

11.1 mM

glu-cose had no insulin responses to

glucose

stimulation as has

been

previously reported (23,

26,

30, 32).

HITcellsaslateas

passage 96 cultured in media

containing

0.8 mM

glucose

had

levels ofinsulin and insulin mRNA that were greater than

thosefound in late passages of cells cultured in 1 1.1 mM

glu-cose. Because

culturing

under conditions of low

glucose

con-centration

caused greater numbers

of

cellstodetach

from

the

walls

of

the culture

flasks,

itwas

necessary

toseed greater

num-Figure 7. Decrements in HIT cell insulin mRNA

in passages87,95and,

&1w £** 96when cultured in lI1.1 mM glucose_and

prevention of

decre-Passage 75 96 96 95 9 5 87 87 mentsby culturingin mM

GLUC

11.1 11.1 0.8 11.1 0.8 11.1 0.8 0.8mMglucose.

bers of cells in earlierpassagesin orderto

harvest comparable

numbers of attached cells

by

the end

of

eachpassage.

Nonethe-less,

cell

population

doubling times and doublinglevelswere

similarata

given glucose

concentrationatpassages75,85,and

95, although

with

increasing

passagedoubling times decreased

comparably

for both

glucose

concentrations.

No

insulin

gene

mutation or

decreased

levels

of

GLUT-2, the functional

glu-cose transporter of

pancreatic

beta

cells,

were

found

in late

passagesof HIT cells

conventionally

cultured with media

con-taining

1

1.1

mM

glucose. Thus,

ourdata

uniquely indicate

that loss of beta cell

function

in HIT

cells

passedserially in

media

containing

a

high glucose

_{concentration is caused by}

prevent-able effects of

chronically culturing

under conditions of

high

glucose

concentrations thatareassociated with loss of

insulin

mRNA and insulincontentand,consequently,lossof

insulin

secretion.

Loss of insulin responses to

glucose

in HIT cells

serially

passed

in media

containing high

glucose concentrations is remi-niscent of abnormal insulin secretion intype IIdiabetic

pa-tients and in rodents that have been made permanently

hyperg-lycemic by

abriefexposuretoexogenouslyinduced high

circu-lating glucose

levels.

Patients

withtypeIIdiabetesmellituslose

first

phase

insulinresponsestoglucose if their fasting

glucose

levelsaregreaterthan 1 5 mg/dl (41). However, thissecretory

defect is

glucose-specific,

since first phase insulinresponsesto

EcoRI PstI HindM BamHI

-14.2 kb

- 8.5 kb

-3.7kb

28

S-18

S-P-71 80 87 98 111 131

Figure 6.Decreasing

levelsof insulin mRNA

withincreasingpassage ofHITcellsserially

-4.7 kb passedinmedia

con-taining 1.1 mM

glu-cose. Arrowdesignates

-1.9 kb hybridization of

oligola-beled humaninsulin ge-nomicDNAprobeto

HITcellinsulin mRNA.

-0.5 kb Hybridizationofprobe

to18SrRNAwas

pro-portionaltototalRNA.

-2.3kb

-1.3kb

Figure8.Restriction enzymeanalysis for insulingenemutationsin

earlyandlatepassagesof HIT cells cultured in RPMI media contain-ing 0.8or 1.1 mMglucose.Thefourrestriction endonucleasesused

areshownaboveeachsetof three lanes. No differences inpatterns were observed.

abilk

-.- .,:---W

Figure 9.Analysis of GLUT-2 levelsby

im-munoblot in early and

late passagesof HIT

55

kD

-

cells cultured in

RPMI

media containing 0.8

or 11.1 mMglucose;

Passage

74

93

93

100 mcgtotal protein

mM GLUC

11.1 11.1 0.8 wasusedineach lane.

non-glucoseagonistssuchasisoproterenol(7), secretin(8), argi-nine (9), and glucagon (10) remain intact in hyperglycemic

diabetic patientseventhough theyaresomewhat diminished. An interesting parallelism exists between these observations

and our previously published report that HIT cells serially passed in media containing high glucose concentrations lose insulinresponsestoglucose before they lose insulinresponses toarginine (30). It is importantto notethat lowering of circu-lating glucose levels into thenormalrangehasbeen shownto

partiallyrestorefirstphase insulinresponsestoglucose in dia-beticpatients (1-5). Likewise, briefintravenousinfusionswith agentssuchasphentolamine (42), sodium salicylate (43), and naloxone (44) have also been shown to partiallyrestore first phase insulinresponsestoglucose despite continued hypergly-cemia. The ability ofthese drugstorestoreresponsesintheface

ofcontinued hyperglycemiasuggeststhepossibilitythat endog-enous a-adrenergic tone, prostaglandins, or endorphins may mediate someof the adverse effects of chronic hyperglycemia onbeta cell function.

The role ofestablished hyperglycemia inthedevelopment of sustained abnormalities in insulin secretion has also been

examined inanimal models. Forexample, Rossetti et al.

in-ducedhyperglycemiaanddefective insulin secretionin ratsby 90% pancreatectomyandreportedthatnormalization of glyce-miabyphlorizintreatmentof thepancreatectomizedanimals restored normal insulin secretion (10). Leahyet al. reported thatratsundergoing 60% pancreatectomy developed persistent

hyperglycemia and abnormal insulin secretion only if they weretemporarilyfed highconcentrationsofsucrose( 12). Ima-muraet al. made similar observations in 80%

pancreatecto-mizeddogs (1 1). However, although studies in animalshave made the point that artificially induced hyperglycemia can causedefective beta cell function, such studiesareintrinsically limited by the short duration over which they can be

con-ducted. SincetypeIIdiabetes mellitus in humans evolvesover

amuchlonger period of time than several weeks,alaboratory model thatcanbestudiedover manymonthsor yearshas clear

advantages in examiningthedimensionof time and precisely chosenglucoseconcentrationsasmajor variablestoassessthe

concept ofglucosetoxicity onthe islet. Wethereforepropose

theHITcellas avaluable laboratorymodelfor examining glu-cosetoxicity of beta cells because putative biochemical

mecha-nisms responsible for this phenomenon and maneuvers

de-signedtosubvert its mechanismscaneasilybeexaminedover

manymonthsandpotentiallyyearsusing thiscell line.Inthis

context,ourfinding of undetectablelevels of insulinmRNAin late HIT cellpassagesthatcanbe prevented by culturingcells

serially in a low glucose concentration points specifically to-wardpotentialgenetic mechanismsof action forglucose

toxic-ity onbetacells, whichinturnleadstodefective insulin

secre-tion.Onepotentialproposal would involveaninhibitoryeffect

ofglucoseoninsulingene

transcription.

This would representa

paradoxical effect

because under normal

conditions,

glucose is

a

physiologic

stimulator

of

insulingene

transcription (35, 45).

Should it be foundthat

prolonged

exposureto

excessively high

glucoselevelscan cause

progressively diminished

ratesof insu-lin gene

transcription,

this would

provide

a

possible

mecha-nism for the decreased insulin mRNA levels and loss

of

insulin

secretion

we have

observed

in HITcells.

Alternatively,

deleteri-ous

effects

on insulin mRNA

stability

and

degradation

rates

needto be

considered,

and

although

our

restriction fragment

analysis studies revealed

no

abnormalities

in the

insulin

genein

late passages

serially

cultured in

high

glucose

concentrations,

this

analytical approach

to

possible

gene

mutations

can notbe

considered definitive. Nonetheless,

our

finding of preventable

decrements in insulin mRNA

associated

with

culturing

under highglucose

conditions points

to

preexocytotic

eventsfor the search for

mechanisms

of

deleterious

effects

of chronic

hyper-glycemia

onbeta cell

function

in typeII

diabetic

patients.

Such

effects

may

contribute

tophenomena thatare

commonly

re-ferredto asglucose

toxicity, glucose desensitization,

orbeta cell

exhaustion.

Acknowledgments

TheexcellenttechnicalassistanceofElizabethOseid,Laurie Pohlman, andBeryl Greenberg, and thesuperbsecretarial skills of Paula Rossin

aregratefully acknowledged.

This workwassupportedby grant

ROl-DK

38325 fromthe Na-tional Institutes of Health.

References

1.Turner, R.C., S.T.McCarthy,R. R.Holman,and E.Harris. 1976. Beta-cellfunction improvedbysupplementingbasalinsulin secretioninmilddiabetes. Br.Med.J. 1:1252-1254.

2.Schmeltz,R., H. J.Wendorff,and J. B. Field. 1978. Effect of control of

bloodglucoseonplasma insulinresponsestovarious stimuliinsecondaryfailures

tooralhypoglycemicagents and innewly diagnosed, maturityonset,

ketosis-re-sistantdiabetics.J.Clin.Endocrinol.Metab. 46:519-527.

3. Kosaka, K., T. Kuzuya, Y. Akanuma, and R. Hagura. 1980.Increase in

insulinresponseaftertreatmentofovertmaturity-onset diabetes isindependent

ofthe modeoftreatment.Diabetologia. 18:23-28.

4.Hidaka,H., M. Nagulesparan, I. Klimes, R. Clark, H.Sasaki,S. L.Aronoff,

B.Vasquez,A.H.Rubenstein,and R. H. Unger. 1982. Improvementofinsulin

secretionbutnotinsulin resistance aftershort term controlofplasma glucose in obeseType IIdiabetics.J.Clin.Endocrinol.Metab. 54:217-222.

5.Vague,P.,andJ.-P.Moulin. 1982. The defectiveglucose sensitivityof the B cell innoninsulin dependent diabetes.Improvement after twenty hoursof

nor-moglycaemia. Metab.Clin. Exp. 31:139-142.

6. Bonner-Weir, S.,D. F.Trent, andG.C. Weir. 1983.Partial pancreatec-tomy in theratandsubsequentdefectinglucose-inducedinsulin release. J. Clin. Invest.71:1544-1553.

7.Unger,R.H.,andS.Grundy.1985.Hyperglycaemiaasan inducer as well

as aconsequenceofimpaired islet cell function and insulin resistance: implica-tions for the management of diabetes.Diabetologia.28:119-121.

8.Giugliano,D., P. Di Pinto,A.Ceriello, G. Paolisso, F. Saccomanno, R. Torella, and F. D'Onofrio. 1985. Glycemic control with an artificial pancreas

improvesinsulin responses to both oral andIV.glucose in nonobese noninsulin-dependent diabetic subjects. ActaDiabetol.Lat. 22:203-213.

9.Glaser, B., G. Leibovich,R.Nesher,S.Hartling,C. Binder, and E. Cerasi. 1988.Improved beta-cell function after intensive insulin treatment in severe non-insulin-dependent diabetes. Endocrinologica.118:365-373.

10.Rossetti, L., G.I. Shulman, W. Zawalich, and R. A. DeFronzo. 1987. Effect of chronichyperglycemiaon in vivo insulin secretion in partially

pancre-atectomizedrats.J.Clin. Invest. 80:1037-1044.

11.Imamura, T., M.Koffler,J. H. Helderman, D. Prince, R. Thirlby, L. Inman, andR.H.Unger. 1988. Severe diabetes induced in subtotally depancrea-tized dogs bysustainedhyperglycemia. Diabetes.37:600-609.

12.Leahy, J. L., S. Bonner-Weir, and G. C. Weir. 1988. Minimal chronic hyperglycemia is a critical determinant of impaired insulin secretion after an incomplete pancreatectomy.J.Clin. Invest. 81:1407-1414.

13.Giroix, M.-H., P.Serradas,and B.Portha. 1989. Thedesensitizationof normal B-cells to glucose invitro is transientand not related tohighglucose levels.Endocrinology. 125:1999-2007.

14._{Grodsky, G. M. 1989. A new phase of insulin secretion. How will it}

contributeto our understandingof(f-ceflfunction? Diabetes.38:673-678. 15.Robertson, R. P. 1989. Type II diabetes, glucose "non-sense," andislet desensitization. Diabetes.38:1501-1505.

16. Rossetti, L., A. Giaccari, and R. A. DeFronzo. 1990. Glucose toxicity.

DiabetesCare. 13:610-630.

17. Korsgren, O., L.Jansson,S.Sandier,and A.Andersson.1990.

Hyperglyce-mia-inducedBcelltoxicity.Thefate ofpancreaticisletstransplanted into diabetic miceis dependent on theirgeneticbackground. J. Clin. Invest.86:2161-2168.

18. Sako, Y., and V. E. Grill. 1990. Coupling ofa-celldesensitization by

hyperglycemiatoexcessive stimulationandcirculating insulinin glucose-infused rats.Diabetes. 39:1580-1583.

19. Robertson, R. P., and D. Porte. 1973. The glucosereceptor.a defective mechanism indiabetes mellitus distinct fromthe beta adrenergic receptor. J. Clin. Invest. 52:870-876.

20. Lerner, R. L. 1979. Augmentedinsulinresponses to glucose after secretin priming indiabetic subjects.J.Clin. Endocrinol. Metab. 48:462-466.

21. Palmer, J. P., J. W. Benson, R. M. Walter, and J. W.Ensinck. 1976.

Arginine-stimulatedacute phaseof insulin and glucagon secretion in diabetic subjects. J. Clin. Invest. 58:565-570.

22.Crockford,P.M., W. R. Hazzard, and R. H. Williams. 1969. Insulin response to glucagon: theopposing effects of diabetesandobesity. Diabetes. 18:216-224.

23.Santerre,R.F., R. A.Cook, R.M. D.Crisel,J. D. Sharp, R. J.Schmidt,

D.C.Williams,and C. P.Wilson. 1981.Insulinsynthesisin aclonalcellline of simian virus 40-transformedhamster pancreatic beta cells. Proc.Natl. Acad.Sci. USA.78:43394343.

24.Swope, S.L., and A.Schonbrunn. 1984. Bombesin stimulatesinsulin

secretion byapancreaticisletcellline.Proc.Natl.Acad. Sci. USA. 81:1822-1826.

25.Hill,R.S., and A. E. BoydIII. 1985. Perifusion ofaclonal cellline of simianvirus40-transformedbeta cell. Insulinsecretory dynamics inresponseto

glucose,3-isobutyl- I-methylxanthine,andpotassium.Diabetes.34:115-120. 26.Ashcroft, S. J. H., P. Hammonds,and D. E.Harrison. 1986. Insulin secretoryresponsesofaclonal cellline ofsimianvirus 40-transformedB cells.

Diabetologia.29:727-733.

27. Lambert, D.G.,K.Hughes,and T. W.Atkins.1986.Insulinreleasefroma clonedhamster B-cell line(HIT-T15).Theeffects of glucose, amino acids, sul-phonylureas andcolchicine. Biochem. Biophys.Res.Commun.140:616-625.

28.Meglasson,M.D.,C.D.Manning,H.Najafi,and F. M.Matschinsky.

1987.Fuel-stimulated insulin secretionbyclonal hamster,-cell lineHITT-15.

Diabetes. 36:477484.

29. Robertson, R. P., P. Tsai, S. A. Little, H.-J. Zhang, and T. F. Walseth. 1987.Receptor-mediatedadenylatecyclase-coupledmechanism forPGE2

inhibi-tionofinsulin secretionin HIT cells.Diabetes.36:1047-1053.

30.Zhang, H.-J.,T.F. Walseth, and R. P. Robertson. 1989. Insulinsecretion

and cAMP metabolism in HIT cells._Reciprocaland serial_{passage-dependent}

relationships.Diabetes.38:44-48.

31.Ullrich,S.,M.Prentki,andC. B.Wollheim.1990.Somatostatin inhibi-tion of

Ca2'-induced

insulin secretion inpermeabilizedHIT-T15 cells. Biochem. J.270:273-276.

32.Seaquist,E.R.,T. F.Walseth,D. M.Nelson,and R. P. Robertson. 1989. Pertussis toxin-sensitive Gproteinmediation ofPGE2inhibition ofcAMP metab-olism andphasic glucose-induced insulin secretion in HIT cells. Diabetes.

38:1439-1445.

33.Chomczynski,P.,and N. Sacchi. 1987.Single-stepmethod of RNA isola-tionbyacidguanidinium thiocyanate-phenol-chloroform extraction. Anal. Bio-chem.162:156-159.

34.Feinberg,A.P.,and B.Vogelstein.1984. Atechniqueforradiolabeling

DNA restrictionendonuclease_fragmentsto_highspecific activity.Anal.Biochem. 137:266-267.

35.Hammonds, P.,P. N.Schofield,and S. J. H. Ashcroft. 1987. Glucose

regulatespreproinsulinmessenger RNA levels inaclonal cell line ofsimian virus

40-transformedB cells. FEBS(Fed.Eur. Biochem.Soc.)Lett.213:149-154.

36. deLeeuw,W. J. F.,P. E.Slagboom,andJ.Vijg. 1989.Quantitative

comparisonofmRNAlevels in mammalian tissues: 28S ribosomal RNA levelas anaccurateinternal control. Nucleic Acids Res. 17:10137-10138.

37.Maniatis, T.,E. F.Fritsch,andJ. Sambrook. 1989. Isolation

ofhigh-molec-ular-weightDNA from mammalian cells. In MolecularCloning.A_Laboratory Manual. ColdSpringHarborLaboratory,ColdSpring Harbor,NY. 9.14-9.24.

38.Southern,E. M. 1975. Detection ofspecificsequences among DNA frag-mentsseparatedby gel electrophoresis.J. Mol. Biol.98:503-517.

39.Walseth,T.F.,H.-J.Zhang,L.K.Olson,W. A.Schroeder,and R. P.

Robertson.1989. Increase inG.andcyclicAMPgenerationin HIT cells. Evi-dence that the 45-kDa a-subunit ofGshas greater functionalactivitythan the

52-kDa a-subunit. J. Biol. Chem. 264:21106-21111.

40.Harris,B.A.,J. D.Robishaw,S.M.Mumby,and A. G. Gilman. 1985. MolecularcloningofcomplementaryDNA forthealphasubunit oftheG-protein

that stimulatesadenylatecyclase.Science(Wash. DC).229:1274-1277.

41.Brunzell,J.D.,R. P.Robertson, R.L.Lerner,W. R.Hazzard,J. W. Ensinck,E. L.Bierman,andD.Porte,Jr. 1976.Relationshipsbetweenfasting plasma glucoselevels and insulin secretionduringintravenousglucosetolerance

tests.J.Clin. Endocrinol. Metab. 42:222-229.

42.Robertson,R.P.,J. B.Halter,and D.Porte,Jr. 1976. A role for alpha-adrenergicreceptors in abnormal insulin secretion in diabetes mellitus. J.Clin.

Invest.57:791-795.

43. Robertson,R.P., and M. Chen. 1977. A role for prostaglandin E in defective insulin secretion andcarbohydrateintolerance in diabetes mellitus.J.

Clin. Invest. 60:747-753.

44.Giugliano, D.,A.Ceriello,P. DiPinto,F.Saccomanno,S.Gentile,and F.

Cappiapuoti. 1982.Impairedinsulin secretion in human diabetes mellitus: the effect ofnaloxone-inducedopiatereceptor blockade. Diabetes.31:367-370.

45.German,M.S.,L. G.Moss,and W. J. Rutter. 1990.Regulationofinsulin geneexpression by glucoseand calcium in transfectedprimaryislet cultures. J.