Multipotential phenotypic expression of genes encoding peptide hormones in rat insulinoma cell lines

Multipotential phenotypic expression of genes

encoding peptide hormones in rat insulinoma

cell lines.

J Philippe, … , W L Chick, J F Habener

J Clin Invest.

1987;

79(2)

:351-358.

https://doi.org/10.1172/JCI112819

.

The developmental origin of the four phenotypically distinct hormone-producing islet cells

(insulin, glucagon, somatostatin, pancreatic polypeptide) is unclear. To investigate the

potential for phenotypic differentiation of islet cells, we prepared several clonal cell lines

from a radiation-induced rat islet tumor and analyzed them for insulin, glucagon, and

somatostatin gene expression by cDNA hybridization, immunocytochemistry, and

radioimmunoassay. We found expression of all three genes in the tumor and in the parental

cell line and mixed variable phenotypes in the clonal lines derived from the parental line.

We also observed the ectopic expression of the angiotensinogen gene in the tumor and the

cell lines. The relative levels of hormonal gene expression differed among the cell lines but

remained fixed during continuous passage. The three islet hormone mRNAs were larger

compared to the pancreas owing to longer poly(A) tracts. These observations indicate that

neoplastic islet cells retain the potential to differentiate into hormone-specific cellular

phenotypes and may mimic developmental pathways of the pancreatic islets.

Research Article

Find the latest version:

Multipotential Phenotypic Expression

of Genes

Encoding

Peptide Hormones

in Rat

Insulinoma

Cell

Lines

Jacques Philippe,*William L.Chick,§and Joel F.Habenert

*Laboratory ofMolecular Endocrinology,Massachusetts GeneralHospitaland tHowardHughesMedicalInstitute, HarvardMedicalSchool, Boston,Massachusetts02114; §Departments ofBiochemistryandMedicine,

UniversityofMassachusetts MedicalSchool, Worcester, Massachusetts 01605

Abstract

The developmental origin of the four

phenotypically

distinct

hormone-producing islet cells (insulin, glucagon,somatostatin, pancreatic polypeptide)isunclear. Toinvestigatethe potential forphenotypicdifferentiation of isletcells,wepreparedseveral clonal cell lines from a radiation-induced rat islet tumor and analyzed them for insulin, glucagon, and somatostatin gene expression by cDNAhybridization,

immunocytochemistry,

and

radioimmunoassay. We found expression ofallthree genes in the tumor and in theparental cellline andmixed variable

phe-notypesinthe clonallinesderived from the parental line. We also observedtheectopicexpressionof the

angiotensinogen

gene in the tumor and the celllines. The relative levels ofhormonal geneexpression differedamong thecell linesbutremained fixed

during continuouspassage. The threeislethormonemRNAswere larger compared to the pancreasowingtolongerpoly(A)tracts. These observationsindicatethatneoplastic isletcellsretain the

potentialtodifferentiate into

hormone-specific

cellular

pheno-types and maymimic developmental pathways ofthepancreatic islets.

Introduction

Inanumberoftissues

terminally

differentiated cellsappearto

arise from

progenitor

stemcells;thedevelopment of the lympho-hematopoieticsystemisaninformative illustration of this con-cept(1). The

origin

and

multipotentiality

ofthe

hormone-pro-ducing

islet cells oftheendocrinepancreashas generatedmuch

discussion. Several hypotheseshavebeen advancedtodelineate

the embryologic origins ofthefour

peptide-producing

cells of the pancreas

(insulin,

glucagon, somatostatin, and pancreatic polypeptide). Current evidence implicatesamultipotent

endo-dermal stem cell, capable of

differentiation

into enterocytes, ducts,acinar,andendocrine cells (2). Alternatively, Pearse

pos-tulatedaneuralorigin of thesecells on thebasis ofthe

APUD'

concept(amineprecursor uptake anddecarboxylation)(3).This conceptderivesfromtheidentificationofcells present in different

Addressreprintrequests to Dr.Philippe.

Received for publication21May1986and in revised form 12 Sep-tember 1986.

1.Abbreviationsused in this paper: APUD, amineprecursoruptake and

decarboxylation;DMEM, Dulbecco's minimalessential medium; DTT, dithiothreitol;SSC,0.15MNaCl/0.015 M Na citrate.

organs whoseprimary function is thesynthesisand release of

polypeptide hormones,but which also have the capacity to syn-thesize

biogenic

amines

(3). Accumulating

data suggestsan en-dodermalorigin of theislet cells andindicates that neural-like

characteristicscanbe shared

by

cellsfrom different

embryological

origins.Forinstance,recentstudiesbyTeitelman(4)have dem-onstrated theexpression, in embryonic mice, ofthe

catechol-amine-synthesizingenzyme,tyrosine hydrolylase,which appears inapopulation ofcellsin the dorsal pancreasatday 11 of ges-tation.By days12to 14, glucagonandtyrosinehydroxylaseare

colocalized in identical cells in upto40% of the cells

containing

eitherprotein.Insulin replaces glucagon in some of thetyrosine hydroxylase-positive cells at day 14. The catecholaminergic phenotype istransient, however, and ceases after thefifteenth embryonic day (4, 5).The adult state of the islet cells is char-acterizedbythe presence ofdopadecarboxylase whichconverts

dopatodopamine (6, 7).A common ancestor for the four

pep-tide-producingcells of the pancreas appears to be a likely

pos-sibility.

Wewished to test this hypothesis of themultipotentialityof the islet cellsbyanalyzing the expression of several genes

en-coding polypeptidehormones in clonal and subclonal cell lines derived from a transplantable rat insulinoma (8, 9). These cells have thecapacitytosynthesize biogenicamines because of their

highcontentofL-dopa decarboxylase (10).Inaddition,because these insulinoma cell lines are transformed cells they arelikely

tobe partially dedifferentiated,and thus be capable ofphenotypic switchingin the expression of the hormone genes. We show here thatatumorderived from a predominantly somatostatin-pro-ducingcellline,first cloned as an individual cell, expresses the insulin, glucagon, andangiotensinogengenes, as well as retaining its somatostatinphenotype. Furthermore, analysesof messenger RNA and hormone synthesis byimmunocytochemistryand

ra-dioimmunoassaysby clonal cell lines derived from the

trans-planted tumor, as well as unrelated cell lines isolated indepen-dently,support the conclusion that a single, transformed islet cell has the potentiality ofdifferentiatinginto cellsexpressingat least three of the islet hormones.

Methods

Tumorand origins of cell lines. The lineage of the tumor and the cell linesanalyzed in thisstudyisillustratedinFig. 1. The original ratislet celltumor arosefollowing 1000radoftotalbodyx-irradiationto ainbred

albino NEDHrat(8).It wasthen developed into a seriallytransplantable

tumorinNEDH rats. Individual cells were subsequently cloned from

the parent cell line (RIN-r), derived from thisserially transplantable tumorline (9). Some representativeclones (1046-38, 1046-44) as well as aseparatechemically inducedsolid tumor (S2181) (14) werefurther analyzed in thisstudy. Anadditionalcellline (1027-B2),expressing

pre-dominantly, somatostatin,was injected subcutaneously in four NEDH rats.3wklater,thetumors (T1056) wereremovedandeither used for J.Clin. Invest.

©DTheAmerican Society forClinical Investigation, Inc. 0021-9738/87/02/0351/08 $1.00

X-ray induced Islet cell tumour

passaged Dy

subcutaneous

transplantation

Parentcell line (RIN-r)

1046-44 1046-38 1027-B2 1046-43 no peptide insulin somatostatin insulin

somatostati n

Tumour 1056 Parent cell line 1056 insulin, glucagon, sooatostatin

1056-A 1056-B 1056-C 1056-D 1056-E 1056-F Glucagon Insulin Soiuatostatin Somatostatin Insulin Insulin Somatostatin Glucagon Somatostatin

Insulin Sonatostaticn

Figure1. Lineage of the islet tumors and cell lines starting from the

originalx-ray induced islet cell tumor. Hormones secreted by each clonal cell line, as measured in the medium by radioimmunoassay, and theidentification number of the specific clones used in the study

areindicated.

RNAextraction or to establish continuous cell lines (1056). Individual cells were, eventually,clonedbylimiting dilutions from the parent 1056 cell line and analyzed forexpression of specific mRNAs and peptide synthesis.

Cellcultureand tumor propagation. The celllinesused in these studies were derived from a continuous islet cell line, RIN-r, which was estab-lishedfrom a transplantable rat islet cell tumor (9). For the present ex-periments, the cells weremaintained in Dulbecco's minimal essential medium (DMEM) (Gibco Laboratories, GrandIsland,NY) at a glucose concentration of4,500 mg/liter, andsupplemented with 10% heat-in-activated fetal bovineserum(GibcoLaboratories), 100 U/ml of penicillin and 100

_gg/ml

ofstreptomycin.Incubations were carried out at 370C

in 95% air:5%CO2.Celllines 1027-B2, 1046-44, and1046-38were cloned from the parent cell line and propagatedtomassculture.

Clone1027-B2, establishedas asomatostatin-producingcellline,was

injected subcutaneously (-4 X 107cells) into the interscapularareaof four albino inbred NEDHrats.20 dlater,tumors wereremoved

asep-tically, minced,andtrypsinizedin ordertodevelopcontinuous celllines, orfrozen inliquid nitrogenand storedat-700C forRNAextraction. Fibroblast cellswereeliminatedby transferring freshly trypsinizedcells

to newdishes.

RNAextractionandNorthernblotanalysis.TotalRNA wasextracted

fromsolidtumorsand cellsbyhomogenizationinguanidine thiocyanate

and sedimentation through a cesium chloride cushion (12). Poly A' RNA wasisolatedbyoligodTcellulosechromatography (13). 20 ,Ag of total RNA from eachsamplewassize fractionatedon a1.4%agarosegel

after denaturation in glyoxal followed by electrotransfer to anylon

membrane (Nytran, Schleicher and Schuell, Keene, NH).Blots were

baked for 2hat80°Cunder vacuum,prehybridized in 1 MNaCl/1%

SDS/10%Dextransulfateat50°Covernightandhybridizedatthesame

temperature for 24hafteraddition of the labeledprobes (3-5 X I05

cpm/ml); theywerethenwashedat55°Ctwice inIXSSC(0.15M_NaCl/

0.015MNacitrate)/l%sodiumdodecylsulfate(SDS)andexposedfor varying timesat-70°C withanintensifyingscreen.

Preparation andradiolabeling

ofoligonucleotides. Oligonucleotides

from 26to 43bases,complementarytodifferentregionsof each ofthe

messenger (m)RNAs, (sixeachforglucagon,somatostatin,and

angio-tensinogenandfour forinsulin)were_{synthesized by}anautomated oli-gonucleotide synthesizer(Applied Biosystems, FosterCity, CA). The cDNAs were 5'end-labeledwith [r32P]ATP by T4-polynrucleotide kinase (Bethesda ResearchLaboratories, Gaithersburg, MD)toa

_{specific activity}

of1-2 x IO' cpm/Mug.

Primer extension assay. The primers were synthetic oligonucleotides cDNAs, as described above, and chosen to have biosynthetic extended fragments of < 250 bases in length; they were labeled as above. DNA-RNAhybridization was performed in a final volume of 10Ml containing 200 mMKC1, 100 mM Tris HCl (pH 8), 1 mM EDTA, 1 pmol of end-labeled primer and 25-50 Mg of total RNA. The reaction mixture was denatured at 1000C for 5 min and incubated at 50'C for 3 h. Extension was carried out in 25Ml of 80 mM KCl, 40 mMTris-HCl (pH 8.0), 1 mM EDTA, 8 mM dithiothreitol (DTT), 8 mMMgC12, 1 mM of each dNTP and 40 U of reverse transcriptase (Life Sciences, Inc., Bayport, NY) at 42°C for 1 h. The extended fragments were purified by phenol extraction, treated with 0.2 N NaOH at 65°C for 1 h and dried down, before they were run on a 5 or 20% (depending on the size of the expected readout) polyacrylamide gel containing 7 M urea.

RNAse Hdigestion ofpoly (A) tracts. 50 to 100,goftotal RNA were hybridized to 10,gof d(pT) 12-18 (Collaborative Research, Inc., Lex-ington, MA) in 100 mM KCl at 37°C for 30min,afterdenaturation at 65°C for 2 min; 2 U of RNAse H (Bethesda Research Laboratories) were then added in 80 mM KC1, 40 mM Tris-HCl pH 7.9, 5 mMMgC12, 1 mM DTT, 0.5 mg/ml bovine serum albumin (BSA) and the reaction was carried on for 30 min. The RNA was phenol-extracted, ethanol-precipitated, and run on agarose gels as described above.

Immunocytochemistry. RIN cells were grown on glass slides. Slides were rinsed in phosphate-buffered saline (PBS), then fixedfor20min in 4%paraformaldehyde in PBS, washed for 10 min in three consecutive batches of PBS, then dipped in water, air-dried, and stored at -80°C.

Serumcontaining biotinylated antibodies (dilution 1/100) was applied to the cells along with 2 drops of 100Mg/mlAvidin D in PBS; after 30

min,slides were drained and covered with primary antibodies(glucagon-,

insul, and somatostatspecific antibodies) diluted in PBS and in-cubated overnight at 4°C. After a 30-min wash in PBS, endogenous peroxidase was blocked in a 0.3% solution ofH202with 20Mug/ml of Biotin (in PBS) for 30min. The slides were then washed and incubated with the biotinylated second step antibody (Vectastain Kit, Vector Lab, Burlingame, CA) for 30min,washed again andincubated with the avidin-biotinylated horseradish peroxidase complex for 45 min. After a final wash, slides were soaked in the staining mixture (0.1 M sodium acetate, pH 5.0, containing 0.015% dimethyl formamide and 0.02% of H202, 0.5% formamide and 0.02% (wt/vol) 3-amino-9-ethylcarbazole) for 3 min, rinsed, and counterstained in hematoxylin for 1 min, dipped in saturated lithium carbonate and mounted in Glycergel (Dako Corp., Santa Barbara, CA). The antibodies were GLP-I (B-5) at a dilution of 1:500 for glucagon (14), insulin specific (from Ling Research, Latham, NY) at a dilution of 1:200 and somatostatin specific (from Immuno-nuclear, Stillwater, MN) at a dilution of 1:200. Control slides were treated similarly with antisera that had been preabsorbed with GLP-I, insulin, andsomatostatin and immunocytochemical analyses of rat pancreas were carried out in parallel.

Radioimmunoassay. Insulin, glucagon, and somatostatin were mea-sured in the cell culture medium byradioimmunoassays, aspreviously described (9, 14).

Results

Morphologyoftheisletcell lines. Despite their common origin, the morphologic features and thegrowth rate of the different cell lines differ markedly. AsshowninFig. 2A, 1027-B2,the

progenitor of 1056 (Fig.2B), grows in clusters with well-defined margins and only rare cytoplasmic processes extending from free cell surfaces; 1046-38 has similar morphological features

(Fig.2C). Clone 1046-44 growsasisolated cells untiltheyreach

confluence(Fig.2D).Thegrowthrateof1046-38 and 1046-44

clones isabouttwiceasfastasthatof1056and 1027-B2

(dou-bling time - 24h).

Incontrast, 1056 consists ofelongatedcells withcytoplasmic

Figure 2.Light photomicrograph of differentRINclonal cell lines in culture. (A) clone 1027-B2, (B) clone 1056, (C) clone 1046-38, (D) clone 1046-44 (X 80).

RIN-m,the first islet cell lines originally reported (9). The

pop-ulation doublingtime is 48 h (as measured by DNA content) compared to 52 h for RIN-r. Clones derived from 1056 had

similar morphologiccharacteristics to the parent cell line. The morphological features do not appear to correspond to any de-fined phenotype and thus seem independent of what the cells secrete.

Phenotypic expression ofthe tumor and cell lines.Expression

of the genescoding forinsulin, glucagon, and somatostatin was assessed by messenger RNA analysis and by measuring their translational products byimmunocytochemicalanalyses of the cells andradioimmunoassay of the culture medium. After

elec-trophoresisandtransfertoanylon membrane, total RNA was hybridized to synthetic oligonucleotide cDNA probes; four cDNAs werecomplimentary to the rat insulin genes, and six to theratglucagon gene. cDNAs matched each of the six exons of theglucagon gene and six cDNAs matched the two exons of the ratsomatostatin gene. In preliminary experiments, the cDNA probes were used one by one toascertaintheauthenticity ofthe mRNAanalyzed by demonstratingthat each cDNAindividually

hybridizedstronglyto thespecific mRNAs (data not shown). Wefoundmarkedvariability inthephenotypic expression ofthe tumor andcell lines derived fromclone 1027-B2(Fig.3). Twoadditional representativecell lines, previously cloned from

RIN-r(1046-38 and 104644) and a solid tumor with a mixed insulin-somatostatin phenotype (S2181) are also represented. Although as a cellline,clone 1027-B2 expressespredominantly somatostatin,once itformsasolid,subcutaneous tumor in the rat host it is capable ofdifferentiation and expresses all three

islethormonesstudied. Thecell linederived fromtumor1056 retains aphenotypicpattern ofgeneexpression that issimilar

tothatofthe tumor.

Oneofthe subclonal celllines,1056A, expressed the glucagon geneat aparticularly highleveland representsapredominantly

A-cellphenotype.Thiscelllineproduces large amounts of glu-cagon and theglucagon-like peptidesprocessed from proglucagon

(15).Thecloningof individualcells from the parent line, 1056,

providedseveralcontinuouscell lines, each expressingdifferent

hormone phenotypes (Fig. 3). These celllines have exhibited

stablephenotypic expression ofthe genes analyzed for -1 yr.

Secondary single cell cloning from 1056 A revealedthe same

qualitativeand

quantitative

variabilityofpolypeptidehormone geneexpression astheprimary cloning.

A

B2 T 5656A 56B565D 56E56F 44 38 81

A

-_G_

B

B2 T 56 56A568 56C

*bdE56F

44 38

jo.

*

I

s

I*

0

p

0

*

81

4.

I9

Figure 3. RNA analyses by Northern blotsofthedifferenttumor and celllines.Each lanecontains20 ugoftotal cellular RNA. Blots were

hybridized with oligonucleotidecDNA probescomplementaryto

insu-lin,glucagon, andangiotensinogenmRNAs (A).After washing offthe

aboveprobes the blotswererehybridized withcDNAs to

angiotensino-genandsomatostatin (B)(seeMethods). Blotswere exposedfor 12h, exceptlane 81 (2h). Apointstotheangiotensinogen mRNA, G to the

glucagonmRNA, I to theinsulinmRNA, and S to thestomatostatin

mRNA. B2: 1027-B2. T: tumor 1056:56:parental 1056 cell line. 56

A-56F: 1056A- 1056F. 38: 1046-38. 44: 1046-44. 81:S2181.

Ofthethree

representative

phenotypesof clones

previously

established (9), 1046-38illustrate theinsulinoma

phenotype (no

detectable expression ofthe glucagon andsomatostatin genes by mRNA

analysis

andonlyrare

glucagon-positive

cells

by

im-munocytochemistry), the amountof insulinmRNA

expressed

differingby60-foldwhencompared with1046-44. TumorS2181

represents themixed

insulin-somatostatin

phenotype,

expressing

roughlyequivalent levels of insulinandsomatostatin mRNAs and 6-foldas much asthe highest

insulin-producing

cell line

analyzed.

Immunocytochemical analyses ofthevarious clonalcell lines revealed that the expressionof detectablehormone

antigens

is

limitedtoselectedsubpopulations ofcells

(Fig.

4).

Approximately

10-20% of the cells arestrongly

immunopositive

and the re-mainder areimmunonegative; there do notappearto be any

intermediatecells. Further, the

immunopositive

cellsoccur in groups among abackground of nonhormone

producing

cells

suggestingtheexistence ofsomeform of

"organizer"

activities.

Withregardtotherelative abundancesanddistributions ofthe glucagon and

somatostatin-specific

cell

phenotypes,

theclonal

lines 1056A and 1027-B2appeartobe "mirror

images"

ofeach other inasmuchas1056Acontainsaratheruniform distribution of

glucagon-positive

cellswithrarenestsof

somatostatin-positive

cells and 1027-B2is

just

thereverseof this.Itisdifficultto

quan-titate theserelative distributionsbut in the 1056A and

1027-B2

lineseverymedium power field (X 100)has numerous glucagon

orsomatostatincells,respectively.The occurrenceofsmallnests

(-3-10cells)ofsomatostatin orglucagon cells appear on the average ofevery 10-12 fields. The presence of insulin-positive

cellsisrarein 1056A;onanentire slide onlyafewinsulincells wereseen.

Multipotentiality of TransformedIslet Cells 353

s

PI

'rop

C..

H

rin

_finAS"

. ,

I

%

_{_w.s}

A

1.

N. I

4"'4t. AV *... 41

Figure4.Immunocytochemical staining of rat islet cell lines. The var-ious celllineswere grown on glass slides and stained with antisera to either glucagon, insulin or somatostatin using the Avidin-Biotin-per-oxidasetechniqueasdescribed in Methods. Note the existence of a mixedheterogenityof hormone-specific cell phenotypes. The cells are either strongly immunopositiveorare immunonegative, the positive

cellsoccurin groups and constitute theminorityof the cell

popula-tions. A, I, D,G;cell line 1056A.B,E,F, cell line 1027-B2.H,I;cell line 1046-38.A-C,stained with antiserumtoglucagon-like peptideI;

D-F, stained with antiserumtosomatostatin, G-I,stained with

anti-serum toinsulin.Magnifications:A,B,D, E,H, X300.C,F, G, I,X

1200.

354 J.Philippe, W. L.Chick, andJ. F. Habener

It"

-V

4.

D

B

E

*.4

A.

S..

CF

_F

_I1

.3

:4,

.-V

Al.,

It is intriguing that these peculiar mixedphenotypesof

hor-mone-positive cellsappears toremain fixed inasmuch asthese precise patterns ofdistributions of the cells has remained un-changed over at least 3-4 mo of repeated immunocytochemical analyses. Furthermore, the relative amounts ofthe different hor-mone-positive cells correlates wellwiththe relative expression

of the genes as measured by both mRNA hybridization and

radioimmunoassays.

Translational productsofthe

specific

mRNAs were measured by radioimmunoassay ofthe culture medium and the results areexpressed per 106 cells/24 h (Table I).Thereisagood

cor-relationbetweentherelative abundanceofthe differentmRNAs

andhormoneconcentrations inthedifferent celllines. Acomparison of hormonegeneexpression between the tu-mor, the cell lines and rat adult pancreas were assessed from

mRNA and immunocytochemical analyses. The intensity of

immunocytochemical

stainingwasroughly equivalent (within

two- tothreefold)between positiveneoplastic and normalislet cells for all three hormones:insulin,glucagon,and somatostatin. The insulin mRNA content ofthe tumor 1056, the cell line 1056Band the adult pancreas were comparable. However, both glucagon andsomatostatinmRNAlevelswerehigher inthe tu-morand cell line 1056 A than in the pancreas.Consideringthat

islettissue accountsforonly 1% ofthe pancreasweight,these

observations indicate thatthefewpercentof the isletcells that expressthehormonegenes, express them atapproximatelythe samelevelsastheyareexpressed innormal ratislets.

Characterization ofthe mRNAs. Somewhatsurprisingly,we

observed that all three mRNAs coding for

insulin,

glucagon, andsomatostatin, isolated from eithertumor orcelllines, are larger by 100 to 300 baseswhen comparedwith the mRNAs extractedfromnormal rat pancreas. Wereasoned that the two most likely explanationstoaccount forthe larger sizes ofthe mRNAs wereutilization ofadifferentupstreamtranscriptional

startsite and/orthe presenceofalonger3'polyadenylationtract.

Weinvestigatedthe firstpossibility by

oligonucleotide-primed

reverse transcription of the

large

mRNAs using reverse

tran-scriptase and compared their transcriptional startpoints to those of the mRNAs in normal pancreas. Both insulin and somato-statinmRNAswereanalyzedin thismannerand the

transcrip-tionalstartsiteswereidentical in the cell lines and pancreas(Fig.

5 B and 5 C). We addressed the second hypothesis by

using

RNAse

H,

anendoribonucleasewhichspecifically degradesthe RNA portion ofan RNA-DNA hybrid. Thus, we hybridized syntheticpoly (dT) oligonucleotidestothepoly (A)tractsof the mRNAs extracted from both cell lines and pancreas before RNAse H digestion and analyzed the sizes of the resulting

mRNAsby Northern blot hybridizations (Fig. 6). The larger size of theglucagon, somatostatin, andinsulin mRNAs in the cell lines is entirely due to a longer en bloc segment of poly A because after RNAse digestion the mRNAs are reduced to a size identical

to the RNAse H treated pancreatic mRNAs (Fig. 6). It is of interest to note that the phenomenon of longer poly (A) tracts appears tobe specific for these mRNAs encoding secreted

pro-teins;wedidnotobserveanydifferencesin thesizesofmRNAs

encoding (3-actin and the c-myc protein in the cell linescompared

topancreas(data notshown).

Ectopic expression ofangiotensinogenin islet cell lines.

Be-cause tumors areknown to ectopically express hormones, we

used a mixture of cDNA oligonucleotides to a number of mRNAs encoding peptide hormones to screen total RNA

ex-tracted from the different cell lines for the expression of mRNAs encoding calcitonin, vasopressin, oxytocin, atrial natriuretic peptide, gastrin-releasing peptide, insulinlike growth factors I and II, proopiomelanocortin, growthhormone-releasing hor-mone,and angiotensinogen. We detected only angiotensinogen mRNA. To confirm thisfinding, weusedexon-specificprobes onebyone todemonstrate theauthenticityof themRNA;probes for all five exons hybridizes to the mRNA (data not shown). Theangiotensinogengene is expressed in every islet cell line that

TableI.Relative Levels ofImmunoreactive Hormones and mRNAs in Clonal Islet Cell Lines

Immunoreactivehorrones mRNA

Insulin Glucagon Somatostatin Insulin Glucagon Somatostatin (uU/106 cells) (pg/106 cells) (pg/106 cells) (RU) (RU) (RU)

1027-B2 63,300 (100) 120 (100)

1056 107(2.2) 283(31) 540 (1) 1 (2) 3(25) 2.5 (2)

1056-A 115(2.4) 854(100) 4,325 (7) 0.5(1) 12(100) 15 (12)

1056-B 504 (10.5) 5 (10)

1056-C - 630 (1) 1 (2) 0.3 (0.2)

1056-D - 959(1.5) 1 (0.8)

1056-E 1,712(40) 535 (62.6) 430(0.4) 26 (48) 6(50) 0.3 (0.2)

1056-F 4,368(91) 2,083 (1.8) 54 (100) 3 (2.4)

1046-44 109(2.2) 0.8(1.5)

1046-38 4,811 (100) 48 (89)

(Left)Immunoreactive insulin,glucagon, and somatostatin secretion from the different clonal cell line. Secretion per 106 cells/24 h. Relative

per-centof theamountof each peptideascompared to its highest equivalent (taken as 100%) is indicated in parentheses. (Right) Amount of insulin,

glucagon,andsomatostatin mRNA from the different clonal cell line, expressed in relativeODunit (RU), asquantitatedbydensitometric

scan-ning of autoradiograms.Relative percent of the amount of each mRNA as compared to its highest equivalent (taken as 100%) is indicated in

parentheses.Both RNA and medium come from the same plates and there is a 10-20% variation in the amounts of specific mRNA levels and peptidesfrom plateto plate. The sensitivity of theradioimmunoassaysis: 100

_,U/ml

for insulin, 50 pg/ml for somatostatin and 20 pg/ml for

c ₅₆ P

B

56 L 56 p

234

w-*

&i

-*--153 40

i

(17)

it alsouses the third site

(at

1800-1803

nucleotides),

as

foundby hybridizationwithanoligonucleotide complementary

to nucleotides 1794-1837. Inasmuch as the

angiotensinogen

mRNAexpressedinthe islet cells is stilllargerthan this latter componentby - 50 nucleotides, it probably uses the last

poly-adenylationsitesat 1839-1842 nucleotides.

Discussion

Figure S.Primer extension analysis of angiotensinogen mRNA from 1056 andliver (A) and of insulin andsomatostatinmRNAs from 1056 and pancreas (B and C,respectively).A synthetic oligonucleotide cDNAof 38 bases was hybridized to the angiotensinogen mRNA 234 bases 3'from the cap site and synthesis of the complementary DNA was allowed by addition of reverse transcriptase and dNTPs; the

ex-tension analysis was analogous for the insulin and somatostatin mRNAs, using a 32-base cDNA probe hybridizing 153 bases 3' and a 30 base cDNA probe hybridizing 40 bases 3' from the cap site, respec-tively. 25

_;&g

of liver, 1056 and pancreas total RNA wereused,except 50

,g

of pancreas RNA in (C). cDNAs were analyzed on a 5% (A and

B)and 20%(C) sequencing gel.

wehave analyzed so far (Fig. 3 and data not shown), albeit at

different

levels. By contrast, the angiotensinogen mRNA was

undetectable in the rat pancreas both by Northern analysis of poly

(A')

RNA and byS1 nucleaseanalysis(16).

Just as wefound forthe mRNAs encoding insulin, glucagon, andsomatostatin, the size of the angiotensinogen mRNA was larger compared to the normal livermRNA. Aprimer readout

study

indicated

identical transcriptionalstartsites forbothislet

cell andlivermRNA(Fig.5 A). Further treatmentofthe RNA

with additionalamountsof RNAse H failed to reduce the size

ofthe mRNAtothatofthelivermRNA treatedsimilarly,

in-dicating

thatthepoly(A) tract was not entirelyresponsible for

theincrease in size of the islet mRNA. Because the rat angio-tensinogengene hasfour polyadenylation sites (17),we

hypoth-esize that one of the distant sites might be usedduring

tran-scription. Although liver angiotensinogenmRNApredominantly

usesthesecond

polyadenylation

site(at1784-1786 nucleotides)

L L 38 385656 81 81 P P

+ +

..

+ +

Aes

P P9381881

'j4*

Figure 6. Northern blotof RNAse H-treated RNAfrom tumor and cell lines. Each control lanecontains20

;g

of total cellular RNA except lanePof (B) (2pgof polyA+RNA) (the

amountof RNA in the RNAse H var-ies from 5 to 16

_;sg).

Blots were hy-bridizedasdescribed in Fig. 3 and ex-posed for4h, except thefour first lanes (12 h) and thelasttwolanes (I h) of (B). 1056wastreatedwith so-dium butyrate 2mMfor72 h to

in-creaseinsulin steady state mRNA (39).+indicate RNAse H-treated sam-ples (see Methods) * points to

an-giotensinogen

mRNA-- _toglucagon

mRNA

.*

toinsulin mRNA and c*to

somatostatin mRNA. L, liver, 38, 1046-38; 56, 1056; 81, S3181;P.

pancreas.

Our results demonstrate the multipotential phenotypiccapability.

ofanindividual transformedratislet cell. Originally, toobtain

single phenotypecelllines, multipleclonesweredeveloped from RIN-r, the parental cell line, and analyzed in

regard

tothe

hor-monessecreted(9). Althoughmostof the clones secreted both insulin and somatostatin, a finding reported by others (10) and

alreadysuggesting themultipotentiality ofthesecells, insulin-and somatostatin-secreting cloneswereobtained,aswellascells which did not produce either of two hormones.

A somatostatin-secreting clone, 1027-B2,wasinjected sub-cutaneously inrats.Solidtumors of2-3 cmin sizedeveloped

atthesite of injection after 3 wk. Not unexpectedly, blood

glu-cose, checked on several occasions, remained within normal limits. Analysis of mRNA encoding insulin, glucagon, and

so-matostatin both from the tumor and the parent cell line, (1056) derived fromit,revealed expression of all three genes. Of interest was the fact that none of these three genes was highly expressed and that the relative abundance of the mRNAs, as measuredby

autoradiographic microdensitometry, was roughly equivalent.

Thus,nodominant phenotype emerged from the tumordespite

the characteristics of itsprogenitor.

Whenweanalyzed expression of thesamegenes in islet cell lines, obtained from single cloned cells, we observed a wide

va-rietyofdifferent phenotypes; from cells ofasingle phenotype (1046-38, 1056B) representing asmall minority of theclones,

tocellsexpressing two or all of the three genes (1056 A, E, F). The same multipotential capability, as seen in the in vivostudy,

canthusbe manifestedduringin vitro culture. Theimplication

ofthese results is that cloned islet cells have given rise to cultures (ortumors) containing mixtures of cells ofdifferent phenotypes, each expressing one hormone.

Wenot only selected these various cell clones to illustrate their different and multiple phenotypes, but also toemphasize

thelarge differences in theamounts of hormones synthesized by them. As an example, there was a 1 0-folddifference in insulin mRNA betweenthehighest and the lowest insulin-producing clones; we have also seen this variability in solidtumors. It ap-pears from theimmunocytochemical analysis that the variability inquantitative gene expressioncanbe accounted for,atleast in part, by the different relative numbers of hormone-expressing cells. Thetopographical arrangement of thehormone-producing

cells in the clonal lines isintriguing. Although onlyaminority (10-20%) of cells in any given cell line expressed a hormone geneasdetermined byimmunocytochemical analyses,the

im-munopositivecellswere notuniformly dispersed throughoutthe cell population but rather occurred in clusters. In the 1056A line the glucagon-expressing cells were dispersed abundantly

throughout the population of cells whereas somatostatin-ex-pressing cells appeared rarely in small nests of a few cells. This

topographyin the

1027-%

linewasjust the reverse; somatostatin cellsweredispersed throughoutandglucagoncellsappearedin

occasionalsmall

clusters,

yet the 1056A linewascloned from

356 J.Philippe, W. L.Chick,andJ. F.Habener

the 1027-B2 line. These observations clearly suggest that these

celllinesareexpressingsome kind of"organizer"principlesbe they either indirect diffusible factors or direct cell contact phe-nomena toexplain theexistence ofsuchmixed, fixedhormone phenotypes.Thequestion arisesregardingwhether or not agiven

cell expressed more than one hormone gene. Because of the marked differences in thetopographical arrangements of the

glucagon-andsomatostatin-expressingcells we can saywith

cer-tainty that in the 1056A lines there exist

glucagon-expressing

cells thatdo not express either somatostatin orinsulin, and in the 1027-B2 line most of the cellsexpressingsomatostatin do not expressglucagon.However, whetheror not the minor cell

populations, somatostatin in 1056A and

glucagon

in 1027-B2

expressglucagon, somatostatin (orinsulin),

respectively,

remains

unknown andmustawait the completionofcolocalization,dual

immunostaining of singlecells.

Theclinicalrelevanceofthetotipotency ofmalignantislet

cellsisillustrated by the

frequent

occurrenceof

pancreatic

islet

tumorsproducing several

peptide

hormones

(18).

Ametastatic tumor,

predominantly

containing

glucagon, hasevenbeen

iden-tifiedintheliver ofa

patient

witha

malignant

insulinoma(19).

It istempting to relate the multipotentiality of neoplastic

islet cells to the developmental pathway of differentiation of

normalislets. Several

lines of

evidence indicate that islet cells arise froma common stem cell, endodermal (2,

20)

in

origin.

Ourneoplasticcells mayhave revertedto a

primitive

precursor stateandretained the

potential

todifferentiate into the different isletcellphenotypes.This potentialappears to beexpressed dur-ingcell growth whetherin vivoor invitro. The factthat these

celllines derived from

single

clonedcellscan secretemultiple hormones,is illustrative of their

multipotentiality.

Itshouldnot

be

considered,

however,as a truedifferentiationevent asthese cellscanalwaysexpresstheir

multipotency.

Embryonic development proceeds from

multiple

successive

steps, during which cells

acquire

certain determinants and

be-come committed. A common endodermal

(20,

21)

stem cell

maygive risetoendocrine cells of both thepancreasandgut

epithelium. Some ofthedeterminants

acquired

during

devel-opment may besharedby

terminally

differentiated cells from different embryological origins, which

might explain

theAPUD

characteristics of theislet cells. Several observations illustrate thepresenceof neuralphenotypicmarkersintissuesnotderived from the neuralcrest.Coexpression of neuron-specific markers

along

with

polypeptide

hormones in cellsof the endocrine sys-tem,has been reported: (a) thepresenceof tyrosine hydroxylase inglucagon- andinsulin-containingcellsoftheembryonicmouse pancreas(4)and

(b)

of synaptophysin,amembraneglycoprotein of presynaptic vesicles in several endocrine epithelial celltypes,

including the pancreatic islets (22);(c) a nerve cell enzyme,

neu-ron-specific enolase,is detectedin awidevarietyoftumors

de-rived fromneural andforegut endocrinecells (23,24). Some of the neural

characteristics

havebeenretainedby theneoplastic

islet cells such as L-dopa decarboxylase (10) indicating that the

acquisition ofthis property occurs early during development. The largeincrease in the poly(A) tract ofthe insulin, glucagon andsomatostatinmRNAsfoundin these cells is intriguing. Most

eukaryoticmRNAspecieshavea sequence of adenylic acid

res-idues added to the 3' ends; newly synthesized mRNAs have

poly(A)tractsof150-200 nucleotides that are shortened to 40-60within afewhours; theshortening of the poly(A)tract

ap-parently

stopsoncethetracthas reached 30-40nucleotides

_(25).

The

_precise

reasonsfor theexistence of the

_poly

_(A)

tractare

still unclear, but a number of recent investigations have shed somelight on its potential role. Two major functions have been proposed: stabilization ofmRNAsand increased translational efficiency (26-30). The fact that among six different mRNAs analyzed only the three encoding the islet hormone mRNAs have alongerpoly (A)tractis even moreintriguing.Theanalogy

to theprimitiveislet cellscanagainbe drawn. Duringfetal de-velopment, the cellsmighthave used theadvantagesprovided by along poly (A) tract to enhance translational efficiency, as hasbeen reportedin other systems(27, 28);the 900-fold increase ininsulin mRNA, accompanied byacorrespondentrise in in-sulin that occursduring ratfetalpancreatic development (31)

may bepossible,in part,byincreased mRNAstability.

It should be noted that increased polyadenylation of the in-sulin mRNA has also been noted in ratyolk sacand in arat

insulinoma line(RIN SF)(31, 32). Theextrapolyadenylation

of the insulin mRNA in the5F line appeared to be dependent uponmatrix substrata upon which the cellswereculturedand tobeassociated withanincreasedresponsivenessof the mRNA

levelstoglucose (33).

Theectopicexpression of theangiotensinogen gene in an isletcelltumoris noveland has notpreviously been reported in islet tumors. Although the major sites of angiotensinogen productionarethe liver andkidneys, it has recently beenfound

that thegeneis expressedin alargenumber of tissuesincluding

thebrain, spinal cord, lung, colon,aorta,heart,and the adrenal

glands, but not in the pancreas (16). The significance of the

derepression oftheangiotensinogengene in these islet cell lines

is unknown, but since it is expressed inevery tumor and cell

linetested,itsexpression mightbelinkedtothetransformation

process.

Theuseofalternative poly(A) site selection during

angio-tensinogen

gene

transcription

in the islet cells is a feature of manyeukaryoticgenes.Thebiologicconsequencesof the

alter-native

polyadenylation

areunknown. Onemay

speculate

that itmayleadtoaltered

stability

of themRNAina

specific tissue,

splicing

modifications

ending

in

particular

polypeptide products

(34-38)orbe without

biologic

consequences (38). The results presented hereprovide

further

informationonthetheory ofa commonisletstemcell and adda newdimension tothe

study

ofthe ratislet cell lines. Thesecell lines should

provide

useful

systemsin which to delineate the

biological

events that take placeduring the developmental pathway

of

theislet cells.

Acknowledgments

We thankAdacie Alan and Heather Hermannforexpertassistance and Janice Canniff fortypingthemanuscript.JacquesPhilippeis supported

byaSwiss NationalFellowship.

The studiesweresupported inpartby NationalInstitues of Health

grantsAM-30457,AM-30834, AM-30846, and AM-32520.

References

1.McCulloch,E. A.1983. Stem cells in normalandleukemic

hemo-poiesis.Blood. 62:1-13.

2.Pictet,R.,and W. J. Rutter. 1972.Developmentof theembryonic

pancreas.Endocrinology. 1:25-66.

3. Pearse,A.G.E. 1966.5-hydroxytryptophanuptake by dogthyroid

Ccells and its possible significanceinpolypeptide hormone production. Nature(Lond.).211:598-600.

of catecholaminergic precursors into glucagon (A) cells in mouse em-bryonic pancreas. Proc. Nati. Acad. Sci. USA. 78:5225-5229.

5. Andrew,A. 1975. APUD cells in the endocrine pancreas and the intestine of chick embryos. Gen. Comp. Endocrinol. 26:485-495.

6. Pearse, A. G. E., and J. M. Polak. 1971. Neural crest origin of the endocrine polypeptide (APUD) cells of the gastro-intestinal tract and pancreas.Gut. 12:783-788.

7. Pearse, A.G. E. 1982. Islet cell precursors are neurons. Nature

(Lond.).295:96-97.

8.Chick, W. L., S. Warren, R. N. Chute, A. A. Like, V. Lauris, and K.C. Kitchen. 1977. A transplantable insulinoma in the rat. Proc.Nati.

Acad. Sci. USA. 74:628-632.

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 so-matostatin-secreting cell lines established from a transplantable rat islet cell tumor. Proc.Natl. Acad. Sci. USA. 77:3519-3523.

10.Oie, H. K., A. F. Gazdar, J. D. Minna, G. L. Weir, and S. B. Baylin. 1983. Clonal analysis of insulin and somatostatin secretion and L-dopa decarboxylase expression by a rat islet cell tumor. Endocrinology. 112:1070-1075.

11.Chick, W. L., M. C. Appel, G. C. Weir, A. A. Like, V. Lasuris, J.G. Porter, and R. N. Chute. 1980. Seriallytransplantable chemically induced rat islet cell tumor. Endocrinology. 107:954-960.

12.Chirgwin, J. M., A. E. Przybyla, R. J. MacDonald, and W. J. Rutter. 1979.Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease.Biochemistry. 18:5294-5299.

13.Aviv, H., and P. Leder. 1972. Purification of biologically active globin messengerRNAbychromatography onoligothymidylic acid-cel-lulose. Proc.Nati.Acad.Sci. USA. 69:1408-1412.

14.Mojsov, S., G. Heinrich, I. Wilson, M. Ravazzola, L. Orci, and J.F. Habener. 1986. Preproglucagon geneexpression in pancreas and intestine diversifies at the level of posttranslational processing. J.Biol. Chem.261:11880-11889.

15. Philippe, J., D. Drucker, S.Mojsov, andJ. F. Habener. 1986. Proglucagon processingin a rat islet cell line resembles phenotype of intestine rather than pancreas.Endocrinology. 119:2833-2839.

16.Campbell, D. J., and J. F. Habener. 1986. The angiotensinogen gene isexpressed anddifferentiallyregulated inmultipletissues of the

rat.J.Clin.Invest.78:31-39.

17.Ohkubo, H., K.Nakayama, T. Tanaka, and S.Nakanishi. 1985. Tissue distribution ofratangiotensinogenmRNAand structuralanalysis ofitsheterogeneity.J.Biol. Chem. 261:319-323.

18. Larson, L. I., L.Grimelius,R.Hakanson, J. F.Rehfeld,F.Stadil,

J.Holst, L.Angervall,and F.Sundler. 1975. Mixedendocrine pancreatic tumorsproducingseveralpeptidehormones. Am. J.Pathol.79:271-279. 19.Friesen,S. R. 1982. Tumorsoftheendocrine pancreas.N.Engl. J.Med. 306:580-590.

20.Pictet, R. L., L. B.Rall,P.Phelps, and W. J. Rutter. 1976. The neuralcrestand theoriginof theinsulin-producing and other gastro-intestinalhormone-producingcells. Science(Wash. DC). 191:191-192. 21.Fontaine,J., C. LeLievre,and N. M. Le Douarin. 1977. What is the developmental fate of the neuralcrestcells whichmigrateinto the pancreas in the avianembryo.Gen.Comp.Endocrinol. 33:394-404.

22.Wiedenmann, B.,W. W.Franke,C.Kuhn, R.Moll,andV. E.

Gould. 1986.Synaptophysin:amarkerproteinforneuroendocrine cells andneoplasms.Proc.Natl. Acad. Sci. USA. 83:3500-3504.

23.Schmechel, D.,P. J.Marangos,andM.Brightman. 1978. Neu-rone-specific enolase is a molecular marker for peripheral and central neuroendocrine cells. Nature (Lond.). 276:834-836.

24.Tapia, F. J., J.M.Polak, and A. J. A. Barbosa. 1981. Neurone-specific enolase is produced by neuroendocrinetumors.Lancet. i:808-811.

25. Mercer, J. F. B., and S.A.Wake. 1985.Ananalysis of the rate ofmetallothionein mRNApoly(A)-shorteningusing RNA blot hybrid-ization. Nuci.Acids Res. 13:7929-7943.

26. Marbaix, G., G. Huez, A. Burny, Y. Cleuter, E. Hubert, M. Leclercq, H. Chantrenne, H. Soreq, U. Nudel, and U. Z. Littauer. 1975. Absence ofpolyadenylate segments in globin messenger RNA accelerates its degradation in xenopus oocytes. Proc.Natl. Acad. Sci. USA. 72:3065-3067.

27.Rosenthal, E. T.,T.R.Tansey, and J. V. Ruderman. 1983. Se-quence-specific adenylations anddeadenylationsaccompanychanges in thetranslationof maternal messenger RNA after fertilization of Spisula oocytes.J. Mol. Biol. 166:309-327.

28.Palatnik, C. M., C. Wilkins, andA.Jacobsen. 1984.Translational controlduring early dictyostelium development: possible involvement of poly(A) sequences. Cell. 36:1017-1025.

29.Doel, M. T., and N. H.Carey. 1976. The translational capacity ofdeadenylated ovalbumin messenger RNA. Cell. 8:51-58.

30.Jacobson, A., and M. Farreau. 1983. Possible involvement of poly(A) inprotein synthesis.Nucl. Acids. Res. 11:6353-6366.

31.Kakita K., S. J.Giddings, P. S. Rotwein, and M. A. Permutt. 1983.Insulin geneexpressioninthedevelopingratpancreas. Diabetes. 32:691-696.

32.Guggenheim, N. Z., and M.A.Permutt. 1986. Identical tran-scriptioninitiation sites forproinsulinmessenger ribonucleic acid in three

insulin-expressingtissues.Endocrinology. 118:1710-1715.

33. Muschel, R., G. Khoury, and L. M. Reid. 1986. Regulation of insulin mRNA abundance and adenylation: dependence on hormones andmatrix substrata.Mol. Cell.Biol. 6:337-341.

34. Amara, S. G., V. Jonas, M. G. Rosenfeld, E. S. Ong, and R. M. Evans. 1982.AlternativeRNAprocessing in calcitonin geneexpression

generates mRNAs encoding different polypeptide products. Nature

(Lond.). 298:240-244.

35.Rosenfeld,M.G.,J. J.Mermod, S. G. Amara, L. W.Swanson, P. E.Sawchenko,J. Rivier,W.W.Vale,and R. M.Evans. 1983. Pro-duction ofanovelneuropeptideencodedbythe calcitonin gene via

tissue-specificRNAprocessing.Nature(Lond.).304:129-135.

36.Alt, F.W.,A.L. M.Bothwell,M.Knapp,E.Siden,E.Mather,

M.Koshland,and D. Baltimore. 1980.Synthesisof secreted and mem-brane-boundimmunoglobulinmuheavychains is directedbymRNAs thatdiffer in their 3' end. Cell. 20:293-301.

37.Schwarzbauer,J.E., J. W.Tamkun,I. R.Lemischka,and R.0.

Hynes. 1983. Three different fibronectin mRNAs ariseby alternative

splicingwithin thecoding region.Cell. 35:421-431.

38.Setzer,D.R., M.McGrogan,and R.T.Schimke. 1982. Nucleotide sequencesurrounding multiplepolyadenylation sites in themouse

di-hydrofolatereductase gene. J. Biol. Chem. 257:5143-5147.

39.Philippe, D., J. Drucker, andJ. F.Habener. 1987.Transcriptional

regulation of genesencodinginsulin, glucagonandangiotensinogen by