Proc. Natl. Acad. Sci. USAVol. 89, pp. 1045-1049, February 1992Biochemistry

c-Jun represses the human insulin promoter activity that dependson multiple cAMP response elements

(cAMP response element/glucose/transcriptional repression/insufinoma cells)

NOBUYA INAGAKI*, TOSHIO MAEKAWAt, TATSUHIKO SUDOt, SHUNSUKE ISHIIt, YUTAKA SEINO*,AND HIROO IMURA**Second Division, Department of Medicine, and *Department of Metabolism and Clinical Nutrition, Kyoto University School of Medicine, Kyoto 606, Japan;and tLaboratory of Molecular Genetics, Tsukuba Life Science Center, The Institute of Physical & Chemical Research (RIKEN), Tsukuba, Ibaraki 305, Japan

Communicated by Donald F. Steiner, August 29, 1991 (received for review June 20, 1991)

ABSTRACT Glucose is known to increase the cAMP con-centration in pancreatic ( cells. To determine the mechanismby which cAMP augments insulin gene expression, we firstidentified the cAMP response elements (CREs) of the humaninsulin gene. In DNase I footprint analysis, the bacteriallysynthesized CRE-binding protein, CRE-BP1, protected foursites: two sites in the region upstream from the insulin corepromoter, one site in the rn-st exon, and one site in the firstintron. To examine the roles of those four sites, we constructeda series of DNA plasmids in which the wild-type and mutantinsulin promoters were linked to the chloramphenicol acetyl-transferase gene. Studies of the transcriptional activity of theseplasmids after transfection into hamster insulinoma (HIT) cellsshowed that these four sites contributed additively to the cAMPinducibility of the insulin promoter. Surprisingly, the c-junprotooncogene product (c-Jun) repressed the cAMP-inducedactivity of the insulin promoter in a cotransfection assay withthe c-Jun expression plasmid. Northern blot analysis demon-strated that the level of c-jun mRNA was dramatically in-creased by glucose deprivation in HIT cells. These resultssuggest that glucose may regulate expression of the humaninsulin gene through multiple CREs and c-Jun.

Insulin is a polypeptide hormone of major physiologicalimportance in the regulation of glucose homeostasis in ani-mals, and its biosynthesis in pancreatic f3 cells is regulatedmainly by blood glucose (1, 2). Although glucose is known toincrease insulin biosynthesis at the transcriptional (3, 4) andtranslational (5) levels, the mechanisms are poorly under-stood. However, the facts that glucose increases the cAMPconcentration in pancreatic 8 cells (6) and that analogues ofcAMP augment insulin mRNA levels in isolated islets (7) andin insulin-secreting clonal cell lines (3, 4, 8) support thepossibility that cAMP may participate in glucose-inducedinsulin gene expression.The cAMP response element (CRE; TGACGTCA) was

first identified as being an inducible enhancer of genes thatcan be transcribed in response to increased cAMP levels (9,10). Mutation analysis of the rat insulin I gene promoterindicated that the rat insulin promoter also contained onefunctional CRE (11). cDNA clones of multiple CRE-bindingproteins such as transcription factors CREB (12, 13), CRE-BP1 (14), and ATFs (15) have been isolated. All of theseproteins have a basic amino acid cluster linked to the leucinezipper as a DNA-binding domain. Among multiple CRE-binding proteins, only the function of CREB has been wellanalyzed, and little is known about the function of otherCRE-binding proteins. Phosphorylation of CREB occurs atSer-133 in response to an increase in the intracellular cAMP

levels, and the phosphorylated form can activate transcrip-tion of the somatostatin gene (16).The c-jun and c-fos protooncogene products (c-Jun and

c-Fos) have DNA-binding domains that are similar to that ofCRE-binding proteins (17-19). The c-Jun-c-Fos heterodimer,AP-1, binds preferentially to the phorbol 12-myristate 13-acetate response element (TGACTCA), which is very similarto the CRE (20, 21) but also has a low but detectable bindingactivity to the CRE (22, 23). Furthermore, c-Jun can form aheterodimer with CRE-BP1 and bind to the CRE with highaffinity (23). These facts suggest that c-Jun may regulatetranscription of genes such as that of insulin through CRE.However, so far little is known about the relationship be-tween c-Jun and transcription of the insulin gene.

In this study, we have identified four CREs in the humaninsulin gene and found that these four CREs are responsiblefor cAMP inducibility and basal promoter activity. Further-more, we have found that c-Jun represses the cAMP-inducedand basal activities of the insulin promoter and that glucosedeprivation increases the c-jun mRNA level in hamsterinsulinoma (HIT) cells.

MATERIALS AND METHODSPlasmid Construction. To generate a series of chloram-

phenicol acetyltransferase (CAT) plasmids (pINCAT1 topINCAT5) shown in Fig. 1A, various lengths of DNA frag-ments containing the human insulin gene promoter wereisolated using appropriate restriction enzymes and insertedinto the HindIII site ofpSVOOCAT (24) using a HindIII linker.To make the plasmids containing the mutated CRE1 andCRE2 shown in Fig. 3A, the PCR and the oligonucleotide-directed mutagenesis (25) were used. To confirm that PCRproducts do not contain a sequence alteration at other thanthe desired location, we sequenced the PCR-derived con-structs.

Preparation of Bacterially Expressed CRE-BP1 and DNase IFootprinting. CRE-BP1 was expressed in Escherichia coliusing a T7 expression vector and purified as described (26, 27).For DNase I footprinting, the HindIfl-HincIl [nucleotides (nt)-339 to -61] fragment or the HindIII-Bgl II (nt + 112 to -169)fragment ofthe plasmid pINCAT1 was 32P-labeled at the 5' endof the upper or lower strand. DNA binding reactions andDNase I digestion were done in 60 ,ul with 1-4 ng ("10 fmol)of an end-labeled DNA fragment and 30 Ag of the bacteriallyexpressed CRE-BP1 as described (14).DNA Transfection and CAT Assay. HIT cells were grown in

RPMI 1640 medium with 10%6 (vol/vol) fetal calf serum.Mixtures of 8 Ag of each CAT plasmid DNA, 8 ,ug of plasmidpacti DNA, which contains the chicken cytoplasmic 8-actinpromoter but no cDNA sequence to be expressed (27), and 4

Abbreviations: Bt2cAMP, N6,02'-dibutyryladenosine 3',5'-cyclicmonophosphate; CAT, chloramphenicol acetyltransferase; CRE,cAMP response element; nt, nucleotide(s).

1045

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 89 (1992)

A 300Pstl Pi

-339

-2

-200 -1 00 0 +100I I I

Givull Hindi BgI 11 Hin-cll P'stI -

58 I HCAi-229

-1 69CAl

-61 +112

B Relative CAT Activity15 0 10.0 5.0 0

x 9.1

x 12.6

x 9.2

0

0

0

00

x6.0_

ELZ- cAMP

E+ cAMP ND

Plasmids

L pINCAT1L pINCAT2F pINCAT3

glyoxal in dimethyl sulfoxide, then electrophoresed on a 1%agarose gel (33), transferred to a nitrocellulose filter, andhybridized to the 32P-labeled mouse c-jun or rat preproinsulinI cDNA (kindly supplied by D. F. Steiner, University ofChicago) (34).

L pINCAT4 RESULTS_ pINCAT5 Deletion Analysis of the 5' Region of the Human Insulin

Gene. To delineate the DNA sequences mediating cAMPinducibility, we fused the 451-base-pair (nt -339 to +112)fragment of the 5' flanking sequence or its 5' deleted frag-ments of the human insulin gene to the bacterial CAT gene(Fig. 1A). Promoter activity and transcriptional regulation by

* pJNCAT1 Bt2cAMP were examined by transient assays after transfec-tion of these constructs into HIT cells (Fig. 1B). When weused pINCAT1, a 9-fold increase in CAT activity over basal

* ]pINCAT2 levels was observed after treatment with Bt2cAMP. Removalof sequences from nt -339 to nt -256 or nt -229 (pINCAT2and pINCAT3) did not decrease the basal promoter activity

J pINCAT3

IpINCAT4

IplNCAT5

FIG. 1. Functional analysis of 5' deletion mutants of the humaninsulin gene. (A) Structure of the 5' deletion mutants of the 5' regionof the human insulin gene. Deletions were generated using therestriction enzyme sites indicated above. (B) CAT activities of the 5'deletion mutants after transfection into HIT cells. A mixture of eachCAT plasmid and pact-/3-gal plasmid was transfected into HIT cells,extracts of transfectants were assayed for CAT activity, and data areshown to the right. Some cell samples were treated with Bt2cAMP asindicated. On the left, relative CAT activity is indicated by a bargraph. ND, not detected.

,ug of internal control plasmid pact-p-gal DNA, in which thechicken cytoplasmic ,3-actin promoter was linked to the ,3-ga-lactosidase gene, were transfected into HIT cells by thecalcium phosphate method (28). To examine the role of c-Junin the insulin promoter activity, mixtures of 8 ,ug of each CATplasmid DNA, 8 Ag of pRSV-c-jun or pRSV-MG, and 4 pg ofpact-3-gal were transfected into HIT cells. In the plasmidpRSV-c-jun, the mouse c-jun cDNA was linked to the longterminal repeat of Rous sarcoma virus (29).The control plas-mid pRSV-MG was generated by joining the mouse P-globingene with the long terminal repeat ofRous sarcoma virus. Wehave confirmed (30) that this control effector plasmid could beused for CAT cotransfection assay to observe the c-Jun-induced trans-activation depending on CRE. To analyze thecAMP inducibility of the insulin gene promoter, cells weretreated with N6,02'-dibutyryladenosine 3',5'-cyclic mono-phosphate (Bt2cAM; final concentration, 2 mM) for 24 hbefore harvesting. Forty-eight hours after transfection, CATwas assayed for 1 h by the procedure ofGorman etaL (31). Theamounts of cell extract used for the CAT assay were normal-ized with respect to p-galactosidase activity. All CAT trans-fection experiments were repeated three or four times, andtypical results are shown here. The differences between eachset of experiments were within 20%.

Preparation ofRNA and Northern Blot Hybridization. HITcells at a density of 5 x 106 cells per 10-cm dish werecultivated in the medium containing various concentrationsof glucose and no serum for 24 h, and then total RNA wasprepared by the guanidine isothiocyanate/cesium chloridemethod (32). Total RNA (20 ug) was denatured with 6 M

A

97CRE-BP 1

66

43 wiw

31 -o

kDa

B (, AF +

+- GG

1tI- |

CRE1 *

-2 3

CRl 1 w

l J a

I

Ii4 ... .

.

CRE3

CRE4 i

5

CCREI L AAGA C .7 -AATGACCCGCTGTC. T %... - "..',

CRE2

CRE3

CRE4

E AGGSJ5GCTGACGACCAAGG AGA>-1".

GGCTGCATCAGAAGAG

*CE-7GCT TCAGlGGI-

CRE1 CRE2 CRE3 CREW

,CG,ACX CA7-TCrT TATA ---iP^

FIG. 2. Binding of CRE-BP1 to the 5' flanking region of humaninsulin gene. (A) Purification of the bacterially expressed CRE-BP1.The purified CRE-BP1 was separated on a SDS/101% polyacrylamidegel and stained with Coomassie brilliant blue. (B) DNase I footprint-ing. The HindflI-HinclI (lanes 1-4) or HindIII-Bgl II (lanes 5-8)fragment of pINCAT1 was 32P-labeled, incubated with 30 jtg ofCRE-BP1 (lanes marked +) or without protein (lanes marked -), anddigested with DNase I. Markers were obtained by the chemicalcleavage of the same end-labeled DNA fragment (lanes 3, 4, 7, and8). The protected regions are depicted on the left. (C) Nucleotidesequences of the CRE-BP1-binding sites. Nucleotide sequences offour sites protected by CRE-BP1 are shown. The sequences similarto the CRE consensus are underlined. A schematic diagram of fourCRE-BP1-binding sites is indicated at the bottom.

1046 Biochemistry: Inagaki et al.

Proc. Natl. Acad. Sci. USA 89 (1992) 1047

A CRE1 CRE2 CRE3 CRE4-339 ____+ 112pINCAT1-wt -GAC- TGA(ACC- TGCATCAG lUCUICAG-

pINCAT1-ml T- ' 'rf

pINCAT1 -m2 -AA AAA

pINCAT1-m3 U Ur a AA

BRelative CAT Activity

10 8 6 4 2 0.

El

E

| | - cAMP L- + cAMP

*

* 0* X* 0

.0 I

.a 1

pINCAT1-wt

pINCAT1-m1

plNCAT1-m2

I pINCAT1-m3

FIG. 3. Effects of mutation of the CRE-BP1-binding site on the insulin promoter activity. (A) Schematic representation of the CAT plasmidscontaining the mutated CREs. Mutations were introduced into four CREs in the 5' region of the human insulin gene. (B) Transient expressionof CAT activity. Results of the CAT assays were presented as in Fig. 1B.

or the cAMP inducibility. Although further deletion to nt-169 (pINCAT4) moderately decreased the basal transcrip-tional activity and cAMP inducibility, a 6-fold increase inCAT activity after Bt2cAMP addition was still observed.Deletion to nt -61 resulted in the almost complete loss ofbasal promoter activity, probably because the cis elements inthe core promoter were removed, as has been reported (35).These results suggest that multiple regions not only upstreambut also downstream from nt -169 conferred the cAMPresponsiveness. This differs from previous results with therat insulin I gene that showed the presence of only one CREbetween nt -177 and -184 (11).

Multiple CRE-BP1-Binding Sites in the Human InsulinGene. To identify the CRE within the human insulin gene, wedid a DNase I footprint analysis using the bacterially syn-thesized CRE-BP1, one of the CRE-binding proteins (Fig.2A). CRE-BP1 protected four regions: two sites in the regionupstream of the RNA start site (CRE1; nt -221 to -195 andCRE2; nt -189 to -167), one site in the first exon (CRE3; nt+16 to +31), and one site in the first intron (CRE4; nt +57to +73) (Fig. 2B). Within the core of each protected region,we found a DNA sequence similar to the CRE consensussequence (5'-TGACGTCA-3') (Fig. 2C). The core sequencewithin the CRE2 (TGACGACC) is similar to the CRE of therat insulin I gene (TGACGTCC) (11), although a 1-basedifference was found.cAMP Inducibility of the Human Insulin Gene Is Mediated

by Four CREs. To find whether or not the cAMP respon-siveness of the human insulin gene is dependent on the fourputative CREs, we constructed mutant pINCAT1 plasmidscontaining the mutated CREs (Fig. 3A). The CAT activitiesexpressed in HIT cells by the wild-type and mutant plasmidsare shown in Fig. 3B. The level of CAT activity expressedfrom the control plasmid pINCAT1 was increased about10-fold by treatment with Bt2cAMP. When both CRE1 andCRE2 were disrupted (pINCAT1-ml), the degree of induc-tion by Bt2cAMP was reduced to a 4-fold increase. Introduc-tion of mutations into both CRE3 and CRE4 (pINCAT1-m2)also reduced the cAMP inducibility to 5-fold and resulted ina 2-fold reduction in basal promoter activity. Disruption of all

four CREs (pINCAT-m3) dramatically reduced the cAMPinducibility to 2-fold. These results indicate that four CREswithin the 5' region, including the first exon and first intron,additively contribute to the basal activity and cAMP respon-siveness of the human insulin promoter.

Repression of Human Inslin Promoter Activity by c-Jun.c-Jun is known to form a heterodimer with CRE-BP1 and bindto CRE. Therefore, we investigated the effects of c-Jun onhuman insulin promoter activity (Fig. 4). Cotransfection ofpINCAT1 and the c-Jun expression plasmid into HIT cellsshowed that c-Jun represses the basal activity and the cAMP-induced activity of human insulin promoter to one-third andone-tenth, respectively. To assess nonspecific effects ofc-Jun on gene expression, the c-Jun expression plasmid wascotransfected into HIT cells with a control plasmid contain-ing the long terminal repeat of Rous sarcoma virus linked tothe CAT gene. No effect could be detected (data not shown).To examine whether the specific sequence in the regionupstream from the core promoter is responsible for thec-Jun-induced repression, three deletion mutants (pINCAT2to pINCAT4) of the human insulin promoter were used forcotransfection experiments. c-Jun represses the CAT activityexpressed from any of these plasmids, indicating that thespecific sequence in the region downstream of nt -169 ormultiple sequences dispersed in the region between nt -339and +112 mediate the c-Jun-induced repression. To investi-gate whether the four CREs mediate the c-Jun-inducedrepression, we used the point mutants ofCREs for a cotrans-fection experiment (Fig. 5). When we used the mutant ofCRE1 and CRE2 (pINCAT1-ml), the repression of basalpromoter activity by c-Jun was similar to that with wild-typepromoter (pINCATi), but the repression of cAMP-inducedactivity by c-Jun was partially relieved, and c-Jun repressedthe activity to one-eighth. Similarly, when the mutant ofCRE3 and CRE4 (pINCAT-m2) was used, the c-Jun-inducedrepression was partially relieved. In this case, c-Jun re-pressed the basal promoter activity and the cAMP-inducedactivity to one-half and one-seventh, respectively. Disrup-tion of all four CREs (pINCAT1-m3) strongly relieved thec-Jun-induced repression, and only slight repressions of the

Biochemistry: Inagaki et al.

Proc. Natl. Acad. Sci. USA 89 (1992)

Relative CAT Activity12 10 8 6 4 2 0L . 1 . 1- Ic-Jun

c

-~~~~~~~~

L

LIIK -cAMP- +cAMP

+

+.

A

*

* pINCATI0

*

* 0* pINCAT2

0

* 0

it pINCAT30

* *1

0

* pINCAT40

0

0 2.8 5.5 8.3 11.1 22.2[G lucose], m M

B c-Jun mRNA

- 28S3.2 kb _2.6 kb +4 * V tw

-1 8 S

0 2.8 5.5 8.3 11.1 222[Glucose], mM

FIG. 4. Repression ofpromoter activity ofthe human insulin geneby c-Jun. A series of CAT plasmids indicated in Fig. 1 werecotransfected with the c-Jun expression plasmid (lanes marked +) orwith the control effector plasmid (lanes marked -) into HIT cells,and CAT activities were measured and are shown on the right.Relative CAT activity is indicated by a bar graph on the left.

basal promoter activity (1.5-fold) and the cAMP-inducedactivity (3-fold) were observed with pINCAT1-m3. Theseresults indicate that c-Jun represses the basal and cAMP-induced promoter activities of the human insulin gene thatdepends on the four CREs.

Effect of the Glucose Concentration on the c-jun mRNALevel. Glucose is known to increase insulin gene transcrip-tion, but its mechanism is poorly understood. Our resultsindicate that c-Jun represses the insulin gene transcription.Therefore, we examined whether glucose affected the level ofc-jun expression. After culture of HIT cells for 24 h in themedium containing various glucose concentrations, the levelof c-jun mRNA was analyzed on Northern blots (Fig. 6). TwomRNA species, 2.6 and 3.2 kilobases, were detected usingthe mouse c-jun cDNA probe, as reported in other species

Relative CAT Activity8 6 4 2LI...X. LI

III

I

LII]- cAMP+ cAMP

°c-Jun, 0

-.

0

+0

0

+ 0

_+0

-

.0+ I---0

+0-

0

piNCAT1-wt

pINCAT1-ml

w* plNCAT1-m2

plNCAT1-m3

FIG. 5. Effects of mutation of CRE on the c-Jun-induced repres-sion. A series ofCAT plasmids indicated in Fig. 3 were cotransfectedwith the c-Jun expression plasmid or the control effector plasmid intoHIT cells, and CAT activities were assayed and are shown on theright. Relative CAT activities are shown by a bar graph on the left.

C Insulin mRNA

0.5 kbb *

0 2.8 5.5 11.1 22.2

[Glucose], mM

FIG. 6. Effects of glucose on c-jun and insulin mRNA in HITcells. HIT cells were cultivated at the glucose concentration indi-cated for 24 h, and total RNAs were prepared. (A) Ethidium bromidestaining of the total cellular RNA. (B) Northern blot analysis wasdone with the mouse c-jun DNA probe. (C) Northern blot analysisusing the rat insulin I DNA probe. kb, Kilobase(s).

(36). The c-jun mRNA level was invariable at 2.8-22.2 mMglucose but was dramatically increased about 10-fold byglucose deprivation. These results suggest that glucose dep-rivation causes a decrease of insulin gene expression byincreasing the level of c-Jun. In addition, after 24 h of culturein the medium containing various glucose concentrations thelevel of insulin mRNA in HIT cells was also analyzed onNorthern blots (Fig. 6C). The insulin mRNA level in HITcells cultured at 2.8 mM glucose was increased =3-foldcompared to cells cultured without glucose. However, onlya little transcriptional response to glucose was observedabove the 2.8 mM glucose concentration.

DISCUSSIONOur results demonstrate that there are four functional CREsin the 5' region of the human insulin gene and that these fourCREs contribute additively to cAMP responsiveness of thehuman insulin gene promoter. Within the core of each CRE,there is a sequence similar to the CRE consensus sequenceand the introduction of mutations into these CRE consensussequences results in a severe decrease in the basal andcAMP-induced promoter activity. Philippe and Missotten(11) reported that a deletion or point mutation ofthe sequenceTGACGTCC between nt -177 and -184 within the ratinsulin I gene promoter markedly decreased the basal activityand cAMP responsiveness, using HIT cells. The nearlyidentical corresponding sequence TGACGACC (nt -182 to-175) is within the CRE2 of the human insulin gene and wasshown to bind to nuclear protein extracted from HIT cells(37). However, deletion or point mutation of this elementonly partially impairs the basal and cAMP-induced activitiesof the human insulin promoter. In addition, there is littlesimilarity between rat and human in the regions correspond-

1048 Biochemistry: Inagaki et al.

Proc. Natl. Acad. Sci. USA 89 (1992) 1049

ing to CREs other than CRE2 of the human insulin gene.These facts suggest that the rat and human insulin genes haveone and four CREs, respectively, and that their expressionsare regulated differently by cAMP.Our results show that c-Jun represses the basal and cAMP-

induced activities of the human insulin promoter. c-Jun andc-Fos are components of the transcription factor AP-1 thatrecognizes the phorbol 12-myristate 13-acetate response el-ement and activates transcription of many genes, such as thecollagenase gene (20, 21). In addition to positive regulatoryeffects, c-Fos has been shown to down regulate severalimmediate early genes, such as the c-fos gene depending ona CC(A/T)6GG (CArG) element (38-40). The results ob-tained by using the c-Jun expression plasmid were in agree-ment; the c-Fos-induced trans-repression of the immediateearly gene promoter was enhanced by c-Jun in some cases(39), but not in other cases (40). We observed that not onlyc-Jun but also c-Fos repressed the insulin promoter activity(data not shown). Since the 5' deletion to nt -169 did notrelieve this c-Jun-induced repression, the DNA sequencesresponsible for this negative regulation seem to exist betweennt -169 and + 112 of the human insulin gene. However, thisregion of the human insulin gene does not contain a sequencesimilar to the CArG element, suggesting that the mechanismof the c-Jun- or c-Fos-induced repression of the humaninsulin promoter is different from that described above.Recently, we have found that the CRE-BP1 homodimer bindsto the CREs of the human insulin gene, but neither thec-Jun-CRE-BP1 heterodimer nor the c-Jun homodimer bindsto the CREs of the human insulin gene (data not shown).These results suggest that c-Jun represses transcription ofthehuman insulin gene by inhibiting some CRE-binding proteinssuch as CRE-BP1 and that this c-Jun-induced repressiondepends on the unique CRE sequences in the human insulingene promoter that differ from both consensus sequences ofCRE and the phorbol 12-myristate 13-acetate response ele-ment. Both the protooncogenes c-jun and c-fos are theimmediate early genes whose characteristic is that theirexpression is low or undetectable in quiescent cells but israpidly induced at the transcriptional level upon stimulationof cellular proliferation: serum, growth factors, or phorbol12-myristate 13-acetate (36, 41, 42). Interestingly, however,in HIT cells glucose deprivation greatly increased the c-junmRNA level. Since c-Jun was observed to repress the basaland cAMP-induced activities of the human insulin promoter,c-Jun might be important in the decrease of insulin geneexpression at low glucose concentrations. Although we founda 10-fold increase of the c-jun mRNA level upon glucosedeprivation, the c-jun mRNA level was not changed dramat-ically from 2.8 to 22.2 mM glucose. Even at 2.8 mM glucose,which is =50 mg/dl and is low by physiological standards, thec-jun mRNA level was not elevated. However, the facts thatin HIT cells the insulin mRNA level was low at 0mM glucoseand was increased at 2.8-22.2 mM glucose and that the c-junmRNA level was, correspondingly, invariable at 2.8-22.2mM glucose but was increased dramatically by glucosedeprivation suggest that reduced expression of c-jun mRNAmay have a role in the glucose-induced increase of insulingene transcription. A study using primary (3 cells that respondto glucose at concentrations within the physiological range(43) will be required to clarify this point. Although furtherstudies are necessary to clarify the relationship betweenc-Jun and glucose-regulated gene expression, our findingsmay provide a clue as to how the expression of the humaninsulin gene is regulated.We thank K. Nakamura for expert assistance and Dr. S. Hirai, Dr.

G. I. Bell, and Dr. Y. Ebina for providing the pRSV-c-jun, the humaninsulin gene and the pSVOOCAT plasmid, respectively. This research

was supported in part by a grant for diabetes research from OtsukaPharmaceutical Co., Ltd., in part by Monbusho International Sci-entific Research Program: Joint Research 03044088, in part byCancer Bio-Science 0325817 from the Ministry of Education, Scienceand Culture of Japan.1. Steiner, D. F. & Tager, H. S. (1979) in Endocrinology, eds. De Groot,

L. J., Cahill, G. F., Jr., Odell, W. D., Martini, L., Potts, J. J., Jr.,Nelson, D. H., Steingerger, E. & Winegrad, A. L. (Grune & Stratton,New York), Vol. 2, pp. 921-934.

2. Permutt, M. A. (1981) in Islets ofLangerhans, eds. Cooperstein, C. J. &Watkins, D. T. (Academic, New York), pp. 75-95.

3. Nielsen, D. A., Welsh, M., Casadaban, M. J. & Steiner, D. F. (1985) J.Biol. Chem. 260, 13585-13589.

4. Welsh, M., Nielsen, D. A., MacKrell, A. J. & Steiner, D. F. (1985) J.Biol. Chem. 260, 13590-13594.

5. Itoh, N. & Okamoto, H. (1980) Nature (London) 283, 100-102.6. Charles, M. A., Lawecki, J., Pictet, R. & Grodsky, G. M. (1975) J. Biol.

Chem. 250, 6134-6140.7. Hammonds, P., Schofield, P. N., Ashcroft, S. J. H., Sutton, R. & Gray,

D. W. R. (1987) FEBS Lett. 223, 131-137.8. Hammonds, P., Schofield, P. N. & Ashcroft, S. J. H. (1987) FEBS Lett.

213, 149-154.9. Comb, M., Birnberg, N. C., Seasholtz, A., Herbert, E. & Goodman,

H. M. (1986) Nature (London) 323, 353-356.10. Montminy, M. R., Sevarino, K. A., Wagner, J. A., Mandel, G. &

Goodman, R. H. (1986) Proc. Nati. Acad. Sci. USA 83, 6682-6686.11. Philippe, J. & Missotten, M. (1990) J. Biol. Chem. 265, 1465-1469.12. Hoeffler, J. P., Meyer, T. E., Yun, Y., Jameson, J. L. & Habener, J. F.

(1988) Science 242, 1430-1433.13. Gonzalez, G. A., Yamamoto, K. K., Fischer, W. H., Karr, K., Menzel,

P., Briggs, W., III, Vale, W. W. & Montminy, M. R. (1989) Nature(London) 337, 749-752.

14. Maekawa, T., Sakura, H., Kanei-Ishii, C., Sudo, T., Yoshimura, T.,Fujisawa, J., Yoshida, M. & Ishii, S. (1989) EMBO J. 8, 2023-2028.

15. Hai, T., Liu, F., Coukos, W. J. & Green, M. R. (1989) Genes Dev. 3,2083-2090.

16. Gonzalez, G. A. & Montminy, M. R. (1989) Cell 59, 675-680.17. Kouzarides, T. & Ziff, E. (1988) Nature (London) 336, 646-651.18. Gentz, R., Rauscher, F. J., III, Abate, C. & Curran, T. (1989) Science

243, 1695-1699.19. Turner, R. & Tjian, R. (1989) Science 244, 1689-1694.20. Bohman, D., Boss, T. J., Admon, A., Nishimura, T., Vogt, P. K. &

Tjian, R. (1987) Science 238, 1386-1392.21. Angel, P., Allegretto, E. A., Okino, S. T., Hattori, K., Boyle, W. J.,

Hunter, T. & Karin, M. (1988) Nature (London) 332, 166-171.22. Rauscher, F. J., III, Voulalas, P. J., Franza, B. R., Jr., & Curran, T.

(1988) Genes Dev. 2, 1687-1699.23. Macgregor, P. F., Abate, C. & Curran, T. (1990) Oncogene 5, 451-458.24. Araki, E., Shimada, F., Shichiri, M., Mori, M. & Ebina, Y. (1988)

Nucleic Acids Res. 16, 1627.25. Sayers, J. R., Schmidt, W. & Eckstein, F. (1988) Nucleic Acids Res. 16,

791-802.26. Studier, F. W. & Moffat, B. A. (1986) J. Mol. Biol. 189, 113-130.27. Takeda, J., Maekawa, T., Sudo, T., Seino, Y., Imura, H., Saito, N.,

Tanaka, C. & Ishii, S. (1991) Oncogene 6, 1009-1014.28. Graham, F. & Van der Eb, A. (1973) Virology 52, 456-457.29. Hirai, S., Bourachot, B. & Yaniv, M. (1990) Oncogene 5, 39-46.30. Matsuda, S., Maekawa, T. & Ishii, S. (1991) J. Biol. Chem. 266,

18188-18193.31. Gorman, C. M., Moffat, L. F. & Howard, B. H. (1982) Mol. Cell. Biol.

2, 1044-1051.32. Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J. & Rutter, W. J.

(1979) Biochemistry 18, 5294-5299.33. McMaster, G. K. & Carmichael, G. G. (1977) Proc. Natl. Acad. Sci.

USA 74, 4835-4839.34. Chan, S. J., Noyes, B. E., Agarwal, K. L. & Steiner, D. F. (1979) Proc.

Natl. Acad. Sci. USA 76, 5036-5040.35. Karlson, O., Edlund, T., Moss, J. B., Rutter, W. J. & Walker, M. D.

(1987) Proc. Natl. Acad. Sci. USA 84, 8819-8823.36. Ryseck, R.-P., Hirai, S., Yaniv, M. & Bravo, R. (1988) Nature (London)

334, 535-537.37. Boam, D. S., Clark, A. R. & Docherty, K. (1990) J. Biol. Chem. 265,

8285-8296.38. Sassone-Corsi, R., Sisson, J. C. & Verma, I. M. (1988) Nature (London)

334, 314-319.39. Lucibello, F. C., Lowag, C., Neuberg, M. & Muller, R. (1989) Cell 59,

999-1007.40. Gius, D., Cao, X., Rauscher, F. J., III, Cohen, D. R., Curran, T. &

Sukhatme, V. P. (1990) Mol. Cell. Biol. 10, 4243-4255.41. Lamph, W. W., Wamsley, P., Sassone-Corsi, P. & Verma, I. M. (1988)

Nature (London) 334, 629-631.42. Quantin, B. & Breathnach, R. (1988) Nature (London) 334, 538-539.43. German, M. S., Moss, L . A& Rutter, W. J. (1990) J. Biol. Chem. 265,

22063-22066.

Biochemistry: Inagaki et al.