Transcriptional regulation of alpha-fetoprotein expression by ...

7 downloads 0 Views 5MB Size Report
nant plasmids, incubated with or without 3 X lo-' M dexamethasone, and then assayed for chloramphenicol acetyltransferase expression. In hepatoma cells that.
THEJOURNAL OF BIOLOGICAL CHEMISTRY Q 1989 by The American Society for Biochemistry and Molecular Biology, Inc

VOl. 264, No. 1, h u e of January 5, pp. 266-271,1989 Printed in U.S.A.

Transcriptional Regulation of a-Fetoprotein Expressionby Dexamethasone in Human Hepatoma Cells* (Received for publication, July 11,1988)

Hidekazu Nakabayashi, KazutadaWatanabeS, Akira Saitos, Akira Otsurull, Kazuyuki Sawadaishill , and TaikiTamaoki** From the Department of Medical Biochemistry,Faculty of Medicine, University of Calgary, Calgary, Alberta, Canada T2N 4N1

The level of a-fetoprotein (AFP) mRNA in HUH-7 been shown that the administration of dexamethasone to human hepatoma cells is elevated by the addition of rodent neonates decreases AFP expression at both protein dexamethasone to the culture medium. To locate the and mRNA levels in the liver (5, 6). DNA region involved in hormonal regulation of the Modulation of AFP gene expression by dexamethasone has AFP gene, we constructed recombinant plasmids in also been observed in cultured hepatoma cells. In the case of which various lengths of the 5”flanking sequence of rat hepatoma cultures, dexamethasone treatment results in the human AFP gene were fused to the CAT gene. either an increase or a decrease inAFP production. For Various cell lines were transfected with the recombi- example, dexamethasone suppresses the secretion of AFP and nant plasmids, incubated with or without 3 X lo-’ M dexamethasone, and then assayed for chloramphenicolthe level of AFP mRNA in the7777 hepatoma (7-9), whereas increases in the level of both AFP secretion and AFP mRNA acetyltransferase expression. In hepatoma cells that produce AFP, the dexamethasone treatment resulted concentration have been observed in the McA-RH8994 hepatoma (10, 11).Increased secretion of AFP is also reported in the stimulated chloramphenicol acetyltransferase expression when the transfected plasmids contained with the H4-II-E-C3-V hepatoma (12), whereas no change in 169 base pairs (bp) or longerAFP 5”flanking se- AFP production has been observed in the AH-66 hepatoma quence. No dexamethasone effect was observed when (13). In the case of HUH-7 and five other human hepatoma the 5”flankingsequence was less than 98 bp long. The cells, dexamethasone treatmenthas invariably resulted in dexamethasone stimulationwas effectively suppressed increased secretion of AFP (14). The reason why both stimby the glucocorticoid antagonistRU486,indicating ulatory and suppressive effects are observed with the rat that thiseffect is mediated by glucocorticoid receptors. hepatoma cell lines, whereas only the stimulatory effect was The 71-bp region between positions -169 and -98 obtained in the human hepatomas, is not known at present. contains a nucleotide stretch which is similar to the In either case, available evidence shows that dexamethasone consensus sequence of the glucocorticoid responsive regulates the AFP gene primarily at thelevel of transcription. element (GRE). Partial alterations of this sequence Transcriptional regulation by steroid hormones is thought resulted in decreased dexamethasone response. The to be mediated by initial binding of the steroid to itsreceptor GRE-containing region stimulated heterologous followed byspecific interaction of the hormone-receptor com(SV40) promoter activity in response to dexamethaplex with a DNA element, termed “glucocorticoid responsive sone treatment in an orientation- and position-independent manner. The GRE and the upstream AFP en- element” (GRE) (15). In this investigation, we used transient transfection analysis to detect a DNA region upstream of the hancer function independently from each other. human AFP gene that mediates positive transcriptional regulation by dexamethasone inhuman hepatoma cells.We report here the localization and characterization of an active GRE that is present in a71-bp regionin thehuman AFP gene a-Fetoprotein (AFP)’ is a major serum protein during the fetal stage, but it is hardly detectable in adult life (1-3). The promoter. AFP expression during development is under the control of several hormones, most notably glucocorticoids (4). It has

* This work was supported by the National Cancer Institute of Canada and the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement’’in accordance with 18U.S.C. Section 1734 solelyto indicate this fact. 4 Present address: Dept. of Biochemistry, Metropolitan Institute of Gerontology, Tokyo, Japan. Present address: National Institute of Agrobiological Resources, Tsukuba Science City, Yatabe, Ibaragi, Japan. 11 Present address: First Dept. of Internal Medicine, Nagasaki University, Nagasaki, Japan. 11 Present address: Dept. of Biotechnology, Kansai New Technology Research Institute, Torishima, Osaka, Japan. ** Terry Fox Cancer Research Scientist of the National Cancer Institute of Canada. To whom reprint requests should be addressed. The abbreviations used are: AFP, a-fetoprotein; GRE, glucocorticoid responsive element; bp, base pairs; kb, kilobase pairs.

MATERIALSANDMETHODS

Cell Cultures-Human hepatoma cell lines, HUH-7 (14, 161, PLC/ PRF/5 (17), and Hep3B (18) were cultured in a chemically defined medium, IS-RPMI (19). HepG2 (18) was subcultured in IS-PRMI containing 1%fetal calf serum for 2 days and thenmaintained in ISRPMI without serum. HeLa, BM314 (human colon carcinoma, a gift from Dr. A. Yachi, Sapporo Medical College, Sapporo, Japan), and Ltk- were cultured in RPMI-1640 supplemented with 5% fetal calf serum. Northern BlotAnulysis-Total RNA wasisolated from cell cultures using the guanidinium isothiocyanate procedure (20). Northern blot analysis of AFP mRNA was conducted according to Thomas (21) using 3ZP-labeledAFP cDNA, pHAF-2 (22), as a probe. The amount of hybrids was quantified by densitometric scanning of the autoradiograms. Construction of CAT Fusion Genes-The structures of CAT fusion plasmids used in this work are shown in Fig. 2. They are designated according to the length (kb) of the AFP 5”flanking DNA; for instance, pAF0.4-CAT and pAF7.5-CAT contained 0.4 and 7.5 kb of the 5”flanking DNA, respectively. The fusion plasmids containing

266

Hormonal Regulation of the Humana-Fetoprotein Gene 0.4-7.5 kb of the human AFP 5'-flanking sequence are described previously (23). To construct the fusion plasmids containing 0.17 kb of the human AFP 5'-flanking sequence (pAFO.l7(Hi)-CAT, pAFO.l7(Bg)-CAT),the ClaI site located 6 bp upstream of the 5' end of the 1-kb AFP flanking sequence in pAF1.O-CAT was converted to a BglII site by the attachment of a synthetic linker. The resultant plasmid is called pAFl.O(Bg)-CAT. This was digested with HindIII to release the 980-bp (positions -951 to +29) AFP 5"flanking sequence. The remaining plasmid is called pBR(Bg)-CAT. The 980-bp fragment was digested with SstIto release the 198-bp fragment (positions -169 to +29). This was treated with mung bean nuclease, and theHindIII linker was attached and inserted at the HindIIIsite of pBR(Bg)-CAT to form pAFO.l7(Hi)-CAT. To construct pAFO.l7(Bg)-CAT, pAFl.O(Bg)-CAT was digested with SstI, followed bymung bean nuclease, and then theBglII linker was attached to it. The DNA was then digested with BgZII to remove the 788-bp BglIIISstI fragment, and the remaining DNA was ligated to form pBRO.l7(Bg)-CAT. To construct pAFO.l(Bg)-CAT, pAFO.l7(Hi)-CAT was digested with S a d , converted to blunt ends by the treatment with the large fragment of DNA polymerase I, and the BglII linker was attached. This DNA was digested with BglII to remove the 77-bp BglIIISauIfragment, and the remaining DNA fragment was ligated. To construct the AFP/SV40 early promoter/CAT fusion plasmids, the 198-bp DNA from positions -169 to +29 was converted to blunt ends by the large fragment of DNA polymerase I, the BglII linker was attached, and introduced to the BglII site of pSV1'-CAT (23) in normal (pSVAF0.17-CAT.) and reverse (pSVAF0.17[R]-CATa)orientations (Fig. 5A). The 198-bp fragment was also inserted to the BamHIsite of pSV1'-CAT in the normal orientation to form pSVAF0.17-CATb (Fig. 5A). Cell Transfection and ChloramphenicolAcetyltransferase AssaysCells were transfected using the calcium phosphateprecipitation method (24) according to Gorman et al. (25) with some modifications as described previously (23). Cells were grownin IS-RPMI andplated a t a densityof 5-7.5 X lo5cells/75 cm2flask 24 h priorto transfection. One hour before transfection, the medium was changed to fresh ISRPMI supplemented with 10% fetalcalf serum. The serum had been treated with charcoal and dextran to remove endogenous steroids (26). Each flask received 20 pg of DNA. Cells were incubated for 4 h at 37 "C, treated with 15% glycerol for 30 s, and rinsed with IS-RPMI medium. The cells were then incubated in IS-RPMI supplemented M dexamethasone for 48 h and analyzed for with or without 3 X chloramphenicol acetyltransferase activity as described previously (23). RU486 was dissolved in dimethyl sulfoxide and diluted with M. phosphate-buffered saline to make a stock solution of7.5 X This was added to theculture following the glycerol shock at thefinal M. concentration of 3 X RESULTS

Dexamethasone Increases AFP mRNA Concentration in HUH-7Cells-Earlier reports show that dexamethasone treatment of HUH-7 cells results in an increase in the AFP concentration in the medium (14). In this studywe analyzed the effect of dexamethasone on the level of AFP mRNA. We M dexamethasone, exincubated HUH-7 cells with 3 x tracted RNA at various times, and analyzed AFP mRNA by Northern blot hybridization (Fig. L4). The amountof hybrids was quantified by densitometric tracing and presented as a function of time (Fig. 1B). The AFP mRNA concentration increased significantly within 12 h of incubation with dexamethasone, reaching a plateau at 24-48 h with an overall increase of about 2.5-fold. In contrast, an increase in the AFP concentration in the medium was observed only after 42 h of dexamethasone treatment (14). Similar delays in the change of secreted protein levels have been reported in other systems (11, 27, 28), which likely reflect the time required for mRNA translation, protein processing, and secretion. In agreement with aprevious report (14),we found that the total number of cells was slightly lower (by about 10%) in dexamethasone-treated cultures than in thecontrol cultures. Delimitation of Glucocorticoid Responsive RegionsUpstream of the AFP Gene-To locate upstream DNA regions that are

267

A 2 kb-

0 12 24 36 48 60

hours

B

h

0

0

10

20

40

30

50

60

TI ME (hours) FIG. 1. Increase in AFP mRNAin HUH-7cells on incubation with dexamethasone. HUH-7 cells were incubated in the presence M dexamethasone. Total RNA was isolated a t indicated of 3 X times and analyzed for AFP mRNA. A , Northern blot analysis. Hours indicate the duration of dexamethasone treatment. B,autoradiograms shown in A were quantified by densitometric scanning and plotted against the time of incubation with dexamethasone. - -2-4K -6K 8b Kb Kbb 5' I

+29 I

l~&

I

A

I

I

cap,

3'

0

E -

pAF7.5-CAT

E pAF5.1-CAT E pAF3.7-CAT - - - -4 pAF3.5[A2]-CAT

1 .

(CATI (CATI

e

pAF2.9-CAT p A F l .E-CAT pAF1.0-CAT pAF0.4-CAT pAFO.l7(Hi)-CAT pAFO.l7(Bg)-CAT pAFO.l(Bg)-CAT

FIG. 2. Fusion genes bearing AFP 5"flanking sequences. Numbers on the top indicate the distance from the cap site of the human AFP gene. A and B indicate two AFP enhancer domains (23). Heavy solidlines indicate AFP 5'-flanking DNA. Broken lines indicate a deleted sequence.

involved in the dexamethasone responsiveness, we conducted transfection experiments using recombinant plasmids containing the CAT gene to which various sizes of the AFP 5'flanking DNA were fused. Fig. 2 shows the structureof these fusion genes. They were introduced into HUH-7cells by the calcium phosphate precipitation method. The transfected M dexamethacells were incubatedwith or without 3 X sone for 48 h and then assayed for chloramphenicol acetyltransferase activity. This concentration of dexamethasone has been shown to induce the highest level of AFP secretion (14). We also found indose-response analysis that chloramphenicol acetyltransferase was stimulated maximally in the presence

268

Human ofa-Fetoprotein the Gene

Hormonal Regulation

M dexamethasone (datanot shown). The Active Region Contains GRE Homologous Sequenceof3-6 X The results of chloramphenicol acetyltransferaseassays The nucleotide sequence of the human AFPpromoter region showed that the dexamethasone treatment stimulated the (29) is shown in Fig. 4A. In the region critical for the dexachloramphenicol acetyltransferase expression in HUH-7cells methasone responsive activity (positions -169 to -981,we transfected with plasmids containing 0.17-7.5 kb of the AFP found a TGTCCT sequence at positions -166 to -161 which 5’-flanking DNA, but not 0.1 kb (Fig. 3, A and B ) . This T is identicalwith the consensus GRE hexamer, TGTdCT indicates that sequences critical for dexamethasone responsiveness reside in the 71-bp region between 98 and 169 bp (30, 31). The 15-bp sequence, AGAGCTCTGTGTCCT (poupstream of the cap site of the AFP gene. pBR-CAT which sitions -175 to -161), that includes this hexamer and the lacks AFP DNA (23) exhibited no chloramphenicol acetyl- adjacent upstream nucleotides, showed 81, 69, and 63% hotransferase activity either in thepresence or absence of dex- mology to the GRE of rat AFP (32), human metallothionein amethasone (Fig. 3B). IIA (33), and growth hormone (34),respectively (Fig. 4B). To Although the dexamethasone effect was observed with a assess the importance of the nucleotides other than thehexwide range of sizes of 5”flanking DNA, the degree of stimu- amer for GRE activity, we altered six nucleotides at the 5’ lation of chloramphenicol acetyltransferase activity by dexa- end of the putative GRE by replacing them with the BglII or methasone varied greatly depending on the size of the 5’- HindIII linker a t position -169.As shown in Fig. 4C, the flanking sequence. This phenomenon will beanalyzed in more BglII linker restores five of six original nucleotides, whereas detail below. the HindIII linker restores only two, without counting one nucleotide gap in bothcases. The sequence modified with the BglII linker was found to retain 85% activity of the original A sequence, whereas the sequence modified with the HindIII linker exhibited only 55% activity (Figs. 3A and 4C). These +3-Ac results are consistent with the report that the nucleotides 5’ to theconsensus GRE hexamer play a role in dexamethasone C 1 -AC responsiveness (35). The Stimulation of Chloramphenicol Acetyltransferase Expression Is Mediated by Glucocorticoid Receptors-The antiglucocorticoid compound RU486 has been shown to strongly interact with the glucocorticoid receptor and to inhibit the action of dexamethasone (36). We examined whether RU486 affects the level of stimulation of chloramphenicol acetyltransferase activity by dexamethasone. RU486 (3 X hi) 1

-

2

3

5

4

6

7

8

9

1 0 1 1 1 2

+ - + - + - + - + - +

”“WL

1 . 82 3 . 9. 53A. 75 . 17 . 5

-120

-130

-140

-110

& x T A & A A T T A ~ ~ ~ G C ~ A A ~ C ~ ( ~ sou I

-100

B

4

~ -60

c

-90

T

A

-80

A

~

-70

~

~

T

~

&

~

~

~

T

A

T

.

~

~

-30

-fO

h A C T A G l T A A C A G 3 X ~ & ~ -10 tl

+ ~ - A c

-20

& m & m d +l-Ac

-70

A

-3‘

B h-AFP

-175 A G A G C T C T G T G T C C T-161

r-AFP

-175 A G T G G T C T

T GT T C C T-161

-262 G G T A C A C T G T G T C C T-248 +93 G G C A C A A T G T G T C C T+107 CONCENSUS G G c A% N N T G T c T

h-MT h-GH

8

~9~

1

2

3

5

4

6

7

8

9

1101 1 2

Dexamethasone

C

responsiveness

- + - + - + - + - + - + ””” 1.0

0.4

0.17(Hi) 0.17(Bg) O.l(Bg)

pBR

FIG. 3. Effect of dexamethasone on chloramphenicol acetyltransferase expressiondirected by various lengthsof AFP 5’-flanking DNA. HUH-7cells were transfected with various fusion M) or without genes shown in Fig. 2, incubated with (3 X dexamethasone for 2 days, and analyzed for chloramphenicol acetyltransferase activity as described under “Materialsand Methods.” Cm, chloramphenicol; I-Ac, 1-acetate chloramphenicol; 3-Ac, 3-acetate and - indicate the presence and absence of chloramphenicol. dexamethasone, respectively. A , lanes 1 and 2, pAF7.5-CAT; lanes 3 and 4 , pAF5.1-CAT; lunes 5 and 6, pAF3.7-CAT; lanes 7 and 8, pAF3.5[A2]-CAT;lanes 9 and 10, pAF2.9-CAT; and, lanes I 1 and 12, pAF1.8-CAT. B, lanes I and 2, pAF1.0-CAT; lanes 3 and 4 , pAF0.4CAT, lanes 5 and 6, pAFO.l7(Hi)-CAT; lanes 7 and 8, pAHO.l7(Bg)CAT; lanes 9 and I O , pAFO.l(Bg)-CAT; lanes 11 and 12, pBR-CAT.

+

~ - A F P -175

... .. A G A G

cT

IC

(%)

T G T G T c c ~1.161

100

85

HdndIII linker

FIG. 4. Nucleotide sequence and location of the human AFP GRE. A, nucleotide sequence of the promoter region of the human AFP gene. The sequence is taken from Sakai et al. (29). The cap site is numbered + I . The 15-bp stretches homologous to the GRE consensus sequence (30, 31) are underlined. Closed circles indicate the CCAAT pentamer. The TATA sequence is bored. B, comparison of the human AFP GRE with several other GREs. h-AFP, human AFP; r-AFP, rat AFP (32); h-MT, human metallothionein IIA (33); h-GH, human growth hormone (34). C, changes in sequence and activity of the human AFP GRE by the attachment of restriction site linkers. Closed circles indicate matched nucleotides.

~

A

A

Hormonal Regulation

Human of the

by itself had no effect on chloramphenicol acetyltransferase expression in HUH-7 cells transfected with pAF1.O-CAT, confirming that RU486 lacks agonist activity (Fig. 5) (26, 36, 37). However, the addition of RU486 to HUH-7cultures simultaneously with dexamethasone suppressed the stimulatory effect of dexamethasone onchloramphenicol acetyltransferase activity (Fig. 5). These results suggest that the stimulation of chloramphenicol acetyltransferase expression by dexamethasone is mediated by the glucocorticoid receptor. Relationship between GRE and Enhancer-Typical GREs have been shown to be similar to classical enhancer elements,

-

3-Ac

c 1 - A ~

“Cm

DEX: RU486:

- - + + - + + -

FIG. 5. Suppression of dexamethasone stimulation of chloramphenicol acetyltransferase activity byRU486. Transfection of HUH-7 cells with pAF1.O-CAT and analysis of chloramphenicol acetyltransferase activity were performed as described in the legend of Fig. 3. + and - indicate the presence and absence of dexamethasone (3 X lo-‘ M ) or RU486 (3 X lo-‘ M), respectively. Cm, 1-Ac, and 3Ac are as described in the legend of Fig. 3. Dex, dexamethasone.

A

-

-169

+29

+29 9f-,

a-Fetoprotein Gene

269

functioningindependently of the orientation and location relative to the gene to be acted on (38). To test whether the AFP GRE behaves in a similar manner, we inserted the 198bp region from positions -169 to +29 of the AFP gene 5’ to the CAT gene in pSVl’-CAT in normal and reverse orientations (Fig. 6 A ) . In another construct, the 198-bp DNA was inserted 3‘ to the CAT gene in normal orientation (Fig. 6 A ) . All these CAT plasmids expressed high chloramphenicol acetyltransferase activity in response to dexamethasone treatment (Fig. 6B). Neither pSV2-CAT (25)nor pSVl’-CAT which lacks the AFPDNA responded to dexamethasone (Fig. 6B). To examine the relationship between the AFP GRE and the upstream enhancer elements, we analyzed the level of stimulation of chloramphenicol acetyltransferase activity by dexamethasone as afunction of the size of the AFP 5’flanking DNA in the fusion genes (Fig. 7). In the absence of dexamethasone, the chloramphenicol acetyltransferase activity was low (about 1 pmol/h/mg protein) with 5”flanking DNA fragments up to 2.9 kb long (without the enhancer). Dexamethasone treatment increased the chloramphenicol acetyltransferase expression 10- to 15-fold. When the size of the 5”flanking DNA increased to 3.5 and 3.7 kb to contain domain B of the AFP enhancer a higher level of chloramphenicol acetyltransferase expression (about 8 pmol/h/mg protein) was obtained in the absence of dexamethasone. This activity was further increased 9- to 10-fold in thepresence of dexamethasone. When the size of the AFP 5’-flanking DNA increased to 5.1 and 7.5 kb to contain domain A as well as domain B of the AFP enhancer (23), the highest level of chloramphenicol acetyltransferase activity (about60 pmol/h/ mg protein) was obtained in the absence of dexamethasone, but the dexamethasone treatment resulted in only a 2- to 3fold increase in chloramphenicol acetyltransferase activity. These results show that the GRE and enhancer elements act

B pSVAF0.17-CATa pSVAFO.l7[R]-CATa

+33-A~

+1 - A C

+ Cm Amp‘

r29

-169

pSVAFO.l7[R]-cATb

+

1

-

2

+ sv2

3

4

5

6

7

8

9 1 0

+ - + - + - + - + ” ” 0.17[R]a 0.17a 0.17[R]b SV1’

FIG. 6. Effect of dexamethasone on chloramphenicol acetyltransferase expressionfrom fusion genes containing a 169-bp AFP 5”flanking sequence and the SV40 early promoter. A, construction of AFP/ SV40 promoter/CAT fusion plasmids. The 198-bp SstI/HindIII fragment between -169 and +29 bp of the AFP gene was inserted a t the BglII site or the BumHI site of pSV1’-CAT. B, HUH-7 cells were transfected with the CAT fusion plasmids described in A, incubated with (3 X lo-‘ M) or without dexamethasone, and then analyzed for chloramphenicol acetyltransferase activity. Cm, 1-Ac, and 3-Ac are as described in the legend of Fig. 3. + and - indicate the presence and absence of dexamethasone, respectively. Lanes 1 and 2, pSV2-CAT; lanes 3 and 4 , pSVAF0.17-CAT.; lunes 5 and 6, pSVAF0.17[R]-CATa;lunes 7 and 8, pSVAFO.l7[R]-CATb;lunes 9 and 10, pSV1’CAT

Hormonal Regulation

270

of Human the

a-Fetoprotein Gene DISCUSSION

Transient transfection analysis has proved to be of use in defining various cis-acting regulatory regions, including hormone responsive elements (28,39,40). In thisstudy, we show that thestimulation of AFP expression by dexamethasone in HUH-7 human hepatoma cells was mediated by the 71-bp region from 98 to 169 bp upstream of the human AFP gene. This stimulation was not due to general effects on cell growth since the total number of cells in the presence of dexamethasone was either similar to or only slightly lower than the controlcultures (14). The dexamethasonestimulation was effectively suppressed by the glucocorticoid antagonist RU486, indicating that thedexamethasone effect is mediated by glucocorticoid receptors. The location of AFP GRE is similar to that of the chicken lysozyme gene (41), the bovine prolactin gene (42), human metallothionein-IIA gene (33), I I I I I I I and several viral genes (15). The location of several other GREs hasbeen shown to vary greatly, from 3 kb upstreamof 0 1 2 3 4 5 6 7 the rabbit uteroglobin (43) and rat tyrosin aminotransferase 5"FLANKING SEQUENCE (kb) (31) genes to 100 bp downstream of the human growth horFIG. 7 . Relationship between dexamethasone stimulation of mone gene (see 33). chloramphenicol acetyltransferase activity and the length of The 71-bp region from positions -169 to -98 of the human AFP 5'-flanking DNA. Chloramphenicol acetyltransferase activities in HUH-7 cells transfected with various fusion genes shown in AFP gene that stimulates chloramphenicol acetyltransferase M) and absence of dexamethasone are expression contains TGTCCT at positions -166 to -161. Fig. 2 in the presence (3 X plotted against the size of AFP 5'-flanking DNA. 0, without dexa- This sequence is identical with the six nucleotides at the 3' methasone; 0, with dexamethasone. end of the GRE consensus sequence. This hexamer is well conserved among known GREs andshown to be essential for A B C D E F mediating dexamethasone effects (30). In addition, several nucleotides immediately upstream of this sequence have been L S - A C shown to play a role in dexamethasone responsiveness (35). Thus, we observed that a changeof four out of six nucleotides ' f l - A c in this region (without counting one gap) resultedin a much greater reduction of dexamethasonestimulation than a change of one nucleotide (Fig. 4C). Strahle et al. (35) have shown that a 15-bp sequence with partial symmetry is sufficient to confer glucocorticoid inducibility on a heterologous promoter. Our results are compatible with this conclusion and F F F F F F F F S F F S F F S strongly suggest that the sequence, AGAGCTCTGTGTCCT + + + + + + (positions -176 to -161), is infactan AFPGRE. This FIG. 8. Expression of the CAT gene from pAF1.O-CAT in conclusion was further supported by the observation that the various cell lines with orwithout dexamethasone. Various cell AFP GRE region was able to activate a heterologous (SV40) lines described below were transfected with pAF1.O-CAT or pSV2CAT in the presence (3 X 1O"j M) and absence of dexamethasone. promoter in response to dexamethasone in an orientationCm, 1-Ac, and 3-Ac are as described in the legend of Fig. 3. The and position-independent manner, thus behaving like classiamounts of cell extracts and the incubation times used in chloram- cal enhancers (15, 38). phenicol acetyltransferase assays were as follows: A, HepG2: 200 pg Chloramphenicol acetyltransferase expression from the fuof protein, 120 min; B, Hep3B: 150 pg of protein, 60 min; C, PLC/ sion genes with or without dexamethasone was seen in AFPPRF/5: 200 pg of protein, 60 min; D,BM314: 400 pg of protein, 90 non-AFP-producing cells. The lack min; E, Ltk-: 200 pg of protein, 135 min; F, HeLa: 1 mg of protein, producing cells, but not in 230 min. + and - indicate the presence and absence of dexametha- of dexamethasone response in non-AFP-producing cells is not due to lack of glucocorticoid receptors in these cells. More sone, respectively. Lane F , pAF1.O-CAT; lune S, pSV2-CAT. likely, the AFPpromoter functions in acell-specific manner, and the GRE cannot modulate a nonfunctioning promoter independently and that the stimulatory effect of dexametha- (15). sone is greater in the absence of the upstream enhancer than We have previously identified a typical enhancer element of the human AFP gene between -3.7 and -3.3 kb (domain its presence. Dexamethasone Effect Is Cell Type-specific-To examine B), but the maximum transcriptional enhancement is obwhether the effect of dexamethasone is dependent on cell tained together with an upstream region from -4.9 to -3.7 type, we transfected pAF1.O-CAT into different cell lines kb(domain A) (23). Analysis of the relationship of these upstream enhancer elements to the GRE showed that the which produce or do not produce AFP. Inthreehuman enhancer and the GREfunction independently. Thus, dexahepatoma cell lines which produce AFP(Hep3B, HepG2, methasone stimulation occurred both in the presence and PLC/PRF/5), dexamethasone treatment resulted in 4- to 12- absence of the enhancer and the effect of the enhancer was fold enhancement of chloramphenicol acetyltransferase activ- seen both in the presence and absence of dexamethasone. ity (Fig. 8). On the other hand, threecell lines which do not However, the magnitude of dexamethasone effect depended produce AFP (Ltk-, BM314, HeLa) expressed no chloram- on the basal level of expression. Thus, thehighest dexamethphenicol acetyltransferase activity in thepresence or absence asonestimulation (10-fold or more) was obtained inthe of dexamethasone. absence of the enhancer and the lowest stimulation (2- to 31

Hormonal Regulation of the Human a-FetoproteinGene fold) was obtained in the presence of the full complement of enhancer elements. It must be noted that although the magnitude of dexamethasone effect was greater in the absence of the enhancer, the overall level of chloramphenicol acetyltransferase activity was much lower (less than one-tenth) in the absence of the enhancer than in its presence. This is because the effect of the enhancer (approximately 150-fold stimulation) is much greater than that of dexamethasone (approximately 10-fold stimulation). It is tempting to speculate that the enhancer plays the major role in AFP gene expression during development (44), whereas the GRE modulates AFP gene activity in adult life. Dexamethasone treatment stimulates AFP synthesis in all human hepatoma cell lines so far examined (14), whereas either positive or negative response has been observed in several rat hepatoma cell lines (see Introduction). Guertin et al. (32) have recently reported that the region between positions -202 and -121 of therat AFP gene binds tothe glucocorticoid receptor and represses chloramphenicol acetyltransferase expression in the presence of dexamethasone. This region contains a sequence (positions -175 to -161) similar to the human AFP GRE (Fig. 4). This suggests that the same GRE element may mediate both stimulatory and suppressive effects of dexamethasone on AFP expression in rat hepatomas. It is not clear at present why no down regulation of the AFP gene by dexamethasone has been observed in human hepatoma cells. Acknowledgments-We wish to thankDr. R. Deraedt for the supply of RU486, and Richard Kennedy and Wendy Matsumoto for excellent technical assistance.

REFERENCES 1. Abelev, G. I. (1971) Adu. Cancer Res. 14,295-355 2. Tilghman, S. M., and Belayew, A. (1982) Proc. Natl. A d . Sci. U.S. A . 79,5254-5257 3. Tamaoki, T., and Fausto, N. (1984) in Recombinant D N A and Cell Proliferation (Stein, G., and Stein, J., eds) pp. 145-168, Academic Press, Orlando, FL 4. BBlanger, L., Hamel, D., Lachance, L., Dufour, D., Trembley, M., and Gagnon, P. M. (1975) Nature 256,657-659 5. Commer, P., Schwartz, C., Tracy, S., Tamaoki, T., and Chiu, J.F. (1979) Biochem. Biophys. Res. Commun. 8 9 , 1294-1299 6. BBlanger, L., Frain, M., Baril, P., Gingras, M.-C., Bartkowiak, J., and Sala-Trepat, J. M. (1981) Biochemistry 20,6665-6672 7. Schwartz, C. E., Burkhardt, A. L., Huang, D.-P., and Chiu, J.-F. (1982) Biosci. Rep. 2 , 777-784 8. Chou, J. Y., Mano, T., and Feldman, M. (1982) J. Cell Bwl. 9 3 , 314-317 9. Huang, D. P., Cote, G. J., Massari, R. J., and Chiu, J.-F. (1985) Nucleic Acids Res. 13,3873-3890 10. DeNechaud, B., Becker, J. E., and Potter, V. R. (1976) Biochem. Biophys. Res. Commun. 68,8-15 11. Cook, J. R., and Chiu, J.-F. (1986) J. Biol. Chem. 2 6 1 , 46634668 12. Tsukada, Y., Hibi, N., and Ohkawa, K. (1985) J. Biol.Chem. 260,16316-16320 13. Isaka, H., Umehara, S., Umeda, M., Hirai, H., and Tsukada, Y. (1975) Gann 66,111-112 14. Nakabayashi, H., Taketa, K., Yamane, T., Oda, M., and Sato, J.

271

(1985) Cancer Res. 45,6379-6383 15. Yamamoto, K. R. (1985) Annu. Reu. Genet. 1 9 , 209-252 16. Nakabayashi, H., Taketa, K., Miyano, K., Yamane, T., and Sato, J. (1982) Cancer Res. 42,3858-3863 17. Alexander, J. J., Bey, E. M., Geddes, E. W., and Lacatsas, G. (1976) S. Afr. Med.J. 5 0 , 2124-2128 18. Aden, D. P., Fogel, A., Plotkin, S., Damjanov, I., and Knowles, B. B. (1979) Nature 2 8 2 , 615-616 19. Nakabayashi, H., Taketa, K., Yamane, T., Miyazaki, M., Miyano, K., and Sato, J. (1984) Gann 7 5 , 151-158 20. Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual, Cold Spring HarborLaboratory, Cold Spring Harbor, NY 21. Thomas, P. S. (1980) Proc. Natl. Acud. Sci. U. S. A . 7 7 , 52015205 22. Morinaga, T., Sakai, M., Wegmann, T. G., and Tamaoki, T. (1983) Proc. Natl. Acud. Sci. U. S. A . 80,4604-4608 23. Watanabe, K., Saito, A., and Tamaoki, T. (1987) J. Biol. Chem. 262,4812-4818 24. Graham, F., and van der Eb, A. (1973) Virology 52,456-467 25. Gorman, C.M., Moffat, L. F., and Howard, B. H. (1982) Mol. Cell. Biol. 2 , 1044-1051 26. Coezy,E., Bouhnik, J., Clauser, E., Pinet, F., Philippe, M., Menard, J., and Corvol, P. (1984) In Vitro 20,528-538 27. Tank, A.W., Curella, P., and Ham, L. (1986) Mol. Phurmcol. 30,497-503 28. Lewis, E. J., Harrington, C. A., and Chikaraishi, D. M. (1987) Prm. Natl. Acud. Sci. U. S. A. 84,3550-3554 29. Sakai, M., Morinaga, T., Urano, Y., Watanabe, K., Wegmann, T. G., and Tamaoki, T. (1985) J. Bwl. Chem. 260,5055-5060 30. Scheidereit, C., Westphal, H. M., Carlson, C., Bosshard, H., and Beato, M. (1986) DNA 5 , 383-391 31. Jantzen, H-M, Strahle, U., Gloss, B., Stewart, F., Schmid, W., Boshart, M., Miksicek, R., and Schutz, G. (1987) Cell 4 9 , 2938 32. Guertin, M., LaRue, H., Bernier, D., Wrange, O., Chevrette, M., Gingras, M., and BBlanger, L. (1988) Mol. Cell. Bwl. 8, 13981407 33. Karin, M., Haslinguer, A., Holtgreve, H., Richards, R., Krauter, P., Westphal, H., and Beato, M. (1984) Nature 3 0 8 , 513-519 34. Slater, E. P., Rabenau, O., Karin, M., Baxter, J. D., and Beato, M. (1985) Mol. Cell. Biol. 5 , 2984-2992 35. Strahle, U., Klock, G., and Schutz, G. (1987) Proc. Natl. Acud. Sci. U. S. A . 84, 7871-7875 36. Moguilewsky, M., and Philibert, D. (1984) J. Steroid Biochem. 20,271-276 37. Oliver, N., Newby, R. F., Furcht, L. T., and Bourgeois, S. (1983) Cell 33,287-296 38. Chandler, V. L., Maler, B. A., and Yamamoto, K. (1983) Cell 3 3 , 489-499 39. Camper, S. A., Yao, Y. A. S., and Rottman, F. M. (1985) J. Biol. Chem. 260,12246-12251 40. Gustafson, T. A., Markham, B. E., Bahl, J. J., and Morkin, E. (1987) J. Biol. Chem. 84,3122-3126 41. von der Ahe, D., Renoir, J.-M., Buchou, T., Baulieu, E.-E., and Beato, M. (1986) Proc. Natl. Acud. Sci. U. S. A . 8 3 , 2817-2821 42. Beato, M., von der Ahe, D., Cato, A. C. B., Janich, S., Krauter, P., Scheidereit, C., Suske, G., Wenz, M., Westphal, H. M., and Willmann, T. (1985) in Glucocorticoid Hormones: Mechanism of Action (Sakamoto, Y., and Isohashi, F., eds) pp. 97-116, Japan Scientific Society, Tokyo 43. Suske, G., Wenz, M., Cato, A. C. B., and Beato, M. (1983) Nucleic Acid Res. 11, 2257-2271 44. Hammer, R. E., Krumlauf, R., Camper, S. A,, Brinster, R. L., and Tilghman, S. M. (1987) Science 235,53-58