Identification of an activating transcription factor (ATF) binding site in ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Vol. 267,No. 28, Issue of October 5, pp. 19938-19943,1992 Printed in U.S.A.

Identification of an Activating Transcription Factor(ATF) Binding Site in theHuman Transforming GrowthFactor-/32 Promoter* (Received for publication, March 6, 1992, and in revised form, June 18, 1992)

Michael A. O’ReillyS, Andrew G. Geiser, Seong-Jin Kim, Leslie A. BruggemanQ, AnhX. Luu, Anita B. Roberts, andMichael B. Sporn From the Laboratory of Chemoprevention, National Cancer Institute and the §Laboratoryof Developmental Biology, National Institute of Dental Research, National Institutes of Health, Bethesda, Maryland20892

Transforming growth factor TGF-82 is encoded by tionally similar. Both TGF-/31 and $2 were originally idenmultiple mRNA transcripts of 5.8, 5.1, 4.0, 3.8, and tified as unique proteins based upon their biologic activities, 2.8 kilobase pairs (kb) that are expressed in various whereas TGF-@3 was identified by cross-hybridization of human andmonkey cells. Northern blot analysis using cDNA clones (3). TGF-82was first identified by its ability to genomic fragments of DNA was used to demonstrate induce the formation of cartilage (4) and subsequently shown that some of this size heterogeneity is due to differences to inhibit cell growth (5, 6) and suppress the function of T in thelength of the 5”untranslatedregion. Probes that cells (7,8). Cloning of human andmouse cDNAs have revealed were colinear with the first 600 nucleotides of the 5‘- that TGF-/32 is expressed from multiple mRNA transcripts; untranslated region detected only the 5.8-, 4.0-, and 3.8-kb transcripts. In order toidentify DNA elements some of this size heterogeneity is due to the utilization of that regulate the transcription of these mRNA tran- different polyadenylation signals (6, 9). Moreover, at least in scripts, deletion constructs of 5’-flanking DNA were humans and monkeys, a 5.1-kb transcript has been shown to ligated to the coding region for chloramphenicol ace- contain an alternatively spliced second exon of 84 nucleotides, tyltransferase (CAT)and analyzed for promoter activ- resulting in an insertion of 28 amino acids in the aminoity in several cell lines. Sequences responsible for pu- terminal prodomain (10). The biological significance of these tative enhancer and silencer regions were identified differences in the structure of the TGF-/32 mRNAand protein between -778 and -40 relative to the transcription remains to be determined. Recent studies have focused on the cell- and tissue-specific initiation site. Addition of a cyclic AMP-responsive element/activating transcription factor-like element at expression of the three TGF-@ isoforms. In situ hybridization, -74 resulted in a 5-10-fold increase in CAT activity Northern analysis, and immunohistochemical studies have over that expressed with a construct that contained clearly demonstrated different patterns of expression for the only the TATA box.This increasein CAT activity was TGF-8 isoforms in the developing mouse and human (11-15). suppressed by the addition of DNA sequences between Moreover, the TGF-@gene promoters have been cloned, and -257 and -187, whereas sequences between -778 and sequence analysis has revealed striking differences in their -257 stimulated CAT activity. Point mutations within structure, suggesting that these differences in expression may the ATF binding site at -74 resulted in a marked be mediated through tissue-specific gene transcription (16decrease in CAT expression. Cotransfection with ATF1 or ATF-2 expression plasmids resulted in both dose- 20). We have recently published the sequence of the human dependent stimulatory and inhibitory activities that TGF-/32 5”flanking DNA and identified a transcription iniwere cell type-dependent. These studies identify mul- tiation site 1357 nucleotides upstream from the coding region tiple transcription initiation sites TGF-82 for and dem- (20). Fragments of 5‘-flanking DNA were cloned into a proonstrate that transcription from one of these promoters moterless pGEM4-SVOCAT plasmid, and modest promoter is dependent upon an ATF binding site located 5’ of activity was obtained in several cell lines with DNA that extended between -1728 and +63. In the present study, we the TATA box. extend those preliminary observations by identifying the mRNA transcripts that are transcribed from this promoter and show that this promoter is regulated, in part, through a Cellular growth and differentiation is a tightly regulated CRE/ATF binding site located 74 nucleotides 5’ of the TATA process that involves the strict coordination of many growth box. factors and hormones. Transforming growth factor-@ (TGFMATERIALS AND METHODS 8)’ is unique in that it acts as a multifunctional regulator of cell growth and differentiation (1, 2). Three distinct TGF-/3 Cell Culture”A549 (human pulmonary adenocarcinoma), BSC-1 genes have been identified in mammals to date. The encoded (African green monkey kidney), CCL-64 (mink lung epithelial), and proteins, TGF-Pl, 4 2 , and $3, are all structurally and func- PC-3 (human prostate adenocarcinoma)cells were obtained from the * 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 18 U.S.C. Section 1734 solely to indicate this fact. $ T o whom correspondence should be addressedLaboratory of Chemoprevention, Bldg. 41, Rm. C629, NCI/NIH,Bethesda,MD 20892. Tel.: 301-496-5391; FAX: 301-496-8395. ’ The abbreviations used are: TGF-/3, transforming growth factor0;CRE/ATF, cyclic AMP-responsive element/activating transcription factor; CAT,chloramphenicolacetyltransferase;kb, kilobase pair(s).

American Type Culture Collection. HeLa cells (human cervical carcinoma) were kindly provide by Dr. John Brady, National Cancer Institute, National Institutes of Health. A549, BSC-1, CCL-64, and HeLa cells were incubated at 37 “C and 5% CO, in Dulbecco’s modified Eagle’s media with 10% fetal bovine serum, 50 units/ml penicillin, and 50 pg/ml streptomycin(GIBCO).PC-3 cells were incubated at 37 “C and5% CO, in Ham’s F-12K media supplemented with 2 mM glutamine, 7% fetal bovine serum, and antibiotics. Cells were maintained in tissue culture flasks and routinely passaged every 3 days. For transfection experiments, cells were plated in 100-mm dishes at a density of 1 X 106/100-mm dish (A549, BSC-11, 7.5 X lo5/ 100-mm dish(CCL-64),or 2 X 106/100-mm dish(PC-3).These

19938

Transcription from Human the

TGF-p2 Promoter

19939

concentrations of cells were chosen based upon preliminary experiEcoRI (5’) and XbaI (3’) site. This insert contains70 nucleotides of ments, which demonstratedthatthe cells were confluent at the 5”untranslated region, the entire coding region, and 19 nucleotides conclusion of the experiment. Cells were transfected the next dayby of the 3’-untranslated region. The human ATF-2 cDNAwas cloned the calcium phosphate coprecipitation technique with 15pg of TGF- into the BglII site of the polylinker in pECE. This insert contains8 ~ 2 / C A Tfusion plasmid DNA and 1 pg of a human growth hormone nucleotides of the 5”untranslated region and the entire coding region expression plasmid under the control of the SV40 promoter (Nichols of the protein. Transcriptionof the inserts is under the control of the Institute, San Juan Capistrano, CA) (20). In some experiments the SV40 early promoter with polyadenylation signalsprovided by SV40 sequences. TGF-PZICAT constructs (15 pg) were cotransfected with different concentrations of expression plasmids containing human ATF-1 or ATF-2 cDNAs. After incubating thecells with the DNA solution for RESULTS 12-16 hours, the cells were washed threetimeswithphosphateIdentification of Multiple 5‘-Ends of the TGF-/32mRNAsbuffered saline and incubated in nutrientmedia for an additional 48 h. Cells were scraped from the culture dishes and lysed by freeze- Previous studies identified a 5579-nucleotide human genomic thawing. Cell lysates (100-400 pg) were incubated with 0.14 pCi of clone that contained the first 116 amino acids encoding [‘’Hlacetylcoenzyme A in a buffer containing 0.1 M Tris, pH7.8, 1.25 preproTGF-/32 (Fig. 1). Atranscriptioninitiationsite was mM chloramphenicol, and 0.11 mg/ml of unlabeled acetyl coenzyme identified 1357 nucleotides upstream from the methionine A (21). The samples were overlayed with Econoflour-2 scintillation initiation codon and promoteractivity using 5’-flanking DNA mixture (Du Pont) and incubated a t 37 “C for 24 h. In order to control was obtained in several cell lines (20). BecauseTGF-PB is fortransfection efficiency, CATactivity wasnormalized tothe translated from multiple mRNA transcripts, we were interamount of growth hormone secreted into themedia. RNA Extraction and Analysis-Total RNA was isolated from fro- ested in determining whichmRNA transcripts were tranzen cell lysates by sonication in 4 M guanidine thiocyanate, 0.5% scribed from this promoter. Poly(A)+RNA wasprepared from N-lauroylsarcosine, 20 mM sodium citrate,0.1 M P-mercapthoethanol, A549 cells and subjected to Northern blot analysis using a and 0.1% antifoam A. RNA was extracted by centrifugation through coding region cDNA probe. This probe detected five distinct a cushion of 5.7 M cesium chloride (22). T h e RNA pelletwas dissolved i n water, extracted three times with phenol/chloroform/isoamyl al- mRNA transcripts of 5.8, 5.1, 4.0, 3.8,and 2.8 kb (Fig. 2, lane cohol (25:24:1), once with chloroform/isoamylalcohol (24:1), and 1 ). The same RNA was also hybridized with probes specific precipitatedwithethanol.Poly(A)+RNA was preparedusing for the 5’-untranslated region of TGF-P2. Only the 5.8-, 4.0-, oligo(dT) columns (Pharmacia LKB Biotechnology Inc.). The amount and 3.8-kb transcripts were detected with an XbaI-BamHI of RNA in an aqueous solutionwas determined by absorbance at 260 probe that contained the previously mapped transcription nm. initiation site andthe first 600 nucleotides of the 5”untransRNA was separated on a 1.2% agarose-formaldehyde gel, transat 80 “C for lated region (Fig. 2, lane 2). All five TGF-P2 mRNA tranferred to Nytran (Schleicher and Schull, Inc.) and baked 2 h. Blots were stained with 0.3 M sodium acetate containing 0.02% scripts were detected using DNA probes that were 3‘ of this met.hylene blue to verify the accuracy of transfer. Blots were prehy- XbaI-BarnHI probe; DNA probes more 5’ failed to hybridize bridized and hybridized in 1% bovine serumalbumin,7%sodium to any of these mRNA transcripts (data not shown). Similar dodecyl sulfate, 0.5 M sodium phosphate, and 1 mM EDTA at 65 “C patterns of hybridization were also observed in other cell lines (23). Hybridized blots were washed in 1%bovine serum albumin, 40 that express these same five TGF-/32 mRNAtranscripts (data mM sodium phosphate, and 2 mM EDTA twice a t room temperature and twice a t 65 “Cfor 15 min before exposing to Kodak XAR-2film. not shown). Therefore, some of the size heterogeneity in the DNA Probes-RNA blots probed for expression of TGF-P2 were TGF-P2 mRNAs is due to different transcription initiation hybridized with a 1.2-kb simian TGF-02 cDNA (5) which contained sites, resulting in differences in the length of the 5”untransonly the coding region or with genomic fragments of DNA that are lated region. colinear with the 5”untranslatedregion, as previously mapped by S1 Basal Activity of the TGF-/32Promoter-In order to identify nuclease analysis (20). RNA blots probed for expression of ATF were DNA regulatory elements that are required for the expression hybridized with a 748-base pair EcoRI-XhoI fragment of the ATF-1 cDNA or a 1.27-kb EcoRI fragment of the ATF-2 cDNA. Blots were of the 5.8-, 4.0-, and 3.8-kb mRNA transcripts, deletion conDNAwere cloned intothe routinely probed with 3-5 X lo6cpm/ml of probe preparedby random structscontaining5”flanking primer labeling (Bethesda Research Laboratories) toa specific activ- promoterless pGEM4-SVOCATplasmid (Fig. 1).All promoter ity of 2 X lo9 cpm/pg using [32P]dCTP (3000 Ci/mmol)asthe fragments were created with a common 3‘-end of +63 with radionucleotide. Blots were stripped in 0.1 X SSPE (1 X SSPE is 150 different 5‘-ends that were designed to investigate putative mM sodium chloride, 10 mM sodium phosphate,pH 7.4,lmM EDTA) enhancer regions identified by computer analysis (20). Thus, and 0.2% SDS at 100 “C for 10 min. CAT Plasmids and Expression Constructs-All DNA constructs W weregenerated by polymerase chainamplification usinggenomic H B X B H 5 DNA containing the 5”flanking DNA and the first exon of the human , TGF-P2 gene as a template (20). DNAs were amplified by cycling 0 2878 3635 5579 ‘956 I temperatures of annealing at 50 “C for 2 min, elongating at 72 “C for 4 min, and denaturing a t 94 “C for 1 min, using Taq polymerase (Perkin-Elmer Cetus). The amplified DNA fragments were cloned into the promoterless CAT expression plasmid (pGEM4-SVOCAT) using HindIII and KpnI restriction sites built into oligonucleotides the used for amplification. All DNA constructs contained unique 5’-ends -778 1 ,, 082-778 pB2-528 with the same 3’-end of +63 relative to the transcription initiation I 1 pB2-437 site. The constructs were named pP2-n, where n is the distance in , I pB2-347 4 p82-257 nucleotides from the transcription initiation site. Constructs pP2-778 f ,4 p02-187 pB2-1 17 and pP2-528 are the constructs previously named pTGF-2 and pTGFc”i p52-77 3 (20). The mutant construct pp2-77M was created using an oligoc3 pB2-62 pB2-40 nucleotide containing the desired mutation (C toT at -71 and T to G a t -69), thereby changing the nucleotide sequence a t -74 from FIG. 1. Human TGF-82 genomic clone and construction of GCACGTCA to GCATGGCA. These oligonucleotides were used to TGF-B2/CAT constructs. A 5579-nucleotide human genomic clone amplify DNA that was cloned into the promoterless CAT expression containing the first coding exon (filled box) wassequenced and a plasmid. All CAT fusion plasmids were confirmed by sequence analy- transcription initiation site (arrow)was identified. ATG is the transsis and purified by two sequential CsCl banding steps for maximum lation initiation codon. Deletion fragments were generated by polympurity. erase chain reaction using different 5”specific oligonucleotides that The expression plasmidspECE-ATF-1andpECE-ATF-2 were span from -778 to -40 and a common 3”specific oligonucleotide at kindly provided by MichaelGreen. These plasmids contained the +63 relative to the transcription initiation site. Amplified DNAs were human ATF-1 and ATF-2 cDNAs (24) inserted into the expression cloned into the promoterless pGEM4-SVOCAT vector. B, BamHI; H , plasmid pECE (25). The human ATF-1 cDNA was cloned into the HindIII; X , XbaI. j

-

Transcription Human thefrom

19940

kb 5.8 5.1

4.0 3.8 2.8

1

2

FIG. 2. Northern blot analysis of TGF-82. Poly(A)+ RNA (5 pg) from A549 cells was separated on a 1.2% agarose-formaldehyde gel and transferred to Nytran. Blots were probed with a ‘”P-labeled simian TGF-P2 open reading frame probe (lane 1 ) or a 1.0-kb XbaIHamHI genomic probe (lane 2 ) thatcontainedthetranscription initiation site and the first 600 nucleotides of the 5”untranslated 1965 and 2878, Fig. 1). Size of the region (genomic DNA between mRNA transcripts isin kilobase pairs (kb).

pp2-40 contained onlya TATA box consensus sequence, pp262 began downstream of an CRE/ATF-like element, and pp277 began 3nucleotides 5‘ of theCRE/ATF-like element. Additional constructs were generated approximatelyevery 90 nucleotides between -77 and -778. The TGF-P2/CAT constructs were transfected into four cell lines which express TGF-02 mRNA and protein. Several patterns of promoter activity were observed that were dependent upon thecell line analyzed. In all cell lines, the two smallest constructs, pp2-40 and pp2-62, did not express significantlygreaterCATactivitythanthepromoterless pGEM4-SVOCAT plasmid (Fig. 3). In A549, CCL-64, and PC3 cells, pp2-77 (containing the CRE/ATF-like element) expressed 5-10-fold more CAT activity than the smaller constructs. Two longer constructs, pp2-117 and pp2-187, expressed similar orslightly less CAT activity in thesecells. In contrast, whereas pp2-77 expresseda modest 2-3-fold increase in CAT activity in BSC-1 cells, the two longer constructs, pp2-117 and p2-187, expressed even greater (8-10fold) CAT activity (Fig. 3B). In all four cell lines, the longer construct pp2-257 expressed significantly less CAT activity than the shorter constructs pp2-187, pp2-117, and pp2-77. Although this decrease in CAT activitywas most dramatic in CCL-64 cells, it was reproducibly observed in all cell lines. A longer construct, pp2-347, consistently expressed more CAT activitythan pp2-257. Finally, the level of CATactivity expressed by the longest constructs pp2-437, pp2-528, and pp2-778 varied and was dependent upon the cell line. Thus, although the overall levels of CAT activity varied between cell lines, some general trends were observed. Promoter sequences less than -62 did not express CAT activity greater than a promoterless construct; sequences between -77 and -62 promoted CAT activity;sequencesbetween -257 and -187 suppressed CAT activity, and sequences longer than -257 expressed increased levels of CAT activity, which was dependent upon thecell line transfected. Characterization of the ATF Element-In all four cell lines, pp2-77 was the smallest construct that expressed CAT activity greater than either the promoterless pGEM4-SVOCAT

TGF-B2 Promoter plasmid or pp2-62. Sequence analysis of the TGF-P2 gene identified a CRE/ATF-like element between -77 and -62 (Fig. 4A). This DNA element contains 6 of the 8 conserved nucleotides observed in other promoters that regulated are by this family of transcription factors (26,27). In order to determine whether thisDNA element is necessary for the expression of TGF-PP, two nucleotides that have been shown to be critical for a functional CRE/ATF binding sitewere mutated (26) (Fig. 4A). This mutant construct, pp2-77M, expressed only 10% of the CAT activity obtained with pp2-77 in A549, CCL-64, and PC-3 cells and approximately 40% of the CAT activity in BSC-1 cells (Fig. 4B). Thus, the CRE/ATF-like element in the TGF-P2 promoter at -74 is required for basal level promoter activity. Recent studies have identified a family of transcription factors that bind the consensus CRE/ATF element (TGACGTCA) (24). These proteins have a similar carboxylterminal leucine zipper motifthrough which the proteinsform homo- and heterodimers. We have recently found that glutathionine S-transferase-ATF-1 and glutathionine S-transferase-ATF-2 fusion proteins bound with similar affinities ina gel mobility shift assay to the TGF-82 CRE/ATF-like element a t -74 (data not shown).Moreover, these fusion proteins did not bind to oligonucleotides containing the mutated CRE/ ATF sequence inpp2-77M. In order to further investigate the role of ATF-1 and ATF-2 in the transcription of the TGF-P2 promoter, cotransfection experiments with expression plasmids containing ATF-1 or ATF-2 cDNAs were performed. A549, BSC-1, CCL-64, and PC-3 cell lines were transfected with pp2-778 in the presence or absence of pECE-ATF-1 or pECE-ATF-2 (Fig. 5). Cotransfection with ATF-2resulted in a marked increase in CAT activity in A549 and PC-3 cells but only a modest 2-3-fold increase in BSC-1 and CCL-64 cells. In A549, BSC-1, and PC-3 cells, cotransfection with ATF-1 resulted in less than a 2-3-fold increaseinCAT expression. In contrast, cotransfection of CCL-64 cells with ATF-1 resulted in a dose-dependentdecrease in promoter activity. Similar results were obtained in all four cell lines when pp2-77 was cotransfected with the ATF expression plasmids (data not shown). Furthermore, cotransfectionswith pp2-77M did not demonstrate significant changes in CAT activity, further supporting the concept that ATF-1 and ATF2 transactivate through the ATF binding site at -74 (data not shown). Several studies have demonstrated that ATF-1 and ATF-2 may form heterodimers with other members of the ATFfamily butnot with eachother (27-30). Inordertodetermine whether ATF-1 and ATF-2 may compete for binding to the ATF site at -74, A549, BSC-1, CCL-64, and PC-3,cells were transfected with pp2-778 and 1.25 pg of both ATF-1 and ATF-2 expression plasmids (Fig. 5, filled columns). The level of CAT activityexpressed by A549, BSC-1, and PC-3cells in the presence of bothATFproteins was not significantly greater than that expressed by 1.25 pg of ATF-2 alone. In contrast, CCL-64 cellscotransfected with both ATF plasmids expressed levels of CAT activity that approximated thelevel observed in the presence of 1.25pg of ATF-1. These data suggest that ATF-1 and ATF-2 compete for binding to the TGF-@2 promoter with ATF-2 having a greater affinity in A549, BSC-1, and PC-3 cells, and ATF-1 having a greater affinity in CCL-64 cells. The differential effects of ATF-1 andATF-2 on theexpression of the TGF-PP/CAT constructs may be due to the endogenous level of expression of ATFs in thevarious cell lines studied. In order to determine the level of ATF-1 andATF-2 expressed by these cell lines, poly(A)+ RNA was hybridized with cDNA probes specific for ATF-1 or ATF-2(Fig. 6). The level of ATF-1 and ATF-2 mRNAexpression was compared

T r a ~ ~Human ~ rthe ~from pt~on h. 1549

TGF-62 Promoter

19941

8. BSC-1 6

12

0

P s

0

5

a

Y 4

a

E 3

2

0

2

zFIG. 3. Basal activity of the TGFp2 promoter. The average CAT activity after normalization to growth hormone obtained from A549 (A), BSC-1 ( B ) , CCL-64 ( C ) ,and PC-3 (D) cells is graph-

2

a

9 1

B

Y 8 0

2

10

6

s 2

4

5) e

2

0

-778 -528 437.347

-257 -187-117-77 -62 . w

o

1

A.

TGF~BLCRVATF

CTAGCKGEACTTT

TGF-E MUTANT

GTAGCAIL%&CTTT

CREmTF

TGACGXA

increase in transactivation of the TGF-OBICAT constructs by ATF-2 in CCL-64 cells compared with A549 cells may be due to the higher level of endogenous ATF-2 expression in CCL64 cells. DISCUSSION

In this study we have identified cis-acting elements in the human TGF-02 gene promoter for which the 5.8-, 4.0-, and 3.8-kb mRNA transcripts are transcribed (Fig. 7). Deletion analysis demonstrated that minimal promoter activity was obtained with a DNA construct that contained a CREIATFlike element at -74. Although this DNA element deviates . . from the consensus CRE/ATF binding site, it was shown to KC-, CCL-fib be critical for promoter activity based upon both point muCELL LINES FIG. 4. Identification of a GRE/ATF sequence that is re- tation analysis and transactivation in the presence of ATF-1 quired for basal level promoter activity. A, the nucleotide se- or ATF-2 expression plasmids. Cotransfection studies idenquence of the TGF-62 promoter between -77 and -62wit.h the tified differential promoter transactivation that was dependunderlined region depicting those nucleotides found in a consensus ent upon the cell line studied. Thus, although ATF-2 stimuCRE/ATF binding site. The TGF-62 CRE/ATF sequence was mu- lated promoter activity (A549, BSC-1, CCL-64, and PC-3), tated attwo nucleotides (lowercase letters) creating a 4 for 8 nucleotide CRE/ATF sequence. B, the average CAT activity after normalization ATF-1 either modestly stimulated (A549, BSC-1, and PC-3) t o growth hormone of pp2-77 and pb2-77M from A549, BSC-1, CCL- or inhibited (CCL-64) transcriptional activity. Since these 64, and PC-3 cells is graphically represented. The values represent a cell lines express similar levels of ATF-1 mRNA, the differpercentage of the activity obtained with pp2-77 and are an average ential activation of the TGF-P2 promoter was most likely due of three experiments with similar results. to distinct cell type-dependent interactions of ATF-1with the transcription initiation complex. with that expressed by HeLa cells, whichare known to express This is the first study to identify a DNA element that is both ATF proteins (24). Hybridization with a cDNA for ATF- necessary for transcription of the human TGF-02 gene. The 1detected a single mRNA transcript of 2.5 kb in all cell lines consensus CRE/ATF element was originally identified as a (Fig. 6A). The level of expression of thistranscript was segment of DNA that was able to confer cAMP responsiveness approximately equivalent in allcell lines but significantly less to the phosphoenolpyruvate carboxykinase (GTP) gene and than that expressed by HeLa cells. Hybridization with a has subsequently been shown to confer cAMP responsiveness cDNA for ATF-2 detected a single mRNA transcript of 4.3 to a wide variety of genes (31). Moreover, transcription of the kb in all cell lines (Fig. 6B). The level of expression of this adenovirus E4 gene in response to E1A is mediated through transcript was approximately 3-&fold greater in CCL-64 cells a cellular protein called activating transcriptionfactor (ATF) than that expressed by the other cell lines. Thus, the stimu- that binds to the same nucleotide sequence. The ATF site is latory andinhibitory activities of ATF-1 on the expression of located adjacent to the TATA box and facilitates the formathe TGF-PBICAT constructs are notdue to differences in the tion of a preinitiation complex at thetranscription initiation level of ATF-1 mRNA expression. Furthermore, the smaller site (32). Molecular cloning and sequence analysis of cDNAs A549

PC3

Transcriptionfrom the Human TGF-@2Promoter

19942

A. A549 20

B. BSC-1

,

4,

I Z

FIG. 5. Promoter activity in the presence of ATF-1 and ATF-2. The

0

t, 3 n

I

3

L 3

IO

0

E

average CAT activity after normaliza- 1 tion to growth hormone of pb2-778 (15 0 LL pg) from A549 ( A ) , BSC-1 ( B ) ,CCL-64 (C), or PC-3 (D)cells cotransfected in 0 the presence or absence of various con0 centrations of pECE-ATF-1 (ticked bar) or pECE-ATF-2 (hatched bar) is graphically represented. The filled bar represents thelevel of CAT activity obtained C. CCL-64 from pP2-778 (15 pg) in the presence of 1.25 pgof pECE-ATF-1 andpECE-ATF2. The values represent ratios of the activity obtained with pp2-778 in the absence of cotransfectant and are an average of two to three cotransfections with similar results. 0

2

0 J

0

LL

1 0

0.31 0.63 1.25

2.5

5

ug DNA

ATF-I ATF-2

0

0.31 0.63 1.25

2.5

5

ATF-I ATF-2

5

ATF-I ATF-2

u g DNA

D. PC-3

5

n 2 0 LL

. 0

0.31 0.63 1.25

2.5

ug DNA

. 5

. ATF-I ATF-2

0

0.31 0.63 1.25

2.5

u g DNA

has identified at least 10 proteins that bind to theCRE/ATF element (24). These proteins have a similar basic DNA binding domain followed by a carboxyl-terminal leucine zipper - 2.5 (bZip) through which hetero- and homodimers may form (2730). Thus, theprotein that binds and regulates transcription in response to CAMP is called CREB, whereas ATF-2 me1 2 3 4 5 diates the simulatory activities of the E1A protein; each of these proteins belong to the bZip family and transactivates through a similarbinding site. The bZip domain has also been B. identified in several other transcription factors, such as Myc, Fos, c-Jun, JunB, Jun D, Fra, andGCN 4. The formation of -4.3 various homo- and heterodimers results in protein complexes that have slightly different stabilities in their interaction with DNA (30,33). More recently, it hasbeen shown that members 1 2 3 4 5 of the basic helix-loop-helix family of transcription factors FIG. 6. Northern blot analysis of ATF-1 and ATF-2. may also form heterodimers with members of the bZip family Poly(A)' RNA (2 pg) from HeLa (lane l ) ,A549 (lane 2), BSC-1 (lane (34). Since many second messenger signaling pathways, such 3 ) ,CCL-64 (lane 4 ) , and PC-3 (lane5) cells was separated on a1.2% as protein kinase A (CREB) andprotein kinase C (JunFos) agarose-formaldehyde gel and transferred to Nytran. Blots were mediate some of their effects through these proteins, the probed with '"P-labeled cDNA for ATF-1 ( A ) or ATF-2 ( B ) . Size of formation of homo- and heterodimers may easily result in the mRNA transcripts is in kilobase pairs (kb). complex gene regulation at a single binding site within a promoter. Numerous attempts have failed todemonstrate transactivation of the TGF-/32 promoter in response to either protein kinase A or C, suggesting that CREB or AP-1 does -778 -257 -187 -74 -30 + I +a10 < not bind to the CRE/ATF element at -74.' However, since E I s I E ~AHT numerous transcription factorsare known to bind the consen5 8 KB 5.1 KB 4 0 KB sus CRE/ATF sequence, it remains to be clarified whether 2.8 KB 3.8 KB second messengers, or other members of the bZip or basic FIG. I. Linear arrayof enhancer andsilencer regions inthe helix-loop-helix family, may regulate the TGF-PP promoter TGF-B2 gene promoter. The 5'-flanking region of the human TGF- through the ATF binding site in different cell types or phys8 2 gene is graphically represented as a line. The filled box represents the first coding exon. The transcription initiation site for the 5.8-, iologic conditions. In addition to theidentification of the CRE/ATF element, 4.0-, and 3.8-kb mRNA transcripts is depicted as an arrow a t +l. Based upon the Northern analysis, transcription of the 5.1- and 2.8- the present study has also identified additional putative enkb mRNA transcripts may initiate from either TATA or TATA-like hancers and silencers in the TGF-82 gene. Mutation analysis elements in the 5"untranslated region, indicated as anarrow at +840 of the CRE/ATF element (Fig. 4) and cotransfection experior from more distal regions of the TGF-02 gene. The cis-acting DNA ments (Fig. 5) suggest that the CRE/ATF element may not elements regulating the expression of the 5.8-, 4.0-,and 3.8-kb mRNA Kb

A.

+/-

0

transcripts, as identified by sequence analysis and functional assays, are represented as boxes. A, A T F E, enhancer; T,TATA box; S, silencer.

* M. A. O'Reilly, A. G. Geiser, S.-J. Kim, L. A. Bruggeman, A. X. Luu, A. B. Roberts, and M. B. Sporn, unpublished observations.

Transcription Human thefrom

TGF-/32 Promoter

19943

be as important for transcription in BSC-1 cells. Deletion been shown by both functional analysis and the conservation analysis revealed that greater promoter activitywas obtained of these elements between avians and mammals. The identiwith DNA extending between -187 and -77, suggesting the fication of trans-acting factors that interact with cis-acting DNA elements in the TGF-P promoters is beginning clarifyto presence of an enhancer in this region that was active in BSC-1 cells. A second putative element was identified in all mechanisms by which the various TGF-Pisoforms have both distinct andoverlapping patterns of expression. cell lines between -257 and -187 that acted to suppress promoter activity. Longer constructs between -778 and -257 Acknowledgments-We t h a n k R. W. Holley for supplying the simexpressed different levels of promoter activity that was deian T G F - 0 2 c D N A ,M. G r e e n and F. Liu for the expression plasmids pendent upon the cell line transfected. The identification of p E C E - A T F - 1 and p E C E - A T F - 2 , a n d J. B r a d y and F. K a s h a n c h i for these putative enhancers and silencers remains to be deterproviding HeLa cells and their critique of this work. mined. REFERENCES The identificationof multiple 5’-ends for the TGF-P2gene 1. Massague, J. (1990) Annu. Reu. Cell Biol. 6 , 597-641 has added additional complexity to the expression of TGF2. Roberts, A. B., and Sporn, M. B. (1990) in Handbook of Experimental p2. Previous studies have shown that TGF-/32 is encoded by A. B., eds)pp. 419-472, Pharmacology (Sporn,M.B.,andRoberts, Heidelberg, Germany several mRNA transcripts; some of this size heterogeneity is 3. tenSpringer-Verlag, Dijke, P., Hansen, P., Iwata, K. K., Pieler, C., and Foulkes,J. G. (1988) Proc. Natl. Acad. Sci. U. S. A. 85,4715-4719 due to the utilization of multiple polyadenylation sites (6, 9). 4. Seyedin, S. M., Segarini, P. R., Rosen, D. M., Thompson, A. Y., Bentz, H., T h e present study has demonstrated that the 5”untranslated and Graycar, J. (1987) J. Biol. Chem. 2 6 2 , 1946-1949 5. Hanks, S. K., Armour, R., Baldwin, J. H., Maldonado, F., Spiess, J., and region for the 5.8-, 4.0-, and 3.8-kb transcripts contain an Holley, R. W. (1988) Proc. Natl. Acad. Sci. U. S. A. 8 5 , 79-82 additional 600 nucleotides not found in the 5.1- and 2.8-kb 6. Madisen, L., Webb, N. R., Rose, T. M., Marquardt, H., Ikeda, T., Twardzik, D., Seyedin, S., and Purchio, A. F. (1988) DNA ( N Y ) 7,1-8 transcripts. Sequence analysis has identified several putative 7. Bodmer, S., Strommer, K., Frei, K., Seipl, C., De Tribolet, N., Heidi, I., TATA and TATA-like elements in the 5”untranslated region and Fontana, A. (1989) J. Immunol. 143,3222-3229 8. Wrann, M., Bodmer, S., de Martin, R., Siepl, C., Hofer-Warbinek, R., Frei, t h a t may indicate transcription initiation sites for the 5.1K., Hofer, E., and Fontana, A. (1987) EMBO J. 6 , 1633-1636 and 2.8-kb transcripts (19, 20). Northern blot analysis has 9. Miller,D. A,, Lee,A,,Pelton, R. W., Cben, E. Y., Moses,H. L., and Derynck, R. (1989) Mol. Endocrinol. 3, 1108-1114 shown that the5.1-kb TGF-P2 mRNA transcript contains an N. R., Madisen, L., Rose, T. M., and Purchio, A. F. (1988) DNA additional 84 nucleotides, resulting ina n insertion of 28 amino 10. Webb, ( N Y ) 7,493-497 Flanders, K. C., Ludecke,G.,Engels, S., Cissel, D. S., Roberts, A. B., 11. acids in the amino-terminal prodomain (10). A recent study Kondaiab, P., Lafyatis, R., Sporn, M. B., andUnsicker, K. (1991) using antisera specific for this insertion suggests that the Development 1 1 3 , 183-191 12. Gatherer, D., ten Dijke, P., Baird, D. T., andAkburst, R. J. (1990) longer TGF-P2 isoform may be differentially proteolytically Development 110,445-460 processed to smaller peptides than the shorter TGF-P2 iso- 13. Millan, F. A,, Denhez, F.. Kondaiah. P.. and Akhurst. R. J. (1991) Deuelnnrnont 1 1 1 131-ldd form (35).3 Althoughthe biologic significance of the two TGFpa, S., Moses, H. L., andHogan, B. L.M. (1989) p2 isoforms is presently notknown, the present study suggests 0 6 , 759-767 D., Bilbe, G., Maier, R., and McMaster, G. K. (1991) that each isform may be under different transcriptional reg!opment 1 1 1 , 117-130 ulation. A. G., Kim, S. J., Roberts, A. B., and Sporn,M. B. (1991) Mol. Cell. 11,84-92 Although the TGF-P1, TGF-P2, and TGF-P3 proteins are J., Glick, A,, Sporn, M. B., and Roberts, A. B. (1989)J. Biol. Chem. 264,402-408 highlyconserved among each other, sequence analysis has 18. Lafyatis, R., Lechleider, R., Kim, S. J., Jakowlew, S., Roberts, A.B., and revealed thatthereissubstantial divergence amongtheir Sporn, M. B. (1990) J. Biol. Chem. 265,19128-19136 19. Malipiero, U., Holler, M., Werner, U., and Fontana, A. (1990) Biochem. respective gene promoters. The TGF-P2 and TGF-P3 proBiophys. Res. Commun. 1 7 1 , 1145-1151 moters containclassic TATA box elements, whereas the TGF- 20. Noma, T., Glick, A. B., Geiser, A. G., O’Reilly, M. A., Miller, J., Roberts, A. B., and Sporn, M. B. (1991) Growth Factors 4 , 247-255 @lpromoter contains a high GC content and lacks a TATA 21. Neumann, J. R., Morency, C. A., and Russian, K. 0. (1987) BioTechniques consensus sequence (16-18,20). We have recently found that R Add -, the transcriptionof TGF-P1 and TGF-P3, but not TGF-P2, 22. is Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J., and Rutter, W. J. (1979) Biochemistry 18,5294-5299 regulated by Sp14 Transcription of the TGF-P1 gene is also 23. Church, G. M., and Gilbert, W. (1984) Proc. Natl. Acad. Sci. U. S. A. 8 1 , 1991-1995 increased by several oncogenes, including Jun and Fos (36), 24. Hai, T., Liu, F., Coukos, W. J., and Green, M. R. (1989) Genes & Deu. 3, Ras (16), and Src (37). Moreover, the product of the retino2083-2090 L., Clauser, E., Morgan, D. O., Edery, M., Roth, R., and Rutter, W. blastoma gene has been shown to stimulate transcription of 25. Ellis, J. (1986) Cell 4 5 , 721-732 the TGF-P1 gene through specific retinoblastoma response 26. Deutsch, P. J., Hoeffler, J. P., Jameson, J. L., Lin, J. C., and Habener, J. F. (1988) J. Biol. Chem. 2 6 3 , 18466-18472 elements (38) and the TGF-P2gene through interaction with 27. Hai, T., and Curran, T. (1991) Proc. Natl. Acad. Sci. U. S. A . 2 8 8 , 37203724 ATF-2 and the CRE/ATF element identified in the present 28. Hursi,-H. C., Totty, N. F., and Jones, N. C. (1991) Nucleic Acids. Res. 19, study (39). Both the human TGF-P2 gene and the human 4601-4609 29. Hoeffler, J. P., Lustbader, J. W., and Cben, C.-Y. (1991) Mol. Endocrinol. TGF-P3 gene contain an upstream CRE/ATF element that is R -, 9.5G-3GG ”” required for basal gene expression (18).The TGF-P3 CRE/ 30. Benbrook, D. M., and Jones, N. C. (1990) Oncogene 5 , 295-302 ATF element ismore homologous (AGACGTCA) to the con- 31. Roesler, W. J., Vandenbark,G. R., and Hanson, R. W. (1988)J. Biol. Chem. 263,9063-9066 sensus CRE/ATF element (TGACGTCA) than the TGF-P2 32. Horidoshi. M.. Hai. T.. Sin. Y.-S.. Green. M. R.. and Roeder. R. G. (1988) . , Cell 54,’ 1033-1042 CRE/ATFelement (GCACGTCA) andimpartsCAMPre33. Flint, K., and Jones, N. C. (1991) Oncogene 6 , 2019-2026 sponsiveness to the TGF-P3 gene. A recent study in which 34. Blanar, M. A,, and Rutter, W. J. (1992) Science 2 5 6 , 1013-1018 the sequence of the human andchicken TGF-P2 and TGF-P3 35. Madisen, L., Lioubin, M. N., Finerty, P. J., Sutter, K., Blake, J., Federick, J., and Purchio, A. F. (1991) Growth Factors 6, 317-325 genes were compared has revealed that although the coding 36. Kim, S.-J., Angel, P., Lafyatis, R., Hattori, K., Kim, K. Y., Sporn, M. B., Karin, M., and Roberts, A. B. (1990) Mol. Cell. Biol. 10, 1492-1497 regions are remarkably conserved between species, only the 37. Birchenall-Roberts, M. C., Ruscetti, F. W., Kasper, J., Lee, H.-D., FriedTATA and CRE/ATF elements conserved are in the promoter man, R., Geiser, A., Sporn, M. B., Roberts, A. B., and Kim, S.-J. (1990) Mol. Cell. B~ol.1 0 , 4978-4983 regions (40). The importance of the CRE/ATF elements for 38. Kim, S.-J., Lee, H.-D., Robbins, P. D., Busam, K., Sporn, M. B., and transcription of the TGF-P2 and TGF-P3 promoters has now Roberts, A. B. (1991) Proc. Natl. Acad. Sci. U. S. A. 88, 3052-3056

___

_”

’

M. A. O’Reilly, unpublished observations. A. G. Geiser, unpublished observations.

’

39. Kim, S.-J., Wagner, S., Liu, F.,O’Reilly, M. A,, Robbins, P. D., and Green, M. R. (1992) Nature 358,331-334 40. Burt, D. W., Paton, I. R., and Dey, B. R. (1991) J. Mol. Endocrinol. 7,175183