atf-ao, a novel variant of the atf/creb transcription factor

0 1994 by The American Society for Biochemistry and Molecular Biology, Inc. THE JOURNAL OF BIOLWICAL CHEMISTRY Vol. 269, No. 2, Issue of January 14, pp. 1159-1165, 1994

Printed in U.S.A.

ATF-aO, a Novel Variant of the ATF/CREB Transcription Factor Family, Forms a Dominant Transcription Inhibitor in ATF-a Heterodimers*

(Received for publication, March 24, 1993, and in revised form, June 7, 1993)

Rosanna Pescini, Wiweka Kaszubska, James Whelan, John F. DeLamarter, and Rob Hooft van HuijsduijnenS From the GLAXO Institute for Molecular Biology, Chemin des Auk , 1228 Plan-les-Ouates, Geneva, Switzerland

We have isolated a cDNA encoding a variant of the transcription factor ATF-a (called ATF-aO) by screening a HeLa cDNA expression library with a regulatory element of the E-selectin promoter, NF-ELAMl/GA. Relative to full-length ATF-a, the ATF-a0 cDNA contains a large in-frame deletion of 525 base pairs that removes the Pis/ T-rich putative transactivation domain. Using reverse- transcription-polymerase chain reaction and Northern blot hybridization to characterize ATF-a0 expression, we found that putative mRNAs for ATF-a0 and ATF-a are present at varying ratios in different tissues. Full-length ATF-a is a transcriptional activator for the NF- ELAMl/GA site of the E-selectin promoter. In contrast, we show ATF-a0 has no measurable transactivating function on this element. Moreover, we demonstrate that co-expressed ATF-a0 and ATF-a preferentially heterodimerize. In the heterodimer ATF-a0 is a dominant inhibitor that completely blocks the transactivating activity of ATF-a. Both forms ofATF-a bind the p50 subunit of NF-lcB as shown by affinity chromatography. ATF-a0 appears to be a splice variant similar to the one found for ATF-2, its closest homologue in structure and function. Taken together, our results suggest that ATF-a0 is an important member of the ATF family with a negative regulatory role in transactivation.

A number of viral and eukaryotic promoters harbor the CAMP responsive element (CRE)l/ATF binding site (TGAC- GTCA) (reviewed in Refs. 14). This site is recognized by a rapidly growing family of related ATF/CREB-like factors (5- 11). Although there is evidence that different members dis- criminate between DNA binding sites in vitro (12) and in vivo (13), selectivity and specificity for ATF transactivation is achieved by regulation at many levels. For example, phos- phorylation by CAMP-activated protein kinase A is required for activation of CREB (8, 14-17) and ATF-1 (18). In contrast, ATF-2 and ATF-a activity is stimulated by direct interaction with the adenovirus E1A (19,201 or retinoblastoma gene product (12). Functional cooperation or binding has been shown between ATFs and the serum response factor SRF (21), CTF/ NF1(22), the HBV X-protein (23), NF-KB (24,25), and nuclear

* 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.

$ To whom correspondence should be addressed. Tel.: 41-22-706-96-

The abbreviations used are: CRE, CAMP responsive element; CAT, chloramphenicol acetyltransferase; IL-1, interleukin 1; HUVEC, human umbilical cord endothelial cell; RT, reverse transcription; PCR, polymerase chain reaction; GST, glutathione S-transferase; bp, base pair(s); kb, kilobase paifis).

66; Fax: 41-22-794-69-65.

matrix proteins (26). In addition, ATF factors selectively heterodimerize with each other or with other members of the leucine zipper family, such as c-Jun (reviewed in Refs. 1 and 271, generating stimulatory or inhibitory complexes. As a conse- quence of these heterodimerizations, transcriptional cross-talk may be expected to occur between different signalling pathways. This has indeed been shown for the protein kinase C-APl( fos / jun) and protein kinase A-ATF-l/CREB pathways (28-30). Finally, the ATF-binding site can be inactivated by DNA methylation at the central CpG residues (31), representing yet another level of regulation.

We have recently described a new ATF-binding site with the recognition sequence TGACATCAT, which differs from the con- sensus by one base substitution (24). This sequence forms the “NF-ELAM1” element of the IL-l-inducible E-selectin promoter, where it cooperates with the neighboring NF-&-site (24). This E-selectin ATF-binding sequence has been conserved in the murine (32, 80) and rabbit (33) E-selectin promoters as well. The same element can also function as an independent enhancer in T-cells as the “GA-element” found in the T-cell receptor-cu and +, and CD3 promoters (34). Screening h g t l l expression libraries with this sequence as a probe yielded multiple ATF-like cDNA clones (35h2 One of the clones we isolated corresponds to a previously characterized ATF member, ATF-a (9). Interestingly, this clone has a very large deletion relative to ATF-a, covering 176 internal amino acids of the protein, with- out changing the following open reading frame. Comparison with other ATF-a variants suggests this form is generated by alternative splicing. ATF-2 (also called CRE-BP1, -2, or mXBP) is the most similar ATF-member to the ATF-a subgroup. Cloned from several mammalian species, ATF-2 cDNAs vary in their coding sequence due to deletions at similar location as in our isolated ATF-a variant (7, 36, 37). The alternative splice pat- tern is conserved through gene duplication and at least 60 million years of evolution as evidenced by ATF-a and ATF-2 cDNAs from various species.

In this report, we further characterize the ATF-a deletion variant (ATF-aO). By RT-PCR and Northern blot hybridization we show that mRNAs of full-length ATF-a and ATF-a0 are co-expressed in various human tissues. We also demonstrate that ATF-a0 is a strong dominant inhibitor of transcription.

EXPERIMENTAL PROCEDURES Agtll Expression Library Screening-A concatenated probe was

made by phosphorylating, annealing, and ligating (38) two oligonucleotides corresponding t o the E-selectin promoter NF-ELAMl site (24), GTAACACAGAGTM’CTGACATCATI’GTAAGC and TACGCTTAAAA’ITACAATGATGTCAGAAACTCTCTG. The sequence in

Kaszubska, W., Hooft van Huijsduijnen, R., Ghersa, P., de Raemy- Schenk, A. M., Chen, B. P. C., Hai, T., DeLamarter, J. F., and Whelan, J. (1993) Mol. Cell. Biol. 13, in press.

1159

1160 Dominant Inhibitor of ATF-a boldface corresponds to the recognition site. The "sticky ends" used for making the concatemers are underlined.

A commercial HeLa cDNA library in Agtll (Clontech) was screened as previously described (39,40). DNA inserts from positive clones were PCR-amplified (41) using the vector-specific oligonucleotides GA'lTG- GTGGCGACGACTCCT and CAACTGGTAATCGTAGCGAC.

In preparation for sequencing, the cloned fragments were biotinylated using PCR with one of the oligonucleotides containing 5"biotin groups. These oligonucleotides were prepared by synthesizing 5'-NHz- oligonucleotides on an Applied Biosystems RNA/DNA synthesizer using AminolinkII (Applied Biosystems) according to the manufacturer's protocols. The 5'-NH2-oligonucleotides (1 mg in 800 pl of 0.1 M NaHCOa buffer, pH 9) were then biotinylated by mixing with 200 pl of 10 mg/ml N-hydroxysuccinimidobiotin (Sigma; a stock solution was prepared in dimethylformamide). After overnight incubation at room temperature, the DNA was purified by chromatography on Sephadex G-50 Fine (Pharmacia LKB Biotechnology Inc.).

The biotinylated insert was then bound to streptavidin-agarose beads and the non-biotinylated strands were eluted using alkali as denaturing agent (42). The single-stranded DNA was sequenced on the resin using vector oligonucleotides as primers (43).

Reverse-Dunscription PCR-Poly(A) RNA was prepared from loga- rithmically growing HeLa and HUVEC cells using a "Quickprep mRNA" kit (Pharmacia). cDNA synthesis was camed out in 100 pl containing 6 pg of RNA, 5 pmoVplATFa-C2 oligonucleotide (see below), 0.5 m~ (each) dNTP, 80 unitdml RNAain (Promega), 1 univpl MuMLV reverse tran- scriptase (Life Technologies, Inc.), buffer, and salts (44). After 1 h at 42 "C, the mixture was heated at 100 "C for 5 min, and 10 pl w e h e d in a 100-pl PCR (41), using oligonucleotides ATFa-C2 and -N2. An' aliquot of the reaction mixture (10 pl) was used for a second PCR, using oligonucleotides ATFa-C1 and -N1. The oligonucleotides for the PCR were as follows (with the position in &A in parentheses; numbering as in Ref. 9): ATFa-Nl, GTAGGATCCGCTCCTCTCTTATATG (-24); ATFa-N2, GCTGGAGACAGATTGTAGGA (-85); ATFa-C1, GCGA- GATCTGGCTGAGGACCTCTCC (~1490); ATFa-C2, GACATCTCTCT- TGGCTCTTG (+1515).

SouthernlNorthern Blot Hybridization-The Southern blot hybridization with PCR products was performed as described elsewhere (44). The nitrocellulose filter was washed at room temperature 20 and 40 min in 2 x SSC, 0.1% SDS; 1 h at 65 "C in 0.2 x SSC, 0.1% SDS; and 30 and 45 min at 70 "C in 0.1 x SSC, 0.1% SDS. The procedure used for the multiple tissue Northern blot was as described by the membrane manufacturer (Clontech), except that the blot was washed as the Southern blot described above. Probes were labeled using random hexamer oligonucleotides as primers according to the protocol from the supplier (New England BioLabs). The PISm-probe, spanning the PISPT-rich exon that is spliced out in ATF-aO, corresponds to the NcoI fragment running from position 680 to 854 bp (numbering as in Ref. 9). The BZIP-probe is a mixture of the EcoRI fragment from -100 to +230 bp and the PstI fragment from position 1035 to 1335. The PISPT-probe corresponds to amino acids 227-285 and the BZIP-probe to amino acids 1-78 and 345-445.

Plasmid Constructs-The ATF-a2 expression vector (pATFa) was a kind gift from Dr. C. Kedinger (Strasbourg). This vector is derived from pSG5 (45) and contains the EcoRIISacI ATF-a fragment (-103 to +1409 bp). We modified this vector for the expression of the ATF-a0 cDNA. First, the BamHI-Sac1 fragment containing the ATF-a2 cDNA with extra 5' polylinker of pSG5 was subcloned into pBluescript SKII' polylinker which had the HincII-EcoRV fragment cut out in order to remove the AccI-site from the polylinker. ATF-a2 has a unique AccI site in the region upstream of the deletion in ATF-aO. The AccI-Sac1 fragment of the ATF-a2-Bluescript construct was substituted with the corresponding fragment from our ATF-a0 cDNA clone. Finally, the BamHI- Sac1 fragment of the ATF-aO-Bluescript construct was recloned into pSG5 vector and sequenced.

Cell Culture-Pmpagation of HeLa cells was ag described elsewhere (32). Interleukin-1 (40 unitalml) and control treatment were performed in fresh medium.

DNA Dansfection and CAT Assays-DNA transfection of HeLa was by the calcium phosphate co-precipitation procedure (46) using CsCI- purified, phenol-extracted plasmid. Per transfection, 2.5 pg of ATF- expression vector were used (or 5 pg when both full-length ATF-a2 and ATF-a0 were tested together) and 5 pg of CAT-expression vector. If necessary, pSG5 plasmid (Stratagene) was included to make the total amount of DNA always 10 pg. CAT enzymatic activity was camed out as previously described (47), using equal amounts of protein as meas- ured by the Bradford assay (48). The radioactivity in the chromato- grams was directly scanned using an Ambis (San Diego, CA) radioana-

lytic imaging system, and the percentage of acetylation was calculated. Band ShiB Assays-Band shift assays were carried out a t room tem-

perature in 25 pl of BCA(10 m~ Tris-C1, pH 7.75, 7.5 m~ EDTA, 10 p~ @-mercaptoethanol, 0.1% Triton X-100,5% glycerol, 80 m~ NaC1.3 m~ MgCl,, 0.5 mgml bovine serum albumin, 0.0025% bromphenol blue, 20 pg/d polyt.C)I (Pharmacia)) with 3 pl of in vitro translation reaction (or twice with 1.5 pl for the mixture of ATF-a2 and ATF-aO) and a double- stranded, end-labeled oligonucleotide (GATCTGACATCATKXATC; 10,000 cpm) carrying the human E-selectin promoter NF-ELAM1 element (24). ARer 20 min of incubation, the samples were loaded on 4% acrylamide gel and run at 30 mAin 0.5 x %s-borate-EDTA(TBE) buffer (44) at room temperature, using a cooling lower compartment (Hoefer). Gels were dried and exposed for 18 h at -80 "C with intensifying screens.

In Vitro DamcriptionlDanslation of ATF-a0 and Full-length ATF-a2-Proteins were synthesized in vitro in the presence of 3sS- labeled methionine using a coupled transcriptiodtranslation reticulo- cyte lysate system according to the manufacturer's instructions (Pro- mega). The products were analyzed by SDS-polyacrylamide gel electrophoresis and quantitated using a radioimaging system (Ambis, San Diego, CA). Equal amounts (countdmin) of radiolabeled proteins were used in the GST or GST-p50 binding assays.

Preparation of GSTpBO Sepharose and Affinity Chromatography- The p50 subunit (amino acids 35-381) of NF-KB was constructed as a fusion with GST in plasmid pGEX-2T and was a gift from R. Hay (St. Andrew's). The purification of GST and GST-p50 was as described elsewhere (49), except that the proteins were not eluted from glutathione- Sepharose beads (Pharmacia). 35S-Labeled proteins (2-5 pl) were di- luted to 100 pl with buffer containing 20 m~ Tris-HC1 (pH 8.0), 2 m~ dithiothreitol, 10 mg/d bovine serum albumin, and 0.1 M NaC1. The affinity beads carrying 12.5 pg of GST or GST-p5O (25 pl of 1:l suspen- sion) were added, and the slurry was incubated overnight on a rotary shaker at 4 "C. The resin was washed with 500-pl aliquots of the same buffer until the unbound proteins were removed (2.0-2.5 ml). The bound proteins were then eluted with 10 m~ reduced glutathione in 50 m~ Tris-HC1 (pH 8.0) and 0.5 M NaCl by rotating the slurry for 30 min at 4 "C. The proteins were subjected to SDS-polyacrylamide gel electrophoresis. The gel was fixed in 40% methanol and 10% acetic acid, incubated in Ample (Amersham Corp.), dried, and autoradiographed.

RESULTS Novel ATF-a Variant Isolated from HeLa Cells-We wished to

identify transcription factors binding the NF-ELAMllGA promoter element. An oligonucleotide probe encompassing this sequence was used to screen a HeLa cDNA Xgtll expression library. From this screening we isolated several cDNA clones of the ATF family (a full accounting will be presented elsewhere).' One of these clones contains a portion of ATF-a, a recently described ATF family member (Ref. 9; EMBL accession nos. X52943 and X57197). However, sequencing of our clone revealed that it has a large internal deletion covering nucleotides 405-929 (numbering as in Ref. 9, with +1 for the A of the initiation codon). The open reading frame following the deletion is still retained. The starting point of this deletion corresponds precisely to the end point of a small deletion found previously in a splice variant ATF-aA, recently renamed ATF-a1 (9, 20). We designate our ATF"a variant with the large deletion as ATF-aO. Based on published work (9) we believe that our cDNA clone results from an alternative splicing event as well. Fig. la gives an overview of these three different variants of ATF-a and the location of the various deletions. Fig. I b shows an amino acid sequence alignment of full-length ATF-a2 with ATF-2, its most closely related family member in structure, relative abundance, and function. Interestingly, for ATF-2 a splice variant with a large deletion has been described as well (7). Comparison of the domains which may be removed by splicing in ATF-2 and ATF-a2 (boxed in Fig. l b ) shows that they occur in the same region, even starting at the same corresponding amino acid position. Fig. I b also shows that the deleted regions in ATF-a0 and the ATF-2 variant are rich in the uncharged amino acids proline, serine, and threonine (P/Sfl'). This characteristic is also found in transactivating domains of other transcription

Dominant Inhibitor of ATF-a 1161

a A T F d ATF-a1 ATF-a0

b ATF-a:

ATF-2

ATF-a:

ATF-2

ATF-a:

ATF-2

ATF-a:

ATFQ:

ATF-a:

ATF-2

ATF-a:

ATF-2

ATF-a:

ATF-2:

ATF-a:

ATF-2

ATF-a:

ATF-2

ATF-a:

ATF-2

ATF-a:

ATF-2

l..................MGDDRPFVCNAPGCGQRFTNEDHLAVHKHKHE

1 M K F K L H V N S A R Q Y K D L H N M S D D K P F L C T A P G C G Q R F T N E D H L A V H K H K H E

3 3 M T L K F G P A R T D S V I I A D Q T P T P T R F L K N C E E V G L F N E L A S S F E H E F K K A A

S l M T L K F G P A R N D S V I V A D Q T P T P T R F L K N C E E V G L F N E L A S P F E N E F K K A S

8 3 D E D E K K A A A G P L D M S L P S T P D I K I K E E E P V E V D S S P P D S P A S S P C S P P L K

101 E D D l K K M . . . P L D L S P L A T P I 1 R S K l EEPSVVETTHQDSPLPHPESTTSD

I : 1 1 : I I : 1 . I I I I I I I I I 1 1 1 1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1 1 1 . 1 1 1 1 : I I I I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 . I I : 1 1 1 1 1 .

: : I I I I I I : I . I I 1 : I I l l I : . . . . I l l . I I . . .

I ~ ~ E E E V T P K P V L I S T P T P T I V R P G S L P L H L G Y D P . . . . . . . L H P T L P S P T S . 1 1 1 1 . . . . I I . . I I I I : I I . : . . . .

1 4 8 E K E V P L A Q . . T A Q P T S A l V R P A S L Q V P N V L L T S S D S S V I I Q Q A V P S P T S S : : . . : I I I I I

1 7 5 . V I T Q A P P S N R Q M G S P T G S L P L V M H L A N G Q T M P V . . . . . L P G P P V O M P S V

1 9 6 T V I T Q A P S S N R P I V P V P G P F P L L L H L P S G Q T M P V A I P A S 1 T S S N V H V P A A I I I I I I . I I I . : . . . . I . : I I : : I I : . I I I I I I : . : . 1 : : l . .

2191 S L A R P V S M V P N I P G I P G P P V N S S G S I S P S G H P I P S E A K M R L K A T L T H O V

L V R P V T M V P S V P G I P G P . . . . . . . . . S S P Q P V Q S E A K M R L K A A L T Q O H I . I I I . I I I . : I I I I I I . I . : I : . 1 1 1 1 1 1 1 1 1 . I I : I

2 6 9 S S I N O G C G U V V G T A S T Y V T A R P E Q S Q I L I Q H P D A P S P A Q P Q $ S P A Q P T P . . . . . I : I I : l . : l . . . I : I . : . . I . I . . : . . . I I I : . l l

2 8 7 P P V T N G . D T V K G H G S G L V R T Q S E E S R P Q S L O Q P A T S T T E T P A S P A H T T P Q

318. . S T G G R R R R T V D E D P D E R R O R F L E R N R A A A S R C R ~ K R K L ~ V S S ~ ~ . ~ . ~ A . E

3 3 6 T Q S T S O R R R R A A N E D P D E K R R K F L E R N R A A A S R C R O K R K V ~ V Q S ~ - ~ U ~ ~ E 1 1 : 1 1 1 1 1 . . : 1 1 1 1 1 : 1 . : 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 : 1 1 I I I I - l - I t

3 6 6 E ~ . ~ t . ~ ~ - ~ ~ ~ ~ ~ - ~ - " W f ~ ~ ~ ~ ~ ~ ~ L , ~ ~ ; Y . ~ - u L ~ . ~ ~ L L A H K D C P V T A L Q K K . . . . T Q G Y L

: I . . . I I I I . l ' l l ~ l t l l l l l l l l l l l l l l l l l l l l : l l l I . : 3esDLSS~L1SQOLO&EVTI ;LRNEVAO'L 'KQCLLAHKDCPVTAUQKKSGYHTADKD

4 1 2 E S P K E S S E P T G S P A P V I Q H S S A T A P S N G L S V R S A A E A V A T S V L T Q M A S Q R

4 3 6 D S S E D I S V P S S P H T E A I Q H S S . V S T S N G V S S T S K A E A V A T S V L T Q M A D Q S

4 6 2 T E L S M P I Q S H V l M T P Q S Q S A G R . .

4 8 5 T E . . . P A L S Q I V Y A P S S Q S Q P S G S

: I . . : I I . : . . . . . I I I I I . . . I I I : I I 1 1 1 1 1 1 1 1 1 1 1 1 1 . 1 .

I I I 1 : : : l . l I l l . . .

32

50

82

100

132

147

174

195

218

245

268

286

317

335

365

385

411

435

461

484

483

505

The location of the BZIP and P/S~T-probes and various domains are indicated. A fourth ATF-a3 species was recently described (20) that cames an FIG. 1. Dinerent forms of ATF-a a, schematic representation of full-length ATF-a2 and the two forms carrying deletions, ATF-a1 and ATF-aO.

extra 10-amino acid insertion at position 90 (not represented here). b, amino acid sequence (ATF-a2, EMBL X52943; ATF-2, EMBL X15875) comparison between full-length ATF-a2 and ATF-2. The boxed area corresponds to the large domain that is deleted in ATF-a0 and mXBP2(CRE- BP2). Shaded areas correspond to the basic region (light gray) and leucine zipper domain (dark gray).

factors, namely Spl(50). CTF/NF-1(51), and Octl(52), and is mRNA. To test whether mRNA for ATF-a0 is expressed, we responsible for the activity of these factors (53). Furthermore, it used the sensitive RT-PCR technique using RNA from two dif- has been shown that the deleted ATF-2 form has reduced trans- ferent human cells types, HeLa and HUVEC (54,55). Two sets activation activity (37). of nested primers in two successive series of amplifications by

Both ATF-a2 and ATF-a0 Occur Naturally in Cellular PCR analysis were used. As the primers are located at the Poly(A) RNA-The ATF-a0 cDNA we isolated could have been a amino and carboxyl termini of the full-length ATF-a2 open cloning artefact rather than representing a naturally occurring reading frame, they should amplify both full-length and inter-

Dominant Inhibitor of ATF-a

a

1162

1.6 1.1 *

P/S/T/-probe BZIP-probe

FIG. 2. RT-PCR producta obtained using poly(A) RNA from HU- VEC or HeLa cells. a , synthesis of cDNA and PCR were performed using nested oligonucleotides just bordering the open reading frame. M, size markers (“1-kb ladder,” Life Technologies, Inc.). The sizes of two major PCR products (kilobase pairs), corresponding to full-length ATF-a2 and ATF-a0 are indicated. h, blot hybridization of the gel shown in ( a ) using the ATF-a2-PISPT-probe. c, blot hybridization of the gel shown in ( a ) using the ATF-a-BZIP-probe.

nally deleted ATF-a variants. This approach yielded a band of 1.6 and 1.1 kb in HUVEC and a band of 1.1 kb in HeLa cells (Fig. 2a). The sizes of these products correspond to full-length ATF-a2 and ATF-aO, respectively. To test whether these bands correspond to ATF-a sequences, the DNA was transferred to a nitrocellulose filter and hybridized with probes corresponding to the variable PISIT domain (PISIT-probe) and to the invari- ant flanking sequences (BZZP-probe; the position of the probes is indicated in Fig. la). As expected, the BZIP-probe hybridizes to both bands, whereas the PISIT-probe recognizes only the largest product (Fig. 2, b and c). The 1.1-kb band in HeLa is a doublet, possibly representing multiple ATF-a0 species. An overexposed film from the hybridization in Fig. 2b demonstrated that the full-length ATF-a form is expressed in HeLa as well, although significantly less ATF-a was PCR-amplified from this cell line (data not shown).

As the RT-PCR results showed that several ATF-a forms are detectable in RNA-samples, we turned to Northern blot hybridization to further support the natural existence of ATF-a0 mRNA and establish its tissue distribution. RNA from various human tissues was hybridized with probes that detect either both ATF-a forms (Fig. 3a, BZZP-probe) or the full-length form only (Fig. 36, PISIT-probe). Both ATF-a-probes detect an ap- proximately 8.3 kb species as reported previously (9). However, only the BZIP-probe reveals a second band of about 1.3 kb, presumably corresponding to ATF-aO-mRNA. The very large size difference between the two mRNAs is remarkable and suggests that the ATF-aO-mRNA is transcribed from a different

9.5 >

4.4 * 7.5 t

2.4 > 1.35 *

b a - C - a

> * * > t

> t t

> t

BZIP-probe P/S/T-probe actin-probe

FIG. 3. Northern blot hybridization of poly(A) RNA from several human tissues. a, the blot was probed with the ATF-a-BZIP- probe; 6 , same as in a , using the ATF-aL-P/Sm-probe; c, same as in a , using a p-actin probe. skel. muscle, skeletal muscle.

transcription start point andlor has undergone extensive proc- essing of its untranslated regions. Alternatively, the 1.3-kb species may have undergone more extensive splicing than that found in our ATF-a0 cDNA clone. In this case ATF-a0 may exist a t a concentration that is below detection by the Northern analysis. The generation of multiple bands by RT-PCR (espe- cially with HeLa cell RNA) suggests that additional RNA species may exist.

Hybridizations shown in Fig. 3 suggest that expression of different ATF-a forms is tissue-specific; the highest levels for the 8.3-kb species are found in muscle, but no signal is obtained from liver. Moreover, the ratio between the upper and lower bands is not constant. Heart, brain and placenta tissue have equivalent amounts of the 8.3-kb species, yet among these the smaller 1.3-kb form is only apparent in heart RNA. In placenta and lung tissue, the P/SIT-probe shows a faint intermediate band a t 2.1 kb. This band might represent yet another splice form, though we have not further investigated this. In addition, the blot was hybridized with a p-actin probe as a control for the RNA-loading (Fig. 3c). The p-actin probe hybridizes to 2.0- and 1.6-kb species, the latter mostly in heart and skeletal muscle (561, and indicates that RNA loaded on the blot has not been degraded.

ATF-a0 Acts as a “Dominant Repressor” Dunscription Factor-To test the biological activity of ATF-aO, an expression vector was constructed for transient transfection assays. The cells were co-transfected with vector carrying the chloramphenicol transferase (CAT) gene under the control of the E-selectid ELAM promoter. The promoter fragment used (233 bp) contains the NF-ELAM116Aelement recognized by ATF-a (24, 32h2 Expression of recombinant, full-length ATF-a2 leads to transcription from this promoter (Fig. 4). We have previously shown that this transactivating activity is mediated by the NF- ELAMlIGA site located at position minus 150 in the E-selectin promoter; mutation of this site abolishes ATF-a’s transactivation.’ When transfected cells were treated with IL-1, the natural inducer of the E-selectin-promoter, an increased response was observed as compared to either IL-1 or ATF-a expression alone. These results suggest that the normal, cellular concentration of ATF-a is suboptimal for full E-selectin activity and precludes “leaky” transcription of the gene. In agreement with this, HeLa cellular levels of ATF-a have been estimated to be about 20-fold lower than either ATF-1 or CREB (20). In contrast, ATF-a0 expression was unable to activate the E-selectin promoter either by itself or in combination with IL-1 (Fig. 4). We next tested the transcriptional activity of an equimolar mixture of full-length ATF-a2 and ATF-a0 expression vectors. As shown in Fig. 4, ATF-a0 dominantly inhibited transcrip-

Dominant Inhibitor of ATF-a 1163

IL-1 + + + + pATF-a2 + + + + pATF-a0 + + + + pBR322

pE-eelectidCAT+ + + + + + + +

”

carrying full-length ATF-a2 or ATF-a0 and an E-selectin pro- FIG. 4. HeLa co-transfection assay using expression vectors

moter-driven CAT vector. The cells were induced or not with IL-1. The lowerpart of the figure is a numerical representation (radioactivity scanning) of the data. Results shown are of one typical experiment out of three.

tional activity of ATF-a2. The simplest explanation is that the two forms heterodimerize, and a single P/S/T domain of the dimer lacks activity when bound to the promoter. The blockage of the normal activity of ATF-a2 by ATF-a0 co-expression strongly suggests that ATF-a0 is a dominant repressor partner in an ATF-a2/ATF-aO heterodimer.

ATF-a0 and ATF-a2 Preferentially Form Heterodimers and Each Alone Can Bind NF-KB-pSGA general characteristic of factors containing leucine zipper domains is their capacity to form dimers. To investigate whether this is the case for ATF-a2 and ATF-aO, we synthesized both factors separately or together in a coupled in vitro transcription-translation system. As shown by translation in the presence of [“%]methionine (see Fig. 61, this method yields predominantly full-length protein products. These proteins were then assayed for their ability to bind the NF-ELAMl/GA sequence element in a band shift assay. Using full-length ATF-a2 or ATF-a0 alone in the assay yielded single bands of different mobility (Fig. 5). When the two forms were co-translated (Fig. 5, lane 41, almost all of the retarded complex migrated with an intermediate mobility. Since co-expression resulted in a band of intermediate mobility as compared to those of ATF-a2 or ATF-a0 alone, our results indicate that both ATF-a forms bind DNA as dimers and not as mono- mers. All complexes formed appear to bind DNA with roughly equal efficiency since equal intensity bands are observed (similar amounts of recombinant protein were loaded in lanes 2-5). In the lane with co-translated products almost all the retarded radioactivity is found in a band with intermediate mobility. This implies that the proteins preferably bind to form heterodimers, as random association would give rise to a theoreti- cal ratio of 1:l for heteromer versus homodimers. The small amount of full-length ATF-a2 homodimer seen in lane 4 is prob- ably due to a slight excess of this form with respect to the deleted species. Mixing of both ATF-a forms after translation does not yield any heterodimers (Fig. 5, lane 5). This result suggests that either dimerization is a co-translational process, or dimers, once formed, do not readily exchange partners.

Recently we have demonstrated that functional cooperation between the E-selectin NF-ELAMlBA and NF-KB elements

1 2 3 4 3

0 ATF-a2-dimer - ATF-a2/ ATF-a0 ATF-aO-dimer

length Am-& (Zue 2 ) or ATF-a0 (Zane 3). Lune 4, both forms FIG. 5. Gel retardation assay with in vitro translated full-

co-translated; lane 5 , mixed after translation; lane 1, control lysate. The probe contained the NF-ELAM1/6AATF-like site.

(24) is established through direct protein-protein interaction between ATF-a and the NF-KB subunits p50 and ~ 6 5 . ~ Using affinity chromatography (as in Fig. 6) we found that p50 bound ATF-a2 but not other leucine-zipper-containing proteins such as ATF-4 and c-Jun. We therefore asked whether ATF-a0 is also able to bind the NF-KB p50 subunit. We used affinity chromatography with p50 bound to beads to measure 35S-labeled ATF-a binding to p50. After extensive washing any complexes formed were eluted from the column and revealed by autora- diography of a denaturing gel loaded with the eluate. Both ATF-a forms, but not luciferase (negative control), bind p50 with similar affinities (Fig. 6). Moreover co-translated ATF-a2/ ATF-a0 heterodimers also bind the column (data not shown). The ability of the co-translated heterodimer to bind p50 with apparently equal efficiency together with our previous results suggest functionally distinct regions of the proteins exist. Dif- ferent domains appear to confer heterodimerization, p50 interaction and direct transactivation of the promoter.

DISCUSSION

We have isolated an ATF-a cDNAclone encoding an apparent splice variant of this factor that lacks 176 amino acids as compared to its full-length counterpart. Using RT-PCR and North- ern blot hybridization we show that both full-length and short forms are expressed in various tissues. A similarly deleted variant has been described for murine and human ATF-2, suggest- ing that these variants represent a relevant feature that is conserved through gene duplication and independent evolution of rodents and primates. Interestingly, for ATF-2/mXBP as well two bands can be detected on Northern blots, of 6.2 and 3.0 kb (57). These bands may correspond to mRNAs encoding full- length and P/S/T-minus species. We also show that full-length ATF-a2 is a strong transactivator binding the variant ATF element TGACATCA, while ATF-a0 is a dominant repressor on this element. Dimeric transcription factor partners that can act as dominant suppressors have been described for members of the leucine zipper (10, 37, 58-60), homeodomain (61-64), and rel-like families (65-72). In most cases they are also derived from the same gene by alternative splicing (see Refs. 27 and 73 for recent reviews). Only one example (LAPLIP) is known where both activator and repressor are translated from the same mRNA, using different in-frame AUG initiation codons

1164 Dominant Inhibitor of ATF-a

GST + + + GST-pSO + + +

elucidation of ATF-aO's tissue-specific splicing mechanism are important areas for future work.

Acknowledgments-We thank Dr. C. Kedinger for kindly providing the full-length ATF-a expression vector and Dr. R. Hay for the GST-p50 fusion vector. We are grateful to J.-C. Ramuz, E. Bader, C. Losberger, and G. Ayala for skilful technical assistance.

Frc. 6. Binding of radiolabeled Am-&, ATF-aO, and luciferase to affinity Sepharose. GST, protein eluted from resin containing GST only; GSTp50, protein eluted from resin containing GST-NF-KB-~~O fusion protein.

(74). Although the physiological role of in-frame tissue-specific splicing of transcription factor mRNAs is not &mpletely un- derstood (e.g. see Refs. 75 and 76), most likely this is an important means for regulating transcription.

The precise mechanism of dominant repression in the case of ATF-a2/ATF-aO is not clear. For a synthetic c-jun monomer containing a duplicated leucine-zipper but a single transactivating domain, the "heterodimer" still has transactivating properties (77). For this factor, at least, the lack of a second activator domain does not lead to dominant repression. Pos- sible mechanisms of ATF-a0 repression are "squelching" of co- factors required for transcription initiation, or "help" for repression by other factors that specifically recognize and bind the ATF-a2/ATF-aO heterodimer. In support of the latter possi- bility is the example of the liver specific POU-transcription factor HNFlLFBl. Dominant inhibition of HNFlLFBl repression is tissue-specific, the active multimer being stabilized by an accessory protein (62, 78).

Our data are illustrative of the modular structure of ATF-a: a P/Sfl"rich domain can be distinguished for independent transactivation; a basic domain for DNA binding; an amino- terminal "zinc finger" required for recruitment of adenovirus E l a protein (20) and a leucine zipper essential for dimerization. Our data show that the domain involved in binding N F - K B - ~ ~ O does not coincide with the transactivating domain, since ATF-a0 binds p50 despite its lack of this region. Recently, direct interaction between ATF-a and adenovirus Ela has been shown to depend on the NH2-terminal zinc finger domain of ATF-a (20). On the other hand, interaction between NF-KB and NF- IL-6 occurs via NF-KB-~~O'S re1 domain and NF-IG6's leucine zipper (79). We have used the NH2-terminal, rel-like portion of N F - K B - ~ ~ O for our affinity column and found that other leucine zipper factors, c-jun, ATF-2, and ATF-3 (but not ATF-4 or homeodomain factor oct-3) also show a low efficiency of binding to ~ 5 0 . ~ Based on these observations, we hypothesize that the interaction between p50's re1 domain and ATF-a depends on the latter's leucine zipper as well, rather than its amino-terminal zinc finger.

Taken together, our results suggest that the ATF-a0 variant of ATF-a we isolated has an important role in transcription regulation. ATF-aO's ability to interact with NF-KB as well as ATF-a suggests several protein-protein interactions occur which could mediate "cross-talk" between external and internal signals at the transcription level. In addition ATF-aO's prefer- ential heterodimerization with and repression of full-length ATF-a is an important example of a means by which transcription is regulated. Other leucine zipper domain factors with which ATF-a is able to heterodimerize are unknown. Any partners of ATF-a2 are also likely to be negatively regulated by ATF-aO. Identification of these binding partners of ATF-a and

1. 2. 3. 4. 5.

6.

7.

8.

9.

10. 11. 12.

13.

15. 14.

16.

17.

18.

20. 19.

21. 22.

23. 24.

25.

26.

27. 28. 29. 30.

31. 32.

33.

34. 35.

36.

37.

38.

39

REFERENCES Ziff, E. B. (1990) 7bends Genet. 6,69-72 Walton, K. M., and Rehfuss, R. P. (1990) Mol. Neurobiol. 4, 197-210 Brindle, P. K.. and Montminy, M. R. (1992) Cum Opin. Genet. Deu. 2,199-204 Hoefiler. J. P. (1992) J . Invest. Dermatol. 98,21S-28S Hai, T., Liu. F., Coukos. W. J.. and Green, M. R. (1989) Genes & Deu. 3,

Hoefiler, J. P., Meyer, T. E.. Yun, Y., Jameson. J. L., and Habener, J. F. (1988)

Ivashkiv, L. B.. Liou, H.-C., Kara. C. J., Lamph, W. W.. Verma. I. M., and

Gonzalez, G . A,, Yamamoto. K. K., Fischer, W. H., Karr. D., Menzel, P., Biggs,

Gaire, M., Chatton. B., and Kedinger, C. (1990) Nucleic Acids Res 18, 3467-

Meyer, T. E., and Habener, J. F. (1992) Nucleic Acids Res. 20,6106 Foulkes, N. S.. Borelli, E., and Sassone-Corsi, P. (1991) Cell 64,739-749

Kim, S. J.. Wagner, S.. Liu, F., 0 Reilly, M. A,, Robbins. P. D., and Green, M. R.

Nichols, M., Weih, F., Schmid. W.. DeVack, C., Kowenz Leutz, E., Luckow, B..

Gonzalez, G . A,, and Montminy, M. R. (1989) Cell 59,675-680 Lee, C. Q.. Yun. Y., Hoemer, J. P., and Habener, J. F. (1990) EMBO J. 9,

Hagiwara, M., Alberts. A., Brindle, P., Meinkoth, J., Feramisco, J.. Deng. T.,

Lamph, W. W.. Dwarki. V. J., Ofir, R.. Montminy, M., and Verma. I. M. (1990)

Rehfuss. R. P., Walton, K. M.. Loriaux, M. M., and Goodman, R. H. (1991) J.

Liu, F., and Green, M. R. (1990) Cell 61, 1217-1224 Chatton. B., Bocco, J. L., Gaire, M., Hauss. C., Reimund, B., Goetz, J., and

Morgan, 1. M.. and Birnie, G . D. (1992) Cell. Proliferation 25,205-215 Kedinger, C. (1993) Mol. Cell. Bid . 13,561-570

Muro, A. F., Bernath. V. A,. and Kornblihtt. A. R. (1992) J. B i d . Chem. 267,

Rossner, M. T. (1992) J. Med. Virol. 36, 101-117 Hooft van Huijsduijnen, R., Whelan, J., Pescini, R., Becker-Andre, M., Schenk,

Lee, M.-R.. Chung, C.-S.. Liou, M.-L., Wu, M., Li, W.-F., Hsueh, Y.-P.. and Lai,

Dworetzky, S. I., Wright, K. L., Frey, E. G., Penman, S., Lian, J. B., Stein, J. L.,

Lee, K. A. (1992) J. Cell Sci. 103.9-14 Masquilier. D.. and Sassone-Corsi, P. (1992) J. B i d . Chem. 267,22460-22466 de Groot. R. P., and Sassone-Corsi, P. (1992) Oncogene 7,2281-2286 Inagaki, N., Maekawa, T., Sudo, T., Ishii, S.. Seino, Y., and Imura, H. (1992)

Whelan, J.. Ghersa, P., Hooft van Huijsduijnen, R., Gray, J., Chandra, G., Iguchi-Ariga, S. M. M., and Schaffner, W. (1989) Genes & Deu. 3,612-619

Larigan, D., Tsang, T. C., Rumberger, J . M.. and Bums, D. K. (1992) DNA Cell Talabot, F., and DeLamarter, J. (1991) Nucleic Acids Res. 19, 2645-2653

Georgopoulos, K., Galson. D., and Terhorst, C. (1990) EMBO J. 9, 109-115 B i d . 11, 149-162

Georgopoulos, K., Morgan, B. A,, and Moore, D. D. (1992) Mol. Cell. Riol. 12,

Maekawa, T., Sakura, H., Kanei-Ishii. C., Sudo, T., Yoshimura, T., Fujisawa,

Ivashkiv, L. B., Fleming, M. D., and Glimcher, L. H. (1992) New Biol. 4,

Kadonaga, J. T.. Carner, K. R., Masian, F. R.. and Tjian, R. (1987) Cell 51,

Singh, H., LeBowitz, J. H.. Baldwin, A. S., and Sharp, P. A. (1988) Cell 52,

2083-2090

Science 242, 1430-1433

Glimcher, L. H. (1990) Mol. Cell. Biol. 10, 1609-1621

W., Vale, W. W.. and Montminy, M. R. (1989) Nature 337.749

3473

(1992) Nature 358,331-334

Boshart, M., and Schutz, G. (1992) EMBO. J . 11,3337-3346

44554465

Karin, M., Shenolikar, S., and Montminy, M. (1992) Cell 70, 105-113

Proc. Natl. Acad. Sci. U. S. A. 87,43204324

Biol. Chem. 266, 18431-18434

12767-12774

A. M.. and DeLamarter. J. F. (1992) J. Riol. Chem. 267,22385-22391

M. Z. (1992) J. Immunol. 146.1906-1912

and Stein, G . S. (1992) Proc. Natl Acad. Sci. U. S. A. 89,4178-4182

Proc. Natl. Acad. Sci. U. S. A. 89, 1045-1049

747-757

J.-I., Yoshida, M., and Ishii. S. (1989) EMBO J. 8.2023-2028

360-368

1079-1090

414-423 40. Vinson, C. R.. LaMarco, K. L.. Johnson, P. F., Landschulz, W. H., and

41. Hooft van Huijsduijnen. R. (1992) Bio'lkchnigues 12,8304332 42. Hultman, T., StAhl, S., and Uhlh, M. (1989) Nucleic Acids Res. 17,49374946 43. Sanger, F., Nicklen, S., and Coulson,A. R. (1977) Proc. Natl. Acad. Sci. U. S. A.

74,5463-5467 44. Ausubel, F. M., Brent. R., Kingston, R. E., Moore, D. D., and Seidrnan. J. G.,

( 1989) Current Protocols in Molecular Biology, Wiley Interscience, New York 45. Green, S., Issemann, I., and Sheer, E. (1988) Nucleic Acids Res. 16, 369 46. Whelan, J., Poon, D., Weil, P. A,, and Stein, R. (1989) Mol. Cell. Riol. 9,

47. Gorman, C. M., Moffat, L. F., and Howard, B. H. (1982) Mol. Cell. Biol. 2,

48. Bradford, M. (1976) Anal. Biochem. 72,248-254 49. Matthews, J. R.. Wakasugi. N., Virelizier, J.-L., Yodoi, J., and Hay, R. T. (1992)

McKnight, S. L. (1988) Genes & Deu. 2,801-806

3253-5259

1044-1051

50. Courey, A. J., and Tjian, R. (1988) Cell 55,887-898 51. Mermod, N., ONeill, E. A,, Kelly, T. J.. and Tjian, R. (1989) Cell 58, 741-753

Nucleic Acids Res. 20,-3821-3830

Dominant Inhibitor of ATF-a 1165 52. Sturm, R. A, Das, G., and Herr, W. (1988) Genes & Deu. 2, 1582-1599 53. Seipel, I€, Georgiev, O., and SfhafTner, W. (1992) EMEO J. 11,4961-4968 54. Ferre, F. (1992) PCR Methods Applications 2,l-9 55. Becker-And&, M., (1993) Absolute Levels of mRNA by Polymerase Chain Re-

oetion-Aided Zkunscript 'IEtmtion Assay, pp. 420445, Academic Press, New York

56. Lamballe, F., Klein, R., and Barbacid, M. (1991) Cell 66, 967-979 57. Liou, H.-C., Boothby, M. R., and Glimcher, L. H. (1988) Science 242.69-71 58. Karpinski, B. A,, Morle, G. D., Huggenvik, J., Uhler, M. D., and Leiden, J. M.

59. Ron, D., and Habener, J. F. (1992) Genes & Lku. 6,439-453 60. Nehlin, J. O., Carlberg, M., and Ronne, H. (1992) Nucleic Acids Res. 20,

61. Treacy, M. H., He, X., and Rosenfeld, M. G. (1991) Nature 360,577-584 62. Nicosia, A,, Tafi, R., and Monaci, P. (1992) Nucleic Acids Res. 20,5321-5328 63. Moms, A. E., Kloss, B., McChesney, R. E., Bancruft, C., and Chasin, L. A.

64. Sun, X.-H., Copeland, N. G., Jenkins, N. A,, and Baltimore, D. (1991) Mol. Cell.

65. Wulnyn, F. G., Naumann, M., and Sfheidereit, C. (1992) Nature %8,597-599 66. Franzoso, G., B o w , V., Park, S. , lbmita-Yamaguchi, M., Kelly, K, and Sieben-

67. Ruben, S . M., Klement, J. F., Coleman, T. A., Maher, M., Chen, C.-H., and

(1992) Proc. Natl. Acad. Sci. U. S. A 89,48204824

5271-5278

(1992) Nucleic Acids Res. 20,1355-1361

Eiol. 11,56033611

lisy, U. (1992) Nature 369,339-342

68. Inoue. J. I., Kerr, L. D., Kakizuka, A., and Verma, I. M..(1992) Cell 68, 1109- Rosen, C. A. (1992) Genes & Deu. 6,745-760

1120 69. Haskill, S., Beg, A. A., lbmpkins, S. M., Moms, J. S., Yurochko, A. D., Samp

s0ndohannes.A.. Mondal, K, Ralph, P., and Ba1dwin.A S. J. (1991) Cell 66, 1281-1289

70. Geisler, R., Bergmann, A., Hiromi, Y., and Niisslein-Volhard, C. (1992) Cell 71, 613-622

71. Kidd, S. (1992) Cell 71, 623-636 72. Baeuerle, P. A,, and Baltimore, D. (1988) Science 242,540-546 73. Foulkes, N. S., and Sassone-Corsi, P. (1992) Cell 68,411-414

75. Li, X.-Y., Hooft van Huijsduiinen, R., Manbvani. R., Benoist, C., and Mathis, 74. Descombes, P., and Schibler, U. (1991) Cell 67, 569-579

76. Ruppert. S., Cole, T. J., Boshart, M., Schmid, E., and Scbiitz, G. (1992) EMBO

77. Deng, T., and Karin, M. (1992) Prm. Natl. Acad. Sci. U. S. A 89,85724676 78. Mendel. D. B., Khavari, P. A., Conley, P. B., Graves, M. K, Hansen, L. P.,

79. LeClair, K. P., Blanar, M. A., and Sharp, P. A. (1992) Proc. Natl. Acad. Sci.

80. Becker-And& M., Hooft van Huijsduijnen, R., Losberger, C., Whelan, J., and

D. (1992) J. Biol. Chem. 267,8964-8990

J. 11,1503-1512

Admon, A,, and Crabtree, G. R. (1991) Science 264.1762-1767

U. S. A 89,8145-8149

DeLamarter, J. (1992) Eur J. Eiochem. 206, 401-411

atf-ao, a novel variant of the atf/creb transcription factor

Documents