Characterization of Human Activating Transcription Factor 4, a ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY © 1997 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 272, No. 38, Issue of September 19, pp. 24088 –24095, 1997 Printed in U.S.A.

Characterization of Human Activating Transcription Factor 4, a Transcriptional Activator That Interacts with Multiple Domains of cAMP-responsive Element-binding Protein (CREB)-binding Protein (CBP)* (Received for publication, May 23, 1997, and in revised form, July 3, 1997)

Guosheng Liang‡ and Tsonwin Hai‡§¶ From the ‡Ohio State Biochemistry Program, §Department of Medical Biochemistry and Neurobiotechnology Center, Ohio State University, Columbus, Ohio 43210

We demonstrate that human activating transcription factor 4 (hATF4), a member of the activating transcription factor/cAMP-responsive element-binding protein (ATF/CREB) family of transcription factors, is a potent transcriptional activator in both mammalian cells and yeast. The N-terminal 113 amino acids of hATF4 activate transcription efficiently, and unexpectedly, the C-terminal bZip DNA binding domain of hATF4 also activates transcription, albeit weakly. Our results indicate that hATF4 interacts with several general transcription factors: TATA-binding protein, TFIIB, and the RAP30 subunit of TFIIF. In addition, hATF4 interacts with the coactivator CREB-binding protein (CBP) at four regions: 1) the KIX domain, 2) a region that contains the third zinc finger and the E1A-interacting domain, 3) a C-terminal region that contains the p160/SRC-1-interacting domain, and 4) the recently identified histone acetyltransferase domain. Interestingly, both the N-terminal and C-terminal regions of hATF4 interact with the above general transcription factors and CBP, providing a mechanistic explanation for their ability to activate transcription. Consistent with its role as a coactivator, CBP potentiates the ability of hATF4 to activate transcription. The potential significance of the interaction between hATF4 and multiple factors is discussed.

Transcription of protein-coding genes is a major regulatory step for gene expression in eukaryotes. Intensive studies in this area have revealed that transcriptional activators exert their effects by contacting general transcription factors (GTFs)1 directly or indirectly (for reviews, see Refs. 1–3). In the case of indirect contact, coactivators mediate the effects from the activators to the general factors. A recently discovered coactivator is the CBP/p300 family of proteins (for reviews, see Refs. 4 and 5). The first member of this family, CREB-binding protein

* This work was supported by National Institutes of Health Grant GM46218 (to T. H.) 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: Room 148, Rightmire Hall, 1060 Carmack Rd., Ohio State University, Columbus, OH 43210; Tel.: 614-292-2910; Fax: 614-292-5379; E-mail: [email protected]. 1 The abbreviations used are: GTF, general transcription factor; ATF, activating transcription factor; hATF, human ATF; bZip, basic region/ leucine zipper; CAT, chloramphenicol acetyltransferase; CRE, cAMPresponsive element; CREB, CRE-binding protein; CBP, CREB-binding protein; CRE-BP1, cAMP-responsive element-binding protein 1; HAT, histone acetyltransferase; GST, glutathione S-transferase; RAP, RNAP (RNA polymerase) II-associating protein; TBP, TATA-binding protein; TF, transcription factor; HTLV, human T-cell leukemia virus.

(CBP), was isolated by its ability to interact with CREB (6), a transcription factor with the basic region/leucine zipper (bZip) DNA binding domain. Strikingly and unexpectedly, CBP/p300 proteins function as coactivators for not only CREB (7–10) but also a variety of other sequence-specific transcription factors, such as c-Jun (8, 11, 12), c-Fos (13), nuclear receptors (14 –16), c-Myb (17, 18), YY1 (19, 20), Sap-1a (21), MyoD (22, 23), Stat2 (24), SREBP (25), and NF-kB (26). These observations provide an explanation for the ability of some of these transcription factors to inhibit each other. For example, nuclear receptors have been demonstrated to inhibit the activity of c-Jun by competing with c-Jun for CBP (14). Significantly, in addition to interacting with sequence-specific transcription factors, CBP/ p300 also interacts with other proteins such as the S6 kinase pp90rsk (27), nuclear receptor coactivator p160/SRC-1 (14, 15), several GTFs (see below), and viral proteins E1A (9, 28, 29), SV40 T antigen (30, 31), and HTLV1 Tax (32). Therefore, CBP/p300 functions as a target for a variety of proteins involved in different regulatory circuits, allowing cells to coordinate or integrate signals from different pathways. Consequently, it has been suggested that it be named an “integrator” (14) or “co-integrator” (16). Although the mechanisms by which CBP/p300 functions as a coactivator remain to be determined, the current understanding suggests several possibilities. One possibility is that it acts as an adapter between the sequence-specific transcriptional activators and GTFs such as TBP (22, 33), TFIIB (7), and RNA polymerase II (34). Another possibility is that CBP/p300 enhances the acetylation of histones; this presumably destabilizes the nucleosome and facilitates access of DNA by regulatory factors (for reviews see Refs. 35 and 36). Recent studies indicate that CBP/p300 can enhance acetylation of histones by at least two mechanisms; it has an intrinsic histone acetyltransferase (HAT) activity (37, 38), and it recruits P/CAF, which also has HAT activity (39). In this report, we demonstrate that human ATF4 (hATF4), a member of the ATF/CRE family of proteins, also interacts with CBP/p300. Binding sites for this family of proteins are present in many cellular and viral promoters (for a review see Ref. 40), suggesting an involvement of this family of proteins in the regulation of many genes. A partial cDNA clone of hATF4 was originally isolated by its ability to bind to the consensus ATF/ CRE site (41). The full-length clone was later isolated and named as TAXREB67 (42) or CREB2 (43). However, the term “CREB2” was also used to refer to CRE-BP1/ATF2 (44) and an alternatively spliced form of CREB that lacks 14 amino acids (45, 46). Therefore, to avoid confusion, we will refer to the human clone as hATF4 in the rest of this report. Three mouse cDNAs, mATF4 (47), mTR67 (48), and C/ATF (49), share over

24088

This paper is available on line at http://www.jbc.org

Multiple Interactions between hATF4 and CBP

24089

FIG. 1. Sequence comparison of various ATF4 proteins. A, amino acid sequence alignment of hATF4 (Refs. 42 and 43 and this study; GenBankTM accession numbers D90209 and M86842), C/ATF (Ref. 49; accession number L13791), mTR67 (Ref. 48; accession number M94087), and mATF4 (Ref. 47; accession number X61507). The bZip region, the second zipper (zipper II), and a potential phosphorylation site for MAP kinase (consensus PX(S/T)P) are indicated. B, comparison of the bZip region and the second zipper of hATF4 and ApCREB2/ATF4 (Ref. 53; accession number U40851). The comparison in panel A and the comparison of the bZip region in panel B were aligned by the MegAlign program; the comparison of the second zipper in panel B was aligned manually.

90% identity to hATF4 (Fig. 1A). Therefore, ATF4 represents a group of conserved proteins. Interestingly, all ATF4 proteins have a second zipper in addition to the C-terminal leucine zipper (Fig. 1A). It is not clear, however, whether this second zipper is functionally important. Although all ATF4 proteins are bZip proteins, the only similarity they share with CREB (50, 51) and other ATF proteins is the conserved residues in the basic region and the conserved leucine residues in the leucine zipper (for reviews see Refs. 40 and 52). In the rest of the bZip domain and outside of the bZip domain, ATF4 proteins are completely different from CREB and other ATF proteins. Therefore, the names of various ATF proteins reflect more the history of discovery rather than the similarities between them. Thus far, the physiological functions of ATF4 proteins have not been defined. Tissue distribution of ATF4 mRNA demonstrated that it is present in most tissues or cell lines examined, revealing little clues about its physiological functions. In this

context, it is interesting to note that aplysia CREB2/ATF4 (ApCREB2/ATF4), which has been demonstrated to repress long term facilitation (53), is 50% identical to hATF4 in the bZip domain and has a second zipper (Fig. 1B). However, it is not clear whether the mammalian ATF4 plays a similar role. In this report, we show that hATF4 is a strong activator in both mammalian cells and yeast. It interacts with several GTFs and with the coactivator CBP at multiple domains including the recently identified HAT domain. EXPERIMENTAL PROCEDURES

Plasmids—N-terminal and C-terminal deletion constructs of hATF4 were generated by ExoIII deletion. GAL4 fusions were generated by placing GAL4 DNA binding domain at either the N or C terminus of various hATF4 fragments, and the resulting fusions are indicated as GAL4-hATF4 or hATF4-GAL4, respectively. To express the proteins, various DNA fragments were cloned into appropriate expression vectors: pCG (from W. Herr) for mammalian cells, pTM1 (from B. Moss) for

24090


reticulocyte translation, pGBT-9 (from S. Fields) for yeast expression, and pET-His (54) for expression in Escherichia coli. Yeast Strains and Expression Vectors—Yeast strain Y190 (from S. J. Elledge), which lacks the endogenous functional GAL4 protein is MATa, leu2–3,112, ura3–52, trp1–901, his3-D200, ade2–101, gal4Dgal80D URA3 GAL-lacZ, LYS GAL-HIS3, cyhr (55). Yeast transformation and the b-galactosidase assay were carried out according to standard protocols (56, 57). Cell Culture, Transfection, and Chloramphenicol Acetyltransferase (CAT) Assay—Cell culture, calcium phosphate transfection, and CAT assays were performed as described previously (58). For transfection, unless otherwise indicated, 1.5 mg of reporter DNA and 0.5 mg of effector DNA expressing the indicated proteins were used for each 60-mm plate of cells. Appropriate amounts of pGEM3 carrier DNA were included to make a total 7 mg of DNA for each transfection. Protein Expression and Purification—hATF4 tagged with six contiguous histidines was expressed in E. coli BL21 (DE3/LysS) (from F. W. Studier) according to Studier et al. (59) and purified at 4 °C as follows. Cell pellets were resuspended in ice-cold 1 M bZip buffer (1 M NaCl, 50 mM sodium phosphate, pH 7.8, 5% glycerol, 2 mM b-mercaptoethanol) containing various protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 0.1 M sodium metabisulfite, 2 mg/ml aprotinin, and 2 mg/ml leupeptin). The high concentration of NaCl appears to increase the solubility of hATF4. Tween 20 was added to a final concentration of 0.1%, and the cell suspension was lysed by sonication on ice. After centrifugation at 15,000 3 g at 4 °C for 15 min, the supernatant was separated from the cell debris and mixed with Ni21 nitrilotriacetic acid-agarose beads (Qiagen) that had been preequilibrated in 1 M bZip buffer. After gentle rocking for 1 h, the beads were packed into column and washed with two column volumes of 1 M bZip buffer, two column volumes of 0.3 M bZip buffer (same as 1 M bZip except NaCl was 0.3 M), and five column volumes of 0.3 M bZip buffer containing 25 mM imidazole. Bound proteins were eluted with 0.3 M bZip buffer containing 0.5 M imidazole and dialyzed against buffer D (20 mM Hepes, pH 7.9, 0.1 M NaCl, 20% glycerol, 0.2 mM EDTA, and 0.5 mM dithiothreitol) containing 0.5 mM phenylmethylsulfonyl fluoride and 0.1 M sodium metabisulfite. Protein concentrations were determined by Bio-Rad assay, and bovine serum albumin was added, if necessary, to make the final protein concentration 1 mg/ml. In Vitro Transcription—In vitro transcription was carried out using primer extension as described previously (60). The CAT primer 59GCCATTGGGATATATCAACGG-39 is complementary to the region from 129 to 149 of the CAT mRNA. Nuclear extracts were made from HeLa cells according to Dignam et al. (61). In Vitro Protein Synthesis—pTM1 derivatives encoding the indicated proteins were transcribed by T7 polymerase and translated in the presence of [35S]methionine using the TNT reticulocyte lysate system (Promega) according to the manufacturer’s instructions. Binding Assays Using GST Fusion Proteins—E. coli cells expressing the indicated GST fusion proteins were collected and resuspended in ice-cold 1 M TED buffer (1 M KCl, 50 mM Tris-HCl, pH 8, 2 mM EDTA, 1 mM dithiothreitol) containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 0.1 M sodium metabisulfite, 2 mg/ml aprotinin, and 2 mg/ml leupeptin). Triton X-100 was added to 1%, and the cell suspension was sonicated on ice. After centrifugation at 15,000 3 g for 15 min at 4 °C, the supernatant was separated from the cell debris. Aliquots were quick-frozen in liquid nitrogen and stored at 280 °C. Each aliquot (0.25–1 ml, depending on the levels of expression) of the frozen supernatant was thawed and incubated at 4 °C for 30 min with 10 ml of glutathione-conjugated agarose beads (Sigma) that had been prewashed in 1 M TED buffer, and the unbound proteins were removed by washes in GST binding buffer (100 mM KCl, 40 mM Hepes, pH 7.5, 5 mM MgCl2, 0.2 mM EDTA, 1 mM dithiothreitol, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 0.1 M sodium metabisulfite, 2 mg/ml aprotinin, and 2 mg/ml leupeptin). The resulting glutathione-agarose beads bound with the fusion protein were used in the binding assay immediately. The beads were incubated with the indicated 35S-labeled proteins at 4 °C in 0.5 ml of GST binding buffer for 1 h to allow binding and washed with GST binding buffer for three times (4 °C, 5 min each time). The bound proteins were eluted by boiling the beads with 20 ml of SDS-PAGE loading buffer and analyzed by electrophoresis. The GST fusion proteins were visualized by Coomassie Blue stain, and the labeled proteins were visualized by autoradiography. RESULTS

hATF4 Is a Potent Transcriptional Activator in both Mammalian Cells and Yeast—To examine the function of hATF4, we

FIG. 2. hATF4 is a transcriptional activator. A, hATF4 activates transcription in vivo. pCG vector or pCG-ATF4 expressing hATF4 was transfected into HeLa cells with the indicated CAT reporters. ATF sites, artificial promoter with three tandem ATF sites and the E1B TATA box; E4, adenovirus E4 promoter, 2330 to 117, with an internal deletion from 2138 to 265; E-selectin, E-selectin promoter, 2383 to 180; ENK, proenkephalin promoter 2193 to 1210; HTLV-1, HTLV-1 promoter, 2438 to 1267; SOM, somatostatin -750 to 150; TAT, tyrosine aminotransferase promoter 24289 to 162. Fold of activation by hATF4 is calculated by defining the reporter activity in the presence of pCG as 1. The result is the average of five experiments. B, hATF4 activates transcription in vitro. The CAT reporter driven by tandem ATF sites was transcribed in vitro using HeLa cell nuclear extracts in the presence of increasing amounts of hATF4 (1, 10, 20, and 40 ng) expressed and purified from E. coli. CAT mRNAs were analyzed by primer extension using a CAT-specific primer. The arrow indicates the expected product. A representative result of three experiments is shown. C, ATF4 activates transcription in yeast. Plasmids expressing the full-length GAL4 (GAL4 FL) and various GAL4-hATF4 fusions (FL, full-length hATF4; N182, hATF4 N-terminal 182-amino acid peptide; DN22, truncated hATF4 lacking the first 22 amino acids) were introduced into yeast strain Y190, which contains a b-galactosidase reporter driven by the GAL1 promoter in its chromosome (55). The result is the average of three experiments.

transfected HeLa cells with pCG-hATF4, which expresses hATF4, and a CAT reporter driven by tandem ATF sites. As shown in Fig. 2A, overexpression of hATF4 activated the reporter more than 150-fold. This activity can be transferred to a heterologous DNA binding domain, the GAL4 DNA binding domain. A plasmid expressing the hATF4-GAL4 fusion protein activated the CAT reporter driven by tandem GAL4 sites. The -fold activation ranged from approximately 60 (Fig. 4) to 700 (Fig. 3), depending on the amount of DNA used (10 and 500 ng, respectively). This activity was comparable with that of GAL4VP16 (data not shown), a well characterized, strong viral activator. Although the DNA binding domain of hATF4 is at its C terminus, the orientation of fusion does not significantly affect the activity (data not shown). Therefore, fusion proteins with either orientation were used in this study. In the rest of this report, “GAL4-hATF4” refers to fusion proteins with a GAL4 DNA binding domain at the N terminus, and “hATF4-GAL4” refers to fusion proteins with a GAL4 DNA binding domain at the C terminus. Because the CAT reporters used above were driven by artificial promoters, we asked whether ATF4 can activate transcription from naturally occurring promoters. We examined the


FIG. 3. Both the N- and the C-terminal regions of hATF4 contain a transcriptional activation domain. A, the N-terminal region is important for the function of hATF4. The CAT reporter driven by three tandem ATF sites was transfected into HeLa cells with pCG vector or pCG derivatives producing the indicated hATF4 proteins. The fold of activation by each derivative is calculated by defining the reporter activity in the presence of pCG as 1. The relative activity is calculated by defining the activity of the full-length hATF4 as 100%. The result is the average of five experiments. B, the bZip region of hATF4 has transcriptional activity when fused to the GAL4 DNA binding domain. The CAT reporter driven by five tandem GAL4 sites was transfected into HeLa cells with pCG vector or pCG derivatives, producing the indicated fusion proteins. The fold of activation and the relative activity are calculated as in panel A. The result is the average of five experiments.

following promoters containing functional ATF or ATF-like sites: adenovirus E4, E-selectin, proenkephalin (ENK), HTLV-1, somatostatin (SOM), and tyrosine aminotransferase (TAT) promoters (Fig. 2). As shown in Fig. 2A, hATF4 activated all of these promoters, ranging from a few to 50-fold. The fold of activation was not as high as that observed on the artificial promoter, possibly because the natural promoters contain binding sites for other transcription factors and have higher basal transcriptional activity, leading to less activation; alternatively, the activation may be lower because hATF4 does not interact well with other factors on the promoters and consequently does not activate the promoter efficiently. To test whether the ability of hATF4 to activate transcription can be recapitulated in vitro, we examined its activity by an in vitro transcription assay. As shown in Fig. 2B, hATF4 activated transcription from a promoter containing tandem ATF sites in a dose-dependent manner. However, when hATF4 was added at a high concentration, the activity of the promoter started to decrease (Fig. 2B, last lane), presumably because hATF4 sequestered factors/co-factors away from the promoter at high concentrations, a phenomenon commonly referred to as “squelching” (62). Because many mammalian transcription factors can function in yeast, we examined whether hATF4 also activates transcription in yeast. We took a GAL4 fusion approach to avoid potential interference from the endogenous yeast ATF-related proteins (63– 66). As shown in Fig. 2C, the full-length (FL) GAL4-hATF4 activated a b-galactosidase reporter driven by tandem GAL4 sites. The activation by GAL4-hATF4 was comparable with that by the full-length GAL4, a well characterized yeast activator, indicating that hATF4 can function as a potent activator in yeast. In summary, we demonstrate that hATF4 can activate transcription in a mammalian system both in vivo and in vitro; in addition, it functions as a potent transcriptional activator in yeast. The N Terminus of ATF4 Contains an Activation Domain—To examine whether hATF4 contains discrete transcriptional activation domains, we made a series of N- and Cterminal deletions and examined their activities by transfection. Fig. 3A shows the results of the N-terminal dele-

24091

FIG. 4. The N-terminal 113 amino acids of hATF4 function as a potent activator. The CAT reporter driven by tandem GAL4 sites was transfected into HeLa cells with 10 ng of pCG or pCG derivatives, producing the indicated hATF4-GAL4 fusion proteins. The fold of activation by hATF4 is calculated by defining the reporter activity in the presence of pCG as 1. The activity of hATF4-(1–339) is then defined as 100%. This construct lacks the last 12 amino acids of hATF4 (to remove the termination codon) and has an activity comparable with that of the full-length hATF4 with GAL4 fused at its N terminus (data not shown). The result is the average of five experiments.

tions. Deletion of the first N-terminal 22 amino acids (DN22) abolished 80% of the activity. Experiments in yeast using GAL4 fusion proteins showed a comparable result (Fig. 2C), indicating that the first 22 amino acids are important for the function of hATF4. Further deletion gradually decreased the activity, with the exception that DN97 (deletion of the first 97 amino acids) is slightly more active than DN22. It is not clear, however, whether this slight increase in activity is significant. Deletion up to amino acid 264 resulted in a low but reproducible activity, 9-fold activation. Because this construct only contains the bZip domain, this result indicates that the bZip domain of hATF4 has the ability to activate transcription, in addition to binding to DNA. This notion is supported by the observation that, when fused to the GAL4 DNA binding domain, this bZip domain activated a reporter gene driven by tandem GAL4 sites about 50-fold (Fig. 3B). When the same amount of DNA (500 ng) was transfected into cells, the GAL4 fusion containing the full-length hATF4 (GAL4-hATF4FL) activated the promoter about 700-fold. The GAL4 fusion containing the bZip domain of another protein (ATF3), however, had no effect (data not shown). This observation is in agreement with previous reports that the bZip domains of some transcription factors have activities other than binding to DNA, such as interacting with RNA polymerase II or E1A (67, 68). Because the DNA binding domain of hATF4 resides at its C terminus, deletion of this region would render the rest of the molecule unable to bind to DNA. To examine the activities of various C-terminal deletions of hATF4, we fused them with the GAL4 DNA binding domain. As shown in Fig. 4, deletion from the C terminus up to amino acid 113 did not decrease the activity; in fact, the remaining molecules appeared to have a higher activity than the full-length hATF4. However, without further analyses, it is not clear whether this result indicates a repression domain between amino acids 113 and 339. Further deletions decreased the activity, and the first 31 amino acids had only 10% of the full-length hATF4 activity. Taken together, the N- and C-terminal deletion analyses indicate that the first 22 amino acids are important for the function of hATF4, and the N-terminal 113 amino acids can activate transcription efficiently. In addition, the C-terminal bZip region of hATF4, previously defined as the DNA binding domain, can activate transcription weakly. hATF4 Interacts with Multiple GTFs—As a first step to elucidate the mechanisms by which hATF4 activates transcription, we examined its ability to interact with several GTFs. We synthesized 35S-labeled hATF4 using reticulocyte lysates and assayed its ability to interact with several GTFs expressed as GST fusion proteins and immobilized on glutathione-conju-

24092


FIG. 5. hATF4 interacts with multiple GTFs. A, Coomassie Blue stain of different GST fusion proteins. B, hATF4 interacts with TBP, TFIIB, and RAP30. Radiolabeled hATF4 synthesized in vitro by reticulocyte lysate was incubated with the indicated GST fusion proteins, washed, eluted, and analyzed on SDS-polyacrylamide gels. The input represents 20% of the protein used in the binding assay. A representative of three experiments is shown. C, schematic diagrams of various hATF4 derivatives are shown. hATF4FL, full-length; hATF4-N, amino acids 1–113; hATF4-C, amino acids 265–351; hATF4-M, amino acids 178 –293.

gated beads: GST-TBP, GST-TFIIB, and GST-RAP30. As shown in Fig. 5B, the full-length hATF4 (hATF4FL) bound to all three factors efficiently. This binding was not due to the GST portion of the fusion proteins, because hATF4FL did not bind to the glutathione beads bound with GST only. To further define the regions of hATF4 that interact with these GTFs, we assayed the radiolabeled polypeptides hATF4-N (amino acids 1–113), hATF4-M (amino acids 178 –293), and hATF4-C (amino acids 265–351) diagrammed in Fig. 5C. hATF4-N and hATF4-C were chosen because they activated transcription in the transfection assays (referred to as N113 in Fig. 4 and bZip in Fig. 3, respectively). Consistent with the results that they activated transcription, both hATF4-N and hATF4-C interacted with TBP, TFIIB, and RAP30 (Fig. 5B). In contrast, hATF4-M did not interact with any of these GTFs, consistent with the observation that the middle region of hATF4 did not activate transcription (data not shown). hATF4 Interacts with the Coactivator CBP at Multiple Domains—As described in the Introduction, some activators exert their effects on the basal transcriptional machinery by interacting with coactivators that serve as bridge molecules between sequence-specific transcription factors and GTFs. Because CBP is a well documented coactivator for bZip proteins such as CREB, c-Jun, and c-Fos, we tested whether hATF4 also interacts with CBP. As shown in Fig. 6B, in vitro translated, radiolabeled hATF4 interacted with various regions of CBP expressed as GST fusions and immobilized on glutathioneconjugated beads. Interestingly, hATF4 interacted with CBP at

FIG. 6. hATF4 interacts with multiple domains of CBP. A, Coomassie Blue stain of different GST-CBP fusion proteins. B, hATF4 interacts with CBP at multiple domains. Radiolabeled hATF4 synthesized in vitro by reticulocyte lysate was incubated with the indicated GST-CBP fusion proteins, washed, eluted, and analyzed on SDS-polyacrylamide gels. hATF4-N, hATF4-C, and hATF4-M are described in the legend to Fig. 5. The input represents 20% of the protein used in the binding assay. A representative of three experiments is shown.

four distinct regions: (a) amino acids 451– 682, the KIX domain, (b) amino acids 1459 –1891, a region that contains the third zinc finger and the E1A-interacting domain (referred to as C/H3 region in the rest of the report for the convenience of discussion), (c) amino acids 1892–2441, a C-terminal region that contains the p160/SRC-1-interacting domain, and (d) amino acids 1069 –1459, a region that contains a portion of the HAT domain (Fig. 6B and summarized in Fig. 9). Furthermore, both hATF4-N and hATF4-C interacted with these CBP domains (Fig. 6B), consistent with the finding described above that these two regions of hATF4 have transcriptional activation activity. The specificity of the interaction described here was indicated by two controls. First, CBP peptide (amino acids 706 –1009), which was expressed at a reasonable level (Fig. 6A), did not interact with hATF4. Second, radiolabeled hATF4-M did not interact with any of these GST-CBP fusion proteins. Analyses of the Interaction between hATF4 and CBP in Vivo—To determine whether the interaction between hATF4 and CBP observed in vitro also occurs in vivo, we took a mammalian two-hybrid approach. We examined two domains: the KIX domain and the C/H3 domain. We used constructs expressing these domains fused with the activation domain of VP16 and examined the ability of these fusion proteins to enhance the transcriptional activities of the following hATF4/GAL4 constructs: hATF4-GAL4 (full-length hATF4), hATF4N-GAL4 (hATF4 amino acids 1–113), and GAL4-hATF4C (hATF4 amino acids 265–351). As described above, these hATF4-GAL4 fusions can activate a CAT reporter driven by tandem GAL4 sites. We arbitrarily defined the activity of these proteins in the absence of CBP-VP16 as 1 (open bars). Co-expressing either KIX-VP16 or C/H3-VP16 further activated the reporter (hatched and filled


FIG. 7. ATF4 interacts with CBP in vivo. CAT reporter driven by tandem GAL4 sites was transfected into HeLa cells with 250 ng of plasmids, producing the indicated proteins: GAL4 DNA binding domain (GAL4 DBD), full-length hATF4 fused with the GAL4 DNA binding domain (hATF4-GAL4); N-terminal amino acids 1–113 of hATF4 fused with GAL4 DNA binding domain (hATF4N-GAL4); C-terminal amino acids 265–351 of hATF4 fused with GAL4 DNA binding domain (GAL4hATF4C). For each set of experiments, 1 mg of pCG vector (open bars) or pCG derivatives expressing different proteins were co-transfected. KIXVP16 expresses the KIX domain (amino acids 461– 662) of CBP fused with the activation domain of VP16 (hatched bars); C/H3-VP16 expresses the C/H3 region (amino acids 1621–1877) of CBP fused with the activation domain of VP16 (filled bars); VP16 expresses only the activation domain of VP16 (stippled bars). The activity of the reporter in the absence of CBP-VP16 is arbitrarily defined as 1 (open bars). The result is the average of three experiments.

bars), supporting the notion that hATF4 can interact with these CBP regions in vivo. The C/H3 region appeared to interact with the full-length hATF4 more efficiently than it does with the hATF4 N-terminal or hATF4 C-terminal region. The specificity of the interaction is indicated by three controls. First, the activation domain of VP16 by itself did not affect the activity of hATF4/GAL4 (stippled bar), indicating that the CBP portion of the VP16 fusion proteins is required for the interaction. Second, neither KIX-VP16 nor C/H3-VP16 affected the activity of the GAL4 DNA binding domain alone (GAL4 DBD; Fig. 7), indicating that the hATF4 portion of the GAL4 fusion proteins is required for the interaction. Third, these VP16 fusion proteins did not affect the activity of the GAL4 fusion protein containing the middle region of hATF4 (hATF4-M, data not shown), consistent with the in vitro finding that this region of hATF4 did not interact with CBP. Because CBP is a coactivator, the interaction between hATF4 and CBP suggests that CBP would potentiate the ability of hATF4 to activate transcription. We therefore carried out the experiment presented in Fig. 8. The difference between this experiment and the above two-hybrid experiment is that wild type CBP, instead of CBP-VP16 fusion, was used. Therefore, it assays the ability of CBP as a coactivator to affect the activity of hATF4, not simply the ability of CBP to interact with hATF4. As shown in Fig. 8, CBP enhanced the ability of hATF4-GAL4 to activate transcription through tandem GAL4 sites. Importantly, CBP had no effect on the GAL4 DNA binding domain alone, indicating that the hATF4 portion of the hATF4-GAL4 fusion is required for the potentiation by CBP. p300, another member of the CBP/p300 coactivator family, also potentiated the ability of hATF4 to activate transcription (data not shown). Furthermore, consistent with the in vitro and in vivo interaction assays, CBP augmented the activation potential of both hATF4-N and hATF4-C (Fig. 8) but not that of hATF4-M (data not shown), which contains a region of hATF4 that does not interact with CBP. The reason for using hATF4-GAL4 fusion, instead of the wild type hATF4, is to avoid the interference from endogenous ATF or ATF-like proteins on the reporter.

24093

FIG. 8. CBP potentiates the ability of hATF4 to activate transcription. CAT reporter driven by tandem GAL4 sites was transfected into HeLa cells with 250 ng of plasmids, producing the indicated proteins (see Fig. 7 legend for nomenclature), in the absence (2) or presence (1) of DNA expressing CBP (1 mg pCG versus pCG-CBP). The activity of the reporter in the absence of CBP is defined as 1. The result is the average of three experiments.

DISCUSSION

hATF4 Functions as a Transcriptional Activator—In this report, we demonstrate that hATF4 is a potent transcriptional activator. However, hATF4 was reported to be a transcriptional repressor previously (referred to as CREB2 in Ref. 43). One possible explanation for this apparent discrepancy is that overexpression of hATF4 resulted in sequestration of factors/cofactors away from the promoter and consequently repression (“squelching”) of the reporter. Consistent with this supposition, titration experiments showed that hATF4 “squelches” its own activity at high concentrations. In an in vitro transcription assay, 40 ng of hATF4 gave rise to a lower activity than 1 ng of hATF4 (Fig. 2B). Furthermore, 100 ng of hATF4 completely abolished the activity (data not shown). In transient transfection assays, 10 –500 ng of DNA expressing hATF4 were used as described above; higher amounts of DNA resulted in lower activities (data not shown). These results are consistent with the observations that hATF4 is a strong activator and interacts with GTFs and the coactivator CBP efficiently. Because of its high affinity with these factors, hATF4, at high concentrations, can presumably sequester them away from the promoters. Therefore, we conclude that hATF4 is a transcriptional activator; however, under certain conditions it may display repression activity due to squelching. The deletion analyses described in this report indicate that the N-terminal 113 amino acids of hATF4 can activate transcription efficiently and that the Cterminal bZip domain can activate transcription weakly. hATF4 Interacts with Multiple GTFs—Our results indicate that hATF4 interacts with multiple GTFs: TBP, TFIIB, and RAP30. Although many eukaryotic transcription factors have been demonstrated to interact with TBP and TFIIB (for reviews see Refs. 69 and 70), fewer have been demonstrated to interact with RAP30 (71). At present, it is not clear whether hATF4 also interacts with other GTFs, because we did not examine them in this study. The observation that hATF4 interacts with multiple GTFs is consistent with the finding that hATF4 functions in both mammalian cells and yeast, because these GTFs are conserved in both systems. We note, however, it is not clear whether any of the interactions we described here are required for the ability of hATF4 to activate transcription. As shown previously, some interactions between transcriptional activators and GTFs are not required for transcriptional

24094


FIG. 9. Summary of the interaction between hATF4 and CBP. A schematic diagram indicates different regions of CBP and their interactions with hATF4. *, the plus sign indicates detectable interactions (see Fig. 6B). However, it does not imply that these regions of CBP interact with hATF4 with similar affinity (see “Discussion”). The minus sign indicates no detectable binding. Some CBP-interacting proteins are included for comparison. RXR, retinoid X receptor; RID, receptor interacting domain; GR, glucocorticoid receptor; P/CAF, P300/CBP-associated factor; RSK, ribosomal protein S6 kinase; SRC-1, steroid receptor coactivator-1; SREBP, sterol regulatory element binding protein.

activation: mutations in TBP that disrupt its interaction with VP16 and p53 do not affect the transcriptional activation by these factors in vivo (72). Therefore, it is possible that some ATF4-GTF interactions may not be required for activation. Further experiments are required to demonstrate the functional relevance of these interactions. hATF4 Interacts with CBP at Multiple Regions—As described in the Introduction, many sequence-specific transcription factors interact with CBP. Most CBP-interacting proteins interact with CBP at one domain, and only TBP, MyoD, and retinoid X receptor have been reported to interact with CBP at two domains (Fig. 9). hATF4 differs from these CBP-interacting proteins in that it interacts with CBP at four regions: the KIX domain, the C/H3 region, the C-terminal region, and the HAT domain. Although hATF4 appears to interact with these regions with different efficiency (Fig. 6B), it is difficult to conclude the relative affinity from our studies, because the levels of various GST-CBP fusion proteins were not exactly the same (Fig. 6A). Furthermore, the HAT construct we used in this study only produces a portion of the HAT domain. Therefore, it remains possible that the interaction between hATF4 and the HAT domain is stronger than that observed in this study. Although the functional significance of this interaction is not clear at present, it is possible that the HAT domain may acetylate hATF4; alternatively, hATF4 may modify the activity of the HAT domain, affecting its ability to remodel nucleosome. In this study, we also demonstrate that both the N- and C-terminal regions of hATF4 interact with CBP. Therefore, hATF4 and CBP interact with each other by multisurface contacts, and the resulting complex may be very stable. Because CBP is a large protein, it may serve as a surface for many proteins to bind. Consistent with this notion, our preliminary results indicate that many radiolabeled proteins from crude cellular extracts bound to the GST-hATF4 affinity column.2 Interestingly, some of these proteins can phosphorylate hATF4 in vitro,2 opening the possibility that phosphorylation may regulate the function of hATF4. In this context, it is interesting to note that phosphorylation has been demonstrated to regulate the activity of several members of the ATF/CREB family of proteins, such as CREB, cAMP-responsive element modulator 2

G. Liang and T. Hai, unpublished results.

(CREM), and ATF2 (see Refs. 40, 73, and 74 for reviews and references therein; also see Ref. 75). Intriguingly, hATF4 contains a potential phosphorylation site (Pro-Leu-Ser-Pro) for MAP kinase (Fig. 1) and several Ser-Pro sequences. The SerPro sequence has been reported to be a potential site for proline-directed kinases (76, 77). We are currently investigating these hATF4-interacting proteins and the potential kinases for hATF4. Acknowledgments—We thank Drs. J. Brady, M. Comb, O.M. Andrisani, and G. Schu¨tz for CAT reporters driven by natural promoters; Drs. R. H. Goodman, T. Kouzarides, D. M. Livingston, B. Lu¨scher, M. R. Montminy, and M. G. Rosenfeld for CBP and p300 constructs; Dr. D. Reinberg for GTF constructs; Drs. S. Fields, W. Herr, and B. Moss for expression vectors; Dr. S. J. Elledge for the Y190 yeast strain; and Dr. F. W. Studier for BL21 bacterial strains. REFERENCES 1. 2. 3. 4. 5. 6. 7.

8. 9. 10. 11. 12. 13. 14.

15. 16. 17. 18. 19. 20. 21. 22.

Goodrich, J. A., Cutler, G., and Tjian, R. (1996) Cell 84, 825– 830 Verrijzer, C. P., and Tjian, R. (1996) Trends Biochem. Sci. 21, 338 –342 Bjo¨rklund, S., and Kim, Y.-J. (1996) Trends Biochem. Sci. 21, 335–337 Janknecht, R., and Hunter, T. (1996) Nature 383, 22–23 Janknecht, R., and Hunter, T. (1996) Curr. Biol. 6, 951–954 Chrivia, J. C., Kwok, R. P. S., Lamb, N., Hagiwara, M., Montminy, M. R., and Goodman, R. H. (1993) Nature 365, 855– 859 Kwok, R. P. S., Lundblad, J. R., Chrivia, J. C., Richards, J. P., Ba¨chinger, H. P., Brennan, R. G., Roberts, S. G. E., Green, M. R., and Goodman, R. H. (1994) Nature 370, 223–226 Arias, J., Alberts, A. S., Brindle, P., Claret, F. X., Smeal, T., Karin, M., Feramisco, J., and Montminy, M. (1994) Nature 370, 226 –229 Eckner, R., Ewen, M. E., Newsome, D., Gerdes, M., DeCaprio, J. A., Lawrence, J. B., and Livingston, D. M. (1994) Genes Dev. 8, 869 – 884 Arany, Z., Sellers, W. R., Livingston, D. M., and Eckner, R. (1994) Cell 77, 799 – 800 Bannister, A. J., Oehler, T., Wihelm, D., Angel, P., and Kouzarides, T. (1995) Oncogene 11, 2509 –2514 Lee, J.-S., See, R. H., Deng, T., and Shi, Y. (1996) Mol. Cell. Biol. 16, 4312– 4326 Bannister, A. J., and Kouzarides, T. (1995) EMBO J. 14, 4758 – 4762 Kamei, Y., Xu, L., Heinzel, T., Torchia, J., Kurokawa, R., Gloss, B., Lin, S.-C., Heyman, R. A., Rose, D. W., Glass, C. K., and Rosenfeld, M. G. (1996) Cell 85, 403– 414 Smith, C. L., On˜ate, S. A., Tsai, M.-J., and O’Malley, B. W. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 8884 – 8888 Chakravarti, D., LaMorte, V. J., Nelson, M. C., Nakajima, T., Schulman, I. G., Juguilon, H., Montminy, M., and Evans, R. M. (1996) Nature 383, 99 –103 Dai, P., Akimaru, H., Tanaka, Y., Hou, D. X., Yasukawa, T., Kanei-Ishii, C., Takahashi, T., and Ishii, S. (1996) Genes Dev. 10, 528 –540 Oelgeschla¨ger, M., Janknecht, R., Krieg, J., Schreek, S., and Lu¨scher, B. (1996) EMBO J. 15, 2771–2780 Lee, J.-S., Galvin, K. M., See, R. H., Eckner, R., Livingston, D., Moran, E., and Shi, Y. (1995) Genes Dev. 9, 1188 –1198 Lee, J.-S., Zhang, X., and Shi, Y. (1996) J. Biol. Chem. 271, 17666 –17674 Janknecht, R., and Nordheim, A. (1996) Oncogene 12, 1961–1969 Yuan, W., Condorelli, G., Caruso, M., Felsani, A., and Giordano, A. (1996)

Multiple Interactions between hATF4 and CBP J. Biol. Chem. 271, 9009 –9013 23. Sartorelli, V., Huang, J., Hamamori, Y., and Kedes, L. (1997) Mol. Cell. Biol. 17, 1010 –1026 24. Bhattacharya, S., Eckner, R., Grossman, S., Oldread, E., Arany, Z., D’Andrea, A., and Livingston, D. M. (1996) Nature 383, 344 –347 25. Oliner, J. D., Andresen, J. M., Hansen, S. K., Zhou, S., and Tjian, R. (1996) Genes Dev. 10, 2903–2911 26. Perkins, N. D., Felzien, L. K., Betts, J. C., Leung, K., Beach, D. H., and Nabel, G. J. (1997) Science 275, 523–527 27. Nakajima, T., Fukamizu, A., Takahashi, J., Gage, F. H., Fisher, T., Blenis, J., and Montminy, M. R. (1996) Cell 86, 465– 474 28. Arany, Z., Newsome, D., Oldread, E., Livingston, D. M., and Eckner, R. (1995) Nature 374, 81– 84 29. Lundblad, J. R., Kwok, R. P. S., Laurance, M. E., Harter, M. L., and Goodman, R. H. (1995) Nature 374, 85– 88 30. Avantaggiati, M. L., Carbone, M., Graessmann, A., Nakatani, Y., Howard, B., and Levine, A. S. (1996) EMBO J. 15, 2236 –2248 31. Eckner, R., Ludlow, J. W., Lill, N. L., Oldread, E., Arany, Z., Modjtahedi, N., DeCaprio, J. A., Livingston, D. M., and Morgan, J. A. (1996) Mol. Cell. Biol. 16, 3454 –3464 32. Kwok, R. P. S., Laurance, M. E., Lundblad, J. R., Goldman, P. S., Shih, H.-M., Connor, L. M., Marriott, S. J., and Goodman, R. H. (1996) Nature 380, 642– 646 33. Abraham, S. E., Lobo, S., Yaciuk, P., Wang, H. G., and Moran, E. (1993) Oncogene 8, 1639 –1647 34. Kee, B., Arias, J., and Montminy, M. R. (1996) J. Biol. Chem. 271, 2373–2375 35. Brownell, J. E., and Allis, C. D. (1996) Curr. Opin. Genet. Dev. 6, 176 –184 36. Wolffe, A. P., and Pruss, D. (1996) Cell 84, 817– 819 37. Ogryzko, V. V., Schiltz, R. L., Russanova, V., Howard, B. H., and Nakatani, Y. (1996) Cell 87, 953–959 38. Bannister, A. J., and Kouzarides, T. (1996) Nature 384, 641– 643 39. Yang, X.-J., Ogryzko, V. V., Nishikawa, J.-I., Howard, B. H., and Nakatani, Y. (1996) Nature 382, 319 –324 40. Sassone-Corsi, P. (1995) Annu. Rev. Cell Dev. Biol. 11, 355–377 41. Hai, T., Liu, F., Coukos, W. J., and Green, M. R. (1989) Genes Dev. 3, 2083–2090 42. Tsujimoto, A., Nyunoya, H., Morita, T., Sato, T., and Shimotohno, K. (1991) J. Virol. 65, 1420 –1426 43. Karpinski, B. A., Morle, G. D., Huggenvik, J., Uhler, M. D., and Leiden, J. M. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 4820 – 4824 44. Benbrook, D. M., and Jones, N. C. (1990) Oncogene 5, 295–302 45. Yoshimura, T., Fujisawa, J. I., and Yoshida, M. (1990) EMBO J. 9, 2537–2542 46. Willems, L., Kettmann, R., Chen, G., Portetelle, D., Burny, A., and Derse, D. (1991) DNA Seq. 1, 415– 417 47. Mielnicki, L. M., and Pruitt, S. C. (1991) Nucleic Acids Res. 19, 6332 48. Chevray, P. M., and Nathans, D. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 5789 –5793 49. Vallejo, M., Ron, D., Miller, C. P., and Habener, J. F. (1993) Proc. Natl. Acad.

24095

Sci. U. S. A. 90, 4679 – 4683 50. Hoeffler, J. P., Meyer, T. E., Yun, Y., Jameson, J. L., and Habener, J. F. (1988) Science 242, 1430 –1433 51. Gonzalez, G. A., Yamamoto, K. K., Fischer, W. H., Karr, D., Menzel, P., Biggs, W., III, Vale, W. W., and Montminy, M. R. (1989) Nature 337, 749 –752 52. Hurst, H. C. (1994) Protein Profile 1, 123–168 53. Bartsch, D., Ghirardi, M., Skehel, P. A., Karl, K. A., Herder, S. P., Chen, M., Bailey, C. H., and Kandel, E. R. (1995) Cell 83, 979 –992 54. Chen, B. P. C., and Hai, T. (1994) Gene (Amst.) 139, 73–75 55. Harper, J. W., Adami, G. R., Wei, N., Keyomarsi, K., and Elledge, S. J. (1993) Cell 75, 805– 816 56. Rose, M. D., Winston, F., and Hieter, P. (1990) Methods in Yeast Genetics: A Laboratory Course Manual, Cold Spring Harbor Laboratory Press, Plainview, NY 57. Guthrie, C., and Fink, G. R. (1991) Methods in Enzymology: Guide to Yeast Genetics and Molecular Biology, Vol. 194, Academic Press, New York 58. Chen, B. P. C., Liang, G., Whelan, J., and Hai, T. (1994) J. Biol. Chem. 269, 15819 –15826 59. Studier, F. W., Rosenberg, A. H., Dunn, J. J., and Dubendorff, J. W. (1990) Methods Enzymol. 185, 60 – 89 60. Hai, T., Horikoshi, M., Roeder, R. G., and Green, M. R. (1988) Cell 54, 1043–1051 61. Dignam, J. D., Lebovitz, R. M., and Roeder, R. G. (1983) Nucleic Acids Res. 11, 1475–1489 62. Gill, G., and Ptashne, M. (1988) Nature 334, 721–724 63. Nojima, H., Leem, S.-H., Araki, H., Sakai, A., Nakashima, N., Kanaoka, Y., and Ono, Y. (1994) Nucl. Acids Res. 22, 5279 –5288 64. Vincent, A. C., and Struhl, K. (1992) Mol. Cell. Biol. 12, 5394 –5405 65. Nehlin, J. O., Carlberg, M., and Ronne, H. (1992) Nucleic Acids Res. 20, 5271–5278 66. Jones, R. H., and Jones, N. C. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 2176 –2180 67. Brandl, C. J., and Struhl, K. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 2652–2656 68. Liu, F., and Green, M. R. (1994) Nature 368, 520 –525 69. Zawel, L., and Reinberg, D. (1995) Annu. Rev. Biochem. 64, 533–561 70. Ranish, J. A., and Hahn, S. (1996) Curr. Opin. Genet. Dev. 6, 151–158 71. Martin, M. L., Lieberman, P. M., and Curran, T. (1996) Mol. Cell. Biol. 16, 2110 –2118 72. Tansey, W. P., and Herr, W. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 10550 –10554 73. Brindle, P. K., and Montminy, M. R. (1992) Curr. Opin. Genet. Dev. 2, 199 –204 74. Kyriakis, J. M., and Avruch, J. (1996) BioEssays 18, 567–577 75. Xing, J., Ginty, D. D., and Greenberg, M. E. (1996) Science 273, 959 –963 76. Joseph, C. K., Byun, H.-S., Bittman, R., and Kolesnick, R. N. (1993) J. Biol. Chem. 268, 20002–20006 77. Vulliet, R., Halloran, S. M., Braun, R. K., Smith, A. J., and Lee, G. (1992) J. Biol. Chem. 267, 22570 –22574