ERM, a PEA3 subfamily of Ets transcription factors, can cooperate with c-Jun.

A human cDNA clone for ERM, a member of the ets gene family, has been obtained by polymerase chain reaction with degenerate primers corresponding to highly conserved regions within an Ets DNA binding domain. ERM mRNA is expressed ubiquitously. The gene was mapped to chromosome 3q27. In in vivo transient-expression assays, ERM induced transcription more efficiently from a synthetic element containing both an ets-binding site (EBS) and a cyclic AMP response element (CRE) than from one containing an EBS alone. The activation of a synthetic EBS-CRE site by ERM was likely to involve a leucine zipper protein capable of dimerizing with CRE-BP1 leucine zipper. Indeed, ERM and c-Jun synergistically activated the EBS-CRE without making an apparent ternary complex. The synergy between c-Jun and ERM may be attributed to the enhancing effect of c-Jun on the transcription activity of ERM, because c-Jun increased ERM transcription activity by more than 20-fold in an assay system using a variety of fusion proteins between a Gal4 DNA-binding domain and a portion of ERM. This enhancing effect of c-Jun required the amino-terminal portion of ERM.

The ets oncogene family members are involved in gene activation through interactions with other gene products (1)(2)(3)(4). The first member of the ets gene family, v-ets oncogene, was identified as a part of the gag-myb-ets fusion oncogene of the avian retrovirus E26, which is responsible for leukemic transformation of chicken erythroblasts and myeloblasts (5,6). Since then, a number of genes related to the v-ets oncogene have been identified in a variety of cell types from Drosophila to human (3,4). The ets gene family is divided into subgroups on the basis of amino acid sequence similarity and overall structure, such as the Ets1/Ets2 (7,8), Elk-1/SAP1 (9, 10), Fli-1/Erg1 (11,12), PEA3/ER81 (13,14), PU.1 (15), GABP␣ (16), and Elf-1/E74 subfamilies (17,18). All of these genes have a highly conserved domain, the Ets domain, which is composed of about 85 amino acids and located near the carboxyl terminus, except in the Elk-1/SAP1 subfamily and Elf-1. The Ets domain is sufficient for sequence-specific binding to DNA at a purine-rich core sequence, GGA(A/T) (19,20).
The interleukin-6 response element in the junB promoter, named JRE-IL6, contains an ets-binding site (JEBS) and a CRE-like site (37). There are constitutive (37) and interleukin-6-inducible JEBS-binding proteins (38). Recently we showed that the major component of the interleukin-6-inducible proteins was Stat3/acute-phase response factor (APRF) in both a human hepatoma cell line, HepG2, and rat liver (38). During the effort to clone a constitutive JEBS-binding protein, we recloned a member of the ets gene family, ERM (39), belonging to the PEA3 subfamily. In this report, we show that ERM and c-Jun synergistically activate transcription from a synthetic element containing an EBS and a CRE without apparent cooperative binding to the corresponding site. Moreover, we show that c-Jun enhances the transcriptional activity of ERM. This functional property is specific to ERM and is not observed with a closely related Ets protein, ER81. Thus ERM is the first ets gene family member that acts synergistically with c-Jun and whose transcriptional activity is enhanced by c-Jun.

MATERIALS AND METHODS
Gene Amplification-To isolate members of the ets gene family expressed in HepG2 cells, we obtained partial cDNA fragments by reverse transcription polymerase chain reaction (PCR) with four mixed degenerate oligonucleotides based on the consensus amino acid sequence corresponding to the two highly conserved regions in the Ets domain, QLW(Q/E)FL(L/V) and MNY(D/E)(K/T)L. The sequences of the upstream and downstream primers are as follows: 5Ј primer ED1, 5Ј-CGGATCCTGG(C/G)AGTT(T/C)CT(T/C/G)CT(G/C)(G/A/C)A-3Ј; 3Ј primer ED2, 5Ј-GGAATTCA(G/T)C(T/G)T(C/G/A)TC(G/A)TA(G/A)TTCAT-3Ј; 5Ј primer ED3, 5Ј-CGGATCCA(A/G)CT(A/G)TGGCA(G/A)T(T/C)(T/ C)T-3Ј; and 3Ј primer ED4, 5Ј-GGAATTCAG(C/T)TT(T/C/G)TC(A/ G)TA(G/A)TTCAT-3Ј. Poly(A) ϩ RNA prepared from HepG2 cells was used to obtain the first strand of cDNA with Superscript reverse transcriptase (Life Technologies, Inc.) and oligo(dT) (Pharmacia Biotech Inc.). DNA fragments encompassing this region with 170 base pairs were generated by PCR with appropriate combinations of primers. Each of these PCR-generated fragments was isolated and subcloned. The analysis of the individual clones revealed that (i) 60% of clones obtained by using combinations of ED3 and ED2 or ED4 corresponded to GABP␣, and 25% of them corresponded to Ets-2; (ii) 10 clones were identical with murine ER81, and 15 clones closely related to ER81 appeared novel and then turned out to be identical with ERM during the course of this study; and (iii) only Ets-2 clones were obtained by using combinations of ED1 and either ED2 or ED4.
Isolation of an ERM cDNA Clone-To obtain the whole coding sequence of the novel (ERM) cDNA, we screened about 5 ϫ 10 5 clones from a human hepatoma gt11 cDNA library (Clontech) with the standard plaque hybridization technique (40) using the [␣-32 P]dCTP-labeled ERM PCR fragment as a probe. Three partial cDNA clones, A6, A9, and A11, were obtained. We rescreened about 5 ϫ 10 5 clones from a human placenta gt11 cDNA library (Clontech) with the 1-kilobase NcoI-NdeI fragment of clone A6 as a probe. Of the resulting clones, B9 had an overlapping region with A6. The whole coding sequence of the ERM cDNA was obtained by combining parts of the A6 and B9 clones. The deduced amino acid sequence of ERM showed that ERM was a member of the PEA3 subgroup, which consists of murine PEA3 (13), its human homolog E1A-F (41), and murine ER81 (14). The Ets domain of ERM showed about 95% similarity to those of E1A-F, PEA3, and ER81. The overall amino acid identity of ERM to ER81, PEA3, and EIA-F was 58, 52, and 52%, respectively. There was one amino acid difference in the deduced amino acid sequences of the ERM cDNA cloned by us (Phe 108 (TTT)) 2 and that cloned by Mónte et al. (39) (Leu 108 (CTT)). This difference was most likely due to polymorphism. ERM mRNA with 4.2 kilobases is expressed ubiquitously in adult human tissues and in a variety of cell lines with different lineages (39). 2 Transfection and Luciferase Assay-NIH 3T3 cells maintained in ␣-minimal essential medium (Life Technologies, Inc.) with 10% fetal calf serum (Life Technologies, Inc.) were plated at 1 ϫ 10 5 cells per 60-mm dish 1 day before transfection, and transfected by the calcium phosphate coprecipitation technique (40) with 5-7 g of plasmid DNA containing 1.2 g of a reporter plasmid, 1 g of pEF-LacZ (37) as a transfection efficiency control, the indicated amounts of various expression vectors, and appropriate amounts of control vectors (pTZ19R or control expression vectors lacking an insert DNA). The cells were incubated with DNA precipitates for 16 h, washed twice with phosphatebuffered saline, and refed with ␣-minimal essential medium containing 0.5% fetal calf serum for 48 h. Luciferase and ␤-galactosidase activities in cell lysates were determined by standard methods (40).
Minimal Promoter Luciferase Gene Constructs and Synthetic Oligonucleotides-A promoterless luciferase construct with a backbone of pBluescript II SK ϩ , pBS luc, was made by subcloning a segment containing the firefly luciferase gene taken from PGV-B (TOYO Ink) into pBluescriptII SK ϩ . A minimal promoter luciferase construct containing the JunB promoter, pBSB luc, was made by inserting a PvuII-PvuII fragment (from Ϫ42 to ϩ136 of the junB gene) upstream of the luciferase gene.
Synthetic response elements are as follows: 3ϫ EBS, 5Ј-TCGAGCAG-GAAGTCAGACTTCCTGCGCAGGAAGT-3Ј; 3ϫ mEBS, 5Ј-TCGAG-CAGCTAGTCAGACTAGCTGCGCAGCTAGT-3Ј; EBS-AP1, 5Ј-CGCG-GAAGTTATAAAGCATGACTCAG-3Ј; EBS-CRE, 5Ј-CGCGGAAGTT-ATAAAGCATGACGTCAG-3Ј. The structure of the EBS-AP1 was equivalent to the PEA3-AP1 motif in the collagenase promoter with a ninebase pair spacing between the two binding sites (35). The typical AP1 site was replaced by a somatostatin CRE (42) for the EBS-CRE. Complementary oligonucleotides were made so that there was an appropriate restriction site at each end. One copy each of 3ϫ EBS and 3ϫ mEBS oligonucleotides was inserted into the SalI site of pBSB luc to make 3ϫ EBS-pBSB luc and 3ϫ mEBS-pBSB luc, respectively. To make 3ϫ EBS-AP1 luc and 3ϫ EBS-CRE luc constructs, the oligonucleotides containing the EBS-AP1 or EBS-CRE with an ApaI and a SalI restriction site at each end were concatamerized and subcloned upstream of the minimal junB promoter at the ApaI and SalI sites of pBSB luc. To make a Gal4 luciferase reporter construct, p5GBS luc, a KpnI-HindIII fragment of G5BCAT (a gift from Dr. M. R. Green) containing five copies of Gal4 binding sites and a TATA box of the adenovirus E1b gene promoter was subcloned into the KpnI-HindIII site of pBS luc.
Expression Vectors-The ERM expression vector, pEF-ERM, was constructed by subcloning a HindIII fragment (Klenow-filled and ligated with BstXI adapters) of pBS-ERM containing a full-length ERM cDNA into the BstXI site of pEF-BOS. A full-length ER81 cDNA was obtained by reverse transcription PCR with poly(A) ϩ RNA prepared from NIH 3T3 cells and the following ER81-specific primers: ER81-A, 5Ј-GAGAATTCAGAGGAGCAGAATGGATGGAT-3Ј and ER81-B, 5Ј-ACTCTAGATGCCTTGCTTGACGGGTACT-3Ј. The sequence of the PCR-amplified ER81 cDNA was verified by dideoxy DNA sequence analysis. The c-Jun expression vector, RSV-c-Jun, the JunD expression vector, RSV-junD, and the c-Fos expression vector, pSV2-c-fos (43,44) were gifts from Dr. S. Hirai. RSV-CREB and RSV-KCREB (45) were gifts from Dr. R. D. Cone. pAct CRE-BP1 (46), a gift from Dr. S. Ishii, contained a CRE-BP1 cDNA under the control of the ␤-actin promoter. NT253 (47), an expression vector encoding an amino-terminal deletion mutant of CRE-BP1, was also a gift from Dr. S. Ishii. EF-CRE-BP1 LZ was made by adding a nuclear localization signal of SV40 large T antigen to the amino terminus of a portion of CRE-BP1 (amino acid residues 378 -507) containing a leucine zipper and a downstream region to the stop codon.
The expression vectors encoding Gal4 (1-147) fusion proteins were as follows. GAL4-ERM (pGAL510 (1-510)) containing a complete coding sequence of ERM was constructed by fusing a HindIII (Klenow-filled) fragment of pBS-ERM to the BamHI site (Klenow-filled) of pSG424 (a gift from Dr. M. Ptashne). pGAL326 was generated by deleting the NdeI-SalI fragment from pGAL510. pGAL276 was constructed by subcloning the SmaI fragment of pGAL510 into the SmaI site of pSG424. The HindIII (Klenow-filled)-PvuII fragment of pBS-ERM and the NdeI-SalI fragment (Klenow-filled) of pGAL510 were used to make pGAL165 and pGAL327-510, respectively. pGAL166 -326 was constructed by subcloning the PvuII-NdeI fragment of ERM, to which had been added an EcoRI linker, to the EcoRI site of pSG424. pGAL-ER81 containing a complete coding sequence of ER81 was constructed by fusing an EcoRI-XbaI fragment of the ER81 cDNA to the EcoRI-XbaI site of pSG424.
Fluorescence in Situ Hybridization of the Human ERM Gene-Fluorescence in situ hybridization was carried out as described previously (48). In brief, metaphase chromosome slides were prepared by the thymidine synchronization, bromodeoxyuridine release technique for delineation of replication G-and/or R-bands. The slides were denatured in 70% formamide, 2 ϫ SSC for 2 min at 75°C, immersed in 70% ethanol at Ϫ20°C, and dehydrated through an ethanol series. A fulllength cDNA designated as pBS-ERM was used as a probe. The probe (1 g) was labeled with biotin-16-dUTP (Boehringer Manheim) by nick translation reaction. The hybridization was performed in a solution containing 50% formamide, 10% dextran sulfate, 2 ϫ SSC, and DNAs dissolved to concentrations as follows: 50 g/ml of biotinylated probe DNA, 0.5 g/ml of sonicated salmon sperm DNA, and 0.5 g/ml of Escherichia coli tRNA. The hybridization signals were detected with fluorescein isothiocyanate-avidin (Boehringer Manheim) and biotinylated anti-avidin (Vector), as described previously (49).

ERM-activated Transcription through EBSs-
To understand the role of the ubiquitously expressed ERM, we characterized the functional properties of ERM. We first examined whether ERM stimulates EBS-driven transcription in NIH 3T3 by using transient transfection assays (Fig. 1). We used an EBS (CAGGAAGT) that is a target site for ER81 (14), since the Ets domain of ERM showed the greatest similarity to that of ER81. Either control reporter pBSB luc or a minimal promoter luciferase construct containing three copies of EBS (3ϫ EBS-pBSB luc) or mutated EBS (CAGCTAGT, 3ϫ mEBS-pBSB luc) was transfected into NIH 3T3 cells together with an ERM expres-sion plasmid (pEF-ERM) or a control expression vector (pEF-BOS). Expression of ERM increased 3ϫ EBS-driven luciferase gene expression by 3.8-fold over the basal level of the reporter without exogenous expression of ERM (Fig. 1A). The increase in luciferase activity by exogenous ERM was not observed when the reporter control (pBSB luc) lacking the ets-binding site or 3ϫ mEBS-pBSB luc was used (Fig. 1A). These results suggested that ERM was, indeed, an EBS-specific transcriptional activator.
ERM Showed Synergy with c-Jun in Activating Transcription through an EBS and a CRE Site-Some of the Ets family members including Ets-1, Ets-2, and Elf-1 have been shown to activate response elements that are composed of an EBS and an adjacent AP1 site (1,35,50). We tested whether ERM efficiently activates a response element containing an EBS and an AP1 site or a related site, CRE (Fig. 1B). We used two reporter plasmids containing either three copies of an EBS-AP1 oligonucleotide or three copies of an EBS-CRE oligonucleotide. The EBS-AP1 had the same orientation and the same ninebase pair spacing as the collagenase PEA3-AP1 (22,35). For an EBS-CRE, the AP1 site was replaced by the typical somatostatin CRE (42). ERM induced transcription more efficiently from the EBS-CRE site or the EBS-AP1 site than the EBS site alone (Fig. 1, A and B). Exogenous expression of ERM increased the EBS-CRE-driven luciferase expression by 25-30-fold over the basal luciferase activity of the reporter without ERM expression, whereas ERM increased the EBS-AP1-driven luciferase expression approximately 5-fold (Fig. 1B). These results suggested that ERM may cooperate with a factor(s) acting on the CRE site and possibly on the AP1 site.
We used the EBS-CRE as a target element for ERM to further study the factor(s) cooperating with ERM. We first tested the effects of several dominant negative forms of CREbinding proteins on the ERM-induced transcriptional activity through the EBS-CRE site (Fig. 1C). Overexpression of either CRE-BP1 LZ without the basal region of CRE-BP1 necessary for DNA binding, or NT253, an amino-terminal deletion mutant of CRE-BP1, efficiently reduced the ERM-activated 3ϫ EBS-CRE-driven gene expression in a dose-dependent manner. CRE-BP1 LZ acted as a dominant negative form by dimerizing with endogenous CRE-BP1 or proteins capable of dimerizing with CRE-BP1 LZ and alleviating their DNA-binding activities to the CRE site. Overexpressed NT253, devoid of transcriptional activity, likely replaces the endogenous CRE-binding proteins. On the other hand, overexpression of KCREB, a dominant negative form of CREB, increased the ERM-activated 3ϫ EBS-CRE-driven gene expression in a dose-dependent manner. These results suggested that cooperation between ERM and a factor(s) acting on the adjacent CRE motif are required for ERM to function efficiently, and the endogenous proteins are most likely ones capable of dimerizing with the CRE-BP1 leucine zipper, possibly CRE-BP1 itself or other proteins including c-Jun.
To determine which transcription factor(s) can cooperate with ERM on the EBS-CRE, NIH 3T3 cells were transfected with the ERM expression vector and various amounts of an expression vector for either CREB, CRE-BP1, c-Jun, JunB, JunD, or c-Fos together with the 3ϫ EBS-CRE luciferase reporter construct. Among the factors tested, only c-Jun showed synergy with ERM ( Fig. 2A). Since this synergy between ERM and c-Jun was not observed when 3ϫ EBS-pBSB luc reporter was used (data not shown), it is unlikely that over-expression of c-Jun enhances the expression level of ERM. The synergy was evident when smaller amounts of c-Jun expression vector were used, suggesting that overexpression of c-Jun may sequestrate another factor(s) necessary for ERM to be transcriptionally active. Exogenous expression of CRE-BP1 showed little effect on ERM-induced transcriptional activity (Fig. 2C). On the other hand, CREB at any concentration tested inhibited ERMinduced transcription (Fig. 2B), consistent with the positive effect of KCREB on ERM-induced transcriptional activity (Fig.  1C). Neither c-Fos, JunB, nor JunD showed cooperative activity with ERM (data not shown). These results suggested that the endogenous protein cooperating with ERM on EBS-CRE might be c-Jun itself. It is also possible that a heterodimer of CRE-BP1 and c-Jun, known to recognize a typical CRE (51), is involved. c-Jun Enhanced the Transcriptional Activity of ERM through the Amino-terminal Region of ERM-We next examined whether ERM and c-Jun with or without CRE-BP1 cooperatively bind to an EBS-CRE in an electrophoretic mobility shift assay using recombinant proteins translated from in vitro transcribed RNA in reticulocyte lysates. Although ERM and c-Jun-CRE-BP1 dimer could bind well to an EBS or a CRE, we did not detect the apparent cooperative binding of the mixture of these recombinant proteins to an EBS-CRE (data not shown).
Then we examined whether exogenous expression of some members of the CREB-activating transcription factor family or the Jun-Fos family enhance the transcription activity of ERM. For that purpose, we used a fusion protein (GAL4-ERM) of the Gal4 (1-147AA) DNA-binding domain and a full-length ERM (Fig. 3). Neither the control vectors, the Gal4 expression vector lacking the ERM cDNA, nor the expression vectors without an insert, showed any effect on the reporter gene expression. The GAL4-ERM fusion protein had little activity and its transcriptional activity dramatically increased by around 30-fold in the presence of exogenously expressed c-Jun (Fig. 3). This increase in transcription activity was not due to the increase in the amount or in the DNA-binding activity of the GAL4-ERM fusion protein, because co-expression of c-Jun did not change either of these factors (data not shown). Neither JunD, c-Fos, CREB, nor CRE-BP1 showed significant effects on the transcriptional activity of the GAL4-ERM fusion protein (Fig. 3). These results indicated that c-Jun specifically enhances the transcriptional activity of ERM.
To investigate which part of ERM has transcription activation domain and which part is required for c-Jun enhancement of the transcriptional activity of ERM, we tested a series of fusion proteins containing the Gal4 DNA-binding domain and different segments of ERM for the ability to activate 5ϫ Gal4site-driven gene expression without or with exogenous c-Jun expression (Fig. 4). Deletion of the carboxyl-terminal region including the ETS domain of ERM (pGAL326) increased transcription activity by 4-fold as compared with the basal activity of Gal4 protein alone. Further deletion of the carboxyl-terminal portion, pGAL276, resulted in a 10-fold increase in transcription activity. The amino-terminal portion containing an acidic region (pGAL165) showed a 69-fold increase in transcription activity. However, regions other than the amino-terminal portion of ERM (pGAL166 -510, pGAL327-510) seemed not to have transcription activity. The part from amino acid 166 to 326 of the ERM protein may contain a negative regulatory domain(s). Among these fusion proteins, the enhancing effect of c-Jun was evident only when the fusion proteins contained both the region of amino acids 166 -276 and the amino-terminal portion of amino acids 1-165, such as in pGAL510, pGAL326, and pGAL276. Consistent with this, c-Jun did not activate the transcription activity of pGAL166 -510 nor that of pGAL327-510. In contrast to the situation with GAL4-ERM, c-Jun had no effect on the transcription activity of GAL4-ER81, a fusion protein of the GAL4 DNA-binding domain, and a full-length ER81 that is closely related to ERM (Fig. 4), suggesting the specificity of ERM-c-Jun cooperation.
Chromosomal Location of ERM-Once we had established the synergy between ERM and c-Jun, which has been shown to be one of important gene products involved in growth, differentiation, and tumorgenicity (52), we decided to determine the chromosomal location of the ERM gene by fluorescence in situ hybridization. Of 61 metaphase cells examined, 14 exhibited twin-spot signals on both homologous chromosomes 3 at q27, and another 19 cells showed twin-spot signals on one chromosome 3q27 and a single spot on another 3q27 (Fig. 5). Such specific accumulation of signals could not be detected on any other chromosomes. These results indicate that the ERM gene is located on chromosome 3 at q27. DISCUSSION ERM together with PEA3 (13), ER81 (14), and EIA-F, the human homolog of PEA3 (41), comprises a subgroup in the ets gene family based on their sequence similarity throughout the entire molecules. In contrast to PEA3, which is expressed in a tissue-restricted manner, most abundant in the mouse epididymis and brain, and in cell lines of fibroblast and epithelial cell origin (13), both ERM and ER81 mRNA are expressed in a wide variety of tissues and cell lines (14,39). 2 The members of the PEA3 subfamily have an acidic region in their amino termini, which has been suggested as a transcription activation domain (13,24). We showed that using a Gal4 system, the amino-terminal portion of ERM did indeed have transcription activity (Fig. 4). Interestingly, in the same assay system, we showed that the region from amino acids 166 -326 of ERM might contain a negative regulatory domain. This notion was supported by our result showing that the Gal4 fusion protein containing the ERM region (amino acids 166 -326) exerted a repressor activity on the herpes simplex virus thymidine kinase promoter in the assay systems using a reporter plasmid that contained five copies of Gal4-binding sites upstream of the thymidine kinase promoter. 2 Since the GAL4-ER81 was not activated by c-Jun and the proline-rich region (amino acids 146 -195), not present in ER81, was located in the region of ERM essential for the activation of ERM by c-Jun (Fig. 4), it was possible that the proline-rich region was one of the targets for the action of c-Jun.
Two major cis-acting motifs, 12-O-tetradecanoylphorbol-13-acetate-responsive element (AP1 site) (53,54) and cAMP-responsive element (55,56) have been shown to play key roles in the signal transduction processes that lead to cell growth or differentiation. In some cases, transcription factors acting on the AP1 site or CRE site cooperate with other transcription factors to achieve efficient gene activation (27,57,58). Conversely, for optimal activity, Ets proteins seem to require another transcription factor binding to an adjacent site, including AP1 (3,4,59). Recent studies have indicated that cooperative functions between specific Ets and AP1 binding proteins are of significant importance in regulating inducible gene expression. For instance, Wasylyk et al. (1) showed that Ets-1 and Ets-2 functionally cooperated with AP1 composed of c-Fos and c-Jun in transcriptional activation from the PEA3-AP1 motif of the polyoma virus enhancer. More recently, Wang et al. (60) showed that in activated T cells, Elf-1, c-Fos, and JunB formed a complex binding to the PB1 motif (purine box 1 in the granulocyte macrophage colony-stimulating factor gene promoter), which is composed of an EBS and an adjacent AP1-binding site, both of which are required for inducible expression of the granulocyte macrophage colony-stimulating factor gene. Our data showed another case of synergy between an Ets family protein and c-Jun.
To elucidate the mechanisms for the synergy between ERM and c-Jun, we examined two possibilities: (i) cooperative binding of ERM and c-Jun to the response element and (ii) enhancement of one's transcriptional activity by the other. The first possibility was unlikely, since we did not detect cooperative binding of ERM and c-Jun to the EBS-CRE even in the presence of CRE-BP1 (data not shown). For the second possibility, we have presented evidence showing that c-Jun enhances the transcription activity of ERM in an assay system using Gal4 fusion proteins. This enhancement required both the aminoterminal transactivation domain and the portion containing a negative regulatory domain of ERM. In the in vitro proteinprotein interaction assays using 35 S-labeled c-Jun and various glutathione S-transferase-ERM fusion proteins, we detected substantial direct interaction between c-Jun and ERM through the carboxyl-terminal portion of ERM containing the Ets domain, although we failed to detect a direct interaction of the amino-terminal portion of ERM, including a negative regulatory domain, and c-Jun. 2 Since the amino-terminal but not the carboxyl-terminal portion of ERM was required for the activation of ERM by c-Jun, a direct interaction between c-Jun and ERM may not occur. It is conceivable that c-Jun may neutralize the squelching effect of the negative regulatory domain of ERM on the transactivation domain. We currently do not know the significance of the direct interaction of c-Jun and the ERM carboxyl-terminal portion.
Recent studies have indicated that phosphorylation of transcriptional activators is important in either translocation of factors to the nuclei, binding to the relevant sites in the control regions, or interaction with other transcription factors or coactivators (57,(61)(62)(63). In the ets gene family, Elk-1, acting on the c-fos serum response element in concert with serum response factor as a ternary complex, has been shown to be transcriptionally activated through phosphorylation by a mitogen-activated protein (MAP) kinase (31,64). PU.1 recruits the binding of a second B cell-restricted nuclear factor, NF-EM5, to a DNA site in the immunoglobulin 3Ј enhancer (32). Recently, phosphorylation of PU.1 at Ser 148 was shown to be necessary for interaction with NF-EM5 (65). Since ERM cooperates with c-Jun, a well-known signal-transducing transcription factor, such a signal-transducing property might be expected for ERM. In relation to this matter, ERM has three putative mitogenactivated protein kinase phosphorylation sites (PX(S/T)P), at FIG. 5. Chromosomal localization of the human ERM gene by fluorescence in situ hybridization. A, twin-spot signals specific for ERM (arrows) were detected on both sister chromatids of the long arm of chromosome 3. B, G-band patterns of the same chromosomes were delineated through a UV filter, indicating that the human ERM gene is located on 3q27.
Thr 135 , Thr 139 , and Ser 142 . The effects of phosphorylation of ERM and c-Jun on their synergistic action should be analyzed to establish the precise role of ERM in signal transduction.
It is of interest that combinations of adjacent Ets and CRE or related motifs are found in several gene promoters, including junB (37), Rb (66), CD8␣ (67), and TCR␣ (28). The significance of the synergy between ERM and c-Jun in transcriptional activation of such cellular genes, actual target genes, remains to be established. Other functional aspects of ERM, such as oncogenicity exerted in concert with c-Jun and the roles of ubiquitously expressed ERM in development will be the focus of our future work. Information on the location of the ERM gene at 3q27 will be useful in searching for naturally occurring tumors caused by the abnormal expression or activity of ERM.