Interaction of the PutativeAndrogen Receptor-SpecificCoactivator ARA70/ELE1a withMultiple Steroid Receptors andIdentification of an InternallyDeleted ELE1b Isoform

Philippe Alen, Frank Claessens, Erik Schoenmakers,Johannes V. Swinnen, Guido Verhoeven, Wilfried Rombauts, andBen Peeters

Division of Biochemistry (P.A., F.C., E.S., W.R., B.P.) andLaboratory for Experimental Medicine and Endocrinology

(J.V.S., G.V.)Faculty of Medicine, Campus GasthuisbergUniversity of LeuvenB-3000 Leuven, Belgium

Steroid-regulated gene transcription requires thecoordinate physical and functional interaction ofhormone receptors, basal transcription factors,and transcriptional coactivators. In this contextARA70, previously called RFG and ELE1, has beendescribed as a putative coactivator that specifi-cally enhances the activity of the androgen recep-tor (AR) but not that of the glucocorticoid receptor(GR), the progesterone receptor, or the estrogenreceptor (ER). Here we describe the cloning of thecDNA for ELE1/ARA70 by RT-PCR from RNA de-rived from different cell lines (HeLa, DU-145, andLNCaP). In accordance with the previously de-scribed sequence, we obtained a 1845-bp PCRproduct for the HeLa and the LNCaP RNA. Startingfrom T-47D RNA, however, an 860-bp PCR productwas obtained. This shorter variant results from aninternal 985-bp deletion and is called ELE1b; ac-cordingly, the longer isoform is referred to asELE1a. The deduced amino acid sequence ofELE1a, but not that of ELE1b, differs at specificpositions from the one previously published by oth-ers, suggesting that these two proteins are en-coded by different nonallelic genes. ELE1a is ex-pressed in the three prostate-derived cell linesexamined (PC-3, DU-145, and LNCaP), and this ex-pression is not altered by androgen treatment. Ofall rat tissues examined, ELE1a expression is high-est in the testis. This is also the only tissue in whichwe could demonstrate ELE1b expression. BothELE1a and ELE1b interact in vitro with the AR, but

also with the GR and the ER, in a ligand-indepen-dent way. Overexpression of either ELE1 isoform inDU-145, HeLa, or COS cells had only minor effectson the transcriptional activity of the human AR.ELE1a has no intrinsic transcription activation do-main or histone acetyltransferase activity, but itdoes interact with another histone acetyltrans-ferase, p/CAF, and the basal transcription factorTFIIB. The interaction with the AR occurs throughthe ligand-binding domain and involves the regioncorresponding to the predicted helix 3. Mutation inthis domain of leucine 712 to arginine greatly re-duces the affinity of the AR for ELE1a but has onlymoderate effects on its transcriptional activity.Taken together, we have identified two isoforms ofthe putative coactivator ARA70/ELE1 that may actas a bridging factor between steroid receptors andcomponents of the transcription initiation complexbut which lack some fundamental properties of aclassic nuclear receptor coactivator. Further ex-periments will be required to highlight the in vivorole of ELE1 in nuclear receptor functioning. (Mo-lecular Endocrinology 13: 117–128, 1999)

INTRODUCTION

The androgen receptor (AR) is a member of the super-family of nuclear receptors (NR), which includes thereceptors for the other steroid hormones, for thyroidhormones, retinoids, and vitamin D (1–4). The NRsconstitute a large group of evolutionarily related pro-teins with a common structural organization, yet di-verse physiological functions. Apart from their role in

0888-8809/99/$3.00/0Molecular EndocrinologyCopyright © 1999 by The Endocrine Society

117

embryonic development, metabolic homeostasis, sexdetermination and differentiation, and fertility, they arealso implicated in a variety of pathologies, includingcancer.

NRs are ligand-inducible transcription factors with amodular structure, encompassing an N-terminal trans-activating domain, a centrally located DNA-bindingdomain (DBD), and a C-terminal ligand-binding do-main (LBD). They have two transcription activationfunctions, AF1 and AF2 (3, 4). AF1 is located in theN-terminal domain and can activate transcription con-stitutively (5, 6), whereas the AF2 function colocalizeswith the LBD and requires the addition of ligand (7).The crystal structures of the LBDs of unliganded hu-man retinoid X receptor-a (hRXRa) (8), of ligand-boundhuman retinoic acid receptor-g (hRARg) (9), thyroidhormone receptor (TR) (10), and progesterone recep-tor (PR) (11) and of agonist and antagonist-boundestrogen receptor (ER) (12) revealed a common overalllayered structure consisting of 12 a-helices. The mostremarkable difference between free and ligand-bound

receptors is the position of the most C-terminally lo-cated helix 12 (H12), which contains the core AF2-domain. In the absence of ligand, H12 is projectedaway from the hydrophobic ligand-binding pocket.Binding of ligand repositions this helix, thereby closingthe hydrophobic core as a lid (13).

Although direct interactions between NRs and com-ponents of the basal transcription machinery havebeen described (14–17), they cannot fully explain theligand-dependent transactivation of AF2. Therefore,another class of proteins called coactivators must betaken into account (18, 19). A number of candidate NRcoactivators have been described, including RIP140(20, 21), TIF1 (22), TRIP-1/SUG1 (23, 24), and a familyof related 160-kDa proteins comprising SRC-1 (25–27), GRIP1/TIF2 (28–30), and RAC3/AIB1/ACTR/TRAM-1/p/CIP (31–35). While the functions of RIP140,TIF1, and TRIP-1/SUG1 remain to be established, thep160 proteins exhibit all the characteristics expectedfor NR coactivators: 1) they interact with the LBD ofNRs in a ligand-dependent and AF2 integrity-depen-dent way both in vivo and in vitro through a conserveda-helical LXXLL-motif (27, 36–40); 2) upon cotransfec-tion in mammalian cells they potentiate the ligand-dependent transcriptional activity of several NRs; 3)they harbor autonomous transcription activation func-tions; and 4) they have intrinsic histone acetyltrans-ferase (HAT) activity and/or interact with yet otherHATs such as CREB-binding protein (CBP) and p300/CBP-associated factor (p/CAF). Thus, apart from therecruitment of basal transcription factors to the pro-moter and the formation of a stable preinitiation com-plex, steroid receptor transactivation of target genes invivo also involves chromatin remodeling, probablythrough targeted histone acetylation by the recruitedcoactivators (41).

Yeh and Chang (42) reported the first receptor-spe-cific coactivator, ARA70 (70 kDa AR-activator), whichstimulates the transcriptional activity of the AR but notthat of the glucocorticoid receptor (GR), the PR, or theER. This protein had originally been identified by oth-ers in thyroid cancer cells and named RFG (43) andELE1 (44, 45). In many cases of thyroid papillary car-cinoma, the RET protooncogene, a transmembranereceptor of the tyrosine kinase family, is found acti-

Fig. 1. ELE1 Is Expressed as at Least Two Different Iso-forms, ELE1a and ELE1b

A, Total RNA from T-47D cells, HeLa cells, and LNCaPcells was reverse transcribed as indicated in Materials andMethods. The first strand cDNA was subsequently PCR-amplified with two ELE1-specific primers. The reaction prod-ucts were separated by electrophoresis in a 1% agarose geland visualized by ethidium bromide staining. The length (inbasepairs) of the bands in the 1-kb DNA ladder (lane M) isdepicted on the right. Arrowheads indicate the positions ofthe 1845-bp ELE1a and the 860-bp ELE1b PCR products. B,Schematic representation of ELE1a and ELE1b and of thechimeric protein RET/PTC 3. RET/PTC 3 consists of the first238 aa of ELE1, fused in frame to the tyrosine kinase domainof the RET protooncogene. ELE1b lacks aa 239–565 fromELE1a. The glutamine at position 238 is the last residueencoded by exon 5 of the ele1 gene.

Table 1. Point Mutations Detected in the ELE1a but Not inthe ELE1b cDNA

Position Base Change Amino Acid Change

281 C to T Ser to Leu460 T to C Phe to Leu483 A to G No change

1048 T to C Cys to Arg1682 T to C Leu to Pro

Irrespective of the cell line used to obtain the RNA or thethermostable polymerase used to amplify the cDNA, fivepoint mutations were consistently detected in ELE1a. In com-parison with the published sequence (42, 43), the basechanges and their location are presented.

MOL ENDO · 1999 Vol 13 No. 1118

vated, due to chromosomal rearrangements resultingin recombinant genes. One such recombination,caused by a paracentric inversion within band q11.2 ofchromosome 10 (45), leads to the fusion of thegenomic region encoding the RET tyrosine kinase (TK)domain with the 59-terminal region of the ele1 gene(43, 44). The resulting protein, called RET/PTC3, isexpressed under the control of the ele1 gene promoterand consists of the first 238 amino acids (aa) of ELE1fused to the RET TK-domain.

The goal of our study was to characterize the puta-tive AR-specific coactivator properties of ARA70/RFG/ELE1. Since our results indicate that it lacks some ofthe fundamental properties of a bona fide NR coacti-vator, we prefer the nomenclature ELE1 to ARA70.

RESULTS

cDNA Cloning

To obtain the ELE1 cDNA, we performed RT-PCR ontotal RNA from HeLa cells, LNCaP cells, and T-47Dcells using primers based on the sequence publishedby Yeh and Chang (42). For the HeLa and LNCaP RNA,a cDNA of the expected 1845 bp was obtained, butsurprisingly, the T-47D RNA generates a 860-bp cDNA(Fig. 1A). The full-length cDNA will further be referredto as ELE1a; accordingly, the shorter variant will becalled ELE1b. Sequence analysis revealed that ELE1bresults from an internal 985-bp deletion from nucleo-tides 790-1674 (the A of the ATG translation start-

codon was arbitrarily given the number 11). This de-letion does not cause a frame shift but results in acDNA encoding a protein of approximately 35 kDa,and the 238 aa before the deletion correspond to theELE1-derived fragment in the RET/PTC3 fusion pro-tein (43). Moreover, Q238 is the last residue encodedby exon 5 of the ELE1 gene (46) (Fig. 1B). Remarkably,the ELE1a cDNA clones derived from HeLa RNA, aswell as those from LNCaP RNA, contain five basesubstitutions (see Table 1) as compared with the se-quence published by Yeh and Chang (42) and by San-toro and co-workers (43). We have analyzed severalELE1a clones from different PCR reactions and con-sistently found the same mutations, irrespective of theRNA or the polymerase used (Taq or Pfu). Further-more, in some clones amplified with Taq polymerasebut not in those obtained with the Pfu polymerase,which has a lower error rate, we found additional ran-dom mutations that only occurred in single clones andare therefore likely due to misreading by the polymer-ase. In the ELE1b cDNA, however, we did not find themutations listed in Table 1.

Expression of ELE1

Because of its putative role in androgen physiologyand the possible implications for the development ofprostate cancer, we examined the expression of ELE1at the RNA level in three different human prostatecancer-derived cell lines (DU-145, LNCaP, and PC-3),grown in the absence or the presence of the syntheticandrogen R1881. As shown in Fig. 2A, all cell lines

Fig. 2. ELE1 Expression in Human Prostate Cancer-Derived Cell Lines and Rat TissuesNorthern blot analysis of total RNA from three human prostate cancer-derived cell lines, LNCaP, PC-3, and DU 145, grown in

either the absence or the presence of 1 nM R1881(panel A), and of total RNA from different rat tissues (panel B). The blots werehybridized with a radiolabeled probe encompassing the first 638 bp of the ELE1a and the ELE1b cDNA. RNA integrity and equalsample loading were verified with a 18S probe. The positions of the 28S and the 18S RNA bands are indicated as size markers.The asterisk in panel B marks a band that possibly corresponds to the ELE1b mRNA.

ELE1/ARA70 Interacts with Steroid Receptors 119

express ELE1 and only contain the ELE1a mRNA ascan be concluded from the size of the radioactiveband. Of the three cell types tested, ELE1a expressionis highest in DU-145 cells, and the expression is notsubstantially affected by androgen treatment.

Since the main targets for androgens are the organsand tissues of the male reproductive tract, we alsoanalyzed the expression of ELE1 in rat prostate andtestis and made a comparison with other tissues. The

blot was probed with a radiolabeled cDNA fragmentencompassing the first 638 bp of the ELE1 cDNA,which can hybridize to both the ELE1a and the ELE1bmRNAs. As shown in Fig. 2B, ELE1a is expressed in alltissues examined at comparable and moderate levels.The lowest expression is found in the brain and theprostate, whereas the testis clearly contains muchhigher levels of ELE1a-mRNA. Moreover, in the testisa second and faster migrating mRNA band appears,

Fig. 3. Analysis of the in Vitro Interaction of ELE1 with NRsA (left panel), In vitro produced hAR was incubated with different concentrations of [3H]mibolerone, ranging from 0.1 to 10 nM,

in the absence or the presence of a 100-fold excess of unlabeled hormone. Free (F) and bound (B) mibolerone were separatedwith dextran-coated charcoal. The Kd (0.46 nM) was calculated from the slope of the Scatchard plot. A (right panel), 35S-labeledhAR was incubated without hormone, with 10 nM of DHT or R1881 or with 10 mM of hydroxyflutamide (HO-F) and subjected toa limited proteolytic digest as described (47). The reaction products were separated by SDS-PAGE and visualized by fluoro-graphy. B, 35S-labeled full-length hAR was produced by in vitro coupled transcription-translation in the absence of ligand or inthe presence of 10 nM R1881 or 10 mM HO-F and incubated with GST, GST-ELE1a, or GST-ELE1b immobilized on GlutathioneSepharose resin. After stringent washings, the resin-bound proteins were eluted with SDS-loading buffer, analyzed by SDS-PAGE, and visualized by fluorography. The input lane represents 10% of the reticulocyte lysate that was used in each experiment.C, 35S-labeled hAR, mER, and rGR were produced in vitro and tested for the interaction with GST, GST-ELE1a, and GST-ELE1bin the absence of ligand as described for panel B and in Materials and Methods. D (upper panel), Linear diagram of the hAR andthe deletion mutants used in this study. The numbers refer to amino acid positions. The thin line represents the N-terminal domain,the shaded box represents the DBD, and the bold line represents the LBD. D, (lower panel), The hAR and the deletion mutantsdepicted in the upper panel were produced and radiolabeled by in vitro transcription translation and subsequently analyzed forinteraction with GST or GST-ELE1a by the GST-pull down assay. Bound proteins were eluted with SDS-loading buffer, separatedby SDS-PAGE, and visualized by fluorography.

MOL ENDO · 1999 Vol 13 No. 1120

possibly corresponding to the shorter ELE1b variant.We could not detect it in any of the other tissuesexamined, even after prolonged exposure. It should benoted that a human cDNA fragment was used to probethe rat multiple tissue RNA blot that therefore had tobe exposed approximately 10 times longer to obtain aclear signal.

In Vitro Interaction of ELE1 with NRs

For the characterization of the interaction betweenELE1 and NRs, we performed in vitro glutathione S-transferase (GST) pull-down experiments. To ensurethat the in vitro produced human AR (hAR) binds hor-mone correctly and that it undergoes the associatedconformational changes, we measured the binding af-finity of the AR for [3H]mibolerone and analyzed thetryptic digest patterns of the ligand-free receptor andits complexes with agonists and antagonists. We mea-sured a Kd for mibolerone of 0.46 nM, which is con-sistent with the values reported by others (47) (Fig. 3A,left panel). Furthermore, in the absence of ligand thereceptor is completely digested by limited amounts oftrypsin. Addition of the agonists dihydrotestosterone(DHT) and R1881 or the antagonist hydroxyflutamide(HO-F), however, results in the protection of small butdifferent ‘cores’ against trypsinization, indicating thatthe specific conformational changes that are associ-ated to ligand binding do occur (Fig. 3A, right panel)(48).

To analyze the ELE1-AR interaction, 35S-labeledfull-length hAR was added to a glutathione Sepharosematrix to which either GST alone or a fusion protein ofGST with ELE1a or ELE1b was coupled. The proteinswere allowed to interact in the absence of ligand or inthe presence of an agonist (R1881) or an antagonist(HO-F). As shown in Fig. 3B, the AR interacts with bothELE1a and ELE1b, although this interaction is notligand dependent and is not antagonized by HO-F. Nodifferences were observed when either R1881 or DHTwere used to activate the receptor (data not shown).Furthermore, the interaction is not specific for the AR,at least in vitro, since the rat GR (rGR) and the mouseER (mER) also bind to ELE1 a and b with affinitiescomparable to that of the hAR (Fig. 3C).

To further characterize the region of the AR that isrequired for this interaction, we performed GST pull-downs with full-length ELE1a and C-terminally trun-cated receptors. In a first series of mutants, we deletedthe C-terminal half of the LBD and the complete LBDand found that the region between aa 563 and 772 issufficient for interaction with ELE1a (data not shown).

Fig. 4. Analysis of the Coactivator Activity of ELE1HeLa cells (upper panel) or COS7 cells (lower panel) were

transfected with 250 ng of pCMV-bGAL, 300 ng of MMTV-lucreporter vector, 100 ng of pSG5-hAR, and 600 ng of eitherempty pSG5, pSG5-ELE1a, pSG5-ELE1b, or pSG5-TIF2. Thecells were treated for 24 h without (open bars) or with (solidbars) 1 nM of R1881 before the luciferase and b-galactosi-dase activities were measured. The results shown are themean 6 SEM of three measurements. The luciferase activitymeasured in the absence of coactivator and in the presenceof R1881 was arbitrarily set to 100.

Fig. 5. ELE1a and ELE1b Interact in Vitro with TFIIB and withp/CAF

p/CAF, the general transcription factor TFIIB, CBP, and thep160 coactivators TIF2, SRC1a, and SRC1e were synthe-sized and radiolabeled in vitro in rabbit reticulocyte lysate.They were allowed to interact with purified GST, GST-ELE1a,or GST-ELE1b (for TFIIB and p/CAF), GST-CBP[2058–2163](for TIF2, SRC1a, and SRC1e), or GST-SRC1[781–988] (forCBP) prebound to a Glutathione Sepharose matrix, under theconditions described in Materials and Methods. After exten-sive washings the bound proteins were eluted, gel electro-phoresed, and visualized by fluorography. The arrows indi-cate the positions of the respective 35S-labeled proteins.

ELE1/ARA70 Interacts with Steroid Receptors 121

Fig. 6. In Vitro and in Vivo Analysis of hAR H3 MutantsA, Sequence alignment of the LBD region of SRs corresponding to the predicted H3, according to Wurtz et al. (49). The numbers

refer to amino acid positions. The mutations are indicated by asterisks and given below the sequence of the hAR. B, COS 7 cellswere transiently transfected with an expression vector for either the wild-type AR or for one of the mutants described in panel A.For the single-point ligand-binding assay, the cells were incubated with 1 nM 3H-labeled mibolerone with or without 100 nM coldmibolerone. After a 2-h incubation at 37 C, they were washed and the radioactivity that was specifically retained within the cells

MOL ENDO · 1999 Vol 13 No. 1122

Next, we further deleted the receptor to P722 (end ofexon 4, mutant D1), P670 (mutant D4), and G626 (endof exon 3, mutant D5). As shown in Fig. 3D, mutant D1still interacts with ELE1a, whereas D4 does not. TheN-terminal domain is not required for the interaction,since deletion of aa 1–537 had no effect on the binding(data not shown). Based on the alignment made byWurtz et al. (49), the region between P670 and P772corresponds to helices 1–3 at the N terminus of theLBD. We therefore generated mutants ending afterhelix 2 (mutant D2, aa 1–696) and after helix 1 (mutantD3, aa 1–683). These truncated receptors do not in-teract with ELE1a (Fig. 3D).

Thus, the domain that is sufficient for the in vitrointeraction of the hAR with ELE1a is located betweenE696 and P722 and corresponds to the predicted helix3 (H3).

Cotransfection of the hAR with ELE1 and TIF2

DU-145 prostate cancer cells were cotransfectedwith the mouse mammary tumor virus (MMTV)-lucif-erase reporter vector and expression plasmids forthe hAR and either ELE1a or ELE1b, but we did notobserve strong coactivator properties (data notshown). Since we have demonstrated that DU-145cells express ELE1a, they offer no obvious advan-tage over HeLa cells for use as a model system toanalyze the properties of ELE1. Therefore, thesecells were also cotransfected with the hAR andELE1a or ELE1b. As shown in Fig. 4 (upper panel),coexpression of either ELE1 isoform results in amaximal 2-fold increase of AR activity (comparelanes 2 and 3 to lane 1). As a control, we includedthe p160 coactivator TIF2 (lane 4) which, under thesame conditions, stimulated the hAR 4-fold. SinceTIF2 has been shown to function better in COS cellsthan in HeLa cells (30), we performed the sameexperiments in COS-7 cells. As shown in Fig. 4(lower panel), also in these cells coexpression ofELE1a or ELE1b has only moderate effects on theactivity of the hAR (2- to 3-fold increase in receptoractivity), whereas coexpression of TIF2 results in analmost 20-fold increase.

In Vitro Interaction of ELE1 with p/CAF and TFIIB

Many transcriptional coactivators have intrinsic HATactivity or act as a platform for the recruitment of otherHATs and the basal preinitiation complex. Using theliquid HAT assay, we could clearly demonstrate HATactivity for bacterially produced p/CAF but not for aGST-ELE1a fusion protein. Single hybrid experimentsin HeLa cells also did not reveal an intrinsic transcrip-tion activation function for a GAL4-ELE1a fusion pro-tein (data not shown).

We next tested the ability of ELE1a and ELE1b tointeract in vitro with p/CAF, TFIIB, the cointegratorCBP, and the p160 coactivators SRC-1a, SRC-1e, andTIF2 (Fig. 5). We observed in vitro binding to p/CAFand to TFIIB. The in vivo relevance of these interac-tions needs to be further analyzed. Coactivators of thep160 family and CBP did not bind to the GST-ELE1a

fusion protein, whereas mutual interactions betweenthe two first components were detected in controlexperiments.

Analysis of H3-Mutants of the hAR

We have shown that the region in the hAR required forthe in vitro interaction with ELE1a corresponds to thepredicted H3 of the LBD (Fig. 3D). To identify singleresidues that are involved in this interaction, we gen-erated point mutations in this domain. Since in vitrothe interaction is not AR specific, we chose to mutatethe following moderately or highly conserved aminoacids in the C-terminal part of H3: leucine 712 toarginine (L712R), valines 715 and 716 to alanine(V715A/V716A), tryptophan 718 to alanine (W718A),alanine 719 to lysine (A719K), and lysine 720 to alanine(K720A) (Fig. 6A). The ability of these mutants to bindhormone was addressed in a single point ligand-bind-ing assay (1 nM mibolerone) of transiently transfectedCOS 7 cells (Fig. 6B). The mutants V715A/V716A andK720A can bind hormone equally well as the wild-typereceptor, whereas L712R has a lower affinity. W718Aand A719K, on the other hand, retained less than 10%of the amount of ligand bound by the wild-type recep-tor. A Western blot with extracts from transfected COS

was measured. The shown values are the mean of five measurements 6 SEM. The activity measured for the wt AR was arbitrarilyset to 100. The lower panel shows a Western blot of extracts of COS 7 cells, probed with an anti-AR antibody, to control theexpression of the respective mutants. C, In vitro interaction assays of GST or GST-ELE1a and 35S-labeled wt or mutated hAR wereperformed as in Fig. 3B and as described in Materials and Methods. The mutations are described in panel A. D, HeLa cells weretransiently transfected with the MMTV-luc (left panel) or the (GRE)2-oct-luc (right panel) reporter vectors (1 mg) together withexpression vectors for either the wt hAR or one of the H3 mutants (100 ng) and pCMV-bGAL (250 ng) as internal control. The cellswere treated with (solid bars) or without (open bars) 1029 M R1881 for 24 h before harvesting and measuring the luciferase andb-galactosidase activities. The activity of the wt hAR in the presence of hormone was taken as 100, and the relative values of testmeasurements were calculated. At least three independent duplicate experiments were performed, and the results are shown asthe mean value 6 SEM. E, HeLa cells were transfected with MMTV-luc and expression vectors for the wild-type AR (triangles) orthe L712R mutant (squares) as in Fig. 6D and treated for 24 h without hormone or with 0.001, 0.01, 0.1 or 1 nM of R1881 beforethe b-galactosidase and the luciferase activity were measured. The activity of the wt hAR in the presence of 1 nM R1881 was takenas 100.

ELE1/ARA70 Interacts with Steroid Receptors 123

cells was performed to ensure adequate expression ofall receptor mutants.

The ability of these mutants to interact with ELE1awas tested in a GST pull-down assay. As shown in Fig.6C, the mutant L712R has a reduced affinity for ELE1acompared with the wild-type receptor, since less pro-tein is retained on the GST-ELE1a beads. The mutantsV715A/V716A, W718A, and A719K have affinitiescomparable to that of the unmodified receptor,whereas that of K720A is slightly higher. We examinedthe transcription-activating properties of these mutantreceptors in transient transfection experiments. HeLacells were cotransfected with MMTV-luc or (GRE)2-oct-luc reporter constructs and expression vectors foreither the wild-type receptor or one of the mutants. Asshown in Fig. 6D, all mutants retain some ability toactivate the transcription of the MMTV-driven reporterin response to 1 nM R1881. Comparable results areobtained with the (GRE)2-oct-luc reporter. Mutation ofleucine 712 to arginine caused only a slight (;35%)decrease in the ligand-dependent transcriptional ac-tivity of the AR, despite the clear effects of this muta-tion on the AR-ELE1a interaction. Surprisingly, muta-tion of the highly conserved lysine at position 720 toalanine, which severely impaired the transcription ac-tivation function of the mER (50), had no effect on thehAR. The double mutant V715A/V716A showed aslightly higher activity on the (GRE)2-oct-luc reporter,but a slightly reduced activity on the MMTV-lucreporter.

Next, we compared the dose-response curve for thewild-type receptor and the L712R mutant in HeLacells, using MMTV-luc as reporter and increasing con-centrations of R1881 (0.001 to 1 nM). As shown in Fig.6E, the response of the L712R mutant to a wide rangeof R1881 concentrations is similar to that of the un-mutated receptor, except that the maximal responseto androgens is allways ;35% lower, which is consis-tent with its reduced affinity for the ligand.

DISCUSSION

The goal of our study was to characterize the interac-tion of ELE1 with the hAR and to evaluate its role as anAR-specific transcriptional coactivator. Therefore wecloned the ELE1 cDNA by RT-PCR. In addition to theexpected full-length cDNA (ELE1a), we obtained aninternally deleted variant (ELE1b). This deletion is likelycaused by the removal of one or more exons, since thelast residue before the deletion (Q238) is the last aminoacid encoded by exon 5 of the ele1 gene (46). More-over, the NH2-terminal region of 238 aa from ELE1(a and b) corresponds to the ELE1-derived fragmentin the RET/PTC 3 fusion-protein (43). Specific basechanges in the ELE1a but not in the ELE1b cDNAindicate, furthermore, that both proteins are encodedby two nonallelic genes with different expression pat-terns and RNA processing characteristics.

In view of its proposed role as an AR-specific co-activator, we analyzed the expression of ELE1 RNA indifferent rat tissues and in three human prostate can-cer-derived cell lines. All cells, including DU-145 cells,were shown to express ELE1a. A possible explanationfor the lack of ELE1a expression in these cells re-ported by Yeh and Chang (42) could be that theseauthors used a high-passage cell line (personal com-munication) with altered characteristics comparedwith our cells, or that they used a different DU-145subclone. The high level of ELE1a expression in the rattestis could be indicative of its role in AR functioning.The prostate, however, another important androgentarget tissue, contains much less ELE1a.

We tested the coactivator properties of ELE1a andELE1b in cotransfection experiments, and for bothisoforms we found a very weak (;2-fold) increase inAR transcripional activity, irrespective of the cell lineused (DU-145, HeLa, or COS7). The very poor effectsof cotransfection of the AR with ELE1a or ELE1b couldbe due to sufficient levels of endogenous protein in allcells. It should be noted, however, that under the sameconditions the p160 coactivator TIF2 effeciently en-hances the AR activity, most notably in COS cells,despite the fact that the p160 family members are alsoubiquitously expressed.

The interaction between the hAR and ELE1 wasreported to be ligand dependent in yeast and specificfor the AR (42). In vitro, we observed a strong interac-tion of both ELE1a and ELE1b with the hAR, althoughthis interaction does not require the addition of ligand.The affinity of the AR is comparable for both ELE1isoforms, indicating that the 328 amino acid regiondeleted in ELE1b is not involved in this interaction.This apparent discrepancy between the in vitro ligand-independent interaction described here and the liganddependency observed in yeast (42, 51) might be ex-plained by the fact that in yeast cells, additional pro-teins may interact with the AR-LBD in the absence ofligand, thereby preventing interaction with ELE1. Ad-dition of hormone could then release this block andallow the LBD to bind to ELE1. Apart from the hAR,other steroid receptors (mER and rGR) also interact invitro with ELE1. Thus, the interaction is not receptorspecific. Furthermore, Treuter et al. (51) recently re-ported the cloning of the ELE1 cDNA by a yeast dou-ble-hybrid screen using the LBD of the peroxisomeproliferator-activated receptor as bait. Thus, ELE1 alsointeracts with NRs that do not belong to the subclassof the SRs. This is not exceptional, however, sincenone of the coactivators that have been described sofar display receptor specificity. The ER, as well as thePR, TR, RAR, and RXR, interacts with all of the p160family members (25, 28–31, 33, 35). Likewise, Trip1interacts with both the TR and RXR (28), and TIF1interacts with RXRa, RARa, vitamin D receptor, PR,and ER (23, 24).

We mapped the domain of the AR that is required forthe interaction with ELE1a to the region between aa696 and 722, which corresponds to the predicted H3

MOL ENDO · 1999 Vol 13 No. 1124

of the LBD (49). This region has been shown to beimportant for transcription activation by the ER (50).Moreover, crystallographic studies on the LBDs of theER, TR, PR, RXR, and RAR have indicated that bindingof the appropriate ligand repositions helix 12, com-prising the core AF2, which subsequently makes directcontacts with amino acids in H3 and H4, thereby pro-viding a new surface for interaction with coactivators(13). Feng et al. (52) systematically mutated specificresidues in the LBD of the TR and found that the AF2is formed by residues from the helices 3, 5, 6, and 12.The AF2 core is dispensable for the hAR-ELE1a inter-action, which again demonstrates that the interactionbetween the AR and ELE1 is clearly different from theinteraction between the p160 coactivators and NRs.Furthermore, both ELE1a and ELE1b contain a LXXLLmotif (LYSLL) between aa 92 and 96. It has beenshown that the integrity of the AF2 core domain isessential for the interaction of NRs with this motif inp160 coactivators (37). Since a truncated receptorlacking H12 can still interact with ELE1a, it is veryunlikely that this binding occurs through the LYSLLmotif.

ELE1 has no intrinsic transactivating properties anddoes not interact with CBP, but it can bind to TFIIBand to p/CAF. Although the in vivo relevance of theseprotein-protein interactions needs to be examined inmore detail, it is tempting to speculate that ELE1 mightfunction as a linker protein between the NR and thebasal transcription machinery, thereby recruiting orstabilizing the PIC on the promoter. Furthermore,through the interaction with p/CAF it might contributeto the recruitment of HATs and the subsequent chro-matin remodeling. Such an activity, however, wouldnot become apparent under the conditions of transienttransfection as used in this work.

We tried to identify single residues that are criticalfor the interaction of ELE1 with the hAR. Sequencecomparison of the H3 regions of steroid receptorsreveals that this domain is best conserved at the Cterminus of the helix. Mutation of a series of moder-ately or highly conserved residues in H3 revealed thatthe leucine at position 712 is essential for ELE1 bind-ing. In the hGR the corresponding residue is a valine,and in the human mineralocorticoid receptor it is amethionine. All these residues have hydrophobic sidechains, indicating that the binding between ELE1a andthe AR could occur through hydrophobic interactions.Despite the dramatic effects of this mutation on theELE1-hAR interaction, the mutated receptor is stilltranscriptionally active on both the MMTV-luc and the(GRE)2-oct-luc reporters, although the maximal activ-ity is ;35% less than that of the wild-type receptor.This correlates well with the reduced ligand-bindingaffinity observed in COS 7 cells. Thus, the slight de-crease in transcriptional activity most probably reflectsa decrease in ligand binding, rather than resulting froman impaired interaction with ELE1. A striking observa-tion in the analysis of the H3 mutants is the fact thatmutation of the highly conserved lysine residue at

position 720 to an alanine (K720A), which severelyimpairs the function of the ER (50), had no effect on thetranscriptional activity of the AR. This indicates thatthe mechanism of activation by the AR may be differ-ent from the ER. Similar conclusions can de drawnfrom other observations: whereas the ER-LBD harborsan autonomous transcription activation function (AF2)when fused to a heterologous GAL4-DBD (7, 50), theAR-LBD requires the NH2-terminal AF1 function togain ligand-dependent transcription activation proper-ties (53).

In conclusion, our results reveal the existence of twononallelic variants of ELE1. They confirm that ELE1interacts with the LBD of the hAR but also demon-strate that, at least in vitro, this interaction is not liganddependent and that ELE1 can also interact with othermembers of the NR superfamily. Although we demon-strate that ELE1 may act as a bridging factor betweensteroid receptors and the basal transcription machin-ery or chromatin-remodeling enzymes, it does not actas a classic coactivator in mammalian cells in ourexperimental conditions. In fact, mutation of specificresidues in the LBD of the hAR that severely impair theAR-ELE1 interaction have only moderate effects onthe AR transcriptional activity. Thus, the in vivo func-tion of ELE1 should be further investigated.

MATERIALS AND METHODS

Materials

All restriction and modifying enzymes were obtained fromeither Pharmacia (Uppsala, Sweden), Life Technologies(Grand Island, NY), or Boehringer (Mannheim, Germany). Cellculture reagents were obtained from Life Technologies. TheTNT rabbit reticulocyte lysate in vitro coupled transcriptiontranslation kit was purchased from Promega (Madison, WI).L-[35S]methionine (.1000 Ci/mmol) and [3H]mibolerone (82.3Ci/mmol) were purchased from Amersham (Buckingham-shire, UK). The hormones 5a-dihydrotestosterone, R1881(methyltrienolone), and mibolerone were from Dupont-NewEngland Nuclear (Boston, MA). Hydroxyflutamide was a kindgift from Schering (Kenilworth, NJ).

Plasmids

For the construction of pSG5-hAR, the hAR cDNA was PCRamplified from pSV-ARO (54) with Expand High Fidelity poly-merase (Boehringer) in the presence of 4% dimethylsulfoxideand using the primers 59-GGTGGATCCATGGAAGTGCAG-TTAGGGCTGGGAAGGTCTAC-39 and 59-TACGTGGATCCT-CACTGGGTGTGGAAATAGATGGGCTTGACT, and the re-sulting PCR product was cloned in the BamHI restriction siteof pSG5. Deletion mutants were made by PCR using the Pfuthermostable polymerase (Stratagene, La Jolla, CA) andpSG5-hAR as template and the following primers: 59-TAG-GATCCATGTTGGAGACTGCCAGGGAC-39 (forward primer;FP), 59-TAAGATCTGGATCCTCAAGGCAAGGCCTTGGCCC-39(reverse primer for D1), 59-TAAGATCTGGATCCTCAAAAG-GAGTCGGGCTGGTT-39 (reverse primer for D2), 59-TAA-GATCTGGATCCTCATACACCTGGCTCAATGGC-39 (reverseprimer for D3), 59-TAAGATCTGGATCCTCAGGGCTGACA-TTCATAGCC-39 (reverse primer for D4), 59-TAAGATCTGGATC-CTCATCCCAGAGTCATCCCTGC-39 (reverse primer for D5).

ELE1/ARA70 Interacts with Steroid Receptors 125

The resulting PCR products were digested with HindIII and BglIIand exchanged for the corresponding fragment in pSG5-hAR.

Site-directed mutagenesis was performed using a stan-dard PCR-based method. The plasmid pCR(538–918;wt) wasconstructed by inserting a fragment encoding the entire DBDand LBD, PCR amplified from pSG5-hAR with the primers FPand 59-GCTGCAATAAACAAGTTCTGC-39 (reverse primer;RP), and subsequently cloned in the pCR-SCRIPT vector(Stratagene). Next, two sets of PCR reactions were per-formed. In the first set, with pSG5-hAR as template, oligonu-cleotide FP was used as forward primer and the followingoligonucleotides as reverse primers: 59-CACGTGTACACG-CTGTCTCTC-39 (PCR a), 59-CTTGTACACGCCGCCAAGT-GGGCCAAG-39 (PCR c), 59-GTGGTCAAGGCGGCCAAG-GCC-39 (PCR e), 59-GTCAAGTGGAAGAAGGCCTTG-39 (PCRg), and 59-CAAGTGGGCCGCGGCCTTGCC-39 (PCR i). Sim-ilarly, the oligonucleotide RP was used as reverse primer andthe following oligonucleotides as forward primers: CACGT-GTACACGCTGTCTCTC-39 (PCR b), 59-CTTGGCCCACTTG-GCGGCGTGTACAAG-39 (PCR d), 59-GGCCTTGGCCGCCT-TGACCAC-39 (PCR f), 59-CAAGGCCTTCTTCCACTTGAC-39(PCR h), and 59-GGCAACGCCGCGGCCCACTTG-39(PCR j). The templates for the second set of reactions, usingthe oligonucleotides FP and RP as primers, were a combi-nation of the reaction products from PCRs a and b, c and d,e and f, g and h, and finally i and j. The PCR products fromthis second set of reactions were digested with HindIII andBglII and exchanged for the corresponding fragment in pCR-(538–918;wt), giving rise to pCR-(538–918;ab), pCR-(538–918;cd), pCR-(538–918;ef), pCR-(538–918;gh), and pCR-(538–918;ij). Finally, Tth111I-NcoI fragments from theseconstructs were exchanged for the corresponding fragmentin pSG5-hAR, resulting in AR expression plasmids carryingthe following mutations: L712R, V715A/V716A, W718A,A719K, and K720A. For the construction of pSG5-TFIIB, theTFIIB cDNA was isolated from pGEX-TFIIB (a gift from Dr. B.O’Malley) and subcloned in pSG5. Expression vectors forTIF2 and SRC-1 were kindly provided by Dr. H. Gronemeyerand Dr. M. G. Parker. The reporter gene construct (GRE)2-oct-luc, which contains two GREs in front of the minimalTATA box derived from the oct6 gene promoter and theluciferase gene, was the kind gift of Dr. A. O. Brinkmann.

Molecular Cloning of the ELE1 cDNA

The ELE1 cDNA was cloned by RT-PCR from RNA derivedfrom HeLa cells, LNCaP cells, and T-47D cells. Briefly, 3 mgof total RNA were reverse transcribed with the primer 59-CTAGGAATTCTCACATCTGTAGAGGAGTTC-39 (30 min at42 C), which generates an EcoRI restriction site immediatelydownstream of the ELE1 stop codon. The remaining RNAwas digested with RNase H for 10 min at 55 C, and thefirst-strand cDNA was PCR amplified using oligonucleotide59-GATCGAATTCATATGAATACCTTCCAAGAC-39 as for-ward primer and the above mentioned oligonucleotide asreverse primer. PCR was performed with either Taq or Pfu(Stratagene) thermostable polymerases. The reaction prod-ucts were gel purified and subsequently cloned in thepGEM-T (Promega) or pCR-SCRIPT (Stratagene) vectors, re-spectively, and analyzed by sequencing. For expression as aGST-fusion protein, the cDNA was cloned in the EcoRI site ofthe pGEX-2TK vector (Pharmacia); for expression in mam-malian cells or in vitro transcription translation, it was clonedin the EcoRI site of pSG5.

Northern Blotting

RNA from LNCaP, PC-3, and DU-145 cells, grown in either theabsence or the presence of 1 nM R1881, was prepared, elec-trophoresed, and blotted as described (55, 56). The probe forNorthern hybridization was prepared by PCR with the primers59-GATCGAATTCATATGAATACCTTCCAAGACC-39 and 59-C-

CACTGGCAGGTTTGCTTCC-39, which generates a fragmentcomprising the first 638 bp of the ELE1 cDNA. One microliter ofthis reaction was used in a 100-ml labeling reaction as previouslydescribed (55). The 18S RNA probe was labeled by randompriming. The membrane was prehybridized for 2 h at 42 C in 5 3SSC (20 3 SSC: 3 M NaCl and 3 M sodium citrate), 2 3 Den-hardt’s (100 3 Denhardt’s: 2% Ficoll 400, 2% polyvinylpyroli-done) containing 50% (vol/vol) formamide, 0.1% SDS, and 100mg/ml sonicated salmon sperm DNA and subsequently hybrid-ized with probe (106 cpm/ml) for 18 h at 42 C in the same buffer.The blot was washed once at room temperature with 2 3 SSCcontaining 0.5% SDS and twice for 20 min at 65 C with 0.2 3SSC containing 0.1% SDS and finally autoradiographed at 280C using intensifying screens.

In Vitro Ligand Binding and Limited TrypsinizationAssay

The in vitro ligand-binding and limited trypsinization assayswere performed as described previously (47, 48).

GST Pull-Down Assay

Expression and purification of GST or the GST fusion proteinsusing the pGEX-system (Pharmacia) and coupled in vitrotranscription-translation with the TNT system (Promega) wereperformed according to the manufacturer’s instructions. 35S-labeled proteins were added to the glutathione Sepharosecontaining either GST or the GST fusion protein in NENT/B[NENT buffer (100 mM NaCl, 1 mM EDTA, 0.02% NP-40, 20mM Tris, pH 8) containing 1 mg/ml BSA] in a total volume of200 ml and incubated for 30 min at room temperature and for30 min at 4 C. The glutathione Sepharose was washed fivetimes with 1 ml of NENT/B. Bound proteins were eluted byboiling the resin in 30 ml of 2 3 SDS loading buffer (20%glycerol, 2% SDS, 0.0025% bromophenol blue, 10% b-mer-captoethanol), separated by SDS-PAGE and visualized byfluorography. For the interactions of ELE1 with TFIIB, SRC-1,TIF2, p/CAF, and CBP, NENT buffer without BSA but con-taining 0.1% NP-40 was used.

Cell Culture

HeLa, T-47D, LNCaP, DU-145, and COS 7 cells were ob-tained from the Amercian Tissue Type Collection (ATTC, Ma-nassas, VA) and were routinely maintained in DMEM (LifeTechnologies) containing 1 g/liter glucose and supplementedwith antibiotics (penicillin, streptomycin; Life Technologies)and 10% heat-inactivated FBS. For DU-145 cells, L-glu-tamine was added to a concentration of 2 mM.

Transfections

Cells were transfected using the standard calcium phosphatecoprecipitation procedure (57). Twenty four hours beforetransfection, HeLa cells, COS 7 cells, or DU-145 cells weretrypsinized and seeded in 24-well culture plates at a densityof 7.5 3 104 cells per well in medium containing 5% dextran-coated charcoal-stripped FBS (DCC-FBS). For cotransfec-tion studies, the following amounts of plasmids were used(per well): 300 ng MMTV-luc, 100 ng pSG5-hAR, and 600 ngof either empty pSG5, pSG5-ELE1a, pSG5-ELE1b, or pSG5-TIF2. For the analysis of the different receptor mutants, HeLacells were transfected with 1 mg of reporter vector (MMTV-lucor (GRE)2-oct-luc) and 100 ng of expression vector for thewild-type or mutated AR per well. p-CMV-b-GAL (250 ng)was always included as internal control. The cells were incu-bated for 24 h in the absence or the presence of 1 nM R1881before the luciferase, b-galactosidase, and protein concen-trations were determined.

MOL ENDO · 1999 Vol 13 No. 1126

For the single-point ligand-binding assay and for Westernblotting, COS 7 cells were transfected with pSG5-hAR orexpression vectors for the respective receptor mutants. Fortyeight hours after transfection, the medium was replaced withDCC-FBS containing 1 nM [3H]mibolerone, either with or with-out a 100-fold excess of cold mibolerone. After 2 h incubationat 37 C, the cells were placed on melting ice, washed twicewith ice-cold PBS, and lysed in 1% Triton X-100, 0.1 NNaOH, and their radioactivity was counted. For Western blotanalysis, the cells were scraped in 1 ml PBS and pelleted bycentrifugation. The pellet was dissolved in 50 ml of high-saltextraction buffer (20 mM Tris, pH 7.8, 420 mM NaCl, 1 mM

EDTA), and the cells were lysed by two cycles of freezing inliquid nitrogen and thawing on ice. Insoluble material waspelleted by centrifugation, and 10 ml of the soluble extractwere separated by SDS-PAGE in a 9% gel and electroblottedonto a polyvinylidene fluoride membrane. The blot wasprobed with a polyclonal antiserum against the N terminus ofthe hAR, and the proteins were visualized with the enhancedchemiluminescence system (Amersham).

Acknowledgments

We thank Dr. A. O. Brinkmann (Erasmus University, Rotter-dam, The Netherlands), Dr. B. W. O’Malley (Baylor University,Houston, TX), Dr. M. G. Parker (Imperial Cancer ResearchFund, London, UK) and Dr. H. Gronemeyer (Institut de Ge-netique et de Biologie Moleculaire et Cellulaire, Illkirch-Ce-dex, France) for the kind gift of plasmids; Dr. Neri (ScheringPlough, Kenilworth, NJ) for the gift of hydroxyflutamide; R.Bollen and H. De Bruyn for excellent technical assistance;and V. Feytons for the expert synthesis of many oligo-nucleotides.

Received June 4, 1998. Revision received August 31,1998. Accepted September 21, 1998.

Address requests for reprints to: Dr. B. Peeters, Depart-ment of Biochemistry, Campus Gasthuisberg, Herestraat49, B-3000 Leuven, Belgium. e-mail: ben.peeters@med.kuleuven.ac.be.

This work was supported in part by a grant ‘Geconcer-teerde Onderzoeksactie van de Vlaamse Gemeenschap,’ bygrants from the Belgian ‘Fonds voor Geneeskundig Weten-schappelijk Onderzoek’ and by a grant ‘Interuniversity Polesof Attraction Programme, Belgian State, Prime Minister’s Of-fice, Federal Office for Scientific, Technical and Cultural Af-fairs.’ P.A. was holder of a scholarship ‘Vlaams Instituut voorde Bevordering van het Wetenschappelijk-TechnologischOnderzoek in de Industrie.’ F.C. and J.V.S. are Senior Re-search Assistants of the Fund for Scientific ResearchFlanders (Belgium).

REFERENCES

1. Evans R 1988 The steroid and thyroid hormone receptorsuperfamily. Science 240:889–895

2. Beato M 1989 Gene regulation by steroid hormones. Cell56:335–344

3. Mangelsdorf DJ, Thummel C, Beato M, Herrlich P,Schutz G, Umesono K, Blumberg B, Kastner P, Mark M,Chambon P, Evans RM 1995 The nuclear receptor su-perfamily: the second decade. Cell 83:835–839

4. Chambon P 1996 A decade of molecular biology of reti-noic acid receptors. FASEB J 10:940–954

5. Jenster G, van der Korput HAGM, Trapman J, Brink-mann AO 1995 Identification of two transcription activa-tion units in the N-terminal domain of the human andro-

gen receptor. J Biol Chem 270:7341–73466. Wilkinson JR, Towle HC 1997 Identification and charac-

terization of the AF-1 transactivation domain of thyroidhormone receptor b1. J Biol Chem 272:23824–23832

7. Danielian PS, White R, Lees JA, Parker MG 1992 Identi-fication of a conserved region required for hormone de-pendent transcriptional activation by steroid hormonereceptors. EMBO J 11:1025–1033

8. Bourguet W, Ruff M, Chambon P, Gronemeyer H, MorasD 1995 Crystal structure of the ligand-binding domain ofthe human nuclear receptor RXR-a. Nature 375:377–382

9. Renaud JP, Rochel N, Ruff M, Vivat V, Chambon P,Gronemeyer H, Moras D 1995 Crystal structure of theRAR-g ligand-binding domain bound to all-trans retinoicacid. Nature 378:681–689

10. Wagner RL, Apriletti JW, McGrath ME, West BL, BaxterJD, Fletterick RJ 1995 A structural role for hormone in thethyroid hormone receptor. Nature 378:690–697

11. Williams SP, Sigler PB 1998 Atomic structure of proges-terone complexed with its receptor. Nature 393:392–396

12. Brzozowski AM, Pike ACW, Dauter Z, Hubbard RE, BonnT, Engstrom O, Ohman L, Greene GL, Gustafsson J-A,Carlquist M 1997 Molecular basis of agonism and antag-onism in the oestrogen receptor. Nature 389:753–758

13. Parker MG, White R 1996 Nuclear receptors spring intoaction. Nat Struct Biol 3:113–115

14. Ing NH, Beekman JM, Tsai SY, Tsai M-J, O’Malley BW1992 Members of the steroid hormone superfamily inter-act with TFIIB. J Biol Chem 267:17617–17623

15. Schulman IG, Chakravarti D, Juguilon H, Romo A, EvansRM 1995 Interactions between Retinoid 3 receptor anda conserved region of the TATA-binding protein mediatehormone-dependent transactivation. Proc Natl Acad SciUSA 92:8288–8292

16. Beato M, Sanchez-Pacheco A 1996 Interaction of steroidhormone receptors with the transcription initiation com-plex. Endocr Rev 17:587–609

17. Fondell JD, Brunel F, Hisatake K, Roeder RG 1996 Un-liganded thyroid hormone receptor a can target TATA-binding protein for transcriptional repression. Mol CellBiol 16:281–287

18. Glass CK, Rose DW, Rosenfeld MG 1997 Nuclear recep-tor coactivators. Curr Opin Cell Biol 9:222–232

19. Shibata H, Spencer TE, Onate SA, Jenster G, Tsai S, TsaiM-J, O’Malley BW 1997 Role of co-activators and co-repressors in the mechanism of steroid/thyroid receptoraction. Recent Prog Horm Res 52:141–165

20. Cavailles V, Dauvois S, Danielian PS, Parker MG 1994Interaction of proteins with transcriptionally active estro-gen receptors. Proc Natl Acad Sci USA 91:10009–10013

21. Cavailles V, Dauvois S, L’Horset F, Lopez G, Hoare S,Kushner PJ, Parker MG 1995 Nuclear factor RIP140modulates transcriptional activation by the estrogen re-ceptor. EMBO J 14:3741–3751

22. Le Douarin B, Zechel C, Garnier JM, Lutz Y, Tora L,Pierrat B, Heery D, Gronemeyer H, Chambon P, LossonR 1995 The N-terminal part of TIF1, a putative mediatorof the ligand-dependent activation function (AF-2) of nu-clear receptors, is fused to B-raf in the oncogenic proteinT18. EMBO J 14:2020–2033

23. Lee JW, Ryan F, Swaffield JC, Johnston SA, Moore DD1995 Interaction of thyroid-hormone receptor with a con-served transcriptional mediator. Nature 374:91–94

24. vom Baur E, Zechel C, Heery D, Heine MSJ, Garnier JM,Vivat V, Le Douarin B, Gronemeyer H, Chambon P 1996Differential ligand-dependent interactions between theAF-2 activating domain of nuclear receptors and theputative transcriptional intermediary factors mSUG1 andTIF1. EMBO J 15:110–124

25. Onate SA, Tsai SY, Tsai M-J, O’Malley BW 1995 Se-quence and characterization of a coactivator for thesteroid hormone receptor superfamily. Science 270:1354–1357

ELE1/ARA70 Interacts with Steroid Receptors 127

26. Takeshita A, Yen PM, Misiti S, Cardona GR, Liu Y, ChinWW 1996 Molecular cloning and properties of a full-length putative thyroid hormone receptor coactivator.Endocrinology 137:3594–3597

27. Kalkhoven E, Valentine JE, Heery DM, Parker MG 1998Isoforms of steroid receptor coactivator 1 differ in theirability to potentiate transcription by the oestrogen recep-tor. EMBO J 17:232–243

28. Hong H, Kohli K, Trivedi A, Johnson DL, Stallcup MR1996 GRIP1, a novel mouse protein that serves as atranscriptional coactivator in yeast for the hormone bind-ing domains of steroid receptors. Proc Natl Acad SciUSA 93:4948–4952

29. Hong H, Kohli K, Garabedian MJ, Stallcup MR 1997Grip1, a transcriptional coactivator for the AF-2 transac-tivation domain of steroid, thyroid, retinoid and vitamin Dreceptors. Mol Cell Biol 17:2735–2744

30. Voegel JJ, Heine MJS, Zechel C, Chambon P, Grone-meyer H 1996 TIF2, a 160 kDa transcriptional mediatorfor the ligand-dependent activation function AF-2 of nu-clear receptors. EMBO J 15:3667–3675

31. Li H, Gomes PJ, Chen JD 1997 RAC3, a steroid/nuclearreceptor-associated coactivator that is related to SRC-1and TIF2. Proc Natl Acad Sci USA 94:8479–8484

32. Anzick SL, Kononen J, Walker RL, Azorsa DO, TannerMM, Guan XY, Sauter G, Kallioniemi OP, Trent JM, Melt-zer PS 1997 AIB1, a steroid receptor coactivator ampli-fied in breast and ovarian cancer. Science 277:965–968

33. Chen H, Lin RJ, Schiltz RL, Chakravarti D, Nash A, NagyL, Privalsky ML, Nakatani Y, Evans RM 1997 Nuclearreceptor coactivator ACTR is a novel histone acetyltrans-ferase and forms a multimeric activation complex withp/CAF and CBP/p300. Cell 90:569–580

34. Takeshita A, Cardona GR, Koibuchi N, Suen C-S, ChinWW 1997 TRAM-1, a novel 160 kDa thyroid hormonereceptor activator molecule, exhibits distinct propertiesfrom steroid receptor coactivator-1. J Biol Chem272:27629–27634

35. Torchia J, Rose DW, Inostroza J, Kmei Y, Westin S, GlassC, Rosenfeld M 1997 The transcriptional coactivatorp/CIP binds CBP and mediates nuclear-receptor func-tion. Nature 382:677–684

36. Le Douarin B, Nielsen AL, Garnier JM, Ichinose H, Jean-mougin F, Losson R, Chambon P 1996 A possible in-volvement of TIF1a and TIF1b in the epigenetic controlof transcription by nuclear receptors. EMBO J 15:6701–6715

37. Heery DM, Kalkhoven E, Hoare S, Parker MG 1997 Asignature motif in transcriptional coactivators mediatesbinding to nuclear receptors. Nature 387:733–736

38. Voegel JJ, Heine MJS, Tini M, Vivat V, Chambon P,Gronemeyer H 1998 The coactivator TIF2 contains threenuclear receptor-binding motifs and mediates transacti-vation through CBP binding-dependent and -indepen-dent pathways. EMBO J 17:507–519

39. Ding XF, Anderson CM, Ma H, Hong H, Uht RM, KushnerPJ, Stallcup MR 1998 Nuclear receptor-binding sites ofcoactivators glucocorticoid receptor interacting protein 1(GRIP1) and steroid receptor coactivator 1 (SRC-1): mul-tiple motifs with different binding specificities. Mol En-docrinol 12:302–313

40. Schmidt S, Baniahmad A, Eggert M, Schneider S, Ren-kawitz R 1998 Multiple receptor interaction domains ofGRIP1 function in synergy. Nucleic Acids Res26:1191–1197

41. Jenster G, Spencer TE, Burcin MM, Tsai SY, Tsai M-J,O’Malley B 1997 Steroid receptor induction of gene tran-scription: a two step model. Proc Natl Acad Sci USA94:7879–7884

42. Yeh S, Chang C 1996 Cloning and characterization of a

specific coactivator, ARA70, for the androgen receptor inhuman prostate cells. Proc Natl Acad Sci USA93:5517–5521

43. Santoro M, Dathan NA, Berlingieri MT, Bongarzone I,Paulin C, Grieco M, Pierotti MA, Vecchio G, Fusco A1994 Molecular characterization of RET/PTC3; a novelrearranged version of the RET proto-oncogene in a hu-man thyroid papillary carcinoma. Oncogene 9:509–516

44. Bongarzone I, Butti MG, Coronelli S, Borrello MG, San-toro M, Mondellini P, Pilotti S, Fusco A, Della Porta G,Pierotti MA 1994 Frequent activation of ret protoonco-gene by fusion with a new activation gene in papillarythyroid carcinomas. Cancer Res 54:2979–2985

45. Minoletti F, Butti MG, Coronelli S, Miozzo M, Sozzi G,Pilotti S, Tunacliffe A, Pierotti MA, Bongarzone I 1994 Thetwo genes generating RET/PTC3 are localized in chro-mosomal band 10q11.2. Genes Chromosom Cancer11:51–57

46. Bongarzone I, Butti MG, Fugazzola L, Pacini F, PincheraA, Vorontsova TV, Demidchik EP, Pierotti MA 1997 Com-parison of the breakpoint regions of ELE1 and RET genesinvolved in the generation of RET/PTC3 oncogene insporadic and in radiation-associated papillary thyroidcarcinomas. Genomics 42:252–259

47. Kuiper GG, de Ruiter PE, Trapman J, Jenster G, Brink-mann AO 1993 In vitro translation of androgen receptorcRNA results in an activated androgen receptor protein.Biochem J 296:161–167

48. Kuil CW, Berrevoets CA, Mulder E 1995 Ligand-inducedconformational alterations of the androgen receptor an-alyzed by limited trypsinization. J Biol Chem 270:27569–27576

49. Wurtz JM, Bourguet W, Renaud JP, Vivat V, Chambon P,Moras D, Gronemeyer H 1996 A canonical structure forthe ligand-binding domain of nuclear receptors. NatStruct Biol 3:87–94

50. Henttu PMA, Kalkhoven E, Parker MG 1997 AF-2 activityand recruitment of steroid receptor coactivator 1 to theestrogen receptor depend on a lysine residue conservedin nuclear receptors. Mol Cell Biol 17:1832–1839

51. Treuter E, Albrektsen T, Johansson L, Leers J, Gustafs-son JA 1998 A regulatory role for RIP140 in nuclearreceptor activation. Mol Endocrinol 12:864–881

52. Feng W, Ribeiro RCJ, Wagner RL, Hguyen H, AprilettiJW, Fletterick RJ, Baxter JD, Kushner PJ, West BL 1998Hormone-dependent coactivator binding to a hydropho-bic cleft on nuclear receptors. Science 280:1747–1749

53. Doesburg P, Kuil CW, Berrevoets CA, Steketee K, FaberPW, Mulder E, Brinkmann AO, Trapman J 1997 Func-tional in vivo interaction between the amino-terminal,transactivating domain and the ligand binding domain ofthe androgen receptor. Biochemistry 36:1052–1064

54. Brinkmann AO, Faber PW, van Rooij HCJ, Kuiper GGJ M,Ris C, Klaassen P, van der Korput JAG M, Voorhorst MM,van Laar JH, Mulder E, Trapman J 1989 The humanandrogen receptor: domain structure, genomic organiza-tion and regulation of expression. J Steroid Biochem34:307–310

55. Swinnen JV, Esquenet M, Heyns W, Rombauts W, Ver-hoeven G 1994 Androgen regulation of the messengerRNA encoding diazepam-binding inhibitor/acyl-CoA-binding protein in the human prostatic adenocarcinomacell line LNCaP. Mol Cell Endocrinol 104:153–162

56. Swinnen JV, Vercaeren I, Esquenet M, Heyns W, Verho-even G 1996 Androgen regulation of the messenger RNAencoding diazepam-binding inhibitor/acyl-CoA-bindingprotein in the rat. Mol Cell Endocrinol 118:65–70

57. Chen C, Okayama H 1987 High-efficiency transformationof mammalian cells by plasmid DNA. Mol Cell Biol7:2745–2752

MOL ENDO · 1999 Vol 13 No. 1128