DNA AND CELL BIOLOGYVolume 19, Number 7, 2000Mary Ann Liebert, Inc.Pp. 389–399

The N-Terminal Domain of Thyroid Hormone Receptor- aIs Required for Its Biological Activities

GUNNAR THUESTAD,1,2 IRENE KRAUS,1,3 JAMES APRILETTI,4 and FAHRI SAATCIOGLU1,5

ABSTRACT

Thyroid hormone (T3) receptors (T3Rs) are ligand-modulated transcription factors that belong to the nuclearreceptor superfamily. Whereas the well-conserved DNA-binding domain and the relatively well-conserved li-gand-binding domain in T3Rs have been characterized in detail, limited information is available on the con-tribution of the variable N terminus to the transcriptional properties of T3Rs. To gain greater insight intothe function of the N terminus, we generated a deletion mutant of T3Ra, T3Ra- D N1, that lacks amino acids7–45 and assessed the effect of this deletion on all known transcriptional activities of T3Ra. Despite the factthat T3Ra- D N1 was expressed and bound T3 with an affinity similar to that of wildtype T3Ra, all of its com-mon transcriptional activities were lost. That is, T3Ra- D N1 did not activate transcription from a positive ornegative T3 response element, and it could not interfere with AP-1 transcriptional activity. Surprisingly, T3Ra-D N1 lost its ability to bind DNA, which can account for its deficiencies as a transcriptional activator. In con-trast, the ability of T3Ra- D N1 to interact with putative coactivators or corepressors was not significantly al-tered from that of wildtype T3Ra. However, overall folding of T3Ra- D N1 was altered, as indicated bydifferential sensitivity to limited protease digestion. These data document that the N terminus of T3Ra, al-beit relatively short and representing a variable and unconserved region when compared with other nuclearreceptors, has a critical role in proper folding of the DNA-binding domain and is required for the biologicalactivities of full-length T3Ra.

389

INTRODUCTION

TH E A CT IO N S O F TH Y RO ID H OR M O N E 3,5,3 9 -triiodo-L -thyro-nine (T3) are mediated through binding to intracellu lar

receptors (T3Rs), which belong to the nuclear receptor su-perfam ily of transcript ion factors, encoded by the c-erbA aand c-erbA b genes (Sap et al., 1986; Weinberger et al., 1986).Nuclear receptors are ligand-act ivated transcript ion factorspossessin g highly conserved DNA-binding domains (DBDs)and moderately conserved ligand-bin ding domains (LBDs),whereas they are divergent in their N-terminal domain (NTD)(Tsai and O’Malley, 1994; Beato et al., 1995; Kastner et al.,1995; Mangelsdor f et al., 1995).

Unliganded T3R a can repress promoters that contain T3REs,and this repression is reversed on T3 binding, resulting in netactivation (Damm et al., 1989; Sap et al., 1989). UnligandedT3R a can also activate transcription from promoters that con-

tain an inverted repeat of a variant half site, and this activationis reversed on T3 binding (Saatcioglu et al., 1993a). Other ex-amples of negative transcriptional regulation by T3Rs have beendescribed, but the response elements mediating these activitiesand the mechanisms that are involved appear to be complex andremain poorly defined (e.g., Carr et al., 1992; Hollenberg et al.,1995; Tagami et al., 1999). The ligand-independen t repressionof basal promoter activity and the ligand-dependent activationfunctions are mediated by sequences within the C-terminal li-gand-binding domain (Saatcioglu et al., 1993b; Barettino et al.,1994; Tone et al., 1994; Baniahm ad et al., 1995), whereas theligand-independe nt activation function appears to be mediatedthrough the NTD (Saatcioglu et al., 1993a).

Another activity of T3Rs is the ability to interfere with othertranscriptional factors, such as AP-1 (for a review, see Saat-cioglu et al., 1994). Conservely, AP-1 complexes, composed ofJun homodimers or Jun/Fos or Jun/ATF2 heterodimers (for a

1Biotechnology Centre of Oslo, Oslo, Norway.Departments of 2Pharmacy and 3Biology and 5Institute for Clinical Medicine, University of Oslo, Oslo, Norway.4Metabolic Research Unit, University of California, San Francisco, California.

review, see Angel and Karin, 1991), interfere with ligand-de-pendent trans-activation by T3Rs (for a review, see Saatciogluet al., 1994). First described for the glucocorticoid receptor(GR), the crosstalk with AP-1 is common to most nuclear re-ceptors (for a review, see Saatcioglu et al., 1994). The inter-ference with AP-1 activity by nuclear receptors is not mediatedby direct binding of nuclear receptors to DNA: several studiesinstead suggest that it may arise from competition for a cofac-tor required for both AP-1 and the nuclear receptor (e.g., Saat-cioglu et al., 1993b; Kamei et al., 1996; Aarnisalo et al., 1998;Frønsdal et al., 1998).

The activity of T3R a is modulated by interactions with otherproteins as well as crosstalk with transcription factors such asAP-1. In vitro, retinoid X receptors (RXRs) form heterodimerswith T3Rs that bind to T3REs more efficiently than do T3Rhomodimers. Consequently, coexpression of T3Rs with RXRsresults in increased ligand-dependent and ligand-independen ttranscriptional activation (for a review, see Mangelsdorf andEvans, 1995). Most recently, proteins that act as putative coac-tivators and putative corepressors have been identified that in-teract with the LBD of T3Rs and other nuclear receptors andhave critical roles in mediating nuclear receptor action (for re-views, see Horwitz et al., 1996; Shibata et al., 1997; Torchiaet al., 1998).

The NTD, which is most divergent among the nuclear re-ceptors, is not as well characterized as the DBD and the LBD.Sequences in nuclear receptors with long NTDs make signifi-cant contributions to transcriptional activation, referred to asactivator function 1 (AF-1), compared with the ligand-depen-dent AF-2, which is in the LBD (for reviews, see Tsai and O’Malley, 1994; Beato et al., 1995; Kastner et al., 1995; Man-gelsdorf et al., 1995). Whereas in some nuclear receptors, theN-terminally deleted forms are active for ligand-dependenttransactivation, in others, N-terminal sequences are required forthis activity (Tsai and O’Malley, 1994; Beato et al., 1995; Kast-ner et al., 1995; Mangelsdorf et al., 1995). Furthermore, cell-type- and promoter-specif ic effects of the NTD on the tran-scriptional activities of nuclear receptors have been documented(e.g., Nagpal et al., 1992; Folkers et al., 1993). Recent studieshave also suggested that functional interactions between theNTD and the LBD may be necessary for transcriptional acti-vation (Langley et al., 1995; McInerney et al., 1996; Ikonen etal., 1997). Mutational studies of the T3R b have documentedthe presence of an AF-1 domain in the NTD (Sjoberg andVennstrom , 1995; Tomura et al., 1995; Hollenberg et al., 1995;Wilkinson and Towle, 1997). Similar studies in T3R a suggestedthat distinct stretches of the NTD are involved in hormone-de-pendent transcriptional activation and specificity of DNA- bind-ing activity (Hadzic et al., 1995; Wong and Privalsky, 1995;Hadzic et al., 1998).

In this study, we assessed the contribution of the NTD ofT3R a to its transcriptional activities. We found that in additionto T3-dependent trans-activation, an intact NTD is required forboth ligand-independen t activation and ligand-dependent tran-scriptional interference. In our search for a mechanism for theloss of these activities, we found that an N-terminally deletedmutant of T3R a (T3R a - D N1) bound to coactivators and core-pressors much as does the wildtype receptor but was impairedin DNA binding. In addition, T3R a - D N1 had an altered sensi-tivity to protease digestion, suggesting changes in receptor con-formation that may alter receptor function. These data under-

score the importance of the NTD in modulating the activity ofother domains of T3R a for all its known biological activities.

MATERIALS AND METHODS

Cell culture, transient transfection, andchloramphenicol acetyltransferase and luciferase assays

HeLa, CV-1, and COS-7 cells were maintained in Dulbecco’sModified Eagle Medium (DMEM) supplemented with 10% fe-tal bovine serum (FBS). The calcium phosphate coprecipitationmethod was used to transfect HeLa and CV-1 cells. For trans-fections with 2XT3RE-LUC and RSV180LUC, 250 ng of re-porter plasmid and 25 and 100 ng of expression vector, re-spectively, and pUC18 to a total of 1 m g of DNA per well in12-well plates was used. Transfection assays with the 2 73Col-CAT reporter contained 670 ng of reporter plasmid, 1 m g ofexpression vectors, and pUC18 to a total of 2 m g of DNA perwell in 6-well plates. After 4 h of incubation with the precipi-tates, cells were treated with 15% glycerol for 2 min, washedonce with phosphate buffered saline (PBS), and then maintainedin DMEM supplemented with 0.5% charcoal-treated FBS in thepresence or absence of T3 (10 2 7 M). After 18 h, cells were har-vested, and chloramphenico l acetyltransferase (CAT) or lu-ciferase (LUC) enzyme activities were determ ined as previouslydescribed (Saatcioglu et al., 1993b).

The COS-7 cells were transfected by a liposome-medi atedtransfection procedure (Lipofectamine ; GIBCO) with 15 m g ofthe indicated expression vector per 10-cm dish. After 5 h, cellswere washed twice with PBS and were maintained in DMEMsupplemented with 10% FBS. After 24 h, the cells were har-vested, and whole-cell extracts were prepared for Westernanalysis.

Plasmids

Reporter plasm ids 2 73Col-CAT (Angel et al., 1987),2XT3RE-LUC (Glass et al., 1989), RSV180-LUC (Maxwell etal., 1989), the expression vector for T3R a (pSG5-c-ErbA a ;Saatcioglu et al., 1993a), GST-GRIP1 (Hong et al., 1996), andGST-CBP (Frønsdal et al., 1998) have been described.

For generation of T3R a - D N1, the BstNI fragment of the Nterminus in pSG5-c-ErbA a was removed, and the vector wasreligated. For the generation of the GST-SMRT (aa1084–1495), pGEX-KG-TRAC2 (Sande and Privalsky, 1996)was cut with XbaI and NotI, filled in with Klenow polym erase,and religated. To generate GST-NCoR (aa 1893–2453), pCEP-NCoR (Horlein et al., 1995) was cut with BglII (partial) andNotI, and the fragment was inserted into the BamHI and NotIsites of pGEX4TI (Promega).

Western analysis

The COS-7 extracts prepared as described above (35 m g)were run on 10% PAGE and transferred to a nitrocellulose fil-ter that was probed with an antiserum raised against either apeptide from the extreme C terminus of T3R a (Saatcioglu etal., 1993a) or a peptide corresponding to the first 12 aminoacids. The enhanced chemilum inescence kit (Amersham) wasused to visualize the T3R a -specific bands.

THUESTAD ET AL.390

Mobility shift analysis

Preparation of whole-cell extracts, recom binant T3R a , andmobility shift analysis were as described previously (Saatciogluet al., 1993b).

T3 binding assay

Assays for T3 binding were done essentially as described(Apriletti et al., 1995). The wild-type and mutant T3R a weretranslated in vitro with the TNT-coupled transcription/ transla-tion system (Promega) and used in the T3 binding assay. TheKd values were calculated with the Prism computer program(GraphPad Software, Inc.).

GST pulldown assay

In vitro interaction of T3R a and T3R a - D N1 with GRIP-1,CBP, and SMRT were examined by the GST pulldown assayas described previously (Hong et al., 1997; Frønsdal et al.,1998; Sande and Privalsky, 1996).

RESULTS

Deletional mutagenesis of T3R a N terminus

To study the role of the T3R a N terminus in various tran-scriptional activities of T3R a , we generated an internal dele-tion mutant that lacks amino acids 7–45, termed T3R a - D N1(Fig. 1A). Western analysis of extracts prepared from COS-7cells transfected with expression vectors encoding either the

wild-type T3R a or T3R a - D N1 using an antiserum raisedagainst a C-terminal peptide (Saatcioglu et al., 1993) showedthat they were expressed as proteins of the expected size (Fig.1B). Even though the level of T3R a - D N1 was lower in this ex-periment, it was comparable to that of wild-type T3R a in im-munoprecipitatio n experiments (data not shown). Repeating theWestern analysis using an antiserum raised against an N-ter-minal peptide (aa 1–12) also resulted in bands of the expectedsize (Fig. 1C). However, the intensity of the T3R a - D N1 bandwas diminished significantly compared with wild-type T3R a ;this result was expected, as T3R a - D N1 has only 7 of the 12residues of the peptide used to raise this polyclonal antiserum .These data indicate that T3R a - D N1 is expressed to an extentsimilar to that of wild-type T3R a and that it contains the ex-pected sequences at its N terminus.

Because the transcriptional activities of T3R a that have beendescribed to date—ligand-dependent activation, interferencewith AP-1 activity, and ligand-independen t activation— aremodulated by hormone binding, we also examined the abilityof T3R a - D N1 to bind T3. Scatchard analysis showed that thedissociation constant of T3R a - D N1 was similar to that of thewild-type receptor (Fig. 1A). Therefore, neither decreased ex-pression nor impaired T3 binding can account for the deficien-cies of T3R a - D N1 activity that are described below.

N Terminus of T3R a is required for its transcriptional activities

To determine how the N-terminal deletion in T3R a - D N1 af-fects its biological activities, we first compared the ability of

N-TERMINAL DOMAIN OF T3Ra AND ITS FUNCTION 391

A

B

C

FIG. 1. Deletional analysis of T3R a N-terminal domain. (A) T3R a and its N-terminal internal deletion mutant T3R a - D N1.The T3 dissociation constants were determined as detailed in Materials and Methods. Standard errors (S.E.) for the binding re-actions are indicated. (B) COS-7 cells were transfected with either an empty expression vector ( 2 ) or expression vectors en-coding the wildtype T3R a (WT) or T3R a - D N1 ( D N1) using the Lipofectamine transfection procedure. After 24 h, whole-cellextracts were prepared for Western analysis with antiserum raised against the C-terminal AF-2 region. Migration positions of themolecular weight standards are indicated to the left. (C) Same experiment except that the antiserum was raised against an N-ter-minal peptide of T3R a . The migration position of a nonspecific band (NS) is indicated; this also served as a control for equalprotein loading.

wild-type T3R a and T3R a - D N1 to stimulate expression of2XT3RE-LUC (Glass et al., 1989) in which two copies of thepalindromic T3RE upstream of the thymidine kinase (tk) pro-moter drive expression of the LUC reporter gene. HeLa cellswere cotransfected with 2XT3RE-LUC and either an empty ex-pression vector or expression vectors encoding the wild-typeT3R a or T3R a - D N1. After transfection, the cells were eitherleft untreated or treated with T3 for 18 h, and LUC activitieswere determined. As shown in Figure 2A, in the presence ofT3, wildtype T3R a activated 2XT3RE-LUC expression ap-proximately 20-fold, whereas T3R a - D N1 was inactive. Theseresults suggest that the N terminus is required for ligand-de-pendent activation by T3R a .

A second function of T3R a is its ability to activate tran-scription in a ligand-independen t manner that is relieved on T3

THUESTAD ET AL.392

FIG. 2. Biological activities of the N-terminal deletion mutant of T3R a . (A) CV-1 cells were cotransfectd with 2XT3RE-LUC(0.5 m g) reporter plasmid and 50 ng of the expression vectors encoding either wild-type T3R a (WT) or T3R a - D N1 ( D N1) or theempty expression vector pSG5 ( 2 ) using the calcium phosphate transfection procedure. Cells were either left untreated or treatedwith T3 (10 2 7 M) as indicated and harvested after 18 h, and LUC activities were determined. The LUC activity in the presenceof transfected wild-type T3R a and T3 is set at 100%. The results represent an experiment done in triplicate. (B) HeLa cells werecotransfected with RSV180-LUC (0.25 m g) reporter plasmid and 100 ng of expression vectors specifying wild-type T3R a (WT)or T3R a - D N1 ( D N1) or the empty expression vector pSG5 ( 2 ). Cells were either left untreated or treated with T3 (10 2 7 M) asindicated and harvested after 18 h, and LUC activities were determined. The LUC activity in the presence of wild-type T3R a andin the absence of T3 is arbitrarily set at 100%. The results represent an experiment done in triplicate. (C) HeLa cells were trans-fected with the 2 73Col-CAT (0.67 m g) reporter and 1.0 m g of expression vectors encoding the wild-type T3R a (WT), the N-ter-minal deletion mutant T3R a - D N1 ( D N1), or the empty expression vector pSG5 ( 2 ) to measure their ability to interfere with AP-1 activity. Cells were either left untreated or treated with T3 (10 2 7 M) as indicated, simultaneously treated with TPA (50 ng/ml),and harvested after 18 h. The CAT activities were determined. Results represent an experiment done in triplicate.

A B

C

binding (e.g., Saatcioglu et al., 1993a). We tested the ability ofT3R a - D N1 to activate the RSV180-LUC reporter, in which 180bp of the RSV long terminal repeat containing a negative T3REdrives expression of the LUC gene (Maxwell et al., 1989). Thewild-type T3R a increased expression of RSV18 0-LUC 25-fold,which was lost in the presence of T3, a result consistent withprevious findings (Fig. 2B) (Saatcioglu et al., 1993a). In con-trast, expression of T3R a - D N1 did not affect RSV180-LUC ex-pression in the presence or absence of T3. These results giveadditional support to our previous findings (Saatcioglu et al.,1993a) that the N-terminal sequences of T3R a are necessaryfor T3-independent activation of RSV180-LUC.

We also tested T3R a - D N1 for its ability to interfere withAP-1 activity. An AP-1-depende nt reporter in which a deletionderivative of the collagenase promoter is fused to the CAT re-

porter gene, 2 73Col-CAT (Angel et al., 1987), was cotrans-fected into HeLa cells with either an empty expression vectoror expression vectors specifying the wild-type T3R a or T3R a -D N1. After transfection, the cells were either left untreated ortreated with T3 for 18 h; to maximize AP-1 activity, all cellswere treated with TPA, and CAT activities were determined.As we have shown previously (Saatcioglu et al., 1993b), wild-type T3R a led to a twofold increase in basal levels in the ab-sence of T3 and a fivefold decrease in 2 73Col-CAT expres-sion in the presence of T3 (Fig. 2C). On the other hand,T3R a - D N1 did not affect -73Col-CAT activity in the absenceor presence of T3. These data suggest that an intact N termi-nus is required for T3R a to interfere with AP-1 activity.

Deletion of T3R a N terminus does not affect in vitrointeractions with coactivators and corepressors

A number of putative coactivator and corepressor moleculeshave been identified that modulate nuclear receptor action (forreviews, see Horwitz et al., 1996; Shibata et al., 1997; Torchiaet al., 1998). We wished to determine whether the loss of thetranscriptional activities of T3R a - D N1 is attributable to its im-paired interaction with these coregulatory molecules. To thatend, we tested the ability of T3R a - D N1 to interact with twoputative coactivators, GRIP1 (Hong et al., 1996) and CBP(Arias et al., 1994; Kwok et al., 1994), as well as two putativecorepressors, SMRT (Chen and Evans, 1995; Sande and Pri-valsky, 1996) and NCoR (Horlein et al., 1995), using the GSTpulldown assay.

The nuclear receptor binding domain of GRIP1 was ex-pressed in Escherichia coli as a GST fusion protein (Hong etal., 1996) and was used in the GST pulldown assay with thecell-free-translated , 35S-labeled wild-type T3R a or T3R a - D N1in the absence or presence of T3. As shown in Figure 3A, GST-GRIP1 bound weakly to both wild-type T3R a and T3R a - D N1in the absence of T3. The basal level of binding in T3R a - D N1was approximately one third of that in T3R a . Binding to GST-GRIP1 was significantly increased for both proteins in the pres-ence of T3, resulting in similar levels of total binding for T3R a -D N1 and wild-type T3R a (when adjusted for input protein).These results suggest that there is no deficiency in the interac-tion of T3R a - D N1 with GRIP1 in vitro .

We then performed a similar experiment with CBP. The Nterminus of CBP, which contains the nuclear receptor-interact-ing domain (Frønsdal et al., 1998), was expressed as a GST fu-sion protein in E. coli and used in the GST pulldown assay asabove. As shown in Figure 3B, there was significant bindingby wildtype T3R a to GST-CBP in the absence of T3. AlthoughT3R a - D N1 also bound to GST-CBP in the absence of T3, thisbinding was approximately fivefold lower than binding of wild-type T3R a . Binding to GST-CBP was significantly increasedfor both wildtype T3R a and T3R a - D N1 in the presence of T3,approximately threefold and sevenfold, respectively. Totalbinding to GST-CBP in the presence of T3 for T3R a - D N1 wasapproximately 50% that of wild-type T3R a . Thus, T3R a - D N1can interact with GST-CBP half as well as does wild-typeT3R a . However, this does not account for the complete loss inthe transcritional properties of T3R a - D N1.

To assess whether deletion of the N terminus of T3R a af-fects interactions with putative corepressors, we again used the

N-TERMINAL DOMAIN OF T3Ra AND ITS FUNCTION 393

GST pulldown assay. The nuclear receptor-binding domain ofSMRT (Chen and Evans, 1995; Sande and Privalsky, 1996) wasexpressed as a GST fusion protein in E. coli and used in theGST pulldown assay with the cell-free-translated , 35S-labeledwild-type T3R a or T3R a - D N1 in the absence or presence ofT3. As shown in Figure 3C, GST-SMRT bound strongly to bothwild-type T3R a and T3R a - D N1 in the absence of T3. How-ever, binding was approximately 40% less for T3R a - D N1 thanT3R a . Binding to GST-SMRT was significantly diminished inthe presence of T3 for the wild-type T3R a and was lost com-pletely for T3R a - D N1. These results indicate that there are mi-nor changes in the ability of T3R a - D N1 to interact with SMRTcompared with wild-type T3R a that are unlikely to account forthe complete loss in its transcriptional activities.

We also tested another corepressor for T3R a , N-CoR (Hor-lein et al., 1995). The nuclear receptor binding domain of N-CoR was expressed as a GST fusion protein in E. coli and usedin the GST pulldown assay with the cell-free-translated, 35S-la-beled wild-type T3R a or T3R a - D N1 in the absence or presenceof T3. As shown in Figure 3D, both wild-type T3R a and T3R a -D N1 bound strongly to GST-NCoR in the absence of T3. How-ever, binding was approximately 30% less for T3R a - D N1 thanfor wild-type T3R a . In the presence of T3, binding to GST-NCoR was lost for both the wild-type T3R a and T3R a - D N1.These results indicate that deletion of the N terminus of T3R adoes not significantly alter its ability to interact with N-CoR.

Deletion of T3R a N terminus impairs DNA binding

The above data indicated that, despite the loss of transcrip-tional activities, T3R a - D N1 retained the ability to interact prop-erly with putative cofactors. We therefore checked another fea-ture of T3R a function: its ability to bind DNA, which isnecessary for at least two of its activities, namely, hormone-dependent and hormone-indepen dent transcriptional activation.To that end, COS-7 cells were transiently transfected with ex-pression vectors specifying the wild-type T3R a or T3R a - D N1,and whole-cell extracts were prepared for the mobility shift as-say with the myosin heavy chain (MHC) T3 response element(T3RE) oligonucleotide as probe. The T3R a expressed in E.coli was used as a control. As shown in Figure 4A, the use ofrecombinant T3R a in the binding reaction resulted in a strongshifted band that represents T3R a homodimers (T-T) (Saat-cioglu et al., 1993a). In contrast, extracts prepared from COS-7 cells transfected with an empty expression vector did not re-sult in any significant binding. However, extracts containingT3R a resulted in weak binding corresponding to the het-erodimers formed with the transfected T3R a and the endoge-nous RXR in COS-7 cells (T-R). Extracts prepared from cellstransfected with T3R a - D N1, on the other hand, did not resultin any significant binding. When the same extracts were incu-bated with a GST fusion construct of RXR a expressed in E.coli, only the COS-7 extracts containing wild-type T3R a re-sulted in significant, slower migrating complexes correspond-ing to T3R a :GST-RXR a heterodimers (T-G). The T3R a - D N1still did not display any significant binding in the presence ofGST-RXR a . Similar results were obtained using a consensuspalindromic T3RE oligonucleotide as probe, and the specificityof the indicated bands has been confirm ed by antibody super-shift and cold probe competition studies (data not shown).

We performed a similar mobility shift experiment using cell-free-translated, 35S-labeled T3R a and T3R a - D N1, which wereexpressed at comparable levels (data not shown). As shown inFigure 4B, unprogrammed lysate gave rise to a nonspecific band(NS), whereas recombinant T3R a expressed in E. coli resultedin a single major band corresponding to the T3R a homodimers(T-T). Lysates containing the wild-type T3R a or T3R a - D N1resulted in no specific binding when incubated with the MHC-T3RE alone. However, when the binding reactions containedGST-RXR a as well, both the recombinant T3R a and the cell-free-translated wild-type T3R a gave rise to significant levelsof slower-m igrating complexes that correspond to T3R a :GST-RXR a heterodimers (T-G). In contrast, reactions containingT3R a - D N1 1 GST-RXR a did not result in any significantbinding. These data, together with those in Figure 4A, indicatethat the N-terminal domain of T3R a is necessary for its abilityto interact with DNA.

N terminal deletion of T3R a alters its sensitivity to proteases

The impaired DNA binding of T3R a - D N1 may provide anexplanation for the loss of DNA-binding-de pendent activitiesof T3R a , but does not explain those activities that do not re-quire DNA binding, such as interference with AP-1 activity (fora review, see Saatcioglu et al., 1994). Because we did not findany differences in the ability of T3R a - D N1 to interact with theputative coactivators that may play a role in mediating theseDNA-binding-independent activities of T3R a , we reasoned thatthere may be differences in the conformation of T3R a - D N1compared with the wild-type T3R a . To assess this possibility,we used the limited proteolytic cleavage assay. Wild-type T3R aand T3R a - D N1 were cell-free-translate d in the presence of 35S-methionine and subjected to partial proteolysis with car-boxypeptidase Y in the presence or absence of T3. As shownin Figure 5, carboxypeptidase Y digested the full-length recep-tor into smaller forms in the absence of hormone, but it waslargely resistant to digestion in the presence of hormone. In con-trast, T3R a - D N1 was strikingly more susceptible to digestionwith carboxypeptidase Y in both the presence and the absenceof T3. For example, at the longest time point and in the pres-ence of T3, whereas 46% of full-length wild-type T3R a re-mained undigested, only 1% of intact T3R a - D N1 was left. Dif-ferences, albeit less striking, were also observed whena 1-chymotrypsin was used as the protease (data not shown).These data suggest that there are significant differences in theaccessibility of proteases to T3R a - D N1 compared with wild-type T3R a , indicating significant differences in their confor-mation.

DISCUSSION

Soon after the discovery of nuclear receptor structure, do-main-swap experiments, as well as the ability of the DBD andLBD to function in isolation, supported the modular structureof nuclear receptors: that is, the distinct activities are carriedout by specified domains that can function independently (Tsaiand O’Malley, 1994; Beato et al., 1995; Kastner et al., 1995;Mangelsdorf et al., 1995). Recent studies have documented that

THUESTAD ET AL.394

FIG. 3. In vitro interactions of T3R a with coactivators andcorepressors is not significantly affected by deletion of its Nterminus. (A) Wild-type T3R a was compared with T3R a - D N1for the ability to interact with the coactivator GRIP1 in the GSTpulldown assay. Both T3R a and T3R a - D N1 were synthesizedin vitro in the presence of 35S-methionine and used with eitherGST as control or GST-GRIP1 in the absence or presence ofT3 (10 2 6 M); 10% of the input is shown to the left of the fig-ure. Bands were quantitated using a PhosphorImag er, and thedata are presented below the lanes as relative values. The gelshown is representative of three independent experiments. (B)Same assay using GST-CBP. (C) Same assay using GST-SMRT. (D) Same assay using GST-NCoR.

A

B

C

D

there may be functional interactions between the different do-mains for some nuclear receptors, particularly between the NTDand the LBD (Langley et al., 1995; McInerney et al., 1996; Iko-nen et al., 1997). In this manuscript, we have shown that oneof these domains is required for the activity of the other do-mains: an N-terminally deleted form of T3R a has lost all of itscommon transcriptional activities. Although T3R a - D N1 re-tained the ability to bind hormone and interact with cofactorsin a manner similar to the wild-type receptor, it surprisingly lostthe ability to bind to DNA. This result contrasts with previousfindings that the T3R a DBD alone can bind DNA efficiently(e.g., Rastinejad et al., 1995) and indicates that the N terminusof T3R a has a distinct role in modulating the activity of theDBD in the context of the full-length receptor.

Previous work on the effect of N-terminal sequences of T3Rson DNA binding has resulted in conflicting findings (Hadzic etal., 1995, 1998; Tomura et al., 1995; Wong and Privalsky, 1995;Judelson and Privalsky, 1996; Wilkinson and Towle, 1997).Whereas two studies using N-terminally truncated T3R a(Hadzic et al., 1995) and T3R b (Wilkinson and Towle, 1997)did not find significant effects on DNA binding, substitutions

in the sequences in the N terminus of T3R a and its oncogenicderivative v-erbA were shown to modulate, but not abrogate,DNA recognition (Wong and Privalsky, 1995; Judelson and Pri-valsky, 1996; Hadzic et al., 1998). The data presented in thispaper are consistent with the latter studies and extend them,showing that the N terminus of T3R a not only modulates, butis required for DNA binding. We do not know the basis for thedifferences between the activities of the previously describeddeletion mutant of T3R a (which lacks aa 1–50; Hadzic et al.,1995) and the T3R a - D N1 that we describe in this manuscript,which lacks aa 7–45. Whereas the internal deletion in T3R a -D N1 completely blocks DNA binding and all three transcrip-tional activities tested, the deletion mutant described earlier cansupport DNA binding and activation of hormone-depend enttranscription, although at a significantly reduced level. The pos-sible effect of this mutation on hormone-indepe ndent transac-tivation or interference with AP-1 activity was not determined.We have used one of the T3REs (T3RE-IR) that was used byHadzic et al. and did not find any significant DNA-binding ac-tivity by T3R a - D N1 in our mobility shift assays under condi-tions where wild-type T3R a was active (data not shown). This

N-TERMINAL DOMAIN OF T3Ra AND ITS FUNCTION 395

A B

FIG. 4. Loss of DNA binding by N-terminally deleted T3R a . (A) COS-7 cells were transfected with 15 m g of expression vec-tor encoding the wild-type T3R a (WT), T3R a - D N1 ( D N1), or the empty expression vector pSG5 using the Lipofectamine trans-fection procedure. 24 h after transfection, whole-cell extracts were prepared. A GST fusion protein of RXR a was expressed inE. coli and purified on glutathione agarose beads. COS-7 extracts were either used directly or were preincubated with GST-RXR aand used in the mobility shift assay with 32P-labeled MHC-T3RE as probe. The migration positions of free probe (F), homod-imeric (T-T), and heterodimeric complexes from endogenous RXRs in COS-7 cells or with GST-RXR a (T-R and T-G, respec-tively) complexes are indicated. (B) Same procedure except that cell-free-translate d T3R a (WT) and T3R a - D N1 ( D N1) wereused in the binding reactions. Unprogramm ed lysate gave rise to a nonspecific band (NS) as indicated.

result suggests that T3R a - D N1 and the deletion construct ofHadzic et al. have differential effects on the folding and activ-ity of the DBD. Additional studies are required to assess thesepossibilities.

The loss of ligand-dependent and ligand-independen t acti-vation by T3R a - D N1 can be explained by the deficiency in itsDNA binding, as both activities require efficient binding toDNA. Although earlier work suggests that the crosstalk betweenAP-1 and nuclear receptors could at least in part be mediatedby the DBD (e.g., Jonat et al., 1990; Yang-Yen et al., 1990;Schule et al., 1990), the bulk of later studies indicates that it ismediated by competition for a common cofactor (e.g., Kameiet al., 1996; Frønsdal et al., 1998; for a review, see Saatciogluet al., 1994). Thus, inability to bind DNA does not explain theloss in the ability of T3R a - D N1 to interfere with AP-1 activ-ity. On the other hand, T3R a - D N1, similar to wild-type T3R a ,can interact with the coactivators GRIP1 and CBP in a hor-mone-regulated manner. These two cofactors or related proteinshave previously been implicated in the crosstalk of nuclear re-ceptors with other transcription factors (Kamei et al., 1996; Aar-nisalo et al., 1998; Frønsdal et al., 1998; Sheppard et al., 1998;Na et al., 1998). However, despite this, T3R a - D N1 cannot in-terfere with AP-1 activity.

There are several possibilities that can account for these find-ings. First, although CBP and GRIP1 are known to be involvedin the crosstalk between nuclear receptors and AP-1, anothercofactor that we have not tested may also be at work in this re-gard for T3R a . It is possible that this other cofactor is no longerable to interact with T3R a - D N1, and therefore, it can no longerinterfere with AP-1 activity. Further work using the other pu-tative cofactors for T3R a and those that becom e available forAP-1 is required to assess this possibility. Second, our in vitrofindings may not accurately reflect the nature of the interac-tions between T3R a - D N1 and the cofactors in vivo . For exam-ple, interactions between the coactivators and T3R a - D N1 may

be much weaker in vivo , perhaps because of the involvementof other factors in such a complex. Further studies are requiredto test this possibility. Third, the removal of most of the N ter-minus may give rise to conformational changes in the rest ofT3R a such that its function is selectively altered. The data wepresent in fact support this possibility, as we found significantdifferences in the protease sensitivities of T3R a - D N1 and wild-type T3R a (see Fig. 5). Fourth, although it is unlikely in viewof previous work, as noted above, it is possible that DNA bind-ing or a specific conformation of the DBD is required for theability of T3R a to interfere with AP-1.

Carboxypeptidas e Y that was used in the partial protease di-gestion experiment sequentially removes amino acids from theC terminus of proteins (for a review , see Jung et al., 1999). Be-cause T3R a - D N1 is much more susceptible to carboxypepti-dase Y than wild-type T3R a , the N terminus of T3R a may in-fluence the conformation of the LBD. This may be achieved bydirect contacts between the NTD and the LBD, as has been sug-gested for some other nuclear receptors (Langley et al., 1995;McInerney et al., 1996; Ikonen et al., 1997). This hypothesismay at first seem unlikely, as the N terminus of T3R a is verysmall compared with the receptors whose N termini have so farbeen shown to interact with the LBD. In support of this hy-pothesis, we did not observe interactions between the NTD andthe LBD of T3R a in a mammalian two-hybrid-assay in pre-liminary experiments (data not shown). Alternatively, the N ter-minus can induce a conformational change in the LBD throughsome other means, perhaps involving other factors. Furtherwork is required to determine the mechanisms that are at thebasis of these observations.

The crystal structures of the well-conserved DBD and therelatively well-conserved LBD of nuclear receptors have pro-vided significant insights into nuclear receptor function (for re-views, see Glass, 1994; Moras and Gronemeyer, 1998). In con-trast, the structure of the N terminus is still not known and is

THUESTAD ET AL.396

FIG. 5. Distinct protease sensitivities of wildtype T3R a and T3R a - D N1. To assess if there are differences in the conformationof wildtype T3R a compared with T3R a - D N1, a limited protease digestion assay was used. Both T3R a and T3R a - D N1 weresynthesized in vitro in the presence of 35S-methionine and either left untreated or treated with T3 (10 2 6 M). They were then sub-jected to proteolysis with carboxypeptidase Y (100 m g/ml) for increasing lengths of time (2, 10, 30, and 60 min). The reactionswere stopped with SDS-PAGE sample buffer and frozen on dry ice. Samples were run on 12% PAGE and visualized with a Phos-phorImager after fluorography and drying. Quantitation of the relative intensities of the full-length bands are presented undereach lane. The gel is representative of an experiment performed two times.

likely to be significantly different among the different nuclearreceptors, as it is very divergent in primary amino acid se-quence. Because of technical limitations in crystallization of acomplete nuclear receptor or of one with two domains, previ-ous structural studies utilized single receptor domains in isola-tion. Our present findings, supported by data from other work(Judelson and Privalsky, 1996; Wong and Privalsky, 1995; Hol-lenberg et al., 1995), suggest that critical differences would beexpected in these domains when the structure of the completereceptor is solved. The crystal structures will conclusively de-termine if there are direct physical interactions between the dif-ferent domains; e.g., between the NTD and the LBD, or if theyare bridged by a cofactor molecule (which could be shown bycocrystallization of the receptor with such molecules).

In examining the in vitro interactio ns between T3R a - D N1and the putative cofactors for T3R a , we have observed thatthe overall qualitativ e interactio ns were similar to those ob-served with wild-type T3R a (see Fig. 3). Thus, in the pres-ence of T3, for both the wild-type T3R a and T3R a - D N1,there was significan tly greater binding to coactivat ors GRIP1and CBP and dissociati on from the corepress ors SMRT andN-CoR. However, the basal levels of binding in the absenceof T3 were consistent ly less for T3R a - D N1 than for the wild-type T3R a . This result suggests that interactio ns of T3R awith cofactors are influenced by its N terminus. For exam-ple, it is possible that a basal level of weak binding of coac-tivators is facilitate d by the NTD and is stabilized when T3is bound to the LBD. Converse ly, binding of corepresso rs toT3R a may be altered by the N-terminal sequences . Furtherstudies with additiona l cofactors are required to assess thefunctiona l significan ce of these difference s. Additional mech-anistic studies will be facilitate d by the future availabili ty ofthe crystal structure of the NTD, NTD plus DBD, and thefull-lengt h T3R a .

ACKNOWLEDGMENTS

We are very grateful to Drs. Richard Eckner, Chris Glass,Richard H. Goodman, Martin Privalsky, Michael Stallcup, andMartin Zenke for generous gifts of plasm ids. This work wassupported by grants from the Norwegian Research Council,University of Oslo, Odd Fellow, and Jahres Foundations to FS.

REFERENCES

AARNISALO, P., PALVIMO, J.J., and JANNE, O.A. (1998). CREB-binding protein in androgen receptor-mediated signaling. Proc NatlAcad Sci USA 95, 2122–2127.

ANGEL, P., IMAGAWA, M., CHIU, R., STEIN, B., IMBRA, R.J.,RAHMSDORF, H.J., JONAT, C., HERRLICH, P., and KARIN, M.(1987). Phorbol ester-inducible genes contain a common cis elementrecognized by a TPA-modulated transacting factor. Cell 49, 729–739.

ANGEL, P., and KARIN, M. (1991). The role of Jun, Fos, and the AP-1 complex in cell proliferation and transformation . Biochim. Bio-phys. Acta 1072, 129–157.

ARIAS, J., ALBERTS, A.S., BRINDLE, P., CLARET, F.X., SMEAL,T., KARIN, M., FERAMISCO, J., and MONTMINY, M. (1994). Ac-tivation of cAMP and mitogen responsive genes relies on a commonnuclear factor. Nature 370, 226–229.

BANIAHMAD, A., LENG, X., BURRIS, T.P., TSAI, Y.T., TSAI, M.J.,

and O’MALLEY, B.W. (1995). The t4 activation domain of the thy-roid hormone receptor is required for release of a putative corepres-sor(s) necessary for transcriptional silencing. Mol. Cell. Biol. 15,76–86.

BARETTINO, D., VIVANCO RUIZ, M.M., and STUNNENBERG, H.(1994). Characterizatio n of the ligand-dependen t transactivation do-main of thyroid hormone receptor. EMBO J. 13, 3039–3049.

BEATO, M., HERRLICH, P., and SCHUTZ, G. (1995). Steroid hor-mone receptors: Many actors in search of a plot. Cell 83, 851–857.

BURRIS, T.P., NAWAZ, Z., TSAI, M.J., and O’MALLEY, B.W.(1995). A nuclear hormone receptor-associated protein that inhibitstransactivation by the thyroid hormone and retinoic acid receptors.Proc. Natl. Acad. Sci. USA 92, 9525–9529.

CHATTERJEE, V.K., LEE, J.K., RENTOUMIS, A., and JAMESON,J.L. (1999). Negative regulation of the thyroid-stimula ting hormonealpha gene by thyroid hormone: Receptor interaction adjacent to theTATA box. Proc. Natl. Acad. Sci. 86, 9114–9118.

CHEN, J.D., and EVANS, R.M. (1995). A transcriptional corepressorthat interacts with nuclear receptors. Nature 377, 454–457.

CLARET, F.-X., ANTAKLY, T., KARIN, M., and SAATCIOGLU, F.(1996). A shift in ligand responsiveness of thyroid hormone recep-tor alpha induced by heterodimeriza tion with retinoid X receptor al-pha. Mol. Cell. Biol. 16, 219–227.

DAMM, K., THOMPSON, C.C., and EVANS, R.M. (1989). Proteinencoded by v-erbA functions as a thyroid-hormo ne receptor antago-nist. Nature 339, 593–597.

DOESBURG, P., KUIL, C.W., BERREVOETS, C.A., STEKETEE, K.,FABER, P.W., MULDER, E., BRINKMANN, A.O., and TRAP-MAN, J. (1997). Functional in vivo interaction between the amino-terminal, transactivation domain and the ligand binding domain ofthe androgen receptor. Biochemistry 36, 1052–1064.

FOLKERS, G.E., VAN DER LEEDE, B.J., and VAN DER SAAG, P.T.(1993). The retinoic acid receptor-beta 2 contains two separate cell-specific transactivation domains, at the N-terminus and in the ligand-binding domain. Mol. Endocrinol. 7, 616–627.

FRØNSDAL, K., ENGEDAL, N., SLAGSVOLD, T., and SAAT-CIOGLU, F. (1998). CREB binding protein is a coactivator for theandrogen receptor and mediates cross-talk with AP-1. J. Biol. Chem.273, 31853– 31859.

GLASS, C.K. (1994). Differential recognition of target genes by nu-clear receptor monomers, dimers, and heterodimers. Endocr. Rev. 15,391–407.

GLASS, C.K., LIPKIN, S.M., DEVARY, O.V., and ROSENFELD,M.G. (1989). Positive and negative regulation of gene transcriptionby a retinoic acid-thyroid hormone receptor heterodimer. Cell 59,697–708.

HADZIC, E., DESAI-YAJNIK, V., HELMER, E., GUO, S., WU, S.,KOUDINOVA, N., CASANOVA, J., RAAKA, B.M., andSAMUELS, H.H. (1995). A 10-amino-acid sequence in the N-ter-minal A/B domain of thyroid hormone receptor alpha is essential fortranscriptional activation and interaction with the general transcrip-tion factor TFIIB. Mol. Cell. Biol. 15, 4507–4517.

HADZIC, E., HABEOS, I., RAAKA, B.M., and SAMUELS, H.H.(1998). A novel multifunctional motif in the amino-termina l A/B do-man of T3R alpha modulates DNA binding and receptor dimeriza-tion. J. Biol. Chem. 273, 10270– 10278.

HOLLENBERG, A.N., MONDEN, T., and WONDISFORD, F.E.(1995). Ligand-independ ent and -dependent functions of thyroid hor-mone receptor isoforms depend upon their distinct amino termini. J.Biol. Chem. 270, 14274– 14280.

HOLLENBERG, A.N., MONDEN, T., FLYNN, T.R., BOERS, M.E.,COHEN, O., and WONDISFORD, F.E. (1995). The human thy-rotropin-releasing hormone gene is regulated by thyroid hormonethrough two distinct classes of negative thyroid hormone responseelements. Mol. Endocrinol. 9, 540–550.

HONG, H., KOHLI, K., TRIVEDI, A., JOHNSON, D.L., and STALL-CUP, M.R. (1996). GRIPI, a novel mouse protein that serves as a

N-TERMINAL DOMAIN OF T3Ra AND ITS FUNCTION 397

transcriptional coactivator in yeast for the hormone binding domainsof steroid receptors. Proc. Natl. Acad. Sci. USA 93, 4948–4952.

HORLEIN, A.J., NAAR, A.M., HEINZEL, T., TORCHIA, J., GLOSS,B., KUROKAWA, R., RYAN, A., KAMEI, Y., SODERSTROM,M., GLASS, C.K., and ROSENFELD, M.G. (1995). Ligand-inde-pendent repression by the thyroid hormone receptor mediated by anuclear receptor co-repressor. Nature 377, 397–404.

HORWITZ, K.B., JACKSON, T.A., BAIN, D.L., RICHER, J.K.,TAKIMOTO, G.S., and TUNG, L. (1996). Nuclear receptor coacti-vators and corepressors. Mol. Endocrinol. 10, 1167–1177.

IKONEN, T., PALVIMO, J.J., and JANNE, O.A. (1997). Interactionbetween the amino- and carboxyl- terminal regions of the rat an-drogen receptor modulates transcriptional activity and is influ-enced by nuclear receptor coactivato rs. J. Biol. Chem. 272, 29821–29828.

JONAT, C., RAHMSDORF, H.J., PARK, K.K., CATO, A.C., GEBEL,S., PONTA, H., and HERRLICH, P. (1990). Antitumor promotionand antiinflammatio n: Down-modulation of AP-1 (Fos/Jun) activityby glucocorticoid hormone. Cell 62, 1189–1204.

JUDELSON, C., and PRIVALSKY, M.L. (1996). DNA recognition bynormal and oncogenic thyroid hormone receptors: Unexpected di-versity in half-site specificity controlled by non-zinc-finger determi-nants. J. Biol. Chem. 271, 10800–10805.

JUNG, G., UENO, H., and HAYASHI, R. (1999). Carboxypeptidas eY: Structural basis for protein sorting and catalytic triad. J. Biochem.(Tokyo) 126, 1–6.

KAMEI, Y., XU, L., HEINZEL, T., TORCHIA, J., KURUKOWA, R.,GLOSS, P., LIN, S.-C., HEYMAN, R.A., ROSE, D.W., GLASS,C.K., and ROSENFELD, M.G. (1996). A CBP integrator complexmediates transcriptional activation and AP-1 inhibition by nuclearreceptors. Cell 85, 403–414.

KASTNER, P., MARK, M., and CHAMBON, P. (1995). Nonsteroidnuclear receptors: What are genetic studies telling us about their rolein real life? Cell 83, 859–869.

KWOK, R.P., LUNDBLAD, J.R., CHRIVIA, J.C., RICHARDS, J.P.,BACHINGER, H.P., BRENNAN, R.G., ROBERTS, S.G., GREEN,M.R., and GOODMAN, R.H. (1994). Nuclear protein CBP is a coac-tivator for the transcription factor CREB. Nature 370, 223–226.

LANGLEY, E., ZHOU, Z.X., and WILSON, E.M. (1995). Evidencefor an anti-parallel orientation of the ligand-activated human andro-gen receptor dimer. J. Biol. Chem. 270, 29983– 29990.

MANGELSDORF, D.J., and EVANS, R.M. (1995). The RXR het-erodimers and orphan receptors. Cell 83, 841–850.

MANGELSDORF, D.J., THUMMEL, C., BEATO, M., HERRLICH,P., SCHUTZ, G., UMESONO, K., BLUMBERG, B., KASTNER, P.,MARK, M., CHAMBON, P., and EVANS, R. (1995). The nuclearreceptor superfamily: The second decade. Cell 83, 835–839.

MAXWELL, I.H., HARRISON, G.S., WOOD, W.M., and MAX-WELL, F. (1989). A DNA cassette containing a trimerized SV40polyadenylation signal which efficiently blocks spurious plasmid ini-tiated transcription. Biotechniques 7, 276–280.

MCINERNEY, E.M., TSAI, M.J., O’MALLEY, B.W., andKATZENELLENBOGEN, B.S. (1996). Analysis of estrogen recep-tor transcriptional enhancement by a nuclear hormone receptor coac-tivator. Proc. Natl. Acad. Sci. USA 93, 10069– 10073.

MORAS, D., and GRONEMEYER, H. (1998). The nuclear receptorligand-binding domain: Structure and function. Curr. Opin. Cell Biol.10, 384–391.

NA, S.Y., LEE, S.K., HAN, S.J., CHOI, H.S., IM, S.Y., and LEE, J.W.(1998). Steroid receptor coactivator-1 interacts with the p50 subunitand coactivates nuclear factor kappaB-mediate d transactivations . J.Biol. Chem. 273, 10831– 10834.

NAGPAL, S., FRIANT, S., NAKSHATRI, H., and CHAMBON, P.(1993). RARs and RXRs: Evidence for two autonomous transacti-vation functions (AF-1 and AF-2) and heterodimeriza tion in vivo .EMBO J. 12, 2349–2360.

RASTINEJAD, F., PERLMANN, T., EVANS, R.M., and SIGLER,P.B. (1995). Structural determinants of nuclear receptor assembly onDNA direct repeats. Nature 375, 203–211.

SAATCIOGLU, F., DENG, T., and KARIN, M. (1993a). A novel ciselement mediating ligand-independen t activation by T3R: Implica-tions for hormonal regulation. Cell 75, 1095–1105.

SAATCIOGLU, F., BARTUNEK, P., DENG, T., ZENKE, M., andKARIN, M. (1993b). A conserved C-terminal sequence that is deletedin v-ErbA is essential for the biological activities of the T3R (thethyroid hormone receptor). Mol. Cell. Biol. 13, 3675–3685.

SAATCIOGLU, F., CLARET, F-X., and KARIN, M. (1994). Negativetranscriptional regulation by nuclear receptors. Semin. Cancer Biol.5, 347–359.

SANDE, S., and PRIVALSKY, M.L. (1996). Identification of TRACs(T3 receptor-associating cofactors), a family of cofactors that asso-ciate with, and modulate the activity of, nuclear hormone receptors.Mol. Endocrinol. 10, 813–825.

SAP, J., MUNOZ, A., DAMM, K., GOLDBERG, Y., GHYSDAEL, J.,LEUTZ, A., BERG, H., and VENNSTROM, B. (1986). The c-erb-A protein is a high affinity receptor for thyroid hormone. Nature 324,635–640.

SAP, J., MUNOZ, A., SCHMITT, J., STUNNENBERG, H., andVENNSTROM, B. (1989). Repression of transcription mediated bya thyroid hormone response element by the v-erb-A oncogene prod-uct. Nature 340, 242–244.

SCHULE, R., RANGARAJAN, P., KLIEWER, S., RANSONE, L.J.,BOLADO, J., YANG, N., VERMA, I.M., and EVANS, R.M. (1990).Functional antagonism between oncoprotein c-Jun and the gluco-corticoid receptor. Cell 62, 1217–1226.

SHEPPARD, K.A., PHELPS, K.M., WILLIAMS, A.J., THANOS, D.,GLASS, C.K., ROSENFELD, M.G., GERRITSEN, M.E., andCOLLINS, T. (1998). Nuclear integration of glucocortic oid recep-tor and nuclear factor-kappaB signaling by CREB-bind ing proteinand steroid receptor coactivator -1. J. Biol. Chem . 273, 29291–29294.

SHIBATA, H., SPENCER, T.E., ONATE, S.A., JENSTER, G., TSAI,S.Y., TSAI, M.J., and O’MALLEY, B.W. (1997). Role of co-acti-vators and corepressors in the mechanism of steroid/ thyroid recep-tor action. Recent Prog. Horm. Res. 52, 141–164.

SJOBERG, M., and VENNSTROM, B. (1995). Ligand-dependen t and-independent transactivation by thyroid hormone receptor beta 2 isdetermined by the structure of the hormone response element. Mol.Cell. Biol. 15, 4718–4726.

TAGAMI, T., PARK, Y., and JAMESON, J.L. (1999). Mechanismsthat mediate negative regulation of the thyroid-stimula ting hormonealpha gene by the thyroid hromone receptor. J. Biol. Chem. 274,22345–22353.

TOMURA, H., LAZAR, J., PHYILLAIER, M., and NIKODEM, V.M.(1995). The N-terminal region (A/B) of rat thyroid hormone recep-tors alpha 1, beta 1, but not beta 2 contains a strong thyroid hor-mone-depende nt transactivation function. Proc. Natl. Acad. Sci. USA92, 5600–5604.

TONE, Y., COLLINGWOOD, T.N., ADAMS, M., and CHATTER-JEE, V.K. (1994). Functional analysis of a transactiva tion dom ainin the thyroid hormone b receptor. J. Biol. Chem . 269, 31157–31161.

TORCHIA, J., GLASS, C., and ROSENFELD, M.G. (1998). Co-acti-vators and co-repressors in the integration of transcriptional re-sponses. Curr. Opin. Cell Biol. 10, 373–383.

TSAI, M.J., and O’MALLEY, B.W. (1994). Molecular mechanisms ofaction of steroid/ thyroid receptor superfamily members. Annu. Rev.Biochem. 63, 451–486.

WAGNER, R.L., APRILETTI, J.W., MCGRATH, M.E., WEAT, B.,BAXTER, J.D., and FLETTERICK, R.J. (1995). A structural rolefor hormone in the thyroid hormone receptor. Nature 378, 690–697.

WEINBERGER, C., THOMPSON, C., ONG, E.S., LEBO, R., GRUOL,

THUESTAD ET AL.398

D.J., and EVANS, R.M. (1986). The c-erbA gene encodes a thyroidhormone receptor. Nature 324, 641–646.

WILKINSON, J.R., and TOWLE, H.C. (1997). Identification and char-acterization of the AF-1 transactivation domain of thyroid hormonereceptor beta1. J. Biol. Chem. 272, 23824– 23832.

WONG, C.W., and PRIVALSKY, M.L. (1995). Role of the N termi-nus in DNA recognition by the v-erb A protein, an oncogenic de-rivative of a thyroid hormone receptor. Mol. Endocrinol . 9, 551–562.

YANG-YEN, H.F., CHAMBARD, J.C., SUN, Y.L., SMEAL, T.,SCHMIDT, T.J., DROUIN, J., and KARIN, M. (1990). Transcrip-tional interference between c-Jun and the glucocorticoid receptor:

Mutual inhibition of DNA binding due to direct protein-protein in-teraction. Cell 62, 1205–1215.

Address reprint requests to:Dr. Fahri Saatcioglu

Biotechnology Centre of OsloUniversity of OsloGaustadalleen 21

0349 Oslo, Norway

E-mail: fahris@ bioslave.uio.no

N-TERMINAL DOMAIN OF T3Ra AND ITS FUNCTION 399