Proc. Nati. Acad. Sci. USAVol. 88, pp. 8587-8591, October 1991Biochemistry

Ligand-binding and heterodimerization activities of a conservedregion in the ligand-binding domain of the thyroidhormone receptor

(thyroid hormone receptor auxiliary protein(s)/heptad repeats/amphipathic a-helix/protein-protein interactions/transactivation)

REMCO A. SPANJAARD*, DOUGLAS S. DARLING, AND WILLIAM W. CHINDivision of Genetics, Department of Medicine, Brigham and Women's Hospital, and Howard Hughes Medical Institute and Harvard Medical School, Boston,MA 02115

Communicated by Donald D. Brown, June 17, 1991

ABSTRACT The ligand-binding domain of the thyroidhormone (3,5,3'-triiodothyronine) receptor (TR) containspoorly characterized subdomains involved with ligand binding,transactivation, and protein-protein interactions. The regionbetween residues 288-331 of rat TRa-1 was analyzed bymodeling and site-directed mutagenesis. Our results suggestthat part of this sequence adopts an amphipathic a-helicalconformation. The integrity of the putative helix is importantfor 3,5,3'-triiodothyronine binding but not necessarily forheterodimerization with nuclear factor(s). Mutants defectivefor both activities were found clustered in a region overlappingthe C-terminal portion of the helix and further downstream.The sequence conservation of this particular region among theentire superfamily suggests a similar role in dimerization inother receptors.

Thyroid hormone (3,5,3'-triiodothyronine; T3) receptors(TR) are members of the nuclear hormone receptor super-family that are homologous, especially in the DNA-bindingregion. More complicated and less well-conserved is the Cterminus, or ligand-binding domain (LBD), where severalfunctional subdomains can be recognized that subserve li-gand binding, ligand-dependent transactivation, and protein-protein interactions (1, 2). At present, the precise nature ofthese subdomains is largely unknown.We have examined a region in the LBD of rat TRa-1

(rTRa-1) located between residues 289-318 and found evi-dence that part of this region adopts an amphipathic a-helicalconformation important for T3 binding and heterodimeriza-tion with nuclear TR auxiliary protein(s) (TRAP) (3-5).Sequences downstream of this putative helix also contributeto these activities.

MATERIALS AND METHODSExpression Vectors for rTRa-1 and Point Mutants. rTRa-1

cDNA (6) was modified by creation of unique HindIII andBamHI sites 3' next to the stop codon (unpublished results).cDNA was digested with EcoRI and BamHI and cloned intothe EcoRI and Bgl II sites of pSG5X, a derivative of pSG5(Stratagene), giving pSGrTRa-1 or clone 1 (wild type). Stan-dard cloning procedures were followed (7). The identity ofeach construct was verified by DNA sequencing.

Site-Directed Mutagenesis. The Xba I-HindIII fragment ofpSGrTRa-1, which contains the LBD, was cloned intoM13mpl8 and used to generate single-stranded DNA inEscherichia coli K-12 strain CJ236 [dut, ung, thi, relA;pCJ105 (Cm9], which served as a template for all oligonu-cleotide-directed mutagenesis (Bio-Rad). Mutations were

confirmed by sequencing and were numbered 2-16. Mutatedfragments were cut out with Xba I-HindIII and used toreplace the wild-type Xba I-HindIII fragment. Subcloning ofmutations was verified by sequencing. Oligonucleotide se-quences can be obtained on request.In Vitro Transcription, Translation, and T3 Binding. Plas-

mids were linearized with HindIII and transcribed with T7RNA polymerase (6). mRNA was translated in rabbit retic-ulocyte lysate (Bethesda Research Laboratories) in the pres-ence of [35S]methionine (Amersham). Incorporation of[35S]methionine into protein was monitored by trichloroace-tic acid precipitation (6). Five microliters of either unpro-grammed or programmed reticulocyte lysate was used forbinding to 0.5 nM 125I-labeled T3 (81.4 Bq/mmol, NEN), andspecific binding was determined as described (6).

Transfections. COS-7 cells were grown in Dulbecco's mod-ified Eagle's medium/l0o fetal calf serum and were trans-fected in 6-cm plates by using the DEAE-dextran method (7).Cells were then grown for 48 hr with or without T3 (100 nM)before harvesting. The same extract was used for luciferase(8) and chloramphenicol acetyltransferase (CAT) activitydeterminations (7). Conversion of [14C]chloramphenicol (2.2GBq/mmol, NEN) was determined by direct scanning ofTLC plates with a Betascope 603 analyzer (Betagen).DNA Binding/Heterodimerization Assay. The avidin-biotin

complex DNA-binding (ABCD) assay used to determine thebinding of in vitro-translated (mutant) receptors to the bio-tinylated rat pituitary glycoprotein hormone a-subunit geneforTR response element (TRE) (nucleotides -74 to -38) andthe preparation of nuclear extract have been described (3, 4).

RESULTSStructural Analysis of the rTRa-1 Region Between Residues

288-331. A schematic representation of rTRa-1 and thelocation and primary sequence of residues 288-331 are givenin Fig. LA. Alignment of the homologous region from otherTRs shows a high degree of protein sequence conservation.Computer calculations [University of Wisconsin GeneticsComputer Group, sequence analysis software package ver-sion 6.2 (16)] predict a potential a-helical region(s) flanked bya major turn at each end. An amphipathic a-helix might existbetween amino acids 292-308, as shown in a helical wheelprojection in Fig. 1B. Leu-292, Ile-299, and Leu-306 wouldappear on the same face of the helix, together with neigh-boring residues Val-295, Leu-302, and Phe-309. Jointly, theywould create an uninterrupted hydrophobic face spanning

Abbreviations: T3, 3,5,3'-triiodothyronine; TR, thyroid hormonereceptor; LBD, ligand-binding domain; TRAP, TR auxiliary pro-tein(s); CAT, chloramphenicol acetyltransferase; rTRa-1, rat TRa-1;ABCD, avidin-biotin complex DNA-binding; TRE, TR responseelement.*To whom reprint requests should be addressed.

8587

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

021

8588 Biochemistry: Spanjaard et al.

rTRa-1

rTRa-1hTRa-1cTRaXTRarTRO-1hTRP-1cTRPXTRO

1 41 121 410

DNA Ligand

T * * * * * * T

KNGGLGLVSQAIFZaLGXSLSAFNLDDTEVALLQAVLLMSTDRSG____________________________________________

-------------D------------------------ S--T--------------D--R--A-------------------S--T--------------D--M---S------------------S--P--------------D--M---S------------------S--P--------------D--M---S------------------S--P--------------D--V---S-S----------------S--P-

C 2 16

V V D M S_t t t t t

KNGGLGVVSDAIFELGKSLSAFNLDDTEVALLQAVLLMSTDRSG

P V E V P V V R7 3 8 4 10 5 6 13

D P P9 14 A R 12

15 11

Hydrophobic

Ser

FIG. 1. (A) Schematic representation of rTRa-1. The DNA- and ligand-binding domains and the location of amino acids (aa) (one-letter code)288-331 in the receptor are shown. Below is the alignment with other TRa and TR,8 sequences. Boldface Ts above the overlined sequence

symbolize turns. Prefixes r, h, c, and X denote rat, human, chicken, and Xenopus, respectively. The sequence in rTRa-1 that might adopt an

amphipathic a-helical conformation is shown in boldface; residues that form the hydrophobic face are indicated by asterisks, and residues inthe hydrophilic face are underlined. rTRa-1, amino acids (aa) 288-331 (9); hTRa-1, aa 287-330 (10); cTRa, aa 290-333 (11); XTRa, gene A andgene B, aa 296-339 (12); rTR,8-1, aa 342-385 (13); hTRf3-1, aa 342-385 (14); cTRf3, aa 250-293 (15); XTRf3, gene A and gene B, aa 218-261 (12).(B) Helical wheel projection of rTRa-1 amino acid sequence 292-309, which might adopt an amphipathic a-helical conformation. ResiduesLeu-292 (N-terminal start point of the helix), Val-295, Ile-299, Leu-302, Leu-306, and Phe-309 form an uninterrupted hydrophobic face spanningfive turns of the helix. Charged residues Asp-297, Glu-301, and Lys-304 constitute the hydrophilic face on the opposite side. (C) Schematicoverview of mutations in rTRa-1 between residues 288-331. Residues in turns are overlined; those in the a-helix are printed in boldface type.Numbers correspond to numbers of respective clones used in text. Double or triple mutations are indicated above the sequence; single mutationsare indicated below. For a list of clones with respective mutations, see Table 1.

five turns of the helix. In contrast, charged residues Asp-297,Glu-301, and Lys-304 are exclusively found on the oppositeside. This kind of structure is thought to be involved inprotein-protein interactions and could constitute part of thetransactivation and/or dimerization interface (17). This hy-pothesis was tested by creating point mutations in rTRa-1that would specifically change a-helical properties of thesequence. Fifteen mutants were constructed (Fig. 1C andTable 1) and assayed for T3 binding, transactivation, DNAbinding, and heterodimerization with TRAP.T3 Binding of rTRa-l and Point Mutants. In vitro-

transcribed mRNA of rTRa-1 and point mutants was trans-lated in vitro with [35S]methionine, and specific binding of125I-labeled T3 was determined. As a control, equal volumesof the same programmed lysate were analyzed by SDS/PAGE. mRNAs were found to be translated with approxi-mately equal efficiency. Densitometric scanning ofautoradio-grams was used to normalize the data obtained by T3 binding(data not shown). The results are given in Fig. 2. Somewhatunexpectedly, this region appears important for hormonebinding. The conservative substitution of valine for Leu-292in clone 3 almost completely destroys T3-binding activity.Moving down the a-helical axis, this effect becomes gradu-ally less severe for the other two amino acids aligned on thehydrophobic surface; clone 4 retains 65% T3-binding activity,and clone 5 is indistinguishable from wild type. In accordancewith this, the independently constructed double mutation inclone 2 shows no appreciable T3 binding. Interestingly,although Leu-306 can be functionally replaced by valine,substitution by a charged residue (arginine) in clone 11, which

would disrupt the hydrophobic face, severely impairs T3binding. Two other mutations were made to test this inter-pretation. The adjacent valine residues in positions 294 and295 were separately mutated into negatively charged aminoacids (clones 8 and 9, respectively). We envisaged twopossibilities: if this region were randomly structured, similarT3-binding activities for both mutants might be expected

150-

10)C

CY)

CU 50

00

1 2 3 4 5 6 7 8 9 101112 13 141516Wt Clone

FIG. 2. Relative T3 binding of receptor mutants compared withwild-type rTRa-1 (clone 1; set at 100%o). T3 binding of clones 2, 7, 10,11, and 13 is undetectable (<5%). Experiments are done in triplicate(SEM <11% of mean values) with in vitro-translated, [35S]methio-nine-labeled receptors and 0.5 nM 1251-labeled T3, and normalized forreceptor quantity (see text). Results represent mean values. Wt,wild-type; other clones are designated in Table 1.

A B

Hydrophilic

Proc. Natl. Acad Sci. USA 88 (1991)

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

021

Proc. Natl. Acad. Sci. USA 88 (1991) 8589

because they are virtually identical. On the other hand, if thisregion were a-helical, a charge (glutamic acid) in the hydro-philic face should have limited effects (clone 8), but a charge(aspartic acid) in the hydrophobic face (clone 9) should havedeleterious consequences. In accordance with our model, thelatter is seen.Another line of evidence for existence of the a-helix was

obtained through mutations that disrupt its continuity. Asexpected, placing proline in the middle of the a-helix in clone10 abolishes T3 binding. The same Leu-302 was changed intoalanine in clone 15 with much less effect. Next, we investi-gated whether proline substitutions could be used to deter-mine the N- and C-terminal boundaries of the helix from afunctional viewpoint because proline substitutions mighthave limited effects in these positions. Asn-289 and Asn-310might be located at the N- and C-terminal borders, respec-tively, and thus form suitable targets for substitution. Inaddition, asparagine is statistically favored at the N-terminalend of helices (18) and thought to angulate the a-helixbetween the basic region and leucine zipper in bZip type oftranscription factors (19). As shown, clone 12 has almostwild-type affinity for T3, suggesting that the helix mightterminate in this position. However, T3 binding in clone 7 isdisrupted, perhaps because of distortion of the turn in whichthe proline residue is presumably placed. Surprisingly, clone14 has lower, but measurable, T3-binding activity. Kinkingthe helix in this proximal position may leave enough C-ter-minal helix intact for T3 binding, whereas disruption of thehydrophobic face, as in clone 9, is more detrimental.

rTR,8-1 has three amino acid differences in the helix,corresponding to rTRa-1 positions 301, 304, and 308. To testthe importance of these variations, a triple mutation wasconstructed (clone 16) that makes rTRa-1 identical to rat TR/3in this region. As shown in Fig. 2, these mutations have nomeasurable effect. Finally, two other constructs were madefurther downstream. Substitution of valine for Leu-311 (clone6) does not change the T3-binding activity. The secondmutation, in clone 13, strongly interferes with T3 binding,showing that the integrity of the whole region is necessary forproper T3 binding.

Transactivation by rTRa-l and Its Derivatives. To examinethe potential role of the region studied here in transactivation,each construct was cotransfected into receptor-deficientCOS-7 cells with a reporter plasmid carrying a T3-induciblepromoter regulating transcription of the bacterial CAT gene(20). CAT activity was measured with and without T3, andfold induction was calculated (Fig. 3). Mutants with >40% T3binding compared with wild type are not significantlychanged, indicating that the substituted residues are notspecifically involved in transactivation. All clones with <5%T3 binding induce significantly less CAT activity than wildtype, but it is impossible to determine whether the measuredeffects are caused by either reduced affinity for T3, animpaired transactivation potential, and/or some other factor.Clones 2, 10, and 13 are completely inactive in both T3binding and transactivation. However, in the absence ofhormone, all constructs can suppress basal transcription ofthe target gene, suggesting that their structure is still largelyintact (data not shown).

Effect of Point Mutations on Heterodimerization withTRAP. Several reports (3-5, 21-23) have shown that TRheterodimerizes with a nuclear factor(s), or TRAP, resultingin greatly increased stability of TR-DNA complexes. Weused the ABCD assay to measure TR-TRAP heterodimerformation of the mutants by using nuclear extract derivedfrom 235-1 cells as a source of TRAP activity. This ratpituitary cell line has no detectable TR activity; the resultsare shown in Fig. 4. In the absence of nuclear extract, 10%oof rTRa-1 is bound to the DNA. As expected, when nuclearextract is added, the number of TR-DNA complexes in-

I-J8<7V O

c-dy b0 5*SI -

C 1 2 3 4 5 6 7 8 9 10111213141516Wt Clone

FIG. 3. Transactivation by rTRa-1 and point mutants. Two andone-half micrograms of receptor DNA was cotransfected into COS-7cells with 5 ,ug of reporter plasmid ptkl4CAT, which contains thebacterial CAT gene under transcriptional control of a TRE (20). Asan internal control for transfection efficiency, 3 j.ug of pRSVluc wasincluded in all experiments, and luciferase activity (Luc) of extractswas used to normalize for CAT activity. CAT activity was measuredwith or without T3 (100 nM), and ratios were determined (foldinduction). Results represent the mean of at least four experimentsdone in duplicate. C, no receptor (pSG5X); Wt, wild type; otherclones are designated in Table 1. *, P < 0.05.

creases S5-fold. Most mutants display basal DNA bindingand TR-TRAP heterodimerization characteristics similar towild type. As has been shown (3), this mechanism is inde-pendent of the ability of the receptor to bind T3. For example,clones 2 and 3 do not bind T3, but the mutations have no effectin this assay. Three mutants (clones 10, 11, and 13) showlower basal DNA binding and are severely impaired inheterodimerization. DNA binding of clone 10, in which thehelix is kinked, is poorly enhanced by nuclear extract,whereas mutation of the same Leu-302 into alanine (clone 15)results in normal levels. Disruption of the helix that isN-terminal of position 302 (clones 7, 9, and 14) has nosignificant effect, but when it is C-terminal (clone 11), basalDNA binding is reduced, and heterodimerization is stronglyimpaired. Clone 5, the control for clone 10, is much lessaffected. Also, substitution for proline at the Asn-310 helixboundary is fairly well tolerated (clone 12). This resultsuggests that within this region the integrity of only theC-terminal portion (residues 302-310) of the helix is impor-tant for TR TRAP heterodimer formation, in contrast tohormone binding, which requires the entire helix. The amino

:s _ I, lI ,

FG4. Heeoierzto of rT 1adpit muatwithX

C' 2 101 1 12 3451

TRAP 3.Lirnearzdcepvtior(murtant) adN wasn mtrantscribed indvtoon-and mirormofecprDNAwasctranslactedinterbieiuoyelste Csysteclswith(5methoninporeridn oflachiprteinATtowhecra coprteinsh

ba-subnit ATR wastestedind theaBscDiptioas iontlt absen (-NE,

aninera coto fo trnfeto efiiny 3 jk ofpS l a

induction. Reut rersn th mea of at leat fou exprimntdoei ulct. C, norcpo..S5) il ye te

FIG.TR-RAHeterodimerization chrTar1andtpoistic smilartwtowidTRAPe.Asneasie beenptown(muan)DNAhwsmecaniscrbe is vinde-penden ofAwstanltditheablt fterecepto rto iculoctFrexIyamplyse,

witherodimerzthionin. DNBinding ofcloheproteintohtecragyortheiwhresubuntTRwastioneof thesABCD assayintheabsie(lnce(-NE

*) or presence (+NE, n) of 40 Ag of nuclear extract (NE) derivedfrom 235-1 cells. Data represent mean values of two experimentalpoints. Wt, wild-type; other clones are designated in Table 1.

Biochemistry: Spanjaard et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

021

8590 Biochemistry: Spanjaard et al.

acid differences between TRa and TR,8 are not significant forTR-TRAP interactions (clone 16). The third mutant that isseriously affected in heterodimerization is clone 13. Whetherthis result is due to a disrupted structure and/or the impair-ment of a specific interaction of residue Leu-318 with TRAPis unknown.

DISCUSSIONAmino acids 288-331 in rTRa-1 are important for T3 bindingand formation ofTR-TRAP heterodimers. We constructed 15different point mutations in rTRa-1 between residues 288 and331 and measured their effects on receptor function withregard to ligand binding, DNA binding, transactivation, andheterodimerization with TRAP (Table 1). Previously, se-quence analysis of the LBD of chicken TRa revealed eight ornine potential a-helical heptad repeats, which were suggestedto constitute the dimerization interface of TR (24, 25). Ourresults are consistent with the existence of an amphipathica-helix within the second and third heptad repeat. However,in addition to its proposed role in dimerization, the integrityof the entire helix strongly correlates with T3-binding activ-ity, and residues outside the helix are also involved. Clearly,the organization of functional subdomains in the LBD of theTR is extremely complicated. Mutations causing a loss inboth hormone binding and TR-TRAP heterodimer formationwere found clustered at the C-terminal portion ofthe putativehelix and further downstream. This configuration is reminis-cent of that in the estrogen receptor, where a major (ho-mo)dimerization domain colocalizes with residues involvedin estrogen binding (26). We speculate that the a-helix formsa structural part of the T3-binding pocket, possibly throughintramolecular peptide-peptide interactions, resembling thecoiled-coil interaction found between leucine zippers (27). Itis particularly interesting that Leu-292 cannot be functionallyreplaced by valine. Computer modeling ofthe C/EBP leucinezipper predicts that valine would block interdigitation, in-stead of locking the coils together (28). Because disruption ofthe turn at the N terminus of the helix has a strong effect onT3 binding (clone 7), loss of a stabilizing interaction near theturn, as in clone 3, may then have the same result. This resultwould also explain why the effect of valine substitutions inclones 4 and 5 becomes gradually less dramatic as the

rTRa-1rTRP- 1hRARahRARPhRARyhVDRhREVhCOUPhGRhPRhMINhARhER

Cons.

LLFFFFFVVI

T

G KG MA NA NA NQ VS E

ES SP QS LS QS S

S LS LQ LQ LQ LG LK LK LE LE FQ FE FR F

SSLLLKNKHVVGR

ASpPPKSARKRwM

F N LF N LL E ML E ML E ML LL LL ML Q VL Q VL Q LL Q IM N L

HT

SSTTQ

D D T ED D A E 'D D T ED D T E

E E [1ED S A E.

Y E EQ E E

P Q EfG

Proc. Natl. Acad. Sci. USA 88 (1991)

Table 1. Summary of phenotypes of rTRa-1 and point mutantsT3 Transacti- Heterodimeri-

Clone Mutation binding vation zation1 Wild-type rTRa-1 ++++ ++++ ++++2 L292,129aTR- +/- +/-++++

V,V3 L292aTR-V + ++4 I29aTR-V +++ ++++ ++++5 LmaTR-V ++++ +++ ++++6 L311aTR-V ++++ ++++ ++7 N289aTR-P +/_ ++++8 V294aTR-E ++++ + ++9 V295aTR-D + + ++++10 L302aTR-P +/- +/_ +/_11 LmaTR-R +/- ++ +/-12 N3lOaTR-p +++ +++ ++13 L318aTR-R +/- +/- +/-14 A2"aTR-P + + + +15 L302aTR-A + + ++16 aTR-(3 ++++ ++++ ++++

Mutations are designated as follows: residue'°aTR-substitution.Clone 16 is a triple mutation E301, K304, A3N aTR-D, M, S, which isidentical to rat TR.8 in the helix. Relative activity compared with wildtype (clone 1) is as follows: + + + +, 75-100%o; + + +, 50-75%; + +,25-50%6; +, 5-25%; +/-, <5%.

distance from the turn increases. At present, it is unclearhow, mechanistically, this region interacts with TRAP, but itmay be a combination of structural component(s) like thea-helix and individual interactions between residues in TRand TRAP.rTRa-1 and rTRf-1 differ by three amino acids in the

helical region, but our data indicate that these variations arenot important. Therefore, an explanation for the defect in onepedigree with generalized resistance to T3 due to a Gly-345(corresponding to Gly-291 in rTRa-1) to arginine substitutionin hTRp3-1 (29) may be that the defect in T3 binding is causedby a disrupted turn (Lys-Asn-Gly-Gly) at the N-terminal endof the helix, or interference with the action of the adjacentleucine residue, as described above. This mutant could stillheterodimerize with TRAP. Unfortunately, the role of TR-TRAP heterodimers in transactivation cannot be evaluatedbecause the mutants that exhibit decreased TR-TRAP for-

VA L L Q A V L L MS TVA L L Q A V L L M S S

G L L S A I C L I CGG L L S A I C L I C G

G L L S A I C L I C Glv L L M A I C I V SPL G L F A V V L V S AY s C L K A I V L F TSY L C M K L L L LSSF L C M KI L L L LTY T I MK1M L L L L STF L C M K A L L L F SIF V C L K I I LL S

X . . . L - - L Q X * * E E T E - X* L C L K A L L LXCS - I P - D R * D GF N D D A M I C

FIG. 5. The region in rTRa-1 involved in heterodimerization with TRAP, aligned with homologous regions in some other (putative) receptors(adapted from ref. 34). Gaps have been introduced for maximum fit. Conserved amino acids are boxed. Critical amino acids for rTRa-1-TRAPheterodimer formation, as described in text, are indicated by vertical arrows. The derived consensus (Cons.) sequence is given below; X*,hydrophobic amino acid. Prefixes r and h denote rat and human, respectively. rTRa-1, amino acids (aa) 302-331 (9); rTR#-1, aa 356-385 (13);hRARa, retinoic acid receptor a, aa 307-336; hRARB, retinoic acid receptor,, aa 305-334; and hRARy, retinoic acid receptor 'y, aa 314-343;hVDR, vitamin D3 receptor, aa 316-345; hREV, Rev-erbAa, aa 523-552 (see ref. 25). hCOUP, human homolog of chicken ovalbumin upstreampromoter transcription factor, aa 305-334; hGR, glucocorticoid receptor, aa 650-679; hPR, progesterone receptor, aa 804-834; hMIN,mineralocorticoid receptor, aa 855-885; hER, estrogen receptor, aa 431-460 (see ref. 34). hAR, androgen receptor, aa 789-819 (35).

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

021

Proc. Natl. Acad. Sci. USA 88 (1991) 8591

mation are also compromised in T3 binding, but clearly themere ability to heterodimerize with TRAP is insufficiept togenerate a transcriptionally active receptor. The qJservationthat the double-defective mutants have relatively lower basalDNA-binding levels may be explained by the presence ofsome TRAP-like activity in reticulocyte lysate, resulting in ahigher base line for receptors that can form heterodimers.Topology and Conservation of Dimerization Subdomains.

The TR can form homodimers (3, 23, 30) and heterodimerswith nuclear hormone receptors (31) and/or with unidentifiednuclear protein(s) (3-5, 21-23) through the LBD. The ques-tion arises how many subdomains are involved and whetherthe same regions/residues are used in all interactions. In theestrogen receptor, two conserved homodimerization do-mains have been identified: (i) one that coincides with thenuclear localization signal (32), and (ii) a stretch of 22 aminoacids located in the C terminps (26, 33) that can be alignedwith the C-terminal heptad repeat in TR (26). With respect toheterodimerization domains, apart from the one reportedhere, there is evidence for the involvement of two otherregions in TR: (i) within the nuclear localization signal, and(ii) one identified in rTR3-1 in the conserved region 286-305(rTRI3-1 amino acid numbering) (3, 5). Thus, with regard tothe primary sequence, four different noncontiguous regionsmay contribute to the homo- and/or heterodimerizationinterface.Comparison of the amino acid sequence of the human

homolog of the chicken ovalbumin upstream promoter tran-scription factor with the sequences of other members of thereceptor superfamily revealed that two regions (II and III) inthe LBp are relatively well-conserved (34). These regionsmap to the heterodimerization domains in the TRidentifiedby us (region III) and others [region II (5)]. Fig. 5 shows thealignment of amino acids' 302-331 of rTRa-1 with the homol-ogous regions of other members of the superfamily; intro-duction of gaps in the alignment permits derivation of anextensive consensus sequence. The alignment clearly showsthat the three amino acids we identified 'as important forTR-TRAP heterodimer formation are highly conserved.Many nuclear hormone receptors have now been found tointeract with different nuclear protein(s) (ref. and thereferences therein). It will be important to determine whichconserved residues play roles and whether the differences insequence contribute to the specificity of binding to differentnuclear protein(s).

This work was supported, in part, by a fellowship from theNetherlands Organization for Scientific Research tq 1Z.A.S.1. Evans, R. M. (1988) Science 240, 889-895.2. Beato, M. (1989) Cell 56, 335-344.3. Darling, D. S., Beebe, J. S., Burnside, J., Winslow, E. R. &

Chin, W. W. (1991) Mol. Endocrinol. 5, 73-84.4. Beebe,' J. S., Darling, D. S. & Chin, W. W. (1991) Mol. En-

docrinol. 5, 85-93.5. O'Donnell, A. L., Rosen, E. D., Darling, D. S. & Koenig,

R. J. (1991) Mol. Endocrinol. 5, 94-99.

6. Lazar, M. A., Hodin, R. A. & Chin, W. W. (1988) Mol. En-docrinol. 2, 893-901.

7. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) MolecularCloning:A Laboratory Manual (Cold Spring Harbor Lab., ColdSpring Harbor, NY), 2nd Ed.

8. Brasier, A. R., Tate, J. E. & Habener, J. F. (1989) BioTech-niques 7, 1116-1122.

9. Thompson, C. C., Weinberger, C., Lebo, R. & Evans, R. M.(1987) Science 237, 1610-1614.

10. Nakai, A., Sakurai, A., Bell, G. I. & DeGroot, L. J. (1988)Mol. Endocrinol. 2, 1087-1092.

11. Sap, J., Mfinoz, A., Damm, K., Goldberg, Y., Ghysdael, J.,Leutz, A., Beug, H. & Vennstrom, B. (1986) Nature (London)324, 635-640.

12. Yaoita, Y., Shi, Y.-B. & Brown, D. D. (1990) Proc. Natl. Acad.Sci. USA 87, 7090-7094.

13. Murray, M. B., Zilz, N. D., McCreary, N. L., MacDonald, M.& Towle, H.; C. (1988) J. Biol. Chem. 263, 12770-12777.

14. Weinberger, C., Thompson, C. C., Ong, E. S., Lebo, R.,Gruol, D. J. & Evans, R. M. (1986) Nature (London) 324,641-646.

15. Forrest, D., Sjoberg, M. & Vennstrom, B. (1990) EMBO J. 9,1519-1528.

16. Devereux, J., Haeberli, P. & Smithies, 0. (1984) Nucleic AcidsRes. 12, 387-395.

17. Ptashne, M. (1988) Nature (London) 335, 683-689.18. Richardson, J. S. & Richardson, D. C. (1988) Science 240,

1648-1652.19. Vinson, C. R., Sigler, P. B. & McKnight, S. L. (1989) Science

246, 911-916.20. Koenig, R. J., Warne, R. L., Brent, G. A., Harney, J. W.,

Larsen, P. R. & Moore, D. D. (1988) Proc. Natl. Acad. Sci.USA 85, 5031-5035.

21. Murray, M. B. & Towle, H. C. (1989) Mol. Endocrinol. 3,1434-1442.

22. Burnside, J., Darling, D. S. & Chin, W. W. (1990) J. Biol.Chem. 265, 2500-2504.

23. Lazar, M. A. & Berrodin, T. J. (1990) Mol. Endocrinol. 4,1627-1635.

24. Forman, B. M., Yang, C.-r., Au, M., Casanova, J., Ghysdael,J. & Samuels, H. H. (1989) Mol. Endocrinol. 3, 1610-1626.

25. Forman, B. M. & Samuels, H. H. (1990) Mo!. Endocrinol. 4,1293-1301.

26. Fawell, S. E., Lees, J. A., White, R. & Parker, M. G. (1990)Cell 60, 953-962.

27. O'Shea, E. K., Rutkowski, R. & Kim, P. S. (1989) Science 243,538-542.

28. Landschultz, W. H., Johnson, P. F. & McKnight, S. L. (1988)Science 240, 1759-1764.

29. Sakurai, A., Takeda, K., Ain, K., Ceccarelli, P., Nakai, A.,Seino, S., Bell, G. I., Refetoff, S. & DeGroot, L. J. (1989)Proc. Natl. Acad. Sci. USA 86, 8977-8981.

30. Holloway, J. M., Glass, C. K., Adler, S., Nelson, C. A. &Rosenfeld, M. G. (1990) Proc. Nat!. Acad. Sci. USA 87,8160-8164.

31. Glass, C. K., Lipkin, S. M., Devary, 0. V. & Rosenfeld,M. G. (1989) Cell 59, 697-708.

32. Kumar, V. & Chambon, P. (1988) Cell 55, 145-156.33. Lees, J. A., Fawell, S. E., White, R. & Parker, M. G. (1990)

Mol. Cell. Biol. 10, 5529-5531.34. Wang, L.-H., Tsai, S. Y., Cook, R. G., Beattie, W. G., Tsai,

M.-J. & O'Malley, B. W. (1989) Nature (London) 340, 163-166.35. Lubahn, D. B., Joseph, D. R., Sullivan, P. M., Willard, H. F.,

French, F. S. & Wilson, E. M. (1988) Science 240, 327-330.

Biochemistry: Spanjaard et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

7, 2

021