THE JOURNAL OF BIOL~XICAL CHEMISTRY 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 268, No. 32, Issue of November 15, pp. 24278-24282, 1993 Printed in U S A .

Distinct Dimerization Domains Provide Antagonist Pathways for Thyroid Hormone Receptor Action*

(Received for publication, June 10, 1993)

Takashi Nagaya and J. Larry Jameson$ From the Thyroid Unit, Massachusetts General Hospital and Haruard Medical School, Boston, Massachusetts 02114

Transcriptional regulation by thyroid hormone is me- diated through its nuclear receptors, which bind to target response elements as homodimers or as het- erodimers with proteins such as retinoid X receptors (RXR). Thyroid hormone response elements exhibit re- markable flexibility in that the receptor binding half- sites can be arranged as direct repeats, inverted repeats, or everted repeats. We report that a limited region at the carboxyl-terminal end of the thyroid hormone receptor differentially contributes to the formation of receptor homo- and heterodimers. The functionally inactive thy- roid hormone receptor splicing variant a2, which is al- tered at the juncture of the homo. and heterodimeriza- tion domains, cannot form homodimers. However, a2 can form a heterodimer when bound to half-sites ar- ranged as a direct repeat spaced by 4 base pairs (Dm), but not with other arrangements of response element half-sites. The cu2-RXFt heterodimer strongly inhibits wild type receptor function mediated by the D M ele- ment, suggesting that the a2 isoform modulates thyroid hormone action by binding as an antagonist to a subset of response elements.

Transcription factors commonly bind to DNA as dimers, pro- viding a potential mechanism for generating regulatory diver- sity. Recently, a subgroup of the nuclear receptor superfamily that includes the thyroid hormone receptors (TR),l retinoic acid receptors (RAR), and vitamin D receptors were found to form homodimers as well as heterodimers with retinoid X receptors ( R X R ) (1-4). Because the DNA binding domains of each of these receptors recognize a similar DNA sequence half-site (AG- GTCA) (5), it is of interest to understand the mechanisms that provide response element specificity. Although the rules that govern selective receptor binding to different DNA target se- quences are still evolving, it is apparent that the spacing be- tween response element half-sites represents a major determi- nant of specificity (6 , 7). For example, when direct repeats of the AGGTCA half-sites are spaced by 3, 4, or 5 base pairs, the response elements are recognized preferentially by vitamin D receptor, TR, and RAR heterodimers with RXR, respectively. The importance of spacing between half-sites suggests that the steric restraints imposed by dimerization may account in part for response element specificity.

* This work was supported in part by National Institutes of Health Grant DK42144. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom reprint requests should be addressed. Present address: Div. of Endocrinology, Metabolism and Molecular Medicine, Tarry 15, Northwestern University Medical School, 303 E. Chicago Ave., Chicago, IL 60611-3008. ”el.: 312-503-0469; Fax: 312-503-0474.

The abbreviations used are: TR, thyroid hormone receptor; RAR, retinoic acid receptor; TRE, thyroid response element; RXR, retinoid X receptor, h, human.

The dimerization domain of the TR involves a broad region of repeated hydrophobic amino acids (heptad repeats) in the car- boxyl terminus that overlaps the ligand binding domain (8). Although multiple domains within the carboxyl terminus of the TR may be involved in protein-protein contacts, deletion stud- ies provide clear evidence that the final, ninth heptad is critical for dimerization and consequently binding to DNA (3, 9-11).

The potential role of this carboxyl-terminal heptad in recep- tor dimerization is of particular interest because this domain is eliminated in a TR splicing variant referred to as a2 (12-15). In a2, the first 370 amino acids are identical to TRal, but the final heptad is disrupted and the remaining 40 residues of a 1 and 120 residues of a2 are distinct. As predicted from receptor mutagenesis studies, a2 does not form effective homodimers and binds poorly to DNA (16). The substitution of a2 sequences in the carboxyl terminus also eliminates thyroid hormone bind- ing (171, resulting in a receptor variant that no longer functions as a hormone-dependent transcription factor (18, 19).

Despite these alterations in the functional properties of a2, there is evidence that it can function as a dominant negative inhibitor of TR action when examined using a thyroid hor- mone response element (TRE) derived from the rat growth hormone promoter (direct repeat spaced by 4 base pairs) (18, 20). However, this dominant negative effect of a2 was not ap- parent with other positive TREs (eg. palindrome) or when a negatively regulated promoter (thyroid-stimulating hormone a ) was examined (19). The observation that a2 could function as a dominant negative inhibitor without the ability to ho- modimerize or bind to DNA led us to investigate the possibil- ity that a2 might retain the ability to form heterodimers with RXR or to bind differentially to various thyroid hormone re- sponse elements.

MATERIALS AND METHODS Plasmid Constructions-Thyroid hormone receptor mutants (Fig.

lA) were constructed by site-directed mutagenesis. Wild type and mu- tant receptor cDNAs were subcloned into pGEM7 (Promega, Madison, WI) for in vitro transcriptiodtranslation and into an Rous sarcoma virus driven vector (21) for transient expression in transfected cells. Thyroid hormone-responsive reporter genes were created by inserting annealed synthetic oligonucleotides corresponding to DR4, DR5, Lap, and Pal (22, 23) into the HindIII-XhoI site of PT109lTUuc (24). The sequences of the thyroid hormone-responsive elements are shown in Fig. lB.

DNA Binding Studies-Gel mobility shift assays were performed to analyze DNA binding and dimerization properties of in vitro translated TR isoforms, mutant receptors, and RARa using radiolabeled synthetic TREs (25). In vitro transcribedtranslated products (4 111) were prepared using the TNT-coupled reticulocyte system (Promega, Madison, WI) and preincubated with or without in vitro translated hFXRa (1 pl) in 20 pl of binding buffer (20 mM Hepes, pH 7.8, 50 mM KCI, 1 mM EDTA, 10% glycerol, 1 mM dithiothreitol, and 50 pg/ml poly(d1,dC) at room tem- perature for 10 min. 32P-Labeled TRE oligonucleotides were added to this reaction and incubated for an additional 20 min. The protein-DNA complexes were analyzed by electrophoresis through a 4% polyacryl- amide gel using 0.5 x TBE (45 mM Tris borate, 1 mM EDTA) buffer containing 2.5% glycerol.

24278

Thyroid Hormone Receptor Action 24279

Dansient Expression Assays-The Luc reporter plasmids (5 pg) and receptor expression plasmids (50 ng) were transfected into JEG-3 cells by the calcium phosphate method as described previously (26). Lucif- erase activity was measured in cell extracts after 24 h of hormone treatment (10 n~ T3 or 100 nM retinoic acid) (27). For studies of domi- nant negative activity, a 5-fold excess (250 ng) of TRa2 or TR mutant plasmids was co-transfected with plasmids (50 ng) expressing wild type TRP or RARa. The total amount of Rous sarcoma virus expression vector was maintained constant by the addition of Rous sarcoma virus CAT plasmid.

RESULTS

Role of the Ninth Heptad Repeat in Homo- and Heterodimer- ization Using Different TREs-The role of distal half of the ninth heptad in receptor dimerization and function was exam- ined using mutant TRPs and the a2 isoform (Fig. U). These mutations were studied in the context of the TRP isoform be- cause of data indicating the importance of dimerization for dominant negative activity of naturally occurring TRj3 mu- tants. TRP-M1 is a single amino substitution that alters one of the hydrophobic leucines that is proposed to exist at the dimer- ization surface (Leu-428 + k g ) . M2 is a deletion (A425-443) that eliminates 19 amino acids beginning at the position anal- ogous to the a2 splice site. Neither of the TRP mutants nor the a2 isoform had measurable T3 binding (data not shown). The DNA binding activity and dimerization properties were studied by gel mobility shift assays using various types of TREs (Fig. lB 1,

Using a direct repeat of the TRE half-site spaced by 4 base

A II * 8

hTRa1 358 € I N I P R " t E 395 hTRaZ 358 ENIPHFWP 395 hTR@ 412 HHVTHIWI-TE 449 hTR$"l 412 BBVTBFWw-m 449 h- 412 HWRWMP-. . . . . . . . . . . . . . . . . . .VECPTE hRARa 338 PsImlmPIQaAQcITD~1"IPGs 375 hR%Ra 409 P E Q P ~ A K L U R U h L R S I ~ I ~ 446

B

"

Pal aqct"

FIG. 1. Mutant thyroid hormone receptor sequences and TREE. A, the amino acid sequences of the carboxyl-terminal regions of the thyroid hormone receptors are aligned. Mutant receptors were con- structed by site-directed mutagenesis, and the leucines that form the borders of the proposed ninth heptad repeat are denoted by asterisks. B , oligonucleotide sequences used for DNA binding studies. Thyroid hor- mone response elements arranged as a direct repeat with t4 (DR4) or +5 spacing (DR5), as inverted palindromic (Lap) sequences, or as pal- indromic (Pal) sequences are shown with the consensus receptor bind- ing half-sites (AGGTCA) indicated with arrows.

AGTC"d

pairs (TRE-DR4), TRal and p bound as a monomer (arrow- head) and as a homodimer (open arrow), although monomer formation was more apparent with TRal (Fig. 2 A ) . Co-incuba- tion with hRXRa formed a slower migrating heterodimer com- plex (solid arrow). In contrast to TRal and P, a2 did not bind to DR4 as a homodimer, suggesting an important role for car- boxyl-terminal receptor sequences in homodimerization. Al- though little or no a2 bound to DR4 when studied alone, it formed a prominent heterodimeric complex when mixed with RXRa and bound to the DR4 response element. The binding properties of TRP-M1 were similar to those of wild type TRP. However, TRP-M2, which deletes a block of amino acids distal to the a2 splice site, formed predominantly heterodimers, simi- lar to the results with a2. These results indicate that distinct amino acid sequences are involved in homo- and heterodimer- ization and that sequences distal to the a2 splice site are re- quired for homodimerization of TRa or TRP isoforms.

When the spacing between direct repeats of the half sites was changed to +5 (TRE-DR51, heterodimerization of the a2 isoform was no longer observed (Fig. 2 B ) . Heterodimerization by the TRP-Ml and "2 was also greatly impaired, even though wild type TRj3 still bound as a heterodimer effectively. This finding suggests that additional sequences in the carboxyl terminus of TRP that are absent in a2 may confer flexibility or stability that allows heterodimerization to half-sites that are spaced further apart. As expected, RARa formed a strong heterodimer with RXRa using both DR4 and DR5 (28, 29).

When the TRE half-sites were arranged in a tail to tail man- ner as an inverted palindrome (TRE-Lap), the binding of TR homodimers was prominent (Fig. 2C). However, the a2 isoform and TRP-M2 did not bind to this element either as a homodimer or a heterodimer. TRP-Ml formed a strong homodimer, but only a small amount of heterodimer with TRE-Lap. Thus, the single amino acid substitution in TRP-M1 (L428R) selectively impairs heterodimerization.

Using a head to head arrangement or palindromic TRE (TRE-Pal), the a2 isoform and TRP-MP bound weakly as mono- mers and heterodimers, without apparent homodimers (Fig. 2 0 ) . TRP-M1 bound to TRE-Pal as a homodimer similar to wild type receptor but formed less heterodimer than the wild type receptor, again indicating that alterations at this site impair heterodimerization. RARa formed weak heterodimer com- plexes with Lap and Pal sequences. The binding data with each of the response elements are summarized in Table I.

TRE-specific Dominant Negative Inhibition by Znactive Het- erodimer-The functional correlation of these DNA binding properties was examined using transient expression assays performed with reporter genes that contain a single copy of DR4, DR5, Lap, or Pal linked to a thymidine kinase promoter- luciferase (TKLuc) vector (Fig. 3, A-D). After treatment with T3, the wild type TRP receptor activated expression by DR4, Lap, and Pal TKLuc by 6.0-, 3.3-, and 5.2-fold, respectively. RARa increased DR5 TKLuc activity 3.0-fold after retinoic acid treatment. As predicted in light of their inability to bind T3, the a2 isoform, TRP-Ml, and TRP-MS did not activate these re- porter genes in response to T3.

Although these mutant receptors were nonfunctional when studied alone, their DNA binding properties predicted that they might function in a dominant negative manner to antago- nize wild type receptor function. Consequently, the mutant and wild type receptors were transfected together and their activi- ties were examined with each of the response elements. The a2 isoform inhibited TRP stimulation of DR4 TKLuc but not activation of DR5 TKLuc. It also weakly inhibited TR activa- tion of Lap and Pal TKLuc. Thus, the dominant negative ac- tivity of a2 correlated well with its ability to form a het- erodimer complex with these TREs. TRp-M1 and TRpM2

24280 Thyroid Hormone Receptor Action

FIG. 2. Interaction of TR isoforms and mutant receptors with different

A, TRE-DR4; B, TRE-DRB; C , TRE-Lap; thyroid hormone response elements.

D, TRE-Pal. The locations of monomer (arrowhead), homodimer (open arrow), and heterodimer (solid arrow) complexes are indicated.

DR4

$on!TRa!TRP,TRw,Ml, ,M2,RAR,a

B DR5

Con!TRa!TRp,TRat,Ml ,,M2,RARy

RXRa "+"+"+"+"+"+"+ RXRa "+"+"+"+-+-+'+

3

. 3

. b r ( J

C , Lap D, Pal , Con!TRy [rRb,TRai ,MI , ,M2 ,RAR,a Con!?Ra!TRl!TRa;,Ml ,,M2,RARy

RXRa -+"+"+"+-+-+-+ RXRo "+"+"+"+"+"+-+

. I & I I

TALILE I Summary of DNA binding properties of receptors

M, monomer; D, dimer; HD, heterodimer.

DR4 DR5 Lap Pal M D HD M D HD M D HD M D HD

interfered the wild type receptor function with DR4 TKLuc but not with DR5, Lap, or Pal TKLuc, again consistent with their ability to bind as heterodimers. The fact that M1 binds well as a homodimer without causing significant inhibition suggests that the heterodimer may be more effective as a dominant negative inhibitor of receptor function (25).

Deletion of the amino terminus and DNA binding domain of a2 (a2DBD-, 82-1201, which eliminates its DNA binding activ- ity (data not shown), relieved the dominant negative activity of the a2 isoform with DR4 TKLuc. This result suggests that the dominant negative activity of the a2-RXR heterodimer is likely by competition with the wild type receptor for binding to target DNA sites (26, 30). Of note, this deletion mutant did not alter the weak inhibition by the a2 isoform that was seen with Lap and Pal TKLuc, suggesting that this effect may involve a dif- ferent mechanism such as titration of essential co-factors (19).

DISCUSSION

There are an increasing number of examples of transcription factors whose activities are modulated by transcriptionally in- active variants. For example, Id blocks the action of Myo D (31) and CREM blocks the action of members of the CREB/ATF family (32). In both of these cases, the inhibitory transcription factor is encoded by a separate gene product. However, alter- nate splicing, as occurs with TRa2, provides an additional mechanism for producing an inhibitory protein.

The principle of dominant negative action is well established for members of the thyroid hormone receptor family. The onco- gene v-erbA, which is a viral homologue of the thyroid hormone receptor that no longer binds T3, is known to block thyroid hormone receptor function by competing for receptor interac- tions with DNA (30). Similarly, an autosomal dominant form of

2428 1

FIG. 3. Ability of a2 and receptor dimerization mutants to inhibit wild type receptor function in transient expression assays. The activity of the a2 isoform and the receptor mutants wa8 examined along with their ability to inter- fere with wild type receptor action in transient expression assays in JEG3 cells. The data are expressed as -fold stimulation by the appropriate ligand (T3 or retinoic acid) and are represented a8 the mean -c S.E. of triplicate transfections from a representative experiment. A, DR4 TKLuc; B, DM TKLuc; C, Lap TKLuc; D, Pal TKLuc.

Thyroid Hormone Receptor Action

DR4 TKLuc A

C Lap TKLuc

B DR5 TKLuc 0 ' J 5 r

. .

a2 M1 M2 a2 + + + +

DBD-

D Pal TKLuc u * I E T

" Con1 p a2 M1 M2 p p p p + + + + a2 M1 M2 a2

DBD-

" Cont p a2 M1 M2 p p p p n2 M1 M2 a2 + + * +

DBD-

thyroid hormone resistance has been attributed to the domi- nant negative activity of mutant thyroid hormone receptors (33, 34). The mutants that cause thyroid hormone resistance retain the ability to dimerize and to bind to DNA, but are transcriptionally inactive, in most cases because of decreased T3 binding (35). Because dimerization is required for dominant negative activity of the thyroid hormone resistance mutants (25), the locations of these naturally occurring mutants help to identify critical receptor domains that are involved in dimer- ization. Of the multiple distinct mutations have been described in the carboxyl terminus of the TR, none have involved the heptad repeats proposed to be required for dimerization (25).

In the current study, we examined the role of the distal half of the ninth heptad repeat in receptor dimerization and func- tion. Although other regions such as the transinhibitory do- mains (9, 36, 37) and heptads 4-6 (38) are also important for receptor dimerizaton, the ninth heptad is of particular interest because it is specifically disrupted in the splicing variant a2 (Fig. lA). Several important conclusions can be derived from the DNA binding characteristics of the mutations introduced into heptad nine. First, a single point mutation (L428R, TRP- M1) altered heterodimerization activity without affecting ho- modimer formation with DR5, Lap, or Pal, providing evidence that domains involved in homo- and heterodimerization are separable. Second, the amino acid sequences distal to leucine 428 in the ninth heptad are required for homodimerization of TRa as well as TRP. For example, the a2 isoform and a deletion mutant (TRP-MB) were unable to form homodimers, even though they formed strong heterodimers with several different response elements. An homologous region in the estrogen re- ceptor is also involved in homodimerization (39). The fact that deletion of the 19 amino acids in TRP-M2 reproduces the bind- ing properties of a2 supports the idea that these TR sequences play an active role in homodimerization rather than the possi- bility that the sequences unique to the a2 carboxyl terminus inhibit the formation of homodimers. Finally, the heterodimer- ization interface appears to vary depending upon the restraints of the response elements. For example, a2 forms a heterodimer with the direct repeat (DR4) but not with TRE-Pal or TRE-Lap.

These results are not surprising, given the different orienta- tions of the response elements, and raise the interesting pos- sibility that the a2-RXR heterodimer may bind selectively to a subset of thyroid hormone response elements to differentially modulate gene expression.

In the transient expression assays, the dominant negative inhibition by the mutant receptors or a2 isoform correlated well with their ability to form heterodimers. The substitution (TRP- M1) or deletion (TRP-MZ) mutants of the ninth heptad each inhibited wild type receptor action selectively with the TRE- DR4 sequence, an element that binds these mutant receptors as heterodimers. The a2 isoform also blocked normal receptor function using a DR4 response element. The fact that deletion of the DNA binding domain of a2 eliminated the dominant negative activity supports the idea that the inhibition occurs as a result of competition for interactions at the DNA target site. In contrast, the relatively weak inhibition by a2 that is seen with Lap and Pal response elements is still seen after deletion of the DNA binding domain. This effect of a2 may reflect titra- tion of co-factors through the dimerization surface or the unique carboxyl terminus of a2 that is retained in the DBD- mutant.

Although the physiologic function of a2 is unknown, the cur- rent study suggests that it can modulate the transcriptional activity of target genes in a selective manner that is dictated by restrictions of the ability of a2 to bind as a heterodimer. Based upon this concept, thyroid hormone-responsive genes that have a DR4 type of response element may be strongly inhibited by the endogenous a2 isoform, whereas other target genes would escape a2 inhibition. This form of regulation may provide a developmental or metabolic switch that is controlled by the relative amounts of a2 and wild type receptor isoforms. Alter- natively, the presence of high levels of a2 in tissues such as brain, testis, and the immune system (40) may allow ambient concentrations of thyroid hormone to affect a limited repertoire of responsive genes. In light of these results, it will be of inter- est to investigate whether a2 causes differential thyroid hor- mone responses of various target genes in uiuo.

24282 Thyroid Hormone Receptor Action Acknowledgments-We are grateful to R. M. Evans for providing TRP 19. Rentournis,A., Chatterjee, V. K. K., Madison, L. D., Datta, S., Gallagher, G. D.,

and RXRa cDNAs, L. J. DeGroot for a1 and a2 cDNAs, and to members DeGroot, L. J., and Jameson, J. L. (1990) Mol. Endocrinol. 4, 1522-1531 of the laboratory for helpful suggestions. 20. Lazar, M. A,, Hodin, R. A,, and Chin, W. W. (1989) Proc. Natl. Acad. Sci.

U. S. A. 86,7771-7714

1

2

3

4.

5.

6.

7.

8. 9.

10.

11.

13. 12.

14.

15.

16. 17.

18.

REFERENCES Mangelsdorf', D. J., Ong, E. S., Dyck, J. A,, and Evans, R. M. (1990) Nature 345, 224-229

Yu, V. C., Delsert, C., Andersen, B., Holloway, J. M., Devary, 0. V., Naar, A. M., Kim, S. Y., Boutin, J. M., Glass, C. K., and Rosenfeld, M. G. (1991) Cell 67,

Leid, M., Kastner, P., Lyons, R., Nakshatri, H., Saunders, M., Zacharewski, T., Chen, J. Y., Staub,A., Gamier, J. M., Mader, S., and Charnbon, P. (1992) Cell

Mangelsdorf, D. J., Borgmeyer, U., Heyman, R. A,, Zhou, J. Y., Ong, E. S., Oro, 88,377-395

Brent, G . A., Moore, D. D., and Larsen, P. R. (1991)Annu. Reu. Physiol. 63, A. E., Kakizuka, A,, and Evans, R. M. (1992) Genes & Deu. 6,329344

Naar, A. M., Boutin, J.-M., Lipkin, S. M., Yu, V. C., Holloway, J. M., Glass, C.

Umesono, K., Murakami, K. K., Thompson, C. C., and Evans, R. M. (1991) Cell

Forman, B. M., and Samuels, H. H. (1990) Mol. Endocrinol. 4, 1293-1301 Marks, M. S., Hallenbeck, P. L., Nagata, T., Segars, J. H., Appella, E., Niko-

Yen, P, M., Sugawara, A,, and Chin, W. W. (1992) J. Biol. Chem. 267,2324&

Zhang, X. K., Hoffmann, B., Tran, P. B., Graupner, G., and Pfahl, M. (1992)

Izumo, S., and Mahdavi, V. (1988) Nature 334,539-542 Benbrook, D., and Pfahl, M. (1987) Science 238, 788-791

Mitsuhashi, T., Tennyson, G. E., and Nikodem, V. M. (1988) Proc. Natl. Acad.

Nakai, A., Seino, S., Sakurai, A,, Szilak, I., Bell, G. I., and DeGroot, L. J. (1988)

Katz, D., Berrodin, T. J . , and Lazar, M. A. (1992) Mol. Endocrinol. 6,805-814 Schueler, P. A., Schwartz, H. L., Strait, K. A., Mariash, C. N., and Oppenhei-

Koenig, R. J., Lazar, M. A,, Hodin, R. A,, Brent, G . A., Larsen, P. R., Chin, W.

1251-1266

17-35

K., and Rosenfeld, M. G. (1991) Cell 66, 1267-1279

66,1255-1266

dem, V. M., and Ozato, K. (1992) EMBO J. 11, 1419-1435

23252

Nature 366,441446

Sci. U. S. A. 85, 5804-6808

Proc. Natl. Acad. Sei. U. S . A. 85,2781-2785

mer, J. H. (1990) Mol. Endocrinol. 4,227-234

W., and Moore, D. D. (1989) Nature 337,65%661

21. Forman, B. M., Yang, C.-R., Stanley, F., Casanova, J., and Samuels, H. H. (1988) Mol. Endocrinol. 2,902-911

22. Baniahmad, A,, Steiner, C., Kohne, A. C., and Renkawitz, R. (1990) Cell 61, 505-514

23. Wahlstmm, G. M., Sjoberg, M., Andemson, M., Nordstrom, K., and Vennstrorn, B. (1992) Mol. Endocrinol. 6, 1013-1022

24. Nordeen, S. K. (1988) Biollkchniques 6,454-457 25. Nagaya, T., and Jameson, J. L. (1993) J. Biol. Chem. 288,1576615771 26. Nagaya, T., Madison, L. D., and Jameson, J. L. (1992) J. Biol. Chern. 267,

27. deWet, J. R., Wood, K V., DeLuca, M., Helinski, D. R., and Subramani, S.

28. Kliewer, S. A., Umesono, K., Nwnan, D. J., Heyman, R. A., and Evans, R. M.

29. Zhang, X. K., Lehmann, J., Hoffmann, B., Dawson, M. I., Cameron, J., Graupner, G., Hermann, T., Tran, P., and Pfahl, M. (1992) Nature 368, 587-591

13014-13019

(1987) Mol. Cell. Biol. 7, 725-737

(1992) Nature 3S8,771-774

30. Damm, K., Thompson, C. C., and Evans, R. M. (1989) Nature 339,593-597 31. Benezra, R., Davis, R. L., Lockshon, D., Turner, D. L., and Weintraub, H.

32. Foulkes, N. S., Borrelli, E., and Sassone-Corsi, P. (1991) Cell 64,739-749 33. Sakurai, A,, Miyamoto, T., Refetoff, S., and DeGroot, L. J. (1990) Mol. Endo.

34. Chatterjee, V. K., Nagaya, T., Madison, L. D., Datta, S., Rentoumis, A., and

35. Parrilla, R., Mixson, A. J., McPherson, J. A., McClaakey, J. H., and Weintraub,

36. ODonnell, A. L., Rosen, E. D., Darling, D. S., and Koenig, R. J. (1991) Mol.

37. Lee, J. W., Gulick, T., and Moore, D. D. (1992) Mol. Endocrinol. 6, 1867-1873 38. Spanjaard, R. A., Darling, D. S., and Chin, W. W. (1991) Proc. Natl. Acad. Sci.

(1990) Cell 61, 49-59

crinol. 4, 198&1994

Jameson, J. L. (1991) J. Clin. Znuest. 87, 1977-1984

B. D. (1991) J. Clin. Inuest. 88,2123-2130

Endocrinol. 6, 94-99

U. S . A. 88,858743591 39. Fawell, S. E., Lees, J. A,, White, R., and Parker, M. G. (1990) Cell 60,953-962 40. Strait, K. A., Schwartz, H. L., Perez, C. A,, and Oppenheimer, J. H. (1990) J.

Biol. Chem. 285, 10514-10521