Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137

Conformational adaptation of nuclear receptor ligand binding domainsto agonists: Potential for novel approaches to ligand design

Marie Togashia, Sabine Borngraeberb, Ben Sandlerb,Robert J. Fletterickb, Paul Webba, John D. Baxtera, ∗

a Department of Medicine and the Diabetes Center, University of California, San Francisco CA 94143, USAb Departments of Biochemistry and Biophysics, University of California, San Francisco CA 94143, USA

Abstract

Ligands occupy the core of nuclear receptor (NR) ligand binding domains (LBDs) and modulate NR function. X-ray structures of NRLBDs reveal most NR agonists fill the enclosed pocket and promote packing of C-terminal helix 12 (H12), whereas the pockets of unligandedNR LBDs differ. Here, we review evidence that NR pockets rearrange to accommodate different agonists. Some thyroid hormone receptor(TR) ligands with 5′ extensions designed to perturb H12 act as antagonists, but many are agonists. One mode of adaptation is seen in aTR/thyroxine complex; the pocket expands to accommodate a 5′ iodine extension. Crystals of other NR LBDs reveal that the pocket cane ealsa nt ofe nhancingt discusss©

K

1

sarmisssaatv

teintings 7%con-d-lter-ntifyother

mul-inster-

ionag-

12rugs

sedstate

0d

xpand or contract and some agonists do not fill the pocket. A TR� structure in complex with an isoform selective drug (GC-24) revnother mode of adaptation; the LBD hydrophobic interior opens to accommodate a bulky 3′ benzyl extension. We suggest that placemextensions on NR agonists will highlight unexpected areas of flexibility within LBDs that could accommodate extensions; thereby ehe selectivity of agonist binding to particular NRs. Finally, agonists that induce similar LBD structures differ in their activities and wetrategies to reveal subtle structural differences responsible for these effects.2005 Elsevier Ltd. All rights reserved.

eywords:Thyroid hormone receptor; Thyroid hormone

. Introduction

The nuclear receptor (NR) family of conditional tran-cription factors comprises one of the largest gene families,nd includes receptors for thyroid hormone (TH), steroids,etinoids, Vitamin D, bile acids, fatty acids, cholesteroletabolites and xenobiotics[1–8]. NRs play widespread and

mportant roles in development, metabolism and homeosta-is and disease. Currently, about 20% of pharmaceutical pre-criptions in the United States are for NR ligands and newelective NR agonists and antagonists with even more desir-ble properties will become available in the future. For ex-mple, tested in rodents and primates, synthetic TH deriva-

ives that bind selectively to the thyroid receptor� (TR�)ersus the TR� isoform and that have selective uptake prop-

∗ Corresponding author. Tel.: +1 415 476 3166; fax: +1 415 564 5813.E-mail address:JBaxter918@aol.com (J.D. Baxter).

erties reduce plasma levels of LDL cholesterol, lipoproLp(a) and triglycerides more effectively than most existreatments and also reduce body weight by as much awithin 1 week (with an even greater decrease in fattent) without obvious ill effects[9–12]. Improved understaning of NR structure and function, and conformational aations that accompany ligand binding, should help us idenew, and equally useful compounds that bind TRs andNRs.

Unlike the case with most classes of pharmaceuticals,tiple X-ray crystal structures of NR ligand binding doma(LBDs) complexed with agonist ligands have been demined [8,13]. These structures provide useful informatabout the nature of the hormone binding pocket, howonists promote folding of the LBD activation helix (H)against the body of the LBD and how to design new dthat bind the pocket and promote or inhibit H12 folding[10].We previously thought that TR agonists must fill the enclopocket and would induce a single active conformational

960-0760/$ – see front matter © 2005 Elsevier Ltd. All rights reserved.

oi:10.1016/j.jsbmb.2005.01.004

128 M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137

(Fig. 1). In the current report, we discuss results with newerTR X-ray structures which suggest that multiple active con-formations exist and that the pocket can adapt to differentagonist ligands in unexpected ways. We discuss the impli-cations of this discovery for understanding receptor functionand pharmaceutical design.

2. Nuclear receptor LBD structures and the ligandbinding pocket

NRs are comprised of single polypeptide chains that canbe separated into three modular domains (Fig. 2) [8]. The C-terminal LBD rearranges in response to ligand binding, withconcomitant influences upon nuclear localization, the choiceof coactivator versus corepressor recruitment, oligomeriza-tion state and activities of the centrally placed DNA bindingdomain (DBD) and N-terminal domain (NTD). Thus, detailedinformation about the structure of unliganded and ligandedNR LBDs could yield significant information about NR func-tion [8,10,13].

2.1. The ligand binding pocket

Our group determined the first X-ray crystal structurefor a liganded NR LBD[14], that of the thyroid hormoner r-m r-r tor-� meu thei con-fi so ing

F unli-g ted that state( es (B)a nfor-m

pocket”, the amino acid residues that directly contact the lig-and.

The fitting of the ligand in the TR hormone binding pocketis rather tight. For example, in the Triac-liganded structure,the volume of the pocket is 600A3 and that of the ligandis 530A3 [14]. Subsequent structures of TR-LBDs in com-plex with other high affinity agonists[16–22], and other NRLBDs in complex with their cognate ligands, also revealedthat agonists usually bind tightly within the enclosed pocket[8,10,13]. Thus, it seemed that agonists must tightly fill arelatively small pocket to induce H12 packing.

In contrast to TR, the peroxisomal proliferator receptor-� (PPAR�) and pregnane X receptor (PXR, Xenobioticreceptor) both contain large ligand binding cavities (of1300 and 1180A3, respectively) within the lower part ofthe LBD, and can accommodate ligands with varied sizes;smaller agonist ligands that act through those receptors of-ten fill only a minor portion of the cavity (developed be-low) [23,24]. Thus, there were hints that NR ligand bindingpockets were more flexible than we had originally antici-pated.

2.2. Ligand binding pocket in the absence of hormone

Presently, X-ray crystal structures are available for onlya few unliganded LBDs. The very first structure of a NRL[ isedp asa -12( ptor( RL nd-b tors,w -p ofR cketm ex-p thel

theP hent sug-g fH -t ptorat nousl hea ce ofl rea ccuroR theo un-l

eceptor-� isoform (TR�) complexed with the thyroid hoone analogue, Triac (Fig. 2). This was published concu

ently with the structure of the liganded retinoic acid recep(RAR �) LBD [15]. Both structures revealed the sa

nexpected finding: the ligand is completely buried innterior of the receptor, a result that has now beenrmed for many NR LBDs[13]. The X-ray crystal structuref liganded LBDs allowed us to define a “ligand bind

ig. 1. The classical and more current views of transition from theanded to the liganded states of NRs. Whereas early models sugges

he active agonist-liganded conformation would correspond to a singleA) we propose that the active NR conformer comprises several statnd that different agonists can preferentially induce particular active coational states or new active conformers (C).

t

BD was that of the unliganded retinoid X-receptor (RXR�)25]. This structure revealed that NR LBDs are comprredominately of�-helices and what, in retrospect, wvery important observation: that the C-terminal helix

H12) was distended away from the body of the recedeveloped below). Of note is that the unliganded RX�BD structure did not reveal the location of the ligainding pocket. Based on structures of liganded recepe now know that two�-helices (H), H11 and H3, iminge into the ligand binding pocket. Thus, admissionXR ligands (such as 9-cis retinoic acid) into the poust involve rearrangements that permit the pocket toand to allow ligand entry and then to collapse around

igand.In contrast to the RXR, the ligand binding pockets of

PAR� and PXR LBDs are large and exist as cavities whe ligand is not present, and the X-ray structure evenests a possible ligand entry route for PPAR� near the base o3 and associated�-sheets[23,24]. An X-ray crystal struc

ure of the liver receptor homolog 1 (LRH-1) orphan recelso reveals a large well formed empty pocket (820A3), al-

hough we do not yet know whether there is an endogeigand for LRH-1[26]. The lack of an obvious pocket in tbsence of ligand in the unliganded RXR, and the presen

arge pockets in PPAR�, PXR and LRH-1 implies that there differences in conformational rearrangements that on hormone binding between, for example, PPAR� versusXR�. The results also show that various NRs differ inrganization of the ligand binding pocket, at least in the

iganded state.

M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137 129

Fig. 2. Linear structure of a NR emphasizing its modular nature with various domains (A) and a three dimensional representation of the TR LBD (B).

2.3. Agonist induced changes in H12 position

Probably the most important rearrangement that occurs onligand binding was revealed with the first liganded LBD struc-tures that could be compared with the unliganded RXR� LBDstructure[14,15,25]. Both groups concluded that a major ef-fect of the ligand is to promote packing of the LBD C-terminalH12 against the body of the LBD. O’Malley and colleagues,using proteolysis and other indirect means, had previouslypredicted that ordering of the receptor’s C-terminus may bea conformational change induced in the LBD in response tothe ligand[27].

We also knew the consequences of the packing of H12 wasto promote release of a corepressor that binds to the receptorin the absence of ligand and cause binding of a coactiva-tor that could promote an increase in transcription (reviewedin [7,10]). We further knew the approximate location of thecoactivator binding surface, but the site where the corepres-sor binds was less clear. Understanding how these surfacesform is critical for understanding how the ligand functions(Fig. 3).

We used our X-ray crystal structures as a guide to placesingle mutations on the surface of the LBD, and examined the

F pack-i core-p g site.

ability of mutant TRs to bind coregulators and activate tran-scription[28–30]. We determined that the coactivator bindingsurface comprises surface exposed residues from H3, H5 andH12, and this surface corresponds to the hormone-dependenttransactivation function (AF-2) within the LBD[28]. Subse-quent crystal structures of TRs[16], and other NRs[23,31],in complex with target peptides from p160 coactivators con-firmed this formulation. We[30], and others[32–36], alsodetermined that the NR corepressor binding surface overlapsthe upper part of AF-2, but extends below the usual positionof H12 in the liganded state. This formulation was confirmedin a structure of PPAR-� in complex with an antagonist anda target peptide from the corepressor SMRT[37]. Thus, H12is displaced in the unliganded state, as seen for RXRs, toexpose the corepressor binding site and agonist binding pro-motes H12 packing against the LBD, occluding the lowerpart of the corepressor binding surface and completing thecoactivator binding site (Fig. 3).

The rearrangements required to convert other unligandedNRs, such as the PPAR�, to the active conformation are prob-ably less extensive than those described above for TR andRAR. Indeed, PPAR� H12 does seem to adopt the agonistposition to some extent in the absence of ligand, as judgedby nuclear magnetic resonance imaging, X-ray crystallogra-phy and direct measurements of mobility of a fluorophoreattached to H12[23,38–40]. However, agonist binding leadst BD[ for-m kingo rsala

3

ep-t BDi f lig-

ig. 3. The coregulator exchange. Agonist binding promotes NR H12ng over the lower part of the corepressor binding surface, leading toressor release and simultaneously completing the coactivator bindin

o significantly increased packing of H12 against the L38–40]. Even though the extent of agonist induced conational rearrangements may differ between NRs, pacf H12 against the body of the LBD seems to be a univend required response to agonist binding.

. The TR ligand binding pocket is flexible

Given that the ligand is completely encircled by the recor in the TR LBD structure, it is apparent that the TR Ls flexible; it must adopt a configuration in the absence o

130 M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137

and that permits ligand entry. However, the tight packing ofthe ligand with little space, for example, for accommodationof bulky side groups suggested that the TR ligand bindingpocket would only be able to accommodate a very limitednumber of ligands. Although that conclusion is correct, ourmore recent data indicate that the pocket harbors much moreflexibility than we originally realized. In the following sec-tions we discuss this unanticipated flexibility, and its possibleimplications for ligand design.

4. The extension hypothesis and antagonist design

The TR LBD structure and realization that hormone de-pendent repositioning of H12 is important for signal trans-duction suggested an obvious way to block NR action; we

proposed in the so-called “extension” hypothesis that ag-onist ligand derivatives with bulky extensions designed todislodge H12 would block coactivator binding[10,41]. In-spection of existing antagonists suggested that they resembleagonists with extension groups[8]. Subsequent solution ofX-ray crystal structures of NRs in complex with antagonistsconfirmed this notion, and it is now known that many ex-isting NR antagonists contain extensions relative to agoniststhat directly perturb H12 in a variety of ways[13].

We previously detailed how we used the extension hy-pothesis to successfully design the first TR antagonists ina previous review[10]. For example, we noticed that the 5′-hydrogen group of Triac fitted tightly against the inner surfaceof TR H12. We took an analogous agonist ligand (MIBRT,Fig. 4) and replaced the 5′ hydrogen with a bulky isopropylgroup[42]. This compound, DIBRT (Fig. 4), is an antagonist.

Fe

Fig. 4. Structures of representative ligands discus

ig. 5. Fitting of T4 into the TR. T3 fits tightly into the ligand binding pocket. T4,xpand the pocket to fit the 5′ iodine group.

sed (triac, T3, T4, MIBRT, DIBRT, GC-1, GC-24).

with a bulky 5′ iodine extension, is accommodated by rearrangements that

M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137 131

We subsequently used the same principle to develop severalother antagonists for the TR[10], including NH-3 [43], aderivative of the TR� selective agonist scaffold GC-1, thatacts as the first TR antagonist that works in animal models[44].

Surprisingly, however, we also found that TR interactingcompounds with bulky 5′ extensions could have a range ofactivities from agonist to partial agonist to antagonist[10].Thus, the nature of the extension and its relationship to the restof the ligand is important for overall agonist/antagonist ac-tivity, as well as its size and position. Further, the fact that theTR ligand binding pocket accommodates these large ligandsindicates that it must rearrange versus previously describedliganded TR LBD structures and still complete the coactiva-tor binding surface. However, the ways that this occurs couldnot be easily predicted from structures of TR ligands without“extensions”.

5. T4, a TR agonist that expands the hormonebinding pocket

How can TRs adapt to the presence of ligands with 5′extensions? Our recent crystal structure of TR� in complexwith major form of thyroid hormone secreted by the thy-roid gland, 3,5,3′5′-tetraiodo-l-thyronine, T or thyroxine,r

to3 lo ytt -c ghc tionoi ldi

r waya andb Rc -i chs intota ationo mot-i da onsi ularT on-fa hea

T -

and binding pocket and H12 was packed against the body ofthe LBD, thereby allowing the coactivator binding surface toexist. The larger T4 ligand was accommodated by expansionof a niche within the pocket that precisely fit the 5′ iodinegroup, created by repositioning of the loop between H11 andH12 and rearrangements of amino acid side chains in thepocket relative to the T3–liganded structure (Fig. 5). Thus, theTR pocket can adapt a unique conformation to bind a bulkysubstituent that should not be expected to fit readily into theligand binding pocket, based on the T3–liganded structure. Itis not yet clear whether TRs use similar mechanisms to adaptto even larger 5′ agonist extensions, or whether the TR canrearrange in other ways.

The ability of TR to accommodate T4 is achieved at acost; H12 is relatively unstable. We observe that the TR�H11-H12 loop and H12 region exhibit higher instability asrevealed by the temperature (B) factors relative to equiva-lent TR�–T3 structures, suggesting that H12 is more mobilein the TR–T4 structure[21]. The TR–T4 complex migratesmore closely to the unliganded TR than typical TR–agonistcomplexes on hydrophobic interaction columns, which pro-vide a crude index of exposed hydrophobic surface on theprotein[21]. Moreover, T4 dissociates from TRs much morerapidly than T3 [21]. Both observations are consistent withthe notion that TR H12 oscillates between the liganded andunliganded configurations in the presence of T, thereby ex-p inga g ofH -e bef e ofh torsa sa

6b

ingp oft eds ffer-e s aren

6

ucee cy-tP ds toh thec ranceo

4eveals one mode (Fig. 5) [21].

T4 is mostly converted in peripheral tissues,5,3′triiodo-l-thyronine (triiodothyronine or T3) by removaf the 5′ iodine group[45]. T4 binds TRs with an affinit

hat is about 0.03–0.1 that of T3 [46]. Although it is usuallyhought that T4 is a prohormone and that T3 activates nulear TRs, free T4 is present in the circulation at high enouoncentrations to conceivably occupy a significant fracf nuclear TRs under certain conditions[45]. We therefore

nvestigated the properties of T4, to understand how it counfluence TR structure and activity.

Our X-ray crystal structure of TR� in complex with T3eveals that it fits into the TR essentially in the sames Triac, in keeping with the notion of a rather fixed liginding pocket[21]. It is therefore difficult to see how Tould accommodate the T45′ iodine group, without dislodgng H12, given that the 5′ iodine group occupies as mupace as the isopropyl substituent that converts MIBRThe agonist DIBRT[42]. We determined that T4 acts as TRgonist and not as an antagonist in vitro, in terms of regulf coactivator binding and corepressor release and in pro

ng dissociation of homodimers from DNA. T4 also behaves a full agonist in transfected cells in vivo, in conditi

n which this activity was not a consequence of intracell4 to T3 conversion. Thus, the TR LBD must assume a c

ormation that differs form that in the T3-LBD structure toccommodate the 5′ iodine extension and yet still adopt tctive LBD conformation.

We solved the X-ray crystal structure of the T4–ligandedR� LBD to 3.1A resolution[21]. T4 was placed in the lig

4osing the hydrophobic interior of the protein and openligand exit route from the pocket. Perhaps poor packin12 could render the TR–T4 complex susceptible to influnces of coregulator levels within the cell; TR H12 could

orced into the unliganded configuration in the presencigh levels of corepressors and/or low levels of coactivand, in these settings, we speculate that T4 could behave apartial agonist, or even as an antagonist.

. Expansion and compression of the NR ligandinding pocket

For TR, we consider that the preferred ligand bindocket is that occupied by T3 and Triac and that expansion

he pocket in the presence of T4 is accompanied by reductability of H12. Nonetheless, other NRs can adapt to dintly sized ligands in other ways, and these adaptationot necessarily accompanied by instability in H12.

.1.1. The PXR pocket: collapse and expansion

PXRs bind harmful chemicals (Xenobiotics) and indxpression of detoxifying enzymes in the liver (such asochrome p450s) that degrade these substances[47,48]. TheXRs also binds a number of pharmaceuticals. This leaarmful drug–drug interactions; one drug that activatesytochrome p450 response can lead to more rapid cleaf other drugs from the body.

132 M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137

Fig. 6. Collapse of the Xenobiotic receptor (PXR) around the body of a ligand. SR12813 (triangle in schematic) fits into the PXR pocket, but binds in multipleconformations. Binding of coactivator peptide from the p160 SRC-1 induces collapse of the ligand binding pocket around the ligand.

Since xenobiotics vary dramatically in size and shape,PXR must adapt to great variety of ligands of different sizesand shapes and yet still attract coactivators and induce tran-scription. The PXR crystal structure was solved in the un-liganded state and in complex with a cholesterol loweringdrug SR12813, both in the presence and absence of a coacti-vator peptide from SRC-1[24,49]. In the absence of SRC-1,the PXR pocket is the same size as that observed in the ab-sence of ligand (Fig. 6). SR12813 occupies only a fraction ofthe pocket and adopts several different conformations. WithSRC-1, the region of the pocket that contacts ligand con-denses, with extensive rearrangements in H6 and H2, trappingthe drug in a single conformation. The structure of PXR wasalso solved in complex with hyperforin, the active ingredientof St. John’s Wort[50]. Here, the pocket expands by 250A3

relative to the unliganded structure and displays increased inorder in H6 and H2, comparable in nature if not in extent tothat observed in the presence of SR12813 and SRC-1. Never-theless, in all structures H12 adopts the agonist conformation.Thus, PXR can tolerate significant expansion and contractionof the pocket, without alterations in H12 position.

6.1.2. Compounds that do not fill the pocket

Another way that NR ligand binding pockets adapt to dif-f f thePo -z ya hterp thel idueo

latedt par-

tial agonists[51]; however, it is clear that TZDs vary in theiractivities in different promoter contexts. It has even been pro-posed that this partial agonist activity may be clinically use-ful; TZDs might enhance PPAR action in useful settings andpartially inhibit PPAR action in harmful settings.

Interestingly, it was apparent as early as the 1970s thatsome antagonists could also be smaller than comparative ag-onist analogues; the androgen receptor (AR) antagonist flu-tamide is one example of this category (reviewed in[10]).These “passive” antagonists do not contain extensions andare thought to occupy the pocket without promoting the con-formational changes necessary to activate the receptor. The

F ,P ocket,b

erent ligands was apparent after the crystallization oPAR� LBD in complex with the anti-diabetic PPAR� ag-nist, rosiglitazone (Fig. 7) [23]. This reveals that rosiglitaone, which binds PPAR� with low affinity, occupies onlbout 40% of the available interior cavity yet induces tigacking of H12 against the LBD by preferentially binding

ower region of the pocket and contacting a tyrosine resn the inner surface of H12.

There is some debate whether rosiglitazone and rehiazolidinediones (TZDs) are true PPAR agonists or

ig. 7. Agonist binding to TRs and PPARs. Whereas, T3 fills the TR pocketPAR ligands such as rosiglitazone occupy a smaller fraction of the put this is sufficient to induce H12 packing into the active state.

M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137 133

best characterized passive antagonist is the estrogen recep-tor (ER) � selective antagonist THC[52]. An ER�–THCstructure confirms that THC occupies the pocket and fails tonucleate formation of a network of contacts with the innersurfaces of H11 that promote packing of H12. Direct mea-surements of association of ER� with coactivator peptidesreveal, however, that THC does permit weak but detectableER�/coactivator association whereas ER� antagonists withextensions, such as raloxifene and tamoxifen, do not. Thus,THC may be better described as an antagonist with weakpartial agonist activity.

Overall, we propose that compounds that do not fit op-timally into NR ligand binding pockets will often prove tobehave as partial agonists/antagonists, but with a spectrumof activities ranging from agonism as seen with the TZDs toalmost complete antagonism as seen with THC.

7. GC-24; extension of the ligand outside of theligand binding pocket in TRs

As developed in the Introduction, and reviewed elsewhere[9–12], TR� selective ligands have great potential medicalutility. We developed an agonist scaffold compound, GC-1,that bound the TR� isoform four to eight-fold more tightlythan to TR� [17,53]. GC-1 forms a network of hydrogenb ingp inT ys ani izec

oned erol-l m-p -s lkye theT iva-t tr sG4 ft parto ngp inderwa acti-v .

7

ithtm[ edi GC-

Fig. 8. Insertion of the side chain of GC-24 between helices 3 and 11 of theTR. Superimposed structures of TR� helix 3, helix 11 and helix 12 com-plexed with GC-24 (dark) and GC-1 (light), respectively. The N-terminus ofhelix 3 and the C-terminus of helix 11 are shifted by outwards 3 and 4A,when GC-24 binds. We suggest that the region that moves may correspondto a possible ligand entry/exit pathway.

1. Moreover, H12 adopts the agonist-position and completesthe coactivator binding surface. The bulky 3′-benzyl groupwas accommodated by a 3–4A shift in the major frameworkhelices 3 and 11 when superimposed with the GC-1 structure(Fig. 8). Even though these helices are key for binding hor-mone and positioning helix 12 for coactivator recognition,the rearrangements involve the lower part of both helicesand do not affect the coactivator binding site. The rearrange-ment of H3 and H11 allows the bulky benzyl group to besandwiched between these helices. The GC-24 benzyl groupis held within a newly formed cluster of six hydrophobicresidues from these two helices (3 and 11) and 12. This re-gion of the TR is not normally accessible to ligand and, thus,GC-24 extends the functional binding pocket to gain extraTR isoform specificity in binding.

7.2. Why is GC-24 TRβ specific?

While GC-24 gains some of its TR� selective bindingfrom contacts with the classical ligand binding pocket, inmuch the same way as GC-1, much of the selectivity comesfrom insertion of the benzyl group into the hydrophobic patchbetween H3 and H11. The residues that contact the GC-24extension are, however, completely conserved between theTR� and TR� isoforms. Thus, we hypothesize that this regiono hei turef rem sf buta cifica

2 29)

onds with amino acids at the top of the ligand bindocket; one TR� residue (Arg 272) that is not conservedR� (where ser227 occupies the equivalent position) pla

mportant, if indirect role, in shaping the pocket to optimontacts with the GC-1 oxoacetic acid group.

Previous attempts to obtain selective thryoid hormerivatives had yielded some compounds with cholest

owering abilities[12,54]. Interestingly, some of these coounds (for example, SFK-94901[55]) contain bulky extenions at the 3′ position of the first thyronine ring. These buxtensions should not be readily accommodated withinR� ligand binding pocket. We synthesized a GC-1 der

ive with a bulky phenyl extension at the 3′ position of the firsing (GC-24)[20]. GC-24 bound to TR� almost as tightly aC-1, and acted as an agonist. It also bound to TR� around0 times more tightly than to TR�. Mutational analysis o

he TR� and TR� pockets suggested that GC-24 derivedf its TR� selectivity from contacts with the ligand bindiocket (analogous to the case with GC-1), but the remaas derived from the 3′ benzyl extension. Thus, TR� must beble to adapt to GC-24 benzyl group and complete the coator binding site and this mechanism must be TR� selective

.1. TRβ/GC-24 structure

We solved the crystal structure of GC-24 in complex whe TR� ligand binding domain to 2.8A resolution usingolecular replacement and refined to anR-factor of 21.6%

20]. As predicted, the GC-1 moiety of GC-24 is positionn the hormone binding pocket much in the same way as

f TR� is less ordered than in TR�, and more accessible to tnsertion. This notion is supported by analysis of temperaactors in TR structures, the TR� H11-H12 loop tends to mo

obile than the equivalent region of TR� [14]. The reasonor this differential stability, however, are not known,re probably related to distant effects of TR isoform spemino acids on H3 and H11 conformation and stability.

One important rearrangement that occurs in the TR�/GC-4 structure involves an Arg residue on H11 (Arg 4

134 M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137

Fig. 9. Rearrangements in charged residues that link TR H11 to H9 and six in the presence of GC-24 and GC-1. Whereas Arg429, Arg 383 and Glu311 forma three way charge cluster in the presence of GC-1, the cluster rearranges in the presence of GC-24. Here, Arg 429 pairs with Asp382 and Arg 383 remainspaired to Glu311.

that usually pairs with arg 383 on H9 and an oppositelycharged Glu residue (Glu311) on H6 in the interior of theTR[14,17,21,56]. This cluster is important; both Arg residuesare mutated in the syndrome of resistance to thyroid hormone[56], and Arg 429 is part of a conserved NR allosteric net-work that dictates whether the RXR partner in heterodimerscan bind its cognate ligand[57]. The charge cluster rearrangesin the presence of GC-24 (Fig. 9); Arg 383 pairs with Glu 311but Arg 429 pairs with Asp328 on H9. Perhaps disruption ofArg429/Glu311 contacts helps H11 to dislocate from the TRLBD.

Provocatively, the Arg429–Asp382 pair juxtaposes two ofthe most highly conserved residues in the NR family (Fig. 10);Arg429 is conserved in about 70% of receptors and Asp 382 in44/48 NRs, even though it has never previously been assigneda function. We do not think that this event occurs by chanceand suggest that GC-24 traps an important conformationalintermediate that arises during the conversion from liganded

TR to the unliganded state, or vice versa, and that the GC-24extension may contact a ligand entry and exit route (Fig. 8).It will be interesting to explore this hypothesis; it suggeststhat rates of formation of this ligand entry/exit route couldvary between the TR isoforms, and contribute to differentialligand effects on the two receptors.

8. Using the extension principle to develop improvedpharmaceuticals

Development of NR ligands with improved specificity iscritical for pharmaceutical design; NRs show great similari-ties in the overall fold and sequence of their LBDs and manyligands exhibit significant cross-reactivity in their binding toclosely related receptors. This has been especially a problemwith the steroid and retinoid receptors; examples of drugsin current use include spironolactone that binds to mineralo-

F 1 (H6) amily. Notet ell cons

ig. 10. Conservation of Charge cluster residues among NRs. Glu 31hat Asp382, which has not previously been assigned a function, is w

, Arg 383, Asp 382 (H9) and Arg 429 (H11) are conserved in the NR ferved.

M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137 135

corticoid and androgen receptors[58,59], RU-486 that bindsto glucocorticoid, androgen and progesterone receptors[60],cortisol that binds to both glucocorticoid and mineralocorti-coid receptors[61,62], and various retinoids that bind to bothRXRs and RARs[1]. An equal problem with specificity is forreceptor subtypes; TRs, RARs, RXRs and PPAR receptors allhave multiple isoforms, and selectively modulating particularisoforms can have clinical advantages[1].

As described above for GC-1, one way to increase selec-tivity is to exploit differences in the ligand binding cavity, butthese approaches are necessarily limited by the enclosed na-ture of the optimal ligand binding pocket for most receptors.The finding that GC-24, with its 3′-extension, exhibited in-creased TR�-selectivity raises the possibility that side chaininsertion could also be exploited to improve selectivity of lig-and binding to other NRs. As an example, the glucocorticoidcortivazol has a bulky ring inserted onto the A ring of thesteroid molecule that should not fit into the glucocorticoidreceptor (GR)[63]. Perhaps this extension could be harboredin an unanticipated area of flexibility within the GR.

We propose that one way to approach this problem of im-proved selectivity is to exploit potential areas of flexibilityin NRs. This would be done by placement of side groups atvarious loci on parent NR ligands, and analysis of their bind-ing and transactivation properties to help us rapidly identifyunanticipated areas of flexibility that could accommodate ex-t tors,t ld in-c

9b

NRL s inL ofr iewo giveu nso 2 ism ithinc tiono mo-b usa andb

g-r nder-s am-i t theL ithT idlyo ce ofT ndst TR

LBD. For example, the TR�/GC-1 and TR�/T3 crystal struc-tures are superimposable, and both compounds behave as ag-onists in most contexts, However, TR represses transcriptionfrom an inverted palindromic TRE in the rat SERCA pro-moter in the presence of GC-1, but not T3 (Gloss, et al., un-published). Thus, the GC-1 and T3-liganded receptors mustdiffer in some subtle way that is not readily observed in X-ray crystal structures but has functionally important conse-quences at this element.

Information about NR LBD dynamics can be obtainedfrom spectroscopic analysis of LBD conformation in solu-tion (see for example[64–67]) [40], use of X-ray crystallo-graphic structural models in molecular dynamics computersimulations[68,69]and, perhaps least obviously, phage dis-play technology as pioneered in the laboratory of Donald Mc-Donnell and at Karo Bio[70–72]. This technology utilizesphage that express a large number of different peptides. Thephage that express peptides that can bind to the NR LBDscan be isolated and the properties of their products exam-ined. Whereas the binding of a number of peptides to NRLBDs bound with different agonists is identical, differencesare sometimes found, and these differences highlight sub-tle differential structural alterations induced by apparentlysimilar agonists. Use of all of these approaches, inc com-bination with X-ray crystallography and mutational analysiswill ultimately tell us much more about the way that differentl s inT ularc

A

281a s in,a BioA rch.

R

Aca-

mily,

e re-.eg-99)

cep-001)

and311.crip-

–141.r lig-999)

ensions. If these regions differ in closely related recephen placement of appropriately placed extensions courease the specificity of ligand binding.

. Conformational changes in the LBD not detectedy X-ray crystallography

A common theme throughout this review is that theBDs are mobile, and that dynamic structural alterationBD structure could play an important role in regulationeceptor activity. X-ray crystallography provides a static vf the receptor in favored configurations and can onlys limited information about protein mobility; comparisof different X-ray crystal structures first revealed that H1obile and analysis of temperature factors (B-factors) w

rystal structures provides information about the distribuf electron density, and can serve as an index of proteinility. Nevertheless, X-ray crystallography will not tellbout dynamic structural alterations that accompany liginding and release, or cofactor association.

It is clearly important to move beyond X-ray crystalloaphy and couple these approaches with an improved utanding of influences of ligands upon the structural dyncs of the TR-LBD. As outlined above, H12 packs againsBD in the TR crystal structures obtained in complex w3 and T4, but other indications suggest that H12 may rapscillate between different conformations in the presen4 [21]. We have also observed different behaviors in liga

hat induce essentially identical conformations within the

igands affect TR structure, how these subtle differenceR structure affect TR function, and how to design particompounds with selective properties.

cknowledgements

This work was supported by NIH grants DK41842, 51nd DK64148 to JDB. Dr. Baxter has proprietary interestnd serves as a consultant and Deputy Director to KaroB, which has commercial interests in this area of resea

eferences

[1] V. Laudet, H. Gronemeyer, The Nuclear Receptor Facts Book,demic Press, London, 2002.

[2] R.M. Evans, The steroid and thyroid hormone receptor superfaScience 240 (1988) 889–895.

[3] R.C.J. Ribeiro, P.J. Kushner, J.D. Baxter, The nuclear hormonceptor gene superfamily, Annu. Rev. Med. 46 (1995) 443–453

[4] N.J. McKenna, R.B. Lanz, B.W. O’Malley, Nuclear receptor corulators: cellular and molecular biology, Endocr. Rev. 20 (19321–344.

[5] A. Chawla, J.J. Repa, R.M. Evans, D.J. Mangelsdorf, Nuclear retors and lipid physiology: opening the X-files, Science 294 (21866–1870.

[6] G.A. Francis, E. Fayard, F. Picard, J. Auwerx, Nuclear receptorsthe control of metabolism, Annu. Rev. Physiol. 65 (2003) 261–

[7] C.K. Glass, M.G. Rosenfeld, The coregulator exchange in transtional functions of nuclear receptors, Genes Dev. 14 (2000) 121

[8] R.V. Weatherman, R.J. Fletterick, T.S. Scanlan, Nuclear-receptoands and ligand-binding domains, Annu. Rev. Biochem. 68 (1559–581.

136 M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137

[9] J.D. Baxter, W.H. Dillmann, B.L. West, R. Huber, J.D. Furlow, R.J.Fletterick, P. Webb, J.W. Apriletti, T.S. Scanlan, Selective modulationof thyroid hormone receptor action, J. Steroid. Biochem. Mol. Biol.76 (2001) 31–42.

[10] P. Webb, N.H. Nguyen, G. Chiellini, H.A. Yoshihara, S.T. CunhaLima, J.W. Apriletti, R.C. Ribeiro, A. Marimuthu, B.L. West, P.Goede, K. Mellstrom, S. Nilsson, P.J. Kushner, R.J. Fletterick, T.S.Scanlan, J.D. Baxter, Design of thyroid hormone receptor antago-nists from first principles, J. Steroid. Biochem. Mol. Biol. 83 (2002)59–73.

[11] J.D. Baxter, P. Webb, G. Grover, T.S. Scanlan, Selective activationof thyroid hormone signaling pathways by GC-1: a new approach tocontrolling cholesterol and body weight, Trends Endocrinol. Metab.15 (2004) 154–157.

[12] P. Webb, Selective activators of thyroid hormone receptors, ExpertOpin. Investig. Drugs 13 (2004) 489–500.

[13] K.W. Nettles, G.L. Greene, Ligand control of coregulator recruitmentof nuclear receptors, Annu. Rev. Physiol. (2004).

[14] R.L. Wagner, J.W. Apriletti, M.E. McGrath, B.L. West, J.D. Baxter,R.J. Fletterick, A structural role for hormone in the thyroid hormonereceptor, Nature 378 (1995) 690–697.

[15] J.P. Renaud, N. Rochel, M. Ruff, V. Vivat, P. Chambon, H. Grone-meyer, D. Moras, Crystal structure of the RAR-gamma ligand-binding domain bound to all-trans retinoic acid, Nature 378 (1995)681–689.

[16] B.D. Darimont, R.L. Wagner, J.W. Apriletti, M.R. Stallcup, P.J.Kushner, J.D. Baxter, R.J. Fletterick, K.R. Yamamoto, Structure andspecificity of nuclear receptor-coactivator interactions, Genes Dev.12 (1998) 3343–3356.

[17] R.L. Wagner, B.R. Huber, A.K. Shiau, A. Kelly, S.T. Cunha Lima,T.S. Scanlan, J.W. Apriletti, J.D. Baxter, B.L. West, R.J. Fletterick,

inol.

[ H.T.roidtancein the03)

[ yen,ith

[ To-gandtor,

[ S.T.J.D.iol.

[ r,son,roid

eptor

[ ert,.V.oxi-137–

[ m-hets of

[ rys-r re-

[26] E.P. Sablin, I.N. Krylova, R.J. Fletterick, H.A. Ingraham, Structuralbasis for ligand-independent activation of the orphan nuclear receptorLRH-1, Mol. Cell. 11 (2003) 1575–1585.

[27] G.F. Allan, X. Leng, S.Y. Tsai, N.L. Weigel, D.P. Edwards, M.J. Tsai,B.W. O’Malley, Hormone and antihormone induce distinct confor-mational changes which are central to steroid receptor activation, J.Biol. Chem. 267 (1992) 19513–19520.

[28] W. Feng, R.C.J. Ribeiro, R.L. Wagner, H. Nguyen, J.W. Apriletti,R.J. Fletterick, J.D. Baxter, P.J. Kushner, B.L. West, Hormone-dependent coactivator binding to a hydrophobic cleft on nuclearreceptors, Science 280 (1998) 1747–1749.

[29] R.C. Ribeiro, W. Feng, R.L. Wagner, C.H. Costa, A.C. Pereira, J.W.Apriletti, R.J. Fletterick, J.D. Baxter, Definition of the surface inthe thyroid hormone receptor ligand binding domain for associationas homodimers and heterodimers with retinoid X receptor, J. Biol.Chem. 276 (2001) 14987–14995.

[30] A. Marimuthu, W. Feng, T. Tagami, H. Nguyen, J.L. Jameson, R.J.Fletterick, J.D. Baxter, B.L. West, TR surfaces and conformationsrequired to bind nuclear receptor corepressor, Mol. Endocrinol. 16(2002) 271–286.

[31] A.K. Shiau, D. Barstad, P.M. Loria, L. Cheng, P.J. Kushner,D.A. Agard, G.L. Greene, The structural basis of estrogen recep-tor/coactivator recognition and the antagonism of this interaction bytamoxifen, Cell 95 (1998) 927–937.

[32] J. Zhang, X. Hu, M.A. Lazar, A novel role for helix 12 of retinoidX receptor in regulating repression, Mol. Cell. Biol. 19 (1999)6448–6457.

[33] L.J. Burke, M. Downes, V. Laudet, G.E. Muscat, Identification andcharacterization of a novel corepressor interaction region in RVRand Rev-erbA alpha, Mol. Endocrinol. 12 (1998) 248–262.

[34] X. Hu, M.A. Lazar, The CoRNR motif controls the recruitment1999)

[ V.m ofptors,

[ ro-.G.essor

[ .G.ket,ill-cruit-813–

[ G.by

n, J.

[ , U.lphalig-ture

[ dy-ation

[ er, J.nds.

[ ell-ro, P.thesis143

Hormone selectivity in thyroid hormone receptors, Mol. Endocr15 (2001) 398–410.

18] B.R. Huber, M. Desclozeaux, B.L. West, S.T. Cunha-Lima,Nguyen, J.D. Baxter, H.A. Ingraham, R.J. Fletterick, Thyhormone receptor-beta mutations conferring hormone resisand reduced corepressor release exhibit decreased stabilityN-terminal ligand-binding domain, Mol. Endocrinol. 17 (20107–116.

19] B.R. Huber, B. Sandler, B.L. West, S.T. Cunha Lima, H.T. NguJ.W. Apriletti, J.D. Baxter, R.J. Fletterick, Two RTH mutants wimpaired hormone binding, Mol. Endocrinol. (2003).

20] S. Borngraeber, M.J. Budny, G. Chiellini, S.T. Cunha-Lima, M.gashi, P. Webb, J.D. Baxter, T.S. Scanlan, R.J. Fletterick, Liselectivity by seeking hydrophobicity in thyroid hormone recepProc. Natl. Acad. Sci. U.S.A. 100 (2003) 15358–15363.

21] B. Sandler, P. Webb, J.W. Apriletti, B.R. Huber, M. Togashi,Cunha Lima, S. Juric, S. Nilsson, R. Wagner, R.J. Fletterick,Baxter, Thyroxine-thyroid hormone receptor interactions, J. BChem. (2004).

22] L. Ye, Y.L. Li, K. Mellstrom, C. Mellin, L.G. Bladh, K. KoehleN. Garg, A.M. Garcia Collazo, C. Litten, B. Husman, K. PersJ. Ljunggren, G. Grover, P.G. Sleph, R. George, J. Malm, Thyreceptor ligands. 1. Agonist ligands selective for the thyroid recbeta1, J. Med. Chem. 46 (2003) 1580–1588.

23] R.T. Nolte, G.B. Wisely, S. Westin, J.E. Cobb, M.H. LambR. Kurokawa, M.G. Rosenfeld, T.M. Willson, C.K. Glass, MMilburn, Ligand binding and co-activator assembly of the persome proliferator-activated receptor gamma, Nature 395 (1998)143.

24] R.E. Watkins, G.B. Wisely, L.B. Moore, J.L. Collins, M.H. Labert, S.P. Williams, T.M. Willson, S.A. Kliewer, M.R. Redinbo, Thuman nuclear xenobiotic receptor PXR: structural determinandirected promiscuity, Science 292 (2001) 2329–2333.

25] W. Bourguet, M. Ruff, P. Chambon, H. Gronemeyer, D. Moras, Ctal structure of the ligand-binding domain of the human nucleaceptor RXR-�, Nature 375 (1995) 377–382.

of corepressors by nuclear hormone receptors, Nature 402 (93–96.

35] L. Nagy, H.Y. Kao, J.D. Love, C. Li, E. Banayo, J.T. Gooch,Krishna, K. Chatterjee, R.M. Evans, J.W. Schwabe, Mechaniscorepressor binding and release from nuclear hormone receGenes Dev. 13 (1999) 3209–3216.

36] V. Perissi, L.M. Staszewski, E.M. McInerney, R. Kurokawa, A. Knes, D.W. Rose, M.H. Lambert, M.V. Milburn, C.K. Glass, MRosenfeld, Molecular determinants of nuclear receptor-coreprinteraction, Genes Dev. 13 (1999) 3198–3208.

37] H.E. Xu, T.B. Stanley, V.G. Montana, M.H. Lambert, BShearer, J.E. Cobb, D.D. McKee, C.M. Galardi, K.D. PlunR.T. Nolte, D.J. Parks, J.T. Moore, S.A. Kliewer, T.M. Wson, J.B. Stimmel, Structural basis for antagonist-mediated rement of nuclear co-repressors by PPARa, Nature 415 (2002)817.

38] B.A. Johnson, E.M. Wilson, Y. Li, D.E. Moller, R.G. Smith,Zhou, Ligand-induced stabilization of PPARgamma monitoredNMR spectroscopy: implications for nuclear receptor activatioMol. Biol. 298 (2000) 187–194.

39] P. Cronet, J.F. Petersen, R. Folmer, N. Blomberg, K. SjoblomKarlsson, E.L. Lindstedt, K. Bamberg, Structure of the PPARaand -gamma ligand binding domain in complex with AZ 242;and selectivity and agonist activation in the PPAR family, Struc(Camb.) 9 (2001) 699–706.

40] B.C. Kallenberger, J.D. Love, V.K. Chatterjee, J.W. Schwabe, Anamic mechanism of nuclear receptor activation and its perturbin a human disease, Nat. Struct. Biol. 10 (2003) 136–140.

41] T.S. Scanlan, J.D. Baxter, R.J. Fletterick, P. Kushner, R. WagnApriletti, B. West, A. Shiau, Thyroid hormone receptors and ligaUniversity of California, International Patent (1996).

42] J.D. Baxter, P. Goede, J.W. Apriletti, B.L. West, W. Feng, K. Mstrom, R.J. Fletterick, R.L. Wagner, P.J. Kushner, R.C.J. RibeiWebb, T.S. Scanlan, S. Nilsson, Structure-based design and synof a thyroid hormone receptor (TR) antagonist, Endocrinology(2002) 517–524.

M. Togashi et al. / Journal of Steroid Biochemistry & Molecular Biology 93 (2005) 127–137 137

[43] N.H. Nguyen, J.W. Apriletti, S.T. Cunha Lima, P. Webb, J.D. Baxter,T.S. Scanlan, Rational design and synthesis of a novel thyroid hor-mone antagonist that blocks coactivator recruitment, J. Med. Chem.45 (2002) 3310–3320.

[44] W. Lim, N.H. Nguyen, H.Y. Yang, T.S. Scanlan, J.D. Furlow, Athyroid hormone antagonist that inhibits thyroid hormone action invivo, J. Biol. Chem. 277 (2002) 35664–35670.

[45] L.E. Braverman, R.D. Utiger, Werner’s and Ingbar’s The Thyroid:A Fundamental and Clinical Text, Lippincott Williams & Wilkins,Philadelphia, 2000.

[46] J.W. Apriletti, N.L. Eberhardt, K.R. Latham, J.D. Baxter, Affinitychromatography of thyroid hormone receptors: biospecific elutionfrom support matrices, characterization of the partially purified re-ceptor, J. Biol. Chem. 256 (1981) 12094–12101.

[47] S. Ekins, E. Schuetz, The PXR crystal structure: the end of thebeginning, Trends Pharmacol. Sci. 23 (2002) 49–50.

[48] R.E. Watkins, S.M. Noble, M.R. Redinbo, Structural insights into thepromiscuity and function of the human pregnane X receptor, Curr.Opin. Drug Discov. Dev. 5 (2002) 150–158.

[49] R.E. Watkins, P.R. Davis-Searles, M.H. Lambert, M.R. Redinbo,Coactivator binding promotes the specific interaction between lig-and and the pregnane X receptor, J. Mol. Biol. 331 (2003) 815–828.

[50] R.E. Watkins, J.M. Maglich, L.B. Moore, G.B. Wisely, S.M. No-ble, P.R. Davis-Searles, M.H. Lambert, S.A. Kliewer, M.R. Red-inbo, 2.1 A crystal structure of human PXR in complex with the St.John’s Wort compound hyperforin, Biochemistry 42 (2003) 1430–1438.

[51] J.M. Olefsky, A.R. Saltiel, PPAR gamma and the treatment of insulinresistance, Trends Endocrinol. Metab. 11 (2000) 362–368.

[52] A.K. Shiau, D. Barstad, J.T. Radek, M.J. Meyers, K.W. Nettles, B.S.ene,

als aol. 9

[ C.lig-99–

[ tivepin.

[ , J.ismcerans-

[ od,, Ayn-tran-

[ truc-et-

[58] J.D. Baxter, J.W. Funder, J.W. Apriletti, P. Webb, Towards selectivelymodulating mineralocorticoid receptor function: lessons from othersystems, Mol. Cell. Endocrinol. 217 (2004) 151–165.

[59] J.A. Delyani, Mineralocorticoid receptor antagonists: the evolutionof utility and pharmacology, Kidney Int. 57 (2000) 1408–1411.

[60] I.M. Spitz, Progesterone antagonists and progesterone receptor mod-ulators: an overview, Steroids 68 (2003) 981–993.

[61] P.L. Ballard, J.P. Carter, B.S. Graham, J.D. Baxter, A radioreceptorassay for evaluation of the plasma glucocorticoid activity of naturaland synthetic steroids in man, J. Clin. Endocrinol. Metab. 41 (1975)290–304.

[62] N.C. Lan, D.T. Matulich, J.R. Stockigt, E.G. Biglieri, M.I. New, J.S.Winter, J.K. McKenzie, J.D. Baxter, Radioreceptor assay of plasmamineralocorticoid activity. Role of aldosterone, cortisol, and deoxy-corticosterone in various mineralocorticoid-excess states, Circ. Res.46 (1980) 94–100.

[63] E.B. Thompson, L.V. Nazareth, R. Thulasi, J. Ashraf, D. Harbour,B.H. Johnson, Glucocorticoids in malignant lymphoid cells: generegulation and the minimum receptor fragment for lysis, J. SteroidBiochem. Mol. Biol. 41 (1992) 273–282.

[64] A. Tamrazi, J.A. Katzenellenbogen, Site-specific fluorescent labelingof estrogen receptors and structure-activity relationships of ligandsin terms of receptor dimer stability, Methods Enzymol. 364 (2003)37–53.

[65] A. Tamrazi, K.E. Carlson, J.A. Katzenellenbogen, Molecular sensorsof estrogen receptor conformations and dynamics, Mol. Endocrinol.17 (2003) 2593–2602.

[66] A. Tamrazi, K.E. Carlson, J.R. Daniels, K.M. Hurth, J.A. Katzenel-lenbogen, Estrogen receptor dimerization: ligand binding regulatesdimer affinity and dimer dissociation rate, Mol. Endocrinol. 16(2002) 2706–2719.

[ rd,ceptorhem-

[ inoicthe

115.[ om

J. 76

[ tlin,n-

duceNatl.

[ Ke-eare li-alpha

[ ua-ep-(1999)

Katzenellenbogen, J.A. Katzenellenbogen, D.A. Agard, G.L. GreStructural characterization of a subtype-selective ligand revenovel mode of estrogen receptor antagonism, Nat. Struct. Bi(2002) 359–364.

53] G. Chiellini, J.W. Apriletti, H.A.I. Yoshihara, J.D. Baxter, R.Ribeiro, T.S. Scanlan, A high-affinity subtype-selective agonistand for the thyroid hormone receptor, Chem. Biol. 5 (1998) 2306.

54] T.S. Scanlan, H.A. Yoshihara, N.H. Nguyen, G. Chiellini, Selecthyromimetics: tissue-selective thyroid hormone analogs, Curr. ODrug Discov. Dev. 4 (2001) 614–622.

55] K. Ichikawa, T. Miyamoto, T. Kakizawa, S. Suzuki, A. KanekoMori, M. Hara, M. Kumagai, T. Takeda, K. Hashizume, Mechanof liver-selective thyromimetic activity of SK&F L-94901: evidenfor the presence of a cell-type-specific nuclear iodothyronine tport process, J. Endocrinol. 165 (2000) 391–397.

56] R.J. Clifton-Bligh, F. de Zegher, R.L. Wagner, T.N. CollingwoI. Francois, M. Van Helvoirt, R.J. Fletterick, V.K.K. Chatterjeenovel TR� mutation (R383H) in resistance to thyroid hormone sdrome predominantly impairs corepressor release and negativescriptional regulation, Mol. Endocrinol. 12 (1998) 609–621.

57] A.I. Shulman, C. Larson, D.J. Mangelsdorf, R. Ranganathan, Stural determinants of allosteric ligand activation in RXR herodimers, Cell 116 (2004) 417–429.

67] K.M. Hurth, M.J. Nilges, K.E. Carlson, A. Tamrazi, R.L. BelfoJ.A. Katzenellenbogen, Ligand-induced changes in estrogen reconformation as measured by site-directed spin labeling, Biocistry 43 (2004) 1891–1907.

68] A. Blondel, J.P. Renaud, S. Fischer, D. Moras, M. Karplus, Retacid receptor: a simulation analysis of retinoic acid binding andresulting conformational changes, J. Mol. Biol. 291 (1999) 101–

69] D. Kosztin, S. Izrailev, K. Schulten, Unbinding of retinoic acid frits receptor studied by steered molecular dynamics, Biophys.(1999) 188–197.

70] L.A. Paige, D.J. Christensen, H. Gron, J.D. Norris, E.B. GotK.M. Padilla, C.Y. Chang, L.M. Ballas, P.T. Hamilton, D.P. McDonell, D.M. Fowlkes, Estrogen receptor (ER) modulators each indistinct conformational changes in ER alpha and ER beta, Proc.Acad. Sci. U.S.A. 96 (1999) 3999–4004.

71] C. Chang, J.D. Norris, H. Gron, L.A. Paige, P.T. Hamilton, D.J.nan, D. Fowlkes, D.P. McDonnell, Dissection of the LXXLL nuclreceptor-coactivator interaction motif using combinatorial peptidbraries: discovery of peptide antagonists of estrogen receptorsand beta, Mol. Cell. Biol. 19 (1999) 8226–8239.

72] J.D. Norris, L.A. Paige, D.J. Christensen, C.Y. Chang, M.R. Hcani, D. Fan, P.T. Hamilton, D.M. Fowlkes, D.P. McDonnell, Ptide antagonists of the human estrogen receptor, Science 285744–746.