THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2005 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 280, No. 27, Issue of July 8, pp. 25665–25673, 2005 Printed in U.S.A.

Rearrangements in Thyroid Hormone Receptor Charge Clusters That Stabilize Bound 3,5؅,5-Triiodo-L-thyronine and Inhibit Homodimer Formation*
Received for publication, February 11, 2005, and in revised form, May 9, 2005 Published, JBC Papers in Press, May 10, 2005, DOI 10.1074/jbc.M501615200

Marie Togashi‡, Phuong Nguyen‡, Robert Fletterick§, John D. Baxter‡¶, and Paul Webb‡ʈ
From the ‡Diabetes Center and the §Department of Biochemistry and Biophysics, University of California, San Francisco, California 94143-0540

In this study, we investigated how thyroid hormone (3,5؅,5-triiodo-L-thyronine, T3) inhibits binding of thyroid hormone receptor (TR) homodimers, but not TRretinoid X receptor heterodimers, to thyroid hormone response elements. Specifically we asked why a small subset of TR␤ mutations that arise in resistance to thyroid hormone syndrome inhibit both T3 binding and formation of TR␤ homodimers on thyroid hormone response elements. We reasoned that these mutations may affect structural elements involved in the coupling of T3 binding to inhibition of TR DNA binding activity. Analysis of TR x-ray structures revealed that each of these resistance to thyroid hormone syndrome mutations affects a cluster of charged amino acids with potential for ionic bond formation between oppositely charged partners. Two clusters (1 and 2) are adjacent to the dimer surface at the junction of helices 10 and 11. Targeted mutagenesis of residues in Cluster 1 (Arg338, Lys342, Asp351, and Asp355) and Cluster 2 (Arg429, Arg383, and Glu311) confirmed that the clusters are required for stable T3 binding and for optimal TR homodimer formation on DNA but also revealed that different arrangements of charged residues are needed for these effects. We propose that the charge clusters are homodimer-specific extensions of the dimer surface and further that T3 binding promotes specific rearrangements of these surfaces that simultaneously block homodimer formation on DNA and stabilize the bound hormone. Our data yield insight into the way that T3 regulates TR DNA binding activity and also highlight hitherto unsuspected T3-dependent conformational changes in the receptor ligand binding domain. Thyroid hormone receptors (TR␣ and TR␤)1 are conditional transcription factors that play important roles in development,
* This work was supported by National Institutes of Health Grants DK41482 and DK51281 (to J. D. B.). 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. ¶Deputy director and consultant to Karo Bio AB, a biotechnology company with commercial interests in nuclear receptors. ʈ To whom correspondence should be addressed: Diabetes Center, University of California School of Medicine, HSW1210, 513 Parnassus Ave., San Francisco, CA 94143-0540. Tel.: 415-476-6789; Fax: 415-5645813; E-mail: 1 The abbreviations used are: TR, thyroid hormone receptor; T3, 3,5Ј,5-triiodo-L-thyronine; RXR, retinoid X receptor; TRE, thyroid hormone response element; RTH, resistance to thyroid hormone syndrome; H, helix; LBD, ligand binding domain; N-CoR, nuclear receptor corepressor; AF, activation function; DR, direct repeat; IP, inverted palindrome; GRIP1, glucocorticoid receptor-interacting protein 1; ANOVA, analysis of variance; PPAR, peroxisome proliferator-activated receptor; NR, nuclear hormone receptor.
This paper is available on line at

metabolism, and homeostasis (1– 4). TRs regulate gene transcription in the presence of 3,5,3Јtriiodo-L-thyronine (T3) and in the absence of ligand (5). Current efforts to modulate TR activities have focused on development of selective agonists that mimic the beneficial effects of T3 upon circulating cholesterol and body weight without producing unwanted effects of the hormone on heart rate (6). However, there is also a need for TR antagonists, which could represent improved and faster acting treatments for hyperthyroidism and cardiac arrhythmias (6, 7). Furthermore observations from TR␣/TR␤ knock-out mice suggest many clinical manifestations of hypothyroidism are due to actions of unliganded TRs (8, 9). Thus, drugs that specifically reverse actions of unliganded TRs could be useful for treating hypothyroidism and would avoid risk of thyroid hormone excess (7). Improved understanding of unliganded TR structure and ways that unliganded TRs rearrange in response to T3 will facilitate development of all of these drugs. Presently the organization of unliganded TR is only partly understood (10, 11). X-ray structures of liganded TR C-terminal ligand binding domains (LBDs) reveal a canonical ␣-helical structure with T3 buried in the core of the protein (12–16), but there are no equivalent structures of unliganded TRs. It has proven possible, however, to use a combination of x-ray structural information and targeted mutagenesis to learn about the organization of unliganded TRs. For example, T3 blocks transactivation and transrepression activities of unliganded TRs by promoting release of corepressors such as N-CoR and SMRT (silencing mediator of retinoid and thyroid receptors) (5) and induces a T3-dependent activation function (AF-2) that binds coactivators such as the p160s (17). Functional analysis of TR mutants reveals that AF-2 is comprised of surface-exposed residues from helices (H) 3, 5, and 12 and that the corepressor binding surface overlaps AF-2 but extends below the position of H12 in the liganded state (18 –21). Thus, it is possible to infer that H12 is displaced in the unliganded state and that T3 binding leads to repositioning of H12 over the lower part of the corepressor binding surface, simultaneously promoting corepressor release and completing the coactivator binding site (5). T3 also regulates TR DNA binding activity (1). TRs utilize their DNA binding domain to recognize specific thyroid hormone response elements (TREs) comprised of AGGTCA repeats and bind these elements either as heterodimers with the closely related retinoid X receptor (RXR) or as homodimers and monomers. T3 does not affect RXR-TR interactions with TREs but does promote release of TR homodimers from some TREs (inverted palindromes (F2/IP-6) and direct repeats (DR-4)) but not from TREs at which TRs bind as monomers or paired monomers (palindromes, TREpal) (22). TR homodimers bind N-CoR more strongly than RXR-TR heterodimers (23, 24), and the extent of TR homodimer binding to different TREs in vitro correlates with the extent of repression from these elements in

Downloaded from at CAPES - UNB, on November 6, 2012


human constitutive androstane receptor.1% Triton X-100. Arg383. We propose that the charge clusters rearrange upon T3 binding to block homodimer formation and create new ionic bonds that stabilize bound hormone. Glutathione S-Transferase Pull-down Assays—Full-length human TR␤ was expressed in a coupled transcription translation system (TNT. and TREpal) upstream of the herpes simplex virus thymidine kinase promoter TATA box linked to luciferase coding sequence. but not heterodimers. 2012 FIG. 10% glycerol. human retinoic acid at CAPES . hLXR. and pellets were lysed by addition of 150 ␮l of 100 mM Tris-HCl. The structural elements that render homodimers sensitive to T3 are not known. charge Clusters 1 and 2 are adjacent to the TR␤ dimer surface. Transfections—HeLa cells were maintained in Dulbecco’s modified Eagle’s H-21 4. For transfection. closer view interactions between the residues that comprise charge Clusters 1 (B) and 2 (C). cells were collected.UNB. hCAR. The figure shows a space-filling model of the TR␤ LBD. 29). and Arg429) are shown in blue (positively charged) and red (negatively charged).25666 Thyroid Hormone Modulation of DNA Binding Downloaded from www. to TREs (30. Lys342. 28. Leu422.. Met423. hER. Here we report that these RTH mutations affect clusters of charged amino acids in the LBD with potential for electrostatic stabilization of TR conformation but that distinct arrangements of charged residues are needed for stable T3 binding and DNA binding by unliganded TRs. and 2 ␮g of pCMV-␤-galactosidase (17). MATERIALS AND METHODS TR Mutants—The pCMX vector was used for expression of the fulllength human TR␤ (17).5 ϫ 107 cells) containing 0. human pregnane X receptor. we utilized targeted mutagenesis to explore elements of the TR that are specifically required for homodimer formation on TREs and tested the hypothesis that the same elements are involved in coupling T3 binding to inhibition of DNA binding. Thus. the bound proteins . human estrogen receptor. it is thought that T3-dependent inhibition of homodimer formation relieves transcriptional repression by unliganded TRs. Beads were washed three times with 200 ␮l of binding buffer. In this study. 1. Pro419.5 ml/4. Nevertheless the mechanisms involved in coupling of T3 binding to inhibition of DNA binding are not clear.8. Inc. 31). Residues in Cluster 1 (Arg338. 1 ␮g of TR expression vector or empty vector control. Promega). CA). a small subset also inhibits binding of TR␤ homodimers. The reporters contained two copies of each TRE (DR-4.jbc. B and C. and negatively charged residues are shown in red. E. hPXR. hRAR. and Met430) are shown in green. pH 7. and plated into 12-well plates. Luciferase and ␤-galactosidase activities were measured by using a luciferase assay system (Promega) and Galacto-Light Plus ␤-galactosidase reporter gene assay system (Applied Biosystems). and Asp355) and Cluster 2 (Glu311. alignment of Cluster 1 residues in TR and other NRs. After incubation for 24 h at 37 °C with T3 or vehicle.2 mM phenylmethylsulfonyl fluoride. Pierce) with 1–2 ␮l of 35S-labeled human TR␤ in 150 ␮l of binding buffer (20 mM HEPES. D. TRs utilize the same surface at the junction of H10 and H11 in homodimer and heterodimer formation on DNA (27). Mutations within TR-encoding sequences were created using the QuikChange XL site-directed mutagenesis kit (Strat- agene). Asterisk represents residues mutated in RTH. Bindings were performed by mixing glutathione-linked Sepharose beads containing 4 ␮g of glutathione Stransferase fusion proteins (Coomassie Plus protein assay reagent. N-CoR (amino acids 1944 –2453) and GRIP1 (amino acids 563–1121) were expressed in Escherichia coli strain BL21 as a fusion protein with glutathione S-transferase according to the manufacturer’s protocol (Amersham Biosciences). 25 mM MgCl2. and protease inhibitors) containing 20 ␮g/ml bovine serum albumin for 1.1% dextrose and typically 4 ␮g of reporter. alignment of Cluster 2 residues in TR and other NRs. human vitamin D receptor. Asp351. vivo (25. 26). F2. 0. 2 mM glutamine. Residues in the dimerization surface (Leu400. A. Hayward. 1 mM dithiothreitol. on November 6.5 g/liter glucose medium containing 10% fetal bovine serum. Whereas most TR␤ mutations that arise in resistance to thyroid hormone syndrome (RTH) reduce the affinity of TR␤ for T3 (3. human liver X receptor. 50 units/ml penicillin. cells were collected and resuspended in Dulbecco’s phosphate-buffered saline (0. Mutation of target sequences was verified by automated DNA sequence (Elim Biopharmaceuticals. Location of RTH mutations in charge clusters that are adjacent to the dimer surface. and 50 mg/ml streptomycin.5 h. hTR. hVDR. Cells were electroporated at 240 V and 960 microfarads. human TR. transferred to fresh media. 150 mM KCl. containing 0. Positively charged residues are shown in blue.

0. we found that another RTH mutation that affects a positively charged Lys residue (K342I) also inhibits homodimer formation on DNA (not shown).jbc.03 program (GraphPad Software. Binding curves were fit by nonlinear regression. Y is total binding (cpm).Thyroid Hormone Modulation of DNA Binding TABLE I Location and conservation of TR charge clusters RTH mutants known to inhibit DNA binding activity are shown in bold. Cluster Residues Location RTH mutants Conservation 25667 1 2 3 4 Arg . 1. The gel was then fixed. Investigation of TR structural models revealed that each of these amino acids lies within separate clusters of closely juxtaposed charged residues (Fig. After 30 min at room temperature.. one-phase exponential decay. the mixture was loaded onto a 5% nondenaturing polyacrylamide gel that was previously run for 30 min at 200 V. Glu311 Arg316. These aliquots were applied to Sephadex G-25 columns. Asp351. 2012 FIG. Lys . 0. 10 ␮M ZnSO4. Inc. His229 Arg410.05). Activities obtained with TR mutants are compared with those obtained with wild type TR␤. with 300. aliquots were taken at the indicated time points to determine how rapidly the labeled ligand dissociates from TR. the gel was run at 4 °C for 120 –180 min at 200 V in a running buffer containing 45 mM Tris borate (pH 8. the amount of specific binding follows an exponential dissociation equation: Y ϭ Span⅐eϪk⅐x ϩ Plateau where x is time (min). Oneway ANOVA with Tukey’s post-test or t test was performed using GraphPad Prism version 3.0) and 1 mM EDTA. and onephase exponential association models. Asp355 Arg429.0 mM MgCl2. 10% glycerol). In cases in which TR ligand binding activity was severely affected by Cluster 1 mutations. Clusters 1 and 2 are comprised of residues that are exposed or partially exposed on the surface of the LBD and are both adjacent to the classical dimer surface at the junction of H10 and H11 (Fig. and R316H) (30.05 was considered statistically significant. no statistical difference was found among mutants and wild type (p Ͼ 0. Asp366. and k is the dissociation rate constant (koff. Asp351 and Asp355. RTH Mutations That Inhibit T3 and DNA Binding Reside in Charge Clusters—RTH mutations that inhibit homodimer formation on DNA affect positively charged Arg residues (R338W. Thr232. R429Q R316H None Charged residues at similar location in several NRs Arg429 conserved in 70% of NRs Polar cluster observed in similar location in many NRs Glu369 well conserved were resuspended in SDS-PAGE loading buffer. and proteins were separated using 10% SDS-polyacrylamide gel electrophoresis and visualized by autoradiography. R429Q. expressed in minϪ1). The TR␤ LBD contains only one similar charge cluster that is not known to be affected by RTH mutations (Cluster 4. Cluster 1 includes Arg338 and Lys342 on H7 and two negatively charged residues on H8. links H7-H8 Partially buried. based on analysis of x-ray crystal structures of the TR␣ LBD (12).05) according to ANOVA and Tukey’s test. and Asp355 is not directly engaged in the cluster (Fig. The EC50 T3 concentration for wild type TR␤ was 15 nM for the F2 driven reporter. In A. wild type. A p value of Ͻ0. To visualize the TR-DNA complexes. 1 mM dithiothreitol. Statistical Analysis—All data are presented as means Ϯ S.000 cpm [␥-32P]ATP-radiolabeled DR-4 and F2 oligonucleotides and 1 ␮g of poly(dI-dC) (Amersham Biosciences) in a 20-␮l reaction (32).UNB. both of which are also mutated in RTH but reported not to affect DNA binding (36). Briefly 15 fmol of each in vitro translated protein were incubated overnight at 4 °C with varying concentrations of [125I]T3 (PerkinElmer Life Sciences) in 100 ␮l of E400 buffer (400 mM NaCl. which is set to 100%. 20 mM KPO4. Glu369 338 342 Surface. 2. 1A and Table I). none of the residues in the clusters directly contacts T3 or comprises part of a known coregulator binding surface. WT or wt. T3 Binding Assay—TRs were expressed using the TNT T7 quick coupled transcription translation system (Promega).D. Cluster 2 includes Arg429 on H11 and Arg383 on H9. Data analyzed referred to at least three independent experiments. Span is the difference between binding at time 0 and plateau (cpm). and is completely surface-exposed. R338L. pH 8. dried. Off rate (koff) was determined by adding a 1000-fold molar excess of unlabeled T3 to a mixture containing TR and 1 nM [125I]T3 incubated previously overnight at 4 °C. Unlike many residues that are affected by RTH mutations. 33–35).org at CAPES . and TR-bound [125I]T3 was quantified using a ␥-counter. Packard Instruments). K342I. As each T3-TR complex dissociates at a random time.) activation and repression (A) and dose of T3 (B) required for half-maximal activation at a T3-regulated reporter gene with an F2/IP-6 TRE. The bound [125I]T3 was isolated by gravity flow through a 2-ml Sephadex G-25 (Amersham Biosciences) column and quantified using a ␥-counter (COBRA. 1A and Table I). In addition. Charge Cluster 1 mutations inhibit T3 activation. Gel Shifts—Binding of TR to DNA was assayed by mixing 20 fmol of TRs produced in a reticulocyte lysate system. 5% glycerol.5 mM EDTA. Maximal (Max. Gln374. CA). TNT T7 (Promega). 1 mM monothioglycerol. links H6 and H9 to H1 Buried. the overall amount of translated TRs in the extracts was also verified independently by Western blot. and exposed for autoradiography. links H9 and H10-H11 loop R338W. The affinities of T3 binding were determined using a saturation binding assay. different letters over bars indicate statistical difference (p Ͻ 0. San Diego. His412. and is partially surface-exposed. on November 6. 1B). contained in the Prism version 3. Arg383. Reinvestigation of TR␤-LBD structures (13) suggested another arrangement: Arg338 and Lys342 both pair with Asp351. 31. and 50 ␮g of calf thymus histones (Calbiochem). We originally suggested that Arg338 and Lys342 engage in parallel ionic pairings with Asp355 and Asp351. respectively. A337del R383H. Here x-ray structures of . see Table I).03 for Windows. RESULTS Downloaded from www. links H11 and H10 to H6 Buried. The binding buffer contained 25 mM HEPES. and dissociation constant (Kd) and koff values were calculated using the one-site saturation binding. In B. respectively. 50 mM KCl. The residues in Clusters 1 and 2 have the potential to engage in electrostatic interactions with each other.1% Nonidet P-40.

2 33 Ϯ 3. Nevertheless several TR␤ Cluster 1 mutants displayed altered T3 concentration dependence (Fig.2 37 Ϯ 16.4 ϫ 10Ϫ12 M and set to 100%. Mutants are compared with values obtained with wild type TR. 1C). TR␤ Cluster 1 mutants did not affect maximal activation of transcription from a TRE-driven reporter (F2) in the presence of saturating T3 or repression of basal transcription in the absence of T3 (Fig. We introduced 1) Ala substitutions. and 2) charge reversal mutations.0b.1. Residues in Clusters 1 and 2 show considerable conservation. which swap a residue with a small neutral side chain for a residue with a charged side chain and thereby eliminate the potential for electrostatic interactions.3a.2 73 Ϯ 9.8a. which should disrupt ionic bonds between oppositely charged residues by juxtaposing residues with like charges.1 100a. Effects of Cluster 1 mutations on activities of liganded TRs in vitro. Values represent the averages of at least three determinations.1 68 Ϯ 11.1b. including retinoic acid receptors and PPARs (Fig. Residues equivalent to those in charge Cluster 1 are conserved on H7 and H8 in other NRs. Fig. Furthermore the fact that mutations in the clusters affect T3 binding and homodimer formation indicates that the clusters must play a role in activities associated with liganded and unliganded TRs. Cluster 1 Is Required for Optimal T3 Binding—We first examined effects of mutations in Cluster 1 on activities of liganded TRs. A.UNB. equilibrium dissociation at CAPES .1 389 Ϯ 98. Residues in Cluster 2 show even better conservation (Fig.05) according to ANOVA and Tukey’s test. TRs bearing Ala substitutions at Arg338 and Asp351 (TR␤R338A and TR␤D351A) exhibited more modest reductions in T3 sensitivity.7a.5a. glutathione S-transferase. In contrast.2 56 Ϯ 11. Finally mutations at Asp355 did not reduce T3 sensitivity.1 112 Ϯ 14. 2 shows effects of mutations on activity of transfected TR␤ in mammalian cells. and different numbers in the same vertical column indicate statistical difference (p Ͻ 0. Shown are autoradiograms of SDS-polyacrylamide gels used to separate labeled TRs bound to bacterially expressed GRIP1 (amino acids 563–1121) and N-CoR (amino acids 1944 –2453) in pull-down assays. kinetics of ligand dissociation from wild type and mutant TRs. Because residues of this cluster are completely surface-exposed it appeared unlikely that these mutations would exert indirect effects on TR function by disrupting internal folding of the LBD. A relatively mild Ala substitution mutation (TR␤K342A) had either no effect or slightly enhanced T3 sensitivity (Table II). In B and C. 3. Mean values Ϯ S. are the average of at least three experiments.7c. different letters over bars indicate statistical difference (p Ͻ 0. mutations in Cluster 1 do not affect coregulator binding. and unappreciated.1 Downloaded from www.7a.1 100a. TR␤K342D. wild type. They are conserved in TRs throughout vertebrate species (not shown). Nevertheless a more severe charge reversal mutation.6a.1 102 Ϯ 27.2 608 Ϯ 185. different mutations at Lys342 exhibited divergent effects.05) according to ANOVA and Tukey’s test. exhibited decreased T3 sensitivity as did the TR␤K342I RTH mutant (not shown).b.5a. TR␤D355A and TR␤D355R either exhibited T3 sensitivity com- .2 308 Ϯ 56.25668 Thyroid Hormone Modulation of DNA Binding TABLE II Average EC50 for T3 response obtained with TRs bearing Cluster 1 mutants at different TREs Values are compared to wild type TR set at 100%. Mutations in two residues (Arg338 and Asp351) led to reduced T3 sensitivity. and TRs with charge reversal mutations at Arg338 and Asp351 (TR␤R338D and TR␤D351R) displayed more marked reductions in T3 sensitivity. role in TR activities.05).1 357 Ϯ 149. which was 161. GST. Values represent the averages of at least three determinations.jbc. 2012 FIG.6b. B. TR␣ and TR␤ indicated that both Arg residues pair with Glu311 on H6 in the LBD core (Fig. 2B and Table II) both in HeLa cells (shown here) and in other cells (U2-OS and CV-1. Different letters in the same horizontal row indicate statistical difference (p Ͻ 0. The result is representative of three experiments. 1E). koff. The TR␤R338W RTH mutant required 17-fold more T3 than wild type TR␤ for half-maximal activation (EC50).2 155 Ϯ 37.1. TRE/TR TR R338A K342A D351A D355A F2 (IP-6) DR-4 TREpal 100a. Kd. not shown).D. 1D). C. Together all of these considerations indicate that the charge clusters play an important. on November 6.5c. WT or wt. 2A).

and Glu311) also exhibited differential effects on activity of liganded TRs and DNA binding. the same mutations that reduced T3 sensitivity in vivo also reduced the affinity of the TR for T3 (Fig. Arg429. Mutations in Cluster 1 Can Either Impair or Enhance Homodimer Binding to DNA. Again these effects were largely homodimer-specific. 5A shows that TR␤E311A exhibited a much larger reduction in T3 sensitivity than TRs bearing mutations at Arg429 and Arg383 (TR␤R429A and TR␤R383H) (Fig. 2B and 4B). 2012 FIG. A. TRs bearing Ala substitution mutants at Arg338 and Asp351 (both required for optimal T3 binding) exhibited reduced homodimer but not heterodimer formation at F2 and DR-4 elements (Fig.Thyroid Hormone Modulation of DNA Binding 25669 FIG. B. Retic. A. an artificial mutation (TR␤L422R) in the classical dimer interface abolished both homodimer and heterodimer formation. Fig. In A. and Effects Do Not Correlate with T3 Binding—Next we examined effects of mutations in Cluster 1 on TR homodimer formation on DNA. 3C). 3B) and increased T3 dissociation rates (Fig. TR␤K342A displayed reduced homodimer formation even though it did not inhibit T3 binding (compare Figs. Together our results confirm that Cluster 1 is required for TR homodimer but not heterodimer formation on DNA. Charge Cluster 2 Residues Differentially Affect T3 Activation and DNA Binding—Mutations in Cluster 2 (Arg383. 5. comparison of TR␤ with charge reversal mutants. By contrast. although TR␤K342D did exhibit somewhat enhanced heterodimer formation. different letters over bars indicate statistical difference (p Ͻ 0. The precise effect of the charge reversal mutants varied. Fig.UNB. which only break one bond. Asp351. 4. TR␤D355R exhibited enhanced DNA binding in the presence of T3. Nevertheless the same data also revealed that different arrangements of charge are needed for optimal DNA and T3 binding. In parallel. 3A) in pull-down assays in vitro. C. and. By contrast. TR␤R338D showed enhanced DNA binding in the absence of at CAPES . whereas TR␤D351R and TR␤K342D showed enhanced DNA binding in the presence or absence of T3. wild type. 4. 3A) or to a corepressor (N-CoR. 2B and Table II). Shown is a summary of relative EC50 values for T3 response obtained in transfections assays performed in HeLa cells with an F2-driven reporter gene as in Fig. By contrast. 4A confirmed that the TR␤R338W RTH mutant exhibits defective homodimer formation at F2 and DR-4 elements along with normal levels of heterodimer formation (30. Lys342 are required for optimal T3 binding and response and that Asp355 is not. Lys342. This is consistent with the results that show no impairment in the maximal effect of the hormone in transfection assays. This finding is consistent with previous observations that RTH mutations in these Arg residues only affect T3 sensitivity weakly (30) and is also consistent with the organization of Cluster 2 in TR␤ structures (Fig. parable to wild type TR␤ or enhanced T3 sensitivity at some reporters (Fig. None of the Cluster 1 mutations impaired binding to a coactivator (GRIP1. WT. comparison of TR␤ with TR␤R338W and a dimer surface mutant (L422R). Mutations in charge Cluster 1 inhibit or enhance homodimerization on DNA. and Asp355 does not (Fig. the charge reversal mutants exhibited enhanced DNA binding (Fig. Cluster 2 requires different charged residues for optimal T3 response and DNA binding just like Cluster 1. whereas TR␤R429A exhibited decreased homodimer formation on DNA (Fig. Mutations in Cluster 2 differentially affect T3 response and DNA binding. 4C) even though most of these mutations inhibit T3 binding (Fig. TR␤E311A (and TR␤R383H) bound to DNA as efficiently as wild type TRs in the absence of T3. The data in Fig. to a lesser extent. . Downloaded from www. Fig. A–C. comparison of TR␤ with Ala substitution mutants as indicated. 1C). Glu311 is required for optimal T3 response. Thus. 4B) just like the TR␤R338W RTH mutant. 1B). 2B).jbc. 4C). reversing the usual effects of T3 on TR DNA binding activity (Fig.. autoradiograph of gel shift assays of labeled F2 (A–C) and DR-4 (A and B) element oligonucleotides with wild type TR (TR␤wt) and various TR mutants in the presence or absence of both T3 and RXR. However. a mutation in Glu311 that breaks electrostatic interactions with both Arg residues exhibited a more severe defect than mutations at Arg429 and Arg383. wild type. 5B). and Asp351 side chains engage in electrostatic interactions with each other. B. 5A). 3. Shown are electrophoretic mobility shift assays to determine binding of Cluster 2 mutants to an F2 oligonucleotide as in Fig. This is consistent with the apparent organization of Cluster 1 in TR␤ crystal structures where Arg338. WT or wt. on November 6.05) according to ANOVA and Tukey’s test. Together our results indicate that Arg338. 31). reticulocyte lysate. Arg429 is required for optimal homodimer formation. More surprisingly.

31). This effect can only be observed.D351R charge reversal mutant (D/R). 3C. Mutations at Lys342 and Asp355 rescued effects of mutations at Arg338 and Asp351. Furthermore a TR␤ double mutant bearing Ala substitutions at residues that are not required for optimal T3 binding (TR␤K342A. Nevertheless Fig. DISCUSSION Downloaded from www. suggesting that they can form a reversible ionic bond that stabilizes liganded TR␤. Furthermore a double mutant that removed both negative charges (TR␤D351A. Ala substitution mutations at Lys342 and Asp355 restore activity of the TR␤R338D. Thus. and TR␤D355A. similar to wild type TR␤. Thus. A. In B.UNB. the largest obtained with any Cluster 1 mutation in this study (not shown).D351R double mutant was surprising. to TREs (30. D and E) and by our studies. An Arg338-Asp351 Pair Stabilizes Bound T3—To learn how individual residues within Cluster 1 interact. 6). K342I. The importance of the clusters is underscored by their conservation both in TRs across evolution (not shown) and in other NRs (Fig.D351R also displayed decreased affinity for T3 and increased dissociation rates of bound T3 with normal levels of coregulator binding and TR homodimer formation on DNA (not shown).D351A) displayed reduced T3 sensitivity and T3 binding and increased dissociation rates of bound at CAPES . Arg338 and Asp351 can be reversed in the absence of interfering charge at Lys342 and Asp355.K342A) exhibited a phenotype that was similar to wild type TR␤. 2012 FIG. slightly increased affinity for T3 (Table III). we asked whether Lys342 and Asp355 might also interfere with TR␤ activity and T3 binding in the context of wild type TR␤.K342A. Nevertheless Cluster 1 is required for activities associated with unliganded TRs: homodimer formation on DNA and transcriptional repression (see “Discussion”). when other charges are removed from the cluster. Data are presented as a comparison of EC50 values for different TR␤ mutants as in Fig. Ala substitution mutations at Lys342 and Asp355 rescue effects of similar mutations at Arg338 and Asp351. 6. The existence of functionally important clusters of charged residues on the TR LBD surface was surprising because proteins are largely stabilized by hydrophobic effects in . however. it is often possible to reverse the positions of residues in ionic pairs and regenerate wild type protein function. on November 6. Nevertheless TR␤4A exhibited strongly reduced homodimer formation on DNA (Fig. First we reversed the positions of two residues that are most important for optimal T3 binding. reduced affinity for T3. We reasoned that these mutations might affect structural elements that are involved in coupling T3 binding to inhibition of DNA binding activity. and Lys342 and Asp355 are not.D351R displayed markedly reduced T3 sensitivity in transfections like the R338D and D351R single mutants. This reduction in homodimer formation. a double mutant that eliminated both positive charges in Cluster 1 (TR␤R338A. to increases and decreases in T3 binding and/or DNA binding. Shown is a comparison of activities of wild type TR␤ and charge reversal mutants as a function of T3 concentration. we asked why some RTH mutations (R316H. Arg338 and Asp351. Together our results show that. but not heterodimers. Transfections were performed in HeLa cells with an F2-driven reporter. 6B shows that introduction of Ala substitutions at Lys342 and Asp355 restored the activity of the TR␤R338D.25670 Thyroid Hormone Modulation of DNA Binding Cluster 1 Is Dispensable for Ligand TR Activity—Because Lys342 and Asp355 interfere with TR␤ activity and T3 binding when the putative Arg338-Asp351 ionic bond is reversed (Fig. 7C). 1. 7B. We report here that each of these RTH mutations affected amino acids that lie within clusters of charged residues with potential for electrostatic interactions between individual residues in the cluster. and normal levels of coactivator and corepressor binding (not shown). we examined how TR DNA binding activity is regulated by its LBD and by ligand.05) according to ANOVA and Tukey’s test. different letters over bars indicate statistical difference (p Ͻ 0. The phenotype of the TR␤R338D. Two of these clusters (1 and 2) are adjacent to the TR dimer/heterodimer surface (Table I and Fig. variously. which revealed that mutations in Clusters 1 and 2 lead. 7A shows that a TR␤ double mutant bearing Ala substitutions at residues that are required for optimal T3 binding (TR␤R338A. These results confirm that Arg338 and Asp351 are needed for optimal hormone binding. To begin to understand this issue. Cluster 1 itself is dispensable for the function of liganded TR.D355A) did not affect TR␤ activity. wild type.jbc.D351R double mutant to near wild type levels but not that of a TR␤ mutant with like charges at both positions (TR␤R338D. 7B). R338W. More surprisingly. Reversibility of the putative Arg338-Asp351 salt bridge.D355A) exhibited a phenotype that was intermediate between TR␤D351A. whereas two individual residues in the cluster (Arg338 and Asp351) are required for optimal T3 response and T3 binding. inset). we examined effects of multiple Ala substitutions in Cluster 1. In this study. and R429Q) that reduce the affinity of TR␤ for T3 also inhibit binding of TR homodimers. B. was paralleled by impaired repression at a TRregulated reporter without T3 (Fig. Fig.D355A). we examined effects of multiple mutations in the cluster upon TR function. Arg338 and Asp351 cannot be reversed without severe disruption of T3 response. 6A shows that TR␤R338D. TR␤4A displayed enhanced T3 sensitivity in transfections (Fig. The fact that some mutations in Cluster 1 rescue effects of others was underscored by the observation that elimination of all charge within Cluster 1 with a quadruple Ala substitution (TR␤4A) failed to inhibit T3 binding or liganded TR␤ function. Fig. wt. TR␤R338D. Arg338 and Asp351 can be reversed without severe loss of TR␤ function. 1). To do this.

one in Cluster 3 (TR␤R316H) and a single surface-exposed ionic bond between Arg243 in H3 and Glu322 at the base of H6 (TR␤R243Q). mutations at Lys342 and Asp355 rescue effects of mutations at Arg338 and Asp351. 2. freely solvated with water (37).jbc.1 Ϯ 23. on November 6. Our mutational analysis supports the notion that Clusters 1 and 2 are stabilizing elements for liganded TR.32 Ϯ 0. representative of several experiments. two RTH mutations that disrupt ionic bonds. C. Furthermore and more strikingly. WT or wt. and koff values obtained with TR mutants with single and double Ala substitutions in Cluster 1. and a charge reversal mutation at Asp355 (D355R) did not affect T3 binding yet enhanced TR homodimer formation on DNA in the presence of T3 (Fig. Shown is a gel shift comparing binding of TR␤ and TR␤4A (4xAla) on an F2-TRE. Kd ϫ10 Ϫ12 M a koff ϫ10 minϪ1 3 TR␤ TR␤4A 134. 15). the clusters adopt an unspecified organi- .6 67. and Lys342 (K342D) all inhibited T3 binding but not DNA binding. Furthermore two Cluster 1 residues (Lys342 and Asp355) must inhibit T3 binding to some extent as judged by the fact that TR␤K342A and TR␤D355A mutants exhibited enhanced sensitivity to T3 in transfections and increased affinity for T3 in vitro and that Ala substitutions at both positions rescued effects of similar mutations at Arg338 and Asp351 (Fig. Nevertheless our results also suggest that the clusters adopt a different organization in unliganded TRs. R316H and R243Q (14. The inset shows maximal (Max. suggesting that both residues are parts of the same structural element (Figs. charge reversal mutations at Arg338 (R338D). 3) TRs with double mutations at Arg338 and Asp351 exhibited phenotypes similar to single mutants. different letters over bars indicate statistical difference (p Ͻ 0. elimination of Cluster 1 inhibits DNA binding. This confirms that electrostatic interactions between oppositely charged TR␤ residues can stabilize the liganded TR␤-LBD. Cluster 1 is dispensable for liganded TR action. are the average of at least three experiments. and 7). 8.05) according to ANOVA and Tukey’s test. repelling charges at Arg338 and Asp351 severely inhibited T3 binding (Fig.72a 1. 2–5).5 Ϯ 44. Other results are hard to reconcile with the simple notion that Clusters 1 and 2 act as static stabilizing elements for liganded and unliganded TRs. 15). In addition. protein-protein interaction surfaces. A. Nevertheless electrostatic interactions between oppositely charged side chains have been shown to provide additional stability to proteins in several contexts. 2 and 3. 2) Arg338 and Asp351 could be reversed without significant disruption of T3 binding. Downloaded from www. B. reduced affinity for T3. In A.05) according to t test.92 Ϯ 0. 6 and 7). 4). data are presented as in Figs. 7.UNB. Shown is a comparison of EC50. and increased T3 dissociation rates. The data represent a single transfection assay in which standard errors are derived from multiple wells. 2). Mutations that disrupted the predicted ionic bond arrangements in Clusters 1 and 2 led to reduced T3 sensitivity. at CAPES . For TRs. We propose that Clusters 1 and 2 are hormone-dependent stabilizing elements for the TR LBD. We suggest that. including particular conformers of allosteric proteins. 6).D. and proteins in thermophilic organisms (37– 40).) activation and repression obtained with TR␤ and TR␤4A (4xA).6a 2. three lines of evidence indicate that Arg338 and Asp351 form an ionic bond required for stable T3 binding. Our hypothesis to explain these observations is outlined in Fig. in the unliganded state. the TR␤K342A mutation inhibited DNA but not T3 binding. Shown are hormone activation profiles for human TR␤ wild type and a quadruple Ala mutant (4xA) at an F2-TRE-regulated reporter. 1) Placement of like. 7). 3. Cluster 1 was dispensable for optimal T3 response and T3 binding (Fig. albeit only in the absence of charge at Lys342 and Asp355 (Fig. The same letters in the same column indicate no statistical difference (p Ͼ 0. wild type. These phenotypes resemble those of aforementioned TR␤ RTH mutations that destabilize the TR LBD by breaking electrostatic interactions. Cluster 1 can be eliminated without loss of function for liganded TR. Asp351 (D351R). 2012 TABLE III Charge Cluster 1 is dispensable for T3 binding Means Ϯ S. lead to broadening of experimental electron density in the lower part of the LBD in x-ray structures (14. 7) even though Arg338 and Asp351 were required for T3 binding (Figs.Thyroid Hormone Modulation of DNA Binding 25671 FIG. Distinct arrangements of charge are required for optimal T3 binding and for DNA binding by unliganded TR homodimers (Figs.56a which hydrophobic residues form the interior of the protein and charged side chains are surface-exposed. Kd.

Nilsson.. Natl. J. 81. M. 1747–1749 18. Cunha-Lima. J. Staszewski. Fletterick. (1999) Annu. D. Zhang. A. S. Mellstrom. J. P. W. G.. Mol... (2001) Mol. E.. Shiau... Baxter.. M. and Fletterick. J.. M. R.... 31– 42 7. P. J. J. Liang.. First charged residues in the region of RXR that is equivalent to H8 rearrange in response to ligand binding (Fig. West. Baxter. R. J.. J. H. J. Biol. C. R. Scanlan. (1995) Nature 378. D. L... S.. L.. Scanlan. and Schwabe... B. S.. J. J. Wang. Ribeiro.. paired in apoRXR. Endocrinol. P. and Rosenfeld.. S.. P. W. P. P. we favor the notion that the clusters comprise homodimer-specific extensions of the dimer surface that engage in flexible contacts with oppositely charged residues in partner LBDs or possibly influence homodimer formation via distant stabilizing effects on the dimer surface. Weatherman.. . Endocrinol. D. L. J. Budny. West. F. Kushner.. Furlow. Cunha Lima.. R. B. Z. Baxter. R.. J. whereas red spheres represent negatively charged residues. Steroid Biochem. J. Milburn. Zhang. 16. Baxter. F. R. (2002) J. Kushner.UNB. S. 13. 398 – 410 14. Chatterjee.. Perissi. L. (1998) Clin. 17.. S2–S11 2. Model to explain coupling of T3 binding to inhibition of TR␤ DNA binding activity. West. T. 690 – 697 13.. Ribeiro. Kindblom. Lambert. T. J.. W. J. Mol. D... A. Kushner. A. 6.. E. Cohen. R. Apriletti. We further suggest that T3 binding promotes structural rearrangements in the TR LBD that reposition the charged residues. 1097–1142 4. B. 93–96 20. B.. Feng. L. Yoshihara. D.. M.. D. S. R. West.. M. Togashi. J.. Marimuthu. L.. R. R. and Scanlan. Ribeiro.. T. Yen. J. Mol. B. J. M.. F. Nagy.. J. Wagner... Kushner. J. B. Nguyen. K.. Jameson.. A schematic shows apoRXR H7 in orange and holoRXR in gray with ligand (9-cis-retinoic acid (RA)) in gray and red. T. R. Scanlan. K. Fletterick. H.. B. V. L. Kushner. T. and Fletterick. J. P. 83. Fletterick. H7 changes backbone conformation: Lys356 and Glu352. 25. 559 –581 12. D. L.25672 Thyroid Hormone Modulation of DNA Binding We recognize that our model cannot yet be verified directly because apoTR dimer structures are not available. B. A. J. Feng. B. E. (1998) Science 280. Scanlan. 53. Physiol. Ribeiro. Acad. Huber. exposing the charged side chains on the protein surface where they can pair with PPARs in RXR-PPAR heterodimers (41– 45). J.. T. 439 – 466 5. and Scanlan. D. E. Rev. Webb. W. (2003) J. T. Nilsson.. R. W. Apriletti. C. on November 6. C. 14. R. J. R. move from inside to outside on ligand binding. L. and Baxter. 121–141 6. Apriletti. H.. Marimuthu. J.. M.. K.. Glass. J. Endocrinol. Huber. S. Wagner. P. J. J. 13.. and H11) must undergo similar ligand-dependent rearrangements... 643– 652 16... and West. (1999) Genes Dev. R. Ligand-dependent rearrangements in RXR H7 (equivalent to TR␤ H8). Desclozeaux. Rev. (2000) Genes Dev. J. R. 107–116 15.. (2000) Annu. (2001) J. 100. (1998) Recent Prog.. X.... and Lazar.. Baxter. Y. G. J.... J.. Investig. S.. S. (1992) Mol.. S. Fletterick. Wondisford. Kim.. and Gardner. G. L. S. S. and Fletterick. P. R.. L. Feng. W. and Fletterick.... The charged residues are in an unspecified organization required for homodimer formation in the absence of hormone and rearrange to form the ionic bond organization detected in our x-ray structures in the presence of hormone. M. Dillmann. K. Apriletti. B. Tagami. T. A. R. B... Fletterick. T. H. Binding of 9-cisretinoic acid twists the helix. S... Apriletti. D.. Ingraham. Scanlan. Kelly. 3) Lys342 and Asp355 inhibited T3 binding because they stabilize the unliganded TR␤ conformer that binds to DNA as a homodimer but only provide limited stability (Lys342) or no additional stability (Asp355) to the liganded TR␤ conformer. 15. and Baxter.. L. Kushner. Borngraeber. G. Nguyen.. W. H. Chiellini. J. (1998) J. (2003) Mol. T. Finally our results also lend support to the notion that TR homodimers are highly active in mediating transcriptional repression in vivo (25... Baxter. V. J. R. (1999) Genes Dev. R. Given the placement of these residues. W. Webb. R. Here mutation of Lys342 and Asp355 enhanced T3 response and T3 binding by counteracting the tendency of the cluster to revert to its organization in unliganded state.. 1) Different charge arrangements are required for T3 binding and homodimer formation on TREs because the clusters adopt different organizations in the presence and absence of T3. L. W. Huber.. Baxter. P. 1049 –1061 24. W.. H. (1999) Genes Dev. W. Biol. Endocrinol. 68. Nilsson.. Exp. Hu. Biochem. C. C. A. Apriletti.. T. Wagner.. J.. and Lazar. Downloaded from www. Webb. Steroid Biochem. Schaufele. Steroid Biochem. S. S. R. simultaneously breaking the interactions that are needed for optimal homodimer formation on DNA and creating new ionic bonds that hold the walls of the hormone binding pocket in an appropriate configuration for stable T3 interactions.. R... D.. 15.. K. H. 2012 FIG. and Baxter. W. A... and Rosenfeld. D. A. M.. S.. M.... S.. L. 62. B.. D. We observed that a TR␤ mutant that strongly inhibited homodimer formation on TREs (TR␤4A) impaired the ability of unliganded TRs to suppress transcription in the absence of hormone (Fig. N. L. M. Ribeiro. West. J. H. J. P.. A. R.. H8... Webb.. Sandler.. Scanlan. B.. Apriletti. 8. E.. C.jbc. the requirement for the stabilizing element is eliminated. at CAPES . H. We suggest that TR␤ charge clusters (on H7. 1142–1152 23. W. 51. R.. D. T. A. L. C.. L. Huber. Nguyen. S. Wondisford. West. Wagner. R.. J. 17. J. M. (2003) FIG. Rev. and Hollenberg. W. Fletterick. West. J. W. Apriletti.. A. J. J. Huber. 13. 3209 –3216 21. J... Marimuthu. Krones. L. A. M. R. Feng. T. Webb. J. R. B. B. Banayo. M. Cunha-Lima. Cunha Lima. B. Wagner. Vennstrom. 9. S. S. Krishna. Suppl. and Baxter.. (2003) Mol. 26). B. (1999) Nature 402. Fletterick. Apriletti. C. Baxter. Ribeiro. C. and Fletterick.. Feng. V. R. The blue spheres represent positively charged residues in Cluster 1. Med.. RXR Glu352 and Lys356 form an ionic bond within the interior of the unliganded LBD.. Gooch. H. J. 7B). We predict that mutations such as those described here that either specifically inhibit or stabilize particular oligomeric forms of TR will help us to further dissect the relative roles of RXR-TR heterodimers and TR homodimers in vivo. Love. C. 1329 –1341 9. Pharmacol. Li.. 65.. Gothe. (2002) Mol. J. J. Wagner. T. 76. Our model suggests explanations for several apparently paradoxical results... Evans. T. C. W. Kao. M. Barros. J. D. Nilsson. If Cluster 1 is eliminated. Apriletti. D. West. J. M.. Sci. V. J. D. and Forrest. D.. 9). Chiellini. J. L.. P.. 215–220 10. 15358 –15363 17. Goede. B. Rose. 271–286 19. Cunha Lima. N. Chorev. R. Horm.. W. T.. and West. McGrath. D. Ohlsson. 133–141 11. Nguyen.. R. S. L. A.. West. McInerney.. zation that is distinct from that observed in x-ray structures of liganded TR␤ LBDs but required for optimal homodimer formation on DNA. J. 351–394 3. (2001) Mol... S. Res. Ng. W. L. Physiol. U... Biol. (2001) Physiol.. R. Endocrinol. L. Endocrinol. 2) Cluster 1 could be eliminated without obvious effect on TR␤ even though mutations in Arg338 and Asp351 inhibit T3 binding because the Arg338-Asp351 ionic bond counteracts the tendency of the Cluster 1 to revert toward the organization in the unliganded state. Brzostek. Glass. J... T. N. P. This model implies that functionally important conformational rearrangements that accompany T3 binding are not restricted to H12 and that T3 induces reorganization of the opposite face of the TR near the dimer surface. F. 59 –73 8. R. Fletterick.. (2003) Proc. Kurokawa. REFERENCES 1. R. 3198 –3208 22. Nguyen.. Nevertheless analysis of liganded and unliganded RXR crystal structures revealed evidence that is consistent with the basic predictions of our model. G. and Baxter.

M. M.. at CAPES . Fletterick. Serrano. T.. Mikami. S. (1996) Proc. R... Bledsoe. Ruff. Fletterick.. T.. 14987–14995 28. M. N. Francois. E. A.. L. W.. Van Helvoirt... W. Gronemeyer. and DeGroot. Wagner. G. F. Montana. D. Zavacki. P. Feng. Gronemeyer. T... Natsume. Y. R. T. G. V.. M. Milburn. Strop. S. Endocrinol.. R... H... R. Ando. K. H. V. S. Wagner. V. S. Apriletti. Miller. G. Lambert. Collingwood. Endocrinol. Chem.. Soc. 297–308 25673 35. J.. 14. J. A.. Ruff.. P. 276. T. Kliewer. (1990) Biochemistry 29. R. Chin... K. de Zegher. A. H. Chambon. Vivat. (1994) Endocrinology 134. Perutz. R. (2000) Mol... K. Biol. J.. J. 278. W. Harney.. 2229 –2241 Downloaded from www. Nagaya.. Williams.. L. (1995) Nature 378. (1996) J. Bourguet.. P. R. and Moras.. A.. Nakamura. D. Nagasawa. T... Gampe.. 42– 44 39. G..jbc. B.. (1999) Biochimie (Paris) 81. Bourguet. G. M.. and Baxter. Takeda. J. T. 151. W. 343–353 32.. D. R. J. 293–300 34. (1998) Mol. Kitajima. M. H. J. C. and Xu. M. V. M. Montana. Miyamoto. L. (2000) Biochemistry 39. Nishiyama. V.. K. M. 1262–1277 31.. Yoh. Gampe... Chem. Suzuki. 16857–16867 27. M. D. Jr.. (1994) Endocrinology 135. Biol. and Xu. S. A. 14. W... R. 681– 689 43.UNB.. Endocrinol. T. Jr. Renaud. Ribeiro. (1995) Nature 375. 545–555 44. W. C. H. Genma. H. Chatterjee. and Baxter. J. V. Ribeiro. L. and Chatterjee. K. J. M. 9343–9352 40. Biol.. B. Kopp. Exp. J. M. S. 211. and Degroot. Horovitz. M. Milburn. Collingwood. Nagasawa. M. Yen. D. K. Takeda. 2012 . Endocrinol.. A. and Beck-Peccoz. and Chatterjee.. T. 276. 377–382 42.. J... 1187–1191 38. and Moras. (1994) Mol.. R. W.. V. Minemura. L. Hashizume. Tone... E.. Adams. 8. (1994) Bailliere’s Clin. C. Chambon.Thyroid Hormone Modulation of DNA Binding J. 11–18 36. M. Sci. Rochel.. H. (1978) Science 201. (2001) J. H. W. Liu. (2000) Genes Dev. L. 609 – 621 37. 1251–1255 41. F. C.. P. and Brent. 8. K. 15073–15083 25. Sasaki.. P. and Jameson. Chambon. A.. 49 – 61 30. and Moras. J. Biol. 12. J. and Yoshimi.. R. Wisely. (1989) Trends Biochem.. Wurtz. (2000) Mol. T. I. Gronemeyer.. P. H. Cell 5. T.. L. P. (2001) J. M. Kitahara. Med. F... V. N. Willson. R.. T. T. H. Avron... S. 267–283 29. Bycroft. 289 –298 45. K. Pereira. M. Clifton-Bligh. V. Apriletti. Metab. L. L. P. (2000) Thyroid 10. Vivat. and Mayo.... and Privalsky. Kitajima. T. Costa... (1995) Thyroid 5. A. Cell 5. Perutz. J.. B. Nagasawa. 2076 –2085 33. and Fersht. R. N.. and Jameson... on November 6. 1888 –1896 26.. K. Lambert. Chem..

Sign up to vote on this title
UsefulNot useful