Hammett analysis of selective thyroid hormone receptor modulators reveals structural and electronic requirements for hormone antagonists

Article abstract of DOI:10.1021/ja0440093

Selective thyroid hormone modulators that function as isoform-selective agonists or antagonists of the thyroid hormone receptors (TRs) might be therapeutically useful in diseases associated with aberrant hormone signaling. The most potent thyroid hormone

Full text of DOI:10.1021/ja0440093

Published on Web 03/11/2005
Hammett Analysis of Selective Thyroid Hormone Receptor
Modulators Reveals Structural and Electronic Requirements
for Hormone Antagonists
Ngoc-Ha Nguyen,†,‡ James W. Apriletti,§ John D. Baxter,§ and
Thomas S. Scanlan*,‡
Contribution from the Program in Chemistry and Chemical Biology, Departments of
Pharmaceutical Chemistry and Cellular and Molecular Pharmacology, UniVersity of California,
San Francisco, California 94143-0446, and Diabetes Center and Metabolic Research Unit,
UniVersity of California, San Francisco, California 94143-0540
Received October 1, 2004; E-mail: scanlan@cgl.ucsf.edu
Abstract: Selective thyroid hormone modulators that function as isoform-selective agonists or antagonists
of the thyroid hormone receptors (TRs) might be therapeutically useful in diseases associated with aberrant
hormone signaling. The most potent thyroid hormone antagonist reported to date is NH-3. To explore the
significance of the 5′-p-nitroaryl moiety of NH-3 and understand what chemical features are important to
confer antagonism, we sought to expand the structure-activity relationship data for the class of
5′-phenylethynyl GC-1 derivatives. Herein, we describe an improved synthetic route utilizing palladium-
catalyzed chemistry for efficient access to a series of 5′-phenylethynyl compounds with varying size and
electronic properties. We prepared and tested sixteen analogues for TR binding and transactivation activity.
Substitution at the 5′-position decreased binding affinity, but retained TRâ-selectivity. In transactivation
assays, the analogues displayed a spectrum of agonist, antagonist, and mixed agonist/antagonist activity
that correlated with electronic character in a Hammett analysis between σ substituent value and TR
modulation. Analogues NH-5, NH-7, NH-9, NH-11, and NH-23 displayed full antagonist activity with reduced
potency compared to NH-3, indicating the nitro group is not required for antagonism. However,
para-substitution with strong electron withdrawing properties on the 5′-aryl extension is important for
antagonist activity, and antagonist potencysbut not ligand receptor bindingswas found to correlate linearly
with the sigma values for the electron withdrawing substituents.
Introduction
Thyroid hormones regulate a multitude of physiologic effects
ranging from embryonic development to maintenance of ho-
meostasis in adults.1-4 Thyroxine (T4) is the major form released
by the thyroid gland (Figure 1). A lesser amount of 3,5,3′-
triiodothyronine (T3) is also released, but the bulk amount is
produced by deiodination of T4 to T3 in peripheral tissues. T3
appears to be the major active form of the hormone in thyroid
target tissues. Its action is primarily mediated through the nuclear
thyroid hormone receptors (TRs) that regulate transcription of
target genes either positively or negatively in response to
hormone binding. There are two TR subtypes (R and â) encoded
† Program in Chemistry and Chemical Biology.
‡ Departments of Pharmaceutical Chemistry and Cellular and Molecular
Pharmacology.
§ Diabetes Center and Metabolic Research Unit.
(1) Nguyen, N. H.; Apriletti, J. W.; Cunha Lima, S. T.; Webb, P.; Baxter, J.
D.; et al. Rational design and synthesis of a novel thyroid hormone
antagonist that blocks coactivator recruitment. J. Med. Chem. 2002, 45,
3310-3320.
Figure 1. Structures of ligands for the thyroid hormone receptors (TR).
T4 and T3 are natural thyroid hormones. GC-1 is a thyromimetic while
GC-14 and NH-3 are antagonists.
(2) Chiellini, G.; Nguyen, N. H.; Yoshihara, H. A.; Scanlan, T. S. Improved
synthesis of the iodine-free thyromimetic GC-1. Bioorg. Med. Chem. Lett.
2000, 10, 2607-2611.
(3) Greenspan, F. S. The Thyroid Gland. Basic & Clinical Endocrinology, 5th
ed.; Appleton & Lange: Stamford, 1997; pp 192-262.
(4) Utiger, L. E. B.; Utiger, R. D. The Thyroid: a Fundamental and Clinical
Text; Lippencott: Philadelphia, 1991.
on separate genes, each having two additional isoforms (TRR1,
TRR2, TRâ1, TRâ2) due to differential splicing.5 Although most
9
10.1021/ja0440093 CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 4599-4608
4599

A R T I C L E S
Nguyen et al.
of these isoforms are widely expressed, there are distinct patterns
of expression that vary with tissue and developmental stage. In
particular, TRâ2 is almost exclusively expressed in the hypo-
thalamus, anterior pituitary, and developing ear. Genes positively
regulated by T3 contain a cis-acting thyroid hormone response
element (TRE) upstream of the promoter region. Unliganded
TRs are bound to the TRE, typically as heterodimers with the
retinoid X receptor (RXR), and are associated with a group of
corepressor proteins to repress the basal transcriptional machin-
ery. Binding of hormone induces TR release of corepressors
and subsequent recruitment of coactivator proteins to enhance
TRE-driven transcriptional activity.6
The crystal structures of several nuclear receptor (NR) ligand
binding domains (LBDs) in the unliganded and liganded (agonist
or antagonist) states suggest a common mode of ligand-regulated
activation and inhibition of the nuclear receptor superfamily.7,8
Binding of an agonist ligand induces rearrangement of the LBD,
most dramatically in the C-terminal helix 12 (H12). H12 acts
as a lid over the ligand binding pocket and contributes to the
formation of a hydrophobic cleft at the receptor surface
accessible for coactivator binding.9-11 Conversely, binding of
an antagonist ligand induces an inactive receptor conformation
by preventing proper packing of H12 to complete the coactivator
binding cleft.12 The crystal structures of the TR LBD bound to
agonists such as T3 and GC-1 have been solved.13,14 However,
no structure of the TR LBD bound to an antagonist is currently
available. Development of potent T3 antagonists is still in its
infancy. To date, most reported T3 antagonists have moderate
to weak potency with IC50 values in the high nanomolar to
micromolar range in cell culture assays, thereby limiting their
characterization in animal models and potential therapeutic
utility.15-19 However, the T3 antagonist NH-3 was the first to
demonstrate potent inhibition of T3 action in both cell culture
and whole animal-based assays.1,20
We have previously shown that the rigid ethynyl moiety of
NH-3 gave improved antagonist efficacy and potency compared
to the parent compound GC-14.1,19 However, the contribution
of the 5′-p-nitroaryl group to antagonist activity is unknown.
To better understand the molecular basis of the antagonistic
activity of NH-3 and the significance of the 5′-p-nitroaryl
pharmacophore, we sought to expand the structure-activity
relationship (SAR) data for the 5′-phenylethynyl series of GC-1
derivatives. We hypothesize that the nitro group is not required
for antagonism and that the electronic nature of the 5′-aryl
extension will dictate a spectrum of agonist versus antagonist
activity. That is, extensions containing electron-donating groups
(EDG) will productively interact with receptor residues to
stabilize an active receptor conformation, while extensions
containing electron-withdrawing groups (EWG) will stabilize
an inactive receptor conformation. Herein, we describe the
synthesis of sixteen 5′-phenylethynyl GC-1 derivatives having
variable electronic properties. We show that the analogues bind
TR with moderate nanomolar affinity and TRâ selectivity,
exhibit selective TR modulation that indeed correlates with
electronic character, and function similar to NH-3 with respect
to ligand-induced TR interaction with coactivators and core-
pressors to neutralize TR transcriptional activity.
Results
Chemical Synthesis. The synthesis previously described for
the 5′-phenylethynyl GC-1 derivatives using the palladium-
catalyzed Suzuki-Miyaura coupling1 was modified to improve
the 5′-iodination reaction to generate the key intermediate 4
(Scheme 1A). The starting GC-1 biarylmethane intermediate 1
was treated with tetrabutylammonium fluoride followed by
alkylation with methyl 2-bromoacetate to generate the 1-oxy-
acetic acid side chain protected as the methyl ester 2. Acidic
hydrolysis of the methoxymethyl (MOM)-protected phenol
allowed efficient iodination at the 5′-position with iodine
monochloride. Reprotection of the phenol 3 as the methoxy-
methyl ether gave intermediate 4, with an overall yield of 51%
from 1. This new route allowed preparation of 4 in multigram
quantities and offered more efficient access to 5′-phenylethynyl
compounds.
Subsequent palladium-catalyzed Suzuki-Miyaura coupling21
of 4 with phenylethynyl boronate derivatives, generated in situ
under basic conditions with MeO-9-BBN, produced 5′-phenyl-
ethynyl analogues in good yields (Scheme 1B). Many of the
starting phenylacetylenic compounds are commercially avail-
able; others were generated via Sonogashira22 or Sandmeyer23,24
(5) Lazar, M. A. Thyroid hormone receptors: multiple forms, multiple
possibilities. Endocr. ReV. 1993, 14, 184-193.
(6) Zhang, J.; Lazar, M. A. The mechanism of action of thyroid hormones.
Annu. ReV. Physiol. 2000, 62, 439-466.
(7) Weatherman, R. V.; Fletterick, R. J.; Scanlan, T. S. Nuclear-receptor ligands
and ligand-binding domains. Annu. ReV. Biochem. 1999, 68, 559-581.
(8) Ribeiro, R. C.; Apriletti, J. W.; Wagner, R. L.; West, B. L.; Feng, W.; et
al. Mechanisms of thyroid hormone action: insights from X-ray crystal-
lographic and functional studies. Recent Prog. Horm. Res. 1998, 53, 351-
392.
(9) Feng, W.; Ribeiro, R. C.; Wagner, R. L.; Nguyen, H.; Apriletti, J. W.; et
al. Hormone-dependent coactivator binding to a hydrophobic cleft on
nuclear receptors. Science 1998, 280, 1747-1749.
(10) Darimont, B. D.; Wagner, R. L.; Apriletti, J. W.; Stallcup, M. R.; Kushner,
P. J.; et al. Structure and specificity of nuclear receptor-coactivator
interactions. Genes and DeVelopment 1998, 12, 3343-3356.
(11) Fondell, J. D.; Ge, H.; Roeder, R. G. Ligand induction of a transcriptionally
active thyroid hormone receptor coactivator complex. Proc. Natl. Acad.
Sci. U.S.A. 1996, 93, 8329-8333.
(12) Schapira, M.; Raaka, B. M.; Samuels, H. H.; Abagyan, R. Rational discovery
of novel nuclear hormone receptor antagonists. Proc. Natl. Acad. Sci. U.S.A.
2000, 97, 1008-1013.
(19) Chiellini, G.; Nguyen, N. H.; Apriletti, J. W.; Baxter, J. D.; Scanlan, T. S.
Synthesis and biological activity of novel thyroid hormone analogues: 5′-
aryl substituted GC-1 derivatives. Bioorg. Med. Chem. 2002, 10, 333-
346.
(20) Lim, W.; Nguyen, N. H.; Yang, H. Y.; Scanlan, T. S.; Furlow, J. D. A
thyroid hormone antagonist that inhibits thyroid hormone action in vivo.
J. Biol. Chem. 2002, 277, 35664-35670.
(21) Soderquist, J. A.; Matos, K.; Rane, A.; Ramos, J. Alkynylboranes in the
Suzuki-Miyaura Coupling. Tetrahedron Lett. 1995, 36, 2401-2402.
(22) Takahashi, S.; Kuroyama, Y.; Sonogashira, K.; Hagihara, N. Convenient
synthesis of ethynylarenes and diethynylarenes. Synthesis Comm. 1980,
627-629.
(23) Suzuki, N.; Kaneko, Y.; Nomoto, T.; Izawa, Y. Synthetic reactions in poly-
(ethylene glycol). Diazotization and Sandmeyer Reactions of anilines in
poly(ethylene glycol)-methylene dichloride. J. Chem. Soc. Chem. Commun.
1984, 1523-1524.
(24) Suzuki, N.; Azuma, T.; Kaneko, Y.; Izawa, Y.; Tomioka, H. Diazotization
and Sandmeyer Reactions of arylamines in poly(ethylene glycol)-methylene
dichloride: usefulness of PEG in synthetic reactions. J. Chem. Soc., Perkin
Trans. 1 1987, 645-647.
(13) Wagner, R. L.; Apriletti, J. W.; McGrath, M. E.; West, B. L.; Baxter, J.
D.; et al. A structural role for hormone in the thyroid hormone receptor.
Nature 1995, 378, 690-697.
(14) Wagner, R. L.; Huber, B. R.; Shiau, A. K.; Kelly, A.; Cunha Lima, S. T.;
et al. Hormone selectivity in thyroid hormone receptors. Mol. Endocrinol.
2001, 15, 398-410.
(15) Yoshihara, H. A.; Apriletti, J. W.; Baxter, J. D.; Scanlan, T. S. A designed
antagonist of the thyroid hormone receptor. Bioorg. Med. Chem. Lett. 2001,
11, 2821-2825.
(16) Webb, P.; Nguyen, N. H.; Chiellini, G.; Yoshihara, H. A.; Cunha Lima, S.
T.; et al. Design of thyroid hormone receptor antagonists from first
principles. J. Steroid. Biochem. Mol. Biol. 2002, 83, 59-73.
(17) Schapira, M.; Raaka, B. M.; Das, S.; Fan, L.; Totrov, M.; et al. Discovery
of diverse thyroid hormone receptor antagonists by high-throughput docking.
Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 7354-7359.
(18) Baxter, J. D.; Goede, P.; Apriletti, J. W.; West, B. L.; Feng, W.; et al.
Structure-based design and synthesis of a thyroid hormone receptor (TR)
antagonist. Endocrinology 2002, 143, 517-524.
9
4600 J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005

Selective Thyroid Hormone Receptor Modulators
A R T I C L E S
Scheme 1. General Synthetic Route to 5′-phenylethynyl GC-1 Derivatives via the Suzuki-Miyaura Couplinga
a (A) Improved synthesis of the 5′-iodinated key intermediate 4. (B) Suzuki-Miyaura coupling of 4 with phenylacetylenic derivatives followed by deprotection
afforded the desired compounds 6a-k. Most phenylacetylenic derivatives were commercially available; others were readily synthesized via standard procedures
as described in the Supporting Information.
conditions. The resulting coupled products 5a-k were then
subjected to acidic hydrolysis of the methoxymethyl phenolic
protecting group followed by basic saponification of the methyl
ester to afford the desired 5′-phenylethynyl GC-1 analogues
6a-k as outlined in Scheme 1B.
Some analogues required additional chemical manipulations
after palladium coupling and/or after deprotection steps. The
5′-p-azidophenylethynyl analogue NH-7 was prepared from the
aniline precursor 5l (Scheme 1B) using Sandmeyer25 conditions
of aqueous sodium nitrite in acid and sodium azide (Scheme
2A). Subsequent deprotection of the MOM group and the methyl
ester gave the final compound NH-7 in high yield.
tected intermediate 7 by oxidation with m-chloroperbenzoic acid
(m-CPBA) in chloroform27 followed by basic hydrolysis of the
methyl ester providing the desired compound in good yield
(Scheme 2C). Attempts to oxidize intermediate 5k (Scheme 1B)
to form the 5′-o-nitrophenylethynyl analogue unexpectedly
resulted in degradation.
The synthesis of analogue NH-9 started with the coupling of
4 with 1-bromo-4-ethynylbenzene to give intermediate 5m
(Scheme 1B). However, the Suzuki-Miyaura coupling did not
go to completion and the coupled product was an inseparable
mixture with the aryl iodide 4. This mixture was carried through
another palladium-catalyzed coupling for the synthesis of
nitroalkane 8, adopting the procedure described by Vogl and
Buchwald28 (Scheme 3A). The reaction yield for this coupling
was low with multiple byproducts, which likely was due to use
of impure bromide 5m. The MOM protecting group was
Analogue NH-16 was obtained by methylation of the 5′-p-
(dimethylamino)-phenylethynyl intermediate 5g with methyl
trifluoromethanesulfonate in refluxing dichloromethane26 (Scheme
2B). NH-16 was not hydrolyzed to the carboxylic acid due to
solubility issues in the workup and isolation of the final
compound. Analogue NH-19 was prepared from MOM-depro-
(26) Oya, S.; Choi, S. R.; Coenen, H.; Kung, H. F. New PET imaging agent for
the serotonin transporter: [(18)F]ACF (2-[(2-amino-4-chloro-5-fluorophen-
yl)thio]-N,N-dimethyl-benzenmethanamine). J. Med. Chem. 2002, 45,
4716-4723.
(27) Craig, J. C.; Purushothaman, K. K. An improved preparation of tertiary
amine N-oxides. J. Org. Chem. 1970, 35, 1721-1722.
(25) Tanno, M.; Sueyoshi, S.; Kamiya, S. Synthesis of arylcyanotriazenes and
related compounds. Chem. Pharm. Bull. 1982, 30, 3125-3132.
9
J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005 4601

A R T I C L E S
Nguyen et al.
Scheme 2. Synthesis of NH-7, NH-16, and NH-19
Scheme 3. Synthesis of NH-9 and NH-23
removed under acidic conditions. Upon treatment with aqueous
lithium hydroxide in methanol for deprotection of the methyl
ester, the p-nitroethane group underwent hydrolysis in a Nef
reaction29 after acidic workup to give the ketone product NH-9
(Scheme 3A).
Scheme 3B. De-silylation with tetrabutylammonium fluoride
generated the free benzyl alcohol, which was readily converted
to the benzyl azide 9 via formation of the benzyl chloride in a
one-pot synthesis with chlorotrimethylsilane in DMSO30 and
sodium azide.31 Subsequent cleavage of the MOM group and
the methyl ester afforded NH-23.
Under the basic conditions of the final saponification step, a
number of 5′-phenylethynyl derivatives underwent cyclization
of the o-alkynylphenolic moiety to form the benzofuran product.
Analogue NH-23 was synthesized from the triisopropylsilyl
(TiPs) protected benzyl alcohol 5n (Scheme 1B) as outlined in
(28) Vogl, E. M.; Buchwald, S. L. Palladium-catalyzed monoarylation of
nitroalkanes. J. Org. Chem. 2002, 67, 106-111.
(29) Smith, M. B.; March, J. Hydrolysis of Aliphatic Nitro Compounds. March’s
AdVanced Organic Chemistry, 5th ed.; Wiley-Interscience: New York,
2001; pp 1178-1179.
(30) Snyder, D. C. Conversion of alcohols to chlorides by TMSCl and DMSO.
J. Org. Chem. 1994, 60, 2638-2639.
9
4602 J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005

Selective Thyroid Hormone Receptor Modulators
A R T I C L E S
Table 1. Binding and Transcriptional Activation Data of 5′-phenylethynyl Derivatives at Human TRR1 and TRâ1
a The KD and standard error (SE) values are expressed relative to the KD of T3 and were calculated by fitting the competition data to the equations of
Swillens32 and using the Graph-Pad Prism computer program (Graph-Pad Software, Inc.). b Luciferase activity of 10-5 M analogue is expressed as a percent
of the TRâ1 or TRR1 response with 10-6 M T3. Values are the mean ( SD for three separate experiments. See Supporting Information for more details.c The
EC50 value is the concentration of ligand required for half-maximum activation, whereas IC50 value is the concentration of ligand required for half-maximum
inhibition in competition experiments with 2 × 10-9 M T3. EC50 and IC50 values were calculated by nonlinear regression with the Graph-Pad Prism computer
program (Graph-Pad Software, Inc.) using a sigmoidal dose-response and single-site competition models, respectively. Values reported are the mean for
three separate experiments with R2 fit of at least 0.85 unless value is expressed as greater than (“>”), indicating a poor R2 fit. d Refer to Chiellini et al.19
e Compounds exhibiting thyromimetic transcriptional activation through hTRâ1 were not further characterized with hTRR1. f For transactivation assays,
HeLa cells were cultured in 10% hormone-depleted, heat-treated (80 °C, 20 min) newborn calf serum during incubation with ligand. Incubation with serum
not heat-treated resulted in little or no ligand activity (data not shown), suggesting serum components are significantly sequestering ligand. g NH-3 was
retested in binding and transactivation assays. Results were comparable to previously reported values of 20 ( 7 nM for hTRâ1 and 93 ( 29 nM for hTRR1
in binding, and IC50 ) 370 nM for hTRâ1 in antagonist potency. h Compound did not induce TR-mediated transactivation at 10-5 M nor antagonize 2 ×
10-9 M T3-induced transactivation in a dose-dependent manner. Therefore, EC50 and IC50 could not be obtained for TRR1 and TRâ1.
The key diagnostic signals in the 1H NMR indicating benzofuran
formation were additional peaks near 1.4 ppm (-CH3, iPr,
doublet), near 3.4 ppm (-CH, iPr, heptet), and a peak ranging
from 7.2 to 7.8 ppm (vinyl -H, singlet). Cyclization generally
occurred at a much slower rate than ester hydrolysis and was
limited by controlling the reaction time. Benzofuran formation
was also observed for some analogues under slight acidic
conditions needed for NMR characterization and consequently
limited acquisition of 13C NMR spectra.
TR Binding and Transactivation Properties. All com-
pounds were tested for binding to hTRR1 and hTRâ1 by in vitro
radioligand displacement assays (see Supporting Information)
and the results are summarized in Table 1. KD values reported
are expressed relative to the KD of T3 and were calculated by
fitting the data to the equations of Swillens.32 Data of T3, GC-
(31) Alvarez, S. G.; Alvarez, M. T. A practical procedure for the synthesis of
alkyl azides at ambient temperature in dimethyl sulfoxide in high purity
and yield. Synthesis 1997, 413-414.
9
J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005 4603

A R T I C L E S
Nguyen et al.
119 and previously published NH compounds (NH-1 through
NH-4)1 are included for comparison. As observed with other
5′-phenylethynyl analogues having the core GC-1 scaffold,
analogues NH-5 through NH-24 retained TRâ-selectivity with
impaired affinity compared to GC-1. The TRâ-selectivity was
consistent with structural and chemical data that suggest the
key molecular determinant of selectivity is located on the
1-oxyacetic acid side chain of the GC-1 core structure.14,33 The
analogues bound hTRâ with low nanomolar affinity with KD
values ranging from 0.5 to 450 nM. Analogue NH-16 with the
masked 1-oxyacetic acid moiety also bound TRâ with reason-
able affinity, suggesting the compound might be partially
hydrolyzed to the acid form under the conditions of the binding
assays. Varying the position of the aryl substituent from para-
to ortho- or meta-positions generally led to a decrease in binding
affinity. For example, NH-19 exhibited almost 2-fold decreased
binding affinity compared to NH-3. Likewise, NH-17 and NH-
18 had 60-fold lower affinity relative to NH-5 at TRâ.
The analogues were then tested in human HeLa cells for
transcriptional transactivation properties using a luciferase
reporter assay. HeLa cells were transiently transfected with
expression plasmids for hTRR1 or hTRâ1 and a TRE-driven
(DR4 element)34 luciferase reporter as described previously (see
Supporting Information).1 Analogues NH-6, NH-16, and NH-
24 exhibited partial agonism at TRâ by at least 60% activity
relative to saturating T3-induced transactivation (Figure 2A).
The EC50 values are reported in Table 1. NH-16 exhibited
similar maximal transactivation activity with reduced potency
compared to NH-24. We surmise that the apparent agonist
activity of NH-16 resulted from partial demethylation of the
trimethylammonium salt and partial hydrolysis of methyl ester
to give NH-24. Analogues NH-8 and NH-14 exhibited mixed
agonist and antagonist activity at TRâ. These compounds alone
were able to induce luciferase expression by approximately 40%
relative to saturating T3-induced transactivation, while in
competition assays they were able to partially block 2 nM T3-
induced activity down to the level of compound alone with
micromolar potency (Figure 2B). EC50 values were not calcu-
lated for NH-6, NH-8, NH-14, and NH-16 at TRR because their
relatively poor affinity and potency for TRR made it difficult
to obtain complete dose response curves. Analogues NH-15,
NH-21, and NH-22 displayed weak activity at TRR and TRâ,
as previously observed with NH-4. These compounds did not
induce TR-mediated transactivation above the level of vehicle
control and failed to compete with 2 nM T3-induced transac-
tivation in a dose dependent manner (Table 1). At 10 µM
concentration, NH-15, NH-21, and NH-22 exhibited mild
antagonism to block T3 activity by approximately 10% (data
not shown). All 5′-phenylethynyl analogues generally became
toxic to the cells at doses above 10 µM.
these compounds exhibited full antagonism in completely
blocking 2 nM T3-induced transactivation in a dose-dependent
manner down to the level of activation observed with vehicle
alone (Figure 2C). The IC50 values for antagonism under these
conditions are shown in Table 1. For TRâ, NH-5 and NH-7
(IC50 ) 590 nM and 630 nM, respectively) had 2- to 3-fold
reduced potency compared to NH-3 (IC50 ) 270 nM). The other
antagonists in this group had IC50 values in the micromolar
range. For TRR, the antagonists exhibited approximately 4- to
5-fold reduced potency compared to TRâ. Thus, these com-
pounds are TRâ-selective in binding affinity as well as
transcriptional activity.
Transactivation data for NH-17, NH-18, and NH-19 revealed
the significance of substitution at the para-position on the 5′-
aryl extension for antagonism. The meta- and ortho-substituted
analogues induced minimal transactivation above that of vehicle
control (Figure 2A). However, these compounds inefficiently
blocked 2 nM T3-induced transactivation and complete dose-
response curves could not be obtained. NH-19 at 10 µM blocked
approximately 60% of T3-induced response at TRâ (data not
shown). Similarly, 10 µM of NH-17 and NH-18 partially
blocked the T3 response by 50% and 30%, respectively, at TRâ
(data not shown). The relative lower affinity of these compounds
can partially account for their impaired transactivation activity
compared to the para-substituted counterparts NH-3 and NH-
5. Combined, these results suggest that substitution at the para-
position on the 5′-aryl extension is optimal for ligand binding
and antagonist activity.
Affect of Antagonists on TR Interaction with NCoR and
GRIP-1. The full antagonists NH-5, NH-7, NH-9, NH-11, and
NH-23 were tested for their affect on TR interaction with
coactivator GRIP-1 and corepressor NCoR in mammalian two
hybrid assays. HeLa cells were transiently transfected with
expression plasmids for the yeast GAL4 DBD linked to either
NCoR (aa1925-2308) or GRIP-1 (aa618-1121), hTRâ-LBD
fused to the VP16 activation domain, and a GAL4-driven
luciferase reporter (see Supporting Information).1 As shown in
Figure 3A, the 5′-phenylethynyl antagonists failed to stimulate
binding of GRIP-1 to TR. The analogues also blocked 10 nM
T3-induced TR-GRIP-1 interactions albeit with weak potency.
Increasing antagonist concentration led to decreased T3-induced
TR-coactivator interaction. Dose-response curves could not be
obtained due to concentration and toxicity limitations. As with
NH-3, the inability of these compounds to promote interaction
between TR and its target coactivators is one of the underlying
mechanisms for their observed antagonist activities.
We then examined the ability of the analogues to promote
release of NCoR upon binding to TR. Like T3 and GC-1, the
antagonists induced TR-NCoR dissociation in a dose-dependent
manner (Figure 3B, 3C). The IC50 values of TR-NCoR interac-
tion with bound GC-1 versus bound 5′-phenylethynyl analogues
vary by at least 1 order of magnitude, consistent with the
difference in their relative binding affinities for TR. These results
suggest that in the presence of NH-5, NH-7, NH-9, NH-11,
and NH-23, TR adopts an inactive conformation where neither
corepressors nor coactivators are recruited and TR transcriptional
activity is nullified. This mechanism of action is not unique to
NH-3 and can be generalized to the class of 5′-para-substituted
phenylethynyl derivatives bearing 5′-electron-withdrawing sub-
stituents.
Analogues NH-5, NH-7, NH-9, NH-11, and NH-23 induced
minimal reporter expression above the level of vehicle control
with either TRR or TRâ (Figure 2A). In competition assays
(32) Swillens, S. Interpretation of binding curves obtained with high receptor
concentrations: practical aid for computer analysis. Mol. Pharmacol. 1995,
47, 1197-1203.
(33) Yoshihara, H. A.; Apriletti, J. W.; Baxter, J. D.; Scanlan, T. S. Structural
determinants of selective thyromimetics. J. Med. Chem. 2003, 46, 3152-
3161.
(34) Umesono, K.; Murakami, K. K.; Thompson, C. C.; Evans, R. M. Direct
repeats as selective response elements for the thyroid hormone, retinoic
acid, and vitamin D3 receptors. Cell 1991, 65, 1255-1266.
9
4604 J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005

Selective Thyroid Hormone Receptor Modulators
A R T I C L E S
Figure 2. (A) Transcriptional activation of T3 and 5′-phenylethynyl analogues in luciferase reporter gene assays at hTRâ1 (see Supporting Information).
Analogues with electron donating character generally have agonist activity, while those with electron withdrawing character are nonagonists. [*] indicates
data obtained using media supplemented with 10% heat-treated (80°C, 20 min.) newborn calf serum. (B) Dose-response curves of mixed agonist/antagonist
NH-8 and NH-14 at hTRâ1 with and without 2 nM T3. NH-8 and NH-14 induced approximately 40% transactivation relative to 2 nM T3. In competition
assays, the compounds blocked T3-induced transactivation to the level of activation observed for compound alone. (C) Dose-response curves of NH-3,
NH-5, NH-7, and NH-23 at hTRâ1 in competition with 2 nM T3. NH-9 and NH-11 show similar competition dose-response curves (data not shown). These
compounds failed to induce transactivation and completely inhibited the T3-induced response to the level of ethanol vehicle control. IC50 values under these
conditions are shown in Table 1. Transactivation induced by 2 nM T3 is defined as 100% maximal luciferase activity. Values are the mean ( SD for three
separate experiments. Dose-response data were fitted by nonlinear regression using the Graph Pad Prism computer program (Graph Pad Software Inc.) for
a single site competition model to generate IC50 values.
Hammett Analysis. On the basis of the transactivation data
of the 5′-para-substituted phenylethynyl derivatives, we gener-
ated a Hammett semilog plot of relative ligand potency [log-
(IC50H/IC50X)] versus the Hammett sigma para (σp) substituent
values35,36 (Figure 4A), which were derived from ionization
constants in water of para-substituted benzoic acids relative to
benzoic acid. These σp values incorporate both inductive and
resonance contributions to electronic effects. Although extended
forms of the Hammett equation incorporating additional pa-
rameters such as steric, hydrophobic, and hydrogen-bonding
effects would be more accurate to derive quantitative structure-
activity relationships (QSAR) for interactions of organic
compounds in biological systems, the standard σp values are
sufficient for a broad qualitative correlation analysis.
A general trend, with the exception of NH-7, was observed
between the positive σp values of substituted aromatics and
ligand antagonist activity where greater electron withdrawing
character translated to improved antagonist potency (Figure 4A).
For example, NH-5 (σp ) +0.54) possessed greater electronic
character relative to NH-11 (σp ) +0.06) and exhibited stronger
antagonist potency. NH-3 had the greatest σp value (σp ) +0.78)
for the 5′-phenylethynyl derivatives and remained the most
potent TR antagonist reported. Antagonist NH-23 was not
included in the Hammett analysis since no comparable σp value
could be found in the literature for the benzyl azide substituent.
Qualitative Hammett analysis of negative σp values and relative
ligand potency revealed no significant correlation (data not
(35) Hansch, C.; Leo, A.; Taft, R. W. A survey of Hammett substituent constants
and resonance and field parameters. Chem. ReV. 1991, 91, 165-195.
(36) Jaffe, H. H. A reexamination of the Hammett equation. Chem. ReV. 1953,
53, 191-261.
9
J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005 4605

A R T I C L E S
Nguyen et al.
Figure 3. Affect of antagonists NH-3, NH-5, NH-7, NH-9, NH-11, and NH-23 on TR interactions with corepressor and coactivator in mammalian two-
hybrid assays (see Experimental section). (A) At 10 µM concentration, the antagonists fail to recruit GRIP-1 to TR. However, increasing antagonist concentration
inhibited with micromolar potency the TR-GRIP-1 interactions induced by 10 nM T3. Competition dose-response curves could not be obtained due to
concentration and toxicity limits. TR-GRIP-1 interaction in the presence of 10 nM T3 is defined as 100% maximal luciferase activity. (B) Like T3 and GC-1,
the antagonists were able to promote NCoR dissociation from TR in a dose-dependent manner. (C) NH-3, NH-5, and NH-7 inhibited TR-NCoR interaction
by 50% (IC50) at 58 nM, 64 nM and 120 nM, respectively. TR-NCoR interaction in the presence of ethanol vehicle control is defined as 100% maximal
luciferase activity. All values are expressed as the mean ( SD for three separate experiments. IC50 values were calculated using the Graph-Pad computer
program (Graph-Pad Software Inc.) by nonlinear regression of a sigmoid dose-response model.
Figure 4. (A) Hammett analysis of electronic character (σp) and relative transcriptional activity (log IC50H/IC50x) showed a linear trend (r2 ) 0.72, slope )
1.52) of improved antagonist potency with increased electron withdrawing character of the 5′-para-substituted aryl extension. IC50H was assigned a value of
10 µM. (B) Hammett analysis of electronic character (σp) and relative binding affinity (log KDH/KDX) revealed essentially no linear correlation (r2 ) 0.17,
slope ) 0.56) between the two parameters. Hammett σp substituent values were based on values reported by Hansch et al.35 and Jaffe:36 X/σp ) H/0; F/0.06;
N3/0.08; CO2CH3/0.45; COCH3/0.50; CF3/0.54; CN/0.66; NO2/0.78.
shown); however, compounds bearing an EDG such as NH-1,
NH-2, NH-6, and NH-24 had partial or full agonist activity.
Similar Hammett analysis of the relative ligand binding
affinity [log(KDH/KDX)] versus the σp parameter revealed no
9
4606 J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005

Selective Thyroid Hormone Receptor Modulators
A R T I C L E S
significant correlation between binding affinity and electronic
character of the 5′-aryl extension (Figure 4B). The presence of
a strong EDG or EWG did not improve nor impair binding.
For example NH-6 (σp ) -0.27) has electron-donating character
while NH-3 (σp ) +0.78) and NH-5 (σp ) +0.54) have
electron-withdrawing character, yet these compounds had
comparable binding affinity and selectivity to TRâ (Table 1).
Lack of correlation between binding affinity and electronic
properties of the 5′-aryl extension suggests that altering the
electronic properties primarily affects TR functional activity
through downstream transactivation signaling events.
antagonist activity and potency. Analogues with EWG in the
ortho- or meta-positions exhibited significantly reduced antago-
nist potency as seen with analogues NH-17, NH-18, and NH-
19 compared to the para-substituted counterparts NH-3 and NH-
5. This result can partially be accounted for by the reduction in
electronic contribution to the aryl extension in the meta- and
ortho-positions relative to the para-position.35 However, com-
pounds with weaker para-electron withdrawing substituents, such
as NH-7 and NH-23, also displayed full antagonist activity
indicating electronic properties alone do not dictate antagonism.
The azido groups of NH-7 and NH-23 may be involved in
stabilizing an inactive TR conformation.
Discussion
To better understand the molecular basis of antagonism for
this class of STRMs, the 5′-phenylethynyl extension was
modeled into the TR ligand binding pocket based on the X-ray
crystal structure of GC-1 bound to hTRâ-LBD.14,37 Modeling
revealed that the 5′-extension clashed with numerous receptor
residues in the active conformation, including Phe455 and
Phe459 of helix H12. In addition to sterically perturbing proper
folding and rearrangement of H12, we postulate that the EWG
on the 5′-aryl extension promotes favorable aromatic π-interac-
tions (center-to-edge, edge-to-face, and/or face-to-face)38 with
the phenylalanine residues to stabilize an inactive receptor
conformation. Given that the 5′-phenylethynyl GC-1 antagonists
induce a conformation that occluded binding of both coactivators
and corepressors, the ligand extension may induce a conforma-
tion of H12 that packs against the hydrophobic groove where
the coactivator and corepressor binding sites overlap.10,39,40 This
mode of “active antagonism”41 is similar to that observed for
the estrogen receptor (ER) bound to selective ER modulators
(SERMs) raloxifene42 and 4-hydroxytamoxifen,43 where H12
adopts an auto-inhibitory conformation to compete with co-
activator recruitment by mimicking the interactions of the
coactivator with the ER-LBD.
In summary, the results of this study confirmed our hypothesis
that the 5′-p-nitroaryl extension is not required for antagonism
and that the size, position, and electronic nature of the 5′-aryl
extension will dictate a spectrum of agonist and antagonist
activity. A better understanding of the pharmacophore important
for T3 antagonism will be useful for the development of
therapeutics for the treatment of diseases associated with
excessive thyroid hormone production and action such as
hyperthyroidism (thyrotoxicosis).44 Current therapies for hy-
In the present study we evaluated, the significance and role
of the 5′-p-nitroaryl moiety of NH-3 for T3 antagonist activity.
We synthesized a series of 5′-phenylethynyl GC-1 derivatives
varying in size and electronic property. All analogues bound
TRR and TRâ with moderate nanomolar affinity and 4-fold to
20-fold TRâ selectivity (Table 1). In transactivation assays with
TRâ (Figure 2), NH-6, NH-16, and NH-24 had partial agonist
activity, while NH-8 and NH-14 exhibited mixed agonist/
antagonist activity. NH-5, NH-7, NH-9, NH-11, and NH-23
were full antagonists with reduced potency relative to NH-3.
We further tested the affect of the antagonists NH-5, NH-7,
NH-9, NH-11, and NH-23 on TR interactions with NCoR and
GRIP-1 (Figure 3). Similar to NH-3, these compounds func-
tioned like agonists to promote corepressor release from TR
but failed to induce recruitment of coactivator to TR. Thus, the
5′-phenylethynyl derivatives containing para-substituted EWG
are a unique class of selective TR modulators (STRMs) that
effectively induce a chemical TR knockout by preventing TR-
mediated hormone transactivation and relieving trans-
repression of positively- regulated target gene expression.
Closer analysis of the chemical features important for TR
modulation by Hammett correlation between TR-mediated
ligand activity and electronic character revealed 5′-aryl exten-
sions with EDG (negative σp value) had agonist activity while
those carrying charge neutral EWG (positive σp value) had
antagonist activity (Figure 4A). Moreover, greater antagonist
potency was observed with greater electron withdrawing
character (greater σp value) as with NH-3, NH-5, and NH-7.
NH-3 had the greatest σp value in this class of compounds and
remained the most potent T3 antagonist. The antagonistic
potency of azide NH-7 did not follow the linear correlation of
the other six antagonists for reasons we do not understand
presently. The combined SAR data reveal the 5′-p-nitroaryl
moiety was not required for antagonism. However, substitution
with a strong EWG on the 5′-aryl extension was essential for
antagonist activity.
(37) Yoshihara, H. A.; Nguyen, N. H.; Scanlan, T. S. Design and synthesis of
receptor ligands. Methods Enzymol. 2003, 364, 71-91.
(38) Cozzi, F.; Annunziata, R.; Benaglia, M.; Cinquini, M.; Raimondi, L.; et
al. Through-space interactions between face-to-face, center-to-edge oriented
arenes: importance of polar-pi effects. Org. Biomol. Chem. 2003, 1, 157-
162.
(39) Marimuthu, A.; Feng, W.; Tagami, T.; Nguyen, H.; Jameson, J. L.; et al.
TR surfaces and conformations required to bind nuclear receptor corepres-
sor. Mol. Endocrinol. 2002, 16, 271-286.
Hammett analysis of the ligand binding affinity relative to
the σp substituent values confirmed that varying the electronic
character of the ligand predominantly affects the downstream
signaling interactions of TR and not binding. As shown in Figure
4B, there was variability in the binding affinities of the
5′-phenylethynyl derivatives but no correlation was observed
between electronic property and changes in binding affinity.
Compounds having EDG aryl extensions bound TR with similar
affinities as those bearing EWG. Moreover, stronger electronic
character does not enhance binding affinity.
(40) Perissi, V.; Staszewski, L. M.; McInerney, E. M.; Kurokawa, R.; Krones,
A.; et al. Molecular determinants of nuclear receptor-corepressor interaction.
Genes DeV. 1999, 13, 3198-3208.
(41) Shiau, A. K.; Barstad, D.; Radek, J. T.; Meyers, M. J.; Nettles, K. W.; et
al. Structural characterization of a subtype-selective ligand reveals a novel
mode of estrogen receptor antagonism. Nat. Struct. Biol. 2002, 9, 359-
364.
(42) Brzozowski, A. M.; Pike, A. C.; Dauter, Z.; Hubbard, R. E.; Bonn, T.; et
al. Molecular basis of agonism and antagonism in the oestrogen receptor.
Nature 1997, 389, 753-758.
(43) Shiau, A. K.; Barstad, D.; Loria, P. M.; Cheng, L.; Kushner, P. J.; et al.
The structural basis of estrogen receptor/coactivator recognition and the
antagonism of this interaction by tamoxifen. Cell 1998, 95, 927-937.
(44) Utiger, R. D. The thyroid: physiology, thyrotoxicosis, hypothyroidism, and
the painful thyroid. Endocrinology and Metabolism; McGraw-Hill: New
York, 1995; pp 435-519.
Our data further lends support to the observation that the
nature and position of the 5′-aryl substituent is critical for T3
9
J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005 4607

A R T I C L E S
Nguyen et al.
perthyroidism usually require weeks before symptoms are
relieved or provide only partial relief. Direct blockade of T3
action with T3 antagonists would bypass such complications and
provide a more effective treatment of hyperthyroidism. Fur-
thermore, TR subtype-selective STRMs will be useful pharma-
cological probes for studying thyroid hormone signaling path-
ways.
Grant DK41842 to J.D.B). We thank Drs. M. Privalsky, R.
Evans, and D. Moore for the gifts of plasmids used in the
transfection and mammalian two-hybrid assays.
Supporting Information Available: Experimental procedures
for chemical synthesis of novel compounds and protocols for
binding and transactivation assays are included (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
Acknowledgment. This work was supported by grants from
the National Institutes of Health (Grant DK52798 to T.S.S;
JA0440093
9
4608 J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005