Discovery of a novel series of 6-azauracil-based thyroid hormone receptor ligands: Potent, TRβ subtype-selective thyromimetics

Article abstract of DOI:10.1016/S0960-894X(02)00947-2

In this communication, we wish to describe the discovery of a novel series of 6-azauracil-based thyromimetics that possess up to 100-fold selectivities for binding and functional activation of the β₁-isoform of the thyroid receptor family. Structure-activity relationship studies on the 3,5- and 3′-positions provided compounds with enhanced TRβ affinity and selectivity. Key binding interactions between the 6-azauracil moiety and the receptor have been determined through of X-ray crystallographic analysis.

Full text of DOI:10.1016/S0960-894X(02)00947-2

Bioorganic & Medicinal Chemistry Letters 13 (2003) 379–382
Discovery of a Novel Series of 6-Azauracil-Based
Thyroid Hormone Receptor Ligands:
Potent, TRꢀ Subtype-Selective Thyromimetics
Robert L. Dow,a,* Steven R. Schneider,a Ernest S. Paight,a Richard F. Hank,a
a
Phoebe Chiang,a Peter Cornelius,a Eunsun Lee,a WilliamP. Newsome,
Andrew G. Swick,a Josephine Spitzer,a Diane M. Hargrove,a Terrell A. Patterson,a
Jayvardhan Pandit,b Boris A. Chrunyk,b Peter K. LeMotte,b Dennis E. Danley,b
Michele H. Rosner,b Mark J. Ammirati,b Samuel P. Simons,b Gayle K. Schulte,b
Bonnie F. Tateb and Paul DaSilva-Jardinea
aCardiovascular and Metabolic Diseases, Pfizer Global Research and Development, Groton, CT 06340, USA
bExploratory Medicinal Sciences, Pfizer Global Research and Development, Groton, CT 06340, USA
Received 5 September 2002; revised 4 November 2002; accepted 6 November 2002
Abstract—In this communication, we wish to describe the discovery of a novel series of 6-azauracil-based thyromimetics that pos-
sess up to 100-fold selectivities for binding and functional activation of the b1-isoformof the thyroid receptor family. Structure–
activity relationship studies on the 3,5- and 30-positions provided compounds with enhanced TRb affinity and selectivity. Key
binding interactions between the 6-azauracil moiety and the receptor have been determined through of X-ray crystallographic
analysis.
# 2002 Elsevier Science Ltd. All rights reserved.
It has been well established in humans that a hyper-
thyroid state leads to increased metabolic rate (thermo-
genesis).1 However, attempts to enhance metabolic rate
in obese patients via treatment with thyroxine (T4),
which is metabolized in various tissues to afford the
active thyroid hormone T3, leads to unacceptable car-
diac side effects. Human cardiac tissue has been shown
to contain predominately TRa1, a major isoform of the
thyroid hormone receptor (TR) family.2 T3 does not
show preferential binding amongst the major TR iso-
forms (TRa1, TRb1, and TRb2), suggesting that identi-
fication of a TRb-specific agonist could lead to an agent
which would retain the thermogenic potential, but
would be devoid of the cardiac side effects.
of our work, investigators at UCSF have reported that
an oxyacetic acid analogue (GC-1) of DIMIT3 is a
potent thyromimetic which exhibits ꢀ10-fold selectivity
in binding to the TRb subtype.4 More recently, HY-1 a
neutral analogue of GC-1 was discovered to be 5-fold
more potent in activating TRb mutant (R320C) relative
to the wild-type.5 In this communication, we wish to
describe the discovery of appropriately functionalized
6-azauracils that are potent and TRb-selective thyro-
mimetics.
To date there have been a limited number of reports on
efforts to identify thyroid ligands that interact selec-
tively with the various TR isoforms. During the course
Screening of the Pfizer sample collection led to the dis-
covery of several 6-azauracil-based diphenyl ethers (1)
*Corresponding author. Tel.:+860-441-4423; fax:+860-715-4610;
e-mail: robert_l_dow@groton.pfizer.com
0960-894X/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00947-2

380
R. L. Dow et al. / Bioorg. Med. Chem. Lett. 13 (2003) 379–382
that bind to TRb1 with submicromolar potencies and
modest (5- to 15-fold) selectivities versus the a1-isoform
of the thyroid receptor.6 These compounds were origin-
ally prepared as part of an anticoccidial programin the
Veterinary Medicine Division.7,8 Froma limited set of
extant compounds, those derivatives containing a 40-
methylthioether substituent (R1=SMe) possessed the
greatest TRb1 potencies and selectivities. Not surpris-
ingly, the most potent analogue (2) identified in this
seen for the azauracil. Further simplification to the six-
membered cyclic urea (5) completely eliminates TR
binding (Ki>10,000 nM). This alignment of N-3 acidity
(3>4ꢂ5) and binding potency further supports the
importance of the Arg320 interaction with an ionized
imide functionality.
Given the strict structural requirements of the 6-azau-
racil required for TR binding, further improvements in
TRb binding potency or selectivity would have to be
driven through modification of the diphenyl ether core.
However, since the only difference between the ligand
binding domains of hTRb and hTRa is a single residue
(Ser277 in TRa for Asn331 in TRb) there was some
question as to whether modifications to the diphenyl
ether would have any impact on TR isoform binding
selectivity. While there have been a number of SAR
reports on modifications to the diphenyl ether core
(ether linker, 3,5- and 30-positions) of T3, few of these
efforts have been directed towards enhancing TR iso-
formbinding selectivity.
initial screen contains
a
lipophilic substituent
(R2=methyl) in the 30-position, which overlaps with the
iodine and isopropyl substituents in T3 and DIMIT,
respectively. Compound 2 possesses good binding affi-
nity (Ki=2 nM, T3 Ki=0.08 nM) for TRb1 and has a
6-fold lower affinity for the a1-isoformof the thyroid
receptor.
Initial efforts around the 30-position focused on profil-
ing a range of substituents, including aminomethyl,
acylamino, sulfonyl, alkyl, aryl, alkylamino, arylamino,
acyl, carboxamido and sulfonamido (data not shown).
0
Fromthis initial evaluation of 3 -functional groups,
carboxamido- and sulfonamido-based substituents
exhibited the greatest potential for identifying potent,
TRb-isoformselective thyromimetics.
Neither 2 nor any of the other initial screening leads
exhibit significant functional agonismof thyroid recep-
tors. The lack of functional activity is due to the absence
of the 40-phenolic functionality present in classical thyro-
mimetics which forms a hydrogen bond with a histidine
residue in the TR binding pocket.9 Incorporation of
the key 40-hydroxyl and 30-isopropyl functionalities (3)
produces ꢀ100ꢁ improvement in TRb1 binding affi-
nity (Ki=0.03 nM) and functional agonismin E25B2
cells.10 Whereas DIMIT and T3 possesses no TR iso-
formselectivity, compound 3 retains a modest 8-fold
TRb1 selectivity, confirming that the 6-azauracil func-
tionality is responsible for the TR isoformselectivity
observed.
To aid in rapid structure–activity relationship develop-
ment in the 30-position, synthetic routes were designed
which allow for the incorporation of the amide or sul-
fonamide functionalities as the final step. The core
6-azauracil-based diphenyl ethers (6) are constructed
utilizing methodologies previously reported.7,8 The 30-
position of the diphenyl ether core can be readily and
selectively functionalized by formylation, chloro-
sulfonylation or nitration, employing 6 or the corre-
sponding phenol (Scheme 1). Oxidation of the 30-formyl
derivative to the carboxylic acid and methyl ether clea-
vage affords 8. Activation of this o-hydroxyacid with
N-hydroxysuccinimide provides a stable, activated ester
To gain insight into the structural features of the
6-azauracil moiety responsible for TR binding affinity,
an X-ray crystal structure of 3 bound to TRb1 was
generated (see Fig. 1).11 In analogy to the binding
modes for T3 and GC-1,12 the ionized imide function-
ality of the 6-azauracil forms a hydrogen-bonding array
with Arg320 of TRb1. In contrast to the structures of T3
and GC-1, an additional hydrogen bond interaction
also occurs between the C-4 carbonyl of 3 and Arg316.
A number of SAR findings relating to the 6-azauracil
moiety are consistent with this binding motif defined by
X-ray crystallography. Elimination of the acidic proton
at N-3 of the azauracil ring (via N-methylation—data
not shown) produces a>50-fold reduction in receptor
affinity. Removal of the nitrogen in the 6-position to
afford the corresponding uracil 4 results in a 120-fold
reduction in binding affinity for the TRb1 receptor.
While 4 loses significant TRb1 potency it still retains a
TRb1/TRa selectivity (9-fold) that is comparable to that
Figure 1. Binding of 3 to human TRb1.11

R. L. Dow et al. / Bioorg. Med. Chem. Lett. 13 (2003) 379–382
381
was held constant while the 3,5-substituents were varied
(7a–c). As is observed for other 30 substituents, potency
and selectivity increases in the following order: Me,
Me<Me,Cl<Cl,Cl. In the case of the 7c binding affi-
nity is restored to a level that is comparable to 3 and 4ꢁ
more potent than T3. In addition to enhanced potency,
7c also possesses an ꢀ4-fold enhanced TRb/TRa selec-
tivity relative to 3.
Attempts to introduce heteroatom functionality (7d–e)
into the 30 site result in significantly reduced binding
affinity. These results are consistent with the highly
lipophilic nature of the 30-receptor binding pocket.9
Sulfonamides derived from primary, aliphatic amines
(e.g., 7f–g) exhibit reduced binding affinities (10–20ꢁ)
relative to tertiary sulfonamides of comparable size,
though binding selectivity is not diminished. TRb bind-
ing is not diminished to such a great extent when a pri-
mary, aromatic amine (7h) is incorporated. Unlike in
the case for aliphatic amine-based sulfonamide sub-
stituents, aromatic amine-derived, tertiary sulfonamides
such 7i or the N-methylaniline analogue (data not
shown) do not improve binding affinity.
Scheme 1. Reagents and conditions: (a) hexamethylenetetramine, TFA,
reflux (65%); (b) NaClO2, 2-methylbutene, KH2PO4, t-BuOH, THF,
H2O (79%); (c) BCl3, CH2Cl2, ꢃ78 ꢄC to rt (78%); (d) N-hydroxy-
succinimide, 1,3-dicyclohexylcarbodiimide, DME, 0 ꢄC to rt (83%); (e)
R3R4NH, TEA, DME (65–90%); (f) BBr3, CH2Cl2, ꢃ78–0 ꢄC (57%);
(g) chlorosulfonic acid, 0 ꢄC (50%); (h) R3R4NH (3 equiv), THF (45–
80%).
that can be employed in the preparation of a range of 30-
amides (9). Synthesis of the corresponding 30-sulfon-
amides (7) is accomplished through initial cleavage of
the methyl ether of 6. Condensation of this phenol with
chlorosulfonic acid provides the 30-chlorosulfonyl inter-
mediate, which can be coupled with amines to form the
target sulfonamides 7.
Based on the above results a range of tertiary sulfon-
amide substituents derived from aliphatic amines were
profiled. In general, 30 substituents more sterically
demanding than that present in 7c lead to a reduction in
TRb1 affinity. One exception to this trend is 7j, which
incorporates a tropinone-based functionality. Because
of the hydrolytic instability of the ketal moiety in 7j the
homologous spiro-tetrahydrofuranyl analogue 7k was
prepared.13 Analogue 7k binds potently and selectivity
(86-fold) to TRb1. In E25B2 cells 7k is a full agonist,
however it is ꢀ30-fold less active than T3 (EC50=62
nM, T3=2 nM) and is 28-fold selective versus TRa
(EC50 ꢀ1.5 mM, T3=2 nM).
Table 1 details a cross-section of a wide range of sulfon-
amide substituents that were profiled in the 30-position
of the diphenyl ether core. Incorporation of the piperi-
dinylsulfonamide group into the core structure (7a)
produces a 15-fold loss in TRb binding affinity relative
to isopropyl analogue 3. As part of the initial profiling
in this series, the piperidinyl sulfonamide substituent
Structure–activity trends for the 30-carboxamido series
were quite distinct fromthat seen for the sulfonamides
(Table 2). Though piperidinyl derivative 9b is more
active than the 3-chloro-5-methyl analogue 9a, it is
ꢀ30-fold less potent in binding to TRb1 than the cor-
responding sulfonamide (7c). Also in contrast to the 30-
sulfonamides, amides derived from primary amines
produced a substantial improvement in potency (9c vs
9b), though a decrease in the lipophilic character (9d)
leads to a 10-fold reduction in potency. Another inter-
esting point of divergence between the two series is that
incorporation of a morpholinyl substituent did not sig-
nificantly reduce potency (9e vs 9b) in the carboxamide
case.
Table 1. Inhibition of [125I]-T3 binding to human TRb1 and hTRa1
by 30-sulfonamides6
Compd
R1
R2
N-R3R4
TR binding
TRb1 Ki
(nM)
a/ba
T3
7a
7b
7c
7d
7e
7f
7g
7h
7i
0.08
0.8
14.8
21.2
30.9
8.2
13
21.3
55
Me
Me
Cl
Me
Me
Cl
Cl
Cl
Cl
Me
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Piperidinyl
Piperidinyl
Piperidinyl
0.46
0.13
0.02
13
27
0.46
0.40
0.10
0.23
4-Methylpiperizinyl
Morpholinyl
Cyclohexylamino
Cyclobutylamino
Anilinyl
14.3
13.5
Indolinyl
Based on 9c, a range of commercially available mono-
and bicyclic-based secondary carboxamide substituents
were evaluated. Fromthis effort several bicyclic amine-
based analogues (9f–9h) were discovered to possess
potent TRb1 binding affinity and selectivities of up to
100-fold versus TRa. As a follow-up to these findings,
the enantiomeric, nopinone-derived amines were incor-
porated to provide 9i and 9j.14 While both are potent
(Ki ꢀ50 pM) binders, the R-enantiomer is substantially
more selective. In functional assays 9i is both a potent
7j
Cl
Cl
Cl
Cl
0.31
0.28
33.0
86
7k
aRatio of TRa Ki/TRb Ki.

382
R. L. Dow et al. / Bioorg. Med. Chem. Lett. 13 (2003) 379–382
Table 2. Inhibition of [125I]-T3 binding to human TRb1 and hTRa1
mmol). High Five insect cell nuclear extract, prepared from cells
engineered to express either human TRa1 or TRb1, was diluted
in assay buffer, then combined with an equal volume of each
compound dilution (final 125I-T3=0.2 nM). After incubating
at room temperature for 90 min, bound 125I-T3 was separated
fromfree using a filtration manifold (Millipore #MHAB N45).
Radioactivity was quantitated in a Wallac Microbeta plate
scintillation counter, and IC50 or Ki values were calculated
using non-linear regression analysis (GraphPad/Prism).
7. Miller, M. W. US Patent 3,905,971; Chem. Abstr. 1975, 82,
171086.
by 30-carboxamides6
Compd
R1
R2
N-R3R4
TR binding
a/ba
TRb1 Ki
(nM)
T3
9a
9b
9c
9d
9e
9f
0.08
1.9
0.8
Me
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Piperidinyl
Piperidinyl
Cyclohexylamino
Cyclobutylamino
Morpholinyl
(ꢅ)-exo-2-Norbornyl
(ꢅ)-endo-2-Norbornyl
(R)-(+)-Bornyl
75.3
66.6
33.8
73.8
30.1
104.8
32.2
49.7
0.63
0.07
0.64
0.72
0.59
0.16
0.37
8. Carroll, R. D.; Miller, M. W.; Mylari, B. L.; Chappel, L. R.;
Howes, H. L.; Lynch, M. J.; Lynch, J. E.; Gupta, S. K.; Rash,
J. J.; Koch, R. C. J. Med. Chem. 1983, 26, 96, and references
cited therein.
9g
9h
9. Wagner, R. L.; Apriletti, J. W.; McGrath, M. E.; West,
B. L.; Baxter, J. D.; Fletterick, R. J. Nature 1995, 378, 690.
10. Cells (E25B2) were transiently transfected (using lipo-
fectin) with cDNAs encoding either human TRa1 or TRb1
together with a plasmid containing firefly luciferase under the
control of a thyroid hormone response element (DR+4).
Following transfection, the cells were distributed into 96-well
plates and compounds and/or T3 were added to the culture
medium and luciferase activity was measured in cell lysates
approximately 18 h later.
9i
9j
Cl
Cl
Cl
Cl
0.06
0.04
58.2
15
aRatio of TRa Ki/TRb Ki.
11. Recombinant human TRb ligand-binding domain (residues
204-461) was expressed in Escherichia coli. Isolation and pur-
ification of the protein was performed using anion exchange, size
exclusion and cation exchange chromatography, maintaining the
presence of 3 throughout the purification. Crystals of the com-
plex were grown by hanging-drop vapor diffusion. The reservoir
solution contained 1.8 M (NH4)2SO4 in 0.1 M Na Hepes, pH 6.0.
The crystals were monoclinic, space group P2(1), with unit cell
agonist (EC50=12 nM) and highly selective (77-fold)
versus TRa.
To date, a detailed receptor structure-based under-
standing of the impact on TR binding of these 30 modi-
fications has been thwarted by an inability to co-
crystallize ligands containing a 30-substituent larger
than isopropyl. It is presumed that modifications of the
30-position produce conformational/positional shifts
that negatively affect binding around the polar pocket
(vicinity of Ser277) of TRa.
ꢄ
˚
dimensions a=42.92, b=105.31, c=55.99 A, b=98.19 , and
contained two molecules of the protein–ligand complex in the
˚
asymmetric unit. Diffraction data to 2.2 A resolution were col-
lected on an RAXIS-IIc detector mounted on a Rigaku Ru200
X-ray generator. The structure was solved by molecular repla-
cement, using the coordinates of a the structure of human TRb
complexed with T3, solved earlier (data not shown). The atomic
coordinates have been deposited in the Protein Data Bank,
http://www.rcsb.org (PDB ID code 1N46).
12. Wagner, R. L.; Huber, B. R.; Shiau, A. K.; Kelly, A.;
Cunha Lima, S. T.; Scanlan, T. S.; Apriletti, J. W.; Baxter,
J. D.; West, B. L.; Fletterick, R. J. Mol. Endocrin. 2001, 15,
398.
A series of 30-sulfonamide and 30-carboxamide ana-
logues of the screening lead 2 have been prepared. Sev-
eral of these compounds have been shown to be potent
and TRb-selective thyromimetics. The potential for
these selective agents to increase metabolic rate, while
limiting cardiovascular side effects, will be the subject of
future reports.
13. Tropane-based amine 11 was synthesized from commer-
cially available N-carboethoxy-4-tropanone via the sequence
outlined below:
References and Notes
1. Sestoft, L. Clin. Endocrin. 1980, 13, 489.
2. Blange, I.; Drvota, V.; Yen, P. M.; Sylven, C. Biol. Pharm.
Bull. 1997, 20, 1123.
3. Jorgensen, E. C.; Wright, J. J. Med. Chem. 1970, 13, 745.
4. Chiellini, G.; Aprilett, J. W.; Yoshihara, H. A.; Baxter,
J. D.; Ribeiro, R. C. J.; Scanlan, T. S. Chem. Biol. 1998, 5, 299.
5. Ye, H. F.; O’Reilly, K. E.; Koh, J. T. J. Am. Chem. Soc.
2001, 123, 1521.
6. Compounds (20 mM in 100% DMSO) were serially diluted in
100% DMSO, then diluted in assay buffer (5 mM Tris–HCl, pH
8.0; 50 mM NaCl; 2 mM EDTA; 10% (v/v) glycerol; 1 mM
DTT) containing 0.4 nM [125I]-T3 (specific activity=2200 Ci/
Reagents: (a) allylmagnesium chloride, THF, ꢃ78 ꢄC (95%);
(63%); (c) p-toluenesulfonyl chloride, pyridine, 5 ꢄC (55%); (d)
hydrazine hydrate, ethylene glycol, reflux (80%).
14. The amines utilized in the preparation of 9i and 9j were
synthesized fromthe enantiomers of nopinone utilizing the
following procedure. Ipaktschi, J. Chem. Ber. 1984, 117, 856.
.
(b) BH3 THF, THF, then NaOH (aq), 3% aq H2O2, MeOH