Thyroid receptor ligands. Part 5: Novel bicyclic agonist ligands selective for the thyroid hormone receptor β

Article abstract of DOI:10.1016/j.bmcl.2005.11.077

Based on the examination of the crystal structure of rat TRβ complexed with 3,5,3′-triiodo-l-thyronine (2) a novel TRβ-selective indole derivative 6b was prepared and tested in vitro. This compound was found to be 14 times selective for TRβ over TRα in bi

Full text of DOI:10.1016/j.bmcl.2005.11.077

Bioorganic & Medicinal Chemistry Letters 16 (2006) 1240–1244
Thyroid receptor ligands. Part 5: Novel bicyclic agonist
ligands selective for the thyroid hormone receptor b
˚
Ana-Maria Garcia Collazo, Konrad F. Koehler, Neeraj Garg, Mathias Fa¨rnegardh,
Bolette Husman, Liu Ye, Jan Ljunggren, Karin Mellstro¨m, Johnny Sandberg,
Marlena Grynfarb, Harri Ahola and Johan Malm*
Karo Bio AB, Novum, Huddinge SE-141 57, Sweden
Received 1 November 2005; revised 20 November 2005; accepted 21 November 2005
Available online 9 December 2005
Abstract—Based on the examination of the crystal structure of rat TRb complexed with 3,5,30-triiodo-L-thyronine (2) a novel TRb-
selective indole derivative 6b was prepared and tested in vitro. This compound was found to be 14 times selective for TRb over TRa
in binding and its b-selectivity could be rationalized through the comparison of the X-ray crystallographic structures of 6b
complexed with TRa and TRb.
Ó 2005 Elsevier Ltd. All rights reserved.
Endogenous thyroid receptor hormones 3,5,30,50-tetra-
I
I
iodo-L-thyronine (T4, 1) and 3,5,30-triiodo-L-thyronine
(T3, 2) exert profound effects on growth, development,
and homeostasis in mammals (Fig. 1).1 Thyroid hor-
mone receptor subtypes (TRa and TRb) mediate differ-
ential functions, suggesting the possibility of developing
selective thyromimetics that cause therapeutic increases
in metabolic rate (anti-obesity effect) and lipid lowering
without deleterious effects on the heart.2 Our hypothesis
is that selective TRb activation can give such a profile.3
I
O
I
I
O
I
NH2
CO2H
NH2
CO2H
HO
HO
I
L-T4 (1)
L-T3 (2)
Figure 1. Chemical structures of L-T4 and L-T3.
replace the iodine atoms with alternative substituents
since iodine is highly susceptible to reductive deiodin-
ation and furthermore hinders biaryl ether formation
due to its large steric bulk. With regard to achieving a
reasonable duration of action in vivo, replacing the io-
dine atoms with alternative substituents eliminates a po-
tential route of metabolic deactivation via enzymatic
deiodination.
TRa and TRb differ in a single amino acid residue in the
hormone binding pocket (Ser-277 vs Asn-331, respec-
tively). It has been proposed that this amino acid differ-
ence accounts for most of the observed selectivity for
TRb-binding in a series of phenyl acetic acid thyromi-
metics.4 This amino acid residue as well as the a-alanine
side chain of 2 is located in a highly flexible region of the
receptor.5 We reasoned that rigidification of the ligand
carboxylic acid side chain through ring formation may
lead to increased opportunities for subtype selectivity.
In addition, the fused ring provides a rigid scaffold on
which substituents could be introduced to more directly
exploit this amino acid difference. We also decided to
An examination of the crystal structure5 of rat TRa1
complexed with 2 (Fig. 2) reveals a significant unoccu-
pied space in the binding cavity next to position-2 of
the ligand. This suggests that a ring fusion between the
benzylic carbon atom of the position-1 substituent and
the position-2 of the aromatic ring would be sterically
tolerated by the receptor.
Based on the observations above, we prepared a number
of TR-analogues in which an additional ring has been
fused to the inner ring of the TR ligand. As outlined
below (Scheme 1), the first synthetic step involved
Keywords: Thyroid hormone receptor agonist; TRb-selectivity; Indole;
Indane; Quinoline; X-ray.
*
Corresponding author. Tel.: +46 8 608 6046; fax: +46 8 774 8261;
e-mail: johan.malm@karobio.se
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.11.077

A.-M. Garcia Collazo et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1240–1244
1241
Figure 2. Comparison of the solvent accessible surface areas of thyroxine (2; yellow surface; ligand atom coloring scheme: white = carbon,
blue = nitrogen, red = oxygen, purple = iodine, and light blue = hydrogen atoms) and the interior binding cavity of TRa (green surface) derived from
the crystallographic structure of rat TRa complexed with thyroxine.5 The solvent accessible surfaces are Z-clipped for the sake of clarity. The
unoccupied space adjacent to position-2 of the ligand suggests that a ring fusion between the benzylic carbon atom of the position-1 substituent and
the position-2 of the aromatic ring would be sterically tolerated by the receptor.
X
X
X
O
X
(d), (e)
(a), (b)
O
X
(c)
(f)
HO
X
O
CO2H
CO2Et
N
HO
N
MeO
H
NO2
MeO
MeO
X
4a-c
NH2
H
6a-b
5a-b
Br
Br
7
Br
Br
(g), (h)
(a)
O
HO
Br
O
O
CO2H
CO2Et
HO
Br
10
Br
MeO
Br
9
O
O
O
O
8
X
X
O
In all structures with "X":
a: X= Br;
O
(i), (j)
(d)
O
4a-c
b: X= Cl;
c: X= Me
HO
X
N
H
CO2H
CO2H
MeO
MeO
X
11a-c
N
H
CO2Et
12a-c
Me Cl
Me
O
O
(k)
(l), (d)
11c
Me
HO
Me
N
CO2Et
N
13
14
Scheme 1. Synthetic route for the preparation of the 6a,b, 10, 12a–c and 14. Reagents and conditions: (a) bis(3-isopropyl-4-methoxyphenyl)iodonium
tetrafluoroborate, Et3N, Cu; (b) SnCl2; (c) (i) HCl; (ii) NaNO2, SnCl2, HCl; (iii) ethyl pyruvate, H2SO4, AcOH; (iv) PPA; (d) BBr3, CH2Cl2; (e)
NaOH, THF; (f) diethylcarbonate, NaH, C6H6; (g) triethylsilane, TFA; (h) BBr3, CH2Cl2, ꢀ78 °C; (i) EtO2CCCCO2Et, MeOH; (j) PPA; (k) POCl3;
(l) H2–Pd–C, KOH.
iodonium salt coupling6 with the appropriate starting
phenols 3a–c and 7, to obtain the corresponding biaryl
ethers 4a–c and 8 in good yield. Synthesis of the final
target ligands 4,6-dibromo-6a and 4,6-dichloroindole
6b7 was accomplished via a modification of the Fisher
Indole synthesis starting from the anilines 4a,b. The
indane-2-carboxylic acid 10 was prepared via direct
2-position carboxylation of the indanone 8.8 The 5,7-
dibromo (12a), 5,7-dichloro- (12b), and 5,7-dimethyl-6-
tetrahydroquinoline-2-carboxylic acids (12c) were all
prepared from 4a–c by reaction with ethynyldicarboxy-
lic acid ethyl ester. The intermediate Schiff base was
treated with PPA to give ring closure to intermediates
11a–c. The 4-oxo substituent could be removed through
treatment of 11c with phosphorus chloride followed by
reduction to give the quinoline-2-carboxylic acid 14.
The results of a radioligand binding assay for the hu-
man TRa1 and TRb1, as well as a reporter cell assay
employing CHOK1-cells (Chinese hamster ovary cells)
stably transfected with hTRa1 or hTRb1 and an alka-
line phosphatase reporter gene linked to a thyroid
response element (TRAFa1 and TRAFb1), are sum-
marized in Table 1.9 The endogenous hormone, 2,
binds to TRa1 and b1 with an IC50 of 0.24 and
0.26 nM, respectively. Compared with 2, the indole
derivative 6a binds with equal affinity for b1, but 10
times lower for a1, thus resulting in a normalized

1242
A.-M. Garcia Collazo et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1240–1244
Table 1. Thyroid hormone receptor binding affinities (IC50s), reporter
cell line potency efficacies (EC50s), and efficacies of synthetic thyromi-
metics 2–14
of the dysmetabolic syndrome prior to establishing its
therapeutic potential.
ID
TRa,
TRb,
a/bb
TRAFa,
TRAFb,
In order to understand the b1-isoform selectivity of 6b,
the crystal structures of the ligand and the TR-isoforms
(TRa1/6b and TRb1/6b) were obtained to resolutions of
a
a
IC50
nM
,
IC50
nM
,
EC50
nM
,
EC50
nM
,
(% agonism)a (% agonism)a
10
˚
2.55 and 2.30 A, respectively. Data collection and
refinement statistics are given in Table 2. As previously
observed in a pair of crystallographic structures of TRa
and TRb complexed with a phenyl acetic acid thyromi-
metic,3 Arg-228 forms two hydrogen bonds with Ser-277
in TRa, while the corresponding Arg-282 in TRb forms
a strong bifurcated salt bridge with one of the carboxyl-
ate oxygen atoms of 6b. The salt bridge, which is present
in the TRb but not in TRa probably, accounts for the
majority of the observed selectivity of 6b for TRb. An
additional interesting observation is that backbone
Leu-276 to Glu-279 of TRb1/6b is differently oriented
compared with Leu-330 to Glu-333 in TRa1/6b
(Fig. 3). This places the relatively acidic C-a methane
2
0.24
0.26
0.28
0.54 1.3 (100)
3.5 (100)
4.2 (96)
4.7 (83)
4.3 (97)
Nt
6a
6b
10
12a
12b 3500
2.5
5.2
10 (85)
10 (80)
4.1 (95)
Ntc
1.4
0.060 14
8.5 2.6
51
38
250
2.9
2.4
1.5
7.8
870
240
15
Nt
Nt
Nt
Nt
12c
14
620
200
83 (95)
49 (96)
a Values are means of two experiments. The variability was on average
25%.
b For an explanation for the calculation of selectivity ratios, see Ref. 3a.
c Nt = not tested.
˚
hydrogen atom of Asn-331 in TRb within 2.3 A of the
selectivity ratio of 5.2. The halogen substitution at the
R4 and R6-positions affects affinity for both TRa1 and
TRb1. When the bromine atoms of 6a are replaced
with chlorines (6b), binding affinity increase for both
isoforms. However, the binding increased is more pro-
nounced for b1, which resulted in an improved selec-
tivity ratio of 14. It can be observed that 6b binds
approximately four times better than the endogenous
hormone 2 for TRb1. A reverse structure–activity rela-
tionship was observed when the affinities of 12a–c
were compared, in view of the fact that the bromo
analogue 12a is the best binder as well as the most
b1-selective ligand in the particular sub-series. Unmis-
takably the 4-oxo group is detrimental for affinity,
most likely through unfavorable desolvation energies,
since 14 gained almost 20 times the affinity for
TRb1 when compared with 12c. The binding data
described above suggest that 5-membered rings fused
to the inner ring are better accommodated in the
receptor than the 6-membered rings. This trend is even
more striking when the binding data of 10 are consid-
ered. Structural demands on the ligand appear to be
significant as 10 drops both in affinity and b1-selectiv-
ity, when compared with 6a,b. This might be due to a
non-optimal orientation of the carboxylate group in
10. In addition, as shown below, the indole NH of
6b is hydrogen bonded to the backbone carbonyl
group of Met-313, whereas this hydrogen bonding
interaction is eliminated in 10. The assay results of
6a,b and 14 in TRAFa1 and b1 reporter cells reflect
to some degree similar trends as the binding assay,
but 10 displayed the highest potency, which might
be a manifestation of its more hydrophobic nature.
The general discrepancy for the other ligands in
potency of TR-binding and TRAF might be due to
the lack of active transporters in the cultivated trans-
activation assay. This discrepancy was also noted in
the KB-141 series.3a The ligands display similar effica-
cy as 2 for the TRAFb1 assay.
carboxylic oxygen atom of 6b, while the corresponding
distance from Ser-277 in TRa to the ligand carboxylic
˚
oxygen atom is 3.2 A. This favorable electrostatic inter-
action in the b-isoform complex may also contribute to
the selectivity of 6b for TRb.
In summary, we have shown that when the inner ring of
a thyromimetic ligand is fused with its R1 side-chain,
thus forming a bicyclic structure, this promotes TRb-se-
lectivity. The in vitro properties of this class of bicyclic
thyromimetics are highly dependent on the size and
shape of the ring as well as the resulting orientation of
its terminal carboxylate group. Selectivity for the b1-iso-
form also depends on nature of the substituents at the
R3- and R5-positions of the ligand. The X-ray crystallo-
graphic structures of the LBD of TRa1 and TRb1 in
complex with 6b reveal that one source of b1-selectivity
is due to a difference in conformation of b turn loop.
Considering the proximity of the backbones to the
Table 2. Data collection and refinement statistics of complexes human
TRa1LBD + 6b and TRb1LBD + 6b
TRa1LBD + 6b
TRb1LBD + 6b
Space group
Unit cell
P6522
C2221
a = b = 109.8, c = 136.8 a = 77.5, b = 107.5,
c = 67.4
˚
˚
˚
˚
Resolution
Redundancy
Completeness (%) 99.7 (99.6)
2.1 A (2.1–2.21 A)
11.7
2.25 A (2.25–2.33 A)
4.0
98.6 (96.8)
10.0 (2.4)
7.3 (26.7)
hIi/hr(I)i
Rsym (%)
Refinement
Rcryst
20.2 (3.0)
6.9 (39.4)
˚
˚
˚
˚
2.55 A (2.55–2.64 A)
23.5 (34.5)
26.3 (36.0)
2.3 A (2.3–2.44 A)
20.2 (24.9)
23.4 (26.7)
2081
Rfree
Number of atoms 2106
Average B-factor 52.6
39.7
Values in parentheses refer to the highest resolution shells.
Rsym
P
P
=
(jI ꢀ hIij)/ (I), where I is the integrated intensity of a given
reflection and hIi is the average intensity over symmetry equivalents.
P
P
Rcryst
=
calculated amplitudes, respectively. Rfree is calculated similarly using a
(jFo ꢀ Fcj)/j (Fo)j, where Fo and Fc are the observed and
Albeit promising in vitro, the effect of the compounds
needs to be further evaluated in relevant animal models
test set of reflections.

A.-M. Garcia Collazo et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1240–1244
1243
Figure 3. Superposition of the crystallographic structures of 6b complexed with TRbLBD (yellow carbon, red oxygen, and blue nitrogen atoms) and
with TRaLBD (white carbon, pink oxygen, and purple nitrogen atoms). The protein backbone of residues Leu-330 to Glu-333 of TRbLBD + 6b is
depicted as a yellow tube and residues Leu-276 to Glu-279 of TRaLBD + 6b shown in white. Hydrogen bond bonding interactions in TRa and TRb
in pink and red, respectively.
land, J. J.; Doweyko, A. M.; Dejneka, T.; Friends, T.
J.; Devasthalea, P.; Mellstro¨m, K.; Sandberg, J.; Gryn-
ligand, this difference can provide further opportunities
for the design of b1-selective ligands.
˚
farb, M.; Sack, J. S.; Einspahr, H.; Fa¨rnegardh, M.;
Husman, B.; Ljunggren, J.; Koehler, K.; Sheppard, C.;
Malm, J.; Ryono, D. E. Bioorg. Med. Chem. Lett. 2004,
14, 3549.
References and notes
4. Wagner, R. L.; Huber, R. B.; 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. Endocrinol. 2001,
15, 398.
5. Wagner, R. L.; Apriletti, J. W.; McGrath, M. E.; West, B.
L.; Baxter, J. D.; Fletterick, R. J. Nature 1995, 378, 690. In
this context it should be observed that the amino acid
sequence of the ligand binding pocket is identical for rat
and human TR.
6. The phenols 2 and 7a,b were coupled with bis(3-isopropyl-
4-methoxyphenyl)iodonium tetrafluoroborate in good
yields according to the procedure described in: Yokoy-
ama, N.; Walker, G. N.; Main, A. J.; Stanton, J. L.;
Morrissey, M. M.; Boehm, C.; Engle, A.; Neubert, A. D.;
Wasvary, J. M.; Stephan, Z. F.; Steele, R. E. J. Med.
Chem. 1995, 38, 695.
7. These compounds were prepared as in PCT
WO2004018421.
8. Nieman, J. A.; Keay, B. A. Tetrahedron: Asymmetry 1995,
6, 1575.
9. TR-binding assay, vector constructs, generation of report-
er cell lines (TRAF), and assay procedures have been
described in detail in Ref. 3 and in Ô Carlsson, B.; Singh, B.
N.; Temciuc, M.; Nilsson, S.; Li, Y.-L.; Mellin, C.; Malm,
J. J. Med. Chem. 2002, 45, 623,Õ and cited refs.
10. Protein expression, purification, and crystallization was
done according to the description given in Ref. 3. X-ray
data collection and structure determination: data of
1. Lazar, M. A. Endocr. Rev. 1993, 14, 184.
2. The literature suggests many or most effects of thyroid
hormones on the heart, and in particular on the heart rate
and rhythm, are mediated through the TRa isoform, while
most actions of the hormones on the liver and other tissues
are mediated more through the b-forms of the receptor: (a)
Wikstro¨m, L.; Johansson, C.; Salto, C.; Barlow, C.;
Campos Barros, A.; Baas, F.; Forrest, D.; Thoren, P.;
Vennstro¨m, B. EMBO J. 1988, 17, 455; (b) Johansson, C.;
Vennstro¨m, B.; Thoren, P. Am. J. Physiol. 1998, 275,
R640; (c) Forrest, D.; Vennstro¨m, B. Thyroid 2000, 10, 41;
(d) Takeda, K.; Sakurai, A.; DeGroot, L. J.; Refetoff, S. J.
Clin. Endocrinol. Metab. 1992, 74, 49. The potential use of
thyroid hormones for the treatment of metabolic diseases
through TRb-activation was recently reviewed in:(e)
Malm, J. Curr. Pharm. Des. 2004, 10, 3522, and Grover,
G. J.; Malm, J. Drug Discovery Today: Ther. Strategies
2005, 2, 137.
3. (a) Ye, L.; Li, Y.-L.; Mellstro¨m, K.; Mellin, C.; Bladh,
L.-G.; Koehler, K. F.; Garg, N.; Garcia Collazo, A.
M.; Litten, C.; Husman, B.; Persson, K.; Ljunggren, J.;
Grover, G.; Sleph, P. G.; George, R.; Malm, J. J. Med.
Chem. 2003, 45, 1580; (b) Ye, L.; Malm, J.; Li, Y.-L.;
Bladh, L.-G.; Mellstro¨m, K.; Sleph, P. G.; Smith, M.
A.; George, R.; Vennstro¨m, B.; Mookhtiar, K.; Horv-
ath, R.; Speelman, J.; Baxter, J. D.; Grover, G. J. Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 10067; (c) Hange-

1244
A.-M. Garcia Collazo et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1240–1244
TRa1LBD + 6b and TRb1LBD + 6b were collected at binding clefts in which 6b was built. Refinement and
beam line X13, DESY, Hamburg, using a MARCCD
detector. Data were collected at 100 K using a wave-
map calculations were performed with CNX (Brunger,
¨
A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.;
Gros, P.; Grosse-Kunstleve, R. W.; Jiang, J. S.;
Kuszewski, J.; Nilges, M.; Pannu, N. S.; Read, R. J.;
Rice, L. M.; Simonson, T.; Warren, G. L. Acta
Crystallogr., Sect. D 1998, 54, 905); All model building
was done with the program ÔOÕ ( Kjeldgaard. Acta
Crystallogr., Sect. A 1991, 47, 110). The figure of
the ligand binding pocket was produced with Molscript
(Kraulis, P. J. Appl. Crystallogr. 1991, 24, 946,) and
rendered with Raster3D (Merrit, E.; Murphy, M. Acta
Crystallogr., Sect. D 1994, 50, 869).
˚
length of 0.8013 A. Data processing was carried out with
the HKL suite (Otwinowski, Z.; Minor, W. Processing
X-ray Diffraction Data Collected in Oscillation Mode.
In Methods in Enzymology; Academic Press: New York,
1997;
p 307 ); Refinement: starting models for the
refinement of the TRa1 and TRb1 structures were TR
complexes previously solved. In both cases, refinement
was initiated with a round of rigid body refinement
followed by simulated annealing starting at 2000 K.
Well-defined electron density was observed in the ligand