FXR agonist activity of conformationally constrained analogs of GW 4064

Article abstract of DOI:10.1016/j.bmcl.2009.06.062

Two series of conformationally constrained analogs of the FXR agonist GW 4064 1 were prepared. Replacement of the metabolically labile stilbene with either benzothiophene or naphthalene rings led to the identification of potent full agonists 2a and 2g.

Full text of DOI:10.1016/j.bmcl.2009.06.062

Bioorganic & Medicinal Chemistry Letters 19 (2009) 4733–4739
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
FXR agonist activity of conformationally constrained analogs of GW 4064
Adwoa Akwabi-Ameyaw a, , Jonathan Y. Bass a,à, Richard D. Caldwell a,§, Justin A. Caravella b,–, Lihong Chen c,
Katrina L. Creech d, David N. Deaton a, Kevin P. Madauss b, Harry B. Marr e, Robert B. McFadyen a,
f
a
g
Aaron B. Miller b, Frank Navas III a, , Derek J. Parks , Paul K. Spearing , Dan Todd ,
*
Shawn P. Williams b, G. Bruce Wisely f
a Department of Medicinal Chemistry, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
b Molecular Discovery Research, Computational and Structural Chemistry Research, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
c Department of Metabolic Diseases, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
d Molecular Discovery Research, Screening & Compound Profiling, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
e Department of Drug Metabolism and Pharmacokinetics, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
f Molecular Discovery Research, Biological Reagents and Assay Development, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
g Department of Pharmaceutical Development, Physical Properties and Developability, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Two series of conformationally constrained analogs of the FXR agonist GW 4064 1 were prepared.
Replacement of the metabolically labile stilbene with either benzothiophene or naphthalene rings led
to the identification of potent full agonists 2a and 2g.
Received 12 May 2009
Revised 12 June 2009
Accepted 15 June 2009
Available online 21 June 2009
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Farnesoid X receptor
FXR
Nuclear receptor
The farnesoid X receptor (FXR) is a human nuclear receptor that
is expressed in a variety of tissues including liver, gall bladder,
intestines, kidney, and adrenal gland.1,2 Activation of FXR by its
natural ligands, bile acids, influences a variety of biochemical
processes through gene transcription, including bile acid3,4 and
glucose homeostasis,5–7 lipid regulation,8–10 and control of
bacterial growth in the intestines.11–13 The many physiological
roles played by this nuclear receptor suggests that FXR modulators
may be useful in treating a wide range of diseases, including diabe-
tes,7 cholestasis,14–16 liver fibrosis,17–19 atherosclerosis,20,21 and
inflammatory bowel disease.11,12
4064 1 is a useful tool compound, it has issues that complicate its
further development. These liabilities include, poor pharmacoki-
netics in the rat (t1/2 <1 h, Cl = 36 mL/min/kg, F <10%),22,23 a poten-
tially toxic stilbene pharmacophore,24,25 and stilbene-mediated UV
light instability.22 In an attempt to address the various liabilities of
its stilbene functional group, GSK researchers prepared a series of
conformationally constrained naphthoic acid analogs.22 Through
this effort, a potent FXR agonist GSK8062 was identified, which
exhibited improved developability parameters and was efficacious
in the ANIT rat model of cholestasis. Herein, these researchers de-
scribe two new series of conformationally constrained analogs of 1,
wherein the central aromatic ring is linked either to the distal stil-
bene carbon or the proximal stilbene carbon as exemplified by
structures 2a and 3a, respectively. As in GW 4064 1, the com-
pounds in series 2 and 3 contain a similar atom count between
the isoxazole and benzoic acid rings. A key difference, however,
is that the connectors in series 2 contain one less rotatable bond
than series 3 or GW 4064 1. The additional conformational con-
straint in series 2 may provide a boost in potency. In both series,
the newly formed bicyclic rings include 5–6, 6–6, and 6–5 ring
systems. Since the meta-carboxylic acid may not be optimal for
activity in these series, the para-isomers of selected compounds
were also prepared.
GlaxoWellcome scientists have reported the discovery of a po-
tent FXR agonist, GW 4064 1, which has been utilized by numerous
investigators to reveal the biological functions of FXR.9 While GW
* Corresponding author at present address: 609 Churchill Drive, Chapel Hill, NC
27517, USA. Tel.: +1 919 968 9181.
E-mail address: frank.navas3@gmail.com (F. Navas III).
Present address: 409 Sir Walker Lane, Cary, NC 27519, USA.
Present address: Department of Chemistry, University of California, Irvine, CA
à
92697, USA.
§
Present address: Biogen Idec, 14 Cambridge Center, Cambridge, MA 02142, USA.
Present address: Biogen Idec, 14 Cambridge Center, Cambridge, MA 02142, USA.
–
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.06.062

4734
A. Akwabi-Ameyaw et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4733–4739
Br
OH
OH
O
OH
O
O
N
O
O
a-d or e,a
Cl
Cl
R1
8a-8b
S
Cl
R3
N
R2
2g-2h
Scheme 2. Reagents and conditions: (a) 5c or 5d, Pd(PPh3)4, 2 M Na2CO3, PhMe or
DME, ;", 18–36%; (b) BBr3, CH2Cl2, 0 °C, 56%; (c) 7a, PPh3, DIAD, CH2Cl2, 0 °C to rt,
59%; (d) 1 N LiOH, THF, rt, 60%; (e) Cs2CO3, DMF, 65 °C; 7b, DMF, 59%.
O
O
O
Cl
Cl
Cl
O
O
O
N
N
N
The isoindolin-1-one 2i was synthesized as illustrated in
Scheme 3. First, the commercially available methyl 3-aminobenzo-
ate 10a (R2 = NH2, R3 = COOMe) was alkylated with the commer-
cially available methyl 2-bromomethyl-4-methoxy-benzoate 9a
(R1 = OMe) in the microwave, then the resulting aniline was cy-
clized to the isoindolin-1-one 11a (R1 = OMe, R3 = COOMe) under
acid catalysis. Lewis acid catalyzed cleavage of the methyl ether
led to partial hydrolysis of the methyl ester. The mixture of acid
and ester was esterified under Fischer conditions to give the phe-
nol 11a (R1 = OH, R3 = COOMe). Finally, the previous conditions,
for the Mitsunobu coupling and ester hydrolysis, were utilized to
provide the carboxylic acid 2i (R3 = COOH).
2a
1
3a
Cl
Cl
Cl
The preparation of compounds 2a–f is depicted in Scheme 1.26
The commercially available 5-methoxybenzothiophene-2-boronic
acid 4a (R1 = B(OH)2, R2 = H, R3 = OMe, X = S, Y = CH) and 6-meth-
oxybenzothiophene-2-boronic acid 4b (R1 = B(OH)2, R2 = OMe,
R3 = H, X = S, Y = CH) were coupled to ethyl 3-iodobenzoate 5a
(R4 = H, R5 = COOEt, R6 = I) and ethyl 4-iodobenzoate 5b (R4 = COO-
Et, R5 = H, R6 = I), employing the Suzuki protocol, then the methyl
aryl ethers were cleaved with boron tribromide to provide the phe-
nols 6a–d. The phenol 6e was synthesized in a similar manner
from 2-bromo-6-methoxybenzothiazole 4c (R1 = Br, R2 = OMe,
R3 = H, X = S, Y = N) and 3-methoxycarbonylphenylboronic acid 5c
(R4 = H, R5 = COOMe, R6 = B(OH)2). The phenol 6f was synthesized
via Suzuki coupling of 1-tert-butyloxycarbonyl-5-benzyloxy-in-
dole-2-boronic acid 4d (R1 = B(OH)2, R2 = OBn, R3 = H, X = CH,
Y = NCOOC(CH3)3) and ethyl 3-iodobenzoate 5a, followed by
hydrogenolysis of the benzyl ether protecting group. The phenols
6a–f were then alkylated with the known alcohol 7a (R7 = OH) un-
der Mitsunobu conditions, followed by hydrolysis of the esters
(and the tert-butylcarbamate of 2f) to afford the acids 2a–f.
The naphthalene analogs 2g and 2h were prepared as shown in
Scheme 2. Employing a similar strategy to Scheme 1, commercially
available 2-bromo-6-methoxynaphthalene 8a (R1 = OMe) was cou-
pled to 3-methoxycarbonylphenylboronic acid 5c (R4 = H,
R5 = COOMe, R6 = B(OH)2), then demethylated, and coupled to the
alcohol 7a (R7 = OH) to give an ester, which upon subsequent hydro-
lysis provided the carboxylic acid 2g (R2 = COOH, R3 = H). More effi-
ciently, alkylation of 6-bromo-2-naphthol 8b (R1 = OH) with
chloride 7b22 (R7 = Cl), followed by coupling of the resulting aryl
ether with 4-carboxyphenylboronic acid 5d (R4 = COOH, R5 = H,
R6 = B(OH)2) afforded carboxylic acid 2h (R2 = H, R3 = COOH),
directly.
The syntheses of the dihydroisoquinolinones 2j and 2k are
shown in Scheme 4. Copper catalyzed coupling of the commercially
available
6-methoxy-3,4-dihydro-2H-isoquinolin-1-one
12a
(R1 = OMe, X = CH2, Y = N, Z = CO) with the iodide 5a (R4 = H,
R5 = COOEt, R6 = I), followed by methyl ether cleavage and reesteri-
fication gave the methyl ester 13a (R1 = OH, R3 = COOMe, X = CH2,
Z = CO). Conversely, the methyl ester 13b (R1 = OH, R3 = COOMe,
X = CO, Z = CH2) was prepared from the commercially available
6-methoxy-isochroman-3-one 12b (R1 = OMe, X = CO, Y = O,
Z = CH2). Lactone ring opening, bromination, and esterification
with thionyl bromide in methanol, followed by alkylation with ani-
line 10a (R2 = NH2, R3 = COOMe) and acid catalyzed cyclization
afforded the methyl ether, which was deprotected to provide the
phenol 13b (R1 = OH, R3 = COOMe, X = CO, Z = CH2). The phenols
13a–b were then alkylated with alcohol 7a (R7 = OH) under Mits-
unobu conditions, followed by hydrolysis of the esters to afford
the acids 2j and 2k.
The preparation of benzothiazole 2l is illustrated in Scheme 5.
The commercially available 2-amino-6-bromobenzothiazole 14a
was coupled to the boronic acid 5c (R4 = H, R5 = COOMe,
R6 = B(OH)2) via the Suzuki protocol. Then, the aldehyde 7c, de-
rived from the alcohol 7a (R7 = OH) by chromium oxidation, was
coupled to the aniline nitrogen via reductive amination. Finally,
hydrolysis of the methyl ester afforded the carboxylic acid 2l.
The synthesis of the benzoxazole 2m is depicted in Scheme 6.
Suzuki coupling of commercially available 4-bromo-2-nitrophenol
R4
R5
R7
Cl
R4
O
N
R5
R4
Y
X
R3
R5
Cl
R6
R3
R1
N
R3
5a-5c
7a-7b
O
O
Y
X
Y
X
Cl
a, b
c, d
Br O
N
O
O
O
O
R2
10a
N
O
Cl
R3
R2
Cl
2a-2f
R3
R2
N
4a-4d
6a-6f
R1
R1
11a
a-c
d-e
Cl
lw, 150 °C; (b) TFA,
CH3CN, rt; (c) BBr3, CH2Cl2, 0 °C to rt; SOCl2, MeOH, ";, 24% over three steps; (d) 7a,
9a
2i
Scheme 1. Reagents and conditions: (a) 5a–c, 2 M Na2CO3, Pd(PPh3)4, PhMe or
DME, ";, 36–89%; (b) BBr3, CH2Cl2, 0 °C, 32–77% or H2, Pd-C, EtOH, EtOAc, rt, 74%; (c)
7a, Ph3P or Ph3P-PS, DIAD, CH2Cl2 or PhMe, rt, 44–66%; (d) LiOH or NaOH, THF or
dioxane, MeOH or EtOH, H2O, rt, 18–99%.
Scheme 3. Reagents and conditions: (a) 10a, Et3N, DMF,
PPh3, DIAD, rt, 37%; (e) 1 N LiOH, 1,4-dioxane, rt, 72%.

A. Akwabi-Ameyaw et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4733–4739
4735
Table 1
Activation of human FXR
R3
N
X
N
R3
Z
O
N
R =
X
Y
X
O
Z
Z
Cl
f, g
a,b or c,d,e,b
Cl
O
Cl
2a 2m
-
N
R1
12a-12b
R1
Cl
13a-13b
2j-2k
#
R
FXR FRET EC50
nMa (%maxb)
FXR TT EC50 nMc
(%maxb)
Scheme 4. Reagents and conditions: (a) 5a, CuI, K2CO3, DMF, 150 °C, 55%; (b) BBr3,
CH2Cl2, 0 °C to rt, 68%; SOCl2, MeOH, ;", 87%; (c) SOBr2, MeOH, PhMe, rt, 90%; (d)
10a, Et3N, PhMe, 90 °C; (e) p-TsOHꢁH2O, PhMe, 100 °C; (f) 7a, PS-PPh3, DIAD, CH2Cl2,
O
O
rt, 65%; (g) 1 N LiOH, MeOH, THF,
l
w, 100 °C, 89%.
1
59 (100)
55 (90)
65 (100)
32 (87)
HO
HO
Cl
Br
O
O
OH
H
N
a-c
N
O
S
O
2a
Cl
S
N
N
Cl
S
S
14a NH2
2l
Scheme 5. Reagents and conditions: (a) 5c, 2 M Na2CO3, Pd(PPh3)4, DME, 85 °C,
26%; (b) 7c, Bu2SnCl, Ph3SiH, rt to 75 °C, 9%; (c) LiOH, THF, rt, 59%.
O
O
2b
2c
280 (80)
140 (92)
160 (89)
63 (93)
S
S
HO
15a with boronic acid 5c (R4 = H, R5 = COOMe, R6 = B(OH)2), fol-
lowed by hydrogenation of the nitro group afforded the aniline
16a. Then, the acid 7d (R7 = CH2COOH), derived from aldehyde 7c
(R7 = O) by aldolization with Meldrum’s acid and in situ hydrolysis,
was activated as the mixed anhydride and coupled to aniline 16a,
followed by acid catalyzed ring closure and ester hydrolysis to give
the acid 2m.
O
HO
O
O
O
2d
2e
200 (95)
410 (72)
220 (79)
410 (86)
The structure–activity relationships for series 2 (analogs 2a–m)
are shown in Table 1. The benzothiophene analogs 2a–d were all
full FXR agonists (%max >80% of GW 4064 1) in both the fluores-
cence resonance energy transfer (FRET) assay for the recruitment
of the SRC-1 co-activator peptide to the AF-2 helix of FXR and
the transient transfection (TT) assay for the FXR-mediated tran-
scription of a luciferase reporter. The meta-substituted carboxylic
acid compounds 2a (TT EC50 = 32 nM) and 2c (TT EC50 = 63 nM)
were slightly more potent than the corresponding para-carboxylic
acid derivatives 2b (TT EC50 = 160 nM) and 2d (TT EC50 = 220 nM),
possibly due to a greater distance separating the negative charge of
the carboxylic acid from its positively charged guanidine interac-
tion partner on 331Arg in the FXR ligand binding domain (LBD)
(vide infra). Moreover, 2a and 2c are equipotent to the GW 4064
1 lead (TT EC50 = 65 nM). Thus, the metabolic liabilities of the stil-
bene moiety in GW 4064 1 can be overcome by its replacement
HO
O
O
HO
HO
O
O
S
N
2f
480 (81)
110 (88)
630 (103)
130 (103)
N
H
O
O
O
2g
HO
O
O
HO
O
2h
440 (68)
980 (170)
N
HO
HO
Br
O
O
a, b
c-e
O
N
Cl
O
O2N
OH
H2N
OH
O
Cl
15a
16a
2m
2i
670 (59)
1100 (61)
N
Scheme 6. Reagents and conditions: (a) 5c, Pd(PPh3)4, DME, 2 M Na2CO3, 85–95 °C;
H2SO4, MeOH, ";, 14%; (b) Pd/C, H2, EtOH, 100%; (c) 7d, i-BuOCOCl, NEt3, CH2Cl2, ꢀ5
to ꢀ15 °C to rt, 85%; (d) propionic acid, 135–150 °C; (e) 1 N LiOH, THF, rt, 3% over
two steps.
O
(continued on next page)

4736
A. Akwabi-Ameyaw et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4733–4739
R2
Table 1 (continued)
R3
R3
Z
FXR TT EC50 nMc
(%maxb)
c-d
#
R
FXR FRET EC50
nMa (%maxb)
R2
Y
OH
W
a-b
or
19a-19c
W
O
O
Y
Z
O
O
e, b
2j
710 (52)
8300 (112)
>10,000
N
N
HO
a, d
or
O
O
N
Y
Z
c
a, g
O
O
Cl
O
Cl
Cl
O
R1
HN
2k
2l
1100 (75)
N
HO
HO
17a-17c
20a
3a-3e
Cl
e, d
R2
R3
O
O
O
H
N
N
X
S
N
W
Cl
960 (91)
380 (123)
710 (75)
O
R2
W
18a-18e
N
O
Z
Cl
R3
f
18d
18e
3f
Y
Scheme 7. Reagents and conditions: (a) 18a, 18b, 18c, or 18e, NaH, DMF, 29–99%;
(b) H2, 10% Pd/C, EtOH, EtOAc, 66–99% or BBr3, CH2Cl2, 0 °C, 65%; (c) 7a, Ph3P or
Ph3P-PS, DIAD, CH2Cl2, rt, or 7b, Cs2CO3, DMF, 65 °C, 30–71%; (d) LiOH or NaOH, THF
or dioxane, MeOH or EtOH, H2O, rt or 100 °C, 10–74%; (e) 18a, (n-Bu)4NI, Zn(OTf)2, i-
Pr2NEt, PhMe, rt, 53–59%; (f) (COCl)2, CH2Cl2, DMF, 0 °C, 99%; (g) TFA, CH2Cl2, 0 °C,
50%.
HO
2m
1200 (41)
a
FXR ligand seeking assay measuring ligand-mediated interaction of the SRC-1
peptide (B-CPSSHSSLTERHKILHRLLQEGSPS-CONH2) with the FXR 237-472LBD, using
5 nM biotinylated FXR LBD coupled to 5 nM allophycocyanin-labeled streptavidin
and 10 nM biotinylated SRC-1 coupled to 5 nM Europium-labeled streptavidin as
reagents in 10 mM DTT, 0.1 g/L BSA, 50 mM NaF, 50 mM MOPS, 1 mM EDTA, and
The indole analogs 3a–f were prepared as shown in Scheme 7.
Coupling of 5-benzyloxyindole 17a (R1 = OCH2Ph, Y = NH, Z = CH)
to methyl 3-bromomethylbenzoate 18a (R2 = H, R3 = COOMe,
W = CH2, X = Br), followed by palladium mediated hydrogenolysis
of the benzyl ether afforded the phenol 19a (R2 = H, R3 = COOMe,
W = CH2, Y = N, Z = CH). Alkylation of the phenol 19a with the chlo-
ride 7b (R7 = Cl), followed by hydrolysis of the ester provided the
indole 3a (R2 = H, R3 = COOH, W = CH2, Y = N, Z = CH). Alternatively,
the indole 3b was prepared by alkylation of 5-hydroxyindole 17b
(R1 = OH, Y = NH, Z = CH) with alcohol 7a via the Mitsunobu proto-
col to afford the aryl ether 20a, which was alkylated with methyl 4-
(bromomethyl)benzoate 18b (R2 = COOMe, R3 = H, W = CH2,
X = Br), followed by hydrolysis of the ester to give the acid 3b
(R2 = COOH, R3 = H, W = CH2, Y = N, Z = CH). Moreover, the isomeric
indole 3c was prepared from 6-benzyloxyindole 17c (R1 = OCH2Ph,
Y = CH, Z = NH) via Lewis acid promoted alkylation with 18a, fol-
lowed by removal of the benzyl protecting group, to afford phenol
19b (R2 = H, R3 = COOMe, W = CH2, Y = C, Z = NH). Then, similarly to
the synthesis of 3a, alkylation with 7a and hydrolysis of the ester
gave isomeric indole 3c (R2 = H, R3 = COOH, W = CH2, Y = C,
Z = NH). The sulfonamide 3d was prepared by sulfonylation of in-
dole 17a with the sulfonyl chloride 18c (R2 = H, R3 = COOMe,
W = SO2, X = Cl), followed by boron tribromide mediated cleavage
of the benzyl ether to yield the phenol 19c (R2 = H, R3 = COOMe,
W = SO2, Y = N, Z = CH). Subsequent alkylation of phenol 19c with
alcohol 7a and hydrolysis as above afforded acid 3d (R2 = H,
R3 = COOH, W = SO2, Y = N, Z = CH). Amide 3e was prepared by
acylation of indole 20a with acid chloride 18e (R2 = H, R3 = COOMe,
W = CO, X = Cl), followed by hydrolysis, to give the indole 3e
(R2 = H, R3 = COOH, W = CO, Y = N, Z = CH). Acid chloride 18e was
prepared from commercially available acid 18d (R2 = H,
R3 = COOt-Bu, W = CO, X = OH) employing oxalyl chloride. Finally,
Lewis acid promoted alkylation of indole 20a with bromide 18a,
followed by hydrolysis, gave the 3-substituted indole 3f (R2 = H,
R3 = COOH, W = CH2, Y = NH, Z = C).
50 lM CHAPS, at pH 7.5. The EC50 values are the mean of at least two assays.
b
Maximum percent efficacy of the test compound relative to FXR activation via
GW 4064 1.
c
FXR transient transfection assay measuring the ligand-mediated luminescense
resulting from FXR-induced transcription of a luciferase reporter. FXR and the
luciferase reporter genes are transfected into African green monkey CV-1 kidney
cells, then treated with test compound. The EC50 values are the mean of at least two
assays.
with a benzothiophene, with no loss in activity or efficacy. In con-
trast, the benzothiazole 2e (TT EC50 = 410 nM) is less potent than
its corresponding benzothiophene 2c, likely due to the entropic
penalty paid to desolvate the nitrogen heteroatom of 2e, relative
to the carbon in 2c, as it binds into a hydrophobic region of the
FXR LBD (vide infra). The reversed benzothiazole 2l (TT
EC50 = 380 nM) has similar FXR affinity as benzothiazole 2e. Simi-
larly, the nitrogen containing indole analog 2f (TT EC50 = 630 nM)
and the nitrogen and oxygen-containing benzoxazole analog 2m
(TT EC50 = 710 nM) are also less active than the benzothiophene 2c.
The naphthalene analogs 2g (TT EC50 = 130 nM) and 2h (TT
EC50 = 980 nM) are slightly less potent than the corresponding ben-
zothiophene analogs 2a and 2b, with the para-derivative 2h once
again being less active than the meta-derivative 2g. Thus, for series
2, 3-carboxylic acid substitution is optimal. In contrast to the more
narrow angle of substituent display by the benzothiophene linker
in 2a, the naphthalene linker in analog 2g (TT EC50 = 130 nM) dis-
plays its appending groups at ꢂ180°. This may force slight altera-
tions in the FXR LBD, possibly explaining the diminished potency
of 2g relative to 2a.
In contrast to the fully aromatic ring system linkers in analogs
2a–h, the isoindolinone 2i (TT EC50 = 1100 nM), and the dihydro-
isoquinolinones 2j (TT EC50 = 8300 nM) and 2k (TT EC50
>10,000 nM) are greater than an order of magnitude less potent
than the benzothiophene 2a. In addition to the higher entropic cost
to desolvate the additional heteroatom(s) in these derivatives, the
loss of planarity due to partial ring saturation alters the angle of
display of the linker ring system substituents, possibly inducing
strain in the protein as it accommodates these ligands in the
LBD, thus diminishing receptor affinity.
=
The preparation of benzimidazole analog 3g is depicted in
Scheme 8. First, alkylation of aniline 21a with bromide 18a, then
reduction of the nitro group afforded an ortho-dianiline. This inter-
mediate was converted to a benzimidazole with formic acid and

A. Akwabi-Ameyaw et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4733–4739
4737
O
OH
OH
O
O
O
O
O
NH
O
O
N
N
O
O
O
N
Br
H2N
10b
O
Br
N
NH
18a
Cl
N
Cl
OH
NH2
NO2
a-c
d-f
OH
a-d
e-f
O
O
N
N
NO2
O
24a
25a NH2
3i
Cl
21a
19d
3g
OH
Cl
Scheme 10. Reagents and conditions: (a) benzyl bromide, K2CO3, CH3CN, 70 °C,
94%; (b) 10b, Pd2(dba)3, BINAP, Cs2CO3, toluene, 100 °C; (c) H2, 10% Pd/C, EtOH; (d)
7d, DCC, HOBt, CH3CN, rt; (e) CH3CH2COOH, 130–150 °C, 7.4% over three steps; (f)
1 N LiOH, THF, rt to 60 °C, 57%.
Scheme 8. Reagents and conditions: (a) 18a, K2CO3, DMF, 110 °C; H2, 10% Pd/C,
EtOH, rt, 21%; (b) SnCl2ꢁ2H2O, EtOH, reflux, 92%; (c) HCOOH, rt, 86%; (d) BBr3,
CH2Cl2, 0 °C; MeOH, H2SO4, reflux, 42%; (e) 7a, PPh3, DIAD, CH2Cl2, 0 °C to rt; (f) 1 N
LiOH, 1,4-dioxane, rt, 25%.
the methyl ether was cleaved to provide the phenol 19d. Mitsun-
obu coupling of 19d with the alcohol 7a and basic hydrolysis of
the ester yielded the acid 3g.
Table 2
Activation of human FXR
As illustrated in Scheme 9, the naphthalene analog 3h was pre-
pared from the naphthol 22a. First, activation of the hydroxy group
as the triflate, then Hartwig–Buchwald coupling with the aniline
10b and boron tribromide mediated hydrolysis of the methyl ether
gave the phenol 23a. Then, O-alkylation of the aniline phenol 23a
with alcohol 7a via the Mitsunobu protocol provided the ester,
which was hydrolyzed to yield the acid 3h.
The synthesis of the benzoxazole analog 3i is shown in Scheme
10. Starting from the commercially available bromide 24a, protec-
tion of the phenol as the benzyl ether, then Hartwig–Buchwald
coupling with the aniline 10b, followed by palladium catalyzed
reduction of the nitro substituent with concomitant cleavage of
the benzyl ether afforded the aniline phenol 25a. Then, coupling
of the acid 7d with the aniline 25a provide an amide, which was
cyclized to the benzoxazole via acid catalysis. Finally, base-pro-
moted hydrolysis of the ethyl ester yielded the acid 3i.
O
N
R =
Cl
Cl
3a-3i
#
R
FXR FRET EC50
nMa (%maxb)
FXR TT EC50 nMc
(%maxb)
O
O
O
1
59 (100)
320 (80)
65 (100)
210 (84)
HO
HO
Cl
O
3a
N
N
The structure–activity relationships for series 3 (analogs 3a–i)
are shown in Table 2. For series 3, none of the analogs were more
potent than the original lead GW 4064 1. The meta- and para-car-
boxy-phenylmethyl-indole analogs 3a (TT EC50 = 210 nM) and 3b
(TT EC50 = 330 nM) were slightly less potent FXR agonists than
O
GW 4064 1, while the regioisomeric indole analog 3c (TT EC50
=
3b
3c
3d
3e
3f
160 (80)
590 (81)
160 (87)
150 (68)
1500 (25)
330 (79)
930 (32)
1300 (63)
860 (63)
>10,000
HO
HO
930 nM, %max = 32) and benzimidazole 3g (TT EC50 = 5000 nM,
%max = 40) were at least an order of magnitude less active and
partial agonists as well. The decreased potency of these analogs
is possibly due to less optimal fit of the indole spacer in the FXR
O
O
O
O
O
NH
HO
O
O
O
O
O
S
N
HO
HO
NH
O
O
O
O
NH2
O
N
OH
NH
O
10b
Cl
a-c,
d, e
O
N
O
22a
23a
3h
OH
Cl
O
HO
Scheme 9. Reagents and conditions: (a) Tf2O, pyridine, CH2Cl2, 0 °C, 81%; (b) 10b,
Pd2(dba)3, BINAP, 2 M Cs2CO3, PhMe, ";, 70%; (c) BBr3, CH2Cl2, 0 °C to rt, 35%; (d) 7a,
Ph3P, DIAD, CH2Cl2, rt, 37%; (e) LiOH, THF, H2O, rt to 60 °C, 54%.
N
O
H
(continued on next page)

4738
A. Akwabi-Ameyaw et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4733–4739
Table 2 (continued)
#
R
FXR FRET EC50
nMa (%maxb)
FXR TT EC50 nMc
(%maxb)
O
O
O
3g
1400 (38)
320 (95)
1700 (19)
5000 (40)
81 (57)
N
HO
HO
N
O
O
H
N
3h
HO
H
N
O
3i
>10,000
N
Figure 1. Superimposed ligand binding domains of the X-ray co-crystal structures
of agonist 2c (ligand 2c carbons colored green) and GW 4064 1 (ligand 1 carbons
colored magenta) complexed with FXR. The FXR carbons from the co-crystal
structure with agonist 2c are colored gray, with the FXR ‘shifted’ residues from the
co-crystal structure with GW4064 1 colored magenta. The hydrogen bonds are
depicted as yellow dashed lines. The coordinates have been deposited in the
Brookhaven Protein Data Bank (1 PDB code 3DCT, 2c PDB code 3HC5). This figure
was generated using PYMOL version 1.0 (Delano Scientific, www.pymol.org).
a
FXR ligand seeking assay measuring ligand-mediated interaction of the SRC-1
peptide (B-CPSSHSSLTERHKILHRLLQEGSPS-CONH2) with the FXR 237-472LBD, using
5 nM biotinylated FXR LBD coupled to 5 nM allophycocyanin-labeled streptavidin
and 10 nM biotinylated SRC-1 coupled to 5 nM Europium-labeled streptavidin as
reagents in 10 mM DTT, 0.1 g/L BSA, 50 mM NaF, 50 mM MOPS, 1 mM EDTA, and
50 lM CHAPS, at pH 7.5. The EC50 values are the mean of at least two assays.
b
Maximum percent efficacy of the test compound relative to FXR activation via
GW 4064 1.
c
In Figure 2, the 3.2 Å X-ray co-crystal structure of indole 3a is
overlaid with benzothiophene 2c. The 3-phenylisoxazole rings of
the two compounds fit into the ligand binding domain in a very
similar manner. However, in other respects, these two compounds
bind to the receptor in distinctly different ways. For example, as
noted above, the oxygen atoms of the benzothiophene carboxylic
FXR transient transfection assay measuring the ligand-mediated luminescense
resulting from FXR-induced transcription of a luciferase reporter. FXR and the
luciferase reporter genes are transfected into African green monkey CV-1 kidney
cells, then treated with test compound. The EC50 values are the mean of at least two
assays.
LBD (vide infra). In addition, the larger desolvation penalties to re-
move water from the indole NH of 3c and extra nitrogen of benz-
imidazole 3g relative to indole 3a, likely explain the further loss
in activity. Indole analogs 3d (TT EC50 = 1300 nM) and 3e (TT
EC50 = 860 nM), where the methylene connector of 3a is replaced
by a sulfonamide or amide linker, were also less potent and effica-
cious than 3a, further supporting the desolvation cost hypothesis
for the spacer region.
The 3,5-disubstituted indole 3f (TT EC50 = >10,000 nM) and the
2,7-disubstituted benzoxazole 3i (TT EC50 = >10,000 nM) display
their carboxylic acid and isoxazole moieties at narrower angles
than the 1,5- and 3,6-disubstituted analogs like 3a and 3c, respec-
tively, and are poor FXR agonists. In contrast, naphthalene 3h (TT
EC50 = 81 nM, %max = 57) places its substituents in similar envi-
ronments as indole 3a and is nearly as potent as the starting lead
1, although it is a partial agonist.
acid coordinate with the terminal NH2 and e-NH of the guanidine
group of 331Arg. In contrast, one oxygen atom of the carboxylic acid
of indole 3a forms both of these interactions while the second oxy-
gen atom coordinates with the proximal nitrogen of the imidazole
of 294His (One oxygen atom of the carboxylic acid 3a is 2.6 Å from
the NH2 nitrogen of 331Arg, and 2.9 Å from the
e-NH nitrogen of
331Arg, while the other oxygen atom is 3.2 Å from the nitrogen of
294His). The additional electrostatic interaction with 294His forces
the carboxylic acid of indole 3a out of the plane of the ring. Another
X-ray co-crystal structures of benzothiophene 2c and indole 3a
with human FXR are shown in Figures 1 and 2, respectively. In
Figure 1, the 2.6 Å X-ray co-crystal structure of benzothiophene
2c is overlaid with the structure of GW 4064 1. The two com-
pounds fit very similarly into the ligand binding domain of the
receptor, which may account for the similar potency of the two
compounds. In both structures, the two oxygen atoms of the
carboxylic acid coordinate to the terminal NH2 and e-NH of the
guanidine group of 331Arg (One oxygen atom of the carboxylic acid
2c is 3.1 Å from the NH2 nitrogen of 331Arg, while the other oxygen
atom is 2.9 Å from the NH2 nitrogen of 331Arg and 2.8 Å from the
e
-NH nitrogen of 331Arg). Also, the carboxylic acids are planar with
the aromatic rings, which are in turn, planar with the stilbene
(GW 4064 1) and benzothiophene ring 2c. Moreover, the isoxazole
ring of the two compounds form edge to face stacking interactions
with 454Trp located on helix 12. The chlorine atoms of the 3-pheny-
lisoxazole force the phenyl ring to twist out of plane relative to the
isoxazole ring. This conformation places the aromatic ring between
328Met and 365Met on helices 5 and 7, respectively.
Figure 2. Superimposed ligand binding domains of the X-ray co-crystal structures
of agonist 2c (ligand 2c carbons colored green) and agonist 3a (ligand 3a carbons
colored cyan) complexed with FXR. The FXR carbons from the co-crystal structure
with agonist 2c are colored gray, with the FXR ‘shifted’ residues from the co-crystal
structure with agonist 3a colored cyan. The hydrogen bonds are depicted as yellow
dashed lines. The coordinates have been deposited in the Brookhaven Protein Data
Bank (2c PDB code 3HC5, 3a PDB code 3HC6). This figure was generated using
PYMOL version 1.0 (Delano Scientific, www.pymol.org).

A. Akwabi-Ameyaw et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4733–4739
4739
Table 3
ner very similar to GW 4064 1. None of the compounds in series 3
were more potent than GW 4064 1.
Pharmacokinetics of FXR agonists in rats
a
b
d
#
t1/2 min
Clc (mL/min/kg)
VSS (mL/kg)
Fe (%)
2a
3a
3g
15
45
31
66
6.7
4.4
720
70
85
8.5
12
26
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.06.062.
a
Rats were dosed at 1 mg/kg (iv) and 5 mg/kg (po).
t1/2 is the iv terminal half-life dosed as a solution. All in vivo pharmacokinetic
b
values are the mean of two experiments.
c
Cl is the iv total clearance.
VSS is the iv steady state volume of distribution.
F is the oral bioavailability.
References and notes
d
e
1. Forman, B. M.; Goode, E.; Chen, J.; Oro, A. E.; Bradley, D. J.; Perlmann, T.;
Noonan, D. J.; Burka, L. T.; McMorris, T.; Lamph, W. W.; Evans, R. M.;
Weinberger, C. Cell 1995, 81, 687.
difference in the binding modes of these two compounds is the
spatial relationship between the benzoic acid ring and the adjoin-
ing benzothiophene ring of 2c and indole ring of 3a. In the case of
benzothiophene analog 2c, these rings are co-planar, while in the
crystal structure of indole 3a, not only are the benzoic acid and in-
dole rings out of plane, but the indole ring is also forced deeper into
the pocket. The greater potency of benzothiophene 2c relative to
indole 3a indicates that a co-planar ligand may more effectively
stabilize the active protein conformer, wherein co-repressors are
shed and co-activators are recruited, thereby promoting gene tran-
scription. Alternatively, although indole 3a is non-planar, the li-
gand causes little perturbation of the FXR ligand binding pocket,
yet its active site conformation, to achieve optimal interactions be-
tween its carboxylic acid and 331Arg and 294His of the FXR ligand
binding pocket, is of higher energy than its bulk solution conforma-
tion. This free energy cost could account for its lower potency rel-
ative to 2c, as well as the reduced activity of all series 3 analogs.
Several compounds from series 2 and 3 were evaluated in phar-
macokinetic studies in rats. As shown in Table 3, benzothiophene
2a had a very high clearance (Cl = 66 mL/min/kg), which was sim-
ilar to the hepatic blood flow in the rat. The high clearance of this
compound, in conjunction with a low volume of distribution
(VSS = 720 mL/kg), resulted in a very short half-life (t1/2 = 15 min).
The benzothiophene 2a was also poorly bioavailable (F = 8.5%). In-
dole 3a had a significantly lower clearance (Cl = 6.7 mL/min/kg)
than the benzothiophene 2a, however, the resulting improvement
in terminal half-life (t1/2 = 45 min) was modest due to a very low
volume of distribution (VSS = 70 mL/kg). The oral bioavailability
(F = 12%) of the indole 3a was low and only slightly better than that
of the benzothiophene. The poor bioavailability of 2a and 3a may
have been due, in part, to their poor aqueous solubility. In an at-
tempt to improve solubility, a second nitrogen atom was incorpo-
rated into the indole ring of 3a to give the benzimidazole 3g. This
change resulted in a twofold improvement in bioavailability
(F = 26%) with little alteration of clearance or terminal half-life.
Although the terminal half-lives of these analogs did not improve
relative to GW 4064 1, the increased oral bioavailability of 3g sug-
gests that absorption of GW 4064 analogs may be dissolution lim-
ited and could be increased by solubility enhancing formulations.
In summary, two series of conformationally constrained analogs
of GW 4064 1 were prepared. Benzothiophene analogs 2a and 2c
and naphthalene 2g were potent, full FXR agonists from the more
highly constrained series 2. An X-ray co-crystal structure of benzo-
thiophene 2c indicates that it binds to the FXR receptor in a man-
2. Higashiyama, H.; Kinoshita, M.; Asano, S. Acta Histochem. 2008, 110, 86.
3. Rizzo, G.; Renga, B.; Mencarelli, A.; Pellicciari, R.; Fiorucci, S. Curr. Drug Targets:
Immune, Endocrinol. Metab. Disord. 2005, 5, 289.
4. Sinal, C. J.; Tohkin, M.; Miyata, M.; Ward, J. M.; Lambert, G.; Gonzalez, F. J. Cell
2000, 102, 731.
5. Cariou, B.; van Harmelen, K.; Duran-Sandoval, D.; van Dijk, T.; Grefhorst, A.;
Bouchaert, E.; Fruchart, J.-C.; Gonzalez, F. J.; Kuipers, F.; Staels, B. FEBS Lett.
2005, 579, 4076.
6. Ma, K.; Saha, P. K.; Chan, L.; Moore, D. D. J. Clin. Invest. 2006, 116, 1102.
7. Zhang, Y.; Lee, F. Y.; Barrera, G.; Lee, H.; Vales, C.; Gonzalez, F. J.; Willson, T. M.;
Edwards, P. A. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 1006.
8. Lambert, G.; Amar, M. J. A.; Guo, G.; Brewer, H. B., Jr.; Gonzalez, F. J.; Sinal, C. J. J.
Biol. Chem. 2003, 278, 2563.
9. Maloney, P. R.; Parks, D. J.; Haffner, C. D.; Fivush, A. M.; Chandra, G.; Plunket, K.
D.; Creech, K. L.; Moore, L. B.; Wilson, J. G.; Lewis, M. C.; Jones, S. A.; Willson, T.
M. J. Med. Chem. 2000, 43, 2971.
10. Watanabe, M.; Houten, S. M.; Wang, L.; Moschetta, A.; Mangelsdorf, D. J.;
Heyman, R. A.; Moore, D. D.; Auwerx, J. J. Clin. Invest. 2004, 113, 1408.
11. Inagaki, T.; Choi, M.; Moschetta, A.; Peng, L.; Cummins, C. L.; McDonald, J. G.;
Luo, G.; Jones, S. A.; Goodwin, B.; Richardson, J. A.; Gerard, R. D.; Repa, J. J.;
Mangelsdorf, D. J.; Kliewer, S. A. Cell Metab. 2005, 2, 217.
12. Inagaki, T.; Moschetta, A.; Lee, Y.-K.; Peng, L.; Zhao, G.; Downes, M.; Yu, R. T.;
Shelton, J. M.; Richardson, J. A.; Repa, J. J.; Mangelsdorf, D. J.; Kliewer, S. A. Proc.
Natl. Acad. Sci. U.S.A. 2006, 103, 3920.
13. Jung, D.; Inagaki, T.; Gerard, R. D.; Dawson, P. A.; Kliewer, S. A.; Mangelsdorf, D.
J.; Moschetta, A. J. Lipid Res. 2007, 48, 2693.
14. Fiorucci, S.; Clerici, C.; Antonelli, E.; Orlandi, S.; Goodwin, B.; Sadeghpour, B. M.;
Sabatino, G.; Russo, G.; Castellani, D.; Willson, T. M.; Pruzanski, M.; Pellicciari,
R.; Morelli, A. J. Pharmacol. Exp. Ther. 2005, 313, 604.
15. Liu, Y.; Binz, J.; Numerick, M. J.; Dennis, S.; Luo, G.; Desai, B.; MacKenzie, K. I.;
Mansfield, T. A.; Kliewer, S. A.; Goodwin, B.; Jones, S. A. J. Clin. Invest. 2003, 112,
1678.
16. Stedman, C.; Liddle, C.; Coulter, S.; Sonoda, J.; Alvarez, J. G.; Evans, R. M.;
Downes, M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 11323.
17. Fiorucci, S.; Antonelli, E.; Rizzo, G.; Renga, B.; Mencarelli, A.; Riccardi, L.;
Orlandi, S.; Pellicciari, R.; Morelli, A. Gastroenterology 2004, 127, 1497.
18. Fiorucci, S.; Rizzo, G.; Antonelli, E.; Renga, B.; Mencarelli, A.; Riccardi, L.;
Morelli, A.; Pruzanski, M.; Pellicciari, R. J. Pharmacol. Exp. Ther. 2005, 315, 58.
19. Fiorucci, S.; Rizzo, G.; Antonelli, E.; Renga, B.; Mencarelli, A.; Riccardi, L.;
Orlandi, S.; Pruzanski, M.; Morelli, A.; Pellicciari, R. J. Pharmacol. Exp. Ther. 2005,
314, 584.
20. Hanniman, E. A.; Lambert, G.; McCarthy, T. C.; Sinal, C. J. J. Lipid Res. 2005, 46,
2595.
21. Zhang, Y.; Wang, X.; Vales, C.; Lee, F. Y.; Lee, H.; Lusis, A. J.; Edwards, P. A.
Arterioscler., Thromb., Vasc. Biol. 2006, 26, 2316.
22. Akwabi-Ameyaw, A.; Bass, J. Y.; Caldwell, R. D.; Caravella, J. A.; Chen, L.; Creech,
K. L.; Deaton, D. N.; Jones, S. A.; Kaldor, I.; Liu, Y.; Madauss, K. P.; Marr, H. B.;
McFadyen, R. B.; Miller, A. B.; Navas, F., III; Parks, D. J.; Spearing, P. K.; Todd, D.;
Williams, S. P.; Wisely, G. B. Bioorg. Med. Chem. Lett. 2008, 18, 4339.
23. Bass, J. Y.; Caldwell, R. D.; Caravella, J. A.; Chen, L.; Creech, K. L.; Deaton, D. N.;
Madauss, K. P.; Marr, H. B.; McFadyen, R. B.; Miller, A. B.; Parks, D. J.; Todd, D.;
Williams, S. P.; Wisely, G. B. Bioorg. Med. Chem. Lett. 2009, 19, 2969.
24. Kuo, C. H.; Hook, J. B.; Bernstein, J. Toxicology 1981, 22, 149.
25. Sugihara, K.; Kitamura, S.; Sanoh, S.; Ohta, S.; Fujimoto, N.; Maruyama, S.; Ito,
A. Toxicol. Appl. Pharmacol. 2000, 167, 46.
26. Akwabi-Ameyaw, A. A.; Deaton, D. N.; McFadyen, R. B.; Navas, F., III. PCT Int.
Appl. WO 005998, 2009; Chem. Abstr. 2009, 150, 121632.