Optimization of triazole-based TGR5 agonists towards orally available agents

Article abstract of DOI:10.1039/c2md20174g

Reported herein is a medicinal chemistry effort towards the identification of orally available TGR5 agonist 12, which served as a dog tool compound for studies to increase confidence in this mechanism. With the challenge of striking the balance of TGR5 potency and desired clearance profile, the screening strategy as well as medicinal chemistry strategy are discussed in this article. The Royal Society of Chemistry 2013.

Full text of DOI:10.1039/c2md20174g

Dynamic Article Links <
C
MedChemComm
Cite this: DOI: 10.1039/c2md20174g
www.rsc.org/medchemcomm
View Article Online
CONCISE ARTICLE
View Journal
Optimization of triazole-based TGR5 agonists towards orally available
agents†
*
Kentaro Futatsugi, Kevin B. Bahnck, Martin B. Brenner, Joanne Buxton, Janice E. Chin, Steven B. Coffey,
Jeffrey Dubins, Declan Flynn, Denise Gautreau, Angel Guzman-Perez, John R. Hadcock, David Hepworth,
Michael Herr, Terri Hinchey, Ann M. Janssen, Sandra M. Jennings, Wenhua Jiao, Sophie Y. Lavergne,
Bryan Li, Mei Li, Michael J. Munchhof, Suvi T. M. Orr, David W. Piotrowski, Nicole S. Roush,
Matthew Sammons, Benjamin D. Stevens, Gregory Storer, Jian Wang, Joseph S. Warmus, Liuqing Wei
and Angela C. Wolford
Received 30th June 2012, Accepted 31st August 2012
DOI: 10.1039/c2md20174g
Reported herein is a medicinal chemistry effort towards the identification of orally available TGR5
agonist 12, which served as a dog tool compound for studies to increase confidence in this mechanism.
With the challenge of striking the balance of TGR5 potency and desired clearance profile, the screening
strategy as well as medicinal chemistry strategy are discussed in this article.
We wished to further investigate the therapeutic potential of
Introduction
non-bile acid TGR5 agonists for the treatment of metabolic
TGR5 is a class A G protein-coupled receptor expressed in liver,
diseases and for this work we sought a tool compound to
skeletal muscle, intestine, brown adipose tissue, as well as
perform studies for increasing confidence in the mechanism. Our
monocytes.1–4 TGR5 agonists have been shown to stimulate
key requirement for a tool compound was to achieve sufficient
secretion of glucagon-like peptide 1 (GLP1) into the blood
systemic free drug concentrations to enable rigorous testing of
stream via activation of receptors in the intestinal tract, sug-
the effects of TGR5 activation. Despite the wide range of
gesting the potential to provide improved glycemic control
reported non-bile acid chemical matter selective for TGR5, our
through glucose-dependent insulin secretion.5 In addition, acti-
data generated from the evaluation of these ligands suggested
vation of TGR5 receptors in brown adipose tissue has been
that few of these agonists possessed sufficient potency and/or
proposed to increase energy expenditure through the induction
metabolic stability to meet the key requirement for a tool
of type 2 iodothyronine deiodinase (D2).6 Therefore, TGR5
compound (vide infra, Fig. 2).
agonists may provide an opportunity for treatment of diabetes
TGR5 receptor amino acid sequences for humans and dogs
with simultaneous management of glucose levels and body
have been reported to be more similar than for human and rat.
weight. Furthermore, recent studies suggest that the activation of
These differences are reflected in the divergent human and rat
TGR5 in macrophages may be of utility in atherosclerosis.7
activity of non-bile acid agonists in their respective assays.11,12
Bile acids have been reported to be endogenous agonists for
Based on the anticipated challenge of identifying rodent tool
TGR5, and much effort has been devoted to studying this class of
compounds with a similar human TGR5 activity profile, we
compounds such as INT-777 (1, Fig. 1).4,7–10 In addition there is
selected dog as a preclinical species, and sought to identify orally
an increasing number of structurally diverse non-bile acid TGR5
available potent agonists for both dog and human receptors.
agonists reported in the literature.4 These agonists offer the
Herein we report the medicinal chemistry effort towards the
potential to (1) enable delivery of a tool compound with higher
identification of such an agent.
selectivity against other bile acid-mediated pathways, (2) avoid
From the reported non-bile acid TGR5 agonists, triazole 4,
potential complications of transporter mediated pharmacoki-
reported by Evans and co-workers at GSK,13 was selected as an
netics, and (3) lead to a more chemically tractable medicinal
attractive starting point for generating a dog tool compound
chemistry program.
based on its favorable physico-chemical properties (molecular
weight ¼ 341, elogD14 ¼ 2.9, and topological polar surface
Pfizer Worldwide Research and Development, Eastern Point Rd, Groton,
CT 06340, USA. E-mail: Kentaro.Futatsugi@pfizer.com
area ¼ 51) and conformational rigidity (number of rotatable
bonds15 ¼ 6). It was hypothesized that the combination of these
† Electronic supplementary information (ESI) available: Detailed
experimental procedures for synthesis of compounds 4, 10–18,
features may improve the probability of designing orally avail-
a
dataset of compounds 2–9 used for a plot in Fig. 2, and protocols for
cAMP assays, human whole blood assay, and in vivo dog PK studies.
See DOI: 10.1039/c2md20174g
able tool compounds, by striking the balance of TGR5 activity
and a clearance profile.
This journal is ª The Royal Society of Chemistry 2012
Med. Chem. Commun.

View Article Online
Fig. 1 Selected TGR5 agonists reported in the literature.
Fig. 2 A scatter plot of representative non-bile acid TGR5 agonists reported in the literature. Y-axis: pEC50 (human TGR5 uninduced cAMP assay).
X-axis: log scale plot of intrinsic clearance (CLint) in human liver microsomes (HLM). Denoted numbers correspond to the compounds 2–9 in Fig. 1, and
compound 12 in Table 1. The arrow indicates the progression of the work reported herein (compounds 4 to 12).
agonist potency and intrinsic activity.16 For the system with a
The screening cascade
high level of receptor expression, a doxycycline-induced CHO
A screening cascade was designed to facilitate our search for a
cell line recombinantly expressing the human TGR5
tool compound that would meet our objectives. The primary
receptor was used with generation of cAMP as the detection
criteria influencing this design were: (1) the human and dog
system (induced cAMP assay). The lower expressing system
TGR5 activities are generally aligned and (2) intrinsic clearance
used the same cAMP detection methodology in cells expressing
in human liver microsomes (HLM) is a good predictor of
lower levels of either the human or dog TGR5 receptor cultured
intrinsic clearance in dog liver microsomes (DLM). Thus,
in the absence of doxycycline (uninduced cAMP assay).
the program could use human reagents in primary screens with
Concurrently, metabolic stability in HLM was monitored
high confidence that both potency and metabolic stability
for improvement. Compounds with higher potency in the
would translate to dog. Given the known assay system-depen-
induced assay and lower intrinsic clearance relative to the lead 4
dent effects on agonist response, we made a strategic decision to
were next profiled in the uninduced human/dog TGR5
develop two TGR5 agonist assays – one with a high level of
cAMP assays, as well as in the DLM clearance assay. With
receptor expression that would provide increased sensitivity
this screening cascade, we initially profiled the literature lead
and a large dynamic range for screening compounds with
4 in the human induced assay and for intrinsic clearance in
widely varying potencies, and also a lower expressing system
HLM.
that would provide
a
more conservative estimate of
Med. Chem. Commun.
This journal is ª The Royal Society of Chemistry 2012

View Article Online
methoxyphenyl)-1-methyl-1H-1,2,3-triazole, which was in turn
synthesized via 1,3-dipolar cycloaddition of 1-ethynyl-3-
methoxybenzene and sodium azide in the presence of a catalytic
copper iodide and iodomethane in 52% yield. Isoxazole analogs
15–17 (Table 2) were prepared from the corresponding carbox-
ylic acid intermediates.
Synthesis
Compounds 10–14 and 18 were prepared as outlined in
Scheme 1. The synthesis of 4-alkyl-1-(3-methoxyphenyl)-1H-
1,2,3-triazole derivatives 10–14 started from a 1,3-dipolar
cycloaddition reaction of 1-azido-3-methoxybenzene and various
alkynes to afford the cycloadducts 19 after separation of their
regioisomers via silica gel chromatography. Alternatively, 19 was
prepared regioselectively through Cu(II)/L-proline catalyzed 1,3-
dipolar cycloaddition of terminal alkynes and in situ generated
1-azido-3-methoxybenzene to obtain cycloadducts 21,17–19 fol-
lowed by the lithiation–carbonylation sequence. This two-step
Results and discussion
Compound 4 was subjected to our screening cascade, and it
exhibited a modest TGR5 activity in the human induced cAMP
assay. It showed high clearance in both DLM and HLM despite
its moderate lipophilicity (Scheme 2).
sequence allowed delivery of the key intermediate 19 (R1
¼
cyclopropyl) regioselectively on a multi-gram scale without the
need to synthesize organic azide. A direct amide bond formation
of the ethyl ester 19 with lithium amides (Li–NMeR2), two-step
ester hydrolysis–amidation sequence, or a three-step ester
Based on the potency/clearance profile of 4, we set the design
strategy to identify in vivo tool compounds by maximizing TGR5
potency and improving clearance through reductions in overall
lipophilicity and addressing metabolically weak positions of the
lead. As the first step in improving the potency of 4 a 3-methoxy
hydrolysis–amidation–methylation
sequence
furnished
compounds 10–14. Compound 18 was prepared from the corre-
sponding ester intermediate 20 by analogy with the sequence
described above. The intermediate 20 was obtained in good yield
through a lithiation–carbonylation sequence starting from 4-(3-
Scheme 2 Genesis of 11 based on the literature lead 4 reported by GSK.
Table 1 Effect of amide modifications on potency and clearance
Scheme 1 Synthesis of 10–14 and 18. Reagents and conditions: (a) 1-
azido-3-methoxybenzene (1 equiv.), ethyl 2-pentynoate (2.5 equiv.),
toluene, 120 ꢀC, 22% (R1 ¼ Et); (b) n-BuLi (1.4 or 1.5 equiv.), amine (1.2
equiv.), THF, ꢁ78 ꢀC, 19–53%; (c) (1) NaOH or LiOH, MeOH–H2O or
THF–H2O, rt to 70 ꢀC, 34–97%; (2) (COCl)2 (1.1–3 equiv.), cat. DMF,
CH2Cl2, rt, followed by HNMeR2 (1.1 or 1.4 equiv.), TEA (3 equiv.),
CH2Cl2 or DME, rt or mwave, 140–190 ꢀC, 8–66%; (d) NaN3 (1 equiv.),
CuI (2.5 mol%), MeI (1 equiv.), H2O, 55 ꢀC, 52%; (e) n-BuLi (1 equiv.),
THF, ꢁ78 ꢀC, followed by methyl chloroformate (1.2 equiv.), ꢁ78 ꢀC to
rt, 72%; (f) 3-iodoanisole (1 equiv.), ethynylcyclopropane (1 equiv.), L-
proline (20 mol%), Na2CO3 (20 mol%), sodium ascorbate (20 mol%),
sodium azide (1.2 equiv.), CuSO4$5H2O (10 mol%), DMSO–H2O, 65 ꢀC,
76% (R1 ¼ cyclopropyl); (g) n-BuLi (1.2 equiv.), THF, ꢁ78 ꢀC to ꢁ60 ꢀC,
followed by ethyl chloroformate (1.05 equiv.), ꢁ78 ꢀC to ꢁ60 ꢀC, 45%;
(h) (1) LiOH, MeOH–H2O, 70 ꢀC, 97%; (2) SOCl2, cat. DMF, CH2Cl2, rt,
followed by R2NH2 (1.2 equiv.), TEA (3.2 equiv.), rt; (3) iodomethane
(2.7 equiv.), KOH (9 equiv.), DMSO, 50 ꢀC, 4% (R1 ¼ cyclopropyl, R2 ¼
6-trifluoromethyl-pyridin-3-yl, compound 14).
b
Human induced
EC50 ꢂ SDa (nM)
HLM CLint
(mL minꢁ1 mgꢁ1
Cpd
)
elogD14
11
12
13
14
1.9 ꢂ 0.75
5.5 ꢂ 2.2
102 ꢂ 7.7
160 ꢂ 17
>320
98
235
30
3.3
2
(1.9)c
(2.7)c
a
Values are arithmetic means of values from at least three experiments.
Intrinsic clearance determined in human liver microsomes. c Calculated
elogD using a model based on Cubist modeling methodology [Rulequest
Research, www.rulequest.com]. The training set for the model was
>45 000 elogD values measured by the method described in ref. 14.
b
This journal is ª The Royal Society of Chemistry 2012
Med. Chem. Commun.

View Article Online
Table 2 Effect of core modifications of 14 on potency and clearance
that of substituted aniline derivatives, and had variable effects
on clearance. For example, 2-amino pyridine analogue 13 led to
a significant reduction in potency with little impact on clearance
compared to compound 11. The lowest clearance in this
compound set was achieved when the 4-chloro phenyl group
was replaced by a 4-trifluoromethyl-3-pyridyl group, albeit at
the expense of a potency reduction (compound 14). While 12
showed the best balance of TGR5 activity and in vitro clearance,
it was reasoned that further optimization of the core hetero-
aromatic ring of 14 might allow for further increases in potency
with maintenance of low clearance. Based on this hypothesis,
several analogues possessing five-membered heteroaromatic ring
cores were examined within a focused range of log D.
Results obtained from these modifications are summarized in
Table 2.
b
Human induced
EC50 ꢂ SDa (nM)
HLM CLint
(mL minꢁ1 mgꢁ1
Cpd
)
elogD14
14
15
16
17
18
160 ꢂ 17
10 ꢂ 6.4
11 ꢂ 5.1
8.2 ꢂ 1.9
135d
30
60
196
135
43
(2.7)c
2.2
3.1
3.4
Replacement of the triazole core by an isoxazole ring led to a
15-fold increase in potency, but at the expense of substantially
increased clearance presumably as a result of increased log D
(compound 14 vs. 17). Based on this result, it was reasoned that
the reduction of log D by down-sizing the cyclopropyl group
pendant to the isoxazole core may strike a better balance of
potency and clearance. Consistent with the log D reduction,
analogues of 17 with smaller alkyl groups at the 5 position
(compounds 15 and 16) of the isoxazole core retained high
potency, with reduced clearance achieved for the 5-Me analogue
15. An attempt to further reduce log D by examining isomeric
triazole 18 led to the anticipated improvement in clearance but a
significant decrease in potency. Based on the improvements of
both TGR5 activity in the induced assay and metabolic stability
in HLM compared to the literature lead 4, compounds 12 and
15 were selected for more detailed profiling in vitro (Table 3).
Compound 12 showed improved potency over 4 in both
human and dog uninduced TGR5 assays. Consistent with HLM
clearance data, 12 showed reduced clearance in DLM compared
to 4. Thus, compound 12 was selected for detailed determination
of the pharmacokinetic properties in dog (Table 4).
1.7
a
Values are arithmetic means of values from at least three experiments
unless otherwise noted. Intrinsic clearance determined in human liver
b
c
microsomes. Calculated elogD using
a model based on Cubist
modeling methodology [Rulequest Research, www.rulequest.com]. The
training set for the model was >45 000 elogD values measured by the
method described in ref. 14. d Value from single experiment.
group was incorporated into the phenyl ring of 4. This structural
modification indeed led to a significant increase in potency as
previously reported in the related isoxazole series,13 although
there was no detectable clearance improvement (compound 10,
Scheme 2). A cyclopropyl variant of 10 was prepared in an
attempt to hinder CYP (Cytochrome P450)-mediated oxidation.
Unfortunately, while excellent potency was maintained as
previously reported for the isoxazole series,13 no positive,
observable impact on clearance was obtained. Based on this
finding, we shifted attention to optimizing the electron-rich
aniline region. It was postulated that introduction of an electron-
withdrawing group and/or polar heteroaromatic rings may
improve the CYP-mediated oxidative metabolism through
reductions of overall lipophilicity and reduced oxidative liability
of this region of the molecule.
Compound 12 showed moderate to high plasma clearance
reflected in the turnover observed in the DLM system. Due to its
moderate to high volume of distribution, however, a reasonable
half-life could be achieved. Gratifyingly, a modest but acceptable
bioavailability was obtained after oral administration of 12.
While not representing an ideal pharmacokinetic profile, the
achievable exposure for this agent following oral dosing was
sufficient to consider it as a tool compound for further charac-
terization of the mechanism.
As shown in Table 1, cyano analogue 12 showed reduced
clearance as anticipated, while maintaining excellent potency.20
Attempts to utilize amino pyridyl groups (compounds 13 and
14) generally led to significant decreases in activity compared to
Table 3 Summary of in vitro profiles of 12 and 15 vs. 4a
Cpd
4
12
15
Human induced EC50 ꢂ SDb (nM)
53 ꢂ 19
1423 ꢂ 810
584 ꢂ 156
>320
620
2.9
5.5 ꢂ 2.2
10 ꢂ 6.4
618 ꢂ 288
1410 ꢂ 668
60
N.D.
2.2
Human uninduced EC50 ꢂ SDb (nM)
212 ꢂ 143
Dog uninduced EC50 ꢂ SDb (nM)
196 ꢂ 93
HLM CLint (mL minꢁ1 mgꢁ1
)
)
98
79
2
c
DLM CLint (mL minꢁ1 mgꢁ1
c
elogD14
a
b
c
N.D. ¼ not determined. Values are arithmetic means of values from at least three experiments. Intrinsic clearance determined in human liver
microsomes and dog liver microsomes.
Med. Chem. Commun.
This journal is ª The Royal Society of Chemistry 2012

View Article Online
Table 4 Mean in vivo pharmacokinetic properties of 12 in doga
Conclusions
CLp
(mL minꢁ1 kgꢁ1
Vdss
(L kgꢁ1
T1/2
(h)
Cmax
(mM)
Tmax
(h)
fu,
plasmab
In conclusion, a potent and orally available triazole-based TGR5
agonist 12 was identified starting from the literature compound
4. 12 demonstrated a robust dog in vivo pharmacokinetic profile
that was suitable to allow for further characterization of the
TGR5 mechanism, and revealed a safety risk associated with
systemic activation of TGR5.
)
)
%F
22
4.6
4.7
0.51
0.25
17
0.27
a
Pharmacokinetic parameters as geometric mean of n ¼ 3 (iv) or n ¼ 2
mg kgꢁ1 (formulation:
(po) animals per group. iv dose
¼
1
30% sulfobutylether-b-cyclodextrin
polyethylene glycol-400 and
(40 : 60, v/v)); po dose
methylcellulose suspension).
determined by equilibrium dialysis.
a
¼
3
mg kgꢁ1 (formulation: 0.5% (w/v)
b
Unbound dog plasma fraction
Notes and references
1 A. Tiwari and P. Maiti, Drug Discovery Today, 2009, 14, 523–530.
2 M. Zhong, Curr. Trends Med. Chem., 2010, 10, 386–396.
3 T. W. Pols, L. G. Noriega, M. Nomura, J. Auwerx and
K. Schoonjans, J. Hepatol., 2011, 54, 1263–1272.
4 Y. Xu, Annu. Rep. Med. Chem., 2011, 46, 69–87.
5 S. Katsuma, A. Hirasawa and G. Tsujimoto, Biochem. Biophys. Res.
Commun., 2005, 329, 386–390.
6 M. Watanabe, S. M. Houten, C. Mataki, M. A. Christoffolete,
B. W. Kim, H. Sato, N. Messaddeq, J. W. Harney, O. Ezaki,
T. Kodama, K. Schoonjans, A. C. Bianco and J. Auwerx, Nature,
2006, 439, 484–489.
7 T. W. Pols, M. Nomura, T. Harach, G. Lo Sasso, M. H. Oosterveer,
C. Thomas, G. Rizzo, A. Gioiello, L. Adorini, R. Pellicciari,
J. Auwerx and K. Schoonjans, Cell Metab., 2011, 14, 747–
757.
8 R. Pellicciari, A. Gioiello, A. Macchiarulo, C. Thomas, E. Rosatelli,
B. Natalini, R. Sardella, M. Pruzanski, A. Roda, E. Pastorini,
K. Schoonjans and J. Auwerx, J. Med. Chem., 2009, 52, 7958–
7961.
9 C. Thomas, A. Gioiello, L. Noriega, A. Strehle, J. Oury, G. Rizzo,
A. Macchiarulo, H. Yamamoto, C. Mataki, M. Pruzanski,
R. Pellicciari, J. Auwerx and K. Schoonjans, Cell Metab., 2009, 10,
167–177.
10 T. Li, S. R. Holmstrom, S. Kir, M. Umetani, D. R. Schmidt,
S. A. Kliewer and D. J. Mangelsdorf, Mol. Endocrinol., 2011, 25,
1066–1071.
Fig. 3 Inhibition of TNFa secretion, ex vivo whole blood LPS challenge
for compound 12. Data based on one donor. Normalized to maximum
TNFa secreted.
11 M. R. Herbert, D. L. Siegel, L. Staszewski, C. Cayanan, U. Banerjee,
S. Dhamija, J. Anderson, A. Fan, L. Wang, P. Rix, A. K. Shiau,
T. S. Rao, S. A. Noble, R. A. Heyman, E. Bischoff, M. Guha,
A. Kabakibi and A. B. Pinkerton, Bioorg. Med. Chem. Lett., 2010,
20, 5718–5721.
12 K. A. Evans, B. W. Budzik, S. A. Ross, D. D. Wisnoski, J. Jin,
R. A. Rivero, M. Vimal, G. R. Szewczyk, C. Jayawickreme,
D. L. Moncol, T. J. Rimele, S. L. Armour, S. P. Weaver,
R. J. Griffin, S. M. Tadepalli, M. R. Jeune, T. W. Shearer,
Z. B. Chen, L. Chen, D. L. Anderson, J. D. Becherer, M. De Los
Frailes and F. J. Colilla, J. Med. Chem., 2009, 52, 7962–
7965.
13 B. W. Budzik, K. A. Evans, D. D. Wisnoski, J. Jin, R. A. Rivero,
G. R. Szewczyk, C. Jayawickreme, D. L. Moncol and H. Yu,
Bioorg. Med. Chem. Lett., 2010, 20, 1363–1367.
14 F. Lombardo, M. Y. Shalaeva, K. A. Tupper and F. Gao, J. Med.
Chem., 2001, 44, 2490–2497.
15 D. F. Veber, S. R. Johnson, H. Y. Cheng, B. R. Smith, K. W. Ward
and K. D. Kopple, J. Med. Chem., 2002, 45, 2615–
2623.
16 Detailed in vitro pharmacology works around TGR5 will be reported
elsewhere.
17 V. V. Rostovtsev, L. G. Green, V. V. Fokin and K. B. Sharpless,
Angew. Chem., Int. Ed., 2002, 41, 2596–2599.
18 A. K. Feldman, B. Colasson and V. V. Fokin, Org. Lett., 2004, 6,
3897–3899.
To increase confidence in functional potency determination by
the primary cAMP assays, compound 12 was evaluated in a
human ex vivo whole blood assay. A dose dependent suppression
of TNFa release was demonstrated with compound 12 after a 30
minute compound pretreatment and stimulation of human whole
blood with LPS (lipopolysaccharides at 10 ng mLꢁ1 for 4–5
hours stimulation). The data indicate that the uninduced EC50 in
the primary cAMP assay correlates well with the functional
activity seen in the whole blood assay (Fig. 3).
The pharmacological selectivity of 12 was demonstrated
through a broad ligand screening panel consisting of 67 targets,
showing <40% inhibition at 10 mM at all of these targets. These
data coupled with the adequate pharmacokinetic profile of 12
allowed us to select this compound for subsequent in vivo studies.
During these studies, treatment dependent changes in heart rate
and blood pressure were observed for 12 as well as for another
systemically available non-bile acid TGR5 agonist from a
structurally distinct chemotype. These data suggest that mecha-
nism-based cardiovascular effects may be associated with
systemic TGR5 agonism. The detailed results of this study are
beyond the scope of this report and will be disclosed in a future
publication.
19 S. Bolvig, J. Andersen and X. Liang, Synlett, 2005, 2941–2947.
20 L. H. Jones, N. W. Summerhill, N. A. Swain and J. E. Mills, Med.
Chem. Commun., 2010, 1, 309–318.
This journal is ª The Royal Society of Chemistry 2012
Med. Chem. Commun.