Discovery of novel quinoline carboxylic acid series as DGAT1 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2014.02.028

Herein we report the design and synthesis of a series of novel bicyclic DGAT1 inhibitors with a carboxylic acid moiety. The optimization of the initial lead compound 7 based on in vitro and in vivo activity led to the discovery of potent indoline and quinoline classes of DGAT1 inhibitors. The structure-activity relationship studies of these novel series of bicyclic carboxylic acid derivatives as DGAT1 inhibitors are described.

Full text of DOI:10.1016/j.bmcl.2014.02.028

Bioorganic & Medicinal Chemistry Letters 24 (2014) 1790–1794
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of novel quinoline carboxylic acid series as DGAT1
inhibitors
a
a
a
a
b
Gang Zhou a, , Pauline C. Ting , Grant Wishart , Nicolas Zorn , Robert G. Aslanian , Mingxiang Lin ,
⇑
Michelle Smith c, Scott S. Walker c, John Cook c, Margaret Van Heek c, Jean Lachowicz c
a Department of Chemical Research, Merck Research Laboratories, 126 E. Lincoln Ave., Rahway, NJ 07065, USA
b Department of Drug Metabolism, Merck Research Laboratories, 126 E. Lincoln Ave., Rahway, NJ 07065, USA
c Department of Biological Research, Merck Research Laboratories, 2015 Galloping Hill Road, Kenilworth, NJ 07033, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Herein we report the design and synthesis of a series of novel bicyclic DGAT1 inhibitors with a carboxylic
acid moiety. The optimization of the initial lead compound 7 based on in vitro and in vivo activity led to
the discovery of potent indoline and quinoline classes of DGAT1 inhibitors. The structure–activity rela-
tionship studies of these novel series of bicyclic carboxylic acid derivatives as DGAT1 inhibitors are
described.
Received 25 December 2013
Revised 7 February 2014
Accepted 10 February 2014
Available online 19 February 2014
Ó 2014 Elsevier Ltd. All rights reserved.
Keywords:
DGAT1 inhibitor
Triacylglyceride
Diacylglycerol acyltransferase inhibitor
Quinoline
Indoline
Cyclohexanecarboxylic acid
Excess accumulation of triglycerides (TG) is often associated
with health risk and can lead to a number of additional conditions
including obesity, type 2 diabetes, atherosclerosis, hypertension,
and cardiovascular disease.1 Current drugs available to treat diabe-
tes and obesity have limitations in terms of long term efficacy and/
or side effects. The unmet need prompts significant research efforts
in this area.2
Dietary TGs are broken down in the gut to monoacylglycerol
and then absorbed in the small intestines. TGs are then reassem-
bled with the sequential addition of two acyl chains. Acyl
CoA:diacylglycerol acyltransferase (DGAT) catalyzes the formation
of TG from diacylglycerol and acyl-CoA, the terminal and commit-
ted step in TG synthesis.3 Although both DGAT1 and DGAT2 are
transmembrane proteins found in white adipose tissue, small
intestine, liver, and mammary gland, they are from different gene
families with distinctive functions. DGAT1 has more sequence
homology to acyl CoA:cholesterol acyltransferase (ACAT1 and
ACAT2), which plays an important role in cholesterol homeostasis.4
DGAT1 knockout mice exhibit resistance to weight gain when fed a
high-fat diet, have increased insulin sensitivity, and have increased
leptin sensitivity.5 In contrast, DGAT2 is critical to survival due to
lipopenia and skin homeostasis abnormalities.6 DGAT1 knock-out
mouse phenotyping studies has shown that inhibition of DGAT1
could be a potential mechanism for the treatment of T2DM and
obesity.
Over the last decade, genetic evidence in this area has inspired
major efforts in identifying small molecule DGAT1 inhibitors for
potential treatment of diabetes and obesity. The discovery of
selective DGAT1 inhibitors has been disclosed in a number of re-
cent reviews and publications from Japan Tobacco, Pfizer, Bayer,
and Abbott, etc.7 A number of drug candidates have been advanced
into clinical trials.8 Most discovered DGAT1 inhibitor (Fig. 1) have a
terminal carboxylic acid moiety which mimics the fatty acid
substrate of DGAT1.7,9
Pharmacophore modelling around carboxylic acid containing
DGAT1 inhibitors 3 (Bayer), 5 (Abbott), 6 (Astrazeneca) was con-
ducted using full conformational analysis to identify a low energy
conformation starting point. All these ligands overlaid to each
other fairly well in superposition. Key elements of the plausible
pharmacophore-based binding area are illustrated in Figure 2.
This pharmacophore modeling inspired us to utilize Abbott
ligand 524 (green color) and AstraZeneca ligand 6 (shown in pink
colour, lit.25 IC50 = 2.5 nM) as the starting point for our ligand de-
sign (Fig. 3). Through ring constraining strategy and hybridization
of fragment from Abbott and AstraZeneca DGAT1 ligands (in the
⇑
Corresponding author. Tel.: +1 9082400657.
E-mail address: gang.zhou@merck.com (G. Zhou).
http://dx.doi.org/10.1016/j.bmcl.2014.02.028
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

G. Zhou et al. / Bioorg. Med. Chem. Lett. 24 (2014) 1790–1794
1791
COOH
COOH
mouse postprandial triglyceridemia (PPTG)10 assay (À45% at
10 mg/kg @ 2 h). Stability issue was a concern with decomposition
of indole 7 in both human and mouse plasma (ꢀ40% loss of parent
compound measured in ex vivo human plasma at 25 °C after 12 h).
SAR has been focused on improving in vitro and in vivo potency
while increasing stability properties.
COOH
NH
2 O
H2N
N
N
N
O
N
N
N
O
2
1
Pfizer
Japan Tobacco
CF3
O
O
Following up the lead compound 7, initial investigations were
carried out by alternating bicyclic indole ring. Representative
examples (8–20) of bicyclic core modification (boxed) are summa-
rized in Table 1. Despite good fit with the consensus pharmaco-
phore obtained from overlaying the Abbott and AstraZeneca
reported inhibitors, benzooxazole 8, quinazoline 9, benzimidazole
10 and indazole 11 showed poor DGAT1 enzyme activity. Changing
the indole substitution pattern as shown in compounds 12 and 13
resulted in a several fold loss in DGAT1 binding potency. Com-
pound 13 which has different substitution pattern and free NH
group, still maintains certain degree of activity against mouse
DGAT1 enzyme, and it is interesting to see how it behaves
in vivo. To our surprise, compound 13 induced an increase in tria-
cylglycerides levels comparing to vehicle control in mouse PPTG
assay at 10 mg/kg. Replacement of the indole core with tetrahydro-
quinoline 14, tetrahydroquinolinone 15, tetrahydroquinoxaline 16
or tetrahydrobenzodiazepine 17 resulted in reduced in vitro affin-
ity and in vivo efficacy.
H
N
O
N
F
OMe
N
S
N
N
Me
N
H
4
F
3
Hoffman-LaRoche / Via
Bayer
O
COOH
COOH
HN
N
F3CO
O
N
O
N
N
H
O
H
F
F
NH
6
5
F
AstraZeneca
Abbott
Figure 1. Selected examples of literature inhibitors for DGAT1.
An important breakthrough came with the design of indoline
compound 18, which displayed very good in vitro inhibition of
the human DGAT1 enzyme and improved ex vivo plasma stability
properties (ꢀ20% loss of parent compound measured in ex vivo hu-
man plasma at 25 °C after 12 h). However, compound 18 showed
low triacylglyceride reduction (À15% at 10 mg/kg) in our mouse
PPTG assay. In contrast, the ring-expanded benzooxazine 19 re-
tained the good in vitro potency profile of indoline 18 as well as
desirable plasma stability (ꢀ80% recovery of parent compound
measured in ex vivo human plasma at 25 °C after 6 days) and
molecular properties while dramatically improving in vivo efficacy.
Triglyceride levels in the PPTG assay were decreased by 52% at
10 mg/kg in comparison to the vehicle-treated group. Interestingly,
the pharmacokinetic properties for both compounds 18 and 19
Figure 2. Pharmocophore model of Abbott (red), Bayer (yellow), AstraZeneca (blue)
DGAT1 inhibitors.
were dramatically increased (rat AUC = 17
18 and AUC = 58 M h, PPB = 99.8% for 19, respectively, at 10 mg/
kg) due to improved solubility (50, 100 M for 18 and 19, respec-
tively), Caco2 permeability (170 and 448 nm/s for 18 and 19,
respectively), low clearance (Clrat/hu = 13:3 l/min/Mcells for 18,
Clrat/hu = 9:11 l/min/Mcells for 19), good rat oral bioavailability
lM h, PPB = 99.8% for
l
l
O
COOH
COOH
F3CO
HN
O
l
N
N
O
N
H
N
H
5
l
6
O
Abbott
F
AstraZeneca
F
NH
(>90%) and half-life (T1/2 = 3 h for 18 and 4 h for 19), and better
hDGAT1 IC50 = 36 nM
plasma stability. Compound 19 also demonstrated acceptable
F
hDGAT1 IC50 = 36 nM (lit. IC50=2.5 nM)
selectivity over hDGAT2 (IC50 > 10
and clean ancillary profile except CYP 1A1 induction (3-fold for
18 and 10-fold for 19 at 30 M). It is worth noting that the design
lM), hACAT2 (IC50 = 9.7 lM)
l
COOH
O
of indoline 18 and benzomorpholine 19 were driven in large part
by the desire to break the aromaticity of the second ring in indole
structure, seeking improvement in physical properties and the po-
tential for novel substitution patterns.
N
N
H
7
hDGAT1 IC50 = 37 nM
hDGAT2 IC50 > 50000 nM
hACAT2 IC50 = 3000 nM
Cold metabolite identification studies of compound 19
confirmed that cleavage of the urea group is the major pathway
in plasma. Therefore another strategy to reduce plasma stability
issues and concomitant formation of aniline-type by-products is
to modify the dihydrobenzoxazine core, replacing the urea moiety
with a less labile amide. We were able to replace the benzomorph-
oline ring with a naphthalene ring, which featured a more stable
amide moiety (compound 20). Unfortunately, lipophilicity increase
due to the presence of the bicyclic aromatic naphthalene in 20
resulted in extremely low permeability (Caco2 = 0 nm/s) and low
Figure 3. Lead structure design based on Abbott and AstraZeneca DGAT1 inhibitors.
box region), indole compound 7 (cyan color) was initially proposed
to probe the effects of torsional constraints on the urea motif by
formation of an additional ring. This modification appeared struc-
turally well tolerated as good alignment of the urea hydrogen bond
donor and acceptor features was observed for compounds 5 and 7.
This initial proposal has led to the identification of the lead
compound 7 with reasonable in vitro DGAT1 inhibitory activity
and good selectivity over DGAT2 and ACAT, as well as moderate
efficacy measured by triacylglyceride reduction in our in vivo
pH 7 solubility (10 lM). This could explain the lack of in vivo effi-
cacy for 20 (absence of plasma TG reduction in mice in PPTG assay),
in spite of excellent potency against DGAT1 enzyme.

1792
G. Zhou et al. / Bioorg. Med. Chem. Lett. 24 (2014) 1790–1794
Table 1
Optimization of the indole core in lead compound 7a
COOH
O
N
N
H
7
Compd
R
Human DGAT1 IC50
(nM)
Mouse DGAT1 IC50
(nM)
mPPTG % reduction of TG
at 2 h (10 mpk) AUC (10 mpk)
N
O
8
2393
5548
2030
3845
NT
NT
NT
N
9
>10,000
>5000
N
H
N
O
10
N
O
O
11
12
13
2735
448
2668
1090
284
NT
N
N
H
O
N
NT
+23%
1.3 lM h
180
NH
O
O
À15%
14
15
16
160
84
N
N
79 lM h
>10,000
7257
1666
4824
NT
O
O
NT
NT
N
N
O
N
17
>10,000
>10,000
N
O
O
À13%
17 M h
18
19
56
65
69
40
N
l
À52%
N
58
0%
lM h
O
O
20
11
8
a
Assay values are the average of at least two independent determinations.
Subsequently decreasing lipophilicity of compound 20 has re-
sulted in a discovery of quinoline series. Selected biological data
are collated in Table 2. Introducing a basic nitrogen into naphtha-
lene (quinoline compound 21a in Table 2) had a significant impact
when considering efficacious DGAT1 inhibitors as tool compounds
that could be used to understand PK/PD relationships (i.e., impor-
tance of local vs systemic exposure for DGAT1 efficacy). Further-
more, a proof-of-concept plasma stability study was conducted
on analog 21b, and demonstrated very good stability in human,
mouse and rat plasma (stable in human plasma at 25 °C for
12 h). Introduction of a methyl group at the quinoline C-2 position
(compound 21c) improved the in vivo efficacy and extended its
duration (À79% TG reduction at 3 mg/kg at the 12 h time-point).
on solubility and permeability (Caco2 = 542 nm/s, sol = 50 lM). As
a result, inhibitors 21a,b showed significant reduction in plasma
TG levels compared to vehicle-treated mice while retaining good
in vitro DGAT1 activity and selectivity. In addition, quinoline 21b
showed low circulating blood levels, which is an attractive feature

G. Zhou et al. / Bioorg. Med. Chem. Lett. 24 (2014) 1790–1794
1793
Table 2
SAR of quinoline amide DGAT1 inhibitors
X
Y
N
R
N
H
O
21
a,b
Compd
X
Y
R
DGAT1 IC50 (nM)
hACAT IC50 (nM)
mPPTG % reduction of
TG at 2 h (dose)c
AUC PO 10 mg/kgc
M h)
(l
OH
41 (h)
23 (m)
À85%
O
21a
H
4175 (ACAT1)
5
(10 mg/kg)
OH
58 (h)
23 (m)
À76%
O
21b
21c
H
0.03
79
(3 mg/kg)
OH
À100%
(3 mg/kg)
À94%
O
28 (h)
10 (m)
Me
1100 (ACAT2)
(1 mg/kg)
À79%
(3 mg/kg, 12 h)
OH
12 (h)
1.4 (m)
À58%
O
21d
Me
NT
10
(3 mg/kg)
Me Me
N
O
275 (h)
69 (m)
À91%
COOH
21e
21f
Me
Me
1555 (ACAT2)
13300 (ACAT2)
34
(3 mg/kg)
O-
N+
Me Me
O
122 (h)
23 (m)
À40%
COOH
NT
(3 mg/kg)
a
b
c
Assay values are the average of at least two independent determinations.
Enzyme species indicated between parentheses: h = human, m = mouse.
Data are the mean standard error of the mean averaged from eight studies, expressed as percentage reduction (p < 0.01), NT = not tested.
Quinoline 21c showed moderate PXR activation issues (1.4Â at
are shown in Schemes 1.12 1,4-trans-Cyclohexylacetate 24 was
obtained by the route depicted in Scheme 1. Wittig–Horner–
Wadsworth–Emmons reaction of cyclohexanone 22 followed by
reduction gave a mixture of cis and trans isomers, which was
isolated by recrystallization to give pure diastereomer trans-1,4-
cyclohexaneacetic ester 23. Triflation of 23 and palladium-
catalyzed cross-coupling reaction gave boronate 24, which was
converted to the corresponding boronate or trifluoroborate 25.
Both key intermediates were used in cross-coupling reactions with
bicyclic core fragments.
The general synthesis of quinoline DGAT1 inhibitors repre-
sented by compound 21c from the readily available bicyclic inter-
mediate 26 is outlined in Scheme 2. Amide formation produced
aryl bromides 27, which were then subjected to Suzuki–Miyaura
palladium-catalyzed cross-coupling reaction with pinacolboronate
10 lM) and CYP 1A1 liver enzyme induction. Seeking to alleviate
potential genotoxic liabilities, we replaced the N-phenylamide
with various length of N-phenylalkylamide (such as compound
21d). This modification resulted in a slight potency improvement
in DGAT1 enzyme inhibition, and more importantly, in retention
of the good in vivo profile with four to six carbon elongation (data
for compounds with C4 and C6 chain did not show here). The
pyridyl ether carboxylic acid piece from Taisho patent11 is also
tolerated in the quinoline series and delivered very promising
compounds including inhibitor 21e which displayed excellent
in vivo efficacy, good PK and physicochemical properties, although
it lacked selectivity against ACAT enzymes. This problem could be
circumvented by introduction of a pyridine N-oxide (compound
21f) though it was accompanied with a 2-fold loss in in vivo
efficacy and stability in human plasma.
Syntheses of key intermediates in the different classes of right-
hand side carboxylic acid-bearing scaffolds of the DGAT1 inhibitors
OH
O
N
Br
N
Br
N
b, c
O
a
CO2Me
f
CO2Me
d, e
CO2Me
a, b, c
O
N
H
O
B
O
HO
HO
HO
O
26
N
H
O
KF3
B
25
22
23
24
21c
compound
27
Scheme 2. Reagents and conditions: (a) aniline, HATU, DIPEA, CH2Cl2 or DMF; (b)
24, PdCl2(dppf)2, K2CO3, dioxane, H2O, heat; or 25, PEPPSI-iPr, K2CO3, EtOH, H2O,
heat; and (c) LiOH, THF/H2O.
Scheme 1. Reagent and conditions: (a) trimethyl phosphonoacetate, NaH, THF; (b)
H2, Pd/C (cat.), EtOAc; (c) recrystallization in EtOAc; (d) Tf2O, Et3N, DCM; (e)
bis(pinacolato)diboron, PdCl2(dppf)2, KOAc, dioxane; and (f) KHF2, H2O, MeOH.

1794
G. Zhou et al. / Bioorg. Med. Chem. Lett. 24 (2014) 1790–1794
24 or trifluoroborate 25 using Molander’s approach.13 Ester sapon-
ification yielded the final acid product 21c.
8. Devita, R. J.; Pinto, S. J. Med. Chem. 2013, ASAP.
9. (a) Fox, B. M.; Furukawa, N. H.; Hao, X.; Lio, K.; Inaba, T.; Jackson, S. M.; Kayser,
F.; Labelle, M.; Kexue, M.; Matsui, T.; McMinn, D. L.; Ogawa, N.; Rubenstein, S.
M.; Sagawa, S.; Sugimoto, K.; Suzuki, M.; Tanaka, M.; Ye, G. Yoshia, A.; Zhang, J.
A. World Patent WO 2004/047755.; (b) Dow, R. L.; Li, J.-C.; Pence, M. P.; Gibbs,
E. M.; LaPerle, J. L.; Litchfield, J.; Piotrowski, D. W.; Munchhof, M. J.; Manion, T.
B.; Zavadoski, W. J.; Walker, G. S.; McPherson, R. K.; Tapley, S.; Sugarman, E.;
Guzman-Perez, A.; DaSilva-Jardine, P. ACS Med. Chem. Lett. 2011, 2, 407; (c)
Smith, R.; Campbell, A.-M.; Coish, P.; Dai, M.; Jenkins, S.; Lowe, D.; O-Connor, S.;
Su, N.; Wang, G.; Zhang, M.; Zhu, L. US Patent 2004/0224997.; (d) Zhao, G.;
Souers, A. J.; Voorbach, M.; Falls, H. D.; Droz, B.; Brodjian, A.; Lau, Y. Y.; Iyengar,
R. R.; Gao, J.; Judd, A. S.; Wagaw, S. H.; Ravn, M. M.; Engstrom, K. M.; Lynch, J.
K.; Mulhern, M. M.; Freeman, J.; Dayton, B. D.; Wang, X.; Grihalde, N.; Fry, D.;
Beno, D. W. A.; Marsh, K. C.; Su, Z.; Diaz, G. J.; Collins, C. A.; Sham, H.; Reilly, R.
M.; Brune, M. E.; Kym, P. R. J. Med. Chem. 2008, 51, 380.
10. The mouse postprandial triglyceride (PPTG) in vivo assay uses C57BL/6J male
mice (7 weeks old, n = 8) which were fasted overnight (ꢀ16 h). The mice were
dosed with 20% hydroxypropyl-b-cyclodextrin vehicle or test compound in
vehicle 15 min before time 0. Mice were challenged with 0.25 mL corn oil bolus
at time 0 then euthanized and terminal blood was collected by heart puncture
at time of 120 min. Plasma triglyceride levels were determined by colorimetric
method using Wako L-Type TG M Enzyme Color assay reagents A and B (Wako
Chemicals USA, Inc. 1600 Bellwood Road, Richmond, VA 23237-1326).
Absorbance values were determined in an automated microplate reader at a
wavelength of 600 nm.
In conclusion, novel DGAT1 inhibitors based on a indoline and
quinoline core have been discovered. Systematic SAR investigation
has optimized the original lead compound 7 into compounds 21c.
Compound 21c exhibits excellent activity and selectivity for
hDGAT1, lower triglyceride levels at rather low dose in our
in vivo mouse PPTG assay, and exhibit minimal off-target issues.
Further research results of this series will be disclosed in the
future.
References and notes
1. Semenkovich, C. F. J. Clin. Invest. 2006, 116, 1813.
2. King, A. J.; Judd, A. S.; Souers, A. J. Expert Opin. Ther. Patents 2010, 20, 19.
3. Cases, S.; Stone, S. J.; Zhou, P.; Yen, E.; Tow, B.; Lardizabal, K. D.; Voelker, T.;
Farese, R. V., Jr. J. Biol. Chem. 2001, 276, 38870.
4. Pramfalk, C.; Eriksson, M.; Parini, P. Eur. J. Lipid. Sci. Tech. 2012, 114, 624.
5. Chen, H. C.; Farese, R. V., Jr. Arterioscler. Thromb. Vasc. Biol. 2005, 25, 482.
6. Stone, S. J.; Myers, H. M.; Watkins, S. M.; Brown, B. E.; Feingold, K. R.; Elias, P.
M., ; Farese, R. V., Jr. J. Biol. Chem. 2004, 279, 11767.
7. (a) Birch, A. M.; Buckett, L. K.; Turnbull, A. V. Curr. Opin. Drug Discov. Dev. 2010,
13, 489; (b) Nakada, Y.; Aicher, T. D.; Le Huerou, Y.; Turner, T.; Pratt, S. A.;
Gonzales, S. S.; Boyd, S. A.; Miki, H.; Yamamoto, T.; Yamaguchi, H.; Kato, K.;
Kitamura, S. Bioorg. Med. Chem. Lett. 2010, 18, 2785; (c) Fox, B. M.; Iio, K.; Li, K.;
Choi, R.; Inaba, T.; Jackson, S.; Sagawa, S.; Shan, B.; Tanaka, M.; Yoshida, A.;
Kayser, F. Bioorg. Med. Chem. Lett. 2010, 20, 6030; (d) Qian, Y.; Wertheimer, S. J.;
Ahmad, M.; Cheung, A. W.-H.; Firooznia, F.; Hamilton, M. M.; Hayden, S.; Li, S.;
Marcopulos, N.; McDermott, L.; Tan, J.; Yun, W.; Guo, L.; Pamidimukkala, A.;
Chen, Y.; Huang, K.-S.; Ramsey, G. B.; Whittard, T.; Conde-Knape, K.; Taub, R.;
Rondinone, C. M.; Tilley, J.; Bolin, D. J. Med. Chem. 2011, 54, 2433.
11. Umemiya, H.; Takahashi, M.; Bohno, M.; Kawabe, K.; Shirokawa, S.; Nagatsuka,
T.; Sasako, S.; Satou, R.; Itoh, S.; Shimizu, T. Patent WO 2009/011285, 2009.
12. (a) Zhou, G.; Wishart, G.; Ting, P.C.; Aslanian, R. G.; Zorn, N.; Kim, D. W. Patent
WO2010/107765, 2010. (b) Zhou, G.; Wishart, G.; Ting, P.C.; Aslanian, R. G.;
Zorn, N.; Cao, J. Patent WO2010/107768, 2010.
13. (a) Molander, G. A.; Figueroa, R. Aldrichim. Acta 2005, 38, 49; (b) Molander, G.
A.; Ellis, N. M. Acc. Chem. Res. 2007, 40, 275.