Design and synthesis of potent carboxylic acid DGAT1 inhibitors with high cell permeability

Article abstract of DOI:10.1016/j.bmcl.2011.12.050

A series of potent carboxylic acid DGAT1 inhibitors with high permeability were developed from a virtual screening hit. Strategies were employed to reduce Pgp substrate recognition and increase passive permeability, resulting in the discovery of a series showing good inhibition of cellular triglyceride synthesis. The mutagenic potential of prospective metabolites was evaluated in the Ames assay, and one aniline was shown to be devoid of mutagenicity.

Full text of DOI:10.1016/j.bmcl.2011.12.050

Bioorganic & Medicinal Chemistry Letters 22 (2012) 824–828
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design and synthesis of potent carboxylic acid DGAT1 inhibitors with high
cell permeability
⇑
Ustav Bali, Oscar Barba, Graham Dawson, William T. Gattrell , James G. Horswill, David A. Pan,
Martin J. Procter, Chrystelle M. Rasamison, Colin P. Sambrook Smith, Amanda Taylor-Warne,
Philippe Wong-Kai-In
Prosidion Ltd, Windrush Court, Watlington Road, Oxford, OX4 6LT, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of potent carboxylic acid DGAT1 inhibitors with high permeability were developed from a virtual
screening hit. Strategies were employed to reduce Pgp substrate recognition and increase passive perme-
ability, resulting in the discovery of a series showing good inhibition of cellular triglyceride synthesis. The
mutagenic potential of prospective metabolites was evaluated in the Ames assay, and one aniline was
shown to be devoid of mutagenicity.
Received 23 November 2011
Revised 9 December 2011
Accepted 10 December 2011
Available online 17 December 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
DGAT1
Obesity
Permeability
Efflux
For most of human history, mankind has struggled with food
scarcity, and historically obesity has been viewed as a sign of
wealth and prosperity. In the western world in the 21st century,
food is abundant for most people, and obesity is now a leading pre-
ventable cause of death worldwide. At least 1.1 billion adults, and
10% of children are now overweight or obese.1 Obesity can lead to
decreased life expectancy due to increased risk of cardiovascular
disease, Type 2 diabetes and certain cancers.
Diacylglycerol acyltransferase 1 (DGAT1) and DGAT2 catalyse
the final and only committed step in triglyceride synthesis.2
DGAT1 deficient mice are resistant to diet induced obesity, and dis-
play increased energy expenditure and reduced adiposity.3 In con-
trast, DGAT2 deficient mice are profoundly lipopenic and die
shortly after birth due to an impaired permeability function of
the skin.4 These findings have lead a number of companies to pur-
sue selective DGAT1 inhibitors for the treatment of obesity and re-
lated metabolic disorders. For example, treatment of rodents with
small molecule 1 has reproduced some characteristics of DGAT1
deficient mice (Fig. 1).5 A number of DGAT1 inhibitors such as
PF-04620110 (2) have entered clinical trials for obesity and meta-
bolic disorders.6
Early profiling showed
3 to have selectivity over DGAT2
(IC50 > 20 M) and aqueous solubility of >1 mg/ml, in part due to
l
a measured LogD of À2.1 at pH 7.4.7 A medicinal chemistry pro-
gram was initiated to build on the encouraging data and properties
for 3. Recent analysis of a large Caco-2 permeability dataset would
suggest that given the molecular weight and polarity of 3, it is un-
likely to be highly permeable.8 Initially we sought to explore the
SAR with an aim to increase the potency of 3 by making modifica-
tions on the left hand side of the molecule via amide formation.
The unusually low LogD of the starting hit, together with the po-
tential need to seek more balanced physiochemical properties lead
us to target analogues with increased lipophilicity/reduced polar-
ity. Indeed, the Pfzier clinical candidate, PF-04620110, a carboxylic
acid with measured LogD À0.15, has reportedly poor passive per-
meability.9 One hazard of pursuing this strategy is that an increase
in lipophilicity is frequently accompanied by increasing clearance,
however this series of molecules exhibited high stability toward
both mouse and human microsomes.
The synthesis of compounds 3–9 is outlined in Scheme 1. Con-
struction of the triazole ring was carried out by reaction of p-nitro-
benzoyl chloride with 4-methylthiosemicarbazide followed by
cyclisation with sodium hydroxide. Functionalisation of the thiol
Ligand-based virtual screening identified triazole 3, a low
molecular weight (MWt 382) inhibitor of the human DGAT1 en-
with an a-bromoester, followed by reduction of the nitro group with
zyme, IC50 0.34
l
M (a ligand efficiency of 0.33 based on the IC50).
tin afforded the aniline building block for amide formation. Finally,
the acids were obtained by hydrolysis of the corresponding ester.
Many of these analogues (Table 1) demonstrated good inhibi-
tion of human DGAT1 enzyme. Hydrophobic substituents generally
gave an enhancement of potency. In particular, extension of the
⇑
Corresponding author.
E-mail address: wgattrell@yahoo.co.uk (W.T. Gattrell).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.12.050

U. Bali et al. / Bioorg. Med. Chem. Lett. 22 (2012) 824–828
825
O
O
OH
O
N
CO2H
N
NH
2 O
S
OH
N
N
N
O
N
O
N
H
O
N
H
N
H
1
2
3
PF-04620110
Figure 1. Structures of some DGAT1 inhibitors.
N
N
N
N
O
(vii), (viii)
SH
S
(i), (ii)
O
N
N
Cl
O
O2N
O2N
O2N
(ix), (x), (vi)
(iii), (iv), (v), (vi)
O
R3
R2
OH
OH
N
N
A
N
N
2 R3
O
S
R
O
N
O
N
O
R1
N
R1
N
H
H
Scheme 1. Reagents and conditions: (i) 4-Methylthiosemicarbazide, pyridine, DCM; (ii) 2 M sodium hydroxide, reflux, 1 h; (iii) a-bromoester, sodium methoxide, methanol;
(iv) tin, hydrochloric acid; (v) acid chloride, THF, triethylamine; (vi) sodium hydroxide, methanol, room temperature ; (vii) MeI, sodium methoxide; (viii) m-CPBA, DCM, room
temperature; (ix) Alcohol, NaH, THF, reflux, 90 min; (x) hydrogen, palladium on carbon, methanol.
Table 1
Effect of structural modifications on in vitro properties
R2
OH
N
N
X
O
N
O
R1
N
H
Compda
R1
R2
X
DGAT1 IC50
(lM)
3T3L1 IC50 (lM)
ACD LogD pH 7.4
b
b
3
4
5
6
7
Phenyl
p-Tolyl
o-Tolyl
2-Naphthyl
2-Benzothiophene
3,4-Dichlorophenyl
3,4-Dicholorbenzyl
3,4-Dichlorophenyl
3,4-Dichlorophenyl
2-Naphthyl
Me
Me
Me
Me
Me
Me
Me
Ph
S
S
S
S
S
S
S
S
S
O
0.35
0.13
1.98
0.02
0.013
0.23
0.038
1.12
0.011
0.144
>20
>20
>20
>20
>20
>20
>20
>20
>20
>10
À0.53
À0.08
À0.08
0.70
1.55
0.88
0.52
2.00
1.44
À0.29
9
10
11
12
13
Et
Me
a
All compounds were >95% pure by LC–MS and characterised by NMR and MS. All compounds are racemates.
N = 2; for assay details see WO Patent 2008099221.
b
benzene to a fused biaryl (eg., 6 and 7) gave at least an order of
magnitude enhancement in potency. The 3,4-dichlorobenzamide,
designed as a biaryl mimic, gave a slight increase in potency. De-
spite excellent inhibition of the DGAT1 enzyme, the compounds
showed no significant inhibition of cellular triglyceride synthesis
in mouse 3T3L1 adipocytes. Caco-2 permeability data for 9 (Table
2) showed it underwent significant efflux (efflux ratio >50).
A number of strategies to prevent P-glycoprotein (Pgp) recogni-
tion of substrates (and hence reduce efflux) are described in the lit-
erature. Compounds with (N+O) P8 and MWt >400 are likely to be
Pgp substrates.10a Reducing heteroatom count could help mitigate
Pgp recognition, and may also increase passive permeability, an
approach to limiting the effect of Pgp.10b A number of structural
modifications were made in an attempt to increase passive perme-
ability, and reduce efflux. To reduce polarity (by increasing the pKa
of the carboxylic acid)11 an additional methylene spacer was intro-
duced between the heteroatom and the acid. At the same time the
sulphur was replaced with oxygen to remove a potential metabolic
liability. The preparation of these analogues is outlined in Scheme
1. The thiol was converted to a leaving group via methylation and

826
U. Bali et al. / Bioorg. Med. Chem. Lett. 22 (2012) 824–828
Table 2
Effect of side chain modification on in vitro properties
O
N
OH
N
X
N
O
R1
N
H
Compd
R1
X
DGAT1 IC50
(l
M)
3T3L1 IC50
(lM)
LogD pH 7.4
Caco-2 permeability
B–A Efflux ratio
A–B
O
O
S
9
3,4-Dichlorophenyl
0.23
>20
0.62
<0.02a
1.8a
>90
OH
OH
O
14
15
2-Naphthyl
2-Naphthyl
0.004
0.167
1.96
4.0
0.79
0.07
0.22b
0.53a
17.6b
3.19a
80
6
O
OH
a
Conducted at Cerep; permeability classification: A–B < 2.0 Â 10À6 cm/s = low; A–B 2.0 Â 10À6 cm/s to 20 Â 10À6 cm/s = medium and A–B > 20 Â 10À6 cm/s = high.
Conducted at Absorption Systems; permeability classification: A–B < 1.0 Â 10À6 cm/s = low and A–B > 1.0 Â 10À6 cm/s = high.
b
N
O
N
N
N
(i), (ii)
NH2
O
(iii), (iv), (v)
O
N
N
O
O
O
O2N
O2N
O2N
(vi), (vii)
N
N
N
N
O
O
N
N
O
O
(iii)
OH
O
N
N
H
H
Scheme 2. Reagents and conditions: (i) Methyl-4-chloro-4-oxobutanoate, DCM, triethylamine; (ii) phosphorous pentoxide, toluene, reflux 2 h; (iii) sodium hydroxide, THF,
water; (iv) methylamine trifluoroacetate, 150 °C, 4 h; (v) methanol, concentrated sulphuric acid, reflux, 8 h; (vi) hydrogen, palladium on carbon, methanol; (vii) 2-naphthyl
chloride, triethylamine, THF.
Table 3
In vitro evaluation of some triazole amides
O
R1
N
N
N
R2
O
N
O
N
H
Compd
R1
R2
DGAT1 IC50
(l
M)
3T3L1 IC50
(lM)
ACD LogD pH 7.4
HBDb
Caco-2 permeabilitya
A–B
0.10
21.9
8.8
19.8
0.21
B–A
Efflux ratio
479
2.6
5.5
1.5
16
17
18
19
20
H
Me
H
H
H
H
Me
iPr
Ph
0.028
0.029
0.039
0.11
6.1
3.9
4.4
4.9
6.2
3.6
3
1
2
2
2
47.9
56.4
48.8
29.3
0.44
2.34
0.22
5.89
(CH2)2NMe2
0.053
8.11
38
a
Conducted at Absorption Systems; permeability classification: A–B < 1.0 Â 10À6 cm/s = low and A–B > 1.0 Â 10À6 cm/s = high; efflux ratio P3 indicates significant efflux.
b
HBD = Hydrogen bond donor count.
subsequent oxidation to the sulphone. The sulphone could then be
displaced with an alcohol in the presence of base. A direct ex-
change of the sulphur atom for oxygen (compare 6 and 13) re-
sulted in drop in potency. However, incorporation of an
a

U. Bali et al. / Bioorg. Med. Chem. Lett. 22 (2012) 824–828
827
N
X
F
(i), (ii)
X
N
X
F
O
(iii), (iv)
X
N
H
R
Br
OH
O
Ar
O
O
N
H
R
OH
X
O
(i), (ii)
OH
(v), (iv)
X
N
H
R
Br
Scheme 3. Reagents and conditions: (i) 4-Aminobenzene boronic acid, Cs2CO3, PdCl2(dppf), dioxan, water 90 °C, 17 h; (ii) 2-naphthoyl chloride, triethylamine, THF; (iii) 3-
hydroxy-2,2-dimethylpropionic acid methyl ester, sodium hydride; (iv) sodium hydroxide, THF, water; (v) 3-bromo-2,2-dimethylpropionic acid methyl ester, sodium
hydride, DMF.
Table 4
Structure–activity relationships for some biaryl acids
OH
O
Ar
O
O
R
N
H
Compd
Ar
R
DGAT1 IC50
(lM)
3T3 IC50
(
lM)
ACD LogD pH 7.4
Caco-2 permeabilitya
A–B
33
B–A
Efflux ratio
N
N
N
21
22
Me
Me
0.115
0.373
0.433
0.709
1.7
1.1
62.1
1.8
1.3
N
N
42.5
19.3
—
55.8
38.8
—
23
24
Me
Me
0.117
0.097
2.16
1.9
2.5
2.0
N
N
0.125
—
N
25
26
Me
Me
2.42
0.52
—
—
2.3
3.3
41.3
—
84.7
—
2.0
—
N
27
28
H
H
0.049
0.102
0.037
0.102
1.8
2.8
7.9
—
5.6
—
0.7
—
a
Conducted at Absorption Systems; permeability classification: A–B < 1.0 Â 10À6 cm/s = low and A–B > 1.0 Â 10À6 cm/s = high; efflux ratio P3 indicates significant efflux.
additional methylene spacer, coupled with installation of a gem-
dimethyl group adjacent to the carbonyl, yielded 14 (Table 2) the
most potent analogue identified in the enzymic assay.12
Substitution of the heteroatom adjacent to the triazole for a
methylene was also accomplished (Scheme 2). Deletion of the het-
eroatom would be anticipated to increase passive permeability as
highlighted above. Starting from 4-nitrobenzhydrazide, the 1,3,4-
oxadiazole was assembled. Reaction of this with methylamine
trifluoroacetate under forcing conditions generated the N-methyl-
1,3,4-triazole. Standard conditions were then used to convert the
nitro group to the desired amide, and finally to hydrolyse the ester
to acid 15. Pleasingly, the sulphur could be replaced with a methy-
lene and still retain reasonable levels of in vitro potency. Unfortu-
nately, both 14 and 15 showed a considerable drop off in potency
when evaluated in the 3T3L1 assay. Permeability data showed 14
and 15 still underwent efflux.
The DGAT1 inhibitors up to this point all contained a carboxylic
acid charged at physiological pH, and thus would be expected to be
highly solvated. To reduce solvation, amide derivatives of highly
potent acid 14 were prepared (Table 3). The amides were typically

828
U. Bali et al. / Bioorg. Med. Chem. Lett. 22 (2012) 824–828
Table 5
against TA98 and TA100 strains S9 metabolic activation. The re-
sults for strain TA98 + S9 (the most sensitive conditions) are shown
in Table 5. The presence of heteroatoms in the distal aromatic ap-
peared to modulate the mutagenic potential of the aniline, with
pyrimidine 31 devoid of mutagenic response in this assay (and
TA100 S9). Shape, lipophilicity and the reactivity of the ultimate
nitrenium species have all been reported to modulate the muta-
genic activity of aromatic amines and the effect of the presence
of the methyl group in 31 cannot be excluded.13 Subsequent to
conducting this study, aniline 30 and an amide derivative was de-
scribed in a competitor DGAT1 patent.14
In summary, a number of strategies were employed to reduce
Pgp substrate recognition and increase passive permeability for a
series of DGAT1 inhibitors. Starting from a virtual screening hit
with poor inhibition of cellular triglyceride synthesis, potent
DGAT1 inhibitors with reduced efflux and high permeability were
obtained. The mutagenic potential of three prospective aniline
metabolites was evaluated in the Ames assay. One aniline was
shown to be devoid of mutagenicity.
Mutagenicity (Ames activity) of anilines for Salmonella typhimurium TA98 in the
presence of rat liver S9; ⁄p < 0.05
Compd
Structure
Max fold increase in
mutation frequency
(TA98 + S9)
OH
OH
OH
O
O
O
24⁄
O
O
29
H2N
H2N
H2N
N
N
30
2.6⁄
O
N
31
No increase
Acknowledgments
The author thanks Mark Saxel and Neil Dyson for construction
of the virtual screening grid, Vicky Gamble for technical assistance,
and Craig Johnstone for helpful discussions with the manuscript.
5 to 10 fold less potent than the parent acid in the DGAT1 enzymic
assay. Primary amide 16 displayed high efflux and reduced cell po-
tency over the parent acid, possibly due to increased hydrogen
bond donor count. For secondary and tertiary amides, improved
passive permeability and cell activity was observed, for example,
References and notes
1. Haslam, D. W.; James, W. P. T. Lancet 2006, 367, 1540.
benzamide 19, DGAT1 IC50 0.11 lM, 3T3L1 IC50 0.22 lM. For this
2. Cases, S.; Smith, S. J.; Zheng, Y. W.; Myers, H. M.; Lear, S. R.; Sande, E.; Novak, S.;
Collins, C.; Welch, C. B.; Lusis, A. J.; Erickson, S. K.; Farese, R. V., Jr. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, 13018.
3. Smith, S. J.; Cases, S.; Jensen, D. R.; Chen, H. C.; Sande, E.; Tow, B.; Sanan, D. A.;
Raber, J.; Eckel, R. H.; Farese, R. V., Jr. Nature Genetics 2000, 25, 87.
4. 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.
limited set of examples, cell potency appeared to correlate with ef-
flux ratio. Sub-micromolar cell potency was only obtained with the
most lipophilic analogues (LogD P 4.4). This, together with the
significant efflux observed for the majority of analogues led us to
seek alternative cores.
5. Zhao, G.; Souers, A. J.; Voorbach, M.; Falls, H. D.; Droz, B.; Brodjian, S.; 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.
6. (a) Birch, A. M.; Buckett, L. K.; Turnball, A. V. Curr. Opin. Drug Disc. 2010, 13,
489; (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.; Wlaker, G.
S.; McPherson, R. K.; Tapley, S.; Sugarman, E.; Guzman-Perez, A.; DaSilva-
Jardine, P. ACS Med. Chem. Lett. 2011, 2, 407.
In the design of a new core, reducing heteroatom count was an
important consideration to mitigate efflux potential. Conforma-
tional analysis suggested the central biaryl in triazole 3 was non-
planar by virtue of the N-methyl group. It was unclear if the triazole
nitrogens were participating in any H-bonding interactions, so in
order to mimic the non-planar orientation, a methyl was appended
to the N-linked phenyl ring. For ease of synthesis, the triazole was
replaced with six-membered heterocycles, as outlined in Scheme
3. The central biaryl was constructed first via Suzuki coupling. Fol-
lowing formation of the 2-naphthyl amide, installation of the 2,2-
dimethylpropionic acid was conducted either via alkylation or SNAr
reaction, depending on the nature of the aryl ring. The results of this
work are shown in Table 4. Although a drop-off in DGAT1 enzymic
potency was observed relative to 14, the compounds showed good
inhibition of cellular triglyceride synthesis in 3T3L1 adipocytes.
Permeability, as determined by Caco-2 was high, with much re-
duced efflux. To establish the effect of shape around the biaryl core,
analogues 27 and 28 lacking the methyl on the N-linked aromatic
were also prepared. Compounds 27 and 28 were found to have en-
hanced activity compared with the methyl analogues 24 and 26,
respectively.
7. The calculated value (ACD) of LogD at pH 7.4 for 3 was À0.5, more than one log
unit higher than measured. For the purposes of this study calculated values
were used as a guide.
8. Waring, M. J. Bioorg. Med. Chem. Lett. 2009, 19, 2844.
9. Dow, R.; Andrews, M.; Aspnes, G. E.; Balan, G. E.; Gibbs, M.; Guzman-Perez, A.;
Kapil, K.; LaPerle, J.; Li, J. C.; Litchfield, J.; Munchhof, M. J.; Perreault, L. P. Bioorg.
Med. Chem. Lett. 2011, 21, 6122.
10. (a) Didziapetris, R.; Japertas, P.; Avdeef, A.; Petrauskas, A. J. Drug Target. 2003,
11, 391–406; (b) Raub, T. Mol. Pharm. 2006, 3, 3–25.
11. Compare 2-mercaptoacetic acid, pKa 3.6 and 3-hydroxypropanoic acid, pKa 4.5.
12. It was not possible to make the analogue lacking the gem-dimethyl group by
this method. Attempted reaction of the sulphone with tert-butyl 3-
hydroxypropionate afforded the triazol-3-ol, presumably via elimination of
tert-butyl acrylate from the desired product.
13. For an example of successfully reducing aniline mutagenicity see: Block, M. H.;
Boyer, S.; Brailsford, W.; Brittain, D. R.; Carroll, D.; Chapman, S.; Clarke, D. S.;
Donald, C. S.; Foote, K. M.; Godfrey, L.; Ladner, A.; Marsham, P. R.; Masters, D. J.;
Mee, C. D.; O’Donovan, M. R.; Pease, J. E.; Pickup, A. G.; Rayner, J. W.; Roberts,
A.; Schofield, P.; Suleman, A.; Turnbull, A. V. J. Med. Chem. 2002, 45, 3509.
14. Ting, P. C., Lee, J. F., Aslanian, R. G.; WO Patent 2011031628, 2011.
Concerns about the potential mutagenic activity of the anilines
released from metabolic cleavage of the amide bond of these com-
pounds led us to evaluate selected anilines 29–31 in the Ames test,