Discovery of Pyrrolidine-Containing GPR40 Agonists: Stereochemistry Effects a Change in Binding Mode

Article abstract of DOI:10.1021/acs.jmedchem.6b01559

A novel series of pyrrolidine-containing GPR40 agonists is described as a potential treatment for type 2 diabetes. The initial pyrrolidine hit was modified by moving the position of the carboxylic acid, a key pharmacophore for GPR40. Addition of a 4-cis-CF₃ to the pyrrolidine improves the human GPR40 binding K_i and agonist efficacy. After further optimization, the discovery of a minor enantiomeric impurity with agonist activity led to the finding that enantiomers (R,R)-68 and (S,S)-68 have differential effects on the radioligand used for the binding assay, with (R,R)-68 potentiating the radioligand and (S,S)-68 displacing the radioligand. Compound (R,R)-68 activates both G_q-coupled intracellular Ca²⁺ flux and G_s-coupled cAMP accumulation. This signaling bias results in a dual mechanism of action for compound (R,R)-68, demonstrating glucose-dependent insulin and GLP-1 secretion in vitro. In vivo, compound (R,R)-68 significantly lowers plasma glucose levels in mice during an oral glucose challenge, encouraging further development of the series.

Full text of DOI:10.1021/acs.jmedchem.6b01559

Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES
Article
Discovery of Pyrrolidine-Containing GPR40 Agonists:
Stereochemistry Effects a Change in Binding Mode
Elizabeth Anne Jurica, Ximao Wu, Kristin N. Williams, Andres S. Hernandez, David S. Nirschl, Richard
A. Rampulla, Arvind Mathur, Min Zhou, Gary Cao, Chunshan Xie, Biji Jacob, Hong Cai, Tao Wang, Brian
J. Murphy, Heng Liu, Carrie Xu, Lori K. Kunselman, Michael B. Hicks, Qin Sun, Dora M. Schnur, Doree
Fay Sitkoff, Elizabeth A. Dierks, Atsu Apedo, Douglas B. Moore, Kimberly A. Foster, Mary Ellen Cvijic,
Reshma Panemangalore, Neil A. Flynn, Brad D. Maxwell, Yang Hong, Yuan Tian, Jason J. Wilkes, Bradley
A. Zinker, Jean M. Whaley, Joel C. Barrish, Jeffrey A. Robl, William R Ewing, and Bruce Alan Ellsworth
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01559 • Publication Date (Web): 23 Jan 2017
Downloaded from http://pubs.acs.org on January 23, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Ewing, William; Bristol-Myers Squibb
Ellsworth, Bruce; Bristol-Myers Squibb, Discovery Chemistry
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Discovery of PyrrolidineꢀContaining GPR40
Agonists: Stereochemistry Effects a Change in
Binding Mode
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Elizabeth A. Jurica,* Ximao Wu, Kristin N. Williams, Andres S. Hernandez, David S. Nirschl,
Richard A. Rampulla, Arvind Mathur, Min Zhou, Gary Cao, Chunshan Xie, Biji Jacob, Hong
Cai, Tao Wang, Brian J. Murphy, Heng Liu, Carrie Xu, Lori K. Kunselman, Michael B. Hicks,
Qin Sun, Dora M. Schnur, Doree F. Sitkoff, Elizabeth A. Dierks, Atsu Apedo, Douglas B. Moore,
Kimberly A. Foster, Mary Ellen Cvijic, Reshma Panemangalore, Neil A. Flynn, Brad D.
Maxwell, Yang Hong, Yuan Tian, Jason J. Wilkes, Bradley A. Zinker, Jean M. Whaley, Joel C.
Barrish, Jeffrey A. Robl, William R. Ewing, Bruce A. Ellsworth
Research and Development, BristolꢀMyers Squibb, Co., P.O. Box 5400, Princeton, New Jersey
08543ꢀ5400, United States
KEYWORDS
GPR40, FFAR1, GPCR, agonist, diabetes, insulin, secretagogue, glucoseꢀdependent insulin
secretion, GLPꢀ1, glucagonꢀlike peptideꢀ1, binding mode, pyrrolidine, radioligand binding
ABSTRACT
ACS Paragon Plus Environment
1

Page 3 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A novel series of pyrrolidineꢀcontaining GPR40 agonists is described as a potential treatment for
type 2 diabetes. The initial pyrrolidine hit was modified by moving the position of the
carboxylic acid, a key pharmacophore for GPR40. Addition of a 4ꢀcisꢀCF to the pyrrolidine
3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
improves the human GPR40 binding K and agonist efficacy. After further optimization, the
i
discovery of a minor enantiomeric impurity with agonist activity led to the finding that
enantiomers (R,R)ꢀ68 and (S,S)ꢀ68 have differential effects on the radioligand used for the
binding assay, with (R,R)ꢀ68 potentiating the radioligand and (S,S)ꢀ68 displacing the
2
+
radioligand. Compound (R,R)ꢀ68 activates both G ꢀcoupled intracellular Ca flux and Gsꢀ
q
coupled cAMP accumulation. This signaling bias results in a dual mechanism of action for
compound (R,R)ꢀ68, demonstrating glucoseꢀdependent insulin and GLPꢀ1 secretion in vitro. In
vivo, compound (R,R)ꢀ68 significantly lowers plasma glucose levels in mice during an oral
glucose challenge, encouraging further development of the series.
INTRODUCTION
The International Diabetes Federation estimates that in 2014 more than 387 million
people were living with diabetes and 4.9 million deaths were attributed to diabetic complications,
1
making diabetes a worldwide epidemic affecting 1 in 11 adults. Nearly half of the population
2
with diabetes is undiagnosed, but the progressive nature of the disease means that their
condition will likely (rapidly) decline, affecting quality of life if left untreated. Type 2 diabetes,
representing 95% of all diabetes cases, is characterized by an insufficiency of insulin action that
results in high blood glucose levels. Increased blood glucose can cause a variety of
microvascular complications such as nephropathy, retinopathy, and neuropathy as well as
3
macrovascular complications such as heart disease and stroke in type 2 diabetes patients.
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
GPR40, also known as free fatty acid receptor 1 (FFAR1), is a nutrientꢀsensing Gꢀ
proteinꢀcoupled receptor (GPCR) that is primarily expressed in the βꢀcells of the pancreas where
4,5
it mediates glucoseꢀstimulated insulin secretion.
GPR40 is commonly reported to be a Gqꢀ
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
coupled GPCR that is most potently activated by longꢀchain unsaturated fatty acids (C>10) at
micromolar concentrations. GPR40ꢀdependent modulation of insulin secretion is glucoseꢀ
dependent, suggesting that drugs targeting this receptor could improve glycemic control with low
6
risk of hypoglycemia in contrast to the clinical experience with nonꢀglucose dependent insulin
secretagogues such as sulfonylureas.
Fasiglifam (1, TAKꢀ875, Figure 1) is a GPR40 agonist that progressed to phase III clinical
trials, validating this target as an approach for the treatment of type 2 diabetes. It is a partial
2
+
agonist of GPR40 that promotes G ꢀdependent Ca accumulation in beta cells, amplifying the
q
7,8
effect of glucose on insulin secretion. In the clinic, 1 reduced glycated hemoglobin (HbA1c),
9
an integrated measure of glycemic control, by ~1% in type 2 diabetic patients. At the 50 mg
dose used in phase III clinical trials, 1 primarily demonstrated improvement on fasting plasma
glucose levels with increased insulin secretion observed. At this dose, there was no effect on the
gut hormones GIP and GLPꢀ1 (vide infra). The development of 1 was halted during phase III
clinical trials due to concerns about liver safety but, since GPR40 is not expressed in liver, the
1
0,11
findings are unlikely to be a direct result of GPR40 agonism.
The clinical efficacy of 1
suggests that a chemically distinct GPR40 agonist with improved liver safety would be a
valuable treatment for type 2 diabetes.
1
2
Recently, GPR40 was shown to have three distinct ligand binding sites. AMG 837 (2) and
AMꢀ1638 (3) were demonstrated, via radioligand binding studies, to occupy distinct sites on
GPR40 (Figure 1). Interestingly, there is crossꢀtalk between these sites, and ligands can exhibit
ACS Paragon Plus Environment
3

Page 5 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
positive cooperative binding. Several compounds, including 3, were found to elicit a greater
Ymax than endogenous fatty acid ligands. These compounds were labeled “superagonists” due
1
3
to their superior ability to control glucose levels. These full activators of GPR40 show greater
potency and efficacy in vivo at lower plasma exposures relative to partial agonists in an oral
glucose tolerance test.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
GPR40 is also expressed in enteroendocrine cells where it is important for secretion of gut
1
3
hormones such as glucagonꢀlike peptideꢀ1 (GLPꢀ1), but many synthetic GPR40 agonists,
including 1, were not known to increase GLPꢀ1 secretion. AMꢀ1638 is the first reported small
13
molecule agonist to demonstrate GPR40ꢀdependent GLPꢀ1 secretion. The ability to promote
GLPꢀ1 secretion brings the possibility of an additional benefit of GPR40 agonism for the
treatment of diabetes, since GLPꢀ1 is known to have effects on gastric emptying in the gut and
glucoseꢀdependent insulin secretion in the pancreas. The native peptide has a short halfꢀlife in
1
4
plasma, but mechanisms that target the preservation of GLPꢀ1, such as DPPꢀ4 inhibitors, and
1
5,16
metabolically stable GLPꢀ1 mimetics are clinically efficacious for the treatment of diabetes.
By developing GPR40 agonists with this dual mechanism of action (i.e. promoting glucoseꢀ
dependent insulin and GLPꢀ1 secretion), we hoped to discover compounds with greater glucose
lowering capacity than 1 or other agonists that employ a single insulin secretagogue mechanism
of action.
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Structures of GPR40 agonists.
To establish a novel starting point for our medicinal chemistry program, we performed a
virtual search of the BristolꢀMyers Squibb compound collection. It was known in the literature
that a carboxylic acid or acid mimetic was required for GPR40 agonism, and several agonists
8,17ꢀ20
(Figure 1) included chirality at the βꢀposition relative to the carboxylic acid.
We searched
for structures with a cyclic constraint bearing a carboxylic acid that might confer GPR40 potency
and efficacy. This exploration led to the discovery of racemic compound 4 that was found to be
a weak partial agonist of the human GPR40 receptor (hGPR40 EC = 5.0 ꢁM). Despite its weak
50
activity, we decided to employ compound 4 as a starting point since we felt the aniline nitrogen
and the novel cyclic structure might serve to constrain the carboxylic acid in a bioactive
conformation. By removing the lactam carbonyl in compound 5, the potency was improved twoꢀ
fold, suggesting that the lactam carbonyl was not important for agonist activity. In a search for
the optimal spacing between the aryl group and the carboxylic acid, we homologated the acid
moiety as found in compound 6, but this modification led to reduced GPR40 agonist activity.
Since the distance between the carboxylic acid and aryl group seemed critical for GPR40
activity, we varied the placement of the acid on the ring, resulting in the discovery of (S)ꢀ7 and
(R)ꢀ7 (Table 1) with submicromolar hGPR40 activity. Due to the improved potency of (S)ꢀ7
ACS Paragon Plus Environment
5

Page 7 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
relative to the starting point (compound 4), we focused on developing the SAR around this
homoproline analog.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. Pyrrolidine series evolution.
While compound (S)ꢀ7 was an attractive starting point, we wanted to improve upon
several features to improve its druggability. Initially, we wished to increase potency versus the
hGPR40 receptor, but we also hoped to improve both potency and efficacy versus the mGPR40
receptor to facilitate preclinical characterization in efficacy models. Additionally, compound
+
(
S)ꢀ7 possessed weak cardiac Na channel antagonist activity (4.0 ꢁM) and weak peroxisome
proliferatorꢀactivated receptor gamma (PPARγ) transactivation (17 ± 10 ꢁM, 54% Ymax), which
was not desired. There is a possible synergy between the insulinꢀsensitizing mechanism of
PPARγ agonism and the increased insulin secretion via GPR40 agonism that could lead to
greater efficacy against diabetes. However, since the safety and efficacy of GPR40 agonists are
not fully clinically validated, we did not want to complicate evaluation of the mechanism of
2
1
action with the regulatory burdens of PPARγ activation.
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
RESULTS AND DISCUSSION
Early syntheses of pyrrolidineꢀ2ꢀacetic acids, such as 7, were (purposefully) racemic so
that both enantiomers could be separated and tested for GPR40 activity. The initial route to
analogs such as 7 employed tosylate 9, the product of the ring opening of tetrahydrofuran 2ꢀ
acetic acid ethyl ester (8) with LHMDS. In a single step, compound 9 was reacted neat with
aniline 10 at 130 ºC to both displace the tosylate and concomitantly cyclize to form pyrrolidine
11 in low to modest yields. The racemic mixture was separated by chiral supercritical fluid
chromatography (SFC) to provide single enantiomers. The (S)ꢀenantiomer was taken forward for
further elaboration. When a benzyl ether was used as a protecting group for analog synthesis, the
benzyl group was removed by hydrogenolysis using Pd/C. Phenol 12 was reacted with benzyl
halide 13 to give ester 14, which was hydrolyzed with LiOH to give the carboxylic acid (S)ꢀ7.
Scheme 1. Representative Synthesis of Pyrrolidine 2ꢀAcetic Acid Analogs, as Depicted for
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a,b
Compound (S)ꢀ7
a
Reagents and conditions. (a) i. LHMDS, THF; ii. TsCl, 60%; (b) 130 °C, neat, 16%; (c) chiral
separation; (d) Pd/C, H , MeOH/EtOH, 88%; (e) K CO , DMF, 93%; (f) LiOH, THF/water,
2
2
3
ACS Paragon Plus Environment
7

Page 9 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
b
4
4%. See supporting information for other compounds that were synthesized via this general
route.
The route depicted in Scheme 1 was short, convergent, and it served our initial needs;
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
but, as we found that the (S)ꢀisomers were consistently more potent (see table in supporting
information), we sought a new route that would tap the chiral pool of substituted pyrrolidines to
synthesize single isomers. Pyrrolidine analogs were synthesized from substituted Lꢀprolines that
were commercially available. Aryl iodides, such as compound 16, could be synthesized by in
situ formation of diazonium salts and displacement with KI (Scheme 2). From substituted
prolines such as 17, the desired prolinol 18 was synthesized via transient (Boc) amine protection
and reduction of the carboxylic acid followed by NꢀBoc deprotection. Unfortunately, direct Nꢀ
arylation of the homologated proline esters via Pdꢀ and Cuꢀcatalyzed couplings mostly led to
oligomerization of the proline, resulting in poor yields of Nꢀarylated homoproline analogs.
These problems could be circumvented by coupling the reduced βꢀamino alcohol 19 to an aryl
22
iodide 16 to give alcohol 20 followed by mesylation and cyanation that, after basic hydrolysis
of nitrile 21, led to homologated acid 22. Acidic (HCl) nitrile hydrolysis conditions were
unacceptable as these conditions led to epimerization of the acetic group. The above methods
form the basis of the synthesis of all Nꢀaryl heterocyclic carboxylic acids reported herein.
Scheme 2. Representative Synthesis of Chiral Pyrrolidine GPR40 Agonists, as Depicted for
a,b
Compound 22
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Reagents and conditions. (a) TsOH, NaNO , KI, 74%; (b) iꢀBuO CCl, NMM, THF; NaBH ,
2
2
4
H O, 95%; (c) HCl, dioxane, quant. (d) CuI, NaOH, iꢀPrOH, 90 °C, 74%; (e) MsCl, NEt ,
2
3
CH Cl ; (f) NaCN, DMSO, 50 °C, 100% (2 steps); (g) KOH, EtOH/H O, 120 °C, 43%.
Similar compounds made using an analogous method; see supporting information.
2
2
2
b
2
+
GPR40 is characterized as a G ꢀcoupled GPCR that promotes intracellular Ca
q
2+
accumulation. As such, we primarily tested compounds in a Ca ꢀcoupled fluorescence assay,
searching for full agonists that elicited high Ymax in CHO cell lines expressing either human or
mouse GPR40. Later (vide infra), we found that the CHO cell line used for screening highly
overexpressed hGPR40, and compounds with lower efficacy were indistinguishable from
compounds with higher efficacy due to receptor reserve. For this reason, Ymax values are not
reported here for human EC values in the CHO cell assay since all compounds produced 100%
50
Ymax. A HEK cell line with an inducible GPR40 expression vector was employed when we
2
3
wished to distinguish partial and full activators, but this assay system had relatively low
throughput and inconsistent receptor expression across different induction experiments so it was
not used for routine screening. For rapid screening, we also employed a CHO cell line
ACS Paragon Plus Environment
9

Page 11 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
expressing the mouse GPR40 receptor, which seemed to be somewhat more discriminatory for
intrinsic efficacy differences (Ymax) among compounds. In practice, we used mGPR40 efficacy
(Ymax) as a surrogate for hGPR40 intrinsic activity and only followed up on select compounds
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
using inducible expression systems to assign compounds as full or partial hGPR40 agonists.
Figure 3. Arbitrary nomenclature (AꢀC) for rings on the pyrrolidine chemotype of GPR40
agonists.
Variations in the Nꢀheterocyclic ring size were explored to determine if a fiveꢀmembered ring
with (S)ꢀstereochemistry had the optimal size and stereochemistry for GPR40 agonist efficacy.
For descriptive purposes, the saturated nitrogen heterocycle is referred to as the A ring, the
central aromatic ring as the B ring, and the appended aryl ring as the C ring (Figure 3). The A
ring was modified to explore the effect of ring size (4ꢀ7) on GPR40 activity in combination with
a 2, 4ꢀdichlorobenzylꢀlinked C ring. Azetidine (S)ꢀ23 had submicromolar GPR40 agonist
activity in both the human and the mouse GPR40 cell lines but the compound proved to be
chemically unstable due to betaꢀelimination to the open form. Azetidine (R)ꢀ23 had no
detectable GPR40 agonist activity (at the concentrations tested). In contrast, pyrrolidine (S)ꢀ7
was equipotent versus hGPR40 but it showed reduced potency versus the mouse receptor as
compared to the corresponding azetidine analog. Both individual isomers of the 6ꢀ and 7ꢀ
membered ring compounds (24 (enantiomer 1), 24 (enantiomer 2), 25 (enantiomer 1), and 25
(enantiomer 2)) lost significant GPR40 agonist activity (absolute stereochemistry was not
ACS Paragon Plus Environment
10

Journal of Medicinal Chemistry
Page 12 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
determined) versus the 5ꢀmembered analogs (S)ꢀ7 and (R)ꢀ7. These results provided confidence
that the original fiveꢀmembered ring with (S)ꢀstereochemistry, as found in (S)ꢀ7, was preferred
due to good hGPR40 activity and chemical stability, so further SAR explorations employed a
pyrrolidine Aꢀring.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 1. Effect of Aꢀring Size on GPR40 Activity
1
3
hEC (µM)
mEC (µM) ±
50
50
2
Compound number
Ring Size
hK (µM)
i
4
4
±
S.D.
S.D. (Ymax)
4
0.44 ± 0.38
(88%)
(S)ꢀ23
0.26 ± 0.12
n.d.
n.d.
(
(
(
n = 0)
4
5
(R)ꢀ23
13
>17
n = 0)
5
1
(S)ꢀ7
0.34 ± 0.20
0.89
4.5 ± 1.8 (66%)
n = 1)
5
5
(R)ꢀ7
0.88 ± 0.061
>33
n.d.
n.d.
>17
(n = 1)
6
2
4 (enantiomer 1)
>33
ACS Paragon Plus Environment
11

Page 13 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
(
(
(
n = 2)
6
24 (enantiomer 2)
>17
n.d.
n.d.
n.d.
>33
>16
n = 2)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
7
5
2
5 (enantiomer 1)
6.4
n = 3)
7
5
25 (enantiomer 2)
11
>33
n = 3)
1
2+
2
3
hGPR40 EC in Ca FLIPR assay. Human binding K versus a radioligand. mGPR40 EC
in Ca FLIPR assay. mean of at least two runs. N = 1.
50
i
50
2
+
4
5
With a focus on optimizing the agonist activity of the pyrrolidineꢀ2ꢀacetic series, we next
sought to optimize the B ring via heteroaryl replacements and pendant substituents. This
investigation was accomplished by substitution of a nitrogen or introduction of a methyl group at
the 2ꢀ and 3ꢀpositions of the aryl ring. For comparison purposes, the effect of these changes is
illustrated with use of either a 2,4ꢀdichlorobenzyl C ring or a 2ꢀmethylbenzyl C ring (Table 2).
While the GPR40 activity of a 2,4ꢀdichlorobenzyl moiety in compound (S)ꢀ7 versus 2ꢀ
methylbenzyl group in compound 28 are not significantly different, the 2,4ꢀdichlorobenzyl
+
analogs had offꢀtarget activities (Na channel antagonist, PXR), which led us to focus upon the 2ꢀ
methyl derivatives for later analogs. Both the 2ꢀ and 3ꢀpyridineꢀcontaining analogs (compounds
2
6 and 29) had similar hGR40 potency to the parent phenyl analog (S)ꢀ7, but they demonstrated
a significant loss of mGPR40 activity. The 2ꢀ and 3ꢀmethyl substitutions (compounds 27 and 30)
were not tolerated on the B ring for either human or mouse GPR40 potency, suggesting that this
group may occupy a very narrow pocket in the receptor.
ACS Paragon Plus Environment
12

Journal of Medicinal Chemistry
Page 14 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 2. SAR of B Ring Analogs
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
3
Compound
Bꢀring
number
hEC50 (µM) ±
mEC50 (µM) ±
2
R
hK
i
(µM)
4
4
S.D.
S.D. (Ymax)
5
(S)ꢀ7
Ph
2,4ꢀCl
2,4ꢀCl
2,4ꢀCl
0.34 ± 0.20
0.89
n.d.
n.d.
4.5 ± 1.8 (66%)
>33
2
2
6
7
2ꢀpyridyl
2ꢀMeꢀPh
0.22 ± 0.047
5
5
8.9
14 (70%)
28
Ph
2ꢀMe
2ꢀMe
2ꢀMe
0.19 ± 0.080
0.45 ± 0.060
>33
n.d.
3.4 ± 2.2 (41%)
2
3
9
0
3ꢀpyridyl
1.8 ± 0.29
>33
>33
3ꢀMeꢀPh
n.d.
1
2+
2
3
hGPR40 EC in Ca FLIPR assay. Human binding K versus a radioligand. mGPR40 EC
50
i
50
2+
4
5
in Ca FLIPR assay. mean of at least two runs. N = 1.
The linkage between the B and C rings was studied to examine the effects of length on SAR
for GPR40 agonist activity (Table 3). A two atom linker, as in benzyl ether 28, showed good
hGPR40 potency. Shortening the linker to a methylene resulted in similar human potency. The
biaryl ether linker 32 provided the best hGPR40 agonist potency. The sulfoxide, sulfone, and
carbonyl linkers (compounds 33, 35, and 36) were not tolerated, suggesting that either a change
in conformation or polarity at this position was not tolerated. These results suggest that a one
atom linker was preferred for GPR40 activity. The one atom linker was also beneficial because
ACS Paragon Plus Environment
13

Page 15 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
S)ꢀ7 (X = CH O) displayed undesired PPARγ transactivation (17 ± 10 ꢁM, 54% Ymax in a
2
Gal4 gene reporter assay); whereas, compound 31 (X = CH ) showed no PPARγ transactivation
2
and compound 32 (X = O) showed only weak PPARγ transactivation (32 ± 19 ꢁM, 27% Ymax).
Table 3. SAR of Bꢀ and CꢀRing Linker, X
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3
Compound
number
mEC50 (µM) ±
1
4
2
X
hEC50 (µM) ± S.D.
hK (µM)
i
4
S.D. (Ymax)
28
31
32
OCH
2
0.19 ± 0.080
0.15 ± 0.075
0.074 ± 0.039
0.57 ± 0.31
>33
n.d.
3.4 ± 2.2 (73%)
CH
O
1.3 ± 0.53
4.4 ± 1.8
n.d.
>33
2
5.6 ± 3.3 (69%)
33
S
>17
>33
>33
>33
5
2
34
SO
SO
n.d.
35
>33
n.d.
6
36
CO
4.6
>33
1
2+
2
3
hGPR40 EC in Ca FLIPR assay. Human binding K versus a radioligand. mGPR40 EC
5
0
i
50
2
+
4
5
6
in Ca FLIPR assay. mean of at least two runs. mixture of diastereomers at sulfur. N = 1.
Having optimized the disposition of the carboxylic acid relative to the AꢀBꢀC rings, we sought
to explore Cꢀring substituents. Relative to the unsubstituted phenyl analog 37 (hGPR40 EC =
5
0
0.56 ꢁM), a 2,4ꢀdichloro group retained similar potency, as found in compound 38. Introduction
of a methyl group at the Cꢀring 2ꢀposition (compound 32) resulted in almost a 10ꢀfold
ACS Paragon Plus Environment
14

Journal of Medicinal Chemistry
Page 16 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
improvement in hGPR40 agonist potency compared to the unsubstituted compound 37. The 2ꢀ
methyl substituent seemed optimal, since a 2ꢀfluoro or a 2ꢀethoxy group resulted in potency
losses (compounds 39 and 40), indicating the GPR40 receptor pocket preferred a small group at
the ortho position. Larger groups, such as 2ꢀtꢀBu and 2ꢀPh (compounds 42 and 43) decreased
potency as well, suggesting a small binding pocket. Small substituents in the 3ꢀ and 4ꢀpositions
were tolerated (compounds 44ꢀ48) but they did not substantially affect potency as compared to
the 2ꢀposition. However, these modifications did not lead to a significant increase in mGPR40
potency and efficacy, which was desired to enable preclinical evaluation of GPR40
pharmacology.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 4. SAR of Substitution on C Ring
1
3
Compound
number
hEC50 (µM) ±
mEC50 (µM) ±
2
R
hK (µM)
i
4
4
S.D.
S.D. (Ymax)
37
38
H
0.56 ± 0.29
6.5 ± 0.33
n.d.
>16
5
5
2,4ꢀCl
2ꢀMe
2ꢀF
0.36
3.2 (120%)
3
3
4
2
9
0
0.074 ± 0.039
0.15 ± 0.10
4.4 ± 1.8
1.3 ± 0.0010
n.d.
5.6 ± 3.3 (69%)
6.9 ± 2.4 (50%)
>33
5
2ꢀOEt
1.7
ACS Paragon Plus Environment
15

Page 17 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
4
1
3ꢀMe
2ꢀtꢀBu
2ꢀPh
0.26 ± 0.17
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
2.4 ± 0.98 (83%)
5
42
43
44
6.5
>17
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3.5 ± 0.48
1.3 ± 0.68 (72%)
3ꢀCl
0.68 ± 0.18
>16
>33
>33
>33
>33
4
4
4
4
5
6
7
8
3ꢀNMe
4ꢀMe
4ꢀCl
2
1.3 ± 0.56
0.49 ± 0.14
0.29 ± 0.089
0.52 ± 0.33
4ꢀOMe
1
2+
2
3
hGPR40 EC in Ca FLIPR assay. Human binding K versus a radioligand. mGPR40 EC
50
in Ca FLIPR assay. mean of at least two runs. N = 1.
50
i
2
+
4
5
Having made significant changes to the linking moiety and Cꢀring substituent, we looked to reꢀ
optimize the pyrrolidine A ring, exploiting the synthetic intermediacy of prolinol derivatives,
many of which are readily available. Introduction of a methyl group at the 2ꢀ or 5ꢀpositions
dramatically reduced potency (compounds 49 and 57, Table 5). These SAR points suggest the
importance of the pyrrolidine heterocycle and the aniline planarity with the aryl ring. Small
substituents at the 3ꢀ and 4ꢀpositions (compounds 50 and 52) improved agonist potency against
both human and mouse GPR40 but unexpectedly the 4ꢀcisꢀfluoro substituent (compound 53),
while enhancing hGPR40 potency, exhibited attenuated potency versus the mouse receptor.
These results prompted us to synthesize the 4ꢀcisꢀCF analog 54, providing for the first time in
3
this series submicromolar agonist potency versus the mouse receptor. A similar result was
obtained with the oxygen linker, as seen in compound 22 (first depicted in Scheme 2). In
ACS Paragon Plus Environment
16

Journal of Medicinal Chemistry
Page 18 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
contrast to the 4ꢀcisꢀCF substituent, the polar 4ꢀcisꢀOMe and larger 4ꢀcisꢀphenyl groups were
3
disfavored (compounds 55 and 56). We hypothesized that one role of the 4ꢀcisꢀCF group is to
3
alter the conformation of the pyrrolidine ring. Indeed, an energetic exploration of the A and B
rings of unsubstituted analog 32 and analog 22 reveals that the unsubstituted pyrrolidine 32
weakly prefers to have the acid group pseudoꢀequatorial (by 2.2 kcal/mol) while the pseudoaxial
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
conformation of the acid group is more accessible with the 4ꢀCF ꢀsubstituted pyrrolidine 22
3
(negligible energy difference for the conformations) (Figure 4). The favorable profile afforded
by the 4ꢀcisꢀCF substituent led us to utilize this moiety for future SAR explorations. With the
3
optimized 4ꢀcisꢀCF group on the pyrrolidine A ring, our attention was again refocused on
3
further substitution of the phenyl C ring (Table 6). The 2ꢀMe group in compound 22 was
required for optimal potency. Substitution at the 3ꢀ, 4ꢀ, and 6ꢀpositions (compounds 60ꢀ64) led
to analogs with similar potency as compound 22.
Table 5. SAR of Substitution on A Ring
1
3
Compound
number
hEC50 (µM)
mEC50 (µM) ±
2
R
X
hK (µM)
i
4
4
± S.D.
S.D. (Ymax)
31
H
CH
2
2
0.15 ± 0.075
>33
1.3 ± 0.53
>33
>33
5
4
9
2ꢀMe
CH
23
ACS Paragon Plus Environment
17

Page 19 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
50
51
52
53
3ꢀtransꢀMe
3ꢀtransꢀPh
4ꢀcisꢀMe
4ꢀcisꢀF
CH
CH
CH
CH
CH
2
2
2
2
2
0.12 ± 0.065 0.31 ± 0.0050 1.9 ± 0.98 (27%)
11 ± 6.1
n.d.
>33
1.1 ± 0.75 (43%)
>33
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
0.14 ± 0.050
0.36
0.074 ± 0.037 0.18 ± 0.067
5
2
4
2
4-cisꢀCF
3
3
0.067 ± 0.026 0.073 ± 0.010 0.71 ± 0.30 (54%)
0.073 ± 0.047 0.035 ± 0.027 0.33 ± 0.33 (84%)
4ꢀcisꢀCF
O
O
55
4ꢀcisꢀOMe
4ꢀcisꢀPh
2.3 ± 1.2
>17
9.0 ± 3.9
>25
0.96 ± 0.17 (76%)
>33
5
5
6
7
CH
2
5
5ꢀcisꢀMe
CH
6.2
n.d.
>33
2
1
2+
2
3
hGPR40 EC in Ca FLIPR assay. Human binding K versus a radioligand. mGPR40 EC
50
i
50
2
+
4
5
in Ca FLIPR assay. mean of at least two runs. N = 1.
Figure 4. Overlay of low energy conformations of compounds 32 and 22 A and B rings only,
demonstrating the influence of a 4ꢀcisꢀCF group on the pyrrolidine ring structure.
3
Table 6. Effect of CꢀRing Substitutions on GPR40 Activity in CF ꢀContaining Pyrrolidine
3
Analogs
ACS Paragon Plus Environment
18

Journal of Medicinal Chemistry
Page 20 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
3
Compound
number
hEC50 (µM) ±
mEC50 (µM) ±
2
R
hK (µM)
i
4
4
S.D.
S.D. (Ymax)
2
5
5
6
6
6
2
8
9
0
1
2
2ꢀMe
0.073 ± 0.047
0.31 ± 0.11
0.039 ± 0.027
n.d.
0.33 ± 0.12 (84%)
1.3 ± 0.26 (42%)
0.74 ± 0.35 (44%)
1.5 ± 0.86 (34%)
0.17 ± 0.041 (41%)
0.50 ± 0.42 (100%)
0.43 ± 0.11 (50%)
0.27 ± 0.049 (36%)
2ꢀCF
3
2ꢀOCF
0.17 ± 0.067
0.12 ± 0.049
0.055 ± 0.025
0.11 ± 0.090
0.082 ± 0.054
0.082 ± 0.024
n.d.
3
2ꢀMe, 4ꢀOBn
2ꢀMe, 4ꢀOMe
2ꢀMe, 3ꢀMe
2ꢀMe, 3ꢀCl
0.010 ± 0.0060
0.039 ± 0.018
0.06 ± 0.02
0.013 ± 0.007
63
64
2ꢀMe, 6ꢀMe
n.d.
1
2+
2
3
hGPR40 EC in Ca FLIPR assay. Human binding K versus a radioligand. mGPR40 EC
50
i
50
2
+
4
in Ca FLIPR assay. mean of at least two runs.
24ꢀ26
While we thought that improved potency was possible,
we hoped that compound 22
would be of sufficient potency to blunt a glucose excursion in an intraperitoneal glucose
tolerance test (iGTT) in a normal lean mouse. The compound achieved a Cmax of 4.7 ꢁM (AUC0ꢀ
=
h
7.0 ꢁM*h) when it was administered orally at 10 mg/kg in mice, but there was no reduction
4
ACS Paragon Plus Environment
19

Page 21 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
in plasma glucose excursion after a glucose challenge. Even a 100 mg/kg dose was not
efficacious, suggesting that greater agonist potency and/or efficacy (Ymax) was required.
2
+
We wondered if the intrinsic efficacy (Ymax = 84%) of compound 22 in the Ca FLIPR assay
may have been misleading due to the artificial overexpression in the CHO cell line outlined
earlier. We suspected the suboptimal intrinsic efficacy may have contributed to the lack of an in
vivo effect in the mouse. As such, we decided to perform an inꢀdepth examination of its agonist
activity in hGPR40ꢀexpressing HEK cells. While it appeared that all of our compounds were full
agonists in the hGPR40 overexpressing CHO cell line, an inducible expression revealed partial
versus full agonists. Under conditions that induced high expression of hGPR40, compound 22
appeared to be a full activator as compared to linoleic acid (Figure 5). However, when tested
under conditions that induced low expression of hGPR40, we found that 22 was a partial
activator of hGPR40. While a similar experiment in the mouse cell line was not performed, we
surmised the lack of in vivo activity in the mouse pharmacology model was due to its partial
agonism. Additionally, when tested for in vitro GLPꢀ1 secretion in STCꢀ1 cells, a GPR40ꢀ
expressing murine enteroendocrine cell line that produces GLPꢀ1, compound 22 did not produce
a response.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We noticed that many GPR40 ligands, including linoleic acid/DHA, are significantly
longer than compound 22 and, given the sensitivity of GPR40 to ligand length (visꢀàꢀvis medium
and shortꢀchain fatty acids), we suspected that we may need to add substituents to the Cꢀring to
improve agonist efficacy and/or potency. Given that the Cꢀring was permissive to substitution at
the 3ꢀ and 4ꢀpositions, it made sense to append aryl and benzyl groups at these positions. In a
similar manner, scientists at Amgen had employed changes (meta versus para) to the central
linkage of their chemotype in conjunction with placement of a fluoroꢀmethoxyphenyl terminal
ACS Paragon Plus Environment
20

Journal of Medicinal Chemistry
Page 22 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
17,27
substituent.
This combination resulted in a more robust agonist efficacy response in their
chemical series. A 2ꢀF, 5ꢀOMe phenyl group was thus appended to biarylmethane and
biarylether substituents in both the unsubstituted and 4ꢀcisꢀCF ꢀsubstituted (S)ꢀpyrrolidines (e.g.
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
compounds 31 and 22). Appending the aryl ring to the 4ꢀposition of the phenyl Cꢀring
(compound 66) resulted in similar human GPR40 potency as compared with compound 31,
however it exhibited surprisingly robust agonist efficacy (Ymax = 160%) at the mouse receptor
with micromolar potency. In the 4ꢀCF ꢀsubstituted pyrrolidine series, connecting a 2ꢀF, 5ꢀOMe
3
phenyl group to the 3ꢀposition of the phenyl C ring resulted in a reduction of hGPR40 activity as
compared to 22, although Ymax (170%) was significantly enhanced. Appending the 2ꢀF, 5ꢀOMe
phenyl group to the 4ꢀposition of the Cꢀring in compound (S,Sꢀ68) resulted in reduced human
and mouse potency but very potent binding.
Table 7. Substitution of Aryl Groups on the Phenyl C Ring
1
3
Compound
number
3
4
hEC50 (µM)
2
mEC50 (µM) ±
S.D. (Ymax)
A
X
R
R
4
hK
i
(µM)
4
± S.D.
31
H
CH
2
H
H
H
0.15 ± 0.075
0.31 ± 0.17
1.3 ± 0.53
>33
6
5
H
O
n.d.
3.0 ± 2.4 (83%)
ACS Paragon Plus Environment
21

Page 23 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
F
5
6
6
H
CH
2
H
H
0.22 ± 0.11
0.31
1.9 ± 0.76 (160%)
0.33 ± 0.12 (84%)
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
0
0
.039 ±
.027
2
6
2
7
CF
3
O
O
H
0.073 ± 0.047
5
1
5
CF
H
0.50
0.049
0.78 (170%)
3
0
0
.045 ±
.010
(S,S)ꢀ68
CF
3
CH
2
H
0.49 ± 0.21
2.4 ± 1.3 (46%)
1
2+
2
3
hGPR40 EC in Ca FLIPR assay. Human binding K versus a radioligand. mGPR40 EC
50
in Ca FLIPR assay. mean of at least two runs. N = 1.
50
i
2
+
4
5
However, it is notable that in the earlier SAR developed for partial activators with the
CF ꢀpyrrolidine, including compound 22, the compounds were synthesized using commercial
3
(
2S,4S)ꢀ4ꢀ(trifluoromethyl)pyrrolidineꢀ2ꢀcarboxylic acid, resulting in enantiopure (>99% ee)
compounds. When compound 68 was initially synthesized, it was the first CF ꢀpyrrolidine
3
analog we made using CF ꢀprolinol that was synthesized inꢀhouse for scaleꢀup and analog
3
2
8
synthesis following a literature route. To evaluate the impact of agonist efficacy on in vivo
pharmacology, we scaled up compound 68 and, upon rigorous characterization, found that 68
ACS Paragon Plus Environment
22

Journal of Medicinal Chemistry
Page 24 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
contained ~10ꢀ20% of its enantiomer as a result of epimerization of an intermediate in the
synthesis of the 4ꢀCF ꢀprolinol.
3
The synthetic route to compound 68 begins with the oxidation of (2S,4R)ꢀ2ꢀbenzyl 1ꢀtertꢀbutyl
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4ꢀhydroxypyrrolidineꢀ1,2ꢀdicarboxylate 69 with PCC to give the carbonyl 70 (Scheme 3). The
tertiary alcohol 71 was formed by reaction with CF TMS and TBAF and it was eliminated
3
through the use of SOCl and pyridine. We postulate that ester 72 was partially racemized at this
2
step since this stereocenter controls formation of the CF stereocenter and no diastereomers were
3
observed in subsequent steps. The olefin was reduced concomitantly with hydrogenolysis of the
benzyl ester via H and Pd/C to give acid 73 possessing a cis relationship between the CF and
2
3
the carboxylic acid functionalities. The carboxylic acid was reduced using mixed anhydride
formation and NaBH reduction and the Boc group was removed to give prolinol salt 75. CF ꢀ
4
3
Prolinol 75 was coupled to 1ꢀbromoꢀ4ꢀiodobenzene followed by homologation via mesylation
and NaCN displacement to provide nitrile 77.
The CD ring system was formed via Suzuki coupling of (2ꢀfluoroꢀ5ꢀ
methoxyphenyl)boronic acid 78 and 4ꢀbromoꢀ2ꢀmethylbenzoic acid 79 in good yield. The
carboxylic acid was reduced with LAH to give alcohol 81, which was converted to bromide 82
via reaction with MsCl and LiBr. An organozinc reagent of bromide 82 was generated using
activated zinc, and this organometallic reagent was reacted with bromide 77 in good yield. Acid
68 was formed by KOHꢀmediated hydrolysis of the nitrile to give 68 as a scalemic mixture (as an
85:15 ratio of S,S:R,R). The enantiomers were separated by chiral SFC. Later, enantiopure
2
9
prolinol was obtained following an alternative literature route and/or via crystallization of a
diastereomeric amine salt with (1R)ꢀ(
̶
)ꢀ10ꢀcamphorsulfonic acid (see supporting information).
a
Scheme 3. Synthesis of (S,S)ꢀ68 and 68
ACS Paragon Plus Environment
23

Page 25 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a
Reagents and conditions. (a) PCC, CH Cl , 80%; (b) CF TMS, TBAF, 79%; (c) SOCl ,
2
2
3
2
pyridine, DIPEA, Boc O, 62%; (d) H , Pd/C, EtOH, 90%; (e) NMM, iꢀBuCOCl, NaBH ,
THF/water, 84%; (f) HCl/dioxane, 93%; (g) 1ꢀbromoꢀ4ꢀiodobenzene, CuI, NaOH, iꢀPrOH, 77%;
(h) MsCl, TEA, CH Cl ; NaCN, DMSO, 87% (2 steps); (i) TBAB, Pd(Ph P) , Na CO , water,
(92%); (j) LAH, THF, 89%; (k) LiBr, MsCl, TEA, THF, 89%; (l) Zn, TMSꢀCl, ethylene
dibromide; Pd(Ph P) , THF, 86%; (m) KOH, EtOH/water, 70%; (n) SFC chiral separation.
2
2
4
2
2
3
4
2
3
3
4
Compound 68 was initially tested in in vitro assays as a scalemic mixture since it appeared to
be very potent and efficacious (high Ymax) in both human and mouse FLIPR assays. The
intrinsic efficacy compound 68 was evaluated using a cell line with inducible expression of
hGPR40 (Figure 5). The inducible assay results showed that scalemic 68 exhibited full
ACS Paragon Plus Environment
24

Journal of Medicinal Chemistry
Page 26 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
activation of GPR40 even at low expression levels of receptor. A similar observation was made
in a mouse inducible expression cell line (data not shown). The interꢀspecies consistency of
GPR40 agonism enabled preclinical characterization of 68 in mouse with improved confidence
that it could model human pharmacology.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. (A) Inducible hGPR40 assay with high expression of GPR40. (B) Inducible hGPR40
assay with low expression of GPR40.
Once the enantiomers were separated, we were surprised to find that (S,S)ꢀ68 (as a single
enantiomer) retained potent hGPR40 binding (0.045 µM) but (R,R)ꢀ68 was the more potent and
efficacious agonist (Figure 6). Furthermore, compound (R,R)ꢀ68 occupied a binding mode that
positively potentiated the radioligand used for the binding assay while (S,S)ꢀ68 displaced the
radioligand (Figure 7).
ACS Paragon Plus Environment
25

Page 27 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
F
F
N
(S)
N
(R)
CF
3
CF
3
(
S)
(S)
(
S,S)-68
OMe
OMe
84
O
OH
HO
O
hGPR40 = 0.49 ± 0.21 ꢀM
mGPR40 = 2.4 ± 1.3 ꢀM (46%)
hGPR40 = 1.7 ꢀM
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
mGPR40 = 1.1 ꢀM (140%)
F
F
N
(R)
N
(S)
CF
3
CF
3
(R)
(R)
(
R,R)-68
85
OMe
OMe
O
O
HO
HO
hGPR40 = 0.11 ± 0.056 ꢀM
mGPR40 = 0.054 ꢀM ± 0.036 (210%)
hGPR40 = 0.15 ± 0.11 ꢀM
mGPR40 = 0.053 ± 0.028 ꢀM (210%)
Figure 6. Impact of pyrrolidine substituent stereoisomers on GPR40 binding potency and
agonist efficacy.
Figure 7. Binding assay results with compounds (S,S)ꢀ68 and (R,R)ꢀ68 (negative slope indicates
radiolabel displacement; positive slope indicates increase in radiolabel binding).
ACS Paragon Plus Environment
26

Journal of Medicinal Chemistry
Page 28 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
In an effort to deduce which chiral center was responsible for the change in binding mode
and agonist efficacy, the (2S,4R)ꢀ and (2R,4S)ꢀ(trifluoromethyl)ꢀanalogs of 68 were synthesized.
Compound 84 with (2S)ꢀstereochemistry remained a micromolar partial activator while (2R)ꢀ
analog 85 retained full activation and good potency, suggesting that the (R)ꢀacetic substituent
potently influenced agonist activity in combination with the 2ꢀF,5ꢀOMe phenyl ring (Figure 6).
We were surprised to find that (R,R)ꢀ68 was the most potent analog, since earlier work in the
chemotype indicated that the (S)ꢀenantiomer was always more potent across 40 different
enantiomeric pairs tested (see supporting information). It was subsequently reported in the
literature that others had also observed that a change in stereochemistry in different chemotypes
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
3,27
could result in complimentary binding modes and full agonist efficacy.
Since free fatty acids are present in circulation, we studied the activation of hGPR40 by
compounds (S,S)ꢀ68 and (R,R)ꢀ68 in combination with DHA, a potent and efficacious
4
endogenous long chain fatty acid (Figure 8). At subꢀefficacious (3 nM) concentrations,
compound (R,R)ꢀ68 significantly enhances the agonist activity of DHA and left shifts the
concentration response curve. On the other hand, (S,S)ꢀ68 does not significantly potentiate the
activity of DHA in vitro. Even concentrations of (S,S)ꢀ68 that show significant agonism do not
shift the DHA curve (data not shown). This data supports a change in binding mode, suggesting
that compound (R,R)ꢀ68 can be a full agonist and a positive allosteric modulator of DHA’s effect
on hGPR40.
ACS Paragon Plus Environment
27

Page 29 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 8. (A) Potency of DHA versus DHA + 3 nM compound (R,R)ꢀ68. (B) Potency of DHA
versus DHA + 10 nM compound (S,S)ꢀ68.
We evaluated the effect of compound (R,R)ꢀ68 on insulin and GLPꢀ1 secretion in vitro.
In isolated primary mouse islets, (R,R)ꢀ68 showed a doseꢀdependent (0.016ꢀ2.0 ꢁM) increase in
insulin secretion (Figure 9), demonstrating that (R,R)ꢀ68 promoted glucoseꢀstimulated insulin
secretion. Compound (R,R)ꢀ68 was also quite potent in the STCꢀ1 cell line, a mouse
30,31
enteroendocrine cell line,
showing a 0.33 ꢁM EC50 for GLPꢀ1 secretion (Figure 10).
Interestingly, enantiopure (S,S)ꢀ68 also exhibited GLPꢀ1 secretion but with an EC of 3.2 ꢁM,
50
(10ꢀfold less potent than its enantiomer). These findings demonstrated that these full activators
show a robust dual mechanism of action as compared to partial agonists, effecting both glucoseꢀ
stimulated insulin secretion as well as GLPꢀ1 secretion. However, it remained unclear if the dual
mechanism of action was due to its binding mode, agonist efficacy, or another determinant.
ACS Paragon Plus Environment
28