N-Thiazolylamide-based free fatty-acid 2 receptor agonists: Discovery, lead optimization and demonstration of off-target effect in a diabetes model

Article abstract of DOI:10.1016/j.bmc.2018.09.015

Free fatty acid-2 (FFA2) receptor is a G-protein coupled receptor of interest in the development of therapeutics in metabolic and inflammatory disease areas. The discovery and optimization of an N-thiazolylamide carboxylic acid FFA2 agonist scaffold is described. Dual key objectives were to i) evaluate the potential of this scaffold for lead optimization in particular with respect to safety de-risking physicochemical properties, i.e. lipophilicity and aromatic content, and ii) to demonstrate the utility of selected lead analogues from this scaffold in a pertinent in vivo model such as oral glucose tolerance test (OGTT). As such, a concomitant improvement in bioactivity together with lipophilic ligand efficiency (LLE) and fraction sp³ content (Fsp³) parameters guided these efforts. Compound 10 was advanced into studies in mice on the basis of its optimized profile vs initial lead 1 (ΔLLE = 0.3, ΔFsp³ = 0.24). Although active in OGTT, 10 also displayed similar activity in the FFA2-knockout mice. Given this off-target OGTT effect, we discontinued development of this FFA2 agonist scaffold.

Full text of DOI:10.1016/j.bmc.2018.09.015

Accepted Manuscript
N-Thiazolylamide-based free fatty-acid 2 receptor agonists: Discovery, lead
optimization and demonstration of off-target effect in a diabetes model
Hamid R. Hoveyda, Graeme L. Fraser, Ludivine Zoute, Guillaume Dutheuil,
Didier Schils, Cyrille Brantis, Alexey Lapin, Julien Parcq, Sandra Guitard,
François Lenoir, Mohamed El Bousmaqui, Sarah Rorive, Sandrine Hospied,
Sébastien Blanc, Jérôme Bernard, Frédéric Ooms, Joanne C. McNelis, Jerrold
M. Olefsky
PII:
S0968-0896(18)30110-X
https://doi.org/10.1016/j.bmc.2018.09.015
BMC 14540
DOI:
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
18 January 2018
13 September 2018
16 September 2018
Please cite this article as: Hoveyda, H.R., Fraser, G.L., Zoute, L., Dutheuil, G., Schils, D., Brantis, C., Lapin, A.,
Parcq, J., Guitard, S., Lenoir, F., Bousmaqui, M.E., Rorive, S., Hospied, S., Blanc, S., Bernard, J., Ooms, F., McNelis,
J.C., Olefsky, J.M., N-Thiazolylamide-based free fatty-acid 2 receptor agonists: Discovery, lead optimization and
demonstration of off-target effect in a diabetes model, Bioorganic & Medicinal Chemistry (2018), doi: https://
doi.org/10.1016/j.bmc.2018.09.015
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

N-Thiazolylamide-based free fatty-acid 2
receptor agonists: Discovery, lead optimization
and demonstration of off-target effect in a
diabetes model
Hamid R. Hoveyda,*,1 Graeme L. Fraser, 1 Ludivine Zoute, 1 Guillaume Dutheuil, 1 Didier
Schils,1, † Cyrille Brantis, 1,■ Alexey Lapin, 1 Julien Parcq, 1 Sandra Guitard, 1 François
Lenoir, 1 Mohamed El Bousmaqui, 1 Sarah Rorive,1 Sandrine Hospied, 1 Sébastien Blanc,1
Jérôme Bernard,1,□ Frédéric Ooms,1,‡ Joanne C. McNelis,2 Jerrold M. Olefsky 2
1 Ogeda SA (formerly Euroscreen), 47, rue Adrienne Bolland, 6041 Gosselies, Belgium
2 University of California, Department of Medicine, San Diego, CA 92161, USA
Keywords: FFA2 agonist, GPR43 agonist, GPCR, OGTT, LLE, Fsp3
1

RECEIVED DATE
Abstract:
Free fatty acid-2 (FFA2) receptor is a G-protein coupled receptor of interest in the
development of therapeutics in metabolic and inflammatory disease areas. The discovery
and optimization of an N-thiazolylamide carboxylic acid FFA2 agonist scaffold is
described. Dual key objectives were to i) evaluate the potential of this scaffold for lead
optimization in particular with respect to safety de-risking physicochemical properties, i.e.
lipophilicity and aromatic content, and ii) to demonstrate the utility of selected lead
analogues from this scaffold in a pertinent in vivo model such as oral glucose tolerance test
(OGTT). As such, a concomitant improvement in bioactivity together with lipophilic
ligand efficiency (LLE) and fraction sp3 content (Fsp3) parameters guided these efforts.
Compound 10 was advanced into studies in mice on the basis of its optimized profile vs
initial lead 1 (∆LLE = 0.3, ∆Fsp3 = 0.24). Although active in OGTT, 10 also displayed
similar activity in the FFA2-knockout mice. Given this off-target OGTT effect, we
discontinued development of this FFA2 agonist scaffold.
1. Introduction
We have had a longstanding interest in free fatty acid-2 (FFA2) receptor (previously
called GPR43 receptor) as a potential GPCR target towards discovery of novel
therapeutics. In 2003, our colleagues1 as well as Brown and co-workers2 reported on the
deorphanization of FFA2 receptor and on the role of short chain fatty acids (SCFAs),
2

consisting of 1-6 carbon atoms, such as acetate and propionate, as potent and selective
endogenous agonists to FFA2. Our subsequent efforts culminated in the discovery of
several agonist series (e.g. 1, Figure 1)3 as well as an antagonist4 to FFA2 receptor (2) thus
far mainly documented in our patent filings. These disclosures have in turn fueled recent
activities by several groups reporting on their findings with our lead structures5 as well as
its widespread citation as FFA2 agonist reference ligands.6
FFA2 receptor is a member of the class A GPCR sub-family that also includes FFA1
(GPR40) and FFA3 (GPR41) receptors, which share 30-40% sequence identity and display
specificity toward free fatty acids (FFAs) of various chain lengths.5 The FFA2 receptor is
expressed in adipose tissue and primary adipocytes and is reported to have a role in
promoting adipogenesis through increasing lipid accumulation and inhibiting lipolysis.7 In
addition, FFA2 receptor is also present in enteroendocrine L-cells,8 wherein it mediates
GLP-1 release, as well as being expressed in pancreatic islets (β-cells).9 Despite the
expression of FFA2 in tissues related to lipolysis and glucose metabolism, it is notable that
phenotyping studies in FFA2-/- mice have yielded contradictory findings with some
reports suggesting that a FFA2 antagonist would be the desired mode of action10,11 whereas
other studies indicating that an agonist would be required.8,12,13,14,15 As there are several
examples of FFA2-/- mice displaying phenotypic differences in glucose tolerance,8,12,13,14 the
oral glucose tolerance test (‘OGTT’) was used as a measure of in vivo compound efficacy
in the current work.
FFA2 receptor is also highly expressed in immune cells such as peripheral blood
mononuclear cells (PBMC) and polymorphnuclear cells (PMN) with a particularly high
3

expression in the neutrophils.1,2 In FFA2 receptor knockout mice results in acute colitis
studies have reported conflicting results with both reduced16 and heightened inflammatory
responses.17 More recently, however, seminal studies reported by Garrett and co-workers
have helped clarify a role for FFA2 receptor activation in terms of regulation of the colonic
Treg pool and protection against colitis.18 The findings by Garrett et al. combined with the
recent lack of success of a FFA2 receptor antagonist in the clinic19 tend to suggest that
agonism of FFA2 receptor is perhaps the relevant mode of action for the treatment of
inflammatory disorders.
Some years ago we initiated a screening of carboxylic acid focused libraries using GTP-
γ35S assay against hFFA2 receptor. Initially these efforts resulted in the discovery of the N-
thiazolylamide-based carboxylic acid agonist series as well as a potent antagonist to FFA2
receptor (Fig 1). Importantly, antagonist 2 proved a valuable tool in our investigations as
it is a potent competitive inhibitor of propionate, the most potent endogenous ligand to
FFA2 receptor.20 Consequently, the [3H]-labeled form of 2 (ES227703) was used in
21
radioligand binding studies where it is displaced by the foregoing N-thiazolylamide
agonists, but not by the previously identified Amgen allosteric agonist 3.22,23
The global objectives for the work described herein were to demonstrate the development
potential of our N-thiazolylamide carboxylic acid FFA2 scaffold in terms of i) lead
optimization of key physicochemical properties that impact drug safety (i.e., lipophilicity
and aromatic content), and ii) validation of the utility of this scaffold in the metabolic axis
using OGTT assay, for which a FFA2-/- mouse model exists. Our results and conclusions
with respect to these objectives are summarized below.
4

Figure 1. Reference ligands herein (nomenclature for SAR discussions indicated).
2. Results and discussion
2.1. Lead optimization strategy
Improving biological activity in parallel to the optimization of physicochemical and
pharmacokinetics (PK) properties is a sound approach in modern drug discovery. Property
optimization integrated early in the discovery process can help reduce overall early phase
attrition due to inadequate safety and/or PK profiles.24 Our lead optimization followed the
same guidelines and aimed to concurrently improve the bioactivity (vs hFFA2 receptor) as
well as the physicochemical properties of the lead structures. Ligand lipophilic efficiency25
(LLE = pEC50 – logD7.4)26 and fraction sp3 (Fsp3 = number of sp3 hybridized carbons/total
carbon count) 27 were principally used to track optimization of properties at this stage of
lead optimization. It is well recognized that increased lipophilicity28 and high aromatic
content (structural flatness)29 correlate with increased safety risks that lead to greater lead
attrition and failure to reach the clinic. It has also been noted that ligands for different
target families can be categorized on the basis of the physicochemical property range.30
Overall, we aimed for improved bioactivity combined with ∆LLE > 0 and ∆Fsp3 > 0 versus
the initial lead 1 (LLE > 4.7 and Fsp3 > 0.19) in our lead progression efforts not only to
5

obtain a proof-of-concept (POC) lead to validate the underlying pharmacology, but also to
evaluate the potential of this agonist scaffold for physicochemical properties optimization
relevant to toxicity de-risking strategy.
2.2. In vitro bioactivity SAR
In vitro bioactivity was tracked primarily using GTP-γ35S against hFFA2 receptor in
Chinese hamster ovary (CHO) cells. GPCR activation by an agonist would ultimately
result in the incorporation of GTP-γ35S (guanosine 5'-O-[gamma-thio]triphosphate) that
cannot be cleaved to continue the GTP to GDP exchange cycle. Thus, accumulation of
GTP-γ35S provides a measure of agonist activity and can be used to derive agonist potency
through EC50 and %Emax values at both human and murine orthologues of FFA2 receptor.
The GTP-γ35S assay on hFFA2 was established using propionate as a reference agonist20 to
which our synthetic agonists were relatively compared. Investigation of ligand-receptor
affinity with the more active structures was achieved through radioligand binding (RLB)
assay using displacement of [3H]-2 from recombinant hFFA2 receptor in CHO cells.21
Since 2 is only active against hFFA2 receptor and not against the murine orthologue, the
RLB data are available merely against hFFA2 receptor (Table 1).31 Linear regression
analysis for the GTP-γ35S versus the RLB assay displayed a good fit with r2 = 0.70. The
GTP-γ35S (pEC50, %Emax) and RLB (pKi) data were obtained on each compound at least in
triplicate with standard deviation • 0.3 (Table 1).
As stated above, GTP-γ35S data against hFFA2 receptor was the primary SAR driver.
Overall, the lead structures herein (1, 4-13 in Table 1) were 0.1−0.9 log right-shifted in
6

comparing human vs murine FFA2 receptor bioactivity data (pEC50, GTP-γ35S). Chiral
SAR was evident since 4, the chiral antipode to 1, displayed a 0.5-log right-shift in the
GTP-γ35S assay and was inactive in the RLB assay. Ester or carboxamide congeners of 1
proved inactive in both assays (data not shown) to thereby establish the indispensability of
the carboxylic acid moiety consistent with the findings of previous reports.32
For brevity, shorthand “Ring A” and “Ring C” notations are used for SAR discussions
herein (defined in Figure 1). Aromatic content reduction, or Fsp3 increase, was feasible
through Ring A modifications. For example, cyclopentyl Ring A analogue 5 displayed
0.5-log and 0.2-log improved bioactivity in both GTP-γ35S and RLB assays against hFFA2
and mFFA2 (Table 1), respectively, when compared against its phenyl Ring A congener 1
(∆Fsp3 = 0.26). However, LLE remained essentially unchanged in this analogy since the
gain in bioactivity (5 vs 1: ∆pEC50,
= 0.5) was essentially nullified by a comparable
hFFAR2
gain in lipophilicity in the phenyl to cyclopentyl replacement (∆logD7.4 = 0.6). An attempt
to remedy this by introducing polar heteroatoms, e.g. cyclopentyl to tetrahydropyran
modification, was unsuccessful due to abrogated bioactivity (6 vs 5: ∆pEC
= −0.8,
50, hFFAR2
∆pEC50,
= −1). In contrast, amide nitrogen substitution provided a means of
mFFAR2
improving bioactivity and LLE. For instance, converting the N-methyl amide in 1 to the
corresponding N-cyclopropyl amide congener 7 resulted in 0.5-log and 0.2-log improved
bioactivity against hFFA2 and mFFA2, respectively, without an adverse lipophilicity effect
(∆logD7.4 = 0). The N-cyclopropyl amide moiety was retained for further rounds of lead
optimization (7-13, Table 1) due to its appreciable impact on bioactivity and the early-
7

stage goal of achieving efficacy in the OGTT assay for pharmacology validation of this
agonist series.
In contrast to the permissive SAR in the Ring A region of the lead structure, the SAR in
the Ring C region was more restrictive, for instance, cycloalkyl or heteroaryl ring
replacement of 2-chlorophenyl (1, Figure 1) abolished bioactivity altogether (data not
shown). An initial notable observation in Ring C SAR was that 2,5-dichloro substitution
on the phenyl Ring C (8) was more active by 0.4-log and 0.5-log against hFFA2 and
mFFA2, respectively, versus its 2-chlorophenyl congener 7. However this enhanced
activity came at the expense of increased lipophilicity (7 vs 8: ∆logD7.4 = +0.7) and
consequently the more active lead 8 proved inferior in LLE to 7 by 0.2-0.3 log units (Table
1). As such, we viewed compound 8 as a prototype early generation lead and sought to use
the 2,5-substitution pattern on the phenyl Ring C to achieve, as much as possible,
concomitant improvement in bioactivity and physicochemical properties.33 After an
extensive systematic investigation, we discovered that replacement of either of the 2- or 5-
chloro substituents in Ring C with a heteroaryl ring was well tolerated. Therefore, we
reasoned that this substitution strategy is potentially useful towards advanced leads with a
more balanced bioactivity and LLE, Fsp3 profiles. For example, the analogy between lead
structures 9 vs 8 revealed the possibility of achieving the same enhanced bioactivity but
with an improved lipophilicity (9 vs 8: ∆logD7.4 = −0.5) as reflected by a 0.5-log improved
LLE, notwithstanding that both Ring C and Ring A modifications were used to achieve the
said results. Ultimately, however, we concluded that the synthetic complexity of the 5-
chloro-2-(6-methoxypyridin-3-yl)phenyl Ring C construct exemplified through analogue 9
combined with the added lipophilicity due to 5-chloro substitution outweighed the
8

improved activity. Therefore, we dropped the 5-chloro moiety in the Ring C construct in 9
in favor of the simpler 2-(6-methoxypyridin-3-yl)phenyl variant and restored the
cyclopentyl Ring A to aid with Fsp3 increase, which altogether resulted in lead 10 that
proved to be the first POC lead in this series as discussed further below. In further
attempts to reduce lipophilicity in the advanced lead structure 10, we used modifications in
both Ring A and Ring C. Replacement of the methoxy moiety in the 2-(6-methoxypyridin-
3-yl)phenyl Ring C construct in 10 (logD7.4 = 2.6) with a pyrazole ring, i.e. 11 (logD7.4
=
3.1) was counterproductive as it led to 0.5-log loss in LLE consistent with the similar gain
in lipophilicity. Moreover, converting the cyclopentyl ring in 10 to the corresponding
isopropyl congener provided compound 12 that displayed a similar bioactivity, but with an
improved LLE (∆LLE = 0.4 based on hFFA2 pEC50). The 11-fold superior plasma
unbound fraction in isopropyl analogue 12 versus the more lipophilic cyclopentyl congener
10 (∆logD7.4 = −0.4) combined with >7.5-fold improved unbound total clearance in the
foregoing analogy was also noteworthy. Similar trends were also noted in comparing the
cyclopentyl analogue 11 to its congener 13 (see Tables 1 and 2).
2.3. In vivo efficacy (OGTT) and verification of mode-of-action in FFA2 knockout mice
The oral glucose tolerance test (OGTT) is a common screen performed in mice to
measure compound effects on fasting glucose relevant to drug discovery in the field of
diabetes and related metabolic disorders. Moreover, phenotypic differences in the OGTT
9

assay have been reported from various groups using FFA2 -/- mice suggestive of a role of
FFA2 in glucose tolerance.8,12,13,14 Thus, the OGTT assay was selected as an approach to
compare the efficacy of compounds in vivo. Selection of compounds for in vivo efficacy
using the OGTT in mice was predicated on the in vitro pharmacology profiling as well as
oral in vivo exposure profile, in particular the dose-normalized unbound oral exposure in
mice (cf. oral AUCu /dose values in Table 2). Although early generation lead structures
such as 1 and 7 had reasonable oral exposure (AUCu) in mouse, these analogues proved to
be OGTT inactive. While various groups have published in vitro results using either
Amgen’s allosteric agonist 3, or our early generation lead 8 in assays such as glucose-
33,34Error! Bookmark not defined.
stimulated insulin secretion,12, lipolysis or GLP-1 release,
neither 3
nor 8 have ever been reported to elicit any in vivo effects, likely due to their rather limited
PK exposure (see Table 2). At this stage of testing, we conjectured that perhaps in order to
achieve in vivo efficacy, further enhanced bioactivity combined with similar or improved
PK exposure vis-à-vis lead 1 was required. Our lead optimization strategy had provided
several second generation lead structures (e.g. 10, 12 and 13) with improved bioactivity vs
1 (∆pEC50 = 0.3−0.4) with similar (10) or improved (12 and 13) dose-normalized unbound
oral exposure in mice vs 1. To our initial satisfaction, these second generation lead
structures, i.e. 9-13 (Table 2), proved active in the OGTT assay.
Once in vivo compound efficacy was established, a subsequent experiment in wild-type
versus FFA2-/- mice was carried out with the POC lead 10, as an OGTT-active lead
representative, in order to evaluate the underlying target specificity. As shown in Figure 2,
10 lowered glucose significantly in both wild-type and in FFA2 receptor knock-out mice
suggesting that the OGTT effect observed is off-target and not FFA2 receptor mediated.35
10

It is important to note that 10 proved inactive at 10 µM against alternate FFA receptors
(FFA1, FFA3, FFA4) as well as various Type A and Type B GPCRs implicated in
metabolic diseases (see Supporting Information for details). Therefore, off-target activity
at these related receptors is unlikely to be the cause of the observed OGTT effect and the
basis of the OGTT response to 10 remains unexplained. We decided at this stage to
discontinue further development of this agonist scaffold based on lack of evidence for
FFA2-specific in vivo efficacy.
11

Table 1. Ligand Efficiency Metrics and In Vitro Bioactivity SAR.
GTP-γ35S assay d
RLB
3
a
logD7.4 Fsp b LLEc
%Emax pKi
Cpd
Structure
see Fig 1
pEC50
4.7
h
m
h
7.0
6.6
7.0
6.8
6.5
6.1
106
81
94
77
57
69
6.1
1
2.3
3.9
2.3
0.19
0.29
0.19
4.3
3.1
2.9
4.2
3.8
-
-
3
4
see Fig 1
m
h
-
< 5
-
S-enantiomer of 1
m
4.6
3.9
5.4
4.5
5.2
4.5
4.9
h
m
h
7.5
6.8
6.7
5.8
7.5
6.8
7.9
84
63
6.6
-
5
6
7
2.9
1.3
2.3
0.45
0.45
0.26
103
76
5.7
-
m
h
104
86
6.4
-
m
h
125
6.7
8
9
3.0
2.5
2.6
0.26
0.26
0.43
4.3
5.4
4.8
m
h
7.3
7.9
7.3
113
126
94
-
7.0
-
m
5.0
4.4
h
7.6
7.0
110
81
6.8
-
10
m
4.5
4.0
h
7.6
7.1
108
72
7.2
-
11
3.1
0.37
m
12

5.4
4.7
5.2
h
m
h
7.6
6.9
7.6
103
84
6.6
-
12
13
2.2
2.4
0.39
0.32
104
6.8
4.6
m
7.0
88
-
a LogD measured by HPLC at pH 7.4, N = 3 and %RSD • 2 (See Supporting Information for
details).26 b Fsp3 or fraction sp3 = number of sp3 hybridized carbons/total carbon count. c LLE =
pEC50 – logD7.4. GTP-γ35S assay data for human (h) and murine (m) orthologues of FFA2
d
receptor provided. GTP-γ35S (pEC50, %Emax) and RLB (pKi) data were obtained on each compound
at least in triplicate with standard deviation • 0.3.
13

Table 2. Plasma Protein Binding, Pharmacokineticsa and OGTT Efficacy Profiles.
mouse PK
(50 mg/kg)
plasma fu
rat mouse
rat PK (iv 1 mg/kg, oral 3 mg/kg)
OGTT %
inhibition
(peak) at 50
mg/kg b
Cpd
iv CLu iv T1/2 %F oral AUCu / oral AUCu
dose
2.4
/dose
2.0
(min)
482
−8 5%
(p >0.05)
1
0.0008 0.0015
738
77
15
7
6%
(p >0.05)
5%
(p >0.05)
5%
3 c 0.0062 0.0052
6666
220
0.08
0.045
1
7
8
0.0016 0.0025
0.0001 0.0001
0.0026 0.0041
0.0006 0.0023
1044
2800
2738
2617
6692
342
319
590
579
240
155
431
280
52
47
38
47
25
35
29
0.70
0.26
0.26
0.29
0.07
2.1
4.7
0.32
1.2
5
(p > 0.05)
32 6%
(p <0.001)
33 7%
(p <0.001)
40 4%
(p <0.001)
19 5%
(p <0.01)
26 5%
(p <0.001)
9
2.1
10
11 0.0013 0.0014
0.0066 0.0107
13 0.0057 0.0075
0.53
4.7
12
1091
0.55
3.3
a PK formulation: 1% DMSO, 9% hydroxypropyl-β-cyclodextrin in 0.9% NaCl. Unbound PK parameters
calculated as follows: iv CLu (mL/min/Kg) = CLT, iv / (plasma fu); AUCu = AUCT (in µM.min) x (plasma fu).b
OGTT formulation is the same as that for PK studies but without DMSO. Statistical analysis performed by 2-
way ANOVA followed by Bonferroni’s posthoc analysis directly provided in parenthesis. OGTT %
c
inhibition presented as data SEM. Statistically significant OGTT data highlighted in bold. Tested at 3
mg/kg in mouse PK and 8.4 mg/kg in OGTT due to limited solubility.
14

(a) Wild-type mice
(b) FFA2-/-mice
Figure 2. Oral glucose tolerance test (OGTT) assay results with 10 (50 mg/kg) in (a) wild-type
and (b) FFA2 receptor knock-out mice.12 Vehicle: 9% hydroxypropyl-β-cyclodextrin in 0.9%
NaCl. The results are shown as mean SEM. Statistical analysis performed by 2-way ANOVA
followed by Bonferroni’s posthoc analysis: *** p< 0.001, **p<0.01.
2.4. Chemistry
The preparative routes used to access the compounds described in this paper are illustrated in
Schemes 1-3. The synthesis of N-thiazolylamide-based analogues required access to two key
reagents: the Ring A (Figure 1) bearing chiral building block, e.g. 17 (Scheme 1), and the 2-
aminothiazole building block bearing Rings B and C, e.g. 23 (Scheme 1). For analogues 1, 4, 7
and 8, the chiral carboxylic acid building block (17, 18) was prepared through Horner-
Wadsworth-Emmons synthesis followed by Noyori asymmetric hydrogenation. The chiral acid-
ester building blocks for the remaining of lead structures herein (5-6, 9-13) were prepared using
the Evans oxazolidinone chiral auxiliary approach (42-45, Scheme 2). The 2-aminothiazole
building blocks (23-25, Scheme 1) were assembled using Hantzch thiazole synthesis through
condensation of substituted thioureas with α-bromoacetophenones. For lead compounds 9-13
(Table 1), the corresponding 2-aminothiazole building blocks 50-53 were prepared by a
15

combination of Hantzsch thiazole synthesis and Pd-catalyzed Suzuki cross-coupling (Scheme 3).
The final target structure was assembled in ester form through HATU-mediated amide coupling
between the chiral carboxylic acid and 2-aminothiazole building blocks at room temperature or
at 60 °C with sterically encumbered N-cyclopropyl-2-aminothiazoles (24-25 and 50-53) to
achieve comparable yields.36 Finally, TFA-mediated deprotection of the tert-butyl ester moiety
afforded the final product in free carboxylic acid form. We favored the use of tert-butyl ester
precursors since the acidolytic deprotection with TFA was less prone to racemization in contrast
to saponification of related methyl ester derivatives in aqueous milieu.
Scheme 1. Synthesis of analogues 1, 4, 7 and 8.a
a
Reagents and conditions: (a) nBuLi, THF, 5-15 °C then PhCHO, THF, 5-15°C to rt; (b)
nBu4OH (1M in water), THF, rt; (c) dicyclohexylamine, RuCl2[(S)-BINAP], H2 (8-10 bar),
MeOH, 55°C; (d) EtOH, reflux; (e) HATU, DIEA, CH3CN, rt (with 23) or 60 °C (with 24 and
25); (f) TFA, DCM, rt.
16

Scheme 2. Preparation of chiral acid building blocks via Evans oxazolidinone auxiliary
approach to synthesize 5, 6, 9-13. a
a Reagents and conditions: (a) TEA, pivaloyl chloride, THF, −78 °C to rt; (b) nBuLi, (S)-4-
benzyloxazolidin-2-one, THF, −78 °C; (c) NaHMDS, tert-butyl 2-bromoacetate; THF, −78 °C to
rt; (d) H2O2/LiOH, THF/H2O, 0 °C to rt.
Scheme 3. Preparation of N-cyclopropropyl-4-biaryl-thiazole-2-amine building blocks used
to synthesize 9-13. a
a Reagents and conditions: (a) (i) AlCl3, Br2, THF, 0 °C to rt; (ii) 1-cyclopropylthiourea, EtOH,
reflux; (b) ArB(OH)2, K2CO3, S-Phos, Pd2(dba)3, toluene, 85 °C; (c) nBuLi, B(Oi-Pr)3, THF,
−78°C to rt; (d) Pd(PPh3)4, ArBr, K2CO3, toluene, 85 °C.
3. Conclusion
As stated in the Introduction, we had two global objectives in the work presented herein. The
first objective was to verify if the N-thiazolylamide carboxylic acid FFA2 agonist scaffold that
17

we had discovered in our HTS campaigns is amenable to physicochemical properties
optimization relevant to safety de-risking strategies. Lipophilicity and aromatic content
optimization were thus undertaken and tracked through the use of LLE and Fsp3 metrics to gauge
whether these properties lend themselves to improvement against the initial lead 1. As depicted
in Figure 3 in the early round of lead optimization, compound 8 emerged as a lead displaying
improved LLE and Fsp3 albeit with the undesirable 0.7-log increase in lipophilicity. For the
latter reason, we viewed structure 8 as a prototype towards more balanced lead structures in
terms of physicochemical properties relevant to safe drug development. Further lead
optimization efforts furnished structures with much improved Fsp3 (10-13) and lipophilicity
similar to the initial lead 1 (10, 12-13). Among these, compound 10 was chosen as the POC lead
candidate. Overall, these results indicated that the physicochemical property optimization of this
scaffold is feasible, although more limited in terms of lipophilicity improvement than aromatic
content.
18

9
12
6
13
7
advanced
lead op
8
initial
lead op
10
Fsp3 = 0.26
logD7.4 = 3.0
1
Fsp3 = 0.43
logD7.4 = 2.6
5
Fsp3 = 0.19
logD7.4 = 2.3
11
4
∆Fsp3 vs. 1
Figure 3. Lead optimization progress as a plot of ∆LLE = LLE (cpd x) – LLE (cpd 1) versus
∆Fsp3 = Fsp3(cpd x) – Fsp3 (cpd 1). LLE and Fsp3 as defined in Table 1. Color code: green = low
lipophilicity (i.e. logD7.4) and red = high lipophilicity. Size of dots refers to hFFA2 pEC50 with
larger dots reflecting more potent compounds in this bioassay and conversely.
The second key objective in this program was to establish in vivo efficacy in an assay such as
OGTT that is both relevant to metabolic disease axis and has been validated using FFA2-/- mice
by other groups. After ensuring that the test compounds possess adequate oral exposure in vivo,
a number of lead structures were thus selected and screened in OGTT assay (Table 2). The
analogs tested encompassed structures from the foregoing scaffold both from the initial (1, 7 and
8) as well as advanced optimization rounds (10-13). As noted in Table 2, OGTT effect was
19

observed merely with the advanced leads (i.e. 9-13) and not with the early lead structures (1, 7
and 8). However, in a subsequent step to validate the mechanism of action underlying
compound activity, the POC lead 10 was observed to have similar efficacy in both wild-type and
FFA2 knockout mice indicating that the OGTT effect is not FFA2 mediated (Fig 2). Irrespective
of the underlying cause for this off-target OGTT effect, we decided to terminate this project as
we failed to demonstrate specific, FFA2-mediated in vivo efficacy. We shall disclose in due
course our findings with an alternative proline-based scaffold that can indeed serve as a bona
fide agonist reference for FFA2 receptor.37
4. Experimental section
4.1 Chemistry
All reactions were performed under a nitrogen atmosphere at rt using anhydrous grade solvents
except as otherwise noted. All reagents were used as supplied by the vendor except as noted.
Compd 2 was prepared as described elsewhere;4 3 was purchased from Enamine. NMR spectra
were recorded on Bruker ARX spectrometer (300 MHz) at rt. Chemical shifts are expressed in
parts per million relative to residual solvent as an internal reference. Peak multiplicities are
expressed as follows: singlet (s), doublet (d), triplet (t), quartet (q), heptuplet (h), multiplet (m),
and broad singlet (bs). HPLC purity analysis was carried out using one of the following methods.
Method A: LC/diode array detector coupled to single quadrupole mass spectrometry
(LC/UV/MS); column: Sunfire C18 3µm 3.0 × 50 mm; mobile phase A: 0.1% TFA in water
(v/v); mobile phase B: 0.1% TFA in CH3CN (v/v); gradient begins at 5% B, hold 0.2 min at 5%
B, increasing to 95% B over 6.0 min, hold at 95% B until 7.75 min; flow rate: 1.0 mL/min.
Method B: LC/diode array detector coupled to single quadrupole mass spectrometry
20

(LC/UV/MS); column: Sunfire C18 3µm 3.0 × 50 mm; mobile phase A: 0.1% TFA in water
(v/v); mobile phase B: 0.1% TFA in CH3CN (v/v); gradient begins at 5% B, hold 0.2 min at 5%
B, increasing to 95% B over 2.0 min, hold at 95% B until 3.75 min; flow rate: 1.0 mL/min.
Enantiomeric excess reported as %ee and determined using one of the following methods.
Method C (chiral): column: Chiralpak ID 5.0µm C18 4.6×250 mm; mobile phase: 0.1% TFA in
hexane/EtOH (95:5 v/v); flow rate: 1.0 mL/min; UV absorbance at 215 nm. Method D (chiral):
column: Chiralpak ID 5.0 µm C18 4.6×250 mm; mobile phase: 0.1% TFA in hexane/i-PrOH
(99:1 v/v); flow rate: 1.5 mL/min; UV absorbance at 254 nm. Method E (chiral): column:
Chiralpak IB 5.0µm C18 4.6×250 mm; mobile phase: 0.1% TFA in hexane/i-PrOH (98:2 v/v);
flow rate: 1.5 mL/min; UV absorbance at 254 nm. Method F: column: Chiralpak IC 5.0µm C18
4.6×250 mm; mobile phase: 0.1% TFA in hexane/i-PrOH (80:20 v/v); flow rate: 1.0 mL/min;
UV absorbance at 254 nm. PerkinElmer ChemDraw Ultra software (ver 15) was used for naming
the compounds described below.
4.2. (R)-3-Benzyl-4-((4-(2-chlorophenyl)thiazol-2-yl)(methyl)amino)-4-oxobutanoic acid (1).
Amide Bond Formation Step: To a solution of carboxylic acid 17 (6 g, 22.6 mmol) in CH3CN
(0.4 M) under Ar was added HATU (8.6 g, 22.6 mmol) and the mixture stirred for 10 min
whereupon aminothiazole 23 (5.1 g, 22.6 mmol) was added followed by DIEA (4.1 mL, 24.8
mmol) and the mixture stirred typically overnight. The volatiles were then removed in vacuo
and the residue thus obtained taken up in DCM (500 mL) and washed with saturated sodium
bicarbonate (500 mL). The organic layers were then washed with 1N HCl (500 mL), brine (500
mL), dried over MgSO4, filtered and concentrated to dryness and the crude thus obtained
subjected to silica gel flash chromatography (10% EtOAc in cyclohexane) to obtain pure tert-
butyl(R)-3-benzyl-4-((4-(2-chlorophenyl)thiazol-2-yl)(methyl)amino)-4-oxobutanoate (26) as
21

white solid (10 g, 94%). LCMS (Method B) m/z 471 (M + 1). HPLC purity > 98% (254 nm). 1H
NMR (300 MHz, CDCl3) 7.93 (dd, J = 1.8, 7.7 Hz, 1H), 7.53 (s, 1H), 7.44 (dd, J = 1.2, 7.7 Hz,
1H), 7.34 - 7.16 (m, 7H), 3.63 (s, 3H), 3.68 – 3.58 (m, 1H), 3.07 – 2.96 (m, 2H), 2.77 (dd, J =
7.2, 13.4 Hz, 1H), 2.47 (dd, J = 4.0, 17.0 Hz, 1H), 1.36 (s, 9H). Tert-butyl Ester Deprotection
Step: To a solution of tert-butyl ester 26 (10 g, 21.2 mmol) in 250 mL DCM was added TFA
(31.7 mL, 0.42 mmol) at rt and the mixture stirred typically for 2-3 h. The mixture was then
diluted with EtOAc (1 L) and the organic phase washed thoroughly with 10% sodium bisulfite
(500 mL) and brine (500 mL). The extracted organic layer was dried (Na2SO4), filtered and
concentrated to dryness in vacuo to obtain 1 as white solid (8 g, 92%). LCMS (Method A) m/z
415 (M + 1). HPLC purity > 99% (254 nm). Chiral LC (Method D) 98.8 %ee (tR(S) = 6.2 min,
1
tR(R) = 9.0 min). H NMR (CDCl3): 7.91 (dd, J = 1.7, 7.8 Hz, 1H), 7.54 (s, 1H), 7.45 (dd, J =
1.2, 7.8 Hz, 1H), 7.34 – 7.20 (m, 5H), 7.25 – 7.17 (m, 2H), 3.72 – 3.64 (m, 1H), 3.63 (s, 3H),
13
3.15 – 3.02 (m, 2H), 2.78 (dd, J = 7.0, 13.9 Hz, 1H), 2.58 (dd, J = 2.1, 18.3 Hz, 1H). C NMR
(CDCl3): 177.5, 174.3, 158.6, 145.5, 137.5, 133.5, 132.1, 131.3, 130.6, 129.0, 128.9, 128.7,
127.3, 127.0, 114.7, 41.3, 38.9, 36.4, 34.9. HRMS: calcd 415.0883, found 415.0876,
C21H20ClN2O3S. Elemental Analysis: calcd C, 60.79; H, 4.62; N, 6.75; found C, 60.11; H, 4.67;
N, 6.49.
4.3. (S)-3-Benzyl-4-((4-(2-chlorophenyl)thiazol-2-yl)(methyl)amino)-4-oxobutanoic acid (4). As
per compd 1 synthesis, amide coupling between acid 18 (77 mg, 0.29 mmol) and aminothiazole
23 (78 mg, 0.35 mmol) furnished tert-butyl(S)-3-benzyl-4-((4-(2-chlorophenyl)thiazol-2-
yl)(methyl)amino)-4-oxobutanoate (27) as a white solid (125 mg, 90%) after flash
chromatography (15% EtOAc in cyclohexane). LCMS (Method B) m/z 471 (M + 1). HPLC
1
purity > 98% (254 nm). H NMR (300 MHz, CDCl3) as above. 27 (125 mg, 0.26 mmol) after
22

TFA deprotection, as per compd 1, afforded 4 as white solid (85 mg, 79%). LCMS (Method A)
m/z 415 (M + 1). HPLC purity: 98% (254 nm). Chiral LC (Method D) 84.2 %ee (tR(S) = 6.2 min,
tR(R) = 9.0 min). Preparative chiral LC purification was performed to afford 4 with HPLC purity >
1
99% (254 nm) and chiral LC (Method D) > 99 %ee H NMR (CDCl3): 7.91 (dd, J = 1.7, 7.8
Hz, 1H), 7.54 (s, 1H), 7.45 (dd, J = 1.2, 7.8 Hz, 1H), 7.34 – 7.20 (m, 5H), 7.23 – 7.15 (m, 2H),
3.71 – 3.65 (m, 1H), 3.63 (s, 3H), 3.15 – 3.02 (m, 2H), 2.78 (dd, J = 7.0, 13.9 Hz, 1H), 2.58 (dd,
13
J = 2.1, 18.3 Hz, 1H). C-NMR (CDCl3): 177.5, 174.3, 158.6, 145.5, 137.5, 133.5, 132.1,
131.3, 130.6, 129.0, 128.9, 128.7, 127.3, 127.0, 114.7, 41.3, 38.9, 36.4, 34.9. HRMS: calcd
415.0883, found 415.0891, C21H20ClN2O3S.
4.4.(R)-4-((4-(2-Chlorophenyl)thiazol-2-yl)(methyl)amino)-3-(cyclopentylmethyl)-4-oxo butanoic
acid (5). As per compd 1 synthesis, amide coupling between acid 42 (50 mg, 0.20 mmol) and
aminothiazole 23 (44 mg, 0.20 mmol) furnished tert-butyl (R)-4-((4-(2-chlorophenyl)thiazol-2-
yl)(methyl)amino)-3-(cyclopentylmethyl)-4-oxobutanoate (54) as white solid (56 mg, 62%) after
flash chromatography (2-20% EtOAc in cyclohexane). LCMS (Method B) m/z 464 (M + 1).
HPLC purity > 98% (254 nm). 54 (58 mg, 0.13 mmol) after TFA deprotection, as per compd 1,
afforded 5 as white solid (20 mg, 39%). LCMS (Method A) m/z 407 (M + 1). HPLC purity >
1
99% (254 nm). Chiral LC (Method D) 99.4 %ee (tR(R) = 5.2 min, tR(s) = 10.2 min). H NMR
(CDCl3): 7.96 (dd, J = 1.5, 7.7 Hz, 1H), 7.54 (s, 1H), 7.46 (dd, J = 1.3, 7.7 Hz, 1H), 7.33 (dt, J
= 1.3, 7.7 Hz, 1H), 7.25 (dt, J = 1.5, 7.7 Hz, 1H), 3.92 (s, 3H), 3.46 - 3.42 (m, 1H), 3.04 (dd, J =
9.9, 17.4 Hz, 1H), 2.63 (dd, J = 4.0, 7.5 Hz, 1H), 1.83 – 1.68 (m, 4H), 1.64 – 1.46 (m, 6H), 1.18
13
– 1.04 (m, 2 H). C NMR (CD3OD): 177.6, 178.6, 160.3, 146.8, 134.9, 133.1, 132.5, 131.5,
130.0, 125.0, 115.5, 40.1, 39.7, 38.9, 37.8, 35.8, 34.2, 33.7, 26.1, 26.0. HRMS: calcd 407.1196,
found 407.1191, C20H24ClN2O3S.
23

4.5.(R)-4-((4-(2-Chlorophenyl)thiazol-2-yl)(methyl)amino)-4-oxo-3-((tetrahydro-2H-pyran-4-yl)
methyl)butanoic acid (6). As per compd 1 synthesis, amide coupling between acid 43 (50 mg,
0.18 mmol) and 23 (45 mg, 0.20 mmol) funrished tert-butyl (R)-4-((4-(2-chlorophenyl)thiazol-2-
yl)(methyl)amino)-4-oxo-3-((tetrahydro-2H-pyran-4-yl)methyl)butanoate (55) as white solid (77
mg, 89%) after flash chromatography (20% EtOAc in cyclohexane). LCMS (Method A) m/z 479
(M + 1). HPLC purity > 98% (254 nm). 55 (77 mg, 0.16 mmol) after TFA deprotection, as per
compd 1, afforded 6 as white solid (26 mg, 38%). LCMS (Method A) m/z 423 (M + 1). HPLC
1
purity > 99% (254 nm). Chiral LC (Method E) 99.3 %ee (tR(R) = 6.4 min, tR(s) = 9.3 min). H-
NMR (CDCl3): 7.95 ((dd, J = 1.5, 7.5 Hz, 1H), 7.54 (s, 1H), 7.46 (dd, J = 1.2, 7.8 Hz, 1H), 7.33
(dt, J = 1.2, 7.5 Hz, 1H), 7.28 – 7.23 (m, 1H), 3.99 – 3.92 (m, 2H), 3.90 (s, 3H), 3.54 – 3.47 (m,
1H), 3.39 – 3.32 (m, 2H), 3.01 (dd, J = 9.5, 17.4 Hz, 1H), 2.59 (dd, J = 4.0, 7.3 Hz, 1H), 1.71 –
1.48 (m, 5H), 1.39 – 1.20 (m, 3H). 13C-NMR (CDCl3): 176.4, 174.9, 158.8, 145.6, 133.5, 132.1,
131.3, 130.7, 128.8, 127.0, 114.8, 67.8, 67.7, 39.4, 38.5, 36.2, 36.0, 35.1, 33.6, 32.9, 32.6, 32.5,
32.2. HRMS: calcd 423.1145, found 423.1151, C20H24ClN2O4S.
4.6.(R)-3-Benzyl-4-((4-(2-chlorophenyl)thiazol-2-yl)(cyclopropyl)amino)-4-oxobutanoic
acid
(7). Using acid 17 (200 mg, 0.76 mmol) and aminothiazole 24 (190 mg, 0.76 mmol) amide
bond formation step was carried out as per compd 1 except at 60 °C to obtain tert-butyl (R)-3-
benzyl-4-((4-(2-chlorophenyl)thiazol-2-yl)(cyclopropyl)amino)-4-oxo butanoate (28) as yellow
oil (232 mg, 62%) after flash chromatography (10% EtOAc in cyclohexane). LCMS (Method B)
m/z 497 (M + 1). HPLC purity > 98% (254 nm). 28 (170 mg, 0.34 mmol) after TFA
derprotection, as per compd 1, afforded 7 as white solid (120 mg, 71%). LCMS (Method A) m/z
441 (M + 1). HPLC purity: 94% (254 nm). Chiral LC (Method E) 95.4 %ee (tR(R) = 4.5 min, tR(S)
= 7.1 min). 1H NMR (CDCl3): 7.91 (dd, J = 1.8, 7.8 Hz, 1H), 7.54 (s, 1H), 7.46 (dd, J = 1.3, 7.8
24

Hz, 1H), 7.34 – 7.14 (m, 7H), 4.19 – 4.11 (m, 1H), 3.09 (dd, J = 6.2, 13.2 Hz, 1H), 3.03 – 2.91
(m, 2H), 2.72 (dd, J = 8.2, 13.2 Hz, 1H), 2.54 – 2.47 (m, 1H), 1.15 – 1.10 (m, 3H), 0.88 – 0.80
(m, 1H). HRMS: calcd 441.1040, found 441.1048, C23H22ClN2O3S.
4.7.(R)-3-Benzyl-4-(cyclopropyl(4-(2,5-dichlorophenyl)thiazol-2-yl)amino)-4-oxobutanoic acid
(8). Using acid 17 (100 mg, 0.38 mmol) and aminothiazole 25 (108 mg, 0.38 mmol) amide bond
formation step was carried out as per compd 1 except at 60 °C to obtain tert-butyl (R)-3-benzyl-
4-(cyclopropyl(4-(2,5-dichlorophenyl)thiazol-2-yl)amino)-4-oxobutanoate (29) as white solid
(152 mg, 76%) after flash chromatography (10% EtOAc in cyclohexane). LCMS (Method B) m/z
1
531 (M + 1). HPLC purity > 98% (254 nm). H NMR (CDCl3): 7.91 (d, J = 2.6 Hz, 1H), 7.55
(s, 1H), 7.31 (d, J = 8.6 Hz, 1H), 7.27 – 7.08 (m, 6H), 4.18 – 4.11 (m, 1H), 2.88 – 2.61 (m, 3H),
2.36 (dd, J = 16.7, 4.9 Hz, 1H), 1.31 (s, 9H), 1.25 – 1.13 (m, 4H), 0.78 – 0.69 (m, 1H). 29 (152
mg, 0.29 mmol) after TFA deprotection, as per compd 1, afforded 8 as white solid (82 mg, 60%).
LCMS (Method A) m/z 475 (M + 1). HPLC purity > 99% (254 nm). 1H NMR (CDCl3): 7.96 (d,
J = 2.3 Hz, 1H), 7.63 (s, 1H), 7.38 (d, J = 8.5 Hz, 1H), 7.28 – 7.18 (m, 4H), 7.18 – 7.10 (m, 2H),
4.21 – 4.15 (m, 1H), 3.14 – 2.86 (m, 3H), 2.73 (dd, J = 8.5, 13.6 Hz, 1H), 2.53 (dd, J = 4.0, 17.4
Hz, 1H), 1.25 – 1.19 (m, 3H), 0.88 – 0.78 (m, 1H). HRMS: calcd 475.0650, found 475.0649,
C23H21Cl2N2O3S.
4.8.(R)-4-((4-(5-Chloro-2-(6-methoxypyridin-3-yl)phenyl)thiazol-2-yl)(cyclopropyl)amino)-3-
(furan-2-ylmethyl)-4-oxobutanoic acid (9). Using acid 45 (100 mg, 0.39 mmol) and
aminothiazole 51 (141 mg, 0.44 mmol) amide bond formation step carried out as per compd 1
except at 60 °C to obtain tert-butyl (R)-4-((4-(5-chloro-2-(6-methoxypyridin-3-yl)phenyl)thiazol-
2-yl)(cyclopropyl)amino)-3-(furan-2-ylmethyl)-4-oxobutanoate) (56) as a white solid (62 mg,
37%) after flash chromatography (20% EtOAc in cyclohexane). LCMS (Method B) m/z 595 (M
25

+ 1). HPLC purity > 98% (254 nm). 56 (85 mg, 0.14 mmol, two batches combined) after TFA
deprotection, as described for 1, afforded 9 as white solid (60 mg, 78%). LCMS (Method A) m/z
538 (M + 1). HPLC purity 95% (254 nm). 1H NMR (CDCl3): 8.02 (d, J = 2.3 Hz, 1H), 7.75 (d,
J = 2.3 Hz, 1H), 7.42 – 7.36 (m, 2H), 7.30 – 7.24 (m, 2H), 6.69 (d, J = 8.5 Hz, 1H), 6.59 (s, 1H),
6.28 – 6.22 (m, 1H), 6.03 (d, J = 3.0 Hz, 1H), 4.23 – 4.16 (m, 1H), 3.93 (s, 3H), 3.02 (dd, J =
5.8, 14.9 Hz, 1H), 2.95 – 2.88 (m, 2H), 2.79 (dd, J = 8.2, 14.9 Hz, 1H), 2.55 (dd, J = 4.3, 17.1
13
Hz, 1H), 1.13 – 1.01 (m, 2H), 0.95 – 0.88 (m, 1H), 0.56 – 0.48 (m, 1H). C-NMR (CD3OD):
178.3, 175.1, 164.7, 153.2, 147.5, 143.0, 141.2, 137.5, 137.1, 134.6, 133.1, 131.4, 131.3, 129.4,
111.5, 111.2, 108.3, 54.2, 40.6, 37.3, 31.7, 30.9, 7.9. HRMS: calcd 538.1204, found 538.1193,
C27H25ClN3O5S.
4.9.(R)-3-(Cyclopentylmethyl)-4-(cyclopropyl(4-(2-(6-methoxypyridin-3-yl)phenyl)thiazol-2-yl)
amino)-4-oxobutanoic acid (10) Using acid 42 (11.5 g, 45 mmol) and aminothiazole 50 (15.95
g, 49 mmol) amide bond formation step was carried out as per compd 1 except at 60 °C to
obtain tert-butyl (R)-3-(cyclopentylmethyl)-4-(cyclopropyl(4-(2-(6-methoxypyridin-3-yl)phenyl)
thiazol-2-yl)amino)-4-oxobutanoate (57) as a beige solid (14.2 g, 56%) after flash
chromatography (10% EtOAc in cyclohexane). LCMS (Method B) m/z 562 (M + 1). HPLC
purity > 98% (254 nm). 1H NMR (300 MHz, CDCl3): 8.08 (d, J = 1.8 Hz, 1H), 7.71 (d, J = 6.9
Hz, 1H), 7.40 – 7.38 (m, 4H), 6.65 (d, J = 6.3 Hz, 1H), 6.59 (s, 1H), 3.92 (s, 3H), 3.92 - 3. 87 (m,
1H), 3.03 – 2.96 (m, 1H), 2.74 – 2.70 (m, 1H), 2.53 – 2.44 (m, 1H), 1.77 – 1.41 (m, 10H), 1.38
(s, 9H), 1.19 -1.07 (m, 4H), 0.58 – 0.49 (m, 1H). Elemental analysis: calcd C, 68.42; H, 7.00; N,
7.48; found C, 68.27; H, 7.11; N, 7.25. 57 (14.3 g, 25 mmol) after TFA deprotection, as
described for 1, afforded 10 as white solid (10.9 g, 86%). LCMS (Method A) m/z 506 (M + 1).
HPLC purity 96% (254 nm). 1H NMR (CDCl3): 8.08 (d, J = 2.3 Hz, 1H), 7.74 – 7.71 (m, 1H),
26

7.45 – 7.36 (m, 4H), 6.69 (d, J = 8.6 Hz, 1H), 6.63 (s, 1H), 3.94 (s, 3H), 3.88 – 3.81 (m, 1H),
3.05-2.99 (m, 1H), 2.91 (dd, J = 9.4, 17.7 Hz, 1H), 2.60 (dd, J = 4.2, 17.5 Hz, 1H), 1.87 – 1.75
(m, 3H), 11.64 – 1.45 (m, 7H), 1.14 – 0.95 (m, 5H), 0.60 – 0.53 (m, 1H). 13C NMR (CD3OD):
179.5, 179.2, 175.7, 164.5, 147.5, 141.4, 138.5, 132.5, 131.7, 131.6, 129.7, 128.9, 111.0, 106.3,
54.3, 42.1, 39.9, 39.6, 39.1, 36.8, 34.3, 33.7, 33.3, 31.7, 30.9, 26.0, 25.9, 7.8. HRMS: calcd
506.2113, found 506.2106, C28H32N3O4S. Elemental analysis: calcd C, 66.51; H, 6.18; N, 8.31;
found C, 66.15; H, 6.12; N, 7.95.
4.10.(R)-4-((4-(2-(6-(1H-Pyrazol-1-yl)pyridin-3-yl)phenyl)thiazol-2-yl)(cyclopropyl)amino)-3-
(cyclopentylmethyl)-4-oxobutanoic acid (11). Using acid 42 (0.51 g, 2 mmol) and aminothiazole
53 (200 mg, 0.56 mmol) amide bond formation step was carried out as per compd 1 except at 60
°C to obtain tert-butyl (R)-4-((4-(2-(6-(1H-pyrazol-1-yl)pyridin-3-yl)phenyl)thiazol-2-
yl)(cyclopropyl)amino)-3-(cyclopentylmethyl)-4-oxobutanoate (58) as white solid (188 mg,
57%) after flash chromatography (10% EtOAc in cyclohexane). LCMS (Method B) m/z 598 (M
+ 1). HPLC purity > 98% (254 nm). 58 (188 mg, 0.31 mmol) after TFA deprotection, as
described for 1, afforded 11 as white solid (46 mg, 27%). LCMS (Method A) m/z 542 (M + 1).
1
HPLC purity 95% (254 nm). H NMR (CDCl3): 8.550 (d, J = 1.9 Hz, 1H), 8.32 (d, J = 1.7 Hz,
1H), 7.84 (d, J = 8.4 Hz, 1H), 7.76 – 7.72 (m, 2H), 7.62 (dd, J = 2.0, 8.6 Hz, 1H), 7.49 – 7.39 (m,
3H), 6.72 (s, 1H), 6.49 – 6.45 (m, 1H), 3.88 – 3.79 (m, 1H), 2.86 – 2.76 (m, 2H), 2.60 (dd, J =
4.0, 17.4 Hz, 1H), 1.83 – 1.69 (m, 3H), 1.63 – 1.45 (m, 6H), 1.13 – 1.02 (m, 2H), 0.97 – 0.92 (m,
2H), 0.84 – 0.76 (m, 1H), 0.62 – 0.55 (m, 1H). 13C NMR (CD3OD): 179.5, 175.9, 162.0, 151.2,
149.3, 143.3, 140.9, 137.9, 137.4, 135.9, 131.7, 131.6, 129.8, 129.4, 128.3, 112.8, 109.0, 40.1,
39.9, 38.8, 37.0, 34.2, 33.3, 31.1, 26.0, 25.9, 7.8. HRMS: calcd 542.2226, found 542.2231,
C30H32N5O3S.
27

4.11.(R)-3-(Cyclopropyl(4-(2-(6-methoxypyridin-3-yl)phenyl)thiazol-2-yl)carbamoyl)-5-methyl
hexanoic acid (12). Using acid 44 (640 mg, 2.78 mmol) and aminothiazole 50 (990 mg, 3.06
mmol) amide bond formation step was carried out as per compd 1 except at 60 °C to obtain tert-
butyl
(R)-3-(cyclopropyl(4-(2-(6-methoxypyridin-3-yl)phenyl)thiazol-2-yl)carbamoyl)-5-
methylhexanoate (59) as white solid (751 mg, 1.40 mmol, 50%) after flash chromatography
(10% EtOAc in cyclohexane). LCMS (Method B) m/z 536 (M + 1). HPLC purity > 98% (254
nm). 59 (751 mg, 1.40 mmol) after TFA deprotection, as described for 1, afforded 12 as white
solid (496 mg, 1.03 mmol, 74%). LCMS (Method A) m/z 480 (M + 1). HPLC purity 99% (254
nm). 1H NMR (CDCl3): 8.09 (d, J = 2.3 Hz, 1H), 7.74 – 7.71 (m, 1H), 7.44 – 7.40 (m, 3H), 7.36
– 7.33 (m, 1H), 6.69 (d, J = 8.5 Hz, 1H), 6.63 (s, 1H), 3.94 (s, 3H), 3.92 - 3.82 (m, 1H), 3.04 –
2.98 (m, 1H), 2.89 (dd, J = 9.8, 17.2 Hz, 1H), 2.56 (dd, J = 4.1, 17.2 Hz, 1H), 1.67 – 1.56 (m,
1H), 1.47 – 1.39 (m, 2H), 1.09 – 0.98 (m, 3H), 0.91 (d, J = 17.4 Hz, 6H), 0.60 – 0.54 (m, 1H).
13C NMR (CD3OD): 179.7, 175.9, 164.6, 162.3, 147.5, 141.3, 138.6, 135.8, 132.5, 131.7, 131.6,
129.7, 128.9, 111.0, 54.2, 42.6, 38.9, 36.8, 31.0, 26.8, 23.8, 22.1, 7.8. HRMS: calcd 480.1957,
found 480.1953, C26H30N3O4S.
4.12.(R)-3-((4-(2-(6-(1H-Pyrazol-1-yl)pyridin-3-yl)phenyl)thiazol-2-yl)(cyclopropyl)carbamoyl)-
5-methylhexanoic acid (13). Using acid 44 (640 mg, 2.78 mmol) and aminothiazole 53 (600 mg,
1.67 mmol) amide bond formation step was carried out was carried out as per compd 1 except at
60 °C to obtain tert-butyl (R)-3-((4-(2-(6-(1H-pyrazol-1-yl)pyridin-3-yl)phenyl)thiazol-2-
yl)(cyclopropyl)carbamoyl)-5-methylhexanoate (60) as white solid (288 mg, 30%) after flash
chromatography (10% EtOAc in cyclohexane). LCMS (Method A) m/z 572 (M + 1). HPLC
1
purity > 98% (254 nm). H NMR (CDCl3): 8.54 (d, J = 2.5 Hz, 1H), 8.30 (d, J = 2.1 Hz, 1H),
7.92 (d, J = 8.4 Hz, 1H), 7.78 – 7.68 (m, 3H), 7.48 – 7.38 (m, 3H), 6.67 (s, 1H), 6.47 (t, J = 2.1
28

Hz, 1H), 3.91 – 3.85 (m, 1H), 2.99 – 2.95 (m, 1H), 2.73 (dd, J = 9.3, 16.6 Hz, 1H), 2.43 (dd, J =
4.9, 16.6 Hz, 1H), 1.63 – 1.53 (m, 2H), 1.39 (s, 9H), 1.10 - 0.94 (m, 3H), 0.89 (d, J = 6.56 Hz,
6H), 0.58 – 0.51 (m, 2H). 60 (288 mg, 0.50 mmol) after TFA deprotection, as described for 1,
afforded 13 as light yellow solid (242 mg, 93%). LCMS (Method A) m/z 516 (M + 1). HPLC
purity 97% (254 nm). 1H NMR (CDCl3): 8.50 (d, J = 2.4 Hz, 1H), 8.37 (d, J = 2.1 Hz, 1H), 7.79
– 7.76 (m, 2H), 7.73 – 7.68 (m, 1H), 7.53 – 7.42 (m, 4H), 6.82 (s, 1H), 6.49 (dd, J = 2.1, 2.4 Hz,
1H), 3.92 - 3.82 (m, 1H), 2.87 – 2.76 (m, 2H), 2.55 (dd, J = 3.7, 16.7 Hz, 1H), 1.67 – 1.58 (m,
13
1H), 1.50 – 1.38 (m, 2H), 0.97 – 0.76 (m, 9H), 0.62 – 0.55 (m, 1H). C NMR (CD3OD): 179.7,
175.6, 151.2, 149.3, 143.3, 140.9, 138.0, 137.5, 135.9, 131.7, 131.6, 129.8, 129.5, 128.4, 112.8,
109.0, 42.6, 38.8, 36.6, 31.0, 26.7, 23.8, 22.1, 7.7. HRMS: calcd 516.2069, found 516.2077,
C28H30N5O3S.
4.13.(R)-2-benzyl-4-(tert-butoxy)-4-oxobutanoic acid (17).
The title compd was prepared
following Scheme 1 (steps a-c). Step a: To a stirred solution of 4-(tert-butyl)-1-methyl-2-
(dimethoxyphosphoryl)succinate (14, Scheme 1) (77 g, 0.26 mol) in THF (1 L), was added n-
BuLi (1.6M in hexanes, 179 mL, 0.29 mol) dropwise (45 min) at 5-15°C (ice bath). The orange
solution thus obtained was stirred at 5-15°C for 90 min. whereupon a THF solution (200 mL) of
benzaldehyde (27.5 g, 0.26 mol) added dropwise (30 min) at 5-10°C. The viscous solution was
then allowed to warm gradually to rt and stirred overnight.
Thereafter the reaction was
quenched with ice/water (300 mL), the volatiles removed in vacuo and the resultant residue
taken up in saturated aq NaHCO3 (400 mL) and then extracted with EtOAc (2 x 400 mL). The
combined organic layers were dried over Na2SO4 and evaporated in vacuo to obtain a brown oil
(85 g) that was purified by silica gel flash chromatography using an increasing gradient from 1-
10% EtOAc /petroleum ether to obtain 4-(tert-butyl) 1-methyl (E)-2-benzylidenesuccinate 15 (31
29