Discovery of structurally novel, potent and orally efficacious GPR119 agonists

Article abstract of DOI:10.1016/j.bmcl.2014.03.023

Screening hit 5 was identified in a biochemical screen for GPR119 agonists. Compound 5 was structurally novel, displayed modest biochemical activity and no oral exposure, but was structurally distinct from typical GPR119 agonist scaffolds. Systematic opti

Full text of DOI:10.1016/j.bmcl.2014.03.023

Bioorganic & Medicinal Chemistry Letters 24 (2014) 2383–2387
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of structurally novel, potent and orally efficacious GPR119
agonists
⇑
Phil Alper , Mihai Azimioara, Christopher Cow, Daniel Mutnick, Victor Nikulin, Pierre-Yves Michellys,
Zhiliang Wang, Esther Reding, Michael Paliotti, Jing Li, Dingjiu Bao, Jocelyn Zoll, Young Kim,
Matthew Zimmerman, Todd Groessel, Tove Tuntland, Sean B. Joseph, Peter McNamara,
⇑
H. Martin Seidel, Robert Epple
Genomics Institute of the Novartis Research Foundation, 10675 John Jay Hopkins Drive, San Diego, CA 92121, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Screening hit 5 was identified in a biochemical screen for GPR119 agonists. Compound 5 was structurally
novel, displayed modest biochemical activity and no oral exposure, but was structurally distinct from
typical GPR119 agonist scaffolds. Systematic optimization led to compound 36 with significantly
improved in vitro activity and oral exposure, to elevate GLP1 acutely in an in vivo mouse model at a dose
of 10 mg/kg.
Received 10 January 2014
Revised 6 March 2014
Accepted 10 March 2014
Available online 27 March 2014
Ó 2014 Elsevier Ltd. All rights reserved.
Keywords:
GPCR agonists
GPR119
Pyrazolopyrimidines
Type 2 diabetes
The need to address the worldwide pandemic of Type 2 diabetes
(T2D) can hardly be overstated.1 The current treatments for this
chronic disorder either fail to be effective over time or present lim-
iting safety risks, emphasizing a strong medical need for novel
matter that did not share structural similarity with known
agonists.
The screen resulted in the identification of multiple (>20) novel
hit scaffolds, which were followed up by routine hit-to-lead medic-
inal chemistry. During this hit-to-lead process we encountered
several recurring challenges across most scaffolds, namely (a) gen-
erally narrow SAR, (b) insurmountable ADMET issues due to the
narrow SAR and the overall hydrophobic nature of the scaffolds,
(c) varying activity across species and (d) poor correlation between
the primary, cellular cAMP assays and the functional hormone
secretion assays in the relevant cell lines.
A hit scaffold identified from a purchased library that merited
further investigation due to its departure from the common phar-
macophore is represented by 5 in Figure 2. The hit displayed mod-
erate activity on both the human and mouse GPR119 receptor and
moderate ability to induce GLP-1 and insulin secretion in the GLU-
Tag and HIT-T15 cell lines respectively. As expected from the fairly
hydrophobic substituents around the pyrazolopyrimidine core, the
high clogP (ꢀ6.5) of 5 led to poor solubility and fairly high extrac-
tion ratios (>0.8) in microsomal stability assays. In addition, the
plasma stability of early analogues was low, postulated to be a re-
sult of either compound aggregation, ester hydrolysis or both.
A flexible, generalized synthetic approach that allowed us to
efficiently explore both pyrazolopyrimidine substituents is shown
in Scheme 1. A methyl ketone 6 was converted to the enamine 7 by
approaches.2,3 The G s-protein-coupled receptor GPR119, predom-
a
inantly expressed in pancreatic b-cells and enteroendocrine L-cells
in the gut, has recently emerged as a promising drug target for T2D
due to its role in regulating glucose homeostasis via modulation of
incretin and insulin secretion.4–7
Since the ground-breaking publications disclosing GPR119 ago-
nists such as 1 and 2 by Arena and Prosidion research groups
respectively,8,9 over 100 patent applications have been filed that
generally revolve around a common pharmacophore shown in
Figure 1.10,11 To date, several organizations have progressed
GPR119 agonists into the clinic. Previously disclosed compounds
3 and 412 are depicted in Figure 1, and structurally also fall under
the common GPR119 pharmacophore. Our own efforts towards
identifying novel agonists of GPR119 that display this ‘classical’
pharmacophore will be disclosed elsewhere. In addition to these
efforts, we undertook a high-throughput screening (HTS) campaign
on a ꢀ2 million compound collection, searching for new chemical
⇑
Corresponding authors. Tel.: +1 858 812 1536; fax: +1 858 812 1648 (P.A.); tel.:
+1 858 812 1720; fax: +1 858 812 1648 (R.E.).
E-mail addresses: palper@gnf.org (P. Alper), repple@gnf.org (R. Epple).
http://dx.doi.org/10.1016/j.bmcl.2014.03.023
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

2384
P. Alper et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2383–2387
R1
N
R2
Linker
R1 = SO2Me, Heterocycle
R2 = CO2R, Heterocycle
N
O
O
O
N
N
O
O
N
F
N
N
N
O
S
O
N
NH
NO2
N
N
1 (AR231453)
hGPR119: 0.003 mM (100%)
2 (PSN632408)
hGPR119: 0.63 mM (85%)
S
O
N
N
O
O
S
N
N
N
N
O
N
N
O
N
N
N
3 (GSK1292263; Phase II)
hGPR119: 0.009 mM (100%)
4 (MBX-2982; Phase II)
hGPR119: 0.005 mM (100%)
Figure 1. Common GPR119 pharmacophore and disclosed GPR119 agonists 1–4, and common pharmacophore of most filed patent applications.13
group to either a phenyl (16) or a methyl group (17) resulted in sig-
nificant loss of activity, as did removal of the chloride in the 3-phe-
nyl substituent (18). Efficacies in the cAMP assay were slightly
lower (50–70%) for this series when compared to typical GPR119
agonist scaffolds (represented by 1 = 100%). Based on prior experi-
ence in the characterization of several GPR119 agonist scaffolds,
we observed that, generally, efficacy >80% in the cAMP assay re-
sulted in robust effects in the downstream functional in vitro and
in vivo assays. Therefore, we focused on compounds with higher
efficacy for further progression.
Another concern was the potential liability of the secondary es-
ter functionality. In fact, the ester group could be chemically re-
moved by spontaneous decarboxylation after saponification of hit
5 to afford 19. The complete lack of activity of this compound indi-
cated that the ester was an important pharmacophore feature. One
way to significantly stabilize the ester was to quaternize the ben-
zylic carbon and make it sterically less accessible. Installing a
methyl group led to 20, which maintained potency but increased
efficacy to almost 80%. After racemate 20 was chromatographically
separated to afford 21 and 22, we observed that GPR119 activity
mainly resided with the R enantiomer (determined crystallograph-
F
O
1
O
hGPR119 :0.20 µM (76%)
2
N
7
mGPR119: 0.076 µM (93%)
GLP-1 - GLUTag: 1.4 µM (105%)
Insulin - HIT-T15: 2.9 µM (68%)
N
3
6
Cl
5
N
4
Cl
5
Figure 2. Structure of novel GPR119 agonist screening hit.14
H2N
R4
O
O
DMF-DMA
HCl, EtOH
N
R1
R2
R1
R2
+
NH
NMe2
R3
6
R3
7
8
R3
R2
R1
N
N
N
R4
Microwave
heating
9
Scheme 1. Representative scheme for the synthesis of analogs of HTS hit 5.
ically) 21, while the S enantiomer 22 had diminished (>10 lM)
treatment with dimethylformamide dimethyl acetal at elevated
temperature. The synthesis of 6 could be tailored to afford
mono-, di- or trisubstituted ketones where various functional
groups could be incorporated either before or after the cyclization
depending on the group. Next, this intermediate was condensed
with a 3-aminopyrrole 8 using an acid in alcoholic solvent to afford
the 3,7-disubstituted pyrazolopyrimidine 9. The R4 substituent was
installed by either carrying through a halogen at this position fol-
lowed by palladium coupling methodology or synthesizing 8 from
a substituted acetonitrile by several literature-known methods.
A summary of SAR on the human GPR119 receptor (cAMP assay)
for 5 and analogues 10–42 is depicted in Table 1.
Substitution of the chlorine atom on the biphenyl unit for
‘essential’ elements seen in the ‘common pharmacophore’ such
as a sulfone (10), a Boc-substituted piperidine (11) or an oxadiaz-
ole (12) resulted in a loss of activity on the GPR119 receptor. Sim-
ilarly, adding the sulfone or a piperidine on the other side of the
molecule (13, 14) also resulted in inferior activity, indicating that
this series does not follow the classical GPR119 pharmacophore.
Removal of the halogens on the phenyl ring of the 7-substituent
gave a roughly equipotent analogue 15. Changing the benzylic
activity. Increasing the size of the quaternizing substituent, exem-
plified by a propyl group (23), resulted in a loss of activity. We
were encouraged by the observation that the ester group of the
quaternized analogue 21 was highly inert to chemical or metabolic
activation, as well as to plasma hydrolases (>99% parent remaining
after 3 h incubation in human and rodent plasma).
With these important findings in hand, we turned our attention
back to further optimization of the aryl substituents around the
pyrazolopyrimidine core. Probing the substitution pattern around
the benzylic group off position 7 of the core heterocycle, we no-
ticed that meta substitution was usually better tolerated than ortho
or para substitution (cf. analogues 25 vs 24 and 26). Substitution of
the benzyl group with a variety of aliphatic and heterocyclic sub-
stituents resulted in significant activity loss (isopropyl- (27), tetra-
hydopyranyl- (28), 3-pyridine- (29) and thiazolo- (30) analogues).
We next turned our attention to the chlorophenyl substituent and
determined that moving the chlorine of the phenyl in position 4 of
the core heterocycle to the 3- (31) or 2- (32) position significantly
reduced activity, as did the elongation to a 4-chlorobenzyl substi-
tuent (33). Changing the 4-chloro-phenyl to a 4-chloro-pyridyl
substitution (34) was fairly well tolerated. Changing the chlorine

P. Alper et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2383–2387
2385
Table 1
SAR of substituents on 3
R3
R2
N
R1
N
R4
N
Entry
R1
R2
R3
R4
Chirality
Racemic
hGPR119
EC50 (nM)
200
Efficacy (%)
F
5
CO2Et
H
4-Cl-Ph
76
Cl
10
11
CO2Et
CO2Et
H
(2-Cl-6-F-Bn)
4-SO2Me-Ph
Racemic
Racemic
1300
560
68
62
BocN
Me
Bn
O
N
12
13
14
CO2Et
CO2Et
CO2Et
Me
Me
Me
Bn
Racemic
Racemic
Racemic
880
43
5
N
4-SO2Me-Bn
4-Cl-Ph
4-Cl-Ph
>10,000
340
MsN
50
15
16
17
18
19
20
21
22
23
24
25
26
CO2Et
CO2Et
CO2Et
CO2Et
H
CO2Et
CO2Et
CO2Et
CO2Et
CO2Et
CO2Et
CO2Et
H
H
H
H
Bn
Ph
Me
Bn
4-Cl-Ph
4-Cl-Ph
4-Cl-Ph
Ph
4-Cl-Ph
4-Cl-Ph
4-Cl-Ph
4-Cl-Ph
4-Cl-Ph
4-Cl-Ph
4-Cl-Ph
4-Cl-Ph
Racemic
Racemic
Racemic
Racemic
Racemic
Racemic
R
16
290
1200
190
>10,000
20
17
>10,000
190
140
40
66
63
32
60
15
79
83
15
68
66
76
73
H
(2-Cl-6-F-Bn)
Bn
Bn
Bn
Bn
4-Me-Bn
3-Me-Bn
2-Me-Bn
Me
Me
Me
Pr
Me
Me
Me
S
Racemic
Racemic
Racemic
Racemic
110
27
28
CO2Et
CO2Et
Me
Me
4-Cl-Ph
4-Cl-Ph
Racemic
Racemic
170
63
56
66
O
N
29
CO2Et
Me
4-Cl-Ph
Racemic
73
40
68
S
30
CO2Et
Me
4-Cl-Ph
Racemic
63
N
Bn
Bn
31
32
33
CO2Et
CO2Et
CO2Et
Me
Me
Me
3-Cl-Ph
2-Cl-Ph
4-Cl-Bn
Racemic
Racemic
Racemic
130
180
>10,000
66
54
10
Bn
N
34
CO2Et
Me
Bn
Racemic
58
69
Cl
35
36
37
CO2Et
CO2Et
CO2Et
Me
Me
Me
Bn
Bn
Bn
4-CF3-Ph
4-CF3-Ph
4-CF3-Ph
Racemic
R
S
11
10
>300
88
80
22
O
N
O
38
Me
Bn
4-Cl-Ph
Racemic
4
81
O
O
39
40
O
Me
Me
Bn
Bn
4-CF3-Ph
4-CF3-Ph
R
R
1
85
65
NHAc
86
O
41
42
Me
Me
Bn
Bn
4-CF3-Ph
4-CF3-Ph
Racemic
Racemic
164
243
50
57
N
SO2Me
to more hydrophilic substituents was not tolerated, but the more
lipophilic CF3-substituent resulted in the racemic analogue 35,
displaying robust potency and efficacy. As observed before, all of
the activity resided with the R isomer 36 and not in the S
isomer 37.
Only certain, fairly conservative changes to the ester were toler-
ated, such as the methoxyethyl ester 38, or the corresponding
methylated amide 39, both leading to a considerable boost of
activity. Interestingly, ‘reverse’ amides such as 40 were also some-
what tolerated, albeit with lower efficacy. Other ester isosteres
such as oxazole 41 or substituting the ester group with other func-
tionalities such as sulfone 42 also gave significantly lower effica-
cies. Many other functionalities were tried, but all without success.
To further probe the scope of this scaffold, we explored several
alternative bicycles to the pyrazolopyrimidine core. Some interest-
ing findings are summarized in Table 2. The pyrazolotriazine 43 ni-
cely retained potency and efficacy, with the difference that the
extra nitrogen in the triazine ring made the core much more basic

2386
P. Alper et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2383–2387
Table 2
SAR of scaffold core
Entry
Structure
Chirality
hGPR119
EC50 (nM)
Efficacy (%)
88
CO2Et
N
N
N
43
44
R
20
62
N
CF3
Cl
CO2Et
N
N
Racemic
73
70
59
N
CO2Et
N
N
45
46
Racemic
Racemic
53
S
Cl
CO2Et
N
N
N
N
1100
OCF3
Table 3
In vitro GLP1 and Insulin secretion, oral exposure and in vivo efficacy of selected compounds
Entry
hGPR119 (cAMP)
mGPR119 (cAMP)
GLP1 secretion (GLUTag)
Insulin secretion
(HIT-T15)
AUCdn
(balb/c) h lM
Cmax
(
l
M) Active GLP-1 elevation
(C57BL/6) (% elevation/dose)
EC50
(l
M) Eff (%) EC50
(l
M) Eff (%) EC50
(l
M)
Eff (%)
EC50
(lM)
Eff (%)
21
36
39
43
0.017
0.010
0.001
0.020
83
80
85
88
0.076
75
82
71
81
0.51
0.33
0.03
0.23
110
130
49
2.2
72
76
74
0.28
2.07
0.50
0.49
1.1
1.7
2.3
2.5
No elevation up to 100 mg/kg
>2 fold at 10 mg/kg
>2 fold at 30 mg/kg
0.019
0.049
0.039
0.37
0.15
0.72
55
103
No elevation up to 100 mg/kg
and amenable to stable salt formation. Especially in light of the
poor overall solubility profile of the series, we thought this might
be of use for formulation optimization in case dissolution proper-
ties play a role in in vivo exposure. The imidazolopyrimidine 44
and 5,5-bicyclic pyrazolothiazole 45 resulted in modestly lower
potency and efficacy, whereas the purine derivative 46 was not
well tolerated.
reached any level of in vivo efficacy. Compound 36 may serve as
a tool compound for further in vivo investigations and as a lead
into further medicinal chemistry optimization toward a drug for
the treatment of T2D.
In summary, we started with the HTS-derived hit 5, which dis-
played moderate in vitro activity, but no oral bioavailability. The
series was subsequently proven not to fall into the common
GPR119 pharmacophore widely reported to date and represents
novel chemical space for further SAR studies into GPR119 agonism.
Through a systematic SAR study of 5, we were able to improve both
in vitro GPR119 agonist activity and functional activity (e.g., GLP-1
and insulin secretion in the relevant cell lines) and achieve high
oral exposure for selected analogues in mice. Furthermore, com-
pound 36 induced robust elevation of active GLP-1 when dosed
acutely at 10 mg/kg in C57BL/6 mice.
Further characterization of selected compounds described in
this Letter can be seen in Table 3. In general, analogues such as
21 that were stabilized by quaternization displayed good-to-excel-
lent human and mouse GPR119 agonist activity. Furthermore,
these compounds also showed robust GLP-1 secretion in GLUTag
cells and insulin secretion in HIT-T15 cells, respectively. Curiously,
switching the phenyl para-substituent from a chloride to a trifluo-
romethyl (21 vs 36) not only reproducibly enhanced the activity on
the mouse receptor, but also raised the exposure levels from mod-
erate to high. This may in part explain why compound 36 is effica-
cious in an acute study measuring active GLP-1 in C57BL/6 mice
after a 10 mg/kg dose, while compound 21 is inactive at doses up
to 100 mg/kg.15 Despite being the most potent analogue on the hu-
man receptor, amide 39 is similarly active on the mouse receptor
compared to the ester analogues, and is efficacious at the higher
30 mg/kg dose, perhaps due to lower exposure versus 36. Com-
pound 43 failed to produce a statistically significant increase of
GLP-1 at doses up to 100 mg/kg, despite a similar in vitro activity
and in vivo PK profile compared to 39. The reasons for this discrep-
ancy are currently not understood.
References and notes
1. For T2D statistics, charts and figures see: www.diabetes.org or www.idf.org.
2. Fisman, E. Z.; Tenenbaum, A. Cardiovasc. Diabetol. 2009, 8, 38.
3. Krentz, A. J.; Patel, M. B.; Bailey, C. J. Drugs 2008, 68, 2131.
4. Jones, R. M.; Leonard, J. N. Annu. Rep. Med. Chem. 2009, 44, 149.
5. Jones, R. M.; Leonard, J. N.; Buzard, D. J.; Lehmann, J. Expert Opin. Ther. Patents
2009, 19, 1339.
6. Shah, U. Curr. Opin. Drug Disc. Dev. 2009, 12, 519.
7. Fyfe, M. C. T.; McCormack, J. G.; Overton, H. A.; Proctor, M. J.; Reynet, C. Expert
Opin. Drug Disc. 2008, 3, 403.
8. Chu, Z.-L.; Jones, R. M.; He, H.; Carroll, C.; Gutierrez, V.; Lucman, A.; Moloney,
M.; Gao, H.; Mondala, H.; Bagnol, D.; Unett, D.; Liang, Y.; Demarest, K.; Semple,
G.; Behan, D. P.; Leonard, J. Endocrinology 2007, 148, 2601.
9. Overton, H. A.; Babbs, A. J.; Doel, S. M.; Fyfe, M. C. T.; Gardner, L. S.; Griffin, G.;
Jackson, H. C.; Procter, M. J.; Rasamison, C. M.; Tang-Christensen, M.;
Widdowson, P. S.; Williams, G. M.; Reynet, C. Cell Metab. 2006, 3, 167.
10. Curr. Opin. Drug Discovery Dev. 2009, 12, 519.
At this point, we concluded that the robust GLP-1 elevation of
36 observed in mice was encouraging for the series, as most other
series we pursued outside the common pharmacophore never

P. Alper et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2383–2387
2387
11. Eur. J. Med. Chem. 2011, 46, 2901.
12. Kang, S.-U. Drug Discovery Today 2013, 18, 1309.
15. Groups of wild-type C57BL/6J mice (n = 8 mice per group) were randomized
into treatment groups based on their initial body weight. Mice were housed
four per cage and orally dosed with vehicle, DPP-4 inhibitor alone, or DPP-4
inhibitor and GPR119 agonist, in a single dose. A glucose bolus (3 g/kg) was
delivered thirty minutes post dosing. Sample was collected 2 min post glucose
bolus. All animals were fasted for 16 h prior to compound administration.
Blood was obtained (via retro-orbital bleeding) to measure plasma levels of
13. The GPR119 agonist activity of the compounds is tested in a CHO stable cell
line overexpressing human GPR119 (Lonza). The compound-induced activation
of GPR119 leads to an increase of cellular cAMP production, which is measured
with a cAMP HTRF kit (Cisbio). The efficacy of the receptor activation is
normalized to AR231453 being 100%.
14. The compound activity on human or mouse GPR119 is measured with a cAMP
HTRF assay in CHO cells overexpressing hGPR119 or mGPR119. Compound-
induced GLP-1 secretion is measured in a mouse enteroendocrine cell line
GLUTag. The secreted GLP-1 in the supernatant of cells treated with GPR119
agonist is detected with a HEK293 reporter cell line that co-expresses GLP-1
receptor and CRE-luciferase constructs. GPR119 agonist-induced insulin
secretion in a hamster beta cell line HIT-T15 is measured with an insulin
HTRF kit from Cisbio. The efficacy of the compounds in these assays is
normalized to AR231453 being 100%.
active GLP-1. Approximately 200 lL samples of blood were removed for
analysis at 62 min post dosing (2 min post glucose bolus). Active GLP-1 was
measured using Glucagon-like peptide-1 (active) ELISA Kit, 96-well plate
(Linco Research, Inc.). All procedures in this study were approved by the GNF
animal care and use committee and were in compliance with the Animal
Welfare Act Regulations 9 CFR Parts 1, 2 and 3, and US regulations (Guide for
the Care and Use of Laboratory Animals, 1995).