Structural Basis for Achieving GSK-3β Inhibition with High Potency, Selectivity, and Brain Exposure for Positron Emission Tomography Imaging and Drug Discovery

Article abstract of DOI:10.1021/acs.jmedchem.9b01030

Using structure-guided design, several cell based assays, and microdosed positron emission tomography (PET) imaging, we identified a series of highly potent, selective, and brain-penetrant oxazole-4-carboxamide-based inhibitors of glycogen synthase kinase-3 (GSK-3). An isotopologue of our first-generation lead, [³H]PF-367, demonstrates selective and specific target engagement in vitro, irrespective of the activation state. We discovered substantial ubiquitous GSK-3-specific radioligand binding in Tg2576 Alzheimer's disease (AD), suggesting application for these compounds in AD diagnosis and identified [¹¹C]OCM-44 as our lead GSK-3 radiotracer, with optimized brain uptake by PET imaging in nonhuman primates. GSK-3β-isozyme selectivity was assessed to reveal OCM-51, the most potent (IC₅₀ = 0.030 nM) and selective (>10-fold GSK-3β/GSK-3α) GSK-3β inhibitor known to date. Inhibition of CRMP2^T514 and tau phosphorylation, as well as favorable therapeutic window against WNT/β-catenin signaling activation, was observed in cells.

Full text of DOI:10.1021/acs.jmedchem.9b01030

Subscriber access provided by BUPMC - Bibliothèque Universitaire Pierre et Marie Curie
Article
Structural Basis for Achieving GSK-3# Inhibition with High Potency,
Selectivity and Brain Exposure for PET Imaging and Drug Discovery
Vadim Bernard-Gauthier, Andrew V. Mossine, Ashley Knight, Debasis Patnaik, Wenning Zhao,
Chialin Cheng, Hema S. Krishnan, Lucius L. Xuan, Peter S. Chindavong, Surya A. Reis, Jinshan
Michael Chen, Xia Shao, Jenelle Stauff, Jana Arteaga, Phillip Steven Sherman, Nicolas Salem,
David Bonsall, Brenda Amaral, Cassis Varlow, Lisa Wells, Laurent Martarello, Shil Patel,
Steven H Liang, Ravi G. Kurumbail, Stephen J. Haggarty, Peter J. H. Scott, and Neil Vasdev
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b01030 • Publication Date (Web): 19 Sep 2019
Downloaded from pubs.acs.org on September 20, 2019
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 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 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.
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 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
Medicine and Molecular Imaging
Kurumbail, Ravi; Pfizer Global Research and Development
Haggarty, Stephen; Massachusetts General Hospital, Center for Genomic
Medicine
7
8
9
Scott, Peter; University of Michigan, Division of Nuclear Medicine,
Department of Radiology; University of Michigan, The Interdepartmental
Program in Medicinal Chemistry
Vasdev, Neil; Centre for Addiction and Mental Health, Azrieli Centre for
Neuro-Radiochemistry; Research Imaging Centre; Massachusetts
General Hospital, Division of Nuclear Medicine and Molecular Imaging;
University of Toronto, Department of Psychiatry/ Institute of Medicine
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 47
1
2
3
4
5
Structural Basis for Achieving GSK-3 Inhibition with High Potency,
Selectivity and Brain Exposure for PET Imaging and Drug Discovery
6
7
8
9
Vadim Bernard-Gauthiera,b,c,1*, Andrew V. Mossined,1, Ashley Knighta,b,e, Debasis Patnaikf, Wen-Ning
Zhaof, Chialin Chengf, Hema S. Krishnanc Lucius L. Xuanf, Peter S. Chindavongf, Surya A. Reis,f Jinshan
Michael Cheng, Xia Shaod, Jenelle Stauffd; Jana Arteagad; Phillip Shermand, Nicolas Salemh, David
Bonsalli, Brenda Amaralh, Cassis Varlow,a Lisa Wellsi Laurent Martarelloh, Shil Patele Steven H. Liangc,
Ravi G. Kurumbailg, Stephen J. Haggartyf, Peter J. H. Scottd,j, Neil Vasdeva,b,c,*
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
aAzrieli Centre for Neuro-Radiochemistry, Research Imaging Centre, Centre for Addiction and Mental
Health, Toronto, ON, M5T 1R8, Canada; bDepartment of Psychiatry/Institute of Medical Science,
University of Toronto, Toronto, ON, M5T 1R8, Canada; cDivision of Nuclear Medicine and Molecular
Imaging, Massachusetts General Hospital and Department of Radiology, Harvard Medical School,
Boston, MA, 02114, USA; dDivision of Nuclear Medicine, Department of Radiology, The University of
Michigan Medical School, Ann Arbor, MI, 48109, USA; eEisai AiM Institute, Boston, MA, 01810, USA;
fChemical Neurobiology Laboratory, Massachusetts General Hospital, Center for Genomic Medicine,
Departments of Neurology & Psychiatry, Harvard Medical School, Boston, MA, 02114, USA; gPfizer
Worldwide Research and Development, Groton Laboratories, Eastern Point Road, Groton, CT, 06340,
USA; hBiogen, Research and Early Development Imaging, Cambridge, MA, 02142; iInvicro, London, UK;
jThe Interdepartmental Program in Medicinal Chemistry, University of Michigan, Ann Arbor, MI, 48109,
USA.
Author contributions: V.B.-G., N.V., A.K., D.P., W.-N.Z., P.C., J.M.C., N.S., L.W., L.M., R.G.K., S.J.H.
designed research; V.B.-G., A.V.M., N.V., A.K., D.P., W.-N.Z., C.C., L.L.X, P.S.C., S.A.R., J.M.C., X.S.,
J.S., J.A., P.S., H.S.K., S.H.L., S.P.; N.S., L.W., L.M. performed research; V.B.-G., N.V., N.S., L.M.,
R.G.K., S.J.H., P.J.H.S. contributed new reagents/analytical tools; V.B.-G., A.V.M., N.V., A.K., D.P., W.-
N.Z., C. C., P.C., J.M.C., S.P., N.S., C.V., L.W., L.M., R.G.K., S.J.H. analyzed data; V.B.-G. wrote the
paper.
Conflict of interest statement: N.V. is a co-founder of MedChem Imaging, Inc. S.H.L. is a consultant for
MedChem Imaging, Inc. S.J.H. is a member of the Scientific Advisory Board of Rodin Therapeutics,
Frequency Therapeutics, and Psy Therapeutics. All other authors declare no conflict of interest.
*To whom correspondence may be addressed. *Email: vadimbgauthier@gmail.com,
neil.vasdev@utoronto.ca
1V.B.-G. and A.V.M. contributed equally to this work.
This article contains supporting information
ACS Paragon Plus Environment

Page 3 of 47
Journal of Medicinal Chemistry
1
2
3
4
Abstract
5
6
7
8
Using structure-guided design, several cell based assays, and microdosed positron emission tomography
(PET) imaging, we identified a series of highly potent, selective and brain penetrant oxazole-4-
carboxamide-based inhibitors of glycogen synthase kinase-3 (GSK-3). An isotopolog of our first-
generation lead, [3H]PF-367, demonstrates selective and specific target engagement in vitro, irrespective
of activation state. We discovered substantial ubiquitous GSK-3-specific radioligand binding in Tg2576
Alzheimer’s disease (AD), suggesting application for these compounds in AD diagnosis, and identified
[11C]OCM-44 as our lead GSK-3 radiotracer, with optimized brain uptake by PET imaging in nonhuman
primates. GSK-3-isozyme selectivity was assessed to reveal OCM-51, the most potent (IC50 = 0.030 nM)
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
and selective (>10-fold GSK-3/GSK-3) GSK-3 inhibitor known to date. Inhibition of CRMP2T514 and tau
phosphorylation, as well as favorable therapeutic window against WNT/-catenin signaling activation was
observed in cells.
Keywords: Glycogen synthase kinase-3; GSK-3; Alzheimer's disease; positron emission tomography;
PET; neuroimaging.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 47
1
2
3
4
Introduction
5
6
7
8
Glycogen synthase kinase-3 and  (GSK-3) are ubiquitous multifunctional serine/threonine kinases
which regulate diverse physiological functions in peripheral organs and the central nervous system
(CNS), ranging from metabolism, cell cycle regulation, to brain development. Appropriately balanced
GSK-3 regulation and signaling is therefore central to human physiology but often compromised in
diseases of diverse etiologies such as diabetes, cancers and neurological disorders.1,2 GSK-3 are
remarkably networked enzymes, with involvement in numerous pathways and over one hundred distinct
putative substrates.3 Also unusual among kinases is the fact that GSK-3 regulation occurs via
inactivation while resting cells show high levels of catalytic activity. The inactivation of GSK-3 has been
well-characterized and revealed primarily to proceeds through N-terminus serine phosphorylation
catalyzed by kinases such as AKT/PKB, p70RSK and p90RSK (at serine 21 and serine 9 for GSK-3 and
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
GSK-3 respectively).4, 5 Of note, GSK-3 and GSK-3 present significant kinase domain homology with
one another, especially within the ATP binding pocket (97% identity).6, 7 Congruently, GSK-3 also
exhibit substantial functional overlap, although some elements of nonredundancy such as GSK-3-
specific perinatal lethality have been characterized from paralog-specific Gsk3 knockout model studies
which together support the notion that individual isozyme contributions should be defined carefully.7-9
Several immunohistochemical, cellular, transgenic and preclinical studies have implicated GSK-3
deregulation as a contributive factor in neurological diseases including Alzheimer’s diseases (AD) and
non-AD tauopathies.10-12 While in most cases, GSK-3 has been the central focus in those studies, both
GSK-3 overexpression and inhibition/knockdown experiments have demonstrated roles in APP
processing and correlated with increased or decreased amyloid- production respectively.13, 14 In turn,
amyloid- injections in rat trigger APP processing changes and pathological tau phosphorylation linked to
GSK-3 overexpression and activation.15 The hyperphosphorylation of tau by GSK-3 also suggests a
central implication in neurofibrillary tangle formation, however whether an increase in GSK-3 levels or
activities alone, rather than parallel changes in phosphatase activities constitute the causative pathogenic
factor in tau hyperphosphorylation and neurofibrillary tangle accumulation remain to be clarified.16-18 Both
ACS Paragon Plus Environment

Page 5 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
inhibition of GSK-3 in transgenic mice overexpressing human tau and overexpressing conditional Gsk3b
mice were associated with changes in hyperphosphorylated tau protein (pTau) levels tightly linked to
GSK-3 levels and activity in vivo.19, 20 Current evidence using human tissue suggest, as with most
immunohistochemical cellular and transgenic rodent analyses, increases in total and active GSK-3 levels
in AD compared to health controls.21-23 Yet, data gathered in human brains have been largely insufficient
to substantiate in vitro preclinical findings while in vivo pharmacological intervention studies in transgenic
mice as well as clinical trials aimed at validating GSK-3 in humans have been obscured by the systematic
use of promiscuous inhibitors.24-26 Indeed, much of the inability to define whether GSK-3 is an actionable
target in neurodegenerative diseases in vivo ensue from the lack of suitable molecular tools. To address
this, highly potent, selective and brain penetrant inhibitors of GSK-3 are needed. However, apart from
rare exceptions where some of these criteria have been met, so far clinical and preclinical GSK-3
inhibitors have been both sub-optimal regarding potency or brain exposure and have displayed, despite
sporadic claims of selectivity in restricted kinase subsets, poor overall kinome selectivity when
systematically evaluated in large scale screening.27-31 Moreover, only one recent report has described
kinome selective GSK-3 inhibitors also achieving -isozyme selectivity, however, none of these
isozyme-selective compounds exhibit single-digit nanomolar potencies or have been characterized in
CNS tissue.32
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
Herein, we used structure guided design and microdosed positron emission tomography (PET)
imaging to develop novel small molecule inhibitors of GSK-3 achieving high potency, selectivity and brain
exposure. We describe the rational development of the novel GSK-3-isozyme selective inhibitor, named
herein as OCM-51 (51), which to the best of our knowledge, constitutes the most potent and selective
GSK-3 inhibitor identified thus far. Biochemical, biophysical, cellular and brain tissue-based assays
collectively were used to confirm highly selective target engagement. Further analysis also demonstrates
differential binding profiles in AD transgenic mice brains versus wild-type. Together, our results describe
finely characterized candidate inhibitors which can lead to the development of GSK-3 specific
therapeutics as well as diagnostic GSK-3 PET molecular probes for neuroimaging.
Results
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 47
1
2
3
4
5
6
7
8
9
[3H]PF-367 Engages GSK-3 with High Specificity in Cells and Human Postmortem Brain.
Previously, we reported the discovery of PF-04802367 (PF-367; 1), a potent type-I oxazole-4-
carboxamide dual GSK-3 inhibitor showing promising ADME properties combined with robust
CNS/peripheral pTau and muscle pGS inhibition in vivo.28 Here, we first tested whether the lead
compound, PF-367, displayed specific GSK-3 binding in mammalian brain tissue. We developed [3H]1
(Fig. 1A) and observed, in rat, marmoset and human brains (frontal cortex and hippocampus), high and
nearly ubiquitous total binding in gray matter consistent with known GSK-3 mRNA and protein levels
(Fig. 1B-F, Fig. S1B,C). Highest radioligand binding was measured in cerebral cortex and consistently
more pronounced within gray matter regions compared to white matter in all species (50–60% difference).
To confirm that our GSK-3 marker provided saturable signal, homologous competition was performed
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
on adjacent sections using unlabeled 1 at 10 M, and resulted in significantly reduced binding compare to
baseline experiments and overall low levels of non-saturable binding (>87%, 66% and 85% reductions in
rat, marmoset and human cerebral cortices, respectively, Fig. 1B,D,E). We then used two structurally
unrelated and well-characterized heterologous pan-GSK-3 competitive inhibitors to determine on-target
binding selectivity. Single concentration cotreatments with the advanced preclinical inhibitor CHIR99021
and the clinical lead LY2090314 both resulted in substantial decreases in tracer binding similar to
unlabeled 1incubations (Fig. 1B-F, Fig. S1A). In all species and brain regions, there were no significant
discrepancies in blocking effects between 1 and the heterologous probes, indicating that GSK-3
specificity could be determined using any of these inhibitors interchangeably.
To further assess our probe in a cellular context, [3H]1 binding in HEK293T P301S hTau cells
overexpressing GSK-3, shRNA induced GSK-3 knockdown and non-overexpressing HEK293T
P301S hTau cells was compared (Fig. 1H-J, Fig. S1F). Significantly higher binding was detected in
overexpressing cells compared to GSK-3 knockdown (three-fold) and control (six-fold) while cold
chase studies (where non-radioactive agent is administered following the radiotracer) resulted in
consistent time-dependent binding reductions in all cells (Fig. 1H,I). Strong, dose-dependent, GSK-3
specific and reversible binding was observed with overexpressing HEK293T P301S hTau cells when
incubated with [3H]1 in competition (co-incubation, 85% displacement at 5 M) and pulse chase (96%
ACS Paragon Plus Environment

Page 7 of 47
Journal of Medicinal Chemistry
1
2
3
4
displacement at 5 M for 5 min) studies with a third heterologous inhibitor (e.g. AR79, Fig. S1A,G).
5
6
7
8
Further evidence of GSK-3 target engagement and selectivity in brain tissue was found when
comparing tumor to normal CNS contralateral tissue binding of [3H]1 in rat brain sections with intracranial
GS9L gliomas in vitro (>40% binding differences, Fig. S1D,E). We next performed binding saturation
experiments using mouse brain homogenates which determined a GSK-3 binding site density (Bmax) of
960.33 ± 234 fmol/mg and an affinity in line with biochemical data (KD = 13.02 ± 3.3 nM) (Fig. 1G).
Consistently throughout these cellular and whole tissue assays, the observed signals are predictably
derived from GSK-3 and GSK-3 – as large-scale proteomic analyses have indicated only about two- to
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
fourfold higher CNS Bmax for GSK-3 when compared to GSK-3.33, 34 Together, these results also
indicated that the 1scaffold aligns suitable target engagement and selectivity properties to serve as a
starting point in our lead-to-candidate optimization efforts.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 47
1
2
3
4
5
6
7
A
B
P
= 0.04
70
60
50
40
30
20
10
0
N
N
N
NS
3H3CO
O
NH
Cl
N
O
[
3H]PF-367
8
9
P
=
0.0001
P
=
0.0020
= 0.0036
0.0007
C
D
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
Rat brain
in vitro (coronal)
P
=
P
20
15
10
5
[
3H]PF-367
Ctx
[
3H]PF-367+ PF-367
CC
(10 M)
CPu
max
min
[
3H]PF-367+ CHIR99021
(10 M)
[
3H]PF-367 + LY2090314

(10 M)
max
min
0
F
Human frontal
cortex in vitro
E
[
3H]PF-367
[
3H]PF-367+ PF-367
P
< 0.0001
GM
(10 M)
14
12
10
8
6
4
2
0
-2
WM
max
NS
min
[
3H]PF-367+ CHIR99021
(10 M)
[
3H]PF-367 + LY2090314
(10 M)


max
min
Gray matter
White matter
1200
G
KD = 13.02  3.3 nM
Bmax = 960.33  234 fmol/mg
800
400
0
100
80
60
40
20
0
0
200
400
600
800
1000
Bound (fmol/mg)
0
20
40
60
80
100
[
3H]PF-367
P
=
0.0005
0.001
= < 0.0001
I
P
=
H
P
14
12
10
8
P
= 0.026
6
GSK-3
-actin
4
2
0
2h
5
min 15 min 30 min
chase chase chase
Fig. 1. [3H]PF-04802367 ([3H]1) displays high specific and selective binding to GSK-3/ in cells and
mammalian brains in vitro. (A) Chemical structure of [3H]1. (B) Quantification for the in vitro
autoradiography of [3H]1 in the normal rat brain (20 m sections, n = 3) under baseline, homologous
blocking (nonradioactive 1, 10 M) and heterologous (LY2090314, 10 M or CHIR99021, 10 M)
conditions. (C) In vitro autoradiography images of [3H]1 normal rat brain (coronal sections) under
baseline, homologous blocking (nonradioactive 1, 10 M) and heterologous (LY2090314, 10 M or
CHIR99021, 10 M) conditions. (D) Quantification for the in vitro autoradiography of [3H]1 in the normal
marmoset brain (20 m sections, n = 3) under baseline, homologous blocking (nonradioactive 1, 10 M)
and heterologous (LY2090314, 10 M or CHIR99021, 10 M) conditions. (E) Quantification for the in vitro
autoradiography of [3H]1 in the normal human brain (20 m sections, frontal cortex, n = 6) under baseline,
ACS Paragon Plus Environment

Page 9 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
homologous blocking (nonradioactive 1, 10 M) and heterologous (LY2090314, 10 M or CHIR99021, 10
M) conditions. (F) In vitro autoradiography images of [3H]1 normal human brain under baseline,
homologous blocking (nonradioactive1, 10 M) and heterologous (LY2090314, 10 M or CHIR99021, 10
M) conditions. (G) Saturation [3H]1 binding curves, Scatchart plot and Bmax determination using
homogenate mouse brain tissue. (H,I) Cell binding assays with [3H]1 in HEK-293T P301S hTau cells,
overexpressing GSK- 3/ and GSK-3α/β knockdown (shRNA). [3H]1 binding under 2h baseline and
following timecourse chase experiments. (J) Western blot analysis of control, GSK-3/overexpressing
and GSK-3/ shRNA cells used for [3H]1 binding studies. Data show mean values  SD. Statistical
significance was calculated one-way ANOVA followed by Dunnett’s test (B,D,E,H).
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
Development and Characterization of Selective Oxazole-4-carboxamide GSK-3 Inhibitors.
The hit-to-lead screening campaign which led to 1 had pointed out the importance of three key fragments
with regard to GSK-3 potency and selectivity: (1) the 1,2,4-triazol-1-yl group involved in a cation-
interaction with Arg141; (2) a critical core oxazole-4-carboxamide moiety providing hinge domain-ligand
hydrogen bond pairing interactions around Val135/Asp133; and (3) the 3-chloro-4-methoxyphenyl ring
mainly involved in hydrophobic contacts within the deep pocket formed between the N- and C-lobes in the
vicinity of Phe67, Lys85 and Leu132 (GSK-3 residues) and corresponding GSK-3 residues (28). On the
basis of the aforementioned in vitro cell and tissue studies, and guided by structural data gained
previously, we undertook structural refinement efforts to improve blood-brain barrier (BBB) permeation,
which is suboptimal for 128, and in vivo properties of our lead while retaining or improving physical and
biochemical properties.28 To identify next-generation inhibitors, >50 optimized compounds were
synthesized and underwent biochemical screening (Fig. 2B, SI Materials and Methods and Table S1-4).
In designing derivatives, emphasis was placed on the exploration of the effects of halogenation patterns,
primarily fluorinations, and the modulation of the cation- interacting group while keeping the oxazole-4-
carboxamide pharmacophore intact and minimizing changes detrimental to calculated Fu_p (unbound
brain fraction) and CNS multiparameter optimization (MPO) scores (properties which are suboptimal for
most known GSK-3 inhibitors, including PET probes evaluated to date, Fig. 2B, SI Materials and Methods
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 47
1
2
3
4
5
6
7
8
9
and Table S1-4). Our structure-activity relationship (SAR) study around the cation- interacting side chain
amide substituents showed that 6-membered nitrogen-containing rings (pyrazine, pyridines) combined
with 3-“chlorine-for-fluorine” substitutions at the 4-methoxyphenyl fragment were well tolerated. We
identified three inhibitors with such structural patterns displaying single-digit nanomolar potencies towards
GSK-3 (namely, PF-618 (28), OCM-37 (37) and OCM-44 (44); IC50 < 2.5 nM) in our ADPGlo screening
assay (Fig. 2A). Gratifyingly, these structural modifications were associated with profound CNS-
penetration enhancements in vivo (vide infra). The high binding affinity and dual GSK-3 inhibitory
nature of these analogs were confirmed in a second biochemical assay format utilizing radiometric
filtration (-[33P]-ATP kinase assay, SI Materials and Methods) (GSK-3 Ki = 3.3/4.2 nM, 28, Fig. 2C,D).
When assayed against a comprehensive panel of 371 kinase targets, these inhibitors displayed distinctive
selectivity, inhibiting only 2 targets (DYRK1A and DYRK1B) beyond IC50 concentrations at 1.0 M (Fig.
2F, Table S5,6). While activity overlap between GSK-3 and the Down syndrome kinases DYRK1s is
common across inhibitors aimed at both targets, we observed that in this case it may be exacerbated by
ligand-receptor  interactions from a conserved DYRK1A/B Asp247/198 mimicking the guanidinium
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
Arg141 cation- interaction (Fig. 2E,G-I).35 However, we found 40-170-fold selectivity for GSK-3 when
measuring specific inhibitory constants even against the closest off-targets (DYRK1A / DYRK1B Ki = 325
nM / 171 nM; Fig. 2C,D), well within the requirements expected for therapeutics and PET radioligand
development.
ACS Paragon Plus Environment

Page 11 of 47
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
120
100
80
60
40
20
0
C
D
Target
PF-618 Ki (nM)
CLK4
DYRK1B
DYRK1A
GSK-3
GSK-3
700
171
325
4.16
3.29
CLK4
DYRK1B
DYRK1A
GSK-3
~40-170-
fold
shift

GSK-3
selectivity (fold)
1.3
GSK-3

DM-1
S
1
O
-10 -9
-8
-7
-6
-5
-4
Log [PF-618] (M)
E
Gatekeeper Hinge 2
GSK-3:
132
238
LDYVPETVYRVA
FEMLSYNLYDLL
DYRK1A:
Met240/
Tyr134
OCM-44
PF-618
(1.0 M)
Phe238/
Leu132
(1.0 M)
~667-fold
~400-fold
G
over GSK-3 IC50 over GSK-3 IC50
F
CDK2 (cyclin A) (44%)
CDK2 (cyclin A1) (41%)
CDK9 (cyclin T1) (49%)
CDK9 (cyclin T2) (60%)
CLK4 (48%)
Tyr243/
Glu137
CDK9 (cyclin T2) (50%)
CLK4 (52%)
Asn244/
Thr138
DYRK1A (86%)
DYRK1B (83%)
GSK-3 (99%)
GSK-3 (98%)
DYRK1A (78%)
DYRK1B (80%)
Asp247/
Arg141
GSK-3 (100%)
GSK-3 (100%)
Tyr246/
Tyr140
PKMYT1 (56%)
2
STK39/STLK3 (49%)
100
0
% Inhibition
OCM-44 / GSK-3
OCM-44 / GSK-3
H
I
Lys85
2
Phe67
1
Val135
3
Arg141
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 47
1
2
3
4
5
6
7
8
9
Fig. 2. Evolution and in vitro pharmacological characterization of a highly potent and selective oxazole-4-
carboxamide-based GSK-3 inhibitors series. (A) Generalized strategy to synthesize arrays of carboxamide-
based GSK-3 inhibitors, as well as structures and properties of selected GSK-3 inhibitors (for details on
chemical synthesis, see Supplementary Schemes 14). IC50 values determined using an ADP-Glo kinase
assay (n = 2). (B) Heat map representing key properties (IC50, CNS MPO and Fu_p) of oxazole-4-
carboxamide-based GSK-3 inhibitors including a comparison with previously described GSK-3 PET
radiotracers (for structures and detailed parameter values, see Supplementary Tab. 14). IC50 values
determined using an ADP-GloTM kinase assay (n = 2). (C and D) The dose-response curves and binding
affinities of the representative inhibitor 28 versus GSK-3, GSK-3 n = 3) and main kinases off targets
DYRK1A, DRYK1B and CLK4 (n = 1) demonstrate high selectivity for GSK-3/ (radiometric filtration
binding assay, [ATP] = Km ATP, data show mean values  SD). (E) Hinge/2 region sequence alignment of
GSK-3 and the main kinase secondary target DYRK1A. (F) Comprehensive kinase selectivity profile of
representative inhibitor 44 and 28 against 371 kinases (kinases are ordered alphabetically and the activity
is relative to control at 1.0 M inhibitor concentration, n = 2, [γ-33P]ATP-based enzymatic assay performed
by Reaction Biology, for detailed data see Supplementary Table S5). (G) Overlay of the hinge/2 region of
GSK-3 (gray, PDB ID code 5K5N) and DYRK1A (yellow, PDB ID code 3ANR). (H) Predicted detailed
interaction of 44 (yellow sticks) with GSK-3 (gray cartoon, PDB ID code 5K5N). Pi-cation interaction is
depicted as a black dash line. (I) Alternate view of the binding mode of 44 (yellow, space fill model) with
GSK-3 (surface rendering, PDB ID code 5K5N). CNS MPO, central nervous system multiparameter
optimization; Fu_p, in silico calculated unbound fraction in plasma.
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
Oxazole-4-Carboxamide Probes Engage npS9-GSK-3 and pS9-GSK-3 irrespective of activation
states.
Pathological changes associated with GSK-3 may be overexpression- or overactivation-driven, therefore
it is crucial to address whether molecular probes engaging GSK-3 do so dependently or independently
of the S9 phosphorylation status in order to extrapolate valid imaging applications (S21 in GSK-3
isozyme). To elucidate possible target engagement discrepancies between npS9-GSK-3 (active
nonphospho-S9) and pS9-GSK-3 (inactive phospho-S9), we developed a high sensitivity microscale
thermophoresis (MST)-based binding assay. Use of MST allows for high-precision determination of
equilibrium dissociation constants by measuring temperature induced changes in a fluorescently-labeled
target protein as a function of small amounts of non-fluorescent small molecules with minimal influence by
ATP concentration and other such factors that impact functional cell-base assays.36 We generated
catalytically inactive pS9-GSK-3 by treating non-phosphorylation active npS9-GSK-3 (Fig. 3B) with AKT
and ATP. When npS9-GSK-3 was treated with AKT and high ATP concentration for 3 h, complete loss of
catalytic activity was observed and high level of pS9-GSK-3 was produced as determined by ADPGlo
kinase and Western blot analyses (Fig. 3C,D). Under control conditions using ATP treatments alone,
GSK-3 catalytic activity remained unaffected (Fig. 3C,D). Characterization by MST of the npS9-GSK-
ACS Paragon Plus Environment

Page 13 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3 and pS9-GSK-3 proteins first confirmed the high binding affinities of oxazole-4-carboxamide inhibitors
towards npS9-GSK-3 (Kd = 0.8 and 4.5 nM for 1 and 44 respectively) and revealed only marginal
differences versus inactive pS9-GSK-3 with regard to target engagement (< two-fold, Fig. 3E-I). Under
the same conditions, CHIR99021 GSK-3 binding was also largely independent of S9 status (Kd = 48.8
and 28.3 nM for pS9-GSK-3 and npS9-GSK-3 respectively, Fig. S2A).
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
Consistent with these MST results, in silico experiments and X-ray crystal structure analyses,
indicated that conformational changes occurring upon N-terminal inhibitory peptide sequence binding do
so remotely from the inhibitor binding site (Fig. 3A). GSK-3 structural remodeling following pS9
autoinhibition primarily affect the C-loop and activation segment which are positioned distantly from the
bound inhibitor and have negligible impact on the ATP binding site conformations, thus having limited
impact on the target engagement of type-I ATP competitive inhibitors as observed here (Fig. 3A).37 On
the basis of these results, we conclude that oxazole-4-carboxamide-based probes, such as [3H]1 and
positron emitter-labeled derivatives thereof, provide a genuine assessment of total GSK-3 density
irrespective of the S9 phosphorylation.
GSK-3 pS9
Activation
loop
A
B
pS9
C-loop
50
40
30
20
10
P = 0.0003
C
N
N-term
Y216
Gly-rich Hinge
loop
N lobe
C-loop
0
D
0 102030 40 50
Time (min)
pS9 N-term peptide
OCM-44
GSK-3
Activation loop
(GSK-3
kDa
48
GSK-3-pS9
C
80
70
60
Axin
P = 0.0018
GSK-3
50
NS
40
30
20
E
MST
d GSK-3
(active -
K
Kd GSK-3
(inactive -
pS9, nM)
0
10 20 30 40 50
0
10 20 30 40 50
Inhibitor
Time (min)
npS9, nM)
GSK-3
+ ATP
+ Akt
GSK-3
+ ATP
PF-367
OCM-44
0.79
4.5
1.69
8.6
F
G
H
I
npS9
npS9
Akt treated
Akt treated
4
2
5
0
0.5
0.0
0.5
0.0
0
-5
-0.5
-1.0
-1.5
-2.0
-2
-4
-6
-8
-0.5
-1.0
-1.5
-10
-15
-20
-11
-10
-9
-8
-7
-6
-5
-11
-10
-9
-8
-7
-6
-5
-11
-10
-9
-8
-7
-6
-5
-11
-10
-9
-8
-7
-6
-5
Log [PF-367] (M)
Log [PF-367] (M)
Log [OCM-44] (M)
Log[OCM-44] (M)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 47
1
2
3
4
5
6
7
8
9
Fig. 3. Oxazole-4-carboxamide-based probes engage both active npS9 and catalytically inactive pS9
GSK-3. (A) Structural basis for the recognition of inhibitors by pS9-GSK-3 44 is depicted as a
representative inhibitor, yellow, space fill model). The predicted binding mode illustrates the distal
positioning of 44 from the activation loop and pS9 auto-inhibitory N-terminal peptide (blue surface) and
the negligible ATP-binding site rearrangement upon binding of the pS9 auto-inhibitory N-terminal peptide
of GSK-3 (gray surface, PDB ID code 4NM3). (B) Timecourse analysis showing GSK-3 activity in the
presence of low ATP content and GSP2 (ADP-Glo-based kinase assay, n = 3). (C) Timecourse analysis
showing GSK-3 activity in the presence of high ATP content and high ATP content combined with AKT
(ADP-Glo-based kinase assay, n = 3). (D) Western blot analysis of enzymatic reaction contents (control,
ATP treated and ATP+AKT treated). (E) Binding affinities of 1 and 44 measured by MST. (F and G) MST
dose-response curves for the binding interaction of between 1 with active npS9-GSK-3 and pS9-GSK-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
(n = 2). (H and I) MST dose-response curves for the binding interaction of 44 with active npS9-GSK-3
and pS9-GSK-3 (n = 2). Data show mean values  SD. Statistical significance was calculated with paired
t test (B and C).
Discovery of Structural Determinants to Generate Picomolar GSK-3Isozyme Selective Inhibitors.
In the course of the initial SAR study, we observed that the orientation of the 3-halogen phenyl
substituents found in this series was optimal with respect to the deep hydrophobic cleft found in GSK-
3 (Fig. 2A,H, Fig. 4C,E-G). Close inspection of GSK-3-1 and other GSK-3 co-crystal structures
indicated that, as recently noted by Wagner et al. 2018, the topology of this pocket is dependent on a
complex backend hydrogen bonding network involving the side chain of Asp133, the only non-conserved
residue from the ATP binding site between GSK-3 and GSK-3 (Glu196 in GSK-3).32 We observed that
the hydrogen bond interactions in GSK-3 which bring the side chains of Asp133, Arg113 and Glu80 in
close contact also directly impact proximal hydrophobic residues from  sheets 5 and 6 (gatekeeper
Leu132, Leu130 and Leu112), pulling these residues away from the ATP binding site and forcing the
opening a deep yet narrow hydrophobic pocket, wherein the paralog GSK-3 residue Glu196 could lack
the ability to secure such interactions in the first place. Finer tuning of the occupation of this pocket using
ACS Paragon Plus Environment

Page 15 of 47
Journal of Medicinal Chemistry
1
2
3
4
subtle variation in halogen patterning, we reasoned, would allow for the possibility of achieving -isozyme
5
6
selectivity at high potency.
7
8
9
To assess this, we developed a subseries of derivatives leading to inhibitor 51 which gratifyingly
provided not only 65-fold improvement in potency compared to 44 (GSK-3 IC50 = 0.031 nM) but also
>10-fold selectivity for GSK-3 over GSK-3 (Fig. 4A,C,E-G, Fig. S2B). Similar results were obtained
using our confirmation [33P]-radiometric binding assays (Ki = 0.10 nM, >10-fold GSK-3 selectivity over
GSK-3, Fig. 4B,D). Increases in potency moving from 44 to 1 (via inhibitor OCM-47 (47)/49 (49)) could
be attributed to both the incorporation of gradually bulky aryl halides substituents (F<Cl<I) and the use of
an optimized 5-fluoropyridin-3-yl cation- interacting moiety (Fig. 4A,D). However, the -isozyme
selectivity occurred specifically with the iodoaryl substitution (Fig. 4E-G). We confirmed this effect in a
similar subseries wherein the cation- interacting fragment was kept untouched (PF-618 (28), OCM-45
(45) and OCM-32 (32) Fig. 4A,D, Table S3,4). As for previous compounds of this series, these inhibitors
displayed exquisite selectivity in broad kinome coverages (Fig. 4I). Inhibitor 51 demonstrated improved
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
selectivity even compared to previous inhibitors from this series (vide supra), with 10,000-fold selectivity
against all 371 kinases tested including the DYRK family. Further analysis of the potent inhibitor 49 as a
representative example of this series in Kinativ proteomic screening using mouse brain lysate revealed
extensive GSK-3 kinase engagement without any detectable off-target interactions (>215 kinases,
Fig. 4H). Notably, the Kinativ profiling used here included >40 CNS-enriched kinases not initially covered
in recombinant enzyme panels. We obtained dissociation constants from MST analysis in line with
biochemical data and showed, as previously noted, no quantifiable affinity differences between npS9-
GSK-3 and pS9-GSK-3 enzymes (51 Kd = 0.39 nM and Kd = 0.49 nM for npS9-GSK-3 and pS9-GSK-
3 respectively; Fig. 4L,M, Fig. S3A). Employing a high concentration cutoff of 10 M, we also found
negligible off-target binding for the representative inhibitor 51 in a CNS-related panel, with only 5 of 80
proteins (including receptors, ion channels and transporters) exhibiting >70% inhibition at this
concentration (Fig. 4J). Follow-up inhibitory constant measurements revealed >1000-20,000-fold gaps in
favor of GSK-3 in all cases providing further proof of selectivity (Fig. 4K). In addition, all representative
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 47
1
2
3
4
5
6
compounds tested (28, 44 and 51) display low hERG inhibition (EC50 = 1.7-62.6 M) and correspondingly
significant in vitro margins with regard to GSK-3 inhibition (8320-56,667-fold, Fig. S2C).
7
8
9
A
F
F
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
N
F
N
N
N
N
O
O
H
3CO
Cl
H
3CO
I
NH
NH
H311CO
Isozyme
Selectivity
Handle
O
O
H
3CO
Cl
NH
NH
N
N
3
I
N
N
O
O
O
O
OCM-49
OCM-50
GSK-3 IC50 = 0.8 nM
CNS MPO = 4.80
Fu_p = 0.03
OCM-32
[
12C]OCM-51/[11C]OCM-51
GSK-3 IC50 = 0.031 nM
CNS MPO = 4.80
GSK-3 IC50 = 0.4 nM
CNS MPO = 4.67
Fu_p = 0.04
GSK-3 IC50 = 0.6 nM
CNS MPO = 4.88
Fu_p = 0.14
Fu_p = 0.03
C
D
GSK-3 GSK-3α
Ki (nM)
GSK-3
Ki (nM) selectivity (fold)
g
g
GSK-3 (OCM-51)
Inhibitor
B
GSK-3 (OCM-51)
120
100
OCM-49
OCM-50
OCM-32
OCM-51
0.36
0.28
0.02
0.10
1.53
0.66
0.15
1.05
4
2
8
GSK-3 (PF-618)
g
GSK-3 (PF-618)
g
Ser78
Glu80
Arg113
Asp133
80
>10
60
40
20
0
10-fold
isozyme
selectivty
∙
Leu112
Leu130
E
Leu132
Glu97
Lys85
DMSO
Asp133
-12 -11 -10 -9 -8 -7 -6 -5
Log [Inhibitor] (M)
Asp200
OCM-51
GSK-3 (>95%)
GSK-3 (>95%)
OCM-51 / GSK-3
OCM-51
(1.0 M)
~10,000-fold
H
I
J
PIP4K2C
(35-50%)
F
5-HT2B
over GSK-3 Ki
120
100
80
GABA-gated-Cl-
NK1
TSPO
KOP
~100,000-fold
A3
CDK9 (cyclin T2) (41%)
DYRK2 (57%)
60
50% Inhibition at 10 M
over GSK-3 Ki
40
20
GSK-3 (96%)
GSK-3 (100%)
HGK/MAP4K4 (44%)
G
0
-20
-40
Y2
MINK/MINK1 (49%)
PAK3 (40%)
TNIK (54%)
100
0
% Inhibition
K
L
M
5
0
5
0
OCM-51
Ki (M)
GSK-3
selectivity (fold)
CNS target
-5
KOP
GABA-gated-Cl-
NK1
0.80
2.20
0.42
>7000
20,000
>3000
-5
-10
Kd GSK-3
K
d GSK-3
-15
-20
-25
(active - npS9)
= 0.39 nM
(inactive - pS9)
= 0.49 nM
-10
TSPO
5-HT2B
0.11
0.16
1000
>1000
-15
-12
-11
-10
-9
-8
-7
-6
-12
-11
-10
-9
-8
-7
-6
Log [OCM-51] (M)
Log [OCM-51] (M)
Fig. 4. Structure-guided design and characterization of the GSK-3 isozyme-selective inhibitors 51. (A)
Structures and properties of selected GSK-3 inhibitors leading to 51 (for details on chemical synthesis,
see Supplementary Schemes 14). IC50 values determined using an ADP-Glo kinase assay (n = 2). (B)
The dose-response curves of 51 versus GSK-3 and GSK-3 n = 3) (compared with 28, radiometric
filtration binding assay, [ATP] = Km ATP, data show mean values  SD). (C) Predicted binding mode of 51
(green sticks) with GSK-3 (gray cartoon, PDB ID code 5K5N and 5KPK). Highlight on the interaction
network associated with Asp133. Hydrogen bonds are depicted as a yellow dash lines. (D) Binding affinity
measurements of inhibitors versus GSK-3 and GSK-3 n = 3). (E-G) Alternate views of the predicted
ACS Paragon Plus Environment

Page 17 of 47
Journal of Medicinal Chemistry
1
2
3
4
contact between 51 and the selectivity pocket of GSK-3. (H) Kinativ data from mice brain lysates
5
6
7
8
showing the selective engagement of the representative inhibitor 49 (1.0 M) with GSK-3 (n = 2, for
detailed data, see Supplementary Table S5). (I) Comprehensive kinase selectivity profile of
representative inhibitor 51 against 371 kinases (kinases are ordered alphabetically and the activity is
relative to control at 1.0 M inhibitor concentration, n = 2, [γ-33P]ATP-based enzymatic assay performed
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
by Reaction Biology, for detailed data see Supplementary Table S9). (J) Selectivity profile of 51 (10.0 M)
against a CNS target panel (n = 2, 80 targets). (K) Binding affinity of 51 against the 5 most inhibited CNS
targets at a concentration 100,000-fold over the Ki of 51 for GSK-3. (L) MST dose-response curve and
binding affinity for the interaction of between 51 with active npS9-GSK-3 and pS9-GSK-3 (n = 2). (M)
MST dose-response curve and binding affinity for the interaction of between 51 with active npS9-GSK-3
and pS9-GSK-3 (n = 2). Data show mean values  SD.
Bioactivities of the Oxazole-4-Carboxamide Inhibitors in Cells.
We next explored the effect of oxazole-4-carboxamide inhibitors on endogenous cellular GSK-3
signaling. To define the cellular effect on pTau, we used high-content imaging of human cortical-like
glutamatergic neurons differentiated from induced pluripotent stem cells through induced expression of
Ngn2 transgene.18, 38 In agreement with high biochemical potency and cellular target engagement, we
found that the two novel representative compounds tested (28 and 51) caused dose-dependent
decreases in pTau levels with efficacy similar to 1 (Fig. 5A-D, Fig. S3C). We employed as negative
control the low potency, trifluoromethoxy-bearing structural analog OCM-29 (29; IC50 = 4.1 M, Fig.S3A)
and observed no appreciable decrease in pTau levels even with doses up to 30 M ( Fig. 5D). We further
evaluated the oxazole-4-carboxamide inhibitors in another neuronal assay, in which the phosphorylation
of collapsin response mediator protein 2 (CRMP2) at Thr-514 (pCRMP2T514) is measured using
immunostaining following 24-hour treatment with test compounds.39 CRMP2 is a direct protein substrate
of GSK-3, not GSK-3, and important for diverse neuronal functions.40, 41 Compounds were tested at 6
doses from 20 μM with 2-fold dilution. Robust dose-dependent effect with EC50 ranging from 0.5 – 1.5 μM
were detected for all active compounds, as well as CHIR99021 control (Fig. 5E-H,I, Fig. S3E,H). In
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 47
1
2
3
4
5
contrast, the low potency, negative control 29 failed to elicit detectable pCRMP2T514 reductions even at
the highest concentration tested (20 μM, Fig. 5H).
6
7
8
9
Given the role of GSK-3 in WNT/-catenin signaling, we also conducted Wnt signaling gene
transcription luciferase reporter assay (Fig. 5N)43. Relative to pTau and pCRMP2T514 inhibition,
significantly higher concentrations of active oxazole-4-carboxamide inhibitors were required to elicit gene
transcription through the WNT/-catenin pathway (>10-25-fold selectivity for pCRMP2T514 and 3-8-fold
selectivity for pTau). While 10 μM CHIR99021 was sufficient to induce 10-fold changes in reporter
induction (Fig. S3D), oxazole-4-carboxamide inhibitors treatment only triggered two- to three-fold changes
under the same conditions. GSK-3 signaling is also essential for coordinating the proliferation and
differentiation of progenitor cells during brain development.2, 42 We used human neural progenitor cells
(NPC) and a proliferation assay to evaluate the pro-neurogenic potential of these GSK-3 inhibitors. In line
with previous results, effect on cell proliferation were however minor and observed only at micromolar
concentrations for all active inhibitors (Fig. 5K,N, Fig. S3I,L). Together, these data indicate these oxazole-
4-carboxamide inhibitors can access endogenous GSK-3 in human neuronal cells and modulate functions
of various GSK-3 complexes with differential efficacy, delineating a potential therapeutic window wherein
the inhibition of GSK-3 leads to the modulation of key neuronal molecular targets while sparing putative
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
mechanism-based -catenin-driven effects.
ACS Paragon Plus Environment

Page 19 of 47
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
Fig. 5. Selective oxazole-4-carboxamide GSK-3 inhibitors modulate both pTau and pCRMP2T514 and
display favorable therapeutic index against Wnt/-catenin signaling in human neuronal cells. (A-D) Dose-
dependent effect of selected compounds on decreasing phosphorylated tau in human neurons. 20hr
treatment with 1, 28, 51, but not 29 dose-dependently decreased mean pTau (PHF-1) intensity in neuronal
cell bodies. Each bar is the mean values ± SEM of 14-21 fields/well. Each dose of each compound was
tested in 1 well per plate. Data from 3 plates in total are shown in the figure. (E-H) High-content imaging
assays using human neurons showed representative inhibitors (1, 28, 51) reduced CRMP2 phosphorylation
in a dose-dependent manner whereas 29 did not. CRMP2 is a direct substrate of GSK-3. Relative level of
pCRMP2T514 intensity was reported as the percentage of the intensity for DMSO-treated (100%) (n = 3
biological replicates, n = 18 images for analysis. Data show mean values  SEM). Relative levels of dendritic
marker MAP2 were not affected by the compound treatments, indicating the health of neurons. (I)
Representative images of 51 treatment on pCRMP2T514 intensity (green) versus MAP2 intensity (red)
illustrating the dose-dependent pCRMP2T514 intensity reduction. (J-M) Pro-neurogenic potentials measured
as NPC proliferation were detected for active oxazole-4-carboxamide GSK-3 inhibitors. Graphs show dose-
dependent increases from PF-367, PF-618 and 51 but not 29. NPC proliferation is expressed as the
percentage of DMSO-treated (100%). (n = 4. Error bars display SEM. Unpaired t test. * 0.01 ≤ p < 0.05, **
0.001 ≤ p < 0.01, *** 0.0001 ≤ p < 0.001, **** p < 0.0001.) (N) Active oxazole-4-carboxamide GSK-3
inhibitors enhanced Wnt/β-catenin signaling in human neural progenitor cells (NPCs). Dose-response
curves of representative inhibitors (1, 28, 51 and inactive 29) with calculated EC50s are shown. n = 4. Data
show mean values  SEM.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 47
1
2
3
4
5
6
Oxazole-4-Carboxamide Probes Display Increased Specific Binding in Tg2576 Mice Brains in vitro
and Robust BBB Permeation in vivo.
7
8
Given the implication of GSK-3 in APP processing and AD pathophysiology, we sought to determine in
9
vitro GSK-3-targeted radioligand binding in brain tissue from transgenic mice with significant amyloid
plaque burden. Remarkably, comparison between Tg2576 mice (15 month) and old wild-type animals (21
and 28 months) revealed significant differences in [3H]1 total and specific binding in brain tissue (Fig. 6A).
As previously, we carried out triple blocking experiments (10 M of 1, CHIR99021 or LY2090314) to
confirm specificity and selectivity (80-91% blocking) in these tissues (Fig. 6A). While we observed a
suitable total and specific binding level correlation between old control mice and other normal brain tissue
from rats, marmosets and humans (vide supra), we discovered important increases in total/specific
binding in Tg2576 mice compared to 21-month controls. Binding changes were drastic and most
pronounced in regions known to exhibit high amyloid plaque deposition such as the cortex (17-fold
increase) while binding in regions which present with lower or negligible plaque burden such as the
hippocampus (six-fold increase) or the caudate putamen (four-fold increase) respectively were
significantly less affected (Fig. 6C)44,45. In addition, regional distribution aligned well with previously
characterized ubiquitous tracer binding in the brains of all other species (Ctx  HP  CB  CPu  WM)
(Fig. 1).
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
With the dual goals of identifying putative GSK-3-targeted therapeutics for CNS diseases and
defining structural basis to achieve PET neuroimaging for this target, we coupled the initial SAR
component of this study with iterative micro-dosing rodent PET imaging experiments to gain insight into
structural factors associated with BBB permeation. As referred to previously (vide supra), this led to the
early realization that 6-membered nitrogen-containing replacements of the triazole ring found in 1, while
upholding or improving other properties from our lead, triggered significant increases in brain exposure.
Among selected inhibitors ([11C]PF-627 (20), [11C]28, [11C]37, [11C]44 and [11C]51, SI Materials and
Methods Scheme S5 and Figure S5-21), uptake of radioactivity in the rat brain following microdosed
peripheral injection (total dose = 0.19-1.75 g i.v.) was variable but in all cases non-negligible with whole
brain SUVpeak ranging from 0.84 to 3.08 followed by rapid washout (Fig. S4A). We observed the highest
ACS Paragon Plus Environment

Page 21 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
BBB permeation with [11C]28 and [11C]44 while the lowest radioactivity brain uptake was seen with the
most potent inhibitor, [11C]51, which may indicate substantial GSK-3 binding in the periphery prior to
reaching the CNS. Alternatively, this may be due to a faster clearance from the blood. Consistently,
regional brain radioactivity following administration was largely homogeneous as observed in vitro. Next,
to measure brain kinetics and determine brain exposure in higher species as a mean to assess the
potential of our inhibitors as therapeutics and radioligands, we performed micro-dosed PET imaging
experiments in non-human primates (one female rhesus macaque, total dose  1.1 g i.v.). The selected
radiolabeled inhibitors, [11C]28, [11C]37 and [11C]44, displayed in rhesus monkey both significant brain
exposure with SUVpeak of 3.0-4.0 SUV and favorable brain kinetic for imaging and CNS-targeted therapy
following injection (Fig. 2B,D, Fig. S4B,C). Radioactivity uptake was equally high in all gray matter regions
but lower in white matter. These results are in contrast with the in vivo profile of [11C]1, previously
reported and re-tested here, which demonstrate only low brain uptake in rhesus macaques (Fig. 6B,E).28
Finally, we conducted neuropharmacokinetic studies with our lead 44 to assess brain exposure over time.
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
Conform with microdosing data, we noted rapid (Tmax = 1.0 h) excellent brain exposure (unbound Cmax
941 nM) following single oral administration in mice (50 mg/kg p.o.; Fig. 6F). The unbound brain-to-
=
plasma ratio (Kp,uu, unbound AUClast ratios) was 1.06 and largely consistent between animals indicating
continued distribution equilibrium between brain and blood compartments and no efflux liability (Fig. 6F).
No clinical sign following oral administration of 44 at therapeutic dose in mice or micro-dosing in non-
human primates for imaging were observed and both procedures were well tolerated.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 47
1
2
3
4
sagittal
A
[
3H]PF-367+ PF-367
(10 M)
[
3H]PF-367 +
[
3H]PF-367 +
LY2090314 (10 M)
[
3H]PF-367 Ctx
CC
Hip
Cb

CHIR99021 (10
M)
Tg2576 mice
15 months
in vitro
max
5
min
Po
Th
CPu
6
7
[
3H]PF-367+ PF-367
M)
[
3H]PF-367 +
CHIR99021 (10
[
3H]PF-367 +
LY2090314 (10 M)
[
3H]PF-367
(10
M)
max
min
C57BL/6 mice
21 months
in vitro
8
9
3H]PF-367
[
3H]PF-367+ PF-367
3H]-PF367_wild-type mice
21 month
[
3H]PF-367 +
CHIR99021 (10
[
3H]PF-367 +
LY2090314 (10 M)
[
[
(10
M)
M)
max
min
C57BL/6 mice
28 months
in vitro
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
5-30 min summed
11C]OCM-37
B
[
11C]PF-367
CC
[
11C]PF-618
CC
[
[
11C]OCM-44
CC
Ctx
Ctx
Ctx
Ctx
CC
2
0
Rhesus macaque
in vivo
(sagittal)
Cb
Cb
11C]PF-618
Cb
11C]OCM-37
Cb
11C]OCM-44
Th
Th
Th
Th
[
11C]PF-367
[
[
[
2
0
Rhesus macaque
in vivo
(coronal)
[
11C]PF-367
[
11C]PF-618
[
11C]OCM-37
[11C]OCM-44
2
0
Rhesus macaque
in vivo
(transverse)
P
≤ 0.0006
Whole brain
Cortex
White matter
Cerebellum
Thalamus
Tg2576
E
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
80
70
60
50
40
30
20
10
0
C
D
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
o Tg2576 (15 months)
o C57BL/6 (28 months)
o C57BL/6 (21 months)
0
10
20
30
40
50
60
Time (min)
0
10
20
30
40
50
60
Time (min)
Plasma (total)
Brain (total)
CSF
Free Plasma
Free Brain
IC50
F
100000
10000
1000
100
10
1
0.1
0
1
2
3
4
5
6
7
8
Time (h)
Fig. 6. Oxazole-4-carboxamide-based probes display robust brain exposure properties in vivo and
differential target engagement in a mouse model of AD in vitro. (A) In vitro autoradiography images of [3H]1
Tg2576 and C57BL/6 mice under baseline, homologous blocking (nonradioactive 1, 10 M) and
heterologous (LY2090314, 10 M or CHIR99021, 10 M) conditions. (B) Summed (5 to 30 min) PET
images showing the distribution of [11C]28, [11C]37 and [11C]44 in comparison with [11C]1 in the rhesus
monkey brain following microdose injections. Images intensity displayed as 0-2 SUV scale. (C)
Quantification for the in vitro autoradiography of [3H]1 in Tg2576 (20 m sections, n = 3) and C57BL/6 (20
m sections, n = 2) mice brains under baseline, homologous blocking (nonradioactive 1, 10 M) and
heterologous (LY2090314, 10 M or CHIR99021, 10 M) conditions. (D) Representative rhesus monkey
regional brain time-activity curves (TACs) after intravenous injection of [11C]44. (E) Representative rhesus
monkey regional brain time-activity curves (TACs) after intravenous injection of [11C]1. (F) Mice exposure
of 44 (C57BL/6 mice, 50 mg/kg, single dose PO, n = 3). Data show mean values  SD.
ACS Paragon Plus Environment

Page 23 of 47
Journal of Medicinal Chemistry
1
2
3
4
Discussion and Conclusions
5
6
7
8
9
Modulating molecular targets efficiently and with high accuracy, particularly in the brain, where the BBB
severely limits the capacity of peripherally administered small molecules to achieve target engagement, is
a significant challenge in CNS drug discovery. To date, the roles of GSK-3 in neurodevelopmental and
neurodegenerative diseases have been difficult to probe pharmacologically owing to the complex nature
of GSK-3-mediated cellular pathways and the lack of suitable inhibitors. The data presented here
describe brain penetrant selective inhibitors, which potently engage and inhibit GSK-3 in mammalian
brain tissue, including human brains, and modulate GSK-3-regulated neuronal pathways in cells. We also
define the structural underpinning for achieving picomolar selective GSK-3 inhibition as well as -
isozyme-specificity exploiting discrete topological variations from the deep ATP binding site of GSK-3
isozymes as previously noted by Wagner et al. 2018.32 While exhibiting GSK-3 selectivity in ranges
described for the few previously known GSK-3 isozyme selective inhibitors, the compounds described
here do so at over two orders of magnitude higher potencies – which may be crucial in achieving a
favorable therapeutic window for this target.32, 26 Whether the maintenance of about the same fold-
differences in isozyme selectivity observed here with double-digit picomolar versus previously described
double-digit nanomolar potency inhibitors represent an intrinsic limitation ensuing from GSK-3 binding
sites dissimilarities, or isozyme selectivity can be improved further, will require further detailed SAR work
based on the results presented here. Previous studies have shown that deletion in embryonic stem cells
of GSK-3 or GSK-3 do not individually alter -catenin/TCF-mediated transcription and that isoform-
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
selective pharmacological inhibition of GSK-3 in acute myeloid leukemia cells prevents Wnt/-catenin
changes at specific concentrations where dual inhibition lacks such ability – thus providing strong
rationale for isozyme-selective inhibition in mitigating putative mechanism-based toxicity driven by -
catenin translocation.9, 32 In the current study, the lack of dual inhibitors equipotent to 51 (or 32) has
impeded the determination of such effect directly. It is important to note however, that inhibitors 51 and 32
only trigger Wnt/-catenin transcriptional changes at higher concentrations than dual and drastically lower
potency controls (1 and CHIR99021). This suggests that the GSK-3-isozyme selective inhibitors
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 47
1
2
3
4
5
6
7
8
9
identified here may in fact cause similar shifts wherein GSK-3 inhibition is achieved while sparing Wnt/-
catenin. Irrespective of this effect, we note for all inhibitors identified here, a significant and favorable gap
between the doses required to trigger Wnt/β-catenin signaling changes and modulate p-CRMP2T514 and
pTau levels.
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
Kinase-targeted in vivo noninvasive neuroimaging to explore intractable neurodegenerative and
psychiatric conditions, constitute an unmet objective despite tremendous progress in the therapeutic and
imaging use of kinase inhibitors outside the CNS.46, 47 Our preclinical data demonstrate that radiolabeled
oxazole-4-carboxamide inhibitors such as [11C]28 and [11C]44 align suitable physicochemical parameters,
kinetics and in vitro/in vivo properties to serve as leads to achieve GSK-3 PET imaging.48 Our analysis of
GSK-3 brain density further supports the view that this target can be imaged using PET. Findings
indicating that oxazole-4-carboxamide probes provide total GSK-3 density readouts and display severe
binding increases in AD mice compared to wild-type animals also provide a basis for studying the
potential of these probes to measure changes in GSK-3 expression in neurodegeneration. A possible
caveat pertaining to the characterization of the activation-independent binding mode of the (radio)ligands
described here is the report of an alternative p38 MAPK-mediated GSK-3 inactivation phosphorylation at
the S389 C terminus residue.48 The relevance of this inactivation pathway has however only been
characterized in mice and may not be relevant in human. Based on the results presented here, detailed
PET quantification studies are currently underway. The major finding of this work is the discovery of orally
available inhibitors with unique pharmacological profiles showing robust GSK-3 target engagement in cell
and tissue as well as high brain exposure which warrant future intervention and imaging investigations.
ACS Paragon Plus Environment

Page 25 of 47
Journal of Medicinal Chemistry
1
2
3
4
Experimental Section
5
6
Chemical Synthesis. Compound PF-04802367 ( 98%) (hereafter called 1), CHIR99021 ( 98%) and
7
8
9
LY2090314 ( 98%) were purchased from Millipore Sigma. Alternatively, 1 was synthesized as described
previously.28 Detailed synthetic procedures for all inhibitors are provided in Scheme S1-S3 and SI
Materials and Methods. Briefly, the lead inhibitor (target compound), 5-(3-fluoro-4-methoxyphenyl)-N-(3-
(pyridin-3-yl)propyl)oxazole-4-carboxamide (44, OCM-44) was synthesized as follows: The title
compound was synthesized from 5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.2 g, 0.84
mmol) and 3-(pyridin-3-yl)propan-1-amine (0.137 g, 1.00 mmol) to give 44 (0.145 g, 48%) as an off white
solid. LCMS (ESI) m/z = 356.10 [M+H]+. HRMS (ESI+): m/z calc. for C19H19FN3O3+ (M + H)+: 356.1410,
found 356.1402. 1H NMR (400 MHz, DMSO-d6) δ 8.55 (s, 1H), 8.49 (t, J = 5.60 Hz, 1H), 8.45 (d, J = 2.00
Hz, 1H), 8.39 (dd, J = 1.40, 4.80 Hz, 1H), 8.35 (dd, J = 2.20, 13.20 Hz, 1H), 7.97 (d, J = 8.40 Hz, 1H),
7.66-7.64 (m, 1H), 7.32-7.28 (m, 2H), 3.91 (s, 3H), 3.34-3.26 (m, 2H), 2.63 (t, J = 7.80 Hz, 2H), 1.84
(quin, J = 7.34 Hz, 2H).13C NMR (100 MHz, DMSO-d6) δ 161.09, 160.65, 150.78 (d, J = 241.3 Hz),
149.80, 149.74, 148.92, 148.30 (d, J = 10.6 Hz), 136.38, 128.31, 124.44 (d, J = 3 Hz), 122.67, 121.15,
119.68 (d, J = 8.1 Hz), 115.49 (d, J = 21.5 Hz), 113.64, 56.08, 38.20, 30.38, 29.58. Purity: HPLC tR = 5.55
min, HPLC purity > 95.4%.
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
All final compounds were determined to have purity >95% by HPLC. A X-select CSH C18 (4.6 x 150 mm
3.5 m) HPLC column was used to determine the chemical purity. Radiosynthesis quality control was
performed on a Phenomenex Luna C18, 5 mm, 4.6 x 150 mm column. Liquid chromatography mass
spectroscopy utilized a X-select CSH C18 (3.0 x 50mm 2.5 m) column.
General procedure for Synthesis of Inhibitors 6-37, 39-42 and 44-51 (OCM-6 - OCM-37, OCM-39 -
OCM-42 and OCM-44 - OCM-51)
To a mixture of an appropriate carboxylic acid (74-84, 1 eq), the corresponding primary amine (1.2 eq)
and HATU (1.2 eq) in DMF (10 vol) was added TEA (5 eq) in dropwise manner. The reaction mixture was
stirred at room temperature for 2 h. The progress of the reaction was monitored by TLC (50% EtOAc in
hexane). After completion, the reaction mass was taken into water and extracted with EtOAc, combined
organic extracts were dried over anhydrous sodium sulfate, filtered and concentrated in vacuum to get
crude mass. Unless otherwise noted, purification by reverse phase column chromatography using
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 47
1
2
3
4
5
6
acetonitrile-water as mobile phase was used to obtain purified final products which were then triturated
with pentane and filtered dried under vacuum to afford 6-37, 39-42 and 44-51 (OCM-6 - OCM-37, OCM-
39 - OCM-42 and OCM-44 - OCM-51).
7
8
9
N-(3-(1H-1,2,4-triazol-1-yl)propyl)-5-(3,4-difluorophenyl)oxazole-4-carboxamide (6, OCM-6). The title
compound was synthesized from 5-(3,4-difluorophenyl)oxazole-4-carboxylic acid (0.15 g, 0.66 mmol) and
3-(1H-1,2,4-triazol-1-yl)propan-1-amine (0.10 g, 0.80 mmol) to give 6 (0.12 g, 51%) as an off white solid.
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
+
LCMS (ESI) m/z = 334.10 [M+H]+. HRMS (ESI+): m/z calc. for C15H14F2N5O2 (M + H)+: 334.1116, found
334.1112. 1H NMR (400 MHz, DMSO-d6) δ 8.66-8.64 (m, 2H), 8.56-8.50 (m, 2H), 8.04-8.02 (m, 1H), 7.96
(s, 1H), 7.63-7.56 (m, 1H), 4.22 (t, J = 6.80 Hz, 2H), 3.27 (q, J = 6.40 Hz, 2H), 2.04 (quin, J = 6.80 Hz, 2H).
13C NMR (100 MHz, DMSO-d6) δ 160.62, 151.37, 150.43 (m), 148.96 (m),148.76 (m), 147.83 (m), 144.05,
129.36, 124.93 (m), 124.3 (dd, J = 11 and 3.48 Hz), 117.83 (d, J = 17.3 Hz), 117.27 (d, J = 20.6 Hz), 46.52,
36.00, 29.39. Purity: HPLC tR = 7.17 min, HPLC purity > 99.7%.
N-(3-(1H-1,2,4-triazol-1-yl)propyl)-5-(perfluorophenyl)oxazole-4-carboxamide (7, OCM-7). The title
compound was synthesized from 5-(perfluorophenyl)oxazole-4-carboxylic acid (0.4 g, 1.43 mmol) and 3-
(1H-1,2,4-triazol-1-yl)propan-1-amine (0.108 g, 0.85 mmol) to give 7 (0.06 g, 11%) as an pink colored solid
+
after SFC purification. LCMS (ESI) m/z = 388.4 [M+H]+. HRMS (ESI+): m/z calc. for C15H11F5N5O2 (M +
H)+: 388.0833, found 388.0825. 1H NMR (400 MHz, CHLOROFORM-d) δ 8.14 (s, 1H), 8.05 (s, 1H), 7.96
(s, 1H), 7.34 (bs, 1H), 4.28 (t, J = 6.40 Hz, 2H), 3.47 (q, J = 6.8 Hz, 2H), 2.20 (quin, J = 6.80 Hz, 2H).
13C NMR not assigned. Purity: HPLC tR = 7.21 min, HPLC purity > 96.5%.
N-(3-(1H-1,2,4-triazol-1-yl)propyl)-5-(3,4,5-trifluorophenyl)oxazole-4-carboxamide (8, OCM-8). The
title compound was synthesized from 5-(3,4,5-trifluorophenyl)oxazole-4-carboxylic acid (0.1 g, 0.41 mmol)
and 3-(1H-1,2,4-triazol-1-yl)propan-1-amine (0.06 g, 0.48 mmol) to give 8 (0.058 g, 41%) as an off white
+
solid. LCMS (ESI) m/z = 352.10 [M+H]+. HRMS (ESI+): m/z calc. for C15H13F3N5O2 (M + H)+: 352.1021,
1
found 352.1020. H NMR (400 MHz, CHLOROFORM-d) δ 8.19-8.14 (m, 3H), 7.95 (s, 1H), 7.85 (s, 1H),
7.53 (br s, 1H), 4.28 (t, J = 6.40 Hz, 2H), 3.49 (q, J = 6.40 Hz, 2H), 2.21 (quin, J = 6.80 Hz, 2H).
13C NMR (100 MHz, CDCl3) δ 160.96, 152.18 (m), 149.75 (m), 148.23 (m), 143.29, 141.96, 139.48 (d, J =
15.7 Hz), 129.65, 122.62(m), 112.69 (m), 47.06, 36.35, 29.92. Purity: HPLC tR = 7.55 min, HPLC purity >
99.6%.
N-(3-(1H-1,2,4-triazol-1-yl)propyl)-5-(3-fluoro-4-(trifluoromethoxy)phenyl)oxazole-4-carboxamide (9,
OCM-9). The title compound was synthesized from 5-(3-fluoro-4-(trifluoromethoxy)phenyl)oxazole-4-
carboxylic acid (0.15 g, 0.51 mmol) and 3-(1H-1,2,4-triazol-1-yl)propan-1-amine (0.09 g, 0.61 mmol) to give
9 (0.090 g, 44%) as an off white solid. LCMS (ESI) m/z = 400.15 [M+H]+. HRMS (ESI+): m/z calc. for
C16H14F4N5O3+ (M + H)+: 400.1033, found 400.1026. 1H NMR (400 MHz, DMSO-d6) δ 8.69 (br. s, 2H), 8.54
(bs, 2H), 8.11 (br s, 1H), 7.97 (br s, 1H), 7.74 (br. s, 1H), 4.23 (bs, 2H), 3.30-3.21 (m, 2H), 2.05 (br s, 2H).
13C NMR not assigned. Purity: HPLC tR = 8.05 min, HPLC purity > 98.2%.
ACS Paragon Plus Environment

Page 27 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
N-(3-(1H-1,2,4-triazol-1-yl)propyl)-5-(3-fluoro-4-(trifluoromethyl)phenyl)oxazole-4-carboxamide (10,
OCM-10). The title compound was synthesized from 5-(3-fluoro-4-(trifluoromethyl)phenyl)oxazole-4-
carboxylic acid (0.2 g, 0.73 mmol) and 3-(1H-1,2,4-triazol-1-yl)propan-1-amine (0.11 g, 0.87 mmol) to give
10 (0.09 g, 33%) as an off white solid. LCMS (ESI) m/z = 384.15 [M+H]+. HRMS (ESI+): m/z calc. for
C16H14F4N5O2+ (M + H)+: 384.1084, found 384.1068. 1H NMR (400 MHz, CHLOROFORM-d) δ 8.37 (d, J =
11.6 Hz, 1H), 8.26 (d, J = 8.0 Hz, 1H), 8.18 (s, 1H), 7.99 (s, 1H), 7.94 (s, 1H), 7.73-7.69 (m, 1H), 7.59 (br.
s, 1H), 4.32 (t, J = 6.80 Hz, 2H), 3.53 (q, J = 6.40 Hz, 2H), 2.25 (quin, J = 6.40 Hz, 2H). 13C NMR (100 MHz,
CDCl3) δ 160.85, 158.26, 152.20, 149.96, 148.75 (d, J = 2.9 Hz), 143.30, 132.32 (d, J = 9.3 Hz), 130.68,
127.29, 123.54 (m), 120.92, 119.30 (dd, J = 45.5 and 12.6 Hz), 116.60 (d, J =24.5 Hz), 47.07, 36.39, 29.93.
Purity: HPLC tR = 7.90 min, HPLC purity > 99.4%.
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
N-(3-(1H-1,2,4-triazol-1-yl)propyl)-5-(3,5-difluoro-4-methoxyphenyl)oxazole-4-carboxamide
(11,
OCM-11).
The title compound was synthesized from 5-(3,5-difluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.2
g, 0.78 mmol) and 3-(1H-1,2,4-triazol-1-yl)propan-1-amine (0.118 g, 0.93 mmol) to give 11 (0.09 g, 32%)
as an off white solid. LCMS (ESI) m/z = 386.25 [M+Na]+. HRMS (ESI+): m/z calc. for C16H16F2N5O3+ (M +
H)+: 364.1221, found 364.1214. 1H NMR (400 MHz, DMSO-d6) δ 8.65-8.69 (m, 1H), 8.64 (s, 1H), 8.53 (s,
1H), 8.19-8.10 (m, 2H), 7.96 (s, 1H), 4.21 (t, J = 6.80 Hz, 2H), 4.01 (s, 3H), 3.27 (q, J = 6.40 Hz, 2H), 1.98-
2.09 (m, 2H).13C NMR (100 MHz, DMSO-d6) δ 160.58, 154.39 (dd, J = 251 Hz and 6.8 Hz), 151.39, 150.67,
150.40, 148.64, 144.07, 129.52, 121.59 (m), 112.08 (m), 61.76, 46.52, 36.03, 29.38. Purity: HPLC tR = 7.38
min, HPLC purity > 96.6%.
N-(3-(1H-1,2,4-triazol-1-yl)propyl)-5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxamide (12, OCM-
12). The title compound was synthesized from 5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.2
g, 0.84 mmol) and 3-(1H-1,2,4-triazol-1-yl)propan-1-amine (0.127 g, 1.00 mmol) to give 12 (0.09 g, 31%)
as an off-white solid. LCMS (ESI) m/z = 346.25 [M+H]+. HRMS (ESI+): m/z calc. for C16H17FN5O3+ (M + H)+:
346.1315, found 346.1302. 1H NMR (400 MHz, DMSO-d6) δ 8.59-8.51 (m, 3H), 8.36-8.33 (m, 1H), 7.98 (s,
1H), 7.96 (s, 1H), 7.30 (t, J = 8.80 Hz, 1H), 4.22 (t, J = 6.80 Hz, 2H), 3.91 (s, 3H), 3.22-3.31 (m, 2H), 2.04
(quin, J = 6.80 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 160.87, 151.39, 150.78 (d, J =241.3 Hz), 149.99,
149.80, 148.36 (d, J =10.6 Hz), 144.04, 128.15, 124.51, 119.63 (d, J =7.6 Hz), 115.53 (d, J =21.5 Hz),
113.65, 56.13, 46.54, 35.96, 29.48. Purity: HPLC tR = 6.95 min, HPLC purity > 96.7%.
(Z)-5-(3-fluoro-4-methoxyphenyl)-N-(3-phenylallyl)oxazole-4-carboxamide (13, OCM-13). The title
compound was synthesized from 5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.2 g, 0.84 mmol)
and (Z)-3-phenylprop-2-en-1-amine (0.134 g, 1.01 mmol) to get 13 (0.040 g, 13%) as an off white solid.
+
LCMS (ESI) m/z = 353.15 [M+H]+. HRMS (ESI+): m/z calc. for C20H18FN2O3 (M + H)+: 353.1301, found
353.1294. 1H NMR (400 MHz, CDCl3) δ 8.21-8.18 (m, 2H), 7.81 (s, 1H), 7.40-7.28 (m, 5H), 7.04 (t, J = 8.8
Hz, 1H), 6.65 (d, J = 11.6 Hz, 1H), 5.83-5.77 (m, 1H), 4.38 (t, J = 6 Hz, 2H), 3.96 (s, 3H). 13C NMR (100
MHz, CDCl3) δ 160.95, 151.84 (d, J = 243.8 Hz), 151.73, 149.12 (d, J = 10.6 Hz), 147.48 (d, J = 4.6 Hz),
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 47
1
2
3
4
5
136.29, 131.85, 128.76, 128.36, 128.03, 127.75, 127.29, 124.91, 119.90 (d, J = 7.6 Hz), 116.04 (d, J = 22
Hz), 112.83, 56.22, 37.68. Purity: HPLC tR = 8.99 min, HPLC purity > 97.0%.
6
7
8
9
N-cinnamyl-5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxamide (14, OCM-14). The title compound
was synthesized from 5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.2 g, 0.84 mmol) and (E)-
3-phenylprop-2-en-1-amine (0.134 g, 1.01 mmol) to get 14 (0.065 g, 21%) as an off white solid. LCMS (ESI)
m/z = 353.20 [M+H]+. HRMS (ESI+): m/z calc. for C20H18FN2O3+ (M + H)+: 353.1301, found 353.1287. 1H
NMR (400 MHz, CDCl3) δ 8.22-8.18 (m, 2H), 7.82 (s, 1H), 7.48-7.23 (m, 6H), 7.05 (t, J = 8.4 Hz, 1H), 6.63
(d, J = 16 Hz, 1H), 6.34-6.27 (m, 1H), 4.25 (t, J = 6 Hz, 2H), 3.96 (s, 3H).13C NMR (100 MHz, CDCl3) δ
160.92, 151.87 (d, J = 243.9 Hz), 151.81, 149.15 (d, J = 10.5 Hz), 147.50 (m), 136.55, 132.26, 128.56,
128.09 127.70, 126.41, 125.32, 124.97 (d, J = 3.4 Hz), 119.93 (d, J = 7.6 Hz), 116.08 (d, J = 19.5 Hz),
112.86, 56.24, 41.26. Purity: HPLC tR = 9.31 min, HPLC purity > 99.4%.
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
5-(3-fluoro-4-methoxyphenyl)-N-phenethyloxazole-4-carboxamide (15, OCM-15). The title compound
was synthesized from 5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.1 g, 0.42 mmol) and (2-
phenylethan-1-amine (0.061 g, 0.504 mmol) to get 15 (0.110 g, 76%) as an off white solid. LCMS (ESI) m/z
= 341.10 [M+H]+. HRMS (ESI+): m/z calc. for C19H18FN2O3+ (M + H)+: 341.1301, found 341.1295.1H NMR
(400 MHz, DMSO-d6) δ 8.53 (s, 1H), 8.44-8.41 (m, 1H), 8.33 (dd, J = 1.6 Hz & 13.2 Hz, 1H), 7.95 (d, J =
8.8 Hz, 1H), 7.32-7.18 (m, 6H), 3.91 (s, 3H), 3.53-3.48 (m, 2H), 2.84 (t, J = 7.2 Hz, 2H).13C NMR (100 MHz,
DMSO-d6) δ 160.57, 150.77 (d, J = 241.3 Hz), 149.88, 149.76, 148.33 (d, J = 10.5 Hz), 139.33, 128.59,
128.32, 128.18, 126.08, 124.44, 119.60 (d, J = 8 Hz), 115.47 (d, J = 21.5 Hz), 113.64, 56.09, 35.08. Purity:
HPLC tR = 9.26 min, HPLC purity > 99.6%.
5-(3-fluoro-4-methoxyphenyl)-N-((1R,2R)-2-phenylcyclopropyl)oxazole-4-carboxamide and 5-(3-
fluoro-4-methoxyphenyl)-N-((1S,2S)-2-phenylcyclopropyl)oxazole-4-carboxamide (16, OCM-16). The
title compound was synthesized from 5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.2 g, 0.84
mmol) and rac-cis-2-phenylcyclopropan-1-amine (0.134 g, 1.01 mmol) to get 16 (0.110 g, 36%) as an off
+
white solid. LCMS (ESI) m/z = 353.10 [M+H]+. HRMS (ESI+): m/z calc. for C20H18FN2O3 (M + H)+:
353.1301, found 353.1290. 1H NMR (400 MHz, CDCl3) δ 8.16-8.13 (m, 2H), 7.65 (s, 1H), 7.35-7.22 (m, 5H),
7.04-6.97 (m, 2H), 3.95 (s, 3H), 3.34-3.28 (m, 1H), 2.47-2.41 (m, 1H), 1.53-1.47 (m, 1H), 1.20-1.16 (m,
1H).13C NMR (100 MHz, CDCl3) δ 162.39, 151.81 (J = 242 Hz), 151.38, 149.05 (d, J = 11 Hz), 147.33 (d, J
= 5 Hz), 136.24, 128.69, 128.36, 127.86, 126.58, 124.80, 119.91 (d, J = 8 Hz), 116.01 (d, J = 20 Hz),
112.79, 56.19, 28.57, 21.85, 12.71. Purity: HPLC tR = 8.96 min, HPLC purity > 99.8%.
5-(3-fluoro-4-methoxyphenyl)-N-((1S,2R)-2-phenylcyclopropyl)oxazole-4-carboxamide and 5-(3-
fluoro-4-methoxyphenyl)-N-((1R,2S)-2-phenylcyclopropyl)oxazole-4-carboxamide (17, OCM-17). The
title compound was synthesized from 5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.2 g, 0.84
mmol) and rac-trans-2-phenylcyclopropan-1-amine (0.134 g, 1 mmol) to get 17 (0.100 g, 33%) as an off
+
white solid. LCMS (ESI) m/z = 353.00 [M+H]+. HRMS (ESI+): m/z calc. for C20H18FN2O3 (M + H)+:
353.1301, found 353.1297. 1H NMR (400 MHz, CDCl3) δ 8.25-8.17 (m, 2H), 7.81 (s, 1H), 7.47 (s, 1H), 7.33-
7.29 (m, 2H), 7.23-7.20 (m, 3H), 7.05 (t, J = 2.2 Hz, 1H), 3.96 (s, 3H), 3.15-3.10 (m, 1H), 2.24-2.19 (m, 1H),
ACS Paragon Plus Environment

Page 29 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
1.40-1.31 (m, 2H).13C NMR (100 MHz, CDCl3) δ 162.16, 151.86 (d, J = 244.2 Hz), 151.77, 149.18 (d, J =
10.6 Hz), 147.43 (d, J = 4.3 Hz), 140.41, 128.41, 127.91, 126.42, 126.16, 125.01, 119.83 (d, J = 8.1 Hz),
116.02 (d, J = 21.5 Hz), 112.86, 56.23, 31.88, 24.70, 16.38. Purity: HPLC tR = 9.31 min, HPLC purity >
99.9%.
5-(3-fluoro-4-methoxyphenyl)-N-(3-phenylpropyl)oxazole-4-carboxamide (18, OCM-18). The title
compound was synthesized from 5-(3-fluoro-4-methoxyphenyl)oxazole-4-carboxylic acid (0.2 g, 0.84 mmol)
using commercial available amine 3-phenylpropan-1-amin (0.136 g, 1.01 mmol) to give 18 (0.09 g, 31%)
as an off white solid. LCMS (ESI) m/z = 355.10 [M+H]+. HRMS (ESI+): m/z calc. for C20H20FN2O3+ (M + H)+:
355.1458, found 355.1453. 1H NMR (400 MHz, DMSO-d6) δ 8.54 (s, 1H), 8.47-8.44 (m, 1H), 8.36-8.32 (m,
1H), 7.96 (d, J = 8.8 Hz, 1H), 7.31-7.25 (m, 3H), 7.20-7.14 (m, 3H), 3.89 (s, 3H), 3.32-3.26 (m, 2H), 2.6 (t,
J = 7.20 Hz, 2H), 1.86-1.78 (m, 2H).13C NMR (100 MHz, CDCl3) δ 161.09, 151.84 (d, J = 243.6 Hz), 151.58,
149.08 (d, J = 10.9 Hz), 147.45, 141.35, 128.44, 128.37 128.21, 125.96, 124.93 (d, J = 3.4 Hz), 119.98 (d,
J = 7.5 Hz), 116.06 (d, J = 21.5 Hz), 112.83, 56.21, 38.87, 33.27, 31.21. Purity: HPLC tR = 8.83 min, HPLC
purity > 99.4%.
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
5-(3-fluoro-4-methoxyphenyl)-N-(3-(2-methyl-1H-imidazol-1-yl)propyl)oxazole-4-carboxamide (19,
PF-981). PF-981 was graciously provided by Pfizer (HPLC purity > 95%). Physical State: White solid.
HRMS (ESI+): m/z calc. for C18H20FN4O3+ (M + H)+: 359.1519, found 359.1514. 1H NMR (400 MHz, CDCl3)
δ 8.14 (dt, J = 2.1, 10.8Hz, 2H), 7.82 (s, 1H), 7.43 (t, J = 5.8 Hz, 1H), 7.12 (d, J = 6.4 Hz, 2H), 7.05 (t, J =
8.5 Hz, 1H), 4.05 (t, J = 7.1 Hz, 2H), 3.95 (s, 3H), 3.51 (m, 2H), 2.62 (s, 3H), 2.12 (t, J = 6.8 Hz, 2H). 13
C
NMR (100 MHz, CDCl3) δ 161.8, 153.1-150.6 (d, J = 245.5 Hz), 152.1 (d, J = 2.7 Hz), 149.3 (d, J = 10.8
Hz), 147.7, 144.1, 127.5, 124.9 (d, J = 3.6 Hz), 122.3, 119.9, 119.5 (d, J = 7.9 Hz), 116.0 (d, J = 21.6 Hz),
112.9, 56.3, 44.5, 36.1, 30.7, 11.7.
5-(3-fluoro-4-methoxyphenyl)-N-(3-(2-oxopyridin-1(2H)-yl)propyl)oxazole-4-carboxamide (20, PF-
627). PF-627 was graciously provided by Pfizer (HPLC purity > 95%). Physical State: White solid. HRMS
(ESI+): m/z calc. for C19H19FN3O4+ (M + H)+: 372.1360, found 372.1367. 1H NMR (400 MHz, CDCl3) δ 8.21
(dd, J = 12.8, 2.1 Hz, 1H), 8.12 (d, J = 8.8 Hz, 1H), 8.1 (s, 1H), 7.84 (s, 1H), 7.55 (d, J = 7.6 Hz, 1H), 7.49
(d, J = 7.7 Hz, 1H), 7.09-6.99 (m, 2H), 6.45 (t, J = 6.3 Hz, 1H), 4.17 (t, J = 6.2 Hz, 2H), 3.94 (s, 3H), 3.47
(m, 2H), 2.08 (t, J = 6.1 Hz, 2H). 13C NMR (100 MHz, CDCl3) δ 163.1, 161.5, 153.1-150.6 (d, J = 245.0 Hz),
151.5 (d, J = 2.4 Hz), 149.0 (d, J = 10.8 Hz), 147.8, 140.4, 137.7, 128.3, 124.8 (d, J = 3.6 Hz), 120.56,
120.1 (d, J = 7.9 Hz), 116.2 (d, J = 21.7 Hz), 112.8, 107.9, 56.2, 47.3, 35.4, 29.8.
5-(3-fluoro-4-methoxyphenyl)-N-(3-(1-methyl-1H-pyrazol-4-yl)propyl)oxazole-4-carboxamide (21, PF-024).
PF-024 was graciously provided by Pfizer (HPLC purity > 95%). Physical State: White solid. HRMS (ESI+):
+
1
m/z calc. for C18H20FN4O3 (M + H)+: 359.1519, found 359.1530. H NMR (400 MHz, CDCl3) δ 8.20-8.13
(m, 2H), 7.79 (s, 1H), 7.40 (s, 1H), 7.31 (br s, 1H), 7.02 (t, J = 8.5 Hz, 1H) 3.94 (s, 3H), 3.91 (s, 3H), 3.46
(m, 2H), 2.57 (t, J = 7.5 Hz, 2H), 1.89 (t, J = 7.3 Hz, 2H). 13C NMR (100 MHz, CDCl3) δ 161.1, 153.1-150.6
(d, J = 245.1 Hz), 151.6 (d, J = 2.4 Hz), 149.1 (d, J = 10.8 Hz), 147.5, 137.5, 129.0, 128.2, 124.9 (d, J = 3.6
Hz), 121.0, 119.9 (d, J = 7.9 Hz), 116.1 (d, J = 21.7 Hz), 112.8 (d, J = 2.1 Hz), 56.2, 38.7, 38.6, 30.8, 21.5.
ACS Paragon Plus Environment