Discovery of a Potent, Orally Bioavailable PI4KIIIβ Inhibitor (UCB9608) Able to Significantly Prolong Allogeneic Organ Engraftment in Vivo

Article abstract of DOI:10.1021/acs.jmedchem.8b00521

The primary target of a novel series of immunosuppressive 7-piperazin-1-ylthiazolo[5,4-d]pyrimidin-5-amines was identified as the lipid kinase, PI4KIIIβ. Evaluation of the series highlighted their poor solubility and unwanted off-target activities. A medicinal chemistry strategy was put in place to optimize physicochemical properties within the series, while maintaining potency and improving selectivity over other lipid kinases. Compound 22 was initially identified and profiled in vivo, before further modifications led to the discovery of 44 (UCB9608), a vastly more soluble, selective compound with improved metabolic stability and excellent pharmacokinetic profile. A co-crystal structure of 44 with PI4KIIIβ was solved, confirming the binding mode of this class of inhibitor. The much-improved in vivo profile of 44 positions it as an ideal tool compound to further establish the link between PI4KIIIβ inhibition and prolonged allogeneic organ engraftment, and suppression of immune responses in vivo.

Full text of DOI:10.1021/acs.jmedchem.8b00521

Subscriber access provided by FORDHAM UNIVERSITY
Article
Discovery of a potent, orally bioavailable PI4KIII# inhibitor (UCB9608)
able to significantly prolong allogeneic organ engraftment in vivo
James Reuberson, Helen Horsley, Richard J Franklin, Daniel Ford, Judi Neuss, Daniel Brookings, Qiuya
Huang, Bart Vanderhoydonck, Ling-Jie Gao, Miyeon Jang, Piet Herdewijn, Anant Ghawalkar, Farnaz Fallah-
Arani, Adnan Khan, Jamie Henshall, Mark Jairaj, Sarah Malcolm, Eleanor Ward, Lindsay Shuttleworth,
Yuan Lin, Shenggiao Li, Thierry Louat, Mark Waer, Jean Herman, Andrew Payne, Tom Ceska, Carl Doyle,
William Ross Pitt, Mark Calmiano, Martin Augustin, Stefan Steinbacher, Alfred Lammens, and Rodger Allen
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b00521 • Publication Date (Web): 28 Jun 2018
Downloaded from http://pubs.acs.org on June 29, 2018
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 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Discovery of a potent, orally bioavailable PI4KIIIβ
inhibitor (UCB9608) able to significantly prolong
allogeneic organ engraftment in vivo
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
*,†
†
†
†
†
James Reuberson , Helen Horsley , Richard J. Franklin , Daniel Ford , Judi Neuss , Daniel
†
¥
¥
¥
¥
Brookings , Qiuya Huang , Bart Vanderhoydonck , LingꢀJie Gao , MiꢀYeon Jang , Piet
¥
¶
†
†
†
Herdewijn , Anant Ghawalkar , Farnaz FallahꢀArani , Adnan Khan , Jamie Henshall , Mark
†
†
†
†
¥
¥
Jairaj , Sarah Malcolm , Eleanor Ward , Lindsay Shuttleworth , Yuan Lin , Shenggiao Li ,
¥
¥
¥
†
†
†
Thierry Louat , Mark Waer , Jean Herman , Andrew Payne , Tom Ceska , Carl Doyle , Will
†
†
ɸ
ɸ
ɸ
Pitt , Mark Calmiano , Martin Augustin , Stefan Steinbacher , Alfred Lammens and Rodger
†
Allen .
†
UCB Pharma, 208 Bath Road, Slough, Berkshire, SL1 3WE, United Kingdom.
¥
KU Leuven, Interface Valorization Platform, IVAP, Campus ST.ꢀRafaël, Blok I, 8°,
Kapucijnenvoer 33 B 7001, 3000 Leuven, Belgium.
¶
SAI Life Sciences Ltd, International Biotech Park, Hinjewadi, Pune 411 057, India.
ɸ
Proteros Biostructures GmbH, Bunsenstrasse. 7a, 82152 Martinsried, Germany.
ACS Paragon Plus Environment
1

Page 3 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
KEYWORDS Immunosuppressive, allogeneic organ engraftment, Phosphoinositol 4ꢀkinase IIIβ,
transplantation, human mixed lymphocyte reaction, selectivity profile, binding mode, solubility.
ABSTRACT The primary target of a novel series of immunosuppressive 7ꢀpiperazinꢀ1ꢀ
ylthiazolo[5,4ꢀd]pyrimidinꢀ5ꢀamines was identified as the lipid kinase, PI4KIIIβ. Evaluation of
the series highlighted their poor solubility and unwanted offꢀtarget activities. A medicinal
chemistry strategy was put in place to optimize physicoꢀchemical properties within the series,
whilst maintaining potency and improving selectivity over other lipid kinases. Compound 22 was
initially identified and profiled in vivo, before further modifications led to the discovery of 44
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(UCB9608), a vastly more soluble, selective compound with improved metabolic stability and
excellent pharmacokinetic profile. A coꢀcrystal structure of 44 with PI4KIIIβ was solved
confirming the binding mode of this class of inhibitor. The muchꢀimproved in vivo profile of 44
positions it as an ideal tool compound, to further establish the link between PI4KIIIβ inhibition
and prolonged allogeneic organ engraftment, and suppression of immune responses in vivo.
INTRODUCTION
The field of transplantation medicine has seen dramatic advances over the last century with
breakthroughs in management of immune responses and the development of genetically
1
engineered animals for xenografting, and the emerging use of organs from living donors . In the
United States alone, nearly 17,000 kidney, 6,700 liver, 2,600 heart, and 1,900 lung transplants
2
are performed annually . Successful management of immunosuppression, heralded by the
3
discovery that corticosteroids could improve graft retention, led to the key discoveries of the
1
980’s, which were crucial to the further development of transplantation science. In particular
4
5
the discovery of cyclosporine A (CSA) and tacrolimus (Tac or FK506) , cyclic peptides that had
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
profound antiꢀcalcineurin activity, were key. These peptidic calcineurin inhibitors (CNIs) were
found to inhibit Tꢀcell activation resulting in a strong immunosuppressive effect. Current
immunosuppressive regimens promoting long term graft survival use these CNIs in combination
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
with steroids and mycoꢀphenolate mofetil (MMF). Although reduced allograft rejection rates
have been achieved, there are still risks associated with CNI therapies, including potential
7
nephrotoxicity , and there remains a case for the discovery of alternative immunosuppressive
8
,9
agents to prevent allograft rejection.
1
2
Figure 1: Previously described 7ꢀPiperazinꢀ1ꢀylthiazolo[5,4ꢀd]pyrimidinꢀ5ꢀamine analogues with immunosuppressive activity
1
0
In 2011 Jang et al reported the discovery of a series of novel 7ꢀpiperazinꢀ1ꢀylthiazolo[5,4ꢀ
d]pyrimidinꢀ5ꢀamine analogues with a novel immunosuppressive effect (Figure 1). The
compounds were shown to be potent in the human mixed lymphocyte reaction (HuMLR) assay,
often used as a surrogate in vitro assay to predict the prevention of rejection in vivo of
11ꢀ13
transplanted organs
. As well as inhibiting the HuMLR response, 2 appeared to prevent the
rejection of a heterotopic murine cardiac allograft from a C57BL/6 donor mouse to a Balb/C Hꢀ2
recipient, confirming that compounds of this class could suppress an allogeneic response in vivo.
RESULTS AND DISCUSSION
1
4
With 2 as a series exemplar, commercial screening platforms (Cerep and DiscoverX’s
1
5
KinomeScan ) were employed to establish the primary target of these novel inhibitors. No
ACS Paragon Plus Environment
3

Page 5 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
noticeable activity was recorded against the Cerep panel, however kinase profiling suggested the
1
6
most likely target of 2 was a member of the lipid kinase family, PI4KIIIβ . PI4KIIIβ is a
phosphatidylinositol kinase widely expressed in mammalian cells, playing an essential role in
1
7
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
membrane trafficking and signal transduction . Sub families include the PI3KC1, C2 and C3
families and the PI4K class II and III’s. The PI4K class II’s are further divided (PI4KIIα and
1
8
PI4KIIβ) as are the PI4K class III’s (PI4KIIIα and PI4KIIIβ) . PI4K’s are essential for the
synthesis of PI4P (phosphatidylinositol 4ꢀphosphate), the most abundant phosphoinositide in
19
eukaryotic cells , and play critical roles in a number of pathological processes including
2
0
mediating the replication of a number of viruses , and in the development of the parasite
2
1
responsible for malaria . PI4KIIIβ is also understood to play a key role in cell
22
compartmentalization, within the Golgi and the transꢀGolgi network (TGN). Here it is recruited
2
3
24
by the Golgi resident ACBD3 protein , and plays a role in lysosomal and lipid transport
2
5
functions . There is significant interest in targeting PI4KIIIα and PI4KIIIβ isoforms as both are
hijacked by multiple viruses which facilitate their entry to target cells and their subsequent
26ꢀ29
replication
. At the time of discovery of 2, there were limited examples of PI4KIIIβ inhibitors
3
0
in the literature. PIK93 (3, Figure 2), originally developed to target PI3KC1 isoforms , showed
concurrent activity against both PI4KIIIα (IC of 1.1 µM) and PI4KIIIβ (IC of 0.019 µM).
5
0
50
31,32
Recently the structure of this panꢀlipid kinase, coꢀcrystalized with PI4KIIIβ was published
.
3
3
Furthermore, analogues of PIKꢀ93 such as 4, have been disclosed with improved PI4KIIIβ
selectivity, with 5, 6 and 7 emerging as further examples of this class of PI4KIIIβ inhibitor,
34,35
36
showing antiꢀhepatitis C
and antiꢀhuman rhinovirus activity respectively. The selective
PI4KIIIβ inhibitor 8, has also been designed to probe the role of phosphatidylinositol signaling in
3
7,38
cancer cell proliferation
. A somewhat structurally related chemotype exemplified by 9, was
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
39
identified as having antiꢀpolio virus activity , and PI4KIIIβ was established as the likely driver
4
0
for this observed effect. The synthesis of related analogues such as 10 and the structurally
4
1
differentiated 11 have also been disclosed as potent and selective PI4KIIIβ inhibitors with antiꢀ
viral activity established against human rhinovirus and the polio virus. Further core modification
of 9 has also been successful in delivering potent inhibitors of PI4KIIIβ, with excellent
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
2ꢀ45
selectivity profiles
. Rationally designed inhibitors such as 12, were shown to be broad
spectrum antiviral agents with excellent selectivity for PI4KIIIβ over other lipid kinases, with the
structure of several inhibitors of this type bound to the complex of PI4KIIIβ/wtRab11 disclosed.
2
6
7
3
4
5
9
10
11
12
8
Figure 2. Compound 2 in a proposed alignment with exemplar structures of PI4KIIIβ inhibitors from literature (3ꢀ12). The nitrogen atoms
postulated to form the key mono or biꢀdentate interaction with the kinase hinge region are shown in red. Atoms in blue, are likely to be involved
in a second critical interaction with the catalytic lysine residue of PI4KIIIβ. It was plausible 2 could adopt multiple binding modes with the
piperazine and 3ꢀpyridyl groups ‘flipping’ to make a putative interaction with the lysine residue (as hiꢀlighted by the blue arrow).
Compounds (2ꢀ12) all share a putative hydrogen bonding interaction with the hinge region of
PI4KIIIβ. A second interaction to a catalytic lysine is also present in compounds 3ꢀ12, however
it was not clear if 2 was deriving its activity by making the same interaction (Figure 2). To this
end, we sought to confirm the binding mode of these novel inhibitors, establish the link between
PI4KIIIβ inhibition and impaired immune cell function, and evaluate if PI4KIIIβ inhibitors from
ACS Paragon Plus Environment
5

Page 7 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
this unique series had the potential to become part of a new CNI sparing immunosuppressive
regimen for the prevention of the rejection of solid organ allografts. The HuMLR continued to be
used in the absence of any cellular assay where PI4KIIIβ target engagement could be directly
measured. As discussed previously, the HuMLR is a phenotypic assay, and used as a predictor of
the classical immune response (i.e. ‘self’ responding to ‘nonꢀself’). To validate the effectiveness
of the HuMLR as a surrogate cellular assay for assessing PI4KIIIβ inhibitors as
immunosuppressive agents, the IC ’s of a selection of differentiated PI4KIIIβ chemotypes from
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
Figure 2 were generated. PIKꢀ93 (3) was found to inhibit the HuMLR with an IC of 28 nM,
5
0
however as discussed previously, PIKꢀ93 is a panꢀlipid kinase inhibitor, designed to primarily
target the PI3K’s. The more selective analogue 5 was also found to inhibit the HuMLR, with an
46
IC of 71 nM, in line with its reported PI4KIIIβ activity . Compound 9 showed modest activity
5
0
in the HuMLR (IC of 2 µM), likely due to its poor cell permeability, whilst 10 inhibited the
5
0
HuMLR with an IC of 8 nM. Finally, 11 was found to have an IC against the HuMLR of 18
5
0
50
4
7
nM. Both 10 and 11 were described as being selective inhibitors of PI4KIIIβ kinase adding
,
weight to the notion that PI4KIIIβ inhibition was responsible for the reduced HuMLR response
of this structurally diverse set of inhibitors (2, 5, 10 and 11). As a means of prioritizing
compounds for evaluation as potential immunosuppressive agents in vivo, the HuMLR was
deemed to be a simplistic and robust assay to continue to screen against, although whether all
PI4KIIIβ inhibitor chemotypes could replicate the effect seen with compound 2 in vivo, remained
to be confirmed. A more inꢀdepth evaluation of analogues of 1 and 2, some of which were
10
detailed in the previous KUL publication , was now undertaken at UCB. This included in vitro
safety profiling (drugꢀdrug interaction (DDI) risk and cardiovascular safety (CVS) risk), in vitro
metabolic stability in Mouse Liver Microsomes (MLM) and Human Liver Microsomes (HLM)
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and measurement of physicoꢀchemical parameters such as LogD and solubility. Of the many
compounds profiled, 2 and 13 stood out as potent examples of amide and urea capped 7ꢀ
piperazinꢀ1ꢀylthiazolo[5,4ꢀd]pyrimidinꢀ5ꢀamine analogues respectively (Table 1).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 1: Early in vitro profiling of 7ꢀpiperazinꢀ1ꢀylthiazolo[5,4ꢀd]pyrimidinꢀ5ꢀamine series
Compound:
2
13
14
15
22
a
PI4KIIIβ IC50 (nM)
10
16
19
5
2
4
11
32
51
53
a
Hu MLR IC50 (nM)
ꢀ
1
ꢀ1
MLM CLint (µl.min .mg )
133
34
72
32
2.0
0.9
3.42
9
87
20
15
ꢀ
1
ꢀ1
HLM CLint (µl.min .mg )
35
13
21
c
hERG IC50 (µM)
ꢀ
2.2
1.6
3.06
32
1.7
>20
28
c
c
CYP3A4 IC50 (µM)
LogD (pH7.4)
ꢀ
>20
1.85
85
c
c
ꢀ
ꢀ
b
c
Solubility (µM)
ꢀ
6
a
b
c
IC50 values are reported as means of values from at least 2 determinations. Kinetic solubility measured from 10mM DMSO stock at pH7.4. Poor
solubility means data was unobtainable or should be treated with caution.
Although both compounds showed promising potency, the solubility of 2 and 13 was poor, and
CL in HLM and MLM moderate to high. Compound 13 was also found to inhibit the hERG
int
channel with an IC of 2 µM (automated patchꢀclamping (QꢀPatch)), indicative of a potential
50
48
49
CVS risk as well as CYP3A4, with an IC of 0.9 µM, indicative of a potential DDI risk . It
5
0
was not possible to profile 2 in these assays due to its poor solubility. Urea analogue 14, showed
improved PI4KIIIβ and HuMLR potency, but remained a strong inhibitor of both hERG and
CYP3A4, with CL in HLM and MLM remaining high. The phenyl analogue, 15 maintained
int
acceptable activity suggesting the 3ꢀpyridyl nitrogen in 14 was not critical for maintaining
ACS Paragon Plus Environment
7

Page 9 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
PI4KIIIβ activity. Encouragingly 15 appeared to be more stable (compared to 14) in HLM and
MLM, whilst CYP3A4 inhibition appeared significantly reduced, although it is noted that the
poor solubility of this compound meant that this data should be treated with caution.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Synthesis of compounds 13, 22 and 23
2
2
13
f
f
1
1
9b X = H
9a X = Boc
e
21b X = H
1a X = Boc
e
2
a
b
c
d
1
6
18
20
a
b
g, h, i
17
2
3
Reagents and conditions: (a) BocꢀPiperazine, 1,4 dioxane, DIPEA, 1 h, 55 °C, quant.; (b) Lawesson’s reagent, 1,4 dioxane, 0.5 h, 65 °C, 93%; (c)
NBS, DMF, 3 h, 55%; (d) 3ꢀ(Et B)ꢀPyridine, Na CO (aq), Pd(0)(PPh , DME, 150°C, CEM microwave, 0.5 h, 45%; (e) 4N HCl in 1,4 dioxane,
DCM, 1ꢀ16 h, 72ꢀ100%.; (f) Isocyanate, DIPEA or Et N, DCM or DMF, 4 h, 20ꢀ95%; (g) Ethyl Isonipecotate, DIPEA, 1,4ꢀdioxane, 12 h, 70°C
0%; (h) 10% NaOH, THF/MeOH (5:1), 6 h, quant; (i) pꢀAnisidine, HOBT.H 0, EDCI, DIPEA 14 h, 81%.
2
2
3
3 4
)
3
5
2
It could not be ruled out that removal of the pyridyl nitrogen, a potential heme binding element
could be playing a significant part in the improved in vitro ADME profile observed with 15. The
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
10
route previously utilized to access compounds 2 and 13 was modified to allow for exploration
of the 2ꢀposition of the thiazolo[5,4ꢀd]pyrimidineꢀ5ꢀamine core. The bocꢀprotected 2Hꢀ
thiazolo[5,4ꢀd]pyrimidineꢀ5ꢀamine 19a was synthesized from commercial 16 in 2
interchangeable steps according to Scheme 1. Bromination of intermediate 19a using NBS in
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
,4ꢀdioxane yielded 20, a flexible intermediate suitable for the exploration of the 2ꢀposition via a
range of coupling methodologies. To validate the route, compound 13 was reꢀsynthesized using
this new approach, coupling diethyl(3ꢀpyridyl) borane with 20. Deprotection of 21a then
subsequent capping with commercial 4ꢀmethylphenyl isocyanate gave 13 in good yields. It was
intended that with 20 in hand, a SAR exploration of the 2ꢀposition could be initiated with the aim
of identifying potent, soluble and more metabolically stable analogues of 14 and 15 that were
free of hERG and CYP liabilities. Interestingly, upon testing the unsubstituted 2Hꢀthiazolo[5,4ꢀ
d]pyrimidineꢀ5ꢀamine analogue 22, an acceptable level of PI4KIIIβ and HuMLR activity could
be maintained. Indeed, with this ring excised, LogD was reduced, and subsequently MLM and
HLM CL were improved. A modest improvement in kinetic solubility was also noted, and both
int
CYP3A4 and hERG inhibition (although not completely ablated), were much reduced (Table 1).
At this juncture, both 13 and 22 were chosen for kinase selectivity profiling. A set of 250
5
0
diverse kinases were screened at a concentration of 10 µM. No activity was noted against any
tested kinase aside from a handful of lipid kinase isoforms. Concentration responses (IC ) were
5
0
5
1
obtained against the 12 available lipid kinases , and a selectivity profile for 13 and 22 generated
5
2
(Figure 3). It appeared 22 was 15ꢀfold selective for PI4KIIIβ over PI3KC2γ , 50ꢀfold over
PI3KC2α and 120ꢀfold over PI3KC2β. Compound 13, with the appended 3ꢀpyridyl ring,
appeared to be more selective for PI4KIIIβ over the PI3KC2 family of kinases in comparison.
The impact of this modest PI3KC2 inhibition was not clear, as the role of these kinases in
ACS Paragon Plus Environment
9

Page 11 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
53
disease, and their subsequent utility as therapeutic targets
is still being explored.
Encouragingly, both 13 and 22 showed no activity against the other PI4K isoforms, and
maintained good selectivity (>100 fold) over the class 1 PI3K α, β, γ and δ isoforms (Figure 3).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Radial plot of lipid kinase selectivity of 13 and 22 for PI4KIIIβ over the other 11 lipid kinase family members tested.
Compound 22 was chosen based on its in vitro profile to evaluate PI4KIIIβ inhibition in murine
models of Tꢀcell mediated immune response. The blood concentration vs time profile in male
Balb/C mice following oral administration at 10, 30 and 100 mpk was determined for 22, prior to
5
4
evaluation in a mouse model of antiꢀCD3 mediated TꢀCell activation . Exposure appeared to
increase in proportion to dose, however there was significant interꢀindividual variation in
exposure observed, likely influenced by the poor solubility of 22 leading to highly variable oral
ACS Paragon Plus Environment
10

Journal of Medicinal Chemistry
Page 12 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
absorption across the doses tested. Cmax (free) was achieved at around 1h at all doses, and for the
1
00 mpk dose, was between 10 and 30ꢀfold over the HuMLR IC .
50
FU
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
.061
(Mouse
Blood)
Dose (po)
10 mpk
30 mpk
100 mpk
CL/F
mL/min/kg
8
0
112
86
(±5)
(±45)
(±56)
(SD)
AUCinf
Free h.nM
333
769
3907
(±18)
(±306)
(±1903)
(SD)
C
nM
max Free
154
444
967
(±24)
(±209)
(±394)
(SD)
Figure 4. Oral PK parameters were determined in male Balb/C mice, n=3, at the specified dose, with crystalline 22 given as a homogeneous
suspension in vehicle (0.1 % (w/v) Tween 80, 0.1 % (w/v) Silicone antifoam in 1 % (w/v) methylcellulose (400 cps) in water). (A) Time course
(in minutes) for assessing the inhibition of antiꢀCD3 induced TꢀCell activation of IFNγ release by 22. (B) Measured levels of IFNγ release
(pg/ml) for negative control, +ve control (vehicle (0.1 % (w/v) Tween 80, 0.1 % (w/v) Silicone antifoam in 1 % (w/v) methylcellulose (400 cps)
in water)), 1, 10 and 100mpk of 22 (all groups n=8). % Inhibitions refer to the mean (± SEM). *P<0.05 Compared to +ve control by Dunnetts
multiple comparison test. ***P<0.001 compared to +ve control by Dunnetts multiple comparison test. PK samples were taken 1h after dosing
with 22 to coincide with Cmax and confirmed exposures achieved in the experiment were in line with those achieved during the PK study.
As seen in Figure 4, 22 significantly inhibited IFNγ release compared to vehicle treated animals
in a dose dependent manor, indicative of 22 having a strong inhibitory effect on TꢀCell
activation. Subsequent evaluation of 22 in a longer term oxazalone (OXA) induced Tꢀcell
5
5
dependent model of antibody response (Figure 5) showed that 22 could also significantly
inhibit IgG1 production at a dose of 100 mpk (PO). With 22 able to inhibit a range of Tꢀcell
mediated antibody responses in vivo, it was left to evaluate if the modified urea 22 remained
ACS Paragon Plus Environment
11

Page 13 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
efficacious in the murine model of cardiac allograft rejection previously described by the team
at KUL. Balb/C mice carrying a heterotopically transplanted heart from a C57B6 donor were
treated once daily with 100 mpk of 22, and long term graft survival rates of ~50%, when
5
6
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
compared to vehicle alone, could be achieved. Significantly animals treated with 22 continued
to maintain their grafts after treatment withdrawal, differentiating them from CSA treated
animals (Figure 5) where rejection occurred days after treatment withdrawal.
Figure 5. (A) Time course (in hours) for assessing the inhibition of OXA induced IgG1release by 22 (n=8). (B) Measured IgG1 levels after daily
dosing of vehicle (0.1 % (w/v) Tween 80, 0.1 % (w/v) Silicone antifoam in 1 % (w/v) methylcellulose (400 cps) in water) and 22 (100 mpk of
crystalline 22 as a homogeneous suspension in vehicle), measured at day 7 and day 14 compared to vehicle control (arbitrary units (A.U.).
***P<0.001 Compared to +ve control by Dunnetts multiple comparison test. (C) Comparison of survival rates for engrafted mice treated with
vehicle (0.1 % (w/v) Tween 80, 0.1 % (w/v) Silicone antifoam in 1 % (w/v) methylcellulose (400 cps) in water), CSA (40 mpk as a solution in
vehicle) and 22 (100 mpk of crystalline 22 as a homogeneous suspension in vehicle). Animals were dosed via oral gavage once a day for 28 days,
or until a transplanted graft had ceased beating, indicative of rejection. Graft survival is defined as a strongly beating heart (as confirmed by
visual inspection and palpitation).
The majority of graft rejection seen with 22 took place during a time critical period postꢀsurgery,
5
7
when risk of acute rejection is at its highest . Given the delicate nature of the surgery involved,
obtaining multiple PK samples during this critical time was challenging, and establishing the
relationship between free drug levels and longꢀterm graft survival was problematical. With the
high interꢀindividual variability in exposure following oral administration of 22, possibly driven
by poor solubility (see Figure 4), there remained uncertainty as to how much drug each engrafted
ACS Paragon Plus Environment
12

Journal of Medicinal Chemistry
Page 14 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
animal was receiving, during the critical postꢀsurgery window. To establish more accurately the
link between PI4KIIIβ inhibition and successful engraftment in this, and other transplantation
models, a tool compound from this series, with a much improved and reproducible PK profile
was required. Further SAR exploration was undertaken to improve potency and solubility whilst
establishing if compounds such as 22, carrying an embedded electron rich anilinoꢀurea, posed
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
58
any toxicity liability. Many aryl urea’s are found in marketed kinase inhibitors , but there
remains a potential for the metabolically triggered release of electron rich anilines, raising a
59
60
potential genotoxicity or liver toxicity risk, as seen with acetaminophen . Considerable effort
was spent seeking more soluble, nonꢀurea equivalents of 22. Firstly, the piperidine amide
equivalent 23, was made as detailed in Scheme 1.
Scheme 2: Synthesis of amide, aryl and heteroaryl urea analogues of 22
2
2
4a
4d
24b
24c
a
24a-f
2
4e
X=
1
9b
2
5a
25b
25c
25d
25e
b, c or d
2
5f
25g
25h
25i
25j
25k
25a-k
Reagents and conditions: (a) Acid, HATU, DIPEA, DMF, 12 h, 50ꢀ90%; (b) Arylꢀisocyanate, DIPEA or Et
3
N, DCM or DMF, 4 h, 20ꢀ95%; (c)
Arylꢀamine, CDI, DMF, DIPEA, 4 h, 20ꢀ95%; (d) Arylꢀamine, PhOCOCl, pyridine, THF then DIPEA, DMSO 3 h 60°C 50 and 90%.
3
Swapping out the piperazine urea nitrogen for an SP carbon was significantly detrimental to
PI4KIIIβ activity (IC50 >6 µM). Subsequently focus turned to making piperazine amides.
ACS Paragon Plus Environment
13

Page 15 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Libraries of aliphatic and aromatic amides were made, however all but a few showed potencies
in the subꢀµM range, with the synthesis of key examples detailed in Scheme 2. These included
heterocyclic amides such as the imidazoꢀpyridine 24a, that also showed a modest improvement
in solubility. Attempts to improve potency by adding back the hydrogen bond donor of 22, were
unsuccessful, with benzimidazole 24b and indole 24c significantly less active (Table 2). The
only amide analogue of 22 that appeared to have any significant potency was 24d, an analogue
of the original lead 2. The IC against PI4KIIIβ was 216 nM but solubility was significantly
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
worse than 22, with subsequent analogues continuing to suffer from modest potency, poor
solubility and metabolic instability. The phenyl acetamide 24e, where the NH of 22 is swapped
for a methylene, was significantly less active, adding to the evidence supporting the importance
of the urea CO and NH in maintaining activity.
Table 2: SAR of nonꢀurea analogues of 22
Cpd
2
4a
24b
24c
24d
24e
a
PI4KIIIβ IC50 (nM)
946
8795
1.58
>350
6138
>10000
2.54
>10000
>10000
2.84
216
635
2.68
13
5743
>10000
1.50
a
HuMLR IC50 (nM)
LogD (pH7.4)
b
Solubility (µM)
38
14
>350
a
b
IC50 values are reported as means of values from at least 2 determinations. Kinetic solubility measured from DMSO stock at pH7.4
Without a published PI4KIIIβ crystal structure (at this time) to aid design, further efforts to seek
urea replacements were halted. The concern around the inherent risk of genotoxicity associated
with embedded electron rich anilines prompted the profiling of 22 in a 3 strain bacterial miniꢀ
6
1
AMES test. It concluded that 22 was nonꢀmutagenic at the top concentrations tested, with and
ACS Paragon Plus Environment
14

Journal of Medicinal Chemistry
Page 16 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
without metabolic activation. There was however literature evidence to suggest 4ꢀmethoxy
6
2
aniline (pꢀanisidine) was a likely genotoxic liability , although metabolite profiling of 22
concluded no liberation of 4ꢀmethoxy aniline in the presence of isolated human liver
6
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
microsomes . A wide range of alternative urea analogues were synthesized to explore SAR,
drive potency and improve solubility, with key examples detailed in Table 3. Unsubstituted
aniline urea 25a was significantly less active, as was the 3ꢀpyridyl urea, 25b, although kinetic
solubility was improved. Potency could be returned to 25b, by making the methoxyꢀpyridine
analogue 25c, which maintained modest potency, although 10ꢀfold less active than 22.
Homologation to the benzyl urea 25d reduced activity, with the more electronꢀwithdrawing 4ꢀ
OCF analogue, 25e 10ꢀfold less active than 22. A simple methyl scan of the aryl urea (25fꢀh)
3
showed a slight preference for ortho or para substitution over meta, with all three compounds
appearing less active than 22, and significantly less soluble.
Cpd
2
5a
25b
25c
25d
25e
25f
25g
25h
25i
25j
25k
PI4KIIIβ IC50
(nM)
#
517
380
178
3387
>5000
>350
611
424
293
1946
1764
33
548
331
12
114
150
33
312
115
34
145
85
9
414
656
173
195
207
Hu MLR IC50
a
35
33
(
nM)
Solubility
b
21
>350
(
µM)
LogD
pH 7.4)
2
.21
1.46
1.79
2.22
3.39
2.12
2.81
2.55
2.03
1.66
1.65
(
Table 3: SAR of aryl and heteroaryl urea analogues of 22
a
b
IC50 values are reported as means of values from at least 2 determinations. Kinetic solubility measured from DMSO stock at pH7.4
ACS Paragon Plus Environment
15

Page 17 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The observation that an orthoꢀmethyl group was tolerated in 25f, was exploited and 25i the 2ꢀ
methylꢀ4ꢀmethyoxy anilino urea found to have an IC of 9 nM against PI4KIIIβ, with activity
5
0
maintained in the HuMLR. Solubility however was poor, although could again be improved by
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
making pyridine analogues 25j and 25k. Again however, this came at the cost of potency, with
2
5k the most active, with an IC of ~200 nM in the HuMLR. Further efforts to solubilize 25i
50
through aryl substitution, aryl ring modification or addition of solubilizing groups to the critical
methyl or methoxy substituents (data not shown) failed to give the balance of potency and
solubility required. Focus next shifted to reducing the planar nature of the compounds by
inducing a twist to the piperazine linker. It is known that by reducing ‘flatness’ or through the
6
4
introduction of more 3ꢀdimensional structure, the solubility of drug like molecules can be
significantly improved. Thus, a broad range of monoꢀprotected substituted piperazines were
identified with the synthesis of key analogues of 22, detailed in Scheme 3. When a methyl was
introduced adjacent to the urea linker (28a), solubility was enhanced, but potency impacted.
When the methyl was added adjacent to the piperazine nitrogen linking to the hinge binding
group (28b), both primary and cellular potency as well as kinetic solubility were much improved
relative to 22. A further set of diꢀmethylated and bridged piperazine linkers were synthesized
(
28cꢀ28g), although none combined the potency and solubility improvements seen with 28b
Table 4). Through use of the respective chiral bocꢀmethyl piperazines, building blocks 30a and
(
3
0b were synthesized to establish if there was any enantiomeric preference. Capping was then
undertaken with the most potent urea to date, derived from 2ꢀmethylꢀ4ꢀmethyoxy aniline. As can
be seen in Table 4, there was a clear preference for the (S) enantiomer over the (R) with 31a
becoming the most potent analogue of 22 made to date.
ACS Paragon Plus Environment
16

Journal of Medicinal Chemistry
Page 18 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Scheme 3: Synthesis of substituted piperazine analogues
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
8a
28b
28c
b,c
a
17
x=
26a,b,c,e,g
28ab,c,e,g
28e
28g
a
c, b
a
d
b,c
1
6
2
6di
26dii
28d
e
2
7
28f
c or f
b
1
7
2
2
9a X = (S)-Me
9b X = (R)-Me
30a X = (S)-Me
0b X = (R)-Me
31a X = (S)-Me, Y= Me, Z = CH
31b X = (R)-Me Y= Me, Z=CH
31c X = (S)-Me, Y= H, Z = CH
3
3
1d X = (S)-Me, Y= Me, Z = N
Reagents and conditions: (a) Substituted Bocꢀpiperazine, DIPEA, 1,4ꢀdioxane 55ꢀ100 °C 12ꢀ100h, 6ꢀ80%; (b) TFA or HCl in 1,4ꢀdioxane, rt 2ꢀ24
h, 80ꢀ100%; (c) Isocyanate, DIPEA, DCM or DMF, rt 10ꢀ24h, 40ꢀ80%, (d) Lawesson’s reagent, THF, 70°C, 3h, quant.; (e) Intermediate 17,
DIPEA, 1,4ꢀdioxane, 100 °C, 100h, 20%; (f) 6ꢀMeOꢀ2ꢀMeꢀpyridinꢀ3ꢀamine, phenyl chloroformate, Pyridine, THF 0°C then addition to 30b in
DMSO with DIPEA, 60°C, 3h 55%.
ACS Paragon Plus Environment
17

Page 19 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Although 31a appeared to show a modest improvement in kinetic solubility, when
6
5
thermodynamic solubility was measured there was no apparent difference in the solubility of
22 and 31a. The direct analogue of 22, bearing a single chiral methyl on the piperazine, 31c was
made, and single digit nM potency maintained with improved thermodynamic solubility
compared to 22.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 4: SAR of substituted piperazine urea analogues
Cpd
28a
28b
28c
28d
28e
28f
28g
31a
31b
31c
31d
=
PI4KIIIβ
IC50
1500
>10000
>350
ꢀ
18
23
2583
2712
>350
ꢀ
1412
543
83
275
381
88
ꢀ
981
635
>350
ꢀ
2577
1768
>350
ꢀ
4
8
316
472
170
ꢀ
8
31
20
60
a
(nM)
Hu MLR
IC50
a
(nM)
Kinetic
Solubility
>350
ꢀ
160
>350
30
>350
362
b
(µM)
Solubility
ꢀ
TD c
ꢀ
3
ꢀ
(µM)
hERG
IC50 (µM)
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
17
>30
>20
CYP3A4
IC50 (µM)
d
>20
>20
a
b
IC50 values are reported as means of values from at least 2 determinations. Kinetic solubility measured from DMSO stock at pH7.4
.
c
d
Thermodynamic solubility (TD) measured from solid stock at pH7.4. Figure should be treated with caution due to low solubility.
The pyridyl analogue 31d showed a 5ꢀfold drop off in PI4KIIIβ activity, but was still more
potent than 22 and exhibited excellent thermodynamic solubility. There was no CYP3A4
inhibition observed with any (S) methyl piperazine analogue tested, and pyridyl urea 31d was
ACS Paragon Plus Environment
18

Journal of Medicinal Chemistry
Page 20 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
clean in hERG. The more lipophilic 31c however did exhibit a stronger hERG signal than 22, but
not enough to warrant concerns at this stage.
Scheme 4: Synthesis of hinge binder analogues of 31a
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
b
c
3
2
33
34
35
c
d, e
3
6
37
38
g, b
c
4
1
42
f
h, b
39
40
c
4
3
44
Reagents and conditions: (a) PyBOP, DBU, acetonitrile, Bocꢀpiperazine 60 °C 72 h, 66%; (b) 4N HCl in 1,4 dioxane, 1ꢀ8 h, quant.; (c) 4ꢀMeOꢀ
ꢀMeꢀphenylisocyanate, DIPEA, DCM or DMF, rt 4ꢀ24 h, 50ꢀ80%; (d) Bocꢀpiperazine, DIPEA, NMP 110 °C 72 h, 49%; (e) 4M HCl in
2
4 2
methanol, 4 h, quant.; (f) Bocꢀpiperazine, DIPEA, 1,4 dioxane, 80 °C 4 h, 93.8%; (g) Sulfur, NH OH, NMP, 90 °C 4 h, 79%; (h) MeNHNH ,
THF, reflux 4 h, 47%.
ACS Paragon Plus Environment
19

Page 21 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Finally, in a bid to further improve potency and solubility within the series, the 5ꢀmembered ring
adjacent to the hinge binding aminopyrimidine was investigated. Keeping the most potent
piperazine urea combinations identified so far, analogues of 31a were synthesized as detailed in
Scheme 4. Deletion of the nitrogen of the thiazole ring to give the fused thiophene 35 gave a
potent analogue, but thermodynamic solubility was negligible, and the hERG signal had further
increased as LogD increased. Swapping the Sulfur of 31a for an NMe group in purine analogue
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
8 (Scheme 4), gave a sub 10nM PI4KIIIβ inhibitor, with HuMLR potency in line with that of
2. Thermodynamic solubility was much improved with this hinge binding heterocycle, and no
2
hERG or CYP3A4 inhibition observed.
Table 5: Profiling of hinge binder analogues of 31a
Cpd
3
5
38
42
44
a
PI4KIIIβ IC50 (nM)
7
8
7
47
11
37
a
Hu MLR IC50 (nM)
13
82
53
b
SolubilityꢀKinetic (µM)
>350
150
1.35
>30
>20
>350
52
>350
110
1.47
>30
>20
c
SolubilityꢀTD (µM)
0
LogD (pH7.4)
2.39
1.96
8
d
hERG IC50 (µM)
8
d
CYP3A4 IC50 (µM)
>20
>20
a
b
IC50 values are reported as means of values from at least 2 determinations. Kinetic solubility measured from DMSO stock at pH7.4
.
c
d
Thermodynamic solubility (TD) measured from solid stock at pH7.4. Figures should be treated with caution due to low solubility.
The isothiazole 43 was also subꢀ10 nM against PI4KIIIβ, although a drop off in the HuMLR was
again noted. Thermodynamic solubility was modest, but as with 35, a significant increase in the
hERG signal was noted, although no CYP3A4 inhibition was present. Finally, the pyrazole
ACS Paragon Plus Environment
20

Journal of Medicinal Chemistry
Page 22 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
analogue 44 (UCB9608) was synthesized, and gave a good balance of potency and solubility,
with no hERG or CYP flags as shown in Table 5. Compounds 31c, 38 and 44 were chosen to be
further evaluated in vivo. Prior to establishing mouse PK parameters, MLM and HLM stability
was assessed. The addition of the chiral methyl piperazine to 22 (giving 31c) had increased
LogD, and metabolic stability was modestly impacted. The modifications made to the hinge
binding region of 38 and 44 lowered LogD, and both MLM and HLM stability improved relative
to 31c. The efflux ratio (ER: Cacoꢀ2) for 22 and 38 indicated that these compounds were
substrates for efflux transporters, whilst 31c and 44 had a lower ER indicative of a reduced risk.
Mouse PK was performed for the three compounds and data is shown in Table 6.
Table 6: Comparison of ADMET properties of 22 and analogues with improved solubility
Compound
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
2
31c
31
38
53
44
37
a
Hu MLR IC50 (nM)
53
b
SolubilityꢀTD (µM)
0
1.85
30
150
110
LogD (pH7.4)
2.32
1.35
1.47
ꢀ
1
ꢀ1
MLM CLint (µl.min .mg )
15
35
10
6
ꢀ
1
ꢀ1
HLM CLint (µl.min .mg )
21
14
9
4
c
Cacoꢀ2 ER
5
1
5
2.6
Fu (Mouse Blood)
0.061
80 (±5)
332 (±18)
154 (±24)
0.020
0.200
0.090
d
CL/F mL/min/kg (SD)
42 (±12)
205 (±56)
194 (±27)
29 (±4.4)
30 (±5)
0.9 (±0.2)
1.0 (±0.2)
46 (±3)
1746 (±112)
1087 (±186)
19 (±4.0)
447 (±86)
0.8 (±0.18)
0.8 (±0.1)
39
5.3 (±0.7)
6941 (±887)
2113 (±259)
5 (±0.4)
684 (±51)
1.4 (±0.12)
0.6 (±0.04)
d
AUCinf Free h.nM (SD)
d
C
max Free nM (SD)
e
f
CL (mL/min/kg)
ꢀ
e
f
AUC Free h.nM (SD)
ꢀ
e
f
t
1/2
h
ꢀ
e
f
Vss L/kg (SD)
ꢀ
g
f
Oral F%
ꢀ
68
~100
a
c
d
IC50 values are reported as means of values from at least 2 determinations. Calculated from Papp (AꢀB)/(BꢀA) in Cacoꢀ2 assay. Oral PK
established in male Balb/C mice, n=3 dosed at 10mpk in vehicle (0.1 % (w/v) Tween 80, 0.1 % (w/v) Silicone antifoam in 1 % (w/v)
e
methylcellulose (400 cps) in water) as homogeneous suspensions of crystalline 31c, 38 or 44. i.v PK established in Male Balb/C mice, n=3 dosed
ACS Paragon Plus Environment
21

Page 23 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
f
g
at 1mpk in vehicle (30% DMA for 31c and 38 or 15% NMP for 44). Too insoluble to formulate for i.v evaluation. Bioavailability extrapolated
from i.v/po experiments.
All three more soluble compounds showed low volume of distribution, and moderate to short
halfꢀlives. The higher LogD 31c, had a higher CL compared to other analogues, and so free drug
b
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
levels following PO administration were no better than those seen with 22, despite being well
absorbed. Compound 38 although having a slightly lower CL than 31c, had a comparable halfꢀ
b
life, impacted by its low volume of distribution. The bioavailability was also lower, which was
unexpected due to the improved solubility and reduced clearance, and may have been due to
reduced intestinal absorption, driven by active transport, in line with Cacoꢀ2 data. Compound 44
had very low CL , commensurate with its low CL in MLM, leading to a halfꢀlife of 1.4h, with
b
int
high bioavailability, and low interꢀindividual variability. With its vastly improved oral PK profile
in comparison to 22, the pyrazolopyrimidine 44 appeared to be an excellent in vivo tool
6
6
compound. AMES MPF screening of 44 showed no flags and metabolic profiling in isolated
67
human microsomes again showed no evidence of urea hydrolysis or aniline derived
metabolites. The embedded 2ꢀmethylꢀ4ꢀmethoxy aniline was also assessed in AMES MPF, and
6
8
appeared free of risk .
ACS Paragon Plus Environment
22

Journal of Medicinal Chemistry
Page 24 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. Radial plot showing the Xꢀfold selectivity of 22 and 44 for PI4KIIIβ over 11 other lipid kinase family members
5
0
Kinome wide screening of 44 was undertaken, and of the 250 kinases tested at 10 µM, only the
PI4KIIIβ and PI3KC2 α, β and γ lipid kinases were inhibited. The selectivity profile of 44 for
PI4KIIIβ over the 11 available lipid kinases confirmed that 44 had a much improved selectivity
6
9
70
profile in comparison to 22 (Figure 6). Throughout the discovery of 44, UCB and Proteros
worked together to deliver a crystal structure of a piperazine urea inhibitor bound to PI4KIIIβ
that would confirm a binding mode and rationalize the observed SAR. Initial efforts to solve the
structure of any protein/ligand complex were hindered by the poor behavior of PI4KIIIβ toward
crystallization. Several disordered regions were identified within the protein and it was
envisaged that through reengineering of these flexible loops (a process also utilized to deliver
3
2,42
recently published PI4KIIIβ/ligand complex structures
), constructs more amenable to
crystallography could be obtained. It therefore proved possible to obtain the first crystals of a
ACS Paragon Plus Environment
23

Page 25 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
urea (44) bound to human PI4KIIIβ. These crystals consisted of 2 monomers of PI4KIIIβ in the
asymmetric unit, with one monomer (Chain A) being well defined, whilst the second monomer
7
1
(Chain B) was highly disordered . Clear density corresponding to the structure of 44 was visible
in the kinase domain of the ordered monomer (Chain A), and was used for all subsequent
analysis of the binding mode. As expected, the amino pyrimidine of 44 makes a biꢀdentate
interaction with the Val613 backbone, consistent with other published structures of amino
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
2,44
heterocyclic inhibitors bound to PI4KIIIβ
. Whereas the aryl and amide side chains of
inhibitors such as 12, reach along the ATP binding pocket toward Lys564 in one direction, and
Asn615 in the other, 44 appears to orient the appended piperazine urea away from the pocket
toward solvent (Figure 7). There is no contact with Lys564, Asn615 or Tyr385, residues believed
42
to be critical for PI4KIIIβ activity , although it could be postulated that earlier urea analogues
bearing an aryl or 3ꢀpyridyl group on the hinge binding core could reach toward Lys564
explaining the difference in potency between 14 and 22 (Table 1).
ACS Paragon Plus Environment
24

Journal of Medicinal Chemistry
Page 26 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. (A) Coꢀcrystal of 44 (Orange) and PI4KIIIβ (green) with surface superimposed in grey (PDB ID 6GL3). (B) The unbiased electron
density (foꢀfc, contoured at 3σ) of 44. (C) Detailed view of the key protein/ligand interactions made by 44 (orange) with PI4KIIIβ (green). Hꢀ
bonds shown as dashed lines, CꢀH/π interactions as dotted lines. (D) Overlay of 44 (orange) and 12 (Purple) from published structure (5FBL). Hꢀ
bond interactions shown as dashed lines (black for shared, orange for 44 specific and purple for 12 specific). Curved arrow denotes loop
movement seen in structure of 44. (E) Detailed overlay of 12 and 44 showing key residues involved in Hꢀbonding, and the different direction of
ACS Paragon Plus Environment
25

Page 27 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
growth from the hinge binding heterocyclic core of each ligand. Images generated using PyMol (The PyMOL Molecular Graphics System,
Schrödinger, LLC).
As can be seen in Figure 7, the urea NH and carbonyl of 44 form Hꢀbonds with the main chain
Gly675 carbonyl and the side chain NH of Asn390 respectively. A water molecule is also
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
visible within hydrogen bonding range of the side chain OH of Ser618 and the urea carbonyl.
The Asn390 residue is part of a loop of protein that has shifted to enclose the electron rich
aromatic ring of the urea in 44, with the side chain methylene of Asn390 and the main chain
methylene of Gly675 available to make putative CꢀH/π interactions from above and below.
These interactions made by the aryl urea of 44 and the loopꢀshifted protein could help rationalize
the previously observed SAR. The conformation of the piperazine, the geometry of the urea
carbonyl and NH, as well as the electronics of the aromatic ring all needed to be optimal to give
3
the most potent compounds. For example, the change from a planar nitrogen to an SP carbon in
amide 23 meant that it could not deliver the ꢀCONHAr group on the same vector as observed
with 44, resulting in a drastic loss of activity. The piperazine amides (Table 2) would also be
unable to replicate the binding mode of 44, and suffer similar drops in activity. The modest
potency observed with 24d was the exception, however it was not possible to confirm a binding
mode for this compound. As discussed previously electron rich aniline urea’s such as 22
appeared more potent against PI4KIIIβ, whilst electron poor urea’s such as 25b or 25e were
significantly less active. This could be rationalized by urea’s bearing electron donating
substituents showing an increased propensity to stabilize the loop movement observed in Figure
7
2
7
, through strengthening of the CꢀH/π interactions with the Asn390 side chain and Gly675
backbone methylene’s. Certainly, the observation that the most potent compounds bear both a
methoxy and a methyl group, suggest the effect is additive, and explains the jump in potency
observed from 22 to 25i, although from examining the structure of 44, there also appears to be an
ACS Paragon Plus Environment
26

Journal of Medicinal Chemistry
Page 28 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
element of space filling with this ortho substituent, suggesting further exploration in this area is
warranted. From the crystal structure of 44, it was also noted that the chiral methyl appended to
the piperazine linker sat axial, and pointed into a small hydrophobic pocket within the active site.
There was obviously a clear difference between the (S) and (R) enantiomers, based on the data
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
73
generated with compounds 31a and 31b and this proved to be the case for 44 as well . From
74
experience and from examination of analogous structures in the CSD the axial conformation
for the chiral methyl attached to piperazine is energetically favored over the equatorial (results
not shown). Docking was difficult for these compounds because the correct geometry for the
piperazine nitrogen is required, and the pocket is open on two sides. A protocol using ring
7
5
76
conformations generated using Corina , minimized using the OPLS3 forcefield and docked
77
with Glide was able to reproduce the crystallographic binding modes. Docking of the distomer
3
1b resulted in only a slightly worse docking score than 31a but with the methyl in the equatorial
conformation. We assume that the high eudysmic ratio between these two compounds derives
from higher strain energy of this conformation, combined with the fact that the methyl in the
distomer does not fit into the hydrophobic pocket shown in the Xꢀray structure of 44. This
apparent preference for an axial substituent fitting into this small hydrophobic pocket goes some
way to explain the loss of activity with the more complex substituted piperazines in Table 4. It
remains to be seen if the binding mode of 44, has any relevance to the biological profile of these
specific PI4KIIIβ inhibitors, and further evaluation of the immunological profile of PI4KIIIβ
inhibitors with different binding modes and selectivity profiles is ongoing, and will be reported
in future publications. To complete the assessment of 44, it was taken into the murine model of
cardiac allograft rejection, at a dose of 5mpk (PO). With the improved oral bioavailability and
reduced interꢀindividual variability in exposure of 44, graft survival rates were significantly
ACS Paragon Plus Environment
27

Page 29 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
improved as illustrated in Figure 8. Indeed, survival rates were unaffected after withdrawal of
drug treatment at d14, with 90% of grafts being retained at day 50 (>36 days drug free)
compared with a 50% survival rate for 22. With 44 confirmed as a potent immunosuppressive
agent, capable of prolonging heterotopic allograft retention in vivo it positions itself as an ideal
tool compound to establish if PI4KIIIβ inhibition can play a critical role in the development of
future clinical immunosuppressive regimens.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Comparison of survival rates for engrafted mice treated with vehicle (0.1 % (w/v) Tween 80, 0.1 % (w/v) Silicone antifoam in 1 %
(w/v) methylcellulose (400 cps) in water), or 44 (5mpk in vehicle as a homogenous suspension). Animals were dosed via oral gavage once a day
for 14 days, or until a transplanted graft had ceased beating, indicative of rejection. Graft survival is defined as a strongly beating heart (as
confirmed by visual inspection and palpitation).
Further studies with 44 (UCB9608) and subsequent analogues are being explored, and will be
discussed in future publications.
CONCLUSION
Within this article, we have described the discovery 44 (UCB9608), an 11 nM PI4KIIIβ inhibitor
that inhibits the HuMLR response with an IC50 of 37 nM. Its potency and excellent ADME
properties make it an ideal compound for future use as an in vitro and in vivo probe to elucidate
ACS Paragon Plus Environment
28