Optimization of Orally Bioavailable PI3KδInhibitors and Identification of Vps34 as a Key Selectivity Target

Article abstract of DOI:10.1021/acs.jmedchem.9b01585

Optimization of a lead series of PI3Kδinhibitors based on a dihydroisobenzofuran core led to the identification of potent, orally bioavailable compound 19. Selectivity profiling of compound 19 showed similar potency for class III PI3K, Vps34, and PI3Kδ, and compound 19 was not well-tolerated in a 7-day rat toxicity study. Structure-based design led to an improvement in selectivity for PI3Kδover Vps34 and, a focus on oral phramacokinetics properties resulted in the discovery of compound 41, which showed improved toxicological outcomes at similar exposure levels to compound 19.

Full text of DOI:10.1021/acs.jmedchem.9b01585

Subscriber access provided by BIU Pharmacie | Faculté de Pharmacie, Université Paris V
Article
Optimization of Orally Bioavailable PI3K# Inhibitors
and Identification of Vps34 as a key off-target activity
Zoë A Henley, Augustin Amour, Nick Barton, Marcus Bantscheff, Giovanna Bergamini, Sophie
Marie Bertrand, Maire A. Convery, Kenneth David Down, Birgit Duempelfeld, Christopher
D. Edwards, Paola Grandi, Paul M. Gore, Steven Keeling, Stefano Livia, David N. Mallett,
Aoife C. Maxwell, Mark Price, Christina Rau, Friedrich Reinhard, James E. Rowedder,
Paul Rowland, Jonathan Taylor, Daniel Thomas, Edith M Hessel, and Nicole Hamblin
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b01585 • Publication Date (Web): 19 Dec 2019
Downloaded from pubs.acs.org on December 22, 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 84
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Optimization of Orally Bioavailable PI3Kδ Inhibitors
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
and Identification of Vps34 as a key off-target activity
†
±
§
∥
∥
Zoë A. Henley* , Augustin Amour , Nick Barton , Marcus Bantscheff , Giovanna Bergamini ,
†
‡
†
∥
⊥
Sophie M. Bertrand , Máire Convery , Kenneth Down , Birgit Dümpelfeld , Chris Edwards ,
∥
†
†
†
⊥
†
Paola Grandi , Paul M. Gore , Steve Keeling , Stefano Livia , David Mallett , Aoife Maxwell ,
○
∥
∥
#
‡
Mark Price , Christina Rau , Friedrich B. Reinhard , James Rowedder , Paul Rowland ,
†
#,◊
±
†,∆
Jonathan A. Taylor , Daniel A. Thomas , Edith M. Hessel , J. Nicole Hamblin
†
§
Global Medicinal Chemistry and Data and Computational Sciences, Medicinal Science and
Technology, GlaxoSmithKline R&D, Gunnels Wood Road, Stevenage, SG1 2NY, U.K.
±Refractory Respiratory Inflammation Discovery Performance Unit, GlaxoSmithKline R&D,
Gunnels Wood Road, Stevenage, SG1 2NY, U.K.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 84
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
#
⊥
Screening, Profiling and Mechanistic Biology, Drug Metabolism and Pharmacokinetics and
‡Protein, Cellular and Structural Sciences, GlaxoSmithKline R&D, Gunnels Wood Road,
Stevenage, SG1 2NY, U.K.
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
○In vitro In Vivo Translation, GlaxoSmithKline R&D, David Jack Centre, Park Road, Ware,
SG12 0DP, U.K.
∥Cellzome GmbH, GlaxoSmithKline, Meyerhofstrasse 1, 69117 Heidelberg, Germany
Abstract
Optimization of a lead series of PI3Kδ inhibitors based on a dihydroisobenzofuran core led to
the identification of potent, orally bioavailable compound 19. Selectivity profiling of compound
19 showed similar potency for the class III PI3K, Vps34, as for PI3Kδ and compound 19 was not
well-tolerated in a 7-day rat toxicity study. Structure-based design led to an improvement in
selectivity for PI3Kδ over Vps34 and a focus on oral PK properties resulted in the discovery of
compound 41, which showed improved toxicological outcomes at similar exposures to
compound 19.
2
ACS Paragon Plus Environment

Page 3 of 84
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
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 84
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
Introduction
The phosphoinositide 3-kinases (PI3Ks) are a family of enzymes that are central to signaling
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
involving intracellular lipids. Several distinct classes of PI3Ks have been identified based upon
their structural features and substrate specificity. The Class I PI3K family includes
phosphoinositide 3-kinase delta (PI3Kδ) and the closely related isoforms α, β and γ, whose
activation leads to the production of phosphatidylinositol-3,4,5-trisphosphate (PtdIns(3,4,5)P3)
2
in vivo. This key signaling molecule acts as a secondary messenger, interacting with the lipid
binding domains of a variety of cellular proteins to influence a diverse array of cellular pathways
that have been shown to be important in disease pathology.3
4
5
PI3Kδ has a role in T- and B-cell antigen receptor signalling, neutrophil chemotaxis, and
inactivation of PI3Kδ leads to impaired allergen-IgE-induced mast cell degranulation and
6
cytokine release. Therefore, the target has been implicated in the development of inflammatory
conditions and a range of cancers.7,8 PI3Kδ expression is restricted to leukocytes and PI3Kδ-
9
deficient mice have been shown to be viable and fertile, hence it is an attractive target for
10
11
respiratory diseases such as asthma, chronic obstructive pulmonary disease (COPD) and
4
ACS Paragon Plus Environment

Page 5 of 84
Journal of Medicinal Chemistry
1
2
3
autoimmune disorders.12 PI3Kγ is also enriched in leukocytes and has been targeted for the
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
13
treatment of inflammatory disease, whereas the α and β isoforms are ubiquitously expressed
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
and the α and β murine knockout phenotypes are embryonically lethal.14 Hence, isoform
selectivity is important when aiming to treat chronic disease.
Several orally bioavailable PI3Kδ inhibitors are advancing through clinical studies for oncology
15
16
17
indications, for example duvelisib (IPI-145) and umbralisib (TGR-1202) , and Idelalisib has
been successfully developed for the treatment of patients with relapsed chronic lymphoid
leukemia and non-Hodgkin lymphoma. More selective, orally bioavailable compounds are also
being progressed for the treatment of inflammatory disorders, for example seletalisib (UCB-
18
19
20
5857) and AM-8508 , and activated PI3K delta syndrome, for example leniolisib (CDZ-173).
Given the significant interest in PI3Kδ as a therapeutic target, a need for orally bioavailable
compounds with further optimized selectivity remains. We previously reported inhaled clinical
21
candidates nemiralisib and GSK2292767 that achieve class-leading levels of selectivity, and
aimed to recapitulate this within a novel, orally bioavailable series. Our studies led to the
discovery of a lead series containing a dihydroisobenzofuran (DHB) core that showed promising
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 84
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
selectivity and in vivo characteristics, as exemplified by compound 1 (Figure 1), which has good
oral bioavailability and encouraging PI3K isoform selectivity.22 Herein we describe the
optimization of this series of PI3Kδ inhibitors to enhance the primary activity and selectivity
profile while maintaining acceptable pharmacokinetic properties for oral dosing.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
N
O
NH
O H
N
O
S
O
O
N
1
PI3K enzyme pIC50 = 6.8
PI3K,, enzyme pIC50 = 5.3,5.0,5.0
PI3Kd hWB pIC50 = 6.5
Cl = 14 ml/min/kg
b
F = 54%
Figure 1. Orally bioavailable PI3Kδ inhibitor.
Results and Discussion
A co-crystal structure of compound 1 showed a binding mode with the DHB acting as the
hinge-binding motif, engaging in a single donor-acceptor interaction in the hinge region of the
22
kinase. The pyridyl sulfonamide group occupied the lipid kinase affinity pocket and the amide-
linked morpholino group extended towards the tryptophan shelf, a region of the protein adjacent
6
ACS Paragon Plus Environment

Page 7 of 84
Journal of Medicinal Chemistry
1
2
3
to Trp760 that has previously conferred selectivity for PI3Kδ over the other PI3K isoforms when
occupied.21,23 However, it was noted that the morpholine substituent itself did not reach the
tryptophan shelf, which led to the hypothesis that further selectivity could be obtained by
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
22
exploiting this region. The results of this SAR exploration are summarized in Table 1, with the
in vitro enzyme potencies of the compounds measured in a homogeneous time-resolved
fluorescence (HTRF) assay as described previously.21
R
O H
N
O
S
O
O
N
pIC50*
Compoun
d Number
hWB clog
R
PI3Kδ PI3Kα PI3Kβ PI3Kγ
IFNγ
Pa
O
6.1b
2
3
7.1
7.3
5.9
5.2
5.1
5.5
5.2
1.7
O
NH
O
5.6
7
6.3
2.8
O
NH
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 84
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
O
N
O
4
5
7.3
7.5
5.1
6.6
4.9
6.3
4.8c
6.5
6.3b
6.0b
2.6
2.4
NH
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
HN
O
O
N
6
7
8
9
6.0
6.3
5.1
5.2
4.9c
5.1
6.0
6.1c
5.6
5.3
5.4c
6.1
4.7d,e 1.8
O
O
N
5.3b
6.2d
6.4
3.8
2.4
3.5
2.2
O
N
7.2
5.9
O
O
N
8.0b,f
7.2
5.5 b,e
6.0c
4.9e
O
N
O
10
5.8
6.0
NH
8
ACS Paragon Plus Environment

Page 9 of 84
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
N
1
1
2
7.8h
6.1h
6.2h
5.3h
6.7
2.8
O
Isomer 1
N
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
8.3h
7.5
7.0h
6.2
6.8h
6.0
5.7h
5.9
7.1
6.3
2.8
2.2
O
Isomer 2
N
O
13
14
O
N
O
8.0
7.0
6.3
5.4
6.1
5.3
6.2
6.6
2.2
0.7
O
N
N
O
1
5
6
5.2c
6.9b
NH
N
N
O
1
7.4
7.0
5.2
5.2
5.0g
4.9c
6.5
1.5
2.1
NH
HO
N
17
5.2c
5.2c
5.9i,b
O
NH
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 84
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
HO
N
1
8
9
7.4
8.4
5.9
5.7
6.0
6.2
5.5
5.4
5.7
2.6
NH
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
HO
N
1
7.5j,k,l 2.9
O
a
*
Data is mean of at least 3 separate test occasions unless stated. NT = not tested. clogP
b
c
calculated using Biobyte model. Did not return a value on one test occasion. Tested <4.5 on
d
e
one test occasion. Did not return a value on three test occasions. Tested <4.4 on two test
f
g
h
occasions. Tested >10.5 on one test occasion. Tested <4.5 on three test occasions. mean of
i
j
2
k
separate test occasions. Tested <5.6 on one test occasion. Tested <4.4 on one test occasion.
l
Tested <5.0 on one test occasion. Tested 53 times, did not return a value on 10 test occasions.
Table 1. 4-position modifications to give compounds 2-19.
Tetrahydropyran analogue 2 showed similar PI3Kδ potency to compound 1, however
increased activity against the other isoforms was observed, particularly PI3Kα. In order to
improve selectivity, a strategy developed previously whereby additional bulk is added to the
2
1
pendant group and is accommodated in PI3Kδ but not the other isoforms, was employed. The
cis-dimethyl-containing compounds 3 and 4 showed up to a 3-fold increase in PI3Kδ potency
and decreases in activity against PI3Kα, β and γ. This results in an improved selectivity profile,
10
ACS Paragon Plus Environment

Page 11 of 84
Journal of Medicinal Chemistry
1
2
3
particularly for compound 4 where ≥100-fold selectivity for PI3Kδ over the PI3K isoforms was
achieved.
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
Changes to the linker from the core to the pendant groups were explored and PI3Kδ potency
was maintained when the amide was reversed in compound 5, however this compound also
showed greater than 10-fold increases in potency against the other isoforms. Carbon-linked
compounds 6 and 7 showed a decrease in potency against PI3Kδ and similar potency against
the other PI3K isoforms. Ether linking groups (compounds 8 and 9) led to up to 5-fold increases
in potency when compared to their amide counterparts (compounds 1 and 4, respectively) and
an increase in PI3K isoform selectivity with increasing steric bulk on the morpholine group.
Pendant groups in this region have been shown to interact with Trp760, increasing PI3Kδ
potency over and above the other isoforms for which access to the tryptophan indole moiety is
23
restricted by a large basic residue. Anilinic and ether linkers have a conformational preference
for co-planarity with the DHB aryl core, potentially pre-organizing the conformation of linkers to
set up an interaction of the pendant group with Trp760. Amide linkers were also shown to offer
some benefit through the pendant moiety, although perhaps the increased rigidity prohibited
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 84
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
interaction as favorable as those of more flexible linkers such as the ether. The SAR seemed to
further confirm that positioning of a cation and increased lipophilicity in the vicinity of Trp760
were advantageous to PI3Kδ potency.
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
Compound 10 showed potency was maintained when the cyclic substituent was directly
attached to the amide linker and the analogous ether compounds 11 and 12 were even further
improved, with one isomer 3-fold more potent than the other, albeit with concomitant potency
increases for the other isoforms. It was found that the R-enantiomer was preferred for PI3Kδ
potency over the S-enantiomer when known isomers 13 and 14 were evaluated, and these data
also showed that a reduction in lipophilicity via replacement of piperidine for morpholine was
tolerated. Other cyclic amines were also potent, for example compounds 15 and 16, where the
larger isopropyl group again confers selectivity for PI3Kδ.
Unfortunately, none of these compounds achieved the desired level of potency in a human
whole blood (hWB) assay (pIC >7.5). This assay involved using cytostim to stimulate cytokine
50
production from the T-lymphocyte compartment and detection of interferon γ (IFNγ), which is
sensitive to PI3Kδ inhibition. Many of the compounds had lower permeability (<100 nm/s in an
12
ACS Paragon Plus Environment

Page 13 of 84
Journal of Medicinal Chemistry
1
2
3
artificial membrane permeability assay, data not shown) than desired, potentially impacting the
whole blood potency due to reduced cell permeation. Therefore, a bulky tertiary alcohol
substituent was incorporated onto the piperidine in order to improve permeability, as previously
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
24
reported , however this was not successful for compound 17. The amine analogue 18 was also
prepared to further assess the linking group and showed similar levels of enzyme potency and
selectivity to the amide but also low whole blood potency. A breakthrough came when the ether
linker was introduced (compound 19), which gave an improvement in PI3Kδ potency, maintained
≥100-fold selectivity over the other isoforms and achieved the desired whole blood activity.
A crystal structure of compound 19 soaked into the active site of PI3Kδ was obtained and
showed the dihydrobenzofuran acting as the hinge-binding motif with Val828 and clear density
for the substituted piperidine moiety occupying the tryptophan shelf (Figure 3). The quality of the
density was sufficient to determine the conformation of the piperazine ring such that the
protonated nitrogen could coordinate an identified water molecule. The conformation of the
isopropoxy group could not be determined directly from the electron density, but a methyl group
is most likely to be pointing at the Trp760 thus maximizing the hydrophobic interactions.
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 84
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 3. X-ray crystal structure of compound 19 (gray, PDB ID: 6TNR) – view showing clear
density (blue lines) for 4-position group and interaction between Trp760 and substituted
piperidine moiety.
Compound 19 was screened in additional assays to understand the wider profile (Table 2). In
vitro intrinsic clearance data for compound 19 of 2.5 ml/min/g of tissue (rat liver) scales to 54%
25
liver blood flow (LBF) using the well-stirred model and a non-restrictive approach, while rat in
vivo clearance of 48 ml/min/kg equates to approximately 62% LBF (based on a liver blood flow
14
ACS Paragon Plus Environment

Page 15 of 84
Journal of Medicinal Chemistry
1
2
3
2
6
in rat of 78 ml/min/kg). This clearance was higher than that of compound 1, however, in general
for this series, the in vitro-in vivo correlation (IVIVC) was good hence it was encouraging that the
in vitro clearance in human hepatocytes was low for compound 19. Compound 19 achieved good
solubility, moderate in vitro passive permeability measured on MDCK cells, and moderate oral
bioavailability in the rat. Compound 19 also had an increased volume of distribution and
improved half-life compared with compound 1, and its overall profile led to its selection as the
first compound from this series to be evaluated in toxicity screening studies.
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
R
O H
N
O
S
O
O
N
R
O
HO
N
O
N
NH
O
Compound Number
1
19
PI3Kδ isolated enzyme pIC50
6.8
8.4
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 84
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
PI3Kδ human whole blood
6
.5
7.5
pIC50
a
IVC rat, minipig, human liver
-
2.5,<0.9,<0.9
hepatocytes (ml/min/g)
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
b
AUC (p.o., ng.h/ml)
673
1.3
14
148
2.1
48
inf
Half life, T1/2 (i.v.) (hr)
Cl (i.v.) (ml/min/kg)
Oral bioavailability, F (%)
Vdss (L/kg)
54
32
1.1
185
5.9
197
c
CLND Solubility (μg/ml)
MDCK passive perm. at pH
122
39
7.4 (nm/s)
Blood binding free fraction,
rat, human (%)
-
15, 4
a
b
IVC = in vitro clearance. In vivo PK data were obtained in the Wistar Han male rat following
c
administration of 1 mg/kg oral and i.v. doses. CLND = Chemiluminescent Nitrogen Detection.
Table 2. Profile of compound 19 compared with progenitor compound 1.
Whilst compound 19 was being evaluated in toxicity screening studies, further exploration
within the series focused on improving the PK profile, particularly reducing the systemic
clearance. Strategies such as reducing lipophilicity and blocking or modulating potential sites of
16
ACS Paragon Plus Environment

Page 17 of 84
Journal of Medicinal Chemistry
1
2
3
metabolism, identified using in silico metabolism predictors,27,28 were employed. The results are
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
summarized in Table 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
O H
R2
N
R1
S
O
O
N
pIC50*
a
IVC rat,
Compound
Number
PI3K PI3K PI3K hWB
human
clogP
R1
R2
PI3Kδ
b
α
β
γ
IFNγ (ml/min/g
tissue)
HO
HO
HO
N
2
0
O
Me
5.5
<4.1c <4.1c
4.1c
<4.4d
NT
3.4
O
Cl
N
2
1
2
O
F
Me
Me
7.6
7.4
5.0
5.9
5.7
6.1
4.7
5.3
6.2
6.3
4.5,<0.9
3.1
2.9
O
N
F
2
NT
O
O
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 84
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
HO
N
2
3
4
Me
Me
7.8
6.3
5.7
6.2
6.2
6.3
NT
NT
2.5
1.5
O
N
O
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
N
O
O
2
4.9e
4.5c
4.9e
5.8f
O
N
HO
N
25
O
Me
8.6
5.6
5.8
5.4
8.1g
9.7,<0.9
NT
3.4
O
HO
N
2
2
6
9.1
8.1
6.5
5.3
7.3
5.9
6.3
5.3
6.6d
7.4h
2.2
3.0
O
N
O
HO
N
O
7
1.6,<0.9
O
O
a
*
Data is mean of at least 3 separate test occasions unless stated. NT = not tested. IVC data
b
c
in liver hepatocytes. clogP calculated using Biobyte model. Tested <4.5 on two test occasions.
d
e
f
Did not return a value on two test occasions. Tested <4.5 on one test occasion. Did not return
g
a value on one test occasion. Tested 26 times, did not return a value on 15 test occasions,
h
tested <5.0 on one test occasion and <4.4 on one test occasion. Tested 50 times, did not return
a value on 5 test occasions.
Table 3. Modifications to give compounds 20-27.
18
ACS Paragon Plus Environment

Page 19 of 84
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
Incorporation of a halide at the DHB 7-position led to detrimental effects on PI3Kδ potency.
Compound 20 showed that a chloride reduced potency significantly and fluoro-containing
compound 21 showed 3- to 8-fold decreases in potency across all the PI3K isoforms.
Fluorination in the 5-position (compound 22) was tolerated for PI3Kδ potency but reduced PI3K
subtype selectivity more than fluorination in the 7-position (compound 21), potentially due to a
conformational effect on the 4-position selectivity vector. Compound 21 achieved >80-fold
selectivity over the other isoforms and moderate hWB activity, however in vitro clearance data
in rat showed an increase compared to compound 19, which may be due to increased
lipophilicity. The less lipophilic nitrogen-containing compounds 23 and 24 showed reduced
potency and selectivity compared with analogous compounds (19 and 14).
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
Incorporation of methyl substituents at the benzylic positions of the DHB core was investigated.
It had been shown on compounds containing a 4-acetamide substituent and a methyl group at
the 1-position (compounds not shown) that the R-enantiomer achieved higher PI3Kδ potency
than the S-enantiomer (PI3Kδ pIC50 7.5 vs 5.4), hence compound 25 was synthesized. This
showed similar PI3Kδ potency to that of compound 19 with an improved selectivity profile. There
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 84
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
was also a beneficial effect on hWB potency, however the in vitro clearance in the rat was very
high, with 9.7 ml/min/g scaling to 82% LBF using the well-stirred model and a non-restrictive
approach.25
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
Functionalization of the sulfonamide was also explored, with small groups such as those
included in compounds 26 and 27 leading to high PI3Kδ potency and good selectivity over the
other PI3K isoforms. Incorporation of the nitrile substituent reduced the lipophilicity such that the
permeability (<3 nm/s in an artificial membrane permeability assay), hence hWB potency, was
impacted. Inclusion of a methoxyethyl substituent in compound 27 led to similar lipophilicity to
compound 19 but lower in vitro clearance, which also translated to a lower in vivo clearance in
the rat following administration of a 1 mg/kg i.v. dose (Cl = 22 ml/min/kg, 28% LBF).
b
Compound 19 was evaluated in a 7-day rat study, at doses of 0, 30, 100 and 300 mg/kg/day
delivered once daily by oral gavage (n=4 per sex per group); however, a dose of 300 mg/kg/day
was not tolerated and the animals were sacrificed after a single dose due to adverse clinical
signs. The principal toxicity noted at necropsy in these deaths was significant distension and/or
fluid contents in the stomach and caecum; subsequent microscopic examination revealed single
20
ACS Paragon Plus Environment

Page 21 of 84
Journal of Medicinal Chemistry
1
2
3
cell necrosis in the stomach and throughout the GI tract. The animals receiving 100 mg/kg/day
survived to scheduled termination after 7 days of dosing but single cell necrosis was observed
in the ileum and colon of some animals, along with inflammatory cell infiltration in the stomach
and GI tract. No adverse effects were observed in the GI tract at 30 mg/kg/day but PK modelling
determined that this would provide insufficient safety cover for predicted clinical exposures and
further evaluation of compound 19 was therefore terminated.
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
Compound 19 had been subjected to more extensive profiling and despite achieving selectivity
against >350 kinases (see Supporting Information), profiling in a chemoproteomics-based lipid
29
kinobeads assay using a mass spectrometry readout showed PI3Kδ binding was inhibited less
than 10-fold over the Vps34 class III PI3K complex (PI3Kδ pKdapp = 9.6 vs. Vps34 pKdapp = 8.7,
Figure 4).
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 84
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 4. Selectivity of compound 19 against lipid and atypical kinases determined using lipid
kinobeads and mass spectrometry readout (n = 2), revealing high potency for PI3Kδ with the
class III PI3K complex Vps34 within a 10-fold window. pKdapp for detected but not competed
proteins set to 5.0.
Vps34 (PIK3C3) is the evolutionary highly conserved, only class III PI3K member, acting via
phosphorylation of the membrane lipid phosphatidylinositol to PtdIns3P. In combination with
other protein binding partners it fulfills several roles in the regulation of autophagy, endosomal
22
ACS Paragon Plus Environment

Page 23 of 84
Journal of Medicinal Chemistry
1
2
3
30
sorting/trafficking, phagocytosis as well as macropinocytosis. Vps34 forms at least two core
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
31
complexes (I and II) that both contain Vps34, beclin 1 (BECN1) and Vps15 (PI3R4). Complex
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
I is additionally composed of ATG14L (BARKOR, KIAA0831) and the accessory subunit NRBF2,
whereas complex II is complemented by UVRAG and rubicon (KIAA0226). In the lipid kinobeads
assay both complexes are captured via Vps34 itself and competed in their binding by compound
1
9 with the same potency, making the complexes indistinguishable. Extensive studies on Vps34
in the last decade have revealed an intricate role of the lipid kinase that is only partially
understood. Both global genomic knockout as well as homozygous expression of kinase-dead
32
Vps34 are embryonic lethal. Heterozygous kinase-dead knock-in mice were found to be viable
and showed an increased sensitivity to insulin, a result that could be mimicked by administration
32
of a tool Vps34 inhibitor at non-saturating concentrations. These results indicate that partial
inhibition of the kinase might be tolerated but must be well-controlled.
Having identified Vps34 as a potential liability, efforts were explored to engineer selectivity
over this related lipid kinase. A crystal structure of Vps34 with a small molecule inhibitor bound
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 84
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
3
3
in the ATP binding site was available (PDB accession code: 3LS8) and was compared to the
crystal structure of PI3Kδ with compound 19 (Figure 5).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Comparison of crystal structures of Vps34 (orange, ligand hidden, PDB ID: 3LS8)33
and compound 19 in PI3Kδ (magenta, PDB ID: 6TNR), highlighting residue differences and
binding site shape similarity. Solvent accessible surface calculated in Molecular Operating
Environment (MOE, version 2018.01).34
24
ACS Paragon Plus Environment

Page 25 of 84
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
After careful superposition of the proteins, some interesting differences were identified (Figure
6
). The sidechain of Phe684 in Vps34 resulted in a much more occluded space close to the
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
hinge binding region of the DHB moiety compared to PI3Kδ (Figure 6, a). It was therefore
proposed that substitution of the core or expansion of the ring size to increase steric bulk would
be better accommodated by PI3Kδ than Vps34. Additionally, it was noted that the Vps34
structure had a more restricted channel where the 4-position substituent of the DHB binds, due
to Phe612 at the top of binding site (Figure 6, b). It is noteworthy that Met804 of PI3Kδ is flexible
and has been observed in various conformations in other ligand-bound PI3Kδ crystal
structures.17,20,35 It was therefore hypothesized that increased steric hindrance close to the DHB
core, or reduced flexibility of the linker, could be better tolerated in PI3Kδ than Vps34.
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 84
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. (a) Top view of the binding site - black arrow highlighting the occluded hinge region in
33
Vps34 (orange, PDB ID: 3LS8) which wraps more closely around the DHB core compared to
PI3Kδ (magenta, PDB ID: 6TNR). (b) Comparison of the binding site cavity surface around the
4-position vector from the DHB core - black arrow indicates a more restricted channel in the
33
Vps34 structure (orange, PDB ID: 3LS8) compared to PI3Kδ (magenta, PDB ID: 6TNR).
The strategies to improve selectivity for PI3Kδ over Vps34 were explored and the results are
summarized in Table 4. To enable a systematic analysis of the SAR of the binding to the Vps34
complexes, a medium-throughput antibody-based chemoproteomic assay was developed.
26
ACS Paragon Plus Environment

Page 27 of 84
Journal of Medicinal Chemistry
1
2
3
Compound 19 showed similar potency against both PI3Kδ (pIC = 8.9) and Vps34 (pIC = 8.7)
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
50
50
in this assay.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O H
N
R2
R1
S
O
O
N
pIC50*
PI3Kδa
Compound
Number
PI3K PI3K PI3K PI3K
hWB
R1
R2
/Vps34
δ
α
β
γ
IFNγ
a
HO
N
or1
8.3/8.8
O
7.1b
2
8
9
7.6
5.0
5.1
5.0
O
c
O
HO
N
or1
O
8.5/8.5 7.6b
2
7.8
7.3
5.2
5.5
5.2
4.9
O
O
HO
N
8.5/7.6
or1
4.8c
5.6c
O
30
6.8
O
c
O
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 84
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
HO
N
or1
O
3
1
6.9
5.3
5.6
5.3
NT
6.4
O
O
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
HO
N
8
.4/6.4
d
3
2
3
Me
8.1
7.7
5.6
5.1
5.6
5.3
5.5
5.4
7.4
7.6
O
O
HO
N
O
3
NT
O
O
*
Data is mean of at least 3 separate test occasions unless stated. NT = not tested.
a
b
Chemoproteomic competition binding assay with antibody-based read-out. Did not return a
c
d
value on one test occasion. Mean of two test occasions. Data from one test occasion.
Table 4. Hinge and ether linker modifications.
The addition of methyl substituents to the 4-position linker moiety demonstrated that steric bulk
adjacent to the nitrogen (compounds 28 and 29) had no effect on Vps34 activity whereas steric
bulk adjacent to the oxygen led to a decrease in Vps34 activity and conferred 8-fold selectivity
for PI3Kδ over Vps34 (compound 30). Compound 31, the opposite isomer of compound 30, was
2
8
ACS Paragon Plus Environment

Page 29 of 84
Journal of Medicinal Chemistry
1
2
3
less potent and the PI3K isoform selectivity of both compounds was not optimal, for example
only 50-fold selectivity for PI3Kδ over PI3Kβ was observed for compound 30.
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
Increasing the ring size of the hinge binder to give the isochroman core (compound 32)
preserved the desired PI3Kδ potency and ensured >100-fold selectivity over the other PI3K
isoforms. Furthermore, this led to a much improved selectivity over Vps34 (100-fold), validating
the hypothesis that this could come from increased ligand size at the hinge region. However,
compound 25 (1-methyl substituted DHB) was also evaluated and was equipotent at both PI3Kδ
(pIC = 8.3) and Vps34 (pIC = 8.3), highlighting the need to add steric bulk in the correct region
50
50
of the binding site.
Despite having the same lipophilicity (clogP = 2.9) as that of compound 19, compound 32 had
very high in vivo clearance of 109 ml/min/kg in the Wistar Han male rat following administration
of a 1 mg/kg i.v. dose. Addition of a methoxyethyl substituent from the sulfonamide in compound
33 led to a similar potency and isoform selectivity profile, as well as similar lipophilicity (clogP =
3.0), but lower in vivo clearance of 41 ml/min/kg (Table 6). Compound 33 was considered for
screening in additional assays to understand the wider profile. Finally, ethers based on
29
ACS Paragon Plus Environment