Optimization of Novel Indazoles as Highly Potent and Selective Inhibitors of Phosphoinositide 3-Kinase δ for the Treatment of Respiratory Disease

Article abstract of DOI:10.1021/acs.jmedchem.5b00767

Optimization of lead compound 1, through extensive use of structure-based design and a focus on PI3Kδ potency, isoform selectivity, and inhaled PK properties, led to the discovery of clinical candidates 2 (GSK2269557) and 3 (GSK2292767) for the treatment of respiratory indications via inhalation. Compounds 2 and 3 are both highly selective for PI3Kδ over the closely related isoforms and are active in a disease relevant brown Norway rat acute OVA model of Th2-driven lung inflammation.

Full text of DOI:10.1021/acs.jmedchem.5b00767

Subscriber access provided by UNIV OF NEBRASKA - LINCOLN
Article
Optimization of Novel Indazoles as Highly Potent and Selective Inhibitors of
Phosphoinostide-3-Kinase Delta for the Treatment of Respiratory Disease
Kenneth David Down, Augustin Amour, Ian R Baldwin, Anthony W.J. Cooper, Angela M Deakin, Leigh M
Felton, Stephen B Guntrip, Charlotte Hardy, Zoë A Harrison, Katherine L. Jones, Paul Jones, Suzanne
E Keeling, Joelle Le, Stefano Livia, Fiona Lucas, Christopher J Lunniss, Nigel J Parr, Ed Robinson,
Paul Rowland, Sarah Smith, Daniel A Thomas, Giovanni Vitulli, Yoshiaki Washio, and Nicole Hamblin
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.5b00767 • Publication Date (Web): 24 Aug 2015
Downloaded from http://pubs.acs.org on August 25, 2015
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 free 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 accessible to all
readers and 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.
Journal of Medicinal Chemistry 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 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 59
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
Optimization of Novel Indazoles as Highly Potent and
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
Selective Inhibitors of Phosphoinositide 3ꢀKinase Delta
for the Treatment of Respiratory Disease
$
$
†
$
¦
Kenneth Down , Augustin Amour , Ian R. Baldwin , Anthony W. J. Cooper , Angela M. Deakin , Leigh
‡
$1
$
$
ǁ
$
M. Felton , Stephen B. Guntrip , Charlotte Hardy , Zoë A. Harrison , Katherine L. Jones , Paul Jones ,
$
#
$
$
$
ǁ
Suzanne E. Keeling , Joelle Le , Stefano Livia , Fiona Lucas , Christopher J. Lunniss , Nigel J. Parr ,
$
†
^
^
$
†
Ed Robinson , Paul Rowland , Sarah Smith , Daniel A. Thomas , Giovanni Vitulli , Yoshiaki Washio , J.
$,
Nicole Hamblin *
$
¦
Refractory Respiratory Inflammation DPU, and Allergic Inflammation DPU, Respiratory Therapeutic
Area, GlaxoSmithKline R&D, Gunnels Wood Road, Stevenage, SG1 2NY, UK.
†
^
#
Molecular Discovery Research, Biological Sciences, and Computational Chemistry, Platform
Technology & Science, GlaxoSmithKline R&D, Gunnels Wood Road, Stevenage, SG1 2NY, UK.
‡
ǁ
Experimental Medicine Unit, and Epinova DPU, ImmunoInflammation Therapeutic Area,
GlaxoSmithKline R&D, Gunnels Wood Road, Stevenage, SG1 2NY, UK.
1
Present addess: Argenta Discovery, Harlow, Essex, CM19 5TR, UK.
ACS Paragon Plus Environment
1

Page 3 of 59
Journal of Medicinal Chemistry
1
2
3
Abstract
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
Optimisation of lead compound 1, through extensive use of structureꢀbased design and a focus on PI3Kδ
potency, isoform selectivity and inhaled PK properties, led to the discovery of clinical candidates 2
(GSK2269557) and 3 (GSK2292767) for the treatment of respiratory indications via inhalation.
Compounds 2 and 3 are both highly selective for PI3Kδ over the closely related isoforms and are active
in a disease relevant Brown Norway Rat acute OVA model of Th2ꢀdriven lung inflammation.
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
Introduction
Phosphoinositide 3ꢀkinase delta (PI3Kδ) is a lipid kinase belonging to the Class 1 PI3K family, along
1,2
with the closely homologous isoforms α, β and γ.
This family of enzymes catalyzes the
phosphorylation of phosphatidylinositolꢀ4,5ꢀbisphosphate to produce the key signalling molecule
phosphatidylinositolꢀ3,4,5ꢀtriphosphate triggering a series of downstream biological events that
3,4,5
ultimately impact on cell growth, proliferation, chemotaxis, differentiation and survival.
PI3Kδ is highly enriched in leukocytes, making it an attractive target for the treatment of
6
7
inflammatory conditions, such as asthma , chronic obstructive pulmonary disease (COPD) and
8
autoimmune diseases. Furthermore, both PI3Kδ knockꢀout and kinase dead knockꢀin mice are viable
9
and have a phenotype consistent with immunological effects on Bꢀcells, Tꢀcells and neutrophils. PI3Kγ
1
0
is also a reported target in inflammatory disease and mutagenic mice are again viable, however PI3Kα
and β knockꢀout mice are embryonically lethal, indicating the importance of subtype specific inhibitors
9
for therapeutic intervention. Targeting PI3Kδ is also of high interest for oncology indications, such that
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 59
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
11
GSꢀ1101 (idelalisib, formally Calꢀ101) has recently been approved for the treatment of patients with
relapsed chronic lymphoid leukemia and nonꢀHodgkin lymphoma.
Herein we describe the medicinal chemistry program, starting from compound 1, that ultimately led to
the discovery of two inhaled clinical candidates GSK2269557 (2) and GSK2292767 (3) (Fig. 1) for the
treatment of respiratory diseases such as asthma and COPD.
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
2
3
Figure 1. Structures of 1, 2 and 3.
Results and Discussion
Compound 1 was a potential lead molecule that was discovered via SAR investigation against an unꢀ
12
related kinase, triggered by data from kinase crossꢀscreening (manuscript in preparation). We selected
compound 1 as a lead due to its promising selectivity profile for PI3Kδ over the closely related PI3K
isoforms, α, β and γ (Table 1). Given the critical role of PI3Kδ in immune responses and keen at the
inception of this work to avoid any potential negative impact of broad systemic inhibition of this
ACS Paragon Plus Environment
3

Page 5 of 59
Journal of Medicinal Chemistry
1
2
3
biology, we embarked upon a lead optimization program with the aim of delivering an inhaled clinical
candidate.
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
We set ourselves a challenging target profile, such that throughout the optimization program we
focused on achieving compounds that had:
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
• High selectivity (>100 fold) for PI3Kδ over its related isoforms (α, β and γ)
•
High cellular potency (pIC > 9)
50
• Moderate to high intrinsic clearance and low oral bioavailability
Inhibition of PI3K enzymatic activity was determined using a homogeneous time resolved
1
3
fluorescence (HTRF) assay format. As well as PI3K isoform selectivity we also sought compounds
that were selective over the wider kinome and we obtained this data on selected compounds using
available inꢀhouse and external assays. To measure cellular activity for compounds of interest, a
peripheral blood mononuclear (PBMC) assay was adopted using cytostim to stimulate cytokine
production from the Tꢀlymphocyte compartment. Interferon gamma (IFNγ) was selected as an optimal
analyte owing to robust stimulation by cytostim and exquisite sensitivity to PI3Kδ inhibition. We
required high potency in this assay in order to minimize the required dose for inhaled delivery. The
target PK profile was intended to minimize systemic circulation after inhaled delivery; moderate to high
intrinsic clearance in order to ensure removal of drug once absorbed through the lung and low oral
1
4
bioavailability to limit absorption of the swallowed fraction of the inhaled dose.
We chose to explore the SAR at the two positions of substitution on the indazole core. To begin with,
a set of 4ꢀposition amide modifications were prepared whilst keeping the 6ꢀindole substituent fixed. The
results are summarized in Table 1. Phenyl analogue 4 showed an approximate 10ꢀfold drop in potency
compared to compound 1, indicating the potential importance of the ortho heteroatom. This was
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 59
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
substantiated by thiazole analogue 5 and 2ꢀmethylpyrazole analogue 6, both of which displayed similar
potency and selectivity to compound 1. We postulated that this may be due to an internal hydrogen bond
that maintains the planarity of the amide and thus enhances the stability of the preferred binding
conformation. Wanting to probe this further and due to the lack of a suitable crystal structure of
PI3Kδ at this time, a homology model was built to enable structureꢀbased design and aid the
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
15
rationalization of SAR. The model (Fig. 2) suggested that the planar conformation of the amide allows
the compounds to be accommodated under the glycine rich loop in the vicinity of bulky residues Met752
and Trp760. The apparent requirement for planarity was further investigated by methylation of the
amide nitrogen atom to give compound 7 which caused a greater than 100ꢀfold reduction in potency.
The 1ꢀmethylpyrazole analogue 8 and the 1ꢀisopropylpyrazole analogue 9 demonstrated a 10ꢀfold drop
in potency compared to compound 6 further corroborating the hypothesis that a planar amide
conformation was important for potency.
1
5
Figure 2. Compound 1 docked into a PI3Kδ homology model displaying the planarity of the amide.
Coloured shading represents a lipophilic (brown), neutral (white) or hydrophilic (blue) region of the
ACS Paragon Plus Environment
5

Page 7 of 59
Journal of Medicinal Chemistry
1
2
3
protein surface. Hydrogen bonds are highlighted with dashed lines between the indazole and the hinge
region (Val828) and the indole and Asp787.
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
Methylation of compounds 1 and 5 to give compounds 10 and 11, respectively, gave an increase in
potency and selectivity that indicated an interesting lipophilic vector to pursue. In contrast, 5ꢀ
methylpyridine analogue 12 showed no increase in potency compared to compound 1, indicating the
specificity of the substitution vector. More importantly, this initial SAR exploration suggested that
PI3Kδ potency could be increased without substantially increasing the potency of the related isoforms,
for instance the selectivity of compound 11 was approximately 100ꢀfold or better against PI3Kα, β and
γ.
Table 1. 4ꢀPosition amide modifications to give compounds 4 – 12.
a
pIC50
Cmpd
R1
R2
H
PI3Kδ
PI3Kα
PI3Kβ
PI3Kγ
b
c
1
7.0 ± 0.14 (5) 5.0 ± 0.10 (4) 5.2 ± 0.25 (4) 5.2 ± 0.11 (5)
c
c
4
H
5.9 ± 0.13 (3)
<4.6 (2)
4.7 (2)
4.9 ± 0.04 (3)
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 59
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
5
6
H
H
6.8 ± 0.20 (4)
<4.6 (3)
<4.6 (3)
4.9 ± 0.16 (3)
5.0 ± 0.15 (4)
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
c
b
7.0 ± 0.11 (4)
4.9 ± 0.07 (4) 5.0 ± 0.12 (4)
7
Me
4.8 (1)
<4.6 (1)
<4.6 (6)
<4.6 (1)
<4.6 (1)
c
8
9
H
H
H
H
H
6.1 ± 0.11 (6)
6.0 ± 0.23 (3)
4.7 ± 0.04 (5) 4.7 ± 0.04 (5)
b
c
<4.6 (3)
4.6 (3)
4.7 ± 0.07 (3)
c
1
0
1
2
7.8 ± 0.09 (6) 4.7 ± 0.06 (4) 5.5 ± 0.05 (4)
6.1 ± 0.08 (5)
c
1
1
7.3 ± 0.24 (7)
<4.6 (3)
<4.6 (2)
5.0 ± 0.04 (3) 5.4 ± 0.11 (7)
6.9 ± 0.07 (2)
4.8 ± 0.06 (2)
5.9 ± 0.02 (2)
a
b
c
Mean ± SEM (number of test occasions). Tested <4.6 on 2 occasions. Tested <4.6 on 1 occasion.
To assess the suitability of the series for inhaled delivery in vitro clearance data in rat microsomes and
subsequently in vivo pharmacokinetic data from Sprague Dawley male rats was obtained. Compounds
ACS Paragon Plus Environment
7

Page 9 of 59
Journal of Medicinal Chemistry
1
2
3
were administered by the oral or intravenous routes, at a dose level of 3 and 1mg/kg respectively (n=2
rats/route). As well as the good correlation between in vitro and in vivo clearance (Table 2), an
interesting correlation between the torsion angle of the amide (Fig. 3) and the in vivo stability of the
series was observed, such that increasing the torsion angle (e.g. compounds 6 to 8 to 9) resulted in
decreased clearance and increased bioavailability. This data implied that amide hydrolysis was a key
metabolic pathway and we were subsequently able to demonstrate this by spiking with the resultant
amine during the analysis of the PK samples. Pleasingly, moderateꢀtoꢀhigh clearance and low oral
bioavailability were observed for preferred compounds 6, 10 and 11, suggesting that maintaining the
planarity of the amide was beneficial not only for potency but also for the intended PK profile. Kinetic
solubility was measured using a high throughput assay with chemiluminescent nitrogen detection
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
1
6
(CLND) . Compounds 10 and 11 both displayed low solubility (2 and 12ꢁg/ml, respectively).
Figure 3. Torsion angle of the amide bond.
Table 2. Amide torsion angle and pharmacokinetic data for selected compounds.
In vitro
metabolism
ml/min/g liver)
Clb
(ml/min/kg)
b
Amide torsion angle
Cmpd
b
F%
c
a
(degrees)
(
6
8
9
NR
2
36
16
9
1
0
19
29
14
23
<1
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 59
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
11
6
49
28
2
2
0
0
11
a
Compounds were tested for metabolic stability in rat microsomes at a concentration of 0.5ꢀM in the
incubation. Compounds were tested in Sprague Dawley male rats and were administered discretely by
the oral and intravenous routes, at a dose level of 3 and 1mg/kg respectively (n=2 rats/route). Dihedral
angle after optimization at the DFTꢀB3LYP/6ꢀ31G** level of theory.
b
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
c
We next switched our attention to the 6ꢀposition of the indazole template in search of further
enhancement of potency and selectivity. Keeping the methyl thiazole amide of compound 11 in place, an
array of compounds was synthesized which are summarized in Table 3.
Table 3. 6ꢀPosition indole modifications to give compounds 13 – 21.
a
pIC50
b
Cmpd
R1
PI3Kδ
PI3Kα
PI3Kβ
PI3Kγ
clogP
4.1
c
1
1
7.3 ± 0.24 (7)
<4.6 (3)
5.0 ± 0.04 (3)
5.4 ± 0.11 (7)
13
6.5 ± 0.08 (2)
4.8 ± 0.07 (2)
5.1 ± 0.02 (2)
4.9 ± 0.13 (2)
3.5
ACS Paragon Plus Environment
9

Page 11 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
d
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
14
5.3 ± 0.04 (4)
<4.6 (4)
<4.6 (4)
<4.6 (4)
5.5 ± 0.15 (5)
<4.6 (1)
4.6
4.6
3.6
3.6
3.2
4.4
5.0
4.2
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
c
1
5
7.7 ± 0.10 (5) 5.1 ± 0.08 (5)
5.5 ± 0.05 (5)
<4.6 (1)
16
6.5 (1)
<4.6 (1)
<4.6 (7)
N
HN
1
1
1
2
7
8
9
0
6.5 ± 0.08 (5)
<4.6 (5)
5.2 ± 0.19 (5)
6.6 ± 0.06 (8)
c
8.1 ± 0.10 (8) 5.1 ± 0.06 (12) 5.3 ± 0.06 (12)
7.8 ± 0.09 (10) 4.9 ± 0.03 (12) 5.5 ± 0.06 (11) 5.5 ± 0.09 (10)
8.1 ± 0.12 (5)
7.9 ± 0.10 (4)
5.3 ± 0.08 (5)
5.2 ± 0.10 (4)
6.0 ± 0.03 (5)
5.6 ± 0.05 (4)
5.4 ± 0.06 (5)
21
5.1 ± 0.05 (4)
a
b
17 c
Mean ± SEM (number of test occasions). Calculated using Biobyte Version 4.0. Tested <4.6 on 1
d
occasion. Tested 4.7 on 1 occasion.
ACS Paragon Plus Environment
10

Journal of Medicinal Chemistry
Page 12 of 59
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
Metaꢀphenol analogue 13 showed a loss of potency relative to compound 11 and this observation held
true for all of the monoꢀcyclic indole replacements we prepared. NꢀMethyl indole analogue 14 showed a
dramatic 100ꢀfold loss of potency. Examination of the docking model (Fig. 4) suggests steric constraints
in this region of the binding site and also that a potentially important Hꢀbond interaction to Asp787
might be disrupted. 2ꢀMethylindole analogue 15 was predicted by modeling to fill a small lipophilic
pocket and gave a small increase in potency, albeit at the price of an increase in lipophilicity.
Benzimidazole analogue 16 and indazole analogue 17 showed reduced potency relative to compound 11,
potentially due to reduced Hꢀbond donating ability compared to the indole or maybe as a consequence of
introducing polarity into a lipophilic pocket. 7ꢀAzaindole analogue 18 displayed an 8ꢀfold increase in
potency, however the PI3Kγ potency also increased and this was indicative of poor wider kinase
selectivity later discovered for this compound. This could potentially be explained by the ability of the
7ꢀazaindole to act as a hingeꢀbinder itself, allowing compound 18 to adopt different binding
conformations across the human kinome. Modeling suggested substitution in the 6ꢀposition of the indole
looked feasible and indeed, incorporation of fluorine (19), chlorine (20) or methoxy (21) substituents all
increased potency whilst maintaining the selectivity profile over the other isoforms. However,
compound 20 also had increased lipophilicity compared to compound 11, so we did not see any
advantage in using this group in subsequent compounds and, whilst compounds 19 and 21 did not
substantially increase lipophilicity, they still did not achieve our target 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
ACS Paragon Plus Environment
11

Page 13 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Closeꢀup of the hinge and back pocket region of compound 1 docked in PI3Kδ. Hydrogen
bonds are highlighted with dashed lines. Coloured shading represents a lipophilic (brown), neutral
(white) or hydrophilic (blue) region of the protein surface.
Unable to significantly improve on the indole moiety at the 6ꢀposition of the indazole template, we
turned our attention back to the 4ꢀposition. Careful reꢀevaluation of SAR, coupled with closer
examination of compounds modeled into PI3Kδ, indicated that there was potentially space to substitute
on the heteroaryl ring to search for further interactions. Table 4 summarizes our efforts in the area.
Unable to differentiate between compounds 10 and 11; extended compounds with both the pyridyl and
thiazole heteroaryl groups in place were prepared.
Table 4. Heteroaryl amide substitutions to give compounds 22 – 29.
ACS Paragon Plus Environment
12

Journal of Medicinal Chemistry
Page 14 of 59
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
a
pIC50
Cmpd
R1
PI3Kδ
PI3Kα
PI3Kβ
PI3Kγ
b
1
1
2
7.3 ± 0.24 (7)
<4.6 (3)
5.0 ± 0.04 (3) 5.4 ± 0.11 (7)
2
6.3 ± 0.03 (2)
7.1 ± 0.26 (2)
<4.3 (1)
4.9 ± 0.24 (2)
<4.6 (2)
5.1 ± 0.38 (2)
b
23
<4.6 (1)
4.9 (2)
b
b
24
25
26
27
8.0 ± 0.16 (7) 4.7 ± 0.03 (3) 4.8 ± 0.04 (3) 4.8 ± 0.08 (6)
c
7.9 ± 0.07 (14)
<4.6 (18)
4.9 ± 0.04 (16)
<4.6 (16)
d
be
8.2 ± 0.07 (8) 5.0 ± 0.04 (8) 5.1 ± 0.08 (8) 5.5 ± 0.06 (10)
8.4 ± 0.03 (5)
5.1 ± 0.04 (6)
5.1 ± 0.03 (5)
5 ± 0.09 (5)
ACS Paragon Plus Environment
13

Page 15 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
b
2
8
8.5 ± 0.01 (2)
8.7 ± 0.08 (7)
5.9 ± 0.01 (2)
5.8 ± 0.03 (2) 5.4 ± 0.07 (3)
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
29
4.9 ± 0.05 (8)
5.3 ± 0.05 (8)
5.6 ± 0.12 (9)
a
b
c
Mean ± SEM (number of test occasions). Tested <4.6 on 1 occasion. Tested <4.6 on 3
d
e
occasions. Tested <4.6 on 2 occasions. Tested <5.2 on 1 occasion.
Adding further aromatic rings in this region did not enhance potency. For instance, changing the
methyl substituent of compound 11 to phenyl (to give compound 22) reduced potency 10ꢀfold, whilst
benzyl substituted analogue 23 was equipotent, despite the large molecular weight increase. A piperidine
substituent on the methyl group of compounds 10 and 11 to give compounds 24 and 25, respectively,
showed an increase in potency whilst also enhancing the selectivity. However compound 25, was
significantly more active in a hERG binding assay (pIC 6) than compound 24 (pIC <4.3), a trend that
5
0
50
was repeated with other extended pyridyl analogues in the series. For that reason subsequent compounds
focused on maintaining the thiazole group. There was little difference in potency or selectivity between
piperidine analogue 24 and morpholine analogue 26 in this region and bicyclic amines could also be
introduced (such as compound 27) suggesting that moderation of the lipophilicity and basicity of the
compounds in this region was achievable, without drastically altering the potency and selectivity profile.
Compound 26 displayed good potency (pIC 7.8) in the PBMC assay, albeit below our stated target,
5
0
low CLND solubility (8ꢁg/ml) and high in vitro clearance (11ml/min/g) in rat microsomes.
ACS Paragon Plus Environment
14

Journal of Medicinal Chemistry
Page 16 of 59
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 5. Comparison of PI3Kδ (left) and PI3Kγ (right) with compound 11 docked in to the active site,
highlighting the different residue aligned adjacent to the conserved tryptophan (Thr750 for PI3Kδ and
Lys802 for PI3Kγ). Hydrogen bonds are highlighted with dashed lines.
On further examination of the modeling we hypothesized that the lipophilic basic amine, which is
predicted to be charged at physiological pH, would align next to Trp760 (conserved in PI3Kα, β and γ),
forming a charge to piꢀelectron cloud interaction. This interaction is accessible in PI3Kδ due to the fact
that the residue adjacent to Trp760 is Thr750. However in PI3Kα, β and γ the residue adjacent to the Trp
is Arg, Lys and Lys, respectively (shown for PI3Kγ in Fig. 5), and we speculated that these larger
residues could clash with the bulky cyclic amine group. This would preclude the compounds from
adopting the same binding mode in these PI3K isoforms compared to PI3Kδ and explains the enhanced
potency increase for PI3Kδ over α, β and γ. This region of the protein is solvent exposed which allows
the compounds still to bind in PI3Kα, β and γ. This Trp760 interaction has been recently reported by
18
Sutherlin et al. for an unrelated series of PI3Kδ inhibitors .
ACS Paragon Plus Environment
15

Page 17 of 59
Journal of Medicinal Chemistry
1
2
3
To take advantage of this hypothesis we next prepared compounds with additional bulk around the
pendant amine, hoping to drive potency and selectivity further and compounds 28 and 29 were indeed
highly potent and exquisitely selective. They were also highly potent in the PBMC assay (pIC 8.5
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
5
0
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=5) and 8.7 (n=8), respectively) and had increased CLND solubility (83 and 20ꢁg/ml, respectively)
compared to compound 26. The wider kinase selectivity profiles of both compounds were encouraging,
with only ALK5 potency (pIC50 7) within 100ꢀfold of compound 28 in a panel of 20 kinases and
1
9
compound 29 was greater than 100ꢀfold selective across all kinases in a similar panel . However these
compounds did not reach our stated potency goal and therefore we continued our exploration of the
SAR.
We next hypothesized that we could potentially access a more direct route to this important Trp760
interaction using a heterocycle (exemplified in Fig. 6), thereby reducing the size of the compounds.
Building on our hypothesis from the amide series that planarity may be required for potency we
prepared a series of 5ꢀmembered heteroaryl groups. These compounds are summarized in Table 5.
ACS Paragon Plus Environment
16

Journal of Medicinal Chemistry
Page 18 of 59
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
Figure 6. Overlay of compound 28 (pink) and compound 2 (green) docked in the PI3Kδ homology
model, highlighting the more direct route to access the selectivity pocket adjacent to Trp760. Hydrogen
bonds are highlighted with dashed lines.
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
Whilst the potency range amongst the heteroaryl groups was narrow, trends did emerge to guide
subsequent compound design. Oxazole analogue 30 demonstrated good potency for PI3Kδ, on a par
with compound 11, albeit with reduced selectivity. Addition of methyl groups to the oxazole in either
the 4ꢀ or 5ꢀposition (compounds 31 and 32, respectively) had no significant impact on potency or
selectivity. Oxadiazole analogue 33 displayed similar potency and selectivity to oxazoles 30 – 33 and in
vitro rat microsomal stability was 7ml/min/kg, demonstrating that removal of the amide to directly
attach the heteroaryl had not impacted on the target profile of high intrinsic clearance. Compounds 30 –
3
4 suggested that the Hꢀbond donor of the amides was not necessary for potency and the presence of an
Hꢀbond acceptor could be sufficient to maintain potency. However, furan analogue 34 was equipotent to
compounds 31 and 33, suggesting that the Hꢀbond acceptor (furan is at best only a very weak Hꢀbond
2
0
acceptor) was also not necessary and we therefore assumed that the key driver for potency was
maintaining the planarity of the heteroaryl ring. In order to probe our hypothesis further we prepared
thiophene analogue 35, predicted to have a nonꢀplanar preferred conformation, and this was 10ꢀfold less
potent than the oxygen containing heterocycles. Further supporting our hypothesis, methylꢀsubstituted
thiazole analogues 36 and 37 recovered some potency relative to compound 35, presumably as some
planarity was reintroduced, however these compounds were still inferior to the oxazole and oxadiazole.
Table 5. Heteroaryl amide replacements to give compounds 30 – 37.
ACS Paragon Plus Environment
17

Page 19 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a
pIC50
Cmpd
R1
PI3Kδ
PI3Kα
PI3Kβ
PI3Kγ
b
11
7.3 ± 0.24 (7)
<4.6 (3)
5.0 ± 0.04 (3)
5.9 ± 0.04 (5)
6.1 ± 0.08 (4)
5.4 ± 0.11 (7)
5.9 ± 0.11 (5)
6.2 ± 0.12 (4)
b
30
7.1 ± 0.21 (5) 5.0 ± 0.10 (5)
6.9 ± 0.14 (4) 4.9 ± 0.05 (4)
31
32
33
7.2 ± 0.13 (4) 5.0 ± 0.07 (4)
5.9 ± 0.13 (4)
5.5 ± 0.06 (5)
5.6 ± 0.22 (3)
6.4 ± 0.12 (4)
5.9 ± 0.09 (5)
5.7 ± 0.13 (3)
c
7.0 ± 0.21 (5)
<4.6 (5)
3
4
5
6.9 ± 0.38 (3) 5.3 ± 0.14 (3)
3
6.1 ± 0.11 (3) 5.1 ± 0.08 (3)
5.2 ± 0.33 (3)
5.2 ± 0.32 (3)
5.4 ± 0.17 (3)
5.6 ± 0.00 (2)
5.4 ±0.08 (3)
d
3
6
7
6.4 ± 0.19 (3)
<4.6 (3)
3
6.6 ± 0.21 (3) 5.0 ± 0.14 (3)
5.2 ± 0.21 (3)
a
b
c
Mean ± SEM (number of test occasions). Tested <4.6 on 1 occasion. Tested 4.8 on 1 occasion.
d
Tested 5.5 on 1 occasion.
ACS Paragon Plus Environment
18

Journal of Medicinal Chemistry
Page 20 of 59
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
From these starting points we hoped that by addition of the groups previously identified to interact
with the tryptophan we could improve the PI3Kδ potency above that of the other isoforms. Compounds
with the extended amine on the oxadiazole and in both the 2ꢀ and 3ꢀpositions of the oxazole were
prepared and are summarized in Table 6. Whilst compound 34 was useful for SAR purposes, the
reported potential toxicological risks of furan dissuaded us from incorporating it into further
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
21
compounds .
Table 6. Extended amine substitution on selected heteroaryl groups to give compounds 2 and 38 – 42.
a
pIC50
PI3Kδ
PBMC
IFNγ
Cmpd
R1
PI3Kδ
PI3Kα
PI3Kβ
PI3Kγ
b
pKi
c
8.4
4.9
5.4
4.8
7.9
3
3
8
9
ꢀ
ꢀ
±
0.00 (3)
± 0.03 (4)
± 0.10 (4)
± 0.02 (4)
± 0.22 (8)
d
8.3
± 0.02 (2)
4.9
(1)
5.9
± 0.03 (2)
5.6
± 0.02 (2)
7.1
± 0.26 (8)
ACS Paragon Plus Environment
19

Page 21 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
e
f
8.9
4.9
± 0.09 (8)
5.5
± 0.11 (6)
5.0
± 0.04 (7)
9.3
4
0
ꢀ
8
9
±
0.05 (5)
± 0.30 (10)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
O
N
e
g
a
f
8
.5
<4.6
6.0
5.7
8.3
4
1
N
ꢀ
±
0.19 (15)
(17)
± 0.07 (17) ± 0.02 (15) ± 0.19 (13)
N
O
eh
e
i
9.9
5.3
5.8
5.2
9.7
±0.21 (10)
2
ꢀ
±
0.09 (4) ± 0.06 (29) ± 0.13 (27) ± 0.05 (33)
O
N
fj
8
.6
4.8
6.0
± 0.08 (3)
5.7
± 0.08 (3)
8.6
4
2
ꢀ
± 0.06 (3)
± 0.09 (3)
± 0.28 (12)
N
O
a
b
Mean ± SEM (number of test occasions). The standard PI3Kδ assay was run at a 2mM ATP
c
d
e
concentration. Tested <4.6 on 2 occasions. Tested >8.1 on 1 occasion. Tested <4.6 on 1 occasion.
f
g
h
i
Tested >8.1 on 3 occasions. Tested average of 4.7 on 4 occasions. Tested <5.3 on 1 occasion. Tested
<4.6 on 3 occasions. Tested >9.8 on 1 occasion and <6 one 1 occasion.
j
Oxazoles substituted in the 4ꢀposition with the preferred amines identified in compounds 28 and 29,
were thus prepared. Comparable potency and selectivity were obtained but, disappointingly compounds
38 and 39 were less active in the PBMC assay than their amide counterparts. Substituted oxadiazoles 40
and 41 had similar potency to compounds 28 and 29 and this activity translated to the PBMC assay.
Compound 42, an oxazole substituted in the 5ꢀposition, again showed similar potency in the enzyme and
ACS Paragon Plus Environment
20

Journal of Medicinal Chemistry
Page 22 of 59
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
PBMC assay to its amide analogue 29. Oxazole compound 2, however, was significantly more potent
than its amide analogue 28 and the 4ꢀsubstituted oxazole 38, in both the enzyme and PBMC assays. The
standard assay was modified to run at 2mM ATP concentration in order to be able to determine a pKi.
These compounds confirmed our approach to successfully shortcut to the tryptophan interaction with an
appropriate heterocycle.
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
Compounds 2 and 40 both met our original potency goal in the PBMC assay (pIC >9) and had
5
0
exquisite selectivity against the PI3K isoforms (Fig. 7; Concentration response plots for compound 2 in
a modified HTRF assay). Both compounds 2 and 40 were also highly CLND soluble (110 and 118ꢁg/ml,
respectively), however compound 2 had marginally less potent hERG binding compared to compound
40 (pIC 5.5 vs. 5.9, respectively) and, coupled with its higher PBMC potency, was selected for wider
5
0
profiling to assess its potential as a clinical candidate.
1
20
PI3Kα
PI3Kβ
PI3Kγ
PI3Kδ
100
80
60
40
20
0
ꢀ
20
0.01 0.1
1
10 100 1000 10000
Concentration (nM)
Figure 7. Concentration response plot of compound 2 against PI3Kα, β, γ and δ in a modified HTRF
2
2
assay using an ATP concentration of 2 mM. The data shown here are from a single experiment
.
ACS Paragon Plus Environment
21

Page 23 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
Wider kinase selectivity was impressive; compound 2 was greater than 1000ꢀfold selective against the
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
2
3
Millipore kinase panel (> 250 kinases in total).
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
The rat PK profile of compound 2 was also encouraging but not ideal based on our stated goals. Low
oral bioavailability (F 2%) was achieved and in vivo clearance of 28ml/min/kg, whilst lower than
initially targeted, still met our criteria for progression. As well as being dibasic and moderately
lipophilic (clogP 4.4), compound 2 also had a high volume of distribution of 6.3L/kg, which we
2
4
hypothesized may lead to beneficial tissue retention properties when delivered to the lung.
We progressed compound 2 to a human lung parenchyma assay wherein finely chopped lung tissue
incubated with the plant lectin phytohaemagglutanin (PHA) for 72 hours induced production of
cytokines including IFNγ and ILꢀ2. This response was inhibited by compound 2 in a concentrationꢀ
dependent manner, returning pIC values of 8.2 (IFNγ) and 8.1 (ILꢀ2).
5
0
In order to further investigate the hERG binding result, compound 2 was profiled in a rabbit cardiac
ventricular wedge assay. Compound 2 induced concentrationꢀdependent increases in QT interval and Tpꢀ
at 0.3 and 1ꢀM (1 and 0.5 Hz) and an increase in QRS at 1ꢀM (2 Hz), however, no treatmentꢀrelated
e
Torsades de Pointes (TdP) arrhythmias were observed in this study. A concentration of 1µM was
calculated to be >300 fold above the predicted human free plasma Cmax at a therapeutic dose of 1 mg per
2
5
day and >150 fold higher than the lung parenchyma assay pIC . As we expected the delivered dose to
5
0
be in the microgram rather than milligram range, we felt this risk was acceptably low.
In a disease relevant Brown Norway Rat acute OVA model of Th2 driven lung inflammation,
compound 2 was shown to protect against eosinophil recruitment with an ED of 67µg/kg (Fig. 8). The
5
0
activity of compound 2 was also assessed using other endpoints in this model including, leukocyte
recruitment to the lung (neutrophils, macrophages, CD4 and CD8 Tꢀlymphocytes at 48 hours) and Th2
ACS Paragon Plus Environment
22

Journal of Medicinal Chemistry
Page 24 of 59
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
cytokines such as ILꢀ13. Compound 2 doseꢀdependently reduced recruitment of all leukocyte
subpopulations and ILꢀ13 in the lungs.
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. Inhibition of eosinophil recruitment by compound 2 in a rat acute OVA model. A trend test
was constructed using analysis of variance which showed a significant decrease of eosinophil numbers
with increasing concentrations of compound 2. Prior to analysis the data from 2 experiments were
standardised so that the average of the control groups were the same. The analysis of variance model
used log eosinophil numbers as the response and the concentrations of compound 2 as a categorical
factor along with the positive control of challenged, untreated animals. An ED was calculated for
50
compound 2 by using the 3 concentrations of compound 2. In addition the positive (OVA challenge, no
treatment) and negative controls (baseline, no challenge) were used to provide more information on the
maximum and minimum responses for the model. The eosinophilic response was logged prior to fitting a
4
parameter logistic model to the data. The fitted curve was used to estimate the ED along with its 95%
50
confidence interval.
ACS Paragon Plus Environment
23

Page 25 of 59
Journal of Medicinal Chemistry
1
2
3
Based on this data, compound 2 was selected for progression to early clinical studies as a novel PI3Kδ
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
inhibitor for the treatment or respiratory diseases by inhaled delivery.
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
Following the successful identification of a first inhaled candidate (compound 2), we set out to
identify a differentiated followꢀup compound, with the combined aims of reducing residual hERG
activity and increasing the in vivo clearance to minimize systemic exposure.
2
6
The classical hERG pharmacophore has been well described and our approach to minimizing
channel blockade focused on modulating the amine pK and reducing lipophilicity. Compound 2 had a
a
measured pK of 8.1 and a clogP of 4.4. In our earlier work at the 4ꢀposition we had identified the
a
morpholine group as a tolerated cyclic amine with reduced basicity; compound 41 for instance had a
calculated pK of 7.1, and so using this a starting point we turned our attention back to replacing the 6ꢀ
a
position indole in order to reduce the lipophilicity. As previously discussed we had been unable to find a
2
7
suitable replacement for the indole, however other internal PI3K programs had since demonstrated
success with a pyridine sulfonamide moiety (Fig. 9) and we looked to explore the utility of this group in
our series, mindful of the need to maintain the exquisite selectivity and potency we had achieved with
the indole group, whilst also targeting reduced lipophilicity. A range of compounds to explore this
region by comparison to compound 11 were prepared, as summarized in Table 7.
Figure 9. Pyridine sulfonamide moiety
Table 7. Exploration of the pyridine sulfonamide moiety to give compounds 43 – 50.
ACS Paragon Plus Environment
24

Journal of Medicinal Chemistry
Page 26 of 59
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
a
pIC50
PBMC
IFNγ
b
Cmpd
R1
PI3Kδ
PI3Kα
PI3Kβ
PI3Kγ
clogP hERG
c
5.0
± 0.04
5.4
7
.3
<4.6
(3)
11
± 0.11
(7)
ꢀꢀꢀꢀ
4.1
2.3
ꢀꢀꢀꢀ
±
0.24 (7)
(3)
d
6
.0
7.3
i
8
.7
6.5
± 0.08 (16)
7.2 ±
0.22 (3)
<4.3
(5)
43
± 0.08
16)
± 0.06
(16)
±
±
0.07 (12)
(
c
6.3
± 0.12
15)
7.6 ±
0.19
(12)
ef
8
.9
6.7
7.3
<4.3
(2)
4
4
4
4
5
6
4.0
3.3
3.4
0.12 (15) ± 0.13 (17)
±0.11 (6)
(
4
.8
± (0.04
2)
6.2
± 0.09 (2)
4.8
± 0.01 (2)
5.4 ±
0.06 (2)
4.9
(1)
ꢀꢀꢀꢀ
(
c
4.7
± 0.01
(2)
6
.0
4.6
5.7
± 0.09 (2)
5.2
(2)
<4.6 (2)
±
0.00 (2)
(2)
6.3
± 0.08 (2)
<4.6
(2)
<4.6
(2)
<4.6
(2)
5.9
± 0.44 (3)
<4.3
(2)
47
3.8
2.0
d
g
d
5.4
± 0.05
7
.5
5.5
6.2 ±
6.4
4.7
(1)
48
±
0.27 (6)
± 0.07 (9)
0.10 (6) ± 0.13 (3)
(7)
ACS Paragon Plus Environment
25

Page 27 of 59
Journal of Medicinal Chemistry
1
2
3
4
5
h
7
.1
5.1
± 0.05 (8)
<4.6
(8)
6.5 ±
0.05 (8) ± 0.12 (5)
5.6
4.6
(2)
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
49
2.1
2.9
±
±
0.05 (8)
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
6
.1
7
.9
6.1
± 0.08 (4)
6.7 ±
0.11 (4) ± 0.15 (8)
7.9
4.7
(1)
50
± 0.08
(4)
0.03 (3)
a
b
17 c
Mean ± SEM (number of test occasions). Calculated using Biobyte Version 4.0. Tested <4.6 on 1
occasion. Tested <4.6 on 2 occasions. Tested <4.3 on 1 occasion. Returned no value on 1 occasion.
Failed to return a value on 2 occasions. Tested 4.58 on 1 occasion. Tested 4.3 on 2 occasions.
d
e
f
g
h
i
Replacing the indole with a pyridine methylsulfonamide to give compound 43 resulted in a greater
than 10ꢀfold increase in potency for PI3Kδ. Compound 43 also had much reduced clogP compared to
compound 11. However, potency at the other PI3K isoforms was also significantly enhanced, such that
compound 43 was no more selective than compound 11. Phenyl sulfonamide 44 demonstrated no
enhancement in potency compared to compound 43 and had the further disadvantage of a clogP similar
to compound 11. Both compounds 43 and 44 had similar PBMC potency with a drop of approximately
30ꢀ40 fold compared to the enzyme assay.
ACS Paragon Plus Environment
26

Journal of Medicinal Chemistry
Page 28 of 59
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
Figure 10. Overlay of docking of compounds 11 (green) and 43 (magenta), highlighting the differences
between the interactions of the pyridine sulfonamide and the indole. Hydrogen bonds are highlighted
with dashed lines.
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
We next sought to understand the key binding elements of the pyridine sulfonamide. Removing the
pyridine nitrogen to give compound 45 resulted in a dramatic loss of potency. We hypothesized that the
nitrogen would make an important interaction through water to backbone residues Tyr813 and Asp787.
An overlay of compounds 11 and 43 (Fig. 10) showed the indole NH of compound 11 could be making a
key Hꢀbond interaction with the backbone, potentially displacing a water molecule with which the
pyridine nitrogen of compound 43 interacts. We hypothesized that the indole would fit the lipophilic
pocket better in this region but, as the pyridinesulfonamide was more potent, there were evidently
multiple factors operating. To probe this further, we next prepared sulfone analogue 46 and Nꢀ
methylsulfonamide analogue 47. These two compounds showed a substantial drop in potency against all
isoforms, suggesting the importance of the sulfonamide NH. Consideration of the binding site led us to
postulate that the sulfonamide (measured pK 6.4, thus predominantly ionized at physiological pH)
a
would form a key chargeꢀcharge interaction with the conserved catalytic lysine (Lys779) and therefore,
disruption of this lysine interaction in compounds 46 and 47 led to a drop in potency. The indole moiety
was not hypothesised to interact with Lys779 and was thus a further explanantion for the enhanced
potency of compound 43 over compound 11. This also provided an explanation for the enhanced
potency in the other PI3K isoforms which are homologous in this region.
To explore further the impact of sulfonamide ionization we investigated the effect of the orthoꢀ
pyridine substituent on potency. As well as moderating the pK of the sulfonamide we postulated that
a
changes in this region would also moderate the Hꢀbond acceptor ability of the pyridyl nitrogen. Methyl
ACS Paragon Plus Environment
27

Page 29 of 59
Journal of Medicinal Chemistry
1
2
3
analogue 48 showed a greater than 10ꢀfold potency drop against all isoforms and this was also reflected
in a drop in the PBMC activity. Hydrogen analogue 49 caused a further small drop in the PI3Kδ potency
which was not reflected in the other PI3K isoforms. However, the PBMC potency dropped a further 10ꢀ
fold. Whilst we had potentially identified a useful indole replacement, the disappointing PBMC potency
of those compounds was considered an issue. A breakthrough came when we introduced an orthoꢀ
methoxy group to give compound 50. Although the PI3K potency of compound 50 was lower at all
isoforms compared to compound 43, compound 50 was actually more potent than compound 43 in 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
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
PBMC assay. We hypothesised that the lower pK of the sulfonamide of compound 43 (measured as 6.4
a
vs 7.8 for compound 50) was impacting on the PBMC potency (presumably through reduced cell
permeation) despite the beneficial effect on PI3Kδ potency. We observed this trend across a number of
compounds such that the lower the pK of the sulfonamide, the greater the difference we saw between
a
the PI3Kδ assay and the PBMC assay. Compound 50 also had a significantly lower clogP than
compound 43, thus confirming that we had successfully discovered an acceptable indole replacement
with reduced lipophilicity, albeit reduced selectivity against the PI3K isoforms. hERG binding was low
for compounds in Table 7, with only compound 46 above pIC 5, however none of these compounds
5
0
contained the extended basic amine.
Finally, we synthesised the fully elaborated compounds (Table 8) with a pendant morpholine in place
based on the hypothesis that this could deliver the target selectivity. Compound 51, the pyridine
sulfonamide equivalent of compound 29, demonstrated significantly enhanced potency both at the
enzyme and in the PBMC assay, whilst also achieving high selectivity over the PI3K isoforms.
Compound 3, the pyridine sulfonamide analogue of compound 42, was both highly potent in the enzyme
and PBMC assay and highly selective over the PI3K isoforms. This addition of the amine had the same
effect as described previously on PI3K isoform selectivity, such that compound 3 was greater than 500ꢀ
ACS Paragon Plus Environment
28