Discovery of a First-in-Class Receptor Interacting Protein 1 (RIP1) Kinase Specific Clinical Candidate (GSK2982772) for the Treatment of Inflammatory Diseases

Article abstract of DOI:10.1021/acs.jmedchem.6b01751

RIP1 regulates necroptosis and inflammation and may play an important role in contributing to a variety of human pathologies, including immune-mediated inflammatory diseases. Small-molecule inhibitors of RIP1 kinase that are suitable for advancement into

Full text of DOI:10.1021/acs.jmedchem.6b01751

Subscriber access provided by UNIV OF NEBRASKA - LINCOLN
Drug Annotation
Discovery of a First-in-Class Receptor Interacting
Protein 1 (RIP1) Kinase Specific Clinical Candidate
(GSK2982772) for the Treatment of Inflammatory Diseases
Philip A. Harris, Scott B. Berger, Jae U Jeong, Rakesh Nagilla, Deepak Bandyopadhyay, Nino Campobasso,
Carol A. Capriotti, Julie A Cox, Lauren Dare, Xiaoyang Dong, Patrick M Eidam, Joshua N. Finger,
Sandra J Hoffman, James Kang, Viera Kasparcova, Bryan W. King, Ruth Lehr, Yunfeng Lan, Lara K
Leister, John D Lich, Thomas T. MacDonald, Nathan A Miller, Michael T. Ouellette, Christina S. Pao,
Attiq Rahman, Michael A. Reilly, Alan R. Rendina, Elizabeth J. Rivera, Michelle C Schaeffer, Clark A.
Sehon, Robert R. Singhaus, Helen H. Sun, Barbara A. Swift, Rachel D Totoritis, Anna Vossenkamper,
Paris Ward, David D Wisnoski, Daohua Zhang, Robert W Marquis, Peter J. Gough, and John J. Bertin
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01751 • Publication Date (Web): 02 Feb 2017
Downloaded from http://pubs.acs.org on February 2, 2017
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 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Zhang, Daohua; GlaxoSmithKline, Pattern Recognition Receptor DPU
Marquis, Robert; GlaxoSmithKline, Medicinal Chemistry
Gough, Peter; GlaxoSmithKline
Bertin, John; GlaxoSmithKline
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 49
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 FirstꢀinꢀClass Receptor Interacting
Protein 1 (RIP1) Kinase Specific Clinical Candidate
(GSK2982772) for the Treatment of Inflammatory
Diseases
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
*†
†
†
†
‡
Philip A. Harris, Scott B. Berger, Jae U. Jeong, Rakesh Nagilla, Deepak Bandyopadhyay,
‡
†
‡
†
†
Nino Campobasso, Carol A. Capriotti, Julie A. Cox, Lauren Dare, Xiaoyang Dong, Patrick
†
†
†
†
†
M. Eidam, Joshua N. Finger, Sandra J. Hoffman, James Kang, Viera Kasparcova, Bryan W.
‡
‡
†
†
†
#
King, Ruth Lehr, Yunfeng Lan, Lara K. Leister, John D. Lich, Thomas T. MacDonald,
†
‡
‡
†
†
Nathan A. Miller, Michael T. Ouellette, Christina S. Pao, Attiq Rahman, Michael A. Reilly,
‡
†
‡
†
Alan R. Rendina, Elizabeth J. Rivera, Michelle C. Schaeffer, Clark A. Sehon, Robert R.
†
†
†
‡
#
Singhaus, Helen H. Sun, Barbara A. Swift, Rachel D. Totoritis, Anna Vossenkämper, Paris
‡
‡
†
†
†
Ward, David D. Wisnoski, Daohua Zhang, Robert W. Marquis, Peter J. Gough and John
†
Bertin
†
Pattern Recognition Receptor DPU and ‡Platform Technology & Science, GlaxoSmithKline,
Collegeville Road, Collegeville, Pennsylvania 19426, United States. #Centre for
Immunobiology, Blizard Institute, Barts and The London School of Medicine and Dentistry,
Queen Mary University of London, E1 2AD London, U.K.
ACS Paragon Plus Environment

Page 3 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ABSTRACT
RIP1 regulates necroptosis and inflammation and may play an important role in contributing to
a variety of human pathologies, including immuneꢀmediated inflammatory diseases. Smallꢀ
molecule inhibitors of RIP1 kinase that are suitable for advancement into the clinic have yet to
be described. Herein we report our lead optimization of a benzoxazepinone hit from a DNAꢀ
encoded library and the discovery and profile of clinical candidate GSK2982772 (compound 5),
currently in phase 2a clinical studies for psoriasis, rheumatoid arthritis and ulcerative colitis.
Compound 5 potently binds to RIP1 with exquisite kinase specificity and has excellent activity in
blocking many TNFꢀdependent cellular responses. Highlighting its potential as a novel antiꢀ
inflammatory agent, the inhibitor was also able to reduce spontaneous production of cytokines
from human ulcerative colitis explants. The highly favorable physicochemical and ADMET
properties of 5, combined with high potency, lead to a predicted low oral dose in humans.
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
INTRODUCTION
RIP1 has emerged as an important upstream kinase that has been shown to regulate
1
inflammation through both scaffolding and kinase specific functions. As a critical regulator of
inflammation, it is positioned as a key node in the innate immune response, which must be
tightly regulated to maintain tissue homeostasis and can have detrimental consequences if
2
,3
dysregulated. To this end, there is strong genetic evidence in humans and mice that mutations
which shunt signaling down the RIP1 kinase pathway result in spontaneous and robust
4ꢀ6
inflammation. Recent work has shown that active RIP1 kinase can drive this inflammation
through directly regulating programmed necrosis, the production of proꢀinflammatory cytokines,
7
ꢀ10
inflammasome assembly and some forms of pathogenic apoptosis. RIP1 has been shown to be
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 49
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
a critical driver of inflammation of various pathways downstream of the death receptors TNFR1,
1
1ꢀ13
FasL and TRAIL, as well as tollꢀlike receptors.
Hence blocking this pathway has the
potential to result in a broad therapeutic benefit for multiple inflammatory diseases.
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
Degterev et al. were first to identify a series of RIP1 inhibitors using a phenotypic cell screen
that measured their ability to block necrotic death induced by TNF and the caspase inhibitor
1
4,15
zVAD.fmk.
The most advanced series, represented by indoleꢀhydantoin 1 (known as Necꢀ1s
or 7ꢀClꢀOꢀNecꢀ1), was subsequently coꢀcrystallized in the RIP1 kinase domain showing it to
1
6
occupy an allosteric lipophilic pocket at the back of the ATP binding site. This allosteric type
17
III binding mode resulted in excellent kinase selectivity. However, a narrow SAR profile
coupled with moderate potency and poor pharmacokinetic properties for this series has limited
1
8
their subsequent development to date.
As part of our initial drug discovery efforts targeting RIP1 inhibitors we screened the library of
GSK kinase inhibitors. This identified a number of potent inhibitors of RIP1 belonging to the
type II class, targeting the “DFGꢀout” conformation of the kinase, as exemplified by 2 in Figure
19
1
. Further development of these type II inhibitors were hampered by high molecular weights
and lipophilicities, coupled with low aqueous solubilities and a range of offꢀtarget kinase
activities. From a separate highꢀthroughput screen of the GSK compound collection we also
identified 1ꢀpivaloylꢀ5ꢀphenyl 4,5ꢀdihydropyrazole (3), known as GSK’963, a small, highly
potent and selective RIP1 inhibitor. Although an excellent tool compound to study RIP1
20
inhibition in vitro, 3 had minimal oral exposure in rodents limiting further development.
2
1
From a third approach, screening of GSK’s collection of DNAꢀencoded libraries identified a
potent and selective RIP1 inhibitor exemplified by benzoxazepinone 4, known as GSK’481
22
(
Figure 1). This series differentiated itself over previously disclosed RIP1 inhibitors in that it
ACS Paragon Plus Environment

Page 5 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
combined high RIP1 potency and kinase selectivity with potential to achieve oral exposure in
rodents. Herein we describe the lead ꢀoptimization of this series that led to the identification of
(S)-5ꢀbenzylꢀN-(5ꢀmethylꢀ4ꢀoxoꢀ2,3,4,5ꢀtetrahydrobenzo[b][1,4]oxazepinꢀ3ꢀyl)ꢀ1Hꢀ1,2,4ꢀ
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
triazoleꢀ3ꢀcarboxamide 5 (GSK2982772). The high RIP1 potency, monoꢀkinase selectivity and
excellent preclinical pharmacokinetic and developability profile of benzoxazepinone 5 led to its
selection for development and it became the firstꢀinꢀclass RIP1 inhibitor to enter clinical trials in
2015 and recently completed phase 1 evaluation.
Figure 1. Structure of RIP1kinase inhibitors.
SAR AND LEAD OPTIMIZATION
The benzoxazepinone 4 hit identified from a DNAꢀencoded library screen possessed excellent
potency in both a RIP1 (1ꢀ375) fluorescence polarization (FP) binding assay and an ADPꢀGlo
22
functional biochemical assay (See Table 1). Potency of 4 was maintained in a human
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 49
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
monocytic U937 cellular assay, measuring the ability to block necrotic cell death induced by
treatment with TNF and the caspase inhibitor QVDꢀOPh. Remarkably for an unoptimized
screening hit, benzoxazepinone 4 showed complete specificity for RIP1 kinase over all other
kinases tested at 10 ꢀM concentration when profiled against 318 kinases using a P33
radiolabeled assay screen at Reaction Biology Corp and 456 kinases using a competition binding
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
assay KINOMEscan at DiscoveRx Corp. This high RIP1 enzymatic and cellular potency,
coupled with complete kinase specificity made this series an excellent choice for further
optimization into a RIP1 clinical candidate. This benzoxazepinone pharmacophore is to our
knowledge a novel kinase inhibitor template. Our initial goals were to explore the SAR in RIP1
and identify a developable candidate that maintained the favorable in vitro profile, whilst
optimizing three key parameters: lipophilicity, solubility and oral exposure in preꢀclinical
species. As a benchmark, benzoxazepinone 4 possessed high lipophilicity (log D = 5.9), low
FaSSIF crystalline solubility (30 ꢀg/mL) and a suboptimal pharmacokinetic profile. In rat 4
exhibited a low oral exposure (AUC0ꢀ∞ = 0.38 ꢁg.h/mL at 2 mg/kg dose) coupled with a high
clearance (69 mL/min/kg) and high volume of distribution (8.5 L/kg).
In the absence of a coꢀcrystal structure, we relied on an homology model of the RIP1 to predict
2
2
how this series was binding in the ATP binding pocket. This model had the lactam carbonyl of
the benzoxazepinone and the exocyclic NH making hydrogenꢀbond interactions with the kinase
hinge at Met95 and Glu93, respectively. Our early SAR investigations began around this putative
hinge binding region. Removal of the lactam N-methyl group gave analog 6 with about a log
drop in enzyme and cell potencies as shown in Table 1, but very encouragingly a substantial
22
increase in rat oral exposure (AUC0ꢀ∞ 2.2 ꢁg.h/mL at 2 mg/kg dose) accompanied this change.
However, increasing the size of the N-substituent to the corresponding ethyl or cyclopropyl
ACS Paragon Plus Environment

Page 7 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
analogs 7 and 8 led to a rapid loss in potency. The S chiral center was established as important
for binding, as the R enantiomer 9 and αꢀmethylation of the chiral carbon leading to 10 were
both inactive. Methylation of the exocyclic NH (benzoxazepinone 11) or removal of the
benzoxazepinone carbonyl to give benzoxazepine 12 both led to complete loss of activity. At the
conclusion of this lead optimization a coꢀcrystal structure of 5 was obtained in RIP1 as discussed
later in this article. Interestingly, this revealed the series does not make any interaction with the
kinase hinge, rather the inhibitor sits deeper in the ATP binding pocket, the benzyl group
occupying an allosteric lipophilic pocket at the back. The benzoxazepinone moiety resides in a
tight pocket formed by two βꢀstrands defined by Leu90ꢀVal91ꢀMet92 and Ile43ꢀMet44ꢀLys45
(see Figure 2). Changes that increase size at either the lactam nitrogen (e.g. 7, 8) or chiral center
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
(e.g. 10, 11), or alter the conformation of the 7ꢀmembered ring (e.g. 9, 12), are not tolerated.
Figure 2. Crystal structure of compound 5 (yellow) in the RIP1 kinase domain (green) with the
benzoxazepinone packed against 2 βꢀstrands defined by Leu90ꢀVal91ꢀMet92 and Ile43ꢀMet44ꢀ
Lys45.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 49
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
a
Table 1. SAR around the lactam amide and chiral center
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
c
Cpd
R
R1 R2 RIP1 FP ADPꢀGlo
U937
IC (nM) IC (nM) IC (nM)
50
50
50
4
6
7
8
9
Me
H
–
–
–
–
–
–
–
10
1.6
10
32
16
–
200
Et
CyPr
H
–
3,162
7,943
d
–
>10,000
>10,000
–
>10,000
>10,000
>10,000
>10,000
>10,000
–
–
1
1
1
0
1
2
–
H
Me
–
Me >10,000
–
–
H
–
>10,000
>10,000
–
–
–
a
Assay protocols are described in Supporting Information; IC values are the average of at
least two determinations. Lower limit of sensitivity is ca. 10 nM. Conventional data analysis
was used for less potent inhibitors (IC > 10 nM); whereas tight binding analyses was used for
5
0
b
c
5
0
d
more potent inhibitors (IC < 10 nM). Cyclopropyl
50
A survey of various replacements for the heteroatom of the benzoxazepinone was carried out
involving synthesis of the benzothiazepinone 13, benzazepinone 14, and benzodiazepinones 15
ACS Paragon Plus Environment

Page 9 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
and 16 analogs (see Table 2). This heterocycle switch maintained the high RIP1 in vitro
potency, presumably because it maintained the required conformation of the sevenꢀmembered
ring. Indeed benzothiazepinone 13 was extremely potent in the ADPꢀGlo assay (IC = 63 pM),
5
0
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
but had no rat oral exposure. The benzodiazepinone 15 did moderately improve lipophilicity,
FaSSIF solubility (97 ꢀg/mL) and rat oral exposure compared to benzoxazepinone 4.
a
Table 2. Replacements of benzoxazepinone heteroatom
b
c
d
Cpd
4
A
ADPꢀGlo
U937
50
AUC0ꢀ∞
log D
IC (nM) IC (nM) ꢁg.h/mL
5
0
O
S
1.6
10
0.38
5.9
6.2
6.0
5.0
6.0
d
13
14
15
16
0.063
0.20
0.79
5.0
2
BLD
0.007
0.80
CH2
NH
NMe
6.3
40
20
0.082
a
Assay protocols are described in Supporting Information; IC values are the average of at
50
b
least two determinations. ADPꢀGlo tight binding analyses was used as the lower limit of
sensitivity was reached in the RIP1 FP assay. Rat oral exposure at 2 mg/kg. Below level of
detection. CHI (chromatographic hydrophobicity index) log D at pH 7.4 was calculated from the
retention time (t ) observed in a fast gradient reverseꢀphase HPLC.
c
d
d
23
R
More substantial modifications of the benzoxazepinone ring were also explored. Expanding the
heterocycle to an eightꢀmembered benzo[1,4]oxazocinone 17, or contracting to a quinolinone 18,
both lost potency. Similarly removal of the benzo function to yield a 1,4ꢀoxazepanꢀ5ꢀone 19 or
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 49
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
its saturation to a cyclohexyl ring 20 resulted in inactive analogs (see Table 3). These findings fit
with the coꢀcrystal structure of 5 in RIP1 in which the benzoxazepinone moiety occupies a tight
pocket, see Figure 2, and is therefore not amenable to modifications that significantly alter the
oxazepinone conformation.
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
Table 3. Modifications of benzoxazepinone heterocycle
b
Cpd
RIP1 FP
U937
AUC0ꢀ∞
IC (nM) IC (nM)
ꢁg.h/mL
5
0
50
6
32
200
2.2
0.38
–
17
18
19
20
200
631
>10,000
>10,000
–
>10,000
>10,000
>10,000
–
a
Assay protocols are described in Supporting Information; IC values are the average of at
5
0
b
least two determinations. Rat oral exposure at 2 mg/kg.
Substitution at the aryl ring of the benzoxazepinone ring was also investigated revealing a
good tolerance for substitution at the 7ꢀ and 8ꢀpositions, which in the coꢀcrystal structure are
orientated towards solvent exposed space. The 6 and 9ꢀpositions had less room for substitutions,
with only fluorine well tolerated. As shown for selected representative examples in Table 4, N-
ACS Paragon Plus Environment

Page 11 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
acetyl (21), N-methyl formate (22) or N-methyl sulfonyl (23) 7ꢀsubstitutions, or attachment of
heterocycles such as pyrazole (24) or 1,3,4ꢀoxadiazolꢀ2ꢀone (25), all maintained comparable
potencies. In particular, the 1,3,4ꢀoxadiazolꢀ2ꢀone (25) possessed good rat total oral exposure
and improved lipophilicity (log D = 4.4), but suffered from a low FaSSIF solubility (8 ꢀg/mL).
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
Table 4. 7ꢀSubstitutions at the benzoxazepinone
b
c
Cpd
Substitution
ADPꢀGlo
IC (nM) IC (nM) ꢁg.h/mL
U937
50
AUC0ꢀ∞
5
0
4
H
1.6
10
0.38
0.45
1.1
ꢀ
21
22
23
24
25
NHCOMe
0.32
0.32
0.63
1.0
7.9
10
NHCO
NHSO
Me
Me
2
13
2
3ꢀpyrazole
6.3
3.2
0.60
2.1
5ꢀ(1,3,4ꢀoxadiazolꢀ2ꢀone)
0.5
a
Assay protocols are described in Supporting Information; IC values are the average of at
least two determinations. ADPꢀGlo tight binding analyses was used as the lower limit of
sensitivity was reached in the RIP1 FP assay. Rat oral exposure at 2 mg/kg
50
b
c
The SAR around the benzyl group, which resides in an allosteric hydrophobic pocket at the
back of the ATP binding site (see Figure 4), was quite narrow and clearly favored lipophilic
groups, as shown in Table 5. Removal of the phenyl group (analog 26) resulted in loss of
activity, and suboptimal activity resulted from replacement of phenyl with nꢀbutyl (27),
isopropyl (28) or cyclohexyl (29) groups. Any substitutions at the phenyl ring other than fluorine
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 49
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
were detrimental to potency, and introduction of polarity such as the N-piperidine functionality
(analog 30) were not tolerated.
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
Table 5. Modifications of benzyl group
Cpd
R
RIP1 FP
U937
IC (nM)
IC (nM)
5
0
50
4
Ph
H
10
10
–
2
2
2
6
7
8
12,600
100
nBu
126
iPr
1,600
251
6310
501
29
30
cyclohexyl
N-piperidine
>10,000
>10,000
a
Assay protocols are described in Supporting Information; IC50 values are the average of at
least two determinations.
Ultimately it was replacement of the isoxazole heterocycle that had the greatest overall impact
on improving the developability profile of this series. The choice of heterocycle had a significant
effect on potency, as shown for selected examples in Table 6. A general trend, although not
absolute, was for heterocycles with a “nonꢀbridging” nitrogen ortho to the amide carbonyl to
possess optimal potency. For example, oxazole 31 (IC = 50 pM) is much more potent than its
5
0
isomer 33 (IC = 794 nM), and the imidazole 35 (IC = 0.4 nM) is much more active than its
5
0
50
isomer 34 (IC = 1.3 ꢀM) or pyrazole analog 36 (IC = 63 nM). This nitrogen adjacent to the
50
50
ACS Paragon Plus Environment

Page 13 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
carbonyl may be imparting a favorable trans orientation due to electron lone pair repulsion
separating the negatively charged nitrogen and amide oxygen atoms thus positioning the benzyl
2
4
group optimally into the tight back pocket. Thiazole 32, with low activity, appears to be an
anomaly to this trend, but likely the large size of the sulfur atom is altering orientation of the
benzyl group resulting in an suboptimal fit into the back pocket. The N-benzylꢀ1,2,3ꢀtriazole 40
displayed encouragingly improved rat oral exposure compared to its N-benzylꢀ1,2ꢀ4ꢀtriazole
isomer 41, but the breakthrough came with preparation of the 3ꢀbenzylꢀ1,2ꢀ4ꢀtriazole isomer 5,
which had the optimal combination of in vitro potency, lipophilicity (log D 3.8) and rat oral
exposure (AUC0ꢀ∞ 2.3 ꢁg.h/mL at 2 mg/kg dose). As can be seen from Figure 3 there is a clear
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
2
3
correlation between lower lipophilicity (log D < 4.3) and improved kinetic aqueous solubilities
2
5
for the compounds profiled in this paper. A recent study of heterocycle containing molecules in
drug developability assays found that 1,2,4ꢀtriazoles possessed optimal profiles including
2
6
solubility. Presumably the lower lipophilicity and increased solubility of 5 resides in an optimal
charge distribution of the 1,2,4ꢀtrizaole leading to larger dipole moment (calculated dipole
27
moment for the unsubstituted 4Hꢀ1,2,4ꢀtriazole is 5.81 Debye). Adding an additional nitrogen
resulted in the tetrazole analog 42, which possessed excellent potency but suboptimal
lipophilicity and oral exposure. Replacing the heterocycle with a 1,3ꢀphenyl analog 43 resulted
in loss of considerable potency, although this could be recovered by introduction of an ortho
nitrogen as shown in pyridine 44. However both analogs are quite lipophilic, with
correspondingly poor solubilities. The lead optimization culminated in selection of 5 to progress
into development based on the inhibitor maintaining the excellent potency and kinase selectivity
of the lead 4, combined with a significantly improved developability profile. This was
demonstrated by a two log improvement in log D and 7ꢀfold improvements in oral exposure in
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 49
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
rat of 5 compared to 4 (see Table 6). We identified a stable nonꢀsolvated crystalline form of 5
with a low melting point (131 °C) and improved FaSSIF solubility (230 ꢀg/mL). As discussed
later in the section on preclinical characterization of 5, this resulted in an excellent dose
proportionality profile in both pharmokinetic and safety studies across a large dose range (2 to
1000 mg/kg).
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
Table 6. Heterocycle replacements
b
c
d
Cpd
A
FP/ADPꢀGlo
IC (nM)
U937
log D
AUC0ꢀ∞
IC (nM)
ꢁg.h/mL
5
0
50
4
1.6
10
1.0
5.9
5.6
6.5
5.4
0.38
3
3
3
1
2
3
0.05
631
794
0.055
794
–
–
2,512
34
1,259
0.4
1,995
6.3
3.8
4.2
4.0
5.4
4.9
–
3
5
6
0.66
2.9
3
63
631
10
37
0.4
0.17
0.13
3
8
0.79
7.9
ACS Paragon Plus Environment

Page 15 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3
4
4
9
0
1
0.5
1.0
5.0
25
5.4
4.6
4.2
3.8
5.0
6.3
0.033
1.3
0.63
1.0
1.6
6.3
2.5
398
0.28
2.3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
42
43
44
0.079
251
0.24
–
2
40
6.4
–
a
Assay protocols are described in Supporting Information; IC5 values are the average of at
0
b
least two determinations. RIP1 FP binding assay was used to assess potency of inhibitors with
IC50 > 10 nM; whereas for more potent inhibitors where the limit of sensitivity FP assay was
reached (ca 10 nM) the ADPꢀGlo tight binding analyses was employed to more accurately
determine potency. A good correlation was observed between both assay formats. CHI
(chromatographic hydrophobicity index) log D at pH 7.4 was calculated from the retention time
c
23 d
(tR) observed in a fast gradient reverseꢀphase HPLC. Rat oral exposure at 2 mg/kg.
Figure 3. Comparison of chromatographic hydrophobicity index (CHI) log D and aqueous
solubility for inhibitors 4-45. Reducing lipophilicity, specifically with a log D < 4.3 (indicated by
red line), leads to higher aqueous solubilities. Benzoxazepinone 5 is colored red. CHI log D at
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 49
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
pH 7.4 was calculated from the retention time (t ) observed in a fast gradient reverseꢀphase
R
2
3
HPLC. Kinetic aqueous solubility was measured by chemiluminescence nitrogen detection
(CLND) from phosphate buffer saline pH 7.4 solution, prepared from DMSO stock solution of
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
25
the inhibitor.
BINDING MODE AND KINASE SELECTIVITY
Subsequent to the identification of 5 as a development candidate we were able to coꢀcrystallize
this template in RIP1. This was achieved by reproducing the published coꢀcrystal structure of
Necrostatinꢀ4
((S)-N-(1ꢀ(2ꢀchloroꢀ6ꢀfluorophenyl)ethyl)ꢀ5ꢀcyanoꢀ1ꢀmethylꢀ1Hꢀpyrroleꢀ2ꢀ
carboxamide) bound to the RIP1 construct (1–294, C34A, C127A, C233A, and C240A) and
1
6
displacing Necrostatinꢀ4 with benzoxazepinone 5. The Xꢀray structure of the RIP1 kinase
domain was refined to 2.6 Å resolution (see Figure 4a). In this structure, benzoxazepinone 5 was
observed to be buried deep in the pocket between the N-terminal and Cꢀterminal domains.
Interestingly the molecule resides further in the binding pocket than we initially modeled making
no interaction with the hinge residues. The triazole and benzyl functionality of 5 occupies the
same allosteric lipophilic pocket at the back of the ATP binding site as Necrostatinꢀ4 (Figure 4b),
2
8
suggestive more of a type III class of kinase inhibitor. Although inhibitor 5 does not occupy the
same space as the adenine ring of ATP, the benzoxazepinone ring does occupy space where the
alpha phosphate would reside so the inhibition is not strictly allosteric. The amide carbonyl
attached to the triazole makes a direct hydrogenꢀbond interaction with the backbone amide NH
of Asp156. The triazole nitrogen makes a waterꢀmediated hydrogenꢀbond to the carbonyl
oxygens of Met67 and Val76. The Cꢀhelix is shifted compared to the bound 1 structure and the
activation loop is more ordered when 5 is bound. Structure determination was done by molecular
ACS Paragon Plus Environment

Page 17 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
replacement using the atomic coordinates of RIP1 kinase domain with Necrostatinꢀ4 (PDB code
4
ITJ) as a search model (see Supplemental Information for details).
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. (a) Coꢀcrystal structure of RIP1 (1–294, C34A, C127A, C233A, and C240A) and
benzoxazepinone 5. (b) Overlaid crystal structures of compound 5 (yellow) in the RIP1 kinase
domain (green) with Necrostatin 4 (cyan) in the RIP1 kinase domain (cyan)
The type III binding mode observed for this pharmacophore in RIP1 results in complete kinase
selectivity as exemplified by the kinase profile of 5 against both a P33 radiolabeled assay screen
at Reaction Biology Corp (339 kinases) and a competition binding assay KINOMEscan at
DiscoveRx Corp (456 kinases). In both assays benzoxazepinone 5 was tested at a concentration
of 10 ꢀM, which represents an estimated >10,000 fold selectivity window based on the RIP1
ADPꢀGlo potency of 1 nM. The Reaction Biology Corp kinase dendrogram is shown in Figure 5.
Details on both profiles are available in the Supporting Information section. This monoꢀkinase
selectivity profile was not observed with any of the type II DFGꢀout RIP1 inhibitors we
1
9
profiled.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 49
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. Kinase selectivity profile of compound 5 as shown by a Reaction Biology Core kinase
panel screen against 339 kinases assayed at 10 ꢀM in duplicate. Compound selectivity is
represented in a dendrogram view of the human kinome phylogenetic tree. All kinases tested
were inactive (< 50%) as indicated by green circles. RIP1 was not part of this panel.
PRECLINICAL CHARACTERIZATION
As would be predicted by the coꢀcrystal structure, benzoxazepinone 5 is an ATP competitive
inhibitor (see Supporting Information). The binding kinetics are similar to benzoxazepinones
ꢀ1
ꢀ1
recently profiled, with a moderate onꢀrate constant (k = 0.045 ꢀM sec ), accompanied by a
on
ꢀ1
slow offꢀrate constant koff = 0.37 hour (t = 112 min), measured by stoppedꢀflow kinetics and
1/2
FP competitive binding, respectively (see Supporting Information). In addition to efficacy
against the immortalized U937 human monocyte cell line, activity of 5 was also assessed in
primary neutrophils isolated from human whole blood. In this assay TNF is coꢀincubated with
both the caspase inhibitor QVDꢀOPh and the SMAC mimetic RMT 5265, which block the
ACS Paragon Plus Environment

Page 19 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
apoptosis and NFꢀκB pathways, respectively, driving the TNF response down the necrosis
2
9
pathway. Inhibitor 5 was able to potently block this response as shown by determination of
overall cell viability as measured by cellular ATP levels (IC = 1.6 nM), cell death as measured
5
0
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
by LDH release (IC = 0.4 nM) and RIP1ꢀdependent inflammatory cytokine MIPꢀ1β production,
5
0
either as absolute levels for protein, or fold changes in mRNA expression (IC = 0.5 nM).
50
A human whole blood stimulation assay was developed in which the necroptosis pathway is
activated in a similar fashion through stimulation with TNF coꢀincubated with the caspase
inhibitor QVDꢀOPh or zVAD.fmk, and the SMAC mimetic RMT 5265. In this assay
benzoxazepinone 5 was also shown to be very potent as measured by inhibition of cytokine MIPꢀ
1β ((IC50 = 2 nM). In a similar cynomolgus monkey whole blood stimulation assay,
benzoxazepinone 5 was equally efficacious (IC = 4 nM).
5
0
In addition to examining potency in stimulated cellular systems, benzoxazepinone 5 was also
able to reduce spontaneous production of cytokines (ILꢀ1β and ILꢀ6) in a concentrationꢀ
dependent fashion from ulcerative colitis explant tissue in overnight incubations (Figure 6). As
previously described, intestinal mucosal tissue obtained from inflamed areas of the gut
spontaneously release significantly elevated levels of inflammatory cytokines compared to
30
healthy mucosa, recapitulating the inflammatory phenotype found in vivo. Although this
human biopsy assay has a high degree of inter subject variability and response rate, the responses
are similar to the steroid positive control prednisolone. The corresponding R enantiomer 45,
inactive against RIP1, was included as a negative control. Although the variability in the data
makes accurate calculation of potency difficult, the data are consistent with the potency of 5 in
other cellꢀbased assays.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 49
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. Inhibition of overnight ILꢀ1β and ILꢀ6 production in human explant samples taken
from ulcerative colitis patients for benzoxazepinone 5 in comparison to prednisolone (predni)
and the corresponding RIP1 inactive R enantiomer 45.
We previously reported this series demonstrates a species selectivity for inhibition of primate
2
2
RIP1 compared to nonꢀprimate RIP1. Consistent with this observation, compound 5 exhibited
approximately equivalent RIP1 FP potency against human and monkey RIP1, but was
significantly less potent against nonꢀprimate RIP1, as shown in Table 7. This species selectivity
was not observed with any of the type II DFGꢀout RIP1 inhibitors we previously profiled (for
1
9
example 2), but appears to be reserved for this type III binding mode. This reduced
biochemical potency also translated to reduced mouse cellular efficacy, as evidenced by a 340
fold reduction in cellular potency for compound 5 in blockage of the necrotic death (induced by
ACS Paragon Plus Environment

Page 21 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
TNF and QVDꢀOPh) in murine fibrosarcoma L929 cells (IC = 1.3 ꢀM), compared to human
5
0
monocytic U937 cells (IC = 0.0063 ꢀM). As detailed in our prior report, selectively mutating
50
key amino acids in the mouse RIP1 activation loop, where nonꢀprimate differed from primate,
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
2
2
increases the potency of benzoxazepinone 4 in mouse cells to that approaching human. This
suggests that the activation loop is the key region in murine RIP1 responsible for the lower
efficacy against this series. We speculate that the differences in amino acid sequence cause
murine RIP1 to have reduced flexibility in adopting the activation loop conformation required
for this inhibitor to bind in its preferred type III conformation.
Table 7. RIP1 species selectivity of compound 5
RIP1 FP
Species
Fold
IC (ꢀM)
5
0
a
Human
Monkey
Rabbit
Rat
0.016
0.020
0.79
2.0
1
a
1.25
49
125
156
313
>625
Mouse
Dog
2.5
5.0
Minipig
>10
a
Values at the lower limit of sensitivity (ca. 10 nM).
The ability to explore an in vitro to in vivo correlation has been limited to acute models, as the
reduction in rodent RIP1 potency meant assessment in more longꢀterm disease models has not
been possible. We had sufficient potency and exposure to evaluate benzoxazepinone 5 in an
acute in vivo mouse model evaluating protection from TNF induced lethal shock, which has been
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 49
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
31
shown to be RIP1 kinase dependent. In this model, injection of TNF leads to a systemic
inflammatory response, characterized by hypotension, hepatitis, hypothermia and bowel necrosis
over 6ꢀ7 hours. In a similar model, injection of TNF combined with the caspase inhibitor zVAD
leads to a similar but earlier onset systemic inflammatory response in about 3 hours. Efficacy can
be measured by the ability of the RIP1 inhibitor to prevent body temperature loss.
Benzodiazepinone 5 was dosed orally 15 min prior to TNF and showed 68, 80 and 87%
protection from temperature loss over 6 hours, at doses of 3, 10 and 50 mg/kg, respectively
(Figure 7). In the corresponding TNF/zVAD model, benzodiazepinone 5 showed 13, 63 and 93%
protection from temperature loss over 3 hours. Assuming the efficacy is Cmax driven, the
correlation of change in body temperature, combined with the estimated blood drug level at the
time of the challenge, leads to an estimated IC50 of 1 and 3.2 ꢀM for both models (Figure 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
This aligns closely with the in vitro assessment of RIP1 mouse L929 cellular activity of 5 (IC =
50
1.3 ꢀM).
ACS Paragon Plus Environment

Page 23 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 7. Evaluation of benzodiazepinone 5 in the TNF (top) and TNF/zVAD (bottom) induced
lethal shock mouse models, measuring reduction in body temperature loss over time.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 8. Dose response curves for benzodiazepinone 5 in the TNF (top) and TNF/zVAD
a
(bottom) mouse induced lethal shock models.
a
Terminal blood concentrations were quantified in the study animals. For each individual
animal, a theoretical Cmax was calculated using Cmax and concentration at 3 h from satellite
animals. These estimated Cmax values were then used to build the dose response model.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 49
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
The assumption that the inhibition in the TNF/zVAD model can be approximated to a direct
effect was assessed by orally dosing 5 at 50 mg/kg and then altering the time of the TNF/zVAD
challenge over a 16 hour period. The measured levels of inhibition of temperature loss were then
compared to the expected levels of inhibition predicted from the oral pharmacokinetic profile
and the standard TNF/zVAD model. As shown in Figure 9, the levels of observed inhibition
correlated reasonably well with the levels of predicted inhibition, and hence the assumption that
the model can be approximated to a direct effect is likely valid within the variability of the 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
Figure 9. Comparison of predicted and actual inhibition of temperature drop following a 50
mg/kg oral dose of 5 in the mouse at various times of TNF/zVAD challenge over 16 hours.
Benzoxazepinone 5 displayed a good free fraction in blood in rat (4.2%), dog (11%),
cynomolgus monkey (11%) and human (7.4%). The inhibitor had a good pharmacokinetic profile
across both rat and monkey (see Table 8). In vitro comparisons of the metabolic stability of 5 in
microsomes and hepatocytes showed it was stable across all species measured including human
(see Supporting Information). All the Phase I and Phase II biotransformations observed in human
ACS Paragon Plus Environment

Page 25 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
were also measured in rat and monkey hepatocytes, as shown in Figure 10. The tissue
distribution of 5 in rats was evaluated following iv infusion over 4 hours before a range of
terminal tissue samples were taken and parent drug concentrations determined and compared to
blood levels (see Supporting Information). This study showed that 5 distributed into a range of
tissues including colon, liver, kidney and heart at concentrations comparable to blood. However,
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
ꢀ6
5
had low brain penetration in rat (4%) despite possessing good cell permeability (21 x 10
cm/s), which is likely due to active extrusion of 5 from brain via the efflux drug transporter Pꢀ
glycoprotein (Pgp) (see Supporting Information).
Table 8. Pharmacokinetic Parameters of 5 after Intravenous or Oral Administration to Rats and
Cynomolgus Monkeys
Route
iv
Parameter
Rat
Monkey
dose (mg/kg)
Cl (mL/min/kg)
Vdss (L/kg)
1.1
1.2
17 ± 5
10 ± 3
2.7 ± 0.6
3.9 ± 0.7
2.1
2.2 ± 0.9
6.5 ± 1.0
1.9
t
1/2 (h)
po
dose (mg/kg)
T
max (h)
0.25 ± 0.0
810 ± 160
2.3 ± 0.8
110 ± 20
1.8 ± 1.1
720 ± 240
2.9 ± 0.8
86 ± 6
C
max (ꢁg/mL)
AUC (ꢁg·h/mL)
bioavailability (%)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 49
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 10. Summary of metabolites of 5 following incubations with cryopreserved hepatocytes
from rats (r), cynomolgus monkeys (m) and humans (h). Compound 5 (10 ꢁM) was incubated in
a hepatocyte suspensions (1 million cells/mL) at 37°C for up to 4 hours under 5% CO :95% O .
2
2
Allometric scaling and in vitro to in vivo extrapolations (using GSK proprietary software) with
and without effects of free fraction were used to generate predictions of the human
pharmacokinetic parameters of 5 (Table 9). These showed a high level of correlation and
predicted 5 to have high bioavailability, moderate to low clearance with a moderate volume and
a terminal half life in the order of 12 hours. The average clearance and volume values were then
used to reconstruct a monoꢀexponential predicted human blood concentration time profile as
shown in Figure 11.
ACS Paragon Plus Environment

Page 27 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 9. Average predicted human PK profile of benzoxazepinone 5 based on allometry and in
vitro to in vivo extrapolation.
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
Parameter
Average
6.3
95%
7.4
ꢀ95%
5.2
Clearance
mL/min/kg)
(
Volume (L/kg)
Half life (h)
2.5
11.5
79
2.7
13.5
86
2.2
9.5
72
Bioavailability (%)
Figure 11. Predicted human blood concentration time profile of benzoxazepinone 5 overlaid
with the human whole blood inhibition IC and IC concentrations
9
0
50
Human PK/PD was modeled using the predicted human PK profile along with a series of in
vitro activity assays. Calculations were made with and without taking into account differences in
free fraction across species, although the lack of apparent shift between binding, cellular and
whole blood numbers makes the true impact of free fraction difficult to interpret. Doses were
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 49
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
then scaled and modeled to achieve 50 and 90% inhibition over 24 hours, as illustrated in Figure
12. Without a clear model to link the required level of inhibition to a validated disease model,
our initial clinical target was to maintain a 90% effect level over 24 hours. Using inhibition of
human whole blood, 5 is predicted to require a 60 mg dose once a day to achieve a greater than
90% pharmodynamic effect over 24 hours with an AUC of 1.7 ꢁg·h/mL.
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 12. Predicted human PD dose effect levels of benzoxaepinone 5 at 6.3 and 60 mg once
daily, achieving 50 and 90% RIP1 inhibition levels over 24 hours, respectively.
Compound 5 demonstrated a clean profile against a range of nonꢀkinase targets. In particular
no inhibitory potential was observed towards recombinant human cytochrome P450 (CYP)
enzymes, with the exception of weak activity against CYP2C9 (IC50 = 25 ꢀM). Compound 5
produced a weak concentrationꢀdependent inhibition of hERG in human embryonic kidney
(
HEKꢀ293) cells, with an estimated IC of 195 ꢀM, and also showed a weak activation of the
5
0
human Pregnane X receptor (hPXR) with an EC of 13 ꢀM.
5
0
ACS Paragon Plus Environment

Page 29 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The nonꢀsolvated crystalline form of 5 has a predicted pKa of 7.7 for the triazole NH and an
aqueous solubility of approximately 0.1 mg/mL across the physiological pH range. The
optimized crystalline solubility of 5 led to excellent dose proportionality comparing solution and
suspension oral dosing ranging from 2 to 1000 mg/kg in both rat and cynomolgus monkey
(Figure 13). This dose proportionality allowed us to evaluate compound 5 across a wide dose
range for preꢀclinical toxicity studies. Due to the significantly lower potency for this series
against nonꢀprimate RIP1, cynomolgus monkey was selected as the second species for preꢀ
clinical safety evaluation. One month safety assessment in rat and monkey provided ample
safety windows to facilitate progression of 5 into phase 1 clinical trials.
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 13. Dose proportionality for 5 in rat and cynomolgus monkey following oral dosing in
pharmacokinetic and toxicology studies, as measured by dose normalized AUC (DNAUC).
CHEMISTRY
The general route to prepare this benzoxazepinone template has been described previously
2
2
starting from BOCꢀLꢀserine and is detailed in Scheme 1. The preparation of the corresponding
ACS Paragon Plus Environment