Discovery of 2-{3-[2-(1-isopropyl-3-methyl-1H-1,2-4-triazol-5-yl)-5,6- dihydrobenzo[f]imidazo[1,2- d ][1,4]oxazepin-9-yl]-1H-pyrazol-1-yl}-2- methylpropanamide (GDC-0032): A β-sparing phosphoinositide 3-kinase inhibitor with high unbound exposure and robust in vivo antitumor activity

Article abstract of DOI:10.1021/jm4003632

Dysfunctional signaling through the phosphoinositide 3-kinase (PI3K)/AKT/mTOR pathway leads to uncontrolled tumor proliferation. In the course of the discovery of novel benzoxepin PI3K inhibitors, we observed a strong dependency of in vivo antitumor activity on the free-drug exposure. By lowering the intrinsic clearance, we derived a set of imidazobenzoxazepin compounds that showed improved unbound drug exposure and effectively suppressed growth of tumors in a mouse xenograft model at low drug dose levels. One of these compounds, GDC-0032 (11l), was progressed to clinical trials and is currently under phase I evaluation as a potential treatment for human malignancies.

Full text of DOI:10.1021/jm4003632

Subscriber access provided by UNIV LAVAL
Article
Discovery of 2-{3-[2-(1-isopropyl-3-methyl-1H-1,2-4-triazol-5-yl)-5,6-
dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepin-9-yl]-1H-pyrazol-1-yl}-2-
methylpropanamide (GDC-0032): A #-sparing phosphoinositide 3-kinase (PI3K)
inhibitor with high unbound exposure and robust in vivo anti-tumor activity
Chudi Ndubaku, Timothy P Heffron, Steven Thomas Staben, Matthew Baumgardner, Nicole
Blaquiere, Erin Bradley, Richard Bull, Steven Do, Jennafer Dotson, Danette Dudley, Kyle Edgar,
Lori Friedman, Richard Goldsmith, Robert Andrew Heald, Aleks Kolesnikov, Leslie Lee, Cristina
Lewis, Michelle Nannini, Jim Nonomiya, Jodie Pang, Steve Price, Wei Wei Prior, Laurent Salphati,
Steve Sideris, Jeffrey Wallin, Lan Wang, Binqing Wei, Deepak Sampath, and Alan G Olivero
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/jm4003632 • Publication Date (Web): 10 May 2013
Downloaded from http://pubs.acs.org on May 14, 2013
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 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of 2ꢀ{3ꢀ[2ꢀ(1ꢀisopropylꢀ3ꢀmethylꢀ1Hꢀ1,2ꢀ4ꢀ
triazolꢀ5ꢀyl)ꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀ
d][1,4]oxazepinꢀ9ꢀyl]ꢀ1Hꢀpyrazolꢀ1ꢀyl}ꢀ2ꢀ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
methylpropanamide (GDCꢀ0032): A βꢀsparing
phosphoinositide 3ꢀkinase (PI3K) inhibitor with high
unbound exposure and robust in vivo antiꢀtumor activity
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
Chudi O. Ndubaku,*,‡ Timothy P. Heffron,*,‡ Steven T. Staben,‡ Matthew Baumgardner,† Nicole Blaquiere,‡
Erin Bradley,‡ Richard Bull,∫ Steven Do,‡ Jennafer Dotson,‡ Danette Dudley,‡ Kyle A. Edgar, ¶ Lori S.
Friedman,¶ Richard Goldsmith,‡ Robert A. Heald,∫ Aleksandr Kolesnikov,‡ Leslie Lee, ¶ Cristina Lewis,§
Michelle Nannini, ¶ Jim Nonomiya,§ Jodie Pang,† Steve Price,∫ Wei Wei Prior,¶, Laurent Salphati,† Steve
°
Sideris, § Jeffery J. Wallin, ¶ Lan Wang,‡ BinQing Wei,‡ Deepak Sampath,¶ Alan G. Olivero‡
Departments of ‡Discovery Chemistry, †Drug Metabolism and Pharmacokinetics, ¶Translational Oncology,
and §Biochemical Pharmacology, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA;
∫Argenta Discovery, 8ꢀ9 Spire Green Centre, Flex Meadow, Harlow, Essex, CM19 5TR, UK
RECEIVED DATE (to be automatically inserted after your manuscript is accepted if required
according to the journal that you are submitting your paper to)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 45
1
2
3
Abstract
4
5
6
7
8
Dysfunctional signaling through the PI3K/AKT/mTOR pathway leads to uncontrolled tumor proliferation.
In the course of the discovery of novel benzoxepin PI3K inhibitors, we observed a strong dependency of in
vivo antiꢀtumor activity on the free drug exposure. By lowering the intrinsic clearance, we derived a set of
imidazobenzoxazepin compounds that showed improved unbound drug exposure and effectively suppressed
growth of tumors in a mouse xenograft model at low drug dose levels. One of these compounds, GDCꢀ0032
(11l), was progressed to clinical trials and is currently under Phase I evaluation as a potential treatment for
human malignancies.
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
Introduction
The phosphoinositide 3ꢀkinase (PI3K) enzymes are lipid kinases responsible for the conversion of
membraneꢀbound 4,5ꢀphosphatidylinositolbisphosphate (PIP2) into 3,4,5ꢀphosphatidylinositoltrisphosphate
(PIP3).1,2 This signaling event occurs in response to various extracellular triggers including activation of
receptor tyrosine kinases (RTKs) and Gꢀprotein coupled receptors (GPCRs). Phosphorylation to PIP3 in turn
promotes downstream activation of Akt and mTOR, which influence several functions important to cellular
integrity including proliferation, evasion of apoptosis, metabolism and angiogenesis. Not surprisingly,
perturbations of this crucial pathway are highly associated with cancer. While there are four class I PI3K
isoforms (α, β, δ and γ) implicated in the signaling cascade, the aberration most common to cancer is the
amplification or activating mutation of PI3Kα.3ꢀ5 Moreover, inactivation or genetic deletion of the negative
regulator of this pathway, phosphatase and tensin homolog (PTEN), is also common.6 As a result, small
molecule inhibitors that target the PI3Ks have been intensely pursued as potential treatments for various
types of cancers and some of these have recently advanced into the clinic.7,8
ACS Paragon Plus Environment

Page 3 of 45
Journal of Medicinal Chemistry
1
2
3
Table 1. Comparison of clinical compound GDCꢀ0941 (1) and thiazolobenzoxepin 2
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
aIC50 data are an average of at least 3 independent experiments. Each value has a standard deviation of less
b
c
than one geometric mean. Determined with ELISA assay. Cell proliferation inhibition over 72–96 h.
e
dCalculated using MoKa® software. Measurements carried out in PBS pH 7.2 buffer (23 ºC) with serial
f
dilution and LCꢀCLND quantification. Adult female nude mice were administered with compound
suspended in MCT via oral gavage. gMeasured logD value.
We recently introduced a novel class of PI3K inhibitors containing a central benzoxepin scaffold (e.g.
compound 2, Table 1).9ꢀ12 This class of inhibitors was originally discovered by structureꢀguided
optimization of hits from high throughput screening. Optimized compounds displayed good levels of
potency towards the PI3Ks as well as high selectivity over offꢀtargets. Strong structural rationale also exists
to suggest why variations of the substituents on the benzoxepin scaffold lead to isoform selectivity for
PI3Kα over PI3Kβ, an important aspect of mitigating potentially adverse effects on glucose metabolism.10
Very recently, we showed that modifications to the heterocyclic central subunit as well as the peripheral
substituents yielded drugꢀlike compounds such as 2 (Table 1), which had sufficient exposure in vivo to
achieve tumor growth inhibition (TGI) in a mouse xenograft study.12 Despite potent in vitro activity, better
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 45
1
2
3
4
5
6
7
8
even than the clinical compound 1 (GDCꢀ0941, Table 1),13 2 only achieved modest tumor growth inhibition
in an MCF7ꢀneo/HER2 xenograft breast cancer model even when dosed at its maximum absorbable dose due
to solubility limitations (100 mg/kg, Figure 1A). In the graph showing the %TGI by drug dose, GDCꢀ0941
achieves greater tumor growth inhibition at lower doses than 2 (Figure 1B). This outcome was evident
despite 2 achieving significantly higher total area under the concentration–time curve (AUCtot) than 1 in vivo
and being slightly more potent in the proliferation assay with the same cell line evaluated in the xenograft
study (Table 1). According to the free drug hypothesis, in the absence of active transport in vivo, only the
steadyꢀstate unbound drug concentration available to engage the target of interest will have the ability to
impart a pharmacological effect.14 Our group had prior experience with this phenomenon during the
discovery of PI3K/mTOR inhibitor GDCꢀ0980.15 We therefore reasoned that the minimal antiꢀtumor activity
observed for 2 was largely due to its low solubility and limited in vivo unbound exposure achieved. The
objective was then to identify molecules with improved solubility and overall lower unbound clearance
(CLu).
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.
ACS Paragon Plus Environment

Page 5 of 45
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
150
100
50
B.
Compound 1
Compound 2
0
Dose (mg/kg)
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 1. A) Doseꢀcurves of tumor volume response of compound 2 against MCF7ꢀneo/HER2 xenografts
in mice relative to vehicle over the course of 21 days (0.5% methycellulose/0.2% Tweenꢀ80) B) Comparison
of % tumor growth inhibition of compound 1 and 2 in MCF7ꢀneo/HER2 xenografts at the four doses
evaluated.
In order to achieve this end, we envisioned using physicochemical parameters to guide these optimizations.
In particular, one property that we quickly deemed was most important was the calculated distribution
coefficient (cLogD), a measure of molecular lipophilicity. While other factors such as acidity and molecular
size and shape can influence properties such as in vitro protein binding and solubility, reducing cLogD has
been shown to correlate with lower unbound clearance.16 Therefore, reducing cLogD became a key strategy
that we employed to increase unbound exposure in vivo. Specifically, the thiazole central subunit was
targeted for modification in order to reduce overall lipophilicity.
In a previous communication,12 we described a set of novel benzoxepin scaffolds with potent activity
against PI3Kα, wherein the thiazolobenzoxepin subꢀseries was selected for further optimization (Figure 2).
All of the scaffolds described were designed to adopt similar binding topologies in their low energy
conformation and make the same requisite polar and Van der Waals interactions in the ligandꢀbinding site. It
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 45
1
2
3
is important to note that for the five scaffolds (A–E) examined, the imidazobenzoxepin (E) was determined
4
5
6
7
to have the lowest inherent cLogD value in analogous cases (e.g.,
• = Me, Figure 2), albeit very similar to
pyrazolobenzoxazepin (C).17
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 2. Scaffold replacements to reduce lipophilic nature of thiazolobenzoxepin B (cLogD values are
given for analogs terminated by methyl groups).
Based on this singular factor alone, the imidazobenzoxazepin (E) may be expected to have decreased
unbound clearance and improved solubility. We targeted analogs within the imidazobenzoxepin scaffold
that were previously determined to be optimal for the thienobenzoxepin (A) and thiazolobenzoxepin (B)
series based upon prior structural knowledge.11,12 In particular, we selected to retain the pyrazole substituent
at the 8ꢀposition and the triazole at the 2ꢀposition (based on the thienobenzoxepin numbering, Figure 2)
because we were content with the benefits that they provided with regards to potency, selectivity and
pharmacokinetics.
ACS Paragon Plus Environment

Page 7 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
Results and Discussion
Chemistry
6
7
8
9
The preparation of imidazobenzoxazepin compounds was carried out in a similar vein to the strategies
described previously.18 One distinct feature of this scaffold is the presence of the bridgehead nitrogen in the
fused tricycle. To fashion the tricyclic core, we subjected commercially available salicaldehyde 3 to a
condensation reaction with glyoxal and two equivalents of ammonia to provide 4 (Scheme 1). Subsequently,
4 was subjected to 7ꢀmembered ring annulation with 1,2ꢀdibromoethane in high dilution to minimize
undesired bimolecular elimination (E2) byꢀproduct. The corresponding imidazobenzoxazepin tricycle (5)
was then functionalized at the Cꢀ4 position of the imidazole ring employing a twoꢀstep sequence. We
originally observed that electrophilic aromatic halogenation occurred preferentially at the Cꢀ5 at a rate faster
than at the Cꢀ4 in line with previous reports.19 Thus, use of at least two equivalents of Nꢀiodosuccinimide
provided the diꢀiodo intermediate 6 in good yield, and no further iodination was observed beyond the diiode.
This compound was readily monoꢀdeiodinated with a Grignard reagent, taking advantage of the heightened
reactivity at Cꢀ5 to adjust the functionality to the desired bromoiodide 7.20
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
Scheme 1. Synthetic route to the imidazobenzoxazepin bromoiodide core
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 45
1
2
3
4
5
6
7
8
9
At this point, the synthesis of the various analogs diverged. The compounds were assembled by
carbonylative amidation of bromoiodide 7 selectively at the iodide to generate the primary amide 8 (Scheme
2).21 Regioselective triazole formation can be carried out with a variety of hydrazines as previously
described.18 Finally, reaction of the bromide with a diversity of pyrazole boronic acids or esters under
Suzuki conditions provided the compounds (11a-l) described in this manuscript.
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
Scheme 2. Elaboration of bromoiodide 7 to PI3K inhibitors (11a-l).
In vitro Activity
Our initial investigations began with measuring the effect that changing the groups on the pyrazole and
triazole substituents had on PI3Kα biochemical potency, selectivity over PI3Kβ and cellꢀbased activity (pꢀ
Akt and MCF7ꢀneo/HER2 proliferation). In addition, the mouse liver microsomal intrinsic clearance (Clint)
and kinetic solubility were also determined (Table 2). In general, it was noted that changes to the triazole
substituents pointing towards the affinity pocket in PI3Kα, R2 and R3, had the greatest impact to potency
ACS Paragon Plus Environment

Page 9 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
whereas the substituents on the pyrazole pointing out towards solvent, R1, were used to tune PI3Kβ
selectivity as well as physicochemical properties, particularly cLogD.10,12 In the variations evaluated, careful
effort was taken in the design process to avoid introduction of severe metabolic soft spots within the
molecules.
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
Table 2 PI3Kα biochemical activity, βꢀsparing ratio, cellular activities and mouse liver microsome intrinsic
clearance as a function of structural changes.a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 45
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
b
aSee Table 1 for assay descriptions. Ki values are an average of 3 independent experiments. Each value
has a standard deviation of less than one geometric mean. cRatio of PI3Kβ IC50 over PI3Kα apparent Ki, x =
fold; dCompounds were incubated with mouse liver microsomes for 1 h and % remaining was determined by
LCꢀMS/MS analysis. eNot tested, the compound was not evaluated in this assay.
ACS Paragon Plus Environment

Page 11 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
We began our survey with the imidazobenzoxazepin 11a bearing an isopropyl substituent on the triazole
similar to the thiazolobenzoxepin 2. Biochemical and cellular potencies were reasonable but intrinsic
microsomal clearance was high at 550 ml/min/kg. This rate of metabolic clearance was not tolerable. A
simple substitution with a 1ꢀmethyl group on the pyrazole near the solventꢀexposed region of the ligand and
a 3ꢀmethyl group on the triazole helped to significantly improve mouse liver microsome (MLM) clearance
(11b). However, the kinetic solubility of 2 ꢀM remained unacceptably low. Placement of a polar substituent
such as a hydroxyethyl tail on the pyrazole near the solventꢀexposed front pocket was effective in reducing
cLogD to 2.6 for 11c while maintaining good biochemical activity. This change was well tolerated and
resulted in a slight improvement of cellꢀbased potency and metabolic clearance. Further reduction of cLogD
by removing the isopropyl substituent on the triazole altogether and replacing with either the less lipophilic
NꢀH (11d) and Nꢀmethyl (11e) led to 50ꢀfold and 5ꢀfold loss in potency, respectively. Reꢀinsertion of the
isopropyl group and removal of the 3ꢀmethyl group resulted in a compound with increased potency and
microsomal clearance (11f vs. 11c). Further modifications to the triazole revealed that a 2,2,2ꢀtrifluoroethyl
substituent in 11g was an effective replacement of the isopropyl, with comparable biochemical and cellular
activity. Metabolic clearance was also slightly improved with this replacement. Next, we investigated the
substituents at the 1ꢀposition of the pyrazole to help improve kinetic solubility. As mentioned previously,
this portion of the inhibitor extends toward solvent, so we reasoned that various groups would be tolerated in
this region. Therefore, a number of polar groups at this position were evaluated with all of the analogs
showing effective kinase inhibition with subꢀnM activity against PI3Kα, good cellꢀbased activity and kinetic
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
solubility (11h-j). Finally, the tertiary alcohol (11k) and the αꢀquaternary amide (11l) pyrazole substituents
were also evaluated. These latter compounds possessed a favorable balance of all of the properties that we
optimized for: enzymatic and cellular activity, microsomal clearance and kinetic solubility. It is noteworthy
that all the compounds described had desirable levels of selectivity over PI3Kβ, which was attributed to the
pyrazole substitution that they shared in common.10,12
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 45
1
2
3
4
5
Pharmacokinetic properties
6
7
8
9
From the data gathered thus far, 11c, 11k and 11l were selected for progression into in vivo
pharmacokinetic (PK) profiling. A single dose of each compound was administered to adult female nude
mice either intravenously at 1 mg/kg or orally at 25 mg/kg. The data for these three compounds are
summarized in Table 3. When compared to the original thiazolobenzoxepin inhibitor 2, the
imidazobenzoxazepin inhibitors (11c, 11k and 11l) all showed improved antiꢀproliferative cellular activity in
MCF7ꢀneo/HER2 line (138, 81 and 25 nM, respectively). Additionally, in line with their low in vitro
clearance, the mouse plasma clearance (CLp) values were also low for these optimized compounds. More
importantly, in each case both the unbound clearance (CLu) and, correspondingly, the unbound oral exposure
(AUCu) were significantly (>10x) improved relative to thiazolobenzoxepin 2. Compound 11l provided the
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
highest unbound exposure of the compounds tested (AUCu = 42 ꢀM hr). Based on these results, we selected
•
11l to evaluate in an in vivo tumor xenograft study in a headꢀtoꢀhead comparison with compound 1.
Table 3. Cellular antiꢀproliferative activity and pharmacokinetic profiles of imidazobenzoxazepins
(compounds 11c, 11k and 11l) compared to thiazolobenzoxepin 2.
ACS Paragon Plus Environment

Page 13 of 45
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
a % Plasma protein binding measured with equilibrium dialysis. b CLu = CLp / fu. c AUCu = AUCtot
X
fu; fu =
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
(100 – %PPB) / 100.
In vivo Efficacy and Pharmacodynamics
The in vivo efficacy of compound 11l was compared to compound 1 in the MCF7ꢀneo/Her2 xenograft model
grown in nude mice. Daily, oral administration of compound 1 at 45 mg/kg resulted in 69% tumor growth
inhibition (TGI) when compared to vehicle treated mice (purple curve, Figure 3). Daily administration of
compound 11l orally at 1.4, 2.8, 5.8, 11.25, or 22.5 mg/kg resulted in doseꢀdependent increase in TGI (19%,
76%, 95%, 103%, and 123%, respectively) and tumor regressions when compared to vehicle treated mice.
The increased potency and unbound concentration of compound 11l resulted in comparable TGI (76%) to
compound 1 (69%) at a dose that is approximately 16ꢀfold lower (2.8 mg/kg versus 45 mg/kg). Treatment of
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 45
1
2
3
both compound 1 and 11l were well tolerated with less than 10% body weight loss observed compared to
4
5
vehicle controls (data not shown).
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 3 Doseꢀresponse curves of fitted tumor volumes in response to 1 and 11l at the doses shown against
MCF7ꢀneo/HER2 xenografts in mice relative to vehicle (MCT; 0.5% methycellulose/0.2% Tweenꢀ80) after
daily (QD) oral (PO) dosing for 20 days.
The relationship between pharmacokinetics (PK) and pharmacodynamics (PD) of compound 11l relative to
compound 1 was also investigated in the MCF7ꢀneo/Her2 xenograft model (Figure 4). The phosphorylation
levels of Akt in MCF7ꢀneo/HER2 tumors, as well as plasma drug concentrations of both compounds, were
evaluated 1 hour after a single dose of vehicle (MCT), compound 1 at 45 mg/kg or compound 11l at 1.4, 2.8,
5.8, 11.25, or 22.5 mg/kg. Compound 1 resulted in 58% decrease of phosphorylation of Akt (pAktSer473
)
levels while a 16ꢀfold lower dose of compound 11l at 2.8 mg/kg resulted in a similar decrease in Akt
phosphorylation (59%). The comparable decreases in pAkt between these doses of compound 11l and 1 is
ACS Paragon Plus Environment

Page 15 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
likely due to similar unbound plasma drug concentrations that were achieved in mice after 1 hour of dosing
(Figure 4).
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) Quantification of pAktSer473 levels in MCF7ꢀneoꢀHer2 tumor xenografts 1 hour after a single dose
of MCT vehicle (0.5% methycellulose/0.2% Tweenꢀ80), compound 1 at 45 mg/kg or compound 11l at 1.4,
2.8, 5.8, 11.125, or 22.5 mg/kg. Phosphorylated Akt (pAktSer473) and total Akt (tAkt) levels were measured
by Mesoscale Discovery assay and values are expressed as ratio of pAkt/tAkt. Error bars represent SEM for
tumor xenograft samples from four different animals. Corresponding plasma free drug concentrations (closed
diamonds) for each dose of compound 1 or compound 11l are plotted on the right yꢀaxis.
To better understand the differences in pharmacokinetic properties between thiazolobenzoxepin (e.g. 2) and
imidazobenzoxazepin (e.g. 11l) inhibitors, it is necessary to consider the impact of reducing cLogD had on
metabolic clearance. This analysis can be applied to all five scaffolds (A–E) previously introduced in Figure
2 in order to generalize the observation.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 45
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 5. Pie charts of mouse liver microsomal clearance categories according to scaffold A–E. Numbers
above individual pie charts correspond to sample size. Green = stable, <27 mL/min/kg; orange = moderate,
27ꢀ63 mL/min/kg; and red = high, >63 mL/min/kg.
When compounds from each of the subꢀseries were incubated with mouse liver microsomes (MLM), the
imidazobenzoxepin (E) had the greatest ratio of compounds (>75%) measured to be in the moderate or low
clearance category (CLhep < 63 mL/min/kg). In contrast, the other scaffolds had more compounds fall in the
high clearance category. This outcome can be attributed to two main factors: first, this series was the least
lipophilic as determined by cLogD; secondly, and more important to our understanding, the oxidative
potential of the αꢀmethylene position vary depending on the adjacent heterocycles present in each core.22 By
blocking this particular metabolic soft spot, we can effectively lower clearance rate, all else being equal. In
silico, the degree to which this oxidation could occur was determined using the computational program
MetaSite® (Figure 6).23 The program indicated that the preferred site of oxidation on the core was at this soft
spot for the compounds analyzed (shaded spheres). The oxidation at this site was severe for
thienobenzoxepin A and thiazolobenzoxepin B. To a lesser degree, pyrazolobenzoxazepin C and
ACS Paragon Plus Environment

Page 17 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pyrazolobenzoxepin D were also calculated to be mildly susceptible to oxidation. However,
imidazobenzoxazepin E was calculated as not experiencing any such oxidation at this αꢀmethylene position.
One possible explanation for the attenuated reactivity at this site for E was that the nitrogen atom directly
attached to the position in question was shielding it from cytochrome P450 enzymes. The polarization of this
methylene group by the imidazole nitrogen can alter its susceptibility towards oxidation, thereby shifting
metabolism on the molecule elsewhere. We confirmed that these calculations were relevant in a real setting
by performing in vitro metabolite identification using compound 2 and 11l following incubation with human
hepatocytes. Indeed, the formation of a substantial amount of the oxidation product corresponding to oxo-2
was observed after 4 h (Scheme 3a). Conversely, exposure of imidazobenzoxazepin 11l to similar conditions
yielded only minimal detectable quantity of the product corresponding to αꢀmethylene oxidation (Scheme
3b).24
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
NH2
NH2
NH2
NH2
NH2
O
S
O
N
O
O
N
O
N
O
O
O
O
O
N
N
N
N
S
N
N
N
N
N
N
N
N
N
N
N
F3C
F3C
F3C
F3C
F3C
N
N
N
N
A
B
C
D
E
Figure 6. MetaSite® predictions for site of CYPꢀmediated oxidation of the αꢀmethyl group adjacent to the
heterocyclic subunit at the center of each scaffold A–E. Red spheres indicate severe magnitude of such
oxidation while gray spheres indicate moderate oxidation.
Scheme 3. Thiazolobenzoxepin 2 was extensively converted to the oxidized product when incubated with
human hepatocytes. The corresponding product is minimal when 11l is subjected to similar conditions.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 45
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
By masking a metabolic site, we managed to reduce both the in vivo clearance and (because we
concomitantly increased free fraction) the unbound clearance as well. To confirm that the improvement to
the unbound plasma clearance was general for the imidazobenzoxazepin (E) scaffold relative to the
thiazolobenzoxepin (B), we plotted CLu for ten matchedꢀpair comparisons where compounds differed only
by the composition of their central core, imidazole versus thiazole (Figure 7). In all cases evaluated, E
consistently showed lower CLu for each pair with direct analogs in both scaffolds. Ultimately, our strategy of
transitioning to the imidazobenzoxazepin scaffold proved beneficial in generating wellꢀbehaved compounds
that achieve high unbound exposures in vivo.
ACS Paragon Plus Environment

Page 19 of 45
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 7. Twoꢀdimensional plot of ten matchedꢀpair comparisons with unbound clearance (CLu) of
imidazobenzoxazepins (E) on the yꢀaxis and thiazolobenzoxepins (B) on the xꢀaxis. The diagonal blue line
represents the line of unity.
Therapeutic candidate profile
Consequently, compound 11l was extensively characterized to determine if it met our criteria for a
therapeutic candidate for use in clinical trials (Figure 8). Earlier, we established that it reached all
expectations with regard to potency, PK/PD and efficacy in human xenograft mouse models. Selectivity
against the broader kinome was also determined by screening against Invitrogen’s SelectScreen panel of
kinases. Out of 235 kinases that were screened with 11l at 1 ꢀM, we found that only 5 kinases were inhibited
>50%: PI3KꢀC2 beta, PI3Kα, PI3Kδ, PI3Kγ and hVPS34 at 80.4%, 98.6%, 97.9%, 90.6% and 69.9%,
respectively. Notably, PI3Kβ was not included in the panel of kinases screened. Measurement of IC50s for
potently inhibited kinases showed that we indeed maintained biochemical selectivity over the nonꢀClass I
PI3Ks, with IC50 for PI3KꢀC2 beta and hVPS34 measuring 292 nM and 374 nM, respectively. Compared to
the 290 pM activity against PI3Kα, we achieve a 1000ꢀfold selectivity against these potential offꢀtargets. The
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 45
1
2
3
4
5
6
7
8
9
favorable pharmacokinetic profile evident in nude mouse was also achieved in Sprague–Dawley rats. We
observed low clearance (CLP = 1.8 ml/min/kg) and high oral bioavailability (F = 99%) consistent with in
vitro experiments carried out in hepatocytes that suggested low metabolic clearance. Auspiciously, the
clearance predicted by human hepatocytes is low, suggesting low plasma clearance could be achieved in
vivo. After these and further analyses, which included toxicological and safety evaluation, we determined
that 11l had a satisfactory profile to progress into human clinical trials.25
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. Preclinical candidate profile of PI3Kβꢀsparing inhibitor 11l.
Conclusions
In summary, we discovered a novel class of compounds for the inhibition of PI3K driven tumors. The key
feature for optimization was improving unbound exposure (AUCu) by reducing cLogD as a design
parameter. This improvement had profound effects on tumor growth inhibition in vivo. Relative to a closely
ACS Paragon Plus Environment

Page 21 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
related benzoxepin scaffold evaluated, the imidazobenzoxazepin had better tumor growth inhibition at even
lower drug doses. In line with the free drug hypothesis, reducing the intrinsic plasma clearance (in this case,
a result of removing metabolic soft spots) and increasing free exposure resulted in more efficacious drugs.
This approach was essential because we had access to a substantial set of data for comparing direct analogs
in several closely related scaffolds. As a result, we quickly settled on a suitable scaffold to optimize in order
to achieve the desired profile. Our optimization efforts led to the identification of 11l (or GDCꢀ0032),
currently undergoing clinical development for use in PI3Kꢀrelated cancers.
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
EXPERIMENTAL SECTION
Computation.
Using MoKa Software (version 1.1.0, Molecular Discovery) and a custom pKa model augmented with
Roche internal data, logarithm of the acid dissociation constant of most basic center (cpKa_MB), logarithm of
the partition coefficient (clogP), and logarithm of the distribution coefficient at pH 7.4 (clogD) were
calculated. Metabolic soft spots were predicted using MetaSite. The program calculates a probability of
CYPꢀbased metabolism at certain sites of a molecule by considering both accessibility of the atoms to the
reactive heme and their reactivity in the appropriate reaction mechanism.23 Accessibility was derived by
comparing GRIDꢀbased fingerprint representations of a CYP450 binding site and a 3D structure of the
substrate. Substrate reactivity was computed from a mixture of molecular orbital calculations and fragment
recognition.
Chemistry. General Methods. All solvents and reagents were used as obtained. Reactions involving air or
moisture sensitive reagents were carried out under nitrogen atmosphere. Microwave reactions were
performed using CEM Discover and Biotage Initiator reactors. NMR spectra were recorded in a deuterated
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 45
1
2
3
4
5
6
7
8
9
solvent with a Bruker Avance 400ꢀ or 500ꢀMHz NMR spectrometer, and referenced to trimethylsilane
(TMS). Chemical shifts are expressed as δ units using TMS as the external standard (in NMR description, s
= singlet, d = doublet, t = triplet, q = quartet, m = multiplet, and br = broad peak). All coupling constants (J)
are reported in Hertz. Mass spectra were measured with a Finnigan SSQ710C spectrometer using an ESI
source coupled to a Waters 600MS HPLC system operating in reverse mode with an Xꢀbridge Phenyl
column of dimensions 150 mm by 2.6 mm, with a 5 ꢀm sized particles. Analytical purity was >95% unless
stated otherwise. The following analytical method was used to determine chemical purity of final compounds
unless otherwise stated: HPLCꢀAgilent 1200, water with 0.05% TFA, acetonitrile with 0.05% TFA, Agilent
Zorbax SDꢀC18, 1.8 ꢀM, 2.1x30 mm, 40 ºC, 3ꢀ95% B in 8.5 min, 95% in 2.5 min, 400 ꢀL/min, 220 nm and
254 nm, equipped with Agilent quadrupole 6140, ESI positive, 110ꢀ800 amu.
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
Ethyl 4-(3-bromophenoxy)butanoate (12)
Solid 3ꢀbromophenol (10.0 g, 58 mmol) was added portion wise to a stirred suspension of K2CO3 in
acetone (100 mL) at room temperature. Sodium iodide (NaI, 1.0 g) was added, followed by ethylꢀ4ꢀ
bromobutyrate (9.2 mL, 64 mmol). The reaction mixture was heated at 80 ˚C overnight, cooled to room
temperature, diluted with water and extracted with ethyl acetate to give ethyl 4ꢀ(3ꢀbromophenoxy)butanoate
(12).
4-(3-bromophenoxy)butanoic acid (13)
Ethyl 4ꢀ(3ꢀbromophenoxy)butanoate (12) was taken up in 100 mL THF and 50 mL water and treated with
lithium hydroxide LiOH (hydrate, 4.9 g). The whole was heated at 50 ˚C for 2 days. The mixture was
cooled to room temperature and acidified to pH 1 with 2N HCl. The aqueous was extracted with
ACS Paragon Plus Environment

Page 23 of 45
Journal of Medicinal Chemistry
1
2
3
ethylacetate. The combined organics were washed with brine and dried over sodium sulfate to give crude 4ꢀ
4
5
1
(3ꢀBromophenoxy)butanoic acid (13) as a sticky solid. H NMR (DMSOꢀd6, 500 MHz) 7.24 (m, 1H), 7.13
6
(m, 1H), 7.11 (m, 1H), 6.95 (m, 1H), 3.99 (m, 2H), 2.37 (m, 2H), 1.94 (m, 2H).
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
8-bromo-3,4-dihydrobenzo[b]oxepin-5(2H)-one (14)
To a stirred suspension of polyphosphoric acid (PPA, ca. 60 g) and celite (ca. 40 g) in 100 mL toluene was
added crude 4ꢀ(3ꢀbromophenoxy)butanoic acid (13) (ca. 58 mmol) in one portion, 10 mL toluene rinse. The
o
resultant suspension was heated at 110 C for 5 hr. The toluene was decanted through a plug of celite and
the remaining slurry was washed repeatedly with toluene and ethylacetate. The eluent was concentrated and
purified by flash column chromatography (4:1 hex:EtOAc) to give 8ꢀbromoꢀ3,4ꢀdihydrobenzo[b]oxepinꢀ
1
5(2H)ꢀone (7 g, ca. 50% yield). H NMR (DMSOꢀd6, 500 MHz) 7.55 (d, J = 8.5 Hz, 1H), 7.37 (d, J = 1.5
Hz, 1H), 7.35 (dd, J = 8.5, 1.5 Hz, 1H), 4.24 (t, J = 6.5 Hz, 2H), 2.79 (t, J = 7.0 Hz, 2H), 2.14 (m, 2H).
4,8-dibromo-3,4-dihydrobenzo[b]oxepin-5(2H)-one (15)
To a stirred solution of 8ꢀbromoꢀ3,4ꢀdihydroꢀ2Hꢀbenzo[b]oxepinꢀ5ꢀone (3.10 g; 12.8 mmol) in Et2O at 0
°C was added Br2 (625 ꢁl; 12.2 mmol) and the reaction mixture was allowed to warm to r.t. over 2 h.
Volatiles were evaporated and the residue purified by flash column chromatography (0ꢀ20% EtOAc in
hexanes) to give 4,8ꢀdibromoꢀ3,4ꢀdihydrobenzo[b]oxepinꢀ5(2H)ꢀone (15) as a colorless solid (4.0 g, 96%
1
yield). H NMR (400 MHz, CDCl3) δ 7.18ꢀ7.21 (m, 2H), 7.55 (d, J = 8.0, 1H), 4.87 (app. t, J = 5.2, 2H),
4.33ꢀ4.39 (m, 1H), 4.09ꢀ4.16 (m, 1H), 2.78ꢀ2.91 (m, 1H), 2.40ꢀ2.49 (m, 1H).
8-Bromo-4,5-dihydro-6-oxa-3-thia-1-aza-benzo[e]azulene-2-carboxylic acid ethyl ester (16)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 45
1
2
3
4
5
6
7
8
9
To a stirred solution of 4,8ꢀdibromoꢀ3,4ꢀdihydrobenzo[b]oxepinꢀ5(2H)ꢀone (2.17 g; 6.8 mmol) in EtOH
(70 ml) was added ethyl thiooxamate (2.72 g; 20.4 mmol) and the reaction mixture was heated at reflux
temperature over 4 d. The reaction mixture was concentrated and the residue purified by flash column
chromatography (0ꢀ20% EtOAc in hexanes) to give 8ꢀBromoꢀ4,5ꢀdihydroꢀ6ꢀoxaꢀ3ꢀthiaꢀ1ꢀazaꢀ
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
1
benzo[e]azuleneꢀ2ꢀcarboxylic acid ethyl ester (16) as a yellow solid (1.08 g, 45% yield). H NMR (400
MHz, CDCl3) δ 8.36 (d, J = 8.4, 1H), 7.16ꢀ7,22 (m, 2H), 4.42 (q, J = 7.2, 2H), 4.31 (app. t, J = 5.2, 2H), 3.32
(app. t, J = 5.2, 2H), 1.38 (t, J = 7.2, 3H).
8-Bromo-4,5-dihydro-6-oxa-3-thia-1-aza-benzo[e]azulene-2-carboxylic acid (17)
To a stirred solution of 8ꢀBromoꢀ4,5ꢀdihydroꢀ6ꢀoxaꢀ3ꢀthiaꢀ1ꢀazaꢀbenzo[e]azuleneꢀ2ꢀcarboxylic acid ethyl
ester in THF (10 ml) and MeOH (5 ml) was added a solution of NaOH (117 mg, 2.9 mmol) in water (2 ml).
The reaction mixture was stirred at room temperature for 2 h upon which time it was acidified with 2 M HCl.
The aqueous was extracted with CH2Cl2 and the combined organics dried (MgSO4) and concentrated to give
8ꢀBromoꢀ4,5ꢀdihydroꢀ6ꢀoxaꢀ3ꢀthiaꢀ1ꢀazaꢀbenzo[e]azuleneꢀ2ꢀcarboxylic acid as a colorless solid (396 mg;
1
83%). H NMR (400 MHz, CDCl3) δ 8.22 (d, J = 8.4, 1H), 7.19ꢀ7.24 (m, 2H), 4.33 (t, J = 5.2, 2H), 3.36 (t, J
= 5.2, 2H).
2-{4-[2-(2-Isopropyl-2H-[1,2,4]triazol-3-yl)-4,5-dihydro-6-oxa-3-thia-1-aza-benzo[e]azulen-8-yl]-
pyrazol-1-yl}-ethanol (2)12
8ꢀbromoꢀ2ꢀ(2ꢀisopropylꢀ2Hꢀ[1,2,4]triazolꢀ3ꢀyl)ꢀ4,5ꢀdihydroꢀ6ꢀoxaꢀ3ꢀthiaꢀ1ꢀazaꢀbenzo[e]azulene (17) was
reacted with ethyl 2ꢀ(4ꢀ(4,4,5,5ꢀtetramethylꢀ1,3,2ꢀdioxaborolanꢀ2ꢀyl)ꢀ1Hꢀpyrazolꢀ1ꢀyl)acetate under typical
Suzuki coupling conditions (vide infra) to give {4ꢀ[2ꢀ(2ꢀIsopropylꢀ2Hꢀ[1,2,4]triazolꢀ3ꢀyl)ꢀ4,5ꢀdihydroꢀ6ꢀ
oxaꢀ3ꢀthiaꢀ1ꢀazaꢀbenzo[e]azulenꢀ8ꢀyl]ꢀpyrazolꢀ1ꢀyl}ꢀacetic acid ethyl ester as a colorless solid. In a separate
ACS Paragon Plus Environment

Page 25 of 45
Journal of Medicinal Chemistry
1
2
3
preparation,
a
solution of {4ꢀ[2ꢀ(2ꢀIsopropylꢀ2Hꢀ[1,2,4]triazolꢀ3ꢀyl)ꢀ4,5ꢀdihydroꢀ6ꢀoxaꢀ3ꢀthiaꢀ1ꢀazaꢀ
4
5
6
7
8
9
o
benzo[e]azulenꢀ8ꢀyl]ꢀpyrazolꢀ1ꢀyl}ꢀacetic acid ethyl ester in THF (200 mL) was cooled to 0 C and treated
drop wise with a solution of 1 M LiAlH4 (in THF, 23 mL, 23 mmol, 2.5 eq). After 1 h, LC/MS indicated
consumption of starting material. A solution of saturated sodium sulfate was added slowly until H2 evolution
ceased. ~30 g of solid magnesium sulfate was added and the whole was stirred for 20 min. Filtration over
celite (EtOAc and DCM) followed by concentration of the filtrate gave a crude residue that was purified by
rpꢀHPLC to give 1.4 g of 2 (43%). 1H NMR (500 MHz, DMSOꢀd6) δ 8.36 – 8.29 (m, 1H), 8.23 (s, 1H), 8.11
(d, J = 1.7 Hz, 1H), 7.97 (s, 1H), 7.44 (dd, J = 8.3, 1.8 Hz, 1H), 7.31 (d, J = 1.8 Hz, 1H), 5.84 (hept, J = 6.6
Hz, 1H), 4.95 (t, J = 5.3 Hz, 1H), 4.39 (t, J = 5.0 Hz, 2H), 4.16 (dd, J = 7.1, 4.3 Hz, 2H), 3.77 (q, J = 5.6 Hz,
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
13
2H), 3.44 (t, J = 5.1 Hz, 2H), 1.56 (d, J = 6.5 Hz, 6H). C NMR (126 MHz, DMSOꢀd6) δ 158.36, 152.50,
150.65, 148.05, 148.01, 145.40, 136.37, 134.51, 134.48, 133.85, 130.12, 128.05, 121.46, 120.50, 119.83,
116.65, 68.70, 59.94, 54.25, 51.58, 40.06, 39.97, 39.90, 39.80, 39.73, 39.63, 39.60, 39.47, 39.39, 39.30,
39.13, 38.97, 29.60, 22.02, 21.99. HRMS (ESI+): m/z (M+Na+) calcd: 445.1423, found: 445.1518. Melting
point: 180 ºC.
5-bromo-2-(1H-imidazol-2-yl)phenol (4)
4ꢀBromoꢀ2ꢀhydroxybenzaldehyde (1.0 g, 5 mmol), 40% aqueous solution of ethanedial (3.6 g, 24.9 mmol)
and 50% aqueous ammonia (2.5 g) in methanol (20 mL) was stirred for 2 h or longer until the reaction is
completed as determined by TLC. The solvent was removed under vacuum pressure and the residue was
partitioned between EtOAc and water. The mixture was then filtered to remove the precipitate. The solution
pH was adjusted to 5ꢀ6 by careful addition of 1 N HCl. The aqueous layer was extracted with EtOAc three
times. The combined organic portions were washed with water, brine and dried over MgSO4. The resulting
residue was purified by ISCO chromatography (30% EtOAc/DCM) to provide 0.9 g (80% yield) of 5ꢀbromoꢀ
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 45
1
2
3
4
2ꢀ(1Hꢀimidazolꢀ2ꢀyl)phenol (4) as a yellow solid. 1H NMR (300 MHz, DMSOꢀd6) δ 11.13 (bs, 1H), 10.23 (s,
1H), 7.57 (d, J = 8.3 Hz, 1H), 7.21 (d, J = 1.8 Hz, 1H), 7.18 – 7.09 (m, 1H), 3.35 (bs, 1H).
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
9-bromo-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepine (5)
A mixture of 5ꢀbromoꢀ2ꢀ(1Hꢀimidazolꢀ2ꢀyl)phenol (0.9 g, 4 mmol), 1,2ꢀdibromoethane (1.3 mL, 15 mmol)
and cesium carbonate (4.9 g, 15 mmol) in DMF (20 mL) was heated to 90 ⁰C for 12 h. The mixture was
partitioned between water and EtOAc. The organic layer was separated, washed with water, brine and dried
over MgSO4. Concentrated in vacuo to yield the pure 9ꢀbromoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀ
d][1,4]oxazepine, 0.8 g (75% yield) (5). 1H NMR (300 MHz, CDCl3) δ 8.41 (d, J = 8.6 Hz, 1H), 7.31 – 7.21
(m, 3H), 7.19 (d, J = 1.1 Hz, 1H), 7.01 (d, J = 1.0 Hz, 1H), 4.59 – 4.26 (m, 4H).
9-bromo-2,3-diiodo-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepine (6)
A mixture of 9ꢀbromoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepine (0.8 g, 3 mmol) and Nꢀ
iodosuccinimide (1.87 g, 8.3 mmol) in DMF was stirred at room temperature for 48 h. The mixture was
diluted with EtOAc, washed with 5% sodium bicarbonate solution, 10% sodium thiosulfate, water and brine
and the organic layer was dried over MgSO4 and concentrated to a solid residue. Purification by ISCO
chromatography (30% EtOAc/heptane) yielded 1.2 g (78% yield) of 9ꢀbromoꢀ2,3ꢀdiiodoꢀ5,6ꢀ
1
dihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepine (6). H NMR (300 MHz, DMSOꢀd6) δ 8.19 (d, J = 8.6 Hz,
1H), 7.35 – 7.18 (m, 2H), 4.58 – 4.24 (m, 4H).
9-bromo-2-iodo-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepine (7)
ACS Paragon Plus Environment

Page 27 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A 3.0 M solution of ethylmagnesium bromide in diethyl ether (1.1 mL, 3.3 mmol) was added dropwise to a
cold suspension of 9ꢀbromoꢀ2,3ꢀdiiodoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepine (1.1 g, 2.2 mmol)
in THF at –15 C. The mixture was stirred at cold temperature and monitored by LC/MS. After 1 h, full
conversion was observed and the reaction was poured into saturated ammonium chloride and extracted with
EtOAc. The organic extracts were washed with water, brine and dried over MgSO4 and concentrated in
vacuo. The crude residue was purified by flash column chromatography (EtOAc) to provide 0.7 g (81%
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
1
yield) 9ꢀbromoꢀ2ꢀiodoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepine (7) as a white solid. H NMR (300
MHz, DMSOꢀd6) δ 8.22 (d, J = 8.5 Hz, 1H), 7.55 (s, 1H), 7.27 (dt, J = 5.5, 1.9 Hz, 2H), 4.43 (q, J = 6.1 Hz,
4H). 13C NMR (75 MHz, DMSO) δ 155.9, 145.1, 131.4, 128.8, 125.8, 123.4, 122.4, 117.4, 83.7, 69.0, 49.8.
9-bromo-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepine-2-carboxamide (8)
9ꢀbromoꢀ2ꢀiodoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepine (1.5 g, 3.8 mmol) was placed in a 100ꢀ
mL rbf and bis(triphenylphosphine) palladium(II) chloride (142 mg, 0.202 mmol) was added followed by
DMF (45 mL) and hexamethyldisilazane (4.34 mL, 20.6 mmol). The entire solution was purged with a CO
balloon and sealed with the CO balloon attached. The reaction flask was heated at 70 °C for 2 h. LC/MS
indicated clean conversion at the end of this period. Cooled to room temperature and poured into 1 N HCl
(30 mL). Stirred for 5 min and neutralized with sat. aq. NaHCO3 soln. Extracted three times with EtOAc,
dried over MgSO4, filtered and concentrated in vacuo. Triturated with IPA and the solids were collected after
filtration and EtOAc wash. This provided 734 mg (62% yield) of 9ꢀbromoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀ
d][1,4]oxazepineꢀ2ꢀcarboxamide (8) as a tan solid. 1H NMR (400 MHz, CDCl3) δ 8.36 (d, J = 8.5, 1H), 7.63
(s, 1H), 7.24 (dd, J = 7.2, 4.2, 1H), 7.09 – 6.99 (m, 1H), 4.51 – 4.36 (m, 4H). LC/MS (ESI+): m/z 310
(M+H).
9-bromo-2-(1-isopropyl-1H-1,2,4-triazol-5-yl)-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepine (9a)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 45
1
2
3
4
5
6
7
8
9
9ꢀbromoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepineꢀ2ꢀcarboxamide (4.93 g, 16.0 mmol, a separate
lot from previously) was taken up in 1,1ꢀdimethoxyꢀN,Nꢀdimethylmethanamine (25 mL, 0.18 mol) and 1,2ꢀ
dimethoxyethane (66.5 mL, 0.640 mol). The heterogeneous mixture was stirred very vigorously and heated
at 65 °C for 1 h. LC/MS showed complete consumption of starting material at the end of this period. The
reaction mixture was concentrated in vacuo and carried on to the subsequent reaction with no further
purification steps applied. The crude product from the previous reaction (5.8 g, 16.0 mmol) was suspended in
glacial acetic acid (53.2 mL) and isopropylhydrazine hydrochloride (4.36 g, 39.4 mmol) was added. The
mixture was heated at 100 °C for 2 h. The reaction vessel was cooled to room temp and the solvent was
removed in vacuo. The resultant residue was dry loaded onto silica gel and purified by ISCO
chromatography (120 g column, 100% EtOAc). Totally, 2.3 g (39% yield) of 9ꢀbromoꢀ2ꢀ(1ꢀisopropylꢀ1Hꢀ
1,2,4ꢀtriazolꢀ5ꢀyl)ꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepine (9a) was isolated over the two steps. 1H
NMR (400 MHz, DMSOꢀd6) δ 8.34 (d, J = 8.6, 1H), 7.95 (s, 1H), 7.91 (s, 1H), 7.36 (dd, J = 8.7, 2.0, 1H),
7.30 (d, J = 2.0, 1H), 5.85 (dt, J = 13.3, 6.6, 1H), 4.55 (d, J = 15.5, 4H), 1.48 (d, J = 6.6, 6H). LC/MS
(ESI+): m/z 376 (M+H).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
9-bromo-2-(1-(2,2,2-trifluoroethyl)-1H-1,2,4-triazol-5-yl)-5,6-dihydrobenzo[f]imidazo[1,2-
d][1,4]oxazepine (9b)
Following the procedure described above, 9ꢀbromoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepineꢀ2ꢀ
carboxamide and (2,2,2ꢀtrifluoroethyl)hydrazine were reacted to give 9ꢀbromoꢀ2ꢀ(1ꢀ(2,2,2ꢀtrifluoroethyl)ꢀ
1
1Hꢀ1,2,4ꢀtriazolꢀ5ꢀyl)ꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepine (9b). H NMR (400 MHz, DMSOꢀ
d6) δ 8.28 (t, J = 21.9, 1H), 8.11 (t, J = 7.9, 2H), 7.51ꢀ7.35 (m, 1H), 7.32 (d, J = 2.0, 1H), 5.88 (q, J = 8.8,
2H), 4.76ꢀ4.29 (m, 4H). LC/MS (ESI+): m/z 413.9 (M+H).
9-bromo-2-(1-isopropyl-3-methyl-1H-1,2,4-triazol-5-yl)-5,6-dihydrobenzo[f]imidazo[1,2-
d][1,4]oxazepine (9c)
ACS Paragon Plus Environment

Page 29 of 45
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
9ꢀbromoꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepineꢀ2ꢀcarboxamide (5.1 g, 16 mmol) was taken up
in toluene (100 mL) and treated with dimethylacetamideꢀdimethylacetal (26.5 mL, 178 mmol). This mixture
was heated 2 h at 120 °C in a sealed vessel. LC/MS indicated complete consumption of starting material and
formation of the acylamidine intermediate together with the hydrolysis byꢀproduct (40:60 mixture). This
crude mixture was concentrated in vacuo to a yellow solid and combined with isopropylhydrazine
hydrochloride (3.59 g, 32.4 mmol) in Acetic acid (50 mL) and heated at 95 °C for 1 h. The reaction was
observed to be completed as determined by LC/MS. The reaction vessel was subsequently cooled to room
temp and the volatiles were evaporated. The solid was dryꢀloaded onto an ISCO column and purified (120 g
column, 10% MeOH/EtOAc) by flash column chromatography. This provided 5.9 g (95% yield) of 9ꢀ
bromoꢀ2ꢀ(1ꢀisopropylꢀ3ꢀmethylꢀ1Hꢀ1,2,4ꢀtriazolꢀ5ꢀyl)ꢀ5,6ꢀdihydrobenzo[f]imidazo[1,2ꢀd][1,4]oxazepine
(9c) as a tanꢀcolored solid. 1H NMR (400 MHz, CDCl3) δ 8.38 (d, J = 8.6, 1H), 7.63 (s, 1H), 7.31 – 7.20 (m,
5H), 5.94 – 5.76 (m, 1H), 5.30 (s, 1H), 4.46 (ddd, J = 7.9, 6.2, 3.6, 4H), 2.41 (s, 3H), 1.67 – 1.45 (m, 7H).
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
1-[2-(Tetrahydro-pyran-2-yloxy)-ethyl]-4-(4,4,5,5-tetramethyl-[1,3,2]dioxaborolan-2-yl)-1H-pyrazole
(10f)
To a solution of 4ꢀpyrazole boronic acid pinacol ester (2.0 g, 10.3 mmol) in anhydrous DMF (20 mL) was
added cesium carbonate (4.03 g, 12.4 mmol) and the mixture stirred at RT for 10 minutes. 2ꢀ(2ꢀ
Bromoethoxy)tetrahydroꢀ2Hꢀpyran (1.87 mL, 12.4 mmol) was added in two portions and the mixture was
heated to 70°C. After heating for 18 hours the mixture was allowed to cool to RT and partitioned between
water (100 mL) and EtOAc (100 mL). The aqueous layer was washed with EtOAc (3 x 20 mL) and the
combined organic layers washed with water (3 x 100 mL) followed by brine, dried (Na2SO4) and
concentrated in vacuo. The resultant residue was purified by flash chromatography (SiO2, gradient 0ꢀ100 %
EtOAc in cyclohexane) to afford the title compound (2.15 g, 65 %). LC/MS: RT = 3.97 min, [M+Na]+ =
345.
ACS Paragon Plus Environment