Discovery of Brigatinib (AP26113), a Phosphine Oxide-Containing, Potent, Orally Active Inhibitor of Anaplastic Lymphoma Kinase

Article abstract of DOI:10.1021/acs.jmedchem.6b00306

In the treatment of echinoderm microtubule-associated protein-like 4 (EML4)-anaplastic lymphoma kinase positive (ALK+) non-small-cell lung cancer (NSCLC), secondary mutations within the ALK kinase domain have emerged as a major resistance mechanism to both first- and second-generation ALK inhibitors. This report describes the design and synthesis of a series of 2,4-diarylaminopyrimidine-based potent and selective ALK inhibitors culminating in identification of the investigational clinical candidate brigatinib. A unique structural feature of brigatinib is a phosphine oxide, an overlooked but novel hydrogen-bond acceptor that drives potency and selectivity in addition to favorable ADME properties. Brigatinib displayed low nanomolar IC₅₀s against native ALK and all tested clinically relevant ALK mutants in both enzyme-based biochemical and cell-based viability assays and demonstrated efficacy in multiple ALK+ xenografts in mice, including Karpas-299 (anaplastic large-cell lymphomas [ALCL]) and H3122 (NSCLC). Brigatinib represents the most clinically advanced phosphine oxide-containing drug candidate to date and is currently being evaluated in a global phase 2 registration trial.

Full text of DOI:10.1021/acs.jmedchem.6b00306

Subscriber access provided by SUNY DOWNSTATE
Article
Discovery of Brigatinib (AP26113), a Phosphine Oxide-Containing,
Potent, Orally Active Inhibitor of Anaplastic Lymphoma Kinase
Wei-Sheng Huang, Shuangying Liu, Dong Zou, R. Mathew Thomas, Yihan Wang, Tianjun Zhou, Jan Romero,
Anna Kohlmann, Feng Li, Jiwei Qi, Lisi Cai, Timothy A. Dwight, Yongjin Xu, Rongsong Xu, Rory Dodd,
Angela V Toms, Lois Parillon, Xiaohui Lu, Rana Anjum, Sen Zhang, Frank Wang, Jeffrey Keats, Scott D.
Wardwell, Yaoyu Ning, Qihong Xu, Lauren E. Moran, Qurish M. Mohemmad, Hyun Gyung Jang, Tim Clackson,
Narayana I. Narasimhan, Victor M. Rivera, Xiaotian Zhu, David C. Dalgarno, and William C. Shakespeare
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b00306 • Publication Date (Web): 04 May 2016
Downloaded from http://pubs.acs.org on May 4, 2016
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 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Discovery of Brigatinib (AP26113), a Phosphine Oxide-Containing, Potent, Orally Active Inhibitor of
Anaplastic Lymphoma Kinase
Wei-Sheng Huang,* Shuangying Liu, Dong Zou, Mathew Thomas, Yihan Wang, Tianjun Zhou, Jan
Romero, Anna Kohlmann, Feng Li, Jiwei Qi, Lisi Cai, Timothy A. Dwight, Yongjin Xu, Rongsong Xu, Rory
Dodd, Angela Toms, Lois Parillon , Xiaohui Lu, Rana Anjum, Sen Zhang, Frank Wang, Jeffrey Keats, Scott
D. Wardwell, Yaoyu Ning, Qihong Xu, Lauren E. Moran, Qurish K. Mohemmad, Hyun Gyung Jang, Tim
Clackson, Narayana I. Narasimhan, Victor M. Rivera, Xiaotian Zhu, David Dalgarno, William C.
Shakespeare
0
1
2
3
4
5
6
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
ARIAD Pharmaceuticals, Inc., 26 Landsdowne Street, Cambridge, MA 02139
1
ACS Paragon Plus Environment

Page 3 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ABSTRACT
In the treatment of EML4-ALK+ non-small-cell lung cancer (NSCLC), secondary mutations within the
anaplastic lymphoma kinase (ALK) kinase domain have emerged as a major resistance mechanism to
both first- and second-generation ALK inhibitors. This report describes the design and synthesis of a
series of 2,4-diarylaminopyrimidine-based potent and selective ALK inhibitors culminating in
identification of the investigational clinical candidate brigatinib. A unique structural feature of brigatinib
is a phosphine oxide, an overlooked but novel hydrogen-bond acceptor that drives potency and
selectivity in addition to favorable ADME properties. Brigatinib displayed low nM IC s against native ALK
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
50
and all tested clinically relevant ALK mutants in both enzyme-based biochemical and cell-based viability
assays, and demonstrated efficacy in multiple ALK+ xenografts in mice, including Karpas-299 (anaplastic
large-cell lymphomas [ALCL]) and H3122 (NSCLC). Brigatinib represents the most clinically advanced
phosphine oxide-containing drug candidate to date, and is currently being evaluated in a global phase 2
registration trial.
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
INTRODUCTION
Anaplastic lymphoma kinase (ALK), a receptor tyrosine kinase of the insulin receptor family, was
originally identified as a part of the fusion protein NPM-ALK resulting from a reciprocal translocation
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
between chromosomes 2 and 5 in anaplastic large-cell lymphomas (ALCL). NMP-ALK has been detected
2
in approximately 75% of ALK+ ALCL patients and its constitutive activation drives disease pathogenesis.
Subsequently, alternate ALK fusion genes were discovered as drivers of inflammatory myofibroblastic
3, 4
tumors and diffuse large B-cell lymphoma. Despite these early findings, the significance of ALK as a
valid oncogenic drug target was not fully recognized until 2007, when the EML4-ALK fusion protein was
5
identified in approximately 5% of non-small-cell lung cancer (NSCLC). Given the large NSCLC patient
population, new cases of ALK+ lung cancer are estimated to exceed 8000 per year in the US,
representing an important unmet medical need. Interestingly, patient populations harboring the EML4-
ALK gene rarely overlap with those carrying epidermal growth factor receptor (EGFR) or kirsten rat
6
sarcoma virus (KRAS) mutations, making EML4-ALK a distinct molecular target for a subset of NSCLC
patients. More recently, amplification or activating point mutations of the ALK gene have been reported
7
8
9
in neuroblastoma, anaplastic thyroid cancer, and ovarian cancer.
1
0, 11
The first-generation inhibitor crizotinib
was approved for clinical use in 2011 and has demonstrated
clear clinical benefit, with responses observed in 60–74% of ALK+ NSCLC patients (median progression-
12, 13
free survival of 8–11 months).
Despite early effectiveness of crizotinib treatment, a majority of
patients develop disease relapse within 1 year post-treatment. Various resistance mechanisms have
been identified, including amplification of the ALK fusion gene, activation of bypass signaling pathways,
inadequate exposure in the central nervous system, and as observed with other targeted therapies,
14, 15
emergence of secondary ALK kinase domain mutations.
Roughly one-third of crizotinib-resistant
patients develop secondary mutations, and ~40% of ALK-positive patients never have an objective
16
response while on crizotinib therapy. Recently, several more potent second-generation inhibitors,
3
ACS Paragon Plus Environment

Page 5 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
7, 18
19, 20
including ceritinib
and alectinib,
have been identified to combat disease resistance. Although
these second-generation inhibitors have demonstrated efficacy against the majority of crizotinib-
resistant mutants, they remain inactive against several mutations, including I1171N/T/S for alectinib,
18, 21, 22
F1174C/V for ceritinib, and G1202R for both agents.
Development of next generation ALK
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
inhibitors capable of potently inhibiting all acquired ALK mutants is therefore highly desirable.
23
Continued search has advanced several compounds to clinical trials including 1 (ASP3026), 2
24
25
(
(
ensartinib, X-396), CEP-37440 (structure undisclosed, series exemplified as compound 3), TSR-011
26
27-29
30, 31
structure undisclosed), 4 (lorlatinib, PF-06463922),
5 (entrectinib, RXDX-101, NMS-E628),
and
11q (brigatinib) (Figure 1). In addition, a number of ALK inhibitors derived from a variety of diversified
32-39
chemical classes have been evaluated at pre-clinical stages.
Figure 1. Structures of select ALK inhibitors
4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Phosphorus, despite its abundance in the human body, is rarely found in drug molecules, with its clinical
utility limited to a few phosphonic or bisphosponic acid-based medicines, acting as phosphate mimics,
40
as well as several phosphonate, phosphinate or phosphate-containing prodrugs. Concerns about poor
cell penetration, low oral bioavailability, or biological instability have limited broader application of
these functional classes in drug design. In our efforts to discover pharmaceuticals with novel
functionality, we have previously explored the potential wider utilization of phosphorus in drug design.
For example, dimethylphosphorylation at the C43-OH of rapamycin led to discovery of ridaforolimus, a
chemically stable mTOR inhibitor that advanced to late-stage clinical trials in oncology, and is currently
0
1
2
3
4
5
6
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
41
being investigated as a component of drug-eluting stents. Introduction of a bone-targeting
arylphosphinylmethylphosphonic acid (PMP) functionality onto multiple templates, generated several
42
series of SRC kinase inhibitors as potential anti-resorptive agents. More recently, we have explored
whether the P=O bond can be exploited as a hydrogen-bond acceptor in kinase inhibitor design. In this
regard, incorporation of a dimethylphosphine oxide (DMPO) moiety in 6 (AP23464) (Figure 2) led to a
43, 44
70-fold increase in SRC potency (IC = 0.45 nM) relative to the unsubstituted analog.
To further
50
explore the potential of this functionality, we have incorporated the DMPO in our design of ALK
inhibitors leading to the discovery of the potent, orally active ALK inhibitor, brigatinib. Detailed
biological characterization of brigatinib as an inhibitor of native ALK and clinically relevant ALK mutants
45
will be reported in a separate publication. Herein we present our medicinal chemistry work leading to
46
the discovery of brigatinib.
5
ACS Paragon Plus Environment

Page 7 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 2. Chemical structures of SRC inhibitor 6 and ALK inhibitors 10a-g
RESULTS AND DISCUSSION
Early ALK Inhibitor Design and Exploration
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
Prior to launching an ALK inhibitor program, inhibitors based on an array of different structural
47
48
scaffolds, including crizotinib (then known as PF-2341066) and 7 (TAE684, Figure 1), had been
reported. Further optimization of 7 to address selectivity and potential idiosyncratic toxicity issues led to
17
the eventual discovery of ceritinib a few years later.
Initially, we explored several parallel paths with the goal to identify ALK inhibitors with multiple distinct
chemotypes. Among others, one early chemotype was based on morphing the crizotinib scaffold into a
49
bicyclic tetrahydropyrazino[2,3,b]-pyrazine template. An alternative, parallel approach focused on
constructing and testing a library of substituted 2-anilinopyrimidine compounds. This design strategy
stemmed from our observation that numerous kinase inhibitors evaluated clinically or at preclinical
stages were derived from this scaffold. We anticipated that introduction of diversified functional groups
at C2, C4, and C5 on the pyrimidine core might favorably alter kinase selectivity of 2-anilinopyrimidine-
based ALK inhibitors. Particularly, we monitored new inhibitor activity toward the insulin receptor kinase
(
InsR) and insulin-like growth factor receptor (IGF1R), as both kinases bear high sequence homology
with ALK and are important for maintaining glucose homeostasis.
To incorporate the DMPO onto the 2-anilinopyrimidine template, our initial efforts, by analogy to 6,
targeted the 4’ position of the 2’-methoxyaniline group at C2 adjacent to the hinge region of the kinase.
Incorporation of the 2’-methoxy group was intended to prevent general kinase promiscuity through
interaction with a small hydrophobic pocket adjacent to the hinge that is unique to certain protein
50
kinases including ALK protein. Gratifyingly, DMPO incorporation at C4’ was found to increase ALK
activity by approximately 7-fold relative to the unsubstituted analog (10a and 10g, Table 1) when
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
investigated in the context of a small focused library at C4. Several C4 ortho-substituted aniline analogs
10a-e exhibited potent ALK inhibition with IC s in the sub-nanomolar range in biochemical assays. Most
50
compounds also effectively inhibited viability of the Karpas-299 ALCL cell line with an IC range of 20–
50
7
40 nM, and with ~4–240-fold decreased activity against the ALK-negative U937 histolytic lymphoma
0
1
2
3
4
5
6
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
cell line. In this series, the isopropylsulfonyl analog 10a, despite displaying the most favorable cellular
potency, did not confer desirable selectivity over IGF1R.
Table 1. Activity Profiles of Inhibitors 10a-g (IC , nM)
50
Enzyme-based kinase activity
Viability of cell lines
(IC , nM)
(
IC , nM)
50
50
Karpas-299
U937
(ALK-)
2286
R
ALK
IGF1R
143
InsR
201
(
ALK+)
365
crizotinib
0.64
1
1
1
1
0a
0b
0c
0d
SO -CHMe2
0.31
0.96
0.79
0.68
0.33
1.06
2.1
1.9
22.1
0.47
0.11
0.40
139
n/d
9.9
55.2
0.88
14.1
4.7
20
4838
2908
1423
2368
4813
13193
20824
2
OMe
278
251
613
405
738
349
C(=O)-Me
C(=O)-NH2
10e
SO -Me
2
1
1
0f
P(=O)Me2
n/a
726
n/d
0g
n/a, not applicable; n/d, not determined
In addition to exploring DMPO substitution on the aniline near the kinase region, homology-based
molecular modeling (Figure 3) revealed that ortho-substituents on the C4 anilines of 10a-e can
potentially form an intramolecular hydrogen-bond to the aniline NH or hydrogen-bond with the Lys-NH
(
K1150) on the protein. This prompted us to explore the H-bonding potential of the DMPO functionality
beyond that noted near the hinge; therefore, a double-phosphorylated analog 10f was synthesized and
7
ACS Paragon Plus Environment

Page 9 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
tested. Although 10f did not exhibit increased ALK activity relative to 10a-e, its improved selectivity over
both IGF1R and InsR was remarkable and unanticipated. Given this finding, to further explore SAR, we
retained the ortho-DMPO-substituted aniline at C4 to maintain selectivity while optimizing C2 and C5
substitutions to improve both potency and the overall drug metabolism and pharmacokinetics (DMPK)
profile.
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. Homology model of 10f bound to ALK. Dotted lines indicate hydrogen bond interactions.
Development of Structure–Activity Relationship (SAR)
Given the hydrophilic nature of the bis-DMPO inhibitor 10f (cLogP = 0.4, ChemBioDraw Ultra 14.0), we
examined how substitution of the C2-DMPO with basic amine side chains might influence ALK activity,
particularly the cellular potency. Based on the para-substitution vector directed at the solvent front, a
large focused library of diverse basic amine-containing solubilizing groups was designed and
synthesized, including N-, O-, and C-linked derivatives. As shown in Table 2, these new analogs were
potent ALK inhibitors and their activities were nearly indistinguishable in biochemical assays because of
their high potency under assay conditions (1 h and 10 µM ATP). Fortunately, potency separation was
more clearly delineated in cell-based viability assays following inhibitor treatment for 72 h. The
discrepancy between enzymatic and cellular potencies may be explained by the higher ATP
8
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
concentrations in the physiological cellular system and/or the structural difference between truncated
enzyme and full-length receptor. Interestingly, this potency mismatch was also observed on structurally
17,48
similar inhibitors such as ceritinib and 7.
In our chemical series, compounds bearing a piperazine
(
11e, 11i) or piperidine ring (11l, 11m, 11n, 11q) directly attached to C2 phenyl displayed excellent cell
0
1
2
3
4
5
6
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
potency. A basic nitrogen atom appeared critical for driving cell-based activity. Substitution of the
terminal piperazinyl nitrogen in 11e by oxygen or carbon atoms (11d, 11g), removal or attenuation of
nitrogen basicity (11f, 11h), all led to reduced ALK potency. 7-membered homopiperazine or 5-
membered pyrrolidine analogs were less active than corresponding 6-membered inhibitors (11j vs. 11e,
11k vs. 11l, 11s vs. 11n) despite possessing equivalently basic nitrogens, highlighting the importance of
ring size. The basic nitrogen can be incorporated into a second ring; however, its basicity (11p vs 11q)
and size (11q vs 11r) influenced activity similar to that observed with the mono-ring analogs (11d, 11e,
11j, 11l). Installation of the solubilizing group identical to crizotinib (11t) resulted in significant potency
loss, presumably due to suboptimal location of the basic nitrogen atom and/or unfavorable aromatic
ring attachment and vectors. An alternate C-linked basic amine-containing inhibitor (11u), designed for
the purpose of mitigating potential idiosyncratic bioactivation associated with the 1,4-diaminophenyl
motif, demonstrated equipotent ALK activity relative to its N-linked counterpart (11e). Relative to other
inhibitors, compounds 11i and 11u exhibited slightly compromised selectivity over InsR. The moderately
increased activity toward InsR could be rationalized by interaction of the hydroxylethyl group in 11i with
Ser1063 of helix D or interaction of the stronger basic amine in 11u with the negatively charged patch in
the ATP ribose binding pocket of InsR. Overall, this solubilizing group screening campaign identified
several amines (11e, 11m, 11n, 11q, 11u) with favorable potency and selectivity profiles.
9
ACS Paragon Plus Environment

Page 11 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 2. Activity Profiles of Inhibitors 11a-u Bearing a Variety of Basic Amines (IC , nM)
50
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
Enzyme-based kinase activity
(IC , nM)
Viability of cell lines
(IC , nM)
50
50
Karpas-299
U937
R
ALK
0.64
IGF1R
143
InsR
201
(
ALK+)
365
(ALK-)
2286
crizotinib
11a
0.22
17.4
284
113
2307
11b
11c
0.33
0.54
100
397
343
65.1
55.5
6334
4523
86.5
1
1d
0.24
0.18
51.4
399
77.4
5994
11e
20.3
32.5
90.6
103
28.3
79.9
1590
5453
0
.59
1
1
1f
1g
0.36
3.1
26.9
125
171
459
28.1
70.4
270
7232
7004
1798
1
1h
11i
11j
0.18
12.8
37.6
0.72
77.5
339
53.9
2968
1
1
1k
1l
0.33
0.07
42.7
3.2
489
100
66.9
41.5
1174
1718
1
1m
0.88
31.6
130
29.9
2725
1
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
11n
0.53
0.28
20.3
43.3
85.3
285
21.1
101
1678
4920
N
1
1o
N
N
0
1
2
3
4
5
6
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
11p
0.49
0.37
0.49
47.3
24.9
38.8
270
196
162
61
29
5682
3194
3389
1
1
1q
1r
105
1
1
1s
1t
0.88
0.76
81.5
98.1
270
702
66.7
322
2390
6226
N
N
HN
1
1u
0.38
8.2
57.2
19.2
3557
To further lead optimization, we next chose to maintain the C2 amine group of compound 11q while
exploring the SAR at other positions on the inhibitor core (Table 3). A chloro substituent at C5 on the
1
pyrimidine ring appears optimal; analogs with a variety of R groups at C5 (12a-d) exhibited reduced
cellular viability despite having similar biochemical enzymatic activity. An acetyl substituent at C5 (12e)
largely ablated activity in the biochemical and viability assays. Increasing the alkyl group size on
phosphorus from methyl to ethyl (12f) led to 5-fold potency loss in Karpas-299 cells. Similarly, fluoro
group introduction para to the methoxy substituent on the 2-substituted aniline (12g) furnished a less
active inhibitor. Interestingly, analog 12h bearing a methyl group at the same position was equipotent to
1
1q in cellular settings, despite exhibiting moderate enzymatic potency loss. Finally, substitution of the
methoxy group by either ethoxy (12i) or isopropoxy analogs (12j) demonstrated enhanced selectivity
over both IGF1R and InsR; however, both inhibitors exhibited reduced (~50%) ALK potency.
1
1
ACS Paragon Plus Environment

Page 13 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 3. SAR at C5 of Pyrimidine Core and Phenyl rings at C2 and C4 (IC , nM)
50
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
Enzyme-based kinase
activity (IC , nM)
Viability of cell lines
(IC , nM)
50
50
Karpas 299 U937
1
2
3
4
R
R
R
R
ALK
IGF1R
143
24.9
434
117
143
128
NT
InsR
(
ALK+)
(ALK-)
crizotinib
Me
0.64
0.37
3.23
1.19
0.39
0.42
291
201
196
367
1684
1683
483
NT
365
29
2286
1
1q
12a
2b
2c
Cl
c-Pr
OMe
F
H
H
Me
Me
Me
Me
Me
Me
Me
Me
Me
Et
3194
10630
13863
24166
5307
Me
278
429
147
46.9
NT
1
Me
H
1
Me
H
1
2d
2e
Me
C(=O)Me
Cl
Me
H
a
1
Me
H
NT
1
2f
2g
12h
Et
H
0.12
1.03
0.86
0.75
0.68
45.1
59.5
27
314
1179
171
253
700
169
162
3465
1231
1
Cl
Me
F
b
b
Cl
Me
Me
H
16
1282
1
2i
2j
Cl
Me
103
203
58.9
54.6
22398
21475
1
Cl
Me
H
i-Pr
a
b
NT = not tested. GI s (concentrations required to inhibit growth by 50%); for comparison, 11q had GI
of 10 and 3194 nM towards Karpas 299 and U937, respectively.
50
50
1
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
To confirm and extend our observations regarding the critical role of the C4 DMPO moiety on both
potency and selectivity (Table 1), analogs of 11q where the DMPO was systematically modified (mostly
with H-bond acceptors) were synthesized and tested (Table 4). While compounds 13c and 13i conferred
reasonable selectivity over IGF1R and InsR kinases, both inhibitors were 5- or 3-fold less active towards
ALK relative to 11q. Interestingly, acetyl inhibitor 13h, while potent against ALK, was completely non-
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
selective over IGFR and InsR relative to its reduced analog 13i, suggesting that sp hybridization at this
position may be differentiating these kinases. Compound 7, a 2-fold more potent inhibitor than 11q,
displayed a narrow selectivity window, and its sulfonamide analog 13g was also poorly selective and
exhibited 5-fold reduced ALK activity. To assess whether the bulkier isopropyl group of 7 was leading to
poor selectivity, analog 13j bearing a methylsulfone group, more closely resembling the DMPO in size,
was tested. 13j demonstrated similar selectivity but reduced activity relative to 7, indicating no
correlation between alkyl substituent size and kinase selectivity. Taken together, independent of the C2
substituents directed at the kinase solvent front, the C4 DMPO functionality was an essential structural
feature necessary to achieve the proper balance between potency and selectivity with this inhibitor
core.
1
3
ACS Paragon Plus Environment

Page 15 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 4 SAR of DMPO Replacements at C4 (IC , nM)
50
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
R
ALK kinase
0.64
IGF-1R kinase
143
InsR kinase
201
Karpas 299 cell
crizotinib
365
29
1
1q
3a
0.37
24.9
196
a
1
H
I
9.9
1.36
1.86
1.88
0.8
51.6
6.2
81.7
13.7
128
26
n/d
13b
3c
3d
13e
3f
705
n/d
851
n/d
1
CN
OMe
51.6
8.7
1
1.4
3.4
1
4.77
0.75
6.17
1.93
6.05
8.44
n/d
n/d
13g
13h
1.59
0.99
0.59
1.33
36.1
0.76
2.16
43.2
5.1
n/d
n/d
n/d
1
1
3i
3j
7
0.15
0.35
1.9
14
a
n/d, not determined.
1
4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Structural Analysis of Brigatinib Binding and Selectivity
Elsewhere, we have reported detailed structural characteristics of the ALK-brigatinib complex (PDB:
4
5
5
J7H). Briefly, the structure shows that brigatinib binds to ALK in the ATP-binding site by adopting a U-
0
1
2
3
4
5
6
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
shaped ligand conformation. While the pyrimidine core of brigatinib binds in the adenosine site, its
methoxy group and DMPO aniline bind in the vicinity of the hinge and DFG sites, respectively (Figure
4A). As revealed in the structure, the potency of brigatinib against ALK can be largely attributed to the
chemical substituents installed on the bisanilinopyrimidine scaffold. These include the molecular
interactions of the methoxy group with the hinge residue L1198, of the C5- chlorine atom with the
gatekeeper residue L1196 and of the DMPO moiety with the GDFG motif (Figure 4A). In addition, the
DMPO moiety fosters intra-molecular interactions and stabilizes the U-shaped conformation of
brigatinib through an O···NH hydrogen bond (Figure 4A). Interestingly, the DMPO aniline in brigatinib is
rotated significantly towards the lower lobe of the ALK when compared with compound 7 (in which the
DMPO is replaced by isopropylsulfone) (Figure 4B). As a result, the DMPO aniline stays further away
from the P-loop relative to the isopropylsulphone aniline in compound 7, which may have led brigatinib
to rely less on P-loop interactions for potency. The observed conformational differences between
brigatinib and compound 7 in this region is likely driven by distinct binding interactions between the
inhibitors and ALK. Apart from the phenyl groups that both bind underneath the P-loop, one of the
methyl groups from the DMPO or the isopropylsulfone occupies a very similar position such that each
methyl makes strong hydrophobic interaction with the side chain of L1256 (from the adenine binding
pocket) (Figure 4B). The solubilizing group, particularly the methyl piperazine, is largely disordered likely
due to lack of strong binding interactions. Of interest, the catalytic lysine (K1150) is also partially
disordered despite being proximal to the DMPO moiety. No co-crystal structures of brigatinib or 7 bound
to insulin kinase are available; however, by analogy, brigatinib may also interact with this part of insulin
kinase differently than the other analogs indicated in Table 4, resulting in the different selectivity profile.
1
5
ACS Paragon Plus Environment

Page 17 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A
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
Figure 4. (A) X-ray crystal structure of ALK in complex with investigational brigatinib. Dotted lines
indicate hydrogen bond interactions. (B) Comparison of brigatinib (pink) and compound 7 (orange)
bound to wild type ALK. Certain residues in the brigatinib-ALK complex are shown in gray; residue L1256
(
green) in the 7-ALK complex is also displayed.
Pharmacokinetic (PK) and Pharmacodynamic (PD) Properties of Selected Inhibitors
During this program, potent ALK inhibitors were evaluated for their PK properties in rats and mice
immediately following assessment of their in vitro activity. A number of inhibitors with diverse
solubilizing groups on the C2 aniline exhibited high levels of oral absorption in both species (Tables 5
4
and 6). Upon dosing intravenously (iv) in rats, N -(2-hydroxyethyl)piperazine inhibitor 11i exhibited
1
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
faster clearance than N -methyl piperazine analog 11e. Similarly, pyrrolidinylpiperidine analog 11n
showed higher clearance relative to the piperazinylpiperdine analog 11q. Both 11i and 11n displayed
lower oral exposures suggesting relatively poor bioavailability. Interestingly, the more basic piperidine
analog 11u, despite displaying higher iv clearance than piperazine inhibitor 11e, had higher systemic
exposure when dosed orally in both rats and mice. To the contrary, the O-linked inhibitor 11c, despite
exhibiting low iv clearance and excellent oral exposure in rats, possessed lower oral exposure in mice
relative to other analogs.
0
1
2
3
4
5
6
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. Mean PK Parameters Following Intravenous (iv) or Oral (po) Dosing in CD Rats
b
c
iv
po
AUC0-∞
ng·h/mL)
CL
t1/2
(h)
Vss
(L/Kg)
Cmax
(ng/mL)
AUC0-∞
(ng·h/mL)
tmax
(h)
t1/2
(h)
F
(%)
(
(
L/h/kg)
1
1e
11i
1g
1n
11c
3092
1326
4546
2170
5035
17889
1550
8619
2268
0.91
2.15
0.64
1.36
0.59
0.17
1.91
0.35
6.6
8.4
5.6
4.7
3.6
3.9
4.0
3.6
4.8
6.12
9.51
2.74
6.61
2.89
0.78
8.55
1.46
254.0
4056
6.0
12.5 39
75.7
777
4.0
2.9
18
64
19
55
73
110
16
52
1
772.3
109.0
658.7
2836.7
601.3
353.0
305.0
9725
1353
9221
43643
5670
4572
3954
0.25 2.7
1
4.0
4.0
2.0
4.0
6.0
4.0
3.7
3.9
6.8
2.5
3.4
3.4
1
1
1p
1u
1
1m
1q
1
0.46
7.80
a
b
c
n = 3 animals per study. Dosed at 3 mg/kg. Dosed at 10 mg/kg.
1
7
ACS Paragon Plus Environment

Page 19 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a
Table 6. Mouse PK Parameters Following Oral Dosing
Cmax
AUC0-24h
t1/2
(
ng/mL)
(ng.h/mL)
(h)
1
1e
1g
418
2943
411
184
421
498
448
5184
13262
3231
951
2.7
2.2
3.1
1.6
5.0
3.1
5.8
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
1
1n
1c
1u
1m
11q
1
1
3451
6494
5772
1
a
3
animals for each of 7 time points;
dose level: 10 mg/kg
Several potent inhibitors (11g, 11q, 11c, 11p, 11u, 11e, 11m, and 11n) bearing diverse solubilizing
groups and exhibiting desirable PK profiles were selected for a single oral (po) dose (25 mg/kg) PK/PD
study in a mouse Karpas-299 model. Although most inhibitors demonstrated strong inhibition of ALK
phosphorylation (p-ALK) for 10 h (Table 7), several (e.g., 11e and 11q) exhibited more sustained signal
suppression (up to 16 h), with 11q (brigatinib) offering the most durable inhibition (p-ALK = 7 and 39% at
1
6 and 24 h, respectively). These data suggested that once-daily po dosing would likely be appropriate
for efficacy studies. Overall, the PK/PD experiments were found to be highly predictive of in vivo efficacy
(vide infra).
1
8
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Table 7. Single-dose (25 mg/kg, po) PD: Percentage of p-ALK at 10, 16 and 24 h
1
1c
11e
11g
11m
11n
11p
11q
11u
b
10 h
23
2
41
13
2
15
1
1
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
6 h
4 h
125
160
14
94
25
60
39
51
52
50
7
69
168
104
39
118
a
Data presented are percent ALK phosphorylation (p-ALK) relative to the p-ALK levels in vehicle-treated
tumors at 2 h post dosing; all p-ALK levels were normalized to the corresponding total ALK levels. 8 h
value.
b
In vivo Efficacy of Selected Inhibitors
Potent inhibitors possessing favorable PK profiles and displaying a strong correlation between PK and
inhibition of ALK phosphorylation were evaluated for their anti-tumor activity in a Karpas-299 xenograft
mouse model expressing the NPM-ALK fusion. Animals were treated orally with inhibitors at selected
doses, once daily for 13 consecutive days. Tumor growth inhibition of treated (T) mice relative to
untreated control (C) was used to assess efficacy and was reported as % T/C in Table 8. Compound 11g
was essentially inactive up to 50 mg/kg, whereas 11n displayed limited but dose-responsive efficacy.
Inhibitors 11e and 11q demonstrated dose-dependent anti-tumor activity, with the latter being more
efficacious at each dose level. While a 50 mg/kg dose of 11e led to tumor stasis, an equivalent dose of
1
1q resulted in tumor regression with no signs of overt toxicity.
a
Table 8. In Vivo Anti-tumor Activity in a Karpas-299 Mouse Model (% T/C)
1
1e
11g
11n
11q
b
b
c
1
2
0 mg/kg
5 mg/kg
81
41
1.4
96
76
50
b
85
52
26
12
b
50 mg/kg
67
-11
a
Unless otherwise indicated, p <0.01. Negative value indicated tumor regression, all other values
b
c
represented tumor growth inhibition. p >0.05. p <0.05.
1
9
ACS Paragon Plus Environment

Page 21 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Next, we evaluated the activity of 11q in a mouse model of NSCLC (H3122 cells expressing the EML4-ALK
fusion). In this model, once-daily oral administration (for 19 days) of 11q resulted in dose-dependent
antitumor activity, with tumor regressions observed at all doses tested (including the lowest 10 mg/kg
dose) (Figure 5). By comparison, animals treated daily with 100 mg/kg crizotinib exhibited 97% tumor
growth inhibition (data not shown). Consistent with the favorable safety profile demonstrated in the
Karpas-299 model, no clinical signs of gross toxicity or body weight loss were observed at any brigatinib
dose level tested, relative to vehicle control-treated animals.
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
1000.00
vehicle
1
2
5
0 mg/kg
5 mg/kg
0 mg/kg
100.00
10.00
0
5
10 15 20 25
Days post initial treatment
Figure 5. In vivo efficacy of 11q (brigatinib) in a H3122 NSCLC model
Kinase Selectivity and Mutant Profile
To assess the kinase selectivity profile of 11q (brigatinib) beyond ALK, IGF1R, and InsR, it was screened
45
against a broad panel of 289 kinases and mutants/isoforms, and the results were reported elsewhere.
In brief, ~4% (11 out of 289) of the kinases in this panel were inhibited with IC s of <10 nM, indicating a
50
relatively high degree of selectivity. Among them, brigatinib potently inhibited FLT3 and ROS1 with IC50
values of 2.1 and 1.9 nM, respectively. Unlike crizotinib, which is characterized as a dual c-Met/ALK
inhibitor that also potently inhibits Ron, 11q did not show significant activity toward c-Met or Ron up to
1
µM.
2
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Kinase domain mutations have been identified as one of the major mechanisms for crizotinib resistance.
Elsewhere, we have reported the cellular activity of brigatinib against a broad panel of clinically relevant
45
EML4-ALK mutants. Briefly, brigatinib was active against 17 clinically and preclinically identified
resistant mutants; these include F1174C/V, I1171N, and G120R mutations, which have been reported in
patients failing crizotinib or the second-generation inhibitors (ceritinib and alectinib). Additionally, in an
accelerated cell-based ENU mutagenesis screen, brigatinib completely suppressed the outgrowth of all
mutants at a concentration of 500 nM indicating its potential as a pan-ALK inhibitor.
0
1
2
3
4
5
6
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
DMPO and its Impact on the Properties of Brigatinib
Brigatinib represents the most clinically advanced phosphine oxide-containing molecule currently in
development. The reason why this functionality has been overlooked by medicinal chemists is unclear; it
is easily synthesized using conventional methods and largely inert once installed. Unlike carbonyls and
alcohols, the PO bond does not undergo redox chemistry and its ability to act as H-bond acceptor makes
it an important complement to existing more conventional H-bond acceptors such as carbonyls and
sulphones. Moreover, phosphine oxides are tetrahedral and there are very few H-bond acceptors that
allow for three points of chemical derivatization. In a study to determine hydrogen-bond basicity, the
equilibrium constants for a series of bases to form 1:1 complexes with 4-fluorophenol were measured
51
by infrared spectroscopy. Trimethylphosphine oxide was found to possess a pKBHX of 3.53, which
indicates its oxygen should be a strong hydrogen-bond acceptor. By comparison, dimethylsulfoxide, N,N-
dimethylacetamide, N,N-dimethylmethanesulfonamide, acetone, and dimethylsulfone have pKBHXs of
2
.54, 2.44, 1.3, 1.2, and 1.1, respectively. Interestingly, brigatinib analogs incorporating some of these
weaker hydrogen-bond acceptors were all less potent against ALK and also displayed narrower
selectivity over IGF1R and InsR (Table 4).
2
1
ACS Paragon Plus Environment

Page 23 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
5
2, 53
Experimental ESR and NMR data,
ab initio NMR calculation and atoms-in-molecules (AIM) study, all
suggest that the PO bond in a phosphine oxide is highly ionic and that its Lewis structure is better
+
-
54-56
represented as R P -O in a manner directly analogous to N-oxides.
This dative, and therefore
3
polarizable, PO bond imparts a number of important drug-like properties to the DMPO-containing
molecules. For example, the increased polarity relative to the isopropyl sulfone derivative 7, has a direct
effect on lipophilicity where the measured LogD (pH 7.0) was reduced by more than a log unit relative to
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
7
(3.02 vs. 4.12). Since lipophilicity and protein binding are positively correlated, it was notable that 11q
is moderately protein bound, with plasma binding being 66, 73, and 64% in human, rat, and mouse,
respectively. In contrast, 7 is highly protein bound, exhibiting binding levels of 99% or higher in all three
species. Finally, 11q demonstrated a marked increase in aqueous solubility relative to 7 at physiological
pH (337 vs 14.6 µg/mL) such that the compound can be formulated as the free base for human clinical
evaluation.
Chemical Synthesis
The inhibitors reported herein were synthesized via a general method in which both chlorines of 2,4-
dichloropyrimidine core were sequentially displaced by a variety of substituted anilines (Scheme 1).
While the first chlorine displacement at C4 was a typical base-mediated S Ar process, subsequent
N
substitution at C2 was realized via an acid-promoted S Ar or a Pd-catalyzed N-arylation method.
N
Syntheses of the requisite substituted aniline intermediates were often key to making these ALK
inhibitors.
Scheme 1
2
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
o
Reagents and conditions: i) K CO , NaH, or LiHMDS, DMF, 65 C; ii) for 11t and 11u, Pd(OAc) , Xantphos,
2
3
2
o
Cs CO , DMF, 110 C; for other inhibitors, 2.5 M HCl in EtOH, dioxane, 2-methoxyethanol, or 2-butanol
2
3
The syntheses of certain requisite C2 anilines are depicted in Scheme 2. Aniline 15 is the common
precursor for inhibitors 10a-f compiled in Table 1; it can be prepared from 2-nitro-5-chloroanisole and
57
dimethylphosphine oxide via Pd-catalyzed C-P bond formation followed by reduction of the nitro
group under hydrogenation. The synthesis of aniline 16 represents a general protocol for making all C2
anilines required for compounds 11d-s. S Ar fluoro displacement of 2-nitro-5-fluoroanisole by secondary
N
amines with K CO as the base, followed by nitro reduction, routinely furnished anilines in high yields.
2
3
The required O-linked anilines for compounds 11a-c can be accessed similarly except that NaH was
employed as the base. C-Linked aniline 17 was obtained by Suzuki coupling of 4-iodo-2-methoxyaniline
with corresponding requisite boronate. Finally, C-linked aniline 18 was prepared by a similar Suzuki
coupling followed by hydrogenation of azacyclohexenyl derivative and exhaustive reduction of Boc
protecting group by lithium aluminum hydride.
2
3
ACS Paragon Plus Environment

Page 25 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
o
Reagents and conditions: i) Pd(OAc) , xantphos, K PO , DMF, 120 C; ii) Pd on carbon, 30 psi H , EtOH,
2
3
4
2
o
o
r.t.; iii) K CO , MeCN, 80 C; iv) 2.5 M HCl in EtOH, 2-methoxyethanol, 120 C; v) Pd(OAc) , PPh , K CO ,
2
3
2
3
2
3
DMF / H O, reflux; vi) lithium aluminum hydride, THF, reflux.
2
Many required substituted anilines at C4 were commercially available with the exception that
phosphorus-containing aniline 19 was made directly from iodoaniline and dimethylphosphine oxide via
Pd-catalyzed P-arylation method (Scheme 3). In this C-P bond-forming reaction, no competing N-
arylation was observed. It represents the first example of specific P-arylation of aryl halides with dialkyl
or diaryl- phosphine oxide in the presence of reactive anilino NH despite similar P-arylation with more
2
58, 59
nucleophilic dialkylphosphonate (also termed as dialkoxy phosphine oxide) being known.
SNAr
reaction of 19 with trichloropyrimidine under basic conditions provided a common intermediate 20,
2
4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
which underwent subsequent acid-promoted chlorine displacement with a range of amines like 16 to
afford inhibitors summarized in Table 2. Similarly, reaction of 20 and 2-methoxy-4-bromo-5-fluoroaniline
generated intermediate 21; subsequent Cu-catalyzed N-arylation of the bromide with 1-methyl-4-(4-
piperidinyl)piperazine furnished analog 12g.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 3
o
Reagents and conditions: i) dimethylphosphine oxide, Pd(OAc) , xantphos, K PO , DMF, 120 C; ii) 2,4,5-
2
3
4
o
o
trichloropyrimidine, K CO , n-Bu NHSO , DMF, 65 C; iii) 2.5 M HCl in EtOH, 2-methoxyethanol, 120 C;
2
3
4
4
o
iv) 1-methyl-4-(piperidin-4-yl)piperazine, CuI, L-proline, K PO , DMSO, 100 C; v) trimethylsilylacetylene,
3
4
o
Pd(PPh ) , CuI, Et N, DMF, 50 C.
3
4
3
Most inhibitors in Tables 3 and 4 were synthesized according to these general procedures except
compounds 12e and 13i. For 12e, Sonagashira coupling of bromide 22 with (trimethylsilyl)acetylene
yielded intermediate 23. Installation of the aniline at C2 under acid conditions concomitantly hydrolyzed
the alkyne moiety generating the C5 acetyl compound 12e (Scheme 3). Reduction of ketone analog 13h
with NaBH provided inhibitor 13i.
4
2
5
ACS Paragon Plus Environment

Page 27 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CONCLUSIONS
Starting from our interest to develop next-generation ALK inhibitors and motivated to further explore
stable phosphorus functionality in drug design, we synthesized and tested a library of phosphine oxide-
containing compounds based on the 2-aminopyrimidine scaffold. Extensive medicinal chemistry efforts
led to the identification of investigational brigatinib, a potent inhibitor of native ALK and a broad panel
of clinically relevant crizotinib-resistant mutations. Brigatinib demonstrated oral efficacy in multiple
ALK+ mouse models, including Karpas-299 (ALCL) and H3122 (NSCLC), while demonstrating no signs of
toxicity, including weight loss. A unique structural feature of brigatinib is the dimethylphosphine oxide, a
novel hydrogen bond acceptor that was critical for balancing ALK potency and selectivity over the highly
homologous IGF1R and InsR kinases. When compared with other hydrogen-bond acceptors, the highly
ionic P=O bond imparts a number of important drug-like properties, including decreased lipophilicity,
increased aqueous solubility, reduced protein binding, and robust metabolic stability. Phosphine oxides
belong to a chemical class seldom employed in drug design; the discovery of brigatinib and its clinical
activity may further inspire application of this unique functional group more broadly in the field of
medicinal chemistry.
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
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
EXPERIMENTAL SECTION
1
General Chemistry. All reagents and solvents were used as received. H NMR spectra were recorded on
a Bruker AVANCE400 spectrometer using TMS as internal standard. MS spectra were recorded on a
Waters Micromass ZQ spectrometer. Elemental analyses were performed by Robertson Microlit
Laboratories in Madison, NJ. HPLC was performed on an Agilent 1100 or an Agilent 1200 HPLC system.
The purity of all SAR compounds was determined to be >95% by reverse phase HPLC (C-18 column,
MeCN/H O with 0.1% CF CO H as the mobile phase). Certain compounds, such as 12a, 12b, 12f and 13d,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
2
were made in a focused library synthesis via standard general procedures and passed purity and mass
check by LC-MS; these inhibitors were not described in the experimental section.
(
3-Methoxy-4-nitrophenyl)dimethylphosphine oxide (14). To a solution of 5-chloro-2-nitroanisole (0.5
g, 2.67 mmol) in 5 mL of DMF was added dimethylphosphine oxide (0.229 g, 2.93 mmol), palladium
acetate (30 mg, 0.13 mmol), xantphos (0.092 g, 0.16 mmol) and potassium phosphate (0.623 g, 2.93
mmol). The mixture was purged with argon, and heated at 120 °C for 18 h. The reaction mixture was
basified with saturated sodium bicarbonate solution, and extracted with ethyl acetate. The organic layer
was concentrated and purified by silica gel column chromatography to give the final product (0.16 g,
1
3
7
0% yield). H NMR (400 MHz, DMSO-d ) δ 7.98 (dd, J=2.76, 8.03 Hz, 1H), 7.64 (dd, J=1.13, 12.17 Hz, 1H),
6
.50 (ddd, J=1.25, 8.03, 10.54 Hz, 1H), 3.98 (s, 3H), 1.71 (d, J=13.55 Hz, 6H). MS/ES+: m/z = 230.1 [M +
+
H] .
(
4-Amino-3-methoxyphenyl)dimethylphosphine oxide (15). To a solution of 14 (0.1 g, 0.44 mmol) in 5
mL of EtOH was added 10% weight of palladium on carbon (0.2 g). The mixture was purged with argon,
and hydrogenated under 30 psi H for 2 h. The mixture was passed through celite and the filtrates were
2
1
concentrated to give the final product (0.088 g, 86% yield). H NMR (400 MHz, DMSO-d ) δ 6.94-7.14 (m,
6
2
H), 6.67 (dd, J=3.26, 7.78 Hz, 1H), 5.24 (s, 2H), 3.79 (s, 3H), 1.53 (d, J=13.05 Hz, 6H). MS/ES+: m/z=200.2
+
[
M + H] .
2
7
ACS Paragon Plus Environment

Page 29 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
-Methoxy-4-(4-(4-methylpiperazin-1-yl)piperidin-1-yl)aniline (16). A suspension of 5-fluoro-2-
nitroanisole (85.6 g, 0.5 mol), 1-methyl-4-(piperidin-4-yl)piperazine (91.7 g, 0.5 mol) and potassium
o
carbonate (138.5 g, 1.0 mol) in MeCN (500 mL) was stirred at 80 C for 13 h. After cooling, the upper
clear layer was decanted and the residue was taken into DCM (1 L). The resulting suspension was
filtrated and the solid was washed with DCM (500 mL). The combined organic layers were washed with
water (400 mL), dried and concentrated to give the nitro intermediate as a yellow solid (163 g, 97%).
This nitro compound (78 g, 0.23 mol), after dissolved in EtOH (800 mL), was subjected to hydrogenation
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
(
20 psi H ) on palladium-activated carbon (4 g, 10%, wet) in a shaker for 2.5 h. The mixture was filtered
2
through celite and the filtrate was concentrated to give a purple solid (69.8 g, 98% yield).
tert-Butyl 4-(4-(4-amino-3-methoxyphenyl)-1H-pyrazol-1-yl)piperidine-1-carboxylate (17). A sealed
tube was charged with 4-iodo-2-methoxyaniline (0.25 g, 1 mmol), tert-butyl 4-(4-(4,4,5,5-tetramethyl-
1,3,2-dioxaborolan-2-yl)-1H-pyrazol-1-yl)piperidine-1-carboxylate ( 0.50 g, 1.4 mmol), palladium acetate
(
0.012 g, 0.05 mmol), triphenylphosphine (0.026 g, 0.10 mmol) and potassium carbonate (0.18 g, 1.3
mmol). After purging with N the tube was sealed and degassed DMF (1.5 mL) and H O (2.0 mL) was
2
2
added. The resulting solution was refluxed for 10 min followed by cooling to rt. The top layer was
separated and concentrated; the resulting residue was purified by prep-TLC furnishing the desired
1
product (95 mg, 26%). H NMR (400 MHz, CDCl ) δ 7.71 (d, J=0.63 Hz, 1H), 7.58 (d, J=0.63 Hz, 1H), 6.86-
3
6
2
.96 (m, 2H), 6.73 (d, J=7.78 Hz, 1H), 4.29 (tdd, J=4.02, 7.64, 15.32 Hz, 3H), 3.91 (s, 3H), 3.66-3.88 (m,
H), 2.92 (t, J=12.42 Hz, 2H), 2.12-2.25 (m, 2H), 1.96 (dq, J=4.39, 12.21 Hz, 2H), 1.50 (s, 9H).
(
2-Aminophenyl)dimethylphosphine oxide (19). To a solution of 2-iodoaniline (8.6 g, 39.3 mmol) and
dimethylphosphine oxide (3.6 g, 46.6 mmol) in DMF (70 mL) were added potassium phosphate (9.2 g,
42.3 mmol), palladium acetate (0.46 g, 2 mmol), and xantphos (1.2 g, 2 mmol). The reaction mixture was
o
purged with nitrogen and then stirred at 120 C for 6 h. Upon cooling, the resulting mixture was filtered
through a pad of celite (5 g) and the solid was washed with ethyl acetate (2 × 25 mL). The combined
2
8
ACS Paragon Plus Environment