Discovery of the Irreversible Covalent FGFR Inhibitor 8-(3-(4-Acryloylpiperazin-1-yl)propyl)-6-(2,6-dichloro-3,5-dimethoxyphenyl)-2-(methylamino)pyrido[2,3-d]pyrimidin-7(8H)-one (PRN1371) for the Treatment of Solid Tumors

Article abstract of DOI:10.1021/acs.jmedchem.7b00360

Aberrant signaling of the FGF/FGFR pathway occurs frequently in cancers and is an oncogenic driver in many solid tumors. Clinical validation of FGFR as a therapeutic target has been demonstrated in bladder, liver, lung, breast, and gastric cancers. Our go

Full text of DOI:10.1021/acs.jmedchem.7b00360

Subscriber access provided by Olson Library | Northern Michigan University
Drug Annotation
Discovery of the Irreversible Covalent FGFR Inhibitor 8-
(3-(4-acryloylpiperazin-1-yl)propyl)-6-(2,6-dichloro-3,5-
dimethoxyphenyl)-2-(methylamino)pyrido[2,3-d]pyrimidin-7(8H)-
one (PRN1371) for the Treatment of Solid Tumors
Ken A. Brameld, Timothy D Owens, Erik Verner, Eleni Venetsanakos, J. Michael
Bradshaw, Vernon T Phan, Danny Tam, Kwan H. Leung, Jin Shu, Jacob LaStant, David G
Loughhead, Dane E Karr, Mary E Gerritsen, David Michael Goldstein, and Jens Oliver Funk
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b00360 • Publication Date (Web): 30 Jun 2017
Downloaded from http://pubs.acs.org on June 30, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of the Irreversible Covalent FGFR Inhibitor 8ꢀ
(3ꢀ(4ꢀacryloylpiperazinꢀ1ꢀyl)propyl)ꢀ6ꢀ(2,6ꢀdichloroꢀ3,5ꢀ
dimethoxyphenyl)ꢀ2ꢀ(methylamino)pyrido[2,3ꢀ
d]pyrimidinꢀ7(8H)ꢀone (PRN1371) for the Treatment of
Solid Tumors
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
Ken A. Brameld,* Timothy D. Owens, Erik Verner, Eleni Venetsanakos, J. Michael Bradshaw,
Vernon T. Phan, Danny Tam, Kwan Leung, Jin Shu, Jacob LaStant, David G. Loughhead, Dane
E. Karr, Mary E. Gerritsen, David M. Goldstein, Jens Oliver Funk
Principia Biopharma, Inc., 400 East Jamie Court, South San Francisco, California 94080, United
States
KEYWORDS: Fibroblast Growth Factor Receptor, FGFR, Antineoplastic Agents, Tumor Cell
Line, Cell Proliferation, Xenograft Model, Antitumor Assays, Pyrimidinone
ABSTRACT
Aberrant signaling of the FGF/FGFR pathway occurs frequently in cancers and is an oncogenic
driver in many solid tumors. Clinical validation of FGFR as a therapeutic target has been
demonstrated in bladder, liver, lung, breast, and gastric cancers. Our goal was to develop an
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 43
1
2
3
4
5
6
7
8
9
irreversible covalent inhibitor of FGFR1ꢀ4 for use in oncology indications. An irreversible
covalent binding mechanism imparts many desirable pharmacological benefits including high
potency, selectivity, and prolonged target inhibition. Herein we report, the structureꢀbased
design, medicinal chemistry optimization and unique ADME assays of our irreversible covalent
drug discovery program which culminated in the discovery of compound 34 (PRN1371), a
highly selective and potent FGFR1ꢀ4 inhibitor.
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
Next generation sequencing of a broad array of solid tumors identified aberrant signaling
of fibroblast growth factor receptor (FGFR) 1, 2, 3, and 4 in 7% of cancers, implicating the
pathway in oncogenesis and as a potential point of therapeutic intervention.1 FGFR signaling is
initiated by binding of an extracellular FGF ligand which leads to receptor dimerization and
crossꢀphosphorylation of the tyrosine kinase domains. Upon activation, the phosphorylated
kinase domains proceed to phosphorylate intracellular substrates such as PLCγ, Gab1, FRS2, and
STAT1. Subsequent downstream signaling is complex and includes activation of the PI3KꢀAkt
and the Ras/Raf/Mek/Erk pathways. In normal cells, FGFR signaling is important for diverse
functions some of which are cell differentiation, wound healing, and tissue homeostasis2. A wide
range of oncogenic alterations involving activation of the FGFR pathway have been identified
including ligand or receptor amplification or overexpression, receptor mutations, and receptor
translocations/fusion.3
Small molecule FGFR tyrosine kinase inhibitors (TKIs) have shown clinical activity in
the treatment of a variety of solid tumors such as metastatic urothelial4, squamous cell lung5,
intrahepatic cholangiocarcinoma6, and breast cancers5. First generation FGFR TKIs are not
ACS Paragon Plus Environment
2

Page 3 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
selective for the FGFR family, often inhibiting a broad range of additional kinases, commonly
VEGFR1ꢀ3, PDGFRꢀα, and PDGFRꢀβ. Several of these agents, such as dovitinib7 and lucitanib,8
are now in advanced clinical trials and have doseꢀlimiting toxicities largely attributed to VEGFR
inhibition. Next generation FGFR TKIs with improved selectivity are in early stage clinical trials
(Chart 1) and promising clinical responses have been reported for 1 (AZD4547)9 and 2
(infigratinib).10 While not entirely selective for FGFR1ꢀ3 and of moderate potency, 1 and 2 do
not significantly inhibit VEGFR activity; thereby further validating FGFR inhibition alone as
therapeutic modality. Single agent treatment with 2 provided objective response rates (ORR) of
22% and 36% in cholangiocarcinoma harboring FGFR2 translocations6 and in metastatic
urothelial cancer harboring FGFR3 mutations or translocations,4a respectively. Small molecule
inhibitors that bind their target irreversibly through covalent bond formation are showing clinical
benefits over their reversible comparators, particularly for oncology indications. First generation
reversible inhibitors such as bortezomib,11 targeting the proteasome, or erlotinib,12 an EGFR
TKI, are being displaced by irreversible inhibitors carfilzomib13 and afatinib,14 respectively, due
to superior efficacy and relative safety profiles. An irreversible binding mechanism is especially
well suited for TKIs that are required to compete with high intracellular ATP concentrations
while maintaining selectively among the 500+ kinase family members.15 Engaging a nonꢀ
conserved cysteine residue for covalent binding within the small molecule binding site can
impart substantial selectivity benefits over a reversible nonꢀcovalent inhibitor. The resulting
sustained inhibition profile of an irreversible inhibitor requires new protein synthesis to restore
pathway activation. Depending upon the rate of target protein turnover, it is possible to discover
drugs that maintain a prolonged high level of target inhibition yet have rapid systemic clearance
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
3

Journal of Medicinal Chemistry
Page 4 of 43
1
2
3
4
5
6
and a low body burden. Such a profile minimizes any reversible offꢀtarget mediated sideꢀeffects
and is ideal for use in combination therapies which are common in oncology.
7
8
9
Chart 1. Chemical Structures of Secondꢀgeneration Clinical FGFR Inhibitors
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
We set out to discover highly selective irreversible inhibitors of FGFR1ꢀ4 as a targeted
therapy for the treatment of solid tumors having FGFR alterations. The heterogeneity of FGFR
isoforms and types of pathway alterations reported in a wide range of tumor types suggested that
an inhibitor of all FGFR isoforms would provide the broadest therapeutic utility. For example,
FGFR1/2 alterations are found in squamous nonꢀsmall cell lung cancer, FGFR2 alterations are
found in gastric and cholangiocarcinoma, FGFR3 alterations are common in urothelial cancer,
and FGFR4 alterations are found in hepatocellular carcinoma.3 Targeting the kinase activity of
these FGFR isoforms ensures pathway inhibition regardless of the mechanism of activation
which may be due to ligand or receptor amplification or overexpression, receptor mutations, and
receptor translocations/fusion. We sought compounds with high cellular potency capable of
achieving complete target inhibition at low exposures followed by rapid systemic clearance.
Herein we describe the design, structureꢀactivity relationship (SAR), pharmacokinetics and in
ACS Paragon Plus Environment
4

Page 5 of 43
Journal of Medicinal Chemistry
1
2
3
4
vivo efficacy of a series of irreversible covalent FGFR1ꢀ4 inhibitors culminating in the selection
5
6
of compound 34 (PRN1371) for advancement into clinical trials.
7
8
9
RESULTS AND DISCUSSION
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
Starting with a focus on the design of compounds that covalently bind to cysteine
residues, our lead generation strategy entailed identifying a suitable cysteine residue within the
ATP binding pocket and applying structureꢀbased design to generate small focused libraries of
possible lead molecules. Within 8 Å of the ATP binding pocket of FGFR, there are three possible
cysteine residues to target for covalent binding (Figure 1a).16 One of these cysteine residues is
only found in FGFR4 and was not pursued further due to the desire to discover an FGFR1ꢀ4
inhibitor. A second cysteine residue that would require a DFGꢀout ligand binding conformation
is present in FGFR1ꢀ4, however it is also conserved in sixteen additional kinases, presenting a
possible selectivity challenge. On this basis we focused upon covalent binding to Cys488
(FGFR1 residue numbering) which is conserved in only six additional kinases. Importantly, this
cysteine is not present in any of the closely related offꢀtargets, VEGFR2, PDGFRα, or PDGFRβ,
for which high selectivity is desired.
Using multiple published Xꢀray crystal structures of reversible nonꢀcovalent small
molecule ligands bound to FGFR1 (Protein Data Bank (PDB): 2FGI, 3GQL, 3C4F) it was
apparent that Cys488 resides in a flexible region of the glycineꢀrich loop (Figure 1b). Structureꢀ
based ligand design suggested the need to sample a range of linker lengths and trajectories
spanning the region between the core molecular scaffold that recognizes the ATP binding pocket
and the electrophile destined to covalently bond to Cys488. At the onset of the program, a
number of diverse scaffolds were evaluated and based upon the available vectors towards
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 43
1
2
Cys488, we quickly focused upon a series of 6ꢀphenylpyridopyrimidines 3ꢀ417 first reported by
researchers at ParkeꢀDavis (Chart 2). SAR from this work established a preference for 2,6ꢀ
dichloroꢀ or 3,5ꢀdimethoxyꢀ substitutions at the 6ꢀphenyl region to impart selectivity and
potency. Inspection of the Xꢀray crystal structure of 4 bound to FGFR1 (PDB: 2FGI)18 revealed
that the tꢀBuꢀurea at the Cꢀ7 position could be used to approach the cysteine, but taking into
consideration properties of druglikeness, we wanted to avoid a biaryl urea functionality in the
lead. In a subsequent publication from ParkeꢀDavis, isosteric 6ꢀphenylpyridopyrimidinones such
as 519 were reported to maintain FGFR activity. The Nꢀ8 position of the pyridopyrimidinone
provided a good vector in the direction of Cys488 and was synthetically tractable for linker
exploration. Furthermore, a covalent inhibitor of FGFR1ꢀ4 had recently been reported (620, Chart
2) for a dihydropyrimidopyrimidinone core scaffold.
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
6

Page 7 of 43
Journal of Medicinal Chemistry
1
2
3
4
Figure 1. Structural assessment of cysteine residues for covalent binding to FGFR1ꢀ4. (A)
5
6
Several cysteine residues are proximal to the ATP binding pocket as illustrated by the spheres
7
8
9
superimposed on the 2FGI Xꢀray structure.
Listed for each highlighted cysteine
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
sequence/structure location are the other kinases which share a cysteine at that position. (B)
Structural alignment and overlay of three Xꢀray crystal structures (PDB codes: 2FGI, 3GQL,
3C4F) of FGFR1 with different ligands occupying the ATP binding site. The highlighted
cysteine residue (Cys488 in FGFR1) targeted for covalent binding by the compounds described
in this work is located in the flexible glycineꢀrich loop and adopts a different conformation in
each structure.
Chart 2. Chemical Structures of Reported FGFR Inhibitors and Irreversible Covalent Inhibitor
Design Hypothesis
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 43
1
2
3
4
5
6
7
8
9
The design hypothesis used for initial lead generation is depicted in Chart 2. Within a
small series of compounds that differed in the region linking the pyridopyrimidinone scaffold to
the electrophile (e.g., 7ꢀ10, Table 1) a sharp SAR emerged. The pꢀphenethyl linker of compound
10 imparted reasonable potency of 38 nM against FGFR1 while maintaining a desirable >25ꢀfold
selectivity window over VEGFR2. Inhibition profiling at 1 ꢁM of 10 against a subset of kinases
(FGR, cꢀSrc, TNK1, and YES) with a conserved cysteine residue in the same location as Cys488
of FGFR1, identified none with >85% inhibition. In contrast, the original lead 5 was reported to
be >10ꢀfold more potent against cꢀSrc than FGFR in a cellular assay.19 With this selectivity and
potency profile, compound 10 was considered to be a viable lead and further optimization was
initiated.
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 1. FGFR1ꢀ4 and VEGFR2 Activity of Representative 8ꢀpyridopyrimidinone Derivatives
FGFR1
FGFR2
FGFR3
FGFR4
VEGFR2
Compound
R
IC50 (nM)a IC50 (nM) IC50 (nM) IC50 (nM) IC50 (nM)
>1000
>1000
250
~1000b
~1000c
369
>1000
>1000
820
>1000
>1000
2772
ꢀꢀ
ꢀꢀ
7
8
9
997
ACS Paragon Plus Environment
8

Page 9 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
38
34
269
526
1035
10
7
8
9
b
aBiochemical potency against FGFR1ꢀ4 and selected offꢀtarget VEGFR2. Estimated from
c
single concentration inhibition of 66% at 1 ꢁM. Estimated from single concentration inhibition
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
of 58% at 1 ꢁM.
Several assays were used to evaluate the SAR and guide lead optimization. In addition to
traditional biochemical and cellular potency measurements, we relied extensively on an FGFR1
biochemical offꢀrate assay which has been described previously.21 Briefly, preꢀincubation of
inhibitor with FGFR1 is followed by dilution and addition of a fluorescent competitive tracer.
Timeꢀdependent binding by the fluorescent tracer to FGFR1 is detected using timeꢀresolved
fluorescence resonance energy transfer (TRꢀFRET) and can only occur upon dissociation of the
preꢀincubated inhibitor. Our goal was to discover inhibitors that would rapidly and completely
form an irreversible covalent bond upon binding to the receptor. Inhibitors with such a profile
have durable and high target occupancy (e.g., > 90%) at all time points of the experiment (15
min – 24 hrs). Tight binding inhibitors that do not form a covalent bond or only partially bond to
a receptor have low target occupancy at all time points or a reduction in target occupancy over
time. This assay does not replace a traditional kinetics analysis measuring kinact/Ki which defines
the second order rate constant for covalent binding to a target protein.22 While measurements of
kinact and Ki are of great utility, they are generally resource intensive, although a high throughꢀ
put method was recently disclosed.23 For the purposes of guiding our SAR optimization, the
occupancy assay was adapted to a moderate throughꢀput format and provided invaluable kinetics
data.
The cellular activity of compounds was routinely evaluated in two assays. Pathway
inhibition in human umbilical vein endothelial cells (HUVEC) was determined by measuring
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 43
1
2
bFGF induced ERK phosphorylation using Alphascreen technology.24 Inhibition of cell
proliferation was evaluated in the SNU16 gastric cancer cell line which overexpresses FGFR2.
3
4
5
6
7
8
9
Two regions of the lead compound 10 were the focus of initial SAR studies.
Modifications at the Cꢀ2 and Nꢀ8 positions were investigated while maintaining a 2ꢀchloroꢀ3,4ꢀ
dimethoxy substitution pattern on the 6ꢀphenyl substituent. Table 2 depicts the SAR observed
when varying the linker between Nꢀ8 of the pyridopyrimidinone scaffold and the electrophilic
acrylamide. The m-benzyl linker (11), which is reportedly optimal for 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
dihydropyrimidopyrimidinone scaffold (6), had only 65% FGFR1 occupancy at 15 min,
indicative of incomplete covalent bond formation. Extending the linker by one carbon to yield
the phenethyl analogs 12 and 16, improved the 15 min occupancy to 82% and 90%, respectively.
Little difference in biochemical potency or occupancy was measured between the meta (12) and
para (16) acrylamide regioisomers. However, the cellular potencies in both the HUVEC and
SNU16 assays were 2ꢀ3 fold improved for the p-phenethyl linker (e.g. 16, SNU16 IC50 = 1.5
nM) when compared to the mꢀphenethyl analog (e.g. 12, SNU16 IC50 = 5.3 nM) perhaps
indicative of a weak preference for the p-phenethyl linker. Introduction of nitrogen in the phenyl
ring ortho or para to the acrylamide (e.g. 14, 15, 18, 19, and 20) had a pronounced impact on the
FGFR1 occupancy with sustained 100% occupancy observed at both the 15 min and 24 hr time
points for all the compounds. This improved FGFR1 occupancy occurred independent of the
acrylamide regioisomer. Methylation of the acrylamide nitrogen reduced the biochemical
potency and occupancy (e.g., 13 and 17).
Table 2. FGFR1 Biochemical Activity and Occupancy and Cellular Activity of Representative
8ꢀpyridopyrimidinone Derivatives
ACS Paragon Plus Environment
10

Page 11 of 43
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
FGFR1 occupancyb
FGFR1
HUVEC
SNU16
Compound
R
IC50 (nM)a
IC50 (nM)c IC50 (nM)d
15 mins
24 hrs
5.3
2.8
9.4
1.9
1.7
1.3
4.7
1.3
1.3
2.1
65%
56%
ꢀꢀ
6.0
ꢀꢀ
ꢀꢀ
11
82%
64%
60%
60%
5.3
ꢀꢀ
12
13
14
15
16
17
18
19
20
104%
107%
90%
102%
102%
52%
2.2
1.6
3.4
ꢀꢀ
4.7
3.5
1.5
ꢀꢀ
65%
49%
112%
108%
102%
105%
102%
97%
1.6
9.5
ꢀꢀ
1.4
6.8
34
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 43
1
2
aBiochemical potency against FGFR1. bFGFR1 biochemical % target occupancy as determined
3
4
5
6
c
by competition with fluorescent tracer following washout at 15 min or 24 hrs. Inhibition of
d
FGFbꢀmediated ERK phosphorylation in HUVEC primary cells. Inhibition of proliferation of
7
8
SNU16 gastric cancer cell line overexpressing FGFR2.
9
In retrospect, it is apparent that the biochemical occupancy data for compounds 12ꢀ20
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
reflects a combination of the geometric preferences of the linker region and the intrinsic
chemical reactivity of the acrylamide functional group. Reaction rate measurements between
glutathione (GSH) and a diverse set of acrylamides have illuminated a broad spectrum of
intrinsic reactivity that is particularly sensitive to substitutions proximal to the acrylamide. In
the study reported by Flanagan et. al., pyridoacrylamide, the electrophilic functional group in 14,
15, 18, and 19, is most reactive and had the shortest measured halfꢀlife of 0.13 h, whereas the Nꢀ
methylacrylamide in 13 and 17 had a much longer halfꢀlife of 3.5 h.25 We observed that the
more reactive pyridoacrylamides suffered from poor whole blood stability (e.g. 18, mouse whole
blood t½ = 51 min) and made the decision to focus exclusively upon a p-phenethyl linker in
subsequent SAR optimization (e.g., 16).
Next we sought to improve upon the poor aqueous solubility of compound 16 (< 1
ꢁg/mL in pH 7.4 PBS buffer). Building upon published SAR19, 26, a series of alkylamino side
chains were introduced at the Cꢀ2 amino position (Table 3). The more strongly basic amines
(e.g., 21, 22, and 24) showed a precipitous 3ꢀ20 fold loss in biochemical and cellular potencies.
The less basic morpholine, methylpiperazine, and acylpiperazine (e.g., 23, 25, 27, and 28)
retained low nanomolar biochemical potency and had improved FGFR1 occupancy at the 24 h
time point (e.g. 23, FGFR1 IC50 = 1.4 nM, FGFR1 24 h occupancy = 97%). The introduction of
the morpholine side chain did not substantially improve the aqueous solubility of compound 23
(< 1 ꢁg/mL). Despite the low aqueous solubility of 23, results from a rat IV (2 mg/kg) and PO
(20 mg/kg) PK study showed encouraging exposure (AUC = 510 hr*ng/ml) and oral
ACS Paragon Plus Environment
12

Page 13 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
bioavailability (%F =21). Further characterization of compound 23 identified a potential hERG
liability with 81% inhibition measured via a patch clamp assay at 1 ꢁM concentration. Profiling
of additional compounds in the hERG assay established that the alkylamino side chain was an
essential feature of the hERG SAR. Removal of the alkylamino side chain (e.g. 16, hERG % inh
= 5%) or replacement with an alkylhydroxyl side chain (e.g. 29, hERG % inh = 5%) largely
eliminated hERG inhibition. Unfortunately, we were unable to identify a suitable sidechain that
maintained acceptable occupancy, cellular potency, oral bioavailability and low hERG
inhibition. While compound 29 addressed the hERG liability and maintained good cellular
potency (e.g. 29, HUVEC IC50 = 2.4 nM, SNU16 IC50 = 3.8 nM), it had no measurable exposure
upon oral dosing in the mouse.
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 3. FGFR1 Biochemical Activity, Cellular Activity, and hERG potency of Representative
2ꢀamino Derivatives
HUVEC c
FGFR1 occupancyb
FGFR1
hERG
Compound
R
SNU16 d
IC50 (nM)a
% inhe
15 mins
24 hrs
IC50 (nM)
3.4
1.5
1.3
4.7
24
90%
52%
5%
16
21
22
23
28
27
ꢀꢀ
ꢀꢀ
ꢀꢀ
ꢀꢀ
ꢀꢀ
17
93%
103%
86%
97%
8.6
0.9
1.4
81%
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 43
1
2
3
4
5
6
7
8
9
ꢀꢀ
26
13
0.8
0.2
1.1
1.5
1.0
0.8
3.9
108%
96%
100%
ꢀꢀ
110%
89%
84%
ꢀꢀ
ꢀꢀ
65%
72%
92%
ꢀꢀ
24
25
26
27
28
29
30
31
ꢀꢀ
2.1
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
9.4
0.3
11
4.1
12
1.2
96%
82%
ꢀꢀ
65%
60%
ꢀꢀ
2.4
3.8
5%
5%
ꢀꢀ
16
11
37
51
ꢀꢀ
ꢀꢀ
aBiochemical potency against FGFR1. bFGFR1 biochemical % target occupancy as determined
c
by competition with fluorescent tracer following washout at 15 min or 24 hrs. Inhibition of
d
FGFbꢀmediated ERK phosphorylation in HUVEC primary cells. Inhibition of proliferation of
SNU16 gastric cancer cell line overexpressing FGFR2. eIn vitro inhibition at 1 ꢁM compound on
the human etherꢀaꢀgoꢀgoꢀrelated gene (hERG) potassium channel current as measured using
QPatch HT patch clamp system.
Oral bioavailability reflects a complex interplay of metabolism and absorption, the latter
of which is largely determined by permeability and solubility. Predicted intestinal permeability
was consistently high across the series with a typical CACOꢀ2 permeability rate >5 x 10ꢀ6 cm/s
(AꢀB) and a low efflux ratio.27 Recognizing the importance of aqueous solubility, a renewed
effort was made to identify more soluble linkers to span between Nꢀ8 of the pyridopyrimidinone
scaffold and the electrophilic acrylamide. We sought to abandon aromatic features in this linker
region and instead focus upon lower molecular weight aliphatic sidechains that could
ACS Paragon Plus Environment
14

Page 15 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
accommodate alkoxy or amino moieties. Table 4 depicts the SAR for a series of compounds with
a 2,6ꢀdichloroꢀ3,4ꢀdimethoxy substitution pattern on the 6ꢀphenyl substituent and a simple
aminomethyl at Cꢀ2. A diverse set of sidechains that spanned a distance of 5ꢀ6 atoms between
the 2ꢀamino and acrylamide nitrogen atoms were discovered that maintained good occupancy
and cellular potency. Comparing ethylpiperazine (e.g. 33, FGFR1 occupancy 24h = 50%,
HUVEC IC50 = 11 nM) to propylpiperazine (e.g. 34, FGFR1 occupancy 24h = 96%, HUVEC
IC50 = 1.5 nM) established a preference for the longer linker length. Reducing the ring size at the
nitrogen of the acrylamide from piperdine (e.g. 37, FGFR1 occupancy 24h = 32%) to pyrrolidine
(e.g. 38, FGFR1 occupancy 24h = 57%) to azetidine (e.g. 39, FGFR1 occupancy 24h = 110%)
revealed a steep improvement in FGFR1 24h occupancy. As hypothesized, the introduction of an
aliphatic sidechain succeeded in improving the aqueous solubility for all of these compounds.
The greatest improvement was measured for the amine linkers (e.g. 34, PBS solubility pH 7.4 =
> 30 ꢁg/mL) and less so for the alkoxy linkers (e.g. 37, PBS solubility pH 7.4 = 3 ꢁg/mL). The
change in sidechain also addressed the potential for a hERG liability, with 36 and 39 both having
< 5% inhibition at 1 ꢁM. This series of compounds represented a significant breakthrough for the
project and warranted further ADME characterization to identify a clinical candidate.
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 4. FGFR1 Biochemical Activity and Occupancy and Cellular Activity of Representative
8ꢀpyridopyrimidinone Derivatives Incorporating an Aliphatic Linker to the Electrophile
FGFR1
occupancyb
FGFR1
IC50
HUVEC SNU16
Solubility
(ꢁM)e
Cmpd
Electrophile
IC50
IC50
(nM)a
(nM)c
(nM)d
15 mins 24 hrs
0.7
108%
43%
ꢀꢀ
1.8
6.5
32
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 43
1
2
3
4
5
6
7
8
2.3
0.6
102%
101%
50%
96%
11
1.2
2.6
49
59
33
34
9
1.1
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
1.5
135%
87%
4.1
3.4
7.4
35
1.0
0.7
1.8
0.4
103%
108%
87%
99%
32%
1.3
5.6
7.8
1.2
3.4
3.5
2.2
0.5
33
3.3
10
18
36
37
38
39
57%
130%
110%
aBiochemical potency against FGFR1. bFGFR1 biochemical % target occupancy as determined
by competition with fluorescent tracer following washout at 15 min or 24 hrs. Inhibition of
c
d
FGFbꢀmediated ERK phosphorylation in HUVEC primary cells. Inhibition of proliferation of
SNU16 gastric cancer cell line overexpressing FGFR2. eKinetic aqueous solubility (pH 7.4).
A widely appreciated potential liability of Michael acceptors, such as acrylamides, is the
possibility for undesirable reactions with offꢀtarget thiols, such as abundant cysteineꢀcontaining
proteins or GSH.23, 28 These offꢀtarget reactions may translate to rapid clearance and poor ADME
properties. We previously have described a reactivity assay used to characterize reversible
covalent inhibitors of BTK.21 The assay measures the Kd between inhibitor and a simple thiol, βꢀ
mercaptoethanol (BME), by LCꢀMS monitoring of the loss of inhibitor and formation of
inhibitor+BME adduct. In the absence of a chemical equilibrium, as is the case for irreversible
inhibitors, we adapted the assay to measure Kd apparent at a fixed incubation time of 2 hours.
ACS Paragon Plus Environment
16

Page 17 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The measured BME Kd apparent for compounds 37ꢀ39 spans three log units from 43 to 4140 ꢁM
(Table 5). Reducing the ring size at the nitrogen of the acrylamide systematically increases the
chemical reactivity. A similar trend is observed between the aminopiperidine (e.g. 35, BME Kd
= 933 ꢁM) and aminoazetidine (e.g. 36, BME Kd = 43 ꢁM) compounds.
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 5 compares the measured BME apparent Kd values to several in vitro ADME
properties. There is an inverse trend between BME reactivity and plasma stability. A subset of
compounds with the highest FGFR1 24h occupancy from Table 4 were found to be most reactive
towards BME and have the greatest plasma instability (e.g. 36, BME Kd = 43 ꢁM, 2h plasma =
67% remaining and 39, BME Kd = 43 ꢁM, 2h plasma = 32% remaining). On this basis, it
became clear that the improved FGFR1 24h occupancy of the azetidine acrylamide compounds
36 and 39 is a likely consequence of heightened chemical reactivity. Human liver S9 stability,
with or without cofactor NADPH, appears to be independent of BME reactivity. This result was
surprising to us as we anticipated the electrophile to be a metabolic softꢀspot for oxidation and
this may have been correlated with chemical reactivity. As a consequence, when profiling the in
vitro clearance properties of irreversible inhibitors, we found it beneficial to include models of
both hepatic and extraꢀhepatic clearance mechanisms, as others also have reported.29 Rat IV (2
mg/kg) and PO (20 mg/kg) PK for compound 36 demonstrated low oral exposure (AUC = 310
hr*ng/ml) and poor oral bioavailability (%F = 12), consistent with the in vitro predictions.
Table 5. Comparison Between Electrophile Reactivity Towards βꢀmercaptoethanol and Selected
ADME Properties
S9 CLint
Stability (%)b
(uL/min/mg protein)c
BME
Compound
Electrophilic Group
Kd (ꢁM)a
Whole
Blood
w/o
w/
Plasma
NADPH NADPH
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 43
1
2
3
4
5
6
7
8
291
164
96%
70%
95%
73%
2.82
0
7.08
1.35
23
34
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
933
100%
87%
2.3
18.4
35
43
4140
951
43
67%
89%
92%
32%
74%
87%
97%
43%
0.5
0
6.3
71.6
36.9
0.2
36
37
38
39
0
0.6
aChemical reactivity assay measuring apparent Kd of compounds towards βꢀmercaptoethanol
(BME) after 2h incubation. Percent compound remaining after 2 hr incubation with human
plasma or human whole blood. Metabolic stability in pooled human liver S9 fraction with (w/)
b
c
and without (w/o) NADPH cofactor.
Compound 34 presents a unique profile of high biochemical and cellular potency (FGFR1
IC50 = 0.6 nM, SNU16 IC50 = 2.6 nM), prolonged target engagement (FGFR1 occupancy 24h =
96%), < 30% hERG inhibition at 1 ꢁM, and good predicted ADME stability with BME reactivity
Kd > 100 ꢁM. A rat IV (2 mg/kg) PK study of compound 34 showed rapid clearance (Cl = 160
mL/min/kg), yet dosing PO (20 mg/kg) demonstrated high oral exposure (AUC = 4348
hr*ng/ml) and a reasonable halfꢀlife (t1/2 = 3.8 h). Broader kinomeꢀwide biochemical profiling of
34 against 251 kinases identified only FGFR1ꢀ4 and CSF1R as being potently inhibited (e.g.,
IC50 < 20 nM) (Table 6 and Supporting Information Table S1). The ATP binding site of CSF1R
ACS Paragon Plus Environment
18

Page 19 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
does not have a proximal cysteine residue and compound 34 binds nonꢀcovalently, as determined
by recovery of kinase activity upon dialysis. Consistent with reversible binding, there is a large
shift between the biochemical potency (8.1 nM IC50) and the cellular potency of CSF1R
inhibition (> 1500 nM IC50) rendering it physiologically irrelevant.
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 6. Kinase Selectivity Profile of Compound 34
Kinasea
FGFR1
FGFR2
FGFR3
FGFR4
CSF1R
BMX
IC50 (nM) ± Std Devb Kinasea
IC50 (nM) ± Std Devb
410 ± 9 (n=2)
0.6 ± 0.1 (n=5)
1.3 ± 0.2 (n=5)
4.1 ± 0.7 (n=5)
19.3 ± 4.7 (n=5)
8.1 ± 0.6 (n=4)
272 ± 7 (n=2)
371 ± 12 (n=2)
FLTꢀ4
JAK3
556 ± 6 (n=2)
LYNB
LYNA
KDR
631 ± 1 (n=2)
639 ± 5 (n=2)
705 ± 63 (n=5)
792 ± 50 (n=2)
1175 ± 43 (n=2)
PERK
BLK
TAOK2
aPotency of biochemical inhibition by compound 34 of kinases that demonstrated > 50%
inhibition at 1ꢁM from a panel of 251 different kinases (see Supporting Information for
b
inhibition data for full kinase panel). IC50 values for kinase inhibition as determined using the
Caliper electrophoresis method. Uncertainties are the standard deviation (Std Dev) for multiple
independent determinations (n).
The preparation of compound 34, described in scheme 1, begins with chlorination of 3,5ꢀ
dimethoxy toluene 42, followed by benzylic bromination with NBS and displacement with
sodium cyanide to afford compound 43 in three steps. Condensation with aldehyde 44 occurs
under basic conditions in DMF to afford imine 45, which is readily hydrolyzed to produce key
intermediate 46. The linker is introduced selectively at the Nꢀ8 position with iodide 41, which
itself is prepared in two steps from compound 40. The resulting thioether 47 is activated at the 2ꢀ
position of the pyrimidine system by oxidization with mꢀCPBA to afford the corresponding
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 43
1
2
sulfone, which is then readily displaced with methyl amine30 to furnish the penultimate
intermediate 48. Boc deprotection is followed by acylation of the piperazine nitrogen with
acryloyl chloride to yield compound 34 in 41% yield after chromatography.
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
Scheme 1. Chemical synthesis of compound 34
In selecting a clinical candidate, we sought to identify a compound with a
pharmacological profile of complete and rapid target inhibition with minimal exposure and low
body burden. The high cellular potency of compound 34 encouraged us to further characterize its
binding kinetics and pharmacokinetics. To achieve rapid and durable target inhibition requires a
fast onꢀrate and slow or nonexistent offꢀrate. By measuring the progress curves of peptide
substrate phosphorylation at different compound concentrations, the initial velocity (Vi), steady
ACS Paragon Plus Environment
20

Page 21 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
state velocity (Vs), and observed constant of inhibition (Kobs) for compound 34 was determined.
The hyperbolic shape of the progress curves indicated timeꢀdependent inhibition which is a
hallmark of irreversible inhibitors and enabled the calculation of kinact, Ki, and kinact/Ki values for
binding to FGFR1ꢀ4 (Table 7).22 Highly efficient covalent target engagement is apparent for
FGFR1ꢀ3 with a calculated Ki of 1.3ꢀ2.2 nM and kinact/Ki ratio of 0.46ꢀ1.2 (ꢁMꢀ1 secꢀ1). While the
observed weaker affinity for FGFR4 (e.g., Ki = 73 nM) is consistent with the biochemical and
cellular IC50 data, the kinetics studies also confirm covalent irreversible binding to FGFR4.
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 7. Kinetics Values of Compound 34 Binding to FGFR1ꢀ4
kinact (secꢀ1)
kinact/Ki (µMꢀ1 secꢀ1)
Target
Ki (nM)
FGFR1
FGFR2
FGFR3
FGFR4
0.0019 ± 0.0015
1.6 ± 1.1
1.2 ± 0.1
0.00092 ± 0.00016 1.3 ± 0.6
0.00097 ± 0.00015 2.2 ± 0.5
0.00066 ± 0.00007 73 ± 22
0.75 ± 0.17
0.46 ± 0.06
0.010 ± 0.003
To minimize offꢀtarget pharmacology and maximize tolerability when used in
combination therapy, the ideal PK profile of a slow offꢀrate inhibitor consists of high oral
absorption followed by rapid systemic clearance. PK studies of compound 34 in rat, dog, and
cynomolgus monkey showed rapid IV clearance in all species; however there were large species
differences in oral exposure and bioavailability for monkey compared to rat and dog (Table 8). In
rat, high exposure upon oral dosing (e.g., Cmax = 1785 ng/ml, AUC = 4348 ng*hr/ml) and >100%
bioavailability (F%) suggested good absorption and partial saturation of clearance mechanisms at
the 20 mg/kg dose. Unique to the rat, there is a large difference in halfꢀlife between the IV (t1/2
0.8 h) and PO (t1/2 = 3.8 h) routes of administration, also indicative of possible saturation of a
=
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 43
1
2
3
4
5
6
7
8
9
clearance mechanism upon oral dosing. In the dog, the same methylcellulose suspension
formulation used for the rat gave low oral absorption and bioavailability (F% < 15%). We
hypothesized that the less acidic gastric and intestinal pH of the dog may be contributing to low
absorption of the free base of compound 34.31 Coꢀadministration of a molar equivalent of citric
acid improved the oral absorption of a 10 mg/kg dose (e.g., Cmax = 1103 ng/ml, AUC = 1134
ng*hr/ml, F% = 94) and brought it in line with the rat PK. Exceptionally low oral exposure in the
monkey (e.g., Cmax = 96 ng/ml, AUC = 84 ng*hr/ml) was initially a cause for concern. We were
able to attribute this to intestinal Cyp3A4 mediated metabolism. It has been reported that the
bioavailability in monkey of compounds that undergo intestinal metabolism is much lower than
that of rat or human.32 For neratinb and ibrutinib, two covalent kinase inhibitors with acrylamide
Michael acceptors that undergo extensive Cyp3A4 mediated metabolism, monkey PK grossly
overestimates clearance and underestimates absorption, making monkey an inappropriate species
for the prediction of human absorption.29 On this basis, we were comfortable with the preclinical
PK of compound 34 as it predicted our desired profile of high oral absorption followed by rapid
clearance.
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 8. Pharmacokinetic Parameters of Compound 34
Parameter
Rata
Dogb
Monkeyc
IV
160
PO
NA
IV
117
PO
NA
IV
PO
NA
Cl (mL/min/kg)
Vss (L/kg)
42.9
1.14
0.349
999
7.20
0.764
596
NA
2.90
0.33
1087
284
NA
NA
t1/2 (h)
3.77
1785
4348
0.58
1103
1344
0.486
95.8
84.4
Cmax (ng/mL)
AUC (ng.hr/mL)
208
472
ACS Paragon Plus Environment
22

Page 23 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
F (%)
NA
208
NA
94
NA
2.3
aFemale SpragueꢀDawley rats dosed at 2 mg/kg IV and 20 mg/kg PO. Male Beagle dogs dosed
at 2 mg/kg IV and 10 mg/kg PO. Male Cynomolgus monkeys dosed at 1.23 mg/kg IV and 10
mg/kg PO.
b
c
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
Several in vivo studies of compound 34 were conducted to understand the relationship
between PK and FGFR pathway inhibition. Exploratory mouse studies identified the chemokine
CCL2 as elevated following tail vein injection of bFGF. Specifically, upon stimulation with 10
µg bFGF, it was found that CCL2 increased from a basal level of 301 ± 101 pg/ml to a level of
1487 ± 537 pg/ml at 4 hours postꢀstimulation. We evaluated the percent inhibition of bFGFꢀ
induced CCL2 at 1, 5, and 12 hrs after oral administration of a 10 mg/kg dose of compound 34 in
rats. The levels of compound 34 in circulation were measured at the same time points to assess
whether CCL2 inhibition was driven by longꢀlasting binding of compound 34 to FGFR or by
compound in circulation. The plasma concentration of compound 34 peaked at 538 ± 88 ng/ml,
1 hr after dosing and fell to a nearly undetectable level after 12 hrs (Figure 2b). Strong inhibition
of CCL2 was sustained from 99.5% ± 3.8% at 1 hr to 93.5% ± 4.7% 12 hrs after dosing,
suggesting longꢀlasting FGFR target inhibition by compound 34. The effect of noncovalent
FGFR inhibitor 2 on CCL2 inhibition was also investigated. In contrast to compound 34, the
level of CCL2 inhibition significantly declined from 98.2% ± 4.1% at 5 hrs to 59.5 % ± 11.6% at
12 hrs for compound 2 (Figure 2a), indicating that the covalent binding mechanism of compound
34 to FGFR is likely responsible for the long lasting inhibition of CCL2.
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 43
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 2. Rat in vivo inhibition of bFGFꢀinduced CCL2 production (black line, right yꢀaxis) by a
10 mg/kg dose of compounds 2 or 34 and associated compound plasma levels (grey line, left yꢀ
axis) at 1, 5, and 12 hrs post dose (n=5 per group). (A) Reversible compound 2. (B) Irreversible
covalent compound 34.
Compound 34 was evaluated in pharmacodynamics and efficacy studies using a SNU16
gastric cancer xenograft mouse model with high overexpression of FGFR2. In nude mice
implanted with subcutaneous SNU16 tumors, pFGFR2 levels in the tumor were measured via
Western blotting at 8 hrs following a 10 mg/kg oral dose (Figure 3a). Low levels of pFGFR2
confirmed the ability of compound 34 to block FGFR2 activity in tumor tissue. Efficacy was
determined by measuring tumor growth inhibition in the same SNU16 xenograft model (Figure
3b). Compound 34 induced a doseꢀdependent reduction in tumor volume and up to 68% tumor
growth inhibition at the highest dose of 10 mg/kg BID following 27 days of treatment. All doses
were well tolerated with no significant body weight loss.
ACS Paragon Plus Environment
24

Page 25 of 43
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 3. PK/PD and efficacy of compound 34 in a SNU16 mouse xenograft model. (A)
Western blots showing blockade of pFGFR within the tumor 8 hrs postꢀdose. Lanes 1ꢀ4 are
individual mice dosed with vehicle and lanes 5ꢀ8 are individual mice dosed with compound 34 at
10 mg/kg QD. (B) Tumor growth inhibition (TGI) dose response upon oral administration of
compound 34 (n=10 per dose group).
With good in vivo efficacy and PK, compound 34 was progressed into preclinical safety
evaluations including 28 day GLP toxicology studies in rats and dogs. The toxicological findings
were consistent with those reported for other FGFR inhibitors, predominantly phosphorus
dysregulation and concomitant soft tissue mineralization.33 Phosphorus homeostasis is dependent
upon FGF23 signaling in the kidney. As a consequence, clinical onꢀtarget mediated effects of
FGFR blockade include elevated serum FGF23, phosphate, and vitamin D.4b, 34 The combination
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 43
1
2
3
4
of a clean preclinical safety profile, favorable human PK projections and efficacy in the
5
6
xenograft models gave us confidence to advance compound 34 into human clinical trials.
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. Phase I clinical data for once daily oral administration of compound 34. (A) Total
plasma concentration over time. (B) Percent change at day 8 of serum phosphate, a
pharmacodynamic marker of FGFR inhibition; (*) indicates prophylactic administration of
phosphate binders.
A phase 1 dose escalation study is being conducted in patients with advanced solid
tumors and metastatic disease to assess pharmacokinetics, tolerability, and objective response
rate, among other endpoints (ClinicalTrials.gov identifier NCT02608125). Compound 34 free
base has been administered orally once daily as powder in a capsule on a 28ꢀday continuous
schedule. Human plasma concentrations for doses ranging from 15 to 35 mg (Figure 4a) confirm
ACS Paragon Plus Environment
26

Page 27 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
good oral exposure, rapid systemic clearance, no accumulation from day 1 to day 15, and a doseꢀ
dependent increase in AUC.35 Serum phosphate, a pharmacodynamic marker of FGFR inhibition,
is increased for all doses studied and shows a doseꢀdependent increase between 20 and 35 mg,
despite the administration of prophylactic phosphate binders (Figure 4b). Additional cohorts and
dosing regimens are being explored in ongoing clinical studies and will be reported in due
course.
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
CONCLUSIONS
Structureꢀbased design and medicinal chemistry optimization that was focused on
maximizing FGFR target occupancy, led to the discovery of compound 34, an irreversible
nanomolar inhibitor of FGFR1ꢀ4. Several specific assays were critical in guiding the lead
optimization process including measures of FGFR1 occupancy, human plasma stability, and
Michael acceptor chemical reactivity. Initial progress towards optimization of a potent lead
molecule (compound 10) having a p-phenethyl linker between Nꢀ8 of the pyridopyrimidinone
scaffold and the electrophilic acrylamide was unsuccessful in identifying a clinical candidate
primarily due to poor solubility and a potential hERG liability. Subsequent incorporation of an
aliphatic amine in place of the p-phenethyl linker led to compound 34 which maintained high
FGFR1 occupancy with improved solubility and exceptional oral bioavailability. A low dose,
high clearance, irreversible covalent drug has the potential for a large therapeutic window due to
durable target inhibition that continues after the drug has cleared systemic circulation.
Compound 34 has this ideal PK profile and as a consequence of its irreversible binding
mechanism, demonstrates extended FGFR inhibition in vivo after plasma concentrations are
below the limits of quantitation. For irreversible inhibitors, the rate of new protein synthesis
largely determines the duration of target inhibition. The protein halfꢀlife of FGFR3 has been
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 43
1
2
reported to be approximately 4 hours and is context dependent.36 PK/PD studies in preclinical
xenograft models confirmed blockade of FGFR signaling and potent tumor growth inhibition.
Initial cohorts of a phase 1 dose escalation trial support the predicted PK profile, were well
tolerated, and warrant further clinical investigation. Expanded trials in a targeted patient
population will be required to demonstrate the clinical benefits of an irreversible covalent
binding mechanism for FGFR inhibition.
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
EXPERIMENTAL SECTION
Cell lines and reagents. The gastric cancer cell line SNU16 was purchased from
American Tissue Culture Collection (ATCC, Manassas, VA) and cultured in RPMIꢀ1640
medium with 10% fetal bovine serum (FBS, Gibco Cat#26140ꢀ079). HUVECs (Genlantis
Cat#PH20005N) were maintained in endothelial cell growth medium (Genlantis Cat#PM211500)
containing 10% FBS. Cells were maintained in standard culture conditions of 37°C, 5% CO2
and 95% humidity and were kept in culture for up to 15ꢀ20 passages.
Kd determination using βꢀmercaptoethanol. Solutions were prepared containing 0,
1.5, 15, 150, and 1500 mM βꢀmercaptoethanol (BME) in a 1:1 mixture of ethanol and phosphate
buffered saline (pH 7.4, PBS). Aliquots of a 10 mM DMSO stock solution of test compound (10
µL) were separately added to 90 µL aliquots of the aboveꢀdescribed ethanol/PBS solutions
containing 0 to 1500 mM BME. After these solutions had been allowed to stand at room
temperature for two hours, they were analyzed using an Agilent 1200 LCMS system equipped
with a 50 x 2 mm Phenomenex Luna 5µ C18 100A column. Samples were eluted using a
gradient of acetonitrile and water, with both solvents containing 0.1% formic acid. Peaks
corresponding to parent and BME adduct were identified by their masses and the percent parent
ACS Paragon Plus Environment
28

Page 29 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
in each sample was determined by measuring the area under the curve for the extracted mass
peaks from the positive ion trace corresponding to parent and BME adduct. Percent parent was
plotted versus the log of the BME concentration using GraphPad Prism to determine an apparent
Kd for the reaction.
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
Kinase assays. Enzyme inhibition was determined using a Caliper capillary
electrophoresis system that separates phosphorylated and nonꢀphosphorylated peptides on the
basis of charge. Different concentrations of inhibitor were first preꢀincubated with enzyme for
15 minutes. The reaction was initiated with addition of peptide substrate, ATP, and Mg2+ and
incubated at 25oC for 3 hr. To stop the reaction, the mixture was quenched with EDTA. The
buffer was 100 mM HEPES, pH 7.5, 0.1% BSA, 0.01% Triton Xꢀ100, 1 mM DTT, 10 mM
MgCl2, 10 mM sodium orthovanadate, 10 µM betaꢀglycerophosphate and 1% DMSO. The ATP
concentration of the reaction was at the preꢀdetermined value of the Km for ATP. All enzyme
inhibition data was acquired by Nanosyn, Inc. (Santa Clara, CA; www.nanosyn.com). Kinase
selectivity was also evaluated using the Nanosyn 250 kinase panel.
ERK phosphorylation in HUVECs. Human umbilical vein endothelial cells (HUVECs)
were incubated in media supplemented with 10% FBS and seeded at 30,000 cells per well in a
96ꢀwell plate overnight. HUVECs were then transferred into serum free media 1 hr before
compound treatment. A compound concentration series was added to cells and incubated for 1
hr at 37oC. Cells were then stimulated with either 50 ng/ml of FGF2 (R&D Systems Cat#233ꢀ
FGꢀ025) or 50 ng/ml of VEGF (R&D Systems Cat#293ꢀVEꢀ050) for 10 min. Ice cold PBS was
added to stop the reaction, and cells were washed three times to remove media. A pERK
SureFire kit (PerkinElmer) was utilized to determine ERK phosphorylation using an Envision
multilabel plate reader (PerkinElmer).
ACS Paragon Plus Environment
29