Discovery of 1-{(3R,4R)-3-[({5-Chloro-2-[(1-methyl-1H-pyrazol-4-yl)amino]-7H-pyrrolo[2,3-d]pyrimidin-4-yl}oxy)methyl]-4-methoxypyrrolidin-1-yl}prop-2-en-1-one (PF-06459988), a Potent, WT Sparing, Irreversible Inhibitor of T790M-Containing EGFR Mutants

Article abstract of DOI:10.1021/acs.jmedchem.5b01633

First generation EGFR TKIs (gefitinib, erlotinib) provide significant clinical benefit for NSCLC cancer patients with oncogenic EGFR mutations. Ultimately, these patients' disease progresses, often driven by a second-site mutation in the EGFR kinase domain (T790M). Another liability of the first generation drugs is severe adverse events driven by inhibition of WT EGFR. As such, our goal was to develop a highly potent irreversible inhibitor with the largest selectivity ratio between the drug-resistant double mutants (L858R/T790M, Del/T790M) and WT EGFR. A unique approach to develop covalent inhibitors, optimization of reversible binding affinity, served as a cornerstone of this effort. PF-06459988 was discovered as a novel, third generation irreversible inhibitor, which demonstrates (i) high potency and specificity to the T790M-containing double mutant EGFRs, (ii) minimal intrinsic chemical reactivity of the electrophilic warhead, (iii) greatly reduced proteome reactivity relative to earlier irreversible EGFR inhibitors, and (iv) minimal activity against WT EGFR.

Full text of DOI:10.1021/acs.jmedchem.5b01633

Subscriber access provided by GAZI UNIV
Article
Discovery of 1-{(3R,4R)-3-[5-Chloro-2-(1-methyl-1H-pyrazol-4-
ylamino)-7H-pyrrolo[2,3-d]pyrimidin-4-yloxymethyl]-4-methoxy-
pyrrolidin-1-yl}propenone (PF-06459988), A Potent, WT Sparing,
Irreversible Inhibitor of T790M-Containing EGFR Mutants
Hengmiao Cheng, Sajiv K Nair, Brion W. Murray, Chau Almaden, Simon Bailey, Sangita Baxi, Douglas
Carl Behenna, Sujin Cho-Schultz, Deepak Dalvie, Dac M Dinh, Martin P. Edwards, Jun Li Feng, Roseanne
Ferre, Ketan S Gajiwala, Michelle D Hemkens, Amy Jackson-Fisher, Mehran Jalaie, Theodore Otto Johnson,
Robert S Kania, Susan Kephart, Jennifer Anne Lafontaine, Elizabeth Lunney, Kevin K. -C. Liu, Zhengyu
Liu, Jean Matthews, Asako Nagata, Sherry Niessen, Martha A. Ornelas, Suvi T M Orr, Mason Pairish,
Simon Planken, Shijian Ren, Daniel Tyler Richter, Kevin Ryan, Neal W Sach, Hong Shen, Tod Smeal,
James Solowiej, Scott C. Sutton, Khanh Tran, Elaine Tseng, William F Vernier, Marlena Walls, Shuiwang
Wang, Scott L Weinrich, Shuibo Xin, Haiwei Xu, Min-Jean Yin, Ru Zhou, Michael Zientek, and John Kath
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.5b01633 • Publication Date (Web): 12 Jan 2016
Downloaded from http://pubs.acs.org on January 23, 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 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Richter, Daniel; Pfizer Global Research and Development, Discovery
Ryan, Kevin; Pfizer
Sach, Neal; Pfizer Worldwide R&D, Medicinal Chemistry
Shen, Hong; Pfizer, Chemistry
Smeal, Tod; Eli Lilly and Company
Solowiej, James; Pfizer
Sutton, Scott; Pfizer, Inc., Cancer Chemistry
Tran, Khanh; Pfizer
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
Tseng, Elaine; Pfizer
Vernier, William; Pfizer
Walls, Marlena; Regulus Therapeutics Inc,
Wang, Shuiwang; WuXi Apptec
Weinrich, Scott; Pfizer
Xin, Shuibo; WuXi App Tec Co., Shanghai, China
Xu, Haiwei; WuXi Apptec
Yin, Min-Jean; Pfizer
Zhou, Ru; Pfizer
Zientek, Michael; Pfizer, PDM
Kath, John; Pfizer Global Research & Development, Worldwide Medicinal
Chemistry
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 77
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 1ꢀ{(3R,4R)ꢀ3ꢀ[5ꢀChloroꢀ2ꢀ(1ꢀmethylꢀ1Hꢀpyrazolꢀ4ꢀylamino)ꢀ7Hꢀpyrrolo[2,3ꢀ
d]pyrimidinꢀ4ꢀyloxymethyl]ꢀ4ꢀmethoxyꢀpyrrolidinꢀ1ꢀyl}propenone
(PFꢀ06459988),
A
Potent, WT Sparing, Irreversible Inhibitor of T790MꢀContaining EGFR Mutants
†
†
†
†
†
Hengmiao Cheng,* Sajiv K. Nair,* Brion W. Murray,* Chau Almaden, Simon Bailey,
†
†
†
†
†
Sangita Baxi, Doug Behenna, Sujin ChoꢀSchultz, Deepak Dalvie, Dac M. Dinh, Martin P.
Edwards, Jun Li Feng, Rose Ann Ferre, Ketan S. Gajiwala, Michelle D. Hemkens, Amy
JacksonꢀFisher, Mehran Jalaie, Ted O. Johnson, Robert S. Kania, Susan Kephart, Jennifer
Lafontaine, Beth Lunney, Kevin K.ꢀC. Liu, Zhengyu Liu, Jean Matthews, Asako Nagata,
Sherry Niessen, Martha A. Ornelas, Suvi T. M. Orr, Mason Pairish, Simon Planken, Shijian
Ren, Daniel Richter, Kevin Ryan, Neal Sach, Hong Shen, Tod Smeal, Jim Solowiej, Scott
Sutton, Khanh Tran, Elaine Tseng, William Vernier, Marlena Walls, Shuiwang Wang, Scott
L. Weinrich, Shuibo Xin, Haiwei Xu, MinꢀJean Yin, Michael Zientek, Ru Zhou, and John
†
†
†
†
†
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
†
†
†
†
†
†
†
†
†
†
†
†
†
†
†
†
≠
†
†
†
†
†
†
†
†
†
†
†
≠
†
≠
≠
†
†
†
†
C. Kath
†
La Jolla Laboratories, Pfizer Worldwide Research and Development, 10770 Science Center
Drive, San Diego, California 92121, United States
≠
Wuxi AppTec, 288 Fute Zhong Road, Waigaoqiao Free Trade Zone, Shanghai 200131, China
Abstract: First generation EGFR TKIs (gefitinib, erlotinib) provide significant clinical benefit
for NSCLC cancer patients with oncogenic EGFR mutations. Ultimately, these patients disease
progresses, often driven by a secondꢀsite mutation in the EGFR kinase domain
(T790M). Another liability of the first generation drugs is severe adverse events driven by
inhibition of WT EGFR. As such, our goal was to develop a highly potent irreversible inhibitor
with the largest selectivity ratio between the drugꢀresistant double mutants (L858R/T790M,
Del/T790M) and WT EGFR. A unique approach to develop covalent inhibitors, optimization of
reversible binding affinity, served as a cornerstone of this effort. PFꢀ06459988, was discovered
as a novel, third generation irreversible inhibitor, which demonstrates (i) high potency and
specificity to the T790Mꢀcontaining double mutant EGFRs, (ii) minimal intrinsic chemical
reactivity of the electrophilic warhead, (iii) greatly reduced proteome reactivity relative to earlier
irreversible EGFR inhibitors and (iv) minimal activity against WT EGFR.
ACS Paragon Plus Environment

Page 3 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
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
Lung cancer is the leading cause of cancerꢀrelated death worldwide, with an estimated over 1
1
million new cases diagnosed and over 1 million deaths each year. In nonꢀsmall cell lung cancer
(
NSCLC) which accounts for 85% of lung cancer, patients harboring mutations in EGFR
constitute between 10ꢀ30% of the overall patient population. Two frequent and mutuallyꢀ
exclusive primary mutations occur either in the activation loop as a point mutation (EGFR
L858R) or by short deletion in exon 19 (EGFRꢀDel), together accounting for approximately 85
2
% of all cases. These somatic EGFR mutants enable constitutive activation and are strong
3
4
predictive biomarkers of response to first generation EGFR TKIs (gefitinib and erlotinib )
Figure 1). In contrast, these drugs are doseꢀlimited by potent inhibition of their originally
(
intended target wild type (WT) EGFR, which causes epitheliumꢀbased toxicities by blocking its
5,6
role in normal adult physiology.
Treatment of EGFRꢀmutant NSCLC patients with the first
generation EGFR TKIs provides excellent response rates and disease control for 11 to 14
months, but patients invariably become resistant to these therapies and their disease progresses.
Of the drugꢀresistant EGFRꢀmutant NSCLC patients, approximately 60% harbor an additional
7
mutation in the EGFR kinase domain (T790M) in cis with the primary activating mutation that
renders the receptor insensitive to inhibition by the EGFR TKIs, and thus, restoring constitutive
EGFR signaling. The resistance mechanism derived from T790M mutation is believed to come
from a number of components, including the increased ATP binding affinity for T790M mutants,
the steric clash between the Met790 gatekeeper side chain and the aniline moiety that is utilized
by the first generation EGFR TKIs to bind in the kinase backꢀpocket and the altered catalytic
8
,9,10
domain conformation.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 77
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The first generation of EGFR drugs relied on reversible affinity to achieve potency. More
recently, a covalent inhibition strategy has been used because it delivers distinct pharmacological
properties. Irreversible kinase inhibitors can achieve advantages over reversible compounds by
delivering complete and sustained target engagement in the presence of high intracellular
concentrations of ATP in the cells, and by requiring the physical turnover of kinase proteins to
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
1,12,13
restore inhibited signaling pathways.
Indeed, second generation, irreversible inhibitors of
1
4
15
16
EGFR WT (canertinib (CIꢀ1033) ; dacomitinib ; afatinib ) (Figure 1) which covalently modify
EGFR Cys797 demonstrate increased cellular potency against T790M mutants relative to their
reversible counterparts . However, because these compounds maintain the aniline moiety that
clashes with Met790 side chain, their T790M activity is less than their activity against the
primary activating EGFR mutants. As a result, their clinical efficacy to treat T790M patients is
limited due to doseꢀlimiting toxicity associated with inhibition of WT EGFR.
st
nd
Figure 1. Summary of 1 generation and 2 generation EGFR inhibitors.
ACS Paragon Plus Environment

Page 5 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In response to this unmet need, and because irreversible inhibitors demonstrated better potency
against EGFR T790M mutants than reversible inhibitors, an effort was initiated to discover
highly potent irreversible inhibitors of the drug resistant T790M mutants that spare EGFR WT
activity to the greatest extent as a means to minimize mechanismꢀbased toxicities. Others have
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
7
pursued irreversible inhibitors of EGFR T790M, resulting in the identification of WZꢀ4002 (7),
1
8
19,20,21
rociletinib (COꢀ1686, 8) and osimertinib
(AZD9291, 9) (Figure 2). Both rociletinib and
osimertinib have progressed to advanced clinical trials. In addition, there are several recent
publications discussing the design and SAR for novel third generation irreversible EGFR
22,23,24,25,26
inhibitors.
The discovery of 1ꢀ{(3R,4R)ꢀ3ꢀ[5ꢀChloroꢀ2ꢀ(1ꢀmethylꢀ1Hꢀpyrazolꢀ4ꢀ
ylamino)ꢀ7Hꢀpyrrolo[2,3ꢀd]pyrimidinꢀ4ꢀyloxymethyl]ꢀ4ꢀmethoxyꢀpyrrolidinꢀ1ꢀyl}propenone
2
7,28
(PFꢀ06459988, 1)
, a highly potent inhibitor of EGFR T790M with unprecedented selectivity
over EGFR WT, is described in this manuscript.
rd
Figure 2. Summary of 3 generation EGFR inhibitors.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 77
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
RESULTS AND DISCUSSION
Objective.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Kinetic mechanism for twoꢀstep covalent inhibition of EGFR.
The inactivation of EGFR occurs through a twoꢀstep process that begins with reversible binding
to EGFR followed by covalent bond formation with the thiol of Cys797 to form a covalent
adduct (Figure 3). Therefore, irreversible inhibitor potency can be increased by maximizing the
reversible binding affinity (K =k /k ), the specific reactivity (kinact), or the inherent chemical
i
off on
reactivity of the electrophile. In addition to optimizing T790M potency and selectivity over WT,
an equally significant emphasis was placed on reducing warhead reactivity to minimize
proteomeꢀwide reactivity and decrease the probability of idiosyncratic toxicity. To achieve
these multiple objectives, a key strategy was to optimize the potency of inhibitors based on their
reversible (nonꢀcovalent) binding affinity to EGFR T790M as opposed to the chemical reactivity
of the warhead. To effectively carry out this strategy, a high capacity binding kinetics assay was
developed which is capable of separating the reversible binding affinity from the inactivation
rate. Typically, irreversible kinetic analysis to determine reversible affinity (K ), and specific
i
reactivity kinact is low throughput and labor intensive. However, to truly leverage binding affinity
ACS Paragon Plus Environment

Page 7 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
SAR it was essential to be able to test all new inhibitors on a weekly basis. To do so, a high
throughput irreversible kinetics screening assay capable of coarsely defining reversible binding
est
affinity (K ) and overall irreversible inactivation (k /I) was developed. Specialized reaction
i
obs
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
29,30,31
conditions were used to enhance the predictability of this method.
Since an inhibitor’s
inactivation rate is governed by a pseudoꢀsecondꢀorder rate constant (kinact/K ), it is highly
i
dependent on the inhibitor concentration and its chemical properties. It was recognized that in
any given week a range of inhibitors would be tested from weak binders to highly optimized
inhibitors. To address this issue, five inhibitor concentrations (0.01, 0.05, 0.1, 1 and 5 µM) were
selected to ensure at least one or two are in the proper range to measure the initial reversible
binding event (between 25ꢀ80% inhibition of the initial linear (60ꢀ120 sec) portion of the
inactivation timeꢀcourse). When the initial region of the inactivation time course is linear, then
the dominant contribution to inhibition is reversible binding affinity. Placing a constraint on the
observed percent inhibition (25ꢀ80%) kept the measurement in an accurate region; enough
inhibition to be confident in the binding affinity but not too much that there is a large
contribution from reactivity or too slow of an enzymatic rate to be accurate. The full inactivation
time course for a given inhibitor concentration is defined by a rate constant (k ). To normalize
obs
the contribution of the inhibitor concentration to the observed inactivation rate, it is divided by
the inhibitor concentration (k /I). This value encompasses contributions from both reversible
obs
binding affinity and chemical reactivity. Under the assay conditions, k /I is correlated to
obs
est
kinact/K . To determine the correlation between Ki and K , both established and absolute values
i
i
were generated for a structurally diverse set of compounds using the L858R/T790M enzyme
est
(
Figure 4). The correlation between these values supported the use of Ki for inhibitor design.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 77
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
2
Figure 4. Correlation between absolute and estimated K values, R = 0.68.
i
The irreversible screening assay provides two important pieces of data from an appropriate
progression curve; 1) the percent inhibition value at the drug concentration used to generate that
est
curve which provides the Ki and 2) the k /I value (which is proportional to k /K ) from that
obs
inact
i
same progression curve which provides a measure of enzyme inactivation. The assay
consistently provided data for multiple compounds each week (medium number of compounds
per week = 14; maximum number of compounds run in one week = 91) and became the primary
assay used throughout the duration of the program.
The irreversible screening assay was supplemented with two additional assays to round out the
set of primary screens used to fully characterize irreversible inhibitor SAR. To further assist in
minimizing warhead reactivity, a glutathione (GSH) conjugation assay was developed which
2
9
monitored parent compound loss in the presence of GSH to establish a halfꢀlife. Canertinib
was used as a control, and the halfꢀlives of test compounds were normalized to that of canertinib
ACS Paragon Plus Environment

Page 9 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
t1/2 ca: 21 min); to describe the intrinsic chemical reactivity of the warhead. To drive toward
the critical objective of achieving the high levels of selectivity for EGFR T790M over EGFR
WT a pair of cellular ELISA assays were used which measure the ability of a compound to
inhibit phosphorylation of EGFR L858R/T790M (H1975 cells) relative to EGFR WT (A549
cells).
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
Establishing Early Leads
The critical structural difference in the surface of the ATP binding pocket between EGFR WT
and the T790M mutant resides at the gatekeeper residue. This mutation changes the amino acid
residue from the polar threonine to a nonꢀpolar methionine. Therefore, to attain the largest
difference in potency between EGFR T790M and EGFR WT, optimizing hydrophobic
interactions between an inhibitor and the Met790 side chain may provide high levels of T790M
potency and selectivity. Modeling studies indicated that the pyrrole portion of a
pyrrolopyrimidine (PP) core would achieve effective hydrophobic interactions with the Met790
side chain and provide an additional hydrogen bond interaction with the backbone carbonyl of
Gln791 (Figure 5).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 77
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Modeled interactions of PP core with Met790 and hinge residue of monomeric L858R/T790M
EGFR. Hydrophobic surface of the binding site near the hinge and T790M is shown.
Initial designs with the PP core led to the identification of an early lead, compound 10 (Figure 6).
This compound had a high binding affinity to the L858R/T790M double mutant (DM) with a
est
Ki of 2 nM (43% inhibition at 0.05 µM), and a substantial inactivation rate with k /I of 51,300
obs
ꢀ1 ꢀ1
M s . The k /I value was obtained at 0.05 µM inhibitor concentration. The substantial
obs
ꢀ1 ꢀ1
inactivation rate for 10 was confirmed by the measured kinact/K rate of 59,123,000 M s .
i
Compound 10 was an attractive lead, potently inhibiting EGFR L858R/T790M
autophosphorylation (IC = 4 nM) in the H1975 cell line but only modestly inhibiting EGFR
5
0
WT autophosphorylation (IC50 = 0.79 µM) in the A549 cell line. Relative to benchmark
compound canertinib in a glutathione conjugation assay, compound 10 exhibits a 6ꢀfold
reduction in intrinsic chemical reactivity with thiol nucleophiles.
ACS Paragon Plus Environment

Page 11 of 77
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 6. Summary of key in vitro potency and WT selectivity data for 10.
3
2
Using our crystallography platform we obtained a cocrystal structure of 10 with monomeric
EGFR L858R/T790M (Figure 7a). The 1.5 Å ligandꢀprotein complex shows covalent
modification of the target residue Cys797 via the Michael acceptor of the ligand. The enzyme is
in a unique conformation for EGFR relative to previous structures with respect to the Asp855ꢀ
Phe856ꢀGly857 (DFG) motif and the activation loop conformation. While the Phe856 side chain
is oriented out of the hydrophobic core of the protein, partially occupying the ATP binding site,
the Asp855 side chain is positioned into the hydrophobic core in a DFGꢀout conformation. The
conformation of the rest of the activation loop is nearly identical to that in the active EGFR
kinase domain. There is an approximately 20 degree rotation of the Nꢀlobe relative to the Cꢀlobe
when compared with the apoꢀprotein structure (Figure 7b) requiring unwinding of the first turn
of the αCꢀhelix in the ligandꢀbound structure.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 77
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7a. Monomeric L858R+T790M ligand bound conformation (yellow) vs. EGFR WT active
(
green). A classic DFG out conformation is observed in the ligand bound T790M structure, however the
activation loop is in a state nearly identical to the active form. The Phe856 side chain is oriented out of
the protein core, partially occupying the ATP binding site, and the Asp855 side chain is positioned into
the hydrophobic core.
ACS Paragon Plus Environment

Page 13 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 7b. Comparison of the monomeric L858R+T790M protein conformations in the apo (blue) and
the ligand bound (yellow) states. Rotation of the Nꢀlobe and αꢀC helix occurs on ligand binding.
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
The cocrystal structure suggests that the binding affinity of 10 is achieved through specific and
nonꢀspecific interactions with the protein independent of the covalent bond formation with
Cys797 (Figure 8). The PP core establishes specific donorꢀacceptorꢀdonor interactions with the
backbone atoms of hinge residues 791 – 793. The phenoxy linker occupies the ATPꢀsugar
binding site and forms a van der Waals interaction with Phe856 from the DFG motif. The Gꢀ
loop is well ordered with the Phe723 side chain turned inward toward the ligand binding site to
make van der Waals contact with Phe856. The phenyl ring of the phenylpiperazine group is
engaged in van der Waals interactions with hinge residues 794 – 796, whilst the piperazine ring
projects into the solvent region without any visible specific interactions with the protein. Two
key water interactions are observed; one between nitrogen of the ligand’s propionamide group,
the backbone atom of the Arg841 and the side chain carbonyl of the Asn842, and a second
between the carbonyl of the acrylamide and the backbone of Cys797. The side chain of the
mutated gatekeeper residue Met790 forms hydrophobic interactions with the pyrrole of the PP
core (distance between the beta carbon of Met790 and the C2 carbon atom of the pyrrole ring is
3.7 Å). In contrast, when this same ligand is modeled in the WT enzyme the C2 atom of the
pyrrole comes within 2.7 Å of the hydroxyl group of the gatekeeper Thr790 residue (The
modeling result is discussed in the supporting information section). This unfavorable interaction
is the presumed origin of the high selectivity for drugꢀresistant T790M mutant EGFR over WT.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 77
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Compound 10 complex with monomeric L858R+T790M KD (1.5 Å). Two views are shown.
Hydrogen bonds to the hinge are shown, with key protein residues highlighted in stick form. Some parts
of the protein are omitted for clarity. Clear covalent modification of Cys797 is observed. The ligand
phenoxy group interacts with Phe856 through VDW interactions. Two key waters are also highlighted
(
red spheres).
An edgeꢀface πꢀinteraction exists between the phenyl rings of the anilinopiperazine and the
phenoxy linker which may assist in optimally positioning the acrylamide warhead to enable
reaction with Cys797.
Next, the primary liabilities of 10 were addressed, specifically poor permeability
33,34
ꢀ6
(RRCK
2x10 cm/sec), potent dofetilide binding activity (IC =1 µM) and high clogP (4.6).
50
The highly basic 4ꢀpiperazinyl–aniline was thought to play a significant role in these liabilities.
Replacement of the aniline with an aminoꢀpyrazole isostere was envisioned to provide a less
lipophilic, metabolically stable surrogate which (unlike the aniline) would be less prone to
35
reactive metabolite formation.
Furthermore, the lower lipophilicity of an aminoꢀpyrazole
functionality might enable removal of the basic amine found in the piperazinylꢀaniline resulting
ACS Paragon Plus Environment

Page 15 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
in a more ligandꢀefficient lead. To investigate this possibility, the N1ꢀmethylꢀ3ꢀaminoꢀ and N1ꢀ
methylꢀ4ꢀaminoꢀpyrazoles 11 and 12 were prepared (Table 1). Both compounds demonstrated
increased permeability and decreased dofetilide binding activity relative to 10 while maintaining
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
good EGFR L858R/T790M potency and selectivity over EGFR WT (>100ꢀfold). Ligand
est
efficiency (LE, calculated from K ) for 11 and 12 are 0.41 and 0.45, respectively, which are
i
significantly higher than LE of 0.37 for 10. The 4ꢀaminoꢀpyrazole 12 better matched the EGFR
T790M binding affinity, inactivation rate (k /I), cellular potency and intrinsic chemical
obs
reactivity to GSH of 10 but showed increased microsomal clearance relative to 10 and 11.
Table 1. Summary of Early Leads.
est
a
c
Compoun
d ID
Structure
Ki
kobs/I
H1975
IC50
A549
IC50
WT/
DM
LogD
clogP HLM RRCK
GSH T
1/2
ꢀ1 ꢀ1
b
ꢀ6
(nM)
M s
Cl
(10
relative to
canertinib
(nM)
(µM)
(ꢁL/
min/
g)
cm/sec)
d
1
1
0
1
2
51,300
4
0.79
198
281
2.7
3.0
4.6
2.7
20
2
6
4
(at 0.05
e
µM)
14
4,080
27
7.6
11
10
(at
1
e
µM)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 77
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
1
2
3
37,500
13
1.6
123
3.2
2.7
66
7
7
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
(at 0.05
e
µM)
a
b
c
log D was measured at pH 7.4. Human liver microsomal intrinsic clearance. Permeability measured
with lowꢀefflux MDCKII cell. Chemical reactivity of warhead with glutathione relative to canertinib at
pH 7.4. Inhibitor concentration at which the k /I is calculated.
d
e
obs
The 1.4 Å coꢀcrystal structure of 12 (Figure 9) is similar to that of 10, with the pyrazole
making a similar edgeꢀface interaction with the phenyl ether bearing the warhead. The two CꢀH
bonds on the pyrazole ring in 12 are polarized and the CꢀH at the 3ꢀposition of the pyrazole
forms a weak nonꢀclassical H bond with the backbone carbonyl of Met793 in the hinge. The
lower binding affinity of the 3ꢀaminopyrazole 11 can be explained by the unfavorable interaction
between the pyrazole 2ꢀN and the backbone carbonyl of M793 in an orientation that maintains
the edgeꢀface interaction between the pyrazole and phenyl ether.
ACS Paragon Plus Environment

Page 17 of 77
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 9. Compound 12 binding interactions with the monomeric EGFR L858R/T790M binding site.
Having established the Nꢀmethylꢀ4ꢀaminopyrazole as a suitable isostere for the 4ꢀ
piperazinylꢀaniline, the linker region of the molecule was optimized as a means to reduce
metabolic clearance and further minimize the reactivity of the warhead.
Aliphatic Linkers to Minimize Warhead Reactivity.
The intrinsic chemical reactivity of a warhead is influenced by the electronic properties of the
linker, in this case the aromatic ring from which the warhead is connected. In order to mitigate
the potential of forming offꢀtarget covalent adducts, the design strategy was to minimize the
warhead’s reactivity by avoiding electron deficient linking motifs. It was predicted that utilizing
aliphatic linkers would remove the conjugative effects of aromatic systems (e.g. phenoxy linker
of 12, anilide of canertinib) on the warhead and afford reduced reactivity. A 3ꢀpyrrolidinemethyl
ether linker was identified to have the appropriate length, orientation and conformational
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 77
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
36
constraint to position the warhead for a reaction with Cys797. Compound 13 (Table 2) was
est
synthesized and tested, however, the Ki of 590 nM and negligible irreversible inactivation
ꢀ1 ꢀ1
(
k /I of 830 M s at 1 µM) indicated that significant optimization was required.
obs
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
To increase binding affinity, a small hydrophobic pocket (identified from the coꢀcrystal
structure of 12) under the Gꢀloop, created by residues Phe723 and Val726 was targeted.
Modelling of 13 (using the protein structure of 12 cocrystal structure) suggested that a small
lipophilic group at the pyrrolidine 4 position would be suited (Figure 10), thus a CF group was
3
installed (chosen for metabolic stability benefits over methyl or ethyl) resulting in compound 14
est
(
Table 2). This molecule provided a 17ꢀfold increase in potency (Ki 35 nM) and an increase in
ꢀ1 ꢀ1
irreversible activation (k /I of 5410 M s at 1 µM) relative to 13. However, introduction of
obs
the CF group was concomitant with higher intrinsic reactivity compared to other aliphatic
3
linkers tested with the intrinsic reactivity of 14 relative to canertinib (9x) similar to that observed
with 12 (7x). In order to improve the potency of the aliphatic linker compounds, alternative
opportunities were sought to increase binding affinity using pyrrolidine substituents that do not
increase the intrinsic reactivity.
ACS Paragon Plus Environment

Page 19 of 77
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 10. Comparison of modeled binding mode of 13 (stick, in blue color) in monomeric
L858R/T790M EGFR and bound conformation of 14 (ball and stick, in green color) in cocrystal structure
with the same protein. Cocrystal structure of 14 with monomeric L858R/T790M EGFR demonstrates that
the CF binds in the G loop hydrophobic pocket.
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 77
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
Table 2: Summary of Initial Aliphatic Linker Designs.
RRCK
b
c
H1975
IC50
WT
HLM Cl
GSH
T
1/2
est
Compound
ID
K
i
k
obs/I
WT/D
M
Log
ꢀ6
Structure
IC50
(µM)
(ꢁL/min/
g)
(10
relative
to
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 ꢀ1
a
(nM)
(M s )
D
e
(nM)
cm/sec canertinib
)
830
1
3
590
(at
1
ND
ND
ND
2.1
56
16
ND
e
µM)
HN
N
5
,410
HN
N
O
CF3
(at
1
14
15
35
47
>5
>106
2.7
26
45
11
13
9
N
N
N
e
µM)
O
1
9,100
8
(at 0.1 45
>3.3
>73
2.91
>19
e
µM)
5
1,400
1
6
3
(at 0.05
8
3.5
438
3.0
37
3
9
e
µM)
a
b
c
log D was measured at pH 7.4. d Human liver microsomal intrinsic clearance. Permeability measured
with lowꢀefflux MDCKII cell. Chemical reactivity of warhead relative to canertinib (halfꢀlife in
glutathione buffer at pH 7.4). Inhibitor concentration at which the k /I is calculated.
e
obs
To help understand the binding mode and ligand conformation of 14, its coꢀcrystal structure was
determined at 2.2 Å resolution (Figure 11). Although the crystal structure does not reflect the
transition state conformation for the covalent bond formation between Cys797 and the warhead
ACS Paragon Plus Environment

Page 21 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
acrylamide, it is evident that the conformation of the linker ꢀOCH ꢀ is constrained in an
2
orientation syn to the nitrogen of the pyrimidine. The pyrrolidine occupies the space extending
from the sugar binding pocket to the triphosphate recognition site of the active site, and the
substituents at the 3 and 4ꢀpositions of the pyrrolidine ring adopt a trans diꢀaxial conformation.
Ligand torsional profile analysis suggests that the minimum energy conformation of 14 in
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
solution has the linker ꢀOCH ꢀ coꢀplanar with the core. In contrast, the bound conformation has
2
the ꢀOCH ꢀ rotated out of conjugation to the core, resulting in a strain energy penalty (~1.4
2
kcal/mol, calculated with Amber force field) which is compensated for by the binding energy
gained from the strong interactions between the CF and the protein.
3
Figure 11. Crystal structure of 14 in complex with monomeric L858R/T790M EGFR (2.2 Å). The CF3
group fills a small hydrophobic pocket affording a 17ꢀfold increase in binding affinity. Note interaction
between the CF3 group and Phe723 and Phe526.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 77
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
5
ꢀCl Pyrrolopyrimidine Scaffold Significantly Increases Intrinsic Binding Affinity.
The cocrystal structure of 14 also revealed an opportunity to increase binding affinity via
targeting an unoccupied hydrophobic pocket near the gatekeeper residue. Placing a halogen at
the 5ꢀposition of the pyrrolopyrimidine was predicted to fill this pocket. A Cl substitution
demonstrated significant increases in both binding affinity and irreversible inactivation (Table
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
). Relative to 13 and 14, introduction of 5ꢀCl to afford 15 and 16 yielded a 74ꢀ and 12ꢀfold
est
increase in Ki respectively. In addition to the enhanced binding affinity, a substantial increase
ꢀ1 ꢀ1
in the inactivation rate was achieved, with k /I values of 19,100 and 51,400 M s respectively.
obs
Computational analysis of ligand conformational energy profile predicts that introduction of the
halogen at the 5ꢀposition will further constrain the linker in an optimal conformation to position
the warhead to react with Cys797 (ligand torsional strain energy analysis to support this
hypothesis is described in supporting information section). This, together with additional
hydrophobic interactions with the G loop, yielded consistently potent aliphatic linkers (see
pairwise plot, Figure 12) and provided the opportunity to optimize the linker in the absence of
electron withdrawing groups. The intrinsic reactivity of compound 15 was more than 19ꢀfold
lower than that of canertinib, indicating that high potency, high inactivation rate and low intrinsic
reactivity could all be achieved within the aliphatic linker subꢀseries.
ACS Paragon Plus Environment

Page 23 of 77
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 12. Pairwise comparison between 5ꢀCl and 5ꢀH pyrrolopyrimidine analogs. Chlorine substitution
of the 5ꢀposition of the PP core afforded a 47ꢀfold average increase in binding affinity.
Lead Optimization.
ꢀ6
Compound 15 had high permeability (RRCK 13 x10 cm/sec) with significant cellular activity
(
pEGFR H1975 IC 45 nM) and high selectivity over EGFR WT (>73x). However, the high
5
0
metabolic clearance needed to be addressed in order to achieve a desirable human PK profile.
The cocrystal structure of 14 reveals that the CF binds very tightly in the Gꢀloop hydrophobic
3
pocket, with the CF directly interacting with the Phe723 side chain phenyl ring (Figure 10). As
3
half of the CF moiety was exposed to solvent, it presented an opportunity to modulate metabolic
3
stability by replacing the CF with small polar groups. Based on the outcome from modeling
3
studies and in silico ADME predictions ꢀF, ꢀOMe, ꢀOH, ꢀCH OH and ꢀCH OMe replacements
2
2
for –CF were explored (Table 3).
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 77
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
Table 3: Summary of Lead Optimization.
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
est
ꢀ
b
d
Compound
ID
R
K
i
k
obs/I (M
H197
A549
IC50
WT/D
M
Log
LipE
HLM
RRCK
GSH
T
1/2
1
ꢀ1
a
c
ꢀ6
(nM)
S )
5 IC50
D
(H1975 Cl
(10
relative to
e
(nM)
(µM)
IC50
5.5
)
(ꢁL/mi
n/g)
cm/sec) canertinib
1
4
40,600
13
5.1
392
467
>48
>14
1140
>83
2.5
2.7
2.0
2.3
2.9
2.4
18
10
15
(at 0.05
f
µM)
17
3
43,500
9
4.2
5.4
4.7
3.9
5.3
4.5
28
12
13
23
24
5
15
(at 0.05
f
µM)
1
1
2
2
8
9
0
1
8
24,800
210
710
8
>10
>10
9.1
3
10
(at 0.1
f
µM)
6,310
(at
15
3
ND
>17
>17
1
f
µM)
4
26,800
16
10
(at 0.05
f
µM)
32
2,010
120
>10
(at
1
f
µM)
a
c
log D was measured at pH 7.4. LipE = ꢀ [log(H1975 cell IC )] – log D. Human liver microsomal
5
0
d
e
intrinsic clearance. Permeability measured with lowꢀefflux MDCKII cell. Chemical reactivity of
f
warhead relative to canertinib (halfꢀlife in glutathione buffer at pH 7.4). Inhibitor concentration at
which the k /I is calculated.
obs
ACS Paragon Plus Environment

Page 25 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Pyrrolidine derivatives 1 (PFꢀ06459988), 17ꢀ20 demonstrate potent affinity and high
inactivation rates with an improvement in metabolic stability relative to compound 16. In the
cellular assay, 1, 17 and 20 were highly potent against L858R/T790M DM and highly selective
over WT. Compound 18 and 19, substituted with an alcohol or hydroxymethyl at the 4ꢀposition
of the pyrrolidine ring, exhibited low permeability and consequently reduced cellular potency.
Compound 1 emerged as providing the best balance of overall properties including potency, WT
selectivity, intrinsic chemical reactivity, ADME properties, and the highest LipE. Its enantiomer
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
(21) exhibited a k /I of 2,010 M s , indicating modest to no irreversible inhibition, and was
obs
est
less potent in both the enzyme assay (Ki = 32 nM) and the cellular assay (H1975 cell IC =
50
120 nM). This suggests the optimum alignment of the acrylamide warhead with Cys797 is
critical for the observed high potency for 1.
A 1.9 Å coꢀcrystal structure of 1 in complex with EGFR L858R/T790M DM shows a great deal
of similarity to that of 14 (Figure 13). The substituents at the 3ꢀ and 4ꢀpositions of the
pyrrolidine ring adopt a transꢀdiaxial conformation, which places the polarized methyl of the
methoxy group at the 4ꢀposition in van der Waals contact with the aromatic side chain of
Phe856, which in turn interacts edge to face with the sidechain phenyl of Phe723. In addition,
the Cl substituent at the 5ꢀposition of the PP core interacts with both the Phe856 side chain and
the Met790 side chain through hydrophobic interactions. Ligand strain energy analysis using
3
7
Bachmin V2.0 indicates that the bound conformation of compound 1 possess minimal strain
energy (see SI for details), which is consistent with the very high binding affinity and cellular
potency that it demonstrates against L858R/T790M mutant.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 77
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 13. Crystal structure of 1 in complex with monomeric L858R/T790M EGFR (1.9 Å). The diaxial
conformation of the pyrroldine places the OMe in van der Waal’s contact with both the Phe856 of the
DFG motif and the Phe723 of the Gꢀloop.
Characterization of compound 1.
Since compound 1 emerged as a project lead by having the best overall balance of properties, it
was characterized in detail in both biochemical and cellular assays (Table4, 5, SIꢀ1, SIꢀ2).
Compound 1 is a potent and efficient inhibitor of EGFR L858R/T790M DM (K = 13 nM, kinact
=
i
ꢀ1
ꢀ1 ꢀ1
est
0
.02 s , k /K = 1,530,000 M s ), and shows modest activity toward WT EGFR (K = 1,600
inact
i
i
est
ꢀ1 ꢀ1
nM, kinact/Ki = 4,520 M s , Table SIꢀ1). Its cellular biochemical potency was characterized in
NSCLC cell lines including H1975 (L858R/T790M), PC9ꢀDRH (Del/T790M), H3255 (L858R),
PC9 (Del), HCC827 (Del) and A549 (WT) (Table 4). rociletinib, osimertinib and erlotinib were
also evaluated in the same NSCLC cell lines. Compound 1, rociletinib and osimertinib all
ACS Paragon Plus Environment

Page 27 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
demonstrate very high cellular potency against T790Mꢀcontaining double mutant EGFRs.
Rociletinib and osimertinib demonstrate good cellular potency against the oncogenic mutants
L858R and Del. In comparison, 1 also demonstrates good cellular potency against L858R
mutant, albeit moderate cellular potency against Del mutant in the PC9 cell line. Erlotinib
demonstrates very high cellular potency against both oncogenic mutants and WT EGFR, but is
inactive against T790Mꢀcontaining double mutants. Compound 1 spares WT EGFR, hence
demonstrates the highest selectivity between the least potent target and WT EGFR.
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: Cellular Potency and Selectivity Profile of 1, Rociletinib, Osimertinib and
Erlotinib
WT/DM
(L858R/
T790M)
L858R/
T790M
Del/
T790M
WT/least
potent target
L858R
Del
Del
WT
H1975
IC50 (nM)
13
PC9ꢀDRH
IC50 (nM)
7
9
13
9,700
H3255
IC50 (nM)
21
46
36
10
PC9
IC50 (nM)
140
79
56
HCC827
IC50 (nM)
90
65
23
8
A549
IC50 (nM)
5,100
1,600
150
Compound
Ratio
Ratio
1
392
100
10
0.0012
36
20
3
Rociletinib
16
Osimertinib 15
Erlotinib 9,000
11
11
1
ProteomeꢀWide Selectivity.
Cysteine is the most intrinsically nucleophilic proteinaceous residue, and many protein classes
38
rely on cysteine for function. Irreversible EGFR inhibitors with high intrinsic reactivity for the
warhead have the potential to crossꢀreact, either specifically or nonspecifically, with other
proteins, including those outside the kinome, which may lead to unanticipated toxicities. To
compare the broad proteome reactivity between 1 (GSH reactivity ratio = 15) and canertinib
(GSH reactivity ratio =1), clickable probes of these two compounds were designed to enable
activityꢀbased protein profiling by installing the alkyne at positions that would neither perturb
the intrinsic reactivity of the Michael acceptors nor interfere with the binding to EGFR.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 77
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
Physiochemical and biological properties for the probe for canertinib (22) and the probe for 1
(
23), including cellular IC against wild type and double mutant EGFR, and K /I and GSH
50 obs
reactivities, closely matched the properties for canertinib and 1, respectively (Table 5).
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
Table 5. Comparison of Properties for 1 and canertinib with Their Corresponding Probes.
b
H1975
IC50
RRCK
GSH
T
1/2
est
Ki
K
obs/I
ꢀ1 ꢀ1
WT IC50
(nM)
a
ꢀ6
ID
Structure
LogD
(10
relative to
(
nM)
M s
c
(nM)
cm/sec) canertinib
1
48,000
(at
Canertinib
0.11
3
6
ND
ND
1
0.01
µM)
d
4
8,600
(at
2
2
2
7
15
4.0
0.4
1.2
0.05
µM)
d
4
0,600
(at
1
2
4
13
5,100
4,200
2.5
3.0
10
15
0.05
µM)
d
1
7,300
3
10
(at 0.1 16
7
>17
d
µM)
a
b
c
log D was measured at pH 7.4. Permeability measured with lowꢀefflux MDCKII cell. Chemical
d
reactivity of warhead with glutathione relative to canertinib at pH 7.4. Inhibitor concentration at which
the k /I is calculated.
obs
ACS Paragon Plus Environment

Page 29 of 77
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Initial experiments were performed in the EGFR L858R/T790M mutant H1975 lung
cancer cell line. In situ reactivity of the two probes by gelꢀbased analysis was derived from an
H1975 in vitro experiment, where cells were treated with 1 µM of compound for 15 minutes 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
24 hours before the H1975 cells were harvested (Figure 14). Fractionated cellular lysates were
reacted with azideꢀrhodamine (N ꢀRh) under standard copperꢀcatalyzed azideꢀalkyne
3
39
cycloaddition (CuACC) conditions.
Covalently probeꢀmodified proteins were revealed by
their inꢀgel fluorescence. Both chemical probes displayed increasing amounts of proteome
reactivity in the soluble (Figure 14, top) and particulate (Figure 14, bottom) proteomes as the
length of probe incubation was increased. Compound 23 (probe to 1) displayed a slower,
gradual increase in reactivity compared to 22 (probe to canertinib). The covalently modified
proteins included both overlapping and unique targets. The identification of the covalent targets
will be revealed in a later manuscript. In agreement with their warhead intrinsic chemical
reactivity, considerably more proteome reactivity was observed for 22 than for 23. This was
especially apparent at a 1 hour timepoint in H1975 cells (Figure SIꢀ1A in SI) where a more than
50% reduction in total offꢀtarget labeling signal was observed (Figure SIꢀ1B in SI). Similarly,
more offꢀtarget labeling in H3255 cells was observed, where H3255 expresses only the L858R
single mutant form of EGFR (Figure SIꢀ1C in SI).
Thus, by minimizing the intrinsic chemical reactivity of the electrophilic warhead, 1
demonstrates significantly increased proteome wide selectivity over canertinib. This is likely
one of the reasons that only minimal findings were observed from toxicity studies with 1, with
ACS Paragon Plus Environment