Discovery of BAY-985, a Highly Selective TBK1/IKK? Inhibitor

Article abstract of DOI:10.1021/acs.jmedchem.9b01460

The serine/threonine kinase TBK1 (TANK-binding kinase 1) and its homologue IKK? are noncanonical members of the inhibitor of the nuclear factor κB (IκB) kinase family. These kinases play important roles in multiple cellular pathways and, in particular, in inflammation. Herein, we describe our investigations on a family of benzimidazoles and the identification of the potent and highly selective TBK1/IKK? inhibitor BAY-985. BAY-985 inhibits the cellular phosphorylation of interferon regulatory factor 3 and displays antiproliferative efficacy in the melanoma cell line SK-MEL-2 but showed only weak antitumor activity in the SK-MEL-2 human melanoma xenograft model.

Full text of DOI:10.1021/acs.jmedchem.9b01460

Subscriber access provided by BIU Pharmacie | Faculté de Pharmacie, Université Paris V
Article
Discovery of BAY-985 a highly selective TBK1/Ikk# inhibitor
Julien Lefranc, Volker K. Schulze, Roman Christian Hillig, Hans Briem, Florian
Prinz, Anne Mengel, Tobias Heinrich, József Bálint, Srinivasan Rengachari, Horst
Irlbacher, Detlef Stöckigt, Ulf Bömer, Benjamin Bader, Stefan Nikolaus Gradl, Carl F.
Nising, Franz von Nussbaum, Dominik Mumberg, Daniel Panne, and Antje Wengner
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b01460 • Publication Date (Web): 20 Dec 2019
Downloaded from pubs.acs.org on December 20, 2019
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 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 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.
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 BAY-985 a highly selective
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
TBK1/IKKε inhibitor
Julien Lefranc,*,# Volker Klaus Schulze,# Roman Christian Hillig,# Hans Briem,# Florian Prinz,#
Anne Mengel,# Tobias Heinrich,# Jozsef Balint,║ Srinivasan Rengachari,§,† Horst Irlbacher,#
Detlef Stöckigt,# Ulf Bömer,# Benjamin Bader,# Stefan Nikolaus Gradl,# Carl Friedrich Nising,#
Franz von Nussbaum,# Dominik Mumberg,# Daniel Panne,§,‡ Antje Margret Wengner#
#Bayer AG, Pharmaceuticals, Research and Development, 13353 Berlin, Germany
║ASCA GmbH (Angewandte Synthesechemie Adlershof), 12489 Berlin, Germany
§Leicester Institute of Structural and Chemical Biology, Department of Molecular and Cell
Biology, University of Leicester, Lancaster Road, Leicester, LE1 7RH, U. K.
‡European Molecular Biology Laboratory, 38042 Grenoble, France
ABSTRACT. The serine/threonine kinase TBK1 (TANK-binding kinase 1) and its homologue
IKKε are non-canonical members of the inhibitor of nuclear factor κB (IκB) kinase family. These
kinases play important roles in multiple cellular pathways and, in particular, in inflammation.
Herein, we describe our investigations on a family of benzimidazoles and the identification of the
potent and highly selective TBK1/IKKε inhibitor BAY-985. BAY-985 inhibits the cellular
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 43
1
2
3
4
5
6
7
8
9
phosphorylation of interferon regulatory factor 3 (IRF3) and displays antiproliferative efficacy in
the melanoma cell line SK-MEL-2, but showed only weak antitumor activity in the SK-MEL-2
human melanoma xenograft model.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
INTRODUCTION
The serine/threonine kinase TBK1 (TANK-binding kinase 1) and its homologue IKKε are non-
canonical members of the inhibitor of nuclear factor κB (IκB) kinase family. These kinases play
important roles in multiple cellular pathways and, in particular, in inflammation.1,2 IKKs are
activated following engagement of pattern recognition receptors such as toll-like receptors (TLRs),
cytosolic DNA receptors, and retinoic acid inducible gene I (RIG-I)-like receptors. Activated non-
canonical IKKs phosphorylate interferon regulatory factors (IRFs) 3, 5, and 7, which then dimerize
and enter the nucleus to induce expression of pro-inflammatory and antiviral genes.1
TBK1 also plays an important role in autophagy,3,4 a quality control process in which the cell
recycles proteins or disposes of excessive or damaged organelles and which contributes to the
clearance of intracellular pathogens.5-9 Furthermore, TBK1 maintains T cell homeostasis and
negatively regulates both autoimmunity and antitumor immunity by promoting immune tolerance
in dendritic cells, likely by regulating a subset of type I interferon receptor induced genes.10
In view of its critical role in inflammatory signaling and oncology, TBK1 has attracted
considerable interest as a potential target in cancer therapy. TBK1 is overexpressed and activated
in bladder, lung, breast, and colon cancers.11–13 Knockdown experiments have shown that TBK1
is crucial for cellular survival and synthetically lethal with mutated KRAS in lung cancer.14 The
RalB/Sec5 effector complex, downstream of KRAS signaling, directly recruits and activates
TBK1, thus preventing induction of apoptosis.15 TBK1 also regulates an autocrine circuit of
ACS Paragon Plus Environment
2

Page 3 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CCL5/RANTES and interleukin 6 to promote KRAS-driven tumorigenesis.16 TBK1 is further
thought to promote vascularization,13 to directly drive oncogenesis by phosphorylating and
activating Akt1,11,17 and to regulate prostate cancer dormancy by inhibition of mTOR.18
Although a number of studies suggest that TBK1 is a potentially interesting target in cancer
therapy, recent data have challenged this view.19,20 Considering that most studies were performed
using relatively unselective TBK1 inhibitors or by protein depletion, it remains unclear if selective
kinase inhibition of TBK1 is a viable therapeutic strategy. Clinical studies targeting TBK1 in
KRAS-mutant cancers using momelotinib, a compound that inhibits both Janus kinases and TBK1,
have been terminated early without release of further information.21 Over the last decade, several
attempts have been made to identify TBK1/IKKε inhibitors; however, it has become apparent that
the identification of such highly selective inhibitors is very challenging.22,23
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
TBK1 is an 84 kD, 729 amino acid protein comprising an N-terminal kinase domain, an
ubiquitin-like domain, an α-helical scaffold dimerization domain, and a C-terminal adaptor
binding domain (CTD).24 Depending on the context, cell type, and stimulus, the CTD binds to a
number of different adaptors including TANK, NFκB activating kinase (NAK)-associated protein
1 (NAP1), similar to NAP1 TBK1 adaptor (SINTBAD), and Optineurin.25–27 Thus, in addition to
its enzymatic activity, TBK1 also has an important scaffolding function, and an unravelling of the
relative contributions to cellular signaling is highly desirable. To that end, we have developed
BAY-985, a highly potent and selective TBK1 inhibitor. BAY-985 has been donated as a probe
molecule to the Structural Genomics Consortium (SGC) to enable access to the scientific
community for future studies of TBK1 and its function. Similarly, selective inhibitors of other
potential drug targets have proven valuable tool compounds for the scientific community,28 and
we anticipate that BAY-985 will broaden the understanding of the biology of TBK1.
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 43
1
2
3
4
5
RESULTS AND DISCUSSION
6
7
8
9
HTS. We first performed a TBK1 high-throughput screening (HTS) campaign using a
biochemical kinase assay as an entry point. Due to the limited set of compounds identified with
this approach, we performed a second HTS, this time with a TBK1-dependent cellular reporter
gene assay. This second approach was expected to allow identification of compounds that inhibit
TBK1 in its cellular environment, thus potentially revealing novel modes of action. The reporter
cell line expressed luciferase under the control of multiple interferon-stimulated response
elements (ISREs). Co-incubation with poly dA:dT stimulates TBK/Ikke driven ISRE activation
and hence luciferase expression. In the HTS we screened 3.05 million compounds for their
ability to inhibit the luciferase signal by more than 40%. Screening hits were tested for toxicity
and unspecific luciferase or transcriptional/translational inhibition (see also supporting
information). They were further characterized using two different biochemical assays to measure
TBK1 inhibition in vitro (referred to as ‘low ATP assay’ and ‘high ATP assay’). These two
assays varied in the ATP concentration (10 µM, low ATP assay; 1 mM, high ATP assay), where
a high potency in the high ATP assay is desirable as this increased ATP concentration reflects
more realistically the situation in the cell and in vivo. The cellular HTS campaign allowed us to
identify compound 1 (Figure 1A) which was ATP competitive in the biochemical assays.
X-ray Crystallography and Initial Optimization. Starting from the HTS hit 1, we explored
the 6-position of the benzimidazole moiety with regard to potency and selectivity toward potential
off-target kinases, most importantly the CDK family of kinases. TBK1 contains a methionine
(Met86, see Figure 1B) at the so-called gatekeeper position, the residue immediately preceeding
the hinge which connects the N- and C-terminal lobe of the kinase domain. This feature can be
exploited to develop potentially selective inhibitors as CDKs contain a phenylalanine at this
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
4

Page 5 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
position. Guided by a co-complex crystal structure published for a similar 6-cyanobenzimidazole
inhibitor binding to the ATP pocket of the kinase CK1, which also contains a methionine
gatekeeper,29 we predicted that the cyano group of compound 1 would not interact with and
modulate the side-chain conformation of the Met gatekeeper of TBK1. Potentially, this could
explain the low selectivity of compound 1 toward kinases which feature a bulky Phe gatekeeper,
such as CDKs (data not shown). As has been shown for JNK3 kinase inhibitors, selectivity can be
improved by exploiting the high flexibility of a Met gatekeeper side chain.30–32 An appropriate
substituent on the inhibitor is expected to push away the Met side chain and to open up the so-
called ‘hydrophobic back pocket’, whereas a more rigid Phe side chain at the gatekeeper position,
such as in CDKs, would restrict access to this back pocket. Based on this rationale, we synthesized
compound 2 where the 6-cyano side chain of 1 was replaced by a bulkier N-cyclopropylmethyl
amide group. While this substitution only modestly improved potency (see Table 1, TBK1 low
ATP assay), it resulted in a suprisingly drastic improvement in selectivity toward the off-target
kinase CDK9 (IC50 = 5 nM for 1 vs >20000 nM for 2).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To understand the binding mode of this compound class, we co-crystallized three compounds
with close to full-length TBK1S172E24 (see Supporting Information, Table S4. TBK1 only
crystallized when ATP site inhibitors were added prior to crystal setup. The vast majority of
crystals diffracted to a minimum Bragg spacing of 4–5 Å which required testing of 20–50
crystals/compound to identify a specimen that diffracted to better than 3.5 Å resolution. We
collected a 3.15 Å data set for compound 2 and determined the structure by molecular replacement.
Despite the modest resolution, omit maps of the ATP site of the kinase domain showed clear
difference density, which allowed modeling of the ligand (Figure 1B). The inhibitor inserts into
the flat and mainly hydrophobic adenine pocket of the ATP-binding site at the hinge region where
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 43
1
2
3
4
5
6
7
8
9
the central conjugated bi-aromatic ring system is sandwiched between Leu15, Val23, and Ala33
at the roof of the ATP site, and Gly92 and Met142 at its floor. The inhibitor forms two hydrogen-
bond interactions to Cys89 of the hinge region of TBK1, one via the central amino group (3.1 Å)
and one via a nitrogen atom of the benzimidazole ring (3.5 Å). The piperazine moiety points
toward the solvent, where the amide oxygen forms a 3.5 Å hydrogen bond to the side chain of
Arg25. The cyclopropyl moiety inserts into the back pocket behind the ATP site and forms a
hydrophobic contact with the gatekeeper residue Met86. The nitrogen atom of the adjacent amide
group forms a hydrogen bond to Thr156 (2.7 Å). However, compound 2 does not interact with the
conserved Lys38 at the back wall of the ATP site nor with Asp157 (of the DFG motif at the start
of the activation loop), two polar and conserved kinase residues which line this back pocket behind
the ATP site (Figure 1B).
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
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 1. (A) Compound 1 identified during our cellular HTS and compound 2 as an early variant.
(B) Crystal structure of TBK1 in complex with 2 (PDB accession code 6RSR). Left, overall view
of one TBK1 monomer with the kinase domain in cyan (N-terminal lobe) and blue (C-terminal
lobe), the ubiquitin-like domain (ULD) in yellow, and the scaffold dimerization domain (SDD) in
green. Right, 2 bound to the hinge region within the ATP site of the kinase domain. The 3.1 Å
resolution mFo-DFc electron density omit map contoured at 2  is shown in blue.
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 43
1
2
3
4
5
6
7
8
A comparison of the binding mode of 2 with the less selective TBK1 inhibitor BX-795 (PDB
accession code 4EUU)33 revealed that, indeed, the gatekeeper Met86 of TBK1 can accommodate
the cyclopropyl moiety of inhibitor 2 by adopting an alternative side-chain conformation (Figure
2) relative to the one observed with BX-795 which does not occupy the back pocket.
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. Conformational flexibility of the TBK1 gatekeeper Met86. Superimposition of the
kinase domains of TBK1 in complex with 2 (PDB accession code 6RSR, green carbon atoms) and
BX-795 (PDB accession code 4EUU, orange carbon atoms). View into the hinge region of the
ATP-binding site. Hydrogen-bond interactions between the inhibitors and TBK1 shown as yellow
dotted lines. The gatekeeper residue Met86 can accommodate back-pocket binders such as 2 by
adopting a new side-chain conformation (red arrow).
Compounds 1 and 2 showed low permeability and strong efflux in Caco-2 assays: Papp A–B = 32
nm/s, Papp B–A = 268 nm/s, ER = 8.3 and Papp A–B = 3.9 nm/s, Papp B–A = 155 nm/s, ER = 40,
ACS Paragon Plus Environment
8

Page 9 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
respectively. We further explored the 6-position of the benzimidazole moiety with the aim of
improving selectivity and potency. We focused our attention on small heterocycles that could
potentially be bioisosteric to an amide side chain. In addition to the desired shift of the Met86
gatekeeper side chain, computational design focused on the addition of a possible hydrogen bond
with the catalytic Lys38, which is in close proximity to the gatekeeper (Figure 1B). This led to the
design of compound 3 where the amide was replaced by its 1,2,4-oxadiazole bioisostere. This
compound inhibited TBK1 and IKKε with IC50 values of 129 nM (low ATP assay) and 168 nM,
respectively (see Table 1). Compound 3 also showed improved permeability and efflux ratio in the
Caco-2 assay (Papp A–B = 42 nm/s, Papp B–A = 142 nm/s, ER = 3.4) compared to compound 1.
In parallel, we tested several of our more selective TBK1 inhibitors for antiproliferative activity
in more than 100 cell lines. Contrary to results we had obtained in RNAi experiments (data not
shown), most of the cell lines tested did not respond to these TBK1 inhibitors. Among the cell
lines with higher sensitivity, we selected two cell lines, ACHN (CDKN2A mutated) and SK-MEL-
2 (NRAS and TP53 mutated),34 as models to further determine the antiproliferative activity.
Compound 3 showed micromolar antiproliferative activity in both cell lines (Table 1).
The selectivity of 3 was further assayed in an internal kinase panel and at 100 nM in the
DiscoverX KINOMEscan (see Figure 3 and Supporting Information Table S1). This analysis
revealed low kinase selectivity, with eight other kinases in the panel being more potently inhibited
than TBK1.
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
SAR. Starting from compound 3, we investigated the amide substitution on the piperazine
moiety (Table 1). Synthetic precursors of 3, namely compounds 4 and 5, with a Boc protecting
group or presenting an unsubstituted piperazine ring, were tested and showed reduced potency in
the TBK1 and IKKε assays. Alkylation of the nitrogen also led to reduced potency compared to
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 43
1
2
3
4
5
6
7
8
9
compound 3: N-propyl as well as N-hydroxyethyl substitution (compounds 6 and 7) resulted in
only micromolar inhibition of TBK1 in both the low and high ATP assays. Compounds 8–10 with
a fluorinated alkyl substituent were also assayed: 8 inhibited TBK1 at an IC50 value of 281 nM in
the low ATP assay and at 5290 nM in the high ATP assay; however, although 8 showed no
antiproliferative activity in SK-MEL-2 cells, it inhibited proliferation in ACHN cells with an IC50
of 350 nM. Compound 10 bearing a shorter trifluoromethylated side chain behaved similarly to 8
in the TBK1 assay, while difluoromethyl analogue 9 had slightly higher potency (IC50 = 98 nM,
low ATP assay) and a moderate antiproliferative activity, with an IC50 of 1.6 µM in SK-MEL-2
cells. We further investigated amide substituents (compounds 11–14). Aromatic amides such as
furan analogue 14 showed reduced potency toward TBK1; however, compounds 11–13, each
bearing a small cycloalkyl substituent, proved to be as potent as compound 3 and didn’t show
major differences with regards to Caco-2 permeability (Papp A–B = 68 nm/s, Papp B–A = 151 nm/s,
ER = 2.2 for 11, Papp A–B = 52 nm/s, Papp B–A = 124 nm/s, ER = 2.4 for compound 12) . Despite
substantial efforts, no substituent superior to the initial trifluoroethylcarbonyl group was found;
therefore, this group was retained throughout the rest of our investigations.
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. SAR at the piperazine ring (Targeting the Entrance of the ATP-Binding Site)a
H
N
H
N
N
N
N
N
NH
N
N
R1
NH
N
N
O
F
F
F
N
R
O
N
1 (R = CN)
2 (R =
3 - 14
)
N
H
O
ACS Paragon Plus Environment
10

Page 11 of 43
Journal of Medicinal Chemistry
1
2
3
4
Compd
R1
IC50 (nM)
5
6
7
TBK1
(low/high ATP)
IKKε
pIRF3
SK-MEL-2c ACHN
8
9
BX-795
8/45
18
17
n.d.
2030
n.d.
n.d.
n.d.
n.d.
n.d.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2
–
–
29/618
93/1390
80
3360
2180 b
n.d.
>33000
9670
n.d.
F
F
3
129/2270
273/>20000
764/14700
168
914
940
n.d.
n.d.
1170
730
n.d.
2750
n.d.
F
O
O
O
4
H
5
n.d.
n.d.
n.d.
b
b
6
1440 /16700
n.d.
n.d.
n.d.
b
b
OH
7
1060 /5110
n.d.
n.d.
n.d.
F
F
8
281/5290
98/2040
5160 b
4780 b
>30000 b
>30000 b
1600 b
>30000 b
350 b
381 b
>30000 b
F
F
9
F
F
F
F
10
188/3740
11
12
13
148/1270
115/1920
146/2660
154
226
433
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
O
O
O
429/5070 b
n.d.
n.d.
n.d.
n.d.
O
14
O
a IC50 values are reported as arithmetic means of several measurements (except stated otherwise);
n.d.: not determined. single measurement. IC50 reported at the median value of several
measurements.
b
c
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 43
1
2
3
4
5
6
7
8
9
Next, we investigated the benzylic position. The introduction of a methyl group at that position
had a significant impact on potency (Table 2). While racemic compound 15 had an IC50 value of
34 nM in the TBK1 low ATP assay, testing of the separated enantiomers revealed that one antipode
is superior: (S)-configured 16 inhibited TBK1 with an IC50 value of 106 nM (low ATP assay),
while (R)-configured 17 displayed a fivefold increase in potency (IC50 = 19 nM, low ATP assay;
372 nM, high ATP assay). Compound 17 was also highly potent toward IKKε (IC50 = 37 nM) and
was active in our cellular mechanistic pIRF3 assay and SK-MEL-2 antiproliferation assay, with
IC50 values of 1.2 μM and 3.5 μM, respectively. An additional benefit of the introduction of a
substituent at the benzylic position was the increased kinase selectivity relative to the unsubstituted
analogue (see Table 3, 17 vs 3). In contrast, replacement of the benzylic methylene by a carbonyl
group (compound 18) significantly reduced potency toward both TBK1 and IKKε. Interestingly,
compound 16 and 17 showed improved permeability compare to 3 in our Caco-2 assay with
Papp A–B = 114 nm/s, Papp B–A = 79 nm/s, ER = 0.7 and Papp A–B = 108 nm/s, Papp B–A = 86
nm/s, ER = 0.8 respectively.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 2. SAR at the Benzylic Positiona
R
H
N
N
N
NH
N
N
F
F
O
F
N
O
N
Compd
R
IC50 (nM)
ACS Paragon Plus Environment
12

Page 13 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
TBK1
(low/high ATP)
SK-MEL-2
IKKε pIRF3
ACHN
b
6
7
8
9
3
H
129/2270
168
593
200
37
2750
2270
3020
1210
2180
n.d.
9670
n.d.
15
16
17
18
Me (rac)
(S)-Me
(R)-Me
C=O
34/829
106/2490
19/372
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
n.d.
n.d.
3530
>25000
8980
>30000
1210/>20000
3880 >10000
a IC50 values are reported as arithmetic means of several measurements; n.d.: not determined. b
IC50 reported at the median value of several measurements.
After some preliminary work on the benzimidazole core (data not shown), our further efforts
were focused on the substituent at the 6-position, as summarized in Table 3. We had previously
established that the introduction of a methyl group at the benzylic position (with (R) configuration)
improved the kinase selectivity profile of our compounds (cf. 3 and 17); however, compound 17
still had micromolar off-target activity toward kinases such as CDK9 or RSK4, and also nanomolar
activity toward FLT3 and DRAK1 (with IC50 values of 30 and 10 nM, respectively). We
hypothesized that a bulkier substituent on the oxadiazole ring might help to improve selectivity.
Compound 19, bearing a cyclopropyl group, and compound 20, substituted with an isobutyl group,
showed improved potency compared to 3 toward TBK1 in the low ATP assay (IC50 = 58 and 72
nM, respectively); however, both 19 and 20 were inactive in the high ATP assay (see Table 3).
Replacement of the oxadiazole moiety by an N-methylpyrazolyl group (21) resulted in a slight
increase in potency in the low ATP assay (IC50 = 55 nM, compared to 129 nM for 3). We next
introduced bulkier substituents at a pyrazole nitrogen: cyclopropylmethyl-substituted 23 showed
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 43
1
2
3
4
5
6
7
8
9
a 10-fold increased potency toward TBK1 compared to the methyl and ethyl analogues (21/22).
Our previous experience that the introduction of a methyl group at the benzylic position increased
potency was confirmed with (R)-configured 24 which had a potency of 1 nM (IC50) and 6 nM in
the TBK1 low ATP assay and the IKKε assay, respectively, and a very high potency (IC50 = 18
nM) in the TBK1 high ATP assay. Unfortunately, the introduction of a pyrazole had a negative
impact on Caco-2 permeability. Compounds 22, 23 and 24 showed lower permeability and higher
efflux with Papp A–B = 17 nm/s, Papp B–A = 173 nm/s, ER = 10 for compound 22, Papp A–B = 32
nm/s, Papp B–A = 125 nm/s, ER = 3.9 for 23 and Papp A–B = 16 nm/s, Papp B–A = 118 nm/s,
ER = 7.5 for 24.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 3. SAR at the 6-position of the benzimidazole core (Targeting the Back Pocket)a
R
H
N
N
N
NH
N
N
F
F
R1
O
F
Compd
R1
R
H
IC50 (nM)
ACHN
TBK1
(low/high
ATP)
SK-
MEL-2
IKKε pIRF3
CDK9 FLT3 RSK4 DRAK1 ULK1
N
N
3
129/2270
19/372
168
37
2750
1210
2180b
3530
n.d.
9670
8980
n.d.
455
1520
n.d.
37
30
8910
2360
n.d.
9c
n.d.
n.d.
n.d.
n.d.
n.d.
O
O
N
N
(R)-
Me
17
19
20
21
10c
n.d.
n.d.
n.d.
N
58/>20000
>30000
H
H
H
41
69
b
b
O
N
N
72/>20000
89
n.d.
n.d.
n.d.
n.d.
>20000
240 b
173
11
n.d.
O
N
55 b /2100 b 197 b
1960 b
n.d.
11200
N
N
ACS Paragon Plus Environment
14

Page 15 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
22
23
24
25
H
H
50/530
5/137
1/18
98
7
1580
368
131
93 b
n.d.
221 b
408
n.d.
1460
9190
17
9
8850
4120
3220
325
n.d.
n.d.
5c
n.d.
28d
9d
N
N
N
7900
5270
5420
N
N
N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(R)-
Me
6
>20000
>20000
18
357
N
N
(R)-
Me
2/6
4
3710 b
15d
n.d.
26
27
28
29
H
H
H
54/391
23/388
15/434
2/17
15
35
16
4
840 b
1260 b
773
n.d.
337 b
172
n.d.
71
263
320
2130
0.3
1
1060
2830
434
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
980d
N
2240 b
15300
2010
N
N
N
9
N
N
(R)-
Me
140
476
10
170
N
N
(R)-
Me
30d
368d
416d
N
30
31
0.7/7
2/5
2
4
63
66
883
470
13400
3870
5110
5
192
O
N
N
(R)-
Me
16100
13
1260
n.d.
O
N
32
33
H
H
10/155
9/167
10
10
1990b
213
245
246
29800b
9690
n.d.
6
177
528
25d
1640d
N
HN
N
>10000d
N
14400
114
252d
N
N
34
(BAY-
985)
(R)-
Me
N
2/30
2
74
900
7260
n.d.
123
276
311d
7930d
N
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 43
1
2
3
4
5
N
(R)-
Me
35
O
2/50
n.d.
119
3970 b
6330
7960
824
n.d.
1
7
n.d.
n.d.
263d
n.d.
N
6
7
8
9
N
36
(BAY-
440)
1140/>200
00
3890 b
1110 b
536
6110
Cl
N
H
1280
N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a IC50 values are reported as arithmetic means of several measurements (except stated otherwise);
n.d.: not determined. b single measurement. c Kd DiscoverX. d Eurofins IC50 values.
We obtained a co-crystal structure of TBK1 bound to compound 24 (see Supporting Information,
Figure S1). As found for 2 (Figure 1), compound 24 adopts an overall similar binding mode within
the ATP-binding site, with two hydrogen-bond interactions to residue Cys89 in the hinge region
of TBK1. The (R)-configured benzylic methyl group appears to improve the potency by stabilizing
the outer piperazine moiety in a conformation which allows both insertion of the trifluoropropionyl
amide moiety into a surface indentation above the entrance to the ATP site and formation of a
hydrogen bond between a piperazine nitrogen and the backbone carbonyl oxygen of Pro90 (see
Supporting Information, Figure S1). This conformation also enables interaction of the propionyl
amide with Arg25 and brings one of the fluoro atoms into a beneficial distance to the amide
nitrogen of Ile14.
Compound 24 was highly potent in the cellular mechanistic IRF3 phosphorylation assay and
showed nanomolar antiproliferative activity in SK-MEL-2 cells but only a micromolar effect on
ACHN cells (Table 3). Here as well, introduction of the benzylic substituent enhanced the
selectivity profile. However, compared to the unsubstituted benzylic analogue 23 (with lower
selectivity), 24 showed a twofold reduction in our SK-MEL-2 antiproliferation assay despite an
almost threefold improvement in the cellular mechanistic pIRF3 assay, suggesting that some of
ACS Paragon Plus Environment
16

Page 17 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
the antiproliferative effect observed might not be only due to TBK1/IKKε inhibition but rather to
inhibition of an as-yet-unidentified off-target kinase.
7
8
9
Compound 24 was further profiled at 100 nM in the DiscoverX KINOMEscan (Figure 3 and
Supporting Information Table S2). In this assay, 24 showed an improved selectivity profile
compared to 3. Only two other kinases were inhibited more potently than TBK1 (DRAK1 and
DRAK2), and two kinases (YSK4 and ULK1) were inhibited as potently as TBK1. Further
attempts to improve the selectivity revealed that increasing the steric bulk of the substituent at the
pyrazole nitrogen (compound 25) led to a similar profile with improved selectivity toward FLT3
(Table 3).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. DiscoverX KINOME trees for compound 3,24 and 34 at 100 nM. The red dots highlight
the kinases that show an inhibition of 65 % or more (for more completelist of the kinases see SI
Tables S1,S2 and S3).
While compounds 26 and 27, bearing a pyridyl substituent at the 6-position, retained potency
toward TBK1, their selectivity was lower, especially regarding CDK9 and FLT3 (Table 3).
Relative to 26, pyrimidinyl-substituted 28 displayed a moderate increase in potency in the
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 43
1
2
3
4
5
6
7
8
9
biochemical TBK1 assay (low ATP). This compound, however, showed a different and improved
selectivity toward CDK9 (IC50 = 320 nM vs 71 nM for 26). Again, the introduction of an (R)-
configured benzylic methyl group improved potency and selectivity: compound 29 showed high
potency toward both our targets, TBK1 and IKKε (IC50 = 2 nM, low ATP assay, and 4 nM,
respectively), and a potency of 140 nM (IC50) in the mechanistic pIRF3 assay. As this was the first
time that selectivity toward ULK1 (50-fold) was observed, we investigated a broader set of
pyrimidine derivatives. Introduction of a bulkier substituent on the pyrimidine moiety (compounds
30 and 31) resulted in very high potency in our biochemical assays as well as in the pIRF3 assay,
along with some antiproliferative effect on SK-MEL-2 cells. As seen before, however, the potency
increase achieved did not translate into an antiproliferative effect on ACHN cells therefore
challenging the hypothesis that the antiproliferative effect observed was due to TBK1/IKKε
inhibition. Despite the good selectivity profile of these compounds, they were still active toward
FLT3 and DRAK1. We observed as well that the change from pyrazole to pyrimidines didn’t
improve Caco-2 permeability with for example compound 30 showing low permeability and strong
efflux (Papp A–B = 14 nm/s, Papp B–A = 218 nm/s, ER = 15). The introduction of an amino
substituent on the pyrimidine moiety (compounds 32 and 33) led to improved selectivity toward
ULK1. The dimethylamino substituent (33) seemed to have a strong influence on FLT3 and
DRAK1 potencies. Combining this substituent with an (R)-methyl group at the benzylic position
led to the identification of BAY-985 (34). Compound 34 showed high potency toward TBK1 and
IKKε (IC50 = 2 nM, low ATP assay; 30 nM, high ATP assay), as well as high potency in the
mechanistic pIRF3 assay (IC50 = 74 nM), and an antiproliferative effect on SK-MEL-2 cells (IC50
= 900 nM). Biochemically, 34 also displayed good selectivity toward FLT3 (60×), RSK4 (138×),
DRAK1 (155×), and ULK1 (~4000×) (Table 3). To confirm the improved selectivity of 34, it was
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
18

Page 19 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
further profiled at 100 nM in the DiscoverX KINOMEscan (see Figure 3 and Supporting
Information Table S3). Compound 34 showed high selectivity with respect to all kinases tested in
the panel, except YSK4. Compound 34 still showed low permeability and strong efflux in the
Caco-2 assay (Papp A–B = 9 nm/s, Papp B–A = 146 nm/s, ER = 17).
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
Further investigations on the pyrimidine family allowed us to also identify compound 35, a close
derivative of 34 in which the pyrimidine moiety is substituted with an isopropoxy group (Table 3).
This compound showed high potency toward TBK1, but reduced selectivity, highlighting the
importance of the dimethylamino substituent for the selectivity. Compound 36 Bearing a Cl on the
pyrimidine ring turned out to be a lot less active. We were able to obtain a co-crystal structure of
compound 35 with TBK1 (Figure 4). In addition to the double hinge interaction of the
aminobenzimidazole moiety to Cys89 of the hinge region, the new pyrimidine moiety enables
formation of a hydrogen bond to the conserved lysine residue (Lys38) at the back wall of the ATP
site. As observed for compound 24 (Supporting Information, Figure S1), the (R)-configured
benzylic methyl group helps to lock the outer piperazine moiety in a conformation which enables
insertion of the trifluoropropionyl amide moiety into a surface pocket above the entrance to the
ATP site, with hydrogen bonds to the side chain of Arg25, the backbone nitrogen of Ile14, and the
carbonyl group of Pro90. Thus, compound 35 successfully combines both SAR insights from the
previous optimization steps: the pyrimidine substituent is bulky enough to fully occupy the back
pocket and push the Met86 side chain in a similar way as seen with compound 2, which induces
selectivity toward kinases with more bulky and rigid gatekeeper residues, such as CDKs;
additionally, the pyrimidine is able to form a hydrogen bond to Lys38 in the back pocket (Figure
4). Although compound 35 retains some residual CDK9 activity (IC50 = 824 nM), the TBK1
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 43
1
2
3
4
affinity was further improved with respect to compound 2 (IC50 = 2 nM vs 93 nM in lw ATP
5
6
assay).
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
20

Page 21 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 4. Crystal structure of TBK1 in complex with 35 (PDB accession code 6RST). Left, view
into the ATP site of TBK1, with the protein shown in surface representation (transparent). Right,
view rotated and surface omitted for clarity. mFo-DFc electron density omit map contoured at 2 
shown around the ligand. Hydrogen-bond interactions between the inhibitor 35 and TBK1 shown
as yellow dotted lines.
Pharmacokinetics. The in vitro metabolic clearance profile [CLb,in vitro(rat hepatocytes) = 1.7
L/h/kg, CLb,in vitro(human liver microsomes) = 0.50 L/h/kg] and CYP450 inhibition potential
(CYP1A2, CYP2D6, CYP3A4; IC50 >20 µM) of compound 34 encouraged further in vivo
evaluation. To this end, 34 was dosed iv/po at 0.30/0.30 mg/kg to male Wistar rats. Compound 34
demonstrated a high clearance (CLb = 4.0 L/h/kg, ca. 95% hepatic extraction) which was higher
than predicted from both rat hepatocytes and rat liver microsomes (CLb,in vitro = 1.7 L/h/kg). The
volume of distribution at steady state was large (Vss = 2.9 L/kg) with a short terminal half-life
(t1/2 = 0.79 h). Following oral administration, 34 showed low bioavailability (F = 11%), consistent
with in vivo blood clearance (Figure 5).
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 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 5. Overall profile for BAY-985 including pharmacological and physicochemical
properties as well as in vitro DMPK properties.
Following administration of a single oral dose of 34 at 100 mg/kg to female NMRI nude mice
the exposure was low to moderate, without hints of accumulation after repeated q.d. (once daily)
administration. Biochemical IC50,u(TBK1) and cell-based IC50,u(TBK1/IKKε) were covered by
plasma exposure for 10 hours and 11 hours, respectively. However, proliferative IC50,u(SK-MEL-
2) was ca. threefold higher than Cmax,u
.
In Vivo Antitumor Efficacy Study. The in vivo efficacy and tolerability of 34 in the cell line
derived SK-MEL-2 human melanoma xenograft model in female NMRI nude mice was
investigated. The maximal tolerated dose of 34 was 200 mg/kg applied twice daily (b.i.d.)
continuously. Treatment resulted in weak antitumor efficacy with a T/Crel.area ratio of 0.8 and
T/Ctumor weight ratio of 0.6, as determined at the study termination (day 111 after tumor
ACS Paragon Plus Environment
22

Page 23 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
inoculation) after 35 treatment days (Figure 5A and 5B). The treatment was well tolerated, with a
maximum body weight loss of less than 10% (Figure 5C).
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
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 5. Antitumor efficacy of 34 as monotherapy in the SK-MEL-2 human melanoma xenograft
model in female NMRI nude mice. Treatment schedule: 34 (200 mg/kg) applied twice daily (b.i.d.)
ACS Paragon Plus Environment
24

Page 25 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
for 35 days. (A) Tumor area (n = 10 mice/group). (B) Tumor weight at study end (day 111). Tumor
areas and weights given as mean ± SD. (C) Body weight change (defined as the percentage change
of the actual mean body weight compared to the maximal mean body weight after start of
treatment).
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
Synthesis of 34 (BAY-985). The synthesis of 34 started with a Suzuki coupling between pinacol
ester 36 (available in one step from the corresponding bromide) and chloropyrimidine 37 (available
in one step from 4,6-dichloropyrimidine) using tetrakis(triphenylphosphine)palladium(0) (Scheme
1). The resulting product 38 was converted into benzimidazole 40 in a two-step procedure: 38 was
first reacted with 1,1'-thiocarbonyldiimidazole (TCDI) and amine 39 (available in five steps from
commercially available reagents, see the Supporting Information) to provide a thiourea
intermediate that was cyclized using N-ethyl-N'-(3-dimethylaminopropyl)carbodiimide (EDCI).
Acid deprotection of the Boc group, followed by amide coupling with 3,3,3-trifluoropropanoic
acid (using HATU), led to 34.
Scheme 1. Synthesis of 34a
NH2
NH2
NH2
NH2
N
Cl
H2N
N
N
N
a
b,c
N
N
N
O
+
+
N
B
O
N
O
O
37
38
39
40
H
N
N
H
N
N
N
N
NH
N
N
F
F
d,e
NH
N
N
O
O
F
N
O
N
N
41
34
N
N
N
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 43
1
2
3
4
5
6
aReagents and conditions: (a) Na2CO3, Pd(PPh3)4, 1,4-dioxane, 90 °C; (b) 1H-imidazole, TCDI,
DCM, 0 °C; (c) EDCI, DCM; (d) HCl, 1,4-dioxane, DCM; (e) 3,3,3-trifluoropropanoic acid,
HATU, NaHCO3, DMF.
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
CONCLUSION
Starting from a HTS campaign, we identified a family of aryl-substituted benzimidazoles as
highly potent TBK1/IKKε inhibitors. Early in our investigations, we observed that achieving a
very high kinase selectivity will be challenging within this compound class. Using X-ray
crystallography, we could confirm the aminobenzimidazole interaction at the hinge as well as
identify the hydrogen bond interaction with the protein from the oxygen of the piperazine amide.
With the help of computational chemistry, we hypothesized that bulky substituents addressing the
back pocket of the protein could improve overall kinase selectivity.
Addressing the back-pocket of the protein with different aromatic substituents turned out to be
key in order to improve the selectivity profile of our compounds. Pyrazole-substituted
benzimidazoles such as compound 24 displayed an improved overall kinase profile with TBK1
and IKKε being the most potently inhibited targets and with a limited number of off-targets
inhibited. Finally, pyrimidine-substituted benzimidazoles appeared to have a superior selectivity
profile. This observation resulted in the identification of compound 34 (BAY-985), a highly potent
and selective ATP-competitive dual inhibitor of TBK1 and IKKε. Compound 34 was active in our
cellular mechanistic assay and showed anti-proliferative activity in a few cancer cell lines.
However, we noticed that the antiproliferative activity observed on ACHN cells did not correlate
with the potency on our targets and appeared to be weaker for compounds showing a high
selectivity. This result highlighted the fact that the antiproliferative activity observed early in our
ACS Paragon Plus Environment
26

Page 27 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
investigations might not be a consequence of TBK1/IKKε inhibition but is more likely resulting
from off-targets effects.
7
8
9
The pharmacokinetic profile of compound 34 was then investigated and a efflux ratio of 17 fold
was observed in the Caco-2 permeability assay.
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
In vivo, compound 34 showed high clearance (CLb = 4.0 L/h/kg, ca. 95% hepatic extraction),
large volume of distribution at steady state (Vss = 2.9 L/kg) and a short terminal half-life (t1/2 = 0.79
h).
Despite the high potency of compound 34, we observed only weak antitumor efficacy in the SK-
MEL-2 human melanoma xenograft model, presumably because we didn’t reach sufficient
exposure to cover the antiproliferative concentration of SK-MEL-2 cells. It is possible that we
haven’t identified the appropriate cell line system characterized by specific oncogenic aberrations
or certain biomarkers related to sensitization to TBK1/IKK inhibition. Possibly, the use of
combination treatments might be beneficial. Therefore, we have not devalidated TBK1/IKKε as
anti-cancer targets but simply have not provided full (in vivo) evidence for it.
Compound 34 was recently accepted by the Structural Genomics Consortium as a dual
TBK1/IKKε probe (along with the negative control BAY-440) and is now available to the
scientific community via the SGC35–43 as a tool compound to aid the understanding of the
pharmacology and potential therapeutic use of TBK1/IKK pathway inhibition in tumors.
EXPERIMENTAL SECTION
Chemistry.
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 43
1
2
3
4
5
6
7
8
9
General methods and materials. Commercially available reagents and anhydrous solvents were
used as supplied, without further purification. All air- and moisture-sensitive reactions were
carried under an inert atmosphere of argon. Reactions were monitored by TLC and UPLC analysis
with a Waters Acquity UPLC MS Single Quad system (see SI for methods details). Flash
chromatography was carried out using a Biotage® Isolera™ One system with 200–400 nm variable
detector. Preparative HPLC was carried out with a Waters AutoPurification MS Single Quad
system; column: Waters XBridge C18 5 µm, 100 × 30 mm; basic conditions: eluent A: H2O + 0.2
vol% aq NH3 (32%), eluent B: MeCN; gradient: 0–0.5 min 5% B, flow: 25 mL/min; 0.51–5.50
min 10–100% B, flow: 70 mL/min; 5.51–6.5 min 100% B, flow: 70 mL/min; acidic conditions:
eluent A: H2O + 0.1 vol% formic acid (99%), eluent B: MeCN; gradient: 0–0.5 min 5% B, flow:
25 mL/min; 0.51–5.50 min 10–100% B, flow: 70 mL/min; 5.51–6.5 min 100% B, flow: 70
mL/min; temperature: 25 °C; DAD scan: 210–400 nm. NMR spectra were recorded at ambient
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
temperature (22 ± 1 °C), unless otherwise noted, on Bruker Avance III HD spectrometers. H
NMR spectra were obtained at 300, 400, 500 or 600 MHz, and referenced to the residual solvent
signal (2.50 ppm for [D6]DMSO). 13C NMR spectra were obtained at 125 MHz and also referenced
to the residual solvent signal (39.52 ppm for [D6]DMSO). 1H NMR data are reported as follows:
chemical shift (δ) in ppm, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, br = broad,
m = multiplet) and integration. Low-resolution mass spectra (electrospray ionization) were
obtained via HPLC-MS (ESI) using a Waters Acquity UPLC system equipped with an SQ 3100
Mass Detector; column: Acquity UPLC BEH C18 1.7 µm, 50 × 2.1 mm; eluent A: H2O + 0.05%
formic acid (99%), eluent B: MeCN + 0.05% formic acid (99%); gradient: 0–0.5 min 5% B, 0.5–
2.5 min 5–100% B, 2.5–4.5 min 100% B; total run time: 5 min; flow: 0.5 mL/min. Melting points
were determined with a Büchi B-540 melting point apparatus. Optical rotations were recorded on
ACS Paragon Plus Environment
28

Page 29 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a JASCO P-2000 polarimeter. The purity of all target compounds was at least 95%, as determined
by 1H NMR spectroscopy and UPLC analysis. Compound names were generated using ICS
software.
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
Only the synthesis of compound 34 (BAY-985) is described in this section. For all other
compounds please see the supplementary information.
Synthesis of 34 (BAY-985).
6-Chloro-N,N-dimethylpyrimidin-4-amine (38)
4,6-Dichloropyrimidine (50.0 g, 336 mmol), dimethylamine (180 mL, 2.0 M in THF, 369 mmol),
and K2CO3 (51.0 g, 369 mmol) were stirred in 1,4-dioxane (400 mL) for 14 h at reflux. Then, the
reaction mixture was cooled to rt and diluted with water. The aqueous phase was extracted with
DCM. The organic layer was dried over a silicone filter and concentrated under reduced pressure
to provide 38 (53.7 g), which was used without further purification. 1H NMR (400 MHz, DMSO-
d6): δ = 3.06 (br s, 6H), 6.75 (d, J = 1.01 Hz, 1H), 8.31 (d, J = 1.01 Hz, 1H). LC-MS (method 2,
see Supporting Information): tR (min) = 0.76. MS (ESI+): m/z = 158 [M + H]+.
4-[6-(Dimethylamino)pyrimidin-4-yl]benzene-1,2-diamine (39)
4-(4,4,5,5-Tetramethyl-1,3,2-dioxaborolan-2-yl)benzene-1,2-diamine (37; 20.0 g, 85.4 mmol)
and crude chloropyrimidine 38 (16.2 g) were dissolved in 1,4-dioxane (500 mL), and Na2CO3 (130
mL, 2.0 M in water, 260 mmol) and Pd(PPh3)4 (9.87 g, 8.54 mmol) were added. The mixture was
stirred for 2 h at 90 °C, then additional Pd(PPh3)4 (2.0 g, 1.73 mmol) was added and stirring
continued overnight at 90 °C. The mixture was cooled to rt and filtered. The filtrate was
concentrated under reduced pressure and the residue was purified by flash chromatography
1
(hexane/EtOAcEtOAc/EtOH) to give 39. Yield 15.4 g (74% yield). H NMR (400 MHz,
DMSO-d6): δ = 3.09 (s, 6H), 4.55 (s, 2H), 4.92 (s, 2H), 6.54 (d, J = 8.36 Hz, 1H), 6.79 (d, J = 1.01
ACS Paragon Plus Environment
29