Structure-Based Optimization of Potent, Selective, and Orally Bioavailable CDK8 Inhibitors Discovered by High-Throughput Screening

Article abstract of DOI:10.1021/acs.jmedchem.6b00597

The mediator complex-associated cyclin dependent kinase CDK8 regulates β-catenin-dependent transcription following activation of WNT signaling. Multiple lines of evidence suggest CDK8 may act as an oncogene in the development of colorectal cancer. Here we describe the successful optimization of an imidazo-thiadiazole series of CDK8 inhibitors that was identified in a high-throughput screening campaign and further progressed by structure-based design. In several optimization cycles, we improved the microsomal stability, potency, and kinase selectivity. The initial imidazo-thiadiazole scaffold was replaced by a 3-methyl-1H-pyrazolo[3,4-b]-pyridine which resulted in compound 25 (MSC2530818) that displayed excellent kinase selectivity, biochemical and cellular potency, microsomal stability, and is orally bioavailable. Furthermore, we demonstrated modulation of phospho-STAT1, a pharmacodynamic biomarker of CDK8 activity, and tumor growth inhibition in an APC mutant SW620 human colorectal carcinoma xenograft model after oral administration. Compound 25 demonstrated suitable potency and selectivity to progress into preclinical in vivo efficacy and safety studies.

Full text of DOI:10.1021/acs.jmedchem.6b00597

Subscriber access provided by Northern Illinois University
Article
Structure-based optimization of potent, selective and orally
bioavailable CDK8 inhibitors discovered by high throughput screening
Paul Czodrowski, Aurélie Mallinger, Dirk Wienke, Christina Esdar, Oliver Poeschke, Michael Busch, Felix
Rohdich, Suzanne A. Eccles, Maria Jesus Ortiz Ruiz, Richard Schneider, Florence I Raynaud, Paul A.
Clarke, Djordje Musil, Daniel Schwarz, Trevor C Dale, Klaus Urbahns, Julian Blagg, and Kai Schiemann
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b00597 • Publication Date (Web): 04 Aug 2016
Downloaded from http://pubs.acs.org on August 5, 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 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
Structure-based optimization of potent, selective and orally bioavailable
CDK8 inhibitors discovered by high throughput screening
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
Paul Czodrowski,*,(1) Aurélie Mallinger,(2) Dirk Wienke,(1) Christina Esdar,(1) Oliver Pöschke,(1) Michael
Busch,(1) Felix Rohdich,(1) Suzanne A. Eccles,(2) Maria-Jesus Ortiz-Ruiz,(2) Richard Schneider,(1) Florence
Raynaud,(2) Paul A. Clarke,(2) Djordje Musil,(1) Daniel Schwarz,(1) Trevor Dale,(3) Klaus Urbahns,(1) Julian
Blagg,(2) and Kai Schiemann*,(1)
(1) Merck KGaA, Darmstadt, 64293, Germany
(2) Cancer Research UK Cancer Therapeutics Unit at The Institute of Cancer Research, London, SW7
3RP, U.K.
(3) School of Bioscience, Cardiff University, Cardiff, CF10 3AX, U.K.
* Corresponding authors: paul.czodrowski@merckgroup.com, kai.schiemann@merckgroup.com
ABSTRACT
The Mediator complex-associated cyclin dependent kinase CDK8 regulates beta-catenin-
dependent transcription following activation of WNT signaling. Multiple lines of evidence
suggest CDK8 may act as an oncogene in the development of colorectal cancer. Here we
describe the successful optimization of an imidazo-thiadiazole series of CDK8 inhibitors that
was identified in a high-throughput screening campaign and further progressed by structure-
based design. In several optimization cycles, we improved the microsomal stability, potency
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 42
1
2
3
4
5
6
7
8
9
and kinase selectivity. The initial imidazo-thiadiazole scaffold was replaced by a 3-methyl-1H-
pyrazolo[3,4-b]-pyridine which resulted in compound 25 (MSC2530818) that displayed
excellent kinase selectivity, biochemical and cellular potency, microsomal stability and is
orally bioavailable. Furthermore, we demonstrated modulation of phospho-STAT1, a
pharmacodynamic biomarker of CDK8 activity, and tumor growth inhibition in an APC-
mutant SW620 human colorectal carcinoma xenograft model after oral administration.
Compound 25 demonstrated suitable potency and selectivity to progress into preclinical in
vivo efficacy and safety studies.
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
In contrast to other members of the CDK family (e.g. CDK1, CDK2 and CDK4/6), neither CDK8
nor its paralog CDK19 are required for the regulation of cell cycle progression. Instead, CDK8
and CDK19 together with MED12/12L and MED13/13L respectively form a regulatory kinase
module for the Mediator complex.1,2 The Mediator complex is a large protein assembly that
couples gene-specific transcriptional regulators to the general RNA polymerase II
transcription machinery.2–7 CDK8 and CDK19 participate in the regulation of transcription
through phosphorylation of RNA polymerase II8 and other transcription factors to regulate
their activity or tag them for proteosomal degradation.9–14
The role of CDK19 in cellular biology and human disease is not well understood and is the
subject of ongoing investigation.1 However, in contrast, CDK8 is much better characterized
and has been proposed to act as an oncogene in colon cancer15 and approximately 60 % of
colorectal cancers have increased CDK8 gene copy number.16 Moreover, a worse prognosis
in colon, breast and ovarian cancer is linked to increased CDK8 expression.17 CDK8 is also
2
ACS Paragon Plus Environment

Page 3 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
essential for cell proliferation in melanoma.18 Additional cancer-relevant activities of CDK8
include growth factor-induced transcription, modulation of TGF beta-signaling,13 and
phosphorylation of the Notch intracellular domain.19,20 Although knockout of CDK8 prevents
embryonic development due to its essential role in the pluripotent stem cell, other studies
have reported that CDK8 depletion does not affect the growth of normal cells.15,21,22
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
CDK8 may also contribute to tumorigenesis through activation of beta-catenin regulated
gene expression,16 a core transcriptional regulator of canonical WNT signaling. Consistent
with this observation, expression of the CDK8 gene is correlated with activation of beta-
catenin in colon and gastric cancers.17,23 Due to its involvement as a regulator of WNT/beta-
catenin signaling, CDK8 may be linked with disorders and diseases in which activation of
WNT/beta-catenin is crucial such as hyperproliferative, inflammatory or degenerative
disorders, and several malignant diseases.19,20
An extensive contemporary review reports the current status of small molecule inhibitors for
CDK8/CDK19 and highlights an increasing interest in targeting CDK8;24 in addition, a recent
report demonstrates a potential use for CDK8 inhibitors in AML.25 We recently described the
discovery of small molecule inhibitors of CDK8 from a cell based high-throughput screen
(HTS) seeking WNT signaling inhibitors and also reported additional differentiated series of
CDK8 inhibitors from a scaffold-hopping approach using the biochemical screen reported
herein.26–30 Here we describe the identification of a novel CDK8/19 chemotype from a
biochemical HTS versus CDK8 and its subsequent optimization, enabled by structure-based
design, to give compound 25, a potent, selective and orally bioavailable CDK8/19 ligand
suitable for further progression into preclinical studies.
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 42
1
2
3
RESULTS & DISCUSSION
4
5
6
7
8
9
In a CDK8 high-throughput screen using a FRET-based Lanthascreen binding competition
assay27, we identified several hit compounds with sub-micromolar activity. Our hit series
contained an imidazo[5,1-b][1,3,4]thiadiazole scaffold, henceforth described as the
“imidazo-thiadiazole” series. Initial hits bore either a 2-substituted benzyl substituent (1 and
2) or a 2-phenylpyrrolidine ring (3 and 4) attached to the imidazo-thiadiazole scaffold via an
amide bond at the 2-position (Table 1). As a cellular readout of WNT signaling, we used our
previously described 7dF3 WNT-specific reporter gene assay31 and observed a 3- to 13-fold
drop off from in vitro CDK8 binding affinity to cell-based activity for compounds 1 – 4.
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 HTS hits from the “imidazo-thiadiazole” series
CLint [ꢀL/min/mg]
CDK8 IC50
[nM]
7dF3 IC50
[nM]
ID
1
Structure
Human
Rat
Mouse
129 ± 27
1650 ± 495
2450 ± 778
>1000
435
119
2
781 ± 114
61
56
38
>1000
145
3
4
77 ± 4
60 ± 6
415 ± 262
290 ± 141
446
38
280
129
Introduction of a methyl group to the alkyl linker resulted in a 6-fold drop in biochemical
potency (compound 2, Table 1) and the pyrrolidine moiety was favored over an alkyl chain
linker by virtue of significantly increased cell-based potency and improved metabolic stability
4
ACS Paragon Plus Environment

Page 5 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(compounds 3 and 4). Hydrogen to fluorine exchange at the para position of the phenyl ring
(compound 3 to 4) did not influence biochemical nor cell-based potency; however, we noted
a slight improvement in microsomal stability (Table 1). To further explore the impact of the
para-substitution and stereochemistry at the pyrrolidine 2-position, the racemic form and
both the enantiomers bearing a p-Cl-phenyl were synthesized (Table 2). The racemic form
(5) showed modest biochemical and cell-based activity. Interestingly, the S-enantiomer (6)
was significantly more active than the R-enantiomer (7); however, the microsomal stability
of both enantiomers was in the same range.
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 Determination of optimal pyrrolidine stereochemistry
CLint [ꢀL/min/mg]
CDK8 IC50
[nM]
7dF3 IC50
[nM]
ID
5
Structure
Human
Rat
Mouse
17 ± 0.02
3.8 ± 1.9
110 ± 15
100 ± 28
70 ± 14
52
719
388
6
7
49
60
655
449
447
514
1,600 ± 0
In order to understand the binding mode and to inform further medicinal chemistry design,
we determined the crystal structure of the most potent enantiomer (6) bound to
CDK8/cyclin C (Figure 1A). Compound 6 is bound to the hinge region of CDK8 (Ala100) via
the nitrogen atom at the 6-position of the “imidazo-thiadiazole” scaffold whilst the 5-methyl
group points away from the gatekeeper residue (Phe97) towards the solvent-exposed
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 42
1
2
3
4
5
6
7
8
9
channel. The carbonyl group of the C2-amide forms a cation-pi interaction to the catalytic
lysine residue (Lys52), too distant (3.6 Å) to be considered as a hydrogen bond. The sulfur
atom of the “imidazo-thiadiazole” scaffold is positioned to form favorable van der Waals
(vdW) interactions to the gatekeeper residue Phe97. The pyrrolidine ring also forms
favorable vdW interactions to Tyr32 and the side chain of Asp173. Furthermore, the
orientation of the pyrrolidine defines the exit vector for the p-Cl-phenyl substituent which
itself forms vdW interactions with Val27, Tyr32, Val35, Leu158 and Arg356. We did not
observe sufficient evidence to support a halogen bond between the chlorine atom and the
backbone C=O of Val27: the distance is 2.9 Å and the angle between the chlorine atom and
the C=O functionality of Val27 is 104.7°.
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
Comparison of the structure of 6 in CDK8/cyclin C with our recently published crystal
structure of a 3,4,5-trisubstituted-2-aminopyridine chemotype27 (compound 8, CCT251921)
revealed few structural differences in the active site. The CDK8 residues in the ATP binding
site with the largest variance between the two crystal structures are shown (Figure 1B).
Despite the different modes of ligand-binding, the protein structure is largely conserved.
Compound 6 and compound 827 both form H-bond acceptor interactions with the backbone
NH of hinge residue Ala100 and interact with the catalytic lysine (Lys52) through the amide
and lactam carbonyl groups respectively. However, interaction with the gatekeeper residue
Phe97 is mediated by the sulfur atom of the thiadiazole scaffold 6 as opposed to the 3-
chlorosubstituent in compound 8. Most notably, the para-chlorophenyl and N-
methylindazole rings that project into a similar region in the CDK8 solvent channel are linked
to the hinge-binding scaffold through different architectures in the two series (Figure 1B).
6
ACS Paragon Plus Environment

Page 7 of 42
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
(A)
(B)
Figure 1 (A) Crystal structure of CDK8/cyclin C complexed with compound 6 (blue). Hydrogen
bonds are displayed as dashed lines. The cation-dipole interaction between the carbonyl
atom and the catalytic Lys52 is also shown. The key amino acids are given in stick
representation. (B) Alignment of compound 6 (blue)/CDK8/cyclin C crystal structure (salmon)
and compound 8 (pink)/CDK8/cyclin C crystal structure (green). The residues which deviate
most are highlighted: Tyr32, Lys52, and Asp173.
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 42
1
2
3
4
5
6
7
8
9
We were concerned that the thiadiazole ring may be prone to oxidative metabolism and
therefore set out to discover a modified scaffold hop in order to mitigate this risk. In
addition, we wished to explore variation of the pyrrolidine ring and its pendant 2-phenyl
substituent in order to optimize potency and metabolic stability.
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) Scaffold hop from the hinge binding motif of the “imidazo-thiadiazole” series
We modified the “imidazo-thiadiazole” scaffold based upon the crystal structure of
compound 6 in CDK8/cyclinC (Figure 1A) according to SZMAP32 and WaterMap33 analysis
(see Figure 2). We ran these calculations for the ligand-deleted form of the crystal structure
of compound 6 with CDK8/cyclin C. Given this calculation set-up, we could analyze the
energy of predicted water sites inside the hinge binding region and identify possible handles
for scaffold variations.
SZMAP [Solvent ZAP (ZAP is A Poisson-Boltzmann solver) MAPping] represents a mixture of
an implicit and explicit solvent model. In brief, a grid with a spacing of 0.25 Å is generated
inside the protein active site and an explicit water molecule is positioned at each grid point
with multiple orientations, followed by an implicit solvent Poisson-Boltzmann
calculation.34,35 These calculations can be done with and without the ligand. For our analysis,
we performed the calculations without the ligand. Once the calculations are completed for
all grid points, an overall energy grid for the presence or absence of water molecules is
computed. The predicted water molecules show different characteristics based on their
energy values: a water molecule is characterized as "hydrophobic" if its energy value is
positive, otherwise it is characterized as "hydrophilic".
The WaterMap methodology is a combination of MD simulation36 with the Inhomogeneous
Fluid Approach37. In brief, this approach calculates the enthalpic and entropic contributions
8
ACS Paragon Plus Environment

Page 9 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
of binding per predicted water molecule. Consistent with the SZMAP analysis, the ligand was
left out for the WaterMap calculations. If the overall free energy per predicted water
molecule is positive, it is specified as "unhappy water", whereas a predicted negative free
energy indicates a "happy water".
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 interpretation of SZMAP and WaterMap calculations can be used for the differentiation
of binding modes observed for highly structurally-related ligands38 and for rationalization of
SAR,39, 40 kinase selectivity,41 and binding kinetics.42 In the course of our studies, we
employed SZMAP and WaterMap for the interpretation of modifications to the hinge binding
scaffold. Applying these computational approaches, we explored changes to the five-
membered ring of the hinge binding motif and selected a range of 5,6-bicyclic scaffolds for
synthesis. The remainder of the molecule was fixed by the S- p-Cl-2-phenyl-pyrrolidine motif
(Table 3) and compounds were scored by WaterMap, see Figure 2(C).
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 42
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
(A)
(B)
(C)
Figure 2 (A) SZMAP and WaterMap calculations have been performed for the crystal structure of
compound 6 complexed to CDK8/cyclin C. For clarity, only part of the ligand denoted in bold (Figure
10
ACS Paragon Plus Environment

Page 11 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2A) is shown in Figures 2(B) and 2(C). (B) Upper panel: Only the hinge region is shown. SZMAP
analysis of the CDK8/compound 6 crystal structure: yellow spheres correspond to water sites bearing
hydrophilic character and pink spheres correspond to water sites bearing hydrophobic character.
Lower panel: The contribution from each ligand atom resulting from water molecule replacement at
the same position is shown. The character of the ligand-displaced water spheres are depicted as
mesh: yellow mesh corresponds to hydrophilic water spheres, green mesh corresponds to
hydrophobic water spheres. The size of the water spheres correspond to the relative energy gain
obtained by replacement of the particular water sphere by each ligand atom. (C) Upper panel: Only
the hinge region is shown for clarity - the full ligand is depicted in Figure 2(A) WaterMap analysis of
the CDK8/compound 6 crystal structure: red spheres correspond to “unhappy” water molecules,
yellow spheres to “happy” water molecules. Lower panel: WaterMap scoring versus CDK8 pIC50 for
the compounds from Table 1 & 2 (blue dots) and the compounds from Table 3 (red dots), r2 = 0.54.
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
Analysis of the predicted water sites via SZMAP (see Figure 2(B)) revealed that hydrophilic
water molecules are favored directly adjacent to the hinge residues Ala100 and Asp98. This
is consistent with the corresponding hydrophilic hinge-binding motifs of our selected
scaffold replacements. The remaining part of the hinge binding region is predicted to be
mostly filled with hydrophobic water molecules. Such water molecules can be replaced by
both hydrophobic and hydrophilic moieties.43–45 Based on this finding, we ranked our
compound ideas by WaterMap scores. In addition, the characterization of predicted water
sites by SZMAP calculations revealed that broad scaffold variations should be possible. The
new compound design ideas were also prioritized based upon their synthetic arose from
chemical tractability and their potential for improved primarily aimed for the optimization of
microsomal stability. Figure 2(B) (lower panel) depicts the contribution from each ligand
atom resulting from water molecule replacement at the same position. This analysis
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 42
1
2
3
4
5
6
7
8
9
underscores the 30-fold difference in CDK8 affinity between compounds 9 and 10: the
favorable binding contribution by addition of a methyl group at the 5-position is driven by
the replacement of an “unhappy” water molecule; this finding is consistent with both the
WaterMap and SZMAPpredictions . Indeed, the WaterMap calculations (see Figure 2(C))
indicate that the hinge-binding region is filled with “unhappy” water molecules. We
performed a ranking of our compounds based on the WaterMap scores for the docking
solutions of compounds from Tables 1 & 2 which adopt a similar hinge-binding mode. Based
on an acceptable trend between WaterMap scores and pIC50, we also scored new compound
ideas from Table 3. The outcome of this ranking is given in Figure 2C (lower panel)
suggesting a weak correlation between the WaterMap-based analysis and the in vitro
potency of compounds from Table 3. We omitted compound 11 from this analysis, because
we anticipate a different binding mode; indeed, a 440-fold decrease in potency was seen
with the introduction of an sp2 nitrogen atom juxtaposed to the C=O backbone group of
Asp98, see compound 11, versus the corresponding hydrogen bond donor in compound 10
(Table 3). However, compound 11 showed good microsomal stability indicating that an
appropriate scaffold hop could significantly improve stability to oxidative metabolism.
Pleasingly, the introduction of a methyl group at the 3-position of the indazole 9 to give 10
resulted in a 22-fold improvement in biochemical potency and a 166-fold increase in cell-
based potency; however, no improvement in microsomal stability was observed consistent
with the increased lipophilicity of compound 10 versus 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
In summary, the potency increase arising from introduction of the C3- methyl group in
compound 10 is consistent with our SZMAP analysis (Figure 2(B). However, we recognized
that the indazole C3-methyl group could be susceptible to oxidative metabolism and
therefore introduced a more polar amino group at this position (compound 13) cognizant
12
ACS Paragon Plus Environment

Page 13 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
that improved hinge-binding interaction with the backbone carbonyl of Ala100 may also be
realized (Figure 1). This tactic maintained in vitro CDK8 potency and improved microsomal
stability as desired but unfortunately led to a 47-fold drop in cellular potency – compared to
compound 10 – despite good Caco2 flux (PappA-B = 33.7 x10-6 cm s-1) and low efflux ratio
(2.2).
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
Table 3 Variation of the hinge binding motif of the “imidazo-thiadiazole” HTS scaffold
CLint [
Human
ꢀL/min/mg]
CDK8 IC50
[nM]
7dF3 IC50
[nM]
ID
9
Structure
Rat
Mouse
32 ± 33
100 ± 28
354
525
94
86
10
1.4 ± 0.1
0.6 ± 0.004
79
244
11
12
3,990 ± 1,151
>10,000
90 ± 14
<10
390
19
18
63
9 ± 1
108
13
4.9 ± 0.5
28 ± 3
23
10
10
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 42
1
2
3
4
5
6
7
8
9
Compound 13 proved optimal in terms of biochemical potency and metabolic stability and
was further profiled in a broad kinase panel to discover any kinase selectivity issues
(Supplementary Material Table S2). Five off-kinases were detected (JAK1, JAK2, JAK3, ROCK-
IIand TYK2) and were subject to IC50 determination to confirm potency (Table 4).
Consequently, we aimed to improve selectivity versus these off-target kinases in subsequent
medicinal chemistry optimization.
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 Kinase profile data for 13 (see also Supplementary Material, Table 2), sorted
alphabetically
kinase
CDK8
JAK1
IC50 (nM)
4.9
% inhibition at 1 ꢀM
100
86
84
79
42
70
90
JAK2
286
JAK3
219
ROCK-II
TYK2
1,330
279
(2) Variation of the ring linker size and halogen substituents
In order to inform our medicinal chemistry design, we calculated Selectivity Grids46 for the
CDK8/compound 6 crystal structure, see Figure 3(A), and selectivity grids for the most
notable off-target kinases, see Figure 3(B). Selectivity grids represent a plausible
interpretation of regions within binding pockets which may be targeted to improve (or to
lose) selectivity.46 These selectivity grids suggested that the para-halogen substituent of the
phenyl ring and the 3-amino group of the indazole scaffold represented possible variation
points to achieve selectivity for CDK8 over the off-target kinases, see Figure 3(A).
Furthermore, this analysis suggested that substitution at the solvent-exposed region of the
14
ACS Paragon Plus Environment

Page 15 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
binding pocket would non-specifically increase potency against the off-target kinases, see
Figure 3(B). Therefore, we decided to vary the para-halogen substituent as well as the
pyrrolidine ring and also to methylate the 3-amino group on the indazole scaffold in order to
improve the selectivity and cell-based activity of compound 13 (Table 5).
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)
(B)
Figure 3 (A) Selectivity Grids for the CDK8-specific regions of the active site. In general,
regions around the para-substituted halogen in the phenyl ring as well as the region
adjacent to the pyrrolidine ring can be considered as potential selectivity handles. (B)
Selectivity Grids of the ATP binding pocket which have the potential to drive increased
potency for off-target kinases such as the JAK family.
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 42
1
2
3
4
5
6
Table 5 Variations of the ring linker size and para-substitution of the phenyl ring.
7
8
9
CDK8
JAK1
JAK2
JAK3
% inhibition at 1 µM
CDK8
7dF3
ROCK-II
TYK2
76
CLint [ꢀL/min/mg]
ID
Structure
IC50 [nM]
Human
Rat
Mouse
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
14
98
88
88
76
65
3.1 ± 1
19 ± 4
<10
<10
14
15
16
93
94
72
63
59
24
58
21
27
12
62
45
32 ± 0.4 120 ± 14
<10
10
<10
<10
<10
10
49 ± 6
435 ± 163
17
18
95
96
21
24
-2
-14
15
22
20
17
23
45 ± 3
24 ± 6
800 ± 283
150 ± 14
<10
<10
15
26
28
19
-14
16
ACS Paragon Plus Environment

Page 17 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
19
99
34
34
37
17
32
5.4 ± 3
12 ± 6
18
28
31
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
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 42
1
2
3
The kinase selectivity data confirmed the hypothesis that the para-substituent could act as
4
5
6
selectivity handle: The %_INHIBITION values, which refer to single point measurements of
7
8
percent inhibition at 1 ꢀM, demonstrate that the off-target activity versus the JAK family
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
decreases as the para-halogen substituent is varied from Br to Cl to F. Selectivity could be
further improved by modifications of the ring linker size or switching from a 1,2-substitution
pattern to a 1,3-substitution pattern of the pyrrolidine linker. Pleasingly, the affinity for off-
target kinases was also reduced by methylating the 3-amino group.
In summary, variation of the halogen substituent in the para-position of the phenyl ring
showed a clear trend – while the on-target biochemical CDK8 and cellular 7dF3 potencies
decreased from Br to Cl to F in the matched trio 14, 13, and 15; less potent off-target activity
versus the JAK family, ROCK-II and TYK2 was also observed. Interestingly, all three
compounds were metabolically stable. Replacement of pyrrolidine (15) with azetidine (16)
retained biochemical potency and microsomal stability with slight loss in cell-based potency.
The same held true when increasing the ring size from a pyrrolidine to a piperidine linker
(17) although cell-based potency was further eroded. Alteration of the substitution pattern
at the pyrrolidine from 1,2 to 1,3 showed a moderate loss of biochemical and cellular
potency (compare 18 with 13) while maintaining good microsomal stability. Finally, the
addition of a methyl group to the 3-amino functionality of the indazole scaffold resulted in
equipotent biochemical and cellular activity (compare 19 with 13) combined with a slight
loss of microsomal stability.
18
ACS Paragon Plus Environment

Page 19 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
We next determined the crystal structure of compound 17 bound to CDK8/cyclin C (Figure
4A) and examined the superposition of 17 with compound 6 (Figure 4B). The indazole hinge
binding motif is located at the same position as the ”imidazo-thiadiazole” scaffold of 6. The
N-atom in the 1-position of the indazole scaffold forms an additional hydrogen bond to the
C=O backbone of Asp98 which cannot be formed by 6. The carbonyl moiety linking the para-
halo -phenyl piperidine to the hinge binding scaffold is located at the same position in both
ligands. Interestingly, the exocyclic amino group in the 3-position of the indazole does not
form a hydrogen bond with the backbone carbonyl atom of Ala100 (the heavy atom distance
= 3.4 Å) and the electron density surrounding this amino group is not consistent with the
placement of a water molecule. These observations are consistent with SZMAP and
WaterMap calculations that did not predict an energetically favorable water molecule in this
region. Compound 6 was 12-fold more potent than 17 versus CDK8 and we reasoned that
this difference may be driven by the halogen substituent; indeed comparison of the matched
molecular pair 15 and 17, which only differ by the ring linker size (pyrrolidine vs piperidine),
were equipotent (32 nM vs 45 nM). Although the ring size is increased from a 5-membered
pyrrolidine linker to a 6-membered piperidine linker, the p-F phenyl ring is placed at an
almost identical position compared to compound 6, see Figure 4(B). Comparing the fluorine
atom (compound 17) to the chlorine atom (compound 6), the halogen to Val27 carbonyl
distance is increased from 2.9 Å (compound 6, chlorine) to 3.7 Å (compound 17, fluorine).
We observed only one other structural difference: the distance between the carbonyl linker
and the catalytic lysine Lys52 (3.6 Å for compound 6 and 4.6 Å for compound 17). Thus, we
attribute the increased potency of compound 6 to the improved interaction of the para-Cl
atom with Val27 compared to the para-F atom in compound 17 together with improved
interaction between the carbonyl linker and Lys52 for compound 6.
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
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 42
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
(A)
(B)
FIGURE 4 (A) Crystal structure of 17 (orange) in complex CDK8/cyclin C. The hydrogen bonds
are indicated by black dotted lines. (B) Superposition of crystal structures of compounds 17
(orange) and 6 (blue).
Although we had improved the kinase selectivity and maintained microsomal stability by
variations of the pyrrolidine linker and para-halogen substituent, the resultant compounds
(16 and 17) lacked the desired biochemical potency and cellular activity. Comparison of the
matched molecular pair C3-amino with C3-methyl (13 vs 10), indicated that the C3-methyl
substitution resulted in good cellular potency, albeit with poorer microsomal stability.
Consequently, we selected the C3-methylindazole scaffold for further progression.
(3) Optimization of microsomal stability
20
ACS Paragon Plus Environment

Page 21 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In vitro metabolite identification studies indicated that the pyrrolidine ring linker is the
major site of oxidative metabolism for compound 10. Thus, we prepared close analogues of
compound 10 with reduced overall lipophilicity with the aim of improving metabolic stability
(Table 6).
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
Replacement of the pyrrolidine by an azetidine (compound 20) did not lead to improved
metabolic stability in all species; however the racemic piperazine and morpholine derivatives
(21 and 22 respectively) resulted in improved CLint profiles across all species; disappointingly,
we observed a decrease in the biochemical CDK8 affinity and the cellular potency (Table 6).
Table 6 Variations of the ring linker size based on the 3-methyl-indazole scaffold.
CLint [ꢀL/min/mg]
CDK8 IC50
[nM]
7dF3 IC50
[nM]
ID
Structure
Human
Rat
Mouse
20
1.7 ± 0.1
2.0 ± 0.4
230 ± 14
908
62
113
21
22
15 ± 2
135
135
40
44
63
29
8.5 ± 2
115 ± 7
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 42
1
2
3
4
5
6
7
8
9
In summary, the microsomal stability could be improved by reducing overall lipophilicity;
however, a corresponding drop in biochemical and cell-based activity was also observed.
Therefore we attempted to improve the metabolic stability by reducing the lipophilicity of
the molecule through the introduction of polarity into the hinge-binding scaffold.
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
(4) Fine-Tuning of the hinge binding motif
The pyrazole hinge binding motif was fixed and the six-membered ring of the indazole
scaffold was varied by introduction of single nitrogen atoms in the 4-, 6- and 7-positions to
reduce the overall lipophilicity of the scaffold (compounds 23 to 25, Table 7). Introduction of
a nitrogen atom at the 5-position of the indazole ring (23) resulted in a significant drop in
biochemical and cell-based potency. Nitrogen at the 4-position of the indazole ring gave the
most potent analog (24); however, this compound showed only modest microsomal stability.
Pleasingly, introduction of a nitrogen atom at the 7-position (25) resulted in acceptable
biochemical CDK8 affinity and furthermore showed improved microsomal stability (Table 7).
22
ACS Paragon Plus Environment

Page 23 of 42
Journal of Medicinal Chemistry
1
2
3
Table 7 Fine-tuning of the hinge binding motif.
4
5
6
7
8
CLint [ꢀL/min/mg]
CDK8 IC50
[nM]
7dF3 IC50
[nM]
ID
Structure
Human
Rat
Mouse
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
23
11 ± 1.6
1.3 ± 0.06
2.6 ± 0.1
125 ± 35
2.8 ± 0.4
6.5 ± 2
206
100
45
70
61
29
24
25
69
<10
<10
The matched molecular pair compound 25 versus compound 6, where only the hinge binding
scaffold (“pyrazolo-pyridine” vs. “imidazo-thiadiazole”) has been changed, showed
equivalent biochemical CDK8 affinity and a 11-fold improvement in cell-based potency
together with a significant improvement in mouse and rat microsomal stability.
We next determined the crystal structure of 25 bound to CDK8/cyclin C (Figure 5A). The
carbonyl atom at C5 of the azaindazole scaffold of 25 forms a hydrogen bond to Lys52 and
the side-chain of Lys52 moves towards the ligand in comparison to the binding mode of
compound 6 in CDK8/cyclin C (Figure 5B). Overall, the orientation of the active site residues
is well conserved between the crystal structures of 25 and 6. The interaction pattern is
identical except for an additional favorable interaction from the N1-indazole hydrogen bond
donor to the backbone C=O atom of Asp98 for 25. Notably, the additional nitrogen atom at
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 42
1
2
3
4
5
6
7
8
the 7-position of the indazole scaffold does not elicit any additional favorable interactions
consistent with the corresponding biochemical potencies: indazole 10 (CDK8 IC50 = 1.4 nM)
versus aza-indazole 25 (CDK8 IC50 = 2.6 nM).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(A)
(B)
Figure 5 (A) Crystal structure of 25 (violet) complexed with CDK8-cyclin C (B) Superposition
of the 25 (violet) and 6 (blue) crystal structures
In a commercially available reporter displacement assay,47 compound 25 binds to CDK8 and
CDK19 with similar affinity (4 nM) - as we have previously observed across multiple
chemotypes27–30 - with residence times of 18 and 32 min respectively. Potent inhibition of
phospho-STAT1SER727, an established biomarker of CDK8 activity,26 in SW620 human
colorectal carcinoma cells was also observed (pSTAT1SER727 IC50 = 8 ± 2 nM). Furthermore,
compound 25 demonstrated potent inhibition of WNT-dependent transcription in human
cancer cell lines that have constitutively activated WNT signalling; for example, 25 inhibited
24
ACS Paragon Plus Environment

Page 25 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the reporter-based luciferase readout in several cell lines bearing activating WNT-pathway
mutations; LS174T (beta-catenin-mutant, IC50 = 32 ± 7 nM), COLO205 (APC-mutant, IC50 = 9 ±
1 nM) and demonstrated inhibition of WNT3a ligand-dependent reporter readout in PA-1
cells (IC50 = 52 ± 30 nM). Compound 25 demonstrated minimal activity in the CEREP panel,
being active on the dopamine transporter (IC50 = 8.5 µM) as the only activity below 10 µM,
and demonstrated minimal hERG inhibition (18%@10uM). Testing against a 264 kinase panel
gave inhibition of only a single kinase by more than 50% at 1 µM (GSK3alpha IC50 = 700).
Furthermore, compound 25 is a soluble CDK8 inhibitor (kinetic solubility >200 µM,
thermodynamic solubilities: 0.399 mg/mL (PBS, pH 7.4), 0.628 mg/mL (FaSSIF, pH 6.5) and
0.836 mg/mL (FeSSIF, pH 5.0)) with high permeability (PappA-B = 44 x10-6 cms-1) and low
efflux ratio in Caco-2 cells (ER = 1.5) and does not inhibit any cytochrome P450 subtypes
(Cyp IC50s > 20 µM). In light of its promising in vitro profile, 25 was further profiled in in vivo
pharmacokinetic and pharmacological studies. Overall, the compound showed acceptable
pharmacokinetics (PK) in all tested preclinical species (Table 8). Prediction of the human PK
was performed by three species allometric scaling corrected for protein binding and
considering interspecies-differences in intrinsic clearance in liver microsomes. The human
clearance (CL) and volume of distribution at steady-state (Vss) were estimated to be low
(~0.14 L/h/kg) and small (0.48 L/kg), respectively, resulting in a short predicted terminal half-
life (2.4 h). Physiological based pharmacokinetics (PBPK) simulations suggested that human
oral bioavailability may be ≥75% up to dose level of 500 mg daily.
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
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 42
1
2
3
Table 8: Pharmacokinetic profile of compound 25
4
5
6
Species
fu (%)
VDss (L/kg) Plasma Cl (L/h/kg) %Qh t1/2 (h)
F
7
8
9
Mouse
Rat
13%
19%
50%
16%
0.59
1.62
1.37
1.50
22.5
34.6
53.8
12.1
0.45
0.86
0.83
~2.4
0.25
0.66
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
Dog
1.47
1.35
0.39
Human Prediction
* Predicted by PBPK modeling (GastroPlusTM)
~0.48
~0.14
>0.75*
Compound 25 was then assessed in vivo in an established SW620 human colorectal cancer
xenograft model in female NCr athymic mice. Tumor-bearing mice were treated orally with
compound 25 (50 mg/kg bid or 100 mg/kg qd), for 16 days. Both schedules gave a similar
reduction in tumor growth (Figure 6 (A)) with T/C ratios (based on final tumor weights) of
49% and 57% respectively. The compound was generally well tolerated, with no effects on
mouse body weight in the q.d. administration schedule and manageable body weight loss
with the bid schedule (Figure 6 (B)). We monitored inhibition of STAT1SER727 phosphorylation
by a Luminex immunoassay; phospho-STAT1SER727 levels were reduced to the limits of
detection 2 h after dosing in both administration schedules and remained below the control
levels even after 24 h, having more pronounced effects in the bid compared with the q.d.
schedule (Figure 6 (C)).
26
ACS Paragon Plus Environment

Page 27 of 42
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
(A)
(B)
(C)
(D)
Figure 6 Efficacy study of compound 25 in SW620 human colorectal cancer xenografts: (A)
Tumor growth rates expressed as volumes (mean±SEM). (B) Mouse body weights expressed
as % of day 0 values. (C) Phospho-STAT1SER727 levels relative to total STAT1 levels and
expressed as % vehicle controls at each time point. (D) Free plasma concentrations of
compound 25 over time at the two dose schedules (50 mg/kg bid or 100 mg/kg q.d.),
CONCLUSIONS
In an HTS campaign using a FRET-based Lanthascreen binding competition assay, we
identified a chemical series comprising an imidazo-thiadiazole scaffold with exemplar
compounds demonstrating IC50 values in the 60 – 780 nM range (Table 1); however, this
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 42
1
2
3
4
5
6
7
8
9
series lacked sufficient microsomal stability. Determination of the crystal structure of
compound 6 in CDK8/cyclin C demonstrated clear differences in binding mode compared to
our previously reported 3,4,5 trisubstituted-2-aminopyridine series.27 SZMAP and WaterMap
calculations were applied to identify potential scaffold replacements to enhance potency
and improve metabolic stability. Applying these computational approaches, we explored
changes to the skeleton of the five-membered ring of the imidazo-thiadiazole hinge binding
motif and discovered a series of indazole-based scaffold replacements. Notably, 3-methyl
indazole 10 demonstrated potent biochemical and cell-based potency; however insufficient
metabolic stability precluded further progression. Replacement of the C3-methyl substituent
with a C3-amino group gave compound 13 which demonstrated optimal biochemical
potency and metabolic stability but suffered from undesirable off-target kinase affinity.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We employed selectivity grids to rationalize key regions for the modulation of kinase
selectivity for the three most notable off-target kinases JAK1, JAK2, JAK3 and TYK2 and
discovered that selectivity could be improved by modifications to the pyrrolidine ring and by
variation of the para-halogen substituent on the pendant phenyl ring to give, for example,
the selective and metabolically stable para-fluorophenylpiperidine analog 17. Crystallization
of 17 in CDK8/Cyclin C revealed a conserved binding mode; however, 17 lacked the desired
biochemical potency and cellular activity.
Revisiting the C3-methylindazole series and applying the principle of reduced lipophilicity to
abrogate oxidative metabolism, enabled the synthesis of potent and selective analogues
with improved cell-based potency compared to 3-amino-indazole 17. In particular, the 7-
azaindole scaffold exemplified by compound 25 which demonstrated an optimal
compromise of biochemical and cell-based potency coupled with acceptable in vitro
metabolic stability that resulted in a favorable in vivo oral pharmacokinetic profile in mouse,
28
ACS Paragon Plus Environment

Page 29 of 42
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
rat and dog. Compound 25 demonstrated reduction of tumor growth rates of established
human SW620 colorectal carcinoma xenografts using two different oral dosing schedules.
We observed concomitant inhibition of a target engagement biomarker phospho-STAT1SER727
consistent with pharmacokinetic exposure.
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 light of its potent and selective profile coupled with good oral pharmacokinetics and
duration of in vivo target engagement on oral dosing, compound 25 (MSC2530818) was
selected for progression into further preclinical in vivo animal efficacy and safety studies.
EXPERIMENTAL SECTION
Chemistry.
Commercially available starting materials, reagents and dry solvents were used as supplied.
Column chromatography was performed on a CombiFlash Companion purification system
using Teledyne Isco silica cartridges. Preparative TLC was performed on Merck plates.
Preparative HPLC was conducted according the following methods: Method A, injections of
the sample were made onto a SunFire C18 OBD column (100 Å, 5 µm, 30 mm x 100 mm).
Chromatographic separation at room temperature was carried out using Agilent
Technologies, 1260 Infinity, acetonitrile:water gradient (both modified with 0.1% trifluoro
acetic acid) at a flow rate of 50 mL/min. Chiral preparative HPLC was conducted according
the following methods: Method B: Injections of the sample were made onto a ChiralPak AD-
H column (5 µm, 4.6 X 250 mm). Chiral chromatographic separation was carried at room
temperature out using SFC Berger Minigram, CO2/Methanol (85:15) + 0.5% diethyl amine, at
a flow rate of 5 mL/min. Method C: Injections of the sample were made onto a ChiralPak AS-
H column (5 µm, 4.6 X 250 mm). Chiral chromatographic separation was carried at room
temperature out using SFC Berger Minigram, CO2/Methanol (60:40), modified with 0.5%
29
ACS Paragon Plus Environment