Studies of CDK 8/19 inhibitors: Discovery of novel and selective CDK8/19 dual inhibitors and elimination of their CYP3A4 time-dependent inhibition potential

Article abstract of DOI:10.1016/j.bmc.2017.03.049

In this article, synthetic studies around a pyridylacrylamide-based hit compound (1), utilizing structure-based drug design guided by CDK8 docking models, is discussed. Modification of the pendant 4-fluorophenyl group to various heteroaromatic rings was c

Full text of DOI:10.1016/j.bmc.2017.03.049

Accepted Manuscript
Studies of CDK 8/19 Inhibitors: Discovery of Novel and Selective CDK8/19
Dual Inhibitors and Elimination of Their CYP3A4 Time-Dependent Inhibition
Potential
Jun Fujimoto, Takaharu Hirayama, Yasuhiro Hirata, Yukiko Hikichi, Saomi
Murai, Maki Hasegawa, Yuka Hasegawa, Kazuko Yonemori, Akito Hata,
Kazunobu Aoyama, Douglas R. Cary
PII:
S0968-0896(16)31317-7
DOI:
http://dx.doi.org/10.1016/j.bmc.2017.03.049
BMC 13647
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
21 December 2016
19 February 2017
22 March 2017
Please cite this article as: Fujimoto, J., Hirayama, T., Hirata, Y., Hikichi, Y., Murai, S., Hasegawa, M., Hasegawa,
Y., Yonemori, K., Hata, A., Aoyama, K., Cary, D.R., Studies of CDK 8/19 Inhibitors: Discovery of Novel and
Selective CDK8/19 Dual Inhibitors and Elimination of Their CYP3A4 Time-Dependent Inhibition Potential,
Bioorganic & Medicinal Chemistry (2017), doi: http://dx.doi.org/10.1016/j.bmc.2017.03.049
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Studies of CDK 8/19 Inhibitors:
Discovery of Novel and Selective CDK8/19 Dual Inhibitors and Elimination of Their CYP3A4
Time-Dependent Inhibition Potential
Jun Fujimotoa,*, Takaharu Hirayamaa, Yasuhiro Hirataa, Yukiko Hikichib, Saomi Muraia, Maki Hasegawaa,
Yuka Hasegawaa, Kazuko Yonemoria, Akito Hataa, Kazunobu Aoyamaa and Douglas R. Carya,†
a Pharmaceutical Research Division, Takeda Pharmaceutical Co., Ltd., Shonan Research Center, 26-1, Muraoka-Higashi 2-Chome,
Fujisawa, Kanagawa 251-8555, Japan
b Product Information Group Japan Oncology Business Unit, Takeda Pharmaceutical Co., Ltd., 12-10, Nihonbashi 2-Chome, Chuo-ku,
Tokyo 103-8686, Japan
* Corresponding author. Tel.: +81 466 32 1086; fax: +81 466 29 4448.
E-mail address: jun.fujimoto@takeda.com (J.Fujimoto)
† Present address: Peptidream Inc.
Abstract
In this article, synthetic studies around a pyridylacrylamide-based hit compound (1), utilizing
structure-based drug design guided by CDK8 docking models, is discussed. Modification of the pendant
4-fluorophenyl group to various heteroaromatic rings was conducted aiming an interaction with the
proximal amino acids, and then replacement of the morpholine ring was targeted for decreasing potential
of time-dependent CYP3A4 inhibition. These efforts led to the compound 4k, with enhanced CDK8
inhibitory activity and no apparent potential for time-dependent CYP3A4 inhibition (CDK8 IC50: 2.5 nM;
CYP3A4 TDI: 99% compound remaining). Compound 4k was found to possess a highly selective kinase
inhibition profile, and also showed favorable pharmacokinetic profile. Oral administration of 4k (15
mg/kg, bid. for 2 weeks ) suppressed tumor growth (T/C 29%) in an RPMI8226 mouse xenograft model.
Keywords
CDK8, CDK19, Cyclin-dependent kinases (CDKs), transcriptional regulation, STAT1, MCM3, RPMI8226,
SW480, DMG, pyridylacrylamide

1. Introduction
Cyclin-dependent kinase 8 (CDK8) belongs to the cyclin-dependent family of protein kinases and
is recently highlighted as oncogene, which forms mediator complex with such as MED12 and MED13.1
The mediator complex phosphorylates C-terminal domain of RNA polymerase (RNAP) II and
transcriptional factors, which regulates oncogene and tumor supressor transcription.2 Other reports have
shown that CDK8 plays roles in modulation of the transcription factors including Wnt/β-catenin pathway,
Notch and TGF-β,3 which supported the cancer-relevant activities of CDK8. In addition, copy number
gain and high expression of CDK8 are observed in 60% of CRC tumors which correlates with poor
prognosis of colorectal cancer patients.4 CDK19 is a close homologue of CDK8 and its biological
function and impact are ill defined.1b
Inhibition of CDK8 kinase activity disrupts transcription, resulting in context dependent cellular
stress.5 So far, several groups have disclosed CDK8 and CDK19 dual inhibitors,6 which includes cortistatin
derivatives,7 SNX-2 class compounds,8 3,4,5-trisubstitutedpyridines,9 thienopyridine analogues10 and
benzylindazole series.11 Although the research around CDK8 is hot topic, only few selective inhibitors
have been discovered.
Herein we describe our efforts to discover and optimize a series of novel and selective CDK8/19
inhibitors.12 Among several classes of hits identified through a high-throughput screening (HTS)
campaign, 4-phenylpyridine compound 1 (Figure 1)13 was judged to be an attractive starting point to
initiate a medicinal chemistry program due to its potency (CDK8 IC50 = 80 nM and CDK19 IC50 = 100 nM),
moderate rate of metabolism in liver microsomes (5.1 mL/min/mg in human and 6.7 mL/min/mg in mouse)
and acceptable physicochemical properties. In addition, 1 showed good selectivity against an in-house
panel of 27 kinases (less than 20% inhibition at 1 µM), which implied intrinsic high kinase selectivity for
this series. Thus 1 was selected as a starting point for further investigation. Our medicinal chemistry
efforts were largely focused on the enhancement of selective enzymatic and cellular potency, in addition to
the concurrent pursuit of favorable pharmacokinetic (PK) and ADME properties, including reduction of
drug-drug interaction (DDI) potential.
Figure 1. Representative hit compound identified through HTS
In the course of profiling a wide variety of HTS hits, we confirmed high correlation between the
inhibitory potencies against CDK8 and CDK19, as expected based on the very high structural similarity of
these two proteins (100% identity in the active site, defined as 4 Å from ligand).14 Hence, during our
optimization campaign we sought to enhance CDK8 inhibitory activity, with CDK19 inhibitory activity

measured as a confirmatory assay. While not discussed in detail here, CDK8 and CDK19 inhibitory
potencies showed an essentially linear relation over the compounds presented in this paper, with IC50 ratios
generally ranging from 1:1 to 1:3.9,11,12
For the identification of potent CDK8/19 inhibitors, we employed a modular structure-based
drug design (SBDD) approach toward introducing modifications into our lead molecule in three separate
regions. Based on a published cocrystal structure of an inhibitor bound to CDK8 (PDB ID: 3RGF)15 that
was obtained at the start of our investigation, a computational model of our lead 1 bound to CDK8 was
constructed, as shown in Figure 2, which suggested that the central pyridine core binds to the hinge region
via a hydrogen bond between the pyridine nitrogen atom and the backbone of Ala100 and this is consistent
with crystallographically observed binding recently reported by other groups.9 According to this model,
the p-fluorophenyl moiety is oriented toward the back pocket region of CDK8, and the proximal Lys52,
Glu66 and Asp173 of the DMG motif of CDK8 seemed attractive targets for forming additional hydrogen
bonds with this molecule (Figure 2A). We therefore designed new substructures in this region as
illustrated by compounds of type 2 (Figure 2B) where the p-fluorophenyl ring is replaced with monocyclic
or fused bicyclic heterocycles with potential hydrogen
A)
B)

Figure 2. (A) Docking model of 1 (green) bound to CDK8, based on the reported structure (PDB ID, 3RGF).
Key amino acid residues are depicted as sticks (gray). (B) General designs for novel CDK8/19 inhibitors.
bond acceptor and/or donor functionalities. Along a different vector, substitution of the hinge-binding
pyridine was conducted with a goal of enhancing potency (generally illustrated by compounds of type 3 in
Figure 2B). As with many reported kinase inhibitors, the possibility of forming hydrogen bonds at two
points in the kinase hinge region was explored. From our docking studies, the backbone carbonyl
oxygens of Asp98 and Ala100 were predicted to sit in close proximity to the 6- and 2-positions of the
pyridine ring, respectively. Thus substituents with potential hydrogen bonding donor ability were
introduced at these sites. An additional modification to the hinge binding motif was the introduction of a
fluoro atom at the 5-position of the pyridine core. This design was based on the prediction of limited
space around this site and the expectation of a favorable hydrophobic interaction between the fluoro atom
with Ile79 and/or Phe97 in this region. The benzamide portion of 1 extends along the front of the ATP
binding site and the morpholino methyl moiety is directed outward the solvent region, forming a CH-π
interaction with Trp105. Thus, this portion of the molecule was expected to allow diverse sets of
substructures in terms of maintaining activity, and we modified this portion of the molecule with an aim
toward improving PK and ADME properties (generally illustrated as compounds of type 4 in Figure 2B).
2. Chemistry
The pyridyl acrylamide derivatives reported in these studies were prepared from commercially
available starting material 4-chloronicotinaldehyde hydrochloride (5) via the modular approach generally
outlined in Scheme 1, with additional examples shown in Schemes 2–8.16 The Horner-Wadsworth-Emmons
reaction of 5 with triethyl phosphonoacetate afforded the desired (E)-alkene 6, with minimal formation of
the (Z)-regioisomer (<5%). Hydrolysis of 5 with sodium hydroxide provided carboxylic acid 6 in nearly
quantitative yield, with these two intermediates acting as the primary starting materials for further
syntheses.
Tributyltin intermediate 7 was prepared from 6 by Pd(PPh3)4 mediated coupling, allowing entry
into isothiazole compound 8a by Pd(PPh3)4 catalyzed coupling with 4-bromoisothiazole. Subsequent
hydrolysis of 8a to 9a with in aqueous sodium hydroxide, followed by amide condensation with
4-(morpholinomethyl)aniline and HATU afforded isothiazole derivative 2a.
Alternatively, palladium-mediated coupling of 6 and the appropriate boronic acids or esters using
SPhos or CyJohnPhos as phosphine ligand afforded a more direct route to formation of pyrazole esters
8b–e. Subsequent hydrolysis of the esters to their corresponding carboxylic acids with sodium hydroxide
was conducted in generally good yields (71–83% for 9c–e, 26% for highly water-soluble derivative 9b).
The preparation of amide derivatives 2g, 2g, 2k and 2m was accomplished by the reaction of
4-(morpholinomethyl)aniline and acids 9b–e, respectively, in the presence of HATU. Reduction of the
acrylamide double-bond was conducted for 2f and 2k to produce 2o and 2p by hydrogenation with Pd/C.
Derivatization of this modular scaffold could also be conducted in the opposite order, starting
from amide bond formation between 10 and 4-(morpholinomethyl)aniline, conducted by formation of the

acid chloride of 10 with the Vilsmeier reagent, producing intermediate 11. Key intermediate 11 was then
subjected to Pd-mediated coupling, using SPhos or CyJohnPhos phosphine ligands, with the appropriate
boronic acids or esters to yield 2d, 2h, 2i, 2l, and 2n. As shown in Scheme 2, inverse coupling of aryl
bromides to boronic ester based intermediate 13 was also found to be feasible. The Heck reaction of aryl
bromide 12 with tert-butyl acrylate in the presence of Pd(OAc)2 and tri-o-tolylphosphine afforded key
boronic ester intermediate 13. This intermediate was then subjected to palladium-catalyzed coupling with
a series of heterocyclic bromides, followed by tert-butyl ester removal, to afford bi-aryl heterocyclic acids
14a–d. These, in turn, were condensed with 4-(morpholinomethyl)aniline under HATU conditions to
afford amide derivatives 2b, 2c, 2e, and 2j.
Scheme 1. Synthesis of 2a, 2d, 2f, 2g, 2i, and 2k−p. Reagents and conditions: (a) ethyl 2-(diethoxyphosphoryl)acetate, NaH,
THF, 0 °C to rt, 86%; (b) 1,1,1,2,2,2-hexabutyldistannane, LiCl, Pd(PPh3)4, toluene, 120 °C, 19%; (c) 4-bromoisothiazole,

cesium fluoride, copper (I) iodide, Pd(PPh3)4, DMF, 100°C, 75%; (d) boronate ester. Method A: SPhos, SPhos precatalyst
1st generation, Cs2CO3, DME, water, 130 °C, microwave irradiation, 66%. Method B: 2-(dicylohexylphosphino)biphenyl,
Pd2(dba)3, K2CO3, DME, water, 75 °C, 74−92%; (e) NaOH, water, THF, EtOH, rt, then HCl or AcOH, 26−83%; (f)
4-(morpholinomethyl)aniline, HATU, DIEA, DMF, rt, 34−83%; (g) H2 (1 atm), Pd/C, EtOAc, 40 °C, 8−9%; (h) NaOH,
water, THF, EtOH, rt, then aq. HCl, 96%; (i) oxalyl dichloride, DMF (5 drops), THF, rt, then 4-(morpholinomethyl)aniline,
DMA, rt, 89%; (j) boronate ester or boronic acid. Method A: Cs2CO3, SPhos, SPhos precatalyst 2nd generation, DME, water
130 °C, microwave irradiation, 43−71%. Method B: [1,1'-biphenyl]-2-yldicyclohexylphosphine, Pd2(dba)3, Cs2CO3, DME,
water, 80 °C, 45%.
Scheme 2. Synthesis of 2b, 2c, 2e and 2j. Reagents and conditions: (a) tert-butyl acrylate, TEA, tri-o-tolylphosphine,
Pd(OAc)2, DMF, 130 °C, 67%; (b) (i) ArBr, Cs2CO3, SPhos, SPhos precatalyst 2nd generation, DME, water, 130 °C,
microwave irradiation, (ii) TFA, rt, 34−95% in 2 steps; (c) 4-(morpholinomethyl)aniline, HATU, DIEA, DMF, rt, 73−100%.
Scheme 3. Synthesis of 3a. Reagents and conditions: (a) ethyl 2-(diethoxyphosphoryl)acetate, NaH, THF, DMF, 0 °C to rt,
98%; (b) 1-methyl-4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)-1H- pyrazole,[1,1'-biphenyl]-2-yldicyclohexylphosphine,
Pd2(dba)3, Cs2CO3, DME, water, 80 °C, 63%; (c) tert-butyl carbamate, XPhos, Pd2(dba)3, Cs2CO3, THF, 70 °C; (d) NaOH,

water, THF, EtOH, rt, then 2 M HCl, 53% in 2 steps; (e) 4-(morpholinomethyl)aniline, HATU, DIEA, DMF, rt; (f) 2 M HCl,
MeOH, 45 °C, 34% in 2 steps.
Compound 3a, an amino-substituted analogue of 2f, was prepared by the synthetic sequence
outlined in Scheme 3. Aldehyde 15 was converted to acrylic ester 16 by Horner-Wadsworth-Emmons
conditions.
Coupling of 16 with the appropriate N-methylpyrazole boronate ester under
Pd2(dba)3/CyJohnPhos mediated conditions afforded chloropyridine intermediate 17. The chloro group
was substituted with tert-butyl carbamate by Pd2(dba)3/XPhos mediated coupling to afford 18, followed by
ester hydrolysis to carboxylic acid 19.
Amide condensation under HATU conditions with
4-(morpholinomethyl)aniline yielded 20, followed by removal of the Boc-protecting group with HCl lead
to the desired compound 3a.
Hydroxypyridyl derivative 3b, perhaps better described as the pyridone tautomer, was prepared
according to the route shown in Scheme 4. Pyridine N-oxide compound 21 was prepared from 8c and
mCPBA, followed by conversion to acetyl-protected hydroxylpyridine compound 2217 by heating in acetic
anhydride. One-step hydrolysis of both the ethyl ester and acetate group of 22 with NaOH, followed by
condensation of the resulting acid 23 with 4-(morpholinomethyl)aniline in the presence of HATU afforded
the desired hydroxypyridyl compound 3b.
Scheme 4. Synthesis of 3b. Reagents and conditions: (a) mCPBA, EtOAc, rt, 66%; (b) acetic anhydride, EtOAc, 120 °C,
43%; (c) 2 M NaOH, THF, EtOH, rt; (d) 4-(morpholinomethyl)aniline, HATU, DIEA, DMF, rt, 20% in 2 steps.
In a procedure analogous to that outlined in Scheme 1, Scheme 5 shows the synthetic route used
to prepare fluoropyridyl derivatives 3c and 3d. Starting material 4-chloro-5-fluoronicotinaldehyde (24)
was reacted with ethyl 2-(diethoxyphosphoryl)acetate or tert-butyl 2-(diethoxyphosphoryl)acetate under
Horner-Wadsworth-Emmons conditions to provide compounds 25a and 25b, respectively. Coupling with

the appropriate pyrazole boronate esters in the presence of Pd2(dba)3/CyJohnPhos catalysis afforded ester
intermediates 26a and 26b. Methyl ester 26a was converted to the corresponding carboxylic acid 27a by
hydrolysis with NaOH, while tert-butyl ester 26b was treated with TFA and neutralized to obtain
carboxylic acid 27b. Acids 27a and 27b were condensed with 4-(morpholinomethyl)aniline in the
presence of HATU to afford amide compounds 3c and 3d.
R'
N
N
Cl
O
Cl
N
O
O
a
b
F
F
R
F
R
H
O
O
N
N
26a (R = Et, R' = Me)
24
25a (R = Et)
26b
(R = tBu, R' = cPr)
25b (R = tBu)
R'
R'
N
N
N N
O
O
N
d
c
F
F
O
OH
N
H
N
N
27a (R' = Me)
27b
3c (R' = Me)
3d (R' = cPr)
(R' = cPr)
Scheme 5. Synthesis of 3c and 3d. Reagents and conditions: (a) for 25a: ethyl 2-(diethoxyphosphoryl)acetate, NaH, THF,
DMF, 0 °C to rt, 90%; for 25b: tert-butyl 2-(diethoxyphosphoryl)acetate, NaH, THF, 0 °C to rt, 77%; (b) boronate ester,
[1,1'-biphenyl]-2-yl dicyclohexylphosphine, Pd2(dba)3, Cs2CO3, DME, water, 85 °C, 75−82%; (c) for 27a: 2 M NaOH, THF,
EtOH, 0 °C to rt, 60%; for 27b: TFA, then 2 M NaOH, 0 °C to rt, 93%; (d) 4-(morpholinomethyl)aniline, HATU, DIEA,
DMF, rt, 69−72%.
Modifications at the morpholine moiety were conducted by the routes shown in Scheme 6.
Compounds 4a, 4b, and 28 were prepared by amide coupling of 9c and the appropriate amines. The
mesylate of 28 was prepared, compound 29, and subjected to nucleophilic substitution with the appropriate
bicyclic morpholine derivative to yield 4e. In a related approach, Boc-protected anilines 33c and 33i
were prepared from amino alcohol 30 by a similar sequence of Boc protection to give 31, followed by
mesylation of the benzylic alcohol to afford 32, and finally nucleophilic substitution with the appropriate
amines to provide the desired intermediates 33c and 33i. Boc-deprotection followed by amide
condensation with 9c to yield products 4c and 4i.
Reductive amination routes to related derivatives were also utilized. Cyclopropyl compound 34
was prepared analogously to compound 28, followed by MnO2 oxidation of the benzylic alcohol to yield
aldehyde 35. Reductive amination with the appropriate amines in the presence of 2-picoline borane
complex afforded compounds 4g and 4h. Alternatively, initial synthesis of aniline intermediates 38d,

Scheme 6. Synthesis of 4a−4k. Reagents and conditions: (a) R-NH2, HOBt, EDCI, DMF, rt, or R-NH2, HATU, DIEA, DMF,
rt, 34−99%; (b) MsCl, NEt3, CH2Cl2, 0 °C to rt; (c) amine hydrochloride, NaH, DMF, rt, 4% in 2 steps; (d) Boc2O, DIEA,
THF, reflux, 59%; (e) MsCl. NEt3, CH2Cl2, rt, 68%; (f) amine hydrochloride, NaH, DMF, rt, 19−21%; (g) (i) HCl, EtOH, rt;
(ii) 9c, HATU, DIEA, DMF, rt, 16−22%; (h) 9d, HATU, DIEA, DMF, rt, 69%; (i) MnO2, THF, 63%; (j) amine
hydrochloride, 2-picoline borane complex, MeOH, AcOH, 19-41%; (k) amine hydrochloride, dichloroethane, rt, then
NaBH(OAc)3, rt, 27−54%; (l) Fe, NH4Cl, EtOH, water, 70 °C, 62−80%; (m) 9c or 9d, HATU, DIEA, DMF, rt, 13−57%.
38f, and 38j could be accomplished by reductive amination of aldehyde 36, followed by reduction of the
nitro group of compounds 37d, 37f, and 37j. These anilines could then be condensed with 9c or 9d in the
presence of HATU to yield amide derivatives 4d, 4f, 4j, and 4k.

Tetrahydroisoquinoline-based derivatives 4l-4n were prepared as shown in Scheme 7. Amide
condensation of amines 39a or 39b with 9c under HATU conditions afforded amine compounds 40a and
40b, which we then subjected to standard Boc-deprotection and alkylation to afford trifluoroethyl
compounds 4l and 4n, respectively. Alternatively, 2,2,2-trichloroethyl carbamate (Troc) protection and
Boc-removal from 39a, followed by alkylation of 42, afforded aniline intermediate 43. This compound
was then condensed with 9d under HATU conditions to afford 4m.
Scheme 7. Synthesis of 4l−4n. Reagents and conditions: (a) 9c, HATU, DIEA, DMF, rt, 76−95%; (b) HCl, EtOAc or CPME,
MeOH, 86%; (c) NEt3, TfOCH2CF3, CH3CN, rt to 50 °C, 32−51%; (d) TrocCl, NEt3, THF then HCl/EtOAc, 90%; (e)
TfOCH2CF3, NEt3, DMF; (f) Zn, AcOH, 69% in 2 steps; (g) 9d, HATU, DIEA, DMF, 65%.
3. Results and discussion
CDK8 and CDK19 binding activities of the newly synthesized compounds were evaluated by
kinase tracer assay, in which inhibition of binding of kinase tracer 236 to CDK8 or CDK19 was detected by
fluorescence measurement. Cellular activity of our compounds was evaluated by an in-cell western
(ICW) assay, in which each compound was incubated with SW480 cells (human colorectal
adenocarcinoma) fixed on a plate, and the inhibition of Ser727 phosphorylation of STAT1 (a direct
substrate of CDK818) following IFN-γ stimulation was measured. In vitro activities measured in the above
assay types are reported as IC50 and EC50 values, respectively.
The effect of replacement of the p-fluorophenyl moiety with various heterocycles is illustrated in
Table 1. According to the strategy described in the Introduction, a variety of heterocyclic derivatives
were prepared, with a particular focus on 5-membered heterocycles (2a–g) and benzimidazoles (2h–j).
First, 4-isothiazoyl and 5-thiazoyl analogues (2a and 2b) were prepared, which showed 2–3 times more

potent CDK8 binding activities than 1. The next approach taken was the introduction of a substituent
containing a hydrogen bonding acceptor into the 5-membered rings (2c–e). Of particular interest was the
finding that 2-methoxythiazole 2c and 2-aminocarboxythiophene 2e displayed more than a 9-fold increase
in binding activity compared to 1, which supported our aim to target Lys52, Glu66 and/or Asp173 as a
hydrogen bonding partner. The difference in potency between 2c and 2d may be derived from the
thiazole nitrogen atom restricting the direction and donor ability of the methoxy oxygen lone pair to
interact with the protein. Encouraged by these results, derivatives containing other 5-membered
heterocycles (pyrazoles 2f and 2g) were prepared. The binding potency of 1-methyl pyrazole 2f was
found to be 12-fold higher than 1, and 1-H pyrazole 2g exhibited a further 12-fold increase in binding
potency, with an IC50 value of less than 1 nM. Although our docking study suggested that these pyrazole
moieties would not be expected to form a direct hydrogen bonding interaction with Lys52, Glu66 or
Asp173, it is thought that they could interact via a water molecule. The increased activity observed for
compounds 2f and 2g was also reflected in the cellular assays; both compounds exhibited more than
17-fold greater potency in the ICW assay relative to 1. The preparation of benzimidazoles (2h–j) was a
further attempt at targeting interaction with Lys52, Glu66 and/or Asp173, with 2h and 2j showing high
potency in both the biochemical and cellular assays. These results supported the validity of our drug
design strategy, however, further study of this compound class was discontinued due to the generally potent
hERG inhibitory activity (data not shown).
Among the most potent compounds identified above (2c, 2f and 2g), methylpyrazole derivative 2f
showed the most favorable PK profile in mouse cassette dosing studies (Table 2 includes the data for 2f)
and became the focus of further optimization efforts. In vivo metabolite analysis of 2f in mice revealed
the primary sites of metabolism to be the morpholine moiety and the 1-methyl group of the pyrazole
(Figure 3). Thus, further modification of the pyrazole 1-methyl group was conducted with a goal of
improving metabolic stability and PK, while maintaining enzymatic potency. The approaches taken in
these studies were 1) blockage of the metabolically unstable methyl site and 2) installation of a bulky
substituent into the methyl group, leading to the preparation of compounds 2k–n (Table 1). Generally
speaking, variation of the alkyl group in this position showed limited tolerance with respect to potency,
namely, cyclopropyl derivative 2k possessed comparable activity to parent compound 2f while other
fluorinated alkyl substituents such as 2l and 2m showed reduced enzymatic potency. Interestingly,
potency was maintained for benzyl derivative 2n, implying that this moiety could be oriented toward the
back pocket region of CDK8. Subsequent mouse cassette dosing studies demonstrated a better PK profile
for cyclopropyl derivative 2k relative to methyl pyrazole 2f, while benzyl compound 2n showed poor

Table 1
Biological activities of 1 and derivatives 2a−n
CDK8a
pSTAT1b
compd.
R
IC50 (nM)
EC50 (nM)
80
(69-92)
370
(201-673)
1
27
(21-35)
110
(51-222)
2a
2b
2c
2d
2e
2f
41
(34-50)
140
(50.2-399)
8.1
(6.5-10.0)
12
(5.4-26)
28
(22-34)
97
(30-315)
8.3
(6.3-11.0)
690
(278-1712)
6.5
(5.8-7.2)
13
(10-18)
0.54
(0.31-0.94)
21
(14-30)
2g

Table 1 . (continued)
CDK8a
pSTAT1b
compd.
R
IC50 (nM)
EC50 (nM)
1.5
(1.0-2.4)
10
(5.7-18)
2h
4800
(3065-7502)
2i
2j
2k
2l
>10000
2.5
(2.1-3.0)
2.4
(1.5-3.9)
6.4
(5.0-8.3)
11
(4.5-26)
24
(21-27)
50
(16-152)
28
(23-35)
150
(78-290)
2m
4.5
(3.6-5.7)
28
(15-54)
2n
a
Inhibition of kinase tracer 236 binding to human CDK8 (n = 2, values in parentheses indicate 95% confidence
interval).
Inhibition of STAT1 phosphorylation (at Ser727) induced by IFN-γ in SW480 cells fixed on a 384-well plate (values
b
in parentheses indicate 95% confidence interval).

Table 2
PK profiles and metabolic stabilities of selected compounds
Metabolic stabilityf
(mL/min/mg)
a
b
F e
Cl
total
Cmax
AUC poc MRT pod
compd.
(mL/h/kg) (ng/mL) (ng/mL*h)
(h)
1.5
1.5
1.2
(%)
21.9
51.2
15.5
human
22
mouse
77
2f
2k
2n
2967
1523
5765
47.5
207.7
15.7
76.9
339.7
26.9
25
64
78
118
−8 h).
PK data obtained from mouse cassette dosing studies (1 mg/kg, po, 0
a Cltotal : mean plasma clearance.
b Cmax: maximum concentration.
c AUCpo.: area under the curve.
d MRTpo.: mean residence time.
e F: bioavailability in mouse cassette dosing studies (0.1 mg/kg, iv and 1 mg/kg, po).
f Metabolic stability in human or mouse hepatic microsomes.
metabolic stability in mouse microsomes. Based on these results, 1-methyl and 1-cyclopropyl pyrazoles
2f and 2k were chosen as templates for further SAR studies in other sections of this chemotype.
As a next focus, modifications were made to the hinge-binding pyridine core with the aim of
improving potency and the resulting compounds 3a–d are shown in Table 3. Our first design strategy was
the introduction of a two-point hydrogen bond utilizing hinge residue Ala100 and proximal Asp98 (Figure
2A), which resulted in the preparation of compounds 3a and 3b possessing an amino or hydroxyl group
adjacent to the pyridine N-position. However, the potencies of both compounds proved to be dramatically
decreased compared to the parent 2f. Subsequently, we undertook the installation of a fluoro atom at the
5-position of the pyridine core, resulting in the preparation of 3c and 3d. The fluoro atom was expected to
sit in a small space adjacent to Ile79 and Phe97 based on our computational model (Figure 2A) and potent
enzymatic activity was maintained for these compounds. In addition, the inhibitory activity of STAT1
phosphorylation (ICW assay) was increased for both compounds with respect to the corresponding
non-substituted analogues (2f and 2k).

Figure 3. Presumed structures for metabolites of 2f (in mouse plasma) based on MS analysis

During the process of profiling our lead compounds 2f and 2k, we found that they exhibit
time-dependent inhibition (TDI) of CYP3A4 in our routine in-house in vitro ADME assay (Table 3). Of
particular interest was the finding that 3-fluoropyridine derivatives 3c and 3d showed increased TDI
potential. TDI is also referred to as mechanism-based inhibition, which results from irreversible binding
of the test compound itself or a reactive metabolite to CYP3A4. Inhibitory activity against CYP3A4 is
considered a substantial concern in terms of drug safety, due to the potential for decreasing drug clearance
and the induction of unpredictable toxicity (hepatotoxicity, idiosyncratic drug toxicity, etc.). Thus, we
sought to reduce the TDI activity of our compounds during optimization. First, we postulated that the
alkene linker between the hinge-binding pyridine and benzamide moiety might be responsible for TDI, but
replacement of this linker with an ethylene structure showed little influence on TDI potential (Figure 4, 2o
and 2p). In addition, these modifications substantially decreased both enzymatic and cellular activities.
Thus our focus was shifted to other
Table 3
Biological activities of 2f, 2k and substituted pyridine derivatives 3a−d
CDK8a
pSTAT1b
CYP3A4 TDIc
1
R
50
50
compd.
IC (nM)
EC (nM)
(% remaining @ 30 µM)
6.5
(5.8-7.2)
13
(9.6-18)
2f
Me
81
54
6.4
(5.0-8.3)
11
(4.5-25)
2k
3a
3b
3c
3d
Pr
c
990
(787-1248)
1300
(392-4389)
NDd
89
Me
Me
Me
> 10000
> 10000
5.2
(4.5-5.9)
3.6
(2.3-5.5)
60
6.5
(5.7-7.5)
2.3
(0.95-5.3)
Pr
46
c
a Inhibition of kinase tracer 236 binding to human CDK8 (n = 2, values in parentheses indicate 95% confidence interval).
b
Inhibition of STAT1 phosphorylation (at Ser727) induced by IFN-γ in SW480 cells fixed on a 384-well plate (values in
parentheses indicate 95% confidence interval).
c
% remaining of CYP3A4 activity. Tested compounds were pre-incubated in human liver microsomes for 60 min.

Testosterone was added to microsomes and incubated for 20 min.
d ND: not determined.
Figure 4. Effects on biological activity and TDI potential upon modification of the alkene linker.
sections of the molecule, especially the right-hand region of this series. As illustrated in Figure 4, one of
the more metabolically labile portions of these compounds was found to be the morpholine moiety, and we
postulated that the morpholine group could be converted to a reactive imine species during the metabolic
process based on metabolite identification studies using LCMS. Therefore, we prepared derivatives in
which the morpholine moiety was replaced with other basic amines, such as those shown in Table 4.
Initially, we prepared compounds 4a and 4b, which demonstrated the importance the of the benazmide
phenyl ring for potent CDK8 binding activity. Compared to 2f, the enzymatic potency of 4b was reduced
by more than a 7-fold and the morpholinomethyl moiety was also found to be important for potent activity.
In addition, the solubility of 4b in pH6.8 was inferior to 2f, which showed that this morpholine was a key to
address good physicochemical property of the compound. While not discussed here in detail, the phenyl
ring was predicted to be close to His106 and important for high binding affinity via a CH-π interaction
based on our computational model, and replacement with aliphatic or heteroaromatic rings was found to
result in significantly reduced enzymatic potency (See Supplementary Figure 2). In terms of CYP3A4
TDI potential, compound 4b was found to show low risk, which supported our assumption that the
morpholine was playing a major role in the TDI potential for this chemical series. Thus, substituted
morpholine analogues 4c and 4d, which were expected to be more metabolically stable, were synthesized
but showed no advantage over 2f in terms of activity or CYP3A4 inactivation. An important clue
regarding the origin of the observed TDI effect was obtained from the preparation and study of bicyclic
morpholine derivatives 4e–h. A previous report from Genentech on mTOR inhibitors suggested that a
bicyclic tropane system could be employed to prevent reactive metabolite formation due to the high strain
energy of the resulting iminium bridgehead double bond (Bredt’s rule).19 Notably, our bicyclic
morpholine compounds showed dramatically improved TDI profiles, with compounds 4f and 4g showing
comparable in vitro activity to 2f and 2k. Although both compounds show somewhat less favorable PK
profiles than 2l (data not shown), these derivatives encouraged us to further investigate pyrrolidine and
azetidine derivatives 4i–k as simplified morpholine analogues with increased ring strain. While 4i did not

display reduced CYP3A4 TDI potential, azetidines 4j and 4k showed almost complete elimination of TDI
risk. We postulate that inclusion of the azetidine ring reduces TDI of CYP3A4 by virtue of the
conformationally restricted 4-membered ring prohibiting double-bond iminium formation. Moreover,
compound 4k showed high potency in both the biochemical and cellular assays, and sufficient level of
solubility in pH6.8. Compounds 4l–n represent further modifications of the right-hand region.
Preparation of these saturated fused ring systems aimed to remove the morpholine structure and gain more
effective CH-π interaction with Trp105. Tetrahydroisoquinoline replacement (4l–n) resulted in high
enzymatic and cellular potency with moderate to strong CYP3A4 TDI potential.
Table 4
Biological activities of 2f, 2k and derivatives 4a−n

CDK8a
pSTAT1b
CYP3A4 TDIc
Solubilitye
compd.
R1
R2
IC50 (nM)
EC50 (nM)
(% remaining @ 30 µM) pH6.8 (µg/mL)
6.5
(5.8-7.2)
13
(10-18)
2f
Me
81
54
68
58
6.4
(5.0-8.3)
11
(4.5-25)
2l
4a
4b
4c
4d
4e
4f
Pr
c
NDd
97
NDd
26
Me
Me
Me
Me
Me
Me
H
> 10000
> 10000
51
(43-61)
110
(52.1-237)
Ph
20
(14-28)
27
(16-47)
59
>75
NDd
NDd
>82
>92
60
6.7
(4.8-9.4)
16
(12-22)
84
6.3
(4.3-9.3)
17
(10-29)
99
16
(13-20)
19
(12-29)
100
101
108
9.6
(8.8-10.6)
3.1
(1.2-7.9)
4g
4h
Pr
Pr
c
c
6.7
(5.0-9.0)
5.5
(2.1-14.4)

Table 4. (continued)
CDK8a
pSTAT1b
CYP3A4 TDIc
Solubilitye
compd.
R1
R2
IC50 (nM)
EC50 (nM)
(% remaining @ 30 µM) pH6.8 (µg/mL)
6.2
(5.1-7.5)
17
(11-25)
4i
Me
69
98
99
91
88
22
33
44
11
(8.6-13)
84
(27-262)
4j
4k
4l
Me
2.5
(2.2-2.8)
8.4
(6.1-11.5)
Pr
30
c
4.5
(3.0-6.8)
7.0
(4.8-10.2)
Me
>81
3.9
8.9
7.1
(5.5-9.1)
14
4m
4n
Pr
c
(7.8-24.6)
3.2
(2.6-4.1)
8.4
(5.1-14.0)
Me
a Inhibition of kinase tracer 236 binding to human CDK8 (n = 2, values in parentheses indicate 95% confidence interval).
b
Inhibition of STAT1 phosphorylation (at Ser727) induced by IFN-γ in SW480 cells fixed on a 384-well plate (values in
parentheses indicate 95% confidence interval).
c
% remaining of CYP3A4 activity. Tested compounds were pre-incubated in human liver microsomes for 60 min.
Testosterone was added to microsomes and incubated for 20 min.
d ND: not determined.
Kinase selectivity of 4k. As a representative of our CDK8/19 inhibitors, compound 4k
(inhibitory potency: CDK8 IC50 = 1.3 nM, CDK19 IC50 = 3.9 nM) was screened against a panel of 456
kinases in a binding assay to determine its selectivity (KINOMEscanTM from DiscoveRx Corp.).20 At a
concentration of 300 nM, compound 4k did not bind to any of the 456 kinases tested with >65% binding
affinity (<35% control), except CDK8 and CDK19 (Supplementary Figure 1). This high level of
selectivity is believed due to a combination of the effective interaction of 4k with several amino acid
residues characteristic for CDK8. For instance, according to our docking studies, gatekeeper Phe97 and
the unusual DMG motif of the CDK8 protein are located in proximity to the inhibitor (Supplementary
Figure 2A). Analysis of kinase sequences3 showed that these features are characteristic of CDK8 and
interaction with these residues would differentiate it from other kinases (Supplementary Figure 2B).

In vitro pharmacology of 4k. According to the results discussed above, we focused further
studies on azetidine compound 4k, which showed potent in vitro activity and virtually no CYP3A4 TDI.
Table 5 shows the results of Western blotting and cell proliferation for three representative cell lines:
SW480 (colon cancer), RPMI8226 (human myeloma) and EOL1 (AML). While phosphorylation
inhibition of STAT1 in SW480 cells was measured above as a primary screen of in vitro cellular activity,
inhibition of cell growth in this cell line could not be suppressed below 50%; therefore, this cell line was
not utilized for measurement of cell growth inhibition. Compound 4k suppressed phosphorylated-STAT1
levels in all cell lines tested, and also demonstrated potent anti-proliferative activity against RPMI8226
(GI50: 21.0 nM) and EOL1 (GI50: 6.5 nM). Based on these results, we have found that phospho-STAT1 is a
target engagement marker, which is suppressed by 4k in these and other cell lines, but does not always
reflect the sensitivity of cell growth. Other predictive markers for cell growth inhibition are under
investigation and will be published in due course.
Table 5
STAT1 suppression after 1 day treatment with 4k (Western blotting: WB) in various cell lines, and the corresponding
anti-proliferative activity (GI50).
a
Cell proliferation was determined by ATP content (SW480 and RPMI8226) or by nuclear count (EOL1). Cells were
exposed to a compound for 168 h.
In vivo pharmacology of 4k. The favorable PK profile of 4k in mice and high metabolic
stability against human microsomes (Table 6) lead us to consider this compound as a promising candidate
for further pharmacological evaluation.

Antitumor efficacy studies were conducted using compound 4k in order to demonstrate the
effects of CDK8/19 inhibition on tumor growth in mice. Compound 4k was administered orally twice a
day at 15 and 30 mg/kg for 2 weeks to mice bearing RPMI8226 tumor xenografts. Tumor growth was
measured over the course of the study and is presented in Figure 5A. When compared to vehicle control,
compound 4k demonstrated 29% and -2% T/C values (ratio of tumor volume for treatment group vs.
control) at doses of 15 and 30 mg/kg, respectively. Furthermore, compound 4k was well tolerated with no
severe changes in body weight at a dose of 15 mg/kg while 12% loss of body weight was observed in 30
mg/kg dosing (Figure 5B). On the last day of the studies, mouse blood was sampled at timepoints of 2 h
and 8 h post-dose, and inhibition of phosphorylation of STAT1 was measured as a PD marker. Compound
4k demonstrated a decrease in production of phosphorylated-STAT1 at each time point (Supplementary
Figure 3), indicating in vivo inhibition of CDK8 by our novel compound.

Table 6
Growth inhibitory activity, PK profile in mouse, and metabolic stability of compound 4k
Metabolic stabilityf
(mL/min/mg)
b
Cell Growtha
GI50 (nM)
Cmax
AUC poc MRT pod
F e
compd.
(%)
(ng/mL) (ng/mL*h)
(h)
human
mouse
4k
21
321 370
1.6
22.5
15
97
a Inhibition of RMPI8226 tumor cell growth.
b
−8h)
Cmax: maximum concentration (1 mg/kg, po, 0
.
c
−8h)
AUCpo: area under the curve (1 mg/kg, po, 0
.
d
−8h)
MRTpo: mean residence time in a discrete study (1 mg/kg, po, 0
.
eF: bioavailability (1 mg/kg, po, 0−8h).
f Metabolic stability in human or mouse hepatic microsomes.
A)
B)
Figure 5. (A) Antitumor activity and (B) body weight loss of 4k in a RPMI8226 mouse xenograft model.
4. Conclusion
In this paper, we report the discovery of a novel class of CDK8/19 inhibitors. Starting from
HTS hit 1, an optimization campaign was initiated with a primary goal of potency enhancement, as well as
achieving favorable PK and ADME profiles. By the utilization of a computational model of compound 1
bound to CDK8, constructed from a reported co-crystal structure of ligand-bound CDK8, we efficiently
optimized the our lead compound using a modular structure-based approach. First, heteroaromatic rings
were introduced as alternatives to the p-fluoro phenyl moiety, leading to the identification of 1-methyl and
1-cyclopropyl pyrazole analogues with substantially enhanced potency and favorable PK properties. In
addition, the introduction of a fluoro atom into the central pyridine core maintained enzymatic activity and
significant enhancement of cellular activity. Although moderate to strong risk of CYP3A4 TDI was
observed in the course of working on this series, changing the pendant morpholine ring to structurally
restricted ring systems (bridged morpholine or azetidine) was found to be an effective method to alleviate

this issue, which finally led to the identification of compound 4k. This compound demonstrated
anti-tumor efficacy in RPMI8226 mouse xenograft studies and excellent selectivity toward CDK8/19
inhibition against a panel of 456 kinases, which supported our expectation that inhibition of CDK8/19
holds potential as a new approach for treating cancer. According to the results discussed in this paper,
compound 4k was selected as a development candidate and progressed into toxicity studies.
5. Experimental section
5.1.
Docking studies. Docking study with reported CDK8 crystal structure (PDB ID; 3RGF) was
performed using GOLD version 5.2 (Cambridge Crystallographic Data Centre: CCDC).
5.2.
Ligand displacement assay against CDK8 and CDK19 kinases. Ligand displacement assay
was conducted with Tb-labeled anti-GST antibody (CisBio), 20 nM Kinase Tracer 236 (ThermoFischer),
20 nM GST-CDK8/CycC (CarnaBio) or GST-CDK19/CycC (CarnaBio), and compounds. All components
were diluted in the assay buffer containing 25 mM HEPES, 10 mM MgCl2, 2 mM DTT, 0.01 % Tween-20.
After 60 min incubation at rt, TR-FRET was measured on EnVision plate reader (PerkinElmer).
5.3.
In-Cell western assay of pSTAT1 in SW480 cells. Human colon adenocarcinoma cell line,
SW480, was maintained in the Leibovits L-15 medium (Gibco) with 10 % FBS at 37 °C. Cells were plated
on 384 well plates at a density of 6000 cells/well. After overnight incubation, cells were treated with
compounds for two hours then treated with 10 ng/mL interferon-γ for one hour. Cells were fixed with
Mildform 20 NM (wako) then washed with PBS. Cells were permeabilized and blocked with 0.1%
Triton-X-100, 0.05% Porcine300, 10% goat serum in PBS at rt for one hour. After discarding the blocking
buffer, cells were treated with phospho-STAT1 (Ser727) (D3B7) Rabbit mAb (Cell Signaling Technology)
at 4 °C. After overnight incubation and wash with 0.1% Triton X-100 in PBS (PBS-T), secondary detection
was conducted with Goat anti-Rabbit IRDye 800CW (LiCor) and DRAQ5 (Cell Signaling Technology) for
one hour at rt. After five washes with PBS-T, pSTAT1 and DNA were visualized using Odyssey Infrared
Imaging Scanner (LiCor).
a
5.4.
Cell growth inhibitory assay. Cells were seeded on a 96-well microplate (5 x102 cells /100 µL
medium). After 24 hours, a test compound was added to them so as to be an appropriate concentration.
After 168 hours incubation, 100 µL CellTiter-Glo (G7573, Promega) was added to each well. After 20
minutes incubation, the luminescence was measured with a microplate reader (ARVO MX 1420). For
nuclear count, 100 µL CyQuant Direct nucleic acid stain (C35011, Invitrogen) was added to each well.
After 1 hour incubation, the fluorescence was measured with ARVO MX 1420.
5.5.
Western blotting of phospho-STAT1. Cell pellets and tumor samples were lysed in lysis buffer
(62.5 mM Tris-Cl[pH 7.5], 1% SDS, 5% Glycerol) and were boiled at 95 °C for 5 minutes. A total of 5 ug
of protein were separated on SDS-polyacrylamidegels. Proteins were transferred with iBlot (Cat#B1001,

Invitrogen) and probed with antibodies. STAT1, Phospho-STAT1Ser727 were all purchased from Cell
signaling. Secondary horseradish peroxidase-conjugated antibodies (Cell signaling) were used, and ECL
plus was used to visualize the bands (LAS-3000, Fujifilm).
5.6.
Solubility. Test compounds in DMSO solutions were added to the JP 2nd fluid for disintegration
test (pH 6.8). After incubation, precipitates were separated from by filtration through a filter plate.
Compound concentrations in the filtrates were measured by plate reader or HPLC.
5.7.
Microsome stability. Human and mouse liver microsomes were purchased from Xenotech, LLC
(Lenexa, KS). An incubation mixture consisted of microsomes in 50 mmol/L KH2PO4–K2HPO4
phosphate buffer (pH 7.4) and 1 µmol/L test compound. The concentration of microsomal protein was 0.2
mg/mL. An NADPH-generating system (5 mmol/L MgCl2, 5 mmol/L glucose-6-phosphate, 0.5 mmol/L
beta-NADP+ and 1.5 unit/mL glucose-6-phosphate dehydrogenase) was added to the incubation mixture
with a half volume of the reaction mixture to initiate the enzyme reaction. The reaction was terminated 15
and 30 minutes after the initiation of the reaction by mixing the reaction mixture with acetonitrile, followed
by centrifugation at 2500 rpm for 10 min. The supernatant was subjected to LC/MS/MS analysis. The
metabolic velocity was calculated as the slope of the concentration-time plot. The in vitro intrinsic
metabolic clearance was calculated by dividing initial metabolic velocity by the test compound
concentration in the incubation mixture.
5.8.
Time-dependent CYP3A4 inhibition (TDI). Human liver microsomes were purchased from
Xenotech, LLC (Lenexa, KS). A mixture of a test compound (30 µM) and microsomes in phosphate buffer
(pH 7.4) was preincubated at 37 °C for 0 or 60 min in the presence of an NADPH-generating system
containing MgCl2, glucose-6-phosphate, β-NADP+, and glucose-6-phosphate dehydrogenase. After
preincubation, enzymatic activity of CYP3A4 in the incubation mixture was determined by measuring
6β-hydroxytestosterone in the reaction with testosterone by UPLC. The activity (% of control) for each
preincubation time was calculated as follows: {(activity with test compound)/(activity with DMSO)} × 100.
The remaining activity (% remaining) after preincubation for 60 min was calculated as follows: {activity at
60min (% of control)}/{activity at 0min (% of control)} ×100.
5.9.
Pharmacokinetic Studies. Test compounds were administered intravenously (0.1 mg/kg) or
orally (1 mg/kg) by cassette dosing to non-fasted mice. After administration, blood samples were collected
and centrifuged to obtain the plasma fraction. The plasma samples were deproteinized by mixing with
acetonitrile, followed by centrifugation. The compound concentrations in the supernatant were measured by
LC/MS/MS.
5.10.
Antitumor efficacy study. Five-week-old female C.B17/Icr-scid/scid Jcl mice purchased from
CLEA Japan Inc. (Tokyo, Japan) were maintained on a 12/12 hr light/dark cycle (lights on at 7 am) with
o
constant temperature (23
2 C). Food and water were available ad libitum. RPMI-8226 cells (1 x

107cells/100 µL HBSS (Gibco): Matrigel (BD Biosciences); 1: 1) were injected subcutaneously into one
flank of each mouse. After tumors had reached approximately 80-140 mm3, test compound (15 mg/kg,
b.i.d., po or 30 mg/kg, b.i.d, po) was administered for 14 days. The tumor size was measured with a caliper
twice a week, and expressed in mm3 using the formula 0.5 x length x width2, where a is the largest
diameter and width is largest diameter perpendicular to a. Body weight was measured at the same time
point of the tumor measurement. Data expresses mean se. All animal studies were conducted under the
guidelines of the Takeda Experimental Animal Care and Use Committee.
Acknowledgements
The authors thank Takashi Motoyaji and Akiyoshi Tani for target identification studies leading us to focus
on lead compound 1. We also would like to express our appreciation to Osamu Ujikawa, Noritaka Kuroda,
Kazuaki Takami, Masanori Okaniwa, and Shiwei Guo for their helpful advice and support in organic
synthesis. We also thank Yukio Toyoda for his assistance in cell Western blotting assay development, and
Tsuyoshi Ishii for his useful discussion about in vitro assay development. We are also grateful to Hiroshi
Miyake, Hikichi Yuichi, and Tomoyasu Ishikawa for helpful discussion and guidance in the overall
execution of this research.
Supplementary data
Supplementary data (chemical synthesis procedures, a docking experiment of 4k, an analysis of the human
kinases, a kinase selectivity profile of 4k and western blotting of phospho-STAT1 in anti-tumor efficacy
test of 4k) associated with this article can be found in "Supplementary material (available online)".
Abbreviations used
ADME, absorption, distribution, metabolism, and excretion; AML, acute myelogenous leukemia; AUC,
area under the blood concentration time curve; b.i.d., twice a day; CDK, Cyclin-dependent kinase; CL
,
total
clearance; Cmax, maximum drug concentration; CRC, colorectal cancer; CTD, C-terminal domain; DDI,
drug-drug interaction; DMA, N,N-dimethylacetamide; DMAP, N,N-dimethyl-4-aminopyridine; DMG loop,
Asp-Met-Glu loop; F, bioavailability; FBS, Fetal bovine serum; GST, glutathione S-transferase; HATU,
O-(7-azabenzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium-hexafluorphosphat;
4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; HOBt, 1-hyfroxybenzotriazole;
HEPES,
HTS,
high-throughput screening; INF, interferon; ICR, Institute for Cancer Research; CyJohnPhos,
2-(dicyclohexylphosphino)biphenyl; MCM3, mini-chromosome maintenance protein; mCPBA,
3
m-chloroperoxybenzoic acid; MED, mediator complex; MRT, mean residence time; ODS, octadecyl silyl;
PBS-T, phsophate buffered saline with Tween 20; PD, pharmacodynamic; PDB, protein data bank; PK,
pharmacokinetic; q.d., once a day; RNAP, RNA polymerase; rt, room temperature; SBDD, structure based
drug design; SPhos, dicyclohexyl(2',6'-dimethoxy-[1,1'-biphenyl]-2-yl)phosphine; STAT, signal
transducers (transduction) and activator of transcription; TR-FRET, time-resolved fluorescence resonance

energy transfer; Vdss, steady state volume of distribution; EDC, water soluble carbodiimide
(1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide); XPhos, 2-dicyclohexylphosphino-2',4',6'-
triisopropylbiphenyl.
References and notes
1. (a) Galbraith, M. D.; Donner, A. J.; Espinosa, J. M. Transcription 2010, 1, 4−12. (b) Tsutsui, T.; Fukasawa, R.; Tanaka, A.; Hirose, Y.;
Ohkuma, Y. Genes Cells 2011, 16, 1208−1218. (c) Rzymski, T.; Mikula, M.; Wiklik, K.; Brzózka, K. Biochimica et Biophysica Acta
2015, 1854, 1617–1629.
2. Rickert, P.; Seghezzi, W.; Shanahan, F.; Cho, H.; Lees, E. Oncogene 1996, 12, 2631−2640.
3. (a) Nemet, J.; Jelicic, B.; Rubelj, I.; Sopta, M. Biochimie 2014, 97, 22−27. (b) Li, N.; Fassl, A.; Chick, J.; Inuzuka, H.; Li, X.; Mansour,
M. R.; Liu, L.; Wang, H.; King, B.; Shaik, S.; Gutierrez. A.; Ordureau, A.; Otto, T.; Kreslavsky, T.; Baitsch, L.; Bury, L.; Meyer, C.
A.; Ke, N.; Mulry, K. A.; Kluk, M. J.; Roy, M.; Kim, S.; Zhang, X.; Geng, Y.; Zagozdzon, A.; Jenkinson, S.; Gale, R. E.; Linch, D. C.;
Zhao, J. J.; Mullighan, C. G.; Harper, J. W.; Aster, J. C.; Aifantis, I.; Boehmer, H.; Gygi, S. P.; Wei, W.; Look, A. T.; Sicinski, P. Nat.
Cell Biol. 2014, 16, 1080−1091.
4. Firestein, R.; Bass, A.J.; Kim, S. Y.; Dunn, I. F.; Silver, S.J.; Guney, I.; Freed, E.; Ligon, A. H.; Vena, N.; Ogino, S.; Chheda, M. G.;
Tamayo, P.; et al. Nature 2008, 455, 547–551.
5. Allen, B. L.; Taatjes, D. J. Nature Reviews Molecular Cell Biology 2015, 16, 155–166.
6. Giovanni, C. D.; Novellino, E.; Chilin, A.; Lavecchia, A.; Marzaro, G. Expert Opinion on Investigational Drugs 2016, 25, 1215-1230.
7. Cee, V. J.; Chen, D. Y.; Lee, M. R.; Nicolaou, K. C. Angew. Chem., Int. Ed. 2009, 48, 8952−8957.
8. Roninson, I. B.; Porter, D. C.; Wentland, M. P. WO2013116786 A1, 2013.
9. (a) Mallinger, A.; Crumpler, S.; Pichowicz, M.; Waalboer, D.; Stubbs, M.; Adeniji-Popoola, O.; Wood, B.; Smith, E.; Thai, C.; Henley,
A. T.; Georgi, K.; Court, W.; Hobbs, S.; Box, G.; Ortiz-Ruiz, M. J.; Valenti, M.; De Haven Brandon, A.; TePoele, R.; Leuthner, B.;
Workman, P.; Aherne, W.; Poeschke, O.; Dale, T.; Wienke, D.; Esdar, C.; Rohdich, F.; Raynaud, F.; Clarke, P. A.; Eccles, S. A.;
Stieber, F.; Schiemann, K.; Blagg, J. J. Med. Chem. 2015, 58, 1717−1735. (b) Mallinger, A.; Schiemann, K.; Rink, C.; Stieber, F.;
Calderini, M.; Crumpler, S.; Stubbs, M.; Adeniji-Popoola, O.; Poeschke, O.; Busch, M.; Czodrowski, P.; Musil, D.; Schwarz, D.;
Ortiz-Ruiz, M.; Schneider, R.; Thai, C.; Valenti, M.; Brandon, A.; Burke, R.; Workman, P.; Dale, T.; Wienke, D.; Clarke, P. A.; Esdar,
C.; Raynaud, F. I.; Eccles, S. A.; Rohdich, F.; Blagg, J. J. Med. Chem. 2016, 59, 1078−1101.
10. Koehler, M. F. T.; Bergeron, P.; Blackwood, E. M.; Bowman, K.; Clark, K. R.; Firestein, R.; Kiefer, J. R.; Maskos, K.; McCleland,M.
L.; Orren, L.; Salphati, L.; Schmidt, S.; Schneider, E. V.; Wu, J.; Beresini, M. H. ACS Med. Chem. Lett. 2016, 7, 223−228.
11. Schiemann, K.; Mallinger, A.; Wienke, D.; Esdar, C.; Poeschke, O.; Busch, M.; Rohdich, F.; Eccles, S. A.; Schneider, R.; Raynaud, F.
I.; Czodrowski, P.; Musil, D.; Schwarz, D.; Urbahns, K.; Blagg, J. Bioorg. Med. Che. Lett. 2016, 26, 1443−1451.
12. For previous related work, see also Ono, K; Banno, H.; Okaniwa, M.; Hirayama, T.; Iwamura, N.; Hikichi, Y.; Murai, S.; Hasegawa,
M.; Hasegawa, Y.; Yonemori, K.; Hata, A.; Aoyama, K.; Cary, D. R. Bioorg. Med. Chem., submitted for publication.
13. Ujikawa, O.; Takami, K.; Takakura, N.; Sakai, N. WIPO Patent WO 2012/008549, Jan 19, 2012.
14. Based on data from UniProt Home Page. http://www.uniprot.org/ and RSCB Protein Data Bank Home Page. http://www.rcsb.org/pdb/
(PDB ID: 3RGF) (accessed October 31, 2016).
15. Schneider, E. V.; Böttcher, J.; Blaesse, M.; Neumann, L.; Huber, R.; Maskos, K. J. Mol. Biol. 2011, 412, 251−266.
16. Partial details of synthetic methods described in Hirayama, T.; Fujimoto, J.; Cary, D. R.; Okaniwa, M.; Hirata, Y. WIPO PCT

WO2015/159938, October 22, 2015.
17. The NMR data of the compound 22 is shown in Supplementary Figure 4.
18. Bancerek, J.; Poss, Z. C.; Steinparzer, I.; Sedlyarov, V.; Pfaffenwimmer, T.; Mikulic, I.; Dölken, L.; Strobl, B.; Müller, M.; Taatjes, D.
J.; et al. Immunity 2013, 38, 250−262.
19. (a) Estrada, A. A.; Shore, D. G.; Blackwood, E.; Chen, Y.-H.; Deshmukh, G.; Ding, X.; DiPasquale, A. G.; Epler, J. A.; Friedman, L.
S.; Koehler, M. F. T.; Liu, L.; Malek, S.; Nonomiya, J.; Ortwine, D. F.; Pei, Z.; Sideris, S.; St-Jean, F.; Trinh, L.; Truong, T.;
Lyssikatos, J. P. J. Med. Chem., 2013, 56, 3090−3101. (b) For a recent publication and reference regarding Bredt's Rule, see Tani, K.;
Stoltz, B. M. Nature. 2006, 441, 731−734.
20. Fabian, M. A.; Biggs, W. H. 3rd; Treiber, D. K.; Atteridge, C. E.; Azimioara, M. D.; Benedetti, M. G.; Carter, T. A.; Ciceri, P.; Edeen,
P.T.; Floyd, M.; Ford, J. M.; Galvin, M.; Gerlach, J. L.; Grotzfeld, R. M.; Herrgard, S.; Insko, D. E.; Insko, M. A.; Lai, A. G.; Lélias, J.
M.; Mehta, S. A.; Milanov, Z. V.; Velasco, A. M.; Wodicka, L. M.; Patel, H. K.; Zarrinkar, P. P.; Lockhart, D. J. Nat. Biotechnol. 2005,
−336.
23, 329

F
N
N
N
N
O
N
O
N
O
N
F
O
O
N
H
N
H
N
H
F
N
N
N
1
2k
4k
CDK8/19: 80 nM/100 nM
ICW(pSTAT1): 370 nM
TDI: 84% remaining @ 60 min
CDK8/19: 6.4 nM/7.6 nM
ICW(pSTAT1): 11 nM
TDI: 54% remaining @ 60 min
CDK8/19: 2.5 nM/3.0 nM
ICW(pSTAT1): 8.4 nM
TDI: 99% remaining @ 60 min