Discovery of coumarin derivatives as potent and selective cyclin-dependent kinase 9 (CDK9) inhibitors with high antitumour activity

Article abstract of DOI:10.1016/j.ejmech.2020.112424

Specific inhibition of CDK9 is considered a promising strategy for developing effective anticancer therapeutics. However, most of the reported CDK9 inhibitors are still at an early stage of development and lack selectivity against other CDKs. Herein, we discovered coumarin derivative 30i as a potent CDK9 inhibitor with high selectivity (8300-fold over CDK7). Binding mode analysis illustrated that the substituent coumarin moiety is a critical group for CDK9 selectivity by occupying a flexible hinge/αD region, which is sterically hindered in other CDKs. Compound 30i showed excellent cellular antiproliferative activity, moderate pharmacokinetic property and low hERG inhibition. Moreover, 30i significantly induced tumour growth inhibition in a dose-dependent manner without causing an obvious loss of body weight in an MV4-11 xenograft mice model. Altogether, these results suggest that 30i may serve as a potential acute myeloid leukaemia (AML) therapeutics by selectively targeting CDK9.

Full text of DOI:10.1016/j.ejmech.2020.112424

Journal Pre-proof
Discovery of coumarin derivatives as potent and selective cyclin-dependent kinase 9
(CDK9) inhibitors with high antitumour activity
Junyu Xu, Xinren Wang, Jianhang Huang, Hongmei Li, Shuwen Li, Chenhe Liu,
Ruinan Dong, Gaoyuan Zhu, Chunqi Duan, Fei Jiang, Yanmin Zhang, Yuqin Zhu,
Tianyi Zhang, Yadong Chen, Weifang Tang, Tao Lu
PII:
S0223-5234(20)30395-0
DOI:
https://doi.org/10.1016/j.ejmech.2020.112424
EJMECH 112424
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 16 December 2019
Revised Date: 24 March 2020
Accepted Date: 4 May 2020
Please cite this article as: J. Xu, X. Wang, J. Huang, H. Li, S. Li, C. Liu, R. Dong, G. Zhu, C. Duan,
F. Jiang, Y. Zhang, Y. Zhu, T. Zhang, Y. Chen, W. Tang, T. Lu, Discovery of coumarin derivatives as
potent and selective cyclin-dependent kinase 9 (CDK9) inhibitors with high antitumour activity, European
Journal of Medicinal Chemistry (2020), doi: https://doi.org/10.1016/j.ejmech.2020.112424.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. 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.
©
2020 Published by Elsevier Masson SAS.

Discovery of Coumarin Derivatives as Potent and
Selective Cyclin-Dependent Kinase 9 (CDK9)
Inhibitors with High Antitumour Activity
†
†
†
†
†
†
†
Junyu Xu, Xinren Wang, Jianhang Huang, Hongmei Li, Shuwen Li, Chenhe Liu, Ruinan Dong, Gaoyuan
†
†
†
†§
†
†
†‡§*
Zhu, Chunqi Duan, Fei Jiang, Yanmin Zhang, Yuqin Zhu, Tianyi Zhang, Yadong Chen,
Weifang
†
‡*
†‡§*
Tang and Tao Lu
†
School of Sciences, China Pharmaceutical University, 639 Longmian Avenue, Nanjing 211198, China.
‡
State Key Laboratory of Natural Medicines, China Pharmaceutical University, 24 Tongjiaxiang, Nanjing
10009, China.
Laboratory of Molecular Design and Drug Discovery, School of Science, China Pharmaceutical University,
39 Longmian Avenue, Nanjing 211198, China.
2
§
6
1

Abstract
Specific inhibition of CDK9 is considered a promising strategy for developing effective anticancer therapeutics.
However, most of the reported CDK9 inhibitors are still at an early stage of development and lack selectivity
against other CDKs. Herein, we discovered coumarin derivative 30i as a potent CDK9 inhibitor with high
selectivity (8300-fold over CDK7). Binding mode analysis illustrated that the substituent coumarin moiety is a
critical group for CDK9 selectivity by occupying a flexible hinge/αD region, which is sterically hindered in
other CDKs. Compound 30i showed excellent cellular antiproliferative activity, moderate pharmacokinetic
property and low hERG inhibition. Moreover, 30i significantly induced tumour growth inhibition in a dose-
dependent manner without causing an obvious loss of body weight in an MV4-11 xenograft mice model.
Altogether, these results suggest that 30i may serve as a potential acute myeloid leukaemia (AML) therapeutics
by selectively targeting CDK9.
Keywords
Coumarin derivatives, CDK9 inhibitor, Selectivity, Structure-activity relationship, Antitumour Activity.
Highlights
ò
ò
ò
A series of novel coumarin derivatives were discovered as CDK9 inhibitors.
Compound 30i showed high selectivity (160- to 8300-fold) over CDK1/2/3/4/5/6/7/8/19.
A substituted coumarin located at the hinge/αD region was suggested to improve selectivity and inhibition
of CDK9.
ò
ò
Compound 30i induced tumour growth inhibition in a dose-dependent manner in an AML mice model.
Compound 30i dose-dependently inhibited CDK9-mediated phosphorylation of RNAPII.
2

1
. Introduction
Cyclin-dependent kinases (CDKs), a family of serine/threonine protein kinases, play a critical role in the
control of cell division cycle progression (e.g., CDK1 − 6) and regulation of gene transcription (e.g., CDK7 −
1
3 and 19) in eukaryotic cells [1,2]. As a CDK family member, CDK9 combines with its corresponding cyclin
partners (cyclin T1, T2a, T2b, or K) to form positive transcription elongation factor b (P-TEFb),
phosphorylating residue Ser2 within the C-terminal domain (CTD) of RNA polymerase II (RNAPII) to
facilitate gene transcription elongation [3-5]. The prevailing view is that many cancer cells are dependent on
the sustained expression of short-lived antiapoptotic proteins (e.g., Mcl-1, Bcl-2 and XIAP); however, CDK9
is a vital regulator that stimulates transcriptional elongation of antiapoptotic proteins [6,7]. Inhibition of
CDK9-mediated phosphorylation of RNAPII can rapidly reduce the levels of antiapoptotic proteins, leading to
the promotion of apoptosis and inhibition of tumour cell proliferation [8-11]. Deregulated CDK9-related
pathways are frequently observed in many human malignancies; thus, CDK9 is considered an attractive cancer
therapeutic target [12-19]. The first-generation clinical CDK9 inhibitor candidates, such as flavopiridol
(
alvocidib) and R-roscovitine (seliciclib), have excellent pharmaceutical properties and in vivo antitumour
efficacy and are associated with the downregulation of CDK9-mediated antiapoptotic protein transcription
10,11,20-22]. However, CDK9 specificity was not achieved, which might cause dose-limiting toxicity and
[
adverse events, and thus result in restricted clinical applications [23].
Designing selective CDK inhibitors has become more prevalent. Intensive attempts of developing
structurally diverse CDK inhibitors have led to approvals of CDK4/6 inhibitors (palbociclib (Pfizer), ribociclib
(Novartis) and abemaciclib (Eli Lilly)) [24-26]. So Several selective CDK9 inhibitors have been actively
pursued by modifying the scaffold of previous pan-CDK inhibitors [27]; however, most of these pan-CDK
inhibitors show cross-reactivity with at least two other CDK family members on the same order of magnitude
or show significant effects on other protein kinases [28-34]. The design of highly selective CDK9 inhibitors
remains a great challenge at the present stage due to the high sequence homology among the CDK family
members [35,36]. Herein, we report our studies to design, synthesize and evaluate a series of coumarin
derivatives, leading to the identification of a highly potent and selective CDK9 inhibitor 30i as a potential
antitumour agent.
3

Figure 1. Chemical structures of representative CDK9 inhibitors recently.
. Results and Discussion
2
2
.1. Discovery and Research of Hit Compound 7
Inspection of published CDK inhibitors revealed that a majority of these molecules occupy the ATP-binding
pocket [27]. Wang et al. reported that selective inhibition of CDK9 by 3 (12U) results from the relative
malleability of the ATP binding area, and a bulky 1,4-diazepan-1-ylaniline moiety can be accommodated
within a less crowded and different electrostatic environment [31,37]. Further comparison of the hinge/αD
region sequence alignment of CDK9/2/1/3/4/5/6/7/8 in the ATP-binding area (Figure 2A/C and supplementary
Figure S1/S2) showed that the Gly112/Ala111 residues of CDK9 are much smaller than the residues at the
same position in other CDKs within this αD-helix, such as the Lys89/Lys88 residues in CDK2, and the
Thr102/Arg101 residues in CDK4 [38]. This indicates the Gly112/Ala111 of αD-helix region in CDK9 may
accommodate bulkier moieties.
Coumarin scaffolds are rarely reported for the design of CDK inhibitors; however, some naturally occurring
4

and synthetic compounds that contain the coumarin core have recently been reported to induce apoptosis of
tumour cells, resulting from the inhibition of CDK-associated kinase activity [39-42]. Moreover, some
structurally related flavanoids including flavopiridol [43], IIIM-290 [29], voruciclib [21], riviciclib [44] and
others [20]) have been successfully designed as CDK inhibitors. In addition, coumarin derivatives have similar
pharmacological effects to CDK9 inhibitors, including antitumour [45,46], antiviral [47], cardioprotective [48]
and other effects [49,50].
Inspired by the above research information, we screened coumarin derivatives from our natural product-like
molecule library for the inhibition of CDKs. As shown in Figure 2D, we identified a hit compound 7 which
had moderate inhibition of CDK9 (CDK9/cyclin T1 IC : 184 nM), and showed selectivity (7-fold) for CDK9
50
over CDK2. In particular, it inhibits CDK7 in uM range (IC50 > 27000 nM) indicating that 7 possessed
selectivity for CDK9 not only against cell cycle CDKs but also against transcriptional CDKs. To better
understand the selectivity of 7 towards CDK9, we docked compound 7 to a CDK9 crystallographic structure
(
PDB code: 4BCG). In Figure 2B, the C2-carbonyl of coumarin forms a hydrogen bond with the NH of
Asp109, and the coumarin moiety is accommodated above the flexible area of CDK9 at the tail of hinge area
and αD-helix. In addition, the N1-pyrimidine moiety accepts a hydrogen bond from the NH of Cys106, and the
C2-NH of the pyrimidine ring donates a hydrogen bond to the carbonyl of Cys106. In conclusion, molecule 7
is strongly fixed in the ATP-binding cleft through three hydrogen bonds at the hinge area. The nonconserved
hydrophobic area and gatekeeper region (Phe103) at the C4- or C5-position of pyrimidine and the solvent
accessible region at the C3- or C4-position of coumarin can be exploited for the design of CDK9 inhibitors.
Therefore, we carried out a structure-activity relationship study of coumarin derivatives with the aim of
identifying novel, potent and selective inhibitors of CDK9 as antitumour agents. The chemical structures and
biological activity results are summarized in Table 1 − Table 3.
5

Figure 2. (A) Overlapping the key amino acid residues of CDK9 and CDK2 at the αD-helix area. (B)
Compound 7 (green stick model) docked into the published crystal complex of CDK9 (PDB code: 4BCG); the
key residues between CDK9 are labelled, and the hinge/αD region is shown in surface representation.
Hydrogen bonds are depicted by black dotted lines. (C) Hinge/αD region sequence alignment of
CDK9/2/1/3/4/5/6/7/8. (D) Chemical structure, inhibition efficacy and properties of compound 7. The mean
*
IC50 values were measured in duplicate. The properties were calculated by ChemDraw Ultra 2017 software;
*
*
ligand Efficiency of CDK9/cyclin T1: LE = -pIC50/n, n = heavy atom count (or non-hydrogen atom count).
.2. Chemistry
2
The desired target compounds 16a-16b were synthesized as depicted in Scheme 1. Commercially available
7
-hydroxy-4-methyl-2H-chromen-2-one (8) was used as the starting material. The hydroxy at the C7-position
was found to interfere in subsequent reactions and was therefore masked as the tert-butyldimethylsilyl ether 9
in the presence of imidazole as a base. Bromination of 8 with NBS at room temperature resulted in bromide 10.
Preparation of key intermediate 13 started from the Stille coupling reaction of 10 with 1-ethoxyvinyltri-n-
butyltin using PdCl (PPh ) as the catalyst, followed by treatment with hydrochloric acid in refluxing methanol
2
3 2
6

to yield 12, which was then esterified with trifluoromethanesulfonic anhydride in the presence of triethylamine
as the base in an ice bath. Compounds 15a-15b were prepared through the Suzuki-Miyaura coupling reaction
of commercial compounds 14a-14b and (2-methoxyphenyl)boronic acid using Pd(dppf)Cl as the catalyst.
2
Target compounds 16a-16b were synthesized by the Buchwald-Hartwig amination reaction of
aminopyrimidine derivatives 15a-15b and key coumarin intermediate 13 using Pd(OAc) as the catalyst.
2
Compound 7 was synthesized under the same conditions with the commercially available 4-(pyridin-3-
yl)pyrimidin-2-amine and intermediate 13.
Scheme 1:
Reagents and conditions: (a) TBDMSCl, imidazole, DMF, r.t., 6 h; (b) NBS, CH CN, r.t., 9 h; (c)
3
PdCl (PPh ) , 1-ethoxyvinyltri-n-butyltin, DMF, 80 °C, 6 h; (d) HCl (6M, in methanol), reflux, 12 h; (e) Tf O,
2
3 2
2
7

Et N, DCM, 0 °C, 1 h; (f) (2-methoxyphenyl)boronic acid, Na CO , Pd(dppf)Cl , 1,4-dioxane : H O = 7 : 1, N ,
3
2
3
2
2
2
9
2 2 3 2
5 °C, 2−6 h; (g) 13, Pd(OAc) , Cs CO , Xantphos, N , 100 °C, 2−16 h.
The synthesis of target compounds 29a-29t and 30a-30p is depicted in Scheme 2. In an effort to further
modify the C3- and C4-positions of the coumarin core, commercially available paeonol was reacted with
diethyl carbonate in the presence of sodium hydride as the base with heating (100 °C, 4 h) to give compound
1
8. Alterations to the C3-position were achieved by a series of modifications to 18; C3-substituted 19a or 19c
were obtained by treating 18 with acetic acid or propionic anhydride in the presence of phosphorus
oxychloride, respectively. Reduction conveniently converted 19a to 19b with sodium cyanoborohydride as the
reductant. In order to introduce the hydrophilic fragment, a nucleophilic substitution reaction was utilized to
install different heterocyclic substituents at the C4-position by refluxing under basic conditions to give 20a-
2
0g. Cleavage of the ether at the C7-position with aluminium chloride in refluxing toluene led to 23a-23g.
Moreover, halogenation of 18 formed 21 with the use of phosphorus oxychloride and triethylamine, and then
cleavage of the ether at the C7-position was carried out under similar conditions as those for compound 23a,
followed by the introduction of heterocyclic substituents to the C4-position, leading to 23h-23j. Finally, a
series of key intermediates 24a-24j were obtained by esterification under similar conditions as those for
intermediate 13.
A series of aminopyrimidine compounds with C-C linkages at the C4-position of pyrimidine were formed by
palladium-mediated chemistry, and the Suzuki-Miyaura coupling reaction was used to install a series of
substituted aromatic groups giving compounds 25a-25q. In an effort to prepare C5-substituted pyrimidine
2
2
derivatives, Suzuki-Miyaura coupling reactions of the commercially inexpensive agent 26a (R = F) or 26b (R
=
Cl) with 4-fluoro-2-methoxyphenylboronic acid were performed with Pd(dppf)Cl as the catalyst. However,
2
further amination reactions of 27a-27b with ammonium hydroxide were performed under heating (85 °C)
conditions, giving aminopyrimidine derivatives 28a-28b. The desired target compounds 29a-29t and 30a-30p
were synthesized by the Buchwald-Hartwig amination reaction using Pd(OAc)
aminopyrimidine derivatives 15a-15b/25a-25q/28a-28b and key coumarin intermediates 24a-24j. In addition,
9d-29e were obtained by treatment with hydrochloric acid at room temperature for deprotection of the N-Boc
group.
2
as the catalyst between the
2
8

Scheme 2:
O
OH
O
OH
O
a
b
R3
O
OH
O
O
O
O
17
18
19a-19c
1
1
1
9a: R3 = acetyl
9b: R3 = ethyl
9c: R3 = propionyl
f
c
Cl
Cl
d
e
O
O
O
HO
O
O
21
22
g
R2
R2
O
R2
O
h
R3
O
R3
O
e
R3
O
TfO
O
HO
O
24a-24j
23a-23j
O
20a-20g
X
Y
N
NH2
R1
15a-15b
R1
R2
O
R3
O
X
Y
Cl
Y
i
k
N
N
X
X
Y
H
R2
N
NH2
N
NH2
R3
29a-29t, 30a-30h were shown
in Table 1 and Table 2
1
4a-14b
25a-25q
TfO
O
O
F
F
F
24a-24j
Cl
i
j
R2
O
R4
O
O
O
N
R4
R3
R4
R4
N
N
N
N
N
Cl
N
N
H
O
N
Cl
NH2
26a-26b
27a-27b
28a-28b
30i-30p were shown in Table 2
3
3
3 2
Reagents and conditions: (a) diethyl carbonate, NaH, toluene, 100 °C, 4 h; (b) POCl , R OH or (R ) O,
2
1
05 °C, 1 h; (c) NaBH CN, AcOH, r.t., 0.5 h; (d) R H, reflux, 2−6 h; (e) AlCl , toluene, 90 °C, 3−6 h; (f)
3
3
2
POCl , Et N, 100 °C, 3 h; (g) R H, Et N, EtOH, reflux, 3 h; (h) Tf O, Et N, DCM, 0 °C, 1 h. (i) phenylboronic
3
3
3
2
3
9

acid derivative, Na CO , Pd(dppf)Cl , 1,4-dioxane : H O = 7 : 1, N , 95 °C, 2−6 h; (j) NH OH, isopropanol, 85
2
3
2
2
2
4
o
C, 6 h; (k) i. relative intermediate 28a−24j, Pd(OAc)
2 2
, Cs CO
3
, Xantphos, N
2
, 100 °C, 2−16 h; ii. or further
deprotection of the N-Boc, HCl (4M, in Dioxane), r.t., 0.5 h.
2
.3. Structure-Activity Relationships (SARs)
Based on the molecular modelling studies of compound 7, the C7-, C4- and C3-positions of coumarin were
modified to explore the hydrophobic area, hinge region and solvent accessible region for potent CDK9
inhibition. All of these coumarin derivatives were screened for their enzymatic activity against CDK9/cyclin
T1 or CDK2/cyclin A in vitro. The initial screening was verified and duplicated at a single concentration of 1.0
μM, followed by selection of the active compound for determination of the IC50 values against CDK9 and
CDK2 enzymatic activities and the antiproliferative activities of the MV4-11 human acute myeloid leukaemia
(AML) cell line.
With the aim of finding more potent analogues, we preferentially evaluated the 2-methoxyphenyl group at
the A position as the hydrophobic region and gatekeeper region (Table 1). Analogues 16a and 16b showed
significant improvement in CDK9 inhibition with an IC50 values of 6.5 nM and 37.6 nM but poor toxicity in
MV4-11 cells. We then introduced a morpholino as the hydrophilic heterocycle towards the solvent-accessible
region in analogue 29a, potently inhibited CDK9 (IC50: 3.1 nM) and improved cellular toxicity 10-fold
3
3
compared with the toxicity of 16a. When changing the acetyl group at R position in 29a, analogues 29b (R =
3
Et) and 29c (R = H) showed over 20-fold and 15-fold losses in CDK9 inhibition, respectively, indicating that
the acetyl group at the C3-position of coumarin was favourable. Notably, the introduction of heterocyclic
piperazin-1-yl (29d) or 1,4-diazepan-1-yl (29e) at the C4-position of coumarin displayed comparable potency
and selectivity for CDK9 and enhanced cellular antiproliferative activity compared to the antiproliferative
2
activity of analogue 16a (R = Me).
Table 1. Structures and biological evaluations of compounds 7, 16a - 16b and 29a – 29e.
1
0

a
Enzymatic IC /nM
(
Cell Inhibitory
50
b
a
% Enzymatic activity)
IC (μM)
2
3
50
Comp.
R
R
X
Y
N
CDK9/
cyclin T1
CDK2/
cyclin A
MV4-11
c
7
Me
Me
acetyl CH
acetyl CH
184
1211
12.88
16a
16b
29a
29b
N
6.5
37.6
3.1
451.2
10.16
＞10
1.01
4.72
4.57
1.34
1.68
Me
acetyl
N
CH
N
(64.5±0.7)
1283
morpholino
morpholino
morpholino
piperazin-1-yl
1,4-diazepan-1-yl
acetyl CH
Et
H
H
H
CH
CH
CH
CH
N
67.3
47.2
25.3
10870
2
9c
N
(57.9±0.3)
3664
2
9d
N
2
9e
N
11.7
2056
a
b
The mean values of IC50 were measured in duplicate. % Enzyme activities were measured at 1.0 μM
c
concentration of test compound, n=2. NA: not determined. A moiety was pyridin-3-yl.
To further explore the hydrophobic area and gatekeeper region for affinity improvement, the diversity of the
substituents was evaluated to investigate how varying the substituents at the phenyl group will affect the
activity against CDK9 (Table 2). Replacement of 2-methoxyl with a 3-methoxyl (29f), 4-methoxyl (29g), 2,4-
1
1

dimethoxyl (29h) or 2,3-dimethoxyl (29i) showed slightly reduced inhibition of CDK9, indicating that an
ortho-substituent on the phenyl ring was most favourable. Interestingly, the introduction of 2-trifluoromethyl
(
29l) or 2-nitryl (29m) resulted in a significant loss in CDK9 inhibition compared with 2-methyl (29n) and 2-
ethyl (29o), suggesting that an electron-deficient phenyl ring at the C4-position of pyrimidine was not tolerated.
This may benefit from the aromatic-aromatic interaction of the electron-rich phenyl group with the Phe103
gatekeeper residue of CDK9. The increased lipophilicity of fluorine (30a-30c) at the 2-methoxyphenyl group
resulted in an improvement in cellular antiproliferative activity possibly due to an increase in cellular
permeability; simultaneously, this modification showed less of an effect on CDK9 inhibitory activity.
To understand the SAR of the pyrimidine of the hinge region, the substituted (C4-pyrimidin-2-yl)amino
group was replaced by the substituted (C6-pyrimidin-2-yl)amino group at the C7-position of coumarin.
Notably, the CDK9 inhibitory activities of analogues 29p – 29t were lost, and this is likely due to the electron
density alteration that disrupted the hydrogen-bond interaction with Cys106 in the hinge region. The electron-
rich phenyl ring was still favourable in the hydrophobic area when the (C6-pyrimidin-2-yl)amino group was
located in the hinge region.
Table 2. Structures and biological evaluations of compounds 29a, 29f – 29t and 30a – 30c.
a
Enzymatic IC50 /nM
Cell Inhibitory
b
a
(%Enzymatic activity)
IC (μM)
1
50
Comp.
R
X
Y
N
N
CDK9/
cyclin T1
CDK2/
cyclin A
MV4-11
2
9a
2-MeO
3-MeO
CH
CH
3.1
1283
1.01
29f
18.4
(54.1±2.3)
1.49
1
2

2
9g
9h
4-MeO
CH
N
N
81.7
54.2
(21.6±0.4)
(46.3±4.0)
NA
6.49
0.82
5.07
NA
2
2,4-2MeO CH
2,3-2MeO CH
2
9i
N
(51.7±4.8)
228
2
9j
H
CH
CH
CH
CH
CH
CH
N
N
NA
2
9k
2,6-2F
N
267
NA
＞10
NA
2
9l
2-CF
3
N
2488
2335
489
NA
2
9m
2-NO
2
N
NA
＞10
NA
2
9n
9o
9p
9q
2-Me
2-Et
N
NA
2
N
66.5
(40.5±1.7)
3415
NA
9.84
＞10
NA
2
2
2-MeO
H
CH
210
N
CH (103.9±7.7)
CH (92.2±1.8)
CH (90.0±0.5)
2
9r
9s
2-Me
2-Et
N
NA
NA
2
N
NA
NA
1
3

2
9t
2-CF
3
N
CH (106.3±3.6)
NA
NA
0.25
1.74
0.52
3
0a
0b
4-F, 2-MeO CH
5-F, 2-MeO CH
6-F, 2-MeO CH
N
N
N
8.8
1328
3
21.8
13.7
(49.6±4.1)
(58.7±1.0)
3
0c
a
b
The mean values of IC50 were measured in duplicate. % Enzyme activities were measured at 1.0 μM
concentration of test compound, n=2. NA: not determined.
1
Although compounds with 4-fluoro-2-methoxyl at the R position and pyrimidin-2-ylamino in the hinge
region showed good potency against CDK9, there was no evidence to clarify which hydrophilic group would
be optimal after considering the cellular toxicity and physical properties. Therefore, hydrophilic heterocycles
were investigated using the optimized structure of 4-(4-fluoro-2-methoxyphenyl)pyrimidin-2-yl as the
hydrophobic segment. The details of further optimization are shown in Table 3. Regrettably, substitution of the
2
morpholino with 4-methylpiperazin-1-yl (30d), 4-ethylpiperazin-1-yl (30e) or piperidin-1-yl (30f) at the R
position did not significantly improve the cellular toxicity or CDK9 inhibitory activity. In the context of 30a,
3
replacement of the 3-acetyl with a 3-propionyl (30g) at the R position exhibited a similar enzymatic potency
and antiproliferative activity profile; however, replacement of the 3-acetyl with a hydrogen (30h) showed a
significant reduction in cytotoxicity.
To further improve cellular toxicity, we introduced fluorine (30i and 30j) or chlorine (30o and 30p) to the
C5-pyrimidine position. This modification had less of an impact on the inhibitory activity against CDK9.
4
4
However, the cellular antiproliferative activity of 30i (R = F) was better than that of both 30o (R = Cl) and
4
3
0a (R = H). In the context of 30i, the substitution with heterocyclic groups or a 3-propionyl group towards
the solvent accessible region was explored. The 4-morpholino and 3-acetyl groups of 30i turned out to be
optimal in terms of improving cellular toxicity.
Table 3. Structures and biological evaluations of compounds 30a, 30d – 30p.
1
4

a
Enzymatic IC /nM
Cell Inhibitory
50
b
a
(
%Enzymatic activity)
IC (μM)
2
3
4
50
Comp.
R
R
R
CDK9/
cyclin T1
CDK2/
cyclin A
MV4-11
3
0a
0d
morpholino
4-methylpiperazin-1-yl
4-ethylpiperazin-1-yl
piperidin-1-yl
acetyl
acetyl
acetyl
acetyl
propionyl
H
H
H
H
H
H
H
F
8.8
1328
0.25
3
16.4
10.5
6.5
(74.8±1.3)
(61.8±8.0)
(73.1±9.4)
1087
0.52
0.74
0.73
0.43
2.28
0.09
0.40
0.55
0.42
3
0e
3
0f
3
0g
0h
morpholino
3.6
3
morpholino
18.2
2.0
1330
3
0i
morpholino
acetyl
acetyl
acetyl
acetyl
320
3
0j
4-methylpiperazin-1-yl
4-ethylpiperazin-1-yl
1,4-diazepan-1-yl
F
1.5
336
3
0k
F
1.2
305
3
0l
F
1.6
496
1
5

3
0m
morpholino
morpholino
propionyl
H
F
F
0.7
1.6
3.6
1.7
338
(22.0±0.1)
401
0.42
3.11
1.01
1.39
3
0n
0o
0p
3
morpholino
acetyl
acetyl
Cl
Cl
3
4-methylpiperazin-1-yl
642
a
b
The mean values of IC50 were measured in duplicate. % Enzyme activities were measured at 1.0 μM
concentration of test compound, n=2.
2
.4. Kinase Selectivity Profile
As the coumarin derivatives demonstrated satisfying CDK9 inhibitory activities, further studies of the kinase
selectivity were conducted. Representative compound 30i was screened in a panel of structurally similar CDK
kinase isoforms, including CDK1/cyclin B, CDK2/cyclin A, CDK2/cyclin E, CDK3/cyclin E, CDK4/cyclin D1,
CDK5/p25, CDK6/cyclin D3, CDK7/cyclin H, CDK8/cyclin C, CDK9/cyclin T1 and CDK19/cyclin C. The
CDK family screening results (Figure 3A/B) showed reasonable CDK9 selectivity. Compound 30i exhibited
the most prominent inhibition of CDK9/cyclin T1 (IC : 2 nM) and presented 160- to 978-fold selectivity over
5
0
CDK1/2/3/4/5/6 (cell cycle CDKs) and more than 3250-fold selectivity over CDK7/8/19 (transcriptional
CDKs), indicating that 30i is highly selective for the inhibition of CDK9 among the CDK family members.
1
6

Figure 3. Selectivity profiling of representative compound 30i. (A) The IC50 values and (B) dose-activity
curves of different CDK family members demonstrated the high selectivity of 30i for CDK9. (C)
Comprehensive kinase selectivity profile of 30i with 372 wild-type kinases. Determination was carried out in
3
3
duplicate at a single dose concentration of 1.0 μM of 30i, and the [γ- P]ATP-based enzymatic assay was
performed by Reaction Biology Corporation. The TREEspot image was generated by the online KinMap
program (http://kinhub.org/kinmap/) and the red circle codes for kinase activity.
In order to further investigate the kinase selectivity among other protein kinases, 30i was screened against a
comprehensive panel of 372 protein kinases using the kinase assay services at Reaction Biology Corporation,
including the kinase families AGC, CAMK, CMGC, CK1, TK, TKL and others (Figure 3C; for detailed data,
see supplementary Table S1). Impressively, 30i showed the most potent inhibition against CDK9, strong
inhibition against GSK-3α/β and MAK (enzymatic activity below 10%), medium inhibition against 8 types of
kinases (enzymatic activity ranging from 10% to 35%), weak inhibition against 76 types of kinases (enzymatic
activity ranging from 35% to 80%), and hardly any inhibition against 282 types of kinases (enzymatic activity
1
7

over 80%). The selectivity screening of CDK family isoforms and protein kinases demonstrates that 30i is a
significantly selective CDK9 inhibitor.
2
.5. Molecular Modelling of Compound 30i with CDK9
Given that compound 30i was highly potent and selective inhibition of CDK9, we tried to understand the
binding mechanism of 30i by molecular modelling. The compound 30i was docked into the active site of
CDK9 (Figure 4A), and 30i adopts the same binding mode as hit compound 7 and is located in a narrow cleft
of the ATP binding site between the N-terminal and C-terminal lobes. The 7-(pyrimidin-2-ylamino)-2H-
chromen-2-one core forms three hydrogen bonds in the hinge region of CDK9 (pyrimidin-2-ylamino 1N···H-N:
Cys106, 2C-N-H···O: Cys106, and coumarin 2C=O···H-N: Asp109). The pyrimidine ring is sandwiched
between the hydrophobic side chains Ala46 and Leu156, forming van der Waals interactions at the adenine
area of ATP (Figure S4). The phenyl ring of coumarin is contacted from above by Ile25 to enable Van der
Waals interactions at the edge of the ATP-binding area. Furthermore, the 4-fluoro-2-methoxyphenyl substituent
of 30i occupies the hydrophobic region and likely induces lowering of the glycine-rich loop (G-Loop) into the
ATP binding site, enhancing interactions within the ATP-binding pocket.
The C4-morpholino and C3-acetyl of coumarin are located in the entranceway of the ATP-binding pocket
and extend into the solvent-accessible area. The C4-morpholino sticks out of the binding cleft and seemingly
does not directly interact with CDK9, which explains the fact that small (i.e., C4-methyl) or bulky (i.e., C4-
morpholino or C4-1,4-diazepan) functional groups are found to be favourable and tolerable. As mentioned,
CDK9 possesses a more flexible and flatter active site than other CDKs [37]. The 3-acetyl-4-morpholino-2H-
chromen-2-one moiety of 30i is available to form a hydrogen bond with Asp109 and occupy the flexible
hinge/αD region of CDK9 comfortably, whereas there is steric hindrance in the other CDK family members.
Overlapping of the crystal structures of CDK2 (cyan, PDB: 4BCP) and CDK9 (grey, PDB: 4BCG), the steric
hindrance from the Lys89 residue of CDK2 obviously hampers the interactions between CDK2 and the 3-
acetyl-4-morpholino-2H-chromen-2-one moiety of 30i (Figure 4C), which results in a structural reversal to
avoid this steric hindrance and creates an unfavourable binding model when 30i is docked into CDK2 (Figure
S4). The steric hindrance also exists in other CDKs, which may explain the highly selective inhibition against
CDK9.
1
8

Figure 4. Binding mode analysis of compound 30i docked into CDK9 kinase (grey ribbon representation,
PDB code: 4BCG). The docking images were generated by the PyMOL molecular graphics system. The key
residues between CDK9 and CDK2 are labelled, and hydrogen bonds in all panels are depicted by black dotted
lines. (A) Compound 30i (green stick model) docked into CDK9. (B) Comparison of the binding mode of 30i
within CDK9 overlaid onto the equivalent region of CDK2. The steric hindrance is shown by red dotted lines.
2
.6. In Vitro Cell Assays
Further cellular antiproliferative activities of 30a and 30i were evaluated against a panel of different types of
cancer cell lines (e.g., pancreatic cancer, gastric cancer, melanoma, liver cancer, breast cancer, colon cancer
and non-small-cell lung cancer and leukaemia). The results are depicted in Table 4. As expected, the growth of
all cancer cell lines was suppressed. 30i displayed much higher antiproliferative activities than 30a. In
particular, the IC50 values of 30i inhibition against BxPC-3 (pancreatic cancer), A-375 (melanoma), Hep G2
(liver cancer) and MV-4-11 (leukaemia) cells were less than 0.1 μM, indicating that 30i might be the most
prominent antitumour agent among this series.
Table 4. Antiproliferative Activities of 30a and 30i against a Panel of Human tumour Cell Lines.
a
IC (μM)
5
0
Panel
Cell line
3
0a
30i
BxPC-3
PANC-1
0.91
1.88
1.60
3.58
0.96
0.08
0.30
0.46
0.60
0.09
Pancreatic Cancer
MIA PaCa-2
NCI-N87
A-375
Gastric Cancer
Melanoma
1
9

Liver Cancer
Breast Cancer
Hep G2
MDA-MB-231
SW620
0.56
1.58
2.70
0.66
2.36
2.15
0.26
0.25
0.10
0.23
0.43
0.28
0.47
0.64
0.26
0.09
Colon Cancer
HCT-116
A549
Non-Small Cell
Lung Cancer
NCI-H460
molm-13
MV-4-11
Leukaemia
a
The mean IC50 values were measured in duplicate.
2
.7. Pharmacokinetics of 30a and 30i in Liver Microsomes of Multiple Species
Following the enzymatic and cellular activities evaluation, a preliminary metabolic stability study of 30a
and 30i was assessed by in vitro liver microsomes of multiple species, including Homo sapiens, monkeys, dogs
and rats. The results are shown in Table 5. These compounds had moderate terminal half-lives (T1/2) with
values of at least 0.5 h in human liver microsomes. This series of compounds has an acceptable level of
metabolic stability for subsequent in vivo evaluations.
Table 5. Pharmacokinetics in Liver Microsomes of Multiple Species.
a
b
T1/2
CLint
Compounds
Species
(
min) (µL/min/mg protein)
3
3.5
2.2
4.1
.61
5.7
4.2
4.0
.01
2.1
.29
.31
.82
41.4
62.5
98.3
144
Human
Monkey
Dog
2
1
9
3
2
1
2
1
6
2
3
3
0a
Rat
38.8
57.3
98.7
689
Human
Monkey
Dog
3
0i
Rat
287
Human
Monkey
Dog
551
c
Midazolam
1499
908
Rat
a
b
c
T1/2 is half-life. CL is the intrinsic clearance. Midazolam was used as control compound.
int
2
.8. hERG assay
Inhibition of the cardiovascular ion channel hERG is the most important mechanism for drug-induced
2
0

prolongation of QT interval [51]. To understand the cardiotoxic effects, we assayed the hERG activity on
cardiovascular ion channels before the compound advanced into in vivo studies. The assay was based on the
competition of a fluorescently labelled tracer binding to a membrane preparation containing hERG. The results
are shown in Figure S5. Compound 30i had low hERG inhibitory activity (IC50: 10080 nM) and over 5000-
fold selectivity for CDK9, indicating that it has low potential for detrimental effects on the heart.
2
.9. Pharmacokinetic Profile of 30i in vivo
As compound 30i showed high potency in vitro, the pharmacokinetic profile was evaluated in ICR mice
through oral gavage (p.o.) and intravenous administration (i.v.). The plasma concentration-time curve is shown
in Figure S6. After a single 30 mg/kg dose administered by oral gavage, the maximum concentration (Cmax
reached 665 ng/mL, the area under the curve (AUC ) was 1200 ng/mL·h and the bioavailability (F%) was
)
0-t
1
4.3%. After intravenous administration of a 2 mg/kg dose, the half-life (T1/2) was found to be 0.23 h, the
clearance (CL) was 3.56 L/h/kg, and the volume of distribution (Vd ) was 0.67 L/kg. The moderate
ss
pharmacokinetic (PK) properties (Table 6) suggested that compound 30i was suitable for intravenous
administration to evaluate its antitumour efficacy in vivo.
Table 6. Pharmacokinetic analysis of 30i in ICR mice by p.o. and i.v. route.
30i
a
Parameters
T1/2
Unit
h
p.o., 30 mg/kg i.v., 2 mg/kg
NR
0.25
-
0.23
-
Tmax
h
-
1
C0
ng·mL
ng·mL
3060
-
-1
Cmax
665
1200
NR
-
-
1
AUC0-t
AUC0-∞
CL
ng·h·mL
560
561
3.56
0.67
-
-
1
ng·h·mL
-
1
-1
L·kg ·h
-
1
Vdss
L·kg
%
-
F
14.3
a
ICR mice (n = 9). Parameters were calculated from composite mean plasma concentration-time data. NR,
no record.
2
.10. In Vivo Efficacy of Compound 30i in the MV4-11 Tumour Xenograft Model
2
1

Compound 30i showed the best tumour cell growth inhibition in vitro, which facilitated us to further study
its potential as an antitumour agent. The antitumour efficacy of 30i was assessed in an MV4-11 mice xenograft
model with intravenous administration for 21 consecutive days. When the tumours grew to an average volume
3
of 180 mm , the mice were randomly assigned to five groups (6 mice per group). Three experimental groups
received 10, 20 or 40 mg/kg (mpk) doses of 30i, and the other two control groups were treated with vehicle
(
saline) as a negative control and a 40 mpk dose of cytarabine as a positive control [52].
The antitumour efficacy is shown in Figure 5. Compound 30i demonstrated marked inhibition of tumour
growth in vivo in a dose-dependent manner. In particular, treatment with 40 mpk doses of 30i significantly
suppressed the tumour progression and tumour growth inhibition (TGI) values up to 100% from days 14 to 32.
Compound 30i was more efficacious than cytarabine at the same dose of 40 mpk. It is worth noting that no
obvious body weight loss or mortality occurred in any of the mice groups during the treatment period,
indicating that there was no general cytotoxicity below a dose of 40 mg/kg/day and a good safety window for
further biological evaluation. The significant reduction in tumour volume was photographed at the end of
treatment. The average tumour weight after treatment with 30i (experimental group of 40 mpk doses) was 0.18
g, which was less than the tumour weights from the positive control group of cytarabine (0.26 g) and negative
control (1.00 g) groups.
2
2

Figure 5. The antitumour efficacy of 30i in the MV4-11 xenograft model. Female BALB/c nude mice
bearing MV4-11 tumour xenografts were treated by intravenous injection (i.v., QD) with 30i at 0 (vehicle), 10,
2
0, or 40 mpk or with cytarabine at 40 mpk. (A) Tumour volumes and (B) body weights were measured in the
MV4-11 xenograft mice after 30i or cytarabine administration. (C) Representative photographs of the tumours
in each group after treatment with 0 (vehicle), 10, 20, or 40 mpk 30i or 40 mpk cytarabine. (D) Comparison of
the final tumour weight in each group after 21 days treatment period. Numbers on the columns indicate the
mean tumour weight in each group. (∗∗) p < 0.01.
2
.11. Cellular Mode of Action
The cellular mode of action of 30i was investigated for CDK9-mediated signalling pathways by western blot
analysis. As expected, after treatment of MV4-11 cells for a period of 24 h, a gradual decrease in
phosphorylation at Ser2 of the CTD of RNAPII (p-RNAPII (S2), a CDK9-mediated phosphorylation site) was
observed in a dose-dependent manner. The expression levels of Mcl-1 and c-Myc were clearly reduced at a
concentration of 0.1 μM 30i, and the complete disappearance of Mcl-1 and c-Myc were observed in the range
2
3

of 0.2 μM to 0.8 μM 30i. Analogous results were also observed with 0.8 μM flavopiridol, such as a decrease in
phosphorylation at Ser2 of the CTD of RNAPII and a reduction in Mcl-1 and c-Myc. Thus, cellular CDK9
inhibition with compound 30i was achieved.
Figure 6. Western blot analysis of MV4-11 cells treated with 30i or Flavopiridol for 24 h. GAPDH
antibody was used as an internal control.
3
. Conclusions
Specific inhibition of CDK9 has been validated as a viable approach for targeted cancer therapy. In this
paper, we discovered a coumarin derivative 7 as a hit molecule with selective inhibition against CDK9. Then, a
series of derivatives of 7-(pyrimidin-2-ylamino)-2H-chromen-2-one and 7-(pyrimidin-4-ylamino)-2H-
chromen-2-one were designed, synthesized and evaluated as antitumour agents. This effort led to the discovery
of a highly promising potent and selective CDK9 inhibitor 30i, which effectively inhibited CDK9 with an IC50
value of 2 nM and presented 160- to 978-fold selectivity versus CDK1/2/3/4/5/6 (cell cycle CDKs) and over
3
250-fold selectivity versus CDK7/8/19 (transcriptional CDKs). Further profiles of kinase selectivity showed
that 30i most selectively inhibit CDK9 among 372 kinases. Analysis of the binding mode demonstrated that the
-acetyl-4-morpholino-2H-chromen-2-one moiety occupied the flexible hinge/αD region of CDK9 which
3
shows the steric hindrance in other CDKs, and this may account for its remarkable selectivity. Furthermore, 30i
demonstrated impressive antiproliferative activity in a panel of tumour cell lines, including leukaemia,
pancreatic cancer, gastric cancer, melanoma, liver cancer, breast cancer, colon cancer and non-small-cell lung
cancer. The moderate metabolic stability in liver microsomes and low hERG inhibition demonstrated
2
4

acceptable drug-like properties. In vivo, 30i exhibited a moderate pharmacokinetic profile, and it significantly
induced tumour growth inhibition in a dose-dependent manner in an MV4-11 (AML) mice xenograft model
without causing mortality or an obvious loss of body weight. The study of cellular mode of action suggests 30i
downregulated Mcl-1 and c-Myc by dose-dependently inhibiting the CDK9-mediated phosphorylation of
RNAPII. In summary, this study identified a promising coumarin-based CDK9 inhibitor 30i with high potency
and specificity which may shed light for developing CDK9-mediated cancer therapies.
2
5

4
. Experimental section
4
.1. General Methods. All reactions were monitored by analytical thin-layer chromatography (TLC), which
was visualized by ultraviolet light (254 nm or 356 nM). All solvents were obtained from commercial sources
and were purified according to standard procedures. Purification of the products was accomplished by flash
chromatography using silica gel (100-200 mesh). Melting points were determined by X-4 digital display
micro-melting point apparatus (Beijing Tech Instrument Co., Ltd.). All NMR spectra were recorded with
Bruker AVANCE AV-600 spectrometer at 300 MHz or 400 MHz in DMSO or CDCl : chemical shifts (δ) are
3
given in ppm, coupling constants (J) in Hz and following abbreviations are used to indicate the multiplicity in
NMR spectra: s (singlet); d (doublet); t (triplet); q (quartet); m (multiplet). The solvent signals were used as
references (residual DMSO in DMSO-d : δ = 2.50 ppm, δ = 39.5 ppm; CHCl in CDCl : δ = 7.26 ppm, δ =
6
H
c
3
3
H
c
7
7.0 ppm). Mass spectra were obtained on the Agilent 1100 LC/MSD mass spectrometer (Agilent, USA) and
Q-tofmicro MS (micromass company). High resolution mass spectrometry (HRMS) was recorded on TOF
+
perimer for ES . The purity of biologically evaluated compounds was >95% as determined by HPLC analysis
(
Agilent C18 column, CH OH and H O as the mobile phase, monitored by UV absorption at 250 and 365 nm).
3 2
4.1.1. 7-((tert-butyldimethylsilyl)oxy)-4-methyl-2H-chromen-2-one (9). 7-hydroxy-4-methyl-2H-chromen-2-
one (1.76 g, 10.00 mmol, purchased from BidePharm Company Inc.) and t-butyldimethylsilyl chloride (4.52 g,
3
5
0.00 mmol) were dissolved in DMF (50 mL) at room temperature and then treated with 1H-imidazole (3.40 g,
0.00 mmol). After 6 hours ethyl ether (50 mL × 3) and water (250 mL) was poured into the solution and the
reaction mixture was transferred into a separation funnel. The reaction mixture was washed with brine (25 mL
3). The organic layer was dried over magnesium sulfate and concentrated to yield the residue, which was
×
chromatographed (ethyl acetate/petroleum, 4%) on silica gel to afford white solid (2.10 g, 72%). ESI-MS m/z:
+
2
91 [M+H] .
4
.1.2. 3-bromo-7-((tert-butyldimethylsilyl)oxy)-4-methyl-2H-chromen-2-one (10). N-bromosuccinimide
(
2.14 g, 12.00 mmol) was added to a solution of 9 (2.90 g, 10.00 mmol) in CH CN (50 mL). The mixture was
3
stirred at room temperature for 9 h, then ethyl acetate (50 mL × 3) and water (250 mL) was poured into the
solution and the reaction mixture was transferred into a separation funnel. The reaction mixture was washed
with brine (25 mL × 3). The organic layer was dried over magnesium sulfate and concentrated to yield the
2
6

residue, which was chromatographed (ethyl acetate/petroleum, 1%) on silica gel to afford white solid (2.96 g,
+
8
0%). ESI-MS m/z: 370 [M+H] .
4
.1.3. 7-((tert-butyldimethylsilyl)oxy)-3-(1-ethoxyvinyl)-4-methyl-2H-chromen-2-one (11). 10 (1.00 g, 2.70
mmol) was combined with tributyl(1-ethoxyvinyl)stannane (1.82 mL, 5.40 mmol) and PdCl (PPh (189 mg,
.27 mmol) in anhydrous DMF (30 mL). The mixture was stirred at 80 °C for 6 h, then ethyl acetate (50 mL ×
) and water (150 ml) was poured into the solution and the reaction mixture was transferred into a separation
2
3 2
)
0
3
funnel. The reaction mixture was washed with brine (50 mL × 3). The organic layer was dried over magnesium
sulfate and concentrated to yield the crude product (0.73 g, 75%), which was available in the next step without
further purification.
4
.1.4. 3-acetyl-7-hydroxy-4-methyl-2H-chromen-2-one (12). The crude intermediate of 11 (500 mg, 0.39
mmol) was dissolved in HCl solution (20 mL, 6M in methanol) and was stirred at 80 °C for 12 h. After
completion of the reaction as monitored by TLC, the mixture was cooled to room temperature. The pH of
solution was adjusted to 7 by NaOH slowly. The solvent was evaporated under reduced pressure. The residue
was purified by chromatography (ethyl acetate/hexane, 25%) on silica gel to afford white solid (187 mg, 62%).
1
H-NMR(300 MHz, DMSO-d
6
) δ 10.81 (s, 1H), 7.75 (d, J = 8.8 Hz, 1H), 6.86 (dd, J = 8.8, 2.3 Hz, 1H), 6.75
+
(
d, J = 2.3 Hz, 1H), 2.46 (s, 3H), 2.34 (s, 3H). ESI-MS m/z: 241 [M+Na] .
4
.1.5. 3-acetyl-4-methyl-2-oxo-2H-chromen-7-yl trifluoromethanesulfonate (13). To an oven-dried 25 mL
round-bottom flask was charged with 12 (110 mg, 0.50 mmol) and anhydrous triethylamine (78 μL, 0.60 mmol)
o
in dry dichloromethane (10 mL). The resulting mixture was stirred at -10 C for 15 mins, then
trifluormethanesulfonic anhydride (100 μL, 0.60 mmol) was added dropwise over a period of 15 min. The
mixture was further stirred at room temperature for additional 1 h until the completion of the reaction as
monitored by TLC. The reaction was quenched with a saturated solution of aqueous NaHCO in an ice bath
3
and washed with water. The organic layer was dried over anhydrous Na SO , and evaporated under reduced
2
4
pressure. The residue was purified by chromatography (ethyl acetate/hexane, 15%) on silica gel to afford white
1
solid (148 mg, 84%). H-NMR(300 MHz, DMSO-d ) δ 8.10 (d, J = 8.9 Hz, 1H), 7.83 (d, J = 2.4 Hz, 1H), 7.59
6
+
(
dd, J = 8.9, 2.4 Hz, 1H), 2.48 (s, 3H), 2.40 (s, 3H). ESI-MS m/z: 373 [M+Na] .
4
.1.6.1. 4-(2-methoxyphenyl)pyrimidin-2-amine (15a). To a round-bottom flask was charged with 4-
2
7

chloropyrimidin-2-amine (500 mg, 3.9 mmol), 2-methoxyphenylboronic acid (608 mg, 4.0 mmol) Pd(dppf)Cl2
0.2 mmol) and base Na CO (7.8 mmol) under nitrogen atmosphere, then 1,4-dioxane (21 mL) and water (3
mL) were added and the vessel was immediately sealed tightly. The resulting mixture was heated at 95 °C for
h until the completion of the reaction as monitored by TLC. The cooled mixture was diluted with water and
(
2
3
2
exhaustively extracted with ethyl acetate (30 mL × 3). The organic phase was washed by brine, dried over
anhydrous Na SO , and evaporated under reduced pressure. Purification by column chromatography (ethyl
2
4
1
acetate/petroleum ether, 30%) on silica gel to afford products white solid (678 mg, 86%). H-NMR (300 MHz,
DMSO-d ) δ 8.23 (d, J = 5.2 Hz, 1H), 7.79 (dd, J = 7.7, 1.7 Hz, 1H), 7.38 – 7.49 (m, 1H), 7.14 (d, J = 8.1 Hz,
6
+
1
H), 7.00 – 7.10 (m, 2H), 6.58 (s, 2H), 3.84 (s, 3H). ESI-MS m/z: 202 [M+H] .
4
.1.6.2. 6-(2-methoxyphenyl)pyrimidin-4-amine (15b). The title compound was prepared according to the
same procedure of 15a on 6-chloropyrimidin-4-amine (500 mg, 3.9 mmol) and (2-methoxyphenyl)boronic acid
(
608 mg, 4.0 mmol). Purification by column chromatography (ethyl acetate/petroleum ether, 30%) on silica gel
1
to afford products white solid (661 mg, 87%). H-NMR (300 MHz, DMSO-d
6
) δ 8.41 (d, J = 1.2 Hz, 1H), 7.85
(
dd, J = 7.7, 1.8 Hz, 1H), 7.44–7.38 (m, 1H), 7.14 (dd, J = 8.4, 1.1 Hz, 1H), 6.99–7.08 (m, 2H), 6.83 (s, 2H),
+
3
.85 (s, 3H).ESI-MS m/z: 202 [M+H] .
.1.7.1. 3-acetyl-7-((4-(2-methoxyphenyl)pyrimidin-2-yl)amino)-4-methyl-2H-chromen-2-one (16a). To a
round-bottom flask was charged with the intermediate 15a (24 mg, 0.12 mmol), 13 (42 mg, 0.12 mmol),
Pd(OAc) (0.01 mmol), Xantphos (0.01 mmol) and base Cs CO (78 mg, 0.24 mmol) under nitrogen
4
2
2
3
atmosphere, then anhydrous toluene (5 mL) was added and the vessel was immediately sealed tightly. The
resulting mixture was heated at 100 °C for 6 h until the completion of the reaction as monitored by TLC. The
cooled mixture was diluted with water and exhaustively extracted with ethyl acetate (10 mL × 3). The organic
phase was washed by brine, dried over anhydrous Na SO , and evaporated under reduced pressure. Purification
2
4
by column chromatography (ethyl acetate/petroleum ether, 50%) on silica gel to afford products (27 mg, 56%
1
yield) as a white solid. mp: 140-142 °C. H-NMR (300 MHz, DMSO-d
6
) δ 10.38 (s, 1H), 8.62 (d, J = 5.2 Hz,
1
7
H), 8.20 (d, J = 1.8 Hz, 1H), 7.88 – 7.96 (m, 1H), 7.84 (d, J = 8.9 Hz, 1H), 7.74 (dd, J = 8.9, 1.9 Hz, 1H),
13
.42 – 7.59 (m, 2H), 7.23 (d, J = 8.3 Hz, 1H), 7.12-7.17 (m, 1H), 3.90 (s, 3H), 2.48 (s, 3H), 2.37 (s, 3H).
C
NMR (100 MHz, DMSO-d ) δ 201.32, 163.71, 159.88, 159.27, 158.45, 158.13, 154.08, 151.40, 145.86, 132.39,
6
2
8

1
30.72, 127.42, 126.17, 123.67, 121.19, 115.82, 114.62, 113.01, 112.73, 104.34, 56.19, 31.45, 15.58. HRMS
+
(
ESI) m/z [M+H] calcd for C23
H
20
N
3
O
4
: 402.1448, found: 402.1456. HPLC retention time 5.85 min, purity
9
7.5%.
.1.7.2. 3-acetyl-7-((6-(2-methoxyphenyl)pyrimidin-4-yl)amino)-4-methyl-2H-chromen-2-one (16b). The
4
title compound was prepared according to the same procedure of 16a on 0.36 mmol scale 15b and 13.
Purification by column chromatography (ethyl acetate/petroleum ether, 50%) on silica gel to afford products
1
(
76 mg, 53% yield) as a white solid. mp: 234-236 °C. H-NMR (300 MHz, DMSO-d
6
) δ 10.29 (s, 1H), 8.87 (s,
1
7
H), 8.19 (d, J = 2.0 Hz, 1H), 8.00 (dd, J = 7.7, 1.7 Hz, 1H), 7.88 (d, J = 8.9 Hz, 1H), 7.65 – 7.56 (m, 2H),
13
.54 – 7.45 (m, 1H), 7.22 (d, J = 8.2 Hz, 1H), 7.08 – 7.13 (m, 1H), 3.93 (s, 3H), 2.49 (s, 3H), 2.38 (s, 3H). C
) δ 201.29, 160.64, 160.32, 159.13, 158.19, 158.13, 153.92, 151.16, 145.16, 132.02,
30.76, 127.62, 125.87, 124.17, 121.15, 116.26, 113.63, 112.67, 108.79, 105.36, 56.27, 31.44, 15.60. HRMS
NMR (100 MHz, DMSO-d
6
1
+
(
ESI) m/z [M+H] calcd for C H N O : 402.1448, found: 402.1450. HPLC retention time 4.75 min, purity
23 20 3 4
9
9.1 %.
.1.8. 4-hydroxy-7-methoxy-2H-chromen-2-one (18). To the solution of paeonol (10.0 g, 60.20 mmol) in
anhydrous toluene (100 mL) was added diethyl carbonate (10.7 g, 90.30 mmol), then the suspension of NaH
60% dispersion in mineral oil, 12.0 g, 300.90 mmol) in anhydrous toluene (50 mL) was added dropwise over
4
(
a period of 30 min and was stirred in an ice bath. The mixture was further stirred at 100 °C for 4 h, then it was
cooled to 0 °C. The residual NaH of mixture was quenched with cool water in an ice bath carefully and washed
with diethyl ether (50 mL × 3). The aqueous phase was acidified to pH 3 with 2M HCl slowly, then the white
precipitation was occurred. The resulting precipitated solid was collected by filtration, washed with water (50
mL × 3) followed by petroleum ether (20 mL × 3), and then dried overnight to afford white solid (10.67 g,
1
9
6
2%). H-NMR(300 MHz, DMSO-d ) δ 12.36 (s, 1H), 7.72 (d, J = 8.5 Hz, 1H), 6.95 (dd, J = 8.5, 2.4 Hz, 1H),
6
+
.91 (d, J = 2.4 Hz, 1H), 5.45 (s, 1H), 3.85 (s, 3H). ESI-MS m/z: 215 [M+Na] .
4
.1.9.1. 3-acetyl-4-hydroxy-7-methoxy-2H-chromen-2-one (19a). To an oven-dried 100 mL round-bottom
flask was charged with 18 (3.00 g, 15.61 mmol), phosphorus oxychloride (5.6 mL) and acetic acid (16 mL),
then the resulting mixture was heated at 105 °C for 1 h. After the completion of the reaction as monitored by
TLC, the mixture was slowly cooled to room temperature and the white precipitation was occurred. The
2
9