Discovery of novel CDK inhibitors via scaffold hopping from CAN508

Article abstract of DOI:10.1016/j.bmcl.2018.02.054

Cyclin-dependent kinases (CDKs) are promising drug targets for various human diseases, especially for cancers. Scaffold hopping strategy was applied on CAN508, a known selective CDK9 inhibitor, and a series of pyrazolo[3,4-b]pyridine compounds were synthe

Full text of DOI:10.1016/j.bmcl.2018.02.054

Accepted Manuscript
Discovery of novel CDK inhibitors via scaffold hopping from CAN508
Liandong Jing, Yanbo Tang, Zhiyan Xiao
PII:
S0960-894X(18)30163-X
https://doi.org/10.1016/j.bmcl.2018.02.054
BMCL 25650
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
20 December 2017
17 February 2018
27 February 2018
Please cite this article as: Jing, L., Tang, Y., Xiao, Z., Discovery of novel CDK inhibitors via scaffold hopping from
CAN508, Bioorganic & Medicinal Chemistry Letters (2018), doi: https://doi.org/10.1016/j.bmcl.2018.02.054
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.

Discovery of novel CDK inhibitors via scaffold
hopping from CAN508
Liandong Jing, Yanbo Tang, Zhiyan Xiao*
Beijing Key Laboratory of Active Substance Discovery and Druggability Evaluation,
Institute of Materia Medica, Chinese Academy of Medical Sciences and Peking Union
Medical College, Beijing 100050, China
* Corresponding author. Tel.: +86 10 63189228; fax: +86 10 63017757. E-mail
address: xiaoz@imm.ac.cn.

Abstract
Cyclin-dependent kinases (CDKs) are promising drug targets for various human
diseases, especially for cancers. Scaffold hopping strategy was applied on CAN508, a
known selective CDK9 inhibitor, and a series of pyrazolo[3,4-b]pyridine compounds
were synthesized and evaluated in vitro as CDK2 and CDK9 inhibitors. Most
compounds exhibited moderate to potent inhibitory activities against both
CDK2/cyclin A and CDK9/cyclin T1 systems. Among them, compound 2e showed
IC50 values of 0.36 M for CDK2 and 1.8 M for CDK9, respectively. Notably, the
scaffold alteration seems to cause a shift in the selectivity profile of the inhibitors. In
contrast to CAN508, compound 2k demonstrated remarkable selectivity toward
CDK2 (265-fold over CDK9). Docking studies on compound 2k provided hints for
further design of more potent and selective CDK2/CDK9 inhibitors.

Members of the cyclin-dependent kinase (CDK) family play critical roles in the
regulation of a wide range of physiological processes, including cell cycle division,
proliferation, apoptosis, gene transcription, and neuronal functions1. In the recent two
decades, accumulative evidences have revealed that CDKs are promising therapeutic
targets for a variety of human diseases, especially different types of cancers2-4. Owing
to the great biological significance of CDKs, a large number of CDK inhibitors have
been developed, and currently there are about twenty CDK inhibitors under different
stages of clinical investigations as cancer therapeutics5,6. Up to now, four CDK
inhibitors have been approved by the U.S. Food and Drug Administration (FDA)
(Figure 1). Flavopiridol (Alvociclib)7 is a pan-CDK inhibitor, and was granted an
orphan drug designation by FDA for the treatment of chronic lymphocytic leukemia
(CLL) in 2014. Palbociclib from Pfizer8 is the first selective CDK4/6 inhibitor
reached market. Two other selective CDK4/6 inhibitors, Ribociclib from Novartis9
and Abemaciclib from Eli Lilly10, have also received FDA approval in the next two
years. All the three selective CDK4/6 inhibitors have been designated for the
treatment of HR+ and HER2- breast cancer. The translational success of the selective
CDK 4/6 inhibitors has inspired extensive research interest in developing selective
CDK inhibitors. More recently, the first selective CDK9 inhibitor, Atuveciclib
(BAY1143572)11, has also entered Phase I clinical trial as an orally available cancer
therapy.

Figure 1. CDK inhibitors approved by FDA
Among all the CDK subtypes, CDK2 and CDK9 have been most intensely studied.
CDK2 is directly involved in the control of multiple events in the cell cycle, and
inactivation of its endogenous inhibitors or overexpression of CDK2 may cause
various malignancies, such as lung carcinoma, melanoma, ovarian carcinoma and
pancreatic carcinoma12. It has also been recently confirmed that inhibition of the
CDK2 kinase activity led to remarkable growth inhibition in a panel of human cancer
cell lines13. In contrast to CDK2, CDK9 participates in the regulation of gene
transcription. As a subunit of positive transcription elongation factor b (P-TEFb),
CDK9 phosphorylates the C-terminal domain of RNA pol II and then promote
transcription elongation14, 15. Inhibition of CDK9 results in the rapid depletion of
short-lived survival proteins (e.g. Mcl-1) and oncogenes (e.g. c-MYC)16. Furthermore,
CDK9 deregulation is also associated with AIDS and other human virus-caused

diseases16,17. Owing to their biological significance, numerous CDK2 and CDK9
inhibitors with divergent selectivity profiles have been identified18-20
.
We report herein the identification of novel CDK2/9 inhibitors via scaffold hopping
based on a known selective CDK9 inhibitor, CAN508 (Figure 2). CAN508 showed
an IC50 value of 0.35 μM against CDK9, and exhibited over 200-fold selectivity
toward CDK9 over CDK221. The availability of CAN508-CDK2/CDK9 complex
structures could shed light on the structural basis for selectivity and make
structure-based selectivity modulation possible22. Furthermore, CAN508 has a
relatively simple scaffold with a molecule weight of only 218, therefore, it is possible
to modulate the selectivity profile by chemical manipulation on the scaffold.
As illustrated in Figure 2, CAN508 assumed similar poses in the ATP binding
pockets of both CDK2 and CDK9. Interestingly, intramolecular hydrogen bonds
between an azo-nitrogen and an amino group on the pyrazole ring were observed, and
a pseudo six-membered ring between the benzene and the pyrazole rings were formed
via conformation locking by hydrogen bonding (Figure 3). Accordingly, a scaffold
hopping strategy was applied to generate the pyrazolo[3,4-b]pyridine molecule 1, in
which a pyridine ring was incorporated to replace the pseudo six-membered ring and
presumably lock the molecule at the active conformation. The alteration in scaffold by
eliminating the N=N moiety could also significantly increase the molecular stability.
Various aromatic rings were introduced on the pyridine ring to explore the
structure-activity relationship (SAR) of this molecular area. Further fragment growth
on the free amino group in 1 was performed to modulate the activity and selectivity

profiles of this compound class (Figure 3).
(a)
(b)
Figure 2. Interaction modes of CAN508 with CDK2 (PDB: 3TNW)22 (a) and CDK9
(PDB: 3TN8)22 (b). Green dashed lines indicate hydrogen bonds, deep purple dashed
lines represent cation-π or π-π stacking interactions and hydrophobic interactions are
shown as light purple dashed lines.
Figure 3. Molecular design based on CAN508
The preparation of target compounds 2 followed the protocol described in Scheme 1.
The starting material 3 was subjected to electrophilic substitution to produce

compound 4. Subsequent treatment with phosphorus oxychloride provided compound
5, which was transferred to intermediates 6 by Suzuki coupling reaction with different
aromatic boric acids. Target compounds 1 were prepared by cyclization of 6 with
hydrazine hydrate under reflux in ethanol. Compounds 1 reacted with acyl chloride to
give target compounds 2 eventually. Compounds 2s, 2t, and 2u-2x were prepared by
nucleophilic substitution from 2q and 2r respectively.
Scheme 1. Synthesis of compounds 1 & 2. *Reagents and conditions: (a) Br2,
CH3COOH, r.t,, overnight; (b) POCl3, reflux, 5 h; (c) ArB(OH)2, (PPh3)2PdCl2,
Na2CO3, H2O, toluene, ethanol, inert atmosphere, 70℃-90℃, overnight; (d) N2H4·H2O,
ethanol, reflux, 6 h; (e) acyl chloride, DMF, trimethylamine, 0℃-r.t., 2 h; (f) DIEA,
DMF, 60℃, 2 h. *Ar and R are defined in Table 1.
Twenty-nine compounds were synthesized and evaluated with previously reported
protocols for their percentage inhibition against CDK2/cyclin A23 and CDK9/cyclin

T124 under the concentration of 10 M (Table 1). In the first round, compound 1a
with the same Ar group as CAN508 was first synthesized. As shown in Table 1, 1a
showed moderate inhibition against the two systems, which implied the feasibility of
the scaffold hopping strategy. The Ar groups were subsequently differed on the
pyridine ring (1b-1e). Compounds 1b and 1c preserved the moderate inhibitory
activity in 1a, while compounds 1d and 1e showed no noticeable inhibition. Such
observations suggested that different Ar group could affect the inhibitory activity of
the compound class. The initial results motivated us to perform further molecular
optimization and SAR exploration on 1a. In the second round, we attempted first to
extend the free amino with cyclohexylacyl (2a) and various aromatic groups were
further introduced as the Ar moiety (2b-2j). The incorporation of cyclohexylacyl did
show remarkable influence on the potency, but the effects varied with different Ar
substitutions. Such a modification could lead to either significantly improved potency
(2a vs 1a; 2e vs 1c) or slightly deceased activity (2d vs 1b). The discrepancy in
potency of 1 and 2 suggested that compounds 1 and 2 might assume different poses in
the binding pocket. Among the various Ar groups, electron-withdrawing substituents
on the benzene ring were generally unfavorable for inhibitory activity (2d, 2f, 2g and
2h), and the position of substitution could also affect the activity (2a vs 2b and 2c).
However, if the substituted benzene ring was replaced by heterocycles, the CDK
inhibition effect could be maintained (2j) or even superior to the benzene ring
congeners (2e). With these results, we inferred that there must be some key
interactions between the Ar group and residues in the binding sites of CDK2/9. For

most compounds in the first two rounds, they showed comparable inhibition against
CDK2 and CDK9 under the tested concentration. In the third round of SAR
exploration, the Ar group was fixed as 4-hydroxyl phenyl, while the R group was
varied. Diverse acyl fragments with cyclohexyl (2a), aliphatic alkyl (2k-2n),
benzyl (2p), heteroatom-contained aliphatic alkyl (2s-2x) or chloro-substituted alkyl
(2q and 2r) were introduced as R group. It seemed that the R group was tolerable to a
range of structural diversity and most compounds preserved CDK inhibitory activity.
Interestingly, several compounds (e.g. 2k, 2n and 2p) obtained in this round showed
inhibitory preference toward CDK2/cyclin A rather than CDK9/cyclin T1. These
observations suggested that the selectivity profile of this compound class might be
sensitive to variation in R group. Although it is not sensible and reliable to deduce
SAR based on inhibition rates under a singular concentration, data in Table 1 did
reveal some SAR trends useful for further molecular design.
Table 1. Percentage inhibition of compounds 1 & 2 against CDK2/cyclin A and
CDK9/cyclin T1
% inhibition*
Compd.
1a
CDK2/
CDK9/
Ar
R
cyclin A
cyclin T1
H
58
48

H
H
51
33
38
50
1b
1c
H
H
NA#
NA
86
NA
NA
70
1d
1e
2a
63
62
2b
80
78
32
2c
NA
2d
96
NA
NA
NA
34
83
NA
NA
NA
31
2e
2f
2g
2h
2i
59
59
2j
63
12
2k

87
77
73
59
2l
2m
54
90
19
70
2n
2o
70
25
2p
ND$
98
ND
73
2q
2r
52
68
39
56
64
64
2s
2t
2u
59
44
51
48
21
49
2v
2w
2x
*
Percentage inhibition was tested under the concentration of 10 M, and the values

were the average of duplicate measurements.
#
NA: not active. Percentage inhibition at the concentration of 10 M was less than
10%.
$ Not determined.
Six compounds (2a, 2c, 2e, 2l, 2o and 2r) with percentage inhibition higher than 70%
against CDK9/cyclin T1 were further tested for their IC50 values (Table 2). Most
compounds showed significant inhibition, and compound 2e was the most potent
CDK9 inhibitor with an IC50 value of 1.8 M. The three most active compounds (2e,
2o and 2r) were further tested for their IC50 values against CDK2/cyclin A to probe
their selectivity profiles. For all the three compounds, they were slightly more active
against CDK2/cyclin A, and the most active compound 2e showed an IC50 value of
0.36 M against CDK2. In combination with the inhibitory preference of compounds
2k, 2n and 2p revealed in Table 1, we inferred that the molecular operation of
scaffold hopping might lead to a shift in the selectivity profile of the inhibitors. To
confirm this assumption, compound 2k was tested against both CDK2/cyclin A and
CDK9/cyclin T1 systems. As expected, it exhibited a 265-fold selectivity for CDK2
(2.0 M) over CDK9 (529.3 M). In fact, for all the four compounds with determined
IC50 values for both CDK2 and CDK9, they were significantly more potent than
CAN508 in CDK2 inhibition, while were much less active than CAN508 in CDK9
inhibition (Table 2). Such results suggested that scaffold hopping is not only a
practical way to generate novel prototype structures, but also a feasible strategy to
modulate pharmacological profiles of template structures. Further scaffold alteration

and molecular optimization on the prototype molecules could provide novel and
diverse CDK inhibitors with different selectivity profiles.
Table 2. IC50 values of selected compounds against CDK2/cyclin A and CDK9/cyclin
T1
IC50(M)*
Compd.
CDK2/cyclin A
13.1  1.2
ND#
CDK9/cyclin T1
0.15  0.01
10.4  0.6
3.6  0.3
CAN508
2a
2c
2e
2l
ND
0.36  0.01
ND
1.8  0.1
3.5  0.1
2o
2r
2k
0.82  0.04
0.66  0.24
2.0  0.6
2.4  0.4
2.4  0.2
529.3  90.8
*
The IC50 value was interpolated from dose-response data with six different
concentrations. Each test was performed in duplicate. Data represent the average of
duplicate measurements.
# ND: Not determined.
Molecular docking of compound 2k was performed to elucidate the possible binding
modes of these compounds and also to provide insights on the structural basis for
selective CDK2 inhibition. Compound 2k was docked into the ATP binding sites of
CDK2 (PDB: 3TNW) and CDK9 (PDB: 3TN8) with the flexible docking mode. The

interaction mode of 2k in CDK2 was depicted in Figure 5a. Compound 2k situated in
the ATP binding site of CDK2 and adopted a pose similar to that of CAN508. Like in
CAN508, the phenol fragment of 2k occupied the deep pocket and interacted with
Lys33 via hydrogen bonding and cation-πinteraction. The pyrazolo[3,4-b]pyridine
scaffold of 2k formed polar contacts with the hinge region of CDK2, and in particular
the pyrazolo ring as well as its 3-amino substitution interacted with Glu81 and Leu83
to form triplet hydrogen bonds. Hydrophobic interactions were also monitored
between 2k and the residues of Ala31, Leu134 and Lys33. The 2-ethylbutanoyl
moiety, which is absent in CAN508, protruded toward the outer part of the binding
site and NH in the amido group formed a hydrogen bond with Leu83. As illustrated in
Figure 5b, the interaction mode of 2k in the binding site of CDK9 was quite similar
to that in CDK2. Phenolic hydroxyl contacted with Lys48, and the 1-NH and 3-amino
groups formed hydrogen bonds with Asp104 and Cys106, respectively. Similar
hydrophobic interactions were also detected between 2k and the residues of Ala46,
Lys48, and Leu156. However, the hydrogen bond observed with the N-2 atom in
CDK2 was absent in CDK9, which might be responsible at least partially for the
relatively low inhibitory activity of 2k against CDK9.
As stated previously in the SAR discussion, we inferred that the selectivity profile of
this compound class was sensitive to the R group, which is 2-ethylbutanoyl in
compound 2k. However, no obvious polar or hydrophobic interaction was observed
between the 2-ethylbutanoyl and the kinases in the docking poses shown in Figure 5.
Recently, a series of substituted purine compounds were reported as selective CDK2

inhibitors25. Crystal structures of inhibitor-kinase complexes indicated that inhibitors
bind more tightly to CDK2 by stabilizing a unique glycine-rich loop conformation
that is preferred in CDK2. Inspired by such results, we deduced that the
2-ethylbutanoyl of 2k might stabilize the glycine-rich loop conformation which is
present in CDK2 but not in CDK9 due to the flexibility and steric effect of
2-ethylbutanoyl. This slight change in binding conformation could lead to the
presence of an extra hydrogen bond between N-2 and Leu83, and eventually to the
alteration in selectivity.
(a)
(b)
Figure 5. Interaction modes of 2k with CDK2 (a) and CDK9 (b) as suggested by the
docking study. Molecule 2k is shown in cyan stick, hydrogen bonds are depicted in
green dot lines. Hydrophobic interactions are shown in light purple dot lines.
In conclusion, a series of pyrazolo[3,4-b]pyridine compounds were designed through
scaffold hopping from the selective CDK9 inhibitor CAN508. Twenty-nine
compounds were synthesized and evaluated as CDK2/9 inhibitors, and preliminary

SAR was also deduced. Most of these compounds showed moderate to potent
inhibition against both CDK2 and CDK9, and the most active compounds exhibited
IC50 values at micromolar or even submicromolar level. In contrast to CAN508,
compound 2k showed 265-fold selectivity toward CDK2. Molecular docking of
compound 2k in the binding sites of CDK2 and CDK9 provided possible new clues
for the design of selective CDK2 or CDK9 inhibitors.
Acknowledgements
We gratefully acknowledge the financial support from National Natural Science
Foundation of China (81172985) and CAMS Innovation Fund for Medical Sciences
(CIFMS, 2016-I2M-3-009).
Supplementary data
1
All target compounds were characterized by melting point, H NMR and HR-MS.
Supplementary information on general preparation procedures and spectroscopic data
of the target compounds are available online. Biological assays were performed by
following experimental protocols in the cited references. Molecular modeling
methods were described.
References
1. Malumbres M, Barbacid M. Mammalian cyclin-dependent kinases. Trends
Biochem Sci. 2005;30: 630-641.

2. Deshpande A, Sicinski P, Hinds PW. Cyclins and cdks in development and cancer:
a perspective. Oncogene. 2005;24: 2909-2915.
3. Lim S, Kaldis P. Cdks, cyclins and CKIs: roles beyond cell cycle regulation.
Development. 2013;140: 3079-3093.
4. Asghar U, Witkiewicz AK, Turner NC, Knudsen ES. The history and future of
targeting cyclin-dependent kinases in cancer therapy. Nat Rev Drug Discov. 2015;14:
130-146.
5. Li T, Weng T, Zuo M, Wei Z, Chen M, Li Z. Recent progress of cyclin-dependent
kinase inhibitors as potential anticancer agents. Future Med Chem. 2016;8:
2047-2076.
6. Tutone M, Almerico AM. Recent advances on CDK inhibitors: An insight by
means of in silico methods. Eur J Med Chem. 2017;142: 300-315.
7. Wiernik PH, Alvocidib (flavopiridol) for the treatment of chronic lymphocytic
leukemia. Expert Opin Investig Drugs. 2016;25:729–734.
8. Mangini NS, Wesolowski R, Ramaswamy B, Lustberg MB, Berger MJ.
Palbociclib: a novel cyclin-dependent kinase inhibitor for hormone receptor-positive
advanced breast cancer. Ann Pharmacother. 2015;49: 1252-1260.
9. Syed YY. Ribociclib: first global approval. Drugs. 2017;77: 799-807.
10. Iriyama N, Hino H, Moriya S, Hiramoto M, Hatta Y, Takei M, Miyazawa K. The
cyclin-dependent kinase 4/6 inhibitor, abemaciclib, exerts dose-dependent cytostatic
and cytocidal effects and induces autophagy in multiple myeloma cells. Leuk
Lymphoma. 2017; doi: 10.1080/10428194.2017.1376741.

11. Lücking U, Scholz A, Lienau P, Siemeister G, Kosemund D, Bohlmann R, Briem
H, Terebesi I, Meyer K, Prelle K. Identification of atuveciclib (BAY 1143572), the
first highly selective, clinical PTEFb/CDK9 inhibitor for the treatment of cancer.
Chemmedchem. 2017;12: 1776-1793.
12. Chohan TA, Qian H, Pan Y, Chen JZ. Cyclin-dependent kinase-2 as a target for
cancer therapy: progress in the development of CDK2 inhibitors as anti-cancer agents.
Curr Med Chem. 2015;22: 237-263.
13. Dai H, Huskey NE, Kusdra L, Wohlbold L, Merrick KA, Zhang C, Creasman KJ,
Shokat KM, Fisher RP, Goga A. Chemical-genetic analysis of cyclin dependent
kinase 2 function reveals an important role in cellular transformation by multiple
oncogenic pathways. Proc Natl Acad Sci U S A. 2012;109: 1019-1027.
14. Garriga J,Grana X. Cellular control of gene expression by T-type cyclin/CDK9
complexes. Gene. 2004;337: 15-23.
15. Price DH. P-TEFb, a cyclin-dependent kinase controlling elongation by RNA
polymerase II. Mol Cell Biol. 2000;20: 2629-2634.
16. Sonawane YA, Taylor MA, Napoleon JV, Rana S, Contreras JI, Natarajan A.
Cyclin dependent kinase 9 inhibitors for cancer therapy. J Med Chem. 2016;59:
8667-8684.
17. Nf DSP, Durvale MC, Canduri F. The emerging picture of CDK9/P-TEFb: more
than 20 years of advances since PITALRE. Mol Biosyst. 2016;13: 246-276.
18. Di Giovanni C, Novellino E, Chilin A, Lavecchia A, Marzaro G. Investigational
drugs targeting cyclin-dependent kinases for the treatment of cancer: an update on

recent findings (2013-2016). Expert Opin Investig Drugs. 2016; 25: 1215-1230.
19. Malínková V, Vylíčil J, Kryštof V. Cyclin-dependent kinase inhibitors for cancer
therapy: a patent review (2009-2014). Expert Opin Ther Pat. 2014; 25: 1-18.
20. Asghar U, Witkiewicz AK, Turner NC, Knudsen ES. The history and future of
targeting cyclin-dependent kinases in cancer therapy. Nat Rev Drug Discov. 2015; 14:
130-146.
21. Krystof V, Cankar P, Frysová I, Slouka J, Kontopidis G, Dzubák P, Hajdúch M,
Srovnal J, Jr DAW, Orság M. 4-arylazo-3,5-diamino-1H-pyrazole CDK inhibitors:
SAR study, crystal structure in complex with CDK2, selectivity, and cellular effects. J
Med Chem. 2006;49: 6500-6509.
22. Baumli S, Hole AJ, Noble ME, Endicott JA. The CDK9 C-helix exhibits
conformational plasticity that may explain the selectivity of CAN508. ACS Chem Biol.
2012;7: 811-816.
23. Meijer L, Borgne A, Mulner O, Chong JP, Blow JJ, Inagaki N, Inagaki M,
Delcros JG, Moulinoux JP. Biochemical and cellular effects of roscovitine, a potent
and selective inhibitor of the cyclin-dependent kinases cdc2, cdk2 and cdk5. Eur J
Biochem. 1997;243: 527-536.
24. de Falco G and Giordano A. CDK9 (PITALRE): A multifunctional cdc2-related
kinase. J Cell Physiol. 1998;177: 501-506.
25. Coxon CR, Anscombe E, Harnor SJ, Martin MP, Carbain B, Golding BT,
Hardcastle IR, Harlow LK, Korolchuk S, Matheson CJ, Newell DR, Noble ME,
Sivaprakasam M, Tudhope SJ, Turner DM, Wang LZ, Wedge SR, Wong C, Griffin

RJ, Endicott JA, Cano C. Cyclin-dependent kinase (CDK) inhibitors:
structure-activity relationships and insights into the CDK-2 selectivity of 6-substituted
2-arylaminopurines. J Med Chem. 2017;60: 1746-1767.

Graphical Abstract