Discovery and SAR of novel pyrazolo[1,5-a]pyrimidines as inhibitors of CDK9

Article abstract of DOI:10.1016/j.bmc.2015.08.035

The serine-threonine kinase CDK9 is a target of emerging interest for the development of anti-cancer drugs. There are multiple lines of evidence linking CDK9 activity to cancer, including the essential role this kinase plays in transcriptional regulation

Full text of DOI:10.1016/j.bmc.2015.08.035

Accepted Manuscript
Discovery and SAR of novel pyrazolo[1,5-a]pyrimidines as inhibitors of CDK9
Louisa J. Phillipson, David H. Segal, Tracy L. Nero, Michael W. Parker, Soo
San Wan, Melanie de Silva, Mark A. Guthridge, Andrew H. Wei, Christopher
J. Burns
PII:
S0968-0896(15)30017-1
DOI:
http://dx.doi.org/10.1016/j.bmc.2015.08.035
BMC 12539
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
2 April 2015
3 August 2015
26 August 2015
Please cite this article as: Phillipson, L.J., Segal, D.H., Nero, T.L., Parker, M.W., Wan, S.S., Silva, M.d., Guthridge,
M.A., Wei, A.H., Burns, C.J., Discovery and SAR of novel pyrazolo[1,5-a]pyrimidines as inhibitors of CDK9,
Bioorganic & Medicinal Chemistry (2015), doi: http://dx.doi.org/10.1016/j.bmc.2015.08.035
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 and SAR of novel pyrazolo[1,5-a]pyrimidines as inhibitors of CDK9
a, b
a, b
c
c, d
a, b
Louisa J. Phillipson , David H. Segal , Tracy L. Nero , Michael W. Parker , Soo San Wan ,
a, b
e
e,f
a,b,g
Melanie de Silva , Mark A. Guthridge , Andrew H. Wei , Christopher J. Burns
*
a
ACRF Chemical Biology Division, Walter and Eliza Hall Institute of Medical Research, 1G Royal
Parade, VIC 3052 Australia
b
Department of Medical Biology, The University of Melbourne, 1G Royal Parade, VIC 3052 Australia
c
ACRF Rational Drug Discovery Centre, St. Vincent’s Institute of Medical Research, 41 Victoria Pde,
Fitzroy, VIC 3065 Australia
d
Department of Biochemistry and Molecular Biology, Bio21 Molecular Science and Biotechnology
Institute, University of Melbourne, 30 Flemington Road, Parkville VIC 3010 Australia
e
Division of Blood Cancers, Australian Centre for Blood Diseases, Monash University, Alfred Medical
Research and Education Precinct, 89 Commercial Rd, Melbourne VIC 3004 Australia
f
Department of Clinical Haematology, The Alfred Hospital, Commercial Rd, Melbourne, VIC 3004
Australia
g
School of Chemistry, Bio21 Molecular Science and Biotechnology Institute, The University of
Melbourne, 30 Flemington Rd, VIC 3010, Australia
1

Corresponding Author
Christopher J. Burns
ACRF Chemical Biology Division, Walter and Eliza Hall Institute of Medical Research, 1G Royal
Parade, VIC 3052 Australia
Ph: +61 (0)3 9345 2957
Fax: +61 (0)3 9347 0852
Email: burns@wehi.edu.au
Keywords
pyrazolo[1,5-a]pyrimidine, CDK9 selectivity, CDK7, PI3Kα, FLT3, molecular modelling
Abstract
The serine-threonine kinase CDK9 is a target of emerging interest for the development of anti-cancer
drugs. There are multiple lines of evidence linking CDK9 activity to cancer, including the essential
role this kinase plays in transcriptional regulation through phosphorylation of the C-terminal domain
(CTD) of RNA polymerase II. Indeed, inhibition of CDK9 has been shown to result in a reduction of
short-lived proteins such as the pro-survival protein Mcl-1 in malignant cells leading to the induction
of apoptosis. In this work we report our initial studies towards the discovery of selective CDK9
inhibitors, starting from the known multi-kinase inhibitor PIK-75 which possesses potent CDK9
activity. Our series is based on a pyrazolo[1,5-a]pyrimidine nucleus and, importantly, the resultant
2

lead compound 18b is devoid of the structural liabilities present in PIK-75 and possesses greater
selectivity.
1. Introduction
The Cyclin-Dependant Kinases (CDKs) are a family of serine-threonine kinases comprising more than
3 members. A number of CDKs (subtypes 1-4, 6 and 11) play a direct role in the regulation of the
1
1
-4
cell cycle, while CDK subtypes 7-9 have been shown to control RNA polymerase II mediated
1
transcription. CDK9 is expressed in a range of human tissues and is highly expressed in terminally
4
differentiated cells. CDK9 is stabilised by the formation of heterodimers; these are either i) transient
complexes with chaperones such as HSP70, HSP90 and Cdc37 or ii) active heterodimers with the T
and K cyclins. The active heterodimers form the complex p-TEFb (positive transcription elongation
4
factor b) where the cyclin forms the regulatory subunit and CDK9 the active enzyme component.
The p-TEFb complex plays two roles in the successful initiation and elongation of RNA transcription:
phosphorylation of the RNA polymerase II C-terminus domain and usurping N-TEF (negative
5
transcription elongation factor) leading to stable elongation of the RNA transcript.
Inhibition of specific CDKs or CDK subsets has been the focus of considerable research and drug
development activity over the last two decades with a number of inhibitors of varying selectivity
3
, 6
profiles progressing to clinical trials, predominantly for the treatment of cancer. While a number
of pan-specific CDK inhibitors (e.g. flavopiridol (1), SNS-032 (2), dinaciclib (3) and AT7519 (4)) have
been examined in clinical trials, dose-limiting toxicities have been reported most likely due to their
7
-9
broad target inhibition. More recently selective targeting of CDK9 for the treatment of various
cancers has gained interest, since blocking RNA synthesis by the inhibition of CDK9 leads to a
reduction in the levels of short-lived pro-survival proteins such as Mcl-1 and the induction of
3
, 4
apoptosis. Of the many CDK inhibitors reported to possess potent activity against the CDK9
2
, 3, 6, 10
11, 12
subtype
few have been reported as CDK9 selective,
one of which is the recently disclosed
3

1
3
LY2857785 (5). The structures for some of the inhibitors with potent CDK9 activity are shown in
Figure 1.
Figure 1.
Structures of known CDK9 inhibitors (1-5) and the PI3Kα inhibitor PIK-75 (6)
As part of a phenotypic screening study looking for drugs targeting Mcl-1 in acute myeloid leukaemia
1
4, 15
(AML), we identified the literature standard PI3K p110α (PI3Kα) inhibitor PIK-75 (6)
as being
16
particularly active against a panel of AML cell lines and primary samples. Further studies
demonstrated that this compound also possessed significant CDK7 and CDK9 inhibitory activity and
that this combination of activities was driving much of the PIK-75 mediated cytotoxicity in AML.
Studies from other laboratories have also demonstrated that PIK-75 possesses significant CDK
1
7-19
inhibitory activity resulting in the induction of apoptosis in diverse cancer cell lines.
In addition,
our group showed in AML that the direct inhibition of CDK9 by PIK-75 causes blockade of the
transcription of the MCL-1 gene thus reducing levels of the short-lived pro-survival protein, Mcl-1,
4

2
0,21
resulting in rapid cell death, given the critical survival role of Mcl-1 in AML.
Similarly, Lemke et al
showed that PIK-75 leads to a reduction of the short-lived proteins Mcl-1 and c-FLIP, and
1
8
subsequently a re-sensitisation of TRAIL resistant cancer cells to TRAIL.
These data encouraged us to initiate a drug discovery program focused on the development of
selective CDK9 inhibitors using PIK-75 as a chemical starting point. This compound, originally
described by Hayakawa et al as a selective inhibitor of the kinase PI3Kα, shows significant activity
1
4-16
against a number of kinases including CDK7, CDK9 and FLT3.
Furthermore, the compound
possesses two non-drug-like structural motifs – a hydrazone linkage and aryl nitro moiety - that are
2
2,23
perceived as developmental liabilities due to potential toxicity concerns.
Presumably as a
consequence of these issues there has been little work published exploring variations on the PIK-75
2
4-28
structure
which offered the possibility to rapidly develop novel chemical matter. Below, we
describe our preliminary work on a series of novel pyrazolo[1,5-a]pyrimidines that show an
improved selectivity profile compared to PIK-75 and that are devoid of the structural liabilities of the
original lead compound.
2
. Results and Discussion
.1 Synthesis
2
A series of sulfonylhydrazones were synthesised (Scheme 1) starting from commercially available 5-
chloropyrazolo[1,5-a]pyrimidine (7) which was converted to the carbaldehyde (8) under Vilsmeier-
2
5
Haack conditions. A one pot, two-step procedure was then used to convert 8 to the desired aryl
sulfonylhydrazones 9a-u via condensation of 8 with methyl hydrazine, followed by sulfonylation with
various sulfonyl chlorides. Use of 2,6-lutidine as base allowed isolation of the desired product in high
purity via filtration, removing the need for aqueous workup and chromatography.
5

Scheme 1
Reagents: (i) POCl
3
, DMF, RT; (ii) MeHNNH
2
.H
2
SO
4
, NaHCO
3
or 2,6-lutidine then 2-
, DMF, RT.
arylsulfonyl chloride, RT; (iii) mCPBA, CH
2
Cl , RT; (iv) HATU, DIPEA, (NH
2
4
)
2
CO
3
The pyridine N-oxide (9n) was synthesised by oxidation of the corresponding pyridine 9m using
mCPBA. Synthesis of the primary amide 9q from the corresponding acid 9p via an amide coupling
reaction with ammonium carbonate proved problematic. Extended reaction times were necessary to
effect complete conversion however these conditions also led to amination at the 5-chloro position
of the pyrazolo[1,5-a]pyrimidine core. Shortening the reaction time resulted in a mixture of
compounds from which 9q could be isolated.
Alterations to the 5-position of the pyrazolo[1,5-a]pyrimidine core were achieved by modification of
the sulfonylhydrazone 9a (Scheme 2). A series of compounds with a C-C linkage in the 5-position
were synthesised using palladium mediated chemistry. First, Suzuki-Miyaura coupling was used to
install aromatic or heteroaromatic substituents in this position giving compounds 10b-e. Synthesis of
2
9
1
0a utilised trimethylboroxane as a methyl source under conditions analogous to a Suzuki coupling,
3
0
whilst palladium catalysed cyanation with zinc cyanide was employed to furnish 11.
6

Scheme 2
Na CO , Pd(PPh
Reagents: (i) Zn(CN)
2
, Pd(PPh
3
)
4
, DMF, 80 °C; (ii) R-pinacolatoboronate, K
2
CO or
3
2
3
3 4
)
, DME or 1,4-dioxane, 80-100 °C.
A second series of compounds were synthesised with a heteroatom linked directly to the 5-position
of the pyrazolo[1,5-a]pyrimidine core (Scheme 3). Sodium methoxide was employed to give
compound 12. Compound 13b was derived from reaction with morpholine under basic conditions
while compound 13a was the unexpected result of attempts to install a primary amine in the 5-
position of 9a, under identical conditions where this addition had been observed (as a side-reaction)
in the synthesis of the primary amide 9q (vide supra).
Scheme 3
Reagents: (i) NaOMe, MeOH, 65 °C; (ii) (NH
4
)
2
CO , DIPEA, DMF, RT; or Morpholine,
3
K
2
CO , DMF, 70 °C.
3
7

Scheme 4
iii) RaNi, H
Reagents: (i) MeNO
2
, MeNH
Cl, NEt
2
.HCl, KOAc, MeOH, RT; (ii) NaCNBH
3
, THF, 0 °C – RT;
(
2
, MeOH, 40 °C; (iv) ArSO
2
3
, CH Cl , 0 °C – RT; (v) MeI, K CO , Acetone, RT.
2
2
2
3
Compounds 18a and 18b were synthesised from the aldehyde 8 (Scheme 4). The aldehyde was first
subjected to a Henry reaction to give the nitroalkene 14. Simultaneous reduction of the alkene and
nitro groups could not be achieved under various conditions (e.g. LiAlH
4
, BH ) with dimerised
3
products and partial nitro reduction being amongst the mixture of products obtained. Consequently,
compound 16 was prepared in a two-step procedure. Reduction of the alkene was achieved using
sodium cyanoborohydride; other reducing agents such as sodium borohydride resulted in compound
degradation. Nitro group reduction was achieved with Raney nickel although some dechlorinated
material, which could be removed by chromatography, was always observed. Treatment of
compound 16 with the requisite sulfonyl chlorides, yielded compounds 17a-b and alkylation of these
using iodomethane furnished the desired compounds 18a-b (respectively).
8

Scheme 5
O, dioxane, 80 °C; (iii) 4 M HCl in dioxane (for Boc protected heterocycles); ArSO
text), CH Cl , RT.
Reagents: (i) NIS, DMF, RT; (ii) Boronic acid/pinacol ester, PdCl
2
(dppf)·CH
2
Cl
2
, K
3
PO ,
4
H
2
2
Cl, base (see
2
2
A series of compounds with a heterocyclic linker from the bicycle were synthesised (Scheme 5).
Initial attempts using 5-chloro-3-bromopyrazolo[1,5-a]pyrimidine in a Suzuki-Miyaura coupling
procedure using Pd(dba)
Pd(PPh gave exclusively reaction at the C-5 position. Consequently, the 3-iodo compound 19 was
prepared from 7 using N-iodosuccinimide. Suzuki-Miyaura coupling between 19 and the desired
heterocyclic boronic acids, or pinacol esters, in the presence of PdCl (dppf)·CH Cl gave reaction
2
as a catalyst gave a mixture of the 3 and 5 regioisomers, whereas
)
3 4
2
2
2
exclusively at the C-3 position (20a-c). Compound 20a was treated with 4 M HCl in 1,4-dioxane to
remove the Boc protecting group prior to reaction with the arylsulfonyl chloride under basic
conditions to give 21a. Compounds 20b and 20c, which had no Boc group in place, were treated with
base and the aryl sulfonyl chloride to give 21b-c. For the synthesis of 21c it was necessary to use the
stronger base sodium hydride to fully deprotonate the heterocycle and drive reaction to completion.
9

2.2 In vitro biology and structure-activity relationships
Preliminary work focused on transitioning from the imidazo[1,2-a]pyridine core of PIK-75 (6) to a
pyrazolo[1,5-a]pyrimidine core, which we reasoned would increase rotational freedom for the 3-
substituent by removing the hydrogen atom from the carbon (C4). Further, the presence of the
bridgehead nitrogen (N7a) in the pyrazolo[1,5-a]pyrimidine should increase the acidity of the
3
1
adjacent CH atom to aid a putative hydrogen bond with the hinge region of the kinase. The direct
pyrazolo[1,5-a]pyrimidine analogue (9a) of PIK-75 (6) was prepared to determine the effect these
changes had against CDK9, and the PIK-75 target kinases (CDK7, PI3Kα and FLT3). Although
compound 9a showed a moderate reduction in potency against all enzymes (Table 1), we deemed
the potency level suitable for further structural elaboration.
We next explored the structure-activity relationships (SAR) around the aryl sulfonamide.
Substitution of this ring appears to be important for the activity of this series as shown by the
general loss of activity for 9b (Table 1). Introduction of a 2-methyl substituent (9c) did not restore
activity to 9b, whereas a 3-nitro substituent (9d) resulted in a compound with moderate levels of
activity for both CDK9 and FLT3. To identify replacements for the nitro group we focused initially on
mono substituted phenyl sulfonamides (Table 1). Substitution at the 3-position of the phenyl ring
(equivalent to the 5-position of the 2,5-disubstituted ring) was largely unsuccessful with the 3-cyano
compound (9h) being the one notable exception. This compound had moderate activity on CDK9 and
FLT3 and importantly had lost measurable activity against CDK7 and PI3Kα. Moving the substituent
to the 4-position (9i) had a detrimental effect on activity, as did disubstitution in the 3- and 4-
positions (9l). Substitution of the phenyl ring for a pyridyl (9m) or the N-oxide (9n) also resulted in a
loss of activity. The data for compounds 9c, 9d and 9h suggests that the 2,5-substitution pattern
plays an important role in the activity of this series by maintaining the optimum orientation of the
phenyl sulfonamide ring to enable the 5-nitro, or 5-cyano, moieties to better interact with the
kinase.
10

Table 1
Single substitutions replacing nitro
IC50 (nM)
Compound
R
CDK9
CDK7
PI3Kα
FLT3
a
PIK-75 (6)
1.37
NT
0.5
5.15
9a
96
841
271
32
9
b
>10000
>10000
291
>10000
>10000
9244
>10000
>10000
4350
4628
9
c
>10000
154
9d
9e
>10000
>10000
>10000
>10000
11

9
f
>10000
>10000
201
>10000
>10000
>10000
NT
>10000
>10000
>10000
4544
>10000
7791
9g
9h
90
9
i
>10000
>10000
>10000
>10000
>10000
8855
9
j
>10000
>10000
NT
5373
>10000
3301
9k
>10000
7840
9
l
>10000
1913
9m
>10000
>10000
12

9n
>10000
>10000
>10000
8882
a
NT = Not tested
In the absence of a crystal structure, molecular modelling and compound docking were used to
examine potential binding modes of PIK-75 (6) and the pyrazolo[1,5-a]pyrimidine compounds in the
ATP pocket of human CDK9. A putative binding mode for PIK-75 in CDK9 is illustrated in Figure 2a, b.
PIK-75 (6) consistently docked into the CDK9 ATP binding site with the 6-bromo moiety pointing
towards the back pocket (also known as the kinase specific affinity pocket; the region beyond the
gatekeeper residue Phe103, near Asp167 of the DFG motif and residues Lys48, Val79 and Ala166).
The PIK-75 imidazole N (N1) interacts with the backbone amide of Cys106 on the CDK9 hinge region
and the imidazo[1,5-a]pyridine ring system forms edge-to-face π-π interactions with Phe103
(gatekeeper residue) and Phe105 (hinge residue). Leu156 forms hydrophobic contacts with the PIK-
75 bicyclic ring system from below. The sulfonamide N-methyl group points up underneath the Gly-
rich loop (also known as the P-loop) and makes hydrophobic interactions with Ile25, Phe30 and
Val33. The sulfonamide oxygen atoms are solvent exposed and make no direct interaction with
CDK9; however, an interaction with Asp109 may be possible through a water molecule. Phe30 on
the Gly-rich loop flips down and becomes part of the ATP binding site. The PIK-75 phenylsulfonamide
ring is located under the Gly-rich loop and makes π-π interactions with Phe30 and the 2-methyl
phenyl substituent makes hydrophobic contacts with Phe30, Val33 and Lys48. Lys151 is within reach
of the 5-nitro phenyl substituent and likely makes a charge-charge interaction. The 2-methyl, 5-nitro
phenyl substituents and the N-methyl of the sulfonamide work in concert to orientate the
phenylsulfonamide ring and maintain a relatively elongated PIK-75 conformation that can access the
ATP binding site.
13

Figure 2.
Putative binding modes of PIK-75 (6) (a) and (b), compound 9a (c) and compound 11
(d) in the human CDK9 ATP binding site. See text for discussion.
Compound 9a can adopt a similar binding mode to that of PIK-75 (6) in the CDK9 ATP pocket (Figure
c). The decreased hydrophobicity of the 5-substituent (Br for PIK-75 (6); Cl for 9a) may, in part,
2
account for the difference in their activity profiles. The 3-nitro (9d) and 3-cyano (9h) compounds
suffer only a ~5-fold reduction in CDK9 and FLT3 activity compared with PIK-75, presumably due to
their ability to adopt low energy binding modes where the phenylsulfonamide substituent can
interact with Lys151. Moving the nitro to the 4-position on the phenylsulfonamide (9j) would result
in a steric clash with the Gly-rich loop, consistent with the loss of kinase activity for this compound.
14

For compounds 9b, 9c, 9e-g, 9i and 9k-n the aryl substituents allow the pyrazolo[1,5-a]pyrimidine
compounds to adopt a wider range of conformations, including those where the aryl sulfonamide
undergoes hydrophobic collapse with the bicycle to generate more compact conformers that would
have difficulty accessing the kinase ATP pocket.
Given the promising CDK9 activity of 9h, and a working understanding of the binding mode, we set
out to explore the effect of different 2,5-substituents on the phenylsulfonamide (Table 2). Replacing
the 5-nitro with the physically larger 5-bromo (9o) or 5-SO Me (9t) substituents reduced activity for
2
all four kinases, possibly due to steric clashes with the Gly-rich loop, Lys151 and Ala153. By contrast,
the absence of any measurable CDK9 activity for compounds 9p and 9q does not appear to be a
steric issue but may be related to a decrease in the hydrophobicity of the 5-substituents for these
two compounds compared to PIK-75 (6) and 9a (clogP 1.5, 0.3, 3.5 and 2.4, respectively) limiting
access to the hydrophobic ATP-binding site. Introduction of a 2-methyl substituent into 9h, to
furnish 9r, gave a 10-fold increase in potency over 9h and the compound retained 100-fold
selectivity over PI3Kα. Exchange of the 2-methyl group for 2-chloro (9u) modestly improved potency
against CDK9 (and FLT3) but significantly reduced selectivity over PI3Kα. Similarly, a 2-methoxy
substituent (9v) led to an increase in the potency against PI3Kα but a slight reduction in activity
against CDK9 and FLT3.
Table 2
Disubstituted ring replacements
15

IC50 (nM)
CDK9
1960
1
2
Compound R
R
CDK7
PI3Kα
>10000
>10000
>10000
2960
2963
9430
29
FLT3
5047
>10000
6264
33
9
9
9
9
9
9
9
o
p
q
r
Me
Me
Me
Me
Me
Me
Cl
Br
>10000
>10000
CO
2
H
>10000
>10000
22
a
C(O)NH
2
NT
CN
F
3590
NT
s
>10000
1064
>10000
6456
21
t
SO
2
Me
NT
u
NO
2
79
NT
9v
OMe
NO
2
174
NT
81
97
a
NT = Not tested
Next we sought to explore the SAR at the 5-position of the pyrazolo[1,5-a]pyrimidine ring (Table 3).
In the modelled binding mode of PIK-75 (6) and compound 9a (Figure 2a-c), the 5-position is located
at the back of the ATP pocket surrounded by Lys48, Val79, Phe103 (gatekeeper residue) and Ala166.
Substitution of the 5-chloro for a 5-methyl (10a) resulted in moderate losses in activity across CDK9,
16

PI3Kα and FLT3. Introduction of the 5-methoxy substituent (12) appeared to reduce both CDK9 and
FLT3 activity with a minimal impact of the PI3Kα activity whilst the 5-N,N-dimethylamino substituent
(
13a) resulted in a significant loss in activity. The introduction of larger substituents to the 5-position
appeared to obliterate all measurable activity with the notable exceptions of the 3-pyridyl
substituent (10c), which exhibited moderate CDK9 activity and selectivity over PI3Kα and FLT3. The
size and planarity of the 5-substituent influences whether it can be accommodated in the back
pocket and, in addition, how readily the compound can access the CDK9 ATP binding site. In the
hinge region CDK9 has a preference for planar compound features and therefore conformations
where the 5-substituent is out of the plane of the pyrazolo[1,5-a]pyrimidine ring will limit access to
the ATP binding site. Thus, the methoxy substituent of 12, the N,N-dimethyl amino substituent of
13a, and the morpholino of 13b are each predicted to have difficulty gaining access to the CDK9 ATP
binding site. Thus, the 3-pyridyl of 10c is at the physical limit of the pocket size, explaining the loss of
CDK9 activity experienced with 10b and 10d. The 5-methyl group of 10a is easily accommodated in
the CDK9 back pocket, however, the CH
3
-π interaction with Phe103 is weaker than the Cl-π
3
2, 33
interaction of 9a
and this likely explains the reduced CDK9 activity. Interestingly, the 5-cyano
substituent of 11 gave a compound equipotent to 9a across the four kinases profiled. The 5-cyano
substituent is accommodated in the back pocket and extends the bicyclic ring π-π interaction with
Phe103 (Figure 2d).
Table 3
Substitutions for chloride in 9a
17

IC50 (nM)
CDK9
1.37
IC50 (µM)
MV4;11
0.003
0.326
0.762
3.12
Compound
R
CDK7
PI3Kα
0.5
FLT3
5.15
32
a
PIK-75 (6)
NT
9
1
1
1
1
1
1
1
1
a
Cl
96
841
271
0a
0b
0c
0d
1
Me
379
8750
>10000
>10000
>10000
1811
NT
902
222
3,4-dimethoxyphenyl >10000
233
3986
874
3-Pyridyl
209
1240
1030
198
7.48
4-Methoxy-3-pyridyl
>10000
43
>10000
82
>10
CN
0.212
1.19
2
OMe
1165
2518
6555
327
512
3a
3b
NMe
2
NT
>10000
>10000
4448
2545
>10
Morpholine
NT
15.1
aNT = Not tested
18

Finally we attempted to replace the metabolically labile hydrazone linking group present in these
compounds (Table 4). Introducing an ethylamino linker (compounds 17a, 18a and 18b) removes the
hydrazone moiety at the expense of some CDK9 activity, though this can be improved through N-
methylation of the sulfonamide. Interestingly, we found that replacement of the linker with a
heterocycle (21a-c) removed all measurable activity in CDK9; however, 21a and 21c both showed
good activity for PI3Kα and are thus selective for PI3Kα over CDK9. The pyrrole 21c also exhibited
inhibitory activity for FLT3 at a similar level as compounds 9a, 9r, 9u, 9v and 11. The reason for the
reduced CDK9 activity of these heterocycle-linked compounds is not immediately clear as modelling
indicates that they can all be accommodated in the CDK9 ATP binding site (Figure 3a). We postulate
that the loss of CDK9 activity arises from their reduced flexibility, which makes it difficult for these
compounds to access the ATP site. Thus, in compounds 21a and 21c, the heterocyclic linker prefers
to be planar with the pyrazolo[1,5-a]pyrimidine ring, resulting in a conformation like that shown in
Figure 3b,c where the sulfonamide substituent blocks access to the CDK9 ATP binding site.
Compound 21b has the additional conformational restriction in that the unsubstituted N of the
pyrazole linker has to be orientated away from the bicycle’s pyrimidine nitrogen (N7α) to avoid
electronic repulsion. The ATP binding sites of PI3Kα and FLT3 are more open (i.e. the Gly-rich loop
generally sits higher than in CDK9) and the heterocyclic linker compounds 21a-c have easier access,
and this likely explains the selectivity profile of these compounds. These data are consistent with the
early work of Waterfield and co-workers who demonstrated that the pyrazole analogue of PIK-75 (6)
1
4
displayed excellent potency for PI3Kα.
Table 4
Linker group substitutions
19

IC50 (nM)
CDK9
IC50 (µM)
MV4;11
Compound R
X
a
CDK7
PI3Kα
FLT3
9
1
1
r
CN
NO
NO
22
3590
2960
33
0.364
0.166
0.374
a
7a
8a
2
1144
265
NT
900
184
516
2
NT
1870
1
8b
1a
CN
203
>10000 >10000 219
0.177
1.90
2
NO
2
2
2
>10000 1680
>10000 NT
>10000 NT
194
361
346
58
2
1b
1c
NO
NO
1070
164
1.28
2
0.123
20

a
NT = Not tested
The ethylamino compounds 18a and 18b represent a significant development in this series. Both
compounds display promising CDK9 affinity (~200nM), and in particular compound 18b possesses
excellent selectivity over PI3Kα and CDK7. Profiling of this compound against kinases known to be
1
8,19
potently inhibited by PIK-75 (6)
). Significantly, compound 18b is devoid of both the problematic hydrazone and aryl nitro moieties
present in PIK-75 (6) and is therefore a more drug-like lead for further optimisation.
indicates that compound 18b shows improved selectivity (Table
5
21

Figure 3.
Modelled CDK9 binding mode (a) and minimum energy conformation (b) and (c) of
the inactive compound 21c. In (b) and (c) CDK9 is depicted as a grey molecular surface and in (c) a
semi-transparent molecular surface for compound 21c is also shown. The views in (b) and (c) differ
by a rotation of -90° about the X-axis. The minimum energy conformation has difficulty gaining
access to the CDK9 ATP binding site and would require the compound to adopt a less favourable
non-planar conformation, as in panel (a), and/or CDK9 to adopt a more open conformation with
greater access to the ATP pocket. In (c) arrows indicate the distance between CDK9 residues Phe30
and Lys151 (5.5 Å), and also the height of the compound (6.4 Å).
Table 5
% Enzyme activity remaining, 18b @ 200 nM
%
Activity
Kinase
CDK1/cyclin A
CDK2/cyclin E
CLK1
90.7
101.1
20.5
102.3
95.1
52.6
34.0
93.7
15.9
c-MET
DNA-PK
DYRK1A
GSK3β
JAK2
PKCε
2.3 Cellular Activity
22

The cellular activity of selected compounds from this work was determined in a proliferation assay
using the leukaemia cell line MV4;11 (Tables 3 and 4). This cell line possesses an activating mutation
34
in FLT3 and thus FLT3 inhibitors potently inhibit the proliferation of these cells. Nonetheless,
compounds from our study show variable activity against this cell line, independent of their FLT3
activity. Thus, for example, while PIK-75 (6) (FLT3 IC50 5 nM) inhibits cell proliferation with an IC50 of
3
nM, 9r (FLT3 IC50 33 nM) is 100-fold less active against cells; the potency of PIK-75 arising from its
1
6
multi-kinase activity. Conversely, potent cellular activity of other compounds with reduced FLT3
activity must result from activities other than FLT3 inhibition. Our previous work has shown that
1
6
MV4;11 AML cells require CDK9 activity for their survival and the significant anti-proliferative
activity of 18b is consistent with its potent inhibition of CDK9. While 200nM 18b inhibits PKC and
3
5
CLK1 by >80% (Table 5), these kinases demonstrate low expression in human AML and so are
unlikely to be key targets. Nonetheless, further studies are required to delineate the role of CDK9
inhibition in the activity of our compound series against leukaemia cell lines and this, along with in
vivo studies, will be the subject of future publications.
3. Conclusions
The development of selective inhibitors of CDK9 represents a novel approach for the treatment of
cancer by targeting the transcriptional machinery. In this work we have demonstrated that the
prototypical PI3Kα inhibitor PIK-75 (6), which is known to also inhibit CDK9, can be used for the
design of novel CDK9 inhibitors with improved selectivity over PI3Kα. Importantly we have shown
that the two major structural liabilities with PIK-75 – the hydrazone and aryl nitro moieties – can be
substituted with more drug-like functionality whilst retaining appreciable CDK9 affinity.
Furthermore, using molecular modelling and compound docking into published crystal structures of
CDK9, we have identified a putative binding mode of the pyrazolo[1,5-a]pyrimidine compounds
which is predictive of their activity. While our current studies define a path for the synthesis and
23

development of novel potent and selective inhibitors of CDK9, future studies assessing their in vivo
pharmacokinetics, pharmacodynamics and toxicity, as well as efficacy using in vivo models of AML,
will be important for determining their therapeutic potential.
4
. Experimental
.1 Chemistry
4
NMR spectra were recorded on a Bruker AV-II DRX 300 NMR spectrometer at 300 K; chemical shifts
are reported in δ ppm using residual solvent as the internal standard. Liquid chromatography mass
spectroscopy (LCMS) was carried out on a Waters ZQ 3100 using reverse phase HPLC (column:
XBridgeTM C18 5 μm 4.6 x 100 mm), Solvent A: Water 0.1% Formic Acid, Solvent B: Acetonitrile 0.1%
Formic Acid, Gradient: 10‐100% B over 10 min, Flow rate: 1.5 mL/min, detection: 100‐600 nm and
ESI in positive mode with source temperature 150 °C. All compounds submitted for biochemical
assay were assessed to have purity ≥95% as measured by HPLC analysis at 254 nm UV absorbance.
High Resolution Mass Spectrometry (HRMS) was conducted on an Agilent Q‐TOF 6200 using positive
mode electrospray ionisation (ESI). Analytical thin‐layer chromatography (TLC) was performed on
Merck silica gel 60 F254 aluminium backed plates and visualised with short wavelength UV (254 nm)
absorbance. Chromatography was performed using either the CombiFlash® Rf purification system
(Teledyne, ISCO, Lincoln, NE, USA) with pre‐packed silica gel columns (particle size 40‐63 m).
Anhydrous solvents were dried using an automated solvent purification system (MBraun SPS,
3
6
Garching, Germany) based upon a technology originally described by Grubbs et al. Reagents were
purchased from a range of chemical suppliers and used without further purification.
Synthesis of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8)
24

5-Chloropyrazolo[1,5-a]pyrimidine (1 g, 6.51 mmol) was dissolved in anhydrous DMF (10 mL) and
cooled to -5 °C. Phosphorous oxychloride (1.82 mL, 19.5 mmol) was added and the reaction was
allowed to warm slowly to RT and stirred for 16 h. The reaction was poured onto ice (~200 g) and the
pH was adjusted to pH 10 with 1 M NaOH. The aqueous suspension was extracted with CH
2
CL (100
2
mL) and the aqueous layer was washed with CH Cl (2 x 50 mL). The combined organic layers were
2
2
washed with water (2 x 70 mL), brine (50 mL), dried (Whatman® PS1 filter paper) and the solvent
was removed in vacuo. The crude product was adsorbed onto silica and purified by automated flash
chromatography on silica gel (0-50% EtOAc – cyclohexane) to yield the title compound (8) (864 mg,
1
7
3%) as a colourless solid. H NMR δ
H
(d
6
-DMSO, 300 MHz) 10.09 (s, 1H), 9.39 (d, J 7.2, 1H), 8.77 (s,
1
H), 7.49 (d, J 7.2, 1H); LC-MS t
R
4.08 min; LC254 99.0%; m/z 182.2/184.1 [M+H].
Preparation of 5-chloropyrazolo[1,5-a]pyrimidin-3-yl sulfonohydrazones
These were made using one of either method A or method B:
General method A: To a solution of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (1 eq) in
anhydrous MeOH (0.1 M) was added NaHCO (5 eq) followed by methylhydrazine sulfate (1.2 eq).
3
After stirring for 2 h at RT the sulfonyl chloride was added (1.3 eq) and the reaction was stirred for a
further hour. The MeOH was removed in vacuo and the orange residue was partitioned between
CH
Cl
2 2
and water. The aqueous phase was washed with CH
2
2
Cl , the organic phases were combined,
dried (Whatman® PS1 filter paper) and the solvent was removed in vacuo. The crude material was
adsorbed onto silica and purified by automated flash chromatography on silica gel (generally 0-50%
EtOAc – cyclohexane) to yield the desired product.
General method B: Using the same procedure as Method A NaHCO was replaced with 2,6-lutidine.
3
Following stirring for up to 16 h the precipitate was isolated by filtration, dried (Whatman® PS1 filter
paper) in vacuo and purified as for Method A.
25

(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-N,2-dimethyl-5-
nitrobenzenesulfonohydrazide (9a)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (168 mg, 0.93 mmol) with 2-
methyl-5-nitrobenzenesulfonyl chloride using method A gave, after chromatography (0-50% EtOAc –
1
cyclohexane) 9a as a yellow solid (123 mg, 32%). H NMR δ
H
(CDCl , 300 MHz) 8.87 (d, J 2.5, 1H), 8.55
3
(d, J 7.3, 1H), 8.37 (s, 1H), 8.28 (dd, J 8.4, 2.5, 1H), 7.92 (s, 1H), 7.49 (d, J 8.4, 1H), 6.85 (d, J 7.2, 1H),
+
3
.42 (s, 3H), 2.85 (s, 3H). LC-MS t 7.28 min; LC254 98.6%; m/z 409.2/411.2 [M+H]; HRMS (ESI ) Calc’d
R
+
for C15
H
13ClN
6
O
4
S [M+H] : 409.0480; found 409.0487.
(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-N-methylbenzenesulfonohydrazide (9b)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with
benzenesulfonyl chloride using method A gave after chromatography (0-50% EtOAc – cyclohexane)
1
9
7
3
b as a yellow solid (118 mg, 77%). H NMR δ
H
(CDCl
3
, 300 MHz) 8.58 (d, J 7.2, 1H), 8.57 (s, 1H), 7.98-
6.75 min; LC254 99.0%; m/z
S [M+H] : 350.0473; found 350.0481.
.93 (m, 3H), 7.63-7.50 (m, 3H), 6.87 (d, J 7.3, 1H), 3.28 (s, 3H). LC-MS t
R
+
+
50.1/352.1 [M+H]; HRMS (ESI ) Calc’d for C14
H
12ClN
5
O
2
(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-N,2-dimethylbenzenesulfonohydrazide
(9c)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with o-
toluenesulfonyl chloride using method A gave after chromatography (0-50% EtOAc – cyclohexane) 9c
1
as a yellow solid (88 mg, 55%). H NMR δ
H
(CDCl
3
, 300 MHz) 8.56 (d, J 7.2, 1H), 8.43 (s, 1H), 8.02 (dd,
26

J 7.9, 1.5, 1H), 7.89 (s, 1H), 7.47 (ddd, J 7.5, 7.4, 1.4, 1H), 7.38-7.29 (m, 2H), 6.85 (d, J 7.2, 1H), 3.38
+
(
s, 3H), 2.75 (s, 3H). LC-MS t
R
7.13 min; LC254 99.0%; m/z 364.2/366.2 [M+H]; HRMS (ESI ) Calc’d for
+
C
15
H14ClN
5
O
2
S [M+H] : 364.0629; found 364.0633.
(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-N-methyl-3-
nitrobenzenesulfonohydrazide (9d)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with 3-
nitrobenzenesulfonyl chloride using method A gave after chromatography (0-50% EtOAc –
1
cyclohexane) 9d as a yellow solid (101 mg, 58%). H NMR δ
H
(CDCl , 300 MHz) 8.80 (dd, J 1.8, 1.8,
3
1
H), 8.57 (d, J 7.2, 1H), 8.52 (s, 1H), 8.43 (ddd, J 8.2, 2.3, 1.1, 1H), 8.31 (ddd, J 7.8, 1.7, 1.1, 1H), 7.97
s, 1H), 7.75 (dd, J 8.2, 7.8, 1H), 6.87 (d, J 7.3, 1H), 3.30 (s, 3H). LC-MS t 6.98 min; LC254 99.0%; m/z
S [M+H] : 395.0324; found 395.0329.
(
R
+
+
395.1/397.1 [M+H]; HRMS (ESI ) Calc’d for C14
H
11ClN
6
O
4
(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-N,3-dimethylbenzenesulfonohydrazide
(9e)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with m-
toluenesulfonyl chloride using method A gave after chromatography (0-50% EtOAc – cyclohexane)
1
9
e as a yellow solid (122 mg, 76%). H NMR δ
H
(CDCl , 300 MHz) 8.58 (d, J 7.2, 1H), 8.58 (s, 1H), 7.94
3
(
s, 1H), 7.76-7.74 (m, 2H), 7.41-7.39 (m, 2H), 6.87 (d, J 7.2, 1H), 3.27 (s, 3H), 2.43 (s, 3H). LC-MS t
R
+
+
7
.12 min; LC254 99.0%; m/z 364.2/366.2 [M+H]; HRMS (ESI ) Calc’d for C15
H
14ClN
5
O
2
S [M+H] :
364.0629; found 364.0633.
27

(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-3-methoxy-N-
methylbenzenesulfonohydrazide (9f)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with 3-
methoxybenzenesulfonyl chloride using method A gave after chromatography (0-50% EtOAc –
1
cyclohexane) 9f as a yellow solid (93 mg, 56%). H NMR δ
H
(CDCl , 300 MHz) 8.57 (s, 1H), 8.56 (d, J
3
7.3, 1H), 7.92 (s, 1H), 7.50 (m, 1H), 7.43-7.37 (m, 2H), 7.09 (ddd, J 8.1, 2.6, 1.0, 1H), 6.85 (d, J 7.2,
+
1
H), 3.83 (s, 3H), 3.25 (s, 3H). LC-MS t 6.92 min; LC254 99.0%; m/z 380.2/382.1 [M+H]; HRMS (ESI )
R
+
Calc’d for C15
H
14ClN
5
O
3
S [M+H] : 380.0579; found 380.0581.
(
E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-N-methyl-3-
trifluoromethoxy)benzenesulfonohydrazide (9g)
(
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with 3-
(trifluoromethoxy)benzenesulfonyl chloride using method A gave after chromatography (0-50%
1
EtOAc – cyclohexane) 9g as a yellow solid (67 mg, 35%). H NMR δ
H
(CDCl , 300 MHz) 8.59 (d, J 7.2,
3
1
H), 8.53 (s, 1H), 7.97 (m, 1H), 7.90 (ddd, J 7.8, 1.6, 1.1, 1H), 7.84 (m, 1H), 7.59 (dd, J 8.2, 8.2, 1H),
.44 (m, 1H), 6.88 (d, J 7.2, 1H), 3.29 (s, 3H). LC-MS t 7.67 min; LC254 99.0%; m/z 434.1/436.1 [M+H];
S [M+H] : 434.0296; found 434.0302.
7
R
+
+
HRMS (ESI ) Calc’d for C15
H
11ClF
3
N
5
O
3
(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-3-cyano-N-
methylbenzenesulfonohydrazide (9h)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with 3-
cyanobenzenesulfonyl chloride using method A gave after chromatography (0-50% EtOAc –
1
cyclohexane) 9h as a yellow solid (96 mg, 58%). H NMR δ
H
(CDCl
3
, 300 MHz) 8.60 (d, J 7.3, 1H), 8.52
28

(s, 1H), 8.26-8.21 (m, 2H), 7.99 (s, 1H), 7.87 (ddd, J 7.8, 1.4, 1.3, 1H), 7.69 (dd, J 7.8, 7.8, 1H), 6.91 (d,
+
J 7.2, 1H), 3.31 (s, 3H). LC-MS t
R
6.68 min; LC254 99.0%; m/z 375.1/377.2 [M+H]; HRMS (ESI ) Calc’d
+
for C15
H
11ClN
6
O
2
S [M+H] : 375.0425; found 375.0436.
(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-3-fluoro-N-
methylbenzenesulfonohydrazide (9i)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with 3-
fluorobenzenesulfonyl chloride using method A gave after chromatography (0-50% EtOAc –
1
cyclohexane) 9i as a yellow solid (82 mg, 51%). H NMR δ
H
(CDCl , 300 MHz) 8.59 (d, J 7.2, 1H), 8.54
3
(s, 1H), 7.95 (s, 1H), 7.76 (ddd, J 7.8, 1.6, 1.0, 1H), 7.67 (ddd, J 8.2, 2.6, 1.7, 1H), 7.52 (ddd, J 8.2, 8.0,
5
9
3
.3, 1H), 7.30 (dddd, J 8.3, 8.3, 2.6, 1.1, 1H), 6.88 (d, J 7.2, 1H), 3.29 (s, 3H). LC-MS t
R
6.13 min; LC254
+
+
9.0%; m/z 368.1/370.2 [M+H]; HRMS (ESI ) Calc’d for C14
H
11ClFN
5
O
2
S [M+H] : 368.0379; found
68.0390.
(E)-N'-((5-Chloropyrazolo[1,5-a]pyrimidin-3-yl)methylene)-N-methyl-4-
nitrobenzenesulfonohydrazide (9j)
Reaction of 5-chloropyrazolo[1,5-a]pyrimidine-3-carbaldehyde (8) (80 mg, 0.44 mmol) with 4-
nitrobenzenesulfonyl chloride using method A gave after chromatography (0-50% EtOAc –
1
cyclohexane) 9j as a yellow solid (59 mg, 34%). H NMR δ
H
(CDCl , 300 MHz) 8.60 (d, J 7.2, 1H), 8.50
3
(
s, 1H), 8.39 (d, J 8.8, 2H), 8.19 (d, J 8.8, 2H), 7.97 (s, 1H), 6.91 (d, J 7.2, 1H), 3.32 (s, 3H). LC-MS t
R
+
+
7
.03 min; LC254 99.0%; m/z 395.2/397.1; HRMS (ESI ) Calc’d for C14
H
11ClN
6
O
4
S [M+H] : 395.0324;
found 395.0330.
29