Cyclin-Dependent Kinase (CDK) Inhibitors: Structure-Activity Relationships and Insights into the CDK-2 Selectivity of 6-Substituted 2-Arylaminopurines

Article abstract of DOI:10.1021/acs.jmedchem.6b01254

Purines and related heterocycles substituted at C-2 with 4′-sulfamoylanilino and at C-6 with a variety of groups have been synthesized with the aim of achieving selectivity of binding to CDK2 over CDK1. 6-Substituents that favor competitive inhibition at

Full text of DOI:10.1021/acs.jmedchem.6b01254

Subscriber access provided by Fudan University
Article
Cyclin-Dependent Kinase (CDK) Inhibitors; Structure-Activity Relationships
and Insights into the CDK-2 Selectivity of 6-Substituted 2-Arylaminopurines
Christopher R Coxon, Elizabeth Anscombe, Suzannah Jane Harnor, Mathew P. Martin,
Benoit Jean-Pierre Carbain, Bernard Thomas Golding, Ian Robert Hardcastle, Lisa K
Harlow, Svitlana Korolchuk, Christopher J. Matheson, David R. Newell, Martin E.M. Noble,
Mangaleswaran Sivaprakasam, Susan J Tudhope, David M Turner, Lan-Zhen Wang,
Stephen R Wedge, Christopher Wong, Roger John Griffin, Jane A. Endicott, and Celine Cano
J. Med. Chem., Just Accepted Manuscript • Publication Date (Web): 22 Dec 2016
Downloaded from http://pubs.acs.org on December 22, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
CyclinꢀDependent Kinase (CDK) Inhibitors; StructureꢀActivity Relationships and Insights
into the CDKꢀ2 Selectivity of 6ꢀSubstituted 2ꢀArylaminopurines
†,‡,%
†,§
‡
∇
Christopher R. Coxon,
Elizabeth Anscombe, Suzannah J. Harnor, Mathew P. Martin, Benoit
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
‡
‡
‡
‡
∇
Carbain, Bernard T. Golding, Ian R. Hardcastle, Lisa K. Harlow, Svitlana Korolchuk,
Christopher J. Matheson, David R. Newell,∇ Martin E. M. Noble,
‡
§,+
Mangaleswaran
Sivaprakasam, Susan J. Tudhope, David M. Turner, Lan Z. Wang, Stephen R. Wedge,∇
‡
∇
‡
∇
‡
‡
ꢀ
,§,+
‡
Christopher Wong, Roger J. Griffin, Jane A. Endicott,* and Céline Cano*
‡
Newcastle Cancer Centre, Northern Institute for Cancer Research, School of Chemistry, Bedson
Building, Newcastle University, Newcastle upon Tyne NE1 7RU UK
§
Department of Biochemistry, University of Oxford, South Parks Road, Oxford, OX1 3QU
∇Newcastle Cancer Centre, Northern Institute for Cancer Research, Paul O’Gorman Building,
Medical School, Framlington Place, Newcastle upon Tyne NE2 4HH UK.
+
Current address: Newcastle Cancer Centre, Northern Institute for Cancer Research, Paul
O’Gorman Building, Medical School, Framlington Place, Newcastle upon Tyne NE2 4HH UK.
%
Current address: School of Pharmacy and Biomolecular Sciences, Liverpool John Moores
University.
ꢀ
Deceased 24 September 2014.
ACS Paragon Plus Environment

Page 3 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Abstract
Purines and related heterocycles substituted at Cꢀ2 with 4’ꢀsulfamoylanilino and at Cꢀ6 with a
variety of groups have been synthesized with the aim of achieving selectivity of binding to CDK2
over CDK1. 6ꢀSubstituents that favour competitive inhibition at the ATP binding site of CDK2
were identified and typically exhibited 10ꢀ80ꢀfold greater inhibition of CDK2 compared to CDK1.
Most impressive was 4ꢀ((6ꢀ([1,1'ꢀbiphenyl]ꢀ3ꢀyl)ꢀ9Hꢀpurinꢀ2ꢀyl)amino) benzenesulfonamide (73)
that exhibited high potency towards CDK2 (IC50 0.044 ꢀM), but was ~ 2000ꢀfold less active
towards CDK1 (IC50 86 ꢀM). This compound is therefore a useful tool for studies of cell cycle
regulation. Crystal structures of inhibitorꢀkinase complexes showed that the inhibitor stabilizes a
glycineꢀrich loop conformation that shapes the ATP ribose binding pocket, and that is preferred in
CDK2, but has not been observed in CDK1. This aspect of the active site may be exploited for the
design of inhibitors that distinguish between CDK1 and CDK2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Key words
CDK / cyclin / Xꢀray crystallography / structureꢀaided inhibitor design / protein kinase
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Introduction
The cyclinꢀdependent kinase (CDK) family of serine/threonine kinases plays an integral role in the
regulation of the eukaryotic cell cycle, as well as having key functions in apoptosis, transcription,
1, 2
differentiation and neuronal function.
Deregulation of the cell cycle, a hallmark of human
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
cancer, is frequently associated with aberrant CDK activity through mechanisms that may include
mutation/overexpression of CDKs, or mutations to genes encoding proteins that directly or
3
ꢀ5
indirectly modulate CDK activity.
CDK2 activation entails association with partner cyclins A
Cip1
Kip1
and E, whereas the endogenous proteins p21
and p27
are inhibitory. Cyclin E overexpression
Kip1
6, 7
and/or p27
suppression are a common feature of many human tumors.
Pharmacological inhibition of CDK family members is a potential therapeutic approach for the
treatment of cancer. In this context, a large number of ATPꢀcompetitive inhibitors have been
5ꢀ9
evaluated.
The clinical development of CDK inhibitors as antitumour agents was originally
hampered by poor kinase selectivity, particularly within the CDK family, and uncertainty as to
3
which CDK constitutes the most appropriate therapeutic target. However, an appreciation of the
roles played by specific CDKs in sustaining signalling through dysregulated pathways that drive
particular cancers is leading to more stringent patient selection and correspondingly improved
efficacy. For example, the highly selective CDK4/6 inhibitor palbociclib has been approved for the
treatment of ERꢀpositive, HER2ꢀnegative breast cancer, while abemaciclib, palbociclib and
ribociclib are in Phase III clinical trials to treat advanced breast cancer in which signalling through
10
the CDK4/6ꢀretinoblastoma axis is critical. Similarly, agents that target a subset of CDKs are
11
being evaluated in a range of other clinically defined segments.
The validity of CDK2 as a cancer therapeutic target was originally called into question both by
CDK2 knockdown experiments in which loss of CDK2 failed to induce cell cycle arrest in a number
12
13, 14
of tumour cell lines, and by mouse knockout experiments where the animals were viable.
ACS Paragon Plus Environment

Page 5 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
However, applying the rationale outlined above, CDK2 inhibitors are also expected to have utility
in settings in which the cancer is addicted to enhanced CDK2 activity or where synthetic lethalities
can be identified. Recent studies employing a chemical genetic approach in which CDK2
expression was maintained, but kinase activity was inhibited, provides compelling evidence that
CDK2 is a valid cancer target. In a panel of human cancer cells transformed with various
oncogenes, highly selective smallꢀmolecule CDK2 inhibition resulted in marked growth inhibition.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
5
CDK2ꢀselective inhibitors may also have applications in combination therapies in appropriate
clinical settings. For example, a recent study demonstrated that a combination of
phosphatidylinositolꢀ3ꢀkinase and CDK2 inhibitors induced apoptosis in malignant glioma
16
xenografts via a syntheticꢀlethal interaction.
Accumulating evidence also implicates CDK2 as a
17
18
19
prospective therapeutic target in BRCAꢀdeficient cancers, neuroblastoma and ovarian cancer.
2
0
These observations, taken together with supportive clinical data,
have led to a resurgence of
21, 22
interest in CDK2 inhibitors as cancer therapeutic agents.
Ideally, such an inhibitor would
discriminate between CDK2 and CDK1 to avoid cell toxicity as a result of inhibiting CDK1 which
is an essential member of the CDK family.
Our previous studies employing a structureꢀlead approach resulted in the development of selective
ATPꢀcompetitive CDK2 inhibitors based on 2ꢀaminoꢀ6ꢀalkoxyꢀ (1; NU2058) and 2ꢀarylaminoꢀ6ꢀ
23, 24
alkoxyꢀ (2) purine derivatives.
These efforts were rewarded by the identification of potent
purineꢀbased inhibitors, exemplified by 6ꢀcyclohexylmethoxyꢀ2ꢀ(4’ꢀsulfamoylanilino)purine (3;
25
NU6102, CDK2 IC50 = 5.0 nM), as well as alkoxypyrimidine inhibitors (e.g. 4, CDK2 IC50 = 0.8
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
6
nM).
Importantly, 3 exhibits selectivity for CDK2 over other CDKꢀfamily members, proving
some 50ꢀfold selective for CDK2 over CDK1 (IC50 = 250 nM), a kinase with close structural
homology to CDK2. In addition, 3 is only weakly active against CDK7 and CDK9 (IC = 4.4 ꢀM
5
0
and 1.1 ꢀM, respectively). These CDKs play a key role in transcriptional regulation through
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
, 27
phosphorylation of RNA polymerase II.
Modulation of protein synthesis by offꢀtarget inhibition
of CDK7/CDK9 has obfuscated previous studies to elucidate the pharmacological effects of
clinically evaluated compounds initially identified as selective CDK2 inhibitors such as seliciclib
3
, 28, 29
(
5), dinaciclib (6) and SNSꢀ032 (7).
Crystal structures of purines 1ꢀ3 in complex with T160ꢀphosphorylated CDK2ꢀcyclin A (CDK2ꢀ
23, 24
cyclin A) provided valuable details of inhibitor interactions within the ATPꢀbinding site.
The
purine heterocycle anchors the inhibitor through a triplet of hydrogen bonds between Nꢀ9, Nꢀ3, and
the 2ꢀamino group, and the backbone carbonyl moiety of Glu81 and amide and carbonyl moieties of
Leu83, respectively, located in the hinge region of the kinase (Figure 2). The greater potency and
selectivity of 3 compared with 2 and the parent 2ꢀaminoꢀ6ꢀcyclohexylmethoxypurine 1, has been
attributed, at least in part, to additional interactions of the 2ꢀsulfanilyl substituent of 3 with the
24, 30
specificity surface of CDK2.
Notably, two hydrogen bond interactions between the
sulfonamide group of 3 and Asp86 facilitate optimal hydrophobic packing of the arylamino ring.
Extensive structureꢀactivity relationship studies (SARs) at the purine 2ꢀposition of 3 have further
25
established the importance of these inhibitorꢀkinase interactions.
By contrast, although the 6ꢀ
ACS Paragon Plus Environment

Page 7 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
cyclohexylmethyl group of 3 occupies the riboseꢀbinding pocket, the precise nature of the
interactions made by substituents at the purine 6ꢀposition with the CDK2 active site remains
uncertain, notwithstanding that a large number of derivatives have previously been evaluated
31
empirically.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In this paper, we report the results of studies to characterise interactions between substituents at the
purine 6ꢀposition and the CDK2 ATPꢀbinding site. The effect upon biological activity of modifying
the core purine heterocycle of 3 is also reported. These studies have resulted in the identification of
4
ꢀ((6ꢀ([1,1'ꢀbiphenyl]ꢀ3ꢀyl)ꢀ9Hꢀpurinꢀ2ꢀyl)amino) benzenesulfonamide 73 as a potent and selective
CDK2 inhibitor (IC50 = 44 nM), exhibiting some 2,000ꢀfoldꢀselectivity over CDK1 (IC50 = 86 ꢀM).
Chemistry
With a view to establishing the overall contribution of the 6ꢀcyclohexylmethoxy substituent of 1
and 3 to inhibitor binding, the simple 6ꢀunsubstituted purines 8, 12 and 13 were required. 2ꢀ
aminopurine (8) was commercially available, while the 2ꢀphenylaminopurine (12) and 2ꢀ
sulfanilylpurine (13) derivatives were synthesised as outlined in Scheme 1. Thus, selective
reductive dehalogenation of 6ꢀchloroꢀ2ꢀfluoropurine (10), prepared from 2ꢀaminoꢀ6ꢀchloropurine
32
(
9) following a literature procedure, afforded 2ꢀfluoropurine (11). Treatment of 11 with aniline or
4ꢀaminobenzenesulfonamide (sulfanilamide) in 2,2,2ꢀtrifluoroethanol (TFE) with catalysis by
3
3
trifluoroacetic acid (TFA) as described previously, gave the respective 2ꢀarylaminopurines 12 and
3 in good yield. The required 2ꢀarylaminoguanine derivatives 15 and 16 were also readily
accessible from commercially available 2ꢀbromohypoxanthine (14) under similar conditions. The 6ꢀ
1
2
3,
alkoxypurine derivatives (30ꢀ35) were synthesised utilising previously optimised methodology.
2
5, 31
Briefly, introduction of the 6ꢀalkoxy substituent was achieved either by direct reaction of the
appropriate sodium alkoxide with 2ꢀaminoꢀ6ꢀchloropurine (9) employing the corresponding alcohol
34
as solvent, or, where necessary, following conversion into the ‘DABCOꢀpurine’ intermediate (17).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Elaboration of the 6ꢀalkoxyꢀ2ꢀaminopurines (18ꢀ23) into the corresponding 2ꢀfluoropurines (24ꢀ29)
under BalzꢀSchiemann conditions, enabled subsequent introduction of the 2ꢀ(4ꢀ
aminobenzenesulfonamide) group using the TFAꢀTFE procedure, to furnish the target purines (30ꢀ
3
5) in good overall yields (Scheme 1).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Reagents and conditions – (i) aq. HBF
MeOH, reflux; (iii) ArNH , TFA, TFE, 90 ºC; (iv) POCl
rac)ꢀcamphorsulfonic acid, EtOAc, 75 °C; (vi) RꢀCCH, PdCl
triisopropylꢀ[1,1'ꢀbiphenyl]ꢀ2ꢀyl)phosphine, Cs CO , MeCN; (vii) TFA, iꢀPrOH, H
DABCO, DMSO, 25 °C; (ix) For 18, 19, 22: ROH, Na, reflux, For 20, 21, 23: ROH, NaH, DMSO,
5 ºC. For definition of R groups see Table 1.
4
, NaNO
, PhNMe
(MeCN)
2
, ꢀ15ꢀ0 °C; (ii) Pd(OH)
, reflux; (v) 3,4ꢀdihydropyran,
, dicyclohexylꢀ(2',4',6'ꢀ
O, reflux; (viii)
2 2 4
, HCO NH ,
2
3
2
(
2
2
2
3
2
2
The 6ꢀethynylpurine derivatives (41 and 42) were synthesised from the THPꢀprotected 6ꢀchloroꢀ2ꢀ
35
fluoropurine (36)
via a Sonogashira alkynylation approach (Scheme 2). Coupling of
triisopropylacetylene with 36, employing Pd(PPh ) Cl /CuI, afforded the protected 6ꢀethynylpurine
3
2
2
9
(37) in excellent yield, and removal of the N ꢀTHP group to give 38 and introduction of the 2ꢀ
ACS Paragon Plus Environment

Page 9 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
arylamino substituent, proceeded in the manner described above to give 39 and 40. Removal of the
TIPS group to furnish the target 2ꢀarylaminoꢀ6ꢀethynylpurines 41 and 42 was achieved under
standard conditions. The 6ꢀpropynylꢀ and 6ꢀphenylethynylꢀpurine derivatives 48 and 49 were
accessible from the guanine derivative 15 by sequential chlorination at the purine 6ꢀposition (44),
THP protection (45), introduction of the appropriately substituted ethynyl group under Sonogashira
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
9
conditions (46, 47) and final N ꢀdeprotection to furnish the target purines 48 and 49 (Scheme 1).
The 6ꢀethylpurine derivative (52) was readily obtained by reduction of the 6ꢀethynyl group of the
9
THP derivative 43 with Lindlar’s catalystꢀquinoline to afford 50, with N ꢀdeprotection (51) and
final introduction of the 2ꢀ(4ꢀaminobenzenesulfonamide) group giving 52 in modest overall yield.
Scheme 2. Reagents and conditions – (i) 3,4ꢀdihydropyran, (rac)ꢀcamphorsulfonic acid, EtOAc,
75 °C; (ii) (iꢀPr) SiCCH, Pd(PPh )Cl , CuI, Et N, THF, 25 ºC; (iii) TFA, iꢀPrOH, H O, reflux; (iv)
PhNH or 4ꢀNH SO NH , TFA, TFE, 90 ºC; (v) (nꢀBu) NF, THF, 25 ºC; (vi) Lindlar’s
catalyst, quinoline, H , EtOAc, RT, 2 h; (vii) RB(OH) or RBF K, Pd(OAc) , dicyclohexyl(2',6'ꢀ
3
3
2
3
2
2
2
C
6
H
4
2
2
4
2
2
3
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
dimethoxybiphenylꢀ2ꢀyl)phosphine, K PO , PhMe, H O, 100 ºC. For definition of R groups see
3
4
2
Table 1.
Efforts to synthesise the 6ꢀcyclopropylpurine (55) employing SuzukiꢀMiyaura conditions were
unsuccessful, likely due to competing protodeboronation of the cyclopropylboronic acid under the
reaction conditions employed. This problem was successfully addressed as reported in the literature,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
36
by converting the cyclopropylboronic acid into the corresponding potassium trifluoroborate,
whereby coupling to the THPꢀprotected 6ꢀchloroꢀ2ꢀfluoropurine (36) occurred to furnish 53 in
acceptable yield. Subsequent transformation into the target 6ꢀcyclopropylpurine (55) was achieved
readily. The introduction of aryl substituents at the purine 6ꢀposition was achieved using Suzukiꢀ
Miyaura crossꢀcoupling chemistry with 6ꢀchloroꢀ2ꢀfluoropurine derivative 36 (Scheme 2).
Optimisation of the reagents and reaction conditions employed for the crossꢀcoupling step, enabled
the preparation of a defined series of 2ꢀfluoroꢀ6ꢀarylpurine derivatives (56ꢀ62) in excellent yields,
and conversion into the requisite 6ꢀarylꢀ2ꢀ(4ꢀsulfonamidophenyl)aminopurines (70ꢀ76) proceeded
25, 33
smoothly under the previously described conditions.
ACS Paragon Plus Environment

Page 11 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 3. Reagents and conditions – (i) PMBꢀCl, K
2 3
CO , DMF, 80 ºC; (ii) NaH,
cyclohexylmethanol, DMSO, 25 ºC; (iii) TFA, 80 ºC; (iv) CH I , isoꢀamyl nitrite, CuI, THF, 80 ºC;
2 2
(v) sulfanilamide, Pd
2
(dba)
3
, XPhos, K
2
CO
3 2
, MeCN, 80 ºC; (vi) LiCl, isoꢀamyl nitrite, SOCl , DMA
or THF, 0 ºC ꢀ 25 ºC; (vii) ArNH
THF, 25 ºC.
2
, TFA, TFE, 90 ºC; (viii) SEMꢀCl, NaH, MeCN, 0 ºC; (ix) TBAF,
The synthesis of prospective CDK2 inhibitors encompassing alternative heterocyclic systems was
undertaken as summarised in Scheme 3. In each case, the parent chloroheterocycles (77, 84, 85, 92,
37ꢀ41
9
9) were synthesised by adapting literature procedures
and the cyclohexylmethyl substituent
9
introduced by treatment with the corresponding alkoxide, with N protected where necessary, as
4
2
described previously. Iodination (81)
or chlorination (88, 89, 96) of the respective 2ꢀ
aminoheterocycles under Sandmeyer conditions facilitated introduction of the 2ꢀarylamino group
under the standard TFAꢀTFE conditions (88, 89), or employing a Buchwald amination procedure
(81, 96), followed by deprotection as necessary to afford the target compounds 83, 90, 91, and 98.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The imidazo[1,2ꢀa]pyrimidine derivatives (101, 102) were prepared from 5,7ꢀdichloroimidazo[1,2ꢀ
4
3, 44
a]pyrimidine (99)
by regioselective 5ꢀalkoxylation to give 100, followed by final introduction
of the arylamino group at the 7ꢀposition under TFAꢀTFE conditions. Given the possibility of
alkoxide attack at the 7ꢀposition of 99 to furnish the alternative regioisomer, the structure of 102
was unambiguously confirmed by Xꢀray crystallography (Supporting Information).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Results and Discussion
The chemical structures, CDKꢀinhibitory activity, and in vitro antitumor activity for the purine
derivatives and alternative heterocycles are summarised in Tables 1ꢀ3, with the values for
compounds 1ꢀ3 included for comparative purposes.
StructureꢀActivity Relationships – The importance of a substituent at the purine 6ꢀposition was
reaffirmed by the simple 2ꢀaminoꢀ and 2ꢀphenylaminoꢀpurines (8 and 12), where a dramatic
reduction in potency was observed compared with the parent 6ꢀcyclohexylmethoxy derivatives 1
and 2, respectively. A similar effect was also evident for the sulfanilylpurine 13, which was some
3
00ꢀfold less active than 3. These results are consistent with the putative interaction of this inhibitor
class with the ATPꢀbinding site of CDK2, which requires that the 6ꢀsubstituent occupies a lipophilic
pocket close to the ribose binding site, thereby orienting the purine to make the triplet of hydrogen
24
bonds with the hinge region. The lowꢀmicromolar potency of the 2ꢀsulfanilyl derivative 13 likely
reflects the contribution to binding arising through additional interactions between the
arylsulfonamide group of 13 with the CDK2 surface adjacent to the ATP binding site on the Cꢀ
terminal lobe (termed the “specificity surface”), despite the absence of a 6ꢀsubstituent. Perhaps not
surprisingly, the guanine derivatives 15 and 16 were only weakly active, attributable to
unfavourable interactions between the 6ꢀoxo functionality and the riboseꢀbinding pocket. These
derivatives also exhibited the very poor aqueous solubility characteristic of many guanine
derivatives.
ACS Paragon Plus Environment

Page 13 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Replacement of the 6ꢀcyclohexylmethoxy group of 3 by smaller alkoxy substituents proved
31
informative, and was broadly consistent with our earlier investigations with derivatives of 1. In
6
that study, 2ꢀaminoꢀ6ꢀmethoxypurine (O ꢀmethylguanine) demonstrated negligible activity (CDK2;
IC > 100 ꢀM), and the relatively poor chemical stability of this purine militated against preparing
5
0
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
the corresponding 2ꢀsulfanilyl analogue. The 6ꢀethoxypurine 30 exhibited high CDK2ꢀinhibitory
activity (IC = 26 nM), with an approximately threeꢀfold increase in potency being observed for
5
0
the nꢀpropoxyꢀ (31; IC50 = 8 nM) and isoꢀpropoxy (32; IC50 = 10 nM) derivatives. Potency was
improved further by increasing the size of the 6ꢀalkoxy group, as demonstrated by the activity of the
isomeric butoxy analogues 33 and 34 with IC50 values of 3 nM and 1 nM, respectively. The
relationship between the bulk of the 6ꢀalkoxy group and the CDK2ꢀinhibitory activity generally
parallels that previously observed for the corresponding 2ꢀaminoꢀ6ꢀalkoxypurines, albeit that the
sulfanilylpurines 30ꢀ34 are several orders of magnitude more potent. For example, 34 (IC50 = 1 nM)
is approximately 50,000ꢀfold more potent than the corresponding 2ꢀaminoꢀ6ꢀsecꢀbutoxypurine
31
(CDK2; IC50 = 49 ± 7 ꢀM), again corroborating the crucial binding contribution made by the
6
arylsulfonamide function. A propargyloxy group at the purine O ꢀposition conferred activity
comparable to the ethoxy derivative (compare 35 and 30), in keeping with the trend reported
3
1
previously for the 2ꢀaminoꢀ6ꢀalkoxypurine series.
The introduction of an ethyl substituent at the purine 6ꢀposition (52) reduced CDK2ꢀinhibitory
activity some 10ꢀfold compared with the 6ꢀethoxypurine (30), and very modest potency was also
observed for the 6ꢀethynyl derivatives 41 and 42. Although 6ꢀalkoxy and 6ꢀalkyl groups differ with
regard to both their electronic and steric character, as well as lipophilicity, the high potency of the
6ꢀtriisopropylsilyethynylpurine (40; IC50 = 17 nM) shows that steric bulk, as well as electronic
factors, is an important feature for binding in the ribose pocket of CDK2. The modest improvement
in potency observed for the propꢀ1ꢀynyl (48) and phenylethynyl (49) derivatives compared with the
parent 6ꢀethynylpurine 41, also suggests that the overall shape of the 6ꢀsubstituent is an important
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
factor for occupancy of the riboseꢀbinding pocket, supported by the activity of the 6ꢀ
cyclopropylpurine 55 (IC50 = 19 nM). These data also suggest that the oxygen atom of the 6ꢀ
alkoxypurines, which would act as a weak hydrogen bond acceptor, does not make a significant
contribution to binding affinity.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The potency of the 6ꢀphenylpurine (70; IC50 = 24 nM) prompted further elaboration, with
comparable activity residing in the 3ꢀmethoxyphenyl (71), 4ꢀmethoxyphenyl (72), and 6ꢀ(3ꢀ
phenylphenyl)purine (73) derivatives. By contrast, the piperonyl derivative (74) was some 20ꢀfold
less potent than 70, and the introduction of still larger groups was detrimental, as evident from the
very weak CDK2ꢀinhibitory activity of the dibenzofuranꢀ1ꢀyl (75) and thianthrenꢀ1ꢀyl (76)
analogues. The bulkier bicyclic and tricyclic heteroaromatic rings of these purines are presumably
poorly accommodated within the riboseꢀbinding pocket. It is of interest to note that the CDK2ꢀ
inhibitory activity of the 6ꢀunsubstituted derivative 13 is equipotent with 75 and superior to that of
7
6, implying that the 6ꢀsubstituents of these purines make a negligible binding contribution.
Table 1. Chemical structures and CDK2ꢀinhibitory activity of purine derivatives
1
2
No.
Structure
R
R
CDK2 Inhibition
a
IC (ꢀM) or Inhibition
5
0
(
%) at 100 ꢀM
b
1
A
−
17 ± 2
ACS Paragon Plus Environment

Page 15 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
3
B
B
H
0.97 ± 0.03
SO
NH
2
0.005 ± 0.001
2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
c
8
A
B
B
B
B
B
B
B
H
H
−
(4 ± 3%)
12
H
61 ± 0.1
1.5 ± 0.1
1
3
5
H
SO
2
NH
2
d
d
1
OH (C=O)
OH (C=O)
OEt
H
(34 ± 1%)
16
SO
SO
SO
SO
2
2
2
NH
2
2
2
(52 ± 4%)
30
31
32
NH
NH
NH
0.026 ± 0.001
0.008 ± 0.002
0.01 ± 0.001
2
2
33
34
B
B
SO
SO
2
2
NH
2
2
0.003 ± 0.0003
0.001 ± 0.0005
NH
(
rac)
35
40
B
SO
SO
2
2
NH
2
2
0.019 ± 0.001
B
B
NH
0.017 ± 0.001
21 ± 9
41
H
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
4
4
5
5
7
2
8
9
2
5
0
B
B
B
B
B
B
B
SO
2
NH
2
0.83 ± 0.03
6.4 ± 0.4
H
H
8.9 ± 0.7
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Et
SO
SO
SO
SO
2
2
NH
2
0.22 ± 0.003
0.019 ± 0.001
0.024 ± 0.002
0.039 ± 0.013
NH
NH
NH
2
Ph
2
2
2
2
71
72
73
74
75
B
B
B
B
SO
SO
SO
SO
2
2
2
2
NH
NH
NH
NH
2
2
2
2
0.052 ± 0.004
0.044 ± 0.012
0.58 ± 0.02
1.4 ± 0.2
Ph
76
B
SO
2
NH
2
6.7 ± 0.5
a
20, 23 b
IC values were determined in accordance with previously described methods.
Data shown
50
c
are the mean of at least two independent experiments ± standard deviation. % inhibition values are
d
in brackets. The oxoꢀtautomers predominate.
ACS Paragon Plus Environment

Page 17 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
With a view to assessing selectivity for CDK2, selected purine derivatives were evaluated against a
panel of CDKs (Table 2). As expected, removal of the 6ꢀcyclohexylmethyl substituent of 3, which
makes a number of interactions with the conserved ATP ribose binding site, to furnish 13
compromised potency against all the CDKs evaluated, although the effect was less pronounced
against CDK2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The 6ꢀalkoxypurines 34 and 35 exhibited good selectivity for CDK2 over the closely related CDK1
(80ꢀfold and 50ꢀfold, respectively), with 34 retaining selectivity over CDKs 4, 7 and 9 comparable
with 3. A similar profile was observed for the 6ꢀphenylpurine 70, which was some 30ꢀfold selective
for CDK2 over CDK1. By contrast, the 6ꢀ([1,1'ꢀbiphenyl]ꢀ3ꢀyl)purine derivative 73 exhibited a very
interesting CDK selectivity profile, proving some 2,000ꢀfold selective for CDK2 over CDK1, whilst
exhibiting only weak inhibitory activity against CDKs 4, 7, and 9. To the best of our knowledge this
level of selectivity for CDK2 over CDK1 is unprecedented.
Table 2. CDK selectivity and cellular activity for selected compounds
a
Compd.
3
R
IC50 (ꢁM) or % inhibition at 100 µM
CDK1/B CDK2/A CDK4/D CDK7/H CDK9/T
0
.25 ±
0.005 ±
0.001
1.45 ±
0.70
4.40 ±
0.90
1.07 ±
0.02
b
0
.05
1
3
(12 ±
H
15.8 ± 1.2 1.5 ± 0.09
75 ± 3
38 ± 11
c
4
%)
34
0
.097 ±
0.0012 ±
0.0005
1.30 ±
0.30
0.57 ±
0.05
2
.8 ± 0.1
0
.040
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3
5
0
0
.019 ±
0
.94 ± 0.1
8.8 ± 1.2
3.8 ± 1.4
−
−
0.001
7
0.73 ±
0.024 ±
0.002
0.57 ±
0.04
Ph
6.8 ± 0.4
0.003
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
.044 ±
.012
7
3
86 ± 4
(26 ± 2%) (28 ± 6%)
25 ± 8
0
a
20, 23 b
IC values were determined in accordance with previously described methods.
Data shown
50
c
are the mean of at least two independent experiments ± standard deviation. % inhibition values are
in brackets.
Modification of the core purine heterocycle of 1 and 3 afforded further interesting results (Table 3).
A comparison of 2ꢀaminoꢀ6ꢀcyclohexylmethylpurine 1 with the alternative heterocycles 86, 87 and
93 revealed that although these compounds exhibited comparable potency against CDK2, activity
against other CDK family members was markedly attenuated. The very weak activity of the
imidazopyrimidine derivatives 101 and 102 is not surprising given the absence of the requisite
donorꢀacceptorꢀdonor motif, and presumably reflects the combined influence arising from the
1
7
absence of the N and N functions found in the corresponding purine derivatives 1 and 3. The
1
modest CDK inhibition profile of imidazopyridine 83 indicates that removal of the N nitrogen is
detrimental to CDKꢀinhibitory activity. However, with the exception of the imidazopyridine (83)
and imidazopyrimidine (102) derivatives, modification of the core purine heterocycle of 3 was
generally tolerated without a marked loss of CDK2ꢀinhibitory activity. Thus, triazolopyrimidine
(90) analogue is 2.5ꢀfold more potent than 3, while pyrazolopyrimidine (91) and pyrrolopyrimidine
(
98) analogues proved only some 4ꢀ5ꢀfold less potent than 3 against CDK2, and retained good
selectivity over all other CDK family members examined. Notably, derivatives 91 and 98 were
approximately 100ꢀfold selective for CDK2 over CDK1.
ACS Paragon Plus Environment

Page 19 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
Table 3. Chemical structures and CDKꢀinhibitory activity of other heterocyclic derivatives
CDK Inhibitory Activity
a
No.
Structure
R
IC (ꢀM) or Inhibition at 10 ꢀM (%)
5
0
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
CDK1
CDK2
17 ± 2
CDK4
CDK5
CDK7
CDK9
b
c
1
3
H
26 ± 4
(33 ± 9%)
97 ± 3
(44 ± 6%)
(42 ±
1
1%)
0.25 ±
0.005 ±
0.001
1.5 ± 0.7
0.48 ±
0.07
4.4 ± 0.9
1.1 ± 0.2
0.05
8
0
H
H
ꢀ
28.4 ± 0.8
0.29 ±
ꢀ
ꢀ
ꢀ
ꢀ
83
11 ± 2
(21 ± 1%)
0.75 ±
23 ± 1%
(37 ± 2%) (37 ± 4%) (42 ± 5%)
(13 ± 1%) (13 ± 6%) (16 ± 8%)
0.01
8
6
18 ± 1
(6.2 ±
3
.6%)
90
0.002 ±
0.0001
9.6 ± 0.5
3.3 ± 0.2
6.8 ± 0.7
0.75 ±
0.03
0
.06
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
Table 3 (continued). Chemical structures and CDKꢀinhibitory activity of other heterocyclic derivatives
CDK Inhibitory Activity
a
No.
Structure
R
IC (ꢀM) or Inhibition at 10 ꢀM (%)
5
0
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
CDK1
CDK2
CDK4
CDK5
CDK7
CDK9
8
7
1
H
(32 ± 1%)
1.6 ± 0.2
16 ± 4
(0%)
(14 ± 9%) (16 ± 7%) (16 ± 2%)
9
0.018 ±
.001
2.8 ± 0.3
5.9 ± 0.1
(12 ± 1%)
4.4 ± 0.6
7.0 ± 0.3
4.6 ± 0.2
(0%)
0
9
3
8
H
H
(11 ± 3%)
2.7 ± 0.2
(41 ±
(5.6 ±
7.9%)
(5.1 ±
5.5%)
d
10%)
9
0.026 ±
.001
1.9 ± 0.1
4.1 ± 0.2
1.0 ± 0.2
0
101
ꢀ
ꢀ
(10 ± 6%)
(36 ± 3%)
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
102
2 2
SO NH
a
20, 23 b
IC50 values were determined in accordance with previously described methods.
Data shown are the mean of at least two independent
c
d
experiments ± standard deviation. % inhibition values are in brackets. Value determined at 100 ꢀM.
ACS Paragon Plus Environment

Page 21 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Cellular studies
The ability of compounds to inhibit cellular proliferation was examined in a 5ꢀday growth assay
using five histologically distinct retinoblastoma protein (Rb)ꢀproficient human tumour lines (A375
melanoma, Caluꢀ6 lung carcinoma, MDAꢀMBꢀ231 breast adenocarcinoma, SJSA1 osteosarcoma
and SKUTꢀ1B uterine corpus leiomyosarcoma). Whilst compound 73 demonstrated appreciable
nanomolar potency versus CDK2 in isolated kinase assays, the maximal concentration tested in
cellular assays (30 ꢀM) had no or limited effects on the growth of cells, thereby preventing the
concentration that induced a 50% growth inhibitory (GI ) effect to be determined (data not shown).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
0
In contrast, compounds 91 and 98, which were only 2.4 and 1.7 fold more potent respectively in
CDK2 kinase assays, doseꢀdependently inhibited the growth of each tumour cell line with GI50
values ranging from 1 – 22 ꢀM. The cellular activity of these compounds compared favourably to
that of compound 3.
45
Figure 1. Activity of compounds 3, 91 and 98 against tumour cell growth. Rbꢀproficient
human tumour cell lines were incubated for 120h with compounds and cellular protein determined
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
by SRB assay. Doseꢀresponse curves and GI bar charts represent the mean ± SEM from 3 or 4
50
independent experiments.
Structure determinationꢀ Compounds within this series show greater activity towards CDK2 than
CDK1, generally varying between 10 and 80ꢀfold (Table 2). However, 73 exhibits exceptional
discrimination being a circa 2000ꢀfold more potent inhibitor of CDK2. In order to confirm the
binding mode and to identify potential interactions that might explain the selectivity of 73 for
CDK2 over CDK1, 73 and 3 were coꢀcrystallised with CDK2ꢀcyclin A and CDK1ꢀcyclin Bꢀ cyclinꢀ
dependent kinases regulatory subunit 2 (CKS2) respectively. The poor potency of 73 towards
CDK1 appears to preclude determination of a CDK1ꢀcyclin BꢀCKS2ꢀ73 coꢀcomplex structure, as
despite repeated attempts we were unable to visualise the compound in the active site. The statistics
for the datasets and for the crystallographic refinements are presented in Table 4. Compound 73
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
4
emulates the interactions made by 3 within the CDK2 active site (Figure 2A). Indeed, the purine
and aniline rings of each inhibitor superimpose very well. The purine rings occupy the CDK2 ATP
binding site and each makes a triplet of hydrogen bonds with the backbone carbonyl of Glu81 and
the amide NH and carbonyl moieties of Leu83 within the hinge sequence (Figure 2A). The
2
respective anilino rings project out of the CDK2 catalytic site so that the sulfonamide NH group
and one of the oxygens in each make hydrogen bonds with the side chain and backbone nitrogen of
Asp86, respectively.
6
The structures of 73 and 3 differ in the substitution present at the purine Cꢀ6 position. The O ꢀ
cyclohexylmethyl substituent of 3 occupies the ATP ribose binding pocket and is complementary in
shape and forms favourable hydrophobic interactions with an apolar pocket created by the
24
conformation of the CDK2 glycineꢀrich loop (residues 9ꢀ19).
This loop adopts an identical
backbone structure when bound to 73, so that the purineꢀproximal phenyl ring emulates the position
of the cyclohexylmethyl substitutent of 3. The distal phenyl ring substituted at the meta position can
ACS Paragon Plus Environment

Page 23 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
then be comfortably accommodated as it twists towards the aniline moiety (Figure 2A). In contrast,
para substitutions of the purineꢀproximal phenyl ring exemplified by 72, 74, 75 and 76 (Table 1)
would be directed out of the CDK2 active site towards the tip of the glycineꢀrich loop. Smaller
substitutions, for example a methoxy in 72 can be accommodated but larger ring systems as
exemplified by 74 and 76 lead to considerable drops in potency suggesting they may sterically clash
with the CDK2 structure in this region (Table 1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To probe further the binding mode of the series, CDK1ꢀcyclin BꢀCKS2 was coꢀcrystallised with 3
(Figure 2B). This structure shows that the purine backbone emulates the interactions made by this
6
inhibitor within the CDK2 binding site and that the O ꢀcyclohexylmethyl substituent occupies the
46
ribose binding pocket. As previously reported, inhibitor binding within the ATP binding site is
accompanied by minor remodelling of the CDK1ꢀcyclin B into a conformation compatible with
catalysis.
All of the sidechains contacted by 73 in the coꢀcomplex with CDK2ꢀcyclin A are conserved in
CDK1, so that selectivity must derive from indirect readout of remote sequence differences. One
mechanism by which such differences can impact inhibitor binding is where they shape the
conformational energy landscape, permitting conformations in one kinase that are precluded in
another. As described above, the glycineꢀrich loop makes a significant contribution to the catalytic
cleft, shaping the binding site that accommodates the purine Cꢀ6 substituents. We note that the
glycineꢀrich loop in the CDK2ꢀcyclin Aꢀ73 structure adopts a conformation in which Tyr15 is
folded into the active site, contacting residues of the Cꢀhelix (Figure 2C). This position for the
sidechain of Tyr15 is seen in a number of other CDK2ꢀcyclin Aꢀinhibitor complexes, suggesting
that it reflects a preferred conformation of CDK2. However, in available structures of apo CDK1ꢀ
46
cyclin B, and CDK1ꢀcyclin bound to compound 23 or 3 (this paper), a comparable location for
the side chain of Tyr15 has not been observed (Figure 2C). We speculate, therefore, that 73 binds
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
more tightly to CDK2 than to CDK1 because in doing so, it stabilises a glycineꢀrich loop
conformation that is preferred in CDK2 but not in CDK1.
Table 4. Xꢀray data collection and refinement statistics
1
CDK2ꢀcyclin Aꢀ73
CDK1ꢀcyclin Bꢀ
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
CKS2ꢀ3
Data collection
Space group
P2
1
2
1
2
1
P1
Unit cell (Å)
a=73.5, b=132.1,
c=149.2
a=65.0, b=67.8,
c=85.1, α=103.9,
β=90.9, γ=90.4
65.76ꢀ2.06
Resolution (Å)
66.03ꢀ2.97
(3.13ꢀ2.97)
226,370 (7244)
30,773 (1098)
0.086
(highest resolution shell)
Total observations
(2.10ꢀ2.06)
169709 (8389)
84841 (4161)
0.092
Unique
a
R
merge
Mean I/σ(I)
Completeness %
Refinement
Total number of atoms
protein
6.7 (1.6)
6.2 (1.2)
100 (99.7)
97.3 (95.5)
8,927
64
20,449
100
other
waters
18
291
R (highest resolution
shell)
0.213
0.198
Rfree (highest resolution
shell)
0.266
0.254
Rmsd bonds (Å)
0.0122
1.770
0.0217
2.233
o
Rmsd angles ( )
1
The structures have been deposited in the PDB with accession codes 5LQE (CDK2ꢀcyclin Aꢀ73)
and 5LQF (CDK1ꢀcyclin BꢀCKS2ꢀ3).
ACS Paragon Plus Environment

Page 25 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A
B
F82
L83
E81
F80
K33
Y15
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
D145
H84
73
Q85
D86
Q131
E81
F82
L83
F80
K33
Y15
S84
M85
D145
3
D86
Q132
C
Gly-rich loop
Tyr15
Gly-rich loop
Tyr15
CDK2/A
CDK1/B
Figure 2. Structures of CDK2ꢀcyclin Aꢀ73 and CDK1ꢀcyclin BꢀCKS2ꢀ3. (A) Surface
representation of CDK2 (ice blue) bound to cyclin A (coral) in complex with 73 (yellow). The inset
shows the interactions of 73 within the ATP site of CDK2ꢀcyclin A. (B) Surface representation of
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
CDK1 (dark cyan) bound to cyclin B (pale crimson) and CKS2 (lemon) in complex with 3 (green).
The inset shows the interactions of 3 within the ATP site of CDK1ꢀcyclin BꢀCKS2. In (A) and (B)
the protein backbone is rendered in ribbon representation and selected CDK2 and CDK1 residues
are drawn respectively with carbon atoms coloured ice blue and dark cyan. The hydrogen bonds are
shown as black dotted lines. (C) The structural conformations of the CDK2ꢀcyclin A and CDK1ꢀ
cyclin BꢀCKS2 glycineꢀrich loops to illustrate the alternative poses of catalytic residue Tyr15. The
left hand side panel compares CDK2ꢀcyclin A in complex with 73 (ice blue) to other inhibitor
bound CDK2ꢀcyclin A complexes (PDB entries 4EOS (green), 3TNW (coral), and 3MY5 (lilac)).
The right hand panel overlays CDK1ꢀcyclin BꢀCKS2 bound to 3 (ice blue) with apo CDK1ꢀcyclin
BꢀCKS2 (PDB entry 4YC3, lilac) and CDK1ꢀcyclin BꢀCKS2 in complex with compound 23 (PDB
entry 5HQ0, green). The conformations of the loop when CDK1 is bound to 3 or Compound 23
cannot be distinguished. Glyꢀrich loop, glycineꢀrich loop.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Conclusions
Novel 6ꢀsubstituted 2ꢀ(4’ꢀsulfamoylanilino)purines have been designed as competitive inhibitors
acting at the ATP binding site of CDK2 with particular attention being given to abrogating activity
against CDK1. A variety of substituents were explored, either attached directly to Cꢀ6 or via an
oxygen link, in the context of their possible interaction with a lipophilic pocket close to the ATP
ribose binding site. The relationship between the size of a 6ꢀalkoxy group and CDK2ꢀinhibitory
activity was found to parallel that previously observed for the corresponding 2ꢀaminoꢀ6ꢀ
alkoxypurines. In general, the new compounds were significantly less potent (typically 10ꢀ80×)
against CDK1 than CDK2. Most impressive was 4ꢀ((6ꢀ([1,1'ꢀbiphenyl]ꢀ3ꢀyl)ꢀ9Hꢀpurinꢀ2ꢀyl)amino)
benzenesulfonamide that was ~ 2000ꢀfold less active towards CDK1 (IC50 86 ꢀM), whilst retaining
high potency against CDK2 (0.044 ꢀM). Compounds substituted with relatively large
conformationally constrained groups, e.g. bicyclic and tricyclic aromatic systems, showed greatly
reduced inhibitory activity, indicating poorly accommodation of these substituents in the lipophilic
ACS Paragon Plus Environment

Page 27 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
binding site. Analogues of 6ꢀcyclohexylmethoxyꢀ2ꢀ(4’ꢀsulfamoylanilino)purine, in which the purine
ring was replaced by a triazolopyrimidine, pyrazolopyrimidine or pyrrolopyrimidine were only
marginally less active against CDK2 than the parent purine. However, replacement with an
imidazopyridine or imidazopyrimidine gave much less potent derivatives. Coꢀcrystal structures of
inhibitors bound to CDK2 and CDK1 revealed that the binding mode of the purineꢀbased inhibitor
series is conserved. We show that inhibitor binding to CDK2 stabilises a glycineꢀrich loop
conformation that shapes the ATP ribose binding pocket, resulting in effective inhibition of CDK2.
We propose that this region of the active site might be the basis of the design of further inhibitors
differentiating between CDK1 and CDK2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Experimental Section
General Synthetic Procedures
Chemicals and solvents were obtained from standard suppliers. Solvents were either dried by
standard techniques or purchased as anhydrous. Reactions needing microwave irradiation were
TM
carried out in an Initiator Sixty Biotage apparatus. Petrol refers to petroleum ether (bp 40–60 °C,
reagent grade, Fisher Scientific). All reactions that required inert or dry atmosphere were carried out
under a blanket of nitrogen, which was dried by passage through a column of phosphorus
pentoxide. Glassware was dried in an oven prior to use. Column chromatography was carried out
using 40ꢀ60 ꢁm mesh silica in glass columns under medium pressure or with a Biotage SP4 flash
purification system using KPꢀSi. Thin layer chromatography (TLC) was performed on 20 mm
precoated plates of silica gel (Merck, silica gel 60F254); visualisation was achieved using
ultraviolet light (254 nm). NMR spectra were recorded on a Bruker Spectrospin AC 300E (300
MHz) NMR Spectrometer or Bruker BioSpin UltraShield Plus 500 MHz using deuterated solvent as
a lock. IR spectra were recorded on a BioꢀRad FTS 3000MX diamond ATR, and UV analysis was
performed using a Hitachi Uꢀ2000 spectrophotometer. LCꢀMS analysis was carried out on a
Micromass Platform instrument operating in positive and negative ion electrospray mode,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
employing a 50 × 4.6 mm C18 column (Supelco Discovery or Waters Symmetry) and a 15 min
gradient elution of 0.05% formic acid and methanol (10–90%). HRMS were measured using a
Finnigan MAT 95 XP or a Finnigan MAT 900 XLT by the EPSRC National Mass Spectrometry
Service Centre (Swansea). The purity of final compounds was assessed by reversedꢀphase HPLC;
all tested compounds were >95% purity. HPLC instrument, Agilent 1200 equipped with a
photodiode array detector (190ꢀ400 nm). Sample temperature, ambient; injection volume, 5 ꢀL;
flow rate, 1mL/min. 5% to 100% MeCN gradient over 9 min and an isocratic hold at 100% MeCN
for 2.5 min, before returning to initial conditions. Mobile phase A = 0.1% ammonia in water or
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
.1% formic acid in water, mobile phase B = MeCN. Column: Waters XSELECT CSH C18, 3.5
ꢁm, 4.6 mm × 150 mm or Waters XTerra RP18, 5 ꢁm, 4.6 mm × 150 mm. Column maintained at
ambient temperature.
General procedure A. To a stirred suspension of the appropriate haloheterocycle (1.0 mol. equiv.)
and 4ꢀaminobenzenesulfonamide (2.0 mol. equiv.) in TFE (25 mL/g of haloheterocycle) was added
TFA (2.5 mL/g of haloheterocycle) dropwise. The resulting solution was heated under reflux for 12ꢀ
48 h under a nitrogen atmosphere. The solvent was removed in vacuo and the residue was
redissolved in EtOAc (10ꢀ20 mL). The solution was washed with saturated aqueous sodium
bicarbonate solution (3 × 10 mL), and the aqueous extracts were combined and washed with EtOAc
(2 × 15 mL). The combined organic layers were dried (Na SO ) and the solvent removed under
2
4
reduced pressure to give a residue that was purified in the manner indicated.
47
6
ꢀChloroꢀ2ꢀfluoroꢀ9
H
ꢀpurine (10). To a stirred solution of HBF (48% aqueous, 120 mL) at 0
4
ºC, was added 2ꢀaminoꢀ6ꢀchloropurine (9) (6.0 g, 35.0 mmol). Over 20 min, a solution of NaNO
2
(4.9 g, 70.0 mmol) in water (200 mL) was added dropwise, ensuring the temperature remained close
to 0 ºC. The pale yellow solution was raised to room temperature and stirred for 18 h. The resulting
2 3
solution was neutralised to pH 7 in an ice bath at 0 °C, by addition of Na CO (6.00 g) in water (200
mL). Solvents were removed in vacuo and the residual solid was redissolved in MeOH (100 mL)
and adsorbed onto silica (250 mL). The crude material was purified by chromatography (silica; 10%
ACS Paragon Plus Environment

Page 29 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
4
7
MeOH:DCM) to afford 10 as a white crystalline solid (4.52 g, 75%); mp 171ꢀ173 ºC (lit., mp 174
ꢀ1
1
6
ºC); UV λmax (EtOH) 393 nm; IR (cm ) 2964, 2785, 1735, 1581; H NMR (500 MHz, DMSOꢀd ) δ
+
+
8
.60 (1H, s, Hꢀ8), 13.9 (1H, s, NH); LRMS (ES ) m/z 172.6 [M+H] .
48, 49
2
ꢀFluoroꢀ9
H
ꢀpurine (11).
To a stirred suspension of 6ꢀchloroꢀ2ꢀfluoropurine (10) (0.30 g, 1.74
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
mmol) and palladium hydroxide on carbon (20% w/w, 0.30 g) in methanol (15 mL) was added
ammonium formate (0.34 g, 5.35 mmol). The suspension was heated under reflux for 1h before
filtering through a pad of Celite, eluting with methanol (20 mL). Removal of volatiles under
48
reduced pressure afforded 11 as a white solid (240 mg, 100%); mp 219 °C (dec.) (lit.,
1
decomposed at 216 °C); H NMR (300 MHz, DMSOꢀd
6
) δ 8.60 (1H, s, Hꢀ8), 9.01 (1H, s, Hꢀ6), 13.9
+
+
(
1H, s, NH); LRMS (ES ) m/z 139.2 [M+H] .
Phenylꢀ(9Hꢀpurinꢀ2ꢀyl)amine (12). Prepared from aniline (0.182 mL, 2 mmol) and 2ꢀfluoroꢀ9Hꢀ
purine (11, 0.138 g, 1.0 mmol) in accordance with general procedure A. The reaction mixture was
stirred under reflux for overnight. After removal of the solvent in vacuo the residue was dissolved
3 4
in EtOAc (10ꢀ30 mL), washed with saturated aqueous NaHCO (3 × 30 mL) and dried (MgSO ).
The crude material was purified by chromatography (silica; 0ꢀ20% MeOH:DCM), followed by
recrystallization from DCM to afford 12 as an offꢀwhite solid (65 mg, 31%); R 0.20 (5%
f
o
ꢀ1
MeOH:DCM); mp 200ꢀ201 C; UV λmax (EtOH) 328, 270, 239, 206 nm; IR (cm ) 3234, 3103,
1
2
804, 1624, 1602, 1539, 1402, 1292, 1217; H NMR (300 MHz, DMSOꢀd
6
) δ 6.91 (1H, t, J = 7.5
Hz, Hꢀ4'), 7.28 (2H, t , J = 7.5 Hz, Hꢀ3' and Hꢀ5'), 7.83 (2H, d, J = 9.0 Hz, Hꢀ2' and Hꢀ6'), 8.82 (1H,
+
+
s, Hꢀ6), 8.24 (1H, s, Hꢀ8), 9.53 (1H, s, NH), 12.9 (1H, s, NH); LRMS (ES ) m/z 212.0 [M+H] ;
+
HRMS calcd for C H N [M+H] 212.0931, found 212.0933.
11
10
5
4
ꢀ(9HꢀPurinꢀ2ꢀylamino)benzenesulfonamide (13). Prepared following general procedure A
2
from 2ꢀfluoroꢀ9Hꢀpurine (11, 0.10 g, 0.725 mmol). Recrystallization from MeOH:H O gave 13 as a
o
pale white solid (74 mg, 35%); R
f
0.40 (20% MeOH:DCM); mp > 320 C (dec); UV λmax (EtOH)
ꢀ1
1
2
87, 213 nm; IR (cm ) 3223, 1585, 1481, 1307, 1251, 1148, 1091; H NMR (300 MHz, DMSOꢀd )
6
δ 7.17 (2H, s, SO
2
NH
2
), 7.73 (2H, d, J = 9.0 Hz, Hꢀ3' and Hꢀ5'); 7.99 (2H, d, J = 9.0 Hz, Hꢀ2' and
ACS Paragon Plus Environment