Benzamide capped peptidomimetics as non-ATP competitive inhibitors of CDK2 using the REPLACE strategy

Article abstract of DOI:10.1016/j.bmcl.2016.05.067

Inhibition of cyclin dependent kinase 2 (CDK2) in complex with cyclin A in G1/S phase of the cell cycle has been shown to promote selective apoptosis of cancer cells through the E2F1 pathway. An alternative approach to catalytic inhibition is to target the substrate recruitment site also known as the cyclin binding groove (CBG) to generate selective non-ATP competitive inhibitors. The REPLACE strategy has been applied to identify fragment alternatives and substituted benzoic acid derivatives were evaluated as a promising scaffold to present appropriate functionality to mimic key peptide determinants. Fragment Ligated Inhibitory Peptides (FLIPs) are described which potently inhibit both CDK2/cyclin A and CDK4/cyclin D1 and have preliminary anti-tumor activity. A structural rationale for binding was obtained through molecular modeling further demonstrating their potential for further development as next generation non ATP competitive CDK inhibitors.

Full text of DOI:10.1016/j.bmcl.2016.05.067

Bioorganic & Medicinal Chemistry Letters xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Benzamide capped peptidomimetics as non-ATP competitive
inhibitors of CDK2 using the REPLACE strategy
y
à
⇑
Padmavathy Nandha Premnath , Sandra N. Craig, Shu Liu , Campbell McInnes
Drug Discovery and Biomedical Sciences, South Carolina College of Pharmacy, University of South Carolina, Columbia, SC 29208, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Inhibition of cyclin dependent kinase 2 (CDK2) in complex with cyclin A in G1/S phase of the cell cycle
has been shown to promote selective apoptosis of cancer cells through the E2F1 pathway. An alternative
approach to catalytic inhibition is to target the substrate recruitment site also known as the cyclin bind-
ing groove (CBG) to generate selective non-ATP competitive inhibitors. The REPLACE strategy has been
applied to identify fragment alternatives and substituted benzoic acid derivatives were evaluated as a
promising scaffold to present appropriate functionality to mimic key peptide determinants. Fragment
Ligated Inhibitory Peptides (FLIPs) are described which potently inhibit both CDK2/cyclin A and
CDK4/cyclin D1 and have preliminary anti-tumor activity. A structural rationale for binding was obtained
through molecular modeling further demonstrating their potential for further development as next gen-
eration non ATP competitive CDK inhibitors.
Received 6 April 2016
Revised 20 May 2016
Accepted 21 May 2016
Available online xxxx
Keywords:
Cyclin dependent kinase
Protein–protein interaction
Cyclin groove
REPLACE
Ó 2016 Elsevier Ltd. All rights reserved.
Inhibitor
The cyclin dependent kinases (CDKs) are activated through
binding to cyclin proteins which act as positive regulatory sub-
units. CDKs are known to play major roles in regulation of both
checkpoint control of the cell cycle1 and transcription initiation
and elongation2 where the latter effects are believed to contribute
to toxicities observed with first generation CDK inhibitors evalu-
ated in the clinic.3,4 In addition to its roles in regulating the G1/S
transition5, CDK2/cyclin A (CDK2A) is intimately involved in pro-
gression through S phase.6 A critical role of this latter complex is
to phosphorylate the E2F1 transcription factor required for expres-
sion of S phase genes in a timely manner and to terminate its activ-
ity through inducing its dissociation from DNA.7 Tumor cells,
which intrinsically possess high levels of E2F1 transcriptional
activity, can be sensitized to apoptosis through maintaining this
protein in its DNA bound state as tumor cells exit S phase.7 CDK2A
inhibition would therefore push tumor cells past the apoptotic
threshold and result in synthetic lethality with tumors that are null
for Rb and/or p53 and therefore have deregulated E2F. Selective
inhibition of cell cycle CDKs resulting in blocking of phosphoryla-
tion of the Rb protein and E2F1 can be attained through the cyclin
groove substrate recruitment site and this protein–protein interac-
tion can potentially be exploited therapeutically.3,7–9 Non-ATP
competitive inhibition through the cyclin groove also avoids com-
peting with high intracellular concentrations of ATP. The previ-
ously optimized octapeptide inhibitor, HAKRRLIF based on a
sequence found in the endogenous CDK inhibitor, p21waf1 binds
potently to the cyclin groove through multiple peptide determi-
nants including ion pairing interactions of the KRR motif and
through the hydrophobic side chains with primary and secondary
lipophilic subsites present (Figs. 1 and 2).10,11 Previous studies
applying the REPLACE strategy9,12–16 have identified capping
groups replacing the N-terminal tetrapeptide that interact with
either the secondary hydrophobic pocket or the critical arginine
binding site but not effectively with both.12–14,16 From this per-
spective, an alternative scaffold for presenting binding groups that
could interact with both regions was sought and a substituted ben-
zoic acid derivatives were generated and appended to a drug-like
peptidomimetic sequence previously identified. The resulting frag-
ment-peptide hybrids were found to potently inhibit CDK2/cyclin
A and CDK4/cyclin D1 further validating the REPLACE strategy
and showing promise as compounds with preliminary anti-prolif-
erative activity useful for development as cell cycle CDK specific
therapeutics.
⇑
Corresponding author. Tel.: +1 (803) 576 5684.
E-mail addresses: pnpremnath@une.edu (P.N. Premnath), shu.liu@ucsfmedctr.
Synthesis of benzoic acid derived N-terminal capping groups:
Furoic acid capped FLIPs containing basic functional groups
designed to interact with the arginine binding site were studied
previously and shown to possess moderate binding and inhibition
of CDK2/cyclin A.13 Phenyl-(1,2,4)-triazole derived capping groups
org (S. Liu), mcinnes@sccp.sc.edu (C. McInnes).
y
Present address: Department of Pharmaceutical Sciences, University of New
England College of Pharmacy, 716 Stevens Ave, Portland, ME, USA.
à
Present address: Department of Surgery, University of California at San Francisco,
USA.
http://dx.doi.org/10.1016/j.bmcl.2016.05.067
0960-894X/Ó 2016 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Premnath, P. N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.05.067

2
P. N. Premnath et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
azine-1-carboxylate was undertaken)18 and alkylation of the free
piperazine nitrogen where required was completed by reductive
amination with the requisite aliphatic aldehyde to generate the
tertiary amine prior to base hydrolysis19 to liberate the capping
group.
The BA2 scaffold (Capping groups 15a, 16a, Table 1, Scheme 2,
SI) was investigated to determine the consequences of restraining
the rotation of the cyclic amine and the effect of this in interacting
with the arginine binding site. Piperidin-4-ylidenemethyl benzoic
acid derivatives were synthesized through Wittig reaction of
methyl 4-((diethoxyphosphoryl) methyl)benzoate with amino
piperidone (Scheme 2). The utility of this reaction and the potential
SAR information could be further extended by high pressure
hydrogenation of the ylidene to reduce the double bond and obtain
20
the BA1 scaffold where X is CH2 (Scheme 2, Table 1).
The BA3 scaffold (capping groups 17a–19a, Table 1, Scheme 3,
SI) was generated to determine the consequences of adding an
amine spacing bridge (between the cyclic amine and the benzoic
acid core) on the ion-pairing interactions with the arginine binding
site. This scaffold also allowed significant diversity to be incorpo-
rated since it allowed not only different piperidine substitutions
to be added but also enabled functionalization of the spacing nitro-
gen. These compounds were synthesized (Scheme 3) by reductive
amination using the requisite amine followed by Boc-protection
as necessary. The synthesis of analogs bis substituted at the 3
and 4 positions of the BA1 scaffold (Capping groups 20a–24a,
Table 2, Scheme 4, SI) was completed in a similar fashion to the
4-(piperazin-1-ylmethyl)benzoic acid derivatives however using
reductive amination18 of 4-formyl-3-hydroxybenzoic acid and the
requisite amine. These were generated as described above to
simultaneously interact with both the arginine binding site and
the minor hydrophobic pocket. As required, the intermediate (i.e.,
4-((4-(tert-butoxycarbonyl)piperazin-1-yl)methyl)-3-hydroxyben-
zoic acid) was alkylated to generate the 3-ethoxy and 3-propoxy
derivatives.
Figure 1. The cyclin groove is located on the cyclin regulatory subunit (blue
secondary structure) of the CDK2 (magenta) complex and is required for substrate
recruitment prior to phosphorylation. Peptidomimetics (shown in yellow) targeting
this site inhibit CDK activity in a non-ATP competitive fashion.
successfully interface with the hydrophobic subsite however do
not make effective contacts with the arginine binding site12,14
and benzoic acid derived cyclin groove inhibitors do not interact
with the minor hydrophobic pocket.17 Based on these precedents,
the design and synthesis of a unique series of benzamide fragment
alternatives was undertaken (Table 1). Since this scaffold has con-
siderable potential to further exploit peptide interacting residues
and therefore more completely mimic the peptide-cyclin interface
compared to previously investigated N-terminal capping
groups12–14, expansion of this as a core structure was investigated
through additional substitutions. In the first instance, a number of
piperazine containing functional groups (BA1 scaffold, Table 1,
Supplementary information, SI) were installed at the 4 position
by reductive amination of 4-formyl methyl benzoate with the
appropriate cyclic amine to form the secondary amine. Capping
groups including a derivatized piperazine group were installed
using reductive amination subsequent to either methyl or benzyl
ester protection and followed by hydrogenation or base hydrolysis
as required (capping groups 2a–14a, Table 1, Scheme 1).
Further bis substituted analogs generated to probe the SAR of
the two subsites were made by applying the Mitsunobu reaction21
(capping groups 25a–28a, Table 2, Scheme 5, SI) and included
4-piperidinyloxy,
3-methoxy-4-piperidinyloxy
and
3,5-bis
piperidinyloxy derivatives.
Structure activity relationship of peptides capped with mono-sub-
stituted benzoic acid derivatives: After generation of the capping
groups, FLIP molecules (Table 1) were synthesized using previously
described methods12 (also see SI) and evaluated in a fluorescence
polarization competitive binding assay shown to be a robust mea-
surement of the ability of compounds to bind to the cyclin groove12
(also see SI) and therefore inhibit CDK activity through blocking of
substrate recruitment. Compounds containing the benzamide scaf-
fold (Table 1) were potent mimics of the N-terminal tetrapeptide
including 1, possessing a guanidinomethyl group at the 4-position
To further functionalize the BA1 scaffold, analogs were incorpo-
rated where the piperazine ring was alkylated at the 3 and/or 4
position (Scheme 1, SI), To achieve this, reductive amination of 4-
formylmethyl benzoate (with e.g., tert-butyl (S)-3-methylpiper-
which was the most effective inhibitor (CDK2/cyclin A 0.69
CDK4/cyclin D1—15.32 M). Incorporation of the piperazine-1yl
group (BA1, 2, 35.16 M) was a less effective substitution at the
lM and
l
l
4-position whereas addition of a methylene spacer into the Ncap
(4-(piperazine-1ylmethyl), 3) resulted in a 7-fold increase in bind-
ing to CDK2A (IC50 = 5.3
(12.87 M). Varying the N-terminal tetrapeptide sequence in the
4-piperazinylmethylbenzamide context was undertaken.
An optimized sequence (RLNpfF, 3, 5.3 M) was found to be 2
fold more potent than the native p21 sequence (RLIF, 4, 12.9 M)
lM) and 2 fold increase in IC50 for CDK4D
l
l
l
for cyclin A whereas replacement of the second arginine (ALIF, 5)
resulted in an inactive FLIP in comparison. Further derivatization
of the core scaffold (BA1, Table 1) was undertaken and included
the synthesis of analogs with varying substitution of the free piper-
azine nitrogen and addition of methyl groups onto the 2-position.
Methylation of the piperazine N in compound 6 resulted in a 2 fold
Figure 2. Calculated binding modes for the capping groups of compounds 2 and 3
with cyclin A (pdb id 2UUE represented as a Connolly surface and as ribbon
diagram), showing an overlay of the 4-piperazinylbenzamide (magenta carbon
atoms) and 4-piperazinylmethylbenzamide groups (cyan carbon atoms), respec-
tively. As highlighted the methylene spacer in compound 3 results in more efficient
ion pairing with Glu220 and 224 therefore leading to the observed potency
increase.
Please cite this article in press as: Premnath, P. N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.05.067

P. N. Premnath et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
3
Table 1
Structure–activity relationship of monosubstituted benzoic acid derived capped peptides
R
H2N
HN
HN R2
N
R
NH
HN R2
X
N
HN R2
R
N
X
(CH2)n
HN R2
Y
O
BA3
R1
BA2
O
O
BA1
O
540
Compound Capping
group
Scaffold
n
X
R
R1
R2
CDK2/cyclin A IC50
M)
CDK4/cyclin D1 IC50
(lM)
(
l
1
2
3
4
5
6
7
1
540
BA1
BA1
BA1
BA1
BA1
BA1
—
0
1
1
1
1
1
—
N
N
N
N
N
N
—
H
H
H
H
CH3
H
—
H
H
H
H
H
(S)
CH3
(R)
CH3
H
RLNpfF
Arg-Leu-Ile-Phe
RLNpfF
Arg-Leu-Ile-Phe
Ala-Leu-Ile-Phe
Arg-Leu-Ile-Phe
Arg-bLeu-N-MePhe-
NH2
Arg-bLeu-N-MePhe-
NH2
Arg-bLeu-N-MePhe-
NH2
Arg-bLeu-3TA-NH2
Arg-bLeu-3TA-NH2
0.69 0.34
35.16 10.17
5.30 0.99
12.89 2.99
>100
15.32 5.09
23.82
2a
3a
3a
3a
6a
7a
12.87 1.67
11.70 5.92
51.49 21.34
11.31 1.48
>100
7.92 1.39
13.39 4.35
8
9
8a
9a
BA1
BA1
1
1
N
N
H
22.28 5.22
8.32 0.36
45.45 0.47
N-
48.89 40.36
methylpiperidine
CH3CH2
CH3CH2
10
11
10a
11a
BA1
BA1
1
1
N
N
H
17.44 13.39
6.78 2.72
74.28 4.45
45.60 13.69
(S)
CH3
H
(S)
CH3
H
—
—
—
12
13
12a
13a
BA1
BA1
-
1
N
N
CH3CH2CH2
CH3CH2CH2
Arg-bLeu-3TA-NH2
Arg-bLeu-3TA-NH2
31.34 15.71
22.42 9.66
8.52 4.43
9.36 6.56
14
15
16
17
14a
15a
16a
17a
BA1
BA2
BA2
BA3, X = CH2,
Y = N
1
1
1
1
CH2
—
—
H
H
CH3
CH3
Arg-bLeu-3TA-NH2
Arg-bLeu-3TA-NH2
Arg-bLeu-3TA-NH2
Arg-bLeu-3TA-NH2
>30
—
>100
69.91 7.07
—
9.16 8.74
26.02 7.13
31
—
18
19
18a
19a
BA3, X = N,
Y = CH2
BA3, X = N,
Y = CH2
1
1
—
—
H
—
—
Arg-bLeu-3TA-NH2
Arg-bLeu-3TA-NH2
>100
>50
—
—
CH3
Scheme 1. Synthesis of BA1 Scaffold N-terminal capping groups.
increase in binding to CDK2A however did not affect interaction
with CDK4D. Subsequent capping group modifications were inves-
tigated in either the Arg-bhomoLeu-N-MePhe-NH2 or Arg-bhomo-
Leu-3-thienylalanine(3TA)-NH2 contexts previously utilized to
improve proteolytic stability and cellular permeability.12 These
sequences have comparable potency in vitro binding as the native
p21 sequence, (RRLIF) to cyclin A even with one less amino acid
residue. The extra methylene of the betahomoleucine residue
mimics the Ile spacer residue in the p21 context, allows more effi-
cient adaptation of the Leu and Phe and hence greater complemen-
tarity with the hydrophobic pocket. Addition of a methyl group
onto the 2-position of the piperazine ring in the S-configuration
resulted in similar CDK2A potency to the parent BA1 scaffold con-
taining FLIP (7, IC50 = 13.4
lM) however the R-isomer was less
active on CDK2A (8, IC50 = 22.3
lM). A further series of N-substi-
tuted piperazine derivatives were evaluated and included the
placement of a piperidinyl group on the 4-N. This analog had
improved activity relative to the free piperazine with just under
a 2-fold increase (9, IC50 = 8.3
tion of the same position with alkyl chains of increasing length
revealed that both N-ethyl (10, IC50 = 17.4 M) and N-propyl (12,
IC50 = 31.3 M) substitutions were deleterious to binding in the
lM) in binding to CDK2A. Substitu-
l
l
Please cite this article in press as: Premnath, P. N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.05.067

4
P. N. Premnath et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
Scheme 2. Synthesis of BA1 and BA2 Scaffold N-terminal capping groups.
Scheme 3. Synthesis of BA3 Scaffold N-terminal capping groups.
CDK2A context. The N-propyl containing FLIP possessed substan-
tially improved binding to cyclin D relative to the N ethyl
Structure activity relationship of peptides capped with 3,4-substi-
tuted benzoic acid derivatives: A further series of FLIPs incorporated
bis substituted benzoic acid core structures in order to provide
greater functionalization and more completely mimic the manifold
interactions observed in the peptide/cyclin crystal structures.10,11
This series includes substituents appended in both the piperazinyl-
methyl and piperidinyloxy contexts (Table 2) where introduction
of substituents at the 3-position can potentially interact with this
pocket and improve potency through additional van der Waals
interactions.
(IC50 = 45.6 vs 8.5
extended by incorporating the 2S-methyl group in both the
N-ethyl (11, IC50 = 6.8 M) and N-propyl (13, IC50 = 22.4 M)
lM). The structure–activity of this series was
l
l
contexts. Each of these analogs were more potent than their
unsubstituted counterparts when considering their CDK2/A
activity with compound 11 attaining low micromolar binding
and showing more than 2-fold greater binding.
As mentioned above, FLIPS were generated containing the BA2
scaffold to determine the consequence of restraining the cyclic
amine relative to the benzamide core structure. The 4-(piperidin-
4-ylidenemethyl)benzamide FLIPs were generated in the free
The contribution of the 3-substituent was examined in the first
instance through addition of a hydroxyl group at this position (20,
IC50 = 4.56
potency enhancement compared to 4 for CDK2A but without a con-
comitant increase in activity towards CDK4D (IC50 = 10.41 M).
Incorporation of a 3-ethoxy substituent in combination with the
4-piperazinylmethyl group (21, IC50 = 5.9 M) lead to similar bind-
ing (compared to 20) for binding to cyclin A and 2 fold better activ-
ity with respect to cyclin D (5.1 M vs 10.4 M). Comparable
lM) and was shown through testing to result in a 2 fold
piperidine (15, IC50 = 9.2
lM) and N-methyl contexts (16,
IC50 = 26.0 M) with the results indicating that the order of
l
l
potency was reversed from the BA1 scaffold. The BA3 scaffold
incorporated both the 4-((piperidin-4-yl)amino)methyl) and 4-
((piperidin-3-yl)amino)methyl)benzoic acid core structures
(Table 1) to probe the spacial requirements of interaction of the
ion pairing of the cyclic amine with the arginine binding site. In
vitro testing of this series (17–19) showed that only weak binding
to cyclin A was observed although the best analog had preliminary
anti-proliferative activity.
l
l
l
results are seen for 22 (RbLNMeF-NH2) versus 20 (RLIF) in both
cyclins where the N-cap is 3-hydroxy-4-(piperazin-1-ylmethyl)
benzamide. The RbLNMeF counterpart (23) to the 3-ethoxy, 4-
(piperazin-1yl)benzamide RLIF (21) has diminished potency for
Please cite this article in press as: Premnath, P. N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.05.067

P. N. Premnath et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
5
Table 2
Structure–activity relationship of 3,4-substituted benzoic acid derived capped peptides
R3
HN R4
R2
O
R1
Compound Capping group R1
R2
R3
H
R4
CDK2/cyclin A IC50 (
4.56 0.74
l
M) CDK4/cyclin D1 IC50 (
lM)
HN
N
N
N
N
N
20
21
22
23
20a
21a
20a
21a
OH
Arg-Leu-Ile-Phe
Arg-Leu-Ile-Phe
10.41 0.61
HN
HN
HN
HN
OCH2CH3
OH
H
H
H
5.95 2.53
7.42 1.69
22.61 1.89
5.14 1.47
11.76 1.01
18.64 9.72
Arg-bLeu-N-MePhe-
NH2
Arg-bLeu-N-MePhe-
NH2
OCH2CH3
24
24a
OH
H
Arg-bLeu-3TA-NH2
13.1 2.1
—
HN
HN
HN
O
O
O
25
26
27
28
25a
25a
27a
28a
H
H
H
H
HN
Arg-Leu-Ile-Phe
31.34 15.71
40.38 17.76
22.42 9.66
>30
8.52 4.43
30.99 1.18
9.36 6.56
—
Arg-bLeu-N-MePhe-
NH2
H
OCH3
HN
Arg-Leu-Ile-Phe
O
O
H
Arg-bLeu-3TA-NH2
Scheme 4. Synthesis of bis substituted N-terminal capping groups based on BA1 Scaffold.
both cyclin A (22.61
version of FLIPs 7 (11.6
l
M) and D (18.64
l
M). The 3,4 bis substituted
M) were generated
cyclin D relative to 25. Bis substituted piperidinyloxy derivatives
were generated (27, 22.4 M and 28, >30 M) and compared to
l
M) and 9 (>100
l
l
l
where the 3-position was hydroxyl in both cases and where activ-
ity was retained in the context of the 2-methylpiperazine but not
in the 4-((4-(1-methylpiperidin-4-yl)piperazin-1-yl)methyl)ben-
zamide capped peptide.
As a means of generating further diversity, FLIPs containing
piperidinyloxy substituents were synthesized in both mono and
bis substituted contexts. Testing of the 4-piperidinyloxy FLIP (25,
their mono 4-substituted counterparts. The 3-methoxy-4-
piperidinyloxy derivative (27) was evaluated and demonstrated
to have increased binding to CDK2A (22
A) and to be essentially equipotent towards CDK4D (9
M). The 3,4 bis piperidinyloxybenzamide capped peptide was
lM vs 31
lM in cyclin
lM vs
8 l
generated and shown to have comparable activity to the analog
mono-substituted at the 4 position.
8.52
azinyl counterpart (2, 23.82
and 2) however was equipotent in binding to cyclin A in its
CDK2 complex. FLIP 26 (IC50 = 31 M) had a decreased binding to
l
M) resulted in 4 fold more potent molecule than its piper-
Structural analysis of FLIP SAR with the cyclin A and D binding
grooves: Previously described capping group scaffolds have led to
potent CDK inhibitors however none have completely mimicked
the interactions of the peptide binding groups. 1-(3,5-dichlorophe-
l
M) in binding to CDK4D (Tables 1
l
Please cite this article in press as: Premnath, P. N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.05.067

6
P. N. Premnath et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
Scheme 5. Synthesis of piperidinyloxy benzoic acid N-terminal capping groups.
nyl)-5-methyl-1H-1,2,4-triazole-3-carboxylate derived capping
groups primarily contact the secondary hydrophobic pocket and
make H-bond interactions with Trp217 however bypass the Lys-
Arg binding site.12,14 Another series containing substituted hetero-
cyclic core structures primarily bind to the acidic region but do not
occupy the secondary pocket.13 The benzoic acid scaffold was
selected as a potentially effective core structure for presentation
of multiple functional groups to interact more completely with
the cyclin groove with the secondary pocket while also making
ion-pairing interactions with acidic residues.
In order to improve the affinity of the N-capping group through
the synthetic versatility of the benzoic acid scaffold, additional
substituents were appended at the 4 position to include groups
capable of ion-pairing with the cyclin groove. In the first instance,
mono substituted benzoic acids with basic groups at the 3 or 4
positions were generated and included (guanidomethyl (1), piper-
azin1-yl (2), piperidin-4yloxy (26) and 4-methylpiperazin-1yl-
methyl (4) (Table 1 and 2). The guanidinomethyl substitution
was generated as a RLNpfF FLIP and the most potent cyclin groove
entropically to binding. Structural analysis indicates confirmation
of the initial design hypothesis in that the piperazine 2-methyl
group of 8 points out of the minor pocket however in 7, the methyl
group interacts effectively with the Trp217 resulting in tighter
binding (Supplementary Fig. 1).
Further modifications included addition of alkyl substituents to
the free piperazine nitrogen in both the unsubstituted and 2-
methyl contexts. In both of these scenarios the N-ethyl substituent
is more favorable for binding and therefore suggests that the pro-
pyl group sterically interferes with effective interaction with the
pocket. This was further illustrated in the binding modes generated
where interference with Glu220 is observed and therefore demon-
strates that the ethyl group is optimal (data not shown). Investiga-
tion of the binding of FLIPs containing the BA2 scaffold (have an
unsaturated link between the cyclic amine and the benzyl group)
strongly suggests that this is beneficial as shown by comparison
of 15 with the saturated version (14). Molecular modeling of these
restrained analogs indicate that the double bond positions the
basic nitrogen of the piperidine closer to both Glu220 and
Glu224 and therefore is more optimal for ion pairing interactions
(see Supplementary Fig. 2). Methylation of the piperidine N leads
to a steric clash in this context providing an explanation for the
decreased activity of 16. Similar activity of the 3-aminopiperidine
analogs (BA3, 17) to 14 and the inactivity of the 4-aminopiperidine
substitutions (BA3, 18, 19) were validated structurally by the sub-
optimal positioning of the cyclic amine (e.g., compared to 4).
In an effort to build on the results obtained for the monosubsti-
tuted benzamide FLIPs, Ncapping groups were generated that com-
bined substituents at the 3 and 4 positions to interact with the
minor hydrophobic pocket and the arginine site. Addition of a phe-
nolic OH group at the 3-position (FLIPs 20, 22, 24 vs 4) in combina-
tion with the 4-piperazinylmethyl substituent leads to a 2–3 fold
potency increase. While it was hypothesized that this group would
increase potency through acting as an H-bond acceptor to the
indole NH group of Trp217, molecular modeling results revealed
that the binding mode is stabilized by an intramolecular H-bond
while maintaining an effective salt bridge (Fig. 3). Alkylation of
the OH to generate the ethoxy derivatives (21, 23) yielded similar
potency in terms of binding to cyclin A however a 2-fold increase
was observed versus cyclin D1. The binding mode of these analogs
suggests that the optimized cyclin D1 affinity results from greater
complementarity with the secondary pocket in this context. The
lack of potency increase for CDK2A compared to the free hydroxyl
(21 vs 20) is suggested by modeling in that the intramolecular
hydrogen bond is not possible in the alkylated context. Overall
though, the activity profiles of these two compounds confirms
the hypothesis that having a basic group at 4 position and an alkyl
group at 3 position improves the activity of benzamido-FLIPs
inhibitor in this series (0.69 lM). The effectiveness of this com-
pound is due to the very strong electrostatic interaction of the
highly basic guanidine with the acidic region of cyclin A (E220,
E224, Fig. 2).
This substitution was not selected for further optimization since
it is likely that the presence of a guanidino group in addition to
Arg5 would prevent cellular penetration. The increased activity
of the 4-piperazinylmethylbenzamide FLIP (4) relative to its 4-
piperazinyl counterpart (2) demonstrates that optimization of
the ion-pairing interaction distance leads to greater binding
(Table 1, Fig. 2). This was relevant in both the cyclin A and cyclin
D1 contexts and structurally validated through the observation of
closer salt bridge contacts in the modeled complexes with cyclin
A (Fig. 2). For 2, the ion pairing distances are 5.3 Å (GLU220) and
4.7 Å (GLU224) whereas in 3, charge–charge contacts are signifi-
cantly reduced to 3.3 Å and 3.8 Å. Compared to the Ncap with
the free piperazine N4 nitrogen, the methylated analog (6) had
increased activity in binding to cyclin A however similar affinity
for cyclin D1. This probably results from a decreased desolvation
penalty of the protonated and methylated amine and through
increased van der Waals interactions (since methylation would
not be expected to change the pKa of this group substantially).
Modifications to the piperazine ring (7, 8) included addition of R
and S methyl group at the 2 position and resulted in almost 2-fold
potency enhancement for the S methyl FLIP (relative to others con-
taining the Arg-bLeu-N-MePhe-NH2 peptide sequence). The struc-
tural basis for this increase was suggested through molecular
modeling in that the methyl group restricts rotation of the piper-
azine due to overlap with the phenyl ring and therefore contributes
Please cite this article in press as: Premnath, P. N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.05.067

P. N. Premnath et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
7
impart drug-like properties, preliminary anti-proliferative activity
was obtained. As a whole the results obtained indicate that this
series has potential for further development as next generation
CDK inhibitors based on non-ATP competitive blocking of kinase
activity through the cyclin groove.
Acknowledgments
We thank Drs. Michael Walla and William Cotham in the
Department of Chemistry and Biochemistry at the University of
South Carolina for assistance with Mass Spectrometry and Helga
Cohen and Dr. Perry Pellechia for NMR spectrometry. This work
was funded by the National Institutes of Health through the
research project grants, R01CA131368 and R41CA189620.
Figure 3. The calculated binding mode for compound 20 indicating the interactions
of the piperazinyl group and the intramolecular H-bond stabilizing the conforma-
tion of the capping group.
Supplementary data
Supplementary data (characterization information for all pep-
tides and FLIPS; additional figures, all synthetic procedures and
characterization data and procedures used for the FP and cellular
assays) associated with this article can be found, in the online ver-
sion, at http://dx.doi.org/10.1016/j.bmcl.2016.05.067.
through ion-pairing and van der Waals interactions. Based on this
observation, FLIPs were generated with piperidinyloxy sub-
stituents at the 3 and/or 4 positions and in combination with a
methoxy group. The 3-fold lower affinity of piperidin-4yloxy (25)
monosubstituted analog for cyclin A was evidenced by suboptimal
contacts with the key subsites. These interactions are more favor-
able in the cyclin D1 context thus correlating well with the
enhanced binding of this analog. Compound 27 showed improved
activity for cyclin A (vs 26) and compares favorably with 21 which
as described earlier effectively interacts with both the arginine and
minor hydrophobic subsites on both cyclins. Inclusion of the
piperidinyloxy group at both the 3 and 5 positions was detrimental
to activity as these groups did not make effective ion pairing inter-
actions with Glu220 and Asp283.
Previously identified N-terminal capping groups based on
phenyltriazole substructures were used as a scaffold to generate
a peptidomimetic that possessed greater drug-like properties and
which demonstrated anti-proliferative activity in U2OS and
DU145 cells.12 The cellular activity (MTT assay) of selected benza-
mide capped FLIPS were assessed and compared with the previ-
ously reported compounds which were shown to have
respectable growth inhibitory activity. For the most part evaluated
compounds were inactive with the exception of FLIP 7 which
exhibited weak but reproducible anti-proliferative activity (U2OS
References and notes
1. Sherr, C. J. Science 1996, 274, 1672.
2. Fischer, P. M.; Gianella-Borradori, A. Exp. Opin. Invest. Drugs 2005, 14, 457.
3. McInnes, C. Drug Discov. Today 2008, 13, 875.
4. Shapiro, G. I. J. Clin. Oncol. 2006, 24, 1770.
5. Harbour, J. W.; Luo, R. X.; Dei Santi, A.; Postigo, A. A.; Dean, D. C. Cell 1999, 98,
859.
6. Kitagawa, M.; Higashi, H.; Suzuki-Takahashi, I.; Segawa, K.; Hanks, S. K.; Taya,
Y.; Nishimura, S.; Okuyama, A. Oncogene 1995, 10, 229.
7. Chen, Y. N.; Sharma, S. K.; Ramsey, T. M.; Jiang, L.; Martin, M. S.; Baker, K.;
Adams, P. D.; Bair, K. W.; Kaelin, W. G., Jr. Proc Natl Acad Sci U S A 1999, 96,
4325.
8. Brown, N. R.; Noble, M. E. M.; Endicott, J. A.; Johnson, L. N. Nat. Cell Biol. 1999, 1,
438.
9. McInnes, C. The REPLACE strategy for generating non-ATP competitive
inhibitors of cell-cycle protein kinases. In Protein-protein interactions in drug
discovery; Mannhold, R., Ed.; Methods and Principles in Medicinal Chemistry;
Wiley-VCH, 2013; pp 291–304.
10. Kontopidis, G.; Andrews, M. J.; McInnes, C.; Plater, A.; Innes, L.; Renachowski,
S.; Cowan, A.; Fischer, P. M. ChemMedChem 2009, 4, 1120.
11. Kontopidis, G.; Andrews, M. J.; McInnes, C.; Cowan, A.; Powers, H.; Innes, L.;
Plater, A.; Griffiths, G.; Paterson, D.; Zheleva, D. I.; Lane, D. P.; Green, S.;
Walkinshaw, M. D.; Fischer, P. M. Structure 2003, 11, 1537.
12. Premnath, P. N.; Craig, S. N.; Liu, S.; Anderson, E. L.; Grigoroudis, A. I.;
Kontopidis, G.; Perkins, T. L.; Wyatt, M. D.; Pittman, D. L.; McInnes, C. J. Med.
Chem. 2015, 58, 433.
IC50 = 228.66 56.5 lM, DU145 IC50 = 204.5 34.64 lM). Another
FLIP from the BA3 series was shown to have activity in the binding
assay but also to have micromolar inhibitory activity in the MTT
13. Premnath, P. N.; Liu, S.; Perkins, T.; Abbott, J.; Anderson, E.; McInnes, C. Bioorg.
Med. Chem. 2014, 22, 616.
assay (compound 17, U2OS IC50 = 153
9 lM) demonstrating weak
but consistent growth inhibition. Interestingly the two other FLIPS
in the BA3 series synthesized were inactive in both the binding and
cellular assays and therefore correlating well in this context. While
the cellular activities are weak, they suggest that further optimiza-
tion of potency and drug-like properties will enable facilitate
promising compounds.
In conclusion, application of the REPLACE strategy to the CDK/
cyclin groove in order to block substrate recruitment resulted in
the elaboration of the 4-substituted benzamide scaffold as a
replacement for the N-terminal tetrapeptide of the HAKRRIF, a
sequence with optimized affinity. FLIPs incorporating derivatized
3,4 bis substituted benzamide groups were identified with low
micromolar activity for both CDK2A and CDK4D with these com-
pounds more completely mimicking the interactions of CGI pep-
tides. When specific FLIPs incorporating these Ncaps were
appended to a peptidomimetic sequence previously shown to
14. Liu, S.; Premnath, P. N.; Bolger, J. K.; Perkins, T. L.; Kirkland, L. O.; Kontopidis,
G.; McInnes, C. J. Med. Chem. 2013, 56, 1573.
15. McInnes, C.; Estes, K.; Baxter, M.; Yang, Z.; Farag, D. B.; Johnston, P.; Lazo, J. S.;
Wang, J.; Wyatt, M. D. Mol. Cancer Ther. 2012, 11, 1683.
16. Andrews, M. J.; Kontopidis, G.; McInnes, C.; Plater, A.; Innes, L.; Cowan, A.;
Jewsbury, P.; Fischer, P. M. ChemBioChem 2006, 7, 1909.
17. Castanedo, G.; Clark, K.; Wang, S.; Tsui, V.; Wong, M.; Nicholas, J.;
Wickramasinghe, D.; Marsters, J. C., Jr.; Sutherlin, D. Bioorg. Med. Chem. Lett.
2006, 16, 1716.
18. Potashman, M. H.; Bready, J.; Coxon, A.; DeMelfi, T. M., Jr.; DiPietro, L.; Doerr,
N.; Elbaum, D.; Estrada, J.; Gallant, P.; Germain, J.; Gu, Y.; Harmange, J. C.;
Kaufman, S. A.; Kendall, R.; Kim, J. L.; Kumar, G. N.; Long, A. M.; Neervannan, S.;
Patel, V. F.; Polverino, A.; Rose, P.; Plas, S.; Whittington, D.; Zanon, R.; Zhao, H. J.
Med. Chem. 2007, 50, 4351.
19. Solladié, G.; Pasturel-Jacopé, Y.; Maignan, J. Application to the synthesis of aryl
cinnamic acids Tetrahedron 2003, 59, 3315.
20. Imamura, S.; Ichikawa, T.; Nishikawa, Y.; Kanzaki, N.; Takashima, K.; Niwa, S.;
Iizawa, Y.; Baba, M.; Sugihara, Y. J. Med. Chem. 2006, 49, 2784.
21. Picard, F.; Barassin, S.; Mokhtarian, A.; Hartmann, R. W. J. Med. Chem. 2002, 45,
3406.
Please cite this article in press as: Premnath, P. N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.05.067