Development of Bifunctional Inhibitors of Polo-Like Kinase 1 with Low-Nanomolar Activities Against the Polo-Box Domain

Article abstract of DOI:10.1002/cbic.201500535

Polo-like kinase 1 (Plk1), a validated cancer target, harbors a protein-protein interaction domain referred to as the polo-box domain (PBD), in addition to its enzymatic domain. Although functional inhibition either of the enzymatic domain or of the PBD has been shown to inhibit Plk1, so far there have been no reports of bifunctional agents with the potential to target both protein domains. Here we report the development of Plk1 inhibitors that incorporate both an ATP-competitive ligand of the enzymatic domain, derived from BI 2536, and a functional inhibitor of the PBD, based either on the small molecule poloxin-2 or on a PBD-binding peptide. Although these bifunctional agents do not seem to bind both protein domains simultaneously, the most potent compound displays low-nanomolar activity against the Plk1 PBD, with excellent selectivity over the PBDs of Plk2 and Plk3. Our data provide insights into challenges and opportunities relating to the optimization of Plk1 PBD ligands as potent Plk1 inhibitors.

Full text of DOI:10.1002/cbic.201500535

DOI: 10.1002/cbic.201500535
Full Papers
Development of Bifunctional Inhibitors of Polo-Like
Kinase 1 with Low-Nanomolar Activities Against the Polo-
Box Domain
Andrej Scharow,[a] Daniel Knappe,[b] Wolfgang Reindl,[c, d] Ralf Hoffmann,[b] and
Thorsten Berg*[a]
Polo-like kinase 1 (Plk1), a validated cancer target, harbors a
protein–protein interaction domain referred to as the polo-box
domain (PBD), in addition to its enzymatic domain. Although
functional inhibition either of the enzymatic domain or of the
PBD has been shown to inhibit Plk1, so far there have been no
reports of bifunctional agents with the potential to target both
protein domains. Here we report the development of Plk1 in-
hibitors that incorporate both an ATP-competitive ligand of
the enzymatic domain, derived from BI 2536, and a functional
inhibitor of the PBD, based either on the small molecule polox-
in-2 or on a PBD-binding peptide. Although these bifunctional
agents do not seem to bind both protein domains simultane-
ously, the most potent compound displays low-nanomolar ac-
tivity against the Plk1 PBD, with excellent selectivity over the
PBDs of Plk2 and Plk3. Our data provide insights into challeng-
es and opportunities relating to the optimization of Plk1 PBD
ligands as potent Plk1 inhibitors.
Introduction
The serine/threonine kinase polo-like kinase (Plk1) is a key reg-
ulator of mitosis.[1] It is upregulated in numerous tumors and
has been identified as a negative prognostic marker for tumor
patients.[2] Because tumor cells are thought to be more depen-
dent on Plk1 than non-transformed cells, much effort has been
directed to the development of small-molecule inhibitors of
Plk1 enzyme activity.[3] One of the most advanced inhibitors of
the enzyme’s ATP binding sites is the compound BI 2536
(Figure 1, below),[4] which has been investigated in clinical
trials.[5] BI 2536 displays low-nanomolar potency against the
kinase domain of Plk1 (IC50 =0.83 nm), but is also active against
the kinase domains of Plk2 (IC50 =3.5 nm) and Plk3 (IC50 =
9.0 nm).[4b] Activity profiling of BI 2536 by a chemical proteo-
mics approach identified additional kinases inhibited by
BI 2536, including death-associated protein kinases (DAPKs)
and calcium/calmodulin-dependent protein kinase kinase 2
(CAMKK2).[6] This illustrates the key problem of ATP-competi-
tive protein kinase inhibitors: because of the large number of
protein kinases (ca. 500) and the presence of other ATP hydro-
lases, all of which contain conserved ATP binding pockets, in
the proteome, even highly optimized compounds suffer from
limited specificity against other kinases.[7]
In the light of these considerations, targeting the polo-box
domain (PBD) has emerged as a promising alternative ap-
proach to the inhibition of Plk1.[8] The Plk1 PBD mediates pro-
tein–protein interactions of Plk1, enabling correct localization
of Plk1 and docking of the enzyme to a subset of its sub-
strates. To date, only four proteins are known to possess a PBD
(Plk1, Plk2, Plk3, and Plk5).[9] This places Plk1 PBD inhibitors in
a privileged position in terms of specificity. Functional inhibi-
tion of the Plk1 PBD by small-molecule agents has been
shown to induce mitotic arrest and apoptosis in tumor cells,[10]
and so has been validated as a useful approach to Plk1 inhibi-
tion. We presented the natural product derivative poloxin[10a]
as the first small-molecule inhibitor of the Plk1 PBD, and re-
cently reported on its development to the optimized analogue
poloxin-2, displaying superior activity and selectivity for the
Plk1 PBD.[11] Additional demonstration of the suitability of the
Plk1 PBD as a target for Plk1 inhibition has been provided by
peptide-based agents,[12] many of which display low-nanomolar
activities against the Plk1 PBD and high selectivities over the
PBDs of Plk2 and Plk3.[13]
[a] Dr. A. Scharow, Prof. Dr. T. Berg
Leipzig University, Institute of Organic Chemistry
Johannisallee 29, 04103 Leipzig (Germany)
E-mail: tberg@uni-leipzig.de
[b] Dr. D. Knappe, Prof. Dr. R. Hoffmann
Leipzig University, Institute of Bioanalytical Chemistry
Center for Biotechnology and Biomedicine (BBZ)
Deutscher Platz 5, 04103 Leipzig (Germany)
[c] Dr. W. Reindl
Max Planck Institute of Biochemistry, Department of Molecular Biology
Am Klopferspitz 18, 82152 Martinsried (Germany)
[d] Dr. W. Reindl
Present address: Evotec AG, Manfred Eigen Campus
Essener Bogen 7, 22419 Hamburg (Germany)
Here we propose a new concept for the design of Plk1 in-
hibitors offering the potential for particularly high activity and
selectivity for Plk1. We present synthetic methodology for the
generation of bifunctional Plk1-targeting agents incorporating
both an inhibitor of its PBD and a ligand of the Plk1 ATP bind-
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201500535.
This manuscript is part of a joint Special Issue between ChemBioChem and
ChemMedChem on Protein–Protein Interactions.
ChemBioChem 2016, 17, 759 – 767
759
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ing site. We designed two classes of target molecules, based
either on a peptidic ligand of the Plk1 PBD or on a non-pepti-
dic functional inhibitor of the Plk1 PBD (Figure 1). The ATP-
competitive Plk1 ligand BI 2536[4b] was incorporated into both
classes of target molecules (Figure 1A). The X-ray structure of
the Plk1 kinase domain in complex with BI 2536 shows that
the methylpiperidino moiety protrudes from the binding
pocket (Figure 1B),[14] and this suggests this part of the mole-
cule as a suitable attachment point for the placement of
a linker molecule. For the peptide-ligand-based molecules, we
chose the peptide sequence PLHSpTA (pT=phosphothreonine,
Thr(PO)(OH)2), which has been reported as a selective Plk1 PBD
ligand.[15] The X-ray structure of the Plk1 PBD in complex with
the peptide Ac-PLHSpT demonstrates that the N terminus of
the peptide extends beyond the PBD (Figure 1C and D).[15]
Therefore, we envisioned the synthesis of bivalent molecules
in which the ATP-competitive ligand would be coupled as a car-
boxylic acid to the N terminus of the peptide sequence (Fig-
ure 1E). For the non-peptide-based inhibitors, we chose the
small molecule poloxin-2 (Figure 1C, E).[11] Because it has
proved challenging to obtain structural information about the
binding site of poloxin[16] and its derivatives,[11] a large range of
linker lengths had to be explored in order to increase the
chances of achieving simultaneous binding of both functional
parts of the molecule to Plk1.
ester 2, which was subjected to reductive amination conditions
in the presence of cyclopentanone, providing the secondary
amine 3 (Scheme 1). Nucleophilic aromatic substitution of 2,4-
dichloro-5-nitropyrimidine with 3 yielded 4, which was cyclized
to the dihydropteridinone 5 by treatment with Fe in acetic
acid. N-Methylation of 5 provided 6, which was converted into
7 by treatment with 4-amino-3-methoxybenzoic acid.[17] Amide
coupling between 7 and 4-amino-1-Boc-piperidine provided 8.
N-Deprotection of 8 afforded the amine 9 (overall yield 22%
over eight steps), which served as the BI 2536-based building
block for the generation of poloxin-2-based bifunctional Plk1
inhibitors. Further modification of 9 with 8-bromooctanoyl
benzoate (10) afforded 11, which was debenzylated by catalyt-
ic hydrogenolysis to afford 12 (overall yield 14% over eleven
steps). Compound 12 was used as the BI 2536-based building
block for the synthesis of the peptide-based bifunctional Plk1
ligands.
Synthesis of the building block 20, based on the Plk1 PBD
inhibitor poloxin-2, was carried out in a six-step procedure
(Scheme 2). 4-Bromo-2-methylbenzoic acid (13) was converted
into its methyl ester 14 and subsequently into the nitrile 15.
Reduction of the nitrile function by hydrogenolysis afforded
the amine 16, which was subsequently coupled to pent-4-
ynoic acid. After ester hydrolysis of 17, the acid 18 was cou-
pled to the oxime 19, providing the desired building block 20
in 18% total yield.
The three linker molecules 24, 28, and 30 were designed for
connecting the poloxin-2-based alkyne 20 and the BI 2536-
derived amine 9 (Scheme 3). The chemical nature of the linker
molecules was guided by our aim to keep the polarity of the
target molecules low to facilitate their purification on silica,
Results and Discussion
The building blocks based on the ATP-competitive ligand
BI 2536 were prepared by starting with the reported synthe-
sis.[17] d-2-Aminobutyric acid (1) was converted into its methyl
Figure 1. Design principle of bifunctional Plk1 inhibitors. A) ATP-competitive ligands such as BI 2536 bind to and inhibit the Plk1 kinase domain (KD). B) Bind-
ing of BI 2536 to the kinase domain of Plk1, based on PBD entry 2RKU.[14] C) Both poloxin-2 and the peptide PLHSpT are functional inhibitors of the polo-box
domain. D) Binding mode of Ac-PLHSpT, based on PBD entry 3HIK.[15] E) Bifunctional ligands incorporating ligands of both the kinase domain and the PBD
have the potential to inhibit the functions of both domains simultaneously. General structures of the target compounds based on poloxin-2 and the PBD-
binding peptide are shown.
ChemBioChem 2016, 17, 759 – 767
www.chembiochem.org
760
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 1. Synthesis of the building blocks 9 and 12 based on BI 2536, a ligand of the ATP-binding-pocket ligand of Plk1. a) SOCl2, MeOH, 08C to reflux,
90 min (quant.); b) cyclopentanone, NaOAc, NaB(OAc)3H, CH2Cl2, RT, 16 h (quant.); c) 2,4-dichloro-5-nitropyrimidine, K2CO3, acetone, 08C to RT, 16 h (54%);
d) Fe powder, HOAc, 70–1108C, 3.75 h (52%); e) DMA, NaH, MeI, À108C to RT, 90 min (quant.); f) 4-amino-3-methoxybenzoic acid, EtOH, H2O, HClconc, reflux,
48 h (85%); g) DMF, HBTU, DIA, RT, 30 min; h) 4-amino-1-Boc-piperidine, RT, 3 h (92%); i) TFA, CH2Cl2, 08C, 3 h; j) K2CO3, MeOH, 08C, 10 min (quant.); k) BnBr,
DMF, K2CO3, RT, 16 h (90%); l) DMF, K2CO3, RT, 36 h (63%); m) EtOH, H2, Pd/C, RT, 30 min (quant.).
whilst exploiting the reliability of amide coupling chemistry.
Synthesis of the linkers was carried out by use of standard cou-
pling techniques in total yields of 65% (for 24), 67% (for 28),
and 29% (for 30).
34), 37 (compound 35), or 49 (compound 36) CÀC or CÀN
bonds, corresponding to maximum distances of approximately
31 to 62 Š in an extended conformation. The synthesis of
longer linkers was hampered by insufficient solubility. The opti-
mal linker length is not known, because there is at present no
available crystal structure of full-length Plk1 from which the
distance between the ligands of the two domains might be
assessed. In addition, the binding site of poloxin and its deriva-
tives on the PBD has not been identified.
Coupling of the two functional parts of the molecule
through the linkers was carried out in a two-step synthesis.
Firstly, the BI 2536-derived amine 9 was coupled with the bro-
moazides 24, 28, and 30, to provide the azides 31, 32, and 33,
respectively (Scheme 4). Subsequently, the alkyne-bearing po-
loxin-2 derivative 20 was coupled to the azides, resulting in
the final bifunctional ligands 34–36. In these bifunctional mol-
ecules, the piperidine nitrogen atom (located at the surface of
the ATP-binding pocket) and the poloxin-2 derivative’s ester
carbonyl group (thought to be attacked by a nucleophilic
amino acid in the Plk1 PBD) are separated by 25 (compound
For the synthesis of the peptide-based building block, the
known PBD ligand PLHSpTA[15] was synthesized by the Fmoc/
tBu-strategy on Tentagel R PHB resin (Scheme 5). A glycine resi-
due was added to the N terminus to facilitate coupling of link-
ers consisting of one, five, 12, and 20 ethylene glycol units.
Ethylene-glycol-based linkers were chosen for the peptide-
ChemBioChem 2016, 17, 759 – 767
www.chembiochem.org
761
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 2. Synthesis of the alkyne-functionalized poloxin-2 derivative 20. a) SOCl2, MeOH, 08C to 508C, 14 h, 96%; b) Zn(CN)2, Pd(PPh3)4, DMF, 1008C, 2 h,
88%; c) H2, Pd/C, MeOH, RT, 12 h, 36%; d) pent-4-ynoic acid, DCC, DMAP, CH2Cl2, RT, 14 h, 72%; e) 1m NaOH, THF, RT, 14 h, 93%; f) DCC, DMAP, CH2Cl2, RT,
14 h, 87%.
Scheme 3. Synthesis of the linker molecules 24, 28, and 30. a) NaN3, H2O, reflux, 7 h; b) KOH, À158C, 10 min; c) EDC, HOBt, DIPEA, CH2Cl2, RT, 14 h; d) 22,
EDC, HOBt, DIPEA, RT; e) TFA, CH2Cl2, 08C, 2.5 h; f) 23, EDC, HOBt, DIPEA, CH2Cl2, RT, 14 h, 81%; g) 25, EDC, HOBt, DIPEA, CH2Cl2, RT, 14 h, 65%; h) 4m HCl,
MeOH, RT, 1 h, 99%; i) 23, EDC, HOBt, DIPEA, CH2Cl2, RT, 14 h, 54%.
based bifunctional compounds to facilitate compound purifica-
tion by reversed-phase HPLC and to promote high solubility of
the final products in aqueous buffer. The BI 2536 derivative 12
was coupled to the amino group of the linkers. After cleavage
from the resin, HPLC purification afforded the target com-
pounds 37a–d in high purities. Depending on the linker
lengths, the distances between the nitrogen atom of the piper-
idine moiety and the carboxylic acid function of the glycine
residue range from 18 CÀC/CÀN bonds (37a) to 75 bonds
(37d) CÀC/CÀN, corresponding to maximum distances of ap-
proximately 23 to 95 Š in an extended conformation.
Analysis of the biological activities of the bifunctional com-
pounds was carried out in a competitive binding assay for the
Plk1 PBD based on fluorescence polarization, with full-length
Plk1. Simultaneous binding of the bifunctional ligands both to
the PBD and to the ATP-binding pocket should manifest itself
in increased inhibitory activity against the Plk1 PBD, relative to
the corresponding PBD ligand alone. However, the bifunctional
ligands 34–36 were less active than the poloxin-2-based build-
ing block 20 (Figure 2A), thus suggesting that the BI 2536-de-
rived part of the molecule does not interact with its target site
when the PBD inhibitor binds to the PBD. The alkyne function-
ChemBioChem 2016, 17, 759 – 767
www.chembiochem.org
762
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 4. Synthesis of the poloxin-2-based bifunctional ligands 34–36. a) K2CO3, DMF, RT, 24 h; b) 20, CuI, DIPEA, HOAc, CH2Cl2, RT, 2.5–14 h.
ality in 20 did not negatively affect the inhibitory activity
against the Plk1 PBD, because 20 displayed inhibitory activity
similar to that of the corresponding poloxin derivative 38 (Fig-
ure S1 in the Supporting Information).[11] One potential explan-
ation for the observation that the activity of 34–36 against the
Plk1 PBD is lower than the activity of the control compound
20 is an increased entropic penalty upon binding of 34–36 to
Plk1 through their PBD targeting moiety, which, in the absence
of simultaneous binding of the BI 2536-derived part of the
molecules, is not compensated by an enthalpic gain.
of PBD ligands against the Plk1 PBD. Simultaneous titration of
the peptide Ac-GPLHSpTA-NH2 in the presence of 160 nm
BI 2536 was less effective than titration with the peptide alone
(IC50 =567Æ80 nm). This observation can be explained by the
concept of autoinhibition, by which the catalytic domain and
the polo-box domain bind to and thereby inhibit one anoth-
er.[8b] Addition of BI 2536 is expected to relieve inhibition of
the Plk1 PBD by the enzyme’s active site, rendering the PBD
more susceptible to binding of ligands. Consistently, BI 2536
increased binding between full-length Plk1 and 5-carboxyfluor-
escein-GPMQSpTPLNG (Figure S2). It is currently not under-
stood why the relief of PBD inhibition by the kinase domain in-
creases the affinity of the PBD for the fluorophore-labeled pep-
tide 5-carboxyfluorescein-GPMQSpTPLNG by a larger extent
than for the peptide Ac-GPLHSpTA-NH2. However, the data
exclude the possibility that the increased activities of 37a–d
for the PBD are merely caused by relief of autoinhibition in the
presence of an ATP binding site ligand.
In contrast to the poloxin-based molecules 34–36, analysis
of the peptide-based bifunctional molecules 37 against full-
length Plk1 revealed that all bifunctional compounds 37a–d
displayed significantly higher activities than the control pep-
tide Ac-GPLHSpTA-NH2 (Figure 2B and Table 1). Notably, the ac-
tivities of compounds 37a–d increased with decreasing linker
length, which argues against the possibility of nonspecific in-
teractions between the linker and the protein as the driving
force for activity. The shortest bifunctional molecule 37a
(IC50 =38Æ2 nm) was an order of magnitude more active than
the peptide Ac-GPLHSpTA-NH2 (IC50 =369Æ15 nm) in terms of
IC50 values, representing a more than 25-fold increase in terms
of the inhibition constants [Ki (37a)=6Æ1 nm; Ki (Ac-
GPLHSpTA-NH2)=166Æ8 nm]. Thus, fusion of the ATP-compet-
itive moiety to the peptide is a valid means to improve activity
To analyze whether the improved binding of the bifunctional
ligands 37a–d to the Plk1 PBD was caused by simultaneous
binding to the ATP-binding site, we carried out competition
assays with the bifunctional ligands 37a–d at a fixed concen-
tration of 160 nm, corresponding to 60–80% inhibition of the
PBD, together with varying concentrations of BI 2536. If simul-
taneous bivalent binding of 37a–d to both protein domains of
ChemBioChem 2016, 17, 759 – 767
www.chembiochem.org
763
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 5. Synthesis of bivalent compounds 37a–d.
Plk1 occurs, titration of free BI 2536 should decrease the activi-
ties of 37a–d to a significant extent. However, in the presence
of up to 5 mm of BI 2536, only a minor reduction in activity
against the Plk1 PBD was observed (Figure S3). Because Ac-
GPLHSpTA-NH2, the peptide part of 37a–d, is also slightly less
active in the presence of BI 2536 (Figure 2 and Table 1), these
data suggest that the bifunctional agents 37a–d were not si-
multaneously targeting the ATP-binding site and the Plk1 PBD.
To interrogate the binding mode of 37a–d in more detail,
we analyzed their effects on the enzymatic activity of full-
length Plk1. Simultaneous inhibition both of the PBD and of
the kinase domain should lead to an increased inhibitory activ-
ity of the bivalent molecules against the enzymatic activity of
Plk1 and/or improved selectivity. To analyze this question, we
tested the ability of Plk1, Plk2, and Plk3 to phosphorylate a suit-
able peptide substrate on a specific serine residue in the pres-
ence of the test compounds. The control compound BI 2536
inhibited the enzymatic activity of Plk1 with an IC50 of 0.78Æ
0.06 nm (Table 2 and Figure S4). Plk2 and Plk3 were inhibited
to lesser extents [IC50 (Plk2): 5.9Æ0.1 nm; IC50 (Plk3): 13.8Æ
0.1 nm], corresponding to selectivity factors of 7 and 17 for
Plk1 against Plk2 and Plk3, respectively. The bifunctional mole-
cules 37a–d displayed slightly reduced activities against Plk1
relative to the reference compound BI 2536, which are likely to
be caused by the attachment of the linker groups. The activi-
ties of the bifunctional molecules against the enzymatic func-
tion of Plk2 and Plk3 were reduced to a similar extent as their
activities against Plk1, resulting in specificity profiles similar to
that of the control compound BI 2536. These data argue
against simultaneous binding of both functional parts of 37a–
d to the enzymes’ ATP binding site and the PBD. However, the
possibility remains that binding of 37a–d might be bivalent
after all, but the large number of rotatable bonds between the
peptide sequence and the BI 2536-derived part might prevent
an additive or synergistic effect of simultaneous binding to
both protein domains.
Activity analysis of the most potent agent 37a against the
isolated Plk1 PBD in a binding assay based on fluorescence
polarization[19] also revealed high potency (IC50 =0.054Æ
0.004 mm, Figure 3 and Table S1), very similar to that obtained
against full-length Plk1 (IC50 =0.038Æ0.002 mm, Figure 2C and
Table 1). In addition, compound 37a was found to display
more than 400-fold selectivity for the Plk1 PBD (Ki =0.012Æ
0.001 mm) over the Plk2 PBD (Ki =5.5Æ0.7 mm) and more than
300-fold selectivity over the Plk3 PBD (Ki =3.7Æ0.2 mm).[20] The
control peptide Ac-GPLHSpTA-NH2 was also found to be highly
selective for the Plk1 PBD (Figure S5 and Table S1), confirming
the high selectivity of the peptide motif PLHSpTA for the Plk1
PBD as reported in the literature.[15]
The improved activities of the bifunctional agents 37a–d
against the PBD of full-length Plk1 (Figure 2C, Table 1) and of
37a against the isolated PBD (Figure 3, Table S1) might be ex-
ChemBioChem 2016, 17, 759 – 767
www.chembiochem.org
764
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 2. Activities of the bivalent agents 37a–d and BI 2536 against the
enzymatic function of full-length Plk1-3.
IC50 [nm]
Plk2
Selectivity factor for
Plk1 over Plk2/Plk3
Cmpd
Plk1
Plk3
BI 2536
37a
37b
37c
37d
0.78Æ0.06
1.95Æ0.43
1.84Æ0.42
2.62Æ0.41
3.53Æ0.38
5.9Æ0.1
13.0Æ1.0
18.7Æ1.5
23.4Æ1.1
39.0Æ2.7
13.8Æ0.1
26.0Æ0.5
41.9Æ0.5
61.2Æ1.2
80.8Æ4.0
8/18
7/13
10/23
9/23
11/23
Figure 2. Activities of bivalent ligands against the PBD of full-length Plk1 as
analyzed in fluorescence polarization (FP) assays. A) Principle of the assay.
B) Activities of poloxin-2-based ligands 34–36 relative to the control com-
pound 20. C) Activities of peptide-based ligands 37a–d relative to the PBD-
binding peptide sequence GPLHSpTA.
Figure 3. The bifunctional peptide 37a selectively inhibits the function of
the Plk1 PBD as analyzed by fluorescence polarization assay. A) Principle of
the assay. The protein constructs used in this assay contain the PBD of the
corresponding Plk only. B) Activity of 37a against the PBDs of Plk1, Plk2, and
Plk3.
required to clarify the binding mode of the bivalent com-
Table 1. IC50 values against the function of the PBD in full-length Plk1
and inhibition constants (Ki) analyzed in a binding assay based on fluores-
cence polarization. Conversion of IC50 values into Ki values was carried
out by use of the published equation.[18]
pounds 37a–d.
Conclusion
Compound
IC50 Plk1 [nm]
Ki Plk1 [nm]
In this study we present a synthetic methodology for the gen-
eration of two classes of bifunctional agents targeting both
protein domains of Plk1. The bifunctional agents 34–37 each
consist of a chemical moiety derived from the ATP-binding-
pocket ligand BI 2536, and either a peptidic or a poloxin-2-
based inhibitor of the Plk1 PBD (Figure 1). Although 34–37 do
not appear to bind to both of the two protein domains simul-
taneously, fusion of BI 2536 to the PBD-binding peptide motif
GPLHSpTA leads to agents with significantly improved activi-
ties against the Plk1 PBD. The most potent agent 37a displays
low-nanomolar activity both against the PBD of full-length Plk1
and against the isolated Plk1 PBD, together with more than
400-fold selectivity over the Plk2 PBD, and more than 300-fold
selectivity over the Plk3 PBD. Once the crystal structure of full-
length Plk1 becomes available, after which the length and
Ac-GPLHSpTA-NH2
37a
37b
37c
369Æ15
38Æ2
166Æ8
6Æ1
57Æ13
81Æ8
15Æ6
27Æ4
37Æ5
262Æ39
37d
102Æ11
567Æ80
Ac-GPLHSpTA-NH2 +160 nm BI 2536
plained by binding of the linker and/or the aliphatic chain of
37a–d to a hydrophobic channel adjacent to the peptide-bind-
ing groove of the Plk1 PBD. Occupation of such channels has
previously been reported for peptides bearing hydrophobic
motifs attached to the side chains of the histidine,[13a] and also
to the proline[13c] residue of the peptide sequence PLHSpT,
which is also contained in 37a–d. Structural analysis would be
ChemBioChem 2016, 17, 759 – 767
www.chembiochem.org
765
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
The synthesis of poloxin-2-based molecules 34–36 is described in
the Supporting Information.
chemical nature of a suitable linker should be ascertainable,
our methodology can be utilized for the rational design of bi-
functional Plk1 inhibitors displaying simultaneous binding of
both ligands to their associated protein domains. Because
many protein kinases harbor functional domains in addition to
their enzymatic domain, and simultaneous targeting of two
domains is a means by which to overcome specificity problems
of ATP-competitive kinase inhibitors, we expect our study to
stimulate further research directed towards the development
of bifunctional inhibitors of protein kinases.
Peptide synthesis: Peptides were synthesized by the Fmoc/tBu-
strategy as C-terminal amides on Tentagel R PHB resin
(0.18 mmolgÀ1) with the aid of a multiple synthesizer (SYRO2000,
MultiSynTech). Amino acid derivatives (10 equiv) were coupled by
use of HBTU (10 equiv) and DIPEA (24 equiv). PEG1 and PEG5
(8 equiv), PEG12 and PEG20 (4 equiv), and compound 12 (6 equiv)
were coupled manually by in situ activation with DIC/HOBt
(8 equiv). Peptides were cleaved with TFA containing 12.5% (v/v)
of a scavenger mixture (ethane-1,2-dithiol/m-cresol/thioanisole/
water 1:2:2:2, v/v/v/v) for 2 h, precipitated with cold diethyl ether,
and dried. Purification was by RP-HPLC with a Jupiter C18 column
(10”250 mm) with use of a linear gradient of aqueous acetonitrile
in the presence of TFA (0.1%) as ion pair reagent. RP-HPLC was
used to evaluate peptide purities with use of a Poroshell SB-C18
column (2”100 mm) and typically a linear gradient from 3 to 57%
acetonitrile in 30 min and a column temperature of 608C. Peptides
were detected by absorbance at 214 nm. Monoisotopic masses
were determined by matrix-assisted laser desorption/ionization
time-of-flight mass spectrometry (MALDI-TOF/TOF-MS; 4700 pro-
teomic analyzer; Applied Biosystems, GmbH, Darmstadt, Germany)
with a-cyano-4-hydroxycinnamic acid [CHCA; 4 gLÀ1 in aqueous
acetonitrile (60%) containing TFA (0.1%)] as matrix.
Experimental Section
Protein expression and purification: Expression and purification
of the PBDs of Plk1, Plk2, and Plk3 has been described.[10a,b,19–20]
Full-length Plk1 for use in binding assays based on fluorescence
polarization was expressed from Sf9 cells.
Binding assays based on fluorescence polarization: Binding
assays for the PBDs of Plk1, Plk2, and Plk3 were performed essen-
tially as described.[10a,b,19,20] In brief, proteins were incubated with
the poloxin-2-based compounds at the indicated concentrations
for 60 min. Subsequently, 5-carboxyfluorescein-labeled peptides
(final concentration 10 nm) were added, and fluorescence polariza-
tion was analyzed after another 60 min by use of a Tecan Infini-
te F500 plate reader. The following carboxyfluorescein-labeled (CF-
labeled) peptides were used: 5-CF-GPMQSpTPLNG-OH for Plk1, 5-
CF-GPMQTSpTPKNG-OH for Plk2, and 5-CF-GPLATSpTPKNG-OH for
Plk3. Compounds were tested at the following final concentrations
of buffer components: NaCl (50 mm), Tris (pH 8.0, 10 mm), EDTA
(1 mm), Nonidet P-40 substitute (0.1%, v/v), and DMSO (2%, v/v).
Proteins were used at the following concentrations corresponding
to their Kd values: 20 nm for Plk1 PBD and 50 nm for full-length
Plk1 from Sf9 cells. Peptide-based compounds were tested as
described above, but with additional dithiothreitol (1 mm) in the
assay buffer, and were analyzed after addition of the carboxyfluor-
escein-labeled peptides. Proteins were used at the following con-
centrations: 15 nm for Plk1 PBD, 60 nm for Plk2 PBD, 200 nm for
Plk3 PBD, and 40 nm for full-length Plk1 from Sf9 cells. All experi-
ments were performed in triplicate. Inhibition curves were fitted
with SigmaPlot (SPSS). Conversion of IC50 values into Ki values was
carried out by use of the published equation.[18]
Acknowledgements
Work in the group of T.B. was generously supported by the Deut-
sche Forschungsgemeinschaft (BE 4572/1–1) and the Department
of Molecular Biology at the Max Planck Institute of Biochemistry.
We extend our thanks to Angela Berg for critical reading of the
manuscript.
Keywords: bioorganic chemistry
· inhibitors · peptides ·
protein kinases · protein–protein interactions
[1] M. Petronczki, P. LØnµrt, J. M. Peters, Dev. Cell 2008, 14, 646–659.
[2] R. Knecht, C. Oberhauser, K. Strebhardt, Int. J. Cancer 2000, 89, 535–
536.
[3] K. Strebhardt, Nat. Rev. Drug Discovery 2010, 9, 643–660.
[4] a) P. LØnµrt, M. Petronczki, M. Steegmaier, B. Di Fiore, J. J. Lipp, M. Hoff-
mann, W. J. Rettig, N. Kraut, J. M. Peters, Curr. Biol. 2007, 17, 304–315;
b) M. Steegmaier, M. Hoffmann, A. Baum, P. Lenart, M. Petronczki, M.
Krssak, U. Gurtler, P. Garin-Chesa, S. Lieb, J. Quant, M. Grauert, G. R.
Adolf, N. Kraut, J. M. Peters, W. J. Rettig, Curr. Biol. 2007, 17, 316–322.
[5] P. Schoffski, Oncologist 2009, 14, 559–570.
[6] M. Raab, F. Pachl, A. Krämer, E. Kurunci-Csacsko, C. Dçtsch, R. Knecht, S.
Becker, B. Küster, K. Strebhardt, Cell. Res. 2014, 24, 1141–1145.
[7] H. Daub, K. Specht, A. Ullrich, Nat. Rev. Drug Discovery 2004, 3, 1001–
1010.
[8] a) A. E. Elia, L. C. Cantley, M. B. Yaffe, Science 2003, 299, 1228–1231;
b) A. E. Elia, P. Rellos, L. F. Haire, J. W. Chao, F. J. Ivins, K. Hoepker, D. Mo-
hammad, L. C. Cantley, S. J. Smerdon, M. B. Yaffe, Cell 2003, 115, 83–95.
[9] G. de Cµrcer, G. Manning, M. Malumbres, Cell Cycle 2011, 10, 2255–
2262.
[10] a) W. Reindl, J. Yuan, A. Krämer, K. Strebhardt, T. Berg, Chem. Biol. 2008,
15, 459–466; b) W. Reindl, J. Yuan, A. Krämer, K. Strebhardt, T. Berg,
ChemBioChem 2009, 10, 1145–1148; c) N. Watanabe, T. Sekine, M.
Takagi, J. Iwasaki, N. Imamoto, H. Kawasaki, H. Osada, J. Biol. Chem.
2009, 284, 2344–2353; d) H. M. Shan, Y. Shi, J. Quan, ChemMedChem
2015, 10, 158–163; e) G. Srinivasrao, J. E. Park, S. Kim, M. Ahn, C.
Cheong, K. Y. Nam, P. Gunasekaran, E. Hwang, N. H. Kim, S. Y. Shin, K. S.
Lee, E. Ryu, J. K. Bang, PLoS One 2014, 9, e107432.
Enzyme activity assays: The IC50 values of BI 2536 and compounds
37a–d against the enzymatic function of Plk1, Plk2, and Plk3 were
determined by use of the Z’-LYTE Ser/Thr 16 peptide kinase assay
kit (Invitrogen) according to the manufacturer’s instructions. The
kinase reaction buffer contained additional DTT (1 mm) and DMSO
(2%). Full-length enzymes were used at the following concentra-
tions: 1.25 ngmLÀ1 for Plk1, 2.5 ngmLÀ1 for Plk2, and 1.25 ngmLÀ1
for Plk3. The final concentration of the Z’-LYTE control phospho-
peptide (ELMEFpSLKDQEA) and the Z’-LYTE peptide (ELMEFSLKD-
QEA) was 2 mm. ATP was used at the following concentrations:
20 mm for Plk1, 60 mm for Plk2, and 20 mm for Plk3. The phosphory-
lation reaction was allowed to proceed for 3 h at room tempera-
ture, followed by addition of development reagent and incubation
for 1 h at room temperature. The reaction was then terminated
with use of the stop reagent. Fluorescence intensities of coumarin
and fluorescein moieties attached to the termini of the peptides
were measured with a Tecan Infinite F500 plate reader (lex =
405 nm, lem =430 and 525 nm). Inhibition curves were fitted by
use of SigmaPlot (SPSS). All experiments were performed in tripli-
cate.
ChemBioChem 2016, 17, 759 – 767
www.chembiochem.org
766
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[11] A. Scharow, M. Raab, K. Saxena, S. Sreeramulu, D. Kudlinzki, S. Gande, C.
Dotsch, E. Kurunci-Csacsko, S. Klaeger, B. Kuster, H. Schwalbe, K. Streb-
hardt, T. Berg, ACS Chem. Biol. 2015, 10, 2570–2579.
[12] S. Richter, I. Neundorf, K. Loebner, M. Graber, T. Berg, R. Bergmann, J.
Steinbach, J. Pietzsch, F. Wuest, Bioorg. Med. Chem. Lett. 2011, 21,
4686–4689.
[14] M. Kothe, D. Kohls, S. Low, R. Coli, G. R. Rennie, F. Feru, C. Kuhn, Y. H.
Ding, Chem. Biol. Drug Des. 2007, 70, 540–546.
[15] S. M. Yun, T. Moulaei, D. Lim, J. K. Bang, J. E. Park, S. R. Shenoy, F. Liu,
Y. H. Kang, C. Liao, N. K. Soung, S. Lee, D. Y. Yoon, Y. Lim, D. H. Lee, A.
Otaka, E. Appella, J. B. McMahon, M. C. Nicklaus, T. R. Burke, Jr., M. B.
Yaffe, A. Wlodawer, K. S. Lee, Nat. Struct. Mol. Biol. 2009, 16, 876–882.
[16] Z. Yin, Y. Song, P. H. Rehse, ACS Chem. Biol. 2013, 8, 303–308.
[17] a) G. Munzert, M. Steegmaier, US-2006/0074088A1, 2006; b) G. Budin,
K. S. Yang, T. Reiner, R. Weissleder, Angew. Chem. Int. Ed. 2011, 50, 9378–
9381; Angew. Chem. 2011, 123, 9550–9553.
[18] Z. Nikolovska-Coleska, R. Wang, X. Fang, H. Pan, Y. Tomita, P. Li, P. P.
Roller, K. Krajewski, N. G. Saito, J. A. Stuckey, S. Wang, Anal. Biochem.
2004, 332, 261–273.
[19] W. Reindl, K. Strebhardt, T. Berg, Anal. Biochem. 2008, 383, 205–209.
[20] W. Reindl, M. Gräber, K. Strebhardt, T. Berg, Anal. Biochem. 2009, 395,
189–194.
[13] a) F. Liu, J. E. Park, W. J. Qian, D. Lim, M. Graber, T. Berg, M. B. Yaffe, K. S.
Lee, T. R. Burke, Jr., Nat. Chem. Biol. 2011, 7, 595–601; b) F. Liu, J. E.
Park, W. J. Qian, D. Lim, A. Scharow, T. Berg, M. B. Yaffe, K. S. Lee, T. R.
Burke, Jr., ChemBioChem 2012, 13, 1291–1296; c) F. Liu, J. E. Park, W. J.
Qian, D. Lim, A. Scharow, T. Berg, M. B. Yaffe, K. S. Lee, T. R. Burke, Jr.,
ACS Chem. Biol. 2012, 7, 805–810; d) W. J. Qian, J. E. Park, D. Lim, C. C.
Lai, J. A. Kelley, S. Y. Park, K. W. Lee, M. B. Yaffe, K. S. Lee, T. R. Burke, Jr.,
Biopolymers 2014, 102, 444–455; e) P. Sledz, C. J. Stubbs, S. Lang, Y. Q.
Yang, G. J. McKenzie, A. R. Venkitaraman, M. Hyvonen, C. Abell, Angew.
Chem. Int. Ed. 2011, 50, 4003–4006; Angew. Chem. 2011, 123, 4089–
4092; f) P. Sledz, S. Lang, C. J. Stubbs, C. Abell, Angew. Chem. Int. Ed.
2012, 51, 7680–7683; Angew. Chem. 2012, 124, 7800–7803; g) M. Ahn,
Y. H. Han, J. E. Park, S. Kim, W. C. Lee, S. J. Lee, P. Gunasekaran, C.
Cheong, S. Y. Shin, Sr. , H. Y. Kim, E. K. Ryu, R. N. Murugan, N. H. Kim,
J. K. Bang, J. Med. Chem. 2015, 58, 294–304.
Manuscript received: October 14, 2015
Final article published: December 4, 2015
ChemBioChem 2016, 17, 759 – 767
www.chembiochem.org
767
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim