Structure-based design of novel human Pin1 inhibitors (I)

Article abstract of DOI:10.1016/j.bmcl.2009.08.034

Pin1 is a member of the cis-trans peptidyl-prolyl isomerase family with potential anti-cancer therapeutic value. Here we report structure-based de novo design and optimization of novel Pin1 inhibitors. Without a viable lead from internal screenings, we de

Full text of DOI:10.1016/j.bmcl.2009.08.034

Bioorganic & Medicinal Chemistry Letters 19 (2009) 5613–5616
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Structure-based design of novel human Pin1 inhibitors (I)
*
*
Chuangxing Guo , Xinjun Hou , Liming Dong, Eleanor Dagostino, Samantha Greasley, RoseAnn Ferre,
Joseph Marakovits, M. Catherine Johnson, David Matthews, Barbara Mroczkowski, Hans Parge,
Todd VanArsdale, Ian Popoff, Joseph Piraino, Stephen Margosiak, James Thomson, Gerrit Los,
Brion W. Murray
Pfizer Global Research and Development, 10770 Science Center Drive, San Diego, CA 92121, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Pin1 is a member of the cis–trans peptidyl-prolyl isomerase family with potential anti-cancer therapeutic
value. Here we report structure-based de novo design and optimization of novel Pin1 inhibitors. Without
a viable lead from internal screenings, we designed a series of novel Pin1 inhibitors by interrogating and
exploring a protein crystal structure of Pin1. The ligand efficiency of the initial concept molecule was
optimized with integrated SBDD and parallel chemistry approaches, resulting in a more attractive lead
series.
Received 18 June 2009
Revised 7 August 2009
Accepted 7 August 2009
Available online 13 August 2009
Keywords:
Pin1
Ó 2009 Elsevier Ltd. All rights reserved.
PPIase
Anticancer
SBDD
Pin1 inhibitor
Originally identified in a yeast two-hybrid screen, Pin1 (Protein
Interacting with NIMA) interacts physically and functionally with
the mitotic kinase NIMA (Never In Mitosis gene A).1 Pin1 is a mem-
ber of peptidyl-prolyl cis–trans isomerase (PPIase) family.1 Pin1
substrate proteins are important in cell-cycle regulation and are
commonly deregulated in human cancer cells. The enzyme plays
a key role in negatively regulating entry into mitosis and is also re-
quired for proper progression through mitosis.2,3 Depletion of Pin1
causes mitotic arrest and apoptosis in budding yeast and tumor
cell lines.1 Pin1 was found to be up-regulated in a wide variety
of human tumors and its over-expression correlates with tumor
grade.4 Inhibition of Pin1 presents a new opportunity for antican-
cer therapy.
Pin1 consists of two structural domains: a catalytic C-terminal
PPIase domain that performs the rotamase function and an N-ter-
minal WW domain5 presumably for substrate recognition. Its cat-
alytic site is unique among other PPIase’s because it recognizes
an unusual phosphorylated Ser/Thr-Pro motif in its substrates.
The first crystal structures of Pin1 have been reported where an
Ala-Pro dipeptide fragment5 was observed in the PPIase catalytic
site.
Pin1 biology. Reversible inhibitor design was initially based on a
substrate inhibitor, Pintide (2, Fig. 1), which achieved low M po-
l
tency.7 Subsequent studies led to reversible phosphorylated inhib-
itors with low nM inhibitions of Pin1, including peptidomimetic
series (3, Fig. 1)8 and non-peptidomimetic series.9,10 The binding
conformation of these phosphorylated series have been character-
ized crystallographically.10,11 Other novel phosphorylated Pin1
inhibitors were also reported.12,13 Recently, moderately potent (Ki
l
M-range) non-phosphorylated small molecule inhibitors14–17
such as thiazolinediones (4, Fig. 1)17 have been reported. We, here-
in describe our design and synthesis of potent, reversible, non-pep-
tidic small molecular Pin1 inhibitors.
The initial goal of our Pin1 program was to discover a Pin1 small
molecular inhibitor that can be used to investigate and exploit po-
tential oncology indications. In order to identify inhibitory hits,
more than one million compounds were screened twice with a
Pin1 fluorescence polarization binding assay18 and a Pin1 enzy-
matic assay.18,19 While there were apparent hits from the high-
throughput screening (HTS), none of these could be confirmed in
secondary assays such as iso-thermal calorimetric titration20 and
NMR study21 on ligand-protein interaction.
Pin1 inhibitor development began with an irreversible, small
To seek a potential starting point for design, we evaluated
inhibitors of other PPIase family members. FKBP-12 is a known
PPIase with well-characterized small molecular inhibitors.22 Since
both Pin1 and FKBP recognize proline residues as part of the
substrate epitope, a collection of more than 200 inhibitors of
FKBP-12 with diverse structures were screened against Pin1.
molecule inhibitor, Juglone (1, Fig. 1),6 that was useful in probing
* Corresponding authors. Tel.: +1 858 622 5927 (C.G.); +1 860 686 0310 (X.H.).
E-mail addresses: alex.guo@pfizer.com (C. Guo), xinjun.hou@pfizer.com (X. Hou).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.08.034

5614
C. Guo et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5613–5616
S
N
OH
O
O
HN
O
O
R2
2: Pintide: Trp-Phe-Tyr-pSer-Pro-
N
S
N
O
O
R2
Phe-Leu-Glu
O
N
O
O
O
O
O
H
O
N
O
H
O
3: Ac-Lys(N-Biotinoyl)-Ala-Ala-Bth-
D-Thr(PO3H2)-Pip-Nal-Gln-NH2
O
O-
R1
1
P
O-
5
6
7
4
Juglone
O
Thiazolinedione
pipecolate core
No Inhibition
Figure 1. Representative known Pin1 inhibitors.
Figure 3. Design based on binding mode of FKBP inhibitors.
Unfortunately, no hit was discovered from this effort. Although the
screening campaign failed, a concurrent protein crystallization ef-
fort provided the necessary starting point to launch a lead discov-
Pipecolate core 6 is a well-known scaffold for FKBP inhibitors.
Compared with FKBP-12’s binding pocket (PDB ID: 1FKG),23 Pin1
active site has additional Ser-154 and Gln-131 residues, which pro-
vides two hydrogen bond donors to the putative ketoamide moiety
in classic FKBP inhibitors. We modeled24 the pipecolate core (7)
into Pin1 catalytic site based on its binding conformation in FKBP
co-crystal structure as a guide. Additional binding interaction
was targeted by building a phenyl fragment to interact with the
canyon formed by Arg-69 and Ser-114 (Fig. 4). This phenyl group
not only could potentially form a hydrophobic stacking interac-
tion25 with the side chain of Arg-69, it could also provide a vector
to reach the hydrophobic pocket formed by Trp-73 and Ala-116. As
a result, compound 11 (Scheme 1) was designed to reach the
charged pocket, the proline-binding site and the Trp-73 canyon
to maximize potential interactions with the R-beta-amino alcohol
appears to fit better.
Compound 11 was synthesized readily from known intermedi-
ate 8 in five steps (Scheme 1). The yields reported here for urea for-
mation (from 8 to 9) and phosphate ester formation (from 9 to 10)
are un-optimized. The five step sequence is mild and robust,9
allowing a number of analogs with various central cores (i.e., pipe-
colate replacements) to be synthesized via a similar route. Com-
pound 11 was determined to be a low micromolar Pin1 inhibitor.
The positive result provided the initial validation for further struc-
ture-based design and optimization of the de novo lead. Analogs of
11 were designed and synthesized, targeting the optimal ester and
proline-derived inhibitor components. Among them, pipecolate
analogs appeared to be most active, with low micromolar to sub-
micromolar Ki in the Pin1 enzymatic assay.
ery effort.
A genetically engineered Pin1 catalytic domain,
construct K7782Q,10 was found to crystallize readily,18 producing
X-ray apo structure (Fig. 2) and co-crystal structure with citric acid
(data not shown). In comparison with the published Pin1 crystal
structure (Fig. 2, orange),5 our apo structure does not contain the
N-terminal WW domain, but aligned well with published structure
in the PPIase domain.
The apo Pin1 K7782Q catalytic domain structure provided a
starting point for inhibitor design. While the apo structure does
not contain sulfate ion, the positively charged recognition pocket
formed by Lys-63, Arg-68, and Arg-69 was occupied by crystal
water molecules or a citric acid molecule. A crystal water molecule
was also located in a canyon region shaped by Arg-69, Ser-114, and
Trp-73. The prolyl binding pocket formed by His-59, Phe-134, and
His-157 is filled with water molecules. In comparison with FKBP-
12,23 the Pin1 catalytic site presents a rich network of H-bond po-
tential with less hydrophobic surface and a somewhat larger bind-
ing site.
Based on the similarities of Pin1 to the FKBP-12 active site, a
proline ester core was investigated. The existence of a positively
charged pocket in the Pin1 active site inspired a simple inhibitor
(5, Fig. 3) which was designed to fill the proline-binding site and
the adjacent charged pocket. It showed no inhibition of the Pin1
catalytic function, suggesting that additional binding interactions
are necessary. For example, the carboxylate may not bind with
affinity comparable to the phosphate of natural substrate. Alterna-
tively, the carboxylate may not be positioned optimally to satisfy
the Pin1 phosphate pocket.
Wefound that the charged interactions are crucial. In comparison
to the phosphate, the corresponding sulfate 12a was significantly
Figure 2. X-ray apo structure of Pin1 K7782Q (light blue, PDB ID: 3IK8) versus full
length Pin1 (orange, PDB ID: 1PIN, Ref. 5). Binding site surface, Lys-63, Arg-68, and
Arg-69 are also shown. The surface is colored by hydrophobicity (brown = hydro-
phobic area).
Figure 4. Predicted binding mode of pipecolate-based Pin1 inhibitors.

C. Guo et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5613–5616
5615
Abs
(s)
1) Dibenzyl
O
N
N,N-diisopropyl-
NH2
(s)
(s)
(R)
Hydrogen
10% Pd/C
(s)
phosphoramidite/
1H-tetrazole
O
Ph
O
Ph
O
O
Ph
4
Ph
OH
N
O
N
N
triphosgen
(R)
N
4
4
(s)
O
O
Ph
4
O
O
O
O
O
O
ethanol
N
Y
Et3N/CH2Cl2 Et3N/CH2Cl2
2) MCPBA
NH
O
NH
O
H
NH
OBn
OBn
OH
OH
O
Ph
O
P
Ph
O P
Ph
OH
(R)
(R)
8
(R)
10 (30% yield)
12a
12b
9
Y=-SO3H Ki = 9.46µM
Y=-PO3H2 Ki = 0.8µM
(12% yield)
11 (88% yield)
Ki = 1.7µM
Ki > 50µM
cell IC50 = 15µM (CA46)
Scheme 1.
less potent (>10Â loss) and the alcohol core 9 was inactive. The
binding of these compounds was further confirmed by both
iso-thermal calorimetric titration and NMR study. Compound 12b
is the most potent compound in the series with an affinity (Ki) of
high-resolution protein structure of Pin1 with a potent small
molecular inhibitor (Fig. 5)18 was available to guide our drug
design.
The crystal structure of 18b revealed a new binding mode in the
PPIase site, far different from our initial model. As shown in Figure 5,
the binding appears to be dominated by three factors. As expected,
the phosphate group fits well in to the charged pocket formed by
Lys-63, Arg-68, and Arg-69 to form hydrogen bonds to all three res-
idues. The phenyl group binds to the ‘prolyl pocket’ formed by His-
59, His-157, Met-130, and Phe-134. The benzene ring forms an
edge-on interaction with His-157. The thiophenyl group binds to
the hydrophobic area formed by Phe-134, Met-130 and Leu-122.
The amide NH makes a weak H-bond to Cys-113 (3.3 Å), while the
amide oxygen is exposed to solvent. This binding mode shows that
while there is a rich hydrogen bonding potential in the pocket,
0.825
l
M. It also inhibited the growth of cancer cells with Pin1
M). As
mechanism-liked phenotypical cell-cycle effects (IC50 = 15
l
expected, this series of compounds demonstrated significant
selectivity over FKBP despite containing a pipecolate fragment, a
putative FKBP binding motif. For example, compound 12b had a
much weaker binding affinity to FKBP-12 (Ki = 137 lM). Com-
pound 12b was synthesized in a route similar to the synthesis of
compound 11.
Optimization of potency and ligand efficiency. Despite achieving
modest cellular activity, the phosphate containing pipecolate Pin1
inhibitors demonstrated shallow SAR and their potency were lim-
ited to low micromolar range. Additionally, the design was focused
on potency, resulting in compound 12b being neither efficient
(LE = 0.19),26 nor drug-like (MW = 636, and Clog P = 7.5). We sus-
pected that the pipecolate core or its ester linkage to the lipophilic
side-chain was not optimal. Attempts to obtain a co-crystal struc-
tures of 11 or analogs were not fruitful. To further improve potency
and drug-like properties, a focused combinatorial library (ꢀ200
compounds) was designed and synthesized (Scheme 2) using the
phenylalaninol phosphate as an ‘anchor’ to search for a pipecolate
replacement. The library core 16 was easily prepared in a three-step
sequence from commercially available amino alcohol 13. The diver-
gent step (from 16 to 17) for library synthesis was conducted in
aqueous 1 M sodium carbonate solution, which was designed to
scavenge any un-reacted acylating (or sulfonylating) reagents and
prevent potential interference with the Pin1 screening assay.
Several potent hits were identified with the high-throughput
PPIase enzymatic assay. The hits were resynthesized and con-
firmed by iso-thermal calorimetric titration. Amide was generally
the best among the three tested linkers-amide, sulfonamide and
urea (17c vs 17a/b, Scheme 2). Compounds 18a and 18b, the fused
bi-aryl amides, are about 10–30-fold more potent than their corre-
sponding pipecolate. The ‘S’ enantiomer of 18a was also synthe-
sized but found to be less active (Ki = 5.6 lM, 56x vs 18a). With
good affinity and solubility, compound 18b was crystallized with
Figure 5. X-ray co-crystal structure of 18b and Pin1 K7782Q construct. (PDB ID:
the Pin1 PPIase K7782Q mutant and its structure was solved. A
3IKD).
Hydrogen
1) Dibenzyl N,N-diisopropyl-
phosphoramidite/1H-tetrazole
NHCBz
NHCBz
OH
NH2
O
P
10% Pd/C
ethanol
86%
CBzCl
Et3N
OBn
OBn
Ph
O
Ph
Ph
Ph
OH
(R)
(R)
(R)
2) MCPBA
95%
13
15
73%
14
R
X
Ki
R
17a
17b
17c
18a
18b
focus
Ph
Ph
Ph
-SO2-
8.77µM
RSO2Cl or
RNCOor
RC(O)Cl
NH2
library
O
P
-NHC(O)- 4.13µM
-C(O)-
-C(O)-
X
OH
OH
O
NH
0.525µM
0.1µM
(R)
O
aq. Na2CO3
OH
OH
2-naphthyl
16 LibraryCore
Ph
O P
(R)
2-benzo[b]thiophenyl -C(O)-
0.179µM
17
Scheme 2.

5616
C. Guo et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5613–5616
Ar
novo design. After further optimization with an integrated SBDD
O
NH2
(R)
and combinatorial library approach, the initial micromolar lead
(11) was developed into a series of potent, non-peptidic, small
molecular Pin1 inhibitors with attractive properties (21a/b). The
discovery improved our confidence in pursuing drug-like Pin1
inhibitors. However, the whole cell activity, presumably due to
poor permeability, remained to be an unsolved problem even for
this series of Pin1 inhibitors. The simplified pharmacophore and
the three Pin1 crystal structures reported herein set the stage for
other drug-like Pin-1 inhibitor designs. In the following communi-
cations, progress toward improving cell permeability of Pin1 inhib-
itors will be reported.
5 steps
NH2
BH3 THF
42%
NH
O
P
R
OH
R
OH
OH
OH
(R)
R
O
(R)
O
similar to Scheme 2
19
20
21-23
Compound
21a
R
Ar
2-naphthyl
Ki (µM)
0.008
0.006
0.032
0.057
0.078
0.089
3-F-phenyl
3-F-phenyl
3-methylphenyl
3-methylphenyl
2,3-diF-phenyl
2,3-diF-phenyl
21b
22a
22b
23a
2-benzothiophenyl
2-naphthyl
2-benzothiophenyl
2-naphthyl
23b
2-benzothiophenyl
Scheme 3.
Acknowledgment
hydrophobic interaction dominates the binding affinity for these
compounds. The channel formed by Ser-114, Arg-69 into the Trp-
73 pocket is harder to access than what we had anticipated.
An analysis of the 18b co-crystal structure revealed a small cav-
ity between to the phenyl C-3 position of the inhibitors and the
Pin1 Phe-134 residue, as shown in Figure 5. Since the cavity ap-
peared hydrophobic, analogs with small substitutions on C-3 such
as F and CH3 were synthesized and tested (Scheme 3). Both analogs
21–22 bound tighter than the corresponding phenyl analogs 18a/b,
with the fluoro analogs being most active. The fluoro atom added
to C-3 position produced a dramatic boost in inhibitory activity
(21b vs 18b, 30x), corresponding to a highly efficient contribution
to binding (DDG = À2.0 kcal/mol).27 The X-ray co-crystal structure
of 22b (Fig. 6) confirmed our hypothesis. Further substitution to
the phenyl ring reduced binding affinity (23a–b).
We thank Dr. Robert Kania for his input and insightful discus-
sion while preparing this Letter.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.08.034.
References and notes
1. Lu, K. P.; Hanes, S. D.; Hunter, T. Nature 1996, 380, 544.
2. Lu, K. P. Prog. Cell Cycle Res. 2000, 4, 83.
3. Winkler, K. E.; Swenson, K. I.; Kornbluth, S.; Means, A. R. Science 2000, 287,
1644.
4. Lu, K.P.; Wulf, G.; Zhou, X. Z. PCT Int. Appl. WO 2001038878, 2001.
5. Ranganathan, R.; Lu, K. P.; Hunter, T. Cell 1997, 89, 875.
6. Hennig, L.; Christner, C.; Kipping, M.; Schelbert, B.; Rucknagel, K. P.; Grabley, S.;
Kullertz, G.; Fischer, G. Biochemistry 1998, 37, 5953.
7. Lu, P. J.; Zhou, X. Z.; Shen, M.; Lu, K. P. Science 1999, 283, 1325.
8. Wildemann, D.; Erdmann, F.; Alvarez, B. H.; Stoller, G.; Zhou, X. Z.; Fanghanel,
J.; Schutkowski, M.; Lu, K. P.; Fischer, G. J. Med. Chem. 2006, 49, 2147.
9. Guo, C.; Dagostino, E. F.; Dong, L.; Hou, X.; Margosiak, S. A. PCT Int. Appl. WO
2004087720, 2004.
10. Matthews, D. A.; Dagostino, E. F.; Ferre, R. A.; Gaur, S.; Guo, C.; Hou, X.;
Margosiak, S.; Mroczkowski, B.; Nakayama, G. R.; Parge, H. E.; Zhu, J. X. PCT Int.
Appl. WO 2004005315, 2004.
11. Zhang, Y.; Daum, S.; Wildemann, D.; Zhou, X. Z.; Verdecia, M. A.; Bowman, M.
E.; Lucke, C.; Hunter, T.; Lu, K. P.; Fischer, G.; Noel, J. P. ACS Chem. Biol. 2007, 2,
320.
Compound 21b, a 3-fluorophenylalanine derivative, is one of
the most potent inhibitors known with
a phosphate group
(Ki = 6 nM). Although it is an efficient ligand with desirable proper-
ties (LE = 0.43, MW = 409, Clog P = 3.1, Clog D = À0.76), compound
21b was inactive in whole cell assay, likely due to the poor perme-
ability conferred by the polar phosphate group.
In conclusion, a series of potent, none peptide-based Pin1 inhib-
itors was designed without a viable lead from HTS. We discovered
the aminophenylpropanol core by utilizing structure-based de
12. Wang, X. J.; Xu, B.; Mullins, A. B.; Neiler, F. K.; Etzkorn, F. A. J. Am. Chem. Soc.
2004, 126, 15533.
13. Siegrist, R.; Zurcher, M.; Baumgartner, C.; Seiler, P.; Diederich, F.; Daum, S.;
Fischer, G.; Klein, C.; Dangl, M.; Schwaiger, M. Helv. Chim. Acta 2007, 90, 217.
14. Do, Q.-Q. T.; Guo, C.; Humphries, P. S.; Marakovits, J. T.; Dong, L.; Hou, X.;
Johnson, M. C. PCT Int. Appl. WO 2006040646, 2006.
15. Bayer, E.; Thutewohl, M.; Christner, C.; Tradler, T.; Osterkamp, F.; Waldmann,
H.; Bayer, P. Chem. Commun. (Camb) 2005, 516.
16. Daum, S.; Erdmann, F.; Fischer, G.; Feaux de Lacroix, B.; Hessamian-Alinejad,
A.; Houben, S.; Frank, W.; Braun, M. Angew. Chem., Int. Ed. 2006, 45, 7454.
17. Bao, L.; Kimzey, A. PCT Int. Appl. WO 2004093803, 2004.
18. See Supplementary data.
19. Zhu, J. X.; Dagostino, E.; Rejto, P. A.; Mroczkowski, B.; Murray, B. Biochem.
Biophys. Res. Commun. 2007, 359, 529.
20. Daum, S.; Fanghanel, J.; Wildemann, D.; Schiene-Fischer, C. Biochemistry 2006,
45, 12125.
21. Namanja, A. T.; Peng, T.; Zintsmaster, J. S.; Elson, A. C.; Shakour, M. G.; Peng, J.
W. Structure 2007, 15, 313. In our NMR studies (unpublished), we observed
ligand-protein interaction similar to this report.
22. Babine, R. E. V. J. E.; Gold, B. G. Exp. Opin. Ther. Patents 2005, 15, 555.
23. Holt, D. A.; Luengo, J. I.; Yamashita, D. S.; Oh, H. J.; Konialian, A. L.; Yen, H. K.;
Rozamus, L. W.; Brandt, M.; Bossard, M. J. J. Am. Chem. Soc. 1993, 115, 9925.
24. For flexible docking method used herein, see: Gehlhaar, D. K.; Verkhivker, G.
M.; Rejto, P. A.; Sherman, C. J.; Fogel, D. B.; Fogel, L. J.; Freer, S. T. Chem. Biol.
1995, 2, 317.
25. Varghese, J. N.; Smith, P. W.; Sollis, S. L.; Blick, T. J.; Sahasrabudhe, A.; McKimm-
Breschkin, J. L.; Colman, P. M. Structure 1998, 6, 735.
26. Hopkins, A. L.; Groom, C. R.; Alex, A. Drug Discovery Today 2004, 9, 430.
27. G = ÀRT ln(Ka) = RT ln(Kd) ꢁ 1.36 log (Ki) [kcal/mol], at T = 298 K. And
Figure 6. X-ray co-crystal structure of 22b and Pin1 K7782Q construct (distance in
Å, PDB ID: 3IKG).
DDG =
D
G(21b) À
DG(18b).