Rational design of urea-based glutamate carboxypeptidase II (GCPII) inhibitors as versatile tools for specific drug targeting and delivery

Article abstract of DOI:10.1016/j.bmc.2014.05.061

Glutamate carboxypeptidase II (GCPII), also known as prostate specific membrane antigen (PSMA), is an established prostate cancer marker and is considered a promising target for specific anticancer drug delivery. Low-molecular-weight inhibitors of GCPII are advantageous specific ligands for this purpose. However, they must be modified with a linker to enable connection of the ligand with an imaging molecule, anticancer drug, and/or nanocarrier. Here, we describe a structure-activity relationship (SAR) study of GCPII inhibitors with linkers suitable for imaging and drug delivery. Structure-assisted inhibitor design and targeting of a specific GCPII exosite resulted in a 7-fold improvement in K_i value compared to the parent structure. X-ray structural analysis of the inhibitor series led to the identification of several inhibitor binding modes. We also optimized the length of the inhibitor linker for effective attachment to a biotin-binding molecule and showed that the optimized inhibitor could be used to target nanoparticles to cells expressing GCPII.

Full text of DOI:10.1016/j.bmc.2014.05.061

Accepted Manuscript
Rational design of urea-based glutamate carboxypeptidase II (GCPII) inhibitors
as versatile tools for specific drug targeting and delivery
Jan Tykvart, Jiří Schimer, Jitka Bařinková, Petr Pachl, Lenka Poštová-
Slavětínská, Pavel Majer, Jan Konvalinka, Pavel Šácha
PII:
S0968-0896(14)00420-9
DOI:
http://dx.doi.org/10.1016/j.bmc.2014.05.061
BMC 11622
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
20 January 2014
26 May 2014
28 May 2014
Please cite this article as: Tykvart, J., Schimer, J., Bařinková, J., Pachl, P., Poštová-Slavětínská, L., Majer, P.,
Konvalinka, J., Šácha, P., Rational design of urea-based glutamate carboxypeptidase II (GCPII) inhibitors as
versatile tools for specific drug targeting and delivery, Bioorganic & Medicinal Chemistry (2014), doi: http://
dx.doi.org/10.1016/j.bmc.2014.05.061
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Rational design of urea-based glutamate carboxypeptidase II (GCPII) inhibitors as versatile
tools for specific drug targeting and delivery
Jan Tykvart‡1,2, Jiří Schimer‡1,2, Jitka Bařinková1, Petr Pachl1,3, Lenka Poštová-Slavětínská1,
Pavel Majer1, Jan Konvalinka1,2, Pavel Šácha*1,2
1Gilead Sciences and IOCB Research Centre, Institute of Organic Chemistry and
Biochemistry, Academy of Sciences of the Czech Republic, Flemingovo n. 2, Prague 6,
Czech Republic
2Department of Biochemistry, Faculty of Natural Science, Charles University, Albertov 6,
Prague 2, Czech Republic
3Institute of Molecular Genetics, Academy of Sciences of the Czech Republic,
Vídeňská 1083, Prague 4, Czech Republic
*corresponding author:
Pavel Šácha, Ph.D.
Institute of Organic Chemistry and Biochemistry, ASCR, v.v.i.
Flemingovo n. 2, Prague 6, 166 10, Czech Republic
tel: 00420 220 183 452, fax: 00420 220 183 578
email: pavelsacha@gmail.com
‡ these authors contributed equally to the presented work
1

Abstract
Glutamate carboxypeptidase II (GCPII), also known as prostate specific membrane antigen
(PSMA), is an established prostate cancer marker and is considered a promising target for
specific anticancer drug delivery. Low-molecular-weight inhibitors of GCPII are
advantageous specific ligands for this purpose. However, they must be modified with a linker
to enable connection of the ligand with an imaging molecule, anticancer drug, and/or
nanocarrier. Here, we describe a structure-activity relationship (SAR) study of GCPII
inhibitors with linkers suitable for imaging and drug delivery. Structure-assisted inhibitor
design and targeting of a specific GCPII exosite resulted in a 7-fold improvement in Ki value
compared to the parent structure. X-ray structural analysis of the inhibitor series led to the
identification of several inhibitor binding modes. We also optimized the length of the
inhibitor linker for effective attachment to a biotin-binding molecule and showed that the
optimized inhibitor could be used to target nanoparticles to cells expressing GCPII.
Keywords
GCPII; PSMA; structure-aided drug design; specific drug targeting
2

List of abbreviations
Avi-GCPII - extracellular portion (amino acids 44-750) of recombinant human GCPII with N-
terminal Avi-tag followed by a TEV cleavage site
GCPII - glutamate carboxypeptidase II
Qdot - quantum dot
LNCaP - lymph node carcinoma of the prostate
2-MPPA - 2-(3-mercaptopropyl)pentanedioic acid
PCa - prostate carcinoma
2-PMPA - 2-(phosphonomethyl)pentanedioic acid
PSMA - prostate specific membrane antigen
rhGCPII - extracellular portion (amino acids 44-750) of recombinant human GCPII
SAR - structure-activity relationship
SPR - surface plasmon resonance
TEV - tobacco etch virus
Graphical abstract
3

1. Introduction
Prostate carcinoma (PCa) is the most commonly diagnosed cancer and the second leading
cause of cancer death among men in the USA; the number of diagnoses and deaths exceeded
240,000 and 28,000, respectively, in 20121. Identification of a molecule that would enable
early diagnosis and reliable detection of PCa metastases, as well as function as a possible
target for specific therapy, is crucial for effective treatment of this cancer. Evidence suggests
that glutamate carboxypeptidase II (GCPII), also known as prostate specific membrane
antigen (PSMA), is a promising candidate for this purpose2-3.
GCPII, a 750-amino-acid type II transmembrane di-zinc metallopeptidase, possesses
two distinct endogenous substrates in the human body. In the brain, it is responsible for
degradation of the peptide neurotransmitter N-acetyl-L-aspartyl-L-glutamate4, and in the
small intestine, it cleaves terminal glutamates from poly-γ-glutamylated folate5. In addition to
its enzymatic roles, GCPII is studied for its overexpression in PCa6. Recently, it has been
shown that GCPII is also expressed at the neovasculature of solid tumors7, which makes it
even more interesting as a diagnostic/therapeutic target. Furthermore, the facts that GCPII is a
transmembrane protein, with a large extracellular domain, and is internalized upon ligand
binding8 make GCPII even more suitable for specific cancer cell targeting.
Antibodies, small-molecule inhibitors, and aptamers have been studied as tools for
specific targeting of GCPII. At present, only one molecule, radioisotope-antibody conjugate
111In-7E11-C5.3, has made it through the development pipeline into practical use as the
imaging agent ProstaScint™9. New generations of diagnostic and therapeutic radiolabeled
antibodies targeting the extracellular portion of GCPII are in various stages of clinical trials,
and diverse radiolabeled small-molecule inhibitors also have been intensively investigated2, 10-
12. Beside radiolabeling approach, a small-molecule inhibitors and in some extent also
4

antibodies modified with various organic dyes have been intensively investigates as a
potential diagnostic agents 13-14
.
In addition to their applications in direct labeling with radioisotope or organic dye,
GCPII-specific antibodies, small-molecule inhibitors, and aptamers also could be used for
targeted delivery of various nanoparticles15, such as gold nanoparticles16-17, polymers18-21
,
nanorods22, and nanobubbles23. Generally, low-molecular-weight ligands have a number of
advantages over antibodies, which are still used more frequently for specific targeting. Small-
molecule ligands are easy to prepare on a large scale, could have very favorable
pharmacokinetic properties (such as bioavailability, half-life, etc.), and might penetrate the
blood-brain barrier24. Moreover, numerous modified analogs of a particular small-molecule
ligand can be easily prepared. Such modifications may lead to enhanced specificity and
affinity.
A number of potent (low nanomolar) inhibitors of GCPII have been described in the
literature. Generally, these GCPII inhibitors contain a glutamate residue or analog, which
binds very specifically in the P1′ site, and a zinc binding group. Based on the nature of the
zinc binding group, GCPII inhibitors can be divided into three groups: (1) phosphonate-,
phosphate-, and phosphoramide-based (e.g., 2-PMPA25); (2) thiol-based (e.g., 2-MPPA26);
and (3) urea-based inhibitors (e.g., Glu-NH-C(O)-NH-Glu27). For more detailed information,
see a recent review summarizing current knowledge about GCPII inhibitors28.
The structure of GCPII has been extensively studied during the past several years,
resulting in deposition of more than 20 X-ray structures, the majority of them solved by the
Barinka and Konvalinka groups. GCPII likely forms homodimer under physiological
conditions, and its substrate binding site has strong specificity in the P1′ pocket, while the P1
pocket is more flexible. Therefore, greater molecular variability is possible in the part of the
inhibitor molecule that binds to the P1 site. Additionally, a deep accessory funnel-shaped
5

tunnel contains several more exosites that could be utilized in design of more potent
inhibitors. For more detailed information, see a recently published structural description of
GCPII molecule29.
In this work, we designed novel inhibitors based on the established urea-based
scaffold27. We performed a structure-activity relationship (SAR) study to utilize the arginine
patch exosite near the S1 site that is able, through conformational change of Arg463, to create
a so-called "S1-accessory hydrophobic pocket"30-31. To maximize inhibitor binding affinity,
we also attempted to exploit a previously described positive effect of halogens attached to the
aromatic moiety of the inhibitor32.
We set out to prepare a generally applicable inhibitor of GCPII that could be used for
facile modification of various nanoparticles using a streptavidin-biotin interaction since it has
been already shown that such modified inhibitors can bind to GCPII 17, 33
.
2. Materials and methods
2.1 Preparation of recombinant GCPII
For SPR measurements, the extracellular part of recombinant human GCPII (rhGCPII)
containing amino acids 44-750 was prepared as previously described34. For inhibition
constant measurements and crystallization trials, the extracellular part of recombinant human
GCPII with an N-terminal Avi-tag (Avi-GCPII) was prepared as previously described35.
2.2 Determination of inhibition constants by HPLC
IC50 values for all inhibitors were determined using an HPLC-based assay. In a 96-
well plate, 0.2 ng of Avi-GCPII in 25 mM Bis-Tris propane, 150 mM NaCl, 0.001% (w/w)
octaethylene glycol monododecyl ether (Affymetrix), pH 7.4, and inhibitors were mixed
together in a final volume of 180 µl. Ten inhibitor concentrations were used to obtain a full
6

inhibition curve. Reactions were pre-incubated at 37 °C for 5 min, started by addition of 20 µl
of 4 µM pteroyl-di-L-glutamate (Schircks Laboratories) and incubated at 37 °C for 20 min.
The reactions were stopped with 20 µl of 25 µM 2-PMPA.
Subsequently, 100 µl of reaction mixture was injected into and analyzed on an Agilent
1200 Series system using an Acquity UPLC HSS T3 1.8 µm column (2.1×100mm, Waters).
The HPLC analysis was run isocratically in 2.7% ACN and 97.3% 20 mM phosphate, pH 6.0.
Absorbance of both substrate and product were detected at 281 nm. Afterwards, data were
processed, and IC50 values were obtained using GraFit v.5.0.11 (Erithacus Software Ltd.).
Kinetic parameters (KM and kcat) of pteroyl-di-L-glutamate cleavage by Avi-GCPII in
the reaction buffer used for IC50 measurement were determined as described above but
without addition of inhibitor and with substrate concentrations ranging from 15 nM to 400 nM
(substrate conversion for all reactions was 13 2 %). From the kinetic parameters of pteroyl-
di-L-glutamate cleavage and by assuming a competitive mode of inhibition, the Ki value for
each inhibitor was calculated using the Cheng-Prusoff equation36.
2.3 Surface plasmon resonance measurements
All surface plasmon resonance measurements were performed on a four-channel SPR
sensor platform developed at the Institute of Photonics and Electronics, Prague37-38. In a
typical experiment, the SPR chip (provided by the IPE) was loaded into a pure ethanol
mixture (7:3) of HS-(CH2)11-PEG4-OH and HS-(CH2)11-PEG6-O-CH2-COOH alkanethiols
(Prochimia) with a final concentration of 0.2 mM and incubated for 1 h at 37 °C. The chip
was rinsed with UV ethanol and deionized water and subsequently dried with flow of
nitrogen. Then, the chip was mounted to the prism on the SPR sensor. The experiment was
performed at 25 °C.
7

The activation of carboxylic terminal groups on the sensor surface was performed
in situ by injecting a 1:1 mixture of 11.51 mg/mL N-hydroxysuccinimide (NHS) and
76.68 mg/mL 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC) in
deionized water (Biacore) for 5 min at a flow rate of 20 µL/min. The remainder of the
experiment was performed at a flow rate of 30 µL/min. Afterwards, a 0.02 mg/mL neutravidin
solution (for qualitative experiments) or a mixture of neutravidin with BSA (molar ratio 1:50)
at a final concentration of 0.04 mg/mL in 10 mM sodium acetate buffer, pH 5.0 (for kinetic
experiments), was loaded for 8 min. Then, high ionic strength solution (PBS with 0.5 M
NaCl) was used to wash out noncovalently bound neutravidin molecules, followed by
1 M ethanolamine (Biacore) for deactivation of residual activated carboxylic groups.
A 1 µM solution of inhibitor in TBS was immobilized on prepared gold chips with
neutravidin or neutravidin/BSA layer in TBS to saturate all biotin binding sites. Afterwards,
rhGCPII (in the same or various concentrations) in TBS was injected for several minutes
(association), and then TBS alone was injected (dissociation).
Kinetic curves of binding were exported and subsequently fitted in TraceDrawer v.1.5
(Ridgeview Instruments AB) to obtain kon and koff parameters.
2.4 Crystallization and data collection
Avi-GCPII stock solution (15 mg/mL) was mixed in a 1:1 ratio with 1-10 mM
inhibitor. The crystallization droplets were prepared by combining 1 µl of the protein/inhibitor
solution with 1 µl of the reservoir solution containing 33% pentaerythritol propoxylate
(Hampton Research), 1-2% PEG 3350 (Fluka), and 100 mM Tris–HCl, pH 8.0. In some cases,
micro-seeding was used to obtain crystals of sufficient size for diffraction. The micro-seeding
solution was prepared by fishing out several smaller crystals, vortexing them in 100 µl of
precipitation solution, and further diluting the solution 20- to 100-fold in precipitation
8

solution. Crystals were grown by the hanging drop vapor diffusion method at 291 K. Crystals
were flash-frozen in liquid nitrogen, and the diffraction data were collected either at 100 K at
beamlines BL14.1 and BL14.2 of BESSY, Berlin, Germany39 or at 120 K using an in-house
diffractometer (Nonius FR 591 generator, 345-mm MarResearch Image Plate detector).
Complete datasets were collected from single crystals. The diffraction data to resolution were
integrated and reduced using XDS40 and its graphical interface XDSAPP41.
2.5 Structure determination and refinement
Structure determination was performed by molecular replacement using the program
Molrep in CCP4 software package42-43. The previously solved structure of rhGCPII in
complex with the inhibitor 2-PMPA (PDB entry 2PVW)44 without the inhibitor and without
water molecules was used as the starting model for structural refinement based on the X-ray
data collected from crystals of the Avi-GCPII/inhibitor complexes. Calculations were
performed with the program Refmac 5.745, and the refinement protocol was interspersed with
manual corrections to the model using WinCoot v.0.746. The final models, together with
experimental amplitudes, were deposited into the RCSB Protein DataBank. A summary of
structural parameters is displayed in Table S1.
2.6 Preparation of nanocarriers specifically recognizing GCPII
A 25 µl aliquot of a 250 nM solution of Qdot® 605 streptavidin conjugate (Life
Technologies) was mixed with a 40-fold molar excess of 22b in TBS and incubated overnight
at 4 °C. Excess unbound 22b was removed by six 10x dilution/concentration cycles using
Amicon Ultra-0.5mL 30K filters (Millipore). Subsequently, the solution was diluted to
100 nM and used as a stock solution for subsequent experiments.
9

2.7 Testing of specific targeting towards prostate cancer cell lines by confocal
microscopy
LNCaP and PC-3 cells were grown in 4-Chamber 35 mm Glass Bottom Dishes (In
Vitro Scientific). LNCaP cells were grown in RPMI-1640 medium (Sigma-Aldrich) with
addition of FBS (final concentration of 10%) while PC3 cells were grown in DMEM-High
Glucose medium (GE Healthcare) with addition of L-glutamine (final concentration of 4 mM)
and FBS (final concentration of 10%). After two days, the medium was aspirated from all
chambers, and cells were washed once with TBS. Afterwards, the 10 nM solution of
appropriate Qdot® 605 streptavidin conjugate in TBS was added to each chamber and the live
cells were incubated for 15 minutes at 37 °C. Then, confocal images (pinhole 1 Airy unit) of
cells in each chamber were taken with a Zeiss LSM 780 confocal microscope (Carl Zeiss
Microscopy) using an oil-immersion objective (Plan-Apochromat 63x/1.40 Oil DIC M27).
The fluorescent images were collected at room temperature using 2% of the 405 nm diode
laser (max. power 30 mW) for excitation and emission spectral detector in the 578-639 nm
range (voltage on detector: 834 V). Simultaneously bright field images were taken using
transmission detector (voltage 342 V). All the images were taken using the same settings. The
microscope was operated and the images were processed by ZEN 2011 software (Carl Zeiss
Microscopy).
3. Experimental section
3.1 Organic synthesis
All chemicals were purchased from Sigma-Aldrich unless otherwise stated. All
inhibitors tested in biological assays were purified using preparative scale HPLC with a
Waters Delta 600 system (flow rate 7 mL/s, gradient shown for each compound - including
tR) using a Waters SunFire C18 OBD Prep Column, 5 µm, 19 x 150 mm. The purity of
10

compounds was tested with an analytical Jasco PU-1580 HPLC (flow rate 1 mL/s, invariable
t
gradient: 2-100% ACN in 30 min, R shown for each compound) using a Watrex C18
Analytical Column, 5 µm, 250 x 5 mm. The purity of all tested inhibitors was at least 99%.
The structures of inhibitors were further confirmed by HRMS on LTQ Orbitrap XL (Thermo
Fisher Scientific) and by NMR (Bruker Avance I™ 500 MHz equipped with Cryoprobe or
Bruker Avance I™ 400 MHz).
The general procedure for preparation of para-substituted branched inhibitors (shown
for the para-bromo derivative) is depicted in Scheme 1.
Scheme 1: Preparation para-bromo substituted branched inhibitor
Compound 21: Di-tert-butyl 2-(3-(6-((4-bromobenzyl)amino)-1-(tert-butoxy)-1-oxohexan-2-
yl)ureido)pentanedioate.
Di-tert-butyl
2-(3-(6-amino-1-(tert-butoxy)-1-oxohexan-2-
yl)ureido)pentanedioate (300 mg, 0.615 mmol, 1.0 eq; prepared as described in47) and 120 mg
(0.646 mmol, 1.05 eq) of 4-bromobenzaldehyde were dissolved in 5 mL methanol in a round-
11

bottom flask. Glacial acetic acid (50 µL) was added, and 120 mg (1.85 mmol, 3.0 eq) of
sodium cyanoborohydride were added in one portion with vigorous stirring. After 12 h, the
reaction mixture was quenched by addition of 10 mL water. The reaction mixture was further
diluted after 10 min with 50 mL water and was extracted with ethyl acetate (EtOAc; 3x25
mL). The organic phase was dried and evaporated, and crude product was purified by
chromatography on silica (mobile phase: EtOAc + 1% saturated ammonia in water, TLC Rf =
0.55). Pure product was isolated in 48% yield (395 mg). Analytical HPLC (gradient: 2-100%
ACN in 30 min) tR = 23.4 min. HRMS (ESI+): calculated for C31H51O7N3Br [M]+ 656.29049.
1
Found 656.29062. H NMR (500 MHz, DMSO-d6): δ 7.47 (m, 2H, m-Ph), 7.27 (m, 2H, o-
Ph), 6.29 (d, 1H, J = 8.5 Hz, HN-Glu-2), 6.24 (d, 1H, J = 8.4 Hz, HN-Lys-2), 4.02 (btd, 1H,
J1 = 8.6 , J2 = 5.1 Hz, Glu-2), 3.96 (td, 1H, J1 = 8.1, J2 = 5.4 Hz, Lys-2), 3.62 (s, 2H, CH2-
Ph), 2.41 (t, 2H, J = 7.1 Hz, Lys-6), 2.25 (ddd,1 H, J1 = 16.6, J2 = 8.8, J3 = 6.8 Hz, Glu-4b),
2.18 (ddd, 1H, J1 = 16.6, J2 = 8.8,J3 = 6.1 Hz, Glu-4a), 1.86 (m, 1H, Glu-3b), 1.66 (m, 1H,
Glu-3a), 1.57 (m, 1H, Lys-3b), 1.49 (m, 1H, Lys-3a), 1.40 (m, 2H, Lys-5), 1.38 (bs, 27 H,
(C(CH3)3), 1.29 (m, 2H, Lys-4).13C NMR (125.7 MHz, DMSO-d6): δ 172.50 (Lys-1), 172.11
(Glu-1), 171.63 (Glu-5), 157.31 (NH-CO-NH), 140.83 (i-Ph), 131.07 (m-Ph), 130.26 (o-Ph),
119.52 (p-Ph), 80.76 (CH(CH3)3), 80.45 (CH(CH3)3), 79.95 (CH(CH3)3), 53.18 (Lys-2),
52.38 (CH2-Ph), 52.36 (Glu-2), 48.49 (Lys-6), 32.17 (Lys-3), 31.07 (Glu-4), 29.24 (Lys-5),
27.93 (CH(CH3)3), 27.84 (CH(CH3)3),27.82 (CH(CH3)3), 27.77 (GLu-3), 23.03 (Lys-4).
Compound 22a: 2-(3-(5-(N-(4-bromobenzyl)acetamido)-1-carboxypentyl)ureido)pentanedioic
acid. Compound 21 (63 mg, 95.9 µmol, 1.0 eq) was dissolved in 0.5 mL DMF, and 23 mg
(144 µmol, 1.5 eq) of N-acetoxysuccinimide were added in one portion along with 25 µL (144
µmol, 1.5 eq) DIEA. The reaction mixture was stirred at room temperature overnight. All
volatiles were removed in vacuum, and the crude product was treated with 0.5 mL TFA, then
12

alternately stirred and sonicated for 15 min. TFA was removed by flow of nitrogen, and the
t
oily product was purified using preparative HPLC (gradient: 15-50% ACN in 60 min, R = 38
min). Sixteen milligrams of product were obtained (isolated yield = 31 %). Analytical HPLC
t
(gradient: 2-100% ACN in 30 min) R = 18.0 min. HRMS (ESI+): calculated for
C21H29O8N3Br [M]+ 530.11325. Found 530.11286. There are two possible rotamers; some
atoms give two signals in NMR spectra. In these cases, the two signals are connected with the
conjunction “and.” 1H NMR (500 MHz, DMSO-d6): δ 7.56 and 7.49 (m, 2H, m-Ph), 7.17 and
7.16 (m, 2H, o-Ph), 6.37-6.25 (m, 2H, NH-Lys-2, NH-Glu-2), 4.51 and 4.44-4.43 (s and m,
2H, CH2-Ph), 4.13-4.00 (m, 2H, Glu-2, Lys-2), 3.21-3.13 (m, 2H, Lys-6), 2.31-2.16 (m, 2H,
Glu-4), 2.06 and 1.97 (s, 3H, CH3-CO), 1.96-1.86 (m, 1H, Glu-3b), 1.75-1.57 (m, 2H, Glu-
3a, lys-3b), 1.57-1.36 (m, 3H, Lys-3a, Lys-5), 1.28-1.16 (m, 2H, Lys-4). 13C NMR (125.7
MHz, DMSO-d6): δ 174.72 and 174.68 (Lys-1), 174.41 (Glu-1), 173.95 (Glu-5), 170.07 and
169.90 (CH3-CO), 157.47 (NH-CO-NH), 138.12 and 137.64 (i-Ph), 131.77 and 131.42 (m-
Ph), 129.97 and 128.98 (o-Ph), 120.37 and 120.13 (p-Ph), 52.39 and 52.24 (Lys-2), 51.85
and 51.83 (Glu-2), 50.60 and 46.99 (CH2-Ph), 47.67 and 45.09 (Lys-6), 32.00 and 31.97
(Lys-3), 30.11 (Glu-4), 27.73 (Glu-3), 27.69 and 26.89 (Lys-5), 22.66 and 22.47 (Lys-4),
21.77 and 21.38 (CH3-CO).
Compound 22b:
(3S,7S)-12-(4-bromobenzyl)-5,13,54-trioxo-59-((3aR,4R,6aS)-2-
oxohexahydro-1H-thieno[3,4-d]imidazol-4-yl)-17,20,23,26,29,32,35,38,41,44,47,50-
dodecaoxa-4,6,12,53-tetraazanonapentacontane-1,3,7-tricarboxylic acid. HOOC-PEG12-biotin
(38 mg, 45.7 µmol, 1.0 eq; IRIS biotech) was dissolved in 0.5 mL DMF, and 16 mg
(49.8 µmol, 1.1 eq) TBTU along with 10 µL DIEA (57.4 µmol, 1.25 eq) were added. The
reaction mixture was stirred for 15 min, and 30 mg (45.7 µmol, 1.0 eq) of 21 (dissolved in
0.2 mL DMF) was added to the reaction mixture in one portion. The reaction proceeded at
13

room temperature overnight, and the mixture was rotary evaporated to dryness. TFA (0.5 mL)
was added, and the reaction mixture was alternately sonicated and stirred for 15 min. TFA
was then removed by flow of nitrogen, and the product was purified using preparative HPLC
t
(gradient: 30-80% ACN in 60 min, R = 18 min). Eighteen milligrams of product were
t
obtained (isolated yield = 33 %). Analytical HPLC (gradient: 2-100% ACN in 30 min) R =
19.0 min. HRMS (ESI+): calculated for C56H92O22N6BrS [M]+ 1311.5174. Found 1311.5164.
4. Results
4.1 Effect of ε-amino group modification and optimization of PEG linker length
First, we prepared several compounds to map the effects of modification at the ε-
amine lysine of the inhibitor scaffold Glu-NH-C(O)-NH-Lys (see Figure 1A). Exchange of a
hydrogen atom (2) for an acetyl group (3) at the ε-amino group resulted in a 2-order-of-
magnitude decrease in Ki value (Ki(2) = 44.3 2.4 nM vs. Ki(3) = 0.174 0.033 nM).
However, substitution of the acetyl group with a PEG linker with biotin (4, 5) led to major
drop in inhibitor potency, showing a negative correlation with the number of PEG units. To
determine the minimal length of the PEG linker (necessary to connect GCPII and biotin-
binding molecule), we used surface plasmon resonance (SPR) to analyze the ability of these
inhibitors, attached via biotin to immobilized neutravidin, to bind GCPII (see Figure 1B). The
difference in association curves as well as saturation signals (4 nm for 5 vs. 0.75 nm for 4)
clearly shows that PEG12-biotin but not PEG8-biotin ensures efficient connection of GCPII
and neutravidin.
14

Figure 1: Kinetic characterization of ε-amino group derivatization of the Glu-NH-C(O)-NH-
Lys inhibitor scaffold and optimization of the PEG linker length. A. Ki values for modified
inhibitors with acetyl or PEG chains of different lengths. The Ki values are presented as mean
#
§
standard deviation. measurements performed in duplicate. measurements performed in at
least triplicate. B. Comparison of recombinant human GCPII (rhGCPII) binding to
immobilized inhibitors 4 and 5 by surface plasmon resonance (SPR). The inhibitors were
immobilized on gold chips via interaction with neutravidin (data not shown). At 0 min,
rhGCPII in TBS (29.4 nM) was injected to both flow-channels (association phase), and at
6 min, TBS was applied to both channels to wash out bound rhGCPII molecules (dissociation
phase). SPR data were processed by TraceDrawer v.1.5 (Ridgeview Instruments AB).
4.2 Reaching the exosite: Connection of the aromatic moiety to the inhibitor scaffold
To further improve inhibitor potency, we attempted to utilize the previously described
"S1-accessory hydrophobic pocket"30. We designed four possible connections of an aromatic
moiety, intended to interact with this pocket, to the inhibitor scaffold. First, we tried a
branching setup at the ε-amino group, substituting hydrogen atom with a benzyl moiety and
the second hydrogen with a linker (acetyl, 15a, or PEG12-biotin, 15b). Second, we used a
15

linear setup connecting the aromatic moiety to the inhibitor scaffold via a methylene carbonyl
(9a, 9b), amide (13a, 13b), or carbonyl group (20(H)a, 20(H)b) with the linker (acetyl or
PEG12-biotin) emerging from the aromatic ring at the ortho position. We compared the Ki
values for all these compounds (see Figure 2). We noted an increase in Ki values when
comparing the original acetylated compound (3) to acetylated inhibitors containing an
aromatic moiety (9a, 20(H)a, 13a, 15a). On the other hand, when we compared inhibitors with
PEG linkers with the original compound with PEG linker (Ki(5) = 10.5 1.0 nM), we
observed improvements in inhibitor potency when the aromatic moiety is attached in the
branched setup (Ki(15b) = 3.42 0.47 nM) and in the linear setup connected via amide group
(Ki(13b) = 7.6 1.2 nM).
Figure 2: Kinetic characterization of inhibitors with aromatic moieties. Inhibitors with
various attachments of an aromatic ring (circled) – methylene carbonyl (9a, 9b), amide (13a,
13b), methylene (15a, 15b), and carbonyl (20(H)a, 20(H)b) linkers – were synthesized and their
Ki values for GCPII determined. The Ki values are presented as mean standard deviation.
16

†
#measurements performed in duplicate. measurements performed in triplicate. The PEG12-
biotin formula is shown in Figure 1.
4.3 Effect of halogen substitution at the para position of the aromatic ring
Based on previously reported data32, we investigated the possible effect of introduction
of halogen atoms at the para position of the aromatic moiety. We prepared and kinetically
characterized a panel of halogenated inhibitors with the aromatic moiety linked via a carbonyl
group (see Figure 3). The carbonyl attachment of the aromatic ring was chosen for its
synthetic availability. In general, halogen substitution showed a negligible effect on Ki values,
with the exception of the bromine derivative. Bromine substitution led to a two-fold increase
in inhibitor potency (Ki(20(Br)a) = 0.268 0.060 nM vs. Ki(20(H)a) = 0.601 0.021 nM).
Figure 3: Kinetic characterization of inhibitors with various halogen substitutions at the para
position of the aromatic ring. The Ki values are presented as mean standard deviation.
#measurements performed in duplicate. †measurements performed in triplicate.
17

Based on these results, we chose the inhibitors with the two most favorable aromatic
moiety linkers and modified them by adding a bromine moiety in the para position of the
aromatic ring (see Figure 4A). Both inhibitors showed superior inhibition potency compared
to their non-halogenated counterparts in both acetylated and PEGylated forms. 4-bromobenzyl
modification with branched attachment of PEG12-biotin moieties at the ε-amino group had the
lowest Ki value (Ki(22b) = 1.570 0.071 nM) of all the PEGylated inhibitors. This
combination of the most favorable aromatic moiety connection and halogen derivatization led
to a 7-fold improvement in inhibitor potency compared to the lead structure
(Ki(5) = 10.5 1.0 nM). Additionally, we confirmed the ability of 22b to mediate the GCPII-
neutravidin interaction and determined its association and dissociation rates by SPR, yielding
a KD value of 1.85 0.42 nM (see Figure 4B).
Figure 4: Kinetic characterization of two inhibitors with different connections of the 4-
bromobenzyl moiety. A. A branched connection or linear connection via amide linker and
bromine placed in the para position of the aromatic ring showed a synergic effect and
provided the tightest binding inhibitors in the panel studied. The Ki values are presented as
†
§
mean standard deviation. measurements performed in triplicate. measurements performed
18

in at least triplicate. The PEG12-biotin formula is shown in Figure 1. B. Kinetic
characterization of the neutravidin-rhGCPII interaction mediated by inhibitor 22b measured
by surface plasmon resonance (SPR). Compound 22b was immobilized on a gold chip via its
interaction with neutravidin (data not shown). At 0 min, four samples with different rhGCPII
concentrations (3.675 nM, 7.35 nM, 14.7 nM, and 29.4 nM) were injected into 4 different
flow-channels (association phase), and after 9 min, TBS was applied to all channels to wash
out bound rhGCPII molecules (dissociation phase). Values of association (kon) and
dissociation (koff) rates displayed here were obtained by fitting four independent
measurements using TraceDrawer v.1.5 (Ridgeview Instruments AB) and represent mean
standard deviation. The dissociation constant (KD) was calculated from kon and koff. For visual
clarity, one representative example of this kinetic measurement is presented.
4.4 Binding modes of inhibitors to GCPII
We determined X-ray structures of GCPII in complex with several inhibitors [13b
(PDB code 4NGS), 22a (4NGM), 22b (4NGQ), 28b (4NGT), 20(H)b (4NGP), 20(Br)
a
(4NGN), and 9b (4NGR)] to investigate how different connections of aromatic moieties (both
unmodified and halogenated) and the presence of the PEG linker influence inhibitor binding
into the GCPII substrate binding pocket (for information about data collection and refinement
statistics, see Table S1). In all structures, the basic inhibitor scaffold binds into the S1′ and S1
sites as previously described30. In general, the accessory tunnel of GCPII exhibited a high
degree of flexibility, enabling several modes of inhibitor binding (see Figure 5A).
Comparison of the structures of GCPII in complex with 22a and 22b revealed
different binding modes of the 4-bromobenzyl moiety (see Figure 5B, C). In the case of
PEGylated inhibitor 22b, the 4-bromobenzyl group is located within the arginine patch of
GCPII created by arginines 463, 534, and 536. The formation of this cavity is enabled by
19

a 2 Å shift of the Arg463 side chain. The conformation is stabilized by a π-cationic interaction
between the guanidinium group of Arg463 and the aromatic ring and also by hydrogen
bonding of the inhibitor’s carbonyl linker with Arg463 κ-NH (see Figure 5B). In contrast, the
4-bromobenzyl group of the acetylated inhibitor (22a) adopted a different conformation
outside the S1-accessory hydrophobic pocket. In this case, the 4-bromobenzyl moiety
extrudes into accessory tunnel of the GCPII binding site with no obvious interaction with the
protein (except a water-mediated hydrogen bond between the inhibitor carbonyl group and ε-
O of Gln254). The arginine patch adopts a more energetically favorable conformation in
which the guanidinium group of Arg463 interacts with the carboxyl groups of Glu457 and
Asp465 (see Figure 5C).
Structures of GCPII in complex with inhibitors with linearly attached linker moieties
(13b, 20(H)b, 28b, 20(Br)a) revealed that these have a different binding mode than inhibitors
with branched attachment of the linker and aromatic moiety (22a,b). In both acetylated and
PEGylated forms, the aromatic moiety did not enter the arginine patch, indicating that the
location of the linker connection at the ortho position of the aromatic ring likely sterically
disables this mode of binding (see Figure 5A).
PEGylated inhibitors with amide (13b, 28b) and carbonyl (20(H)b) connections
showed similar modes of binding to GCPII. The aromatic moieties of these inhibitors were
located in the hydrophobic pocket created by the side chains of Trp541 and Phe546. Trp541
and Phe546 are part of the so-called GCPII "entrance lid," which in these complexes is in
closed conformation, enabling the creation of this exosite (see Figure 5D). Addition of
bromine (28b) led to slight movement of the aromatic ring towards the protein backbone. This
inhibitor was found in two different conformations. These conformations were stabilized
either by π-cationic interaction of the 4-bromobenzyl moiety with the guanidinium group of
Arg463 and bromine ionic interaction with a sodium cation (28b - cyan) or by interaction of
20

bromine with the carbonyl oxygen of Lys539 by halogen bond and with γ-O of Thr538 by
hydrogen bond (28b - green) (see Figure 5E)48. In the case of PEGylated inhibitor with
methylene carbonyl connection of the aromatic ring (9b), the electron density for the aromatic
moiety was not visible in the structure, suggesting that this moiety is not bound in one stable
conformation (data not shown). This observation might explain the lower potency we
observed for inhibitors with methylene carbonyl connections.
Figure 5: Structural analysis of several inhibitors binding into the GCPII active site. The
backbone of GCPII is depicted in cartoon representation in gray. Interacting amino acid side
chains are shown as lines, and inhibitor molecules are in ball-and-stick representation. The
carbon atoms of interacting amino acid side chains and the inhibitor molecules are colored
orange (22b), yellow (22a), magenta (20(H)b), and cyan/green (28b). Oxygen atoms are
depicted in red, nitrogen atoms in blue, and bromine atoms in black. Zinc ions are shown as
gray spheres, chlorine ions as green spheres, sodium ion as a violet sphere, and water
molecule as a red sphere. A. An overlay of the crystal structures of inhibitors 22a, 22b,
20(H)b, and 28b in complex with GCPII and their different binding modes into the accessory
21

binding tunnel. The major conformational shift of the Arg463 side chain is indicated with a
black arrow. The gray circles highlight the S1 and S1′ binding sites of the inhibitor molecules.
For clarity, only one conformation of GCPII in complex with 28b is shown (cyan). B. The
bromobenzyl group of 22b is bound within the arginine patch and stabilized by π-cationic
interaction with the Arg463 guanidinium group and by hydrogen bonding between κ-NH of
Arg463 and the carbonyl oxygen of the inhibitor molecule. C. The bromobenzyl moiety of
22a is bound outside the arginine patch, which is in "closed" conformation with the Arg463
side chain stabilized by hydrogen bonding with the Glu457 and Asp465 carboxyl groups. D.
Binding of the 20(H)b aromatic moiety into the GCPII hydrophobic pocket created by the
Phe546 and Trp541 side chains. The GCPII entrance lid (Trp541-Gly548) adopts a closed
conformation, enabling formation of this hydrophobic pocket. E. In the co-crystal structure of
GCPII, with inhibitor 28b was found in two different conformations (the inhibitor molecule,
together with interacting amino acids, is depicted in cyan and green). In one conformation
(cyan), the aromatic moiety is moved above the Arg463 side chain and stabilized by π-
cationic interaction with its guanidinium group, and in addition, the bromine atom interacts
with a sodium ion. The second conformation (green) is stabilized by bromine interaction with
the Lys539 carbonyl oxygen of the protein backbone and with the hydroxyl group of the
Thr538 side chain. In both conformations, the GCPII entrance lid is in "closed" conformation,
enabling additional stabilization by hydrophobic interactions of the aromatic ring with the
side chains of Phe546 and Trp541 (not shown). Only the part of the inhibitor molecule for
which the electron density could be mapped is shown in C, D, and E. The PEG chain is
depicted schematically. The pictures were processed with PyMOL Molecular Graphics
System, Version 1.3.0 Schrödinger, LLCPyMOL. (color reproduction in print)
22

The structure of GCPII in complex with 20(Br)a (20(H)a modified with bromine in the
para position) revealed yet another conformation of the aromatic moiety. The entrance lid in
this structure is highly disordered, and the inhibitor position is stabilized by bromine
interactions with water molecules and by a hydrogen bond between the inhibitor’s terminal
carbonyl group and the Tyr234 hydroxyl group (data not shown).
4.5 Visualization of GCPII on cancer cells using specific nanocarriers
We prepared GCPII-specific nanocarriers from commercially available Qdot® 605
streptavidin conjugate and 22b. As shown in Figure 6, such nanoparticles specifically
recognized LNCaP cells (endogenously expressing GCPII) and not PC-3 cells (lacking GCPII
expression). Additionally, Qdots without 22b bound to neither LNCaP nor PC-3 cells. These
results support the concept that a small-molecule GCPII inhibitor could be used for
nanoparticle targeting and visualization of GCPII-positive cells.
Figure 6: Confocal images of LNCaP and PC-3 cells treated with Qdot® 605 streptavidin
conjugate (red color in the picture) with 22b to visualize GCPII. Cells were treated with
Qdots with or without bound inhibitor 22b for 15 minutes. Confocal images were merged
with bright field images and processed using ZEN 2012 software. All images were taken in
23

the same microscope set-up. For more details, see the Experimental section. (color
reproduction in print)
5. Discussion
We set out to develop low-molecular weight inhibitor of GCPII as a molecular
recognition tag suitable for facile modification of various nanoparticles through well-
established biotin-avidin interaction. We performed structure-assisted design of this tag in
order to maximize its affinity towards GCPII upon binding to the nanoparticle.
First, we mapped the effects of modifying our inhibitor scaffold, Glu-NH-C(O)-NH-
Lys (2), at the lysine ε-amino group. Acetylation of the amine led to a major improvement in
inhibitor potency, confirming the fact that GCPII does not favor positively charged amino
acids in the P1 and P1′ positions34. In agreement with the findings of Zhang et al.49, we
observed a drop in inhibitory activity after addition of a PEG linker (in proportion to the
number of PEG units). Most likely, the long disordered PEG chain entropically disfavors
inhibitor binding to GCPII. We attempted to reduce the negative effect of the PEG chain
linker by shortening it to a minimal length. It has already been shown that 12 PEG units are
sufficient and 4 PEG units are insufficient for interaction of GCPII with streptavidin17, 33. In
this work, we found out that even 8 PEG units in the linker chain are insufficient to mediate
proper connection of GCPII and the biotin-binding molecule.
To further improve inhibitor potency, we attempted to utilize the “S1-accessory
hydrophobic pocket”30. We chose two different approaches (branched and linear) to introduce
an aromatic moiety and PEG linker or acetyl moiety into the inhibitor molecule. In the linear
setup, we chose 3 different moieties to link the aromatic ring with the inhibitor scaffold.
These moieties differ in length (1 atom for the carbonyl vs. 2 atoms for the amide and
methylene carbonyl moieties), rigidity (planar carbonyl and amide vs. methylene carbonyl),
24

and hybridization. Amide attachment of the aromatic ring yielded inhibitors (acetylated and
PEGylated) with the best potencies, showing that 2 atoms and a more rigid linker improves
inhibitor potency (additionally, the nitrogen atom may participate in hydrogen bonding and
thus improve inhibitory activity). This observation is in agreement with previously published
data32. The PEG linker was attached to the ortho position of the aromatic moiety. We chose
the ortho position based on visual examination of the published crystal structure (PDB entry
3D7H) of GCPII in complex with the inhibitor DCIBzL30.
Halogen bonds are increasingly recognized as important contributors to the specific
binding of drug-like molecules to their targets50-51. Maresca et al. showed that halogen
substitution on the aromatic ring in urea-based inhibitors can lead to significant improvement
of inhibitor potency32. Therefore, we prepared inhibitors with the aromatic ring modified in
the para position with halogen atoms (for ease of synthesis, we decided to initially test the
hypothesis with 20(H)a). Modification with bromine (20(Br)a) gave the best results, suggesting
that the positive effect of a halogen atom on inhibitor potency is spatially restricted. Even
though the results from crystallographic studies showed quite different binding modes for
20(Br)a compared to 22a, 22b, and 28b, we also observed a positive effect of bromine addition
to the aromatic ring for the latter compounds, suggesting that the effect of this substitution is
generally applicable. Bromine substitutions could therefore be utilized in future design of
GCPII inhibitors with similar scaffolds.
The X-ray structures of GCPII in complex with various inhibitors revealed that the
aromatic moiety was targeted inside the "S1-accessory hydrophobic pocket" in only one case
(22b). This inhibitor showed the best inhibitory activity among the PEGylated inhibitors. The
acetylated form of this inhibitor (22a) had the lowest Ki value of all inhibitors tested. The 4-
bromobenzyl moiety of 22a does not bind inside the arginine patch, as aimed for, but
protrudes into the wide accessory tunnel. The reason for this binding mode may be the fact
25

that the opening of the arginine patch requires repositioning of the Arg463 side chain, which
is typically strongly coordinated with the Glu457 and Asp465 side chains. Moreover, the 4-
bromobenzyl moiety displaces several water molecules that would otherwise be strongly
coordinated in the accessory tunnel, making this conformation more entropically favorable. In
the case of 22b, the long PEG chain forbids such a conformation (the acetyl moiety of 22a
points inside the accessory binding tunnel), and therefore, the 4-bromobenzyl moiety adopts
another energetically favorable position – inside the S1-accessory hydrophobic pocket.
Even though inhibitors with the PEG linker emerging from the aromatic ring (9b, 13b,
20(H)b, 28b) do not utilize the S1-accessory hydrophobic pocket, the most promising one
(28b) possesses significantly more potent inhibitory activity than the lead structure (5). With
the exception of 9b, the electron density for the aromatic ring was visible in all the crystal
structures. Generally, the binding mode of the aromatic moiety is mediated through the highly
flexible GCPII "entrance lid" (Trp541 - Gly548), which in closed conformation creates a
hydrophobic binding pocket for the inhibitor aromatic moiety, which can be stabilized by the
side chains of Trp541 and Phe546. This observation is supported by the fact that, in structures
with clearly defined electron density for the aromatic moiety (13b, 20(H)b, 28b), the entrance
lid is found in closed conformation, while in the structure with poorly defined electron density
of the aromatic moiety (9b), it adopts the open conformation. It should be noted that in all
structures except the complex with 20(H)b the entrance lid is highly flexible with high B
factors and poor electron density maps of amino acid side chains. The various binding modes
of the inhibitors studied clearly show and confirm the ability of the GCPII accessory binding
tunnel (mainly the entrance lid) to adopt diverse conformations and thus form various
transient exosites.
Finally, we conducted a proof-of-concept experiment showing that small molecules
can be used for nanoparticle targeting. We prepared GCPII-specific nanoparticles bearing the
26

novel inhibitor 22b as a homing device. We chose Qdot® 605 streptavidin conjugate for its
commercial availability. After conjugation with 22b, the nanoparticles were able to
specifically bind GCPII. These nanoparticles specifically recognized GCPII on the surface of
prostate cancer cell lines endogenously expressing GCPII (LNCaP cells) and did not bind to a
prostate cancer cell line that lacks GCPII expression (PC-3 cells; compare Fig. 6). This simple
experiment confirms our initial assumption that our engineered inhibitors can function as
simple GCPII-specific homing devices for nanoparticles.
Generally, nanoparticles using the inhibitors as specific targeting moieties will face the
similar problem of cross-reactivity as the antibodies. For antibodies, the similarity of primary
and secondary structures predetermines a possible cross-reactivity of their original antigen
while in the case of inhibitor-guided nanoparticles, a substrate specificity will play a key role
in assessing the possible cross-reactivities. In this particular case, we can expect that the close
GCPII homologs which were shown to possess similar enzymatic activity (e.g., mouse GCPII
(Knedlik et al. manuscript in preparation) or human GCPIII52) will most probably be also
recognized by our engineered nanoparticles. On the other hand, even though the homologs
have the same substrate binding side, they may differ slightly in the conformation of their
exosites. Therefore by further fine-tuning of human GCPII specific inhibitors we will likely
also diminish their cross-reactivity towards GCPII homologs.
The aim of this study, preparation of a more potent inhibitor that could utilize the S1-
accessory hydrophobic pocket of GCPII and also enable functional connection of GCPII to a
biotin-binding molecule, was successful. The final inhibitor 22b possesses 7-fold greater
inhibitory potency than the starting inhibitor 5. Even though a 7-fold improvement in
inhibitory activity may seem small, these types of inhibitors are intended to be used as part of
a larger complex (e.g., nanoparticles) to ensure the specificity of the complex for its target (in
this case, GCPII). Ideally, such a complex will contain more than one inhibitor molecule. In
27

this scenario, due to the multiple binding effects, the final inhibitory potency of the complex
is likely to be significantly higher than that of the inhibitor itself. Therefore, it can be
expected that even a minor improvement in the small-molecule inhibitor’s potency will lead
to more significant improvement in the overall binding of the final complex.
6. Conclusion
We have prepared a series of GCPII inhibitors enabling effective connection of this
molecule to any biotin-binding protein or nanoparticle. Through SAR study of the distal part
of inhibitor we improved its affinity towards GCPII. The X-ray structure analysis showed that
the accessory binding tunnel of GCPII is very flexible and is able to adopt several different
conformations, thus creating various exosites that can bind the distal part of the inhibitor
molecule. Finally, we successfully tested the ability of our designed inhibitor to effectively
modified biotin-binding nanoparticle (Qdot) and function there as GCPII specific anchor.
Acknowledgement
The authors would like to thank Jana Starková and Karolína Šrámková for expert technical
help and Hillary Hoffman for language editing. We also acknowledge the Ministry of
Education of the Czech Republic, Grant No. P208-12-G016 (Center of Excellence) and OPPK
project CZ.2.16/3.1.00/24016 for financial support.
28

References and notes
1. Siegel, R.; Naishadham, D.; Jemal, A. Cancer Statistics, 2012. Ca-a Cancer Journal
for Clinicians 2012, 62, 10-29.
2. Foss, C. A.; Mease, R. C.; Cho, S. Y.; Kim, H. J.; Pomper, M. G. GCPII Imaging and
Cancer. Current Medicinal Chemistry 2012, 19, 1346-1359.
3. Ghosh, A.; Heston, W. D. W. Tumor target prostate specific membrane antigen
(PSMA) and its regulation in prostate cancer. Journal of Cellular Biochemistry 2004, 91, 528-
539.
4. Robinson, M. B.; Blakely, R. D.; Couto, R.; Coyle, J. T. Hydrolysis of the Brain
Dipeptide N-Acetyl-L-Aspartyl-L-Glutamate - Identification and Characterization of a Novel
N-Acetylated Alpha-Linked Acidic Dipeptidase Activity from Rat-Brain. Journal of
Biological Chemistry 1987, 262, 14498-14506.
5. Pinto, J. T.; Suffoletto, B. P.; Berzin, T. M.; Qiao, C. H.; Lin, S. L.; Tong, W. P.; May,
F.; Mukherjee, B.; Heston, W. D. W. Prostate-specific membrane antigen: A novel folate
hydrolase in human prostatic carcinoma cells. Clinical Cancer Research 1996, 2, 1445-1451.
6. Mhawech-Fauceglia, P.; Zhang, S.; Terracciano, L.; Sauter, G.; Chadhuri, A.;
Herrmann, F. R.; Penetrante, R. Prostate-specific membrane antigen (PSMA) protein
expression in normal and neoplastic tissues and its sensitivity and specificity in prostate
adenocarcinoma: an immunohistochemical study using mutiple tumour tissue microarray
technique. Histopathology 2007, 50, 472-483.
7. Haffner, M. C.; Kronberger, I. E.; Ross, J. S.; Sheehan, C. E.; Zitt, M.; Muhlmann, G.;
Ofner, D.; Zelger, B.; Ensinger, C.; Yang, X. M. J.; Geley, S.; Margreiter, R.; Bander, N. H.
Prostate-specific membrane antigen expression in the neovasculature of gastric and colorectal
cancers. Human Pathology 2009, 40, 1754-1761.
8. Liu, H.; Rajasekaran, A. K.; Moy, P.; Xia, Y.; Kim, S.; Navarro, V.; Rahmati, R.;
Bander, N. H. Constitutive and antibody-induced internalization of prostate-specific
membrane antigen. Cancer Research 1998, 58, 4055-4060.
9. Hinkle, G. H.; Burgers, J. K.; Neal, C. E.; Texter, J. H.; Kahn, D.; Williams, R. D.;
Maguire, R.; Rogers, B.; Olsen, J. O.; Badalament, R. A. Multicenter
29