2-Hexylthio-β,γ-CH2-ATP is an effective and selective NTPDase2 inhibitor

Article abstract of DOI:10.1021/jm401933c

NTPDase2 catabolizes nucleoside triphosphates and consequently, through the interaction of nucleotides with P2 receptors, controls multiple biological responses. NTPDase2 inhibitors could modulate responses induced by nucleotides in thrombosis, inflammation, cancer, etc. Here we developed a set of ATP analogues as potential NTPDase inhibitors and identified a subtype-selective and potent NTPDase2 inhibitor, 2-hexylthio-β,γ-methylene-ATP, 2. Analogue 2 was stable to hydrolysis by NTPDase1, -2, -3, and -8. It inhibited hNTPDase2 with K_i 20 μM, while only marginally (5-15%) inhibiting NTPDase1, -3, and -8. Homology models of hNTPDase1 and -2 were constructed. Docking and subsequent linear interaction energy (LIE) simulations provided a correlation with r² = 0.94 between calculated and experimental inhibition data for the triphosphate analogues considered in this work. The origin of selectivity of 2 for NTPDase2 over NTPDase1 is the thiohexyl moiety of 2 which is favorably located within a hydrophobic pocket, whereas in NTPDase1 it is exposed to the solvent.

Full text of DOI:10.1021/jm401933c

Article
pubs.acs.org/jmc
2‑Hexylthio-β,γ-CH2‑ATP is an Effective and Selective NTPDase2
Inhibitor
Irina Gillerman,†,⊥ Joanna Lecka,‡,§,⊥ Luba Simhaev,†,⊥ Mercedes N. Munkonda,‡,§ Michel Fausther,‡,§
Mireia Martín-Satue,‡,§,∥ Hanoch Senderowitz,*,† Jean Sevigny,*,‡,§ and Bilha Fischer*,†
́
́
†Department of Chemistry, Bar-Ilan University, Ramat-Gan 52900, Israel
‡Dep
artement de microbiologie-infectiologie et d’immunologie, Faculte de Med
§Centre de recherche du CHU de Queb
ec, Canada
́
́
́ ́ ́ ́
ecine, Universite Laval, Quebec City, Quebec, Canada
́
́ ́
ec, Quebec City, Queb
∥Department of Pathology and Experimental Therapeutics, University of Barcelona, Barcelona, Spain
S
* Supporting Information
ABSTRACT: NTPDase2 catabolizes nucleoside triphosphates and consequently,
through the interaction of nucleotides with P2 receptors, controls multiple biological
responses. NTPDase2 inhibitors could modulate responses induced by nucleotides in
thrombosis, inflammation, cancer, etc. Here we developed a set of ATP analogues as
potential NTPDase inhibitors and identified a subtype-selective and potent NTPDase2
inhibitor, 2-hexylthio-β,γ-methylene-ATP, 2. Analogue 2 was stable to hydrolysis by
NTPDase1, -2, -3, and -8. It inhibited hNTPDase2 with Ki 20 μM, while only marginally
(5−15%) inhibiting NTPDase1, -3, and -8. Homology models of hNTPDase1 and -2
were constructed. Docking and subsequent linear interaction energy (LIE) simulations provided a correlation with r2 = 0.94
between calculated and experimental inhibition data for the triphosphate analogues considered in this work. The origin of
selectivity of 2 for NTPDase2 over NTPDase1 is the thiohexyl moiety of 2 which is favorably located within a hydrophobic
pocket, whereas in NTPDase1 it is exposed to the solvent.
actions and influence electrical impulse conduction as well as
metabolic processes.11,14 Thus, a selective inhibitor for NTPDase2
may exhibit cardioprotective properties.
INTRODUCTION
■
Hydrolysis of extracellular nucleotides plays a critical role in
physiological phenomena, by creating a balance between the
excitatory actions of ATP (and ADP) at P2 receptors and the
often inhibitory actions of its breakdown product, adenosine,
at P1 receptors. NTPDases, EC 3.6.1.5,1−3 are responsible for
the rapid hydrolysis of the extracellular nucleotides.3 Eight
members of the E-NTPDase family are known (NTPDase1−8)
among which four are membrane-bound glycoproteins with an
extracellular active site (NTPDase1−3, −8).52 NTPDase2 is
expressed in several tissues. In the vasculature it is expressed in
the adventitia of arteries and on subendothelial cells of veins
where it was proposed to facilitate platelet aggregation.6 Indeed,
upon injury, large amounts of ATP are secreted to the extra-
cellular milieu and are converted to the platelet activation
factor, ADP, by NTPDase2, thus activating P2Y1 and P2Y12
receptors on thrombocytes.6
In the central nervous system, NTPDase2 is expressed on
astrocytes.11 During cerebral ischemia, extracellular ATP con-
centrations rise and it is predominantly hydrolyzed by
NTPDase2. The resulting ADP is a potent inhibitor of ecto-
5′-nucleotidase. This enzyme dephosphorylates AMP to the
cytoprotective adenosine.12 Thus, NTPDase2 inhibitors may
be useful for the treatment of ischemic conditions in brain.
There is an accumulating evidence of overexpression of
NTPDase2 in gliomas.13 Because ADP derived from NTPDase2
activity stimulates platelet migration to the tumor area,
NTPDase2, by regulating angiogenesis and inflammation,
seems to play an important role in tumor progression.
Taken together, selective and potent NTPDase2 inhibitors
may serve as potential drug candidates for the treatment of
health disorders such as thrombosis, ischemia, and cancer.
Ideally, NTPDase inhibitors should have no effect on P2
receptors and should be stable to hydrolysis by ectonucleoti-
dases. So far only a few NTPDase inhibitors have been described
in the literature. Recently we reported on BuS-adenine
nucleotide analogues that specifically inhibit NTPDase114,15
and are not agonists of P2Y receptors. Other non P2 receptor
agonist reported inhibitors are ARL 67156,16,17 polyoxometalates
An imbalanced ATP/ADP hydrolysis ratio, in favor of ATP,7
was observed in patients with coronary artery disease leading
to prothrombotic ADP levels which could not be effectively
removed. In addition, ADP (as well as ATP and ADP-α,β-CH2)
potently inhibits ecto-5′-nucleotidase,4−7 thereby preventing the
extracellular formation of adenosine and impeding adenosine
receptor-effected vasodilation and antithrombotic properties.8−10
Specific localization of NTPDase2 in the heart suggests
that this enzyme may regulate functions other than platelet
activation. Extracellular ATP, and its degradation product,
adenosine, exert pronounced inotropic and chronotropic
Received: December 18, 2013
Published: June 27, 2014
© 2014 American Chemical Society
5919
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
(POMs),15 and PSB-6426 (Ki 8.2 μM).18 However, these
compounds are not NTPDase-subtype selective.
δ: 8.35 (s, 1H, H2), 6.1 (d, 1H, J = 6.0 Hz,, H1′), 4.54 (t, 1H, J = 2.4 Hz,
H2′), 4.36 (m, 1H, H3′), 4.2 (m, 1H, H4′), 3.1 (m, 2H, H5′, H5″),
3 (t, 2H, J = 15.0 Hz, SCH2), 2.3 (t, 2H, J = 20.4 Hz CH2 phosphonate),
1.73 (m, 2H, SCH2CH2), 1.45 (m, 2H, SCH2CH2CH2), 1.2 (m, 2H,
SCH2CH2CH2CH2), 0.96 (m, 2H, SCH2CH2CH2CH2CH2), 0.90 (3H,
CH3) ppm; 31P NMR (D2O, 121 MHz, pH 8.5) δ: 15 (s), 9.8 (m),
−10.2 (d) ppm; MS (TOF ES−): 617 (M−H+); HRMS: calcd for
C17H26N5O12P3S 617.4031 found, 617.5034.
Here, we report on the synthesis and evaluation of a series of
adenosine 5′-tri- and -monophosphate analogues as potential
and selective inhibitors of NTPDase2. Specifically, we studied
the inhibitory effect of analogues 1−9 on NTPDase-1, -2, -3, -8
and evaluated their selectivity vs NPP1 and NPP3. Next, we
determined the Ki value of the most promising and selective
NTPDase2 inhibitor and studied its inhibition of ATPase
activity on various cell surfaces. Finally, we studied computa-
tionally the most active and the least active compounds
identified here together with NTPDase1 inhibitors previously
reported by us, and tested for NTPDase2 inhibition,17 to
develop a model for predicting NTPDase2 inhibition by
nucleotide analogues.
N6′,N6″-Diethyladenosine, 16. 6-Chloropurine riboside 2′,3′,5′-
triacetate (382 mg, 0.93 mmol) was dissolved in DMF (7 mL), and
diethylamine was added (290 μL, 2.8 mmol, 3 equiv). The reaction
mixture was stirred for 24 h at 60−70 °C in a sealed tube. The solvent
was evaporated, and the residue was triturated with ethanol. The
mixture was separated on a silica gel column eluted with chloroform.
The product was obtained in a 27% yield (80 mg). 1H NMR (CD3OD,
300 MHz) δ: 8.20 (s, 1H, H8), 8.10 (s, 1H, H2), 6.00 (d, 1H, J =
5 Hz, H1′), 4.90 (br s, 4H) 4.75 (m, 1H, J = 5 Hz, H2′), 4.2 (m, 1H,
H3′), 3.87 (m, 2H, H5′, H5″), 3.85 (m, 3H, H4′, NCH2) 1.26 (t, J =
6.9 Hz, 6H) ppm; 13C NMR (CD3OD, 50.3 MHz) δ: 154.31 (C-6),
152.2 (C-2), 150.9 (C-4), 141.58 (C-2), 120.3 (C-5), 89.3 (C-1′), 86.4
(C-4′), 74.13 (C-2′), 72.6 (C-3′), 62.4 (C-5′) 43.40 (2CH2), 13.01
(2CH3) ppm; MS (TOF ES+): 324 (M−H+). HRMS: calcd for
C14H21N5O4 323.1476 found, 323.1608.
EXPERIMENTAL SECTION
■
General. All air- and moisture-sensitive reactions were carried out
in flame-dried, argon-flushed, two-necked flasks sealed with rubber
septa, and the reagents were introduced with a syringe. TLC analysis
was performed on precoated Merck silica gel plates (60F-254).
Visualization was accomplished with UV light. New compounds were
characterized (and resonances assigned) with proton nuclear magnetic
resonance using AC-200 spectrometers, Bruker DPX-300, or DMX-
N6′,N6″-Dibenzyladenosine, 17. 6- Chloropurine riboside 2′,3′,5′-
triacetate (296 mg, 0.64 mmol) was dissolved in DMF (7 mL), and
dibenzylamine was added (371 μL, 1.93 mmol, 3 equiv). The reaction
mixture was stirred for 24 h at 60−70 °C. The mixture was evaporated
and separated on a silica column with 90:10 CHCl3:MeOH. The product
was obtained in a 44% yield (127 mg). Spectroscopic data were
consistent with that from the literature.21
1
600. H NMR spectra were recorded in D2O, and the chemical shifts
are reported in ppm relative to HDO (4.78 ppm) as an internal
standard. Nucleotides were also characterized with 31P NMR in D2O,
using 85% H3PO4 as an external reference, on a AC-200 spectrometer
at pH 8. All final products were characterized on an AutoSpec-E Fision
VG high-resolution mass spectrometer with chemical ionization and
high-resolution mass spectrometry. Nucleotides were desorbed from a
glycerol matrix under FAB (fast atom bombardment) conditions in low
and high resolution. Primary purification of the nucleotides was
achieved on an LC (Isco UA-6) system using a Sephadex DEAE-A
25 column, which was swelled in 1 M NaHCO3 at RT overnight. Final
purification of the nucleotides was achieved on a high-performance
liquid chromatography (HPLC) (Merck-Hitachi) system with a semi-
preparative reverse-phase (Gemini 5 μm C-18 110 Å 250 × 10 mm,
Phenomenex, Torrance, CA). For analytical purposes, a Gemini (5 μm,
C-18, 110 Å, 150 × 4.6 mm, Phenomenex, Torrance, CA) was used.
The purity of the nucleotides was evaluated on an analytical column in
two different solvent systems. Peaks were detected by UV absorption
using a diode array detector. Solvent system I consisted of (A) CH3CN
and (B) 0.1 M TEAA (pH 7). Solvent system II consisted of (A) 5 mM
tertabutylammonium phosphate (TBAP) in methanol and (B) 60 mM
ammonium phosphate and 5 mM TBAP in 90% water/10% methanol.
See gradient details below. 2-Hexylthioadenosine22 and 8-azaadeno-
sine19 were prepared according to literature methods.
Phosphorylation Reaction. General Procedure. Nucleotides 1,
6, 7, and 11 were prepared by our previously published procedure.22
8-Aza-2-hexylthioadenosine 5′-triphosphate, 1,. was obtained as
an off-white solid at a 52% yield (100 mg, 0.16 mmol) from 8-aza-2-
hexylthioadenosine (120 mg, 0.31 mmol) yield. Final separation was
achieved on HPLC, applying a linear gradient of TEAA/CH3CN 70:30
to 55:45 in 25 min (flow rate 6 mL/min), tR 14 min (92% purity).
Compound purity was evaluated by applying a linear gradient of
50 mM phosphate buffer (pH = 7)/CH3CN 70:30 to 60:40 in 25 min
(flow rate 1 mL/min) tR 4.5 min (90% purity). 1H NMR (D2O,
300 MHz) δ: 6.15 (d, 1H, J = 4.5 Hz, H1′), 4.49 (t, 1H, J = 4.8 Hz,
H2′), 4.43 (m, 1H, H3′), 4.3 (m, 1H, H4′), 3.12 (m, 2H, H5′, H5″),
3.03 (m, 2H, SCH2), 1.73 (m, 2H, SCH2CH2), 1.45 (m, 2H,
SCH2CH2CH2), 1.2 (m, 2H, SCH2CH2CH2CH2), 0.96 (m, 2H,
SCH2CH2CH2CH2CH2), 0.9 (3H, CH3) ppm; 31P NMR (D2O, 81
MHz, pH 8.5) δ: −23 (t), 10.2 (d), −9.8 (d) ppm; MS (TOF ES−):
620 (M+H−). HRMS: calcd for C15H23N6O13P3S, 620.3640; found,
620.3420.
N6′,N6″-Diethyladenosine 5′-triphosphate, 6,. was obtained as a
white solid at a 73% yield (81 mg, 0.15 mmol) starting from N6′,N6″-
diethyladenosine, 21 (65 mg, 0.2 mmol). Final purification was
achieved on HPLC, applying a linear gradient of TEAA/CH3CN 88:12
to 78:22 in 25 min (flow rate 6 mL/min), tR 14 min (99.8% purity);
linear gradient of phosphate buffer/CH3CN 95:5 to 90:10 tR 4 min
2-Hexylthioadenosine-5′-(β,γ-methylene-triphosphate), 2.
2-Hexylthioadenosine20 (120 mg, 0.31 mmol) was dissolved in dry
trimethyl phosphate (1 mL) and cooled to −15 °C. Proton sponge
was added (105 mg, 0.49 mmol, 1.6 equiv). After the mixture was
stirred for 20 min, POCl3 (85 μL, 0.93 mmol, 3 equiv) was added
dropwise, and the reaction mixture was stirred for 2 h. A solution of
1 M bis-tributylammionium salt of methylene diphosphonic acid in
DMF (2.2 mL, 2.2 mmol, 7 equiv) and Bu3N (300 μL, 1.26 mmol,
4 equiv) was added at once. After the mixture was stirred for 6 min,
1 M TEAB (5 mL) was added, and the clear solution was stirred for
45 min at RT. The solution was freeze-dried. The semisolid obtained
after freeze-drying was chromotographed on an activated Sephadex
DEAE-A 25 column using a two-step gradient of NH4HCO3 (200 mL
of 0−0.2 M and then 350 mL of 0.2−0.4 M). The relevant fractions
were freeze-dried repeatedly to yield the product as a white solid
in a 70% yield. Final separation was achieved by HPLC, applying a
linear gradient of TEAA/CH3CN 80:20 to 60:40 in 25 min (flow rate
6 mL/min), retention time (tR) 15.7 min (98% purity); linear gradient
of phosphate buffer/CH3CN 80:20 to 60:40 in 15 min (flow rate
1
(92% purity). H NMR (D2O, 300 MHz) δ: 8.46 (s, 1H, H2), 8.2 (s,
1H, H8), 6.11 (d, 1H, J = 5.5 Hz, H1′), 4.8 (t, 1H, J = 5.5 Hz, H2′),
4.45 (m, 1H, H3′), 4.35 (m, 1H, H4′), 4.2 (m, 2H, H5′, H5″), 3.8 (m,
4H, CH2), 1.4 (t, 6H, J = 6.7 Hz, CH3) ppm; 31P NMR (D2O, 81
MHz, pH 8.5) δ: −10 (d), −10.5 (d), −22.5 (t) ppm; MS (TOF ES+):
560 (M+H−). HRMS: calcd for C14H20N5O13P3, 559.0304; found,
559.0271.
N6′,N6″-Dibenzyladenosine 5′-triphosphate, 7,. was obtained
as a white solid at a 68% yield (70 mg) starting from N6′,N6″-
dibenzyladenosine (127 mg, 0.28 mmol). Final purification was
achieved on HPLC, applying a linear gradient of TEAA/CH3CN 70:30
to 55:45 in 25 min (6 mL/min), tR 14 min (90% purity) linear
gradient of phosphate buffer/CH3CN 80:20 to 60:40 in 25 min
(1 mL/min) tR 5 min (92% purity). Spectral data were consistent with
that from the literature.23
1
1 mL/min), tR 4.3 min (91% purity). H NMR (D2O, 600 MHz)
5920
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
8-Azaadenosine 5′-triphosphate,23 11,. was prepared from 8-
azaadenosine (25.7 mg, 0.096 mmol) and obtained at a 45% yield.22
Spectral data were consistent with that from the literature.25
Plasmids. The plasmids used in this study have all been described
in published reports: human NTPDase1 (GenBank accession no.
U87967),24 human NTPDase2 (NM_203468), human NTPDase3
(AF034840),25 human NTPDase8 (AY430414),26 human ecto-5′-
nucleotidase (NM_002526),27 human NPP1 (NM_006208),28 and
human NPP3 (NM_005021).29
Cell Transfection and Protein Preparation. COS-7 cells were
transfected with an expression vector (pcDNA3) containing the cDNA
encoding for each NTPDase using lipofectamine (Invritrogen) and
harvested 40−72 h later, as previously described.24 For the preparation
of protein extracts, transfected cells were washed three times with Tris-
saline buffer at 4 °C, collected by scraping in harvesting buffer (95 mM
NaCl, 0.1 mM PMSF, and 45 mM Tris, pH 7.5) and washed twice by
centrifugation (300g, 10 min, 4 °C). The cells were then resuspended
in the harvesting buffer supplemented with 10 μg/mL aprotinin to
block proteinases and sonicated. Nucleus and cellular debris were
discarded by another centrifugation (300g, 10 min, 4 °C), and the
resulting supernatant (thereafter called protein extract) was aliquoted
and stored at −80 °C until used. Protein concentration in the protein
extracts was estimated by Bradford microplate assay using bovine
serum albumin as a standard.30
Enzymatic Activity Assays. NTPDases (EC 3.6.1.5) and Ecto-5′-
nucleotidase (EC 3.1.3.5). Activity was measured as described
previously3 in 0.2 mL of incubation medium (5 mM CaCl2 and
80 mM Tris, pH 7.4) or Tris-Ringer buffer (in mM, 120 NaCl, 5 KCl,
2.5 CaCl2, 1.2 MgSO4, 25 NaHCO3, 5 mM glucose, 80 Tris, pH 7.4)
at 37 °C with or without analogues 2−9. Ectonucleotidase protein
extracts were added to the incubation mixture and preincubated
at 37 °C for 3 min. The reaction was initiated by the addition of
10−100 μM ATP or AMP for NTPDases and ecto-5′-nucleotidase,
respectively, with or without analogues and stopped after 15 min with
50 μL of malachite green reagent. The released inorganic phosphate
(Pi) was measured at 630 nm according to Baykov et al.31 The type of
inhibition was determined by Dixon32 and Cornish−Bowden35 plots
of four independent experiments, and Ki values were obtained using
nonlinear regression.
Protein extracts from nontransfected COS-7 cells exhibited ATP or
AMP hydrolysis less than 5% of that of COS-7 cells transfected with
either NTPDases or ecto-5′-nucleotidase expressing plasmid. Hence,
NTPDase or ecto-5′-nucleotidase activity of native COS-7 cells was
considered as negligible.
Figure 1. Adenosine mono- and triphosphate analogues 1−9 explored
in this study.
NPPs (EC 3.1.4.1; EC 3.6.1.9). Evaluation of the effect of 2 on
human NPP1 and NPP3 activity was carried out with p-nitro-
phenylthymidine 5′-monophosphate (pNP-TMP) as the substrate.28,33
The reactions were carried out at 37 °C in 0.2 mL of the following
incubation mixture, in mM: 1 CaCl2, 140 NaCl, 5 KCl, and 50 Tris,
pH 8.5, with or without analogues. Analogues 2 and pNP-TMP were
used at a concentration of 100 μM. Human NPP1 or NPP3 extract
was added to the incubation mixture and preincubated at 37 °C for
3 min. Reaction was started by adding pNP-TMP, and the production
of p-nitrophenol was measured at 410 nm, after 15 min. The protein
extracts from nontransfected COS-7 cells exhibited less than 5%
of pNP-TMP hydrolysis after transfection with either NPP1 or NPP3
expressing plasmid and, as such, their NPP1/3 activity was considered
as negligible.
Studio (DS) version 2.5.537 using default settings. Six conserved water
molecules within the catalytic site of the NTPDase2 were identified
by superposing four structures of Rattus norvegicus NTPDase2 (PDB
codes: 3CJ1, 3CJA, 3CJ7, 3CJ9) based on multiple sequence alignment
using DS protocols, and their coordinates were used in course of the
homology modeling procedure. Two of these water molecules are
involved in the catalytic reaction, while the additional four coordinate
the Ca2+ ion in the catalytic site.34 These water molecules are also
largely conserved in the crystal structures of Legionella pneumophila
and Toxoplasma gondii NTPDase1. The alignment, the conserved
water molecules, and the crystal ligand were used as input to the
MODELER12 program as implemented in DS version 2.5.5. Default
settings were used throughout the modeling procedure (optimization
level = high; no. of models =10; refine loops = false). The highest
ranked model was tested for stereochemical quality using Procheck42
and Prosa,38,39 and selected for subsequent analysis.
Statistical Analysis. Statistical analysis was performed with
Student’s t test (Microsoft Office OneNote 2003). P values below
0.05 were considered statistically significant.
Homology Modeling of NTPDase2. A ligand-supported model
of human NTPDase2 was built based on the crystal structure of
rat NTPDase234 (PDB code 3CJA; sequence identity to human
NTPDase2 of 82%). This template was selected because of its
favorable resolution (2.1 Å) and because it was solved in the presence
of an ATP analogue (AMPPNP) and a cofactor (Ca2+). The sequence
of the target protein (accession number Q9Y5L3) was downloaded
from the UniprotKB/Swiss-Prot35 server and aligned with the target
sequence using the ClustalW36 method as implemented in Discovery
Homology Modeling of NTPDase1. A ligand-supported model
of human NTPDase1 was built in a manner similar to that described
above for NTPDase2 using the NTPDase1 sequence (accession number
P49961) and the rat NTPDase2 structure (sequence identity to human
NTPDase1 of 44%).
Docking. Docking simulations were performed using Glide40−42
XP (extra precision sampling and scoring using the OPLS-AA force
field with energy penalties for nonplanar amide bonds; flexible ligand
5921
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
a
Scheme 1. Synthesis of 2-Hexylthio-β,γ-methylene-ATP Analogues 1 and 2
a
Reagents and conditions: (a) POCl3, proton sponge, −15 °C, 2 h; (b) (NHBu3)H2P2O7/NBu3, −10 °C, 6 min; (c) 1 M TEAB, RT, 45 min, overall
yield of 1:52%; (d) (Bu3NH)2CH2P2O6H2/NBu3, −10 °C, 6 min, overall yield for 2, 70%.
a
Scheme 2. Synthesis of N6-Substituted ATP and AMP Analogues 4 and 9
a
Reagents and conditions: (a) NHR2R3/EtOH, 100 °C, 16 h (70% for 14, 50% for 15, 27% for 16, 44% for 17); (b) POCl3, proton sponge, −15 °C,
2 h; (c) (NHBu3)H2P2O7/NBu3, −10 °C, 6 min; (d) 1 M TEAB, RT, 45 min, overall yield for steps b−d: 74% for 4, 73% for 5, 73% for 6, 68% for
7, 30% for 8, and 32% for 9.
with sampling of nitrogen inversion and ring conformation, keeping
the best 800 poses for postdocking minimization) as implemented in
Maestro 9.0,43 and were initiated from ligand-specific conformational
ensembles generated with the LMCS/MCMM conformational search
procedure as implemented in Maestro 9.0.49 Prior to docking, the
model was prepared using the protein preparation wizard in Maestro
9.0. Both lowest energy poses and poses with binding modes similar to
the crystallographic ligand (AMPPNP) were selected for further
analysis.
and its distances to four crystallographic water molecules and to the
chelating oxygens of the ligands were restrained to their initial values.
Sodium ions were added to the system to maintain charge neutrality.
For the solvated protein−ligand complexes, all atoms outside the 20 Å
were tightly restrained to their initial coordinates and excluded from
the nonbonded interactions list. While such restraints may prevent
large scale interdomain movements known to occur in NTPDases
upon ligand binding, the purpose of the LIE simulations was not to
describe the binding process but rather the binding of ligands to a
preformed site. Furthermore, because of the high similarity between
the different analogues considered in this work we expect that for all,
the binding site adopts a similar conformation. For the free ligand in
water, a weak harmonic restraint with a force constant of 10 kcal/mol·Å2
was applied to the atoms of ligands to keep it near the sphere center.
All the systems were heated from 1 to 300 K in four steps followed by
100 ps long equilibration at 300 K. During the heating phase, restraints
on heavy atoms within the simulation sphere were applied and gradually
reduced to zero.
LIE Calculations. Linear interaction energy (LIE)44,45 calculations
were performed using the program Q46 and the OPLS-AA 200147
force field implemented therein. Missing ligand parameters were
obtained from the implementation of this force field in MacroModel.
Two sets of MD simulations were performed, one for the ligands
in a 20 Å sphere of TIP3P water (free state) and one for the ligands
bound to NTPDase2 where the complex was immersed in a similar
sphere (bound state). The surface of the sphere was subjected to
radial and polarization restraints according to the SCAAS48 model as
implemented in Q. Nonbonded interactions were calculated for all
atoms within the sphere. To avoid overpolarization effects, charged
amino acids outside the simulation sphere and within 2−4 Å of the
boundary were neutralized. The Ca ion was assigned a charge of +2,
This was followed by a 3 ns long production run (among the
longest production runs reported for LIE calculations with Q). Ligand-
surrounding interaction energies were taken as an average over the
5922
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Table 1. Stability of Analogues 1−9 to Hydrolysis by Human
RESULTS
■
a
NTPDases
Rational Design of Potential NTPDase Inhibitors. To
identify NTPDase inhibitors based on an adenine nucleotide
scaffold, we explored the inhibitory effect of adenine nucleo-
tides 1−9 modified at the adenine and 5′-phosphate moieties,
Figure 1. Specifically, analogues 4−9 modified at the N6′,N6″-
position were synthesized to evaluate the recognition of the
N6-region by NTPDases. SAR study of this moiety in adenine
nucleotides was triggered by the structure of NTPDase
inhibitor ARL 67156.17 This compound exhibits its inhibitory
effect in the micromolar range and is disubstituted at the
N6 position (N6′,N6″-diethyl) of the adenine moiety. For this
purpose we substituted ATP analogues at the N6-position by
one or two methyl substitutions (analogues 4 and 5).
Furthermore, we compared the inhibitory effect of N6′,N6″-
disubstituted (Me, Et, or Bz groups) analogues (analogues 5 vs
6 and 7, respectively). In addition, we studied the effect of
the related monophosphate nucleotide analogues 8 and 9 on
NTPDase activity. The SAR of the adenine moiety was studied
as well with 8-azaadenine nucleotide analogues 1 and 3, and
2-hexylthioadenine nucleotide analogues 1 and 2. Finally, the
contributaion of Pα,β-nonhydrolyzable bridge to generation of
NTPDase inhibitor was evaluated as well, compound 2.
hydrolysis (in %) vs control
compound
NTPDase1
NTPDase2
NTPDase3
NT
NTPDase8
1
2
3
4
5
6
7
8
9
44 5.1
2.3 0.1
73 5.7
87 3.0
74 1.0
22 1.0
14 0.7
11.5 4.2
2.8 0.1
38.5 6.2
77 1.0
73 5.0
NT
0.8
NT
NT
NT
2
0
3.0 0.2
NT
NT
NT
0
3
0
3
0
0.1
0.4
0.2
20 1.0
13 0.6
0.6
0
8
0
6
0
4
2
5
0.1
0.2
0.3
0.2
10 0.6
a
Each analogue was tested as a substrate for the indicated recombinant
human NTPDase and was compared with the activity obtained with
ATP which was set as 100% and which corresponded to 680 34,
930
for NTPDase1, -2, -3, and -8, respectively. Results are expressed in
SD of three independent experiments, each performed in
56, 200
36, and 130
10 nmol Pi·min−1·mg protein−1
%
duplicate. NT = not tested.
production phase of the MD trajectory of each complex, and the
binding free energy was calculated according to eq 1:
Synthesis. A series of N6′,N6″-disubstituted ATP and AMP
analogues, C2-substituted-ATP, or -8-aza-ATP has been synthesi-
zed (Scheme 1). 2-Hexylthio-8-aza-ATP analogue, 1, was
synthesized from 2-hexylthio-8-azaadenosine, 10.49 Triphosphor-
ylation of 10 using our improved Ludwig’s conditions22 resulted
in product 1 in 52% yield (Scheme 1). 2-Hexylthioadenosine
vdW
vdW
ΔGbind = α(⟨V
⟩
− ⟨V
⟩
)
l‐s
bound
l−s free
el
l‐s
el
+ β(⟨V ⟩bound − ⟨V ⟩
)
(1)
l‐s free
where Vl‑s are MD time averages of ligand−solvent (TIP3P water in
the free state and protein + TIP3P water in the bound state) non-
bonded interactions. The “el” and “vdW” terms refer to electrostatic
and van der Waals interactions, respectively. In accord with the
original LIE publications,45 the scaling parameters were set to α = 0.18,
β = 0.5. LIE simulations were initiated from the selected poses as
described above.
5′-β,γ-methylene triphosphate, 2, was obtained in 70% yield starting
20
from 2-hexylthioadenosine.
8-Aza-ATP, 3, was synthesized
according to literature methods.19 6 -Chloroadenosine-2′,3′,5′-
acetate, 13, was used as a common synthetic intermediate for
the preparation of N6-substituted-adenosine 5′-triphosphate and
Figure 2. Influence of adenosine nucleotide derivatives 6−9 on recombinant NTPDases activities. Enzymatic assays were carried out using protein
extracts from COS-7 cells transfected with an expression vector encoding each enzyme separately. Both substrate and derivatives were used at
100 μM. The 100% activity was set with ATP alone that was 1270 35, 928 55, 202 37, and 129 11 [nmol Pi·min−1·mg protein−1] for
NTPDase1, -2, -3, and -8, respectively. Data are presented as the mean SD of three experiments carried out in triplicate.
5923
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
monophosphate derivatives, 4−9 (Scheme 2). N6′,N6″-Dieth-
yladenosine, 6, was obtained in 27% yield from 13 by treatment
with diethylamine in DMF at 70 °C for 24 h in a sealed tube.
These conditions also resulted in the removal of the ribose
protecting groups. Similarly, N6′,N6″-dibenzyladenosine, 17, was
obtained in 44% yield upon treatment of 13 with dibenzylamine.
N6-Methyl- and N6′,N6″-dimethyladenosine 5′-triphosphate, 4
and 5, were prepared as we previously reported (Scheme 2).50
N6′,N6″-Dimethyladenosine 5′-monophosphate, 8, was obtained
as a byproduct of the triphosphorylation reaction of N6′,N6″-
dimethyladenosine. Compounds 8 and 9 were obtained in 73%
and 68% yield, respectively, from N6′,N6″-dimethyladenosine
and N6′,N6″-diethyladenosine (Scheme 2).51
Effect of Nucleotide Derivatives on Recombinant
Ectonucleotidase Activity. As described in Experimental
Section the protein extracts of nontransfected COS-7 cells
showed less than 5% of the activity obtained with COS-7 cells
transfected with each ectonucleotidase (NTPDases, ecto-5′-
nucleotidase, and NPPs), thus allowing the analysis of each
ectonucleotidase in its native membrane-bound form. Previous
work demonstrated that NTPDase1 and NTPDase2 possess
similar ATPase molecular activity.52 In our assays here we
had a similar level of ATPase activities of both NTPDase1 and
NTPDase2, suggesting that similar levels of each NTPDase
were used. Western blot confirmed the expression of each
protein (data not shown).
Out of the nine analogues studied here, analogues 2, 6, 7, 8,
9 were reasonably stable to hydrolysis by human NTPDase1−3
and -8 (Table 1). However, the ATPase activity assays per-
formed with equal concentrations (100 μM) of ATP and these
analogues revealed that N6′,N6″-diethyladenosine, 6 (Figure 2),
and 2-hexylthio-β,γ-CH2-ATP, 2 (Figure 3A) inhibited NTPDase2
(about 40% and 70%, respectively) while the other three
analogues were either extremely weak inhibitors of NTPDase2
(7 and 9) or did not inhibit it (8) (Figure 2). 2-Hexylthio-β,γ-
CH2-ATP inhibited human NPP1 activity by >50% (Figure 3B).
The enzymatic activity of human ecto-5′-nucleotidase was barely
affected by compound 2 (10% inhibition of AMPase activity).
Kinetic analysis was performed only for compound 2, which
exhibited the best inhibitor characteristics. We have tested the
kinetic parameters (Ki and IC50) of NTPDase2 inhibition with
ATP as substrate. The inhibition constant (Ki) was evaluated
to be 21 μM, while the IC50 was 40 μM. Cornish−Bowden
plot analyses showed a mixed-type inhibition which was mainly
competitive (Figure 4A,B) and therefore partly reversible.
Homology Modeling. To create a predictive tool for the
design of novel NTPDase2 inhibitors, we studied the two most
potent and the two least active analogues in this work
(analogues 2, 6, 7, and 9), as well as several other NTPDase1
inhibitors recently reported by us and also tested for NTPDase2
inhibition (analogues 18, 19, 20; Figure 5).17 A model of human
NTPDase2 was generated based on the X-ray crystal structure of
rat NTPDase2 (3CJA)37 (Figure 6). The model was found to
have good stereochemical quality (99.5% of residues residing in
the most favored and additional favored regions of the
Ramachandran plot and an overall G-factor of −0.05) and a
Prosa profile similar to that of the crystal structure of the
template (Figure S1, Supporting Information). The template/
target main chain RMSD was found to be 0.48 Å, suggesting
that the model has the correct NTPDase2 topology. The
structure of the model is consistent with the nucleotide binding
pattern of NTPDase2 and its proposed catalytic mechanism.34
The catalytic binding site in the NTPDase2 model contains one
Figure 3. Influence of 2-hexylthio-β,γ-CH2-ATP, 2, on recombinant
ectonucleotidase activities. Enzymatic assays were carried out with
protein extracts from COS-7 cells transfected with an expression
vector encoding each enzyme separately. All substrates (ATP, AMP, or
pNP-TMP) and 2 were used at 100 μM. For all panels, data are
presented as the mean
triplicate. (A) 100% Activity was set with ATP alone: 1270
928 55; 202 37; 129
11 [nmol Pi·min−1·mg protein−1] for
SD of three experiments carried out in
35;
human NTPDase1, -2, -3, and -8, respectively. (B) The activity of
human NPP1 and NPP3 with pNP-TMP as substrate and human ecto-
5′-nucleotidase with AMP as substrate in the absence of compound 2
was set as 100% of activity, which was 30
2 and 61
4 [nmol
p-nitrophenol·min−1·mg protein−1] for NPP1 and NPP3, respectively,
and 1010
33 [nmol Pi·min−1·mg protein−1] for human ecto-5′-
nucleotidase.
calcium ion, which is coordinated by four water molecules and
two phosphate oxygens of the ligand (Figure 6C). The adenine
base forms a π−π interaction with Tyr350 and a π−cation
interaction with Arg394. The 3′ hydroxyl group is hydrogen
bonded to Arg245. Several hydrogen bonds are formed between
the phosphate chain and the backbone amides or side chains of
several residues: Ser48, Ser49, His50, Thr122, Gly204, Ala205,
and Ser206 (Table S1, Supporting Information).
To identify structural factors which may contribute to ligand
selectivity, a homology model of human NTPDase1 was generated.
The selection of template for this model merits some discussion.
Three crystal structures of NTPDase1 are currently available in
the PDB, namely rat (3ZX0),53 Legionella pneumophila (4BRA)54
5924
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Figure 5. Structures of 8-BuS-ATP and 8-BuS-ADP derivatives, 18−
20, which were originally reported in ref 18 and which were subjected
to docking and LIE simulations in this study.
stereochemical quality (98.5% of residues residing in the most
favored and additional favored regions of the Ramachndran plot
and an overall G-factor of −0.1), a Prosa profile similar to that
of the crystal structure of the template (Figure S1, Supporting
Information), and a template/target main chain RMSD value of
0.62 Å. A sequence and structural comparison between the two
models is presented in Figure 7. Both models have very similar
binding sites with residues forming stabilizing interactions with
the crystallographic ligand (AMPPNP) in NTPDase2 highly
conserved in NTPDase1 (residues Ser48, Ser49, His50, Thr122,
and Ala205 in NTPDase2 correspond, respectively, to residues
Ser57, Ser58, His59, Thr131, and Ala217 in NTPDase1; see
Table S1, Supporting Information), yet they differ by several
residues both within and in close vicinity to the site. In particular,
Arg245, Tyr350, Asp353, and Arg394 in NTPDase2 are
respectively replaced by Lys258, Phe365, Lys368, and Tyr408
in NTPDase1. While the first two changes are conservative
(Arg → Lys, Tyr → Phe), the last two may affect ligand binding
and form the basis for ligand selectivity.
Docking Simulations. Docking simulations for all ligands
were initiated from conformational ensembles generated by the
conformational search procedure. This approach provides the
docking tool with multiple starting points including different
ring conformations. Docking simulations with Glide success-
fully reproduced the crystallographic pose (i.e., binding mode)
of AMPPNP in the rat NTPDase2 binding site as the lowest
energy structure with an heavy atoms RMSD of 0.61 Å (data
not shown).
Figure 4. Ki determination for 2-hexylthio-β,γ-CH2-ATP, 2, by Dixon
(A) and Cornish−Bowden (B) plot. ATP concentration was 50, 100, and
250 μM, in the presence of 0, 10, 50, 100, and 250 μM compound 2.
Data are presented as the mean SD of three experiments carried out in
triplicate.
and Toxoplasma gondii (4KH6). The rat NTPDase structure
shares the highest sequence identity with the human protein
(73%), yet it was solved in the absence of a nucleotide analogue.
The Legionella pneumophila and Toxoplasma gondii structures
were solved in the presence of a metal cofactor and a nucleotide
analogue (AMPNP and AMPPNP for Legionella pneumophila
and Toxoplasma gondii, respectively), yet they share a much
lower sequence identity with human NTPDase1 (25% and 18%
for Legionella pneumophila and Toxoplasma gondii, respectively).
An initial attempt to build a model for human NTPDase1 based
on the Legionella pneumophila structure resulted in a model
with a poor stereochemical quality, presumably due to the low
template-target sequence identity. A second attempt at model
construction was performed based on the rat NTPDase1
template. However, docking of analogue 2 into this model
resulted in an orientation of the triphosphate chain different from
that observed in the Legionella pneumophila NTPDase1 or in
rat NTPDase2 structures. This is likely the consequence of
differences in side-chain conformations between the binding sites
due to the absence of ligand in the rat NTPDase structure. Thus,
we chose to construct a model of human NTPDase1 based
on the structure of rat NTPDase2 which, as discussed above,
was solved in the presence of a Ca2+ cofactor and a relevant
ligand (AMPPNP). The resulting model was found to have good
Encouraged by these results, NTPDase2 inhibitors and related
molecules (analogues 2, 6, 7, 9, 18, 19, 20; Figures 1 and 6)
were docked into the binding site of the human NTPDase2
model. All analogues occupied the catalytic binding site forming
5925
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Figure 6. Homology model of human NTPDase2. A: Sequence alignment of human NTPDase2 and rat NTPDase2 (3CJA). Identical, similar, and
nonsimilar residues are depicted in dark blue, light blue, and white, respectively. B: 3D model of human NTPDase2 with AMPPNP in its catalytic
site. The protein is shown as a ribbon diagram color coded according to its secondary structure, the Ca2+ ion is shown in green, and the AMPPNP
molecule is depicted in blue. C: 3D model of the catalytic site of human NTPDase2 highlighting the interactions between AMPPNP and binding site
residues. The Ca2+ ion is shown as a green sphere, the six conserved water molecules are represented by red spheres (the water molecules that
coordinate the Ca2+ ion are W1−W4), and the AMPPNP molecule is colored according to atom types (nitrogen atoms in blue, oxygen atoms in red,
carbon atoms in gray, phosphate atoms in purple). Hydrogen bonds and π−π and π−cation interactions are shown in green and orange, respectively.
interactions with binding site residues similar to those formed
by AMPPNP, suggesting that, in principle, they could compete
with the endogenous ligand (ATP) for binding site interactions.
Analogues however differ in the relative orientations of the
ribose and base moieties. Analogues with substitution at the
C8 position (analogues 18, 19, 20) adopt a syn conformation
5926
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Figure 7. A comparison between NTPDase2 and NTPDase1 homology models. (A) Sequence alignment between human NTPDase1 and
NTPDase2. The overall sequence identity is 41.4%. The sequence identity between binding sites residues (boxed regions) is 84.2%. Different
residues in both the binding site and in close vicinity to it are bolded in red. (B) Superposition of the binding sites of human NTPDase2 (blue) and
NTPDase1 (green). The corresponding pairs Lys258 on NTPDase 1 vs Arg245 on NTPDase2, Phe365 on NTPDase1 vs Tyr350 on NTPDase2,
Lys368 on NTPDase1 vs Asp353 on NTPDase2, and Tyr408 on NTPDase1 vs Arg394 on NTPDase2 are circled.
whereas analogues lacking such substitutions (2, 6, 7, 9) adopt
an anti conformation.20 To obtain insight into the origin of the
inhibitory activity, an attempt was made to correlate the docking
scores for all analogues with their % inhibition values. This
attempt however was unsuccessful (data not shown). Similarly, a
detailed interaction analysis for all analogues (see Table S1,
Supporting Information) failed to reveal the source of their
differential activities.
which describes the averaged ligand−water vdW energy). In
terms of the electrostatic energy, only analogue 2 (the most
active compound) preferred the protein over the water
environment (the ⟨Vl‑sel⟩p term which describes the averaged
el
ligand−protein electrostatic energy was lower than the ⟨Vl‑s
⟩
w
which describes the averaged ligand−water electrostatic
energy for analogue 2 only and higher for all other analogues).
These data suggest that improving the inhibition potency of
NTPDase2 inhibitors may require further optimization of their
electrostatic interactions with the bind site. The correlations
between the LIE-based binding free energies and the
experimental % inhibition data for the triphosphate, triphos-
phate diphosphate, and for all analogues studied in this work are
presented in Figure 9, parts a, b, and c, respectively. In contrast
with the poor results obtained with the docking simulations
(see above), these good correlations (r2 = 0.94, r2 = 0.86, and
r2 = 0.68) lend credit to our hypothesis concerning the need to
use a proper treatment of molecular flexibility and solvation
effects for the accurate prediction of binding of ATP analogues
to NTPDase2.
LIE Simulations. Assuming that the efficacy of the highly
charged and flexible ATP analogues considered in this work
depends on factors not accounted for by docking simulations,
for example, desolvation energies and entropy penalties, we
turned our attention to MD simulations using the linear inter-
action energy (LIE) approach.51,52 Simulations were initiated
from the lowest energy poses and from the AMPPNP-like poses
of analogues 2, 6, 7, 9, 18, 19, and 20 (Figure 8 and Table S1,
Supporting Information), yet only the latter poses provided
good correlation with the experimental binding free energy data
and will be further discussed. Table 2 lists the individual energy
components obtained from the LIE simulations as well as the
computed binding free energies (ΔGLIE). In terms of the van der
Waals energies, all ligands preferred the protein over the water
Docking Simulations at NTPDase1. To gain insight into
the structural factors governing NTPDase2/NTPDase1 selec-
tivity, analogue 2, the most active and selective compound
discovered in this work, was docked into the catalytic site of the
environment (the ⟨Vl‑svdW⟩p term which describes the averaged
vdW
ligand−protein vdW energy was always lower than the ⟨Vl‑s
⟩
w
5927
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Figure 8. Starting points for LIE simulations. Panels A−G present the starting points for the LIE simulations for ligands 2, 6, 7, 9, 18, 19, and 20,
respectively. The Ca2+ ion is shown as a green sphere, the six conserved water molecules are represented by red spheres, and the ligands are colored
according to atom types (nitrogen atoms in blue; oxygen atoms in red; carbon atoms in gray; phosphate atoms in purple). Hydrogen bonds and π−π
and π−cation interactions are shown in green and orange, respectively. Panel H presents the superimposition of the starting points for all analogues
2 (green), 6 (yellow), 7 (purple), 9 (orange), 18 (red), 19 (cyan), 20 (magenta), and the crystal AMPPNP (blue).
human NTPDase1 model. Figure 10 provides a comparison
between representative binding modes of analogue 2 in the
catalytic sites of the two enzyme models. In NTPDase2, the
adenine ring is tightly stacked between Tyr350 and Arg394,
forming π−π and π−cation interactions, respectively
(Radenine‑Tyr350 = 4.19 Å; Radenine‑Arg394 = 4.38 Å). In NTPDase1,
Arg394 is replaced by Tyr408, leading to a loss of a π−cation
interaction and to a potential gain of a π−π interaction.
However, the aromatic ring of Tyr408 is too remote from the
adenine ring of the inhibitor, 2, to effectively interact with it
5928
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Table 2. Calculated Binding Energies (kcal/mol) and % Inhibition for Analogues 2, 6, 7, 9, 18, 19, and 20 Using the LIE
a
Approach
vdW
el
vdW
el
compound
percent of inhibition
⟨Vl‑s
⟩
⟨Vl‑s
⟩
⟨Vl‑s
⟩
⟨Vl‑s ⟩
p
ΔGLIE
w
w
p
2
70
42
0
10.0 0.6
11.3 0.9
0.9 0.5
−1640 10
−1653 15
−1686 13
−588.6 6.3
−1668 14
−1097 5.5
−1710 21
−29.6 0.6
−25.6 1.0
−31.9 4.1
−29.5 0.9
−31.1 1.6
−36.1 0.4
−30.6 1.9
−1647
−1621 11
−1583
−563
−1631
−1057
7
−10.4
9.3
6
7
3
8
6
4
45.4
8.7
9
10
30
18
18
−6.7 0.4
13.1 0.4
2.2 0.4
18
19
20
10.4
13.0
22.9
14.7 0.8
−1648 11
a
The Vl‑s terms represent MD time averages of analogue−solvent nonbonded interactions where solvent is taken as TIP3P water for the free state
and the protein immersed in TIP3P water for the bound state. The “el” and “vdW” terms refer to electrostatic and van der Waals interactions,
respectively. The coefficients α and β take the values of 0.18 and 0.5, respectively.
(Radenine‑Tyr408 = 5.52 Å). This leads to a shift in the position of
the adenine ring and to a concomitant change in the position
of the thiohexyl chain. Thus, while in NTPDase2 this chain is
deeply buried within a hydrophobic pocket within the binding
site, in NTPDase1, it is much more exposed to the solvent. We
therefore propose that the unfavorable exposure of a hydro-
phobic part of analogue 2 to the solvent in NTPDase1 is at
least partially responsible for its selectivity toward NTPDase2.
NTPDase2, NTPDase1, NTPDase3, and NTPDase8 by 70%,
<10%, ∼15%, and <10%, respectively (Figure 3A). It did not
affect the activity of ecto-5′-nucleotidase and partially inhibited
human NPP1 (Figure 3C). Kinetic analysis using the Dixon32
and Cornish−Bowden method32,60 revealed that compound 2
displayed a mixed-type, predominantly competitive, inhibition
toward recombinant human NTPDase2 (Figure 4 A,B).
Interestingly, ATP-β,γ-CH2 does not decrease the ATPase
activity of NTPDase234,61 and, in this light, substitution at the
C2 position by a thiohexyl group seems to be crucial for this
inhibition.
The kinetic parameters of analogue 2 were found to be
superior to those of the nonspecific inhibitors ARL 67156 and
(PSB-6426).18 The potency of compound 2 (Ki = 21 μM) is of
the same order as that of PSB-6426, (Ki = 8.2 μM), but the
selectivity of 2 toward NTPDase2 vs NTPDase1 was better
than that of PSB-6426. The latter inhibited NTPDase1 by 50%
at the same concentration it fully inhibited NTPDase2.
The importance of developing potent and selective NTPDase2
inhibitors, and not only general NTPDase inhibitors, is high-
lighted by the opposite effects of NTPDase1 and NTPDase2 on
platelet aggregation. NTPDase1 hydrolyzes both ATP and ADP,
whereas NTPDase2 is a preferential nucleoside triphosphatase-
hydrolyzing enzyme. Thus, by hydrolyzing ADP, which is an
aggregation factor, NTPDase1 inhibits platelet aggregation, while
in contrast, NTPDase2 promotes platelet aggregation in the
presence of ATP. Differential cellular expression of NTPDases
in the vasculature suggests a spatial regulation of nucleotide-
mediated signaling. In this context, NTPDase1 should abrogate
platelet aggregation and recruitment in intact vessels by the
conversion of ADP to adenosine monophosphate, while
NTPDase2 expression would promote platelet microthrombus
formation at sites of extravasation following vessel injury.
Thus, selectivity of the inhibitor toward these two enzymes is
critical.
DISCUSSION
■
NTPDase2 has been detected on blood vessels,55 cultured brain
astrocytes,56 neuronal progenitor,57 and in cancer cells.58,59
Thus, NTPDase2 can potentially be the target for novel drugs
for the treatment of cancer and cardiovascular or central
nervous system disorders. Potent and selective NTPDase2
inhibitors are therefore required. With this in mind, we have
studied C2- and N6-substituted ATP derivatives. Nucleotides
with modified N6′,N6″-position were synthesized to evaluate
the tolerance of NTPDases to the presence of N6-methyl,
-ethyl, or -benzyl groups (compounds 4−9). The SAR of
the adenine C8- and C2-positions was studied as well by
8-azaadenine nucleotide analogues 1 and 3, and 2-thiohexyl
nucleotide analogues, 1 and 2.
First, nucleotide derivatives 4−9 were studied as NTPDases
substrates. N6′,N6″-Methyl- and dimethyl-substituted ATP
derivatives 4 and 5 were hydrolyzed to ca. 80% by all
NTPDases, while compounds 6−9 were only modest substrates
(Table 1). The latter compounds were therefore further tested
as NTPDase inhibitors (Figure 2). Compounds 7−9 did not
inhibit NTPDases, while N6′,N6″-diethyl-ATP, 6, had a modest
nonspecific inhibitory effect on all tested NTPDases, similar
to the closely related ARL 67156.17A bulkier substitution
(dibenzyl) makes the compound 7 nonactive both as substrate
and inhibitor, showing that NTPDase2 is tolerant to Me or Et
groups but is not tolerant to steric hindrance by two benzyl
groups at the N6-position.
2-Hexthio-8-aza-ATP, 1, substituted at both C2 and 8
positions, was hydrolyzed by NTPDase1 and NTPDase2 in
about 40% and 10%, respectively. To isolate the effect of the
8-aza or 2-hexylthio group on recognition and subtype selectivity,
we synthesized and evaluated two additional compounds:
8-aza-ATP, 3, and 2-hexylthio-β,γ-CH2-ATP, 2. Compound 3
was hydrolyzed by both NTPDase1 and -2, indicating that the
2-hexylthio group of 1 provided resistance to hydrolysis by
NTPDase2. As expected, compound 2 was not hydrolyzed by
NTPDases1−3 and -8. Furthermore, it proved to be an effective
NTPDase2 inhibitor (Ki = 21 μM). Compound 2 was the
most selective inhibitor tested here, inhibiting (at 100 μM)
To provide structural insight into the activities of NTPDase2
inhibitors, a ligand-supported homology model of the protein
was generated based on the high resolution crystal structure of
rat NTPDase2.34 Docking simulations using Glide successfully
reproduced the crystallographic pose of AMPPNP within the
rat NTPDase2 binding site (heavy atoms RMSD of 0.61 Å).
Encouraged by these results, compounds 2, 6, 7, 9, 18, 19, and
20 were docked into the binding site of the model, yet attempts
to correlate the docking scores or the interaction patterns with
% inhibition data have proven unsuccessful. Assuming that this
lack of correlation results from factors not accounted for by
docking simulations such as desolvation energies and entropy
penalties, we have turned our efforts to LIE simulations.44,45
5929
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Figure 9. Correlation between calculated free energy of binding and % inhibition. Correlation between binding free energies calculated using the LIE
approach and experimentally determined % inhibition for (A) the triphosphate analogues 2, 6, 7, 18, and 20, (B) the triposphate and diphospate
analogues (2, 6, 7, 18, 19, 20), and (C) all analogues. % Inhibition for the inactive analogue 7 was set to zero.
Indeed, when initiated from AMPPNP-like (but not from
lowest energy; results not shown) poses, these simulations pro-
vided a very good correlation between experiment (% inhibition)
and calculations. This correlation is particularly good when
considering only the triphosphate analogues (analogues 2, 6, 7,
18, 20; r2 = 0.94; Figure 9a) and slightly decreases when
5930
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Figure 10. A comparison of representative binding modes of analogue 2 in the catalytic site of the human NTPDase2 and NTPDase1 homology
models. (A) Surface of the human NTPDase1 model colored according to its hydrophobicity (hydrophilic and hydrophobic regions colored in blue
and brown, respectively). (B) Surface of the human NTPDase2 model colored according to its hydrophobicity (color coding as in A). (C)
Superposition of representative poses of analogue 2 in the catalytic site of NTPDase2 (green) and NTPDase1 (orange). Also shown are key residues
which interact with the adenine ring (NTPDase2: Tyr350 and Arg394 colored in green; NTPDase1: Phe365 and Arg394 colored in orange). The
different position of the adenine rings in the two sites is clearly visible.
including the diphosphate analogue (19; r2 = 0.86; Figure 9b)
Table 2 demonstrates the expected trend between the formal
charges of the analogues and their electrostatic energies, both
in solution and in the protein environment. Thus, the average
and the monophsphate analogue (9, r2 = 0.68; Figure 9c).
Indeed, LIE simulations work best for congeneric series.62
5931
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
electrostatic energy for the triphosphate analogues (2, 6, 7, 18,
20; formal charge (−4)) is −1671 (solution) and −1626
(protein) kcal/mol, that of the diphosphate analogue (19,
formal charge (−3)) is −1097 (solution) and −1057 (protein)
kcal/mol and that of the monophosphate analogue (9, formal
charge (−2)) is −588 (solution) and −563 (protein) kcal/mol.
From all analogues, only analogue 2 has a negative ΔVel
(= ⟨Vl‑s ⟩p − ⟨Vl‑s ⟩w) term, while for the rest of the ligands
this term is positive. Thus, this is the only ligand which prefers
the protein over the solvent environment in terms of electro-
static interactions. Furthermore, the least potent analogue, 7,
ASSOCIATED CONTENT
* Supporting Information
Additional information as noted in text. This material is
available free of charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Authors
*Tel: 972-3-5318303. Fax: 972-3-6353907. E-mail: bilha.
fischer@mail.biu.ac.il.
*Tel: 418-525-4444. Fax: 418-654-2765. E-mail: Jean.Sevigny@
crchul.ulaval.ca.
*Tel: 972-3-7384588. Fax: 972-3-7384588. E-mail:
hsenderowitz@gmail.com.
■
el
el
has the largest electrostatic preference for solution (ΔVel
=
103 kcal/mol). Controlling the protein/solvent electrostatic
interactions balance may therefore be key to the design of more
efficacious NTPDase2 inhibitors. For example, replacement of
any polar moiety within a ligand which favors the solvent over
the protein by a nonpolar group may improve binding by
reducing the desolvation penalty. While an atom by atom
analysis of the MD trajectory for analogue 2 revealed that in
this case all polar moieties prefer the protein environment, a
similar analysis of other ligands may point to modifications
which may increase the binding affinity. We therefore propose
that the LIE method could be used as a design tool for both
predicting ligand−protein electrostatic energies as well as the
overall NTPDase2 binding energies of potential NTPDase2
inhibitors.
Author Contributions
⊥I.G., J.L., and L.S. contributed equally to this work
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by grants from the Heart & Stroke
Foundation (HSF) of Quebec and from the Canadian Institutes
of Health Research (CIHR; MOP-102472, IRO-102806) to JS.
M.F. was the recipient of a Doctoral Scholarship from the
Government of Gabon, M.M.-S. of a fellowship from the
Spanish Ministry of Education and Science (MEC, Jose
Castillejo Program), and J.S. of a senior Scholarship from the
Given the importance of NTPDase1/2 inhibitor’s selectivity,
a homology model of NTPDase1 was developed and compared
to that of NTPDase2. Both models have highly conserved
binding sites, yet they differ by several residues (Figure 7). In
particular, Arg245, Tyr350, Asp353, and Arg394 in NTPDase2
are respectively replaced by Lys258, Phe365, Lys368, and
Tyr408 in NTPDase1. As shown in Figure 10, these changes
result in different interactions of the adenine ring with its
“flanking” residues (Tyr350 and Arg394 in NTPDase2 and
Phe365 and Tyr408 in NTPDase1) which ultimately lead to
an unfavorable shift of the hydrophobic thiohexyl chain from
a buried conformation in NTPDase2 to a solvent-exposed
conformation in NTPDase1. This may at least partially explain
the selectivity of analogue 2 toward NTPDase2.
Finally, the differences between the binding sites of
NTPDase1 and 2 could be utilized to confer NTPDase1 or
NTPDase2 selectivity by further modifications to the inhibitors
described in this study. For example, substituting the C2
position of the adenine ring by a positive- or negative-charge
terminated thioalkyl chain, which would interact with Asp353
or Lys368, may lead to selectivity toward NTPDase2 or
NTPDase1, respectively (Figure 7). Such design efforts are
currently being conducted in our laboratory.
In summary, 2-hexylthio-β,γ-methylene-ATP, 2, may be used
to selectively block NTPDase2 ectonucleotidase activity.
Compound 2 may be a good starting point for the development
of drugs for the treatment of certain cardiovascular disorders,
especially when controlling the ADP level is essential as in
atherosclerosis.40,42 Balancing ADP levels may indeed be
important for the therapy of diseases where platelet dysfunction
is a key component and also for cancer therapy and cardio-
protection. Such development efforts could benefit from
the homology models of NTPDase1 and 2 and from the
computational models for binding free energy prediction
developed in this work.
Fonds de recherche du Queb
́
ec − Sante
́
(FRQS).
ABBREVIATIONS USED
■
E-NPP, ecto-nucleotide pyrophosphatase/phosphodiesterase;
E-NTPDase, ecto-nucleoside triphosphate diphosphohydrolase;
Pi, inorganic phosphate; pnp-TMP, p-nitrophenylthymidine 5′-
monophosphate; POM, polyoxometalates; TEAA, triethylam-
monium acetate; TBAP, tetrabutylammonium phosphate
REFERENCES
■
(1) Yegutkin, G. G. Nucleotide- and nucleoside-converting
ectoenzymes: Important modulators of purinergic signalling cascade.
Biochim. Biophys. Acta 2008, 1783, 673−694.
(2) Zimmermann, H.; Zebisch, M.; Strater, N. Cellular function and
molecular structure of ecto-nucleotidases. Purinergic Signalling 2012, 8,
437−502.
́
(3) Kukulski, F.; Lev
́
esque, S. A.; Lavoie, E. G.; Lecka, J.; Bigonnesse,
igny, J.
F.; Knowles, A. F.; Robson, S. C.; Kirley, T. L.; Sev
́
Comparative hydrolysis of P2 receptor agonists by NTPDases 1, 2,
3 and 8. Purinergic Signalling 2005, 1, 193−204.
(4) Zimmermann, H. 5′-Nucleotidase: Molecular structure and
functional aspects. Biochem. J. 1992, 285, 345−365.
(5) Iqbal, J.; Jirovsky, D.; Lee, S. Y.; Zimmermann, H.; Muller, C. E.
̈
Capillary electrophoresis-based nanoscale assays for monitoring ecto-
5′-nucleotidase activity and inhibition in preparations of recombinant
enzyme and melanoma cell membranes. Anal. Biochem. 2008, 373,
129−140.
(6) Strater, N. Ecto-5′-nucleotidase: Structure function relationships.
Purinergic Signalling 2006, 2, 343−350.
(7) Krug, U.; Patzschke, R.; Zebisch, M.; Balbach, J.; Strater, N.
̈
Contribution of the two domains of E. coli 5′-nucleotidase to substrate
specificity and catalysis. FEBS Lett. 2013, 587, 460−466.
(8) Shryock, J. C.; Belardinelli, L. Adenosine and adenosine receptors
in the cardiovascular system: Biochemistry, physiology, and
pharmacology. Am. J. Cardiol. 1997, 79, 2−10.
̈
(9) Koszalka, P.; Ozuyaman, B.; Huo, Y.; Zernecke, A.; Flogel, U.;
̈
̈
́
Braun, N.; Buchheiser, A.; Decking, U. K. M.; Smith, M. L.; Sevigny, J.;
Gear, A.; Weber, A.-A.; Molojavyi, A.; Ding, Z.; Weber, C.; Ley, K.;
5932
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
Zimmermann, H.; Godecke, A.; Schrader, J. Targeted disruption of
(28) Belli, S. I.; Goding, J. W. Biochemical characterization of human
PC-1, an enzyme possessing alkaline phosphodiesterase I and
nucleotide pyrophosphatase activities. Eur. J. Biochem. 1994, 226,
433−443.
(29) Jin-Hua, P.; Goding, J. W.; Nakamura, H.; Sano, K. Molecular
cloning and chromosomal localization of PD-Iβ(PDNP3), a new
member of the human phosphodiesterase I genes. Genomics 1997, 45,
412−415.
̈
cd73/ecto-5′-nucleotidase alters thromboregulation and augments
vascular inflammatory response. Circ. Res. 2004, 95, 814−821.
(10) Kawashima, Y.; Nagasawa, T.; Ninomiya, H. Contribution of
ecto-5′-nucleotidase to the inhibition of platelet aggregation by human
endothelial cells. Blood 2000, 96, 2157−2162.
(11) Wink, M. R.; Braganhol, E.; Tamajusuku, A. S. K.; Lenz, G.;
́
Zerbini, L. F.; Libermann, T. A.; Sevigny, J.; Battastini, A. M. O.;
Robson, S. C. Nucleoside triphosphate diphosphohydrolase-2
(NTPDase2/CD39L1) is the dominant ectonucleotidase expressed
by rat astrocytes. Neuroscience 2006, 138, 421−432.
(30) Bradford, M. M. A rapid and sensitive method for the
quantitation of microgram quantities of protein utilizing the principle
of protein-dye binding. Anal. Biochem. 1976, 72, 248−254.
(31) Baykov, A. A.; Evtushenko, O. A.; Avaeva, S. M. A malachite
green procedure for orthophosphate determination and its use in
alkaline phosphatase-based enzyme immunoassay. Anal. Biochem.
1988, 171, 266−270.
̌
(12) Sali, A.; Blundell, T. L. Comparative protein modelling by
satisfaction of spatial restraints. J. Mol. Biol. 1993, 234, 779−815.
(13) Braganhol, E.; Morrone, F. B.; Bernardi, A.; Huppes, D.;
Meurer, L.; Edelweiss, M. I. A.; Lenz, G.; Wink, M. R.; Robson, S. C.;
Battastini, A. M. O. Selective NTPDase2 expression modulates in vivo
rat glioma growth. Cancer Sci. 2009, 100, 1434−1442.
(32) Dixon, M. Graphical determination of equilibrium constants.
Biochem. J. 1965, 94, 760−762.
(14) Lecka, J.; Gillerman, I.; Fausther, M.; Salem, M.; Munkonda, M.
(33) Vollmayer, P.; Clair, T.; Goding, J. W.; Sano, K.; Servos, J.;
Zimmermann, H. Hydrolysis of diadenosine polyphosphates by
nucleotide pyrophosphatases/phosphodiesterases. Eur. J. Biochem.
2003, 270, 2971−2978.
N.; Brosseau, J.-P.; Cadot, C.; Martín-Satue,
́
M.; d’Orlea
́
ns-Juste, P.;
́
Rousseau, E.; Poirier, D.; Kunzli, B.; Fischer, B.; Sevigny, J. 8-BuS-ATP
́
̈
derivatives as specific NTPDase1 inhibitors. Br. J. Pharmacol. 2013,
169, 179−196.
(34) Zebisch, M.; Strater, N. Structural insight into signal conversion
̈
(15) Muller, C. E.; Iqbal, J.; Baqi, Y.; Zimmermann, H.; Rollich, A.;
̈
̈
and inactivation by NTPDase2 in purinergic signaling. Proc. Natl. Acad.
Sci. U.S.A. 2008, 105, 6882−6887.
Stephan, H. Polyoxometalates-a new class of potent ecto-nucleoside
triphosphate diphosphohydrolase (NTPDase) inhibitors. Bioorg. Med.
Chem. Lett. 2006, 16, 5943−5947.
(35) Gasteiger, E.; Gattiker, A.; Hoogland, C.; Ivanyi, I.; Appel, R. D.;
Bairoch, A. ExPASy: the proteomics server for in-depth protein
knowledge and analysis. Nucleic Acids Res. 2003, 31, 3784−3788.
(36) Larkin, M. A.; Blackshields, G.; Brown, N. P.; Chenna, R.;
McGettigan, P. A.; McWilliam, H.; Valentin, F.; Wallace, I. M.; Wilm,
A.; Lopez, R.; Thompson, J. D.; Gibson, T. J.; Higgins, D. G. Clustal
W and Clustal X version 2.0. Bioinformatics 2007, 23, 2947−2948.
(37) Accelrys. Discovery studio modeling environment, 2.5; 2009.
(38) Sippl, M. J. Recognition of errors in three-dimensional
structures of proteins. Proteins 1993, 17, 355−362.
(16) Crack, B. E.; Pollard, C. E.; Beukers, M. W.; Roberts, S. M.;
Hunt, S. F.; Ingall, A. H.; McKechnie, K. C.; IJzerman, A. P.; Leff, P.
Pharmacological and biochemical analysis of FPL 67156, a novel,
selective inhibitor of ecto-ATPase. Br. J. Pharmacol. 1995, 114, 475−
481.
́ ́
(17) Levesque, S. A.; Lavoie, E. G.; Lecka, J.; Bigonnesse, F.; Sevigny,
J. Specificity of the ecto-ATPase inhibitor ARL 67156 on human and
mouse ectonucleotidases. Br. J. Pharmacol. 2007, 152, 141−150.
(18) Brunschweiger, A.; Iqbal, J.; Umbach, F.; Scheiff, A. B.;
Munkonda, M. N.; Sev
́
igny, J.; Knowles, A. F.; Muller, C. E. Selective
(39) Wiederstein, M.; Sippl, M. J. ProSA-web: Interactive web service
for the recognition of errors in three-dimensional structures of
proteins. Nucleic Acids Res. 2007, 35, W407−W410.
̈
nucleoside triphosphate diphosphohydrolase-2 (NTPDase2) inhib-
itors: Nucleotide mimetics derived from uridine-5′-carboxamide. J.
Med. Chem. 2008, 51, 4518−4528.
(40) Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J. J.; Mainz,
D. T.; Repasky, M. P.; Knoll, E. H.; Shelley, M.; Perry, J. K.; Shaw, D.
E.; Francis, P.; Shenkin, P. S. Glide: A new approach for rapid, accurate
docking and scoring. 1. Method and assessment of docking accuracy. J.
Med. Chem. 2004, 47, 1739−1749.
(19) Seela, F.; Munster, I.; Luchner, U.; Rosemeyer, H. 8-
̈
̈
Azaadenosine and its 2′-deoxyribonucleoside: Synthesis and oligonu-
cleotide base-pair stability. Helv. Chim. Acta 1998, 81, 1139−1155.
(20) Halbfinger, E.; Major, D. T.; Ritzmann, M.; Ubl, J.; Reiser, G.;
Boyer, J. L.; Harden, K. T.; Fischer, B. Molecular recognition of
modified adenine nucleotides by the P2Y1-receptor. 1. A synthetic,
biochemical, and NMR approach. J. Med. Chem. 1999, 42, 5325−5337.
(21) Endo, T.; Zemlicka, J. Oxidation of N6,N6-dialkyl-2′,3′,5′-tri-O-
acyladenosines with ruthenium tetroxide and a novel selective N-
monodealkylation sequence. J. Org. Chem. 1979, 44, 3652−3656.
(22) Gillerman, I.; Fischer, B. An improved one-pot synthesis of
nucleoside 5′-triphosphate analogues. Nucleosides, Nucleotides Nucleic
Acids 2010, 29, 245−256.
(41) Glide, 5.5; Schrodinger, Inc., New York, 2009.
̈
(42) Friesner, R. A.; Murphy, R. B.; Repasky, M. P.; Frye, L. L.;
Greenwood, J. R.; Halgren, T. A.; Sanschagrin, P. C.; Mainz, D. T.
Extra precision glide: docking and scoring incorporating a model of
hydrophobic enclosure for protein−ligand complexes. J. Med. Chem.
2006, 49, 6177−6196.
(43) Donohue, J.; T, K. N. Base pairing in deoxyribonucleic acid. J.
Mol. Biol. 1960, 2, 363−371.
(44) Åqvist, J.; Medina, C.; Samuelsson, J.-E. A new method for
predicting binding affinity in computer-aided drug design. Protein Eng.
1994, 7, 385−391.
(45) Hansson, T.; Marelius, J.; Aqvist, J. Ligand binding affinity
prediction by linear interaction energy methods. J. Comput.-Aided Mol.
Des. 1998, 12, 27−35.
(46) Marelius, J.; Kolmodin, K.; Feierberg, I.; Aqvist, J. Q: a
molecular dynamics program for free energy calculations and empirical
valence bond simulations in biomolecular systems. J. Mol. Graph
Model. 1998, 16, 213−225.
(47) Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. Development
and resting of the OPLS all-atom force field on conformational
energetics and properties of organic liquids. J. Am. Chem. Soc. 1996,
118, 11225−11236.
(48) King, G.; Warshel, A. A surface constrained all-atom solvent
model for effective simulations of polar solutions. J. Chem. Phys. 1989,
91, 3647−3661.
(23) Freist, W.; Cramer, F. Synthesis of AMP and ATP analogs; Wiley:
New York, 1978; Vol. 2, pp 827−836.
́
(24) Kaczmarek, E.; Koziak, K.; Sevigny, J.; Siegel, J. B.; Anrather, J.;
Beaudoin, A. R.; Bach, F. H.; Robson, S. C. Identification and
characterization of CD39/vascular ATP diphosphohydrolase. J. Biol.
Chem. 1996, 271, 33116−33122.
(25) Smith, T. M.; Kirley, T. L. Cloning, sequencing, and expression
of a human brain ecto-apyrase related to both the ecto-ATPases and
CD39 ecto-apyrases. Biochim. Biophys. Acta, Protein Struct. Mol.
Enzymol. 1998, 1386, 65−78.
(26) Bigonnesse, F.; Lev
S. C.; Fernandes, M. J. G.; Sev
́
esque, S. A.; Kukulski, F.; Lecka, J.; Robson,
igny, J. Cloning and characterization of
́
mouse nucleoside triphosphate diphosphohydrolase-8. Biochemistry
2004, 43, 5511−5519.
́
(27) Lecka, J.; Rana, M. S.; Sevigny, J. Inhibition of vascular
ectonucleotidase activities by the pro-drugs ticlopidine and clopidogrel
favours platelet aggregation. Br. J. Pharmacol. 2010, 161, 1150−1160.
5933
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934

Journal of Medicinal Chemistry
Article
(49) Gillerman, I.; Fischer, B. Investigations into the origin of the
molecular recognition of several adenosine deaminase inhibitors. J.
Med. Chem. 2011, 54, 107−121.
(50) Gillerman, I.; Fischer, B. An improved one-pot synthesis of
nucleoside 5′-triphosphate analogues. Nucleosides, Nucleotides Nucleic
Acids 2010, 29, 245−256.
(51) Ludwig, J. A simple one flask synthesis of nucleoside 5′-
triphosphates from unprotected nucleosides via nucleoside 5′-
cyclotriphosphates. In Biophosphates and Their AnaloguesSynthesis,
Structure, Metabolism and Activity; Proceedings of the Second
International Symposium on Phosphorus Chemistry Directed
Towards Biology, Lodz, Poland, Sep 8−12, 1986; Elsevier Science
Publishers B. V.: Amsterdam, Netherlands,1987; p 201.
́
(52) Kukulski, F.; Sevigny, J.; Komoszynski, M. Comparative
hydrolysis of extracellular adenine nucleotides and adenosine in
synaptic membranes from porcine brain cortex, hippocampus,
cerebellum and medulla oblongata. Brain Res. 2004, 1030, 49−56.
(53) Zebisch, M.; Krauss, M.; Schaefer, P.; Straeter, N. Crystallo-
graphic evidence for a domain motion in rat nucleoside triphosphate
diphosphohydrolase (NTPDase) 1. J. Mol. Biol. 2012, 415, 288−306.
(54) Zebisch, M.; Krauss, M.; Schaefer, P.; Lauble, P.; Straeter, N.
Crystallographic snapshots along the reaction pathway of nucleoside
triphosphate diphosphohydrolases. Structure (Oxford, U. K.) 2013, 21,
1460−1475.
́
(55) Sevigny, J.; Sundberg, C.; Braun, N.; Guckelberger, O.;
Csizmadia, E.; Qawi, I.; Imai, M.; Zimmermann, H.; Robson, S. C.
Differential catalytic properties and vascular topography of murine
nucleoside triphosphate diphosphohydrolase 1 (NTPDase1) and
NTPDase2 have implications for thromboregulation. Blood 2002, 99,
2801−2809.
(56) Wink, M. R.; Braganhol, E.; Tamajusuku, A. S. K.; Lenz, G.;
́
Zerbini, L. F.; Libermann, T. A.; Sevigny, J.; Battastini, A. M. O.;
Robson, S. C. Nucleoside triphosphate diphosphohydrolase-2
(NTPDase2/ CD39L1) is the dominant ectonucleotidase expressed
by rat astrocytes. Neuroscience 2006, 138, 421−432.
(57) Langer, D.; Ikehara, Y.; Takebayashi, H.; Hawkes, R.;
Zimmermann, H. The ectonucleotidases alkaline phosphatase and
nucleoside triphosphate diphosphohydrolase 2 are associated with
subsets of progenitor cell populations in the mouse embryonic,
postnatal and adult neurogenic zones. Neuroscience 2007, 150, 863−
879.
(58) Knowles, A. F.; Chiang, W.-C. Enzymatic and transcriptional
regulation of human ecto-ATPase/E-NTPDase 2. Arch. Biochem.
Biophys. 2003, 418, 217−227.
(59) Buffon, A.; Ribeiro, V. B.; Wink, M. R.; Casali, E. A.; Sarkis, J. J.
Nucleotide metabolizing ecto-enzymes in Walker 256 tumor cells:
molecular identification, kinetic characterization and biochemical
properties. Life Sci. 2007, 80, 950−958.
(60) Cornish-Bowden, A. A simple graphical method for determining
the inhibition constants of mixed, uncompetitive and non-competitive
inhibitors. Biochem. J. 1974, 137, 143−144.
(61) Lecka, J.; Bloch-Boguslawska, E.; Molski, S.; Komoszynski, M.
Extracellular purine metabolism in blood vessels (Part II): Activity of
ecto-enzymes in blood vessels of patients with abdominal aortic
aneurysm. Clin. Appl. Thromb./Hemostasis 2010, 16, 650−657.
(62) Boukharta, L.; Keranen, H.; Stary-Weinzinger, A.; Wallin, G.; de
Groot, B. L.; Aaqvist, J. Computer simulations of structure−activity
relationships for hERG channel blockers. Biochemistry 2011, 50,
6146−6156.
5934
dx.doi.org/10.1021/jm401933c | J. Med. Chem. 2014, 57, 5919−5934