Highly Selective and Potent Ectonucleotide Pyrophosphatase-1

Apr 22, 2018 - Centre de Recherche du CHU de Québec, Université Laval, Québec , QC , Canada. § Department of ... Phone: 972-3-5318303. E-mail: Bil...
1 downloads 3 Views 2MB Size
Article pubs.acs.org/jmc

Cite This: J. Med. Chem. 2018, 61, 3939−3951

Highly Selective and Potent Ectonucleotide Pyrophosphatase‑1 (NPP1) Inhibitors Based on Uridine 5′‑Pα,α-Dithiophosphate Analogues Vadim Zelikman,† Julie Pelletier,‡ Luba Simhaev,† Aviad Sela,† Fernand-Pierre Gendron,§ Guillaume Arguin,§ Hanoch Senderowitz,† Jean Sévigny,‡,∥ and Bilha Fischer*,† †

Department of Chemistry, Bar-Ilan University, Ramat-Gan 5290002, Israel Centre de Recherche du CHU de Québec, Université Laval, Québec, QC, Canada § Department of Anatomy and Cellular Biology, Université de Sherbrooke, 3201 Rue Jean-Mignault, Sherbrooke, QC J1E 4K8, Canada ∥ Département de Microbiologie-Infectiologie et d’Immunologie, Faculté de Médecine, Université Laval, Pavillon CHUL, 2705 Boulevard Laurier, Local T1-49, Québec, QC G1V 4G2, Canada ‡

S Supporting Information *

ABSTRACT: Ectonucleotide pyrophosphatase/phosphodiesterase-1 (NPP1) hydrolyzes phosphodiester bonds of nucleotides such as ATP, resulting mainly in the formation of AMP and pyrophosphate. NPP1 activity plays a deleterious function in calcified aortic valve disease and calcium pyrophosphate deposition disease. Thus, inhibitors of NPP1 represent a medical need. We developed novel NPP1 inhibitors based on uridine 5′-Pα,α-dithiophosphate analogues, 9−12. All these analogues potently inhibited hNPP1 (80−100% inhibition) at 100 μM, with no, or minimal, inhibition of NPP3 and other ectonucleotidases (NTPDase1,2,3,8). These compounds showed nearly no activity at uracil-nucleotide sensitive P2Y2,4,6-receptors and thus represent highly selective NPP1 inhibitors. The most promising inhibitor was diuridine 5′-Pα,α,5″-Pα,α-tetrathiotetraphosphate, 12, exhibiting Ki of 27 nM. Analogues 9−12 proved to be highly stable to air oxidation and to acidic and basic pH. Docking simulations suggested that the enhanced NPP1 inhibitory activity and selectivity of analogue 12 could be attributed to the simultaneous occupancy of two sites (the AMP site and an alternative site) of NPP1 by this compound.



INTRODUCTION Ectonucleotide pyrophosphatase/phosphodiesterase1 (NPP1) is a member of the ectonucleotide pyrophosphatase/phosphodiesterase (eNPP) family. It is a membrane glycoprotein with Zn(II)-binding extracellular catalytic-site. NPP1 hydrolyzes bonds of nucleotides such as ATP.1 Hydrolysis of ATP by NPP1 results mainly in the formation of AMP and pyrophosphate (PPi) and also ADP and inorganic phosphate (Pi).2 It is noteworthy that PPi can be further converted to Pi by alkaline phosphatase. AMP produced by NPP1 can be hydrolyzed by 5′-nucleotidase (5′-NT, CD73) or alkaline phosphatase to adenosine.3 An abnormally high level of NPP1 is found in calcific aortic valve disease (CAVD),1 which is the most frequent heart valve disorder.4 Overexpression of NPP1 leads to increased mineralization of the aortic valve and to valvular interstitial cell apoptosis.5,6 NPP1 is also a major contributor to the pathological NPPase activity and elevated PPi levels are © 2018 American Chemical Society

observed in a joint pathology, calcium pyrophosphate dihydrate (CPPD) disease.7−9 In addition to NPP1, the phylogenetically related enzyme, NPP3, also catalyzes the hydrolysis of phosphodiester bonds.10 Yet, while excess activity of NPP1 results in health disorders such as CAVD or CPPD, excess activity of NPP3 triggers an allergic response.11,12 Currently, there are no known therapies to slow down CAVD progression, making surgical valve replacement as the only effective treatment for aortic stenosis.13 Likewise, today there are no CPPD disease modifying drugs, and the only treatments are symptomatic. Hence, a promising avenue for a novel pharmacological treatment for either CAVD or CPPD could be selective NPP1 inhibitors. Received: December 24, 2017 Published: April 22, 2018 3939

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

Figure 1. Previously reported NPP1 inhibitors.

Figure 2. Uridine 5′-Pα,α-dithiophosphate analogues synthesized and evaluated in this study.

analogue 1-B isomer, analogue 2-A isomer, and analogue 2-B isomer, respectively. Importantly, these nucleotides were selective for NPP1 vs NPP3 as at the concentration of 100 μM they inhibited human NPP1 activity by 83−93% and human NPP3 activity by only 15−18%.20 This selectivity is significant because overexpression of NPP1 is responsible for the osteoarthritis ailment, as well as CAVD, while NPP3 activation is assumed to be a trigger for an allergic response.11,12 In addition, we identified di-2′-deoxyadenosine α,β-δ,εdimethylenepentaphosphonate, 3, as a selective inhibitor of NPP1 vs NPP3, showing an inhibition of 99% vs 18%, respectively, at the concentration of 100 μM. The IC50 of 3 was 13 μM and the Ki was 9 μM for NPP1. Moreover, 3 was devoid

Previously, we developed biocompatible, water-soluble, and selective NPP1 inhibitors that do not affect other ectonucleotidases such as ectonucleoside triphosphate diphosphohydrolase (NTPDases) and ecto-5′-ectonucleotidase (5′-NT) and that do not interfere with nucleotide receptors, P2-Rs, activation14−16 (e.g., 1−3, Figure 1). Activation of P2-Rs, divided to P2Y-R and P2X-R subfamilies, which are expressed in many tissues, may trigger various health disorders. For example, activation of P2Y2-R by extracellular nucleotides results in the inhibition of bone formation.17 On the other hand, activation of P2Y2-R inhibits aortic valve calcification.18,19 Specifically, analogues 1 and 2 were found as NPP1 inhibitors with Ki values of 56 μM, 0.5 μM, and 7 μM for 3940

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

Scheme 1. Synthesis of Uridine 5′-O-(2-thio)-1,3,2-dithiaphospholane, 13a

a

Reagents and conditions: (a) DIPEA, CH3CN, −38 °C, 3 h; (b) S8, overnight, rt.

Scheme 2a

a

Reagents and conditions: DBU, CH3CN, rt, 0.5−15 h.

sulfamide, 8, was found as a potent inhibitor of NPP1 with Ki = 0.036 μM, but this compound was also a potent ligand for the hERG potassium channel, which plays a key role in regulating cardiac rhythm.23 The above-mentioned Pα-boranophosphate- or Pα-thiophosphate-nucleotide NPP1 inhibitors, 1, 2, and 5, are characterized by the presence of an asymmetric center at Pα. Since usually only one of these diastereomeric pairs is biologically active, half of the synthesized material is wasted. Therefore, here we designed uridine 5′-Pα,α-dithiophosphate analogues 9−12 (Figure 2) to maintain the desired pharmacological properties of nucleoside 5′-Pα-phosphorothioates (e.g., 5) while avoiding the presence of a chiral center and the loss of half of the product. Specifically, we report the synthesis of uridine Pα,α-dithiophosphate analogues, 9−12, their inhibitory activity at NPP1, and their selectivity vs related ectonucleotidases, NPP3, NTPDase1,-2,-3,-8, and uridine nucleotide sensitive purinergic receptors (P2Y2,4,6-Rs). In addition, we analyzed and rationalized the inhibitory activity

of inhibitory activity at the related ectonucleotidases NTPDase1, -2, -3, and -8.21 Later, we identified ATP-α,β-CH2-γ-thio, 4, as a potent NPP1 inhibitor with Ki = 0.02 μM and IC50 = 0.39 μM.15 However, analogue 4 was not NPP1 selective and inhibited tissue nonspecific alkaline phosphatase (TNAP), NTPDase1, and NTPDase3 activity by 17%, 60%, and 40%, respectively. However, ATP-α-thio-β,γ-CCl2, 5, A-isomer, was found to be a selective NPP1 inhibitor (Ki = 0.68 μM and IC50 = 0.57 μM) and had only a minor inhibitory effect on NTPDase1, -2, -3, -8 activity (0−25% inhibition) and on TNAP activity (8% inhibition). Both 4 and 5 were found to be poor or nonagonists of P2Y1/P2Y2 receptors. Furthermore, we showed that 5A was able to lower extracellular pyrophosphate levels in human cartilage.15 Non-nucleotide NPP1 inhibitors, e.g., 6 and 7, were reported to be noncompetitive NPP1 inhibitors with IC50 of 66 and 368 μM, respectively.22 However, their selectivity was not determined. In addition, quinazolin-4-piperidin-4-ethyl3941

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

Figure 3. Inhibition of NTPDases and NPPs by analogues 9−12 using as substrates ATP (100 μM) for NTPDase1, -2, -3, -8 and pNP-TMP (100 μM) for NPP1,3 (A). ATP (100 μM) was used as the substrate for NPP1 (B). Results were obtained from at least two independent experiments carried out in triplicate. In panel A, the activity with 100 μM ATP was set as 100%, which was 287 ± 31, 395 ± 67, 278 ± 22, 119 ± 26 nmol of Pi· min−1·(mg of protein)−1 for human NTPDase1, -2, -3, and -8, respectively, and the activity with 100 μM pNP-TMP was set as 100% which was 67 ± 5, 21 ± 1 nmol of p-nitrophenol·min−1·(mg of protein)−1 for human NPP1 and NPP3, respectively. In panel B, the activity with 100 μM ATP was set as 100% which was 34 ± 7 nmol of PPi·min−1·(mg of protein)−1 for NPP1.

catalytic site. In addition, the O → S replacement may allow for a better coordination of the Zn(II) ions in NPP1’s catalytic site, thus competing with ATP and preventing its hydrolysis. Furthermore, we focused here on a uridine- rather than adenine-nucleotide scaffold since NPP1 hydrolyzes also UTP. UTP analogues are recognized by fewer proteins than ATP analogues and thus are expected to increase selectivity. Synthesis of Uridine 5′-Pα,α-Dithiophosphate Analogues 9−12. The synthesis of analogues 9−12 involved uridine 2′,3′-O-methoxymethylidene-5′-O-(2-thio)-1,3,2-dithiaphospholane, 13, as a common synthetic intermediate (Scheme 1). Uridine 2′,3′-O-methoxymethylidene, 14, was treated with 1,3,2-dithiaphospholane,30 15, at −38 °C to obtain intermediate 16. The latter was oxidized by powdered elemental sulfur at rt to give product 13 at 55% yield. Uridine Pα,α-dithiophosphate analogues 9−12 were synthesized from thiodithiaphospholane intermediate, 13, via DBUmediated nucleophilic dithiaphospholane ring-opening.30 Specifically, we applied orthophosphate/pyrophosphate/ methylenediphosphonate as nucleophiles (Scheme 2). The reactions were performed at rt with complete exclusion of moisture. The time of the reactions depended on the nucleophilicity of the phosphate reactant and was in the range of 0.5−15 h (the longest reaction time was for orthophosphate, and the shortest reaction time was for methylenediphosphonate). The nucleophilic ring-opening reaction was followed by deprotection of the methoxymethylidene group. The crude material was separated on an anion exchanger and eluted with a linear gradient of

of 9−12 by molecular modeling. Finally, we report on the chemical stability of 9−12 under various conditions (air oxidation, acidic pH, and basic pH).



RESULTS AND DISCUSSION Design of Uridine 5′-Pα,α-Dithiophosphate Analogues 9−12. The replacement of a nucleotide phosphate group by a thiophosphate group24−26 significantly affects the pharmacological properties of a nucleotide due to the different physicochemical properties of these groups, e.g., different steric requirements of P−S vs P−O bond and different affinity of the “soft’” sulfur atom vs the “hard” oxygen atom toward zinc ions.27 Specifically, due to the larger volume and polarizability of the sulfur vs oxygen atom, a thiophosphate vs phosphate group in a nucleotide can contribute to a tighter interaction with a specific protein. In addition, the sulfur atom reduces the pKa value of the phosphate moiety (by ∼1 log unit)29 and leads to a higher electron density of the thiophosphate vs phosphate moiety, which may also increase electrostatic interactions with a protein binding site. Indeed, previously we reported enhancement of biochemical/pharmacological activity of nucleotides upon replacement of a phosphate group by thiophosphate.1116 Furthermore, we and others characterized nucleoside 5′phosphorothioate analogues, as biocompatible Zn(II)-ion chelators.28,19 Hence, our working hypothesis here was that the replacement of both nonbridging oxygen atoms at Pα of a nucleotide by sulfur atoms may confer unique and possibly beneficial pharmacological properties to the resulting compound. Increase of the volume of the Pα,α-dithiophosphate vs thiophosphate group may lead to a tighter fit to the NPP1 3942

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

ammonium bicarbonate solution with no need for further purification by HPLC (94−98% purity). Compounds 10 and 12 were obtained by reaction of 13 with pyrophosphate tris-tributylammonium salt (1 and 0.5 equiv, respectively) at 40% and 31% overall yield, respectively. Compounds 9 and 11 were obtained by reaction of 13 with orthophosphate bis-tributylammonium salt or methylenediphosphonate tris-tributylammonium salt at 18% and 44% overall yield, respectively. Analogues 9−12 Are Potent and Selective NPP1 Inhibitors. Analogues 9−12 were evaluated for their ability to inhibit human NPP1. We found that all these analogues at 100 μM potently inhibited human NPP1 (80−100% inhibition) using thymidine 5′-monophosphate p-nitrophenyl ester, pNPTMP, at 100 μM as the substrate (Figure 3A). In particular, uridine 5′-Pα,α-S,S-triphosphate, 10, proved to be a highly potent and selective NPP1 inhibitor (100% inhibition at 100 μM), with no inhibition of the related NPP3 and other ectonucleotidases (NTPDase1, -2, -3, -8). Analogues 9 and 11 had negligible effects on the ATPase activity of human NTPDases and a minor effect on human NPP3: 15% and 21% inhibition for analogues 9 and 11, respectively. Analogue 12 effectively blocked human NPP1 activity using either pNPTMP or ATP (100 μM) as the substrates (Figure 3A and Figure 3B). Next, IC50 and Ki values were determined for the promising analogues, 10−12. Analogue 12 exhibited Ki = 27 nM with pNP-TMP as the substrate, being approximately 10 and 40 times more effective than analogues 10 and 11, respectively (Table 1). Analogue 12 was selective to NPP1 vs human NTPDase1, -2, -8 but mildly inhibited human NTPDase3 and human NPP3 by 24% and 27%, respectively.

Table 2. Stability of Analogues 9−12 to Hydrolysis by Human Ectonucleotidasesa relative activity ± SEM of ATP hydrolysis (%) ectoenzyme

9

NTPDase1 NTPDase2 NTPDase3 NTPDase8 NPP1

0 0.6 ± 0.2 1.0 ± 0.2 0.2 ± 0.3 0

10 30.8 32.0 39.7 16.5 0

± ± ± ±

11 3.2 3.3 1.9 2.8

0 0 0 0 0

12 2.7 2.0 3.2 1.1 0

± ± ± ±

0.4 0.2 0.6 0.6

a

Analogues 9−12 were incubated in the presence of the indicated ectonucleotidase at a concentration of 100 μM. The activity with 100 μM ATP was set as 100%, which was 329 ± 108, 302 ± 34, 263 ± 51, 195 ± 76 nmol of Pi·min−1·(mg of protein)−1 for NTPDase1, -2, -3, and -8, respectively, and 34 ± 7 nmol for NPP1. Data presented are the mean ± SEM of results from two experiments carried out in triplicate.

pressed in AD293 cells. The compounds were tested at 100 μM, and P2YR activation was determined by measuring variations in intracellular calcium concentration (Δ[Ca2+]i) using UDP or UTP as control as described. All of the compounds had no effect at P2Y2 and P2Y4-Rs and a weak agonist activity at P2Y6-R (data not shown). Then, we switched to a more sensitive assay that measures the accumulation of inositol 1-phosphate (IP1), which is also directly linked to Gq activation.32 Analogues 9, 10, and 12 showed limited agonist activities at P2Y6-R, whereas analogue 11 was devoid of any P2Y6-R activity (Figure 4).

Table 1. IC50 and Ki Values Obtained for NPP1 Inhibition by the Promising Analogues 10−12 Using pNP-TMP as a Substratea inhibitor

IC50 (μM)

Ki (μM)

10 11 12

1.2 ± 0.1 4.3 ± 0.1 0.125 ± 0.004

0.24 ± 0.02 1.11 ± 0.19 0.027 ± 0.009

a Data presented are the mean ± SEM of at least three independent experiments conducted in triplicate.

Figure 4. P2Y6R agonist activity of analogues 9−12 vs UDP was determined by measuring the production of inositol 1-phosphate (IP1).

Analogues 9−12 Resist Hydrolysis by NPP1 and NTPDases. Uridine 5′-Pα,α-phosphorodithioate analogues, 9− 12, were tested for their stability to degradation by NPP1 and NTPDases. These enzymes are involved in the metabolism of extracellular nucleotides. Analogues 9−12 were not hydrolyzed by hNPP1 (Table 2). Analogues 9, 11, and 12 resisted hydrolysis by hNTPDase1, -2, -3, -8 (Table 2). Analogue 10 was hydrolyzed by hNTPDase1, -2, and -3 at 30−39% the rate of ATP and was slightly hydrolyzed by human NTPDase8 at 16% the rate of ATP. Uridine 5′-Pα,α-Dithiophosphate Analogues Have Only Minor Effects on the Uracil Nucleotide Responding P2Y2,4,6 Receptors. Uracil nucleotides are not only substrates of ectonucleotidases but also agonists of P2Y2,4,6-Rs 31 and thus could potentially exert additional biological effects. For instance, activation of P2Y2 receptor results in the inhibition of bone formation.17 To evaluate NPP1 selectivity of analogues 9−12, we tested them also for their ability to activate the uridine-nucleotide sensitive human recombinant P2Y2,4,6-receptors stably ex-

Analogues 9, 10, 11, and 12 do not have any, or very weak, agonistic activities at uracil nucleotide responding receptors when compared to their natural ligands. Unlike the previously reported uridine 5′-Pα-S-triphosphate diastereoisomers, which are highly potent agonists at P2Y2-R and P2Y4-R,26 the presence of two sulfur atoms at Pα-phosphate, as in compounds 9−12, resulted in loss of activity at both P2Y2-R and P2Y4-R. Docking Simulations of 9−12 at Human NPP1 and NPP3 Molecular Models. We used docking simulations to provide insight into the inhibitory activity and NPP1 selectivity of the analogues studied here. For this purpose, we used our previously reported homology models of human NPP1 and NPP3 based on the crystal structures of mouse NPP1.33 First, we confirmed the ability of Glide to reproduce the binding mode of AMP in the crystal structure of mouse NPP1 (PDB code 4GTW). This binding mode was indeed reproduced to within 1.1 Å by the pose with the best (lowest) Emodel energy (Emodel is the recommended scoring function 3943

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

Figure 5. AMP-like and non-AMP-like binding modes of analogues 9−12 (A−D, respectively) shown together at the human NPP1 homology model. Carbon atoms are colored in violet in the AMP-like poses and in green in the non-AMP-like poses. The two Zn(II) ions are shown as purple spheres. Hydrogen bonds are shown as orange dashed lines.

for comparing different poses of the same ligand; see https:// www.schrodinger.com/kb/1027) suggesting that Glide is a suitable docking tool for this system. Interestingly, Glide also predicted an alternative binding mode where the monophosphate moiety largely maintains its position as well as chelation of the zinc ions but the nucleotide moiety adopts an alternative conformation primarily interacting with residues H362, K273, N259, and K510. Next, analogues 9−12 were docked into the hNPP1 and hNPP3 binding sites and their lowest energy poses were analyzed. The results are presented in Figures 5 and S1−S3 and Table SI1. In general, two types of binding modes were observed: one that highly resembles the binding mode of AMP in the crystal structure of mouse NPP1 (AMP-like conformations), and an alternative binding mode also observed for AMP (see above). In the case of NPP1, all ligands adopt both AMP-like and nonAMP-like conformations (Figures 5 and S1), whereas for NPP3 all analogues adopt AMP-like conformation, and only analogue 11 also adopts a non-AMP-like conformation (Figures S2 and S3). Interestingly, the diuridine tetrathiotetraphosphate analogue 12 occupies both binding regions simultaneously (Figures 5, S1, S4). These binding modes suggest that all analogues could compete with ATP for binding-site interactions at NPP1. In agreement with the experimental data (Table 1), Glide predicts analogue 12 to have the highest affinity to NPP1, followed by analogues 9−11 which all have similar Glide scores (Table SI1). The improved binding of analogue 12 to NPP1 could be attributed to its simultaneous occupancy of both binding regions leading to enhanced electrostatic interactions (primarily salt bridges and H-bonds, Table SI1). Furthermore, all analogues are predicted to be poorer binders to NPP3 than to NPP1, again in agreement with the experimental data (Tables 1 and SI1). A comparative examination of the interaction patterns (Figures 5 and S1−S3 and Table SI1) suggests that similar to the trend observed in NPP1 potency, NPP1 selectivity could be primarily attributed to the effect of electrostatic interactions. Thus, all analogues form a larger

number of salt bridges and H-bonds with the NPP1 site than with the NPP3 site. We have previously shown that the ligand binding site in NPP1 is enriched with Lys residues, in comparison with NPP3 (7 vs 1 Lys residues in NPP1 and NPP3, respectively).33 In accord with this observation, the electrostatic potential in the ligand binding site of NPP1 is more positive than in that of NPP3, making it a more suitable binding site for the negatively charged nucleotides. Examining the Glide generated poses suggests that analogues 10 and 12 demonstrated preference to chelation of zinc ions via their phosphate nonbridging oxygen atoms for both NPP1 and NPP3, whereas analogues 9 and 11 demonstrate no preference for S-chelation over O-chelation for NPP1 and preference for O-chelation for NPP3. Other poses for all analogues feature either O- or S-chelation for the two proteins. Since Zn(II) is known to be a thiophilic ion, these Zn(II)-chelation patterns may seem counterintuitive yet they echo previous docking results as well as the analysis of similar complexes from the PDB.15 We note however that the computational study of the exact Zn(II) ions chelation patterns of the studied analogues may require a quantum-mechanical (QM) treatment (e.g., QM docking). Table SI1 lists the interactions formed between analogues 9−12 at the human NPP1 and NPP3 homology models. Evaluation of the Chemical Stability of Uridine 5′-Pα,αDithiophosphate Analogues. Orally administered pharmacologically active molecules should resist acidic and slightly basic hydrolysis in the different parts of the gastrointestinal tract and should be stable against oxidizing agents. The newly synthesized uridine 5′-Pα,α-dithiophosphate analogues have a number of chemical functionalities that could be hydrolyzed under acidic or basic conditions including phosphodiester bonds, glycosidic bond, and a dithiophosphate moiety that can undergo desulfurization. The latter moiety may also undergo oxidation to the corresponding disulfide product. Previously, relatively high stability was reported for the related nucleotide, 2-benzylthio-ATP-α-S, under pH 1.4 and 37 3944

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

Figure 6. Kinetic profiles showing the changes in the percentage of 10 (A), UTP (B), UDP (C) under acidic conditions, pD 2.1 (pH 1.7), as monitored by 31P NMR at 243 MHz at 298 K for 23 days.

°C, exhibiting half-life of 264 h.34 We found that 2-methylthioADP-α-S was less stable than 2-benzylthio-ATP-α-S: after 48 h at pH 7.4, 22% of the material decomposed.24 To evaluate the chemical stability of uridine 5′-Pα,αdithiophosphate analogues, we measured by 31P NMR the time-dependent percentage of the analogues remaining under basic and acidic conditions and due to air oxidation. Specifically, the stability of UTP 5′-Pα,α-dithiophosphate, 10, at rt and pD 2.1 (pH 1.7) was monitored for 23 days. During that time only 15% of the starting material decomposed (Figure 6A, Figure S5). Compound 10 under these acidic conditions was slightly less stable than the parent compound, UTP, 11% of which decomposed after 20 days (Figure 6B), indicating that the phosphorodithioate moiety does not significantly reduce the chemical stability of UTP. At pD 2.1 UDP was more stable than UTP (only 7% decomposition after 28 days) (Figure 6C). Mass spectrum (ESI negative) and 31P NMR analysis of freeze-dried 10 after 23 days at pD 2.1 revealed hydrolysis products apart from 10: uridine 5′-Pα,α-dithiodiphosphate, 9, and inorganic phosphate (Figure S6).

Compound 10 was found to be relatively stable also under basic conditions, pD 12.1. After 25 days, 31P NMR spectrum showed that only 24% of 10 underwent decomposition (Figure 7). The hydrolysis products included uridine 5′-Pα,α-dithiodi-

Figure 7. Kinetic profile showing the changes in the percentage of 10 under basic conditions, pD 12.1 (pH 11.7), as monitored by 31P NMR at 243 MHz at 298 K for 25 days. 3945

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

of the phosphodiester bond between the diphosphonate moiety and Pα,α-dithiophosphate followed by hydrolysis of the dithiophosphate group. The absence of hydrolysis products containing a dithio- or thiophosphate moiety is probably due to further hydrolysis of sulfur-containing intermediates to orthophosphate (Figure 9). Compound 11 was found to be relatively stable also under basic conditions, pD 12.1. After 25 days at rt, only 23% of 11 underwent decomposition (Figure S15) giving rise to uridine 5′-Pα,α-dithiomonophosphate, 17, inorganic dithiophosphate, 18, uridine 5′-Pα,α-dithio(mercaptoethylene)monophosphate, 20; and diphosphonate 21 (Figure S16). Under neutral pH (air-flow, 3 weeks) 11 showed complete stability. In summation, uridine 5′-Pα,α-dithiophosphate analogues demonstrated a much greater stability under acidic and basic conditions than the related nucleotides, 2-benzylthio-ATP-Pα-S and 2-methylthio-ADP-Pα-S.24,34 This increased stability likely results from the additional sulfur atom attached to the Pα, possibly due to the greater volume of the dithiophosphate vs thiophosphate moiety which sterically hinders attack by a water molecule or hydroxide ion. As expected, the most stable analogue of this series under acidic conditions was 11, due to replacement of Pβ,Pγ-bridging oxygen atom by a methylene group (Table 3). Under basic conditions, diuridine (5′,5″-Pα,α-dithio)tetraphosphate, 12, showed the highest stability, due to the absence of a terminal phosphate group. Under oxidizing conditions the compounds were completely stable for at least 3 weeks. This increased stability to oxidizing conditions possibly resulted from steric hindrance at Pα phosphate. Apparently, the phosphorodithioate moiety in analogue 10 only slightly reduced the chemical stability as compared to UTP, unlike the related 5′-Pα-thiophosphate-containing compounds where the thiophosphate group significantly lowered the stability compared to the parent compounds (Table 3).24,34 Yet, compounds 11 and 12 showed higher stability than that of UTP.

phosphate, 9; uridine 5′-Pα,α-dithiomonophosphate, 17; inorganic dithiophosphate, 18; UMP, 19; and uridine 5′-Pα,αdithio(mercaptoethylene)monophosphate, 20, resulting from dithiaphopholane ring opening without thiirane release; and inorganic phosphate (Figure S7). Interestingly, the phosphorodithioate moiety in 17 and 18 remained intact and did not undergo basic hydrolysis to the corresponding phosphorothioate and phosphate analogues. Under these basic conditions 10 is less stable than UTP (which exhibited only 10% decomposition after 20 days) (Figure S8).

Compound 10 also exhibited remarkably high stability to air oxidation. An aqueous solution of 10 was subjected to a constant air flow. After 3 weeks at rt 31P NMR spectrum of the solution showed no new signals, indicating no tendency to disulfide bond formation due to air oxidation. When compound 12 was subjected to pD 1.8 (pH 1.4) for 16 days at rt, 31 P NMR spectrum showed only 6% decomposition of 12 (Figures S9, S10). The only hydrolysis product was inorganic phosphate, probably due to a series of hydrolytic steps described in Figure 8. Compound 12 was found to be relatively stable also under basic conditions, pD 12.6. After 16 days, 31P NMR spectrum showed a loss of only 13% of 12 (Figure S11) and formation of uridine 5′-Pα,αdithiomonophosphate, 17, inorganic dithiophosphate, 18, and inorganic phosphate (Figure S12). The decomposition products of 12 under basic conditions differed from those under acidic conditions, indicating a much more rapid conversion of dithiophosphate to inorganic phosphate in acidic environment than under basic conditions. Uridine 5′-Pα,α-dithiomonophosphate, 17, and inorganic dithiophosphate, 18, are intermediates giving rise to inorganic phosphate. In addition, compound 12 was completely resistant to air oxidation for at least 3 weeks. Likewise, 11 showed only 6% decomposition after 25 days at pD 2.1 (Figure S13). Hydrolysis products included methylenediphosphonate, 21 (Figure 10), and inorganic phosphate (Figures S14 and S15). These products are due to hydrolysis



CONCLUSIONS Calcific aortic valve disease and calcium pyrophosphate dihydrate disease, whose pathology is related to abnormally high level of NPP1, are considered as unmet medical needs. Therefore, the need for biocompatible, potent, selective, and stable NPP1 inhibitors is urgent. Here, we developed novel NPP1 inhibitors based on uridine 5′-Pα,α-dithiophosphate analogues, 9−12. These analogues were designed to maintain

Figure 8. Hydrolysis products of 12 obtained under acidic conditions. 3946

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

Figure 9. Hydrolysis products of 11 under acidic conditions for 25 days.

Inhibitor 10, although 10-fold less active than 12, is more selective than 12, exhibiting practically no inhibition of NPP3, as well as of NTPDase1, -2, -3, -8. The latter enzymes were activated by 10 by 7−27%. Analogue 12 represents a second-generation NPP1 inhibitor with superior pharmacological properties as compared to those of the first generation NPP1 inhibitors based on adenosine Pαphosphorothioate analogues (e.g., 5).15 Thus, 12 exhibited Ki = 27 nM with pNP-TMP as the substrate, as compared to Ki = 685 nM for the first generation inhibitor, ATP-α-thio-β,γ-CCl2, 5A. Furthermore, 12 at 100 μM was selective to NPP1 vs human NTPDase1, -2, -8. In addition 12 only mildly inhibited human NPP3 by 27%, whereas 5A inhibited NPP3 activity by 43%. Moreover, unlike 5 where half of the nucleotide is wasted due to chirality of Pα group and inactivity of one of the diastereoisomers, Pα moiety of 12 is nonchiral, resulting in easy purification and no waste of product. In summary, analogue 12 represents a highly active, selective, and stable NPP1 inhibitor. Its evaluation in CAVD animal models and in pathological human chondrocytes will be reported in due course.

Figure 10. Structures of decomposition products of 11 subjected to hydrolysis at pD 12.1 (pH 11.7) at rt for 25 days.

Table 3. Comparison of Percent of Degradation of Nucleotide Analogues after 5 Days under Basic/Acidic Conditions no.

compd

basic pH

acidic pH

1 2 3 4 5 6

UTP 10 11 12 2-Bz-S-ATP-α-S34 2-MeS-ADP-α-S24

8% 13% 8% 9% N/A 22% after 2 d

5% 11% 2% 6% 25% N/A



the desired pharmacological properties of adenosine 5′-Pαphosphorothioate analogues, which we reported before,15 while avoiding the presence of a chiral center at Pα-phosphate group. Analogues 9−12 proved to be highly stable to air oxidation and significantly chemically stable to acidic and basic pH, turning them into good candidates for further drug development. These findings prompted the evaluation of the inhibitory activity of analogues 9−12 at NPP1. All these analogues potently inhibited hNPP1 (80−100% inhibition) at 100 μM, with nearly no effect or small effect on the related NPP3, as well as other ectonucleotidases (NTPDase1, -2, -3, -8), or uridine nucleotide sensitive purinergic receptors (P2Y2,4,6-Rs). A most promising analogue was diuridine 5′-Pα,α,5″-Pα,αtetrathiotetraphosphate, 12, exhibiting Ki of 27 nM. Docking simulations have indicated that the enhanced NPP1 inhibitory activity and selectivity of 12 could be attributed to a larger number of interactions formed between analogue 12 and the NPP1 site in comparison with the other analogues, likely due to the simultaneous occupancy of both binding regions of NPP1 by this compound. Likewise, more interactions are formed between analogue 12 and the NPP1 vs the NPP3 catalytic site.

EXPERIMENTAL SECTION

Synthesis. General. All air and moisture sensitive reactions were carried out in flame-dried, argon-flushed, two-neck flasks sealed with rubber septa, and the reagents were introduced by syringe. Progress of reactions was monitored by TLC on precoated Merck silica gel plates (60F-254). Visualization was accomplished by UV light. Flash chromatography was carried out on silica gel (Davisil Art. 1000101501). Alternatively, separation was carried out using an HPFC automated flash purification system (Biotage SP1 separation system (RP)). Compounds were characterized by NMR using Bruker spectrometers (300, 400, 600, 700 MHz). Nucleotides were characterized also by 31P NMR in D2O, using 85% H3PO4 as an external reference. High resolution mass spectra were recorded on an AutoSpec Premier (Waters UK) spectrometer by chemical ionization. Nucleotides were analyzed under ESI (electron spray ionization) conditions on a Q-TOF microinstrument (Waters, UK). Primary purification of the nucleotides was achieved on a LC (Isco UA-6) system using a Sephadex DEAE-A25 column, swollen in 1 M NaHCO3 at +4 °C for 1 day. The resin was washed with deionized water before use. The LC separation was monitored by UV detection at 280 nm. A buffer gradient of NH4HCO3 was applied as detailed below. Final purification of the nucleotides was achieved on an HPLC (MerckHitachi) system, using a semipreparative reverse-phase column (Gemini 5u C-18 110A, 250 mm × 10.00 mm, 5 μm, Phenomenex, 3947

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

The residue was dissolved in water and extracted with DCM (×3, 20 mL). The aqueous phase was freeze-dried. The crude residue was deprotected by the general procedure for methoxymethylidene deprotection, as described above for 6. Separation of the crude material was performed by liquid chromatography on a Sephadex DEAE-A25 column with a linear gradient of 0−0.3 M NH4HCO3 buffer (pH 7.5, 600 mL of water and 600 mL of the buffer). The product eluted at 0.27 M buffer. The fractions containing the product were freeze-dried. The excess of the buffer was removed by repeated freeze-drying cycles (dissolution of a solid residue each time in deionized water followed by freeze-drying). The product was obtained in 40% yield (28.7 mg). 1H NMR (D2O, 300 MHz): δ 8.07 (d, J = 7.8 Hz, 1H), 6.02 (d, J = 4.8 Hz,1H), 5.93 (d, 7.8 Hz, 1H), 4.39 (m, 5H). 31 P NMR (D2O, 242.9 MHz): δ 100.7 (d, J = 31.3 Hz), −5.8 (d, J = 13.1 Hz), −22.3 (dd, J1 = 31.3 Hz, J2 = 13.1 Hz) ppm. 13C NMR (D2O, 600 MHz): δ 166.96 (s), 152.49 (s), 142.82 (s), 103.29 (s), 89.26 (s), 83.69 (s), 74.58, 70.39 (s), 65.43 (s) ppm. (ESI negative): m/z calcd for C9H15N2O13P3S2, 514.9156; found 514.9160. Purity data were obtained on an analytical column: tR = 2.56 min (95% purity) using solvent system I (98:2 TEAA(pH 7.4)/CH3CN over 10 min, 1 mL/min); tR = 2.36 min (97% purity) using solvent system II (99:1 0.01 M KH2PO4 (pH 4.5)/CH3CN over 10 min, 1 mL/min). UTP 5′-O-Pα,α-Dithiophosphate-β,γ-methylene (11). Tetrabasic methylenediphosphonate tetrasodium salt was converted to tristributylammonium salt by elution of an aqueous solution of the sodium salt through DOWEX hydrogen form cation exchange resin. The obtained solution was freeze-dried. Uridine 2′,3′-methoxymethylidene-5′-O-(2-thio-1,3,2-dithiaphospholane), 13 (200 mg, 0.45 mmol), and methylenediphosphonate tris-tributylammonium salt (659 mg, 0.9 mmol) were mixed in a 2-necked flask and dried overnight in a vacuum oven over P2O5. After the flask was filled with argon and closed with a rubber septum, a dry acetonitrile (1 mL) was injected and the residue was dissolved after a short stirring. Freshly distilled DBU (137 mg, 0.9 mmol) in dry acetonitrile (0.5 mL) was added dropwise during 10 min at rt. After 30 min of a stirring under argon, typical signals of the product were detected at 100 ppm region by 31P NMR spectrum. The stirring was continued for additional 2 h with 31P NMR monitoring. No further reaction progress was observed. The reaction was stopped, the solvent was evaporated, and the residue was extracted with water and DCM (×3, 20 mL). The aqueous phase was freeze-dried. The methoxymethylidene group was removed as described above. Separation of the crude material was performed by liquid chromatography on a Sephadex DEAE-A25 column with linear gradient of 0−0.5 M NH4HCO3 buffer (pH 7.5, 700 mL of water and 700 mL of the buffer). The product was eluted at 0.34 M buffer. The fractions containing the product were freeze-dried. The excess of the buffer was removed by repeated freeze-drying cycles. The product was obtained in 44% overall yield (101 mg) after separation. 1H NMR (D2O, 400 MHz): δ 8.1 (d, J = 8.0 Hz, 1H), 5.94 (d, J = 4.4 Hz, 1H), 5.90 (d, J = 8.0 Hz, 1H), 4.35 (m, 2H), 4.27 (m, 3H), 2.27 (t, J = 20.8 Hz, 2H). 31P NMR (D2O, 242.9 MHz): δ 99.0 (d, J = 35.9 Hz), 13.9 (d, J = 35.9 Hz), 12.4 (m) ppm. 13C NMR (D2O, 600 MHz): δ 167.75, 153.26, 143.61, 104.07, 90.09, 84.44, 75.39, 71.19, 66.29, 30.759 (dd, J = 116.8 Hz) ppm. HRMS (ESI): calcd for C10H14N2O12P3S2, 513.9447; found, 512.9375 (M − H+). Purity data were obtained on an analytical column: tR = 2.28 min (92% purity) using solvent system I (98:2 TEAA (pH 7.4)/CH3CN over 10 min, 1 mL/min); tR = 2.13 min (98% purity) using solvent system II (99:1 0.01 M KH2PO4 (pH 4.5)/CH3CN over 10 min, 1 mL/min). Diuridine 5′-O-Pα,α,5″-O-Pα,α-Tetrathiotetraphosphate (12). Uridine 2′,3′-methoxymethylidene-5′-O-(2-thio-1,3,2-dithiaphospholane), 13 (200 mg, 0.45 mmol), and pyrophosphate tristributylammonium salt (360 mg, 0.45 mmol) were mixed in 2-neck flask and dried overnight in a vacuum oven over P2O5. After the flask was filled with argon and closed with a septum, dry acetonitrile (1 mL) was injected and the residue was dissolved after a 10 min stirring. Freshly distilled DBU (137 mg, 0.9 mmol) in dry acetonitrile (0.5 mL) was added dropwise at rt. After stirring for 1 h under argon, typical signals of the product were detected by 31P NMR. The stirring was continued for additional 4 h with 31P NMR monitoring. After 4 h the

Torrance, USA). The purity of the nucleotides was evaluated with an analytical reverse-phase column (Gemini 5u C-18 110A, 150 mm × 4.60 mm; 5 μm; Phenomenex, Torrance, CA) using two solvent systems: solvent system I, (A) 1 M triethylammonium acetate (TEAA), pH 7, and (B) CH3CN; solvent system II, (A) 0.01 M KH2PO4, pH 4.5, and (B) CH3CN. The details of the solvent system gradients used for the separation of each product are given below. The purity of the nucleotides was generally ≥95%. All commercial reagents were used without further purification, unless otherwise noted. All reactants in moisture sensitive reactions were dried overnight in a vacuum oven. All phosphorylation reactions were carried out in flamedried, argon-flushed, two-neck flasks sealed with rubber septa. Nucleosides were dried in vacuum overnight. Phosphorus oxychloride and phosphorus trichloride were distilled and kept under nitrogen. 1,8Diazabicyclo[5.4.0]undec-7-ene (DBU) was distilled under reduced pressure before use. CHCl3 was distilled over P2O5. UDP 5′-O-Pα,α-Dithiophosphate (9). Uridine 2′,3′-methoxymethylidene-5′-O-(2-thio-1,3,2-dithiaphospholane), 13 (63 mg, 0.14 mmol), and a phosphate bistributylammonium salt (131 mg, 0.28 mmol) were mixed in a 2-neck flask and dried for 4 h in a vacuum oven over P2O5. After the drying, the flask was filled with argon and the residue was dissolved in dry acetonitrile (1 mL) at rt. DBU (43 mg, 0.28 mmol) was injected dropwise via a septum and the solution immediately turned slightly turbid. After 1 h of stirring at rt and under positive pressure of argon, typical signals of the product were detected by 31P NMR: 100.5 (d), −11 (d) ppm. The reaction mixture was stirred overnight followed by extraction with DCM and water. The aqueous phase was freeze-dried. The end of the reaction was confirmed by 31P NMR spectrum after 20 h reaction. The solvent was evaporated; the residue was dissolved in water and extracted with DCM. The aqueous phase was freeze-dried. The crude residue was subjected to deprotection of methoxymethylidene group as described below. The separation of UDP 5′-O-α,α-dithiophosphate was performed by liquid chromatography on Sephadex DEAE A-25 column with a linear gradient of 0−0.3 M NH4HCO3 buffer (pH 7.5, 600 mL of water in one beaker and 600 mL of the buffer in other one). Fractions containing the product were collected and freezedried. An excess of the buffer was removed by repeated freeze-drying cycles (dissolution of a solid residue each time in deionized water followed by freeze-drying). The product was obtained at an 18% overall yield after separation (12 mg). 1H NMR (D2O, 600 MHz): δ 8.0 (d, J = 6.6 Hz, 1H), 5.93 (d, J = 4.8 Hz, 1H), 5.85 (d, J = 8.0 Hz, 1H), 4.36 (m, 1H), 4.31 (m, 1H), 4.26 (m, 3H) ppm. 31P NMR (D2O, 162 MHz): δ 96.6 (d, J = 34.6 Hz), −7.2 (d, J = 34.6 Hz) ppm. 13C NMR (CDCl3, 600 MHz): δ 165.6, 151.5, 142.2, 101.9, 86.7, 83.6, 81.0, 58.9, 53.1, ppm. HRMS (ESI): calcd for C9H14N2O10P2S2, 435.9567; found, 434.9494 (M − H+). Purity data were obtained on an analytical column: tR = 1.78 min (93% purity) using solvent system I (98:2 TEAA (pH 7.4)/CH3CN over 10 min, 1 mL/min); tR = 1.51 min (99% purity) using solvent system II (99:1 0.01 M KH2PO4 (pH 4.5)/CH3CN over 10 min, 1 mL/min). General Procedure for Deprotection of Uridine 2′,3′-OMethoxymethylidene Nucleotides. Protected nucleotide 9 was dissolved in water (2 mL). 10% HCl was added dropwise at rt until pH 2.3 was obtained. The reaction solution was stirred for 2.5 h, and then 24% NH4OH was added at rt until pH 9.0 was obtained. After 45 min of stirring, the residue was freeze-dried and applied on a Sephadex DEAE A-25 column for LC purification. UTP 5′-O-Pα,α-Dithiophosphate (10). Uridine 2′,3′-methoxymethylidene-5′-O-(2-thio-1,3,2-dithiaphospholane), 13 (90 mg, 0.2 mmol), and a pyrophosphate tris-tributylammonium salt (294 mg, 0.4 mmol) were mixed in a 2-neck flask and dried overnight in a vacuum oven over P2O5. The flask was filled with argon and closed with a rubber septum. Dry acetonitrile (0.5 mL) was injected via the septum, and after a short stirring the reagents were added. Freshly distilled DBU (61 mg, 0.4 mmol) was added dropwise in dry acetonitrile (0.5 mL) at rt during 15 min. After 25 min of stirring under argon, typical signals of the product were detected by 31P NMR spectrum. The stirring was continued for additional 1.5 h without any change of 31P NMR spectrum. The reaction was stopped by solvent evaporation. 3948

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

heating the samples 2 min at 95 °C and pyrophosphate produced was measured using the pyrophosphate assay kit from Abcam (ON, Canada) according to their instructions. The analogues replaced ATP when tested as a potential substrate. NTPDase Enzymatic Assays. Activity was measured in 0.2 mL of Tris-Ringer incubation buffer (in mM, 120 NaCl, 5 KCl, 2.5 CaCl2, 1.2 MgSO4, 25 NaHCO3, 5 glucose, 80 Tris, pH 7.4) at 37 °C with or without analogues. NTPDase’s cell lysate (1−3 μg) was added to the incubation mixture and preincubated at 37 °C for 3 min. The reaction was initiated by the addition of the substrate (100 μM ATP), and stopped after 10 min with 50 μL of malachite green reagent. The released inorganic phosphate (Pi) was measured at 630 nm by the malachite green method.36 The analogues were also tested as potential substrate. For the blank samples, the enzymes were added after the reaction was stopped (this blank was subtracted for each condition). Evaluation of Compounds 9−12 as P2Y2,4,6-Receptor Ligands. Compounds 9−12 were tested as ligand for the human P2Y2, P2Y4, and P2Y6 receptors. The compounds were tested at 100 μM, and we measured the calcium response as previously described.37 A second method which measures the accumulation of inositol 1phosphate (IP1) was also used. The technique is also directly linked to Gq activation. This latter second method has the advantage over the calcium assays of being more sensitive. For testing the antagonist activity of compound 9 at P2Y6-R, the analogue was added at a concentration of 100 μM prior to the addition of 0.03 μM UDP, and IP1 was evaluated. Molecular Modeling. Docking. Three-dimensional (3D) models of human NPP1 and NPP3 were previously reported based on the crystal structure of mouse NPP1 (PDB code 4GTW).33 These models were used for the docking of analogues 9−12. Prior to docking, the protein structures were prepared using the Protein Preparation Wizard in Schrodinger’s Maestro. Docking simulations were performed using Glide as implemented in Maestro 9.8. Glide’s grid box was centered on the binding sites as deduced from the mouse NPP1 crystal structure (PDB code 4GTW). A docking grid was generated within the docking box, and ligands were docked into the binding site using Glide’s Standard Precision (SP) option with default parameters. For each ligand, poses were selected based on Glide Emodel Score energies (as recommended in https://www.schrodinger.com/kb/1027) and analyzed for their interactions with binding site residues. However, for ranking the interactions of the different analogues in the binding site, Glide Score (gscore) was used. Analysis of Chemical Stability of Analogues 10−12. The stability of 2 mM solutions of 10−12 at pD 1.8−2.1 (pH 1.4−1.7), or pD 12.1, was monitored by 31P NMR at 243 MHz, at 298 K, and MS (ESI negative) for 23−25 days. The percentage of remaining 10−12 was obtained from the ratio between the integration of signals of 10− 12 and all signals in the spectrum. A 1.8 mM D2O solution of 10−12 at 298 K was subjected to a constant air flow for 3 weeks to evaluate resistance to air oxidation using 31P NMR and MS.

reaction progress has stopped. The solvent was evaporated and the residue was extracted with water and chloroform (×3, 20 mL). The aqueous phase was freeze-dried. The methoxymethylidene group was removed as described above. Separation of the crude material was performed on a Sephadex DEAE-A25 column with linear gradient of 0−1 M of NH4HCO3 buffer (pH 7.5, 900 mL of water and 900 mL of the buffer). The product eluted at 0.8 M buffer. The fractions containing the product were freeze-dried. An excess of the buffer was removed by repeated freeze-drying cycles. The product was obtained with overall yield of a 31% (118.6 mg) after separation. 1H NMR (D2O, 400 MHz): δ 8.0 (d, J = 7.8 Hz, 2H), 5.94 (d, J = 4.8 Hz, 2H), 5.86 (d, J = 8.0 Hz, 2H), 4.35 (m, 2H), 4.3 (m, 6H), 4.28 (m, 2H). 31P NMR (D2O, 162 MHz): δ 100.6 (d, J = 32 Hz), −24.9 (d, J = 32 Hz) ppm. 13C NMR (D2O, 600 MHz): δ 164.2, 151.76, 139.03, 103.52, 87, 89, 84, 68, 71.65, 68.11, 62.29 ppm. HRMS (ESI positive: m/z calculated for C18H22N4O19P4S4 849.8779, found 849.8901 (M − 4H+)4−. Purity data were obtained on an analytical column: tR = 12.23 min (93% purity) using solvent system I (98:2 TEAA (pH 7.4)/ CH3CN over 10 min, 1 mL/min); tR = 11.68 min (97% purity) using solvent system II (99:1 0.01 M KH2PO4 (pH 4.5)/CH3CN over 10 min, 1 mL/min). Uridine 2′,3′-Methoxymethylidene-5′-O-(2-thio-1,3,2-dithiaphospholane) (13). DIPEA (225 mg, 1.74 mmol) was added to a dry 2-neck flask with uridine 2′,3′-methoxymethylidene (450 mg, 1.57 mmol) in dry acetonitrile (2 mL) under argon pressure. The reaction mixture was cooled to −38 °C (acetonitrile−dry ice bath), and chloro1,3,2-dithiaphospholane30 (248 mg, 1.57 mmol) in dry acetonitrile (0.5 mL) was injected through a septum. The mixture was stirred at −38 °C for 2.5 h, and then 31P NMR spectrum of the crude mixture showed the disappearance of the starting material signal at 170 ppm and a new signal at 150 ppm. Dry sulfur (755 mg, 23.6 mmol) was added as grounded powder. The reaction temperature was elevated to rt. After an overnight reaction, the intermediate uridine 2′,3′methoxymethylidene-5′-O-(1,3,2-dithiaphospholane) was completely consumed, and the signal of the product was detected at 124 ppm (two diastereomers). The solvent was evaporated. DCM/EtOAc (1:1, 5 mL) mixture was added to the crude mixture, and the residue was filtered through Buchner funnel. The filtrate was evaporated. The crude residue was separated on a silica gel column with gradient of DCM/EtOAc mixture (from 0% to 80% of EtOAc) under TLC monitoring (DCM/EtOAc 1:1, Rf = 0.55). Fractions containing the product were collected and evaporated. The product was obtained at a 55% (561 mg) overall yield after separation as two diastereomers. 1H NMR (CDCl3, 600 MHz): δ 9.68 (s, 1H), 7.38 (d, 1H, J = 13 Hz), 5.96 (2s, 1H), 5.73 (d, 1H, J = 13 Hz), 5.70 (m, 1H), 5.05 (m, 1H), 5.00 (m, 1H), 4.9 (m, 1H), 4.42 (m, 1H), 3.45 (s, 4H), 3.31 (s, 3H) ppm. 31P NMR (CDCl3, 162 MHz): δ 124.06 (s), 123.93 (s) ppm. 13C NMR (CDCl3, 600 MHz): δ 165.6, 151.5, 142.2, 125.9, 101.9, 101.2, 86.7, 83.6, 81.0, 58.9, 53.1, 32.5, 28.8 ppm. MS (+ESI): 462.98 (M + Na+). Preparation of Enzymes. Protein extracts were prepared with human NTPDase1, -2, -3, -8 and human NPP1, -3 expressing plasmids as previously described.35 NPP Enzymatic Assays. Evaluation of the effect of analogues on human NPP1 and NPP3 activity was carried out with thymidine 5′monophosphate p-nitrophenyl ester (pNP-TMP) or ATP as the substrate. 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 (100, 50, and 25 μM were also used to evaluate the kinetic parameters). Human NPP1 (4−5 μg) or NPP3 (12−14 μg) cell lysate was added to the incubation mixture and preincubated at 37 °C for 3 min. Reaction was initiated by the addition of 100 μM pNP-TMP, and the production of p-nitrophenol was measured after 15 min reaction at 405 nm. Blank values were subtracted from these data. For the blank samples, the enzyme was added at the end of the reaction and read immediately. The Ki values were calculated by the Dixon plot using three different concentrations of pNP-TMP (25, 50, and 100 μM), with an hNPP1 protein amount about 10 times lower than mentioned above. In the assays where ATP was used as the substrate, the reaction was stopped after 25 min by



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jmedchem.7b01906. Molecular modeling data (2D interaction maps and binding modes of studied analogues), kinetic profiles of of studied analogues under acidic or basic conditions, and molecular formula strings (PDF) Molecular formula strings (CSV)



AUTHOR INFORMATION

Corresponding Author

*Fax: 972-3-6354907. Phone: 972-3-5318303. E-mail: Bilha. [email protected]. 3949

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

ORCID

(13) Goding, J. W.; Grobben, B.; Slegers, H. Physiological and pathophysiological functions of the ecto-nucleotide pyrophosphatase/ phosphodiesterase family. Biochim. Biophys. Acta, Mol. Basis Dis. 2003, 1638, 1−19. (14) Iung, B.; Baron, G.; Butchart, E. G.; Delahaye, F.; GohlkeBärwolf, C.; Levang, O. W.; Tornos, P.; Vanoverschelde, J.-L.; Vermeer, F.; Boersma, E. A prospective survey of patients with valvular heart disease in Europe: The Euro heart survey on valvular heart disease. Eur. Heart J. 2003, 24, 1231−1243. (15) Nadel, Y.; Lecka, J.; Gilad, Y.; Ben-David, G.; Förster, D.; Reiser, G.; Kenigsberg, S.; Camden, J.; Weisman, G. A.; Senderowitz, H.; Sevigny, J.; Fischer, B. Highly potent and selective ectonucleotide pyrophosphatase/phosphodiesterase I inhibitors based on an adenosine 5′-(α or γ)-thio-(α, β-or β, γ)-methylenetriphosphate scaffold. J. Med. Chem. 2014, 57, 4677−4691. (16) Rajamannan, N. M. Calcific aortic stenosis: lessons learned from experimental and clinical studies. Arterioscler., Thromb., Vasc. Biol. 2009, 29, 162−168. (17) Orriss, I. R.; Utting, J. C.; Brandao-Burch, A.; Colston, K.; Grubb, B. R.; Burnstock, G.; Arnett, T. R. Extracellular nucleotides block bone mineralization in vitro: Evidence for dual inhibitory mechanisms involving both P2Y2 receptors and pyrophosphate. Endocrinology 2007, 148, 4208−4216. (18) Côté, N.; El Husseini, D.; Pépin, A.; Guauque-Olarte, S.; Ducharme, V.; Bouchard-Cannon, P.; Audet, A.; Fournier, D.; Gaudreault, N.; Derbali, H.; McKee, M. D.; Simard, C.; Després, J.P.; Pibarot, P.; Bossé, Y.; Mathieu, P. ATP acts as a survival signal and prevents the mineralization of aortic valve. J. Mol. Cell. Cardiol. 2012, 52, 1191−1202. (19) Qian, S.; Regan, J. N.; Shelton, M. T.; Hoggatt, A.; Mohammad, K. S.; Herring, P. B.; Seye, C. I. The P2Y 2 nucleotide receptor is an inhibitor of vascular calcification. Atherosclerosis 2017, 257, 38−46. (20) Eliahu, S. E.; Camden, J.; Lecka, J.; Weisman, G. A.; Sevigny, J.; Gelinas, S.; Fischer, B. Identification of hydrolytically stable and selective P2Y1 receptor agonists. Eur. J. Med. Chem. 2009, 44, 1525− 1536. (21) Eliahu, S.; Lecka, J.; Reiser, G.; Haas, M.; Bigonnesse, F.; Levesque, S. A.; Pelletier, J.; Sevigny, J.; Fischer, B. Diadenosine 5 ′,5 ″-(boranated)polyphosphonate analogues as selective nucleotide pyrophosphatase/phosphodiesterase inhibitors. J. Med. Chem. 2010, 53, 8485−8497. (22) Khan, K. M.; Fatima, N.; Rasheed, M.; Jalil, S.; Ambreen, N.; Perveen, S.; Choudhary, M. I. 1,3,4-Oxadiazole-2(3H)-thione and its analogues: A new class of non-competitive nucleotide pyrophosphatases/phosphodiesterases 1 inhibitors. Bioorg. Med. Chem. 2009, 17, 7816−7822. (23) Patel, S. D.; Habeski, W. M.; Cheng, A. C.; de la Cruz, E.; Loh, C.; Kablaoui, N. M. Corrigendum to "Quinazolin-4-piperidin-4-methyl sulfamide PC-1 inhibitors: Alleviating hERG interactions through structure based design". Bioorg. Med. Chem. Lett. 2010, 20, 2685−2685. (24) Azran, S.; Danino, O.; Förster, D.; Kenigsberg, S.; Reiser, G.; Dixit, M.; Singh, V.; Major, D. T.; Fischer, B. Identification of highly promising antioxidants/neuroprotectants based on nucleoside 5′phosphorothioate scaffold. Synthesis, activity, and mechanisms of action. J. Med. Chem. 2015, 58, 8427−8443. (25) Jacobson, K. A.; Boeynaems, J. M. P2Y nucleotide receptors: promise of therapeutic applications. Drug Discovery Today 2010, 15, 570−578. (26) Jacobson, K. A.; Costanzi, S.; Ivanov, A. A.; Tchilibon, S.; Besada, P.; Gao, Z.-G.; Maddileti, S.; Harden, T. K. Structure activity and molecular modeling analyses of ribose-and base-modified uridine 5′-triphosphate analogues at the human P2Y2 and P2Y4 receptors. Biochem. Pharmacol. 2006, 71, 540−549. (27) Frey, P. A.; Sammons, R. D. Bond order and charge localization in nucleoside phosphorothioates. Science 1985, 228, 541−545. (28) Sigel, R. K. O.; Song, B.; Sigel, H. Stabilities and structures of metal ion complexes of adenosine 5′-O-thiomonophosphate (AMPS2) in comparison with those of its parent nucleotide (AMP2-) in aqueous solution. J. Am. Chem. Soc. 1997, 119, 744−755.

Hanoch Senderowitz: 0000-0003-0076-1355 Bilha Fischer: 0000-0001-8837-0978 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS J.S. received support from the Canadian Institutes of Health Research (CIHR), and he was the recipient of a “Chercheur National” research award from the Fonds de Recherche du Québec, Santé (FRQS).



ABBREVIATIONS USED NPP1, nucleotide pyrophosphatase/phosphodiesterase-1; CAVD, calcific aortic valve disease; CPPD, calcium pyrophosphate dihydrate; NTPDase, nucleoside triphosphate diphosphohydrolase; 5′-NT, 5′-ectonucleotidase; P2-R, purinergic receptor



REFERENCES

(1) Bollen, M.; Gijsbers, R.; Ceulemans, H.; Stalmans, W.; Stefan, C. Nucleotide pyrophosphatases/phosphodiesterases on the move. Crit. Rev. Biochem. Mol. Biol. 2000, 35, 393−432. (2) Yegutkin, G. G. Nucleotide-and nucleoside-converting ectoenzymes: important modulators of purinergic signalling cascade. Biochim. Biophys. Acta, Mol. Cell Res. 2008, 1783, 673−694. (3) Albright, R. A.; Ornstein, D. L.; Cao, W.; Chang, W. C.; Robert, D.; Tehan, M.; Hoyer, D.; Liu, L.; Stabach, P.; Yang, G.; De La Cruz, E. M.; Braddock, D. T. Molecular basis of purinergic signal metabolism by ectonucleotide pyrophosphatase/phosphodiesterases 4 and 1 and implications in stroke. J. Biol. Chem. 2014, 289, 3294−3306. (4) Freeman, R. V.; Otto, C. M. Spectrum of calcific aortic valve disease. Circulation 2005, 111, 3316−3326. (5) El Husseini, D.; Boulanger, M.-C.; Mahmut, A.; Bouchareb, R.; Laflamme, M.-H.; Fournier, D.; Pibarot, P.; Bossé, Y.; Mathieu, P. P2Y2 receptor represses IL-6 expression by valve interstitial cells through Akt: implication for calcific aortic valve disease. J. Mol. Cell. Cardiol. 2014, 72, 146−156. (6) Côté, N.; El Husseini, D.; Pépin, A.; Bouvet, C.; Gilbert, L. A.; Audet, A.; Fournier, D.; Pibarot, P.; Moreau, P.; Mathieu, P. Inhibition of ectonucleotidase with ARL67156 prevents the development of calcific aortic valve disease in warfarin-treated rats. Eur. J. Pharmacol. 2012, 689, 139−146. (7) Hashimoto, S.; Ochs, R. L.; Rosen, F.; Quach, J.; McCabe, G.; Solan, J.; Seegmiller, J. E.; Terkeltaub, R.; Lotz, M. Chondrocytederived apoptotic bodies and calcification of articular cartilage. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 3094−3099. (8) Johnson, K.; Hashimoto, S.; Lotz, M.; Pritzker, K.; Goding, J.; Terkeltaub, R. Up-regulated expression of the phosphodiesterase nucleotide pyrophosphatase family member PC-1 is a marker and pathogenic factor for knee meniscal cartilage matrix calcification. Arthritis Rheum. 2001, 44, 1071−1081. (9) Pattrick, M.; Hamilton, E.; Hornby, J.; Doherty, M. Synovial fluid pyrophosphate and nucleoside triphosphate pyrophosphatase: comparison between normal and diseased and between inflamed and noninflamed joints. Ann. Rheum. Dis. 1991, 50, 214−218. (10) Stefan, C.; Jansen, S.; Bollen, M. NPP-type ectophosphodiesterases: unity in diversity. Trends Biochem. Sci. 2005, 30, 542−550. (11) Shayhidin, E. E.; Forcellini, E.; Boulanger, M. C.; Mahmut, A.; Dautrey, S.; Barbeau, X.; Lagüe, P.; Sévigny, J.; Paquin, J. F.; Mathieu, P. Quinazoline-4-piperidine sulfamides are specific inhibitors of human NPP1 and prevent pathological mineralization of valve interstitial cells. Br. J. Pharmacol. 2015, 172, 4189−4199. (12) Ye, Y.-M.; Yang, E.-M.; Yoo, H.-S.; Shin, Y.-S.; Kim, S.-H.; Park, H.-S. Increased level of basophil CD203c expression predicts severe chronic urticaria. J. Korean Med. Sci. 2014, 29, 43−47. 3950

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951

Journal of Medicinal Chemistry

Article

(29) Sayer, A. H.; Itzhakov, Y.; Stern, N.; Nadel, Y.; Fischer, B. Characterization of complexes of nucleoside-5′-phosphorothioate analogues with zinc ions. Inorg. Chem. 2013, 52, 10886−10896. (30) Okruszek, A.; Olesiak, M.; Balzarini, J. The synthesis of nucleoside 5′-O-(1, 1-dithiotriphosphates). J. Med. Chem. 1994, 37, 3850−3854. (31) von Kügelgen, I.; Hoffmann, K. Pharmacology and structure of P2Y receptors. Neuropharmacology 2016, 104, 50−61. (32) Trinquet, E.; Bouhelal, R.; Dietz, M. Monitoring Gq-coupled receptor response through inositol phosphate quantification with the IP-One assay. Expert Opin. Drug Discovery 2011, 6, 981−994. (33) Lecka, J.; Ben-David, G.; Simhaev, L.; Eliahu, S.; Oscar, J., Jr; Luyindula, P.; Pelletier, J.; Fischer, B.; Senderowitz, H.; Sévigny, J. Nonhydrolyzable ATP analogues as selective inhibitors of human NPP1: a combined computational/experimental study. J. Med. Chem. 2013, 56, 8308−8320. (34) Hillaire-Buys, D.; Shahar, L.; Fischer, B.; Chulkin, A.; Linck, N.; Chapal, J.; Loubatières-Mariani, M. M.; Petit, P. Pharmacological evaluation and chemical stability of 2-benzylthioether-5′-O-(1thiotriphosphate)-adenosine, a new insulin secretagogue acting through P2Y receptors. Drug Dev. Res. 2001, 53, 33−43. (35) Eliahu, S.; Lecka, J.; Reiser, G.; Haas, M; Bigonnesse, F.; Levesque, S. A.; Pelletier, J.; Sévigny, J.; Fischer, B. Diadenosine 5′,5″(boranated)polyphosphonate analogues as selective nucleotide pyrophosphatase/phosphodiesterase inhibitors. J. Med. Chem. 2010, 53, 8485−8497. (36) Baykov, A.; Evtushenko, O.; Avaeva, S. A malachite green procedure for orthophosphate determination and its use in alkaline phosphatase-based enzyme immunoassay. Anal. Biochem. 1988, 171, 266−270. (37) Meltzer, D.; Ethan, O.; Arguin, G.; Nadel, Y.; Danino, O.; Lecka, J.; Sévigny, J.; Gendron, F.-P.; Fischer, B. Synthesis and structure− activity relationship of uracil nucleotide derivatives towards the identification of human P2Y6 receptor antagonists. Bioorg. Med. Chem. 2015, 23, 5764−5773.

3951

DOI: 10.1021/acs.jmedchem.7b01906 J. Med. Chem. 2018, 61, 3939−3951