Exploration of Pyrazolo[1,5‐a]pyrimidines as Membrane‐Bound Pyrophosphatase Inhibitors

Abstract Inhibition of membrane‐bound pyrophosphatase (mPPase) with small molecules offer a new approach in the fight against pathogenic protozoan parasites. mPPases are absent in humans, but essential for many protists as they couple pyrophosphate hydrolysis to the active transport of protons or sodium ions across acidocalcisomal membranes. So far, only few nonphosphorus inhibitors have been reported. Here, we explore the chemical space around previous hits using a combination of screening and synthetic medicinal chemistry, identifying compounds with low micromolar inhibitory activities in the Thermotoga maritima mPPase test system. We furthermore provide early structure‐activity relationships around a new scaffold having a pyrazolo[1,5‐a]pyrimidine core. The most promising pyrazolo[1,5‐a]pyrimidine congener was further investigated and found to inhibit Plasmodium falciparum mPPase in membranes as well as the growth of P. falciparum in an ex vivo survival assay.


Introduction
Parasitic human diseases, such as malaria, leishmaniasis, trypanosomiasis and toxoplasmosis, represent a severe global health concern. The life cycles of protozoan parasites, including Plasmodium spp., Leishmania spp., Trypanosoma spp. and Toxoplasma spp., are rather complex and typically comprise of transitions between hosts and vectors as well as intracellular and extracellular environments. To survive these changes, the protozoan cell must be able to adjust e. g. for fluctuating osmotic pressures. The key mechanism involves membranebound pyrophosphatases (mPPases). [1] These enzymes are located in the cell membrane of bacteria and archaea, but can also be found in the protist acidocalcisome, and in the Golgi apparatus and/or the vacuole of plants and algae. [2][3][4][5][6][7][8] mPPases are important for many protists by acting as ion pumps through pyrophosphate hydrolysis. [9][10][11][12] As shown by knock-out and knock-down studies in P. falciparum, [13] T. brucei [14] and T. gondii, [15] mPPases are required for maintaining in vitro asexual blood stage growth, acidocalcisome acidification and parasitic virulence. Moreover, T. gondii intracellular proliferation was retarded by mPPase-inhibiting bisphosphonate derivatives. [16] In recent years, various crystal structures of mung bean Vigna radiata [17,18] and hyperthermophilic bacterium Thermotoga maritima [17,19,20] mPPases have been solved by us and others.
mPPases are promising therapeutic targets as they play an important role in the lifecycle of many protists but do not exist in multicellular animals. [9][10][11][12] In addition, their three-dimensional structures have been solved in diverse conformations, which makes them amenable to structure-based design. Only a few small molecule inhibitors have been however reported so far. A main class is formed by nonhydrolyzable pyrophosphate (PP i ) analogues, which have the drawback of interfering with several human enzymes hydrolyzing or producing PP i . Examples include ectonucleotide pyrophosphatases, phosphodiesterases and inorganic pyrophosphatases. [21] Previously, we developed a 96-well plate in vitro screening assay using thermostable Thermotoga maritima mPPase (TmPPase) as a model enzyme, and discovered several classes of nonphosphorus inhibitors. [22,23] By screening of commercially available compounds as well as synthetic medicinal chemistry, we explored and identified several low micromolar hits (compounds 1-5; Figure 1). [24] Among those, the isoxazolebased compound 2 retained activity ex vivo against P. falciparum. [24] The binding location of the isoxazoles, whether at the catalytic site or elsewhere, could not be determined experimentally. In contrast, we have demonstrated using X-ray crystallography that the 2-aminobenzothiazole 1 binds allosteri-cally outside the catalytic side, near the so-called exit channel (PDB code: 6QXA). [20] Here, starting from 2 and its congeners, we aimed to identify novel molecular scaffolds that would lead to gain in binding affinity. These could be translated into therapeutic molecules but could also have a higher probability to be cocrystallized at the mPPase catalytic site, which so far has eluded us. Following a screening of 52 compounds, we set to explore the pyrazolo[1,5-a]pyrimidine core. An advantage of swapping the central core structure is that it would allow further chemical expansion of the 2-position in pyrazolo[1,5-a]pyrimidines via Suzuki coupling or other coupling methods. Furthermore, pyrazolo[1,5-a]pyrimidines [25] have been reported to have antibacterial activity against Mycobacterium tuberculosis, [26] antiparasitic activity including antileishmanial, [27] antimalarial (P. falciparum dihydroorotate dehydrogenase inhibitors) [28] and antitrypanosomal, [27] as well as antiviral activity against HIV. [29] Results and Discussion

Exploration of the isoxazole and sulfonamide space through screening
We started from the previously identified 5-arylisoxazole-3carboxylate core present in 2-5 ( Figure 1, highlighted in blue). Using KNIME, [30] we did a substructure search for commercial analogues in the ZINC12 database [31] From our experience, computational methods (including docking simulations) do not offer a robust and/or reliable mean of prioritization at the mPPase. Therefore, we manually picked compounds from a diverse collection provided by a single vendor (Ambinter, 174 matches for the substructure query). Out of the search results, we selected fifteen isoxazoles for testing in vitro. For all compounds in this manuscript, we tested activity using a 96well based assay that detects the inhibition of TmPPase. [22][23][24] In this first set (Supporting Information, Table S1 and S2), eleven of the fifteen isoxazoles had a sulfonamide-linked 2methoxyphenyl moiety. The sulfonamides 6 and 7 ( Figure 2) were the two best compounds with 6 (IC 50 = 5.4 μM) being one of the most potent TmPPase inhibitors found this far. Interestingly, in comparison with 7-10, the sulfonamide-linked 2,3,4,5tetrahydrobenzo[b] [1,4]thiazepine moiety found in 6 is critical for mPPase activity. Even the simple addition of a methyl group (as in 8) was not tolerated. Also, minor alterations of the sevenmembered ring (as in 9 and 10) caused the loss of all TmPPase inhibitory activity. Other ring modifications also proved unsuccessful (Table S2). Because of the sharp, difficult to rationalize, structure-activity relationships of the isoxazole-sulfonamide core, we in this manuscript decided to pursue other scaffolds (see below).
We also tested another set of 18 sulfonamides (Table S3) and 19 amides (Table S4) that were readily available to us and thus were candidates for repurposing. These compounds share some chemical features with 2, with many having a central polar core bearing N and/or O atoms surrounded by more  [20,24] Blue highlight, common substructure used for similarity searches in this manuscript; IC 50 , half maximal inhibitory concentration; CI 95% , half maximal inhibitory concentration expressed as a 95 % confidence interval (given in square brackets).  hydrophobic/aromatic functional groups. Of this set, compounds 11 and 12 ( Figure 3) were the most potent with IC 50 values below 25 μM. Sulfonamide 11 (IC 50 = 14 μM; ligand efficiency (LE) = 0.23) share the 2-methoxyphenyl-sulfonamide moiety of 6 but lacks its isoxazole core, and therefore shows potential for further exploration. Compound 12 (IC 50 = 25 μM) is structurally different from the hitherto most potent nonphosphorus TmPPase inhibitors and thus was picked as a template for further design. Compound 12 was the only compound from this subset with a 4,5-dihydropyrrolo [3,4-c]pyrazol-6(1H)-one core.

Chemical exploration of the pyrazolo[1,5-a]pyrimidine core
Next, we decided to further explore the chemical space around the recently discovered 4,5-dihydropyrrolo [3,4-c]pyrazol-6(1H)one scaffold, using combinations of substituents already found to be favorable for activity. We nonetheless decided to swap the original bicyclic scaffold with a related nitrogen-rich, pyrazolo[1,5-a]pyrimidine core. This approach was synthetically more achievable, allowing easy modifications at three positions. In addition, the synthetic exploration could take advantage of readily available starting materials, such as various aromatic or aliphatic ketones. The scaffold change can be rationalized by comparing the pharmacophores of 17 a with 2 and 12 (Supporting Information, Figure S1 and S2).
We relied on a method reported by Childress et al. [32] as we could adapt the first two steps of their synthesis route to access the key intermediate 15 (Scheme 1). Crossed condensation at room temperature of commercially available 3,5-dimethylacetophenone or acetylcyclopropane with diethyl oxalate and potassium tert-butoxide in THF gave 14. Subsequent ring condensation of 14 with 3-bromo-1H-pyrazol-5-amine in refluxing ethanol resulted in the formation of pyrazolo[1,5-a] pyrimidine 15. Hydrolysis of the formed ethyl ester 15 using lithium hydroxide in a mixture of ethanol and water gave the corresponding carboxylic acid 16, which was reacted with 2bromophenol and 1-[bis(dimethylamino)methylene]-1H-1,2,3triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU) and 2-bromophenol to give the desired 2-bromophenyl carboxylate 17.
The bromine substituent in the 2-position of 15 could serve as a coupling handle for further exploration of the chemical space by various Pd-mediated cross-coupling reactions. Suzuki coupling of ethyl ester 15 in n-propanol with commercially available potassium vinyltrifluoroborate was done under microwave conditions. The product obtained, 18, could be transformed into the corresponding 2-bromophenyl ester 20 following the same hydrolysis and esterification procedures used for 16 and 17, respectively. However, changing trimethylamine used in the Suzuki coupling to ethylenediamine serendipitously led to both Suzuki coupling and amidation occurring in the same reaction mixture, yielding 21 in a single step.

Biological activity of the pyrazolo[1,5-a]pyrimidine core
We started our exploration by taking inspiration from 12 (IC 50 = 25 μM) and changing the isoxazole core of 2 (IC 50 = 6.9 μM) to the corresponding pyrazolo[1,5-a]pyrimidine analogue 17 a (Table 1). This scaffold change was relatively well tolerated, with just a 2-fold loss in activity compared to 2. As previously noticed for the isoxazole series, [24] the corresponding carboxylic acid 16 a was inactive, but the ethyl ester 15 a retained weak activity.
Since the central pyrazolo[1,5-a]pyrimidine core is bulkier than the original isoxazole moiety (the molecular weight of 17 a is approximately 500 Da), we tried to introduce lighter substitutions. The exchange of the R-group at the 7-position, from a 3,5-dimethylphenyl group to a cyclopropyl ring, was only slightly effective for the ethyl carboxylate 15 b. Furthermore, it proved completely unsuccessful for the carboxylic acid 16 b and the 2-bromophenyl carboxylate 17 b. Similarly, replacing the Rsubstituent with other phenyl moieties than the original 3,5dimethylphenyl substituent (unsubstituted or bearing electron withdrawing/donating groups), were generally not well tolerated for the isoxazole derivatives. [24] We next studied further functionalization of the 2-position in pyrazolo[1,5-a]pyrimidines. Bromine atoms are very useful in X-ray crystallography (due to their anomalous scattering, which can aid in identifying the presence of the compound as well as its orientation in low-resolution, 3.5 Å or worse). Additionally, aryl bromides are highly useful e. g. in Suzuki coupling with various organoboron substrates. We introduced a well-accepted vinyl group at the 5-position, showing nearly no loss of activity for 20 a in comparison to the hitherto best pyrazolo[1,5-a] pyrimidine 17 a (IC 50 = 14 μM). Moreover, there was a 2.2-fold improvement in the inhibition comparing the ethyl ester 18 a to the 2-bromo analogues 15 a. Interestingly, the 2-vinyl-substituted carboxylic acid 19 a (IC 50 = 14 μM) was as active as the best pyrazolo[1,5-a]pyrimidines and superior to its inactive 2bromo analogue 16 a. As presented above, the corresponding 2-vinyl-substituted cyclopropyl analogues 18 b-20 b were all inactive. In the same way, the 3,5-dimethylphenyl-substituted amide 21 a showed 3.6-fold higher inhibition than the corresponding cyclopropyl-substituted analogue 21 b.

Follow-up studies of 12, 17 a, 19 a and 20 a
In order to rule out a cause of false positives, colloidal aggregation in the TmPPase model assay was evaluated for compounds 17 a, 19 a and 20 a at six concentrations (100 μM, 50 μM, 20 μM, 10 μM, 1 μM, and 0.1 μM) using the assay conditions ( Figure S3). Compounds 17 a and 20 a showed aggregate formation at concentrations above 20 μM, which is above their IC 50 values. Compound 19 a showed no detectable aggregation.
Further hit validation was done against the purified mPPase from P. falciparum (PfPPase-VP1) expressed in baculovirusinfected insect cells. Compound 17 a was able to inhibit the PfPPase-VP1 activity with an IC 50 of 58 μM ( Figure 4A). Compounds 19 a and 20 a had lower inhibition activities with IC 50 values of 130 μM and 74 μM, respectively ( Figure S5). Overall, these compounds have higher IC 50 values for PfPPase-VP1 than for TmPPase. In a survival assay in erythrocytes culture 17 a was able to inhibit the growth of P. falciparum with an IC 50 of 31 μM ( Figure 4B), better than its inhibition on the PfPPase-VP1. This could mean that the compound inhibits other proteins in the parasite, e. g. through soluble pyrophosphatase, or via some other mechanism, which could be linked to colloidal aggregation. Interestingly, compound 12 was able to inhibit the growth of P. falciparum with the IC 50 of 3.6 μM ( Figure S6) even though the activity on PfPPase-VP1 ( Figure S4) was comparably weak. No hemolysis of human erythrocytes was observed, suggesting no significant cytotoxicity. [33] Conclusion Altogether, this manuscript presents novel scaffolds with potential for further exploration in the drug discovery against parasitic diseases. Using a screening approach (52 compounds) together with a medicinal chemistry exploration (14 compounds) and the TmPPase test system, we discovered new TmPPase inhibitors: the 4,5-dihydropyrrolo[3,4-c]pyrazol-6(1H)one core (compound 12) and pyrazolo[1,5-a]pyrimidines (17 a, 19 a and 20 a). We explored the SARs around this latter core and maintained low micromolar activity (IC 50 = 14-18 μM) for three of the synthesized pyrazolo[1,5-a]pyrimidines. Molecular modelling suggests that the substrate binding sites are highly conserved in many protozoan pathogens, which should allow transferability of the findings. [11,34] Indeed, compound 17 a binds to the PfPPase-VP1 with an IC 50 of 58 μM and inhibits parasite growth.

Computational methods
Pharmacophore modelling was conducted using the Schrödinger Maestro suite. [35] Substructure searches were conducted from the ZINC12 database [31] (clean drug-like subset; 13,195,609 compounds; downloaded on 2018.11.11). Ligand efficiencies were computed using the pIC 50 and the "Heavy Atom Count" normalization method with Accelrys's Discovery Studio. [36] Chemistry General experimental methods: All chemicals were available from commercial vendors and used without any further purification. Anhydrous reactions were conducted in oven-dried (130°C, > 24 h) glassware that were purged with argon prior to use. Microwave reactions were done in sealed reaction vials using a Biotage ® Initiator + instrument (Uppsala, Sweden). The progress of the reactions was monitored using thin-layer chromatography on silica gel 60-F 254 aluminum plates and visualized by a dual short/long wave (254/366 nm) UV lamp. Combined organic solutions from extractions were dried over anhydrous Na 2 SO 4 , filtered and concentrated with a rotary evaporator at reduced pressure. Flash SiO 2 column chromatography was performed with automated high performance flash chromatography, Biotage ® Isolera™ Spektra Systems with ACI™ and Assist (ISO-1SW Isolera One) equipped with a variable UV-VIS (200-800 nm) photodiode array (Uppsala, Sweden) using SNAP KP-Sil/Ultra 10, 25, 50 or 100 g cartridges and the indicated mobile phase gradient. The reactions were not optimized and all the yields are given for purified products.
The synthesized products were characterized by NMR and MS analysis. 1   General procedure for synthesis of compound 14: The synthesis of compound 14 was adapted from a previously described method. [32] In brief, t-BuOK (2 equiv) was dissolved in anhydrous THF (1.2 mL/mmol) under argon, followed by addition of diethyl oxalate (2 equiv) and the mixture was stirred for 30 min. Then, ketone 13 (1 equiv) dissolved in anhydrous THF (2.2 mL/mmol) was added dropwise and stirred for 1 h. The reaction was quenched by addition of a 1 M solution of HCl in H 2 O (3.0 mL/mmol).

Biological activities
All commercially obtained compounds had purities > 90 % as specified by the vendor. Synthesized compounds were > 95 % pure as determined by LC-MS and characterized by HRMS. The inhibition assay was carried out with purified TmPPase as previously reported. [22][23][24] The best hits 17 a, 19 a, and 20 a, were validated using eight compound concentrations in triplicate. Further hit validation was done against the purified mPPase from P. falciparum (PfPPase-VP1) expressed in baculovirus-infected insect cells (expression, purification and further enzyme analysis will be described in more detail elsewhere). 4 μL of the purified PfPPase-VP1 (2.0 mg/ mL) was reactivated in 96 μL of reactivation buffer, i. e. 20 mM 2-(Nmorpholino)ethanesulfonic acid (MES) pH 6.5, 3.5 % (v/v) glycerol, 2 mM dithiothreitol (DTT) and 12 mg/mL l-α-phosphatidylcholine from soybean. For each tube strip, 3 μL of the reactivated PfPPase-VP1 was added into 12 μL of reaction buffer (200 mM Tris-Cl pH 8.0, 8.0 mM MgCl 2 , 333 mM KCl, 67 mM NaCl) and 25 μL of inhibitor solution. The reaction mixtures were incubated at 50°C for 5 min and the assay was started with the addition of 10 μL of 2 mM sodium pyrophosphate and further incubation at 50°C for 45 min. The reaction termination and colour development were done as earlier described. [22][23][24] Plasmodium falciparum parasite strain 3D7 was used for testing the antiplasmodial activity of the test compounds. The parasite was maintained in culture as described previously. [37] Briefly, parasites were cultured in O + human erythrocytes at 5 % haematocrit in RPMI-1640 medium supplemented with 0.5 % Albumax II (Gibco, Carlsbad, CA, USA), 200 μM hypoxanthine (Sigma, St. Louis, MO, USA) and 20 μg/mL gentamycin (Gibco). Parasites were synchronized with the sorbitol method as described. [38] Test compounds were dissolved in DMSO to make a stock solution of 10 mM or 50 mM. Two-fold serial dilutions of the test compounds were made in the culture medium to cover either the range of 100 μM-100 nM or 400 μM-390 nM. 50 μL of each dilution were mixed in 96-well plate with 150 μL of 2 % haematocrit of 0.5 % parasitemia (ring stage). Compound dilutions were also mixed with 150 μL of 2 % haematocrit of uninfected erythrocytes to serve as baselines for the activity of the coloured compounds. DMSO concentration in the highest compound concentration was 0.8 %. After 72 h incubation at 37°C, the parasite growth was quantified using fluorescent SYBR Green I®-based assay as described. [39] The half-maximal inhibitory concentration (IC 50 ) of the test compounds was assessed using the non-linear regression fit model in Prism 7 (GraphPad Software, San Diego, CA, USA). The potent anti-malarial drug, artemisinin, was used in parallel as a positive control. Each concentration of the test compounds was tested in duplicate and the assay was repeated three times. The mean and standard deviation of the three repeats were used to calculate the IC 50 .