Development of Anthraquinone Derivatives as Ectonucleoside Triphosphate Diphosphohydrolase (NTPDase) Inhibitors With Selectivity for NTPDase2 and NTPDase3

Ectonucleoside triphosphate diphosphohydrolases (NTPDases) catalyze the hydrolysis of nucleoside tri- and di-phosphates to mono-phosphates. The products are subsequently hydrolyzed by ecto-5′-nucleotidase (ecto-5′-NT) to nucleosides. NTPDase inhibitors have potential as novel drugs, e.g., for the treatment of inflammation, neurodegenerative diseases, and cancer. In this context, a series of anthraquinone derivatives structurally related to the anthraquinone dye reactive blue-2 (RB-2) was synthesized and evaluated as inhibitors of human NTPDases utilizing a malachite green assay. We identified several potent and selective inhibitors of human NTPDase2 and -3. Among the most potent NTPDase2 inhibitors were 1-amino-4-(9-phenanthrylamino)-9,10-dioxo-9,10-dihydroanthracene-2-sulfonate (20, PSB-16131, IC50 of 539 nM) and 1-amino-4-(3-chloro-4-phenylsulfanyl)phenylamino-9,10-dioxo-9,10-dihydroanthracene-2-sulfonate (48, PSB-2020, IC50 of 551 nM). The most potent NTPDase3 inhibitors were 1-amino-4-[3-(4,6-dichlorotriazin-2-ylamino)-4-sulfophenylamino]-9,10-dioxo-9,10-dihydroanthracene-2-sulfonate (42, PSB-1011, IC50 of 390 nM) and 1-amino-4-(3-carboxy-4-hydroxyphenylamino)-9,10-dioxo-9,10-dihydroanthracene-2-sulfonate (33, PSB-2046, IC50 of 723 nM). The best NTPDase2 inhibitor 20 showed a non-competitive inhibition type, while the NTPDase3 inhibitor 42 behaved as a mixed-type inhibitor. These potent compounds were found to be selective vs. other NTPDases. They will be useful tools for studying the roles of NTPDase2 and -3 in physiology and under pathological conditions.


Material and Methods
All materials were used as purchased (Acros, Alfa Aesar, Merck, or Sigma-Aldrich, Germany). Thin-layer chromatography was performed using TLC aluminum sheets silica gel 60 F 254 or TLC aluminum sheets reversed phase (RP) silica gel 18 F 254 (Merck, Darmstadt, Germany). Colored compounds were visible at daylight; other compounds were visualized under UV light (254 nm). Flash chromatography was performed on a Büchi system using silica gel RP-18 (Merck, Darmstadt, Germany). 1 H and 13 C NMR data were collected on either a Bruker Avance 500 MHz NMR spectrometer at 500 MHz ( 1 H) or 126 MHz ( 13 C), respectively or a 600 MHz NMR spectrometer at 600 MHz ( 1 H) or 151 MHz ( 13 C), respectively. Deuterated dimethyl sulfoxide (DMSO-d 6 ) or chloroform-d (CDCl 3 ) were used as a solvent. Chemical shifts are reported in parts per million (ppm) relative to the deuterated solvent, i.e., DMSO, d 1 H 2.49 ppm; 13 C 39.7 ppm, coupling constants J are given in Hertz, and spin multiplicities are given as s (singlet), d (doublet), t (triplet), q (quartet), sext (sextet), m (multiplet), and br (broad).
The purities of isolated products were determined by high performance liquid chromatography (HPLC) coupled with electrospray ionization mass spectrometry (ESI-MS) and ultraviolet (UV) detector using the following procedure: the compounds were dissolved at a concentration of 0.5 mg/mL in H 2 O/MeOH = 1:1, containing 2 mM NH 4 CH 3 COO. Then, 10 mL of the sample was injected into an HPLC column (Phenomenex Luna 3 m C18, 50 mm × 2.00 mm). Elution was performed with a gradient of water:methanol (containing 2 mM NH 4 CH 3 COO) from 90:10 to 0:100 starting the gradient immediately at a flow rate of 250 mL/min for 15 min, followed by washing with 100% methanol for another 15 min. The purity of the compounds proved to be ≥95%. For microwave reactions, a CEM Focused Microwave Synthesis Type Discover apparatus was employed. A freeze-dryer (CHRIST ALPHA 1-4 LSC) was used for lyophilization.
General Procedure A: Preparation of 4-Substituted 1-Aminoanthraquinone-2-sulfonate Derivatives (11-51) To a 5 mL microwave reaction vial, equipped with a magnetic stirring bar, were added 1-amino-4-bromo substituted anthraquinone compounds [bromaminic acid sodium salt (10a) or 1-amino-2,4-dibromoanthraquinone (10b)] (0.1−0.3 mmol) and the appropriate aniline or amine derivative (1.5−9.0 equiv), followed by a buffer solution of Na 2 HPO 4 (pH 9.6) (5.0 mL) and NaH 2 PO 4 (pH 4.2) (1.0 mL) and a finely powdered elemental copper (0.002−0.003 g, 5−10 mol%). The mixture was capped and irradiated in the microwave oven (80−100 W) for 5−24 min at 100−120°C. The reaction mixture was cooled down to room temperature (rt), and the product was purified using the following procedure. The contents of the vial were filtered to remove the elemental copper. Then, ca. 200 mL of water was added to the filtrate, and the aqueous solution was extracted with dichloromethane (200 mL). The extraction procedure was repeated until the dichloromethane layer became colorless (two to three times). The aqueous layer was reduced by rotary evaporation to a volume of 10−20 mL, which was subsequently submitted to flash column chromatography using RP-18 silica gel and water as an eluent. The polarity of the eluent was then gradually decreased by the addition of acetone in the following steps: 5, 10, 20, 40, and 60%. Fractions containing blue product were collected. For some compounds the last step of purification (RP-18 flash chromatography) had to be repeated two to three times to obtain pure product (≥95% purity as determined by HPLC-UV-MS). The pooled product-containing fractions were evaporated under vacuum to remove the acetone and reduce the water volume. The remaining water was subsequently removed by lyophilization to yield up to 80% of the product as blue powder (Scheme 1 and Table 1).

General Procedure B: Preparation of 2-Substituted 1-Amino-4-anilinoanthraquinone Derivatives (52-56)
A round bottom flask (25 mL) equipped with a magnetic stirring bar was charged with one equivalent of starting material (10b or 1-amino-4-bromo-2-methylanthraquinone (10c)), an excess of appropriate aniline derivative (15 equiv.) and copper(I) acetate (10 mol%) in the presence of 2.25 equiv. of potassium acetate (Scheme 1). The resulting mixture was heated at 110°C under an argon atmosphere for 2−15 h, and the progress of the reaction was monitored by TLC using 10% dichloromethane/cyclohexane as eluent. The reaction mixture was then let to cool down to room temperature, followed by the addition of ethanol (5 mL), and the blue-colored precipitate was filtered off and washed successively with ethanol, 0.1 M HCl, and water (ca. 15 mL each), and then the solid material was dried at 70°C in the oven for 16 h. The product was then purified by silica gel column chromatography using dichloromethane/cyclohexane (9:1) as eluent. The desired products (52−56) were obtained in high yields (Scheme 1 and Table 1).

General Procedure C: Preparation of 4-Substituted
Anthraquinone-2-sulfonate Derivatives (57 and 58) To a 50 mL round bottom flask equipped with a magnetic stirring bar, 0.1 mmol of 1-aminoanthraquinone derivative (21 or 33) was added, followed by 5 mL of 1 M hydrochloric acid. The solution was cooled to 0−5°C in an ice bath, and a previously cooled solution of NaNO 2 (13.8 mg, 0.2 mmol, 2 equiv) in 0.5 mL of distilled water was added dropwise. After 5 min, the mixture was allowed to warm up to rt, followed by addition of 30 mg of zinc powder (1.0 mmol, 10 equiv) and 5 mL of ethanol. The resulting mixture was then allowed to stir at rt for ca. 30 s. The mixture was filtered off, and the purple-colored filtrate was then purified by flash column chromatography on a reversed phase silica gel (RP-18) using a gradient of acetone in water (5 and 20%) as the eluent. Fractions containing the purple product were collected and evaporated in vacuum to remove acetone and decrease the volume of water to ca. 10-20 mL. Complete drying was achieved with a freeze-dryer, affording purplecolored products in excellent yields (Scheme 2 and Table 1).

Malachite Green Assay to Investigate NTPDase Inhibitors
Membrane preparations expressing human NTPDase1, -2, -3, or -8, respectively, were obtained as previously described (Sevigny et al., 1997;Cogan et al., 1999;Kukulski et al., 2005;Lecka et al., 2013;Lee et al., 2018). Enzyme inhibition assays were performed using the malachite green assay in analogy to published procedures with some modifications (Dou et al., 2018). The reaction buffer contained 10 mM HEPES, 2 mM CaCl 2 , and 1 mM MgCl 2 (pH 7.4) in a final volume of 50 mL in transparent 96well half-area plates. The compounds were initially tested at a final concentration of 2 µM using a COS-7-cell membrane preparation expressing the appropriate NTPDase isoenzyme (protein amount: 143 ng for NTPDase1, 175 ng for NTPDase2, 152 ng for NTPDase3, and 175 ng for NTPDase8). Preincubated of the enzyme preparations was perfomred at 37°C in the presence or absence of test compounds with gentle shaking (Eppendorf Thermomixer comfort at 500 rpm) for 5 min. The reaction was initiated by the addition of 50 µM ATP [K m (CD39) = 17 µM] for NTPDase1 or 100 µM ATP for NTPDase2, -3, and -8 [K m (NTPDase2) = 70 µM; K m (NTPDase3) = 75 µM; K m (NTPDase8) = 46 µM] (Kukulski et al., 2005). After 15 min of incubation at 37°C with gentle shaking, the reaction was stopped by the addition of the detection reagents (20 µL malachite green solution, 0.6 mM, and 30 µL of ammonium molybdate solution, 20 mM, in 1.5 M sulfuric acid). The released inorganic phosphate was quantified after 20 min of gentle shaking at 25°C by measuring the absorption of the malachite green-phosphomolybdate complex at 600 nm using a BMG PheraStar FS plate reader (BMG Labtech GmbH, Ortenberg, Germany). The corrected absorption was calculated by subtracting the absorption of the negative control samples, which were incubated with previously denatured enzyme (90°C, 15 min). Full concentration-inhibition curves were determined with inhibitor concentrations ranging from 0.03 to 30 µM in the presence of 2% DMSO. Inhibition-type experiments were performed with 25, 50, 100, 150 and 200 µM ATP as substrate for NTPDase2 in the presence of inhibitor 20 (0, 0.25, 0.5, and 1 µM) and 25, 50, 100 and 150 µM ATP substrate for NTPDase3 and compound 42 (0.25, 0.5, and 1 µM). For all of the presented data, at least three independent experiments were performed, and IC 50 values were calculated by GraphPad Prism 8 software.

RESULTS AND DISCUSSION
A library of 48 anthraquinone derivatives was synthesized and tested at human NTPDase1, -2, -3, and -8, which are ectoenzymes hydrolyzing extracellular nucleotides, using the malachite green assay. Subsequently, inhibition curves for compounds showing above 50% inhibition at 2 µM test concentration were determined.
Compounds 52-56, bearing a bromo or methyl residue at the 2-position, were synthesized starting from 10b or 10c, respectively, with excess of the appropriate aniline derivatives (15 eq.) under argon in the presence of potassium acetate and copper(I) acetate (CuOAc) as a catalyst, upon heating at 110°C for 2-15 h (Scheme 1).
In order to investigate the role of the amino group at the 1position of the anthraquinone moiety, two anilinoanthraquinone derivatives (21 and 33) were treated with sodium nitrite in hydrochloric acid solution (1 M) at 0-5°C for 5 min, then allowed to warm up to room temperature, followed by the addition of ethanol and an excess of zinc powder (10 equiv.) to achieve deamination within 30 seconds (Baqi and Müller, 2012), affording the desired products 57 and 58 in excellent yields (Scheme 2).

Biological Studies
Inhibition of human NTPDases was performed using the malachite green assay, which was established on a robotic system (Z' factors > 0.70) (Baykov et al., 1988;Fiene et al., 2015). The malachite green assay enables the detection of the phosphate produced by the enzymatic hydrolysis of nucleotides. A fixed substrate concentration of 50 µM ATP for NTPDase1 and 100 µM for NTPDase2, -3, and -8 was employed. Test compounds were initially screened at a concentration of 2 mM.
For compounds that showed about 50% inhibition or more, concentration-dependent inhibition curves were determined, and IC 50 values were calculated. A total of 48 synthesized anthraquinone derivatives including 14 new compounds not previously described in the literature were evaluated for their inhibitory activity at human NTPDase1, -2, -3, and -8 (for results see Table 1).

Structure-Activity Relationships (SARs)
The anthraquinone derivative reactive blue-2 (RB-2 (5), Figure 1 and Table 1) showed the highest potency at NTPDase3 (IC 50 of 0.942 µM) followed by NTPDase2 and was inactive at NTPDase8 ( Table 1). RB-2 is a relatively large molecule (molecular weight of >800 g/mol) with high polarity bearing three negatively charged sulfonate (SO 3 Na) groups. Therefore, smaller and less polar anthraquinone derivatives were designed, synthesized, and evaluated as NTPDases inhibitors (see Table S1 in Supplementary Materials for clogD values of all anthraquinone derivatives discussed in the present study).
In our previous study, we had investigated a smaller series of anthraquinone derivatives at ecto-NTPDases of rat, which had led to the identification of PSB-071 (6) bearing a m-methyl substituent on the 4-anilino group. This inhibitor was slightly selective for rat NTPDase2 (12.8 mM) (Baqi et al., 2009b;Zebisch et al., 2014) vs. rat NTPDase1 and -3, while in the present study, it showed no significant inhibitory activity on all tested human NTPDases (compound 6, Table 1), except for NTPDase2, at which it displayed very moderate potency.
Interestingly, a combination between structures of 1-naphthyl and 2-naphthyl resulting in phenanthryl derivative 20, yielded a potent inhibitor of NTPDase2 which displayed no activity vs. NTPDase1, -3, and -8 at the tested concentration. This is probably due to the presence of a large lipophilic pocket present in human NTPDase2. This presence of a lipophilic pocket in NTPDase2 was confirmed with compound 16 (IC 50 of 5.62 µM, Table 1), which is bearing an extra lipophilic methyl group in the 2position of the 1-naphthyl moiety; again, this compound was found to be selective vs. the other investigated human NTPDases (-1, -3, and -8). Introduction of polar and negatively charged groups, SO 3 H (17) or CO 2 H (19) on the naphthyl moiety shifted the inhibitory activity towards human NTPDase3.
In the next step, we introduced different substituents on phenyl ring D (compounds 21-38, Table 1). Mono-substitution of the aromatic ring D with Br (22), NO 2 (24), CO 2 H (26), or CH 2 CO 2 H (30) in the meta-position led to selective inhibition of NTPDase3, while other mono-substitutions including m-F, p-NH 2 , o-CH 2 OH, m-CH 2 OH, and p-CH 2 OH resulted in no inhibition at all tested NTPDases. On the other hand, di-substitution with polar functions, e.g., NH 2 , SO 3 H, and OH, on the meta-and para-position of the phenyl ring restored the inhibitory potency towards NTPDase2, especially compound 33 showing inhibitory potency at submicromolar concentration. Any polar substituent in the ortho-position and in combination with a substituent in the meta-or para-position led to inactive derivatives. The introduction of lipophilic substituents in the ortho-and paraposition shifted the inhibitory potency towards NTPDase2, see compound 38 (Table 1).
Next, we introduced an additional aromatic residue, ring E. Lipophilic substitution in the meta-and para-position resulted in moderate to good potency at NTPDase2 (39-41 and 43-51, Table 1), with potencies reaching the submicromolar range (IC 50 of 0.551 mM, 48), while a m-dichlorotriazinyl moiety in combination with a p-SO 3 H group furnished the most potent compound of the present anthraquinone series at NTPDase3 (42, IC 50 of 0.390 mM, Table 1).
Any modification on the anthraquinone moiety, such as removal of the amino group in position 1 or replacement of the sulfonate function in position 2 of the anthraquinone core by bromo or methyl abolished the inhibitory activity (see compounds 52-58, Table 1).
The SARs for human NTPDase2 and -3 are summarized in Figure 4. Large and lipophilic substituents have led to selectivity for NTPDase2 ( Figure 4A), while smaller and polar substituent have provided selectivity for NTPDase3 ( Figure 4B).
We previously published articles highlighting the fact that the anthraquinone scaffold represents a privileged scaffold in  Table 1. medicinal chemistry targeting different nucleotide-binding proteins including ecto-5'-nucleotidase, P2Y 12 and P2X2 receptors (Baqi, 2016;. However, this does not mean that potent compounds are non-selective. In fact, selectivity for specific targets has been achievable (Baqi et al., 2009a;Baqi, 2016;Rafehi et al., 2017a;Rafehi et al., 2017b). Compounds that are highly potent at a specific target typically also have shown selectivity. In a study published in 2010, we reported the first SARs of anthraquinone derivatives as inhibitors of rat ecto-5'-nucleotidase (CD73) . The observed SARs were clearly different from the SARs of anthraquinone derivatives as NTPDase inhibitors. For example, compound 15 displayed an IC 50 of 0.53 µM at rat CD73 but was virtually inactive at NTPDases, while compound 20, found to be a potent inhibitor of human NTPDase2 in the present study, was shown to be only weakly active against CD73 (58% inhibition at 1 mM concentration) . The compounds have not yet been tested at alkaline phosphatase, but this enzyme has a very high K m value for adenine nucleotides, and its significance in the context of extracellular nucleotide metabolism and signaling in inflammation is therefore questionable. Nevertheless, ancillary activities of NTPDase inhibitors as blockers of CD73 or alkaline phosphatase would not be detrimental, but might even enhance their over-all effects leading to an accumulation of immunostimulatory, proinflammatory nucleotides while inhibiting the final production of immunosuppressive adenosine. Future studies might therefore be directed at multi-target drugs inhibiting more than one single ectonucleotidase.

Mechanism of Enzyme Inhibition
In previous studies at rat NTPDase2 and -3, selected small 1amino-4-anilino-2-sulfoanthraquinone derivatives were found to display a competitive inhibition mechanism (Baqi et al., 2009b;Zebisch et al., 2014). In the present study at human NTPDases, the most potent inhibitors at NTPDase2, compound 20, and at NTPDase3, compound 42, were investigated with regard to their inhibition mechanism (see Figure 5). NTPDase2 inhibitor 20 displayed non-competitive inhibition, while the larger NTPDase3 inhibitor 42 showed a mixed inhibition type. Together with previous results (Baqi et al., 2009b;Zebisch et al., 2014), these data show that anthraquinone derivatives may inhibit NTPDase isoenzymes with different inhibition mechanisms depending on the compound's substitution pattern and perhaps also the NTPDase subtype and the species.

CONCLUSIONS
Ectonucleoside triphosphate diphosphohydrolase (E-NTPDase) plays a major role in controlling extracellular nucleotide levels. NTPDase inhibitors have potential as novel drugs, for example, for the treatment of inflammation, neurodegenerative diseases and cancer. In the present study, we synthesized and investigated a series of 48 anthraquinone derivatives as potential inhibitors of NTPDases, 14 of which are novel compounds. The synthesized compounds showed no inhibitory activity on NTPDase1 (CD39) or NTPDase8, while potent inhibitors for NTPDase2 or -3 were identified. The most potent inhibitors exhibited selectivity for either NTPDase2 or -3. It was noticed that human NTPDase2 features a lipophilic pocket that accommodates polynuclear-aromatic rings such as phenanthryl or naphthyl bearing lipophilic substituents such as chloro or methyl. In contrast, NTPDase3 was found to accommodate smaller hydrophilic functions such as hydroxyl, carboxyl or sulfonate. These NTPDase3-inhibitors were selective (>10-fold) vs. other NTPDases. Although inhibitors bearing polar sulfonate functions cannot be expected to be brain-penetrant, they will be useful tools for studying peripheral effects, or maybe even used to study central effects after direct application to the brain.