Introduction

Extracellular nucleotides such as ATP, adenosine diphosphate (ADP), uridine triphosphate (UTP), and uridine diphosphate (UDP) act as extracellular signaling molecules in virtually all tissues and organs [1]. Especially ATP is involved in a variety of physiological and pathological functions in the body, e.g., as a neurotransmitter and neuromodulator in the central and peripheral nervous system, as a key messenger in nociception, as a tumor-inhibiting agent, in the control of secretion from a variety of endocrine glands, in the modulation of platelet aggregation by ADP, in chloride transport in the airway epithelia, in renal function, in bone and cartilage disease, and as a proinflammatory agent in the immune system [2, 3]. Extracellular nucleotides are efficiently inactivated by surface-located enzymes which sequentially hydrolyze the phosphate groups eventually resulting in the formation of the respective nucleoside and inorganic phosphate [4]. Nucleoside 5′-triphosphates and 5′-diphosphates may be hydrolyzed by members of the E-NTPDase (ectonucleoside triphosphate diphosphohydrolase) family, the E-NPP (ectonucleotide pyrophosphatase/phosphodiesterase) family, and by alkaline phosphatases. Nucleoside 5′-monophosphates are subject to hydrolysis by ecto-5′-nucleotidase and also by alkaline phosphatases [5]. Members of the various ectonucleotidase families reveal overlapping substrate specificity and tissue distribution whose functional significance needs to be further elucidated [4].

E-NTPDases comprise eight members, four of which have two plasma membrane-spanning domains with an active site facing the extracellular milieu [4–6], namely, NTPDase1 (CD39), NTPDase2 (CD39L1), NTPDase3 (CD39L3), and NTPDase8 [7, 8]. In contrast, NTPDase4, 5, 6, and 7 are intracellular membranes proteins. Even though NTPDases5 and 6 may be present at the surface of the plasma membrane and secreted as soluble enzymes following a proteolytic cleavage, their high Km values and low specific activities [9] make it unlikely that these enzymes regulate P2 receptor signaling. NTPDases share five highly conserved sequence domains (apyrase conserved regions) that are presumably of major relevance for their catalytic activity [10]. They contain the actin-HSP 70-hexokinase β- and γ-phosphate binding motif [11, 12].

Members of the E-NTPDase family can hydrolyze nucleoside 5′-triphosphates and nucleoside 5′-diphosphates albeit with varying preference for the individual type of nucleotide [10, 11]. NTPDase1 (CD39, ectoapyrase, ecto-ATP diphosphohydrolase) hydrolyzes ATP and ADP at a molecular ratio of about 1:0.5 to 1:0.9 [13]. In contrast, NTPDase2 has a strong preference for ATP with molecular ratios of ATP:ADP of 1:0.03 or less. NTPDase3 is a functional intermediate and reveals a molecular ratio of ATP:ADP of approximately 1:0.3. The activity of all three types of ectonucleotidases depends on millimolar concentrations of divalent cations such as Ca2+ or Mg2+ [11, 13, 14]. The enzymes hydrolyze not only ATP or ADP but have broad substrate specificity towards purine and pyrimidine nucleotides.

The analysis of nucleotide release requires the availability of inhibitors of ectonucleotidases that should ideally have no effect on P2 receptor activation [6]. It has been demonstrated that stable analogs of ATP can elicit tissue contractions up to 100 times more effectively than ATP. This suggests that the effects of exogenously applied ATP on P2 receptors are limited by its enzymatic degradation. Inhibitors of ectonucleotidases could thus serve as drugs that increase the lifetime of extracellular ATP (or ADP, UTP, or UDP) in situ. They could act in a site- and event-specific manner since they would only have an effect when nucleotides are present.

NTPDase1 is the major physiological E-NTPDase member showing a nearly ubiquitous tissue distribution. It is constitutively expressed in microvascular endothelium and certain immune cells and plays an important role in thromboregulation and immune function [15]. The distribution of NTPDase2 appears to be more restricted. NTPDase2 has been detected on blood vessels [16], cultured brain astrocytes [17], and on certain neuronal progenitor [18] and cancer cells [19, 20]. In contrast, NTPDase3 is expressed in the brain exclusively in neurons, with a predominantly axonal localization. Its colocalization with hypocretin-1/orexin-A indicates that it may be involved in the modulation of feeding, the sleep-wake cycle, or other behavioral components [21]. NTPDase3 has also been reported to be involved in auditory neurotransmission since it is highly expressed in the cochlea [22].

Potent and selective inhibitors for the individual NTPDase isoenzymes are required as pharmacological tools to investigate the (patho)physiological roles of NTPDases. Furthermore, such compounds are required to study their potential as novel drugs, e.g., for the treatment of cancer, as immunomodulatory agents, and for the treatment of cardiovascular or central nervous system disorders. So far only few classes of NTPDase inhibitors have been described [23] (Fig. 1).

Fig. 1
figure 1

Structures of NTPDase inhibitors

Full size image

These include nucleotide derivatives and analogs, mainly derived from ATP. Until recently, the only compound that has been shown to effectively inhibit the hydrolysis of ATP in a variety of tissues (albeit with moderate potency) without significantly acting on P2 receptors was the structural analog of ATP, ARL 67156 (1, FPL 67156) (N 6-diethyl-β,γ-dibromomethylene-ATP) [24–26]. However, ARL 67156 was recently found to be nearly inactive at rat, mouse, and human NTPDase2 [27, 28]. Although 1 is stable versus ecto-alkaline phosphatases and E-NTPDases, it might be hydrolyzed by E-NPPs, which directly attack the oxygen bridge between the α- and β-phosphorus atom of ATP and related compounds. However, Sévigny and colleagues showed that human NPP1 and NPP3 were unable to hydrolyze ARL 67156 [28]. 8-Thioether-ATP derivatives, found inactive at P2Y and P2X receptors, and stable to NTPDase hydrolysis, have also been described as NTPDase inhibitors. 8-Butylthioadenosine-5′-triphosphate (compound 2) acts as a competitive inhibitor of ATP with an estimated Ki value of 10 μM at NTPDase from bovine spleen [29]. Although the compound appears to be stable against NTPDases, it is likely to be hydrolyzed by the ubiquitously present ectoalkaline phosphatases which accept a very broad variety of substrates. Very recently, we have developed metabolically stable, uncharged nucleotide mimetics derived from uridine 5′-carboxylate and identified PSB-6426 (3) as a selective inhibitor of human NTPDase2 [30].

Besides inhibitors derived from nucleotides, non-nucleotide NTPDase inhibitors have been described. These include suramin (4) and related compounds [13, 31–33], sulfonate dyes such as reactive blue 2 (RB-2, 5) [34–37], and PPADS (6, pyridoxal phosphate-6-azophenyl-2′,4′-disulfonic acid) [13, 38, 39], which can all attenuate the hydrolysis of ATP. However, all of these compounds are nonselective, as they are also potent antagonists at P2 receptors. We recently identified a novel class of NTPDase inhibitors, the polyoxometalates (POMs), which are inorganic anionic metal complexes, e.g., POM-4 (7) [40]. These compounds have turned out to be useful pharmacological tools [41–43]. However, they have high molecular weights, bear several negative charges, and are not drug-like molecules.

The present study was aimed at identifying the pharmacophore of RB-2 for inhibiting NTPDases and at investigating the structure-activity relationships of truncated RB-2 derivatives. Since RB-2 shows some selectivity for rat NTPDase3 (Ki 1.10 μM, approximately 20-fold selectivity vs NTPDase1 and 2) [27, 44], we expected that this strategy might lead to the identification of an NTPDase3-selective inhibitor.

Results and discussion

Syntheses

The 1-amino-4-anilino-2-sulfoanthraquinone derivatives 11–19 were synthesized as recently described by combinatorial parallel synthesis using the Ullmann coupling reaction [45]. The new compounds 20–22 were obtained by two different methods. The 1-naphthyl (20) and the 3,4-dimethoxyphenyl derivative (22) were synthesized by the classic Ullmann reaction from bromaminic acid (8) and 1-naphthylamine, or 3,4-dimethoxyphenethylamine, respectively, in water in the presence of potassium carbonate and copper sulfate at 100–105°C (see Fig. 2). The 2-naphthyl derivative (21) was obtained by a new microwave-catalyzed Ullmann reaction in phosphate buffer using elemental copper as a catalyst [46] within only 5 min reaction time (Fig. 2). The 1-amino-2-methyl-4-arylaminoanthraquinone derivatives 23–26 were obtained by solvent-free reaction of 1-amino-4-bromo-2-methylanthraquinone (9) with the appropriate arylamine derivative in the presence of copper acetate and potassium acetate at 110°C for 5–24 h according to a procedure described by Harris et al. [47] (see Fig. 2).

Fig. 2
figure 2

Synthesis of aryl- and arylalkylaminoanthraquinone derivatives 20–22 and 23–26 (for R see Table 1)

Full size image

Anthraquinone sulfonic acid esters 27–30 and sulfonamides (31–33) were obtained by chlorination of bromaminic acid (8) with phosphorus pentachloride in phosphorus oxychloride to yield the sulfonic acid chloride 10 as a reactive intermediate, which was subsequently treated with the appropriate alcohol or phenol, or ar(alkyl)amine under basic conditions (Fig. 3). When p-ethoxy- or o-methoxy-substituted aniline was used for the condensation with 10, the desired simple sulfonamide derivatives were not obtained. Mass spectral and nuclear magnetic resonance (NMR) analysis revealed that bissulfonamides 34 and 35 had formed instead, although the anilines had been present in the reaction mixtures in excess. The explanation may be an increased nucleophilicity of the aniline nitrogen atom due to the positive mesomeric effects of the substituents (o-methoxy, p-ethoxy).

Fig. 3
figure 3

Synthesis of bromaminic acid esters 27–30, bromaminic acid amides 31–33, and bis(bromaminic acid) amides 34 and 35

Full size image

Solubility and stability of selected anthraquinone derivatives

It was observed that some anthraquinone sulfonic acid sodium salts, which were initially well soluble in water, formed precipitates during the testing. One explanation may be that the high salt concentrations in the buffer solutions, e.g., the presence of calcium ions, which may form hardly soluble calcium sulfonates, reduced their solubility. However, precipitation might also be due to another chemical reaction. Therefore, selected 1-amino-4-phenylamino-2-sulfoanthraquinone derivatives (12, 15, 17, 18) bearing different substituents on the phenyl ring (o-methyl, o, p-dimethyl, o-ethoxy, and p-chloro residues) were investigated for their stability under the test conditions. Aqueous solutions (10 mM) were prepared and diluted with the N-[2-hydroxyethyl]piperazine-N′-2-ethanesulfonic acid (HEPES) buffer that was used in the assays, to obtain a compound concentration of 3 mM. After a short period of time after dilution, compound 18 precipitated, while the other compounds stayed in solution. After 15 h at room temperature, compound 17 also started to precipitate, while 12 precipitated after 48 h and 15 after about 1 week. The precipitated compounds were collected by filtration, dried, and NMR spectra were recorded showing that the precipitated compounds were structurally identical to the original anthraquinone derivatives. This result indicates that the compounds were not degraded, but precipitated upon dilution with buffer solutions due to the lower solubility in the presence of high salt concentrations probably due to the formation of poorly soluble salts of the anthraquinone sulfonic acids with cations in the buffer (HEPES and/or calcium). Since precipitation in most cases only occurred after several hours, testing of high concentrations was always performed with freshly prepared solutions.

Investigation of anthraquinone derivatives as potential inhibitors of nucleoside triphosphate diphosphohydrolases by capillary electrophoresis

Inhibition of rat NTPDases1, 2, and 3 was performed essentially as previously described using a capillary electrophoresis (CE) method [27]. Membrane preparations derived from transfected cells containing rat NTPDase1, NTPDase2, or NTPDase3 were employed. A fixed substrate concentration of 400 μM was used. Test compounds were initially screened at a concentration of 1 mM. For compounds which showed about 50% inhibition or more, IC50 values were determined with six to eight different concentrations of inhibitor spanning about three orders of magnitude. After stopping the reaction by heating, an aliquot was subjected to CE analysis with UV detection at 210 nm. Under the applied conditions less than 10% of substrate was converted by the enzymes. Ki values were calculated from IC50 values as previously described assuming a competitive inhibition mechanism [27].

Structure-activity relationships

Even though the investigated 1-amino-2-sulfo-4-ar(alk)-ylaminoanthraquinone derivatives exhibited only a partial structure of RB-2 and were much smaller than the parent compound, many of them were similarly or even more active as NTPDase inhibitors (Table 1). For example, compound 18 bearing a p-chlorophenylamino residue was a potent, nonselective NTPDase inhibitor (Ki 16–18 μM). While NTPDase2 and 3 were quite tolerant with regard to the substituent in the 4-position accepting a large variety of differentially substituted arylamino residues and even a 3,4-dimethoxyphenethylamino residue, structure-activity relationships (SARs) of NTPDase1 were more restricted: phenylamino (11), m-methylphenylamino (13), and particularly o,p-dimethylphenylamino (15) and p-chlorophenylamino (18) were best tolerated, while monosubstitution in the ortho-position of the phenylamino residue (12, 16, 17) or replacement of the phenyl ring by 1-naphthyl (20) or a 3,4-dimethoxyphenethyl residue (22) were not well tolerated by rat NTPDase1. On the other hand, a 2-naphthyl ring was very well tolerated by NTPDase1: compound 21 was the most potent NTPDase1 inhibitor of the present series with a Ki value of 0.328 μM; it was less potent at NTPDase3 (Ki 2.22 μM) and much less potent at NTPDase2 (19.1 μM). Small p-substituents on the phenyl ring (p-chloro, p-methyl) were well accepted by NTPDase1, while an acetylamino residue in the p-position was detrimental. Compound 20 was the most potent and selective inhibitor of rat NTPDase3 and not inhibitory to rat NTPDase1 and 2 even at high, millimolar concentrations. Rat NTPDase3 appears to have a lipophilic pocket which can accommodate a 1-naphthylamino residue in the 4-position. Concentration-inhibition curves of the most potent nonselective NTPDase inhibitor 18 (PSB-069) at the three investigated NTPDases and of the NTPDase3-selective inhibitor 20 (PSB-06126) are shown in Fig. 4.

Fig. 4
figure 4

Concentration-inhibition curves of selected anthraquinone derivatives. Top: curves of the nonselective NTPDase inhibitor 18 at NTPDase1, 2, and 3. Bottom: curve of the NTPDase3 inhibitor 20

Full size image
Table 1 Inhibitory potencies of anthraquinone derivatives at rat NTPDase1, 2, and 3 determined by capillary electrophoresisa
Full size table

An unsubstituted sulfonate group in the 2-position of the anthraquinone scaffold was essential for inhibitory activity. Replacement of the sulfonate by a methyl group (23–26) abolished activity. Bromaminic acid derivatives, in which the sulfonic acid group was converted to sulfonic acid esters (27–30) or sulfonamides (31–33), were also totally inactive. Thus, a negative charge (a free sulfonate group) appeared to be required for NTPDase inhibition. This is in agreement with previous studies on sulfonate dyes and related compounds, which showed that sulfonate groups in certain positions of the molecules were important for ectonucleotidase inhibitory activity [23, 34].

In order to investigate the inhibition mechanism of this class of NTPDase inhibitors, kinetics of NTPDase3 were exemplarily determined in the absence and in the presence of compound 18 (10, 20, and 30 μM). While the Vmax value was not affected by the inhibitor, the Km value increased with increasing concentration of inhibitor indicating a competitive mechanism of inhibition (for Lineweaver-Burk plot see Fig. 5).

Fig. 5
figure 5

Lineweaver-Burk plot for rat NTPDase3 kinetics in the absence and presence of different concentrations of inhibitor 18

Full size image

The present study is the first one to systematically investigate the SARs of anthraquinone derivatives derived from RB-2 at defined NTPDase isoenzymes. We could confirm previous findings by Tuluc et al. [37] that anthraquinone derivatives with only one substituted phenyl ring at the 4-amino group, which are much smaller than RB-2, can exhibit ectonucleotidase inhibitory activity. However, compound 11 (also known as acid blue 25) was inactive (at 100 μM) in their study investigating ATP degradation in rat vas deferens tissue [37]. Like RB-2, its truncated derivatives also interact with P2 receptors [37, 45, 48]. However, the SARs are clearly different. For example, 16 (PSB-716) was shown to be a potent P2Y2 antagonist (IC50 ca. 9 μM) [45], but it is only a weak NTPDase1 inhibitor. Compound 19 was also a potent inhibitor at P2Y2 (IC50 6–12 μM) [45], but nearly inactive at all three NTPDases tested.

Conclusions

We have investigated 25 anthraquinone derivatives related to RB-2 for their potency at inhibiting rat NTPDases1, 2, and 3 in order to obtain initial information about their structure-activity relationships. While 11 of the compounds had previously been reported, 14 compounds are new and their syntheses and physicochemical characterization are described for the first time. Certain 1-amino-2-sulfo-4-arylaminoanthraquinones were identified as potent NTPDase inhibitors. While the 4-chlorophenylamino derivative 18 was a nonselective inhibitor (Ki 16–18 μM), the 4-(1-naphthylamino) derivative 20 was very potent and selective for rat NTPDase3 versus rat NTPDase1 and 2. In contrast, its 2-naphthylamino isomer 21 was a very potent NTPDase1 inhibitor (Ki 0.328 μM). For future studies it would be of interest to investigate potential species differences of these and other NTPDase inhibitors. Furthermore, it will be important to investigate the potency of some of these compounds at other ectonucleotidases, including ecto-5′-nucleotidase, nucleotide pyrophosphatases/phosphodiesterases (NPPs), and alkaline phosphatases, and to determine their profile at the various P2Y and P2X receptor subtypes, in order to be able to fully assess their usefulness as pharmacological tools. The compounds described will be good starting points for further optimization.