4-Amino-substituted Benzenesulfonamides as Inhibitors of Human Carbonic Anhydrases

A series of N-aryl-β-alanine derivatives and diazobenzenesulfonamides containing aliphatic rings were designed, synthesized, and their binding to carbonic anhydrases (CA) I, II, VI, VII, XII, and XIII was studied by the fluorescent thermal shift assay and isothermal titration calorimetry. The results showed that 4-substituted diazobenzenesulfonamides were more potent CA binders than N-aryl-β-alanine derivatives. Most of the N-aryl-β-alanine derivatives showed better affinity for CA II while diazobenzenesulfonamides possessed nanomolar affinities towards CA I isozyme. X-ray crystallographic structures showed the modes of binding of both compound groups.


Introduction
Since the discovery of the antibacterial activity of protonsil in 1932 [1], a wide array of aryl-and heteroarylsulfonamides exhibiting a variety of biological activities including antibacterial, antimicrobial, antifungal, anti-HIV, antihelmintic, anti-inflammatory, phytotoxic, cytotoxic, and radiosensitizing properties has been synthesized [2][3][4][5][6][7]. Most of such compounds were secondary sulfonamides. The primary sulfonamides, however, are well known as carbonic anhydrase (CA) inhibitors that are clinically used as antiglaucoma agents, antiephileptic and antiobesity drugs, as well as diuretics [8][9][10][11]. There are 15 CA isoforms, out of which 12 are catalytically active in the human body and several of them are highly over-expressed in various diseases [12][13][14]. Currently used drugs inhibit most of these isoforms with insufficient selectivity, thus causing numerous side effects. The design of isoform-selective CA inhibitors could lead to drugs with fewer side effects compared to the currently used ones [13].
Benzenesulfonamides are the most extensively investigated class of CA inhibitors. Several excellent reviews on CA inhibitors [15][16][17] have discussed the main structural features of such compounds revealing that substituents at the para-position of benzenesulfonamides (halogens, acetamido, alkoxycarbonyl moieties, carboxy-, hydrazido-, ureido-, thioureido-, as well as methylamine moieties) are related with good CA-inhibitory properties. To date many CA II-benzenesulfonamide inhibitor crystal structures show similarities in their interaction modes, with the sulfonamide group involved in the coordination of the Zn 2+ catalytic ion while the phenyl ring establishes several van der Waals interactions with the residues Gln92, Val121, Phe131, Leu198, and Thr200. The substituents on the benzenesulfonamide ring form additional contacts with the hydrophobic and/or the hydrophilic region of the active site, thus contributing favorably to the binding affinity and selectivity.
Novel design of drugs containing benzenesulfonamide scaffolds is often based on the "tail" strategy [17,18], which consists of attaching moieties that provide the desired physico-chemical properties to the sulfonamides possessing free amino groups. The orientation of the tail moiety, the nature of the spacers between the head and the tail moieties, and the presence of polar groups in the tail, strongly influence the potency and selectivity of benzenesulfonamides for the inhibition of CA isoforms [19].
Many benzenesulfonamides were synthesized using sulfanilamide as a starting compound leading to N-4-mono-substituted or 4-N,N-disubstituted derivatives (such compounds were reviewed in [20]). Here we report the synthesis of new N-aryl-β-alanine derivatives containing a primary sulfonamide moiety. We investigated these compounds as inhibitors of six carbonic anhydrase isoforms: CA I, II, VI, VII, XII, and XIII.
Triazenes have been extensively studied as agents displaying antitumor activity [21,22]. Several 3,3-dialkyltriazenes bearing sulfonamide groups on the benzene ring were synthesized by Afghani [23] and tested in vivo for antitumor activity against mouse lymphoid leukemia, showing no activity.
However these compounds were never tested as CAs inhibitors. We synthesized a series of 4-substituted diazobenzenesulfonamides with cyclic amines (pyrrolidine, piperidine, morpholine, and hexamethyleneimine) of various ring sizes as well as with aliphatic amines (dimethylamine and diethylamine) as described by Dabbagh [24] and measured their binding affinity for CA I, II, VI, VII, XII, and XIII, concluding that 4-substituted diazobenzenesulfonamides exhibited significantly higher binding affinity toward tested CAs than N-aryl-β-alanine derivatives.

Binding Studies
The synthesized compounds could be arranged into two groups. The first group of compounds is the N-aryl-β-alanine derivatives containing the primary sulfonamide group 3-25 (except in the compound 4 which is a tertiary sulfonamide) and the second group of compounds-diazobenzenesulfonamides 26-33. Compounds such as benzenesulfonamide (BSA) and acetazolamide (AZM) were used as controls. The binding affinity of these compounds to six CA isoforms I, II, VI, VII, XII, and XIII was determined by the fluorescent thermal shift assay (FTSA) and confirmed for two key compounds 18 and 31 by the isothermal titration calorimetry (ITC). The dissociation constants (Kd) of these compounds for the six CA isoforms are listed in Table 1.
Most N-aryl-β-alanine derivatives 3-25 were weak inhibitors of all tested CAs exhibiting the Kd higher than 1 µM. Only three compounds 5, 6, and 18 exhibited the Kd for CA II of 0.83, 0.67, and 0.67 µM, respectively. In most cases, the Kd for any CA isoform were in the comparable ranges as for BSA and sulfanilamide (SA). For example, only two compounds 6 and 18 bound CA I with more than four times stronger Kd (1.11 and 1.67 µM, respectively) than BSA (7.14 µM). For other N-aryl-β-alanine derivatives, the Kd for CA I were in the range of 5.88-170 µM. Only compound 10 bound CA I 1.7 times more weakly than the SA (Kd were 170 and 100 µM, respectively). Our binding data for SA are quite different from the data reported in [33,34]. The previously reported inhibition constant Ki for CA I was reported to be 25 µM, for CA II-0.24 µM, for CA VI-0.9 µM, for CA VII-0.07 µM, for CA XII-0.037 µM, and for CA XIII-0.035 µM. However, Murakami and Sly [35] showed that SA was less effective inhibitor of CA I, II, and VI (Ki was 50, 2.0, and 9.9 µM, respectively). Their inhibition constants were more comparable with our Kd values for CA I, II, and VI and differed 2 times for CA I, 6.5 times for CA II and 5.7 times for CA VI. These discrepancies between inhibition and binding data could be attributed to the different conditions used in assays (i.e., temperature and pH). ND-not determined due to the limited solubility of the compound; a-entire compound structure, not the R group; b-compound synthesis and binding data described in [36]; c-data taken from [37].
Most N-aryl-β-alanine derivatives bound to CA II with higher affinities than to other tested CAs. Interestingly, compound 4, bearing the tertiary sulfonamide group, bound only to CA II (Kd = 40 µM). Compound 20 showed better binding potency to CA XII (Kd 1.85 µM) than for other isoforms.
The diazobenzenesulfonamides 26-32 were significantly more potent CA inhibitors than N-aryl-β-alanine derivatives, especially for CA I and XIII. Figure 1  with the FTSA data and confirmed higher binding affinity of diazobenzenesulfonamides rather than the N-aryl-β-alanine derivatives ( Figure 2). The increase in binding affinity for CA I and CA XIII was evident with increasing the tail hydrophobicity from dimethylamine group in compound 26 to hexamethyleneimine in 31 (16.7 and 18 times stronger in Kd for CA I and XIII, respectively). The Kd for CA II, VII, and XII did not significantly depend on the hydrophobicity of substituent, whereas such hydrophobic substitutions did not entirely affect the binding affinity to CA VI (Kd were in the range of 10.0-14.3 µM for compounds 26-31). Compounds 27, 30, and 31 demonstrated some selectivity for CA I with respect to other isoforms (affinities between at least two isoforms differed at least four-fold). For example, 30 bound to CA I 9.4 times stronger than to CA II, 1500 times stronger than to CA VI, 18 times stronger than to CA VII, 115 times stronger than to CA XII, and 10 times stronger than to CA XIII.
The influence of the diazo group on the binding affinity could be seen comparing the binding data for compounds 32 (bearing the diazo group) and 34 (without the diazo group). As seen in Table 1, the binding affinities are quite similar with the exception of CA I where the binding was about six-fold  Similar comparison concludes that the presence of amino group diminished the affinity of a compound for CAs. For example, SA bound 5-10 times more weakly than the BSA. Similarly, compound 33 bearing the amino group bound 2-12 times more weakly than compound 32 bearing the hydroxy group.

Crystallography
The crystal structures of compounds 31 and 18 bound in the active site of recombinant CA II were solved by X-ray crystallography. Both ligands are well defined in the crystal structures. The difference electron density maps of the compounds calculated from the models omitting the ligand are shown in Figure 3a,b. Benzene rings of 18 and 31 in CA II coincide well with the position of benzenesulfonamide (PDB ID 2WEJ) [39] as shown in Figure 3c. Benzene rings of both ligands make van der Waals contacts with Val121, Leu198, and Thr200. The para-substituents of both compounds are fixed between Pro202 and Phe131. Similar orientation was observed for para-substituted [(2-pyrimidinylthio)acetyl]benzenesulfonamides described in [36]. Thus, para-groups are located in the CA II active site in mostly hydrophobic environment and contact the residues Phe131, Val135, Leu198, Pro202, and Leu204. The azepane group in para-position of 31 is more hydrophobic than of 18 and this probably improves the binding of 31 to CA II as compared to 18 ( Table 1). The crystallographic data collection statistics and the PDB IDs are listed in Table 2.

General Information
The melting points were determined on a MEL-TEMP (Electrothermal, A Bibby Scientific Company, Burlington, NJ, USA) melting point apparatus and are uncorrected. The 1 H and 13

Chemistry
Diethyl 4-Aminophenylsulfonylcarbonimidodithioate (2): 4-Aminobenzene-1-sulfonamide (1, 17.2 g, 0.1 mol) was dissolved in DMSO (50 mL), the solution was cooled down to 0 °C and 30% aqueous NaOH solution (10 mL) was added. The mixture was cooled to 0 °C and stirred at this temperature for 20 min. Afterwards, CS2 (6 mL, 0.1 mol) was added dropwise and the stirring was continued for 20 min at 0 °C. 30% aqueous NaOH solution (10 mL) was added again, the mixture was stirred for 20 min and CS2 (6 mL, 0.1 mol) was added dropwise again. Temperature of the reaction mixture was raised to room temperature and it was stirred for 1 h. Ethyl iodide (16 mL, 0.2 mol) was added dropwise and the reaction mixture was kept at room temperature for 3 h. Afterwards, it was poured into ice-cold water (200 mL) and conc. HCl was added to pH 2. The precipitate was filtered off and recrystallized from ethanol. Yield   (4): A mixture of sulfonamide 2 (3.04 g, 10 mmol), acrylic acid (1.7 mL, 25 mmol), toluene (15 mL) and glacial acetic acid (5 mL) was heated under reflux for 20 h. After solvent evaporation, the residue was dissolved in 10% aqueous NaOH solution (20 mL) and extracted with ethyl ether (2 × 20 mL). The aqueous solution was acidified with acetic acid to pH 5. The precipitate was filtered off and recrystallized from water.  (5) (6) (7): A mixture of acid 3 (6.1 g, 25 mmol), methanol (20 mL) and conc. H2SO4 (0.3 mL) was heated under reflux for 18 h. After solvent evaporation, water (25 mL) was added to the residue and the obtained solution was neutralized with 10% aqueous Na2CO3 solution to pH 8-9. The precipitate was filtered off and recrystallized from propan-2-ol. Yield 6.  (9) (10): Method A. Semicarbazide 9 (1.59 g, 5 mmol) was dissolved in conc. H2SO4 (12 mL) at 5 °C and the mixture was stirred at room temperature for 15 h. Afterwards, the reaction mixture was poured into ice-cold water (100 mL) and ammonia was added to pH 8−9. Precipitate was filtered off and recrystallized from propan-2-ol. Yield 0.51 g (34%). Method B. Acid 3 (1.22 g, 5 mmol) was dissolved in conc. H2SO4 (7 mL), the mixture was cooled down to room temperature, thiosemicarbazide (0.46 g, 5 mmol) was added and the reaction mixture was stirred at 70 °C for 3 h. Afterwards, the reaction mixture was poured onto grinded ice (300 mL) and neutralized with ammonia to pH 8−9. Precipitate was filtered off and recrystallized from propan-2-ol.

Protein Preparation
Expression and purification of CA I, II, VI, VII, XII, and XIII was performed as previously described: CA I and VI in [40], CA II in [36], CA VII and XIII in [41], and CA XII in [42].

Isothermal Titration Calorimetry
Isothermal titration calorimetry (ITC) experiments were performed using VP-ITC instrument (Microcal, Inc., Northampton, MA, USA) with 5-10 µM protein solution in the cell and 50-100 µM of the ligand solution in the syringe. A typical experiment consisted of 25 injections (10 µL each) within 3 min intervals. Experiments were performed at 37 °C in a 50 mM sodium phosphate buffer containing 100 mM NaCl at pH 7.0, with a final DMSO concentration of 2%, equal in the syringe and the cell.

Crystallization
CA II in 20 mM Hepes buffer pH 7.5 and 50 mM NaCl was concentrated by ultrafiltration to the concentration 50 mg/mL. CA II crystals were obtained by sitting drop vapor diffusion method mixing 4 μL of protein solution with 4 μL of crystallization buffer. Crystallization buffer contained 0.1 M sodium bicine, pH 9.0 and 2.0 M sodium malonate, pH 7.5. Crystals were grown at 20 °C for several weeks. Complexes were prepared by soaking crystals with 0.5 mM solution of ligand in crystallization buffer (prepared by mixing of 50 mM stock solution of ligand in DMSO with crystallization solution).

Data Collection and Structure Determination
Diffraction data were collected at the EMBL beam line P13 at the storage ring PETRAIII (DESY, Hamburg, Germany). Datasets were processed using XDS [44], TRUNCATE [45], and SCALA [46,47]. Molecular replacement was performed using MOLREP [48]. The protein moiety from PDB entry 3HLJ was used as an initial model. Model building and refinement was carried out with COOT [49] and REFMAC [50], respectively. Atomic coordinates of ligands were generated using molecular editor Avogadro [51]. Descriptions of ligand geometry for structure refinement were generated with LIBREFMAC [52]. The data collection and refinement statistics as well as PDB access codes (4Q6D, 4Q6E) are presented in Table 2. Graphic representations of crystal structures are prepared with MOLSCRIPT [53], RASTER3D [54], and BOBSCRIPT [55].

Conclusions
A series of N-aryl-β-alanine and diazobenzenesulfonamides were synthesized and tested as binders of CA isozymes I, II, VI, VII, the catalytic domain of XII, and XIII. The most potent diazobenzenesulfonamides, possessing hexamethylimine and piperidine groups (compounds 30 and 31), displayed nanomolar binding affinity toward CA I. The X-ray crystallographic cocrystal structures of compounds 18 and 31 bound to CA II showed that the location of the benzene ring did not depend on the para-substituent. The more hydrophobic para-substituent of inhibitor 31 improved the binding properties due to additional hydrophobic interactions.