Kinetic and in silico studies of hydroxy-based inhibitors of carbonic anhydrase isoforms I and II

Abstract

A series of hydroxy and phenolic compounds have been assayed for the inhibition of two physiologically relevant carbonic anhydrase (CA, EC 4.2.1.1) isozymes, the cytosolic human isozymes I and II. The investigated molecules showed inhibition constants in the range of 1.07–4003 and 0.09–31.5 μM at the hCA I and hCA II enzymes, respectively. In order to investigate the binding mechanisms of these inhibitors, in silico studies were also applied. Molecular docking scores of the studied compounds are compared using three different scoring algorithms, namely Glide/SP, Glide/XP and Glide/IFD. In addition, different ADME (absorption, distribution, metabolism and excretion) analysis was performed. All the examined compounds were found within the acceptable range of pharmacokinetic profiles.

• ADME
• CA inhibitors
• carbonic anhydrase
• molecular docking simulations

Introduction

The carbonic anhydrases (CAs, EC 4.2.1.1) are zinc-containing metaloenzymes which participate in the maintenance of pH homeostasis in the human body, catalyzing the reversible hydration of carbon dioxide in a two-step reaction, to yield bicarbonate and protons1–3. So far, 16 CA isozymes have been described, that differ in their subcellular localization and catalytic activity. These isozymes play important roles in different tissues2–5. CAs constitute interesting targets for the design of pharmacological agents that are useful in the treatment or prevention of a variety of disorders such as glaucoma, acid–base disequilibria, epilepsy and other neuromuscular diseases, altitude sickness, edema and obesity3–9.

Many hydroxy compound derivatives have been widely used as pro-drugs or drugs. For instance, venlafaxine is used as an antidepressant. Another hydroxy group containing molecule, tetracycline, is mainly used for the treatment of bacterial infections. Our group recently investigated the interaction of CA isozymes I, II with salicylic acid derivatives, antioxidant phenolic compounds, organic nitrates, organic sulphate, hydroxy group containing compounds, etc. (Figure 1)6–12. We wanted to extend these earlier investigations to some hydroxy compounds in order to discover powerful CA inhibitors (CAIs) which might have implications in medicine6–10.

Figure 1. Chemical structures of some carbonic anhydrase inhibitor hydroxy compounds.

In this work, previously reported synthesized hydroxy including derivatives (i.e. compounds 7 and 8) are tested to check their CA inhibition profiles. These compounds were evaluated for their ability to inhibit human CA I and II enzymes. Inhibition is reported as K_I (μM) and the results are the average of at least three independent experiments. Since X-ray structures of these small molecules with CA targets are not yet available, in silico studies are performed for the investigation of their binding mechanisms.

Materials and methods

Chemicals

The CNBr-activated Sepharose 4B, protein assay reagents, p-aminobenzene sulphonamide, l-tyrosine, 4-nitrophenylacetate (4-NPA) and chemicals for electrophoresis were purchased from Sigma-Aldrich Co. (St. Louis, MO). All other chemicals were of analytical grade and obtained from either Sigma or Merck (Darmstadt, Germany).

Synthesis of compounds 7, 8 and 12: Detailed synthetic procedures for the preparation of (1-methyl-4-hydroxy-4-phenylpiperidin-3-yl)(phenyl)methanone (7) and (1-ethyl-4-hydroxy-4-phenylpiperidin-3-yl)(phenyl)methanone (8) can be found in Sharma et al13. Detailed synthetic procedures for the preparation of 1R(S),2R(S)-2-(3-bromo-4,5-dimethoxybenzyl)cyclohexanol (12) can be found in Balaydin et al8.

CA purification assay

Purification of two human CA isozymes (hCA I and hCA II) was previously described with a simple one step method by a Sepharose-4B aniline–sulfanilamide affinity column chromatoghrapy14.

CA activity assay and kinetic studies

Carbonic anhydrase activity was assayed by following the change in absorbance at 348 nm of 4-nitrophenylacetate (4-NPA) to 4-nitrophenylate ion over a period of 3 min at 25 °C using a spectrophotometer (Shimadzu UV–VIS) according to the method described by Verpoorte et al15. The inhibitory effects of compounds 7 and 8 were examined and their inhibitory profiles are compared with previously tested chemicals (Figure 2). All compounds were tested in triplicate at each concentration used. Control cuvette activity in the absence of inhibitor was taken as 100%. For each inhibitor an Activity %–[Inhibitor] graph was drawn. To determine K_I values, three different inhibitor concentrations were tested: In these experiments, 4-NPA was used as substrate at five different concentrations (0.15–0.75 mM). The Lineweaver–Burk curves were drawn16. Regression analysis graphs were drawn for IC₅₀ using inhibition % values by a statistical package (SPSS-for windows; version 10.0) on a computer (Student’s t-test; n: 3).

Figure 2. Chemical structures of compounds 7–15 and AZA.

In silico methods

In silico approaches were applied to estimate binding energy values at the ligand-binding domain, critical amino acid residues in binding site, ligand–protein interactions and ADME (Absorption, Distribution, Metabolism and Excretion) profiles of synthesized compounds.

Ligand setting

All chemical structure preparation procedure for the used ligands in this work was performed using MarvinSketch17 software. Consequence ligand preparations steps are follow as: (i) 2D structure of each ligand was sketched, (ii) 3D structures of all ligands were generated using molecular mechanical (MM) geometry optimization approach, (iii) all ligands were then set to the physiological pH (pH 7.4) at the protonation state step and geometry optimization is repeated.

Receptor preparation

Crystal structures of CA I and II are retrieved from Protein Data Bank server (PDB: 2NMX and 3M04)18, respectively. Both structures were submitted in protein preparation procedure using Schrodinger software19 which includes: (i) removing of all water molecules from crystal structure (implicit solvation was used), (ii) assignment of bond orders, (iii) adding hydrogen atoms, (iv) setting of physiological pH (pH 7) by protonation states of corresponding amino acid residues using implemented PROPKA software20 in Schrodinger molecular modeling package, (v) and finally, restrained minimization of added hydrogen atoms.

Molecular docking simulations

Glide/SP and Glide/XP docking protocols21 are applied for the prediction of topologies as well as binding energies (i.e. docking scores) of used compounds at the active sites of CA I and CA II enzymes. Active site of the enzyme is defined from the co-crystallized ligands from PDB files. Throughout the docking simulations both partial flexibility and full flexibility around the active site residues are performed by Glide/SP/XP and Induced Fit Docking (IFD)22 approaches. IFD procedure was established in three consequence stages which involves: (i) docking of all ligands into the generated model, (ii) refining of amino acid residues within 4 Å of docked poses, (iii) re-docking of docked ligands against refined protein.

ADME analysis

In order to estimate biological, pharmaceutical and drug similarity of synthesized compounds, ADME prediction analysis was carried out using QikProp program23.

Results and discussion

The rationale of investigating phenols as CAIs is due to the fact that the simple phenol (11) has been shown to be the only competitive inhibitor with CO₂ as substrate for the main isoform of CA, i.e. human CA II (hCA II)24. Our group investigated the interactions of a few simple phenols, salicylic acid derivatives, antioxidant phenolic compounds, some natural product polyphenols and phenolic acids with all mammalian isozymes, CA I–XV24–26. Results for these molecules were in low micromolar/submicromolar inhibitions and derived molecules may assist to design isozyme selective CAIs. Indeed, the inhibition profile of various isozymes with this class of agents was diverse, with inhibition constants ranging from the millimolar to the submicromolar for many simple phenols24–27.

We report here the first study on the inhibitory effects of compounds 7 and 8 on the esterase activity of hCA I and II. Compounds 7–15 and AZA were docked at the binding site of the targets (hCA I and II). Glide SP/XP docking scores of docked inhibitors at hCA I and II targets and corresponding binding interactions are tabulated in Table 1. The previous reports by Innocenti et al25 investigated other phenol derivatives (including compounds 13–15) using a stopped flow, CO₂ hydration assay for monitoring CA inhibition. Data of Table 1 show the following regarding inhibition of hCA I and II with compounds 9–12, by an esterase assay8, with 4-nitrophenylacetate (4-NPA) as substrate:

Table 1. K_I values obtained from regression analysis graphs for hCA I and hCA II in the presence of different inhibitor concentration (µM).

	hCA I			hCA II
Compound	Exp (μM) KI	Glide SP/XP	IFD	Exp (μM) KI	Glide SP/XP	IFD
7	13.86	−5.40/−5.36	−7.42	31.5	−6.33/−5.77	−8.17
8	9.90	−5.37/−7.08	−8.47	10.19	−6.44/−5.74	−8.56
9	1.04*	−5.32/−5.07	−4.61	0.38*	−3.97/−4.27	−4.59
10	7.56*	−4.06/−3.40	−4.95	0.34*	−4.68/−3.91	−4.81
11	7.64†	−4.59/−3.11	−5.30	0.38†	−4.88/−3.61	−3.91
12	10.2*	−4.75/−2.64	−7.28	5.50*	−4.40/−3.65	−6.87
13	4003†	−4.84/−4.03	−5.21	9.90†	−4.37/−4.08	−5.37
14	795†	−5.27/−3.28	−4.85	7.70†	−5.88/−3.18	−4.80
15	10.7†	−4.83/−3.11	−5.23	0.09†	−5.00/−2.40	−4.54
AZA	0.25†	–	–	0.012†	–	–

Docking results are in kcal/mol. Mean from at least three determinations. Errors in the range of ±3% of the reported value (data not shown).

*From Balaydin et al8.

†From Innocenti et al25.

(i) Against the slow cytosolic isozyme hCA I, compounds 13 and 14 behave as moderate inhibitors, with K_I values in the range of 795–4003 μM, similarly to the structurally related compounds 7–11, 15 and acetazolamide (AZA) (K_Is of 0.25–13.86 µM). It is interesting to note that the compounds 7–12 and 15 were much better hCA I inhibitors as compared to the corresponding compounds 13 and 14. Kinetic investigations (Lineweaver–Burk plots, data not shown) indicate that similarly to sulfonamides and inorganic anions5–12^,25–30, all the investigated natural compounds act as competitive inhibitors with 4-NPA as substrate (i.e. they bind in different regions of the active site cavity as compared to the substrate). However, the binding site of 4-NPA itself is unknown, but it is presumed to be in the same region as that of CO₂, the physiological substrate of this enzyme25–28.

(ii) A better inhibitory activity has been observed with compounds 9–11, 15 and AZA investigated here for the inhibition of the rapid cytosolic isozyme hCA II (Table 1). Structure–activity relationship (SAR) is thus quite sharp for this small series of hydroxylic compounds: the 4-hydroxy-4-phenylpiperidin-3-yl)(phenyl)methanone containing compounds 7 and 8 are ineffective leads, with two dihydroxy moieties containing compounds 13 and 14 is already a submicromolar hCA II inhibitor. The best hCA II inhibitor in this series of derivatives were compounds 9–11 and 15, with a K_I of 0.09–0.38 µM. It must be stressed that K_Is measured with the esterase method are always in the micromolar range because hCA I and II are weak esterases25–28.

In addition to partial flexible docking simulation, flexible receptor docking at the active site was also carried out to enhance fitting treatment of amino acid residues within 4 Å of the ligand. The IFD results are tabulated in Table 1 for both CA I and II receptors (top docking scores were considered). Revealed binding energies from IFD simulations are lower (i.e. higher docking scores) than the results of Glide/SP and XP results. In the case of compounds 7 and 8 (synthesized inhibitors), lowest binding energies against both CA I and II receptors were connected to these ligands. 2D and 3D ligand-binding interaction diagrams of docking poses of both inhibitors into the ligand binding site of CA I and II receptors are represented in Figure 3. These ligand diagrams clearly identify key amino acid residues which participate in ligand-binding interactions for both receptors. Hydrophobicity and polarity of binding pockets were identified for each complex in the diagrams. H-bonding and π–π stacking interactions were displayed by red and blue lines, respectively. Zinc metal ion in all complexes was located in the active site and the neighborhood of inhibitor. His94 amino acid residue was constrained in all complexes, which plays a critical role in CA I and II inhibitory processes. In the case of CA I, Gln92 amino acid residue forms a H-bond with hydroxyl group of compounds 7 and 8. In the case of CA II, Thr200 has participated in ligand–receptor interaction and form H bonds with hydroxyl groups of compounds 7 and 8. Figures 4 and 5 display 3D docking poses of ligands 7 and 8 at the binding pockets of CA I and II, respectively. In order to determine active sites and orientations of ligands, positions of the docked ligand were zoomed graphically for each complex and the corresponded residues around 4 Å of ligands were labeled. According to the superimposition data of the complexes, in individual CA I and II receptors, both inhibitors have shared the similar positions in ligand-binding pockets.

Figure 3. 2D docking poses of compounds 7 and 8 into the hCA I and II receptors.

Figure 4. 3D docking poses of compounds 7 and 8 into the hCA I receptor. Key amino acid residues were determined in the 3D positions in active sites. Superimposition of docked compounds 7 and 8 into the hCA I (lower panel).

Figure 5. 3D docking poses of compounds 7 and 8 into the CA II receptors. Key-amino acid residues were determined in the 3D positions.

In order to examine the biological and pharmaceutical characteristics of synthesized molecules, different ADME tests such as hydrophobic solvent accessible surface area (SASA), hydrophilic SASA, central nervous system (CNS) activity, blockage of hERG K+ channel activity and percentage of human oral absorption were performed for all compounds. The results are tabulated in Table 2. All the compounds were located in an acceptable range on the base of these estimations. In addition, their standard drug similarity was predicted, and five approved similar drugs (based on used ligand similarity indices of >80%) for individual inhibitors has been reported (Table 3). In order to examine the CA I and II inhibitory capacity of approved drugs from drug similarity analysis, several drugs were randomly selected and docked into the receptors, the results are tabulated in Table 4.

Table 2. Predicted ADME properties of synthesized compounds.

Compound	MW*	Hydrophobic SASA†	Hydrophilic SASA‡	CNS¶	QploghERG§	QplogCaco||	QplogBB#	Percent oral absorp**
7	295.380	177.890	52.977	1	−5.026	776.996	0.358	100.000
8	309.407	207.371	47.314	1	−4.927	879.261	0.346	100.000
9	114.187	285.911	41.696	1	−2.492	3985.599	0.189	100.000
10	142.241	347.269	41.284	1	−3.007	4021.648	0.049	100.000
11	94.113	0.000	54.672	1	−3.441	3002.271	0.098	100.000
12	329.233	387.001	34.985	1	−4.107	4614.673	0.122	100.000
13	110.112	0.000	101.351	−1	−3.338	1083.400	−0.316	85.925
14	110.112	0.000	109.311	−1	−3.344	910.548	−0.379	84.596
15	110.112	0.000	109.336	−1	−3.344	910.046	−0.379	84.591

*Molecular weight of the molecule (reasonable value <500).

†Hydrophobic component of the SASA (saturated carbon and attached hydrogen).

‡Hydrophilic component of the SASA (SASA on N, O and H on heteroatoms).

¶Predicted central nervous system activity on a −2 (inactive) to +2 (active) scale.

§Predicted blockage of hERG K+ channel (reasonable value <−7).

||Predicted apparent Caco-2 cell permeability in nm/s (reasonable value >25).

#Predicted brain/blood partition coefficient.

**Percentage of human oral absorption (<25% is weak and >80% is strong).

Table 3. Drug similarity using QikProp algorithm.

Compound	Approved drugs
7	Diphenylpyraline	Chlorcyclizine	Bamipine	Nepinalone	Ketotifen
8	Diphenylpyraline	Bamipine	Benztropine	Thenalidine	Chlorcyclizine
9	Methimazole	Isaxonine	Ethosuximide	Fomepizole	Nicotinyl
10	Cyclopentamine	Phenylpropanol	Isaxonine	Hexapropymate	Paroxypropione
11	Methimazole	Fomepizole	Nicotinyl	Isaxonine	Chlorzoxazone
12	Venlafaxine	Tramadol	Brompheniramine	Chlorpheniramine	Bromhexine
13	Gimestat	Nicotinyl	Metacetamol	Acetaminophen	Ethionamide
14	Gimestat	Nicotinyl	Metacetamol	Acetaminophen	Ethionamide
15	Gimestat	Nicotinyl	Metacetamol	Acetaminophen	Ethionamide

Table 4. Docking scores of approved drugs within hCA I and II enzymes (docking scores are in kcal/mol).

	CA I		CA II
Compound	Glide/SP	Glide/XP	Glide/SP	Glide/XP
Bromhexine	−4.749	−6.007	−4.527	−3.615
Nepinalone	−4.252	−5.466	−4.727	−4.543
Diphenylpyraline	−4.418	−5.184	−4.775	−4.828
Tramadol	−5.165	−4.908	−5.165	−5.407
Bamipine	−4.447	−4.826	−4.975	−4.579

In a recent study, it was reported that simple compounds such as catechol (13) and resorcinol (14)25 as CA inhibitors do not involve sulfonamide, sulfamate or related functional groups. These compounds can be used as starting point for a new class of CA inhibitors that may have advantages for patients with sulfonamide allergies25^,31. The sulfonamide zinc-binding group is thus superior to the thiol one (from the thioxolone hydrolysis product) for generating CA inhibitors with a varied and sometimes isozyme-selective inhibition profile against the mammalian enzymes. However, it is critically important to explore further classes of potent CAIs in order to detect compounds with a different inhibition profile as compared to the sulfonamides and their bioisosteres and to find novel applications for the inhibitors of these widespread enzymes.

Conclusions

Compounds used in this study affect the activity of CA isozymes due to the presence of the different functional groups (i.e. OH, phenyl, benzyl and cyclohexyl) present in their aromatic and nonaromatic scaffolds. In addition to the well-known sulfonamides and sulfamates, in this study another class of CAIs is presented. Indeed, some hyroxylic compounds investigated here showed effective hCA I and II inhibitory activity, in the low micromolar range, by the esterase method which usually gives K_Is an order of magnitude higher as compared to the CO₂ hydrase assay24–32. These findings point out that substituted hydroxylic compounds may be used as leads for generating potent CAIs eventually targeting other isoforms which have not been assayed yet for their interactions with such agents.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.