Novel Potential Binding Sites for Selective Inhibitor Design of Human Soluble Epoxide Hydrolase

Abstract

Human soluble epoxide hydrolase (hsEH) has been known to be involved in the hydrolysis of epoxyeicosatrienoic acids (EETs), which are anti-inflammatory and cardioprotective signalling molecules. Since the conversion of EETs into the corresponding diols generates non-bioactive molecules, the enzyme’s inhibition would be beneficial. hsEH quickly became the molecular target, however, the proposed inhibitors directed toward this enzyme have not undergone clinical trials, mostly due to their low solubility in water.

Our goal is to indicate new regions, distinct from those already investigated, that could be considered during the development of novel and perhaps more soluble inhibitors of hsEH. To fulfil this goal, we used a combination of classical and mixed-solvent molecular dynamics (cMD and MixMD) simulations and a ligand tracking approach. Such a strategy, considers conformational changes in the interior of the enzyme, extends conformational space of binding cavities, and provides insight into different sets of potential binding modes. By analysis of the local distribution of different molecular probes (water and organic cosolvent molecules), we were able to identify potential binding sites (hot-spots), which can indicate the location of the most important functional groups for the binding of designed inhibitors. The uniqueness of the binding spots was examined by superpositioning the indicated hot-spots with a map of chemical interactions between the known inhibitors and hsEH. We analysed all available crystal structures of hsEHs with inhibitors deposited in the PDB database to gain insight into the binding modes of the known inhibitors and to find similarities and dissimilarities between them. By comparing the results obtained based on the map of chemical interactions, with the regions identified by using cMD and MixMD, we were able to identify regions that have not been targeted by ligands until now and which are promising possible binding sites for polar groups that would be beneficial for inhibitors solubility.

Introduction

The human soluble epoxide hydrolase (hsEH, EC 3.3.2.10) is a bifunctional homodimeric enzyme weighing 62 kDa. It is a member of the epoxide hydrolase superfamily and is composed of two independently folded domains: N-terminal domain (NTD) with phosphatase activity and C-terminal domain (CTD) with hydrolase activity. hsEH is encoded by the EPHX2 gene and can be found in both cytosol and peroxisomes[1–3]. The enzyme shows the highest activity in the liver; however, it is distributed in the majority of tissues, including the kidney, intestine, adipocytes, and vessels[4–6]. The clinical relevance of hsEH is linked with a C-terminus motif which performs the hydrolysis of arachidonic acid epoxides and other natural epoxy-fatty acids, especially the transformation of four regioisomers of epoxyeicosatrienoic acids (EETs): 5,6-, 8,9-, 11,12-, and 14,15-EETs to the corresponding dihydroxyeicosatrienoic acids (DHETs)[7, 8]. The conversion of EETs into the corresponding diols performed by hsEH generates non-bioactive molecules, thus the enzyme inhibition would be expected to enhance the EETs bioavailability and their beneficial properties[9, 10]. EETs have been found to modulate the activities of many molecular targets and signalling pathways. They have been shown to possess vasodilator and anti-inflammation properties, as well as act as mediators of autocrine and paracrine[9, 11, 12]. These characteristics are why many studies implicate their significant meaning in several cardiovascular disorders, e.g. blood pressure, hypertension, and atherosclerosis[13, 14]. EETs have also been found in the brain, where they regulate cerebral blood flow and protect against ischemic brain injury[15, 16]. Several studies were carried out to investigate the effect of the sEH gene deletion[17–19]. Samokhvalov et al. reported that such a deletion provides a protective effect against lipopolysaccharide (LPS)-induced cardiotoxicity by maintaining mitochondrial function[20]. In their research associated with Parkinson's disease, Ren et al. reported that sEH gene deletion or enzyme inhibition may protect against MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) induced neurotoxicity[21]. EETs from the cytochrome P450/soluble epoxide hydrolase pathway, together with the prostaglandins derived from the cyclooxygenase (COX) pathway, have also been found to be important eicosanoids that may regulate angiogenesis and tumorigenesis[22, 23]. Therefore, a great effort has been made to develop inhibition strategies towards sEH.

Some of the earliest sEH inhibitors were chalcone oxide, which can form a covalent intermediate with the enzyme. However, they were slowly hydrolysed and then released from the active site due to time-dependent inhibition and they were rather unstable, making in vivo usage limited[9, 24–26]. Urea, carbamate, and amide compounds with the hydrophobic group substitution were the first successful generation of hsEH inhibitors. They were identified as significantly more potent inhibitors and still represent the most popular class of hsEH inhibitors[26, 27]. These compounds can mimic epoxide binding and are not released from the active site. However, they have rather poor physical properties which include, low solubilities and high melting points, resulting in inappropriate bioavailability[26, 28, 29]. To improve their properties, while maintaining a high potency, a polar functional group at a proper distance from the urea or amide moiety was added. Soon after, the second generation of inhibitors, based on the 12-(3-adamantan-1-ylureido)dodecanoic acid (AUDA) and 1-adamantan-1-yl-3-{5-[2-(ethoxyethoxy)ethoxy]pentyl} urea (AEPU), were developed[9, 26, 28, 30–32]. But, these compounds remain flawed, due to their rapid in vivo metabolism and limited pharmacokinetics parameters.

Bruce Hammock’s group proposed the so-called “Pharmacophore Model”, which has been widely used to develop new sEH inhibitors[33]. The “Pharmacophore Model” is as follows: the “primary pharmacophore” (P1) may consist of carbamate, amide, or urea (preferably) and bear a bulky, hydrophobic substituent, whereas the “secondary pharmacophore” (P2) should consist of a polar group, located at a distance of five or six carbons (7.5 Å) from the urea carbonyl group. The “secondary pharmacophore” could be any group that is capable of serving as a proton acceptor (e.g., amide, ketone, alcohol, etc.); however, moieties that are too polar offered less potent inhibition. The model may also show an optional “tertiary pharmacophore” (P3), located at least 12 Å from the urea carbonyl group. The hydrophobic linkers, L1 and L2, should connect the polar moieties from the primary, secondary and tertiary pharmacophores[33–35].

Several pharmaceutical companies and academic institutions and pharmaceutical companies, such as the University of California-Davis, Columbia University, Chinese Academy of Sciences, Arête, Boehringer Ingelheim, Taisho, Takeda, Merck, Roche, etc., have entered the arena of developing sEH inhibitors. Thousands of various hsEH inhibitors were synthesised over the last three decades, but only individual compounds have been qualified into clinical trials. Such an example is the 1-(1-acetyl-piperidin-4-yl)-3-adamantan-1-yl-urea (AR9281) inhibitor from Arête Therapeutics[36]. The compound has completed phase I of the clinical trial successfully[37] but has failed the IIa clinical trial due to lack of efficiency[38]. Esters and salts of AUDA, commonly used as an experimental sEH reference inhibitor, are good inhibitors of sEH but their clinical use has been restricted due to metabolic instability and limited solubility in water and many organic solvents[33, 34, 39]. Also, GSK2256294 ((1R,3S)-N-(4-cyano-2-(trifluoromethyl)benzyl)-3-((4-methyl-6-(methylamino)-1,3,5-triazin-2-yl)amino)cyclohexanecarboxamide) inhibitor has shown promising effects in two phase I studies, however, there is no information on the results of phase II clinical trials[40].

Various research groups began using the fragment-based drug discovery with biophysical or biochemical assays and a high-concentration of fragments to speed up the research of new inhibitors. For instance, Amano et al. conducted fragment screening against sEH using X-ray crystallography together with biochemical assays[41, 42]. They were able to identify numerous fragments that inhibit sEH and reveal their binding modes. Using the optimisation techniques, they determined structures of fragment-enzyme complexes and discovered a novel scaffold for sEH inhibitors. In 2015, Öster et al. also presented a fragment-based crystallisation approach, where a fragment-based lead generation was executed with a limited high-throughput screening against the sEH[43]. The obtained complexes extended the structural knowledge about possible dual conformations or multiple binding sites. In 2016, Xue et al. used a similar approach to completely map the active site pocket to gain a deeper knowledge of ligand interactions and to guide the rational structure-based drug design[44].

In the last decade, research groups also applied various approaches to discover new hsEH inhibitors based primarily on in silico studies. In 2011, Tanaka et al. identified almost 70 new sEH inhibitors using a docking-based virtual screening[45]. They analysed different physicochemical properties of the hit compounds, e.g., size and lipophilicity, to find a compound with high ligand efficiency. Xing et al. used a similar approach to construct combinatorial libraries of potential inhibitors based on a benzoxazole template[46]. The libraries were further evaluated by docking and by analysing the capacity to make hydrogen bonds. Research was then expanded into exploring the structure-activity relationship by the amide bond formation. In 2012, Moser et al. developed a pharmacophore model based on co-crystallised inhibitors of hsEH[47]. They used a combination of Protein Ligand Interaction Fingerprints with the QueryGenerator tool to model the shape of the binding pocket indicated as excluded volumes. Then, they proposed a new pharmacophore model (with a special focus on the catalytic triad, as described by Morisseau and Hammock[26]) by examination of the clusters of receptor pharmacophore annotation points within the binding pocket. This pharmacophore model was later used by Moser et al. to find dual 5-lipoxygenase (5-LO)/soluble epoxide hydrolase inhibitors[48]. In 2016, Waltenberger et al. created multiple structure- and ligand-based pharmacophore models to cover different putative binding modes of the ligands and to cover the active space more thoroughly[49]. In 2018, Tripathi et al. applied a 3D-QSAR approach using sEH inhibitors from literature (with a wide range of activity and structural diversity) to design a pharmacophore model[50]. A new pharmacophore hypothesis was presented by Bhagwati et al. in recent research, based on the molecular modelling studies, which include ligand-based pharmacophore modelling, virtual screening, molecular docking, and all-atom molecular dynamics simulations[51].

The use of in silico methods has undoubtedly helped propose new hsEH inhibitors, however, the research was mainly focused on exploring the main binding pocket and developing new pharmacophore models based on the structures of known inhibitors binding in this area. Exploration of other parts of the hsEH enzyme, e.g., accessible after the conformation changes that could host additional functional groups, was omitted. For the exploration of new, potential binding sites in the protein, molecular dynamics simulation with (co)solvent molecules could be applied. Such an approach, called mixed-solvent molecular dynamics (MixMD) or cosolvent-based molecular dynamics (CMD) simulations, mixes small organic molecules with a water solvent to stimulate dynamic protein motions and induce partial conformational changes of binding pocket residues appropriate for the binding of diverse ligands[52–56]. Based on such abundant information, it is possible to identify the ligand binding modes, preferable binding regions (often called hot-spots), and present structural characteristics of a given place to build a pharmacophore model[57–59]. But, the MixMD has several issues. There is no universal procedure or methodology for performing this type of simulation. Various research groups select different types of cosolvent molecules and the concentration of cosolvent in the systems can vary significantly[52, 53, 67–69, 57, 60–66]. One problem is the selection of cosolvents to properly map all of the possible protein-ligand interactions and regions, such as hydrogen bonding (donors and acceptors), hydrophobic regions, and positive and negative ionic interactions[60, 70, 71]. Moreover, the water:cosolvent ratio could also affect the results, since a higher concentration of particular molecules could disturb the protein’s structure and the cosolvent molecules could aggregate[54, 72–74]. MixMD simulations were usually used for relatively small proteins, such as thermolysin[52, 63, 75–77], p53[76, 78, 79], HIV-1 protease[80–82], hen egg-white lysozyme[83], Hsp90[81], and several kinases[72, 84] to identify potential binding regions on the protein’s surface exclusively. Therefore, an analysis for proteins with buried active sites (as in the case of hsEH) remains a challenge. For a long time, no bioinformatics tools were available that would allow accurate tracking of the behaviour of solvent molecules inside the protein’s core and its further analysis from an intramolecular voids the point of view during the molecular dynamics simulations. The situation changed when new and easy-to-use AQUA-DUCT (AQ) software appeared[85]. This tool was primarily used to track the molecular trajectories of ligands (solvent molecules) throughout molecular dynamics simulations. The new version, released in 2019, also allows the identification of structurally important residues and regions of macromolecules responsible e.g., for ligand trapping, which can vary depending on the tracked ligand[86]. Thus, such functionality can be used for in silico drug design and for identifying new potential binding sites for hsEH inhibitors.

In this paper, we have addressed two main questions: i) is it possible to expand the targeted space of the hsEH interior to propose unique binding sites for new classes of inhibitors? and ii) is it possible to identify amino acids that, so far, were not involved with inhibitors’ interactions and could accept functional groups to increase the inhibitors solubility? To address these questions, we used the combination of small molecules tracking and classical and mixed-solvent molecular dynamics simulations to analyse the intramolecular voids of the hsEH. Such an approach allows us to analyse the local distribution of particular solvent molecules employed as molecular probes. The densest regions of water and cosolvent pathways (hot-spots), are used to predict particular regions of protein-ligand interactions and could be used for further pharmacophore design. Since hsEH has a relatively high number of crystallographic protein-inhibitor complexes deposited at the Protein Data Bank (PDB) database[87], we could use this abundant information to gain structural insight into the binding of known inhibitors and compare them with the obtained results. We believe that such an approach can be applied for any molecular target and can work as a universal pipeline in the drug discovery pathway.

Results

Discussion

A combination of experimental and computational approaches has accelerated the discovery of numerous compounds inhibiting the human soluble epoxide hydrolase. However, none of the identified compounds which have qualified for clinical trials have passed the required tests allowing it to be marked as a drug. One of the biggest problems is low solubility of the proposed inhibitors that comes from the predominantly hydrophobic characteristics of the binding site of hsEH[41, 94, 95]. To shed light on new possibilities in drug design, we performed a comprehensive review of all deposited structures of hsEH with inhibitors and crash these analyses with the molecular probe tracking results in classical and MixMD simulations.

Based on studies by other researchers a relatively large number of crystal structures are available. In 2015, Amano et al. used X-ray crystallographic fragment screening to identify fragment hits and describe their binding modes[42]. They identified eight fragment hits and determined their co-crystal structures. They presented N-ethylmethylamine as a promising scaffold for inhibitors of soluble epoxide hydrolase and they selected compounds containing this scaffold for biochemical assays. Although the starting fragment had a weak inhibitory activity (IC50: 800 uM), they were able to identify compounds with more than a 1500-fold increase in the inhibitory activity of IC50: 0.51 uM. They characterised the binding site for each fragment in their work[42]. In the same year, Öster et al. also presented a fragment-based crystallisation approach, where a fragment-based lead generation was executed with a limited high-throughput screening against the sEH[43]. They obtained 55 structure-ligand complexes, with a wide range of size and affinity. Superimposition of all structures indicated that the entire binding pocket is covered by ligands. They chose to deposit 52 structures within the hsEH dataset, including those containing some ambiguities, such as dual conformations or multiple binding sites, to provide the additional structural information[43]. In 2016, Xue et al. also used a similar approach to completely map the active-site pocket to gain a deeper knowledge of ligand interactions and to guide the rational structure-based drug design[44]. They obtained more than 50 structures of ligand complexes and through the analysis, they characterised preferred binding sites with some consensus hot-spots, in addition to the central channel. They identified two scaffolds: oxoindoline and 2-phenylbenzidazole-4-sulfonamide as a new series for potential hsEH inhibitors[44].

Most of the scientific research conducted so far for hsEH inhibitors design highlights the importance of the side chain residues composing the active site, D335 from the catalytic triad, as well as the stabilizing residues Y383 and Y466. The hydrogen-bonding network between the known inhibitors and these residues is a highly conserved feature across the reported hsEH inhibitors. This was demonstrated in numerous research studies. For instance, Gomez et al. presented that the binding of N-cyclohexyl-N′-(iodophenyl)urea (CIU) inhibitor is stabilised by a hydrogen bond interaction between the urea oxygen and T383 and Y466, as well as between the urea NH groups and D335[96]. In other work, Gomez et al. again stressed the importance of the hydrogen bond interaction of the above-mentioned residues with 4-(3-cyclohexylureido)-butyric acid (CU4) and 4-(3-cyclohexylureido)-hexanoic acid (CU6)[94]. Furthermore, X-ray crystallography research conducted by Tanaka et al. revealed the binding mode of inhibitors in which the amide group creates a hydrogen bonding network with the sEH active site residues[45]. Other studies have also been conducted using fragment-based drug discovery or in silico methods and have also pointed out the importance of D335, T383, and Y466[44, 97, 98]. Our studies confirm these findings, since the largest number of interactions (both non-bonded and hydrogen bonds) were found for these amino acids. Other important places, e.g. the hydrophobic region described in various studies and defined by F267, Y383, L408, M419, V498, and W525 (also known as F267 pocket), as well as the hydrophobic surface defined by W336, M339, and L499 (also known as W336 niche) are also reflected in our research by a high number of identified interactions between these residues and known inhibitors (Supplementary Information, Table S2).

However, the interaction with active site residues is not imperative. All deposited crystal structures underline the location of the binding site inside the protein’s interior, a common feature in enzymes. According to the study by Pravda et al., over 64% of enzymes contain the buried active site[99]. Although such a location adds additional constraints for an inhibitors design, it may also provide new opportunities. As described in the latest review by Marques et al., inhibitors can target not only the active site itself but also the tunnels providing access to it[92]. Indeed, our analysis of the deposited structures of hsEH with bound inhibitors indicates such compounds. Inhibitors belonging to the C-III cluster may be described as the ones that fill the internal pockets of the proteins, but as can be seen, their ends are placed within the Tc/m and Tm1 entry/egress areas. Inhibitors from C-I, C-II, and C-IV clusters are located exactly within the entry/egress areas with most near the Tm1 and Tg tunnels. In turn, inhibitors that have not been classified into any major cluster are located deeper in the internal pocket.

One limitation of all research conducted so far is the fact that they focus only on the main pocket in which the natural substrate binds, without the exploration of the potential conformational changes of the protein’s interior. This omits the possibility of designing compounds that would deeper penetrate the enzyme’s interior and would bind to the unexplored regions of hsEH. In a case of selective inhibitors design for nitric oxide synthase, even rotation of the side chain of a single amino acid can provide access to the additional pocket and provide a new anchor stabilising the ligand binding[100, 101]. Therefore, we aimed to identify new pockets that extend the interior of the well-known “L”-shaped binding cavity of hsEH. The identification of new, unique regions, where the binding of new compounds may potentially occur, was achieved by applying the combination of tracking small molecules used as specific molecular probes (water molecules during cMD and cosolvent molecules during MixMD simulations) together with the local distribution approach. Such a combination provides information about an overall distribution of solvent in the protein’s interior (pockets) as well as identifies areas with significant density (hot-spots) of selected molecules, which can be considered as regions of special interest. The uniqueness of this approach lies in the fact that the identification of such sites can take place in the protein’s interior and not only on its surface, as done in previous studies[54, 57, 102–105]. This is particularly important for enzymes that have a buried active site. The detected additional pockets filled with particular (co)solvent hot-spots may provide an indicator of cavities which can host additional (polar) functional groups of potential inhibitors and could increase the solubility of the newly discovered compounds, thus overcoming the limitations that make inhibitors fail the clinical trials.

The combination of various chemical probes also addresses one more important issue of drug design. Mostly due to their low solubility in water and organic solvents, the chemical properties of amino acids affect the nature of the internal pocket (hydrophobic), leading to problems. The solution of such a problem could be the insertion of some polar fragments into the existing inhibitors that would improve solubility. The analysis of known inhibitors indicates that they do not fully occupy the available internal pocket (this is the case for both branches of “L”-shaped pocket). This suggests that there is an additional potency, which may be gained by filling these places with additional substituents. Admittedly, in their work, Öster et al. as well as Xue et al. claim that the fragment binders sample the entire binding pocket[43, 44]. However, our analysis shows that there is still some unused space at the ends of both short and long branches of the “L”-shaped pocket. Nevertheless, their research has greatly contributed to understanding the internal binding pocket of hsEH and resulted in obtaining numerous structure-ligand complexes, with a wide range of size and affinity. It is also worth noticing that Amano et al. presented an interesting structural insight into the binding of inhibitors by fragment screening and X-ray crystallography[41, 42]. They identified 126 hsEH-fragment complexes, among which almost 50 do not strictly interact with the amino acids from the active site. Instead, they extend their occurrence to long and short branches, which further indicates that it is reasonable to design novel inhibitors outside the active site. These fragments could be expanded toward further protein exploration.

The computational tools were dedicated to tackle the issues e.g., by application of the Lipinski’s rule of five. In recent years, several in silico approaches have been applied to discover new hsEH inhibitors. In 2011, Tanaka et al. identified 68 new inhibitors (out of a 735 compound library) using a docking-based virtual screening[45]. They analysed different physicochemical properties of the hit compounds, e.g., size and lipophilicity, to find a compound with high ligand efficiency. Then, the lead compound was optimised to yield a more potent compound, characterised by good ADMET profile. Xing et al. used a similar approach to construct combinatorial libraries based on a benzoxazole template[46]. The library was evaluated by docking and by analysing the capacity to make hydrogen bonds. The first library focused on the coupling of 591 acids with the 5-chlorobenzoxazoleamine. The results showed that 343 out of 383 synthesised compounds were active in a biological assay. IC50 determination confirmed 23 compounds as single-digit nanomolar hsEH inhibition. The second library focused on exploring the structure-activity relationship by the amide bond formation between a subset of the acids and eight (aza)benzoxazole amines. 322 out of 840 compounds were active against hsEH[46]. The large study by Moser et al. involved a massive virtual screening of the “merged libraries” database (Asinex, > 430,000 molecules)[47]. Using this database, they performed a pharmacophore search. In the first step of analysis 3,191 virtual hits were found. Hits were further limited to 89 by applying the modified Lipinski’s rule of five and Oprea’s test for lead-likeness. Within this set, nine compounds were tested experimentally, of which two of the most potent hits showed an inhibition of over 50% at a single concentration of 30 µM. The pharmacophore model was later used by Moser et al. to find dual 5-lipoxygenase (5-LO)/soluble epoxide hydrolase inhibitors[48]. In 2016, Waltenberger et al. created multiple structure- and ligand-based pharmacophore models to cover different putative binding modes of the ligands and to cover the active space more thoroughly[49]. They used nine crystal structures and 68 actives from the literature to model the structure-based pharmacophore using LigandScout (LS). Based on the PDB entries, 22 structure-based pharmacophores were generated and the resulting model set was able to recognise 55 out of 68 actives. Because part of the active compounds were not covered, additional three ligand-based pharmacophore models were generated using LS. 25 combined pharmacophore models were able to find 58 out of 86 active compounds. The remaining structures (containing aliphatic chains and ring systems) were not recognised as hydrophobic by LS version 3.03. For this reason, Discovery Studio software was used to create an additional three structure-based and two ligand-based models. Finally, 30 pharmacophore models were evaluated and 67 active compounds were found to interact with them. Further elimination of models with low enrichment factor (EF) qualities was performed. In the end, eight models (five structure-based and three ligand-based) were able to cover 65 active molecules. Models were used to perform the prospective virtual screening of Specs database. Six compounds per model (48 total) were experimental and 19 new sEH active molecules, with a total hit rate of 40% were found. Six out of 19 compounds displayed IC50 values in the low nanomolar range. In 2018, Tripathi et al. applied a 3D-QSAR approach using 26 sEH inhibitors from literature (with a wide range of activity and structural diversity) to design a pharmacophore model containing four features: one hydrogen bond acceptor, two hydrophobic, and one ring aromatic[50]. The selected model was validated using both internal and external test set prediction, Fischer’s randomization, Rm2 metrics, and Güner-Henry scoring method. The chosen model was used to search the National Cancer Institute (NCI) and Maybridge datasets. The hits were screened based on estimated activity and fit value, along with Lipinski’s rule of five. Three hits with very high potency and fit value were chosen for sEH enzyme-based assay and rat aortic ring based vasodilation assay. They were also docked into the she. According to the most recent research, Bhagwati et al. presented a new pharmacophore hypothesis based on the molecular modelling studies, including ligand-based pharmacophore modelling, virtual screening, molecular docking, and all-atom molecular dynamics simulations[51]. For the pharmacophore model generation, they chose the top three inhibitors from the available crystal structures based on their IC50 values and interaction with the active site residues, D335 from the catalytic triad, along with Y383 and Y466. The ligand-based approach was used to produce three pharmacophore models using phase module. Then, the best alignment and common feature methods were used to create the pharmacophoric sites. For pharmacophore validation, the test set and decoy set method (with 28 confirmed actives and 1,232 decoys or inactives) were used. Validated pharmacophore was then used to screen the Maybridge database (consisting of around 53,000 small drug-like compounds) to obtain the hit compounds through the phase module. 1,802 hits were further subjected to molecular docking using an extra precision mode of Glide module. The top 100 hits from docking were visually analysed for their molecular interactions with the binding site. After that, nine hits were selected for in vitro inhibitor screening assay. Five out of nine compounds showed more than 50% inhibition at 25 µM concentration. Within this group, three compounds indicated inhibition greater than 60% and were selected for secondary screening at 5 µM concentration. Two out of three compounds showed inhibition of more than 50% at 5 µM concentration. To explore the mechanism of inhibition, as well as to analyse the binding modes, a docking procedure was applied. To get even better insight into the validity of docking and biological assay results, the molecular dynamics simulations of the three most active protein-ligand complexes were performed to examine their complexes and to compare them with the reference co-crystallised inhibitor. However, none of the above mentioned methods and trials have considered significant conformational space sampling or alternative binding regions.

In the proposed approach, we focused on searching for regions (pockets identified by cosolvent hot-spots) that due to native or inhibitors induced dynamics of the protein could host inhibitors or their functional groups. Such information was extracted from classical and mixed-solvent MD simulations and we were able to detect potential sites for placing the new functional groups. For instance, different hot-spots have been found at the ends of both branches of the “L”-shaped pocket. After applying the known inhibitors’ structures with identified hot-spots, it can be seen that only a few compounds reach these areas. A particularly interesting idea seems to be the expanding of inhibitors belonging to the C-IV cluster so that their side chains would be directed more towards the cap-domain (within the identified Tcap1 tunnel), where the acetonitrile hot-spot was found. The P1 pocket is surrounded by six amino acids that could be involved in new inhibitor binding – L417, S418, V422, A425, G427, and L428. The mechanism of the cap domain opening could then be blocked.

Another option to consider is to pull out the inhibitors from the binding pockets towards the P2 pocket and position them within the side entry/egress area instead – next to the Tm3 tunnel. Such inhibitors could perhaps fill the gorge between the cap and main domains of hsEH. Within the P2 pocket we identified five amino acids that could be engaged into the inhibitors’ binding – M441, C522, G523, W525, and D529. Three of them provide functional groups that can address strong polar or even covalent interactions with designed inhibitors. The inhibitors classified in the C-IV cluster could probably be used as starting points of synthesis with an aliphatic linker crossing near the W525 residue and polar functional group targeting P2 pocket. The third option would utilise the small inhibitors clustered into the C-I cluster as a starting point for designing new compounds. The detected P3 pocket provides a space that could host additional functional groups to increase solubility and selectivity of designed inhibitors.

Conclusions

Identifying hot-spots for different (co)solvents showed the importance of protein exploration using different molecular probes. By carrying out the analysis using several molecular probes with different chemical properties, it was possible to identify specific (possible) sites of interactions. Some of them were occupied by more than one type of (co)solvent, whereas some were more specific. This may mean that there are some regions in the protein structure, where the nature of the introduced groups can be various, while other places have a more specific and defined chemical character.

Methods

List of Abbreviations

hsEH, human soluble epoxide hydrolase; EETs, epoxyeicosatrienoic acids; cMD, classical molecular dynamics simulations; MixMD, mixed-solvent molecular dynamics simulations; NTD, N-terminal domain; CTD, C-terminal domain; EPHX2, bifunctional epoxide hydrolase 2 gene; DHETs, dihydroxyeicosatrienoic acids; sEH, soluble epoxide hydrolase; 5-LO, 5-lipooxygenase; CMD, cosolvent-based molecular dynamics simulations; AQ, AQUA-DUCT; ACN, acetonitrile; DMSO, dimethylsulfoxide; MEO, methanol; PHN, phenol; URE, urea; 3D-RISM, Three-Dimensional Reference Interaction Site Model; LS, LigandScout

Declarations

Availability of data and materials

Not applicable.

Competing interests

Not applicable.

Funding

The authors’ work was supported by the National Science Centre, Poland (grant no. DEC-2013/10/E/NZ1/00649).

Authors' contributions

Conceptualization, A.G.; methodology, M.B., and K.M.; software, M.B. and K.M.; validation, M.B., K.M. and A.G.; formal analysis, M.B.; investigation, M.B.; resources, M.B and K.M.; data curation, M.B and K.M.; writing—original draft preparation, M.B.; writing—review and editing, K.M. and A.G.; visualization, M.B., and A.G; supervision, A.G.; project administration, A.G.; funding acquisition, A.G. All authors have read and agreed to the published version of the manuscript.

Acknowledgements

The authors would like to thank Tomasz Magdziarz for his helpful advice in clustering analysis of data.

References

1. Gomez GA, Morisseau C, Hammock BD, Christianson DW (2004) Structure of Human Epoxide Hydrolase Reveals Mechanistic Inferences on Bifunctional Catalysis in Epoxide and Phosphate Ester Hydrolysis. Biochemistry 43:4716–4723. https://doi.org/10.1021/bi036189j
2. Newman JW, Morisseau C, Hammock BD (2005) Epoxide hydrolases: Their roles and interactions with lipid metabolism. Prog Lipid Res 44:1–51. https://doi.org/10.1016/j.plipres.2004.10.001
3. Harris TR, Hammock BD (2013) Soluble epoxide hydrolase: Gene structure, expression and deletion. Gene 526:61–74. https://doi.org/10.1016/j.gene.2013.05.008
4. Norwood S, Liao J, Hammock BD, Yang G-Y (2019) Epoxyeicosatrienoic acids and soluble epoxide hydrolase: Potential therapeutic target for inflammation and its induced carcinogenesis. Am J Transl Res 2:447–457
5. Cronin A, Decker M, Arand M (2011) Mammalian soluble epoxide hydrolase is identical to liver hepoxilin hydrolase. J Lipid Res 52:712–719. https://doi.org/10.1194/jlr.M009639
6. Liu J-Y (2019) Inhibition of Soluble Epoxide Hydrolase for Renal Health. Front Pharmacol 9:. https://doi.org/10.3389/fphar.2018.01551
7. Seubert JM, Zeldin DC, Nithipatikom K, Gross GJ (2007) Role of epoxyeicosatrienoic acids in protecting the myocardium following ischemia/reperfusion injury. Prostaglandins Other Lipid Mediat 82:50–59. https://doi.org/10.1016/j.prostaglandins.2006.05.017
8. Wagner K, Vito S, Inceoglu B, Hammock BD (2014) The role of long chain fatty acids and their epoxide metabolites in nociceptive signaling. Prostaglandins Other Lipid Mediat 113–115:2–12. https://doi.org/10.1016/j.prostaglandins.2014.09.001
9. Spector AA, Fang X, Snyder GD, Weintraub NL (2004) Epoxyeicosatrienoic acids (EETs): metabolism and biochemical function. Prog Lipid Res 43:55–90. https://doi.org/10.1016/S0163-7827(03)00049-3
10. Larsen BT, Gutterman DD, Hatoum OA (2006) Emerging role of epoxyeicosatrienoic acids in coronary vascular function. Eur J Clin Invest 36:293–300. https://doi.org/10.1111/j.1365-2362.2006.01634.x
11. Thomson SJ, Askari A, Bishop-Bailey D (2012) Anti-Inflammatory Effects of Epoxyeicosatrienoic Acids. Int J Vasc Med 2012:1–7. https://doi.org/10.1155/2012/605101
12. Jones RD, Liao J, Tong X et al (2019) Epoxy-Oxylipins and Soluble Epoxide Hydrolase Metabolic Pathway as Targets for NSAID-Induced Gastroenteropathy and Inflammation-Associated Carcinogenesis. Front Pharmacol 10:. https://doi.org/10.3389/fphar.2019.00731
13. Tacconelli S, Patrignani P (2014) Inside epoxyeicosatrienoic acids and cardiovascular disease. Front Pharmacol 5:. https://doi.org/10.3389/fphar.2014.00239
14. Yang L, Mäki-Petäjä K, Cheriyan J et al (2015) The role of epoxyeicosatrienoic acids in the cardiovascular system. Br J Clin Pharmacol 80:28–44. https://doi.org/10.1111/bcp.12603
15. Imig JD, Simpkins AN, Renic M, Harder DR (2011) Cytochrome P450 eicosanoids and cerebral vascular function. Expert Rev Mol Med 13:e7. https://doi.org/10.1017/S1462399411001773
16. Liu X, Davis CM, Alkayed NJ (2018) P450 Eicosanoids and Reactive Oxygen Species Interplay in Brain Injury and Neuroprotection. Antioxid Redox Signal 28:987–1007. https://doi.org/10.1089/ars.2017.7056
17. Zuloaga KL, Zhang W, Roese NE, Alkayed NJ (2015) Soluble epoxide hydrolase gene deletion improves blood flow and reduces infarct size after cerebral ischemia in reproductively senescent female mice. Front Pharmacol 5:. https://doi.org/10.3389/fphar.2014.00290
18. Wu C, Shyue S, Hung T et al (2017) Genetic deletion or pharmacological inhibition of soluble epoxide hydrolase reduces brain damage and attenuates neuroinflammation after intracerebral hemorrhage. 1–21. https://doi.org/10.1186/s12974-017-1005-4
19. Darwesh AM, Keshavarz-Bahaghighat H, Jamieson KL, Seubert JM (2019) Genetic Deletion or Pharmacological Inhibition of Soluble Epoxide Hydrolase Ameliorates Cardiac Ischemia/Reperfusion Injury by Attenuating NLRP3 Inflammasome Activation. Int J Mol Sci 20:3502. https://doi.org/10.3390/ijms20143502
20. Samokhvalov V, Jamieson KL, Darwesh AM et al (2019) Deficiency of Soluble Epoxide Hydrolase Protects Cardiac Function Impaired by LPS-Induced Acute Inflammation. Front Pharmacol 9:. https://doi.org/10.3389/fphar.2018.01572
21. Ren Q, Ma M, Yang J et al (2018) Soluble epoxide hydrolase plays a key role in the pathogenesis of Parkinson’s disease. Proc Natl Acad Sci 115:E5815–E5823. https://doi.org/10.1073/pnas.1802179115
22. Zhang G, Kodani S, Hammock BD (2014) Stabilized epoxygenated fatty acids regulate inflammation, pain, angiogenesis and cancer. Prog Lipid Res 53:108–123. https://doi.org/10.1016/j.plipres.2013.11.003
23. Zhang G, Panigrahy D, Hwang SH et al (2014) Dual inhibition of cyclooxygenase-2 and soluble epoxide hydrolase synergistically suppresses primary tumor growth and metastasis. Proc Natl Acad Sci 111:11127–11132. https://doi.org/10.1073/pnas.1410432111
24. Mullin CA, Hammock BD (1982) Chalcone oxides—potent selective inhibitors of cytosolic epoxide hydrolase. Arch Biochem Biophys 216:423–439. https://doi.org/10.1016/0003-9861(82)90231-4
25. Morisseau C, Du G, Newman JW, Hammock BD (1998) Mechanism of Mammalian Soluble Epoxide Hydrolase Inhibition by Chalcone Oxide Derivatives. Arch Biochem Biophys 356:214–228. https://doi.org/10.1006/abbi.1998.0756
26. Morisseau C, Hammock BD (2005) EPOXIDE HYDROLASES: Mechanisms, Inhibitor Designs, and Biological Roles. Annu Rev Pharmacol Toxicol 45:311–333. https://doi.org/10.1146/annurev.pharmtox.45.120403.095920
27. Morisseau C, Goodrow MH, Dowdy D et al (1999) Potent urea and carbamate inhibitors of soluble epoxide hydrolases. Proc Natl Acad Sci 96:8849–8854. https://doi.org/10.1073/pnas.96.16.8849
28. Morisseau C, Hammock BD (2008) Gerry Brooks and epoxide hydrolases: four decades to a pharmaceutical. Pest Manag Sci 64:594–609. https://doi.org/10.1002/ps.1583
29. Ghosh S, Chiang P-C, Wahlstrom JL et al (2008) Oral Delivery of 1,3-Dicyclohexylurea Nanosuspension Enhances Exposure and Lowers Blood Pressure in Hypertensive Rats. Basic Clin Pharmacol Toxicol 102:453–458. https://doi.org/10.1111/j.1742-7843.2008.00213.x
30. Imig JD (2006) Cardiovascular Therapeutic Aspects of Soluble Epoxide Hydrolase Inhibitors. Cardiovasc Drug Rev 24:169–188. https://doi.org/10.1111/j.1527-3466.2006.00169.x
31. Xu D, Li N, He Y et al (2006) Prevention and reversal of cardiac hypertrophy by soluble epoxide hydrolase inhibitors. Proc Natl Acad Sci 103:18733–18738. https://doi.org/10.1073/pnas.0609158103
32. Ulu A, Davis BB, Tsai H-J et al (2008) Soluble Epoxide Hydrolase Inhibitors Reduce the Development of Atherosclerosis in Apolipoprotein E-Knockout Mouse Model. J Cardiovasc Pharmacol 52:314–323. https://doi.org/10.1097/FJC.0b013e318185fa3c
33. Kim I-H, Morisseau C, Watanabe T, Hammock BD (2004) Design, Synthesis, and Biological Activity of 1,3-Disubstituted Ureas as Potent Inhibitors of the Soluble Epoxide Hydrolase of Increased Water Solubility. J Med Chem 47:2110–2122. https://doi.org/10.1021/jm030514j
34. Kim I-H, Heirtzler FR, Morisseau C et al (2005) Optimization of Amide-Based Inhibitors of Soluble Epoxide Hydrolase with Improved Water Solubility. J Med Chem 48:3621–3629. https://doi.org/10.1021/jm0500929
35. Li H-Y, Jin Y, Morriseau C et al (2006) The 5-substituted piperazine as a novel secondary pharmacophore greatly improving the physical properties of urea-based inhibitors of soluble epoxide hydrolase. Bioorg Med Chem 14:6586–6592
36. Anandan S-K, Webb HK, Chen D et al (2011) 1-(1-Acetyl-piperidin-4-yl)-3-adamantan-1-yl-urea (AR9281) as a potent, selective, and orally available soluble epoxide hydrolase inhibitor with efficacy in rodent models of hypertension and dysglycemia. Bioorg Med Chem Lett 21:983–988. https://doi.org/10.1016/j.bmcl.2010.12.042
37. Chen D, Whitcomb R, MacIntyre E et al (2012) Pharmacokinetics and Pharmacodynamics of AR9281, an Inhibitor of Soluble Epoxide Hydrolase, in Single- and Multiple-Dose Studies in Healthy Human Subjects. J Clin Pharmacol 52:319–328. https://doi.org/10.1177/0091270010397049
38. Www.clinicaltrials.gov (2009) The trial was terminated in November 2009 according to www.clinicaltrials.gov and no efficacy results have been reported
39. Morisseau C, Goodrow MH, Newman JW et al (2002) Structural refinement of inhibitors of urea-based soluble epoxide hydrolases. Biochem Pharmacol 63:1599–1608. https://doi.org/10.1016/S0006-2952(02)00952-8
40. Lazaar AL, Yang L, Boardley RL et al (2016) Pharmacokinetics, pharmacodynamics and adverse event profile of GSK2256294, a novel soluble epoxide hydrolase inhibitor. Br J Clin Pharmacol 81:971–979. https://doi.org/10.1111/bcp.12855
41. Amano Y, Yamaguchi T, Tanabe E (2014) Structural insights into binding of inhibitors to soluble epoxide hydrolase gained by fragment screening and X-ray crystallography. Bioorganic Med Chem 22:2427–2434. https://doi.org/10.1016/j.bmc.2014.03.001
42. Amano Y, Tanabe E, Yamaguchi T (2015) Identification of N-ethylmethylamine as a novel scaffold for inhibitors of soluble epoxide hydrolase by crystallographic fragment screening. Bioorganic Med Chem 23:2310–2317. https://doi.org/10.1016/j.bmc.2015.03.083
43. Öster L, Tapani S, Xue Y, Käck H (2015) Successful generation of structural information for fragment-based drug discovery. Drug Discov Today 20:1104–1111. https://doi.org/10.1016/j.drudis.2015.04.005
44. Xue Y, Olsson T, Johansson CA et al (2016) Fragment Screening of Soluble Epoxide Hydrolase for Lead Generation-Structure-Based Hit Evaluation and Chemistry Exploration. ChemMedChem 11:497–508. https://doi.org/10.1002/cmdc.201500575
45. Tanaka D, Tsuda Y, Shiyama T et al (2011) A Practical Use of Ligand Efficiency Indices Out of the Fragment-Based Approach: Ligand Efficiency-Guided Lead Identification of Soluble Epoxide Hydrolase Inhibitors. J Med Chem 54:851–857. https://doi.org/10.1021/jm101273e
46. Xing L, McDonald JJ, Kolodziej SA et al (2011) Discovery of Potent Inhibitors of Soluble Epoxide Hydrolase by Combinatorial Library Design and Structure-Based Virtual Screening. J Med Chem 54:1211–1222. https://doi.org/10.1021/jm101382t
47. Moser D, Achenbach J, Klingler F-M et al (2012) Evaluation of structure-derived pharmacophore of soluble epoxide hydrolase inhibitors by virtual screening. Bioorg Med Chem Lett 22:6762–6765. https://doi.org/10.1016/j.bmcl.2012.08.066
48. Moser D, Wisniewska JM, Hahn S et al (2012) Dual-Target Virtual Screening by Pharmacophore Elucidation and Molecular Shape Filtering. ACS Med Chem Lett 3:155–158. https://doi.org/10.1021/ml200286e
49. Waltenberger B, Garscha U, Temml V et al (2016) Discovery of Potent Soluble Epoxide Hydrolase (sEH) Inhibitors by Pharmacophore-Based Virtual Screening. J Chem Inf Model 56:747–762. https://doi.org/10.1021/acs.jcim.5b00592
50. Tripathi N, Paliwal S, Sharma S et al (2018) Discovery of Novel Soluble Epoxide Hydrolase Inhibitors as Potent Vasodilators. Sci Rep 8:14604. https://doi.org/10.1038/s41598-018-32449-4
51. Bhagwati S, Siddiqi MI (2019) Identification of potential soluble epoxide hydrolase (sEH) inhibitors by ligand-based pharmacophore model and biological evaluation. J Biomol Struct Dyn 1–11. https://doi.org/10.1080/07391102.2019.1691659
52. Seco J, Luque FJ, Barril X (2009) Binding Site Detection and Druggability Index from First Principles. J Med Chem 52:2363–2371. https://doi.org/10.1021/jm801385d
53. Lexa KW, Carlson HA (2011) Full Protein Flexibility Is Essential for Proper Hot-Spot Mapping. J Am Chem Soc 133:200–202. https://doi.org/10.1021/ja1079332
54. Ghanakota P, Carlson HA (2016) Driving Structure-Based Drug Discovery through Cosolvent Molecular Dynamics. J Med Chem 59:10383–10399. https://doi.org/10.1021/acs.jmedchem.6b00399
55. Ghanakota P, van Vlijmen H, Sherman W, Beuming T (2018) Large-Scale Validation of Mixed-Solvent Simulations to Assess Hotspots at Protein–Protein Interaction Interfaces. J Chem Inf Model 58:784–793. https://doi.org/10.1021/acs.jcim.7b00487
56. Uehara S, Tanaka S (2017) Cosolvent-Based Molecular Dynamics for Ensemble Docking: Practical Method for Generating Druggable Protein Conformations. J Chem Inf Model 57:742–756. https://doi.org/10.1021/acs.jcim.6b00791
57. Ung PMU, Ghanakota P, Graham SE et al (2016) Identifying binding hot spots on protein surfaces by mixed-solvent molecular dynamics: HIV-1 protease as a test case. Biopolymers 105:21–34. https://doi.org/10.1002/bip.22742
58. Sabanés Zariquiey F, de Souza JV, Bronowska AK (2019) Cosolvent Analysis Toolkit (CAT): a robust hotspot identification platform for cosolvent simulations of proteins to expand the druggable proteome. Sci Rep 9:19118. https://doi.org/10.1038/s41598-019-55394-2
59. Lee JY, Krieger JM, Li H, Bahar I (2020) Pharmmaker: Pharmacophore modeling and hit identification based on druggability simulations. Protein Sci 29:76–86. https://doi.org/10.1002/pro.3732
60. Ghanakota P, Carlson HA (2016) Driving Structure-Based Drug Discovery through Cosolvent Molecular Dynamics. J Med Chem 59:10383–10399. https://doi.org/10.1021/acs.jmedchem.6b00399
61. Alvarez-Garcia D, Barril X (2014) Relationship between Protein Flexibility and Binding: Lessons for Structure-Based Drug Design. J Chem Theory Comput 10:2608–2614. https://doi.org/10.1021/ct500182z
62. Yu W, Lakkaraju SK, Raman EP et al (2015) Pharmacophore Modeling Using Site-Identification by Ligand Competitive Saturation (SILCS) with Multiple Probe Molecules. J Chem Inf Model 55:407–420. https://doi.org/10.1021/ci500691p
63. Yang C-Y, Wang S (2010) Computational Analysis of Protein Hotspots. ACS Med Chem Lett 1:125–129. https://doi.org/10.1021/ml100026a
64. Yang C-Y, Wang S (2011) Hydrophobic Binding Hot Spots of Bcl-xL Protein – Protein Interfaces by Cosolvent Molecular Dynamics Simulation. ACS Med Chem Lett 2:280–284. https://doi.org/10.1021/ml100276b
65. Bakan A, Nevins N, Lakdawala AS, Bahar I (2012) Druggability Assessment of Allosteric Proteins by Dynamics Simulations in the Presence of Probe Molecules. J Chem Theory Comput 8:2435–2447. https://doi.org/10.1021/ct300117j
66. Huang D, Rossini E, Steiner S, Caflisch A (2014) Structured Water Molecules in the Binding Site of Bromodomains Can Be Displaced by Cosolvent. ChemMedChem 9:573–579. https://doi.org/10.1002/cmdc.201300156
67. Tan YS, Spring DR, Abell C, Verma C (2014) The Use of Chlorobenzene as a Probe Molecule in Molecular Dynamics Simulations. J Chem Inf Model 54:1821–1827. https://doi.org/10.1021/ci500215x
68. Tan YS, Spring DR, Abell C, Verma CS (2015) The Application of Ligand-Mapping Molecular Dynamics Simulations to the Rational Design of Peptidic Modulators of Protein–Protein Interactions. J Chem Theory Comput 11:3199–3210. https://doi.org/10.1021/ct5010577
69. Prakash P, Sayyed-Ahmad A, Gorfe AA (2015) pMD-Membrane: A Method for Ligand Binding Site Identification in Membrane-Bound Proteins. PLOS Comput Biol 11:e1004469. https://doi.org/10.1371/journal.pcbi.1004469
70. Arcon JP, Defelipe LA, Modenutti CP et al (2017) Molecular Dynamics in Mixed Solvents Reveals Protein–Ligand Interactions, Improves Docking, and Allows Accurate Binding Free Energy Predictions. J Chem Inf Model 57:846–863. https://doi.org/10.1021/acs.jcim.6b00678
71. Graham SE, Leja N, Carlson HA (2018) MixMD Probeview: Robust Binding Site Prediction from Cosolvent Simulations. J Chem Inf Model 58:1426–1433. https://doi.org/10.1021/acs.jcim.8b00265
72. Tan YS, Śledź P, Lang S et al (2012) Using Ligand-Mapping Simulations to Design a Ligand SelectivelyTargeting a Cryptic Surface Pocket of Polo-Like Kinase 1. Angew Chem Int E 51:10078–10081. https://doi.org/10.1002/anie.20120567
73. Lexa KW, Goh GB, Carlson HA (2014) Parameter Choice Matters: Validating Probe Parameters for Use in Mixed-Solvent Simulations. J Chem Inf Model 54:2190–2199. https://doi.org/10.1021/ci400741u
74. Defelipe L, Arcon J, Modenutti C et al (2018) Solvents to Fragments to Drugs: MD Applications in Drug Design. Molecules 23:3269. https://doi.org/10.3390/molecules23123269
75. English AC, Groom CR, Hubbard RE (2001) Experimental and computational mapping of the binding surface of a crystalline protein. Protein Eng Des Sel 14:47–59. https://doi.org/10.1093/protein/14.1.47
76. Lexa KW, Carlson HA (2013) Improving Protocols for Protein Mapping through Proper Comparison to Crystallography Data. J Chem Inf Model 53:391–402. https://doi.org/10.1021/ci300430v
77. English AC, Done SH, Caves LSD, et al (1999) Locating interaction sites on proteins: The crystal structure of thermolysin soaked in 2–100% isopropanol. Proteins Struct Funct Genet 37:628–640. https://doi.org/10.1002/(SICI)1097-0134(19991201)37:4<628::AID-PROT13>3.0.CO;2-G
78. Ho WC, Luo C, Zhao K et al (2006) High-resolution structure of the p53 core domain: implications for binding small-molecule stabilizing compounds. Acta Crystallogr Sect D Biol Crystallogr 62:1484–1493. https://doi.org/10.1107/S090744490603890X
79. Basse N, Kaar JL, Settanni G et al (2010) Toward the Rational Design of p53-Stabilizing Drugs: Probing the Surface of the Oncogenic Y220C Mutant. Chem Biol 17:46–56. https://doi.org/10.1016/j.chembiol.2009.12.011
80. Miranker A, Karplus M (1991) Functionality maps of binding sites: A multiple copy simultaneous search method. Proteins Struct Funct Genet 11:29–34. https://doi.org/10.1002/prot.340110104
81. Alvarez-Garcia D, Barril X (2014) Molecular Simulations with Solvent Competition Quantify Water Displaceability and Provide Accurate Interaction Maps of Protein Binding Sites. J Med Chem 57:8530–8539. https://doi.org/10.1021/jm5010418
82. Raman EP, Yu W, Guvench O, MacKerell AD (2011) Reproducing Crystal Binding Modes of Ligand Functional Groups Using Site-Identification by Ligand Competitive Saturation (SILCS) Simulations. J Chem Inf Model 51:877–896. https://doi.org/10.1021/ci100462t
83. Hall DH, Grove LE, Yueh C et al (2011) Robust Identification of Binding Hot Spots Using Continuum Electrostatics: Application to Hen Egg-White Lysozyme. J Am Chem Soc 133:20668–20671. https://doi.org/10.1021/ja207914y
84. Ghanakota P, Carlson HA (2016) Moving Beyond Active-Site Detection: MixMD Applied to Allosteric Systems. J Phys Chem B 120:8685–8695. https://doi.org/10.1021/acs.jpcb.6b03515
85. Magdziarz T, Mitusińska K, Gołdowska S et al (2017) AQUA-DUCT: a ligands tracking tool. Bioinformatics 33:2045–2046. https://doi.org/10.1093/bioinformatics/btx125
86. Magdziarz T, Mitusińska K, Bzówka M et al (2020) AQUA-DUCT 1.0: structural and functional analysis of macromolecules from an intramolecular voids perspective. Bioinformatics 36:2599–2601. https://doi.org/10.1093/bioinformatics/btz946
87. Berman HM (2000) The Protein Data Bank. Nucleic Acids Res 28:235–242. https://doi.org/10.1093/nar/28.1.235
88. Mitusińska K, Magdziarz T, Bzówka M et al (2018) Exploring Solanum tuberosum Epoxide Hydrolase Internal Architecture by Water Molecules Tracking. Biomolecules 8:143. https://doi.org/10.3390/biom8040143
89. Klausen MS, Jespersen MC, Nielsen H et al (2019) NetSurfP-2.0: Improved prediction of protein structural features by integrated deep learning. Proteins Struct Funct Bioinforma 87:520–527. https://doi.org/10.1002/prot.25674
90. Hopmann KH, Himo F (2006) Theoretical Study of the Full Reaction Mechanism of Human Soluble Epoxide Hydrolase. Chem - A Eur J 12:6898–6909. https://doi.org/10.1002/chem.200501519
91. Gora A, Brezovsky J, Damborsky J (2013) Gates of Enzymes. Chem Rev 113:5871–5923. https://doi.org/10.1021/cr300384w
92. Marques SM, Daniel L, Buryska T et al (2017) Enzyme Tunnels and Gates As Relevant Targets in Drug Design. Med Res Rev 37:1095–1139. https://doi.org/10.1002/med.21430
93. Huson DH, Scornavacca C (2012) Dendroscope 3: An Interactive Tool for Rooted Phylogenetic Trees and Networks. Syst Biol 61:1061–1067. https://doi.org/10.1093/sysbio/sys062
94. Gomez GA, Morisseau C, Hammock BD, Christianson DW (2006) Human soluble epoxide hydrolase: Structural basis of inhibition by 4- (3-cyclohexylureido) -carboxylic acids. 58–64. https://doi.org/10.1110/ps.051720206.58
95. Chen H, Zhang Y, Ye C et al (2014) Insight into the binding modes and inhibition mechanisms of adamantyl-based 1,3-disubstituted urea inhibitors in the active site of the human soluble epoxide hydrolase. J Biomol Struct Dyn 32:1231–1247. https://doi.org/10.1080/07391102.2013.812981
96. Gomez GA, Morisseau C, Hammock BD, Christianson DW (2004) Structure of Human Epoxide Hydrolase Reveals Mechanistic Inferences on Bifunctional Catalysis in Epoxide and Phosphate Ester Hydrolysis † ‡. Biochemistry 43:4716–4723. https://doi.org/10.1021/bi036189j
97. Lo HY, Man CC, Fleck RW et al (2010) Substituted pyrazoles as novel sEH antagonist: Investigation of key binding interactions within the catalytic domain. Bioorg Med Chem Lett 20:6379–6383. https://doi.org/10.1016/j.bmcl.2010.09.095
98. Takai K, Chiyo N, Nakajima T et al (2015) Three-dimensional rational approach to the discovery of potent substituted cyclopropyl urea soluble epoxide hydrolase inhibitors. Bioorg Med Chem Lett 25:1705–1708. https://doi.org/10.1016/j.bmcl.2015.02.076
99. Pravda L, Berka K, Svobodová Vařeková R et al (2014) Anatomy of enzyme channels. BMC Bioinformatics 15:379. https://doi.org/10.1186/s12859-014-0379-x
100. Garcin ED, Arvai AS, Rosenfeld RJ et al (2008) Anchored plasticity opens doors for selective inhibitor design in nitric oxide synthase. Nat Chem Biol 4:700–707. https://doi.org/10.1038/nchembio.115
101. Cheshire DR, Åberg A, Andersson GMK et al (2011) The discovery of novel, potent and highly selective inhibitors of inducible nitric oxide synthase (iNOS). Bioorg Med Chem Lett 21:2468–2471. https://doi.org/10.1016/j.bmcl.2011.02.061
102. Tuncbag N, Gursoy A, Keskin O (2009) Identification of computational hot spots in protein interfaces: combining solvent accessibility and inter-residue potentials improves the accuracy. Bioinformatics 25:1513–1520. https://doi.org/10.1093/bioinformatics/btp240
103. Zerbe BS, Hall DR, Vajda S et al (2012) Relationship between Hot Spot Residues and Ligand Binding Hot Spots in Protein–Protein Interfaces. J Chem Inf Model 52:2236–2244. https://doi.org/10.1021/ci300175u
104. Brenke R, Kozakov D, Chuang G-Y et al (2009) Fragment-based identification of druggable ‘hot spots’ of proteins using Fourier domain correlation techniques. Bioinformatics 25:621–627. https://doi.org/10.1093/bioinformatics/btp036
105. Kozakov D, Grove LE, Hall DR et al (2015) The FTMap family of web servers for determining and characterizing ligand-binding hot spots of proteins. Nat Protoc 10:733–755. https://doi.org/10.1038/nprot.2015.043
106. The PyMOL Molecular Graphics System, Version 2.0 Schrödinger; Schrödinger LLC.: New York, NY, USA
107. Gordon JC, Myers JB, Folta T et al (2005) H++: a server for estimating pKas and adding missing hydrogens to macromolecules. Nucleic Acids Res 33:W368–W371. https://doi.org/10.1093/nar/gki464
108. Luchko T, Gusarov S, Roe DR et al (2010) Three-Dimensional Molecular Theory of Solvation Coupled with Molecular Dynamics in Amber. J Chem Theory Comput 6:607–624. https://doi.org/10.1021/ct900460m
109. Sindhikara DJ, Yoshida N, Hirata F (2012) Placevent: An algorithm for prediction of explicit solvent atom distribution-Application to HIV-1 protease and F-ATP synthase. J Comput Chem 33:1536–1543. https://doi.org/10.1002/jcc.22984
110. Case DA, Babin V, Berryman JT et al (2014) AMBER 14
111. Pence HE, Williams A (2010) ChemSpider: An Online Chemical Information Resource. J Chem Educ 87:1123–1124. https://doi.org/10.1021/ed100697w
112. Nikitin AM, Lyubartsev AP (2007) New six-site acetonitrile model for simulations of liquid acetonitrile and its aqueous mixtures. J Comput Chem 28:2020–2026. https://doi.org/10.1002/jcc.20721
113. Wang J, Wang W, Kollman PA, Case DA (2006) Automatic atom type and bond type perception in molecular mechanical calculations. J Mol Graph Model 25:247–260. https://doi.org/10.1016/j.jmgm.2005.12.005
114. Gasteiger J, Marsili M (1980) Iterative partial equalization of orbital electronegativity—a rapid access to atomic charges. Tetrahedron 36:3219–3228. https://doi.org/10.1016/0040-4020(80)80168-2
115. Martínez L, Andrade R, Birgin EG, Martínez JM (2009) PACKMOL: A package for building initial configurations for molecular dynamics simulations. J Comput Chem 30:2157–2164. https://doi.org/10.1002/jcc.21224
116. Archelas A, Iacazio G, Kotik M (2016) Epoxide hydrolase and its application in organic synthesis. In: Green Biocatalysis
117. Wallace AC, Laskowski RA, Thornton JM (1995) LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions. Prot Eng 8:127–134. https://doi.org/10.1093/protein/8.2.127
118. McDonald IK, Thornton JM (1994) Satisfying Hydrogen Bonding Potential in Proteins. J Mol Biol 238:777–793. https://doi.org/10.1006/jmbi.1994.1334
119. Köhn H-F, Hubert LJ (2015) Hierarchical Cluster Analysis. In: Wiley StatsRef: Statistics Reference Online. John Wiley & Sons, Ltd, Chichester, pp 1–13
120. LEVANDOWSKY M, WINTER D (1971) Distance between Sets. Nature 234:34–35. https://doi.org/10.1038/234034a0
121. Müllner D (2011) Modern hierarchical, agglomerative clustering algorithms. ArXiv