A Bioinformatics Search for Selective Histamine H4 Receptor Antagonists Through Structure-Based Virtual Screening Strategies


Corresponding author: M. Xavier Suresh, bioxavi@hotmail.com


The prevalence of allergic disease is increasing dramatically in the developed world. Studies of allergic diseases have clearly demonstrated that histamine plays an important role in the pathogenesis of the early-phase allergic response. Histamine effects are mediated by H1, H2, H3, and H4 receptors. The presence of the histamine H4 receptors on leukocytes and mast cells suggests that the new histamine receptor H4 plays an important role in the modulation of the immune system. Thus, histamine H4 receptor is an attractive target for anti-allergic therapy. In our present study, we have generated a histamine H4 receptor model using I-TASSER based on human B2-adrenergic G-protein-coupled receptor. Structurally similar compounds of the three known antagonists JNJ777120, thioperamide, and Vuf6002 were retrieved from PubChem, and database was prepared. Virtual screening of those databases was performed, and six compounds with high docking score were identified. Also the binding mode revealed that all the six compounds had interaction with Asp94 of the receptor. Our results serve as a starting point in the development of novel lead compounds in anti-allergic therapy.

Histamine is one of the most important anciently known inflammatory mediators and explicitly studied molecules in biological systems (1). It is an important mediator of allergic responses in the airway, considering the positive correlation between its abundance and asthma severity (2). It exerts many physiological and pathological processes, and some of the best characterized roles of histamine are those in inflammation, gastric acid secretion, and as a neurotransmitter. The diverse biological effects of histamine are mediated through histamine receptors. These receptors are members of G-protein-coupled receptor (GPCR) superfamily, whereas to date, four subtypes (H1R, H2R, H3R, and H4R) have been identified. Histamine H4 receptor (H4R) is the novel member of the histamine receptor family. Earlier H1R antagonists have been used for asthma and other allergic diseases. Although H1R antagonists offer symptomatic relief of several physiological events in allergic rhinitis, such as edema and vasodilatation, they do not substantially ameliorate early- or late-phase bronchoconstrictive events in the asthmatic airway in response to allergen challenge. The discovery of this fourth histamine receptor, and the evidence that it is expressed on many cell types involved in allergic responses, suggested that the H4R may play an important role in mediating the histamine effects in asthma and other allergic diseases (3).

The H4R receptor is more widely distributed, especially in organs associated with the immune system (4–6). It is preferentially expressed in intestinal tissue, spleen, thymus, medullary cells, bone marrow, and peripheral hematopoietic cells, including eosinophils, basophils, mast cells, T lymphocytes, leukocytes, and dendritic cells (7,8). These cell types are primarily involved with the development and continuation of allergic responses. All the leukocyte chemo-attractant properties of histamine, that is, agonist-induced actin polymerization, mobilization of intracellular calcium, alteration in cell shape, and upregulation of adhesion molecule expression and the selective recruitment of eosinophils, are mediated by H4R subtype (8–10).

Much of the recent drug research in H4R field is focused on antagonists, mainly because of the prospect of new pharmacotherapies for the treatment for inflammatory diseases. H4R characterization clearly indicates the potential of this receptor as a novel drug target for treating allergy and inflammation. So, more effective search for potent and selective H4R antagonists is in progress to explore the therapeutic potential of such compounds (11).

Structure-based virtual screening (SBVS) is a widely accepted approach to identify new potential lead molecules from large databases of compounds that can subsequently synthesized and tested for their biological activities (12,13). The high-throughput screening led to the identification of the first H4R-specific antagonist JNJ7777120 (5-chloro-2-((4-methylpiperazin-1-yl)carbonyl)-1H-indole). On the basis of the structure of JNJ7777120, varieties of structurally diverse classes of compounds have also been developed (14). One such modification is Vuf6002 (JNJ10191584) having anti-inflammatory property (15). Another known antagonist is thioperamide, which is a high-affinity antagonist/inverse agonist for H3R and H4R (16). More recently, a large-scale SBVS study reported by Kiss et al. (17) yielded 16 compounds with significant H4R affinity that may serve as leads in the development of selective H4R ligands. The promising effect of these compounds both in vivo and in vitro has made researchers to move in a fast pace to design or screen more active compounds.

In this study, we have initially predicted the 3D structure of the human H4 receptor based on the crystal structure of human B2-adrenergic receptor. Molecular docking–based virtual screening of a precompiled ligand dataset was carried out against the H4R receptor mode using the ligand fit module of Discovery Studio version 2.0 (Accelrys, Inc., San Diego, CA, USA). Our study led to the identification of six compounds with high docking scores, which can be potent H4R antagonists.

Materials and Methods

Sequence analysis and topology prediction

The amino acid sequence of H4R (Q9H3N8) was downloaded from the Swissprot database. H4R is the novel member of the histamine receptor family that belongs to the GPCR super family. To determine the location of transmembrane helices, several tools were employed. TM prediction tools such as Transmembrane prediction (TMPRED) (18), Transmembrane hidden Markov model (TMHMM) (19), HMMTOP (20) and SOSUI (21) were used, and the results were compared with the structural model developed. The BLASTP was carried out against protein data bank (PDB) with BLOSUM62 matrix to search for the closely related proteins structures.

3-D structure prediction and model validation

We exploited I-TASSER, a web-based structure prediction server, to predict the tertiary structure of H4R protein. I-TASSER is an automated prediction tool that is based on the sequence to structure prediction paradigm (22–24). The H4R protein sequence was submitted to the server. On the basis of amino acid sequence, three-dimensional atomic models were generated from multiple threading alignment and iterative structural assembly simulations. The output from the server run contained full-length secondary and tertiary structure predictions, functional annotations on ligand-binding sites, Enzyme Commission numbers, and Gene Ontology terms. An estimate of accuracy of the predictions is provided based on the confidence score of the modeling (25). The models generated using I-TASSER were validated by procheck and ERRAT to make sure that it was of reasonable quality for analyzing ligand binding and to check their ability for virtual screening (26,27).

All the H4R models generated were validated by two independent tests, from which the best model was chosen for further studies. The first test assessed the stereochemical quality of the models by PROCHECK, which indicates the amino acid with unusual backbone conformation. Here, the Phi/Psi Ramachandran plots of the models were computed by comparing them to the preferred values of the protein in the protein data bank (26). The next test ERRAT calculated the non-bonded interactions between different atom types (27).

Identification of ligand-binding site

Identification of binding site is an important step in analyzing ligand binding interactions. The amino acid residues in the binding site were predicted with the help of binding pocket detection server tools, such as pocketfinder and Q-site finder (http://www.modelling.leeds.ac.uk/qsitefinder). In addition to that, the binding pockets of the receptor were also determined by using Accelrys Discovery Studio. It has been previously reported that Asp94 and Glu182 form part of the ligand-binding site and exhibit interactions with ligands (17,28), which in turn is most valuable in locating the exact binding site.

Preparation of ligand database

As a practical aspect in the preparation of the database, compounds that were commercially or academically available were preferred. The 2-D structure of the three antagonists JNJ7777120, thioperamide and Vuf 6002 were drawn and optimized with ChemSketch. Then the model and the ligands were subjected to energy minimization using the CHARMm forcefield implemented in the Discovery Studio software package (Accelrys, Inc., San Diego, CA, USA). PubChem, a free database of chemical and small molecule structures, was used for the database preparation (29). Each ligand was given as an input in the PubChem structure search as a structure file, and the output contained similar structure with the Tanimoto score of similarity >0.9 of the ligands in .sdf format. For JNJ777720, the similarity percentage used was 95%, whereas for the thioperamide and Vuf 6002, the similarity was kept as 90%. By this, 120 similar structures of JNJ7777120 were obtained. Similarly for thioperamide and Vuf6002, 49 and 168 similar structures were acquired, respectively. Thus, three different databases were prepared.

Screening of ligands

Docking calculations were performed by Discovery Studio version 2.0. Prior to docking, both the predicted H4R model and the structure of the compounds retrieved from the database were minimized to their low energy using CHARMm forcefield implemented in the Discovery Studio. The convergence gradient was set to 0.01 kcal/mol and 1000 steps of steepest descent algorithm followed by 50 000 or more steps of conjugate gradient algorithm. A spherical cutoff of 14 Å was used for non-bonding interactions, and other parameters were set as default. To validate the docking protocol, prior to screening, the known antagonists JNJ7777120, thioperamide, and Vuf6002 have been docked in the ligand-binding site and the results were compared with earlier results. Then the databases were subjected to virtual screening against the receptor model. Ligand fit module in the Discovery Studio is used for docking the compound databases. The LigandFit docking procedure consists of two major parts: (i) specifying the region of the receptor to use as the binding site for docking; (ii) docking the ligands to the specified site (30). The top 10 docked poses were allowed to be saved. The successful poses were evaluated using a set of scoring functions as implemented in Discovery Studio program including LigScore1, LigScore2, PLP1, PLP2, and PMF, whereas the candidate ligands in the binding site are prioritized according to the DockScore function.

ADMET prediction

Absorption, Distribution, Metabolism, Excretion and Toxicity (ADMET) describe the kinetics of drug exposure to the tissues and pharmacological activity of the compounds. ADMET properties of the compounds were tested using ADMET descriptors in Discovery Studio. Models of intestinal absorption, blood–brain barrier penetration, cytochrome P450 2D6 inhibition, and hepatotoxicity were tested for our compounds using ADMET descriptor module of Discovery Studio.

Molecular dynamics

Molecular dynamics (MD) simulations were performed using the simulation module in Discovery Studio with the standard CHARMm forcefield parameters. The six top scored receptor–ligand complexes were used for performing MD simulations. Implicit solvent with a distance dependent dielectric constant of 4*r was used. Temperature was maintained at 300 K, and 14 Å cutoff for non-bonding interactions were used. A total of 6 nanoseconds simulations were performed for the six complexes. The results were analyzed using the analysis protocol.

Results and Discussion

Topology prediction

The structural and visual analysis of the model 2 with Discovery Studio showed that TMH1 is present between 16 and 42, TMH2 between 51 and 75, TMH3 between 86 and 121, TMH4 between 130 and 151, TMH5 between 172 and 195, TMH6 between 305 and 329, and TMH7 between 339 and 369 (Figure 1), which is in accordance with the prediction by the servers shown in Table 1.

Figure 1.

 Transmembrane helices of model 2 predicted by SOSUI.

Table 1.   Prediction of transmembrane regions using various web servers
Helix 118–4220–4216–4219–41
Helix 251–7351–7352–7253–75
Helix 386–10988–11091–10999–121
Helix 4132–153130–152131–151130–151
Helix 5174–193172–194173–195173–195
Helix 6305–327307–329306–328306–328
Helix 7342–361339–369341–362341–362

Generation of H4R model

Structure modeling of HR4 has been a great concern since the first crystal structure of GPCR was determined. Many research groups have developed homology models for HR4 based on the crystal structure of bovine rhodopsin. But in this we have generated models by threading alignment tool I-TASSER. It generated five models based on the input sequence (Q9H3N8). I-TASSER output also contained top ten ranks of templates used for the structure prediction. The top template used is the high-resolution crystal structure of human B2-adrenergic GPCR (PDB ID: 2rh1A). The remaining templates used for threading are the isoforms of human B2-adrenergic GPCR. This remains contradictory to the blast search result that shows human adrenoreceptor (PDB ID: 2R4R A) as the closely similar protein with 23% identity. This suggests that human adrenoreceptor has the close sequence identity with the histamine H4R, whereas human B2-adrenergic receptor has close structural similarity (28%).

It is the first report where human GPCR is used as a template for developing the H4R model wherein previous studies have used the structure of bovine rhodopsin in structure prediction using homology modeling (17,28). Moreover, the output also contained functional annotations on ligand-binding sites, Enzyme Commission numbers, and Gene Ontology terms of the top models. Accuracy of the predictions was also provided based on the confidence score (C-score), which is an estimate of the confidence of structure prediction. C-score is typically in the range −5 to 2; a higher score reflects a model of better quality. In general, models with C-score >−1.5 are considered to have a correct fold. Model 1 has the highest score followed by model 2, which signifies the highest confidence than the other models as shown in Table 2.

Table 2.   Main geometric parameters of the model prediction and validation
Predicted modelsC scoreResults of Ramachandran’s plot by PROCHECKQuality factor by ERRAT
Core %Allowed %Generously allowed %Disallowed %
Model 1−0.5276.
Model 2−0.7477.618.62.21.796.597
Model 3−1.8472.622.43.31.789.527
Model 4−4.1572.920.
Model 5−569.323.53.63.690.052

Model validation

The geometry of the finally defined models was evaluated with Ramachandran’s plot calculation computed with the procheck program. The results of procheck analysis obtained for the models are collected in Table 2. The Phi/Psi distribution shows that 96.2% of the residues in the model 2 are in the most favored or allowed regions, which is high when compared to the other models (model 1: 93.9%, model 3: 95%, model 4: 93%, model 5: 92.8%). This high percentage of Phi/Psi angles in the allowed and the disallowed regions suggest model 2 to be superior to the others (Figure 2A,B).

Figure 2.

 (A) 3D structure of model 2 predicted by I-TASSER. (B) Ramachandran plot developed by procheck program for the model 2.

The second program used for the validation is ERRAT. Good high-resolution structures generally produce values around 95% or higher. For lower resolutions (2.5–3 Å), the average overall quality factor is around 91%. The overall quality factor of model 2 is 96.597, which is high when compared to other models. Thus, the analysis of the validation results allowed us to select model 2 for further analysis.

Binding site analysis of H4R

The ligand-binding sites were predicted with the help of binding pocket detection server tools such as pocket finder and Q-site finder (http://www.modelling.leeds.ac.uk/qsitefinder). In addition to that, the binding pockets of the receptor were also determined by using Discovery Studio. It has been previously reported that Asp94 and Glu182 form part of the ligand-binding site and exhibit interactions with ligands (17,28). This served as valuable evidence in locating the exact binding site. Moreover, based on the ligands bound in the crystal structure of human B2 adrenergic GPCR, the binding site residues Phe193, Tyr199, Ser103, Ser204, Asn393, Tyr308, Trp109, Asn312, Trp109, Asn312, Asp103, Val104, Phe289, Phe290, Thr118, Val117, Trp286, and Tyr316 were identified. As the H4R and B2 adrenergic GPCR shares structural similarity, the superimposition of these two structures revealed the corresponding amino acid residues in H4R. The corresponding amino acid residues in the model are Phe168, Ile174, Tyr340, Trp90, Phe344, Lys84, Tyr319, Ser320, Glu182, Thr88, Cys98, Trp316, and Trp348. It has been reported previously that JNJ777120 forms interactions with ASP94 and GLU182 and forms lipophilic interactions with Val64, Phe312, Trp316, Tyr319, and Trp348 (27). Among the 11 biding sites predicted by the ligand fit module of Discovery Studio, site 2 was found to posses most of the key residues that were identified in earlier analysis. Hence, site 2 is considered as the best binding site for further docking studies and is depicted in the following figure (Figure 3).

Figure 3.

 Predicted ligand-binding site.

Structure-based virtual screening

We performed the molecular docking using Discovery Studio against the compounds from PubChem. Docking was carried out separately for the three sets of databases prepared. A total of 150 JNJ7777120 analogues, 49 thioperamide analogues, and 193 Vuf6002 analogues were docked on to the H4R model. Among the 392 compounds, 378 compounds were successfully docked into the binding site. From each dataset, the top four compounds are inspected and the details are discussed below. Previously, molecular docking was performed for the identification of inverse antagonist for H1R using the same strategy, but in this, a pharmacophore model was set up to screen the ligands from three chemical databases namely Maybridge database, Chembridge database, and NCI2003 (31).

JNJ7777120 analogues

Molecular docking with the compound database of the analogues structures of JNJ7777120 revealed that 148 of 150 compounds were successfully docked, and the top four compounds with high dock scores are shown in the Table 3. Six aromatic acids formed the inner lining of the binding pocket. These amino acids give hydrophobic environment to the binding site and contribute to the lipophilic interactions through which the ligand stabilizes its position inside the binding pocket. Apart from that, Asp94 forms hydrogen bonding with the ligands. Compound A formed hydrogen bond interaction with Asp94 (2.18 Å) (Figure 4A) of the receptor, whereas other compounds did not reveal any such interaction. Compound A is similar to JNJ7777120 but lacks the methyl piperizine moiety. Moreover, it has only a slight modification of a compound that has been reported by Kiss et al. (17). The compound identified by them lacked the 5-chloro substituent on the indole ring and contained a simple ethylamine side chain, whereas our compound possessed both the simple ethylamine side chain and the chlorine substituents on the indole ring. This clearly highlights that modification of the structure of JNJ777120 can lead to the development of more potent H4R antagonists.

Table 3.   Scores of JNJ 7777120, thioperamide, Vuf 6002 analogue structures Thumbnail image of
Figure 4.

 Binding mode of (A) Compound A (B) Compound E (C) Compound F (D) Compound K (E) Compound L (F) Compound M at H4 H4R showing interactions with Asp94.

Thioperamide analogues

Similarly, in molecular docking of the database of thioperamide analogue structures, 42 of 49 compounds were successfully docked. Five compounds with high docking scores are given in the Table 3. Compound E, (4-(2-(1H-Imidazol-5-yl)-ethyl)-piperidinium), which has high docking score, only formed hydrogen bond interaction with Asp94 (2.35 Å) (Figure 4B) of the receptor model. An analogue of this compound is Immepip (4-(3H-imidazol-4-ylmethyl)piperidine) and Immethridine (4-(1H-Imidazol-4(5)-ylmethyl)pyridine). Immepip is a selective H3 agonist whereas immetherdine has been identified as a novel potent and highly selective H3 agonist with a 300-fold selectivity over the closely related H4R (32,33). The Compound F (4-(1H-imidazol-5-ylmethyl) piperidin-1-ium) has the next highest dock score and forms hydrogen bond interaction with Asp94 (2.33 Å) (Figure 4C). Its structure is similar to the Compound E with difference in the position of imidazole and methyl group (Figure 4B). Thus, our findings suggest that Compounds E and F can be an effective ligand for both H3R and H4R, but further in vivo studies have to be carried out to prove its efficacy.

Vuf 6002 analogues

In the third set of database containing Vuf 6002, the molecular docking results show that 193 compounds in a total of 198 compounds were successfully docked. Four compounds with high docking score are shown in Table 3. Hydrogen bond interactions were found with Compounds K, L, and M. These three compounds formed interaction with Asp94. (Figure 4D–F). Compound K (6-chloro-2-((1R)-1-piperazin-4-ium-1-ylethyl)-1H-benzimidazole), Compound L, and Compound M have the same core benzimidazole structure with modification in the ring structure. Hence, the modification at those positions can lead to the identification of a potent H4R ligand. The nitrogen in the imidazole ring has a possibility of forming hydrogen bond (4.00 Å) with the Phe168. Similarly, there are also C–C interaction in the central amino group of the ligand with Leu71 (3.73 Å) and Thr76 (4.86 Å). This attributes to the hydrophobic interlock that could lead to multifold selectivity of the ligand. Table 4 shows the six compounds with high docking scores.

Table 4.   The top six compounds of high docking score and their interactions with Asp94 Thumbnail image of

Summarizing the analysis of ligand binding, all the six compounds created an interaction with Asp94, preferably a hydrogen bond, situated at reasonably acceptable distance. This shows clearly that this residue plays crucial role in ligand binding to H4R. It is worthy of note that the compounds with highest scores in the respective databases are known to have similar structural core as that of the original compound. Moreover, all these compounds found to have interactions as listed here have not been tried earlier for any antihistaminic activity. Thus, these hits can form basis for lead development of novel antagonists. However, further invivo and in vitro studies have to be carried out to confirm its action as that of the original compound.

ADMET prediction

ADMET descriptor includes models that produce the characteristics of ADMET properties. The plot of polar surface area (PSA) versus AlogP is shown in Figure 5. Compound E and F are located out of the accepted region of ADMET in the plot which indicates that other compounds have good ADMET properties.

Figure 5.

 ADMET Plot of polar surface area versus AlogP.

Molecular dynamics

To ascertain the conformational variations of the H4R-ligand complexes, MD simulations have been carried out for the top scored receptor–ligand complexes. Analysis of the 1 nseconds MD trajectories for each complex structure reveals that the complexes were well stabilized at the active site. Figure 6 shows the total energy, kinetic energy, and potential energy profile over the period of simulations during production run. It can be noted that there is not much deviation indicates that the compounds bind in a better position. Further analyses reveals that the binding modes of the compounds established after the MD simulation are nearly the same as that obtained of molecular docking. The compounds are stabilized by intermolecular hydrogen bonds. The interactions with the residue Asp94 are seen throughout the dynamics trajectory of all the complexes illustrating its pivotal role in ligand binding at the active site (Table 4). Invariably in all the complexes, the interaction with the key residues are well coincided with the molecular docking studies and may provide guidance for the rational design of more potent H4R inhibitors.

Figure 6.

 The energy profile of the top six complexes obtained during 1 nseconds of MD simulations (A) Compound A; (B) Compound E; (C) Compound F; (D) Compound K; (E) Compound L; (F) Compound M.


H4R is an effective candidate in treating diseases associated with chronic pruritus and asthma. Structure-based virtual screening is applied in this study. Our work aimed at developing a homology model, and this H4R model can reliably be used in SBVS. A series of six potential leads of histamine H4R antagonists by virtual screening was identified. Molecular docking revealed the potential interactions of the filtered compounds and their affinity toward H4R. Our results also helped to find out novel scaffolds to design potent and selective H4R antagonist. Further, invitro functional assays have to be performed for the compound identified to prove its effectiveness.


The bioinformatics computational facilities available at Department of Bioinformatics, Sathyabama University, and the extreme support provided by the management of Sathyabama University (Mr. Marie Johnson & Ms. Mariazeena Johnson, Directors) are greatly acknowledged. The authors also thank the anonymous reviewers for their valuable comments and suggestions.