Developing potential drugs for insomnia through computational analysis

Retrieving the drug models

The canonical smiles of Suvorexant, Ramelteon, Zolpidem and Triazolam were individually searched and copied from PubChem online library (RRID:SCR_004284). The Canonical smiles obtained were pasted into the SwissSimilarity search bar. The class of drugs was set as commercial. The combined, Zinc-Drug like was selected with combines 2D and 3D. This was used as a query to generate similar molecular structures from the ZINC database through ligand based virtual screening. The red search button of the page was clicked to allow the query to begin, which took a few minutes to generate the final results. The generated results were downloaded and saved as an excel file.

Retrieving and preparation of the retrieving the protein models

The receptors were obtained from the protein data bank GABA-A receptor, Orexin receptor, and melatonin receptor were individually searched on protein data bank (RRID:SCR_012820) search bar and downloaded using the ‘Download file’ option available on the Protein Data Bank. The receptors were saved in the project directory. The receptors were “Cleaned” by removing all non-standard compounds by clicking on the “Select” option on chimera, then selecting all non-standard compounds, and finally deleting using the ‘Actions’ option on chimera software (RRID: SCR_004097).

Ligand preparation

The first 20 canonical smiles of each parent drug compound, were pasted in the search bar at PubChem Sketcher and enter button clicked to allow the generation of a to a 2D molecular structure. The generated structure was saved as “All files” format by clicking the export button present in PubChem Sketcher to the Project folder in the desktop. PubChem Sketcher was refreshed before the next canonical smile was inserted for the next query.

The Avogadro software (RRID: SCR_011958) was opened. The software was used to open the 2D files which were generated from the project folder in the Desktop. The 2D structures of the analogs and the drug compounds were converted to 3D structures automatically, by opening them individually, and clicking “yes” on the prompt by Avogadro to generate a 3D model of the 2D structure that was being opened. The 3D generated structure was optimized by clicking the auto optimization option provided in Avogadro. The settings were set to MMF94s force field, and fixed atoms movable, on the Algorithm settings. The start button was clicked and the structure was allowed to form the most stable conformation. The most stable conformation was determined when the Auto optimization energy became constant. The resulting structure was saved as a.mol file to be used in the next step in chimera.

The optimized ligands were then opened and minimized (to add charges and hydrogen) using the Chimera software. This was done by clicking tools, then structure editing, using the steric only settings and then Gasteiger as the other residues, in the setting section. The resulting minimized structure was saved in the project folder to for the next process of docking

Docking

The minimized ligands and the cleaned receptor were opened simultaneously in Chimera. The docking process was initiated where the ligands were docked with their respective receptors using the chimera software, embedded with AutoDock Vina (RRID: SCR_011958). The docked compounds with better docking scores than the parent compounds were afterwards analyzed using the Biovia Discovery studio (RRID: SCR_01565) to determine the interaction between the receptor and the ligand using the 2D and 3D models of the complexes.

Determination of the pharmacokinetic properties

The canonical smiles of compound found to have a better docking score than the parent compounds were pasted on SwissADME online tool search box and queried to determine possible pharmacokinetic properties. The results were saved as an excel file. The results contained parameters such as Blood brain barrier permeability, the Lipinski’s violations, Gastrointestinal tract absorption, p-glycoprotein substrate, metabolism and synthesizability index which were used in the study.

Determination of the toxicity profile

The Protox online tool was used to predict the toxicities of the compounds. The canonical smiles of the parent drug compounds and the analogues were pasted individually on the search bar in the Protox server and queried for the toxicity profiles prediction. The parameters that were analyzed included the individual toxicities based on the LD50 were automatically generated with the Protox which was directly obtained from the search results.

Data presentation

Tables of the specific analogues (Table 1, Table 2, Table 3 and Table 4) were recorded showing various data obtained. The data include the docking scores, the pharmacokinetic profile, and the toxicities profile.

Suvorexant and its analogues

Suvorexant interacts with the orexin receptor through van der Waal forces hydrogen bonds, carbon-sulfur, Pi-sulfur, Pi-Pi alkyl, alkyl, and Pi alkyl (Figure 1). The increase in interaction forces between the receptor and the drug molecule increases the duration of activity of the drug in the receptor hence the activity of the drug and decreasing the concentration of the drugs needed to provide therapeutic effects.⁴

Figure 1. Suvorexant interaction with and orexin receptor.

Docking scores, pharmacokinetic and toxicity profiles of 20 compounds similar to suvorexant were analyzed. From the 20 molecules, 12 compounds (Table 1) showed better docking scores compared to the parent drug. All the molecules showed higher Gastrointestinal absorption indicating that the molecules can easily be administered orally.³⁴ The parent molecule had a poor penetration to the central nervous system (CNS) which is their site of action. five molecules showed higher CNS bioavailability as compared to the parent compound. They include ZINC000206774725, ZINC000206774070, ZINC000206774330, ZINC000206774469 and ZINC000253409181.

Suvorexant

ZINC000206774070 and ZINC000206774469 have good CNS penetration ability but those molecules are good substrates of P-glycoprotein. P-glycoprotein is a transporter that decreases the absorption of a drug by pumping the molecule back into the lumen. As a substrate of P-glycoprotein, the drug concentration may not be enough to cause the desired therapeutic effects.³¹

All the molecules do not have any Lipinski violation. A molecule that violates Lipinski’s rule of five indicates its limited bioavailability and oral absorption, these substances may need to be administered via other methods. In drug development, the rule is often applied.³⁵

ZINC000206774725, ZINC000206774330, and ZINC000253409181 are not a substrate of P-glycoprotein (Table 3). These molecules on the other hand have a greater Cytochrome P450 (CYP) enzyme interaction compared to the parent molecule. They all interact with CYP 1A2,2C9, 2C19 2D6,3A4 as an inhibitor. These enzymes are essential in the metabolism of many drugs.³⁶ Inhibition of those molecules may cause a serious interaction with other drugs which are administered to the patient. The molecules can all be synthesized. They all have a synthesizability index of above 3.2.

LD50 was used in the estimation of the molecule’s toxicity profile. This is a dose that is necessary to kill 50% of the population or test subjects. As a result, a higher LD50 value implies that a drug is less dangerous since a larger dose is required to kill half of the population or test subjects.³⁷ ZINC000206774725, ZINC000206774330, and ZINC000253409181 have a higher LD50 compared to the parent compound; suvorexant indicating that the molecules are safer compared to the parent molecule.

Ramelteon and its analogues

Docking scores, pharmacokinetic and toxicity profiles of Ramelteon interacts with the melatonin receptor by the van der Waal forces hydrogen bonds, alkyl, and Pi alkyl (Figure 2). Van der Waal forces are the weakest involved in the interaction while the hydrogen bonds are the strongest forces involved.³⁵

Figure 2. Ramelteon interaction with melatonin receptor type 1B.

Ramelteon

20 molecules similar to Ramelteon were analyzed. Thirteen compounds showed better docking scores compared to the parent molecule Ramelteon (Table 2). Out of the 13 compounds, two compounds had better pharmacokinetic properties compared to the parent compound. The two compounds ZINC000003825731, and ZINC000004222622 (Table 2) were affecting metabolism by affecting only the CYP2D6 enzyme compared to the parent compound affected which inhibits two enzymes CYP 2D6 and CYP 2C19.

Inhibition of the CYP enzymes may cause increased cases of drug-drug interactions associated with a particular drug. The parent drug interacts with more cytochrome enzymes compared to the ZINC000003825731, and ZINC000004222622 which only interact with one enzyme.

Ramelteon is a substrate for p-glycoprotein and hence may lead to the reduction of the concentration of the drugs in the CNS. P-glycoproteins proteins work to reduce the concentration thus reducing intracellular drug accumulation leading to treatment failure or increased concentration of the drugs used. The two molecules show fewer interactions with this protein which may improve the bioavailability after administration and at the same time decrease dose of administration of the drug and better clinical performance.

The molecules also have good oral bioavailability. They also pass and also cross the blood-brain barrier where they can execute their function.

They are both synthesizable. The synthesizability of ZINC000003825731 is 2.91 and ZINC000004222622 is 3.28 compared to that of 2.91 of Ramelteon. The synthesizability index of a molecule indicates how easy a drug is to be synthesized.

Triazolam and its analogues

Triazolam interacts with the GABA A receptor through van der Waals forces (Figure 3). These are forces that attract neutral molecules to one another in gases solids and liquids. The binding may be weaker than that of other forms of interactions, such as hydrogen bonding or electrostatic interactions. This may reduce the drug’s potency and at the same time increase the drug selectivity.³⁷

Figure 3. Triazolam receptor interaction and GABA A receptor.

Triazolam had a docking score of -8.1 (Table 3). Nine molecules had a better docking score compared to triazolam. Of the nine molecules, only one molecule, ZINC000000702142, has poor penetration to the CNS this indicates that the molecule may not be able to reach the site of action. All the molecules have good oral bioavailability. All the molecules are substrates of the P-glycoprotein. This indicated that the molecules have less bioavailability.³⁴

Triazolam

ZINC000000000903, ZINC000010152022, and ZINC000000347721 do not interact with the CYP 3A4. The CYP3A4 enzyme is involved in the metabolism of the majority of drugs³⁶ including Statins, Calcium channel blockers, Benzodiazepines, Macrolide Antibiotics, Antidepressants, Antipsychotics, Immunosuppressants, and Opioids. This makes it a potential for interaction with any drugs hence the need to be careful during co-administration. ZINC000000000903 interacts with only one enzyme CYP 1A2. This molecule has then very few interactions as compared to the other molecules.³⁶ These molecules studies have a higher LD50; ZINC000000000903(1080 mg/kg), ZINC000010152022(1080 mg/kg), ZINC000000347721(1080 mg/kg) compared to triazolam (695 mg/kg) (Table 3).

Zolpidem and its analogs

Zolpidem interacts with the GABA A receptor through hydrogen bonds and van der Waal forces (Figure 4). This provides a stronger receptor drug-receptor complex as compared to triazolam. interactions with, arginine, serine, methionine, and leucine at positions 5 1, and 3, and two interactions between proline and arginine at locations 49 and 50 and 6 and 7 respectively.

Figure 4. Zolpidem receptor interaction and GABA A receptor.

Zolpidem has a docking score of -8.0. 10 molecules had a better docking score compared to the parent molecule (Table 4). Of the 10 molecules ZINC000041288365 and ZINC000426519011 are substrates of P-glycoprotein transporters indicating that the rest of the molecules have a good oral bioavailability. ZINC000065743121 and ZINC000022054496 have very little interaction with CYP 2D6 molecule but interact with CYP 1A2,2C9, 2C19, and 3A4. The rest of the molecules interact with CYP 1A2,2C9, 2C19 2D6, and 3A4 as inhibitors. This increases the range of drugs the molecules can interact with during the administration.

Zolpidem

Zolpidem has an LD50 of 695 mg/kg. ZINC000022054496 has the highest LD50 value of 1190 mg/kg. ZINC000065743121 has an LD50 of 1000 mg/kg. which is better than the parent molecule. The two molecules can be given at high concentrations compared to zolpidem without the patient experiencing serious adverse effects.³⁸ GI absorption of the drugs of all the molecules is good indicating they can be issued orally. The molecules have no Lipinski violations with a synthesizability index of above 2.5. This is an indication that the drug molecules can be synthesized.

Source data

The canonical smiles of the Parent compounds (Ramelteon, Zolpidem, Suvorexant and Triazolam) were obtained from PubChem

Structural analogues of the Parent compounds were obtained from SwissSimilarity

The pharmacokinetic profiles of the structural analogues and the Parent compounds were obtained from SwissADME

Protein Receptors of the parent drugs compounds and their analogues were obtained from the Protein data bank

Toxicity profile of the parent compounds and their structural analogues were obtained from the Protox server

Underlying data

Figshare: Developing potential drugs for insomnia through computational analysis: https://doi.org/10.6084/m9.figshare.23674614.³⁹

This project contains the following underlying data

Extended data

Figshare: Developing potential drugs for insomnia through computational analysis: https://doi.org/10.6084/m9.figshare.23674614.³⁹

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).