Machine learning study of the extended drug-target interaction network informed by pain related voltage-gated sodium channels

ABSTRACT

Pain is a significant global health issue, and the current treatment options for pain management have limitations in terms of effectiveness, side effects, and potential for addiction. There is a pressing need for improved pain treatments and the development of new drugs. Voltage-gated sodium channels, particularly Nav1.3, Nav1.7, Nav1.8, and Nav1.9, play a crucial role in neuronal excitability and are predominantly expressed in the peripheral nervous system. Targeting these channels may provide a means to treat pain while minimizing central and cardiac adverse effects. In this study, we construct protein-protein interaction (PPI) networks based on pain-related sodium channels and develop a corresponding drug-target interaction network to identify potential lead compounds for pain management. To ensure reliable machine learning predictions, we carefully select 111 inhibitor data sets from a pool of more than 1000 targets in the PPI network. We employ 3 distinct machine learning algorithms combined with advanced natural language processing (NLP)-based embeddings, specifically pretrained transformer and autoencoder representations. Through a systematic screening process, we evaluate the side effects and repurposing potential of more than 150,000 drug candidates targeting Nav1.7 and Nav1.8 sodium channels. In addition, we assess the ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties of these candidates to identify leads with near-optimal characteristics. Our strategy provides an innovative platform for the pharmacological development of pain treatments, offering the potential for improved efficacy and reduced side effects.

Integrating virtual screening, pharmacoinformatics profiling, and molecular dynamics: identification of promising inhibitors targeting 3CLpro of SARS-CoV-2

Introduction: The pursuit of effective therapeutic solutions for SARS-CoV-2 infections and COVID-19 necessitates the repurposing of existing compounds. This study focuses on the detailed examination of the central protease, 3-chymotrypsin-like protease (3CLpro), a pivotal player in virus replication. The combined approach of molecular dynamics simulations and virtual screening is employed to identify potential inhibitors targeting 3CLpro. Methods: A comprehensive virtual screening of 7120 compounds sourced from diverse databases was conducted. Four promising inhibitors, namely EN1036, F6548-4084, F6548-1613, and PUBT44123754, were identified. These compounds exhibited notable attributes, including high binding affinity (ranging from -5.003 to -5.772 Kcal/mol) and superior Induced Fit Docking scores (ranging from -671.66 to -675.26 Kcal/mol) compared to co-crystallized ligands. Results: In-depth analysis revealed that F6548-1613 stood out, demonstrating stable hydrogen bonds with amino acids His41 and Thr62. Notably, F6548-1613 recorded a binding energy of -65.72 kcal/mol in Molecular Mechanics Generalized Born Surface Area (MMGBSA) simulations. These findings were supported by Molecular Dynamics simulations, highlighting the compound's efficacy in inhibiting 3CLpro. Discussion: The identified compounds, in compliance with Lipinski's rule of five and exhibiting functional molecular interactions with 3CLpro, present promising therapeutic prospects. The integration of in silico methodologies significantly expedites drug discovery, laying the foundation for subsequent experimental validation and optimization. This approach holds the potential to develop effective therapeutics for SARS-CoV-2.

Probing the potential of bioactive compounds of millets as an inhibitor for lifestyle diseases: molecular docking and simulation-based approach

Millets are becoming more popular as a healthy substitute for people with lifestyle disorders. They offer dietary fiber, polyphenols, fatty acids, minerals, vitamins, protein, and antioxidants. The nutritional importance of millets leads to the present in-silico study of selective bioactive compounds docked against the targets of lifestyle diseases, viz., diabetes, hypertension, and atherosclerosis using molecular docking and molecular simulations approach. Pharmacokinetic analysis was also carried out to analyse ADME properties and toxicity analysis, drug-likeliness, and finally target prediction for new targets for uncharacterized compounds or secondary targets for recognized molecules by Swiss Target Prediction was also done. The docking results revealed that the bioactive compound flavan-4-ol, among all the 50 compounds studied, best docked to all the four targets of lifestyle diseases, viz., Human dipeptidyl peptidase IV (-5.94 kcal mol-1 binding energy), Sodium-glucose cotransporter-2 (-6.49 kcal mol-1) diabetes-related enzyme, the Human angiotensin-converting enzyme (-6.31 kcal mol-1) which plays a significant role in hypertension, and Proprotein convertase subtilisin kexin type 9 (-4.67 kcal mol-1) for atherosclerosis. Molecular dynamics simulation analysis substantiates that the flavan-4-ol forms a better stability complex with all the targets. ADMET profiles further strengthened the candidature of the flavan-4-ol bioactive compound to be considered for trial as an inhibitor of targets DPPIV, SGLT2, PCSK9, and hACE. We suggest that more research be conducted, taking Flavon-4-ol into account where it can be used as standard treatment for lifestyle diseases.

Novel computational and drug design strategies for inhibition of monkeypox virus and Babesia microti: molecular docking, molecular dynamic simulation and drug design approach by natural compounds

Background

The alarming increase in tick-borne pathogens such as human Babesia microti is an existential threat to global public health. It is a protozoan parasitic infection transmitted by numerous species of the genus Babesia. Second, monkeypox has recently emerged as a public health crisis, and the virus has spread around the world in the post-COVID-19 period with a very rapid transmission rate. These two novel pathogens are a new concern for human health globally and have become a significant obstacle to the development of modern medicine and the economy of the whole world. Currently, there are no approved drugs for the treatment of this disease. So, this research gap encourages us to find a potential inhibitor from a natural source.

Methods and materials

In this study, a series of natural plant-based biomolecules were subjected to in-depth computational investigation to find the most potent inhibitors targeting major pathogenic proteins responsible for the diseases caused by these two pathogens.

Results

Among them, most of the selected natural compounds are predicted to bind tightly to the targeted proteins that are crucial for the replication of these novel pathogens. Moreover, all the molecules have outstanding ADMET properties such as high aqueous solubility, a higher human gastrointestinal absorption rate, and a lack of any carcinogenic or hepatotoxic effects; most of them followed Lipinski's rule. Finally, the stability of the compounds was determined by molecular dynamics simulations (MDs) for 100 ns. During MDs, we observed that the mentioned compounds have exceptional stability against selected pathogens.

Conclusion

These advanced computational strategies reported that 11 lead compounds, including dieckol and amentoflavone, exhibited high potency, excellent drug-like properties, and no toxicity. These compounds demonstrated strong binding affinities to the target enzymes, especially dieckol, which displayed superior stability during molecular dynamics simulations. The MM/PBSA method confirmed the favorable binding energies of amentoflavone and dieckol. However, further in vitro and in vivo studies are necessary to validate their efficacy. Our research highlights the role of Dieckol and Amentoflavone as promising candidates for inhibiting both monkeypox and Babesia microti, demonstrating their multifaceted roles in the control of these pathogens.

Insilico generation of novel ligands for the inhibition of SARS-CoV-2 main protease (3CLpro) using deep learning

The recent emergence of novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causing the coronavirus disease (COVID-19) has become a global public health crisis, and a crucial need exists for rapid identification and development of novel therapeutic interventions. In this study, a recurrent neural network (RNN) is trained and optimized to produce novel ligands that could serve as potential inhibitors to the SARS-CoV-2 viral protease: 3 chymotrypsin-like protease (3CL^pro). Structure-based virtual screening was performed through molecular docking, ADMET profiling, and predictions of various molecular properties were done to evaluate the toxicity and drug-likeness of the generated novel ligands. The properties of the generated ligands were also compared with current drugs under various phases of clinical trials to assess the efficacy of the novel ligands. Twenty novel ligands were selected that exhibited good drug-likeness properties, with most ligands conforming to Lipinski's rule of 5, high binding affinity (highest binding affinity: -9.4 kcal/mol), and promising ADMET profile. Additionally, the generated ligands complexed with 3CL^pro were found to be stable based on the results of molecular dynamics simulation studies conducted over a 100 ns period. Overall, the findings offer a promising avenue for the rapid identification and development of effective therapeutic interventions to treat COVID-19.

Corrigendum: Anti-viral drug discovery against monkeypox and smallpox infection by natural curcumin derivatives: a computational drug design approach

[This corrects the article DOI: 10.3389/fcimb.2023.1157627.].

Discover the Medication Potential of Algerian Medicinal Plants Against Sars-Cov-2 Main Protease (Mpro): Molecular Docking, Molecular Dynamic Simulation, and ADMET Analysis

At the end of 2019, the world faced a big challenge and crisis caused by the SARS-CoV-2 virus. It spreads rapidly and is contagious; no treatment has officially been found. Algeria has used medicinal plants native to the country to defend against this pandemic. The objective of this paper is based on a molecular docking study of the active compounds of five Algerian medicinal plants with their target Sars-2Cov-2 virus protease to assess their potential antiviral activity against COVID-19. Innovative software and computerized databases were introduced into the in-silico domain, mainly the Auto-Dock software version 1.5.6. Similar results were obtained for all ligands, with a better chemical affinity of − 5.600 kcal/mol for the protease target 6LU7 and − 5.700 kcal/mol for the protease target 6WTT, with an average of − 4.227 kcal/mol and − 4.221 kcal/mol, respectively. The protease targets 6LU7 and 6WTT. In the ADME-Tox study, the active compounds of Algerian medicinal plants also demonstrated an excellent pharmacokinetic and toxic profile. Best scores were noted for cedrol, camphor, and eucalyptol. A molecular dynamics simulation showed the stability of camphor-6LU7 and cedrol-6LU7 complexes, favoring the biological potential of white artemisia and cypress plants.

Mechanistic inhibition of Monkeypox and Marburg virus infection by O-rhamnosides and Kaempferol-o-rhamnosides derivatives: a new-fangled computational approach

The increasing incidence of Monkeypox virus (Mpox) and Marburg virus (MARV) infections worldwide presents a significant challenge to global health, as limited treatment options are currently available. This study investigates the potential of several O-rhamnosides and Kaempferol-O-rhamnosides as Mpox and MARV inhibitors using molecular modeling methods, including ADMET, molecular docking, and molecular dynamics/MD simulation. The effectiveness of these compounds against the viruses was assessed using the Prediction of Activity Spectra for Substances (PASS) prediction. The study's primary focus is molecular docking prediction, which demonstrated that ligands (L07, L08, and L09) bind to Mpox (PDB ID: 4QWO) and MARV (PDB ID: 4OR8) with binding affinities ranging from -8.00 kcal/mol to -9.5 kcal/mol. HOMO-LUMO based quantum calculations were employed to determine the HOMO-LUMO gap of frontier molecular orbitals (FMOs) and to estimate chemical potential, electronegativity, hardness, and softness. Drug similarity and ADMET prediction assessments of pharmacokinetic properties revealed that the compounds were likely non-carcinogenic, non-hepatotoxic, and rapidly soluble. Molecular dynamic (MD) modeling was used to identify the most favorable docked complexes involving bioactive chemicals. MD simulations indicate that varying types of kaempferol-O-rhamnoside are necessary for successful docking validation and maintaining the stability of the docked complex. These findings could facilitate the discovery of novel therapeutic agents for treating illnesses caused by the Mpox and MARV viruses.

Recent advances in functionalized quinoline scaffolds and hybrids-Exceptional pharmacophore in therapeutic medicine

Quinoline is one of the most common nitrogen-containing heterocycles owing to its fascinating pharmacological properties and synthetic value in organic and pharmaceutical chemistry. Functionalization of this moiety at different positions has allowed for varying pharmacological activities of its derivative. Several publications over the last few decades have specified various methods of synthesis. This includes classical methods of synthesizing the primary quinoline derivatives and efficient methods that reduce reaction time with increased yield employing procedures that fulfill one of the twelve green chemistry principles, "safer solvent". The metal nanoparticle-catalyzed reaction also serves as a potent and effective technique for the synthesis of quinoline with excellent atom efficiency. The primary focus of this review is to highlight the routes to synthesizing functionalized quinoline derivatives, including hybrids that have moieties with predetermined activities bound to the quinoline moiety which are of interest in synthesizing drug candidates with dual modes of action, overcoming toxicity, and resistance amongst others. This was achieved using updated literature, stating the biological activities and mechanisms through which these compounds administer relief. The ADMET studies and Structure-Activity Relationship (SAR) of novel derivatives were also highlighted to explore the drug-likeness of the quinoline-hybrids and the influence of substituent characteristics and position on the biological activity of the compounds.

Anti-viral drug discovery against monkeypox and smallpox infection by natural curcumin derivatives: A Computational drug design approach

Background

In the last couple of years, viral infections have been leading the globe, considered one of the most widespread and extremely damaging health problems and one of the leading causes of mortality in the modern period. Although several viral infections are discovered, such as SARS CoV-2, Langya Henipavirus, there have only been a limited number of discoveries of possible antiviral drug, and vaccine that have even received authorization for the protection of human health. Recently, another virial infection is infecting worldwide (Monkeypox, and Smallpox), which concerns pharmacists, biochemists, doctors, and healthcare providers about another epidemic. Also, currently no specific treatment is available against Monkeypox. This research gap encouraged us to develop a new molecule to fight against monkeypox and smallpox disease. So, firstly, fifty different curcumin derivatives were collected from natural sources, which are available in the PubChem database, to determine antiviral capabilities against Monkeypox and Smallpox.

Material and method

Preliminarily, the molecular docking experiment of fifty different curcumin derivatives were conducted, and the majority of the substances produced the expected binding affinities. Then, twelve curcumin derivatives were picked up for further analysis based on the maximum docking score. After that, the density functional theory (DFT) was used to determine chemical characterizations such as the highest occupied molecular orbital (HOMO), lowest unoccupied molecular orbital (LUMO), softness, and hardness, etc.

Results

The mentioned derivatives demonstrated docking scores greater than 6.80 kcal/mol, and the most significant binding affinity was at -8.90 kcal/mol, even though 12 molecules had higher binding scores (-8.00 kcal/mol to -8.9 kcal/mol), and better than the standard medications. The molecular dynamic simulation is described by root mean square deviation (RMSD) and root-mean-square fluctuation (RMSF), demonstrating that all the compounds might be stable in the physiological system.

Conclusion

In conclusion, each derivative of curcumin has outstanding absorption, distribution, metabolism, excretion, and toxicity (ADMET) characteristics. Hence, we recommended the aforementioned curcumin derivatives as potential antiviral agents for the treatment of Monkeypox and Smallpox virus, and more in vivo investigations are warranted to substantiate our findings.

Modified coptisine derivatives as an inhibitor against pathogenic Rhizomucor miehei, Mycolicibacterium smegmatis (Black Fungus), Monkeypox, and Marburg virus by molecular docking and molecular dynamics simulation-based drug design approach

During the second phase of SARS-CoV-2, an unknown fungal infection, identified as black fungus, was transmitted to numerous people among the hospitalized COVID-19 patients and increased the death rate. The black fungus is associated with the Mycolicibacterium smegmatis, Mucor lusitanicus, and Rhizomucor miehei microorganisms. At the same time, other pathogenic diseases, such as the Monkeypox virus and Marburg virus, impacted global health. Policymakers are concerned about these pathogens due to their severe pathogenic capabilities and rapid spread. However, no standard therapies are available to manage and treat those conditions. Since the coptisine has significant antimicrobial, antiviral, and antifungal properties; therefore, the current investigation has been designed by modifying coptisine to identify an effective drug molecule against Black fungus, Monkeypox, and Marburg virus. After designing the derivatives of coptisine, they have been optimized to get a stable molecular structure. These ligands were then subjected to molecular docking study against two vital proteins obtained from black fungal pathogens: Rhizomucor miehei (PDB ID: 4WTP) and Mycolicibacterium smegmatis (PDB ID 7D6X), and proteins found in Monkeypox virus (PDB ID: 4QWO) and Marburg virus (PDB ID 4OR8). Following molecular docking, other computational investigations, such as ADMET, QSAR, drug-likeness, quantum calculation and molecular dynamics, were also performed to determine their potentiality as antifungal and antiviral inhibitors. The docking score reported that they have strong affinities against Black fungus, Monkeypox virus, and Marburg virus. Then, the molecular dynamic simulation was conducted to determine their stability and durability in the physiological system with water at 100 ns, which documented that the mentioned drugs were stable over the simulated time. Thus, our in silico investigation provides a preliminary report that coptisine derivatives are safe and potentially effective against Black fungus, Monkeypox virus, and Marburg virus. Hence, coptisine derivatives may be a prospective candidate for developing drugs against Black fungus, Monkeypox and Marburg viruses.

Metals-triggered compound CDPDP exhibits anti-arthritic behavior by downregulating the inflammatory cytokines, and modulating the oxidative storm in mice models with extensive ADMET, docking and simulation studies

Triggering through abiotic stress, including chemical triggers like heavy metals, is a new technique for drug discovery. In this research, the effect of heavy metal Nickel on actinobacteria Streptomyces sp. SH-1327 to obtain a stress-derived compound was firstly investigated. A new compound cyclo-(D)-Pro-(D)-Phe (CDPDP) was triggered from the actinobacteria strain SH-1327 with the addition of nickel ions 1 mM. The stress compound was further evaluated for its anti-oxidant, analgesic, and anti-inflammatory activity against rheumatoid arthritis through in-vitro and in-vivo assays in albino mice. A remarkable in-vitro anti-oxidant potential of CDPDP was recorded with the IC₅₀ value of 30.06 ± 5.11 μg/ml in DPPH, IC₅₀ of 18.98 ± 2.91 against NO free radicals, the IC₅₀ value of 27.15 ± 3.12 against scavenging ability and IC₅₀ value of 28.40 ± 3.14 μg/ml for iron chelation capacity. Downregulation of pro-inflammatory mediators (NO and MDA), suppressed levels of pro-inflammatory cytokines (TNF-α, IL-6, IL-Iβ) and upregulation of expressions of anti-oxidant enzymes (GSH, catalase, and GST) unveiled its anti-inflammatory potential. CDPDP was analyzed in human chondrocyte cell line CHON-001 and the results demonstrated that CDPDP significantly increased cell survival, and inhibited apoptosis of IL-1β treated chondrocytes and IL-1β induced matrix degrading markers. In addition, to evaluate the mitochondrial fitness of CHON-001 cells, CDPDP significantly upregulated pgc1-α, the master regulator of mitochondrial biogenesis, indicating that CDPDP provides protective effects in CHON-001 cells. The absorption, distribution, metabolism, excretion, and toxicity (ADMET) profile of the CDPDP showed that CDPDP is safe in cases of hepatotoxicity, cardiotoxicity, and cytochrome inhibition. Furthermore, docking results showed good binding of CDPDP with IL-6-17.4 kcal/mol, and the simulation studies proved the stability between ligand and protein. Therefore, the findings of the current study prospect CDPDP as a potent anti-oxidant and a plausible anti-arthritic agent with a strong pharmacokinetic and pharmacological profile.

A Report on Multi-Target Anti-Inflammatory Properties of Phytoconstituents from Monochoria hastata (Family: Pontederiaceae)

This study aims to investigate the potential analgesic properties of the crude extract of Monochoria hastata (MH) leaves using in vivo experiments and in silico analysis. The extract, in a dose-dependent manner, exhibited a moderate analgesic property (~54% pain inhibition in acetic acid-induced writhing test), which is significant (** p < 0.001) as compared to the control group. The complex inflammatory mechanism involves diverse pathways and they are inter-connected. Therefore, multiple inflammatory modulator proteins were selected as the target for in silico analysis. Computational analysis suggests that all the selected targets had different degrees of interaction with the phytochemicals from the extract. Rutin (RU), protocatechuic acid (PA), vanillic acid (VA), and ferulic acid (FA) could regulate multiple targets with a robust efficiency. None of the compounds showed selectivity to Cyclooxygenase-2 (COX-2). However, regulation of COX and lipoxygenase (LOX) cascade by PA can reduce non-steroidal analgesic drugs (NSAIDs)-related side effects, including asthma. RU showed robust regulation of cytokine-mediated pathways like RAS/MAPK and PI3K/NF-kB by inhibition of EGFR and IKBα (IKK), which may prevent multi-organ failure due to cytokine storm in several microbial infections, for example, SARS-CoV-2. Further investigation, using in vivo and in vitro experiments, can be conducted to develop multi-target anti-inflammatory drugs using the isolated compounds from the extract.

Molecular docking and dynamics study to explore phytochemical ligand molecules against the main protease of SARS-CoV-2 from extensive phytochemical datasets

BACKGROUND

The high transmission and pathogenicity of SARS-CoV-2 has led to a pandemic that has halted the world's economy and health. The newly evolved strains and scarcity of vaccines has worsened the situation. The main protease (M^pro) of SARS-CoV-2 can act as a potential target due to its role in viral replication and conservation level.

METHODS

In this study, we have enlisted more than 1100 phytochemicals from Asian plants based on deep literature mining. The compounds library was screened against the M^pro of SARS-CoV-2.

RESULTS

The selected three ligands, Flemichin, Delta-Oleanolic acid, and Emodin 1-O-beta-D-glucoside had a binding energy of -8.9, -8.9, -8.7 KJ/mol respectively. The compounds bind to the active groove of the main protease at; Cys145, Glu166, His41, Met49, Pro168, Met165, Gln189. The multiple descriptors from the simulation study; root mean square deviation, root mean square fluctuation, radius of gyration, hydrogen bond, solvent accessible surface area confirms the stable nature of the protein-ligand complexes. Furthermore, post-md analysis confirms the rigidness in the docked poses over the simulation trajectories.

CONCLUSIONS

Our combinatorial drug design approaches may help researchers to identify suitable drug candidates against SARS-CoV-2.

Discovering Potential RNA Dependent RNA Polymerase Inhibitors as Prospective Drugs Against COVID-19: An in silico Approach

COVID-19, caused by Severe Acute Respiratory Syndrome Corona Virus 2, is declared a Global Pandemic by WHO in early 2020. In the present situation, though more than 180 vaccine candidates with some already approved for emergency use, are currently in development against SARS-CoV-2, their safety and efficacy data is still in a very preliminary stage to recognize them as a new treatment, which demands an utmost emergency for the development of an alternative anti-COVID-19 drug sine qua non for a COVID-19 free world. Since RNA-dependent RNA polymerase (RdRp) is an essential protein involved in replicating the virus, it can be held as a potential drug target. We were keen to explore the plant-based product against RdRp and analyze its inhibitory potential to treat COVID-19. A unique collection of 248 plant compounds were selected based on their antiviral activity published in previous literature and were subjected to molecular docking analysis against the catalytic sub-unit of RdRp. The docking study was followed by a pharmacokinetics analysis and molecular dynamics simulation study of the selected best-docked compounds. Tellimagrandin I, SaikosaponinB2, Hesperidin and (-)-Epigallocatechin Gallate were the most prominent ones that showed strong binding affinity toward RdRp. All the compounds mentioned showed satisfactory pharmacokinetics properties and remained stabilized at their respective binding sites during the Molecular dynamics simulation. Additionally, we calculated the free-binding energy/the binding properties of RdRp-ligand complexes with the connection of MM/GBSA. Interestingly, we observe that SaikosaponinB2 gives the best binding affinity (∆Gbinding = -42.43 kcal/mol) in the MM/GBSA assay. Whereas, least activity is observed for Hesperidin (∆Gbinding = -22.72 kcal/mol). Overall our study unveiled the feasibility of the SaikosaponinB2 to serve as potential molecules for developing an effective therapy against COVID-19 by inhibiting one of its most crucial replication proteins, RdRp.

Fr-PPIChem: An Academic Compound Library Dedicated to Protein-Protein Interactions

Protein-protein interactions (PPIs) mediate nearly every cellular process and represent attractive targets for modulating disease states but are challenging to target with small molecules. Despite this, several PPI inhibitors (iPPIs) have entered clinical trials, and a growing number of PPIs have become validated drug targets. However, high-throughput screening efforts still endure low hit rates mainly because of the use of unsuitable screening libraries. Here, we describe the collective effort of a French consortium to build, select, and store in plates a unique chemical library dedicated to the inhibition of PPIs. Using two independent predictive models and two updated databases of experimentally confirmed PPI inhibitors developed by members of the consortium, we built models based on different training sets, molecular descriptors, and machine learning methods. Independent statistical models were used to select putative PPI inhibitors from large commercial compound collections showing great complementarity. Medicinal chemistry filters were applied to remove undesirable structures from this set (such as PAINS, frequent hitters, and toxic compounds) and to improve drug likeness. The remaining compounds were subjected to a clustering procedure to reduce the final size of the library while maintaining its chemical diversity. In practice, the library showed a 46-fold activity rate enhancement when compared to a non-iPPI-enriched diversity library in high-throughput screening against the CD47-SIRPα PPI. The Fr-PPIChem library is plated in 384-well plates and will be distributed on demand to the scientific community as a powerful tool for discovering new chemical probes and early hits for the development of potential therapeutic drugs.

Fingerprint-based in silico models for the prediction of P-glycoprotein substrates and inhibitors

P-Glycoprotein (P-gp, ABCB1) plays a significant role in determining the ADMET properties of drugs and drug candidates. Substrates of P-gp are not only subject to multidrug resistance (MDR) in tumor therapy, they are also associated with poor pharmacokinetic profiles. In contrast, inhibitors of P-gp have been advocated as modulators of MDR. However, due to the polyspecificity of P-gp, knowledge on the molecular basis of ligand-transporter interaction is still poor, which renders the prediction of whether a compound is a P-gp substrate/non-substrate or an inhibitor/non-inhibitor quite challenging. In the present investigation, we used a set of fingerprints representing the presence/absence of various functional groups for machine learning based classification of a set of 484 substrates/non-substrates and a set of 1935 inhibitors/non-inhibitors. Best models were obtained using a combination of a wrapper subset evaluator (WSE) with random forest (RF), kappa nearest neighbor (kNN) and support vector machine (SVM), showing accuracies >70%. Best P-gp substrate models were further validated with three sets of external P-gp substrate sources, which include Drug Bank (n = 134), TP Search (n = 90) and a set compiled from literature (n = 76). Association rule analysis explores the various structural feature requirements for P-gp substrates and inhibitors.