Ligand Binding Affinity Research Articles

Control of polymer-protein interactions by tuning the composition and length of polymer chains.

Nanomoduling the 3D shape and chemical functionalities in a synthetic polymer may create recognition cavities for biomacromolecule binding, which serves as an attractive alternative to natural antibodies with much less cost. To obtain fundamental understanding and predict molecular design rules of the polymer antibody, we analyze the complex structure between the biomarker protein epithelial cell adhesion molecule (EpCAM) and a series of polymer ligands via molecular dynamics (MD) simulations. For monomeric ligands, strong enrichment of aromatic residues in protein binding sites is revealed, in line with the reported observations for natural antibodies. Yet, for linear polymers with a growing degree of polymerization, for the first time, a drastic change is revealed on the type of enriched protein residues and the location of protein binding sites, driven by the increasing steric hindrance effect that makes the adsorption of the polymer in the protein exterior feasible. Varying the polymer length and monomeric composition also significantly affects the ligand binding affinity. Here, we have captured three distinct dependences of the ligand binding free energy on the degree of polymerization: for NIPAm based hydrophilic polymers, TBAm dominated hydrophobic polymers and AAc dominated charged polymers. These results can be rationalized by the complex structure and the composition of protein residues at the binding interface. The entire analysis demonstrates unique binding features for polymer ligands and the possibility to modulate their binding sites and affinity by engineering the polymer structure.

TrGPCR:GPCR-ligand Binding Affinity Predicting based on Dynamic Deep Transfer Learning.

Predicting G protein-coupled receptor (GPCR)-ligand binding affinity plays a crucial role in drug development. However, determining GPCR-ligand binding affinities is time-consuming and resource-intensive. Although many studies used data-driven methods to predict binding affinity, most of these methods required protein 3D structure, which was often unknown. Moreover, part of these studies only considered the sequence characteristics of the protein, ignoring the secondary structure of the protein. The number of known GPCR for affinity prediction is only a few thousand, which is insufficient for deep learning training. Therefore, this study aimed to propose a deep transfer learning method called TrGPCR, which used dynamic transfer learning to solve the problem of insufficient GPCR data. We used the Binding Database(BindingDB) as the source domain and the GLASS(GPCR-Ligand Association) database as the target domain. We also introduced protein secondary structures, called pockets, as features to predict binding affinities. Compared with DeepDTA, our model improved by 5.2% on RMSE(root mean square error) and 4.5% on MAE(mean squared error).

Alchemical approach performance in calculating the ligand-binding free energy.

Alchemical binding free energy calculations are one of the most accurate methods for estimating ligand-binding affinity. Assessing the accuracy of the approach over protein targets is one of the most interesting issues. The free energy difference of binding between a protein and a ligand was calculated via the alchemical approach. The alchemical approach exhibits satisfactory accuracy over four targets, including AmpC beta-lactamase (AmpC); glutamate receptor, ionotropic kainate 1 (GluK1); heat shock protein 90 (Hsp90); and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) main protease (Mpro). In particular, the correlation coefficients between calculated binding free energies and the respective experiments over four targets range from 0.56 to 0.86. The affinity computed via free energy perturbation (FEP) simulations is overestimated over the experimental value. Particularly, the electrostatic interaction free energy rules the binding process of ligands to AmpC and GluK1. However, the van der Waals (vdW) interaction free energy plays an important role in the ligand-binding processes of HSP90 and SARS-CoV-2 Mpro. The obtained results associate with the hydrophilic or hydrophobic properties of the ligands. This observation may enhance computer-aided drug design.

Exploration of Secondary Metabolites in Flower-Petal Annona muricata as Agonists for Peroxisome Proliferator-Activated Receptor-Alpha (PPARα) for Liver Function

The expression of PPARα in the liver is significantly increased in both non-alcoholic fatty liver disease (NAFLD) patients and experimental models. Animal studies have shown promising outcomes in improving histological conditions, such as fibrosis, through the use of PPARα agonists. This particular petal to act as agonists for PPARα. Molecular docking and Prime MM-GBSA (Molecular Mechanics-Generalized Born Surface Area) were employed to analyze the ligand binding affinity, atomistic interactions, and protein stability. Additionally, we conducted evaluations of the identified PPARα agonist candidates to assess their toxicity and pharmacological profiles were conducted. The hit compounds exhibit favourable binding affinity and thermodynamics stability, and interact effectively with key residues in the binding site. Furthermore, the safety assessment indicates minimal to non-acute toxicity and favourable drug-like properties for these compounds. Secondary metabolites in the extract are potential drug candidate. They demonstrate drug-like properties as they adhere to the Lipinski rule.

Improving protein-ligand docking results using the Semiempirical quantum mechanics: testing on the PDBbind 2016 core set

Molecular docking techniques are routinely employed for predicting ligand binding conformations and affinities in the in silico phase of the drug design and development process. In this study, a reliable semiempirical quantum mechanics (SQM) method, PM7, was employed for geometry optimization of top-ranked poses obtained from two widely used docking programs, AutoDock4 and AutoDock Vina. The PDBbind core set (version 2016), which contains high-quality crystal protein − ligand complexes with their corresponding experimental binding affinities, was used as an initial dataset in this research. It was shown that docking pose optimization improves the accuracy of pose predictions and is very useful for the refinement of docked complexes via removing clashes between ligands and proteins. It was also demonstrated that AutoDock Vina achieves a higher sampling power than AutoDock4 in generating accurate ligand poses (RMSD ≤ 2.0 Å), while AutoDock4 exhibits a better ranking power than AutoDock Vina. Finally, a new protocol based on a combination of the results obtained from the two docking programs was proposed for structure-based virtual screening studies, which benefits from the robust sampling abilities of AutoDock Vina and the reliable ranking performance of AutoDock4. Communicated by Ramaswamy H. Sarma

Structural Insights into the Activation of Human Aryl Hydrocarbon Receptor by the Environmental Contaminant Benzo[a]pyrene and Structurally Related Compounds

The aryl hydrocarbon receptor (AHR) is a ligand-dependent transcription factor belonging to the bHLH/PAS protein family and responding to hundreds of natural and chemical substances. It is primarily involved in the defense against chemical insults and bacterial infections or in the adaptive immune response, but also in the development of pathological conditions ranging from inflammatory to neoplastic disorders. Despite its prominent roles in many (patho)physiological processes, the lack of high-resolution structural data has precluded for thirty years an in-depth understanding of the structural mechanisms underlying ligand-binding specificity, promiscuity and activation of AHR. We recently reported a cryogenic electron microscopy (cryo-EM) structure of human AHR bound to the natural ligand indirubin, the chaperone Hsp90 and the co-chaperone XAP2 that provided the first experimental visualization of its ligand-binding PAS-B domain. Here, we report a 2.75 Å resolution structure of the AHR complex bound to the environmental pollutant benzo[a]pyrene (B[a]P). The structure substantiates the existence of a bipartite PAS-B ligand-binding pocket with a geometrically constrained primary binding site controlling ligand binding specificity and affinity, and a secondary binding site contributing to the binding promiscuity of AHR. We also report a docking study of B[a]P congeners that validates the B[a]P-bound PAS-B structure as a suitable model for accurate computational ligand binding assessment. Finally, comparison of our agonist-bound complex with the recently reported structures of mouse and fruit fly AHR PAS-B in different activation states suggests a ligand-induced loop conformational change potentially involved in the regulation of AHR function.

Machine learning models to predict ligand binding affinity for the orexin 1 receptor

The orexin 1 receptor (OX1R) is a G-protein coupled receptor that regulates a variety of physiological processes through interactions with the neuropeptides orexin A and B. Selective OX1R antagonists exhibit therapeutic effects in preclinical models of several behavioral disorders, including drug seeking and overeating. However, currently there are no selective OX1R antagonists approved for clinical use, fueling demand for novel compounds that act at this target. In this study, we meticulously curated a dataset comprising over 1300 OX1R ligands using a stringent filter and criteria cascade. Subsequently, we developed highly predictive quantitative structure-activity relationship (QSAR) models employing the optimized hyper-parameters for the random forest machine learning algorithm and twelve 2D molecular descriptors selected by recursive feature elimination with a 5-fold cross-validation process. The predictive capacity of the QSAR model was further assessed using an external test set and enrichment study, confirming its high predictivity. The practical applicability of our final QSAR model was demonstrated through virtual screening of the DrugBank database. This revealed two FDA-approved drugs (isavuconazole and cabozantinib) as potential OX1R ligands, confirmed by radiolabeled OX1R binding assays. To our best knowledge, this study represents the first report of highly predictive QSAR models on a large comprehensive dataset of diverse OX1R ligands, which should prove useful for the discovery and design of new compounds targeting this receptor.

Assessment of Quercetin Antiemetic Properties: In Vivo and In Silico Investigations on Receptor Binding Affinity and Synergistic Effects.

Quercetin (QUA), a flavonoid compound, is ubiquitously found in plants and has demonstrated a diverse range of biological activities. The primary objective of the current study is to assess the potential antiemetic properties of QUA using an in vivo and in silico approach. In this experiment, 4-day-old chicks were purchased to induce emesis by orally administering copper sulfate pentahydrate (CuSO4·5H2O) at a dose of 50 mg/kg (orally). Domperidone (DOM) (6 mg/kg), Hyoscine (HYS) (21 mg/kg), and Ondansetron (OND) (5 mg/kg) were treated as positive controls (PCs), and distilled water and a trace amount of Tween 80 mixture was employed as a negative control (NC). QUA was given orally at two distinct doses (25 and 50 mg/kg). Additionally, QUA (50 mg/kg) and PCs were administered separately or in combination to assess their antagonistic or synergistic effects on the chicks. The binding affinity of QUA and referral ligands towards the serotonin receptor (5HT3), dopamine receptors (D2 and D3), and muscarinic acetylcholine receptors (M1-M5) were estimated, and ligand-receptor interactions were visualized through various computational tools. In vivo findings indicate that QUA (25 and 50 mg/kg) has a significant effect on reducing the number of retches (16.50 ± 4.65 and 10.00 ± 4.19 times) and increasing the chick latency period (59.25 ± 4.75 and 94.25 ± 4.01 s), respectively. Additionally, QUA (50 mg/kg) in combination with Domperidone and Ondansetron exhibited superior antiemetic effects, reducing the number of retches and increasing the onset of emesis-inducing time. Furthermore, it is worth noting that QUA exhibited the strongest binding affinity against the D2 receptor with a value of -9.7 kcal/mol through the formation of hydrogen and hydrophobic bonds. In summary, the study found that QUA exhibited antiemetic activity in chicks, potentially by interacting with the D2 receptor pathway.

Virtual Screening, Pharmacokinetic Prediction, Molecular Docking and Dynamics Approaches in the Search for Selective and Potent Natural Molecular Inhibitors of MAO-B for the Treatment of Neurodegenerative Diseases

This research aims to find natural product compounds that have the potential to act as MAO-B inhibitors that are useful in the treatment of neurodegenerative diseases, through the stages: a) Virtual Screening, b) Molecule Docking, c) Pharmacokinetic and toxicity prediction, and d) Simulation approach Molecular dynamics.The research steps include the following steps: a) searching for molecules in the ZINC15 data base that are similar to the natural ligand molecule (safinamide) obtained from the protein data bank (PDB code: 2v5z) and the control ligand L-DOPA. A total of 481 molecules were downloaded from the data base and then molecular docking was carried out using the autodock program on the MAO-B target in the 2v5z receptor. After carrying out the docking analysis, 48 ligand molecules were selected which had a binding affinity (DG/kcal/mol) that was smaller than the DG of the natural ligand and the control ligand and nine (9) ligand molecules were taken to be tested: (i) ligand-ligand interactions MAO-B with discovery studio, (ii) Absorption, Distribution, Metabolism and Excretion properties with SwissADME and (iii) toxicity using PROTOX-II. Molecular dynamics simulations were carried out to determine the stability of ligands in proteins. Ligand complex [CC(C)c1ccc(NC(=O)Cn2cnc3c2c(=O)n(C)c(=O)n3C)cc1]-2v5z was chosen to simulate for 100 ps. The results of the molecular docking study showed that there were 9 molecules that had binding affinity values that were smaller than the binding affinity of the natural ligand and the control ligand. Ligand and residue interactions are dominated by hydrogen bonds, donor-donor and pi-pi stacked interactions. Based on SwissADME, the Blood Brain Barrier (BBB) permeant on ligand number 1, 2, 3, 5, 6, and 9 shows that it is orally active and cannot pass through the BBB and will not cause any side effects, whereas ligand number 4, 7, 8, 10, and 11 can cross the BBB and may cause side effects. Based on the results of toxicity prediction (PROTOX-II), it is known that there are four (4) ligands in class V, five (5) ligands in class IV and the rest in class II. Hepatotoxicity, carcinogenicity, and Phosphoprotein (Tumor Suppressor) p53 in eleven ligands are predicted to be inactive and have a small probability. The stimulated [CC(C)c1ccc(NC(=O)Cn2cnc3c2c(=O)n(C)c(=O)n3C)cc1]-2v5z complex apparently still shows ligand-protein fluctuations so that its conformation is still unstable.

AutoDock Koto: A Gradient Boosting Differential Evolution for Molecular Docking

Molecular docking plays a vital role in modern drug discovery, by supporting predictions of the binding modes and affinities of ligands at the binding site of target proteins. Several docking programs have been developed for both commercial and academic applications. Typically, a docking program’s performance depends on the sampling algorithm used to generate the ligand’s potential conformations and the scoring function applied to evaluate and rank these conformations. Evolutionary algorithms are widely used as sampling algorithms in docking programs. However, both the linkage problem and the dimensionality degenerate the search ability of evolutionary algorithms in the docking process. Therefore, a newly designed docking program named AutoDock Koto was developed in this study, which adopts a novel gradient boosting differential evolution algorithm to effectively address these issues. Experimental results show that compared with commonly used docking programs, AutoDock Koto yields dramatic improvements in docking performance based on an extensive dataset of 285 protein-ligand complexes. In addition, due to its strong docking ability, AutoDock Koto was used to identify potential drugs for COVID-19 based on a virtual screening of all approved drugs in our experiments. Sixteen drugs are found to possess low binding energy to the main target protease of SARS-CoV-2, and thus have the potential to treat COVID-19 as antiviral drugs. The source code of AutoDock Koto can be downloaded for free from. https://github.com/codezhouj/Molecular_Docking.

Tubulin-gene Mutation in Drug Resistance in Helminth Parasite: Docking and Molecular Dynamics Simulation Study

Background: Drug resistance is an important phenomenon in helminth parasites. Microtubules are among the key chemotherapeutic targets, mutations of which lead to drug resistance. Objectives: The present study investigated the role of F167Y, E198A, and F200Y mutations in β- tubulin protein and their effect on albendazole binding. Methods: Brugia malayi β-tubulin protein models were generated using the SwissModel platform by submitting amino acid sequences. Mutations were carried out at amino acid sequences by changing F167Y, E198A, and F200Y. All the model proteins (one wild and three mutated) were docked with the anthelmintic drug albendazole using AutoDock vina-1.1.5. Docking complexes were further investigated for their binding stability by a Molecular Dynamic Simulation study using Gromacs-2023.2. The binding free energies of protein-ligand complexes were analyzed using the MM/PBSA package. Results: The docking study observed decreased ligand binding affinity in F167Y and E198A mutant proteins compared to wild proteins. MD simulation revealed the overall structural stability of the protein complexes during the simulation period. The simulation also observed more stable binding of albendazole in the active pocket of mutant proteins compared to wild-type proteins. Like ligand RMSD, wild-type protein also showed higher amino acid residual flexibility. The flexibility indicates the less compactness of wild β-tubulin protein complexes compared to mutant proteinligand complexes. Van der Waals and electrostatic interactions were found to be the major energy in protein-ligand complexes. However, due to higher solvation energy, wild-type protein showed more flexibility compared to others. Conclusion: The study, therefore, concludes that mutations at positions 167 and 198 of the β- tubulin protein contribute to resistance to albendazole through weakened binding affinity. However, the binding of albendazole binding to the proteins leads to structures becoming more stable and compact.

The evolutionary divergence of receptor guanylyl cyclase C has implications for preclinical models for receptor-directed therapeutics

Mutations in receptor guanylyl cyclase C (GC-C) cause severe gastrointestinal disease, including meconium ileus, early onset acute diarrhea, and pediatric inflammatory bowel disease that continues into adulthood. Agonists of GC-C are FDA-approved drugs for the treatment of constipation and irritable bowel syndrome. Therapeutic strategies targeting GC-C are tested in preclinical mouse models assuming that murine GC-C mimics human GC-C in its biochemical properties and downstream signaling events. Here, we reveal important differences in ligand binding affinity and guanylyl cyclase activity between mouse GC-C and human GC-C. We generated a series of chimeric constructs of various domains of human and mouse GC-C to show that the extracellular domain of mouse GC-C contributed to log-orders lower affinity of mouse GC-C for ligands when compared to human GC-C. Further, the Vmax of the murine guanylyl cyclase domain was lower than that of human GC-C, and allosteric regulation of the receptor by ATP binding to the intracellular kinase-homology domain also differed. These altered properties are reflected in the high concentrations of ligands required to elicit signaling responses in the mouse gut in preclinical models and the specificity of a guanylyl cyclase inhibitor towards human GC-C. Therefore, our studies identify considerations in using the murine model to test molecules for therapeutic purposes that work as either agonists or antagonists of GC-C, and vaccines for the bacterial heat-stable enterotoxin that causes watery diarrhea in humans.

Unveiling the anticancer mechanism of 1,2,3-triazole-incorporated thiazole-pyrimidine-isoxazoles: insights from docking and molecular dynamics simulations

Cancer is a major global health concern, and the constant search for novel, selective anticancer compounds with low toxicity is never ending. Nitrogen heterocyclic compounds such as pyrimidine and triazole have been identified as potential candidates for cancer treatment. A novel series of 1,2,3-triazole incorporated thiazole-pyrimidine-isoxazole derivatives 10 (a-j) were designed, synthesized and evaluated for antitumorigenic activities against human breast cancer (MCF-7), human lung cancer (A549) and human prostate (PC3 & DU-145) various cell-lines by employing MTT assay using etoposide as the positive control. The synthesized hybrids yielded decent efficacy, which was further compared with the standard drug. Among all the molecules, 10h revealed the more potent anticancerous activities, having IC50 values ranging from 0.011 ± 0.0017 µM; 0.063 ± 0.0012 µM; 0.017 ± 0.0094 µM and 0.66 ± 0.072 µM with DU145, PC3, A549, and MCF7 cell-lines, respectively. Tubulin, being a major protein involved with diverse biological actions, also serves, as a crucial target for several clinically practiced anticancer drugs, was utilized for docking analyses to evaluate the binding affinity of ligands. Docking results demonstrates that the selected hybrids 10 (g-j) exhibited good binding affinities with protein. Subsequently, drug likeness studies were carried out on the synthesized compounds to evaluate and analyze their drug like properties such as absorption, distribution, metabolism, excretion, and toxicity (ADMET) for toxicity prediction. Based on these analyses, the selected complexes were further employed for molecular dynamic simulations to analyze stability via an exhaustive cumulative 200 nanoseconds simulation. These results suggest that the selected compounds are stable and might serve as potential inhibitors to tubulin complex. In conclusion, we propose these synthesized compounds 10 (g-j) might provide new insights into cancer treatment and have potential for future development. Communicated by Ramaswamy H. Sarma

Molecular insights into disease-associated glutamate transporter (EAAT1 / SLC1A3) variants using in silico and in vitro approaches

Glutamate is an essential excitatory neurotransmitter and an intermediate for energy metabolism. Depending on the tumor site, cancer cells have increased or decreased expression of excitatory amino acid transporter 1 or 2 (EAAT1/2, SLC1A3/2) to regulate glutamate uptake for the benefit of tumor growth. Thus, EAAT1/2 may be an attractive target for therapeutic intervention in oncology. Genetic variation of EAAT1 has been associated with rare cases of episodic ataxia, but the occurrence and functional contribution of EAAT1 mutants in other diseases, such as cancer, is poorly understood. Here, 105 unique somatic EAAT1 mutations were identified in cancer patients from the Genomic Data Commons dataset. Using EAAT1 crystal structures and in silico studies, eight mutations were selected based on their close proximity to the orthosteric or allosteric ligand binding sites and the predicted change in ligand binding affinity. In vitro functional assessment in a live-cell, impedance-based phenotypic assay demonstrated that these mutants differentially affect L-glutamate and L-aspartate transport, as well as the inhibitory potency of an orthosteric (TFB-TBOA) and allosteric (UCPH-101) inhibitor. Moreover, two episodic ataxia-related mutants displayed functional responses that were in line with literature, which confirmed the validity of our assay. Of note, ataxia-related mutant M128R displayed inhibitor-induced functional responses never described before. Finally, molecular dynamics (MD) simulations were performed to gain mechanistic insights into the observed functional effects. Taken together, the results in this work demonstrate 1) the suitability of the label-free phenotypic method to assess functional variation of EAAT1 mutants and 2) the opportunity and challenges of using in silico techniques to rationalize the in vitro phenotype of disease-relevant mutants.

Fluorescent Tools for the Imaging of Dopamine D2 -Like Receptors.

The family of dopamine D2 -like receptors represents an interesting target for a variety of neurological diseases, e. g. Parkinson's disease (PD), addiction, or schizophrenia. In this study we describe the synthesis of a new set of fluorescent ligands as tools for visualization of dopamine D2 -like receptors. Pharmacological characterization in radioligand binding studies identified UR-MN212 (20) as a high-affinity ligand for D2 -like receptors (pKi (D2long R)=8.24, pKi (D3 R)=8.58, pKi (D4 R)=7.78) with decent selectivity towards D1 -like receptors. Compound 20 is a neutral antagonist in a Go1 activation assay at the D2long R, D3 R, and D4 R, which is an important feature for studies using whole cells. The neutral antagonist 20, equipped with a 5-TAMRA dye, displayed rapid association to the D2long R in binding studies using confocal microscopy demonstrating its suitability for fluorescence microscopy. Furthermore, in molecular brightness studies, the ligand's binding affinity could be determined in a single-digit nanomolar range that was in good agreement with radioligand binding data. Therefore, the fluorescent compound can be used for quantitative characterization of native D2 -like receptors in a broad variety of experimental setups.

β-adrenergic receptor regulates embryonic epithelial extensibility through actomyosin inhibition

During morphogenesis, epithelial tissues reshape and expand to cover the body and organs. The molecular mechanisms of this deformability remain elusive. Here, we investigate the role of the β-adrenergic receptor (ADRB) in orchestrating actomyosin contractility, pivotal for epithelial extensibility. Chemical screens on Xenopus laevis embryos pinpointed ADRB2 as a principal regulator. ADRB2 promotes actomyosin relaxation, facilitating apical cell area expansion during body elongation. In contrast, ADRB2 knockdown results in heightened cell contraction, marked by synchronous oscillation of F-actin and myosin, impeding body elongation. ADRB2 mutants with reduced affinity for ligand binding lack the function to induce cellular relaxation, highlighting the ligand's essential roles even in the developing epidermis. Our findings unveil ADRB2's critical contribution to extensibility of the epidermis and subsequent body elongation during development. This study also offers insights into the physiology of mature epithelial organs deformed by the smooth muscle response to the adrenergic autonomic nervous system.

Comprehensive elucidation of experimental results of polytype ZnS‐10H quantum dots produced in single stride

Polytypes of a material exhibit different structures (nanosheets, nanorods, nanospheres etc.), leading to distinct properties pertinent to their respective fields. Theoretical studies using computational methods have elaborated their properties and essence; however, only a few studies were presented experimentally. Ab-initio synthesis of wurtzite ZnS-10 H polytype was carried out and an extensive study was done to unfold the properties to manoeuvre in different fields. Being a potential material for quantum dot sensitized solar cells (QDSSC), ZnS-10 H was synthesized at 70 °C to enhance the properties and in conjunction with the sustainability and durability of the device. Wurtzite crystal structure of polytype ZnS-10 H was confirmed from the studies. Capping agent manifests its role by passivating the material’s surface and the efficacy of the capping agent depends on the binding affinity of ligands and the concentration of the capping agent. Photoluminescence (PL) of uncapped ZnS-10 H shows its luminescence similar to ZnS, and the effect of 2-mercaptoethanol (2-ME) is obtained with 2-ME variations, after a limit (i.e., at 6 mL of capping agent) core-shell structure formation was seen with no emission and no excitation indicating the light capturing ability. Current findings suggest the possible applications of polytype ZnS-10 H in QDSSC, photovoltaic devices and photocatalytic activities.

Free energies at QM accuracy from force fields via multimap targeted estimation.

Accurate predictions of ligand binding affinities would greatly accelerate the first stages of drug discovery campaigns. However, using highly accurate interatomic potentials based on quantum mechanics (QM) in free energy methods has been so far largely unfeasible due to their prohibitive computational cost. Here, we present an efficient method to compute QM free energies from simulations using cheap reference potentials, such as force fields (FFs). This task has traditionally been out of reach due to the slow convergence of computing the correction from the FF to the QM potential. To overcome this bottleneck, we generalize targeted free energy methods to employ multiple maps-implemented with normalizing flow neural networks (NNs)-that maximize the overlap between the distributions. Critically, the method requires neither a separate expensive training phase for the NNs nor samples from the QM potential. We further propose a one-epoch learning policy to efficiently avoid overfitting, and we combine our approach with enhanced sampling strategies to overcome the pervasive problem of poor convergence due to slow degrees of freedom. On the drug-like molecules in the HiPen dataset, the method accelerates the calculation of the free energy difference of switching from an FF to a DFTB3 potential by three orders of magnitude compared to standard free energy perturbation and by a factor of eight compared to previously published nonequilibrium calculations. Our results suggest that our method, in combination with efficient QM/MM calculations, may be used in lead optimization campaigns in drug discovery and to study protein-ligand molecular recognition processes.

Exploring the composition of protein-ligand binding sites for cancerous inhibitor of PP2A (CIP2A) by inhibitor guided binding analysis: paving a new way for the Discovery of drug candidates against triple negative breast cancer (TNBC)

Triple-negative breast cancer (TNBC) is associated with high-grade invasive carcinoma leading to a 10% to 15% death rate in younger premenopausal women. Targeting cancerous inhibitors of protein phosphatase (CIP2A) has been a highly effective approach for exploring therapeutic drug candidates. Lapatinib, a dual tyrosine kinase inhibitor, has shown promising inhibition properties by inducing apoptosis in TNBC carcinogenesis in vivo. Despite knowledge of the 3D structure of CIP2A, no reports provide insight into CIP2A ligand binding sites. To this effect, we conducted in silico site identification guided by lapatinib binding. Four of the five sites identified were cross-validated, and the stem domain revealed more excellent ligand binding affinity. The binding affinity of lapatinib in these sites was further computed using the Molecular Mechanics/Poisson-Boltzmann Surface Area (MM/PBSA) approach. According to MM/PBSA//200 ns MD simulations, lapatinib exhibited a higher binding affinity against CIP2A in site 2 with ΔG critical values of −37.1 kcal/mol. The steadiness and tightness of lapatinib with CIP2A inside the stem domain disclosed glutamic acid-318 as the culprit amino acid with the highest electrostatic energy. These results provide clear information on the CIP2A domain capable of ligand binding and validate lapatinib as a promising CIP2A inhibitor in TNBC carcinogenesis.

Expanding Training Data for Structure-Based Receptor-Ligand Binding Affinity Regression through Imputation of Missing Labels.

The success of machine learning is, in part, due to a large volume of data available to train models. However, the amount of training data for structure-based molecular property prediction remains limited. The previously described CrossDocked2020 data set expanded the available training data for binding pose classification in a molecular docking setting but did not address expanding the amount of receptor-ligand binding affinity data. We present experiments demonstrating that imputing binding affinity labels for complexes without experimentally determined binding affinities is a viable approach to expanding training data for structure-based models of receptor-ligand binding affinity. In particular, we demonstrate that utilizing imputed labels from a convolutional neural network trained only on the affinity data present in CrossDocked2020 results in a small improvement in the binding affinity regression performance, despite the additional sources of noise that such imputed labels add to the training data. The code, data splits, and imputation labels utilized in this paper are freely available at https://github.com/francoep/ImputationPaper.