Machine Learned Classification of Ligand Intrinsic Activities at Human µ-Opioid Receptor.

Opioids are small-molecule agonists of µ-opioid receptor (µOR), while reversal agents such as naloxone are antagonists of µOR. Here, we developed machine learning (ML) models to classify the intrinsic activities of ligands at the human µOR based on the SMILES strings and two-dimensional molecular descriptors. We first manually curated a database of 983 small molecules with measured Emax values at the human µOR. Analysis of the chemical space allowed identification of dominant scaffolds and structurally similar agonists and antagonists. Decision tree models and directed message passing neural networks (MPNNs) were then trained to classify agonistic and antagonistic ligands. The hold-out test AUCs (areas under the receiver operator curves) of the extra-tree (ET) and MPNN models are 91.5 ± 3.9% and 91.8 ± 4.4%, respectively. To overcome the challenge of a small data set, a student-teacher learning method called tritraining with disagreement was tested using an unlabeled data set comprised of 15,816 ligands of human, mouse, and rat µOR, κOR, and δOR. We found that the tritraining scheme was able to increase the hold-out AUC of MPNN models to as high as 95.7%. Our work demonstrates the feasibility of developing ML models to accurately predict the intrinsic activities of µOR ligands, even with limited data. We envisage potential applications of these models in evaluating uncharacterized substances for public safety risks and discovering new therapeutic agents to counteract opioid overdoses.

Force Field X: A computational microscope to study genetic variation and organic crystals using theory and experiment.

Force Field X (FFX) is an open-source software package for atomic resolution modeling of genetic variants and organic crystals that leverages advanced potential energy functions and experimental data. FFX currently consists of nine modular packages with novel algorithms that include global optimization via a many-body expansion, acid-base chemistry using polarizable constant-pH molecular dynamics, estimation of free energy differences, generalized Kirkwood implicit solvent models, and many more. Applications of FFX focus on the use and development of a crystal structure prediction pipeline, biomolecular structure refinement against experimental datasets, and estimation of the thermodynamic effects of genetic variants on both proteins and nucleic acids. The use of Parallel Java and OpenMM combines to offer shared memory, message passing, and graphics processing unit parallelization for high performance simulations. Overall, the FFX platform serves as a computational microscope to study systems ranging from organic crystals to solvated biomolecular systems.

QM/MM Simulations of Afatinib-EGFR Addition: The Role of ß-Dimethylaminomethyl Substitution.

Acrylamides are the most commonly used warheads of targeted covalent inhibitors (TCIs) directed at cysteines; however, the reaction mechanisms of acrylamides in proteins remain controversial, particularly for those involving protonated or unreactive cysteines. Using the combined semiempirical quantum mechanics (QM)/molecular mechanics (MM) free energy simulations, we investigated the reaction between afatinib, the first TCI drug for cancer treatment, and Cys797 in the EGFR kinase. Afatinib contains a ß-dimethylaminomethyl (ß-DMAM) substitution which has been shown to enhance the intrinsic reactivity and potency against EGFR for related inhibitors. Two hypothesized reaction mechanisms were tested. Our data suggest that Cys797 becomes deprotonated in the presence of afatinib, and the reaction proceeds via a classical Michael addition mechanism, with Asp800 stabilizing the ion-pair reactant state ß-DMAM+/C797- and the transition state of the nucleophilic attack. Our work elucidates an important structure-activity relationship of acrylamides in proteins.

QM/MM simulations of EFGR with afatinib reveal the role of the ß-dimethylaminomethyl substitution.

Acrylamides are the most commonly used warheads of targeted covalent inhibitors (TCIs) directed at cysteines; however, the reaction mechanisms of acrylamides in proteins remain controversial, particularly for those involving protonated or unreactive cysteines. Using the combined semiempirical quantum mechanics (QM)/molecular mechanics (MM) free energy simulations, we investigated the reaction between afatinib, the first TCI drug for cancer treatment, and Cys797 in the EGFR kinase. Afatinib contains a ß-dimethylaminomethyl (ß-DMAM) substitution which has been shown to enhance the intrinsic reactivity and potency against EGFR for related inhibitors. Two hypothesized reaction mechanisms were tested. Our data suggest that Cys797 becomes deprotonated in the presence of afatinib and the reaction proceeds via a classical Michael addition mechanism, with Asp800 stabilizing the ion-pair reactant state ß-DMAM+/C797- and the transition state of the nucleophilic attack. Our work elucidates an important structure-activity relationship of acrylamides in proteins.

Machine Learned Classification of Ligand Intrinsic Activities at Human µ-Opioid Receptor.

Opioids are small-molecule agonists of µ-opioid receptor (µOR), while reversal agents such as naloxone are antagonists of µOR. Here we developed machine learning (ML) models to classify the intrinsic activities of ligands at the human µOR based on the SMILE strings and two-dimensional molecular descriptors. We first manually curated a database of 983 small molecules with measured E max values at the human µOR. Analysis of the chemical space allowed identification of dominant scaffolds and structurally similar agonists and antagonists. Decision tree models and directed message passing neural networks (MPNNs) were then trained to classify agonistic and antagonistic ligands. The hold-out test AUCs (areas under the receiver operator curves) of the extra-tree (ET) and MPNN models are 91.5±3.9% and 91.8± 4.4%, respectively. To overcome the challenge of small dataset, a student-teacher learning method called tri-training with disagreement was tested using an unlabeled dataset comprised of 15,816 ligands of human, mouse, or rat µOR, κOR, or δOR. We found that the tri-training scheme was able to increase the hold-out AUC of MPNN to as high as 95.7%. Our work demonstrates the feasibility of developing ML models to accurately predict the intrinsic activities of µOR ligands, even with limited data. We envisage potential applications of these models in evaluating uncharacterized substances for public safety risks and discovering new therapeutic agents to counteract opioid overdoses.

Machine Learning Models to Interrogate Proteome-Wide Covalent Ligandabilities Directed at Cysteines.

Machine learning (ML) identification of covalently ligandable sites may accelerate targeted covalent inhibitor design and help expand the druggable proteome space. Here, we report the rigorous development and validation of the tree-based models and convolutional neural networks (CNNs) trained on a newly curated database (LigCys3D) of over 1000 liganded cysteines in nearly 800 proteins represented by over 10,000 three-dimensional structures in the protein data bank. The unseen tests yielded 94 and 93% area under the receiver operating characteristic curves for the tree models and CNNs, respectively. Based on the AlphaFold2 predicted structures, the ML models recapitulated the newly liganded cysteines in the PDB with over 90% recall values. To assist the community of covalent drug discoveries, we report the predicted ligandable cysteines in 392 human kinases and their locations in the sequence-aligned kinase structure, including the PH and SH2 domains. Furthermore, we disseminate a searchable online database LigCys3D (https://ligcys.computchem.org/) and a web prediction server DeepCys (https://deepcys.computchem.org/), both of which will be continuously updated and improved by including newly published experimental data. The present work represents an important step toward the ML-led integration of big genome data and structure models to annotate the human proteome space for the next-generation covalent drug discoveries.

Drug resistance revealed by in silico deep mutational scanning and mutation tracker.

As COVID-19 enters its fifth year, it continues to pose a significant global health threat, with the constantly mutating SARS-CoV-2 virus challenging drug effectiveness. A comprehensive understanding of virus-drug interactions is essential for predicting and improving drug effectiveness, especially in combating drug resistance during the pandemic. In response, the Path Laplacian Transformer-based Prospective Analysis Framework (PLFormer-PAF) has been proposed, integrating historical data analysis and predictive modeling strategies. This dual-strategy approach utilizes path topology to transform protein-ligand complexes into topological sequences, enabling the use of advanced large language models for analyzing protein-ligand interactions and enhancing its reliability with factual insights garnered from historical data. It has shown unparalleled performance in predicting binding affinity tasks across various benchmarks, including specific evaluations related to SARS-CoV-2, and assesses the impact of virus mutations on drug efficacy, offering crucial insights into potential drug resistance. The predictions align with observed mutation patterns in SARS-CoV-2, indicating that the widespread use of the Pfizer drug has lead to viral evolution and reduced drug efficacy. PLFormer-PAF's capabilities extend beyond identifying drug-resistant strains, positioning it as a key tool in drug discovery research and the development of new therapeutic strategies against fast-mutating viruses like COVID-19.

Constant-pH Simulations with the Polarizable Atomic Multipole AMOEBA Force Field.

Accurately predicting protein behavior across diverse pH environments remains a significant challenge in biomolecular simulations. Existing constant-pH molecular dynamics (CpHMD) algorithms are limited to fixed-charge force fields, hindering their application to biomolecular systems described by permanent atomic multipoles or induced dipoles. This work overcomes these limitations by introducing the first polarizable CpHMD algorithm in the context of the Atomic Multipole Optimized Energetics for Biomolecular Applications (AMOEBA) force field. Additionally, our implementation in the open-source Force Field X (FFX) software has the unique ability to handle titration state changes for crystalline systems including flexible support for all 230 space groups. The evaluation of constant-pH molecular dynamics (CpHMD) with the AMOEBA force field was performed on 11 crystalline peptide systems that span the titrating amino acids (Asp, Glu, His, Lys, and Cys). Titration states were correctly predicted for 15 out of the 16 amino acids present in the 11 systems, including for the coordination of Zn2+ by cysteines. The lone exception was for a HIS-ALA peptide where CpHMD predicted both neutral histidine tautomers to be equally populated, whereas the experimental model did not consider multiple conformers and diffraction data are unavailable for rerefinement. This work demonstrates the promise polarizable CpHMD simulations for pKa predictions, the study of biochemical mechanisms such as the catalytic triad of proteases, and for improved protein-ligand binding affinity accuracy in the context of pharmaceutical lead optimization.

Why is the Omicron main protease of SARS-CoV-2 less stable than its wild-type counterpart? A crystallographic, biophysical, and theoretical study of the free enzyme and its complex with inhibitor 13b-K.

During the continuing evolution of SARS-CoV-2, the Omicron variant of concern emerged in the second half of 2021 and has been dominant since November that year. Along with its sublineages, it has maintained a prominent role ever since. The Nsp5 main protease (Mpro) of the Omicron virus is characterized by a single dominant mutation, P132H. Here we determined the X-ray crystal structures of the P132H mutant (or O-Mpro) as free enzyme and in complex with the Mpro inhibitor, the alpha-ketoamide 13b-K, and we conducted enzymology, biophysical as well as theoretical studies to characterize the O-Mpro. We found that O-Mpro has a similar overall structure and binding with 13b-K; however, it displays lower enzymatic activity and lower thermal stability compared to the WT-Mpro (with "WT" referring to the original Wuhan-1 strain). Intriguingly, the imidazole ring of His132 and the carboxylate plane of Glu240 are in a stacked configuration in the X-ray structures determined here. The empirical folding free energy calculations suggest that the O-Mpro dimer is destabilized relative to the WT-Mpro due to the less favorable van der Waals interactions and backbone conformation in the individual protomers. The all-atom continuous constant pH molecular dynamics (MD) simulations reveal that His132 and Glu240 display coupled titration. At pH 7, His132 is predominantly neutral and in a stacked configuration with respect to Glu240 which is charged. In order to examine whether the Omicron mutation eases the emergence of further Mpro mutations, we also determined crystal structures of the relatively frequent P132H+T169S double mutant but found little evidence for a correlation between the two sites.

Effect of Acetylation on the Nanofibril Formation of Chitosan from All-Atom De Novo Self-Assembly Simulations.

Chitosan-based materials have broad applications, from biotechnology to pharmaceutics. Recent experiments showed that the degree and pattern of acetylation along the chitosan chain modulate its biological and physicochemical properties; however, the molecular mechanism remains unknown. Here, we report, to the best of our knowledge, the first de novo all-atom molecular dynamics (MD) simulations to investigate chitosan's self-assembly process at different degrees and patterns of acetylation. Simulations revealed that 10 mer chitosan chains with 50% acetylation in either block or alternating patterns associate to form ordered nanofibrils comprised of mainly antiparallel chains in agreement with the fiber diffraction data of deacetylated chitosan. Surprisingly, regardless of the acetylation pattern, the same intermolecular hydrogen bonds mediate fibril sheet formation while water-mediated interactions stabilize sheet-sheet stacking. Moreover, acetylated units are involved in forming strong intermolecular hydrogen bonds (NH-O6 and O6H-O7), which offers an explanation for the experimental observation that increased acetylation lowers chitosan's solubility. Taken together, the present study provides atomic-level understanding the role of acetylation plays in modulating chitosan's physiochemical properties, contributing to the rational design of chitosan-based materials with the ability to tune by its degree and pattern of acetylation. Additionally, we disseminate the improved molecular mechanics parameters that can be applied in MD studies to further understand chitosan-based materials.

Mechanism of Dimer Selectivity and Binding Cooperativity of BRAF Inhibitors.

Aberrant signaling of BRAFV600E is a major cancer driver. Current FDA-approved RAF inhibitors selectively inhibit the monomeric BRAFV600E and suffer from tumor resistance. Recently, dimer-selective and equipotent RAF inhibitors have been developed; however, the mechanism of dimer selectivity is poorly understood. Here, we report extensive molecular dynamics (MD) simulations of the monomeric and dimeric BRAFV600E in the apo form or in complex with one or two dimer-selective (PHI1) or equipotent (LY3009120) inhibitor(s). The simulations uncovered the unprecedented details of the remarkable allostery in BRAFV600E dimerization and inhibitor binding. Specifically, dimerization retrains and shifts the αC helix inward and increases the flexibility of the DFG motif; dimer compatibility is due to the promotion of the αC-in conformation, which is stabilized by a hydrogen bond formation between the inhibitor and the αC Glu501. A more stable hydrogen bond further restrains and shifts the αC helix inward, which incurs a larger entropic penalty that disfavors monomer binding. This mechanism led us to propose an empirical way based on the co-crystal structure to assess the dimer selectivity of a BRAFV600E inhibitor. Simulations also revealed that the positive cooperativity of PHI1 is due to its ability to preorganize the αC and DFG conformation in the opposite protomer, priming it for binding the second inhibitor. The atomically detailed view of the interplay between BRAF dimerization and inhibitor allostery as well as cooperativity has implications for understanding kinase signaling and contributes to the design of protomer selective RAF inhibitors.

Machine Learning Models to Interrogate Proteomewide Covalent Ligandabilities Directed at Cysteines.

Machine learning (ML) identification of covalently ligandable sites may accelerate targeted covalent inhibitor design and help expand the druggable proteome space. Here we report the rigorous development and validation of the tree-based models and convolutional neural networks (CNNs) trained on a newly curated database (LigCys3D) of over 1,000 liganded cysteines in nearly 800 proteins represented by over 10,000 three-dimensional structures in the protein data bank. The unseen tests yielded 94% and 93% AUCs (area under the receiver operating characteristic curve) for the tree models and CNNs, respectively. Based on the AlphaFold2 predicted structures, the ML models recapitulated the newly liganded cysteines in the PDB with over 90% recall values. To assist the community of covalent drug discoveries, we report the predicted ligandable cysteines in 392 human kinases and their locations in the sequence-aligned kinase structure including the PH and SH2 domains. Furthermore, we disseminate a searchable online database LigCys3D (https://ligcys.computchem.org/) and a web prediction server DeepCys (https://deepcys.computchem.org/), both of which will be continuously updated and improved by including newly published experimental data. The present work represents a first step towards the ML-led integration of big genome data and structure models to annotate the human proteome space for the next-generation covalent drug discoveries.

Quantum Descriptors for Predicting and Understanding the Structure-Activity Relationships of Michael Acceptor Warheads.

Predictive modeling and understanding of chemical warhead reactivities have the potential to accelerate targeted covalent drug discovery. Recently, the carbanion formation free energies as well as other ground-state electronic properties from density functional theory (DFT) calculations have been proposed as predictors of glutathione reactivities of Michael acceptors; however, no clear consensus exists. By profiling the thiol-Michael reactions of a diverse set of singly- and doubly-activated olefins, including several model warheads related to afatinib, here we reexamined the question of whether low-cost electronic properties can be used as predictors of reaction barriers. The electronic properties related to the carbanion intermediate were found to be strong predictors, e.g., the change in the Cß charge accompanying carbanion formation. The least expensive reactant-only properties, the electrophilicity index, and the Cß charge also show strong rank correlations, suggesting their utility as quantum descriptors. A second objective of the work is to clarify the effect of the ß-dimethylaminomethyl (DMAM) substitution, which is incorporated in the warheads of several FDA-approved covalent drugs. Our data suggest that the ß-DMAM substitution is cationic at neutral pH in solution and promotes acrylamide's intrinsic reactivity by enhancing the charge accumulation at Cα upon carbanion formation. In contrast, the inductive effect of the ß-trimethylaminomethyl substitution is diminished due to steric hindrance. Together, these results reconcile the current views of the intrinsic reactivities of acrylamides and contribute to large-scale predictive modeling and an understanding of the structure-activity relationships of Michael acceptors for rational TCI design.

Integrative Approach to Dissect the Drug Resistance Mechanism of the H172Y Mutation of SARS-CoV-2 Main Protease.

Nirmatrelvir is an orally available inhibitor of SARS-CoV-2 main protease (Mpro) and the main ingredient of Paxlovid, a drug approved by the U.S. Food and Drug Administration for high-risk COVID-19 patients. Recently, a rare natural mutation, H172Y, was found to significantly reduce nirmatrelvir's inhibitory activity. As the COVID-19 cases skyrocket in China and the selective pressure of antiviral therapy builds in the US, there is an urgent need to characterize and understand how the H172Y mutation confers drug resistance. Here, we investigated the H172Y Mpro's conformational dynamics, folding stability, catalytic efficiency, and inhibitory activity using all-atom constant pH and fixed-charge molecular dynamics simulations, alchemical and empirical free energy calculations, artificial neural networks, and biochemical experiments. Our data suggest that the mutation significantly weakens the S1 pocket interactions with the N-terminus and perturbs the conformation of the oxyanion loop, leading to a decrease in the thermal stability and catalytic efficiency. Importantly, the perturbed S1 pocket dynamics weaken the nirmatrelvir binding in the P1 position, which explains the decreased inhibitory activity of nirmatrelvir. Our work demonstrates the predictive power of the combined simulation and artificial intelligence approaches, and together with biochemical experiments, they can be used to actively surveil continually emerging mutations of SARS-CoV-2 Mpro and assist the optimization of antiviral drugs. The presented approach, in general, can be applied to characterize mutation effects on any protein drug targets.

Electro-Biofabrication. Coupling Electrochemical and Biomolecular Methods to Create Functional Bio-Based Hydrogels.

Twenty years ago, this journal published a review entitled "Biofabrication with Chitosan" based on the observations that (i) chitosan could be electrodeposited using low voltage electrical inputs (typically less than 5 V) and (ii) the enzyme tyrosinase could be used to graft proteins (via accessible tyrosine residues) to chitosan. Here, we provide a progress report on the coupling of electronic inputs with advanced biological methods for the fabrication of biopolymer-based hydrogel films. In many cases, the initial observations of chitosan's electrodeposition have been extended and generalized: mechanisms have been established for the electrodeposition of various other biological polymers (proteins and polysaccharides), and electrodeposition has been shown to allow the precise control of the hydrogel's emergent microstructure. In addition, the use of biotechnological methods to confer function has been extended from tyrosinase conjugation to the use of protein engineering to create genetically fused assembly tags (short sequences of accessible amino acid residues) that facilitate the attachment of function-conferring proteins to electrodeposited films using alternative enzymes (e.g., transglutaminase), metal chelation, and electrochemically induced oxidative mechanisms. Over these 20 years, the contributions from numerous groups have also identified exciting opportunities. First, electrochemistry provides unique capabilities to impose chemical and electrical cues that can induce assembly while controlling the emergent microstructure. Second, it is clear that the detailed mechanisms of biopolymer self-assembly (i.e., chitosan gel formation) are far more complex than anticipated, and this provides a rich opportunity both for fundamental inquiry and for the creation of high performance and sustainable material systems. Third, the mild conditions used for electrodeposition allow cells to be co-deposited for the fabrication of living materials. Finally, the applications have been expanded from biosensing and lab-on-a-chip systems to bioelectronic and medical materials. We suggest that electro-biofabrication is poised to emerge as an enabling additive manufacturing method especially suited for life science applications and to bridge communication between our biological and technological worlds.

Analysis of the ERK Pathway Cysteinome for Targeted Covalent Inhibition of RAF and MEK Kinases.

The ERK pathway is one of the most important signaling cascades involved in tumorigenesis. So far, eight noncovalent inhibitors of RAF and MEK kinases in the ERK pathway have been approved by the FDA for the treatment of cancers; however, their efficacies are limited due to various resistance mechanisms. There is an urgent need to develop novel targeted covalent inhibitors. Here we report a systematic study of the covalent ligandabilities of the ERK pathway kinases (ARAF, BRAF, CRAF, KSR1, KSR2, MEK1, MEK2, ERK1, and ERK2) using constant pH molecular dynamics titration and pocket analysis. Our data revealed that the hinge GK (gate keeper)+3 cysteine in RAF family kinases (ARAF, BRAF, CRAF, KSR1, and KSR2) and the back loop cysteine in MEK1 and MEK2 are reactive and ligandable. Structure analysis suggests that the type II inhibitors belvarafenib and GW5074 may be used as scaffolds for designing pan-RAF or CRAF-selective covalent inhibitors directed at the GK+3 cysteine, while the type III inhibitor cobimetinib may be modified to label the back loop cysteine in MEK1/2. The reactivities and ligandabilities of the remote cysteine in MEK1/2 and the DFG-1 cysteine in MEK1/2 and ERK1/2 are also discussed. Our work provides a starting point for medicinal chemists to design novel covalent inhibitors of the ERK pathway kinases. The computational protocol is general and can be applied to the systematic evaluation of covalent ligandabilities of the human cysteinome.

Structure-Kinetics Relationships of Opioids from Metadynamics and Machine Learning.

The nation's opioid overdose deaths reached an all-time high in 2021. The majority of deaths are due to synthetic opioids represented by fentanyl. Naloxone, which is an FDA-approved reversal agent, antagonizes opioids through competitive binding at the mu-opioid receptor (mOR). Thus, knowledge of opioid's residence time is important for assessing the effectiveness of naloxone. Here we estimated the residence times of 15 fentanyl and 4 morphine analogs using metadynamics, and compared them with the most recent measurement of the opioid kinetic, dissociation, and naloxone inhibitory constants (Mann, Li et al, Clin. Pharmacol. Therapeut. 2022). Importantly, the microscopic simulations offered a glimpse at the common binding mechanism and molecular determinants of dissociation kinetics for fentanyl analogs. The insights inspired us to develop a machine learning (ML) approach to analyze the kinetic impact of fentanyl's substituents based on the interactions with mOR residues. This proof-of-concept approach is general; for example, it may be used to tune ligand residence times in computer-aided drug discovery.

Structure-Kinetics Relationships of Opioids from Metadynamics and Machine Learning Analysis.

The nation's opioid overdose deaths reached an all-time high in 2021. The majority of deaths are due to synthetic opioids represented by fentanyl. Naloxone, which is a FDA-approved reversal agent, antagonizes opioids through competitive binding at the µ-opioid receptor (mOR). Thus, knowledge of the opioid's residence time is important for assessing the effectiveness of naloxone. Here, we estimated the residence times (τ) of 15 fentanyl and 4 morphine analogs using metadynamics and compared them with the most recent measurement of the opioid kinetic, dissociation, and naloxone inhibitory constants (Mann et al. Clin. Pharmacol. Therapeut. 2022, 120, 1020-1232). Importantly, the microscopic simulations offered a glimpse at the common binding mechanism and molecular determinants of dissociation kinetics for fentanyl analogs. The insights inspired us to develop a machine learning approach to analyze the kinetic impact of fentanyl's substituents based on the interactions with mOR residues. This proof-of-concept approach is general; for example, it may be used to tune ligand residence times in computer-aided drug discovery.

An Integrative Approach to Dissect the Drug Resistance Mechanism of the H172Y Mutation of SARS-CoV-2 Main Protease.

Nirmatrelvir is an orally available inhibitor of SARS-CoV-2 main protease (Mpro) and the main ingredient of PAXLOVID, a drug approved by FDA for high-risk COVID-19 patients. Recently, a rare natural mutation, H172Y, was found to significantly reduce nirmatrelvir's inhibitory activity. As the COVID-19 cases skyrocket in China and the selective pressure of antiviral therapy builds up in the US, there is an urgent need to characterize and understand how the H172Y mutation confers drug resistance. Here we investigated the H172Y Mpro's conformational dynamics, folding stability, catalytic efficiency, and inhibitory activity using all-atom constant pH and fixed-charge molecular dynamics simulations, alchemical and empirical free energy calculations, artificial neural networks, and biochemical experiments. Our data suggests that the mutation significantly weakens the S1 pocket interactions with the N-terminus and perturbs the conformation of the oxyanion loop, leading to a decrease in the thermal stability and catalytic efficiency. Importantly, the perturbed S1 pocket dynamics weakens the nirma-trelvir binding in the P1 position, which explains the decreased inhibitory activity of nirmatrelvir. Our work demonstrates the predictive power of the combined simulation and artificial intel-ligence approaches, and together with biochemical experiments they can be used to actively surveil continually emerging mutations of SARS-CoV-2 Mpro and assist the discovery of new antiviral drugs. The presented workflow can be applicable to characterize mutation effects on any protein drug targets.